`
`
`
`
`observation that matrix effects are the dominant factor in peptide
`detectibility, and that there is a linear range for good peptide
`C§l.ial‘i’l'll.'aElOl‘..[l36]
`Usually, one tries to inject the same total amount of protein.
`Conceptually, this would work ifthere were only small changes be-
`tween samples, but how well would it workifthere were changes in
`abundant proteins? One solution is to use internal referen 'e stan-
`dards for alignment or for normalizing the different rLins.ll37'l38l
`l\/lass-spectrometry based guantitation poses significant statisti-
`cal challenges. Because ofthe high cost peranalysis, and the often
`limited amounts ofsample,there are very few studies that have ad--
`dressed the issues of biological variability (samples from different
`patients or animals), and technical variability (the same biological
`sample, split and processed independently), and experimental
`variability (different analyses of the sarne processed sample). No—
`table exceptions are the papers by Gan em}. for iTRAQ”3"l and
`a recent paper by Li e3tai.W’Gl on label—free guantitation, where
`the authors o'eveloped a method to determine statistical signifi»
`cance and false positives usir-.g AiVll'-based lahel~free data. A ‘false
`positive’ defined as a misassignment of differential expression.
`These authors also discuss the challenges of performing statistics
`on label—free analysis, as well as the added difficulties of per-
`forming statistical analysis on peptide-based guantitation data in
`general, because different numbers measurements are made
`on different proteins, which is not the cas
`for microarray data. A
`fold-change cutoff was not found to be sufficient — an additional
`statistical test, performed at the peptide level, was found to be
`necessan/lwl This method, however, was still not sufficient to de-
`termine the false discovery rate (FDR) and statistical significance
`of relative expression data from label--free experiments. Although
`the authors recognize that often only one analysis per biological
`sample is normally performed,they found that the minimum num-
`ber ofanalyses required obtaining these statistics was two LC,/IVES
`analyses (ie. two experimental replicates), spiked with the same
`level of 15 l\i-labeled internal standard. By duantitating the labeled
`and unlabeled sample separately, they were able to produce four
`possible pairings.Three parameters were used to determine differ-
`ential expression: fold—change, the 1‘-test orwilcoxon ranksum test,
`and a minimum number of permuted statistical pairings (ii./iPSPs).
`Using the internal standard as the control, unlabeled protein
`found to he differentially labeled was considered to be a posi-
`tive, while the labeled internal standard found to be differentially
`labeled, was considered to be a false positive. interestingly, at a
`confidence level of 95%, a critical fold-change was below
`which there was a drop in the number
`positives, while the
`
`number of false positives stayed constant which was dependent
`on the number ofanalyses (MPSPS).
`.'his critical fold-change was
`2.75,. 2.5, 2.5 and 2.9 for :'v'lPSPs of l, 2, 3 and 4, respectively, which
`corresponded to FDRs of22, 'l S, 8.7 and 4.2%.V4"l
`
`Metabolic versus noimmetabolic labeling
`
`applied to any source of biological
`labeling can
`Chemical
`lvletabolic labeling using SlLAC can be used for cell
`material.
`culture, but
`is not effective for au-totropl‘.ic organisms such
`as plants or bacteria. For these autotrophic organisms.
`l5i\l
`labeling is preferred. Sll_AC typically works well for mammalian
`cell
`lines, which do not
`all of the amino acids, and
`so incorporation of the labeled amino acids from the growth
`medium is more cornpreherisive.l‘l7l
`Sll_AC has been used for one study involving plant cell culture (A.
`rhaiiana), and although the average incorporation ofl3C5-arginine
`
`was only 75%, it allowed the study of differential expression of
`glutathione S~transferase ir-. response to sialic acid treatment.l55l
`To our knowledge, this is the only report about the use ofSlLAC
`plant proteornics. in contrast, virtually complete labeling (§5% and
`higher) of proteins in both suspension cultures and entire plants
`of A. tnaiiarra has been attained using ‘SN isotopes.l5_
`There have been a few reports on stable isotope labeled
`protein guantitation
`‘unusual’ organisms. Drosophila was also
`the first multicellular model organism subjected to labeling
`with ‘SN — Heck arid collaborators labeled D. meionogaster and
`C. eiegarrs with "5 l\l.l53l Proteornic studies in Drosophila are rare, and
`only a few guantitative proteomics studies have been performed.
`Aebersold
`co-worlters have used 4-vplex iTF{AQ
`protein
`--1:.
`phosphatase treatment for specific substrates in Drosophila cell
`lines.‘-'
`Our laboratory was involved in art ETRAQ study on
`!.eishrno'r7io',
`in which 21% of the proteome was identified and
`guantified over seven tirriepoints.l“’3l Siuzdak and co—workers
`used stable isotope labeling to monitor the expression kinetics of
`viral proteins, changes in the expression levels of cellular proteins,
`and fluctuations ir-. metabolites in response to l-"lock House Virus
`(Fl-l‘v",i viral infection.ll4‘”
`Yates and collaborators applied the l5l\l metabolic labeling
`technique to fiatttis riorvegicus by feeding them a l5l\l-enriched
`diet. The strategy was employed to generate internal standards to
`quantify proteins in mammalian tissues. This worl< provided the
`proof-of--principle that metabolic labeling of whole organisms is
`feasible in mammals,as had already been demonstrated for worms
`and flies, and opened up new possibilities for similar applications.
`Mann and co—worl<ers have established protocols forthe Slt.AC—
`labeling of rnice.ll‘l5l Labeling ofwhole animals is based on a special
`diet containing either the natural or the "'3C5-vsubstitiited version
`of lysine. Labeling was carried out over four generations, with no
`effect on development, growth or behavior. Full incorporation of
`Sll_AC amino acids was achieved for all organs in the F2 generation
`animals. However, metabolic labeling strategies for animals are
`often impractical, due to the high cost ofthe diet and the long time
`required for labeling (full incorporation is typically not achieved i
`.
`the first or ever-. the second generation ofanimals).
`Because
`if the high cost of isotopically labeled materials,
`metabolic labeling studies tend to be used for pathway determina—
`tion. iTl'iAQ and other chemical-based isotopic-labeling methods
`are used for biomarker discovery, and :'v’iP.i\/i methods are used for
`biomarker verification or validation.
`
`Label-free methods for biomarker discoveryare currently receiv-
`ing a lot ofattention, because oftheir simplicityand low cost. How—
`ever,the lackoflabeled internal standards makes them susceptible
`to suppression effects from other components in the sample. Of
`course, at the ‘discovery stage’ one cannot add internal standards
`because one does not yet know what standards you will need. This
`conundrum reflects the current status of protein quantitation.
`Also, attl ris poir-.t,there is no single method that will identifyand
`duantitate all of the proteins
`the sample ~ different techniques
`will find different proteins. Several studies illustrate this point. in
`the first example, a study of insect salivary gland extracts, the
`iTl'-RAQ technique identified 43 proteins not observed using the
`LC/i‘v’iS/'l‘vlS analysis of salivary gland extracts from insects of the
`same age. This result is consistent with the previous observations
`that betterfragmentation is obtained using this teclinology, giving
`more peptides per protein arid allowing the identification of less~
`abundant proteins. ETRAQ labeling led to the identification of 78
`proteins, 39 ofwhich were not identified byin a standard LC./'i\/iS/MS
`
`J, /l/lass. Spectrom. 203$, 44,
`
`7 — i 660
`
`Copyright
`
`2009 John Wiley & Sons, Ltd,
`
`www.interscience.wiiey.com/journal/inns
`
`0000051
`
`
`
`
`iiof-
`ivi. i-i. Eliiott etai.
`
`
`
`(rs;
`
`
`
`Labeiniree
`
`(A)
`
`Comparison of ingte) ratios Expression to iTRAQ
`
`ETRAQ
`
`
`
`Laoei—iree
`
`Figure 16. A). The correlation "between the iTRAQ resnits and the label»-free results, 8). Venn tliagram of proteins identified by the three techniques. (J).
`Venn diagram of proteins identified by the three techniques, requiring at ieast 2 peptides for an identification. Reprinted from [M6], with pernrission.
`
`
`
`as 5
`
`
` eeeee
`
`eeeeeeeeeeee
`
`eeeeeeeeeeee
`
`r-'.~‘o'.{w
`
`-‘.c\~;\\\'ss.
`
`high—corrlio'ence proteins identified by i\/‘..L\SCCiT in each of the five labeling experiments. i\IiS./MS
`Figure ‘i7.The total nurnbei
`against E. coir‘ database. The ‘l :25 ratio gave anomalous results for the ECPL iabeling, so the ‘l :25 ratio is not inciuded.
`
`were searched
`
`anaiysis, iiiustrating the vaiue of using both two technologies in
`parailei for rnaxirnum proteome coverage.
`
`Comparison of methods
`
`in a very recent comparison of iTR/\Q, label-free (ion accoi,-rating‘).
`and geLC by the Patei etai.,.”‘l5l the expression ratios were higher
`for the ia'pei—free analyses than for iTRA-Q, as had been noted
`in other studies (note the siope of the iine). The "Venn diagram
`in Fig. 145 shows the number of proteins identified by the three
`techniques (incitiding identifications based on a singie peptide).
`it
`clear from these studies that, at this point, there
`no one
`technique that can qtiantitate w or even detect - every protein.
`in a recent study in our laboratory, five of the most cornrnon
`labeling TeCl‘ii’iiC§LieS- lCPt__, clCAT (cieavable KAT},
`iTRAQ, "30,
`and acetyiatior‘. mwere compared on air‘. E. coii tryptic digest to
`determine the method that identifies
`highest number of
`proteins and provides the most acc-.ir'ate quantitatiorr.
`in this
`study, the highest n=.irnToer of proteins was identified with the
`iTRA-Q iabelirig system, foiiowed by lCPi.. The peptides in these
`two methods. however, were separated by 2D--LC unlike the other
`experiments which were done using ii)-LC, thus demonstratirig
`the advantages for prefractionation of peptides in complex
`sarnpies. The three other laheiing systems (lat), acetyiation and
`
`clCAT) resuited in approxirnateiy the same number of proteir‘.
`identifications (Fig. 17).
`Peptides labeled at i :1 and i :3 ratios with CECAT, iTRAQ and
`acetyiation were quantified with reasonable accuracy. i-iowever,
`only the highest-confidence proteins in i'l'RAQ~iabeied sampies
`resuited in an acceptahie arnount of variation when iabeied at a
`ratio of 1:10 (Fig. 18). We were unabie to anaiyze the ‘SO and
`ICPL data as we couid not find or modify any of our software
`to accept these iabels with
`O_Star data flies. The variation
`observed in these experiments ciearly demonstrates the need for
`both technical and hioiogicai replicates.
`The advantages and disadvantages of each procedure, as found
`in our study, are compared in Table i.
`There have been several other corriparisons of different lahei
`and iabel-free methods, where the same samples were anaiyzed
`through various quantitation techniques. The results of these
`cornparisons are shown in given in the tabie in the Supporting
`information.
`in ETRAQ, where the protein identification is done
`on
`same set of
`ialoeled peptides as the quantitation,
`there is an inverse relationship between the confidence
`the
`identification and the nurnher
`proteins on which Q-.ianti‘tati\/e
`data can be obtained.
`in a set of ten experiments on the
`reprodu-zibiiity of iTRr’\Q analyses, Gan etai. found that allowing
`a :4}.-50% change in expression ratio between bio!ogic'ai repiicates
`
`
`J. A/iass. Spectrom 2009, =44, 1637-1660
`Copyright
`2009 John Wiley & Sons, Ltd.
`www.interscience.wiiey.com[iournai/jms
`
`
`
`0000052
`
`
`
`O_uantitative proteomics
`
`
`
`
`0
`
`20
`
`40
`
`60
`
`80
`
`100
`
`0
`
`1 O0
`
`200
`
`300
`
`400
`
`0
`
`40
`
`80
`
`1 20
`
`coil’ proteins at i :1 ratios were calculated by F‘roteinPiiot (ICAT,
`Figure 18. The experimental average ratio of ciCA‘i (A), iTRAC2 (B), and acetyl (C) iabeiecl
`iTRAQ) and MSQUANT tacetylit The experirnerital average ratio o‘l‘ciC/-\T (D), iTRAQ (E) and acetyl (F) iabeleol E. coli proteins at i :3 ratios were calculated
`by ProteinPilot (ECAT, ETRAQ) and i\/lSQUAi\iT (acetyl). The experimentai average ratio of CECAT (3), iTP.AO_ (H) and acetyl (E) iabeied E. coil’ proteins at i :10
`ratios were caicuiated by Pr'oteinPilot (ECAT, URAQ) and i\/LSQLJ Ai\iT -(acetyi).
`
`produced by peak area intensity measurements compared to
`spectrai counting is the restriction by different software pack-
`ages on the number of peptides required for a protein to be
`considered ’detected’.i99] A requirement for a larger number
`of peptides per protein discriminates against lower~abundance
`proteins, thus removing the larger expression differences. The
`larger the difference
`abundance ratios,
`the more reliapiy
`this difference couid be detected through lahei-free techniques.
`Ti‘-.e study by Liu etai. reported that the ntrrnber of spectra
`produced was a reiiable indication of expression ratio ii‘ the
`concentration difference was :>5.l93i Other factors inch,-de the
`
`size of the protein, the number of tn/ptic cieavage sites and
`the amount of protein that can be ioaded onto a capiliary LC
`cpiurnn.ll49]
`
`resulted in 88% protein coverage, ailowing a :_t:30% change in
`expression ratio between technicai replicates resuited in 95%
`protein coverage, with oniy 3, iO.‘l% variance coming from the
`i\."iS.ll38] Liu eta.l.i98] found that the correiatiori of abundance
`with the r-.um'per of spectra observed, was better than that
`based on % sequence coverage or the number of peptides
`identified per protein.
`in a comparison of spectrai counting
`versus peptide ion intensities, Xia e3ta:.W’7l found that spectral
`counting gave better agreement with the true protein ratios.
`A recent comparison of studies using the spectrai counting
`and ion intensity-based methods
`label—free quantitation, with
`respect to dynamic range of quantitation and dynamic range of
`protein detection was done by Wong eraI.“4ii3 Both methods
`were abie to detect changes in protein levels
`approxirnateiy
`2.5. However, machine learning rnetheds and methods
`peptide ion intensity were computationaliy more difficuit. This
`study conciuded. however, that these iabel--free approaches were
`cornpiernentary, and recornrnended using both for increased
`confidence in the resuits.
`Sorne labei--free methods, however, were found to underes-
`timate expression ratios if the true ratios were :>2.5.“493 One
`reason that has been proposed for the lower range of ratios
`
`The chaiienge of cornparing duantitation methods is shown in
`Fig. i9.m5] in this study, three different methods were compared,
`and all
`three gave different expression rations~which one
`is correct? in this particular‘ study, the authors were able to
`experimentaiiy confirm that the spectrai 'l"iC method (using the
`average MS/i\IiS TEC) was correct, and they attributed the lower
`ratios obtained from SELAC and spectrai counting to compression.
`i-iowever, most studies are performed without
`validation
`
`www.interscienceiwiley.corn/journal/Ems
`
`
`J, /l./i(.'SS. Spectrom. 203$, 44,
`7 — i riot‘.-
`Copyright
`2009 John Wiiey & Sons, Ltd,
`
`0000053
`
`
`
`
`
`i\/it HK Eiiiott grain
`
`
`ittsisttm sttssmtstseee R.st§::- at
`
`estate ?3’§?i€§3§?i$
`
`ate‘
`
`
`'
`
`
`Figure ‘E9. Reiative expression ratios of _o'i'yrosinev—binding proteins using three different quantitation methods. Reprinted from Ref. E125] with permission.
`
`
`
`
`.
`ECPL
`n
`«
`I
`5. mt digest time
`
`-
`HRAQ
`
`'
`.
`.
`-
`.
`-
`Large nuinbei of peptides
`per protein utentr
`-3
`increase corziideriae and
`quantitation statistics
`Mtiitipiex with 4/8 iabeis
`
`:
`5
`:
`‘
`
`crcAT
`
`Advantages
`
`K
`.
`Complex sarnpies are
`simphfieci
`
`quantitatiori can
`reduce cc-stiy
`.
`.
`i
`anaiysrs tirne
`
`.
`
`-
`.
`
`5
`
`.
`
`E
`:
`t
`“
`§
`§
`§
`.
`§
`§
`Lysis buffer contains
`§
`guarzidiite which is
`§
`exceiierit for
`*-
`;_
`.
`.
`..
`.
`§
`so:-.ib:i:zatron
`§
`."/‘iS—based
`i‘\/iS—based quarititation.
`§
`quantitation can
`can reduce costiy
`anaiysis time
`i'€GLit'E‘,‘ costly
`§
`2
`~.
`5
`anaiysrs time
`
`
` : :
`:
`-
`.
`-
`.
`.
`.
`-
`-
`.
`.
`-
`A
`t
`.
`.
`‘
`.
`"
`-
`.r
`-
`.
`K
`.
`-
`.
`Disacivantages Very few peptrcies
`Fractionation required
`Software avariabie oniy
`Long :ncuba—
`Requires
`§
`identined per
`for certain mstruments
`tior‘./ciigestaon
`tractionation
`protein ieads to
`times
`inw -tnnfidence
`and poor statistics
`Not aii proteins
`contain cysteiries
`
`3
`‘*0
`
`I.
`v
`'-
`GLi'EihiG::’1t’:“ can be
`used t-3r
`
`'
`'
`
`_
`Acetyiation
`.
`-
`inexpensive
`
`i’\/iS—basec:'
`quantitation can
`reduce costly
`.
`__.
`analysis iime
`
`Short preparation
`tiine
`
`§
`§
`.
`
`5
`.
`.
`5
`
`§
`Pooled sainpie is only Avaiiabiiity of
`§
`stabie for i
`it
`software for data
`.
`§
`anaiysas
`.
`§
`Acetic anh Idride is a
`iiabiiit I of
`§
`controlled
`soitware tor
`:
`:
`substance
`analysis
`1ttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttt\\~
`
`MS/!V‘:S quantitatiori is costiy
`in terms of MS time
`
`Requires tractionation
`
`Ex ensive
`
`Ex vensive
`
`
`
`At,
`
`of true protein concentrations. Cieariy, more work needs to be
`derie ta validate different quarttitation methcscis using mixtures of
`proteins with known concentration ratiosi
`5
`,
`_
`5
`.
`5
`Aithougn there is general agreement that comparison with
`r-.=r'-. "1
`=1
`*
`1:
`t me‘.
`r
`_
`_ d +
`d
`d
`54
`5
`5“
`_
`5
`id
`4
`_, £98512,
`_
`5
`i
`2’ aha‘ _s“‘n Br mm ‘ G E" S“
`‘ 90
`‘ta ‘ 3 "'
`‘t
`is
`“gt a gmbai Eechmqua H°“"’e"'e“»' mamy Smdies h"-“'9 Comiuded
`that Ehié Z'I"€!'?(j$ observed
`iabei-free EXp€3i'iYT‘!El"!tS
`BFEE vaiid
`Ln
`(Le. overexpression versus underexpres ion) even if the actual
`observed expression ratios rnay be corr1pressed.”4‘**'
`
`Seftware Censiéeratiens
`5
`5
`5
`5
`5
`5
`5
`Berore you start a project, it is rriiportant to me certain that the
`software ‘matches’ the iabei you are trying to use. Certain software
`_
`5
`__
`_
`_
`5
`5
`5
`5
`5
`_
`_.
`_i
`55 5
`.
`.
`-
`r
`.
`.
`.
`packages, for exoimpie, can oniy handie certain types of ianeieo
`amin-:>acicistorSiLAC. eertain companies make proprietary iabeis.
`whitn may only be abie to be aitaiyzed using their own software.
`i-I:Q\I\,'f§‘\[@[" thggrgg
`‘((3 be 5 yveiggme gjhafiggg tgyvard ‘gpggnr-.955’
`now,anci"iTRAC-basses‘ quantitation.forexampiecan be ainaiyzed
`using an iricreasirig number of instruments and softvr/are.
`
`
`vvwwiriterscience.Wiiey.com,{iournai,/jms
`Copyright
`2009 John Wiiey & Sons, Ltd.
`J. /\./iass. Spectrom. 2009, -44, 1637-1660
`
`
`
`0000054
`
`
`
`O_uantitati\.Ie proteomics
`
`3‘N
`
`
`
`.,¥,.
`.\\\.
`
`- -
`-~+~-
`
`LAC
`ETRef:
`£0‘
`is-¥$w‘St°.t°A2t£5\
`tats-::«~.‘Ir~z~a
`
`
`
`
`
`ffiltrtbbf137F"|).-‘nil¢3'!3¢r3.'l2
`
`Kb)
`
`<_t§{r
`
`1(0)
`
`651
`
`'i‘§3'r$l2‘.W'.23')'l
`
`zine
`
`.2l>:.>3
`
`.?i)<\.£
`‘:"¢z9.\‘
`
`.2:>s'i$
`
`em exe
`
`.3926‘
`
`Figure 2(Il.’i'rends in rnass»spectrornetry»- ased guantitation. Publications
`per year, based on a lteyword search in SciFinder Scholar“ Note that
`comparisons over time are probably more accurate than comparisons
`betin/een techniques, due to the difficulties in finding lreyvvords to catch
`every ieierenrge. The number of 2009 publications was extrapolated from
`the number published by August 2009.
`
`Mascot,l94l for example, can perform quantitation of a variety of
`‘precursor’ methods (where the duantitation based on shifts in the
`molecular weight), as well as’l"i\/lTand iTRA’)—ty_oe labels where the
`quantitation is based on l\IlS./ll./lS reporter ions.ll5”l However. addi-
`tional software packages, such as ll./lascot Distiller rnay be required.
`They specifically note that for
`iVlAl_Dl--TOF/TGIF data, there is
`a special software available ( S2i\/lascot) that should be used be—
`cause the standard GPS explorer lvlascot data is de--isotoped.
`labeling may pose a particular problem because every amino acid
`will get labeled, but new software (QuantiSpec‘ has recently been
`written to enable interpretation and quantitation of "'5l\l-labeled
`mass spectra.l"5ll ln general, it is still prudent to make certain to
`select a label that your data~processing software can handle.
`it/lultifi,-nctional so\‘tware packages are being developed to han—
`clle data from label-free and stable-labeled samples, and from a
`variety of instrument platforms. These include the Pr-oteinQuant
`Suite,ll3°l developed by the Novotny group, Census sol‘twarell52l
`developed by the Yates group, the PatternLab software‘-’53l also
`
`
`
`developed by the Yates group for normalizing spectral CQlJl"i‘£ data,
`and the Corra software detleloped by the Aebersold group, which
`in addition, produces protein interaction networks from the dift"er—
`entially expression data.ll’54'l’55l IVES-Biomarlter Discovery Platform
`(lv’lS-BlDl,ll56l frorn the Aebersold group, is designed for r_letermin-
`ing peptides that discriminate between treatment groups. l\Ilarl<--
`erview software (Applied Biosystems) is also designed to facilitate
`detection of biornarker peptides that correlate with treatment.l"57l
`
`ilonclusions
`
`it would have been nice to be able to end this article with a
`
`recommendation for a single method. l-iowever, as is clear from
`the above data and discussion, there is really no one single method
`that will solve all ofthe analytical problems associated with protein
`qtiantitation. This is partly because ’quantitation" means so many
`different things
`global or targeted, absolute or relative.
`The lCPl_ and i'i'RACz methods (from our study) and the ion-
`accounting label—free method (from the Patel study‘) seemed
`to identify and quantify significantly more proteins than the
`other methods in these studies.
`it should be remembered,
`however. that the variability of the enzymatic digestion step
`can affect all of the cherriical labeling te-:hnid_ues and the label-
`free methods by leading to analytical variability. We and others
`are actively exploring solutions to this problem, including the
`
`use of microwave digestion, detergents, pr .5“ -re, and a variety of
`solvents,chaotropic agents and denaturants.ll53l We are confident,
`therefore, that this problem will be able to be solved (or at least
`reduced) in the nearl‘utLire. Another significant source ol‘variability
`comes from depletion steps. This variability, however, can be
`reduced by the stringent use of well--developed SOPs.
`l.abel—tree methods are based on less-rigorous mass spe-:—
`trometry, with more reliance on bioinformatics and separation
`techniques. Factors that have to be considered when selecting a
`method are the number oftreatments, the cost ofthe experiment,
`the complexity of the sample, the biological source of the sample
`and whetherthe experiment will be done in
`ClJltUi’(’:‘. These ulti-
`mately will be the determining factors in choosing the appropriate
`ouantitation method.
`For this review, we used SciFinder Scholarllllgl to count the
`number of publications per year using these various types of
`
`
`
`ta r:.&n§te t§t‘€3&_§
`
`fiémwaar>3?gmaatm
`wt/vw.interscience.wiley.<:om/journal/Ems
`
`Figure 21. -.’)istrilotition of proteins predicted from the $.’w’_.gei/a dysenreriae genome (5-Di , blue), and found by either 20 gel electrophoresis (L~’.»-DE, yellow)
`or LC/i\.'lS,"l‘\/l5 (blue), Reprinted from Ref. [128] with permission.
`
`
`J, A/lass. Spectrum. 203$, 44,
`7 -1660
`Copyright
`2009 John Wiley & Sons, Ltd.
`
`0000055
`
`
`
`
`
`lvi. i-i. Elliott etai.
`
`
`quantitative proteornics (Fig. 20). /-‘xithough it was impossible to
`seiect ‘keywords’ to inciude every reievant publication, it was ciear
`that all of the methods described above are stili in use. Moreover,
`
`with the possibie exception of 5N metabolic labeiing and lCAT,the
`use of most of these methods is increasing each year. in fact, the
`numbers of pubiications using DIGE, iTRAQ. labei-free methods
`are ail increasing at approximately the
`rate -- on the order of
`50% per year.
`in conciusion, even though there is, as yet, no ‘one perfect
`method’, this does not mean that there are no usabie methods. it is
`stili possibie to obtain useful reiative and/or absoiute quantitative
`data by matching the bioiogicai probiem tothe proper‘-o_uanti‘tative
`proteomics approach. Furthermore, mass spectrometers continue
`to be improved with respect to sensitivity, dynamic range, mass
`accuracy and scan rate. We are aiso certain that new rnuitiplex
`labeiing techniqties will continue to be developeciand that better
`SOPs wiii bring improvements in reproducibiiity. This wiii certainiy
`lead to changes and significant improvements in the heid of
`quantitative proteomics.
`We shouid remember, too, that mass spectrometric coverage
`of the proteome stili needs to he irnproved. As can be seen from
`Fig.2'i, mass spectrometry is still not ‘catching’ iarge numbers
`of iower-molecuiar weight proteins which are predicted from
`the genome, and these proteins are underrepresented in current
`proteornics studies. This might be because insufficient nurnbers
`of tryptic peptides of the appropriate size are produced from
`these srnailer proteins, as discussed above. Whether ’top—down’
`techniques or other methods, such as the use of proteotypic
`synthetic peptide standards and l\/ERM techniques, as proposed
`by Aebersold as part of the Peptideittlas project can fill
`gap
`still remains to be seen.
`it is important, however, to remember
`that rnass-spec:rornetry—hased protein cruantitation stili requires
`mass spectrometric peptide or protein detection, and much work
`remains to he done this area as weii.
`
`Acknowiedgements
`
`This work was funded by a platform grant from Genome Canada
`and Genome British Coiumbia.
`
`Supporting information
`
`Supporting information may be found in the online version of this
`article.
`
`Reterences
`
`ii] S. Orig, M. Mann. Mass spectrometry—based proteomics turns
`qiiantitati\Je. Nature Chemicoi .6‘ioiogy 2905, 1,252.
`[2] T. A. Prokhorova, K.T.6.Rigbo!t,. P.T.Johansen, J.l-iehningsen,
`i.Kratchrnaro\Ia,
`ix/l.Kassern, B.Blagoev. Stabie isotope labeling
`by amino acids in ceii cuiture (Sit.A-C.) and quantitative comparison
`oi the membrane proteomes of seir"—renewing and difiei'eritiatirig
`human embryonic stem ceiis. ./V iecuicir and Ceiiuiar Proteomics
`20{39,8,9‘59.
`K, Biemann,
`J. E. Biller,
`F3] A,\/an
`Langenhove, C.E.Costeiio,
`
`T.R. Browne. A gas chromatographic./mas ' ettrornetricmeth-ad
`
`for
`the simiiltaneous quantitation of
`dipltenyihydantoin
`(phenytoin), its para—hyr.lroxylated metaboiite and their stable
`isotope labelied anaiogs. Ciinicr; Chimica Acra, inrermxttionaijournai
`of Clinical (.'hern:’.s?ry 13981, .7 .7/5,. 2675.
`
`i4] SJJ3 skeii, K. Rollins, R.'\/\/. Smith, C, E. Parker. Determination oi
`serum cortisoi by therrnospray liquid chrornatography/rnass
`spectrometry:
`comparison with
`gas
`chromatography/mass
`spectrometry. Biornedicai and Erivironmerimi Ii/lass Spect.rorr;et.ry
`1987, 74, 7l 7.
`
` \\
`
`
`kc
`
`[25]
`
`
`J. /l./iass. Spectrom. 2009, -44, 1637-1660
`Copyright 2009 John Wiiey & Sons, Ltd.
`vvvvW.interscience.Wiiey.com[iournai/jms
`
`E25}
`
`gs'5:
`
`F3.’
`
`'35
`
`E55
`
`[liii
`
`['l 1
`
`H2.
`
`H3.
`
`.a $2»
`
`_«.a P‘.P]
`
`_s NJ
`
`l8]
`
`l9,
`
`ix)5?
`
`21
`
`.
`
`j
`
`A. l.. Yergey, N.V. Esteban, D. J. Liberate. Metaboiic kinetics and
`ouarititative anaiysis by isotope dilution thermospray LC/MS.
`Biorriedica! and Envirorirrientai /lriass Spectr-Jmetry K387, 74, 623.
`J. R. Barr,
`V. i.. Maggio,
`iJ.G...=. i—‘atterson,
`er oi.
`isotope
`dilution—mass spectrometric oyuarititication of speciiic proteins:
`mode! appiication with apoiipoprotein A—i.iE»76—82. Clinical
`a’fherniSr:'ji 1996, 42, 'l 676.
`ivi. iviann. Un'piased quantitative
`i_._l. Foster, C. Lde Hoog,
`proteomics of lipid rafts reveais high specificity for sighaiing
`factors. P:'oc/eedings of ie i‘v'a?ionai A(-‘Jdt?i‘i’.lj/ of .‘§ciences of the
`United States oi‘Americ(.I 2003, 100, 5813.
`N. Kitteringham, R. E.Jeni<ins, (1.5. Lane, V. L. Eiiiett, B. it. Paris.
`iviuitiple reaction monitoring for quantitative. biomarker anaiysis
`in proteomics and rnetaboiornics. Jourrmi of Ch!'Oir1-Ji‘Og‘rc7phy
`B-Anaiyticai Technoiogies in the Biomedical and Life Sciences 2969,
`8/7}’, 1229‘.
`V. Lange, P. Picotti, B, Dornon, Ft. Aebersold. Selected monitoring.
`monitoring for quantitative preteornics: a tutoriai.
`ivloiecuiar
`Systerns Biology 25308, 4,. i .
`S. P.i3ygi.
`S. A. Gerber,
`J. Rush, O.Stemrnan, M. W. Kirschner,
`Abs-aiute quantification of proteins and phosphoproteins from
`ceii lysates by tandem !viS. Proceedings of the National Academy of
`Sciences ofthe United’ States of/imerico 2903, 1100, 694-1).
`D. S. Kiri<patricir., S. A. Gerber, 5. P. Gygi. The absoiute quantifica-
`tion strategyta genera! proce scireforthequantification ofproteins
`(5.
`and post—transiationai mo rticatiorrs.!'»~’,’etiioa's 2095, 35, 265.
`!\Ii.Kuzyi<, 3.Sn”iith,
`J. Yang,
`er 0:‘. MRM-based,
`.'\Iiuitiple:»<ecl,
`A'osoiute Quantitation of 45 proteins in hurnan piasma. ii/ioiecuiar
`and Ceiiuiar Proteorriics 2999, 8, i860.
`C.A.i\/iueiiei‘, W.\/lleinrnann,
`:’~..Dresen, A.Schreiber, r’\/l.Gergov.
`ifieveiopment of a muiti»-target screening anaiysis for 301 drugs
`using a Qirap iiquid chrornatography/tandern mass spectrometry
`system and automated library searching. Rapid Communications
`in Mass Spectrometry 2905, .79, 13.35..
`P.i\/lallicié,
`l\Ii.Schirie, S.S.Chen, etai. Computational prediction
`of proteetypic peptides for quantitative proteornics. Nature
`Biotecimoiogy 2007, .12, 125.
`Pepti-:ieAtias. vi/ww.pei:>tideatiasorg. [Last accessed: 2G09].
`.'vi.itohi, Recllich, i\f..Eisenacher, etai. Autornatecl caicuiation
`oi unique peptide sequences for unambiguous identification oi
`highly homoiogous proteins by mass spectrometry. Journal of
`Prczteomics and Bioinformatrcs 2698, 7, (306.
`W. Tirnm, A. Scherbart,S. Bocker,O. Kohibacher,.T. W. Nattkernper,
`
`Peak intensity prediction in ix/lAL£2-E—TOF .'T
`s spectrometry: A
`
`m .
`ine iearning study to support quantitative proteomi-:5. .‘3.’l/iC
`Bioirrforrriatics 2608, 8, 443.
`A. M. F-ranié. Predicting intensity ranks of peptide ‘lragrnent ions.
`Journal of Proteome r’?eser1rch 25339, 8, 2226.
`Y. Xu,
`i\/l, Lazar. :‘\/lit!‘/i screening/biornarker discovery with linear
`ion trap MS: a library of human cancenspecific peptides. B/\/if
`(.'ancer2t3(39,£ 96.
`et-2!, A database oi mass
`Picotti, H.Lam_. D. Carripbeii,
`spectrometric assays for the yeast proteome. Nature /\/iethods
`2908, 5,. 9'13.
`:‘\/l.'\/\/aish, S.i_fn, D. M. Evans, A,i\'hosrovi—Eghbal, R.C.Beavis,
`
`J.Kast.
`irnpie.
`tation of a data repository—driven approach
`for
`targeted proteornics experirnents by muitipie reaction
`rnonitoring.Joumai ofProteomic5 2093, 72, 838.
`Skyiine.SRi\/i/MRi\/l-Builder.
`29053,
`https:/./'prendanx—uwigs.
`wa s h i ng ton .ed u/i a bkey/wi ki,/ho me/softwa re/'5 kyl in e/pag e.
`vie\:v?narnezdeiault.
`i\.'iacl_ean, D. Tornazeia, G. Finney, et ai. Automated creation and
`refinement of compiex scheduied srm methods for targeted
`proteomics. ln Presented at the 5.711? A57‘/IS Coriier‘ence on Mass
`S;3ectron7etr;.r and Aiiied Topics. Phiiadeiphia, PA, i\/iay 31»-Jun 4,
`2309.
`, A.Praltash, D.i\/‘..Torriaze!a, B.i-‘rewen, eta}. Expediting the
`deveioprnent of targeted SRi‘v’l assays: using data from shotgun.
`proteomics
`to autornate rnethod deveioprnent.
`journaf of
`Proteome Research 2009, 8, 2.733.
`i\.'iPJ\Iipilot.
`https://productsappliedbiosysterns.corn/ab/en/US,’
`adirect/ah?cmd:catNavigate2&catiD=6ti53S4&tab=Detaiiinfo.
`[2 " Agiient_.’ViRM. wvvvv.chem—agiient.corn/pdf/QQQ,oeptide_quant_
`-3n_eserninar.pdf.
`
`0000056
`
`
`
`O_uantitative proteornics
`
`
`
`
`
`
`J. A/lass. Spectrom. 200$, 44, 631 1 660
`Copyright
`2009 John Wiley & Sons, Ltd.
`
`[27
`
`[28
`
`[30
`
`[34
`
`[42]
`
`-C. Sherwood, A. Eastham, A. Peterson, et al. i\:".aRii‘»Illaa: a software
`applicationfor specti
`' bran/—based i\/lRi\.'ltr'arrsition listassemply.
`_I'ourna!ofP.roteome R_searclr 2(it39,8,/-i396.
`
`J. A. Me;
`',. L. Bianco, V.Ottone, er al. i\f.RlVlaid,tl1e vxiela»-'oased tool
`for designing multiple reaction rnonitoring (ii/ERM)
`transitions.
`.'l/lolecular and Cellular Proteorrrlcs 2999, 8, 696.
`[29 D. B. ivlartin, T. Holzman, T). ivlay, erol.
`lVlRi‘v’ler, an interactive,
`open source and cross—platt"orrn system .or data extraction
`and visualization of multiple reaction monitoring experiments.
`llxiolecular and Cellular Proteornics 2069, 7, 2270.
`B. it. Franza Jr, Y. P. Rorgnon. Stable isotope metabolic labeling tor
`analysis of laiopolymers. Patent No. 6653076. Assignees: The
`Regents ofthe University ofWasnington,fiied 20s’)l,issued 2003.
`[J Automaton.
`nttp:/'/wvv‘vv‘3.appliedbiosysternscorn/cmslgroupsl
`psm_marl<eting/clocuments/generaldocuments/cms 039}’98.pdf.
`[last accessed: 2()t‘:9].
`[.22 M. Fonsi, F. Fiore, P. Jones, et al. lvletabolisiwrelated liabilities ofa
`potent histone deacetylase r’,i*lDAC) inhrbitorand relevance or't“:e
`route ofadrninistration on its metabolicfate. Xenolmiotlca 2069, 'l.
`[33 M. Bantscheif. M. Schirle,
`G. Sweetman,
`J, Rick,
`B. Kuster.
`Quantitative mass spectrometry in proteon 's' a critical review.
`Armlytlcaland8loor:alyticalC'hemistry2