Regulation of p53 by the ubiquitin-conjugating enzymes UbcH5B/C in vivo.

p53 levels are regulated by ubiquitination and 26 S proteasome-mediated degradation. p53 is a substrate for the E3 ligase Mdm2, however, the ubiquitin-conjugating enzymes (E2s) involved in p53 ubiquitination in intact cells have not been defined previously. To investigate the E2 specificity of Mdm2 we carried out an in vitro screen using a panel of ubiquitin E2s. Of the E2s tested only UbcH5A, -B, and -C and E2-25K support Mdm2-mediated ubiquitination of p53. The same E2s also support Mdm2 auto-ubiquitination. Small interfering RNA-mediated knockdown of UbcH5B/C causes accumulation of Mdm2 and p53 in unstressed cells. We show that suppression of UbcH5B/C inhibits p53 ubiquitination and degradation. Despite up-regulating the level of nuclear p53, UbcH5B/C knockdown does not on its own result in an increase in p53 transcriptional activity or sensitize p53 to activation by the therapeutic drugs doxorubicin and actinomycin D. We provide evidence that Mdm2 is responsible, at least in part, for repression of the transcriptional activity of the accumulated p53. In MCF7 cells levels of UbcH5B/C are reduced by doxorubicin and actinomycin D. This observation and the sensitivity of p53 expression to levels of UbcH5B/C raise the possibility that E2 regulation could be involved in signaling pathways that control the stability of p53. Our data indicate that UbcH5B/C are physiological E2s for Mdm2, which make a significant contribution to the maintenance of low levels of p53 and Mdm2 in unstressed cells and that inhibition of p53 ubiquitination and degradation by targeting UbcH5B/C is not sufficient to up-regulate p53 transcriptional activity.

In normal unstressed cells the levels and activity of the p53 tumor suppressor are kept low. p53 is stabilized, and its transcriptional activity is up-regulated following diverse stresses, including ionizing and UV radiation, genotoxic drugs, and the inappropriate activation of oncogenes. This regulates the expression of multiple target genes and leads to cell cycle arrest or apoptosis, thus preventing damaged cells from proliferating (1,2). Most tumor cells have escaped this p53 surveillance. p53 is frequently inactivated by mutation. However, there are also many tumors expressing wild-type p53, which is inactivated by other mechanisms, including overexpression of Mdm2 and loss of p14ARF. Activation of the endogenous p53 pathway is an attractive approach to treatment of tumors expressing wildtype p53 (3,4). The precise mechanisms by which the activity of p53 is regulated are thus the focus of much interest. The maintenance of low levels of p53 in cells involves its ubiquitination and consequent targeting for degradation by the 26 S proteasome (5). Because ubiquitination is involved in many key processes, there is also considerable general interest in components of the ubiquitination pathway as therapeutic targets for treatment of diverse diseases. Clinical trials have been carried out with the proteasome inhibitor PS-341. This displays antitumor activity and has been approved for therapy of patients with multiple myeloma (6 -8).
Ubiquitination of proteins occurs through the sequential actions of three enzymes (9,10). Initially ubiquitin is activated by the ubiquitin-activating enzyme (E1). 1 Ubiquitin is then transferred from the E1 to a ubiquitin-conjugating enzyme (E2). A ubiquitin ligase (E3) then facilitates transfer of ubiquitin from an E2 to the substrate. There is one human ubiquitin E1 and multiple ubiquitin E2s (10,11). Mdm2 acts as an E3 ligase for p53 (12) and can promote its ubiquitination and degradation in vivo (13,14). Mdm2 also "auto-ubiquitinates" and is itself targeted for degradation by the proteasome (15). In addition to regulating p53 protein expression Mdm2 can inhibit p53 transcriptional activity by binding to its transactivation domain (16,17), and it may directly repress basal transcription from p53-responsive promoters (18). We have recently observed that Mdm2 can promote conjugation of the ubiquitin-like protein NEDD8 (neural precursor cell-expressed developmentally down-regulated) to p53 (19). This modification also represses the transcriptional activity of p53. The relative contribution to inhibition of p53 of Mdm2-dependent ubiquitin and NEDD8 conjugation and transcriptional repression through direct binding to p53 are not clear. Loss of Mdm2 in mice is embryonic lethal in a p53-dependent manner (20,21). In transgenic adult mice, which express lower levels of Mdm2 than wild-type mice, the transactivation and apoptotic activities of p53 are enhanced (22). Agents that bind to Mdm2 and block its association with p53 increase both the levels and transcriptional activity of p53 in vivo. These include high affinity Mdm2 binding peptides, the Mdm2-specific antibody 3G5 (23), and the Nutlin family of small molecules (24). These and other observations indicate that Mdm2 is a key negative regulator of p53 (25,26).
Mdm2 is itself a target for p53 transcriptional regulation. Mdm2 is induced by p53 resulting in a negative feedback loop (27). The Mdm2 structural homologue MdmX is also an important regulator of p53. MdmX binds to p53 and inhibits its transcriptional activity but does not itself ubiquitinate p53 in cells (28), although a low level of E3 ligase activity has been observed in vitro (29). Mdm2 and MdmX heterodimerize through their ring fingers (30). Strikingly, ablation of MdmX also results in p53-dependent embryonic lethality in mice (31,32).
Although auto-ubiquitination of Mdm2 and Mdm2-mediated ubiquitination of p53 have been reconstituted in vitro using purified E2s (12,33,34), the E2 specificity of Mdm2 has not been investigated in detail, and the one or more ubiquitin E2s involved in regulating the Mdm2/p53 pathway in intact cells have not been determined previously. Ubiquitin E2s consist of a conserved core domain of ϳ150 amino acids. They have an active site cysteine residue, which forms a thiolester bond with ubiquitin transferred from the E1. Ubiquitination, and by inference ubiquitin E2s, are involved in regulating many cellular processes. However, the precise physiological roles of many of the ubiquitin E2s are not well defined, particularly in mammalian cells. There is evidence that E2s have unique functions, presumably arising from their specificity for particular E3s. Genetic studies in Saccharomyces cerevisiae indicate that Ubc2 (RAD6) is central to DNA repair, Ubc3 (CDC34) is required for the G 1 -S transition, and Ubc4/5 are required for viability (10).
Regulation of E2 levels or activity could provide a mechanism to alter the stability of key proteins. Studies using rat tissue extracts indicate that E2s can be rate-limiting for ubiquitination (35). The expression of S. cerevisiae Ubc4 and -5 is growth-and heat shock-regulated (36). The levels of specific mammalian E2s have also been reported to be regulated by the cell cycle and by agents, including interferons ␣ and ␥, insulin, insulin-like growth factor 1, amyloid-␤, and herpes simplex virus (37)(38)(39)(40)(41)(42)(43)(44)(45). E2 activity can be influenced by phosphorylation. CDC34 and its homologue Ubc3b are substrates for protein kinase CK2 (46), and UbcH1 (HHR6) is phosphorylated by CDK1 and -2 (47).
In this study we identified ubiquitin E2s involved in the Mdm2/p53 pathway in intact cells. A panel of E2s were tested for their ability to support the ubiquitin ligase activity of Mdm2 in vitro. We show that there is specificity in E2 usage by Mdm2. Of the E2s tested only members of the UbcH5 family (A-C) and E2-25K are able to support the ubiquitin ligase activity of Mdm2. We observed that the pattern of Mdm2-mediated ubiquitination of p53 in vitro is dependent on the E2 used. Using siRNA-mediated knockdown we show that UbcH5B/C make a significant contribution to the maintenance of low levels of Mdm2 and p53 in unstressed cells. In contrast, suppression of UbcH5A or E2-25K has little effect on expression of Mdm2 and p53 in the cell lines examined. Levels of UbcH5B/C protein expression were found to be higher than those of UbcH5A, possibly accounting for the preferential usage of UbcH5B/C in intact cells. The balance of activities of the Mdm2/p53 pathway is such that the p53 accumulated following UbcH5B/C knockdown is not transcriptionally active, the level of Mdm2 being sufficient to maintain transcriptional repression of p53. We find that expression of UbcH5B/C is reduced by the p53 stabilizing therapeutic drugs doxorubicin and actinomycin D. Because p53 expression is sensitive to the levels of UbcH5B/C, it is conceivable that E2 regulation could play some part in signaling to the p53 pathway.

EXPERIMENTAL PROCEDURES
Antibodies-The antibodies used were 4B2 for Mdm2; DO-1 for p53 unless otherwise indicated; H-81 for CDC34 (Santa Cruz Biotechnology); UBC4 (C-15) for UbcH5B/C (Santa Cruz Biotechnology), which was precleared by absorption with GST-UbcH5A; Ab-1 for p21 (Oncogene); L27 for CD20 (BD Biosciences); and Ab-1 for actin (Calbiochem). A rabbit polyclonal anti-E2-25K antibody was purchased from Affiniti Research, a rabbit anti-ubiquitin antiserum was purchased from Sigma and a sheep anti-UbcH5A polyclonal antiserum raised against purified full-length UbcH5A was provided by Dr. R. T. Hay. This was affinity-purified and then precleared by absorption with GST-UbcH5B and -C. Where indicated, cross-reactivity was removed by incubating antibodies (15-30 g) three times for 4 h with GST-E2s (ϳ50 g) bound to glutathione-Sepharose beads in 0.5 ml of 5% dried milk, 0.1% Tween 20 in PBS.
Gel Electrophoresis and Western Blotting-Cells were washed twice with PBS at 4°C. Cell extracts were prepared by direct lysis into SDS-urea electrophoresis sample buffer: 100 mM Tris, pH 6.8, 4% SDS, 8 M urea, 20% glycerol, 20 mM EDTA, 0.014% bromphenol blue. DNA was sheared by passage through a 25-gauge needle, and protein concentrations were measured using the BCA protein assay (Pierce). Dithiothreitol was added to a final concentration of 100 mM, samples were heated for 5 min at 95°C, and proteins were resolved by SDS-PAGE. Gels were transferred onto nitrocellulose for 16 h, at 25 mA, or for 1 h at 150 mA, and membranes were processed as described in Midgley et al. (34). Membranes probed for ubiquitin were boiled in de-ionized water prior to blocking to expose epitopes in ubiquitin. Peroxidaseconjugated secondary antibodies were supplied by Jackson Immuno-Research Laboratories and used at a dilution of 1/10,000. Bound antibodies were detected by enhanced chemiluminescence (Amersham Biosciences) or using SuperSignal West Dura Extended Duration Substrate (Pierce).
Expression and Purification of Recombinant Proteins-Untagged human Mdm2 was prepared from bacterial inclusion bodies as described previously (34). Human E1 was purified from recombinant baculovirusinfected insect cells by affinity chromatography on ubiquitin-Sepharose as described in Desterro et al. (51). Human GST-HHR6A, HHR6B, UbcH2 (52), UbcH5A, UbcH5B, UbcH5C, and UbcH8 were expressed in bacteria and purified as described previously (53). GST-ubiquitin containing a protein kinase A (PKA) consensus site between the ubiquitin and the tag was expressed in bacterial, purified, and phosphorylated by PKA as described in Tongaonkar et al. (54). GST was cleaved from E2s and ubiquitin by thrombin digestion prior to use in ubiquitination assays, and free thrombin was removed by incubation with benzamidine-Sepharose. UbcH6, -7, and -10 and E2-25K proteins were supplied by Affiniti Research, and CDC34 was donated by Dr. R. T. Hay. His 6 -p53 was expressed in bacteria. Following induction with isopropyl-1thio-␤-D-galactopyranoside at room temperature p53 in the soluble fraction was partially purified on a heparin-Sepharose column as described p53 Regulation by UbcH5B/C previously (55). Fractions containing full-length p53 were further purified on Ni 2ϩ -NTA-agarose.
Cell Culture and Transfection-MCF7 breast tumor cells and U2-OS osteosarcoma cells were cultured in DMEM and H1299 lung carcinoma cells in RPMI. Media were supplemented with 10% fetal calf serum and gentamycin (50 g/ml), and cells were kept at 37°C, 5% CO 2 in a humidified atmosphere. Unless otherwise indicated MCF7 and U2-OS cells were transfected with 1.5 g of pSUPER construct per 10-cm dish. Transfection of these cells lines with plasmids was performed with FuGENE 6 transfection reagent (Roche Applied Science) following the manufacturer's instructions. Cells were washed after 16 h, and fresh medium was added. Unless otherwise indicated H1299 cells were transfected with 7 g of pSUPER constructs per 10-cm dish. Transfection of this cell line was performed using the calcium phosphate method essentially as described in Xirodimas et al. (15). Transfection with single siRNA synthetic duplexes (100 nM) or SMART pools (200 nM) was carried out using Oligofectamine (Invitrogen) according to the manufacturer's instructions.
CD20 Enrichment of Transfected Cells-MCF7 cells in 10-cm dishes were transfected with pSUPER constructs and pCMVCD20 (1 g). At the indicated time after transfection, cells were detached by washing with 3 mM EDTA in PBS. Cells were resuspended thoroughly in DMEM, 10% FCS, and incubated for 20 min at 4°C with magnetic Dynabeads Pan Mouse IgG (Dynal Biotech) coated with anti-CD20 antibody (1 ϫ 10 7 beads and 1 g of antibody per dish). The beads were isolated with a magnet, and the selected cells were washed twice with PBS before lysis for RNA extraction or Western blotting.
Immunofluorescence-Cells were seeded onto NUNC Permanox slides. 36 h after transfection with synthetic siRNA duplexes, cells were fixed with ice-cold methanol-acetone and incubated with primary antibodies followed by fluorescence isothiocyanate-conjugated donkey antimouse secondary antibodies (Jackson ImmunoResearch) as described previously (15).
Quantitative Detection of ␤-Galactosidase-Cells in 24-well plates were washed twice with PBS and lysed on the plate with 150 l of passive lysis buffer (Promega). 150 l of substrate solution containing 80 g/ml chlorophenol red ␤-D-galactopyranoside (Roche Applied Science) 0.5 mM MgCl 2 , 23 mM ␤-mercaptoethanol in 0.1 M sodium phosphate buffer, pH 7.5, was added to 15 l of extract in a 96-well plate. The assay was quantitated by absorbance measurements at a wavelength of 590 nm in a plate reader. Luciferase activities were measured using the Berthold microplate luminometer LB 96V and used to control for transfection efficiency.
RNA Preparation and Real-time RT-PCR-Total RNA was extracted using RNeasy columns (Qiagen) according to the manufacturer's instructions, including an on-column DNase treatment step. RNA (300 -600 ng) was incubated with random hexamers and Superscript II reverse transcriptase (Invitrogen) to generate cDNA. Real-time PCR was carried out with an ABI Prism 7700 sequence detector using the following protocol: 50°C for 2 min, 95°C for 10 min, and 40 cycles of 95°C for 15 s and 60°C for 1 min. For ubiquitin E2s, primers and 6-FAM/TAMRA-labeled probes were as used in Okamoto et al. (56). Other specific probes and primers were as follows: p53 primers: 5Ј-CAGCCAAGTCTGTGACTTGCA-3Ј, 5Ј-GTGTGGAATCAACCCACAGCT-3Ј, probe

RESULTS
The Ubiquitin Ligase Activity of Mdm2 Is Specifically Supported by UbcH5A, -B, and -C and E2-25K in Vitro-Mdm2mediated ubiquitination of p53 has previously been reconstituted in vitro with the E2s UbcH5A and -B (12,34). However, the ubiquitin E2 specificity of Mdm2 has not been investigated in detail. To address this we screened a range of characterized E2s for the ability to support the ubiquitin ligase activity of Mdm2. Purified E2s were matched for protein concentration and tested for functionality by their ability to form thiolesters with 32 P-labeled ubiquitin (data not shown). We first looked at Mdm2 auto-ubiquitination in the absence of p53. Following incubations with purified E1, [ 32 P]ubiquitin and the indicated E2s, proteins were resolved by SDS-PAGE under reducing FIG. 1. In vitro specificity of Mdm2 for ubiquitin E2s. A, Mdm2 auto-ubiquitination. In vitro ubiquitination assays were carried out with E1, 32 P-labeled ubiquitin, and the indicated E2s in the presence or absence of Mdm2. Gels were run under reducing conditions, and [ 32 P]ubiquitin was detected by phosphorimaging. B, Mdm2-mediated ubiquitination of p53. In vitro ubiquitination assays were carried out with p53 and the indicated E2s (where s is for sugar) in the presence or absence of Mdm2. p53 was detected by Western blotting with the monoclonal antibody DO-1.
p53 Regulation by UbcH5B/C conditions (Fig. 1A). Ubiquitinated Mdm2 migrated as a very high molecular weight species near the top of the gel. The identity of this species was confirmed by immunodepletion using a mixture of anti-Mdm2 antibodies (data not shown). Of the E2s tested, only UbcH5A, -B, and -C and E2-25K were able to support auto-ubiquitination of Mdm2. E2-25K has not previously been shown to support the ubiquitin ligase activity of Mdm2. Typically, the incorporation of ubiquitin into Mdm2 was 7-to 8-fold higher with UbcH5 family members than with E2-25K. Mdm2-mediated ubiquitination of p53 was also observed exclusively with members of the UbcH5 family and E2-25K (Fig. 1B). In these assays p53 was detected by Western blotting using the anti-p53 antibody DO-1. In contrast to a recent report (57) we did not observe Mdm2-mediated ubiquitination of p53 with HHR6A or -B. The pattern of p53 ubiquitination was similar with UbcH5A, -B, and -C and consisted predominantly of relatively low molecular weight ubiquitinated species. E2-25K supported predominantly high molecular weight ubiquitination of p53. These observations show that there is specificity for E2 usage by Mdm2, that the E2s which support Mdm2 auto-ubiquitination can also support Mdm2mediated ubiquitination of p53, and that Mdm2 can display different biochemical properties dependent on the E2 that is used.
Members of the UbcH5 family are highly homologous. UbcH5B and -C are 97% identical at the amino acid level, and with a difference of only four amino acids, it is likely that UbcH5B and -C have similar activities. UbcH5A is more divergent, being 89 and 88% identical to UbcH5B and -C, respectively, at the amino acid level (58,59). UbcH5A and UbcH5B/C have been observed to have different biochemical properties in vitro (60).
siRNA-mediated Knockdown of UbcH5B/C Causes Accumulation of p53 and Mdm2 Protein-To investigate the involvement of UbcH5 family members and E2-25K in regulation of p53 and Mdm2 expression in intact cells, we used RNA interference (RNAi) to knockdown specific E2s. Target sequences were inserted into the pSUPER vector, which directs the synthesis of small interfering RNA (siRNA) (50). We targeted the E2s able to support the ubiquitin ligase activity of Mdm2 in vitro and as a negative control CDC34 with which no ubiquitin ligase activity of Mdm2 was observed. Due to their high degree of homology and the consequent likelihood of functional redundancy between UbcH5B and -C, we generated single pSUPER constructs that would suppress both UbcH5B and -C. We employed MCF7 human breast adenocarcinoma cells, which express wild-type p53 and Mdm2. In these cells p53 levels and activity are sensitive to agents that disrupt the p53-Mdm2 interaction, therapeutic drugs, and p14ARF. These cells were co-transfected with pSUPER constructs and a plasmid encoding the CD20 cell surface marker. Cells were harvested after 48 h, and the transfected population of cells was captured using anti-CD20 antibody bound to magnetic beads. For each of the E2s targeted several sequences were tested. pSUPER plasmids were selected that gave 75% or greater knockdown of their target mRNA ( Fig. 2A).
To facilitate detection of siRNA-mediated knockdown of E2 proteins, anti-UbcH5 antibodies were characterized using purified recombinant E2s (Fig. 2B). A UbcH5A-selective antibody was generated by preclearing a polyclonal antiserum raised against purified UbcH5A by adsorption of cross-reacting antibodies with GST-UbcH5B and -C. A UbcH5B/C-selective antibody was generated by preclearing UBC4 (C-15) anti-peptide polyclonal by absorption of cross-reacting antibodies with GST-UbcH5A. To date we have been unable to generate an antibody that can distinguish between the highly homologous E2s UbcH5B and -C. Transfection of MCF7 cells with the pSUPER constructs specifically reduced expression of their target proteins (Fig. 2C). CDC34, UbcH5A, UbcH5B/C, and E2-25K protein expression was reduced by a similar extent. Knockdown of UbcH5C reduced the total level of UbcH5B/C immunoreactivity, whereas this was unaffected by UbcH5B knockdown. This suggests that in MCF7 cells the level of UbcH5C protein expression is higher than that of UbcH5B. Knockdown of UbcH5C or UbcH5B/C resulted in the accumulation of endogenous p53 and Mdm2 protein (Fig. 2C). To confirm that the accumulation of p53 and Mdm2 was not due to an off-target effect of the siRNA, two additional pSUPER constructs that target UbcH5B/C were tested. These pSUPER constructs reduced the mRNA and protein levels of UbcH5B/C and caused accumulation of p53 and Mdm2 protein (Fig. 2D). To investigate the generality of these observations, the effects of ubiquitin E2 knockdown were also determined in U2-OS human osteosarcoma cells, which express wild-type p53 and Mdm2, and in H1299 human lung carcinoma cells, which express Mdm2 and are p53 null. In these cell lines transfection with pSUPER constructs specifically reduced expression of the target proteins (Fig. 3A). Knockdown of UbcH5B/C caused accumulation of endogenous p53 and Mdm2 protein in U2-OS cells and of endogenous Mdm2 in H1299 cells. To confirm that UbcH5B/C are required for Mdm2-dependent degradation of p53, the effect of UbcH5B/C knockdown on the ability of ectopically expressed Mdm2 to promote degradation of p53 was examined. U2-OS cells were transfected with a transcriptionally inactive p53 mutant (His-273) to avoid the deleterious effects associated with prolonged expression of wild-type p53. Overexpression of Mdm2 promoted degradation of exogenous p53 (Fig. 3B). UbcH5B/C knockdown caused accumulation of ectopically expressed Mdm2 and attenuated Mdm2-dependent p53 degradation. These data indicate that UbcH5B/C are required for the maintenance of low levels of p53 and Mdm2 in a range of cell lines. In MCF7 cells this principally involves UbcH5C. UbcH5B/C knockdown caused accumulation of Mdm2 in a p53 null cell line showing that UbcH5B/C can regulate Mdm2 independently of p53.
In the cell lines examined, no effect on p53 and Mdm2 expression was observed following knockdown of the other E2s targeted, including UbcH5A and E2-25K. Knockdown of UbcH5A or E2-25K in combination with UbcH5B/C, by cotransfection with pSUPER plasmids, did not increase p53 and Mdm2 levels more than suppression of UbcH5B/C alone (data not shown). In vitro UbcH5A supported the ubiquitination of p53 equally as well as UbcH5B/C. To investigate whether differences in E2 usage in vivo could be related to differences in protein expression, protein levels of UbcH5 family members were determined by Western blotting using recombinant proteins as standards. The level of protein expression of UbcH5B/C was 3-to 5-fold higher than that of UbcH5A in the cell lines examined (Fig. 3C).

UbcH5B/C Knockdown Attenuates p53 Ubiquitination without Causing Gross Changes in the General Pattern of Ubiquitin
Conjugates-To investigate the effect of E2 suppression on the ubiquitination of endogenous p53, MCF7 cells were transfected with pSUPER plasmids and harvested by direct lysis into SDSurea electrophoresis sample buffer. This denaturing buffer inhibits de-ubiquitinating enzymes and thus preserves ubiquitinprotein conjugates (61). A ladder of high molecular weight p53 species was detected in control samples. This ladder results from conjugation of multiple ubiquitin or ubiquitin-like proteins to p53. Using pSUPER constructs that target two different sequences in UbcH5B/C, to control for off-target effects, we observed that UbcH5B/C knockdown reduced the relative lev-p53 Regulation by UbcH5B/C els of the highest molecular weight species of p53. To demonstrate this more clearly the amount of control extract loaded onto SDS-PAGE gels was increased 3-fold so that the level of unmodified p53 was equivalent to that in extracts from pSUPER-UbcH5B/C transfected cells (Fig. 4A). The proportion of p53 species with electrophoretic mobility consistent with conjugation of three or more ubiquitin molecules was markedly reduced by UbcH5B/C suppression. In contrast, the proportion of p53 species with mobility consistent with conjugation of one or two ubiquitin molecules was increased to some extent. Knockdown of UbcH5A or E2-25K in combination with UbcH5B/C did not reduce the intensity of these species (data not shown). The effect of UbcH5B/C suppression on the pattern of p53 modification was also examined in cells treated with the proteasome inhibitor MG132 (Fig. 4B). MCF7 cells were transfected with a control synthetic siRNA duplex or a duplex that targets the same sequence as the UbcH5B/C pSUPER construct previously used. In the presence of MG132, UbcH5B/C suppression decreased the abundance of p53 conjugates with molecular weights consistent with modified by two or more ubiquitin molecules. To confirm that the high molecular weight species represent ubiquitinated forms of p53, MCF7 cells were Western blots were probed with an antibody raised against full-length UbcH5A before and after preclearing with UbcH5B/C or UBC4 (C-15) and before and after preclearing with UbcH5A. C, Western blots for target E2s, p53 and Mdm2. MCF7 cells were harvested 64 h after transfection with CD20 plasmid and pSUPER constructs. Transfected cells were enriched by CD20-mediated capture and endogenous protein expression determined by Western blot analysis. p53 was detected using the rabbit polyclonal antibody CM1. D, targeting UbcH5B/C with additional pSUPER constructs. MCF7 cells were transfected with pSUPER plasmids targeting alternative sequences within UbcH5B/C and analyzed for expression of the indicated proteins and for E2 mRNA levels as described in C.
p53 Regulation by UbcH5B/C co-transfected with His 6 -tagged ubiquitin and pSUPER plasmids. His 6 -tagged conjugates were isolated on Ni 2ϩ -agarose and analyzed by Western blotting for p53. The electrophoretic mobility of high molecular weight p53 species in whole cell lysates from cells treated with MG132 was similar to the mobility of ubiquitinated p53 species isolated on Ni 2ϩ -agarose (Fig. 4C). The presence of these p53 species in the Ni 2ϩ -agarose-purified material was dependent on transfection with His 6ubiquitin indicating that they contain ubiquitin. Consistent with observations made with whole cell lysates, UbcH5B/C suppression reduced the relative amount of conjugates containing three or more ubiquitin molecules but increased the intensity of conjugates containing one or two ubiquitin molecules. The mono-and di-ubiquitinated species of p53 present after UbcH5B/C suppression could represent ubiquitination of p53 that is supported by residual UbcH5B/C, because the knockdown is incomplete. E3 up-regulation following UbcH5B/C knockdown may also be a contributory factor.
Human UbcH5A, -B, and -C are closely related to S. cerevisiae Ubc4 and Ubc5. Simultaneous deletion of these E2s in S. cerevisiae results in a severe reduction in high molecular weight ubiquitin conjugates (36). We examined the effect of E2 knockdown on the general pattern of ubiquitination in MCF7 cells. The transfection efficiency was sufficient to detect a reduction in UbcH5B/C expression in SDS-urea whole cell lysates (Fig. 4D, top panel). There were no dramatic changes in the pattern or level of conjugates detected by Western blot analysis using an anti-ubiquitin antibody (Fig. 4D, middle panel).
UbcH5B/C Suppression Stabilizes p53 and Causes Accumulation of p53 in the Nucleus-To determine the effect UbcH5B/C suppression on p53 protein stability, 36 h after transfection with siRNA duplexes MCF7 cells were incubated with cycloheximide (10 g/ml) to inhibit de novo protein synthesis and harvested after the indicated times (Fig. 5A). UbcH5B/C suppression increased the half-life of endogenous p53 from 25 min to over 180 min.
It has been reported that Mdm2-mediated mono-ubiquitination of p53 causes its export from the nucleus (62). Given that UbcH5B/C knockdown results in an increase in the proportion of mono-and di-ubiquitinated species of p53, it was of interest to determine the intracellular localization of the accumulated p53. MCF7 cells were transfected with synthetic siRNA duplexes. Cells were fixed in acetone-methanol and stained for p53 or Mdm2. Suppression of UbcH5B/C caused accumulation of both p53 and Mdm2 in the nucleus (Fig. 5B). A number of agents, including the proteasome inhibitor MG132 and p14ARF, cause an increase in nucleolar Mdm2 (15). p53 and Mdm2 did not accumulate in the nucleolus following E2 knockdown.
p53 Accumulated following UbcH5B/C Knockdown Is Transcriptionally Repressed-Reporter assays were carried out in MCF7 cells to determine whether UbcH5B/C suppression caused an increase in p53 transcriptional activity. The effect of E2 knockdown was compared with that of super thioredoxin insert protein (STIP). STIP consists of a high affinity Mdm2 binding peptide, fused to thioredoxin, which disrupts the interaction between Mdm2 and p53 (23). pSUPER constructs and STIP were co-transfected with a p53-responsive reporter construct RGC⌬-Fos-LacZ and pSVE-luciferase to monitor transfection efficiency. At the indicated times post-transfection cells were harvested for Western blotting and for ␤-galactosidase assays. Expression of STIP increased p53 protein levels more rapidly than UbcH5B/C knockdown (Fig. 6A), presumably reflecting the time taken for the E2 levels to decrease as a result of the RNAi response. ␤-Galactosidase reporter activity was increased by STIP but was not affected by UbcH5B/C suppression (Fig. 6B). These data indicate that knockdown of UbcH5B/C does not increase p53 transcriptional activity.
Mdm2 accumulates following UbcH5B/C knockdown. A pos- p53 Regulation by UbcH5B/C sible explanation for the lack of p53 activation following knockdown of UbcH5B/C is that it is held inactive through binding to Mdm2. To address this possibility we used STIP, which disrupts the interaction between p53 and Mdm2. Cells were first transfected with control or UbcH5B/C siRNA duplexes. After 24 h the cells were replated and then transfected with empty thioredoxin vector (TRX) or STIP along with RGC⌬-Fos-LacZ and pSVE-luciferase. 20 h later cells were harvested for Western blotting and for ␤-galactosidase assays. STIP up-regulated p53 reporter activity in cells in which UbcH5B/C was suppressed (Fig. 6D). Reporter activity in UbcH5B/C siRNA-and STIP-transfected cells was higher than in cells transfected with control siRNA and STIP, correlating with the higher level of p53 protein expression caused by UbcH5B/C knockdown, at the time point examined (Fig. 6C). This suggests that the transcriptional activity of the p53 accumulated following suppression of UbcH5B/C is repressed, at least in part, by Mdm2. STIP can also probably target MdmX (63), and it is thus possible that p53 Regulation by UbcH5B/C MdmX could contribute to inactivation of the accumulated p53.
The effect of UbcH5B/C knockdown on the mRNA levels of endogenous p53 target genes was also investigated. We compared this to the effect of Mdm2 knockdown. MCF7 cells were transfected with siRNA duplexes and harvested the indicated time after transfection (Fig. 7A). An increase in p53 protein expression was observed following suppression of Mdm2, which occurred within 24 h of transfection with Mdm2 siRNA. However, p53 expression declined at longer time points, even though levels of Mdm2 remained low. UbcH5B/C knockdown resulted in a robust increase in p53 protein levels, which was sustained for at least 72 h. mRNA levels of endogenous p53 target genes were determined by real-time RT-PCR 36 h after transfection of MCF7 cells with synthetic siRNA duplexes (Fig.  7B). At this time point levels of p53 protein were maximal following UbcH5B/C knockdown, and there was an increase in protein expression of the CDK inhibitor p21 but levels of p53 following suppression of Mdm2 were no longer elevated. Suppression of UbcH5B/C did not result in a significant increase in the mRNA levels of the p53 target genes Mdm2, p21, or BAX confirming that p53 transcriptional activity is not increased. In contrast, up-regulation of p21 and BAX mRNA occurred following Mdm2 knockdown. Suppression of UbcH5B/C thus resulted in an increase in protein levels of p21 without an increase in p21 mRNA levels (Fig. 7, A and B). p21 is itself ubiquitinated and undergoes proteasomal degradation (64). UbcH5B/C may thus also be involved in regulation of p21 protein stability.
Knockdown of UbcH5B/C Does Not Enhance the Effect of Doxorubicin and Actinomycin D on p53 Transcriptional Activity-We speculated that accumulation of p53 as a result of suppression of UbcH5B/C might sensitize p53 to activation by anti-tumor therapeutic drugs. Doxorubicin or actinomycin D were added to MCF7 cells 24 h after transfection with control or UbcH5B/C siRNA duplexes. After a further 16 h cells were harvested for analysis of mRNA and protein expression (Fig. 8). Doxorubicin binds DNA and inhibits topoisomerase II resulting in alterations in DNA transcription and replication. Actinomycin D intercalates into DNA and at concentrations of 1-10 nM is regarded as being a selective inhibitor of ribosomal RNA transcription and at concentrations above 10 nM also inhibits RNA polymerase II-dependent transcription (65). UbcH5B/C knockdown did not reduce the concentration of doxorubicin or actinomycin D required for up-regulation of p53 target genes. In the case of doxorubicin, maximal induction of p53 target gene mRNA levels was reduced by suppression of UbcH5B/C (Fig. 8A). The combination of actinomycin D and UbcH5B/C knockdown caused a greater increase in p53 protein levels than actinomycin D treatment alone (Fig.  8B). Despite this, UbcH5B/C knockdown did not enhance the effect of actinomycin D on p53 target gene expression. Induction of BAX mRNA by both doxorubicin and actinomycin D was especially sensitive to inhibition by down-regulation of UbcH5B/C. The effect of doxorubicin and actinomycin D on p53 target gene mRNA levels was biphasic, induction of mRNA being followed by decrease at higher drug concentrations. Mdm2 mRNA and protein levels were particularly sensitive to this effect. This downregulation of Mdm2 may be a signaling event that contributes to regulation of p53 (66,67); however, p53 induction can occur in the absence of a decrease in Mdm2 protein expression (Fig. 9).
Treatment with genotoxic drugs stabilizes p53, but some genotoxic agents have been shown to promote degradation of Mdm2 despite causing an increase in Mdm2 protein expression due to up-regulation of Mdm2 mRNA (68). In cells in which UbcH5B/C was suppressed there was a dissociation between Mdm2 mRNA and protein levels following drug treatment. This was most clear for cells treated with doxorubicin. In the presence of 1 M doxorubicin mRNA levels of Mdm2 were somewhat higher than those in cells that had not been treated with drugs, but the level of Mdm2 protein expression was significantly lower. This suggests that destabilization of Mdm2 following treatment with genotoxic drugs may still occur in cells in which UbcH5B/C is suppressed (Fig. 8A).
Expression of UbcH5B/C Is Regulated by Therapeutic Drugs-p53 protein accumulates following a variety of stresses, including exposure to genotoxic therapeutic drugs. The regulatory events that are involved in this stabilization of p53 remain to be fully determined. We have observed that p53 levels in unstressed cells are sensitive to suppression of UbcH5B/C. The p53 pathway could thus respond to stressinduced changes in UbcH5B/C levels or activity. MCF7 cells were treated with doxorubicin (1 M) or actinomycin D (10 nM). A decrease in UbcH5B/C protein level was observed following treatment with the drugs (Fig. 9). This was rapid, occurring before increases in p53 and Mdm2 protein expression. Levels of UbcH5A and of E2-25K were not reduced. The decrease in UbcH5B/C expression was similar in magnitude to that observed in knockdown experiments, which was sufficient to lead to accumulation of p53 and Mdm2 (Fig. 2, C and D). The kinetics and magnitude of the effects of drug treatment on UbcH5B/C expression are thus consistent with an involvement of E2 regulation in the complex sequence of events regulating the activity of p53. At the drug concentrations used, Mdm2

FIG. 5. Knockdown of UbcH5B/C inhibits p53 degradation and causes accumulation of nuclear p53 and Mdm2 in MCF7 cells. A,
UbcH5B/C suppression inhibits p53 degradation. 36 h after transfection with control or UbcH5B/C synthetic siRNA duplexes MCF7 cells were incubated with cycloheximide (10 g/ml) for the indicated times. p53 protein expression was determined by Western blotting. Because UbcH5B/C knockdown increases expression of p53 the Western blot shown in the upper panel (control siRNA duplex) was exposed for longer than that in the lower panel (UbcH5B/C siRNA duplex) to allow comparison of similar intensities of p53 signal. B, immunostaining for p53 and Mdm2. MCF7 cells were transfected with control, UbcH5A, or UbcH5B/C synthetic siRNA duplexes and analyzed 36 h post-transfection by indirect immunofluorescence using DO-1 to detect endogenous p53 and 4B2 to detect endogenous Mdm2.
p53 Regulation by UbcH5B/C levels did not decrease prior to p53 induction. This indicates that, although down-regulation of Mdm2 protein expression may play some part in control of p53, it is not an obligatory step for induction of p53 in MCF7 cells.

DISCUSSION
In this study we show that UbcH5B/C are required for the maintenance of low levels of p53 and Mdm2 in unstressed cells. UbcH5B/C participate in p53 ubiquitination and in Mdm2mediated degradation of p53 in vivo, indicating that they are physiological E2s for Mdm2. Suppression of UbcH5B/C results in the accumulation of nuclear p53 but does not on its own increase p53 transcriptional activity. A reduction in UbcH5B/C expression occurs following treatment with doxorubicin and actinomycin D raising the possibility that E2 regulation may participate in signaling to p53.
We examined the ubiquitin E2 specificity of the E3 ligase Mdm2. In vitro auto-ubiquitination of Mdm2 and Mdm2-dependent ubiquitination of p53 were supported by UbcH5A, -B, and -C and by E2-25K. The pattern of Mdm2-mediated p53 ubiquitination was similar with UbcH5 family members, which gave predominantly low molecular weight ubiquitinated species. This is consistent with the observation that UbcH5B supports multiple mono-ubiquitination of p53 in vitro (69). Mdm2 promoted the formation of high molecular weight p53-ubiquitin conjugates in the presence of E2-25K. This may be related to the propensity of E2-25K to form chains of ubiquitin. E2-25K displays the unique property of mediating the formation of unanchored Lys-48-to Gly-76-linked polyubiquitin chains in vitro without the requirement for an E3 or acceptor substrate (70). Only a small proportion of the total ubiquitin added to in vitro reactions was converted to unanchored polyubiquitin chains (data not shown). Therefore E2-25K may selectively transfer polyubiquitin chains to p53 in preference to single ubiquitin molecules, or E2-25K may efficiently extend chains of ubiquitin conjugated to p53.
We show that UbcH5B/C play a significant role in regulating both p53 and Mdm2 protein levels in intact cells. In vitro studies with human UbcH5 family members and genetic studies involving their homologues in lower organisms clearly indicate that they are involved in the ubiquitination of multiple proteins (36,60). Dominant negative UbcH5B and -C have been shown to block degradation of IB␣ (60). The UbcH5 family is closely related to Ubc4 and -5 of S. cerevisiae, which together are involved in degradation of the majority of short-lived proteins in yeast (36). Suppression of UbcH5B/C altered the pattern of p53 ubiquitination without causing dramatic changes in the general profile of ubiquitin conjugates. This is consistent with the observation that overexpression of dominant negative FIG. 6. UbcH5B/C suppression alone does not increase p53-responsive reporter activity due to repression of the accumulated p53 by Mdm2. A, Western blot for p53. MCF7 cells were transfected with a p53-responsive reporter construct RGC⌬-Fos-LacZ, pSVE-luciferase, and the indicated plasmids. STIP but not the TRX control disrupts the interaction between Mdm2 and p53. Cells were harvested at the indicated time after transfection, and p53 protein expression was determined by Western blotting. B, p53-responsive reporter assays. In the same experiments described in A samples were assayed for ␤-galactosidase activity. Results are normalized to luciferase activity and expressed as a percentage of the activity in control transfected cells. Values are means Ϯ S.D. of three independent experiments each performed in duplicate. C, Western blots for p53 and Mdm2. MCF7 cells were transfected with the indicated siRNA duplex, replated after 24 h, and then transfected with TRX or STIP and RGC⌬-Fos-LacZ and pSVE-luciferase. Cells were harvested 20 h after the second transfection, and expression of the indicated endogenous proteins was determined by Western blotting. D, p53-responsive reporter assays. In the same experiments described in C, samples were assayed for ␤-galactosidase activity. Results are normalized to luciferase activity and expressed as a percentage of the activity in cells transfected with control siRNA followed by TRX vector. Values are means Ϯ range of two independent experiments each performed in duplicate.
p53 Regulation by UbcH5B/C UbcH5B or -C does not change the pattern of ubiquitination in HeLa cells (60). UbcH5B/C may thus be required for the ubiquitination of a specific subset of proteins in vivo. The human genome encodes for more E2s than that of S. cerevisiae and contains several homologues of S. cerevisiae Ubc4/5 (11,56). This provides the scope for both divergence and for greater redundancy of E2 function in humans. Alternatively, ubiquitination of p53 may be more sensitive to changes in UbcH5B/C levels than the ubiquitination of the major UbcH5B/C substrates. Consistent with this the K m of UbcH5B for Mdm2mediated ubiquitination of p53 is ϳ3 M (69), whereas the apparent K m of a rat homologue for stimulation of ubiquitination of its major substrates in rat tissue extracts is in the region of 50 nM (35).
Both UbcH5A and E2-25K were able to support the ubiquitin ligase activity of Mdm2 in vitro; however, in the cell lines tested no up-regulation of Mdm2 or p53 was observed following suppression of these E2s. This dependence on UbcH5B/C rather than UbcH5A in vivo could reflect differences in expression of the E2s. In MCF7, U2-OS, and H1299 cells UbcH5B/C protein expression was higher than that of UbcH5A. This may generally be the case, because previous studies have shown that mRNA levels of UbcH5B/C are higher than those of UbcH5A in a range of tissues (56,59). We cannot exclude the possibility that UbcH5A or E2-25K may participate in the regulation of Mdm2 and p53 under different conditions or in different cell types. Overexpression of Mdm2 to moderate levels stimulates principally multiple mono-ubiquitination of p53, however overexpression to higher levels has been reported to promote p53 polyubiquitination (62). Polyubiquitinated rather than monoubiquitinated p53 may be targeted for proteasomal degradation. Further studies are required to determine whether E2-25K plays a part in this polyubiquitination of p53. It is of interest to note that E2-25K is induced by low density lipoprotein in human macrophages (71). Up-regulation of E2-25K under these circumstances is associated with a decrease in p53 levels and with a decrease in the apoptotic potential of the cells, however, the direct causal linkage of these events has not been demonstrated. We did not detect p53 ubiquitination with HHR6A or -B in vitro; however, these E2s have been found in complexes with p53 in cells (57), and it remains to be determined whether they could also be involved in p53 ubiquitination in vivo.
Despite causing accumulation of nuclear p53, suppression of UbcH5B/C did not on its own increase p53 transcriptional activity, and it did not enhance the sensitivity of p53 to activation by doxorubicin or actinomycin D. Suppression of UbcH5B/C resulted in an increase in the level of endogenous Mdm2. Experiments involving the expression of STIP, which disrupts the interaction between Mdm2 and p53, indicated that in cells in which UbcH5B/C are suppressed, the accumulated p53 is repressed at least in part, by Mdm2. This repression p53 Regulation by UbcH5B/C could be mediated directly through the interaction of Mdm2 with the transactivation domain of p53. It could also involve Mdm2-dependent NEDDylation of p53, which we have shown inhibits the transcriptional activity of p53 (19). This provides further evidence that stress-dependent activation of p53 is unlikely to be mediated solely by a general inhibition of the ubiquitin ligase activity of Mdm2, leading to both reduced Mdm2 auto-ubiquitination and p53 ubiquitination. These observations may also have implications for therapeutic approaches based on non-genotoxic inhibitors of the E3 ligase activity of Mdm2. Drugs that block both Mdm2 auto-ubiquitination and Mdm2-mediated ubiquitination of p53 may be relatively ineffective at activating p53. Mdm2 auto-ubiquitination and ubiquitination of p53 can, however, be dissociated. We have previously observed that they can be differentially regulated by p14ARF and MdmX in vivo (15,28). Compounds have been identified that can block Mdm2-mediated ubiquitination of p53 in vitro without effects on Mdm2 auto-ubiquitination (72); however, the effect of these compounds on the p53 pathway in cells has not been reported.
We compared the effects of UbcH5B/C and Mdm2 knockdown. In MCF7 cells the increase in p53 protein levels resulting from suppression of Mdm2 was more transient than that observed following UbcH5B/C suppression. Because Mdm2 is itself a p53 target gene the activation of p53 following Mdm2 knockdown could lead to a partial restoration of Mdm2 mRNA levels. Although we observed a sustained decrease in Mdm2 protein expression, it is possible that a feedback loop is p53 Regulation by UbcH5B/C achieved that limits p53 induction. As well as p53 degradation mediated by viral proteins, there are a number of reports of Mdm2-independent p53 degradation in the normal cellular context (73)(74)(75). Pirh2 and COP1 can stimulate the ubiquitination and degradation of p53 independently of Mdm2 in mammalian cells (76,77). Pirh2 and COP1 are themselves p53 target genes and may be up-regulated in response to Mdm2 knockdown. It has been shown in U2-OS cells that knockdown of COP1 and Mdm2 in combination has a greater effect on p53 stability than knockdown of Mdm2 alone (77). UbcH5 family members support ubiquitination of p53 in vitro mediated by Pirh2 and COP1. If this is the case in vivo, then targeting of UbcH5B/C may have a more robust effect on p53 stability than targeting individual p53 E3 ubiquitin ligases.
p53 protein expression is sensitive to the level of UbcH5B/C indicating that the activity of UbcH5B/C is close to rate-limiting for p53 degradation in the cell lines examined. Changes in the level or activity of UbcH5B/C could thus be involved in regulating the p53 pathway. E2 regulation might be involved in responses to cellular stress. It could participate in a general stress response, which affects multiple UbcH5B/C substrates, however it may selectively affect particular pathways dependent for example, on the affinity of UbcH5B/C for the E3s involved. The p53 pathway appears to be tuned such that it is sensitive to modulation of UbcH5B/C levels. We observed a decrease in UbcH5B/C expression following treatment with the therapeutic drugs doxorubicin and actinomycin D. This preceded induction of p53 by the drugs and its magnitude was sufficient to influence p53 levels. Consistent with an involvement of E2 down-regulation in signaling to p53, accumulation of p53 following exposure to doxorubicin and actinomycin D is associated with attenuation of its ubiquitination and consequent protein stabilization (78). However, our data indicate that the decrease in UbcH5B/C levels could not account for the magnitude of the increase in p53 levels following treatment with these drugs, and in addition we have demonstrated that a reduction in E2 levels would not on its own lead to an increase in p53 transcriptional activity. It is possible that a reduction in UbcH5B/C expression could participate in regulation of p53, but additional signaling events are undoubtedly required to achieve full p53 activation. It has recently been shown that, under conditions where several genotoxic agents increase Mdm2 protein expression due to up-regulation of its mRNA levels, Mdm2 protein stability is actually reduced through an incompletely defined mechanism (68). It has also been reported that genotoxic agents, including doxorubicin, promote proteasomal degradation of MdmX and that this has a requirement for Mdm2 (79). These effects on Mdm2 and MdmX stability serve to regulate the ratio of these repressors to p53, a parameter which our study confirms is a critical determinant of p53 activity. This destabilization could involve stress-dependent modification of Mdm2, which changes the sites or pattern of Mdm2 auto-ubiquitination or which selectively increases the efficiency of auto-ubiquitination through an increase in the rate of transfer of ubiquitin to Mdm2 from the remaining UbcH5B/C. It could result from a stress-dependent decrease in Mdm2 de-ubiquitinating activity or signaling events, which enhance proteasomal recognition of Mdm2. Enhanced degradation of Mdm2/MdmX could also result from the targeting of Mdm2 by another E2/E3. It is conceivable that a reduction in UbcH5B/C levels could participate in a switch from Mdm2/p53 to Mdm2/MdmX degradation. Although UbcH5B/C make a major contribution to the maintenance of low levels of Mdm2 and p53 in unstressed cells, it remains a possibility that another E2 may participate in Mdm2/MdmX degradation following cell stress. One possible model is that a correctly timed decrease in UbcH5B/C expression could be involved in facilitating the interaction of Mdm2 with a distinct stress-activated E2, which favors Mdm2/MdmX degradation over p53 degradation. p53 is activated by diverse stresses and is regulated by multiple signaling events (1,80,81). Further studies will be required to determine the role of E2 regulation and of modulation of the interaction between Mdm2 and E2s in the signaling pathways controlling p53 and to determine the mechanism of the drugdependent changes in expression of UbcH5B/C.