Crystal structure of the catalytic domain of UCHL5, a proteasome-associated human deubiquitinating enzyme, reveals an unproductive form of the enzyme

Abstract

Ubiquitin carboxy-terminal hydrolase L5 (UCHL5) is a proteasome-associated deubiquitinating enzyme, which, along with RPN11 and USP14, is known to carry out deubiquitination on proteasome. As a member of UCH family, UCHL5 is unusual because, unlike UCHL1 and UCHL3, it can process polyubiquitin chain. However, it does so only when it is bound to the proteasome; in its free form, it is capable of releasing only relatively small leaving groups from the C-terminus of ubiquitin. Such a behavior might suggest at least two catalytically distinct forms of the enzyme, an apo form incapable of chain processing activity, and a proteasome-induced activated form capable of cleaving polyubiquitin chain. Through the crystal structure analysis of two truncated constructs representing the catalytic domain (UCH domain) of this enzyme, we are able to visualize a state of this enzyme that we interpret as its inactive form, because the catalytic cysteine appears to be in an unproductive orientation. While this work was in progress, the structure of a different construct representing the UCH domain was reported; however, in that work the structure reported was that of an inactive mutant (catalytic Cys to Ala, Nishio K, Kim SW, Kawai K, Mizushima T, Yamane T, Hamazaki J, Murata S, Tanaka K and Morimoto Y (2009) Biochem Biophys Res Commun 390, 855-860), which precluded the observation that we are reporting here. Additionally, our structures reveal conformationally dynamic parts of the enzyme that may play a role in the structural transition to the more active form.

Introduction

Much of the regulated proteolysis in eukaryotic cells is mediated by the ubiquitin-proteasome system. The degradation of a polypeptide by the eukaryotic proteasome generally requires it to be modified with covalently attached polyubiquitin chain before it is brought to the proteasome for degradation [1–3]. The polyubiquitin chain on a protein marked for degradation is essential for its recognition by the proteasome, and consequently, is responsible for the selectivity of the proteasome as a protein degradation machinery. Following its recognition by the proteasome via interaction of proteasome-resident ubiquitin receptors with the attached polyubiquitin chain, the substrate is unfolded by proteasome-resident ATPases and is subsequently translocated into the hydrolytic core of the proteasome to be degraded [1]. However, ubiquitin itself is spared from degradation and is recovered as intact ubiquitin and/or ubiquitin chain, perhaps to increase economy by its recycling. Substrate deubiquitination on the proteasome is achieved by three distinct deubiquitinating enzymes (DUBs) associated with the regulatory particle (RP, also known as the 19S subunit) of proteasome, RPN11, USP14 and UCHL5 (also known as UCH37) [4–13]. RPN11, a Zn-dependent metalloprotease, belongs to the JAMM-family of DUBs [10, 14], whereas USP14 and UCHL5 are cysteine proteases belonging to the USP and UCH family of DUBs, respectively [15–18]. RPN11 appears to cleave at the base of the ubiquitin chain, where it is linked to the substrate, whereas USP14 and UCHL5 are known to mediate stepwise removal of ubiquitin from the substrate by disassembling the chain from its distal end [6].

The proteasome-associated DUB activity assigned to UCHL5 is particularly intriguing. In general, the UCH family of DUBs have been known to be incapable processing polyubiquitin chains, based on previous biochemical and structural studies on UCHL1 and UCHL3 which strongly suggest that the active-site crossover loop, a feature present in all UCH enzymes, limits the size of P′ side of the substrate to either a small-molecule or a short peptide leaving group [19–25]. In fact, in its apo form (a form defined as when UCHL5 is not bound to its activating protein partners on proteasome), UCHL5 behaves in a manner similar to its smaller counterparts, UCHL1 and UCHL3: it can release aminomethyl coumarin (AMC) group from ubiquitin-AMC (Ub-AMC), but does not cleave isopeptide-linked diubiquitin [4], a model substrate for polyubiquitin chain. However, binding to proteasome via its interaction with RPN13 can activate the chain-processing activity of UCHL5 [4]. Consistent with this behavior, one would expect UCHL5 to exist in at least two forms, one incapable of chain processing activity, and an activated form, induced upon binding to proteasome-resident RPN13, capable of processing of polyubiquitin chain.

Through the crystal structure analysis of two truncated constructs representing the catalytic domain (UCH domain) of this enzyme, we are able to visualize a state of this enzyme that we interpret as its inactive form, because the catalytic cysteine appears to be in an unproductive orientation. Comparison of these structures with that of previously published structures of a different construct representing the UCH domain (reference 28) and the full-length protein (deposited in Protein Data Bank as entry 3IHR, Burgie et al. 10.2210/pdb3ihr/pdb) reveals variable conformations of the crossover loop, changing from completely disordered to partially ordered states. This indicates that the loop can be substantially dynamic, capable adopting multiple conformational states, some of which could regulate the access of substrate into the active site. Indeed, in one of our structures, we are able to visualize a segment of the crossover loop blocking the active-site cleft, though we cannot rule out the influence of packing forces in the crystal on this observed conformation.

Results

UCHL5N240 (the first 240 amino acids of UCHL5), one of the constructs we have crystallized, comprises the UCH domain (defined, based on sequence alignment with other UCH enzymes, as the first 228-aminoacid-containing portion of UCHL5) plus an additional 12-residue segment from the C-terminal tail of full-length UCHL5. This construct displays almost the same level of enzymatic activity in Ub-AMC hydrolysis assay as the full-length protein (Fig. 1 and Table 1) suggesting that the C-terminal tail of UCHL5 does not a play a significant role in the catalysis of Ub-AMC hydrolysis. Additionally, both the full-length protein and the truncated catalytic domain were inactive toward a diubiquitin substrate (Lys48-linked) (data not shown and reference 4). Together, these data suggest that the structure of the catalytic site in the crystallized domain is a true representation of that in the full-length protein, at least in its apo form.

Figure 1.

Progress curve of hydrolysis of ubiquitin-AMC (Ub-AMC) catalyzed by UCHL5 constructs: full-length UCHL5 (filled triangle), UCHL5N240 (open circle), UCHL5N240^C88S (Cys88 to Ser mutant, open square), UCHL5N240^H164N (closed diamond), UCHL5N240^D179N (open triangle).

Table 1.

Kinetic parameters of UCH enzymes

Enzyme	KM (µM)	kcat(s−1)	kcat/KM (× 105 M−1 s−1)
UCHL1	0.03	0.029	8.87
	0.04a	0.020a	5.10a
UCHL3	0.02a	5.9a	2773a
UCHL5N240	10.7	8.7	8.1
UCHL5	n.d*	n.d*	12.0*
GST-UCHL5	12.70b	0.28b	0.22b

^aReference [41];

^bReference [4]. Actual values reported in this reference were for V_max (33.4 nM.min⁻¹) and V_max/K_M (2.6 × 10⁻³ min⁻¹) (enzyme concentration used was 2 nM);

^*The values of k_cat and K_M could not be determined individually, but the ratio k_cat/K_M was, from the slope of the line that resulted when initial velocity was plotted with substrate concentration. We could not reach high enough substrate concentration to get to the plateau region of the Michaelis-Menten plot for determining both the parameters separately.

The crystal structure of selenomethionine-labeled UCHL5N228^C88S (the Cys88 to Ser mutant of UCHL5N228, the first 228 amino acids of UCHL5) was solved at 2.0 Å by molecular replacement (MR) using the structure of UCHL3 (PDB ID: 1UCH) as the search model [19]. The structure was refined to a final R_crys value of 17.5 % and R_free value of 22.2 %. More than 98 % of non-glycine 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 [26] (Table 2). The structure of UCHL5N240 was solved by MR employing MOLREP [27] using the structure of UCHL5N228^C88S as the search model. The structure was refined to a final R_crys value of 20.4 % and R_free value of 25.8 %, with good stereochemistry (Table 2). The four copies of the protein in the asymmetric unit are very similar (Fig S1) except for the crossover loop (defined as amino acids 143–162), which varied in the extent to which it was ordered in different subunits of the asymmetric unit. We will therefore focus our discussion on only one of the subunits.

Table 2.

Data collection and refinement statistics

	UCHL5N240	UCHL5N228C88S
Data collection
Space group	P21	P21
Cell dimensions
a, b, c (Å)	87.4, 41.7, 138.7	45.4, 99.0, 48.1
αβγ(°)	90.0, 89.9, 90.0	90, 96.4, 90.0
Resolution (Å)	41.7–2.4 (2.49–2.40)*	41.0–2.0 (2.07–2.00)
Rmerge (%)	8.4 (64.8)	12.2 (46.2)
I / σI	14.7 (1.6)	12.9 (2.8)
Completeness (%)	99.1 (92.1)	96.8 (93.1)
Redundancy	3.5 (2.6)	6.0 (3.2)
Refinement
Resolution (Å)	2.40	2.00
No. Reflections	37595	27355
Rwork (%) / Rfree (%)	20.4 / 25.8	17.5 / 22.2
No. Atoms
Protein	7191	3347
Ion	3	1
Water	246	282
R.m.s. deviations
Bond lengths (Å)	0.008	0.003
Bond angles (°)	1.045	0.658
Ramachandran Plot
Favorably allowed (%)	93.2	94.9
Additionally allowed (%)	6.4	5.1
Disallowed (%)	0.4	0.0

^*Values in parentheses are for the highest-resolution shell

Overall Structure

Like other structurally characterized members of the UCH family, such as UCHL1 and UCHL3, the three-dimensional structure of the UCH domain of UCHL5 is organized around a six-stranded antiparallel β-sheet sandwiched by α-helices on either side to form an α-β-α fold (Fig. 2). The overall arrangement of secondary structural elements in UCHL5N240 gives an appearance of a bilobal architecture, with one lobe consisting of five α helices (α1, α3, α4, α5, and α6), and the other consisting of two helices (α2 and α7) and the 6-stranded β sheet (Fig. 2). The 12-residue C-terminal extension after the UCH domain, α8, projects into solvent running in the same direction as the terminal β-strand, β6, of the UCH domain. The active site, harboring the nucleophile Cys88 (on α3), the general base His164 (on β3) and Asp179 (on loop L11), is located in a cleft between the two lobes of the molecule. The so-called active-site crossover loop, residues 143 to 162, connects α6 from one lobe to β-strand 3 (β3) on the other and is only partially visualized in UCHL5N240 structure. The conformation of this loop, the parts that are visible, shows substantial variability among the structures discussed herein, the two structures solved by us, UCHL5N240 and UCHL5N228^C88S, a previously published structure of a different UCH-domain construct, UCHL5N225^C88A [28], (it was referred to as UCHL5N in reference [28], but we choose to refer to it as UCHL5N225^C88A here to distinguish the different N-terminal constructs of UCHL5 discussed in this manuscript) and full-length UCHL5, indicating that the loop is conformationally flexible, capable of adopting multiple conformations that may co-exist in equilibrium in solution. As expected from sequence similarity (Fig. S2), the overall structure of the UCH domain of UCHL5 is similar to UCHL1, UCHL3 and the yeast ubiquitin hydrolase Yuh1, with a major difference seen in the conformation of the 20-amino acid polypeptide segment following β2 (L5 in Fig. 2), as noted by Nishio et al and confirmed by the two structures presented here (for a superposition of the UCH domain of UCHL5 with UCHL1, UCHL3 and Yuh1 please see Fig. S3).

Figure 2.

Ribbon representation of the X-ray structure of the catalytic domain of UCHL5 (UCHL5N240). Secondary structures are labeled. The catalytic triad residues and the oxyanion-stabilizing side chain are shown in stick representation. (B) A different view of the structure of UCHL5N240 obtained by rotating the picture in A by 90 degree around the axis indicated here.

Arrangement of the active-site residues

Like other members of the UCH family, UCHL5 is a cysteine protease with a catalytic triad consisting of a cysteine (Cys88), a histidine (His164), and an aspartate (Asp 179). Mutation of these residues results in a significant loss of enzymatic activity (Fig. 1). Figure 3 shows the arrangement of active-site residues in UCHL5N240. The distance between the members of the triad is consistent with a canonical arrangement of the triad [29]; the Nδ1 atom of the imidazole ring of His164 is within hydrogen-bonding distance from the thiol group of Cys88 (the Nδ1-Sγ distance is 3.5 Å) and His164 is within hydrogen-bonding distance from Asp179 (Nε2-Oδ1 distance of 2.6 Å). However, the orientation of the side-chain of the catalytic cysteine with respect to the general base histidine (His164) is quite different from that seen in the active forms of other structurally characterized UCH enzymes, which have their cysteine Sγ atoms pointing toward the open space in the catalytic cleft (Fig. 3B and 3C). In contrast, the catalytic cysteine in UCHL5N240 has an opposite orientation pointing toward the interior of the protein (Fig. 3B). The Sγ atom in this opposite orientation is held tightly by a network of hydrogen-bonding interactions involving the Nδ1 atom of His164, the backbone NH group of Phe165 (N-Sγ distance of 3.5 Å) and the side-chain carbonyl oxygen atom of Gln82, the oxyanion-stabilizing side chain (Sγ-Oδ distance of 3.4 Å). The side-chain carbonyl oxygen of Gln82 is in short contact with the Cε of His161, which can be interpreted as a CH---O hydrogen bond [30–32], an observation consistent with a previous finding of a CH---O contact in the active site of serine proteases involving the general base histidine and a nearby backbone carbonyl oxygen [33]. An identical orientation of the catalytic cysteine and the equivalent serine residue, stabilized by the same network of hydrogen-bonding interactions, is also observed in the full-length UCHL5 and in UCHL5N228^C88S structures, respectively (Fig. 4B). In UCHL5N225^C88A, since the catalytic cysteine was mutated to alanine [28], we are unable to determine if such an orientation exists for this structure.

Figure 3.

Unproductive form of active-site triad in UCHL5. (A) Active-site residues in UCHL5240N (shown as sticks). Carbon atoms are shown in gray, nitrogen in blue, oxygen in red and sulfur in orange. Electron density (2Fo-Fc map contoured at 1.0 σ) is shown in purple line. (B) Superposition of active-site residues in UCH enzymes. Structures overlaid are that of full-length UCHL5 (gray sticks), UCHL5N240 (gray sticks), UCHL5N228C88S (gray sticks), UCHL3 (cyan sticks), UCHL1 (magenta sticks) and Yuh1 (yellow sticks). Distances indicated correspond to UCHL5N240. (C) Superposition of active-site residues of UCHL5N240 with 15 other cysteine proteases from Merops database. Distances indicated correspond to UCHL5N240. In addition to UCH enzymes, the other cysteine proteases used in the superposition are papain, actinidain, chymopapain, caricain, cathepsinF, cathepsinV, cathepsinX, zingipain, USP7, USP2, USP8 and ataxin-3. (D) Interactions stabilizing the non-canonical orientation of Cys88 side chain observed in the crystal structure of UCHL5N240.

Figure 4.

Conformational dynamics in UCHL5. (A) Comparison of structures of UCHL5 constructs. Ribbon representation of backbone superposition of UCHL5N240 (gray), UCHL5N228^C88S (cyan), UCHL5N225^C88A (yellow) and full-length UCHL5 (magenta). Ubiquitin (Ub, green ribbon) is modeled into its binding site on UCHL5. (Inset) Expanded view of helix 6 (α6) as observed in structures of different UCHL5 constructs. (B) Steric clash (indicated as a box) between C-terminus of ubiquitin and a part of the crossover loop folding into a 3₁₀ helix as seen in the X-ray structure of UCHL5N228^C88S. Ubiquitin (shown in green ribbon) is placed in its binding site by modeling. (C) Interactions of side chains (shown as sticks) on the 3₁₀ helix (shown in B clashing with the C-terminus of ubiquitin) with nearby residues in UCHL5N228^C88S. (D) Superposition of X-ray structures of UCHL5N240 (gray ribbon, except α8, which is shown in orange) and full-length UCHL5 (magenta ribbon, except for α8, which is shown in cyan). The arrow indicates the relative displacement (of approximately 45 degree) of α8 between the two structures.

The observed orientation of the catalytic cysteine appears to be unique to UCHL5. Figure 3C shows the comparison of the orientation of four catalytic residues of UCHL5 with 15 other cysteine proteases from the Merops database [34]. The orientation of the side chain of Cys88, both in UCHL5N240 as well as that in full-length UCHL5, is unique and is opposite to all others (Fig. 3C). We propose that this orientation represents an unproductive form of the catalytic thiol because it is facing away from the open space in the active-site cleft, pointing into the interior of the protein. In such an orientation, the thiol group can be regarded as a partially buried nucleophile. Thus, despite having canonical distance between the catalytic residues, the active-site triad of UCHL5 may still be misaligned for catalysis.

Conformational flexibility of the crossover loop

As noted before, we have crystallized two different forms of the catalytic domain of UCHL5, UCHL5N240 and UCHL5N228^C88S. These structures can now be compared with a previously reported structure of the Cys88Ala mutant of another construct representing the UCH domain, UCHL5N225^C88A [28], and full-length UCHL5 (deposited in Protein Data Bank as entry 3IHR, Burgie et al). In the UCH domain, except for the crossover loop (residues 143–162), the secondary structures and the way they are arranged are very similar in different structures (average rmsd of Cα atoms of 0.73 Å over 203 matching Cα atoms, calculated using the program Superpose in CCP4) (Fig. 4A), suggesting that these structural units are conformationally rigid. The conformation of the crossover loop, in contrast, shows significant variability. In UCHL5N225^C88A and full-length UCHL5, this loop is mostly disordered and is not visualized in the crystal structures. In UCHL5N240, only the N-terminal part of the loop (residues 143–149) is ordered, folded into almost two complete turns of an α-helix, which, together with residues 134 to 142, form α6 (Fig. 2 and Fig. 4A). In contrast, most of the same part of the loop in UCHL5N228^C88S is disordered and therefore is not visible (from residues 145–158), whereas the C-terminal segment of the loop in this structure is folded into a two-turn 3₁₀ helix (residues 158–163) (Fig. 4B). This 3₁₀ helix presents three contiguous acidic side-chains for ionic and hydrogen-bonding interactions with nearby residues: Glu159, Glu160 and Asp161. Glu159 interacts with Arg217, Glu160 with Ser13 and Asn219 (these interactions are not present in the other subunit in the asymmetric unit. Instead, Glu160 is interacting with Arg217 in that subunit), and Asp161 with the backbone NH of Ser13 and with Arg145 from the other end of the crossover loop (Fig. 4C). These interactions appear to hold the 3₁₀ helix in such an orientation that it sits directly over the active-site cleft blocking the C-terminus of ubiquitin from approaching the catalytic cysteine, as observed when ubiquitin was modeled into its binding site on UCHL5 (Fig. 4B).

Apart from the crossover loop, the helix immediately after the UCH domain, α8, also appears to be dynamic (Fig. 4D). When the structure of UCHL5N240 is superimposed on that of full-length UCHL5, α8 appears to be rotated by approximately 45 degree between the two structures. Since α8 in UCHL5N240 is involved in intermolecular contacts in the crystal, it is possible that the observed orientation could be due to crystallographic packing forces. However, the possibility that α8 can adopt both these orientations in solution cannot be ruled out.

Discussion

From the X-ray structural analysis of the catalytic constructs we have presented herein, it appears that the crystallized forms represent an unproductive form of the active site of UCHL5, with the catalytic cysteine adopting a non-canonical orientation in which the thiol group faces a direction opposite to the catalytic cleft and is partially buried. Unproductive active sites in their free form appear to be a common feature of most structurally characterized DUBs, particularly UCH enzymes. In UCHL1, the catalytic histidine is situated almost 8Å away from the catalytic cysteine, consistent with an unproductive arrangement [23]. In UCHL3, residues Leu10 and Glu11 in the N-terminal end are too close to the active site cysteine occluding substrate access [19]. Ubiquitin binding in both cases transforms the active site from unproductive to productive form, with the catalytic histidine moving in closer in UCHL1-UbVME structure [24] and the N-terminal peptide moving away in UCHL3-UbVME structure [20]. A similar mechanism may exist for UCHL5. Upon ubiquitin binding, the catalytic cysteine may adopt an orientation seen in other productive cysteine proteases (Fig. 4C). A mechanism of this type will likely confer selectivity on UCHL5 and other UCH enzymes, such that these cysteine hydrolases behave as ubiquitin hydrolases rather than non-specific proteases.

As expected, the crossover loop is mobile in UCHL5; it showed the greatest variability between structures of different constructs of the protein. Crystal structure determination of two UCH-domain-containing constructs presented here has allowed visualization of different conformational states of the crossover loop. In one observed conformation, the C-terminal part of the loop folds into a two-turn 3₁₀ helix and is situated in a position where it would cause steric clash with the C terminus of ubiquitin (Fig. 4B). In another, the N-terminal part of the loop is folded into a two-turn α-helix. Although the possibility that these conformations are a result of crystallographic packing forces cannot be ruled out, it is likely that the observed conformations represent some of the actual conformational states the loop can sample in solution. We speculate that the crossover loop in UCHL5 may exist in at least two conformational states: one in which the loop is fully stretched, corresponding to an open active-site cleft, and in the other, the loop is restricted in length due to coiling up of the two ends into helical conformations, corresponding to a closed active site. Because of the contribution of such restrictive conformations, the average picture of the active-site cleft on UCHL5 may look somewhat obstructed, translating into reduced affinity for the substrate. This could be one of the factors behind a much higher value of K_M with Ub-AMC as the substrate, ~10 µM for UCHL5 in contrast to ~30 nm for UCHL1 and UCHL3 (Table 1). (In UCHL3, there too is occlusion imposed by the two residues from the N-terminus, Leu 10 and Glu 11, but it is not as extensive as in UCHL5.)

Although ubiquitin binding may be sufficient to transform the active site of UCHL5 into a more productive form for Ub-AMC hydrolysis to occur, additional changes are required to process a diubiquitin substrate. It is tempting to speculate that its C-terminal tail may play a role in this. Like the crossover loop, the tail may sample different conformational states as well. In one conformational state, perhaps represented by the crystal structure of the full-length construct, the tail may be in close contact with the crossover loop, as can be seen when UCHLN240 is superimposed with full-length UCHL5 (Fig. 4A). In another putative conformation, α8 is displaced from its position seen in full-length UCHL5 structure to the position seen in the crystal structure of UCHL5N240 (Fig. 4D). Since α8 is a part of a helix-turn-helix arrangement with α9 in full-length UCHL5, it is possible that the displacement of α8 would move the entire C-terminal tail as a rigid body from its location in the full-length structure to a direction indicated in Figure 4D, and along with the tail, the crossover loop may be pulled off from its obstructing position. Future studies aimed at co-crystallizing UCHL5 with ubiquitin and its proteasome partner RPN13 will shed light on the mechanism of UCHL5’s activation.

Materials and methods or Experimental procedures

Cloning Expression and Purification of Catalytic Domain of UCHL5

UCHL5N240 and UCHL5N228 were sub-cloned from full-length UCHL5 DNA (UCHL5-pGEX-6P1, kindly gifted by Sarah Luchansky, Harvard Medical School) into pGEX-6P1 vector using standard cloning protocol. UCHL5N228^C88S mutant was generated by PCR mutagenesis using QuickChange site-directed mutagenesis kit (Stratagene, Santa Clara, CA). Plasmids containing these genes were transformed into E. coli Rosetta cells (Novagen, Gibbs Town, NJ). The cells were grown in LB medium containing 100µg/ml ampicillin and 50µg/ml chloramphenicol at 37°C. The cells were induced at OD₆₀₀ ~ 0.6 with 1mM isopropylβ-D-thiogalactoside (IPTG) and induction of protein expression was continued at 18°C overnight with shaking. The cells were then harvested by centrifugation at 7000g for 15min at 4°C. The pellet was resuspended in PBS (phosphate buffer saline) buffer containing 400mM KCl, lysed using French press and were spun at a centrifugal force corresponding to 22,000g for 1h at 4°C. The supernatant was then loaded onto a glutathione-sepharose column, which was previously equilibrated with PBS containing 400mM KCl. The column was then washed with the same buffer and the GST (glutathione S-transferase)-fused protein was eluted with the elution buffer (50mM Tris.HCl, pH 8.0, 20mM reduced glutathione and 500mM KCl). The eluted protein was subsequently dialyzed against PBS containing 400mM KCl and 5mM DTT (dithiothreitol) overnight at 4°C. During dialysis, the GST tag was cleaved from the fusion protein using PreScission protease (GE Biosciences, Piscataway, NJ). After dialysis, the protein was separated from the GST tag and PreScission protease by passing the solution through a glutathione-sepharose column. The subtracted protein was then concentrated and loaded onto Superdex 75 gel filtration column and eluted with the buffer 50mM Tris.HCl, pH 7.4, 150mM NaCl and 5mM DTT. The purity of the protein was checked by sodium dodecyl sulfate-polyacrylamide gel electrophoresis. The purified protein was concentrated using Amicon centrifugal concentrator (Millipore). GST-fused selenomethionine-labeled UCHL5N228^C88S was prepared by growing cells in M9 minimal medium using standard protocol and was purified as described above, with the only difference that the buffers used in the purification included 10 mM DTT.

Crystallization, Data collection and structure determination

Both UCHL5N240 and selenomethionine-labeled UCHL5N228^C88S were concentrated to ~20mg/ml in a solution of 50mM Tris.HCl, pH 7.4, 150mM NaCl and 5mM DTT (for the latter protein, 10 mM DTT was used). Crystals of UCHL5N240 were grown at room temperature by hanging drop vapor diffusion method from a solution containing 1.6M ammonium sulfate, 0.1M Tris.HCl, pH 8.5. The crystals were briefly soaked in 20% glycerol as the cryoprotectant and plunged into liquid nitrogen for flash cooling. X-ray diffraction data (up to 2.4Å) were collected at 100K on Mar 300 CCD detector (Mar USA) at the beam line 23ID-B at the Advanced Photon Source (Argonne National Laboratory, IL). The data were processed with the HKL2000 program [35]. The crystal belongs to the space group P2₁ with four molecules per asymmetric unit (please see Table 2 for data collection statistics). Crystals of selenomethionine-labeled UCHL5N228^C88S were grown by hanging drop vapor diffusion method from a solution that contained 30% PEG (polyethylene glycol) 3350, 0.1M MgCl₂, 0.1M Tris-HCl, pH 8.5 and 3% trimethylaminooxide as an additive. Crystals were briefly soaked in 25% ethylene glycol as the cryoprotectant and plunged into liquid nitrogen for flash cooling. Diffraction data up to 2.0Å were collected and processed by HKL2000 [35] (Table 2).

The crystal structure of selenomethionine-labeled UCHL5N228^C88S was solved at 2.0 Å by molecular replacement (MR) using the structure of UCHL3 (PDB ID: 1UCH) as the search model [19]. A homology model of UCHL5N228^C88S was built based on the crystal structure of UCHL3 using the online model-building server SWISS-MODEL [36]. The segment comprising amino acids 134–165 was deleted from the homology model, which was then used as the search model in MR using Phaser, part of the CCP4 suite of programs [37]. The search identified two subunits in the asymmetric unit. Rigid-body followed by restrained refinement of this model yielded a crystallographic R factor (R_crys) and free R (R_free) of 40.5 and 50.7 %, respectively. Refinement using Refmac [38] after rounds of model building in Coot [39] brought R_crys and R_free to 20.4 and 26.5 %, respectively. The ordered parts of the deleted segment were built in during the course of model building and refinement. Final rounds of refinement conducted using Phenix [40] resulted in a model with final R_crys of 17.5 and R_free of 22.2 %, respectively. The current refined model includes the residues 6–146 and 157–227 in one subunit and 7–145 and 153–227 in the other. The structure of UCHL5N240 was solved by MR employing MOLREP [27] using the structure of UCHL5N228^C88S as the search model. Cross-rotation followed by translation identified four copies of the protein in the asymmetric unit. Further refinement using Refmac and Phenix, after rounds of model building using Coot, resulted in a crystallographic tetramer with R_crys and R_free of 20.4 and 25.8 %, respectively (Table 2). The current refined model includes the residues 7–154 and 161–239 in the chain A, 7–153 and 160–239 in chain B, 7–152 and 162–239 chain C and 7–151 and 162–238 in chain D. Coordinates and structure factors have been deposited in PDB under the entries 3RII (UCHL5N228^C88S) and 3RIS (UCHL5N240).

Enzymatic activity assay

Stock solution of full-length UCHL5, UCHL5N240 and the active-site mutants UCHL5N240^C88S, UCHL5N240^H164N, UCHL5N240^D179N were diluted with reaction buffer (50mM Tris.HCl, pH 7.6, 5mM DTT, 1mM EDTA, 0.1mg/ml BSA) in the individual wells of a 96-well plate to the final concentration of 1nM. Ub-AMC was added to that well in the final concentration of 600nM. The reaction volume in the each well was kept at 100µl. The rate of Ub-AMC hydrolysis was monitored at 25°C by TECHAN GENios fluorescence micro plate reader with excitation at 380nm and emission at 465nm. The amount of released AMC was quantified using 7-amido-4- methylcoumarin as a standard (Sigma Aldrich). K_m and k_cat/K_m values for UCHL5N240 were obtained from the measurement conducted at constant enzyme concentration (500pM) and varying Ub-AMC concentration (0–15µM). Initial velocity was calculated from the initial slope of each progress curve. The values of k_cat and K_M were determined from double-reciprocal plots assuming Michaelis-Menten model.

Abbreviations

UCH

Ubiquitin carboxy-terminal hydrolase

UCHL1

Ubiquitin carboxy-terminal hydrolase L1

UCHL3

Ubiquitin carboxy-terminal hydrolase L3

UCHL5

Ubiquitin carboxy-terminal hydrolase L5

AMC

Aminomethylcoumarin

RMSD

Root mean square deviation

VME

Vinyl methyl ester