UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

High throughput antibody-based methods to track the architecture and operations of cell signalling systems Yue, Hon Lam Lambert 2020

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Notice for Google Chrome users:
If you are having trouble viewing or searching the PDF with Google Chrome, please download it here instead.

Item Metadata

Download

Media
24-ubc_2020_november_yue_honlamlambert.pdf [ 21.04MB ]
Metadata
JSON: 24-1.0394777.json
JSON-LD: 24-1.0394777-ld.json
RDF/XML (Pretty): 24-1.0394777-rdf.xml
RDF/JSON: 24-1.0394777-rdf.json
Turtle: 24-1.0394777-turtle.txt
N-Triples: 24-1.0394777-rdf-ntriples.txt
Original Record: 24-1.0394777-source.json
Full Text
24-1.0394777-fulltext.txt
Citation
24-1.0394777.ris

Full Text

  HIGH THROUGHPUT ANTIBODY-BASED METHODS TO TRACK THE ARCHITECTURE AND OPERATIONS OF CELL SIGNALLING SYSTEMS by Hon Lam Lambert Yue B.Sc. The University of British Columbia, 2016  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF  THE REQUIREMENTS FOR THE DEGREE OF  DOCTOR OF PHILOSOPHY  in The Faculty of Graduate and Postdoctoral Studies (Experimental Medicine)  THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver)  October 2020  © Hon Lam Lambert Yue, 2020  ii The following individuals certify that they have read, and recommend to the Faculty of Graduate and Postdoctoral Studies for acceptance, the dissertation entitled:  High throughput antibody-based methods to track the architecture and operations of cell signalling systems   submitted by Hon Lam Lambert Yue                   in partial fulfillment of the requirements for   the degree of Doctor of Philosophy            in              Experimental Medicine           Examining Committee:    Prof. Steven Pelech, Department of Medicine          Supervisor    Prof. Christopher Ong, Department of Urologic Sciences         Supervisory Committee Member     Prof. Vincent Duronio, Department of Medicine          Supervisory Committee Member     Prof. Michael Cox, Department of Urologic Sciences         University Examiner     Prof. Shoukat Dedhar, Department of Biochemistry and Molecular Biology      University Examiner    Prof. James D. Woodgett, Department of Medical Biophysics, University of Toronto     External University Examiner   Additional Supervisory Committee Members:    Prof.  Gregg Morin, Department of Medical Genetics         Supervisory Committee Member      iii Abstract  The antibody microarray platform has emerged as an innovative and economical tool for ultra-sensitive, semi-quantitative interrogation of protein expression, post-translational modifications such as phosphorylation, and interactions with other proteins and small molecules. While this technology possesses unique strengths such as compatibility to directly assay biofluids as well as tissue and cell specimens without enrichment, the interpretation of antibody microarray findings have been plagued by the biological complexity of protein-protein interactions, poor sample handling, and inadequately characterized antibodies. Further refinement of the antibody microarray platform can identify proteomic cell signalling events in a targeted approach that is still high throughput and relatively accurate. To this end, this thesis describes recent innovations in the design, production and characterization of over 600 phosphosite-specific antibodies used in printing antibody microarrays, as well as refinements in sample preparation and detection systems to yield significant improvements in data quality. The resulting KAM-1325 antibody microarray was then used to investigate the associated similarities and differences between epidermal growth factor and insulin-mediated metabolic signal transduction pathways, as well as the effects of inhibition of protein-tyrosine phosphatases and protein synthesis. Taken together, along with over 250 different experiments performed with the same array with samples from diversified model systems and treatments, I used correlative analysis to propose physiological protein-protein relationships with a focus on kinase-substrate connections. I have also developed maps of common cell signalling networks of known and proposed novel protein connections relating to the targets that are tracked with antibodies on the KAM-1325 antibody microarray. To validate some of these connections, in  iv vitro kinase assays were performed. These experiments involved direct incubation of cell lysates with purified recombinant protein kinases and surveillance of enhanced phosphorylation of substrate proteins with the antibody microarray. The continued elucidation of cell signalling systems will help discern between healthy and pathological states of cells and organisms, so that future therapies that utilize these signalling pathways to restore homeostasis and wellbeing can be developed.    v Lay Summary  Antibodies are a popular research tool used to help identify how proteins communicate between each other within cells of the body for exacting the appropriate functions in specific kinds of situations, such as when insulin is produced and circulated throughout the body after eating a meal to enhance glucose uptake. The use of large collections of antibodies printed on microscope slides to track biological specimens, a promising and powerful methodology known as antibody microarrays, has stagnated in recent years as researchers shifted their preference to adoption of alternative technologies like mass spectrometry. The work in this thesis highlights why antibody microarray technology is still worthy of application and development. I propose new ways to overcome previous hurdles and demonstrate the potential of the improved technology to advance the current understanding of cellular biology by uncovering novel connections between proteins.      vi Preface  The experiments presented in this collection of work were all designed by me under the supervision and assistance of Dr. Steven Pelech in his laboratory at Kinexus Bioinformatics Corporation, which is also where his University of British Columbia academic group is located.   All peptide-membranes used in Chapter 3 and peptides used for the production of antibodies printed on the antibody microarrays (KAM-900P, KAM-1150, and KAM-1325) were synthesized by Dr. Dirk Winkler. I partook in the preparation of the majority of the peptide antigens that were sent for immunization in rabbits and was involved with the majority of the work in purifying and characterizing the antibodies. The work in Chapter 3 regarding the purification and characterization of a generic anti-phosphotyrosine antibody was initiated by Dr. Shenshen Lai, and was partially published as an appendix in Lai, S. (2015). Investigations of the origin, regulation, and substrate specificities of protein kinases in the human kinome (Doctoral Dissertation). University of British Columbia. Retrieved from https://open.library.ubc.ca/ collections/ubctheses/24/items/1.0167195. I worked with Dr. Lai during her PhD studies in Dr. Pelech’s laboratory, and have since expanded the scope of the study, the findings of which are reported in this thesis.   The results and figures in Chapter 4 have been partially previously published and were adapted from Yue L., Sam C., Arora N., Winkler D., & Pelech, S. (2017) Antibody microarray and immunoblotting analyses of the EGF signaling phosphorylation network in human A431 epidermoid carcinoma cells. Clinical Protemoics Bioinformatics, 2: 1-10. doi:  vii 10.15761/CPB.1000119, as well as Yue L. & Pelech, S. (2018) Applications of High Content Antibody Microarrays for Biomarker Discovery and Tracking Cellular Signaling. Advances in Proteomics Bioinformatics, 2: 1-24 DOI: 10.29011/APBI-107.100007. I conducted most of the experimental work in these two publications and drafted both manuscripts together with Dr. Pelech. The bioinformatics analyses adapted from this publication were performed by Ms. Christine Sam and Dr. Pelech.   I was the primary investigator for Chapters 5-8 and performed all the antibody microarray experiments related to this work. I had assistance with the Western blotting and generation of lysates by cell culture from a team of undergraduate co-op students of whom I helped train and mentored. The DU145 and LNCaP cell lysates used in Chapter 5 were generated by Dr. Jim Peacock, but the antibody microarray experiments were performed by me. The data regarding those two cell lines are part of a larger study of which Dr. Christopher Ong is the principal investigator.   I consulted on the design of the statistical methodology for Chapter 8 with Mr. Peiyuan Zhu and Mr. Giorgio Sgarbi from the Statistical Opportunity for Students program at UBC but performed all of the analyses myself.  The KAM-1325 datasets used in the meta-analysis of correlating antibody pairs were either generated by me or were sourced from clients and collaborators of Kinexus Bioinformatics Corporation. The in vitro kinase-assay experiments using mouse brain lysates were performed by me but are part of a larger study in collaboration with Dr. Hava Gil Henn, who is the principal investigator behind that study.    viii Table of Contents Abstract ……………………………………………………………………………………………………………………… iii Lay Summary ………………………………………………………………………………………………………………  v  Preface …………..…………………………………………………………………………………………………………..  vi Table of Contents ………………………………………………………………………………………………………..  viii  List of Tables ……………………………………………………………………………………………………………….  xv List of Figures ………………………………………………………………………………………………………………  xviii  List of Abbreviations ………………………………………………………………………………………………….. xxiii  Acknowledgements ……..…………………………………………………………………………………………….. xxvi Dedication …………………..…………………………………………………………………………………………….. xxviii Chapter 1: Introduction ……………………………………………………………………………………………… 1 1.1        Protein phosphorylation in cell signalling ………………………………………………. 1 1.2        Protein kinase and protein phosphatases ………………………………………………. 4 1.3        EGF signalling and cancer ………………………………………………………………………. 10 1.4        Insulin signalling and diabetes ………………………………………….……………………. 13 1.5        Control of protein synthesis by phosphorylation ……………………………………. 15 1.6        Research strategies for tracking protein phosphorylation ………………………. 16 1.7        Research antibodies to track protein expression and phosphorylation …… 18 1.8        Phosphoproteome analysis ……………………………………….…………………………… 19 1.9        Network analysis ……………………………………………………………….…………………… 23 1.10 Research aims …………………………………………………………………………………………       25   ix Chapter 2: Materials and Methods…………………………………………………………………………….. 29 2.1 Modified Bradford protein assay …………………………………………………………….       29 2.2 Antibody production, purification, and characterization ……....………………..       29 2.3  Western blotting and variants ………………………………………………………….……..       30  2.3.1  Peptide dot blots ………………………….……………………………………………..        31  2.3.2  SPOT membranes ………………………………………………………………………..       31 2.4  Kinase substrate profiling assays ……………………………………………………………..       31 2.5  Cell culture ………………………………………………………….…………………………………..       32  2.5.1  Cell treatment and conditions ……………………………………………………..       32  2.5.2 Cell lysate preparation for Western blotting ………………………………..       34  2.5.3  Cell lysate preparation for microarray without fragmentation …....       34  2.5.4 Cell lysate preparation for microarray with fragmentation …………..       34 2.6 Kinex™ Antibody microarrays ……………………………………………………….…………       35  2.6.1  Sample labelling ………………………………………..…………………………………       35  2.6.2 Reporter antibody preparation …………………………………………………….       35  2.6.3  Probing and imaging …………………………………………………………………….       36  2.6.4  Microarray slide signal quantification …………………………………………..       37  2.6.5  Metadata statistical analysis ……………………….……………………………….       37 Chapter 3: Characterization of Phosphosite-specific Antibodies …………………………….……       38  3.1  Rationale …………………………………………………………………………………..…………....       38 3.2  Consensus sequence analysis using peptide dot blots ……………………………..       39  x 3.3  Characterization of immunoreactivity of residues N- and C-terminal to a  phospho-acceptor amino acid residue with SPOT membranes …..………..….       44 3.4  Antibody specificity for phosphoserine, phosphothreonine, and   phosphotyrosine residues …………………………………………………………………..……       48 3.5  Antibody yields versus specificity …………………………………………………………….       53 3.6  Production and characterization of polyclonal generic phosphotyrosine-    specific antibody ……………………………………………………...………………………….....       56 3.7  Discussion ……………………………………………………………………………………………....       60 Chapter 4: Optimization of High Content Antibody Microarrays …………………………………. 65  4.1 Rationale ………………………………………………………………………..………..……………..       65  4.2  Lability of phosphorylation sites ……………………………………..………..……………..       66  4.3  Maintenance of phosphorylation states by chemical cleavage …………….....       69  4.4  Antibody microarray analysis of EGF-treated A431 cells ……….…………….……       73  4.5  Validation of antibody microarray leads by Western blotting …………………..       76  4.6  Sandwich antibody microarray ……………………..……………………………..………….       79 4.7  Discussion ……………………..………………………………………………………………..………       84 Chapter 5: Investigations of EGF and Insulin Signalling Systems  …………………………………       89  5.1 Rationale ………………………………………………….……………..………………………………       89 5.2 KAM-900P analysis of EGF-signalling in A431, A549, DU145, HeLa, and          LNCaP cells ………………………………………………………………….…………………………..       90  5.3 KAM-1325 analysis of EGF-signalling in A431, A549, and HeLa cells …………       95 5.4  Western blotting analysis of A431, A549 and HeLa cells treated with EGF..       98  xi 5.5 KAM-1325 analysis of insulin-signalling in HepG2, HeLa, Jurkat and MCF7     cells ………………………………………………………………………………………………………….      105 5.6  Western blotting analysis of HepG2, HeLa, Jurkat and MCF7 cells treated         with insulin ………………………………………………………………………………………………      109 5.7 EGFR and INSR in vitro kinase assays for substrates in lysates from HeLa           cells and identification by KAM-1325 antibody microarrays …………………….      117 5.8 Comparison of EGFR and INSR signalling pathways ……………………………….…      120  5.9  Discussion ………………………………………………………………………………………………..      127 Chapter 6: Effect of Protein-Tyrosine Phosphatase Inhibition on Protein                      Phosphorylation ………………………………………………………………………………………… 131  6.1  Rationale ………………………………………………………………………….……..…………..….       131 6.2  Trends in the number of phosphorylated protein-tyrosine and protein-serine/threonine phosphosites in response to PAO and vanadate …………..       132 6.3 Trends in the stoichiometry of phosphorylated protein-tyrosine and          protein-serine/threonine-phosphosites in response to PAO and       vanadate ………………………………………………………………………….……………..……….      134 6.4  Western blotting of HeLa cells treated with PAO and sodium            orthovanadate …………………………………………………..…………..………………………..      137 6.5 Mapping changes in protein phosphorylation in response to PAO and      vanadate treatment in HeLa cells ………………………………………….…………………       145  6.6 Discussion .……………………………………………………………………………………………….      148   xii Chapter 7: Effect of Protein Synthesis Inhibition on Protein Phosphorylation ……..……… 151  7.1  Rationale .………………………………………………………………………………………..………       151 7.2  KAM-1325 antibody microarray analysis of HeLa cells treated with         anisomycin and cycloheximide ………………………………………………………………..      152 7.3  Western blotting analysis of HeLa cells treated with anisomycin ………..…..      156 7.4  Effect of anisomycin in HeLa cells on protein-serine, protein-threonine,            and protein-tyrosine phosphorylation ……………………………………………………..      161 7.5 Discussion ………………………………………………………………………………………………..      164 Chapter 8: Meta-analyses of Diverse Treatments of HeLa Cells for Mapping Protein                      Phosphorylation Networks……………………………………………………………..……..…..      168 8.1  Rationale ……………………………………...…………………………………………………………      168 8.2  Evaluation of the most reliable antibodies ……….……………………….……..……..      169 8.3 Identification of the best positively correlated protein pairs ..………………….      171 8.4  Evaluation of potential kinase-substrate relationships ……...…………………….      173 8.5  Validation of kinase-substrate relationships using in vitro kinase assay    data …………………………………………………………………………………………………….…..      196 8.6  Discussion …….………………………………………………………………………………………….      198 Chapter 9: Future Directions and Conclusions………………………………………….…………………..      203 9.1  Characterizations of polyclonal antibodies …………………………………….………..      203 9.2  Improved application of antibody microarrays …………...…………………………..      204 9.3  Exploration of cell signalling architectures …...……………………………..………….      206 9.4  Follow up validation of the antibody microarray ……………………………………..     207  xiii 9.5 Conclusion ……………………………………………………………………………………………….       210 Bibliography ………………..…………………………………………………………………………………………….. 212 Appendices ………………..……………………………………………………………….……………………………... 229 Appendix A  Kinex G-series peptide macroarray details ……….………………………….       229 Appendix B  Key EGF-induced changes in protein phosphorylation in A431 cells   as detected with the Kinex™ KAM-900P antibody microarray ….….       230 Appendix C  Key EGF-induced changes in protein tyrosine phosphorylation in       A431 cells as detected with the Kinex™ KAM-1150 antibody  microarray ……………………………………………………………………………….....       233 Appendix D  Target proteins and phosphorylation sites phosphorylated or dephosphorylated in response to EGF treatment …………………………       235 Appendix E Target phosphorylation sites phosphorylated or dephosphorylated          in response to EGF treatment from Western blot analysis …………...       258 Appendix F  Target phosphorylation sites phosphorylated or dephosphorylated          in response to insulin treatment from Kinex™ KAM-1325 antibody         microarray analysis ………………………………………………………………………       272 Appendix G  Target phosphorylation sites phosphorylated or dephosphorylated          in response to insulin treatment by Western blotting ………………….       288 Appendix H  Target protein phosphorylation sites phosphorylated by          recombinant EGFR or INSR as identified by the Kinex™ KAM-1325              antibody microarray …………………………………………………………………….       292  xiv Appendix I  Protein targets changing in phosphorylation or expression in HeLa       cells treated with protein translation inhibitors anisomycin and cycloheximide …..………………………………………………………………………….       309 Appendix J Western blotting of HeLa cells treated anisomycin indicating           changes in protein phosphorylation and expression …..……………....       318 Appendix K  Listing of 320 antibodies classified as reliable based on internal correlation with other antibodies targeting the same protein ……..       322     xv List of Tables  Table 2.1  Treatment conditions, durations, in serum-free or supplemented media...          33 Table 3.1  Frequency of amino acids identified positionally to a central phosphosite  as part of a consensus sequence for phosphoserine-, phosphothreonine-  and phosphotyrosine site-specific antibody immunoreactivity ……..…………         41 Table 3.2  Frequency of amino acids identified as part of a consensus sequence for antibody immunoreactivity compared to original frequency in all          immunizing peptides ……………………………………………………………………………….         43 Table 3.3  Number and percent frequency of critical and non-critical amino acid       residues …………………………………………………………………………………………………..         48 Table 3.4  Phosphosite-specific antibody yields by phosphoamino acid specificity …..         55 Table 3.5  Generic and site-specific phosphotyrosine antibody yields ……………….…….         56 Table 4.1  Comparison of lead generation from different homogenization and         detection systems using EGF-treated A431 cells and the Kinex™ KAM-              900P antibody microarray ……………….……..…………………….…………………………         75 Table 4.2  Assessment of EGF-induced changes in protein phosphorylation in A431          cells as detected with the Kinex™ KAM-900P antibody microarray and       Western blotting ……………..………………………………………………………………………         79 Table 5.1  Kinex™ KAM-900P antibodies revealing increased or decreased     phosphorylation in response to EGF treatment …………………..…………………..         93 Table 5.2  Kinex™ KAM-1325 antibodies revealing increased or decreased    phosphorylation in response to EGF treatment ……..………………………………..         96  xvi Table 5.3  Kinex™ KAM-1325 target phosphosites in HeLa cells significantly increased         or decreased in phosphorylation in response to EGF incubation ……………..         97 Table 5.4  Phosphosite-specific antibodies revealing increased or decreased phosphorylation in response to EGF treatment by Western blotting ……….      102 Table 5.5 Kinex™ KAM-1325 antibodies revealing increased or decreased     phosphorylation in response to insulin treatment ………………….……………….      107 Table 5.6 Phosphosite-specific antibodies revealing increased or decreased phosphorylation in response to insulin treatment by Western blotting …..      111 Table 5.7  Kinex™ KAM-1325 tyrosine phosphosite-specific antibodies revealing     increased phosphorylation following recombinant kinase incubation with              fragmented HeLa cell lysate proteins ……………………………….……………………..      119 Table 6.1 Kinex™ KAM-1325 antibodies revealing increased or decreased protein    phosphorylation in response to protein-tyrosine phosphatase inhibition ..      134 Table 6.2  Average and median %CFC values of Kinex™ KAM-1325 antibodies in  response to PAO and vanadate treatment ……………………………………………….      135 Table 6.3 Top 5% of Kinex™ KAM-1325 phosphoantibodies that indicated increases           or decreases with PAO and vanadate treatment of HeLa cells ……….………..      136 Table 6.4 Phosphoantibodies revealing increases or decreases in phosphorylation in response to PAO and sodium orthovanadate treatment in HeLa cells by    Western blotting …………………………………………………………………………..….……..      138 Table 6.5 Increased phosphorylation of sites on protein phosphatases in response              to PAO and sodium orthovanadate by Western blotting …….……………………      145  xvii Table 7.1  Kinex™ KAM-1325 antibodies revealing changes in HeLa cells in response            to anisomycin or cycloheximide treatment ………………………..…………………….      154 Table 7.2  Western blotting leads with HeLa cells treated with anisomycin ……………..      159 Table 7.3  Type of phosphosite increased or decreased in phosphorylation in            response to anisomycin treatment in HeLa cells by Kinex™ KAM-1325 analysis and by Western blotting …………………………………………………….…….…      162 Table 8.1  Number of significant correlated antibody pairs in meta dataset and HeLa        cell dataset ………………………………………………………..………………………….…………      172 Table 8.2  High confidence kinase-substrate phosphosite correlated pairs validated            by empirical studies …………………………………………………………..…………………….      173 Table 8.3  Kinase-phosphosite pairs where the correlated kinase was ranked within         the top 10 predicted kinases on PhosphoNET ………………………………………….      175 Table 8.4 Possible protein pathways connecting kinases and correlated substrate       protein phosphosites ……………………………………………………………………………….      181 Table 8.5 Comparison of kinase-assay data of HeLa or mouse brain lysates             incubated with recombinant kinases analyzed on the microarray with the correlated kinase-phosphosite leads generated from microarray              metadata analysis …………………………………………………………………………………...      197    xviii List of Figures  Figure 1.1  Alignment of the catalytic domain of human protein kinases that define              12 highly conserved subdomains …………………………………………………………….. 6 Figure 1.2 Schematic of EGFR signalling pathways ……………………………………………………         11 Figure 3.1  Representative SPOT peptide membranes featuring positional glycine substitutions probed with phosphosite-specific antibodies ……………………..         45 Figure 3.2  Average signal intensity of phosphoserine, phosphothreonine, and phosphotyrosine site-specific antibodies on peptides featuring singular       glycine substitutions from three residues N-terminal to three residues                 C-terminal to the phospho-accepting amino acid residue ………..………………  46 Figure 3.3 Representative SPOT peptide membranes featuring phosphoamino acid substitutions probed with phosphosite-specific antibodies ……………………..         49 Figure 3.4  Average signal intensity of SPOT membranes featuring phosphoacceptor          site substitutions probed with phosphosite specific antibodies ……………….         51 Figure 3.5  Venn diagrams of cross-reactivities of phosphoserine, phosphothreonine          and phosphotyrosine-site antibodies …………………………..………………………….         52 Figure 3.6  Production yield plotted against the number of defined residues in        consensus sequences ………………………………………………………………………………         54 Figure 3.7  PYK and P-Tyr-1000 generic phosphotyrosine antibody specificities on   Kinex™ G Macro phosphopeptide arrays ……………….………………………………..         59 Figure 3.8  Western blotting of A431 and HeLa cells with PYK antibody ……………………         60  xix Figure 4.1  Lability of EGF receptor phosphosites in lysates prepared from EGF-            treated A431 cells homogenized in standard buffers ………………………………         67 Figure 4.2  Solubilization of EGFR in Triton X-100 and SDS-PAGE sample buffer ………. 68 Figure 4.3 Schematic representation of alternative lysate sample and antibody      microarray processing strategies ……………………………………………………………..         69 Figure 4.4 Scanned images of Kinex™ KAM-900P antibody microarrays incubated             with lysates from EGF-treated A431 cells subjected to alternative lysate     sample and antibody microarray processing strategies ……………………….…..         72 Figure 4.5  Western blotting validation of antibody microarray leads of EGF induced  protein phosphorylation ………………………………………………………………………….         77 Figure 4.6  Kinex™ KAM-1150 antibody microarray analyses of human A431 cervical carcinoma cell lysates following treatment of the cells with and without           100 ng/ml EGF for 5 min for protein expression levels as well as tyrosine phosphorylation …..………………………………………………………………………….………         81 Figure 4.7  Kinex™ KAM-1150 antibody microarray analyses of human HeLa cervical carcinoma cells treated with 50 µM of the proteasome inhibitor MG-132           for 4 hours ……………………………………………………….………………………………………         83 Figure 5.1  Number of Kinex™ KAM-900P target protein phosphorylation sites             phosphorylated or dephosphorylated in response to EGF treatment ……...        94 Figure 5.2  Western blotting with EGFR phosphosite-specific antibodies in A431,              A549 and HeLa cells ……………..………………………………………………………………….        99  xx Figure 5.3  Western blotting with phosphosite-specific antibodies with lysates from       A431, A549 and HeLa cells following EGF treatment ……………….………………. 100 Figure 5.4  Number of target proteins and phosphorylation sites phosphorylated or dephosphorylated in response to EGF treatment by Western blotting …….       103 Figure 5.5  Analysis of true positive, true negative, false positive and false negative      results of the EGF-signalling antibody microarray results validated by        Western blotting ……………………………………………………………………………………..       105 Figure 5.6  Number of Kinex™ KAM-1325 target protein phosphorylation sites phosphorylated or dephosphorylated in response to insulin treatment ….        108 Figure 5.7 Western blotting with RPS6 phosphosite-specific antibodies in HeLa cells..      109 Figure 5.8 Western blotting of protein phosphosites in Jurkat and MCF7 cells up-    regulated with insulin treatment ……………..……………………………………………..        112 Figure 5.9 Western blotting of protein phosphosites in Jurkat and MCF7 cells down-regulated with insulin treatment …………………………………………………………….        113 Figure 5.10 Number of target protein phosphorylation sites phosphorylated or dephosphorylated in response to insulin treatment by Western blotting .. 114 Figure 5.11 Analysis of true positive, true negative, false positive and false negative      results of the insulin-signalling antibody microarray results validated by    Western blotting …………………………………………………………………………………….. 116 Figure 5.12 Number of target protein phosphorylation sites phosphorylated by   recombinant EGF receptor or insulin receptor as identified by the Kinex™  KAM-1325 antibody microarray ……………………………………………..…………...….       120  xxi Figure 5.13  Protein phosphosite leads identified as EGF or insulin responsive …………… 121 Figure 5.14 Predicted interconnectivity of EGF-responsive cell signalling proteins ……..       123 Figure 5.15 Predicted interconnectivity of insulin-responsive cell signalling proteins …        125 Figure 5.16  Interconnectivity of cell signalling proteins that responded similarly or       contrarily to EGF and insulin ………………………………………………….…………………       126 Figure 6.1  Western blotting of protein-tyrosine phosphosites up-regulated with PAO       and vanadate treatment ………………………………………………………………………….         139 Figure 6.2  Western blotting of protein-serine and threonine phosphosites up-         regulated with PAO and vanadate treatment …………………………………………..       140 Figure 6.3  Western blotting of protein phosphosites down-regulated with PAO and vanadate treatment …………………………………………………………………………………        141 Figure 6.4 Validation of true positive, true negative, false positive and false negative antibody microarray phosphorylation results from lysates of HeLa cells       treated with PAO and sodium orthovanadate for 15 min by Western blotting ….…………………………………………………………………….………………………….        143 Figure 6.5  Interconnectivity of proteins changing in phosphorylation in response to      hyper-protein-tyrosine phosphorylation due to protein-tyrosine          phosphatase inhibition in HeLa cells ………………………………………………………..       147 Figure 7.1 Number of protein targets increased or decreased in phosphorylation and expression with anisomycin or cycloheximide treatment of HeLa cells ..….        155 Figure 7.2  Representative Western blotting results from lysates of HeLa cells treated      with anisomycin for 1-2 h for increases in protein phosphorylation …………      157  xxii Figure 7.3  Representative Western blotting results from lysates of HeLa cells treated  with anisomycin for 1-2 h for decreases in protein phosphorylation and expression. …………………………….……………………………………………………………….. 158 Figure 7.4  Validation of true positive, true negative, false positive and false negative antibody microarray results from lysates of HeLa cells treated with       anisomycin for 2 h by Western blotting ………………………………………….……….        160 Figure 7.5  KiNetscape map of the interconnections between proteins that are             altered in protein expression and/or phosphorylation with anisomycin   treatment of HeLa cells ……………………………………………………………………….…..        163 Figure 8.1  Heatmap of pair-wise correlations between all antibodies on the Kinex™        KAM-1325 antibody microarrays ……………………..………………………………………       170 Figure 8.2  Interconnectivity of kinases and substrates in predicted cell signalling      networks ………………………………………………………………………………………………….       178 Figure 9.1 Annual publications relating to antibody microarrays ………………………………      210      xxiii List of Abbreviations  %CFC percent change from control 4EBP1 eukaryotic translation initiation factor 4E-binding protein  A alanine ADP adenosine diphosphate  ANOVA analysis of variance Ala alanine Arg arginine Asp aspartic acid ATP adenosine triphosphate  BSA bovine serum albumin  C cysteine cAMP cyclic-adenosine monophosphate CCC cysteine chemical cleavage   CCCB cysteine chemical cleavage buffer CD cluster of differentiation Cys cysteine D aspartic acid DMEM Dulbecco’s Modified Eagle Medium  E glutamic acid ECL enhanced chemiluminescence EGF epidermal growth factor EGFR/ErbB1/HER1 epidermal growth factor receptor EGR1 early growth response 1  ELISA enzyme-linked immunosorbent assay ePK eukaryotic protein kinase ERK1 extracellular regulated kinase-1 F phenylalanine FAK focal adhesion kinase  FBS fetal bovine serum FCP TFIIF-associating component of RNA polymerase II CTD phosphatase FOXO1 forkhead box O1 FSBA 5ʹ-(4-fluorosulphonylbenzoyl)adenosine hydrochloride G glycine Gln glutamic acid Gln glutamine Gly glycine Grb2 growth factor receptor-bound protein 2   xxiv h hour H histidine His histidine HLA human leukocyte antigen HPLC high-pressure liquid chromatography  I isoleucine IEG immediate early genes  IGF1R insulin-like growth factor 1 receptor  Ile isoleucine INSR insulin receptor  IRS insulin receptor substrate  JNK c-Jun N-terminal kinase K lysine KLH keyhole limpet hemocyanin  KSR Kinase-substrate relationship L leucine Leu leucine Lys lysine M methionine min minute MAP2K/MEK mitogen-activated protein kinase kinase  MAPK mitogen-activated protein kinase MEM Minimum Essential Media Met methionine MS mass spectrometry mTOR mammalian target of rapamycin NAD+ nicotinamide adenine dinucleotide NTCB 2-nitro-5-thiocyanotabenzoic acid  P proline p90S6K p90 ribosomal S6 kinase  PAO phenylarsine oxide PBS phosphate-buffered saline  PDK1 3-phosphoinositide-dependent kinase 1 PEPCK phosphoenolpyruvate carboxykinase  PH pleckstrin homology  Phe phenylalanine PI3K phosphatidylinositol 3-kinase PIP3 phosphatidylinositol-3,4,5-biphosphate PKB/Akt protein kinase B  PPM protein phosphatase Mg2+ or Mn2+ dependent  xxv PPP phosphoprotein phosphatase Pro proline pS phosphoserine pT phosphothreonine PTB protein-tyrosine binding  PTM post translational modification PTP protein-tyrosine phosphatase pY phosphotyrosine Q glutamine R arginine RPMI Rosewell Park Memorial Institute  RPS6 ribosomal protein S6 RTK receptor-tyrosine kinase S serine SDS sodium dodecylsulphate SDS-PAGE Sodium dodecylsulphate-polyacrylamide gel electrophoresis sec second SEMA3C semaphorin 3C Ser serine SH2 Src-homology-2  SHB Standard Homogenizing Buffer T threonine TCEP Tris(2-carboxyethyl) phosphine hydrochloride Thr threonine TKI protein-tyrosine kinase inhibitor TORC1 target of rapamycin complex 1  Trp tryptophan TSC  tuberous sclerosis protein Tyr tyrosine V valine Val valine W tryptophan Y tyrosine    xxvi Acknowledgements  I am deeply grateful for the enlightening support from many inspiring mentors who have guided me along this journey. I thank my chemistry teacher, Dr. Terry Jarvis, who recognized my curiosity for the sciences and planted the idea of obtaining a PhD. I also thank Ms. Milah Woo for diverting my attention into joining a co-op program and leading me to my initial collaboration with Dr. Steven Pelech’s lab that ultimately grew into the thesis presented here.   I thank all the friends I have made along the way, from fellow Pelech Lab mates to colleagues in other labs, for their companionship, words of support, and practical aid in my research. I would also like to thank Kinexus Bioinformatics Corporation, the University of British Columbia, and NSERC for their financial support and scholarship titles.   I am deeply indebted to the outstanding support and opportunities offered to me by Dr. Steven Pelech, Ms. Catherine Sutter, and Dr. Dirk Winkler, who have all been my academic parents, raising me since my junior undergraduate years. They treated me with true compassion and expended significant effort to provide the best experience that I could have. Their advice was invaluable throughout my graduate studies and in broader life. I am also honored to have worked alongside Dr. Shenshen Lai, who taught me how to work in the lab and to be a critical thinking scientist. I thank my supervisory committee members, Dr. Christopher Ong, Dr. Gregg Morin, and Dr. Vincent Duronio, for their kind advice and constructive critiques.    xxvii Lastly, I would like to thank my family, for their encouragement and support. Above all, I thank my spouse for being my emotional pillar and the foundation to the nest where I can lick my wounds after trials and tribulations, and for the life we continue to build.       xxviii              This thesis is dedicated to all my past and present mentors who signalled to me along my path.     1 Chapter 1: Introduction  1.1  Protein phosphorylation in cell signalling Cell signalling is the process of communicating information via molecular interactions throughout the cell and is the chief means by which all basic life functions are governed. An example of signal transduction may begin with the recognition of a stimulus from outside of the cell, such as the binding of an extracellular mediator with a membrane receptor, which precipitates a cascade of subsequent events that are ideally appropriate under the specific circumstances for that cell. This signal is translated into a series of downstream interactions that includes regulation of enzyme activities, and ultimately evokes diverse responses within the cell such as post-translational modifications and gene expression changes. Cell signalling systems encompass all classes of biomolecules, ranging from small calcium ions, secondary messengers like 3’-5’-cyclic-adenosine monophosphate (cAMP) and bio-lipids, to large effector protein complexes like ribosomes and transmembrane receptors and associated proteins (Bootman et al., 2001; Hou et al., 2016; Lemmon and Schlessinger, 2016; Yan et al., 2016). Regulation of cell signalling can occur at the protein level by managing the rates of a protein’s synthesis and degradation, functional efficiency, subcellular location, and interactions controlled by covalent modifications.   Protein phosphorylation is one of the most prevalent mechanisms for reversible post-translational regulation and control of cell signalling in eukaryotic cells (Pawson and Scott,  2 2005). This involves the addition of a highly charged phosphoryl group onto a protein by kinases, and its removal by protein phosphatases (Cohen, 2002). More than 20 other common types of covalent modifications of cellular proteins have also been reported, and these include most notably ubiquitination, sumoylation, S-nitrosylation, O- and N-linked glycosylation, acetylation, methylation and lipidation (Garcia, 2019). However, protein phosphorylation is by far the most prevalent.  The majority of protein phosphorylation occurs on serine, threonine, and tyrosine residues, although phosphorylation on histidine, lysine, and arginine have also been reported (Matthews, 1995; Needham et al., 2019). In chick embryo fibroblasts, approximately 92% of phosphorylation of proteins occurs on serine residues, 8% on threonine residues and 0.04% on tyrosine residues by weight (Hunter and Sefton, 1980). Major phosphoproteome databases, such as PhosphoSitePlus (www.phosphosite.org) and PhosphoNET (www.phosphonet.ca), have documented at least 293,897 unique phosphosites from over 22,000 research articles as well as thousands of mass spectrometry (MS) datasets. Prediction algorithms that have been used in PhosphoNET indicate that the actual number of human phosphosites may be closer to a million, but the vast majority of these are likely to be of low stoichiometry and inconsequential. Of the identified phosphosites by MS, 59.6% were serine phosphorylations, 24.7% were on threonine, and 15.7% were on tyrosine (Hornbeck et al., 2019). However, a large fraction of MS studies used enrichment techniques to isolate phosphotyrosine-containing peptides from proteins. Without such enrichment, it appears that around 76%, 20% and 4% of the reported phosphosites are found on serine, threonine and tyrosine residues, respectively (S. Pelech,  3 unpublished data). Compared to serine and threonine sites, tyrosine phosphorylation sites have been proposed to be more functional and better conserved within the evolutionary phylogeny of vertebrates (Miao et al., 2018). However, protein-tyrosine phosphorylation is almost exclusively associated with multicellular organisms. Thus, serine and threonine phosphorylation sites are more conserved in eukaryotes in general. In particular, functional threonine phosphorylation sites tend to be especially conserved and 8-times more likely to be activatory than inhibitory (Safaei et al., 2011b).   The coupling of a phosphoryl group onto a protein imparts a strong negative charge that could induce significant conformational changes into the local 3D structure of the protein. Phosphorylation can lead to an opening of an enzyme’s structure to increase substrate access to its catalytic active site, such as occurs with Thr-202 and Tyr-204 phosphorylation in extracellular-regulated kinase-1 (ERK1), or inactivation of the enzyme due to a new stabilized structure that obstructs the active site, such as Tyr-527 phosphorylation in proto-oncogene-encoded protein-tyrosine kinase Src (c-Src) (Day et al., 2016). Tyrosine phosphorylation can often drive formation of protein complexes between signalling proteins due to phosphotyrosine recognition binding domains such as Src-homology-2 (SH2) and protein-tyrosine binding (PTB) domains (Wagner et al., 2013). At least 119 SH2 domains have been recognized in 109 different human proteins, while 54 PTB domains have been identified in another 54 human proteins (SafaeiMehranpour, 2015). In addition to SH2 and PTB domains, several atypical binding domains with affinity to phosphotyrosine have also been documented (Kaneko et al., 2012). Proteins can also feature domains that recognize phosphorylated serine or threonine residues  4 for binding. Seven 14-3-3 isoforms have been recognized to interact with over 200 different proteins, rendering the interfacing phosphorylation sites inaccessible to other binding partners (Mhawech, 2005). While some phosphosites can be analogous to “on” or “off” switches, most proteins features multiple phosphorylation sites, such as Thr-24, Ser-256, and Ser-319 on Forkhead box protein O1, which fine tune different catalytic functions for the most appropriate response within a given environmental context (Cohen, 2000; Li et al., 2017). Hyper-phosphorylation of a protein at multiple sites is also a mechanism that has evolved to tag proteins for their degradation (Varedi et al., 2010). Phosphorylation can serve to alter proteins by regulating their enzymatic activities, binding capabilities to other proteins, or by facilitating their protein stability or degradation. These outputs all function in networks and cascades to produce appropriate responses to diverse stimuli that the cell may receive.   1.2  Protein kinase and protein phosphatases Over 535 protein kinases are encoded by the human genome, which accounts for about 2.5% of all human protein coding genes (Manning et al., 2002, www.kinasenet.ca). Departure from the normal functioning of kinases can lead to diseases, and tumours in particular, lending them to become attractive targets for development of counteractive therapeutics (Ardito et al., 2017). Functionally, kinases can be largely organized into three main types based on their substrate specificities. Protein kinases typically have specificity for either serine and threonine residues, tyrosine residues, or all three in which they have been referred to as “dual” specificity kinases (Ubersax and Ferrell, 2007). It is believed that protein-tyrosine kinases actually evolved from protein-serine/threonine kinases as a major driving force in the evolution of early metazoans  5 (Darnell et al., 1997; Rokas et al., 2005). The phylogeny of human protein kinases is categorized into two main groups: atypical kinases, and eukaryotic protein kinases (ePKs), the latter of which all share sequence similarity in a stretch of approximately 247 amino acids that forms the catalytic kinase domain (Hanks et al., 1988). This sequence contains 12 catalytic subdomains marked by around 30 highly conserved amino acids, such as the subdomain I consensus sequence GXGXXG, the DFG motif of subdomain VII, and the APE motif of subdomain VIII (Figure 1.1). The glycine residues within subdomain I are critical in facilitating kinase interaction with the ATP molecule, and any phosphorylation sites found here are predicted to disrupt the ATP binding ability of the kinase and be inhibitory of its overall function (Hanks et al., 1998). The region between subdomains VII and VIII houses the activation loop, which in its inactive form, is protected inside the protein structure in a hydrophobic pocket (Scheeff et al., 2009). Phosphorylation in this region, as exemplified with phosphorylation of the Thr-202 and Tyr-204 sites of ERK1 by mitogen-activated protein kinase kinase (MAP2K/MEK1), exposes the catalytic portion of many kinases and leads to their enzymatic activation (Ubersax and Ferrell, 2007). By contrast, the flanking Thr-207 and Tyr-210 sites in ERK1 seem to be important in the proper structural configuration of the active site, since mutation of either of these residues results in loss of activity. Furthermore, autophosphorylation of these residues is known to decrease phosphotransferase activity, and phosphorylation of these highly conserved phosphosites in most protein-serine/threonine kinases could serve as a general intrinsic shut off mechanism of this class of protein kinases (Lai and Pelech, 2016).  6   7  Figure 1.1. Alignment of the catalytic domain of human protein kinases that define 12 highly conserved subdomains. Alignment is based on 497 protein kinase domains identified in 485 human protein kinases. The alignment of 9 representative kinases are shown here. The number of amino acid residues of gap regions are highlighted in yellow. Amino acids with side chains of similar properties are colour coded as follows: Pink and red – acidic amino acids, blue – basic amino acids, green – bulky hydrophobic side chains, brown – proline.    8 Alignment of all the typical human protein kinase catalytic domains with proteins most related to protein kinases in six other very diverse species indicates that a common ancestor related to glutaminyl aminoacyl-tRNA synthetases may have given rise to both ePKs and choline kinases (Lai et al., 2016). Phylogenetic analysis of the catalytic kinase domain further categorizes the ePK superfamily into eight major groups, which are then also subdivided into 132 families, based on the KinBase database (http://kinase.com). The largest group, the protein-tyrosine kinases, are further divided into receptor and non-receptor varieties. While non-receptor protein tyrosine kinases are fully intracellular, receptor-tyrosine kinases (RTKs) are a type of cell-surface receptor with an extracellular ligand binding domain, a single transmembrane helix with hydrophobic amino acid residues, an intracellular tyrosine kinase domain, and a carboxyl terminus region. In general, binding of ligands like growth factors induces receptor dimerization (or in some cases oligomers), activation of intrinsic kinase domains, and cross phosphorylation in trans (Lemmon and Schlessinger, 2010).   Protein phosphorylation is dependent on the activities of both protein kinases and protein phosphatases and is the result of the relative balance of both types of enzymatic activities. Not counting regulatory subunits, 194 human proteins were found to contain phosphatase catalytic domains, and at least 154 human protein-phosphatases have been documented (Liberti et al., 2013; Sacco et al., 2012, SafaeiMehranpour, 2015). Unlike protein kinases, many protein phosphatases seem to have evolved separately and are mechanically distinct in their mode of hydrolysis of the phosphoryl group. While there are three main families of protein-serine/threonine phosphatases (PPP, PPM, and FCP), the protein-tyrosine phosphatases (PTPs)  9 are characterized by a consensus motif of HCX5R, in which the cysteine is critical in the phosphatase reaction (Cohen, 2004; Tonks, 2006). Protein phosphatases can also be “dual” specific and dephosphorylate both serine/threonine as well as tyrosine phosphorylated residues. Like protein-tyrosine kinases, protein-tyrosine phosphatases also come in transmembrane receptor-like and non-transmembrane varieties. The catalytic domain of classical PTPs spans about 280 residues and feature conserved motifs that are functional, many of which participate in recognition of phosphotyrosine (Andersen et al., 2001). Protein-tyrosine phosphatase inhibitors, phenylarsine oxide (PAO) and vanadate, as well as the protein-serine/threonine protein phosphatase inhibitor okadaic acid, all affect the insulin signalling pathway and glucose metabolism in part through inhibition of phosphatases (Carey et al., 1995; Li et al., 1996). PAO is known to inhibit a variety of protein phosphatases such as CD45 and PTPN1, as well as other phosphatases including acid phosphatase-1 through oxidative stress of critical cysteine residues required for phosphatase activity (Garcia-Morales et al., 1990; Kim et al., 2011; Liao et al., 1991). Vanadate, which can also induce oxidation stress, also functions as a competitive inhibitor by acting as a phosphate analogue (Huyer et al., 1997). Sodium orthovandate, as well as protein-serine/threonine phosphatase inhibitors like beta-glycerophosphate and sodium fluoride, are often common ingredients used in buffer formulations for cell lysis in order to preserve the phosphorylation states of cell systems during treatment conditions of interest (Hulley et al., 1998; Jaumot and Hancock, 2001). The relatively lesser number of phosphatases than kinases and the apparently generally looser substrate specificity of protein phosphatases are challenges to finding very specific phosphatase inhibitors. This has resulted in less progress for using phosphatase inhibitors as therapies for  10 diseases as compared to kinase inhibitors. Two notable exceptions are cyclosporine and tacrolimus, which are potent inhibitors of protein phosphatase 2B (calcineurin) and were proven to be very useful drugs for immune suppression and organ transplantation (Azzi et al., 2013). However, since protein phosphorylation accounts for the regulation of so many critical cell signalling events, such as cell cycle regulation, research in this field is still promising and an incredibly important endeavour (Lazo et al., 2017; Vintonyak et al., 2009).  1.3  EGF signalling and cancer The signalling pathways of the protein-tyrosine kinase epidermal growth factor receptor (EGFR/ErbB1/HER1) have been intensively studied for understanding growth factor signal transduction and regulation of gene expression to modulate cellular growth, survival, proliferation, and differentiation in mammalian cells in normal physiology (Oda et al., 2005). Although EGFR was originally recognized as a receptor for EGF, it can also bind many other mitogens including transforming growth factor-alpha, amphiregulin, epiregulin, heparin-binding EGF-like growth factor, beta-cellulin and glia growth factor. Other highly related receptors include ErbB2, ErbB3 and ErbB4. ErbB4 binds heregulin, whereas ErbB2 and ErbB3 do not appear to engage growth factors. The binding of cognate ligands to their respective ErbB family receptor induces the formation of homo receptor dimers and hetero receptor dimers with ErbB2 and ErbB3, leading to cross phosphorylation and the activation of their kinase catalytic domains (except for ErbB3, which is believed to be catalytically inactive). The resulting phosphorylated cytoplasmic domains also serve as docking sites for various proteins that feature SH2 and PTB domains. Through recruitment of adaptor proteins such as growth factor  11 receptor-bound protein 2 (Grb2) and Shc, the Ras>Raf>MEK>mitogen-activated protein kinase (MAPK) pathway becomes activated for induction of cell proliferation, while the phosphatidylinositol 3-kinase (PI3K)/Akt pathway is activated for cell survival through the recruitment of the PI3K p85 regulatory subunit (Normanno et al., 2006). The EGFR signalling network has become a well-recognized paradigm for the highly complex web of positive and negative feedback interactions that can regulate the propagation and the eventual termination of the signalling activity. An example of a few of the well-known proteins involved in pathways downstream of EGFR can be found in Figure 1.2.  Figure 1.2. Schematic of EGFR signalling pathways. Activation of EGFR leads to dimerization, phosphorylation of tyrosine residues, and recruitment of SH2 domain-containing proteins. Kinases are shown as ovals, transcription factors as purple rectangles, and protein biosynthesis-related proteins as green squares. Phosphorylation events are shown as grey arrows.   12 Early in vivo and in vitro studies have illuminated how EGFR hyperactivity and EGF-like ligands can induce cell transformation (Di Fiore et al., 1987; Velu et al., 1987). Overexpression of these cellular components is involved in many human carcinomas, and activating mutations are frequently observed in cancer patients (Normanno et al., 2006). Cancer therapies designed to mitigate aberrant EGFR activity are facilitated through the use of monoclonal antibodies and small-molecule tyrosine kinase inhibitors (TKIs). While anti-EGFR antibodies like cetuximab bind to the extracellular domain, hinder ligand attachment, and facilitate receptor internalization, TKIs like gefitinib work by inhibiting the phosphotransferase activity of the intracellular kinase domain (Seshacharyulu et al., 2012). Many prevalent gain-of-function mutations of EGFR have been characterized in cancer patients. A study involving crystallization of ErbB family dimers has shown that EGFR harbouring the L834R mutation stabilizes a configuration that is more suitable for accepting phosphorylation from the other dimer counterpart and is better primed for activation (Red Brewer et al., 2013). The T790M mutation, also known as the “gatekeeper mutation”, increases the mutant kinase’s affinity for ATP and decreases the inhibitory effects of TKIs like gefitinib that work as competitive inhibitors, hence facilitating drug resistance (Dagogo-Jack et al., 2017). Treatment of patients with resistant cancers harbouring L834R and T790M mutations in EGFR can be a challenge. Irreversible TKIs like osimertinib, which bind covalently to the Cys-797 residue, can be used as a second mode of therapy (Jiang et al., 2017). However, tumours can also be resistant to treatment with these new generation TKIs by evolving new C797S mutations, prompting a never-ending arms-race of EGFR inhibitors and function-rescuing mutations (Madic et al., 2018). This quagmire highlights the need for combination therapies that utilize multi-pronged approaches to overcome tumours that  13 harbour drug-resistance related mutations. Combination therapies involving EGFR-TKIs with radiotherapy, chemotherapy, and targeted therapy can be used for more effective treatment of cancers (Yang and Tam, 2018). Radiotherapy and cytotoxic chemotherapy treatments are classic general tumour-combating strategies (along with removal surgery), but targeted therapy is more specialized to single out proteins that also participate within the EGFR signalling network to work synergistically with EGFR-TKIs. For example, MEK and Raf inhibitors have been administered in conjunction with TKIs to treat EGFR T790M mutation-positive cancer patients (Liao et al., 2019, Tricker et al., 2015). Understanding the larger network connectivity of EGFR signalling can help support improved successful combination therapies, particularly for treatment resistant tumours with acquired mutations.   1.4  Insulin signalling and diabetes  Diabetes is a disease where insulin signalling is compromised in the cellular functioning of the body. The root of this could stem from either insufficient insulin production from the pancreas in regulating glucose metabolism, known as Type 1 diabetes, or if cells in the body stop responding effectively to it, which is categorized as Type 2. People living with diabetes are twice as likely to suffer from heart attacks and strokes, and the disease is associated with many other risks, including blindness, kidney failure and amputation of extremities (Bourne et al., 2013; Saran et al., 2015; The Emerging Risk Factors Coalition, 2010). Expanded comprehension of insulin signalling pathways could unveil new treatment opportunities and counteractive drug development.    14 The insulin receptor (INSR) and related insulin-like growth factor 1 receptor (IGF1R), which binds insulin with 500-fold lower affinity, are receptor-tyrosine kinases, but their mode of activation and signalling is slightly different from EGFR. These receptors are found normally in the cell plasma membrane as disulphide-linked homodimers with spatially separated transmembrane domains. Binding of the ligand stabilizes a new structural rearrangement that allows the transmembrane domains to interact closer with each other, ultimately leading to enzymatic activation of their intracellular kinase domains (Gutmann et al., 2018). Phosphorylation of the activation loop of the kinase domain as well as other tyrosine residues outside the kinase domain leads to binding of insulin receptor substrate family proteins. Unlike EGFR, which bind many signalling proteins directly, the insulin receptor substrate (IRS) proteins 1-6 form the main hubs where other signalling cascades are initiated, such as the PI3K and ERK-MAPK cascades (De Meyts, 2000, updated 2016). While these two cascades are also activated in EGFR signalling, the primary role of insulin signalling in the physiology of cells is to control metabolism and glucose homeostasis through regulation of glucose uptake, glycogen synthesis, and lipogenesis, among others. This is all mediated through phosphorylation of initiation and elongation factors, as well as ribosomal protein S6 leading to increased protein synthesis and functionality (Posner and Mounier, 2006; Radimerski et al., 2000). The divergent outcomes resulting from the similar signalling cascades activated by EGFR and INSR is a function of cross talk between other concurrent cascades, as well as the duration and the intensity of signal transduction. Understanding the finely tuned orchestration of the insulin signalling network will increase comprehension of Type 2 diabetes and permit exciting new strategies for the treatment of the hundreds of millions of people who are afflicted by this disease.   15 1.5  Control of protein synthesis by phosphorylation  Phosphorylation of transcription factors and mRNA translation machinery is one of the key mechanisms by which protein levels in the cell can be regulated through control of protein synthesis. For example, through the Ras>Raf>ERK MAPK pathway, activation of Elk1 transcription factor through phosphorylation of Ser-383 and Ser-389 by ERK1, and related isoform ERK2, promotes transcription of genes like c-fos, early growth response 1 (EGR1), and JunB, also known as immediate early genes (IEGs) (Gille et al., 1993, 1995; Gregg and Fraizer, 2011). It has also long been established that both the EGFR and INSR signalling networks activate the 90 kDa ribosomal S6 kinase (p90S6K), leading to enhanced protein synthesis (Nilsen-Hamilton et al., 1982; Novak-Hofer et al., 1988).   In INSR signalling, through the PI3K pathway, generation of phosphatidylinositol-3,4,5-trisphosphate (PIP3) recruits protein kinase B (PKB/Akt) and 3-phosphoinositide-dependent kinase 1 (PDK1) to the plasma membrane through binding to their pleckstrin homology (PH) domains (Cantley et al., 2006). Subsequent to phosphorylation by PDK1 at Thr-308, Akt then activates mammalian target of rapamycin (mTOR) though phosphorylation and inactivation of tuberous sclerosis proteins 1/2 (TSC1/2) (Hay and Sonnenberg, 2004). TSC1/2 in their dephosphorylated forms are inhibitors of Ras homology enriched in brain small GTPase (Rheb), and the newly uninhibited Rheb then activates the target of rapamycin complex 1 (TORC1) (Yang et al., 2006). This then leads to phosphorylation of eukaryotic translation initiation factor 4E-binding protein (4EBP1) and p70S6K (Hay and Sonnenberg, 2004). Activated Akt isoforms also phosphorylate forkhead box O1 (FOXO1), leading to its binding to 14-3-3 proteins and  16 sequestration within the cytosol. This obstructs FOXO1 from traveling to the nucleus, where it normally can promote transcription of gluconeogenesis genes, including phosphoenolpyruvate carboxykinase (PEPCK) (Tran et al., 2003).   Anisomycin, a bacterial antibiotic isolated from Streptomyces griseolus, is a ribotoxin that interferes with peptidyltransferase activity upon binding to 60S ribosomal subunits (Grollman, 1967). Cycloheximide is another compound that also inhibits protein synthesis at the ribosomal level by binding to the E-site, which competes with and hinders the binding of tRNAs (de Loubresse et al., 2014). In addition to being potent protein synthesis inhibitors, anisomycin and cycloheximide are also known as a potent activators of MAPK cascades such as c-Jun N-terminal kinase (JNK), p38 and ERK1/2, which can sensitize malignant cells for apoptosis and reduce tumours sizes (Macias-Silva et al., 2010; Sah et al., 2003).   1.6  Research strategies for tracking protein phosphorylation  Through the use of radioactive labelled adenosine triphosphate ([g-32P]ATP), Fischer et al. (1959) demonstrated the direct transfer of phosphate from ATP onto phosphorylase b. Since then, the methodology of tracking 32P isotope radioactivity has been regarded as the gold standard of kinase assays (Hastie et al., 2006). However, due to special radioactivity licensing requirements, complexity and inconveniences of handling radioisotopes and disposal issues, alternative techniques to assess the kinase reaction have since been developed. Many of these methods are focused on monitoring either the generation of adenosine diphosphate (ADP) or detection of the phosphopeptide/phosphoprotein, for example, with phosphosite-specific  17 antibodies. A popular replacement strategy known as ADP-Glow is used in a kit developed by ProMega. With this method, at the conclusion of the kinase reaction, the unused ATP substrate is hydrolyzed and the ADP product from the kinase reaction is converted back into ATP, which is then used by luciferase with luciferin to generate light that can be quantified (Zegzouti et al., 2009). Many other creative methods have been developed, such as converting ADP to nicotinamide adenine dinucleotide (NAD+) and detection of NAD+ fluorescently or by coupling a secondary reaction with ADP in which the beta-phosphate is released and observed with malachite green (Rojas et al., 2018; Wu, 2011). Peptide-focused methods to monitor phosphorylation involve the use of affinity reagents. Popular reagents like Pro-Q Diamond and pIMAGO reagent are proprietary nanoparticles that have affinity to the phosphate moiety and can detect many phosphoproteins and peptides (Agrawal and Thelen, 2005; Iliuk and Tao, 2015). Metal oxides, such as TiO2, are prevalently used in affinity columns to enrich phosphopeptides prior to their detection by MS (Chen and Hoehenwarter, 2019; Gates et al., 2010). Generic phosphotyrosine-specific antibodies are also often used to enrich samples prior to analysis on MS. However, the best application of antibodies is rooted by their affinity for specific epitopes. A gamut of research techniques like Western blotting, cell-based enzyme-linked immunosorbent assay (ELISA) , intracellular flow cytometry, immunohistochemistry, and antibody microarrays, all exploit the specific affinity of antibodies to localized epitopes on phosphoprotein targets, and are very powerful tools currently used in phosphoproteomics research (Hofman and Taylor, 2013; Kohl and Ascoli, 2017; Mahmood and Yang, 2012; Menon et al., 2014; Yue and Pelech, 2018).    18 1.7  Research antibodies to track protein expression and phosphorylation  The variety of antibodies typically used for research are usually monoclonal, polyclonal, or single-chain recombinant. Although monoclonal antibodies can excel with regards to specificity in their affinity for a singular epitope, monoclonal antibodies are at least 5-times more expensive and less humane to produce than polyclonal antibodies. The uniform specificity of monoclonal antibodies may also be undesirable in certain research applications due to minor changes in antigen structure across different model systems, or other factors that may affect protein conformation, including covalent post translational modifications (PTMs), pH, salt concentration, and fixation. These could all affect the ability of the antibody to reliably bind to its target. Polyclonal antibodies are less challenged in this regard, because they recognize multiple epitopes, but suffer from greater chances of cross reactivity with similar epitopes on other antigens. However, polyclonal antibodies developed against short peptides are likely to be much less cross-reactive than polyclonal antibodies raised against whole proteins. Since the avidity and stability of antibody-antigen complexes are determined by both the affinity of their binding as well as the valency of the binding, polyclonal antibodies with similar specificity to monoclonal antibodies are expected to have better antigen-capturing capacity due to the formation of more stable complexes over multiple epitopes. Lian et al. (2010) reported the superiority of polyclonal antibodies over monoclonal antibodies in capturing toxins on microarray slides, despite having similar affinities as determined by ELISA. Recombinant antibodies have the potential to be mutated to further improve specificity and potency. However, their extremely high costs limit their suitability primarily to therapeutic purposes.    19 1.8  Phosphoproteome analysis  While highly focused kinase-substrate relationship assays are still incredibly important to perform, there has been a much greater shift towards high throughput phosphoproteome analysis. Mass spectrometry (MS) has become a popular technique in this area of research, and MS datasets have been used extensively to populate many phosphoproteomic database websites. However, despite the sensitivity and technical marvel that is MS, this technique is subject to many limitations and difficulties. The prohibitive cost of MS equipment, which is also not user friendly, the requirement for efficient phosphopeptide enrichment, the lability of the phosphate in experimental environments, the difficulty in ideal protein fragmentation, and the difficulty in analyzing complex biofluids are a few of the challenges faced by MS studies (Iliuk et al., 2014; Solari et al., 2015; Zhvansky et al., 2019). Hsu et al. (2017) reported that high speed and accuracy mass spectrometers could only identify 20-40% of total phosphopeptides, were unable to detect low abundance proteins with low phosphorylation stoichiometry, and were additionally troubled by complicated analysis of the data due to additional modifications of the peptides. Tabb et al. (2010) also found that technical replicates overlapped only 35-60%, depending on the specific equipment used. Finally, enrichment of phosphopeptides for MS makes it very difficult to evaluate whether increased detection of a phosphopeptide reflects altered stoichiometry of phosphorylation and/or a change in the expression level of the protein from which the phosphopeptide was derived.  In light of these hurdles associated with MS phosphoproteomic research, antibody microarrays were selected as the platform of preference for phosphoproteome analysis in this thesis,  20 because of its targeted nature, flexibility of the platform in using crude cell lysates or biofluids, sensitive detection of labelled antigens with 10-100 µg crude cell or tissue lysates, and the convenience of using the same antibody probes for follow-up studies (Chen et al., 2018; Delfani et al., 2016). The measurement of a target protein on an antibody microarray requires about 100-times less antibody than required for a typical ELISA, resulting in a much lower cost per data point than other immuno-based assays (Schroder et al., 2010). One of the earliest forms of the antibody microarray technology was based on 400 antibodies that were spotted onto glass slides and were used to help classify immune cells by tracking the expression of allotypic cluster of differentiation (CD) and human leukocyte antigen (HLA) and other cell surface markers (Chang, 1983). Following this, work by Roger Ekins and colleagues fostered more appreciation of the platform. Coupled with new advancements in microarray printing technologies, the number of publications related to “antibody microarrays” steadily escalated until reaching stagnating numbers after 2008 (Ekins 1989; Ekins and Chu 1991; Ekins 1998; Yue and Pelech 2018). Current applications of antibody microarrays have featured densities of as much as 60,000 antibodies printed onto nitrocellulose covered slides for extremely high coverage proteomic analysis (Wang et al., 2020). However, challenges that researchers using the antibody microarray platform have faced are similar to those ailing the MS approaches. The sheer complexity of the proteome, with its roughly 21,300 genes and increased complexity of alternative splicing isoforms and post-translational modifications, amounts in distinct protein entities or proteoforms that are potentially in excess of 10 million, since about 95% of multi-exon genes have been estimated to undergo around 100,000 variable splicing events (Aebersold et al., 2015; Pan et al., 2008).   21 The selection of candidate phosphorylation sites for the development of antibodies has been primarily dominated by phosphosites that are well demonstrated to be functionally important in literature. Similar sites that are found in cousin proteins are popular candidates, as well as phosphorylation sites that have been repeatedly identified in MS studies. However, datasets from MS studies often reveal the same subset of frequently phosphorylated human proteins, like cytoskeletal and scaffolding proteins and metabolic pathway enzymes (Yue and Pelech, 2018). This proteomic bias, as well as the common use of anti-phosphotyrosine antibodies for enrichment in the MS process, leads to overrepresentation of identified phosphotyrosine sites of proteins that are ubiquitously expressed in high abundance in cells and tissues, and may not be selective biomarkers for diagnostic purposes. This is reflected in finding that out of the 2742 phosphosites found to be regulatory in literature in PhosphoSitePlus, 22% have not been reported by MS studies, while another 16% have only been reported once. While MS is an incredibly powerful tool, antibody microarrays provide a much more directed approach in tracking specific proteins and phosphorylation sites of interest.   Protein-protein interactions are another complication within the overwhelmingly complex nature of the human proteome. Most antibody microarrays capture lysate proteins in their native states in the presence of only mild detergents due to incompatible effects associated with stronger detergents, such as protein denaturation and precipitation. Dye labelled proteins in large complexes will result in signals that are less indicative of the target protein but also of their associated proteins, and it is also possible that many antibody epitopes could be shielded by other interacting proteins. The formation of phosphorylation-driven complexes such as  22 those involving SH2 and PTB domains, as well as 14-3-3 proteins, would be expected to render certain phosphorylation sites inaccessible for capture onto the array.   To improve the reliability of antibody microarrays for proteomic research, various challenges still need to be addressed. Improvements to the quality of data produced by antibody microarrays can be achieved by optimizing the preparation of lysate samples, capture and detection of substrate onto the antibody microarray, and validation of the analysis leads. One of the methods to reduce false positive results on antibody microarrays has been to fragment lysate proteins chemically at cysteine residues by successive treatment with Tris(2-carboxyethyl) phosphine hydrochloride (TCEP) and 2-nitro-5-thiocyanotabenzoic acid (NTCB) (Yue et al., 2017; Zhang et al., 2016). Fragmentation is beneficial in dissociating protein complexes and facilitates exposure of previously masked epitopes. The length of peptides produced by cysteine cleavage of the human proteome is predicted to vary but will mostly feature about 30-40 amino acid residues (based on the average length of predicted fragments from 21,000 human proteins that are cut at cysteine residues in silico). This allows the amount of signal recorded for individual antibody spots to be more reflective of the molar concentration of the target protein and less on the size of the full protein, which could vary by several orders of magnitude between proteins. Cysteine residues are often avoided in the epitope sequences used to generate antibodies, because they are usually used for coupling to keyhole limpet hemocyanin (KLH) for immunization and to resins for affinity purification. This enables the preservation of peptide fragments that retains the epitopes targeted by the antibodies, while producing lysate samples that are stable after their preparation due to the  23 destruction of kinases, phosphatases, proteases and other enzymes that could otherwise alter their structures. Furthermore, by fragmenting the proteins, competition of antibodies for different epitopes on the same target proteins can be avoided, which can produce higher signals and cross-validation of protein changes detected with the antibody microarray.  While traditional approaches for detection of captured analytes on antibody microarray have involved labelling the analytes with fluorescent dyes such as Cy3 and Cy5, biotin-labelling has grown in popularity as an alternative strategy (Delfani et al., 2016; Olah et al., 2015; Shi et al., 2011; Zhu et al., 2015). The detection of biotin-labelled peptide fragments with dye-labelled anti-biotin antibodies allows for greater non-specific binding than dye-labelled avidin, and stronger signals due to greater dye coupling to the larger immunoglobulin G protein (Yue et al., 2017). With these kinds of improvements in the antibody microarray technology, the evolved platform permits even greater potential in the screening of clinical biomarkers for diseases, drug targets, and for analysis of cell signalling networks.   1.9  Network analysis The type of cellular response that is produced by stimuli is finely orchestrated though a network of messengers, proteins, phosphorylation sites, and other processes. Cataloguing and characterizing the vast connectivity of cell signalling proteins has been a daunting task for researchers. Many proteomic studies have used MS as the primary method to generate large datasets to track protein expression and phosphorylation for specific cell responses, such as the EGFR signalling cascade (Pandey et al., 2000; Reddy et al., 2016). This approach can provide  24 detailed insight into the mechanisms of specific signalling cascades and how cells respond to particular types of stimuli. Other studies have used MS to analyze panels of different model systems to compare what these different models share in similarity, and what the molecular differences are (Kim et al., 2014; Lundby et al., 2012). This can reveal a proteomic fingerprint of what signalling connections are unique to each tissue phenotype, which could be helpful for looking at biomarkers for tissue-specific diseases. However, MS is not really practical in tracking specific phosphosites of interest in a variety of model systems, because of its random nature of detection, non-quantitative character, high equipment costs and the larger amounts of starting material that is required.   There are computational methods of predicting upstream and downstream kinase substrate specificities based on sequence motifs. Many programs have increased in popularity to make sense of large MS data sets and can generate lists of kinases predicted to be the most active based on the phosphopeptides detected. Some widely used programs include PhosphoPredict, Quokka, GPS5.0, and the Kinexus Kinase-Substrate Predictor featured in PhosphoNET (Li et al., 2018; Safaei et al., 2011a; Song et al., 2017, Wang et al., 2020). Other programs have even been integrated as free packages for the popular open-sourced statistics platform “R”, such as KinSwingR (Engholm-Keller et al., 2019). While these have been instrumental in helping researchers to make sense of large data sets and evaluating potential clinical biomarkers and drug targets, the strength of these algorithms depends on the quality of empirical kinase-substrate relationships used in their training. More targeted research on kinase-substrate  25 epitope specificities will enhance the capacity of these programs to model signalling networks more accurately.   1.10  Research aims The central hypothesis of my thesis research was that if high content antibody microarrays can track hundreds of regulatory proteins simultaneously, then it may be feasible to elucidate the composition and architecture of signalling networks in an unbiased manner. Subsets of proteins that become altered in their expression and/or phosphorylation are likely to be interconnected within local networks. Examination of these linkages could validate existing notions of signal transduction networks and uncover novel protein interactions that have evaded discovery so far. The wide-spread application of antibody microarrays has been hampered by several issues such as protein-protein interactions that can mask capture epitopes or lead to co-capture and competition of the same targets by antibodies that target different epitopes, that have questioned the reliability of this powerful technology. However, the untapped potential of the antibody microarray lies in its targeted nature for specific proteins and epitopes as opposed to the random detection associated with MS.  In this thesis, I describe the experiments that I have performed to optimize the antibody microarray platform as a more reliable research tool. Firstly, in Chapter 3, I have explored the specificity of approximately 600 phosphosite-specific antibodies, many of which I have helped to purify and characterize, to get a better sense of the limitations of their applications. Secondly, in Chapter 4, I have described the studies that I carried out to develop improvements  26 in cell and tissue lysate preparation, and optimization of the detection of peptides captured on antibody microarrays from these biospecimens. This work was primarily undertaken in the A431 human epidermoid carcinoma cell line, in which the EGF receptor is over-expressed. I also developed a sandwich antibody microarray to track how tyrosine phosphorylation is generally affected in target proteins adsorbed on the microarray with capture antibodies. As a reporter antibody, I used a generic phosphotyrosine-specific antibody, which was developed in house at Kinexus, to detect tyrosine-phosphorylated proteins (Lai, 2015). This type of microarray could be adapted to monitor these proteins for other types of covalent modifications such as ubiquitination, and for protein-protein interactions.   As described more fully in Chapter 5, I applied the optimized antibody microarray to explore the signalling systems that are utilized by EGF and by insulin in several other different human cancer cell lines to see how they were differentially affected in accordance with cell type. This also included the A431, A549, DU145, HeLa, and LNCaP, cell lines for EGF treatments and the HeLa, HepG2, Jurkat, and MCF7 cell lines for responses to insulin. As these are well studied growth factors, this afforded the opportunity to both validate the utility of the microarray and the means to identify other novel signalling proteins that are regulated by these important hormones. To verify the key changes flagged by these antibody microarray experiments, I also performed extensive Western blotting validation studies with the lead antibodies.  EGF and insulin are agents that bind to and activate receptor-tyrosine kinases and are known to increase protein-tyrosine phosphorylation inside cells. As presented in Chapter 6, I examined  27 how general inhibitors of protein-tyrosine phosphatases that induce even greater hyper-tyrosine phosphorylation of proteins would affect downstream protein-serine/threonine phosphorylation. This was of interest, since the cell would be expected to have counter-responses to deal with exessive tyrosine phosphorylation of cellular proteins, as typically occurs with neoplastic transformation in cancer cells.  As described in Chapter 7, I also explored how the phosphorylation of proteins was affected when protein synthesis or protein degradation was inhibited. With a reduction of protein synthesis induced with agents such as anisomycin and cycloheximide, I hypothesized that the remaining protein may undergo altered phosphorylation to retain their needed levels of functionality. It was interesting to identify these phosphoproteins, since they may have direct relevance in the control of protein synthesis as well as other anabolic pathways.  As presented in Chapter 8, I combined the data from over 250 datasets generated with the same antibody microarray across many different model systems that involved a diverse multitude of treatments to establish the most reliable antibody pairs for future antibody microarray development, and to verify known and uncover novel kinase-substrate connections. In particular, this included over 26 different treatments of the well-studied HeLa cervical carcinoma cell line to identify subsets of signalling proteins that showed a high degree of correlation across these treatments, which could be indicative of co-regulation within common signalling pathways. Using correlative analysis, I sought to determine the protein kinases that were the most associated with specific phosphosites on other targets in response to a plethora  28 of treatments in different model systems. Next, I compared and validated the correlated leads with known kinase-substrate interactions from the literature or those that were predicted from algorithms developed in our laboratory and described previously (Safaei et al., 2010, 2011a). I also performed in vitro phosphorylation experiments with purified recombinant active preparations of protein kinases with ATP to phosphorylate fragmented cell lysate proteins that were subsequently captured on the antibody microarray to identify kinase-substrate connections implicated from the in vivo studies with the same microarrays.   In the final Chapter of my thesis, I have summarized the key conclusions from my Ph.D. studies and propose future work that can build on my findings.     29 Chapter 2: Materials and Methods  2.1  Modified Bradford protein assay  Bovine serum albumin (BSA, Sigma-Aldrich, Missouri, MO) protein standards were made up to 50 µL at 0.1, 0.2, 0.3, 0.5, and 0.7 mg/mL in phosphate-buffered saline (PBS [135 mM NaCl, 1.8 mM KH2PO4, 10 mM Na2HPO4, 2.7 mM KCl, pH 7.4]), including 10 µL of lysis buffer (or equivalent that the unknown sample is known to be in). Typically, 10 µL of the unknown sample would be diluted to 50 µL, in duplicate. Each sample was added to 2.5 mL of Bradford’s Reagent (10% [v/v] phosphoric acid, 0.01% [g/v] Coomassie Brilliant Blue G250, 5% [v/v] ethanol), mixed thoroughly, and left for 10 min in the dark. Each solution was re-vortexed and transferred to a cuvette for measurement at 595 nm in a Spectronic Helios Gamma UV-Vis Spectrophotometer (Thermo Fisher) (Bradford, 1976). If the unknown sample reading was not within the range of the absorbance readings generated by the 0.1-0.7 mg/mL protein standards, a new dilution of the unknown sample was made and assayed. The protein standard readings were plotted in Microsoft Excel with a best line of fit, and the protein concentration of the unknown sample was extrapolated from the line.   2.2  Antibody production, purification, and characterization Antigen phospho- and nonphosphopeptides were synthesized using the Fmoc solid-phase synthesis strategy as described in Fields and Noble (1990) on a Multipep peptide synthesizer and purified by reverse phase high-pressure liquid chromatography (HPLC). The purity and identity of the synthesized peptides were validated by analytical HPLC as well as by mass  30 spectrometry for a final purity between 70-98%. Each peptide was conjugated to keyhole limpet hemocyanin (KLH, Sigma-Aldrich) and was subcutaneously injected into New Zealand White rabbits every month for 4 months (Pacific Immunology Corp., Ramona, CA, USA). Any sera from rabbits immunized with a phosphopeptide antigen was subjected to an additional negative purification with agarose columns to which phosphotyrosine-beta-alanine-cysteine was coupled through a thio-ester linkage prior to affinity purification with agarose coupled with the immunizing antigen peptide. Captured antibodies were eluted from the columns with 0.1 M glycine (pH 2.5) and collected in 0.5 mL fractions in ice-cooled tubes and were neutralized to pH 7.0 with 1 M Tris buffer pH 8.8. Peak fractions were identified by the modified Bradford Protein Assay procedure as indicated, then pooled and concentrated with Centricon plus-70 centrifugal filters (Millipore).   2.3  Western blotting and variants Protein lysate samples were subjected to standard Western blotting where different lysate samples were deposited into separate lanes (~20 µg protein per lane), resolved by multi-lane SDS-PAGE, and probed with a primary antibody after their electrotransfer on to a nitrocellulose membrane. Membranes were later rinsed with Tween Tris-buffered saline pH 7.4 (20 mM Tris base, pH of 7.4, 500 mM NaCl, 0.05% Tween 20), and then incubated with horseradish peroxidase conjugated secondary anti-rabbit IgG antibody for 30 min at room temperature. The immunoblots were developed with enhanced chemiluminescence (ECL) Plus reagent (Amersham, Arlington Heights, IL), and signals were captured by a Fluor-S MultiImager and quantified using Quantity One software (Bio-Rad, Hercules, CA). All of the pan- and  31 phosphosite-specific rabbit polyclonal antibodies used for immunoblotting were developed in-house at Kinexus Bioinformatics and are commercially available from the company and described at www.kinexusproduct.ca.  2.3.1  Peptide dot blots Purified peptides were hand spotted onto nitrocellulose membrane (approx. 10 ng/spot) and probed using the Western blotting protocol.   2.3.2  SPOT membranes Peptide macroarrays were produced by direct peptide synthesis onto cellulose membranes by the SPOT technique (Hilpert et al., 2007). The yield of peptide per spot was estimated at 500 to 1000 ng/spot. SPOT membranes were blocked with Tris-buffer saline-Tween with 5% sucrose and 4% BSA. Primary antibody probing of SPOT membranes were incubated at 0.5 µg/mL antibody in blocking buffer overnight at 4°C, followed with 30 min at room temperature with 20 ng/mL horseradish peroxidase–coupled anti-rabbit secondary antibody in blocking buffer. Detection was carried out by ECL incubation for 15 sec and scanning at every 15 sec on a Fluor-S Max Multi-Imager (Bio-Rad, Hercules, CA).   2.4  Kinase substrate profiling assays For the phosphotransferase assays, 100 µg of crude cell lysate were incubated in 0.4 mM Tris(2-carboxyethyl)phosphine hydrochloride (TCEP) at pH 9 in room temperature for 10 min, followed with incubation for 30 min with 6 mM 2-nitro-5-thiocyanatobenzoic acid (NTCB) at  32 37°C for protein fragmentation. The cleaved lysates were then passed through a Sephadex G-25 spin column for buffer exchange with PBS and incubated with 0.5 µg of purified active recombinant human protein kinases (SignalChem, Richmond, B.C., Canada) as described in the recommended protocol supplied for each kinase for 30 min at 37° C. Following kinase incubation, the samples were then biotinylated and processed as described in the antibody microarray protocol.   2.5  Cell culture  Human A431 skin epithelial carcinoma cells (American Type Culture Collection, Manassas, VA; tested for, and found free of, Mycoplasma contamination) were grown in 100 x 20 mm tissue culture dishes (Sarstedt) in Dulbecco’s Modified Eagle Medium (DMEM) (all medias were sourced from ThermoFisher, Rockford, IL) supplemented with 10% fetal bovine serum (FBS) (ThermoFisher) up to ~70% confluence. Human A549 lung epithelial carcinoma cells were grown in similar fashion in Ham's F12 Nurtient Mix Media, human HeLa cervical carcinoma cells were grown in Minimum Essential Media (MEM), while human HepG2 liver carcinoma cells and human cancer MCF7 breast cancer epithelial cells were grown in MEM supplemented with 10 µg/mL porcine insulin (Sigma-Aldrich) in addition to FBS. Human T lymphocyte Jurkat cells were grown in Rosewell Park Memorial Institute (RMPI) 1640 Medium with FBS.   2.5.1  Cell treatment and conditions Please see Table 1 for a list of treatment conditions of concentration and duration. Cells cultures that were ~ 70% confluent were either treated directly with reagents diluted in FBS- 33 supplemented media or were changed to serum-free media for 16-18 h and then incubated with treatments in serum-free media.  Table 2.1. Treatment concentrations, durations, in serum-free or supplemented media.  Treatment Treatment Concentration Treatment Duration Serum-free / Supplemented Anisomycin 10 µg/mL 120 min Supplemented Calyculin A 25 ng/mL 15 min Supplemented Cycloheximide 10 µg/mL 120 min Supplemented Epidermal growth factor (EGF) 100 ng/mL 15 min Serum-free Forskolin 20 µM 15 min Supplemented H202 200 nM 120 min Supplemented Insulin (higher dose) 1 µg/mL 15 min Serum-free Insulin (lower dose) 100 ng/mL 15 min Serum-free MG132 50 µM 120 min Supplemented Nocodazole 200 ng/mL 24 h Supplemented Oleic acid 1 mM 15 min Supplemented Phenylarsine oxide (PAO) + sodium orthovanadate (shorter duration) 25 µg/mL + 50 µg/mL 15 min Supplemented Phenylarsine oxide (PAO) + sodium orthovanadate (longer duration) 25 µg/mL + 50 µg/mL 30 min Supplemented PMA phorbol ester 20 ng/mL 15 min Supplemented Rapamycin 250 nM 15 min Serum-free Rapamycin + insulin 250 nM + 1 µg/mL 15 min Serum-free Serum starvation (long term) N/A  4 h in new media Serum-free Serum starvation (short term) N/A 4 h in new media Supplemented, then serum-free Serum supplementation (long term) 10% FBS 4 h in new media Supplemented Serum supplementation (short term) 10% FBS 4 h in new media Serum-free, then Supplemented Sodium arsenite 20 µM 120 min Supplemented Sodium nitroprusside 20 µM 15 min Supplemented Staurosporine (long term) 0.5 µg/mL 24 h Supplemented Staurosporine (short term) 0.5 µg/mL 15 min Supplemented UV exposure 25 mJ/cm2 Culture for additional 2 days after exposure Supplemented   34 2.5.2  Cell lysate preparation for Western blotting The cells were treated as listed in Table 1 and washed with ice-cold PBS. After removal of the PBS by aspiration, the cells were scraped from the dishes with a rubber policeman into 100 µL of SDS Homogenizing Buffer (25 mM Tris-HCl, pH 6.8, 10% glycerol, 1% beta-mercaptoethanol; 0.8% sodium dodecylsulphate, 0.016% bromophenol blue).  2.5.3  Cell lysate preparation for microarray without fragmentation Cells were treated as indicated and scraped with 100 µL of ice-cold Standard Homogenizing Buffer (SHB) [1% Triton X-100, 5 mM EDTA, 2 mM EGTA, 20 mM MOPS, 25 mM NaF, 25 mM Na4P2O7, 1 mM Na3VO4, 60 mM beta-glycerophosphate, 50 nM phenylarsine oxide, 1 mM Pefabloc, 3 mM benzamidine, 10 µM leupeptin, 0.5 µM aprotinin, 1 mM dithiothreitol, 100 mM NaCl]. These buffer components were sourced from Sigma-Aldrich. The cell samples were subjected to sonication for 40 sec in intervals of 10 sec with 10 sec intermissions over ice and were then immediately centrifuged for 30 min at 4°C at 50,000 rpm in a Sorvall Discovery M120 Ultracentrifuge in a S120AT2-0278 titanium rotor. The resulting supernatant fractions were stored at -80°C until further use.  2.5.4  Cell lysate preparation for microarray with fragmentation Cells were treated as above and scraped with 100 µL of ice-cold Cysteine chemical cleavage buffer (CCCB) [SHB that also included 0.4 mM TCEP at pH 9]. After sonication as described above, the homogenates were incubated for 30 min with 6 mM NTCB (Cat. N7009, Sigma-Aldrich) at 37°C and then subjected to ultracentrifugation for 30 min at 20°C at 50,000 rpm. The  35 pH of the TCEP prepared supernatants was neutralized to pH 7.0, and these preparations were stable at room temperature for at least a week but were kept at -20°C for longer term storage.  2.6  Kinex™ antibody microarrays The three antibody microarrays used in this thesis were all commercially available products from Kinexus Bioinformatics Corporation. A full listing of the antibodies printed onto the Kinex™ KAM-900P, KAM-1150, and KAM-1325 arrays can be downloaded using the following links: http://www.kinexus.ca/pdf/KAM900P-Kit_Product_Information.pdf, http://www.kinexus.ca/pdf/KAM1150-Service_Information.pdf, and http://www.kinexus.ca/pdf/KAM-1325-Service_Information.pdf .   2.6.1  Sample labelling Approximately 50 µg of crude cell lysate were incubated with either 40 µg of AlexaFluor546/Cy3 (1:1) mixture or 20 µg of NHS-Biotin (APExBIO) and adjusted to pH 8.5 using 1 M NaHCO3 for 1 h. Subsequently, labelled lysate proteins were purified using Sephadex G-25 Spin columns (Sigma-Aldrich) and collected. The lysate sample was then diluted to 400 µL with PBS with 0.05% Tween and 50 mM ethanolamine for incubation with the antibody microarray.   2.6.2  Reporter antibody preparation Anti-biotin antibody (Sigma-Aldrich, B3640) that was dissolved to 1 mg/mL in 0.135 M NaCl or generic anti-phosphotyrosine PYK antibody (Kinexus, PG001) was adjusted to pH 8.5 using 1 M NaHCO3, coupled to 400 µg of Alexa546/Cy3 mixture for an hour, and passaged down a 10 mL  36 Sephadex G-100 (Sigma-Aldrich) column. The initial wash through was collected in 0.5 mL fractions and the protein concentration was subsequently determined with the modified Bradford protein assay. The peaks containing the bulk of the labelled antibody was dialyzed with Spectra/Por tubing of 6-8,000 molecular weight cut-off overnight in PBS containing 50 mM ethanolamine. The dialyzed antibody was then assayed again for protein concentration and was diluted to 0.1 µg/mL and frozen in 1 µg aliquots.   2.6.3  Probing and imaging The labeled protein lysate samples and labeled detection antibody were both diluted in PBS with 0.05% Tween-20 and 50 mM ethanolamine (Cat. E9508, Sigma-Aldrich). The microarray was blocked with 25 mM ethanolamine in pH 8.5 and 100 mM sodium borate for 1 h at room temperature, and then washed in three baths of PBS with 0.05% Tween-20 with a final wash in distilled water. The microarray slide was then placed onto a dual well FAST slide incubation chamber (Cat. Z721336, Sigma-Aldrich) with a FAST frame slide holder (Cat. Z721212, Sigma-Aldrich) and incubated with two diluted cell lysate protein samples (400 µL of each) in a humidity chamber for 2 h. For dye-labeled protein samples, the slides were washed and dried after incubation and scanned at 543 nm with a ScanArray GX Microarray scanner (Perkin-Elmer, Wellesley, MA). For biotin-labelled or non-labelled protein samples, the slides were washed, dried, secured into the FAST frame with a single well incubation chamber, and incubated with 1 µg of dye-labeled reporter antibody (anti-biotin antibody if lysates were biotinylated) in 800 µL over the entire slide for 10 min in the humidity chamber. After incubation, the slides were washed, dried under nitrogen, and scanned at 543 nm.   37 2.6.4  Microarray slide signal quantification  The fluorescent signals from the scanned microarrays that were captured in a TIFF file image were quantified with ImaGene 9.0 microarray image analysis software (BioDiscovery, El Segundo, CA). The signal strength of each spot was calculated with a function of the net median signal and spot size and was normalized according to the sum of all of the recorded signals per field. The values for duplicate measurements were averaged. In the case of at least 3 replicate experiments, two-tailed homoscedastic Student T tests were performed to evaluate the statistical significance of changes in protein expression and phosphorylation between the control and treatment samples.   2.6.5  Metadata statistical analysis All analyses were done using Microsoft Excel and statistical computing software R (R Core Team 2019). Microsoft Excel datasets containing the percentage change from control (%CFC) for every antibody for every treatment condition were pooled together and imported into R. Spearman’s correlations were done using the rcorr function within the Hmisc package which provided Spearman’s correlation coefficients and p values (Harrell, 2019). The p values were then analyzed with the Qvalue package for q values and exported back into Microsoft Excel (Storey et al., 2019).    38 Chapter 3: Characterization of Phosphosite-specific Antibodies  3.1  Rationale The effective use of antibody microarrays in tracking proteins and other biomolecules, as with any other research technique that are dependent on the use of antibodies, is contingent on the epitope specificity that the antibodies have affinity for. Many antibodies sold commercially are poorly characterized, and most only feature Western blotting images with a detected band close to the predicted size of the target protein after SDS-polyacrylamide gel electrophoresis as evidence of the performance of the antibody. Monoclonal antibodies feature a singular population of antibody and possesses affinity for theoretically one epitope, but they are often produced through immunization of large protein fragments or entire proteins into mice and are also not generally characterized for their epitope specificities. Through my own research, I have been misguided by antibodies that have cross-reacted with a protein of the approximately predicted size of my intended target by Western blotting, only to realize later through the use of other antibodies for the same target, that the original antibody was tracking a separate protein. The need for proper characterization of antibodies is critical for the reliable use of antibodies in research, and for reproducible and accurate results that support the same scientific outcomes among the research community. In this chapter, I summarize the screening of over approximately 600 phosphosite-specific antibodies for their performance on membranes with peptides that were curated after their immunizing sequences as well as related sequences featuring alterations to gain insight on their actual epitope  39 immunoreactivity. My findings provide a general sense of phosphosite-specific antibody potency and specificity, and this work may guide the development of more successful antibodies for target phosphosites in future studies.  3.2  Consensus sequence analysis using peptide dot blots  At the time that I commenced this study of phosphosite-specific antibody potency and specificity, there were about 400 phosphopeptides then available that were originally used to elicit phosphosite-specific antibody production in rabbits. This afforded the opportunity to use dot blotting with nitrocellulose membranes that were hand spotted with phosphopeptides that featured highly similar amino acid sequences to the original immunogenic peptides used to immunize rabbits and affinity purify antibodies from serum.  This permitted limited evaluation of the consensus immunoreactivity epitopes for individual antibodies.  The dot blots featured twelve spots, with approximately 10 ng of peptide per spot, and were carefully selected based on similarity for every antibody tested. The first phosphopeptide spotted corresponded to the original phosphopeptide that was utilized to produce the phosphosite-specific antibody tested on each dot blot. To reduce possible ambiguities, antibodies that targeted only a singular phosphosite were considered. Based on the immunoreactivity of the various peptide spots on each membrane, a tentative consensus sequence of the key amino acid residues that form the immunoreactive epitope was deduced from three amino acid residues on both the N- and C-terminal sides of the phosphorylated residue. Positional residues that could not be deduced were marked as “Any” or an “X”.  40 Deduced consensus sequences that were not specific to a single phosphoacceptor were marked as “p-Any”. This analysis was performed for a total of 412 phosphosite-specific antibodies, with 127 targeting phosphoserine sites, 83 for phosphothreonine sites, and 202 for phosphotyrosine sites. The frequency of amino acid residues occurring in the deduced consensus sequence for antibody immunoreactivity were tabulated and expressed as a percent (Table 3.1). From this analysis, it was observed that amino acids in the -2 position were the most likely to have a defined residue as part of the consensus sequence, therefore indicating that the amino acids typically found in this position relative to phosphosites are generally important in epitope recognition. This observation was common to phosphoserine-, phosphothreonine-, as well as phosphotyrosine site-specific antibodies. The second position to be most likely to have a defined amino acid as part of the consensus recognition sequence was the -3 position for phosphoserine and phosphothreonine site-specific antibodies, and the -1 position for phosphotyrosine site-specific antibodies. Based on the rotational turn of the side groups of consecutive amino acids in a chain, the side chain of the -2 amino acid residue is likely to be closer directionally to the phosphosite. This improves the possibility of interacting with an antibody and may explain why this position is more important in antibody binding over the -1 and -3 positions.    41 Table 3.1. Frequency of amino acids identified positionally to a central phosphosite as part of a consensus sequence for phosphoserine, phosphothreonine and phosphotyrosine site-specific antibody immunoreactivity.    Frequencies are expressed as percentage. Consensus sequences were deduced by unique peptide dot blots featuring the original immunizing peptide as well as eleven similarly sequenced peptides for every antibody tested. This included 127 phosphoserine (pS) site-specific antibodies, 83 phosphothreonine (pT) site-specific antibodies, and 202 phosphotyrosine (pY) site-specific antibodies.  42 Overall, the data indicated that the N-terminal amino acids were more important for consensus epitope formation among the phosphoserine, phosphothreonine, and phosphotyrosine site-specific antibodies tested. The next question that I considered was which of the 20 possible amino acid residues appeared to be the most favoured for antibody recognition. The frequency of every amino acid residue occurring in the original immunizing peptides in any position was compiled, and this was expressed as a percentage of all of the possible residues. Since a limited pool of around 400 phosphopeptides was used to create the dot blots, there was already a pre-existing bias for the frequency of each of the 20 amino acid residues amongst the immunizing peptides. This inherent expected frequency of each of the amino acid residues was also calculated (Table 3.2). After comparison of the frequencies of amino acids occurring in the deduced consensus sequences with the expected frequencies, it was apparent that arginine, lysine, isoleucine, valine, proline and to a lesser extent phenylalanine were among the amino acid residues that were more frequently incorporated into consensus sequences for antibody reactivity, while glutamine, asparagine, and histidine were among the least frequent. These differences from the expected frequencies were relatively small, and a caveat for this type of analyses is the limitations on the number of phosphopeptides that were available to derive consensus recognition sequences for each antibody. Nevertheless, the importance of hydrophobic residues and arginine, but not lysine was an interesting observation of this meta-analysis.     43 Table 3.2. Frequency of amino acids identified as part of a consensus sequence for antibody immunoreactivity compared to original frequency in all immunizing peptides.   Amino Acid Position -3 -2 -1 +1 +2 +3 -3 to +3 Residue occurrence in Immunizing Epitopes Frequency of amino acid found in consensus subtracted by predicted occurrence  A 1.7 7.5 2.4 3.2 8.3 2.7 6.3 6.6 -0.4 C 0.2 0.2 0.7 0.0 5.6 0.5 1.8 1.9 -0.1 D 5.6 8.0 7.0 4.1 2.9 2.9 7.4 7.0 0.4 E 7.8 2.2 4.1 6.3 5.1 1.5 6.5 6.9 -0.3 F 0.5 0.7 3.6 8.0 1.5 0.7 3.7 3.4 0.2 G 4.6 6.8 2.2 0.7 2.4 11.9 7.0 7.0 0.0 H 1.7 0.7 1.7 1.0 0.5 0.7 1.5 2.2 -0.7 I 1.7 0.5 5.3 4.4 4.1 2.9 4.6 3.7 0.9 K 7.8 3.9 2.9 2.2 3.2 3.2 5.6 5.9 -0.3 L 4.9 8.0 10.4 5.6 3.2 4.4 8.8 7.4 1.4 M 1.0 0.7 1.5 1.0 1.0 1.2 1.5 2.0 -0.5 N 4.6 0.5 2.7 0.5 0.7 1.0 2.4 3.4 -1.0 P 1.7 11.2 4.1 14.1 4.4 5.8 10.0 9.3 0.8 Q 0.7 1.0 1.7 1.0 2.9 1.7 2.2 3.6 -1.4 R 14.3 9.0 2.2 2.2 3.4 3.4 8.4 6.7 1.6 S 2.9 5.3 3.6 2.9 5.1 4.6 6.0 6.3 -0.3 T 2.4 5.1 3.9 1.9 1.9 3.4 4.5 5.5 -1.0 V 2.7 4.9 5.6 7.0 6.3 7.5 8.3 7.4 0.9 W 0.2 0.0 0.2 0.7 0.0 0.2 0.4 0.5 -0.2 Y 1.7 1.2 4.4 2.9 1.2 1.7 3.2 3.3 -0.1  Frequencies are expressed as percentages. Consensus sequences were deduced by unique peptide dot blots featuring the original immunizing peptide as well as eleven similarly sequenced peptides for every antibody tested. This analysis was performed for 412 phosphosite-specific antibodies.   44 3.3  Characterization of immunoreactivity of amino acid residues N- and C-terminal to a phospho-acceptor amino acid residue with SPOT membranes To increase confidence in my findings of the relative importance of particular amino acid residue positions in the formation of an immunoreactive epitope, I advanced to the use of SPOT membranes in which peptide analogues were systematically synthesized as separate spots in small arrays. The amount if peptide present in a spot on the SPOT cellulose membranes can be up to 500-fold higher than available in peptide-printed nitrocellulose membranes. These arrays featured the original immunizing phosphopeptide sequences as well as analogues that contained a single glycine substitution from the -3 to +3 positioning residues flanking the phospho-accepting amino acid. The rationale was that substitution of an important amino acid residue for antibody recognition would markedly reduce antibody binding and signal detection. A total of 155 single-phosphosite-specific antibodies were tested with individually tailored SPOT membranes. A representative figure of 75 of these immunoblots is shown in Figure 3.1. The signal intensity of every peptide spot was quantified manually on a scale of 0-5, with 5 representing maximal signal and 0 indicating no signal (Figure 3.2).   45  Figure 3.1. Representative SPOT peptide membranes featuring positional glycine substitutions probed with phosphosite-specific antibodies. The target phosphosite name, Kinexus antibody codes, as well as the immunizing peptide sequence for each antibody are listed. The full list included 60 phosphoserine site-specific antibodies, 30 phosphothreonine site-specific antibodies, and 65 phosphotyrosine site-specific antibodies, but only 25 from each category are shown here. WT corresponds to the original wild-type sequence.  46  Figure 3.2. Average signal intensity of phosphoserine, phosphothreonine, and phosphotyrosine site-specific antibodies on peptides featuring singular glycine substitutions from three residues N-terminal to three residues C-terminal to the phospho-accepting amino acid residue. Error bars represent standard deviation in the average signal intensities. For this analysis, 60 phosphoserine site-specific antibodies, 30 phosphothreonine site-specific antibodies, and 65 phosphotyrosine site-specific antibodies were tested. Brackets indicate statistically significant differences in average signal intensities between positions at p ≤ 0.05.   An analysis of variance (ANOVA) on the average signal strengths recorded per flanking position yielded significant variation (p ≤ 0.05) among the different positions for all three phosphosite-specific antibody types. A post hoc Tukey-Kramer test was performed to determine positional pairs that differ significantly at p ≤ 0.05 and these are indicated in square brackets in Figure 3.2. This analysis indicated that a glycine substitution in the -1 amino acid position was apparently the most disruptive for antibody recognition among phosphotyrosine antibodies. The -1 position also trended to be the most disruptive for recognition by phosphoserine and phosphothreonine site-specific antibodies, but these differences were not large enough to be statistically significantly different from some other postions. Taken together, these findings  47 from the glycine substitution experiment have some overlapping agreement with the consensus sequence analysis with the phosphopeptide dot blots, but the order of the importance of N-positional residues for antibody recognition could not be definitively ranked.   Next, I also focused on the amino acid residues that resulted in the greatest loss of signal intensity when substituted with glycine as well as those that were not affected by glycine substitution. This was done to determine if the specific amino acids that are generally more important in immuno-epitope formation and those that are typically much less critical could be identified. To that end, the frequency of amino acids that yielded signal intensities of 0 and 1 when substituted with glycine in any position, as well as those that yielded intensities of 4 and 5 due to substitution were compiled and expressed as a percent over the frequency of these amino acids occurring in these two groups (Table 3.3). From this analysis, the amino acid residues that frequently resulted in low signal intensity when substituted with glycine, and most likely to be critical for antibody recognition in general, were phenylalanine, isoleucine, threonine, leucine, glutamine, and arginine. The least critical amino acids were cysteine, alanine, asparagine, serine, and methionine. These results, while slightly different with respect to the importance of glutamine and threonine, tended to agree with the observations arising from the consensus sequence analyses for favoured amino acid residues shown in Table 3.2. In short, hydrophobic amino acid residues and arginine were favoured for phosphosite-specific antibody epitope recognition by both approaches.     48 Table 3.3. Number and percent frequency of critical and non-critical amino acid residues Amino Acid A C D E F G H I K L M N P Q R S T V W Y 0 + 1 Signal Intensities 8 1 14 21 17 1 4 17 9 28 4 9 33 14 31 10 13 21 1 14 % 25 10 33 47 81 3 50 65 31 61 27 26 42 61 58 26 62 51 33 52 4 + 5 Signal Intensities 24 9 28 24 4 36 4 9 20 18 11 26 45 9 22 29 8 20 2 13 % 75 90 67 53 19 97 50 35 69 39 73 74 58 39 42 74 38 49 67 48  Frequencies of amino acids that yielded signal intensities of 0 and 1, or 4 and 5 when substituted with glycine on SPOT peptide membranes probed with phosphorylation site-specific antibodies were tabulated from 155 experiments.   3.4 Antibody specificity for phosphoserine, phosphothreonine and phosphotyrosine residues  As evidenced in the consensus sequence analysis, there were some cross-reactivity of phosphosite-specific antibodies with peptides featuring alternative phosphorylated amino acids. To determine the extent and likelihood of this cross-reactive phenomenon, peptide arrays that featured the immunizing phosphopeptide sequence as well as substitutions with other phosphorylated amino acids were produced by SPOT synthesis and were tested with phosphosite-specific antibodies. The phospho-accepting amino acid residue within each peptide was replaced by either phosphoserine, phosphothreonine, phosphotyrosine, unphosphorylated serine, threonine, tyrosine residues, or glutamic acid (as a phospho-mimetic). The signal intensity for each peptide spot were quantified on a scale of 0-5, with 5 representing the greatest signal strength and 0 for no signal. A total of 456 antibodies were tested, consisting of 155 phosphoserine site-specific antibodies, 99 phosphothreonine site- 49 specific antibodies, and 202 phosphotyrosine site-specific antibodies. Figure 3.3 provides a representative figure with SPOT blot images of 75 tested phosphosite-specific antibodies.  Figure 3.3. Representative SPOT peptide membranes featuring phosphoamino acid substitutions probed with phosphosite-specific antibodies. The central phosphoamino acid residue within each peptide was replaced by either phosphoserine (pS), phosphothreonine (pT), phosphotyrosine (pY), unphosphorylated serine (S), threonine (T), tyrosine (Y) residues, or glutamic acid (E).   50 An ANOVA on the average signal strengths recorded per amino acid subtitution yielded significant variation (p ≤ 0.05) among the different substitutions for all three phosphosite-specific antibody types. A post hoc Tukey-Kramer test was performed to determine significantly differing amino acid substitutions at p ≤ 0.05 and significantly different pairs of substitutions are indicated in square brackets in Figure 3.4. Analysis of these SPOT membranes revealed that phosphoserine site-specific antibodies are likely to cross react to a lesser degree with peptides featuring phosphothreonine substitutions, as well as glutamic acid substitutions (Figure 3.4). Phosphoserine site-specific antibodies were also likely to cross-react weakly with phosphorylated tyrosine and unphosphorylated serine. Phosphothreonine site-specific antibodies were more specific for recognizing phosphorylated threonine, but also exhibited weaker recognition of phosphoserine and phosphotyrosine residues. Phosphotyrosine site-specific antibodies were the least promiscuous in affinity for phosphoserine, phosphothreonine, glutamic acid, as well as the unphosphorylated counterparts. Figure 3.5 shows Venn diagrams as to the overall specificities of the three phosphosite-specific antibody categories. For this analysis, the qualification of a reactivity was defined as medium strength or stronger, or a quantified signal strength of 3 or greater from the 0-5 scale in the previous analysis. Phosphoserine-site antibodies were found to be the least specific, with only about 8% of tested antibodies uniquely specific for phosphoserine. The majority of phosphoserine site-specific antibodies appeared to be reactive towards peptides with phosphothreonine substitutions (71%), and 39% were reactive with glutamic acid replacements. A lesser portion of 17% were reactive for phosphotyrosine, and 19% of tested antibodies were reactive for unphosphorylated serine. The phosphothreonine site-specific antibodies were less cross- 51 reactive, with 38% of the tested phosphothreonine site-specific antibodies demonstrating reactivity towards phosphoserine substituted peptides, while 12% and 10% cross-reacted with phosphotyrosine and glutamic acid, respectively. A subset of 50% were specific for only phosphothreonine, and 3% cross-reacted with unphosphorylated threonine. The phosphotyrosine site-specific antibodies were generally the most selective, with 73% of all tested phosphotyrosine site-specific antibodies indicating specificity for only phosphotyrosine. A smaller portion of 18% and 13% of tested antibodies cross-reacted with phosphoserine and phosphothreonine substitution peptides, respectively, while 2% and 3% cross-reacted with the glutamine substitution and unphosphorylated tyrosine peptides.   Figure 3.4. Average signal intensity of SPOT membranes featuring phosphoacceptor site substitutions probed with phosphosite-specific antibodies. A total of 456 antibodies were tested, consisting of 155 phosphoserine site-specific antibodies, 99 phosphothreonine site-specific antibodies, and 202 phosphotyrosine site-specific antibodies. Error bars represent standard deviation of the average signal intensities. Brackets indicate statistically significant differences in average signal intensities between categories at p ≤ 0.05.  52   Figure 3.5. Venn diagrams of cross-reactivities of phosphoserine, phosphothreonine and phosphotyrosine-site antibodies. A total of 456 antibodies were tested on peptide membranes where the central phosphoamino acid residue within each peptide was replaced by either phosphoserine (pS), phosphothreonine (pT), phosphotyrosine (pY), unphosphorylated serine (s), threonine (T), tyrosine (Y) residues, or glutamic acid (E). These Venn diagrams and in following figures were all designed with software created by Heberle et al. (2015).   53 3.5  Antibody yields versus specificity With the production of rabbit polyclonal phosphosite-specific antibodies, profound differences in their overall yields were observed. Anecdotally, it appeared that the lower the yield of a particular phosphosite-specific antibody, the poorer its potency. Consequently, I further investigated whether there was a general relationship between antibody yield and antibody specificity. Since all of the tested antibodies were all affinity purified using peptide columns, it could be rationalized that a high yielding antibody preparation could potentially be indicative of a better performing antibody with greater specificity due to increased competition binding to the limited peptide agarose that is saturated with rabbit sera during column loading. To answer this question, I reviewed the consensus sequences deduced by comparison of peptides with sequence similarity in dot blots as performed in Section 3.2. While variable, the average production yield trends to increase slightly with antibody specificity, as measured as the number of defined residues deduced as part of the defined consensus recognition sequence for each phosphosite-specific antibody (Figure 3.6).   54  Figure 3.6. Production yield plotted against the number of defined residues in consensus sequences. Consensus sequences were deduced based on the relative signal intensities of twelve similar peptides on dot blots per phosphosite-specific antibody. A total of 412 antibodies were tested. Error bars represent standard deviations.   Focusing on yields of the antibodies that targeted single or multiple phosphosites revealed that phosphothreonine site-specific antibodies were on average higher yielding than phosphoserine site-specific antibodies, followed by phosphotyrosine site-specific antibodies which were generally lowest yielding (Table 3.4). The analysis represented previously in Figure 3.4 supports the idea that phosphothreonine site-specific antibodies are generally more specific than phosphoserine site-specific antibodies, and here, phosphothreonine site-specific antibodies were found to have greater average production yield. Phosphotyrosine site-specific antibodies were indicated to be the most specific in Figure 3.4, but the data here does not conclusively R² = 0.49540.01.02.03.04.05.06.00 1 2 3 4 5 6Average Production Yield (mg) Number of Defined Residues in Consensus SequenceAllAntibodiespS SiteAntibodiespT SiteAntibodiespY SiteAntibodiesYield VSSpecificity(AllAntibodies) 55 support these antibodies to have the highest yields. This is likely because phosphotyrosine site-specific antibodies were subjected to additional negative purifications with phosphotyrosine-agarose during our purification procedures in an effort to extract out phosphotyrosine antibodies that were not site-specific and generally reacted with phosphotyrosine in any context in proteins and peptides; these would be considered as generic phosphotyrosine-specific antibodies.  Table 3.4. Phosphosite-specific antibody yields by phosphoamino acid specificity  pS Sites pT Sites pY Sites pS+pT SItes pS+pY Sites pT+pY Sites Number of Antibodies: 262 127 324 39 14 17 Average Yield (mg): 2.3±2.5 3.4±3.3 2.0±1.9 2.3±2.3 3.8±2.5 2.5±1.9 Median Yield (mg): 1.5±2.5 2.7±3.3 1.5±1.9 1.8±2.3 3.1±2.5 2.3±1.9 Min Yield (mg): 0 0.1 0 0 0.2 0.4 Max Yield (mg): 20.9 18.0 13.8 12.3 9.9 5.9 Total Yield (mg): 611.3 430.6 642.0 89.0 53.4 41.8  Numbers following ± represents standard deviation of the average yield.   An analysis of 38 batches of rabbit sera immunized with only one singular-phosphotyrosine site peptide per rabbit revealed that an average and median yield of 1.2 and 0.8 mg, respectively, of generic anti-phosphotyrosine antibody were co-purified for every 1 mg of phosphotyrosine site-specific antibody (Table 3.5).  56 Table 3.5. Generic and site-specific phosphotyrosine antibody yields.  Generic P-Tyr Antibodies Specific P-Tyr Antibodies Ratio – PYK / pY-Site specific Number of Antibodies: 38 38  Average Yield (mg): 1.1 ± 1.2 1.4 ± 1.3 1.2 Median Yield (mg): 0.6 ± 1.2 1.1 ± 1.3 0.8 Min Yield (mg): 0 0 0 Max Yield (mg): 5.4 6.0 6.8 Total Yield (mg): 43.7 58.5    Numbers following ± represents standard deviation of average yield.   Without the additional generic phosphotyrosine antibody purification, the total phosphotyrosine site-specific antibody yields would typically be greater on average than phosphothreonine site-specific antibodies, yet still potentially showing specificity for phosphorylated tyrosine and no other amino acids. However, the existence of generic phosphotyrosine antibody populations would result in a less phosphosite-specific antibody product. Taken together, the data seems to be supportive of a weak trend that antibodies with higher yields from purification production is loosely associated with greater antibody specificity.   3.6  Production and characterization of polyclonal generic phosphotyrosine-specific antibody In light of the importance and diverse impacts of tyrosine phosphorylation in a plethora of physiological and pathological conditions, many generic phosphotyrosine research antibodies, such as the 4G10 (EMD Millipore, Mississauga, ON, Canada), PY20 (EMD Millipore), P-Tyr-100 and P-Tyr-1000 (Cell Signaling Technologies, Whitby, ON, Canada) antibodies, are available commercially. These reagents have been invaluable tools for detection and enrichment of  57 tyrosine-phosphorylated proteins. Efforts to characterize the performance and individual epitope preferences of these monoclonal antibodies have already been described in literature and have revealed problematic aspects, such as the influence of neighbouring amino acid sequence flanking the phosphotyrosine residue (Tinti et al., 2012). Due to intrinsic preference of these generic phosphotyrosine antibodies for specific and less generic phosphotyrosine epitopes, the use of these monoclonal antibodies in research could lead to inaccuracies in the investigation of protein phosphorylation at tyrosine sites in cell signalling processes in some applications. It was hypothesized that a polyclonal anti-phosphotyrosine antibody would be better adapted to produce a broader general binding to more phosphotyrosine sites, and our laboratory developed our own polyclonal version of a generic phosphotyrosine antibody (PYK antibody, Cat# PG001, Kinexus Bioinformatics Corporation, Vancouver, BC, Canada), and evaluated the performance of this antibody as compared to some established contemporary monoclonal anti-phosphotyrosine antibodies.   Using the rabbit sera from the development of other polyclonal phosphotyrosine site-specific antibodies, over 100 different preparations of sera were passed through phosphotyrosine-beta-alanine-cysteine-coupled agarose columns for affinity purification of generic phosphotyrosine antibody populations, which were then pooled together (antibody purification methodology is described in Section 2.2). In 2015, Dr. Shenshen Lai in our lab had previously described the epitope characterization of the PYK antibody in comparison to the 4G10, PY20, and P-Tyr-100 monoclonal antibodies by their respective performances on peptide microarrays with 6099 phosphotyrosine peptides modelled after physiological phosphorylation sites from Jerini  58 Peptide Technologies. She found that the PYK antibody outperformed the other three commercial alternatives in the number of phosphotyrosine peptides that each antibody successfully bound to (both overall and uniquely) as well as also exhibiting the greatest signal intensities on the array. She also reported that the PYK antibody exhibited the least overall bias for adjacent amino acids surrounding the phosphotyrosine residue, as well as a much lower negative bias towards aspartic acids and glutamic acids, which are commonly found near physiological phosphotyrosines, than the other three commercial monoclonal antibodies. Protein-tyrosine kinases typically have a preference for acidic residues flanking the phosphoacceptor site for their substrate recognition (Safaei et al., 2011a).   Since then, I have re-purified the PYK antibody through phospho-serine-beta-alanine-cysteine agarose and phospho-threonine-beta-alanine-cysteine agarose to negatively purify any populations of antibody that exhibited affinity for the other phosphorylated amino acids. I have also expanded the comparison of the performances of PYK antibody to other commercial antibodies, including the newer rabbit monoclonal P-Tyr-1000 antibody from Cell Signalling Technologies. Kinex G Peptide Macro phosphopeptide arrays (Appendix A), produced by spotting purified phosphopeptides onto nitrocellulose membranes, were probed with either PYK antibody at 0.5 µg/ml or P-Tyr-1000 at 1:1000 dilution (Figure 3.7). The P-Tyr-1000 antibody demonstrated high selectivity and potency against phosphotyrosine peptides on the array. While the strongest immunoreactivity of the PYK antibody were towards phosphotyrosine peptides, there was slightly more cross-reactivity against phosphoserine and phosphothreonine peptides than with P-Tyr-1000. Nevertheless, the polyclonal nature of the  59 PYK antibody preparation would likely be able to detect more tyrosine phosphorylation sites in more proteins that P-Tyr-1000, be much cheaper to purchase, and be stable to repeated freeze-thaw cycles than the monoclonal antibody. An application of the PYK antibody for Western blotting for detection of phosphorylation changes in human cancer cells lines with a growth factor or protein-tyrosine phosphatase inhibitors is presented in Figure 3.8.   Figure 3.7. PYK and P-Tyr-1000 generic phosphotyrosine antibody specificities on   Kinex™ G Macro phosphopeptide arrays. PYK antibody was used at 0.5 µg/ml (1:2000 dilution), and P-Tyr-1000 at 1:1000 dilution. The locations of phosphotyrosine-containing peptides are indicated within the outlined area, whereas other phosphoserine- and phosphothreonine-containing peptides were printed outside of this zone. More information of the sequences of peptides printed on this nitrocellulose array can be found in Appendix A.  60  Figure 3.8. Western blotting of A431 and HeLa cells with PYK antibody. Lysates of overnight serum starved A431 cells treated with 100 ng/ml EGF and HeLa cells treated with 25 µg/mL phenylarsine oxide and 50 µg/mL sodium orthovanadate were probed by Western blotting with PYK antibody at 2 µg/mL.   3.7  Discussion In the consensus sequence analysis of twelve related peptides by dot blotting per phosphosite-specific antibody, it is acknowledged that there could be a bias in the rationalized consensus sequences based on the particular peptide sequences chosen for testing as well as the other eleven peptides available per immunizing sequence. I was limited by the number of peptides available for this purpose, but the purity of these phosphopeptides was 95% or greater, and their sequences were all verified by mass spectrometry. To control for the internal bias of  61 amino acid representation in the sequences of all peptides used in the dot blots, ideally I should have calculated the predicted occurrence of each amino acid residue at each position for direct comparison with the observed occurrence of these amino acid residues in the deduced consensus sequences for antibody recognition. However, the consensus sequence data was primarily used to identify the amino acid positions that would be most and least critical for antibody recognition in general, and the predicted occurrence of amino acids found in any position in the immunizing peptides was deemed to be sufficiently representative. Another limitation to the experiment is that these dot blot membranes were prepared using purified peptides of varying ages, and antibody non-reactivity may have been partially reflective of unstable peptide preparations that had been degraded. These reasons are why I sought to analyze the positional importance of amino acids flanking the phosphoacceptor site by a second strategy that was more systemic and prepared membranes featuring peptides with glycine substitutions that were all synthesized directly on the membranes used for probing at all around the same time. However, the purity and yield of each peptide spot on the SPOT membranes could not be determined. Furthermore, it should be appreciated that a loss or the lack of a loss of signal due to glycine substitution is dependent on how similar a glycine residue can structurally substitute the original wildtype amino acid. This may result in similar amino acids like alanine to be falsely concluded to be non-critical for epitope recognition, while it could in fact be important and was just well substituted by a glycine residue. Glycine substitution could have also caused negative hinderance of antibody epitope recognition when substituting a non-critical amino acid, leading to a result that may indicate the original amino acid as being critical. Some of these issues related to both the peptide dot blot consensus  62 sequence analysis and the glycine substitution experiments may explain why there are conflicting conclusions of glutamine and threonine being found to be critical in epitope recognition in one method but not the other, as well as the variation in the trending ranks of the most important amino acid positions in relation to the phosphosite for antibody recognition. Nevertheless, the peptide dot blot and glycine substitution experiments both revealed that the -1 and the -2 positions are generally the most important out of the 3 flanking residues N and C-terminal to the phosphoacceptor amino acid in creating the antibody recognition epitope. While my experiments were primarily designed to test the positional importance of flanking residues, both experiments are in agreeance that hydrophobic amino acid residues like phenylalanine, isoleucine, leucine and valine, as well as arginine are generally important for phosphosite-specific antibody recognition, while alanine, methionine and tryptophan seemed to be less important. Further experimentation specifically designed to test the immunogenic antibody recognition value of specific amino acids can be performed with SPOT synthesis to more thoroughly address this research question in the future.   The SPOT membrane experiments of peptides featuring phosphoacceptor amino acid substitutions revealed that phosphoserine site-specific antibodies seem to commonly be cross-reactive with phosphothreonine sites and appreciably with glutamic acid residues. By contrast, the phosphothreonine and phosphotyrosine site-specific antibodies seem to be more selective for their respective phosphoamino acids. A potential explanation could be that antibodies raised against phosphoserine-sites are less specific have a more generic affinity for the phospho-moiety because of the size of the phosphoserine amino acid. Phosphothreonine- and  63 especially phosphotyrosine-binding antibodies are perhaps more specific to the bulkier side chains on these amino acids and are less likely to cross-react with other phosphoacceptors as well as glutamate. Given appreciable suitability of glutamate to mimic a phosphoamino acid, in a future experiment, it would be interesting to test if these antibodies have affinity to an aspartate residue. The results shown here could be an indication that other commercial phosphoserine site-specific antibodies may not be as specific as preferred and could lead to inaccuracies when trying to draw conclusions from the results from experiments utilizing these antibodies. As evidenced with the generation of phosphotyrosine site-specific antibodies, it may also be potentially beneficial to negatively purify out non-specific antibodies from phosphoserine site-specific antibodies. However, in my experience, very low yields of generic phosphoserine site-specific antibodies are recovered from the application of phosphoserine-beta-alanine-agarose columns on serum from rabbits immunized with phosphoserine-containing peptides. For every 1 mg of phosphotyrosine site-specific antibody that was produced, an average of 1.1 mg of generic phosphotyrosine PYK antibody was also purified. The high ratio of generic tyrosine antibodies that could be purified from rabbit sera immunized by a single phosphotyrosine-containing peptide per animal indicates that polyclonal antibodies sold commercially are probably not very phosphotyrosine-site specific, unless they have gone thorough negative generic phosphotyrosine antibody purification as performed here. This could contribute to significantly increased cross-reactivities with off-target proteins with many of the phosphotyrosine site-specific antibodies used in scientific studies, which would lead to errors in the interpretation of Western blotting even if the reactive protein band appears to be in the expected range of the target protein. A typical cell lysate resolved by SDS-PAGE may be  64 expected to feature thousands of different protein forms generated by alternative mRNA splicing and post-translational modifications, and the migration of the intended phosphoprotein target band may differ from what is expected based on its predicted size.   The experimentation with the PYK antibody demonstrated here shows that a useful polyclonal generic phosphotyrosine antibody is possible, and that the performance of such an antibody is not critically inferior to commercial monoclonal antibody alternatives. Due to a broader affinity profile, a polyclonal generic phosphotyrosine antibody may be more beneficial in applications such as immunoprecipitation and phosphotyrosine-peptide enrichment. Although slightly less specific to only phosphotyrosine peptides than the P-Tyr-1000 antibody, the PYK antibody was in essence generated as a by-product from the purification and refinement of phosphosite-specific antibodies, and required minimal resources to produce as compared to generating B-cell hybridoma clones in culture from the spleen of an animal. The approach taken here is a resource-friendly and humane alternative in generating a useful generic phosphotyrosine affinity reagent that I found to be compatible for applications in Western blotting, microarrays, ELISA and immunoprecipitation experiments.   Ultimately, the findings of this antibody characterization work will help guide the interpretation of my research as I aimed to use antibody-based techniques to assay biological model systems to elucidate cell signalling networks.     65 Chapter 4: Optimization of High Content Antibody Microarrays  4.1  Rationale The use of both pan and phosphosite-specific antibodies in antibody microarrays allows the simultaneous quantification of protein expression and phosphorylation to provide valuable insight into the signalling networks inside of cells. However, the interpretations of the findings are all influenced by general issues related to using antibodies for research, as well as challenges that are unique to antibody microarrays. Some of these barriers in getting high quality reliable data include the instability of phosphorylation sites in lysate proteins post homogenization, cross-reactivity with off-target proteins, protein-protein interactions that can hinder accessibility of epitopes from capture by antibodies, and unintended non-specific binding interactions with dye-labelled analytes.   The vast majority of proteins within the human proteome are subject to phosphorylation, and protein phosphorylation may be the predominant mechanism for reversible post-translational regulation of proteins (Humphrey et al., 2015; Jünger and Aebersold, 2018). Most commonly used antibody microarray methods involve the adsorption of lysate proteins in native states in buffers containing mild detergents. The work in this chapter demonstrates that homogenization of cells in standard buffer containing mild detergents and phosphatase inhibitors are insufficient to preserve most phosphorylations, and that fragmentation of proteins at cysteine residues is a compelling strategy to maintain the phosphorylated state of proteins for their  66 detection with antibody microarrays. Further advancements of the antibody microarray technology described here include optimization of target protein and phosphorylation signal strength while minimizing false signals generated from the unspecific binding of dye-labelled analytes, as well as new applications of the array in sandwich-style assays similar to existing ELISA techniques but in larger scale for improved throughput.   The majority of work described in this chapter used the human epidermoid cervical carcinoma A431 cell line, which overexpress EGFR through gene amplification, as well as a truncated isoform, and tracks the phosphorylation of EGFR and downstream signalling proteins (Haigler et al., 1978; Xu et al., 1984). While the data generated by current antibody microarray applications still leaves room for improvement, the potential of the platform is underscored by its ability to simultaneously track hundreds of diverse proteins of interest in a targeted manner with high sensitivity and speed. Once fully developed, antibody microarrays can have an even greater impact in research for reliable analysis of clinical biomarkers, drug targets, and improving the understanding of cell signalling systems.   4.2  Lability of phosphorylation sites To examine pronounced activation of EGF signalling systems and phosphorylation of EGFR, cell cultures of A431 were subjected to serum starvation overnight and briefly exposed to 100 ng/ml EGF for up to 5 min prior to processing with Standard Homogenization Buffer (SHB), which contains many standardly utilized inhibitors of protein-serine/threonine phosphatases and protein-tyrosine phosphatases and non-ionic detergent. Western blotting was performed  67 on these lysate samples using five different antibodies that were affinity-purified for various phosphosites within the ~160 kDa receptor (Figure 4.1A-E). Parallel cultures of A431 cells, which were treated in the same fashion but directly homogenized into SDS-PAGE loading buffer, were also analyzed by Western blotting (Figure 4.1F-J). Out of five phosphosites, it was evident that phosphorylation of EGFR induced by EGF simulation at these phosphosites in A431 cells were lost during the processing of the lysates in SHB. While the improved performance of the phosphosite-specific antibodies can be partially explained by better extraction of EGFR from the plasma membranes due to the stronger anionic detergent SDS, the amount of Triton-X100 used in the standard buffer was also tested to be comparatively proficient in solubilizing the total amount of the receptor (Figure 4.2).    Figure 4.1. Lability of EGF receptor phosphosites in lysates prepared from EGF-treated A431 cells homogenized in standard buffers. Lysates from A431 cells that were serum-starved for 18 h and then treated without and with 100 ng/mL EGF for 1 and 5 min were prepared by harvesting in Standard Homogenizing Buffer (SHB) that included inhibitors of protein phosphatases and proteases (A-E) or directly in sodium-dodecylsulphate-polyacrylamide gel electrophoresis (SDS-PAGE) buffer (F-J). The detergent-solubilized lysates were subjected to SDS-PAGE and Western blotted with five different phosphosite-specific antibodies that targeted distinct EGF receptor phosphosites.   68   Figure 4.2. Solubilization of EGFR in Triton X-100 and SDS-PAGE sample buffer. Lysates from A431 cells that were serum-starved for 18 h and then treated without and with 100 ng/mL for 5 min were prepared by harvesting in Standard Homogenizing Buffer (SHB) containing Triton X-100 or directly in sodium-dodecylsulphate-polyacrylamide gel electrophoresis (SDS-PAGE) buffer. After collection of the supernatant fractions post ultracentrifugation of the Triton X-100 detergent solubilized lysates, the residual pellets were re-processed into SDS-PAGE buffer and were also analyzed in parallel by Western blotting with five different phosphosite-specific antibodies that targeted distinct EGF receptor phosphosites as well as four different pan- specific EGFR antibodies. All antibodies used here are available from Kinexus Bioinformatics Corporation.    69 4.3  Maintenance of phosphorylation states by chemical cleavage As using lysates solubilized in buffers containing SDS would not be conducive for use on an antibody microarray, since SDS would interfere with antigen-antibody interactions, an alternate strategy would have to be developed. Using the Kinex™ KAM-900P antibody microarray, I examined four different methodologies in the sample processing and their effectivity in monitoring changes in protein phosphorylation and expression (Figure 4.3).  Figure 4.3. Schematic representation of alternative lysate sample and antibody microarray processing strategies. With Method 1 (A steps), cells were harvested in Standard Homogenizing  70 Buffer (SHB) that included protein phosphatase and protease inhibitors and then the lysate proteins were directly labelled with fluorescent dye prior to incubation with the microarray. With Method 2 (B steps), lysates were initially prepared in SHB and then after ultracentrifugation subjected to cysteine chemical cleavage (CCC). The fragmented lysate proteins were then dye-labelled and captured on the antibody microarray. With Method 3 (B steps), the cells were directly homogenized into CCC Buffer (CCCB) and the fragmented lysate proteins were subsequently dye-labelled and incubated with the antibody microarray. With Method 4 (C steps), the cells were directly homogenized into CCCB and the fragmented lysate proteins were then biotinylated and incubated with the antibody microarray. Subsequently, the microarray was probed with dye-labelled anti-biotin antibody. With all of these procedures, the free dye or free biotin was resolved from the dye-labelled or biotinylated proteins by gel filtration chromatography prior to incubation with the antibody microarrays.   Signals detected from captured proteins on antibody microarrays can be influenced by many variables that could produce false-positive and false-negative results. Such factors include the size of the target protein, the amount of dye that it binds, and its interactions with other proteins in complexes. Moreover, endogenous phosphatases and kinases may induce post-homogenization changes in protein phosphorylation. Many of these issues can be mitigated by cleaving the protein samples into peptide fragments, thus normalizing the amount of dye bound to the peptide epitopes and breaking protein-protein interactions. By fragmenting proteins, antibodies that were raised against different epitopes on the same target protein do not have to compete during the incubation of the lysate sample. Disrupting protein-protein interactions will also help uncover epitopes that were previously masked by such interactions.    71 Previously, work in our lab and others has found that successive treatment with Tris (2-carboxyethyl) phosphine hydrochloride (TCEP) and 2-nitro-5-thiocyanatobenzoic acid (NTCB), which cleaves proteins after cysteine residues, could be used compatibly with antibody microarrays to reduce the interference of protein-protein interactions (Wu et al., 1996; Zhang et al., 2006). Many of the short epitope sequences used to generate the antibodies used on the Kinex™ KAM-900P microarray avoided the inclusion of cysteine residues, and peptide fragments generated by the CCC method would retain these epitopes for recognition.  To investigate the effectiveness of the various CCC methods, EGF-stimulated A431 cells were prepared for analysis with the KAM-900P antibody microarray. The introduction of post-lysis CCC revealed a marked reduction in the signal intensities of spots (Method 2), based on scanned images of fluorescent dye-labelled proteins captured with this microarray. Quantification of the individual antibody spots demonstrated an overall 70% reduction in fluorescent dye signals as compared to usage with standard homogenizing buffer (SHB) and no protein fragmentation (Method 1; Figure 4.4A, B). If the CCC was performed at the time of homogenization and during the initial sample processing (Method 3), then the strengths of the individual fluorescent dye signals, visually and as quantified, were comparable to those observed without CCC (Figure 4.4C). Since the majority of antibodies on the array were phosphosite-specific, the increase in signal intensity generated by lysates that were subjected to CCC during the initial homogenization as opposed to cleavage after homogenization with SHB supports the notion that protein phosphorylation is better preserved through earlier fragmentation of proteins. This is likely explained by the concurrent fragmentation of  72 endogenous protein phosphatases that could still rapidly dephosphorylated lysate proteins when samples were prepared in SHB.   Figure 4.4. Scanned images of Kinex™ KAM-900P antibody microarrays incubated with lysates from EGF-treated A431 cells subjected to alternative lysate sample and antibody microarray processing strategies. A431 cells were serum-starved for 18 hours and then treated with 100 ng/mL EGF for 5 min and subjected to the four different protocols that are outlined in Figure 1. The strongest signals appear red, intermediate signals are yellow and then green, and the weakest signals are blue.   73 4.4  Antibody microarray analysis of EGF-treated A431 cells By comparing A431 cells treated with 5 min of EGF with control treatments (A431 cells with serum starvation) on the microarray, I was able to evaluate EGF-induced changes. Dramatic inductions of phosphorylation, especially on tyrosine-sites were expected (Hunter and Cooper, 1981). However, when the data from replicated studies with fluorescent dye-labeled lysate proteins were analyzed in aggregate for EGF-induced changes, relatively few leads of 35% or greater in changes in protein expression or phosphorylation were observed to be statistically significant in three or more replicated experiments (Table 4.1).   One of the major issues associated with the detection of fluorescent dye-labelled lysate proteins, or fragments following CCC, was carryover of unreacted dye following the labeling reaction and resolution from dye-bound protein over a Sephadex G25 spin-column. While separation by gel filtration effectively removed the majority of unreacted dye, residual free dye appeared to still be able to cause partial dye labelling of the capture antibodies on the microarray, despite the inclusion of 50 mM ethanolamine in an attempt to neutralize this carry-over dye by quenching any remaining reactive groups. This resulted in higher background signals and minimized the apparent lead changes that resulted with respect to protein expression or phosphorylation. To address this problem, I developed a biotin-labelling method after chemical cleavage of proteins at the time of cell homogenization (Method 4). Following capture on the KAM-900P antibody microarray, the biotinylated proteins were subsequently detected with a fluorescent dye-labelled anti-biotin antibody that was subjected to more stringent purification procedures as well as more extensive free-dye neutralization. I found this  74 worked much better than fluorescent dye-labelled avidin for detection, as this produced much lower backgrounds in control experiments. Since the size of immunoglobulin G was approximately thrice the size of avidin, the use of dye-labelled anti-biotin antibody also increased the amount of signal detection upon successful capture of biotin-labelled analytes on the array. Further optimization of the incubation concentration and time of the dye-labelled anti-biotin antibody revealed that a shorter incubation period of 10 min of only 1 µg of labelled reporter antibody per chip, as opposed to usual incubation protocols of at least 1 h using greater amounts, was the most effective at reducing background signal and maximizing the lead changes in protein expression and phosphorylation generated by the array.   Application of the CCC protein fragmentation method with biotin labelling and detection with anti-biotin antibody (Method 4) from EGF-treated A431 cells produced more than 20-times the number of statistically significant leads than observed with fluorescent dye-labelled Method 3, most likely due to reduced non-specific background signals (Table 4.1). Over 20% of the 265 pan- and 613 phosphosite-specific antibodies on the KAM-900P microarray revealed EGF-induced changes of 35% or greater, and this was visually evident from the scanned images of the arrays as well as in the data generated by the quantification of these images from replicate experiments. All of the following studies described hereafter were based on comparison with the results from experiments with antibody microarrays using the protein fragment biotin-labelling Method 4 unless indicated.    75 Table 4.1. Comparison of lead generation from different homogenization and detection systems using EGF-treated A431 cells and the Kinex™ KAM-900P antibody microarray   Lysates from A431 cells that were serum-starved for 18 h and then treated without and with 100 mg/mL EGF for 5 min were prepared by harvesting in Standard Homogenizing Buffer (SHB) or homogenized directly into Cysteine Chemical Cleavage Buffer (CCCB). For Method 1, lysate proteins from SHB processed cells were directly dye-labelled and incubated with the KAM-900P antibody microarrays. For Method 2, lysate proteins from SHB cells were subsequently subjected to Cysteine Chemical Cleavage (CCC), and then dye-labelled prior to KAM-900P incubations. For Method 3, A431 cells were homogenized in CCCB, and then directly dye-labelled for KAM-900P analyses. For Method 4, A431 cells were homogenized in CCCB, and then biotin labelled. Following initial incubation with KAM-900P microarrays, the captured biotinylated lysate proteins were detected with dye-labelled anti-biotin antibodies. Each set of three to five lysates from untreated and three to five lysates from EGF-treated were separately analysed, and duplicate measurements were carried out with each antibody microarray for each lysate. Only antibodies that revealed EGF induced percentage changes from controls (%CFC) of 35% or greater that were statistically significant with p values of ≤ 0.05 with the paired, two-tailed Student T-test are shown.    76 4.5  Validation of antibody microarray leads by Western blotting There were more than 107 different phosphosite-specific antibodies that showed EGF-induced changes in phosphorylation of captured lysate proteins on the KAM-900P microarray. I then endeavored to validate of many of the key positive and negative results to establish the reliability of the antibody microarray data from application of Method 4 with CCC and biotin labelling. Key EGF-induced changes as identified by the antibody microarray can be found in Appendix B. Representative Western blots with many of the phosphosite-specific antibodies that presented EGF-induced changes are shown in Figure 4.5.     77  Figure 4.5. Western blotting validation of antibody microarray leads of EGF induced protein phosphorylation. Lysates from A431 cells that were serum-starved for 18 h and then treated without and with 100 ng/mL EGF for 1-5 min were prepared by directly harvesting the cells into SDS-PAGE sample buffer (A-N) or Standard Homogenizing Buffer (O-AC). These lysates were then Western blotted with phosphosite-specific antibodies. Antibody target proteins are indicated with arrows based on their immunoreactivity and expected sizes on the immunoblots.   78 Of the 91 tested antibodies that indicated changes (increased or decreased phosphorylation) on the microarray, only 27 (30%) of these could be confirmed by Western blotting. For 30 (33%) of the tested antibodies in this “positive” set, Western blotting was not sensitive enough to reveal any immunoreactive proteins so many of these leads could still be correct. For the remaining 34 antibodies (37%) in this “positive” set, although changes in the target proteins could not be confirmed by Western blotting, it appeared these false positives probably arose from cross-reactivity with off-target proteins, and in particular with the EGF receptor itself. I then proceeded to examine 83 phosphosite-specific antibodies that did not show any significant changes with EGF treatment of A431 cells on the KAM-900P antibody microarray. Of these, 74 were either validated to have no changes (23%) or were of insufficient sensitivity for target detection (77%) by Western blotting. Of the other 9 antibodies from this “negative” set, 8 of these successfully detected their targets by Western blotting, which were actually observed to exhibit alterations in phosphorylation with EGF treatment. Overall, nearly 60% of the results from the antibody microarray appeared to reflect true positives and true negatives, and all of the false positives that yielded immune reactivity on Western blotting were due to off-target proteins that were also affected by EGF treatment of the A431 cells. Table 4.2 summarizes the results from these immunoblot tests with 174 different phosphosite-specific antibodies.        79 Table 4.2. Assessment of EGF-induced changes in protein phosphorylation in A431 cells as detected with the Kinex ™ KAM-900P antibody microarray and Western blotting. Validation result Number % Number target detections by immunoblotting Increased EGFR detection by immunoblotting Increased detection of non-EGFR off-targets by immunoblotting True positive 27 15.5 27 9 6 True negative 74 42.5 17 12 12 False positive (FPc) 64  (34) 36.8 (17.2) 6 25 9 False negative 9 5.2 8 2 4  True positive and true negative correspond to correlation between the positive and negative effects, respectively, of EGF treatment between the antibody microarray and Western blotting results for the target protein. False positive results correspond to an apparent change with EGF treatment that could not be confirmed in the target protein by Western blotting. FPc = false positive, but a similar EGF-induced change in one or more unintended cross-reactive protein targets on the Western blot that correlates with the antibody microarray results. False negative corresponds to an EGF- associated change on the Western blot that was not evident with the antibody microarray.  4.6  Sandwich antibody microarray Many generic phosphotyrosine-site antibodies have been used as affinity reagents to enrich phosphotyrosine containing peptides prior to MS analyses. However, they can also be used as reporter antibodies to detect differential phosphorylation of lysate proteins on antibody microarrays. A previous application of this concept was performed for the detection of protein-tyrosine phosphorylation with 35 capture antibodies on a microarray incubated with lysate proteins from Bcr-Abl-expressing RT10+ cells as well as EGF-treated HeLa cells probed with Cy5- 80 labelled custom anti-phosphotyrosine monoclonal antibody (Gembitsky et al., 2004). Gembitsky and colleagues recorded reductions of at least 45% in the tyrosine phosphorylation signals for 23 of the antibody targets when the RT10+ cells were pretreated for 2 hours with the protein-tyrosine kinase inhibitor Gleevec, as well as 2-fold increases in tyrosine phosphorylation signals with 16 antibodies following 1 hour exposure of EGF in HeLa cells that had been previously serum-starved for 24 h. Using the Kinex™ KAM-1150 antibody microarray, which features only pan-specific antibodies for capture of protein targets regardless of phosphorylation state, I have also detected increased tyrosine phosphorylation of lysate proteins in A431 cells exposed to EGF for 0, 1, 5, 10, and 20 min using a similar approach with the generic polyclonal anti-phosphotyrosine PYK antibody (as described in Chapter 3) used as a reporter antibody.   In my experiments, 100 µg of lysate protein from serum starved A431 cells treated with or without EGF for 1-20 min were prepared with SHB and subjected to antibody microarray analyses by direct dye labelling as described as Method 1 from Figure 4.3. Replicate cell lysates were prepared that were not cleaved nor tagged and were also used for capture onto the microarray and detected with dye-labelled PYK antibody (Figure 4.6). Chemical cleavage of the lysate proteins was avoided in this case, because I wanted to detect additional tyrosine phosphorylated sites on the target proteins that may not have directly flanked the epitopes of the capture pan-specific antibodies on the KAM-1150 antibody microarray. As a control, another microarray field was incubated with only the dye-labelled anti-phosphotyrosine polyclonal rabbit antibody in the absence of cell lysates, and these immunoreactive antibodies  81 are circled in Figure 4.6 (Panels E, J and D). However, these non-specific background signals were not saturating and increased signals could be detected when exposed to lysate.    Figure 4.6. Kinex™ KAM-1150 antibody microarray analyses of human A431 cervical carcinoma cell lysates following treatment of the cells with and without 100 ng/mL epidermal growth factor (EGF) for 5 minutes for protein expression levels as well as tyrosine phosphorylation. Only 3 representative grids of the 16 distinct antibodies grids printed in a single field are shown as scanned images in this figure; each field was replicated four times per array. The strongest signals appear red and the weakest signals are blue. There were little if any apparent changes in protein expression in A431 cells with short term exposure to EGF (Panels A, B, F, G, K and L), but the treatment increased the phosphorylation signals for several of the captured proteins (Panels C, D, H, I, M and N).   82 Appendix C includes all the key results where percent changes from control (%CFC) of -35 and lesser or +35 or greater were observed in at least one of the time points (1-20 min EGF treatment), and were found to be statistically significant (p <0.05) based on analysis with a Student’s T test. Considering only the number of significant changes of +35% or greater, there was an increase in tyrosine phosphorylation identified in 25 antibody spots when A431 cells were treated with EGF for 1 min, 30 spots by 10 min, which tapered back down to 12 spots by 20 min. In contrast, looking at only the number of significant reductions of -35% or more, there was only 1 antibody spot identified to exhibit a decrease in tyrosine phosphorylation with 1 min of treatment, which increased to 6 spots by 20 min of EGF exposure. This indicates that feedback mechanisms to reduce tyrosine phosphorylations were increasingly activated by 20 min of treatment.   In later experiments, I was able to further increase the signal sensitivity for tyrosine phosphorylation detection by using biotinylated PYK antibody and detecting the presence of this reporter antibody with fluorescent dye-conjugated anti-biotin antibody. In addition, by labelling the lysate protein in either Cy3 or Cy5 dye, and labelling the PYK reporter antibody with the other dye, I also found that it was feasible to analyze changes in protein phosphorylation and expression in the regular antibody microarray analysis, coupled with simultaneous investigation of changes in tyrosine phosphorylation on the same array by switching the wavelength of the laser in the microarray scanner. I also found that any microarray chips that had been previously ‘used’ in the regular fashion could be re-probed with a compatible dye-labelled PYK antibody for retro-fit experimentation. This type of sandwich  83 assay application with the antibody microarray allows investigation of other types of covalent modification using the appropriate reporter antibodies. An as example I monitored changes in the ubiquitination of proteins using the KAM-1150 antibody microarray and dye-labelled anti-ubiquitin antibody as a reporter antibody (Figure 4.7).  Figure 4.7. Kinex™ KAM-1150 antibody microarray analyses of human HeLa cervical carcinoma cells treated with 50 µM of the proteasome inhibitor MG-132 for 4 hours. Cell lysates with 1.0% Triton-X100 were prepared and 100 µg were incubated with the KAM-1150 chip, which was subsequently probed with anti-ubiquitin antibody (Cat. No. SC-8017, Santa Cruz Biotechnology) that was dye-conjugated. Inhibition of protein degradation with the MG-132 resulted in increased detection of several proteins on the KAM-1150 antibody microarray as shown in 3 representative grids of distinct 16 grids of antibody spots, which were each  84 replicated in four fields on the array. The strongest signals appear red and the weakest signals are blue.   4.7  Discussion  From the Western blotting analyses, the identification of the EGF receptor as a cross-reactive protein was a common observation that resulted in many false positive hits on the antibody microarray analysis with phosphotyrosine site-specific antibodies. Out of 118 non-EGF receptor-based phosphotyrosine site-specific antibodies, 30 candidates were strongly cross-reactive with the EGF receptor in lysates from EGF-treated A431 cells, and 6 yielded weak, but detectable signals. Furthermore, 11 phosphoserine as well as 9 phosphothreonine site-specific antibodies also showed immunoreactivity with the EGF receptor in these immunoblots. All of these phosphosite-specific antibodies had already undergone prior negative purification with phosphotyrosine-agarose twice to reduce non-specific phosphotyrosine-binding. These antibodies do not typically cross-react with EGFR-sized bands in other model systems and the cross-reactivity observed in this experiment with phosphosite-specific antibodies may be a reflection of the very high level of EGFR over-expression in A431 cells, as well as the pronounced hyperphosphorylation of this receptor in response to EGF (Hunter and Cooper, 1981; Linsley and Fox, 1980). At least 32 serine, 23 tyrosine, and 17 threonine phosphorylation sites in the human EGF receptor have been documented in databases such as PhosphoNET (www.phosphonet.ca) and PhosphoSitePlus (www.phosphosite.org) by mass spectrometry studies, and phosphorylation is often also induced at serine and threonine sites in addition to tyrosine sites upon engagement of the EGF receptor with its ligands. Furthermore, based on the studies shown in Chapter 3 of this thesis, it is apparent that many phosphosite-specific  85 antibodies do display cross-reactivities, and the phosphotyrosine site-specific antibodies tend to be particularly reactive with similar phosphotyrosine sites on off target proteins.   In addition to problematic cross-reactivities with the phosphosite-specific antibodies leading to false positives, there was also an issue related to false negatives. This was especially reflected with the lack of detection of changes in the phosphorylation of the MAP kinases ERK1 and ERK2 with the KAM-900P antibody microarray, despite clear increases in the phosphorylation of these kinases by Western blotting. Potentially, some phospho-epitopes may be blocked by local binding of the oxygen atoms of the phosphate from the phosphosite with nitrogen atoms in flanking arginine and lysine residues. My attempts in disrupting such internal charged interactions without also causing loss of the binding of phosphofragments to the capture antibodies on microarrays were unsuccessful. This is a significant issue that not only can affect the reliability of antibody microarrays, but also other immunological techniques such as immunoprecipitation, ELISA, and immunohistochemistry that rely of detection of certain phospho-epitopes with antibodies. With extensive denaturing with SDS prior to electrophoresis, and subsequent Western blotting, it appears that such internal interactions of phosphosites with flanking basic residues can be successfully disrupted. Parallel to the chemical cleavage methodology, I have also found that enzymatic cleavage of lysate proteins with trypsin, which normally hydrolyzes proteins after basic amino acid residues, also failed to reveal EGF-induced ERK1 and ERK2 phosphorylation. In fact, at least 4-fold fewer EGF-induced changes were evident with the KAM-900P antibody microarrays using tryptic fragments as compared with the cysteine chemical cleavage Method 4 (data not shown).   86 One of the most surprising outcomes of this study was the large number of EGF-induced changes in protein levels as revealed with pan-specific antibodies. With 5 min of exposure of EGF to the A431 cells, 60 pan-specific antibodies showed EGF-induced increases or decreases of 35% or more in apparent protein expression. Amongst the most consistent EGF changes observed with two independent pan-specific antibodies that target different epitopes on the same protein were: decreases in EGF receptor; and increases in HSP90, ERK1, JAK2, and p38-beta MAP kinase. However, conflicting changes with different pan-specific antibodies were observed for other target proteins such as MST1, PDGFRB, and RafB. One possible explanation for observed changes in protein expression could be the masking of epitopes on CCC-generated protein fragments by adapter/scaffolding protein domains, that might not be as susceptible to CCC. This could include phosphotyrosine-sites that are targeted by SH2 and PTB family phosphotyrosine-binding domains. Binding of intact SH2 and PTB domains to phosphotyrosine sites on CCC-generated peptides could also exaggerate the detection these peptides on the microarray if these domains were also biotinylated.  To investigate the possibility of SH2 and PTB domains surviving our chemical cleavage protocols functionally intact, our lab identified the number and location of cysteine residues in 116 human SH2 domains and 56 human PTB domains. Approximately 84% of the PTB domains featured 3 or more cysteine residues, which might be expected to be sensitive to CCC, resulting in loss of functionally active structures. By contrast, two-thirds of the SH2 domains had two or less cysteine residues with 18 of these having none at all. Therefore, it is possible that certain SH2 domains might still be able to bind to phosphotyrosine sites on CCC-generated peptides  87 and prevent their capture onto antibodies on the microarray. This could be a confounding factor in the interpretation of results from similar studies conducted with other ligand-activated receptor-tyrosine kinases to explore their signaling networks in cells.  To follow up the study of analyzing tyrosine phosphorylation of A431 cells treated with EGF in the sandwich antibody microarray application, many avenues could be researched. Leads generated from the study that are not known as either direct substrates of EGFR, or involved in the EGFR signalling pathway, can be validated by immunoprecipitation of the target using the capture antibody printed on the array, and subsequent Western blotting with the PYK generic phosphotyrosine antibody, or by MS analysis for identification of phosphotyrosine peptide fragments. Since non-fragmented protein lysates were used on the arrays, it is very likely that many phosphorylation sites of interest would have already been lost by the time these lysates were analyzed on these arrays. It is possible that the protein undergoing tyrosine phosphorylation is not the target protein of the capture antibody, but in fact other proteins that are in complex with the target protein. While in later experiments I have found that it is still possible to use fragmented lysates in the sandwich antibody microarray technique with PYK anti-phosphotyrosine antibody, that would limit the number of total tyrosine residues available for detection. However, fragmentation of the lysate prior to their capture on the array could also help localize candidate tyrosine sites based the epitope of the capture antibody in relation to the nearest up- and down-stream cysteine residue in the amino acid sequence. Phosphorylation sites are apparently clustered in proteins. Although this study focused on EGF-regulation of phosphorylation in living cells, similar strategies can be adapted to evaluate the  88 substrate specificity of protein-tyrosine kinases in in vitro kinase assays of lysate proteins that are subsequently captured on antibody microarrays, as well as for uncovering protein-protein interactions with specific targets of interest, given that a suitable reporter antibody for that target protein is available.   Taken together, the sample preparation and handling refinement work executed here will aid in the pursuit of generating high quality reliable data using antibody microarrays so that I can explore cell signalling networks and detailed relationships between proteins in model systems that are indicative of physiological processes in nature.    89 Chapter 5: Investigations of EGF and Insulin Signalling Systems   5.1 Rationale  The EGF signalling network serves as a well-known paradigm for receptor-tyrosine kinase signalling mechanics, and it forms one of the principal systems for regulating cellular proliferation, growth, and differentiation during normal physiology. Due to its strong role in growth and proliferation, hyperactivity of this signalling system is often involved in growth of tumours and malignant cancers. Over-expression of EGFR has been associated as a marker for poor prognosis and response to endocrine therapy in patients with breast and ovarian cancer, among others (Barlett et al., 1996; Salomon et al., 1995). In a similar theme, the importance of insulin receptor (INSR) signalling is paramount in the cellular functioning of glucose metabolism of the body. Insufficient production of insulin or aberrant loss of function of signalling pathways downstream from INSR are at the root of the pathology of diabetes. While there are treatment options to mitigate the negative effects of this metabolic disease, diabetes is still presently incurable. The incidence of diabetes in Canada has been expanding at an epidemic rate and the negative complications associated with diabetes pose severe risks to the health of individuals (Pelletier et al., 2012; Bilandzic and Rosella, 2017). EGFR and INSR are both receptor-tyrosine kinases that activate many shared pathways like those involving PI3K and ERK MAPK’s. The activation of similar signalling responses by these two receptors partially accounts for why insulin signalling molecules, like insulin receptor substrate-2 (IRS2), are also associated with the pathology of cancers, such as prostate cancer and non-small cell lung cancer, and why drugs  90 used to treat diabetes, like metformin, have found new applications in treatment of cancers (Gonzalez-Angulo et al., 2010; Heckman-Stoddard, 2017; Piper et al., 2019). The interplay between of EGF-and insulin-related signalling pathways have a common area of study (Cui et al., 2006; Ochnik and Baxter, 2016; Voudouri et al., 2015). Continuing the effort to defining the similarities and differences of these related signalling systems can help us understand the level of crosstalk between the two networks and provide insights into why EGF is more growth oriented, while insulin is more metabolic focused, despite the activation of many of the same key pathways by these growth factors. Improved understanding of the interconnectivity of the two signalling axes can also yield insight into how manipulation of the EGFR signalling pathway could have greater impacts on diabetes treatment, and how knowledge of INSR signalling systems might be exploited for improved cancer therapy.   5.2 KAM-900P analysis of EGF-signalling in A431, A549, DU145, HeLa, and LNCaP cells KAM-900P antibody microarrays were initially used to examine EGF signalling networks of a variety of cancer cell lines. These lysates were all fragmented by chemical cleavage to best preserve labile phosphorylation sites at the time of cell lysate homogenization prior to their labelling and processing onto the antibody microarray. The rationale for conducting these experiments was to identify a core set of signalling proteins that were more common for mediation of the actions of EGF, and those that are unique to different cancer cell lines. Every cell line was expected to differ in the expression of many proteins, and this might be expected to alter signal transduction from the same growth factor receptor. The cell lines that I examined included overnight serum-starved human epidermoid carcinoma A431 cells, lung epithelial  91 carcinoma A549 cells, and cervical epithelial adenocarcinoma HeLa cells (all treated without and with 100 ng/mLof EGF from 1-10 min), and in addition prostate epithelial metastatic carcinoma DU145 and LNCaP cells (both treated with and without 20-25 ng/mL of EGF for 5 min). The A431, A549, and HeLa cell lines were chosen for this investigation, because of previous studies that characterized the overexpression of EGFR in A431 cells, as well as the EGFR-positive, but lesser expression of the growth factor receptor in A549 and HeLa cells (Zhang et al., 2005). Utilizing cell lines that would respond to EGF in different intensities was anticipated to permit further understanding of the dynamics of EGFR signalling in greater detail. Each cell line was incubated with EGF for either 1, 5, or 10 min. The data for every time point was averaged together and compared with the data from the control cells that were not incubated with EGF. This was done to collect data representative of the earliest EGF signalling responses such as phosphorylation of EGFR, as well as later phosphorylation events downstream of EGFR activation, such as activation of ERK1 and ERK2. The DU145 and LNCaP cell lines received a different treatment concentration of EGF, because these experiments were performed for comparison with other treatments in those cells for another study. The investigations of EGF signalling in DU145 and LNCaP cells were part of a larger study that evaluated the signalling similarities and differences between EGF, hepatocyte growth factor, and semaphorin 3C in these prostate cancer cell lines, which was an investigation that was led Dr. Christopher Ong. Increased expression levels of the secreted protein ligand, semaphorin 3C, is able to activate RTK’s including EGFR, ErbB2, as well as c-Met, and is also associated with castration-resistant prostate cancer.    92 A breakdown of the results of the KAM-900P antibody microarray experiments can be found in Table 5.1. Out of the 611 phosphosite-specific antibodies printed on the KAM-900P, only antibodies reporting a signal strength of 250 relative units or higher were deemed as detecting a suitable substrate. Overall, all 5 cell lines displayed strong antibody performance with 83-96% of phosphosite-specific antibodies indicating suitable signals, and the A549 and HeLa cell systems had the highest percentages of antibodies capturing labelled substrates. The LNCaP cell experiments revealed the most antibodies that demonstrated increases in protein phosphorylation in response to EGF treatment, while the HeLa cell system showed the most antibodies that exhibited protein dephosphorylation in response to treatment. Student T tests were used to determine p values of leads generated from the averaged A431, A549 and HeLa cell line data. The number of target protein phosphosites that were increased or decreased in phosphorylation in response to EGF treatment that were unique or common to the 5 different cell lines are shown in Figure 5.1. A listing of the specific target phosphosites found to be changed in response to EGF treatment for each cell line can be found in Appendix D.   The KAM-900P antibody microarray permitted uniform analysis of the same antibody probes between all five cell lines. Despite expected differences in protein expression, most of these antibody probes were sensitive enough to detect suitable substrates. Despite successful tracking of over 500 protein phosphosites on each cell line, comparison of the results revealed that the phosphosites found to be changed the most in response to EGF treatment were fairly unique to each cell line, and little was shared. This finding indicates that while there most likely is a canonical EGF signalling pathway that is somewhat universal to every cell, each cancer cell  93 line potentially has individual adaptations to EGFR signalling and may also indicate that the EGF signalling mechanisms of different cells in the body may also be fairly variable and tissue-specific.   Table 5.1. Kinex™ KAM-900P antibodies revealing increased or decreased phosphorylation in response to EGF treatment.   A431 A549 DU145 HeLa LNCaP Total number ofhosphosite-specific antibodies on KAM-900P 611 Number of phosphosite-specific antibodies with appreciable signals 528 571 508 584 507 % of antibodies where %CFC ≥ 40 with EGF treatment 7.8% (1.3%) 6.1% (0.7%) 2.4% 3.9% (0.2%) 30.6% % of antibodies where %CFC ≤ -40 with EGF treatment 3.4% (0.9%) 6.0% (0.5%) 1.2% 8.4% (5.5%) 2.4%  The data represented here for the A431 cell experiments represents pooled data from 9 microarray comparisons. Pooled data from 3 microarray comparisons each are represented for A549 and HeLa cell experiments for calculation of the percent change from the non-treated control (%CFC). Data for DU145 and LNCaP cells are from 1 microarray comparison each. Percentages in parenthesis denote antibodies with a p value of 0.05 or lesser. Signal strengths of 250 relative units or higher on the antibody microarray were regarded as appreciable. Background signals were usually less than 50. 94  Figure 5.1. Number of Kinex™ KAM-900P antibody target protein phosphorylation sites phosphorylated or dephosphorylated in response to EGF treatment. A listing of the target phosphorylation sites, and Uniprot IDs for each cell line can be found in Appendix D.  95 5.3  KAM-1325 analysis of EGF-signalling in A431, A549, and HeLa cells Following the production and availability of the KAM-1325 antibody microarray, which features greater coverage with 1350 antibodies printed to allow duplicate measurements per sample, I repeated the analysis of A431, A549 and HeLa cell lysates of cells treated with 100 ng/mL EGF for 5 min using the newer antibody microarray. These lysates were also fragmented by chemical cleavage to best preserve labile phosphorylation sites at the time of cell lysate homogenization prior to their labelling and processing onto the antibody microarray. Table 5.2 summarizes the number of phospho-site specific antibodies that revealed increased and decreased phosphorylation in response to EGF incubation with the A431, A549, and HeLa cells. Similar to the KAM-900P experiments, 87-89% of antibodies reported signal intensities of 250 relative units or higher for all three cell lines. The A431 cell line was found to have the most phosphosite-specific antibodies that revealed increases of 40% or greater over the control. While the HeLa cell system resulted in the most phosphosite-specific antibodies that showed decreases in phosphorylation with the KAM-900P chip, the A549 cell system exhibited the most reductions in phosphorylation on the KAM-1325 microarray. Since the KAM-900P experiments averaged data from 1-10 min of EGF treatment for each cell line, many of the decreases in phosphorylation in response to 10 min of EGF treatment with the HeLa cells were not evident in the KAM-1325 experiments, which consisted of data for only 5 min of EGF treatment. Between the KAM-1325 and KAM-900P antibody microarrays, 522 phosphosite-specific antibodies were common to both platforms. Among the antibodies that generated leads using the more extensive KAM-1325 array, about 61-71% of these antibodies were also tested with the KAM-900P, and 10-12% of these antibodies reproduced changes of ±40% or more using both arrays.   96 Table 5.2. Kinex™ KAM-1325 antibodies revealing increased or decreased phosphorylation in response to EGF treatment.   A431 A549 HeLa Total number of phosphosite-specific antibodies on KAM-1325 892 Number of phosphosite-specific antibodies with appreciable signals 774 782 796 % of antibodies where %CFC ≥ 40 with EGF treatment  9.3% 4.7% 3.8% % of antibodies where %CFC ≤ -40 with EGF treatment 1.7% 4.2% 1.0% Phosphosite-specific antibodies common to KAM-900P and KAM-1325 522 Total antibody leads from KAM-1325  85 70 38 Number of KAM-1325 lead antibodies that are also on the KAM-900P 52 49 27 % of common leads identified by both KAM-900P and KAM-1325 10% 12% 11%  The data represented here for the A431 and A549 cell experiments represents data from one KAM-1325 microarray comparison. Pooled data from four KAM-1325 microarray comparisons are represented here for the HeLa cell experiments. An appreciable antibody signal is defined as 250 relative units or higher on the antibody microarray. Leads were defined as antibodies that reported changes of ±40% or more.   A two-tailed paired Student’s T test was performed on the averaged data from four biological replicate experiments performed using HeLa cells. Out of the 792 phosphosite-specific antibodies that reported appreciable signals from the averaged HeLa cell data, 128 reported statistically significant changes. Of these, 27 were increases of 40% or greater, while 6 were decreases of -40% or more. These antibodies are listed in Table 5.3. Any of these antibodies  97 that were also found to also be altered by ±40% or greater in the A431 and A549 experiments are also indicated.   Table 5.3. Kinex™ KAM-1325 target phosphosites in HeLa cells significantly increased or decreased in phosphorylation in response to EGF incubation.  Target Name P-Site Antibody Codes UniProt ID Average %CFC HeLa Starved vs EGF HeLa T Test p value Effect of EGF in A431 Effect of EGF in A549 Effect of EGF in HeLa Adducin a/g (ADD3) S726 PN003-PN004 Q9UEY8 55 0.043 + + + BLK Y389 PK543 P51451 51 0.001 +  + CLNE T395 PN191 P24864 43 0.013 +  + CrkL Y251 PN507 P46109 -40 0.045   - DDR2 (Tyro10) Y740 PK594 Q16832 59 0.026 +  + EGFR (ErbB1) Y1172 PK601 P00533 49 0.000   + ERK1 (MAPK3) T202 +Y204 PK621 P27361 43 0.000 +  + FAK (PTK2) Y397 PK627 Q05397 51 0.000   + FAK (PTK2) Y576+ Y577 PK151 Q05397 44 0.007 + - + FBPase 2 (FBP2) Y216 PN700 O00757 63 0.002 +  + G6PD Y503+ Y507 PN701 P11413 41 0.005   + H2AFX S139 PN036 P16104 -66 0.001  - - HSP27 (HSP28; HSPB1) S78 PN041 P04792 83 0.003 + + + IGF1R Y1280 PK152 P08069 59 0.001   + IGF1R Y1346 PK658 P08069 45 0.045   + IRS1 S639 PN118 P35568 -56 0.012 - - - JAK1 Y1022 PK126 P23458 -40 0.006  + - JNK1 (MAPK8) T183+ Y185 PK035-2 P45983 52 0.031   + Jun (c-Jun) S63 PN213 P05412 44 0.011   + Jun (c-Jun) Y170 PN155 P05412 78 0.016   + KHS1 (MAP4K5; KHS) Y31 PK672 Q9Y4K4 47 0.000   +  98 Target Name P-Site Antibody Codes UniProt ID Average %CFC HeLa Starved vs EGF HeLa T Test p value Effect of EGF in A431 Effect of EGF in A549 Effect of EGF in HeLa Lck Y394 PK149 P06239 68 0.013  + + MEK1 (MKK1, MAP2K1) T386 PK048-1 Q02750 41 0.048   + Met Y1230+ Y1234+ Y1235 PK055-1 P08581 58 0.003   + NMDAR1 S896 PN055-1 Q05586 -42 0.004 - - - p70S6K (S6Ka, RPS6KB1) S447 PK156 P23443 -54 0.001   - PBRM1 S948 PN714 Q86U86 67 0.000   + PLCG2 Y753 PN143 P16885 50 0.032   + PTPN11 (PTP1D; PTP2C; SHP2; SHPTP2; Syp) T542 PP008 Q06124 41 0.005   + Rb T356 PN264 P06400 56 0.005 + + + Rb S608 PN262 P06400 45 0.011 + + + Rb S795 PN263 P06400 45 0.010 + + + Rb T821+ T826 PN265 P06400 75 0.001   +  Pooled data from four KAM-1325 microarray comparisons are represented here for the HeLa cell experiments, and one experiment each for A431 and A549. Positive or negative changes are defined as altered by ±40% or more.   5.4  Western blotting analysis of A431, A549 and HeLa cells treated with EGF In addition to the experiments performed using the two generations of antibody microarrays, extensive Western blotting was also performed using SDS-PAGE sample buffer-solubilized lysates generated from A431, A549 and HeLa cells treated with or without 100 ng/mL EGF. EGFR phosphosite-specific antibodies were used to validate phosphorylation and activation of EGFR following EGF treatment in these cell lines (Figure 5.2). A total of 1239 Western blotting  99 analyses were performed with the three cell lines using phosphosite-specific antibodies. Representative Western blots from these extensive analyses are shown in Figure 5.3.    Figure 5.2. Western blotting with EGFR phosphosite-specific antibodies in A431, A549 and HeLa cells. Western blotting was also performed using SDS-PAGE sample buffer-solubilized lysates generated from A431, A549 and HeLa cells treated with or without 100 ng/ml EGF. The Kinexus antibody codes for these antibodies are shown, and full blots are available for inspection on-line at www.kinexusproducts.ca.   100  Figure 5.3. Western blotting with phosphosite-specific antibodies with lysates from A431, A549 and HeLa cells following EGF treatment. Western blotting was also performed using SDS-PAGE sample buffer-solubilized lysates generated from A431, A549 and HeLa cells treated with or without 100 ng/mL EGF. A total of 1239 Western blots were performed, and the clearest EGF-induced changes in phosphorylation are shown here. The Kinexus antibody codes for these antibodies are provided and full blots are available for inspection on-line at www.kinexusproducts.ca.   101 A summary of the aggregate results of these experiments can be found in Table 5.4. In general, most antibodies revealed increases rather than decreases in phosphorylation of immunoreactive protein bands in the approximate expected sizes of the target proteins following EGF treatment. Consistent with the antibody microarray results, A431 cells had the highest percentage (68%) of phosphosite-specific antibodies that showed increases in phosphorylation, followed by HeLa cells (57%) and A549 cells (43%). A549 cells had reported more increases in phosphorylation from the microarray experiments than HeLa cells, but my focus of only large changes in phosphorylation (%CFC of 40% or greater) may account for the discrepancy between trends resulting from the antibody microarray with the Western blotting data. Similar to the KAM-900P antibody microarray results, HeLa cells were once again found to have the most antibodies that indicated decreases in phosphorylation in response to EGF treatment, followed by A549 cells, then A431 cells. The number of unique or shared protein phosphosites that were phosphorylated or dephosphorylated in response to EGF among A431, A549 and HeLa cell lines are shown in Figure 5.4. A total of 193 lead phosphosite-specific antibodies were identified from EGF treatment of A431, A549 and HeLa cells using the KAM-1325 antibody microarray and a total of 22% of these lead antibodies were tested by Western blotting using SDS-solubilized cell lysate respective of the cell line that the lead was found in. Of the tested lead antibodies, a total of 45% these appeared to successfully pick up a protein band in the appropriate predicted size range of the target protein, and 68% of these antibodies that were successful in tracking what appeared to be the target protein validated the trends indicated by the KAM-1325 antibody microarray.   102 Table 5.4. Phosphosite-specific antibodies revealing increased or decreased phosphorylation in response to EGF treatment by Western blotting.   A431 A549 HeLa Total number of antibodies tested 452 380 397 Number of antibodies with sensitivity for target-sized band 244 309 282 % Antibodies indicating increases in phosphorylation in target-sized band 67.3% 42.3% 57.4% % Antibodies indicating decreases in phosphorylation in target-sized band 5.8% 14.1% 15.7% Leads resulting from KAM-1325 microarray 85 70 38 Lead antibodies   tested by Western blotting 25 10 7 Lead protein targets detectable by Western blotting 13 1 5 Microarray leads validated by Western blotting 9 0 4  Western blotting was also performed using SDS-PAGE sample buffer-solubilized lysates generated from A431, A549 and HeLa cells treated with or without 100 ng/mL EGF. Antibody leads are defined as altered by ±40% or more from the KAM-1325 antibody microarray experiments.  103  Figure 5.4. Number of target proteins and phosphorylation sites phosphorylated or dephosphorylated in response to EGF treatment by Western blotting. A listing of the target phosphorylation sites, and Uniprot IDs for each cell line can be found in Appendix E.  104 The performance of the other non-lead antibodies that were tested by the KAM-1325 antibody microarray experiments were evaluated and summarized in Figure 5.5. While an antibody lead was defined as antibodies reporting ±40% CFC on the microarray, for the purposes of evaluating Western blotting correlation to antibody microarray results, antibodies indicating changes in phosphorylation were defined as ±30% CFC or more. Antibodies reporting % CFC values between -30 and +30 were considered to be unchanging. From a total of 937 Western blotting experiments that used phosphosite-specific antibodies also present on the KAM-1325 antibody microarray, 317 of these detected suitable target protein bands by Western blotting. Immunoblotting experiments with 95 of these antibodies revealed trends that were consistent with the antibody microarray and were either true positive or true negative results. Another set of 222 Western blotting experiments revealed trends that were contrary (i.e. no changes or opposite changes) to those reported by the antibody microarray and were false-positive and false-negative results.   105  Figure 5.5 Analysis of true positive, true negative, false positive and false negative results of the EGF-signalling antibody microarray results validated by Western blotting. Antibodies indicating increases of decreases in phosphorylation were defined as ±30% CFC or more on the KAM-1325 antibody microarray. Data labels within the bands represent the actual number of antibodies in each category.   5.5 KAM-1325 analysis of insulin-signalling in HepG2, HeLa, Jurkat and MCF7 cells To compare EGFR and INSR signalling responses, parallel studies were also performed using chemically cleaved lysates made from overnight serum-starved HeLa, HepG2, Jurkat, and MCF7 cells treated without and with 1 µg/mL insulin for 15 min using the KAM-1325 antibody 1428 134162094121310534630% 20% 40% 60% 80% 100%A431A549HeLa% of total antibodies% of increased targets by microarray validated by Western blot (True Positive)% of decreased targets by microarray validated by Western blot (True Positive)% of unchanged targets by microarray validated by Western blotting (True Negative)% of increased targets by microarray not validated Western blot (False Positive)% of decreased targets by microarray not validated Western blot (False Positive)% of unchanged targets by microarray not validated by Western blotting (FalseNegative) 106 microarray. The HeLa cell system was chosen as a common cell line to test both EGF and insulin for more direct comparison of the two signalling systems and five replicate experiments were performed. The HepG2, Jurkat, and MCF7 cell lines were chosen due to their relatively abundant levels of INSR expression, well documented responsiveness to insulin treatment in cell culture in promoting tumour growth, or the ability to mimic in vivo models of gluconeogenesis (Sefried et al., 2018; Wei et al., 2017; Wang et al., 2019). One KAM-1325 microarray analysis was performed for each of the HepG2, Jurkat and MCF7 cell lines. Out of the 892 phosphosite-specific antibodies that are on the KAM-1325 microarray, about 80-90% yielded signal intensities of 250 relative units or greater and were deemed as appreciable and reliable. Antibodies revealing changes of ±40% change from control (CFC) were considered as leads. A total of 258 leads between the four cell lines were recorded. Among these, the insulin-treated HepG2 cell lysates revealed the most leads that indicated increases, followed by MCF7, Jurkat, and lastly HeLa cells (Table 5.5). The MCF7 cells revealed with the greatest number of antibody leads that reported decreases in phosphorylation, followed by HepG2, Jurkat, and lastly HeLa cells. The HeLa cells exhibited the least number of changes in phosphorylation and this was expected, because the other three cell lines were picked based on their known responsiveness to insulin signalling. However, among the replicate experiments performed with HeLa cells, Student’s T tests were used to identify three leads as decreasing 40% or greater in phosphorylation with statistical significance (p ≤ 0.05), namely Tyr-14 in Caveolin-1, Ser-200 in Myoblast determination protein 1, and Ser-966 in Mitogen-activated protein kinase kinase kinase 5. The shared or unique target protein phosphosites that were differentially  107 phosphorylated or dephosphorylated with insulin treatment among the four cell lines are indicated in Figure 5.6.   Table 5.5. Kinex™ KAM-1325 antibodies revealing increased or decreased phosphorylation in response to insulin treatment.   HepG2 HeLa Jurkat MCF7 Total number of phosphosite-specific antibodies on KAM-1325 892 Number of phosphosite-specific antibodies with appreciable signals 811 798 741 813 % of antibodies reporting %CFC ≥ 40 with insulin treatment 11.3% 0.6% 3.6% 4.9% % of antibodies reporting %CFC ≤ -40 with insulin treatment 2.8% 1.1% 2.8% 5.0%  The data represented here for the HepG2, Jurkat and MCF7 cell experiments represents data from one microarray comparison. Pooled data from five microarray comparisons are represented here for the HeLa cell experiments. An appreciable antibody signal is defined as 250 relative units or higher on the antibody microarray.    108  Figure 5.6. Number of Kinex™ KAM-1325 target protein phosphorylation sites phosphorylated or dephosphorylated in response to insulin treatment. A listing of the target phosphorylation sites, and Uniprot IDs for each cell line can be found in Appendix F.  109 5.6 Western blotting analysis of HepG2, HeLa, Jurkat and MCF7 cells treated with insulin. To confirm that the insulin treatments were working as intended on HeLa cells, which demonstrated relatively small responses to insulin treatment with the antibody microarray, pan- and phosphosite-specific antibodies for the ribosomal protein S6 (RPS6) were used to confirm RPS6 phosphorylation in HeLa cells treated with insulin, and subsequent dephosphorylation when treated with a combination of insulin and rapamycin (Figure 5.7)  Figure 5.7. Western blotting with RPS6 phosphosite-specific antibodies in HeLa cells. Western blotting was performed using SDS-PAGE sample buffer-solubilized lysates generated from HeLa cells treated with 250 nM rapamycin and 1 µg/mL insulin for 15 min.  110 A separate analysis of 1032 Western blots of SDS-PAGE buffer solubilized lysates made from HepG2, HeLa, Jurkat, and MCF7 cells treated without and with 1 µg/mL insulin for 15 min and subsequently probed with diverse cell signalling protein antibodies was performed (Table 5.6). For the Jurkat and MCF7 cells, approximately 35% of phosphosite-specific antibodies indicated increases in phosphorylation in response to insulin treatment, while about 20% of antibodies revealed decreases in phosphorylation. Representative Western blots revealing phosphorylation and dephosphorylation events in Jurkat and MCF7 cells are displayed in Figures 5.8 and 5.9. There were less experiments performed with HepG2 and HeLa cells, because of the milder effect of insulin treatments with HeLa cells and smaller number of changes seen in HeLa cells from the KAM-1325 antibody microarray analysis, as well as the relatively slow growth rate of HepG2 cells in cell culture compared to Jurkat and MCF7 cells. From the 258 antibody microarray leads that were generated between the four cell lines, 37 of these were tested within Western blotting experiments, 9 of which were sensitive towards a protein band of the estimated molecular mass of the target. From this subset, only Kinexus antibody PK804, which targeted Ser-221 of ribosomal S6 protein-serine kinase 1, indicated an increase in phosphorylation in insulin-treated HepG2 cells that was validated by Western blotting. The number of target phosphosites that were found to be commonly or uniquely changed in phosphorylation levels in response to insulin treatments among the four cell lines is visualized in Figure 5.10.     111 Table 5.6. Phosphosite-specific antibodies revealing increased or decreased phosphorylation in response to insulin treatment by Western blotting.   HepG2 HeLa Jurkat MCF7 Total number of antibodies tested 36 5 424 566 Number of antibodies with sensitivity for target-sized protein 11 2 44 47 % Antibodies indicating increases in phosphorylation in target-sized protein 18.2% 100% 34.4% 35.9% % Antibodies indicating decreases in phosphorylation in target-sized protein 36.4% 0% 23% 18.8% Leads reported by the KAM-1325 microarray 115 14 48 81 KAM-1325 leads tested by Western blotting 6 0 2 29 Lead protein targets detectable by Western blotting 1 0 1 7 Microarray leads validated 1 0 0 0  Western blotting was performed using SDS-PAGE sample buffer-solubilized lysates generated from HepG2, HeLa, Jurkat and MCF7 cells treated with or without 1 µg/mL insulin for 15 min. Antibody leads are defined as altered by ±40% or more from the KAM-1325 antibody microarray experiments.  112   Figure 5.8. Western blotting of protein phosphosites in Jurkat and MCF7 cells up-regulated with insulin treatment. Western blotting was performed using SDS-PAGE sample buffer-solubilized lysates generated from Jurkat and MCF7 cells treated with 1 µg/mL insulin for 15 min. Full length images of these Western blots are available for viewing on the www.kinexusproducts.ca website.  113  Figure 5.9. Western blotting of protein phosphosites in MCF7 and Jurkat cells down-regulated with insulin treatment. Western blotting was performed using SDS-PAGE sample buffer-solubilized lysates generated from Jurkat and MCF7 cells treated with 1 µg/mL insulin for 15 min. Full length images of these Western blots are available for inspection on the www.kinexusproducts.ca website. 114  Figure 5.10. Number of target protein phosphorylation sites phosphorylated or dephosphorylated in response to insulin treatment by Western blotting. A listing of the target phosphorylation sites, and Uniprot IDs for each cell line can be found in Appendix G.   115 The performance of the other non-lead antibodies that were tested by the KAM-1325 antibody microarray experiments were evaluated and summarized in Figure 5.11. While an antibody lead was defined as antibodies reporting ±40% CFC on the microarray, for the purposes of evaluating Western blotting correlation to antibody microarray results, antibodies indicating changes in phosphorylation were defined as ±30% CFC or more. Antibodies reporting values in between -30 and +30 were considered to be unchanging. From Western blotting experiments that used a total of 767 phosphosite-specific antibodies that were also present on the KAM-1325 antibody microarray, 104 of these detected suitable target protein bands by Western blotting. A total of 36 antibodies in these Western blotting experiments revealed trends that were consistent with the antibody microarray and were either true positive or true negative results. Another set of 68 antibodies in the Western blotting experiments revealed trends that were contrary (unchanged or changing oppositely) to those reported by the antibody microarray and were false positive and false negative results. The HeLa cell data was omitted from Figure 5.8, but there were only two antibodies that were sensitive for the target protein by Western blotting in HeLa cells, and both were false negatives.    116  Figure 5.11. Analysis of true positive, true negative, false positive and false negative results of the insulin-signalling antibody microarray results validated by Western blotting. Antibodies indicating increases of decreases in phosphorylation were defined as ±30% CFC or more on the KAM-1325 antibody microarray. Data labels within the bands represent the actual number of antibodies in each category.  Overall, the results with the KAM-1325 antibody microarray as well as from Western blotting indicate that while insulin-signalling resulted in the phosphorylation of many protein targets, there were proportionally more dephosphorylation events than indicated by EGF-signalling. This may reveal that the activation of protein phosphatases may play a bigger role in insulin 1111 5151212111426210% 20% 40% 60% 80% 100%HepG2JurkatMCF7% of total antibodies% of increased targets by microarray validated by Western blot (True Positive)% of decreased targets by microarray validated by Western blot (True Positive)% of unchanged targets by microarray validated by Western blotting (True Negative)% of increased targets by microarray not validated Western blot (False Positive)% of decreased targets by microarray not validated Western blot (False Positive)% of unchanged targets by microarray not validated by Western blotting (FalseNegative) 117 signalling network than the EGF signalling network. Like EGF signalling, the insulin signalling fingerprint is also fairly variable and cell-line specific. This kind of biological variance is likely to be another form of regulation in which various cell types and tissues are evolved to behave differently despite being in the same environment within the body. This kind of cell-type specific response may be mediated by differential protein expression, protein compartmentalization, splicing isoforms, or protein phosphorylation as evidenced here, that prompt divergent signalling events and ultimately distinct responses to the same signalling ligands that are circulated in the bloodstream.  5.7  EGFR and INSR in vitro kinase assays for substrates in lysates from HeLa cells and identification by KAM-1325 antibody microarrays The previous studies were performed with living cells that were exposed to insulin and EGF to uncover similarities and differences in cell signalling events. As the receptors for these mitogens were both protein-tyrosine kinases, I explored how different their direct targets could be in lysates that contained the same cellular proteins. To identify direct substrates of EGFR and INSR, 0.5 µg of active recombinant EGFR and INSR were incubated with 1 mM ATP and 100 µg of chemically fragmented proteins in HeLa cell lysates. The resulting lysates were then biotinylated and then captured on KAM-1325 antibody microarrays and compared to control lysates that were also fragmented and incubated with ATP without kinases. Three replicates of the EGFR assay experiments were conducted, while the INSR assay experiment was conducted once. Out of the 355 phosphosite-specific antibodies on the microarray that track at least one tyrosine-phosphosite, 72-74% reported signal intensities of 250 relative units or higher for both  118 the EGFR and INSR incubation experiments (Table 5.7). Among these antibodies, 64% of antibodies showed increases of 40% or greater from the control following EGFR incubation (4.2% were statistically significant with p values of 0.05 or less based on a paired Student T test), and 39% of antibodies revealed INSR induced increases. Using the Kinase Substrate Predictor algorithm on the PhosphoNET database that evaluates 90 potential human protein-tyrosine kinases, I also identified the predicted scores for EGFR or INSR as the upstream kinase for every tyrosine phosphosite that was tracked by phosphotyrosine-specific antibodies on the array. For antibodies tracking more than one tyrosine-phosphosite, the the phosphosite that presented the best score was used. Among the phosphosite-specific antibodies that reported increases of 40% of greater following EGFR incubation, EGFR was predicted to be within the top 10% of upstream kinase candidates for 83% of these phosphosites. For INSR, 50% of phosphosites showing 40% increases or greater following kinase incubation were within the top 10% of predicted upstream kinases. Of the 176 and 100 tyrosine phosphosites that were increased in phosphorylation following EGFR or INSR incubation, respectively, 77 were common to both kinases (Figure 5.12). A listing of these target phosphosites that were identified as candidate substrates of EGFR and INSR can be found in Appendix H.  By comparing the identified substrates of recombinant EGFR and INSR, the majority of strong phosphorylation events were found to be unique to each kinase with a substantial subset that was conserved between the two kinases. It is acknowledged that these substrates identified by in vitro incubation of exogenous recombinant kinases may not match the physiological signalling networks due to various explanations. The kinase assay experiments were performed  119 with fragmented peptides that cannot fully replicate the 3-dimensional structure of phosphosites in proteins. By lysing cell samples with detergent, I also eliminate the physiological occurrence of cell compartmentalization and may have exposed non-regular substrates to the kinase. Ultimately, it is not surprising that the substrate specificity among both kinases are largely shared, but also different, as the two signalling systems are already known to activate the same major signalling proteins while achieving different cellular responses as downstream effects.  Table 5.7. Kinex™ KAM-1325 tyrosine phosphosite-specific antibodies revealing increased phosphorylation following recombinant kinase incubation with fragmented HeLa cell lysate proteins.    EGFR INSR Total number of tyrosine phosphosite-specific antibodies on KAM-1325 355 Number of tyrosine phosphosite-specific antibodies with appreciable signals 262 254 % of antibodies where %CFC ≥ 40 with kinase incubation 64.1% 39.4% % of antibody phosphosites that are predicted in silico to be substrates 82.7% 50.0%  Data for EGFR incubation with fragmented HeLa cell lysate proteins were averaged from three replicate experiments. Data for INSR incubation is representative of a single experiment. A signal strength of 250 relative units or greater was considered to be appreciable. Antibody phosphosites predicted in silico to be substrates are defined as phosphosites where the kinase was predicted to be within the top 10% of upstream kinase candidates by PhosphoNET.    120  Figure 5.12. Number of target protein phosphorylation sites phosphorylated by recombinant EGF receptor or insulin receptor as identified by the Kinex™ KAM-1325 antibody microarray. A detailed listing of the target phosphorylation sites, and Uniprot IDs for the results of each kinase can be found in Appendix H.   5.8  Comparison of EGFR and INSR signalling pathways Based on the KAM-900P and KAM-1325 antibody microarray experiments as well as the Western blotting evidence, Venn diagrams detailing the overlap in leads identified as EGF responsive or insulin responsive across all tested cell lines were created (Figure 5.13). In the case of conflicting leads reported by the different assay methods, priority was given to leads indicated from Western blotting, then leads resulting from the KAM-1325 array, then leads  121 from the KAM-900P. Leads that reported divergent trends in different cell lines were removed. The antibody microarray and Western blotting studies have overall revealed that phosphorylation events are more common than dephosphorylation events in response to EGF or insulin signalling among the phosphosites that could be monitored. The majority of EGF-induced phosphorylation events were unique to EGF, while the majority of insulin-induced events were common with EGF-induced events, with a smaller subset that were unique to insulin signalling. For dephosphorylation events, there was much less similarity between the two signalling systems. These similarities and differences in identified leads are a strong insight into the overlapping yet distinct nature of these signalling systems physiologically and can potentially illustrate cellular outcomes that are unique to each system or are shared based on the known functions of the proteins involved with either signalling network.   Figure 5.13. Protein phosphosite leads identified as EGF or insulin responsive. Protein phosphosite leads reported here were pooled from the KAM-900P and KAM-1325 microarray analysis across all tested cell lines as well as by Western blotting.   122 Based on known kinase-substrate relationships already reported in literature, signalling network KiNetscape maps were created to illustrate the connectivity of protein leads that were generated from this analysis of EGF or insulin treatment of the tested cell lines. These maps detail the possible interconnectivity of kinases and other signalling proteins that were evidently changed as monitored with the phosphosite-specific antibodies used on the microarrays as well as from Western blotting and were predicted to be regulated. Novel substrates that were identified as direct targets of EGFR and INSR from the kinase assay experiments that validated observed increases in phosphorylation when cells were treated with EGF or insulin are also included in these maps. Figure 5.14 depicts the interconnectivity of EGF-responsive lead signalling protein phosphosites, while Figure 5.15 depicts the connective network of insulin-responsive lead signalling protein phosphosites identified from the parallel studies. From the microarray and Western blotting experiments, some protein phosphorylation sites were found to be changing in the same direction in response to both EGF and insulin, while some behaved oppositely between EGF and insulin. Figure 5.16 depicts the connectivity between the proteins that were found to be phosphorylated or dephosphorylated similarly in response to EGF and insulin, and those that were opposite.   123  Figure 5.14. Predicted interconnectivity of EGF-responsive cell signalling proteins. Phosphosites found to be increased in response to EGF are indicated with orange text and decreases are in blue. Arrows represent known kinase-substrate relationships that have been pre-established in literature as well as novel substrate connections identified as potential direct targets of EGFR from kinase assay experiments. Functionally inhibitory phosphorylations are depicted with red lines, stimulatory phosphorylations in green, and unclear effects in grey. Protein-serine/threonine kinases are represented by light blue circle icons, protein-tyrosine  124 kinases in purple circles, protein phosphatases in red circles, transcription factors in orange squares, metabolic proteins in green squares, and structural proteins in brown squares. Phosphosite numberings presented are based on protein sequences predicted by human gene sequences. This map was built from the identification of 162 phosphosite-specific antibodies that were selected as most reliable for EGF-induced in substrate phosphorylation at specific sites. Based on these specific phosphosites, 363 potential kinase-substrate (phosphosite) relationships (KSR) that included these phosphosites were retrieved from the Kinexus KiNector database (www.kinector.ca) of over 21,000 KSR’s. This list was further reduced to 207 KSR’s by consideration of only those entries where EGF appeared to have an effect on the expression level and/or phosphorylation state of the kinase in the KSR or was unidentified. The initial linkages between the proteins were initially first generated using Cytoscape v3.4.0 software and then further manipulated with Adobe Illustrator 2019 software.    125  Figure 5.15. Predicted interconnectivity of insulin-responsive cell signalling proteins. Phosphosites found to be increased in response to insulin are indicated with orange text and decreases in blue. Phosphosite numberings presented are based on protein sequences predicted by human gene sequences. Arrows represent known kinase-substrate relationships that have been pre-established in literature as well as novel substrate connections identified as direct targets of INSR from kinase assay experiments. Proteins icons and connectivity arrows are colourized as described in Figure 5.14. This map was built from the identification of 82 phosphosite-specific antibodies that were selected as most reliable for insulin-induced in substrate phosphorylation at specific sites. Based on these specific phosphosites, 237 potential kinase-substrate (phosphosite) relationships (KSR) that included these phosphosites were retrieved from the Kinexus KiNector database (www.kinector.ca) of over 21,000 KSR’s. This list was further reduced to 75 KSR’s by consideration of only those entries where insulin appeared  126 to have an effect on the expression level and/or phosphorylation state of the kinase in the KSR or was unidentified.  Figure 5.16. Interconnectivity of cell signalling proteins that responded similarly or contrarily to EGF and insulin. Phosphosite numberings presented are based on protein sequence predicted by human gene sequences. Phosphosites becoming increased in phosphorylation are indicated with orange text and those undergoing dephosphorylation are shown in blue. Arrows Y1238+, Y1239+ <ïT183+Y398T183+S184+T191+S439T304S460Y76T232 T232S380+S221+,S380+S380+S380+S380+Y570T160+7ïT160+T160+S236+S235+, S236+      S235+, S236+S227+6ï6ï<ïY1346+Met EGFRAbl1LKB1AMPK_1NEK7UnidentifiedPPP2C`RonMBPRSK1JAK2ERK2Smad3CDK2ERK1JNK1RPS6PKCbPKC_ Akt1p70S6KAurKBTAK1MAPKAPK3BrkPLCb1IGF1RInsRS235+, S236+S235+, S236+S235+, S236+Dok3Common Signalling - EGF and InsulinT356 Y119T298+T341S463Y685+,     Y686+T634S392+T182+T182+S392+ S392+S392+Y284+S392+ PPAT PPP5CNLKMAT1APPP3CCUnidentifiedPKC_PRMT5Tyro3CDK7SIK ERK1PKRLKB1ACK1GSK3`TP53 CK2_1EGF and Insulin Opposite Signalling - EGF EffectT356 Y119T298+T341S463Y685+,     Y686+T634S392+T182+T182+S392+ S392+S392+Y284+S392+ PPAT PPP5CNLKMAT1APPP3CCUnidentifiedPKC_PRMT5Tyro3CDK7SIK ERK1PKRLKB1ACK1GSK3`TP53 CK2_1EGF and Insulin Opposite Signalling - Insulin Effect 127 represent known kinase-substrate relationships that have been pre-established in literature. Proteins icons and connectivity arrows are colourized as described in Figure 5.14. This map was built from the identification of 242 phosphosite-specific antibodies that were selected as most reliable for EGF- and/or insulin-induced in substrate phosphorylation at specific sites. Based on the specific phosphosites that were altered by both treatments, the potential kinase-substrate (phosphosite) relationships (KSR) that included these phosphosites were retrieved from the Kinexus KiNector database (www.kinector.ca). This list was further reduced to 99 KSR’s by consideration of only those entries where insulin and EGF appeared to have either a common or opposite effect on the phosphorylation status of these phosphosites.  5.9  Discussion Overall, these experiments were fairly successful in cataloguing a broad list of phosphosites that were affected by EGF and insulin signalling. I was able to identify specific phosphosites that were activated in both signalling systems, and those that were unique to either. These studies indicate that at the receptor-tyrosine kinase level, there were many substrate targets that were common to EGFR and INSR, but more profound differences emerged between the two signalling networks downstream.   There were some inconsistencies between leads generated from the KAM-900P microarray and the KAM-1325 which in part could be due to differences in protocol (such as a reduced incubation time with the dye-labelled anti-biotin reporter antibody) that had been optimized further by the time the KAM-1325 arrays were developed. Other differences between the results arising from the two microarray platforms and Western blotting may stem from variation in the state of dynamic phosphorylations biologically between different preparations  128 of cell lysates. While cell lysates prepared for the antibody microarray experiments were homogenized in cysteine chemical cleavage buffer to best preserve phosphorylations, the incubation time of the cleavage reaction may still require further optimization. Also, while much more sensitive than Western blotting, the antibody microarray signals are representative of a sum of all reactive protein fragments, whereas target proteins can be separated from cross-reactive proteins of different sizes by Western blotting for greater accuracy. Ideally, more replicate experiments coupled with greater statistical analyses could more accurately filter out true treatment-responsive changes from those generated by chance. However, it should be appreciated that these high antibody microarrays are still expensive consumables, and the experiments described here were already fairly costly to perform, given the number of chips that were used in this overall thesis.   Nonetheless, the use of antibody microarray allowed for consistent interrogation using the same set of antibodies for each model system and demonstrated clear differences in phosphoproteomic responses to EGF and insulin whether or not the intended target proteins were tracked. With this benefit, I was able to compare the EGF and insulin-mediated changes in protein phosphorylation that were consistent or cell-line specific with ease. Due to my strict cut-off criteras for being categorized as leads, there could have been far greater overlap of leads generated between cell lines with the same treatment than indicated. Western blotting of the A431 cell lysates treated with EGF revealed that a very significant proportion of antibodies cross-reacted with a band that was similarly sized to EGFR. While by Western blotting, this would not affect my analysis of the target protein greatly, provided that the target was not  129 predicted to be similarly sized as EGFR, but could have contributed to false positives in the microarray experiments with A431 cells. From extensive Western blotting studies, 1102 cross-reactive non-target-sized bands were observed to be changing in phosphorylation in response to EGF. Similarly, 243 cross-reactive non-target bands were documented to change in phosphorylation in response to insulin treatment. The antibodies that detected these cross-reactive bands may be used to immunoprecipitate these proteins for identification with mass spectrometry in future studies.  Despite confirming the activation of the insulin pathway as evidenced by ribosomal protein S6 phosphorylation, insulin treatment at 1 µg/mL for 15 min of serum-starved HeLa cells did not reveal many large changes in protein phosphorylation on the antibody microarrays. Although a more dramatic response would have been ideal, overstimulation of cells with too much insulin was to be avoided. Insulin-like growth factor receptor could be activated at high concentrations of insulin, and this study was targeted towards INSR-mediated activity. While INSR signalling was ideally to be compared to EGFR signalling, a caveat to this study is that EGFR often dimerizes with other ErbB family receptors such as ErbB2 and ErbB3. The levels of these other ErbB family members might be expected to varying amongst the various cell lines that were tested. The leads observed with EGF stimulation could be partially attributed to the signalling of the other ErbB receptors. I have attempted to solve this problem with direct incubation of recombinant EGFR and INSR with cell lysates. However, since the EGFR preparations were already activated, it is unlikely that these preparations purified from insect cells engineered to overexpress the human protein are perfect homodimers of EGFR, as they could have bound to  130 endogenous insect ErbB members upon their activation and be co-purified with the tagged EGFR.  Nonetheless, the work in this chapter demonstrates the successful application of antibody microarray technology to explore and map out cell signalling events in response to the activation of major receptor tyrosine kinases, and the benefits of such a platform in generating data that is easy to compare and contrast between different datasets.    131 Chapter 6: Effects of Protein-Tyrosine Phosphatase Inhibition of Protein Phosphorylation   6.1  Rationale Protein-tyrosine kinases are amongst the most important regulators of signalling cascades that control physiological processes like cellular reproduction, differentiation, and programmed death in multi-cellular organisms. Abnormal expression of protein-tyrosine kinases or their constitutive activation can lead to neoplastic cell transformation and the growth of tumours and cancers. Protein-tyrosine kinase inhibitors (TKIs) represent a wide class of pharmacological drugs that are used to inhibit malignant growth by inhibiting processes like cellular proliferation and angiogenesis by reducing protein-tyrosine phosphorylation. Given the gravity of the impact that hyper-tyrosine phosphorylation could invoke on a cell, I was interested in what kind of innate cellular responses exist in regulating and mitigating the effects of overactive tyrosine kinases.  In this chapter, the protein-tyrosine phosphatase inhibitors phenylarsine oxide (PAO) and sodium orthovanadate were both used to treat HeLa cells that were growing in culture in serum-supplemented media during the exponential growth phase to induce a build-up of protein-tyrosine phosphorylation. Fragmented lysate proteins from the treated cells were captured onto the KAM-1325 antibody microarray to determine how hyper-tyrosine phosphorylation would affect downstream processes affecting protein phosphorylation. It was  132 rationalized that countermeasures might be evoked in cells to compensate for aberrant protein-tyrosine phosphorylation, in an attempt to restrain the inappropriate signalling that arose from what might appear to be overly active protein-tyrosine kinases. In this vein, it was intriguing to explore how downstream signalling through protein-serine/threonine phosphorylation might also be affected.     6.2  Trends in the number of phosphorylated protein-tyrosine and protein-serine/threonine phosphosites in response to PAO and vanadate  HeLa cells were treated with 25 µg/mL PAO and + 50 µg/mL sodium orthovanadate for 15 or 30 min whilst growing in FBS-supplemented media prior to the capture of their lysate proteins onto KAM-1325 antibody microarrays. Out of the 1350 different antibodies printed on the array, 323 were characterized as phosphotyrosine site-specific, while 538 were specific for serine- and threonine-phosphosites. Antibodies that tracked epitopes containing both tyrosine- and serine/threonine-phosphosites that were juxtapositioned next to or near each other were excluded from this analysis. With the protein-tyrosine phosphatase inhibitor treatment, 11% of tyrosine-phosphosite antibodies displayed increases of 25% CFC or greater within 15 min of PAO and vanadate treatment, which grew to 23% of tyrosine-phosphosite antibodies by 30 min treatment (Table 6.1). This was predicted, since sustained protein-tyrosine phosphatase inhibitor treatment should result in more phosphotyrosine sites remaining phosphorylated. Of the phosphotyrosine-site antibodies that showed a decrease of -25% CFC or more following protein-tyrosine phosphatase inhibitor treatment, 4% were evident with 15 min treatment and 15% by 30 min. This result also indicated that with sustained phosphatase inhibitor exposure, a  133 counter response was activated to reduce phosphorylation of other protein-tyrosine phosphosites where regulation is possible, either by decreasing the activities of protein-tyrosine kinases and/or increasing of the activities of protein-tyrosine phosphatases that were less sensitive to the phosphatase inhibitors that were used. Apart from changes in enzymatic activities of these enzymes, some of these counter responses may have also involved changes in protein expression.   Following sustained treatment using the protein-tyrosine phosphatase inhibitors, a small increase of protein-serine and threonine phosphorylation was also observed. With 15 min of treatment, 13% of serine and threonine phosphosite-specific antibodies exhibited increases of 25% CFC or greater, and this rose to 19% of antibodies by 30 min. However, the percentage of protein-serine/threonine phosphosite-specific antibodies that showed reductions of -25% or more also increased from 5% to 23% from 15 to 30 min of treatment. This revealed that in addition to reducing phosphorylation at phosphotyrosine sites where control was possible, the cell also responded to hyper tyrosine-phosphorylation by down-regulating protein-serine/threonine phosphorylations.         134 Table 6.1. Kinex™ KAM-1325 antibodies revealing increased or decreased protein phosphorylation in response to protein-tyrosine phosphatase inhibition.   pY-sites only (15 min) p-Y-sites only (30 min) pS+pT sites only (15 min) pS+pT sites only (30 min) Total number of antibodies on KAM-1325 323 538 Number of antibodies with appreciable signals 295 300 473 502 % of antibodies reporting %CFC ≥ 25 with treatment 12.5% 25.0% 15.0% 19.9% % of antibodies reporting %CFC ≤ -25 with treatment 4.1% 16.0% 5.7% 24.7%  HeLa cells were treated with 25 µg/mL PAO and + 50 µg/mL sodium orthovanadate for 15 or 30 min while growing in FBS-supplemented media prior to their capture on KAM-1325 antibody microarrays. Data shown here is from one microarray experiment per time point each compared to its own control. An appreciable antibody signal is defined as 250 relative units or higher on the antibody microarray.   6.3  Trends in the stoichiometry of phosphorylated protein-tyrosine and protein-serine/threonine-phosphosites in response to PAO and vanadate  In addition to the number of phosphosites that were phosphorylated or dephosphorylated in response to PAO and vanadate treatment, I was also interested in the intensities of the changes in phosphorylation. Focusing on tyrosine phosphosite-specific antibodies that indicated increases of 25% CFC or more, the average and median %CFCs increased from 15 to 30 min of tyrosine phosphatase inhibitor treatment (Table 6.2) However, the average and median %CFC of the phosphotyrosine site-specific antibodies that indicated decreases in phosphorylation remained similar from 15 to 30 min of treatment. For phosphoserine and phosphothreonine  135 site-specific antibodies that reported increases in phosphorylation, the average and median intensity of change also increased dramatically from 15 to 30 min of treatment. Similar to the phosphotyrosine site-specific antibodies, the average and median %CFCs of phosphoserine and phosphothreonine site-specific antibodies that indicated dephosphorylation events from 15 to 30 min of treatment also did not change dramatically. Taken together, while the number of phosphosites being phosphorylated and the intensity of phosphorylation in response to PAO and vanadate can be increased with sustained treatment, the intensity of phosphosites being dephosphorylated remains consistent, although more phosphosites become dephosphorylated with prolonged treatment.   Table 6.2. Average and median %CFC values of Kinex™ KAM-1325 antibodies in response to PAO and vanadate treatment.    pY sites only (15 min) pY sites only (30 min) pS+pT sites only (15 min) pS+pTsites only (30 min) Average %CFC of all antibodies  7% (±19%) 13% (±54%) 5% (±30%) 238% (±3284%) Average %CFC of antibodies reporting %CFC ≥ 25 with treatment 42% (±20%) 79% (±75%) 51% (±52%) 1338 (±7549%) Average %CFC of antibodies reporting %CFC ≤ -25 with treatment -38% (±16%) -41% (±13%) -41% (±13%) -41% (±13%) Median %CFC of all antibodies  6% 3% 1% -4% Median %CFC of antibodies reporting %CFC ≥ 25 with treatment 36% 57% 38% 63% Median %CFC of antibodies reporting %CFC ≤ -25 with treatment -30% -38% -37% -39%   136 HeLa cell lysates of cells treated with 25 µg/mL PAO and 50 µg/mL sodium orthovanadate for 15 or 30 min. while growing in FBS-supplemented media were captured on KAM-1325 antibody microarrays. The symbol ± denotes standard deviation.   Next, I focused on those antibodies that revealed the greatest increases and decreases in phosphorylation with phosphatase treatment, with the hypothesis that these target proteins could potentially act within common signalling pathways. From the averaged data from both 15 and 30 min time points of treatment, the top 5% of target phosphosites found to be changed the most in either direction are indicated in Table 6.3.   Table 6.3. Top 5% of Kinex™ KAM-1325 phosphosite-specific antibodies that indicated increases or decreases with PAO and vanadate treatment of HeLa cells. Phosphosites phosphorylated in response to PAO and vanadate treatment Phosphosites dephosphorylated in response to PAO and vanadate treatment ATF2 (CRE-BP1) S112 P15336 Abl (Abl1) Y413 P00519 AurKB (Aurora B, AIM-1) S227 Q96GD4 Akt1 (PKBa) S473/S474/S472 P31749 B-Myb (MYBL2) T487 P10244 ALOX5 (5-LO) S272 P09917 BRCA1 S1423 P38398 AR S310 P10275 Caveolin 2 S36 P51636 ASK1 (MAP3K5) S83 Q99683 CDK1 (CDC2) T161 P06493 Bad S75 Q92934 CDK1 (CDC2) Y15 P06493 CaMK2a T286 Q9UQM7 CDK10 (PISSLRE) T196 Q15131 CaMKK1 (CaMKK) S74 Q8N5S9 CLNB1 S147 P14635 Cbl Y674 P22681 DDR2 (Tyro10) Y740 Q16832 Cbl Y700 P22681 EGFR (ErbB1) Y1110 P00533 CFL1 S3 P23528 EGFR (ErbB1) Y1172 P00533 Cip (WAF1; p21) p21(WAF1/CIP1) S146 P38936 eIF4E S209 P06730 Cip (WAF1; p21) p21(WAF1/CIP1) T145 P38936 eIF4G (eIF4G1) S1106 Q04637 CPS2 (CAD) T456 P27708 eIF4G (eIF4G1) S1231 Q04637 eEF2K S366 O00418 FAK (PTK2) Y576+Y577 Q05397 eIF4B S422 P23588 FBPase 2 (FBP2) Y216 O00757 Elk-1 S383 P19419 FGFR3 Y647+Y648 P22607 EphA2 Y772 P29317 Fgr Y208+Y209 P09769 EphB2 Y780 P29323  137 Phosphosites phosphorylated in response to PAO and vanadate treatment Phosphosites dephosphorylated in response to PAO and vanadate treatment Fgr Y412 P09769 ERK1 (MAPK3) S283 P27361 Flt3 (STK1) Y842 P36888 ERK1 (MAPK3) T202+Y204 P27361 FOXO1A (FKHR) S319 Q12778 ERK1 (MAPK3) Y204 P27361 Fyn Y420 P06241 EZR Y146 P15311 GluR1 S849 P42261 Fyn T12 P06241 GUK1 Y53 Q16774 Gab1 Y406 Q13480 IRS1 S312 P35568 GATA3 S369 P23771 IRS1 S639 P35568 GATA4 S262 P43694 Lck Y394 P06239 GYS2 Y45 P54840 MEK2 (MKK2, MAP2K2) T394 P36507 HePTP (PTPN7) S44 P35236 NBN (p95NBS1) S343 O60934 IkBa Y42 P25963 PAK1 (PAKa) S141 Q13153 MEF2A T108 Q02078 PKCd (PRKCD) S645 Q05655 NLK T298 Q9UBE8 RSK1 (RPS6KA1, p90RSK) S221/S227 Q15418 p38d MAPK (MAPK13) T180+Y182 O15264 RSK1 (RPS6KA1, p90RSK) T348 Q15418 PFN1 Y129 P07737 Shc1 (Shc) Y349 P29353 PKCd (PRKCD) S645 Q05655 TBK1 S172 Q9UHD2 PKCd (PRKCD) Y313 Q05655 TNK1 Y277 Q13470 PKCg (PRKCG) T514 P05129 TORC2 S433 Q53ET0 PKCi (PRKCiota) T564 P41743 TP53BP1 T1056 Q12888 PKCt (PRKCQ) S695 Q04759 TRIM28 (TIF1B) Y517 Q13263 PKCt (PRKCQ) T538 Q04759 TrkA (NGFR; NTRK1) Y680+Y681 P04629 Raf1 (RafC) S296 P04049 TSSK3 T168 Q96PN8 RARA S77 P10276 YAP1 T119 P46937 Tau S721 P10636  VEGFR2 (KDR) Y1214 P35968  Averaged data with lysate proteins from HeLa cells treated with 25 µg/mL PAO and 50 µg/mL sodium orthovanadate for 15 or 30 min while growing in FBS-supplemented media prior to their capture on KAM-1325 antibody microarrays are represented here. The UniProt ID codes for these target proteins are also shown.  6.4  Western blotting of HeLa cells treated with PAO and sodium orthovanadate HeLa cells treated with PAO and sodium orthovanadate were also solubilized in SDS-PAGE sample buffer and Western blotted with 492 phosphosite-specific phosphoantibodies, of which  138 193 were specific for only phosphotyrosine sites, and 253 were specific for phosphoserine and/or phosphothreonine sites in target proteins (and 46 targeted mixed sites). From the Western blotting analysis, inhibition of tyrosine phosphatases in HeLa cells were more likely to induce increases in tyrosine phosphorylation, although substantial increases in certain serine and threonine phosphorylation sites were also recorded (Table 6.4). However, serine and threonine phosphosites were more commonly observed to be dephosphorylated following treatment than tyrosine phosphosites, and these trends are in concordance with the KAM-1325 antibody microarray results. Representative Western blots of revealing up- and down-regulation of protein-tyrosine, protein-serine and protein-threonine phosphorylation in response to treatment can be found in Figures 6.1-6.3.  Table 6.4. Phosphoantibodies revealing increases or decreases in phosphorylation in response to PAO and sodium orthovanadate treatment in HeLa cells by Western blotting.    All pYsites only pS+pT sites only Total number of antibodies tested 492 193 253 Number of antibodies with clear sensitivity for target-sized protein 197 94 78 % Antibodies indicating increases in phosphorylation in target-sized protein 73.6% 84.0% 57.7% % Antibodies indicating decreases in phosphorylation in target-sized protein 8.1% 5.3% 14.1%  Western blotting was also performed using SDS-PAGE sample buffer-solubilized lysates generated from HeLa cells treated with 25 µg/mL PAO and 50 µg/mL sodium orthovanadate for 15 min while growing in FBS-supplemented media.  139  Figure 6.1. Western blotting of protein-tyrosine phosphosites up-regulated with PAO and vanadate treatment. Western blotting was also performed using SDS-PAGE sample buffer-solubilized lysates generated from HeLa cells treated with 25 µg/mL PAO and 50 µg/mL sodium  140 orthovanadate for 15 min while growing in FBS-supplemented media. An intermediate lane was removed between the two lanes shown in the generation of these images, but each image was extracted from same Western blot. Full length images of these Western blots are available on the www.kinexusproducts.ca website.   Figure 6.2. Western blotting of protein-serine and threonine phosphosites up-regulated with PAO and vanadate treatment. Western blotting was also performed using SDS-PAGE sample buffer-solubilized lysates generated from HeLa cells treated with 25 µg/mL PAO and 50 µg/mL sodium orthovanadate for 15 min while growing in FBS-supplemented media. An intermediate lane was removed between the two lanes shown in the generation of these images, but each image was extracted from same Western blot. Full length images of these Western blots are available for inspection on the www.kinexusproducts.ca website.   141   Figure 6.3. Western blotting of protein phosphosites down-regulated with PAO and vanadate treatment. Western blotting was also performed using SDS-PAGE sample buffer-solubilized lysates generated from HeLa cells treated with 25 µg/mL PAO and 50 µg/mL sodium orthovanadate for 15 while growing in FBS-supplemented media. An intermediate lane was removed between the two lanes shown in the generation of these images, but each image was extracted from same Western blot. Full length images of these Western blots are available for viewing on the www.kinexusproducts.ca website.   142 There were 147 lead antibodies that showed changes of ±25% or more from the antibody microarray experiments using lysates from HeLa cells treated with PAO and sodium orthovanadate for 15 min. Among the Western blotting experiments performed, 33 were lead antibodies, and 11 were successful in detecting an appropriately sized band predicted to be the target protein. All 11 of these lead antibodies from Western blotting experiments successfully validated the changes indicated by the same antibody on the KAM-1325 microarray. In a broader analysis of the 399 antibodies tested by Western blotting, which were also featured on the KAM-1325 antibody microarray, 171 were successful in detecting a target-sized protein band. About 24% of these antibodies validated the antibody microarray results and were either true-positives or true-negatives (Figure 6.4). The remaining 76% of these antibodies showed changes by Western blotting, but no treatment-related differences on the antibody microarray and were false-negatives. No Western blotting experiments indicated that false-positive results from the microarray were observed.      143   Figure 6.4. Validation of true positive, true negative, false positive and false negative antibody microarray phosphorylation results from lysates of HeLa cells treated with PAO and sodium orthovanadate for 15 min by Western blotting. Antibodies indicating increases of decreases in phosphorylation were defined as ±25% CFC or more on the KAM-1325 antibody microarray. Data labels represent the actual number of antibodies in each category, with the percentage in parenthesis.  The antibody microarray and Western blotting data both indicated substantial reduction of protein-serine and threonine phosphorylation of many proteins following protein-tyrosine phosphatase inhibitor treatment in HeLa cells. Mechanistically, it was hypothesized that enhanced tyrosine phosphorylation of protein-serine and threonine phosphatases could be 11 (6%) 1 (1%)29 (17%)130 (76%)% of increased targets by microarray validated by Western blot (True Positive)% of decreased targets by microarray validated by Western blot (True Positive)% of unchanged targets by microarray validated by Western blotting (True Negative)% of unchanged targets by microarray not validated by Western blotting (FalseNegative) 144 activatory in some cases, and this could account for the marked dephosphorylation of some proteins at serine and threonine phosphorylation sites.   Among the 492 phosphosite-specific antibodies used in the Western blotting analysis, 54 were raised against peptides patterned after phosphosites in human protein phosphatases. Focusing on the data generated by the phosphatase-specific phosphosite-specific antibodies, 17 demonstrated increases in phosphorylation in HeLa cells treated with 15 min of 25 µg/mL PAO and + 50 µg/mL sodium orthovanadate by Western blotting (Table 6.5). Out of these, phosphorylation of Tyr-66 on PTP1B, Tyr-536 and Tyr-564 on PTPN6, and Tyr-546 and Tyr-584 on PTPN11, are all documented on PhosphoNET as sites in protein-tyrosine phosphatases known to be correlated with increased phosphatase activity (Frank et al., 2004, Liu and Chernoff, 1997, Sha et al., 2013). The phosphorylation of these proteins in response to phosphatase inhibitors may be part of an innate negative feedback system evolved to increase the enzymatic activity of these protein-tyrosine phosphatases in response to mitigate abnormally high protein-tyrosine phosphorylation.          145 Table 6.5. Increased phosphorylation of sites on protein phosphatases in response to PAO and sodium orthovanadate by Western blotting. Protein phosphatases phosphorylated in response to PAO and Vanadate Treatment Phosphatase Type Functional Effect HePTP T66 P35236 p-Tyr Promotes dissociation from substrate HePTP Y149 P35236 p-Tyr Unknown PPP2CA Y307 P67775 p-Ser/pThr Inhibits phosphatase activity PPP2R5E Y99 Q16537 p-Ser/pThr Unknown PPP3CC S463 P48454 p-Ser/pThr Unknown PPP5C Y119 P53041 p-Ser/pThr Unknown PPPM1A Y362 P35813 p-Ser/pThr Unknown PTP1B Y46 P18031 p-Tyr Unknown PTP1B Y66 P18031 p-Tyr Stimulates phosphatase activity PTPN11 Y546 Q06124 p-Tyr Stimulates phosphatase activity PTPN11 Y62 Q06124 p-Tyr Unknown PTPN11 Y584 Q06124 p-Tyr Stimulates phosphatase activity PTPN2 S304 P17706 p-Tyr Unknown PTPN22 Y499 Q9Y2R2 p-Tyr Unknown PTPN6 Y536 P29350 p-Tyr Stimulates phosphatase activity PTPN6 Y564 P29350 p-Tyr Stimulates phosphatase activity PTPRA Y798 P18433 p-Tyr Reorganizes cytoskeletal organization.  Western blotting was also performed using SDS-PAGE sample buffer-solubilized lysates generated from HeLa cells treated with 25 µg/mL PAO and 50 µg/mL sodium orthovanadate for 15 min while growing in FBS-supplemented media. Descriptions of functional phosphosites were retrieved from PhosphoNET.  6.5 Mapping changes in protein phosphorylation in response to PAO and vanadate treatment in HeLa cells The Western blotting leads that revealed changes in phosphorylation were then pooled with antibody microarray leads that were validated internally by at least 2 different antibodies for  146 the same phosphosite and both indicating changes of ±40% CFC or more in the same direction. Based on the pooled leads, a KiNetscape map illustrating the interconnectivity of the lead protein targets based on known established kinase-substrate relationships from literature was created (Figure 6.5). This map only includes kinases that were regulated by changes in phosphorylation and their substrates and may not include other signalling pathways that exist. A notable observation that became apparent by drawing these connections was the conflicting signalling events relating to the regulation of protein synthesis. Ribosomal protein S6 was found to be dephosphorylated at Ser-235 and 236, both of which facilitate the activation and assembly of the ribosome upon phosphorylation, while many of the kinases that target these sites appear to become phosphorylated at activating sites, such as Thr-497 on protein kinase C alpha, Tyr-313 on protein kinase C-delta, Thr-198 on c-AMP-dependent protein kinase subunit alpha, Ser-221 on RSK1, and Thr-308 on Akt1. Conversely, dephosphorylation of Ser-380 on RSK1 and Ser-473 on Akt1, which are also activatory sites were observed. This may indicate that a phosphatase that targets ribosomal protein S6 and some sites on RSK1 and Akt1 may have been activated in response to hyper-protein-tyrosine phosphorylation to reduce protein synthesis. This may be hypothesized to be an adaptive safety response evolved to inhibit cells from growing out of control by curtailment of protein synthesis when protein-tyrosine kinases that are commonly strong promoters of proliferation are over-active.   147  Figure 6.5. Interconnectivity of proteins changing in phosphorylation in response to hyper-protein-tyrosine phosphorylation due to protein-tyrosine phosphatase inhibition in HeLa cells. Phosphosites found to be increased in phosphorylation in response PAO and orthovanadate treatment for 15 min are indicated with orange text and those that showed reductions in phosphorylation are shown in blue. Phosphosite numberings presented are based on protein sequences predicted by human gene sequences. Arrows represent known kinase-substrate relationships that have been pre-established in literature. Proteins icons and connectivity arrows are colourized as described in Figure 5.14. This map was built from the identification of 159 phosphosite-specific antibodies that were selected as most reliable for PAO- and orthovanadate-induced in substrate phosphorylation at specific sites. Based on these specific phosphosites, 427 potential kinase-substrate (phosphosite) relationships (KSR) that included these phosphosites were retrieved from the Kinexus KiNector database  148 (www.kinector.ca) of over 21,000 KSR’s. This list was further reduced to 163 KSR’s by consideration of only those entries where for PAO- and orthovanadate appeared to have an effect on the expression level and/or phosphorylation state of the kinase in the KSR or was unidentified. The initial linkages between the proteins were initially first generated using Cytoscape v3.4.0 software and then further manipulated with Adobe Illustrator 2019 software.  6.6  Discussion  The microarray experiments of HeLa cells treated with protein-tyrosine phosphatase inhibitors in two time points indicated that with increased tyrosine hyperphosphorylation, some protein phosphatases may become activated in order to control important regulatory phosphosites in signalling pathway proteins. Based on phosphatase substrate promiscuity, protein-serine/threonine phosphatases would seem to have a broader range of potential substrates than do protein-tyrosine-phosphatases. Only about 4% of the number of known human phosphosites are on tyrosine residues, and there are more phosphotyrosine-specific protein phosphatases (i.e. 56) than phosphoserine/threonine-specific protein phosphatases (i.e. 41), and dual-specificity protein phosphatases (i.e. 41). It could be reasoned that protein-serine/threonine phosphatases must be less specific given the smaller ratio of protein phosphatases that target phosphoserine/threonine sites to the number of phosphoserine/threonine sites that exist. The distinct increase in the number of protein-dephosphorylation events as evidenced by the antibody microarray and by Western blotting is in keeping with the hypothesis that protein-tyrosine as well as protein-serine/threonine phosphatases could be activated in a compensatory response to hyper-tyrosine phosphorylation. I was able to identify a few protein-tyrosine phosphatases that appear to be  149 activated by treatment, as suggested by phosphorylation of sites which have been characterized to be activatory, but I could not confirm activation of any protein serine-threonine phosphatases. However, the phosphorylation of some of these phosphatases on tyrosine-sites of unknown function were documented. Many of these phosphosite-specific antibodies were originally created based on homology with known functional sites on related proteins, their evolutionary conservation in diverse species, and on the frequency of these phosphosites being identified as commonly phosphorylated in mass spectrometry studies. Further studies to investigate the functionality of these phosphosites may reveal key mechanistic explanations for why marked reductions on protein-serine and threonine phosphorylation sites are observed following protein-tyrosine-phosphatase inhibition. From the Western blotting studies, hundreds of cross-reactive bands identified as changing in response to PAO and vanadate treatment were detected by 301 phosphosite-specific antibodies. These cross-reactive bands may partially explain some of the inconsistencies between the signals generated by the antibody microarray, which would be representative of the sum of all reactive entities, and the Western blotting with regards with trends observed in target bands. The protein identities of these cross-reactive bands could be worthwhile to follow up in future studies for their proper identification.   These experiments also indicated that not only were there increases in the number, but also in the intensity of phosphorylation of many protein-serine/threonine phosphosites from 15 to 30 min of treatment. The increase in phosphorylated serine and threonine phosphosites may be especially biased towards functional sites that activate protein phosphatases, deactivate  150 protein kinases, or could also be a protective mechanism where hyperphosphorylation of regulatory proteins could potentially facilitate their degradation. The antibody microarray experiments indicated that 16 pan-specific antibodies revealed decreases in protein expression (%CFCs of -25 or more) after 15 min of PAO and vanadate treatment, which escalated to 102 pan-specific antibodies with 30 min of treatment. Taken together, this is supportive of a hypothesis that hyperphosphorylation could be driving protein degradation in order to reduce the functionality of overactive signalling proteins. These experiments have also revealed that hyper-protein-tyrosine phosphorylation appeared to induce the dephosphorylation of ribosomal protein S6 in order to block over-active cells from growing malignantly by limiting protein synthesis or at least critical proteins for cellular growth. Follow up experiments focusing on the identity of these proteins which are down-regulated in expression the most following PAO and vanadate treatment may reveal even more insights into this predicted cellular response. A caveat to these experiments is that stress responses caused by PAO and vanadate are likely to produce other feedback circuts that complicate our analysis of the cause and effect interactions of tyrosine phosphatase inhibition.   Ultimately, the antibody microarray and Western blotting techniques have successfully catalogued the cell signalling systems that may be involved in response to tyrosine hyperphosphorylation as a follow up investigation to the previous investigation of signalling networks downstream of receptor-tyrosine kinases in Chapter 5. The work in this chapter has continued to demonstrate the ability of the antibody microarray as a useful tool in exploring cell signalling systems of low abundance proteins in a targeted nature in high throughput.    151 Chapter 7: Effect of Protein Synthesis Inhibition on Protein Phosphorylation   7.1  Rationale Control of protein turnover via regulation of the rates of protein synthesis and degradation is one of the principal means by which cell signalling can be modulated. The up- or down-regulation of certain signal transduction systems can be achieved not only by increasing the intrinsic functional activity of proteins by phosphorylation, but also by altering the expression of those proteins. Many tumours that arise from over activation of growth pathways often overexpress pro-growth proteins, and generally have higher rates of protein synthesis to maintain their constant state of growth. Pharmacological agents that inhibit protein synthesis like omacetaxine, a protein translation inhibitor, have also been explored as potential treatments to combat growth of tumours and hematological cancers in particular (Gandhi et al., 2014). However, development of this class of drugs has been partially setback by cytotoxic effects that can occur to normal functioning cells (Lindqvist et al., 2012). In this chapter, I hope to expand current scientific knowledge about the consequences of protein synthesis inhibition at the cell signalling level, which could spark new interest and insights into how protein synthesis inhibitors can be made safer for use in therapeutic treatments for diseases. Using HeLa cells treated with protein translation inhibitors such as anisomycin and cycloheximide as an experimental model, I hypothesized that cells would respond to protein synthesis inhibition with changes in the phosphorylation of remaining proteins to retain their functionality and  152 preserve overall homeostasis. In Chapter 6, it was revealed that an inhibition of protein-tyrosine phosphatases in the HeLa cell model system resulted in enhanced tyrosine as well as serine and threonine phosphorylation of many cell signalling proteins. This was accompanied by what appeared to be enhanced degradation of many proteins. Here, I explored how inhibition of protein synthesis might affect their phosphorylation status.  7.2  KAM-1325 antibody microarray analysis of HeLa cells treated with anisomycin and cycloheximide HeLa cells growing in FBS-supplemented media during the exponential expansion phase were treated with either 10 µg/mL anisomycin or cycloheximide for 1-2 h. Following cell lysis with protein fragmentation and biotin-labelling, the samples were captured onto KAM-1325 antibody microarrays for detection. A summary of the number of lead antibodies that reported increases or decreases of ±40% or more are shown in Table 7.1. Surprisingly, with the signal transduction protein focus of the KAM-1325 microarray, there actually appeared to more increases in protein expression than decreases evident with the pan-specific antibodies printed on the microarray. It should be appreciated that signalling proteins are typically expressed at levels that are typically magnitudes lower than structural and metabolic pathway enzymes, so overall protein synthesis could still have been reduced with the protein synthesis inhibitors. Antibodies revealing changes that were common or unique to anisomycin and cycloheximide treatment are shown in Figure 7.1. Anisomycin treatment appeared to induce more profound changes in protein phosphorylation and protein expression than cycloheximide treatment amongst the targets tracked. Nine of the protein targets found to be changing in  153 phosphorylation or expression with anisomycin or cycloheximide treatment were common to both treatments, but most observed changes were unique to each. This indicates that while there is potentially a small subset of protein changes that may constitute a generic response to protein synthesis inhibition, there could also be different additional responses depending on the exact mechanism of inhibition. Focusing on the pan-specific antibodies, the effects on protein expression by both treatments were more generally similar, with 6-8% of pan-specific antibodies reporting increases and 1-3% showing decreases. It should be appreciated that regardless of whether the KAM-1325 antibodies were truly reflective of the phosphorylation and expression levels of their intended targets, there was clearly enhanced protein phosphorylation of many phosphoproteins, and apparently more protein expression increases than decreases. It is evident that despite my treatments with protein synthesis inhibitors, the HeLa cells were able to partially circumvent their effects and increase the expression of what could likely be critical proteins.      154 Table 7.1. Kinex™ KAM-1325 antibodies revealing changes in HeLa cells in response to anisomycin or cycloheximide treatment.    Anisomycin treatment Cycloheximide treatment Total number of phosphosite-specific antibodies on KAM-1325 892 Number of phosphosite-specific antibodies with appreciable signals 781 787 % of phosphosite-specific antibodies reporting %CFC ≥ 40 with treatment 9.1% 3.9% % of phosphosite-specific antibodies reporting %CFC ≤ -40 with treatment 3.3% 1.7% Total number of pan-specific antibodies on KAM-1325 458 Number of pan-specific antibodies with appreciable signals 363 367 % of pan-specific antibodies reporting %CFC ≥ 40 with treatment 7.7% 6.3% % of pan-specific antibodies reporting %CFC ≤ -40 with treatment 2.5% 1.1%  KAM-1325 antibody microarrays were used to analyze fragmented HeLa cell lysates made of cells treated with 10 µg/mL anisomycin or cycloheximide for 2 h. Data from one microarray comparison per treatment are represented here. An appreciable signal on the array were defined as 250 relative units or greater. Background signals on the KAM-1325 antibody microarray were less than 50 relative units.     155   Figure 7.1. Number of protein targets increased or decreased in phosphorylation and expression with anisomycin or cycloheximide treatment of HeLa cells. KAM-1325 antibody microarrays were used to analyze fragmented HeLa cell lysates made of cells treated with 10 µg/mL anisomycin or cycloheximide for 2 h. A detailed listing of all the protein targets represented here along with phosphorylation sites and Uniprot IDs can be found in Appendix I.    156 7.3  Western blotting analysis of HeLa cells treated with anisomycin Western blotting was performed with lysates from HeLa cells that were treated for 1 to 2 h with anisomycin and then directly homogenized into SDS-PAGE sample buffer. Western blotting analyses were completed with a total of 596 different pan- and phosphosite-specific antibodies using these lysates. Examples of target proteins that displayed increases and decreases in signal detection with phosphosite-specific antibodies are provided in Figure 7.2 and Figure 7.3, respectively. In agreeance with trends produced by the KAM-1325 antibody microarray analysis, the Western blotting of treated cells revealed slightly more increases in phosphorylation than decreases (Table 7.2). However, slightly more decreases in protein expression than increases were observed by Western blotting, contrary to the antibody microarray results. This supports the hypothesis that as protein synthesis is inhibited and overall protein expression declines as proteins are being degraded but not replenished, many remaining proteins become enhanced in the stoichiometry of their phosphorylation to potentially increase their functional activity to preserve cellular homeostasis.   Out of the 134 antibody leads identified from the microarray analysis of HeLa cells treated with anisomycin, 24 of these antibodies were also tested by Western blotting. Of these, 6 antibodies detected protein bands that were the approximate size of their respective target proteins, but these particular experiments were not able to validate the trends indicated by the antibody microarray. A full listing of the antibodies that indicated changes in protein phosphorylation and expression by Western blotting in HeLa cells following anisomycin treatment are found in Appendix J.   157  Figure 7.2. Representative Western blotting results from lysates of HeLa cells treated with anisomycin for 1-2 h for increases in protein phosphorylation. Full listing of antibodies that indicated changes can be found in Appendix J. Full images of these Western blots can be viewed on the www.kinexusproducts.ca website.  158  Figure 7.3. Representative Western blotting results from lysates of HeLa cells treated with anisomycin for 1-2 h for decreases in protein phosphorylation and expression. Full listing of antibodies that indicated changes can be found in Appendix J. Full images of these Western blots are available for inspection on the www.kinexusproducts.ca website.     159 Table 7.2. Western blotting leads with HeLa cells treated with anisomycin.    All Phosphosite-specific Pan-specific Total number of antibodies tested 596 536 60 Number of antibodies with clear sensitivity for target-sized protein 153 139 14 % Antibodies indicating increases in phosphorylation or expression in target-sized protein 37.9% 38.8% 28.6% % Antibodies indicating decreases in phosphorylation or expression in target-sized protein 30.7% 30.2% 35.7%  Western blotting was performed with lysates of HeLa cells treated without and with 10 µg/mL anisomycin for 1 to 2 h.   Out of the 596 antibodies tested in Western blotting experiments, 475 of these had been printed on the KAM-1325 antibody microarray. From these, 115 antibodies were successful in detecting an appropriate protein band predicted to be the intended target. Out of these, 31% of the antibodies tested in Western blotting experiments validated the trends in protein phosphorylation and protein expression indicated by the microarray analysis and were true-positives or true-negatives (Figure 7.4). The remaining 69% of the antibodies tested in Western blotting experiments indicated trends that did not confirm those reported by the antibody microarray and appeared to be false-positives or false-negatives. A potential explanation for some of the false-positives and false-negatives indicated by the antibody microarray may have arisen from detection of peptide fragments from partially degraded proteins. While these protein fragments would have been separated from intact proteins following electrophoresis on gels and were therefore not detected by Western blotting, these fragments may have retained intact epitopes recognized by the antibodies on the microarray. This may partially  160 explain why probing with many antibodies revealed decreases in protein expression by Western blotting but reported unchanging signals or even increased signals with the antibody microarray.    Figure 7.4. Validation of true positive, true negative, false positive and false negative antibody microarray results from lysates of HeLa cells treated with anisomycin for 1-2 h by Western blotting. Antibodies indicating increases of decreases were defined as ±30% CFC or 6, 5%2, 2%28, 24%8, 7%7, 6%64, 56%% of increased targets by microarray validated by Western blot (True Positive)% of decreased targets by microarray validated by Western blot (True Positive)% of unchanged targets by microarray validated by Western blotting (True Negative)% of increased targets by microarray not validated Western blot (False Positive)% of decreased targets by microarray not validated Western blot (False Positive)% of unchanged targets by microarray not validated by Western blotting (FalseNegative) 161 more on the KAM-1325 antibody microarray. Data labels represent the actual number of antibodies in each category, followed with the percentage.   7.4  Effect of anisomycin in HeLa cells on protein-serine, protein-threonine, and protein-tyrosine phosphorylation  From the phosphosite-specific antibodies that indicated changes in phosphorylation of ±40% or more from the antibody microarray and those that indicated changes in phosphorylation by Western blotting, I focused on those that were specific for phosphoserine-sites, phosphothreonine-sites, or phosphotyrosine-sites. Antibodies targeting combination neighbouring phosphorylation sites of mixed types were excluded from this analysis. The antibody microarray results indicated that phosphoserine-sites were the most likely to be phosphorylated in HeLa cells after treatment with anisomycin, followed by phosphothreonine-sites, then by phosphotyrosine-sites (Table 7.3). For dephosphorylations, there were no large differences in the percentages between the three phosphosite categories. Western blotting, however, revealed that phosphotyrosine-sites were more likely to be phosphorylated in response to anisomycin treatment over phosphoserine and phosphothreonine-sites, and the latter two were more likely to be dephosphorylated with treatment than phosphotyrosine-sites.        162  Table 7.3. Type of phosphosite increased or decreased in phosphorylation in response to anisomycin treatment in HeLa cells by Kinex™ KAM-1325 analysis and by Western blotting.    pS sites only pT sites only pY sites only Total number of antibodies tested by KAM-1325  337 168 323 Number of antibodies with appreciable signals 281 147 292 Total number of antibodies tested by Western blotting 176 93 230 Number of antibodies with clear sensitivity for target-sized proteins 47 35 42 % of antibodies reporting increases in phosphorylation by antibody microarray 14.2% 8.2% 5.5% % of antibodies reporting decreases in phosphorylation by antibody microarray 3.2% 2.7% 3.8% % of antibodies reporting increases in phosphorylation by Western blotting 27.7% 22.9% 35.7% % of antibodies reporting decreases in phosphorylation by Western blotting 46.8% 42.9% 26.2%  Changes in protein phosphorylation indicated by the KAM-1325 antibody microarray are defined as changes of ±40% or more. Antibodies targeting multiple phosphosites of combination categories were omitted from this analysis.   Based on the changes in protein phosphorylation and expression that were reported by multiple antibodies for the same target using the microarray, or were reported by Western blotting experiments, a KiNetscape map of the interconnectivity of these reported changes is presented in Figure 7.5 based on known kinase-substrate relationships reported in literature. This map only shows the known connections of kinases and substrates that were regulated in phosphorylation, and may not include other signaling pathways that could exist. These maps  163 also show possible kinase-substrate connections that may not reflect true physiological connections in this model system.   Figure 7.5. KiNetscape map of the interconnections between proteins that are altered in protein expression and/or phosphorylation with anisomycin treatment of HeLa cells. Phosphosites found to be increased in phosphorylation in response anisomycin treatment for 1 to 2 h are indicated with orange text, and those that showed reductions in phosphorylation are shown in blue. Increases in protein expression are shown as orange icons, reductions in expression are represented as blue icons, unchanged proteins appear as purple icons, and proteins that were not tracked for expression changes are shown as grey icons. Protein kinases  164 and protein phosphatases appear as circular icons and other classes of proteins are represented as squares. Phosphosite numberings presented are based on protein sequences predicted by human gene sequences. Arrows represent known kinase-substrate relationships that have been pre-established in literature, and activatory phosphorylations appear in green, inhibitory in red and phosphorylation of unknown effect are in grey. This map was initially built from the identification of 181 proteins that revealed altered phosphorylation with phosphosite-specific antibodies and 45 proteins with clear changes in protein expression that was evident with pan-specific antibodies. Based on these specific phosphosites, 243 potential kinase-substrate (phosphosite) relationships (KSR) that included these phosphosites were retrieved from the Kinexus KiNector database (www.kinector.ca) of over 21,000 KSR’s. The initial linkages between the proteins were initially first generated using Cytoscape v3.4.0 software and then further manipulated with Adobe Illustrator 2019 software.  7.5  Discussion Overall, these experiments documented many proteins that were either increased or decreased in expression in HeLa cells in response to protein synthesis inhibition. This reveals that while the inhibitors were effective, there seemed to be signalling systems that have evolved to respond to inhibition and actually increase protein synthesis of important compensating proteins where regulation is possible. This may also be achieved in part by a reduction of degradation of some proteins by proteolysis. The signalling events captured in these experiments would also include stress responses and feedback interactions that were induced by treatment.  Western blotting analysis revealed that protein-tyrosine phosphorylations were slightly preferred following anisomycin treatment, and were more likely to be induced, while protein-serine and threonine phosphosites were more likely to be dephosphorylated. Future studies  165 could attempt to identify the protein-tyrosine kinases that are activated in response to anisomycin treatment, as well as protein-serine and protein-threonine phosphatases that are activated by protein-tyrosine phosphorylation. As protein-tyrosine phosphorylation sites are generally more likely to be activating than protein-serine and protein-threonine phosphorylations, this data is supportive of my hypothesis that remaining proteins would be activated in response to protein synthesis inhibition in order to maintain cellular homeostasis. As proposed in Chapter 6, it was suspected that a major general physiological function of protein-serine and threonine phosphorylation is to facilitate proteins for degradation, and the profound dephosphorylation of a large number of protein-serine and threonine sites observed in these experiments may actually be an attempted response to decrease protein degradation in light of reduced protein synthesis. The larger relationship between protein-serine and threonine phosphorylation and protein degradation would also be an interesting avenue to explore in future studies.   Amongst the most dramatic reductions in protein levels observed following the anisomycin treatments of HeLa cells were reductions in the levels of the protein-serine/threonine kinases ASK1, ICK, LKB1, p70 S6 kinase and Raf1. Interestingly, the phosphorylation of the main target of p70 S6 kinase, the serine phosphorylation sites at the C-terminus of ribosomal protein S6, were observed to be enhanced. This may have been achieved via phosphorylation by RSK isoforms, which also showed enhanced phosphorylation at their own sites of activation, at least at the Ser-380 site of RSK2, which was noted with the KAM-1325 antibody microarray analyses. The phosphorylation of S6 has been proposed to enhance the translation of 7-methyl G capped  166 mRNA’s that are important for protein synthesis (Hutchinson et al., 2011). This may be one of the mechanisms adopted by cells to compensate for inhibition of general protein synthesis.  As shown in Figure 7.5, among the protein kinases that seemed to show increased phosphorylation with anisomycin treatment at activatory sites were Blk, Btk, BubR1, PERK (EIF2AK3), GSK3a, Insulin receptor, MST3, NLK, p38a MAPK, PKCa and SLK. PERK is one of the protein kinases that target phosphorylation of the human eukaryotic translation initiation factor 2-alpha subunit at Ser-49 and Ser-52, which is expected to cause global inhibition of protein synthesis by stabilizing the eIF2A/eIF2B GDP-bound complex and impairing the recycling of eIF2 for successive rounds of eIF2-induced initiation. This seems counter-intuitive as a response to anisomycin-induced inhibition of protein synthesis. Anisomycin is known to bind and inhibit the peptidyl transferase activity of the 60S ribosomal subunit. However, it appears to have other off-targets too (Macias-Silva et al., 2010). The activation of p38 MAPK isoforms in response to anisomycin is expected to induce the phosphorylation of hundreds of other target proteins in cells. While RSK isoforms have not been recognized as typical targets of p38 MAPK’s in the literature, they feature several activatory phosphosites that are predicted in silico as shown on the www.phosphonet.ca website to be substrates for p38 MAPK’s. For example, the activatory Thr-359 and Ser-363 phosphosites in human RSK1 have comparable prediction scores to ERK1 and ERK2 MAP kinases that are well documented to also phosphorylate these sites.  In addition to potentially activated kinases, this study also uncovered several protein kinases that were predicted to have reduced enzymatic activity in response to anisomycin treatment of  167 HeLa cells. This included kinases that had increased phosphorylation of inhibitory phosphosites in protein kinase C isoforms such as Ser-661 of PKCb, Tyr-545 of PKCq, and Ser-205 of PKCµ (PKD1), as well as kinases that had reductions in the phosphorylation of activatory phosphosites, including Ser-435 of ATR, Thr-500 of PKCb, Thr-444 and Ser-447 of p70 S6K, Thr-172 of AMPKa2, Thr-383 of Chk2, Thr-170 of CDK7, Tyr-185 of JNK1, Ser-171 of NEK2 and Thr-161 of MLTK.   Our knowledge of the architecture of the phosphorylation networks that operate to maintain cellular operations under conditions during which protein synthesis is restricted should benefit from these studies. However, understanding how these cell signalling changes precisely permit compensatory responses to protein synthesis inhibition will require more information about the interconnections of kinases, phosphatases and their phosphoprotein substrates. In the next chapter, I will describe meta-analyses of the results from cell lysates from over 250 diverse treatments that were subjected to KAM-1325 antibody microarray analysis to forge new pathway connections.    168 Chapter 8: Meta-analyses of Diverse Treatments in Hela Cells for Mapping Protein Phosphorylation Networks   8.1  Rationale One of the greatest benefits of using the antibody microarray platform as the primary investigative tool for my thesis work was the ability to assay with a large and consistent set of antibodies for both expression and phosphorylation changes in hundreds of different signalling proteins in diverse model systems. I hypothesized that by evaluating the relative performance of the same antibodies within a sufficiently large number of related studies, that correlations between the antibody targets could reveal connective relationships, such as between protein kinases and specific phosphosites on other proteins, and that these relationships could be used to construct predictive maps of the signalling architecture inside of cells. To evaluate if these correlations could be true direct kinase-substrate connections, or just co-regulated in the same pathway or perhaps artefacts of antibody cross-reactivities, I determined whether these proposed connections were already described in the scientific literature. If not, I also investigated if these correlative connections were strongly predicted, based on the PhosphoNET kinase-substrate predictor algorithm for every potential substrate phosphosite, and if connections between kinases and correlated phosphosites could be connected in a pathway through intermediate kinases based on pre-established kinase-substrate relationships. Lastly, I looked at the data generated from several in vitro kinase assays that I had performed using active recombinant protein kinase preparations with fragmented HeLa cell and mouse  169 brain lysate proteins using the antibody microarray approach to identify direct kinase-substrates, to determine if I could find more evidence for novel proposed kinase-substrate relationships that had previously not been reported to my knowledge.  8.2  Evaluation of the most reliable antibodies To narrow down the most reliable antibodies that target epitopes on the same protein, Spearman’s rank-order correlations of the %CFC of every pair-wise antibody combination out of the 1350 antibodies on the Kinex™ KAM-1325 antibody microarray from 254 unique treatment-control comparisons were performed (Figure 8.1). For this study, I accepted any Spearman’s coefficient with a p and q value of less or equal to 0.05 to be statistically significant. While the p value is a function of the probability of obtaining a given result observed if the null hypothesis is true (i.e. probability of a false positive), the q value is the probability of a false discovery based on the hits that are accepted to pass a given hypothesis from the accepted p value (Storey and Tibshirani, 2003). With these criteria, 912 antibodies in the array were found to be significantly correlated to another antibody for the same target internally in the array. Thus, these antibodies appeared to cross-validate each other and were likely able to detect their intended target proteins. Nevertheless, every experimental cell model system varies in the expression and phosphorylation states of proteins, so antibody cross-reactivity always remained a possibility that could thwart interpretation of the results. For some of the antibodies, if there was only one antibody on the KAM-1325 array for the intended target protein, these were also eliminated, even if the results with such antibodies were actually reflective of the target protein.  170   Figure 8.1. Heatmap of pair-wise correlations between all antibodies on the Kinex™ KAM-1325 antibody microarrays. Spearman’s rank-order correlation coefficients are represented here and are sorted by hierarchial clustering of correlating antibodies.   Given that my goal was to ultimately look at the relationships between the antibody targets that were demonstrated to be significantly changing in correlation with each other in response to diverse treatments, I then re-analyzed the meta dataset and retained only the antibody rows that had an absolute (both positive and negative) average %CFC value of 15 or more. This step was taken, because within a given antibody microarray experiment with a treatment, it was  171 typical for less than 5% of the tested antibodies to actually display %CFC values that were considered substantial, i.e. ± 30%. A large number of the antibodies pairs would have shown high correlations, simply because they did not demonstrate any marked changes in any of the particular experimental model systems that were tested. Out of the 1350 antibodies on the KAM-1325, 709 were retained by application of this filter. From the dataset of 709 antibodies in the 254 treatments, I further filtered the results to retain only the significant positive correlations for antibody pairings for the same target proteins, revealing a short list of 301 antibody pairings that represented 320 unique antibodies. These were categorized going forwards as the most reliable antibodies, and the listing of them can be found in Appendix K.  8.3  Identification of the best positively correlated protein pairs  Once I had established the antibodies that were predicted to be the most reliable at capturing their intended target proteins, I then pooled all significant correlating antibody pairs of which both antibodies target different proteins, and at least one antibody of the pair was deemed to be reliable. This yielded 57,028 pairing entries. The number of instances in which the same pair of different protein targets (as tagged by their Uniprot IDs) were found to be significantly correlating within this list of entries were recorded, and the average Spearman’s coefficient of all the antibody pairs targeting the same protein pair was calculated for each protein pair. A total of 3886 pairings were identified to have an average Spearman’s coefficient of 0.4 or greater between the two protein targets being tracked (Table 8.1). The amino acid sequences from 7 residues down and upstream from the central phosphoacceptor amino acid residue for all phosphoepitopes that were tracked by phosphosite-specific antibodies were retrieved.   172  Using only HeLa cells that I had treated with 26 different perturbation conditions for this study (Table 2.1), as well as the data of various treatments on HeLa cells that I had performed in other collaborative studies, I repeated the Spearman’s rank correlation analysis to generate the Spearman’s coefficients for all 910, 575 possible pairings between the 1350 antibodies printed on the array, as well as p and q values. Only the antibody pairs with a Spearman’s coefficient of 0.4 or greater, with a p and q value of 0.05 or lesser were retained. Out of the previous short list of 3886 high confidence correlated protein pairs, 52% of these were re-observed from the HeLa cell data (Table 8.1).   Table 8.1. Number of significant correlated antibody pairs in meta dataset and HeLa cell dataset. Spearman's Coefficient of Pair Number of Correlating Pairs  Number of Pairings Re-observed in HeLa Dataset 0.4-0.5 2757 1478 (53.6%) 0.5-0.6 796 421 (53.9%) 0.6-0.7 244 113 (46.3%) 0.7-0.8 63 11 (17.5%) 0.8-0.9 24 6 (25.0 %) 0.9-1 2 0 (0%) 0.4-1 (Total Pairs) 3886 2029 (52.2%)  Spearman’s coefficient indicated here is the average Spearman’s coefficient of all antibody pairings tracking the same protein target pair and were statistically significant. Significant pairs were evaluated as having a p and q value of 0.05 or lesser.      173 8.4  Evaluation of potential kinase-substrate relationships Within the shortlist of 3886 correlating pairs that I had the most confidence in from the meta dataset, I singled out all the pairings of where one is a protein kinase and the other is a phosphosite in another protein. These 1235 antibody pairs representing 208 unique kinases were then compared to a database of current kinase-substrate relationships that are recorded in literature. Of these, only 16 kinase-substrate phosphosite pairs were found to be validated empirically in literature and documented in the PhosphoSite website (Table 8.2).   Table 8.2. High confidence kinase-substrate phosphosite correlated pairs validated by empirical studies.  Kinase Target Kinase Uniprot ID Kinase Type Substrate Target Substrate Phosphosite(s) Substrate Uniprot ID Frequency of Antibody Pairs with Same Kinase-Substrate Correlation Same Pair Spearman Average Abl (Abl1) P00519 PYK PLCG1 Y771 P19174 10 0.508 Akt1 (PKBa) P31749 PSTK ASK1 (MAP3K5) S83 Q99683 4 0.449 Akt1 (PKBa) P31749 PSTK Bad S75 Q92934 4 0.471 Akt1 (PKBa) P31749 PSTK H3F3A S10 P84243 1 0.413 CaMK1a CAMK1 Q14012 PSTK MRLC2 (MLC) S19 P19105 1 0.470 CDK1 (CDC2) P06493 PSTK GFAP S8 P14136 2 0.728 EphA4 P54764 PYK CDK5 Y15 Q00535 2 0.423 EphB1 P54762 PYK STAT3 Y705+T708 P40763 2 0.402 ERK1 (MAPK3) P27361 PSTK CDC5L T385 Q99459 1 0.571 ERK1 (MAPK3) P27361 PSTK STAT1 S727 P42224 1 0.691 IKKb (IkBKB) O14920 PSTK IKKg (NEMO) S376 Q9Y6K9 2 0.608 PKCd (PRKCD) Q05655 PSTK PKD1 (PRKCM S738+S742 Q15139 11 0.461  174 Kinase Target Kinase Uniprot ID Kinase Type Substrate Target Substrate Phosphosite(s) Substrate Uniprot ID Frequency of Antibody Pairs with Same Kinase-Substrate Correlation Same Pair Spearman Average PKCm PRKD1) PKCd (PRKCD) Q05655 PSTK STAT1 S727 P42224 1 0.395 RSK1 (RPS6KA1 p90RSK) Q15418 PSTK MDM2 S166 Q00987 1 0.503 RSK2 (RPS6KA3) P51812 PSTK RPS6 S236 P62753 4 0.416 YSK1 (STK25 SOK1) O00506 PSTK PKCd (PRKCD) S645 Q05655 1 0.681  In addition to being reported previously in literature, each kinase-substrate pair indicated here were found to be correlated by the antibodies that track these targets on the KAM-1325 microarray with a Spearman’s coefficient of 0.4 or greater and p and q value of 0.05 or less.   To evaluate the likelihood of the other as yet unconfirmed substrate-kinase connections, the score and rank of the kinase as being the potential phosphorylating enzyme for every significantly correlated phosphosite were retrieved from the Kinase Substrate Predictor algorithm on PhosphoNET. For antibody phosphoepitopes that contained more than one phosphosite, the highest score out of all tracked phosphosites was evaluated. Out of 1235 kinase-substrate pairs, the corresponding kinases for 254 substrate phosphosite pairs were ranked within the top 50 predicted kinases out of 500 possible kinases. In particular, 149 protein kinases were ranked within the top 25 (top 5%) for their respective substrate phosphosite, and 56 of these kinases ranked in the top 10 (top 2%) (Table 8.3).     175 Table 8.3. Kinase-phosphosite pairs where the correlated kinase was ranked within the top 10 predicted kinases on PhosphoNET. Kinase Target Kinase Uniprot ID Kinase Type Substrate Target Substrate Phosphosite(s) Substrate Uniprot ID PhosphoNet Prediction Kinase Rank Same Pair Spearman Average Abl (Abl1) P00519 PYK PTPN22 Y499 Q9Y2R2 1 0.687 Abl (Abl1) P00519 PYK ROCK2 (ROKa) Y722 O75116 1 0.534 Abl (Abl1) P00519 PYK PLCG1 Y771 P19174 8 0.508 Abl (Abl1) P00519 PYK ACTB Y53 P60709 9 0.396 Abl2 (Arg) P42684 PYK IGF1R Y1161+T1163 P08069 1 0.419 ALK Q9UM73 PYK Abl (Abl1) Y139 P00519 4 0.447 Bmx (Etk) P51813 PYK Btk Y223 Q06187 4 0.498 BRK Q13882 PYK BLK Y187 P51451 4 0.466 BRK Q13882 PYK ACP1 Y133 P24666 5 0.477 BRK Q13882 PYK ACLY Y682 P53396 7 0.716 BRK Q13882 PYK ACTB Y53 P60709 8 0.401 BRK Q13882 PYK A6 (Twinfilin 1) Y309 Q12792 9 0.619 Btk Q06187 PYK ANXA2 Y238 P07355 2 0.460 Btk Q06187 PYK Bmx (Etk) Y40 P51813 4 0.498 Btk Q06187 PYK ROCK2 (ROKa) Y722 O75116 5 0.411 Btk Q06187 PYK A6r (Twinfin 2) Y309 Q6IBS0 6 0.445 CK2a1 (CSNK2A1) P68400 PSTK SRPK1 S222 Q96SB4 2 0.426 EGFR (ErbB1) P00533 PYK CDK1 (CDC2) Y15 P06493 4 0.442 EphA2 P29317 PYK ERK1 (MAPK3) Y204 P27361 3 0.524 EphA2 P29317 PYK ERK1 (MAPK3) T202+Y204 P27361 3 0.524 EphA4 P54764 PYK ERK1 (MAPK3) Y204 P27361 10 0.436 EphA4 P54764 PYK ERK1 (MAPK3) T202+Y204 P27361 10 0.436 EphB2 P29323 PYK ERK1 (MAPK3) Y204 P27361 4 0.538 EphB2 P29323 PYK ERK1 (MAPK3) T202+Y204 P27361 4 0.538 EphB2 P29323 PYK ERK1 (MAPK3) Y204+T207 P27361 4 0.538  176 Kinase Target Kinase Uniprot ID Kinase Type Substrate Target Substrate Phosphosite(s) Substrate Uniprot ID PhosphoNet Prediction Kinase Rank Same Pair Spearman Average EphB3 P54753 PYK ERK1 (MAPK3) Y204 P27361 8 0.512 EphB3 P54753 PYK ERK1 (MAPK3) T202+Y204 P27361 8 0.512 EphB3 P54753 PYK ERK1 (MAPK3) Y204+T207 P27361 8 0.512 ErbB2 (HER2, Neu) P04626 PYK ErbB3 (HER3) Y1328 P21860 3 0.484 ErbB2 (HER2, Neu) P04626 PYK GSK3a Y279 P49840 9 0.443 ERK1 (MAPK3) P27361 PSTK PTPN2 S304 P17706 1 0.654 ERK1 (MAPK3) P27361 PSTK CDC5L T385 Q99459 5 0.571 ERK1 (MAPK3) P27361 PSTK PKCh (PRKCeta) T655 P24723 7 0.783 ERK1 (MAPK3) P27361 PSTK STAT1 S727 P42224 10 0.691 ERK2 (MAPK1) P28482 PSTK FASN (FAS) S207 P49327 6 0.725 Hck P08631 PYK IGF1R Y1161+T1163 P08069 10 0.583 IKKb (IkBKB) O14920 PSTK IKKg (NEMO) S376 Q9Y6K9 5 0.608 IKKe (IkBKE) Q14164 PSTK Nek7 T191+S195 Q8TDX7 8 0.463 PAK1 (PAKa) Q13153 PSTK SIK3 (QSK) T411 Q9Y2K2 2 0.428 Pim2 Q9P1W9 PSTK ERK2 (MAPK1) Y263+S266 P28482 6 0.399 PKCd (PRKCD) Q05655 PSTK PKD1 (PRKCM, PKCm, PRKD1) S738+S742 Q15139 5 0.461 PKCg (PRKCG) P05129 PSTK PKD1 (PRKCM, PKCm, PRKD1) S738+S742 Q15139 9 0.431 PKG1a (PRKG1A) Q13976.1 PSTK SLK S189 Q9H2G2 2 0.638 PKG1a (PRKG1A) Q13976.1 PSTK PKD1 (PRKCM, PKCm, PRKD1) S738+S742 Q15139 3 0.485 PKG1a (PRKG1A) Q13976.1 PSTK HGK (ZC1) T187 O95819 4 0.397 PRKACA (PKA) P17612 PSTK PTPN1 (PTP1B) S50 P18031 10 0.530  177 Kinase Target Kinase Uniprot ID Kinase Type Substrate Target Substrate Phosphosite(s) Substrate Uniprot ID PhosphoNet Prediction Kinase Rank Same Pair Spearman Average RSK1 (RPS6KA1, p90RSK) Q15418 PSTK MDM2 S166 Q00987 6 0.503 SLK Q9H2G2 PSTK SRPK1 S587 Q96SB4 6 0.621 TrkC (NTRK3) Q16288 PYK Syk Y525+Y526 P43405 2 0.468 TrkC (NTRK3) Q16288 PYK Src Y530 P12931 4 0.423 TXK P42681 PYK Mos Y263 P00540 3 0.439 Tyro3 Q06418 PYK TTK Y833+Y836 P33981 3 0.614 Tyro3 Q06418 PYK STAT3 Y705 P40763 6 0.426 Yes P07947 PYK Kit Y730 P10721 7 0.527 Yes P07947 PYK TEC Y519 P42680 10 0.436 Yes P07947 PYK VEGFR2 (KDR) Y1054 P35968 10 0.475  Each kinase-substrate pair were found to be correlated by the antibodies that track these targets on the KAM-1325 microarray with a Spearman’s coefficient of 0.4 or greater and p and q value of 0.05 or less. Candidate upstream kinase rankings for every phosphosite were evaluated using the Kinase-Substrate Predictor algorithm on the PhosphoNET website.   Based on the kinase-substrate connections that are predicted here by correlative analysis, KiNetscape signalling network maps were produced (Figures 8.2). The purpose of these maps was to represent the potential interconnectivity of these kinases with each other and their non-kinase targets, and to summarize my findings visually. These signalling maps could be used in the future to represent possible interactions between leads that are identified in specific experimental model systems that are analyzed with antibody microarrays that feature reliable antibodies against the targets represented on these maps. It should be appreciated that these interactions are not based on established in vitro connections between these signalling  178 proteins, but rather in vivo correlations arising from the meta-analysis of many studies based on treatments of living cells.     179  Figure 8.2. Interconnectivity of kinases and substrates in predicted cell signalling networks.  Predicted connections are represented by dashed lines, and connections that were also confirmed by reports in the scientific literature are in solid lines. Inhibitory interactions are depicted in red lines, stimulatory interactions in green and unclear interactions in grey. Protein-serine/threonine kinases are in light blue circles, protein-tyrosine kinases are in purple circles, protein phosphatases are in red circles, transcription factors are in orange squares, metabolic proteins in green squares, adaptor and scaffolding proteins in yellow squares and structural proteins in brown squares. Phosphosite numberings presented are based on the protein sequences predicted by human gene sequences.   180 I had hypothesized that although direct kinase substrate connections may not exist for some predicted kinase-substrate showing high correlation, these correlated kinase-substrate pairs may be connected through intermediate kinases. Using the Kinector connection database (www.kinector.ca), potential signalling pathways up to 4 degrees of separation using only confirmed kinase-substrate connections reported in literature were queried for every kinase-substrate connection predicted by correlative analysis. The Kinector database revealed possible signalling pathways for 211 pairs where the kinases were ultimately connected to the correlated substrate phosphosites through 4 or less intermediate kinases (Table 8.4) From this list, 21 pairs were connected through a minimum of one intermediate kinase, 69 pairs were connected through a minimum of 2 kinases, 59 pairs were connected through a minimum of 3 kinases, and 55 pairs were connected through a minimum of 4 kinases.    181 Table 8.4. Possible protein pathways connecting kinases and correlated substrate protein phosphosites ordered by substrate.  Kinase Name Kinase Uniprot ID Substrate Name Substrate Phosphosite Substrate Uniprot ID Lowest Degrees of Separation Potential Pathway 1 Potential Pathway 2 Potential Pathway 3 ILK1 Q13418 AAK1 S637 Q2M2I8 3 ILK>Akt1>MST1> NDR1>AAK1     BRK Q13882 Abl1 Y139 P00519 4 Brk>Akt1>COT>Chk1> Hck>Abl1     ASK1 (MAP3K5) Q99683 ACC1 (ACACA)  S80 Q13085 4 ASK1>JNK1>p70S6K> LKB1>AMPKa1> ACACA ASK1>p38a MAPK>MSK1>LKB1>AMPKa1>ACACA ASK1>JNK1> RSK1>LKB1> AMPKa2> ACACA ERK1 (MAPK3)  P27361 ACLY S455 P53396 2 ERK1>MAPKAPK2> Akt1>ACLY ERK1>PKCa>Akt1> ACLY ERK1>CDK2> Chk1>Src>PDPK> Akt3>ACLY ERK2 (MAPK1) P28482 ACLY S455 P53396 2 ERK2>CK2a1>Akt1> ACLY ERK2>PKCa>Akt1> ACLY ERK2> MAPKAPK2> Akt1>ACLY PKCa (PRKCA)  P17252 ACLY S455 P53396 1 PKCa>Akt1>ACLY PKCa>Src>Akt1> ACLY PKCa>Src> PDPK1>Akt3> ACLY PKCh (PRKCeta) P24723 ACLY S455 P53396 3 PKCh>PKD1>Akt1> ACLY     PKCz (PRKCZ)  Q05513 ACLY S455 P53396 2 PKCz>PKD1>Akt1> ACLY     PKD1 (PRKCM PKCm PRKD1)  Q15139 ACLY S455 P53396 1 PKD1>Akt1>ACLY PKD1>ASK1>p38a MAPK>Akt1>ACLY PKD1>ASK1> MKK3>p38a MAPK>Akt1> ACLY TAK1 (MAP3K7) O43318 ACLY S455 P53396 2 TAK1>MKK6>Akt1> ACLY TAK1>MKK6>p38a MAPK>Akt1>ACLY TAK1>MKK6> JNK2> MAPKAPK5> Akt1>ACLY Abl1  P00519 ACP1 Y132 P24666 2 Abl1>PKCd>Src>ACP1 Abl>PDGFRB>Src> ACP1 Abl1>Met>FAK> Src>ACP1 Abl1  P00519 ACP1 Y133 P24666 2 Abl1>Met>Src>ACP1 Abl1>PKCd>Src> ACP1 Abl1>PDGFRB> Src>ACP1 Abl2 (Arg) P42684 ACP1 Y133 P24666 2 Arg>PDGFRB>Src> ACP1 Arg>PDGFRB>PKCd> Src>ACP1 Arg>PDGFRB> FAK>Src>ACP1 Abl2 (Arg)  P42684 ACP1 Y132 P24666 2 Arg>PDGFRB>Src> ACP1 Arg>PDGFRB>PKCd> Src>ACP1 Arg>PDGFRB> FAK>Src>ACP1  182 Kinase Name Kinase Uniprot ID Substrate Name Substrate Phosphosite Substrate Uniprot ID Lowest Degrees of Separation Potential Pathway 1 Potential Pathway 2 Potential Pathway 3 BRK Q13882 ACP1 Y133 P24666 4 Brk>Akt1>COT>CDK1> Src>ACP1 Brk>Akt1>COT>Chk1>Src>ACP1   BRK Q13882 ACP1 Y132 P24666 4 Brk>Akt1>COT>CDK1> Src>ACP1 Brk>Akt1>COT>Chk1>Src>ACP1   Mos P00540 ADD1 (Adducin a) S726 P35611 3 MEK1>ERK2>PKCa> ADD1 MEK1>ERK1>PKCa> ADD1 MEK1>ERK2>CDK2>Chk1>ADD1 ASK1 (MAP3K5) Q99683 Akt1 (PKBa) T450 P31749 2 ASK1>p38a MAPK>Akt1>IKKa ASK1>JNK1>p70S6K>mTOR>IKKa ASK1>p38a MAPK>MAPKAPK2>Akt1>IKKa CDK1 (CDC2) P06493 AKT1S1 (PRAS40) T246 Q96B36 2 CDK1>CK2a1>Akt1> Akt1S1 CDK1>ATR>Akt1> Akt1S1 CDK1>Src>Akt1> Akt1S1 Akt1 (PKBa) P31749 AR S310 P10275 2 Akt1>COT>CDK1>AR     NIK (MAP3K14 NLK LAK1) Q99558 AR S310 P10275 3 NIK>IKKb>COT>CDK1>AR     PKCa (PRKCA)  P17252 AR S310 P10275 3 PKCa>Akt1>COT> CDK1>AR PKCa>IKKb>COT> CDK1>AR PKCa>Src>PKCi> CDK7>CDK5>AR PKCa (PRKCA)  P17252 ASK1 (MAP3K5) S83 Q99683 1 PKCa>Akt1>ASK1 PKCa>PKD1>Akt1> ASK1 PKCa>Src>RSK2> ASK1 CaMK1a (CAMK1) Q14012 ATF2 (CRE BP1)  T71 P15336 4 CaMK1a>CaMKK1> Akt1>COT>Plk1>ATF2 CaMK1a>CaMKK1> Akt1>PAK1>Plk1> ATF2   Bmx (Etk) P51813 Btk Y223 Q06187 2 Bmx>FAK>Src>Btk Bmx>Met>Src>Btk Bmx>Met>Src> MST1>Abl1>Btk Chk1 (CHEK1) O14757 CaMK2a T286 Q9UQM7 3 Chk1>Src>PDPK1>PKCa>CaMK2a Chk1>Src>Ret>ERK1>PKCa>CaMK2a Chk1>Src>IGF1R> PDPK1>PKCa> CaMK2a CDK5 Q00535 CBS S227 P35520 4 CDK5>Src>PDPK1> PKCa>PKG1>CBS     PAK1 (PAKa)  Q13153 CBS S227 P35520 4 PAK1>MEK1>ERK1> PKCa>PKG1>CBS PAK1>MEK1>ERK2> PKCa>PKG1>CBS   CDK5 Q00535 CDC5L T385 Q99459 3 CDK5>Src>Ret>ERK1> CDC5L CDK5>ATM>Abl1> JAK2>ERK1>CDC5L CDK5>Src>Raf1> MEK1>ERK1> CDC5L PKD1 (PRKCM PKCm PRKD1)  Q15139 CDC5L T385 Q99459 4 PKD1>Akt1>COT> MEK1>ERK1>Cdc5L PKD1>Akt1>PAK1> MEK2>ERK1>Cdc5L    183 Kinase Name Kinase Uniprot ID Substrate Name Substrate Phosphosite Substrate Uniprot ID Lowest Degrees of Separation Potential Pathway 1 Potential Pathway 2 Potential Pathway 3 EGFR (ErbB1)  P00533 CDK1 (CDC2) Y15 P06493 2 EGFR>ATM>Chk1> CDK1 EGFR>FAK>Src>CDK1   MEKK1 (MAP3K1)  Q13233 CDK1 (CDC2)  T161 P06493 2 MEKK1>IKKb>COT> CDK1 MEKK1>MEK1> ERK1>Akt1>COT> CDK1 MEKK1>MEK1> ERK2>CDK2> CDK7>CDK1 MKK3 (MAP2K3 MEK3) P46734 CDK1 (CDC2)  T161 P06493 4 MKK3>p38a MAPK>MAPKAPK2> Akt1>COT>CDK1 MKK3>p38a MAPK>MAPKAPK5> Akt1>COT>CDK1   MKK3 (MAP2K3 MEK3)  P46734 CDK1 (CDC2)  Y15 P06493 3 MKK3>p38a MAPK>Akt1>COT> CDK1 MKK3>p38a MAPK>Akt1>COT> Chk1>CDK1   MKK4 (MAP2K4 MEK4)  P45985 CDK1 (CDC2)  T161 P06493 3 MKK4>p38a MAPK>Akt1>COT> CDK1 MKK4>JNK2>MAPKAPK5>Akt1>COT> CDK1 MKK4>p38d MAPK>PKD1>Akt1>COT>CDK1 MKK4 (MAP2K4 MEK4)  P45985 CDK1 (CDC2)  Y15 P06493 3 MKK4>p38a MAPK>Akt1>COT> CDK1 MKK4>p38a MAPK>Akt1>COT> Chk1>CDK1   PAK1 (PAKa)  Q13153 CDK12 (Cdc2L7) T893 Q9NYV4 4 PAK1>MEK1>ERK1/2> CDK2>CDK7>CDK12     Chk1 (CHEK1)  O14757 CDK5 S159 Q00535 3 Chk1>Src>PKCi>CDK7>CDK5     CK2a1 (CSNK2A1)  P68400 Cip1 (WAF1 p21  T145 P38936 2 CK2a1>p70S6K>Akt1> CDKN1A CK2a1>ERK2>PKCa> Akt1>CDKN1A CK2a1>ERK2> PKCa>CDK2>Chk1> CDKN1A p38a MAPK (MAPK14) Q16539 Cip1 (WAF1 p21 T145 P38936 2 p38a MAPK>MAPKAPK2> Akt1>CDKN1A p38a MAPK>MAPKAPK5> Akt1>CDKN1A   Akt1 (PKBa)  P31749 CPS2 (CAD)  T456 P27708 3 Akt1>COT>MEK1> ERK1>CAD Akt1>mTOR>SGK1> ERK2>CAD Akt1>MST1>Abl1>JAK2>ERK1>CAD CDK5 Q00535 CPS2 (CAD) T456 P27708 4 CDK5>ATM>Abl1> JAK2>ERK1>CAD CDK5>Src>Raf1>MEK1>ERK1>CAD CDK5>p70S6K> mTOR>SGK1> ERK2> CAD CK2a1 (CSNK2A1)  P68400 CPS2 (CAD) T456 P27708 1 CK2a1>ERK2>CAD CK2a1>mTOR>ERK2>CAD CK2a1>Akt1>COT>MEK1>ERK1> CAD PKCa (PRKCA)  P17252 CPS2 (CAD) T456 P27708 3 PKCa>Raf1>MEK1> ERK1>CAD PKCa>IKKb>COT>MEK2>ERK1>CAD PKCa>Src>FAK> Ret>ERK1>CAD IKKa (IkBKA)  O15111 CRYAB S45 P02511 3 Ikka>mTOR>SGK1> ERK2>CRYAB      184 Kinase Name Kinase Uniprot ID Substrate Name Substrate Phosphosite Substrate Uniprot ID Lowest Degrees of Separation Potential Pathway 1 Potential Pathway 2 Potential Pathway 3 CDK5 Q00535 Dynamin I S795 Q05193 3 CDK5>Src>PDPK1> PKCa>DYN1 CDK5>Src>Ret>ERK1>PKCa>DYN1 CDK5>Src>Ret> ERK2>PKCa> DYN1 CK2a1 (CSNK2A1)  P68400 Dynamin I S795 Q05193 2 CK2a1>ERK2>PKCa> DYN1     CDK5 Q00535 eEF2K S366 O00418 1 CDK5>p70S6K>eEF2K CDK5>Src>RSK2> eEF2K CDK5>ATM>LKB1>AMPKa1>eEF2K Chk1 (CHEK1)  O14757 eEF2K S366 O00418 2 Chk1>Src>p70S6K> eEF1K Chk1>AurKb>ATM> LKB1>AMPKa1> eEF2K Chk1>Src>Akt1> COT>RSK1> eEF1K MLK3 (MAP3K11) Q16584 eEF2K S366 O00418 1 MLK3>AMPKa1> eEF2K MLK3>MKK4>JNK1> p70S6K>eEF2K MLK3>MKK7>JNK2>RSK1>eEF2K ERK1 (MAPK3)  P27361 eIF2a S52 P05198 1 ERK1>RSK2>eIF2A ERK1>Akt1>PAK1> MEK1>ERK2>eIF2A ERK1>PKCa>PKD1>ASK1>p38a MAPK>eIF2A PKCd (PRKCD)  Q05655 eIF4B S422 P23588 2 PKCd>Src>p70S6K> eIF-4B PKCd>PKD1>Akt1> eIF-4B PKCd>JNK1>RSK1>eIF-4B EphB2 P29323 ERK1 (MAPK3)  Y204 P27361 2 EphB2>FAK>Ret> ERK1 EphB2>Src>PYK2> JAK2>ERK1 EphB2>Src>Plk1>MEK2>ERK1 EphB2 P29323 ERK1 (MAPK3) T202 Y204 P27361 2 EphB2>FAK>Ret> ERK1 EphB2>Src>PYK2> JAK2>ERK1 EphB2>Src>Plk1>MEK2>ERK1 EphB2 P29323 ERK1 (MAPK3) Y204 T207 P27361 2 EphB2>FAK>Ret> ERK1 EphB2>Src>PYK2> JAK2>ERK1 EphB2>Src>Plk1>MEK2>ERK1 PKCh (PRKCeta)  P24723 ERK3 (MAPK6)  S189 Q16659 3 PKCh>PKD1>Akt1> PAK1>ERK3 PKCh>PKD2>PKD1> Akt1>PAK1>ERK3   CDK5 Q00535 ERK4 (MAPK4)  S186 P31152 3 CDK5>ATM>Akt1> PAK1>ERK4 CDK5>Src>Bmx> PAK1>ERK4 CDK5>Src>PDPK1>PAK2>ERK4 PKCa (PRKCA)  P17252 ERK4 (MAPK4) S186 P31152 2 PKCa>Akt1>PAK1> ERK4 PKCa>Src>Bmx> PAK1>ERK4 PKCa>Src>PDPK1>PAK2>ERK4 PKD1 (PRKCM PKCm PRKD1)  Q15139 ERK4 (MAPK4) S186 P31152 2 PDK1>PAK1>ERK4     TAK1 (MAP3K7)  O43318 ERK4 (MAPK4) S186 P31152 4 TAK1>MKK4/6>p38a MAPK>Akt1>PAK1> ERK4     ERK3 (MAPK6)  Q16659 EZH2 T487 Q15910 4 ERK3>MAPKAPK5> Akt1>COT>CDK1> EZH2      185 Kinase Name Kinase Uniprot ID Substrate Name Substrate Phosphosite Substrate Uniprot ID Lowest Degrees of Separation Potential Pathway 1 Potential Pathway 2 Potential Pathway 3 TTK P33981 EZH2 T487 Q15910 4 TTK>Abl1>Met>Src> PKCi>EZH2 TTK>Abl1>PDGFRB> Src>PKCi>EZH2 TTK>Abl1>PKCd> Src>PKCi>EZH2 EphB2 P29323 EZR Y146 P15311 1 EphB2>Src>EZR EphB2>FAK>Src> EZR   Fyn P06241 EZR Y146 P15311 2 Fyn>PKCd>Src>EZR Fyn>FAK>Src>EZR Fyn>Abl1>Met>Src>EZR EphB2 P29323 Gab1 Y406 Q13480 2 EphB2>Src>EGFR> GAB1 EphB2>Src>Bmx> Met>GAB1 EphB2>MST1> Abl1>Met>GAB1 Fyn P06241 Gab1 Y406 Q13480 2 Fyn>Abl1>EGFR> GAB1 Fyn>PKCd>Src> EGFR>GAB1 Fyn>FAK>Src>Met>GAB1 MKK3 (MAP2K3 MEK3)  P46734 GFAP S8 P14136 4 MKK3>p38a MAPK>Akt1>COT> CDK1>GFAP     EphB2 P29323 GSK3a Y279 P49840 3 EphB2>Src>PDPK1> MEK1>GSK3a EphB2>Src>Akt1> COT>MEK2>GSK3a EphB2>Src>Raf1>MEK1>GSK3a Fyn P06241 GSK3a Y279 P49840 3 Fyn>Abl1>Plk1>MEK1>GSK3a Fyn>FAK>Ret> PDPK1>MEK2> GSK3a Fyn>p38a MAPK>Akt1>COT>MEK2>GSK3a MAPKAPK2 (RPS6KC1)  P49137 GSK3b S9 P49841 1 MAPKAPK2>Akt1> GSK3b MAPKAPK2>RSK2> GSK3b MAPKAPK2>Plk1>MEK2>ERK2> PKCa>GSK3b MEK1 (MKK1 MAP2K1)  Q02750 H2B S14 P33778 4 MEK1>ERK1> MAPKAPK2>Akt1> MST1>H2B MEK1>ERK2>CK2a1>Akt1>MST1>H2B MEK1>ERK2>Src> PKCd>H2B CK2a1 (CSNK2A1)  P68400 H3F3A S10 P84243 1 CK2a1>Akt1>H3F3A ERK2>MSK1/2> H3F3A CK2a1>p70S6K> RSK1/2>H3F3A MEK1 (MKK1 MAP2K1)  Q02750 H3F3A S28 P84243 1 MEK1>ERK1>H3F3A MEK1>ERK2>H3F3A   ERK2 (MAPK1)  P28482 HDAC4/5/9 S246 S259 S220 P56524 3 ERK2>MSK1>LKB1> MARK2>HDAC4 ERK2>RSK1>LKB1> MARK2>HDAC4 ERK2>p70S6K> LKB1>MARK2> HDAC4 ERK1 (MAPK3)  P27361 HDAC5 S498 Q9UQL6 2 ERK1>PKCa>PKD3> HDAC5 ERK1>PKCa> CaMK2a>HDAC5 ERK1>PKCa> PKD1>HDAC5 Akt3 (PKBg)  Q9Y243 HePTP (PTPN7)  T66 P35236 4 Akt3>IKKa>mTOR> SGK1>ERK2>HePTP     CaMKK1 (CaMKK)  Q8N5S9 HePTP (PTPN7)  T66 P35236 4 CaMKK1>Akt1>MST1>Abl1>p38a MAPK>HePTP CaMKK1>Akt1> mTOR>SGK1>ERK2>HePTP CaMKK1>AMPKa1>Raf1>MEK1> ERK1>HePTP  186 Kinase Name Kinase Uniprot ID Substrate Name Substrate Phosphosite Substrate Uniprot ID Lowest Degrees of Separation Potential Pathway 1 Potential Pathway 2 Potential Pathway 3 CK2a1 (CSNK2A1)  P68400 HMGCR S872 P04035 2 CK2a1>ERK2>PKCa> HMGCR CK2a1>p70S6K> LKB1>AMPKa1>HMGCR CK2a1>ERK2> MSK1>LKB1> AMPKa1>HMGCR Abl2 (Arg)  P42684 IGF1R Y1161 T1163 P08069 2 Arg>PDGFRB>Src> IGF1R Arg>PDGFRB>PKCd> Src>IGF1R Arg>PDGFRB>FAK>Src>IGF1R CaMK2a Q9UQM7 IGF1R Y1161 T1163 P08069 2 Arg>PDGFRB>Src> IGF1R Arg>PDGFRB>FAK> Src>IGF1R Arg>PDGFRB> PKCd>Src>IGF1R Hck P08631 IGF1R Y1161 T1163 P08069 3 HCK>Abl1>PKCd>Src> IGF1R HCK>Abl1>Met>Src> IGF1R HCK>Abl1> PDGFRB>FAK>Src>IGF1R IKKe (IkBKE)  Q14164 IGF1R Y1161 T1163 P08069 4 IKKe>Akt1>COT>CDK1 or Chk1>Src>IGF1R     IKKe (IkBKE)  Q14164 IKKg (NEMO)  S376 Q9Y6K9 4 IKKe>Akt1>mTOR> SGK1>Ikkb>Ikkg     IKKe (IkBKE)  Q14164 IKKg (NEMO)  S377 Q9Y6K9 4 Ikke>Akt1>COT>MEK1>ERK1  Ikke>Akt1>PAK1> MEK1>ERK1   ILK1 (ILK)  Q13418 IKKg (NEMO)  S376 Q9Y6K9 4 ILK>Akt1>mTOR>SGK1>IKKb>IKKg     ILK1 (ILK) Q13418 IKKg (NEMO)  S377 Q9Y6K9 4 ILK>Akt1>COT>MEK1>ERK1>IKKg ILK>Akt1>COT>MEK2>ERK1>IKKg ILK>Akt1>PAK1> MEK1>ERK1>IKKg ERK1 (MAPK3)  P27361 IL6ST (GP130)  S782 P40189 2 ERK1>PKCa>CaMK2a> gp130     AurKB (Aurora B AIM-1)  Q96GD4 IRS1 S312 P35568 3 AurKB>Plk1>MEK1> ERK1>IRS1 AurKB>ATM>Akt1> IKKa>IRS1 AurKB>ATM>Abl1>PKCd>JNK1>IRS1 Lck P06239 JNK1 (MAPK8)  Y185 P45983 3 Lck>PKCd>PKD1> ASK1>JNK1 Lck>PKCd>Src>Ret> JNK1 Lck>ERK1>Akt1> COT>MKK6>JNK1 ATR Q13535 c-Jun  T91 P05412 4 ATR>Akt1>COT>MKK6>JNK1>Jun ATR>ATM>Abl1> PKCd>JNK1>Jun ATR>Chk1>Src> Ret>JNK1>Jun CDK5 Q00535 c-Jun S63 P05412 1 CDK5>JNK3>Jun CDK5>Src>PKD1> Jun CDK5>Src>Ret> JNK1>Jun CDK5 Q00535 c-Jun T91 T93 P05412 3 CDK5>Src>Ret>JNK1> Jun CDK5>Src>PKCd> JNK1>Jun CDK5>ATM>Abl1>PKCd>JNK1>Jun Ksr1 Q8IVT5 c-Jun S63 P05412 3 Ksr1>Raf1>MEK1> ERK1>Jun Ksr1>Raf1>MEK1> ERK2>Jun   LKB1 (STK11)  Q15831 c-Jun S63 P05412 4 LKB1>AMPKa1> B-Raf>MEK1>ERK1>Jun LKB1>AMPKa2>IKKb>COT>CDK1>Jun LKB1>MELK> ASK1>MKK4> JNK1>Jun  187 Kinase Name Kinase Uniprot ID Substrate Name Substrate Phosphosite Substrate Uniprot ID Lowest Degrees of Separation Potential Pathway 1 Potential Pathway 2 Potential Pathway 3 LKB1 (STK11)  Q15831 c-Jun T91 T93 P05412 3 LKB1>MELK>ASK1> JNK1>Jun LKB1>MELK>ASK1> MKK4>JNK1>Jun LKB1>MELK> ASK1>MKK6> JNK1>Jun ILK1 (ILK)  Q13418 KCND2 (KV4 2)  T607 Q9NZV8 4 ILK>Akt1>COT>MEK1>ERK1>KCND2 ILK>Akt1>PAK1> MEK2>ERK1> KCND2 ILK>Akt1>mTOR> SGK1>ERK2> KCND2 MAPKAPK2 (RPS6KC1)  P49137 KCND2 (KV4 2)  T607 Q9NZV8 3 MAPKAPK2>Plk1> MEK1>ERK1>KCND2 MAPKAPK2>Plk1> MEK2>ERK2> KCND2 MAPKAPK2>Akt1>mTOR>SGK1> ERK2>KCND2 Chk1 (CHEK1)  O14757 LKB1 (STK11)  S428 Q15831 2 Chk1>AurKB>ATM> LKB1 Chk1>Src>P70S6K> LKB1 Chk1>Src>PDPK1>RSK1>LKB1 MAPKAPK2 RPS6KC1  P49137 LKB1 (STK11) S428 Q15831 3 MAPKAPK2>Akt1> mTOR>P70S6K>LKB1 MAPKAPK2>Akt1> COT>RSK1>LKB1 MAPKAPK2>Plk1>MEK1>ERK2> MSK1>LKB1 TTK P33981 LKB1 (STK11) S428 Q15831 3 TTK>Abl1>p38a MAPK>MSK1>LKB1 TTK>Abl1>MST1> AurKB>ATM>LKB1 TTK>Abl1>PKCd> JNK1>RSK1>LKB1 LKB1 (STK11)  Q15831 MAPKAPK2 (RPS6KC1)  T334 P49137 3 LKB1>MELK>ASK1> p38a MAPK> MAPKAPK2 LKB1>AMPKa1>Raf1>MEK1>ERK1> MAPKAPK2 LKB1>AMPKa1>B-Raf>MEK2>ERK2>MAPKAPK2 Raf1 (RafC)  P04049 MAT1A T341 Q00266 3 Raf1>MEK1>ERK1/2> PKCa>MAT1A     MEK1 (MKK1 MAP2K1)  Q02750 MCM2 Y137 S139 P49736 2 MEK1>ERK1>CDK2> MCM2 MEK1>ERK2>CK2a1>MCM2 MEK1>ERK2>CDK2>MCM2 MEK1 (MKK1 MAP2K1) Q02750 MCM2 S40 S41 P49736 2 MEK1>ERK1>CDK2> MCM2 MEK1>ERK2>CK2a1>MCM2 MEK1>ERK2>CDK2>MCM2 MST1 (STK4 Krs2)  Q13043 MCM2 S40 S41 P49736 4 MST1>Abl1>JAK2> ERK1>CDK2>MCM2 MST1>Abl1>JAK2> ERK2>CK2a1>MCM2   PDGFRa P16234 MCM2 Y137 S139 P49736 3 PDGFRA>Src>PKCi> CDK7>MCM2 PDGFRA>Src>Ret> ERK1>CDK2>MCM2 PDGFRA>Src>Ret>ERK2>Ck2a1> MCM2 ILK1  Q13418 MEF2C S396 Q06413 3 ILK>Akt1>COT>CDK1>MEP2C     MST1 (STK4 Krs2)  Q13043 MEK1 (MKK1 MAP2K1)  S222 Q02750 2 MST1>Abl1>Plk1> MEK1 MST1>AurKB>Plk1>MEK1 MST1>Abl1>JAK2>PAK1>MEK1 MST1 (STK4 Krs2) Q13043 MEK1 (MKK1 MAP2K1) S298 Q02750 3 MST1>Abl1>JAK2> PAK1>MEK1 MST1>Abl1>PKCd> Src>PAK2>MEK1 MST1>Abl1>PDK1>Akt1>PAK1> MEK1  188 Kinase Name Kinase Uniprot ID Substrate Name Substrate Phosphosite Substrate Uniprot ID Lowest Degrees of Separation Potential Pathway 1 Potential Pathway 2 Potential Pathway 3 CDK1 (CDC2)  P06493 MKK3 (MAP2K3 MEK3) S218 P46734 3 CDK1>ATR>Akt1>COT>MKK3 CDK1>Src>IKKb>COT>MKK3 CDK1>Src>PKD1> ASK1>MKK3 mTOR (FRAP)  P42345 MKK3 (MAP2K3 MEK3)  S218 P46734 2 mTOR>Akt1>COT> MKK3 mTOR>SGK1>IKKb> COT>MKK3 mTOR>p70S6K> LKB1>MELK1> ASK1>MKK3 PKCe (PRKCE)  Q02156 MKK3 (MAP2K3 MEK3) S218 P46734 2 PKCe>IKKb>COT>MKK3 PKCe>Akt1>COT> MKK3 PKCe>PKD1>ASK1>MKK3 MST1 (STK4 Krs2)  Q13043 MSK2 (RPS6KA4) T687 O75676 2 MST1>Abl1>p38a MAPK>MSK2 MST1>AurKB>Plk1> p38a MAPK>MSK2 MST1>AurKB>Plk1>MEK1>ERK1> MSK2 PAK1 (PAKa)  Q13153 MSK2 (RPS6KA4)  T687 O75676 2 PAK1>MEK1>ERK1> MSK2 PAK1>Raf1>MEK1> ERK2>MSK2 PAK1>Plk1>MEK2>ERK2>MSK2 TTK P33981 MSK2 (RPS6KA4) T687 O75676 2 TTK>Abl1>P38a MAPK>MSK2 TTK>Abl1>JAK1> ERK1>MSK2 TTK>Abl1>PDK1> ASK1>P38a MAPK>MSK2 MKK3 (MAP2K3 MEK3)  P46734 mTOR (FRAP)  S2448 P42345 2 MKK3>p38a MAPK>Akt1>mTOR MKK3>p38a MAPK>MAPKAPK2> Akt1>mTOR MKK3>MEKK3> MKK6>JNK1> p70S6K>mTOR MLK3 (MAP3K11) Q16584 mTOR (FRAP) S2448 P42345 3 MLK3>AMPKa1>PIP5K>Akt1>mTOR MLK3>MKK7>JNK1> p70S6K>mTOR MLK3>MKK4> JNK2>MAPKAPK5>Akt1>mTOR MLK3 (MAP3K11) Q16584 mTOR (FRAP) S2478 S2481 P42345 3 MLK3>AMPKa1>PIP5K>Akt1>mTOR MLK3>MKK3>p38a MAPK>Akt1>mTOR MLK3>MKK7> JNK2>p70S6K> mTOR CLK1/CLK4 P4975/Q9HAZ1 Myc T58 S62 P01106 3 CLK1>Abl1>JAK2> ERK2>Myc CLK1>Abl1>Plk1> MEK1>GSK3a>Myc CLK1>Abl1>Plk1>MEK2>GSK3b> Myc MLK3 (MAP3K11)  Q16584 Myc T58 S62 P01106 2 MLK3>MKK7>JNK3> Myc MLK3>MKK4>JNK3>Myc MLK3>AMPKa1> Raf1>MEK1>ERK2>Myc NIK (MAP3K14 NLK LAK1)  Q99558 Myc T58 S62 P01106 4 NIK>IKKa>mTOR> SGK1>ERK2>Myc NIK>IKKb>COT> MEK1>GSK3a>Myc NIK>IKKb>COT> MKK6>JNK3>Myc PKCh (PRKCeta) P24723 Myc T58 S62 P01106 4 PKCh>PKD1>ASK1> MKK4>JNK3>Myc PKCh>PKD1>ASK1> MKK6>JNK3>Myc PKCh>PKD1> ASK1>MKK7> JNK3>Myc ROCK1 (ROKb)  Q13464 Myc T58 S62 P01106 3 ROCK1>FAK>Ret> ERK2>Myc ROCK1>FAK>Src> Ret>ERK2>Myc    189 Kinase Name Kinase Uniprot ID Substrate Name Substrate Phosphosite Substrate Uniprot ID Lowest Degrees of Separation Potential Pathway 1 Potential Pathway 2 Potential Pathway 3 MKK3 (MAP2K3 MEK3)  P46734 MYPT1 (MBS)  T696 O14974 3 MKK3>p38a MAPK>Akt1>PAK1> MYPT1 MKK3>p38a MAPK>MAPKAPK2> Akt1>PAK1>MYPT1 MKK3>p38a MAPK>MAPKAPK2>Plk1>ROCK2> MYPT1 ASK1 (MAP3K5)  Q99683 NFAT3 (NFATc4)  S168 S170 Q14934 1 ASK1>p38a MAPK>NFAT3 ASK1>MKK3>p38a MAPK>NFAT3 ASK1>JNK1> p70S6K>mTOR> NFAT3 CK2a1 (CSNK2A1)  P68400 NFAT3 (NFATc4) S168 S170 Q14934 2 CK2a1>p70S6K> mTOR>NFAT3 CK2a1>Akt1>mTOR>NFAT3 CK2a1>Akt1>IKKa>mTOR>NFAT3 CLK1/CLK4 P49759/Q9HAZ1 NFAT3 (NFATc4) S168 S170 Q14934 2 CLK1>Abl1>p38a MAPK>NFAT3 CLK1>Abl1>PKD1> ASK1>p38a MAPK>NFAT3 CLK1>Abl1>PKD1>Akt1>mTOR> NFAT3 NIK (MAP3K14 NLK LAK1)  Q99558 NFAT3 (NFATc4) S168 S170 Q14934 2 NIK>IKKa>mTOR> NFAT3 NIK>IKKb>Cot> MKK3>p38a MAPK>NFAT3 NIK>IKKb>Cot> MKK6>p38a MAPK>NFAT3 PKCa (PRKCA)  P17252 NFAT3 (NFATc4) S168 S170 Q14934 2 PKCa>Akt1>mTOR> NFAT3 PKCa>Src>p70S6K> mTOR>NFAT3 PKCa>PKD1> ASK1>p38a MAPK>NFAT3 MST1 (STK4 Krs2)  Q13043 NFkB p50 S337 P19838 4 MST1>Abl1>JAK2> ERK1>PKCa>NFkB MST1>Abl1>JAK2> ERK2>PKCa>NFkB   MELK Q14680 NFKB p65 (Rel A)  S536 Q04206 4 MELK>ASK1>p38a MAPK>Akt1>IKKa> NFkB     MEK5 (MAP2K5 MKK5)  Q13163 NFKB1 (NFkB p105) S932 P19838 4 MEK5>ERK5> MAPKAPK5>Akt1> IKKa>NFkB     MLK3 (MAP3K11)  Q16584 NFKB1 (NFkB p105) S932 P19838 4 MLK3>AMPKa1>PIP5K>Akt1>IKKa MLK3>MKK3>p38A MAPK>Akt1>IKKa MLK3>MKK4> p38a MAPK>Akt1> IKKa MST1 (STK4 Krs2)  Q13043 NFKB1 (NFkB p105) S932 P19838 4 MST1>AurKB>ATM> Akt1>IKKa>NFkB MST1>Abl1>PKCd> Src>IKKb>NFkB MST1>Abl1>PKD1>Akt1>IKKa>NFkB NIK (MAP3K14 NLK LAK1) Q99558 NFKB1 (NFkB p105) S932 P19838 1 NIK>IKKa>NFkB NIK>IKKb>NFkB   MAPKAPK2 (RPS6KC1)  P49137 NMDAR1 S896 Q05586 4 MAPKAPK2>Plk1> MEK1>ERK1>PKCa> NMDAR1 MAPKAPK2>Plk1> MEK2>ERK2>PKCa> NMDAR1    190 Kinase Name Kinase Uniprot ID Substrate Name Substrate Phosphosite Substrate Uniprot ID Lowest Degrees of Separation Potential Pathway 1 Potential Pathway 2 Potential Pathway 3 CDK5 Q00535 NOS3 T495 P29474 3 CDK5>ATM>LKB1> AMPKa1>eNOS CDK5>p70S6K> LKB1>AMPKa1> eNOS CDK5>Src>PDPK1>PKACa>eNOS MLK3 (MAP3K11)  Q16584 NOS3 T495 P29474 1 MLK3>AMPKa1>eNOS     p38a MAPK (MAPK14)  Q16539 p38b MAPK (MAPK11) T180 Y182 Q15759 4 p38a MAPK>MAPKAPK2> Akt1>COT>MKK6> p38b MAPK p38a MAPK>MAPKAPK5> Akt1>COT>MKK6> p38b MAPK   PKCd (PRKCD)  Q05655 p38d MAPK (MAPK13)  T180 Y182 O15264 3 PKCd>PKD1>ASK1> MKK6>p38d MAPK PKCd>PKD1>HPK1>MLK3>MKK4>p38d MAPK PKCd>Src>Akt1> COT>MKK6>p38d MAPK PKD1 (PRKCM PKCm PRKD1)  Q15139 p38d MAPK (MAPK13)  T180 Y182 O15264 2 PKD1>ASK1>MKK6> p38d MAPK PKD1>HPK1>MLK3>MKK4>p38d MAPK PKD1>Akt1>COT>MKK6>p38d MAPK p38d MAPK (MAPK13)  O15264 p53 (TP53)  S392 P04637 3 p38d MAPK>PKD1>Akt1> NuaK1>TP53     ATM Q13315 p70S6K (S6Ka RPS6KB1)  T252 P23443 3 ATM>Chk1>Src>PDPK1>p70S6K ATM>Abl1>PKCd>Src>PDPK1>p70S6K ATM>Chk1>Src> PDPK1>Ret> p70S6K ACK1 (TNK2)  Q07912 PBK Y272 Q96KB5 4 ACK1>Akt1>COT> CDK1>Src>PBK  ACK1>Akt1>COT> CHK1>Src>PBK   CDK5 Q00535 PFKFB2 (PFK2)  S483 O60825 1 CDK5>p70S6K> PFKFB2 CDK5>Src>RSK2> PFKFB2 CDK5>ATM>Akt1>PFKFB2 CK2a1 (CSNK2A1)  P68400 PFKFB2 (PFK2)  S483 O60825 1 CK2a1>Akt1>PFKB2 CK2a1>p70S6K> PFKB2 CK2a1>ERK2> RSK2>PFKB2 PKCa (PRKCA)  P17252 PFKFB2 (PFK2) S483 O60825 1 PKCa>Akt1>PFKFB2 PKCa>Src>p70S6K> PFKFB2 PKCa>Src>PDPK1>Akt3>PFKFB2 RSK1 (RPS6KA1 p90RSK)  Q15418 PFKFB2 (PFK2) S483 O60825 4 RSK1>LKB1>AMPKa1>PIP5K>Akt1>PFKFB2     p38a MAPK (MAPK14) Q16539 PFKFB3 S461 Q16875 1 p38a MAPK>Akt1>PFKFB3 p38a MAPK>MAPKAPK2> PFKFB3 p38a MAPK>MSK1> AMPKa1>PFKFB3 p70S6K (S6Ka RPS6KB1)  P23443 PFKFB3 S461 Q16875 2 p70S6K>mTOR>Akt1> PFKFB3 p70S6K>LKB1> BRSK1>AMPKa1> PFKFB3 p70S6K>mTOR> SGK1>ERK2> PKCa>PFKFB3  191 Kinase Name Kinase Uniprot ID Substrate Name Substrate Phosphosite Substrate Uniprot ID Lowest Degrees of Separation Potential Pathway 1 Potential Pathway 2 Potential Pathway 3 PAK1 (PAKa)  Q13153 PFKFB3 S461 Q16875 3 PAK1>ERK3> MAPKAPK5>Akt1> PFKFB3 PAK1>MEK1>ERK1>MAPKAPK2>PFKFB3 PAK1>Plk1>MEK2>ERK2>PKCa> PFKFB3 PYK2 (PTK2B)  Q14289 PFN1 Y129 P07737 1 PYK2>Src>PFN1 PYK2>FAK>Src> PFN1 PYK2>JAK2>ERK1>PKCa>Src>PFN1 ERK2 (MAPK1)  P28482 PKCd (PRKCD)  T507 Q05655 2 ERK2>PKCa>Src> PKCd ERK2>CDK2>Chk1> Src>PKCd ERK2>CDK2> CDK7>CDK1>Src> PKCd PKD1 (PRKCM PKCm PRKD1)  Q15139 PKCd (PRKCD)  T507 Q05655 4 PKD1>Akt1>COT> CDK1>Src>PKCd PKD1>Akt1>COT> Chk1>Src>PKCd   ERK2 (MAPK1)  P28482 PKCg (PRKCG) T514 P05129 4 ERK2>PKCa>Src> IGF1R>PDPK1>PKCg ERK2>PKCa>Src> Ret>PDPK1>PKCg   PKCa (PRKCA)  P17252 PKCg (PRKCG) T514 P05129 2 PKCa>Src>PDPK1> PKCg PKCa>Src>IGF1R> PDPK1>PKCg PKCa>Src>FAK> Ret>PDPK1>PKCg PKCa (PRKCA) P17252 PKCz (PRKCZ)  T410 Q05513 2 PKCa>Src>PDPK1> PKCz PKCa>Src>IGF1R> PDPK1>PKCz PKCa>Src>Ret> PDPK1>PKCz PKCz (PRKCZ)  Q05513 PKD1 (PRKCM PKCm PRKD1)  S205 Q15139 3 IKKb>COT>MKK6> p38d MAPK>PKD1     PKD1 (PRKCM PKCm PRKD1)  Q15139 PKM2 S37 P14618 1 PKD1>Akt1>PKM2 PKD1>ASK1>p38a MAPK>Akt1>PKM2 PKD1>ASK1>p38a MAPK>MAPKAPK2>Akt1>PKM2 PKD3 (PRKCN)  O94806 PKM2 S37 P14618 2 PKD3>PKD1>Akt1> PKM2 PKD3>PKD1>ASK1> p38a MAPK>Akt1> PKM2   PKCa (PRKCA)  P17252 PRC1 T481 O43663 3 PKCa>Akt1>COT> CDK1>PRC1 PKCa>IKKb>COT> CDK1>PRC1 PKCa>Src>PKCi> CDK7>CDK1> PRC1 CLK1/CLK4 P49759/Q9HAZ1 PTEN S380 P60484 2 CLK1>Abl1>Plk1> PTEN CLK1>Abl1>p38a MAPK>MSK1>LKB1> PTEN CLK1>Abl1>JAK2>ERK2>CK2a1> PTEN p38b MAPK (MAPK11) Q15759 PTPN1 (PTP1B)  S50 P18031 2 p38b MAPK>MAPKAPK5> Akt1>PTP1B     ERK1 (MAPK3) P27361 PTPN2 S304 P17706 1 ERK1>CDK2>PTPN2 ERK1>Akt1>COT> CDK1>PTPN2 ERK1>IKKb>COT> CDK1>PTPN2  192 Kinase Name Kinase Uniprot ID Substrate Name Substrate Phosphosite Substrate Uniprot ID Lowest Degrees of Separation Potential Pathway 1 Potential Pathway 2 Potential Pathway 3 ERK3 (MAPK6)  Q16659 PTPN2 S304 P17706 4 ERK3>MAPKAPK5> Akt1>COT>CDK1> PTPN2     PKCa (PRKCA)  P17252 PTPN2 S304 P17706 3 PKCa>Akt1>COT> CDK1>PTPN2 PKCa>IKKb>COT> CDK1>PTPN2 PKCa>Raf1>MEK1>ERK1>CDK2> PTPN2 TAK1 (MAP3K7)  O43318 PTPN2 S304 P17706 4 TAK1>AMPKa2>IKKb> COT>CDK1>PTPN2     MST1 (STK4 Krs2) Q13043 PU 1 S146 P17947 4 MST1>Abl1>JAK2> ERK2>CK2a1>PU 1     Raf1 (RafC) P04049 PU 1 S146 P17947 3 Raf1>MEK1>ERK2> CK2a1>PU 1     TTK P33981 PU 1 S146 P17947 4 TTK>Abl1>JAK2>ERK2>CK2a1>PU 1     LRRK2 (PARK8)  Q5S007 Raf1 (RafC)  S301 T303 P04049 4 LRRK2>Akt1>COT> MEK1>ERK1>Raf1 LRRK2>Akt1>COT> MEK2>ERK1>Raf1 LRRK2>Akt1>PAK1>MEK1>ERK1> Raf1 CDK1 (CDC2)  P06493 RBM9 iso6 T7 O43251 2 CDK1>CK2a1>ERK2> RBM9 CDK1>Src>Ret>ERK2>RBM9 CDK1>Src>Ret> ERK1>RBM9 Chk1 (CHEK1)  O14757 RIOK2 S332 S335 S337 Q9BVS4 2 Chk1>AurKB>Plk1> RIOK2 Chk1>Src>Plk1> RIOK2 Chk1>Hck>Abl1> Plk1>RIOK2 Raf1 (RafC)  P04049 RIOK2 S332 S335 S337 Q9BVS4 4 Raf1>MEK1>ERK1> MAPKAPK2>Plk1> RIOK2 Raf1>MEK1>ERK2> MAPKAPK2>Plk1> RIOK2   TTK P33981 RIOK2 S332 S335 S337 Q9BVS4 2 TTK>Abl1>Plk1>RIOK2 TTK>Abl1>JAK2> PAK1>Plk1>RIOK2 TTK>Abl1>Mer> Src>Plk1>RIOK2 Abl1  P00519 ROCK2 (ROKa)  Y722 O75116 1 Abl1>Src>ROCK2 Abl1>PKCd>Src> ROCK2 Abl1>Met>Src> ROCK2 CaMK2a Q9UQM7 RPS6 S236 P62753 4 CaMK2a>BRSK1> AMPKa1>PIP5K>Akt1>RPS6     Ksr1 Q8IVT5 RPS6 S236 P62753 4 Ksr1>Raf1>MEK1> ERK1>Akt1>RPS6 Ksr1>Raf1>MEK1> ERK1>DAPK1>RPS6 Ksr1>Raf1>MEK1>ERK1>RSK1> RPS6 PBK Q96KB5 RPS6 S235 S236 P62753 2 PBK>p38a MAPK>Akt1>RPS6 PBK>p38a MAPK>MAPKAPK5> Akt1>RPS6 PBK>p38a MAPK>MAPKAPK2>RSK2>RPS6 Akt3 (PKBg)  Q9Y243 RSK1 (RPS6KA1 p90RSK)  T573 Q15418 4 Akt3>IKKa>mTOR> SGK1>ERK2>RSK1      193 Kinase Name Kinase Uniprot ID Substrate Name Substrate Phosphosite Substrate Uniprot ID Lowest Degrees of Separation Potential Pathway 1 Potential Pathway 2 Potential Pathway 3 IKKb (IkBKB)  O14920 RSK1 (RPS6KA1 p90RSK)  S380 Q15418 2 IKKb>COT>RSK1     Mos P00540 RSK1 (RPS6KA1 p90RSK)  S380 Q15418 2 Mos>MEK1>ERK1> RSK1 Mos>MEK1>ERK2> RSK1   Tyro3 Q06418 Shc1  Y349 Y350 P29353 4 Tyro3>Akt1>MST1> Abl1>EGFR>Shc1 Tyro3>Akt1>COT>CDK1>Src>Shc1 Tyro3>Akt1>COT>Chk1>Src>Shc1 CDK5 Q00535 SIN3A S832 Q96ST3 4 CDK5>Src>Ret>ERK2> CK2a1>SIN3A     PKD1 (PRKCM PKCm PRKD1)  Q15139 SIN3A S832 Q96ST3 4 PKD1>Akt1>COT> CDK1>CK2a1>SIN3A     ERK2 (MAPK1)  P28482 SRPK1 S587 Q96SB4 1 ERK2>CK2a1>SRPK1     CDK5 Q00535 STAT1 S727 P42224 3 CDK5>Src>PKCd> STAT1 CDK5>ATM>Abl1> p38a MAPK>STAT1 CDK5>Src>Ret> ERK1>STAT1 PKCa (PRKCA)  P17252 STAT1 S727 P42224 1 PKCa>CAMK2a> STAT1 PKCa>Src>PKCd> STAT1 PKCa>Raf1>MEK1>ERK1>STAT1 PKD1 (PRKCM PKCm PRKD1) Q15139 STAT1 S727 P42224 2 PKD1>ASK1>JNK1> STAT1 PKD1>ASK1>p38a MAPK>STAT1 PKD1>ASK1> MKK6>p38b MAPK> STAT1 TAK1 (MAP3K7)  O43318 STAT1 S727 P42224 2 TAK1>MKK6>JNK1> STAT1 TAK1>MKK6>p38a MAPK>STAT1 TAK1>MKK6>p38b MAPK>STAT1 p38b MAPK (MAPK11)  Q15759 STAT3 Y705 T708 P40763 3 p38b MAPK>MAPKAPK5> Akt1>PKM2>STAT3     Plk3 (CNK) Q9H4B4 STAT3 Y705 T708 P40763 4 Plk3>Chk2>TTK>Abl1> JAK2>STAT5     Syk P43405 STAT3 Y705 P40763 3 Syk>PKCb1>Akt1> PKM2>STAT3 Syk>PKCb1>PKCa>Src>STAT3   Syk P43405 STAT3 Y705 T708 P40763 3 Syk>PKCb1>Akt1> PKM2>STAT3 Syk>PKCb1>PKCa>Src>STAT3   TTK P33981 STAT3 Y705 P40763 2 TTK>Abl1>JAK2> STAT3 TTK>Abl1>Met>Src> STAT3 TTK>Abl1>p38a MAPK>MAPKAPK2>STAT3  194 Kinase Name Kinase Uniprot ID Substrate Name Substrate Phosphosite Substrate Uniprot ID Lowest Degrees of Separation Potential Pathway 1 Potential Pathway 2 Potential Pathway 3 TTK P33981 STAT3 Y705 T708 P40763 2 TTK>Abl1>JAK2> STAT3 TTK>Abl1>Met>Src> STAT3 TTK>Abl1>p38a MAPK>MAPKAPK2>STAT3 Tyro3 Q06418 STAT3 Y705 P40763 2 Tyro3>Akt1>PKM2> STAT3 Tyro3>Akt1>MST1> Abl1>JAK2>STAT3 Tyro3>Akt1>COT>CDK1>Src>STAT3 IKKe (IkBKE)  Q14164 STAT4 S721 Q14765 4 Ikke>Akt1>MST1>Abl1>p38a MAPK>STAT4 Ikke>Akt1>COT> MKK3 or MKK6>p38a MAPK>STAT4   TTK P33981 STAT5A Y694 P42229 2 TTK>Abl1>Btk> STAT5A TTK>Abl1>JAK2> STAT5A TTK>Abl1>Met> Src>STAT5A CDK1 (CDC2)  P06493 Tau S516 P10636 2 CDK1>CDK7>CDK5> Tau CDK1>Src>Raf1> MEK1>GSK3b>Tau CDK1>Src>Plk11>MEK2>GSK3b>Tau MAPKAPK2 (RPS6KC1)  P49137 Tau T522 P10636 3 MAPKAPK2>Plk1> MEK2>GSK3b>Tau MAPKAPK2>Plk1> MEK1>GSK3b>Tau MAPKAPK2>Akt1>COT>MEK1> GSK3b>Tau PKD1 (PRKCM PKCm PRKD1)  Q15139 TRIM28 (TIF1B)  S473 Q13263 3 PKD1>Akt1>COT> Chk1>TRIM28 PKD1>ASK1>p38a MAPK>MAPKAPK2> TRIM28 PKD1>Akt1>PAK1>Plk1>Chk2> TRIM28 TTK P33981 TRIM28 (TIF1B) S473 Q13263 1 TTK>Chk2>TRIM28 TTK>Abl1>p38a MAPK>MAPKAPK2> TRIM28 TTK>Abl1>Mst>Src>PKCd>TRIM28 TTK P33981 Vav Y174 P15498 1 TTK>Abl1>VAV1     Tyro3 Q06418 Vav Y174 P15498 3 Tyro3>Akt1>MST1> Abl1>VAV1 Tyro3>Akt1>COT> CDK1>Src>VAV1 Tyro3>Akt1>COT>Chk1>Src>VAV1 MKK3 (MAP2K3 MEK3)  P46734 WASP Y291 P42768 4 MKK3>p38a MAPK>Akt1>COT> Abl1>WASP MKK3>p38a MAPK>Akt1>MST1> Abl1>WASP   MKK3 (MAP2K3 MEK3)  P46734 YAP1 S127 P46937 3 MKK3>p38a MAPK>MAPKAPK2/5> Akt1>YAP1     MKK3 (MAP2K3 MEK3)  P46734 YAP1 T119 P46937 3 MKK3>MEKK3>MKK6>JNK1>YAP1 MKK3>MEKK3> MKK7>JNK2>YAP1 MKK3>p38a MAPK>Akt1>COT>CDK1>YAP1 MKK4 (MAP2K4 MEK4)  P45985 YAP1 S127 P46937 2 MKK4>p38a MAPK>Akt1>YAP1 MKK4>JNK1>MST1> NDR1>YAP1 MKK4>JNK2> p70S6K>mTOR> Akt1> YAP1  195 Kinase Name Kinase Uniprot ID Substrate Name Substrate Phosphosite Substrate Uniprot ID Lowest Degrees of Separation Potential Pathway 1 Potential Pathway 2 Potential Pathway 3 PKCe (PRKCE)  Q02156 YAP1 S109 P46937 3 PKCe>Akt1>MST1> NDR1>YAP1 PKCe>PKCd>Src> PDPK1>PKCz> YAP1 PKCe>IKKb>COT> Chk1>LATS2> YAP1 PKCe (PRKCE)  Q02156 YAP1 S127 P46937 1 PKCe>Akt1>YAP1 PKCe>PKD1>Akt1> YAP1 PKCe>AurKB>ATM>Akt1>YAP1  Potential signalling pathways using only confirmed kinase-substrate connections reported in literature of up to 4 intermmediate kinases between kinases and substrate phosphosites were retrieved from the Kinector connection database (www.kinector.ca).   196 8.5 Validation of kinase-substrate relationships using in vitro kinase assay data Active recombinant kinases, namely EGFR and INSR, were used in kinase assays with HeLa cell lysates and analyzed with the microarray in Chapter 5. In addition, preparations of active recombinant focal adhesion kinase (FAK), Fyn, and Pyk2, were incubated with ATP in 100 µg mouse hippocampus brain lysates and analyzed on the microarray concurrently with corresponding lysate that had been incubated with ATP but no kinase. These kinase assays were performed by me as part of a larger study spearheaded by Dr. Hava Gil Henn from the Azrieli Faculty of Medicine, Bar-Ilan University, Safed, Israel. With these data, I was able to evaluate if my predicted kinase substrate relationships could be validated empirically with in vitro kinase assays with the KAM-1325 antibody microarray. Since the EGFR-kinase assays were performed in triplicate experiments, p values from a Student T test could be assigned to determine any statistically significant leads. For the FAK, Fyn, INSR, and Pyk2 experiments, the sum of the % error within the duplicate antibody measurements for each control and kinase-treated sample was reported. Some of the data had low signal intensities on the microarrays (evaluated as less than the half the typical median value observed) and were flagged. Although not statistically significant, four phosphosite-specific antibodies displayed reasonable increases in phosphorylation after in vitro lysate incubation with the active kinases and were of high data quality (high signal intensity on the microarray), and these are highlighted by red text in Table 8.4.   197 Table 8.5. Comparison of kinase-assay data of HeLa cell or mouse brain lysates incubated with recombinant kinases analyzed on the microarray with the correlated kinase-phosphosite leads generated from microarray metadata analysis. Kinase Target Kinase Uniprot ID Kinase Type Substrate Target Substrate Phosphosite Antibody 2 Code Substrate Uniprot ID Sum % Error from Duplicate Signals %CFC Recombinant Kinase Assay P Value/ Signal Intensity of Lysate PhosphoNet Prediction Kinase Rank  Same Pair Spearman Average EGFR (ErbB1) P00533 PYK CDK1 (CDC2) Y15 PK007-1 P06493   -4 0.937 4 0.442 EGFR (ErbB1) P00533 PYK Yes Y222 Y223 PK858 P07947   57 0.059 13 0.460 EGFR (ErbB1) P00533 PYK VEGFR2 (KDR) Y1054 PK852 P35968   28 0.312 33 0.419 FAK (PTK2) Q05397 PYK VEGFR1 (Flt1) Y1053 PK851 P17948 13 44   19 0.542 FAK (PTK2) Q05397 PYK ERK2 (MAPK1) Y263 S266 PK880 P28482 10 26   28 0.413 FAK (PTK2) Q05397 PYK STAT2 Y690 PN668 P52630 6 82 Low Signal 30 0.407 FAK (PTK2) Q05397 PYK Cas L Y166 PN505 Q14511 55 97   40 0.792 Fyn P06241 PYK EphB2 Y780 PK610 P29323 23 249 Low Signal 35 0.410 Fyn P06241 PYK Gab1 Y406 PN516 Q13480 18 128   50 0.466 Fyn P06241 PYK EphA2 Y772 PK607 P29317 26 547 Low Signal >50 0.405 Fyn P06241 PYK EphB3 Y600 PK611 P54753 6 117 Low Signal >50 0.500 Fyn P06241 PYK EZR Y146 PN224 P15311 8 25 Low Signal >50 0.440 Fyn P06241 PYK GSK3a Y279 PK649 P49840 11 279 Low Signal >50 0.405 IR (INSR) P06213 PYK JAK2 Y570 PK668 O60674 24 -31   >50 0.402 PYK2 (PTK2B) Q14289 PYK PFN1 Y129 PN524 P07737 46 120   >50 0.445 Each kinase-substrate pair were found to be correlated by the antibodies that track these targets on the KAM-1325 microarray with a Spearman’s coefficient of 0.4 or greater and p and q value of 0.05 or less. Candidate upstream kinase rankings for every phosphosite were evaluated using the Kinase-Substrate Predictor algorithm of the PhosphoNET database. Kinase assays done with HeLa cell lysates incubated with EGFR were performed in triplicate and the results were averaged. A paired Student T test were performed to generate p values for the EGFR assays. INSR assays were performed with HeLa cell lysates, while FAK, Fyn, and Pyk2 assays were performed with mouse hippocampus brain lysates.   198 8.6 Discussion While correlative analysis is not as convincing as other statistical tests in demonstrating bonafide protein-protein relationships, this approach was ultimately agreed upon as the best method to generate candidate relationship leads for this particular study during my consultation meetings with Statistics graduate students Mr. Peiyuan Zhu and Mr. Giorgio Sgarbi, from the Statistical Opportunity for Students program at UBC. Analysis via Spearman’s rank correlation was chosen as the best method to model the strength of correlations between antibody pairs, because it is robust to extreme differences in antibody performances across different datasets, such as differences in the signal intensities and %CFCs. A large caveat of such an analysis would be that a significant correlation alone, no matter the strength as represented by the Spearman’s coefficient, would not be sufficient to prove an interactive connection between two proteins. A significant correlation between two proteins may also be explained by indirect relationships or co-regulation in the same signalling pathway. Proving a direct interactional relationship between two proteins would require a follow up study designed to test such hypotheses.   In the correlative analysis, given there were 1350 antibodies on the array that translated to 910,575 possible pairs of antibody combinations, I had to implement parameters to reduce the dataset through stringent filters so that meaningful conclusions could be exposed more easily. Firstly, I took out 641 antibodies from the analysis, because they were observed to change with an average %CFC of less than 15% (in absolute value) across 254 datasets, and because I wanted to focus on the correlations between antibodies that showed greater changes in  199 response to treatments. However, these antibodies, while not changing as dramatically across the different datasets, could still have provided evidence of strong protein-protein correlations. Secondly, I narrowed down the number of pairings to evaluate by only focusing on those related to the list of what I had evaluated as reliable antibodies. Reliable antibodies were ones that had at least one other statistically significant positive correlating antibody for the same target protein on the array. By removing 641 antibodies from the analysis previously, I limited the number of antibodies that had validating same-target antibodies on the array. This ultimately reduced the analysis from looking at the correlations between the combinations of 1350 antibodies, down to only the significant positive correlating pairs of 320 antibodies. This was done, because despite removing the relatively non-changing antibodies, I still wanted to filter out the data to focus only on the pairings of antibodies that had an indication of picking up their intended targets and were of the strongest positive correlations. While it is most likely that valuable data may have been removed, such as significant correlations with coefficients of less than 0.4, as well as the significant negative correlations that imply mutual exclusion-type regulation between those proteins, I wanted to be stringent in my analysis to focus on the best leads from the data, which would have the greatest likelihood to be validated here or in the future.   In the data of significant kinase-phosphosite correlating pairs, these could be indicative of a forward kinase-substrate relationship, but also of a reverse relationship where the phosphorylation of the phosphosite on a substrate might lead to the increase in expression levels of the kinase, especially if the substrate is a transcription factor. This may explain why  200 many of these correlations were not validated by the empirical kinase-substrate relationships already documented in literature. Within the identified correlating kinase-substrate pairs that were validated by documented kinase-substrate relationships, nine of these established relationships were predicted by the PhosphoNET upstream kinase identifier algorithm as ranked within the top 50 kinase candidates for the given phosphosite. Six of the other relationships validated by literature were ranked by the algorithm as below the top 50 predicted candidates. This demonstrates that relationships reported in literature, especially with purified preparations of kinase and substrate, may not be as accurate as what happens in nature, where there is compartmentalization of proteins and differences in substrate availability, competition between many possible substrates, and only the substrates with the best affinity may be phosphorylated by the kinase. It is also possible that the Kinase-Substrate Predictor algorithm developed for the PhosphoNET website was not accurate for some kinases, in part due to insufficient training data available for substrates of these kinases. This is why antibody microarrays are actually well suited for discovery of new substrates for protein kinases from in vitro studies, since they can be used to confirm substrates that are affected by these same kinases when the kinases are specifically stimulated or inhibited in vivo. This can be achieved through cell manipulation with mitogens and other agents that directly target a kinase of interest for its stimulation, or inhibition of the kinase through specific drugs. Genetic approaches such as gene transfection with activated mutant forms of the kinase or knockouts such as with CRISPR-Cas9 gene editing can also be used to specifically alter the activity of a kinase and the phosphorylation status of a suspected substrate.    201 While my kinase assay testing of a few recombinant protein-tyrosine kinases was limited in confirming leads from the meta-analyses of possible kinase-substrate phosphosite relationships, it should be appreciated that the KAM-1325 antibody microarray may have featured phosphosite-specific antibodies for relatively few direct substrates of the EGFR and INSR protein-tyrosine kinases in HeLa cells. Likewise, Pyk2, FAK and Fyn may not be prominent kinases in the HeLa cells and may not have as many substrates in this cell line. It is feasible that statistically significant candidate direct-kinase substrate relationships may be more evident in other cell model systems with future research. While this experiment was designed to identify direct substrates of the tested kinases, the candidates identified here may not be true physiological substrates or be representative of true physiological signalling outcomes resulting from activity of these kinases. For example, the activity of these kinases physiologically may involve the other proteins, such as the recruitment of Src kinase through autophosphorylation of FAK.   Overall, this chapter highlights one of the greatest advantages of the antibody microarray, which is its ease in generating large datasets that are compatible for comparison with each other. This work highlights the exponential value that antibody microarrays can have when used in large scale and when the data can be collected for analysis in aggregate. If antibody microarray data could be collected and catalogued in databases as mass spectrometry datasets are currently, profound insights may be able to be extracted from this wealth of scattered data. This work epitomizes of how antibody-based methods can elucidate complex cell signalling  202 networks without bias from preconceived notions of the relationships between target proteins through consistent interrogation of the same protein targets using the same reagent probes.    203 Chapter 9: Future Directions and Conclusions  9.1  Characterization of polyclonal antibodies Antibodies are invaluable research tools and have also been growing in popularity as therapeutic drugs. The use of antibodies in technologies and techniques such as antibody microarrays, ELISA, immunoprecipitation, or Western blotting all depend on the quality of the antibodies used. Sourcing high quality, well characterized antibodies for use in these applications persists as a significant hurdle in producing reliable and accurate results. Some efforts have been made to document many of the more commonly used antibodies reported in the scientific literature such as CiteAb (www.citeab.com). However, most commercial antibodies are actually relatively poorly characterized. The value and potential of antibody microarrays is pivotal on the antibodies that are used for its printing.   The desperate need for high quality antibodies is an industry-wide phenomenon and printing an array requires substantial amounts of capital upfront to purchase and test large quantities of antibodies. While monoclonal antibodies can have strategic benefits in certain aspects, the work in this thesis demonstrate that polyclonal antibodies continue to be worth producing and using due to lower financial cost, ethical production, and the fact that well characterized polyclonal antibodies function effectively better than monoclonal antibodies in many situations. The strategy our lab has undertaken in producing polyclonal antibodies by immunizing rabbits with up to five different peptides at a time, has further reduced the costs and loss of life associated with generating animal antibodies. The use of SPOT peptide membranes is a  204 particularly effort- and cost-effective way to screen antibodies against peptide arrays for their comparative performance. My systematic tests have demonstrated how phosphoserine site-specific antibodies are particularly susceptible to cross-reactivities with non-target phosphoepitopes, which is problematic unless properly identified. By negatively purifying out rabbit sera through phosphotyrosine-agarose columns, I was able to reduce non-specific anti-phosphotyrosine antibody populations from phosphosite-specific preparations and produce a valuable generic phospho-tyrosine antibody as a side product. All these steps taken together takes the route of antibody production into the next level of reliable and economical research that can still be extensive and high throughput.   9.2  Improved application of antibody microarrays  Through improvement of sampling handling procedures, I was able to better preserve protein phosphorylation events, reduce false positives from proteins found in complexes being co-captured on the microarray, reduce competition between antibodies printed in the same field for the same target protein, and better normalize signal intensity strengths to the stoichiometry of target proteins rather than their native sizes. The strategies in signal detection that I have developed have helped improve sensitivity, decrease unspecific background signals, and demonstrate innovative ways to incorporate antibody microarrays for different kinds of analysis, such as profiling of protein-tyrosine phosphorylation or protein-protein interactions. All of these incremental improvements are aimed at polishing the technology to produce better reliable tools for research of clinical biomarkers and diagnostics.    205 While experiments described in this thesis have demonstrated the feasibility of using other reporter antibodies to profile general tyrosine phosphorylation as well as ubiquitination of proteins, the sandwich antibody microarray method can be used to interrogate protein-protein interactions. For example, to examine the specificity of 14-3-3 proteins, dye-labelled isoform-specific antibodies can be used to identify binding partners in HeLa cell lysates by capturing the unfragmented protein targets on to the array and probing with the antibody. This could be further developed in the future as a larger scale study, especially with proteins of unknown binding partners, such as orphan kinases and their unknown substrates.   In addition to the various techniques previously mentioned, pilot experiments were also performed to investigate the feasibility of using the KAM-1325 antibody microarray to screen for drug targets. In these experiments, unfragmented HeLa cells were captured onto the antibody microarray and were subsequently incubated with biotinylated ATP-analogue 5ʹ-(4-fluorosulfonylbenzoyl)adenosine hydrochloride (FSBA). After the covalent binding of the biotinylated FSBA to active kinases captured onto the array, the array was then incubated with fluorescent dye-labelled anti-biotin antibody. The method proved to be successful in pilot experiments, and warrants follow-up studies in optimizing this protocol with known as well as novel kinase inhibitors.   In this thesis, I used the antibody microarray as an explorative tool to study basic fundamental cellular phenomena. From the analysis of the over 250 sets of KAM-1325 antibody microarray analyses, there were several model systems and treatments that produced numerous and  206 intense changes, like EGF stimulation in A431 cells, and others that featured almost none, like site-directed mutagenesis of selected proteins in cell lines. Through comparison of EGF and insulin signalling, I was able to identify profound changes in phosphorylations of these similar growth factors, which help make some sense of why EGF ultimately is mitogenic, while insulin is more metabolic. I also studied the effects of phosphatase inhibition as well as protein-synthesis inhibition, and identified substantial changes in phosphorylation, which allowed me to hypothesize biochemical mechanisms that cells may employ to maintain homeostasis.   9.3   Exploration of cell signalling architectures While studies in this thesis have provided examples of the type of signalling responses that cells can have towards treatment with EGF, insulin, PAO + vanadate, anisomycin, cycloheximide, and others, the overarching purpose of these studies was to see if visual maps could be generated to help navigate known protein-protein relationships and propose novel ones for further testing. Based on the correlative performance of these antibodies within over 250 sets of data generated with the same KAM-1325 antibody microarrays, I was able to model the generic protein-protein associations common to most cell and tissue types. With in silico analysis, I predicted which of these proposed kinase-substrate connections were the most likely to be authentic in normal physiology, as well as propose pathway connections between indirect relationships. Some of the kinase-substrate connections involving a few protein-tyrosine kinases were then validated with kinase assays. The next step to validate more of these novel relationships could be undertaken by a larger scale study involving the incubation of lysates  207 pre-phosphorylated by more purified protein kinases onto higher content antibody microarrays that feature even more phosphosite-specific antibodies.   In my studies, I have utilized the Kinase-Substrate Predictor algorithm from PhosphoNET. As these new kinase assay studies are performed, that data would be valuable to retrain the algorithm for better accuracy and precision. The KiNET Kinector pathway identifier also emerged as an extremely powerful tool to identify potential direct and indirect pathways that connect two proteins, and the ways that they can regulate each other. Other bioinformatics tools that could be informative include the analysis of gene ontology terms associated with the protein targets being tracked on the antibody microarray. Methods like these can dramatically help researchers digest large datasets resulting from antibody microarray or mass spectrometry experiments more efficiently in understanding big picture biological events. This is particularly helpful for researchers who are less familiar with the particular protein leads that could be generated from such analyses and improves overall user friendliness and approachability. Taken together, from the visual maps, gene ontology terms, and ease of analyzing biofluids or cell and tissue samples without special enrichment, the antibody microarray truly emerges as an attractive platform for analysis of overarching cell-signalling architectures as well as detailed events such as site-specific protein phosphorylation.   9.4  Follow up validation of the antibody microarray Despite the extensive efforts taken to characterize the majority of antibodies used in printing the antibody microarray, there were challenges in validating the leads from the microarray  208 analysis with Western blotting. This immunoblotting was very laborious, and often unable to yield results that were reconcilable with the antibody microarray, primarily because of cross-reactive proteins or insufficient sensitivity in visualizing bands where I expected the target to be. The primary question that I was faced with was if using the antibody microarray was worthwhile if the resulting leads cannot be consistently validated. While the reputational trustworthiness of targets identified by antibody microarrays is not as great as mass spectrometry, the work undertaken in this thesis demonstrates the idea that antibodies can still be reliable and straight-forward tools when used appropriately. Antibodies are biological macromolecules that have affinity for a certain peptide epitope, and once those epitopes are properly defined, any assay using that antibody to identify the presence of peptide fragments can be equivalent to the identification of the same peptide fragment by mass spectrometry through the interpretation of mass-to-charge ratios. While the antibody microarray is typically used in conjunction with relatively strong treatments to dramatize potential changes in cell signalling for easier identification, the technology was still able to generate reproducible reliable results with very similar samples among the datasets used in the meta-analysis. To improve the confidence of any biological trends concluded from microarray analyses, more antibodies targeting the same proteins can be used as internal cross-validation.   The gold standard of Western blotting validation for antibody probes is perhaps antiquated as Western blotting almost seems akin to validating minute changes in pH of buffers with pH paper after having used more advanced technologies like a pH meter. There needs to be new improved methods of validating biological trends with antibodies other than by Western  209 blotting, and our lab has been active in collaborations to develop new technologies that couples antibody microarray immune-capture with mass spectrometry identification for every individual antibody. This combination could theoretically be the best of both worlds, with antibody probes offering targeted capture of low abundance cell signalling proteins, and the reputational trustworthiness of MS identification. Another option could be to separately validate leads generated by the microarray with a form of ELISA that uses pairs of different antibodies that target the same protein to confirm results. Compared to Western blotting, ELISA utilizes much less lysate material, and can be performed in much greater assay capacities especially when capture antibodies can be robotically printed onto microtiter plates.   Since 2008, research publications involving antibody microarrays have stagnated (Figure 9.1). The rise and fall of antibody microarrays in research draw parallels with oligonucleotide gene chips that have somewhat given way to other microfluidic technologies such as quantitative real-time polymerase chain reaction. However, the use of oligonucleotides to target nucleic acid sequences of interest still remains, as will the use of antibodies to target proteins and protein-phosphorylations of interest. While the future of antibody use may also evolve from printed arrays on glass slides, ultimately, the antibody microarray remains as an effective explorative tool to assay many antibodies in a quick and easy manner across many different kinds of biological samples. Building on the those that are already discussed in this thesis, I still believe that continued improvements to the reliablility and accuracy of antibody microarrays will allow researchers to access the significant untapped potential of this technology in greater research and clinical applications.   210  Figure 9.1. Annual publications relating to antibody microarrays. The search term “antibody microarray” was queried on Google Scholar and tabulated. Redundant entries were removed.  9.5  Conclusion In this body of work, I set out to track the composition and architecture of cell signalling networks in an unbiased manner by using antibody-based methodology. High content antibody microarrays were selected a primary tool for tracking hundreds of low-abundance cell signalling proteins simultaneously among various model systems and cellular treatments. Through extensive characterization of the antibodies used to print the KAM-1325 antibody microarray, I defined common strengths, weaknesses, and characteristics of rabbit polyclonal protein phopshosite-specific antibodies. I improved sample handling techniques in order to generate  211 higher quality signals to produce more reliable data that improved capture of phosphorylation events that are easily lost with typical procedures. The antibody microarray was ultimately successful in its application to elucidate protein expression and phosphorylation changes of cells in response to treatments and made the resulting datasets easy to compare to each other, because of consistent interrogation of the same target proteins using the same reagent probes across model systems. This unique feature allowed me to distill valuable insight into novel potential kinase-substrate candidate relationships, without bias towards pre-existing relationships, and I was able to follow up these candidates using computational bioinformatic tools as well as empirical kinase assays. While the scope of my data analysis was limited to primarily phosphorylation events as well as kinase-substrate signalling networks, because of the predominant role that phosphorylation plays in facilitating signal transduction, there were many other aspects of cell signalling that can also be explored from these datasets. This body of work truly highlights the flexibility of the antibody microarray platform, the wealth of information that can be successfully accessed easily, quickly, and at reasonably low cost, and the immense potential that this research technique still holds for the future of proteomics research.         212 Bibliography Aebersold, R., Agar, J. N., Amster, I. J., Baker, M. S., Bertozzi, C. R., Boja, E. S., Costello, C. E., Cravatt, B. F., Fenselau, C., Garcia, B. A., Ge, Y., Gunawardena, J., Hendrickson, R. C., Hergenrother, P. J., Huber, C. G., Ivanov, A. R., Jensen, O. N., Jewett, M. C., Kelleher, N. L., Kiessling, L. L., Krogan, N. J., Larsen, M. R., Loo, J. A., Ogorzalek Loo, R. R., Lundberg, E., MacCoss, M. J., Mallick, P., Mootha, V. K., Mrksich, M., Muir, T. W., Patrie, S. M., Pesavento, J. J., Pitteri, S. J., Rodriguez, H., Saghatelian, A., Sandoval, W., Schlüter, H., Sechi, S., Slavoff, S. A., Smith, L. M., Snyder, M. P., Thomas, P. M., Uhlén, M., Van Eyk, J. E., Vidal, M., Walt, D. R., White, F. M., Williams, E. R., Wohlschlager, T., Wysocki, V. H., Yates, N. A., Young, N. L., & Zhang, B. (2018). How many human proteoforms are there? Nature Chemical Biology, 14(3), 206-214. doi:10.1038/nchembio.2576   Agrawal, G. K., & Thelen, J. J. (2005). Development of a simplified, economical polyacrylamide gel staining protocol for phosphoproteins. Proteomics, 5(18), 4684-4688. doi:10.1002/pmic.200500021   Alessi, D. R., Andjelkovic, M., Caudwell, B., Cron, P., Morrice, N., Cohen, P., & Hemmings, B. A. (1996). Mechanism of activation of protein kinase B by insulin and IGF-1. The EMBO Journal, 15(23), 6541-6551. doi:10.1002/j.1460-2075.1996.tb01045.x   Andersen, J. N., Mortensen, O. H., Peters, G. H., Drake, P. G., Iversen, L. F., Olsen, O. H., Jansen, P. G., Andersen, H. S., Tonks, N. K., & Møller, N. P. (2001). Structural and evolutionary relationships among protein tyrosine phosphatase domains. Molecular and Cellular Biology, 21(21), 7117-7136. doi:10.1128/MCB.21.21.7117-7136.2001   Ardito, F., Giuliani, M., Perrone, D., Troiano, G., & Lo Muzio, L. (2017). The crucial role of protein phosphorylation in cell signaling and its use as targeted therapy (Review). Int J Mol Med, 40(2), 271–280. doi: 10.3892/ijmm.2017.3036  Azzi, J. R., Sayegh, M. H., & Mallat, S. G. (2013). Calcineurin inhibitors: 40 years later, Can’t live without. The Journal of Immunology, 191(12), 5785-5791. doi:10.4049/jimmunol.1390055   Bartlett, J., Langdon, S., Simpson, B., Stewart, M., Katsaros, D., Sismondi, P., Love, S., Scott, W. N., Williams, A. R., Lessells, A. M., Macleod, K. G., Smyth, J. F., & Miller, W. (1996). The prognostic value of epidermal growth factor receptor mRNA expression in primary ovarian cancer. British Journal of Cancer, 73(3), 301-306. doi:10.1038/bjc.1996.53   Bootman, M. D., Collins, T. J., Peppiatt, C. M., Prothero, L. S., MacKenzie, L., De Smet, P., Travers, M., Tovey, S. C., Seo, J. T., Berridge, M. J., Ciccolini, F., & Lipp, P. (2001). Calcium signalling—an overview. Seminars in Cell and Developmental Biology, 12(1), 3-10. doi:10.1006/scdb.2000.0211   213 Bourne, R. R. A., Stevens, G. A., White, R. A., Smith, J. L., Flaxman, S. R., Price, H., Jonas, J. B., Keeffe, J., Leasher, J., Naidoo, K., Pesudovs, K., Resnikoff, S., & Taylor, H. R. (2013). Causes of vision loss worldwide, 1990–2010: A systematic analysis. Lancet Global Health, 1(6), 339-349. doi:10.1016/S2214-109X(13)70113-X   Bradford, M. M. (1976). A rapid and sensitive method for the quantitation of microgram quantities of protein utilizing the principle of protein-dye binding. Analytical Biochemistry, 72(1-2), 248-254. doi:10.1006/abio.1976.9999  Bilandzic, A., & Rosella, L. (2017). The cost of diabetes in canada over 10 years: Applying attributable health care costs to a diabetes incidence prediction model. Health Promotion and Chronic Disease Prevention in Canada, 37(2), 49-53. doi:10.24095/hpcdp.37.2.03   Cantley, L. C., Engelman, J. A., & Luo, J. (2006). The evolution of phosphatidylinositol 3-kinases as regulators of growth and metabolism. Nature Reviews Genetics, 7(8), 606-619. doi:10.1038/nrg1879   Carey, J. O., Azevedo, J. L. Jr, Morris, P. G., Pories, W. J., & Dohm, G. L. (1995). Okadaic acid, vanadate, and phenylarsine oxide stimulate 2-deoxyglucose transport in insulin-resistant human skeletal muscle. Diabetes, 44(6), 682-688. doi:10.2337/diab.44.6.682   Chang, T. (1983). Binding of cells to matrixes of distinct antibodies coated on solid surface. Journal of Immunological Methods, 65(1), 217-223. doi:10.1016/0022-1759(83)90318-6 Chen, Y., & Hoehenwarter, W. (2019). Rapid and reproducible phosphopeptide enrichment by tandem metal oxide affinity chromatography: Application to boron deficiency induced phosphoproteomics. The Plant Journal, 98(2), 370-384. doi:10.1111/tpj.14215   Chen, Z., Dodig-Crnković, T., Schwenk, J. M., & Tao, S. (2018). Current applications of antibody microarrays. Clinical Proteomics, 15(1), 7-15. doi:10.1186/s12014-018-9184-2   Cohen, P. (2000). The regulation of protein function by multisite phosphorylation – a 25 year update. Trends in Biochemical Sciences (Amsterdam. Regular Ed.), 25(12), 596-601. doi:10.1016/s0968-0004(00)01712-6   Cohen, P. (2002). The origins of protein phosphorylation. Nature Cell Biology, 4(5), E127-E130. doi:10.1038/ncb0502-e127  Cohen P. T. (2004) Overview of protein serine/threonine phosphatases. In Ariño J., Alexander D.R. (Eds.), Protein Phosphatases. Springer, Berlin: Heidelberg.    214 Cui, X., Kim, H., Kuiatse, I., Kim, H., Brown, P. H., & Lee, A. V. (2006). Epidermal growth factor induces insulin receptor substrate-2 in breast cancer cells via c-jun NH2-terminal Kinase/Activator protein-1 signaling to regulate cell migration. Cancer Research (Chicago, Ill.), 66(10), 5304-5313. doi:10.1158/0008-5472.CAN-05-2858   Dagogo-Jack, I., Engelman, J. A., & Shaw, A. T. (2017). Overcoming on-target resistance to tyrosine kinase inhibitors in lung cancer. Annual Review of Cancer Biology, 1(1), 257-274. doi:10.1146/annurev-cancerbio-050216-122044   Darnell, J. E. Jr. (1997). Phosphotyrosine signaling and the single cell:metazoan boundary. Proc. Natl. Acad. Sci. U.S.A. 94, 11767–11769. doi:10.1073/pnas.94.22.11767   Day, E. K., Sosale, N. G., & Lazzara, M. J. (2016). Cell signaling regulation by protein phosphorylation: A multivariate, heterogeneous, and context-dependent process. Current Opinion in Biotechnology, 40, 185-192. doi:10.1016/j.copbio.2016.06.005   De Loubresse, N., Prokhorova, I., Holtkamp, W., Rodnina, M., Yusupova, G., & Yusupov, M. (2014). Structural basis for the inhibition of the eukaryotic ribosome. Nature, 513(7519), 517-517. doi:10.1038/nature13737   De Meyts, P. (2000, updated 2016). The insulin receptor and its signal transduction network. In Feingold KR, Anawalt B, Boyce A, et al. (Eds). Endotext. South Dartmouth, MA: MDText.com, Inc. Available from: https://www.ncbi.nlm.nih.gov/books/NBK378978/   Delfani, P., Dexlin Mellby, L., Nordström, M., Holmér, A., Ohlsson, M., Borrebaeck, C. A. K., & Wingren, C. (2016). Technical advances of the recombinant antibody microarray technology platform for clinical immunoproteomics. PloS One, 11(7), e0159138. doi:10.1371/journal.pone.0159138   Di Fiore, P. P., Pierce, J. H., Fleming, T. P., Hazan, R., Ullrich, A., King, C. R., Schlessinger, J., & Aaronson, S. A. (1987). Overexpression of the human EGF receptor confers an EGF-dependent transformed phenotype to NIH 3T3 cells. Cell, 51(6), 1063-1070. doi:10.1016/0092-8674(87)90592-7   Ekins, R. P. (1989). Multi-analyte immunoassay. J Pharm Biomed Anal, 7: 155-168. doi: 10.1016/0731-7085(89)80079-2   Ekins, R. P. (1998). Ligand assays: From electrophoresis to miniaturized microarrays. Clinical Chemistry, 44(9), 2015-2030. doi:10.1093/clinchem/44.9.2015   Ekins, R., & Chu, F. (1991). Multianalyte microspot immunoassay-microanalytical "compact disk" of the future. Clinical Chemistry, 37(11), 1955-1967. doi:10.1093/clinchem/37.11.1955    215 Engholm-Keller, K., Waardenberg, A. J., Müller, J. A., Wark, J. R., Fernando, R. N., Arthur, J. W., Robinson, P. J., Dietrich, D., Schoch, S., & Graham, M. E. (2019). The temporal profile of activity-dependent presynaptic phospho-signalling reveals long-lasting patterns of poststimulus regulation. PLoS Biology, 17(3), e3000170. doi:10.1371/journal.pbio.3000170   Frank, C., Burkhardt, C., Imhof, D., Ringel, J., Zschornig, O., Wieligmann, K., Zacharias, M., & Bohmer, F. (2004). Effective dephosphorylation of src substrates by SHP-1. Journal of Biological Chemistry, 279(12), 11375-11383. doi:10.1074/jbc.M309096200  Fields, G. B., & Noble, R. L. (1990). Solid phase peptide synthesis utilizing 9-fluorenylmethoxycarbonyl amino acids. International Journal of Peptide and Protein Research, 35(3), 161-214. doi:10.1111/j.1399-3011.1990.tb00939.x   Fischer, E. H., Graves, D. J., Crittenden, E. R. S., and Krebs, E. G. (1959). Structure of the site phosphorylated in the phosphorylase b to a reaction. J. Biol. Chem, 234, 1698–1704.  Gandhi, V., Plunkett, W., & Cortes, J. E. (2014). Omacetaxine: A protein translation inhibitor for treatment of chronic myelogenous leukemia. Clinical Cancer Research : An Official Journal of the American Association for Cancer Research, 20(7), 1735-1740. doi:10.1158/1078-0432.CCR-13-1283   Garcia-Morales, P., Minami, Y., Luong, E., Klausner, R. D., & Samelson, L. E. (1990). Tyrosine phosphorylation in T cells is regulated by phosphatase activity: Studies with phenylarsine oxide. Proceedings of the National Academy of Sciences of the United States of America, 87(23), 9255-9259. doi:10.1073/pnas.87.23.9255   Garcia, B. A. (2019). Post-translational modifications that modulate enzyme activity. San Diego: Elsevier Science & Technology.   Gates, M. B., Tomer, K. B., & Deterding, L. J. (2010). Comparison of metal and metal oxide media for phosphopeptide enrichment prior to mass spectrometric analyses. Journal of the American Society for Mass Spectrometry, 21(10), 1649-1659. doi:10.1016/j.jasms.2010.06.005   Gembitsky, D. S., Lawlor, K., Jacovina, A., Yaneva, M., & Tempst, P. (2004). A prototype antibody microarray platform to monitor changes in protein tyrosine phosphorylation. Molecular & Cellular Proteomics, 3(11), 1102-1118. doi:10.1074/mcp.M400075-MCP200   Gille, H., Kortenjann, M., Thomae, O., Moomaw, C., Slaughter, C., Cobb, M. H., & Shaw, P. E. (1995). ERK phosphorylation potentiates Elk-1-mediated ternary complex formation and transactivation. The EMBO Journal, 14(5), 951-962. doi:10.1002/j.1460-2075.1995.tb07076.x    216 Gille, H., Sharrocks, A. D., & Shaw, P. E. (1992). Phosphorylation of transcription factor p62TCF by MAP kinase stimulates ternary complex formation at c-fos promoter. Nature, 358(6385), 414-417. doi:10.1038/358414a0   Gonzalez-Angulo, A. M., & Meric-Bernstam, F. (2010). Metformin: A therapeutic opportunity in breast cancer. Clinical Cancer Research, 16(6), 1695-1700. doi:10.1158/1078-0432.CCR-09-1805   Gregg, J., & Fraizer, G. (2011). Transcriptional Regulation of EGR1 by EGF and the ERK Signaling Pathway in Prostate Cancer Cells. Genes & cancer, 2(9), 900–909. doi:10.1177/1947601911431885   Grollman, A. P. (1967). Inhibitors of protein biosynthesis. II. Mode of action of anisomycin. Journal of Biological Chemistry, 242(13), 3226.   Gutmann, T., Kim, K. H., Grzybek, M., Walz, T., & Coskun, Ü. (2018). Visualization of ligand-induced transmembrane signaling in the full-length human insulin receptor. Journal of Cell Biology, 217(5), 1643-1649. doi:10.1083/jcb.201711047   Haigler, H., Ash, J. F., Singer, S. J., & Cohen, S. (1978). Visualization by fluorescence of the binding and internalization of epidermal growth factor in human carcinoma cells A-431. Proceedings of the National Academy of Sciences - PNAS, 75(7), 3317-3321. doi:10.1073/pnas.75.7.3317  Hanks, S. K., Quinn, A. M., & Hunter, T. (1988). The protein kinase family: Conserved features and deduced phylogeny of the catalytic domains. Science, 241(4861), 42-52. doi:10.1126/science.3291115   Harrell, F. E. Jr, Dupont, C, et al. (2019). Hmisc: Harrell Miscellaneous. R package version 4.3-0. Retreived from: https://CRAN.R-project.org/package=Hmisc   Hastie, C. J., Cohen, P., & McLauchlan, H. J. (2006). Assay of protein kinases using radiolabeled ATP: A protocol. Nature Protocols, 1(2), 968-971. doi:10.1038/nprot.2006.149   Hay, N., & Sonenberg, N. (2004). Upstream and downstream of mTOR. Genes & Development, 18(16), 1926-1945. doi:10.1101/gad.1212704   Heberle, H., Meirelles, G. V., da Silva, F. R., Telles, G. P., & Minghim, R. (2015). InteractiVenn: A web-based tool for the analysis of sets through venn diagrams. BMC Bioinformatics, 16(1), 169. doi:10.1186/s12859-015-0611-3  Heckman-Stoddard, B. M., DeCensi, A., Sahasrabuddhe, V. V., & Ford, L. G. (2017). Repurposing metformin for the prevention of cancer and cancer recurrence. Diabetologia, 60(9), 1639-1647. doi:10.1007/s00125-017-4372-6  217   Hilpert, K., Winkler, D.F., Hancock, R.E. (2007). Peptide arrays on cellulose sup- port: SPOT synthesis, a time and cost efficient method for synthesis of large numbers of peptides in a parallel and addressable fashion. Nature Protocols, 2, 1333–1349. doi:10.1038/nprot.2007.160   Hofman, F. M., & Taylor, C. R. (2013). Immunohistochemistry. Current Protocols in Immunology, 103, 21.4.1.   Hornbeck, P. V., Kornhauser, J. M., Latham, V., Murray, B., Nandhikonda, V., Nord, A., Skrzypek, E., Wheeler, T., Zhang, B., & Gnad, F. (2019). 15 years of PhosphoSitePlus®: Integrating post-translationally modified sites, disease variants and isoforms. Nucleic Acids Research, 47(D1), D433-D441. doi:10.1093/nar/gky1159   Hou, Q., Ufer, G., & Bartels, D. (2016). Lipid signalling in plant responses to abiotic stress. Plant, Cell & Environment, 39(5), 1029-1048. doi:10.1111/pce.12666   Hsu, C., Xue, L., Arrington, J. V., Wang, P., Paez Paez, J. S., Zhou, Y., Zhu, J., & Tao, W. A. (2017). Estimating the efficiency of phosphopeptide identification by tandem mass spectrometry. Journal of the American Society for Mass Spectrometry, 28(6), 1127-1135. doi:10.1007/s13361-017-1603-5   Hulley, P. A., Gordon, F., & Hough, F. S. (1998). Inhibition of mitogen-activated protein kinase activity and proliferation of an early osteoblast cell line (MBA 15.4) by dexamethasone: Role of protein phosphatases. Endocrinology, 139(5), 2423-2431. doi:10.1210/en.139.5.2423   Humphrey, S. J., James, D. E., & Mann, M. (2015). Protein phosphorylation: A major switch mechanism for metabolic regulation. Trends in Endocrinology & Metabolism, 26(12), 676-687. doi:10.1016/j.tem.2015.09.013  Hunter, T., & Cooper, J. A. (1981). Epidermal growth factor induces rapid tyrosine phosphorylation of proteins in A431 human tumor cells. Cell, 24(3), 741-752. doi:10.1016/0092-8674(81)90100-8   Hunter, T., & Sefton, B. M. (1980). Transforming gene product of rous sarcoma virus phosphorylates tyrosine. Proceedings of the National Academy of Sciences of the United States of America, 77(3), 1311-1315. doi:10.1073/pnas.77.3.1311   Hutchinson, J. A., Shanware, N. P., Chang, H., & Tibbetts, R. S. (2011). Regulation of ribosomal protein S6 phosphorylation by casein kinase 1 and protein phosphatase 1. The Journal of Biological Chemistry, 286(10), 8688-8696. doi:10.1074/jbc.m110.141754   218 Huyer, G., Liu, S., Kelly, J., Moffat, J., Payette, P., Kennedy, B., Tsaprailis, G., Gresser, M. J., & Ramachandran, C. (1997). Mechanism of inhibition of protein-tyrosine phosphatases by vanadate and pervanadate. Journal of Biological Chemistry, 272(2), 843-851. doi:10.1074/jbc.272.2.843   Iliuk, A. B., & Tao, W. A. (2015). Universal non-antibody detection of protein phosphorylation using pIMAGO. Current Protocols in Chemical Biology, 7(1), 17. doiI:10.1002/9780470559277.ch140208   Iliuk, A. B., Arrington, J. V., & Tao, W. A. (2014). Analytical challenges translating mass spectrometry-based phosphoproteomics from discovery to clinical applications. Electrophoresis, 35(24), 3430-3440. doi:10.1002/elps.201400153   Jaumot, M., & Hancock, J. F. (2001). Protein phosphatases 1 and 2A promote raf-1 activation by regulating 14-3-3 interactions. Oncogene, 20(30), 3949-3958. doi:10.1038/sj.onc.1204526   Jiang, X., Xu, Y., Huang, M., Zhang, L., Su, M., Chen, X., & Lu, J. (2017). Osimertinib (AZD9291) decreases programmed death ligand-1 in EGFR-mutated non-small cell lung cancer cells. Acta Pharmacologica Sinica, 38(11), 1512-1520. doi:10.1038/aps.2017.123   Jünger, M. A., & Aebersold, R. (2014). Mass spectrometry-driven phosphoproteomics: Patterning the systems biology mosaic. Wiley Interdisciplinary Reviews: Developmental Biology, 3(1), 83-112. doi:10.1002/wdev.121   Kaneko, T., Joshi, R., Feller, S. M., & Li, S. S. (2012). Phosphotyrosine recognition domains: The typical, the atypical and the versatile. Cell Communication and Signaling, 10(1), 32-32. doi:10.1186/1478-811X-10-32   Kim, H., Seo, H., & Cho, S. (2011). Phenylarsine oxide inhibits acid phosphatase-1. Bulletin of the Korean Chemical Society, 32(11), 4103-4105. doi:10.5012/bkcs.2011.32.11.4103   Kim, M., Pinto, S. M., Getnet, D., Nirujogi, R. S., Manda, S. S., Chaerkady, R., Madugundu, A. K., Kelkar, D. S., Isserlin, R., Jain, S., Thomas, J. K., Muthusamy, B., Leal-Rojas, P., Kumar, P., Sahasrabuddhe, N. A., Balakrishnan, L., Advani, J., George, B., Renuse, S., Selvan, L. D. N., Patil, A. H., Nanjappa, V., Radhakrishnan, A., Prasad, S., Subbannayya, T., Raju, R., Kumar, M., Sreenivasamurthy, S. K., Marimuthu, A., Sathe, G. J., Chavan, S., Datta, K. K., Subbannayya, Y., Sahu, A., Yelamanchi, S. D., Jayaram, S., Rajagopalan, P., Sharma, J., Murthy, K. R., Syed, N., Goel, R., Khan, A. A., Ahmad, S., Dey, G., Mudgal, K., Chatterjee, A., Huang, T., Zhong, J., Wu, X., Shaw, P. G., Freed, D., Zahari, M. S., Mukherjee, K. K., Shankar, S., Mahadevan, A., Lam, H., Mitchell, C. J., Shankar, S. K., Satishchandra, P., Schroeder, J. T., Sirdeshmukh, R., Maitra, A., Leach, S. D., Drake, C. G., Halushka, M. K., Prasad, T. S. K., Hruban, R. H., Kerr, C. L., Bader, G. D., Iacobuzio-Donahue, C. A., Gowda, H., & Pandey, A. (2014). A draft map of the human proteome. Nature, 509(7502), 575-581. doi:10.1038/nature13302    219 Kohl, T. O., & Ascoli, C. A. (2017). Direct and indirect cell-based enzyme-linked immunosorbent assay. Cold Spring Harbor Protocols, 2017(5). doi:10.1101/pdb.prot093732  Lai, S. (2015). Investigations of the origin, regulation, and substrate specificities of protein kinases in the human kinome (Doctoral Dissertation). University of British Columbia. Retrieved from https://open.library.ubc.ca/collections/ubctheses/24/items/1.0167195   Lai, S., & Pelech, S. (2016). Regulatory roles of conserved phosphorylation sites in the activation T-loop of the MAP kinase ERK1. Molecular Biology of the Cell, 27(6), 1040-1050. doi:10.1091/mbc.E15-07-0527   Lai, S., Safaei, J., & Pelech, S. (2016). Evolutionary ancestry of eukaryotic protein kinases and choline kinases. The Journal of Biological Chemistry, 291(10), 5199-5205. doi:10.1074/jbc.M115.691428   Lazo, J. S., McQueeney, K. E., & Sharlow, E. R. (2017). New approaches to difficult drug targets: The phosphatase story. Los Angeles, CA: SAGE Publications. doi:10.1177/2472555217721142   Lemmon, M. A., & Schlessinger, J. (2010). Cell signaling by receptor tyrosine kinases. Cell, 141(7), 1117-1134. doi:10.1016/j.cell.2010.06.011   Li, F., Li, C., Marquez-Lago, T. T., Leier, A., Akutsu, T., Purcell, A. W., Ian Smith, A., Lithgow, T., Daly, R. J., Song, J., & Chou, K. (2018). Quokka: A comprehensive tool for rapid and accurate prediction of kinase family-specific phosphorylation sites in the human proteome. Bioinformatics, 34(24), 4223-4231. doi:10.1093/bioinformatics/bty522   Li, J., Elberg, G., & Shechter, Y. (1996). Phenylarsine oxide and vanadate: Apparent paradox of inhibition of protein phosphotyrosine phosphatases in rat adipocytes. BBA - Molecular Cell Research, 1312(3), 223-230. doi:10.1016/0167-4889(96)00038-9   Li, Y., Zhou, X., Zhai, Z., & Li, T. (2017). Co-occurring protein phosphorylation are functionally associated. Plos Computational Biology, 13(5), e1005502. doi:10.1371/journal.pcbi.1005502   Lian, W., Wu, D., Lim, D. V., & Jin, S. (2010). Sensitive detection of multiplex toxins using antibody microarray. Analytical Biochemistry, 401(2), 271-279. doi:10.1016/j.ab.2010.02.040   Liao, B. C., Griesing, S., & Yang, J. C. (2019). Second-line treatment of EGFR T790M-negative non-small cell lung cancer patients. Therapeutic Advances in Medical Oncology, 11. doi:10.1177/1758835919890286   Liao, K., Hoffman, R. D., & Lane, M. D. (1991). Phosphotyrosyl turnover in insulin signaling. characterization of two membrane-bound pp15 protein tyrosine phosphatases from 3T3-L1 adipocytes. Journal of Biological Chemistry, 266(10), 6544.   220 Liberti, S., Sacco, F., Calderone, A., Perfetto, L., Iannuccelli, M., Panni, S., Santonico, E., Palma, A., Nardozza, A. P., Castagnoli, L., & Cesareni, G. (2013). HuPho: The human phosphatase portal. The FEBS Journal, 280(2), 379-387. doi:10.1111/j.1742-4658.2012.08712.x  Lindqvist, L. M., Vikström, I., Chambers, J. M., McArthur, K., Ann Anderson, M., Henley, K. J., Happo, L., Cluse, L., Johnstone, R. W., Roberts, A. W., Kile, B. T., Croker, B. A., Burns, C. J., Rizzacasa, M. A., Strasser, A., & Huang, D. C. S. (2012). Translation inhibitors induce cell death by multiple mechanisms and mcl-1 reduction is only a minor contributor. Cell Death & Disease, 3(10), e409-e409. doi:10.1038/cddis.2012.149   Linsley, P. S., & Fox, C. F. (1980). Direct linkage of EGF to its receptor: Characterization and biological relevance. Journal of Supramolecular Structure, 14(4), 441. doi:10.1002/jss.400140404  Liu, F., & Chernoff, J. (1997). Protein tyrosine phosphatase 1B interacts with and is tyrosine phosphorylated by the epidermal growth factor receptor. The Biochemical Journal, 327(1), 139-145. doi:10.1042/bj3270139   Lundby, A., Secher, A., Lage, K., Nordsborg, N. B., Dmytriyev, A., Lundby, C., & Olsen, J. V. (2012). Quantitative maps of protein phosphorylation sites across 14 different rat organs and tissues. Nature Communications, 3(1) doi:10.1038/ncomms1871   Macias-Silva, M., Vazquez-Victorio, G., & Hernandez-Damian, J. (2010). Anisomycin is a multifunctional drug: More than just a tool to inhibit protein synthesis. Current Chemical Biology, 4: 124. doi:10.2174/2212796811004020124   Madic, J., Jovelet, C., Lopez, J., André, B., Fatien, J., Miran, I., Honoré, A., Mezquita, L., Besse, B., Lacroix, L., & Droniou, M. (2018). EGFR C797S, EGFR T790M and EGFR sensitizing mutations in non-small cell lung cancer revealed by six-color crystal digital PCR. Oncotarget, 9(100), 37393-37406. doi:10.18632/oncotarget.26446   Mahmood, T., & Yang, P. (2012). Western blot: Technique, theory, and trouble shooting. North American Journal of Medical Sciences, 4(9), 429-434. doi:10.4103/1947-2714.100998    Manning, G., Whyte, D. B., Martinez, R., Hunter, T., & Sudarsanam, S. (2002). The protein kinase complement of the human genome. Science, 298(5600), 1912-1934. doi:10.1126/science.1075762   Matthews, H. R. (1995). Protein kinases and phosphatases that act on histidine, lysine, or arginine residues in eukaryotic proteins: A possible regulator of the mitogen-activated protein kinase cascade. Pharmacology and Therapeutics, 67(3), 323-350. doi:10.1016/0163-7258(95)00020-8    221 Menon, V., Thomas, R., Ghale, A. R., Reinhard, C., & Pruszak, J. (2014). Flow cytometry protocols for surface and intracellular antigen analyses of neural cell types. Journal of Visualized Experiments : JoVE, (94) doi:10.3791/52241   Mhawech, P. (2005). 14-3-3 proteins—an update. Cell Res, 15, 228–236. doi:10.1038/sj.cr.7290291 Miao, B., Xiao, Q., Chen, W., Li, Y., & Wang, Z. (2018). Evaluation of functionality for serine and threonine phosphorylation with different evolutionary ages in human and mouse. BMC Genomics, 19(1), 431-9. doi:10.1186/s12864-018-4661-6   Needham, E. J., Parker, B. L., Burykin, T., James, D. E., & Humphrey, S. J. (2019). Illuminating the dark phosphoproteome. Science Signaling, 12(565). doi:10.1126/scisignal.aau8645   Nilsen-Hamilton, M., Hamilton, R. T., Ross Allen, W., & Potter-Perigo, S. (1982). Synergistic stimulation of S6 ribosomal protein phosphorylation and DNA synthesis by epidermal growth factor and insulin in quiescent 3T3 cells. Cell, 31(1), 237-242. doi:10.1016/0092-8674(82)90423-8   Normanno, N., De Luca, A., Bianco, C., Strizzi, L., Mancino, M., Maiello, M. R., Carotenuto, A., De Feo, G., Caponigro, F., & Salomon, D. S. (2006). Epidermal growth factor receptor (EGFR) signaling in cancer. Gene, 366(1), 2-16. doi:10.1016/j.gene.2005.10.018   Novak-Hofer, I., Küng, W., & Eppenberger, U. (1988). Role of extracellular electrolytes in the activation of ribosomal protein S6 kinase by epidermal growth factor, insulin-like growth factor 1, and insulin in ZR-75-1 cells. The Journal of Cell Biology, 106(2), 395-401. doi:10.1083/jcb.106.2.395  Ochnik, A. M., & Baxter, R. C. (2016). Combination therapy approaches to target insulin-like growth factor receptor signaling in breast cancer. Endocrine-Related Cancer, 23(11), R527-R550. doi:10.1530/erc-16-0218   Oda, K., Matsuoka, Y., Funahashi, A., & Kitano, H. (2005). A comprehensive pathway map of epidermal growth factor receptor signaling. Molecular systems biology, 1. doi:10.1038/msb4100014   Olah, Z., Kalman, J., Toth, M., Zvara, A., Santha, M., Ivitz, E., Janka, Z., & Pakaski, M. (2015). Proteomic analysis of cerebrospinal fluid in alzheimer's disease: Wanted dead or alive. J Alzheimer’s Dis, 44: 1303-1312. doi: 10.3233/JAD-140141   Pan, Q., Shai, O., Lee, L. J., Frey, B. J., Blencowe, B. J. (2008) Deep surveying of alternative splicing complexity in the human transcriptome by high-throughput sequencing. Nature Genetics, 40(12): 1413-1415. doi:10.1038/ng.259    222 Pandey, A., Podtelejnikov, A. V., Blagoev, B., Bustelo, X. R., Mann, M., & Lodish, H. F. (2000). Analysis of receptor signaling pathways by mass spectrometry: Identification of vav-2 as a substrate of the epidermal and platelet-derived growth factor receptors. Proceedings of the National Academy of Sciences of the United States of America, 97(1), 179-184. doi:10.1073/pnas.97.1.179   Pawson, T., & Scott, J. D. (2005). Protein phosphorylation in signaling – 50 years and counting. Trends in Biochemical Sciences (Amsterdam. Regular Ed.), 30(6), 286-290. doi:10.1016/j.tibs.2005.04.013  Peacock, J. W., Takeuchi, A., Hayashi, N., Liu, L., Tam, K. J., Al Nakouzi, N., Khazamipour, N., Tombe, T., Dejima, T., Lee, K. C., Shiota, M., Thaper, D., Lee, W. C., Hui, D. H., Kuruma, H., Ivanova, L., Yenki, P., Jiao, I. Z., Khosravi, S., Mui, A. L., Fazli, L., Zoubeidi, A., Daugaard, M., Gleave, M. E., & Ong, C. J. (2018). SEMA3C drives cancer growth by transactivating multiple receptor tyrosine kinases via Plexin B1. EMBO molecular medicine, 10(2), 219–238. doi:10.15252/emmm.201707689   Pelletier, C., Dai, S., Roberts, K. C., Bienek, A., Onysko, J., & Pelletier, L. (2012). Report summary. diabetes in canada: Facts and figures from a public health perspective. Chronic Diseases and Injuries in Canada, 33(1), 53-54.   Piper, A. J., Clark, J. L., Mercado-Matos, J., Matthew-Onabanjo, A. N., Hsieh, C., Akalin, A., & Shaw, L. M. (2019). Insulin receptor substrate-1 (IRS-1) and IRS-2 expression levels are associated with prognosis in non-small cell lung cancer (NSCLC). PloS One, 14(8). doi:10.1371/journal.pone.0220567   Posner, B. I., & Mounier, C. (2006). Transcriptional regulation by insulin: From the receptor to the gene. Canadian Journal of Physiology and Pharmacology, 84(7), 713-724. doi:10.1139/Y05-152   R Core Team. (2019). R: A language and environment for statistical computing. Retrieved from: https://www.R-project.org/.    Radimerski, T., Mini, T., Schneider, U., Wettenhall, R. E. H., Thomas, G., & Jenö, P. (2000). Identification of insulin-induced sites of ribosomal protein S6 phosphorylation in drosophila melanogaster. Biochemistry, 39(19), 5766-5774. doi:10.1021/bi9927484   Red Brewer, M., Yun, C. H., Lai, D., Lemmon, M. A., Eck, M. J., & Pao, W. (2013). Mechanism for activation of mutated epidermal growth factor receptors in lung cancer. Proceedings of the National Academy of Sciences of the United States of America, 110(38), E3595–E3604. doi:10.1073/pnas.1220050110    223 Reddy, R. J., Gajadhar, A. S., Swenson, E. J., Rothenberg, D. A., Curran, T. G., & White, F. M. (2016). Early signaling dynamics of the epidermal growth factor receptor. Proceedings of the National Academy of Sciences of the United States of America, 113(11), 3114-3119. doi:10.1073/pnas.1521288113   Rojas, B. E., Santin, F., Ulloa, R. M., Iglesias, A. A., & Figueroa, C. M. (2018). A fluorometric method for the assay of protein kinase activity. Analytical Biochemistry, 557, 120-122. doi:10.1016/j.ab.2018.07.018   Rokas, A., Krüger, D., & Carroll, S. B. (2005). Animal evolution and the molecular signature of radiations compressed in time. Science, 310(5756), 1933-1938. doi:10.1126/science.1116759   Sacco, F., Perfetto, L., Castagnoli, L., & Cesareni, G. (2012). The human phosphatase interactome: An intricate family portrait. FEBS Letters, 586(17), 2732-2739. doi:10.1016/j.febslet.2012.05.008   Safaei, J., Manuch, J., Gupta, A., Stacho, L., & Pelech, S. (2010). Prediction of human protein kinase substrate specificities. 2010 IEEE International Conference on Bioinformatics and Biomedicine Proceedings, 259-64. doi:10.1109/BIBM.2010.5706573   Safaei, J., Maňuch, J., Gupta, A., Stacho, L., & Pelech, S. (2011a). Prediction of 492 human protein kinase substrate specificities. Proteome Science, 9: Suppl 1-6. doi:10.1186/1477-5956-9-S1-S6   Safaei, J., Manuch, J., Gupta, A., Stacho, L., & Pelech, S. (2011b). Evolutionary conservation of human phosphorylation sites. Proceedings - 2011 IEEE International Conference on Bioinformatics and Biomedicine, 222-227 doi:10.1109/BIBM.2011.58   SafaeiMehranpour, J. (2015). Modelling human cell protein phosphorylation networks (Doctoral Dissertation). University of British Columbia. Retrieved from: https://open.library.ubc.ca/cIRcle/collections/ubctheses/24/items/1.0166161   Sah, N. K., Munshi, A., Kurland, J. F., McDonnell, T. J., Su, B., & Meyn, R. E. (2003). Translation inhibitors sensitize prostate cancer cells to apoptosis induced by tumor necrosis factor-related apoptosis-inducing ligand (TRAIL) by activating c-jun N-terminal kinase. Journal of Biological Chemistry, 278(23), 20593-20602. doi:10.1074/jbc.M211010200  Salomon, D. S., Brandt, R., Ciardiello, F., & Normanno, N. (1995). Epidermal growth factor-related peptides and their receptors in human malignancies. Critical Reviews in Oncology and Hematology, 19(3), 183-232. doi:10.1016/1040-8428(94)00144-I      224 Saran, R., Li, Y., Robinson, B., Ayanian, J., Balkrishnan, R., Bragg-Gresham, J., Chen, J. T. L., Cope, E., Gipson, D., He, K., Herman, W., Heung, M., Hirth, R. A., Jacobsen, S. S., Kalantar-Zadeh, K., Kovesdy, C. P., Leichtman, A. B., Lu, Y., Molnar, M. Z., Morgenstern, H.,  Nallamothu, B., O’Hare, A. M., Pisoni, R., Plattner, B., Port, F. K., Rao, P., Rhee, C. M., Schaubel, D. E., Selewski, D. T., Shahinian, V., Sim, J. J., Song, P., Streja, E., Kurella Tamura, M., Tentori, F., Eggers, P. W., Agodoa, L. Y. C., & Abbott, K. C. (2015). US renal data system 2014 annual data report: Epidemiology of kidney disease in the united states. American Journal of Kidney Diseases, 66(1), A7-A7. doi:10.1053/j.ajkd.2015.05.001   Scheeff, E. D., Eswaran, J., Bunkoczi, G., Knapp, S., & Manning, G. (2009). Structure of the pseudokinase VRK3 reveals a degraded catalytic site, a highly conserved kinase fold, and a putative regulatory binding site. Structure, 17(1), 128-138. doi:10.1016/j.str.2008.10.018   Schröder, C., Jacob, A., Tonack, S., Radon, T. P., Sill, M., Zucknick, M., Rüffer, S., Costello, E., Neoptolemos, J. P., Crnogorac-Jurcevic, T., Bauer, A., Fellenberg, K., & Hoheisel, J. D. (2010). Dual-color proteomic profiling of complex samples with a microarray of 810 cancer-related antibodies. Molecular & Cellular Proteomics, 9(6), 1271-1280. doi:10.1074/mcp.M900419-MCP200   Sefried, S., Häring, H., Weigert, C., & Eckstein, S. S. (2018). Suitability of hepatocyte cell lines HepG2, AML12 and THLE-2 for investigation of insulin signalling and hepatokine gene expression. Open Biology, 8(10), 180147. doi:10.1098/rsob.180147   Seshacharyulu, P., Ponnusamy, M. P., Haridas, D., Jain, M., Ganti, A. K., & Batra, S. K. (2012). Targeting the EGFR signaling pathway in cancer therapy. Expert Opinion on Therapeutic Targets, 16(1), 15-31. doi:10.1517/14728222.2011.648617   Sha, F., Gencer, E. B., Georgeon, S., Koide, A., Yasui, N., Koide, S., & Hantschel, O. (2013). Dissection of the BCR-ABL signaling network using highly specific monobody inhibitors to the SHP2 SH2 domains. Proceedings of the National Academy of Sciences of the United States of America, 110(37), 14924-14929. doi:10.1073/pnas.1303640110  Shi, W., Meng, Z., Chen, Z.,Luo, J., & Liu, L. (2011). Proteome analysis of human pancreatic cancer cell lines with highly liver metastatic potential by antibody microarray. Molecular and Cellular Biochemistry, 347(1), 117-125. doi:10.1007/s11010-010-0619-y   Solari, F. A., Dell'Aica, M., Sickmann, A., & Zahedi, R. P. (2015). Why phosphoproteomics is still a challenge. Molecular BioSystems, 11(6), 1487-1493. doi:10.1039/C5MB00024F   Song, J., Wang, H., Wang, J., Leier, A., Marquez-Lago, T., Yang, B., Zhang, Z., Akutsu, T., Webb, G. I., & Daly, R. (2017). PhosphoPredict: A bioinformatics tool for prediction of human kinase-specific phosphorylation substrates and sites by integrating heterogeneous feature selection. Scientific Reports, 7(1), 6862-19. doi:10.1038/s41598-017-07199-4    225 Storey, J. D., Bass, A. J., Dabney, A., & Robinson, D. (2019). qvalue: Q-value estimation for false discovery rate control. R package version 2.18.0. Retrieved from: http://github.com/jdstorey/qvalue  Storey, J. D., & Tibshirani, R. (2003). Statistical significance for genomewide studies. Proceedings of the National Academy of Sciences of the United States of America, 100(16), 9440-9445. doi:10.1073/pnas.1530509100   Tabb, D. L., Vega-Montoto, L., Rudnick, P. A., Variyath, A. M., Ham, A. L., Bunk, D. M., Kilpatrick, L. E., Billheimer, D. D., Blackman, R. K., Cardasis, H. L., Carr, S. A., Clauser, K. R., Jaffe, J. D., Kowalski, K. A., Neubert, T. A., Regnier, F. E., Schilling, B., Tegeler, T. J., Wang, M., Wang, P., Whiteaker, J. R., Zimmerman, L. J., Fisher, S. J., Gibson, B. W., Kinsinger, C. R., Mesri, M., Rodriguez, H., Stein, S. E., Tempst, P., Paulovich, A. G., Liebler, D. C., & Spiegelman, C. (2010). Repeatability and reproducibility in proteomic identifications by liquid Chromatography−Tandem mass spectrometry. Journal of Proteome Research, 9(2), 761-776. doi:10.1021/pr9006365   The Emerging Risk Factors Coalition. (2011). Diabetes mellitus, fasting blood glucose concentration, and risk of vascular disease: A collaborative meta-analysis of 102 prospective studies. Journal of Vascular Surgery, 53(2), 548-549. doi:10.1016/j.jvs.2010.12.016   Tinti, M., Nardozza, A. P., Ferrari, E., Sacco, F., Corallino, S., Castagnoli, L., & Cesareni, G. (2012). The 4G10, pY20 and p-TYR-100 antibody specificity: Profiling by peptide microarrays. New Biotechnology, 29(5), 571-577. doi:10.1016/j.nbt.2011.12.001   Tonks, N. K. (2006). Protein tyrosine phosphatases: From genes, to function, to disease. Nature Reviews Molecular Cell Biology, 7(11), 833-846. doi:10.1038/nrm2039   Tran, H., Brunet, A., Griffith, E. C., & Greenberg, M. E. (2003). The many forks in FOXO's road. Science's Signal Transduction Knowledge Environment, 2003(172), RE5. doi:10.1126/stke.2003.172.re5   Tricker, E. M., Xu, C., Uddin, S., Capelletti, M., Ercan, D., Ogino, A., Pratilas, C. A., Rosen, N., Gray, N. S., Wong, K., & Jänne, P. A. (2015). Combined EGFR/MEK inhibition prevents the emergence of resistance in EGFR-mutant lung cancer. Cancer Discovery, 5(9), 960-971. doi:10.1158/2159-8290.CD-15-0063    Ubersax, J. A., & Ferrell, J. E. (2007). Mechanisms of specificity in protein phosphorylation. Nature Reviews Molecular Cell Biology, 8(7), 530-541. doi: 10.1038/nrm2203   Varedi K, S. M., Ventura, A. C., Merajver, S. D., & Lin, X. N. (2010). Multisite phosphorylation provides an effective and flexible mechanism for switch-like protein degradation. PloS one, 5(12), e14029. doi:10.1371/journal.pone.0014029    226 Velu, T. J., Beguinot, L., Vass, W. C., Willingham, M. C., Merlino, G. T., Pastan, I., & Lowy, D. R. (1987). Epidermal growth factor-dependent transformation by a human EGF receptor proto-oncogene. Science, 238(4832), 1408-1410. doi:10.1126/science.3500513   Vintonyak, V. V., Antonchick, A. P., Rauh, D., & Waldmann, H. (2009). The therapeutic potential of phosphatase inhibitors. Current Opinion in Chemical Biology, 13(3), 272-283. doi:10.1016/j.cbpa.2009.03.021  Voudouri, K., Berdiaki, A., Tzardi, M., Tzanakakis, G. N., & Nikitovic, D. (2015). Insulin-like growth factor and epidermal growth factor signaling in breast cancer cell growth: Focus on endocrine resistant disease. Analytical Cellular Pathology (Amsterdam), 2015, 975495-10. doi:10.1155/2015/975495   Wagner, M. J., Stacey, M. M., Liu, B. A., & Pawson, T. (2013). Molecular mechanisms of SH2- and PTB-domain-containing proteins in receptor tyrosine kinase signaling. Cold Spring Harbor Perspectives in Biology, 5(12), doi:10.1101/cshperspect.a008987   Wang, C., Xu, H., Lin, S., Deng, W., Zhou, J., Zhang, Y., Shi, Y., Peng, D., & Xue, Y. (2020). GPS 5.0: An update on the prediction of kinase-specific phosphorylation sites in proteins. Genomics, Proteomics & Bioinformatics, doi:10.1016/j.gpb.2020.01.001   Wang, J., Xue, H., Chen, Y., Xu, H., Lin, S., Tang, X., & Chen, C. (2019). Evaluation of insulin-mediated regulation of AKT signaling in childhood acute lymphoblastic leukemia. Journal of Pediatric hematology/oncology, 41(2), 96-104. doi:10.1097/MPH.0000000000001425   Wang, Z., Li, Y., Hou, B., Pronobis, M. I., Wang, M., Wang, Y., Cheng, G., Weng, W., Wang, Y., Tang, Y., Xu, X., Pan, R., Lin, F., Wang, N., Chen, Z., Wang, S., Ma, L. Z., Li, Y., Huang, D., Jiang, L., Wang, Z., Zeng, W., Zhang, Y., Du, X., Lin, Y., Li, Z., Xia, Q., Geng, J., Dai, H., Yu, Y., Zhao, X., Yuan, Z., Yan, J., Nie, Q., Zhang, X., Wang, K., Chen, F., Zhang, Q., Zhu, Y., Zheng, S., Poss, K. D., Tao, S., & Meng, X. (2020). An array of 60,000 antibodies for proteome-scale antibody generation and target discovery. Science Advances, 6(11), doi:10.1126/sciadv.aax2271   Wei, M. L., Duan, P., Wang, Z. M., Ding, M., & Tu, P. (2017). High glucose and high insulin conditions promote MCF-7 cell proliferation and invasion by upregulating IRS1 and activating the Ras/Raf/ERK pathway. Molecular medicine reports, 16(5), 6690–6696. doi:10.3892/mmr.2017.7420   Wu, J., Gage, D. A., & Watson, J. T. (1996). A strategy to locate cysteine residues in proteins by specific chemical cleavage followed by matrix-assisted laser Desorption/Ionization time-of-flight mass spectrometry. Analytical Biochemistry, 235(2), 161-174. doi:10.1006/abio.1996.0108   Wu, Z. L. (2011). Phosphatase-coupled universal kinase assay and kinetics for first-order-rate coupling reaction. PloS One, 6(8), e23172. doi:10.1371/journal.pone.0023172   227 Xu, Y., Richert, N., Ito, S., Merlino, G. T., & Pastan, I. (1984). Characterization of epidermal growth factor receptor gene expression in malignant and normal human cell lines. Proceedings of the National Academy of Sciences - PNAS, 81(23), 7308-7312. doi:10.1073/pnas.81.23.7308  Yan, K., Gao, L., Cui, Y., Zhang, Y., & Zhou, X. (2016). The cyclic AMP signaling pathway: Exploring targets for successful drug discovery (review). Molecular Medicine Reports, 13(5), 3715-3723. doi:10.3892/mmr.2016.5005   Yang, Q., Inoki, K., Kim, E., & Guan, K. (2006). TSC1/TSC2 and rheb have different effects on TORC1 and TORC2 activity. Proceedings of the National Academy of Sciences of the United States of America, 103(18), 6811-6816. doi:10.1073/pnas.0602282103   Yang, Z., & Tam, K. Y. (2018). Combination strategies using EGFR-TKi in NSCLC therapy: Learning from the gap between pre-clinical results and clinical outcomes. International Journal of Biological Sciences, 14(2), 204-216. doi:10.7150/ijbs.22955   Yue, L., & Pelech, S. (2018). Applications of high content antibody microarrays for biomarker discovery and tracking cellular signaling. Advances in Proteomics Bioinformatics, APBI-107. DOI: 10.29011/APBI -107. 100007   Yue, L., Sam, C., Arora, N., F.H. Winkler, D., & Pelech, S. (2017). Antibody microarray and immunoblotting analyses of the EGF signaling phosphorylation network in human A431 epidermoid carcinoma cells. Clinical Proteomics and Bioinformatics, 2(1), 1-10. doi:10.15761/CPB.1000119   Zegzouti, H., Zdanovskaia, M., Hsiao, K., & Goueli, S. A. (2009). ADP-glo: A bioluminescent and homogeneous ADP monitoring assay for kinases. Assay and Drug Development Technologies, 7(6), 560-572. doi:10.1089/adt.2009.0222   Zhang, F., Wang, S., Yin, L., Yang, Y., Guan, Y., Wang, W., Xu, H., & Tao, N. (2015). Quantification of epidermal growth factor receptor expression level and binding kinetics on cell surfaces by surface plasmon resonance imaging. Analytical Chemistry, 87(19), 9960-9965. doi:10.1021/acs.analchem.5b02572   Zhang, H., Shi, X., & Pelech, S. (2016). Monitoring protein kinase expression and phosphorylation in cell lysates with antibody microarrays. Methods in Molecular Biology 1360: 107-122. doi:10.1007/978-1-4939-3073-9_9   Zhu, Z., Tang, J., Gang, D., Wang, M., Wang, J., Lei, Z., Feng, Z., Fang, M., & Yan, L. (2015). Antibody microarray profiling of osteosarcoma cell serum for identifying potential biomarkers. Molecular Medicine Reports, 12(1), 1157. doi:10.3892/mmr.2015.3535    228 Zhvansky, E. S., Pekov, S. I., Sorokin, A. A., Shurkhay, V. A., Eliferov, V. A., Potapov, A. A., Nikolaev, E. N., & Popov, I. A. (2019). Metrics for evaluating the stability and reproducibility of mass spectra. Scientific Reports, 9(1), 914-8. doi:10.1038/s41598-018-37560-0      229 Appendices  Appendix A. Kinex G-series peptide macroarray details.   230 Appendix B. Key EGF-induced changes in protein phosphorylation in A431 cells as detected with the Kinex™ KAM-900P antibody microarray.    231     232 Averages are calculated from at least three separate microarray analyses with duplicate measurements performed for each analysis and the % standard deviations are shown. "True positive" (TP) and "true negative" (TN) correspond to correlation between the positive and negative effects, respectively, of EGF treatment between the antibody microarray and Western blotting results for the target protein. "False positive" (FP) results correspond to an apparent change with EGF treatment that could not be confirmed in the target protein by Western blotting. FPc = false positive, but a similar EGF-induced change in one or more cross-reactive proteins on the Western blot that correlates with the antibody microarray results. "False negative" (FN) corresponds to an EGF- associated change on the Western blot that was not evident with the antibody microarray. ‡ Reacts with EGF receptor (EGFR) in EGF-treated A431 cells on Western blots or a protein that migrates on SDS-PAGE gels like the EGFR. Significance * p≤0.05, ** p≤0.01, *** p≤0.005, **** p≤0.001 based on paired, two-tailed Student T-test.           233  Appendix C. Key EGF-induced changes in protein tyrosine phosphorylation in A431 cells as detected with the Kinex™ KAM-1150 antibody microarray.       234   A431 cervical carcinoma cells were serum starved and treated with and without 100 ng/ml epidermal growth factor (EGF) for 1-20 min and incubated onto the KAM1150 antibody microarray, which consists of all pan specific antibodies. Subsequent to incubation with dye-labelled generic phosphotyrosine antibody, quadruplicate measurements for each spot was averaged and the % change from control (%CFC) values for antibody spot for each treatment length were calculated. Student T tests were conducted to generate p values, and any antibody leads that changed -35% or lesser or +35% and greater, which were significant (p<0.05), within any timepoint were pooled.  235 Appendix D. Kinex™ KAM-900P antibody microarray target phosphorylation sites changed in response to EGF treatment in cancer cell lines.  Target Name P-site Uniprot ID Antibody Code Direction of Change % CFC Standard Deviation / Sum % Error p Value Cell Line Treatment Detection Tool A6r Y309 Q6IBS0 PK502 Up 78 100   LNCaP EGF treatment 900P Microarray AAK1 S637 Q2M2I8 PK503 Up 50 11 0.047 A549 EGF treatment 900P Microarray AAK1 S637 Q2M2I8 PK503 Up 121 18   LNCaP EGF treatment 900P Microarray Abl1 Y139 P00519 PK504 Up 67 26   DU145 EGF treatment 900P Microarray Abl1 Y226 P00519 PK505 Down -46 -14 0.016 HeLa EGF treatment 900P Microarray Abl1 Y257 P00519 PK506 Up 91 206   LNCaP EGF treatment 900P Microarray Abl1 Y264 P00519 PK507 Up 404 82   LNCaP EGF treatment 900P Microarray Abl1 Y469 P00519 PK508 Up 108 15   LNCaP EGF treatment 900P Microarray AcCoA carboxylase S80 Q13085 PN002 Up 95 12   LNCaP EGF treatment 900P Microarray ACK1 Y284 Q07912 PK511 Up 71 45 0.091 A431 EGF treatment 900P Microarray ACK1 Y518 Q07912 PK512 Up 77 74   LNCaP EGF treatment 900P Microarray ACTB Y53 P60709 PN501 Down -53 -14 0.006 HeLa EGF treatment 900P Microarray ACTB Y294 P60709 PN500 Up 118 11   LNCaP EGF treatment 900P Microarray ACTB Y53 P60709 PN501 Up 188 14   LNCaP EGF treatment 900P Microarray ACTN1 Y246 P12814 PN502 Up 118 36   LNCaP EGF treatment 900P Microarray Adducin a S726 Q9UEY8 PN003-PN004 Up 265 30   LNCaP EGF treatment 900P Microarray  236 Target Name P-site Uniprot ID Antibody Code Direction of Change % CFC Standard Deviation / Sum % Error p Value Cell Line Treatment Detection Tool Akt1 Y326 P31749 PK517 Down -46 -20 0.027 HeLa EGF treatment 900P Microarray Akt1 T308 P31749 PK515 Up 73 25   LNCaP EGF treatment 900P Microarray ALK Y1096 Q9UM73 PK519 Up 60 30 0.039 A431 EGF treatment 900P Microarray ALK Y1092 Q9UM73 PK518 Up 47 12   LNCaP EGF treatment 900P Microarray ANKRD3 S438 P57078 PK523 Up 351 29   LNCaP EGF treatment 900P Microarray ANXA1 Y207 P04083 PN503 Up 112 28   LNCaP EGF treatment 900P Microarray ANXA2 Y238 P07355 PN504 Up 105 24   LNCaP EGF treatment 900P Microarray APP T743 P05067 PN189 Up 170 22   LNCaP EGF treatment 900P Microarray Arrestin b1 S412 P49407 PN133 Up 95 266 0.616 A431 EGF treatment 900P Microarray Arrestin b1 S412 P49407 PN133 Up 43 25   LNCaP EGF treatment 900P Microarray ASK1 T838 Q99683 PK525 Up 41 150   LNCaP EGF treatment 900P Microarray ATF2 S112 P15336 PN115 Up 95 23   LNCaP EGF treatment 900P Microarray ATM S1981 Q13315 PK526 Up 81 15   LNCaP EGF treatment 900P Microarray ATM Y2969 Q13315 PK527 Up 85 24   LNCaP EGF treatment 900P Microarray ATR S435+S436 Q13535 PK528 Up 49 23 0.181 A549 EGF treatment 900P Microarray AurKA T287+T288 O14965 PK529 Up 92 46   LNCaP EGF treatment 900P Microarray Axl Y702+Y703 P30530 PK533 Up 49 21 0.282 A549 EGF treatment 900P Microarray  237 Target Name P-site Uniprot ID Antibody Code Direction of Change % CFC Standard Deviation / Sum % Error p Value Cell Line Treatment Detection Tool B-Raf S729 P15056 PK535 Down -49 -35 0.028 A431 EGF treatment 900P Microarray B-Raf S729 P15056 PK535 Up 141 33   LNCaP EGF treatment 900P Microarray B23 (NPM) T199 P06748 PN008 Up 64 74   LNCaP EGF treatment 900P Microarray B23 (NPM) T234+T237 P06748 PN009 Up 109 35   LNCaP EGF treatment 900P Microarray BARK1 S670 P25098 PK536 Down -54 -36 0.014 A431 EGF treatment 900P Microarray BARK1 S670 P25098 PK536 Down -45 -14 0.025 HeLa EGF treatment 900P Microarray BARK1 S670 P25098 PK536 Up 46 45   LNCaP EGF treatment 900P Microarray BARK1 Y356 P25098 PK537 Up 64 34   LNCaP EGF treatment 900P Microarray Bcr Y177 P11274 PK538 Up 91 75   LNCaP EGF treatment 900P Microarray Bcr Y591 P11274 PK539 Up 120 14   LNCaP EGF treatment 900P Microarray BLK Y389 P51451 PK543 Up 114 51 0.091 A431 EGF treatment 900P Microarray BLK Y188 P51451 PK542 Up 51 22 0.250 A549 EGF treatment 900P Microarray BLK Y389 P51451 PK543 Up 93 71 0.140 A549 EGF treatment 900P Microarray BLK Y188 P51451 PK542 Up 75 64   LNCaP EGF treatment 900P Microarray BLK Y389 P51451 PK543 Up 93 31   LNCaP EGF treatment 900P Microarray BLNK Y84 O75498 PN013 Up 195 28   LNCaP EGF treatment 900P Microarray BMPR2 S375 Q13873 PK544 Up 46 13 0.228 A549 EGF treatment 900P Microarray  238 Target Name P-site Uniprot ID Antibody Code Direction of Change % CFC Standard Deviation / Sum % Error p Value Cell Line Treatment Detection Tool BMPR2 S375 Q13873 PK544 Up 55 111   LNCaP EGF treatment 900P Microarray Bmx Y40 P51813 PK545 Down -44 -11 0.005 HeLa EGF treatment 900P Microarray BMX (Etk) Y40 P51813 PK003 Down -53 -55 0.364 A549 EGF treatment 900P Microarray BRK S446+Y447 Q13882 PK547 Down -45 -33 0.082 A431 EGF treatment 900P Microarray BRK S446+Y447 Q13882 PK547 Up 56 15 0.132 A549 EGF treatment 900P Microarray BRK Y342 Q13882 PK548 Down -41 -17 0.117 HeLa EGF treatment 900P Microarray BRK S446+Y447 Q13882 PK547 Down -42 -21 0.022 HeLa EGF treatment 900P Microarray BRK S446+Y447 Q13882 PK547 Up 41 8   LNCaP EGF treatment 900P Microarray BRSK1 T189 Q8TDC3 PK549 Up 55 26 0.135 A549 EGF treatment 900P Microarray Btk Y223+Y225 Q06187 PK550 Up 53 17 0.083 A549 EGF treatment 900P Microarray Btk Y223+Y225 Q06187 PK550 Down -54 -22 0.120 HeLa EGF treatment 900P Microarray CaMK1d T180 Q8IU85 PK554 Down -40 -17 0.042 HeLa EGF treatment 900P Microarray CaMK4 T200 Q16566 PK556 Up 124 96   LNCaP EGF treatment 900P Microarray Cas-L Y166 Q14511 PN505 Down -67 -63 0.095 A431 EGF treatment 900P Microarray Cas-L Y166 Q14511 PN505 Up 47 33 0.774 A549 EGF treatment 900P Microarray Cas-L Y166 Q14511 PN505 Down -44 -17 0.122 HeLa EGF treatment 900P Microarray Catenin a S641 P35221 PN162 Up 132 19   LNCaP EGF treatment 900P Microarray  239 Target Name P-site Uniprot ID Antibody Code Direction of Change % CFC Standard Deviation / Sum % Error p Value Cell Line Treatment Detection Tool Caveolin 2 S36 P51636 PN018 Down -49 586   LNCaP EGF treatment 900P Microarray CDC7 T376 O00311 PK558 Down -41 -55 0.372 A431 EGF treatment 900P Microarray CDK1 T14+Y15 P06493 PK560 Up 123 234 0.378 A431 EGF treatment 900P Microarray CDK1 T14 P06493 PK559 Up 46 6 0.152 A549 EGF treatment 900P Microarray CDK1 T161 P06493 PK008-1 Up 55 38 0.404 A549 EGF treatment 900P Microarray CDK1 Y19 P06493 PK563 Up 61 59 0.574 A549 EGF treatment 900P Microarray CDK1 T14 P06493 PK559 Up 56 31   LNCaP EGF treatment 900P Microarray CDK1/2 T14+Y15 P06493 PK006 Up 213 251 0.462 A549 EGF treatment 900P Microarray CDK11A T583 Q9UQ88 PK565 Down -47 30   DU145 EGF treatment 900P Microarray CDK12 S383+S385 Q9NYV4 PK566 Down -58 -42 0.018 A431 EGF treatment 900P Microarray CDK12 S383+S385 Q9NYV4 PK566 Up 86 12   LNCaP EGF treatment 900P Microarray CDK2 T160 P24941 PK568 Down -61 -33 0.005 A431 EGF treatment 900P Microarray CDK2 T160 P24941 PK568 Up 44 52   DU145 EGF treatment 900P Microarray CDK2 T160 P24941 PK568 Up 86 69   LNCaP EGF treatment 900P Microarray CDK5 Y15 Q00535 PK570 Up 86 27   LNCaP EGF treatment 900P Microarray CDK9 T186 P50750 PK575 Up 46 196   LNCaP EGF treatment 900P Microarray CDKL5 Y171 O76039 PK576 Up 45 16 0.201 A431 EGF treatment 900P Microarray  240 Target Name P-site Uniprot ID Antibody Code Direction of Change % CFC Standard Deviation / Sum % Error p Value Cell Line Treatment Detection Tool CDKL5 Y171 O76039 PK576 Up 42 25   LNCaP EGF treatment 900P Microarray Chk1 S317 O14757 PK578 Up 47 23 0.285 A549 EGF treatment 900P Microarray Chk1 S280 O14757 PK162 Up 162 274   LNCaP EGF treatment 900P Microarray Chk1 S317 O14757 PK578 Up 73 44   LNCaP EGF treatment 900P Microarray Chk2 T383 O96017 PK580 Up 50 70   LNCaP EGF treatment 900P Microarray CK2a T360+S362 P68400 PK167 Down -44 18   DU145 EGF treatment 900P Microarray CLK1 S337 P49759 PK583 Up 44 19 0.155 A549 EGF treatment 900P Microarray CLK1 S337+T338 P49759 PK584 Up 61 35 0.146 A549 EGF treatment 900P Microarray Crystallin aB S19 P02511 PN025 Up 48 36 0.404 HeLa EGF treatment 900P Microarray Crystallin aB S45 P02511 PN110 Up 66 19 0.376 HeLa EGF treatment 900P Microarray Crystallin aB S19 P02511 PN025 Up 45 56   LNCaP EGF treatment 900P Microarray CSF1R Y699 P07333 PK587 Down -46 -23 0.045 HeLa EGF treatment 900P Microarray CSF1R Y809 P07333 PK588 Up 148 93   LNCaP EGF treatment 900P Microarray Csk Y184 P41240 PK589 Up 61 17 0.158 A549 EGF treatment 900P Microarray Csk Y184 P41240 PK589 Up 57 49   LNCaP EGF treatment 900P Microarray DAPK S269 Q12852 PK590 Up 61 32   LNCaP EGF treatment 900P Microarray DDR1 Y797 Q08345 PK592 Down -41 -14 0.013 HeLa EGF treatment 900P Microarray  241 Target Name P-site Uniprot ID Antibody Code Direction of Change % CFC Standard Deviation / Sum % Error p Value Cell Line Treatment Detection Tool DDR2 Y740 Q16832 PK594 Up 325 274 0.038 A431 EGF treatment 900P Microarray DDR2 Y736 Q16832 PK593 Down -56 -25 0.063 HeLa EGF treatment 900P Microarray eEF1A1 Y141 P68104 PN509 Down -40 42   DU145 EGF treatment 900P Microarray EFNB2 Y316 P52799 PN173 Down -41 41   DU145 EGF treatment 900P Microarray EGFR Y1069 P00533 PK599 Down -56 -346 0.771 A431 EGF treatment 900P Microarray EGFR Y1110 P00533 PK123 Up 57 36 0.214 A431 EGF treatment 900P Microarray EGFR Y869 P00533 PK602 Up 700 1059 0.236 A431 EGF treatment 900P Microarray EGFR Y1069 P00533 PK599 Down -52 72   LNCaP EGF treatment 900P Microarray eIF2a S52 P05198 PN028-2 Down -63 -45 0.065 A549 EGF treatment 900P Microarray eIF2a S52 P05198 PN028-2 Up 52 12 0.147 HeLa EGF treatment 900P Microarray eIF4G S1231 Q04637 PN193 Down -42 -26 0.289 A549 EGF treatment 900P Microarray eIF4G S1231 Q04637 PN193 Up 75 35   LNCaP EGF treatment 900P Microarray Elk-1 S389 P19419 PN170 Up 42 30   LNCaP EGF treatment 900P Microarray EML4 Y226 Q9HC35 PN510 Up 45 18 0.121 A549 EGF treatment 900P Microarray ENO1 Y44 P06733 PN511 Up 41 1495 0.885 A549 EGF treatment 900P Microarray ENO2 Y25 P09104 PN512 Down -44 -16 0.021 HeLa EGF treatment 900P Microarray EphA1 Y781 P21709 PK605 Up 42 18 0.159 A431 EGF treatment 900P Microarray  242 Target Name P-site Uniprot ID Antibody Code Direction of Change % CFC Standard Deviation / Sum % Error p Value Cell Line Treatment Detection Tool EphA1 Y781 P21709 PK605 Up 71 38   LNCaP EGF treatment 900P Microarray EphA2 Y772 P29317 PK607 Down -41 -19 0.109 HeLa EGF treatment 900P Microarray EphA2 Y772 P29317 PK607 Up 47 21   LNCaP EGF treatment 900P Microarray EphA2 Y588 P29317 PK606 Up 275 50   LNCaP EGF treatment 900P Microarray EphB2 Y780 P29323 PK610 Down -48 -17 0.065 HeLa EGF treatment 900P Microarray ErbB2 Y1248 P04626 PK613 Up 48 10 0.031 A431 EGF treatment 900P Microarray ErbB2 Y877 P04626 PK615 Down -42 -20 0.050 HeLa EGF treatment 900P Microarray ErbB2 Y877 P04626 PK615 Up 59 24   LNCaP EGF treatment 900P Microarray ERBB2IP Y1104 Q96RT1 PN513 Up 222 9   LNCaP EGF treatment 900P Microarray ErbB3 Y1289 P21860 PK616 Up 56 22 0.160 A431 EGF treatment 900P Microarray ErbB3 Y1307 P21860 PK617 Up 100 26   LNCaP EGF treatment 900P Microarray ErbB3 Y1328 P21860 PK163 Up 170 26   LNCaP EGF treatment 900P Microarray ERK1 Y204 P27361 PK864 (PK014-6) Down -42 -31 0.213 A431 EGF treatment 900P Microarray Erk1 T202 P27361 PK170-PK171 Down -42 -22 0.215 A549 EGF treatment 900P Microarray Erk1 T202 P27361 PK170-PK171 Up 91 68   LNCaP EGF treatment 900P Microarray Erk1 Y204 P27361 PK168-PK169 Up 62 25   LNCaP EGF treatment 900P Microarray ERK1 Y204+T207 P27361 PK866 (PK014-8) Up 46 18   LNCaP EGF treatment 900P Microarray  243 Target Name P-site Uniprot ID Antibody Code Direction of Change % CFC Standard Deviation / Sum % Error p Value Cell Line Treatment Detection Tool ERK3 S189 Q16659 PK623 Up 64 14   LNCaP EGF treatment 900P Microarray ERK5 Y221 Q13164 PK626 Down -43 -15 0.022 HeLa EGF treatment 900P Microarray ERK5 Y221 Q13164 PK626 Up 89 22   LNCaP EGF treatment 900P Microarray Ezrin T567 P15311 PN174 Up 49 17 0.423 HeLa EGF treatment 900P Microarray FAK Y397 Q05397 PK627 Down -44 -42 0.139 A431 EGF treatment 900P Microarray FAK Y397 Q05397 PK017-1 Up 50 24 0.146 A431 EGF treatment 900P Microarray FAK Y397 Q05397 PK627 Up 50 30 0.358 A549 EGF treatment 900P Microarray FAK S722 Q05397 PK020-3 Up 57 37 0.466 HeLa EGF treatment 900P Microarray FAK Y576+Y577 Q05397 PK151 Up 125 71 0.169 HeLa EGF treatment 900P Microarray FAK Y397 Q05397 PK017-1 Up 54 18   LNCaP EGF treatment 900P Microarray FAK Y397 Q05397 PK627 Up 69 38   LNCaP EGF treatment 900P Microarray FAK S722 Q05397 PK020-3 Up 164 11   LNCaP EGF treatment 900P Microarray FES Y713+S716 P07332 PK633 Up 64 19   LNCaP EGF treatment 900P Microarray FGFR2 Y656+Y657 P21802 PK635 Down -45 -17 0.035 HeLa EGF treatment 900P Microarray FGR Y208+Y209 P09769 PK638 Up 46 15 0.049 A431 EGF treatment 900P Microarray FGR Y412 P09769 PK639 Down -57 -67 0.434 HeLa EGF treatment 900P Microarray Flt3 Y842 P36888 PK640 Up 49 13 0.307 A431 EGF treatment 900P Microarray  244 Target Name P-site Uniprot ID Antibody Code Direction of Change % CFC Standard Deviation / Sum % Error p Value Cell Line Treatment Detection Tool Fos T232 P01100 PN033 Up 96 22   LNCaP EGF treatment 900P Microarray Fused S159 Q9NRP7 PK643 Up 86 29   LNCaP EGF treatment 900P Microarray Fyn Y213+Y214 P06241 PK644 Down -45 -43 0.287 HeLa EGF treatment 900P Microarray Fyn Y213+Y214 P06241 PK644 Up 41 24   LNCaP EGF treatment 900P Microarray G6PD Y401 P11413 PN515 Up 80 35   LNCaP EGF treatment 900P Microarray GATA1 S142 P15976 PN196 Up 375 134 0.227 HeLa EGF treatment 900P Microarray GATA1 S142 P15976 PN196 Up 118 126   LNCaP EGF treatment 900P Microarray GCK S170 Q12851 PK646 Down -40 -34 0.102 A431 EGF treatment 900P Microarray GFAP S8 P14136 PN034 Up 53 28 0.203 HeLa EGF treatment 900P Microarray GRK2 (BARK1) S670 P25098 PK025 Up 42 41   LNCaP EGF treatment 900P Microarray GSK3a Y279 P49840 PK028-PK029-1 Down -44 -16 0.084 A549 EGF treatment 900P Microarray GSK3a Y279 P49840 PK028-PK029-1 Up 70 19 0.126 HeLa EGF treatment 900P Microarray GSK3a Y284+Y285 P49840 PK650 Up 43 19 0.864 HeLa EGF treatment 900P Microarray GSK3a S278+Y279 P49840 PK647 Up 132 71   LNCaP EGF treatment 900P Microarray GTF2F1 S385+T389 P35269 PK651 Up 42 10 0.072 A549 EGF treatment 900P Microarray HCA59 Y147 Q9NZ63 PN518 Up 42 80   DU145 EGF treatment 900P Microarray HCA59 Y147 Q9NZ63 PN518 Down -48 24   LNCaP EGF treatment 900P Microarray  245 Target Name P-site Uniprot ID Antibody Code Direction of Change % CFC Standard Deviation / Sum % Error p Value Cell Line Treatment Detection Tool IGF1R Y1161+ T1163 P08069 PK657 Down -58 -53 0.314 HeLa EGF treatment 900P Microarray IGF1R Y1280 P08069 PK152 Down -47 60   LNCaP EGF treatment 900P Microarray IKKa T23 O15111 PK154 Down -48 28   LNCaP EGF treatment 900P Microarray ILK1 Y351 Q13418 PK662 Down -42 -28 0.059 A431 EGF treatment 900P Microarray ILK1 Y351 Q13418 PK662 Up 51 43   DU145 EGF treatment 900P Microarray IR (INSR) Y999 P06213 PK032-1 Up 56 83   LNCaP EGF treatment 900P Microarray IRAK1 T387 P51617 PK664 Up 44 77   DU145 EGF treatment 900P Microarray IRAK1 T387 P51617 PK664 Up 53 16   LNCaP EGF treatment 900P Microarray IRS1 S312 P35568 PN117 Down -43 -25 0.001 HeLa EGF treatment 900P Microarray IRS1 Y612 P35568 PN045 Down -42 -13 0.046 HeLa EGF treatment 900P Microarray ITK Y512 Q08881 PK666 Down -96 8414   DU145 EGF treatment 900P Microarray JAK1 Y1034 P23458 PK126 Down -42 -17 0.042 HeLa EGF treatment 900P Microarray JAK2 Y1007+ Y1008 O60674 PK667 Down -43 -25 0.035 HeLa EGF treatment 900P Microarray JAK3 Y980+Y981 P52333 PK669 Up 64 12   LNCaP EGF treatment 900P Microarray JNK1 Y185 P45983 PK670 Down -40 71   LNCaP EGF treatment 900P Microarray JNK1/2/3 T183+Y185 P45983 PK035-2 Up 50 18 0.281 HeLa EGF treatment 900P Microarray Jun Y170 P05412 PN155 Up 42 27 0.172 A431 EGF treatment 900P Microarray  246 Target Name P-site Uniprot ID Antibody Code Direction of Change % CFC Standard Deviation / Sum % Error p Value Cell Line Treatment Detection Tool Kit S821+Y823 P10721 PK674 Up 128 36   LNCaP EGF treatment 900P Microarray Kit Y936 P10721 PK038 Up 79 67   LNCaP EGF treatment 900P Microarray Ksr1 S404 Q8IVT5 PK675 Up 93 34   LNCaP EGF treatment 900P Microarray LATS1 S464 O95835 PK677 Down -53 -23 0.024 HeLa EGF treatment 900P Microarray Lck S158 P06239 PK039 Up 149 64   LNCaP EGF treatment 900P Microarray Lck Y192 P06239 PK040 Up 48 35   LNCaP EGF treatment 900P Microarray Lck Y192 P06239 PK679 Up 153 120   LNCaP EGF treatment 900P Microarray Lck Y505 P06239 PK041 Up 94 19   LNCaP EGF treatment 900P Microarray LOK T952 O94804 PK686 Down -72 -71 0.110 A431 EGF treatment 900P Microarray LOK S191 O94804 PK685 Up 45 22 0.099 A431 EGF treatment 900P Microarray Lyn Y508 P07948 PK688 Up 68 11   LNCaP EGF treatment 900P Microarray MAK T157 P20794 PK689 Up 77 234   DU145 EGF treatment 900P Microarray MAPKAPK2 T222 P49137 PK690 Up 180 12   LNCaP EGF treatment 900P Microarray MAPKAPK2a T334 P36507 PK049-PK112-2 Up 73 25   LNCaP EGF treatment 900P Microarray MAPKAPK3 Y76 Q16644 PK692 Up 52 5 0.010 A549 EGF treatment 900P Microarray MAPKAPK5 T186 Q8IW41 PK693 Up 51 10   LNCaP EGF treatment 900P Microarray MARK3 T507 P27448 PK697 Up 44 7 0.032 A549 EGF treatment 900P Microarray  247 Target Name P-site Uniprot ID Antibody Code Direction of Change % CFC Standard Deviation / Sum % Error p Value Cell Line Treatment Detection Tool MEK1 S222 Q02750 PK698 Up 59 8   LNCaP EGF treatment 900P Microarray MEK1 (MAP2K1) S298 Q02750 PK047-2 Up 50 36 0.530 A549 EGF treatment 900P Microarray MEK1 (MAP2K1) T292 Q02750 PK046-1 Up 291 42   LNCaP EGF treatment 900P Microarray MEK1 (MAP2K1) T386 Q02750 PK048-1 Up 119 42   LNCaP EGF treatment 900P Microarray MEK1/2 (MAP2K1/2) S218+S222 Q02750 PK045-PN007 Up 63 64 0.561 HeLa EGF treatment 900P Microarray MEK1/2 (MAP2K1/2) S218+S222 Q02750 PK045-PN007 Up 46 99   LNCaP EGF treatment 900P Microarray MEK2 (MAP2K2) T394 P36507 PK049-2 Up 41 25 0.409 HeLa EGF treatment 900P Microarray MEK2 (MAP2K2) T394 P36507 PK050 Up 47 16   LNCaP EGF treatment 900P Microarray MEK3/6 (MAP2K3/6) S218/S207 P46734 PK051-4 Down -61 -39 0.161 A549 EGF treatment 900P Microarray MEK3/6 (MAP2K3/6) S218/S207 P46734 PK051-4 Up 138 34   LNCaP EGF treatment 900P Microarray MEK5 S311 Q13163 PK699 Up 57 60   LNCaP EGF treatment 900P Microarray MEKK2 S239 Q9Y2U5 PK700 Up 68 25   LNCaP EGF treatment 900P Microarray MERTK Y749+ Y753 Q12866 PK703 Up 49 7   LNCaP EGF treatment 900P Microarray Met Y1230 P08581 PK709 Up 71 22 0.035 A431 EGF treatment 900P Microarray Met Y1230+ Y1234+ Y1235 P08581 PK055-1 Up 186 275 0.933 A431 EGF treatment 900P Microarray Met T1241 P08581 PK706 Up 55 27 0.192 HeLa EGF treatment 900P Microarray Met T1241 P08581 PK706 Down -47 63   LNCaP EGF treatment 900P Microarray  248 Target Name P-site Uniprot ID Antibody Code Direction of Change % CFC Standard Deviation / Sum % Error p Value Cell Line Treatment Detection Tool Met T1355+ Y1356 P08581 PK707 Up 50 11   LNCaP EGF treatment 900P Microarray MKK4 S257 P45985 PK715 Down -45 -15 0.054 HeLa EGF treatment 900P Microarray MLC(MLRC2) S19 P19105 PN051-1 Up 44 67 0.571 A431 EGF treatment 900P Microarray MLK3 S281 Q16584 PK718 Up 93 35   LNCaP EGF treatment 900P Microarray MLTK T161+T162 Q9NYL2 PK719 Up 191 25   LNCaP EGF treatment 900P Microarray Mos Y263 P00540 PK722 Up 54 17 0.226 A549 EGF treatment 900P Microarray MSK2 T687 O75676 PK725 Up 105 252 0.557 A431 EGF treatment 900P Microarray MST3 T184 Q9Y6E0 PK727 Up 72 83   DU145 EGF treatment 900P Microarray MST3 T190 Q9Y6E0 PK728 Up 102 12   LNCaP EGF treatment 900P Microarray MST3 T184 Q9Y6E0 PK727 Up 122 18   LNCaP EGF treatment 900P Microarray Myc T58 P01106 PN199 Down -42 -28 0.316 A549 EGF treatment 900P Microarray Nek2 T170+S171 P51955 PK733 Down -62 -45 0.048 A431 EGF treatment 900P Microarray Nek2 S171 P51955 PK732 Up 109 29   DU145 EGF treatment 900P Microarray Nek2 T170+S171 P51955 PK733 Down -44 -20 0.048 HeLa EGF treatment 900P Microarray Nek2 S171 P51955 PK732 Up 42 26   LNCaP EGF treatment 900P Microarray Nek6 S206 Q9HC98 PK734 Down -41 -17 0.017 HeLa EGF treatment 900P Microarray NFkappaB p65 S276 Q04206 PN053-1 Up 46 20   LNCaP EGF treatment 900P Microarray  249 Target Name P-site Uniprot ID Antibody Code Direction of Change % CFC Standard Deviation / Sum % Error p Value Cell Line Treatment Detection Tool OSR1 T185 O95747 PK738 Down -41 -18 0.005 HeLa EGF treatment 900P Microarray p38a MAPK T180+Y182 Q16539 PK060-3 Up 173 13   LNCaP EGF treatment 900P Microarray p38d Y182 O15264 PK743 Up 76 30 0.358 A431 EGF treatment 900P Microarray p38d T180+Y182 O15264 PK742 Up 216 44   LNCaP EGF treatment 900P Microarray p53 S33 P04637 PN158 Up 47 43   DU145 EGF treatment 900P Microarray p53 S37 P04637 PN159 Up 64 46   LNCaP EGF treatment 900P Microarray p53 S6 P04637 PN160 Up 487 124   LNCaP EGF treatment 900P Microarray p53 S33 P04637 PN158 Up 136 32089   LNCaP EGF treatment 900P Microarray p70S6K T252 P23443 PK744 Up 44 9   LNCaP EGF treatment 900P Microarray p70S6KB S423 Q9UBS0 PK747 Up 101 222   LNCaP EGF treatment 900P Microarray PAK1 S144 Q13153 PK748 Up 69 151   LNCaP EGF treatment 900P Microarray PAK4 S474 O96013 PK752 Up 276 131   LNCaP EGF treatment 900P Microarray Paxillin 1 Y31 P49023 PN059 Up 55 38 0.097 A431 EGF treatment 900P Microarray Paxillin 1 Y118 P49023 PN060-1 Up 49 9 0.103 A549 EGF treatment 900P Microarray Paxillin 1 Y118 P49023 PN060-1 Up 55 34   LNCaP EGF treatment 900P Microarray PDGFRa Y754 P16234 PK063 Up 64 54   LNCaP EGF treatment 900P Microarray PDGFRa Y762 P16234 PK758 Up 57 74   LNCaP EGF treatment 900P Microarray  250 Target Name P-site Uniprot ID Antibody Code Direction of Change % CFC Standard Deviation / Sum % Error p Value Cell Line Treatment Detection Tool PDGFRb Y716 P09619 PK065 Up 71 45 0.137 HeLa EGF treatment 900P Microarray PDK1 S241 O15530 PK760 Up 47 44 0.405 A431 EGF treatment 900P Microarray PDLIM5 Y251 Q96HC4 PN522 Up 58 3   LNCaP EGF treatment 900P Microarray PECAM-1 Y713 P16284 PN523 Up 47 33 0.219 A431 EGF treatment 900P Microarray PGK1 Y196 P00558 PN525 Up 45 16 0.080 A431 EGF treatment 900P Microarray PIK3R1 Y467 P27986 PN526 Up 90 68   LNCaP EGF treatment 900P Microarray PIK3R2 Y464 O00459 PN528 Up 51 30   LNCaP EGF treatment 900P Microarray Pim2 T195 Q9P1W9 PK761 Up 527 66   LNCaP EGF treatment 900P Microarray PKA Ca/b T198 P17612 PK067 Down -50 -34 0.254 A549 EGF treatment 900P Microarray PKA Cb S339 P22694 PK068 Down -73 -77 0.276 A549 EGF treatment 900P Microarray PKA Cb S339 P22694 PK068 Up 113 27   LNCaP EGF treatment 900P Microarray PKCb2 T642 P05771 PK076-2 Up 46 30 0.148 A431 EGF treatment 900P Microarray PKCb2 T642 P05771 PK076-2 Down -41 -34 0.390 A549 EGF treatment 900P Microarray PKCb2 T642 P05771 PK076-2 Up 131 75   LNCaP EGF treatment 900P Microarray PKCd Y313 Q05655 PK768 Up 49 38 0.365 A431 EGF treatment 900P Microarray PKCd Y334 Q05655 PK769 Down -52 -23 0.022 HeLa EGF treatment 900P Microarray PKCd S645 Q05655 PK079-1 Up 148 60   LNCaP EGF treatment 900P Microarray  251 Target Name P-site Uniprot ID Antibody Code Direction of Change % CFC Standard Deviation / Sum % Error p Value Cell Line Treatment Detection Tool PKCd Y313 Q05655 PK077-2 Up 54 95   LNCaP EGF treatment 900P Microarray PKCe S729 Q02156 PK081-1 Down -44 29   DU145 EGF treatment 900P Microarray PKCg T514 P05129 PK082-1 Up 51 20 0.116 A431 EGF treatment 900P Microarray PKCg T655 P05129 PK083 Up 51 34 0.721 HeLa EGF treatment 900P Microarray PKCg T514 P05129 PK082-2 Up 44 35   LNCaP EGF treatment 900P Microarray PKCg T674 P05129 PK084 Up 135 28   LNCaP EGF treatment 900P Microarray PKCg T514 P05129 PK082-1 Up 209 98   LNCaP EGF treatment 900P Microarray PKCh T656 P24723 PK085 Up 326 9   LNCaP EGF treatment 900P Microarray PKCl/i T564 P41743 PK087 Down -43 -41 0.196 A549 EGF treatment 900P Microarray PKCl/i T564 P41743 PK087 Up 144 53   LNCaP EGF treatment 900P Microarray PKCm (PKD) S738+S742 Q15139 PK092 Down -54 -27 0.112 A549 EGF treatment 900P Microarray PKCm (PKD) S910 Q15139 PK093-1 Up 57 8   LNCaP EGF treatment 900P Microarray PKCq S695 Q04759 PK772 Up 75 48 0.138 A431 EGF treatment 900P Microarray PKCq Y545 Q04759 PK773 Up 44 19 0.096 A431 EGF treatment 900P Microarray PKCq S695 Q04759 PK772 Up 184 89   LNCaP EGF treatment 900P Microarray PKCq Y545 Q04759 PK773 Up 43 16   LNCaP EGF treatment 900P Microarray PKCz S262+Y263 Q05513 PK774 Up 101 23   LNCaP EGF treatment 900P Microarray  252 Target Name P-site Uniprot ID Antibody Code Direction of Change % CFC Standard Deviation / Sum % Error p Value Cell Line Treatment Detection Tool PKR1 T446 P19525 PK777 Down -49 -19 0.017 HeLa EGF treatment 900P Microarray PKR1 T446 P19525 PK132 Up 65 23   LNCaP EGF treatment 900P Microarray PLCg1 Y771 P19174 PN165 Up 55 42   LNCaP EGF treatment 900P Microarray PLCG1 Y783 P19174 PN530 Up 42 55   LNCaP EGF treatment 900P Microarray PPP1R11 Y64 O60927 PN532 Down -74 -80 0.283 A549 EGF treatment 900P Microarray PPP1R11 Y64 O60927 PN532 Down -40 -14 0.022 HeLa EGF treatment 900P Microarray PRAS40 T246 Q96B36 PN062 Down -83 -101 0.272 A549 EGF treatment 900P Microarray PRK1 T774 Q16512 PK781 Up 46 28   LNCaP EGF treatment 900P Microarray PRKX T201+T203 P51817 PK785 Up 65 78   LNCaP EGF treatment 900P Microarray PRP4K Y849 Q13523 PK786 Down -45 -16 0.039 HeLa EGF treatment 900P Microarray PTEN S380+T382+T383 P60484 PP006 Down -53 -26 0.126 A549 EGF treatment 900P Microarray PTEN S380+T382+T383 P60484 PP006-1 Down -67 -66 0.270 A549 EGF treatment 900P Microarray PTEN S380+T382+T383 P60484 PP006 Up 43 29 0.043 HeLa EGF treatment 900P Microarray PTEN S380+T382+T383 P60484 PP006-1 Up 47 31 0.793 HeLa EGF treatment 900P Microarray PTRF Y308 Q6NZI2 PN533 Down -45 -20 0.148 HeLa EGF treatment 900P Microarray Pyk2 Y579 Q14289 PK097-3 Up 162 45   DU145 EGF treatment 900P Microarray PYK2 Y402 Q14289 PK787 Up 218 15   LNCaP EGF treatment 900P Microarray  253 Target Name P-site Uniprot ID Antibody Code Direction of Change % CFC Standard Deviation / Sum % Error p Value Cell Line Treatment Detection Tool Raf1 S296 P04049 PK791 Down -87 -104 0.246 A549 EGF treatment 900P Microarray Raf1 S296 P04049 PK791 Down -46 -15 0.013 HeLa EGF treatment 900P Microarray Raf1 S301+T303 P04049 PK792 Down -43 -13 0.010 HeLa EGF treatment 900P Microarray Rb S612 P06400 PN066 Down -46 -13 0.046 A549 EGF treatment 900P Microarray Rb S780 P06400 PN067 Down -43 -18 0.144 A549 EGF treatment 900P Microarray Rb S795 P06400 PN131-1 Down -72 -84 0.326 A549 EGF treatment 900P Microarray Rb S807 P06400 PN068 Down -84 -101 0.276 A549 EGF treatment 900P Microarray Rb T821 P06400 PN070 Down -91 -120 0.278 A549 EGF treatment 900P Microarray Rb S807 P06400 PN068 Down -42 -14 0.041 HeLa EGF treatment 900P Microarray Rb S780 P06400 PN067 Up 88 7   LNCaP EGF treatment 900P Microarray RelB S573 Q01201 PN151 Up 42 35 0.795 HeLa EGF treatment 900P Microarray RIOK1 Y466 Q9BRS2 PK794 Down -57 -40 0.111 HeLa EGF treatment 900P Microarray RIPK1 Y384 Q13546 PK795 Down -93 -123 0.271 A549 EGF treatment 900P Microarray RIPK1 Y384 Q13546 PK795 Down -42 -19 0.100 HeLa EGF treatment 900P Microarray RIPK1 Y384 Q13546 PK795 Down -43 42   LNCaP EGF treatment 900P Microarray RIPK2 S176 O43353 PK796 Down -60 -56 0.313 A549 EGF treatment 900P Microarray Ron Y1238 +Y1239 Q04912 PK801 Down -77 -84 0.267 A549 EGF treatment 900P Microarray  254 Target Name P-site Uniprot ID Antibody Code Direction of Change % CFC Standard Deviation / Sum % Error p Value Cell Line Treatment Detection Tool ROR2 Y645+Y646 Q01974 PK802 Up 59 14   LNCaP EGF treatment 900P Microarray Ros Y2114+ Y2115 P08922 PK803 Up 69 43 0.195 A431 EGF treatment 900P Microarray RSK1 S380 Q15418 PK805 Up 55 38 0.151 A431 EGF treatment 900P Microarray RSK1 T573 Q15418 PK806 Up 64 77 0.457 A431 EGF treatment 900P Microarray RSK1 S380 Q15418 PK805 Down -71 -68 0.250 A549 EGF treatment 900P Microarray RSK1 T573 Q15418 PK806 Down -76 -77 0.243 A549 EGF treatment 900P Microarray RSK1/2 S363/S369 Q15418 PK100-2 Up 59 125 0.592 A431 EGF treatment 900P Microarray RSK1/2 S380/S386 Q15418 PK101-2 Up 48 61 0.412 A431 EGF treatment 900P Microarray RSK1/2 S221/S227 Q15418 PK099 Down -65 -5 0.382 A549 EGF treatment 900P Microarray RSK1/2 S363/S369 Q15418 PK100-2 Down -62 -27 0.068 A549 EGF treatment 900P Microarray RSK1/2 S380/S386 Q15418 PK101-2 Down -42 -14 0.034 HeLa EGF treatment 900P Microarray RSK1/2/3 T573/T577/T570 Q15418 PK102 Up 513 793 0.475 HeLa EGF treatment 900P Microarray S6 S235 P62753 PN073 Up 53 34 0.693 HeLa EGF treatment 900P Microarray S6Kb1 T252 P23443 PK145 Up 74 83   LNCaP EGF treatment 900P Microarray Sgk223 Y413 Q86YV5 PK810 Down -40 -20 0.031 HeLa EGF treatment 900P Microarray SgK269 Y635 Q9H792 PK811 Down -40 -16 0.062 HeLa EGF treatment 900P Microarray Shc1 Y349 P29353 PN161 Down -46 0 0.772 A549 EGF treatment 900P Microarray  255 Target Name P-site Uniprot ID Antibody Code Direction of Change % CFC Standard Deviation / Sum % Error p Value Cell Line Treatment Detection Tool Shc1 Y349+Y350 P29353 PN074 Down -40 -35 0.375 A549 EGF treatment 900P Microarray SHIP-2 Y886 O15357 PN534 Up 74 33   LNCaP EGF treatment 900P Microarray SIK T182 P57059 PK812 Down -44 -21 0.304 A549 EGF treatment 900P Microarray SIK T182 P57059 PK812 Down -48 -24 0.092 HeLa EGF treatment 900P Microarray SIK3 T163 Q9Y2K2 PK814 Up 45 10 0.096 A549 EGF treatment 900P Microarray SIK3 T411 Q9Y2K2 PK815 Up 67 52 0.205 A549 EGF treatment 900P Microarray SIK3 T411 Q9Y2K2 PK815 Down -64 -72 0.357 HeLa EGF treatment 900P Microarray SIT Y95 Q9Y3P8 PN536 Up 41 22 0.216 A549 EGF treatment 900P Microarray SIT Y90 Q9Y3P8 PN535 Up 128 228   DU145 EGF treatment 900P Microarray SLK S189 Q9H2G2 PK816 Up 42 22 0.209 A431 EGF treatment 900P Microarray SLK S189 Q9H2G2 PK816 Up 43 7 0.945 A549 EGF treatment 900P Microarray SMC1 S957 Q14683 PN125 Down -61 -52 0.190 A431 EGF treatment 900P Microarray SNCA S129 P37840 PN197 Up 67 42   LNCaP EGF treatment 900P Microarray snRNP 70 Y126 P08621 PN537 Up 81 17 0.056 A549 EGF treatment 900P Microarray Src Y419 P12931 PK107 Up 43 22 0.089 A431 EGF treatment 900P Microarray Src Y419 P12931 PK818 Down -43 -23 0.007 HeLa EGF treatment 900P Microarray Src Y419 P12931 PK818 Up 52 35   LNCaP EGF treatment 900P Microarray  256 Target Name P-site Uniprot ID Antibody Code Direction of Change % CFC Standard Deviation / Sum % Error p Value Cell Line Treatment Detection Tool STAT1a S727 P42224 PN078-PN135 Up 50 40 0.269 A431 EGF treatment 900P Microarray STAT1a Y701 P42224 PN079-PN136 Up 45 29 0.576 HeLa EGF treatment 900P Microarray STAT2 Y690 P52630 PN080 Up 63 45   LNCaP EGF treatment 900P Microarray STAT3 Y705 P40763 PN082-1 Up 71 117   LNCaP EGF treatment 900P Microarray Syk Y352 P43405 PK822 Down -41 -22 0.007 HeLa EGF treatment 900P Microarray TAK1 S439 O43318 PK824 Down -57 -37 0.085 A431 EGF treatment 900P Microarray TAK1 S439 O43318 PK824 Up 50 21 0.105 A549 EGF treatment 900P Microarray TAK1 T184+T187 O43318 PK825 Up 45 58   DU145 EGF treatment 900P Microarray Tau S516 P10636  PN085 Down -53 -19 0.048 A549 EGF treatment 900P Microarray Tau S721 P10636  PN092 Up 56 59   LNCaP EGF treatment 900P Microarray Tau S739 P10636  PN107 Up 60 20   LNCaP EGF treatment 900P Microarray Tau T522 P10636  PN121 Up 280 133   LNCaP EGF treatment 900P Microarray TBC1D7 Y14 Q9P0N9 PN540 Up 61 42 0.491 A549 EGF treatment 900P Microarray TBC1D7 Y14 Q9P0N9 PN540 Up 46 77 0.788 HeLa EGF treatment 900P Microarray TBK1 S172 Q9UHD2 PK828 Down -46 -22 0.104 HeLa EGF treatment 900P Microarray TBK1 S172 Q9UHD2 PK828 Down -46 10   LNCaP EGF treatment 900P Microarray TEC Y519 P42680 PK829 Up 71 13   LNCaP EGF treatment 900P Microarray  257 Target Name P-site Uniprot ID Antibody Code Direction of Change % CFC Standard Deviation / Sum % Error p Value Cell Line Treatment Detection Tool TNK1 Y277 Q13470 PK832 Up 174 70 0.022 A431 EGF treatment 900P Microarray TNK1 Y277 Q13470 PK832 Down -52 22   LNCaP EGF treatment 900P Microarray TrkB Y706 Q16620 PK160 Up 235 158 0.003 A431 EGF treatment 900P Microarray TYK2 Y1054+ Y1055 P29597 PK845 Up 42 23   LNCaP EGF treatment 900P Microarray Tyro3 Y685+Y686 Q06418 PK848 Down -41 30   LNCaP EGF treatment 900P Microarray Vimentin S34 P08670 PN094 Up 92 45 0.163 A549 EGF treatment 900P Microarray Wee1 S642 P30291 PK854 Down -45 27   LNCaP EGF treatment 900P Microarray WNK1 T60 Q9H4A3 PK856 Down -62 -50 0.147 A431 EGF treatment 900P Microarray WNK1 T60 Q9H4A3 PK856 Up 71 25   LNCaP EGF treatment 900P Microarray Yes Y222+Y223 P07947 PK858 Up 95 80   LNCaP EGF treatment 900P Microarray ZAP70 Y492+Y493 P43403 PK863 Down -43 -34 0.362 A431 EGF treatment 900P Microarray ZAP70 Y292 P43403 PK861 Down -73 -74 0.253 A549 EGF treatment 900P Microarray     258 Appendix E. Target phosphorylation sites phosphorylated or dephosphorylated in response to EGF treatment from Western blot analysis.  Target Name P-site Uniprot ID Antibody Code Direction of Change Cell Line Treatment Detection Tool Abl (Abl1) Y393+T394 P00519 PK873 Up A431 EGF treatment Western Blotting Abl (Abl1) Y264 P00519 PK507 Up A431 EGF treatment Western Blotting Abl (Abl1) Y264 P00519 PK507 Up A549 EGF treatment Western Blotting Abl (Abl1) Y393+T394 P00519 PK873 Up HeLa EGF treatment Western Blotting Abl2 (Arg) Y439+T440 P42684 PK510 Up A431 EGF treatment Western Blotting Abl2 (Arg) Y439+T440 P42684 PK510 Up A549 EGF treatment Western Blotting ACK1 (TNK2) Y284 Q07912 PK511 Up A431 EGF treatment Western Blotting ACK1 (TNK2) Y859+Y860 Q07912 PK513 Up A431 EGF treatment Western Blotting ACK1 (TNK2) Y284 Q07912 PK511 Up A549 EGF treatment Western Blotting ACK1 (TNK2) Y284 Q07912 PK511 Up HeLa EGF treatment Western Blotting ACLY Y682 P53396 PN686 Up A549 EGF treatment Western Blotting ACS1 (ACSL1) Y567 P33121 PN688 Up A431 EGF treatment Western Blotting Akt1 (PKBa) Y326 P31749 PK517 Up A549 EGF treatment Western Blotting ALK Y1096 Q9UM73 PK519 Up HeLa EGF treatment Western Blotting ALK Y1507 Q9UM73 PK520 Up HeLa EGF treatment Western Blotting  259 Target Name P-site Uniprot ID Antibody Code Direction of Change Cell Line Treatment Detection Tool ALK Y1092 Q9UM73 PK518 Up HeLa EGF treatment Western Blotting AMPKa1 (PRKAA1) T183+S184 Q13131 PK521 Up A431 EGF treatment Western Blotting ANKRD3 (RIPK4, DIK) S438 P57078 PK523 Up A431 EGF treatment Western Blotting ATASE (PPAT) T356 Q06203 PN742 Down A431 EGF treatment Western Blotting ATASE (PPAT) T356 Q06203 PN742 Up HeLa EGF treatment Western Blotting AurKB (Aurora B, AIM-1) S227 Q96GD4 PK530 Down A431 EGF treatment Western Blotting BLK Y188 P51451 PK542 Down A431 EGF treatment Western Blotting BRK S446+Y447 Q13882 PK547 Up A431 EGF treatment Western Blotting BRK S446+Y447 Q13882 PK547 Down A549 EGF treatment Western Blotting BRK S446+Y447 Q13882 PK547 Up HeLa EGF treatment Western Blotting Btk Y551 Q06187 PK551 Up HeLa EGF treatment Western Blotting CD45 (PTPRC; Receptor-type tyrosine-protein phosphatase C) Y1216 P08575 PP527 Up A431 EGF treatment Western Blotting CD45 (PTPRC; Receptor-type tyrosine-protein phosphatase C) Y1216 P08575 PP527 Up HeLa EGF treatment Western Blotting CDC5L T385 Q99459 PN576 Down HeLa EGF treatment Western Blotting CDK1 (CDC2) T14+Y15 P06493 PK560 Up A431 EGF treatment Western Blotting CDK2 T160 P24941 PK568 Up A431 EGF treatment Western Blotting CDK6 Y24 Q00534 PK572 Up A431 EGF treatment Western Blotting  260 Target Name P-site Uniprot ID Antibody Code Direction of Change Cell Line Treatment Detection Tool Chk1 (CHEK1) S317 O14757 PK578 Up A431 EGF treatment Western Blotting CK2a1 (CSNK2A1) Y255 P68400 PK582 Up A431 EGF treatment Western Blotting COT (MAP3K8, TPL2) S334 P41279 PK585 Up A431 EGF treatment Western Blotting CREB1 S129+S133 P16220 PN577 Up A431 EGF treatment Western Blotting CREB1 S129+S133 P16220 PN577 Down HeLa EGF treatment Western Blotting CRYAB S45 P02511 PN110 Up A549 EGF treatment Western Blotting CRYAB S45 P02511 PN110 Down HeLa EGF treatment Western Blotting Csk Y184 P41240 PK589 Up A431 EGF treatment Western Blotting DDR2 (Tyro10) Y740 Q16832 PK594 Down HeLa EGF treatment Western Blotting DLK (ZPK) S269 Q12852 PK590 Up A431 EGF treatment Western Blotting Dok3 Y398 Q7L591 PN508 Up A549 EGF treatment Western Blotting Dok3 Y398 Q7L591 PN508 Up HeLa EGF treatment Western Blotting EEF2 T57 P13639 PN555 Up A549 EGF treatment Western Blotting EEF2 T57 P13639 PN555 Down HeLa EGF treatment Western Blotting EGFR (ErbB1) Y1110 P00533 PK600 Up A431 EGF treatment Western Blotting EGFR (ErbB1) Y998 P00533 PK603 Up A431 EGF treatment Western Blotting EGFR (ErbB1) Y1069 P00533 PK599 Up A431 EGF treatment Western Blotting EGFR (ErbB1) Y869 P00533 PK602 Up A431 EGF treatment Western Blotting  261 Target Name P-site Uniprot ID Antibody Code Direction of Change Cell Line Treatment Detection Tool EGFR (ErbB1) Y1172 P00533 PK601 Up A431 EGF treatment Western Blotting EGFR (ErbB1) Y1069 P00533 PK599 Up A549 EGF treatment Western Blotting EGFR (ErbB1) Y1172 P00533 PK601 Up A549 EGF treatment Western Blotting EGFR (ErbB1) Y1069 P00533 PK599 Up HeLa EGF treatment Western Blotting EGFR (ErbB1) Y869 P00533 PK602 Up HeLa EGF treatment Western Blotting EGFR (ErbB1) Y1172 P00533 PK601 Up HeLa EGF treatment Western Blotting ENO1 Y44 P06733 PN511 Down A549 EGF treatment Western Blotting ENO2 Y25 P09104 PN512 Up A431 EGF treatment Western Blotting ErbB2 (HER2, Neu) Y735 P04626 PK614 Up A431 EGF treatment Western Blotting ErbB2 (HER2, Neu) Y1248 P04626 PK613 Up A431 EGF treatment Western Blotting ErbB2 (HER2, Neu) Y877 P04626 PK615 Down HeLa EGF treatment Western Blotting ErbB2 (HER2, Neu) Y1248 P04626 PK613 Up HeLa EGF treatment Western Blotting ERBB2IP (Erbin) Y1104 Q96RT1 PN513 Up A431 EGF treatment Western Blotting ERBB2IP (Erbin) Y1104 Q96RT1 PN513 Up A549 EGF treatment Western Blotting ERBB2IP (Erbin) Y1104 Q96RT1 PN513 Up HeLa EGF treatment Western Blotting ErbB3 (HER3) Y1307 P21860 PK617 Up A431 EGF treatment Western Blotting ErbB3 (HER3) Y1328 P21860 PK618 Up A431 EGF treatment Western Blotting ErbB3 (HER3) Y1289 P21860 PK616 Up A431 EGF treatment Western Blotting  262 Target Name P-site Uniprot ID Antibody Code Direction of Change Cell Line Treatment Detection Tool ErbB4 (HER4) Y875 Q15303 PK620 Up A431 EGF treatment Western Blotting ErbB4 (HER4) Y733 Q15303 PK619 Up A431 EGF treatment Western Blotting ErbB4 (HER4) Y733 Q15303 PK619 Up A549 EGF treatment Western Blotting ErbB4 (HER4) Y733 Q15303 PK619 Up HeLa EGF treatment Western Blotting ERK3 (MAPK6) S189 Q16659 PK623 Up A431 EGF treatment Western Blotting ERK5 (MAPK7) Y221 Q13164 PK626 Down A431 EGF treatment Western Blotting FBPase (FBP1) Y265 P09467 PN699 Up A431 EGF treatment Western Blotting FBPase (FBP1) Y265 P09467 PN699 Down A549 EGF treatment Western Blotting FBPase 2 (FBP2) Y216 O00757 PN700 Down A431 EGF treatment Western Blotting FBPase 2 (FBP2) Y216 O00757 PN700 Down A549 EGF treatment Western Blotting FBPase 2 (FBP2) Y216 O00757 PN700 Up HeLa EGF treatment Western Blotting Fes Y713+S716 P07332 PK633 Up HeLa EGF treatment Western Blotting FGFR3 Y647+Y648 P22607 PK636 Up A431 EGF treatment Western Blotting Fyn Y420 P06241 PK881 Up A431 EGF treatment Western Blotting G6PD Y503+Y507 P11413 PN701 Up A431 EGF treatment Western Blotting G6PD Y503+Y507 P11413 PN701 Up HeLa EGF treatment Western Blotting GATA3 S369 P23771 PN702 Up A431 EGF treatment Western Blotting GCN2 (EIF2AK4) T667 Q9P2K8 PK877 Up A431 EGF treatment Western Blotting  263 Target Name P-site Uniprot ID Antibody Code Direction of Change Cell Line Treatment Detection Tool GRK2 (BARK1, ADRBK1) Y356 P25098 PK537 Down HeLa EGF treatment Western Blotting GSK3a Y284+Y285 P49840 PK650 Up A431 EGF treatment Western Blotting GSK3a Y284+Y285 P49840 PK650 Up HeLa EGF treatment Western Blotting GYS2 Y45 P54840 PN704 Up A431 EGF treatment Western Blotting HCFC1 S1507 P51610 PN604 Up A431 EGF treatment Western Blotting HePTP (PTPN7) T66 P35236 PP501 Up A431 EGF treatment Western Blotting HGK (ZC1) T187 O95819 PK653 Up A431 EGF treatment Western Blotting HMGCR S872 P04035 PN705 Down A431 EGF treatment Western Blotting IGF1R Y1161+T1163 P08069 PK657 Up A431 EGF treatment Western Blotting IGF1R Y1346 P08069 PK658 Up A431 EGF treatment Western Blotting IGF1R Y1346 P08069 PK658 Up HeLa EGF treatment Western Blotting IKKa (IkBKA) T23 O15111 PK154 Up A431 EGF treatment Western Blotting IRF3 T135 Q14653 PN610 Up A431 EGF treatment Western Blotting IRF3 T135 Q14653 PN610 Up A549 EGF treatment Western Blotting IRF3 T135 Q14653 PN610 Down HeLa EGF treatment Western Blotting ITSN2 Y968 Q9NZM3 PN521 Up A431 EGF treatment Western Blotting JAK1 T1107 P23458 PK895 Down HeLa EGF treatment Western Blotting JAK2 Y570 O60674 PK668 Up A431 EGF treatment Western Blotting  264 Target Name P-site Uniprot ID Antibody Code Direction of Change Cell Line Treatment Detection Tool JAK2 Y570 O60674 PK668 Down A549 EGF treatment Western Blotting JAK2 Y570 O60674 PK668 Up HeLa EGF treatment Western Blotting JNK1 (MAPK8) Y185 P45983 PK670 Down A549 EGF treatment Western Blotting JNK1 (MAPK8) Y185 P45983 PK670 Up HeLa EGF treatment Western Blotting Jun (c-Jun) S243 P05412 PN614 Up A431 EGF treatment Western Blotting Jun (c-Jun) S243 P05412 PN614 Up A549 EGF treatment Western Blotting Jun (c-Jun) S243 P05412 PN614 Up HeLa EGF treatment Western Blotting Kit Y721 P10721 PK885 Up A431 EGF treatment Western Blotting LATS1 S909 O95835 PK678 Up A431 EGF treatment Western Blotting LKB1 (STK11) S31 Q15831 PK682 Up A431 EGF treatment Western Blotting LKB1 (STK11) S428 Q15831 PK683 Up HeLa EGF treatment Western Blotting MAFG S124 O15525 PN617 Up HeLa EGF treatment Western Blotting MAPKAPK3 Y76 Q16644 PK692 Up A431 EGF treatment Western Blotting MAPKAPK3 Y76 Q16644 PK692 Up HeLa EGF treatment Western Blotting MAPKAPK5 (PRAK) T186 Q8IW41 PK693 Up A431 EGF treatment Western Blotting MAT1A T341 Q00266 PN759 Down A431 EGF treatment Western Blotting MBP T232 P02686 PN558 Up A431 EGF treatment Western Blotting MCM2 Y137+S139 P49736 PN620 Up HeLa EGF treatment Western Blotting  265 Target Name P-site Uniprot ID Antibody Code Direction of Change Cell Line Treatment Detection Tool MEF2A T108 Q02078 PN622 Up A431 EGF treatment Western Blotting MEF2A T108 Q02078 PN622 Up HeLa EGF treatment Western Blotting MEF2D S180 Q14814 PN625 Down HeLa EGF treatment Western Blotting Met T1241 P08581 PK706 Up A431 EGF treatment Western Blotting Met S1236 P08581 PK705 Up A431 EGF treatment Western Blotting Met Y1234 P08581 PK710 Up A431 EGF treatment Western Blotting Met Y1234+Y1235 P08581 PK711 Up A431 EGF treatment Western Blotting Met Y1234+Y1235 P08581 PK711 Up A549 EGF treatment Western Blotting Met Y1234+Y1235 P08581 PK711 Up HeLa EGF treatment Western Blotting Nek7 T191+S195 Q8TDX7 PK735 Down A431 EGF treatment Western Blotting NF2 S518 P35240 PN709 Down HeLa EGF treatment Western Blotting NLK T298 Q9UBE8 PK736 Down A431 EGF treatment Western Blotting NMDAR2A (GRIN2A) Y943 Q12879 PN710 Up A431 EGF treatment Western Blotting NMDAR2A (GRIN2A) Y943 Q12879 PN710 Up A549 EGF treatment Western Blotting NMDAR2A (GRIN2A) Y943 Q12879 PN710 Up HeLa EGF treatment Western Blotting NOS1 S746 P29475 PN762 Up A431 EGF treatment Western Blotting p38d MAPK (MAPK13) T180+Y182 O15264 PK742 Down A549 EGF treatment Western Blotting p38d MAPK (MAPK13) T180+Y182 O15264 PK742 Up HeLa EGF treatment Western Blotting  266 Target Name P-site Uniprot ID Antibody Code Direction of Change Cell Line Treatment Detection Tool p53 (TP53) S392 P04637 PN640 Down HeLa EGF treatment Western Blotting p70S6K (S6Ka, RPS6KB1) S447 P23443 PK156 Up A431 EGF treatment Western Blotting p70S6K (S6Ka, RPS6KB1) T444+S447 P23443 PK746 Up A549 EGF treatment Western Blotting p70S6K (S6Ka, RPS6KB1) T444+S447 P23443 PK746 Up HeLa EGF treatment Western Blotting p70S6KB (RPS6KB2) S423 Q9UBS0 PK747 Up HeLa EGF treatment Western Blotting PAH S16 P00439 PN713 Up A431 EGF treatment Western Blotting PAK1 (PAKa) S144 Q13153 PK748 Up A431 EGF treatment Western Blotting PAK4 S474 O96013 PK752 Down HeLa EGF treatment Western Blotting PBRM1 S948 Q86U86 PN714 Up A431 EGF treatment Western Blotting PCTK1 (PCTAIRE1, CDK16) Y176 Q00536 PK755 Up A431 EGF treatment Western Blotting PGK1 Y196 P00558 PN525 Up A431 EGF treatment Western Blotting PKCb (PRKCB1) T500 P05771 PK766 Up A549 EGF treatment Western Blotting PKCd (PRKCD) Y313 Q05655 PK768 Up A431 EGF treatment Western Blotting PKCd (PRKCD) Y313 Q05655 PK768 Up HeLa EGF treatment Western Blotting PKCt (PRKCQ) Y545 Q04759 PK773 Up A431 EGF treatment Western Blotting PKCt (PRKCQ) S695 Q04759 PK772 Up A431 EGF treatment Western Blotting PKCt (PRKCQ) S695 Q04759 PK772 Up A549 EGF treatment Western Blotting PKCt (PRKCQ) S695 Q04759 PK772 Up HeLa EGF treatment Western Blotting  267 Target Name P-site Uniprot ID Antibody Code Direction of Change Cell Line Treatment Detection Tool PKD1 (PRKCM, PKCm, PRKD1) S738+S742 Q15139 PK771 Down A549 EGF treatment Western Blotting PKR1 (PRKR; EIF2AK2) T446 P19525 PK132 Up A431 EGF treatment Western Blotting PLCD1 S460 P51178 PN720 Up A431 EGF treatment Western Blotting PLCG1 Y977 P19174 PN722 Up HeLa EGF treatment Western Blotting PLCG2 Y759 P16885 PN531 Up A431 EGF treatment Western Blotting PLCG2 Y753 P16885 PN723 Up A431 EGF treatment Western Blotting PP2A/Ca (PPP2CA) Y307 P67775 PP504 Up A431 EGF treatment Western Blotting PP2A/Ca (PPP2CA) Y307 P67775 PP504 Up HeLa EGF treatment Western Blotting PP2Ca (PPM1A; PPPM1A)/'PP2Cb (PPM1B; PPPM1B) Y362 P35813 PP508 Up A431 EGF treatment Western Blotting PPP1R12B T646 O60237 PP503 Down A431 EGF treatment Western Blotting PPP1R12B T646 O60237 PP503 Down HeLa EGF treatment Western Blotting PPP2CB T304 P62714 PP505 Up HeLa EGF treatment Western Blotting PPP3CC (Calcinerin) S463 P48454 PP506 Up A549 EGF treatment Western Blotting PPP5C Y119 P53041 PP507 Up A431 EGF treatment Western Blotting PPP5C Y119 P53041 PP507 Up A549 EGF treatment Western Blotting PPP5C Y119 P53041 PP507 Up HeLa EGF treatment Western Blotting PRMT5 T634 O14744 PN549 Up A431 EGF treatment Western Blotting  268 Target Name P-site Uniprot ID Antibody Code Direction of Change Cell Line Treatment Detection Tool PTPN11 (PTP1D; PTP2C; SHP2; SHPTP2; Syp) Y62 Q06124 PP512 Up A431 EGF treatment Western Blotting PTPN11 (PTP1D; PTP2C; SHP2; SHPTP2; Syp) Y62 Q06124 PP512 Up HeLa EGF treatment Western Blotting PTPN14 S486 Q15678 PP514 Up HeLa EGF treatment Western Blotting PTPN22 Y499 Q9Y2R2 PP517 Up HeLa EGF treatment