UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Multivalent pIII phage display libraries : selected issues and applications Wilson, Daniel Ross 1996

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

Item Metadata

Download

Media
831-ubc_1997-25187X.pdf [ 24.94MB ]
Metadata
JSON: 831-1.0088345.json
JSON-LD: 831-1.0088345-ld.json
RDF/XML (Pretty): 831-1.0088345-rdf.xml
RDF/JSON: 831-1.0088345-rdf.json
Turtle: 831-1.0088345-turtle.txt
N-Triples: 831-1.0088345-rdf-ntriples.txt
Original Record: 831-1.0088345-source.json
Full Text
831-1.0088345-fulltext.txt
Citation
831-1.0088345.ris

Full Text

M U L T I V A L E N T PIII P H A G E DISPLAY LIBRARIES: SELECTED ISSUES AND APPLICATIONS by DANIEL ROSS WILSON B.Sc , University of Guelph, 1990 A THESIS SUBMITTED IN PARTIAL FULFILMENT OF T H E REQUIREMENTS FOR T H E D E G R E E OF DOCTOR OF PHILOSOPHY in T H E F A C U L T Y OF G R A D U A T E STUDIES Department of Microbiology and Immunology and the Biotechnology Laboratory We accept this thesis as conforming to the required standard T H E UNIVERSITY OF BRITISH COLUMBIA August 1997 © Daniel Ross Wilson, 1997 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Department of The University of British Columbia Vancouver, Canada DE-6 (2/88) 11 Abstract By the use of model Ff phage libraries displaying (i) variants of the Plasmodium falciparum circumsporozoite protein (CSP) immunodominant tetrapeptide repeats and (ii) peptides derived from the gene encoding Bordetella pertussis filamentous hemagglutinin (FHA), the application potential of phage display was illustrated and several issues related to the utility of phage display were examined. Studies with the P. falciparum model, together with a statistical analysis of conformations adopted by aspartate/asparagine-proline residue pairs in proteins deposited in the Brookhaven Protein Data bank, provided insight into the structure of the CSP repeats and into the epitopes recognized by a panel of anti-CSP monoclonal antibodies. These studies also served to demonstrate the technical simplicity and general utility of phage display when applied to simple models. The more complex model, involving phage display of B. pertussis FHA-derived peptides, served as a model for genomic libraries. Work with this model led to the identification of fourteen antigenic regions within the large (200 kDa) FHA protein; these findings were both confirmed by and greatly extend previous antigenic analyses of FHA. This model also demonstrated that while construction of genomic phage display libraries is technically achievable, limitations imposed by phage and host biology may limit their utility and broad applicability. Difficulties encountered while exploring this model indicate that vector design is important to library construction and employment, and that attempts to display some foreign peptides on the phage surface can so profoundly affect phage/host biology that in some cases incompatible peptides will be "corrected" so as to enable their display. These and other issues of phage/host biology and their impact on phage display are discussed, including the need for fundamental studies into the influence of the nature (composition and sequence) of a peptide on its ability to be incorporated into virions. Ill Table of Contents Abstract ii Table of Contents i i i List of Tables x List of Figures xi Acknowledgements xiv Chapter 1 - Introduction 1 1.1. O V E R V I E W 1 1.2. F F P H A G E B I O L O G Y A N D E A R L Y P H A G E D I S P L A Y 5 1.2.1. Overview 5 1.2.2. fd-tet and other ori(-) mutants 8 1.2.3. p i l l 9 1.2.4. Development of pill-based phage display 10 1.2.5. The first random peptide libraries 11 1.3. R O L E OF THE H O S T C E L L 12 1.3.1. Signal peptides 12 1.3.2. Other aspects of membrane insertion/translocation 15 1.3.3. Genomic libraries, membrane anchors and secondary signal sequences . . . . 16 1.3.4. Codon usage and context 17 1.3.5. Molecular chaperones and export targeting 18 1.3.6. Proteases 18 1.4. IMPACT OF P H A G E A N D HOST B I O L O G Y O N P H A G E D I S P L A Y 19 1.5. APPLICATIONS A N D INNOVATIONS IN P H A G E DISPLAY 20 1.5.1. Random peptide libraries 20 1.5.2. Phage-displayed proteins and alternative methods of display 25 1.5.3. "Directed evolution of a protein" 29 1.5.4. Gene fragment libraries 29 1.5.5. Innovations in peptide presentation 32 1.5.6. Innovations in identification of target clones 34 1.6. T H E PRESENT S T U D Y 35 1.6.1. Project goals 35 1.6.2. Project accomplishments in context of the concurrent work of others 36 1.6.3. Thesis layout 37 Chapter 2 - Materials and Methods 38 2.1 . B A C T E R I A L STRAINS, PLASMIDS A N D B A C T E R I O P H A G E 38 2 .1 .1 . Culture of bacteria 38 IV 2.2. G E N E R A L M E T H O D S 41 2.2.1. Common reagents 41 2.2.2. D N A manipulation 41 2.2.3. Agarose gel electrophoresis 43 2.2.4. D N A quantification 43 2.2.5. Oligonucleotides 43 2.3. D N A E X T R A C T I O N F R O M VIRION HOSTS A N D VIRIONS 44 2.3.1. RF D N A extraction from host cells 44 2.3.2. S.s. D N A extraction from virions 44 2.4. SEQUENCING 45 2.4.1. Sequencing primers 45 2.4.2. Sequencing reactions and gels 45 2.5. H O S T C E L L TRANSFORMATION A N D INFECTION 46 2.5.1. Calcium chloride transformation 46 2.5.2. Electroporation 47 2.5.3. Infection of F-piliated host cells with virions 48 2.6. H A R V E S T I N G OF VIRIONS 49 2.6.1. Harvesting virions from bacterial growth on solid media 49 2.6.2. Separation of virions in culture or wash supernatant from cells by centrifugation 49 2.6.3. One-stage PEG precipitation 49 2.6.4. Two-stage PEG precipitation . . . 50 2.6.5. CsCl density gradient centrifugation 50 2.7. VIRION QUANTIFICATION A N D A N A L Y S I S 51 2.7.1. Quantification of virion particles by ultraviolet ( U V ) spectroscopy 51 2.7.2. Quantification of plaque-forming units (pfu) and transducing units (TU) . . . 51 2.7.3. Agarose gel electrophoresis of virions 53 2.8. V E C T O R CONSTRUCTION 54 2.8.1. Conventional vector fDRW20 54 2.8.2. Conventional vectors fDRW21 and fDRW22 55 2.8.3. Amber vector fDRW5 55 2.8.4. Amber vector fDRW613 56 2.8.5. Amber vector fDRW613C 59 2.8.6. Amber vector fDRW70 59 2.8.7. Conventional vector series fDRW8wi 60 2.9. V E C T O R STABILITY ISSUES 61 2.9.1. Initial studies of frameshift vectors 61 2.9.2. Evaluation of E. coli host strains for fDRW5 propagation (Table 7-1) 64 2.9.3. Comparative analysis of frameshift and amber vector stability (Figure 7-15) 64 V 2.9.4. Effects of SupF and wild-type p i l l on pseudorevertant production 66 2.9.5. Followup analysis of pseudorevertants and comparison of fUSE5 with fDRW5 (Figures 7-24 to 7-26) 67 2.10. CONSTRUCTION OF A L I B R A R Y OF V A R I A N T PEPTIDES D E R I V E D F R O M THE P. FALCIPARUM CIRCUMSPOROZOITE PROTEIN ( C S P - L I B R A R Y ) 69 2.10.1. Construction of a clone displaying P N A N P N A N P N A 69 2.10.2. Construction of a first CSP-library 69 2.10.3. Unsuccessful construction of a second, proline-biased CSP-library 71 2.10.4. Successful construction of a third, proline-biased CSP-library (Table 7-II) . . 72 2.11. B. PERTUSSIS FHAB RESTRICTION F R A G M E N T LIBRARIES 73 2.11.1. Construction and analysis of B. pertussis fhaB Sau3Al fragment (FHA-S) libraries 73 2.11.2. Construction and analysis of B. pertussis fhaB HhallHinVWHpall restriction fragment (FHA-H/H/H) libraries (Figures 7-17 to 7-22) 76 2.11.3. Followup study of FHA-H/H/H library clones (Figures 7-21 and 7-22) . . . . 81 2.12. CONSTRUCTION OF B. PERTUSSIS FHAB D N A S E I F R A G M E N T LIBRARIES 84 2.12.1. Construction and analysis of F H A series 70 libraries (Table 7-V) 84 2.12.2. Construction of FHA series 80 libraries (Table 7-VI) 86 2.13. ANTIBODIES A N D I M M U N O L O G I C A L R E A G E N T S 87 2.13.1. Rabbit pAbs produced against wild-type phage f l 87 2.13.2. Mouse mAbs produced against P. falciparum CSP 88 2.13.3. Rabbit pAbs produced against B. pertussis FHA 89 2.13.4. Secondary antibodies 92 2.13.5. Immunochemicals, recombinant CS protein and related material 92 2.14. BIOPANNING (AFFINITY SELECTION OF T A R G E T C L O N E S ) 93 2.14.1. Biopanning random peptide libraries with oc-CSP mAbs 93 2.14.2. Biopanning FHA-70 and-80 libraries (Figure 7-30) 94 2.15. P L A Q U E LIFTS A N D R E L A T E D M E T H O D S E M P L O Y E D TO IDENTIFY A N T I B O D Y - R E A C T I V E C L O N E S IN FHA-70 A N D -80 LIBRARIES 96 2.15.1. Methodology issues 96 2.15.2. Identification of antibody-reactive clones in FHA-70 libraries and assessment of the effect of library amplification on the fraction of antibody-reactive clones recovered 98 2.15.3. Identification of antibody-reactive clones in FHA library 80-A 102 2.15.4. Large-scale propagation of selected FHA-70 and -80 clones 103 2.15.5. Assessment of biopanning enrichment (Figure 7-31) 105 2.16. ELISA M E T H O D " A " : IMMOBILIZED VIRIONS 106 2.16.1. Methodology . 106 2.16.2. ELISA of CSP-library clones (Chapter 4) 107 VI 2.16.3. Preliminary ELISA of random peptide library clones with a-CSP mAbs . . 109 2.16.4. Initial ELISA and B C A assays of FHA-70 and -80 clones (Figures 6-3 and 6-4) 110 2.16.5. Competition ELISA and related B C A assays of FHA-70 and -80 clones . . 112 2.17. ELISA M E T H O D " B " : IMMOBILIZED ANTIBODIES 113 2.17.1. Capture ELISA of selected CSP-library clones (Figure 4-7) 113 2.17.2. Capture ELISA of random peptide library biopan eluates (Figure 5-2) . . . . 114 2.18. IMMUNOBLOTS 114 2.18.1. Assessment of binding to nitrocellulose of FHA-70 and -80 clones (Figure 6-4) and immunoblots of antibody-reactive clones (Figure 6-6, assay (i)) 114 2.18.2. Immunoblots of antibody-reactive FHA-70 and -80 clones (Figure 6-6, (ii)) 115 2.19. A N A L Y S I S OF A S X - P R O T U R N S 116 2.20. PREDICTED S I G N A L PEPTIDE C L E A V A G E 116 Chapter 3 - The "Asx-Pro turn" as a local structural motif stabilized by alternative patterns of hydrogen bonds, and a consensus-derived model of the sequence Asn-Pro-Asn 120 3.1. A B S T R A C T 120 3.2. INTRODUCTION 120 3.3. A N A L Y T I C A L M E T H O D S 122 3.3.1. Molecular databases and datasets 122 3.3.2. Computer software 124 3.3.3. Hydrogen bonds 124 3.4. RESULTS 129 3.4.1. Residues not preceding proline favor /<— /+4 H-bonds; proline favors i*—i + 3 H-bonds 129 3.4.2. Short polar sidechains in residues preceding proline favor + 3 H-bonds 133 3.4.3. Asx-Pro favors a combination of sidechain-backbone and backbone-backbone H-bonds 134 3.4.4. Similar structures are formed by a variety of Asx-Pro H-bond conformations 134 3.4.5. Asx-Pro turns have diverse roles 137 3.4.6. Backbone and sidechain geometry of Asx-Pro turns are not unusual 138 3.4.7. A consensus-derived model for Asn-Pro-Asn agrees with experimental data 139 3.4.8. The proportion of turns versus non-turns increases with structure resolution 142 3.5. DISCUSSION ; 143 vii Chapter 4 - Recognition of phage-expressed peptides containing Asx-Pro sequences by monoclonal antibodies produced against Plasmodium falciparum circumsporozoite protein 145 4.1. A B S T R A C T 145 4.2. INTRODUCTION 145 4.3. RESULTS 149 4.3.1. ELISA overview 149 4.3.2. Variant peptides recognized by Group 4 mAbs 152 4.3.3. Similar effects of residue substitutions on recognition by Group 4 mAbs . . 152 4.3.4. Bivalent PflB2.2 and Pf5A4.1 versus decavalent Pf2Fl . l 156 4.4. DISCUSSION 160 4.4.1. Importance of peptide conformation and tetrapeptide cadence 160 4.4.2. Peptide acetylation may confound epitope analysis . 161 4.4.3. Is peptide length important? 163 4.4.4. Structural implications of the recognition of variant peptides by Group 4 mAbs 164 4.4.5. Multiple Asx-Pro repeats as immunogens 164 4.4.6. Concluding remarks 165 Chapter 5 - Biopanning random peptide libraries (RPLs) with monoclonal antibodies raised against Plasmodium falciparum circumsporozoite protein suggests inherent practical limitations of RPLs 166 5.1. A B S T R A C T 166 5.2. INTRODUCTION 166 5.3. RESULTS 168 5.3.1. Progressive enrichment for target clones with each round of biopanning . . 169 5.3.2. Screening of 96 clones yielded only a limited set of sequences 169 5.4. DISCUSSION 173 5.4.1. Practical issue 173 5.4.2. Biopanning results are essentially unpredictable 174 Chapter 6 - Antigenic analysis of Bordetella pertussis filamentous hemagglutinin using phage display libraries and rabbit oc-FHA polyclonal antibodies 176 6.1. A B S T R A C T 176 6.2. INTRODUCTION 176 6.2.1. Bordetella pertussis vaccinas 176 6.2.2. Role of FHA in pathogenesis 178 6.2.3. Rationale for this study 181 6.3. RESULTS 181 Vl l l 6.3.1. Library construction and evaluation 181 6.3.2. Selection of clones for characterization 182 6.3.3. Sequencing 184 6.3.4. Preliminary ELISA 187 6.3.5. Immunoblots and ELISA 190 6.3.6. Signal peptide cleavage predictions 197 6.4. DISCUSSION 200 6.4.1. Comparison with other studies 200 6.4.2. Candidate immunogens for eliciting protective antibodies 200 6.4.3. Issues related to phage display technology 205 6.4.4. Related studies and concluding remarks 207 Chapter 7 - Issues in vector stability and library construction 209 7.1. A B S T R A C T 209 7.2. INTRODUCTION 209 7.2.1. The expected utility of fUSE« vectors 209 7.2.2. The outcome 212 7.3. RESULTS 212 7.3.1. Early experimentation with frameshift vectors 212 7.3.2. A library of variants of the P. falciparum CSP immunodominant repeats, and development of amber vector fDRW5 219 7.3.3. A B. pertussis fhaB Sau3Al fragment library constructed with conventional vectors 224 7.3.4. Amber vector fDRW613, for Sau3A\ fragment libraries 228 7.3.5. Comparative assay of frameshift and amber vector pseudorevertant production 230 7.3.6. Is a gene fragment library possible? B. pertussis fhaB restriction fragment libraries constructed with frameshift vector fUSEl 234 7.3.7. A strategy for gene fragment and genomic libraries 243 7.3.8. Design of amber vector fDRW70 244 7.3.9. B. pertussis fliaB DNase I fragment libraries constructed with fDRW70 . . 250 7.3.10. B. pertussis fliaB DNase I fragment libraries constructed with fDKWSnn vectors 252 7.3.11. Assessment of FHA-70 and -80 libraries with a-FHA polyclonal antibodies 255 7.4. DISCUSSION 260 7.4.1. General findings 260 7.4.2. Vector issues 262 7.4.3. Concluding remarks 267 ix Chapter 8 - Summary and General Discussion 269 Literature Cited 274 Appendix 297 List of Tables Chapter 2 Table 2-1. E. coli strains used in these studies 39 Table 2-11. Plasmids 40 Table 2-III. Phage display vectors 41 Table 2-IV. Evaluation of host strains and media for plaque formation by amber vectors. 52 Table 2-V. Virions employed in biopanning FHA-70 and -80 libraries 95 Table 2-VI. Predicted signal peptidase cleavage of antibody-reactive FHA-70 and -80 clones 117 Chapter 3 Table 3-1. Effect of proline on backbone hydrogen bonding of preceding residue. . . . 130 Chapter 4 Table 4-1. Summary of a-CSP mAb recognition of CS repeat-derived peptides 161 Chapter 6 Table 6-1. Peptides displayed by FHA-70 and -80 libraries 181 Table 6-11. Summary of antibody-reactive clones selected from FHA-70 libraries 182 Table 6-III. Summary of antibody-reactive clones identified in F H A library 80-A by plaque lifts 183 Table 6-IV. F H A regions grouped by patterns of recognition by antibody 201 Chapter 7 Table 7-1. Evaluation of E. coli host strains for fDRW5 propagation 223 Table 7-11. Construction and assessment of third CSP-library 224 Table 7-III. Characteristics of FHA-S libraries 226 Table 7-IV. FHA-H/H/H libraries: distribution of Hhal/HinPll/Hpall fragment sizes. . 235 Table 7-V. Construction of B. pertussis fhaB DNase I fragment (FHA-70) libraries. . . 250 Table 7-VI. Construction of B. pertussis fhaB DNase I fragment (FHA-80) libraries. . . 254 xi List of Figures Chapter 1 Fig. 1-1. Wild-type Ff phage and multivalent p i l l phage display 2 Chapter 2 Fig. 2-1. Growth curves and estimated doubling times (T d) of E. coli strains 40 Fig. 2-2. fDRW6«« amber vector series for constructing SauiAl restriction fragment libraries 57 Fig. 2-3. Flow chart for experiment summarized in Figure 7-15 65 Fig. 2-4. Partial digest of pASlOO with Sau3Al 73 Fig. 2-5. Relative mobility (Rf) of Haelll fragments of FHA-H/H/H library clones . . . . 80 Fig. 2-6. Host cell doubling times (T d) for FHA-H/H/H library clones 81 Fig. 2-7. Flow chart for followup study of FHA-H/H/H library clones 82 Fig. 2-8. Effect of a-FHA sera purification and biotinylation on recognition of F H A . . . 91 Fig. 2-9. Alternative methods of applying nitrocellulose to lawns 97 Fig. 2-10. Plaque lifts of CSP-library and wild-type f l virions 98 Fig. 2-11. Plaque lifts of biopanned FHA-80 library clones 99 Fig. 2-12. Evaluation of Immulon plates for ELISA 105 Fig. 2-13. B C A assay for quantification of virions bound to Immulon plate wells: effect of incubation time 107 Fig. 2-14. Titration of recognition of CSP-library clones by a-CSP mAbs 108 Fig. 2-15. Control for B C A assay 113 Chapter 3 Fig. 3-1. Local H-bonds in Asx-Pro sequences, Dataset A 126 Fig. 3-2. Local H-bonds in Asx-Pro-Asx sequences, Dataset C 127 Fig. 3-3. Effect of proline on backbone hydrogen bonding 131 Fig. 3-4. Local H-bonds in Asx-Pro, Ser-Pro and His-Pro sequences 135 Fig. 3-5. Alternative hydrogen bonding patterns in Asx-Pro turns 136 Fig. 3-6. Superimposed wireframe models of Asx-Pro turns 137 Fig. 3-7. Ramachandran plot of (j),ij/ dihedral angles in Asx-Pro turns and other proline-containing sequences 138 Fig. 3-8. Sidechain X\ angles of Asx in Asx-Pro turns 139 Fig. 3-9. Superimposed a-carbon traces of Asx-Pro helix caps and flanking residues . . 139 Fig. 3-10. Local H-bonds in Asx-Pro-Asx sequences 140 Fig. 3-11. Superimposed wireframe models of Asx-Pro-Asx peptides 141 Fig. 3-12. H-bonds in Asn-Pro-Asn sequences that form Asx-Pro turns 142 Fig. 3-13. Influence of structure resolution on identification of Asx-Pro turns 143 Chapter 4 Fig. 4-1. Postulated hydrogen bonds within the sequence Asn-Pro-Asn-Ala 147 Fig. 4-2. Phage-displayed CSP-related peptides 148 Xll Fig. 4-3. ELISA with recombinant CS protein and phage-displayed native peptides . . . 150 Fig. 4-4. ELISA with phage-displayed native peptides and single-residue variants . . . . 151 Fig. 4-5A. Group 4 mAb PflB2.2 ELISA with phage-displayed native peptides as well as one-, two- and three-residue variants 153 Fig. 4-5B. Group 4 mAb Pf5A4.1 ELISA with phage-displayed native peptides as well as one-, two- and three-residue variants 154 Fig. 4-5C. Group 4 mAb Pf2Fl . l ELISA with phage-displayed native peptides as well as one-, two- and three-residue variants 155 Fig. 4-6. Comparison of ELISA results for Group 4 mAbs 156 Fig. 4-7. Group 4 mAb PflB2.2, comparison of ELISA methods A and B 157 Fig. 4-8. Relative quantities of virions bound to microtiter plate wells 158 Fig. 4-9. Effect on ELISA signal of varying the quantity of virions 159 Fig. 4-10. Inhibition of ELISA signal by preincubation of a-CSP mAbs with (NPNA) 3 peptide 159 Chapter 5 Fig. 5-1. Overview of 16 biopans 167 Fig. 5-2. Apparent enrichment for target clones with multiple rounds of biopanning . . 168 Fig. 5-3. Initial screen of clones selected from third round of biopanning with Pf2A10 170 Fig. 5-4. Initial screen of clones selected from third round of biopanning with PflB2.2 171 Chapter 6 Fig. 6-1. Sequences of antibody-reactive FHA-70 and -80 clones 185 Fig. 6-2. Distribution of antibody-reactive FHA-70 and-80 clones 187 Fig. 6-3. Early ELISA of FHA-70 and -80 clones . . •. 188 Fig. 6-4. Uniform binding of FHA-70 and -80 clones to nitrocellulose 190 Fig. 6-5. Example immunoblot of FHA-70 and -80 clones 191 Fig. 6-6. Immunoblots of antibody-reactive FHA-70 and -80 clones 192 Fig. 6-7. Variable binding of FHA-70 and -80 clones to Immulon plates 194 Fig. 6-8. Inhibition of ELISA signal by preincubation of a-FHA antibodies with purified FHA 195 Fig. 6-9. Predicted signal peptide cleavage of antibody-reactive FHA-70 and -80 clones 198 Fig. 6-10. Positional comparison of antibody-reactive clones with results of other studies 199 Fig. 6-11. Sequence overlaps in region I and XI clones 202 Chapter 7 Fig. 7-1. Phage display vectors provided by G. P. Smith 211 Fig. 7-2. Early efforts to prepare linear fragments of fUSE5 RF D N A 213 Fig. 7-3. Early preparations of fUSEn RF D N A 214 Fig. 7-4. Pseudorevertant production by fUSEn frameshift vectors 216 Fig. 7-5. Use of chloramphenicol-containing medium to reduce s.s. D N A species . . . . 217 Fig. 7-6. Effects of SI nuclease and restriction endonucleases on fUSE5 D N A preparations 217 Fig. 7-7. Elution of s.s. D N A with 1.3M NaCl in Qiagen column wash 218 Xl l l Fig. 7-8. Construction of phage displaying CSP-related peptides 219 Fig. 7-9. Conversion of frameshift vector fUSE5 to amber vector fDRW5 220 Fig. 7-10. fDRW5 RF D N A preparations for constructing CSP-peptide libraries 221 Fig. 7-11. Assessment of clonal bias in first CSP-library 221 Fig. 7-12. Conventional vectors fDRW20, fDRW21 and fDRW22 225 Fig. 7-13. Assessment by agarose gel electrophoresis of FHA-S library virions . . . . . . 226 Fig. 7-14. Design of amber vector fDRW613 and two-amber vector fDRW613C 229 Fig. 7-15. Frameshift and amber vector stability as determined by pseudorevertant production 231 Fig. 7-16. Effects of SupF and wild-type p i l l on pseudorevertant production 234 Fig. 7-17. Screen for virion-producing transformants in FHA-H/H/H libraries 235 Fig. 7-18. Virion production by selected FHA-H/H/H library clones 236 Fig. 7-19. Assessment of influence of insert size on virion production by selected FHA-H/H/H library clones 237 Fig. 7-20. Correlation of virion production with host cell doubling time, selected FHA-H/H/H library clones . 238 Fig. 7-21. Virion production by 1° and 2° clones of FHA-H/H/H library members . . . . 239 Fig. 7-22. Characterization of selected FHA-H/H/H library 1° and 2° clones 242 Fig. 7-23. Design of amber vector fDRW70 244 Fig. 7-24. Stability of fUSE5 and amber vectors, as determined by pseudorevertant production 245 Fig. 7-25. RF D N A of fUSE5 and fDRW5 propagated in E. coli LE392 247 Fig. 7-26. Pseudorevertants of frameshift and amber vectors 248 Fig. 7-27. A screen for virion-producing clones in FHA-70 libraries 251 Fig. 7-28. A novel class of fDRW70 pseudorevertant 252 Fig. 7-29. Design and construction of conventional vector series fDRW8nn 253 Fig. 7-30. Output from a single round of biopanning FHA-70 and -80 libraries 256 Fig. 7-31. Effect of biopanning on fraction of antibody-reactive clones recovered from FHA-70 libraries: unenriched library versus biopanning output 257 Fig. 7-32. Effect of library amplification on fraction of antibody-reactive clones recovered from FHA-70 libraries 258 xiv Acknowledgements I have incurred a debt of gratitude to many individuals during my studies and research. Many in the B. B. Finlay laboratory have provided practical advice and assistance with my project, and I thank them for this. I wish to specifically thank • Brett Finlay, my supervisor, for the opportunities he provided; • Annette Siebers, for generous provision of antibodies, D N A fragments and other material that made the FHA study possible; • Sharon Ruschkowski, for her cheerful commitment to ensuring that materials I required were made available at unreasonably short notice; • Nancy Carpenter and Mitchell Uh, for indispensible technical assistance that substantially shortened the duration of my project; as well as my supervisory committee • R. E. W. Hancock, N . Reiner and R. A . J. Warren, each of whom contributed to the success of my project; and other individuals, including • G. P. Smith (University of Missouri) and J. Scott (Simon Fraser University) for bacterial strains, phage vectors and random peptide libraries; • R. A. Wirtz (Walter Reed Army Institute of Research) for monoclonal antibodies and recombinant circumsporozoite protein; • U . Hobohm, G. Vriend and R. Hooft (EMBL, Germany), P. Walters and M . Stahl (University of Arizona, U.S.A.) and R. Sayle (University of Edinburgh, Scotland) for computer software; • L . L . Walsh (University of Illinois, U.S.A.) for her useful "Annotated PDB File Listing"; • L . Mcintosh (U. B. C.) for his critical reading of an early draft of material presented in Chapter 3. The studies reported in Chapter 4 were made possible by materials developed or provided by R. S. Nussenzweig and New York University (mAb Pf2A10), Y . Charoenvit and the Naval Medical Research Institute (mAb PvNSV3) and SmithKline Beecham (R32tet32 and PvNSlv20 recombinant proteins). MAbs used in this study were produced under the United Nations Development Programme/World Bank/WHO Special Programme for Research and Training in Tropical Diseases Grant 880068. M y studies were made possible by a 1967 Science and Engineering Scholarship awarded by the Natural Sciences and Engineering Research Council of Canada. Much of the research was supported by an operating grant to B. B. Finlay from the Canadian Bacterial Diseases Network Centre of Excellence. 1 Chapter 1 Introduction 1.1. O V E R V I E W The publication in 1990 of three studies describing peptides displayed on the surface of recombinant bacteriophage brought to the attention of many scientists a novel, simple and powerful approach to the problem of identifying or optimizing ligands for antibodies and other biomolecules. In these model studies, three groups of researchers had each constructed a vast library of >107 six- or fifteen-residue random sequence peptides surface-displayed as N-terminal fusions to a minor coat protein (pill) of an Ff (Ml3 , f l and fd) bacteriophage (Figure 1-1). By technically simple affinity selection methods, they identified novel sequences that bound specifically to cognate antibodies (Cwirla et al. 1990; Scott & Smith 1990) or to streptavidin (Devlin et al. 1990). Several aspects of Ff phage biology made these studies possible and the concept powerful. First, because Ff phage are small and large numbers can be readily accommodated in a small volume ( » 1 0 mL" ), constructing libraries of the required diversity (>10 unique clones; accomplished by "shotgun" cloning of degenerate oligonucleotides into the 5' region of g i l l 1 , the gene encoding p i l l ; Figure 1-1C) was technically achievable. Second, because a virion is structurally simple (Figure 1-1B), the displayed peptides were presented in a comparatively simple (with respect to many expression systems) physicochemical environment, allowing successful affinity selections (called "biopanning" by Scott & Smith 1990) with minimal 'Commonly, phage proteins are identified with Roman numerals prefixed with a lower case "p" {e.g., pi 11, pVIII) while their encoding genes are identified simply with Roman numerals (e.g., "gene III", "gene VIII"). For clarity and convenience, I have employed the prefix "g" to identify phage genes (e.g., "g i l l " , "gVIII"). 2 Fig. 1-1. Wild-type Ff phage and multivalent p i l l phage display. A, B. Wild-type filamentous phage that infect F-piliated Escherichia coli (Ff phage). A. F f phage are narrow, flexible rods of the indicated dimensions. B. Cartoon illustrates inherent simplicity of virion structure: a single-stranded circular genome is surrounded by -2700 copies of the major coat protein pVIII and 4 or 5 copies of each of four species of minor coat proteins, including p i l l which binds to a host cell F-pilus. C. Multivalent pi II phage display vectors fUSE5 and fDRW5 and others employed in this study are derivatives of fd-tet (Zacher et al. 1980) and accordingly possess a tetracycline-resistance gene (Tet R) inserted into the major intergenic (I.G.) region. Inserting foreign D N A into g i l l , at a point corresponding to the p i l l N-terminal signal peptidase cleavage site, results in the display of a foreign peptide as an N-terminal fusion to each mature p i l l molecule, as illustrated in D. As described in the text, a key feature of phage display libraries is that an affinity selectable target peptide and its encoding D N A are packaged together (D). 3 background. Finally and most importantly, each phage-displayed peptide was physically linked to the gene encoding its production: affinity selection of a target peptide corresponded to cloning its encoding gene (Figure 1-1D). Importantly, the displayed peptides appeared to have little effect on phage biology, either during the processes that lead to incorporation of p i l l into a virion, or on the ability of pi l l to initiate the infection cycle by adsorbing to the tip of a host cell F-pilus (Parmley & Smith 1988; Scott & Smith 1990). As well, certain technical aspects of the design of some vectors (to be discussed later) appeared to ensure that libraries free of nonrecombinants could be constructed and to allow that large peptides or proteins might be successfully displayed. Given the power of the concept, the demonstration of its success and the technical merits of certain vectors, it thus seemed possible that phage display libraries (PDLs) could be applied to a variety of research problems in several disciplines, including pathogenesis. Early studies bore this out. Bass et al. (1990) soon demonstrated pill-display of the 191-residue disulfide bond-containing human growth hormone in a properly folded form, while McCafferty et al. (1990) showed that single-chain Fv antibodies could be displayed on p i l l . Later, Kang et al. (1991) showed that functional Fab fragments could be displayed as fusions to the major coat protein pVIII, while Roberts et al. (1992a) showed that a library of randomized pill-displayed variants of bovine pancreatic trypsin inhibitor could be employed to identify a variant possessing >106-fold greater affinity for human neutrophil elastase than the parental peptide. It was in context of these early successful studies that my thesis project began. My general goal was that of exploring, by means of relatively simple models the feasibility of exploiting PDLs in the study of pathogenesis, with the eventual goal of constructing phage-displayed genomic libraries that could, given an adequate molecular probe, be used to identify pathogen-encoded proteins with a specific role in pathogenesis. 4 Concurrently, there has been growth in both numbers and diversity of applications as well as advances in methodology. A sampling of papers published during the past six years shows that epitope characterization (45 papers) and other applications (30 papers) of random peptide libraries (RPLs) are popular. Studies related to display of large polypeptides or proteins, including affinity-maturation studies such as the "directed evolution of a protein" theme of Roberts et al. are also common (34 papers). Studies concerning methodology, such as conformational constraint of displayed peptides (17 papers), are more limited. Although the first PDL was constructed with gene fragments (Smith 1985), few studies (seven papers) have explored this idea further. A number of reviews of phage display have been published, including general overviews (Wells & Lowman 1992; Barbas 1993; Smith & Scott 1993; Clackson & Wells 1994; Perham et al. 1995), reviews of RPLs (Scott 1992; Lane & Stephen 1993; Cortese et al. 1994, 1995, 1996; Scott & Craig 1994; Burritt et al. 1996) and of display of proteins (Dunn 1996), as well as more focussed reviews concerned with "directed evolution" and affinity-maturation themes (O'Neil & Hoess 1995), display of enzymes (Soumillion et al. 1994a) and antibodies (Marks et al. 1992; Winter et al. 1994), design of DNA-binding proteins by use of zinc finger libraries (Choo & King 1995) and applications in neurobiology (Bradbury & Cattaneo 1995). Thus, our initial optimism has been largely justified by successful application of phage display to a variety of biological questions. At the same time, as reviewed later and illustrated in this thesis, it has become clear that not every application attempted is successful and that phage biology may play a larger role in this than initially believed. As with most expression systems, the principal issues of phage display relate to the nature of the displayed foreign peptide or protein and its influence on successful expression and incorporation into virions. Given my goal of assessing the feasibility of constructing phage-display genomic libraries, 5 which requires the display of diverse collections of heterologous peptides, it becomes especially important to understand the ways in which phage and host cell biology are likely to hinder this objective. The following sections review filamentous bacteriophage biology, related aspects of host cell biology, and phage display applications with the goals of (i) providing a general overview of these subjects, (ii) illustrating ways in which phage/host biology might be expected to hinder phage display, and (iii) providing specific and relevant examples of difficulties experienced by others. 1.2. F F P H A G E B I O L O G Y AND E A R L Y P H A G E DISPLAY 1.2.1. Overview M l 3 , f l and fd 2 (reviewed in Rasched & Oberer 1986; Model & Russel 1988; Russel 1991, 1994,1995) are long, narrow bacteriophages of about 7 nra diameter and 900 nm length (Figure 1-1 A) . They are collectively called Ff phage because of their filamentous appearance and almost total dependence on the F-pilus for infection of their E. coli host. The Ff virion consists of a single-stranded (s.s.) D N A genome of roughly 6400 nucleotides packaged in a tube comprised of -2700 copies of the major coat protein pVIII and closed at the ends by four or five copies of each of four species of minor coat proteins (Figure 1-1B). While most bacterial viruses assemble in the host cell cytoplasm and are released by cell lysis, Ff phage do not kill their hosts but rather are continuously extruded through the host cell membrane in a process that couples assembly with export. The process is efficient and virion titers commonly exceed 10 1 2 mL"1 after overnight culture in a suitable host. The infection cycle begins when pi l l (the adsorption protein, seen as a knob-on-stem or 2Except where noted, material presented in this section is derived from the cited reviews. 6 "lollipop" in electron micrographs) adsorbs to the tip of a host cell F-pilus. Through a process that is not yet completely understood, the virion is brought to the cell surface, the s.s. genome is delivered to the host cell cytoplasm, and virion proteins are deposited in the cytoplasmic membrane for re-use during subsequent virion production. Immediately after infection, host polymerases employ the s.s. genome (the (+) strand) as a template for synthesis of a complementary or (-) strand, thereby creating a double-stranded (d.s.; or replicative form, RF) phage genome. Synthesis of the (-) strand is initiated at ori(-), one of several sites within the major intergenic region. Because of the d.s. nature of two adjacent stem loops formed at ori(-), the sequence and conformation of ori(-) bears some resemblance to the (double-stranded) -35 and -10 regions of RNA promoters; these similarities are important for initiation of primer synthesis by RNA polymerase (Higashitani et al. 1997). Following (+) strand complementation, host-mediated transcription from the plasmid-like RF molecule leads to expression of ten genes encoding the coat proteins (pil l , pVI, pVII, pVIII, pIX), assembly/export proteins (pi and pIV) and proteins involved in D N A replication and its control (pll, pV and pX). As presently understood, the relative levels of these proteins are mainly determined by a mechanism that, while simple in concept, is extraordinarily complex in detail. In simplistic terms, the genome is organized into two main coding regions, one for proteins required in large quantities (such as pVIII) and possessing a relatively strong initial promoter; the other for proteins required in fewer numbers (such as pill) and possessing a less efficient initial promoter. Multiple promoters within each region create a cascade effect whereby 3' genes are transcribed more frequently than 5' genes. A host mechanism that degrades mRNA transcripts from the 5' end magnifies the cascade effect, with the net result that a finely tuned balance of synthesis of the ten phage proteins is achieved with little apparent feedback control (see Fulford & Model 1988a, 1988b for exceptions to this). 7 Synthesis of the (+) strand is initiated by p l l , which nicks the (+) strand at ori(+), located in the major intergenic region, allowing the freed 3' end to serve as a primer for rolling circle replication on the (-) strand template. As the strand is completed, p l l ligates the molecular ends to form a closed circle. The fate of new (+) strands varies. Early in infection these serve as templates for the formation of additional RF molecules in the manner described earlier. As the levels of the s.s. D N A binding protein (SSB) pV increase (as a function of time, increases in RF D N A copy number, and consequent increase in RF-directed gV mRNA synthesis), s.s.-to-RF conversion is arrested as newly-synthesized (+) strands become sequestered by pV. Virion assembly and export take place in the host cell membrane. Following synthesis, all five coat proteins are inserted into the inner membrane (Endemann & Model 1995). While pVI, pVII and pIX lack signal peptides, p i l l and pVIII are synthesized with N-terminal signal peptides that are cleaved upon membrane insertion. Anchored by C-terminal hydrophobic regions, both p i l l and pVIII are oriented with their N-termini in the periplasm. Membrane insertion of pVIII is Sec-independent (Pugsley 1993; Kuhn 1995), while pi l l may require the Sec apparatus (Peters et al. 1994). In current models, pi and pIV form a gated, multimeric export channel with pi and pIV localized to the inner and outer membranes respectively. Assembly/export is initiated by interactions among (i) a "packaging" signal within the intergenic region, (ii) export channel protein pi, (iii) host thioredoxin, and (iv) membrane-inserted minor coat proteins pVII and pIX. In a process that requires ATP hydrolysis and in which proton motive force is also important (Feng et al. 1997), phage are extruded through the pI/pIV export channel with the concomitant replacement of pV SSB by pVIII. Virion assembly is completed by addition of pVI and p i l l to the end of the extruding virion. 8 1.2.2. fd-tet and other ori(-) mutants Surprisingly, mutants with defects in ori(-) are viable. Kim et al. (1981) constructed a number of viable M l 3 mutants containing deletions in ori(-), while Zacher et al. (1980; see also Smith 1988) constructed the cloning vector fd-tet (the parent of many phage display vectors) by splicing a 2.8 kbp tetracycline-resistance gene into ori(-) at a position that interrupts one of the two stem loops required (Higashitani et al. 1997) for primer synthesis. In the M13 mutants and fd-tet, initiation of (-) strand replication is delayed and proceeds slowly by an unidentified alternative mechanism (Kim et al. 1981; Smith 1988). Relatively recently, M13 mutants with ori(-) defects have been shown to induce the E. coli SOS response (Higashitani et al. 1992, 1995), presumably because of delays in (-) strand synthesis or related events leading to accumulation of s.s. DNA, the apparent SOS system inducer (Walker 1996). Considering the phenotype similarities of M l 3 ori(-) mutants and fd-tet, it seems likely that fd-tet also induces the SOS response. Ff mutations that block phage assembly while allowing D N A replication to continue are normally lethal to the host cell because of intracellular accumulation of phage D N A and gene products (Smith 1988). In fd-tet, because of its ori(-) defect, the RF D N A copy number is reduced about 13-fold, with the result that while RF DNA levels increase in morphogenetically defective mutants of fd-tet, these levels do not exceed those of wild-type phage (Smith 1988). As described later, Parmley and Smith (1988) exploited the viability of fd-tet mutants in designing phage display vectors. Compared to wild-type Ff, fd-tet plaques are extremely small (for an example, see Materials and Methods, Figure 2-10) and turbid, virion production after overnight culture is reduced four-fold, and yields of RF and s.s. D N A are reduced more than 10-fold (Smith 1988). M l 3 ori(-) mutants have similar plaque phenotypes and reduced virion yields (Kim et al. 1981). 9 Infectivity (infectious units per physical particle) is reduced in fd-tet to as little as 2.5% of wild-type levels (Smith 1988) although infectivities as high as 50% of wild-type levels have been reported for the fd-tet derivative fUSE2 (Parmley & Smith 1988). 1.2.3. pill pil l is required not only for its role in F-pilus adsorption but also for terminating virion assembly and stabilization of the virion particle. Delays in supplying p i l l during assembly/export lead to production of multilength virion particles (polyphage) containing two or more unit-length phage genomes (Model & Russel 1988). Supernatants of wild-type Ff phage routinely contain about 5% polyphage. The knob-on-stem appearance of pi l l in electron micrographs reflects its two functional domains. The N-terminal two-thirds of the molecule mediates virion adsorption while the C-terminal one-third functions in virion assembly and structure (Armstrong et al. 1981; Nelson et al. 1981; Crissman & Smith 1984; Model & Russel 1988). Thus, mutants unable to produce at least the C-terminal portion of p i l l produce virions containing multiple unit-length genomes (polyphage), while mutants producing p i l l that lacks 169 residues from the middle of the N-terminal domain (mature pi l l contains 406 residues) produce normal numbers of unit-length virions (monophage) but are non-infectious (Crissman & Smith 1984). p i l l contains two sets of repeats of GGGSE (in one-letter amino acid code) or related variants. The second and longer of these repeats occurs about two-thirds (Beck & Zink 1981) from the N-terminus of the mature protein. Given its high content of glycine and polar residues (deemed important in linker design, as reviewed in Alfthan et al. [1995]) and its position (which matches roughly with the estimated size of the N-terminal domain), it seems likely that these repeats serve as a flexible linker connecting the N-terminal knob and C-terminal stem. Ff phage-infected cells exhibit a number of membrane-associated phenotypes including 10 accumulation of intracellular membranes and increased leakage of periplasmic proteins. Many of these effects are mediated by p i l l , especially by its N-terminal 98 amino acids (Boeke et al. 1982; Rampf et al. 1991). p i l l additionally mediates resistance to Ff superinfection and several other phenotypes (Boeke et al. 1982). 1.2.4. Development of pill-based phage display The first PDL was a gene fragment library constructed by Smith (1985), who cloned fragments of EcoRI endonuclease and methylase enzymes genes into a natural BamRl site of gil l of wild-type phage f 1, creating a small library of virions displaying peptides in the middle of pi l l . Although identification of a single clone possessing a 171 -bp insert reactive with a-EcoRL antibody demonstrated that "phage display" held promise, several problems related to phage biology required attention. Plaques of the identified clone were very small or barely detectable, the stationary phase titer was 100 times lower than wild-type levels, infectivity was 25-fold lower, the incidence of polyphage was higher, and recombinants in which the original insert was lost or altered were found to possess a marked selective advantage during virion propagation. Later, Parmley and Smith (1988) showed that it was possible to affinity select rare target o virions displaying peptides of interest from a "mock library"containing a 10 -fold excess of virions lacking the target peptide. Their success derived in part from improvements in vector design that reflected several aspects of phage biology. First, the cloning site was moved from mid-gill to a position corresponding to the N-terminus of mature p i l l , with the expectations that displayed peptides would interfere less with p i l l adsorption function and enjoy greater conformational freedom. Secondly, fd-tet (versus wild-type fl) was employed as a parent for new vectors, with the expectation that the reduced gene expression and virion production characteristic of fd-tet (Smith 1988) would diminish the host-toxicity of difficult-to-export 11 peptides (Smith & Scott 1993). Finally, knowing from earlier studies that pill-deficient mutants of fd-tet were viable, Parmley and Smith developed a "frameshift" vector (fUSEl) in which a gi l l frameshift at the cloning site prevented production of p i l l and thus of infectious virions, unless gil l possessed a foreign insert able to restore the gi l l reading frame. This would allow construction of libraries where each infectious virion could be considered a recombinant, a clear advantage (i) in eliminating overgrowth of nonrecombinants during library propagation, and (ii) assessing cloning success and the "completeness" of a library. Some problems remained. The already-low infectivity of fd-tet (Smith 1988) was further reduced in certain fUSEl recombinants. Thus, while a 20-bp insert (derived from a synthetic T7 R N A polymerase promoter sequence) had no apparent effect on infectivity, 54-bp (P. falciparum CSP repeats) and 71-bp inserts (from E.coli lacZ) reduced infectivity by 50% (compared to the non-frameshift vector 1TJSE2), while a 335-bp insert (from lacZ) reduced infectivity 20-fold. 1.2.5. The first random peptide libraries The three earlier-mentioned RPL applications of Cwirla et al. (1990), Devlin et al. (1990) and Scott and Smith (1990) validated Parmley and Smith's (1988) proposal that large libraries of phage-displayed peptides could be used to identify epitopes, mimotopes (Geysen et al. 1987) or other mimetic peptides. Some of this success would appear to derive from continued consideration of phage biology. Thus, Scott and Smith, in constructing their RPL of 4 x 10 hexapeptides (-69% of the 20 6 possible sequences), employed a new vector (fUSE5) that incorporated several features intended to facilitate peptide display. To minimize problems in signal peptidase processing of recombinant p i l l , the first two residues following the signal peptide cleavage site were retained (with Asp substituting for the normal Glu in Ala-Glu) in their clones; these were followed by a random hexapeptide sandwiched between Gly-Ala and 12 Gly-Ala-Ala-Gly-Ala "linker" peptides. Other features of fUSE5, including a g i l l frameshift, ensured that non-recombinant virions were excluded from their library. Cwirla et al. (1990) also used an fd-tet-derived frameshift vector to construct their hexapeptide RPL; curiously, they inserted random peptides immediately following the signal peptide cleavage site. In the RPL of Devlin et al. (1990), constructed without a frameshift vector, 15-residue peptides were separated from the signal peptide cleavage site by Ala-Glu and from the remainder of p i l l by Pro 6. In spite of these and the previously-noted early successes with display of proteins such as human growth hormone, bovine pancreatic trypsin inhibitor variants and antibodies, questions remained concerning whether phage display would find broad applicability. Indeed, a review of the issues involved might predict that many applications might not succeed, being limited by the biology of the host cell. 1.3. R O L E O F T H E H O S T C E L L As with other applications involving production of engineered proteins, peptide or protein display by Ff phage requires host cell processes for both synthesis and export of recombinant p i l l and pVIII molecules. Thus the same cellular processes that sometimes interfere with recombinant technology generally can be expected to interfere with phage display. Importantly, phage display involves N-terminal fusions to phage coat proteins p i l l or pVIII, and a number of studies have shown that the N-terminal mature region of a protein plays a role in its export. 1.3.1. Signal peptides Both p i l l and pVIII are synthesized with signal peptides. Indeed, much of our understanding of signal peptide processing has derived from studies of this process in pVIII. 13 Both processing by signal peptidase and subsequent virion production are affected in some pVIII mutants in which residues following the cleavage position -- i.e., residues +1, +2, of the mature protein -- are altered. Thus, while pVIII preprotein cleavage is efficient when Ala at position +1 is replaced with Phe, Leu, Val, Gly or Ser, cleavage is reduced by substituting Cys and abolished by Pro or Thr. Glu—»Leu and Glu—>Tyr substitutions at position +2 also retard pVIII processing and, in the case of Leu, affect the kinetics of virion production (Boeke et al. 1980; Russel & Model 1981). Although most residue substitutions in positions +1 to +5 have no apparent effect on virion production (Iannolo et al. 1995; Williams et al. 1995), certain substitutions (most notably by Tyr) can reduce virion titers dramatically, to <102 mL" 1 culture supernatant (Iannolo et al. 1995). Whether this was due to hindered signal peptidase activity was not determined. Early difficulties {e.g., Greenwood et al. 1991; Felici et al. 1991; see also Perham et al. 1995) in employing pVIII for multivalent display of peptides larger than six amino acids raised the question of whether this was due to issues such as (i) a "size" restriction related to virion assembly {e.g., steric interference in pVIII-pVIII interactions) or export through the pI/pIV channel, or (ii) issues related to pVIII synthesis and processing. Using X-ray diffraction studies and model building, Malik et al. (1996) have recently shown that it is theoretically possible to display much larger structures on pVIII than it has been thus far possible to accomplish, suggesting that a size restriction of practical significance does not exist. Importantly, they also showed that (for the small number of recombinants analysed) the degree of success in display of peptides on pVIII correlated better with efficiency of preprotein processing by signal peptidase, rather than with the size of the displayed peptide. Studies of the N-terminal mature region of p i l l are limited to a single report (Peters et al. 1994), showing that (i) that positively charged (Arg) residues within the N-terminal decamer 14 of an inserted peptide reduced virion titers in a dose-related manner from ~10 1 0 m l / 1 to as little as 103 mL"1, and (ii) that prlA (secY) suppressors could restore virion production of most defective clones 106- to 107-fold to wild-type levels. Charged residues at the N-terminal mature region have been shown to hinder translocation of other recombinant proteins (e.g., Li et al. 1988; Yamane & Mizushima 1988). It has been suggested (i) that these disrupt a net dipole formed around the hydrophobic domain of the signal peptide and required for efficient translocation (reviewed in Boyd & Beckwith 1990; Izard & Kendall 1994) and (ii) that clusters of these residues may, by conforming to the "positive inside" rule derived by von Heijne (1994) from study of periplasmic and cytoplasmic loops of membrane proteins, serve to convert a signal peptide to a membrane anchor. The demonstrated role of the N-terminal mature region is apt to be important in phage display, for peptides displayed in nearly all RPLs are entirely contained within a postulated N -terminal "critical 30-residue export initiation domain" (Andersson & von Heijne 1991; Nielsen et al. 1997; also see Summers & Knowles 1989). Notably, the popular display vector f A F F l (Cwirla et al. 1990) allows foreign peptide insertion at, rather than a few residues away from (as in fUSE5, Scott & Smith 1990), the +1 position. Although Burritt et al. (1996) have suggested that there is no compelling evidence that this is problematic, data reviewed above suggest otherwise. Moreover, a survey by Nielsen et al. (, 1997 #196; see also von Heijne 1986) of prokaryotic signal peptides and flanking sequences reveals a marked positional distribution of amino acids around the signal peptidase cleavage site, including not only amino acid preferences in preprotein positions (-1, -2, ...) but also in the first positions of the mature protein. Thus, Pro is relatively common in positions -6 to -4 and +2 to +6 but rare (mol% <0.53) in positions -3 to +1. Ala summary data derived from raw data employed in the study by Nielsen et al. and provided by the authors via their FTP server at ftp://virus.cbs.dtu.dk/pub/signalp. 15 (mol%=38.3) is the most common residue at +1, while Asp (16.2%) and Glu (15.0%)) predominate at +2. Notably, these or similar residue pairs initiate mature p i l l of wild-type Ff phage (Ala-Glu) and of recombinants in two frequently employed RPLs, those of Devlin et al. (1990, Ala-Glu) and Scott and Smith (1990, Ala-Asp). The possible effects of defects or delays in p i l l or pVIII signal peptide cleavage are interesting. Because both molecules possess hydrophobic domains at their N - and C-termini (signal peptide and membrane anchor, respectively), they could readily be converted to polytopic membrane proteins by recombinant inserts that affected signal peptidase activity. Considering that phage gene expression is crudely regulated, an anticipated result would be accumulation within the host membrane of p i l l (which at normal levels mediates a number of membrane effects) or pVIII (-2700 copies are required per virion), with consequent pleiotrophic effects on host viability. 1.3.2. Other aspects of membrane insertion/translocation The mature protein sequence can influence more than signal peptidase activity. For example, the normally Sec-independent membrane insertion of pVIII becomes Sec-dependant when the periplasmic loop (transiently formed between the hydrophobic core of the signal peptide and the C-terminal membrane anchor) is increased in size from 20 to 118 residues, and in a related way, Sec-dependance of a leader peptidase construct (studied as a model membrane protein rather than for its physiological role) increases concomitantly with gradual increases in size - from 25 to 65 residues — of its periplasmic loop (reviewed in von Heijne 1994). Several studies (reviewed or reported in von Heijne 1994; Cao et al. 1995) point to a role for both negatively and positively charged residues in markedly influencing the efficiency and success of membrane translocation. Thus, increasing the negative charge of the periplasmic loop in pVIII and leader peptidase constructs results in a corresponding increase in dependance on the 16 electrochemical membrane potential for efficient translocation or, in the case of some pVIII constructs, on the Sec apparatus (Cao et al. 1995). Finally, inserting as few as four or five tandem Asp and Glu residues at the beginning or middle of the periplasmic loop of pVIII can inhibit membrane insertion or processing by leader peptidase to as little as 18% of wild-type levels (Cao et al. 1995). In context of these findings, it is interesting that Grihalde et al. (1995) constructed a 30-residue RPL in which variable residues followed the N-terminal mature region sequence D Y K D D D D K A E T A . This sequence includes an epitope tag (underlined residues) incorporated with the idea that loss of the tag would be diagnostic for N-terminal proteolysis of the p i l l fusions. Surprisingly, the authors noted without comment that only 5-10% of virions displayed an intact tag and that infectivity (the fraction of physical particles that yield productive infections) of pooled library virions was only 0.4%. Conceivably, their findings may reflect a cellular response to problems in membrane translocation of the charge-rich sequence. Indeed, the neural networks developed by Nielsen et al. (1997) for predicting signal peptides conclude4 that the pi l l preprotein sequence encoded by the Grihalde et al. construct lacks a signal peptide. 1.3.3. Genomic libraries, membrane anchors and secondary signal sequences A related problem is likely to arise in genomic PDLs, for a surprisingly large fraction (20%>) of random chromosomally-derived sequences code for peptides that can serve as N-terminal signal peptides (Kaiser et al. 1987). Given also that (i) signal peptides and membrane anchors are difficult to distinguish by prediction algorithms (Nielsen et al. 1997), suggesting a relative ease of interconversion, and (ii) sequence requirements for transmembrane segments 4Prediction was made using the WWW server (http://www.cbs.dtu.dk/) of the Center for Biological Sequence Analysis using the Grihalde et al. preprotein sequence V K K L L F A I P L V V P F Y S H S - D Y K D D D D K A E T A , where '- ' indicates the cleavage site in wild-type p i l l . The sequence yielded a cleavage site score well below (66% of) the minimum default cutoff value. 17 include only minimal length (>12 residues) and hydrophobicity (Pugsley 1993), then a potentially large fraction of genomic PDL clones might be expected to contain motifs that hinder translocation, serve as anchors, or yield unexpected cleavage of p i l l or pVIII fusion products. Presumably, the fraction of clones affected will increase with gene fragment size. In terms of library "completeness" (the fraction of coding sequences successfully displayed on a virion surface), effects will likely be greatest in libraries of low redundancy (/. e., there is little sequence overlap among clones) such as those constructed with restriction fragments, and least where redundancy is greatest, such as those derived from random DNase I digestion (Anderson 1981) of chromosomal DNA. 1.3.4. Codon usage and context Codon usage and translational context in E. coli (reviewed^ briefly in Smith & Smith 1996; Berg & Silva 1997) may adversely affect synthesis of recombinant pVIII (especially; in view of the -2700 copies required per virion) and pi l l (less so; only 4-5 copies per virion), particularly for genomic PDLs derived from genomes with substantially different G+C content from that of E. coli, and with correspondingly different codon preferences. Codon usage within the Bordetella pertussis (%G+C = 66-70 versus -48-52% for E. coli; Pittman 1984) fliaB gene, which was employed in phage display studies presented in this thesis, provides a relevant example. Analysis of 196 fhaB Arg codons shows that the codon most common in E. coli (CGU) tends to be avoided, and while this is largely compensated for by use of an alternate E. co//-favoured codon (CGC), fhaB employs other Arg codons rarely used by E. coli. These include 15 occurrences of A G G , the rarest of all sense codons (with a similarly rare cognate tRNA) employed by E. coli under varying growth conditions (Dong et al. 1996) and known to adversely affect recombinant gene expression in E. coli (reviewed in Kane 1995). Notably, E. coli avoids not only A G G but also the A G - G triplet (where ' - ' denotes a boundary between 18 codons), for reasons that appear related to translation (Smith & Smith 1996; Berg & Silva 1997). That the A G - G triplet accounts for fewer than 4% (versus an expected 25%) of a dataset of 742 E. coli tetranucleotides (Smith & Smith 1996) but 38% of a similar fhaB dataset (my data not shown) suggests difficulties in phage display of fhaB-derived peptides. 1.3.5. Molecular chaperones and export targeting Nascent and completed proteins interact with a variety of host molecules with roles in folding and export. These include molecular chaperones such as GroEL/GroES as well as SecB and other components of the Sec apparatus (Pugsley 1993; Harti 1996; Mayhew & Hard 1996; Murphy & Beckwith 1996). While it is not yet clear how cytoplasmic proteins are distinguished from those destined for export, the mere attachment of a signal peptide to a normally cytoplasmic protein does not ensure its export (Summers & Knowles 1989; Andersson & von Heijne 1991). Rather, the protein must possess certain' general motifs (patterns of hydrophobicity, positively-charged and flexible polypeptide segments) that allow maintenance, by molecular chaperones, of an unfolded state required for membrane translocation. Although chaperone-assisted phage display of antibodies (see, for example, Lah et al. 1994; Soderlind et al. 1995) and of recombinant proteins generally (reviewed in Wall & Pluckthun 1995) have been explored (with mixed results), little is understood regarding how export of recombinant pi l l and pVIII are likely to be influenced by molecular chaperones. Conceptually, fusion of small peptides to large, normally exported molecules such as p i l l (424 residues) should have little effect on export competence. Conversely, fusion of a large, normally cytoplasmic protein to the relatively smaller pVIII (50 residues) would be expected to be pose greater difficulties. 1.3.6. Proteases Protein degradation (reviewed in Miller 1996) is an essential and highly active component 19 of E. coli metabolism. Although much of this degradation is used to regulate levels of functional proteins, much also is concerned with eliminating damaged proteins. While the ways in which targets for degradation are identified are not well understood, certain kinds of alterations tend to increase the rate of degradation of otherwise stable proteins, including the production of longer or shorter versions of a normal protein, or amino acid substitutions that lead to exposure of structural elements not normally accessible in a native protein. Proteins and proteases have evolved together, it has been suggested, such that protein evolution (and thus recombinant technology) is constrained by proteolytic machinery effectively designed to recognize "non-native" elements. In context of these ideas, recombinant p i l l and pVIII, appearing abnormal, may be candidates for degradation. Although there is some evidence, illustrated by later examples, that protein degradation is an issue in phage display, the issue is seldom examined. Indeed, our understanding of the extent to which protein degradation occurs may be obscured by the common practice of inferring the sequence of a displayed peptide from that of its encoding insert into g i l l or gVIII, without corresponding assessment (such as by SDS-PAGE and Western blotting) of p i l l or pVIII fusion products. 1.4. I M P A C T O F P H A G E A N D H O S T B I O L O G Y O N P H A G E DISPLAY Although in some applications of recombinant gene expression, host biology-related problems with the quantity and quality of exported proteins can be (or, out of practical necessity, must be) ignored, such matters assume greater importance with PDLs. This is because much of the utility of a PDL will derive from unbiased and reliable display of all possible encoded proteins. While issues such as library completeness, bias and clonal competition have been raised from time to time in applications of phage display, such as in explaining unexpected results, few studies have addressed the issues directly. As suggested by 20 material reviewed in the preceding sections, there is evidence that phage and host biology may interfere with successful phage display of certain peptides and proteins to a greater extent than commonly recognized. In reviewing applications of phage display, the following section provides a sense of (i) the degree to which such interference is in issue, and (ii) the extent to which it has been examined. 1.5. APPLICATIONS AND INNOVATIONS IN P H A G E DISPLAY 1.5.1. Random peptide libraries (a) Library diversity and bias. RPLs displaying peptides of lengths 6 to 38 (36 residues randomized) have been described. A concern, particularly with longer peptides, is whether a library can contain all possible sequences. Although a hexapeptide library containing all of the 20 6 (6.4 x 107) possible clones can be constructed with only moderate difficulty (e.g., Scott and Smith [1990] plated transformants from multiple electroporations on 64 large N U N C culture dishes to construct their 4 x 107-member library) and larger libraries are achievable, a practical upper limit (~109 unique clones) dictates that libraries of peptides longer than seven residues risk being "incomplete". "Completeness" is not essential, for libraries of limited diversity yield useful results. Thus, although the library of Devlin et al. (1990) displays oniy 1/109 of the 20 1 5 possible 15-residue peptides, Motti et al. (1994) successfully employed it to identify a MAb-reactive peptide that, although it shared only six (of 13) residues with the native antigen, could be used to induce antigen-reactive antibodies. Similarly, Schellekens et al. (1994) employed the Devlin library to identify a MAb-reactive peptide that shared only three residues with the native antigen (glycoprotein D of herpes simplex virus type 1, HSV-1) but, when used as an immunogen (as a BSA-hapten conjugate), nevertheless protected mice against subsequent HSV-21 1 challenge. Still, lack of diversity may have limited these successes for in each study only a single sequence was isolated from among the 39 (Motti et al. 1994) or 5 (Schellekens et al. 1994) antibody-reactive clones analysed. A related issue concerns over- and under-representation of specific amino acids. Much of this is unavoidable and reflects the bias inherent in the genetic code. Commonly, RPLs are constructed with only 32 codons (e.g., of the form N N K , where K = G or T) with the effect of reducing inherent bias from a maximum amino acid molar ratio of 6:1 (e.g., Arg versus Met) to 3:1. Of greater concern is bias due to biological selection against specific residues or sequences, such as position-specific bias against residues in the early mature region of p i l l or pVIII fusion products. Realistic assessment of such bias requires that relatively large numbers of randomly selected library members be analysed. Except for a limited but otherwise careful study by DeGraaf et al. (1993) showing no apparent bias, such analyses have not been reported. Crude measures of bias, possibly indicative of sequence- or position-specific effects, can be derived from analysis of the overall (i. e., position-independent) distributions of amino acids in a small number of randomly selected clones. Cursory analyses suggest that bias is not extreme. Cwirla et al. (1990) analysed 52 randomly selected clones and found that the amino acid content of the displayed peptides was not entirely random (a %2 test on their data yields p < 0.01), Gly and Lys occurring ~1.8-fold more often than expected, and Pro about half as often as expected. Similar over- and under-representation of Gly and Pro has been found in other RPLs (Jellis et al. 1993; Burritt et al. 1996). analyses of data published to show lack of bias surprisingly show that Gly occurred 2.1-fold more often than expected in one RPL (Dybwad et al. 1995a) and that less than one-third the expected Cys occurred in a second RPL (Dybwad et al. 1995b). Blond-Elguindi et al. (1993) also found bias favouring Gly and strongly disfavouring Cys in their RPL, as well as position-specific effects. Also, and in a manner 22 consistent with their expected influence on membrane insertion and signal peptide cleavage, Asp and Glu were slightly more abundant towards the N-terminus, while the reverse was true for Lys. Interestingly, Kay et al. (1993) found apparent selection against odd numbers of cysteines in their 38-residue RPL and reasoned that an unpaired cysteine could form a bond with one of the eight cysteines of native p i l l , with a subsequent effect on phage infectivity. Zhong et al. (1994) have also found a bias against unpaired cysteines. (b) Representative results - biopanning with antibodies. An example of a straightforward and successful application of RPLs is provided in a study by Conley et al. (1994), who biopanned a 15-residue RPL of limited diversity (9 x 107 unique clones) with an anti-HIV-1 (human immunodeficiency virus-1) mAb. Conventional biopanning methods (Smith & Scott 1993) followed by immunoscreening (colony lifts of phage-secreting host colonies) of putative positive clones yielded 20 mAb-reactive clones that could be classified into four consensus groups displaying X X D K W (8 clones), X L D R / K W (5), E X D R / K W (4) and E L D R / K W (3) and subsequently used to identify E L D K W as the native epitope. Not all biopanning experiments are so successful. For example, after three rounds of biopanning the hexapeptide RPL of Scott and Smith (1990) with oc-keratin mAbs, Bottger and Lane (1994) found that only one of seven mAbs had substantially enriched for target clones. In later studies, Bottger et al. (1995b) had greater success with a 20-residue RPL than with a 12-residue RPL; nevertheless, the small number of 20-residue sequences recovered and their limited sequence similarity with the native protein were not enough to identify an epitope. Rather than a consensus sequence, Sioud et al. (1994) found a compositional "theme" after biopanning a RPL with TNF-a (tumor necrosis factor) human autoantibodies. Thus, of 63 affinity selected clones chosen for analysis, 46 displayed Ser- and Pro-rich sequences ASSLLASSP (20 clones), N S S P Y L N T K (16) or PQSPGSSFP (10) and the remaining 17 23 clones also contained Ser or Pro. Some biopannings of RPLs with antibodies (or with other ligates) fail to yield meaningful results (Lane & Stephen 1993; Smith & Scott 1993) or yield misleading results. Smith and Scott (1993) have noted that two consensus sequences arise repeatedly with a variety of antibodies and other ligates. These, PWXWL (X is usually A or E) and GDWVFI, presumably bind some other component of the system (possibly streptavidin, Roberts et al. 1993). Indeed, phage-displayed peptides that bind to these components have themselves been identified by biopanning RPLs. These include streptavidin (Devlin et al. 1990; Kay et al. 1993), biotin (Saggio & Laufer 1993), mouse IgG Fc region (Kay et al. 1993) and plastic (certain Tyr- and Trp-rich sequences; Adey et al. 1995). Comprehensive studies (Bottger et al. 1995a; Stephen et al. 1995) of 6-, 12- and 20-residue RPLs biopanned with a total of 24 mAbs suggest other reasons for a lack of success in some biopannings, including that the epitopes recognized by some mAbs, such as the "linear but discontinuous" K K X 1 . 4 S T S X 0 . 4 H X K variants of a p53 epitope, are substantially longer than those provided in libraries commonly employed. (c) Biopanning with non-antibody molecules. Similar results derive from panning RPLs with molecules other than antibodies. That is, as illustrated below, the sequences of the affinity selected peptides match those of the native ligate in varying degrees or not at all. Thus, "good" matches were found by Cheadle et al. (1994) and Sparks et al. (1994), who independently biopanned RPLs with fusion proteins containing the Src homology 3 (Src SH3) domain and identified a common motif (RPLPXXP) corresponding to a previously identified Src SH3 domain binding motif. Similar success was enjoyed by Healy et al. (1995), who biopanned 6- and 15-residue RPLS with the integrin ocvP3 and identified a diverse collection of sequences containing the RGD triplet. In contrast, Smith et al. (1993) biopanned a hexapeptide RPL with S-protein (a fragment of bovine pancreatic ribonuclease) and identified 24 a sequence motif, F / Y N F E / V ' / V L / V , that bore little resemblance to the natural ligand. A n even less-defined but important motif was found by Blond-Elguindi et al. (1993) who, by biopanning 8- and 10-residue RPLs with the molecular chaperone BiP, identified a heptameric motif (best described as 0 w / x 0 X 0 X 0 , where 0 = Trp, Leu, Phe or other large hydrophobic amino acid) that fits well with the role of BiP in recognizing partially folded polypeptides. Studies by Kay et al. (1993) illustrate several aspects of biopanning RPLs. Panning their 38-residue RPL with streptavidin yielded two classes of sequences, the first providing a consensus sequence (HP^/ M X, X = a non-polar residue) similar to that previously identified by phage display (Devlin et al. 1990), the second providing no clear motif. Intriguingly, clones of the non-motif class bound to streptavidin with ~5-fold greater affinity than selected H P ^ / M X clones. The importance of context and a possible advantage of disulfide-mediated constrained presentation was suggested by the finding that the affinity selected sequence *CHPQAC* (* denotes potential disulfide bond partners) bound 100-fold less efficiently after treatment with a reducing agent. (d) Substrate phage. "Substrate phage" are a class of RPLs pioneered by Matthews and Wells (1993) to identify peptide substrates of proteases. In a study that illustrates their use, Matthews et al. (1994) constructed a substrate phage library by inserting a randomized pentapeptide "substrate" flanked by di-alanine residues between (i) a binding domain (a variant of human growth hormone, hGH) and (ii) p i l l . After being immobilized on hGH receptor molecules, the library was treated with furin to release clones bearing a furin substrate. Six rounds of enrichment led to a consensus motif which was used as a "seed" for a second series of randomized libraries. Furin selection from these libraries yielded a more defined motif, L / p R R F % R P . 25 1.5.2. Phage-displayed proteins and alternative methods of display Much of the success in phage display of larger peptides and proteins has been attributed to the development of alternative vectors. These can be classified (Smith 1993) according to (i) the choice of pi l l or pVIII as carrier molecule, (ii) the number of fusion products displayed per virion, and (iii) the vector strategy employed. In the Type 3 (or multivalent pil l) systems commonly employed in RPLs, vectors are constructed by inserting a cloning site into g i l l with the result that all 4-5 copies of p i l l display a foreign peptide (Figure 1-1C and D). In Type 3+3 ("monovalent" or phagemid) display, recombinant gil l is carried on a phagemid (a plasmid containing the Ff origins of replication and packaging signal). Superinfection of phagemid-harboring cells by a helper phage (an Ff phage with a defective packaging signal) results in synthesis of phagemid s.s. D N A and its preferential packaging into infectious virion-like particles comprised of both wild-type and recombinant p i l l . Recombinant g i l l expression is commonly controlled so that, on average, no more than one recombinant p i l l molecule is incorporated per phagemid particle. In Type 33 display, the genes encoding wild-type and recombinant p i l l are included in a single phage genome; not uncommonly, recombinant g i l l is inserted into the multiple cloning site of an M13 cloning vector. Type 8, 8+8 and 88 vectors are the pVIII multivalent and oligovalent counterparts to the multivalent and monovalent p i l l vectors. The perceived relative advantages of these systems relate to two main issues. The first concerns intrinsic affinity versus functional avidity: affinity selection from monovalent PDLs, it is believed, leads to identification of target clones with higher affinity for their cognate molecule than selection from multivalent libraries where multiple low-affinity interactions create high functional avidity. The second issue reflects phage and host biology. Here the belief is that reducing the copy number of a pi l l or pVIII fusion product can improve the odds 26 that an intrinsically toxic or difficult-to-export product will be successfully displayed on the virion surface. Accordingly, monovalent vectors are often employed for display of larger polypeptides or proteins. Type 8 display is uncommon because of early difficulties in displaying peptides longer than five or six residues. In contrast, proteins as large as antibody Fab fragments (Kang et al. 1991) have been displayed with type 8+8 and 88 display (Smith 1993). (a) Type 3 display. Both low infectivity and proteolysis have been described in several multivalent p i l l systems. For example, in otherwise successful display of catalytically active E. coli alkaline phosphatase (AP), infectivity of the AP-pIII recombinants was only 0.3-0.4% (one-fifteenth of that of the parent vector) and proteolysis of AP-pIII fusion proteins was considerable (McCafferty et al. 1991). In early preparations, only 5-10%) of the fusions were intact; in later preparations, 30-60%). Similar degradation was seen in virions displaying the 32 kDa B chain of ricin (a lectin from castor beans): only one copy of intact B chain-pill fusion protein was present for every three copies of proteolyzed product (Swimmer et al. 1992). Proteolysis was also suspected in phage displaying the 58-residue bovine pancreatic trypsin inhibitor (Roberts et al. 1992b). In contrast, no significant degradation products were seen in Type 3 display of correctly folded and fully active P-lactamase (Soumillion et al. 1994b). Human interleukin 3 (hIL3) has also been displayed in a Type 3 vector (Gram et al. 1993), but it is unclear what fraction of virions displayed intact hIL3. (b) Type 3+3 display. Among early phage display papers was that describing Type 3+3 display of the 191-residue disulfide-containing human growth hormone (hGH), accomplished by fusion of hGH to a truncated p i l l , the C-terminal domain (pIIIc) required for virion morphogenesis (Bass et al. 1990). Importantly, phagemid-displayed hGH-pIII c could bind to hGH receptor and was recognized by mAbs whose epitopes are sensitive to hGH conformation. 27 Although phagemid titers were sensitive to the ratio of recombinant versus wild-type p i l l molecules incorporated into each particle — indicating an undefined but possibly translocation-related effect of hGH-pIII c on phage or host biology - this ratio could be controlled so that, for example, only one in ten particles contained a single hGH-pIII c molecule. Immunoblots showed little evidence of hGH-pIII c degradation. Other molecules displayed as Type 3+3 p i l l ( fusions include (i) the a subunit of the high-affinity receptor for IgE (FceRI) in a form recognized by IgE (Scarselli et al. 1993) and (ii) the 23 kDa human ciliary neurotrophic factor (CTNF) in a form that bound to its natural receptor (Saggio et al. 1995). Degradation of the pIII c fusions was not examined in these latter papers. Other methods of Type 3+3 display involve fusions to the complete p i l l molecule. One such study (Eerola et al. 1994) involved display of the 33 kDa prostate specific antigen (PSA; a serine protease) in a form capable of being recognized by antibodies recognizing non-linear epitopes. Nevertheless, Western blotting showed degradation of many of the PSA-pIII fusion products. Staphylococcal protein A (SPA) has been displayed in this manner (Kushwaha et al. 1994) but it is unclear what fraction of phagemid particles displayed intact SPA. Both Type 3+3 and Type 3 display have been employed to display the first 176 residues of human CD4 ( C D 4 M 7 6 ) in a form capable of capture by gpl20 of the human immunodeficiency virus (Abrol et al. 1994). Infectivity of the C D 4 M 7 6 Type 3 recombinants was only ~5% of that of the parent vector. (c) Types 33 and 88 display. In a study (Corey et al. 1993) exploring both Type 33 and Type 88 display, sequences encoding a signal peptide and trypsin were fused to those encoding pIII c or mature pVIII and cloned into M13mpl8, resulting in production of virions with hybrid coats comprised of trypsin-pIIIc or trypsin-pVIII as well as wild-type p i l l and pVIII. Importantly, the displayed trypsin fusion proteins possessed near wild-type enzyme activity. 28 However, <20% of virions displayed fusion proteins, and mean infectivity (plaque-forming units per virion, independent of whether a virion displayed a recombinant protein) was less than one-quarter of that of the non-recombinant vector. These are suggestive of profound effects of the fusions on host metabolism. The earlier and careful construction by Markland et al. (1991) of a Type 88 vector and its employment to display bovine pancreatic trypsin inhibitor (BPTI) in a functional form illustrates several aspects of vector design and host cell biology. In a manner similar to that described for trypsin display, Markland et al. inserted sequences encoding BPTI-pVIII fusions into M13mpl8. To avoid homologous recombination involving identical gVIII coding sequences, and in contrast to exclusive use of wild-type gVIII sequences employed by Corey et al. for trypsin display, Markland et al. used alternative codons to construct a synthetic gene VIII for the BPTI-pVIII fusions. Their first constructs, encoding a fusion of the pVIII signal peptide, BPTI and mature pVIII, failed to produce virions displaying BPTI-pVIII fusion products. Rather, Western blots of host cell lysates showed only a single protein species with a size characteristic of unprocessed fusion protein. Replacement of the natural pVIII signal peptide with an alternative signal sequence resulted in partial processing of the fusion protein, but reasonable levels of (and greatly enhanced) processing and incorporation of BPTI-pVIII into virions could only be achieved by use of an E. coli prlA (secY) host. The 1:100 to 1:50 ratio thus achieved of BPTI-pVIII to wild-type pVIII incorporated into virions was increased to 1:30 by changing the start codon of the wild-type gene VIII from A T G (Met) to CTG (Leu). Notably, this ATG—•CTG mutation also resulted in a 10-fold reduction in virion yields. (d) Type 8+8 display. An innovative example of Type 8+8 display of an antibody Fab fragment is provided in a study by Kang et al. (1991). Having employed a phagemid to coexpress (i) heavy chains (as V H - C H 1 ) as fusions between a PelB signal peptide and pVIII, 29 and (ii) light chains (V L -C L ^ as fusions to a second PelB signal peptide, they expected the V H -C H 1-pVIII and V L - C L molecules to associate in the periplasm and form a properly folded Fab fragment. Careful studies, including electron microscopy of phagemids decorated with antigen-Fab complexes, validated their expectations. Apparent adaptations of this approach, also relying on periplasmic association of components that become incorporated into virions, are reviewed in later sections. 1.5.3. "Directed evolution of a protein" A number of phage applications have derived from a novel approach, described as "directed evolution of a protein", in which Roberts et al. (1992a) constructed a PDL of 1000 protease inhibitor variants derived from the sequence of wild-type bovine pancreatic trypsin inhibitor (BPTI) by limited randomization of five residue positions believed important for interaction between BPTI and human neutrophil elastase (FINE). Subsequent affinity selection with HNE-agarose yielded a variant with >106-fold greater affinity for FINE than the parental sequence. Similar studies, often with multiple libraries of greater diversity, have yielded other peptides or proteins with altered target specificity including (i) other protease inhibitors with increased affinity for their target (Dennis & Lazarus 1994a, 1994b; Wang et al. 1995a), (ii) an enzyme with altered specificity for active-site ligands (Widersten & Mannervik 1995) and (iii) zinc fingers with modified specificity in one or more fingers (reviewed in Choo & K l u g 1995). The popularity of PDLs as tools for in vitro selection of peptides and proteins with altered properties is reflected in the subheading "phage fever" employed by Clackson and Wells (1994) in their review of the field. 1.5.4. Gene fragment libraries Constructed by Smith (1985), the first PDL was a gene fragment library displaying 30 fragments of EcoRl endonuclease and methylase enzymes in the middle of p i l l of wild-type phage f l . Notwithstanding the publication by Smith (1992), in his widely-distributed handbook, of protocols oriented to creating such libraries, they have found surprisingly little employment. Indeed, they appear to be limited to the following few examples. (a) Multivalent pill-display gene fragment libraries screened with antibodies. In an early demonstration of the utility of subgenomic libraries, Bleul et al. (1991) cloned 200 bp DNase I-generated fragments of subgenomic D N A of the human papilloma virus type 18 (HPV-18) into the Type 3 frameshift vector fUSEl (described earlier; see also Parmley & Smith [1988]), creating a library of ~105 transformants of which <2% (so it appears from their data) possessed inserts capable of directing display of a HPV-18-related peptide. Screening by plaque lifts (rather than affinity selection) with sera against HPV-18 E6 and E7 fusion proteins identified two unique 30- to 50-residue antibody-reactive peptides for each serum. Similarly, Wang et al. (1995b) cloned 100-200 bp fragments of the gene encoding the bluetongue virus capsid protein VP5 into a conventional (non-frameshift) Type 3 vector, fUSE2 (described earlier; see also Parmley & Smith [1988]), creating a library of 5,500 transformants, of which only 200 were believed to display a VP5-related peptide. Biopanning with a VPS-specific mAb led to the identification of two overlapping (44- and 50-residue) MAb-reactive peptides. Petersen et al. (1995) used a related approach to construct Type 3 libraries (in fUSE5, Scott & Smith 1990) of 50-200 bp gene fragments encoding peptides for (i) Drosophila R N A polymerase II, (ii) human 53 protein and (iii) human cytokeratin 19 protein. One round of biopanning followed by colony blot screening led to identification of epitopes recognized by mAbs raised against these proteins. Three cytokeratin 19 libraries were constructed with different peptide linkers flanking the cloned insert. Interestingly, only one of these libraries 31 yielded MAb-reactive clones; this library used Cys-containing linkers to present peptides in a disulfide-constrained loop. (b) Identification of ligand-binding domains with a phagemid system. Type 3+3 display has also been employed in gene fragment libraries. In a model system exploring the feasibility of constructing prokaryotic genomic libraries, Jacobsson and Frykberg (1995) cloned 100-700 bp sonication products of Staphyloccocus aureus genomic D N A into a phagemid vector. Biopanning the resulting 107-member library with human IgG and fibronectin and subsequent screening by (i) colony blotting with these same molecules, and (ii) hybridization with oligonucleotide probes corresponding to fibronectin binding domains, led to the identification of sequences corresponding to known binding regions of staphylococcal protein A and fibronectin binding proteins. Oddly, the cloned inserts of all five fibronectin-binding clones contained +1 or -1 frameshifts, and 47 of 50 clones derived from other libraries possessed similar frameshifts. (c) cDNA libraries by means of Jun and Fos leucine zippers. An innovation system developed by Crameri and Suter (1993) for phage display of cDNA fragments exploits the interaction of the Jun and Fos leucine zippers, in a manner conceptually similar to the approach adopted by Kang et al. (1991) for display of Fab fragments by means of intraperiplasmic association of antibody heavy and light chains. In the Crameri and Suter phagemid system, Jun and Fos are expressed from pelB-jun-gIIIc and pelB-fos-cDNA fusions, respectively, with the result that intraperiplasmic, high-affinity interactions between Fos and Jun allow display of cDNA-derived peptides on the phagemid surface. To prevent exchange of fos-cDNA fusion products among phagemid particles, cysteines were added to the Jun and Fos N - and C-termini to allow intermolecular disulfide bond formation. Notably, Jun-Fos display involves C-terminal fusions (to Fos) rather than insertions (near the N-terminus of pill) , thus allowing expression 32 of cDNA or other gene fragments containing translation stop sites or other non-coding sequences. In a followup study, Crameri et al. (1994) confirmed the utility of Jun-Fos display by screening an Aspergillus fumigatus cDNA library with human serum IgE, thereby identifying sequences of putative allergens. 1.5.5. Innovations in peptide presentation Several approaches have been adopted to the construction of PDLs in which peptides are displayed in a constrained, often defined, conformation. This has proved advantageous in some cases but not in others. (a) Disulfide-constrained peptides. Because, prior to their incorporation into virions, p i l l and pVIII are inserted into the inner membrane with their N-termini within the periplasm, paired cysteine residues flanking randomized sequences (e.g., * C X 6 C * , where * denotes a potential disulfide bond and X is any amino acid) can be used to present peptides in a disulfide-bonded (Luzzago et al. 1993) loop. Cys-constrained RPLs have been employed with remarkable success. In a preface to two papers (co-published in Science: Livnah et al. 1996; Wrighton et al. 1996) describing a peptide derived from such libraries, Wells (1996) wrote that the articles "are enough to reinstate one's belief in Santa Claus", for the authors had found a 20-residue disulfide-constrained peptide capable of forming dimers and subsequently dimerizing the erythropoietin receptor, leading to its activation. The isolation of brain- and kidney-targeting phage after intravenous injection of Cys-constrained RPLs into mice (Pasqualini & Ruoslahti 1996) illustrates another novel application of disulfide-constrained libraries. Other examples illustrate the relative merits of disulfide-constrained and unconstrained libraries. McLafferty et al. (1993) biopanned a X * C X 4 C * X RPL with an anti-p-endorphin mAb and found that the disulfide bond was required for high-affinity binding of the MAb-reactive peptides (*CYG / A FC*) . In contrast, Felici et al. (1993) biopanned both constrained 33 (*CX 9 C*) and unconstrained ( X 9 ) PvPLs and, having identified three MAb-reactive peptides from each library, found that the unconstrained peptides were better recognized in immunoblots than those from the constrained library. In reviewing constrained display, Ladner (1995) has suggested that increasing the constraint imposed on a peptide segment (e.g., RGD in xxxRGDxxx, *CxxRGDxxC* and *Cx*CRGDC*xC*) decreases the likelihood that it will bind to any particular target, but that if a target is found the peptide will bind more tightly and specifically, and more can be learned about the nature of the binding. (b) Presentation on carrier molecules. Tendamistat, a 74 amino acid molecule comprised of two three-strand P-sheets, has been employed as a scaffold in which two adjacent strand-connecting loops (13 and 6 residues) can be randomized (McConnell & Hoess 1995) in Type 33 display. A perceived advantage of constrained display involving two (versus one) randomized loops is that it may better approximate protein surface structures involved in inter-molecular recognition, such as between discontinuous epitopes and antibody. The 61-residue "minibody" (Pessi et al. 1993) is similar. Engineered from an antibody V H domain and comprised of two three-strand P-sheets, the minibody retains the regions corresponding to two adjacent hypervariable loops (HI and H2) of the parental antibody. Its utility was demonstrated (Martin et al. 1994) by the identification, from a Type 3 library displaying minibodies in which one or both of these loops had been randomized, of an H2 variant that bound human interleukin-6 and inhibited its biological activity. A more potent variant was identified in a study employing Type 3+3 display of minibodies in which the HI loop of the previously identified H2 variant was randomized (Martin et al. 1996). Phage display of randomized a-helices has been explored by Bianchi et al. (1995), who randomized five helix positions of a Cys 2 His 2 zinc finger that are most exposed in the helix and which cluster on one side. Since zinc coordination and folding are coupled in zinc fingers, 34 zinc-dependent binding (between a randomized finger and its cognate molecule) was used as a built-in control against structurally undefined sequences. Phage display involving randomizing solvent-accessible residues of a surface comprised of two of the three a-helices of a monovalent Fc-binding domain of staphylococcal protein A has also been explored (Nord et al. 1995). 1.5.6. Innovations in identification of target clones Two innovative methods of identifying target clones exploit the concept employed by Kang et al. (1991) for Fab display, and by Crameri and Suter (1993) for Jun-Fos-mediated display of cDNA fragments, viz., the periplasmic association of separately exported molecules to form a complex that becomes incorporated into a virion. The innovation lies in the physical separation of the two functional domains of p i l l , the N-terminal (pIIIN) domain which plays a role in F-pilus adsorption, and the C-terminal(pIIIc) domain which functions in virion morphogenesis. In the "direct interaction rescue" system of Gramatikoff et al. (1994), an invariant "bait" molecule such as c-Jun (Gramatikoff et al. 1995) is cloned as an N-terminal fusion to pIII c while variant fragments such as those from a human cDNA library are cloned as C-terminal fusions to pIIIN. In cells harboring a cDNA fragment encoding a peptide that interacts with the Jun bait, periplasmic association of Jun-pIIIc and pIII N-cDNA fusion products yields infectious virions constructed with the heterodimeric complexes; in other cells, non-infectious virions containing only Jun-pIIIc are produced. The similar system of Krebber et al. (1995) employs single-chain antibodies (scFv) fused to pIII c, and infectivity is restored by cognate pairs of pIIIN-antigen and scFv-pIIIc. Notably, antigen or antibody sequences can be randomized or otherwise varied. The system is similar to that developed by Duenas and Borrebaeck (1994) to mimic clonal selection by the immune system. 35 1.6: T H E P R E S E N T S T U D Y 1.6.1. Project goals: "Issues and Applications" This thesis project was initiated with the goal of exploring the use of phage display in the study of pathogenesis, with a view in mind of demonstrating the feasibility of constructing genomic PDLs that could be used to identify pathogen-encoded proteins with roles in pathogenesis. The systematic working through of two model systems was intended not only to illustrate the application potential of phage display, but also to identify limitations and to explore ways around these limitations. The simpler of these models involved constructing a 32-member library displaying variants of the immunodominant tetrapeptide repeats of the Plasmodium falciparum (malaria parasite) circumsporozoite protein (CSP) and using this library to, at a minimum, characterize the epitope specificities of a panel of cc-CSP mAbs. Given that, as reviewed in Chapter 4, something was already known about the structure of CSP and the epitope specificities of the a-CSP mAbs, it seemed possible to both (i) confidently explore basic methodology with a model expected to yield a narrow range of results, and (ii) provide new insights into an important malaria parasite protein. The second, more complex model involved constructing a series of PDLs displaying random peptide fragments (size-fractioned by library) of filamentous hemagglutinin (FHA), a multifaceted adhesin of the respiratory pathogen Bordetella pertussis and an important component of acellular pertussis vaccines (reviewed in Chapter 6). These libraries were to be used for two purposes. First, biopanning with rabbit oc-FHA polyclonal antibodies and subsequent immunocharacterization of affinity selected clones was expected to provide insight into the antigenic makeup of FHA. Second, characterization of randomly selected library 36 members was expected to reveal something about the nature of peptides that could and could not be successfully displayed on phage. The large size of FHA (220 kDa in its mature, processed form) and its encoding gene (fhaB, 10 kbp) ensured that yfazfi-derived PDLs would serve as a non-trivial models of genomic libraries. 1.6.2. Project accomplishments in context of the concurrent work of others Concurrent with my work towards illustrating the potential of phage display and examining its limitations, it has become clear ~ as the technology pioneered by Smith has been extended by others — that phage display can be used to answer a variety of biological questions. Also evident, however, is that phage and host biology impose greater limitations than were suggested by the successes of the early RPL studies and those that shortly followed. Notably, few studies have specifically addressed phage- or host-imposed limitations. Equally noteworthy, and although phage display was pioneered with a gene fragment library, there have been few reports of gene fragment or genomic PDLs. Possibly, this reflects research objectives or a perceived lack of utility. Alternatively, it may reflect difficulties inherent in employing these libraries. Indeed, the heterologous nature (peptide fragment size and sequence diversity) of peptides in genomic PDLs sets these libraries apart from nearly all other PDLs, which characteristically display (i) peptides of great diversity but of small fixed size, as in RPLs, or (ii) large peptides or proteins in which only a small number of residues are varied within an otherwise uniform scaffold, as in affinity maturation studies or antibody display. Given this distinction, construction and employment of a genomic PDL might be seen as a kind of "extreme test" of phage display. In this context, my study of phage display of FHA-derived peptides has been an "extreme test", particularly so because it involved the exclusive use of multivalent (Type 3) vectors and, as reviewed earlier, monovalent vectors are believed better suited to the display 37 of larger peptides and proteins. The vectors employed (the fUSEn derivatives of fd-tet) were chosen for several reasons (reviewed in Chapter 7), including that their design incorporated features deemed important or essential to the success of several components of my project. Although unexpected problems with these vectors made it difficult to meet some project objectives, in particular the characterization of randomly sampled F H A library clones, the problems themselves provided surprising insight into certain aspects of the impact of host cell biology on phage display. 1.6.3. Thesis layout The layout of this thesis reflects the. dual themes of my project, "issues" and "applications". Applications are described first. Chapter 3 does not relate directly to phage display, but rather provides a statistical analysis (based on the Brookhaven Protein Data Bank) of structures formed by Asx-Pro dipeptides, and develops a model for folding of the Asn-Pro-Asn-Ala sequences that comprise the P. falciparum CSP tetrapeptide repeats. This model is further developed by the experimental data of Chapter 4, derived from characterization of 32 phage-displayed variants of these repeats with a panel of a-CSP mAbs. Experiments summarized in Chapter 5, derived from biopanning RPLs with two a-CSP mAbs, serve to illustrate an inherent limitation of RPLs. Chapter 6 provides an antigenic analysis of FHA derived from biopanning a series of FHA-derived PDLs with rabbit a-FHA polyclonal antibodies. Chapter 7, in describing various difficulties encountered during the work that led up to the successful applications of Chapters 4 and 6, deals with issues of vector stability and the construction of libraries. Finally, Chapter 8 reviews my overall findings in context of their significance to phage display technology. 38 Chapter 2 Materials and Methods 2.1. B A C T E R I A L STRAINS, PLASIVODS AND B A C T E R I O P H A G E Escherichia coli strains are described in Table 2-1; growth characteristics of selected strains are shown in Figure 2-1. Plasmids are outlined in Table 2-II, phage display vectors in Table 2-111. Wild-type bacteriophage f l (Zinder et al. 1963) and a random hexapeptide phage-display library, derived from the library described in Scott and Smith (1990) by amplification in E. coli K91, were gifts from G. P. Smith, University of Missouri. A second phage-display library of random 15-residue peptides was a gift from J. Scott, Simon Fraser University. 2.1.1. Culture of bacteria E. coli strains were routinely cultured in L B broth or other media such as 2xYT (Sambrook et al. 1989) at 37°C with shaking at 125-200 rpm, or on plates (commonly L B , or other rich medium) containing 1.5% agar. For broth cultures, commonly, >1 mL of a 4 h to overnight "starter" culture was used as inoculum for 30-50 mL (125 mL Erlenmeyer flask) to 600 mL (2 L flask) 10-24 h cultures; these are referred to as "serial overnight culture in mL)" where n = final culture volume. Strains harboring fUSEn or fDRWfl vectors or their recombinants were cultured in media containing 20 (ig tetracycline mL" 1 (LB-Tet). As appropriate, E. coli K91-Kan was cultured in media containing 100 jig kanamycin mL" 1; strain MCI061 in media containing 25-100 |_ig streptomycin mL" 1; strains harboring pACYC184 in media containing 25 [ig tetracycline mL"1 and 35 u.g chloramphenicol mL" 1; strains harboring pASlOO, pJB61, pNK1759 or pNK2859 in media containing 50 or 100 [ig ampicillin mL" 1. 39 Table 2-1. E. coli strains used in these studies. Strain F-piliation Amber suppression Principal use Source Reference K 3 7 a HfrC supD titration of amber 0 vector virions A T C C d A T C C 33626 K37 / p A C Y C 1 8 4 HfrC supD titration of amber vector virions, particularly of virions in undiluted culture supernatant containing tetracycline this study K802 F-minus supE CaCl 2-transformation host, vector and virion propagation G . P. Smith, U . of Missouri Smith & Scott 1993 K802 recA F-minus supE single experiment (Figure 7-16) A T C C A T C C 47026 K 9 1 a HfrC none virion propagation and titration G . P. Smith Smith & Scott 1993 K 9 1 / pNK1759 HfrC none single experiment (Figure 7-16) this study K 9 1 / pNK2859 HfrC supF titration of amber vector virions this study K91/pJB61 HfrC none single experiment (Figure 7-16) this study K 9 1 - K a n b HfrC none virion propagation and titration G . P. Smith Smith & Scott 1993 K91-Kan / p A C Y C 1 8 4 HfrC none virion titration, particularly for undiluted culture supernatant containing tetracycline this study LE392 F-minus supE, supE CaCl 2-transformation host, vector propagation A T C C A T C C 33572 MC1061 F-minus none electroporation host, virion propagation G . P. Smith Smith & Scott 1993 aE. coli K37 and K91 are derivatives of K38 (Smith 1988). b derived from E. coli K91 by G . P. Smith (Smith & Scott 1993); carries kanamycin resistance within a mini-transposon inserted into lacZ. c Amber vectors are described in §7.3.2.b. See also Table 2-III. d A T C C , American Type Culture Collection. 40 E c o 0.1-0.02-K802 I 1 1 1 1 I " 1 K802 recA I 1 1 •11 1 1 1 11 1 1 1 1 i 1 T d = 40 min. 0.1-0.02- T 1 T"1 T 1 K91-Kan 50 100 150 200 250 0 200 250 50 100 150 200 2500 Time (minutes) Fig. 2-1. Growth curves and estimated doubling times (T d) of E. coli strains. Duplicate 0.5 m L samples from an overnight culture (2 m L L B ; 3 7 ° C with shaking) of each strain were diluted 1:100 into fresh medium (50 m L L B pre-warmed to 37°C) ; these cultures were incubated with shaking at 3 7 ° C , and optical densities at 600 nm ( O D 6 0 0 ) were measured at -20 min. intervals using 1 mL samples (discarded) in 1 cm disposable cuvettes. Doubling time (T d ) for each duplicate was calculated on the basis of the minimum slope - of every set of 3, 4 and 5 consecutive data points — of the least squares linear regression of l o g ( O D 6 0 0 ) versus time. Reported T d are arithmetic means for the duplicate cultures. Table 2-II. Plasmids. Plasmid Description Source Reference p A C Y C 1 8 4 carries chloramphenicol resistance gene from transposon Tn9 and tetracycline-resistance gene f r o m p S C l O l New England Biolabs Chang & Cohen 1978 pASlOO p T Z l 8R carrying (i) the 10 kilobase pair B. pertussis fhaB EcoRI fragment sub-cloned (A. Siebers) from clone CI-5 of a Sau3k\ library of B. pertussis 18323 chromosomal D N A (M.J . Brennan, F D A , Bethesda, M D ) ; this fragment is described in Delisse-Gathoye et al. (1990); (ii) ampicillin resistance A . Siebers A . S.and B. Finlay, unpublished pJB61 pBR322 carrying (i) F f gene III, (ii) ampicillin resistance A T C C A T C C 39162 pNK1759 pBR322 carrying (i) kanamycin resistance within a mini-TnlO transposon, (ii) transposase, (iii) ampicillin resistance; designed for transposon mutagenesis A T C C A T C C 77352 pNK2859 pBR322 carrying (i) supF within a mini-TnlO transposon, (ii) transposase; (iii) ampicillin resistance; designed for transposon mutagenesis A T C C A T C C 77338 41 Table 2-III. Phage display vectors. Vector Description 3 Source Derived from Reference f D R W 2 0 fDRW21 f D R W 2 2 conventional1 3 vector this study f U S E l Figure 7-12, §2.8.1 , §2.8.2 f D R W 5 amber 0 vector fUSE5 Figure 7-9, §2.8.3 fDRW613 amber vector f D R W 5 Figures 2-2 and 7-14, §2.8.4, §2.8.5 f D R W 6 1 3 C two-amber d vector f D R W 7 0 amber vector Figure 7-23, §2.8.6 f D R W 8 w i series conventional vectors Figure 7-29, §2.8.7 f U S E l frameshift6 vector G . P. Smith, U . of Missouri fd-tet Figure 7-1; Smith 1985; Parmley & Smith 1988; Scott & Smith 1990; Smith 1992; Smith & Scott 1993 fUSE2 conventional vector fUSE3 frameshift vector fUSE5 frameshift vector a Each of these vectors, as a derivative of fd-tet (Zacher et al. 1980), carries a tetracycline-resistance gene. b Conventional , °amber, drwo-amber and eframeshift vectors are explained in text and figures describing these vectors (see column "Reference"). See also Chapter 7. 2.2. G E N E R A L M E T H O D S 2.2.1. Common reagents Common reagents include: PBS, phosphate buffered saline, 12 m M phosphate, 157 m M Na + , 4.4 mM K + , 140 mM CI", pH 7.4; phenol, reagent-grade phenol equilibriated with 0.1 M Tris pH 8.0, or a similar commercial preparation; TBS, 50 mM Tris-Cl pH 7.5, 150 m M NaCl; T E , 10 m M Tris-Cl, 1 mM ethylene diamine tetraacetate (EDTA), pH 7.4 or 8.0. 2.2.2. D N A manipulation (a) Restriction digests. Digestions of D N A with restriction endonucleases (Boehringer, New England Biolabs, Pharmacia, Promega or Stratagene) were normally carried out in manufacturer-supplied buffer at recommended temperatures in 10-20 |J,L volumes. In most cases for incubation temperatures >50°C, reaction mixtures were overlaid with a small volume of sterile mineral oil. Commonly, <1 \xg D N A possessing one or a few restriction sites per 42 molecule was digested ~1 h with a large excess (4-10 U) of restriction endonuclease. (b) Ligations. Ligations with T4 ligase (GIBC07BRL, NEB, Promega) were carried out in manufacturer-supplied buffer in a total volume of 20 uL. Ligase units (U) reported are Weiss units. (c) Phenol and chloroform extraction of DNA from reaction mixtures. Most commonly, an equal volume of phenol was added to a DNA-containing sample (volume commonly adjusted to 100-450 uL with TE pH 7.4 or 8.0); these were mixed by repeated inversion or gentle vortexing. After brief centrifugation, the aqueous phase was similarly extracted with chloroform. Alternatively, a D N A sample was similarly treated with a mixture of equal parts of phenol and chloroform or a 25:24:1 mixture of phenol/chloroform/isoamyl alcohol. (d) Precipitation with ethanol. D N A samples, most commonly in volumes <450 uL, were precipitated by addition of 2.2 volumes reagent grade ethanol and one-tenth volume 3M sodium acetate pH 5.2-6. After overnight incubation at 4°C (most commonly) or -20°C, or at least 1 h on ice (uncommonly), samples were centrifuged at maximum speed in an Eppendorf 5415C microfuge >0.5 h at 4°C. After washing in 70% ethanol and centrifuging at maximum speed >3 min. at room temperature or at 4°C, pellets were dried at room temperature (in some cases under low vacuum) before addition of a small volume (10-20 \xL) of 10 m M Tris-Cl pH 8, TE pH 7.4 or 8, other slightly alkaline buffer (as employed for restriction digests) or water. Samples were stored overnight at 4°C to allow rehydration of DNA. (e) Precipitation with isopropanol. D N A samples such as of amber vectors digested with restriction endonucleases to excise "stuffer" fragments (for examples, see Figures 2-2, 7-9 and 7-23) were precipitated with isopropanol to eliminate these small fragments, which tend to remain in solution. After adjusting the sample volume to (typically) 200-450 uL with TE pH 7.4 or 8.0, two-thirds volume of isopropanol and one-ninth volume of 3M sodium acetate 43 pH 6 were added. D N A was precipitated, washed and rehydrated as described for ethanol precipitation. 2.2.3. Agarose gel electrophoresis D N A samples were routinely analysed by agarose gel electrophoresis at 2-7 V cm"1 in 5 cm x 8 cm or 11 cm x 14 cm 0.7%-0.75% agarose gels containing 0.5 [ig ethidium bromide (EtBr) mL" 1, using T A E running buffer (Sambrook et al. 1989) containing 0.5 fig EtBr mL" 1. 2.2.4. DNA quantification (a) By UV spectroscopy. Estimates assumed that an absorbance of 1.0 at 260 nm ( A 2 6 0 ) corresponded to 50 [ig mL"1 d.s. or 33 (ig mL"1 s.s. DNA. Relative purity of a preparation was assessed by the ratio of absorbances at 260 ( A 2 6 0 ) and 280 nm. A 2 6 o / A 2 8 o ratios for d.s. and s.s. D N A preparations were commonly ~1.8 and -1.7 respectively. (b) By agarose gel electrophoresis. Samples estimated to contain <200 ng of linear D N A fragments, together with commercially purchased standards of known quantity (e. g., Hindlll- or ZtoEII-digested bacteriophage X DNA) were analysed by agarose gel electrophoresis and the quantities of D N A determined by comparing the intensities of fluorescence of EtBr-stained samples with those of the standards. (c) By an agarose gel "spot" assay. Samples (-5 uL) of dilutions of D N A to be quantified, together with commercially purchased standards of known quantity, were applied to solidified 0.7% agarose (with 0.5 [ig EtBr mL" 1) and allowed to stand >0.5 h. Relative quantities of D N A were determined by intensities of fluorescence of samples and standards under U V light. 2.2.5. Oligonucleotides (a) Synthesis and purification. Most synthetic oligonucleotides were prepared by the Nucleic Acid Protein Synthesis service (NAPS) unit of the University of British Columbia (UBC) Biotechnology Laboratory; in one case, by the R.E.W. Hancock laboratory at UBC. 44 Following synthesis, oligonucleotides were purified by the method of Sawadogo (1991). (b) Annealing. Roughly equimolar mixtures of pairs or other combinations of synthetic oligonucleotides to be annealed (per sample: -25-150 uM in 50-100 uL 200 m M Tris pH 7.5, 20 mM MgCl 2 , 500 mM NaCl) were held in a TempBlock 80°-85°C for 1.5-15 min. Samples were cooled (over a -1-1.5 h period) by removing the TempBlock from the heat source and placing it on the bench until the temperature reached ~30°C. Samples were stored at -20°C or 4°C. 2.3. D N A E X T R A C T I O N F R O M VIRION H O S T S A N D VIRIONS 2.3.1. R F D N A extraction from host cells Except as noted elsewhere, RE D N A of phage vectors or recombinants was extracted from 30-37.5 mL serial overnight culture ( O D 6 0 0 commonly ~ 0.9 to 1.3) using a Qiagen Tip 20 plasmid D N A extraction/purification kit, following the manufacturer's recommended protocol for extraction of plasmid DNA, except as follows. First, cell pellets in the more recent preparations were washed in STE (100 mM Tris, 0.2 m M EDTA, 200 m M NaCl, pH 8.0) as recommended (Qiagen 1993) for M13 RF DNA. Second, volumes of reagents PI , P2 and P3 (similar to those used in common alkaline lysis protocols) were scaled up (from the recommended 0.3 mL to 0.9-4 mL, with 1.8 mL commonly used in recent preparations) to compensate for increased culture volume (30-50 mL versus the recommended 3-6 mL). 2.3.2. S.s. D N A extraction from virions (a) Phenol extraction offUSEn s.s. DNA. For each of fUSEl , fUSE3 and fUSE5, s.s. D N A (packaged into virions by G. P. Smith by means of a pill-encoding helper plasmid) was extracted from Smith-supplied virions by phenol and chloroform extraction as described by Smith (1992). (b) Preparation of s.s. sequencing template. S.s. D N A was extracted from two-stage (most commonly; §2.6.4) or one-stage (§2.6.3) PEG-precipitated virions derived from at least 45 15 mL (commonly 30-37 mL) serial overnight culture (§2.1.1) using a Qiagen Tip 20 plasmid D N A extraction/purification kit following the manufacturer's recommended procedures (Qiagen 1991) for extracting s.s. DNA from M13 virions except as follows. In most cases, particularly in more recent preparations, reagents L3, L4 and L5 were scaled up from 1.0 mL to 1.8 mL. Also, Qiagen reagent L5 was occasionally replaced with the similar reagent P3 (intended for RF D N A preparations) as convenient. Finally, in many early preparations, s.s. D N A recovered from Qiagen columns was additionally phenol/chloroform extracted and ethanol precipitated; such preparations generally yielded superior sequencing chromatograms. 2.4. S E Q U E N C I N G 2.4.1. Sequencing primers Primers for sequencing d.s. RF or s.s. D N A included the synthetic oligonucleotide 5' -CCCTCATAGTTAGCGTAACG-3 ' (standard " A " primer) and the alternative (and generally less satisfactory) 5 ' -TGAATTTTCTGTATGAGG-3 ' (alternative " B " primer). These primers, designed for sequencing s.s. DNA of fUSEn recombinants (Smith 1992), are complementary to wild-type g i l l sequences (of the virion-encapsidated or "sense" strand) and prime from a position 77-86 bases (primer "A") or 18-32 bases (primer "B") downstream (i.e., in the 3' direction of the "sense" strand) of fUSEn cloning sites. Sequencing of one set of fDRW70 recombinants possessing relatively large B. pertussis yMB-derived inserts (siblings of clone ID I-a; Figure 6-1) also employed the primers 5 ' -GCTGACCGCCTCTCCACC-3 ' and 5 ' - C C T G C G G C A A C C A C G G T C - 3 ' ; these are complementary to sequences within the fhaB-derived insert of clone I-a. 2.4.2. Sequencing reactions and gels Sequencing was performed using Applied Biosystems D N A sequencers, Applied Biosystems Taq and Taq F S polymerase and DyeDeoxy Terminator cycle sequencing reagents generally in accordance with the manufacturer's recommended protocols, except as noted 46 below. Sequencing performed with Applied Biosystems Taq (but not Taq F S) polymerase employed, for s.s. template, roughly 5-fold more (or, in one particularly successful set of sequences, 2,000-fold more) sequencing primer than recommended. Most sequencing reactions were carried out with the recommended 25 thermocycles; in some cases, this was reduced to 20 cycles. Unincorporated nucleotides were removed using Select-D G-50 (5 Prime -> 3 Prime) or Centri-Sep (Princeton Separations) spin columns. Sequencing reactions (with Applied Biosystems Taq polymerase) of CSP-library clones, most fDRWn vectors, some fDRW/z pseudorevertants5, and many B. pertussis fhaB restriction fragment library clones were analysed using an Applied Biosystems model 3 70A sequencer. Other sequencing reactions (in particular those of B. pertussis fhaB DNase I fragment library clones; carried out with Taq or T a q F S polymerase) were analysed by the Nucleic Acid Protein Synthesis service unit of the U B C Biotechnology Laboratory using Applied Biosystems sequencers. Where sequence ambiguities existed, these were resolved by re-sequencing or by examination of sequencing chromatograms in context of documentation (Applied Biosystems Inc.) of anomalous sequence-dependant incorporation of nucleotides. 2.5. H O S T C E L L T R A N S F O R M A T I O N AND INFECTION 2.5.1. Calcium chloride transformation (a) CaCl2-competent host cells. E. coli K37, K802, K91 and LE392 were cultured in L B to an O D 6 0 0 of -0.35-0.55; the more slowly growing (Figure 2-1) K802 recA, to an O D 6 0 0 of 0.2-0.5. After chilling on ice, 20 mL culture volumes were centrifuged 10-15 min. at -500 x g m a x (4°C) and cell pellets were gently resuspended in 0.5 to 2 volumes cold 0.1 M C a C l 2 After incubating >20 min. on ice and centrifuging as before, the resulting cell pellets were gently resuspended in <1 mL cold 0.1 M CaCl 2 and stored (4°C) up to five days before use. 5Pseudorevertants are described later in this chapter, and discussed in Chapter 7. 47 (b) Transformation. Small samples (commonly unqualified, in the ng to |ig range) of s.s. or RF DNA were diluted to 200 uL with TE pH 8 and chilled in glass culture tubes before combining with 200 \iL cold CaCl2-competent E. coli host cells. After incubating 30 s at 37°C, >30 min. on ice, 1.5-2 min. at 42°C, >1 min. on ice or at room temperature, 1.6 mL SOC (Sambrook et al. 1989) or L B (containing 0.2 [ig tetracycline6 mL"1 for fUSEn or fDRWn and their recombinants) was added and the cells incubated 0.5-2 h (commonly 1 h) with shaking (<225 rpm) at 37°C before spreading (commonly as 1, 10 and 100 | iL aliquots) on selective medium (LB-Tet plates for fUSEn or fDRW/z or their recombinants) and incubating overnight at 37°C. In some early transformations, medium containing 0.75% agar was overlaid on spread plates (Smith 1992). 2.5.2. Electroporation (a) Electrocompetent host cells. Several related protocols were employed for preparing electrocompetent E. coli MCI061. In each case, large volume cultures (typically 500 mL L B [= 1 volume] in a 2 L Erlenmeyer flask) were inoculated from <10 mL overnight "starter" cultures and cultured to O D 6 0 0 = 0.45-0.6. After 15 min. incubation on ice, 460 mL culture was transferred to chilled 250 mL centrifuge bottles and centrifuged 10-15 min. at 3,800-7,500 x g m a x (4°C) to recover cells. These were washed once or twice by resuspension in 0.5-1.0 volume ice-cold water or 1 mM HEPES pH 7 followed by centrifugation as before, and additionally washed once by resuspension in 0.03-0.04 volume ice-cold 10% glycerol and centrifugation for 10-15 min. at 4,000-6,000 x gmax (4°C). After a final resuspension of cells in 0.002-0.003 volume ice-cold 10% glycerol, 50 uL aliquots (in 500 uL Eppendorf tubes) were frozen by pushing the tubes into pulverized dry ice. These were stored at -70°C to -80°C until immediately before use. This concentration of tetracycline does not harm host cells but induces expression of phage-encoded tetracycline resistance (see Smith, 1992). 48 (b) Electroporation. Electroporations were performed using a Biorad Gene Pulser with external Pulse Controller and 0.2 cm electroporation cuvettes. For each electroporation, up to 2.5 [iL of D N A in ligation or other buffer, or up to 3.5 uL of D N A in water was added to 50 [iL frozen electrocompetent E. coli MCI061 in 500 uL Eppendorf tubes (held on pulverized dry ice until immediately before use). After each cell/DNA mixture was thawed (by holding the tube between fingers) and mixed (by flicking the tube each -10 s during thawing) and subsequently incubated on ice -30 s, the mixture was transferred to a chilled cuvette within a chilled holder and electroporated at 2.5 kV, 200 or 400 Q, 25 [iF. Time constants (Miller 1994) were typically 4.5-4.6 (electroporations at 200 Q) or 8.9-9.0 (at 400 Q) for electroporations with D N A samples in water, and 4.1-4.3 (200 Q) or 8.3-8.7 (400 Q) for samples in ligation buffer. Immediately after electroporation, each mixture was transferred to 2 mL L B or SOC (pre-aliquoted in glass tubes, at room temperature; with 0.2 u.g tetracycline mL" 1 for fUSEn or fDRWn vectors and their recombinants, as in §2.5.1 .b) and incubated 0.5-1 h at 37°C with rapid shaking (175-225 rpm) before spreading on LB-Tet in standard petri dishes (commonly 1 [iL to 100 \xL aliquots) or 245 mm x 245 mm N U N C culture dishes (<2 mL aliquot); these were incubated >12 h at 37°C. 2.5.3. Infection of F-piliated host ceils with virions After a small quantity (1-40 uL) of generally unqualified virions (fUSE/z, fDRWrc or their recombinants) in culture supernatant, L B broth, PBS or TBS were combined with 50 uL to 2 mL E. coli K37, K91 or K91-Kan (all F-piliated) grown to visible turbidity ( O D 6 0 0 <0.9, commonly -0.2-0.3), the cell/virion mixture was incubated 20 min.-1.5 h (37°C, standing) in medium containing 0.2 \ig tetracycline mL"1 (see §2.5.l.b) to allow infection and expression of phage-encoded tetracycline resistance before transferring to L B broth containing 20 \xg tetracycline mL"1 or spreading on LB-Tet plates. 49 2.6. H A R V E S T I N G O F VIRIONS 2.6.1. Harvesting virions from bacterial growth on solid media Virions were harvested from colonies on solid media by adding TBS or PBS (most commonly), or L B to the media surface (e.g., 10-30 mL for a 245 mm x 245 mm N U N C culture dish, 3-4 mL for a standard petri plate) and washing bacterial growth from the surface with a glass spreading rod. Virions within washings were recovered by centrifugation, as below. 2.6.2. Separation of virions in culture or wash supernatant from cells by centrifugation Virions in culture supernatant or culture plate washings (§2.6.1) were separated from host cells by centrifugation (i) >3 min at maximum speed (4°C or room temperature) in an Eppendorf 5415C centrifuge, for volumes <2 mL, or (ii) >10 min at 3000-7500 x g m a x (4°C) for volumes >2 mL. In some cases supernatants were incubated >10 min. at 68-70°C to kill residual host cells and centrifuged as before. 2.6.3. One-stage PEG precipitation After cultures of phage-infected E. coli were centrifuged (as immediately above), virions were precipitated from culture supernatant with one of two formulation of polyethylene glycol (PEG)/sodium chloride: (i) PEG formulation " A " (30% w/v polyethylene glycol 8000, 3M NaCl (Qiagen 1991) or (ii) PEG formulation " B " (16.7% w/v polyethylene glycol 8000, 3.3 M NaCl; Scott & Smith 1990; Smith 1992). Typically, virions being precipitated for preparation of sequencing template were precipitated by adding 0.2 volume of PEG " A " to culture supernatant, mixing by 100 inversions, storing on ice >2 h (commonly >4 h) or at 4°C overnight, and centrifuging as described below; virions being precipitated for use in ELISA or other assays were commonly precipitated with PEG " B " . After incubation on ice or at 4°C, (i) virion/PEG samples of volume <1.6 mL (in 1.5 or 2 mL Eppendorf tubes) were centrifuged 50 0.25 to 1 h (4°C) at maximum speed in an Eppendorf 5415C centrifuge; (ii) samples of volume <37.5 mL, at 12,000-34,000 x g m a x (in recent preparations, 34,000 x g m a x ) >0.25 h (in recent preparations, 0.5-1 h) at 4°C in 50 mL Oakridge tubes; and (iii) samples >37.5 mL, >0.5 h (typically >1 h) at 12,000-22,000 x g m a x (most commonly at 15,000 x g m a x ) at 4°C in 250 mL centrifuge bottles. Pellets were resuspended in 0.02-0.1 volume PBS, TBS or Qiagen buffer L3 (§2.3.2.b) as convenient and appropriate. In most cases samples were centrifuged again to remove insoluble debris: (i) volumes <2 mL were centrifuged >3 min. at maximum speed in an Eppendorf centrifuge (4°C or room temperature); (ii) larger volumes, >10 min. at 3,000-6,000 x g m a x (4°C). 2.6.4. Two-stage PEG precipitation Except in some cases where virions were to be further purified by CsCl density gradient (§2.6.5) or were being prepared to serve as a source of s.s. sequencing template, virions precipitated once by PEG precipitation were precipitated again by addition of PEG "B" , incubation on ice >2 h or at 4°C overnight, and centrifugation as before. After resuspension in <1.8 mL PBS, TBS or Qiagen buffer L3, samples were centrifuged >3 minutes at maximum speed (4°C or room temperature) in an Eppendorf 5415C centrifuge to recover virion-containing supernatant. 2.6.5. CsCl density gradient centrifugation CSP-library clones and selected other phage precipitated by two-stage (most cases) or one-stage PEG precipitation were further purified by CsCl density gradient centrifugation (Smith 1992). Virion yields from (typically) 200-600 mL overnight culture were suspended in 31% CsCl (w/w) in TBS (density = 1.30 g mL" 1) and centrifuged 48 h at 237,000 x g m a x (37,000 rpm) in a SW41Ti rotor (5°C) before recovering a single faintly bluish band (when illuminated by lighting from one side) as CsCl-purified virions. After washing in ~20 mL TBS and centrifugation for 4 h at 257,000 x g m a x (4°-5°C), virions were resuspended in 0.5 to 1.5 mL 51 PBS or TBS. 2.7. VIRION QUANTIFICATION A N D A N A L Y S I S 2.7.1. Quantification of virion particles by ultraviolet (UV) spectroscopy (adapted from Smith 1992; see also Yamamoto et al. 1970) Dilutions of PEG-precipitated or CsQ-purified virions in TBS or PBS were quantified by deducting sample absorbance at 320 nm (A320, believed due to light scattering of impurities; Smith 1992) from absorbance at 269 nm ( A 2 6 9 ) and multiplying the net value (net A 2 6 9 ) by (i) 197 \ig mL"1 to estimate total phage protein or (ii) 6.5 x 10 1 2 virions mL"1 to estimate numbers of physical particles as fUSE2 (Figure 7-1) equivalents. Because genome size and thus virion length and protein content of a virion increase to accommodate foreign D N A inserted into g i l l (with a concomitant increase in A 2 6 9 per virion), the latter quantity — physical particles as fUSE2 equivalents — nominally over-estimates the numbers of physical particles of recombinant phage. For example, quantities of recombinants possessing 600 base pair (bp) inserts are overestimated by ~6%; those with 100 bp inserts by ~1%. 2.7.2. Quantification of plaque-forming units (pfu) and transducing units (TU) (a) Full-plate plaque assay (adapted from Smith 1992). After adding 400 uL F-piliated E. coli K37, K91, K91-Kan, or K91/pNK1759 grown to an O D 6 0 0 of 0.6-0.9 (most commonly 0.7-0.8, corresponding to late exponential or early stationary growth phase; Figure 2-1) to 100 U.L culture supernatant or serial 10-fold or other dilutions (in culture media, PBS or TBS), 3 mL medium (commonly LB) containing 0.7-0.75% soft agar held at 48°-50°C was added by pipettor or by pouring from aliquots. The cell/virion/agar mixture was poured onto plates containing L B or other medium with 1.5%> agar (in most experiments these plates had been warmed to 37°C) and held at room temperature until the soft agar solidified. Plates were incubated >7 h (37°C) before plaque-forming units (pfu) were counted. Pfu of wild-type phage f l , conventional vectors (e.g., fUSE2), and recombinants and 52 pseudorevertants of amber7 and frame-shift8 vectors were enumerated using E. coli K91 or K91-Kan as convenient. Early assays to enumerate total pfu (i.e., both pseudorevertant and non-pseudorevertant pfu) of amber vectors employed the amber-suppressing E. coli K37. Because plaques on K37 are difficult to discern (indeed, fDRW613C produces no plaques on this strain; see Table 2-IV and Figure 7-14), later assays employed E. coli K91/pNK1759 which produces readily discernible plaques for fDRW613C and a greater number (than K37) of discernible plaques for other amber vectors (Table 2-IV). Table 2-IV. Evaluation of host strains and media for plaque formation by amber vectors. Plaque formation 3 by host strain / medium E. coli ¥31 E. coli K 9 1 / p N K l 759 L B 2 x Y T L B 2 x Y T fDRW613 85 ± 2 6 difficult to discern b or none discernible 388 ± 4 7 405 ± 3 7 f D R W 6 1 3 C none discernible b 451 ± 6 3 447 ± 5 0 f D R W 7 0 605 ± 26 1 1 1 9 ± 6 1 1181 ± 7 3 a For each offDRW613, fDRW613C and fDRW70, full-plate plaque assays were performed in triplicate for 100 u.L samples of a dilution of overnight culture supernatant, for (i) each of the two indicated host strains, and (ii) each of the two indicated media. Body of table shows mean number of plaques ± one standard deviation. b Compared to plaques on E. coli K 9 1 / p N K l 759, those on E. coli K37 are faint and difficult to discern, particularly for fDRW613C. (b) Pfu "spot" assay. Lawns of F-piliated E. coli were prepared in essentially the same manner as described above (§2.7.2.a) except that L B broth was substituted for dilutions of culture supernatant. After allowing the top agar to solidify, 4-5 uL samples (singly, or in duplicate or triplicate) of undiluted (requires E. coli K37/pACYC184 or K91-Kan/pACYC184 or another tetracycline-resistant F-piliated strain) or diluted (tetracycline-resistant strain not 7 Virions produced by amber vectors (containing an amber codon in gill) nominally plaque only on amber-suppressing host strains. However, mutations commonly arise in which the amber codon is changed to a sense codon (e.g., encoding Tyr). These "pseudorevertants to wild-type" are accordingly able to plaque on non-amber-suppressing host strains. 8Virions produced by frame-shift vectors (vectors containing a frameshift in glII) are nominally non-infectious and accordingly produce no plaques. However, mutations commonly arise in which the gi l l reading frame is restored. These "pseudorevertants to wild-type" are accordingly able to produce plaques. 53 required) culture supernatant were applied, as "spots", to the lawn. After excess liquid had been absorbed by the media, plates were incubated >7 h (37°C) before plaque-forming units were counted. This "spot" assay allowed up to 24 samples to be applied to a standard plate; in many cases, 40 or more plaques could be distinguished in a single spot. (c) TU "spot" assay. This "transducing unit" (TU) assay derives from the ability of fUSE/z and fDRWn vectors and recombinants to transduce tetracycline resistance into host cells. For recombinants deficient to some degree in their abilities to infect host cells, assays of TU are in principle more sensitive than assays of pfu, since the former require only a single infection event to give rise to a colony, while the latter require multiple rounds of virion propagation and infection to give rise to a plaque (Smith 1992). Seventy-five to 100 uL E. coli K91-Kan grown to visible turbidity ( O D 6 0 0 = 0.25-0.7) were combined with 5-20 \xL samples of serial dilutions (commonly 5- or 10-fold in L B , PBS or TBS) of culture supernatant in sterile 96-well microtiter plates. After 0.5-2 h incubation (37°C) in medium containing 0.2 u,g tetracycline mL"1 (see §2.5.1 .b), 4-5 u.L samples ("spots") were applied, singly or in duplicate, to LB-Tet plates; these were incubated overnight before tetracycline-resistant colonies were counted as TU. This "spot" assay allowed up to 24 samples to be applied to a standard plate; in many cases, 20 or more colonies (TU) could be distinguished in a single spot. 2.7.3. Agarose gel electrophoresis of virions (adapted from Griess et al. 1990) PEG-precipitated virions (§2.6.4) with sample loading buffer (0.25% bromophenol blue in 30% glycerol diluted 1:5 with virion sample) were electrophoresed >1.5 h at ~7 V cm"1 in 0.7%) agarose, stained -1.5 h in 0.05% (w/v) Coomassie, 10%) glacial acetic acid, and de-stained with several washes in 10% glacial acetic acid over a period of several hours, or overnight, until virion bands could be satisfactorily discerned. 54 2.8. V E C T O R C O N S T R U C T I O N 2.8.1. Conventional vector fDRW20 Ten fmol fUSEl RF D N A extracted from cultures screened for pseudorevertant production, digested to apparent completion with Pvall and treated with alkaline phosphatase (Boehringer), was combined with 1 umol (thus, a 100:1 molar insert to vector ratio) of an annealed (§2.2.5.b) self-complementary 5'-phosphorylated oligonucleotide (NEB) containing a BgUl restriction site (Figure 7-12). Vector and insert were ligated (1 U T4 ligase, §2.2.2.b) 25 h at 14°-16°C before electroporating (§2.5.2) one-eighth of the ligation products into E. coli MC1061. After 48 transformants (isolates) cultured individually overnight in 200 |_iL LB-Tet in wells of microtiter plates were combined into 6 pools of 8 isolates, 0.8 mL of each pool was transferred to 32 mL fresh medium and incubated overnight. Samples of 5g/II-digested RF D N A extracted (§2.3.1) from these cultures were analysed by agarose gel electrophoresis to determine which pool contained the largest fraction of transformants possessing Bglll sites. Each of the 8 isolates comprising the pool so identified was cultured by serial overnight culture (§2.1.1, 50 mL) before harvesting (§2.3.1) RF D N A from the final cultures. With the goal of removing multimeric inserts, unquantified (assumed to be 5-6 u.g) RF D N A was digested with Bglll (50 U per isolate, 2-3 h) and precipitated with isopropanol to remove excised BglW fragments. After samples were analysed by agarose gel electrophoresis, 17 fmol RF D N A of each of the five (of eight) isolates that yielded the expected restriction pattern (a single 9.4 (kbp fragment) was re-ligated (10 U T4 ligase, §2.2.2.b) for 3.5 h at 22°-23°C and electroporated (§2.5.2) into E. coli MC1061. Six transformants were selected for each of the 5 isolates; after overnight culture in 200 uL LB-Tet in microtiter plate wells, 5 uL culture supernatant from each transformant was applied to a lawn of E. coli K91-Kan/pACYC184 in a manner similar to pfu "spot" assays (§2.7.2.b). After serial overnight culture (§2.1.1, 50 mL) of four of the 17 virion-producing transformants, s.s. D N A extracted 55 (§2.3.2.b) from virions harvested from culture supernatants by two-stage PEG precipitation (§2.6.4) was employed as sequencing template (§2.4) to identify all of the four sequenced clones as possessing the" target fDRW20 sequence. 2.8.2. Conventional vectors fDRW21 and fDRW22 In separate ligations, 11 fmol fUSEl RF D N A extracted from cultures that screened for low pseudorevertant production and digested to apparent completion with PvuW was combined with 1 umol and 5 umol (thus 100:1 and 500:1 molar insert:vector ratios) of an annealed (§2.2.5.b) oligonucleotide pair containing a BgUl restriction site (Figure 7-12). Each mixture was ligated (10 U T4 ligase, §2.2.2.b) 22 h at 14°-18°C before electroporating (§2.5.2) one-eighth of the ligation products into E. coli MC1061. After culturing 80 transformants (isolates) overnight in 200 uL LB-Tet, virion-producing isolates were identified by applying 10 \xL samples of culture supernatant to lawns of E. coli K91-Kan/pACYC184 in a manner similar to pfu "spot" assays (§2.7.2.b). After serial overnight culture (§2.1.1, 50 mL) of each of the 11 plaque-producing isolates, Z?gfll-digested samples of RF D N A extracted (§2.3.1) from these cultures were analysed by agarose gel electrophoresis. For each of the seven isolates whose RF D N A had digested with BgUl, 10 uL culture supernatant was used to infect (§2.5.3) 50 uL E. coli K91-Kan, and infected cells were plated on LB-Tet and incubated overnight. For each of the seven isolates, a single colony was used as a source of inoculum for a serial overnight culture (§2.1.1, 37 mL). S.s. D N A extracted (§2.3.2.b) from virions harvested from culture supernatants by one-stage PEG precipitation (§2.6.3) was employed as sequencing template (§2.4) to identify three of the sequenced clones as possessing the target fDRW21 sequence and three others, the fDRW22 sequence. 2.8.3. Amber vector fDRW5 In separate ligations, 10 fmol fUSE5 RF D N A extracted from cultures screened for low pseudorevertant production, digested to apparent completion with Sfil and precipitated with 56 isopropanol to eliminate "stuffer" fragments (Figure 7-1B), was combined with 30 and 90 fmol of an annealed (§2.2.5.b) oligonucleotide pair constituting a replacement amber-containing "stuffer" fragment (Figure 7-9). These mixtures were ligated (10 U T4 ligase, §2.2.2.b) 22 h at 10°-16°C before transforming (§2.5.1) one-half of the ligation products into E. coli K802. Virions produced by transformants were harvested by washing bacterial growth from plates (§2.6.1) and subsequent centrifugation. After heating wash supernatants 30 min. at 70°C (to ki l l residual host cells), 40 \iL of a 10"2 dilution of supernatant was used to infect (§2.5.3) 50 uL E. coli K37, and dilutions of the infected cells were spread on LB-Tet and incubated overnight. After serial overnight culture (§2.1.1, 45 mL) of selected transformants, virions were recovered from culture supernatants by two-stage PEG precipitation (§2.6.4) and s.s. D N A was extracted (§2.3.2.b) and stored (4°C) for later use. After transforming this s.s. D N A into E. coli K802, single colonies (one per original transformant) were used as a source of inocula for serial overnight cultures (§2.1.1, 40 mL). S.s. D N A extracted (§2.3.2.b) from virions harvested from culture supernatants by one-stage PEG precipitation (§2.6.3) was employed as sequencing template (§2.4) to identify the two successfully sequenced clones as possessing the target fDRW5 sequence. 2.8.4. Amber vector fDRW613 In separate ligations, 17 fmol fDRW5 RF D N A digested to apparent completion with Sfil and precipitated with isopropanol to eliminate "stuffer" fragments (Figure 7-9) was combined with 51 and 170 fmol of annealed (§2.2.5.b) oligonucleotide sets L I , M3 and R (Figure 2-2). These mixtures were ligated (0.5 U T4 ligase, §2.2.2.b) 20 h at 15°-18°C before transforming (§2.5.1) one-half of the ligation products into E. coli K802. Virions produced by transformants were harvested by washing bacterial growth from plates (§2.6.1) and subsequent centrifugation. Five uL of each wash supernatant was used to infect (§2.5.3) 120 [iL E. coli host K37, and dilutions of the infected cells were spread on LB-Tet and 57 X • CD X 3 a, s; O o << s .bo .2 cd o • CO .a u fa ^ y CD . o t/) 0) (D — .•3 O t/1 ~ o CD > O -a c cd 3 o bp ~o CD M 2 Br) . - H <D ex £8 CD i CD e v " o 2 3 £ a. cu . 3 «2 -a U) 3 a o +3 eo -g 2 Cd o eo a s O bO « w t/i t/) ' E D . <u 3 CD O c CD 3 a-CD C/5 T3 CD X O PQ 3 o T 3 O o o a. O a c CD o c CD 3 C T CD .3 60 - 3 CD > o 3 U c H O ea < cd in o > <U CD X ! g S • cd O s n -5 CD E o o so £ .S o £ -a <N CD <i OS CD 3 Cd I K C to ^ X! -a CD -*-» O 3 • 3 o 3 60 n in i i 13 U EH in m m in i i • O • <C • EH • U u o EH <; u o EH <; o u CD u in ro T 3 m S >—1 «<+-< — o CD -r-« Q. CD a 5 § o u CD c« .SP ^ "O O H <o - t + CD P < K C T 3 ^ "u C J S o C M - • s i i - l r<l 'tt <D CD - e g <N tfc ^ ^ 0 0 ^ c^-S ^ '-3 •= a § 1 en -a 1 = I = o a-bo CD IS ts> O CD CD O. >> o ^ CD 5 a CD >— c« CD •r- bD E .5 8 s •< CD i - 3 O —' CD J 3 T 3 CD C 3 C3 CD T 3 o CD -a CD • 22 + K ^ 5 CD CD Pi 03 + c c 2 c J 3 CD C/5 — CD <tt 5 E -a ™ <u .S S T 3 S "J ca CD 1 - 2 — w .2 t . - 3 <D _ cn . 3 _S CD 2 C+H o CD E CD co CD O 3 3 3 3 3 — O c/l 6C — — ro O O CD in CD -a •a CD (J- K ^ -> o '+3 3 O C CD Si o 3 -a •— CD en > <D E cd cn CD 4 3 a: Q 3 60 CD T 3 .s a S O w CD M > °- S E o. w i CD t . T 3 ^ '> bD o .S "a I c3 C at t -M CD § E 1 1 S i "> .5 8 3 3 o • 3 cd <D O a. 3 O CD 3 3 cd C+H K O ro .X - 3 '53 "EJD >> S O o J 3 VO l e cd CD o > &>" o ^ O 3 3 • -"Z ~° C CD « X -CD 3 CD T 3 J D Xn T 3 3 O 3 3 3 cn ^ O ^ cd 58 © c j I— <C CD C D < 'TC C J C D 1— <c <c \— C D o 1— < C J C D h- <C C J C D 1— < C D C J C D O I— < CD O CD C J O CD CD O CD O \— <. CD C J CD O < £ I— < I— CD C J C J CD C J CD I— < C J CD CD C J CD C J 0£ © O I— < C D C J © C J C D O C J C J CD 1— < < h -CD C J CD O C J CD I— < (_) CD I— < CD C J C J CD 1— < < I— CD C J < I— C J CD I— < £ C J CD ro ro ro ro rO CD C D zs CO =3 (O .— cn C D ro zi ro =3 C D ro cn C D ro Cn C D +-> ,— +-> r -+-> 4 J CO +-> i— +-> •— cn CT> => +-> ro +-> ra cn > cn => 4-> 4-> S-C J u -*-> +-> s_ CD f— TO (— u CO sz. \— u ra sz L _ CD Z5 ro 13 A3 1— ra i— ( D ro =3 C D C D cn C D ro cn C D ro cn C D U ro C J ra CJ C J i— u ro C J ra C D < cn < C u Cn < C J cn < CO cn >•> CTJ >— cn <— cn cn C D 1— C D ! _ C D C D 1— s -Oi C D C D cn C D C D C J CD C J Q J (— t_ h - S_ 1— 00 h - oo O CD C J QJ I— OO \— OO r - >1 1— C D C D C D C D <— C D C D 1— C D > > h -C D >. C J C J > , C D C D C D C D C D C D C D C D <— C D C D C J > ^ C J >, C D C D h - 1— >> C D C D C D C D C D C D i— C D C D C D C D 1— 1— C D >. C D ZJ < =3 C D C D C D C D < i— < r— C D C D C D C D < rs < < zs C J C - C J C C D C D C D C D C J O J C J QJ 1— OO \— OO O C J S -QJ C J C J s_ aj C J > > C J > , 1— 00 1— OO C D 1— C D r— C D C D C D C D C J C D > > C J C D >^  1— >1 1— >^  C D C D C D C D C D 1— C D r— C D C D C D C D h -C D 1— C D > > C J > i C J > > C D C D C D C D C D C D . — C D C D C D C D C J C D > > O C D C J 1_ C J 0J C D C D C D C D C J CD OO <c —• C J QJ C D C < ^ C D =^ C D C r < .— h-< ~ < C J C D C D C D C J C D C D sz 1— C L to C J 1— =3 cu h - t -C J C J < C J C D < C D < base C J 1 CO CO 1— OO CO h- S- ~ o base |— X I C D > > C D . — CD CO ra C J <D CO O ( U C D CNI \— OO CD CO ro 1— OO + C D C D + C D C D C D C J i— co C D C D - Q C D '— h- fO < - < s C D C D + C D 1— h - ^ CO CD E C D CO C D e O >•> C J > , CNj" < < e TO C D C D c C D r— CD CO h -C D C D C D C D E CD O ) 2 C J >, E C J > ^ C D c T J C D C 4= C D 1— ro C D read <C 1— ro cn C D C D >*— C D C D read C J C D CD C J C D c: cn read X J ra CD < t_ C J QJ h~- C L h - Q - C J 0) X J ro CD h -<C LO <C to 1— OO < i—i _ C D < _ C D < < >> C D SZ 1— s_ |— j_ C D < i— O QJ C J Q J C D C D C J C D I— OO 1— OO C J C J zs h - S_ |— C J i— I— <u C J QJ 1— ro h— ra O 1 1— OO C D > C D > C D C D cn > > C D C D > . C D r -Cn C D y— C D cn > > C D C D C D cn > > C D C J C L u C L C J C L C J C L rn I/) ra in CO CO ro CO Cn <C cn <C Cn <C Cn <£ C J ro a ro C J ro C J ro C J U i— C J C J C D < Cn < Cn < E cn <; <- cn •4 cn < cn cn s - S- S-u QJ u aj C J aj C J Q) +-> OO +J OO 4-> OO +-> OO c n <— C D C D CD CD CD <: =3 CD CD CD CD I BJD C J i -C J Q> Q Q Q Q Q CD O 59 incubated overnight. Eight transductants (four from each ligation) were streaked on LB-Tet and, for each transductant, a single isolate was used as a source of inoculum for a serial overnight culture (§2.1.1, 42 mL). RF D N A extracted (§2.3.1) from these cultures was screened for the presence of a Bsal restriction site by agarose gel electrophoresis of Bsal-digested RF DNA. For each of the six isolates whose RF D N A had digested with Bsal, s.s. D N A extracted (§2.3.2.b) from virions harvested from'supernatants of the 42 mL cultures by two-stage PEG precipitation (§2.6.4) was employed as sequencing template (§2.4) to identify four of the six isolates as possessing the target fDRW613 sequence. 2.8.5. Amber vector fDRW613C In separate ligations, 10 fmol fDRW5 RF D N A digested to apparent completion with Sfil and precipitated with isopropanol to eliminate "stuffer" fragments (see Figure 7-9) was combined with 50 and 250 fmol annealed (§2.2.5.b) oligonucleotide sets L I , MC3 and R (Figure 2-2). These mixtures were ligated (5 U T4 ligase, §2.2.2.b) 18 h at 14.5°-15°C before transforming (§2.5.1) one-half of the ligation products into E. coli LE392. Twelve transformants (isolates; six per ligation) were cultured by serial overnight culture (§2.1.1, 40 mL) and ifoal-digested RF DNA extracted (§2.3.1) from each of these cultures was screened for the presence of a Bsal restriction site by agarose gel electrophoresis. Each of the eight isolates whose RF DNA had digested with Bsal was again cultured by serial overnight culture (§2.1.1, -38 mL) and s.s. DNA extracted (§2.3.2.b) from virions harvested by two-stage PEG precipitation (§2.6.4) of culture supernatants was employed as sequencing template (§2.4) to identify eight of the 12 isolates as possessing the target fDRW613C sequence. 2.8.6. Amber vector fDRW70 In duplicate ligations, 12 fmol fDRW5 RF D N A derived from cultures previously screened for low pseudorevertant production, digested to apparent completion with Sfil and precipitated with isopropanol to eliminate "stuffer" fragments (Figure 7-9) was combined with 60 36 fmol of an annealed (§2.2.5.b) oligonucleotide pair constituting a replacement amber-containing stuffer fragment (Figure 7-23A). These mixtures were ligated (10 U T4 ligase, §2.2.2.b) overnight at >14°C before transforming (§2.5.1) one-half of the ligation products into E. coli LE392. RF DNA harvested (§2.3.1) from serial overnight cultures (§2.1.1, 40 mL) of six transformants (isolates) was screened for the presence of Fspl, Pvull and Xbal restriction sites and for the absence of an Sfil site by agarose gel electrophoresis of RF D N A digested with these enzymes. For each of the four isolates whose RF D N A possessed the desired restriction patterns, s.s. D N A extracted (§2.3.2.b) from virions harvested from culture supernatants by one-stage PEG precipitation (§2.6.3) was employed as sequencing template (§2.4) to identify isolates possessing the target fDRW70 sequence. 2.8.7. Conventional vector series fDRW8nn In separate ligations, 12 fmol fDRW5 RF DNA, derived from cultures of isolates whose sequences had been confirmed to contain an intact amber codon within the 5/zI-excisable "stuffer" fragment (see Figure 7-9), digested (possibly not to completion) with an excess of Sfil (> 100 U [ig"1 DNA) and precipitated with isopropanol to remove stuffer fragments, was combined with 12, 60, 300 and 1500 fmol semi-degenerate, incompletely complementary oligonucleotides (Figure 7-29B). These mixtures were ligated (5 U T4 ligase, §2.2.2.b) 4 h at 15°-16°C before precipitating the ligation products (with 10 [ig tRNA added, per sample, as carrier) with ethanol and transforming (§2.5.1) the resuspended ligation products into E. coli LE392. Only 30 transformants were recovered. RF D N A extracted (§2.3.1) from serial overnight cultures (§2.1.1, 40 mL) of 16 of these was screened for the presence of a putative Srfl restriction site by agarose gel electrophoresis of SmaZ-digested RF D N A (Figure 7-29B). For each of the 12 isolates whose RF D N A was digested with Smal, undigested RF D N A was transformed into E. coli LE392 and a single transformant was selected and propagated by serial overnight culture (§2.1.1, 42 mL). S.s. D N A extracted (§2.3.2.b) from virions harvested from 61 culture supernatants by two-stage PEG precipitation (§2.6.4) was employed as sequencing template (§2.4) to identify the six unique fDRW8wz sequences shown in Figure 7-29. RF D N A extracted (§2.3.1) from the corresponding 42 mL cultures was employed (§2.12.3.1) to construct the B. pertussis fhaB DNase I fragment libraries summarized in Table 7-VI. 2.9. V E C T O R STABILITY ISSUES 2.9.1. Initial studies of frameshift vectors (a) JVSE5 propagated in E . coli K91-Kan (Figure 7-2A and B). fUSE5 RF D N A was extracted from three serial overnight cultures (§2.1.1, 500 mL) of E. coli K91-Kan that had been infected (§2.5.3) with fUSE5 virions (fUSE5 s.s. D N A packaged by G. P. Smith by means of a pill-encoding helper plasmid), using a Qiagen Tip 500 DNA extraction/purification. Two series of Sfil digests of this RF DNA were performed. In the first, ~90 [ig RF D N A was digested with 100 U Sfil (GIBCO/BRL) in a total volume of 290 uL. After 1 h and 2 h, a 3 [iL sample was withdrawn, an additional 100 U Sfil (in 10 \iL) added, and digestion continued for an additional hour before analysing samples of undigested and digested D N A by agarose gel electrophoresis (Figure 7-2A). In the second series, ~1 [ig samples were digested 2 h with 5, 10 and 20 U Sfil (GIBCO/BRL) before analysis by agarose gel electrophoresis (Figure 7-2B). (b) fUSEn propagated in E . coli K91-Kan (Figures 7-2C and 7-3). After infecting (§2.5.3) 0.5 mL E. coli K91-Kan with ~107 fUSEl, fUSE2, fUSE3 and fUSE5 virions (except for fUSE2, these had been packaged by G. P. Smith using a pill-encoding helper plasmid), infected cells were spread on LB-Tet plates and incubated overnight. After washing the resulting bacterial growth from plates with cold PBS (20 mL per fUSEn vector), RF D N A was extracted from the wash using a Qiagen Tip 100 DNA extraction/purification kit using volumes of reagents appropriate for 180 mL overnight broth cultures. In a followup to the experiment described earlier (§2.9.1.a), ~1 [ig aliquots of fUSE5 RF D N A were digested 2 h with 0, 0.5, 62 1, 2, 4 and 8 U Sfil (Pharmacia) and analysed by agarose gel electrophoresis (Figure 7-2C). Separately, samples of fUSEl , fUSE2, fUSE3 and fUSE5 RF D N A were digested with Pvull, BgUl, Xhol or Sfil as appropriate to each vector's cloning site and, together with undigested samples, were analysed by agarose gel electrophoresis (Figure 7-3). (c) Initial assessment of pseudorevertant production by fUSE5 that had been passaged in E. coli K91-Kan (Figure 7-4A). After -1-3 [ig RF D N A of each of fUSE2 and fUSE5 propagated in E. coli K91-Kan (§2.9.1 .b) was electroporated (§2.5.2) into E. coli MC1061, isolated colonies of E. coli MCI061 transformed with fUSE2 (two transformants) and fUSE5 (20 transformants) were cultured overnight in 2 mL LB-Tet. For each isolate, a single 5 [iL sample of a 100-fold dilution of culture supernatant was applied to a lawn of E. coli K91-Kan in a manner similar to pfu "spot" assays (§2.7.2.b) and rough estimates of virion production were obtained from the numbers of plaques formed after >24 h incubation (37°C). (d) Initial assessment of pseudorevertant production by fUSEn vectors provided by Smith (Figure 7-4B and C). After s.s. D N A extracted from Smith-supplied virions (§2.3.2.a) was transformed (§2.5.1) into E. coli K802 and transformants spread on plates of L B + 40 |ig tetracycline mL"1, selected transformants were assayed twice for virion production. In the first assay (Figure 7-4B) 12 isolates of each of fUSEl , fUSE3 and f(JSE5 were cultured overnight in 2 mL LB-Tet. For each isolate a single 5 [iL sample of undiluted culture supernatant was applied to a lawn of E. coli K91-Kan/pACYC184 in a manner similar to pfu "spot" assays (§2.7.2.b), and rough estimates of virion production obtained from the numbers of plaques formed after overnight incubation (37°C). In the second assay (Figure 7-4C), 18 isolates of fUSEl and 24 isolates of each of fUSE3 and fUSE5 were cultured overnight in 2 mL LB-Tet. For each isolate, duplicate 5 uL samples of undiluted culture supernatant were applied to lawns of E. coli K91-Kan/pACYC184 and virion production assessed after overnight incubation. (e) Use of chloramphenicol to reduce s.s. DNA species (Figure 7-5). After culturing 63 (to OD600=1.3-1.5) four isolates of E. coli K802/TUSE5 that had been screened for low pseudo-revertant production, cultures were pooled and the pool used as an inoculum (1:50 dilution) for duplicate 110 mL LB-Tet cultures (labelled A and B). These were incubated 4 h (to O D 6 0 0 ~0.7) before adding chloramphenicol to culture B (see Figure 7-5) to a final concentration of 15 fig mL"1 and incubating an additional 1 h (culture A , final O D 6 0 0 =1.2) or 1.75 h (culture B, OD 6 0 0=0.8). RF D N A extracted from these cultures (using a Qiagen Tip 100 D N A extraction/purification kit with reagent PI, P2 and P3 volumes scaled up 1.2-fold for culture A) was analysed by agarose gel electrophoresis (Figure 7-5). (f) Identification of s.s. bands with SI nuclease (Figure 7-6). In a followup to the experiment described above (§2.9.1.e), 0.8-0.9 \ig samples of D N A derived from culture A were treated: (i) with 7.2 U SI nuclease (10 min. at room temperature, volume = 14 uL; in 12 mM Tris, 0.2 mM EDTA, 0.9 mM MgCl 2 , 29 mM sodium acetate, 43 m M NaCl, 0.03 m M ZnS0 4 , pH >4.5) , (ii) with 10 U BamHl (1 h, 37°C) and (iii) with 10 U EcoRl (1 h, 37°C). Samples of D N A so treated, together with samples of undigested D N A from cultures A and B, were analysed by agarose gel electrophoresis (0.83 V cm"1 for 12 h, Figure 7-6). (g) Elution of s.s. DNA with 1.35 M NaCl in Qiagen column wash (Figure 7-7). One-mL samples of 12 h cultures of 4 isolates of each of E. coli K802/fUSEl and E. coli K802/TLTSE5 that had been screened for low pseudorevertant production were diluted 1:50 into fresh medium (LB-Tet) and incubated to O D 6 0 0 -0.65. After chloramphenicol was added to a final concentration of 15 \xg mL"1, cultures were incubated an additional 2.5 h (final O D 6 0 0 = 0.7-0.9) before harvesting D N A using a Qiagen Tip 20 D N A extraction/purification kit generally in accordance with the manufacturer's recommended procedures but with the following modifications: (i) 3 mL volumes of reagents PI, P2 and P3 were used in place of the 0.3 mL volumes recommended for 3-6 mL cultures; (ii) after the recommended wash with Qiagen reagent QC, an additional wash was performed with 3.8 mL QC-1.35 (reagent QC with 64 NaCl increased from 1.0 M to 1.35 M). Samples of D N A so extracted, together with samples of the QC-1.35 eluate (concentrated using a Centricon 30 microconcentrator) were analysed by agarose gel electrophoresis (Figure 7-7). 2.9.2. Evaluation of E. coli host strains for fDRW5 propagation (Table 7-1) fDRW5 RF DNA was transformed (§2.5.1) into E. coli K37, K802 and LE392 and, for each strain, 6 transformants were selected for analysis. For each transformant 150-200 \xL of overnight culture was used to inoculate a 50 mL LB-Tet culture. After 24 h incubation, RF DNA was extracted (§2.3.1) from 35 mL culture and samples of undigested RF D N A evaluated by the location and appearance of bands after agarose gel electrophoresis. Virions capable of plaquing on a non-amber-suppressing host (viz., pseudorevertants) were counted by a pfu "spot" assay (§2.7.2.b) using duplicate samples of undiluted culture supernatant applied to E. coli K91-Kan/pACYC184. 2.9.3. Comparative analysis of frameshift and amber vector stability (Figure 7-15) For each of fUSEl , fUSE2, fUSE3 and fUSE5, s.s! DNA extracted from Smith-supplied virions (§2.3.2.a) was transformed (§2.5.1) into E. coli K802 and — in a preliminary analysis of vector stability (not shown) — selected transformants (isolates) were assayed for virion production. For each vector, two or three isolates that produced few or no pseudorevertants were selected. RF D N A was extracted (§2.3.1) from serial overnight cultures (§2.1.1) of these isolates and stored (4°C) for later use, as follows. For the analyses summarized in Figures 7-15 and 7-16, aliquots of this RF DNA, as well as of s.s. DNA that had been employed as template for sequencing fDRW5, fDRW613 and fDRW613C, were transformed (§2.5.1) into E. coli K802 (fUSEl, fUSE2, fUSE3, fUSE5, fDRW5), E. coli K37 and K91 (fDRW5), and E. coli LE392 (fDRW5, fDRW613 and fDRW613C). For each of these nine host/vector combinations, three transformants were streaked on LB-Tet and three single colony isolates (one per transformant) selected for analysis; these were cultured overnight (37°C, standing) in 200 uL 65 LB-Tet in microtiter plates. For each isolate (see Figure 2-3 for a flow chart summarizing this and the following procedures) a 20 [iL aliquot of the overnight culture was used to inoculate 43-45 mL LB-Tet (path A in Figure 2-3) while 10 | iL aliquots were used to inoculate four replicate 200 | iL cultures in microtiter plates (path B in Figure 2-3). for each of 9 host/ vector combinations: 3 isolates for each of the 3 isolates: inoculate 200 jiL LB-Tet overnight culture (37'C, standing) - employ as inoculum -inoculate one 43-45 mL culture in 125 mL flask one 43-45 mL overnight culture (37'C, shaking] titre virions produced by pseudorevertants; harvest and analyse RF DNA inoculate four 200 \il cultures in microtiter plate wells four replicate 200 \iL overnight cultures (37°C, standing) titre virions produced by pseudorevertants Fig. 2-3. Flow chart for experiment summarized in Figure 7-15, provided as an accompaniment to §2.9.3, "Comparative analysis of frameshift and amber vector stability". 4 — Exclude data of path A from summary Include data of path A in summary Summarize (Figure 3-14) (a) Path A (Figure 2-3). After 21.5 h incubation, virions within supernatant of each 43-45 mL culture were counted by a full-plate plaque assay (§2.7.2.a) using single samples of 10-fold serial dilutions of culture supernatant; E. coli K37 and K91-Kan were employed to determine total and pseudorevertant virion production respectively (Figure 7-15). A sample of undigested D N A extracted (§2.3.1) from each culture was analysed by agarose gel electrophoresis to confirm that no isolate was an obvious "deletion" mutant (seen previously 66 in certain frameshift vector preparations; see Figures 7-5 to 7-7) and to provide crude approximations of the quantities of RF D N A recovered. (b) Path B (Figure 2-3). After -30 h incubation (standing), virions within culture supernatant of the four replicate 200 [iL cultures were similarly titered by a full-plate plaque assay (§2.7.2.a). These titers were used to assess whether any of the cultures used as inocula for the 43-45 mL cultures (path A) contained a large fraction of pseudorevertants, suggesting that the isolate itself may have been a pseudorevertant. In one case (an isolate of host/vector combination K37/fDRW5), titers for all four replicate 200 [iL cultures were high, -10 9 pfu mL" 1 culture supernatant. Accordingly, results for this isolate were eliminated from the analysis (Figure 7-15) derived from cultures in path A. 2.9.4. Effects of SupF and wild-type pill on pseudorevertant production (a) Assays of pseudorevertant production (Figure 7-16A). S.s. D N A that had been employed as template for sequencing fDRW613, and RF D N A of fDRW5 isolates screened for low pseudorevertant production, were transformed (§2.5.1) into E. coli K802 and K802 recA. CaCl2-competent cells prepared from these transformants and E. coli K91 were transformed with pJB61 (K802 and K802 recA), pNK1759 (K802 and K91) and pNK2859 (K802) and plated on L B + 100 [ig ampicillin mL"1 + 20 |ig tetracycline mL" 1. In separate assays for strains harboring fDRW5 and fDRW613, four isolates (isolated colonies of original transformants, or isolates derived from streak-purification of transformants) of each host/vector/plasmid combination were cultured (37°C, shaking) -24 h in 2 mL L B + 20 [ig tetracycline mL"1 and (for plasmid-harboring strains) 50 |ig ampicillin mL "! A 50 [iL aliquot of each culture was used to inoculate 40 mL (fDRW5) or 50 mL (fDRW613) L B containing 20 [ig tetracycline mL" 1 and (plasmid-harboring strains) 50 |ig ampicillin mL "I After -24 h incubation, virions were enumerated by a full-plate plaque assay (§2.7.2.a) using single samples of 10-fold serial dilutions of culture supernatant. E. coli K91/pNK1759 and K91 were 67 employed as plaquing hosts to measure total virion production and pseudorevertant production, respectively. (b) Confirmation of gill expression by pJB61 (Figure 7-16B). Because production of pi l l prevents superinfection of host cells (Boeke et al. 1982), g i l l expression in strains harboring pJB61 can be confirmed by comparing the abilities of virions to plaque on hosts possessing (plaques not expected) or lacking (plaques expected) pJB61. Accordingly, three serial 10-fold dilutions of fUSE2 (plaques on F-piliated host) and fDRW613 (requires amber-suppressing F-piliated host) expected to yield ~10 -10 plaques on a suitable host were plaqued in a manner similar to full-plate plaque assays (§2.7.2.a) on E. coli K37 (amber-suppressing), K37/pJB61, K91-Kan (not amber-suppressing) and K91-Kan/pJB61, and plaque formation or lack thereof was determined. 2.9.5. Followup analysis of pseudorevertants and comparison of fUSE5 with fDRW5 (Figures 7-24 to 7-26) (a) Initial cultures. Aliquots of fUSE5 s.s. D N A extracted (§2.3.2.a) from Smith-supplied virions, s.s. D N A that had been employed for sequencing fDRW5, fDRW613 and fDRW613C, and fDRW70 RF DNA extracted from cultures screened for low pseudorevertant production was transformed (§2.5.1) into E. coli LE392. For each vector, four transformants (isolates) were cultured (standing) in 200 U.L LB-Tet in microtiter plates. After 24 h each culture was diluted serially by six ten-fold and ten two-fold dilutions (in LB-Tet) in microtiter plates. These were incubated 24 h (standing) before using (for each isolate) 100 \xL of the highest two-fold dilution that showed clear evidence of growth (as determined by eye) to inoculate 2 mL LB-Tet. After 24 h incubation (with shaking) of the 2 mL cultures, (i) 0.5 mL culture supernatant was heated to 70°C (10 min.) and stored at 4°C for later use, and (ii) 1.5 mL culture used to inoculate 48 mL LB-Tet; these -50 mL cultures were incubated 24 h (with shaking). 68 (b) Assay of pseudorevertant production (Figure 7-24). Virions present in culture supernatant of the 2 mL and 50 mL cultures were enumerated by a full-plate plaque assay (§2.7.2.a) using E. coli host strains K91 and K91/pNK1759 and, for each isolate, three samples from a single 10-fold dilution series of culture supernatant. For fUSE5 and fDRW5, each of the four isolates was titered; for the remaining vectors, three of the four isolates were titered. Virions produced during the 24 h incubation of 50 mL cultures were estimated by subtracting, for each isolate, mean titers of 3 samples of the 2 mL cultures from the corresponding titers of the 50 mL cultures (Figure 7-24). (c) Analysis of JVSE5 and fDRW5 RF DNA (Figures 7-24 and 7-25). After RF D N A of each fUSE5 and fDRW5 isolate was extracted (§2.3.1; 4 mL PI, P2, P3) from a serial overnight culture (§2.1.1, 38 mL), one-half of the harvested D N A was treated with Plasmid-Safe ATP-dependent DNase (Epicentre Technologies; this enzyme degrades s.s. and linear d.s. forms of DNA; per isolate: 10 U , 37°C, 0.5 h, in Boehringer restriction endonuclease buffer M supplemented with 1 mM ATP, volume = 49 uL), and heated to 70°C for 15 min.. After subsequent digestion with BamHl (20 U , 1 h, volume=51 \iL) to create linear fragments, each isolate's RF D N A was quantified and examined for "deletion" mutants (seen previously in certain frameshift vector preparations, see Figures 7-5 to 7-7) by agarose gel electrophoresis. (d) Sequencing of pseudorevertants (Figure 7-26). For each vector, three ~2 mm x ~2 mm agar blocks containing plaques were excised from lawns used to titer pseudorevertants. These were incubated overnight (4°C) in 1 mL PBS before using 50 uL of each resulting eluate to infect (§2.5.3) 50 uL E. coli K91-Kan, and subsequently using infected cells to inoculate 50 mL overnight (shaking) cultures. S.s. DNA extracted (§2.3.2.b) from virions harvested from culture supernatants by two-stage PEG precipitation (§2.6.4) was employed as template for sequencing (§2.4) the cloning site and flanking bases of each pseudorevertant. 69 2.10. C O N S T R U C T I O N O F A L I B R A R Y O F V A R I A N T PEPTIDES D E R I V E D F R O M THE P. FALCIPARUMClRCUMSPOROZOITE PROTEIN (CSP-LlBRARY) 2.10.1. Construction of a clone displaying P N A N P N A N P N A In each of two ligations, 10 fmol f(JSE5 RF D N A derived from a culture screened for low pseudorevertant production, digested to apparent completion with Sfil and precipitated with isopropanol to remove "stuffer" fragments, was combined with 30 fmol of an annealed (§2.2.5.b) oligonucleotide pair encoding the circumsporozoite protein (CSP) sequence PNANPNANPNA (Figure 7-8A). These mixtures were ligated (0.5 U T4 ligase, §2.2.2.b) 18 h at 13.5°-17.5°C before electroporating (§2.5.2) one-eighth of the ligation products into E. coli MCI061. After overnight culture (standing) of 96 transformants (48 per ligation) in wells of microtiter plates (200 uL culture"1) virion-producing clones were identified by applying 5 \xL culture supernatant to lawns of E. coli K91-Kan/pACYC184 in a manner similar to pfu "spot" assays (§2.7.2.b). Five of the 50 transformants that produced relatively higher numbers of plaques were selected for further analysis. RF D N A extracted (§2.3.1) from serial overnight cultures (§2.1.1, 50 mL) of these was digested with AlwNl (diagnostic for the target recombinant) to confirm that all possessed the desired insert. Sequencing template was prepared by (i) extracting s.s. D N A in the manner described earlier (§2.3.2.a) from one-stage PEG-precipitated virions (§2.6.3) derived from serial overnight cultures (§2.1.1, 100 mL) of these clones, and (ii) subsequently purifying D N A with a Qiagen Tip 20 D N A extraction/purification kit. For four transformants, this s.s. D N A was employed as sequencing template (§2.4) to confirm that all possessed the desired insert. 2.10.2. Construction of a first CSP-library (a) fDRWS propagation in E. coli K802. After transforming E. coli K802 with an aliquot of fDRW5 s.s. D N A that had been employed in sequencing this vector, transformants 70 were cultured (as a pool) by serial overnight culture (§2.1.1, 450 mL). Virions were harvested from culture supernatant by two-stage PEG precipitation (§2.6.4) and stored (4°C) for later use (§2.10.3.1). RF D N A was extracted from cells using, for each 210 mL culture volume, a Qiagen Tip 100 D N A extraction/purification kit generally in accordance with the manufacturer's recommendations, but with an additional column wash with 1.35 M NaCl (§2.9.l.g). After digesting -20 u.g fDRW5 RF D N A with 200 U Sfil (>1 h; N E B , volume = 200 \xL), digested D N A was purified using a Qiagen Tip 20 D N A extraction/purification kit. (b) Ligation and electroporation. In three ligations, 13 fmol fDRW5 RF D N A digested with Sfil (possibly incompletely: see Figure 7-1 OA) and precipitated with isopropanol to remove "stuffer" fragments was combined with 13, 130 and 1,300 fmol annealed (§2.2.5.b) semidegenerate oligonucleotides encoding variants of the CSP immunodominant repeats (Figure 7-8B). These mixtures were ligated (5 U T4 ligase, §2.2.2.b) 17 h (15°-19°C) before electroporating (§2.5.2) one-eighth of the ligation products into E. coli MCI061. (c) Isolation of virion-producing clones. For each ligation, virions comprising the resulting library were harvested by washing transformants (6 x 104 to 2.5 x 105 cfu) from plates (§2.6.1) and subsequent centrifugation. Virions were recovered by one-stage PEG precipitation (§2.6.3) of wash supernatants. For each library, an aliquot of harvested virions was used to infect (§2.5.3) E. coli K91-Kan and dilutions of infected cells (10° to 10"4) were spread on LB-Tet plates and incubated overnight. For each library, 52 isolates selected from these plates were transferred as short streaks to master "stock" plates and incubated overnight before storing (4°C) for later use as a source of candidate clones for sequencing. (d) Assessment of recombinant fraction. RF D N A extracted (§2.3.1) from serial overnight cultures (§2.1.1) of 12 candidate clones (4 per library) derived as described above was digested with Haell to determine that, based on loss of a 3,965 base pair (bp) fragment and concomitant gain of -2,890 and 1,075 bp fragments, all 12 clones were recombinants. 71 (e) Identification of unique clones by sequencing (Figure 7-11). Twenty-six of 32 possible (Figure 7-8) unique clones and one product of "illegitimate" recombination were identified by sequencing (§2.4) the gi l l inserts of 63 candidate clones. Usually this was done by sequencing s.s. D N A extracted (§2.3.2.b) from one- or two-stage PEG-precipitated virions (§2.6.3, §2.6.4) produced by (1°) transductants (selected from master "stock" plates) in serial overnight cultures (§2.1.1, 30-50 mL). In other cases, such as when sequences were ambiguous, E. coli K802 was transformed with s.s. or RF D N A derived from these 1° transductants or E. coli K91-Kan was infected (§2.5.3) with their virion progeny, and sequencing template was prepared from virions produced by these 2° transformants or transductants in serial overnight cultures (§2.1.1, 30-50 mL). 2.10.3. Unsuccessful construction of a second, proline-biased CSP-library (a) fDRW5 propagation in E. coli K37. After infection (§2.5.3) of 2 mL E. coli K37 with an aliquot of PEG-precipitated fDRW5 virions recovered (§2.10.2.a) from cultures used for propagation of fDRW5 RF DNA that had been employed in successful construction of the first CSP-library, infected cells were transferred to 40 mL fresh medium and incubated 6.5 h before using 20 mL of the resulting cultures to inoculate each of two 900 mL cultures. These were incubated 19 h before harvesting RF D N A using a Qiagen Tip 500 D N A extraction/purification kit. RF D N A was further purified by CsCl density gradient centrifugation by centrifuging recovered D N A (in 3.6 mL TE pH 8.0 with 50% w/w CsCl; 20°C) 19-20 h at 45,000 rpm (194,000 x g m a x ) in a Beckman VTi65 rotor, extracting the bottom most pronounced bands three times with NaCl-saturated butanol, and precipitating extracted D N A with ethanol. That the Sfil restriction sites and amber codon within the Sfil-excisable "stuffer" fragment were intact was confirmed by sequencing (§2.4) with 1.0 |_ig of CsCl-purified RF DNA. (b) Ligation and electroporation. In separate ligations, 17 fmol CsCl-purified fDRW5 72 RF DNA, digested to apparent completion with Sfil and precipitated with isopropanol to remove "stuffer" fragments (see Figure 7-1 OB), was combined with 51 and 170 fmol of an annealed (§2.2.5.b) semidegenerate oligonucleotide pair (encoding variants of the CSP immunodominant repeats, as before) with a bias in the bases (C or G) encoding proline versus alanine (see Figure 7-8) favouring proline over alanine -4:1. These mixtures were ligated (5 U T4 ligase, §2.2.2.b) 20 h at >15°C before electroporating (§2.5.2) one-twentieth of the ligation products and, as a control, -0.5 fmol undigested fDRW5 RF D N A into E. coli MC1061. (c) Isolation and sequencing of virion-producing clones. For each electroporation, virions were harvested from bacterial growth on plates (1.5 x 10 to 2.3 x 10 cfu for ligation products, 4 x 105 cfu for undigested fDRW5 RF DNA) and virion-producing clones were selected and sequenced in a manner similar to that described for the first CSP-library (§2.10.2.c, §2.10.2.e). 2.10.4. Successful construction of a third, proline-biased CSP-library (Table 7-II) (a) Ligation and electroporation. In three separate ligations, 10 fmol fDRW5 RF D N A propagated in E. coli LE392, digested to apparent completion with Sfil and precipitated with isopropanol to eliminate "stuffer" fragments was combined with 0 (control), 30 and 90 fmol (i.e., molar insertvector ratios of 0:1, 3:1 and 9:1) of an annealed (§2.2.5.b) semidegenerate oligonucleotide pair prepared with a bias in the bases (C or G) encoding proline versus alanine (§2.10.3.b). These vector/insert mixtures were ligated (5 U T4 ligase, §2.2.2.b) 18 h at 15°-19°C before electroporating (§2.5.2) the following into E. coli MCI061: (i), (ii) and (iii), one-tenth of the products of ligations with 0:1, 3:1 and 9:1 molar insertvector ratios respectively; (iv) one-twentieth of the products of the ligation with a 9:1 molar insert:vector ratio combined with 0.13 fmol undigested fDRW5 RF DNA; and (v) 0.13 fmol undigested fDRW5 RF DNA. (b) Isolation and sequencing of virion-producing unique clones. For each electroporation, virions comprising the resulting library were harvested by washing 73 transformants from plates and PEG precipitating wash supernatant essentially as described for the first CSP-library (§2.10.2.c). Virion yields were estimated by a pfu "spot" assay (§2.7.2.b) using duplicate -5 uL samples of 100-fold serial dilutions of wash supernatant applied to lawns of E. coli K91-Kan. Virion-producing recombinants were isolated essentially as described earlier (§2.10.2.c). RF D N A extracted (§2.3.1) from serial overnight cultures (§2.1.1) of six candidate clones was digested with Haell to determine that, based on loss of a 3,965 bp fragment and concomitant gain of -2,890 and 1,075 bp fragments, all six candidate clones were recombinants. Four previously unidentified recombinant clones were identified by sequencing the inserts into gi l l of 19 candidate clones. MW. pASlOO Sau3AI digest 4072 3054 2036 1636 1018 517, 506 tt m Fig. 2-4. Partial digest of pASlOO with Sau3Al. Agarose gel shows samples from one of two sets of partial digests: -5.5 u.g samples of pASlOO were incubated (0.5 h, 60 u.L volumes) with (left to right) 10, 7, 5, 3.4, 2.4, 1.7, 1.2, 0.8, 0.6 and 0.4 U Sau3M. 2.11. B. PERTUSSIS FHAB RESTRICTION FRAGMENT LIBRARIES 2.11.1. Construction and analysis of B. pertussis fhaB Saw3AI fragment (FHA-S) libraries (a) pASlOO propagation and partial digestion with Sau3A7. pASlOO (Table 2-II) D N A was extracted from E. coli DH5oc/pAS100 (a gift from A. Siebers) cultured to OD 6 0 0=0.5 in 500 mL L B containing 100 p,g ampicillin mL"1 using a Qiagen Tip 500 D N A 74 extraction/purification kit. Two sets of partial digests of pASlOO were performed. In the first, ten 5.5 [ig samples (60 [iL volumes) of pASlOO were digested 0.5 h with two-fold serial dilutions of Sau3Al (10 to 0.2 U). In the second set, ten 5.5 |ig samples (in 60 | iL volumes) were digested 0.5 h with a second set of 10 serial dilutions of Sau3Al (10, 7, 4.9, 0.4 U). After analysis (Figure 2-4) by 1.5% agarose gel electrophoresis, digests were pooled and fragments separated on a preparative scale by 1.5% agarose gel electrophoresis (2 V cm"1, 4-4.5 h). Gel bands containing fragments of -1300-2300 bp were excised and placed in sections of dialysis tubing. Fragments were electroeluted (Sambrook et al. 1989) onto dialysis membrane by placing the closed tubing under TAE buffer in an electrophoresis chamber and applying 2.5 V cm"1 for -2 h. After the current was reversed for 2 min., the fluid contents of the tubing together with 5 mL T A E washes were filtered through 0.2 [iM Acrodisc filters and extracted repeatedly with sec-butanol until the sample volume was reduced to -3.2 mL. Samples were subsequently ethanol precipitated and resuspended in TE pH 8 before being additionally purified by Select-D G-50 (5 Prime -> 3 Prime) spin columns and concentrated by ethanol precipitation. (b) Ligation and electroporation. In four separate ligations (I, II, III and IV), 16 fmol of an equimolar mixture of fDRW20, fDRW21 and fDRW22 (Figure 7-12) RF D N A digested to apparent completion with Bglll was combined with 21, 105, 560 and 2560 ng (roughly corresponding to 1:1, 5:1, 25:1 and 125:1 molar insert:vector ratios) of size-fractioned (1,300-2,300 bp) Sau3Al fragments of pAS 100 (§2.11.1.a). These mixtures were ligated (10 U T4 ligase, §2.2.2.b) overnight at >14°C before electroporating (§2.5.2) one-eighth of the ligation products into E. coli MCI061. (c) Initial characterization of virion production and virion-producing clones. For each resulting library (I, II, III and IV), 96 of the 1.4 x 105 to 1.8 x 105 transformants (Table 7-III) were cultured overnight in 200 [iL LB-Tet in microtiter plate wells. For each of the 384 75 transformants so cultured, 5 uL of culture supernatant was applied to lawns of E. coli K91-Kan/pACYC184 in a manner similar to pfu "spot" assays (§2.7.2.b) to identify and characterize virion-producing clones. Twenty-four transformants (isolates) that produced a relatively diminished number of virions (<2 x 104 pfu mL"1 culture supernatant, explained in Table 7-III footnotes), together with a transformant that produced >2 x 104 pfu mL" 1 and two that produced no detectable plaques (<500 pfu mL"1), were selected for further analysis. After streaking on LB-Tet plates, a single-colony isolate of each was used as a source of inoculum for serial overnight culture (§2.1.1, 50 mL). RF D N A extracted (§2.3.1) from these cultures was digested with EcoKl, EcoKV and BamHl and analysed by 0.6% agarose gel electrophoresis (i) to show that, except for 6 of the clones that produced a diminished number of virions, all of the 26 clones selected for analysis could be identified as recombinants, and (ii) to quantify RF D N A for use as sequencing template. (d) Initial assessment of fl and library virion migration in agarose gels (Figure 7-13). To assess virion migration (Griess et al. 1990) in agarose gels (Figure 7-13A), 32, 6.4, 1.3 and 0.3 fig samples of CsCl-purified (§2.6.5) f l virions were electrophoresed and stained with Coomassie Blue (§2.7.3). In a followup experiment (Figure 7-13B), 6 and 0.6 [ig samples of PEG-precipitated virions of each of the four FHA-S libraries were similarly electrophoresed and stained. (e) Assessment by agarose gel electrophoresis of virion production by selected clones. Single colony isolates of six clones that produced fewer than 2 x 104 pfu mL" 1 culture supernatant (a "diminished number of virions", §2.11.1.c) and one clone that produced >2 x 104 pfu mL"1 were cultured by serial overnight culture (§2.1.1, 50 mL). Separately, "stock" cultures of seven CSP-library clones (§2.10) were similarly used as inocula for serial overnight culture (§2.1.1, 40 mL). Virions of the 7 FHA-S library clones were harvested from culture supernatants by one-stage PEG precipitation (§2.6.3) and resuspended in 0.01 volume TBS. 76 The relative quantities of virions (as physical particles) within 13 uL samples of these putative 100-fold concentrated preparations and within 13 uL unconcentrated culture supernatant of two CSP-library clones were compared by examining Coomassie-stained agarose gels of electrophoresed virions (§2.7.3). (f) Attempted enrichment for virion-producing clones by agarose gel electrophoresis. Three samples (6, 6 and 40 fig) of PEG-precipitated virions of library I (§2.11.1.e, Table 7-III) in sample loading buffer (§2.7.3) were loaded into the two sample and single preparative wells, respectively, of a 0.7% low melting point (LMP) agarose gel cast with a preparative comb, and electrophoresed ~4 h at 2 V cm-1 before staining the sample lanes with Coomassie Blue (§2.7.3). A 0.9 g agarose section of the unstained preparative lane, corresponding to the region between the trailing edge of the fast-running monophage band (Figure 7-13) and the leading edge of the fastest-running putative polyphage band, was excised and heated at 70°C for 15 min. (to melt LMP agarose; virions are relatively insensitive to heating: see Salivar et al. 1964) in 2 mL L B before cooling at 37°C for 10 min. (LMP agarose remains fluid at this temperature) and subsequently adding 2 mL E. coli K91-Kan ( O D 6 0 0 = 0.75). After 20 min. (37°C) the agarose/virion/cell mixture was spread on LB-Tet and incubated overnight (37°C). Virions were harvested by washing transductants from plates (§2.6.1) and subsequent two-stage PEG precipitation (§2.6.4) of wash supernatant. The effect of this enrichment method was examined by agarose gel electrophoresis and subsequent Coomassie staining (§2.7.3) of 6 and 0.6 |ig samples of these "size-enriched" library I virions, together with 6 and 0.6 |ig samples of unenriched library I virions. 2.11.2. Construction and analysis of B. pertussis fhaB HhallHinY11/Hpall restriction fragment (FHA-H/H/H) libraries (Figures 7-17 to 7-22) Methods of library construction described below were intended to allow the construction of a set of clones producing virions displaying an assortment (in both amino acid content and 77 peptide length) of peptides derived from the B. pertussis fhaB gene, i. e., a "library" of diverse peptides. Certain methods reflect a goal of eliminating non-recombinants without regard to the effect of these methods on library content; viz., not all possible peptides needed to be represented. (a) fUSEl propagation. After E. coli K802 was transformed with fUSEl RF D N A extracted (§2.3.1) from a culture of an isolate previously screened for low pseudorevertant production, serial overnight cultures (§2.1.1, 40 mL) were prepared for three transformants. Fifty-[iL samples (one per transformant) of ten-fold serial dilutions (100-10"4) of culture supernatants were assayed for production of pseudorevertant, using a full-plate plaque assay (§2.7.2.a); no plaques were detected (limit of detection = 20 pfu mL"1 culture supernatant) for the three transformants. For each transformant, ~1 [ig RF D N A extracted (§2.3.1) from the 40 mL culture was incubated with 5 U Mung Bean Nuclease (NEB; in 10 mM Tris, 10 m M M g C l 2 , 100 m M NaCl , 1 m M DTT, 1 m M ZnCl 2 ; pH 7.9; 70 [iL volume) for 30 min. at 30°C to digest s.s. species in the preparation and, after recovering D N A by phenol/chloroform extraction and ethanol precipitation, each sample was digested with 10 U Pvull (2 h). In spite of an excess of Pvull (10 U per [ig DNA), each preparation showed evidence of an incomplete digest as judged by agarose gel electrophoresis. Accordingly, samples were pooled and the -3 [ig RF D N A digested again with 100 U PvwII (2 h, 300 [iL volume) before recovering D N A by phenol/chloroform extraction and ethanol precipitation. (b) pASlOO propagation and generation of blunt-end fragments. Plasmid pASlOO, carrying the 10 kbp EcoRl fragment of B. pertussis fhaB (see Table 2-II), was transformed (§2.5.1) into E. coli K802, transformants were plated on L B with 100 [ig ampicillin mL" 1 , and plasmid D N A extracted from serial overnight cultures (§2.1.1, 125 mL L B + 50 [ig ampicillin mL"1) of each of two transformants. For each of Hhal, HinPll and Hpall, -10 [ig 78 of pooled pASlOO D N A was digested 1 h with -30 U restriction endonuclease before recovering D N A by phenol/chloroform extraction and ethanol precipitation. For each digest, -5-10 fig D N A was incubated with 10 U Mung Bean Nuclease (NEB) (30°C, 1 h, in 50 mM sodium acetate, 30 m M NaCl, 1 mM ZnS0 4 , pH 5; volume = 200 uL) to remove 3' and 5' overhangs. As well, -1 |ig of each of the HinPll and Hpall digests was incubated with 5 U E. coli D N A Polymerase I large (Klenow) fragment (NEB) (15 min., 20°-25°C, in 10 m M Tris, 5 m M MgCl 2 , 7.5 mM DTT, 0.2 mM dNTPs, pH 7.5, volume = 60 uL), to fill in 3' recessed ends. DNA in each sample of Mung bean nuclease (MBN)- and Klenow-treated fragments was recovered by phenol/chloroform extraction and ethanol precipitation. (c) Primary libraries: ligation, re-digestion with Pvu/7 and transformation of E . coli K802. Five libraries were constructed: libraries I, II and III, with MBN-treated Hhal, HinPll and Hpall fragments, respectively; and libraries IV and V , with Klenow-treated HinPll and Hpall fragments. For each library, ligations (§2.2.2.b) with 3:1 and 9:1 molar insert:vector ratios (per ligation: -75 or -225 fmol restriction fragments, 25 fmol fUSEl , 1.8 U T4 ligase) were carried out for 21 h at 13.5°-17.5°C. Ligation products were heated 10 min. at 70°C to inactivate T4 ligase and digested with 10 U Pvull (volume = 200 \xL) to diminish non-recombinants. For each library, one-fourth of the ligation products were pooled and transformed (§2.5.1) into E. coli K802. For each transformation, 5 x 10 to 2 x 10 transformants were recovered on LB-Tet plates. (d) Secondary libraries: re-digestion of DNA harvested from transformants with Pvull and subsequent transformation of E. coli K802. For each primary library, transformants were washed from plates (§2.6.1) and RF DNA harvested (§2.3.1) from these washings was digested with an apparent excess of Pvull (26 U , 2 h, volume = 30 uL) to diminish non-recombinants. Products of this digestion were transformed (§2.5.1) into E. coli K802. For each transformation, 4 x 104 to 4 x 105 transformants were recovered on LB-Tet plates. 79 (e) Screening for virion production by a TU assay (Figure 7-17). Twenty-four transformants from each secondary library were cultured overnight in 1 mL LB-Tet and the numbers of virions produced estimated by a TU "spot" assay (§2.7.2.c) using a single sample for each 10-fold serial dilution (10"' to 10"4) of culture supernatant. Each of the 45 clones that produced a detectable number of virions (limit of detection = 104 T U mL" 1 culture supernatant) was streaked on LB-Tet and incubated overnight before transferring individual colonies as short streaks to a "master" plate (LB-Tet). After overnight incubation, this master plate was stored at 4°C and used as a source of inocula for subsequent experiments. (f) Extraction of RF DNA from, characterization of virion production by (Figure 7-18), and preparation of PEG-precipitated virion stocks of virion-producing clones. Each of 45 virion-producing clones identified in Figure 7-17 was cultured 36 h (Experiment 1) or 24 h (Experiment 2) in 1 mL LB-Tet. One mL (Experiment 1) or 0.1 mL (Experiment 2) of the resulting cultures was diluted into 40 mL LB-Tet and incubated 20 h. RF D N A extracted (§2.3.1; 3 mL PI, P2, P3) from cultures for Experiment 1 was stored at 4°C for later analysis (§2.11.2.7.a, §2.11.3.1). The numbers of virions produced by the 45 clones (Experiments 1 and 2) were estimated by a TU "spot" assay (§2.7.2.c) using a single sample for each 5-fold serial dilution (5° to 5"") of culture supernatant. For Experiment 1, virions were harvested from ~30 mL culture supernatant by two-stage PEG precipitation (§2.6.4) and stored at 4°C for later use. (g) Determination of insert size by electrophoresis of HaeHI fragments. Haelll digestion of fUSEl produces 8 fragments (352 to 2528 bp) including an 850 bp fragment which contains the Pvull cloning site. With the goal of determining the sizes of inserts on the basis of changes in size of the 850 bp fragment, RF D N A of virion-producing clones (§2.11.2.f) was digested with Haelll (3-5.25 h) and restriction fragments were analysed by 1.7% agarose gel electrophoresis. Only 17 clones produced restriction patterns that could 80 reasonably be analysed; for the remaining clones, D N A preparations were of poor quality (a "smear" of high- to low-molecular weight species was pronounced, unexpected species were present) or of low yield, or no new fragments could be discerned within the range of molecular weight markers employed and under the electrophoresis conditions employed. For each of these 17 clones, sizes of fragments containing inserts were estimated using the relative mobility (Rf) of these fragments compared to the Rf of known fragments, by means of a least squares regression of a fourth order polynomial (see legend to Figure 2-5). Standard curves based on this regression analysis are shown in Figure 2-5. 3.5T I 3.5 Rf Fig. 2-5. Relative mobility (Rf) of Haelll fragments of F H A - H / H / H library clones. Figure is derived from agarose gel electrophoresis of Haelll restriction fragments of 17 virion-producing clones (§2.11.2 .g) . Clones are shown in groups; in each group a common range of known fragment sizes was used to derive a fourth order polynomial employed in estimating sizes of pASlOO-derived inserts: log(L) = a0+a, R f +a 2 R f 2 +a 3 R f 3 +a 4 R f 4 where L = length of fragment in base pairs; Rf= relative mobility; a 0 , a l , a 2 , a 3 and a 4 are constants. (h) Determination of insert size by sequencing. Nine virion-producing clones were selected for sequencing. For each, PEG-precipitated virions (§2.11.2.f) were amplified (§2.5.3) in E. coli K91-Kan, and s.s. D N A was extracted (§2.3.2.b) from virions harvested from the resulting culture supernatant by two-stage PEG precipitation (§2.6.4). Of the nine samples, only five produced s.s. DNA of sufficient quantity (this correlated roughly with levels of virion production) or quality to be successfully employed as sequencing template (§2.4). 81 Time (minutes) Fig. 2-6. Host cell doubling times (T d) for F H A - H / H / H library clones. Culture conditions and other methodology are described in §2.11.2.i. Shaded bands highlight a common range of O D 6 0 0 values employed in calculating T d values for each of nine clones. T d , and T d 2 are estimates for duplicate cultures. (i) Relationship between virion production and host cell doubling time (Figure 7-20). For 12 clones, samples of duplicate 20 h cultures (1 mL LB-Tet) inoculated from a "master" plate (§2.11.2.e) were diluted into 40 mL fresh medium to yield a starting OD 6 0 0~0.006; these cultures were incubated with shaking (175 rpm) and O D 6 0 0 measured at -30 min. intervals using 1 mL samples (discarded) in 1 cm cuvettes. Doubling time (T d) for each duplicate was calculated on the basis of a least squares linear regression of log(OD 6 0 0 ) versus time for three data points with O D 6 0 0 values between 0.04 and 0.3 and with log(OD 6 0 0 ) versus time roughly forming a straight line (Figure 2-6). In a separate assay, duplicate small volume overnight cultures of the 12 clones were diluted 1:400 into 40 mL fresh medium (LB-Tet) and incubated overnight. The numbers of virions produced by these clones were estimated by a T U "spot" assay (§2.7.2.c) using a single sample for each 5-fold serial dilution (5° to 5"11) of supernatant of each duplicate culture. 2.11.3. Followup study of FHA-H/H/H library clones (Figures 7-21 and 7-22) Figure 2-7 provides an overview of the methods described in the following sub-sections. 82 virion-producing library transformant I transfer to "master" plate I "original library member" I initial 40 mL culture cells I extract RF I RF I transform E. coli K802 I sefecf 3 transformants I • transfer to "master" plate culture supernatant precipitate virions PEG-preci pita ted virions infect E.coliK9l sefecf 3 transductants transfer to "master" plate three 1 clones three 2 clones assay each for virion production sequence 4-1or2 (for selected library members) assay each for virion production —• sequence tori (for selected library members) Fig. 2-7. Flow chart for followup study of F H A - H / H / H library clones, provided as an accompaniment to §2.11.3, the text of which is summarized in B. Origin of the material used to recover the "original" library members and derive sub-clones from their virion progeny is described in §2.11.2.f ("Extraction of R F D N A from ....") and is outlined in A. (a) Recovery of "original" library members and derivation of sub-clones from their virion progeny. Forty "original" library members were recovered by transforming (§2.5.1) E. coli K802 with RF DNA extracted from the initial cultures of virion-producing FHA-H/H/H library clones (§2.11.2.f); for each of the 40 transformations, three transformants ("1° clones") were transferred as short streaks to "master" plates (LB-Tet), incubated overnight and stored (4°C) for later use. Sub-clones of these "original" library members were derived by infecting (§2.5.3) 100 [iL E. coli K91-Kan with small aliquots of PEG-precipitated virions harvested from the same initial cultures (§2.11.2.f), and spreading serial 10-fold dilutions of infected cells on LB-Tet plates. After overnight incubation, three transductants ("2° clones") of each library member were transferred to "master" plates (LB-Tet), incubated overnight and stored (4°C) for 83 later use. As required, master plates were sub-cultured onto fresh LB-Tet plates and again stored (4°C) after overnight incubation. (b) Comparison of virion production by 1° and 2° clones (Figure 7-21). For each of three 1° clones (in E. coli K802) and three 2° clones (in E. coli K91) of the 40 library members, sterile toothpicks were touched to "master" plates and used to inoculate 2 mL LB-Tet cultures. After 18 hours incubation, virions produced in these cultures were estimated by a T U "spot" assay (§2.7.2.e) using a single sample for each 10-fold serial dilution (10"' to 10"6) of culture supernatant. (c) Sequencing of selected 1° and 2° clones (Figure 7-22). Nine library members were selected for sequencing. For four of these, virion titers of the 1° clones were substantially lower than those of the corresponding 2° clones. For the remaining five, titers of the 1° and 2° clones were similar. For the former group, attempts were made to sequence two 1 ° and two 2° clones; for the latter group, a single 1 ° and single 2° clone. Two methods were employed to extract RF D N A from these clones for use as sequencing template. In the first of these, employed for 26 preparations, RF DNA was extracted from 18 mL of serial overnight cultures (§2.1.1, 35-40 mL) inoculated from master plates, using Nucleobond AX-20 columns according to the manufacturer's recommended protocol for a three-fold scale-up of reagents. Extracted DNA was treated with ATP-dependant DNase (see §2.9.5.c; per preparation: 10 U , 37°C, 1 h, in manufacturer-supplied buffer supplemented with 1 m M ATP, volume=405 uL), incubated 20 min. at 70°C, and again purified (to remove enzyme and degradation products) with Nucleobond AX-20 columns. In the second method, employed for 25 preparations, RF D N A was extracted using Qiagen reagents (§2.3.1; 4 mL PI, P2, P3). After treating extracted D N A with ATP-dependant DNase (see §2.9.5.c; 5 U , 37°C, 4.75 h, volume=12.5 uL), D N A was recovered from reaction mixtures by phenol/chloroform extraction and ethanol precipitation. Sequencing (§2.4, using primer A for most samples, primer B for selected samples) results 84 were generally poor and repeated efforts to sequence certain clones were unsuccessful. Of 51 sequencing samples, only 16 (30%) provided sequencing chromatograms that were marginally acceptable or better. Limited sequencing data that included sequences for 1° clones were obtained for only five of the nine library members selected for analysis. 2.12. C O N S T R U C T I O N O F B. PERTUSSIS FHAB D N A S E I F R A G M E N T LIBRARIES 2.12.1. Construction and analysis of FHA series 70 libraries (Table 7-V) (a) JDRW70 propagation. Four samples of RF D N A extracted from cultures that had been used to prepare sequencing template during fDRW70 vector construction were transformed (§2.5.1) into E. coli LE392 and for each sample a single transformant was used as a source of inoculum for a serial overnight culture (§2.1.1, 30 mL). Samples of RF D N A extracted (§2.3.1) from each of these cultures were digested ~2 h with Fspl, Pvull and Xbal (separately) and analysed by agarose gel electrophoresis to confirm that the Fspl and Pvull sites flanking the excisable stuffer fragment and amber codon (Xbal used for diagnosis) within the stuffer fragment (Figure 7-23A) were intact. Larger-scale samples (~10 fig) were doubly digested 3.25 h with Fspl and Pvull (50 U each, 100 \xL total volume), phenol/chloroform extracted and isopropanol precipitated (to remove "stuffer" fragments); these preparations were pooled and used in library construction, as below. (b) Ligation, electroporation and harvesting of virions (Table 7-V). Four libraries were constructed with DNase I-generated fragments (Anderson 1981; Smith 1992) of the 10 kbp B. pertussis fhaB EcoRl restriction fragment (see Table 2-II, entry for pASlOO). These fragments (a gift from A. Siebers) had been size-fractioned (extracted from polyacrylamide gels after electrophoresis, by A . Siebers) into groups of 30-75 bp (used for library 70-A), 75-150 bp (70-B), 150-300 bp (70-C) and 300-600 bp (library 70-D). For each library, two ligations were performed, each with 30 fmol fDRW70 RF D N A and -90 fmol of B. pertussis fhaB fragments (thus roughly 3:1 molar insert:vector ratios). Two control ligations (70-X and 70-Y) 85 were also carried out, each consisting of 30 fmol fDRW70 RF D N A alone (i.e., no fhaB fragments were included). After ligating (2 U T4 ligase, §2.2.2.b) each vector/insert mixture or control 33 h at >13°C, one-tenth (first ligation for each library) or one-twentieth (second ligation) of the ligation products were electroporated (§2.5.2) into E. coli. For each ligation, the number of transformants was estimated by counting colonies on a single LB-Tet plate spread with a 10 uL sample (Table 7-V). After washing transformants from plates (§2.6.1) and subsequent centrifugation of the washings, virion yields were estimated by full-plate plaque assays (§2.7.2.a) using single samples of 100-fold serial dilutions of wash supernatant (Table 7-V). Virions were subsequently harvested by two-stage PEG precipitation (§2.6.4) of wash supernatant. (c) Library evaluation (Figures 7-27 and 7-28). To identify virion-producing transformants by a tetracycline-resistance transduction assay, toothpicks were used to transfer growth from colonies of 24 transformants derived from each ligation for libraries 70-A, 70-B, 70-C, 70-D and 70-X to 20 uL E. coli K91-Kan ( O D 6 0 0 ~ 1.0) in L B + 0.2 |ig tetracycline mL" 1 + 100 fig kanamycin mL"1 (to inhibit growth of E. coli MCI061) in wells of 96-well microtiter plates. After 0.5 h incubation (standing), 180 [_iL L B + 20 \xg tetracycline mL"1 + 100 fig kanamycin mL" 1 was added to each well and plates were incubated 14 h (standing) before transferring 10 uL of the subsequent culture to 190 uL fresh medium in microtiter plates. After an additional 25 h incubation (standing), the optical density at 595 nm ( O D 5 9 5 ) was determined for each culture using a Biorad model 3550 microplate reader (Figure 7-27). Twenty four selected clones (typically, those with the greatest O D 5 9 5 values; see Figure 7-27) were cultured by serial overnight culture (§2.1.1, 30 mL). S.s. D N A extracted (§2.3.2.b) from virions harvested by two-stage PEG precipitation (§2.6.4) of culture supernatants was used as sequencing template (§2.4) to determine that all of the selected clones were (unexpectedly) of a novel class of pseudorevertant (Figure 7-28). 86 2.12.2. Construction of FHA series 80 libraries (Table 7-VI) (a) fDRW8nn propagation. fDRW836, fDRW861, fDRW863, fDRW864 and fDRW867 (Figure 7-29) extracted from cultures used to prepare sequencing template for these vectors (§2.8.7) was digested to apparent completion with Smal and treated with Mung Bean Nuclease (NEB; 1 U per fag DNA, 30°C, 30 min.) to eliminate s.s. DNA, before recovering D N A by phenol/chloroform extraction and ethanol precipitation. (b) Two-stage ligations. Four libraries (80-A to -D) were constructed, each with a different size-fractionated set of DNase I-generated fragments of the 10 kbp B. pertussis fhaB EcoBl restriction fragment (§2.12.1.b). For each library, three ligations were performed, each with 21 fmol of an equimolar mixture of five fDRW8/z/z vectors (§2.12.2.a) and 63 fmol, 210 fmol or 1.1 pmol (roughly 3:1, 10:1 and 50:1 molar insert:vector ratios) of B. pertussis fhaB fragments. Two control ligations (80-X and 80-Y) were also performed, each with 21 fmol of fDRW8nn RF D N A alone (i.e., no fhaB fragments were included). After ligating (3 U T4 ligase, §2.2.2.b) each vector/insert mixture or control 12 h at 16.5°-19°C, and with the goal of diluting each mixture to favour circularization of ligation products, an additional 3 U T4 ligase was added and the volume adjusted to 150 uL. After an additional 14 h incubation, ligation products for the three insert:vector ratios were pooled, yielding a single pool for each of libraries 80-A, -B, -C and -D. (c) Srf/ digestion of ligation products. To reduce the numbers of non-recombinants, ligation products for each library (80-A, 80-Y) were incubated 10 min. at 65°C to inactivate T4 ligase, precipitated with isopropanol (with 5 fig tRNA as carrier) and digested (except library 80-X) 1.5 h with 4 U Sffi before incubating 20 min. at 65°C to inactivate Srfl, ethanol precipitating and resuspending in 7 pL water. (d) Electroporation and harvesting of virions. Two electroporations were carried out for each library (80-A, 80-Y). In each, one-half (3.5 (iL) of the ligation products were 87 electroporated (§2.5.2) into E. coli MC1061. For each ligation, the number of transformants was estimated by counting colonies on a single LB-Tet plate spread with a single 10 \xL sample; values are reported in Table 7-VI. To confirm that these estimates were reasonable, 16-32 cm 2 areas of plates (245 mm x 245 mm NUNC culture dishes) spread with ~2 mL (90%) of transformants were also counted. For each library, virions were harvested by washing transformants from plates (§2.6.1) and subsequent two-stage PEG precipitation (§2.6.4) of wash supernatant; yields (as physical particles) were estimated by U V spectroscopy (§2.7.1, Table 7-VI). 2.13. ANTIBODIES AND I M M U N O L O G I C A L R E A G E N T S 2.13.1. Rabbit pAbs produced against wild-type phage fl Rabbit a-fl polyclonal antibodies were obtained from New Zealand White rabbits. Each of two rabbits was immunized with 0.1 mg (as determined by a B C A protein assay, similar to that described later, §2.16.1.2) CsCl-purified (§2.6.5) f l virions delivered with RIBI adjuvant via intraperitoneal, intramuscular, intradermal and subcutaneous routes, following the adjuvant manufacturer's recommended protocol; this immunization was repeated four weeks later. After an additional two weeks, rabbits were sacrificed and antibodies purified as described below. After centrifuging (3000 x g m a x , 4°C, 30 min.) 11 mL serum, selected serum proteins were precipitated from the resulting supernatant by slowly adding with stirring 0.5 volume saturated ( N H 4 ) 2 S 0 4 (76.1% w/w in H 2 0 ) and storing the resulting mixture >24 h at 4°C. After centrifugation as before to remove the precipitate, antibodies within supernatant were precipitated by adding 1.0 volume saturated ( N H 4 ) 2 S 0 4 and storing the resulting mixture overnight at 4°C. After centrifugation as before, the antibody-containing precipitate was resuspended in 4 mL PBS; this was dialysed against PBS (1 L, 4°C) three times over a 24 hour period. In each of two preparations, half of the dialysed sample was purified by Protein A affinity chromatography using a prepacked column (Pierce Immunochemicals, binding capacity 88 = 34 mg IgG) following the manufacturer's recommended procedures for antibody purification and column regeneration. Recovered antibodies were concentrated by precipitation with saturated (NH 4 ) 2 S0 4 . After centrifugation, the antibody-containing precipitate was resuspended in 4.2 mL PBS; this was dialysed three times against PBS (1.5 L , 4°C) before adding NaN 3 to a final concentration of 0.02% and storing (as 50 pL and 250 pL aliquots) at -20°C. This final preparation (~2.2-fold less volume than the starting serum) contained (estimated by A280; Harlow & Lane 1988) 4.7 mg IgG mL" 1. 2.13.2. Mouse mAbs produced against P. falciparum CSP Pf4C11.6 (IgG2a), PHG3.4 (IgGl), P f5Cl . l (IgGl), Pf2A10 (IgG2a), Pf5G5.3 (IgGl), PflB2.2 (IgGl), P54A4.1 (IgG3), Pf2Fl.l (IgM), PvNSV3 (IgG) and Pb4B10 (IgG), made and screened against P. falciparum (Pf), P. vivax (Pv) and P. berghei (Pb) sporozoites, recognize CS protein (Wirtz et al. 1987; Burkot et al. 1991; Wirtz et al. 1991; R. A. Wirtz, unpublished data) and were provided by R. A. Wirtz, Walter Reed Army Institute of Research (WRAIR). P f 2 F l . l was used as unpurified ascitic fluid; other mAbs had been purified by Protein A affinity chromatography. Pf2A10 and PflB2.2 were biotinylated by Kirkegaard & Perry Laboratories, Inc., Gaithersburg, M D . (a) Confirmation and determination of mAb isotypes. For each of affinity-purified Pf2A10, PHB2.2, Pf4C11.6, Pf5G5.3, PHG3.4, P f5Cl . l and Pf5A4.1 and for Pf2F l . l ascitic fluid, ten wells of an Immulon-1 microtiter plate were coated with 100 ng affinity-purified mAb or 0.5 pL Pf2Fl . l ascitic fluid (in PBS, 100 uL well"1). After 1 h incubation at 37°C, plates were washed three times with wash buffer (PBS + 0.05% Tween 20). Each a-CSP antibody was probed with each of six goat a-mouse isotype mAbs (a-IgGj, a-IgG 2 A , a-IgG 2 B , a-IgG3, a-IgM and a-IgA; Sigma isotyping kit ISO-2) by adding 100 pL of a 1:3,000 dilution (in wash buffer) of each goat a-mouse mAb to duplicate wells. After 0.5 h incubation at room temperature, plates were washed three times before addition of peroxidase-conjugated mouse 89 a-goat IgG 2° antibodies (1:5,000 dilution in wash buffer, 100 uL well" 1, §2.13.4). After 15 min., plates were washed four times before developing with substrate (OPD, §2.13.5) and reading colored reaction products as outlined later (§2.16.2.a). Wells with the greatest A 4 9 0 values were used to confirm (see references cited in preceding section) or determine that isotypes are: Pf2A10, IgG 2 A ; PflB2.2, IgG 3; P f2F l . l , IgM; Pf4C11.6, I g G 2 A ; Pf5G5.3, IgG,; PflG3.4, IgG,; P f 5 C l . l , IgG,; Pf5A4.1, IgG 3 . (b) Quantification of PJ2F1.1. The concentration of IgM in Pf2Fl . l ascitic fluid was estimated by ELISA using purified IgM (Sigma M-5170) as a standard: 5-fold serial dilutions of PfZFl. l and the IgM standard were coated onto wells of Immulon 3 plates and probed with peroxidase-conjugated goat cc-mouse IgM mAbs (Sigma) using OPD (§2.13.5) as substrate in a manner similar to that described later (§2.16.2.1). 2.13.3. Rabbit pAbs produced against B. pertussis FHA Three rabbit a-FHA sera were a gift from A. Siebers. These had been prepared in New Zealand White rabbits using two preparations of FHA purified by heparin-sepharose affinity chromatography (Menozzi et al. 1991). One such preparation, of so-called "native" FHA, had been eluted from the heparin-sepharose column using a NaCl gradient according to the published (Menozzi et al. 1991) procedure. A second preparation, of so-called "SDS-denatured" FHA, could be removed from the heparin-sepharose column only by use of sodium dodecyl sulfate (A. S. and B. Finlay, unpublished). Sera from two rabbits immunized with "native" FHA are identified as "FN 1/4" and "FN2/4". Serum from a third rabbit immunized with "SDS-denatured" FHA is identified as "FS1/4". Following a recommendation by A.S. (based on the sera's recognition of E. coli expressing B. pertussis _/7zaB-derived fragments using a Pseudomonas aeruginosa OprF expression system), sera FN2/4 and FS1/4 were employed for most work described here. 90 (a) a-FHA antibody purification and biotinylation for biopanning and plaque lifts. Antibodies employed for biopanning FHA-70 and -80 libraries and subsequent plaque lifts were purified as follows. For each of FN2/4 and FS1/4, -950 uL serum was centrifuged 30 min. at 3000 x g m a x in an Eppendorf 5415C microfuge; selected serum proteins were precipitated by diluting the resulting supernatant to 3.8 mL in PBS + 0.02% N a N 3 and slowly adding, with stirring, 1.9 mL saturated (NH 4 ) 2 S0 4 and storing the resulting mixture overnight at 4°C. After centrifugation to remove the precipitate, antibodies were precipitated from supernatant by adding an additional 1.9 mL saturated ( N H 4 ) 2 S 0 4 and storing the resulting mixture overnight at 4°C. After centrifugation, the antibody-containing precipitate was resuspended in 750 [iL PBS and dialysed against PBS (400 mL, 4°C) three times over a -24 h period. Each dialysed sample was absorbed against E. coli antigens using a Pierce Immunochemicals immobilized E. coli lysate kit in accordance with the manufacturer's recommended procedures. One-half of each E. co/z-absorbed serum was purified by Protein A affinity chromatography using a prepacked column (Pierce Immunochemicals, binding capacity = 34 mg IgG) following the manufacturer's recommended procedures. The remaining volume of each E. co//-absorbed serum and the entire volume of each Protein A-purified pAbs were separately precipitated with 1.0 volume ( N H 4 ) 2 S 0 4 and resuspended in 0.5 mL PBS before washing each sample by centrifuging three times at 3,000-4,350 x g m a x (4°C) in a Centricon 50 microconcentrator, with 1 mL and 2 mL volumes of PBS being applied after the first and second centrifugations respectively. For each Centricon 50-treated Protein A-purified serum, -2 mg IgG (as determined by A280; Harlow & Lane 1988) was biotinylated using a Pierce Immunochemicals Sulfo-NHS-biotinylation kit, as follows. After -2 mg of antibody in 1 mL PBS was combined with 20 uL (0.4 mg) Sulfo-NHS-biotin and incubated 2 h on ice, excess Sulfo-NHS-biotin was removed by repeated washings in a Centricon 50 microconcentrator. Avidin and H A B A (2-4'-hydroxyazobenzene benzoic acid) were used in accordance with the 91 manufacturer's recommended procedures to determine that the molar ratios of "surface-exposed" or "available" (viz., to streptavidin) biotin to antibody were 2.3:1 and 1.8:1 for FN2/4 and FS1/4 respectively. Each of the (i) E. coli-absorbed sera, (ii) protein A-purified and (iii) biotinylated IgG preparations were employed in later experiments. Volume changes have been considered in reporting the dilutions employed for each serum or antibody; i.e., each dilution is reported in terms of an effective dilution of crude serum. The effects of the various stages of purification on the abilities of FN2/4 and FS1/4 to recognize FHA were determined (Figure 2-8) by applying serial dilutions (3.1 to 100 ng in 2 uL PBS) of heparin-sepharose affinity-purified F H A and, as a control, recombinant CSP protein PfR32tet32, to nitrocellulose and probing these samples in a manner similar to that described later (§2.18, Immunoblots), with 1:2,000 and 1:10,000 effective dilutions of sera or antibodies from each level of purification. FS1/4 1 : 10,000 1 : 2,000 1 I-N2/4 1 : 10,000 1 : 2,000 • • 1 1 1 . 3 if 3 8 I T O (t — 32 CD L 1 Fig. 2-8. Effect of a -FHA sera purification and biotinylation on recognition of F H A . Samples (~2 uL) of heparin-sepharose affinity-purified F H A (and recombinant C S P as controls; not shown) were applied to gridded (5 mm x 5 mm) nitrocellulose in the pattern F H A (ng) 100 12.5 50 6.3 25 3.1 and probed with FN2/4 and FS1/4 a - F H A crude and purified sera and antibodies at the indicated dilutions, using a method similar to that described for F H A controls in §2.15.2.b (see also Figure 2-11). (b) a-FHA antibody purification for ELISA and immunoblots. FN2/4 and FS1/4 (crude) sera were absorbed against E. coli antigens (by means of a Pierce Immunochemicals immobilized E. coli lysate kit used as before) for use in ELISA and immunoblots. Volume changes have been considered in reporting the dilutions employed for these sera; i.e., each 92 dilution is reported in terms of an effective dilution of crude serum. FS2/4 serum was not pre-absorbed against E. coli antigens. 2.13.4. Secondary antibodies Peroxidase-conjugated goat a-mouse IgG or goat a-rabbit IgG polyclonal antibodies (Jackson ImmunoResearch or GIBCO/BRL) were employed as 2° antibodies for ELISAs in which virions coated onto wells of Immulon plates were probed with mouse mAbs or rabbit pAbs. For other ELISAs in which cross-reactivity of the 2° antibody with antigen (components of mouse ascitic fluid, §2.13.2.a) or capture antibody (mouse mAbs, §2.17.1, §2.17.2) was of concern, peroxidase-conjugated mouse a-goat IgG (§2.13.2.a) or a-rabbit IgG (§2.17.1, §2.17.2) pAbs that had been absorbed against serum proteins of mouse and other species; (Jackson ImmunoResearch) were employed; these 2° antibodies were also employed as a matter of convenience in certain conventional ELISAs. Alkaline phosphatase-conjugated goat a-rabbit IgG pAbs (GIBCO/BRL) were employed in plaque lifts and immunoblots. 2.13.5. Immunochemicals, recombinant CS protein and related material OPD (Sigma, o-phenylenediamine, 1 mg mL" 1 in 0.1 M citrate pH 4.5 with 0.012% H 2 0 2 ) was employed as substrate for peroxidase-conjugated 2° antibodies (ELISA). BCIP/NBT (5-bromo-4-chloro-3-indolyl phosphate /7-toluidine salt, 50 \xg mL" 1; nitroblue tetrazolium chloride, 100 \xg mL" 1, in substrate buffer: 100 mM Tris pH 9.6, 40 m M MgCl 2 ) was employed as substrate for alkaline phosphatase-conjugated 2° antibodies (plaque lifts and immunoblots). Nitrocellulose discs employed in plaque lifts and immunoblots were purchased from Schleicher and Schleicher. PfR32tet32, Pbtet32 and PvNSlv20, recombinant CS proteins of P. falciparum, P. berghei and P. vivax respectively, provided by R. A. Wirtz (WRAIR), have been previously described (Young et al. 1985; Egan et al. 1987; Gordon et al. 1990). Bovine serum albumin (BSA) was purchased from Sigma (as fraction V) or from Intergen (as "bovuminar standard powder"). 93 2.14. BIOPANNING (AFFINITY S E L E C T I O N O F T A R G E T C L O N E S ) 2.14.1. Biopanning random peptide libraries with cc-CSP mAbs Sixteen sets of biopans (affinity purifications of target clones from libraries) were carried out (Chapter 5, Figure 5-1). Thus, each of two random peptide libraries (§2.1) was biopanned with four quantities (1.2 fig - 8 pmol, 120 ng - 0.8 pmol, 12 ng - 80 fmol and 1.2 ng- 8 fmol) of each of two a-CSP mAbs, Pf2A10 and PflB2.2. (a) First round of biopanning. For each biopan, the indicated quantity of biotinylated mAb was combined with 2.6 x 10 1 0 virions (physical particles, "biopan input") of the 6-mer library or 2.5 x 10 1 1 virions (physical particles) of the 15-mer library in a final volume of 80 fiL PBS with 1% extensively dialysed BSA (d-BSA = 5 g BSA mL" 1 in 25 mL PBS dialysed three times over two days against 4 L PBS). Samples (in 500 (iL Eppendorf tubes) were incubated -20-24 h at 4°C on a slowly rotating culture tube carrier. After blocking (1 h, 37°C, 0.1% d-BSA in PBS, 200 fiL well - 1) microtiter plate wells to which streptavidin (SA) had been covalently bound (Pierce Reacti-bind SA-coated polystyrene microtiter plates, 400 ng - 7 pmol SA per well, 28 pmol biotin binding capacity per well) and subsequently washing each well 3 times with wash buffer (PBS + 0.5% Tween 20, 200 fiL well"1), virion/antibody mixtures were added to wells (one sample per well) and incubated 15 min. at room temperature with gentle motion on a platform shaker. For each biopan sample, 50 fiL of mixture ("unbound virions") was removed and stored for later use as a control. After wells were washed 10 times with wash buffer (200 fiL well"1), 200 fiL 0.1 N HC1 (adjusted to pH 2.2 with glycine) was added to each well to elute bound virions. After 15 min. elution with gentle motion on a platform shaker, 12 pL 2 M Tris base (unadjusted pH) was added to each well. Each sample of eluted and pH-neutralized virions was recovered immediately and combined with 1.5 mL E. coli K91-Kan ( O D 6 0 0 = 0.3). After incubating 1 h (37°C, standing, in L B + 0.2 fig tetracycline mL" 1), each virion/cell mixture was transferred to 30 mL L B + 20 pg 94 tetracycline mL"1 and incubated 23 hours (shaking) before virions were harvested (as "amplified biopan output") from culture supernatant by two-stage PEG precipitation (§2.6.4). For use as controls, four 10 \xL aliquots of "unbound virions" for each antibody/library combination were pooled and used to infect (§2.5.3) -2 mL E. coli K91-Kan. After infected cells were transferred to 100 mL L B + 20 u.g tetracycline mL"1 and incubated overnight (shaking), virions were harvested from culture supernatants by two-stage PEG precipitation (§2.6.4). (b) Subsequent rounds. The second and third rounds of biopanning were carried out in the same manner as the first round, except that 2.5 x 10" virions (physical particles, as quantified by U V spectroscopy, §2.7.1) of the 1 x 10 1 4 - 3 x 10 1 4 virions harvested from the previous round as "amplified biopan output" were employed as "biopan input". (c) Selection and propagation of selected clones. For each of the four biopans that employed 120 ng mAb, approximately 1-2 x 109 virions (physical particles; volume = 1 uJL) of amplified output from the third biopan were used to infect (§2.5.3) 1 mL E. coli K91-Kan, and infected cells were spread on LB-Tet plates and incubated overnight. After streaking on LB-Tet to obtain well-isolated colonies, 24 clones were cultured by serial overnight culture (§2.1.1, 30-50 mL) and virions were harvested from culture supernatants by two-stage PEG precipitation (§2.6.4). Virion yields were estimated by U V spectroscopy (§2.7.1). 2.14.2. Biopanning FHA-70 and -80 libraries (Figure 7-30) For each of eleven FHA peptide and "control" libraries (Table 2-V), four biopans were carried out with 5 x 108 to 7 x 1 0 n virions and 3.6 [ig (-24 pmol), 360 ng (-2.4 pmol), 36 ng (^240 fmol) and 3.6 ng (^24 fmol) of an equimolar pool of biotinylated protein-A purified rabbit a-FHA sera FN2/4 and FS1/4 (§2.13.3.a). For each biopan, the indicated quantities of virions and antibodies were combined in a final volume of 100 uL PBS with 1% extensively dialysed BSA (§2.14.1.a) and incubated -16 h (4°C, in 500 uL Eppendorf tubes) on a slowly rotating tube carrier. After blocking (2 h, 37°C, 1.0% d-BSA in PBS, 200 uL well"1) streptavidin-coated microtiter plate wells (28 pmol biotin binding capacity per well, §2.14.1.a), blocking solution was discarded and virion/antibody mixtures were added to wells (one sample per well) and incubated 20 min. at room temperature with gentle motion on a platform shaker. After wells were washed 10 times with wash buffer (PBS + 0.5% Tween 20, 200 pL well, ~5 min agitation on a platform shaker for each wash), 150 uL 0.1 N HC1 (adjusted to pH 2.2 with glycine) was added to each well to elute bound virions. After 20 min. elution on a platform shaker, 9 uL 2 M Tris base (unadjusted pH) was added to each well and each sample of eluted and pH-neutralized virions recovered immediately and stored (4°C) in microtiter plates that had been previously blocked with 1% BSA in PBS (200 pL well"1). The numbers of virions recovered for each biopan (Figure 7-30) were estimated by a TU "spot" assay (§2.7.2.c) using single 5 p.L samples of 10-fold serial dilutions of eluted virions. Table 2-V. Virions employed in biopanning FHA-70 and -80 libraries. Library Insert size (base pairs) Virions employed, log (physical particles) 70-X no insert 8.8 70-A 30-75 8.8 70-B 75-150 8.8 70-C 150-300 8.7 70-D 300-600 9.2 80-X no insert 11.6 80-Y no insert 10.9 80-A 30-75 11.4 80-B 75-150 11.4 80-C 150-300 11.1 80-D 300-600 11.8 96 2.15. P L A Q U E L I F T S AND R E L A T E D M E T H O D S E M P L O Y E D T O IDENTIFY A N T I B O D Y - R E A C T I V E C L O N E S IN F H A - 7 0 A N D -80 LIBRARIES 2.15.1. Methodology issues Two methods of using plaque lifts to identify antibody-reactive clones within biopan eluates of FHA-70 and -80 libraries were explored using wild-type phage f l and a-fl pAbs. For each method, two samples of f l virions sufficient to yield 200-400 plaques were plaqued on E. coli K91-Kan in a manner similar to full-plate plaque assays (§2.7.2.a). The two resulting lawns (A and B) were treated differently with respect to application of nitrocellulose discs (§2.13.5). In method A (Figure 2-9A) a nitrocellulose disc was laid down on lawn A after 1 h incubation (37°C; as in a protocol for plaque lifts of Agt l l expression libraries) and the lawn with applied nitrocellulose was incubated overnight (37°C). In method B (Figure 2-9B), lawn B was incubated overnight (37°C) before application of a nitrocellulose disc with slight pressure applied by a glass spreading rod. Following overnight incubation, both lawns with applied nitrocellulose were incubated 10 min. at 4°C (to allow nitrocellulose to be removed without lifting the [less fluid] top agar) before removing and placing each nitrocellulose disc in a Petri dish. After being washed three times with gentle shaking in 10 mL blocking buffer (5% skim milk powder, 0.1% Tween 20 in PBS; 30 min wash"1), each disc was cut in two parts and each part incubated 1 h at room temperature with gentle shaking in 10 mL of a 1:5,000 or 1:50,000 dilution of Protein A -purified a-fl (1°) pAbs (§2.13.1) in blocking buffer. After washing three times with blocking buffer (10 mL wash"1, 10 min. wash"1), each piece was incubated 1 h at room temperature with gentle shaking in 10 mL alkaline phosphatase-conjugated 2° antibodies (1:3,000 dilution in blocking buffer; §2.13.4). After three washes (10 mL wash"1, 10 min. wash"1) in blocking buffer and one in substrate buffer (§2.13.5), 8 mL BCIP/NBT substrate (§2.13.5) was added 97 to each disc, and reactions were allowed to proceed 2.5-8.5 min. before being stopped by rinsing the nitrocellulose pieces in water. Based on the puzzling and unsatisfactory results for method A and the conversely satisfactory results for method B (Figure 2-9), method B was chosen for subsequent plaque lifts. Fig. 2-9. Alternative methods of applying nitrocellulose to lawns. Plaque lifts of f l virions probed with a-fl pAbs (§2.15.1). In method A , a nitrocellulose disc was applied after 1 h incubation of the phage/bacteria lawn and the disc/lawn subsequently incubated overnight. In method B, the disc was applied after overnight incubation of the lawn. The resulting plaque lifts were probed with 1:5,000 (illustrated here) and 1:50,000 (not shown; results were similar) dilutions of a-fl pAbs. Alternative blocking and washing conditions were explored using CSP-library clones (§2.10) and a-CSP mAb P G A 10. TWO-LIL samples (five-fold serial dilutions from 1 [ig to 8 ng protein) of CSP-library clones displaying the peptides N A N P N A N (recognized by Pf2A10) and N A N A N A N (not recognized), N P N A N P N (recognized) and recombinant CSP proteins R32tet32 (recognized) and Pbtet32 (not recognized) were applied to nitrocellulose and processed in a manner similar to that described in the preceding section (§2.15.1), except that four combinations of blocking and washing methods were explored. Method A employed 5% skim milk powder + 0.1% Tween 20 in PBS for all washes and antibody dilutions. Method B employed 3% BSA, 1% skim milk powder for the first wash and antibody dilutions, and 0.1%o Tween 20 in PBS for remaining washes. Methods C and D were similar to method B, except skim milk powder (Method C) or Tween 20 (Method D) were omitted. Based on minimal background staining of nitrocellulose and appropriate signals for CSP-library clones and recombinant CSP, method B was chosen for subsequent plaque lifts. Results of a followup experiment using method B to probe plaque lifts of CSP-library clones are shown in Figure 2-10. 98 i J ! 1 ; , I [ IK N A N P N A N B 9 T V ' i bo ><trt 1 1 J 1 1 f l I r NANANAN , • « I 1 " J *- fj • • 1 • VcJ r / N P N A N P N • i Fig. 2-10. Plaque lifts of CSP-library and wild-type f l virions. A . Plaque lifts — performed using squares of gridded (5 mm x 5 mm) nitrocellulose - of CSP-library clones probed with a-CSP m A b Pf2A10. Plaque lifts of N A N P N A N and N A N A N A N were expected to be negative; the lift of N P N A N P N was expected to be positive. B. Concurrent plaque lifts, from the same lawns as in A , probed with a-fl pAbs to demonstrate that plaques existed and were successfully transferred to nitrocellulose. C. Wild-type f l virions probed with a-fl pAbs illustrates the small size of plaques of recombinant fd-tet derivatives (B) compared to those of wild-type F f phage (C). 2.15.2. Identification of antibody-reactive clones in FHA-70 libraries and assessment of the effect of library amplification on the fraction of antibody-reactive clones recovered (a) Library amplification. For each of libraries 70-A, -B, -C and -D that had been biopanned with 3.6 \ig and 360 ng of pooled FN2/4 and FS1/4 a-FHA pAbs (§2.14.2), 40 uL biopan eluate (-^lO3 TU) was combined with 1 mL E. coli K91-Kan (OD 6 0 0=0.2, thus, ~2 x 10 cells) and the resulting mixture incubated 0.5 h at 37°C before plating on LB-Tet. After 18 h incubation (37°C), virions were harvested by washing bacterial growth (2 x 102 to 3 x 103 cfu per eluate) from plates (§2.6.1) and subsequent centrifugation. (b) Plaque lifts of unamplified and amplified FHA-70 eluates (Figure 7-32). For each of libraries 70-A, -B, -C and -D, quantities of unamplified eluates (from biopanning with 3.6 |ig and 360 ng pooled FN2/4 and FS1/4 a-FHA pAbs, §2.14.2) and amplified eluates (immediately preceding section) sufficient to yield <2 x 103 plaques on each of 4 lawns were 99 plaqued on E. coli K91-Kan in a manner similar to full-plate plaque assays (§2.7.2.a). After overnight incubation to allow plaque development, plates were equilibriated to room temperature before applying a nitrocellulose disc (§2.13.5) to each lawn with light pressure applied by a glass spreading rod. After 40 min. incubation at 4°C (in some later experiments a-FHA 80-D a-FHA 80-X a-f1 80-X Fig. 2-11. Plaque lifts of biopanned FHA-80 library clones. A. Disc of gridded (5 mm x 5 mm) nitrocellulose identifying the location of the four-by-four matrices shown in B to D . B . Initial plaque lifts employed to identify antibody-reactive clones included 100, 20, 4 and (optionally) 0.8 ng samples of heparin-sepharose affinity-purified F H A as positive controls. C and D. Plaque lifts from early trials in which relatively low dilutions (1:4,000) of each of FN2/4 and FS1/4 (as a pool) were employed as 1° antibodies; later trials included higher dilutions (1:8,000 each) of pooled antibodies and 1:8,000-1:32,000 dilutions of FN2/4 alone. C and D are derived, respectively, from the "large fragment" (300-600 bp inserts) library 80-D and the control (no insert) library 80-X. The relatively greater intensity of staining of F H A controls in D compared to C reflects the greater color development time ( B C I P / N B T substrate with alkaline phosphatase-conjugated 2° antibodies) of the 80-X versus 80-D plaque lifts. E . Plaque lift o f control library 80-X probed with rabbit a-fl pAbs. This concurrent control for the corresponding plaque lift shown in D , intended to show that plaques existed and were transferred to nitrocellulose, was included only in early experiments. this time was arbitrarily increased to up to 1.5 h), nitrocellulose discs were removed and dried (room temperature, 30 min.) before applying 2 uL samples (100, 20, 4 ng) of heparin-sepharose affinity-purified FHA as positive controls (Figure 2-11) and allowing these to dry 100 (room temperature, 20 minutes). Each disc was incubated 1 h at room temperature with gentle agitation in 10 mL blocking buffer (1% skim milk powder; 3% BSA, reduced to 2% B S A in some later experiments) before storing overnight (4°C) in blocking buffer (overnight storage was for convenience and not always repeated in later experiments). After equilibriation (20 min., with gentle agitation) to room temperature, each disc was washed three times (per wash: 10 mL PBS + 0.05% Tween 20, 15 min. with gentle agitation) before addition of (i) 1:8,000 dilutions (in blocking buffer, 8 mL per disc) of each of E, co//-absorbed protein A-purified FN2/4 and FS1/4 pAbs (§2.13.3.a), and (ii) 1:8,000, 1:32,000 and 1:128,000 dilutions of FN2/4 alone. After 1 h incubation (room temperature) with gentle agitation, each disc was washed three times before addition of 10 mL alkaline phosphatase-conjugated 2° antibodies (1:3,000 dilution in blocking buffer; §2.13.4). After 1 h incubation with gentle agitation, each disc was washed three times with wash buffer and once (15 min. with gentle agitation) with 10 mL substrate buffer (§2.13.5) before addition of 8 mL BCIP/NBT substrate (§2.13.5). Reactions were allowed to proceed until (after 22-96 min.) adequate signal had developed for an apparently significant fraction of plaques; reactions were stopped by rinsing the nitrocellulose discs in water. Each set of amplified and unamplified samples was developed for the same period of time. Plaques visible (even faintly so) on nitrocellulose were counted as "antibody-reactive"; plaques visible on the corresponding bacterial lawn were counted (or estimated in some cases, by counting plaques on one-quarter of the lawn) as "total" plaques (Figure 7-32). Average color +intensity of antibody-reactive plaques varied considerably from library to library, such that plaques of the "larger-fragment" libraries 70-C and 70-D developed more quickly and to more intense color than did those of the "smaller-fragment" libraries 70-A and 70-B; indeed, plaques of the latter libraries were commonly faint and difficult to discern, even with a hand-held magnifying lens. Example plaque lifts from early trials with library 80-D are shown in 101 Figure 2-11. (c) Selection and plaque purification of antibody-reactive FHA-70 clones. As summarized in Table 6-II (Chapter 6), 56 antibody-reactive clones were selected from FHA-70 libraries for further analysis. These were chosen from those that reacted most strongly with antibody (as judged by plaque lift color intensity at a given antibody dilution, or by any visible reactivity at a high antibody dilution). For each selected clone, a ~2 mm x 2 mm plaque-containing agar plug (identified by the position of the target plaque on gridded nitrocellulose and corresponding grid marks left on the lawn by applying nitrocellulose with light pressure) was excised and transferred to 1 mL PBS. After overnight incubation (4°C) to allow elution of virions, 5-fold serial dilutions of each virion eluate were plaqued on E. coli K91-Kan in a manner similar to full-plate plaque assays (§2.7.2.a). After overnight incubation (37°C) to allow plaque development, plaque lifts of sections of the resulting lawns were performed essentially as described above (§2.15.2.b), except (i) all plaque lifts were probed with a 1:8,000 dilution of Protein A-purified FN2/4 pAbs, and (ii) in many cases, plaque lifts were performed with one or several 2 to 2.5 cm square pieces of nitrocellulose (versus a complete disc). This procedure was repeated once or twice in essentially the same manner (exceptions are noted below) until an apparently clonal population was isolated for each of 51 of the originally selected 56 clones, viz., until the isolation of a single plaque that, after excision and elution, gave rise to plaque lifts in which all plaques were antibody-reactive. During one round of plaque purification, plaque lifts for seven weakly-reactive clones were performed using a 1:4,000 dilution of FN2/4, and those for ten other weakly-reactive clones were performed using an equimolar pool of 1:4,000 dilutions of each of FN2/4 and FS1/4. At some time during the repeated rounds of plaque purification, reactivity with antibody was "lost" for five of the originally selected 56 clones. Conceivably, this may reflect errors in selecting plaques or recombination within B. pertussis ^ TzaB-derived 102 inserts; these five were not considered in subsequent analyses. As a final clonal isolation step, 20 uL samples of 5-fold serial dilutions (5"1 to 5"6) of each apparently monoclonal virion eluate were used to infect (§2.5.3) 80 uL E. coli K91-Kan, and infected cells were spread on LB-Tet plates. After overnight incubation, a single colony derived from each clone was transferred as a short streak to an LB-Tet plate. After overnight incubation these "FHA-70 master stock plates" were stored at 4°C and used as inocula for subsequent virion propagation. (d) Virion propagation. For each of the 51 clones isolated as described above, colonies on "FHA-70 master stock plates" were used as a source of inocula for serial overnight cultures (§2.1.1, 50 mL). Virions were harvested from -32 mL of each resulting culture by two-stage PEG precipitation (§2.6.4) and quantified by U V spectroscopy (§2.7.1). These virions were used (i) as a source of template (§2.3.2.b) for sequencing (§2.4) /TzaB-derived inserts, (ii) in ELISA (§2.16.4.2) and (iii) as a source of virions used to infect E. coli K91-Kan in later large-scale preparations of selected clones (§2.15.4). 2.15.3. Identification of antibody-reactive clones in FHA library 80-A (a) Identification and selection' of clones. Quantities of FHA library 80-A biopan eluates (derived from biopanning this library with 3.6 p,g and 360 ng pooled FN1/4 and FS2/4 a-FHA pAbs, §2.14.2) sufficient to yield -200 plaques on each of six lawns were plaqued on E. coli K91-Kan in a manner similar to that described for full-plate plaque assays (§2.7.2.a). After overnight incubation (37°C) plaque lifts of these lawns were performed essentially as described earlier (§2.15.2.b) except that for each eluate, duplicate plaque lifts were probed with 1:8,000, 1:32,000 and 1:128,000 dilutions of protein A-purified FN2/4 pAbs (§2.13.3.a). Positive plaques and total plaques (Chapter 6, Table 6-III) were counted as described earlier. (b) Clonal isolation by limiting dilution transduction. For each of 58 clones selected from among those identified with a 1:32,000 dilution of FN2/4 (Chapter 6, Table 6-III), a ~2 mm x 2 mm plaque-containing agar plug was excised, transferred to 0.5 mL L B and stored 103 overnight (4°C) to allow elution of virions. For each clone, 10 uL samples of 10-fold serial dilutions (10"' to 10 " l2) of these virion eluates were transferred to 90 \iL E. coli K91-Kan (OD 6 0 0=0.2) in microtiter plate wells. After 1 h incubation (37°C, in L B + 0.2 \xg tetracycline mL" 1), 10 [iL of each well was transferred to 90 \xL LB-Tet in microtiter plate wells and, after overnight incubation (standing), the O D 5 9 5 of each culture was read to determine the highest dilution of each clone showing obvious host cell growth. A single sample of each such dilution was streaked on LB-Tet and, after overnight incubation, a single colony was selected and employed in virion propagation, as below. (c) Virion propagation. Colonies selected as described immediately above were used as a source of inocula for serial overnight cultures (§2.1.1, 40 mL). Virions harvested from ~32 mL of each culture supernatant by two-stage PEG precipitation (§2.6.4) and quantified by U V spectroscopy (§2.7.1) were used (i) as a source of s.s. template (§2.3.2.b) for sequencing (§2.4) B. pertussis fhaB-derived inserts, (ii) in ELISA (§2.16.4.2)and (iii) as a source of virions used to infect E. coli K91-Kan in later large-scale preparations of selected clones (§2.15.4). 2.15.4. Large-scale propagation of selected FHA-70 and -80 clones Because of concerns regarding batch-to-batch variability (Chapter 6, §6.3.4) among previous virion preparations (§2.15.2.d, §2.15.3.c) of antibody-reactive clones (particularly with respect to varying quantities of non-virion contaminants that may have interfered with virion binding to Immulon plates), 44 selected B. pertussis _/7z<2B-derived clones, together with two fDRW70 pseudorevertants (to be employed as controls) were propagated in large volume and harvested from culture supernatant in such a way as to minimize variation among the five batches of such preparations. (a) Culture of phage-infected E. coli K91-Kan. For each clone, one uL PEG-precipitated virions from a preparation that had served as a source of sequencing template for identification of unique clones was used to infect (§2.5.3) 1 mL E. coli K91-Kan, and infected 104 cells were transferred to 200 mL 2xYT + 20 pg tetracycline mL"1 in 1 L Erlenmeyer flasks. After 20.5-22.5 h incubation ( O D 6 0 0 = 1.5-1.8 for 23 arbitrarily selected cultures), virions were harvested from culture supernatant and purified as follows. (b) PEG precipitation and syringe filtration. After centrifuging 195.0 ± 0.5 mL of each culture in a 250 mL centrifuge bottle (15 min, 5,500 x g m a x , 4°C), 190.0 ± 0.5 mL culture supernatant and 28.5 mL PEG formulation " B " (§2.6.3) were combined and mixed by 100 inversions. After overnight incubation at 4°C, each sample was centrifuged (1 h, 15,300 x g m a x , 4°C), supernatant was discarded and each bottle drained (>0.5 h) on paper towel. After addition of 12.5 mL TBS to each pellet, each sample was rocked gently (room temperature) for 0.5 h before transferring resuspended virions into 50 mL centrifuge bottles. After centrifugation (15 min., 4,400 x g m a x , 4°C), 12.5 ± 0.5 mL cleared supernatant was transferred to a 50 mL centrifuge bottle and precipitated again, as follows. After addition of 1.9 mL PEG formulation B to each bottle and subsequent mixing (by 100 inversions), each sample was incubated overnight at 4°C. After centrifugation (0.5 h, 35,000 x g m a x , 4°C), supernatant was discarded and each bottle drained (>0.5 h) on paper towel. After addition of 3.0 mL TBS, each bottle was incubated overnight at 4°C before transferring resuspended virions as two 1.5 mL aliquots into two 2-mL Eppendorf tubes. After centrifuging each tube 3 min. at maximum speed in an Eppendorf 5415C microfuge, 1.4 mL of cleared supernatant was transferred to a fresh Eppendorf tube and precipitated again, as follows. After addition of 210 pL PEG formulation B to each tube and subsequent mixing (by 100 inversions) each sample was incubated 4 h on ice before centrifugation (30 min., maximum speed in an Eppendorf 5415C microfuge, 4°C). After careful removal of supernatant by pipettor, tubes were centrifuged briefly and residual supernatant again removed by pipettor. After addition of 525 \xL PBS, each sample was incubated overnight at 4°C. Virions within each sample were resuspended by gentle pipetting, incubation for 1 h at 37°C, and repeated pipetting. After centrifugation 105 (10 min., maximum speed in a 5415C, 4°C), 500 [iL cleared supernatant of each of the two samples for each clone were pooled into a 1.5 mL Eppendorf tube, and centrifuged as before. Centrifugation (15 min. at room temperature) and subsequent transfer of cleared supernatant to fresh tubes was repeated twice. Finally, each sample was filtered with a 0.2 u,m syringe filter (VWR). Recovered virions were quantified by U V spectroscopy (§2.7.1). 2.15.5. Assessment of biopanning enrichment (Figure 7-31) For each of libraries 70-A and 70-B, quantities of unenriched library virions (viz., not biopanned) and biopan eluates (§2.14.2) sufficient to yield <5 x 10 plaques on each of two lawns were plaqued on E. coli K91-Kan in a manner similar to full-plate plaque assays (§2.7.2.a). After overnight incubation (37°C), plaque lifts of these lawns were performed essentially as described earlier (§2.15.2.b), except that plaque lifts were probed with a 1:8,000 dilution of Protein A-purified FN2/4 pAbs (§2.13.3.a) and color development was stopped after 90 minutes. Positive plaques and total plaques were counted as described earlier (§2.15.2.b). Additionally, concurrent plaque lifts of selected library 70-A and 70-B virions were probed with Protein A-purified a-fl pAbs (1:10,000 dilution, §2.13.1) to demonstrate adequacy of plaque transfer (illustrated in Figure 2-1 IE). O J -3-Fig. 2-12. Evaluation of Immulon plates for ELISA. Wells of Immulon-1, -2, -3 and -4 microtiter plates were coated with the indicated quantities of CsCl-purified fUSE2 virions and probed with Protein A-purified rabbit a-fl (1°) pAbs and peroxidase-conjugated 2° antibodies (§2.15.5.a). Values shown are means of triplicate wells ± two standard deviations. 10 5 2.5 1.25 0.630 31 0 Immulon 4 • 1 : 4,000 O 1 : 16,000 • 1 : 64,000 a-f1 dilution fUSE2 (n,g protein well"1) 106 2.16. E L I S A M E T H O D " A " : IMMOBILIZED VIRIONS. 2.16.1. Methodology (a) Evaluation of Immulon plates and determination of optimal virion quantity for coating plates (Figure 2-12). Triplicate wells of Immulon-1, -2, -3 and -4 plates were coated with 0-10 fig CsCl-purified (§2.6.5) fUSE2 virions in 100 pL PBS and incubated -20 h at 4°C before washing three times (PBS + 0.05% Tween 20). After addition of blocking buffer (PBS + 1% BSA, 200 [iL well"1), plates were incubated 1.1 h at 37°C before washing three times. After addition of four-fold dilutions (1:4,000-1:64,000) of protein A-purified rabbit a-fl pAbs (§2.13.1), plates were incubated 1.25 h at 37°C before washing three times. After addition of peroxidase-conjugated 2° antibodies (1:3,000 dilution in blocking buffer, 100 pL well" 1; §2.13.4), plates were incubated 1 h at 37°C before washing six times, developing with OPD (§2.13.5) and reading colored reaction products as described later (§2.16.2.1). Although mean absorbance values varied little among plates (Figure 2-12), coefficients of variation varied considerably from plate to plate and with the different quantities of virions used to coat wells. Considering this, Immulon-2 plates coated with 2 [ig phage protein well"1 were used in subsequent ELI SAs. (b) BCA assay for quantifying virions bound to Immulon plates (Figure 2-13). To determine the feasibility of using the colorimetric bicinchoninic acid (BCA) assay (Smith et al. 1985) to quantify virions (as total protein) bound to Immulon-2 plates, triplicate wells were coated 36 h at 4°C with 0-2.5 [ig CsCl-purified (§2.6.5) fUSE2 virions (in PBS, 100 [iL well"1) and washed in the manner described for ELISA (§2.16.1.a). After adding a mixture of 10 [iL H 2 0 and 200 [iL B C A reagent (Sigma bicinchoninic acid kit for protein determination) to each well, plates were incubated at 37°C and sample absorbances ( A 5 7 0 ) measured at one- to eight-hour intervals (Figure 2-13A) using a Biorad 3550 microplate reader. Concurrently, 0-5 [ig of (i) fUSE2 virions and (ii) BSA (in PBS, 10 [iL well"1) were loaded into triplicate wells of 107 Immulon-2 plates and, after adding 200 (iL B C A reagent to each well (without wells having been washed as for ELISA), plates were incubated at 37°C and A 5 7 0 determined as described above. A Immobilized phage tf) ™ 0.20 o C D .E 0.15 0.10 0.05 A o 0.00 0 2 i 1 i ' i ' i 1 i 1 i 4 6 8 10 12 14 B 1.2 n 1.0 0.8 0.6 0.4 0.2 0.0 Incubation time (h) with BCA reagent -1 Phage protein loaded (ug well ) • 1.0 SI 1.5 • 2.5 Phage in solution 6 8 10 12 14 < C O m (N LU w < o C Phage versus BSA 1.5 1.0 0.5 0.0 0 2 4 6 8 10 12 14 Incubation time (h) with BCA reagent Phage protein or BSA (u,g well 1) X 0.0 • 0.5 H 1.0 H 2.5 • 5.0 Fig. 2-13. B C A assay for quantification of virions bound to Immulon plate wells: effect of incubation time. A . CsCl-purified fUSE2 virions coated onto wells of Immulon-2 microtiter plates and assayed (using a colorimetric B C A assay) for total protein remaining after washing plates in the manner employed for E L I S A . B. fUSE2 virions and B S A assayed in the same manner as in A except that plates were not washed. Values in A and B are means of triplicate wells ± two standard errors. Although A 5 7 0 values for virions in solution (Figure 2-13, plot B) were higher than those for virions bound to wells (plot A; this may reflect a low fraction of virions that bind to wells), in both cases (A and B) A 5 7 0 was proportional to virion concentration over a wide range of incubation times, and the rate of change in A 5 7 0 had begun to diminish by 4 h. As well, the ratio of A 5 7 0 values for virions in solution versus BSA (BSA chosen to be illustrative of a conventional application of a B C A assay) had reached a near plateau by 4 h (Figure 2-13C). These findings indicated that virions bound to Immulon-2 plate wells could be quantified using a B C A assay with >4 incubation at 37°C. 2.16.2. ELISA of CSP-library clones (Chapter 4) (a) Standard ELISA (Figures 4-3 to 4-5). Triplicate wells of Immulon-2 plates were coated with (i) 2 fig CsCl-purified (§2.6.5) virions of CSP-library clones or fUSE2 (negative 108 control) in 100 uL PBS, or (ii) 100 ng recombinant CSP protein PfR32tet32 or PvNSlV20 or a 1:500 dilution of culture supernatant containing PBtet32, in 100 [iL PBS, or (iii) as a control, PBS only. After >16 h incubation at 4°C, plates were washed three times (PBS + 0.05% Tween 20) before addition of blocking buffer (PBS + 1% BSA + 0.5% bovine milk casein, 200 pL well"1), incubation for 0.5-1.5 h at 37°C, and washing as before. After addition of (i) three dilutions (see §2.16.2.2) of a-CSP (1°) mAbs (in blocking buffer, 100 pX well' 1) and (ii) as a control, no 1° mAbs, plates were incubated ~1 h at 37°C before washing three times. After addition of peroxidase-conjugated 2° antibodies (1:3,000 dilution in blocking buffer, 100 u.L well" 1; §2.13.4), plates were incubated ~1 h at room temperature before washing 6 times. After addition of OPD(§2.13.5), reactions were allowed to proceed <35 min. before being stopped by addition of 3N H 2 S 0 4 (100 \xL well"1). Colored reaction products were measured at 490 nm using a Biorad model 3550 microplate reader. Antibody concentration (nM) Fig. 2-14. Titration of recognition of CSP-library clones by a-CSP mAbs. Immulon-2 plates were coated with CsCl-purified CSP-library clones (2 irg well" 1) and probed with the indicated quantities of a-CSP mAbs and peroxidase-conjugated 2° antibodies (§2.16.2.a). Values shown are means ± two standard errors. Most means are of triplicate wells («=3) coated with virions from a single clone or a pool of two independently isolated and processed clones displaying the same peptide. Other means are of triplicate wells o f each of two independently isolated, processed and assayed clones displaying the same peptide (n=6). Encoded peptides are: V 0 1 , N A N P N P N ; V 0 3 , N A N P D P N ; V 0 5 , N A N A N P N ; V 0 7 , N A N A D P N ; V 1 7 , N P N P N P N ; V I 9 , N P N P D P N ; V 2 1 , N P N A N P N ; V 2 3 , N P N A D P N . Data shown here derive from the experiment summarized in Figures 4-5B and C (Chapter 4). 109 (b) Titration curves for CSP-library clones probed with a-CSP mAbs (Figure 2-14). Antibody concentrations employed in the ELISAs described above were chosen to emphasize clone-to-clone differences in peptide-MAb interaction. In most cases, the ELISA signal was titerable in a near linear fashion around the second of the three chosen mAb concentrations; viz., increasing or decreasing mAb concentration yielded a corresponding increase or decrease in ELISA signal. This is illustrated in Figure 2-14 for CSP-library clones that reacted strongly (Chapter 4, Figure 4-5) with two mAbs employed in this study, Pf2Fl . l and Pf5A4.1. (c) Quantification of CSP-library clones bound to Immulon plates (Figure 4-8). Immulon-2 plates were coated and washed as described for ELISA (§2.16.2.a). After a mixture of 10 [iL H 2 0 and 200 \xL B C A reagent (§2.16.1 .b) was added to each well, plates were incubated 4 h at 37°C before measuring relative quantities of bound virion protein as described earlier (§2.16.1.b). (d) Effect on ELISA of varying virion quantities, selected CSP-library clones (Figure 4-9). These ELISAs were performed essentially as described earlier (§2.16.2.a), except that wells were coated with 4, 2, 1, 0.5, 0.25 and 0 fig CsCl-purified (§2.6.5) virions. (e) Competition ELISA, selected CSP-library clones (Figure 4-10). These ELISAs were performed generally as described earlier (§2.16.2.a) except that each dilution of each 1° mAb was preincubated (2 h at room temperature) with varying quantities (0 to 100 u,M in blocking buffer) of the synthetic peptide N P N A N P N A N P N A (Sigma). As well, virions used in these assays were purified only by two-stage PEG precipitation (§2.6.4), not by CsCl density gradient centrifugation. 2.16.3. Preliminary ELISA of random peptide library clones with a-CSP mAbs (a) ELISA "A" (Figures 5-3 and 5-4, graphs labelled "A"). For each of the four random peptide library biopannings that employed 120 ng mAb (§2.14.1), individual wells of 110 Immulon-2 plates were coated with (i) 2 u.g two-stage PEG precipitated virions (§2.6.4) of 24 clones selected from the third round of biopanning and (ii) as a control, coating buffer only (PBS). After overnight incubation (4°C), plates were washed three times (PBS + 0.05% Tween 20) before adding 200 \iL well"1 blocking buffer (PBS + 1% BSA + 0.5% bovine milk casein) and incubating 1 h at 37°C. After washing three times, a relatively high concentration of 1° antibodies (1:333 dilution in blocking buffer of Pf2A10, PHB2.2 or Pb4B10, as appropriate) was added to each well (100 [iL well"1) and plates were incubated 0.75 h at 37°C before washing three times. After addition of peroxidase-conjugated 2° antibodies (1:3000 dilution in blocking buffer, 100 [iL well"1; §2.13.4), plates were incubated 0.75 h at 37°C before washing, developing with OPD (§2.13.5) and reading reaction products as described earlier (§2.16.2.a). (b) ELISA "B" (Figures 5-3 and 5-4, graphs labelled "B"). These ELISAs were performed as described above (§2.16.3.a), except (i) 1° antibodies were used at a 1:500 dilution; (ii) additional controls were employed in which 1° antibodies Pf2A10 and PflB2.2 were pre-incubated with the recombinant CSP protein R32tet32 (10 ng and 1 |j.g mL" 1) ~2 h at room temperature; and (iii) plates were incubated only 0.5 h with 1° and 2° antibodies. 2.16.4. Initial ELISA and BCA assays of FHA-70 and -80 clones (Figures 6-3 and 6-4) (a) Pooling of siblings and amplification of selected clones. Two-stage PEG-precipitated virions amplified from initial antibody-reactive plaques selected from plaque lifts of FHA-70 and -80 libraries (§2.15.2.d, §2.15.3.c), and which had been employed as a source of sequencing template to identify unique clones, were pooled or amplified as follows. For selected clones for which multiple siblings existed, virion preparations from up to 22 siblings were pooled. Other clones, particularly those possessing unique sequences (i.e., no siblings existed), were amplified by infecting (§2.5.3) E. coli K91-Kan with a small aliquot of virions and culturing infected cells overnight in LB-Tet (35-50 mL) before harvesting virions by two-I l l stage PEG precipitation (§2.6.4). (b) ELISA and concurrent BCA assay (Figure 6-3). For each unique clone, triplicate wells of Immulon-2 microtiter plates were coated with (i) 1 fig PEG-precipitated virions (in PBS, 50 fiL well-1) of antibody-reactive clones (a single pool, as described in the preceding section, or one or more individual siblings) and, as controls, (ii) virions of fDRW70 pseudo-revertants and (iii) PBS only. After overnight incubation at 4°C, plates were washed three times (PBS + 0.05% Tween 20) before adding blocking buffer (PBS + 1% B S A + 1% skim milk powder, 150 fiL well"1) and incubating 2 h at 37°C. After plates were washed three times, 1° antibodies (1:2,000 and 1:8,000 dilutions in blocking buffer, E. co//-absorbed FN2/4 and FS1/4 a-FHA pAbs, §2.13.3.a; 100 fiL well"1) were added and plates were incubated 1 h at 37°C. After plates were washed 3 times, peroxidase-conjugated 2° antibodies were added (1:3,000 dilution in blocking buffer, 100 fiL well" 1; §2.13.4) and plates were incubated 1 h at 37°C before washing, developing with OPD (§2.13.5) and reading reaction products (Figure 6-3B) as outlined earlier (§2.16.2.a). Concurrently, a second set of Immulon-2 plates was coated and washed as for ELISA, and quantities of bound virions were determined by a B C A protein assay performed essentially as described earlier (§2.16.2.c). In a followup ELISA (Figure 6-3D), selected clones were assayed in the same manner as before, except that virions were probed with (i) FN2/4 pAbs (at a 1:2,000 dilution) and (ii) as a control to demonstrate lack of recognition of virions by 2° antibodies, with no 1° antibodies. (c) Clone-to-clone variability of binding to Immulon plates. Selected clones (specifically, three that bound poorly to Immulon-2 plates, four that bound well) assayed as described in the preceding section were again assayed, except that varying quantities of virions were used to coat Immulon-2 plate wells. Duplicate plates (with duplicate wells for each clone on each plate) were coated with 5 times (5x), lx and 0.2x the quantity (1 fig in 50 fiL PBS) of virions employed previously. An ELISA with a 1:2,000 dilution of E. co//-absorbed protein 112 A-purified FN2/4 (one plate) and a B C A assay (the 2nd plate) were performed essentially as described in the preceding section. 2.16.5. Competition ELISA and related BCA assays of FHA-70 and -80 clones (a) Clone-to-clone variability of binding to Immulon plates (Figure 6-7). For each of 44 antibody-reactive clones and for two pseudorevertants of fDRW70 (A and B), duplicate wells of Immulon-2 plates were coated with two-fold serial dilutions (250 ng to 8 pg well"1 , 100 [iL well"1) of PEG-precipitated and syringe-filtered virions (§2.15.4). After overnight incubation at 4°C, plates were washed and a B C A protein assay performed essentially as described earlier (§2.16.2.c). (b) Competition ELISA. a-FHA antibodies for this "competition" ELISA were prepared by preincubating 1:5,000 dilutions of each of E. co/z'-absorbed FN2/4 and FS1/4 (§2.13.3.a; in blocking buffer with the concentration of NaCl adjusted to 233 mM) with heparin sepharose-affinity purified FHA at final concentrations of 30 \ig, 5 |_ig, 833 ng and 0 ng mL"1 for 2.75 h at 37°C before diluting the antibody preparations 2-fold immediately prior to their use in ELISA. For each antibody-reactive clone or control (fDRW70 pseudorevertant), duplicate wells of Immulon-2 plates were coated with 1 |j.g PEG-precipitated and filtered (§2.15.4) virions (100 [XL well"1 in PBS). After overnight incubation at 4°C, plates were washed 3 times (PBS + 0.05% Tween 20) before adding blocking buffer (PBS + 1% BSA + 1% skim milk powder, 200 uL well"1) and incubating 1 h at 37°C. After plates were washed 3 times, a-FHA antibodies prepared as described above were added (100 [iL well"1) and plates were incubated 0.75 h at 37°C. After plates were washed 3 times, peroxidase-conjugated 2° antibodies (1:3,000 dilution in blocking buffer, 100 pL well" 1, §2.13.4) were added and plates were incubated 1 h at 37°C before washing, developing with OPD (§2.13.5) and reading reaction products (Figure 6-8B) as described earlier (§2.16.2.a). Concurrently, an additional set of Immulon-2 plates were coated with virions and washed as for ELISA before quantifying the 113 relative quantities of virions remaining bound to wells (Figure 6-8C) by a B C A protein assay as described earlier (§2.16.2.c). Concurrent controls for this B C A assay, generally illustrative of those periodically employed for other B C A assays of bound virion protein, are shown in Figure 2-15. 0.25-o CO .a o CO < 0.0-0.01-• 1DRW70-prA • fDRW70-prB 0.8 0.1 1 P h a n e Drotein. un/well 10 30 Fig. 2-15. Control for B C A assay. In an assay performed concurrently with those of §2.16.5.b, the indicated quantities of virions of fDRW70 pseudorevertants " p r A " and "p rB" were loaded into triplicate wells of Immulon-2 plates. Total virion protein was assayed (without washing wells as in the corresponding assay of bound virions) using a colorimetric B C A protein assay. Data in A and B illustrate overlapping sets of data points derived from the same assay; note that axes in A are linear-linear, while those in B are log-log. Values shown are means of triplicate wells. Although error bars representing two standard errors were plotted, these are not evident because of their small size: they are effectively "hidden" by the data point symbols 2.17. E L I S A M E T H O D " B " : IMMOBILIZED ANTIBODIES 2.17.1. Capture ELISA of selected CSP-library clones (Figure 4-7) After addition of biotinylated mAbs (Pf2A10, 2.5 ng ~ 17 fmol well" 1; PflB2.2, 250 ng ~ 1.7 pmol well"1) in wash buffer (100 fiL well"1; 25 mM Tris pH 7.6, 150 m M NaCl, 0.1% B S A , 0.05%o Tween 20) to wells of streptavidin-coated 96-well microtiter plates (28 pmol biotin-binding capacity per well, §2.14.1.a), plates were incubated 2 h at room temperature before washing three times (200 fiL well"1). After addition of CsCl-purified (§2.6.5) virions (0.08-2 fig well"1 in 100 fiL wash buffer), plates were incubated 30 min. at room temperature before washing three times. After addition of Protein A-purified rabbit a-fl pAbs (1:5,000 dilution in wash buffer; 100 fiL well"1; §2.13.1), plates were incubated 30 min. at room 114 temperature before washing three times. After addition of peroxidase-conjugated 2° antibodies (1:5,000 dilution in wash buffer, 100 uL well" 1, §2.13.4), plates were incubated 30 min. at room temperature before washing three times, developing with OPD (§2.13.5) and reading reaction products generally as described earlier (§2.16.2.a). 2.17.2. Capture ELISA of random peptide library biopan eluates (Figure 5-2) After addition of biotinylated mAbs (Pf2A10 or PflB2.2, 120 ng = 0.8 pmol well"1) in wash buffer (100 uL well"1, as in preceding section) to wells of streptavidin-coated microtiter plates (28 pmol biotin binding capacity per well, §2.14.1.a), plates were incubated and washed as described above (§2.17.1). After addition of 3.6 u.g PEG-precipitated virions (amplified output from third round of biopanning or unbound fraction from first round, §2.14.1), plates were incubated and washed as described above (§2.17.1). After addition of rabbit a-fl 1° Abs (as in §2.17.1, but additionally purified to remove antibodies that may have reacted with E. coli contaminants in PEG-precipitated virion preparations, using a Pierce Immunochemicals immobilized E. coli lysate kit in accordance with the manufacturer's instructions), plates were incubated 0.75 h at room temperature before washing, adding 2° antibodies and performing other procedures as described above (§2.17.1). 2.18.. I M M U N O B L O T S 2.18.1. Assessment of binding to nitrocellulose of FHA-70 and -80 clones (Figure 6-4) and immunoblots of antibody-reactive clones (Figure 6-6, assay (i)) For F H A library clones and for two fDRW70 pseudorevertants (A and B), triplicate ~2 uL samples (800 ng virion protein) of PEG-precipitated and filtered virions (§2.15.4) were applied to nitrocellulose discs (§2.13.5) in the pattern shown in Figure 6-5 (Chapter 6). Triplicate ~2 \xL samples of two-fold serial dilutions (per 2 uL sample: 6 ng to 3.2 |ig) of PEG-precipitated and filtered virions (§2.15.4) of fDRW70 pseudorevertants were similarly applied to nitrocellulose discs, but in a grid fashion (Chapter 6, Figure 6-4). After samples had 115 dried, each nitrocellulose disc (or "blot", contained in a standard petri dish) was incubated 1 h at room temperature with gentle rocking in 10 mL blocking buffer (PBS + 3% BSA + 1% skim milk powder). After three 10 min. washes at room temperature in 10 mL wash buffer (PBS + 0.05% Tween 20), each blot was incubated 1 h at room temperature with gentle rocking in 10 mL primary antibody (per blot, one of (i) 1:8,000, (ii) 1:32,000 or (iii) 1:128,000 dilution in blocking buffer of protein A-purified a-fl pAbs (§2.13.1); additionally, for library clones, 1:2,000 dilution of E. co/7-absorbed (§2.13.3.a) (iv) FN2/4 or (v) FS1/4 pAbs or (vi) crude FN 1/4 a-FFIA serum, or (vii) as a control, no antibody). After three 10 min. washes, each blot was incubated 1 h at room temperature with gentle rocking in 10 mL alkaline phosphatase-conjugated 2° antibodies (1:3,000 dilution in blocking buffer; §2.13.4). After three 10 min. washes in wash buffer followed by a single 10 min. wash in 10 mL substrate buffer (§2.13.5), 10 mL BCIP/NBT substrate (§2.13.5) was added to each blot. Reactions were allowed to proceed 5 min. (blots probed with 1:8,000 and 1:32,000 dilutions of a-fl pAbs), 9 min. (1:128,000 of a-fl pAbs) or 14 min. (a-FHA pAbs or serum) before being stopped by rinsing blots in water. 2.18.2. Immunoblots of antibody-reactive FHA-70 and -80 clones (Figure 6-6, (ii)) For each of the antibody-reactive clones shown in Figure 6-5 (Chapter 6), and for two fDRW70 pseudorevertants (A and B), triplicate ~2 uL samples (800 ng phage protein) of PEG-precipitated and filtered virions (§2.15.4) were applied to nitrocellulose discs in the pattern indicated in Figure 6-5. After drying, blots were probed with 1:8,000, 1:32,000 or 1:128,000 dilutions in blocking buffer of (i) protein A-purified a-fl pAbs (§2.13.1) or E. co/i-absorbed (§2.13.3.a) (ii) FN2/4 or (iii) FS1/4 a-FHA pAbs or (iv) crude F N 1/4 a-FHA serum. Blots were washed, blocked, probed with 2° antibodies as described above (§2.18.1). Blots probed with a-fl pAbs were stopped after 6.5 min., while those probed with a-FHA pAbs or serum were stopped after 13.5 min. 116 2.19. ANALYSIS O F A S X - P R O T U R N S Methods employed in analysis of Asx-Pro turns are described in Chapter 3. 2.20. P R E D I C T E D SIGNAL PEPTIDE C L E A V A G E Each of the peptides encoded by the antibody-reactive FHA-70 and -80 library clones identified in Figure 6-1 (Chapter 6) were examined for signal peptide cleavage expected from their display in three alternative vectors: (i) fDRW70, which separates displayed peptides from signal peptides with the linker sequence A D G A G A ; (ii) an fDRW8nn vector, which separates displayed peptides with ADGP; and (iii) a vector which uses no linker, but rather fuses the displayed peptide directly to the signal peptide. Each of the peptide sequences shown in Table 2-VI reflects an FHA library clone as it would be expressed in one of these three vectors; these sequences were analysed by the neural networks developed by Nielsen et al. (1997) and available via the WWW server of the Center for Biological Sequence Analysis (http://www.cbs.dtu.dk/). C-scores (cleavage-site scores, one of three scores provided by the networks) and most likely cleavage positions are tabulated in Table 2-VI and summarized in Figure 6-9 (Chapter 6). 117 Table 2-VIh Predicted" signal peptidase cleavage of antibody-reactive FHA-70 and -80 clones. Clone ID b Last residue of native preprotein + 30 residues of N-terminal mature region0 C-scored I - a S . A D G A G A L S V S A G G N L R A K Q L V S S A Q L E V R G S . A D G P L S V S A G G N L R A K Q L V S S A Q L E V R G Q R L S V S A . G G N L R A K Q L V S SAQLEVRGQREVAL 0 . 5 2 1 0 . 427 0 . 2 7 4 I - b S . A D G A G A I A P R G W R A T T I S Q V D L H D L S A A R G S . A D S P I A P R G W R A T T I S Q V D L H D L S A A R G A D S I A . P R G W R A T T I S Q V D L H D L S A A R G A D I S G G 0 . 521 0 . 427 0 . 156 I - c S . A D G A G A D L S A A R G A D I S G E G R G A G S G A E T V S . A D G P D L S A A R G A D I S G E G R G A G S G A E T V E S S . D L S A A R G A D I S G E G R G A G S G A E T V E S C L A K 0 . 5 2 1 0 . 459 0 . 188 I - d S . A D G A G A L S A A R G A D I S G G V G T G A E T V E S C L S . A D G P L S A A R G A D I S G G V G T G A E T V E S C L A K S L S A . A R G A D I S G G V G T G A E T V E S C L A K P H T E 0 . 521 0 . 427 0 . 615 I - e S . A D G A G A S A A R G A D I S G G A G S G A E T V E S C L A S . A D G L S A A R G A D I S G G A G S G A E T V E S C L A K P S S A A . R G A D I S G G A G S G A E T V E S C L A K P H T E N 0 . 521 0 . 427 0 . 5 1 8 I - f S . A D G A G A L S A A R G A D I S G E G R G V G S G A E T V E S . A D G P L S A A R G A D I S G E G R G V G S G A E T V E S C S L S A . A R G A D I S G E G R G V G S G A E T V E S C L A K P 0 . 521 0 . 4 2 7 0 . 615 I - g S . A D G A G A L S A A R G A D I S G E G R G A G S G A E T V E S . A D G P L S A A R G A D I S G E G R G A G S G A E T V E S C S L S A . A R G A D I S G E G R G A G S G A E T V E S C L A K P 0 . 521 0 . 4 2 7 0 . 615 I - h S . A D G A G A S A A R G A D I S G E G G A G T G A E T V E S C S . A D G P S A A R G A D I S G E G G A G T G A E T V E S C L A S S A A . R G A D I S G E G G A G T G A E T V E S C L A K P H T 0 . 521 0 . 427 0 . 5 1 8 I I S . A D G A G A A L S I D S M T A L G A G A G A E T V E S C L A S . A D G P S A A R G A D I S G E G G A G T G A E T V E S C L A S S A A . R G A D I S G E G G A G T G A E T V E S C L A K P H T 0 . 5 2 1 0 . 427 0 . 518 I l l - a S .ADGAGAMTVRDVAAAADLALQAGDASGAGA S .ADGPMTVRDVAAAADLALQAGDASGAGAET S . M T V R D V A A A A D L A L Q A G D A S G A G A E T V E S C 0 . 521 0 . 5 1 4 0 . 5 1 4 I l l - b S .ADGAGAMTVRDVAAAADLALQAGDALPGAG S . A D G PMTVRDVAAAADLALQAGDAL PGAGAE S . M T V R D V A A A A D L A L Q A G D A L P G A G A E T V E S 0 . 5 2 1 0 . 427 0 . 2 3 6 I I I - c S . A D G A G A R D V A A A A D L A L Q G V G T G A E T V E S C S . A D G P R D V A A A A D L A L Q G V G T G A E T V E S C L A S . R D V A A A A D L A L Q G V G T G A E T V E S C L A K P H T 0 . 521 0 . 427 0 . 370 I V S . A D G A G A I V V E A G E L V S H A G G A G S G A E T V E S S . A D G P I V V E A G E L V S H A G G A G S G A E T V E S C L S I V V E A . G E L V S H A G G A G S G A E T V E S C L A K P H 0 . 521 0 . 427 0 . 3 1 6 118 V S .ADGAGATKGEMQIAGKGGGSPGAGSGAETV S .ADGPTKGEMQIAGKGGGSPGAGSGAETVES S T K G E . M Q I A G K G G G S P G A G S G A E T V E S C L A K 0 . 521 0 . 427 0 . 2 1 7 V l - a S A D G . A G A D Y T V S A D A I A L A G V G T G A E T V E S C S . A D G P D Y T V S A D A I A L A G V G T G A E T V E S C L A S . D Y T V S A D A I A L A G V G T G A E T V E S C L A K P H T 0 . 521 0 . 4 5 9 0 . 2 0 8 V l - b S . A D G A G A T V S A D A I A L A A Q G A G T G A E T V E S C S . A D G P T V S A D A I A L A A Q G A G T G A E T V E S C L A S T V S . A D A I A L A A Q G A G T G A E T V E S C L A K P H T 0 . 5 2 1 0 . 427 0 . 4 8 7 V I I S . A D G A G A N K I R L M G P L Q G V G S G A E T V E S C L A S . A D G P N K I R L M G P L Q G V G S G A E T V E S C L A K P S . N K I R L M G P L Q G V G S G A E T V E S C L A K P H T E N 0 . 5 2 1 0 . 4 2 7 0 . 2 1 1 V I I I S . A D G A G A I T V T S R G G F D N E G K M E S N K A G A G A S . A D G P I T V T S R G G F D N E G K M E S N K A G A G A E T S . I T V T S R G G F D N E G K M E S N K A G A G A E T V E S C 0 . 521 0 . 427 0 . 179 I X - a S.ADGAGAHLRNTGQVVAGQGVGTGAETVESC S .ADGPHLRNTGQVVAGQGVGTGAETVESCLA S H L R . N T G Q V V A G Q G V G T G A E T V E S C L A K P H T 0 . 5 2 1 0 . 427 0 . 2 4 6 I X - b S A D G . A G A V V A G H D I H I M G V G T G A E T V E S C L A S . A D G P V V A G H D I H I M G V G T G A E T V E S C L A K P S W A . G H D I H I M G V G T G A E T V E S C L A K P H T E N 0 . 5 2 1 0 . 4 2 7 0 . 618 X - a S A D G . A G A N K L G R I R A G E D M Q G A G A E T V E S C L S . A D G P N K L G R I R A G E D M Q G A G A E T V E S C L A K S . N K L G R I R A G E D M Q G A G A E T V E S C L A K P H T E 0 . 5 2 1 0 . 427 0 . 2 0 6 X - b S A D G . A G A N K L G R I R A G E D M H L D A P R I E G V G T S . A D G P N K L G R I R A G E D M H L D A P R I E G V G T G A S . N K L G R I R A G E D M H L D A P R I E G V G T G A E T V E 0 . 5 2 1 0 . 4 2 7 0 . 145 X l - a S . A D G A G A L D Q N R Y E Y I W G L Y P G A G A G A E T V E S . A D G P L D Q N R Y E Y I W G L Y P G A G A G A E T V E S C S . L D Q N R Y E Y I W G L Y P G A G A G A E T V E S C L A K P 0 . 5 2 1 0 . 4 2 7 0 . 2 4 9 X l - b SADG.AGADQNRYEYIWGLYQGVGSGAETVES S . A D G P D Q N R Y E Y I W G L Y Q G V G S G A E T V E S C L S D Q N . R Y E Y I W G L Y Q G V G S G A E T V E S C L A K P H 0 . 5 2 1 0 . 459 0 . 311 X I - c S . A D G A G A N R A G L S P A T W N F Q S T Y E L L D Y L L D S . A D G P N R A G L S P A T W N F Q S T Y E L L D Y L L D Q N S N R A . G L S P A T W N F Q S T Y E L L D Y L L D Q N R Y E Y 0 . 521 0 . 427 0 . 4 8 8 X l - d S . A D G A G A E Y I W G L Y Q T Y T E W S V N T L K N L D L G S . A D G P E Y I W G L Y Q T Y T E W S V N T L K N L D L G Y Q S . E Y I W G L Y Q T Y T E W S V N T L K N L D L G Y Q A K P A 0 . 5 2 1 0 . 4 7 9 0 . 2 0 2 X I - e S . A D G A G A T E W S V N T L K N L D L G Y Q A K P A P T A P S . A D G P T E W S V N T L K N L D L G Y Q A K P A P T A P P M S . T E W S V N T L K N L D L G Y Q A K P A P T A P P M P K A P 0 . 5 2 1 0 . 4 2 7 0 . 2 0 8 X I I S .ADGAGAVHDQLGQRYGKALGGMDAETKEVD S .ADGPVHDQLGQRYGKALGGMDAETKEVDGI S V H D . Q L G Q R Y G K A L G G M D A E T K E V D G I I Q E F 0 . 5 2 1 0 . 427 0 . 5 1 4 X I I I S . A D G A G A G R P E G L K I G A H S A T S V S G S F D A L P S . A D G P G R P E G L K I G A H S A T S V S G S F D A L P G A S . G R P E G L K I G A H S A T S V S G S F D A L P G A G A G A 0 . 5 2 1 0 . 427 0 . 2 0 1 119 X I V S . A D G A G A L A A V L V N P H I F T R I G A A Q T S L A D G 0 . 521 S . A D G P L A A V L V N P H I F T R I G A A Q T S L A D G A A 0 . 427 S L A A . V L V N P H I F T R I G A A Q T S L A D G A A G P A G 0 . 327 Predicted signal peptide cleavage, based on the neural networks of Nielsen et al. (1997). b A s shown in Figure 6-1 (Chapter 6). cEach table entry shows three sequences, each including up to 30 residues o f the N-terminal mature region of a displayed peptide. One o f the sequences is that o f an antibody-reactive FHA-l ib ra ry clone. The other two reflect the FHA-derived peptide as it would be expressed in an alternative vector, as described in §2.20. The first line of each entry represents the peptide in fDRW70 , which separates displayed peptides from signal peptides with a A D G A G A linker peptide. The second line represents the peptide in an fDRW8«« vector, in which a A D G P linker is employed, while the third line represents the peptide in a vector in which no linker is employed. The symbol " ." indicates the position yielding the highest C-score and the putative cleavage site for C-scores >0.49. In wild-type F f phage, cleavage is believed to occur on the C-terminal side of the first Ser residue shown. Each sequence was preceded by the preprotein sequence V K K L L F A I P L V V P F Y S H when submitted to the neural networks. d C-score, or cleavage-site score, as described in Nielsen et al. (1997) and in §2.20. 120 Chapter 3 The "Asx-Pro turn" as a local structural motif stabilized by alternative patterns of hydrogen bonds, and a consensus-derived model of the sequence Asn-Pro-Asn 3.1. A B S T R A C T Analyses of databases derived from the Brookhaven Protein Data Bank have identified a set of related turn structures formed by the sequence Asx-Pro-Xxx n . In a variety of flanking structural contexts, more than 60 per cent of Asx-Pro sequences adopt a turn conformation stabilized by a set of alternative hydrogen bonds among the sidechain O and backbone C=0 carbonyl oxygens of Asx (residue /) and the backbone N H of residues i+2, i+3 and in some cases i+4. In contrast, the structures adopted by Ser-Pro, His-Pro and other Xxx-Pro sequences reflect more heterogenous hydrogen-bonding patterns. As expected, structures formed by Asx-Pro-Asx are similar to those formed by Asx-Pro-Xxx n but in some cases additional hydrogen bonds are formed between the Asx sidechains. Hydrogen bond patterns within Asx-Pro and Asn-Pro-Asn turns are consistent with published N M R studies of helical (Asn-Pro-Asn-Ala) n peptides, indicating that a consensus structure reflecting these hydrogen bonds can serve as a partial model of the Asn-Pro-Asn-Ala tetrapeptide repeats of Plasmodium falciparum circumsporozoite protein. 3.2. INTRODUCTION Much of our understanding of protein structure has derived from analyses of databases of known three-dimensional structures, particularly the Brookhaven Protein Data Bank (PDB; Bernstein et al. 1977), such as those of backbone dihedral angles (Ramachandran et al. 1963; Ramachandran & Sasisekharan 1968), amino acid sidechain conformations (McGregor et al. 1987; 121 Schrauber et al. 1993; Dunbrack & Karplus 1994), and backbone and sidechain hydrogen bonds (Kabsch & Sander 1983; Ippolito et al. 1990; Stickle et al. 1992; McDonald & Thornton 1994). From these and other studies, it is clear that some amino acids influence local conformation more than others, proline being an obvious example: its pyrrolidine ring limits both its own (b,i|/ conformational space and that of the preceding residue (MacArthur & Thornton 1991). Amino acids with short polar sidechains such as aspartate, asparagine and serine are other examples. These, together with proline, occur frequently in B-turns (Chou & Fasman 1978, 1979; Wilmot & Thornton 1988) and at the N-termini of a- and 310-helices, and are strongly associated with certain clusters of backbone conformations (Presta & Rose 1988; Richardson & Richardson 1988; Karpen et al. 1992; Dasgupta & Bell 1993). As well, certain structural motifs have been shown to be defined by local sequence motifs. Examples include (a) the "tyrosine corner" (Hemmingsen et al. 1994), in which a tyrosine (in a sequence which includes proline) hydrogen bonds its sidechain OH to the backbone N H or CO of a preceding residue; (b) type VI turns formed by aromatic amino acids flanking proline (Yao et al. 1994a, 1994b); (c) ring motifs formed by hydrogen bonding of Asn or Gin sidechains to backbone atoms (Le Questel et al. 1993); and (d) the S X X E "capping box" in which reciprocal sidechain-mainchain H-bonds (Harper & Rose 1993), together with hydrophobic interactions (Seale et al. 1994), stabilize an a-helix cap. The current study describes another local sequence and structural motif, identified in a narrowly-focussed study of Asx-Pro and other Xxx-Pro sequences. Interest in these specific sequences arose from investigations (Chapter 4; see also Wilson et al. 1997) of Asx-Pro-containing peptides derived from the circumsporozoite protein of Plasmodium falciparum. The current study was prompted (a) by the intriguing findings of Richardson and Richardson (1988) and related studies (Presta & Rose 1988; Karpen et al. 1992; Lyu et al. 1992; Chakrabartty et al. 1993; Dasgupta & Bell 1993; Forood et al. 1993) that there is a striking propensity for asparagine, 122 aspartate and serine at an a-helix N-cap and for proline in the first helix position, (b) by a similar positional distribution of these residues in fl-turns (Chou & Fasman 1978, 1979; Wilmot & Thornton 1988) and (c) by the Presta and Rose hypothesis (1988) that an a-helix N-cap can be stabilized by hydrogen bonding between the sidechain of a polar residue flanking the helix and the initial backbone N H group. Given that characteristics of asparagine, aspartate, serine and proline must independently lend themselves to formation of H-bonds required for a turn or helix cap, expected conformations of these residues in combination (Asx-Pro or Ser-Pro) should be few and predictable. Because previous studies have not focussed on these amino acids in combination and, moreover, have generally focussed on a single structure class (a-helix, 3 1 0-helix or B-turn), several PDB-derived databases representative of all protein structure classes were analysed to specifically examine H-bonds formed by these residues in combination. 3.3. A N A L Y T I C A L M E T H O D S 3.3.1. Molecular databases and datasets Three datasets were employed, each consisting of Brookhaven PDB coordinates for a number of proteins, their corresponding secondary structure profiles (DSSP profiles) defined by the program DSSP (Kabsch & Sander 1983), and additional H-bond information as described later. PDB coordinates were obtained from the Internet archive of the Brookhaven National Laboratory. Most DSSP profiles were obtained from the Internet archive of the European Molecular Biology Laboratory (EMBL) at Heidelberg, Germany; others were generated in-house. Each of datasets A and B was chosen to be representative of all protein structure classes. Dataset A consists of 80 non-homologous proteins, 77 of which had been refined to a resolution <3.0 A; their Brookhaven codes are as listed in Table 1 of Adzhubei and Sternberg (1993), except that entries 1HMQ and 4FD1 were replaced by more current entries 2HMQ and 123 5FD1, respectively. Dataset B consists of 102 non-homologous proteins refined to 3.0 A resolution or better, as described in Boberg et al. (1992). Brookhaven codes for dataset B (chain designations in parentheses) are: 1ACX, 1BP2, 1CCR, 1CLA, 1CSE(E,I), 1CTF, 1CTX, 1ECA, 1ETU, 1FC2(C,D), 1GCR, 1GD1(0), 1GOX, 1GP1(A), 1HIP, 1HOE, IIIB, 1LDM, 1LRD(4), 1LZ1, 1NXB, 1PCY, 1PHH, 1PRC(C,L,M,H), 1PYP, 1R69, 1RHD, 1SGT, 1SN3, 1TIM(A), 1TNF(A), 1UBQ, 1UTG, 1WSY(A,B), 256B(A), 2AAT, 2AZA(A), 2CAB, 2CCY(A), 2CDV, 2CI2, 2CPP, 2CTS, 2CYP, 2FB4(H,L), 2FXB, 2GBP, 2GN5, 2HHB(A,B), 2HLA, 2HMG(A,B), 2LH2, 2LTN(A,B), 2MHR, 20VO, 2PAB(A), 2PAZ, 2PFK(A), 2RNT, 2RSP(A), 2SGA, 2SNS, 2SOD(0), 2SSI, 2TAA, 2TS1, 2WRP, 3ADK, 3APP, 3B5C, 3BCL, 3FXC, 3FXN, 3GAP(A), 3GRS, 3INS(A,B), 3 L Z M , 3PGK, 3PGM, 3RN3, 3TLN, 451C, 4CHA(A), 4HVP(A,I), 4PEP, 4XIA(A), 5CPA, 5CPV, 5FD1, 5PTI, 5RXN, 6ACN, 6AT1(A,B), 7API(A,B), 8ADH, 6AT1(A,B), 8CAT(A), 8DFR, 9PAP and 9WGA(A). Most analyses were performed twice, once for each dataset, to assess bias in choice of dataset. Although there is considerable overlap between these datasets (they have 55 proteins in common, and an undetermined number of proteins share at least limited local homology), the different methods (Boberg et al. 1992; Adzhubei & Sternberg 1993) used to select the proteins included in each dataset and differences in content were expected to preclude a biased analysis. Dataset C consists of 39 non-homologous protein fragments containing Asp-Pro-Asn (15 fragments), Asn-Pro-Asn (9 fragments) and Asn-Pro-Asp (15 fragments). These were identified by searching Release 14 of the NRL3D (Pattabiraman et al. 1990) database (from the Internet archive of the National Center for Biotechnology Information) for all occurrences of these tripeptides in proteins refined to better than 3.0 A resolution. Highly homologous or identical fragments were eliminated by choosing the entry with the highest resolution or, for multichain proteins, the lowest chain code. Only structures determined by X-ray crystallography were included. Structures with only C a coordinates were eliminated. A fragment in entry 4ENL 124 was eliminated because of high temperature factors. 3.3.2. Computer software The program DSSP was provided by U . Hobohm (EMBL, Germany) under academic licence. The program B A B E L version 1.05, provided by P. Walters and M . Stahl (Dolata Research Group, University of Arizona, U.S.A.), was used to calculate interatomic distances from PDB coordinates. The program WHATIF, provided by G. Vriend (EMBL, Germany) under academic licence, was used to identify H-bonds. The program R A S M O L version 2.3, provided by R. Sayle (1992) (University of Edinburgh, Scotland), was used to view and manipulate three-dimensional models of PDB coordinates, and to produce the molecular models shown in Figures 3-6, 3-9, 3-11 and 3-12. 3.3.3. Hydrogen bonds The program DSSP identifies backbone H-bonds using a simple energy model that considers both the distance between oxygen and nitrogen atoms in backbone C=0 and N H groups and the deviation from an optimal (for H-bond formation) angle between these groups. Using a relatively low cutoff value of -0.5 kcal mol"1 for binding energy (a good H-bond is about -3 kcal mol"1), DSSP allows C=0 and N H groups to be misaligned up to 63° at the ideal (2.9 A) O-N distance, and allows O-N distances of up to 5.2 A for perfect alignment. H-bonds identified by the program DSSP were used in the analysis presented in Table 3-1 and Figure 3-3, and for identification of backbone-backbone H-bonds in analyses that also included sidechain-backbone H-bonds (Figures 3-1, 3-2, 3-4, 3-5, 3-10 and 3-11). Because DSSP identifies only backbone H-bonds, a separate criterium was employed to identify sidechain-backbone H-bonds: a sidechain-backbone H-bond was considered to exist i f the interatomic distance between a sidechain donor/acceptor heavy atom (O or N) and a backbone acceptor/donor heavy atom was <3.5 A. This mixture of criteria appears to have been adequate and seems warranted by five 125 sa}ON Is 5 5 - ° cr CD ro c CD CD <0 0) ">'(/) O CD i t o > < c ro < ° CD Q - L U Q O CJ : =c nz h— •= =c oo : x x x x x u I X I X I I X h ' ' i ' ' i OO UJ C D X X C O O O : rc re - 3Z in _<c<i: LU Li I <c _ i u _ ^ >-< ^ 0 : x o x ^ < a < i a _ i ^ -— i r - > : U J i—i ^ : . CL. C L . C L . c Q_ <t I <£ oO = C o o s : —i :=> Q_ OO > —I OO L U U _ O > - L U J OO "Z. Q_ CD >-UJ H Lf) If) H Lf) exj ro »—i c\j Cd OO OO i—I H U U U - i W X U U i - H C \ J C \ J C \ J C \ J ^ J - C \ J C \ J o C J Cc^-O CD CO oo r--C J o_ C J C Q u. CD HHHHHH HHHHHH HHHHHH S-BSSS c n in rr C0-SS C0-1 S3 i C £ - C J 1— zlz 1— re c t - C J OO C £ - C J L U l i l HI- 1— ZLZ 1— rc ~r~ ~r~ •NH-:EE NH-L U UJ co- rc nr CO-UJ C0-UJ <£ UJ >- _j Q i - C J ICL 1 c * : - C J C D OO / ' Z ' o i—i u_ C D < \— OO C J ) S <t 3 1 CL. CL. ! Q ^ Q _ OP NH ' C D ^ NH C D , - nr :=> CO-FDKVI LAAY1 CO-CO- OA] q , o-' C J FDKVI LAAY1 / ci \ o cb V ! !! o <£ CTI ro 1 130 I 66 | -<_>-(_) ==: i lj) -<_>-o <=> i o CD 1ECD 3C2C X Q s: Q O OO CM OO O0| C D y C D h C D o C J o CJ a;-CJ c o -o V CJ-CJ-CJ S9J0N CT Cl) ro c = 1 8 s <D CO W "D 0) TD 13 C-D ro'5) O CD i t O ) x.E (/) ^ < 1= ro <2 K < ° CQ c r -Q ~ ro « c-c= CJ o;-CJ O CJ o^-CJ c o CJ -CJ o -zn oo i U J ni oo zlz i zlz ' re r— X X X i I ZCOOh— LU=C ih— ' X ( / ) X f - X X X ( / ) X ZC i f— LUZCOOI— I— X O O X X X X X i i i CD i » LU rC CD I— CD ZLZ > O I X X I i CQ i CDCDCDCDCDCDCDCDCDCDCDnZICIClX; iO0 • CjDCJ3CDOOCJ3CDC3CDCDCDniZCZn|—1—(—h— O O O O O O C D C D C D C D C D I — I— 1— I— I— I— I— i OO ' ' i i ' ' OO i OO OO CQ i i J 00 h OO L J OO i i X > 00 OO ' < ZLZ ZC 00 L U I— I— ZLZ ZLZ UJ ZLZ HZ ZLZ ZLZ ZLZ L Q O0 >- •—• —' 0O O Q_i—(Q_^CD—J — I Q X L L I Q 3 h u j > Q i Q < ^ s t o - i - i a : ^ ^ : CC\— O O ^ : ^ — I C D C D o O ^ O — \ ^ C d —loOoOO < i—i(/)<0 — I <C >- CD >- Q —IU_U_CD^ — < L L O h - J f - Z y - I Q C O J Z Z > - < 0 0 a : y z < y c o o o < ^ o < z > u j Q ^ z f -L L D . C L C L Q . Q . Q . Q . C L Q . D . Q . C L C L L L Q _ Q . C L CDh—5-5-<U_2:oOQ->Cli2:>.>.UJCDCD> • s : ^ i / i L i J Z L i . z i - u z c / ) Q ; x x t o L i j > i—( —I I— > C D I— L U CO CJ L U I I—I <C !^ ^ ^ c£ <c o oo > u_ ad i—< i— z^d^z—i L U <c C D —ii— C D > — < ^ c i . ^ _ i > < O X O Q ; _ I < ^ C D <C oo a. s; <c u_ oo u_ C D c o < u < > ^ u . C N J ^ J - L O L O u O C O r - O y D ' ^ - O CT> CT) L O CO '—I CT) < < < C Q < H H Q : C D C Q C Q C L Z : Q . X I — X C Q C Q H - o C Q >—< X Q - Q - < < < > - - J C O Q < Q X X < C v J U D -L L ( i 3 < U U U U h < < U < I I U U ^ < T^r^ojouc\joocM(^cocooooo^^cororor--. Cu o < cx) o C J O i C J ; c * ; - C J I c U - -Z. Li_ Q < C J U _ 3 Q S I I— — I > C D >—• >- =» S Q OO DC LU O LU <C I— U J > -u- Q ; n: >-=» —I I— CD < S> O O CJ t— < X LiJ C ) LO C > i-H CXJ J^" CO U0 Fig. 3-1. continued on following page 126 sajON CD Q-•3 w § Q .t >> 05 XI > N T 3 II o co C O "o (fl CD X I _ ! C X I 1 0 (fl o a) x.E co __ < 1= to m < ° pt: « Q - L U Q c. c; i n i X X X r -X X X I — —I — I I— o u_ <—. u_ : 3 < <£ —•! LU |— u _ X | _J CD I— h - o O Z O l CO CO cvi O o_ < < f— < o . a . < o u u u co L O L D co J co L U | — i : J L U CO L U CO CD -J CO > L U L U o : J i CO L U L U CD f-CO i L U CO CO I-) CO CO | — i CO CO h CO CO • . CO CO LU CO CO LU CO ' LU CO I— LU CO I— ~> |— co |— . : - L U i — C D i : - L U x cD co : - L U x CD i : - L U x C D i : •> L U x • co : - Q - U O • 3 > > ^ C D a Q X •=» h - ^ l — C D ^ —I — CO —I <t —I > i — y— > L U C D > < C Q Cd <—1 1 — • L U L U Q Q U J I i — • Cr. ^ <C >- CL I— 321 i—< Q_ Cd >- o_ <t > - L U > CO 3*3 <C I— Li_ LU Cd O =^  —I LU X CD ^ LU —I Li_ Q_ > i—i CO <£ —I Q I Q_ LU >— Q LU Cd > 3> ^ X CL CL i— 3ZI > - O O —I o z : ^ L U — I X a O - C L C L o_ o_ a_ z z: o Q _ Q _ Q _ a , Q _ Q _ o z o z : o o Q _ Q _ Q _ Q _ Q _ Q _ Q _ Z Q Z Z O O Z a_ a- a_ z: o z : C J ) 1 i to 3*3 Q_ I— < — • > C D U _ •—• O C D — LU ^ —i CD a_ h- < <_> 21 ^ O I— CD CO CD >- CO > i—y\— CD I— Q ^ Q L L < > ^ Q | — <£ > L L Q O Z < X CD U I Q_ X LU > CD ^ CD ^ <C <C CJ <C ^ LU —I ^ >- < > - < \— ^d CD Oi > U_ C L — <C U— LU >—iLJOZ< CD Cd. C(L\ CD => CD \— h- Q > J— <—I - J CO i CO CT) IJD r-. co «3 o -=3- I Ch LO O <^ > <-Q CO T - I CO r-. ^j- CO CO r-t CO CJ co i—i c\j n r n H H i < o < < <t • h s : o Q O o o < < m u z Q i 3) O X - J —JcOLOr— h— h-COCJ)CD HC\JC\JCSJC\JOJC\JC\JCMC\J(^COCO o < - i r o n n v o r o c \ j r - . o O-^-r-HC0r-HC0C0C00JC\J CL CL Q_ |— |— L L < U U U cocococorOLncocococo S9J0|SJ 0) Q_ B Q >> CO -O ^x> 1| 8 <u co -a T 3 (fl CD c-g o CD i t o> < s < s Q-uJo o CO i X X CD LU CO CD i X X CD LU CD CD i X X CD LU CD h CD CO X • i LU CD h CD CO I— i i LU CD I— I— I— I— I— CDCDr-- C D C D (• C D • X X I — ' CD CO h X X X i CD CO r X X X i CD h-3 3 - J > - 0 - l > > 0 a : i z - i L j _ x < i - Q -— o — i < Q h ^ —is: C D co •—"—11— • L U Q z : O h - Q U _ C D — I C O L L C O C L C L Q . C L Q _ C L Q . C L C L Q Z : Z : Q Z : Q O Z O O i > > Z : > _ i _ t L i _ L i _ > e J a > t _ 5 < D . Q . < > _ l H - i < O L L L U U I Q; _j > Q ^ ^ : •-. o < o a o o < > < > OCOOCOCvJ-^-COChCO T-ir—ir—i CO C\J CsJ CO < < < < i - I X Q - Q Z I U h h r^JQ<tO_XQr— <t<C —I — I O C D Q - < < U U T-HCNjco^ rLococococo Fig. 3-1 continued from previous page. C ON .9 ^ 1 ^ ' m 0> (33 (« u c 01 3 O" 0> 05 O u OH * c O .o o cd • c o x> o C 3 X) T3 C a T3 S O 9S CJ o co 0 L / -i-2 ^ O <D cn < cn X) TO 'o « O ui co 3 ' lo C (U . C s ° ° 1 J3 O O C •a c o a. CO (U fc O o T3 C eS <u c o X) M o ed X)  c3 u -a -S Z c o ° * H-r ^  " 2 T 3 Q . 53 M » co Lfl" 0.-T3 3 (S 2 o iX) 6 b • o =3 O o S t o .—t CO a) TO x: x i T 3 CJ s C CJ T3 a C J 2 T3 C O n. CO CJ £ o o CJ CT X 1 TO o C J CO 3 C J a - o CJ a c a> 3 cr CJ (O r-m O OH I a < T3 C ca Q , 1 ^ bfl co cj o M is T3 TO fJ X i CO £ I " < II . O O O •9 O 53 O X ! cj c o XI o cd X) i cj C o X) ^ o cd Xi oo -a . „ c x cd "a> co xT 2 " 1 Cd a" o CJ D . on , i o CJ o c CJ 3 cr CJ - H ^ CJ C i co co cj 2 §• CL CJ I co < PH M 5 ON O N " o 'C CJ cd ^ G 'co G >• 3 C co o c/T C § c2 Xi 53 xi CJ ,—, C ON o -8 <~ a ° TO CN X) v-* 1 -^3 cu S CJ CJ 3 + •S _ i > c ° S E 3^ —7 cd A. T3 1 11 cd G « O S cd X o 1^ -o ° r •k < c « 2 B CD ^ 3 PH GO cd _o (30 _ C •3 c o X) c CJ 0 0 o u. T 3 x; CQ o a -a ' c ^ TO CO CJ CJ 3 bO TD co 3 =0 •SPW" co M - ' Ui r- CJ CJ Q . CN CN . 5 O X) r o . O O O Q a. 3 o 6b CQ 0 . XI tin cj cd X) 1 S TO x; r- c/l O co <U T3 " 2 3 'cn O X) ^ 5 3 CO CJ •a o o cj c o XI o TO X! CJ S G + u: •I H o o + I 7 S CJ <D § G CJ - «i 3 -3 CN + r o c co B P. CJ O -o -a *- •§ "° co cd c s- -73 XI •S 53 -o c cj X) O XT 7 3 d c 3 O 3 - 9 -c 53 g c o <o X 5 <_ 5 o cd t/i cd ' V •S o i _ cj o G 3 7 'i 5 ^ 53 CJ f> c o •a — & 15 cd ^ o aj XJ CJ 3 G cu <Z> CJ 0 0 g Q T 3 127 B D PDB Entry (Chain) Asx a Asx-Pro-Asx and b in £ pos'n flanking residues o PDB Entry (Chain) Asx Asx-Pro-Asx and £ pos'n flanking residues o •z. x r o - N r o c r > \ ? R R - N H - C - C 0 - N - C - C 0 - N H - C - C O - N H - C - C 0 - N H - C - C 0 -1GAL 41 AARLTE NPN ISVLVI 1LGA(A) 278 LTQLGQ DPN AMTDCS f 1THG 493 SFANHH OPN VGTNLL c Y O ? R R - N H - C - C O - N - C - C O - N H - C - C O - N H - C - C O - N H - C - C O -11MDA(L) | 66 | S C Y N P P i D P N j K Y I T A Y | H N r 0 x r o •N-c.O c <"> C R R - N H - C - C O - N - C - C O - N H - C - C O - N H - C - C O - N H - C - C O -c R 1AAK 117 SLLCDP NPN SPANSE c - c - c o - N - C - C O - N H - C - C O - N H - C - CO-NH-C -C0-- 1CAU 123 A ILVLV NPD GRDTYK 1PK4 53 TMNYCR NPD ADKGPW c 11NDK | 150 | LTEVKP!NPN I LYE L i M 3GAP(A) 109 RQLIQV NPD ILMRLS c - C - C O - N - C - C O - N H - C - C O - N H - C - C O - N H - C - C O -3ICD 268 PWLKVKj NPNITGKEIV f x r o --NH c r > f R R • C - C O - N - C - C O - N H - C - C O - N H - C - C O - N H - C - C O -1ATN(D) 59 DYLNQD DPN TYHYVV c 1 FN R 160 EMLMPK DPN AT I IML 1FXD 35 . DKAVVI NPD SDLDCV 1LF I 424 QQSSDP DPN CVDRPV 1RBP 140 SFVFSR DPN GLPPEA 2APR 59 SGQTKY DPN QSSTYQ 2FCR 125 KPVGFS NPD DYDYEE 2FXB 76 DEPFDG DPN KFE e 2GST(A) 116 LIMLCY NPD FEKQKP 2TPR(A ) 141 LVRESA DPN SAVLET e 3PSG 57 SDHNQF NPD DSSTFE 4 ICB 19 YAAKEG DPN QLSKEE c 4MT2 2 M DPN CSCATD c t • C -I4TMS - N H - C O - N - C - C O - N H - C - C 0 - N H - C - C 0 - N H - C - C O -285 APTLQL i NPD i KHDIFD c C ^ \ C R R - N H - C - C O - N - C - C O - N H - C - C O - N H - C - C O - N H - C - C O -| 3 S D H ( A ) | 86 I F l DQLD i NPD!DLVCVV I c N r 0 R - N H - C - C O - N - C - C O - N H - C - C O - N H - C - C O - N H - C - C O -|1ATR 364 ELNKSI INPD IEAVAYG d x r o N T 0 9 r > C, - N H - C - C O - N - C - C O - N H - C - C O R R N H - C - C O - N H - C - C O -2BB2 95 K ITLYE NPN iFTGKKM d 3PGM 141 ERYKYV DPN IVLPETE d 5CPA 112 F L E I V T NPN IGFAFTH K N r 0 ( O - N H - C - C O - N - C II ACE I 1GPB N^.0 C R R • C O - N H - C - C 0 - N H - C - C 0 - N H - C - C0-481 558 TFAKTG EYKVHI NPN NPN ESKWP SLFDVQ ,«o- (v c .O-C ^ ' , C \ R R - N H - C - C O - N - C - C O - N H - C - C O - N H - C - C O - N H - C - C O -I 1 G D K 0 ) I 22 I FRAALKINPD l l E V V A V I M -NH x r o • C - C O - N - C -x- c.o C R R C O - N H - C - C O - N H - C - C O - N H - C - C O -1AAK 115 IQSLLC DPN PNSPAN 1ACE 85 SGSEMW NPN REMSED 1BAA 139 GVDLLA NPD LVATDA 1CAU 150 T P F Y L I NPD NNQNLR 1NN2 339 SNSNCR DPN NERGTQ 2HPD(A) 192 NKLQRA NPD DPAYDE 3TLN 183 EFYANK NPD WEIGED Fig. 3-2. Local H-bonds in Asx-Pro-Asx sequences, Dataset C. Classification of local sidechain-backbone and backbone-backbone H-bonds formed in 39 Asx-Pro-Asx sequences identified in Release 14 of the N R L 3 D database. Groups A to F correspond to the 51 per cent of (20 of 39) Asx-Pro-Asx sequences that form both sidechain-backbone and backbone-backbone H-bonds, summarized in Figure 3-10, part A ; groups G to J, to the 23 per cent (9 of 39) that form only backbone-backbone H-bonds; groups K and L , to the 8 per cent (3 of 39) that form only sidechain-backbone H-bonds; and group M , to the 18 per cent (7 of 39) that form no local H-bonds. Superimposed wireframe models of groups A to F are shown in Figure 3-11 (parts A and B) ; groups G to M in Figure 3-11 (C and D). aPosition of Asx f in the sequence AsXj-Pro-AsXj. bUnderlined residues are within 3 | 0 - or a-helices c d T h e indicated i-i+2 (note c) or i-i+3 (note d) backbone-backbone H-bonds does not meet all criteria o f W H A T I F , but meets criteria of DSSP. e ' f The indicated sidechain-backbone (note e) or sidechain-sidechain (note f) H-bond does not meet all criteria of W H A T I F , but meets theinteratomic distance criterium. 128 considerations, (i) H-bond geometry is quite variable (McGregor et al. 1987; Ippolito et al. 1990; Jeffrey & Saenger 1991; Stickle et al. 1992). (ii) Possible biases in protein crystal structure refinement procedures such as the more accurate location of backbone versus side chain atoms (Richardson 1981) and stricter constraints for covalent versus H-bonds, as well as the relatively low absolute resolution of most structures, suggest that stricter criteria should not be applied, in particular to the identification of sidechain-backbone H-bonds. (iii) The program WHATIF was employed for an independent validation of the mixed criteria. Importantly, most H-bonds identified by DSSP or by the simple distance criterium were also identified by WHATIF, which considers H-bond geometry; differences have been noted in the detailed analyses presented in Figures 3-1 and 3-2. (iv) For Asx-Pro sequences in dataset A , C=0(i)<—NHf;'+«n=3 o r ^ backbone H-bonds identified by DSSP (Figure 3-1) were compared with those identified by the simple length criterium. For 61 of the 79 sequences, the methods agreed; disagreement in 14 of the remaining 18 sequences concerned whether an Asx C=0(i) group formed H-bonds with both versus only one of NH(/+5) and NH(/ + 4). Importantly, discrepancies between the two methods would not have affected the identification of 36 Asx-Pro sequences as "Asx-Pro turns." (v) Importantly, superimposed wireframe models of structures identified as "Asx-Pro turns" (Figures 3-6 and 3-11) strongly suggest that the criteria employed for H-bond identification were reasonable. Nitrogen and oxygen atoms are not readily distinguished by protein X-ray crystallography techniques and these two atoms are usually assigned on the basis of hydrogen bonding. Considering this, and that McDonald and Thornton (1995) have found that 15 per cent of Asn sidechains would be more favourably oriented for hydrogen bonding i f the nitrogen and oxygen designations were reversed, the estimates of H-bonds involving Asn sidechains would seem conservative. 129 3.4. R E S U L T S Many elements of secondary structure can be defined by their hydrogen bonding patterns. Thus the DSSP (Kabsch & Sander 1983) algorithm defines turns and helices in terms of local H-bonds between backbone C=0 and N H groups of residues i and i+n (3<rc<5) respectively. A single such i<—i+n H-bond defines a turn (/z-turn) while repeating turns define 3 ] 0 - , a- and 7i-helices as a series of 3-, 4- and 5-turns. This definition allows that statistical analyses of differences between local H-bonds formed by residues preceding a proline versus a non-proline can provide convenient and appropriate measures of proline's influence on local structure. Such an analysis was performed for two independently-selected sets of proteins, datasets A (80 proteins) and B (102 proteins), each broadly representative (Boberg et al. 1992; Adzhubei & Sternberg 1993) of all protein structural classes. To avoid bias, two datasets were employed; since the separate analyses yielded similar results, only parenthetical reference has been made to the results for database B. 3.4.1. Residues not preceding proline favor i*—i+4 H-bonds; proline favors i*— i+3 H-bonds When averaged across all amino acids, residues preceding a non-proline favor i*—i+4 over other local (*'«—/+nn=2 3 5 ) backbone H-bonds (Table 3-Ia). Thus, the 23.4 per cent (25.0%, dataset B) of residues forming /«—i+4 H-bonds is 2.9-fold greater than the 8.2 per cent (8.3%, dataset B) forming z<—/ + 3 H-bonds, the next favoured conformation. The importance of i*—i+4 H-bonds to protein structure is apparent: they account for 39.6 per cent (41.2%, dataset B) of all (local and non-local) backbone H-bonds. The importance of local H-bonds is similarly evident: 63.5 per cent (64.7%, dataset B) of backbone H-bonds are local, a finding consistent with a previous survey (Stickle et al. 1992). In contrast, residues preceding proline favor i<—i+3 over other local backbone H-bonds (Table 3-Ib): the 21.1 per cent (20.0%, dataset B) of residues forming i*—i + 3 H-bonds is 1.5-130 Table 3-1. Effect3 of proline on backbone hydrogen bonding of preceding residue. Number of residues Per cent of total number of residues forming H-bonds forming C=0(^~NH(7+«J backbone H-bonds b , per 100 residues0 Dataset A Dataset B Dataset A Dataset B (a) Residue i+l is not proline Local H-bonds C = 0 $ - N H ( / + 2 ; 4.9 4.8 8.3 7.9 C = 0 $ - N H ( i + 3 ) 8.2 8.3 13.9 13.7 C=0(^«-NH(7+4; 23.4 25.0 39.6 41.2 C=O^-NH^/+5; 1.0 1.1 1.7 1.8 Local H-bonds 37.5 39.2 63.5 64.7 Non-local H-bonds C=0(i)-NH(i+n), n>5 or n<0 21.6 21.4 36.5 35.3 Local + Non-local H-bonds 59.1 60.6 100.0 100.0 (b) Residue /+1 is proline Local H-bonds C = 0 $ ~ N H ( / + 2 ) 4.7 4.3 8.3 7.6 C = 0 $ - N H f i + 3 ; 21.1 20.0 37.2 35.4 C=0(i)^U(i+4) 14.2 15.0 25.0 26.5 C=0(z>NH(7+5j 0.2 0.2 0.4 0.4 Local H-bonds 40.2 39.5 70.9 69.9 Non-local H-bonds C=0(i)-NH(i+n), n>5 or « < 0 16.5 17.0 29.1 30.1 Local + Non-local H-bonds 56.7 56.5 100.0 100.0 a Table provides an overview of the effect of proline on local backbone hydrogen bonding by allowing comparison of the fractions of residues at position / that form C = 0 ( 7 ) " - N H ( 7 + « 2 < n S 5 / ) backbone H-bonds when position /+1 is occupied by (i) non-proline or (ii) proline residues. Numbers in bold are referred to specifically in the text of the Results section. b F o r simplicity, only a single H-bond (the strongest, according to the program DSSP) was counted for each C = 0 group. Because few C=0 groups form multiple H-bonds and similar results (not shown) were thus obtained when up to two H-bonds were counted for each C = 0 group, I consider that these statistics give an adequate measure of backbone H-bonds. cSummary is based on analysis of (a) 13,900 (dataset A ) and 21,395 (dataset B) residues preceding a non-proline residue, and (b) 660 (dataset A ) and 1,065 (dataset B) residues preceding proline. 131 so bo <3 Q, bo K ? O K O K to <u 3 X m Si x Cd •3 -3 2 •-Q. o 0) co - -a o ^ « c '« £ — x> § p cd e cd T 3 3 CO X CQ 3 x H eio B •3 e o e <u 6D O u 12 5-u g i - o O ^ in *0> ' O ^ O >~> - ° cd T 3 <C. .a a. 3 1 ^ § 2 1 «> o 1 1 1 C*H o o ca. co X d o o c to CJ . CD > U H • - o 2 : * B o 03 o. 3 CD X O CL, s—' "O .5 o CO o C o IS c ca -P o o O H bO -a CO cd C3 fS CO _ 3 a. "2 CO 12 c £ o o o in °-& e 2. 1 0 S fj tUO £ CO •> co « CO o X x CO hH o .Q .* u « -Q e o o a w T 3 c/i C CO o -a "? 'in X CO 1- CO _ c (S3 O 3 X -O •— o .> cd T 3 - ° •s ^ 2 S £ X •c ^ E ° 3 © = <+-CO O 00 i m a z ca ~s X) ca CO o ca co co X CO 3 ca -a o 73, O H <D « 1: •° o O " CO m T - , 3 e -o rt « U CO CO 3 co r s J 3 co CO c * -j= c2 ^ tyi i - 3 _CO O ca o B 2 CO (30 Sa .2 « T 3 "O C c o X o o CO X c o x ca o Q. CO /--> 3 T 3 CO -o CO ca T3 =a co < ^ 2 • - too • O ca o E c UJ ca IZ n. i E ^ So 3 ° u E "ca c£ ca ca -a _bo : ca co 3 -3 H CO co 3 CO c« 3 =a T 3 J> •55 E u- ca + ca co C C 4 . x O O CO £ § 1 O c ° - ' b S^? „ cd u td D . 3 <« E 3 "N O 1 / 1 CO •2 CO j— CO ~ ^ o ^ S ca o CLT ca O co 3 2^ i CO o <-O C cd ~ CO o o a. >> flgUl cd tive flgUl C O tive C O CO o CO X L H CO C O -a D. m #C X 3 o CO T 3 >? X m" rate CO X rate CO CO CO X c E :d int bo < C O enu :d int ack CO CO C O X m o C O one-cd "ca c CO scu one-T 3 L H _c cd -a m o suo C cd cd suo 'cd X td ofb o X ofb o CO tion ca cd tion cC 'co CO tion ese •T3 cd ese X co 3 X ned X cS ned CO CO i— td X 3 f i c2 •4—» m T 3 0C( ds _ 3 an ses c m ses bo O echain crea I X ided echain . 3 td _> pis Xj CJ _C pis "cd c CO . 3 _N c/l c cd m m m cd E X> X o ive mp hain' td _cd CO hain' _cd o o CO L H ca CO Q. L . < ^ o O 132 E c o ro si ro o cu ro " O o CO a. c V } 3 S E } B Q 133 fold greater than the 14.2 per cent (15.0%, dataset B) forming /<—i+4 H-bonds. These findings hold true for individual residues: for each of glycine, nine polar residues and for groupings of other residues, /<—z+4 backbone H-bonds predominate over i*—i+3 H-bonds by a factor of 2.1:1 (Ser) to 5.3:1 (Arg) when these residues precede a non-proline (Figure 3-3a). Similarly, for most residues (Glu being a notable exception), i<—i + 3 H-bonds predominate over /<—i+4 H-bonds by factors as high as 2.2:1 (Asp), 3:1 (Arg) and 4:1 (Lys) when these residues precede proline (Figure 3-3b). 3.4.2. Short polar sidechains in residues preceding proline favor i*—i+3 H-bonds The effect of proline in favoring i*—i + 3 backbone H-bonds is most pronounced for residues with short polar sidechains and for glycine (Figure 3-3b and c). Thus, the fractions of serine, threonine, aspartate, asparagine, histidine and (anomalously) glycine forming i*—i+3 H-bonds increase by 14.6 (Thr) to 36.1 (Asp) per cent (dataset A; Figure 3-3c) when preceding proline versus non-proline, while the fractions forming i*—i+4 H-bonds are reduced by only 0.9 (Thr) to 4.7 (His) per cent. Overall "gains" for local H-bonds are greatest for aspartate and asparagine: 73.8 and 62.2 per cent, respectively, of these residues form i-*—i+n2<n<5 H-bonds when preceding proline (dataset A , Figure 3-3b). In contrast, residues with polar atoms more distant from the cc-carbon tend to score only modest gains in those fractions forming i<—i+3 backbone H-bonds and to score "losses" of i<—i+4 H-bonds. For example, the fraction of glutamine residues forming i<—i + 3 H-bonds increases only 5.8 per cent when glutamine precedes proline versus non-proline, while the fractions of glutamate, glutamine, arginine and lysine forming i<—i+4 H-bonds decrease 8.8 to 27.3 per cent when these residues precede proline versus non-proline (dataset A , Figure 3-3c). A similar trend occurs with other residues (A, V , I, L , P, M , F and W in one-letter code; detailed data not shown, see Figure 3-3 for summary data). Moreover, the total number of local H-bonds for this heterogenous group of residues is smaller when these residues precede 134 proline versus non-proline. Small increases in local H-bonds occur for cysteine and tyrosine (data not shown). 3.4.3. Asx-Pro favors a combination of sidechain-backbone and backbone-backbone H-bonds Since the greatest propensity to form local backbone H-bonds occurs with aspartate, asparagine, serine and histidine preceding proline, the analysis was extended to include potentially turn-stabilizing i*—i+n2<n<5 H-bonds from the polar sidechains of these residues to the peptide backbone. Significantly, Asx-Pro sequences were found to have a greater tendency to form combined sidechain-backbone and backbone-backbone H-bonds than do Ser-Pro and His-Pro sequences (Figure 3-4). Thus, 68 per cent (64%, dataset B) of Asx-Pro peptides form C=O0')*-NH(/+n 2 < n < 5 ) backbone H-bonds, 49 per cent (52%, dataset B) form O NH(/+n n =2 3) sidechain-backbone H-bonds and 47 percent (46%, dataset B) form these H-bonds in combination. In contrast, and although 61 per cent (55%, dataset B) of Ser-Pro peptides form C=0(i)<—NHfi+/i2<n<5 ) backbone H-bonds, only 15 per cent (18%, dataset B) form both sidechain-backbone and backbone-backbone H-bonds. Moreover, those sidechain-backbone H-bonds that are formed are more diverse, involving backbone N H and C=0 groups, reflecting alternative roles of serine's polar sidechain as H-bond donor or acceptor. Similarly, only 21 per cent (15%, dataset B) of His-Pro peptides form a combination of complementary (and diverse) sidechain-backbone and backbone-backbone H-bonds. 3.4.4. Similar structures are formed by a variety of Asx-Pro H-bond conformations The 47 per cent of Asx-Pro sequences (46%, dataset B) that form complementary sidechain-backbone and backbone-backbone H-bonds represents 37 of the 79 Asx-Pro sequences of dataset A (56 of 121, dataset B). Importantly, 36 of these 37 sequences (55 of 56, dataset B), form turn structures stabilized by a limited set of H-bonds among the sidechain O 5 and backbone C=0 carbonyl oxygens of Asx(i) and the backbone N H of residues i+2, i+3 135 V leseiea a laseiea t > Q, % CO •a c o x i i c CO - C o c co co o o 'O 00 cn CO CN II c N-*~ CO T _^ CM Csl CO LO with: +3), ;'+3) lone erall ^ ' > o II o O CO CO CO II c m T CN m LO T — CO 0 + CN ^ + o n o o a) > o % o CD cn II c K % V CO CD CNI CNI CNI 0 0 C D 0 (;') with: +2) or (/' +3) None Overall CD CM CD T I -II LO CO c C O cn CO LO N (;') with: NH(/+2),(/+3), C=0(/+1)or (;'+3) None Overall LO CO in II c CO h-CN| CO co LO LO 0 (/') with: NH(/'+2),(/'+3), /+2) or(/+4) None Overall C=0( CNI LO CN II C D O CO c CD CO C D 5 0 (/') with: +2) or (; +3) None Overall to CD X J CL CD Q . O C L to I to CD X J o. CD Q . to CD X ) t l CD Q . O CL i X < O C o • --—I to CL ° to O v - ° a ° !•? C J X. c o o cS X) W o . CO X) -a e cd ID c o X) o X> " 3 . u O ° '—1 t-< O eg > <D ^ C-> 7T x o u • • O <D CO to O 3 > i to O C3 X3 C C c • - ca o <D _ • ca o L> a o s O " o to O i_ CD cn hi 8 x. « X) 3 -t-< x> c 3 " ^ s O «J T 3 ^ N « s5 « -3 S CD 03 T3 0 to ca 1 ° ^ _2 </5 . O o ca ca p X3 5 _ 3 "o CJ Q. C o o to CD g = O" C J <D >-O h= cj 73 X3 fi to O <D a © , i— o B J g • C J SP 3 c o co T3 C o x> i °r 1 to 3 cj i s C PH to v c o • g ca o cd o ^ 1 y C i o ca cj x : c73 P H - C J <D o _e to 0) o •2 < g" ? " £ E x i i O J , cj O X 00 SpUOq -H U|BL|0U!B1AJ - U|BL|09P!S |B0O"| 136 CD S -CD CL. 8 cu 9-1 " to ;± o = o 1 o=o [ 1 o - x cc 1 - o - x 1 x - x -1 x - x o=o 0 ' 1 o=o I _ . o - x cc 1 - o - x G--HN-1 ' X - -X O = 0 ( D : 1 CN . O - X , 1 i cn t i - o 1 - o - x (39) 1 i X ; . X 1 - z i 0 = 0 ! ; o=o < CD no role as helix cap a-helix cap X "CD Q. %S ro" " c << • m : x to" << • in o b Data set < CD o c r - o -I I cr z - x I o=o I c t - o - x o=o I - o - x o=o I Q C - O - X cr in ^ Q b < CD X - o - o - x I z - x ~ b < CD o=o I c e - o - x I o=o I c c - o - x I X - - . I c c - o - x ^ c w - o - o - x z - x I K x~r°" o - o - o - x z - x I I - o -I cr %- o - o -I o"5 CD Q -•f 8 8 2 8 •g x o <E I o=o I - o - x o=o l a : - o - x / a i o - o - o x' I I o=o I _ o ' ) - X no role as helix cap • • a-helix cap i CN X "as o . =^ CD CO" Asx = Asn • CN Asx = Asp ' -i CO Data set < CD c U I c e - o - x I z - x • I _ . - " X - - V l-o' I c c - o - x I . x - z I o=o b - o - o -I o= I C E - O -I z -I o=o I c c - o - x I . x - z I o=o I b -< CD o - o -I in r~, (to o O X 05 9" < CD 0> 9" i - o I Q : - O - X I z - x -I o = o I o c - o - x I C K - O -I \ X I o - o - o - x o = o I c c - o - x I z - x • o = o I Q C - O - X I . - ' X - - -t-a I c c - o - x I o=o I , <->= b -< CD CD s bO O O X ^ o x ? _ g s § X «D 2 O O H C x 'S3 C O X +- 1 2 X 'co .SP <u CD X CD o cd CD CD X o g cd 2 "8 _ m g n « a. «J eg - 1-3 - I • | £ K > Cd tti •rt ° ^3 * ^ i _ " « CD 2 S - ° CO <+H CD JJ CD C • -fr-1 x .5 3 c CD X CD to O e <+-i CO 3 -a ** c ° g U X> s " 'cyj « C .> W SS CD .3 § 1 8 C CD an •< 2 1 * . a « s • ~ . CD • O * C oo cd cd C/5 CD CD X C 1 5 .3 c S E 3 3 e CU >n td .SP -S E "a c o X) + § oo C L CD « -5 ^ oo O CD i % O cD oo >, CL X M td z - s CD r-B £ ° 3 O O cd "2 ^ C CD cd x> T - N S < ^ 3 X C « CD < XJ <+- oo O £ O .9 II T 3 U .5 o « 15 e <D C c/5 3 cd cd -a oo ^—/ S3 m . -a E § 2 < S3 8 °-§ c/T C L cd ^ « "o OX) 2 1 O cd 137 and i+4 (Figure 3-5B). The most common H-bonds, individually and in combination, are 06(z)<-NH(/ + 2) and C=0(/)^NH(/ + 3). Significantly, the eight hydrogen-bonding patterns adopted by Asx-Pro peptides (Figure 3-5A) form virtually indistinguishable structures (Figure 3-6), suggesting that they form a canonical structure stabilized by alternative hydrogen bonding patterns. For want of a better name, this canonical structure has been called an "Asx-Pro turn". Fig. 3-6. Superimposed wireframe models of Asx-Pro turns. Models were prepared from P D B coordinates of the 36 A s x - P r o - X x x , - X x x 2 turns identified in dataset A and superimposed by hand using general-purpose graphics software. In view A, proline N - C a and C a - C bonds lie in the plane of the paper. Rotating the models 90° around the X-axis (away from the viewer) yields view B and a subsequent 75° Y-axis rotation (to the viewer's right) yields view C. In views A and B, sidechains of X x x j and X x x 2 are omitted. In view C , the sidechains of Asx and X x x , are omitted and the sidechain of X x x 2 is truncated at the B-carbon. Red, oxygen atoms; blue, nitrogen. 3.4.5. Asx-Pro turns have diverse roles Although most Asx-Pro turns function as 3 1 0 - or a-helix caps (Figure 3-5, tabular data), 138 more than one-fifth of these structures function as fl-turns. No clear association was found between the eight H-bond patterns (Figure 3-5A) and their roles as helix caps or as B-turns. Similarly, no clear association could be found between Asx-Pro turns and flanking structural environments. Asx-Pro turns are, for example, found between B-strand and helix structures (Figure 3-1B, 1FX1: position 129), in areas of geometric bends and undefined structure (1GP1(A):142), in a B-strand/turn/helix/B-strand configuration (2APR:59) and as a turn connecting two helices (8CAT(A):256). A . Asx/Xxx, position / B. Pro, position i+1 C . Xxx, position i+2 180 90 \ J / 0 -90 -180 ' ||... -180 -90 0 90 1 80 -180 -90 0 4> 90 180 -180 . a L . . . a 90 180 Fig. 3-7. Ramachandran plot of (|),I|J dihedral angles in Asx-Pro turns and other proline-containing sequences, "x", individual <j),v|f pairs for the 36 Asx-Pro turns identified in dataset A . Contour plots reflect all Xxx-P ro -Xxx sequences of dataset A other than Asx-Pro-Xxx; in each plot, the outermost contour line of each separate area reflects a common low density. Contour plots illustrate the general restriction to B-conformational space of residues preceding prolines (panel A), the narrow range of (|) angles adopted by prolines (panel B) and the restriction of residues generally (whether fol lowing proline or not) to a-, B- and (in some cases) to aL-conformational space (panel C); they reasonably reflect detailed experimental and theoretical plots of permitted (j),i|j angles (MacArthur & Thornton 1991; Richardson 1981; Schulz & Schirmer 1979). 3.4.6. Backbone and sidechain geometry of Asx-Pro turns are not unusual While the geometry of Asx-Pro turns makes intuitive sense, backbone 4>,T|J angles of residues in Asx-Pro turns were nevertheless compared with those of other proline-containing sequences, and X\ angles of Asx in Asx-Pro turns were compared with published values for aspartate and asparagine (McGregor et al. 1987), to confirm that Asx-Pro turn geometry is not unusual. Indeed, backbone 4>,t|/ pairs for Asx residues in Asx-Pro turns fall within common 139 (Schulz & Schirmer 1979; Richardson 1981; MacArthur & Thornton 1991) areas o f conformat ional space (Figure 3-7). A s w e l l , most A s x - P r o turns adopt A s x angles (Figure 3-8) corresponding to one (the t conformation) o f two common conformations observed for aspartate and asparagine sidechains (McGregor et al. 1987), and a l l possess A s x X\ angles consistent w i t h s idecha in conformations observed (Dunbrack & Karplus 1994) for the d),i|/ B-conformat ion adopted by A s x i n these turns. Fig. 3-8. Sidechain Xi angles of Asx in Asx-Pro turns. angles in the 36 Asx-Pro turns identified in dataset A , converted to the range 0°-360° for comparison with published global distributions of Xi angles (McGregor et al. 1987). Most angles (n = 33, mean = 185° , s = 9.3°) correspond to the t conformation; remaining angles correspond to the g- conformation. Fig. 3-9. Superimposed cc-carbon traces of Asx-Pro helix caps and flanking residues. These views, rotated 90° relative to one another, are of six 3 ] 0 - and six a-helix caps among 36 Asx-Pro turns (dataset A ) that adopt three different hydrogen bonding patterns. X l 5 residue immediately preceding helix cap. B 2 , Asx , helix capping position. P 3 , Pro, first helix position. X 4 to X 6 , residues in first helix turn. A further feature o f A s x - P r o turn geometry is noteworthy: a l though these turns commonly function as hel ix caps, stabilizing helices in the C- te rmina l d i rect ion, they are he l ix -terminat ing i n the N - t e r m i n a l di rect ion (Figure 3-9). 3.4.7. A consensus-derived model for Asn-Pro-Asn agrees with experimental data Chapte r 4 describes studies o f the expression, on the surface o f f i lamentous 140 bacteriophage, o f "shaped" peptides containing the sequence A s n - P r o - A s n - A l a ( N P N A ) , tandem repeats o f w h i c h comprise most o f an immunodominant region o f the P. falciparum circumsporozoite protein. The structure o f these repeats has been the focus o f several studies, including two that predicted minimum-energy conformations o f these repeats (Gibson & Scheraga 1986; Brooks et al. 1987). However , neither o f these predicted conformations is consistent w i t h N M R studies (Dyson et al. 1990) that identif ied a l imi ted set o f H-bonds possible i n N P N A -containing peptides. Specifically, these N M R studies a l l owed that H-bonds cou ld f o r m among the sidechain O o f A s n f / j , backbone C = 0 o f residues i-l and /, and backbone N H groups o f residues i + 2 and i + 3. Later studies (Satterthwait et al. 1990) addi t ional ly a l l o w e d that H-bond(s) c o u l d f o r m between the sidechains o f asparagine residues f l ank ing pro l ine . Mainchain - Mainchain H-bonds 5 O (/)<-NH(/+2) or(/+3) 51 8 59 None 2 3 18 Overall 74 n = 39 B H , N 0 - - ' I C H , I (3) ..(3) \ ( 1 ) - I , I N - N P N , n = 3 \ cf \ I H C H , ', R R i I ' I I - N - C - C - N — C - C - N - C - C - N H - C - C — N — C — C -I I II I II . I II I II I I II H H O H O H O H O H H O - (2) ' ' - : ; ; ; " --'(2) HO cr \T i C H , I Z (11)" .(2) - H 2 V D P N , n= 11 H C H , R R i I I I - N - C - C - N — C - C - N — C - C - N H - C - C — N - C - C -I I II I II I II ; I II I I II H H O H O H O - H O H H O . . . - - ' ( 1 1 ) H,N o r C H , (6) ' • - H O NPD, n = 6 H C H , ', R R I ' \ / i I 1 I I V N - C - C - N - C - C - N - C - C - N H - C - C - N - C - C -I I II I II I II ' I II I I II H H O H O H O . . ' H O H H O .. d) Fig. 3-10. Local H-bonds in Asx-Pro-Asx sequences. The matrix in part A shows, per 100 residues, the number of Asx-Pro-Asx sequences identified in dataset C that form (i) local sidechain-backbone and backbone-backbone H-bonds in combination, (ii) sidechain-backbone H-bonds only, (ii i) backbone-backbone H-bonds only, or (iv) no local H-bonds. Part B provides a summary of H-bonds formed by the 51 per cent of (20 of 39) Asx-Pro-Asx sequences that form Asx-Pro turns. Numbers in parentheses indicate the number of H-bonds formed among the 3 Asn-Pro-Asn, 11 Asp-Pro-Asn and 6 Asn-Pro-Asp sequences. 141 Fig. 3-11. Superimposed wireframe models of Asx-Pro-Asx peptides. Models were prepared from P D B coordinates of 39 Asx-Pro-Asx sequences identified in dataset C and superimposed by hand using general-purpose graphics software. Views A and B represent 20 Asx-Pro-Asx sequences that form Asx-Pro turns, while views C and D show 19 other Asx-Pro-Asx sequences. Atoms comprising the C = 0 group of the 2nd A s x residue are omitted. In views A and C, proline N - C a and C a - C bonds lie in the plane of the paper. A 9 0 ° rotation of these models around the X-axis (away from the viewer) yields views B and D, respectively. Red, oxygen atoms; blue, nitrogen. Anticipating that a consensus structure derived from the PDB would accord with these N M R findings and more precisely describe possible hydrogen bonding patterns within NPNA repeats, the PDB was surveyed for structures formed by Asx-Pro-Asx. To allow for Asx-Asx sidechain-sidechain H-bonds, the study was restricted to sequences where both Asx residues were not aspartate. Not unexpectedly, the results (Figure 3-10) are similar to the findings for Asx-Pro: 51 per cent (20 of 39) of Asx-Pro-Asx sequences form H-bonds among sidechain 0 6 (/) , backbone C=0(i) and backbone NH( /+n 2 < n < 4 ) groups. The most common or 142 "consensus" H-bonds are 06(i)*-N¥L(i+2) and C=0(i)<-NH(i+3). Importantly, geometries adopted by Asx-Pro-Asx turns are strikingly similar (Figure 3-11). Fig. 3-12. H-bonds in Asn-Pro-Asn sequences that form Asx-Pro turns. The molecular geometries adopted by the three Asn-Pro-Asn sequences identified in dataset C are striking similar. A and B, PDB entry 1 GAL, residues 41-43; C and D, 1NDK, 150-152; E and F, 3ICD, 268-270 (Figure 3-2). In views A, C and E proline N - C a and C a -C bonds lie in the plane of the paper. A 90° rotation of these models around the X-axis yields views B, D and F respectively. H-bonds in views A and B are consistent with those allowed by NMR (Dyson et al. 1990) and related (Satterthwait et al. 1990) studies of peptides containing Asn-Pro-Asn-Ala repeats. Although only four sequences were found in which Asx-Asx sidechain-sidechain H-bonds were formed, it may be significant that all of the three Asn-Pro-Asn turns possessed these H-bonds (Figure 3-12). H-bonds in the first of these (Figure 3-12A and B) include the two "consensus" H-bonds described above. That these H-bonds are consistent with N M R (Dyson et al. 1990) and related (Satterthwait et al. 1990) studies of NPNA-containing peptides suggests that this structure can serve as a consensus-derived partial model for a minimum energy conformation of NPNA. 3.4.8. The proportion of turns versus non-turns increases with structure resolution Structures with relatively low resolution (up to 3 A) were included in this study in order to analyse a reasonably large dataset: at this low resolution, a significant number of Asx-Pro turns were identified. Notably, however, the proportions of Asx-Pro and Asx-Pro-Asx sequences forming turns increase with improved resolution. Thus, only 45 per cent of Asx-Pro 143 sequences form turns in structures resolved to <3.0 A, while 63 per cent form turns in structures resolved to <1.8 A (Figure 3-13). Similarly, 51 per cent of Asx-Pro-Asx sequences form Asx-Pro turns in structures resolved to < 3.0 A, but 73 per cent form turns in structures resolved to <1.8 A (Figure 3-13). X c to j -< 3 o I CL O to « < £ ° CD tz o o c ~ CD O =3 to cr to 0.8 0.7 0.6 0.5 0.4 | Dataset A, Asx-Pro I Dataset C, Asx-Pro-Asx 3.0 2.8 2.6 2.4 2.2 2.0 1.8 Resolut ion (A) Fig. 3-13. Influence of structure resolution on identification of Asx-Pro turns. Figure shows the fraction of Asx-Pro sequences identified as Asx-Pro turns as a function of structure resolution. Bars summarizing structures resolved to <3.0 A resolution reflect all of the 79 Asx-Pro sequences identified in dataset A and all of the 39 Asx-Pro-Asx sequences identified in dataset C. Bars summarizing higher resolution structures reflect progressively fewer sequences; thus the bars for structures solved to < 1.8 A are derived from only 27 Asx-Pro sequences and 11 Asx-Pro-Asx sequences. 3.5. DISCUSSION This study has shown that, in contrast to a relatively more diverse assortment of conformations adopted by other Xxx-Pro sequences (including those where Xxx = Ser or His), 63 per cent or more of Asx-Pro sequences and 73 per cent or more of Asxj-Pro-AsXj (where ASXJ * ASXJ) sequences, in a variety of flanking structural contexts, adopt strikingly similar conformations. Their common feature is a canonical turn geometry stabilized by a set of alternative H-bonds among the sidechain O and backbone C=0 of Asx (residue /) and backbone N H groups of residues i+2, i+3 and i+4. That alternative H-bonds are possible with changes in geometry would seem to reduce the sensitivity of Asx-Pro turns to perturbation, allow greater flexibility in local packing and greater freedom to substitute neighbouring residues, whether by evolutionary mechanisms or deliberate engineering. That a majority of Asx-Pro sequences form these turn structures is not surprising. Anecdotal experience, combined with studies of turn-forming propensities of these residues 144 considered separately, dictate that they "must" form turns. Indeed, the molecular geometry of an Asx-Pro turn makes almost intuitive sense and would seem to reflect simple molecular modelling concepts such as (a) proline-induced and more general steric constraints that limit backbone and sidechain combinations to a small number of preferred choices, and (b) those characteristics of asparagine and aspartate sidechains (highly polar O 6 atoms, sidechain length and geometry) that make these residues uniquely suited to forming turn-stabilizing, alternative sets of H-bonds. In contrast, the polar sidechains of serine, threonine and histidine are presumably too short, lack a required conformational freedom or (compared to aspartate) are too versatile, being ab