(PDF) A family of unconventional deubiquitinases with modular chain specificity determinants ARTICLEA family of unconventional deubiquitinases withmodular chain specificity determinantsThomas Hermanns 1, Christian Pichlo 2, Ilka Woiwode 1, Karsten Klopffleisch 1, Katharina F. Witting3,Huib Ovaa3, Ulrich Baumann2 Kay Hofmann 1Deubiquitinating enzymes (DUBs) regulate ubiquitin signaling by trimming ubiquitin chains orremoving ubiquitin from modified substrates. Similar activities exist for ubiquitin-relatedmodifiers, although the enzymes involved are usually not related. Here, we report humanZUFSP (also known as ZUP1 and C6orf113) and fission yeast Mug105 as founding members ofa DUB family different from the six known DUB classes. The crystal structure of humanZUFSP in covalent complex with propargylated ubiquitin shows that the DUB family shares afold with UFM1- and Atg8-specific proteases, but uses a different active site more similar tocanonical DUB enzymes. ZUFSP family members differ widely in linkage specificity throughdifferential use of modular ubiquitin-binding domains (UBDs). While the minimalisticMug105 prefers K48 chains, ZUFSP uses multiple UBDs for its K63-specific endo-DUBactivity. K63 specificity, localization, and protein interaction network suggest a role for ZUFSPin DNA damage response.DOI: 10.1038/s41467-018-03148-5 OPEN1Institute for Genetics, University of Cologne, Zülpicher Str. 47a, 50674 Cologne, Germany. 2Institute of Biochemistry, University of Cologne, Zülpicher Str.47, 50674 Cologne, Germany. 3Department of Chemical Immunology, Leiden University Medical Center, Einthovenweg 20, 2333 ZC Leiden, TheNetherlands. Correspondence and requests for materials should be addressed to K.H. (email: kay.hofmann@uni-koeln.de)NATURE COMMUNICATIONS | (2018) 9:799 |DOI: 10.1038/s41467-018-03148-5 |www.nature.com/naturecommunications 11234567890():,;Content courtesy of Springer Nature, terms of use apply. Rights reserved The covalent attachment of ubiquitin to proteins via the ε-amino group of substrate lysine residues is—besidesphosphorylation—the most important posttranslationalmodification for regulating protein signaling and homeostasis1.The ability of ubiquitin to target other ubiquitin molecules atvarious lysine residues, giving rise to ubiquitin chains of differentlinkage types, contributes substantially to the versatility of theubiquitination system2. Deubiquitinating enzymes (DUBs) areisopeptidases that can deconjugate single ubiquitin units or entireubiquitin chains from proteins, thereby erasing or modulating theubiquitin signal. In addition, several DUBs are also able to cleavepeptide bonds at the C terminus of ubiquitin, an activity requiredfor processing the primary translation products of the ubiquitingenes and thus essential for the entire ubiquitination cascade. Thehuman genome encodes approximately 100 DUB enzymesbelonging to 6 different families, which exhibit distinct butoverlapping cleavage preferences3,4. Five of these families arecysteine proteases, some of which show weak but significantsequence similarity to each other5. The sixth family belongs to themetalloproteases and appears to be evolutionary more ancient,since it contains bacterial and archaeal members that are likely toact on prokaryotic ubiquitin-fold proteins6. Besides ubiquitin,there are several ubiquitin-like modifiers (UBLs), which alsorequire proteases for processing their immature precursors and/or for their deconjugation. Despite their mechanistic similarities,UBL proteases typically belong to families distinct from DUBs7:SUMO and NEDD8 are cleaved by members of the SENP/ULPfamily, UFM1 is cleaved by the UFSP family, while the autophagymodifiers ATG8 and ATG12 are cleaved by members of theautophagin (ATG4) family8,9. Besides their catalytic domains,many DUBs and UBL proteases harbor domains or motifs forrecognizing the modifier to be cleaved, or the substrate fromwhich the modifier is removed. This trend is particularly pro-nounced for DUBs, where the presence of multiple ubiquitin-binding domains (UBDs) can confer specificity for ubiquitinchains of a particular linkage type10.Here, we describe the biochemical and structural character-ization of a seventh deubiquitinase family, which is distantlyrelated to proteases for ubiquitin-like modifiers, but has a dif-ferent active site architecture and is truly specific for the cleavageof ubiquitin. We provide a detailed analysis of ZUFSP (Zn-fingerand UFSP domain protein), the singular human member of thisfamily, which contains multiple UBDs responsible for the specificaction on K63-linked chains. By contrast, Mug105—a K48-preferring ZUFSP homolog from the fission yeast Schizo-saccharomyces pombe—lacks all UBDs and consists only of thecore catalytic domain. A comparison of ZUFSP and Mug105offers unique insights into the mechanism of how the evolu-tionary loss (or gain) of non-catalytic ubiquitin-binding domainscan profoundly change the specificity of a deubiquitinase.ResultsZUFSP and Mug105 are related to UFM1/Atg8 proteases. Thecatalytic domain of the UFM1-specific protease (UFSP) familyhas been reported to be structurally related to that of the ATG4family, enzymes that process the autophagy modifiers Atg8 andAtg1211. By performing sensitive sequence analysis using thegeneralized profile method12, we found this structural similarityto be mirrored by a distant but highly significant sequence rela-tionship (p 0.001) between these families. Concurrently, a thirdprotein family containing the uncharacterized human proteinC6orf113/ZUFSP was found to exhibit significant sequencesimilarity (p 0.001) to both UFSP and ATG4 families of cysteineproteases. As shown in Fig. 1a, the active site Cys and Asp resi-dues of UFSP and ATG4 are conserved in members of the ZUFSPfamily, while the catalytic His is absent. Hypothesizing thatZUFSP might nevertheless be an active protease with arearranged active site, we analyzed ZUFSP family members froma wide range of species and identified two highly conservedhistidine residues, with His-491 being the best candidate forcompleting the active site of human ZUFSP (SupplementaryFig. 1). While the conserved catalytic domain is common to allZUFSP 322Mug105 1UFSP2 256ATG4A 34ZUFSP 381Mug105 64UFSP2 313ATG4A 119ZUFSP 467Mug105 147UFSP2 383ATG4A 187ZUFSP 549Mug105 217UFSP2 451ATG4A 257ZUFSP 1ZUFSP 2ZUFSP 3ZUFSP 4POLHZUFSPMINDY-2MINDY-1RNF168SPRTNRFC1RABEX5abcFig. 1 ZUFSP and Mug105 are related to UFM1/Atg8 proteases. aStructurally correct alignment of the catalytic domains of human ZUFSP (this work),mouse UFSP2 (3OQC) and human ATG4A (2P82). The S. pombe Mug105 sequence was added by sequence similarity to ZUFSP. Invariant andconservatively replaced residues are shown on black or gray background, respectively. Catalytic residues are highlighted in blue. bConservation of the fourUBZ-like zinc fingers of ZUFSP, in comparison to the structurally characterized UBZ finger (3WUP). cConservation of the MIU domain of ZUFSP, incomparison to other human MIU domainsARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-03148-52NATURE COMMUNICATIONS | (2018) 9:799 |DOI: 10.1038/s41467-018-03148-5 |www.nature.com/naturecommunicationsContent courtesy of Springer Nature, terms of use apply. Rights reserved 252217115846320 36036036036 036 036036hK6 K11 K29 K33 K48 K63 M1252217115846320K1130 120 0 30 120 0 30 120 0 30 120 MinK48 K63 M12522171103603603603 6 0 3603 603 6hK6 K11 K29 K33 K48 K63 M1Ub-PAK48 diUb-VME2522171132–abcedfZUFSPUb2UbZUFSPUb2UbUb4Ub3Mug105Ub2UbMug105K63 diUb-VMEg252217115846320 30 120601 2.5 5 10 15 MinZUFSP800500010,00015,00020,00025,00030,00035,00040,00045,00050,000(RFU)05001000150020002500300035004000450050000102030(RFU)500100015002000250030003500400045005000(RFU)0500100015002000250030003500400045005000(RFU)0500100015002000250030003500400045005000(RFU)(Min)0102030(Min)0102030(Min)0500100015002000250030003500400045005000(RFU)0500100015002000250030003500400045005000(RFU)0102030(Min)0102030(Min)0102030(Min)0102030(Min) RLRGGUbiquitinSUMO1 SUMO2NEDD8 ISG15Ub2UbUb3Ub4Ub5Ub6CoomassieCoomassieCoomassieCoomassieCoomassieLC3A100705535251510130ZUFSPZUFSP + Cy5-Ub-PAZUFSP C360A + Cy5-Ub-PAZUFSP + Rho-Ufm1-PAUfsp2Ufsp2 + Rho-Ufm1-PAUfsp2 + Cy5-Ub-PA**Fluorescent scanh25221711584632–Ub-PALC3B-PA–Ub-PALC3B-PAZUFSP ATG4BCoomassie**ZUFSPMug105ZUFSPMug105ZUFSPMug105ZUFSPMug105ZUFSPMug105ZUFSPMug105ZUFSPMug105Fig. 2 ZUFSP and Mug105 are ubiquitin-specific proteases with different chain specificity. aActivity assays with Ub-/UbL-AMC substrates shown asreleased fluorescence (RFU) over time (min) with ZUFSP or Mug105. Shown RFU values are the means of triplicates. bFluorescent scan of a suicide probereaction of ZUFSP or Ufsp2 with Cy5-Ub-PA (arrow) and Rho-Ufm1-PA (arrowhead). Asterisks (*) mark the shifted band after reaction. cSuicide probereaction of ZUFSP or ATG4B with Ub-PA and LC3B-PA. Asterisks (*) mark the shifted band after reaction. d,eLinkage specificity analysis with ZUFSP. Apanel of di-ubiquitin (d) or tetra-ubiquitin (e) chains was treated with ZUFSP for the indicated time points. fTime course of cleavage of K63-linked Ub6+chains by full-length ZUFSP. gLinkage specificity analysis with Mug105. A panel of Ub2was treated with Mug105 for the indicated time points. hSuicideprobe reaction with Mug105 and K48-diUb-VME, K63-diUb-VME or Ub-PA. Arrows mark the shifted bands after reactionNATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-03148-5 ARTICLENATURE COMMUNICATIONS | (2018) 9:799 |DOI: 10.1038/s41467-018-03148-5 |www.nature.com/naturecommunications 3Content courtesy of Springer Nature, terms of use apply. Rights reserved members of the ZUFSP family, there are substantial differences inthe non-catalytic regions. Some members, including Mug105from the fission yeast S. pombe, consist solely of the catalyticdomain, while most members contain one or more sequenceregions with significant similarity to known ubiquitin-bindingdomains. Human ZUFSP contains four predicted C2H2 zincfingers related to the ubiquitin-binding UBZ-class13, followed bya predicted MIU (motif interacting with ubiquitin) region14(Fig. 1b, c). Thus, bioinformatical sequence analysis suggests thatZUFSP could be an active protease that is distantly related tovarious UBL-specific proteases, but has the potential to bind—and maybe cleave—polyubiquitin chains.ZUFSP and Mug105 are DUBs. Since ZUFSP is related to UBLproteases and contains putative ubiquitin-binding domains, wetested bacterially expressed ZUFSP proteins for catalytic activityagainst ubiquitin and ubiquitin-related modifiers. Both humanZUFSP and the S. pombe homolog Mug105 were able to liberatethe fluorophore AMC (7-amino-4-methylcoumarin) from aubiquitin-AMC fusion, indicating a cleavage after the C-terminalGly-76 of ubiquitin (Fig. 2a). The specific activities of ZUFSP andMug105 against this substrate were 2.3 and 4.1nmol substrate permg enzyme per second, respectively (Supplementary Fig. 2a). Bycontrast, analogous AMC fusions of the ubiquitin-like modifiersSUMO1, SUMO2, NEDD8, ISG15 or LC3A were not processed(Fig. 2a). Surprisingly, both ZUFSP and Mug105 were highly activeagainst RLRGG-AMC, a fusion of AMC, to a pentapeptide derivedfrom the ubiquitin C terminus (Fig. 2a). This reaction follows theMichaelis–Menten kinetics with KM=50.4 μM, kcat=4.9 s−1forZUFSP, and KM=12.2 μM, kcat=7.2 s−1for Mug105 (Supple-mentary Fig. 2b). The RLRGG-AMC peptide is not cleaved bytypical DUBs, which require an intact ubiquitin moiety for activity,as shown here for USP21 (Supplementary Fig. 2c). While bacte-rially expressed ZUFSP did not react with C-terminally pro-pargylated UFM1 (UFM1-PA, Fig. 2b) or the propargylated Atg8homolog LC3B-PA (Fig. 2c), it readily reacted with propargylatedubiquitin (Ub-PA, Fig. 2b, c). Ub-PA is a covalent inhibitor that ishighly selective for thiol DUBs15; the analogous inhibitor UFM1-PA reacts only with UFM1 proteases such as UFSP2, while LC3B-PA reacts with Atg8 proteases such as ATG4B (Fig. 2b, c).When tested against a panel of di-ubiquitin species of differentlinkage types, ZUFSP showed a moderate activity towards K63-linked di-ubiquitin, with some minimal activity towards K11- andK48-linked species (Fig. 2d). When using tetra-ubiquitin sub-strates, ZUFSP showed K63-specific cleavage with a markedlyincreased activity: while K63-Ub4was completely hydrolyzed after30 min (Fig. 2e), the hydrolysis of K63-Ub2was not completedafter 6 h under comparable reaction conditions (Fig. 2d). Thispreference for Ub4over Ub2suggests an ‘endo-cleavage’mode forZUFSP, which became more evident when using longer K63-chainsubstrates: while the upper bands disappear within the firstminutes, Ub2is remarkably inert and mono-ubiquitin appears latein the time course (Fig. 2f). The specificity for K63 cleavage ismaintained in ZUFSP truncations: gradual shortening of the N-terminal region leads to decreased activity, but no change in thelinkage preference (Supplementary Fig. 3a-c).Unlike the K63-specific ZUFSP, the compact homolog Mug105from S. pombe cleaved K48-linked di-ubiquitin better than otherlinkage types, with some residual activity against K63 andK11 species (Fig. 2g). A concordant observation was made whentesting Mug105 with covalent DUB-specific inhibitors containinga reactive group between two ubiquitin units16. Mug105 reactedwith the mono-ubiquitin-targeted probe Ub-PA and with theK48-targeted probe K48-diUb-VME, but not with the corre-sponding K63-targeted probe K63-diUb-VME (Fig. 2h). Thus,both ZUFSP and Mug105 are linkage-specific DUBs, albeit withdifferent specificities. The multitude of predicted ubiquitin-binding domains in ZUFSP, all absent from Mug105, mightaccount for the strikingly different chain preference.ZUFSP and ATG4/UFSP structures differ in the active site. Forgetting further insights into the unusual active site geometry ofZUFSP and the basis for its K63 specificity, we solved the crystalstructure of a covalent complex between human ZUFSP (residues232–578) and Ub-PA to a resolution of 1.7 Å (Table 1). Theasymmetric unit contains one ZUFSP ubiquitin conjugate, whichwas almost completely resolved in the electron density. Only ashort flexible loop (AA: 468–473) and eight amino acids at thevery N terminus were not defined well enough in the electrondensity to permit reliable modeling. The ZUFSP fragment usedfor coupling to Ub-PA and subsequent crystallization startsbefore the predicted MIU domain and encompasses the con-served C-terminal region of human ZUFSP, including the cata-lytic core domain and adjacent elements predicted to bestructured. The reaction product of thiol DUBs with Ub-PAresembles an intermediate stage of the protease reaction, in whichthe distal ubiquitin (poised for removal) occupies the S1 site ofthe enzyme. The ZUFSP structure forms a globular α/β-foldedcore with two prominent helical protrusions (Fig. 3a). The coreTable 1 Data collection and refinement statisticsZUFSP S-SAD ZUFSP nativeData collectionSpace group P6522 P6522Unit cell constantsa,b,c(Å) 84.2, 84.2, 201.8 84.2, 84.2, 201.8α;β;γ(°) 90, 90, 120 90, 90, 120Wavelength (Å) 2.07 0.98Resolution (Å) 59.13–2.3 (2.38–2.3) 49.45–1.73 (1.79–1.73)No. of observations 4,098,044 (364,012) 484,411 (48,504)No. of unique reflections 35,609 (3576)a44,866 (4337)Multiplicity 115.1 (101.8) 10.8 (11.2)Completeness (%) 100 (100) 100 (99)Rmerge(%) 11.7 (39.5) 8.4 (82.4)Rmeas(%) 11.8 (39.7) 14.3 (86.4) I/σ(I) 47.7 (13.9) 16.6 (2.5)CC1/2(%) 100 (99.7) 99.8 (79.5)Anomalous signalCCano (59.13–3Å[3.08–3.0 Å]) (%)56 (31)Rpim(59.13–3Å[3.08–3.0 Å]) (%)0.8 (1.4)RefinementReflections used inrefinement44,862Number of testreflections1930Rwork/Rfree(%) 17.8/20.7Root-mean-square deviationsBond lengths (Å) 0.005Bond angles (°) 0.6Average Bfactor (Å2)All macromoleculeatoms37.0Solvent molecules 39.5Other atoms 24.3Ramachandran plot (%)Most favored 97.7Additionally allowed 2.2Disallowed 0.0aFriedel pairs were kept separateARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-03148-54NATURE COMMUNICATIONS | (2018) 9:799 |DOI: 10.1038/s41467-018-03148-5 |www.nature.com/naturecommunicationsContent courtesy of Springer Nature, terms of use apply. Rights reserved catalytic domain is clearly related to that of the UFM1-proteaseUFSP2 (PDB: 3OQC)11; the two structures can be superimposedwith an RMS (root mean square deviation) distance of 3.65 Åover 200 residues (Fig. 3b and Supplementary Fig. 4a). A long 29-residue helix (α1) points away from the active site and providesthe S1 binding surface for the outgoing distal ubiquitin—hereoccupied by the covalently coupled Ub-PA. The α2 and α3 helicesform a hairpin-like structure protruding from the catalytic core inthe opposite direction. This region has little contact with theremainder of the structure and might form the S1’site for bindingthe proximal (substrate) ubiquitin. Of particular interest is theorganization of the active site (Fig. 3c). As expected fromsequence analysis, the positions of the catalytic Cys-360 (start ofα5) and Asp-512 (behind β5) are analogous to UFSPs andATG49,11. However, the catalytic His-491 of ZUFSP is providedby the start of β4, whereas the catalytic histidine of UFSP andATG4 proteases would have been expected between β5 and α10.In accordance with its catalytic role, His-491 is universallyconserved within the ZUFSP family (Supplementary Fig. 1),and its essentiality for ZUFSP activity could be established(Supplementary Fig. 2d). Through the use of a different catalytichistidine, the active site geometry of ZUFSP resembles that ofpapain17 and is almost a mirror image of the UFSP and ATG4active sites (Fig. 3c). Both histidine positions appear equallysuited for acidifying the catalytic cysteine. Nevertheless, a fun-damental change in active site architecture, as observed herebetween the ZUFSP and UFSP families, is a rare event for evo-lutionary related proteases that have maintained similar activities.The oxyanion hole of ZUFSP has one remarkable difference toother cysteine proteases with similar active site architecture. InUFSPs and papain, the side chains of Tyr-282 and Gln-19respectively provide polarized hydrogens for the electrophilicpocket. Mutating these polar residues abolishes the proteolyticactivity11,18. In ZUFSP, the corresponding position is occupied bySer-351, which cannot act as hydrogen donor due to its smallerside chain. Consequently, the S351A mutation showed no loss inactivity (Supplementary Fig. 5a, b). Attempts to improve thecatalytic rate by introducing more suitable side chains (S351Q orS351Y) proved unsuccessful (Supplementary Fig. 5a, b), furthersupporting the idea that the oxyanion hole in ZUFSP is differentfrom that in UFSPs and papain.Contribution of the N terminus to ZUFSP catalytic activity.While the crystal structure of ZUFSP in covalent complex withubiquitin offers interesting insights into the catalytic mechanismand linkage specificity, the construct used for crystallization lacksthe predicted UBZ-like zinc fingers. Moreover, the MIU domain,which forms the N-terminal part of the crystallized fragment, isnot contacting the distal S1 ubiquitin. For assessing the con-tributions of the predicted UBDs to ZUFSP activity, a series oftruncation mutants were analyzed in a time course experiment(Fig. 4a, b). When incubating full-length ZUFSP with long K63-linked ubiquitin chains (Ub6+), most of the high-molecular-weight material was reduced to Ub6and shorter forms within 5min, with very little mono-ubiquitin being generated. After 60min, most of the substrate was present as di-ubiquitin, which—being a poor ZUFSP substrate—required several hours for furtherdegradation. A ZUFSP construct starting at position 148, andthus lacking the first two zinc fingers, was slightly more activethan full-length ZUFSP (Fig. 4b). By contrast, the construct usedfor crystallization, starting at position 232 before the MIUα1α2α3α4α10η1β1α5α6α7α8β2α9β3β4α11η2α12β5β7‘β1‘β2‘α1‘β3‘η1‘β4UbiquitinzUBDα2/α3abcUfsp2Standard Front 90° 90°+ZUFSPPA*C360C25‘S294D512N175D418 H491Q19Y282S351H159H420G75* R74*W423G65W342β6Fig. 3 Crystal structure of ZUFSP232-578 in covalent complex with Ub-PA. aOverview of the crystal structure in cartoon representation. The catalyticcore of ZUFSP is shown in gray, ubiquitin in blue. The MIU region on helixα1 of ZUFSP is colored green, the zUBD region in cyan. The putative S1’ubiquitin-binding α2/α3 helices are shown in red. The catalytic triad isshown as sticks and colored orange. bStructural superposition of thecatalytic domain of ZUFSP (blue) and UFSP2 (3OQC, cyan) in twoperspectives. RMS distance is 3.65 Å over 200 residues. cMagnification ofthe active site of ZUFSP (gray). The catalytic triad is colored orange,putative components of the oxyanion hole in green and Trp-423, closing thesubstrate binding groove directly next to the active site, is in dark pink.Ubiquitin is shown in blue color. The active sites of Ufsp2 (cyan) andpapain (violet) are superimposed. Structurally equivalent residues of Ufsp2and ZUFSP are shown as sticks. Important residues of ZUFSP (black), Ufsp2(cyan) and papain (violet) are labeled. Important ubiquitin residue labelscontain asterisks. In the available structure (3OQC) of UFSP2, the catalyticcysteine was mutated to a serine, indicated here as ‘S294NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-03148-5 ARTICLENATURE COMMUNICATIONS | (2018) 9:799 |DOI: 10.1038/s41467-018-03148-5 |www.nature.com/naturecommunications 5Content courtesy of Springer Nature, terms of use apply. Rights reserved domain, was substantially less active against K63 chains. Evenshorter versions, such as the one starting at position 249 after theMIU, or starting at 274 after the α1 helix, were hardly activeagainst K63 chains (Fig. 4b). The same trend was observed whentesting the truncation mutants against Ub4and Ub2(Supple-mentary Fig. 3d, e).The shortest fragment tested for activity starts at position 310and corresponds to the core catalytic domain, analogous to thewild-type Mug105 protein. While both Mug105 and the humancore fragment were active against ubiquitin-AMC (Fig. 4c) andthus properly folded, only Mug105 was able to cleave ubiquitinchains (Fig. 4d). Thus, the surface features conferring K48 spe-cificity to Mug105 are not conserved in the human core fragment.The catalytic activity of the ZUFSP truncations is closely mirroredby the ability of the corresponding proteins to bind ubiquitinchains. In a pull-down experiment with immobilized His-taggedZUFSP truncations and K63-liked ubiquitin chains, only the full-length protein and the truncation starting at 148 showed robustubiquitin-binding (Fig. 4e).Determinants of ZUFSP specificity. Analysis of the contactsurface between ZUFSP and the covalently bound ubiquitinrevealed two major substrate recognition modalities. The inter-face most critical for proteolytic activity is formed by salt bridgesinvolving two acidic residues of ZUFSP (Asp-406 and Glu-428)and the two arginine residues within the C-terminal tail of ubi-quitin (Arg-72 and Arg-74) (Fig. 5a). These side-chain interac-tions position the ubiquitin C terminus next to the catalyticcysteine and are thus absolutely crucial for the cleavage of ubi-quitin chains and the model substrate RLRGG-AMC (Fig. 5b, c).In both assays, mutation of either D406A or E428A rendersZUFSP as inactive as the active site mutant C360A. This Arg–Argrecognition, together with a narrow hydrophobic tunnel directly05001000150020002500300035004000450050000 1020304050(RFU)(Min)Mug105ZUFSP310–578dcea25221711584632800 5 15 30 60 180 O/N 0 5 15 30 60 180 O/N 0 5 15 30 60 180 O/N 0 5 15 30 60 180 O/N 0 5 15 30 60 180 O/NZUFSP ZUFSP148–578 ZUFSP232–578 ZUFSP249–578 ZUFSP274–5781 578ZUFSP:2 29 57 151 183 223 233 248 276 310 578271ZZZZ MzUBDα-2/3 Catalytic domainb2522171103 h18 0 3 18 0 3 18 0 3 18K48 K48K63 K63Mug105 ZUFSP310–578Ub2Ub25225846322522584632Short exposure Long exposure584680α-Smt3α-Ubiquitinα-Ubiquitin–ZUFSPZUFSP148–578ZUFSP232–578ZUFSP248–578Coomassie Coomassie CoomassieUbUb2Ub3Ub4CoomassieMinFig. 4 ZUFSP UBDs contribute to chain cleavage and specificity. aSchematic representation of ZUFSP domain architecture. UBZ-like zinc fingers (Z), MIUdomain (M), novel ZUFSP ubiquitin-binding domain (zUBD) and α2/α3 region are shown as boxes. bActivity of ZUFSP FL and truncations lacking theUBDs against K63-linked Ub6+chains. Positions of the truncated ZUFSP proteins are indicated by arrows. cComparison of Mug105 and ZUFSP310-578activity against ubiquitin-AMC. The shown RFU values are the mean of triplicates. dChain specificity of ZUFSP catalytic core (ZUFSP310-578) compared tofull-length Mug105. Both DUBs were tested against K48- and K63-linked Ub2for the indicated time points. ePull-down analysis of ZUFSP (full-length andtwo N-terminal truncations) against a mixture of K63-linked Ub4and Ub5chainsARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-03148-56NATURE COMMUNICATIONS | (2018) 9:799 |DOI: 10.1038/s41467-018-03148-5 |www.nature.com/naturecommunicationsContent courtesy of Springer Nature, terms of use apply. Rights reserved before the Ser-360, defines the specificity of ZUFSP for the Cterminus of ubiquitin (RLRGG). The tunnel itself is formed bythe core of the cysteine peptidase as well as the side chains ofTyr-267, Trp-423, Gln-489 and Gly-490. At least in the con-formation found in our crystal structure, the space within thetunnel only permits glycine to pass (Supplementary Fig. 4b). Asecond interaction surface is formed by the second half of theZUFSP α1 helix, which binds the Ile-44 patch of the S1ubiquitinin an orientation similar, but not identical to that of the Rabex-5MIU domain (Fig. 5d). The ubiquitin-binding part of the α1 helix,here denoted as zUBD (ZUFSP ubiquitin-binding domain),contacts the Ile-44 patch of ubiquitin at an angle tilted by 20°relative to the MIU orientation. There is also a substantialdifference in the sequence consensus (Fig. 5e), emphasizingthat—despite the similar binding mode–the zUBD is not just anunusual MIU. Close contacts between the zUBD and ubiquitinare formed by two acidic residues (Glu-256 and Glu-259)(Fig. 5f), which differ in their importance for ubiquitin cleavage:while the E259A mutant was hardly active against Ub4, the E256Amutant showed only a modest reduction in activity (Fig. 5g). TheN-terminal half of the ZUFSP α1 helix is formed by its canonicalMIU domain, but is not bound to ubiquitin and only partiallyresolved in the available structure. Nevertheless, the canonicalMIU is ideally positioned to bind a distal S2-ubiquitin K63-linkedto the outgoing S1 ubiquitin present in the structure. A doublemutant targeting the highly conserved Leu-240/Gln-241 residuesof the canonical MIU motif was equally active as wild-typeZUFSP, suggesting that MIU activity is dispensable for chain0200040006000800010,00012,00014,00016,00018,00020,000100 20304050(RFU)(Min)ab252217115846320 3 60 36036 03 6hZUFSP232–578 ZUFSP232–578D406AZUFSP232–578E428AZUFSP232–578C360Aceg25221711584632ZUFSP148-578ZUFSP148–578,Δ–α2/30 10 30 60 120 MinhZUFSPUb2UbUb4Ub3252217115846320 30 Min60 0 30 60 0 30 60 0 30 60ZUFSP148–578 ZUFSP148–578L240A/Q241AZUFSP148–578E256AZUFSP148–578E259AZUFSPUb4Ub3Ub2UbMIUzUBDdUb4Ub3Ub2UbWTD406AE428AC360ACoomassieCoomassieCoomassieZUFSP148-578ZUFSP148–578,Δ–α2/3ZUFSP148-578ZUFSP148–578,Δ–α2/3 ZUFSP148-578ZUFSP148–578,Δ–α2/3 ZUFSP148-578ZUFSP148–578,Δ–α2/3 R72R74L73Q408Q412D406E428Q487Q549Q547Ubiquitin R42zUBD20°Rabex-5 MIUUbiquitin4.03.02.0Bits1.00.04.03.02.0Bits1.00.05101520WebLogo 3.5.05101520WebLogo 3.5.0K63E259Y267F260K6E256L263A46G47I44V70L8 H68 E252fFig. 5 Determinants of chain specificity. aRecognition of ubiquitin C terminus by the catalytic core of ZUFSP. Ubiquitin (blue) and ZUFSP (gray/green)shown in cartoon representation with key residues highlighted as sticks. Blue and black residue labels refer to ubiquitin and ZUFSP, respectively. Saltbridges are indicated by dotted lines. bActivity of C-terminus recognition mutants (ZUFSP232–578 D406A or E428A) against K63-linked Ub4wascompared to ZUFSP232–578 and inactive ZUFSP (ZUFSP232–578 C360A). cActivity of mutants described in bagainst RLRGG-AMC. The RFU values shownare the means of triplicates. dStructural superposition of ubiquitin-binding interfaces to zUBD (cyan, this work) and the MIU domain of Rabex-5 (2FIF,orange). Orientation of the two helical ubiquitin-binding domains differ by 20°. eSeqLogo45 representation of the consensus sequences for the MIUmotif14 (top) and the zUBD derived from the ZUFSP family as shown in Supplementary Fig. 1(bottom). fMagnification of the interaction interface betweenubiquitin and zUBD. Relevant residues are shown as sticks and labeled black in case of ubiquitin and cyan in case of zUBD. Electrostatic interactions areindicated as dotted lines. gActivity of a MIU mutant (ZUFSP148–578 L240A/Q241A) and two zUBD mutants (ZUFSP148–578E256A and E259A) on K63-linked Ub4, in comparison to wild-type ZUFSP148-578.hActivity time course of the α2/3-deletion mutant ZUFSP148–578; Δ-α2/3 on K63-linked Ub4chains,compared to activity of the parental ZUFSP148–578 constructNATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-03148-5 ARTICLENATURE COMMUNICATIONS | (2018) 9:799 |DOI: 10.1038/s41467-018-03148-5 |www.nature.com/naturecommunications 7Content courtesy of Springer Nature, terms of use apply. Rights reserved cleavage (Fig. 5g). The positioning of the α2/α3 hairpin protru-sion is expected to contact the proximal (S1’) ubiquitin, which isnot part of the available structure. A ZUFSPΔ277-310 mutantlacking this region showed a strongly reduced activity againstUb4chains (Fig. 5h), suggesting a contribution of the α2/α3region to ubiquitin recognition. All mutants tested in Fig. 5g, hremained fully active against the model substrate RLRGG-AMC(Supplementary Fig. 2e, f), suggesting that the observed differencein chain cleavage activity reflects substrate recognition rather thanprotein misfolding.abde–ZUFSPZUFSP148–578ZUFSP232–578–ZUFSPZUFSP148–578ZUFSP232–578–ZUFSPZUFSP148–578ZUFSP232–578–ZUFSPZUFSP148–578ZUFSP232–578ss DNA ds DNA OriL OriL+6Active ZUFSPZUFSP C360ARPA1UbiqUbiqUbiqUSP11LGALS7 TCEAL1LRRC15C2CD6 KIAA1211BCL11ASERPINB5SSBP1RPA2RPA1RPA1SSBP1SSBP1RPA2RPA2RPA3RPA3RPA3UBR5USP11 USP11UBR2UBR5UBR5TCEAL1 TCEAL1KRT6BMYO18B MYO6TOP3AZUFSP+NucleaseAverage log2(intensity ratio ZUFSP/control)Average log10(intensity)Before AftereGFPmVenus-ZUFSPZUFSP-eGFPmVenus-ZUFSPZUFSP DAPI PhaCo2580584632254632α-FLAGα-RPA32– – + + FLAG-ZUFSP––+ + Nuclease c11.511.010.510.09.59.08.58.07.57.0–20246–20246–20246Fig. 6 ZUFSP localization and interaction network. aFLAG-tagged versions of inactive ZUFSPC360A (left) and active ZUFSP (middle, right) were expressedin HEK293T cells and co-precipitating proteins quantified by mass spectrometry. Log2 enrichment ratios relative to uninduced/untransfected controls areplotted against log10 signal intensity. The right panel shows the results after nuclease treatment. Consistently enriched proteins are labeled in color, red forDNA-dependent and blue for DNA-independent enrichment. The bait ZUFSP is off-scale and hence not shown; the x/ycoordinates are 9.7/11.1 (left), 11.0/11.2 (middle), and 10.5/11.6 (right panel). bFLAG-tagged ZUFSP was immunoprecipitated from HEK293T cells in the presence or absence of nuclease.Coimmunoprecipitated endogenous RPA32 was visualized with α-RPA32 4E4. cElectrophoretic mobility shift assay (EMSA) comparing the DNA-bindingpreferences of the full-length ZUFSP to the N-terminal truncations (ZUFSP148–578 and ZUFSP232–578). All constructs were tested against a panel ofoligonucleotides previously tested for SSBP1 binding28, including ssDNA, dsDNA, short hairpin (OriL), and long hairpin (OriL+6). dLocalization of ZUFSPN-terminally fused to mVenus or C-terminal fused to eGFP (green) was visualized in fixed U2OS cells. Cells are shown in phase contrast (PhaCo) andnuclei are stained with DAPI (blue). Scale bar =10 µm. eLocalization of mVenus-tagged ZUFSP to sites of NIR laser-induced DNA damage in U2OS cells(top panels), as compared to eGFP alone (bottom panels). Images were taken immediately before (left) and 10s after 800 nm laser irradiation (right).Scale bar =10 µmARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-03148-58NATURE COMMUNICATIONS | (2018) 9:799 |DOI: 10.1038/s41467-018-03148-5 |www.nature.com/naturecommunicationsContent courtesy of Springer Nature, terms of use apply. Rights reserved ZUFSP localization and interaction network. For gettinginsights into the biological function of ZUFSP, cellular interactorsof this DUB were identified. To that end, FLAG-tagged full-lengthZUFSP was stably expressed in HEK293T cells under tetracyclinecontrol. At 24 h after induction, the tagged ZUFSP construct wasimmunopurified together with its binding partners and analyzedby mass spectrometry. For comparison, a transiently transfectedC360A mutant was immunopurified and analyzed accordingly.Relative to the uninduced and untransfected controls, both wild-type and inactive ZUFSP co-purified with a nearly identical set ofhighly enriched proteins, consisting of RPA1, RPA2, RPA3(constituents of replication protein A (RPA)), SSBP1 (mito-chondrial replication protein), the deubiquitinase USP11, theubiquitin ligase UBR5, ubiquitin itself, and the uncharacterizedprotein TCEAL1 (Fig. 6a). Since the RPA proteins and SSBP1 areknown to bind single-stranded DNA (ssDNA), we tested theinteractions for DNA dependence by repeating the co-purification in the presence of nuclease. After removal of DNA,all four ssDNA binding proteins were no longer enriched, whilebinding to USP11, TCEAL1, and ubiquitin was not affected. TheDNA-dependent binding of ZUFSP to the RPA complex wasconfirmed by co-immunoprecipitation experiments. Full-lengthZUFSP expressed in HEK293T cells co-precipitated endogenousRPA2, as visualized by an α-RPA32 antibody, while no RPA2 co-precipitation was observed in the presence of nuclease (Fig. 6b).To test whether ZUFSP is itself an ssDNA binding protein,electrophoretic mobility shift assays (EMSAs) were performed.Full-length ZUFSP was able to partially shift ssDNA, hairpin anddouble-stranded DNA (dsDNA) oligonucleotides, while the232–578 truncation was a better binder (Fig. 6c). The poorbinding of full-length ZUFSP and the preference for structurescontaining dsDNA makes it unlikely that a direct ssDNA bindingaccounts for the DNA-dependent interaction with RPA subunitsand SSBP1. Since available ZUFSP antibodies failed to stainendogenous ZUFSP in immunofluorescence experiments, thesubcellular localization in U2OS cells was determined for ecto-pically expressed ZUFSP fused to fluorescent proteins at either Nor C terminus. Both constructs showed a uniform distributionthroughout cytoplasm and nucleus (Fig. 6d). Two-photon near-infrared (NIR) laser microirradiation of U2OS nuclei lead to apartial and short-lived ( 1 min) recruitment of ZUFSP to irra-diated areas (Fig. 6e).DiscussionThe discovery that C6orf113/ZUFSP and its homologs form aseventh class of deubiquitinating enzymes is interesting in severalrespects, including DUB evolution, catalytic mechanism, and therole of ubiquitin-binding domains in determining DUB specifi-city. The existence of—so far—seven different DUB classes alsoraises the question of how the different deubiquitination tasks aredistributed between the different classes.In sequence databases, the previously uncharacterized proteinC6orf113 is referred to as ZUFSP (zinc finger with UFM1-specificpeptidase domain protein) because in the Pfam database19 itssequence scores significantly against a Hidden Markov Model(PF07910) describing the UFM1 proteases. However, dependingon the exact sequence search method and parameters, the ZUFSPfamily appears to be nearly equidistant to the UFSP and ATG4families, with a distance resembling that between UFSP andATG4 (Supplementary Fig. 6). Thus, ZUFSP is clearly not amember of the UFSP family, but rather forms a superfamily withthe processing enzymes for UFM1 and Atg8/Atg12. Both UFM1and Atg8/Atg12 are extremely divergent members of theubiquitin-like modifier family with no overt sequence similarityto ubiquitin. Unlike the canonical GG motif found at the Cterminus of most ubiquitin-like modifiers, UFM1 and Atg8/Atg12have only the second glycine residue conserved. Considering thehigh divergence between the three modifier types, it was sur-prising that one superfamily member is a linkage-specific endo-cleaving ubiquitin isopeptidase. A possible explanation for thesubstrate switch might be the different active site geometry, incombination with the presence of multiple UBDs that determinecleavage specificity within the ZUFSP/Mug105 family.In evolutionary terms, the ZUFSP family is rather ancient, withrecognizable members from all eukaryotic kingdoms. However,the phyletic distribution is characterized by many independentgene loss events, suggesting that ZUFSP function is either notuniversally required or that another DUB has been co-opted incertain lineages. ZUFSP family members are widespread in ani-mals, plants, and fungi; they are also found in several other taxa,including Cryptophyta,Alveolata,Amoebozoa, and Rhizaria.Gene loss events affect many common model organisms: unlikesome other nematodes, Caenorhabditis elegans lacks a ZUFSP-like protein; the same is true for the insect model Drosophilamelanogaster and the budding yeast Saccharomyces cerevisiae,which both lack this DUB despite clear ZUFSP homologs beingpresent in other Dipterans and Ascomycetes. This heterogeneity isalso seen in the surprisingly diverse domain architectures ofZUFSP family members. All family members share the conservedcatalytic domain anchored to the C terminus. Several protistshave minimalistic proteins resembling Mug105, while the N-terminal regions of most taxa contain at least one ubiquitin-binding domain. Recurring architectures include 1xUBZ (mostplants), 2xUBZ (most ascomycete fungi), 5xUBZ and 1xzUBD(insects), or 3xMIU but no zinc fingers in several alveolates(Supplementary Fig. 7). The UFSP proteases have also been lostin many lineages, together with the entire UFM1-modificationsystem, but there is little correspondence to the species lackingZUFSP family proteases. The absence of UFM1 and UFSPs fromthe yeast S. pombe underscores once more that Mug105 has noconnection to UFM1 signaling. Despite their overall similarity instructure and sequence, ZUFSP proteases use an active site his-tidine residue different from that used in the UFSP and ATG4families. Apart from the active site residues, all three familiesshow a structural fold similar to the papain-type proteases, a foldthat is also used by most other DUBs and UBL proteases20.Infact, the active site arrangement of ZUFSP, with the catalytichistidine upstream of the catalytic aspartate (C-H-D), is typical ofpapain fold proteases including USP-type deubiquitinases. It istherefore likely that the UFSP and ATG4 families evolved from aZUFSP-like ancestor, and that the C-D-H active site of the twoUBL-protease families is a derived feature.The globular portion of ZUFSP, which is resolved in thecovalent complex structure, has two regions of contact with thedistal (S1) ubiquitin, which—at least partially—explains the spe-cificity for K63-linked ubiquitin chains. The recognition of thetwo Arg residues in the ubiquitin C terminus through salt bridgesis not unusual for DUBs. A similar arrangement can be seen inthe substrate complex of USP21 and other USP-type deubiqui-tinases21. The resulting interactions position the ubiquitin Cterminus favorably relative to the active site and should help inthe recognition of the sequence R-x-R-G-G. Since mutating onlyone of the two acidic recognition counterparts is sufficient toabrogate the activity, recognition of both C-terminal arginineresidues is required. Accordingly, the peptide-based model sub-strate RLRGG-AMC is cleaved very effectively, while NEDD8(ending on ALRGG), SUMO1 (EQTGG), and SUMO2 (QQTGG)are not processed. However, ISG15 also ends on RLRGG but isnot recognized by ZUFSP, showing that a second recognitionlayer must be in place. Contact site analysis suggests that thebinding of the newly defined zUBD domain to the Ile-44-patch ofNATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-03148-5 ARTICLENATURE COMMUNICATIONS | (2018) 9:799 |DOI: 10.1038/s41467-018-03148-5 |www.nature.com/naturecommunications 9Content courtesy of Springer Nature, terms of use apply. Rights reserved the S1 ubiquitin is responsible for the second recognition event,most likely enhanced by binding of the adjacent MIU domain tothe S2-ubiquitin, which is not part of the structure. The zUBDdomain binds ubiquitin in an approximately MIU-like manner,albeit with a different angle and a different consensus sequencefor the contacting residues. The combination of zUBD and MIUon a contiguous helix forms a suitable surface for the recognitionof K63 linkages, analogous to the recognition of K63 chains by thetandem-UIM domains of Rap8022. The available structure doesnot reveal how this second binding interface excludes ISG15, aubiquitin fold with the proper RLRGG C terminus but withoutIle-44 patch, from being cleaved by ZUFSP. One attractive modelwould posit that in the absence of ubiquitin, the α1-helix wouldfold back and obstruct access to the active site—at least forsubstrates larger than the RLRGG-AMC pentapeptide. On theother hand, the α1-deficient ZUFSP274–578 variant is severelydefective for ubiquitin chain cleavage, suggesting that the zUBD/MIU binding to the Ile-44 patch, together with hydrogen bondsformed by the ubiquitin-specific Arg-42 residue (Fig. 5a), mightjust be a necessary contribution to the overall substrate affinity.Particularly intriguing is the different linkage specificity ofhuman ZUFSP and its S. pombe homolog Mug105. While inZUFSP, the K63 specificity can be largely rationalized by the rigidpositioning of recognizable UBDs relative to the active site, thelack of a Mug105 crystal structure precludes the identification ofthe S1 and S1’sites required for K48 specificity. Since Mug105 is aradically reduced version of the ZUFSP architecture, it is safe toassume that the substrate recognition sites of Mug105 are foundon surfaces that are not accessible in the human ZUFSP protein.A truncation of human ZUFSP310-578, down to the core peptidasedomain as found in Mug105, did not recapitulate the shift toK48 specificity, but rather abrogated chain cleavage altogether.Apparently, the surface required for K48 recognition is notconserved in the human protein. While the maximally shortenedZUFSP310–578 no longer cleaved ubiquitin chains, it was stillactive against ubiquitin-AMC and RLRGG-AMC, showing thatthe truncated protein is properly folded and has maintained afunctional active site. In summary, human ZUFSP has a uniquemodular architecture, consisting of a catalytic core made forcleaving after R-x-R-G-G motifs, but not able to efficiently recruitubiquitin chains to the active site. The necessary capability forubiquitin recruitment and linkage specificity is conferred to thecore domain by a number of ubiquitin-binding domains, such asthe rigidly linked zUBD and MIU, as well as the flexibly linkedUBZ-like domains. It will be interesting to study if the variousUBD classes linked to the catalytic core of other ZUFSP familymembers might confer yet different linkage specificities.Despite their different domain architecture, human ZUFSP andS. pombe Mug105 appear to be orthologs, although it cannot beformally ruled out that both metazoan and fungal lineages haveexperienced gene losses of the true orthologs, leaving ZUFSP andMug105 as pseudo-orthologs23. In either case, the difference inlinkage specificity implies that ZUFSP and Mug105 have eitherassumed different biological roles or work in a biological path-way, which in humans and S. pombe is governed by differentchain types. Neither ZUFSP nor Mug105 are functionally char-acterized. The only fact reported about Mug105 is its transcrip-tional upregulation during meiosis24, which is not informativesince several other meiotically upregulated genes in S. pombe areinvolved in pathways without obvious connections to meiosis.Our interaction studies in HEK293T cells found the threesubunits of RPA as the most enriched interactors, along with themitochondrial ssDNA binding protein SSBP1. This interaction iscorroborated by a recently reported incidental finding of ZUFSPin a screen for RPA interactors25. The co-precipitation of endo-genous RPA with ectopically expressed ZUFSP further supportsthe authenticity of this interaction. RPA and SSBP1 are known toperform analogous roles during replication and homologousrecombination in the nucleus and mitochondria, respectively.During meiotic recombination and recombination repair, the twofactors perform a similar role in stabilizing the ssDNA of thedisplacement loop (D-loop), which arises during strand inva-sion26,27. Due to their different localization, it is unlikely thatRPA and SSBP1 form a complex, suggesting that ZUFSP recog-nizes these two similar factors independently. Both SSBP1 andRPA1 are known to bind ssDNA, which raises the question ofwhether their interaction with ZUFSP is direct or possiblybridged by ssDNA. Indeed, when repeating the interactionexperiment in the presence of a non-specific bacterial nuclease,the enrichment of SSBP1 or RPA components was lost, whileother interaction partners such as USP11 and TCEAL1 were notaffected. These findings suggest that ZUFSP either recognizes anucleic acid containing complex of RPA and SSBP1 or thatZUFSP itself is a DNA-binding protein. As seen in the EMSAexperiments, ZUFSPs do in fact bind to DNA, but severalobservations make it unlikely that ZUFSP DNA-binding prop-erties can explain the DNA-dependent association to RPA andSSBP1. When comparing the DNA-binding profile of ZUFSP tothat of SSBP1 using the same set of oligonucleotides28, it becomesapparent that full-length ZUFSP binds DNA rather poorly andprefers dsDNA-containing binding partners over pure ssDNApartners. By contrast, SSBP1 does not bind to dsDNA and prefersthe ssDNA oligonucleotide over those containing hairpin struc-tures28. Thus, a more complex recognition mode of the replica-tion and recombination factors RPA and SSBP1 has to beassumed.Summing up, the highly reproducible and DNA-dependentinteraction of ZUFSP with nuclear and mitochondrial replicationfactors suggests a role of this DUB class in the regulation ofreplication and/or homologous recombination. This idea iscompatible with the presence of ZUFSP in the nucleus and itstransient recruitment to NIR-induced DNA damage sites. Thebiochemical properties of ZUFSP suggest that it generally targetsK63 chains, rather than particular substrates, after being recruitedto its site of action by one or more of its interaction partners.Ubiquitination by K63-linked chains are a hallmark of severalDNA damage pathways29,30. While the processes activating thedamage-responsive ubiquitin ligases are reasonably well under-stood, the removal of K63 chains during or after resolution of thedamage is more enigmatic. The discovery of ZUFSP as a K63-DUB, making DNA-mediated interactions with RPA and SSBP1,will be instrumental for addressing these important questions.MethodsConstructs and cloning. ZUFSP was cloned from HEK293 complementary DNAand Mug105 from S. pombe genomic DNA (kind gift from J. Dohmen, Universityof Cologne) using Phusion DNA Polymerase (New England Biolabs). For proteinpurification, all constructs were cloned in the pOPIN-S vector31 using the In-Fusion®HD cloning system (Takara Clontech). Point mutations were generatedusing the QuikChange Lightning kit (Agilent Technologies). For the generation ofthe stable cell line full-length ZUFSP was cloned into the pCDNA5/FRT/TO vector(ThermoFisher). ZUFSP was cloned into popinE-3C-eGFP (kind gift from RayOwens, OPPF UK) and pCL-Neo-mVENUS (gift from Niels Gehring, University ofCologne) for localization studies. Constructs for ubiquitin-PA purification(pTXB1-ubiquitin1–75) and pOPIN-S USP21196-565 were a kind gift of D.Komander (MRC LMB Cambridge).Protein purification. Full-length ZUFSP, all truncations and mutants, and Mug105were expressed from pOPIN-S vector with an N-terminal 6His-Smt3-tag. Allconstructs were transformed into Escherichia coli (Strain: Rosetta (DE3)pLysS).Then, 6–12 l cultures were grown in LB media at 37 °C until the OD600 of 0.8 wasreached. The cultures were cooled down to 18 °C and protein expression wasinduced by addition of 0.2 mM isopropyl β-D-1-thiogalactopyranoside (IPTG). Forfull-length ZUFSP and ZUFSP148-578 constructs, 0.1 mM ZnSO4was added inaddition to the IPTG. After 16 h, the cultures were harvested by centrifugation atARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-03148-510 NATURE COMMUNICATIONS | (2018) 9:799 |DOI: 10.1038/s41467-018-03148-5 |www.nature.com/naturecommunicationsContent courtesy of Springer Nature, terms of use apply. Rights reserved 5000 × gfor 15 min. After freeze thaw, the pellets were resuspended in bindin gbuffer (300 mM NaCl, 20 mM TRIS pH 7, 20 mM imidazole, 2 mM β-mercap-toethanol) containing DNase and Lysozyme, and lysed by sonication. Lysates wereclarified by centrifugation at 50,000 × gfor 1 h at 4 °C and supernatant was used foraffinity purification on HisTrap FF columns (GE Healthcare) according to themanufacturer s instructions. For all constructs, except the ones used for pull-downanalysis, the 6His-Smt3 tag was removed by incubation with Senp1415-644 andconcurrent dialysis in binding buffer. The liberated affinity-tag and Senp1 wereremoved by a second round of affinity purification with HisTrap FF columns (GEHealthcare). If necessary, proteins were further purified by anion exchange chro-matography (HiScreen Q HP, GE Healthcare) or cation exchange chromatography(HiScreen SP HP, GE Healthcare). All proteins were finally subjected to sizeexclusion chromatography (HiLoad 16/600 Superdex 75 or 200 pg) in 20 mM TRISpH 7, 150 mM NaCl, 2 mM dithiothreitol (DTT), concentrated using VIVASPIN20 Columns (Sartorius), flash frozen in liquid nitrogen, and stored at −80 °C.MBP-Ufsp2 was expressed and purified as described previously32. In brief,MBP-Ufsp2 was expressed as indicated above. Bacteria were lysed in binding buffer(20 mM TRIS pH 7.5, 200 mM NaCl, 2mM DTT) and the supernatant was usedfor affinity purification on a MBPTrap (GE Healthcare). MBP-Ufsp2 was elutedwith binding buffer containing 10 mM maltose, subjected to size exclusionchromatography (HiLoad 16/600 Superdex 200 pg; GE Healthcare) in 20 mM TRISpH 7.5, 150 mM NaCl, 2 mM DTT, concentrated using VIVASPIN 20 Columns(Sartorius), flash frozen in liquid nitrogen, and stored at −80 °C.AMC activity assays. Activity assays of DUBs against AMC-labeled Ub/UbLsubstrates were performed using reaction buffer (150 mM NaCl, 20 mM TRIS pH7, 10 mM DTT) 1 µM DUBs and 10 µM zRLRGG-AMC (BACHEM AG, Swit-zerland), 1 µM Sumo1-AMC, 1 µM Sumo2-AMC (Boston Biochem, Inc., USA), 1µM ISG15-AMC (Boston Biochem, Inc., USA), 1 µM Nedd8-AMC (ENZO LifeSciences GmbH, Germany), LC3A-AMC (Boston Biochem, Inc., USA), or 5 µMUb-AMC (UbiQ-Bio, The Netherlands). The reaction was performed in black 96-well plates (Corning) at 30 °C and released fluorescence was measured using theInfinite F200 Pro plate reader (Tecan) equipped for excitation wavelength of 360nm and an emission wavelength of 465 nm. The measurements were performed intriplicate and the mean is presented.Kinetics. For determination of the specific activity against Ub-AMC the initialvelocity was determined from a measurement of 100 nM DUB against 5 µM Ub-AMC in reaction buffer (150 mM NaCl, 20 mM TRIS pH 7, 10 mM DTT). Steady-state kinetics of ZUFSP or Mug105 against RLRGG-AMC were measured inreactions containing 100 nM DUB and the indicated concentrations of RLRGG-AMC in reaction buffer. Measurements were performed at 30 °C in triplicate.Initial velocities were plotted against the RLRGG-AMC concentrations and fitted tothe Michaelis–Menten equation using Prism 6 (GraphPad) software.Ub and LC3B-PA synthesis. The constructs pTXB1-ubiquitin1–75 pTXB1-LC3B1–119 were used to express ubiquitin or LC3B as a C-terminal intein fusionprotein as described in ref. 33. In brief, the fusion protein was affinity purified inbuffer A (20 mM Hepes, 50 mM sodium acetate, pH 6.5, 75mM NaCl) fromclarified lysates using Chitin Resin (New England Biolabs) following the manu-facturer s protocol. On-bead cleavage was performed by incubation with cleavagebuffer (buffer A containing 100 mM MesNa (sodium 2-mercaptoethanesulfonate))for 24 h at room temperature (RT). Resin was washed extensively with buffer A andpooled fractions were concentrated and subjected to size exclusion chromato-graphy (HiLoad 16/600 Superdex 75) with buffer A. To synthesize Ub/LC3B-PA,300 µM Ub/LC3B-MesNa were reacted with 600 µM propargylamine hydro-chloride (Sigma Aldrich) in buffer A containing 150 mM NaOH for 3 h at RT.Unreacted propargylamine was removed by size exclusion chromatography andUb/LC3B-PA was concentrated using VIVASPIN 20 Columns (3 kDa cutoff,Sartorius) flash frozen and stored at −80 °C.Rho-UFM1-PA synthesis. UFM1 was synthesized by total linear solid-phasepeptide synthesis (SPPS) on a Syro II MultiSyntech Automated Peptide synthesizerusing standard 9-fluorenylmethoxycarbonyl (Fmoc) based solid-phase peptidechemistry at a 40 μmol scale, using fourfold excess of amino acids relative to pre‐loaded Fmoc amino acid trityl resin (0.2 mmol/g, Rapp Polymere GmbH). Peptidecouplings were performed using benzotriazol-1-yl-oxytripyrrolidinophosphoniumhexafluorophosphate (PyBOP, 4 equivalent (equiv)) and N,N-diisopropylethyla-mine (DiPEA, 8 equiv) in N-methyl-2-pyrrolidone (NMP) for 45 min. Fmocremoval was executed using 20% pi peridine in NMP for 2 × 2 and 1 × 5 min. 5-Carboxy-Rhodamine-110 was coupled to the N terminus of resin-bound UFM1and subsequently the fluorescently labeled UFM1 was cleaved off the resin usinghexafluoroisopropanol (HFIP) in dichloromethane (DCM) (1:4 v/v) for 2 times for20 min and filtered, thereby only liberating the C-terminal carboxylic acid whileleaving all other protective groups in place. The flow-through was collected andconcentrated in vacuo, followed by coevaporation with dichloroethane (3×) toremove residual HFIP. Subsequently, the protected peptide was dissolved in DCMand reacted with PyBOP (5 equiv), DiPEA (15 equiv), and propargylamine (15equiv) for 16 h. The reaction was concentrated in vacuo and treated with 90.5%trifluoroacetic acid, 5% water, 2.5% phenol, and 2% tri-isopropylsilane for 2.5 h toglobally remove all protective groups. The fully deprotected peptide was pre-cipitated from Et2O/Pentane (1:1, v/v) and subsequently redissolved in dimethylsulfoxide/water (1:9, v/v) and purified using reverse-phase high-performance liquidchromatography. Lyophilisation of the appropriate fractions yielded the targetactivity-based probe, which was analyzed by liquid chromatography–mass spec-trometry (LC-MS; Waters 2795 Separation Module (Alliance HT), Waters 2996Photodiode Array Detector (190–750 nm), Phenomenex Kinetex C18 (2.1 × 100,2.6 μm) column, and LCTTM orthogonal acceleration time of flight massspectrometer.Suicide probe assay. DUBs were prediluted to 2× concentratio n (10 µM) inreaction buffer (20 mM TRIS pH 7, 150 mM NaCl and 10mM DTT) and 1:1combined with 100 µM Ub-PA, LC3B-PA, 2K48-VME (UbiQ-Bio), or 2K63-VME(UbiQ-Bio). After 16 h of incubation at 4°C, the reaction was stopped by additionof Laemmli buffer, resolved by sodium dodecyl sulfate–polyacrylamide gel elec-trophoresis (SDS-PAGE), and Coomassie stained. For demonstrating Ub specificityof ZUFSP, full-length ZUFSP was incubated with either Cy5-Ub-Propargylamine15(Cy5-Ub-PA) (10 µM) or Rho-UFM1-Propargylamine (Rho-Ufm-PA ) (10 µM) in50 mM Tris-HCl pH 7.5, 100 mM NaCl, and 2 mM DTT at 37 °C for 30 min.The reaction was quenched by the addition of reducing sample buffer and heatingat 95 °C for 3min. Samples were resolved by SDS-PAGE and labeled enzymes werevisualized by in-gel fluorescence scanning using the Typhoon FLA imaging system(GE Healthcare Life Sciences) (λex/ λem =625/ 680 nm and λex/λem=480/ 530nm) and subsequent silver staining.Chain generation. Met1-linked di-ubiquitin was expressed as a linear fusionprotein and purified by ion exchange chromatography and size exclusion chro-matography. K11-, K48-, and K63-linked ubiquitin were enzymatically assembledusing UBE2SΔC (K11), CDC34 (K48), and Ubc13/UBE2V1 (K63) as previouslydescribed34,35. In brief, ubiquitin chains were generated by incubation of 1 µM E1,25 µM of the respective E2 and 2 mM ubiquitin in reaction buffer (10 mM ATP, 40mM TRIS (pH 7.5), 10 mM MgCl2, 1 mM DTT) for 18 h at RT. The reaction wasstopped by 20-fold dilution in 50 mM sodium acetate (pH 4.5) and chains ofdifferent lengths were separated by cation exchange using a Resource S column (GEHealthcare). Elution of different chain lengths was achieved with a gradient from 0to 600 mM NaCl.Chain cleavage assays. DUBs were preincubated in 150 mM NaCl, 20 mM TRISpH 7, and 10 mM DTT for 10 min. The cleavage was performed for the indicatedtime points with 5 µM DUBs and either 25 µM di-ubiquitin (K11, K63, andK48 synthesized as described above, others from Boston Biochem) or 5 µM tetra-ubiquitin (Boston Biochem) at RT, stopped with Laemmli buffer, resolved by SDS-PAGE, and Coomassie stained.Crystallization. 100 µM ZUFSP232-578 was incubated with 200 µM ubiquitin-PAfor 18 h at 4 °C. Unreacted ZUFSP and Ub-PA were removed by size exclusionchromatography. The covalent ZUFSP232-578 and Ub-PA complex (12 mg/ml) wascrystallized using the vapor diffusion sitting drop method. Crystallization trialswere set up with drop ratios of 1:2, 1:1, 2:1 protein solution to precipitant solutionwith a total volume of 300 nl. Initial crystals appeared in PEG/Ion (HamptonResearch) E4 (0.2 M sodium malonate pH 5, 20% polyethylene glycol 3350 (PEG3350)) after 2 days at 4 °C. Optimization was carried out with 3 µl drops (protein/precipitant ratios: 2:1, 1:1 and 1:2) and precipitant solutions varying in pH or PEG3350 concentration respectively. Best crystals were obtained from the crystal-lization trial where protein solution was mixed 1:1 with precipitant solutioncomposed of 0.2 M sodium malonate pH 5, 20% PEG 3350. Crystals were flash-cooled in reservoir solution containing 20 % (v/v) glycerol.Data collection, phasing, model building, and refinement. Diffraction data werecollected at beamline P13 at EMBL Hamburg, Deutsches Elektronen-Synchrotron(DESY), Hamburg, Germany, and processed using X-ray detector software (XDS)36.For S-SAD (single-wavelength anomalous dispersion of S atoms) phasing, a highlyredundant dataset at 2.06 Å was collected at 6 different κ-angles with a 360° sweepeach, and subsequently initial phases were determined using the autosol.phenixroutine37. Afterwards, the model was built manually in Coot and with autobuilt.phenix38,39. Iterative cycles of refinement between building steps were performed withphenix.refine40 using the high-resolution dataset (1.7 Å) recorded at a wavelength of0.97 Å. Restrains of the propargyl moiety were calculated using phenix.elbow41.Ubiquitin-binding assay. A total of 20 µl Nickel resin (MagneHis™Protein Pur-ification System, Promega) was saturated with 6His-Smt3 tagged ZUFSP trunca-tions in 200 µl binding buffer (20 mM TRIS pH 7.5, 150 mM NaCl, 20 mMimidazole and 0.1% NP-40) and incubated for 1 h at 4 °C. All truncations con-taining the catalytic domain were inactivated by a C360A mutation. The resin waswashed three times with binding buffer and afterwards incubated with the twofoldmolar excess of K63-linked ubiquitin chains for 2 h at 4 °C. The washing steps wererepeated and the protein was eluted from the beads by addition of 30 µl LaemmliNATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-03148-5 ARTICLENATURE COMMUNICATIONS | (2018) 9:799 |DOI: 10.1038/s41467-018-03148-5 |www.nature.com/naturecommunications 11Content courtesy of Springer Nature, terms of use apply. Rights reserved buffer. The proteins were separated via SDS-PAGE and the subsequent westernblots were visualized with α-Smt3 (1:10000; kind gift of J. Dohmen, University ofCologne) or α-ubiquitin P4D1 antibody (1:5000; 3936S; Cell Signaling Technol-ogy), respectively.Cell culture, transfection, and stable cell line generation. HEK293T (ATCC®CRL-3216™), HEK293 Flp-In T-REx (ThermoFisher Scientific), and U20s (ATCC®HTB-96™) were maintained by serial passage in Dulbecco’s modified Eagle smedium, high glucose (Invitrogen Life Technologies) supplemented with 2 mM ofL-glutamine, 1 mM of sodium pyruvate, 1× minimal essential medium non-essential amino acids, 100 U/ml of penicillin, 100 µg/ml of streptomycin (PAA),and 10% fetal calf serum (Biochrom). The stable 3xFLAG-ZUFSP cell line wasgenerated by co-transfection of HEK293 Flp-In T-Rex with pCDNA5 /FRT/TO3xFLAG-ZUFSP and pOG44 vectors followed by selection with 0.1 mg/mlHygromycin B (Sigma Aldrich). Single colonies were picked and expression of3×FLAG-ZUFSP was induced by incubation with 1 µg/ml tetracycline for 24 h.Expression of 3xFLAG-ZUFSP was tested by western blot with α-FLAG M2antibody (1:3000; F1804; Sigma Aldrich). Transfections for the generation of stablecell lines or transient expression were performed using TransIT-LT1 (Mirus Bio)according to the manufacturer s protocol.Immunofluorescence microscopy and laser-induced DNA damage. For locali-zation studies, U2OS cells on coverslips were fixed with 3% paraformaldehyde 24 hafter transfection with the respective constructs. Fixed cells were permeabilizedwith 0.1% Saponin in 1× phosphate-buffered saline and stained with DAPI (4′,6-diamidine-2′-phenylindole dihydrochloride; Sigma Aldrich) for 15 min. Subse-quently, cells were mounted on slides using ProLong Gold antifade reagent andanalyzed. Fluorescence images were obtained and processed using a LMS 710confocal scanning laser microscopy system (Zeiss). For NIR laser irradiation forinducing DNA damage, a Leica TCS SP8 MP-OPO confocal scanning lasermicroscope was used at the excitation wavelength of 488 nm and a wavelength of800 nm for inducing the DNA damage. U2OS Cells were seeded in glass-bottomeddishes 35 mm (Ibidi) 2 days before the experiments and transfected with theenhanced green fluorescent protein (eGFP)/mVenus expression vectors 1 daybefore the DNA damage experiments. Time series of 2 to 3 min were performed bytaking a picture every 10 s.Purification and MS analysis of interacting proteins. Flp-In T-REx 293 cells orFlp-in T-REx 293 cells with stable 3xFLAG-ZUFSP integration were seeded in 10 cmdishes and protein expression was induced by addition of 1µg/ml tetracycline.Alternatively, HEK293T cells were transiently transfected with pCMV2b 3xFLAG-ZUFSP using TransIT-LT1 transfection reagent. At 24h after induction or transfec-tion, cells were harvested and lysed in lysis buffer (20 mM TRIS pH 7, 150 mM NaCl,0.1% NP-40, complete protease inhibitor (Roche)) for 30min at4°C, and briefly sonicated before centrifugation at 13,000× gfor 10 min. Thesupernatant was incubated with prewashed magnetic anti-FLAG M2 beads (SigmaAldrich) for 2 h at 4 °C. Beads were washed three times with lysis buffer for 5 min.Last washing step was performed with lysis buffer without NP-40. For western blotanalysis with rabbit α-FLAG (1:3000; F7425, Sigma Aldrich) and rat α-RPA32 (4E4)(1:1000; 2208S; Cell Signaling Technology) proteins were eluted by addition of 2×Laemmli buffer. For MS analysis, bound proteins were eluted with 200 µg/ml 3xFLAGpeptide (Sigma Aldrich) in 6 M urea, 2 M thiourea. Eluted proteins were prepared formass spectrometry by incubation with 1 mM DTT for 1 h, 55 mM iodoacetamide for45 min, 0.005 ng/µl Lys-C for 2 h and 0.005 ng/µl Trypsin for 18 h. For LC-MS/MSanalysis, an EASY-nLC 1000 chromatograph (Thermo Scientific) was coupled to thequadrupole-based Q Exactive Plus (Thermo Scientific) instrument by a nano-sprayionization source. Peptides were separated on a 50 cm in-house-packed column usinga two-solvent buffer system: buffer A (0.1% formic acid) and B (0.1% formic acid inacetonitrile). The amount of buffer B was increased from 7% to 23% within 40 min,followed by an increase to 45% in 5 min, and a washing and re-equilibration stepbefore the next sample injection. The mass spectrometer operated in a Top 10 data-dependent mode, using the following settings: MS1: 70,000 (at 200 m/z)resolution,3e6 AGC target, 20 ms maximum injection time, 300–1750 scan range; MS2: 35,000(at 200 m/z) resolution, 5e5 AGC target, 120ms maximum injection time, 1.8 Thisolation window, 25 normalized collision energy. Data analysis was performed byMaxQuant software, V 1.5.4.742 using the Andromeda search engine43 against thehuman proteome reference data (including splice variants) from Uniprot. Defaultmass tolerance and modification settings were used. Re-quantify, label-free quantifi-cation, and match between runs were enabled. ‘OxidationonM’,‘Phosphorylation onS,T,Y’,and‘GlyGly on K’were allowed as variable modifications. Intensities wereaveraged over biological triplicates, and the log2 of the intensity ratio ‘sample average/control average’was used for enrichment quantification. To account for missingvalues, pseudocounts corresponding to the minimal observed intensity were added tosample and control averages.Electrophoretic mobility shift. The EMSA assays were carried out as previouslydescribed44 using oligonucleotides described previously to bind SSBP128. In brief,0.5 µM of ssDNA (5′-GGGCTTCTCCCGCCTTTTTTCCCGGCGGCGGGAGAAGTAGATTGAAG-3′), labeled at the 5′-end with 6-Fam, was incubated with 5µM of the respective ZUFSP truncation in 40 µl reaction buffer (10 mM T RIS pH7.5, 0.2 mM DTT, 5 µM ZnSO4) for 1 h at 4 °C. The DNA-protein complexes wererun on an 1.5% agarose gel and visualized by a ChemiDoc MP imaging system(BioRad). The dsDNA was created by annealing the 6-Fam labeled ssDNA with areverse complementary oligo. In addition, the stem-loop forming oligonucleotidesOriL: 5′GGGCTTCTCCCGCCTTTTTTCCCGGCGGCGGGAGAAGTAGATTGAAG-3′andOriL+6: 5′GGGCTTCTCCCGCCCCCGCCTTTTTTCCGGCGGCGGGGGCGGGAGAAGTAGATTG-3′were annealed and used as described above. Gels were stained by SybrGold(NEB) and analyzed on the ChemiDoc system.Data availability. The structure of ZUFSP232-578 in covalent complex with Ub-PAhas been deposited in the Protein Data Bank under the accession code 6EI1. Themass spectrometry data have been deposited to the ProteomeXchange Consortiumvia the PRIDE partner repository (https://www.ebi.ac.uk/pride) under the accessioncode PXD008731. All other data supporting the presented findings are availablefrom the corresponding author upon request.Received: 5 October 2017 Accepted: 23 January 2018References1. Komander, D. Rape, M. The ubiquitin code. Annu. Rev. Biochem. 81,203–229 (2012).2. Swatek, K. N. Komander, D. Ubiquitin modifications. Cell Res. 26, 399–422(2016).3. Abdul Rehman, S. A. et al. MINDY-1 is a member of an evolutionarilyconserved and structurally distinct new family of deubiquitinating enzymes.Mol. Cell 63, 146–155 (2016).4. Clague, M. J. et al. Deubiquitylases from genes to organism. Physiol. Rev. 93,1289–1315 (2013).5. Scheel, H., Tomiuk, S. Hofmann, K. Elucidation of ataxin-3 and ataxin-7function by integrative bioinformatics. Hum. Mol. Genet. 12, 2845–2852(2003).6. Burroughs, A. M., Iyer, L. M. Aravind, L. Structure and evolution ofubiquitin and ubiquitin-related domains. Methods Mol. Biol. 832,15–63(2012).7. Ronau, J. A., Beckmann, J. F. Hochstrasser, M. Substrate specificity of theubiquitin and Ubl proteases. Cell Res. 26, 441–456 (2016).8. Ha, B. H. et al. Structural basis for Ufm1 processing by UfSP1. J. Biol. Chem.283, 14893–14900 (2008).9. Satoo, K. et al. The structure of Atg4B-LC3 complex reveals the mechanism ofLC3 processing and delipidation during autophagy. EMBO J. 28, 1341–1350(2009).10. Mevissen, T. E. et al. OTU deubiquitinases reveal mechanisms of linkagespecificity and enable ubiquitin chain restriction analysis. Cell 154, 169–184(2013).11. Ha, B. H. et al. Structure of ubiquitin-fold modifier 1-specific protease UfSP2.J. Biol. Chem. 286, 10248–10257 (2011).12. Bucher, P., Karplus, K., Moeri, N. Hofmann, K. A flexible motif searchtechnique based on generalized profiles. Comput. Chem. 20,3–23 (1996).13. Vogt, B. Hofmann, K. Bioinformatical detection of recognition factors forubiquitin and SUMO. Methods Mol. Biol. 832, 249–261 (2012).14. Penengo, L. et al. Crystal structure of the ubiquitin binding domains of rabex-5 reveals two modes of interaction with ubiquitin. Cell 124, 1183–1195 (2006).15. Ekkebus, R. et al. On terminal alkynes that can react with active-site cysteinenucleophiles in proteases. J. Am. Chem. Soc. 135, 2867–2870 (2013).16. Mulder, M. P., El Oualid, F., ter Beek, J. Ovaa, H. A native chemical ligationhandle that enables the synthesis of advanced activity-based probes:diubiquitin as a case study. Chembiochem 15, 946–949 (2014).17. Yamamoto, D. et al. Refined x-ray structure of papain.E-64-c complex at 2.1-Aresolution. J. Biol. Chem. 266, 14771–14777 (1991).18. Menard, R. et al. Contribution of the glutamine 19 side chain to transition-state stabilization in the oxyanion hole of papain. Biochemistry 30, 8924–8928(1991).19. Finn, R. D. et al. Pfam: the protein families database. Nucleic Acids Res. 42,D222–D230 (2014).20. Komander, D., Clague, M. J. Urbe, S. Breaking the chains: structure andfunction of the deubiquitinases. Nat. Rev. Mol. Cell Biol. 10, 550–563 (2009).21. Ye, Y. et al. Polyubiquitin binding and cross-reactivity in the USP domaindeubiquitinase USP21. EMBO Rep. 12, 350–357 (2011).ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-03148-512 NATURE COMMUNICATIONS | (2018) 9:799 |DOI: 10.1038/s41467-018-03148-5 |www.nature.com/naturecommunicationsContent courtesy of Springer Nature, terms of use apply. Rights reserved 22. Sato, Y. et al. Structural basis for specific recognition of Lys 63-linkedpolyubiquitin chains by tandem UIMs of RAP80. EMBO J. 28, 2461–2468(2009).23. Koonin, E. V. Orthologs, paralogs, and evolutionary genomics. Annu. Rev.Genet. 39, 309–338 (2005).24. Ikebe, C., Konishi, M., Hirata, D., Matsusaka, T. Toda, T. Systematiclocalization study on novel proteins encoded by meiotically up-regulatedORFs in fission yeast. Biosci. Biotechnol. Biochem. 75, 2364–2370 (2011).25. Tkac, J. et al. HELB Is a feedback inhibitor of DNA end resection. Mol. Cell.61, 405–418 (2016).26. Ribeiro, J., Abby, E., Livera, G. Martini, E. RPA homologs and ssDNAprocessing during meiotic recombination. Chromosoma 125, 265–276 (2016).27. Richard, D. J. et al. Single-stranded DNA-binding protein hSSB1 is critical forgenomic stability. Nature 453, 677–681 (2008).28. Miralles Fuste, J. et al. In vivo occupancy of mitochondrial single-strandedDNA binding protein supports the strand displacement mode of DNAreplication. PLoS Genet. 10, e1004832 (2014).29. Hofmann, K. Ubiquitin-binding domains and their role in the DNA damageresponse. DNA Repair (Amst.) 8, 544–556 (2009).30. Jackson, S. P. Durocher, D. Regulation of DNA damage responses byubiquitin and SUMO. Mol. Cell 49, 795–807 (2013).31. Berrow, N. S. et al. A versatile ligation-independent cloning method suitablefor high-throughput expression screening applications. Nucleic Acids Res. 35,e45 (2007).32. Kang, S. H. et al. Two novel ubiquitin-fold modifier 1 (Ufm1)-specificproteases, UfSP1 and UfSP2. J. Biol. Chem. 282, 5256–5262 (2007).33. Borodovsky, A. et al. Chemistry-based functional proteomics reveals novelmembers of the deubiquitinating enzyme family. Chem. Biol. 9, 1149–1159(2002).34. Bremm, A., Freund, S. M. Komander, D. Lys11-linked ubiquitin chainsadopt compact conformations and are preferentially hydrolyzed by thedeubiquitinase Cezanne. Nat. Struct. Mol. Biol. 17, 939–947 (2010).35. Komander, D. Barford, D. Structure of the A20 OTU domain andmechanistic insights into deubiquitination. Biochem. J. 409,77–85 (2008).36. Kabsch, W. Xds. Acta Crystallogr. D Biol. Crystallogr. 66, 125–132 (2010).37. Terwilliger, T. C. et al. Decision-making in structure solution using Bayesianestimates of map quality: the PHENIX AutoSol wizard. Acta Crystallogr. DBiol. Crystallogr. 65, 582–601 (2009).38. Emsley, P., Lohkamp, B., Scott, W. G. Cowtan, K. Features and developmentof Coot. Acta Crystallogr. D Biol. Crystallogr. 66, 486–501 (2010).39. Terwilliger, T. C. et al. Iterative model building, structure refinement anddensity modification with the PHENIX AutoBuild wizard. Acta Crystallogr. DBiol. Crystallogr. 64,61–69 (2008).40. Afonine, P. V. et al. Towards automated crystallographic structure refinementwith phenix.refine. Acta Crystallogr. D Biol. Crystallogr. 68, 352–367 (2012).41. Moriarty, N. W., Grosse-Kunstleve, R. W. Adams, P. D. Electronic ligandbuilder and optimization workbench (eLBOW): a tool for ligand coordinateand restraint generation. Acta Crystallogr. D Biol. Crystallogr. 65, 1074–1080(2009).42. Cox, J. Mann, M. MaxQuant enables high peptide identification rates,individualized p.p.b.-range mass accuracies and proteome-wide proteinquantification. Nat. Biotechnol. 26, 1367–1372 (2008).43. Cox, J. et al. Andromeda: a peptide search engine integrated into theMaxQuant environment. J. Proteome Res. 10, 1794–1805 (2011).44. Lopez-Mosqueda, J. et al SPRTN is a mammalian DNA-binding metalloproteasethat resolves DNA-protein crosslinks. eLife 5, pii: e21491 (2016).45. Crooks, G. E., Hon, G., Chandonia, J. M. Brenner, S. E. WebLogo: asequence logo generator. Genome Res. 14, 1188–1190 (2004).AcknowledgementsThis work was supported in part by a DFG grant (SPP1365) to K.H., and by a Vici grantfrom the Netherlands Foundation for Scientific Research to H.O. We thank ChristianeHorst and Claudia Poschner for expert technical assistance and members of the Hof-mann lab for discussions. We thank David Komander, Jürgen Dohmen, and NielsGehring for constructs and reagents. We thank Gerbrand van der Heden van Noort forthe synthesis of ubiquitin-AMC. We thank the CECAD proteomics and microscopyfacilities, as well as the beamline staff from PETRA III, DESY Hamburg, for support.Funding for the synchrotron visit has been provided by the iNEXT initiative (EU pro-gram Horizon 2020).Author contributionsT.H. performed most of the biochemical work including crystallization, C.P. helped withcrystallization and solved the structure, I.W. contributed to biochemical work and per-formed functional studies, K.K. contributed to functional studies, K.F.W. and H.O.performed UFM1 work and contributed important reagents, U.B. supervised and helpedwith the structural work, K.H. conceived the project, performed bioinformatical analyses,and wrote the manuscript. All authors edited and contributed to the manuscript.Additional informationSupplementary Information accompanies this paper at https://doi.org/10.1038/s41467-018-03148-5.Competing interests: The authors declare no competing financial interests.Reprints and permission information is available online at http://npg.nature.com/reprintsandpermissions/Publisher s note: Springer Nature remains neutral with regard to jurisdictional claims inpublished maps and institutional affiliations.Open Access This article is licensed under a Creative CommonsAttribution 4.0 International License, which permits use, sharing,adaptation, distribution and reproduction in any medium or format, as long as you giveappropriate credit to the original author(s) and the source, provide a link to the CreativeCommons license, and indicate if changes were made. The images or other third partymaterial in this article are included in the article’s Creative Commons license, unlessindicated otherwise in a credit line to the material. If material is not included in thearticle’s Creative Commons license and your intended use is not permitted by statutoryregulation or exceeds the permitted use, you will need to obtain permission directly fromthe copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.© The Author(s) 2018NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-03148-5 ARTICLENATURE COMMUNICATIONS | (2018) 9:799 |DOI: 10.1038/s41467-018-03148-5 |www.nature.com/naturecommunications 13Content courtesy of Springer Nature, terms of use apply. Rights reserved 1.2.3.4.5.6.Terms and Conditions Springer Nature journal content, brought to you courtesy of Springer Nature Customer Service Center GmbH (\"Springer Nature”). Springer Nature supports a reasonable amount of sharing of research papers by authors, subscribers and authorised users (\"Users”), for small-scale personal, non-commercial use provided that all copyright, trade and service marks and other proprietary notices are maintained. Byaccessing, sharing, receiving or otherwise using the Springer Nature journal content you agree to these terms of use (\"Terms”). For thesepurposes, Springer Nature considers academic use (by researchers and students) to be non-commercial. These Terms are supplementary and will apply in addition to any applicable website terms and conditions, a relevant site licence or a personalsubscription. These Terms will prevail over any conflict or ambiguity with regards to the relevant terms, a site licence or a personal subscription(to the extent of the conflict or ambiguity only). For Creative Commons-licensed articles, the terms of the Creative Commons license used willapply. We collect and use personal data to provide access to the Springer Nature journal content. We may also use these personal data internally withinResearchGate and Springer Nature and as agreed share it, in an anonymised way, for purposes of tracking, analysis and reporting. We will nototherwise disclose your personal data outside the ResearchGate or the Springer Nature group of companies unless we have your permission asdetailed in the Privacy Policy. While Users may use the Springer Nature journal content for small scale, personal non-commercial use, it is important to note that Users maynot: use such content for the purpose of providing other users with access on a regular or large scale basis or as a means to circumvent accesscontrol;use such content where to do so would be considered a criminal or statutory offence in any jurisdiction, or gives rise to civil liability, or isotherwise unlawful;falsely or misleadingly imply or suggest endorsement, approval , sponsorship, or association unless explicitly agreed to by Springer Nature inwriting;use bots or other automated methods to access the content or redirect messagesoverride any security feature or exclusionary protocol; orshare the content in order to create substitute for Springer Nature products or services or a systematic database of Springer Nature journalcontent. In line with the restriction against commercial use, Springer Nature does not permit the creation of a product or service that creates revenue,royalties, rent or income from our content or its inclusion as part of a paid for service or for other commercial gain. Springer Nature journalcontent cannot be used for inter-library loans and librarians may not upload Springer Nature journal content on a large scale into their, or anyother, institutional repository. These terms of use are reviewed regularly and may be amended at any time. Springer Nature is not obligated to publish any information orcontent on this website and may remove it or features or functionality at our sole discretion, at any time with or without notice. Springer Naturemay revoke this licence to you at any time and remove access to any copies of the Springer Nature journal content which have been saved. To the fullest extent permitted by law, Springer Nature makes no warranties, representations or guarantees to Users, either express or impliedwith respect to the Springer nature journal content and all parties disclaim and waive any implied warranties or warranties imposed by law,including merchantability or fitness for any particular purpose. Please note that these rights do not automatically extend to content, data or other material published by Springer Nature that may be licensedfrom third parties. If you would like to use or distribute our Springer Nature journal content to a wider audience or on a regular basis or in any other manner notexpressly permitted by these Terms, please contact Springer Nature at onlineservice@springernature.com