Genetically Encoded Crosslinking Enables Identification of Multivalent Ubiquitin-Deubiquitylating Enzyme Interactions

Abstract

Ubiquitin (Ub) proteoforms control nearly every aspect of eukaryotic cell biology through their diversity. Inspired by the widely used Ub C-terminal electrophiles (Ub-E), here we report the identification of multivalent binding of Ub with deubiquitylating enzymes (Dubs) using genetic code expansion (GCE) and crosslinking mass spectrometry. While the Ub-Es only gather structural information with the S1 Dub sites, we demonstrate that GCE of Ub with p-benzoyl-L-phenylalanine enables identification of interaction modes beyond the S1 site with a panel of Dubs of both eukaryotic and prokaryotic origin. Collectively, this represents the next generation of Ub-based affinity probes with a unique ability to unravel Ub interaction landscapes beyond what is afforded by cysteine-based chemistries.

Graphical Abstract

Structurally guided genetic incorporation of p-benzoyl-L-phenylalanine into ubiquitin paired with bottom-up proteomics leads to the next-generation of ubiquitin probes for elucidating interacting interfaces, a feature distinguished from its first-generation counterparts. This study uses these probes to profile interactions of ubiquitin with deubiquitylating enzymes capable of multivalent interactions.

Introduction

Many essential features of life processes at the molecular level are achieved through a diverse network of protein-protein interactions (PPIs). Directing PPIs has been recognized as a challenging yet promising therapeutic feat with proximity-based therapeutic modalities such as PROteolysis TArgeted Chimeras (PROTAC) and DeUBiquitinase TArgeted Chimeras (DUBTAC) having gained attention for their ability to elicit post-translational modifications (PTMs) (1, 2). These modalities take advantage of the ubiquitin (Ub) machinery responsible for attaching and detaching Ub, essentially driving or disrupting downstream PPIs. Protein ubiquitylation and deubiquitylation is an important PTM that regulates key processes in protein degradation and trafficking, mitophagy, cell-cycle control, and DNA damage response (3, 4). The joint effort between E1 Ub activating enzymes, E2 Ub conjugating enzymes, and E3 Ub ligases attach Ub to substrate protein lysine residues. This ligation machinery modularly recognizes their substrates to assemble the many different flavors of Ub modification. Assembling a sense of a “Ub code”, substrates can be monoubiquitylated or polyubiquitylated through a broad range of controlled Ub oligomerization (Met1, Lys6, Lys11, Lys27, Lys29, Lys33, Lys48, and Lys63) (5, 6). Varying permutations, introduction of branching points, and additional PTMs like phosphorylation and acetylation on Ub subunits add to the complexities of Ub chains and contributes to multivalent features that directly impact the relayed signals. Understanding how these ubiquitin proteoforms contribute to overall signaling and how Ub-related machineries encompass these multivalences has been a long-standing question in PTM controlled PPIs and a limiting factor in PPI driven therapeutic intervention (7).

Proteolytic deubiquitinating enzymes (Dubs) must dynamically recognize the diverse set of Ub modifications to render this process reversible (Figure 1). There are nearly 100 human Dubs that are classified into seven different subfamilies: the Ubiquitin C-terminal Hydrolases (UCHs), Ubiquitin Specific Proteases (USPs), Ovarian Tumor pro-teases (OTUs), the JAB1/MPN/MOV34 metalloenzymes (JAMM), Machado-Josephin Domain proteases (MJDs), Zinc-finger containing Ubiquitin Specific Peptidases (ZUFSP), and the Motif Interacting with ubiquitin-containing novel Dub family (MINDY) (8). Dubs play an important role in cellular homeostasis across all eukaryota with their deregulation often accompanied by disease pathogenesis (9). Interestingly, several pathogenic bacteria such as the Legionnaire’s causative agent Legionella pneumophila and the scrub typhus causing Orientia tsutsugamushi have evolved to manipulate and recognize the host ubiquitinated proteome (10–16), a feature also seen in a diverse set of viral lineages (17–19). Both eukaryotic and prokaryotic Dubs recognize different Ub modifications. Dubs OTULIN (20), USP30 (21), Cezanne (22), TRABID (23), and OTUB1 (8) among others display selectivity for Ub chains (Met1, Lys6, Lys11, Lys29, and Lys48 respectively). Further PTMs factor into substrate selectivity as previous genetically encoded phosphorylation of Ub chains has helped reveal discrepancies towards Dub selectivity and activity (24). Collectively, Dubs accommodate multivalent features for Ub modification to render this PTM reversible.

Figure 1.

(A) Ub C-terminal electrophile probes target the active site cysteine on Dubs, profiling only the S1 site. (B) This work strives to develop probes that can profile Dub – Ub interactions that vary from chain recognition to allostery and modular recognition. The strategy of profiling interaction used in this study combines genetic code expansion to incorporate Bpa and bottom-up proteomics to profile the interaction interfaces.

Dub probes are largely centered around molecules that target the active site cysteine residue. These include both small molecule inhibitors and ubiquitin-electrophiles (Ub-Es) generated from an intein-based chemical ligation methods with glycine-mimetic Michael acceptors (25, 26). These activity-based probes react with the nucleophilic catalytic cysteine residue and inactivate the Dub (Figure 1A). Ub-Es are particularly useful as proteomics tools for the identification of new Dubs (27) and as crystallographic tools for elucidating the structural-basis of Ub-Dub active site interaction (28–30). However, there is a need for probes that assess interaction modes outside the active site and independent from thiol-based chemistry. In this study, we use genetic code expansion (GCE) to create Ub proximity-based photocrosslinking probes to characterize Ub multivalent binding modes on Dubs (Figure 1B). The unnatural amino acid p-benzoyl-L-phenylalanine (Bpa) generates a reactive triplet diradical that can carry out C-H bond insertion when exposed to 365 nm light (31). General incorporation of Bpa into different sites of Ub has previously served as interaction probes for Ub binding domains (32), RING E3 ligases (33), assessing the Ub interaction interface with the kinase PINK1 (34), and has reveal structural effects of ubiquitination (35). Here, we show that genetically encoded Bpa in the Thr-9 position serves as a pan-Dub probe, as seen with the linkage specific Ub binding domain probes previously reported (32). This placement allowed for structurally templated crosslinking to any amino acid residue, a feature distinct from nucleophile driven crosslinking with the widely used disuccinimidyl sulfoxide in crosslinking mass spectrometry. We couple this reaction scheme with bottom-up proteomics to give solution phase, structural insight into Ub-Dub interactions. Through this approach, we discerned multivalent interaction interfaces by identifying crosslinked peptides with representative members of five of the seven human Dub families and three pathogenic bacterial Dubs. We observed both S1 and S1’ site interactions in Dubs that possess linkage specificity, allostery in the multi-modular Dub USP5, and even support previously observed sites for possible debranching as in the case of UCHL5’s back site binding. Taken together, these findings demonstrate that GCE coupled with LC-MS/MS gives structural insight into multivalent Ub-Dub binding modes and depict a more dynamic representation of Ub-Dub interactions.

Results and Discussion

The mutant orthogonal Methanococcus jannaschii tRNA_CUA/tyrosyl – tRNA synthetase pair enabled us to site selectively incorporate Bpa into Ub (36–38). Most Ub interactions are mediated by two overlapping hydrophobic patches (Figure 2A), one centered around isoleucine-36 and the other around isoleucine-44. Leucine-8, projecting from a β-hairpin turn is shared by both and contributes to interactions involving these patches (39). Additionally, the β-hairpin turn is oriented towards the C-terminus, another major recognition motif for many Dubs. Leveraging this strategic location, we hypothesized that placing Bpa in the threonine-9 position would offer a pan Ub-protein interaction probe.

Figure 2.

(A) Major structural elements control Ub-protein interactions to relay function: Ile36 patch (red), Ile44 patch (green), and the C-terminus (teal). Of these inter-action hot spots, Thr9 bridges the gap between both hydrophobic patches and it’s resident β-hairpin is directionally parallel to its C-terminus. (B) Table of representative Dubs spanning 5 of the 7 Dub families used in this study. (C) Denaturing SDS-PAGE gels that indicate Ub T9Bpa can react with every eu-karyotic and bacterial Dub tested here in a UV dependent fashion.

We carried out in vitro photocrosslinking reactions of Ub T9Bpa with human Dubs UCHL1 (UCH), USP7 (USP), OTULIN (OTU), AMSH-LP (JAMM), and ZUFSP (ZUFSP) along with the bacterial pathogenic Dubs OtDub, LotB and SdeA^1–200. Figure 2B and 2C shows the crosslinking reactions between the representative Dubs and Ub T9Bpa as visualized by SDS-PAGE Coomassie-stained gels. Despite these structurally and functionally diverse set of enzymes interacting with Ub in divergent modes, in each case, we observe the formation of a species shifted approximately 10kDa from the unmodified enzyme, suggesting addition of the probe following UV irradiation. The efficiency of crosslinking varies based on the Dub. We suspect that these differences may reflect the disparity in the engagement of the threonine residue at the interaction interface and/or extent to which the Bpa substitution is tolerated at that position without impairing binding affinity. Moreover, this probe is recognized by Dubs and is subject to proteolytic cleavage of the His₆ tag from the C-terminal Gly-Gly as evident by the lower molecular weight Ub T9Bpa species produced throughout the span of the crosslinking reaction. This is most evident with Dubs like SdeA^1–200 and UCHL3 as they have higher activity compared to that of UCHL1. Top-down proteomics provided more accurate measurement of these masses along with confirming expected addition of the Ub to both cysteine proteases and Zn-dependent metalloproteases (Supplemental Figures 2, 3). Interestingly, we find that crosslinking with this probe may lead to several different bands as shown for instance with LotB. This variable shift in electrophoretic migration led us to believe that this encompasses discrepancies in interprotein crosslinking species, processing of the C-terminal His₆ tag, or even multivalency in Ub binding. Motivated to identify specific sites of crosslinking, we implemented a bottom-up proteomics approach to reveal putative PPI surfaces.

Initially, we focused on UCHL1. Its interaction is structurally well characterized with the Ub-bound structure having been solved by both X-ray crystallography and cryo-electron microscopy (CryoEM) (28, 40). Our Ub T9Bpa crosslinking results show that there is one equivalent of Ub binding as suggested by SDS-PAGE gel. Photocrosslinking reaction mixtures were subjected to tryptic digestion and analysis by LC-MS/MS. By using MaxQuant’s dependent peptide search functionality, UCHL1 peptides conjugated to the Ub tryptic peptide containing Bpa (TLBpaGK, +668.34 Da and dehydrated TLBpaGK, +650.36) were identified (Figure 3A). Alignment of crosslinked UCHL1 peptides with its structure localizes regions of interaction with the T9 position on Ub (Figure 3B). Fragmentation data in MS2 gives fragment ions within 5ppm of the expected values for more than five ions from the Dub chain (designated as the β chain) and the four diagnostic ions expected from the Ub-Bpa T7-K11 peptidyl fragment (denoted as the α chain) (Figure 3C and Supplemental Figure 4). Taken together the mass shift and the fragmentation, these peptides corroborate structures from both crystallography and CryoEM. Our results support the presence of a persistent, single S1 binding site. In select cases, fragment ion analysis provides residue-level localization of crosslinking sites.

Figure 3.

(A) Table describing the identified peptides by MaxQuant with poten-tial miss cleavages bearing the modification of +668.34 Da. (B) Lo-calized regions identified from UbT9Bpa-UCHL1 WT crosslinking mapped on to the crystal structure of UCHL1 in complex with Ub vinyl methyl ester (PDB 3KW5). (C) Observed fragmentation data (left) reveals extensive coverage of the observed crosslinked peptide and its parent ion of a m/z = 732.13 (+4) observed in the MS1 (right)

A persistent challenge has been the identification of binding sites beyond the dominant canonical S1 site. This site is occupied by the most distal Ub in a chain, the Ub with its C-terminal Gly76 residue as an amide bond with a Lys residue on the neighboring Ub. However, Dubs have evolved to recognize different chain types, harbor allosteric binding sites, and recognize ubiquitylated POIs. These PPIs impart selectivity by the recognition of proximal Ub moieties (neighboring Ub moieties in a chain closer to or with a free C-terminus), monomeric proteoforms of Ub, and interactions with POIs respectively. The photocrosslinking and peptide analysis approach described above is well suited to help give insight into both dominant and transient binding sites. ZUFSP is a recently described Lys63 specific Dub by Kulathu and coworkers that constitutes its own family of Dubs (41). The X-ray crystal structure of ZUFSP in complex with a Ub-E shows that Ub interacts distinctly with the noncanonical Ub binding domain termed the ZUFSP helical arm for distal Ub engagement. However, only inclusions of the Ub-binding ZNF domain ZNF4 of ZUFSP elicit Lys63 chain cleavage activity. As depicted in Figure 4, Ub T9Bpa captured peptides correlating to the already structurally described interaction with the ZUFSP helical arm (crosslinked peptides 4 and 5) and peptides in what was previously hypothesized to be the proximal Ub binding site comprised by α helices 1 and 2 of the catalytic domain (crosslinked peptide 3) (41). Collectively, these data point to specific binding events consistent with ZUFSP distal and proximal interactions that contribute to isomeric selectivity of ZUFSP.

Figure 4.

(A) Table describing the identified peptides by MaxQuant with potential miss cleavages bearing the modification of +668.34 Da. (B) Localized regions identified from UbT9Bpa-ZUFSP crosslinking mapped on to the crystal structure of ZUFSP in complex with Ub vinyl methyl ester (PDB 6FGE). (C) Observed fragmentation data reveals extensive coverage of the observed crosslinked peptide for both distal (left) and proximal (right) binding sites (inset: parent ion of a observed in the MS1.

There are only four members of the UCH family of Dubs in humans. UCHL1, UCHL3, UCHL5, and BAP1 all share a core catalytic papain-like protease domain that exhibits a knotted structure (42–46). While these enzymes share a crossover loop that bridges the active site to control substrate selectivity, their Ub binding properties seem to be drastically different. This prompted us to compare Ub T9Bpa crosslinking with UCH members UCHL1, UCHL3, and UCHL5 to probe these differences. Figure 5A shows an SDS-PAGE Coomassie-stained gel of crosslinking reactions with UCHL1, UCHL3, and UCHL5. Titration of Ub T9bpa to a higher concentration conferred greater occupancy in capturing multiple binding sites (right). It has very recently been shown that the proteosome associated Dub UCHL5 harbors a backside binding site distal from the canonical S1 site for debranching activity (47), which stands in stark contrast to UCHL1 binding. While our work has suggested the presence of only one S1 Ub binding site on UCHL1, high-confidence peptide spectral matches support the presence of a back side binding site for Ub on UCHL5. Crosslinked peptide 14 corresponds to α-helices 5 and 6 and is described by Streiter and coworkers to be the site for Lys48 polyUb interacting at the cryptic binding site, crucial for debranching activity on the proteosome (47). Along with this noncanonical S1 site, the canonical S1 site designated by crosslinked peptides 12, 16 and 19 and a previously uncharacterized site in the helical extension of the ULD domain (peptide 17) was also observed (Figure 5B). To our surprise, the homolog UCHL3’s crosslinking with Ub T9Bpa yields six PSMs consistent with back site binding as shown in our results with UCHL5. While peptides 8 and 11 represent crosslinking for distal site Ub binding, several peptides (peptides 6, 7, 9, and 10) are displaced from the known T9 interface (Figure 5B). The region described by peptide 9 is homologous to the α-helices 5 and 6 of UCHL5 and fragmentation ion spectra shows crosslinking far displaced from the original binding site (Figure 5C and Supplemental Figure 6). We believe these results indicate that this region on UCHL3, much like UCHL5, but not on UCHL1 harbors these previously unknown Ub-binding properties.

Figure 5.

(A) Ub T9Bpa crosslinking reaction with UCH Dubs UCHL1, UCHL3, and UCHL5 both 1:1 (left) and 1:3 (right) equivalence of probe to Dub. (B) Tables including peptides with potential miss cleavages bearing the modification of +668.34 Da as identified by MaxQuant. (C) Crosslinked peptides observed mapped onto the crystal structures of Ub vinyl methyl ester bound to UCHL3 (PDB 1XD3), left, and Ub propargyl bound to UCHL5 (PDB 4UEL), right. (D) Ub T9Bpa crosslinking reaction with USP5. (E) Table of identified peptides including potential miss cleavages bearing the modifica-tion of +668.34 Da from MaxQuant (* denotes regions in the crystal structure that is not resolved). (F) Crosslinked peptides identified from MaxQuant mapped onto the crystal structure of the covalent complex of USP5 and Ub (PDB 3IHP).

There are several Dubs that recognize the different Ub proteoforms. We asked whether this probe could help gather insight into modular Ub recognition. Accordingly, we chose USP5 as a prime example. USP5 contains 4 different Ub binding sites ZnF-UBP, UBP, UBA1, and UBA2 that have previously all been shown to interact with Ub (48). This enzyme hydrolyzes unanchored Met1, Lys6, Lys29, Lys48, and Lys63 polyUb chains to replenish the free monoUb pool. However, binding specificity, as measured by isothermal calorimetry (ITC) (49, 50), has shown higher affinity towards Lys48 and Lys63 chains. Tryptic digest followed by LC-MS/MS analysis of Ub T9Bpa crosslinking (Figure 5D) with the full-length USP5 gives PSMs that indicate crosslinking to all but one previously described binding positions (Figures 5E and 5F). Our results show interaction of Ub to the ZnF-UBP, UBP, and UBA2 domains. Peptides 21 and 29 indicate crosslinking of Ub T9Bpa with the ZnF-UBP. Peptides 24, 25, and 27 show crosslinking with the UBP and UBA2 domains. We have also reported several other regions shown by peptides 20, 22, 23, 26 and 28 that are not modelled in the previously reported X-ray crystal structure (PDB 3HIP). Peptide 20 describes an unmodelled region that exists between the UBP and ZnF-UBP whereas peptides 22, 23, 26, and 28 describe regions on loops of the USP catalytic code, ZnF-UBP, in between the USP and ZnF-UBP domains, and a loop on the UBP domain respectively. Fragment ion analysis of the crosslinked peptide shows more than 5 ions for each site (Supplemental Figure 7). UBA1 was the only domain for which crosslinked peptides were not observed. In addition, we also do not observe crosslinking in the distal site. There is a requirement for the proximal Ub to have an intact, free C-terminus as binding to the ZnF-UBP domain is needed for high affinity engagement with Ub chains (49). Binding at the S1’ site is required for efficient hydrolysis at the S1 site, leading us to believe that there is low occupancy of binding at that site. Taken together, this probe accounts for USP5’s intriguing multivalency and modularity by structurally guided crosslinking.

Conclusion

In summary, this study demonstrates gathering structural information of multivalent and transient PPIs from structurally templated photocrosslinking by means of GCE and LC-MS/MS. We applied it to the Ub field where Ub proteoforms must be recognized by versatile Dubs to render Ub modification reversible.

While leveraging the commonly recognized β-hairpin formed by the β1 and β2 strands, we introduce a bulky, hydrophobic amino acid in Bpa over the endogenous threonine. This may effectively perturb native interaction modes in the context of the experiments carried out here. Further validation from discovered interacting interfaces is required in the native system to support the structural hypotheses developed from the identification of the crosslinked peptide. However, binding ensembles of Ub-Dub interactions captured here confer consistency with known interactions of UCHL1’s S1 site, the S1 and S1’ for ZUFSP, support previously observed sites for possible debranching as in the case of UCHL5 and depict allosteric binding for USP5. In performing these experiments, a similar interaction mode observed in UCHL5 is also present in homolog UCHL3 but not in UCHL1. While further investigations need to be performed to understand the structural basis and functional relevance of these interactions, we believe that these probes, along with a mutagenized library, represent a new generation to the widely used Ub-electrophiles such as Ub-glycine vinyl methyl ester and Ub-glycine vinyl sulfone. Their ability to assess multivalency and tandem application with LC-MS/MS bottom-up proteomics gives a unique dimension to investigate Ub-protein interaction modes.

Experimental Section

Genetic code expansion and purification of ubiquitin

A ubiquitin construct was cloned into pET-28a with an N-terminal Human Influenza Hemagglutinin (HA) tag and a C-terminal hexa-histidine tag. The amber stop codon was introduced in the Thr-9 position of Ub yielding the plasmid pET-28a HA-Ub-His T9TAG (Bioneer). This plasmid was cotransformed into chemically competent E. coli BL21 DE3 cells (Novagen) with the plasmid pEvol-Mj-BpaRS, a generous gift from Professor Abhishek Chatterjee at Boston College, and plated on Luria Broth (LB) – agar containing 50 ug/mL kanamycin and 34 ug/mL chloramphenicol. A single colony was inoculated into a starter culture of LB media containing the same concentrations of kanamycin and chloramphenicol and grown for 16 h at 37 °C. The next day, expression cultures with the respective antibiotics were inoculated (1:100) into LB media and grown to an OD600 = 0.6–0.8 at 37 °C. A final concentration of 1 mM Bpa (Bachem), dissolved initially in 1 M NaOH, was added to the culture followed by neutralizing the pH through addition of equal equivalents of aqueous HCl. To the culture, a final concentration of 1 mM isopropyl β-d-1-thiogalactopyranoside (IPTG) and 0.2% (w/v) L-arabinose was added, and protein expression took place for 16 h at 30 °C in the dark while shaking. The cells are harvested by centrifugation at 7000 rpm for 10 min at 4 °C. The harvested cells were resuspended in pH 7.4 phosphate-buffered saline (PBS) with 400 mM KCl containing 0.5 mg/mL lysozyme and lysed under high pressure using a French press. Lysate was clarified by ultracentrifugation for 1hr at 100,000 × g at 4 °C and applied to 5 mL of Ni-NTA agarose (Qiagen) resin pre-equilibrated with the respective lysis buffer. The resin is washed with 20 column volumes (CV) of 1X PBS 400 mM KCl, 20 CV of 1X PBS with 400 mM KCl and 25 mM imidazole, 10 CV of 1X PBS with 400 mM KCl and 50 mM imidazole and eluted with 8 CV of 1X PBS with 400 mM KCl and 300 mM imidazole. The elution was concentrated, and buffer exchanged into 50 mM Tris pH 7.4 containing 100 mM NaCl using an Amicon 3 kDa molecular weight cut-off concentrator (Millipore-Sigma).

Cloning, expression, and purification of recombinant proteins

Cloned into pGEX-6P-1 (GE-Healthcare), constructs of UCHL1FL, ZUFSP184–578, UCHL3FL, UCHL5FL, SdeA1–200, and LotB15–330 were all expressed as glutathione S-transferase (GST) fusion proteins in E. coli BL21 DE3 cells (Novagen). pGEX-6P-1 ZUFSP184–578 was generously provided by Professor Yogesh Kulathu from the University of Dundee. AMSH-LP231–436 was cloned into pCold-GST (Takara Bio) but expressed and purified the as the other GST fusion protein. Cells were grown while shaking at 37 °C until an OD600 = 0.6–0.8 was reached. Protein expression was induced by the addition of 0.35 mM IPTG and carried out for 18 h at 18 °C. Cells were harvested at 7000 rpm for 10 min and resuspended in pH 7.4 1X PBS with 400 mM KCl and 0.5 mg/mL lysozyme. Clarified lysate was retrieved by ultracentrifugation at 70,000 × g for 1 h and was applied to 5 mL of glutathione-Sepharose resin (GE Healthcare). The resin was washed with 25 CV of pH 7.4 1X PBS with 400 mM KCl and eluted in 250 mM Tris pH 8.0 buffer containing 500 mM KCl and 10 mM reduced glutathione. Tag cleavage was carried out by the addition of GST-tagged Precision Protease to the elution as per the manufacturer’s recommendation (GE Healthcare) during two rounds of dialysis against 4 L of dialysis buffer (pH 7.4 1X PBS with 400 mM KCl and 1 mM dithiothreitol) overnight. Tag subtraction proceeded with applying the dialyzed sample to regenerated glutathione-Sepharose resin twice. The subtracted sample is subject to size-exclusion chromatography (HiLoad 16/600 Superdex 75 pg) by FPLC with a mobile phase of 50 mM Tris pH 7.4 buffer containing 100 mM NaCl, and 1 mM DTT (GE Healthcare). Fractions of interest are pooled and concentrated prior to use.

USP5FL, USP7207–560, OTULINFL and OtDub1–259 cloned into pET-28a, pET-11a, pCold-HisSUMO, and pET-28a respectively. USP5 was a gift from Cheryl Arrowsmith (Addgene plasmid # 25299; http://n2t.net/addgene:25299; RRID:Addgene_25299). These proteins were expressed as His-tagged proteins and lysed similarly as described above. Clarified lysate was applied to 5 mL of Ni-NTA resin (Qiagen) and proteins were purified as reported for Ub T9Bpa above.

Photocrosslinking reactions

Reactions between HA-Ub-His T9Bpa and respective Dubs were carried out in a 1:1 mol ratio (or 1:3 as indicated). In reaction buffer (50 mM Tris pH 7.4, 100 mM NaCl, 1 mM DTT) a total of 15 μM of Dub is incubated with 15 μM of HA-Ub-His T9Bpa (or 45 μM as indicated) for 30 min on ice. The sample is siphoned into two aliquots where one is kept on ice and the other was subjected to irradiation of 365 nm light using a UVP Blak-Ray Lamp (Model XX-15L) (Upland, CA) in a 96-well clear flat-bottom plate (Thermo-Fisher Scientific) for 45 min at 4 °C. These reactions were analyzed by SDS-PAGE through Coomassie staining. In the cases where crosslinked proteins were observed, aliquots of the reaction were collected for further analysis by intact LC-MS and bottom-up proteomics.

Bottom-up and top-down mass spectrometry and analysis

Proteins were denatured in 50 mM ammonium bicarbonate buffer containing 8 M urea 5 mM dithiothreitol at room temperature for 30 min. The reaction was diluted 8-fold with 50 mM ammonium bicarbonate buffer and digested with 1/50 trypsin/protein ratio over night at 37 °C. Reactions were terminated by adding trifluoroacetic acid (0.5% final concentration), and peptides were desalted by microspin C18 cartridges (NEST Group) and dried in a vacuum centrifuge. Peptides dissolved in 0.1% trifluoroacetic acid were separated using a nanoACQUITY UPLC system (Waters).

For bottom-up MS analysis, an Q Exactive HF-X hybrid quadrupole-Orbitrap mass spectrometer (Thermo Scientific) was equipped with a pulled borosilicate capillary ion source and connected to a nanoACQUITY UPLC system (Waters). Peptides were loaded onto a trap column (5 cm × 360 μm OD × 150 μm ID) packed with Aeries C18 (3.6-μm, Phenomenex) with 1% B (mobile phase A was 0.1% formic acid in water, and mobile phase B was 0.1% formic acid in acetonitrile) and were separated on an analytical column (70 cm × 360 μm OD × 75 μm ID packed with 3-μm Phenomenex Jupiter C18 stationary phase) with the following gradient : 1–8% B in 2 min, 8–12% B in 18 min, 12–30% B in 55 min, 30–45% B in 22 min, 45–95% B in 3 min, hold for 5 min in 95% B and 99–1% B in 10 min. The capillary temperature was set to 300 °C and the spray voltage was set to 2.2 kV, and S-lens RF level set to 40%.

MS1 scans were acquired in the Orbitrap with a resolving power of 60,000 FWHM at 200 m/z from mass range 300–1800 m/z. For MS1 scans the AGC target was set to 3 × 106 ions with a max fill time of 20 ms. MS2 scans were acquired using the TopN method with a loop count of 12, an isolation width of 0.7 m/z in the quadrupole, and fragmentation in the HCD cell at a normalized CE of 30. MS2 scans were acquired in the Orbitrap at a resolving power of 45,000 FWHM at 200 m/z with a fixed first mass of 110 m/z. The MS2 AGC was set to 1 × 105 target ions and a max fill time of 100 ms. Monoisotopic precursor selection was enabled, only MS1 signals exceeding 5 × 104 counts triggered MS2 scans, and +1, +7, +8, and unassigned charge states were not selected for MS2 analysis. Dynamic exclusion was enabled with a repeat count of 1 and exclusion duration of 45 seconds.

Data analysis was carried in MaxQuant v2.0.3.0. using its dependent-peptide search functionality. MaxQuant parameters were set up as follows: peptide mass tolerance in first search, 20 ppm; peptide mass tolerance in the main search, 4.5 ppm; PSM and protein FDRs were both set to 0.01; carbamidomethyl on Cys as fixed modification; oxidation on Met and acetyl at protein N-terminus as variable modifications; Trypsin/P and number of max missed cleavages 2; dependent peptide search. Fasta sequence files of protein sequences (Supplementary Information) were used to search against the bottom-up data. The search led to the identification of peptides bearing the +668.34 m/z modification. Identified peptides were cross referenced to ensure the presence of peptide fragments from both the protein sequence and the Ub TLBpaGK peptide along with the presence of diagnostic ion 187.14 m/z.

For top-down MS analysis, an Orbitrap Eclipse Tribrid mass spectrometer (Thermo Scientific) was equipped with a nanoelectrospray source and attached to a nanoACQUITY UPLC system (Waters). Reversed phase separation was carried out on an in-house packed C2 column (100 μm i.d., ~50 cm long, packing material SMTC2MEB2–3-300 from Separation Methods Technologies, Newark, DE). Mobile phases were 0.2% formic acid in water (A) and 0.2% formic acid in acetonitrile (B). After online desalting on a C2 trap column (same material and i.d. as separation column, 5 cm long) for 6 min, a linear gradient with a flow rate of 0.3 μL/min was run from 10 – 50% mobile phase B over 30 min. The capillary temperature was set to 300 °C, the spray voltage was set to 1.85 kV, and S-lens RF level set to 50%. The mass spectrometer was operated in Intact Protein mode set to Standard pressure (8 mTorr in IRM).

MS1 scans were acquired in the Orbitrap with a resolving power of 120,000 FWHM at 200 m/z from mass range 500–2000 m/z. For MS1 scans the AGC target was set to 8 × 105 ions with a max fill time of 50 ms and 5 microscans. MS2 scans were acquired using the TopSpeed method with a cycle time of 3 seconds and an isolation width of 1.6 m/z in the quadrupole. Precursor ions were fragmented using both HCD and ETD independently. For HCD fragmentation, precursor ions were dissociated with stepped normalized collision energies of 15%, 30%, and 45%. For ETD fragmentation, precursor ions reacted reagent ions (target 2 x105 ions) for 15 ms. Fragment ions were analyzed by the Orbitrap with a resolving power of 120,000 FWHM at 200 m/z from mass range 150–2000 m/z. For MS2 scans the AGC target was set to 2.5 × 105 ions with a max fill time of 246 ms. Monoisotopic precursor selection was enabled, only MS1 signals exceeding 5 × 104 counts triggered MS2 scans, and charge states 5–50 were included. Dynamic exclusion was enabled with a repeat count of 1, an exclusion duration of 60 seconds, and limited selection of one charge state per precursor.

For top-down mass spectrometry analysis, neutral monoisotopic masses of precursor ions were calculated from high-resolution data with FreeStyle 1.8 SP2 using the Xtract algorithm. Neutral average masses of precursor ions were calculated from medium-resolution data with Unidec 5.2.1. Neutral monoisotopic masses of fragment ions were calculated with the Xtract algorithm. Fragment ion maps and P-scores were calculated with ProSight Lite 1.4 with a fragment mass tolerance of 10 ppm.