Divergent N-terminal Sequences of a Deubiquitinating Enzyme Modulate Substrate Specificity*

Ubiquitin-specific processing proteases (UBPs) are characterized by a conserved core domain with surrounding divergent sequences, particularly at the N-terminal end. We previously cloned two isoforms of a testis UBP, UBP-t1 and UBP-t2, which contain identical core regions but distinct N termini that target the two isoforms to different subcellular locations (Lin, H., Keriel, A., Morales, C. R., Bedard, N., Zhao, Q., Hingamp, P., Lefrancois, S., Combaret, L., and Wing, S. S. (2000) Mol. Cell. Biol. 20, 6568–6578). To determine whether the N termini also influence the biochemical functions of the UBP, we expressed UBP-t1, UBP-t2, and the common core domain, UBP core, in Escherichia coli. The three isoforms cleaved branched triubiquitin at >20-fold faster rates than linear diubiquitin, suggesting that UBP-testis functions as an isopeptidase. Both N-terminal extensions inhibited the ability of UBP-core to generate free ubiquitin when linked in a peptide bond with itself, another peptide, or to small adducts. The N-terminal extension of UBP-t2 increased the ability of UBP-core to cleave branched triubiquitin. UBP-core removed ubiquitin from testis ubiquitinated proteins more rapidly than UBP-t2 and UBP-t1. Thus, UBP enzymes appear to contain a catalytic core domain, the activities and specificities of which can be modulated by N-terminal extensions. These divergent N termini can alter localization and confer multiple functions to the various members of the large UBP family.

The ubiquitin-proteasome pathway of protein degradation is a major mechanism for intracellular protein catabolism (reviewed in Refs. [1][2][3]. Proteins destined to be degraded through the ubiquitin-proteasome pathway are first covalently ligated with ubiquitin, a 76-amino acid peptide. This reaction involves the sequential action of three enzymes. Ubiquitin is first activated by ubiquitin-activating enzyme (E1) 1 (4) and is then transferred to a specific cysteine residue of one of a family of ubiquitin-conjugating enzymes (E2s) (5). Although some E2s can transfer ubiquitin to substrates directly in vitro, most E2s support ubiquitin conjugation to substrates by interaction with one of the many ubiquitin protein ligases (E3s) (6 -8). These E2/E3 enzymes form an isopeptide bond between the C terminus of ubiquitin and the ⑀-amino group of the side chain of lysine residues of the target protein. A branched polyubiquitin chain is then formed on the protein through the ligation of additional monomers of ubiquitin to the side chain of a lysine residue of the previous ubiquitin in successive rounds of ubiquitination (5). The covalent attachment of a polyubiquitin chain to proteins generally acts as a signal for their degradation by a multisubunit protease, the 26 S proteasome (9,10).
In addition to the families of enzymes involved in conjugation of ubiquitin, a very large family of deubiquitinating enzymes has recently been identified from various organisms (reviewed in Refs. [11][12][13]. These enzymes have several possible functions. First, they may have peptidase activity and cleave the products of ubiquitin genes. Ubiquitin is encoded by two distinct classes of genes. One is a polyubiquitin gene, which encodes a linear polymer of ubiquitins linked through peptide bonds between the C-terminal Gly and N-terminal Met of contiguous ubiquitin molecules (14). Each copy of ubiquitin must be released by precise cleavage of the peptide bond between Gly-76 -Met-1 of successive ubiquitin moieties (15). The other class of ubiquitin genes encodes ubiquitin C-terminal extension proteins, which are peptide bond fusions between the C-terminal Gly of ubiquitin and N-terminal Met of the extension protein (16 -18). To date, the extensions described are ribosomal proteins consisting of 52 or 76 -80 amino acids (19,20). These ubiquitin fusion proteins are processed to yield ubiquitin and the corresponding C-terminal extension proteins (21). Second, deubiquitinating enzymes may have isopeptidase activities. When a target protein is degraded, deubiquitinating enzymes can cleave the polyubiquitin chain from the target protein or its remnants (22)(23)(24). The polyubiquitin chain must also be disassembled by deubiquitinating enzymes during or after proteolysis by the 26 S proteasome, regenerating free monomeric ubiquitin (25,26). In this way, deubiquitinating enzymes can facilitate the ability of the 26 S proteasome to degrade ubiquitinated proteins. Third, deubiquitinating enzymes may hydrolyze ester, thiolester, and amide linkages to the carboxyl group of Gly-76 of ubiquitin (27)(28)(29). Such nonfunctional linkages may arise from reactions between small intracellular compounds such as glutathione and the E1-, E2-, or E3-ubiquitin thiolester intermediates. Fourth, deubiquitinating enzymes may compete with the conjugating system by removing ubiquitin from protein substrates, thereby rescuing them from degradation or any other function mediated by ubiquitination. Thus generation of ubiquitin by deubiquitinating enzymes from the linear polyubiquitin and ubiquitin fusion proteins and from the branched polyubiquitin ligated to proteins should be essential for maintaining a sufficient pool of free ubiquitin. Many deubiquitinating enzymes exist, suggesting that these deubiquitinating enzymes recognize distinct substrates and are therefore involved in specific cellular processes (11)(12)(13). Although there is recent evidence to support such specificity of these deubiquitinating enzymes (30 -32), the structure-function relationships of these enzymes remain poorly studied.
Deubiquitinating enzymes can be divided broadly on the basis of sequence homology into two classes, the ubiquitinspecific processing protease (UBP or USP, also known as type 2 ubiquitin C-terminal hydrolase (type 2 UCH)) and the UCH, also known as type 1 UCH) (12,13). UCH (type 1 UCH) enzymes hydrolyze primarily C-terminal esters and amides of ubiquitin (27) but may also cleave ubiquitin gene products and disassemble polyubiquitin chains (30). They have in common a 210-amino acid catalytic domain, with four highly conserved blocks of sequences that identify these enzymes. They contain two very conserved motifs, the CYS and HIS boxes. Mutagenesis studies revealed that the two boxes play important roles in catalysis (33,34). Some UCH enzymes have significant Cterminal extensions (12,32). The functions of the C-terminal extensions are still unknown but appear to be involved in proper localization of the enzyme (24,32). The active site of these UCH enzymes contains a catalytic triad consisting of cysteine, histidine, and aspartate and utilizes a chemical mechanism similar to that of papain (33,34).
UBP (type 2 UCH) enzymes are capable of cleaving the ubiquitin gene products (21) and disassembling polyubiquitin chains after hydrolysis (15). It appears that there is a core region of about 450 amino acids delimited by CYS and HIS boxes. Many of these isoforms have N-terminal extensions and a few have C-terminal extensions (12). In addition, there are variable sequences in the core region of many of the isoforms. The functions of these divergent sequences remain poorly characterized. Recently, we identified UBP-t, an UBP enzyme that is primarily expressed in the testis as two isoforms with the same core region (347 residues) but distinct N termini. The N-terminal extension of UBP-t1 has 49 residues and that of UBP-t2 has 271 residues. The divergent N termini were found to target distinct subcellular compartments (35). UBP-t1 is located primarily in the nucleus, whereas UBP-t2 is found primarily in a perinuclear location and can be associated with the centrosome. To evaluate whether in addition these N termini have functions in substrate specificity, we have tested the abilities of the two isoforms as well as the common core domain to cleave natural and semi-synthetic ubiquitin substrates.

MATERIALS AND METHODS
Preparation of Recombinant UBP-core, UBP-t1, and UBP-t2-To express full-length UBP-t1 and UBP-t2 and only the common core region of UBP-t1 and UBP-t2 (UBP-core), DNA fragments encoding the proteins were amplified by polymerase chain reaction and subcloned into the bacterial expression vector pET11-d (Novagen). Plasmids were sequenced to confirm accurate amplification and cloning and then transformed into Escherichia coli BL21 (DE3). After induction of expression with 1 mM isopropyl-1-thio-␤-D-galactopyranoside for 2 h at 28°C, cells were harvested from 800 ml of culture and rinsed with PBS, and the cell pellets were frozen at Ϫ20°C. Subsequent manipulations were on ice or at 4°C. The frozen pellets were resuspended in 1 ⁄10 of the original culture volume of 50 mM Tris, pH 7.5, 1 mM DTT, 5 mM EDTA, 0.5 mM phenylmethylsulfonyl fluoride, and 5 g/ml pepstatin A and lysed by sonication. The lysate was clarified by centrifugation at 100,000 ϫ g for 1 h, and then proteins were differentially precipitated by ammonium sulfate (ICN Ultra Pure). The UBP-core was enriched in the 30 -50% ammonium sulfate fraction. This fraction was dialyzed against 20 mM Mes, pH 6.2, 1 mM DTT overnight before being applied to a 5 ϫ 100-mm propylsulfonic acid cation exchange column (Waters SP/HR15) equilibrated in the same buffer. Bound proteins were eluted with a 0 -0.5 M NaCl gradient (10 mM/min) in the same buffer. UBP-core was eluted at ϳ0.14 M NaCl and in most fractions was Ͼ99% pure as evaluated by Coomassie Blue-stained acrylamide gels. The 30 -40% ammonium sulfate fraction containing the UBP-t1 was dialyzed against 50 mM Tris, pH 7.5, 1 mM DTT overnight before being applied to a quaternary amine anion exchange column (Amersham Pharmacia Biotech MonoQ) equilibrated in the same buffer. Full-length UBP-t1 was eluted at ϳ0.28 M NaCl. The 35-45% ammonium sulfate fraction containing the UBP-t2 was dialyzed against 50 mM Tris, pH 7.5, 1 mM DTT overnight before being applied to the quaternary amine anion exchange column equilibrated in the same buffer. Full-length UBP-t2 eluted at ϳ0.31 M NaCl. The fractions of each step were screened by both Western blots with anti-UBP-core-specific antibody and activity assays. The activities of the enzymes in different fractions were monitored by determining their abilities to hydrolyze 125 I-labeled Ub-PESTc (ubiquitin extended at the C terminus by MHISPPEPESEEEEEHYC) (36). These purification steps did not yield homogeneously pure UBP-t1 or UBP-t2 but succeeded in removing partially degraded forms. Control lysates prepared from uninduced cells show no activity under these conditions.
The concentrations of active UBP-core, UBP-t1, and UBP-t2 were determined by inhibitor titration with ubiquitin aldehyde (24). The enzymes were preincubated at 37°C with different amounts of ubiquitin-aldehyde for 5 min, and then the isopeptidase activities were assayed with 2 M 125 I-labeled branched triubiquitin (Ub 3 ) as described below.
Preparation and Use of Antibodies against the Core Region of UBP-t1 and UBP-t2-Antibodies specific for the core region of UBP-t1 and UBP-t2 were prepared by immunizing rabbits with Freund's adjuvant mixed with UBP-core bearing an N-terminal His 6 tag. Antibodies were affinity-purified by passing crude anti-serum over an Affi-Gel 10 (Bio-Rad) column coupled to glutathione S-transferase fused to the same UBP-core sequence.
To detect UBP-core, UBP-t1, and UBP-t2, protein samples were resolved by SDS-PAGE on 10% acrylamide gels and transferred to nitrocellulose membranes. Membranes were probed with anti-UBP-core antibody followed by incubation with horseradish peroxidase-protein A and detection by chemiluminesence (ECL, Amersham Pharmacia Biotech).
Iodination of Proteins-The chloramine-T method was used to label ubiquitin, diubiquitin, and triubiquitin with Na 125 I. Ub-PESTc was labeled with Na 125 I using Iodobeads (Pierce). Unincorporated 125 I was removed by passing the reaction products over a Sephadex G-25 column.
Ubiquitin Peptidase and Isopeptidase Assays-To measure the abilities of UBP-core, UBP-t1, and UBP-t2 to cleave some natural substrates and semi-synthetic ubiquitin derivatives, the different enzymes were incubated in a total volume of 20 l as follows: 100 mM Tris, pH 7.5, 1 mM DTT, 1 mM EDTA, 1 mg/ml bovine serum albumin or ovalbumin, 2 M 125 I-labeled substrate. After incubating at 37°C for various times, the reactions were quenched as indicated below. For the substrate Ub-PESTc, the reaction was terminated by adding 100 l of 2.5% bovine serum albumin and 1 ml of 20% (w/v) trichloroacetic acid. After incubation on ice for 30 min, the samples were centrifuged, and the resulting supernatants were counted for their radioactivities to detect the PESTc peptide bearing the iodinated tyrosine. For the linear diubiquitin (Ub 2 ) or branched Ub 3 substrates, the reaction was stopped with Laemmli sample buffer containing 2-mercaptoethanol, and the products were resolved by SDS-PAGE on 20% acrylamide gels and detected by autoradiography. After detection by autoradiography, the monoubiquitin (Ub) band and Ub 2 band (in the assay with Ub 3 ) were excised from the dried gel and counted. The rate of cleavage of peptide bonds for linear Ub 2 substrate was calculated as half of the Ub produced in this reaction. The rate of cleavage of isopeptide bonds for branched Ub 3 substrate was calculated as (2/3 (Ub Ϫ Ub 2 )) ϩ Ub 2 produced in this reaction. To test the abilities of these UBP enzymes to cleave ubiquitin ester and ubiquitin linked to lysine by the ␣ or ⑀ amino groups, they were incubated with these substrates (16 M) separately in 10 l of 50 mM Tris, pH 7.6, 3 mM DTT, 5 mM MgCl 2 at 37°C for 20 or 30 min. The reactions were quenched with 40 l of 0.1 N HCl and then injected onto a 4.6 ϫ 250-mm C8 column equilibrated and eluted with 60% acetonitrile in 25 mM perchlorate to resolve the free ubiquitin product from the substrates (29). Fluorometric assays used ubiquitin-AMC as the substrate (37). In a typical assay, 80 l of assay buffer (50 mM Tris, pH 7.8, 100 g/ml ovalbumin, 10 mM DTT) containing 6.6 nM enzyme were added to a 100-l cuvette. When present, diubiquitin analogs were added, and after a 5-min preincubation at 37°C to achieve thermal equilibrium, ubiquitin-AMC was added to a final concentration of 40 nM. Reaction progress was monitored by the increase in fluorescence emission at 450 nm ( ex ϭ 355 nm) that resulted from the cleavage of AMC from the substrate.
To compare the abilities of these enzymes to cleave the substrates Ub-PESTc, linear Ub 2 , and branched Ub 3 , equivalent amounts of these enzymes (38 nM) were incubated with 2 M substrate. The initial rates obtained with the different enzymes were compared. To compare the abilities of UBP-core, UBP-t1, and UBP-t2 to interact with the Ub-PESTc substrate, the apparent K m values for this substrate were determined. Variable concentrations of the substrate were assayed in reactions as described above. The products were monitored by removing aliquots from the reaction at various times. Initial velocities were calculated from the time courses and used in double-reciprocal plots (Lineweaver-Burk) to determine apparent K m values and maximal velocities.
To determine the dissociation constants for the inhibition of UBPcore by nonhydrolyzable diubiquitin, reactions were conducted using ubiquitin-AMC as substrate at five or more different concentrations of dimers. The dissociation constant was calculated by a non-linear curvefitting analysis using fitting function where v o is the uninhibited rate, v f is the rate at high concentration of inhibitor, and K i is the inhibition constant. In all cases inhibition was fully competitive, and v f ϭ 0.
To evaluate the abilities of these enzymes to remove ubiquitin from testis ubiquitinated proteins, rat testis extracts were prepared by homogenizing the tissues in 5 vol/g of wet weight of 0.25 M sucrose, 50 mM Tris, pH 7.5 (at 4°C), 5 mM N-ethylmaleimide (to inactivate endogenous isopeptidases), 1 mM EDTA, and protease inhibitors (1 mM phenylmethylsulfonyl fluoride, 10 g/ml pepstatin A, and 10 g/ml leupeptin). The extracts were clarified by centrifugation at 3000 ϫ g for 20 min at 4°C and then frozen at Ϫ80°C until analysis. Before assaying, the excess N-ethylmaleimide in the samples was neutralized by adding DTT to a final concentration of 5 mM. Aliquots (50 g) of the protein were then incubated with equal amounts of each of the UBP enzymes (38 nM) for various times at 37°C. The reaction was stopped with Laemmli sample buffer containing 2-mercaptoethanol, and the proteins were resolved by SDS-PAGE on 10% acrylamide gels and transferred to nitrocellulose membranes. The ubiquitinated proteins were detected by Western blotting with anti-ubiquitin antibodies (Sigma) followed by protein A coupled to horseradish peroxidase and a chemiluminescent detection method (ECL, Amersham Pharmacia Biotech).

RESULTS AND DISCUSSION
Preparation of UBP-core, UBP-t1, and UBP-t2-To characterize the biochemical properties of UBP-t1 and UBP-t2 as well as their common core region, UBP-core (Fig. 1A), the proteins were expressed in E. coli. Since expressed proteins were susceptible to degradation, purification of the various enzymes was undertaken to remove degradation products (Fig. 1B). UBP-core was purified to apparent homogeneity, as estimated by Coomassie Blue-staining of the protein on a polyacrylamide gel. UBP-t1 and UBP-t2 were enriched to levels of specific activity that were 17-and 27-fold greater, respectively, than in the crude bacterial lysates. Control lysates not expressing enzymes did not show activity in these assays, and the activities of all isoforms were found to be highly susceptible to inhibition by ubiquitin aldehyde, and therefore, they were quantitated by titration with this inhibitor.
N-terminal Extensions Inhibit the Peptidase Activity of UBPcore-Since a key function of many deubiquitinating enzymes is to process the products of ubiquitin genes, we tested whether UBP-t1 and UBP-t2 can generate free ubiquitin from a fusion protein with another peptide, as in the case of the model substrate Ub-PESTc. UBP-core readily cleaved the ␣-peptide bond between ubiquitin and the PESTc extension, indicating that the core region itself has activity and therefore likely represents a functional domain. However, both UBP-t1 and UBP-t2 had very low activities (Ͻ20% that seen with UBP-core) with this substrate (Fig. 2). Thus the N-terminal extensions inhibited the hydrolytic activity inherent in the core domain. To test whether the N-terminal extensions of UBP exert this inhibitory effect by decreasing the abilities of the UBP core to interact with Ub-PESTc, we measured apparent K m values of Ub- FIG. 1. A, schematic of the UBP-t1, UBP-t2, and UBP-core isoforms. UBP-t1 and UBP-t2 consist of identical core regions containing the conserved motifs of the UBP family (CYS and HIS boxes are indicated) but distinct N termini. B, the three isoforms were expressed in E. coli, and the crude extracts were purified sufficiently to remove partially degraded forms. UBP-t1 and UBP-t2 were analyzed by immunoblotting with anti-UBP-core antibodies. UBP-core was analyzed by SDS-PAGE followed by staining with Coomassie Blue. PESTc for UBP-core, UBP-t1, and UBP-t2. Indeed, UBP-t2 and UBP-t1 had greater than 10-fold higher K m values with this substrate than did UBP-core (Table I). However, the V max value of UBP-t2 was similar to that of the UBP-core enzyme, indicating that the N-terminal extensions of UBP can negatively influence binding of the core domain to the substrate without significantly affecting catalytic function.
To test whether these enzymes may be involved in processing the linear polyubiquitin that arises from the polyubiquitin genes, we tested the activities of these enzymes against linear Ub 2 . UBP-t1, UBP-t2, and UBP-core all had low abilities to generate free ubiquitin from Ub 2 . Interestingly, as occurred for Ub-PESTc, the initial velocities of UBP-t1 and UBP-t2 to cleave this ␣-peptide bond were lower than that of UBP-core (Fig. 3). This confirms the previous observation (Fig. 2) that N-terminal extensions inhibit the ability of the core region to cleave linear peptide fusions. All of the above results would suggest that both UBP-t1 and UBP-t2 do not generate free ubiquitin from ubiquitin gene products. They also do not appear to be involved in processing the precursor forms of the ubiquitin-related proteins SUMO1 (sentrin) or NEDD8, as none of the three isoforms were able to cleave proSUMO1 or proNEDD8 (data not shown). Thus the core region alone has activity, but the N-terminal extensions limit that activity presumably by imposing constraints and selectivity on the ability of substrates to access the core domain.
Abilities of UBP-core, UBP-t1, and UBP-t2 to Cleave Ubiquitin Fused to Small Adducts-It has been postulated that ubiquitin may become linked to small thiols and amines in the cell arising from nucleophilic attack of these compounds on the reactive thiol ester linkage between ubiquitin and E1, E2s, or some E3s. Deubiquitinating enzymes may play an important role in regenerating ubiquitin from these nonproductive adducts. Indeed, the deubiquitinating enzymes UCH-L1 and UCH-L3 can efficiently hydrolyze these small thiols and amines from ubiquitin in vitro. To test the potential function of our UBP enzymes to hydrolyze such compounds, Ub ester and N␣-Ub-L-lysine were used as substrates. UBP-core had higher activity in hydrolyzing the ester and lysine from ubiquitin than both UBP-t1 and UBP-t2 (Table II). This indicates that the N-terminal extensions of both UBP-t1 and UBP-t2 inhibit the ability of the core region to hydrolyze small thiols and amines from ubiquitin and further supports the role of these extensions in imposing substrate selectivity.
Ubiquitin is conjugated to proteins through an isopeptide bond (through the ⑀-amino group of lysine). So it was of interest to examine whether UBP-t1 and UBP-t2 could cleave Ub-⑀amino lysine derivatives. When N⑀-Ub-L-lysine was used as such a model isopeptidase substrate, UBP-core still had higher activity than UBP-t1 or UBP-t2 against the ⑀-isopeptide bond (Table II). Thus, at least for a small leaving group, UBP-core does not discriminate between isopeptide and peptide linkages. This is similar to UCH enzymes tested to date that also do not appear to discriminate between these two model substrates (30).
Branched Ub 3 Is Preferentially Cleaved by UBP-t1 or UBP-t2 Isoforms-Most biological functions of ubiquitin are mediated by the linkage of the C-terminal glycine of ubiquitin in an isopeptide bond with the ⑀-amino group of the side chain of lysine residues of the protein substrate. Recognition of ubiquitinated proteins for degradation by the 26 S proteasome generally requires the presence of a polyubiquitin chain on the protein substrate in which each ubiquitin is linked to each other via these isopeptide linkages. To evaluate whether UBPcore, UBP-t1, and UBP-t2 can potentially disassemble such a polyubiquitin chain, Ub 3 , a branched triubiquitin chain linked via isopeptide bonds between the ⑀-amino group of lysine 48 in one ubiquitin molecule and the C-terminal of another ubiquitin molecule, was used as a substrate. All three enzymes had activity against this substrate, and the rates of cleavage were similar to that for Ub-PESTc but 1-2 orders of magnitude higher than that seen for linear Ub 2 (Fig. 4). Thus, the UBP core domain appears to prefer branched rather than linear ubiquitin polymers as substrate. The higher rate of cleavage of triubiquitin compared with diubiquitin was not due to the extra moiety of ubiquitin in the former, as diubiquitin did not accumulate over time in the reactions with triubiquitin as substrate (data not shown). In addition, UBP-t2 had higher activity than UBP-core. This indicates that the N-terminal extension of UBP-t2 not only suppresses peptidase activity but enhances isopeptidase activity. In contrast, UBP-t1 had lower activity    than UBP-core in cleaving branched Ub 3 (Fig. 4). This indicates that the N-terminal extensions can modulate the intrinsic activity of the core domain both positively and negatively.
Although ubiquitin moieties joined by Gly-76 -Lys-48 linkages appear to be most efficient at targeting for proteasomalmediated degradation, linkages of Gly-76 to other lysine residues of ubiquitin have been observed. To evaluate whether these enzymes show preference for specific linkages, the ability of nonhydrolyzable ubiquitin dimer analogs joined in various linkages (38) to inhibit these enzymes was tested. These dimer analogs were synthesized by using dichloroacetone to cross-link ubiquitin containing a terminal cysteine to another ubiquitin in which individual lysines have been mutated to cysteine. To determine inhibitory constants easily and accurately, a sensitive continuous fluorometric assay using ubiquitin-AMC as substrate was employed (Table III). At the low concentrations of ubiquitin-AMC used, only UBP-core and UBP-t2 had significant activities, and so UBP-t1 was not tested. The apparent K m of ubiquitin-AMC for UBP-core was similar to that of Ub-PESTc. However, in contrast to Ub-PESTc, where the affinity for UBP-t2 decreased 1 order of magnitude compared with UBP-core, the affinity of ubiquitin-AMC actually increased approximately 3-fold. When the ubiquitin dimer analogs were tested, diubiquitin linked by lysine 63 was found to be most inhibitory. However, the differences between the inhibition constants of the analogs were relatively small, suggesting that there is no significant preference of the enzymes for a particular linkage. Furthermore, even for the lysine 63-linked analog, the K i value was relatively high (10 Ϫ7 M range) compared with an enzyme such as isopeptidase T in which diubiquitin linked by lysine 48 inhibits with a K i in the range of 10 Ϫ8 M (38). It is conceivable that UBP-t2 shows enhanced activity against branched Ub 3 due to the presence in this isoform of a binding site for an additional ubiquitin moiety. To evaluate this possibility, the K i values of monoubiquitin and the dimers for UBPcore and for UBP-t2 were compared to see if the dimer K i values were lower for the UBP-t2 isoform. The ratios of K i for UBPcore to K i for UBP-t2 were similar for ubiquitin monomer and for all of the dimers, suggesting that the N-terminal extension in UBP-t2 does not confer an additional binding site for ubiquitin (Table III).
UBP-core, UBP-t2, and UBP-t1 Can Remove Ubiquitin from Endogenous Testis Proteins-Since the various isoforms could disassemble a short branched polyubiquitin chain, we tested whether they had isopeptidase activity against a broader spectrum of ubiquitinated proteins in a testis extract. The UBPcore and UBP-t2 enzymes deubiquitinated ubiquitin from high molecular weight ubiquitinated proteins, but the UBP-t1 enzyme had relatively low activity (Fig. 5). This is consistent with the observation that UBP-t1 was the least efficient among the three isoforms in cleaving branched triubiquitin (Fig. 4). UBPcore was most active, suggesting that the N termini impose varying degrees of substrate specificity on the isopeptidase activity. Without the addition of the enzymes, the levels of ubiquitinated proteins were stable, confirming that endogenous deubiquitinating enzymes had been inactivated by the presence of N-ethylmaleimide in the homogenization buffer. The activities of the UBP isoforms were inhibited by ubiquitin aldehyde, indicating that these cleavages were unlikely to be due to any contaminating bacterial proteases.
In summary, the UBP core region containing the conserved elements of the UBP family of enzymes and delimited by the CYS and HIS boxes contains ubiquitin-specific protease activity. Thus we provide the first direct evidence that this core region represents a functional and quite possibly a structural domain also. Interestingly, this core domain appears to preferentially cleave isopeptide rather than peptide linkages (compare rates in Figs. 3 and 4), thus suggesting that the UBP family functions primarily in either the regeneration of ubiquitin from polyubiquitin chains produced after the action of the 26 S proteasome or in the editing or rescuing of ubiquitinated protein substrates before the action of the 26 S proteasome. Indeed, the isoforms were capable of removing ubiquitin from endogenous testis proteins (Fig. 5). The lower K i values for the ubiquitin dimers compared with ubiquitin alone (Table III) would also be consistent with preferential binding of branched polyubiquitin to the core region of the enzyme.
We have previously shown that the N-terminal extensions of UBP-t1 and UBP-t2 serve to localize the enzyme to different compartments of the cell (35). It should be noted that UBP-t2 is localized in a perinuclear pattern resembling that of the proteasome and is also the most active on branched polyubiquitin conjugates. It is intriguing to speculate that these indicators

TABLE III
Kinetic constants of ubiquitin-AMC for UBP-core and UBP-t2 and inhibition of these isoforms by ubiquitin and nonhydrolyzable ubiquitin dimer analogs Various concentrations of ubiquitin-AMC were incubated with the indicated enzymes (6.6 nM), and the release of fluorescent AMC was followed continuously as described under "Materials and Methods." To determine inhibitor constants of ubiquitin and various ubiquitin dimer analogs, the rates of cleavage of ubiquitin-AMC were measured in the presence of various concentrations of the inhibitor and then analyzed by a nonlinear curve-fitting analysis as described under "Materials and Methods." may suggest a role for UBP-t2 in metabolism of branched polyubiquitin at or near the proteasome. Our data demonstrate clearly and for the first time that these divergent N termini can also modulate the activity of the core domain. Interestingly, the activity can be both positively and negatively affected by the extension depending on the particular substrate. For the substrates for which measures of affinity could be determined, the parameters obtained were relatively high (in the micromolar range). Rates of cleavage were also relatively low. Thus, there are probably specific substrates for these enzymes, and the N-terminal extensions may play roles in recognizing these specific ubiquitinated proteins. Indeed, without N-terminal extensions, UBP-core efficiently removed ubiquitin from endogenous proteins. However, with N-terminal extensions, UBP-t2 did this at a slower rate, and UBP-t1, hardly at all (Fig. 5). Thus, it is quite possible that the N-terminal extensions are positioned near the S1Ј site, and the different effects of the UBP-t2 extension on affinity for ubiquitin-AMC and ubiquitin-PESTc would support this model (Tables I and III). In addition, given their precise localization in the cell, the enzymes may be colocalized with their substrates in specific compartments. Such co-localization would permit privileged delivery of substrates to the enzymes and diminish dependence of access to the enzymes on factors such as affinity constants and diffusion rates. Finally, the observations on UBP-testis in this and our previous work (35) indicate that the numerous distinct N-terminal extensions of UBP enzymes could permit specific spatial, temporal, and kinetic modulation of UBP function and allow these enzymes to mediate quite precise functions in the ubiquitin-dependent proteolytic pathway.