transcription factors, their activity depends on the recruitmentof coactivators or corepressors, and it has been suggested thatPTMs regulate these interactions.7 Moreover, the localizationof FoxO transcription factors is also regulated by PPIs thatdepend on PTMs. Similarly, the binding partners of RNApolymerase II, which include transcription factors, chromatin-modifying factors, and RNA processing factors, are thought torecognize PTM patterns on its C-terminal domain, and awriter−reader−eraser model has been proposed.8 Themodulation of pRb PPIs by PTMs has been carefully studied.9

In short, its interaction with E2F-1 is dependent on thephosphorylation state of pRb, and its interactions with HP1,L3MBTL1, and 53BP1 are modulated by three distinctmethylation sites.9,11 Moreover, acetylation of pRb increasesits affinity for Mdm2 and has also been proposed to recruitbromodomain-containing proteins.9 Finally, the high number ofPTMs on chaperone proteins, some of the most highlyinteracting of all proteins due to their role in protein folding,12

suggests that a PTM code could participate in recruitingdifferent cochaperones and allocating each chaperone modformto a different type of protein.10 In fact, it is known that differentcochaperones can recruit different client proteins12 and thatPTMs can regulate the binding of chaperones to cochaperones.For example, C-terminal phosphorylation of Hsp70 and Hsp90shifts their binding affinity from the cochaperone CHIP to thecochaperone HOP,13 and acetylation of Hsp90 increases itsaffinity to several cochaperones.14 In short, all of theaforementioned proteins are known to participate in manyPPIs and to feature several PTMs; there is also evidence thattheir affinity to different interactors is PTM-regulated.

■ INTERACTION CODES

Here we propose that PTM codes that specifically modulatePPIs, such as those previously referred to6−10 and moregenerally proposed by Wilkins and Kummerfeld,15 be termedinteraction codes. We also propose that they will involvespecific types of PTMs rather than all types of PTMs that occurwithin the cell. The histone code and other newly proposedinteraction codes should thus not be considered as isolated or

rare examples; interaction codes of varying degrees ofcomplexity are likely to be widespread in the proteome. Aminimal interaction code would consist of two PTM sites, ofthe same or different type, that modulate the interactions of aprotein with at least two different interactors. More complexcodes will involve a greater number of PTMs, which modulate agreater number of interactions (Figure 1A). Conceptually,different types of codes can be imagined. In a protein with oneinteraction interface but multiple interaction partners, thePTMs on that interface, alone or in combination, can modulateinteractions in a mutually exclusive manner. Alternatively, aprotein with multiple interaction interfaces but only onepartner per interface might have one PTM per interface; thePTMs would switch each interface’s particular interaction on oroff. A third scenario, which is in between the previous two, alsoexists whereby a protein has multiple interaction interfaces andmore than one PTM and interactor per interface. In this case,the total interactions of the protein will depend on the type andtotal number of PTMs present.The search for new interaction codes could be accelerated if

promising protein candidates were easily identifiable. Thechallenge is how can these proteins be found? We suggest thatPPI networks, coanalyzed with PTM and other relevant data,can serve as a means to predict which proteins are likely tofeature an interaction code. We will consider the situation ofthe complex code rather than the minimal interaction code;these involve proteins with many interactions and many PTMsand should thus be more easily found.Hubs in Protein Networks Are Likely Candidates forInteraction Codes

In a PPI network, proteins with a large number of interactions(five or more in yeast) are termed hub proteins or hubs. Theseshould be evaluated as candidates for interaction codes.However, there are two types of hub proteins, “date” and“party” hubs (Figure 2). “Date hub” proteins are those thathave a large number of interaction partners but a small numberof interaction interfaces.16 They cannot interact with all of theirpartners at once, raising the possibility that PTMs are used tospecifically control their interactions at one or more interfaces.

Figure 1. Logic of interaction codes. (A) Interaction codes are likely to exist on proteins that have two or more PTMs and few interaction surfaces. Aminimal code is composed of two PTMs regulating two different interactions. Two types of minimal interaction codes can be imagined: (i) the twoPTMs occur on the same interaction interface, thereby modulating interactions in a mutually exclusive manner, or (ii) they can instead occur on twodistinct interfaces and switch each interaction on or off independently. (In the illustrated example, the yellow PTM promotes an interaction, whereasthe blue PTM blocks an interaction.) More complex types of interaction codes (iii) would feature several PTMs per interface, and PTMs might act incombination to promote or block interactions. Gray disks represent a protein; colored disks represent interaction partners; hatched areas representinteraction interfaces; colored symbols represent PTMs; and dotted colored symbols represent unmodified PTM sites. (B) PTM crosstalk underpinsinteraction codes by limiting the number of modforms of a given protein. PTMs might block or promote each other in a number of ways: (i) PTMscan compete for the same residue; (ii) a PTM can recruit a modifying enzyme to further modify a protein; (iii) a PTM can block a modifyingenzyme, preventing further modification of the protein; and (iv) a PTM can promote structural changes in a protein, exposing (or concealing, notshown) another PTM site.

Journal of Proteome Research Perspective

dx.doi.org/10.1021/pr500585p | J. Proteome Res. 2014, 13, 5333−53385334

Dow

nloa

ded

by N

AT

L L

BR

Y O

F SE

RB

IA o

n Se

ptem

ber

11, 2

015

| http

://pu

bs.a

cs.o

rg

Pub

licat

ion

Dat

e (W

eb):

Nov

embe

r 12

, 201

4 | d

oi: 1

0.10

21/p

r500

585p

“Party hub” proteins are members of large, stable molecularmachines. In networks, these are identifiable as proteins with acharacteristically large number of PPIs shared among all ormany members of the molecular machine. The “party hub”proteins tend to interact with all of their partners at the sameplace and time,16 having less biological need for regulation ofinteractions. The search for interaction codes should thus focuson “date hubs”; for example, there are 91 of these reported inyeast,16 of which 36 have just one or two interaction interface.17

Hubs like these should then be examined in the context of theirknown PTMs.Modifications and Enzyme−Substrate RelationshipsHighlight Interaction Codes

Do all types of PTMs participate in interaction codes? In thehistone code, as well as in the recently described interactioncodes, it is notable that a small subset of PTMs is used in all fiveknown codes. To date, these are phosphorylation, acetylation,methylation, O-GlcNAc, and ubiquitination or ubiquitin-likemodifications (Table 1). ADP-ribosylation is also part of threecodes. This is not to say that interaction codes are necessarilylimited to only these modifications, but these PTMs are likelyto form the “core” alphabet. Interestingly, PTMs involved inknown codes occur on only a few amino acids: lysine(methylation, acetylation, ubiquitin-like modifications), serine,threonine (phosphorylation, O-GlcNAc), arginine (methyla-tion, ADP-ribosylation), and tyrosine (phosphorylation).Moreover, except for ADP-ribosylation, these types of PTMwere shown to coevolve, indicating some sort of functionalassociation.18 It is also notable that code-associated PTMsaffect a range of physicochemical properties of their covalentlylinked amino acids. This concurs with the idea that PTMsinvolved in interaction codes are able to create or block

interactions by changing the shape, hydrophobicity, or chargeof a surface. Ultimately, this can affect domain−domain ordomain−motif interactions; for example, the phosphorylationof a tyrosine can modulate interactions with an Src-homology-2(SH2) domain-carrying partner, and trimethylation of lysinecan be specifically recognized by the chromodomain.19

The existence of a core set of modifications in interactioncodes suggests that date hubs should be examined for thesePTMs. Those hubs that have two or more core PTMs maycarry an interaction code, with those carrying larger numbersbeing of most interest. PTM-associated information can beintegrated with PPI networks in two ways to assist in thediscovery of proteins with interaction codes. First is theintegration of PTM types and their incidence into networks, asdemonstrated by us in 200820 and by Woodsmith and Stelzl in2014.21 Many proteome-scale studies of PTMs now exist, inmodel systems and human cells,4 providing extensive data forthis purpose. Second is the integration of PTM-associatedenzyme−substrate relationships with PPI networks, such askinase−substrate or methyltransferase−substrate interactions.22These enzyme−substrate relationships can be easily visualizedin PPI networks, for example, with the use of colored edges(Figure 2), and constitute a useful strategy to identify “datehubs” targeted by relevant enzymes. As previously suggested,hubs subject to many enzyme−substrate interactions will bethose most likely to carry an interaction code. Unfortunately,however, there is a need for more large-scale enzyme−substratedata concerning protein modifications. Two-hybrid screens,used to detect PPIs, do not typically confirm enzyme−substraterelationships because modifying enzymes also interact withnonsubstrate partners and enzyme−substrate interactions canbe too transient to be detected by two-hybrid techniques.23

Large-scale unbiased enzyme−substrate mapping studies,employing other techniques such as the proteome arrays usedin the global analysis of protein phosphorylation in yeast,24 willbe critical for the construction of modifying enzyme-proteinsubstrate relationships that can be integrated with PPInetworks.

Filtering of Candidates with Structural and LocalizationData

Candidate proteins for interaction codes, discovered aspreviously described, can finally be filtered with structural andintracellular localization data. Structural data, when available,are useful to verify that PTM sites are located at interactionsurfaces and thus accessible to potential interactors.25 Sites thatare buried in the core of proteins, although rare, are unlikely toparticipate in interaction codes. Data on intrinsic disorder ofproteins are also relevant. In fact, many types of PTMs have apreference for occurrence in disordered regions;26 such regionscan participate in PPI via disorder-to-order transitions, andPTMs have been proposed as a mechanism to modulate such

Figure 2. Integration of PPI networks and enzyme−substraterelationships can reveal interaction codes. (A) PPI network (whitenodes, proteins; gray edges, PPIs) contains “party hubs” (protein A;note the dense network) and “date hubs” (proteins B and C). (B)Integration of enzyme−substrate relationships into a PPI network(white nodes, proteins; colored nodes, modifying enzymes; gray edges,PPIs; colored directed edges, enzyme−substrate relationships) revealsthat protein B is likely to feature an interaction code.

Table 1. PTMs Involved in Interaction Codes

protein modificationsa references

histones Pho, Ac, Met, O-GlcNAc, Ubq, ADP-ribosylation 3,39FoxO transcription factors Pho, Ac, Met, O-GlcNAc, Ubq, ADP-ribosylation 7,40p53 Pho, Ac, Met, O-GlcNAc, Ubq, ADP-ribosylation 6Hsp90 Pho, Ac, Met, O-GlcNAc, Ubq 13,14,41pRb Pho, Ac, Met, O-GlcNAc, Ubq 9,42RNA polymerase II Pho, Ac, Met, O-GlcNAc, Ubq 8,43

aAbbreviations: Pho, phosphorylation; Ac, acetylation; Met, methylation; Ubq, ubiquitination.



Dow

nloa

ded

by N

AT

L L

BR

Y O

F SE

RB

IA o

n Se

ptem

ber

11, 2

015

| http

://pu

bs.a

cs.o

rg

Pub

licat

ion

Dat

e (W

eb):

Nov

embe

r 12

, 201

4 | d

oi: 1

0.10

21/p

r500

585p

interactions.27 Finally, data on subcellular localization or tissue-specificity of interaction partners and modifying enzymes maybe helpful to filter out false-positives. If, for example, a humandate hub has many interactors but its expression occurs inmutually exclusively tissues, an interaction code would not berequired to control interaction specificity. The different types ofdata to integrate with PPI networks in the search for interactioncodes and the corresponding lines of evidence are summarizedin Table 2.

■ CROSSTALK IN INTERACTION CODESIn addition to directly regulating interactions, PTMs can alsoregulate each other. This is known as PTM crosstalk (Figure1B) and will be featured in codes on interaction interfaces thatinvolve many modifications and interactors. Because PTMcrosstalk limits the number of possible modforms of a protein,not all PTMs can occur on a protein at the same time. As aconsequence, neither can all PTM-mediated PPIs.The simplest type of PTM crosstalk is where more than one

type of PTM can compete for the same residue (Figure 1B,example i). This type of crosstalk is most striking in the case oflysine residues, which can harbor many different types of PTMs(see previous).28 For example, in p53, there exists competitionbetween methylation, acetylation, and ubiquination on severalC-terminal lysines, modulating the fate of the protein.29 A morecomplex type of crosstalk is where one PTM is required topromote the modification of a second, often nearby PTM site(Figure 1B, example ii). This is the case of methylation ofarginine 27 of STAT6, which is required for STAT6phosphorylation.30 Alternatively, a PTM can block anotherPTM on a proximal site (Figure 1B, example iii). In this case,the crosstalk can be bidirectional, whereby each PTM will blockthe other, or unidirectional, where one PTM can block theother, but not the converse. In pRb, methylation of lysine 810

prevents phosphorylation of nearby serine 807 by disturbingthe interaction between pRB and the Cdk kinase. Phosphor-ylation of pRb, however, has no impact on methylation of lysine810,31 making this an example of unidirectional crosstalk.Finally, it should be noted that crosstalk is possible betweendistant PTM sites through allosteric effects32 (Figure 1B,example iv) or when two PTM sites that are distant on theprimary sequence of a protein are close in 3-D structure.To summarize, there are complexities of protein post-

translational modifications that constitute crosstalk. Some ofthese will affect the interactions of modifying enzymes withtheir substrate proteins. Equally, PTM crosstalk can affectnonenzyme protein−protein interaction specificity and, in thecontext of date hub proteins, is likely to be associated withmany interaction codes.

■ NEW CANDIDATES FOR INTERACTION CODESThe principles of interaction-code-containing proteins, pre-viously outlined, are best illustrated by example. Erce et al.recently constructed a yeast methylproteome network,embedded within a kinase−substrate network.22 We haveextracted two hub proteins from the methylproteome networks(histone Hht1p, well-characterized as featuring an interactioncode and mRNA transport protein Npl3p) and covisualizedthem with one hub protein extracted from the entire yeastinteractome (spliceosome-associated protein Brr2p) andknown enzyme−substrate relationships (Figure 3). A hubprotein that features an interaction code, namely, histoneHht1p, is different from a hub protein that does not, such asspliceosome-associated protein Brr2p. Both proteins participatein more than 10 PPIs; however, Brr2p harbors only a fewphosphorylation sites by Cdk1, whereas Hht1p is targeted bymany enzymes and carries many more PTMs. Brr2p featuresthe typical topology of a “party hub”, with many interactions

Table 2. Data Integration for the Discovery of Interaction Codes

data evidence for an interaction code

PPI data sets node appears as a “date hub”enzyme−substrate relationships node is targeted by two or more modifying enzymes; a greater number indicates greater likelihood of an interaction codePTM sites two or more PTMs; a greater number indicates greater likelihood of an interaction codePTM sites−interaction interface PTM sites occur on interaction surfaces; evidence is stronger if several types of PTM occur on the same interaction interfacePTM sites−structural disorder PTM sites occur on intrinsically disorganized domains; evidence is stronger if several types of PTM occur on the same domainsubcellular localization interaction code-containing protein candidate, its modifying enzymes, and its interactors are colocalized

Figure 3. Examples of hub proteins from the yeast interactome. Histone protein Hht1p features an interaction code. Note the number of interactionsand enzymes targeting Hht1p. Npl3p features a “date hub” topology and is a promising candidate for a new interaction code. In contrast, Brr2pfeatures the topology of a “party hub” protein, with tens of interactions but harbors only one type of PTM and is unlikely to feature an interactioncode. White nodes, proteins; orange nodes, kinases; red nodes, methyltransferases; green nodes, acetyltransferases; gray edges, PPIs; orange edges,phosphorylation; red edges, methylation; green edges, acetylation. Data from Erce et al.,22 UniProtKB, and NetworKIN, visualized with Cytoscape.38



Dow

nloa

ded

by N

AT

L L

BR

Y O

F SE

RB

IA o

n Se

ptem

ber

11, 2

015

| http

://pu

bs.a

cs.o

rg

Pub

licat

ion

Dat

e (W

eb):

Nov

embe

r 12

, 201

4 | d

oi: 1

0.10

21/p

r500

585p

and many PPIs between its interactors, resulting in a very densenetwork that indicates the existence of a large molecularmachine (in this case, the spliceosome). The network topologyof Hht1p, at first glance, might indicate a stable complex aswell. However, the network topologies of “date hubs” are quitevaried33 and Hht1p is, in fact, such a “date hub”. The density ofits network is due to the fact that Hht1p participates in manydifferent complexes, as opposed to simple PPIs. The topologyof the network in Figure 3 suggests Npl3p as an interestingcandidate for an interaction code. Npl3p is a highly interactingRNA-binding protein that is targeted by one methyltransferase,several kinases, and one phosphatase. Moreover, it features thetypical topology of a “date hub”, with many interactions butvery few PPIs between its interactors. Using a conditional two-hybrid system, Erce et al. demonstrated that methylation ofNpl3p by Hmt1p modulates several, but not all, of its PPIs.34

The fact that Npl3p can also be phosphorylated by a number ofkinases is likely to explain this observation, as phosphorylationmay modulate some of its other interactions. This is consistentwith the phosphorylation−methylation interplay that regulatesNpl3p import and export from the cell nucleus.35 Npl3p is notcurrently known to harbor as many types of PTMs as thehistones or p53, but not all interaction codes will necessarilyshare the same degree of complexity. What is important tonote, however, is that Npl3p harbors at least two physicochemi-cally distinct types of PTMs and has a relatively large number ofknown modification sites (18 methylarginines36 and at least 2phosphoserines37). A careful examination of integrated net-works, such as those in yeast from high-throughput protein−protein interactions and enzyme−substrate interactions, alongwith known PTMs from large-scale mass spectrometric screensand other relevant data, should reveal many other proteins thatare subject to interaction codes.

■ CONCLUSIONS

We have discussed how PPI networks, integrated withknowledge of protein PTMs, will help identify proteins withnew interaction codes. Minimal interaction codes are likely toexist on proteins that have two or more modifications on aninteraction interface and two or more interaction partners. Bycontrast, complex interaction codes are likely to be found ondate hub proteins that have many PTMs or are targeted bymany modifying and demodifying enzymes. Increasinglycomplete and accurate proteome-scale data for PTMs and theestablishment of detailed relationships between PTM modify-ing enzymes and their substrates will underpin this process ofdiscovery.

■ AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected].

Funding

D.L.W. acknowledges support from the University of NewSouth Wales. M.A.E. acknowledges support from the Universityof New South Wales. M.R.W. acknowledges support from theAustralian Research Council, the Australian Federal Govern-ment EIF Super Science scheme and the New South WalesState Government Science Leveraging Fund.

Notes

The authors declare no competing financial interest.

■ ACKNOWLEDGMENTS

D.L.W., M.A.E., and M.R.W. acknowledge the assistance ofAidan Tay in the generation of Figure 3.

■ ABBREVIATIONS

Ac, acetylation; Met, methylation; O-GlcNAc, O-linked N-acetylglucosamination; Pho, phosphorylation; PPI, protein−protein interaction; PTM, post-translational modification; Ubq,ubiquitination

■ REFERENCES(1) Jensen, O. N. Interpreting the protein language using proteomics.Nat. Rev. Mol. Cell Biol. 2006, 7 (6), 391−403.(2) Prabakaran, S.; Lippens, G.; Steen, H.; Gunawardena, J. Post-translational modification: nature’s escape from genetic imprisonmentand the basis for dynamic information encoding. Wiley Interdiscip. Rev.:Syst. Biol. Med. 2012, 4 (6), 565−583.(3) Rando, O. J. Combinatorial complexity in chromatin structureand function: revisiting the histone code. Curr. Opin. Genet. Dev. 2012,22 (2), 148−155.(4) (a) Humphrey, S. J.; Yang, G.; Yang, P.; Fazakerley, D. J.; Stockli,J.; Yang, J. Y.; James, D. E. Dynamic adipocyte phosphoproteomereveals that Akt directly regulates mTORC2. Cell Metab. 2013, 17 (6),1009−1020. (b) Choudhary, C.; Kumar, C.; Gnad, F.; Nielsen, M. L.;Rehman, M.; Walther, T. C.; Olsen, J. V.; Mann, M. Lysine acetylationtargets protein complexes and co-regulates major cellular functions.Science 2009, 325 (5942), 834−840. (c) Guo, A.; Gu, H.; Zhou, J.;Mulhern, D.; Wang, Y.; Lee, K. A.; Yang, V.; Aguiar, M.; Kornhauser,J.; Jia, X.; Ren, J.; Beausoleil, S. A.; Silva, J. C.; Vemulapalli, V.;Bedford, M. T.; Comb, M. J. Immunoaffinity enrichment and massspectrometry analysis of protein methylation. Mol. Cell. Proteomics2014, 13 (1), 372−387.(5) (a) Venne, A. S.; Kollipara, L.; Zahedi, R. P., The next level ofcomplexity: Crosstalk of posttranslational modifications. Proteomics2013;. (b) Sims, R. J., 3rd; Reinberg, D. Is there a code embedded inproteins that is based on post-translational modifications? Nat. Rev.Mol. Cell Biol. 2008, 9 (10), 815−820.(6) Gu, B.; Zhu, W. G. Surf the post-translational modificationnetwork of p53 regulation. Int. J. Biol. Sci. 2012, 8 (5), 672−684.(7) Calnan, D. R.; Brunet, A. The FoxO code. Oncogene 2008, 27(16), 2276−2288.(8) Eick, D.; Geyer, M. The RNA polymerase II carboxy-terminaldomain (CTD) code. Chem. Rev. 2013, 113 (11), 8456−8490.(9) Munro, S.; Carr, S. M.; La Thangue, N. B. Diversity within thepRb pathway: is there a code of conduct? Oncogene 2012, 31 (40),4343−4352.(10) Cloutier, P.; Coulombe, B. Regulation of molecular chaperonesthrough post-translational modifications: decrypting the chaperonecode. Biochim. Biophys. Acta 2013, 1829 (5), 443−454.(11) Carr, S. M.; Munro, S.; Zalmas, L. P.; Fedorov, O.; Johansson,C.; Krojer, T.; Sagum, C. A.; Bedford, M. T.; Oppermann, U.; LaThangue, N. B. Lysine methylation-dependent binding of 53BP1 tothe pRb tumor suppressor. Proc. Natl. Acad. Sci. U.S.A. 2014, 111 (31),11341−11346.(12) Rohl, A.; Rohrberg, J.; Buchner, J. The chaperone Hsp90:changing partners for demanding clients. Trends Biochem. Sci. 2013, 38(5), 253−262.(13) Muller, P.; Ruckova, E.; Halada, P.; Coates, P. J.; Hrstka, R.;Lane, D. P.; Vojtesek, B. C-terminal phosphorylation of Hsp70 andHsp90 regulates alternate binding to co-chaperones CHIP and HOPto determine cellular protein folding/degradation balances. Oncogene2013, 32 (25), 3101−3110.(14) Scroggins, B. T.; Robzyk, K.; Wang, D.; Marcu, M. G.;Tsutsumi, S.; Beebe, K.; Cotter, R. J.; Felts, S.; Toft, D.; Karnitz, L.;Rosen, N.; Neckers, L. An acetylation site in the middle domain ofHsp90 regulates chaperone function. Mol. Cell 2007, 25 (1), 151−159.



Dow

nloa

ded

by N

AT

L L

BR

Y O

F SE

RB

IA o

n Se

ptem

ber

11, 2

015

| http

://pu

bs.a

cs.o

rg

Pub

licat

ion

Dat

e (W

eb):

Nov

embe

r 12

, 201

4 | d

oi: 1

0.10

21/p

r500

585p

mailto:[email protected]

(15) Wilkins, M. R.; Kummerfeld, S. K. Sticking together? Fallingapart? Exploring the dynamics of the interactome. Trends Biochem. Sci.2008, 33 (5), 195−200.(16) (a) Han, J. D.; Bertin, N.; Hao, T.; Goldberg, D. S.; Berriz, G.F.; Zhang, L. V.; Dupuy, D.; Walhout, A. J.; Cusick, M. E.; Roth, F. P.;Vidal, M. Evidence for dynamically organized modularity in the yeastprotein-protein interaction network. Nature 2004, 430 (6995), 88−93.(b) Kim, P. M.; Lu, L. J.; Xia, Y.; Gerstein, M. B. Relating three-dimensional structures to protein networks provides evolutionaryinsights. Science 2006, 314 (5807), 1938−1941.(17) Goel, A.; Wilkins, M. R. Dynamic hubs show competitive andstatic hubs non-competitive regulation of their interaction partners.PloS One 2012, 7 (10), e48209.(18) Minguez, P.; Parca, L.; Diella, F.; Mende, D. R.; Kumar, R.;Helmer-Citterich, M.; Gavin, A. C.; van Noort, V.; Bork, P.Deciphering a global network of functionally associated post-translational modifications. Mol. Syst. Biol. 2012, 8, 599.(19) Seet, B. T.; Dikic, I.; Zhou, M. M.; Pawson, T. Reading proteinmodifications with interaction domains. Nat. Rev. Mol. Cell Biol. 2006,7 (7), 473−483.(20) Ho, E.; Webber, R.; Wilkins, M. R. Interactive three-dimensionalvisualization and contextual analysis of protein interaction networks. J.Proteome Res. 2008, 7 (1), 104−112.(21) Woodsmith, J.; Stelzl, U. Studying post-translational mod-ifications with protein interaction networks. Curr. Opin. Struct. Biol.2014, 24, 34−44.(22) Erce, M. A.; Pang, C. N.; Hart-Smith, G.; Wilkins, M. R. Themethylproteome and the intracellular methylation network. Proteomics2012, 12 (4−5), 564−586.(23) Westermarck, J.; Ivaska, J.; Corthals, G. L. Identification ofprotein interactions involved in cellular signaling. Mol. Cell. Proteomics2013, 12 (7), 1752−1763.(24) Ptacek, J.; Devgan, G.; Michaud, G.; Zhu, H.; Zhu, X.; Fasolo, J.;Guo, H.; Jona, G.; Breitkreutz, A.; Sopko, R.; McCartney, R. R.;Schmidt, M. C.; Rachidi, N.; Lee, S. J.; Mah, A. S.; Meng, L.; Stark, M.J.; Stern, D. F.; De Virgilio, C.; Tyers, M.; Andrews, B.; Gerstein, M.;Schweitzer, B.; Predki, P. F.; Snyder, M. Global analysis of proteinphosphorylation in yeast. Nature 2005, 438 (7068), 679−684.(25) (a) Pang, C. N.; Hayen, A.; Wilkins, M. R. Surface accessibilityof protein post-translational modifications. J. Proteome Res. 2007, 6(5), 1833−1845. (b) Vandermarliere, E.; Martens, L. Protein structureas a means to triage proposed PTM sites. Proteomics 2013, 13 (6),1028−1035.(26) Gao, J.; Xu, D. Correlation between posttranslationalmodification and intrinsic disorder in protein. Pac. Symp. Biocomput.2012, 94−103.(27) Hsu, W. L.; Oldfield, C. J.; Xue, B.; Meng, J.; Huang, F.;Romero, P.; Uversky, V. N.; Dunker, A. K. Exploring the bindingdiversity of intrinsically disordered proteins involved in one-to-manybinding. Protein Sci. 2013, 22 (3), 258−273.(28) Yang, X. J.; Seto, E. Lysine acetylation: codified crosstalk withother posttranslational modifications. Mol. Cell 2008, 31 (4), 449−461.(29) Marouco, D.; Garabadgiu, A. V.; Melino, G.; Barlev, N. A.Lysine-specific modifications of p53: a matter of life and death?Oncotarget 2013, 4 (10), 1556−1571.(30) Chen, W.; Daines, M. O.; Hershey, G. K. Methylation of STAT6modulates STAT6 phosphorylation, nuclear translocation, and DNA-binding activity. J. Immunol. 2004, 172 (11), 6744−6750.(31) Carr, S. M.; Munro, S.; Kessler, B.; Oppermann, U.; LaThangue, N. B. Interplay between lysine methylation and Cdkphosphorylation in growth control by the retinoblastoma protein.EMBO J. 2011, 30 (2), 317−327.(32) Nussinov, R.; Tsai, C. J.; Xin, F.; Radivojac, P. Allosteric post-translational modification codes. Trends Biochem. Sci. 2012, 37 (10),447−455.(33) Chang, X.; Xu, T.; Li, Y.; Wang, K. Dynamic modulararchitecture of protein-protein interaction networks beyond thedichotomy of ‘date’ and ‘party’ hubs. Sci. Rep. 2013, 3, 1691.

(34) Erce, M. A.; Abeygunawardena, D.; Low, J. K.; Hart-Smith, G.;Wilkins, M. R. Interactions affected by arginine methylation in theyeast protein-protein interaction network. Mol. Cell. Proteomics 2013,12 (11), 3184−3198.(35) Yun, C. Y.; Fu, X. D. Conserved SR protein kinase functions innuclear import and its action is counteracted by arginine methylationin Saccharomyces cerevisiae. J. Cell Biol. 2000, 150 (4), 707−718.(36) Hart-Smith, G.; Low, J. K.; Erce, M. A.; Wilkins, M. R.Enhanced methylarginine characterization by post-translationalmodification-specific targeted data acquisition and electron-transferdissociation mass spectrometry. J. Am. Soc. Mass Spectrom. 2012, 23(8), 1376−1389.(37) Albuquerque, C. P.; Smolka, M. B.; Payne, S. H.; Bafna, V.; Eng,J.; Zhou, H. A multidimensional chromatography technology for in-depth phosphoproteome analysis. Mol. Cell. Proteomics 2008, 7 (7),1389−1396.(38) Shannon, P.; Markiel, A.; Ozier, O.; Baliga, N. S.; Wang, J. T.;Ramage, D.; Amin, N.; Schwikowski, B.; Ideker, T. Cytoscape: asoftware environment for integrated models of biomolecularinteraction networks. Genome Res. 2003, 13 (11), 2498−504.(39) (a) Sakabe, K.; Wang, Z.; Hart, G. W. Beta-N-acetylglucosamine(O-GlcNAc) is part of the histone code. Proc. Natl. Acad. Sci. U.S.A.2010, 107 (46), 19915−19920. (b) Hottiger, M. O. ADP-ribosylationof histones by ARTD1: an additional module of the histone code?FEBS Lett. 2011, 585 (11), 1595−1599.(40) (a) Housley, M. P.; Rodgers, J. T.; Udeshi, N. D.; Kelly, T. J.;Shabanowitz, J.; Hunt, D. F.; Puigserver, P.; Hart, G. W. O-GlcNAcregulates FoxO activation in response to glucose. J. Biol. Chem. 2008,283 (24), 16283−16292. (b) Sakamaki, J.; Daitoku, H.; Yoshimochi,K.; Miwa, M.; Fukamizu, A. Regulation of FOXO1-mediatedtranscription and cell proliferation by PARP-1. Biochem. Biophys. Res.Commun. 2009, 382 (3), 497−502. (c) Yamagata, K.; Daitoku, H.;Takahashi, Y.; Namiki, K.; Hisatake, K.; Kako, K.; Mukai, H.; Kasuya,Y.; Fukamizu, A. Arginine methylation of FOXO transcription factorsinhibits their phosphorylation by Akt. Mol. Cell 2008, 32 (2), 221−231.(41) (a) Donlin, L. T.; Andresen, C.; Just, S.; Rudensky, E.; Pappas,C. T.; Kruger, M.; Jacobs, E. Y.; Unger, A.; Zieseniss, A.; Dobenecker,M. W.; Voelkel, T.; Chait, B. T.; Gregorio, C. C.; Rottbauer, W.;Tarakhovsky, A.; Linke, W. A. Smyd2 controls cytoplasmic lysinemethylation of Hsp90 and myofilament organization. Genes Dev. 2012,26 (2), 114−119. (b) Overath, T.; Kuckelkorn, U.; Henklein, P.;Strehl, B.; Bonar, D.; Kloss, A.; Siele, D.; Kloetzel, P. M.; Janek, K.Mapping of O-GlcNAc sites of 20 S proteasome subunits and Hsp90by a novel biotin-cystamine tag. Mol. Cell. Proteomics 2012, 11 (8),467−477. (c) Kundrat, L.; Regan, L. Identification of residues onHsp70 and Hsp90 ubiquitinated by the cochaperone CHIP. J. Mol.Biol. 2010, 395 (3), 587−594.(42) Wells, L.; Slawson, C.; Hart, G. W. The E2F-1 associatedretinoblastoma-susceptibility gene product is modified by O-GlcNAc.Amino Acids 2011, 40 (3), 877−883.(43) (a) Schroder, S.; Herker, E.; Itzen, F.; He, D.; Thomas, S.;Gilchrist, D. A.; Kaehlcke, K.; Cho, S.; Pollard, K. S.; Capra, J. A.;Schnolzer, M.; Cole, P. A.; Geyer, M.; Bruneau, B. G.; Adelman, K.;Ott, M. Acetylation of RNA polymerase II regulates growth-factor-induced gene transcription in mammalian cells. Mol. Cell 2013, 52 (3),314−324. (b) Ranuncolo, S. M.; Ghosh, S.; Hanover, J. A.; Hart, G.W.; Lewis, B. A. Evidence of the involvement of O-GlcNAc-modifiedhuman RNA polymerase II CTD in transcription in vitro and in vivo. J.Biol. Chem. 2012, 287 (28), 23549−23561.



Dow

nloa

ded

by N

AT

L L

BR

Y O

F SE

RB

IA o

n Se

ptem

ber

11, 2

015

| http

://pu

bs.a

cs.o

rg

Pub

licat

ion

Dat

e (W

eb):

Nov

embe

r 12

, 201

4 | d

oi: 1

0.10

21/p

r500

585p

a web of possibilities-network-based discovery of protein

Documents

new interaction codes

interaction partners

interaction presenceabsence

protein interaction

complex interaction

minimal interaction

unique proteins

date hub proteins