Seminar 1

Page 1

R

Tb

AD

a

ARRAA

KMPPSE

0d

Analytica Chimica Acta 703 (2011) 19– 30

Contents lists available at ScienceDirect

Analytica Chimica Acta

j ourna l ho me page: www.elsev ier .com/ locate /aca

eview

ools for analyzing the phosphoproteome and other phosphorylatediomolecules: A review

lexander Leitner ∗,1, Martin Sturm1,2, Wolfgang Lindnerepartment of Analytical Chemistry, University of Vienna, Vienna, Austria

r t i c l e i n f o

rticle history:eceived 9 May 2011eceived in revised form 7 July 2011ccepted 10 July 2011vailable online 19 July 2011

eywords:ass spectrometry

rotein phosphorylationroteomics

a b s t r a c t

Enrichment, separation and mass spectrometric analysis of biomolecules carrying a phosphate groupplays an important role in current analytical chemistry. Application areas range from the preparativeenrichment of phospholipids for biotechnological purposes and the separation and purification of plasmidDNA or mRNA to the specific preconcentration of phosphoproteins and -peptides to facilitate their lateridentification and characterization by mass spectrometry. Most of the recent improvements in this fieldwere triggered by the need for phosphopeptide enrichment technology for the analysis of cellular proteinphosphorylation events with the help of liquid chromatography–mass spectrometry. The high sensitivityof mass spectrometry and the possibility to combine this technique with different separation modes inliquid chromatography have made it the method of choice for proteome analysis. However, in the case

ample preparationnrichment

of phosphoprotein analysis, the low abundance of the resulting phosphopeptides and their low qual-ity fragment spectra interfere with the identification of phosphorylation events. Recent developmentsin phosphopeptide enrichment and fragmentation technologies successfully helped to overcome theselimitations. In this review, we will focus on sample preparation techniques in the field of phosphopro-teomics, but also highlight recent advancements for the analysis of other phosphorylated biomolecules.

Alexander Leitner studied Chemistry in Vienna andobtained his PhD in Analytical Chemistry under Prof.Lindner in 2004. After continuing his research inVienna, he joined the group of Prof. Ruedi Aeber-sold at the Institute of Molecular Systems Biologyat ETH Zurich as a postdoctoral researcher in 2008.His general interest is the development of analyticalmethodology for applications in proteomics, with astrong focus on mass spectrometry. Currently, he isworking on the development of new techniques forthe structural analysis of proteins and protein com-plexes using chemical cross-linking and advanced MS.

∗ Corresponding author. Current address: Institute of Molecular Systems Biology, ETH ZE-mail address: [email protected] (A. Leitner).

1 These authors contributed equally.2 Current address: ofi - Austrian Research Institute for Chemistry and Technology, Vien

003-2670/$ – see front matter © 2011 Elsevier B.V. All rights reserved.oi:10.1016/j.aca.2011.07.012

© 2011 Elsevier B.V. All rights reserved.

Martin Sturm studied Biochemistry at the Univer-sity of Vienna and obtained his Master degree at theMedical University of Vienna in 2006. After one yearresearch work at the Vienna Biocenter in the lab ofProf. Ammerer he started his doctoral thesis in Ana-lytical Chemistry under Prof. Lindner, obtaining hisPhD 2010. Currently he is employed at the AustrianResearch Center OFI as an expert for LC–MS. Therehe is working on establishing multicomponent anal-ysis methods for extractables and leachables studiesby LC–MS.

urich HPT E 53, Wolfgang-Pauli-Str. 16, 8093 Zurich, Switzerland.

na, Austria.

Page 2

2 Chimica Acta 703 (2011) 19– 30

1

1

ebtRwtmmWtaoavTAbboafmib(ri

1(

LgmAtmpoaacopi

Fig. 1. The phosphate group as enrichment target. The selective recognition ofthe phosphate group by various approaches is illustrated: Structure (antibodies);charge and polarity (ion exchange chromatography, partially also antibodies and

0 A. Leitner et al. / Analytica

Wolfgang Lindner holds since 1996 a Chair of Ana-lytical Chemistry at the University of Vienna, Austria.Throughout his career he got strongly influenced bylife science themes spanning from pharmaceuticalanalysis, metabolomics, proteomics, etc. and separa-tion science methodologies as of HPLC, GC, CE/CEC andLC–MS. In this context particular interest developedalso towards non-covalent interactions and molecularrecognition phenomena with focus on stereochem-istry leading to the development of novel syntheticselectors useful for enantioselective separation tech-niques. Working on the interface of organic, analyticaland biological chemistry characterizes best his scien-

tific credo supported by a rich portfolio of publications.

. Introduction

.1. The biological role of phosphates

Phosphate groups are a very common functional group in a vari-ty of different biological compounds. They form the hydrophilicackbone of DNA or RNA poly- and oligomers and representhe hydrophilic group in many amphiphilic membrane lipids.eversible protein phosphorylation works as a molecular switch,hich allows the regulation of metabolism and signal transduc-

ion in cells. Phosphorylation marks the activation of sugars foretabolism, and the triphosphate nucleotide ATP serves as theain energy carrier and phosphate group donor in all organisms.estheimer summarized some of the unique chemical proper-

ies of the phosphate group in a seminal paper [1]. Phosphatesre capable to carry two negative charges depending on the pHf the solvent. Typical pKa values are 2.2 for the singly chargednd 7.2 for the doubly charged free phosphate [2], although thesealues can vary depending on the specific chemical environment.his ensures that phosphates remain ionized at physiological pH.nother important structural feature is that phosphate groups areasically tetrahedral, with a symmetry depending on the num-er of substituents on the O-atoms [3]. This allows the formationf geometrically restrained hydrogen bond networks. Phosphateslso exhibit chemical reactivity that may be particularly attractiveor carrying out and/or preventing biologically relevant transfor-

ations; for example the slow rate of phosphate ester hydrolysisn DNA ensures its long-term stability. Although phosphorous haseen deemed an essential element in living organisms, the recentand controversial) discovery of a microorganism that is able toeplace phosphorous with arsenic [4] represents an important find-ng that this may not necessarily be the case.

.2. The phosphate group as a target for enrichment methodsFig. 1)

Many enrichment methods take advantage of the ionic andewis base (electron-pair donating) character of the phosphateroup for interaction. Therefore, methods based on ion exchangeechanisms are generally suitable for the analysis of phosphates.dditionally, the Lewis base properties allow coordinative binding

o positively charged iron or gallium central atoms in a chelatingatrix, a concept called immobilized metal affinity chromatogra-

hy (IMAC). Recently, new metal oxide materials like TiO2, ZrO2r SnO2, which possess Lewis acid and ion exchange properties,re frequently used for selective phosphopeptide enrichment. Inddition to the methods mentioned above, there are different

hemical tagging techniques that take advantage of the reactivityf the phosphate group. Finally, antibodies may be raised againsthosphate group-containing motifs and are frequently used for

mmunoprecipitation of phosphoproteins.

metal affinity techniques); metal affinity (IMAC and MOAC) and chemical reac-tivity (chemoselective reactions, see also Fig. 3). IMAC, immobilized metal affinitychromatography; MOAC, metal oxide affinity chromatography.

Some of the techniques that were primarily developed for phos-phoproteomic analysis may also be applied to other classes ofphosphorylated biomolecules. Due to the fact that there is anincreased interest of academic researchers, but also from the phar-maceutical industry for the selective enrichment and separationof phospholipids, DNA (plasmids), or phosphorylated metabolites,we will also cover selected recent applications and developmentsin these fields at the end of this review.

1.3. Phosphoproteomics

The majority of enrichment and separation strategies weredeveloped as a result of the need for the selective enrich-ment of phosphopeptides to facilitate their identification bymass spectrometry. Reversible protein phosphorylation is themost widespread post-translational protein modification (PTM) in(eukaryotic) cells, and it is the main chemical protein modificationinvolved in cellular signaling, metabolism, protein transport or celldivision and apoptosis, when proteins interact with each other [5].Recent proteomic research revealed that the onset of many severediseases, especially many types of cancers, is influenced by theactivity of tyrosine kinases in certain regulation/signal transductionpathways [6,7]. Recently developed anticancer drugs like imatinib,dasatinib or bosutinib act as direct inhibitors of the respective pro-tein tyrosine kinases in cancer specific signaling networks [8,9].New proteins in protein interaction pathways could successfully beinferred by LC–MS/MS identification coupled with phosphopeptideenrichment methods [10–13]. These achievements show that phos-phoprotein identification and phosphorylation site determinationis of high clinical and research interest.

O-phosphorylation is the most common type of protein phos-phorylation. It occurs mainly on the hydroxyl group-containingamino acids serine, threonine and tyrosine. Although widely dif-fering ratios have been reported, phosphotyrosines are clearly the

least abundant residues among the three. Far less frequently, N-, S-and acyl-phosphorylation occurs on histidine, lysine, cysteine andaspartic or glutamic acid residues [14]. The covalent attachmentof the phosphate group to the respective amino acid residues is

Page 3

Chimi

cohstol

twidrarupc

2

aotpadibtmtppsibm

emmpp(tCfahmbplbdf

aMtait

A. Leitner et al. / Analytica

atalyzed by a class of enzymes called kinases which use the energyf adenosine triphosphate (ATP) or guanine triphosphate (GTP)ydrolysis for the transfer of the phosphate group onto the sub-trate protein [15,16]. This may cause a conformational change andhus alteration of the activity of the respective substrate proteinsr attracts phosphospecific binding domains from other proteinseading to a protein–protein interaction [17].

Although some proteins remain permanently phosphorylated,he larger fraction, especially those involved in signaling path-ays show a highly dynamic and regulated pattern resulting from

nteractions between kinases and phosphatases, which rapidlyephosphorylate the proteins after a phosphorylation event,esembling an “on/off” mechanism. Even though phosphoproteinsccount for at least 30% of the eukaryotic proteome, based onecent comprehensive studies, the ratio of the phosphorylated tonphosphorylated form of the protein is rather small, so that thehosphoproteins will usually be present in substoichiometric con-entrations.

. The role of mass spectrometry in phosphoproteomics

Mass spectrometric analysis of phosphopeptides poses somedditional challenges compared to the characterization of mixturesf unmodified peptides. The widely held opinion of reduced ioniza-ion sensitivity of phosphopeptides is mostly related to the fact thathosphorylation is a substoichiometric event and phosphopeptidesre generally less abundant [18]. At similar concentrations, theetection sensitivity of singly and doubly phosphorylated peptides

s not necessarily dramatically different. Similarly, the retentionehavior of peptides is not significantly altered upon phosphoryla-ion for the majority of peptides; in fact, some phosphopeptides

ay be even more retained in reversed-phase chromatographyhan their unmodified counterparts depending on the mobilehase composition. In contrast, the situation is different for highlyhosphorylated peptides that are usually detectable only with con-iderably reduced sensitivity in positive ion mode while negativeonization is commonly avoided for practical reasons (reduced sta-ility). Therefore, the actual presence of highly modified peptidesay be severely underestimated.Such a bias could also result from a bias introduced during

nrichment steps (see below), but also as a result of subopti-al fragmentation properties. In fact, the latter constitutes theost significant obstacle in mass spectrometric analysis of phos-

hopeptides [19,20]. During collision-induced dissociation, manySer- and pThr-peptides exhibit a neutral loss of phosphoric acidH3PO4, −98 Da) as the dominant fragmentation pathway—in con-rast to pTyr peptides where the modification remains stable.onsequently, informative fragments stemming from backbone

ragmentation may be present only at low intensities or some cleav-ges may be entirely suppressed. A substantial amount of researchas therefore been contributed to improving phosphopeptide frag-entation. To this end, chemical removal of the phosphate group by

eta-elimination has been proposed [21,22], which is however notractical for complex samples due to possible side reactions. Simi-

arly, altering the general ionization properties of a phosphopeptidey reducing the basicity of the protonation site upon arginineerivatization has been shown to reduce neutral loss, although onlyor model peptides [23].

More attention has been directed to the development of refinednd new fragmentation mechanisms [19,20]. Neutral loss drivenS3 is one way to deal with the loss of the phosphate group, where

he loss of H3PO4 from a precursor is detected in real time andnother stage of fragmentation is induced. Although this features now routinely implemented on many tandem mass spectrome-ers containing an ion trap analyzer and can be used for improving

ca Acta 703 (2011) 19– 30 21

the statistical certainty at the data analysis stage, this comes at aprice of reduced scan speeds and more complex bioinformatic pro-cessing. Alternatively, multi-stage activation has been introducedon linear ion trap systems, where instead of two separate isola-tion and fragmentation steps, the precursor and the neutral lossreaction product are activated subsequently without an intermit-tent isolation step. Therefore, a “hybrid” spectrum containing MS2

and MS3 information is obtained, this facilitates data analysis asonly one level of fragmentation data needs to be analyzed [24].Both approaches, however, fail in the case of highly phosphory-lated peptides characterized by multiple successive losses of thephosphates.

CID can be performed either in ion traps or in dedicated colli-sion cells, and the differences in energy deposition (“slow heating”in ion traps and “fast heating” in collision cells, reviewed in [25])result in differences in the appearance of mass spectra, especiallyfor modified peptides such as phosphopeptides. Therefore, beam-type collision cell CID is not only used in quadrupole-time-of-flightinstruments, but has also been introduced on the highly popularOrbitrap hybrid instruments (termed HCD for “higher-energy CID”[26]). As a result of the faster energy deposition and the occurrenceof consecutive dissociation, such collision cell CID spectra maycontain less abundant neutral loss signals and more interpretablesequence information [19]. For example, Mann and co-workersused HCD on a new Orbitrap instrument to identify more than16,000 phosphorylation sites in HeLa cells following multidimen-sional enrichment/fractionation of phosphopeptides [27].

The introduction of the complementary electron transfer disso-ciation technique [28] has also had enormous impact on large-scalephosphopeptide analysis. In contrast to CID, ETD does not causeneutral loss of phosphoric acid, thereby dramatically improvingspectral quality in certain cases. While it was initially thought thatETD would to some extent replace CID, especially for the character-ization of modified peptides, several years of research have madeapparent some limitations of the technique: Slower scan speeds(due to longer reaction times), lower fragmentation efficiencies andconsiderable m/z and charge state dependence of spectral quality.Coon and co-workers therefore introduced a decision tree method[29] that switches between CID and ETD according to precursor m/zand charge state. Furthermore, most widely used search enginesperform suboptimal for ETD data as they do not take into accountpeculiar properties of ETD spectra, e.g. the formation of hydro-gen rearrangement products [30]. Spectral postprocessing and thedevelopment of new search algorithms [31,32] attempt to over-come these issues, although there is no universal “gold standard”yet.

Finally, quantitation of phosphopeptides or, more general, thephosphorylation state of a protein or even a proteome remains anarea of intense research. Due to the enormous biological signifi-cance of phosphorylation dynamics in signaling events, a numberof approaches have been used to study quantitative changes inphosphosite occupancy between different biological conditions.For details, the reader is referred to recent reviews on this topic[33–35]. While quantitative data can be obtained by variousmeans of stable isotope labeling, e.g. metabolic or chemical label-ing using established techniques, it has to be considered thatchanges in the relative abundance of phosphopeptides may notonly result from up- or down-regulation of a particular phosphory-lation site. Additionally, phosphorylation on additional sites withinthe same peptide may lead to the peptide evading detection inits multiphosphorylated form. Furthermore, in some experimen-tal set-ups changes in absolute protein amounts may skew theresults, as recently demonstrated by Gygi and co-workers [36].Ideally, both the unmodified form of the peptide and all possible

phosphoisoforms would need to be evaluated to obtain definitiveinformation.

Page 4

22 A. Leitner et al. / Analytica Chimica Acta 703 (2011) 19– 30

F most

I fractiH hroma

3

eiwtepIaittsabttuopppb

3

wkt

ig. 2. Commonly used enrichment workflows in phosphoproteomics. Only the

mmunoprecipitation and IMAC/MOAC enrichment with or without additional preILIC, hydrophilic interaction chromatography; IMAC, immobilized metal affinity c

. Enrichment techniques in phosphoproteomics

To facilitate the identification of phosphopeptides, many differ-nt enrichment strategies were developed and are in permanentmprovement. Fig. 2 gives an overview about frequently used

orkflows that are discussed in detail in the following chap-ers. Enrichment techniques in phosphoproteomics are usuallymbedded in elaborate workflows that involve numerous samplereparation steps, i.e. protein extraction, enzymatic cleavage, etc.

n the context of analyzing protein phosphorylation it is essentiallthough not necessarily straightforward to minimize the activ-ty of phosphatases that are released during cell lysis, to avoidhe generation of artifacts during workup [37]. Similarly, pro-ease inhibitors should be added to avoid degradation of theample. Most enrichment techniques (with immunoprecipitations a notable exception) are used at the peptide level, not onlyecause these methods work more efficiently but also because ofhe more substantial depletion of unphosphorylated compoundshat is achievable. Proteolytic digestion, however, leads to annavoidable loss of contextual information on the interdependencef modifications on the protein level, for example, whether a givenhosphorylation site is only occupied when another one is alreadyresent. Techniques that study the interplay of different types ofost-translational modification are still in their infancy, but of highiological significance [38].

.1. Immunoaffinity chromatography

Immunoaffinity chromatography or immunoprecipitation is aidely used enrichment strategy in biochemistry. Antibodies are

nown to have a high affinity and selectivity for their target epi-opes. Commonly used for trapping whole proteins, they are also

commonly used approaches relying on peptide-level enrichment are illustrated:onation (ion exchange chromatography or HILIC). For additional details, see text.tography; MOAC, metal oxide affinity chromatography.

used for enrichment of smaller molecules like peptides or aminoacids, although the generation of good working antibodies forsmaller molecules is more elaborate.

In a traditional experiment to analyze protein phosphoryla-tion the setup focused on a single phosphoprotein of interest. Fordetection and purification specific antibodies directed against thephospho-epitopes had to be generated. The production and val-idation of proper antibodies is a time consuming and expensiveprocess taking up to one year to generate and validate an effectiveantibody [39]. Antibodies used for phosphopeptide or phosphopro-tein enrichment on a proteomic scale should therefore be directedsolely against the phosphorylated amino acid, i.e., they should havethe same or similar affinity and selectivity for all the phospho-peptides or -proteins in a biological sample to obtain a balancedand unbiased enrichment of all phosphorylated analytes. How-ever, the recognition of phosphate groups in peptides is affectedby the surrounding amino acids so it is difficult to generate ade-quate antibodies. Although there are antibodies against the threemost common phosphoamino acids, phosphoserine (pS), phospho-threonine (pT) and phosphotyrosine (pY), their use is frequentlylimited by the specificity [40,41]. Highly specific pY antibodies existand are predominantly used. Tyrosine kinases play an importantrole in human cancer, taking part in oncogenic signaling for cel-lular proliferation and survival. Therefore an unregulated tyrosinekinase activity can lead to malignancy and tumor formation [42].In a global phosphoproteomic approach the number of identifica-tions of phosphotyrosine containing peptides would be rather smallbecause of its low abundance, therefore pTyr-selective methods are

essential.

As an example, a comprehensive proteomic study of tyrosinephosphorylation in Jurkat cells was done by Rush et al. [42]. Phos-photyrosine containing peptides from a cell digest of pervanadate

Page 5

Chimi

(icepsiscatdwiiE

3

wwabufiteeivatnaopma

dttdaptdreTtcs

ptTmbaFbba

A. Leitner et al. / Analytica

a tyrosine phosphatase inhibitor) treated Jurkat cells weremmunoprecipitated with P-Tyr-100, a phosphotyrosine spe-ific antibody non-covalently coupled to protein G agarose. Thenriched peptides were later on analyzed by conventional reversedhase chromatography–tandem mass spectrometry. Using thistrategy, 688 pTyr-containing peptides and 628 pTyr-sites could bedentified. The same strategy was used by Rikova et al. for a largecale analysis of tyrosine kinase activity in non-small cell lung can-er (NSCLC) cell lines [43]. Known tyrosine kinases such as EGFRnd c-Met as well as new oncogenic tyrosine kinases not knowno play a role in lung cancer were identified. In summary over 50ifferent tyrosine kinases and over 2500 downstream substratesere identified by this approach. White and co-workers combined

mmunoprecipitation with relative quantitation by using isobaricTRAQ tags to elucidate time-resolved phosphorylation events inGFR signaling [44].

.2. Immobilized metal affinity chromatography (IMAC)

Cysteines and histidines are known to form stable complexesith zinc and copper ions in an aqueous solution [45]. This effectas used for selective enrichment of proteins via transition met-

ls trapped in a chelating matrix. The concept, termed IMAC, maye seen as a sort of pseudo-affinity chromatography and is widelysed for the enrichment of recombinant histidine-tagged proteinsor biotechnological purposes. A string of six His residues is genet-cally attached to the C- or N-terminus of proteins of interest andhe high affinity of histidine to covalently fixed Ni2+ is used fornrichment of the tagged protein. The concept was expanded fornrichment of phosphorylated proteins by Andersson and Porathn the year 1986 [46]. In their experiments they used Fe3+ boundia iminodiacetic acid (IDA) to a sepharose matrix. Phosphorylatedmino acids like phosphoserine, phosphothreonine or phospho-yrosine were retained by the chromatographic material whereason-phosphorylated amino acids were not, or in some cases, likespartic acid or glutamic acid, only weakly bound. A separation ofvalbumin phosphoisoforms, carrying different numbers of phos-hates, succeeded using this technology. An advantage of theethod was that all steps could be carried out in water or buffer

nd no protein denaturing components were needed.Over the years, a variety of different supports have been intro-

uced for IMAC-based enrichment, and the technique has long beenhe most frequently used method for enrichment of phosphopep-ides, also due to the fact that commercial kits are available fromifferent suppliers. However, the high level of unspecific binding ofcidic peptides via their carboxylic residues remains a challengingroblem. A method to circumvent unspecific binding of acidic pep-ides was introduced by Ficarro et al. [47]. In their approach a trypticigest of a yeast whole cell lysate was analyzed and carboxylicesidues on peptides were chemically derivatized by O-methylsterification with methanolic HCl before Fe3+-IMAC enrichment.he esterification led to an elimination of unspecifically bound pep-ides without loss in sensitivity. Using this method the authorsould identify 200 phosphopeptides from S. cerevisiae which was aubstantial achievement at that time.

Although this technology increased the selectivity of IMAC phos-hopeptide enrichment considerably and has been used frequently,he esterification step might be incomplete and cause sample loss.herefore, additional ways to improve the specificity of the enrich-ent were evaluated, such as adjusting the pH of the loading

uffer or using different chelated ions. Kokubu et al. used 50%cetonitrile, 0.1% trifluoroacetic acid (TFA) as loading buffer for

e3+-IMAC enrichment when analyzing a tryptic digest of a mouserain protein extract [48]. With TFA, selective protonation of car-oxylic residues on peptides was reported; additionally, the highcetonitrile content reduced hydrophobic interactions between

ca Acta 703 (2011) 19– 30 23

peptides and the IMAC resin. Using this loading buffer a reduc-tion of nonspecific binding of acidic peptides was obtained. Just 96nonphosphopeptides and 1654 phosphopeptides were assigned byMascot from a mouse brain sample. After manual validation 166phosphosites on 135 different proteins were identified using thisapproach.

Although iron is used as central ion in most IMAC methods,other metal ions have been evaluated for selective phosphateaffinity. Posewitz tested different metal ions like Ga, Sn, Ge, Fe,and others for their applicability in IMAC phosphopeptide enrich-ment [49]. With Ga3+ a better selectivity compared to conventionalFe3+-IMAC was reported when analyzing a tryptic digest of phos-phoproteins. An interesting approach was reported from the groupof Zou [50,51]. They used a phosphate polymer to coordinativelybind Ti4+ or Zr4+ ions, and the resulting IMAC resin was used forphosphopeptide enrichment and compared to Fe3+-IMAC, TiO2 andZrO2 enrichment methods. For the preparation of the IMAC resinthey used a special chemistry to attach the phosphate groups tothe matrix via a linker to improve the accessibility of the inter-action sites and to provide a beneficial structural orientation forthe selective binding of phosphorylated biomolecules. Additionally,phosphates form a stable MO6 octahedron structure with Ti and Zrions. So the metal ions that are stably bound to the phosphonategroups form an interface on the surface of the resin on which phos-phate ions may form a stable double layer. To illustrate the potentialof this material, a mix of a standard phosphoprotein digest and aBSA digest as control in a ratio of 1 to 500 was analyzed. Addition-ally, the method was applied to the phosphoproteome analysis ofmouse liver. The method was reported to outperform conventionalmetal oxide enrichment (see following chapter) and Fe3+-IMAC interms of efficacy and selectivity, even when using optimized con-ditions for the respective methods. A bias of Ti4+-IMAC towardsmonophosphorylated peptides and of Zr4+-IMAC towards multi-ply phosphorylated peptides was reported. The results of the newmethods were ascribed on one hand to the highly specific inter-action between the Ti and Zr ions and the phosphate groups onpeptides and on the other hand to the novel resin design.

Notable IMAC-based studies on complex biological sampleswere able to identify several hundred to thousands of phospho-rylation sites [52–54]. A way to further increase the specificityand efficacy of the IMAC technology is to combine it with otherenrichment or fractionation methods. In most cases ion exchangechromatography or hydrophilic interaction liquid chromatography(HILIC) chromatography is used for prefractionation before specificenrichment for phosphopeptides with IMAC or metal oxides is per-formed. A detailed discussion is given in the chapter on combinedenrichment methods.

3.3. Metal oxide affinity chromatography (MOAC)

Since the beginning of the 1990s, titanium dioxide (titania,TiO2) and zirconium dioxide (zirconia, ZrO2) were increasingly usedas new chromatographic materials for high performance liquidchromatography in normal phase mode. The advantages of thesemetal oxides included large adsorption capacities, chemical stabil-ity when used under extreme pH ranges, mechanical stability andunique amphoteric ion exchange properties [55–59]. Phosphatesare known to bind to metal oxide materials [3] and in 1990 Matsudaet al. [56] reported the selective adsorption of organic phosphatesto ceramic TiO2 material. Henceforward several studies concerningthe enrichment of phosphate group-containing biomolecules withmetal oxide materials emerged [58,59], but not a lot was known

about the surface chemistry and binding properties of these mate-rials. Connor and McQuillan [3] investigated phosphate adsorptiononto TiO2 from aqueous solutions with infrared spectroscopy. Theyproposed a pH-dependent bidentate binding of monosubstituted

Page 6

2 Chimi

pfa[pnaBbp

piaiieTtcOZacbalefafpcafgsfip

eaccmmpfiarpTlt1lwpdltcOii

4 A. Leitner et al. / Analytica

hosphate groups to TiO2 sol–gel films. In another work, the sur-ace chemistry of metal oxide materials was characterized by theircidity and basicity which are of Lewis and Bronsted base type60]. Alumina (Al(OH)3), TiO2 and ZrO2 possess strong Lewis acidroperties with increasing Lewis acidity from alumina to zirco-ia. In addition, the high coordination numbers of Ti–O and Zr–Ore responsible for strong complexation properties of these oxides.oth the Lewis acid and the strong complexation properties maye a possible explanation for the affinity of some metal oxides forhosphate molecules.

Tani and Miyamoto [61] characterized the chromatographicroperties in terms of ion exchange and ligand exchange behav-

or of different calcinated titania materials. TiO2 was calcinated inn oven at 200–800 ◦C and a strong temperature dependence of theon and ligand exchange behavior could be observed. At 700 ◦C TiO2s completely converted from anatase to rutile form and no ion-xchange or ligand-exchange behavior could be detected anymore.he authors ascribed the absence of these properties to the loss ofhe surface hydroxyl groups and coordinatively bound water byalcination at high temperatures. In another experiment Tani andzawa [62] compared the chromatographic behavior of TiO2 andrO2 with hydroxy acids and other substituted aliphatic carboxyliccids. The highest retention times were achieved with �-hydroxyarboxylic acids because they can form a stable five-membered ringetween metal ion and acid. �-hydroxy acids were later introduceds additives to avoid unspecific binding of acidic unphosphory-ated peptides to TiO2 to increase selectivity in phosphopeptidenrichment ([63], see further discussion below). The stable ringormation can be a possible explanation why �-hydroxy acids areble to displace acidic peptides from TiO2 or ZrO2. A work with aocus to screen differently treated metal oxide materials for theirhosphate affinity was performed by Leitner et al. [64]. They alsoompared different surface treated tin dioxide microspheres [65],n alternative metal oxide used for phosphopeptide enrichment,or phosphopeptide affinity and selectivity purposes. The resultsave evidence that parameters like calcination temperature andurface treatment with acid or base can differently affect the sur-ace chemistry and morphology of metal oxide materials, whichn turn will affect the affinity and selectivity of the materials tohosphates.

The first proteomic application of TiO2 for phosphopeptidenrichment was reported by Pinkse et al. [66]. In this work a novelutomated method for the enrichment of phosphopeptides fromomplex mixtures with TiO2 was developed. A two dimensionalhromatographic setup with titanium dioxide-based solid-phaseaterial (Titansphere) as the first dimension and reversed-phaseaterial as the second dimension was employed. Phosphorylated

eptides were separated from non-phosphorylated peptides in therst dimension by trapping them under acidic conditions (0.1 Mcetic acid) on the TiO2 precolumn. Nonphosphopeptides were notetained in the first dimension but trapped in the second dimensionrecolumn before they were analyzed by nanoflow LC–ESI–MS/MS.he phosphopeptides were eluted from the column under alka-ine conditions (ammonium bicarbonate pH 9.0), concentrated onhe second dimension and analyzed by nanoflow LC–ESI–MS/MS.25 fmol of a phosphopeptide in a 1:1 mixture of the phosphory-

ated and unphosphorylated form could be successfully identifiedith a recovery rate of above 90%. Additionally a novel autophos-horylation site could be identified from a digest of the cGMPependent protein kinase. A drawback of this strategy was the high

evel of unspecific binding of acidic non-phosphorylated peptideso TiO2 under these conditions which were reported to be on a

omparable level as for IMAC. The authors initially recommended-methyl esterification [47] to reduce unspecific binding. However,

n the following years numerous refined protocols were reportedn the literature.

ca Acta 703 (2011) 19– 30

Larsen et al. [67] used 2,5-dihydroxybenzoic acid (DHB), amatrix compound widely used in MALDI mass spectrometry, toincrease selectivity without a reduction in phosphopeptide recov-ery. In this procedure, peptides from a tryptic digest of thephosphoproteins ovalbumin and �-and �-casein were used forphosphopeptide enrichment with TiO2 columns. Peptides wereloaded onto TiO2 columns with different concentrations of DHB(0–200 mg mL−1 in 80% acetonitrile, 0.1% TFA) in the loading buffer.The column was washed with 20 mg mL−1 DHB in 50% acetoni-trile and subsequently with an 80% acetonitrile/0.1% TFA solution.Bound peptides were eluted with NH4OH, pH 10.5. With increas-ing concentrations of DHB in the loading buffer the number ofnon-phosphorylated peptides bound to the column decreaseddramatically with no unspecifically bound peptides observed at200 mg mL−1 DHB. When using very complex samples the authorrecommended the use of even higher concentrations of DHB toexclude unspecific binding of non-phosphorylated peptides. A pos-sible reason why DHB is able to avoid unspecific binding of acidicnon-phosphorylated peptides without reducing phosphopeptidebinding was explained by different binding properties of thesemolecules to TiO2. DHB was proposed to bind to TiO2 via a chelat-ing bidentate geometry whereas phosphates bind in a bridgingbidentate way. The assumption was that the two molecules tar-get different binding sites on the TiO2 surface and DHB directlycompetes with carboxylic residues from acidic peptides. A big dis-advantage of this protocol is that complete removal of DHB beforeLC–MS analysis is necessary to avoid interference with the chro-matographic separation and ionization in the mass spectrometer.

In the mean time, several attempts to increase selectivity with-out a decrease in sensitivity were carried out by screening differentloading buffer additives to suppress unspecific binding [63,68].Mazanek et al. [68] suggested using a mix of DHB and octanesul-fonic acid (OSA), an ion pairing agent used for improved peptideseparation in reversed phase chromatography, to reduce unspe-cific binding. The additives were used in lower concentration andshould therefore be less problematic for the following analysis.Additionally, OSA should improve the quality of the reversed phaseseparation by acting as an ion pairing agent. The protocol was testedwith a mix of phosphorylated and non-phosphorylated peptidesand the applicability of the protocol to more complex sampleslike a tryptic HeLa digest was demonstrated. More recently, thismethod was further optimized in terms of selectivity using slightlyincreased concentrations of DHB and OSA, and with the additionof heptafluorobutyric acid [69]. The optimized protocol allowedimproved recovery of phosphorylated peptides and a reduction ofunspecific binding from a trypsinized bovine serum albumin matrixwhen compared to a previous protocol [68]. The new method wasapplicable to both titania and zirconia enrichment. The usefulnessfor more complex samples was demonstrated with the identi-fication of in vivo phosphorylation sites of the affinity enrichedanaphase promoting complex (APC/C) and the mitotic complex,condensin-I. Sugiyama et al. screened different hydroxy acids asalternative displacement additives in TiO2 and ZrO2 enrichment[63]. The best results in terms of selectivity were achieved whenusing lactic acid in a concentration of 300 mg mL−1 in 80% acetoni-trile/0.1% TFA for TiO2 and 100 mg mL−1 �-hydroxypropanoic acidin 80% ACN/0.1% TFA for ZrO2 enrichment. Again the applicabilityto complex samples was successfully demonstrated by analyzing atryptic HeLa sample [63]. Lactic acid modified TiO2 enrichment gavethe highest selectivity as well the largest number of identified phos-phopeptides. In total, 1100 phosphopeptides could be identified byfour replicate experiments using TiO2 and ZrO2 enrichment.

An alternative to optimizing the loading buffer composition forTiO2 enrichment is the use of other metal oxide materials for phos-phopeptide enrichment purposes [70]. Soon after the introductionof TiO2-MOAC, ZrO2 was also evaluated due to its structural,

Page 7

Chimi

cttfdhTpfpwtodpil[od

oa[trMtacn

tcpcbttIdi

3

mpgiu

tcFisptppstcdw

A. Leitner et al. / Analytica

hemical and physical similarities [71]. Initially, a different biasowards the enrichment of mono- and polyphosphorylated pep-ides was reported for titania and zirconia. Subsequent studiesailed to reproduce these tendencies which may be attributed toifferences in other experimental parameters. The Ishihama groupas published a series of studies where ZrO2 was used alongsideiO2 to provide a more comprehensive coverage of the phospho-roteome [63,72]. Porous tin dioxide microspheres, a new materialor phosphopeptide enrichment, were tested in our group as novelhosphoaffinity chromatographic material [64,65]. The spheresere manufactured by the nanocasting process [73] which allows

he fine tuning of material properties like particle size, morphol-gy and pore size of different metal oxide materials. Using a trypticigest of 10 proteins with ovalbumin as the only phosphorylatedrotein, phosphopeptide enrichment with SnO2 and a simple load-

ng buffer composition without additives performed on the sameevel to TiO2 with the use of 300 mg mL−1 lactic acid as additive65] which was used as benchmark protocol. Later, a tryptic digestf a HeLa cell lysate was used for a comparison of SnO2 with twoifferent types of titania materials [74].

In addition to the three materials, a number of other metalxides have now been shown to possess some phosphopeptideffinity (reviewed in ref. [70]). These include Al(OH)3 [75], Nb2O576], Ta2O5 [77], Fe2O3 [78] and ZnO [79], among others. In addi-ion to conventional micrometer-sized particles, several groupseported the preparation of magnetic metal oxide nanoparticles forOAC enrichment. Although these materials offer potential advan-

ages in terms of capacity (due to the high surface-to-volume-ratio)nd rapid binding kinetics (due to short diffusion paths), appli-ations to more complex biological samples remain limited up toow.

Summing up, metal oxide materials possess a high affinityowards phosphates, exhibit chemical and physical stability andan be manufactured in different morphologies and sizes. Theseroperties make it the materials of choice in many current appli-ations. Additionally, MOAC has shown to be more tolerant againstuffer additives like salts, detergents and denaturing componentshan IMAC enrichment. These unique properties make metal oxideshe ideal materials for automated online enrichment/LC–MS [66].n this regard, Agilent Technologies has introduced a chip-basedevice with integrated TiO2 enrichment and RP-LC separation that

s now commercially available [80,81].

.4. Ion exchange chromatography

Different forms of ion exchange chromatography are com-only used in a two-dimensional chromatography set-up in

roteomics. Because of the strong negative charge of the phosphateroup, most phosphopeptides show particular retention behaviorn ion exchange chromatography in comparison to the majority ofnmodified peptides.

Beausoleil et al. were the first to exploit this in a phosphopro-eomic approach [10] when analyzing a tryptic digest of a HeLaell lysate using strong cation exchange (SCX) chromatography.or loading of peptides on the SCX material, peptides were acid-fied to a pH of 2.7 and bound peptides were eluted with increasingalt and pH. At the pH 2.7 Lys, Arg, His and the N-terminus ofeptides are positively charged. Tryptic proteolysis produces pep-ides with an Arg or Lys on the C-terminus, consequently, trypticeptides should carry a solution charge of +2 and because phos-hate groups retain their negative charge at this pH, the net chargetate of phosphopeptides should be +1. Thus the phosphopeptides

end to elute earlier from the SCX column. The phosphopeptide-ontaining fractions were analyzed by LC–MS2 and additionalata-dependent MS3 for improved phosphopeptide identificationas performed. This chromatographic setup combined with elab-

ca Acta 703 (2011) 19– 30 25

orate mass spectrometric identification allowed the determinationof 2002 phosphorylation sites on 967 phosphoproteins. But thecombination of SCX with RP chromatography also revealed somedifficulties, especially when both columns were coupled on-line.For example, the interaction of peptides with the SCX resin is pro-posed to be (mainly) of electrostatic and (partially) of hydrophobiccharacter, due to some residual hydrophobicity of the polymericstationary phase, so that structurally similar peptides with the samenet charge may be separated to some degree. For the increase ofpeptide recovery from the SCX column, the use of organic mod-ifier has been recommended [82], which can on the other handinterfere with the following RP separation. As a consequence bothdimensions cannot be optimized independently [83].

In contrast, when using strong anion exchange chromatogra-phy (SAX) instead of SCX, phosphopeptides are strongly retained.Zhang et al. demonstrated the applicability of anion exchange chro-matography (AEC) for selective phosphopeptide enrichment whenanalyzing a tryptic digest of the standard phosphoprotein �-casein[84]. Later, SAX was used for the enrichment and separation ofphosphopeptides from a digest of human liver cancer tissue byGu and co-workers [85]. In contrast to SCX chromatography, theauthors claimed that not only phosphopeptide enrichment, but alsofractionation according to the number of phosphate groups waspossible. This was demonstrated when analyzing a tryptic digest of�- and �-casein. For separation of bound phosphopeptides a saltgradient of NH4Cl was used and the pH of the solvent was set to avalue of 4. At this pH phosphopeptides remain ionized and henceare able to interact with the SAX resin via their phosphate groups.Consequently, multiply phosphorylated peptides should show ahigher affinity to the SAX resin than singly phosphorylated ones.The performance of the technology was compared to a Fe3+-IMACenrichment with a commonly used protocol. For comparison ofthe two methods only one fraction (5–28 min retention time) ofretained peptides was collected from the SCX column due to theassumption that nonphosphopeptides would not be retained bythe SCX column and will therefore elute with the flow through.A higher number of unique phosphopeptides, 47 compared to 24with IMAC, could be identified using the SAX method and an overlapof 12 unique phosphopeptides was identified with both methods.Although both methods suffer from unspecific binding, the non-phosphopeptides identified with SAX were mainly acidic peptideswhereas IMAC enriched peptides were reported to be more hetero-geneous as a result of histidine binding to IMAC. Additionally, theability for fractionating phosphopeptides was investigated. Two-minute fractions from the liver digest were taken and analyzed byMALDI–MS. This led to the identification of 274 phosphorylationsites from 305 unique phosphopeptides corresponding to 168 pro-teins at a false discovery rate of less than 1%. A disadvantage arisingfrom this method is that the solvents used in SAX chromatographyare not optimal for online LC–MS/MS coupling because the weaklyacidic to neutral pH and the aqueous buffer lower the ionizationefficiency when LC–ESI–MS is used.

3.5. Hydrophilic interaction chromatography and relatedtechniques

Beyond partitioning by charge state, hydrophilic interactionchromatography (HILIC) provides an additional orthogonal sepa-ration tool to reversed phase chromatography. In HILIC, analytesare separated according to their polarity. HILIC is a special formof “normal phase” chromatography using mobile phases that are5–40% buffered aqueous solutions. In HILIC, the dominant retention

mechanism is claimed to be a partitioning of the analytes betweena polar, water-rich stationary phase layer and an organic modifier-rich mobile phase. In addition, adsorption phenomena with themodified silica surface may occur, leading to a mixed-mode

Page 8

26 A. Leitner et al. / Analytica Chimica Acta 703 (2011) 19– 30

) Beta

rbd(rtpsbNIodbtAig

i[smHtucpcoo5l1a

Fig. 3. Chemoselective approaches targeting the phosphate group. (a

etention mechanism. In general, the more polar the eluentecomes the more elution strength it has. Typical starting con-itions for HILIC separations are up to 95% organic solventacetonitrile). HILIC as part of a multidimensional peptide sepa-ation strategy was used for the phosphoproteomic analysis of aryptic HeLa cell digest [86]. Phosphopeptides with the highly polarhosphate group should therefore be strongly retained on the HILICtationary phase. IMAC phosphopeptide enrichment was addedefore or after the HILIC step to increase selectivity of the setup.early 100% selectivity for phosphopeptides was achieved when

MAC enrichment was performed after HILIC chromatography, andver 1000 phosphorylation sites on 914 peptides were identified,emonstrating the use of HILIC as a powerful prefractionation toolefore selective phosphopeptide enrichment is performed. Addi-ionally phosphopeptides elute at a buffer composition of 70–50%CN containing 0.1% TFA which is asserted to be the optimal load-

ng buffer composition for IMAC phosphopeptide loading. Otherroups have recently adopted HILIC in their protocols [87,88].

Somewhat similar to HILIC, electrostatic repulsion-hydrophilicnteraction liquid chromatography (ERLIC), introduced by Alpert89], uses electrostatic repulsion as an additional chromatographictationary phase property to adjust selectivity in HILIC chro-atography. By superimposing the properties of ion exchange andILIC chromatography, selectivity can be adjusted by changing

he organic content or the pH of the solvent and additionally bysing a salt gradient. In an elaborate comparative study, ERLIC wasompared to SCX and SCX-IMAC for phosphopeptide enrichmenturposes by Gan et al. [90]. In their approach human epithelial car-inoma A431 cells were analyzed leading to a total identificationf 2058 unique phosphopeptides of which 1801 were identified bynly one of the methods. ERLIC accounted for 38%, SCX-IMAC for

7% and SCX alone for 5% of the phosphopeptide pool. The over-

ap between ERLIC and SCX-IMAC identified phosphopeptides was2% showing that for a comprehensive phosphoproteome analysis

combination of different enrichment strategies is sometimes nec-

-elimination and Michael addition, (b) phosphoramidate chemistry.

essary. Recently, a similar approach was applied to the analysis ofrat kidney tissue [91].

3.6. Chemical modification (Fig. 3)

In addition to chromatography-based methods, tagging phos-phate species with certain compounds using a specific chemicalderivatization reaction is another strategy for phosphopeptideenrichment (reviewed in ref. [92]). Site-specific modification ofphosphoseryl and phosphothreonyl residues [21] using a combi-nation of �-elimination and Michael addition is a way to introducephosphosite specific tagging. Chait and co-workers used this tech-nology to introduce a biotin tag for biotin/avidin enrichment [93].Under strongly alkaline conditions the phosphate moiety fromphosphoserine or phosphothreonine undergoes �-elimination toform a dehydroalanyl or �-methyldehydroalanyl residue, respec-tively. These �,�-unsaturated residues are Michael acceptors,which can react with nucleophiles like ethandithiole, which is fur-ther coupled to biotin. The benefits for this method are that onecan selectively enrich via different types of tags available, and thetags can be isotope labeled for quantification purposes [93] or carryfunctional groups to increase the ionization efficiency or to facili-tate phosphorylation site determination [22,94,95]. However, somedrawbacks are associated with this procedure. The high pH neces-sary for efficient �-elimination can cause some protein or peptidedegradation. In addition, O-glycosylated serines can be convertedto dehydroalanine as well. These drawbacks can be avoided byadditional protecting reactions or method modifications [96] butthis may lead to a more complicated workflow and potential sam-ple loss. In addition, only phosphoserine and phosphothreoninecontaining peptides can be enriched using �-elimination/Michael

addition because tyrosine is not able to form an �,�-unsaturatedbond.

Aebersold and co-workers used a carbodiimide-catalyzedreaction for the reversible capture of phosphate groups on

Page 9

Chimi

pamrdiwdmcfiUieatsic

3

plepptawl2idtweurc

3

emhsewitemIdiwolfmtop

A. Leitner et al. / Analytica

hosphopeptides using phosphoramidate chemistry [97]. Initially, laborious six step protocol was developed that only yielded aoderate number of identifications. Later, this method was further

efined and relative quantitation was implemented by the intro-uction of isotope labeling [98]. The resulting three step strategy

ncludes the protecting of free carboxylate groups by methylationith methanolic HCl (which also offers the possibility for the intro-uction of deuterium labeled methyl groups), the coupling of theethylated peptides to a dendrimer via their phosphate groups,

atalyzed by imidazole and carbodiimide in a one pot reaction, andnally the release of the peptides by acidic hydrolysis with 10% TFA.sing a tandem purification protocol with anti-phosphotyrosine

mmunoprecipitation followed by phosphoserine and -threoninenrichment by implementing the above stated protocol, 80 serinend threonine phosphorylation sites on 97 tyrosine phosphopro-eins from Jurkat T cells could be identified. A later application of aimilar protocol was reported by Bodenmiller et al. when perform-ng an elaborate phosphosite assignment of D. melanogaster KC167ells [99].

.7. Ca and Ba precipitation

In 1994 Reynolds et al. used an excess of Ca2+ in 50% ethanol forrecipitating phosphopeptides from a tryptic casein hydrolysate. At

ower pH, only peptides containing multiple phosphoserines werenriched. At a pH of 8 all phosphopeptides except two monophos-horylated ones could be found in the precipitate [100]. Calciumhosphate/calcium phosphopeptide coprecipitation in combina-ion with two subsequent IMAC enrichment steps was used whennalyzing the phosphoproteome of rice embryonic cells. Peptidesere mixed with disodium phosphate and ammonia solution fol-

owed by the addition of calcium chloride at a pH of 10. In total42 phosphopeptides representing 125 phosphoproteins could be

dentified [101]. Although no tyrosine phosphorylated peptide wasetected in this study, this was attributed to the low abundance ofhese in rice, rather than lacking selectivity for it. A similar methodas used for the phosphoproteomic analysis of human brain by Xia

t al. allowing the identification of 551 phosphopeptides with 466nique phosphorylation sites [102]. Recently, also Ba2+ has beeneported to precipitate phosphopeptides in a study from Yates ando-workers [103].

.8. Method comparisons and combined methods

Multiple phosphospecific enrichment techniques can bemployed by combining different methods in parallel to obtain aore comprehensive view of the phosphoproteome. On the other

and it is possible to combine them in a serial workflow to maximizeelectivity. For example, several groups reported that using TiO2nrichment preferably singly phosphorylated peptides are isolated,hereas IMAC enrichment shows a bias towards multiply, predom-

nantly doubly phosphorylated peptides. This may be explained byhe different binding properties and/or some microstructural pref-rences, but also by the different peptide-to-resin ratios of bothaterials [104]. In many recent phosphoproteome studies TiO2 and

MAC enrichment are carried out in parallel [11,105] utilizing theifferent preferences of the methods. A comprehensive compar-

son of TiO2, IMAC and chemical (phosphoramidate) enrichmentas carried out by Bodenmiller et al. In their experiments they

btained good reproducibility within the same method when ana-yzing repeated isolates but the overlaps of peptide identificationsrom different methods were rather low (around 30% between all

ethods) [105]. In a recent work from our group the phosphopro-eome of HeLa cells was screened by combining different metalxide protocols (nanocast SnO2 and TiO2, commercial TiO2) in aarallel setup [74]. In sum, 1595 unique phosphopeptides were

ca Acta 703 (2011) 19– 30 27

identified. The overlap of identifications between the methods wasless than one third; almost 140 phosphopeptides were exclusivelyidentified by SnO2, but more surprisingly also both titania materi-als contributed more than 20% unique identifications each. Theseand other studies indicate that for a complementary coverage ofthe phosphoproteome, a combined approach may be necessary.

For two-dimensional fractionation/enrichment schemes, LCbased methods are used in the first dimension and fractions are col-lected according to resolving power of the respective method used.One has to pay attention that the elution conditions used for the firstdimension do not interfere with the following enrichment step. Pri-marily high salt concentration or low pH elution conditions, whichmay be used to elute peptides from prefractionation columns, cancause some problems with following enrichment methods. There-fore a desalting step or other sample buffer adjustment may berequired to allow further phosphopeptide enrichment. Trinidadet al. analyzed the postsynaptic density, a part of the mammaliancentral nervous system using SCX or SCX-IMAC phosphopeptideenrichment [106]. 88 SCX fractions were collected and subjectedeither to direct LC–MS analysis or to further IMAC enrichment. Inthe early fractions the highest numbers of phosphopeptides werefound, but phosphopeptides were identified throughout the wholefractionation. Thus 311 phosphopeptides were identified with SCXalone and with the addition of IMAC chromatography a three-fold increase of identifications could be achieved, resulting in atotal of 998 unique phosphopeptide identifications on 287 phos-phoproteins out of 1263 proteins which were identified in totalwith this approach. Using SILAC peptide labeling in combinationwith SCX-IMAC phosphopeptide enrichment and high accuracymass spectrometry with MS3 peptide sequencing, double aux-otroph yeast strains were analyzed for their pheromone responseby the Jensen group [53]. More than 700 phosphopeptides could beidentified of which 18% were regulated by pheromone response.Another comprehensive phosphoproteomic approach using SILAC-quantification, but using SCX-TiO2 enrichment and high resolutionmass spectrometry with multistage activation peptide sequencing[107] for accurate phosphopeptide identification was performedby Olsen et al. [13]. In this analysis of EGF-treated HeLa cells, morethan 99% confidence of phosphopeptide identifications could beachieved by combining high precursor mass accuracy informationand two stage fragmentation of peptides and 6600 phosphoryla-tion sites on a total of 2244 proteins could be identified. Fourteenpercent of identified phosphorylation sites were found to be mod-ulated more than twofold upon EGF stimulation. SCX fractionationand a combination of IMAC and TiO2 enrichment with high massaccuracy peptide identification and SILAC quantification was usedfor the analysis of the phosphorylation dependent cell cycle regu-lation of HeLa cells [11]. Cells were arrested in the G1 or M-phaseof the cell cycle and grown in media supplemented with differ-ently isotope labeled amino acids to be able to distinguish betweenthe different cell cycle phases. After SCX chromatography, frac-tions were split into halves and equally enriched with either TiO2or IMAC. Differently enriched fractions were pooled again andanalyzed by mass spectrometry. The combined setup allowed theidentification of over 14,000 sites of phosphorylation on 3682 pro-teins with over 1000 proteins showing increased phosphorylationduring mitosis. The authors reported nearly an order of magnitudeincrease in the number of sites identified from nocodazole arrestedHeLa cells compared to previous work [10].

Thingholm et al. developed a strategy for the sequentialseparation of monophosphorylated peptides and multiply phos-phorylated peptides by combining IMAC and MOAC in a sequential

fashion. The strategy, termed sequential elution from IMAC(SIMAC), uses IMAC in the first dimension and TiO2 for secondaryenrichment of IMAC flow through and eluates [108]. Peptides wereeluted from IMAC under acidic conditions (1% TFA, pH 1.0)—when

Page 10

2 Chimica Acta 703 (2011) 19– 30

pbpS(pimawto

4b

brpaI

4o

doUaaarpOaesm

4

msa[sstcmut

cosoiTyttC

8 A. Leitner et al. / Analytica

referably monophosphorylated peptides should elute, followedy TiO2 enrichment to remove remaining non-phosphorylatedeptides which may coelute from IMAC under these conditions.ubsequently basic elution with ammonium hydroxide solutionpH 11.3) is applied to IMAC to recover remaining multiply phos-horylated peptides. Combining the advantages of IMAC and TiO2

t was possible to identify 306 monophosphorylated peptides, 186ultiply phosphorylated peptides and 716 unique phosphosites of

tryptic digest of human mesenchymal stem cells. These numbersere in contrast to 232 monophosphorylated and just 54 mul-

iphosphorylated peptides and 350 unique sites when using anptimized TiO2 protocol alone.

. Analytical strategies for other classes of phosphorylatediomolecules

As already mentioned in the Introduction, other phosphorylatediomolecules such as phospholipids, DNA/RNA, and phospho-ylated metabolites play an important role in biological andharmaceutical research as well. Several strategies introducedbove have been adapted for the analysis of these analyte classes.n the following, we will only highlight a few examples.

.1. The phosphate group as affinity handle for the enrichment ofther PTMs

In a recent interesting application, Parker et al. intro-uced a phosphate group as an affinity tag for the analysisf peptides carrying an O-GlcNAc modification (Fig. 4) [109].sing an established chemoenzymatic labeling technique, thezidosugar, N-azidoacetylgalactosamine, was attached to the N-cetylglucosamine moiety. The azide group can be reacted withlkynes in a bioorthogonal 1,3-cycloaddition (“click chemistry”)eaction. In this case, cyclohexylammonium pent-4-ynyl phos-hate was used as the alkyne probe, enabling the isolation of-GlcNAc peptides using titanium dioxide-based MOAC. For a sep-rate enrichment of phosphopeptides and O-GlcNAc peptides, annzymatic dephosphorylation step was introduced. It seems fea-ible to use a similar approach also for other post-translationalodifications.

.2. Phospholipids

Phospholipids (PLs) are the main constituent of the cellularembrane bilayer but also play an important role in cellular

ignaling [110]. Lipidomics provides analytical strategies for thenalysis of lipid metabolism and lipid mediated signaling processes111]. Mass spectrometry is currently the method of choice fortraightforward lipidomic analysis [112,113], however, elaborateeparation or specific enrichment is also in this case essen-ial for efficient MS identification. Analytical tools such as gashromatography–mass spectrometry (GC–MS) and thin layer chro-atography (TLC) in combination with MALDI-TOF–MS can be

sed, but HPLC–ESI–MS is currently the most preferred combina-ion for lipidomic analysis [110,111,114].

In general, total lipids are extracted from biological samples likeells or tissue and extracts are separated by normal phase (on diolr amino phases) or reversed phase chromatography prior to masspectrometric identification [114]. For a more specific enrichmentf phospholipids, MOAC has been employed. This method was firstntroduced by Ikeguchi and Nakamura. In their approach they usediO2 for selective enrichment of PLs from egg yolk prior to LC anal-

sis. Recovery rates up to 70% for diverse PLs could be achieved andhe method was also effective for removal of fatty acids and neu-ral lipids [59]. A further refinement of the method was done byalvano et al. [115]. TiO2 was packed into frits allowing a fast and

Fig. 4. Enrichment O-GlcNAc-peptides by introduction of a phosphate group asaffinity handle after chemoenzymatic attachment of N-azidoacetylgalactosamine.Adapted from ref. [109].

uncomplicated SPE enrichment of PLs from dairy products. Simi-lar to the protocol of Larsen for phosphopeptide enrichment theyused DHB to reduce unspecific binding from non-phosphorylatedlipids. Only PLs could be identified in the elution fraction, prov-ing the high selectivity for PLs of the method. In a recent workfrom our group, ZrO2 packed into SPE cartridges was used to selec-tively isolate phosphatidylcholines (PCs) from natural samples likemilk, or human or mouse plasma [116]. Recovery rates up to 100%were achieved using an optimized extraction, incubation and elu-tion protocol. Using the protocol optimized for ZrO2, TiO2 and SnO2were also tested for PC enrichment, however ZrO2 performed supe-rior compared to other materials tested.

4.3. Nucleotides and nucleosides

The negatively charged phosphate diester backbone of DNA canalso serve as a ligand for purification with phosphoaffine enrich-ment materials. Due to the advancements of gene therapy andgenetic vaccines during the last decade, the demand for highly puri-fied plasmid DNA (pDNA) is still increasing [117]. pDNA is mainlyproduced by recombinant E. coli fermentation, so for research and

clinical applications pDNA has to be purified and separated fromgenomic DNA, RNA, remaining proteins and endotoxins. Secondly,only the super coiled conformation of pDNA (sc pDNA) is biolog-ically active and hence has to be separated from other isoforms

Page 11

Chimi

[socdftadnpnpw

ciga

atmbat

5

drrpptpatttpf

R

A. Leitner et al. / Analytica

118]. Therefore chromatographic methods such as size exclu-ion, hydrophobic interaction, hydroxyapatite, reversed phase,r thiophilic adsorption and affinity chromatography, mostly inombination with ion exchange chromatography as the secondimension are used [119]. To date, no phosphate specific methodor pDNA purification was reported, although Sousa et al. reportedhe complete separation of sc pDNA and open circular pDNA byrginine affinity chromatography. The specific recognition wasescribed to be the result of multiple interactions between argi-ine and pDNA, including electrostatic interaction with the pDNAhosphate backbone and also some degree of biorecognition ofucleotide bases by the arginine ligand [120]. A similar recognitionrinciple has been proposed for phosphoproteomic applications asell, using polyarginine or poylethyleneimine materials [121,122].

RNA–protein interactions have been probed by Urlaub ando-workers using UV-induced cross-linking and enrichment ofnteracting peptide-RNA oligonucleotide fragments using an inte-rated two-dimensional LC set-up that included a TiO2 column forffinity enrichment of the conjugates [123].

These examples show that also phosphate group-containingnalytes other than proteins and peptides need to be isolated, iden-ified and quantified in low concentrations in a complex biological

atrix. Therefore, specific enrichment methods for phosphorylatediomolecules that were initially developed for phosphoproteomenalysis may help to overcome limitations which currently hamperheir effective LC–MS analysis.

. Conclusions and outlook

The biological relevance of the phosphate group has led to theevelopment of various analytical tools to probe biomolecules car-ying this functional group. Mass spectrometry plays a centralole in these analytical workflows but researchers need to employrefractionation and enrichment methods to deal with the sam-le complexity in biological systems. In this respect, establishedechnologies continue to be refined while promising emergingrotocols are quickly adopted. Although impressive results havelready been obtained using available tools, the future will cer-ainly see the development of additional ones. We expect thathe increased emphasis on quantitative workflows in phosphopro-eomics will put the robustness and reproducibility of the completerotocol, from sample preparation to final MS analysis, into theocus.

eferences

[1] F.H. Westheimer, Science 235 (1987) 1173–1178.[2] R.N. Goldberg, N. Kishore, R.M. Lennen, J. Phys. Chem. Ref. Data 31 (2002)

231–370.[3] P.A. Connor, A.J. McQuillan, Langmuir 15 (1999) 2916–2921.[4] F. Wolfe-Simon, J. Switzer Blum, T.R. Kulp, G.W. Gordon, S.E. Hoeft, J. Pett-

Ridge, J.F. Stolz, S.M. Webb, P.K. Weber, P.C.W. Davies, A.D. Anbar, R.S.Oremland, Science 332 (2011) 1163–1166.

[5] T. Hunter, Cell 100 (2000) 113–127.[6] Y.P. Lim, Clin. Cancer Res. 11 (2005) 3163–3169.[7] H.C. Harsha, A. Pandey, Mol. Oncol. 4 (2010) 482–495.[8] A. Levitzki, Acc. Chem. Res. 36 (2003) 462–469.[9] T. Force, K. Kuida, M. Namchuk, K. Parang, J.M. Kyriakis, Circulation 109 (2004)

1196–1205.[10] S.A. Beausoleil, M. Jedrychowski, D. Schwartz, J.E. Elias, J. Villén, J. Li, M.A.

Cohn, L.C. Cantley, S.P. Gygi, Proc. Natl. Acad. Sci. U. S. A. 101 (2004)12130–12135.

[11] N. Dephoure, C. Zhou, J. Villén, S.A. Beausoleil, C.E. Bakalarski, S.J. Elledge, S.P.Gygi, Proc. Natl. Acad. Sci. U. S. A. 105 (2008) 10762–10767.

[12] S. Lemeer, R. Ruijtenbeek, M.W.H. Pinkse, C. Jopling, A.J.R. Heck, J. den Hertog,F.M. Slijper, Mol. Cell. Proteomics 6 (2007) 2088–2099.

[13] J.V. Olsen, B. Blagoev, F. Gnad, B. Macek, C. Kumar, P. Mortensen, M. Mann,

Cell 127 (2006) 635–648.

[14] J. Reinders, A. Sickmann, Proteomics 5 (2005) 4052–4061.[15] H.R. Bourne, Philos. Trans. R. Soc. Lond. B Biol. Sci. 349 (1995) 283–289.[16] H.R. Bourne, Nature 376 (1995) 727–729.[17] O. Lichtarge, H.R. Bourne, F.E. Cohen, J. Mol. Biol. 257 (1996) 342–358.

ca Acta 703 (2011) 19– 30 29

[18] H. Steen, J.A. Jebanathirajah, J. Rush, N. Morrice, M.W. Kirschner, Mol. Cell.Proteomics 5 (2006) 172–181.

[19] P.J. Boersema, S. Mohammed, A.J.R. Heck, J. Mass Spectrom. 44 (2009)861–878.

[20] A.M. Palumbo, S.A. Smith, C.L. Kalcic, M. Dantus, P.M. Stemmer, G.E. Reid, MassSpectrom. Rev. 30 (2011) 600–625.

[21] H. Jaffe, Veeranna, H.C. Pant, Biochemistry 37 (1998) 16211–16224.[22] C. Klemm, S. Schröder, M. Glückmann, M. Beyermann, E. Krause, Rapid Com-

mun. Mass Spectrom. 18 (2004) 2697–2705.[23] A. Leitner, A. Foettinger, W. Lindner, J. Mass Spectrom. 42 (2007) 950–959.[24] P.J. Ulintz, A.K. Yocum, B. Bodenmiller, R. Aebersold, P.C. Andrews, A.I.

Nesvizhskii, J. Proteome Res. 8 (2009) 887–899.[25] J.M. Wells, S.A. McLuckey, Meth. Enzymol. 402 (2005) 148–185.[26] J.V. Olsen, B. Macek, O. Lange, A. Makarov, S. Horning, M. Mann, Nat. Methods

4 (2007) 709–712.[27] N. Nagaraj, R.C.J. D‘Souza, J. Cox, J.V. Olsen, M. Mann, J. Proteome Res. 9 (2010)

6786–6794.[28] J.E.P. Syka, J.J. Coon, M.J. Schroeder, J. Shabanowitz, D.F. Hunt, Proc. Natl. Acad.

Sci. U. S. A. 101 (2004) 9528–9533.[29] D.L. Swaney, G.C. McAlister, J.J. Coon, Nat. Methods 5 (2008) 959–964.[30] K. Kandasamy, A. Pandey, H. Molina, Anal. Chem. 81 (2009) 7170–7180.[31] S. Kim, N. Mischerikow, N. Bandeira, J.D. Navarro, L. Wich, S. Mohammed,

A.J.R. Heck, P.A. Pevzner, Mol. Cell. Proteomics 9 (2010) 2840–2852.[32] R.-X. Sun, M.-Q. Dong, C.-Q. Song, H. Chi, B. Yang, L.-Y. Xiu, L. Tao, Z.-Y. Jing, C.

Liu, L.-H. Wang, Y. Fu, S.-M. He, J. Proteome Res. 9 (2010) 6354–6367.[33] F.M. White, Curr. Opin. Biotechnol. 19 (2008) 404–409.[34] B. Macek, M. Mann, J.V. Olsen, Ann. Rev. Pharmacol. Toxicol. 49 (2009)

199–221.[35] G. Palmisano, T.E. Thingholm, Expert Rev. Proteomics 7 (2010) 439–456.[36] R. Wu, N. Dephoure, W. Haas, E.L. Huttlin, B. Zhai, M.E. Sowa, S.P. Gygi, Mol.

Cell. Proteomics, in press, doi:10.1074/mcp.M111.009654.[37] T.E. Thingholm, M.R. Larsen, C.R. Ingrell, M. Kassem, O.N. Jensen, J. Proteome

Res. 7 (2008) 3304–3313.[38] N.L. Young, M.D. Plazas-Mayorca, B.A. Garcia, Expert Rev. Proteomics 7 (2010)

79–92.[39] M.O. Collins, L. Yu, J.S. Choudhary, Proteomics 7 (2007) 2751–2768.[40] A. Pandey, A.V. Podtelejnikov, B. Blagoev, X.R. Bustelo, M. Mann, H.F. Lodish,

Proc. Natl. Acad. Sci. U. S. A. 97 (2000) 179–184.[41] M. Gronborg, T.Z. Kristiansen, A. Stensballe, J.S. Andersen, O. Ohara, M. Mann,

O.N. Jensen, A. Pandey, Mol. Cell. Proteomics 1 (2002) 517–527.[42] J. Rush, A. Moritz, K.A. Lee, A. Guo, V.L. Goss, E.J. Spek, H. Zhang, X.M. Zha, R.D.

Polakiewicz, M.J. Comb, Nat. Biotechnol. 23 (2005) 94–101.[43] K. Rikova, A. Guo, Q. Zeng, A. Possemato, J. Yu, H. Haack, J. Nardone, K. Lee,

C. Reeves, Y. Li, Y. Hu, Z.P. Tan, M. Stokes, L. Sullivan, J. Mitchell, R. Wetzel, J.MacNeill, J.M. Ren, J. Yuan, C.E. Bakalarski, J. Villén, J.M. Kornhauser, B. Smith,D. Li, X. Zhou, S.P. Gygi, T.-L. Gu, R.D. Polakiewicz, J. Rush, M.J. Comb, Cell 131(2007) 1190–1203.

[44] Y. Zhang, A. Wolf-Yadlin, P.L. Ross, D.J. Pappin, J. Rush, D.A. Lauffenburger,F.M. White, Mol. Cell. Proteomics 4 (2005) 1240–1250.

[45] J. Porath, J. Carlsson, I. Olsson, G. Belfrage, Nature 258 (1975) 598–599.[46] L. Andersson, J. Porath, Anal. Biochem. 154 (1986) 250–254.[47] S.B. Ficarro, M.L. McCleland, P.T. Stukenberg, D.J. Burke, M.M. Ross, J. Sha-

banowitz, D.F. Hunt, F.M. White, Nat. Biotechnol. 20 (2002) 301–305.[48] M. Kokubu, Y. Ishihama, T. Sato, T. Nagasu, Y. Oda, Anal. Chem. 77 (2005)

5144–5154.[49] M.C. Posewitz, P. Tempst, Anal. Chem. 71 (1999) 2883–2892.[50] H. Zhou, M. Ye, J. Dong, G. Han, X. Jiang, R. Wu, H. Zou, J. Proteome Res. 7

(2008) 3957–3967.[51] S. Feng, M. Ye, H. Zhou, X. Jiang, X. Jiang, H. Zou, B. Gong, Mol. Cell. Proteomics

6 (2007) 1656–1665.[52] M.O. Collins, L. Yu, M.P. Coba, H. Husi, L. Campuzano, W.P. Blackstock, J.S.

Choudhary, S.G.N. Grant, J. Biol. Chem. 280 (2005) 5972–5982.[53] A. Gruhler, J.V. Olsen, S. Mohammed, P. Mortensen, N.J. Faergeman, M. Mann,

O.N. Jensen, Mol. Cell. Proteomics 4 (2005) 310–327.[54] X. Li, S.A. Gerber, A.D. Rudner, S.A. Beausoleil, W. Haas, J. Villén, J.E. Elias, S.P.

Gygi, J. Proteome Res. 6 (2007) 1190–1197.[55] M. Kawahara, H. Nakamura, T. Nakajima, Anal. Sci. 5 (1989) 485–486.[56] H. Matsuda, H. Nakamura, T. Nakajima, Anal. Sci. 6 (1990) 911–912.[57] K. Tani, Y. Suzuki, Chromatographia 46 (1997) 623–627.[58] Y. Ikeguchi, H. Nakamura, Anal. Sci. 13 (1997) 479–483.[59] Y. Ikeguchi, H. Nakamura, Anal. Sci. 16 (2000) 541–543.[60] M. Grün, A.A. Kurganov, S. Schacht, F. Schüth, K.K. Unger, J. Chromatogr. A 740

(1996) 1–9.[61] K. Tani, E. Miyamoto, J. Liq. Chromatogr. Relat. Technol. 22 (1999) 857–

871.[62] K. Tani, M. Ozawa, J. Liq. Chromatogr. Rel. Techn. 22 (1999) 843–856.[63] N. Sugiyama, T. Masuda, K. Shinoda, A. Nakamura, M. Tomita, Y. Ishihama,

Mol. Cell. Proteomics 6 (2007) 1103–1109.[64] A. Leitner, M. Sturm, J.-H. Smått, M. Järn, M. Lindén, K. Mechtler, W. Lindner,

Anal. Chim. Acta 638 (2009) 51–57.[65] M. Sturm, A. Leitner, J.-H. Smått, M. Lindén, W. Lindner, Adv. Funct. Mater. 18

(2008) 2381–2389.[66] M.W.H. Pinkse, P.M. Uitto, M.J. Hilhorst, B. Ooms, A.J.R. Heck, Anal. Chem. 76

(2004) 3935–3943.[67] M.R. Larsen, T.E. Thingholm, O.N. Jensen, P. Roepstorff, T.J.D. Jorgensen, Mol.

Cell. Proteomics 4 (2005) 873–886.

Page 12

3 Chimi

0 A. Leitner et al. / Analytica

[68] M. Mazanek, G. Mitulovic, F. Herzog, C. Stingl, J.R.A. Hutchins, J.-M. Peters, K.Mechtler, Nat. Protocols 2 (2007) 1059–1069.

[69] M. Mazanek, E. Roitinger, O. Hudecz, J.R.A. Hutchins, B. Hegemann, G. Mit-ulovic, T. Taus, C. Stingl, J.-M. Peters, K. Mechtler, J. Chromatogr. B 878 (2010)515–524.

[70] A. Leitner, Trends Anal. Chem. 29 (2010) 177–185.[71] H.K. Kweon, K. Håkansson, Anal. Chem. 78 (2006) 1743–1749.[72] N. Sugiyama, H. Nakagami, K. Mochida, A. Daudi, M. Tomita, K. Shirasu, Y.

Ishihama, Mol. Syst. Biol. 4 (2008) 193 (article number).[73] J.-H. Smått, N. Schüwer, M. Järn, W. Lindner, M. Lindén, Microporous Meso-

porous Mater. 112 (2008) 308–318.[74] A. Leitner, M. Sturm, O. Hudecz, M. Mazanek, J.-H. Smått, M. Lindén, W. Lind-

ner, K. Mechtler, Anal. Chem. 82 (2010) 2726–2733.[75] F. Wolschin, S. Wienkoop, W. Weckwerth, Proteomics 5 (2005) 4389–4397.[76] S.B. Ficarro, J.R. Parikh, N.C. Blank, J.A. Marto, Anal. Chem. 80 (2008)

4606–4613.[77] H.-Y. Lin, W.-Y. Chen, Y.-C. Chen, Anal. Bioanal. Chem. 394 (2009) 2129–

2136.[78] A. Lee, H.-J. Yang, E.-S. Lim, J. Kim, Y. Kim, Rapid Commun. Mass Spectrom. 22

(2008) 2561–2564.[79] W.-Y. Chen, Y.-C. Chen, Anal. Bioanal. Chem. 398 (2010) 2049–2057.[80] S. Mohammed, K. Kraiczek, M.W.H. Pinkse, S. Lemeer, J.J. Benschop, A.J.R. Heck,

J. Proteome Res. 156 (2008) 5–1571.[81] R. Raijmakers, K. Kraiczek, A.P. de Jong, S. Mohammed, A.J.R. Heck, Anal. Chem.

82 (2010) 824–832.[82] Y. Wagner, A. Sickmann, H.E. Mayer, G. Daum, J. Am. Soc. Mass Spectrom. 14

(2003) 1003–1011.[83] A. Motoyama, T. Xu, C.I. Ruse, J.A. Wohlschlegel, J.R. Yates III, Anal. Chem. 79

(2007) 3623–3634.[84] K. Zhang, Anal. Biochem. 357 (2006) 225–231.[85] G. Han, M. Ye, H. Zhou, X. Jiang, S. Feng, X. Jiang, R. Tian, D. Wan, H. Zou, J. Gu,

Proteomics 8 (2008) 1346–1361.[86] D.E. McNulty, R.S. Annan, Mol. Cell. Proteomics 7 (2008) 971–980.[87] S.D. Garbis, T.I. Roumeliotis, S.I. Tyritzis, K.M. Zorpas, K. Palavkis, C.A. Con-

stantinides, Anal. Chem. 83 (2011) 708–718.[88] C.-J. Wu, Y.-W. Chen, J.-H. Tai, S.-H. Chen, J. Proteome Res. 10 (2011)

1088–1097.[89] A.J. Alpert, Anal. Chem. 80 (2008) 62–76.[90] C.S. Gan, T.N. Guo, H. Zhang, S.K. Lim, S.K. Sze, J. Proteome Res. 7 (2008)

4869–4877.[91] P. Hao, T. Guo, S.K. Sze, PloS ONE 6 (2011) e16884 (article number).[92] A. Leitner, W. Lindner, Methods Mol. Biol. 527 (2009) 229–243.[93] Y. Oda, T. Nagasu, B.T. Chait, Nat. Biotechnol. 19 (2001) 379–382.

[94] Z.A. Knight, B. Schilling, R.H. Row, D.M. Kenski, B.W. Gibson, K.M. Shokat, Nat.

Biotechnol. 21 (2003) 1047–1054.[95] G. Arrigoni, S. Resjo, F. Levander, R. Nilsson, E. Degerman, M. Quadroni, L.A.

Pinna, P. James, Proteomics 6 (2006) 757–766.[96] D.T. McLachlin, B.T. Chait, Anal. Chem. 75 (2003) 6826–6836.

ca Acta 703 (2011) 19– 30

[97] H. Zhou, J.D. Watts, R. Aebersold, Nat. Biotechnol. 19 (2001) 375–378.[98] W.A. Tao, B. Wollscheid, R. O’Brien, J.K. Eng, X.J. Li, B. Bodenmiller, J.D. Watts,

L. Hood, R. Aebersold, Nat. Methods 2 (2005) 591–598.[99] B. Bodenmiller, L.N. Mueller, P.G.A. Pedrioli, D. Pflieger, M.A. Jünger, J.K. Eng,

R. Aebersold, W.A. Tao, Mol. BioSyst. 3 (2007) 275–286.[100] E.C. Reynolds, P.F. Riley, N.J. Adamson, Anal. Biochem. 217 (1994)

277–284.[101] X. Zhang, J. Ye, O.N. Jensen, P. Roepstorff, Mol. Cell. Proteomics 6 (2007)

2032–2042.[102] Q. Xia, D. Cheng, D.M. Duong, M. Gearing, J.J. Lah, A.I. Levey, J. Peng, J. Proteome

Res. 7 (2008) 2845–2851.[103] C.I. Ruse, D.B. McClatchy, B. Lu, D. Cociorva, A. Motoyama, S.K. Park, J.R. Yates

III, J. Proteome Res. 7 (2008) 2140–2150.[104] Q.-R. Li, Z.-B. Ning, J.-S. Tang, S. Nie, R. Zeng, J. Proteome Res. 8 (2009)

5375–5381.[105] B. Bodenmiller, L.N. Mueller, M. Mueller, B. Domon, R. Aebersold, Nat. Methods

4 (2007) 231–237.[106] J.C. Trinidad, C.G. Specht, A. Thalhammer, R. Schoepfer, A.L. Burlingame, Mol.

Cell. Proteomics 5 (2006) 914–922.[107] M.J. Schroeder, J. Shabanowitz, J.C. Schwartz, D.F. Hunt, J.J. Coon, Anal. Chem.

76 (2004) 3590–3598.[108] T.E. Thingholm, O.N. Jensen, P.J. Robinson, M.R. Larsen, Mol. Cell. Proteomics

7 (2008) 661–671.[109] B.L. Parker, P. Gupta, S.J. Cordwell, M.R. Larsen, G. Palmisano, J. Proteome Res.

10 (2011) 1449–1458.[110] C. Wang, H.W. Kong, Y.F. Guan, J. Yang, J.R. Gu, S.L. Yang, G.W. Xu, Anal. Chem.

77 (2005) 4108–4116.[111] C. Wang, J. Yang, J. Nie, Biomed. Chromatogr. 23 (2009) 1079–1085.[112] P.T. Ivanova, S.B. Milne, D.S. Myers, H.A. Brown, Curr. Opin. Chem. Biol. 13

(2009) 526–531.[113] M.B. Khalil, W. Hou, H. Zhou, F. Elisma, L.A. Swayne, A.P. Blanchard, Z. Yao,

S.A.L. Bennett, D. Figeys, Mass Spectrom. Rev. 29 (2010) 877–929.[114] H.K. Min, G. Kong, M.H. Moon, Anal. Bioanal. Chem. 396 (2010) 1273–1280.[115] C.D. Calvano, O.N. Jensen, C.G. Zambonin, Anal. Bioanal. Chem. 394 (2009)

1453–1461.[116] A. Gonzalvez, B. Preinerstorfer, W. Lindner, Anal. Bioanal. Chem. 396 (2010)

2965–2975.[117] J. Urthaler, R. Schlegl, A. Podgornik, A. Strancar, A. Jungbauer, R. Necina, J.

Chromatogr. A 1065 (2005) 93–106.[118] F. Sousa, D.M. Prazeres, J.A. Queiroz, Biomed. Chromatogr. 23 (2009) 160–165.[119] J. Stadler, R. Lemmens, T. Nyhammar, J. Gene Med. 6 (2004) S54–S66.[120] F. Sousa, T. Matos, D.M.F. Prazeres, J.A. Queiroz, Anal. Biochem. 374 (2008)

432–434.

[121] C.-K. Chang, C.-C. Wu, Y.-S. Wang, H.-C. Chang, Anal. Chem. 80 (2008)

3791–3797.[122] C.-T. Chen, L.-Y. Wang, Y.-P. Ho, Anal. Bioanal. Chem. 399 (2011) 2795–2806.[123] F.M. Richter, H.-H. Hsiao, U. Plessmann, H. Urlaub, Biopolymers 91 (2009)

297–309.