Ubiquitin vinyl methyl ester binding orients the misaligned active site of the ubiquitin hydrolase UCHL1 into productive conformation

Abstract

Ubiquitin carboxy-terminal hydrolase L1 (UCHL1) is a Parkinson disease-associated, putative cysteine protease found abundantly and selectively expressed in neurons. The crystal structure of apo UCHL1 showed that the active-site residues are not aligned in a canonical form, with the nucleophilic cysteine being 7.7 Å from the general base histidine, an arrangement consistent with an inactive form of the enzyme. Here we report the crystal structures of the wild type and two Parkinson disease-associated variants of the enzyme, S18Y and I93M, bound to a ubiquitin-based suicide substrate, ubiquitin vinyl methyl ester. These structures reveal that ubiquitin vinyl methyl ester binds primarily at two sites on the enzyme, with its carboxy terminus at the active site and with its amino-terminal β-hairpin at the distal site—a surface-exposed hydrophobic crevice 17 Å away from the active site. Binding at the distal site initiates a cascade of side-chain movements in the enzyme that starts at a highly conserved, surface-exposed phenylalanine and is relayed to the active site resulting in the reorientation and proximal placement of the general base within 4 Å of the catalytic cysteine, an arrangement found in productive cysteine proteases. Mutation of the distal-site, surface-exposed phenylalanine to alanine reduces ubiquitin binding and severely impairs the catalytic activity of the enzyme. These results suggest that the activity of UCHL1 may be regulated by its own substrate.

UCHL1, a member of the UCH (ubiquitin C-terminal hydrolase) family of deubiquitinating enzymes (DUBs), is a 223-amino acid protein found abundantly and selectively expressed in brain, constituting up to 1–2% of total brain protein (1, 2). In vivo studies suggest that UCHL1 is involved in regulation of ubiquitin pool, apoptosis, and learning and memory, and its absence in mice because of spontaneous intragenic deletions yields phenotypes with neurological defects (3). Mutations in UCHL1 have been implicated in Parkinson disease (PD). A point mutation near the active site that changes Ile93 to Met (I93M) has been linked to an increased risk of developing an autosomal-dominant form of PD (4). Conversely, a common S18Y polymorphism reduces susceptibility to PD (5, 6) and Alzheimer’s disease (7). In addition to its association with neurodegenerative diseases, abnormal expression of UCHL1 is found in many forms of cancer, including lung, colorectal, and pancreatic cancers, and may be related to tumor progression (8, 9). The normal function of UCHL1, however, is not known. Also unknown are how the activity of this abundant neuronal enzyme is regulated and what its true physiological substrates are, although biochemical studies have indicated that UCHL1 can accept short-peptide (α or ϵ-amino-linked) or small-molecule C-terminal conjugates of ubiquitin as substrates, cleaving, as its name suggests, the amide bond following immediately after the C-terminal glycine (Gly76) of ubiquitin (10).

Like other members of the UCH family, UCHL1 is a cysteine protease, with a catalytic triad consisting of a cysteine (Cys90), a histidine (His161), and an aspartate (Asp176). The overall structure of UCHL1 is very similar to that of its nearest UCH relative, UCHL3, which shares 51% of sequence identity with UCHL1. However, unlike UCHL3, the catalytic triad in UCHL1 is misaligned. Whereas the His161-Asp176 distance (2.7 Å) is very much within the expected range, the critical Cys90-His161 pair is separated by 7.7 Å (the Sγ atom of Cys90 from the Nδ1 atom of His161), a distance far greater than expected of a productive cysteine protease (11). This observation raises the question as to how this enzyme would catalyze the hydrolysis of an amide bond as a cysteine protease and how it would act as a ubiquitin hydrolase. To resolve this issue, we sought to determine the three-dimensional structure of the enzyme bound to a substrate analog. Here we report the X-ray structures of the wild-type UCHL1 and its two PD-associated variants, S18Y and I93M, hereafter referred to as UCHL1S18Y and UCHL1I93M, bound to a substrate mimic, ubiquitin vinyl methyl ester (UbVMe).

Results

Structure Determination.

To prepare a stable complex with ubiquitin, UCHL1 and its variants were allowed to react with UbVMe, a derivative of ubiquitin with a glycyl vinyl methyl ester (GlyVMe) group substituting for Gly76 of ubiquitin (Fig. 1A). UbVMe acts as a suicide substrate for cysteine protease DUBs by targeting the active-site cysteine leading to the formation of a covalently bonded DUB–UbVMe complex, in which a thioether bond links the Sγ atom of the active-site cysteine of the DUB to the Cβ atom of the VMe moiety (Fig. 1A) (12). The thioether linkage mimics the thioester reaction intermediate proposed to exist during catalysis of the peptide bond hydrolysis by a cysteine protease. UCHL1 and its variants reacted with UbVMe nearly quantitatively, as judged by a mobility shift on a sodium dodecyl sulfate polyacrylamide gel of approximately 8 kDa, the expected molecular mass of ubiquitin (molecular mass, 8564.5 Da), allowing isolation of the complexes by size exclusion chromatography in milligram quantities suitable for crystallization.

Fig. 1.

Structure of UCHL1S18Y bound to the suicide substrate UbVMe. (A) Schematic representation of the UCHL1S18Y–UbVMe complex formed by covalent attack of the catalytic cysteine of UCHL1S18Y at the α,β-unsaturated bond of the VMe moiety. (B) Ribbon representation of the structure of UCHL1S18Y–UbVMe complex at 2.4 Å resolution. UbVMe is shown in green. Secondary structures of UCHL1S18Y discussed in the text are labeled.

The UCHL1S18Y–UbVMe complex structure was solved at 2.4 Å by molecular replacement using the wild-type UCHL1 and ubiquitin (residues 1–75) as search models (11, 12). Cross-rotation followed by translation search identified a single copy of the complex in the asymmetric unit. The electron density map generated after rigid-body followed by restrained refinement of the model, obtained as the solution to molecular replacement, was interpretable, indicating the success of the molecular replacement search. Refinement using Refmac (13) after rounds of model building in Coot (14) yielded a final model with a crystallographic R factor of 20.9% and an R_free of 25.6% (Table S1). The final model contains a complete UCHL1S18Y chain (residues 1–223), a ubiquitin chain (residues 1–75), the GlyVMe group (modeled as 4-amino methyl butanoate), and 51 ordered solvent molecules. More than 98% of nonglycine residues are placed within the most favorable and additionally allowed regions of the Ramachandran plot, and less than 0.5% are located in the disallowed areas, as defined by the program PROCHECK (15). Structure of apo UCHL1I93M (2.80 Å) was solved by molecular replacement using the wild-type UCHL1 as the search model (SI Methods and Table S1). Structures of UCHL1-UbVMe (2.85 Å) and UCHL1I93M-UbVMe (2.80 Å) were solved by molecular replacement using the UCHL1S18Y–UbVMe complex as the search model (SI Methods and Table S1). The structures of UCHL1-UbVMe and UCHL1I93M-UbVMe are very similar to that of UCHL1S18Y-UbVMe (Fig. S1). We therefore chose to focus our discussion only on the UCHL1S18Y–UbVMe complex because it was determined at the highest resolution.

Overall Structure of UCHL1S18Y in the Complex.

UCHL1S18Y is composed of two lobes, one consisting of five α helices (α1, α3, α4, α5, and α6) and the other consisting of two helices (α2 and α7) and a 6-stranded β-sheet (Fig. 1B). These secondary structures together form an α-β-α sandwich fold that characterizes most structurally known members of the papain family of cysteine proteases, including UCHL3 (16) and the yeast ubiquitin C-terminal hydrolase Yuh1 (17). Between the two lobes lies a relatively deep cleft that harbors the active site, which is composed of three secondary structure elements: a helix (α3), a strand (β3), and a loop (L9) on which the members of the catalytic triad Cys90, His161, and Asp176 reside, respectively. Tyr18 is located on the solvent-exposed face of the first N-terminal helix (α1). This residue is involved in tight hydrogen-bonding interactions with Arg27 with a distance of 2.8 Å separating the hydroxyl oxygen and one of the guanidinium nitrogens. In contrast, the corresponding residue in the wild-type protein, Ser18, is completely solvent-exposed and is not engaged in any interaction with other protein atoms. The overall architecture of UCHL1S18Y in UbVMe-bound complex is quite similar to that of apo wild-type UCHL1 (11), with C^α rmsd of 1.50 Å (Fig. S2). The arrangement of active-site residues, however, is quite different than the apo form, which has the general base His161 at 7.7 Å from the nucleophile Cys90, consistent with an inactive state of the enzyme (16, 18, 19). In the complex, the catalytic residues have adopted a canonical arrangement found in active cysteine proteases with His161 at 3.9 Å from the catalytic Cys90.

Specific Interactions of UbVMe with UCHL1S18Y.

Binding of UbVMe with UCHL1S18Y is substantial, burying 2,548 Å² solvent-accessible surface area. Complex formation involves binding of UbVMe primarily at two sites on the enzyme: the active-site cleft and a solvent-exposed hydrophobic crevice 17 Å away from the active-site cysteine, hereafter referred to as the distal site (Fig. 2A). In terms of the buried accessible surface area on UCHL1S18Y, the UbVMe-binding interface is split almost evenly between the two sites, with the contributions from the active site and the distal site being ∼45% and ∼55%, respectively. The C-terminal segment of UbVMe (Val70 to GlyVMe76) is deeply buried in the active-site cleft, accounting for nearly 46% of the total buried accessible surface. With its backbone in an extended β conformation, this segment of UbVMe is extensively coordinated by hydrogen-bonding and salt-bridge interactions to residues lining the active-site cleft of the enzyme (Fig. 2B), with the active-site cysteine linked to the VMe moiety of the suicide substrate via a thioether bond. In addition to these contacts, intermolecular van der Waals contacts also contribute to the specific recognition of the C-terminal segment of UbVMe. The two hydrophobic residues, Leu71 and Leu73, point into the catalytic cleft by making direct van der Waals contacts to the corresponding UCHL1S18Y residues (Leu71 of UbVMe with Val212 and Arg213 of UCHL1S18Y, and Leu73 of UbVMe with Ile8, Leu52, and Phe160 of UCHL1S18Y). The last two residues of UbVMe, Gly75 and GlyVMe76, fit in the narrowest region of the catalytic cleft (Fig. 2A). The space surrounding the backbone C^α atoms of these Gly and GlyVMe residues is insufficient to accommodate any other side chain, consistent with the selectivity displayed by UCH enzymes for cleaving the amide bond immediately following the terminal Gly-Gly motif of ubiquitin. The UCHL1S18Y-interacting UbVMe residues determined in this study are mostly in agreement with a previous mutational analysis delineating the side chains of ubiquitin required for its recognition by UCHL1 (20). For example, mutation of residues Leu71, Leu73, and Gly76 to Ala on ubiquitin-tryptophan as the substrate significantly reduced the value by approximately 50-, 100-, and 300-fold, respectively (20).

Fig. 2.

Intermolecular contacts between UCHL1S18Y and UbVMe. (A) Molecular surface representation of UCHL1S18Y structure as observed in the crystal structure of its complex with UbVMe (shown as a green ribbon), illustrating the distal- and active-site binding of UbVMe. The active-site cysteine is indicated in yellow. The cross-over loop of UCHL1S18Y (L8 in Fig. 1B) is shown as a purple ribbon. The N terminus of UbVMe is indicated. (Inset) An expanded view of the distal binding site on UCHL1S18Y showing the Phe214 side-chain (Orange) compared to its position in apo UCHL1 (Blue). (B) Active-site interactions of the C-terminal segment of UbVMe (Green Ribbon) with the UCHL1S18Y (Gray Ribbon). Backbone and side chains of interacting residues are shown in stick representation. Oxygen atoms are shown in red, nitrogen atoms in blue, and carbon atoms of UCHL1S18Y and UbVMe (1–75) in gray and green, respectively. Carbon atoms of the GlyVMe residue are shown in yellow. The segment comprising residues 156–159 of UCHL1S18Y is removed for clarity.

The so-called active-site cross-over loop is an important structural feature shared uniquely among the members of the UCH family of DUBs (11, 17). This loop has been proposed to play a pivotal role in substrate selection by UCH enzymes (17, 21). The cross-over loop in UCHL1, comprising of residues Gly150 to Lys157 (loop L8, Fig. 1B), connects the helix α6 on one lobe and the strand β3 on the other and strings across the active-site cleft (11) forming an arch directly over the catalytic Cys90 (Fig 2A). The position and the dimension of this loop in apo UCHL1 are similar to that in UCHL1S18Y–UbVMe complex (Fig S2), with the diameter of the opening enclosed by the loop (the distance measured between the C^α atoms of Glu7 and Val154, the pair of atoms with widest separation across the loop) being approximately 9 and 13 Å in the apo and UbVMe-bound forms, respectively. In the complex, the loop appears to have opened up a little to embrace the C terminus of UbVMe. Comparison of the apo and UbVMe-bound structures suggests that the cross-over loop of UCHL1 is relatively rigid, which may serve the function of a stereochemical gate for selecting substrates; only those ubiquitin conjugates whose C-terminal extension at ubiquitin (the P1' portion of the substrate) can thread through the narrow arch of the loop would be accepted (Fig. 2A). This inference is consistent with a previous biochemical analysis showing that UCHL1 can preferentially cleave small leaving groups such as amino acids and polypeptides from the C terminus of ubiquitin (10). In contrast, UCHL3 possesses a longer (residues 146–167) and more flexible cross-over loop, which can adopt a wide range of conformational states allowing the enzyme to accept larger extensions at the C terminus of ubiquitin (12). In addition to its apparent role as a stereochemical gate for substrate selection, the cross-over loop also contributes key interactions to the specific recognition of the C-terminal segment of UbVMe. For example, the side chain of Arg74 of UbVMe is engaged in electrostatic interaction with that of Asp155 of the cross-over loop, and, reciprocally, the side chain of Arg153 of the cross-over loop is hydrogen-bonded to the backbone carbonyl of Arg72 of UbVMe (Fig. 2B).

Alignment of Active-Site Residues is Induced by Binding of UbVMe.

The distal site on UCHL1S18Y is occupied by the two-residue turn segment (Leu8-Thr9) of the N-terminal β-hairpin of ubiquitin, with Leu8 nestled in the surface-exposed hydrophobic pocket lined by Val31, Leu34, Leu51, Phe214, and Ala216 of the enzyme. Comparison of UCHL1S18Y–UbVMe structure with that of UCHL1 in apo form reveals remarkable differences in the position of the side-chain rings of three residues: Phe214, Phe53, and the general base His161 (Fig. 3A). In the apo form of the enzyme, the aromatic ring of Phe214 is solvent-exposed, facing away from the inner core of the protein (Fig. 2A Inset). Binding of ubiquitin pushes this ring, causing it to swing inward 7.1 Å (Cζ-Cζ distance) away from its position in the apo form of the enzyme. The displaced ring of Phe214, in turn, causes the aromatic ring of Phe53 to swivel away by 7.8 Å (Cζ-Cζ distance) relative to its unbound position to avoid steric overlap. The aromatic ring of Phe53 is now too close to His161, which, in order to accommodate the Phe53 ring, moves into the free space in the vicinity of Cys90 reorienting itself such that the Nδ1 atom of the imidazole ring faces the Sγ atom of Cys90 with a distance of 3.9 Å between the two. Consequently, the enzyme adopts a catalytically competent, papain-like arrangement of the catalytic triad (Fig. 3B). The possibility that the change in the environment surrounding Cys90, brought about by its reaction with the VMe group, might have attracted His161 to its proximity causing the observed rearrangement cannot be ignored. This possibility could be addressed in the future by cocrystallizing UCHL1 with a short vinyl methyl ester such as Gly-VMe.

Fig. 3.

Binding of ubiquitin induces a conformational relay leading to the alignment of the catalytic triad. (A) Conformational changes of three side-chain rings of UCHL1S18Y induced by UbVMe binding. Superposition of apo UCHL1 (Orange) and UCHL1S18Y in UbVMe-bound form (Gray) showing the relative positions of Phe214, Phe53, His161, and Cys90. Electron density (contoured at 1.0σ) is shown as blue line. UbVMe is shown in green (without the VMe portion). For clarity, the following segments were removed: residues 31–39, 54–57, 147–160, and 207–212 of UCHL1S18Y and apo UCHL1. (B) Comparison of the active-site triad of UCHL1S18Y in UbVMe-bound form with that of apo UCHL1. Oxygen atoms are shown in red, nitrogen atoms in blue, sulfur in yellow, and carbon in gray (UCHL1S18Y bound) and orange (apo UCHL1).

In addition to the UCHL1S18Y–UbVMe complex, we have also solved the structures of the wild-type UCHL1–UbVMe (2.85 Å) and UCHL1I93M–UbVMe (2.80 Å). Comparison of structures of these complexes with their respective apo enzymes reveals that the relative movement of Phe214, Phe53, and His161 upon UbVMe binding is a common feature in all the UCHL1 variants studied (Fig. 4). To corroborate the crystallographic observations, we sought to carry out mutational analysis by substituting Phe214 with alanine (the F214A mutant). As shown in Fig. 5, the activity of the F214A mutant in ubiquitin aminomethyl coumarin (UbAMC) hydrolysis assay is significantly reduced relative to the wild-type UCHL1. This difference in the catalytic activity of the F214A mutant relative to the wild-type enzyme is not because of the difference in structures of these proteins; the overall three-dimensional structure of the F214A mutant is nearly identical to that of the wild type as judged by inspection of the far-UV circular dichroism spectra (Fig. S3).

Fig. 4.

Concerted movement of Phe214, Phe53, and His161 side chains relative to the apo form is also observed in the crystal structures of the wild-type UCHL1 and its PD-associated variant I93M bound to UbVMe. Superposition of the structures of apo UCHL1 (Orange), UCHL1-UbVMe (Yellow), apo UCHL1I93M (magenta), and UCHL1I93M-UbVMe (Cyan) are shown in ribbon representations. UCHL1S18Y-UbVMe is also shown (Gray Ribbon) for comparison. UbVMe from the UCHL1S18Y complex is shown in green. For clarity, the following segments were removed from the structures: residues 54–56, 210–213, and 154–160 and the VMe moiety.

Fig. 5.

Comparison of the enzymatic activity of the wild-type UCHL1 and the F214A and C90S mutants. Reaction progress curves showing AMC released vs. time for the cleavage of Ub-AMC by UCHL1 (Cyan), UCHL1F214A (Purple), and the catalytically inactive mutant UCHL1C90S (Red). The substrate and enzyme concentrations for all reactions were 600 and 3 nM, respectively.

The loss of activity observed upon mutation of Phe214 could be attributed to the impairment of the concerted movements of phenylalanine residues or the disruption of binding at the distal site. To explore this further, we performed isothermal titration calorimetry (ITC) to estimate the binding affinity of ubiquitin toward the wild-type UCHL1 and the F214A mutant (Fig. 6 and Table S2). Analysis of the ITC data showed that UCHL1 binds ubiquitin with a dissociation constant (K_d) of 385 nM, whereas the F214A mutant binds with approximately 60-fold less affinity (K_d = 22 μM). This reduced affinity for ubiquitin displayed by the F214A mutant suggests that Phe214 is essential for distal-site substrate binding, and the loss of catalytic activity of the mutant may reflect its reduced affinity toward the substrate. This observation corresponds to the previous biochemical analysis by Luchansky et al. demonstrating that Leu8 of ubiquitin is critically required for its interaction with UCHL1 (20).

Fig. 6.

Binding isotherm of the titration of UCHL1 and UCHL1F214A with ubiquitin. (A) Binding of ubiquitin (100 μM) to UCHL1 (10 μM). (B) Binding of ubiquitin (576 μM) to UCHL1F214A (56 μM). The top panel for each figure shows the raw data, and the lower panel shows the integrated heat data as enthalpy as a function of molar ratio of ligand to protein. The solid line in the bottom panel represents the best curve fit to the data by using a one-site binding model.

Discussion

UCHL1 is a neuron-specific DUB that has been linked to neurodegenerative diseases and cancer. In particular, the two point mutations I93M and S18Y have been linked to the early onset of and protection from PD, respectively (4–6). Yet the normal function of UCHL1 and how the activity of this cysteine protease is regulated remain elusive. The crystal structure of UCHL1 in apo form revealed that the active-site triad is misaligned for catalysis. In traditional cysteine proteases, the nucleophilicity of the catalytic cysteine is enhanced by the abstraction of the proton from the thiol group by a proximal histidine, the general base catalytic residue (22). Such a relationship between the nucleophilic cysteine and the general base histidine would require them to be within ∼4 Å of each other, allowing effective hydrogen-bonding interactions (16). In the apo UCHL1 structure, this distance (between Cys90 and His161) is 7.7 Å, far greater than expected for any productive interaction. To understand how this enzyme functions as a cysteine protease, we have crystallized and solved the structures of the wild-type UCHL1 and its two PD-associated variants, UCHL1S18Y and UCHL1I93M, bound to the suicide substrate UbVMe. The structures of these complexes reveal a previously unanticipated feature of the enzyme, a substrate-mediated distal-site effect leading to the transition of the active site of the enzyme from an unproductive to its productive form.

The binding of the suicide substrate reveals two dominant substrate binding sites on the enzyme: the active-site cleft and a distal site 17 Å away from the active site. Intermolecular interactions at the active-site cleft ensure that the scissile peptide bond is positioned appropriately relative to the catalytic cysteine. The active-site cross-over loop may help select the appropriate substrate—a yet to be identified ubiquitinated species with small-molecule or short/unfolded polypeptide as the C-terminal extension of ubiquitin. However, no useful chemistry can occur without the presence of the general base histidine, properly oriented, in the proximity of the catalytic thiol and the scissile peptide bond. The task of aligning the general base relative to the reactive moieties appears to be left to the interactions at the distal site. A conformational relay, presumably starting at a surface-exposed phenylalanine in the distal site, is triggered by the binding of the N-terminal β-hairpin of ubiquitin resulting in the placement of the active-site histidine in the correct location.

Ligand-dependent alignment of an active-site Cys-His pair has been observed previously in the case of two cysteine proteases: USP7, a DUB in the USP family (18), and μ-calpain, a calcium-activated cysteine protease (19). USP7 is a much larger protein with a very different binding site for ubiquitin. Crystal structures of USP7 in free and ubiquitin-aldehyde (Ubal)-bound form revealed that the C-terminal segment of Ubal induces significant backbone conformational changes in the vicinity of the active site leading to alignment of the catalytic triad (18). In μ-calpain, calcium binding changes the relative orientation of the two domains of the protease leading to the alignment of active-site residues (19). In contrast, ubiquitin-induced conformational changes in UCHL1 appear to originate at a distal site and involve swiveling motion of the side-chain rings of three residues that work in a concerted manner to avoid steric clash, with relatively little backbone reorganization, consistent with the observation that the UCHL1 backbone is extensively knotted (23).

The pair of phenylalanine residues (Phe53 and Phe214) involved in the active-site alignment in UCHL1 is highly conserved in different vertebrates (Fig. S4) as well as in other UCHs (Fig. S5). Although the general mode of ubiquitin binding observed in the UCHL1–UbVMe complex, including the interactions involving ubiquitin’s Leu8-Thr9 at the distal site, is very similar to that seen in UCHL3–UbVMe (12) and Yuh1–Ubal complexes (17), the concerted movement of phenylalanines leading to active-site alignment is observed only in the case of UCHL1. Comparison of the structures of UCHL3 in apo (16) and UbVMe-bound form (12) reveals that the orientation and the position of the side-chain rings of the phenylalanine pair and the general base histidine are nearly identical in both forms, overlapping with those observed in the UbVMe-bound form of UCHL1 (Fig. S6). This observation suggests that there is no distal-site conformational relay in UCHL3, consistent with a preorganized active-site triad in this enzyme (16). Considering that UCHL1 and UCHL3 share a high level of sequence and structural similarity and their ubiquitin-binding interfaces are nearly identical (this study and ref. 12), the difference in the way they are regulated is remarkable. Although the biological relevance of this is unclear at the moment, this difference may translate into a different function and could explain their distinct tissue specificity (unlike UCHL1, UCHL3 is expressed ubiquitously in all tissues) (3, 24).

Compared to other UCHs such as UCHL3, whose active-site triad is prearranged in a catalytically productive conformation, UCHL1’s active site is misaligned. Therefore, ubiquitin-mediated activation of UCHL1 appears to be necessary for the catalytic activity of this enzyme. The structural studies presented herein provide a mechanism to this end. However, whether this confers an advantage to UCHL1 over the selective mechanism of other ubiquitin hydrolases needs to be investigated in the future.

UCHL1 is abundantly expressed in neurons and has been known to undergo several types of posttranslational modification, such as farnesylation and monoubiqutination, that may regulate its localization and enzymatic activity (25, 26). Our crystallographic results suggest that the activity of UCHL1 is intrinsically regulated by its own substrate and provide a structural basis for the enzyme’s specificity for ubiquitin. Interestingly, these studies also indicate that the PD-associated mutants of UCHL1 (I93M and S18Y), like the wild-type enzyme, can adopt productive catalytic triads when bound to ubiquitin. The biochemical basis of how these mutations are linked to PD is yet to be elucidated. By showing that the mutants can have a similar productive triad, and hence similar catalytic properties as the wild type, this study raises an intriguing question as to what could then be the difference between the mutants and the wild-type protein.

The true in vivo function of UCHL1 remains unclear. Although it was proposed to be a deubiquitinating enzyme, primarily because of its sequence similarity with UCHL3, a widely accepted deubiquitinating enzyme, several lines of evidence suggest that UCHL1 may have other alternative functions. For example, a dimerization-dependent ligase activity was previously proposed for UCHL1 (27). More recently, a report suggests that UCHL1 can inhibit microtubule formation in a ubiquitination-dependent manner. The authors of this report suggest that UCHL1 may increase ubiquitination (and hence behave as a ligase) of microtubule components (28). In light of these studies, and others that indicate that UCHL1 may have functions independent of the ubiquitin-proteasome system (29, 30) and the misaligned active site observed in the structure of the apo form of the protein, the demonstration that UCHL1 can adopt a productive conformation as a hydrolase when bound to ubiquitin assumes particular significance.

Although UCHL1 is normally expressed in the brain, abnormal overexpression of this enzyme has been found in many forms of cancer, including lung and colorectal cancer (8, 9). Identification of high-affinity small-molecule inhibitors as pharmacological agents is highly desirable to elucidate the pathophysiological roles of UCHL1. The crystallographic studies along with the mutational data presented here suggest that the perturbation of enzyme–substrate interactions at the distal site of UCHL1 could be detrimental to its enzymatic activity. Targeting the distal site with small-molecule binders that can perturb these interactions could be envisioned as an attractive strategy for UCHL1 inhibition.

Experimental Procedures

Cloning, Expression, and Purification.

UCHL1S18Y was subcloned from a pcDNA-UCHL1S18Y vector into a pGex-6P-1 vector (GE Biosciences) by using standard cloning protocols. The resulting N-terminally fused glutathione S-transferase (GST)-tagged UCHL1S18Y protein was expressed in Escherichia coli Rosetta cells (Novagen) and purified with a glutathione-Sepharose column (GE Biosciences) following manufacturer’s instructions. For expression and purification of other variants of UCHL1 described in this paper, please see SI Methods.

The UCHL1S18Y–UbVMe complex was prepared according to the method described by Misaghi et al. with slight modifications (12). In brief, 1 M excess of UbVMe solution (pH adjusted to 8.0 by adding 1 M NaHCO₃) was added to the UCHL1S18Y protein solution (in 50 mM Tris.HCl, 150 mM NaCl, pH 7.4). The mixture was incubated for 5 h at room temperature. The formation of the complex was verified by an 8-kDa shift of the UCHL1 protein band in an SDS-PAGE gel. The UCHL1S18Y–UbVMe complex was purified by size exclusion chromatography with a Superdex S75 column (GE Biosciences). UbVMe used in the preparation of the complex was obtained by intein-mediated semisynthesis using a previously published procedure (31, 32).

Crystallization and Structure Determination.

The UCHL1S18Y–UbVMe complex was concentrated to ∼25 mg/mL in a solution of 50 mM Tris.HCl (pH 7.4), 150 mM NaCl, and 10 mM DTT. Crystals were grown at room temperature by the hanging drop vapor diffusion method from a solution that contained 2.4 M ammonium sulfate and 0.1 M bicine (pH 9.0). Crystals grew over 2 months to a final dimension of approximately 200 × 200 × 100 μm. Crystals were briefly soaked in the cryoprotectant solution (2.9 M sodium malonate, pH 7.5) and plunged into liquid nitrogen for flash cooling. X-ray diffraction data (up to 2.4 Å) were collected at 100 K on a Mar300 CCD detector (Mar USA) at the beam line 23-ID-D at the Advanced Photon Source of Argonne National Laboratory. The data were processed with the program HKL2000 (33). The crystals belong to the space group R32, with unit cell dimensions a = 87.3 Å, b = 87.3 Å, c = 193.5 Å, α = 90.0°, β = 90.0°, and γ = 120.0°, with one UCHL1S18Y–UbVMe complex per asymmetric unit.

The structure was determined by molecular replacement employing Molrep (34) using the wild-type UCHL1 and human ubiquitin (residues 1–75) as search models. Cross-rotation and translational searches identified a single copy of the complex in the asymmetric unit. Rigid-body refinement of this model followed by restrained refinement brought the crystallographic R factor to 31.7% and R_free to 40.5%. The electron density map at this stage showed clear density corresponding to the side chain of Tyr at position 18 on UCHL1S18Y and the VMe part of UbVMe. Subsequent refinement was performed with Refmac (13) and model building with Coot (14). The final crystallographic R factor (R_cryst) and R_free are 20.9% and 25.6%, respectively (Table S1). The model contains the complete 223-residue UCHL1S18Y chain, 1–75 residues of ubiquitin, the GlyVMe moiety (modeled as 4-amino methyl butanoate), and 51 ordered solvent molecules. More than 98% of the nonglycine residues are placed within the most favorable and additionally allowed areas of Ramachandran plot, and less than 0.5% are located in the disallowed areas, as defined within the program PROCHECK (15). The first five N-terminal residues carried over from the GST-tagged cloning vector were disordered and, therefore, were not included in the model. Graphical analysis was done with the program PYMOL (DeLano Scientific).

Enzymatic Activity Assay.

Stock solutions of wild-type UCHL1 and the F214A and C90S mutants were diluted into the reaction buffer (50 mM Tris.HCl, pH 7.4, 1 mM DTT, and 1 mM EDTA) in individual wells of a 96-well plate to the final concentration of 3 nM. UbAMC was added to these wells to yield a final concentration of 600 nM to initiate the enzymatic reaction. The rate of AMC cleavage was monitored at 25 °C by a TECAN Genios microplate spectrofluorometer with excitation at 380 nm and emission at 465 nm.

ITC.

ITC was carried out by using a VP-ITC Microcal calorimeter (MicroCal) at 24 °C. Ubiquitin from bovine erythrocytes was purchased from Sigma-Aldrich. Amino acid sequences between human and bovine ubiquitin are identical. Samples of UCHL1, UCHL1F214A, and ubiquitin were extensively dialyzed against the buffer 50 mM Tris-HCl, pH 7.6 (buffer A) and then degassed to remove dissolved air. Titrations consisted of 10-μL injections of ubiquitin into the sample cell containing the protein, at time intervals of 4 min to ensure each peak returned to baseline. Each UCHL1 sample was followed by a background titration of an equal volume of ubiquitin being titrated into a sample cell containing buffer A to account for the heat of dilution, which was subtracted from the UCHL1–ubiquitin data. All data were analyzed by using the program Origin, version 7.0, included with the system. The data were fitted with a one-site binding model (one molecule of ubiquitin binding to one molecule of UCHL1). Binding constants and thermodynamic parameters are given in Table S2.