The Lysine 48 and Lysine 63 Ubiquitin Conjugates Are Processed Differently by the 26 S Proteasome*

The role of Lys-63 ubiquitin chains in targeting proteins for proteasomal degradation is still obscure. We systematically compared proteasomal processing of Lys-63 ubiquitin chains with that of the canonical proteolytic signal, Lys-48 ubiquitin chains. Quantitative mass spectrometric analysis of ubiquitin chains in HeLa cells determines that the levels of Lys-63 ubiquitin chains are insensitive to short-time proteasome inhibition. Also, the Lys-48/Lys-63 ratio in the 26 S proteasome-bound fraction is 1.7-fold more than that in the cell lysates, likely because some cellular Lys-63 ubiquitin conjugates are sequestered by Lys-63 chain-specific binding proteins. In vitro, Lys-48 and Lys-63 ubiquitin chains bind the 26 S proteasome comparably, whereas Lys-63 chains are deubiquitinated 6-fold faster than Lys-48 chains. Also, Lys-63 tetraubiquitin-conjugated UbcH10 is rapidly deubiquitinated into the monoubiquitinated form, whereas Lys-48 tetraubiquitin targets UbcH10 for degradation. Furthermore, we found that both the ubiquitin aldehyde- and 1,10-phenanthroline-sensitive deubiquitinating activities of the 26 S proteasome contribute to Lys-48- and Lys-63-linkage deubiquitination, albeit the inhibitory extents are different. Together, our findings suggest that compared with Lys-48 chains, cellular Lys-63 chains have less proteasomal accessibility, and proteasome-bound Lys-63 chains are more rapidly deubiquitinated, which could cause inefficient degradation of Lys-63 conjugates.

Protein ubiquitination is a posttranslational modification catalyzed by a cascade of enzymatic reactions involving a ubiquitin (Ub) 4 -activating enzyme (E1), a Ub-conjugating enzyme (E2), and a Ub ligase (E3) (1). Ub is conjugated onto protein substrates by formation of an isopeptide bond between the carboxyl group of the C-terminal glycine residue of Ub and the ⑀-amino group of a lysine residue in the substrate. Furthermore, a polyubiquitin (polyUb) chain is formed by conjugating the carboxyl group of the C-terminal glycine residue of Ub to the ⑀-amino group of one of the seven internal lysines in the preceding Ub. In addition, linear polyUbs are linked by amide bonds formed between the C-terminal glycine residue of Ub and the N-terminal methionine residue of a following Ub (2). Thus, at least eight different polyUb linkages exist in cells.
Protein ubiquitination plays diverse roles in regulating cellular activities. Monoubiquitination does not support degradation, but it is involved in regulating membrane trafficking, gene transcription, DNA repair, and DNA replication (3). As for polyubiquitination, it seems that different linkages have distinct functions; polyUbs linked through Lys-48 are the primary targeting signals for proteasomal degradation (1), whereas polyUbs linked through Lys-63 recruit other binding partners and execute many functions including kinase activation (4), protein synthesis (5), DNA repair (6,7), and chromosome segregation (8). The functions of other polyUbs have also been suggested such as the involvement of the Lys-6 linkage in regulating DNA repair (9). Interestingly, the Ub linkage on a modified substrate can be switched in response to different functions. For instance, in tumor necrosis factor ␣-stimulated activation of the NFB gene (10), a Lys-63 chain on receptor interacting protein (RIP) functions as a signaling element to recruit the IB kinase complex, which phosphorylates IB␣ (an inhibitor protein of NFB) and triggers its Ub-dependent degradation. During the stimulation, A20 deubiquitinates Lys-63linked RIP and then assembles Lys-48 polyUbs on RIP, promoting RIP degradation (10). In addition to Lys-48 polyUbs, a recent proteomic study found that Ub chains linked by Lys-6, -11, -27, -29, or -33 could also serve as proteolytic signals (11). Lys-11 polyUbs were found to mediate degradation of proteins involved in endoplasmic reticulum-associated degradation, cell cycle progression, and other functions (11,12), whereas Lys-29 polyUbs may promote Ub fusion degradation (13).
The role of Lys-63 polyUbs in targeting proteins for proteasomal degradation is still unclear. Some studies suggested that Lys-63 polyUbs are competent proteolytic signals. For example, in vitro studies have shown that Lys-63 polyUbs are able to target degradation of several proteins including Sic1, cyclin B1, dihydrofolate reductase, and troponin I (14 -17). In Saccharomyces cerevisiae, partial degradation of the transcription factor Mga2, which releases the N-terminal p90 activator domain, can be processed by overexpression of the UbLys-48R mutant that promotes the formation of Lys-63 ubiquitinated Mga2 (14). Also, inhibition of the proteasome by MG132 in S. cerevisiae or mammalian cells causes an increase of both the Lys-48 and Lys-63 Ub conjugates as detected by mass spectrometric analysis (14). However, to our knowledge, physiological substrates that exclusively depend on the Lys-63 linkage for degradation have not been identified. In contrast to the findings that suggest a role of Lys-63 polyUbs in targeting proteolysis, Xu et al. (11) proposed that the Lys-63 polyUbs are not proteolytic signals in S. cerevisiae based on quantitative proteomic studies. They also suggest that all other Ub linkages can support degradation and have partially redundant functions in proteolysis (11).
In this study we systematically compared proteasomal processing of Lys-63 polyUbs with that of the primary proteolytic signal of Lys-48 polyUbs in the aspects of binding/recognition, deubiquitination, and targeting for degradation. Our results suggest that cellular Lys-63 Ub chains have less proteasomal accessibility than Lys-48 chains, likely because some cellular Lys-63 ubiquitin conjugates are sequestered by Lys-63 chainspecific binding proteins, such as NEMO. In vitro Lys-63 and Lys-48 Ub chains bind the 26 S proteasome comparably, but Lys-63 chains are deubiquitinated 6-fold more rapidly than Lys-48 chains. Both the ubiquitin aldehyde (Ubal)-and 1,10phenanthroline-sensitive deubiquitinating activities of the 26 S proteasome contribute to Lys-48-and Lys-63-linkage deubiquitination, albeit their inhibitory extents are different. Moreover, we found that rapid deubiquitination of Lys-63 chains could cause inefficient degradation of their conjugates.

Reagents, Plasmids, Recombinant Protein Purification, and
Ubiquitination of UbcH10-See supplemental "Experimental Procedures." Proteasomal Degradation and Deubiquitination Assays-The bovine 26 S proteasome and PA700 were purified according to methods established by DeMartino and co-workers (18,19). Proteasomal degradation and deubiquitination were performed at 30°C in degradation buffer (20 mM Tris, pH 7.2, 20 mM NaCl, 5 mM MgCl 2 , 2 mM ATP, 1 mM ␤-mercaptoethanol, and 5% glycerol). Reaction mixtures usually contained 13.5 nM 26 S proteasome and 100 nM polyubiquitinated UbcH10 or other substrates as specified in the legends to Figs. 2, 3, and 6. Samples were withdrawn at each designated time point and added into 5ϫ SDS sample buffer to stop the reaction immediately. Usually samples of time 0 represented a reaction of about 15 s except in Fig. 5, C and D, and Fig. 6C where samples at time 0 were prepared by adding the substrates directly into 1ϫ SDS sample buffer with the 26 S proteasome. For reactions containing epoxomicin (100 M), Ubal (2.5 M) or 1,10-phenanthroline (5 mM), the 26 S proteasome was preincubated with the corresponding inhibitors for 10 min before the supplementation of substrates.
Size-exclusion Spin Column Assay-12.5 nM 26 S proteasome was preincubated with 2.5 M Ubal and 5 mM 1,10-phenanthroline for 10 min at 30°C in degradation buffer. 80 nM Lys-48 Ub 4 or Lys-63 Ub 4 was then mixed with the preinhibited protea-some and incubated for 2 min at room temperature. 60-l mixtures or Ub 4 alone were loaded into Micro Bio-Spin P-30 chromatography columns (Bio-Rad) and centrifuged according to the manufacturer's instruction. The flow-through was eluted directly into a 1.5-ml micro tube with 20 l of 5ϫ SDS sample buffer. 30-l samples were resolved by SDS-PAGE for immunoblotting assays. To determine the binding between the 26 S proteasome (13.5 nM) and Ub 4 (Lys-48 or Lys-63)-UbcH10 (100 nM), we used homemade Sephadex G-100 spin columns (exclusion limit is 80 kDa). After centrifugation, all resulting mixtures (75 l) were concentrated to about 30 l by heating. Proteins are resolved in SDS-PAGE and immunoblotted with an anti-UbcH10 antibody. Similar assays were also used to determine the interaction between the 26 S (SW) proteasome or PA700 and Usp14.
Purification of the 26 S Proteasome-bound Ubiquitinated Proteins-We obtained the pQCXIP viral expression plasmid that expresses S13/Rpn11-HTBH or the HTBH tag alone from Dr. L. Huang at the University of California at Irvine. The HTBH tag includes a His 6 tag, tobacco etch virus protease site, in vivo biotinylation sequence, and another His 6 tag. Stable HeLa cell lines that express S13-HTBH or the HTBH tag were established according to a published method by using HEK293 10A1 packaging cells (20). To purify the 26 S proteasome, HeLa cells were grown in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum and 1% penicillin/streptomycin. At 90% confluency, three 10-cm plates of cells were treated with 30 M MG132 or 0.3% DMSO for 45 min, then washed twice with phosphate-buffered saline before harvest. Cells were lysed with lysis buffer (20 mM Tris, pH 7.2, 50 mM NaCl, 10% glycerol, 2 mM ATP, 5 mM MgCl 2 , 2 mM ␤-mercaptoethanol, 10 mM iodoacetamide, 2 mM 1,10-phenathroline, and the protease inhibitor mixture (Roche Applied Science)). The lysates were cleared by centrifugation, and the supernatants were incubated with 50 l of streptavidin-agarose for 2 h at 4°C. The resins were then washed three times with the lysis buffer followed by two additional washes with the lysis buffer without iodoacetamide. Finally, the streptavidin-bound 26 S proteasome was released by incubation with 50 ng/l tobacco etch virus overnight at 4°C.
Quantitative Mass Spectroscopic Analysis-The whole cell lysates and the purified 26 S proteasome described above were resolved in SDS-PAGE. The gel regions containing the vast majority of ubiquitinated proteins (Ͼ70 kDa), judged by immunoblotting, were excised followed by in-gel trypsin digestion. Because trypsin cleaves ubiquitin to a small GG tag on modified lysine residues, the abundance of polyUb linkages is represented by the level of the GG peptides. Stable isotope-labeled GG peptides were chemically synthesized, quantified, and added into the samples as internal standards (21,22). The samples were fractionated by reverse-phase liquid chromatography coupled with a hybrid mass spectrometer (LTQ-Orbitrap; Thermo Fisher Scientific). The instrument was operated in the mode of selective reaction monitoring to increase sensitivity. The native GG peptides and the labeled internal standards were eluted together and separated by mass spectrometry due to mass difference, enabling relative quantification.

Neither the Cellular Level nor the 26 S Proteasome-bound
Level of Lys-63 PolyUbs Increased in Response to Short-time Proteasome Inhibition-If a specific Ub linkage targets proteins for proteasomal degradation, its cellular level and the 26 S proteasome-bound fraction should elevate in response to shorttime proteasome inhibition. To test this idea, we first generated stable HeLa cell lines that express S13/Rpn11-HTBH or the HTBH tag alone according to a previous report (20) (data not shown). This allowed us to rapidly purify the 26 S proteasome using affinity purification. We next examined whether treating these stable HeLa cell lines with 30 M MG132 for 45 min would increase cellular Ub conjugates. We found that the high molecular weight Ub conjugates in the whole cell lysates significantly increased as determined with immunoblotting of Ub (Fig. 1A). Quantitative mass spectrometry (21) that uses heavy isotope-labeled Ub linkage peptides as internal standards identified Lys-11-, Lys-48-, and Lys-63-linked polyUbs in both the DMSO-(control) and MG132-treated whole cell lysates. Interestingly, the relative amounts of the Lys-11 and the Lys-48 linkages increased 1.8-and 2.3-fold, respectively, upon MG132 treatment (Fig. 1B). In contrast, the amount of the Lys-63 linkages did not change (Fig. 1B). The other polyUb linkages were undetectable in the whole cell lysates. Thus, the cellular level of the Lys-63 linkages does not elevate in response to short-time proteasome inhibition, whereas the Lys-11 and Lys-48 linkages promptly increase.
Next, we examined the amount of 26 S proteasome-bound polyUb linkages to evaluate which Ub linkages target substrates to the 26 S proteasome. To this end we purified the 26 S proteasome from the established stable HeLa cells by utilization of the HTBH tag according to a previous report (20). To protect the cellular polyUbs from deubiquitination, the whole cell lysates were treated with inhibitors of deubiquitinating enzymes (iodoacetamide and 1,10-phenanthroline). Coomassie-stained SDS-PAGE and immunoblotting confirmed that the 26 S proteasome was purified from HeLa cells expressing S13-HTBH, but not from the HTBH tag alone (Fig. 1, C and D). Immunoblotting of the purified 26 S proteasome with an anti-Ub antibody found that the purified 26 S proteasome contained significant amounts of high molecular weight Ub conjugates, with more in the MG132-treated preparation (Fig. 1E). Next, we used mass spectrometry to quantify the proteasome-bound polyUb linkages. After enriching on the proteasome, the Lys-6, -11, -27, -29, -48, and -63 linkages were detectable, and their relative amounts were quantified (Fig. 1F). MG132 treatment increased the 26 S proteasome-bound Ub conjugates of the Lys-6-, -11-, -27-, -29-, and -48-linked at a range of 1.7-2.5-fold, whereas the Lys-63linked only had a slight increase (Fig. 1F). Thus, the Lys-63 linkage is the only detectable form in the 26 S proteasomebound fraction that does not significantly increase in response to short-time proteasome inhibition. Moreover, the amount of Lys-48 polyUbs bound to the 26 S proteasome is 6-and 12.7fold more than the Lys-63 polyUbs in the DMSO-and MG132treated preparations, respectively (Fig. 1G). This could explain why immunostaining of HeLa cells with Ub linkage-specific antibodies found that Lys-48 polyUbs colocalize with the 26 S proteasome, whereas the Lys-63 polyUbs do not (23). Furthermore, the Lys-48/Lys-63 ratios in the 26 S proteasome-bound fraction are 1.7-fold higher than those in the whole cell lysates in both DMSO-and MG132-treated preparations (Fig. 1G). Thus, cellular Lys-48 polyUbs appear to have more proteasome accessibility than the Lys-63 polyUbs.
Both the Lys-48 and Lys-63 PolyUbs Bind the 26 S Proteasome Comparably, but the Lys-63 PolyUbs Are Rapidly Deubiquitinated-In the DMSO-treated HeLa cells, 11 and 68% of the 26 S proteasome-bound Ub conjugates are the Lys- 63-and Lys-48-linked forms, respectively (data not shown). This indicates that both Ub linkages can bind the 26 S proteasome. We, therefore, evaluated the ability of the 26 S proteasome to process both Ub linkages from the aspects of proteasomal binding, deubiquitination, and degradation of conjugates. To obtain large quantities of the 26 S proteasome for biochemical characterization, we purified the 26 S proteasome from bovine red blood cells according to our previous report (18) (supplemental Fig. 1). Mass spectroscopy and immunoblotting determined that the purified 26 S proteasome contained three deubiquitinating enzymes: Uch37, Usp14, and S13 (Rpn11) (supplemental Fig. 1, C-F and data not shown). Notably, purified PA700 (19) contained both Uch37 and S13 but no Usp14 (supplemental Fig. 1C). Usp14 is a proteasome-associating protein, and its association is salt concentration-sensitive (24). It is likely that Usp14 dissociates from PA700 during the anion exchange purification step. To examine the ability of the mammalian 26 S proteasome to bind Lys-48 and Lys-63 polyUbs, we performed a size-exclusion spin column assay. We found that both Lys-48 and Lys-63 Ub 4 had comparable binding efficiency to the 26 S proteasome ( Fig. 2A). In the spin column assay, Lys-63 or Lys-48 Ub 4 are trapped inside the P-30 spin column (40-kDa exclusion limit) after centrifugation (lane 2 in Fig. 2A). Coelution of Ub 4 with the 26 S proteasome in the flow-through indicates a direct binding between them (lane 3 in Fig. 2A). It is noteworthy that the 26 S proteasome was incubated with both Ubal and 1,10-phenanthroline to inhibit its deubiquitinating activities in these binding assays. Given the fact that both Lys-48 and Lys-63 Ub 4 efficiently bind the 26 S proteasome, the two deubiquitinating inhibitors likely have no effect on binding of polyUbs to the 26 S proteasome.
We next evaluated 26 S proteasome-mediated deubiquitination of Lys-48 or Lys-63 Ub 4 . Surprisingly, we found that Lys-63 Ub 4 was deubiquitinated about 6-fold more rapidly than Lys-48 Ub 4 ( Fig. 2B and supplemental Fig. 2). Coomassiestained SDS-PAGE showed that both Lys-48 and Lys-63 Ub 4 were eventually deubiquitinated into monomeric Ub (supplemental Fig.  2). The more rapid deubiquitination of Lys-63 Ub 4 correlated with its higher V max value as compared with Lys-48 Ub 4 , whereas both chains had similar K m values ( Fig. 2C and supplemental Fig. 3). Together, these results indicate that the 26 S proteasome binds both Lys-48 and Lys-63 polyUbs comparably, but it deubiquitinates Lys-63 polyUbs much more rapidly.
Both the Ubal-and 1,10-Phenanthroline-sensitive Deubiquitinating Activities Contribute to Deubiquitination of Lys-48 and Lys-63 PolyUbs-We next aimed to determine which of the deubiquitinating activities of the 26 S proteasome is responsible for deubiquitination of Lys-48 or Lys-63 polyUbs. The 26 S proteasome contains two cystine-protease deubiquitinating enzymes (Uch37 and Usp14) that are sensitive to Ubal inhibition. It also contains a Zn 2ϩ -dependent metalloprotease (S13/Rpn11) that is inhibited by the metal chelator 1,10-phenanthroline. Uch37 was shown to have a Ub chain trimming activity that initiates cleavage from the distal site (25). S13 has a proximal deubiquitinating activity that cleaves an entire polyUb from the substrates (26). Ubal had a concentration-dependent inhibitory effect on 26 S proteasome-catalyzed deubiquitination (supplemental Fig.  4). Accordingly, we used 2.5 M Ubal to block the activity of Uch37/Usp14 and 5 mM 1,10-phenanthroline to inhibit the activity of S13. For Lys-48 Ub 4 , at a reaction time point when more than 85% of Ub 4 was deubiquitinated if no Ubal or 1,10phenanthroline was added, adding Ubal exhibited nearly complete inhibition of both the 26 S proteasome-and PA700-mediated deubiquitination, whereas 1,10-phenanthroline had a more mild inhibitory effect (upper panels in Fig. 3, A and B). For Lys-63 Ub 4 , neither Ubal nor 1,10-phenanthroline alone completely inhibited deubiquitination, with Ubal having a stronger effect (lower panels in Fig. 3, A and B). In contrast, deubiquitination of Lys-63 Ub 4 was abolished by a combination of these two inhibitors (lanes 10 and 11 in Fig. 3B). Thus, both the Ubaland 1,10-phenanthroline-sensitive deubiquitinating enzymes contribute to deubiquitination of Lys-48 and Lys-63 polyUbs, but they have different inhibitory extents. The 26 S Proteasome Catalyzes an Usp14-dependent deISGylation Activity-To investigate whether both of the thiol proteases (Uch37 and Usp14) of the proteasome contribute to the Ubal-sensitive deubiquitination, we prepared Usp14-depleted 26 S proteasome by taking advantage of the fact that its residence on the 26 S proteasome is salt concentration-sensitive (24). To remove Usp14, we incubated our purified 26 S proteasome with 160 mM NaCl. Dissociated salt-sensitive binding proteins were further separated by gel filtration (data not shown). We found that, compared with the purified intact 26 S proteasome, the salt-treated 26 S proteasome contained 9% Usp14 and 66% Uch37 (Fig. 4A). Other intrinsic proteasomal subunits including the S13 subunit remained intact ( Fig. 4A and supplemental Fig. 5A). Thereafter, the salt-washed 26 S proteasome was referred to as the 26 S (SW) proteasome. Importantly, salt treatment did not disrupt proteasome integrity as both the regularly purified 26 S and the 26 S (SW) proteasomes consisted of ϳ50% double-capped and 50% singlecapped proteasomes (Fig. 4B). Also, salt treatment did not affect the peptidase activity of the proteasomes (values listed under the gel in Fig. 4B and supplemental Fig. 5B). Next, we examined whether adding back recombinant Usp14 increases the deubiquitinating activity of the 26 S (SW) proteasome. The size exclusion spin column assay demonstrated that the recombinant Usp14 bound the 26 S (SW) proteasome and PA700 (Fig.  4C). Moreover, the 26 S and PA700-bound Usp14 was activated at least partially as probed by Ub vinyl sulfone (Fig. 4D). However, binding of Usp14 did not promote the 26 S (SW) proteasome or PA700 to deubiquitinate Ub-Amc and Lys-48 Ub 4 ( Fig.  4E and supplemental Fig. 5, C and D). Consistent with these findings, the loss of the deubiquitination activity of the 26 S (SW) proteasome (39%) when using Ub-Amc as the substrate is proportional to the loss of the Uch37 subunit (34%) but not Usp14 (91%) (supplemental Fig. 5D). Uch37 associates with the 26 S proteasome through interacting with Adrm1. Moreover, Adrm1 activates Uch37 deubiquitinating activity (27)(28)(29). In contrast to Usp14, adding back purified Adrm1/Uch37 to the 26 S (SW) proteasome significantly stimulated deubiquitination of Ub-Amc (supplemental Fig. 5D), although no effect was observed for deubiquitination of Lys-48 or Lys-63 Ub 4 (data not shown). The deubiquitination discrepancy between Ub-Amc and polyUbs is unclear and under investigation. A recent study found that proteasome-bound Usp14 is reactive to the catalytic site probes of both Ubvinylmethyl ester and ISG15-vinyl sulfone (30). ISG15 is an Ub-like modifier found only in vertebrates, and its expression is induced by type I interferons and viral or bacterial infection. We, therefore, examined if Usp14 catalyzes deISGylation. PA700 did not catalyze deISGylation of ISG15-Amc. Adding back recombinant Usp14 to PA700 stimulated deISGylation of ISG15-Amc (Fig. 4F). Similarly, adding back recombinant Usp14 stimulated the 26 S (SW) proteasome deISGylation activity (Fig. 4F). Thus, Usp14 catalyzes the deISGylation activity of the 26 S proteasome.

Rapid Deubiquitination Causes Inefficient Degradation of Some Lys-63 PolyUb Substrates-Because
Lys-48 and Lys-63 polyUbs have obviously different deubiquitination rates catalyzed by the 26 S proteasome, we next compared the ability of Lys-48 and Lys-63 Ub 4 to target proteins for degradation. To do this, we conjugated Lys-48 or Lys-63 Ub 4 to the physiological substrate, UbcH10 (31), using immunoprecipitated Xenopus anaphase-promoting complex/cyclosome as the E3 ligase (18). The size exclusion spin column assay demonstrated that both Lys-48 and Lys-63 Ub 4 -UbcH10 efficiently bound to the 26 S proteasome (Fig. 5A). Consistent with our earlier report (18), Lys-48 Ub 4 -UbcH10 was efficiently degraded by the purified 26 S proteasome as judged by the fact that UbcH10 only accumulated in the reaction containing the proteasome inhibitor, epoxomicin (comparing lanes 1 and 3 in Fig. 5B). Concomitantly, more Lys-48 Ub 4 -UbcH10 remained in the reaction containing epoxomicin (lane 1 in Fig. 5B), suggesting that deubiquitination is impaired as a secondary consequence of inhibited proteolytic activity (18,32). Surprisingly, Lys-63 Ub 4 -UbcH10 was rapidly deubiquitinated by the purified 26 S proteasome and mainly accumulated as monoubiquitinated UbcH10 (Ub-UbcH10) (Fig. 5C). Because monomeric Ub cannot efficiently bind the 26 S proteasome, rapid deubiquitination into Ub-UbcH10 could cause the substrate to dissociate from the 26 S proteasome without degradation. The rapid deubiquitination of Lys-63 Ub 4 -UbcH10 was inhibited by the addition of both Ubal and 1,10-phenanthroline (lane 5 in Fig. 5C). Consistent with the results from previous deubiquitination inhibition assays (Fig. 3), Ubal had a stronger effect than 1,10-phenanthroline on inhibition of the deubiquitination of Lys-63 Ub 4 -UbcH10 (lanes 6 and 7 in Fig. 5C). Notably, about 12% of Lys-63  DECEMBER 18, 2009 • VOLUME 284 • NUMBER 51

JOURNAL OF BIOLOGICAL CHEMISTRY 35489
Ub 4 -UbcH10 was degraded as determined by densitometric analysis (comparing lanes 4 and 8 in Fig. 5C). Together, these results show that rapid deubiquitination of polyUbs could cause inefficient degradation of some of its conjugates.
Because the majority of Lys-63 Ub 4 -UbcH10 was rapidly deubiquitinated into Ub-UbcH10 without degradation, we hypothesized that the 26 S (SW) proteasome, which has decreased deubiquitination activity, would retain Lys-63 Ub 4 -UbcH10 on the proteasome long enough to promote degradation. This idea turns out to be true as we found that the 26 S (SW) proteasome efficiently bound Lys-63 Ub 4 -UbcH10 (supplemental Fig. 6) and catalyzed efficient degradation of Lys-63 Ub 4 -UbcH10 (Fig. 5D). Degradation was abolished by adding either Ubal or 1,10-phenanthroline (lanes 5 and 6 in Fig. 5D), indicating that both of the deubiquitination activities are required for mediating degradationcoupled deubiquitination. Interestingly, degradation of Lys-63 Ub 4 -UbcH10 by the 26 S (SW) proteasome occurred even more rapidly than the degradation of Lys-48 Ub 4 -UbcH10 by either the 26 S (SW) (supplemental Fig. 7) or the regular 26 S proteasome (compare Fig. 5D to 5B), indicating that deubiquitination could be the rate-limiting step in degradation of some proteins.

NEMO Protects Lys-63-linked PolyUbs from Proteasomal
Deubiquitination in Vitro-The above studies indicate that both Lys-48 and Lys-63 polyUbs bind the 26 S proteasome equally well in vitro ( Figs. 2A and 5A), whereas Lys-48 polyUbs are found to be more abundant at the proteasome in vivo (Fig.  1G). Because Lys-63 polyUbs often form a complex with their binding partners when performing non-proteolytic functions, we speculated that Lys-63 polyUb-interacting proteins sequester Lys-63 polyUbs and limit their accessibility to the 26 S proteasome or other deubiquitinating enzymes. In this regard, in tumor necrosis factor ␣-stimulated activation of the NFB gene, IB kinase ␥ (IKK␥) (NEMO), the non-catalytic subunit of the IKK kinase complex, binds the Lys-63 polyUb chain(s) on RIP (33), and this interaction is required for stabilization of RIP (33). This stabilization effect is likely mediated by blocking deubiquitination of Lys-63 polyubiquitinated RIP by A20 (33). Accordingly, we hypothesized that NEMO is capable of protecting Lys-63 polyUbs from deubiquitination by the 26 S proteasome and other deubiquitinating enzymes such as A20. To test this hypothesis, we synthesized Lys-63 polyUb mixtures, Ub nϾ6 . Glutathione S-transferase-NEMO pulldown experiments demonstrated that NEMO preferred to bind long Lys-63 chains (Fig. 6A). Consistent with our hypothesis, preincubation of Lys-63 Ub nϾ6 with NEMO blocked 26 S proteasome-mediated deubiquitination, and this effect was more pronounced on longer polyUbs (Fig. 6B). We next conjugated Lys-63 Ub nϾ6 to UbcH10 using immunoprecipitated Xenopus anaphase-promoting complex/cyclosome as the E3 Ub ligase. Lys-63 Ub nϾ6 -UbcH10 was rapidly deubiquitinated into Ub-UbcH10 by the 26 S proteasome without obvious degradation (comparing lanes 2 and 3 in Fig. 6C), consistent with the result obtained from Lys-63 Ub 4 -UbcH10 (Fig. 5C). Preincubation of Ub nϾ6 -UbcH10 with NEMO blocked proteasome-mediated deubiquitination (lanes 4 -6 in Fig. 6C). A longer time course experiment showed that inhibition of the deubiquitination of Lys-63 Ub nϾ6 -UbcH10 by NEMO was extremely effective (supplemental Fig. 8). Furthermore, the inhibitory effect was due to the specific interaction between NEMO and Lys-63 polyUbs because a single residue substitution (L329P) in NEMO that disrupts the interaction between NEMO and Lys-63 polyUbs (33) abrogated the inhibitory effect on deubiquitination of Lys-63 Ub nϾ6 -UbcH10 (lanes 7-9 in Fig. 6C). Moreover, we found that NEMO protected Lys-63 Ub nϾ6 -UbcH10 from A20mediated deubiquitination (supplemental Fig. 9). Together, these results demonstrate that Lys-63 polyUb-interacting proteins can protect Lys-63 polyUbs from deubiquitination by the 26 S proteasome and other deubiquitinating enzymes.

NEMO Protects Its Bound Linear PolyUbs from Deubiquitination in
Vivo-Next, we sought to examine whether NEMO protects its bound polyUbs from deubiquitination in vivo. Linear Ub chains were recently discovered to regulate activation of the NFB pathway by modification of NEMO (34). Structural studies revealed that NEMO binds linear di-Ub through its CC2-LZ (also called UBAN) domain (35,36). We, therefore, examined whether NEMO can stabilize linear Ub 6 in vivo. HEK293 cells were transfected with HA-Ub 6 and FLAG-NEMO, FLAG-NEMO (L329P), or a combination of both. Immunoblotting the whole cell lysates with an anti-HA antibody found no obvious accumulation of HA-Ub 6 in cells when it was transfected alone (lane 2 in Fig. 6D). In contrast, HA-Ub 6 and its conjugates accumulated when cotransfected with NEMO but not NEMO (L329P) (lanes 5 and 6 in Fig. 6D). These results suggest that NEMO protects linear Ub 6 from deubiquitination or degradation in cells. To further examine whether the protection is from binding of HA-Ub 6 with NEMO, we immunoprecipitated FLAG-NEMO or FLAG-NEMO (L329P) and found that NEMO indeed precipitated HA-Ub 6 and its conjugates (Fig. 6E). These results suggest that NEMO is able to protect its bound polyUbs from deubiquitination in cells.

Lys-48 and Lys-63 Ub Chains Have Different Proteasomal Accessibility, Deubiquitination Rates, and Abilities to Target
Proteins for Proteolysis-In this study we systematically compared Lys-63 polyUbs with Lys-48 polyUbs in proteasomal binding, deubiquitination, and in targeting proteins for degradation. In vitro, Lys-48-linked and Lys-63-linked tetraubiquitin bound the 26 S proteasome equally well ( Fig. 2A). In contrast, quantitative mass spectrometry determined that the 26 S proteasome-bound Ub chains had a higher Lys-48/Lys-63 ratio than that in the whole cell lysates (Fig. 1G), indicating that cellular Lys-63 Ub chains have less proteasomal accessibility than Lys-48 chains. Also, we found that a Lys-63 chain-specific binding protein, NEMO, protected Lys-63 or linear Ub chains from deubiquitination by the 26 S proteasome and/or deubiquitinating enzymes (Fig. 6). Thus, some cellular Lys-63 Ub chains could be sequestered by their binding partners. This protecting effect might explain why our mass spectrometric analyses determined that Lys-63 polyUbs have less proteasomal accessibility than Lys-48 polyUbs in cells. In response to a short time of proteasome inhibition, Lys-63 polyUbs are the only detectable Ub linkage in HeLa cells that did not increase in the cellular level or the 26 S proteasomebound fraction. In contrast, other detectable linkages including the Lys-6, -11, -27, -29, and -48 increased promptly. Under a severe proteasome inhibition condition (10 M MG132 for 15 h) we found that the cellular levels of all detectable Ub linkages (Lys-11, -48, and -63) increased (data not shown), consistent with two recent studies in yeast and mammalian cells (37,38). The increase of the cellular level of Lys-63 polyUbs in the later time of proteasome inhibition would suggest that most Lys-63 chains are not used as signals for proteasomal degradation; otherwise, they would have promptly increased after proteasome inhibition in a similar manner as the Lys-48 polyUbs. The delayed accumulation of Lys-63 polyUbs in response to proteasome inhibition might come from a secondary effect of an impaired Ub-proteasome pathway (37). For instance, Lys-63 polyUbs are likely protected from deubiquitination/degradation by binding to their partners. Therefore, impaired degradation of their binding partners under severe proteasome inhibition could result in the accumulation of more Lys-63 polyUbs in cells. Certainly, a small portion of proteasome-bound Lys-63 Ub conjugates could be degraded (see below); thus, severe proteasome inhibition would eventually result in an increase of Lys-63 polyUbs as well.
Interestingly, we found that the 26 S proteasome deubiquitinates Lys-48 and Lys-63 Ub chains differently. Although both chains were bound similarly by the 26 S proteasome, Lys-63linked Ub 4 was deubiquitinated six times more rapidly than Lys-48 Ub 4 . When conjugated to UbcH10, Lys-48 Ub 4 effi-ciently targeted UbcH10 for degradation, whereas only 12% of Lys-63 Ub 4 -UbcH10 was degraded. The majority of Lys-63 Ub 4 -UbcH10 was rapidly deubiquitinated into Ub-UbcH10 without degradation (Fig. 5), possibly because low proteasomal binding affinity of monomeric Ub causes dissociating Ub-UbcH10 from the 26 S proteasome. Rapid deubiquitination of Lys-63 polyUbs compared with Lys-48 polyUbs might result from the difference in topology of these chains. The isopeptide bonds in Lys-63 polyUbs are exposed in an open conformation, whereas they are buried in Lys-48 polyUbs (39). The open conformation might make Lys-63 polyUbs more accessible to the deubiquitinating enzymes of the 26 S proteasome than Lys-48linked. Certainly, we cannot exclude the possibility that the different proteasomal binding geometries, if any, among the different Ub linkages might also cause varied accessibility to the deubiquitinating enzymes. Accordingly, the topologies of the Ub linkages might determine their rates of deubiquitination by the 26 S proteasome, conferring a layer of substrate selectivity for proteasomal degradation.
Degradation of polyubiquitinated proteins requires highly coordinated actions including substrate binding, deubiquitination, unfolding, translocation, peptide hydrolysis, and ATP hydrolysis (18). Disturbing this process by disrupting any one of these actions could be detrimental to proteasomal degradation. In the case of Lys-63 polyUb conjugates, rapid deubiquitination could cause the conjugates to be released from the proteasome before being unfolded for translocation and degradation. In this regard we would expect that Lys-63 polyUbs could target degradation of unfolded proteins much more efficiently than that of stably folded ones. Using this same line of reasoning, reducing the deubiquitination activity of the 26 S proteasome would facilitate the degradation of Lys-63 polyUb conjugates. We show this to be true using a 26 S proteasome preparation that had decreased deubiquitinating activity (Fig. 5D). Therefore, it is not surprising that Lys-63 polyUbs were found to target several proteins for degradation in vitro in cases where the substrates are not well folded (14), the deubiquitinating activity of the 26 S proteasome is partially inhibited by Ubal (15), or deubiquitination activity is reduced by depletion of the deubiquitination enzymes (32). In addition to targeting proteins to the proteasome, Lys-63 Ub chains play a role in endosomal sorting and could target proteins to lysosomal degradation (40 -42), a process that is also regulated by endosome-residing deubiquitinating enzymes (40).
Both the Ubal-and the 1,10-Phenanthroline-sensitive Deubiquitinating Activities of the 26 S Proteasome Contribute to Lys-48-and Lys-63-linkage Deubiquitination-We found that the Ubal-and the 1,10-phenanthroline-sensitive deubiquitinating activities of the 26 S proteasome have different inhibitory effects on Lys-48 and Lys-63 polyUbs. Simultaneous inhibition of both the Ubal-and the 1,10-phenanthroline-sensitive activities were required for complete inhibition of Lys-63 chain deubiquitination. In contrast, Ubal alone exhibited nearly complete inhibition of Lys-48 chain deubiquitination, whereas 1,10-phenanthroline had a more mild inhibitory effect. To distinguish the role of the two Ubal-sensitive enzymes (Uch37 and Usp14), we found that add-back of Uch37/Adrm1 to the 26 S (SW) proteasome was able to stimulate deubiquitination. In contrast, add-back of Usp14 to the 26 S (SW) proteasome or PA700 did not stimulate deubiquitination of any tested substrates ( Fig. 4 and supplemental Fig. 5). This may imply that Usp14 is not a major deubiquitinating enzyme of the mammalian proteasome. However, we cannot exclude the possibilities that Usp14 may have specific activity against other untested Ub linkages; that Usp14 is activated by an unknown protein that is not present in our in vitro system and/or Usp14 is redundant when coexisting with Uch37. Additionally, Usp14 did stimulate a modest deIS-Gylation activity at the proteasome and may very well be an authentic deISGylating enzyme, but further investigation is required.
A recent study reported that neither N-ethylmaleimide (a cysteine modifier that inhibits thiol proteases) nor Ubal block bovine PA700 or 26 S (unspecified source)-catalyzed deubiquitination of Lys-63 Ub 2 , whereas1,10-phenanthroline does (43). Thus, the authors proposed that S13 is responsible for Lys-63 chain deubiquitination of the 26 S proteasome. It seems unlikely that the different conclusions originate from different proteasome sources because we purified PA700 and the 26 S proteasome from bovine red blood cells as well. Different experimental setup might explain why the previous study did not observe the activity of Uch37 in catalyzing deubiquitination of Lys-63 polyUbs. First, we used 2.5 M Ubal to inhibit Uch37 activity in this study, whereas the other study used 0.5 M Ubal (43). A concentration of 0.5 M Ubal cannot efficiently block the chain-trimming activity of the 26 S proteasome (supplemental Fig. 4). Second, we used Lys-63 Ub 4 to evaluate proteasomal deubiquitination activities, whereas Lys-63 Ub 2 was used in the other study (43). It is possible that deubiquitination of Lys-63 polyUbs by Uch37 depends on efficient proteasomal binding, which requires a minimal chain length of four Ubs (44). Unfortunately, we were not able to evaluate the effect of N-ethylmaleimide treatment on deubiquitination because our purified 26 S proteasome was disassembled when incubated with 2 mM N-ethylmaleimide (data not shown). Presumably, N-ethylmaleimide modifies cysteine residues in some subunits that are essential for maintaining proteasome integrity.
Uch37 and S13 belong to two different deubiquitinating enzyme families; Uch37 is a thiol protease, and S13 is a Zn 2ϩdependent metalloprotease and a member of the JAMM/ MPNϩ deubiquitinating family. Interestingly, the deubiquitinating activities of both Uch37 and S13 are activated when integrated into the 26 S complex. The JAMM/MPNϩ family members, including AMSH and Brcc36, have been shown to have specificity for Lys-63 polyUbs (43,45). Not surprisingly, S13 catalyzes deubiquitination of Lys-63 polyUbs (43). In contrast to the specificity of the JAMN/MPNϩ deubiquitinating enzymes for Lys-63-linked chains, the thiol-utilizing deubiquitinating enzymes have diverse deubiquitination specificities. For example, Usp2, Usp5, and Usp15 can deubiquitinate both the Lys-48 and Lys-63 polyUbs (46), whereas CYLD only cleaves Lys-63 polyUbs (46). The data from our study reveals that Uch37 belongs to a thiol-dependent deubiquitinating enzyme group that cleaves both the Lys-48 and Lys-63 linkages.