Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

A quantitative proteomics analysis of human cells undergoing apoptosis Anthony, Joseph Stephan 2010

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

Item Metadata

Download

Media
24-ubc_2010_fall_anthony_joseph.pdf [ 33.41MB ]
Metadata
JSON: 24-1.0069929.json
JSON-LD: 24-1.0069929-ld.json
RDF/XML (Pretty): 24-1.0069929-rdf.xml
RDF/JSON: 24-1.0069929-rdf.json
Turtle: 24-1.0069929-turtle.txt
N-Triples: 24-1.0069929-rdf-ntriples.txt
Original Record: 24-1.0069929-source.json
Full Text
24-1.0069929-fulltext.txt
Citation
24-1.0069929.ris

Full Text

   A QUANTITATIVE PROTEOMICS ANALYSIS OF HUMAN CELLS UNDERGOING APOPTOSIS.  by  JOSEPH STEPHAN ANTHONY    A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF  DOCTOR OF PHILOSOPHY   in   THE FACULTY OF GRADUATE STUDIES  (Experimental Medicine)    THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver)   April, 2010    © Joseph Stephan Anthony, 2009      ii  ABSTRACT  Elucidating the events and mechanism of regulation of apoptosis is of wide interest to the scientific community, and to humanity, since apoptosis, so important for proper development and maintenance of an organism, is also responsible for disease when the process goes awry. In this thesis, a proteomics investigation into changes in protein concentrations and half-lives in early apoptosis is presented, enhancing our understanding of this process.  Using Stable Isotope Labeling with Amino Acids in Cell Culture (SILAC), cytokine withdrawal-induced apoptosis of a human hematopoietic cell line, TF-1, was studied. This is a useful model in which signaling pathways regulating apoptosis have been extensively studied previously. A study such as this can be considered “hypothesis-generating”, but at the outset the hypothesis is that proteins whose functions are closely tied to regulation of apoptosis will show detectable changes in quantity in cells undergoing apoptosis.  Initially three biological replicates were performed, comprising 200 samples in all, analyzed using an FT-ICR mass spectrometer. Relative abundance of 1451 proteins identified in common between three biological replicates was determined, and 124 proteins showing the largest concentration changes in response to cytokine withdrawal are discussed in more detail. A subsequent effort investigated protein half-life changes in response to cytokine withdrawal and identified 255 proteins for which half-lives were calculated. The apparent changes in protein half-life in response to cytokine withdrawal are discussed.     iii A high level of coverage of the proteome was achieved, giving a large number of protein identifications and relative quantitations. Further I have been able to identify several apparently synchronous changes in concentration between proteins with related functions, suggesting possible interactions not previously described, or identified as playing a role in cell survival, proliferation or death. Further, I observed cytokine withdrawal-induced alterations in concentration in some proteins for which little is known. The proteomic analysis of apoptosis using SILAC to determine protein half-life data is also a novel approach. Together, the work in this thesis suggests numerous avenues of investigation potentially leading to novel findings regarding cells undergoing apoptosis; and also suggests a potentially fruitful avenue of investigation for clinical management of patients undergoing chemotherapy.    iv TABLE OF CONTENTS ABSTRACT ............................................................................................................................ ii! TABLE OF CONTENTS ...................................................................................................... iv! LIST OF TABLES.................................................................................................................. x! LIST OF FIGURES.............................................................................................................. xii! LIST OF ABBREVIATIONS .............................................................................................. xv! ACKNOWLEDGEMENTS ............................................................................................... xvii! DEDICATION ..................................................................................................................... xix! 1! Introduction ...................................................................................................................... 1! "#"! $%&%'&()( #########################################################################################################################################*! "#"#"! $%&''()(*&+(,-!,)!$.%%!/.&+0 ##################################################################################################################1! "#"#2! 34(.)!5('+,4(*&%!67.47(.8#####################################################################################################################9! "#"#:! /.'*4(;+(,-!,)!<;,;+,'(' ######################################################################################################################=! "#"#>! $&';&'.'####################################################################################################################################################""! "#"#1! ?-(+(&+(,-!@0&'.!&-A!BC.*D+(,-!@0&'.##########################################################################################"2! "#"#E! ?-+.4-&%!@&+08&F!!"#!BC+.4-&%!@&+08&F######################################################################################":! "#"#9! G.HD%&+(,-!,)!$&';&'.'#######################################################################################################################"I! "#"#I! $&';&'.J(-A.;.-A.-+!@&+08&F!K<?L!M.A(&+(,-N ####################################################################"=! "#"#=! L4,M!@&+0,%,HF!+,!O,%.*D%&4!3(,%,HF #########################################################################################2P! "#"#=#"! 3*%J2!L&M(%F##################################################################################################################################################### 2P! "#"#=#2! @4,J!&-A!<-+(J<;,;+,+(*!O.MQ.4' ######################################################################################################## 2P! "#"#=#:! LD-*+(,-!,)!3*%J2!L&M(%F!O.MQ.4'######################################################################################################## 22! "#+! ,-./0!,1'2314############################################################################################################################# +.! "#2#"! <*+(7&+(,-!,)!@?:JR ###############################################################################################################################2>! "#2#2! /.&*+(7&+(,-!,)!@?:JR ##########################################################################################################################21! "#2#:! /,8-'+4.&M!,)!@?:JR##########################################################################################################################2E! "#2#>! /F'4.HD%&+(,-!,)!@?:JRS@R3 ############################################################################################################2=! "#2#1! TOJ$UL!U(H-&%%(-H#################################################################################################################################:P! "#2#1#"! TOJ$UL!U(H-&%(-H!7(&!B4V########################################################################################################################### :"! "#2#1#2! TOJ$UL!U(H-&%(-H!7(&!W<RSUX<X############################################################################################################# :2! "#.! 51674!$%&%'&()(!&6!8,69/$%&%'&()(: ##################################################################################### ..!    v "#:#"! @4,+.(-!$,-*.-+4&+(,-!$0&-H.'!#$%"&"!@XO' ###########################################################################:1! "#;! ,6&'9&<)=(#################################################################################################################################### .>! "#>#"! O&''!U;.*+4,M.+4F!Y!34(.)!5('+,4(*&%!67.47(.8 #####################################################################:9! "#>#2! U?Z<$ ##########################################################################################################################################################:I! "#>#2#"! /.+.4M(-(-H!+0.!<**D4&*F!,)!@4,+.(-![D&-+(+&+(,-######################################################################## >2! "#>#:! O&''!U;.*+4,M.+4F!Y!L,D4(.4!X4&-'),4M!Y!?,-!$F*%,+4,-!G.',-&-*.###########################>:! "#>#:#"! L,D4(.4!X4&-'),4M!?,-!$F*%,+4,-!G.',-&-*.!O&''!U;.*+4,M.+4F!Y!<!34(.)!67.47(.8## >1! "#>#:#2! R.F!O&''!U;.*+4,M.+.4!@&4&M.+.4'!),4!@4,+.,M(*'##################################################################### >I! "#>#:#:! \3,++,MJ];^!_.4'D'!\X,;J/,8-^!<;;4,&*0##################################################################################### >=! "#>#:#>! /&+&Q&'.!U.&4*0(-H!&-A!A.!-,7,!U.`D.-*(-H!KOUSOUN############################################################### 1P! "#*! ?29!$%%6&1=2 ############################################################################################################################## *.! "#>! @A<<164 ####################################################################################################################################### *;! 2! Materials and Methods .................................................................................................. 56! +#"! B977!BA7'A69 ################################################################################################################################### *>! +#+! B977!BA7'A69!C!@'1D79!-(&'&%9!E1D977)FG################################################################################ *>! +#.! BHI-!C!,69%161')&F!1FJ!K(1G9 ############################################################################################## >L! +#;! B977!E4()(!MANN96!O@&7AD)7)P1')&F!DANN96Q!R!@1<%79!E&1J)FG!MANN96 ########################### >L! +#*! @S@/,$H5!C!H97!,69%161')&F!1FJ!@'1)F)FG ######################################################################### >"! +#>! -F/H97!?64%()F!S)G9(')&F!1FJ!,9%')J9!5T'61=')&F############################################################ >.! +#U! @S@/,$H5!C!-<<AF&D7&'')FG################################################################################################## >>! +#V! -<<AF&D7&'')FG!OW9('96F!D7&'')FGQ#################################################################################### >U! +#X! $F')D&J)9(##################################################################################################################################### >X! +#"L! Y7&3!B4'&<9'64 ######################################################################################################################## UL! +#""! M61JN&6J!,6&'9)F!B&F=9F'61')&F!$((14 ############################################################################ U"! +#"+! $=/S5ZS/%[$!B1(%1(9!$((14 ############################################################################################### U+! +#".! I9'21F&7/B27&6&N&6<!,6&'9)F!,69=)%)'1')&F ################################################################# U.! +#";! S1'1!\1FJ7)FG ########################################################################################################################### U.! 2#">#"! a,4V(-H!8(+0!+0.!L(-&%!Z('+!,)!@4,+.(-' ###################################################################################91! 2#">#2! 3%&'+2T6#################################################################################################################################################99! 2#">#:! ?-H.-D(+F!b!@&+08&F!<-&%F'(' #####################################################################################################99! 3! Proteomic Analysis of A Hemopoietic Cell Line Undergoing Apoptosis................... 79! .#"! -F'6&JA=')&F################################################################################################################################# UX! .#+! 5T%96)<9F'17!]^96^)93!1FJ!M1=_G6&AFJ!I1'96)17 ######################################################## V;!    vi :#2#"! L%,8!$F+,M.+4F!<-&%F'('!,)!<;,;+,'('!(-!XLJ"!$.%%'#############################################################I1! :#2#2! $&';&'.!<*+(7(+F!<-&%F'('!(-!XLJ"!$.%%' #######################################################################################I9! :#2#:! /.+.4M(-&+(,-!,)!$,-*.-+4&+(,-'!,)!<4H(-(-.!&-A!ZF'(-.!G.`D(4.A!),4!$.%%!T4,8+0! (-!G@O?J"E>P #########################################################################################################################################################I=! :#2#>! ?-*,4;,4&+(,-!,)!":$JZ&Q.%%.A!<M(-,!<*(A' ##############################################################################I=! :#2#1! 6;+(M(c&+(,-!,)!U&M;%.!L4&*+(,-&+(,-!$,-A(+(,-' #################################################################=P! :#2#1#"! U+4,-H!<-(,-!BC*0&-H.!$04,M&+,H4&;0F!U.;&4&+(,-################################################################### =P! :#2#1#2! U/UJ@<TB!U.;&4&+(,- ################################################################################################################################## =2! :#2#E! L,D4(.4JX4&-'),4M!?,-!$F*%,+4,-!G.',-&-*.!O&''!U;.*+4,M.+4F!,)!XLJ"!@4,+.(-' ### ! #######################################################################################################################################################################=>! .#.! ,6&'9)F(!@2&3)FG!B&F()('9F'!B21FG9(!)F!B&F=9F'61')&F!)F!`9(%&F(9!'&!B4'&_)F9! W)'2J61317 ###########################################################################################################################################"L+! :#:#"! G.'D%+' #################################################################################################################################################### "P>! :#:#2! X0.!$D44.-+!a,4V!(-!+0.!$,-+.C+!,)!@4.7(,D'!U+DA(.'####################################################### """! .#;! B&F=7AJ)FG!`9<16_( ###############################################################################################################"+.! 4! Bioinformatics ............................................................................................................... 125! ;#"! -F'6&JA=')&F###############################################################################################################################"+*! ;#+! @'1')(')=17!S)('6)DA')&F!&N!'29!S1'1#####################################################################################"+U! ;#.! S9'96<)F1')&F!&N!Y17(9!,&()')^9!,6&'9)F!-J9F')N)=1')&F( ############################################"+X! >#:#"! $&%*D%&+(,-!,)!+0.!L&%'.!/('*,7.4F!G&+.!KL/GN ##################################################################### ":>! ;#;! ,6&'9)F!aA1F')'1')&F!C!`1')&!?9(')FG#################################################################################".U! ;#*! S9'96<)F)FG!E9^97(!&N!@)GF)N)=1F=9####################################################################################";L! ;#>! ,6&'9)F!B1'9G&6)P1')&F!M4!H9F9!]F'&7&G4!OH]Q# ############################################################";"! >#E#"! 3(,%,H(*&%!@4,*.''############################################################################################################################## ">I! >#E#"#"! X,+&%!@4,+.(-!$,M;&4.A!8(+0!UDQ'.+!+0&+!U0,8'!\?-*4.&'.A!$,-*.-+4&+(,-^!Y!3(,%,H(*&%! @4,*.''! #############################################################################################################################################################################"1>! >#E#"#2! @4,+.(-'!6Q'.47.A!+,!?-*4.&'.!(-!$,-*.-+4&+(,-d!&-A!@4.7(,D'%F!/.'*4(Q.A!<'!5&7(-H!<! G,%.!(-!$.%%!$F*%.d!$.%%!@4,%().4&+(,-!,4!$.%%!/.&+0############################################################################################"E"! >#E#"#:! @4,+.(-'!6Q'.47.A!+,!?-*4.&'.!(-!$,-*.-+4&+(,-d!&-A!e,+!@4.7(,D'%F!/.'*4(Q.A!<'!3.(-H! ?-7,%7.A!(-!$.%%!UD47(7&%d!$.%%!$F*%.!,4!$.%%!/.&+0# ###########################################################################################"EE! >#E#"#>! @,+.-+(&%!G.%.7&-*.!,)!X0.'.!6Q'.47&+(,-'#####################################################################################"92! >#E#"#1! X,+&%!@4,+.(-!$,M;&4.A!8(+0!+0.!UDQ'.+!+0&+!U0,8'!\/.*4.&'.A!$,-*.-+4&+(,-^!Y! 3(,%,H(*&%!@4,*.'' ###########################################################################################################################################################"9>!    vii >#E#"#E! @4,+.(-'!6Q'.47.A!+,!/.*4.&'.!(-!$,-*.-+4&+(,-d!&-A!@4.7(,D'%F!/.'*4(Q.A!<'!5&7(-H!<! G,%.!(-!$.%%!$F*%.d!$.%%!@4,%().4&+(,-!,4!$.%%!/.&+0############################################################################################"I1! >#E#"#9! @4,+.(-'!6Q'.47.A!+,!/.*4.&'.!(-!$,-*.-+4&+(,-d!&-A!e,+!@4.7(,D'%F!/.'*4(Q.A!<'! 5&7(-H!<!G,%.!(-!$.%%!$F*%.d!$.%%!@4,%().4&+(,-!,4!$.%%!/.&+0########################################################################"="! >#E#"#I! @,+.-+(&%!G.%.7&-*.!,)!X0.'.!6Q'.47&+(,-'#####################################################################################"=I! >#E#2! O,%.*D%&4!LD-*+(,- ########################################################################################################################### 2PP! >#E#2#"! X,+&%!@4,+.(-!$,M;&4.A!8(+0!UDQ'.+!+0&+!U0,8'!\?-*4.&'.A!$,-*.-+4&+(,-^!Y!O,%.*D%&4! LD-*+(,-! #############################################################################################################################################################################2P>! >#E#2#2! X,+&%!@4,+.(-!$,M;&4.A!8(+0!UDQ'.+!+0&+!U0,8'!\/.*4.&'.A!$,-*.-+4&+(,-^!Y! O,%.*D%&4!LD-*+(,- ########################################################################################################################################################2P9! 4.7! Pathway Analysis Using Ingenuity ® Pathway Analysis (IPA) ########################################+"L! ;#V! B&F=7AJ)FG!`9<16_( ###############################################################################################################++"! >#I#"! UDHH.'+(,-'!),4!LD+D4.!/(4.*+(,-'############################################################################################## 22>! 5! Validation of the Mass Spectrometry Results ............................................................ 226! *#"! -F'6&JA=')&F###############################################################################################################################++>! *#+! Z17)J1')&F!&N!,6&'9)F!B21FG9(!SA6)FG!$%&%'&()( ##########################################################++X! 5.3! Immunoblotting N&6!'29 Validation &N Proteins###################################################################+.L! 1#:#"! \];JG.HD%&+.A^!@4,+.(-'!Y!@4,+.(-'!U0,8(-H!<-!?-*4.&'.!(-!$,-*.-+4&+(,- ########## 2:"! 1#:#2! \/,8-JG.HD%&+.A^!@4,+.(-'!Y!@4,+.(-'!+0&+!U0,8!<!/.*4.&'.!(-!$,-*.-+4&+(,-# # 2:I! 1#:#:! ]-*0&-H.A!@4,+.(-'########################################################################################################################## 2>"! *#;! $F1<&6()F ##################################################################################################################################+;*! 1#>#"! <-&M,4'(-!(-!$F+,V(-.Ja(+0A4&8&%!Y!O&''!U;.*+4,M.+4F!G.'D%+'# ########################### 2>E! 1#>#2! <-&M,4'(-!/.*4.&'.'!(-!G.';,-'.!+,!$F+,V(-.!a(+0A4&8&%# ######################################## 21>! 1#>#:! B7(A.-*.!),4!@0,';0,4F%&+(,-###################################################################################################### 21>! 1#>#>! <-&M,4'(-!(-!G.';,-'.!+,!$F+,V(-.!a(+0A4&8&%!8(+0!&-A!a(+0,D+!Zf2=>PP2!&-A! ]6"2E!X4.&+M.-+# ############################################################################################################################################ 21I! 1#>#1! <-&M,4'(-!5&%)J%().########################################################################################################################### 21I! 1#>#E! <-&M,4'(-!Y!$,-*%D'(,-'!&-A!LD+D4.!/(4.*+(,-' ################################################################ 21=! *#*! ,6&G61<<9J!B977!S91'2!;!O,SBS;Q ###################################################################################+>+! 1#1#"! <''(H-M.-+!,)!@/$/>!@.;+(A.'!X,!<!T.-.!@4,AD*+!J!U&M;%.!<-&%F'('!]'(-H!+0.! T%,Q&%!@4,+.(-!O&*0(-.!KT@ON#################################################################################################################### 29E! 1#1#2! @,''(Q%.!@/$/>!@4,+.(-!/.H4&A&+(,- ####################################################################################### 2I9! 1#1#:! O(H0+!+0.!<;;&4.-+!@4,+.(-!G&+(,!3.!/('+,4+.A!3F!?-*,-'('+.-+![D&-+(+&+(,-! G&+(,'g!################################################################################################################################################################### 2I=!    viii 1#1#>! O(H0+!@/$/>!Z.7.%'!3.!Z,8!(-!XLJ"d!Z.&A(-H!X,!?-&**D4&+.![D&-+(+&+(,-g ########### 2I=! 1#1#1! $,-'('+.-*F!,)!U&M;%.!@4.;&4&+(,-!3.+8..-!BC;.4(M.-+' ############################################ 2=P! 1#1#E! @,''(Q%.!G,%.!),4!UDQ*.%%D%&4!/('+4(QD+(,-!,)!@/$/>######################################################### 2=P! 1#1#9! /()).4(-H!X.M;,4&%!/F-&M(*' ###################################################################################################### 2=P! 1#1#I! ?'!+0.!<-+(Q,AF!/.+.*+(-H!+0.!$,44.*+!@4,+.(-g#################################################################### 2="! *#>! B&F=7A()&F( ################################################################################################################################+X;! 6! Determination of Rates of Protein Synthesis – Exploratory Data ........................... 295! >#"! -F'6&JA=')&F###############################################################################################################################+X*! >#+! 5T%96)<9F'17!]^96^)93 #########################################################################################################+XV! >#.! `9(A7'(##########################################################################################################################################.L.! E#:#"! @4,+.(-'!?A.-+()(.A!(-!\$,-+4,%^!U&M;%.################################################################################### :P:! E#:#2! @4,+.(-'!?A.-+()(.A!(-!\U+&47(-H^!U&M;%. ################################################################################# :"E! E#:#:! @4,+.(-!5&%)JZ().!/&+&!#"'!@4,+.(-!$,-*.-+4&+(,-!/&+& ###################################################### :2I! >#;! B21FG9(!)F!'29!,1''96F!&N!S)('6)DA')&F!&N!,6&'9)F(!'&!H9F9!]F'&7&G)9( ################..U! >#*! S)(=A(()&F###################################################################################################################################.;.! E#1#"! @4,+.(-'!8(+0!<!$0&-H.!(-!3,+0!$,-*.-+4&+(,-!&-A!(-!5&%)JZ(). ################################### :>E! E#1#2! @4,+.(-'!8(+0!e,!$0&-H.!(-!$,-*.-+4&+(,-!&-A!<!$0&-H.!(-!5&%)JZ().####################### :1"! E#1#:! @4,+.(-'!8(+0!<!$0&-H.!(-!$,-*.-+4&+(,-!&-A!e,!$0&-H.!(-!5&%)JZ().####################### :11! E#1#>! @4,+.(-'!L,D-A!(-!+0.!5&%)JZ().!BC;.4(M.-+!&-A!e,+!L,D-A!(-!X04..!$,-*.-+4&+(,-! BC;.4(M.-+' ######################################################################################################################################################## :1E! E#1#1! T.-.4&%!$,MM.-+' ############################################################################################################################ :1=! E#1#E! /('+4(QD+(,-!<M,-H!Z.7.%!>!T.-.!6-+,%,H(.'!(-!$,-+4,%!&-A!U+&47(-H!$.%%' ########### :1=! >#>! B&F=7A()&F( ################################################################################################################################.>*! 7! Conclusions and Future Directions............................................................................. 368! U#"! ?29!`971')&F(2)%!M9'399F!'29!,69^)&A(!B21%'96(!C!-F'9G61')&F!&N!'29!I1'96)17#! #### ! ########################################################################################################################################################.UL! U#+! ?29!BA669F'!W&6_!)F!'29!]^96177!B&F'9T'!&N!'29!Y)97J# ################################################.U"! U#.! @'69FG'2(!1FJ!E)<)'1')&F(!&N!'29!BA669F'!@'AJ4 ############################################################.U;! U#;! @%9=)N)=!\4%&'29(9(!1FJ!YA'A69!S)69=')&F( ####################################################################.U>! REFERENCES ................................................................................................................... 385! Appendix 1. Proteins Observed to Increase in Concentration. ...................................... 412!    ix Appendix 2. Proteins Observed to Decrease in Concentration. ..................................... 432! Appendix 3 - Derivation of the Protein Half-Life Equations.......................................... 462!     x LIST OF TABLES TABLE 2.1 DETAILS OF THE CHROMATOGRAPHY RUNS ON THE 1100 HPLC ............................ 66 TABLE 2.2 PRIMARY ANTIBODIES USED IN THE EXPERIMENTS REPORTED IN THIS THESIS. ........ 69 TABLE 3.1 THE NUMBER OF PROTEINS IDENTIFIED IN EACH EXPERIMENT, AND THE NUMBER OF PEPTIDES USED TO IDENTIFY THESE PROTEINS................................................................ 100 TABLE 3.2 RELATIVE 13 C/ 12 C RATIOS AT VARIOUS LIKELIHOODS OF CHANCE........................ 104 TABLE 3.3 THE LIST OF PROTEINS OBSERVED TO INCREASE IN RESPONSE TO CYTOKINE WITHDRAWAL. ............................................................................................................... 106 TABLE 3.4 THE LIST OF PROTEINS OBSERVED TO DECREASE IN RESPONSE TO CYTOKINE WITHDRAWAL. ............................................................................................................... 108 TABLE 3.5. DIRECTION OF CHANGE OF PROTEINS SHOWING GREATER THAN 30% CHANGE IN ANTHONY, COMPARED WITH OTHER PUBLISHED RESULTS ............................................. 113 TABLE 3.6 RECENT PROTEOMICS PUBLICATIONS REFERRED TO IN THE TEXT, WITH DETAILS OF EACH STUDY. ................................................................................................................. 114 TABLE 3.7 COMPARISON OF PROTEINS AND QUANTITATION RATIOS FOUND IN THE CURRENT STUDY WITH PREVIOUSLY PUBLISHED WORK. ................................................................ 115 TABLE 4.1 SUMMARY OF DETAILS FOR CALCULATION OF FALSE DISCOVERY RATE (FDR). .. 136 TABLE 4.2 DIFFERENCES BETWEEN MEASURED RATIOS AND EXPECTED RATIOS FOR HSP-! AND " AT DIFFERENT MIXTURE RATIOS OF 12 C WITH 13 C........................................................ 138 TABLE 4.3 NUMBER OF STANDARD DEVIATIONS FOR DIFFERENT LEVELS OF SIGNIFICANCE – NORMALLY DISTRIBUTED DATA. .................................................................................... 140 TABLE 4.4 CUT-OFFS FOR I/W RATIOS AT DIFFERENT LEVELS OF SIGNIFICANCE. .................... 141 TABLE 4.5  RAW DATA FROM WHICH FIGURE 4.4 HAS BEEN DRAWN. ..................................... 152 TABLE 4.6 RAW DATA FROM WHICH FIGURE 4.5 HAS BEEN DRAWN....................................... 153 TABLE 4.7 RAW DATA FROM WHICH FIGURE 4.6 HAS BEEN DRAWN....................................... 177 TABLE 4.8. RAW DATA FROM WHICH FIGURE 4.7 HAS BEEN DRAWN...................................... 178 TABLE 4.9 RAW DATA FROM WHICH FIGURE 4.8 HAS BEEN DRAWN....................................... 203 TABLE 4.10 THE TOP TEN FOLD CHANGES OBSERVED AMONGST PROTEINS SHOWN TO INCREASE IN CONCENTRATION IN RESPONSE TO CYTOKINE WITHDRAWAL...................................... 211 TABLE 4.11 THE TOP TEN FOLD CHANGES OBSERVED AMONGST PROTEINS SHOWN TO DECREASE IN CONCENTRATION IN RESPONSE TO CYTOKINE WITHDRAWAL...................................... 212    xi TABLE 5.1. DETAILS OF THE PEPTIDES USED TO IDENTIFY ANAMORSIN IN THREE SILAC EXPERIMENTS ................................................................................................................ 253 TABLE 5.2 TWENTY PEPTIDES USED TO IDENTIFY PDCD4 IN THREE MASS SPECTROMETRY EXPERIMENTS. ............................................................................................................... 265 TABLE 5.3. RESULTS OF ANALYSIS OF RAW DATA USING THE GPM....................................... 278 TABLE 6.1 LIST OF PROTEINS FROM CONTROL SAMPLE AT 15 HOURS, IDENTIFIED AND QUANTITATED AS HAVING A RATIO OF LESS THAN 0.67.................................................. 304 TABLE 6.2 LIST OF PROTEINS FROM CONTROL SAMPLE AT 15 HOURS, IDENTIFIED AND QUANTITATED A RATIO OF GREATER THAN 1.85............................................................. 310 TABLE 6.3 LIST OF PROTEINS FROM “STARVE” SAMPLE AT 15 HOURS, IDENTIFIED AND QUANTITATED AS HAVING A RATIO OF LESS THAN 0.67.................................................. 317 TABLE 6.4 LIST OF PROTEINS FROM “STARVE” SAMPLE AT 15 HOURS, IDENTIFIED AND QUANTITATED AS HAVING A RATIO OF GREATER THAN 1.84 .......................................... 321 TABLE 6.5. PROTEINS FOUND IN SILAC PROTEIN CONCENTRATION EXPERIMENTS AND IN HALF- LIFE EXPERIMENTS SHOWING A CHANGE IN BOTH CONCENTRATION AND IN HALF-LIFE .. 329 TABLE 6.6. PROTEINS FOUND IN SILAC (PROTEIN CONCENTRATION) EXPERIMENTS AND IN HALF-LIFE EXPERIMENTS SHOWING NO CHANGE IN CONCENTRATION AND A CHANGE IN HALF-LIFE ...................................................................................................................... 332 TABLE 6.7. PROTEINS FOUND IN SILAC (PROTEIN CONCENTRATION) EXPERIMENTS AND IN HALF-LIFE EXPERIMENTS SHOWING A CHANGE IN CONCENTRATION AND NO CHANGE IN HALF-LIFE ...................................................................................................................... 335 TABLE 6.8. PROTEINS FOUND IN SILAC (PROTEIN CONCENTRATION) EXPERIMENTS AND IN HALF-LIFE EXPERIMENTS SHOWING NO CHANGE IN CONCENTRATION OR IN HALF-LIFE. . 336 TABLE 6.9. NINE PROTEINS FOUND IN COMMON BETWEEN PUBLISHED PROTEIN DEGRADATION DATA AND HALF-LIFE DATA COMPUTED FROM OUR DATA SET........................................ 344    xii LIST OF FIGURES  FIGURE 1.1 APOPTOSIS SIGNALING PATHWAYS........................................................................ 17 FIGURE 1.2 THE BCL-2 FAMILY OF PRO-SURVIVAL AND PRO-DEATH PROTEINS........................ 21 FIGURE 1.3 FT-ICR MASS SPECTROMETRY. ............................................................................. 47 FIGURE 2.1 EXPERIMENTAL OVERVIEW FOR THREE CYTOKINE-WITHDRAWAL EXPERIMENTS. . 59 FIGURE 3.1 FLOW CYTOMETRY OF PROPIDIUM IODIDE-STAINED TF-1 CELLS. .......................... 86 FIGURE 3.2 DEVD-PNA CASPASE ACTIVITY ASSAY. ............................................................. 88 FIGURE 3.3 PROTEIN LOAD (#G) OF WHOLE CELL LYSATE AND BOVINE SERUM (BSA)............ 93 FIGURE 3.4 A 12% MINI-GEL WAS USED AS THE 40-SLICE GEL IN THE FIRST EXPERIMENT. ...... 95 FIGURE 3.5 AN 8% - 16.5% LARGE FORMAT GEL WAS USED AS THE 90-SLICE GEL FOR THE SECOND EXPERIMENT....................................................................................................... 98 FIGURE 3.6 AN 8%-16.5% LARGE FORMAT GRADIENT GEL WAS USED FOR THE 70-SLICE GEL FOR THE THIRD EXPERIMENT............................................................................................ 99 FIGURE 4.1 HISTOGRAM AND PROBIT CURVE OF DATA DISTRIBUTION. .................................. 129 FIGURE 4.2 LEVEL 4 BIOLOGICAL PROCESS GENE ONTOLOGIES OF SUBSET OF PROTEINS SHOWING AN INCREASE IN CONCENTRATION IN RESPONSE TO CYTOKINE WITHDRAWAL.146 FIGURE 4.3 LEVEL 4 BIOLOGICAL PROCESS GENE ONTOLOGIES OF SUBSET OF PROTEINS SHOWING A DECREASE IN CONCENTRATION IN RESPONSE TO CYTOKINE WITHDRAWAL.. 147 FIGURE 4.4 PROTEIN CATEGORIZATION BY GENE ONTOLOGY (I) – CATEGORIZATION BY LEVEL 4 TERMS OF A BIOLOGICAL PROCESS DIRECTED ACYCLIC GRAPH. ..................................... 150 FIGURE 4.5 PROTEIN CATEGORIZATION BY GENE ONTOLOGY (II) – CATEGORIZATION BY LEVEL 4 TERMS OF A BIOLOGICAL PROCESS DIRECTED ACYCLIC GRAPH. ..................................... 151 FIGURE 4.6 PROTEIN CATEGORIZATION BY GENE ONTOLOGY (III) – CATEGORIZATION BY LEVEL 4 TERMS OF A BIOLOGICAL PROCESS DIRECTED ACYCLIC GRAPH.................................... 175 FIGURE 4.7 PROTEIN CATEGORIZATION BY GENE ONTOLOGY (IV) – CATEGORIZATION BY LEVEL 4 TERMS OF A BIOLOGICAL PROCESS DIRECTED ACYCLIC GRAPH.................................... 176 FIGURE 4.8 PROTEIN CATEGORIZATION BY GENE ONTOLOGY (V) – CATEGORIZATION BY LEVEL 4 TERMS OF A MOLECULAR FUNCTION DIRECTED ACYCLIC GRAPH.................................... 202 FIGURE 4.9. FIRST EXAMPLE OF A FUNCTIONAL NETWORK GENERATED USING INGENUITY PATHWAY ANALYSIS ®. ................................................................................................ 214    xiii FIGURE 4.10. SECOND EXAMPLE OF A FUNCTIONAL NETWORK GENERATED USING INGENUITY PATHWAY ANALYSIS ®. ................................................................................................ 215 FIGURE 4.11. PROPOSED MODEL OUTLINING OBSERVED CHANGES IN PROTEIN CONCENTRATIONS AS THEY RELATE TO THE INDUCTION OF APOPTOSIS BY ALTERATIONS IN MEVALONATE PATHWAY AND CDC42 SIGNALING. ................................................................................ 221 FIGURE 5.1 IMMUNOBLOT OF WHOLE CELL LYSATES - CONTROL WITH HOURS OF CYTOKINE WITHDRAWAL AS SHOWN............................................................................................... 232 FIGURE 5.2 REPRESENTATIVE MASS SPECTRUM OF CDC42  SHOWING RATIO OF HEAVY TO LIGHT ISOFORMS OF PEPTIDE YVECSALTQK......................................................................... 233 FIGURE 5.3 IMMUNOBLOT OF WHOLE CELL LYSATES - CONTROL (CONT.) WITH 15- AND 18- HOURS OF CYTOKINE WITHDRAWAL............................................................................... 235 FIGURE 5.4 REPRESENTATIVE MASS SPECTRUM OF HMGB2 ................................................. 236 FIGURE 5.5 WESTERN BLOT SHOWING LEVELS OF SHIP IN RESPONSE TO 15-HOURS OF CYTOINE WITHDRAWAL. ............................................................................................................... 237 FIGURE 5.6 REPESENTATIVE MASS SPECTRUM OF SHIP......................................................... 238 FIGURE 5.7 THYMIDYLATE SYNTHASE (TS) IMMUNOBLOT SHOWING CONTROL AND 15-HOURS OF CYTOKINE WITHDRAWAL. ......................................................................................... 239 FIGURE 5.8 REPRESENTATIVE MASS SPECTRUM OF THYMIDYLATE SYNTHASE ...................... 239 FIGURE 5.9 WESTERN BLOT SHOWING LEVELS OF CLEAVED CASPASE 3 IN RESPONSE TO CYTOKINE WITHDRAWAL FOR THE TIMES SHOWN........................................................... 240 FIGURE 5.10 REPRESENTATIVE MASS SPECTRUM OF VINCULIN. ............................................ 242 FIGURE 5.11 REPRESENTATIVE MASS SPECTRUM OF HSP90-ALPHA (!) ................................. 243 FIGURE 5.12 REPRESENTATIVE MASS SPECTRUM OF HSP90-BETA (") ................................... 244 FIGURE 5.13 EIGHT PEPTIDES WERE USED IN THE IDENTIFICATION OF ANAMORSIN................ 248 FIGURE 5.14 ANAMORSIN IMMUNOBLOT IN FDCP-1 CELLS................................................... 254 FIGURE 5.15 (A) & (B). ANTI-ANAMORSIN IPS....................................................................... 255 FIGURE 5.16. ANTI-ANAMORSIN IMMUNOBLOT, 12% LOW-BIS GEL. ...................................... 255 FIGURE 5.15A ANTI-ANAMORSIN IP WITH 4G10 PROBE......................................................... 254 FIGURE 5.15B ANTI-ANAMORSIN IP WITH ANAMORSIN PROBE............................................... 254 FIGURE 5.17. ANTI-ANAMORSIN IP, ANTI-PHOSPHOSERINE PROBE......................................... 256 FIGURE 5.18. ANTI-ANAMORSIN IP TREATED WITH CALF INTESTINAL PHOSPHATASE. ........... 257    xiv FIGURE 5.19. ANAMORSIN RESPONSE TO CYTOKINE WITHDRAWAL FOR 6 AND 12 HOURS, AND TO LY294002 AND UO126............................................................................................ 258 FIGURE 5.20. ANAMORSIN EXPRESSION IN FDCP-1 CELLS FOLLOWING CYCLOHEXIMIDE TREATMENT. .................................................................................................................. 259 FIGURE 5.21. REPRESENTATIVE SPECTRA IDENTIFYING TWELVE OF THE PEPTIDES FROM TABLE 5.2 ................................................................................................................................. 269 FIGURE 5.22. IMMUNOBLOT SHOWING PDCD4 IN TF-1 CELLS - CONTROL WITH 15- AND 18- HOURS OF CYTOKINE WITHDRAWAL............................................................................... 273 FIGURE 5.23. WB SHOWING PDCD4 IN CONTROL AND 15 HOURS CYTOKINE WITHDRAWAL (ST). .............................................................................................................................. 274 FIGURE 5.24. WB SHOWING PDCD4 IN CONTROL WITH 15-HOURS CYTOKINE WITHDRAWAL. CELLS LYSED DIRECTLY INTO HOT LOADING BUFFER. .................................................... 275 FIGURE 6.1 LEVEL 4 DAG FOR BIOLOGICAL PROCESS GENE ONTOLOGIES. PROTEINS IN CONTROL AND STARVING CONDITIONS WITH HALF-LIFE LESS THAN 15 HOURS. ............. 339 FIGURE 6.2. LEVEL 4 DAG FOR MOLECULAR FUNCTION GENE ONTOLOGIES. PROTEIN IN CONTROL AND STARVING CONDITIONS WITH HALF-LIFE LESS THAN 15 HOURS. ............. 340  341 FIGURE 6.3. LEVEL 4 DAG FOR BIOLOGICAL PROCESS GENE ONTOLOGIES. PROTEINS IN CONTROL AND STARVING CONDITIONS WITH HALF-LIFE GREATER THAN 24 HOURS. ...... 341 FIGURE 6.4. LEVEL 4 DAG FOR MOLECULAR FUNCTION GENE ONTOLOGIES. PROTEIN IN CONTROL AND STARVING CONDITIONS WITH HALF-LIFE GREATER THAN 24 HOURS. ...... 342       xv LIST OF ABBREVIATIONS  AIF apoptosis inducing factor APAF-1 apoptotic protease activating factor 1 Bcl-2 B-cell lymphoma-2 BH Bcl-2 homology CARD caspase recruitment domain CGMI BHK cell line stably transfected with human GM-CSF gene Da Dalton (unit of mass) DAG directed acyclic graph DD death domain DISC death-inducing signaling complex Erk extracellular signal-related kinase ESI electrospray ionization FADD Fas associated death domain Fas-L Tumor necrosis factor ligand superfamily, member 6 FDR false discovery rate FT-ICR Fourier transform ion cyclotron resonance mass spectrometry GM-CSF granulocyte-macrophage colony stimulating factor GO gene ontology ICAT isotope-coded affinity tags IL-3 interleukin-3 IL-5 interleukin-5 IPA Ingenuity® Pathway Analysis JAK Janus kinase LC-MS/MS liquid chromatography-tandem mass spectrometry LTQ-FT Thermo Electron IonTrap / FT-ICR mass spectrometer MALDI matrix-assisted laser desorption and ionization MAPK Mitogen-activated protein kinase MRM multiple reaction monitoring MS mass spectrometry    xvi NF-$B nuclear factor kappa b OD optical density PBS phosphate buffered saline PDK-1 3-phosphoinositide-dependent protein kinase 1 PH pleckstrin homology PI-3,4-P2 phosphatidylinositol-3,4-bisphosphate PI-3,4,5-P3 phosphatidylinositol-3,4,5-trisphosphate PI3-K phosphatidylinositol-3 kinase PKB protein kinase B PTEN Phosphatase and tensin homologue PTM post-translational modification PTP (mitochondrial) permeability transition pore SDS-PAGE sodium dodecylsulfate polyacrylamide gel electrophoresis SH-2 Src-homology-2 SHIP SH2 domain-containing inositol-5'-phosphatase SILAC stable isotope labeling with amino acids in cell culture STAT Signal transducer and transcription activator TBS Tris buffered saline TBST Tris buffered saline with Tween® TF-1 an erythroleukemia human cell line Th Thomson (unit of mass-to-charge ratio) TNF tumour necrosis factor TOF time of flight TRADD TNF receptor activated death domain TRAIL TNF-related apoptosis-inducing ligand     xvii ACKNOWLEDGEMENTS   Dr Vincent Duronio – for the opportunity he has given me; who, through the trust he has placed in me (as in all his students), has allowed me to pursue this long-held dream, and who has imbued in me some of his own integrity, and scholarship, as well as his rigor and critical thinking – enhancing not only my knowledge and skill, but my understanding of what it means to be a scientist.  Dr Juergen Kast, a member of the Supervisory Committee, without whose willingness to collaborate, this project could not have been completed; who has always been ready to give of his time, to advise and assist.  Dr Youssef Av-Gay and Dr Aly Karsan, members of the Supervisory Committee, for the scholarship and enthusiasm they have consistently brought to this endeavour. Thank you.  Dr Leonard Foster, who has always been gracious and so generous with the gifts of his time and knowledge.  Jason Rogalski, for his technical skill and support, as well as his engaging discussions.  Shujun Lim, for her technical expertise, her patience, and her willingness to help.  Payman Hojabrpour, for his technical assistance, and his skill in managing the Duronio laboratory.  Penny Brasher, PhD, for her statistical insights, and her willingness to be available to discuss this project.  David Creese, for his skill in translating my Latin dedication.     xviii Andrei Godoroja, for his friendship and support, and for the Perl ® scripts which made some of the data handling possible.  Matthew Sniatynski, for his Python ® script which saved me so much time.  My colleagues and fellow lab members: Dr Sarwat Jamil, Sherry Wang, Ivan Waissbluth and Stefanie Cheah for their encouragement and support during the difficult times, and their laughter during the good times.  David Shih, for his help with some of the Western blot images in Chapter 5.  Greg Martin, PhD, for his friendship, and mathematical insights, especially the equations for the protein half-life calculations.  Elizabeth Dean, P.T., PhD, Ginny Mulhall, Sheila Mannell for their friendship and support, without which, this undertaking would have been much more difficult.   xix DEDICATION     “Dicebat Bernardus Carnotensis nos esse quasi nanos, gigantium humeris insidentes, ut possimus plura eis et remotiora videre, non utique proprii visus acumine, aut eminentia corporis, sed quia in altum subvenimur et extollimur magnitudine gigantea.” John of Salisbury [d.1180], Metalogicon [III, lv, 900a]      Eis in quorum vestigiis semitisque ambulavi gratias ago et haec ego scripta dedico.*              *I thank those in whose footsteps and paths I have trod, and dedicate these writings to them.      1 1 Introduction  Apoptosis, one form of programmed cell death, is essential for life. This natural process of cell loss is critically important to the function of the organism, in such diverse ways as (i) embryonic development, e.g. for neural development (Putcha & Johnson, 2004) and for the functional development of the hands and feet (Jacobson, Weil, & Raff, 1997), (ii) cellular (tissue) homeostasis (Green & Evan, 2002), or (iii) tissue remodeling during healing or functional regression, e.g. cessation of lactation after weaning (Baxter, Neoh, & Tevendale, 2007), (iv) for defense, e.g. for the control of lymphocyte proliferation (Newton & Strasser, 2000), (v) for preventing the development of auto-immunity by the deletion of self-reactive T-cells (McCaughtry & Hogquist, 2008), (vi) as a defense against pathogens (Lilley, Schwartz, & Weitzman, 2007), and (vii) in aging (Y. Zhang & Herman, 2002)). The process of apoptosis is highly conserved throughout evolution, with evidence for apoptosis found in all metazoan organisms from marine sponges to man (Earnshaw, Martins, & Kaufmann, 1999; Twomey & McCarthy, 2005; Wiens, Krasko, Muller, & Muller, 2000). Indeed, for several years evidence has been accumulating to suggest that much simpler organisms, such as the unicellular eukaryotes Leishmania sp. and yeast, undergo a form of apoptosis  (Arnoult et al., 2002; Buttner et al., 2006; Eisenberg, Buttner, Kroemer, & Madeo, 2007; Frohlich, Fussi, & Ruckenstuhl, 2007).  Apoptosis, the process that is so critically important for the development and homeostasis of an organism, can cause morbidity when it goes awry. For example, in the case of cells that suffer DNA damage that is so severe that the cellular DNA repair machinery cannot repair the damage, the cells should be directed towards apoptosis. If, for some reason, cells escape    2 the signal to undergo apoptosis or have defective apoptosis machinery, allowing survival and subsequent division, then the damaged DNA may be copied and transmitted to daughter cells following mitosis. Further DNA damage can result from this reproduction of abnormal genetic material. Evading apoptosis is the classic pathway to oncogenesis (Bellamy, Malcomson, Harrison, & Wyllie, 1995; Letai, 2008). However, DNA damage is not the only cause of apoptosis - viral infection can induce apoptosis (Lilley et al., 2007), as can stress, for example, starvation (Schamberger, Gerner, & Cerni, 2005), oxidative stress (Buttke & Sandstrom, 1995) or nitrosative stress (Thomas et al., 2008). In other cases, cells that are healthy may improperly be directed towards apoptosis, which results in the loss of functional tissue. This occurrence has been implicated in conditions such as Alzheimer’s disease (Lafay-Chebassier et al., 2005), Parkinson’s disease (Burke, 2008), AIDS (Alimonti, Ball, & Fowke, 2003) and osteoarthritis (Del Carlo & Loeser, 2008).  Although the process we understand as apoptosis has been described since the mid- nineteenth century (Clarke & Clarke, 1995), a deeper understanding of this process has only been possible with the development of the more sophisticated histological studies in the 1970s, and biochemical studies beginning in the early 1980s. Despite decades of work, there remain many unanswered questions concerning the details of the signaling pathways and molecular interactions involved in apoptosis. An enhanced understanding of the complex process of apoptosis may allow better treatments of the diseases caused by “inappropriate” apoptosis, or those caused by evasion of apoptosis. “Many human diseases can be attributed directly or indirectly to a derangement of apoptosis, resulting in either cell accumulation, in which cell eradication or cell turnover is impaired, or cell loss, in which the apoptotic    3 programme is inadvertently triggered” (Fadeel & Orrenius, 2005). However, there remains always more work to be done to understand the many cellular responses that occur in cells undergoing apoptosis. The control of any complex process such as apoptosis is achieved via hundreds of signalling proteins. These multiple signalling components make up a cellular machinery that is not composed simply of linear signaling elements or pathways, but rather, one which can be seen as a complex network of interconnections (Pawson & Saxton, 1999). Perturbation of one element of the network disturbs the whole “structure” – much like plucking one strand of a spider-web may distort the whole web.   A number of techniques have been developed for the investigation of signaling networks, which involve perturbation of one element of the network. For example, by selective inhibition of one  “node” in the network by using a pharmacological small molecule inhibitor; or by enhancing the translation of one gene, and over-expressing its protein product; or by inhibiting translation of one protein, and determining the effects of these changes on the whole system. Antibodies to a protein of interest are frequently used to visualize the results of any perturbation of the system. These techniques have worked extremely well – the thousands of publications generated by scientists using these well- established techniques are a tribute not only to the ingenuity of these investigators, but to the effectiveness of this approach in dissecting the elements of the apoptosis machinery. Indeed, such is the state of knowledge generated by this “single element” approach to studying apoptosis that the scientific community is now able to begin to investigate cellular function from a more “global” perspective; we understand many of the individual elements well enough now to start to see the picture as a whole.    4  One more recently devised approach is to use techniques that take “snapshots” of the whole cellular system at one point. In this regard, recent advances in experimental techniques have allowed for the development of the field of proteomics  - analysis of the whole protein content of a group of cells (or as much of the whole as may be visualized with current technology) as an approach to answer questions of cellular function.  The work described in this thesis has made use of a proteomics approach to further our understanding of the signaling pathways involved in apoptosis, specifically, the changes encountered in early or “pre-apoptosis”, using cytokine-dependent hematopoietic cells as an experimental model. The PI3-K pathway is an important element in this approach because of the pivotal role that PI3-K and its key downstream kinase, PKB/Akt, plays in cellular survival (reviewed in (Cantley, 2002; Duronio, 2008). While the ultimate goal of apoptosis research may be considered the development of new therapeutic approaches for the treatment of cancer and other diseases of inappropriate apoptosis, new therapeutics need to be built upon a molecular understanding of apoptosis (Antignani & Youle, 2005; Labi, Grespi, Baumgartner, & Villunger, 2008) and it is in this area, the molecular understanding of apoptosis, that this current work has been able to make a contribution. The results shown are unexpected in the wide range of proteins affected by withdrawal of pro-survival cytokine signaling. In over 1450 proteins found in common in three biological replicates of our experiment, totalling 200 individual LC-MS/MS runs, we observe both increases and decreases in the concentrations of a large percentage of proteins, and we have classified these proteins with altered concentrations into groups based on gene ontologies, to allow further    5 analysis of the complex response to cytokine withdrawal. Further, we have been able to demonstrate that the techniques utilized in this work are sufficiently robust to be applied to answer questions such as the ones we posed. As well, we show some preliminary data obtained using stable isotope labelling, concerning apparent changes in protein half-life in response to cytokine withdrawal.  Although the TF-1 cell line does not perfectly mimic human primary cells, the study is valuable in as much as the TF-1 cell line is a human cell line, hence similarities in proteome are to be expected. As well, the cytokine dependence of the TF-1 cell line remains a relevant model in which to assess apoptosis. However, there are inherent limitations in studying cell lines, and eventually these data would need to be compared with those obtained by the study of primary cells. This notwithstanding, the proteomics approach has been useful in helping to elucidate the complex changes occurring in protein concentrations, resulting either from changes in translation or in degradation, that occur when cells undergo apoptosis, and further, has provided valuable data which will help to direct future studies. 1.1 Apoptosis 1.1.1 Classification of Cell Death  Cell death may be broadly classified as occurring in one of two modes; either through programmed cell death, or through necrosis, although the distinction is becoming increasingly blurred as our understanding of these processes deepens. Variations of programmed cell death have been described over the years, using different means of classification. The current classification for cell death modes, based on the morphological    6 changes observed, consists of – apoptosis, autophagy, necrosis and mitotic catastrophe (Galluzzi et al., 2007). Other forms of cell death, such as anoikis, and caspase-independent programmed cell death have been described (Bursch, Ellinger, Gerner, Frohwein, & Schulte- Hermann, 2000; Frisch & Screaton, 2001; Kroemer & Martin, 2005). These latter modes of cell death are based upon observations other than morphological changes, and so are not included in the classification above. A further form of programmed cell death, called pyroptosis, has also been recently described. Pyroptosis is a novel form of cell death induced by infections with Salmonella and Shigella species (Fink & Cookson, 2005). These various types of cell death differ substantially in process, although it is not always easy to distinguish one type from another (Galluzzi et al., 2007).    7 1.1.2 Brief Historical Overview  Karl Vogt (1817-1895), a German physician and scientist who published in the fields of zoology, physiology and geology, first reported apoptosis while studying the development of the tadpole of the midwife toad (Alytes obstetricians). Vogt’s description is found in his work entitled “Untersuchungen über die Entwicklungsgeschichte der Geburtshelferkröte“ (Alytes obstetricians) (Jent & Gassmann, Solothurn, 1842) (Vogt, 1842). In this work, Vogt describes the “resorption” (or absorption) of the cells of the notochord in response to pressure exerted by the “rampant growth” of the surrounding cartilage (p 86 – my loose translation).  Walther Flemming, the German physician and scientist who first stained and coined the term “chromatin” and also coined the term “mitosis”, gave a more detailed description of the process of apoptosis in his observations on the degeneration of antral follicles in the rabbit ovary ((Flemming, 1885), quoted in (Clarke & Clarke, 1996)). Flemming named the process he observed “chromatolysis” because of the changes observed in nuclear chromatin.  These are but two of the early pioneers in the study of apoptosis. There are very many others who have contributed to the field over the past hundred and fifty years. An extensive review of the nineteenth century literature concerning naturally occurring cell death was undertaken by Clarke and Clarke (Clarke & Clarke, 1996). This review makes fascinating reading for those interested in the early history of the field of apoptosis. Another informative document is the 1997 historical review by Häcker and Vaux (Hacker & Vaux, 1997). The work described in these early accounts was based on observation, mostly using tissue staining and    8 light microscopy, and so it was to remain until the early 1970s, when the concept of apoptosis as we understand it began to form in the minds of John Kerr, Andrew Wylie, and Alastair Currie.  In 1962, as part of his Ph.D. work, John Kerr began investigating the cellular processes involved in the shrinkage of liver tissue following the induction of ischemia. Kerr observed the induction of classical necrosis in the ischemic tissue, but also observed a different type of cell death – some dying cells were converted to small, round masses, with condensed nuclear chromatin. These cell remnants were engulfed by other hepatocytes, and by phagocytes. Histopathology of the shrunken cells showed that the lysosomes were intact, as well as the mitochondria and ribosomes, a finding that suggested that these shrunken cells were not undergoing a form of necrotic cell death. This “newly” observed cell death was initially named “shrinkage necrosis” (Kerr, 1965).  Electron microscopy of shrinkage necrosis In 1971, Kerr collaborated with Andrew Wyllie and Alastair Currie in work which showed that shrinkage necrosis was observed in the adrenal cortex of rats treated with prednisolone, as well as in neonatal rats where the levels of adrenocorticotrophic hormone (ACTH) taper physiologically. Shrinkage necrosis was also observed in breast carcinomas of female rats that had been treated by oophrectomy. Discussion with Allison Crawford at that time brought to light the concept of cell death during normal embryonic development – a concept previously unknown to most who were not developmental biologists. From this “serendipitous confluence of ideas” (Kerr, 2002), the concept of apoptosis was born (Kerr,    9 Wyllie, & Currie, 1972). The word “apoptosis” was coined by Professor James Cormack, of the Department of Greek in the University of Aberdeen (from the Greek !"#"$%&'(, which is derived from !")- in Greek, apo-, in transliteration =  “from”, and "$*&'( in Greek, pt!sis, in transliteration =  “falling” in the sense of leaves falling from trees in the autumn) (Kerr et al., 1972). 1.1.3 Description of Apoptosis  Apoptosis has been extensively characterized in many sources. Two (of many) helpful reviews of apoptosis have been published by Majno and Joris (Majno & Joris, 1995) and by Hengartner (Hengartner, 2000). Briefly, apoptosis involves the tidy elimination of cells by an ordered process which is characterized histologically by cell shrinkage, cell membrane blebbing, nuclear chromatin condensation (pyknosis) and margination (becoming packed against the inner nuclear membrane), and by nuclear DNA cleavage into segments that are approximately 185 bp long, or multiples of this. The cell generates membrane bound “processes”, by budding, which can detach from the cell body as vesicles, to become apoptotic bodies.  The vesicles often contain pyknotic nuclear fragments. Apoptotic bodies undergo phagocytosis by macrophages, or other specialized cell types. Biochemically, these changes are brought about principally by the activation of eight proteolytic enzymes from the family of caspases (see 1.1.4 below). Caspases selectively cleave a subset of proteins causing inactivation (in most cases). For example, the caspase-activated DNase (CAD) is responsible for cutting genomic DNA between nucleosomes, leading to DNA fragments of approximately 185 bp or integer multiples of this. Caspase cleavage of nuclear lamins causes nuclear shrinking and budding, while cleavage of cytoskeletal proteins such as fodrin and gelsolin is thought to be responsible for loss of the cell shape. Cleavage of PAK2 (a member    10 of the p21-actvated protein kinase family), which causes constitutive activation, seems to be responsible for active blebbing. Apoptosis is tightly regulated and morphologically uniform across cell types. Despite this, events that occur downstream of caspase activation have not yet been fully decribed. Apoptosis can occur very quickly – it has been observed to progress from the onset of budding to complete cell breakup within thirty-four minutes. The process takes place in the absence of inflammation - apoptosis is a non-inflammatory cell death process. In contrast, necrosis is an inflammatory process, morphologically distinct from apoptosis and autophagy. Necrosis is a descriptive term used when dead cells or tissues are present (Fink & Cookson, 2005) – the remnants of “untidy” cell death. Necrotic cell death is marked by cytoplasmic vacuolation, plasma membrane breakdown and the presence of inflammation around the dying cell, due to release of pro-inflammatory cytokines from the cell (Edinger & Thompson, 2004). Necrosis differs biochemically from apoptosis and autophagy by the absence of cytochrome c release from mitochondria, and the absence of caspase activation and DNA fragmentation (Krysko, Vanden Berghe, D'Herde, & Vandenabeele, 2008).  Biochemical studies of the molecular process of apoptosis first started to appear in the early 1980s with publication of the discovery that one early feature of apoptosis is the activation of a Ca ++ -  and Mg ++ -dependent DNA-endonuclease (Duke, Chervenak, & Cohen, 1983). It was not long before other enzymes were discovered which become activated as part of the apoptosis process: the lysosomal protease cathepsin D (Levy-Strumpf & Kimchi, 1998; Tanabe, Lee, & Grayhack, 1982); tissue-type plasminogen activator (Rennie, Bouffard, Bruchovsky, & Cheng, 1984); collagenases and metalloproteinases. RNase has been shown    11 to play a role in apoptosis by degrading mRNA and ribosomal RNA. Another important biochemical change in early apoptosis is the switch in orientation of phosphatidylserine (PS) from the inner leaflet of the lipid bilayer cell membrane to the outer leaflet (Homburg et al., 1995; S. J. Martin et al., 1995). It was thought that this process served to “mark” the apoptotic cells for phagocytosis by macrophages (Fadok et al., 1992). However, recent evidence suggests that the presence of PS on the outer leaflet of the bilayer cell membrane is due to a failure of the unidirectional aminophospholipid translocase, which normally excludes PS from the outer membrane and which requires ATP to function (reviewed in (Vance, 2008). Indeed the identity of a PS receptor on macrophages remains controversial.  The aforementioned discoveries were of enormous importance in defining apoptosis, and in delineating the process. However, it was the discovery of the caspases that was critical to understanding the “mechanics” of the process of apoptosis. 1.1.4 Caspases  The product of the gene ced-3, was known to be essential for the developmental apoptosis that takes place in the nematode Caenorhabditis elegans (Ellis & Horvitz, 1986). In 1992, it was recognized by Horvitz and colleagues that the mammalian equivalent of CED-3 is interleukin-1!-converting enzyme (ICE, or caspase-1) (Thornberry et al., 1992; Yuan, Shaham, Ledoux, Ellis, & Horvitz, 1993), an enzyme involved in inflammation (although not directly in cell death, unless over-expressed). Soon after this discovery, it was learned that over-expression of murine ICE in rat cells caused the induction of apoptosis (Miura, Zhu, Rotello, Hartwieg, & Yuan, 1993). ICE (as caspase-1) thus became the founding member of a family of cysteine-aspartic acid proteases, the caspases, of which there are now fourteen    12 members (Degterev, Boyce, & Yuan, 2003). Caspases function in both inflammation and cell death, and may be classified accordingly (see, e.g. (Algeciras-Schimnich, Barnhart, & Peter, 2002; Newton & Strasser, 2003; Siegel, 2006)). Most of the morphological changes observed in cells undergoing apoptosis can be explained by the action of the caspases (Hengartner, 2000), although cells can survive low levels of caspase activation  and caspases do not provide the only pathway to programmed cell death – see caspase-independent cell death (below). A large number of caspase substrates have been identified (Earnshaw et al., 1999; Stroh & Schulze-Osthoff, 1998; Utz & Anderson, 2000), and the list continues to grow (but see also the caveat in (Timmer & Salvesen, 2007)). Recent evidence suggests that the caspases also play roles in development, activation and differentiation of cells of the immune system (Newton & Strasser, 2003; Siegel, 2006), and perhaps other tissues (Algeciras- Schimnich et al., 2002). 1.1.5 Initiation Phase and Execution Phase  Apoptosis is described as having two distinct phases, an initiation phase and an execution phase. (Leber, Lin, & Andrews, 2007; Zimmermann & Green, 2001). During the initiation phase, caspase-8 or caspase-9 become activated by protein-protein interactions. Two pathways have been described; the internal pathway (for activation of caspase-9), and the external pathway (for activation of caspase-8) (see discussion below). During the execution phase, downstream caspases become activated by up-stream caspase(s), leading to a cascade of signals, causing the induction of apoptosis.     13 1.1.6 Internal Pathway Vs. External Pathway  Figure 1.1 shows an overview of the internal and external pathways to the activation of the caspase cascade. i. Internal or intrinsic pathway (mitochondrial mediated) The internal or intrinsic pathway to apoptosis is mediated by the mitochondria, beginning with permeabilization of the outer mitochondrial membrane. Mitochondria undergo two changes early in apoptosis – permeabilization of the outer mitochondrial membrane, regulated by protein-protein and protein-membrane interactions of Bcl-2 family members (Chipuk & Green, 2008; Scorrano & Korsmeyer, 2003); and a decrease in mitochondrial transmembrane potential (%&m), critically important for oxidative phosphorylation (Green & Reed, 1998). A decrease in %&m has been observed in many different cell types early in apoptosis. (Deckwerth & Johnson, 1993; Zamzami, Marchetti, Castedo, Decaudin et al., 1995; Zamzami, Marchetti, Castedo, Zanin et al., 1995) The reduction in mitochondrial transmembrane potential causes a decrease in mitochondrial protein translation and gene transcription (Vayssiere, Petit, Risler, & Mignotte, 1994). While it has been argued that a reduction in %&m allowed the escape of cytochrome c from the mitochondria via the formation of a permeability transition pore (PTP), Green and colleagues have convincingly shown that cytochrome c release is independent of the change in %&m(Bossy-Wetzel, Newmeyer, & Green, 1998). Thus, the role of mitochondrial membrane depolarization in apoptosis remains unclear. However, the permeabilization of the outer mitochondrial membrane is still considered to be the “point of no return” in many models of apoptosis (reviewed in (Leber et al., 2007)).    14 Cytochrome c, released from the space between the inner and outer mitochondrial membranes, binds in the cytoplasm with ”apoptotic protease activating factor 1” (APAF-1), in the presence of deoxyadenosine triphosphate (X. Liu, Kim, Yang, Jemmerson, & Wang, 1996), and these two molecules oligomerize, causing a conformational change in APAF-1, which exposes the CARD (caspase recruitment domain), allowing pro-caspase-9 to bind and activate its proteolytic function (P. Li et al., 1997). Activated caspase-9 cleaves pro-caspase- 3, which becomes active caspase-3. Active caspase-3 cleaves the CAD/I CAD  complex, freeing CAD (caspase-activated deoxyribonuclease), which causes nuclear DNA fragmentation (X. Liu, Zou, Slaughter, & Wang, 1997). Caspase-3 also activates caspase-7 and caspase-6, each of which has multiple protein targets, causing the cell to proceed to apoptosis (Nicholson & Thornberry, 1997). A second activator of the caspase cascade, called SMAC/Diablo, is released along with cytochrome c from the mitochondria during apoptosis (Du, Fang, Li, Li, & Wang, 2000; Verhagen et al., 2000). SMAC/Diablo binds to members of the “inhibitors of apoptosis” (IAPs) family, which are endogenous protein inhibitors of active caspases, normally associated with caspase-9, caspase-3 and caspase-7 (reviewed in (Richter & Duckett, 2000; Vaux & Silke, 2005)). SMAC/Diablo disrupts the association of XIAP with cleaved caspase- 9, allowing caspase-9 to activate caspase-3 (Ekert, Silke, Hawkins, Verhagen, & Vaux, 2001). ii. External pathway (receptor mediated) External signals, mediated by molecules such as TNF, TRAIL/Apo-2L and Fas-L, can induce apoptosis in target cells expressing the corresponding cell surface receptor to which these ligands can bind.    15 TNF – the prototypic member of a family of pro-apoptosis cytokines - was discovered as the TNF-' and TNF-( forms in 1968 and 1975 respectively (Carswell et al., 1975; Kolb & Granger, 1968). TNFs are mainly produced by activated macrophages, but also by other cell types, including mast cells, endothelial cells and fibroblasts. TNF causes apoptosis of certain tumour cells, modulates immune function and mediates the inflammatory response (Locksley, Killeen, & Lenardo, 2001). TNF has been recognized as a major cytokine in the pathogenesis of chronic inflammatory disease (Clark, 2007). The TNFs can have opposing signalling effects in target cells, depending on the intracellular signalling molecules engaged by the receptor, but this discussion will focus on the pro-apoptotic effects. A large family of receptors binds TNF family members – the TNF Receptor Superfamily. The binding of ligand to receptor exposes a ‘death domain’ (DD), which engages the cytoplasmic TNF receptor activated death domain-containing (TRADD) protein and Fas-associated death domain protein (FADD). The complex of these proteins leads directly to the activation of caspase 8, and to indirect activation of other caspases (Locksley et al., 2001). Specifically, procaspase-8 is activated by association with the co-factor FADD, binding via the receptor’s DD (Boldin, Goncharov, Goltsev, & Wallach, 1996; Muzio et al., 1996). Procaspase-8 molecules, when in proximity, can activate each other, generating caspase-8. Active caspase- 8 cleaves Bid, creating active tBid (X. Luo, Budihardjo, Zou, Slaughter, & Wang, 1998), which induces Bax oligomerization (Roucou, Montessuit, Antonsson, & Martinou, 2002), allowing Bax to permeabilize the outer mitochondrial membrane, causing the release of cytochrome c (Roucou et al., 2002). In a similar way, FasL, the ligand that binds the Fas receptor, induces the formation of a death-inducing signaling complex (DISC), by recruiting FADD to the Fas receptor’s DD.    16 The complex, consisting of Fas receptor, FADD and recruited procaspase-8, similarly causes the activation of caspase-8 and leads to apoptosis.    17   Figure 1.1 Apoptosis signaling pathways. Cellular stress induces proapoptotic Bcl-2 family members to translocate from the cytosol to the mitochondria, where they induce the release of cytochrome c. Cytochrome c catalyzes the oligomerization of Apaf-1, which recruits and promotes the activation of procaspase-9. This, in turn, activates procaspase-3, leading to apoptosis. Alternatively, the activation of caspase-8 by ligation of the death receptor Fas is illustrated. Ligated Fas recruits FADD to the intracellular region, which in turn recruits procaspase-8. The procaspase-8 transactivates, and the mature caspase now can cleave and activate procaspase-3, leading to apoptosis. Signaling from the Fas receptor to mitochondria involves cleavage of the BH3-only protein, Bid, by caspase-8. Bid subsequently induces cytochrome c release and downstream apoptotic events. (Reprinted from Journal of Allergy and Clinical Immunology, Vol 108/Issue 4, Katja C. Zimmerman and Douglas R. Green, How Cells Die: Apoptosis pathways, Pages S99-S103, Copyright (2001), with permission from Elsevier).     18 1.1.7 Regulation of Caspases  Caspases are tightly regulated, since the improper timing of caspase activation would have disastrous consequences for a cell. Caspases are produced as inactive zymogens, constitutively expressed, and become active by proteolytic cleavage. The initiator caspases are activated as a result of their associations with each other (which brings them into proximity so that they undergo self-proteolysis), and the effector caspases are activated by the action of the initiator caspases, releasing the inhibitory portion of the molecule (Duan, Chinnaiyan et al., 1996; Duan, Orth et al., 1996). Inhibition of caspase activity is achieved by  (i) regulating the interaction between caspases and their cofactors, (ii) by compartmentalizing the various elements of a caspase activation complex, or (iii) by inhibiting the effector caspases (or procaspases) directly. Caspase-8 may be inhibited by FADD-like ICE inhibitory proteins (FLIPs) (Irmler et al., 1997) – probably by competitive inhibition between caspase-8 and its cofactor FADD. Caspase-9 appears to be inhibited by ARC (apoptosis repressor with caspase recruitment domain), a CARD-domain-containing competitive inhibitor (Koseki, Inohara, Chen, & Nunez, 1998). Early work by Liu and colleagues suggests that another means of control of caspase activation is the compartmentalization of the elements required to activate caspases - specifically the presence of cytochrome c being sufficient to activate caspase-9 in the presence of dATP and APAF-1 in vivo, suggesting that the presence of the elements that make up the caspase activation complex is sufficient to cause activation of the caspase (X. Liu et al., 1996). Consequently by keeping the elements apart, caspase activation may be inhibited.    19 Lastly, the discovery of selective inhibition of caspase-3 and caspase-7 has led to an understanding of negative regulation of these initiator caspases by specific IAPs (inhibitors of apoptosis) (Deveraux et al., 1998), which function to inhibit a broad spectrum of apoptotic stimuli (reviewed in (Deveraux & Reed, 1999; LaCasse, Baird, Korneluk, & MacKenzie, 1998). 1.1.8 Caspase-independent Pathway (AIF mediation)  Other pathways to programmed cell death exist, which are not mediated by the caspases. Two families of proteases, the calpains, and the cathepsins (lysosomal proteases), as well as AIF (apoptosis inducing factor) have been shown to cause programmed cell death without activation of the caspases. For example, cathepsin D has been shown to trigger Bax activation and mitochondrial translocation, causing the selective release of AIF from the mitochondrial inner membrane space, inducing apoptosis in a caspase-independent manner (Bidere et al., 2003). There is some evidence to suggest that the calpains can induce cell death without the activation of the caspases, at least under some circumstances (Mathiasen, Lademann, & Jaattela, 1999; Mathiasen et al., 2002; Narvaez & Welsh, 2001) implicating the endoplasmic reticulum in some forms of programmed cell death. Lastly, AIF appears to be capable of causing “large scale DNA fragmentation and peripheral chromatin condensation” in a caspase-independent manner (Susin et al., 2000). However, it should be borne in mind that in some cases, caspase-independent apoptosis might have appeared to be “caspase- independent” because of the inability to detect low levels of caspase activity that could have been present, and still have played a part in inducing cell death.    20 1.1.9 From Pathology to Molecular Biology 1.1.9.1 Bcl-2 Family  David Vaux and others in 1988 identified B-cell lymphoma-2 (BCL-2) as a pro- survival gene (Vaux, Cory, & Adams, 1988). The BCL-2 gene product was found not to cause cell proliferation, as most of the oncogenes discovered to that time had done, rather, overexpression of BCL-2 caused an inhibition of programmed cell death (Vaux et al., 1988). In the intervening years it has been demonstrated that members of the Bcl-2 family can play both positive and negative roles in regulation of apoptosis (Adams & Cory, 1998). 1.1.9.2 Pro- and Anti-Apoptotic Members  The Bcl-2 family is currently comprised of twenty members. All contain '-helical sequences homologous to that of the Bcl-2 Homology (BH) domains in the founding member, Bcl-2. Domains are labeled BH1, BH2, BH3 and BH4. The Bcl-2 family members are categorized as pro-survival, all of which contain at least 3 BH domains (e.g. Bcl-2, Bcl-XL, Mcl-1 (mcl-1 doesn’t have a BH4)), or pro-death. Pro-death Bcl- 2 family members are subdivided into those containing several BH domains (e.g. Bax and Bak) and those containing only the BH3 domain (e.g. Bid, Bim, Bad). The Bcl-2 family members are shown in figure 1.2, reproduced with permission from a recent review article by Youle and Strasser (Youle & Strasser, 2008).      21                                     Figure 1.2 The Bcl-2 family of pro-survival and pro-death proteins. (Reprinted by permission from Macmillan Publishers Ltd: Nature Reviews Molecular Cell Biology, Youle RJ and Strasser A, vol. 9 (1) pp. 47-59, © 2008.)      22 1.1.9.3 Function of Bcl-2 Family Members  The function of the Bcl-2 family members remains a source of some controversy. The consensus opinion seems to be that interactions between pro- and anti-apoptotic members of the Bcl-2 family largely determine cell fate. The balance between the function of the pro- survival and the pro-death members of the Bcl-2 family is an important determinant of cell survival or death. The pro-death members are thought to function by binding via their BH3 domain to pro-survival members of the family (Zha, Aime-Sempe, Sato, & Reed, 1996), and thus inhibiting the function of the pro-survival proteins, especially as they guard Bax and Bak (Willis et al., 2007). The activation of Bax and Bak is generally held to allow the formation of the mitochondrial outer membrane permeabilization (MOMP) (Kluck, Bossy- Wetzel, Green, & Newmeyer, 1997; Wei et al., 2001). The MOMP forms in response to pro- apoptotic signals, and allows proteins such as cytochrome c and SMAC/Diablo to leak from the mitochondrial intramembrane space. Once this occurs, the mitochondrial pathway to apoptosis proceeds (Martinou & Green, 2001; Waterhouse, Ricci, & Green, 2002). The precise mechanism by which this occurs remains unclear. However, it is known that over- expression of Bcl-2 prevents the release of cytochrome c, as well as SMAC/DIBALO from the mitochondria, and inhibits apoptosis (Adrain, Creagh, & Martin, 2001; J. Yang et al., 1997), at least under some circumstances.  Three models have been proposed to describe the interaction between pro-survival and pro- apoptotic members of the Bcl-2 family (Chipuk & Green, 2008)– firstly the “rheostat model”, in which the cell exists in a steady state where pro-apoptotic and anti-apoptotic Bcl- 2 proteins are actively functioning, but are held in balance, with the result being cell survival.    23 However, disruption of the balance, such as by DNA damage, or cytokine withdrawal (in cytokine dependent cells) causes the balance to shift in favour of the pro-apoptotic proteins, and cell death follows. The second model is the “anti-apoptotic protein neutralization model”. Here the activation of Bax and Bak is continually inhibited by the action of the pro- survival Bcl-2 family members. When the function of the pro-survival Bcl-2 proteins is inhibited, for example by DNA damage or other stress, the inhibition ceases, and Bax and Bak become activated, with the formation of the MOMP following. In this case, Bcl-2 pro- survival members block apoptosis by sequestering BH3 domain-only molecules in stable complexes, located within the mitochondria, thus preventing the BH3-only molecules from interacting with, and activating Bax and Bak (Cheng et al., 2001). Lastly, the “direct activation of Bax and Bak” model proposes that Bax and Bak undergo conformational changes and become active, by the action of the BH-3 only proteins Bim and Bak, In addition to activation by Bim and Bid, there is some evidence to suggest that Bax/Bak may be activated by PUMA (Kim et al., 2006).  1.2 PI3-K Pathway  The phosphatidylinositol-3 kinases are a group of phosphoinositide kinases - heterodimeric proteins, each containing a catalytic subunit of 110-120 kDa, and a regulatory subunit of molecular mass varying between 50 and 101 kDa (reviewed in (Fruman, Meyers, & Cantley, 1998)). The catalytic subunit functions as a lipid kinase, while the regulatory subunit functions as an adaptor protein that allows appropriate associations that lead to activation of the kinase subunit. Three classes of PI3-Ks have been described, however, when considering    24 cytokine or mitogen signaling, the Class I PI3-Ks are the main group responsible for transducing these signals (Vanhaesebroeck & Waterfield, 1999), and this brief discussion will focus on this sub-class of the PI3-K family of enzymes.  Phosphatidylinositol (PI) is a cellular acidic phospholipid. This diacylglycerophospholipid has an inositol ring on the third position of the glycerol backbone, linked through a phosphodiester bond (Lehninger, Nelson, & Cox, 1993). PI3-K phosphorylates the free 3- position on the inositol ring of PI-4-phosphate, or PI-4,5-bisphosphate to generate PI-3,4- bisphosphate (PI-3,4-P2) or PI-3,4,5-trisphosphate (PI-3,4,5-P3) respectively. Another important, and perhaps the most physiological, means of generating PI-3,4-P2 is by dephosphorylation of PI-3,4,5-P3 (the main in vivo product of the class I PI 3-kinases), by 5- phosphatases, which will be discussed below. PI-3,4-P2 and PI-3,4,5-P3 are important signaling molecules, being critically involved in cell cycle progression, protein synthesis, cell survival and proliferation, actin polymerization and more (reviewed in (Vanhaesebroeck et al., 2001)). As one of the major signaling pathways involved in survival signaling, and as one of the major pathways through which GM-CSF is known to signal, the PI3-K pathway warrants discussion in more detail.  1.2.1 Activation of PI3-K  The binding of cytokines to their respective transmembrane receptors in many cases causes the receptor '-chain to heterodimerize with the common (-chain. The cytokines interleukin-3 (IL-3), interleukin-5 (IL-5) and granulocyte-macrophage colony stimulating factor (GM- CSF) provide classic examples of this signaling process (reviewed in (Martinez-    25 Moczygemba & Huston, 2003b)). Activation of PI3-K causes a strong survival signal. Activting mutations of PI3-K are responsible for disease. For example, constitutive activation of PI3-K is detectable in 50% of acute myeloid leukemias (S. Park et al., 2009). The PI3-K pathway is one of three major pathways stimulated by cytokine receptors (the other two are the JAK/STAT pathway and the MAPK pathway). The p110 subunit of Class IA PI3-Ks is found associated in a complex with an adaptor protein (Vanhaesebroeck & Waterfield, 1999) that contains two Src-homology-2 (SH-2) domains (Sadowski, Stone, & Pawson, 1986). The SH-2 domains bind to phosphorylated tyrosine residues in a specific amino acid milieu, recognizing between three and six specific amino acid residues C- terminal to the phosphorylated tyrosine (Songyang et al., 1993). Where tyrosine phosphorylation has occurred by the activation of tyrosine kinase receptors, the interaction of the SH-2-containing adaptor protein brings the p110 subunit of PI3K into approximation with the inner leaflet of the cell bilayer lipid membrane. The phosphatidylinositol lipid substrates of PI3-K are then in close physical proximity for the enzyme to modify. 1.2.2 Deactivation of PI3-K  While phosphorylation is the “on” switch to activate a signaling pathway, there also exists a corresponding “off” switch, to deactivate the pathway and turn off the signal. In the case of PI3-K, the only major phosphatase that exists to directly reverse and turn the signal off is PTEN (phosphatase and tensin homologue) (J. Li et al., 1997; Steck et al., 1997). However, there are other important 5-phosphataes (Astle, Horan, Ooms, & Mitchell, 2007; Astle et al., 2006).     26 PTEN, a phosphoinositide 3-phosphatase, is a tumour suppressor protein that hydrolyzes PI- 3,4,5-P3 to PI-4,5-P2, and so acts in direct opposition to the function of PI3-K. The ubiquitously expressed PTEN can promote cell cycle arrest, induce apoptosis, inhibit cell motility, and inhibit angiogenesis when its activity is elevated (Chow & Baker, 2006; Engelman, Luo, & Cantley, 2006; Sansal & Sellers, 2004). On the other hand, loss of PTEN activity allows PI3-K activation to continue unchecked. As a result, almost 50% of human cancers have been found to contain inactivated PTEN (Cantley & Neel, 1999), contributing to overactive downstream pathways controlled by PI3-K.  SHIP, a phosphoinositide 5-phosphatase, is found only in hematopoietic cells, and functions to hydrolyze PI-3,4,5-P3 to PI-3,4-P2 (Damen et al., 1996). SHIP is included in the discussion here because of its role in decreasing the levels of the PI3-K product, PI-3,4,5-P3. SHIP functions in mast cells, controlling the level of degranulation, cytokine production, and adhesion (Sly, Rauh, Kalesnikoff, Buchse, & Krystal, 2003). There is some evidence that SHIP may function as a tumour suppressor in hematopoietic progenitor cells (J. M. Luo et al., 2003; Wisniewski et al., 1999). However, it is worth remembering that since SHIP generates PI-3,4-P2, which may act as a second messenger in some cells, by its ability to attract PH-containing proteins (A. J. Marshall, Krahn, Ma, Duronio, & Hou, 2002; Scheid et al., 2002), the evidence for SHIP functioning as a tumor suppressor is controversial. 1.2.3 Downstream of PI3-K  The lipid products of PI3-K, phosphatidylinositol-3,4-bisphosphate and phosphatidylinositol- 3,4,5-trisphosphate, contain binding sites for molecules containing one of two distinct lipid binding domains – the Pleckstrin homology (PH) domain (Gray, Van Der Kaay, & Downes,    27 1999; Haslam, Koide, & Hemmings, 1993; Ingley & Hemmings, 1994), or the FYVE zinc finger domain (Stenmark, Aasland, Toh, & D'Arrigo, 1996). These binding domains allow specific interactions between the phosphatidylinositol lipids and downstream signaling molecules. PI3-K signaling is held to be one of the most important pathways for cell survival and proliferation (reviewed in (B. D. Manning & Cantley, 2007; Toker & Cantley, 1997)).  The importance of phosphatidylinositol-3 kinase signaling in cell survival was discovered in 1995, and published in two important works. Yao et al (R. Yao & Cooper, 1995) determined that nerve growth factor (NGF) could prevent apoptosis in a rat cell-line, but not if small molecule inhibitors of PI3-K, specifically wortmannin (Ui, Okada, Hazeki, & Hazeki, 1995) and LY294002 (Vlahos, Matter, Hui, & Brown, 1994), were included. At the same time, Duronio’s group discovered that PI3-K signaling is important for the inhibition of apoptosis in hematopoietic cells (Scheid, Lauener, & Duronio, 1995). Specifically, the actions of cytokines that are critical for cell survival were shown to require the activity of PI3-K. Duronio’s group also discovered that because of cross-talk between signaling pathways, downstream signaling through PI3-K differs depending on the cytokine used to activate the kinase (Scheid et al., 1995). PI3-K signaling for survival has subsequently been extensively studied and is known to be mediated via several signaling pathways (Downward, 2004; Duronio, 2008; McCubrey et al., 2008; Song, Ouyang, & Bao, 2005; Toker & Cantley, 1997).  One of the key events following the generation of PI-3,4-P2 and PI-3,4,5-P3 is  the activation of PDK1 (3-phosphoinositide dependent kinase 1) (Alessi et al., 1997), which contains a PH    28 domain, and is therefore bound to the plasma membrane in areas where PI3-K activation has occurred. The putative autophosphorylation site on PDK1 is phosphorylated by PDK1 itself, probably through both cis- and trans- mechanisms (Wick et al., 2003), and PDK1 also phosphorylates most of the kinases of the AGC family – including PKB, several protein kinase C (PKC) isoforms p70 S6 kinase, RSK, SGK as well as itself (reviewed in (Bayascas, 2008)).  Critically important to PI3-K survival signaling is the activation of PKB/Akt - the mammalian homologue of v-akt (Bellacosa, Testa, Staal, & Tsichlis, 1991; Coffer & Woodgett, 1991; Jones, Jakubowicz, Pitossi, Maurer, & Hemmings, 1991), the N-terminus of which contains a PH domain (Ferguson et al., 2000; Lietzke et al., 2000). PKB is recruited via its PH domain to the plasma membrane along with PDK1, where activation of PKB takes place (Franke et al., 1995). PKB is a key mediator of PI3-K-dependent cell survival (Downward, 2004; Lawlor & Alessi, 2001; Marte & Downward, 1997; Song et al., 2005) and is activated by phosphorylation by PDK-1 at Thr308. A second site of phosphorylation in the hydrophobic domain (also common to all AGC family members) is Ser473. Phosphorylation at that site is mediated by one or more kinases, which are still not completely identified in terms of their upstream regulation (for further discussion, see (Duronio, 2008)). PKB also functions in other important cellular roles, such as growth (i.e. protein synthesis), proliferation, angiogenesis, metabolism and migration (reviewed in (Engelman et al., 2006; B. D. Manning & Cantley, 2007)).     29 As well as activating PDK-1 and PKB/Akt signaling, the lipid products of PI3-K activity are important for activating the following kinases - PLC-) (Falasca et al., 1998), the Tec kinases Btk, Tec and Itk (see, e.g. (Lindvall et al., 2005)), the Rac GEFs, e.g. Vav-1 (Reynolds et al., 2002), and cdc42 (reviewed in (Ward, 2004).  Activation of PI3-K/PKB also causes the inhibition of BAD, BAX, FOXO-3a, TSC1/2 and GSK-3, resulting in inhibition of pro-apoptotic activity and increased protein translation, and repression of transcriptional events that can contribute to apoptosis (reviewed in (Duronio, 2008)).  The regulation of protein synthesis by PI3-K/PKB via TOR has been extensively reviewed (see, e.g. (Duronio, 2008; Franke, 2008; Ruggero & Sonenberg, 2005; Wymann & Marone, 2005)). Being such an important element in the control of protein synthesis, the PI3-K pathway is centrally positioned to form a critical regulator of the cellular proteome. The outcome of the activating and inhibiting effects of the lipid products of PI3-K is to determine the balance between the pro-survival and pro-apoptotic forces within the cell. 1.2.4 Dysregulation of PI3-K/PKB  Dysregulation of PI3-K/PKB signaling has been implicated in diverse human diseases ranging from cancer and cardiovascular disease to diabetes, acute leukemia, non-Hodgkin’s lymphoma and more (Chang et al., 2003; Fresno Vara et al., 2004; Oudit et al., 2004). It is well recognized that in most diseases in which PI3-K/PKB is disrupted, it is the alteration in the survival/apoptosis signaling by PI3-K/PKB that is most important. Such is the abundance of evidence linking dysregulation of the PI3-K/PKB pathway “by various genetic and    30 epigenetic mechanisms in a wide range of tumor types [that] there is now extensive evidence validating various components of this pathway as molecular targets in cancer.” (Yap et al., 2008)  With evidence of the far-reaching effects of PI3-K signaling, effects on cell survival, proliferation, metabolism, growth, angiogenesis and cytoskeletal structure and function, as well as the importance of the PI3-K signaling pathway as a regulator of protein synthesis, it is apparent that there is much potential for fruitful investigation of this pathway. 1.2.5 GM-CSF Signalling  Granulocyte macrophage colony stimulating factor (GM-CSF) was originally discovered for its ability to generate granulocytes and macrophages from progenitor cells in mouse bone marrow (Burgess, Camakaris, & Metcalf, 1977). However, GM-CSF was soon found to function in mature cells, such as granulocytes, macrophages and eosinophils. GM-CSF is understood to be a major regulator of cell number for cells of the granulocyte and macrophage lineage and to affect the level of immune cell activation (Hamilton & Anderson, 2004). That is, GM-CSF functions in developing the tissue inflammatory reaction and in host protection, hence GM-CSF appears to be involved in inducible hematopoiesis in response to infection, and also in allergic and inflammatory reactions. It has been determined that, in the presence of GM-CSF, circulating neutrophil half-life is approximately 10 hours in both the normal and activated states (Dale, Liles, Llewellyn, & Price, 1998).  GM-CSF is secreted by macrophages, activated T-cells, mast cells, endothelial cells, bone marrow stromal cells and basophils (Martinez-Moczygemba & Huston, 2003a), and GM-    31 CSF signals through a unique '-receptor, which heterodimerizes with a common beta chain (shared with IL-3 and IL-5) (Martinez-Moczygemba & Huston, 2003a).  Apart from signaling via the PI3-K pathway (described above), the binding of GM-CSF to its cognate receptor causes signalling via the extracellular signal-regulated kinase (ERK), and the Janus Kinase (Jak)/signal transducers and activators of transcription (STAT) pathways. These pathways will be briefly reviewed. 1.2.5.1 GM-CSF Signaling via Erk  The mitogen-activated protein kinase (MAP Kinase) family consists of six sub-groups – ERK1/2, JNKs, p38 isoforms, ERK5, ERK3/4 and ERK 7/8. The subgroup containing the kinases ERK1 and ERK2 has been shown to be activated by IL-3, IL-5 and GM-CSF signaling, and to result in survival signaling (i.e. inhibition of apoptosis) and cell cycle progression (Crews, Alessandrini, & Erikson, 1992; Egan & Weinberg, 1993; McCubrey, May, Duronio, & Mufson, 2000; Weinstein-Oppenheimer, Blalock, Steelman, Chang, & McCubrey, 2000).  Binding of GM-CSF to the GM-CSF receptor causes the inactive ras-GDP complex to become active ras-GTP. Ras-GTP activates raf, which then signals via MEK to ERK. Erk targets transcription factors such as NF-$B, AP-1, c-Myc and Ets-1. However this is by no means an exhaustive list – a recent review lists 160 well-documented substrates of Erk1/2 (Yoon & Seger, 2006).     32 As well as causing the activation of ERK1/2, there is evidence that p38 may be activated by GM-CSF in human neutrophils (Nahas, Molski, Fernandez, & Sha'afi, 1996). As well, there are reports of activation of JNK/SAPK in some cells by IL-3/IL-5/GM-CSF (de Groot et al., 1997; Foltz & Schrader, 1997; Terada, Kaziro, & Satoh, 1997).  1.2.5.2 GM-CSF Signaling via JAK/STAT  The Janus kinases (JAKs) are a family of four cytosolic tyrosine kinases (JAK1, JAK2, JAK3, and Tyk2)  that bind to cytokine receptors and play an important role in cytokine signaling (Darnell, 1997; Schindler & Darnell, 1995; Watanabe, Itoh, & Arai, 1996). Inactive JAKs bind to the cytoplasmic tail of receptors for cytokines such as IL-3, IL-5 and GM-CSF. Upon activation of the receptor, two JAKs trans-phosphorylate each other to become active. Activated JAKs phosphorylate and activate several substrates, one of which is the cytokine receptor itself (Darnell, 1997; Schindler & Darnell, 1995; Watanabe et al., 1996). Phosphorylated tyrosine residues on the receptor form a docking site for molecules containing an SH-2 domain. One class of such molecules is the family of transcription factors known as STATs (signal transducers and activators of transcription)(Darnell, 1997; Schindler & Darnell, 1995). The binding of ligand to receptors for IL-3, IL-5 and GM-CSF activates JAK2. GM-CSF activated JAK2 forms a binding site for (predominantly) STAT5. Binding of a STAT5 monomer to JAK2 results in phosphorylation of the STAT5 monomers, which dimerize, and translocate to the nucleus. Once within the nucleus, dimerized STAT5 plays a role in cell proliferation, via such gene targets as c-fos, pim-1, osm and cis (Basham et al., 2008), which is important in the context of the work described here.     33 The TF-1 cells used in many of the studies reported here are dependent on a continuous supply of IL-3, GM-CSF, or EPO for survival (see discussion below)(Kitamura et al., 1989). Withdrawal of the cytokine results in the cell colonies undergoing apoptosis within a period of approximately 18-30 hours. Of importance for these studies, the addition of GM-CSF to hemopoietic cells such as the TF-1 cell line, results in the activation of  the PI3-K pathway, some JAK/STAT pathways, and some MAPK pathways, which promote cell survival and growth.  1.3 Early Apoptosis or “Pre-Apoptosis” The primary goal of this study was to examine changes in protein concentrations in early apoptosis, so-called “pre-apoptosis”; that is, after the cells have committed to apoptosis, but before the cells have begun to be dismantled by the process. Analysis of the observed concentration changes could then lead to a new understanding of early apoptosis. The induction of apoptosis was brought about by the withdrawal of mitogens, in this case GM- CSF, required by the cells for survival and growth. Under these conditions, the cells are in the presence of the full complement of nutrients and growth factors derived from serum, and yet we know that these cells will undergo apoptosis if specific cytokines are not present. Perhaps this loss of cytokine stimulation induces a “stress” response in the cell? If so, this may be because the cellular “wiring” is set to require a specific complement of signals to be triggered, and if these signals are not present, the threshold of the Bcl-2 family proteins is set close to the point where apoptosis will then be initiated. Another possibility needs to be considered. Under conditions of cytokine withdrawal, it is possible that the cells might    34 initially develop an autophagic response, in an attempt to survive. Autophagy, also called type II programmed cell death, is a highly conserved cellular mechanism which involves the “recycling” of long-lived proteins and cellular organelles, so that the components may be used for basal synthesis of proteins and other molecules, and the generation of energy under conditions of stress (Ferraro & Cecconi, 2007). Autophagy is, under most circumstances, a survival process; activated in response to stress condtions. However, autophagy may also play a role in cell death; where cellular damage is too extensive for survival, or if apoptosis is compromised, autophagy may function to kill the cell (Klionsky, 2007; Thorburn, 2008). While there is evidence that nutritional deprivation brings about autophagy, at least in some situations, the question of whether or not cytokine withdrawal will cause the same response as nutrient deprivation has not been addressed. A study by Li, et. al. was able to provide some evidence that cell death caused by growth-factor withdrawal (in this case, withdrawal of both fetal calf serum and IL-2 from the murine T-cell line, D10) may be autophagic (C. Li et al., 2006). Lum and co-workers describe a study of Bax/Bak double knock-out mice in which immortalized IL-3 dependent cell lines were generated from the bone marrow.  IL-3 was withdrawn and the cells were able to survive for several weeks using autophagy of organelles to provide a source of ATP and nutrients (Lum et al., 2005). In this case, autophagy allowed cell survival. Transfection of either Bax or Bak into the cells restored the apoptosis response to IL-3 withdrawal. (Lum et al., 2005). Bearing this in mind, we must allow for the possibility that some of the concentration changes observed in response to cytokine withdrawal might be due to an autophagic response to cytokine withdrawal. Determining the presence of autophagy in response to cytokine withdrawal may be attempted in two ways – morphologically and molecularly. Morphological evidence is usually obtained    35 by using electron microscopy. The presence of autophagic vesicles within cells, as well as the absence of cell engulfment through phagocytes, are considered hallmarks of autophagy (at least during the early stages of cell death) (Conradt, 2009). Molecularly, the use of an LC3-based assay (mammalian Atg8 homologue) to observe flux in the levels of LC3 following cytokine withdrawal is considered standard (Klionsky, 2007).  1.3.1 Protein Concentration Changes versus PTMs  The major means of transmitting information via cell signaling networks is through post- translational modifications (PTMs) of proteins (see, e.g. (Hunter, 1995; Wold, 1981)). It has been estimated that as many as three hundred post-translational modifications of proteins occur physiologically (Witze, Old, Resing, & Ahn, 2007). The most common PTM is protein phosphorylation, which affects about one-third of all proteins, and is the most widely studied PTM (Cohen, 2001).  Apart from phosphorylation, there are at least twenty other forms of post-translational modification commonly seen in proteins, and many more possible modifications, as noted above. Modifications such as methylation, acetylation, glycosylation and lipid modifications such as farnesylation and gernaylgeranylation are frequently encountered, and function to regulate the biological activity of the protein. From our understanding of the importance of PTM in cell signaling, it is reasonable to assume that many of the changes involved in signaling for apoptosis will occur via modifications of proteins already present in the cell. This is not to understate the importance of changes in gene expression levels that will occur as a result of activation and repression of various signaling pathways during early apoptosis., leading to changes in the concentrations of proteins. With this in mind the present study provides interesting results, since I have clearly    36 shown changes in the concentrations of many proteins in response to cytokine withdrawal – that is, that the process of inducing apoptosis causes both increases and decreases in the concentrations of proteins. These changes have occurred in addition to the post-translational modifications expected, and analysis of these changes in protein concentrations adds to our understanding of apoptosis. 1.4 Proteomics Proteomics is perhaps the most important approach to the study of cellular function in post- genomic biology. The term was first coined in 1994 by Marc Wilkins (Wilkins, 1994) and the first study utilizing the term “proteomics” was published in 1995 (Wasinger et al., 1995). Some of the techniques have existed since much earlier – for example 2-D gel electrophoresis (Margolis & Kenrick, 1969; O'Farrell, 1975); and mass spectrometry was first developed by J.J. Thomson in the early Twentieth century (Thomson, 1913). However, it was not until the 1990s that “biological” mass spectrometry began to emerge as a powerful tool for the analysis of complex biological samples (Pandey & Mann, 2000).  The proteome was defined by Wilkins et. al. as the proteins expressed by the genome of an organism, or, in multi-cellular organisms, as the protein complement expressed by a tissue or differentiated cell (quoted in (Haynes, Gygi, Figeys, & Aebersold, 1998). Proteomics involves a study of the totality of proteins present within an organism or a tissue or differentiated cells, under experimental and control conditions, and has been defined as “the use of quantitative protein-level measurements of gene expression to characterize biological processes (e.g., disease processes and drug effects) and decipher the mechanisms    37 of gene expression control” (Anderson & Anderson, 1998). While the early definitions included the term “quantitative”, early proteomics was a field of investigation which sought to define the proteome, and as such, was originally mostly descriptive – presenting a catalogue of the proteins identified in cells under different conditions. Indeed the term “descriptive proteomics” is frequently seen in the literature. With the advent of new technologies, and the advancement of existing technologies, the field has progressed, and now seeks quantitative data, despite the enormous technical challenges in such a goal (Bantscheff, Schirle, Sweetman, Rick, & Kuster, 2007). Ong and Mann summarize the potential of proteomics by observing that “proteomics directly addresses the level of gene products present in a given cell state and can further characterize protein activities, interactions and subcellular distributions.” (Ong & Mann, 2005). Two-dimensional gel electrophoresis and mass spectrometry are both valuable tools for proteome profiling. Each has strengths and weaknesses, and there are some in the field who argue that the techniques are complimentary, and should be used together to seek the answer to certain questions. However, the work described in this thesis was carried out using mass spectrometry, and so this will be the focus of the ensuing discussion. 1.4.1 Mass Spectrometry – Brief Historical Overview  J.J. Thomson designed and built what is considered the first mass spectrometer - an instrument capable of separating the ionized components of a mixture of gases, according to the mass of the ions, thereby allowing determination of the mass-charge ratios of the ions (Thomson, 1913). Thomson’s instrument was refined by A.J. Dempster, who built the first modern mass spectrometer in 1913 (Dempster, 1918). The British chemist and physicist F.W. Aston, beginning in 1919, further developed mass spectrometry by refining the magnetic    38 field, and the exit grate, and so was able to examine the isotopic composition of over fifty elements (Aston, 1919). The field of mass spectrometry progressed with the development of the quadrupole mass spectrometer (Wolfgang Paul and Helmut Steinwedel, 1953) and the ion trap (Wolfgang Paul and Hans G. Dehmelt). However it was not until the development of soft ionization techniques in the late 1980s – in particular, electrospray ionization (ESI) and matrix-assisted laser desorption and ionization (MALDI) (Fenn, Mann, Meng, Wong, & Whitehouse, 1989; Karas & Hillenkamp, 1988), that it became possible to analyze very complex protein mixtures using mass spectrometry (Siuzdak, 1994). Even with these technical advances, it wasn’t until the completion of the genome sequencing projects (McPherson et al., 2001; Venter et al., 2001) that it became possible to adopt a high- throughput “shotgun” approach to the field, and thereby to gather information on the products of large portions of the genome from a single experiment. This shotgun approach allows for the visualization of a much greater portion of the proteome at one time, and hence for the large scale examination of the perturbations of the proteome induced by various experimental conditions, allowing for the function of genes and cells to be determined directly at the protein level (Aebersold et al. 2003). The field of proteomics continues to develop, with the emphasis now being on improving techniques to encompass more of the proteome, and to improve protein quantitation.  1.4.2 SILAC  Being able to determine quantitative changes in the concentrations of proteins in cells under various conditions is a major goal in systems biology. In our work, having a clear identification of proteins that exhibit concentration changes under conditions of cytokine    39 withdrawal provides information on cellular function at the level of gene products, rather than post-translational modification. Such information will provide a clearer understanding of the changes at the level of gene expression induced by the signaling pathways involved in apoptosis and, at the same time, provide information on potential targets for drug development, or potential targets for different uses of already existing drugs. The ultimate goal is to provide better means of dealing with the diseases of inappropriate apoptosis.  For several reasons, it is not possible to quantitate peptides based on the intensity of MS signals - different peptides ionize with different efficiencies (Gygi & Aebersold, 2000), some peptides are too small, or too large to be analyzed by mass spectrometry, and ion suppression can cause apparent differences in the strength of peptide signals, unrelated to the amount of peptide present (Jessome & Volmer, 2006). The size, charge and hydrophobicity of peptides lead to different behaviour in the mass spectrometer, and so to different signal strengths (Bantscheff et al., 2007). Furthermore, peptides may behave differently at different stages of sample preparation, leading to an accumulation of nonsystematic errors (Ong & Mann, 2005). Hence, quantitative proteomics approaches are based on techniques other than the comparison of peptide signal intensities.  The use of known quantities of synthetic internal standards mixed with the sample of interest to achieve absolute quantitation using mass spectrometry has been used since the early 1980s. The technique is becoming more broadly applied, and in proteomics is known as AQUA (absolute quantitation of proteins) (Gerber, Rush, Stemman, Kirschner, & Gygi,    40 2003). One disadvantage concerning the use of AQUA is the need for a priori knowledge of the protein to be quantitated.  Relative quantitation is currently used in all cases where quantitation involves unknown proteins. Current techniques for the relative quantitation of proteins may be broadly divided into two classes – labelling and label-free. Labelling may be subdivided into either metabolic labelling (i.e. in vivo) or chemical labelling (in vitro) (reviewed in (Ong & Mann, 2005)). Examples of metabolic labelling include incorporation of 15 N (Wu, MacCoss, Howell, Matthews, & Yates, 2004), and SILAC (see discussion below).  Post-harvest (chemical) labelling techniques include ICAT (Gygi et al., 1999), iTRAQ TM  (Ross et al., 2004) and 18 O labelling (Mirgorodskaya et al., 2000; X. Yao, Freas, Ramirez, Demirev, & Fenselau, 2001).  Metabolic labelling techniques have the advantage of being able to label almost all proteins in a cell, however the techniques cannot be applied to tissue samples. Post-harvest labelling can be applied, in theory, to any sample. The disadvantage of post-harvest techniques is that not all peptides will be labelled.  Label-free techniques include peptide spectral counting (H. Liu, Sadygov, & Yates, 2004; Old et al., 2005; Zybailov, Coleman, Florens, & Washburn, 2005) in which the number of spectra identifying a protein are counted. For simple mixtures, there is a near linear relationship over two orders of magnitude between the number of copies of a spectrum and    41 the relative abundance of the protein identified by the spectrum (H. Liu et al., 2004).  A second label-free technique, utilizes a calculated “protein abundance index” (PAI). PAIs are based on the observed fact that as the amount of protein increases so does the number of peptides detected. Since a large protein will generate more peptides than a small protein, the scores need to be normalized (Rappsilber, Ryder, Lamond, & Mann, 2002).  Stable Isotope Labelling by Amino Acids in Cell Culture (SILAC) was developed by Mann’s group as a means of characterizing changes in the concentrations of proteins present in cells under different conditions, so called “expression proteomics” (Ong et al., 2002). The technique involves growing cells in culture containing amino acids synthesized using stable isotopes of biologically common atoms – usually carbon-13 or nitrogen-15. After six doublings of the cell population, greater than 98% of the protein content of the cells will contain labelled amino acid residues – that is less than two percent of the proteins will be original (unlabelled) material. Concurrently with stable isotope labelling, cells are grown in media formulated with “wild-type”, i.e. 12 C- or 14 N-containing amino acids. Amino acids chosen for labelling are usually arginine and lysine – selected because trypsin, which is commonly used for the generation of peptides from proteins for mass spectrometry analysis, cleaves at the C-terminus side of an arginine or lysine reside (Olsen, Ong, & Mann, 2004). After labelling has occurred, “experimental” cells are perturbed in some way, and compared with “control” cells. After the intervention, cells are lysed, the protein concentration determined and lysates mixed in a 1:1 ratio based on protein concentration. The lysates can then be fractionated either by chromatography or by SDS-PAGE and the fractions    42 trypsinized, prior to mass spectrometry analysis. It is the analysis of SILAC data that will form a major part of this study.  1.4.2.1 Determining the Accuracy of Protein Quantitation  The protein ratio (“control” versus “experimental”) is determined by averaging the individual peptide ratios – matching the intensities of pairs of heavy- and light-labelled occurrences of the same peptides. Under optimal conditions, very high quantitation accuracies may be achieved – in fact, it is possible to achieve accuracies which are greater than the variability introduced by biological or sample preparation factors. Under ideal conditions the accuracy of quantitation involves an error of less than five percent (Ong, Kratchmarova, & Mann, 2003).  Accuracy is affected by the following factors:  a) Low abundance peptides may appear in only one or two consecutive  spectra, therefore the average peptide ratios will be less precise  b) If a protein is identified with only a few peptides, the protein quantitation may be less accurate if one peptide quantitation is of poor quality  c) The co-elution of two or more peptides in a complex mixture may lead to overlap of the isotope clusters identified by the mass spectrometer. This  overlap leads to errors in quantitation  d) The signal-to-noise ratio for each of the two peaks used in the comparison is important. If both peaks have a high signal-to-noise ratio, there would not usually be a problem with the quantitation calculated from data, however, if    43 one peak is small, the contribution of background noise to peak area (or height) can be significant. This may create a ratio of heavy to light peaks that was lower than the correct value. At the other extreme, an abundant peak might cause overload of the detector, and so the true value of the intensity  might be under-reported, leading to a reduction in the value of the reported ratio compared with the true value.  e) In some cell lines, arginine is converted to proline, and so 13 C-Arg becomes 13 C-Pro. In this case, the 13 C-Pro must be counted with the 13 C-Arg to  accurately quantitate the two cell states in the experiment. In summary, the accuracy of protein quantitation depends on:  (i) The number of peptides used  (ii) The accuracy of the data for each peptide (especially if there are a low  number of peptides) (iii) Isotopic peak clusters standing alone, that is, no peak overlaps from co- eluting peptides  (iv) Good signal-to-noise ratio (in practice, better than 3:1)  1.4.3 Mass Spectrometry – Fourier Transform – Ion Cyclotron Resonance  Fourier-transform ion cyclotron resonance mass spectrometry (FT-ICR MS) is the most sensitive mass spectrometry method currently used (A.G. Marshall, 2000). FT-ICR developed from ion cyclotron resonance (ICR) mass spectrometry, which was first described in the late 1940s and early 1950s (Sommer, Thomas, & Hipple, 1951). ICR mass spectrometry technique was regarded with mainly academic interest until Allan G. Marshall    44 and M. Comisarow at the University of British Columbia, Vancouver, applied the Fourier transform (Cooley & Tukey, 1965) methods that had previously been used successfully with nuclear magnetic resonance spectroscopy (Comisarow & Marshall, 1974). The ion detection capabilities of FT-ICR allow mass resolution of greater than 0.003 Da (A G Marshall & Hendrickson, 2002), with sensitivity down to attomole quantities (Quenzer, Emmett, Hendrickson, Kelly, & Marshall, 2001). Coupled with on-line chromatography, this technique allows the analysis of complex protein samples and as such, is particularly suited to experiments involving analysis of whole cell lysates. The analysis problem caused by differences in protein abundance in a sample can be at least partially addressed by using an instrument with the resolving power and sensitivity of the FT-ICR. Indeed “FT-ICR MS offers 10 –100 times higher mass resolution, mass resolving power, and mass accuracy than any other mass analysis technique.” (A.G. Marshall, 2000)  In an “on-line” experiment, such as is necessary for the analysis of a whole cell lysate, current mass spectrometers are limited to the analysis of the peptides with the strongest signals present in the instrument at one time. This constraint is the result of the continuous in-flow of sample into the instrument, and the time taken for a mass spectrum to be generated. Because there are more of them, peptides of an abundant protein will frequently give a much stronger signal than peptides from a low-abundance protein. It has been determined that protein expression covers greater than ten orders of magnitude in human serum (Patterson & Aebersold, 2003b). While this figure is reported to be lower in a whole cell lysate (perhaps “only” 7-8 orders of magnitude in human cells (Anderson & Anderson, 1998)), many of the signaling proteins in cells are present in lower copy numbers, and it is    45 the signaling proteins, rather than the structural or other ”housekeeping” proteins, which will be of interest in an experiment in which the proteins that play key regulatory roles in apoptosis are the focus. 1.4.3.1 Fourier Transform Ion Cyclotron Resonance Mass Spectrometry – A Brief Overview  An excellent primer on technological aspects of FT-ICR, was published by Marshall and Hendricksen in 1998 (A. G. Marshall, Hendrickson, & Jackson, 1998a). The FT-ICR consists of an analyzer (cyclotron) cell (also called a Penning trap), surrounded by a superconducting magnet that generates a spatially uniform static magnetic field. The Penning trap is maintained under vacuum to allow ions introduced into the cell to orbit (typically more than 100,000 times) without collision with gas molecules. The Penning trap has two sets of opposing plates (electrodes) arranged perpendicularly at the perimeter – one set is used to excite the ions introduced into the trap, and the other set detects the tiny current generated by the charged ions moving in a static magnetic field. The magnetic field strength determines instrument sensitivity, resolution and mass accuracy; and determining efficient ways to increase the magnetic field strength drives development in this field. The largest FT- ICR magnet currently in existence is at the National High Magnetic Field Laboratory in Florida, and has a field strength of 14.5 Tesla.  Ions enter the spatially uniform, static magnetic field generated by the superconducting magnet. As noted above, the charged particles moving in a static magnetic field generate an electric field, of frequency dependent on the mass and charge of the charged particle. This so-called cyclotron frequency is also dependent on the strength of the magntic field – as the    46 magnetic field strength increases, so does the cyclotron frequency for any charged particle. Further, as the cyclotron frequency increases, so does the difference between any two ICR frequencies, making detection of mass differences easier. As the ions enter the Penning trap, they are guided into a circular orbit, and the ion orbital motion is compressed or restrained to become spatially coherent (that is, to orbit in phase) by applying a spatially uniform radio frequency (RF) electric field at the same frequency as the ion orbital frequency. This RF field is applied via the excitation electrodes. Orbiting ions pass the two opposing parallel electrodes (detector plates) and induce an oscillating signal - an electric field in sine wave form - with frequency dependent on the mass and charge of the circulating ions. The generated electric field can be detected and analyzed. The technique is non-destructive, as the ions are trapped and can be re-analyzed, allowing for increased sensitivity and resolution. The frequency of the ion signal (the generated electric field sine wave) is a function of the ion mass and charge state, and the magnetic field strength. Where all the ions are of a single mass and all carry the same charge, a single frequency will be generated. In this case, it is trivial to convert the single frequency to mass units. However, where the circulating ions are of different masses, or carry different charges, a much more complex electric field is generated. Each ion produces its own sine wave electric field, with characteristic frequency and amplitude, and each individual sine wave becomes superimposed (linear superposition) to produce a complex waveform. In order to reduce this complex waveform back into its component simple sine waves, a Fourier transform is applied to the oscillating signal. The Fourier transform generates a frequency-domain spectrum, which can then be converted to a mass-domain spectrum. (See Figure 1.3)     47 It is because of the use of frequency to determine mass, rather than the use of time (in a TOF mass analyzer, for example) that FT-ICR is capable of such high resolution – it is possible to measure frequency much more accurately than to measure time.   Figure 1.3 FT-ICR mass spectrometry. (a) Schematic representation of excited ion cyclotron rotation, (b) time- domain image-current signal from opposed detection electrodes, (c) frequency-domain spectrum obtained by fast Fourier transform of the digitized time-domain signal, and (d) Fourier transform-ion cyclotron resonance m/z spectrum obtained by calibrated frequency-to-m/z conversion. A full-range mass spectrum (including computation) is typically generated in ~1 s. Reprinted, with permission, from the Annual Review of Analytical Chemistry, Volume 1 © 2008 by Annual Reviews   www.annualreviews.org.      48 1.4.3.2 Key Mass Spectrometer Parameters for Proteomics  The following list outlines the key mass spectrometry parameters of relevance to proteomics. i. Mass accuracy – the absolute deviation of measured mass from true mass (Balogh, 2004), can be reported in four ways – as Daltons (Da), as absolute milli-mass units (mmu), or in ppm or as a fraction expressed as a percent. Mass accuracy is determined by performing an internal calibration using a calibrant of known mass. Generally a mass accuracy of less than 10 ppm is desirable. FT-ICR MS is capable of mass accuracy at sub-parts-per-million (Alan G. Marshall & Hendrickson, 2008).  ii. Mass resolution – Marshall defines mass resolution as “the minimum mass difference at which two equal-height peaks may be base-line resolved” (A.G. Marshall, 2000). FT-ICR MS is capable of mass resolution to 0.000 45 Da (F. He, Hendrickson, & Marshall, 2001). This number is smaller than the mass of an electron (0.000 55 Da)!  Mass-resolving power, a related value, is defined as R = M/*M, where *M is determined by full peak width at half maximum height (FWHM), as above (A. G. Marshall et al., 1998a). Since it is the usual practice to define resolution as the full width of a peak at half the peak height, in this case *m is written as *m50% (A. G. Marshall, Hendrickson, & Jackson, 1998b). For an ion with m/z of 1000, and using a 3 s observation, FT-ICR is capable of mass-resolving power (m/*m50%) of  "400,000 (Alan G. Marshall & Hendrickson, 2008).    49 iii. Sensitivity – a measured value. Progressively smaller amounts of material are loaded into the mass spectrometer, until a point is reached where the material can no longer be detected. The FT-ICR is capable of achieving attomole sensitivity (S. E. Martin, Shabanowitz, Hunt, & Marto, 2000).  1.4.3.3 “Bottom-Up” versus “Top-Down” Approach  Proteomics experiments may be designated as either “bottom-up” or “top-down”. In a “bottom-up” approach, intact proteins are digested with trypsin, or another cleavage-specific protease, and the peptides are introduced into the mass spectrometer. Proteins are identified based on the peptides present using either  “peptide mass fingerprinting”, or  “tandem mass spectrometry”. The work described in this thesis utilizes the latter approach – MS/MS or tandem mass spectrometry – to infer the presence of various protein species based on the peptides identified in the mass spectrometer.  The “bottom-up” approach contrasts with the “top-down” approach, in which intact proteins are ionized and introduced into the mass spectrometer for analysis. The bottom-up approach has several advantages over the top-down approach for the analysis of complex mixtures of proteins: the identification of intact proteins from cell lysates is difficult (Mann, Hendrickson, & Pandey, 2001); sensitivity of electrospray ionization mass spectrometry is poorer for large molecules than it is for peptides, because the signal is “diluted” by being split over several charge states. Also, the heterogeneity of the protein (with various post- translational modifications) spreads the signal (Mann et al., 2001). The size limit for both electrospray and MALDI ionizations is about 100 kDa (Bakhtiar & Nelson, 2000). The    50 molecular weight of a protein cannot be predicted from its database entry (sequence), because of various post-translational modifications, chemical modifications (oxidation of methionine, for example), and N- and C-terminal processing (Mann et al., 2001). Hence for complex mixtures of proteins, the bottom-up approach is more practical and gives more reliable results.  1.4.3.4 Database Searching and de novo Sequencing (MS/MS)  With the enormous increase in the amount of data that it is possible to generate from a modern proteomics experiment, it has been essential to develop computational tools capable of handling large data sets. The generation of such large numbers of mass spectra requires an automated approach to analysis of the spectra. The automated analysis of uninterpreted fragment spectra may be achieved by utilizing one of two approaches – the spectra may be compared with spectra in an in silico digest of a protein sequence database, or alternatively, the peptide identity may be determined without reference to a database – so called “de novo” sequencing. These approaches will be discussed briefly below. a) Peptide identity may be determined without reference to a database – so called de novo sequencing. With the human genome project nearing completion, most, if not all protein sequences are available in sequence databases. However, for proteins whose sequence is not known, the only way to reach an identification is by using de novo sequencing. In the de novo sequencing approach, the experimental peptides masses are matched against a “universal” database that contains all linear amino acid combinations. Many software programs are available for performing de novo sequencing. For review see (Palagi, Hernandez, Walther, &    51 Appel, 2006; Xu & Ma, 2006). This approach was not necessary for the work presented in this thesis, since the human genome has been sequenced.  b) Peptides may be searched against a sequence database: Three approaches have been developed for the identification of proteins based on spectra searches against an in silico digest of a sequence database.  (i) Peptide sequence tags were developed by Mann’s group in 1994 (Mann & Wilm, 1994). The peptide sequence tag consists of a “short, easily identifiable series of sequence ions” from the fragmentation spectrum, plus the masses of the fragment on either side of the identifiable series. It was discovered that this information could be used as a “highly specific identifier of the peptide” (Mann & Wilm, 1994). Implementation of the algorithm was achieved using the software PeptideSearch (Mann, Hojrup, & Roepstorff, 1993) which correlates mass spectrometric and sequence (MS/MS) data with a sequence database, and calculates the likelihood that a match between the experimental and predicted peptides is a random event.  (ii) A second early approach is the cross-correlation method, developed in the Yates lab, and offered to the community as the SEQUEST program (Eng, McCormack, & JR Yates, 1994). Using this approach, the experimentally derived spectra are cross-correlated against predicted spectra. The quality of the correlation is determined, and the results ranked on the basis of this determination. (iii) A third approach is to use probability-based matching. The most commonly used probability-based software is MASCOT (Perkins, Pappin, Creasy, & Cottrell, 1999). In this approach, a probability score is derived, indicating the likelihood that a match between    52 the experimentally derived peptides and the sequence database-derived peptides is correct. In MASCOT, the probability scores are derived using a modified derivation of the “molecular weight search” MOWSE probability model (Pappin, Hojrup, & Bleasby, 1993), in which experimentally derived peptide masses are compared with the sequence database and the probability that a match between the two is a random event is computed. The match with the lowest probability of being random is reported as the best match (Perkins et al., 1999). The MASCOT implementation of the MOWSE probability scoring has been described for peptide mass fingerprinting, and it is generally held that the aforementioned is a correct description of the functioning of MASCOT for the interpretation of MS/MS spectra, but the details of exactly how MASCOT works for tandem mass spectrometry data have never been published.  Given that the above algorithms all utilize some form of probability scoring in determining the likelihood that the match between experimental and predicted peptides is a random event, there will occur, in the spread of scores, some overlap between positive (correct) identifications and false positive results. Under these circumstances, it is necessary to set a score threshold so that as many as possible of the correct identifications are included in the results, and a few as possible of the incorrect assignments are included.  One way in which the score thresholds may be set is by using a reversed database concatenated with the forward database, and searching both simultaneously (R. E. Moore, Young, & Lee, 2002). Extensions of this concept have arisen, by using a randomized database instead of a reversed one, or a randomized database with arginine and lysine in the    53 same relative positions. By using one of these approaches, the number of false positive matches may be minimized, by selecting threshold scores that keep the ratio of false to correct identifications at less than 5% (MacCoss, 2005).   1.5 The Approach  My approach to the questions posed in this thesis was to analyze proteins obtained from stable isotope labelling of cells, which were caused to undergo cytokine withdrawal, and the induction of apoptosis. This enabled us to study the effects of withdrawal of cytokine binding to cell surface receptors. The normally induced tyrosine-kinase phosphorylation, causing a cascade of further signaling events that regulate many biological processes, was interrupted. The effects of this interruption were to induce apoptosis, and as part of that process, to induce changes (both up and down) in the concentrations of many proteins in the cell. I have been able to visualize some of the effects of changes in the activation of various signaling pathways, especially the PI3-K, JAK/STAT and MAPK pathways. One advantage of this approach is that it requires minimal a priori knowledge of the changes induced by cytokine binding to receptor. I have been able to uncover a set of proteins associated with the inhibition of the pathways activated by GM-CSF. Future work will be required to test the function of this set of proteins by “targeted” follow-up experiments, using alternative approaches. In this work, I pose the following questions:    54 - what changes occur in cellular protein concentrations in response to cytokine  withdrawal-induced apoptosis? - can a global view of these protein concentration changes be interpreted in any  way that suggests a new understanding of the complex interactions occurring  between proteins as cells commit to undergoing apoptosis? - do any specific proteins appear in these data that have previously been  unrecognized as being involved in the PI3-K signaling pathway? - and finally, does the current technology allow adequate analysis of complex  protein mixtures to permit the above questions to be addressed in this way? The answers to these questions appear in the following pages, along with the supporting material used to discover the answers.  1.6 Summary  Apoptosis is an important field of study, because of the impact of abnormal apoptosis on human health (the diseases of “inappropriate apoptosis”). Apoptosis is a complex, well- regulated, highly conserved process, mediated by multiple interactions between elements of the pathways for cell survival and proliferation, as well as specific signaling related directly to apoptosis.  Because of this complexity, a proteomics approach is a useful means of addressing questions of global changes during apoptosis. Cytokine withdrawal-induced apoptosis is a useful model for the investigation of apoptosis.     55 We chose to investigate early apoptosis (pre-apoptosis), in the hope of identifying important changes in protein concentrations which might point the way to further work, leading ultimately to an advancement in the therapeutic approach to some of the diseases of inappropriate apoptosis.     56 2 Materials and Methods  2.1 Cell Culture TF-1 cells (American Type Culture Collection, Manassas, VA) were grown in RPMI 1640 medium supplemented with 500 mg/L L-Glutamine (Sigma-Aldrich Canada, Oakville, ON), 50 µM (-mercaptoethanol (Mallinckrodt Baker Inc. Phillipsburg, NJ), 10% (v/v) Fetal Bovine Serum (Invitrogen Canada, Burlington, ON) and 1% (v/v) CGMI cell conditioned medium (a source of human GM-CSF (see below)). Cells were sub-cultured every 2-3 days to ensure that cell density was never greater than 8 x 10 5  cells/mL. Cells were grown in 10cm- or 15cm-diameter sterile Petri dishes in 5% CO2 at 37° C in a humidified incubator. 2.2 Cell Culture – Stable Isotope Labelling  For the SILAC experiments, TF-1 cells (American Type Culture Collection, Manassas, VA) were grown as described above in either “ 12 C medium” or “ 13 C medium” for labelling. These media were formulated from specially manufactured RPMI 1640 medium that had been prepared without L-Arginine, L-Lysine or L-Glutamine (JR Scientific, Woodland, CA).  The 12 C medium was supplemented with 100 mg/L of 12 C-L-Arginine and 20 mg/L of 12 C-L- Lysine (both Sigma-Aldrich Canada, Oakville, ON). These concentrations of supplemental amino acids are half the normal concentrations of L-arginine and L-lysine found in RPMI 1640, since I was able to confirm in early trials that TF-1 cells would grow normally in media containing reduced amounts of L-arginine and L-lysine.     57 For the 13 C medium, the amino acid deficient RPMI was supplemented with 100 mg/L of 13 C-L-Arginine and 20 mg/L of 13 C-L-Lysine (Cambridge Isotope Laboratories, Andover, MA). Both isotopes are uniformly labeled, i.e. each of the six carbon atoms per molecule in both isotopes is substituted with 13 C. Stock solutions at 1000X concentrations of both 12 C- and 13 C-L-arginine and 12 C- and 13 C-L-Lysine were made up in phosphate buffered saline (PBS).  Both the 12 C and the 13 C media formulations were supplemented with 500 mg/L L-Glutamine (Sigma-Aldrich Canada, Oakville, ON), 50 µM (-mercaptoethanol (Mallinckrodt Baker Inc. Phillipsburg, NJ), 10% (v/v) dialyzed, heat inactivated fetal bovine serum (Invitrogen Canada, Burlington, ON) (fetal bovine serum was heat-inactivated by placing in a water bath at 55 ° C for 30 minutes), and 1% (v/v) CGMI cell conditioned medium as a source of human GM-CSF (see below). The constituents were mixed together and then filtered through a 0.2 µM filter with vacuum. Prepared media were stored at 4° C and warmed to 37° C immediately prior to use.  Cells were cultured for six-doublings, by which time greater than 98% of all proteins would have incorporated labels. I was able to confirm very high incorporation rates by examining 13 C-labelled cell lysate using mass spectrometry.  After six doublings, cells were washed three times in sterile PBS to deplete growth factors (especially GM-CSF). Cells labeled with 13 C were returned to the 13 C medium from which they had been removed for washing. Cells labeled with 12 C were re-suspended in fresh 12 C    58 medium and “starved” for fifteen hours by omitting GM-CSF from the medium. At fifteen hours, both groups of cells were spun down and washed once with PBS, then lysed in solubilization buffer (recipe below). Cells were disrupted with an ultrasound sonicator for three x 10 seconds on ice. Lysates were cleared by centrifugation at 13,200 rpm, 4ºC for 5 minutes, and the supernatant collected. Protein concentration in the supernatant was determined using the Bradford Assay (BioRad Laboratories (Canada), Mississauga, ON) (see protocol below). 13 C-labelled and 12 C-labelled lysates were mixed in a ratio of 1:1 based on protein concentration. Three experiments were carried out using the above procedure. The experimental procedure is outlined in Figure 2.1 below.                  59   Figure 2.1 Experimental overview for three cytokine-withdrawal experiments. Forty, 90 and 70 fractions were subjected to in-gel trypsin digestion and peptide extraction, giving 200 samples in total, which were examined using an FT-ICR mass spectrometer.    13 C - Control 12 C - Starve Lyse cells Determine Protein Conc’n Mix 1:1 SDS-PAGE 40, 90 or 70 Fractions Lyse cells Determine Protein Conc’n 15 hr Starvation    60 2.3 CGMI – Preparation and Usage  CGMI is the name given to baby hamster kidney (BHK) cell line that has been stably transfected with the human GM-CSF gene, enabling these cells to secrete hGM-CSF. CGMI cells were a kind gift of Dr C. Brown, University of Calgary. Cells were grown to confluence in 15cm-diameter tissue culture dishes in Dulbecco’s Modified Eagle’s Medium (DMEM, Invitrogen Canada, Burlington ON), supplemented with 500 mg/L L-Glutamine (Sigma- Aldrich Canada, Oakville, ON), 50 µM (-mercaptoethanol (Mallinckrodt Baker Inc. Phillipsburg, NJ), 10% (v/v) fetal bovine serum (Invitrogen Canada, Burlington, ON) in 5% CO2 at 37° C. After the cells reached confluence, they were left for another two days. The medium was decanted, spun down at 6000 rpm at room temperature, and filtered through a 0.2 µM filter with vacuum. This conditioned medium contained hGM-CSF, and was used at a final concentration of 1% (v/v), after confirming its activity in a bioassay of TF-1 cell proliferation.   2.4 Cell Lysis Buffer (Solubilization buffer) & Sample Loading Buffer  Cell lysis buffer consisted of 50mM Tris-HCl, pH 7.7, 1% Triton X-100, 10% Gycerol, 100 mM NaCl, 2.5 mM EDTA, 10 mM NaF, ddH2O. Immediately prior to use, this buffer was supplemented with 0.2 mM sodium orthovanadate, 1 mM sodium molybdate and a 1:1000 dilution of Protease Inhibitor Cocktail (Sigma-Aldrich Canada, Oakville, ON). This buffer was used throughout this series of experiments at 50-+L lysis buffer for each 1 x 10 6  cells as determined by automated cell counting.  Sample loading buffer (4X) was prepared according to the following recipe.    61 For 40 mLs of sample buffer: 100% glycerol, 20 mLs, 1M Tris-HCl pH 6.8, 5 mLs, 20% SDS, 8 mLs, 1% Bromophenol blue, 3.2 mLs, Distilled water, 1.8 mLs  Immediately prior to use, (-mercaptoethanol at 5% (v/v) was added.   2.5 SDS-PAGE – Gel Preparation and Staining  Gels were cast in a BioRad Mini-Protean II apparatus  (BioRad Laboratories (Canada), Mississauga, ON) unless otherwise stated. Solutions used were as follows: 30% acrylamide/Bis-acrylamide, 37.5:1 (2.6%C) (BioRad Laboratories (Canada), Mississauga, ON) TEMED (N,N,N’,N’-tetra-methyl-ethylenediamine) (BioRad Laboratories (Canada) Mississauga, ON) Ammonium persulfate (BioRad Laboratories (Canada), Mississauga, ON).  Mini-gels were prepared using a 9% or 12% (as noted) resolving gel with a 4% stacking gel according to the method described by Laemmli (Laemmli, 1970).  All gel casting components were cleaned with detergent and water, thoroughly rinsed in distilled water, and dried prior to assembling the casting chamber. All reagents were combined, leaving the initiators (APS and TEMED) until just before pouring. When ready to pour the gel, the APS and TEMED were added to the other components in solution, and gently mixed (to avoid oxygenating the solution) before being poured into the gel cast(s).     62 Large format gradient gels were prepared according to the methods of the manufacturer (BioRad Laboratories (Canada), Mississauga, ON) with a linear gradient from 8%-16.5%, using separate acrylamide-bis-acrylamide solutions of 8% acrylamide and 16.5% acrylamide. As with the preparation of the mini-gels, all gel casting components were cleaned with detergent and water, thoroughly rinsed in distilled water, and dried prior to assembling the casting chamber. All reagents were combined, leaving the initiators (APS and TEMED) until just before pouring. When ready to pour the gel, the APS and TEMED were added to both solutions, which were gently mixed (to avoid oxygenating the solutions) and poured into the appropriate chambers of the gradient pourer (BioRad Laboratories (Canada), Mississauga, ON). A stir bar, which had been previously placed in the mixing chamber of the gradient pourer, mixed the acrylamide solutions, which were allowed to feed by gravity into the casting chamber.  Coomassie R-250 gel staining  Gels were washed for 5 minutes in purified water (Milli-Q water, Millipore Corporation, Billerica, MA), three times, prior to being fixed in an aqueous solution of 40% methanol, 10% acetic acid for one hour. Gels were rinsed for 5 minutes in water three times, prior to being covered in Coomassie ®  Blue R-250 solution (2 gm Coomassie R-250 (Sigma-Aldrich Canada, Oakville, ON), 100 mL glacial acetic acid, 250 mL methanol, 150 mL water), and left to rotate gently for 45 minutes. Gels were rinsed for 5 minutes in water three times prior to de-staining overnight in 40% methanol, 10% acetic acid with gentle rotation. Gels were rinsed in water prior to cutting.    63 2.6 In-Gel Trypsin Digestion and Peptide Extraction  The procedure used for in-gel trypsin digestion and peptide extraction was developed by the European Molecular Biology Laboratory – Protein and Peptide Group in 1997, and published in modified form by Pandey et. al. in 2000 (Pandey, Andersen, & Mann, 2000). I received training in this procedure, but it was felt that the results might be more reproducible if this procedure were to be left to the hands of Ms Shujun Lin, technician in the laboratory of Dr Juergen Kast. I was responsible for the gel slicing. In brief, the gel was rinsed with water, and the bands excised with a clean scalpel. A hair cover, face-mask and gloves were worn in addition to a clean laboratory coat, to prevent or diminish potential keratin contamination. The bands were chopped into approximately 1 x 1 mm cubes and transferred to 0.5 mL Eppendorf tubes. Gel pieces were washed with 100-150 µL distilled-deionized (dd) H20 for 5 minutes. Tubes were centrifuged and the supernatant removed. Acetonitrile (at 3-4 times the volume of the gel pieces) was added. Gel fragments were incubated for approximately 15 minutes until gel pieces had shrunk, becoming white and sticking together. Tubes were centrifuged and the liquid removed and discarded. Gel fragments were dried in a vacuum centrifuge at 40°C for approximately 10 minutes. Gel fragments were swollen by the addition of 10 mM dithiothrietol/0.1M NH4HCO3 solution (just enough volume to cover the gel pieces), and incubated for 30 minutes at 56° C to reduce the proteins. Tubes were spun down, the liquid removed, and 100 µL acetonitrile added to shrink the gel pieces. After incubation for 15 minutes at room temperature, the supernatant was removed and replaced with 100 µL 55 mM iodoacetamide/0.1M NH4HCO3 to alkylate the proteins. Tubes were incubated for 20 minutes at room temperature, in the dark. Tubes were centrifuged, supernatant was removed and gel fragments washed using 150-200 µL 0.1M NH4HCO3 for    64 15 minutes. Tubes were once again centrifuged, liquid removed and gel fragments shrunk using acetonitrile (100 µL for 15 minutes). After removal of the liquid, since the fragments were still blue from Coomassie staining, the gel fragments were rehydrated in 150 µL of 0.1M NH4HCO3 for 15 minutes, then an equal volume of acetonitrile was added (for a 1:1 mixture). Tubes were placed into a thermal mixer and mixed for 15 minutes, spun, and the liquid removed. Gel fragments were shrunk by adding 100 µL acetonitrile. Tubes were spun down, liquid removed and fragments dried in a vacuum centrifuge at 40°C for approximately 10 minutes. Gel fragments were rehydrated in digestion buffer containing 50 mM NH4HCO3, 5 mM CaCl2 and 12.5 ng/µL of trypsin at 4°C (on ice) for 40 minutes (after 15-20 minutes, tubes were checked, and more digestion buffer was added if all the liquid had been absorbed by the gel fragments). After 40 minutes total, any remaining supernatant was removed and 5- 25µL of the same digestion buffer, but without trypsin, was added (enough to cover the gel pieces). Tubes were incubated at 37°C overnight. Tryptic peptides were extracted from the gel fragments as follows. Ten -15 µL of 25 mM NH4HCO3 was added to the buffer in each tube and the tubes incubated at 37°C for 15 minutes. Acetonitrile was added (1-2 times the volume of the gel fragments). Tubes were incubated at 37°C for a further 15 minutes with shaking. Supernatants were transferred to separate tubes and 40 µL of 5% formic acid added. Tubes were placed into a thermal mixer and mixed for 15 minutes at 37°C. Acetonitrile was added (1-2 times the gel fragments volume). Tubes were incubated at 37°C for a further 15 minutes with shaking, spun down and both supernatants (i.e. pre-extraction and post- extraction from the same gel fragments) were pooled together. Supernatants were dried down in a vacuum centrifuge. Prior to use, desiccated peptides were reconstituted with 5% formic acid.    65  Peptides extracted from each gel slice were loaded onto an in-house-manufactured reverse phase C-18 column, 15 cm length, 75uM OD, 3µM ID for nano-flow liquid chromatography using a 1100 HPLC (Agilent Technologies, Canada, Mississauga, ON) as a front end interface to mass spectrometry analysis on a Thermo LTQ-FT instrument (Thermo Fisher Scientific, Waltham, MA). Sample runs were 105 minutes in length. The mobile phase A consisted of 0.5% acetic acid in H20 and the phase B was 0.5% acetic acid in 80% acetonitrile + 19.5% H20. Detailed information on the chromatography is shown in Table 2.1.    66  Time Module Setting Value t(min) %A %B Flow µL/min µL A µL B 0.00 Nano Pump Column Flow 0.6 µL 0.00 Nano Pump %B 6.0% 0.00 Micro WPS External Contact Closed 0.50 Micro WPS External Contact 1 Open new 20.00 Nano Pump Column Flow 0.6 µL/min 0 94.00 6.00 0.60 11.28 0.72 20.00 Nano Pump %B 6.0% 20 94.00 6.00 0.60 0.19 0.21 21.00 Nano Pump Column Flow 0.2 µL/min 21 94.00 6.00 0.20 3.78 7.02 75.00 Nano Pump %B 30.0% 75 70.00 30.00 0.20 0.90 1.10 85.00 Nano Pump Column Flow 0.2 µL/min new 85 20.00 80.00 0.20 0.40 1.60 85.00 Nano Pump %B 80.0% 90 20.00 80.00 0.60 0.60 2.40 90.00 Nano Pump Column Flow 0.6 µL/min new 95 20.00 80.00 0.60 0.34 0.26 95.00 Nano Pump %B 80.0% modified 96 94.00 6.00 0.60 5.08 0.32 96.00 Nano Pump %B 6.0% modified 105 94.00 6.00 0.60  Table 2.1 Details of the chromatography runs on the 1100 HPLC (Agilent Technologies, Canada, Mississauga, ON) used in each of the three FT-ICR experiments.   2.7 SDS-PAGE – Immunoblotting  In preparation for immunoblotting (Western blotting (Burnette, 1981)), cells were counted, then spun down (6000 rpm, 4°C, 5 minutes) and the supernatant removed without disturbing the pellet. Cells were washed once in ice cold PBS (Sigma-Aldrich Canada, Oakville, ON). Cells were spun again (6000 rpm, 4° C, 5 minutes) and the supernatant removed, without disturbing the pellet, before being lysed on ice in cell lysis buffer (recipe 2.4 above) using 50    67 µL lysis buffer for each 1 x 106 cells as determined by an automated cell counter (Beckman- Coulter Z-1 particle counter). Cells were vortexed three times for ten seconds over three minutes, then sonicated for 3 x 10 seconds on ice. Lysates were spun at 13,200 rpm for 5 minutes at 4°C to remove debris. Supernatants were carefully removed. A Bradford protein assay (see below) was performed at this point, if required. Lysates were mixed in ratio of 3:1 (v/v) with 4X loading buffer (recipe below) prior to “boiling” for five minutes in a heat- block set to 95°C. Samples were loaded onto a gel (mini-gel) using a wet-loading technique, and the gel was transferred to nitrocellulose using a semi-dry transfer on a Pharmacia-LKB MultiPhor unit as recommended by manufacturer. 2.8 Immunoblotting (Western blotting)  Following transfer the nitrocellulose membranes were stained using a Ponceau-S solution (Sigma-Aldrich Canada, Oakville, ON) to verify transfer of proteins. Membranes were rinsed prior to being blocked in 5% (w/v) skim milk in TBST for one hour at room temperature (TBS is 50 mM Tris, 150 mM NaCl, pH 7.6. TBST has Tween-20 ® added to TBS at 0.0005% v/v). Membranes were rinsed for five minutes in TBST, three times. Primary antibody (usually at 1:1000, but actual dilutions are given for each antibody used – see Table 2.2 below) were diluted in TBS containing 5% BSA or skim milk powder, depending on antibody, plus 0.02% sodium azide and sufficient liquid added to cover the membranes. Membranes were incubated either overnight at 4°C or for two hours at room temperature, depending on the antibody, with gentle rocking. Membranes were washed three times five minutes in TBST. Secondary antibody (dilutions varied from 1:2000 to 1:5000 in TBST with 5% skim milk powder) was added for one hour at room temperature. Membranes were washed three times five minutes in TBST prior to the application of Enhanced    68 Chemiluminescence solution (0.5 mL each solution A & B) (Amersham ECL, GE Healthcare Bio-Sciences Corp, Piscataway, NJ), and exposure to X-ray film, or image was captured using a high sensitivity camera.    69 2.9 Antibodies  The following primary antibodies were used in the work reported in this thesis:  Table 2.2 Primary antibodies used in the experiments reported in this thesis.  Antibody Supplier Species Type Conditions; size of protein detected Anti- phosphotyrosine, clone 4G10 Upstate Cell Signaling Solutions, Lake Placid, NY Mouse Monoclonal 1:1000; Various mW Overnight, 4°C Anti-PDCD4  Abcam Inc., Cambridge, MA Rabbit Polyclonal 1:1000; 52 kDa 2 hours, RT Anti-HMGB2 Abcam Inc., Cambridge, MA Mouse Monoclonal 1:200; 24 kDa Overnight, 4°C Anti-Thymidylate synthase Abcam Inc., Cambridge, MA Mouse Monoclonal 1:200; 35 kDa 2 hours, RT Anti-SHIP A kind gift from Dr Gerald Krystal Rabbit Polyclonal 1:1000; 145 kDa 2 hours, RT Anti-actin (I-19) Santa Cruz Biotechnology, Santa Cruz, CA Goat polyclonal 1:1000; 43 kDa 2 hours, RT Anti-vinculin Sigma-Aldrich (Canada), Oakville, ON Mouse Monoclonal 1 µg/mL; 130 kDa 2 hours, RT Anti-cdc42 Santa Cruz Biotechnology, Santa Cruz, CA Rabbit Polyclonal 1:500; 25 kDa Overnight, 4°C Anti-p85 (PI3-K) Cell Signaling Technology, Beverly, MA Rabbit Polyclonal 1:1000; 85 kDa Overnight, 4°C    70 Secondary antibodies were obtained from Dako Cytomation (Dako Denmark, Glostrup, Denmark) and were peroxidase-conjugated goat anti-rabbit, goat anti-mouse or rabbit anti- goat immunoglobulins – determined according to the primary antibody used. Dilutions were 1:3000 to 1:5000 depending on results obtained. Antibodies were diluted in TBST with 3% skim milk powder for the anti-rabbit and anti-mouse antibodies, and 1% BSA for the anti- goat secondary.  2.10  Flow Cytometry  Cells were stained with propidium iodide for analysis of apoptosis using flow cytometry using an adaptation of the protocol given in Current Protocols (Darzynkiewicz & Huang, 2004). One-milliliter aliquots of cells at approximately 0.5 x 10 6  per mL were removed for each condition to be tested. Cells were centrifuged in 15 mL conical tubes at 400 x g for 5 minutes at room temperature. The cell pellets were washed once in 1 X PBS (Ca 2+ -, Mg 2+ - free, + 0.1% glucose). Cells were centrifuged at 400 x g for 5 minutes at room temperature. Supernatant was removed (except for approximately 100 µL liquid, in order to avoid disturbing the pellet). The pellet was vortexed briefly to resuspend the pellet and to break up confluent cells. One milliliter of ice-cold 70% ethanol was added drop-wise while vortexing. Cells were fixed for a minimum of one hour at 4°C.  For propidium iodide staining, cells were vortexed briefly. Tubes were centrifuged at 3000 rpm for 5 minutes at room temperature (because cells are more buoyant in ethanol). All but 100 µL of supernatant was removed. The pellets were resuspended in one milliliter of    71 propidium iodide solution containing 1 mL PBS, Ca 2+ -, Mg 2+ - free, + 0.1% glucose, 10 µL 10 mg/mL RNase A, 5 µL 10 mg/mL propidium iodide (Sigma-Aldrich Canada, Oakville, ON). Cells were allowed to stain in the dark for at least 30 minutes prior to analysis with the flow cytometer.  2.11  Bradford Protein Concentration Assay  The Bradford protein assay (Bradford, 1976; Stoscheck, 1990) (Bio-Rad, Hercules, CA) measures the change in the colour of Coomassie ® Brilliant Blue G-250 dye as it binds with protein. The dye absorption shifts from 465nm to 595 nm with protein binding. The procedure followed was as outlined in the Bio-Rad Protein Assay instruction manual (catalogue number 500-0006). Briefly, a series of albumin protein standards were prepared at 2000, 1000, 500, 250, 125 +g/mL. A blank was also prepared. Cell lysates were prepared as described elsewhere. Lysates were diluted 1:20 in water. Ten +L of each protein standard, the blank, and each unknown sample were pipetted into a microtitre plate, in triplicate. Two hundred +L of the diluted Bio-Rad Bradford solution (5X stock solution, diluted in water) was added to each well containing sample or standard. The plate was shaken, and allowed to rest as colour developed over ten minutes. The plate was read at 595 nm using a microtitre plate reader.      72 2.12  Ac-DEVD-pNA Caspase Assay  Ac-DEVD-pNA is a colorimetric substrate for caspase-3. The sequence is based on the poly- ADP-ribose polymerase (PARP) cleavage site Asp-216. Cleavage of the artificial substrate by active caspase-3 results in development of a colorimetric reaction in a quantitative manner. Release of pNA is monitored by absorbance at 405 nm. The DEVD-pNA stock solution is 10 mM in DMSO (Calbiochem, San Diego, CA). Buffer for DEVD-pNA activity is 50 mM HEPES, pH 7.6, 5 mM EDTA, 0.5% Triton X-100 and 2 mM DTT.  TF-1 cells were seeded at a density of 2 x 10 6  /mL in 4 mL RPMI 1640 in 5-cm Petri dishes. Cells were incubated in the presence and absence of 1% CGMI. At the time-points indicated, cells were harvested, washed once in PBS and the pellets resuspended in 30 µL DEVD-pNA activity buffer. Cells were rotated at room temperature for 10 minutes, centrifuged for 10 minutes at 19,000 x g and the supernatants collected.  Two +L of supernatant were used for the Bradford protein assay. The DEVD-pNA substrate was diluted 1:100 in buffer and 100 µL added for each 50 µg protein. Samples were incubated in the dark at 37°C for 30 minutes, prior to reading absorbance at 405 nm.      73 2.13  Methanol-Chloroform Protein Precipitation  Sample volume was determined prior to protein precipitation. Three volumes of methanol and one volume of chloroform were added to the samples and vortexed. Three volumes of purified water were added, and the samples vortexed for one minute. Samples were centrifuged for five minutes at 10,000 x g. The upper phase was discarded. Three volumes of methanol were added. Samples were vortexed briefly. Samples were centrifuged for 5 minutes at 10,000 x g. The supernatant was removed and discarded. The pellet was allowed to air dry, prior to storage at -80 °C.  2.14  Data Handling  Mass spectrometry raw data files from the LTQ-FT were obtained in a proprietary .raw format. The raw data were converted to “.dta” format using extract_MSN.exe (Thermo Fisher Scientific, Waltham, MA). Using the software program DTA-Supercharge (available from http://msquant.sourceforge.net), the .dta files were converted to “.mgf” files for the next step in the analysis.  Files in the .mgf format were submitted for analysis with MASCOT (Matrix Science, www.matrixscience.com) for protein identification. MASCOT parameters included:  use of the MSDB database,  taxonomy homo sapiens,  allow up to 1 missed cleavage,  fixed modification: carbamidomethyl,    74  variable modifications: oxidized methionine, lysine-6-13C, arginine-6-13C,  peptide tolerance ± 5ppm,  MS/MS tolerance ± 0.6Da,  peptide charge 2+ and 3+. Analysis with MASCOT returned an “.html” file which was then analyzed using MSQuant (Schulze & Mann, 2004) which returns the list of identified proteins with information on the relative quantitation of the 12 C-labelled proteins and the 13 C-labelled proteins.  The peptide information was obtained from the MSQuant program by loading the original MSQuant-generated “.m3b” file and exporting the protein and peptide information. This yielded a redundant list of peptides identified by the mass spectrometer.  To derive the final definitive protein list, a non-redundant peptide list was generated by sorting the raw data peptides according to descending Mascot MOWSE scores, and then removing redundant peptides. This procedure generated a list of the highest-scoring peptides, with each peptide occurring only once. The list of peptides was then searched against the list of all human proteins in “fasta” format using the program “finalList.pl”. The fasta file used was dated 13th March, 2008, the list of human proteins used was the IPI human database (Kersey et al., 2004), version 3.40, containing 69965 entries. The perl program, finalList.pl, was used to find the smallest number of protein sequences that can be used to explain all the peptide data (Dobreva, Fielding, Foster, & Dedhar, 2008; Rogers & Foster, 2007). The resultant list of proteins identified was based on each peptide being used once only, to identify a single protein. (finalList-mod.pl was a kind gift from Dr Leonard Foster, University of British Columbia).    75  Next, the quantitation data for each peptide was extracted from the MSQuant .m3b file, using the peptide sequence information and the i/w average of ratios data. Entries without quantitation data were removed, and redundant entries were again removed.  The above two steps generated a final list of proteins identified from a non-redundant list of the highest scoring peptides, and with associated quantitation ratios for each protein.  2.14.1 Working with the Final List of Proteins  The final list of proteins required some manual reviewing to ensure that the data were satisfactory. The following steps were undertaken: 1. Proteins with outlying ratios were checked manually – outlying ratios are often not accurate. For example, keratin contamination will sometimes appear with a very low isotope/wild-type ratio. In these cases, the keratin is a contaminant. Note that the labeled cells can have intrinsic keratin identified, but in these cases of genuine labelling, the keratin ratio does not change much, and so is found closer to a ratio of 1:1 (and not 20:1 as was sometimes reported for the contaminants). For each protein with the highest or lowest ratios the spectra were manually reviewed. Spectra from the list of proteins with outlying ratios were reviewed sequentially, starting with the proteins showing the highest and lowest ratios, and continuing until the protein spectra showed a clear indication that the data were genuine, and not the result of noise or contamination. The procedure was as follows:  a. copy peptide used for ID    76  b. search against the raw data from the .mb3 file to locate the protein identified by MSQ/Mascot  c. open MSQ and search for the spectra for that protein  d. determine if the spectra represent a real occurrence, or artifact  e. remove protein entry if artifact 2. remove proteins identified by only one peptide 3. remove proteins identified by two peptides where one (or both) of the two are less than seven (7) residues long 4. calculate the average i/w for the remaining proteins 5. using the average above, correct the i/w ratios for pipetting error and any other error in determining protein concentration.    77 2.14.2  Blast2GO  Lists of IPI accession numbers were obtained from each experiment as described above, and used to determine the gene ontologies (Ashburner et al., 2000) of the proteins detected. Lists were converted to fasta format using the EMBL-EBI website “dbfetch” function (http://www.ebi.ac.uk/cgi-bin/dbfetch). The fasta lists were imported into Blast2GO (Ver. 2.2.3) (Conesa et al., 2005). The non-redundant blast database at the NCBI was used as the reference for a protein blast (blastp). Settings for blastp were:  expectation value minimum: 1 x e-3  high scoring segment pair cut-off: 33. Default parameters were used to make the Annotations:  pre-eValue-Hit-Filter: 1 x e-6,  Annotation cut-off: 55,  GO Weight:  5. Directed acyclic graphs were then prepared using a cut-off (sequence filter) of 5, a score # of 0.6 and a node score filter of 0. This allowed us to generate directed acyclic graphs at level 4 gene ontologies.  2.14.3 Ingenuity ® Pathway Analysis  Functional and pathway analysis of proteins identified as being cytokine withdrawal- regulated was carried out using Ingenuity Pathway Analysis, version 7.5 (IPA, Ingenuity  ® Systems, www.ingenuity.com). IPA is a tool for the description of molecular networks and signaling pathways. The IPA proprietary database has been extracted from the experimental    78 literature by manual evaluation of the full texts of papers (less than 10% of the database has been automatically extracted and modeled).  The database contains only results that have been confirmed by multiple laboratories, thus controversial results are not included in an IPA analysis.  Analysis was carried out using the 41 proteins identified as increasing in concentration in response to cytokine withdrawal, and the 82 proteins identified as decreasing in concentration in response to cytokine withdrawal. Proteins were assigned to various functions, and networks were generated. The Functional Analysis identified the biological functions that were most significant to the data set. Proteins used for the analysis were those that had previously met the determined cut-off for inclusion, and were associated with biological functions and/or diseases in the Ingenuity Pathways Knowledge Base. Fischer’s exact test was used to calculate a p-value determining the probability that each biological function and/or disease assigned to that data set is due to chance alone.      79 3 Proteomic Analysis of a Hemopoietic Cell Line Undergoing Apoptosis. 3.1 Introduction  Cell lines derived from hematopoietic lineages have served as useful models for studying the onset of apoptosis. In hematopoietic cells that are dependent on specific cytokines for growth and survival, the removal of the cytokine, or cytokine starvation, can result in apoptosis, even though the cells are in the presence of all other nutrient and serum-derived factors that are normally present in growth medium (see, e.g. (Brach, deVos, Gruss, & Herrmann, 1992; Lotem & Sachs, 1996; Minshall, Arkins, Freund, & Kelley, 1996; J. R. Park, 1996)). By virtue of this behaviour, hematopoietic cell lines are good models of inflammatory cells in the bloodstream that normally have a very short half-life, and are dependent for survival upon the presence of cytokines, which they may encounter at a site of infection or inflammation. Hence this experimental model mimics the natural process of lymphocyte blood cell homeostasis through cytokine-withdrawal apoptosis that occurs in the body in the aftermath of infection; for example, when increased numbers of neutrophils previously generated to fight infection are no longer required and undergo apoptosis as a result of decreased cytokine levels generated to support the (temporarily enlarged) cell population (Blank et al., 1997; Pellegrini, Belz, Bouillet, & Strasser, 2003).  We now know that a major reason for cells undergoing apoptosis is the loss of survival signals, many of which stem from activation of the PI3-K/PKB signaling pathway (Alessi & Cohen, 1998; Klein et al., 2000; Marte & Downward, 1997; Rameh & Cantley, 1999). In order to have a more complete understanding of the specific events that are critical for    80 survival of a particular cell type, it was felt that a global analysis of the proteome of cells undergoing apoptosis would provide many leads for further investigation.  The studies reported here were carried out using a human erythroleukemia cell line - TF-1 – established in 1989 in the Takaku lab using mononuclear cells from a 35-year-old male with erythroleukemia (Kitamura et al., 1989). TF-1 cells are cytokine-dependent, that is, they require the continued presence of cytokine for survival and proliferation. In the case of TF-1 cells, the required cytokines are (one of) interleukin-3 (IL-3), interleukin-5 (IL-5), granulocyte-macrophage colony stimulating factor (GM-CSF), or erythropoietin (EPO – although EPO only sustains the short-term growth of these cells)(Kitamura et al., 1989). A study by Rosas et. al. in 2005 investigating cytokine-withdrawal apoptosis in TF-1 cells showed that IL-5 withdrawal led to dephosphorylation and activation of the Forkhead transcription factor FOXO3a, leading to increased expression of Bim, and that increased levels of Bim were sufficient to cause apoptosis in these cells (Rosas, Birkenkamp, Lammers, Koenderman, & Coffer, 2005). This study demonstrated the importance of activation of the PI3-K pathway in causing inhibition of apoptosis in TF-1 cells. In this work, normal growth media was supplemented with recombinant hGM-CSF, produced by cells of the CGMI cell line (Schubert & Duronio, 2001). GM-CSF signaling in hematopoietic cells is mediated by several pathways, including; PI3-K, MAPK, JAK/STAT and PKC (de Groot, Coffer, & Koenderman, 1998; Wheadon, Roberts, Watts, & Linch, 1999). GM-CSF has been shown to inhibit apoptosis in neutrophils (Klein et al., 2000; Yasui et al., 2002) and macrophages. Apoptosis begins in TF-1 cells following several hours of cytokine    81 withdrawal. Cells show visually detectable signs of apoptosis by about 18 hours and the colonies proceed to die completely within the period from 24 to about 30 hours.  The use of TF-1 cells in these experiments offered several advantages: (i) TF-1 cells undergo cytokine-withdrawal apoptosis more slowly than most murine cytokine-dependent hematopoietic cell lines, giving a larger “window” in which to observe protein changes under conditions of cytokine withdrawal, (ii) TF-1 cells have been fairly well characterized in terms of signaling events, and (iii) because TF-1 are human cells, it is hoped that our data might be of use to the Human Proteome Organization (HuPO – http://www.hupo.org).  Several previous studies have characterized some aspects of changes to the proteome during apoptosis (see, e.g.(An & Seong, 2006; Bruneel et al., 2005; Gerner et al., 2000; Herzog, Kuntz, Daniel, & Wenzel, 2004; Hwang et al., 2006b; Kozielski et al., 2008; Lau, He, & Chiu, 2004; S. C. Lee, Chan, Clement, & Pervaiz, 2006; Machuy et al., 2005; Qiu, Gao, Li, & Shen, 2008; Short et al., 2007; Thiede, Dimmler, Siejak, & Rudel, 2001; Thiede, Siejak, Dimmler, & Rudel, 2002; Winkelmann, Nassl, Daniel, & Wenzel, 2008)). This thesis describes work performed using cytokine-withdrawal for the induction of apoptosis.  By adopting this approach, we have been able to observe the effects on the proteome of the withdrawal of survival signaling, rather than the imposition of a death-inducing signal by such means as the use of Fas-L, staurosporine, camptothecin or oligomycin for the induction of apoptosis. As well, many of the published studies have used two-dimensional gel electrophoresis, which has been an enormously valuable tool in proteomics. However, we believe that mass spectrometry offers significant advantages over two-dimensional gel    82 electrophoresis for the study of whole cell lysates – greater depth of discovery of the proteome, along with increased opportunities to observe membrane-associated proteins, and proteins with a high pKa. Further, this study was performed using three biological replicates – giving increased confidence in the identification of the proteins and in the quantitations reported.  It was decided at the outset of these experiments to use whole cell lysates in the analysis. This was decided for two reasons. Firstly, sub-cellular fractionation at this point would have introduced more experimental variability, and would have complicated quantitation by producing a wider margin of error (however, fractionation could be introduced as an additional step in later experiments, once the basic protocol had been established). Secondly, while recognizing that whole cell lysates are more complex than lysates from sub-cellular fractions, and that this complexity would necessarily limit the total number of proteins identified, and hence the coverage of the proteome, sub-cellular fractionation as part of the sample preparation would have led to a very large number of samples for the analysis of the complete cellular proteome. Such a large number of samples would make it impossible to get sufficient mass spectrometry instrument time for analysis. Hence for experimental design, and for pragmatic reasons, whole cell lysates were used.  In this chapter, the background work done in order to optimize the conditions for the mass spectrometry analysis is presented, followed by the most important results obtained from the mass spectrometry analysis – the lists of proteins which have shown an important increase or decrease in concentration (i.e. apparent up-regulation or down-regulation) in response to    83 fifteen hours of cytokine withdrawal. Results from three biological replicates of this experiment are presented.  The protein abundance figures give an indication of the technically demanding nature of experiments of this type, with protein concentrations varying by up to nine orders of magnitude in serum (Adkins et al., 2002), and at least six orders of magnitude within cells (Schirle, Heurtier, & Kuster, 2003). Initial experiments to determine changes in protein concentration in response to IL-3 stimulation were disappointing. My first approach was to use two-dimensional gel electrophoresis, comparing separate gels used for control and starving lysates, and using Surface Enhanced Laser Desorption/Ionization (SELDI) mass spectrometry and peptide mass fingerprinting for protein identification. No attempt at quantitation could be made with the technology at the time. A typical experiment highlights the low yield from this approach – thirty-eight proteins shown to increase in response to cytokine stimulation, and eight shown to decrease. Apart from the difficulty in visualizing a large enough number of proteins (due to the dynamic range required of the technique, as stated above), one of the problems with this approach was that spots frequently contained more than one protein. A second difficulty was that smaller spots did not always contain sufficient material to allow analysis with the SELDI instrument. The most interesting change detected in this series of experiments was an increase in the concentration of the highly abundant protein hnRNP-A1 in response to cytokine withdrawal. When it became apparent that this method would be inadequate for my needs, I undertook to investigate changes in the proteome in response to cytokine withdrawal using a Q-Star instrument (quadrupole time-of- flight). This also was unsatisfactory. I was unable to see sufficient detail of changes in the    84 proteome in response to cytokine withdrawal. By using FT-ICR mass spectrometry for analysis of the stable isotope-labeled samples it became possible to visualize the proteome down to the level of some of the less abundant signaling molecules; and by doing so, to view changes in relative quantitation of these lower abundance proteins in response to various experimental conditions.  3.2 Experimental Overview and Background Material  The details of the experimental procedure are given in chapter 2. Briefly, using the SILAC procedure, TF-1 cells were labelled, incorporating a stable isotope of carbon ( 13 C), to enable comparison with TF-1 cells which had been labeled with wild-type carbon ( 12 C), and starved of cytokine for fifteen hours. Cells were lysed, and whole cell lysates were prepared and fractionated on SDS-PAGE gels prior to separation on a reverse phase chromatography column for analysis using an FT-ICR mass spectrometer. This procedure was carried out three times, giving a series of biological replicates.  In this series of experiments, the time-point chosen for investigation, i.e. fifteen hours of cytokine withdrawal, was chosen because of the opportunities it presented to study early apoptosis, or what might be referred to as “pre-apoptosis”. – Pre-apoptosis can be defined as that period in the process of apoptosis after which the cells have committed to apoptosis, but (prior to) or (much earlier than) the point at which to fragmentation of the cells can be observed. In TF-1 cells, fifteen hours of cytokine withdrawal is sufficient to cause activation of caspases, hence it is likely after the “point of no return”, but is not so far along in the process of apoptosis that the cells are visibly fragmenting. TF-1 cells undergoing factor    85 withdrawal for 18-24 hours begin to show microscopically apparent cell death – cell shrinkage and obvious fragmentation.  3.2.1 Flow Cytometry Analysis of Apoptosis in TF-1 Cells  Figure 3.1 shows the results of analysis of TF-1 cells stained with propidium iodide, and analyzed using flow cytometry. Control cells show about 4% sub-diploid DNA, while cells undergoing cytokine withdrawal for 18 and 24 hours show 28% and 36% sub-diploid DNA respectively. This work was done in conjunction with the caspase activity assay (below), to determine an appropriate time-point for investigation.   86    F ig u r e  3 .1  F lo w  c y to m e tr y  o f p r o p id iu m  i o d id e -s ta in e d  T F -1  c e ll s.  C o n tr o l v e r s u s  1 8 - an d  2 4 -h o u rs  c y to k in e w it h d ra w al . R ep re se n ta ti v e o f 3  e x p er im en ts . C y to k in e w it h d ra w al  c au se d  c el l d ea th , in d ic at ed  b y  t h e p re se n ce  o f su b -G 1  D N A , as  w el l as  l iv e ce ll  a cc u m u la ti o n  i n  G 1 .  Cell Count Cell Count Cell Count      87 3.2.2 Caspase Activity Analysis in TF-1 Cells  Caspase activity was visualized by using a colorimetric assay that is dependent on the activation of caspase-3 – the Ac-DEVD-pNA caspase-3 activity assay. Caspase-3 is the most abundant caspase in the cell, and is the one ultimately responsible for most of the effects of apoptosis (although caspase-6 and caspase-7 also play important roles) (Zimmermann, Bonzon, & Green, 2001). Activation of caspase-3 is considered by many to be a hallmark of early apoptosis. Relative levels of caspase-3 activation were determined after 9, 12, 15 and 18 hours of cytokine withdrawal, and compared with control cells – in the presence of normal amounts of cytokine added, as a negative control; Staurosporine (1 !M for 3 hours) was added to induce apoptosis as a positive control (1 µM staurosporine for three hours). Results (shown in Figure 3.2) indicate that by 12 hours, levels of activated caspase-3 were twice baseline levels (52.41 +/- 0.22 versus 25.77 +/- 0.06), and further increased by 15 hours (61.4 +/- 9.28). The apparent decrease at 18 hours (55.14 +/- 10.74) may be attributable to the fact that almost 30% of these cells have begun to fragment (based on flow cytometry data). At 15 hours, almost all the cells still looked relatively normal microscopically, although the cells did appear to be somewhat smaller in size than control cells. It was thus confirmed that caspase-3 was activated at 15 hours, and so this was the time point chosen for investigation.             88       Figure 3.2 DEVD-pNA Caspase Activity Assay. Upper panel shows Caspase 3 activity in response to cytokine withdrawal. Lower panel shows Caspase 3 activity in respones to staurosporine. Note difference in scales used. TF-1 cells were washed 3 times in PBS, then incubated for the time indicated in medium in the absence of GM- CSF, or in the presence of 1 !M staurosporine (Sts) for 3 hours. Control cells were kept in the presence of complete medium.  Caspase activity was determined as described in Chapter 2.     89 3.2.3 Determination of Concentrations of Arginine and Lysine Required for Cell Growth in RPMI-1640  In preparation for cell metabolic labelling with stable isotopes, cells were tested at different concentrations of arginine and lysine – 0%, 10%, 25%, 50%, 75% and 100%. This was done to determine a minimal level of amino acids that still allow normal proliferation of cells, in order that the labelling media might be prepared at a lower concentration of 13 C-labelled arginine and lysine, which would reduce the expense involved, yet not interfere with cell growth. Cells survived, and appeared to grow normally at concentrations of arginine and lysine at 25% of normal levels found in the growth medium (RMPI 1640), hence it was decided that arginine and lysine at 50% of normal concentration would allow an adequate supply of these essential amino acids for cell metabolism, and yet still afford substantial savings in the quantities of labeled amino acids used. A further advantage of using reduced concentrations of arginine and lysine is that at reduced concentrations, any conversion of arginine to proline is reduced as well.  3.2.4 Incorporation of  13 C-Labelled Amino Acids  A test was performed to determine levels of incorporation of 13 C-labelled amino acids in cellular proteins. A sample of whole cell lysate from 13 C-labelled TF-1 cells was analyzed using the FT-ICR mass spectrometer. Results indicated that over 95% of the peptides detected had been labelled with 13 C amino acids    90 3.2.5 Optimization of Sample Fractionation Conditions  The fractionation of whole cell lysates was an important step in our experimental design. Since we had decided to use whole cell lysates, careful attention would be required for “pre- fractionation” of the lysates prior to analysis using the mass spectrometer. Some preliminary work was carried out using a quadrupole time-of-flight mass spectrometer (Q-Star, Applied Biosystems) to analyze whole cell lysates in control cells compared with cells undergoing cytokine withdrawal for six hours and fifteen hours. Two different types of sample fractionation were tested – strong anion exchange (SAX = Mono-Q), and SDS-PAGE. 3.2.5.1 Strong Anion Exchange Chromatography Separation  Whole cell lysate from cells that had undergone six-hours of cytokine withdrawal was separated by strong anion exchange chromatography (SAX). Control and labelled cells were counted, using approximately 42 x 10 6  of each for approximately 84 x 10 6  cells total. Cells were lysed in 500 µL of solubilization buffer, yielding about 8.0 mg of protein. The sample was loaded onto a 5 mL Mono-Q column, and separated over 30 minutes using a step-wise gradient (150mM, 200 mM, 300 mM, 400 mM and 600 mM sodium chloride).  Thirty fractions of one millilitre were collected. Proteins were precipitated using a methanol- chloroform precipitation protocol and the samples dried. Some samples were combined, based on estimates of the amount of material in each of the thirty samples, to give a total of eighteen samples with approximately equal quantities of material in each. Samples were subjected to in-solution trypsin digestion and the peptides run (in eighteen separate runs) via a reverse-phase column into the Q-Star for mass analysis. A total of 997 proteins were identified, of which 436 or 44% were identified based on a single peptide. The number of   91 proteins with a relative quantitation ratio of < 0.67 (i.e. decreased 30% or more in response to starvation), and with identification based on greater than one peptide was 9. The number of proteins with a relative quantitation ratio of > 1.3 (i.e. increased 30% or more in response to starvation), and with identification based on greater than one peptide was 3.    92 3.2.5.2 SDS-PAGE Separation  While the approximate load capacity of the strong anion exchange column I proposed to use was known, the question of how much protein could be loaded in a gel lane had to be addressed. The need to maximize the amount of material separated on the gel – to give the best possible chance for detection of as many proteins as possible - needed to be weighed against the loss of protein separation that occurs when an SDS-PAGE gel is overloaded with material. Initially this question was posed using the mini-gel format (gel dimensions approximately 8.5cm x 6 .5 cm, BioRad Mini-Protean II). Whole cell lysate from TF-1 cells was prepared as described in Chapter 2. Samples were loaded on a 1.5 mm thickness 9% SDS-PAGE mini-gel in the amounts described under each lane in Figure 3.3 (!g protein). As will be noted, it appeared that band resolution started to become less distinct at an 80-!g total protein load. From this it was determined that a load of 70-!g protein per well would be used.          93   Figure 3.3 Protein load (µg) of whole cell lysate and bovine serum (BSA) Comparison was made between a simple protein mix and a complex protein mix to determine maximum protein load per lane. Protean II mini-gels shown. Images show increasing protein loads, and the comparison between a complex protein mixture (whole cell lysate) and a simple protein mix (BSA). Different protein loads were tested to determine how much material could be loaded before protein separation became less distinct – determined by the visual appearance (i.e. blurring) of the Coomassie! R250-stained protein bands.    Once the maximum protein load had been determined, the second approach to sample fractionation was tested using whole cell lysate from cells that had undergone fifteen-hours of cytokine withdrawal. In this instance, the sample was separated by using polyacrylamide gel electrophoresis. Two hundred micrograms of protein (100 µg from each of the labeled conditions) was split into three aliquots and each aliquot loaded into one of three lanes of a 1.5mm thick, 9% polyacrylamide gel. The gel was cut (across all three lanes) into 18 vertical slices, giving 18 samples for in-gel trypsin digestion and peptide extraction. A total of 616 proteins were identified, of which 222 or 36% were identified with only a single peptide. The number of proteins with a relative quantitation ratio of < 0.67 (i.e. decreased 30% or more in response to starvation), and with identification based on greater than one peptide was 4. The number of proteins with a relative quantitation ratio of  > 1.3 (i.e. increased 30% or more in response to starvation), and with identification based on greater than one peptide was 6.   94  Based on the results of these two experiments, it was apparent that with the current experimental design, and using the technology available at the time, the results detected were not significant enough to warrant further investigation. The total number of proteins identified and quantitated by more than one peptide was disappointingly small. In all cases, it appeared that the proteins identified were highly abundant proteins, and so less likely to be involved in the survival signaling I wished to investigate.  3.2.6 Fourier-Transform Ion Cyclotron Resonance Mass Spectrometry Of TF-1 Proteins  The acquisition of a Fourier-transform ion-cyclotron resonance mass spectrometer (FT-ICR) gave hope that the question of what proteins change in concentration in early apoptosis in response to cytokine withdrawal might be answered in more detail. For the initial FT-ICR experiment a 12% mini-gel was used, with the same protein load (70 µg per lane, in three lanes). However, for this initial FT-ICR experiment, the gel was cut into forty slices – in an attempt to gain further pre-fractionation of the whole cell lysate. This experiment was designed in such a way that samples would be run on both the Q-Star and the FT-ICR in parallel, for a comparison of both instruments under these experimental conditions.  In brief, whole cell lysate was fractionated using one-dimensional polyacrylamide gel electrophoresis in a 12% polyacrylamide mini-gel. Two sets of three lanes were loaded with 70-µg total protein per lane. After separation, the gel was cut into forty horizontal slices across the six lanes. Each horizontal slice (six lanes) was considered as one sample, and was   95 divided into half, giving two sets of three lane’s worth of material – three lane’s worth for each mass spectrometer, and a total of forty samples for each mass spectrometer. An image of the gel, showing the approximate position of the slices, is shown as Figure 3.4.   Figure 3.4 A 12% mini-gel was used as the 40-slice gel in the first experiment. Scan of stained gel used for the first experiment. Values for the molecular weight markers are shown on the right side. Approximate cut points for the 40 fractions (slices) are shown on the left. The six lanes are identical – each loaded with 70-µg (total) protein. All six lanes were used for analysis. See text for details.  Early results showed promise using the FT-ICR. Analysis of nine of the forty fractions showed that by using the Q-Star I was able to identify 321 proteins, however, by using the FT-ICR I was able to identify over 1,000 proteins with approximately 75% being identified   96 using more than one peptide. This difference was such that it was decided not to continue with the Q-Star analysis, and to concentrate efforts on analysis of data the FT-ICR.  On the basis of the results from the earlier mass spectrometry experiments, and in discussion with other researchers who had had more experience with SILAC, it was decided to try a large format gradient gel, in an attempt to “drill deeper” into the proteome by using greater protein separation and further increasing the number of slices into which the lysate was fractionated. The large format gradient gel was used  (see Materials & Methods – Chapter 2). For this large format gel, the well size was calculated to be approximately 20mm x 22 mm x 1.5 mm for a total volume of 660 mm 3 . Up to 500 !L could be loaded comfortably into a well of this volume. Using TF-1 whole cell lysate, a series of protein loads was tested - in this case 400-, 450-, 500-, 550-, 600-, 650-, 750- and 850-!g protein was loaded into wells.  The results of the protein load tests indicated that acceptable resolution was maintained up to at least 600-!g protein load per well, but that beyond 600-!g protein load, the resolution appeared to diminish, with the visible bands becoming more blurred. It was determined that for the proposed series of experiments, the maximum protein load under these gel conditions would be 600-!g per well.  The question arose as to whether there would be sufficient sodium dodecylsulfate (SDS) contained in the loading buffer to bind adequately to all the protein contained in the sample (inadequate SDS could be one explanation for the blurring observed at higher protein loads). Sample protein was determined to be 5.2 µg/µL, and so it was calculated that the ratio of   97 SDS to protein in sample prepared above was 2.6:1 – well above the 1.4:1 ratio required for complete binding of SDS to polypeptide. The blurring observed at the higher protein loads was interpreted as indicating that the gel was overloaded.  For the second biological replicate using the FT-ICR, whole cell lysate was fractionated using a large-format polyacrylamide gradient gel, with gradient from 8%-16.5% acrylamide, prepared in-house. Two lanes of the large format gel were loaded with 600-µg total protein in each lane. Two lanes were loaded so that one could kept as a “back-up”, for use if necessary. After separation, the gel lane was cut into ninety horizontal slices, giving ninety samples. The gel was cut into ninety slices in order that no individual slice would contain too large a volume of gel, which might cause technical difficulties with the in-gel trypsin digestion and peptide extraction later in the sample handling. An image of the second gel and the approximate position of the slices is shown as Figure 3.5.         98   Figure 3.5 An 8% - 16.5% large format gel was used as the 90-slice gel for the second experiment. Approximate cut points for the 90 slices are shown on the image. Two lanes of a large format gel were loaded with 600-µg (total) protein in each lane. Two lanes were loaded so that one could kept as a “back-up”, for use if necessary. Only the contents of the right-hand lane were used for mass spectrometry analysis.   The third biological replicate of the FT-ICR experiment also used a large format gradient gel. In this case the gel was cut into seventy horizontal slices. Cutting the gel into seventy slices was chosen as a compromise between the enhanced numbers of proteins identified by cutting the gel into more slices, and the instrument time required to run the number of samples on the mass spectrometer. An image of the third gel and the approximate position of the slices is shown as Figure 3.6.     99  Figure 3.6 An 8%-16.5% large format gradient gel was used for the 70-slice gel for the third experiment. Approximate cut points for the 70 slices are shown on the image. Two lanes of a large format gel were loaded with 600-µg (total) protein in each lane. Two lanes were loaded so that one could kept as a “back-up”, for use if necessary. Only the contents of the right-hand lane were used for mass spectrometry analysis.    Gel slices from each of the three experiments were treated in the same way: slices were subjected to in-gel trypsin digestion and peptide handling as described in Chapter 2. Samples were separated using HPLC, on a reverse-phase chromatography column as a front end to the FT-ICR mass spectrometer. The final number of proteins identified and quantitated is shown in Table 3.1. These numbers represent the final totals, after data handling and processing as outlined below, and described in detail in the data handling section of Chapter 2. In table 3.1, protein identifications are based on two or more high confidence unique peptides, that is, the   100 number of peptides shown in the table refers to the actual number of peptides used in the identification of the proteins shown.   Experiment #1 #2 #3 Number of slices 40 90 70 Number of proteins identified 1987 2186 2149 Number of peptides used in protein identification 24425 27067 27784  Table 3.1 The number of proteins identified in each experiment, and the number of peptides used to identify these proteins. See text in “Materials & Methods” for details of data handling used to identify these numbers.   As can be observed from the table, the number of proteins identified follows the same pattern as the number of slices used in each experiment, however the differences in number identified are not as great as might be expected – only about a 10% increase in the number of proteins identified, from an approximately 125% increase in the number of slices used between experiments #1 and #2. It should be noted that the protein and peptide counts have been made after the following data handling procedures, which are covered in detail in Chapter 2:   101 i. removal of redundant peptides – only unique peptides are used in the analysis ii. Mascot score set to 25 iii. assignment of unique peptides to proteins in the most efficient manner by the Perl® script “finalList.pl”, which functions to find the smallest number of protein sequences that can be used to explain all the peptide data (see Chapter 2) iv. removal of protein entries that refer to keratin contamination v. removal of protein entries identified using only single peptides vi. removal of protein entries identified by only two peptides, where each peptides was not greater than seven amino acid residues in length vii. removal of protein entries where quantitation was based on artifact (noise), as determined by manual inspection of the spectra. These proteins were usually associated with ratios that were quite different from the normal range of ratios seen in this experimental design (for example, 20:1, or 1:16). For all proteins with “outlying” ratios, manual inspection of the spectra of all peptides used to identify the “outlying” proteins was undertaken. Spectra that contained obviously overlapping peaks were excluded from the ratio determination, and spectra   102 in which the signal-to-noise ratio was less than 3:1 were excluded from the ratio determination. viii. For any protein that was to be investigated in more detail (for example by immunoblotting) the spectra were reviewed manually to ensure that the quality of the mass spectrometry data was sufficient to warrant more detailed investigation. 3.3 Proteins Showing Consistent Changes in Concentration in Response to Cytokine Withdrawal  The total number of proteins identified and quantitated in common in all three experiments, using two or more high confidence unique peptides, was 1451. Proteins showing important changes in concentration in response to cytokine withdrawal are presented in detail in appendices 1 and 2, and discussed below and in Chapters 4 and 5.  In order to determine the level of change in protein concentration that might be important, I used the average change in protein concentration, and the standard deviation of the changes in concentration across all three experiments. From these I calculated concentration changes at different levels of “significance” - from a change that might occur by chance in 5% of instances to a change that might occur by chance in only 0.5% of instances. These calculations are shown below. The average of the mean i/w ratios from the three experiments was 0.99, with the standard deviation of 0.18. Using these figures, the following table (Table 3.2) shows the i/w ratios both greater than and less than 1.0 (no change in concentration) that are determined by the calculations above. Hence the i/w ratios shown represent a decrease and an increase in protein concentration for each of several levels of “confidence” from 95%   103 improbable by chance to 99.5% improbable by chance. The table also shows the number of proteins identified and quantitated at each of the levels.   104  i/w ratio cutoffs 13 C/ 12 C ratio Number of proteins identified and quantitated* Increased concentration 0.5% 0.54 1 1.0% 0.58 3 2.5% 0.65 13 5.0% 0.70 42  Decreased concentration 0.5% 1.46 3 1.0% 1.42 27 2.5% 1.35 50 5.0% 1.30 82  Table 3.2 Relative 13 C/ 12 C ratios at various likelihoods of chance. Probabilities that the change reported occurred at random; along with the number of proteins identified at each level. * numbers are cumulative  – that is, the numbers shown at the extremes include the proteins at less extreme ratios.  3.3.1 Results  Two tables of results are presented as appendices. These tables show details of the proteins of interest – proteins identified with a change in relative quantitation of up to 30% (increase or decrease) in starving cells compared with control cells. This represents all results at the level of 5% chance of occurring at random. Ratios are expressed as isotype/wild-type (i/w), or 13 C/ 12 C, that is, control/starving. The proteins identified in the tables have been found in all three experiments. In the tables, the column marked “avg (3) corrected i/w” shows the   105 average 13 C/ 12 C (i.e. control/starving) ratio for each protein in the three experiments. The tables were felt to be too large to be included with the text of this chapter, and so they are included as appendices, however, a list containing only the protein names and relative quantitation ratios is presented here. The list of 42 proteins shown to increase in response to cytokine withdrawal is given as Table 3.3 and the list of 82 proteins shown to decrease is given as Table 3.4. Appendix 1 contains the details of the proteins that showed an increase in concentration of 30% or more in response to cytokine withdrawal for fifteen hours, while Appendix 2 contains the details of the proteins determined to decrease in concentration by 30% or more in response to cytokine withdrawal.     106 Table 3.3 The list of proteins observed to increase in response to cytokine withdrawal. The i/w ratio is the inverse of the change in concentration.    107   ID Avg (3) corrected i/w CBX3; LOC653972 Chromobox protein homolog 3  0.52 PSMD10 26S proteasome non-ATPase regulatory subunit 10  0.55 HLTF Isoform 1 of Helicase-like transcription factor  0.56 HIST1H1B Histone H1.5  0.61 ELOVL1 Elongation of very long chain fatty acids protein 1  0.61 CDC42 Isoform 2 of Cell division control protein 42 homolog precursor  0.61 PDCD4 Programmed cell death protein 4  0.62 FECH ferrochelatase isoform a precursor  0.62 FUS Isoform Short of RNA-binding protein FUS  0.62 MGC4172 Isoform 1 of Dehydrogenase/reductase SDR family member 11 precursor 0.63 H3F3A;H3F3B Histone H3.3  0.63 ACSM3 Isoform 1 of Acyl-coenzyme A synthetase ACSM3  mitochondrial precursor 0.63 METTL7A Methyltransferase-like protein 7A precursor  0.65 ATP6V1H Isoform 2 of Vacuolar proton pump subunit H  0.65 52 kDa protein  0.66 ARHGDIB Rho GDP-dissociation inhibitor 2  0.66 HIST1H1D Histone H1.3  0.66 HIST1H1C Histone H1.2  0.66 NNT NAD(P) transhydrogenase mitochondrial precursor  0.67 C2orf47 Uncharacterized protein C2orf47  mitochondrial precursor  0.67 HMGB2 High mobility group protein B2  0.67 TXNRD1 Isoform 4 of Thioredoxin reductase 1  cytoplasmic  0.68 HDGF Hepatoma-derived growth factor  0.68 PSMF1 Proteasome inhibitor PI31 subunit  0.69 NME6 nucleoside diphosphate kinase type 6  0.69 GRPEL1 GrpE protein homolog 1  mitochondrial precursor  0.69 MRPL22 Isoform 1 of 39S ribosomal protein L22  mitochondrial precursor  0.69 ADH5 Alcohol dehydrogenase class-3  0.69 EXOSC8 Exosome complex exonuclease RRP43  0.69 FKBP3 FK506-binding protein 3  0.69 MT-CO2 Cytochrome c oxidase subunit 2  0.69 HCFC1 Isoform 1 of Host cell factor  0.69 RSL1D1 RSL1D1 protein  0.69 30 kDa protein  0.69 ALAD delta-aminolevulinic acid dehydratase isoform b  0.70 PDHB Isoform 1 of Pyruvate dehydrogenase E1 component subunit beta mitochondrial precursor 0.70 HMGB1 High mobility group protein B1  0.70 ANXA7 Isoform 1 of Annexin A7  0.70 CDC37 Hsp90 co-chaperone Cdc37  0.70 ATPAF2 ATP synthase mitochondrial F1 complex assembly factor 2 mitochondrial precursor 0.70 TMPO Lamina-associated polypeptide 2 isoform alpha  0.70 TAF15 Isoform Short of TATA-binding protein-associated factor 2N  0.70   108 Table 3.4 The list of proteins observed to decrease in response to cytokine withdrawal. The i/w ratio is the inverse of the chang