Perspective

Abstract

The addition of ubiquitin (Ub) and ubiquitin-like (Ubl) modifiers to proteins serves to modulate function and is a key step in protein degradation, epigenetic modification and intracellular localization. Deubiquitinating enzymes and Ubl-specific proteases, the proteins responsible for the removal of Ub and Ubls, act as an additional level of control over the ubiquitin-proteasome system. Their conservation and widespread occurrence in eukaryotes, prokaryotes and viruses shows that these proteases constitute an essential class of enzymes. Here, we discuss how chemical tools, including activity-based probes and suicide inhibitors, have enabled (i) discovery of deubiquitinating enzymes, (ii) their functional profiling, crystallographic characterization and mechanistic classification and (iii) development of molecules for therapeutic purposes.

Introduction

Post-translational modifiers can alter the function of proteins in many different ways. Covalent attachment of ubiquitin has been recognized as a key step in the specific destruction of proteins, whereas ubiquitin-like modifiers have more specialized roles in the regulation of nuclear factors (for example, SUMO1), in the fine tuning of ubiquitination (Nedd8) and in the cell's capability to destroy or recycle its own organelles (autophagy-related modifiers including ATG12 and others)^{1, 2}. The Ub/Ubl modifiers are evolutionarily ancient and share a similar fold (Fig. 1a,b). Ubiquitin, a 76-amino-acid protein, is covalently linked to lysine residues of substrate proteins in a multistep process. The manner of activation and conjugation is shared with that of the Ubl modifiers. Ubiquitin-activating enzymes (E1 enzymes) activate the C terminus of ubiquitin, which results in ATP-dependent thioester formation with the E1 active site cysteine (Fig. 1c). Ubiquitin is then transferred to the active site cysteine of an E2 conjugating enzyme and finally conjugated to its substrate by an E3 ligase, which confers substrate specificity. The human genome contains an estimated 500–600 Ub and Ubl ligases (a number comparable to the total number of predicted kinases³), which is consistent with a substrate-specific function in conjugation. Ub ligases, and their involvement in signaling pathways and targeting of proteins for degradation, have been reviewed recently^{2, 4}.

Figure 1: Ub/Ubl modifiers have a comparable fold and are activated and conjugated similarly.

(a) Ribbon diagram of ubiquitin. Ub and Ubls consist of a similar structural domain, either in the form of a monomer or as a dimer (for example, FAT10 and ISG15). (b) Cladogram representation of the human Ubl family based on amino acid sequences. The bootstrap consensus tree was calculated with a minimum evolution algorithm, using 100 iterations and the JTT substitution matrix. The proposed functional classification is simplified. For a more detailed discussion see ref. 2 and references therein. (c) The ubiquitin-proteasome pathway. (1) DUBs/ULPs are required to process Ub/Ubl precursors, thereby generating the active Ub/Ubl for conjugation. (2) Ub/Ubl ligation takes place in a three-step process using a series of ligases via enzyme thioester intermediates. Ub may be added to a substrate to generate mono-, multi- (several distinct lysine residues in a single protein are ubiquitinated) and polyubiquitinated species (a polymer of Ub created by its addition to Lys48 of Ub on another protein), which may directly regulate protein function. (3) DUBs are required for recycling of Ub from Ub-protein conjugates. (4) Many poly-Ub proteins are sent to the 26S proteasome for degradation. The 19S portion of the proteasome has both Ub-binding and DUB activity to aid in disassembly and unfolding of Ub substrates before proteolysis.

Full size image (65 KB)

Ub/Ubl ligases are counteracted by deubiquitinating enzymes (DUBs) and Ubl-specific proteases (ULPs)—proteases that serve to deconjugate the Ub/Ubl-modified substrates (Fig. 1c). In the case of DUBs, such functions serve to rescue substrate proteins from proteasomal degradation⁵, to recycle Ub (which is otherwise transcribed at relatively low levels^{6, 7}) or even to control protein trafficking⁸. DUBs and ULPs are central to the control of cell cycle regulatory proteins⁹ and can also act either as oncoproteins or as tumor supressors^{10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20}. Because Ub and Ubl modifiers are produced as inactive fusion proteins, DUBs and ULPs also generate the mature Ub and Ubl molecules (Fig. 1c), thereby adding an additional level of regulation. There are now approximately 100 human DUBs identified by in silico efforts and activity-based profiling^{21, 22}. These proteases belong to five distinct subclasses: (i) Ub C-terminal hydrolases (UCHs), (ii) Ub-specific proteases (USPs), (iii) Machado-Joseph disease protein domain proteases (MJDs), (iv) ovarian tumor proteases (OTUs) and (v) JAMM motif (zinc metallo) proteases. Two subclasses not found in mammals have also emerged: herpesvirus tegument USPs (htUSPs)²³ and the SUMO-specific ULP/SENP family (bacterial and primitive eukaryotes)²⁴, the latter being only distantly related to the mammalian ULP/SENP family^{24, 25}. A comparison of secondary structure included in a recent survey of known human DUBs (for both substrate specificity and function) shows remarkable diversity between the subclasses²¹. Though it is tempting to assign specificity of a DUB to Ub based on sequence similarity alone, the structural resemblance between Ub and Ubls by no means excludes specificity for a Ubl, and we shall expand on this point below. Because few studies systematically compare the action of a proposed DUB on modifiers other than Ub, it is not clear to what extent such assignments are correct.

While the importance of DUBs is well documented, our understanding of DUB substrate specificity and contribution to regulation of the Ub-proteasome system is comparatively poor. We stand to benefit most from studies that take a molecular approach toward identifying DUBs and ULPs and that confirm activity using Ub/Ubl-based probes. This strategy complements in silico methods by a more direct assessment of function. It further allows for recognition of proteases that have little homology to known DUBs/ULPs, but are nonetheless specific for Ub or a particular Ubl. We review here the chemical tools that have been useful in the analysis of DUBs and ULPs, including identification, mechanistic studies and inhibition by small molecules.

Activity-based protein profiling to identify DUBs

A variety of proteomic techniques, including LC-MS and protein micro-array analysis, have increased our understanding of protein expression and function in vitro, but determining the activity of proteins within a cell requires further efforts. Function-dependent methodologies are most useful in assessing biological activity profiles of expressed proteins in cells and tissues. This approach, termed activity-based protein profiling (ABPP), uses chemical probes directed at enzyme active sites to profile the activity of proteins in complex mixtures (ref. 26 and references therein). In addition to aiding in the identification of proteins with a particular enzymatic function, this type of profiling can paint a more complete picture of enzyme activity (as opposed to mere expression) under a particular set of physiological conditions. ABPP has been used to study a variety of enzyme classes, including proteases, kinases, phosphatases, glycosidases and oxidoreductases—a range of target activities that continues to expand²⁶.

ABPP has been used to identify new enzymes of the DUB/ULP family²². Epitope-tagged Ub (ref. 27) and Ubl (refs. 28,29) probes were synthesized to contain a variety of thiol-reactive groups as replacements for C-terminal glycine in the native Ub/Ubl structure (Fig. 2a). These probes included a vinyl methylsulfone, a vinyl methylester, several alkyl halides and a vinyl cyanide, all capable of behaving as suicide substrates when positioned within the DUB/ULP active site (Fig. 2b). Although the reactivity of these probes toward particular DUBs/ULPs does depend on the size and electrophilicity of the C-terminal group, variation in cellular expression levels of particular DUBs/ULPs remains a confounding variable.

Figure 2: ABPP approach for identifying DUBs.

(a) Synthesis of HAUb-derived probes using the intein-based chemical ligation. Recombinant HA-tagged Ub/Ubls are expressed with C-terminal intein-chitin binding domains for purification and introduction of reactive groups. Addition of -mercaptoethane sulfonic acid (MESNa) to the chitin column-bound fusion protein results in a thioester easily subject to chemical ligation, which generates the desired HAUb-containing probe. CBD, chitin-binding domain; CA, chitin agarose. (b) A table of the glycine-based electrophiles synthesized for attachment to the C terminus of HAUb probes. All probes listed have been conjugated with Ub (ref. 27). Ubls (Nedd8, SUMO1, IS615, URM1, Fau, GATE-16, MAP1-LC3, Apg8L and GABARAP) have been examined as VS conjugates^{28, 29}.

Full size image (43 KB)

After incubation of the hemagglutinin-tagged (HA) Ub-derived probes with cell lysates, protein complexes containing modified enzymes may be recovered via anti-HA immunoprecipitation, resolved by SDS-PAGE and visualized. Polypeptides are excised from the gel, trypsinized and analyzed by MS/MS. MS/MS data are then subjected to database searches against the US National Center for Biotechnology Information Expressed Sequence Tag (NCBI EST) databases. Using these first-generation probes, 23 DUBs were isolated and identified by MS/MS from mouse lymphoma cell lysates²⁷ (Table 1). Additionally, new gene products (Otubain) containing the ovarian tumor domain were identified as DUBs based on their reactivity with these probes²⁷. As most Ub ligases also contain an active site cysteine, one may predict that the design of new and more potent electrophilic probes will lead to the recovery of ligases, using an immunoprecipitation strategy similar to the one described above.

Table 1: DUBs identified from mouse and human cells

Full table

Profiling DUBs and ULPs in various tumors has uncovered remarkable differences in enzyme activity, which are not evident when comparing gene expression levels¹⁰. Using the power of ABPP to screen malignancies for their protease activity in vitro might be a future strategy for analyzing the susceptibility of tumors to cell-permeable protease inhibitors that are currently under development.

Activity-based probes also have proven instrumental in the crystallization of several DUBs. In the case of M48, a viral DUB (see below), crystallized protein was not obtained in the absence of a suicide inhibitor; covalent adduct formation was required for crystal formation²³. Addition of a Ub suicide substrate (UbVME) to UCH-L3 yielded better diffracting crystals that improved the structural resolution and with it the mechanistic understanding of the enzyme's active site³⁰.

So far, DUBs and ULPs have been analyzed only in conjunction with isopeptide-linked or linearly linked fusion proteins. More research is required to investigate the role of these proteases in the emerging group of thioester-modified proteins, in which Ub has been attached to a cysteine instead of a lysine. Enzymes such as E1, E2 and HECT (homologous to the E6-AP C terminus)-domain E3s are not the only molecules that can carry a thioester-linked Ub—even substrates may be modified to include a thioester-linked Ub (refs. 31,32). Bioinformatics suggests that conjugation of Ub to these two side chains is functionally distinct, as modified lysines avoid the vicinity of neighboring sulfhydryl groups³³. New classes of substrates are required to study differences in the chemical specificity of proteases toward isopeptide versus thioester bonds.

Pathogens express DUBs and ULPs

An ABPP approach combined with bioinformatic analysis was used to identify DUBs and ULPs in the viral family Herpesviridae^{34, 35}, in Chlamydia trachomatis²⁵ and other pathogenic bacteria^{24, 36, 37}, and in eukaryotic pathogens such as Plasmodium falciparum³⁸ and Toxoplasma gondii³⁹. Identification of a DUB within the N-terminal 500 residues of the large tegument protein of herpes simplex virus 1 (HSV-1) required a chemistry-based method, as the active DUB fragment identified bears no homology to known DUBs or Ub-binding proteins and initially escaped detection using a bioinformatics approach (Fig. 3)³⁴. Sequence alignment of the large tegument protein across the family Herpesviridae indicated the conservation of catalytic residues, and recombinantly expressed homologs from mouse cytomegalovirus (MCMV) and Epstein-Barr virus (EBV) show deubiquitinating activity in vitro³⁵. Crystallization of the DUB domain of MCMV M48 in complex with a Ub suicide substrate (UbVME) showed a unique binding mode with ubiquitin that relies on extensive hydrophobic contacts distinct from those observed for other DUBs²³.

Figure 3: Topological comparison of the M48^USP structure to other DUBs.

(a) Ribbon diagrams of M48^USP (Protein Data Bank (PDB) code 2J7Q) and representatives of the USP (HAUSP, PDB code 1NBF) and UCH (UCH-L3, PDB code 1XD3) families of DUBs are shown with bound inhibitors. In each case, the DUB is shown in blue with active site residues highlighted. Bound Ub-based suicide probes are shown in red. (b) Topology of the protein structures shown in a. -helices are indicated by rectangles, -strands by arrows. Secondary structural elements conserved among the three classes of DUBs are colored red (helices) or blue (strands). N and C termini are denoted by N and C, respectively, and catalytic residues are shown in magenta circles. For M48^USP and HAUSP, -strands contributing to either the -hairpin or finger domains are colored yellow. Diagrams for HAUSP and UCH-L3 were adapted with permission from ref. 92. (c) The sequential order of residues in the catalytic triad of each enzyme also shows the divergence of the herpesvirus family of DUBs, of which M48^USP is representative.

Full size image (77 KB)

The existence of DUBs in non-eukaryotic pathogens is noteworthy, as these organisms do not contain machinery for ubiquitin conjugation. This has led to speculation that DUBs exist in pathogens as a means for immune evasion or to manipulate the host in other ways^{40, 41, 42}. Unique inhibition of pathogen-specific DUBs could provide a novel means of therapy with minimal side effects on the host, given the phylogenetic distance between pathogen and host DUBs.

Evolution of DUBs and phylogenetic comparisons

Beyond enhancing our knowledge of the physiology of proteases, ABPP also represents a tool for phylogenetic studies. DUBs and ULPs are remarkably widespread among proteobacteria and Chlamydiae and were probably acquired from host cells by lateral gene transfer^{43, 44, 45}. An analysis of the catalytic core sequences of bacterial peptidases suggests an origin in the ULP/SENP family of proteases, which have distinct specificity for SUMO and Nedd8 in higher eukaryotes (Fig. 4). Several bacterial peptidases, however, also hydrolyze Ub-bound substrates. Notably, this apparent lack of substrate preference can also be observed in more deeply rooted eukaryotes²⁴. Though the origin of Ub and Ubls is still under debate, some argue that the high conservation of Ub indicates that this protein is the oldest and most important modifier in eukaryotes^{46, 47, 48}. It is possible that proteases with specificity for Ubls are derived from DUBs—the catalytic features consistent with this suggestion are still preserved in 'old' eukaryotes and in bacteria. The detailed analysis of prokaryote protease specificity not only allowed an estimate of when the genes encoding these enzymes were acquired by lateral gene transfer, but also showed that ancient eukaryote ULP/SENP proteases might have originated from a DUB.

Figure 4: Cladogram representation of proteases forming covalent adducts with Ub-based probes.

Most DUBs represented are mammalian and viral (C76) papain-like proteases. Some adenain-like proteases of the ULP/SENP family (previously believed to be specific for the Ubls SUMO (ref. 93) and Nedd8 (ref. 94)) were recently found capable of hydrolyzing Ub adducts. Examples include the SENP8 homolog of at least one fungus and several bacteria. Represented here are ElaD from Escherichia coli and ChlaDUBs from pathogenic Chlamydiae.

Full size image (49 KB)

DUBs and ULPs have been subject to evolutionary change, and they have also shaped the sequences of their targets—Ub and Ubls. For instance, several Ubls end with a diglycine motif at their C terminus. Making a terminal G76A substitution does not affect the conjugation of Ub (or SUMO) to substrates. This mutation does, however, render the modifier more resistant to proteolysis—in other words, Ub or Ubls ending with Gly-Ala instead of Gly-Gly result in irreversible modification^{49, 50}. The fact that this mutation rarely occurs in nature is a testimony to the importance of DUBs and ULPs, and it highlights their role in the dynamics of this post-translational modification. Evolution has witnessed a recent explosion in the occurrence of Ub fusion proteins, in which a Ub domain is linked to the termini of enzymes^{51, 52}. Even though these domains originated from Ub, they are not processed by DUBs—possibly because of inaccessibility or random mutations. The failure to remove the Ub moiety from a fusion protein exempts this domain from the selective pressure to which it is otherwise subjected. This can lead to drastic deterioration of the amino acid sequence during evolution, as exemplified by uncleavable Ub precursors in some organisms^{53, 54}.

Using Ub probes to study DUB mechanisms and kinetics

The mechanism by which DUBs hydrolyze amide bonds has been examined extensively using Ub-based probes⁵⁵. Mechanistic studies began with the realization that UCH-type DUBs can be inactivated by the addition of sodium borohydride, but only when Ub is present⁵⁶. This inhibition is due to the generation of Ub-aldehyde (Ub-al), and accordingly the mechanism that underlies hydrolysis by a DUB was interpreted as one of nucleophilic catalysis requiring an acyl-Ub-enzyme intermediate^{56, 57}. Ubiquitin ethyl ester was similarly used to characterize UCH-type DUB mechanisms⁵⁸. More recent studies using a variety of fluorescent Ub-based probes (discussion below) have shown that hydrolysis is a three-step process that consists of DUB-substrate binding, generation of an acyl-enzyme intermediate and hydrolysis of the acyl-enzyme intermediate to liberate Ub and free enzyme.

The use of peptide-aminomethylcoumarin (AMC) substrates based on the C-terminal residues of ubiquitin showed that USP isopeptidase T (IPaseT; also known as IsoT or USP5) is both activated and inhibited by Ub in a concentration-dependent manner⁵⁹. This led to the postulation that IPaseT has two binding sites for Ub, an "activation site" and an "inhibitory site," that contribute to its regulation. Hydrolysis of full-length Ub-AMC had distinctly different kinetics than that of truncated versions, showing that DUBs can use the free energy released from remote interactions with Ub-containing substrates for transition-state stabilization⁶⁰. The crystal structures of UCH-L3, both unliganded and in a complex with a suicide inhibitor, suggest a mechanism in which binding of the substrate induces a change in the active site crossover loop that stabilizes the C terminus of Ub, thus allowing hydrolysis of the UCH-L3-Ub covalent intermediate^{30, 61}.

Another cost-saving method for assaying DUB kinetics is the use of a linear fusion of polyHis-glutathione-S-transferase-Ub-ecotin (His-GST-Ub-ecotin) as a substrate⁶². The activity of DUBs is measured indirectly by determining the ability of ecotin, which is released by DUB activity, to inhibit trypsin. USPs, including USP69 from rat muscle, USP46 from chick muscle and USP7(HAUSP) from mouse and human, were assayed using this method. A potential limitation to Ub-ecotin fusions for assaying DUB activity is that these constructs cannot be used to detect the activity of UCH family members, as UCHs generally are ineffective at cleaving Ub fusions with extensions larger than 20 kDa. Given that only four mammalian UCH-type DUBs have been characterized so far (out of more than 80 functional DUBs), the limitation of this assay is not a major one.

The importance of specific Ub side chains during the cleavage reaction catalyzed by UCH-L1 was investigated by directed alanine scanning⁶³. C-terminal Ub fusions containing naturally fluorescing tryptophan residues (UbW and UbAW) were generated and tested as inhibitors of the reaction between Ub-AMC and UCH-L1. Based on previous mutational analysis and crystallographic experiments with UCH-L3, mutagenesis of UbW was focused to specific regions of Ub that potentially interface with UCH-L1. Of the 15 Ub variants generated, only three substitutions had a significant effect on the kinetics of complexation between UbW and UCH-L1. Several variants with minor effects on UCH-L1 kinetics were accepted as wild-type substrates for UCH-L3, which suggests that these two DUBs have subtly distinct interfaces with their substrates and that identifying selective inhibitors for a particular enzyme may be possible even between closely related proteases (UCH-L1 and UCH-L3 are 84% identical). These data show an interesting parallel with the systematic mutational analysis undertaken by Hicke et al., who showed that many residues of Ub tolerate substitution without loss of function⁶⁴.

Truncated versions of the C-terminally modified Ub and Ubl probes were used by Borodovsky et al. to examine the peptide length requirement for binding and inhibition of DUBs and ULPs (ref. 65). A series of peptide vinyl sulfones containing various portions of the Ub/Ubl C terminus were synthesized using Kenner's safety-catch linker and standard Fmoc-based solid-phase peptide synthesis. Inhibitory activity was achieved using a minimum of the 12 C-terminal Ub residues. Extension of these peptides to a length of 14 to 18 residues did not improve inhibitory activity. In the case of the Ubl SUMO1, a five-amino-acid probe was sufficient to achieve selective targeting of SUMO1 protease in cell lysate, but failed to label recombinant UCH-L3. This underscores the selectivity embedded in the C-terminal residues of the Ubl modifiers.

Kinetic assays for HTS of DUBs/ULPs

As described above, Ub-AMC is often used to test the enzymatic activity of a purified DUB. Ub-AMC is less suitable, however, in high-throughput screening (HTS) of DUBs for potential inhibitors, because of the short excitation wavelength (known to excite a number of compounds present in such screens) and the absence of a Ub linked via an -amino group, as found under physiologic conditions (whether this is a purely formal objection must be settled by experiment). Synthesis of fluorogenic Ub substrates via conjugation of -NH₂-tetramethylrhodamine-lysine has allowed the development of new assays⁶⁶. The Ub variant containing the large N-terminal NusA tag and the C-terminal fluorophore currently provides the highest dynamic range in fluorescence polarization readout and is being used to identify inhibitors of UCH-L3 and USP2, enzymes claimed to be involved in tumor growth promotion. A time-resolved fluorescence resonance energy transfer (FRET)-based assay using a yellow fluorescent protein–labeled Ub fusion containing a C-terminal terbium group also may be suitable for the identification of UCH-L3 inhibitors⁶⁷.

Developing small-molecule inhibitors for DUBs

There is increasing evidence that links ubiquitination to certain pathologies. Deficiencies in E3 ligases have been noted to cause a variety of disorders, including Angelman's syndrome (E6AP)⁶⁸, Parkinson's disease (Parkin)⁶⁹ and even cancer (BRCA1/BARD1)⁷⁰. Compromised DUB activity can also contribute to disease, as a result of either a loss of function or a gain of function. Some currently known examples of DUBs implicated in diseases are CYLD (refs. 17,71), a tumor suppressor that prevents cylindromatosis, UCH-L1 (refs. 72,73), which has been linked to Parkinson's disease, and USP14 (refs. 74,75) and Ataxin-3 (refs. 76,77), both of which are linked to ataxias. Development of small molecules for the treatment of these and other conditions arising from faulty DUBs is unlikely to be successful when a restoration of function is needed, and it remains challenging in cases in which function needs attenuation.

Given that Ub-al, like most proteins, is not suitable for studies in intact cell or animal models because it is not cell-permeable as such, development and implementation of high-throughput assays to screen libraries of small molecules for potential cell-permeable DUB inhibitors is the obvious alternative. Screening of 42,000 drug-like compounds led to the identification of a class of O-acyloxime derivatives of isatins as UCH-L1 inhibitors⁷⁸. Examination of structure-activity relationships allowed the development of compounds with improved inhibitory potency and selectivity for UCH-L1 over UCH-L3 (Fig. 5a), with compounds 1 and 3 showing a strong preference for UCH-L1. Treatment of lung tumor cell lines with these inhibitors indicated that UCH-L1 activity is antiproliferative. Although UCH-L1 is considered an enzyme most highly expressed in neuronal tissue, its elevated expression has been noted also in malignancies of hematologic origin¹⁰. During the screening process, some compounds (4–6) showed as much as 100-fold selectivity for UCH-L3 over UCH-L1 (ref. 78).

Figure 5: Small-molecule inhibitors of DUBs.

(a) Improvement on isatins identified in an initial screen of 42,000 compounds led to the development of 1–3. These three inhibitors show a strong preference for UCH-L1 (IC₅₀ values in black) over UCH-L3 (IC₅₀ values in red). Three compounds (4–6) from the initial screen were selective for UCH-L3 (IC₅₀ values in red) over UCH-L1 (IC₅₀ values in black)—by as much as 100-fold in the case of 6. (b) UCH-L1 inhibitor LDN-91946 (7). (c) Chemical structures of cyclopentenone prostaglandins (8–11). ¹²-PGJ₂ (11) inhibits cellular ubiquitin isopeptidase activity. (d) Chemical structures of punaglandins (12–15). Punaglandins have been examined for enhanced ubiquitin isopeptidase inhibition and antiproliferative effects relative to J- and A-series prostaglandins. IC₅₀, half-maximal inhibitory concentration.

Full size image (73 KB)

An HTS campaign at Harvard's Laboratory for Drug Discovery in Neurodegeneration identified the first reversible, low-molecular-weight inhibitor of UCH-L1 (Fig. 5b). LDN-91946 (7) is a heteroaryl carboxylic acid with a K_i,app of 2.3 M for inhibiting UCH-L1 (ref. 79). Kinetic studies indicate an uncompetitive mechanism (K_i = 2.8 M) and suggest that the inhibitor binds only to the Michaelis complex and not to free enzyme. LDN-91946 is a selective inhibitor, showing K_i,app values in excess of 100 M toward other enzymes with active site cysteine residues, including caspase-3, tissue transglutaminase, UCH-L3 and isopeptidase T. A program of medicinal chemical optimization is likely to yield even better results.

Proteasome inhibitors also have received much attention, as protein degradation is an essential component of cell cycle progression and regulation and an immediate downstream consequence of Ub addition. Proteasome inhibitors, though useful for the study of the components comprising protein degradation, are also quite toxic. Observed toxicities could be mitigated by inhibition of a particular DUB. J-series prostaglandins (8–11) were inferred to preferentially inhibit ubiquitin isopeptidase activity of the 26S proteasome pathway^{80, 81} (Fig. 5c) based on slowed polyubiquitin disassembly in treated cell lysates. After this result, the antiproliferative effects of punaglandins (12–15)—highly functionalized prostaglandins chlorinated at the endocyclic -carbon—were examined⁸² (Fig. 5d). The punaglandins were more potent than either A- or J-series prostaglandins against human colon tumor cells; thus, they are potentially a new class of anticancer therapeutics. A discussion of the very active field of proteasome inhibitors is beyond the scope of this review (see ref. 83 and references therein). Because of the intimate interactions between Ub addition, removal and proteasomal proteolysis, these areas have great potential for synergy where manipulation with small-molecule inhibitors is concerned.

Concluding remarks and future perspectives

As large-scale DNA sequencing projects have led to the identification of open reading frames encoding newly discovered proteins, many scientific databases contain more proteins of unknown function than of known function. In an effort to fully annotate genomes, the determination of function of unknown enzymes in various organisms is a priority. Functional genomics is an important strategy in deciphering the functions of a variety of enzymes, in particular DUBs and ULPs. C-terminally modified Ub/Ubl probes have helped identify DUBs and ULPs from various organisms, even those that themselves do not contain the complete machinery of the ubiquitin-proteasome system. Although many DUBs and ULPs can be identified solely on the basis of sequence alignment with other family members, some DUBs and ULPs, such as the herpesviral M48, may be identified only using activity-based profiling. Ub/Ubl-based probes also interact with and irreversibly inhibit Ub/Ubl ligases. An ABPP approach can confirm the activity for both DUBs/ULPs and ligases already assigned a particular function based on bioinformatic analysis. New probes containing a more diverse set of C-terminal electrophiles may help to ferret out nuances of the active sites of DUBs in terms of both selectivity and reactivity. The possibility of combining cell-permeable peptides (ref. 84 and references therein) with Ub- or Ubl-derived probes may ultimately allow the extension of ABPP of DUBs and ULPs to live cells.

Crystallographic and mechanistic studies complement functional assays of DUBs. Ub-based probes facilitate the crystallization of particular DUBs and improve our understanding of the DUB active site. Crystal structures of DUBs with probes bound in the active site have clarified reorganizations required during isopeptide bond cleavage. Crystal structures of DUBs complexed with very short truncations of Ub as suicide inhibitors may even serve to make clear the positioning of the isopeptide bond in the DUB active site and could provide additional leads that might yield cell-permeable and cell-specific inhibitors.

Ub-based fluorophores have been developed and used in assays to examine enzyme kinetics and for HTS of small molecules as inhibitors. HTS has identified several molecular scaffolds that upon further modification have yielded specific inhibitors of UCH-type DUBs. Further optimization of these compounds may segue to DUB/ULP inhibitors that are effective as therapeutics, as DUBs and ULPs are involved in a variety of disease states including cancer and neurodegenerative conditions. The divergence of DUBs and ULPs in non-eukaryotic pathogens from their eukaryotic counterparts makes the microbial enzymes particularly attractive targets for treatment of infectious diseases.

One outstanding problem in the field is that of DUB substrate identification. Until recently, substrates used to examine DUBs, both for kinetic and mechanistic analysis, have been primarily linear C-terminal extensions. The new development of a photolytically removable ligation auxiliary for the site-specific ubiquitination of lysine residues in a target peptide will allow for the investigation of potential substrates with native, isopeptide fusions between Ub and a peptide of desired length (Fig. 6)^{85, 86}. This technique is a significant improvement over existing enzymatic methods for isopeptide substrate synthesis, which consist of mixing Ub with the recipient peptide in the presence of appropriate E1 and E2 ligases to achieve the native linkage³⁰. One can now imagine the synthesis of DUB substrate libraries using solid-phase peptide synthesis to generate a peptide library and expressed protein ligation to link the library members to Ub. Taking cues from the successful substrate profiling of other cysteine proteases^{87, 88}, expressed protein ligation could be combined with positional scanning of the residues flanking the isopeptide linkage, potentially leading to the identification of unique specificities of particular DUBs and aiding in inhibitor development.

Figure 6: Ligation of an auxiliary-containing peptide with a C-terminally truncated Ub-thioester to achieve isopeptide-linked DUB substrates.

Treatment of the ligated product with UV light allows the removal of the auxiliary. Figure adapted with permission from ref. 85.

Full size image (26 KB)

Finally, one of the most exciting outlooks is the possibility that DUBs and ULPs might recognize and cleave substrates unrelated to Ub or Ubls. In the absence of straightforward assays that might help demonstrate such off-target specificity, a few reports on autocleavage of DUBs have emerged^{89, 90}—and they might just be the tip of the iceberg. As an extreme example, if we assume that the Gly-Gly motif at the C terminus of Ub/Ubls is necessary for cleavage by the majority of DUBs and ULPs, there could be more than 40,000 putative substrates in the human proteome bearing this mini-motif. One could envision using activity-based probes not to hunt for proteases with a known substrate, but to hunt for DUB/ULP specificity with a random pool of peptide substrates. A split-mix synthesized library of electrophilic peptides could be used to analyze off-target specificity of a defined DUB/ULP, following MS/MS identification of the probe used as a bootstrap for identification of new physiological substrates.

Acknowledgments

We thank T. DiCesare (Whitehead Institute Bioinformatics & Research Computing Group) for help with the figures. K.R.L. is a US National Institutes of Health postdoctoral fellow (F32 AI63854). C.S. is supported by a European Molecular Biology Organization long-term fellowship (187-2005).

References

1. Kirkin, V. & Dikic, I. Role of ubiquitin- and Ubl-binding proteins in cell signaling. Curr. Opin. Cell Biol. 19, 199–205 (2007). | Article | PubMed | ISI | ChemPort |
2. Kerscher, O., Felberbaum, R. & Hochstrasser, M. Modification of proteins by ubiquitin and ubiquitin-like proteins. Annu. Rev. Cell Dev. Biol. 22, 159–180 (2006). | Article | PubMed | ISI | ChemPort |
3. Wilkinson, K.D., Ventii, K.H., Friedrich, K.L. & Mullally, J.E. The ubiquitin signal: assembly, recognition and termination - symposium on ubiquitin and signaling. EMBO Rep. 6, 815–820 (2005). | Article | PubMed | ISI | ChemPort |
4. Pickart, C.M. & Eddins, M.J. Ubiquitin: structures, functions, mechanisms. Biochim. Biophys. Acta 1695, 55–72 (2004). | Article | PubMed | ISI | ChemPort |
5. Rumpf, S. & Jentsch, S. Functional division of substrate processing cofactors of the ubiquitin-selective Cdc48 chaperone. Mol. Cell 21, 261–269 (2006). | Article | PubMed | ISI | ChemPort |
6. Swaminathan, S., Amerik, A.Y. & Hochstrasser, M. The Doa4 deubiquitinating enzyme is required for ubiquitin homeostasis in yeast. Mol. Biol. Cell 10, 2583–2594 (1999). | PubMed | ISI | ChemPort |
7. Ryu, K.Y. et al. The mouse polyubiquitin gene UbC is essential for fetal liver development, cell-cycle progression and stress tolerance. EMBO J. 26, 2693–2706 (2007). | Article | PubMed | ISI | ChemPort |
8. Nikko, E. & Andre, B. Evidence for a direct role of the Doa4 deubiquitinating enzyme in protein sorting into the MVB pathway. Traffic 8, 566–581 (2007). | Article | PubMed | ISI | ChemPort |
9. Stegmeier, F. et al. Anaphase initiation is regulated by antagonistic ubiquitination and deubiquitination activities. Nature 446, 876–881 (2007). | Article | PubMed | ISI | ChemPort |
10. Ovaa, H. et al. Activity-based ubiquitin-specific protease (USP) profiling of virus-infected and malignant human cells. Proc. Natl. Acad. Sci. USA 101, 2253–2258 (2004). | Article | PubMed | ChemPort |
11. Machida, Y.J. et al. UBE2T is the E2 in the Fanconi anemia pathway and undergoes negative autoregulation. Mol. Cell 23, 589–596 (2006). | Article | PubMed | ISI | ChemPort |
12. Nicholson, B., Marblestone, J.G., Butt, T.R. & Mattern, M.R. Deubiquitinating enzymes as novel anticancer targets. Future Oncol. 3, 191–199 (2007). | Article | PubMed | ChemPort |
13. Brooks, C.L. & Gu, W. p53 ubiquitination: Mdm2 and beyond. Mol. Cell 21, 307–315 (2006). | Article | PubMed | ISI | ChemPort |
14. Stevenson, L.F. et al. The deubiquitinating enzyme USP2a regulates the p53 pathway by targeting Mdm2. EMBO J. 26, 976–986 (2007). | Article | PubMed | ISI | ChemPort |
15. Graner, E. et al. The isopeptidase USP2a regulates the stability of fatty acid synthase in prostate cancer. Cancer Cell 5, 253–261 (2004). | Article | PubMed | ISI | ChemPort |
16. Janssen, J.W.G., Schleithoff, L., Bartram, C.R. & Schulz, A.S. An oncogenic fusion product of the phosphatidylinositol 3-kinase p85 beta subunit and HUMORF8, a putative deubiquitinating enzyme. Oncogene 16, 1767–1772 (1998). | Article | PubMed | ISI | ChemPort |
17. Trompouki, E. et al. CYLD is a deubiquitinating enzyme that negatively regulates NF-kappa B activation by TNFR family members. Nature 424, 793–796 (2003). | Article | PubMed | ISI | ChemPort |
18. Brummelkamp, T.R., Nijman, S.M.B., Dirac, A.M.G. & Bernards, R. Loss of the cylindromatosis tumour suppressor inhibits apoptosis by activating NF-kappa B. Nature 424, 797–801 (2003). | Article | PubMed | ISI | ChemPort |
19. Kovalenko, A. et al. The tumour suppressor CYLD negatively regulates NF-kappa B signalling by deubiquitination. Nature 424, 801–805 (2003). | Article | PubMed | ISI | ChemPort |
20. Jensen, D.E. & Rauscher, F.J. Defining biochemical functions for the BRCA1 tumor suppressor protein: analysis of the BRCA1 binding protein BAP1. Cancer Lett. 143, S13–S17 (1999). | Article | PubMed | ISI | ChemPort |
21. Nijman, S.M.B. et al. A genomic and functional inventory of deubiquitinating enzymes. Cell 123, 773–786 (2005). | Article | PubMed | ISI | ChemPort |
22. Hemelaar, J. et al. Chemistry-based functional proteomics: mechanism-based activity-profiling tools for ubiquitin and ubiquitin-like specific proteases. J. Proteome Res. 3, 268–276 (2004). | Article | PubMed | ISI | ChemPort |
23. Schlieker, C. et al. Structure of a herpesvirus-encoded cysteine protease reveals a unique class of deubiquitinating enzymes. Mol. Cell 25, 677–687 (2007). | Article | PubMed | ISI | ChemPort |
24. Catic, A., Misaghi, S., Korbel, G.A. & Ploegh, H.L. ElaD, a deubiquitinating protease expressed by E. coli. PLoS ONE 2, e381 (2007). | Article | PubMed | ChemPort |
25. Misaghi, S. et al. Chlamydia trachomatis-derived deubiquitinating enzymes in mammalian cells during infection. Mol. Microbiol. 61, 142–150 (2006). | Article | PubMed | ISI | ChemPort |
26. Evans, M.J. & Cravatt, B.F. Mechanism-based profiling of enzyme families. Chem. Rev. 106, 3279–3301 (2006). | Article | PubMed | ISI | ChemPort |
27. Borodovsky, A. et al. Chemistry-based functional proteomics reveals novel members of the deubiquitinating enzyme. Chem. Biol. 9, 1149–1159 (2002). | Article | PubMed | ISI | ChemPort |
28. Hemelaar, J., Lelyveld, V.S., Kessler, B.M. & Ploegh, H.L. A single protease, Apg4B, is specific for the autophagy-related ubiquitin-like proteins GATE-16, MAP1–LC3, GABARAP, and Apg8L. J. Biol. Chem. 278, 51841–51850 (2003). | Article | PubMed | ISI | ChemPort |
29. Hemelaar, J. et al. Specific and covalent targeting of conjugating and deconjugating enzymes of ubiquitin-like proteins. Mol. Cell. Biol. 24, 84–95 (2004). | Article | PubMed | ISI | ChemPort |
30. Misaghi, S. et al. Structure of the ubiquitin hydrolase UCH-L3 complexed with a suicide substrate. J. Biol. Chem. 280, 1512–1520 (2005). | PubMed | ISI | ChemPort |
31. Cadwell, K. & Coscoy, L. Ubiquitination on nonlysine residues by a viral E3 ubiquitin ligase. Science 309, 127–130 (2005). | Article | PubMed | ISI | ChemPort |
32. Ravid, T. & Hochstrasser, M. Autoregulation of an E2 enzyme by ubiquitin-chain assembly on its catalytic residue. Nat. Cell Biol. 9, 422–427 (2007). | Article | PubMed | ISI | ChemPort |
33. Catic, A., Collins, C., Church, G.M. & Ploegh, H.L. Preferred in vivo ubiquitination sites. Bioinformatics 20, 3302–3307 (2004). | Article | PubMed | ISI | ChemPort |
34. Kattenhorn, L.M., Korbel, G.A., Kessler, B.M., Spooner, E. & Ploegh, H.L. A deubiquitinating enzyme encoded by HSV-1 belongs to a family of cysteine proteases that is conserved across the family Herpesviridae. Mol. Cell 19, 547–557 (2005). | Article | PubMed | ISI | ChemPort |
35. Schlieker, C., Korbel, G.A., Kattenhorn, L.M. & Ploegh, H.L. A deubiquitinating activity is conserved in the large tegument protein of the Herpesviridae. J. Virol. 79, 15582–15585 (2005). | Article | PubMed | ISI | ChemPort |
36. Rytkonen, A. et al. SseL, a Salmonella deubiquitinase required for macrophage killing and virulence. Proc. Natl. Acad. Sci. USA 104, 3502–3507 (2007). | Article | PubMed | ChemPort |
37. Zhou, H.L. et al. Yersinia virulence factor YopJ acts as a deubiquitinase to inhibit NF-kappa B activation. J. Exp. Med. 202, 1327–1332 (2005). | Article | PubMed | ISI | ChemPort |
38. Artavanis-Tsakonas, K. et al. Identification by functional proteomics of a deubiquitinating/deNeddylating enzyme in Plasmodium falciparum. Mol. Microbiol. 61, 1187–1195 (2006). | Article | PubMed | ISI | ChemPort |
39. Frickel, E.M. et al. Apicomplexan UCHL3 retains dual specificity for ubiquitin and Nedd8 throughout evolution. Cell. Microbiol. 9, 1601–1610 (2007). | Article | PubMed | ISI | ChemPort |
40. Sulea, T., Lindner, H.A. & Menard, R. Structural aspects of recently discovered viral deubiquitinating activities. Biol. Chem. 387, 853–862 (2006). | Article | PubMed | ISI | ChemPort |
41. Angot, A., Vergunst, A., Genin, S. & Peeters, N. Exploitation of eukaryotic ubiquitin signaling pathways by effectors translocated by bacterial type III and type IV secretion systems. PLoS Pathog. 3, e3 (2007). | Article | PubMed | ChemPort |
42. Rytkönen, A. & Holden, D.W. Bacterial interference of ubiquitination and deubiquitination. Cell Host Microbe 1, 13–22 (2007). | ChemPort |
43. Horn, M. et al. Illuminating the evolutionary history of Chlamydiae. Science 304, 728–730 (2004). | Article | PubMed | ISI | ChemPort |
44. Cazalet, C. et al. Evidence in the Legionella pneumophila genome for exploitation of host cell functions and high genome plasticity. Nat. Genet. 36, 1165–1173 (2004). | Article | PubMed | ISI | ChemPort |
45. Wolf, Y.I., Aravind, L. & Koonin, E.V. Rickettsiae and Chlamydiae: evidence of horizontal gene transfer and gene exchange. Trends Genet. 15, 173–175 (1999). | Article | PubMed | ISI | ChemPort |
46. Graciet, E. et al. From the cover: aminoacyl-transferases and the N-end rule pathway of prokaryotic/eukaryotic specificity in a human pathogen. Proc. Natl. Acad. Sci. USA 103, 3078–3083 (2006). | Article | PubMed | ChemPort |
47. Iyer, L., Burroughs, A.M. & Aravind, L. The prokaryotic antecedents of the ubiquitin-signaling system and the early evolution of ubiquitin-like beta-grasp domains. Genome Biol. 7, R60 (2006). | Article | PubMed | ChemPort |
48. Xu, J. et al. Solution structure of Urm1 and its implications for the origin of protein modifiers. Proc. Natl. Acad. Sci. USA 103, 11625–11630 (2006). | Article | PubMed | ChemPort |
49. Hodgins, R.R.W., Ellison, K.S. & Ellison, M.J. Expression of a ubiquitin derivative that conjugates to protein irreversibly produces phen