DOI:10.1002/cbic.201900611 Communications

AClick--Based Enrichable Crosslinker for Structural and Protein Interaction Analysis by Mass Spectrometry Michael Stadlmeier+,[a, b] Leander SimonRuntsch+,[a] Filipp Streshnev,[a] MartinWühr,[b] and Thomas Carell*[a]

Mass spectrometry is the method of choice for the characteri- and demethylation.[3–5] This showcasesthe need to study the sation of proteomes. Most proteins operate in protein com- complex environment of agiven protein to analyse its function plexes, in which their close associationmodulates their func- and activity state. Protein crosslinking, in combination with tion. However,with standard MS analysis, information on pro- analysisbymass spectrometry (XL-MS), is ideally suited as a tein–protein interactions is lost and no structural information is methodtogain information about protein structure and the retained. To gain structuraland interactome data, new cross- composition of protein complexes.[6–9] For XL-MS, specialised linking reagents are neededthat freeze inter- and intramolecu- chemicalreagents, crosslinkers, are required, which are able to lar interactions. Herein, the development of anew reagent, covalently connect protein residues that are in close proximity, which has severalfeaturesthat enable highly sensitive cross- for example, interacting in acomplex.[10,11] Characterisation of linking MS, is reported.The reagent enables enrichmentof the crosslinked peptides by meansofmass spectrometry,how- crosslinked peptides from the majority of background peptides ever,poses aformidable challenge for two reasons. Firstly, to facilitate efficient detection of low-abundant crosslinked crosslink identification has to be achieved by analysing frag- peptides. Due to the special cleavable properties, the reagent ment ions of not only one, buttwo, connected peptides, can be used for MS2 and potentially for MS3 experiments. Thus, thereby massively complicating MS2 spectra. Secondly,the the new crosslinking reagent, in combination with high-end crosslinked peptidespecies have only avery low abundance MS, should enable sensitive analysis of interactomes, which and are part of apeptide mixture that is overwhelmingly do- will help researchers to obtainimportant insights into cellular minatedbynon-crosslinked peptides. In many cases, this leads states in health and diseases. to dramatic information loss that hampers accurate crosslink identification, and therefore, interactome analysis. Afew cross- linking reagents were developed, which introduced MS-cleav- Proteinsneed to interactwith other proteins to form function- able groups, and the possibility for XL enrichmenttotackle al complexes. In many cases, protein function inside cells these problems.[12–18] cannotbeunderstood without information about the protein Herein, we report the development of anew crosslinking re- structure and knowledge of in whichcomplex the protein is agent that is enrichable and has cleavage properties which situated.[1,2] Thisis, for example, of paramount importance for allow accuratecrosslink identification. The designed cliXlink (1) proteins that modulate epigenetic informationonDNA. Almost is depicted in Figure 1. The cell-permeable reagent (Figure 1A all of these chromatin-modifying proteins require intensivein- and Figure S1 in the Supporting Information) features two suc- teraction with metabolic enzymes,which provide the cofactors cinimidyl units that can react with nucleophilesinpro- neededfor histone acetylation, deacetylation or methylation teins and are spacedroughly 9 Š apart;thus enabling the fixa- tion of short distances to gain structuralinformation on pro- + + [a] Dr.M.Stadlmeier, L. S. Runtsch, F. Streshnev,Prof. Dr.T.Carell teins and to capture proteins in close proximity to each other. Department of Chemistry,Ludwig-Maximilians-UniversitätMünchen Butenandtstrasse5–13,81377Munich(Germany) Efficient MS cleavability is ensured by the well-established sulf- E-mail:[email protected] oxide group, which can be cleaved under low-energy collision- [b] Dr.M.Stadlmeier,+ Prof. Dr.M.Wühr induceddissociation (CID) conditions prior to peptide fragmen- Department of Molecular Biology and tation. In addition, an unit allows an enrichment moiety, the Lewis-Sigler Institute for Integrative Genomics, Princeton University for example, biotin, to be attached to the crosslinked sites Washington Road,Princeton, NJ 08544 (USA) with the helpofCuAACreaction.[19] [+] These authors contributed equally to this work. Application of the crosslinker can follow the workflowde- Supporting information and the ORCID identification numbers for the authors of this article can be found under https://doi.org/10.1002/ picted in Figure 1B.Itinvolves the addition of reagent 1 to a cbic.201900611. complex proteome, fixing the native state distance information  2019 The Authors. Published by Wiley-VCH Verlag GmbH&Co. KGaA. of peptides in close proximity and preserving them throughout This is an open access article under the terms of the Creative Commons At- the further workflowfor MS analysis. Attachment of the affinity tributionNon-Commercial License,whichpermitsuse, distributionand re- group for enrichmentisperformed after crosslinking to avoid production in any medium, provided the original work is properly cited and is not used for commercial purposes. interference from bulky enrichmentgroups. By modification of This article is part of aSpecial Issue commemorating the 20th Anniversary the crosslinked peptides at the protein level, excess small-mol- of ChemBioChem. To view the complete issue, visit our homepage ecule reagents can readily be removed by acetone precipita-

ChemBioChem 2020, 21,103 –107 103  2019 The Authors. PublishedbyWiley-VCH Verlag GmbH &Co. KGaA, Weinheim Communications

to aclean separation of the peptides (a, b)and generates two fragments foreach peptide:analkene and asulfenic acid frag- ment;the latter formsathial upon water loss. This provides two mass pairs with acharacteristic Dm/z 32. Importantly,we  observed that fragmentation pathway a(Figure 1C)predomi- nated.Therefore, for the a-peptide the fragment, and for the b-peptide the thial fragment, are the main cleavage products.Due to the asymmetric cleavage properties, mostof the signal intensity is retained on one fragment per peptide, which should provide excellent sensitivity in MS3 experiments. The synthesis of reagent 1 is straightforward(Scheme1). The startingpoint is the oxidised dimer of homocys- tein 2,which is treatedwith 4-pentynoic acid to give disulfide

Scheme1.Synthesis of cliXlink (1). a) 4-Pentynoicacid, N-(3-dimethylamino- propyl)-N’-ethylcarbodiimide hydrochloride (EDC·HCl), 1-hydroxybenzotri-

azole hydrate (HOBt·H2O), NEt3,DMF,RT, 15 h, 71%; b) over two steps:

1) tris(2-carboxyethyl)phosphine hydrochloride (TCEP·HCl), NaHCO3, DMF,

H2O, RT,3h; 2) methyl 3-bromopropionate, 45 8C, 44 h, 55%; c) LiOH,THF,

H2O, 08CRT, overnight (o/n), 91%; d) N-hydroxysuccinimide (NHS), , Figure 1. A) Structureof1.Su: succinimidyl. B) Workflowfor XL-MS experi- trifluoroacetic anhydride,MeCN, 08CRT, o/n, 78%; e) meta-chloroperoxyben- ments. After the addition of 1,protein sites in close proximity are covalently zoic acid (mCPBA), AcOEt,RT, 30 min,57%. linked.The generated crosslinks can be functionalised with acopper-cata- lysed Huisgenreaction (CuAAC). Excess reagents are removed by acetone precipitations. After enzymaticdigestionand enrichment on magnetic strep- tavidin beads,crosslinkedpeptides were cleavedoff under mild conditions 3.Reductivecleavage of the disulfide and alkylation of the and subsequently analysed by means of LC-MS2.C)Fragmentation pathways generated with methyl 3-bromopropionate generates of the crosslinked peptides,forming two peptidepairs with acharacteristic compound 4.Saponification of both methyl to 5 Dm of 31.9721 u. rep and conversionof5 into activated bissuccinimidylester 6,fol- lowed by oxidation of the thioether to the , gives re- tion. This is followed by enzymatic digestionofthe proteins. agent 1 in atotal yield of 16 %. Of particularimportance is that The crosslinkedpeptides are in this way labelled, for example, the reagent is very pure and the reactive ester units are in with biotin, whichallows for their enrichment. Afterwards, they place at both sides. Partial hydrolysis needs to be avoided. This are analysed by means of mass spectrometry.Because the indi- was ensured by afinal precipitation purification.Reagent 1 vidual peptidemasses in acrosslink are not immediately acces- was dissolved in amixture of ethyl acetate and dichlorometh- sible, crosslink assignment is difficult, especially in complex ane and precipitated upon addition of hexanes. samples. This requires the use of MS-cleavable reagents,which We next investigated the MS properties of 1.Tothis end, we facilitateMS2 identification by forming specific fragment added 1 to the commonly used modelprotein,bovine serum ions.[20,21] At best, the reagent should also enable MS3 experi- albumin(BSA). We followed the workflow depicted in Fig- ments,which requires that the crosslinker is cleaved before ure 1B withoutperforming the enrichmentstep. In short,after the peptides. This allows peptides to be separated for individu- the addition of 1 to the protein solution and reaction for 1h al MS3-based identification. Our reagent 1 contains b-hydrogen at room temperature and physiological pH, we precipitated atoms on both sides of the sulfoxide, so that fragmentation the protein with acetone, resuspended it in buffer (see the occurs in two directions, as depicted in Figure 1C.This leads

ChemBioChem 2020, 21,103 –107 www.chembiochem.org 104  2019 The Authors. PublishedbyWiley-VCH Verlag GmbH &Co. KGaA, Weinheim Communications

Supporting Information) and digested the protein subsequent- ly with amixture of trypsin and Lys-C. The obtained peptide mixturewas desalted with C18 Tip columns and analysed by meansofHPLC-MS2 (Figure 2). Fig- ure 2A shows two crosslinkedBSA peptides (a+b)asanex- ample.After determining the exact mass of the intact crosslink (m/z 677.1;Figure 2B), we performed both CID and HCD frag- mentation on the precursor to evaluate the fragmentation properties (Figure 2C). More selective CID fragmentation (reso- nance excitation of the crosslink ion;Figure 2C,top spectrum) at alow normalised collision energy of 25%provides, as expected, only asmall set of intense signals. Twoprominent signal pairs, with Dm/z of 32 (Dmrep), are obtained due to crosslinker fragmentation, which results in athial(a-/b-thial) and alkene(a-/b-alkene) fragment for each peptide. As mentioned before, fragmentation pathway a(Figure 1C)ispre- ferred, leading to an asymmetric intensity distribution. The dominant formation of the expected intact, separated peptides consequently makesthe reagent amenable for more sophisti- cated MS3 experiments. For our study,weused HCD fragmentation. This method provides simultaneouscrosslinker cleavage and formation of the peptide fragments required for identification (Figure 2C, bottom spectrum). To help with crosslink/peptide recognition, it is important that HCD fragmentation still provides the alkene and thial fragments. The HCD data, therefore, allow the iden- tification of peptidesbyMS2. Using this method, we analysedthe data with the freely availableMeroX software, strictly filtering all crosslink identifi- cations (score >50, false discovery rate (FDR) 1%)and requir- ing the presence of at least three out of four ions from the two characteristic mass pairs.[23, 24] Without exploitingthe possi- bility for CuAAC-based enrichment, we wereable to identify 61 unique BSA crosslinks in asingle measurement (Fig- ure 2D).[25] This result is representative for measurements con- ducted with purified BSA in our laboratory.Additionally,we depicted the crosslinks found in the BSA crystal stucture and noted that the majority of measured distances were reasona- ble. Importantly,the crosslinks show amean C C distance aÀ a between the crosslinkedamino acids of 22.1 Š and adistance distribution that is in perfect agreement of what is expected for crosslinkingwith areagent that spaces the active esters by about 9 Š (Figures 2E and S2)[26] and taking into account side- chain lengths and protein molecular dynamics.[27] Crosslinking occurred mostly between two Lysresidues (49%), but we also Figure 2. Analysis of crosslinked BSA prior to enrichment. A) Sequences of a detected Lys–Thr (30%), Lys–Ser (13%) and Lys–Tyr (8%) con- representative crosslink within BSA:peptide a:VHKECCHGDLLECADDR(IAA- modified on Cys) and b:ALKAWSVAR. B) The isotopeenvelopeofthe 5+ nections, in agreement with the reactivity of the succinimidyl precursor is shown. C) Fragmentation of the crosslinkedpeptides under CID [28] esters (Figure 2F). (25%, top spectrum) and steppedhigher energy collisional dissociation This success allowed us to validate the new crosslinker in a (HCD)conditions (25/30/32%,bottom spectrum) showing the product ions more complex environment. We particularly wanted to show of sulfoxide cleavage.The different fragmentation pathways are shownin orangeand purple (Figure1). D) MS2 identification of BSA crosslinkingsites the additional value of the CuAAC-based enrichmentpossibili- prior to enrichment.The xVis representationshows the positionsofcross- ty.For this experiment, we again crosslinkedthe BSAprotein linked aminoacids as red lines (left). The corresponding C C distances are aÀ a (10 mg), precipitated it with acetone, redissolved the cross- depicted in the BSA crystalstructure(PDBID: 4F5S[22]). E) Histogramofthe measured Euclidean C C distances in the crystalstructure;the majority linked protein and subsequently performed the click reaction aÀ a are below 40 Š,with ameanmeasureddistanceof22.1 Š.F)Distributionof (Figure S3A) by adding 7 (Figure 3A)and CuSO4/tris(3-hydroxy- the identified crosslinking sites between Lys–Lys, Lys–Thr,Lys–Ser and Lys– propyltriazolylmethyl) (THPTA) followedbyreduction of Tyr. CuII to CuI upon addition of sodium ascorbate. After 1h at

ChemBioChem 2020, 21,103 –107 www.chembiochem.org 105  2019 The Authors. PublishedbyWiley-VCH Verlag GmbH &Co. KGaA, Weinheim Communications

In summary,our new crosslinker 1 combines the following advantages:firstly,the size of the reagent is ideally suited to gain valuable distance information for structural proteomics. In addition, the molecule is cell permeable and the alkyne unit does not cause major interference with the crosslinking reac- tion. Furthermore, the robust, efficient sulfoxide fragmentation simplifies crosslink identification. The possibility of functionalis- ing peptides crosslinked with 1 by click chemistryallows di- verse modifications and enrichment strategies that enable analysisoflow-abundant crosslinks to be employed. To con- clude, our reagent 1 now paves the way for complex interac- tome analysisbyusing MS2 and MS3 experiments.

Acknowledgements

We thank the Deutsche Forschungsgemeinschaft for financial Figure 3. Enrichment of crosslinked BSA from acomplex sample. A) Struc- supportthrough SFB1309 (TP-A4), SFB1032 (TP-A5) and SPP1784. ture of 7,with the biotin group (pink), the disulfide bond(green)and the Further supportwas obtainedfrom the EinzelverfahrenCA275- azide moiety (blue). Upon reduction of the disulfide, the biotin moiety is re- 11/1 and from the European Union through the Marie Curie Inter- moved from the crosslink. B) Depictionofthe spike-in experiment. 10 mgof crosslinkedand CuAAC-modified BSA digest were added to 435 mgofHEK national Training and Mobility Network “LightDynamics” (grant digest as acomplex background. Magnetic streptavidin beads were used to no. 765866). Research was furthermore supported by aEuropean enrich the crosslinkedpeptides. C) Resultsofthe HPLC-MS2 analysis, showing Research Council (ERC) advanced grant under the European the effect of enrichment. The experiment was performed in technicaltripli- Union’s Horizon 2020 research and innovation programme cates;numbers showthe mean valueand the standard deviation is indicat- ed by the error bar. (grantagreement no. EPiR 741912) and theVolkswagenfounda- tion (Initiative “Life”: EcoRib). M.W.thanks the U.S. Department of Energy(DOE grant DE-SC0018260) and the National Institutes of room temperature,weperformed another acetone precipita- Health(NIH grant R35GM128813) for support. M.S. and L.S.R. tion to remove excess biotin azide. We then added trypsin and thank the Fonds der Chemischen Industrie for predoctoral fellow- Lys-C for digestion. ships. The mass spectrometry proteomics data have been depos- [29] These peptides we next combinedwith aprotein digest ob- ited with the ProteomeXchange Consortium through the PRIDE tained from aHEK cell extract (435 mgofprotein;Figure 3B). partner repository with the accession no. PXD015080. This creates amassive background of non-crosslinked pep- tides. If we now analysed the crosslinks within this large back- Conflict of Interest ground (Figure 3C), we identified only avery small number of crosslink spectral matches (mean CSMs=5.0), uniquecross- The authors declare no conflict of interest. linked sites (meanunique XLs=3.7) as well as monolinks (mean= 5.3), in which one succinimidylester hydrolysed before reactionwith the protein. However,ifweperformed Keywords: click chemistry · crosslinking mass spectrometry · the enrichmentwith streptavidin-coated magnetic beads, the interactome analysis · proteomics · structure elucidation number of crosslink identifications improved dramatically.Fol- lowing incubation of the peptidemixture with the magnetic [1] I. N. Berezovsky,E.Guarnera, Z. Zheng,B.Eisenhaber,F.Eisenhaber, particles;extensive washing (6”)ofthe beads with buffer (see Curr.Opin. Struct. Biol. 2017, 42,67–74. [2] H. Currinn, T. Wassmer, Biochem. Soc. Trans. 2016, 44,185–190. the SupportingInformation) and mild, reductive cleavage of [3] S. Allard, R. T. Utley,J.Savard, A. Clarke, P. Grant, C. J. Brandl, L. Pillus, the disulfide to liberate the “captured” crosslinks and thereby J. L. Workman,J.CôtØ, EMBO J. 1999, 18,5108 –5119. removing the biotin moiety (Figure S3B), we now,onaverage, [4] Y. Xue, J. Wong, G. T. Moreno,M.K.Young, J. CôtØ,W.Wang, Mol. Cell detected 95.0 CSMs and 37.0 unique XLs together with 184.3 1998, 2,851 –861. [5] M. Sauvageau, G. Sauvageau, PLoS Biol. 2008, 6,e113. monolinks. The number of identified uniqueXLs was lower [6] A. Leitner,T.Walzthoeni, A. Kahraman, F. Herzog, O. Rinner,M.Beck, R. than that for the purifiedBSA (Figure 2D), which could be ex- Aebersold, Mol. Cell. Proteomics 2010, 9,1634 –1649. plained by inefficiences in the CuAACreaction or interference [7] A. Sinz, Angew.Chem.Int. Ed. 2018, 57,6390 –6396; Angew.Chem. of the non-crosslinked complex background.However,this 2018, 130,6498 –6504. [8] J. Rappsilber, J. Struct. Biol. 2011, 173,530–540. result shows that our new crosslinker, 1,isabletoenrich cross- [9] J. D. Chavez, J. E. Bruce, Curr.Opin.Chem.Biol. 2019, 48,8–18. linked peptides in avast excessofunmodified peptides;thus [10] F. J. O’Reilly,J.Rappsilber, Nat. Struct. Mol. Biol. 2018, 25,1000–1008. facilitating the analysisofcomplex samples with low-abundant [11] A. Leitner,M.Faini, F. Stengel, R. Aebersold, Trends Biochem. Sci. 2016, crosslinks. Originally,weweresceptical aboutthe asymmetric 41,20–32. [12] A. Kao, C.-l. Chiu, D. Vellucci, Y. Yang, V. R. Patel, S. Guan, A. Randall, P. structureof1.The data, however,show that, despite this, a Baldi, S. D. Rychnovsky, L. Huang, Mol. Cell.Proteomics 2011, 10, large number of crosslinks are identified. M110.002212.

ChemBioChem 2020, 21,103 –107 www.chembiochem.org 106  2019 The Authors. PublishedbyWiley-VCH Verlag GmbH &Co. KGaA, Weinheim Communications

[13] M. Q. Müller, F. Dreiocker,C.H.Ihling, M. Schäfer,A.Sinz, Anal. Chem. [24] M. Gçtze,C.Iacobucci, C. H. Ihling,A.Sinz, Anal. Chem. 2019, 91, 2010, 82,6958 –6968. 10236 –10244. [14] A. M. Burke, W. Kandur,E.J.Novitsky,R.M.Kaake,C.Yu, A. Kao, D. Vel- [25] M. Grimm, T. Zimniak, A. Kahraman, F. Herzog, NucleicAcids Res. 2015, lucci, L. Huang, S. D. Rychnovsky, Org. Biomol. Chem. 2015, 13,5030 – 43,W362 –W369. 5037. [26] E. D. Merkley,S.Rysavy,A.Kahraman, R. P. Hafen, V. Daggett, J. N. [15] X. Tang, J. E. Bruce, Mol. Biosyst. 2010, 6,939 –947. Adkins, Protein Sci. 2014, 23,747 –759. [16] R. Huang, W. Zhu, Y. Wu, J. Chen,J.Yu, B. Jiang,H.Chen,W.Chen, [27] Z. Orbµn-NØmeth, R. Beveridge, D. M. Hollenstein, E. Rampler,T.Stranzl, Chem. Sci. 2019, 10,6443 –6447. O. Hudecz, J. Doblmann, P. Schlçgelhofer,K.Mechtler, Nat. Protoc. 2018, [17] M. Rey,M.DuprØ,I.Lopez-Neira, M. Duchateau, J. Chamot-Rooke, Anal. 13,478–494. Chem. 2018, 90,10707 –10714. [28] S. Kalkhof, A. Sinz, Anal. Bioanal. Chem. 2008, 392,305 –312. [18] B. Steigenberger, R. J. Pieters, A. J. R. Heck,R.A.Scheltema, ACS Cent. [29] Y. Perez-Riverol, A. Csordas, J. Bai, M. Bernal-Llinares, S. Hewapathirana, Sci. 2019, 5,1514–1522. D. J. Kundu, A. Inuganti,J.Griss, G. Mayer, M. Eisenacher, E. PØrez, J. [19] M. Ehrlich, M. J. Gattner,B.Viverge, J. Bretzler,D.Eisen, M. Stadlmeier, Uszkoreit, J. Pfeuffer,T.Sachsenberg, S¸ .Yılmaz,S.Tiwary,J.Cox, E. M. Vrabel, T. Carell, Chem. Eur.J.2015, 21,7701 –7704. Audain, M. Walzer,A.F.Jarnuczak, T. Ternent, A. Brazma, J. A. Vizcaíno, [20] F. Liu, P. Lçssl,R.Scheltema, R. Viner,A.J.R.Heck, Nat. Commun. 2017, Nucleic Acids Res. 2019, 47,D442–D450. 8,15473. [21] O. Klykov,B.Steigenberger,S.Pektas¸, D. Fasci,A.J.R.Heck,R.A.Schel- tema, Nat. Protoc. 2018, 13,2964–2990. Manuscript received:October8,2019 [22] A. Bujacz, Acta Crystallogr.Sect. DBiol. Crystallogr. 2012, 68,1278 –1289. [23] M. Gçtze, J. Pettelkau,R.Fritzsche, C. H. Ihling, M. Schäfer, A. Sinz, J. Accepted manuscript online:October 8, 2019 Am. Soc. Mass. Spectrom. 2015, 26,83–97. Version of record online:November 28, 2019

ChemBioChem 2020, 21,103 –107 www.chembiochem.org 107  2019 The Authors. PublishedbyWiley-VCH Verlag GmbH &Co. KGaA, Weinheim