INTRODUCTION

The interaction of cytokines with their cognate receptors induces the growth and differentiation of specific hematopoietic cells. Receptor activation results in the rapid induction of new mitogenic signaling pathways. The interaction of IL-2¹ with the IL-2 receptor, for example, activates multiple signal transduction pathways (1) including the Ras/Raf/mitogen-activated protein kinase pathway (2) and the JAK/STAT pathway (3, 4, 5). In addition to these general mitogenic pathways activated by all cytokines, IL-2 induces specific signals that are unique to its function in T cell development (6) (7). IL-2-specific immediate-early genes presumably control signaling mechanisms that are unique to the IL-2 receptor and not utilized by other cytokine receptors.

Cytokine receptor activation also results in the induction of new genes regulating intracellular proteolysis. The ubiquitin-mediated proteolytic pathway, for example, can be modified by cytokine stimuli (8, 9). The ubiquitin-mediated proteolytic pathway has recently been implicated in multiple cellular processes, including cell cycle regulation (10), transcriptional activation (11), and antigen presentation (12). Proteins targeted for degradation are initially conjugated to the 76-amino acid ubiquitin (Ub) polypeptide tag. Several classes of enzymes, including ubiquitin-activating enzymes (E1), ubiquitin carrier proteins (E2), and ubiquitin ligases (E3) are required to attach Ub to protein targets (12, 13, 14, 15). Over 50 enzymes are known to be involved in ubiquitin conjugation, and most are highly conserved among species. Polyubiquitinated proteins are next recognized and degraded by the proteasome, a multisubunit protein degradation complex (16, 17, 18).

Protein ubiquitination also serves regulatory functions in the cell that do not involve proteasome-mediated degradation (19). For example, Hicke and Riezman (20) have recently demonstrated ligand-inducible ubiquitination of the Ste2 receptor in yeast. Ubiquitination of the Ste2 receptor triggers receptor endocytosis and receptor targeting to vacuoles, not proteasomes. Also, Chen et al. (21) have demonstrated that activation of the IB kinase requires a rapid, inducible ubiquitination event. This ubiquitination event is a prerequisite for the specific phosphorylation of IB and does not result in subsequent proteolysis of the kinase complex. The ubiquitination of Ste2 and IB kinase appears reversible, perhaps resulting from the action of a specific deubiquitinating enzyme.

A large superfamily of genes encoding deubiquitinating enzymes, or ubps, has recently been identified (17, 22, 23, 24). Ubps are ubiquitin-specific thiol-proteases that cleave either linear ubiquitin precursor proteins or post-translationally modified proteins containing isopeptide ubiquitin conjugates. The large number of ubps suggests that protein ubiquitination, like protein phosphorylation, is a highly reversible process that is regulated in the cell.

Interestingly, ubps vary greatly in length and structural complexity, suggesting functional diversity. While there is little amino acid sequence similarity throughout their coding region, sequence comparison reveals two conserved domains. The Cys domain contains a cysteine residue that serves as the active enzymatic nucleophile (24). The His domain contains a histidine residue that contributes to the enzyme's active site. More recent evidence demonstrates six homology domains contained by all members of the ubp superfamily (22). Mutagenesis of conserved residues in the Cys and His domains has identified several residues that are essential for ubp activity (24) (25).

Deubiquitinating enzymes have multiple functional roles within the cell. Some deubiquitinating enzymes, such as isopeptidase T (22, 26) hydrolyze branched polyubiquitin chains and thereby regulate cellular pools of free monomeric ubiquitin. Other deubiquitinating enzymes, such as faf (25, 27), remove Ub from cellular target proteins, and thereby prevent their proteasome-mediated degradation. Still other deubiquitinating enzymes, such as Doa-4 (24), remove ubiquitin from Ub-peptide degradation products produced by the proteasomes and thereby accelerate proteasome-mediated degradation.

Specific deubiquitinating enzymes regulate cellular growth. The mammalian protooncogene, tre-2, for example, encodes a deubiquitinating enzyme, and the tre-2 oncoprotein exhibits transforming activity in 3T3 fibroblasts (24, 28). The unp gene encodes a deubiquitinating enzyme and is tumorigenic in transgenic mice (29, 30). The Drosophila faf gene determines cell growth and cell differentiation during Drosophila eye development (27). Interestingly, the encoded faf deubiquitinating enzyme functions at a preproteasomal level (25).

We have recently identified a growth regulatory deubiquitinating enzyme, DUB-1, that is rapidly induced in response to cytokine receptor stimulation (31). DUB-1 is specifically induced by the receptors for IL-3, granulocyte macrophage-colony-stimulating factor, and IL-5, suggesting a specific role for the c subunit shared by these receptors (32). In the process of cloning the DUB-1 gene, a family of related, cross-hybridizing DUB genes was identified. We reasoned that these other DUB genes might be induced by different growth factors. Here we report the identification of an IL-2-inducible DUB enzyme, DUB-2. DUB-1 and DUB-2 are more related to each other than to other members of the ubp superfamily and thereby define a novel subfamily of deubiquitinating enzymes.

MATERIALS AND METHODS

Ba/F3 is an IL-3-dependent murine pro-B cell line. Ba/F3 cells were maintained in RPMI 1640 medium supplemented with 10% (v/v) fetal calf serum and 10% conditioned medium from WEHI-3B cells as a source of murine IL-3. CTLL-2 cells were maintained in RPMI 1640 medium supplemented with 10% (v/v) fetal calf serum and 2 units/ml of murine recombinant IL-2 (Boehringer Mannheim). For induction, CTLL cells were starved in RPMI 1640, 10% fetal calf serum for 8 h and stimulated for various times with IL-2-containing medium.

For Northern blots, RNA samples (10-30 µg) were electrophoresed on denaturing formaldehyde gels and blotted onto Duralon-UV membranes (Stratagene). For Southern blots, genomic DNA (10 µg) was digested with the indicated restriction enzymes, electrophoresed on 1% agarose gel, and blotted onto Duralon-UV membranes (Stratagene). The indicated cDNA probes, purified from agarose gels (Qiagen), were radiolabeled and hybridized for 1 h to the membranes in a 68 °C oven. Hybridized filters were washed at room temperature in 0.1 × SSC and 0.1% sodium dodecyl sulfate.

Immunoprecipitation and immunoblotting of the DUB-2 polypeptide was performed as described previously (32) using a cross-reacting anti-DUB-1 antiserum.

A cDNA for DUB-2 was isolated by RT-PCR, using total cellular RNA prepared from IL-2-stimulated CTLL cells. Primers were derived from the DUB-1 sequence (31). The 5 primer was 5-TTTGAAGAGGTCTTTGGAGA-3, which is upstream of the ATG start codon. The 3 primer was 5-GTGTCCACAGGAGCCTGTGT-3, which is downstream of the TGA stop codon. The PCR reaction was performed at 94 °C for 1 min, 55 °C for 1 min, and 72 °C for 2 min for a total of 35 cycles. The amplified cDNA was cloned into pCRII (Invitrogen) and sequenced by the dideoxynucleotide method. PCR errors were eliminated by confirming the sequence of six independent cDNA clones and by comparing the sequence of the DUB-2 genomic clone.

A mouse genomic library prepared from ES cells and cloned in Lambda FIX II (Stratagene, La Jolla, CA) was screened with the open reading frame (ORF) of the DUB-1 cDNA. The probe was labeled with [³²P]dCTP by the random primer method (Pharmacia Biotech Inc.), and hybridization was performed in 0.8 M NaCl, 0.02 M Pipes, pH 6.5, 0.5% SDS, 50% deionized formamide, and 100 µg/ml denatured, sonicated salmon sperm DNA for 16 h at 42 °C. A total of 1 × 10⁶ recombinant phage were screened, and four positive clones were identified. By sequence analysis, one phage clone was found to contain the full-length DUB-1 coding region (32). A second phage clone was found to contain the full-length DUB-2 coding region. The other two phage clones contained different genes with ORFs bearing considerable homology to DUB-1 and DUB-2 (approximately 90% amino acid identity).² These genes are DUB subfamily members, and we refer to them as DUB-3 and DUB-4.

The deubiquitination assay of ubiquitin--galactosidase fusion proteins has been described previously (23, 24). A 1638-base pair fragment from the wild-type DUB-2 cDNA (corresponding to amino acids 1-545) and a cDNA containing a missense mutation (C60S) were generated by PCR and inserted, in frame, into pGEX (Pharmacia), downstream of the glutathione S-transferase (GST) coding element. Ub-Met--gal was expressed from a pACYC184-based plasmid. Plasmids were co-transformed as indicated into MC1061 Escherichia coli. Plasmid-bearing E. coli MC1061 cells were lysed and analyzed by immunoblotting with a rabbit anti--gal antiserum (Cappel), a rabbit anti-GST antiserum (Santa Cruz), and the ECL system (Amersham Corp.).

Interspecific backcross progeny were generated by mating (C57BL/6J × Mus spretus)F₁ females and C57BL/6J males as described (33). A total of 205 N₂ mice were used to map the DUB loci (see text for details). DNA isolation, restriction enzyme digestion, agarose gel electrophoresis, Southern blot transfer, and hybridization were performed essentially as described (34). All blots were prepared with Hybond-N⁺ nylon membrane (Amersham). Several different DUB probes were used in these studies. The initial probe was a mouse DUB-1 cDNA, containing the full-length ORF of DUB-1. This probe detected multiple fragments, suggesting that the probe recognized more than one DUB locus. However, all polymorphic fragments co-segregated, suggesting that all DUB loci were tightly linked. The probe used in the studies reported here was an ~800-base pair HindIII/KpnI genomic fragment which localized 7 kb downstream of the DUB-2 coding region. This probe only detects DUB-2 and DUB-3. Fragments of 6.6 and 3.2 kb were detected in SacI-digested C57BL/6J DNA, and fragments of 2.7 and 1.7 kb were detected in SacI-digested M. spretus DNA. The presence or absence of the 2.7- and 1.7-kb SacI M. spretus-specific fragments, which again co-segregated, was followed in backcross mice.

A description of the probes and restriction fragment length polymorphisms for the loci linked to the DUB loci has been reported previously (35).³ Recombination distances were calculated as described (37) using the computer program SPRETUS MADNESS. Gene order was determined by minimizing the number of recombination events required to explain the allele distribution patterns.

RESULTS

DUB-1 was originally cloned as an IL-3-inducible immediate-early gene (31). To identify related mRNAs that are specifically induced by other growth factors, CTLL cells were removed from growth factor and restimulated with IL-2 (Fig. 1). An inducible 2.9-kb mRNA (DUB-2) was identified that weakly hybridized with the full-length DUB-1 cDNA probe. The 2.9-kb DUB-2 mRNA was distinct from the 3.1-kb DUB-1 mRNA described previously (data not shown). DUB-2 mRNA levels were rapidly induced within 30 min of IL-2 restimulation but declined after 6 h, similar to the induction kinetics observed for DUB-1 mRNA in IL-3-responsive cells (31). The DUB-2 mRNA was superinduced in the presence of cycloheximide (10 µg/ml), thereby defining DUB-2 as an immediate-early gene. The DUB-2 mRNA (2.9 kb) was also expressed in the murine T cell line 3DO (38) but not in IL-3-dependent murine hematopoietic cell lines, including 32D and Ba/F3. The DUB-2 mRNA was expressed in murine primary T cells, but was not expressed in other normal murine tissues (data not shown).

To identify the DUB-2 protein expressed in CTLL cells, we utilized an anti-DUB-1 antiserum (32) (Fig. 1B). IL-2 induced the expression of a 62-kDa protein after 30 min of cell stimulation (lane 2). The DUB-2 protein was distinct from the 59-kDa DUB-1 protein described previously (compare lanes 8 and 10). DUB-2 protein levels were induced and declined after 6 h, similar to the induction kinetics observed for the DUB-2 mRNA.

Using oligonucleotide primers derived from the DUB-1 cDNA sequence and RNA from IL-2-stimulated CTLL cells, we isolated the DUB-2 cDNA by RT-PCR (Fig. 2). The isolated cDNA fragment contained a single open reading frame, predicting a protein similar in size to that observed in Fig. 1B. Six independent cDNA clones generated by RT-PCR from IL-2-stimulated CTLL cells were sequenced, and all were DUB-2. The DUB-2 cDNA was not amplified by RT-PCR from unstimulated CTLL cells or from IL-3-stimulated Ba/F3 cells or 32D cells. In addition, six independent cDNA clones derived by RT-PCR from IL-3-stimulated Ba/F3 cells were all DUB-1 (data not shown). Taken together, these results demonstrate that DUB-2 is induced by IL-2 and is expressed in T lymphocytes.

In order to verify the sequence of the DUB-2 cDNA and to obtain additional 5- and 3-untranslated sequence and intron sequence, we isolated a full-length genomic clone for murine DUB-2 (Fig. 2A). The nucleotide sequence of the DUB-2 cDNA was identical to the coding region of the DUB-2 genomic clone. The DUB-2 gene contains a small exon (exon 1) encoding amino acids 1-9 and a larger exon (exon 2) encoding amino acids 10-545, similar to the genomic organization of the DUB-1 gene (32). The single intron is 838 base pairs. The sequence of the intron-exon junction conforms to a consensus sequence for a eukaryotic splice site. A region of the genomic clone 5 to the ATG start site contained a stop codon.

The DUB-2 cDNA encodes a polypeptide of 545 amino acids (62 kDa) (Fig. 2B), consistent with its observed electrophoretic mobility (Fig. 1B). DUB-2 has 93% amino acid similarity and 88% amino acid identity to DUB-1. Both DUB-1 and DUB-2 polypeptides contain the highly conserved Cys and His domains (24). These domains are likely to form the enzyme's active site. The putative active site nucleophile of DUB-2 is a cysteine residue (Cys^-60) in the Cys domain. In addition, DUB-1 and DUB-2 have a lysine rich region (Lys domain; amino acids 374-384 of DUB-2) and a short hypervariable region (amino acids 385-451 of DUB-2), in which the DUB-1 and DUB-2 sequences diverge considerably. The hypervariable (HV) region of DUB-2 contains a duplication of the eight-amino acid sequence: PQEQNHQK.

Sequence alignment of DUB-1, DUB-2, and several other deubiquitinating enzymes identifies the six homology domains of the ubp superfamily, as described previously (22). Interestingly, DUB-1 and DUB-2 are more related to each other than to other members of the ubp superfamily (Fig. 3A). DUB-1 was 88% identical to DUB-2 and 48% identical to d38378, a putative human DUB protein found in the GenBank^TM. In contrast, DUB-1 was only 18-32% identical to other known members of the ubp superfamily. DUB-1, DUB-2, and d38378 therefore define a discrete DUB subfamily of the ubp superfamily. Other subfamilies within the ubp superfamily are evident from an evolutionary display of ubp sequences (Fig. 3B). For instance, the yeast ubps, doa-4 and p39944, are highly related to each other (44% identical).

To further characterize the HV domain of the DUB subfamily, we have aligned this region for three DUB members (Fig. 3C). Included in this comparison is the sequence of DUB-3, an additional DUB gene isolated by our genomic screen. All three of these DUB proteins are functional deubiquitinating enzymes in vitro (see below). The HV domain extends from amino acid 385 to 434 of DUB-1. DUB-1 and DUB-2 have more sequence identity in this region relative to DUB-3, except for an 11-amino acid insertion in DUB-2. The function of the HV domain is unknown, but may confer substrate specificity.

In order to determine whether DUB-2 has deubiquitinating activity, we expressed DUB-2 as a GST fusion protein (Fig. 4). The open reading frame of DUB-2 was subcloned into the bacteria expression vector pGEX. pGEX-DUB-2 was co-transformed into E. coli (MC1061) with a plasmid encoding Ub-Met--gal, in which ubiquitin is fused to the NH₂ terminus of -galactosidase. As shown by immunoblot analysis (Fig. 4A), a cDNA clone encoding GST-DUB-2 fusion protein resulted in cleavage of Ub-Met--gal (lane 5) to an extent comparable with that observed with GST-DUB-1 (lane 3). A GST-DUB-3 fusion protein, synthesized from the coding region of the DUB-3 genomic clone, had similar activity (lane 7). As control, cells transformed with the pGEX vector (lane 1) or pBluescript vector with a nontranscribed DUB-2 insert (lane 2) failed to cleave Ub-Met--gal. A mutant DUB-2 polypeptide, containing a C60S mutation, was unable to cleave the Ub-Met--gal substrate (lane 6). Taken together, these results demonstrate that DUB-2 has deubiquitinating activity and that Cys^-60 is critical for its thiol protease activity. An anti-GST immunoblot confirmed that the GST-DUB-1 and GST-DUB-2 proteins were synthesized at comparable levels (Fig. 4B).

As further evidence for a DUB subfamily of deubiquitinating enzymes, we have isolated four genomic clones encoding DUB enzymes by screening a murine genomic library.² The DUB-1 cDNA probe hybridized with DUB-1, DUB-2, and two additional genomic clones that encode novel deubiquitinating enzymes (DUB-3 and DUB-4). The DUB-3 and DUB-4 genes are highly related to DUB-1 and DUB-2 (32). All four DUB genes have two exons and encode proteins with extensive homology (approximately 90%) throughout their primary amino acid sequence. Interestingly, our genomic library screen with the full-length DUB-1 cDNA probe identified only DUB subfamily genes. Other ubp genes in the mouse genome, such as unp (29), have more distant DNA sequence homology and did not hybridize to this probe under our screening conditions. Therefore, an operational definition of a DUB gene is a ubp gene that hybridizes to the DUB-1 cDNA.

In order to estimate the number of DUB subfamily members, we performed a genomic Southern blot with the DUB-1 ORF probe (data not shown). Analysis of murine genomic DNA cut with multiple restriction enzymes revealed four to six major bands in each case. These data suggest that the DUB subfamily is small, perhaps consisting of only four to six gene members. The expression patterns of the other DUB genes remain unknown. These genes may be induced by other cytokines or may be pseudogenes.

The mouse chromosomal location of the DUB loci was determined by interspecific backcross analysis using progeny derived from matings of (C57BL/6J × M. spretus)F₁ × C57BL/6J mice. This interspecific backcross mapping panel has been typed for over 2200 loci that are well distributed among all the murine autosomes as well as the X chromosome (33). C57BL/6J and M. spretus DNAs were digested with several enzymes and analyzed by Southern blot hybridization for informative restriction fragment length polymorphisms, using various mouse DUB probes (see "Materials and Methods"). The mapping results were consistent using different probes, indicating that the DUB loci are tightly linked to each other and located in the central region of mouse chromosome 7, linked to Omp, Nup98, and Pth. Although 177 mice were analyzed for every marker and are shown in the segregation analysis (Fig. 5), up to 189 mice were typed for some pairs of markers. Each locus was analyzed in pairwise combinations for recombination frequencies using the additional data. The ratios of the total number of mice exhibiting recombinant chromosomes to the total number of mice analyzed for each pair of loci and the most likely gene order is: centromere-Omp-3/189-DUB 0/178-Nup98-6/179-Pth. The recombination frequencies (expressed as genetic distances in centimorgans ± the standard error) are: Omp, 1.6 ± 0.9; DUB, Nup98 - 3.3 ± 1.4 - Pth. No recombinants were detected between the DUB loci and Nup98 in 178 animals typed in common, suggesting that these loci reside within 1.7 centimorgans of each other (upper 95% confidence interval).

The central region of mouse chromosome 7, containing the DUB genes, shares a region of homology with human chromosome 11 (summarized in Fig. 5). In particular, Nup98 has been placed on human chromosome 11p15. The tight linkage between Nup98 and the DUB loci in mice suggests that the DUB loci will reside on human 11p as well. Interestingly, this region of human chromosome 11 is a frequent site of translocations in human leukemias (39, 40, 41) and is also a site of several tumor suppressor genes (42, 43, 44, 45).

That all polymorphic DUB gene restriction fragments mapped to mouse chromosome 7 is consistent with the hypothesis that the DUB genes are all closely linked and have evolved through a series of tandem gene duplication events. These mapping results indicate that the DUB gene subfamily is unlinked to other ubp members such as Tre2 (28) or unp (29, 30).

DISCUSSION

In the current work, we have cloned a novel deubiquitinating enzyme, DUB-2, that is rapidly induced by IL-2 in cytokine-dependent CTLL cells. The rapid induction and short half-life of the DUB-2 mRNA suggest that DUB-2 may play a regulatory role in the initial events of the IL-2-mediated growth response. The DUB-2 protein is highly related to the IL-3-inducible DUB-1 protein throughout its primary sequence and is more distantly related to other deubiquitinating enzymes.

We propose that DUB-1 and DUB-2 are members of a discrete subfamily of deubiquitinating enzymes, called the DUB subfamily. DUB subfamily members contain distinct structural features that distinguish them from other ubps. First, DUB subfamily members are comparatively small enzymes of approximately 500-550 amino acids. Second, DUB subfamily members share amino acid similarity not only in the Cys and His domains but also throughout their primary amino acid sequence. For instance, DUB proteins contain a lysine-rich region (Lys domain) and a HV domain near their carboxyl terminus.

On the basis of these structural criteria, additional members of the DUB subfamily can be identified in the GenBank^TM. For instance, one submission (accession number d38378) is a 529-amino acid human polypeptide with considerable homology (48% amino acid identity) to murine DUB-1 (Fig. 3A). The highest degree of homology is in the Cys and His domains. Additionally, this putative human DUB protein contains a Lys domain (amino acids 400-410) and a hypervariable region (amino acids 413-442).

Wilkinson et al. (22) have recently described six homology domains found in all ubp superfamily members. These six homology domains comprise the minimum catalytic core of the ubp. The catalytic core spans the sequence from the active site cysteine in the Cys domain to the carboxyl-terminal end of the His domain. For DUB enzymes, this minimum catalytic domain is 293 amino, making it the smallest functional ubp catalytic domain identified to date. In addition to the minimum catalytic core, ubps have NH₂- or COOH-terminal extensions. For instance, human isopeptidase T and its yeast homologue, ubpC, have a 300-amino acid NH₂-terminal extension. In contrast, DUB proteins have a COOH-terminal extension, containing the Lys domain and the HV region.

DUB subfamily members differ from other ubps by functional criteria as well. DUB subfamily members are cytokine-inducible, immediate-early genes and may therefore play regulatory roles in cellular growth or differentiation. Also, DUB proteins are unstable and are rapidly degraded by ubiquitin-mediated proteolysis shortly after their induction.⁴

Interestingly, at least two DUB genes (DUB-2 and DUB-3) map to the same region of murine chromosome 7. Their sequence similarity and chromosome co-localization suggest that the DUB genes arose by a tandem duplication of an ancestral DUB gene. Other polymorphic DUB gene restriction fragments also map to this region of mouse chromosome 7, suggesting the presence of a larger subfamily of DUB genes at this locus. DUB-3 and DUB-4 have a similar two exon structure and contain the Cys, His, Lys, and HV domains found in other DUBs. We predict that other cytokines induce the expression of DUB-3 and DUB-4.

The ubiquitin-mediated proteolytic pathway can be selectively modified by inducible protein components. For instance, interferon  induces the expression of two major histocompatibility class-encoded proteins, LMP2 and LMP7 (8, 9, 46) that serve as alternate  subunits of the proteasome. Replacement of the normal  subunits with LMP2 and LMP7 results in an altered proteasome that more efficiently cleaves viral proteins. In this way, interferon mediates an enhanced antiviral response, generating viral peptides that are more efficiently translocated to the endoplasmic reticulum and more efficiently bound to the peptide groove of the class I major histocompatibility class.

Our data demonstrate that specific cytokines, such as IL-2 and IL-3, induce specific deubiquitinating enzymes (DUBs). The DUB proteins may modify the ubiquitin-proteolytic pathway and thereby mediate specific cell growth or differentiation signals. These modifications are temporally regulated. The DUB-2 protein, for instance, is rapidly but transiently induced by IL-2. Interference of DUB enzymes with specific isopeptidase inhibitors may block specific cytokine signaling events.

The large number of deubiquitinating enzymes suggests that these proteases have narrow substrate specificity and are highly specific in their cellular functions. The hypervariable region at the COOH terminus of the DUB proteins further suggests a mechanism of substrate specificity. If DUB substrates are specific, several classes of substrates can be envisioned. First, DUB enzymes may deubiquitinate cell surface growth factor receptors, thereby prolonging receptor half-life and amplifying growth signals. At least some growth factor receptors, including the T cell receptor (47), the platelet-derived growth factor receptor (48), the c-kit protein (49), and the growth hormone receptor (50) are ubiquitinated. Ubiquitination of growth factor receptors is one known mechanism for regulating growth factor receptor surface expression (20). Second, DUB enzymes may deubiquitinate proteins involved in signal transduction. Recent studies have demonstrated that the signaling protein, c-cbl, is reversibly ubiquitinated (51). Also, the transcription factor, STAT-1, is ubiquitinated (52). Third, DUB enzymes may deubiquitinate cyclin-CDK inhibitors such as p27 (10). Deubiquitination of p27 at a post-proteasomal level could potentially increase degradation of p27, thereby driving cells from a resting state into a growing state (36, 53). The identification of specific substrate(s) of DUB enzymes will elucidate the biological role of the enzymes in regulating cell growth.