Identification of the Preferential Ubiquitination Site and Ubiquitin-dependent Degradation Signal of Rpn4*

Lysine selection is a long-standing problem in protein ubiquitination catalyzed by the RING ubiquitin ligases. It is well known that many substrates carry multiple lysines that can be ubiquitinated. However, it has seldom been addressed whether one lysine is preferred for ubiquitin conjugation when all other lysines exist. Here we studied the mechanism underlying ubiquitin-dependent degradation of Rpn4, a transcription activator of the Saccharomyces cerevisiae proteasome genes. We found that the ubiquitin-dependent degradation of Rpn4 can be mediated by six different lysines. Interestingly, we showed through in vivo and in vitro assays that lysine 187 is selected for ubiquitination when all other lysines are available. To the best of our knowledge, this is the first demonstration of a preferential ubiquitination site chosen from a group of lysines susceptible for ubiquitination. We further demonstrated that lysine 187 and a proximal acidic domain constitute a portable degradation signal. The implications of our data are discussed.

The ubiquitin (Ub) 2 system is responsible for selectively marking abnormal and regulatory proteins for degradation by the proteasome (1)(2)(3). Ubiquitination of a protein substrate is a consecutive process involving multiple enzymes (4). Ub is first activated by the Ub-activating enzyme (E1), forming a thioester between the C-terminal carboxyl group of Ub and a specific cysteine of the E1. The Ub moiety of the E1ϳUb thioester is thereafter transferred to one of the Ub-conjugating enzymes (E2). The Ub moiety of the E2ϳUb thioester is conjugated, via an isopeptide bond, to the ⑀-amino group of a lysine residue (Lys) of a substrate or a preceding Ub molecule conjugated to the substrate, the latter reaction resulting in a substrate-linked multi-Ub chain. Most E2s function in complex with one of the E3 enzymes or Ub ligases. A Ub ligase also denotes an E2⅐E3 complex. Besides this common scenario, multiubiquitination of a subset of substrates also requires a Ub chain elongation factor (E4) (5). Ubiquitination of a specific substrate is mainly regulated through modulation of its degradation signal (degron) and through control of the activity of its cognate E3 (4, 6 -8). The isopeptide bond between Ub and a substrate can be hydrolyzed by deubiquitinating enzymes, which provides yet another layer of regulation for substrate ubiquitination (9).
Most E3s are grouped into two families (HECT E3 and RING E3) based on their catalytic modules and features of sequence and structure (4,7). A HECT E3 can accept Ub moiety from an associated E2ϳUb thioester, forming an E3ϳUb thioester intermediate and acting as a proximal Ub donor to the substrate that it selects. By contrast, formation of thioester between a RING E3 and Ub has not been detected. The precise mechanism by which a RING E3 catalyzes the transfer of Ub from the E2ϳUb thioester to the substrate is still poorly understood. Many known substrates of the RING E3s carry multiple lysines that can be ubiquitinated (10 -12). It is unclear whether one lysine is preferred for Ub conjugation when all other lysines susceptible for ubiquitination are also available in the substrate. This question remains one of the central issues in regard to the mechanism of substrate ubiquitination by the RING E3s. It is also uncertain whether or not formation of multiple Ub chains is required for substrate degradation by the proteasome. Failure to address this problem partly results from biased experimental designs that emphasized the essential but not the sufficient lysine for substrate ubiquitination and degradation.
Rpn4 (also named Son1 and Ufd5) is a transcription activator required for normal expression of the Saccharomyces cerevisiae proteasome genes (13,14). Interestingly, Rpn4 is extremely short-lived and degraded by the proteasome (14). These observations and subsequent reports demonstrated that the proteasome homeostasis is regulated by a negative feedback circuit in which Rpn4 up-regulates the proteasome genes and is destroyed by the proteasome (15,16). Intriguingly, the proteasomal degradation of Rpn4 can be mediated by two distinct pathways (17). One pathway is Ub-independent, whereas the other involves lysine ubiquitination. Recently, we demonstrated that the Ub-dependent degradation of Rpn4 is mediated by the Rad6/Ubr2 Ub ligase (18). Ubr2, a RING E3, is a sequence homolog of Ubr1, the E3 component of the N-end rule pathway (18 -20). The Ubr2-mediated ubiquitin-dependent degradation of Rpn4 has been shown to play an important role in controlling the steady-state levels of Rpn4 and the proteasome (18).
In the current work, we investigated the molecular mechanism underlying the Ub-dependent degradation of Rpn4. We found that the Ub-dependent degradation of Rpn4 can be mediated by six different lysines. Remarkably, among these six lysines, Lys-187 is the preferential ubiquitination site. We further demonstrated that Lys-187 and a proximal acidic domain constitute a portable Ubr2-dependent degradation signal.
Pulse-Chase and Immunoprecipitation Analysis-S. cerevisiae cells from 10-ml cultures (A 600 of 0.8 -1.0) in SD medium containing 0.1 mM CuSO 4 and essential amino acids were harvested. The cells were resuspended in 0.3 ml of the same medium supplemented with 0.15 mCi of [ 35 S]methionine/cysteine (Expre 35 S 35 S labeling mix, PerkinElmer Life Sciences) and incubated at 30°C for 5 min. The cells were then pelleted and resuspended in the same SD medium with cycloheximide (0.2 mg/ml) and excessive cold L-methionine/L-cysteine (2 mg/ml L-methionine and 0.4 mg/ml L-cysteine) and chased at 30°C. An equal volume of sample was withdrawn at each time point. Labeled cells were harvested, lysed in an equal volume of 2ϫ SDS buffer (2% SDS, 30 mM dithiothreitol, 90 mM Na-HEPES, pH 7.5), and incubated at 100°C for 3 min. The supernatants were diluted 20-fold with buffer A (1% Triton X-100, 150 mM NaCl, 1 mM EDTA, 50 mM Na-HEPES, pH 7.5) before being applied to immunoprecipitation with anti-HA antibody (Sigma) or anti-␤-galactosidase antibody (Promega) combined with protein A-agarose (Calbiochem) or anti-FLAG M2 affinity-agarose (Sigma). The volumes of supernatants used in immunoprecipitation were adjusted to equalize the amounts of 10% trichloroacetic acid-insoluble 35 S. The immunoprecipitates were washed three times with buffer A and resolved by SDS-PAGE followed by autoradiography and quantitation with a PhosphorImager (Amersham Biosciences).
Pulldown-Ubiquitination Assay-The pulldown-ubiquitination assay was carried out as described previously (18). Specifically, ubr2⌬ cells harboring plasmid 425GAL1FLAGUBR2 overexpressing N-terminally FLAG-tagged Ubr2 from the GAL1 promoter in a high copy vector were grown in synthetic selective medium containing 2% galactose to an A 600 of 1.8. Cells were spun down and manually ground to fine powder with a pestle. The pellet being ground was kept frozen by liquid nitrogen. Crude extracts were prepared by incubation of the powder in buffer B (0.2% Triton X-100, 150 mM NaCl, 50 mM HEPES, pH 7.5) plus protease inhibitor mix (Roche Diagnostics). Approximately, 0.2 mg of crude extract was used for each pulldown. After a standard pulldown, the beads were washed three times with buffer C (25 mM HEPES, pH 7.5, 25 mM KCl, 5 mM MgCl 2 , 2 mM ATP, and 0.1 mM dithiothreitol) and then subjected to ubiquitination reactions containing 100 nM Uba1 and 1 M Rad6 with or without 7 M ubiquitin in buffer C at 30°C for 45 min. The GST fusion proteins were resolved by SDS-PAGE followed by immunoblotting with anti-GST antibody (Sigma). Rad6 was overexpressed and purified from vector pET-11d in bacteria, whereas N-terminally hexahistidine-tagged Uba1 was overexpressed from YEplac181 in S. cerevisiae and purified by nickel-nitrilotriacetic acid chromatography according to the manufacturer's instructions (Qiagen).

Ub-dependent Degradation of Rpn4 Mediated by Different Lysines with Lys-187
Being the Most Efficient-Our early work has shown that proteasomal degradation of Rpn4 can be mediated by two distinct mechanisms, Ub-dependent and -independent (17). Taking advantage of the observation that deletion of the N-terminal 10 amino acids substantially inhibits the Ub-independent degron but displays no effect on the Ub-dependent degradation of Rpn4 (17,18), we adopted Rpn4 ⌬1-10 to study the molecular mechanism underlying the Ub-dependent degradation of Rpn4.
We first tried to define the lysine(s) required for Rpn4 ⌬1-10 degradation. Rpn4 possesses 35 lysines throughout its 531-amino acid sequence. Lys-9 was already deleted in Rpn4 ⌬1-10 . Previous study and current pulse-chase analysis showed that the turnover of Rpn4 ⌬1-10/10R , a mutant of Rpn4 ⌬1-10 with 10 N-terminal lysines (Lys-47 to Lys-187, referred to the positions in wild type Rpn4, see Fig. 1A) mutated to arginines, was severely inhibited (Fig. 1, B and E), indicating that the major ubiquitination site or sites are among the 10 N-terminal lysines. Note that the band just above the position of Rpn4 ⌬1-10 and Rpn4 ⌬1-10/10R represents a phosphorylated form of these two proteins, which is sensi-tive to treatment with calf intestinal alkaline phosphatase (data not shown). To determine whether all 10 N-terminal lysines are required for Rpn4 ⌬1-10 degradation, we constructed two Rpn4 ⌬1-10 mutants that bear Lys-to-Arg mutations for the first four lysines (N-4R) and for the fifth to tenth lysines (C-6R), respectively. Pulse-chase experiments were carried out to measure the stability of these two proteins. Although C-6R remained unstable, its turnover was much slower than that of N-4R, which was degraded as rapidly as Rpn4 ⌬1-10 (Fig. 1C). These results indicate that whereas the four N-terminal lysines are dispensable, one or more of the next six lysines (Lys-123 to Lys-187) are essential for wild type level degradation of Rpn4 ⌬1-10 .

(lanes
Lys-187 Is the Preferential Ubiquitination Site of Rpn4-The more efficient degradation of Rpn4 ⌬1-10 mediated by Lys-187 than by other lysines promoted us to examine if Lys-187 is a preferential ubiquitination site. We first sought to compare the ubiquitination efficiency of various Rpn4 ⌬1-10/10R derivatives that carry one of the lysines from Lys-132 to Lys-187. As the measurement of multiubiquitinated substrates formed in vivo can be complicated because of their degradation by the proteasome and rapid deubiquitination by the deubiquitinating enzymes, we decided to assess the efficiency of conjugation of the first Ub moiety to different lysines added back to Rpn4 ⌬1-10/10R . To this end, we took advantage of Ub K48R/G76A , a ubiquitin mutant that inhibits the formation of a Lys-48-linked multi-Ub chain (21). Our early work has shown that the Ub-dependent degradation of Rpn4 is through Lys-48linked multiubiquitination (17). Moreover, Ub K48R/G76A is a poor substrate for deubiquitinating enzymes (22) such that substrates conjugated with a moiety of Ub K48R/G76A can be well preserved in a short pulse-labeling period. Therefore, the ratio of monoubiquitinated to nonubiquitinated species of a specific substrate largely reflects the ubiquitination efficiency at a defined lysine.
The N-terminal Acidic Domain (NAD) Is Essential for the Ub-dependent Degradation of Rpn4-Our recent study has located the major, if not exclusive, degron of Rpn4 to its N-terminal 229-residue domain (17). To further define the minimal sequences required for the Ub-dependent degradation of Rpn4, we created a series of N-terminally truncated mutants (Fig. 3A). Pulse-chase analysis showed that deletion up to the N-terminal 148 amino acids had no effect on the turnover of Rpn4 (Fig.  3B, lanes 1-12), implying that the Ub-dependent degron was located within residues 149 -229. Not surprisingly, further deletion to residue 190 or 205 inhibited the degradation (Fig. 3B, lanes 13-18) as the preferential and all alternative ubiquitination sites were removed. Interestingly, the sequence of residues 211-229 is rich in acidic amino acids, a characteristic of transcription activation domain (Fig. 3A). We suspect that this domain might act as a ubiquitination signal because the ubiquitination signals of transcription activators are often overlapped with their activation domains (24). To test this hypothesis, we first compared the stability of two truncated Rpn4 mutants containing residues 11-229 (Rpn4 11-229 ) and 11-210 (Rpn4 11-210 ), respectively, using pulse-chase analysis (Fig. 3C). Whereas Rpn4 11-229 was extremely short-lived, Rpn4 11-210 was barely degraded, suggesting that residues 211-229 serve as a ubiquitination signal in Rpn4  .
We then examined the role of residues 211-229 in the context of Rpn4 ⌬1-10 . Sequence analysis showed that Rpn4 carries two domains rich in acidic amino acids (Fig. 3A). NAD consists of residues 211-229, whereas the C-terminal acidic domain (CAD) lies between residues FIGURE 3. NAD is required for the Ub-dependent degradation of Rpn4. A, schematic representation of the Rpn4 deletion mutants. The sequences of NAD and CAD are also displayed. B, pulse-chase analysis of the N-terminal deletion mutants of Rpn4. All mutants were C-terminally tagged with an HA epitope and expressed from the CUP1 promoter in a low copy vector. C and D, NAD is required for the Ub-dependent degradation of Rpn4. The degradation of C-terminally FLAG-tagged Rpn4   (lanes 1-3) and Rpn4   (lanes 4 -6) was assessed by pulsechase analysis (C). The stability of C-terminally FLAG-tagged Rpn4 ⌬1-10/⌬211-229 (lanes 1-3), Rpn4 ⌬1-10/⌬291-311 (lanes 4 -6), and Rpn4 ⌬211-229 (lanes 7-9) was measured by pulse-chase analysis (D). The CUP1 promoter was used to express these truncated mutants in a low copy vector. 300 -312. We, therefore, generated two mutants from Rpn4 ⌬1-10 via internal deletion of residues 211-229 and 291-313, respectively (Fig.  3A). Remarkably, pulse-chase analysis showed that deletion of NAD but not CAD inhibited the degradation of Rpn4 ⌬1-10 (Fig. 3D, compare  lanes 1-3 with 4-6), indicating that NAD is required for the Ub-dependent degradation of Rpn4. The proximal position of NAD to Lys-187 may functionally distinguish itself from CAD. To examine whether NAD also plays a role in the Ub-independent degradation of Rpn4, we deleted NAD in the context of full-length Rpn4 and measured the turnover of Rpn4 ⌬211-229 by pulse-chase analysis. As shown in Fig. 3D (lanes  7-9), deletion of NAD displayed no effect on the Ub-independent degradation of Rpn4, indicating that NAD is specific for the Ub-dependent degradation of Rpn4.
Lys-187 and NAD Constitute a Portable Ubr2-dependent Degron-A Ub-dependent degron is composed of a ubiquitination site and a ubiquitination signal, usually a short primary sequence or a structural feature recognized by a specific Ub ligase. The identification of Lys-187 as the preferential ubiquitination site and NAD as an essential sequence for the Ub-dependent degradation of Rpn4 prompted us to determine whether Lys-187 and NAD (including minimal flanking sequences) constitute a sufficient degron. To this end, we fused residues 172-229 of Rpn4 to the otherwise stable ␤-galactosidase reporter protein and measured the stability of the resulting fusion protein Rpn4 172-229 -␤galactosidase. As shown in Fig. 4 (lanes 1-4), Rpn4 172-229 -␤-galactosidase Ubr2 was rapidly degraded, indicating that Lys-187 and NAD do form a sufficient and portable degron. Interestingly, substitution of Lys-187 with arginine inhibited the degradation of Rpn4 172-229 -␤-galactosidase (Fig. 4, lanes 9 -12), indicating that the implanted Lys-187 functions as the ubiquitination site in the fusion protein. To rule out the possibility that a cryptic degron was created in the Rpn4 172-229 -␤-galactosidase fusion, we evaluated the stability of Rpn4 172-229 -␤-galactosidase in ubr2⌬, where Ubr2, the cognate Ub ligase for Rpn4, was absent. Pulse-chase analysis revealed that Rpn4 172-229 -␤-galactosidase was stabilized in the ubr2⌬ mutant (Fig. 4, lanes 5-8). Thus, Lys-187 and NAD constitute a portable degron specifically targeted by Ubr2. We postulate that NAD acts as the ubiquitination signal.

DISCUSSION
Lysine selection by the RING E3s is a central question in the field of protein ubiquitination. Although several models exemplified by the "effective concentration" and the "hit-and-run" models have been proposed to explain the process of lysine selection (25)(26)(27)(28), these models can only interpret the data of ubiquitination of one substrate but not the other. It is well known that many substrates of the Ub-proteasome system can be ubiquitinated at multiple lysines. However, in every case that we are aware of, it has never been determined if one lysine is preferred when all other lysines are also available. Extensive analysis of several substrates including Gcn4, p53, and c-Jun show that no single Lys-to-Arg substitution affects protein degradation, suggesting that it is rather nonselective with respect to which lysine of these substrates is ubiquitinated (11,12,29). However, strictly speaking, this type of analysis only demonstrates whether one of the lysines is essential for substrate ubiquitination but is unable to define a potential preferential ubiquitination site. Using an approach of adding back individual lysines to a lysine-free Sic1 allele, Petroski and Deshaies (10) recently demonstrated that each one of the six N-terminal lysines is ubiquitinated and is sufficient to mediate the degradation of Sic1. Although these authors showed that the Ub chains attached at different N-terminal lysines influence the rate of in vitro proteasomal proteolysis, the in vivo half-lives of the Sic1 mutants carrying different lysines are quite similar (10). It is unknown whether one of these N-terminal lysines of Sic1 behaves as a preferred ubiquitination site.
Using a similar "adding-back" approach, we identified six lysines in the N-terminal domain of Rpn4, each of which was able to sustain the degradation of Rpn4. Remarkably, among these six lysines, Lys-187 was ubiquitinated more efficiently than the others, and the turnover rate of Rpn4 via Lys-187 was also markedly higher than that by other individual lysines. Consistently, Lys-187 and NAD constitute an efficient and portable Ubr2-dependent degron. More importantly, both in vivo and in vitro ubiquitination assays demonstrated that Lys-187 is selected for ubiquitination when all other lysines are also available. To the best of our knowledge, this is the first demonstration of a preferential ubiquitination site that is selected from a group of lysines susceptible for ubiquitination in a substrate. It is currently unclear how the Ub ligase Ubr2/ Rad6 selects Lys-187 over other lysines. One possibility is that Lys-187 is proximally positioned to the active center of the Ubr2/Rad6 Ub ligase, as the primary sequence of Rpn4 suggests that Lys-187 is closer to the NAD ubiquitination signal than other lysines. Further investigation of the molecular details of Rpn4 ubiquitination, e.g. deciphering the atomic structure of the Ubr2⅐Rad6⅐Rpn4 complex, will shed new light on the general mechanism of lysine selection.
It remains an open question whether the presence of multiple Ub chains on a protein substrate is essential for its degradation by the proteasome. Our current study demonstrates that the Ub-dependent degradation of Rpn4 can be mediated efficiently by six individual lysines. Consistent with our data, a recent report has shown that each one of the six N-terminal lysines is sufficient for the degradation of Sic1 (10). Therefore, it appears that a single multi-Ub chain is sufficient to sustain efficient degradation of at least a subset of proteasomal substrates. What would be the physiological meaning of having multiple Ub chains on a substrate? Given the presence of multiple Ub chain receptors including the intrinsic proteasome subunits (Rpn10 and Rpt5) and proteasomeassociated UBA domain proteins in the cell (30), it is likely that multiple Ub chains may slow the dissociation of a substrate from the proteasome, providing the proteasome with sufficient time to unfold and translocate the substrate into the proteolytic chamber. This high affinity conferred by multiple Ub chains, although not an absolute requirement for the degradation of some substrates such as Rpn4 and Sic1, may be essential for others, especially those that are difficult to unfold. We reason that the structural features of protein substrates determine whether a single Ub chain is enough or multiple Ub chains are required for their degradation. For instance, a single multi-Ub chain may be sufficient for the degradation of substrates with loosely structured segments such as unfolding initiation sites.