The ubiquitin-conjugating enzyme UBE2O modulates c-Maf stability and induces myeloma cell apoptosis

Primary bone marrow species were collected from the Department of Hematology, the First Affiliated Hospital of Soochow University. The use of primary bone marrow cells was approved by the Review Board and Ethical Committee of Soochow University, and each patient provided written informed consent to donate 2–5 ml of bone marrow for this study after diagnostic and clinical procedures in accordance with the Declaration of Helsinki. Mono-nuclear cells were isolated by Lympholyte® Cell Separation (Cedarlane, Canada) [13].

c-Maf and MafA were cloned as described previously [14]. UBE2O cDNA (Genebank accession No. BC051868.2) was obtained from the SPARC BioCentre, The Hospital for Sick Children, Toronto, Canada.

To generate UBE2O-expressing lentivirus, the cDNA fragment of UBE2O was generated by using the following primers: 5′-TCGAGCTCAAGCTTATGACCTCAGCCGACGTGATG-3′ (forward) and 5′-CTCACCATGACCCATGACCGGTGGATCCTCCTTGTCCTCTGTGCACTCCG-3′ (reverse) and inserted into pLVX-AcGFP lentiviral vector (Clontech) within the EcoRI and BamHI sites. The specific protocol for packaging of the virus and preparation of lentiviral particles were described previously [12].

The AP/MS process and data analysis were performed as described previously [12].

HEK293T cells were transfected with c-Maf and UBE2O plasmids for 24–48 h followed by immunoprecipitation as described as previously [15] using specific primary antibodies as needed followed by incubation with 40 μl of a 50% slurry of protein A+G agarose beads with gentle rotation at 4 °C for 2 h. Agarose beads were collected and washed five times with the lysis buffer [12], followed by re-suspension in 20 μl of 2× SDS loading buffer. Samples were then boiled before being subjected to fractionation on sodium dodecyl sulfate poly-acrylamide gel electrophoresis (SDS-PAGE) and immunoblotting (IB) analysis.

This assay was adapted from a previous report [16]. Briefly, HA-c-Maf and Flag-UBE2O plasmids were transfected into HEK293T cells, respectively. Forty-eight hours later, cells were treated with MG132 for 2 h, followed by cell lysate preparation. To enrich and purify c-Maf and UBE2O proteins, individual cell lysates were subjected to immunoprecipitation with HA- (for c-Maf) or Flag- (for UBE2O) antibody-conjugating agarose beads, respectively, at 4 °C for overnight. After that, the beads were washed four times with an immunoprecipitation lysis buffer, twice with 1× ubiquitin reaction buffer (Boston Biochem, Boston, MA) and then re-suspended in 20 μl of 1× ubiquitin reaction buffer containing 200 ng of recombinant E1, 250 ng of recombinant UbcH5c, 10 μg of ubiquitin, 0.5 mM ATP, and 1× Energy Restoration System (Boston Biochem, Boston, MA). The reaction was carried out at 30 °C for 2 h and then terminated by boiling in the 2× SDS loading buffer. Ubiquitinated products were resolved by SDS-PAGE and detected by immunoblotting analysis.

All the IB and CHX chase assays were performed as described previously [15].

MM cells were infected with lentiviral UBE2O for 1 - 7 d before being subjected to MTT (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide) assay as described previously [17, 18].

MM cells were infected with lentiviral UBE2O for 48 h; cells were collected and treated with cold 70% ethanol and stained with propidium iodide (PI). Cell cycle was analyzed on a cytometer (FACSCalibur; BD Biosciences, San Jose, CA) as described previously [19].

MM cells were infected with lentiviral UBE2O at various periods; cells were then stained with 7-ADD and Annexin V-PE (MultiSciences BiotechCo., Ltd., Hangzhou, China) followed by analysis on a cytometer as described previously [20].

The luciferase reporter plasmid pGL4-CCND2-Luc was constructed as described previously [15]. The DNA sequence of CCND2 promoter contained a c-Maf recognition element (MARE) which could be recognized by c-Maf protein [10, 21]. To examine the effect of UBE2O on c-Maf biological function, HEK293T cells were co-transfected with pGL4-CCND2, c-Maf, and UBE2O, and a β-gal expression vector (100 ng). Luciferase and β-gal expression were measured 36 h after transfection according to the manufacturer’s protocols (Promega, Madison, WI, USA). Firefly luciferase activity was normalized to β-gal expression for each sample [15]. All transfection experiments were performed in duplicates.

The DNA microarray dataset from primary MM patients and healthy donors was retrieved from Gene Expression Omnibus (GEO) databases (http://​www.​ncbi.​nlm.​nih.​gov/​gds) [22]. Log₂(UBE2O mRNA level) was reported.

Total RNA was extracted using Trizol® (Transgene, Beijing, China). RNA (2.5 μg) was reversely transcribed using a Superscript^TM-III kit (Invitrogen) according to the manufacturer’s instruction. PCR amplification was carried out in 25 μL of PCR reaction mixture containing 10 mM Tris-HCl (pH 8.3), 50 mM KCl, 2 mM MgCl₂, 20 pmol of each primer set, two units of Taq DNA polymerase (Transgene, Beijing, China), 0.2 mM dNTPs, and 2 μL cDNA. The primers for UBE2O were 5′-ACATCCGCTCCAACGAC-3′ and 5′-GCTGGTGCTGCCTTCTAC-3′, and the primers for GAPDH were 5′-CCAGCCGAGCCACATCGCTC-3′ and 5′-ATGAGCCCCAGCCTTCTCCAT-3′. PCR products were visualized by ethidium bromide staining after electrophoresis on 1.5% agarose gels. The optical densities of the genes were quantified using the ImageJ software (National Institutes of Health, Bethesda, MD) [19].

LP1 cells (2 × 10⁷) were s.c. inoculated into the right flanks of nude mice (The SLAC Experimental Animal Co., Shanghai, China). When the tumors were palpable, 10 μg of UBE2O plasmids or empty vectors in 100 μl of In Vivo-jetPEI® Delivery Reagent (N/P = 6) (Polyplus-transfection Inc., New York, USA) [23, 24] were injected into tumors twice a week for continued 3 weeks [17]. Tumor sizes and mice body weights were monitored every other or 3 days. This xenograft study was approved by the Review Board of Animal Ethics of Soochow University. At the end of the experiment, all tumor species were excised for immunoblotting analysis against specific antibodies as indicated.

Student’s t test was used to calculate P values for differences. Differences were considered significant at P < 0.05.

To find out specific enzymes responsible for c-Maf ubiquitination, a LC/MS/MS assay was performed after affinity purification with a c-Maf specific antibody [12]. The LC/MS/MS identified 104 proteins specifically associated with c-Maf that were enriched into two “pathways” by KEGG: the “proteasome pathway” containing 24 proteins and the “ubiquitin-mediated proteolysis” containing 6 proteins [12] from which UBE2O was found to be the only ubiquitin-conjugating enzyme. As shown in Fig. 1a, unique UBE2O peptides were identified from the co-immunoprecipitates of c-Maf protein. A typical peptide (SGYPDIGFPLFPLSK) identified by MS was shown in Fig. 1b, and specific MS data on UBE2O was presented in Table 1.

To verify the interaction between UBE2O and c-Maf, a UBE2O and a c-Maf plasmid were co-transfected into HEK293T cells, from which the lysates were co-immunoprecipitated with a specific anti-HA antibody followed by immunoblotting. As shown in Fig. 1c, UBE2O was found in the c-Maf precipitates but not in controls. To confirm this finding, a reciprocal Co-IP was performed with a specific antibody against UBE2O, followed by immunoblotting against c-Maf and the result showed that c-Maf was present in the UBE2O precipitates (Fig. 1d). To further check this finding in MM cells, lentiviral UBE2O was infected in a MM cell line LP1. In this experiment, c-Maf was used as a baitor. As shown in Fig. 1e, UBE2O was identified in the c-Maf precipitates. Therefore, these immunoprecipitation/immunoblotting assays suggested that c-Maf interacted with UBE2O.

Because UBE2O is a ubiquitin-conjugating enzyme and it has been reported to mediate ubiquitination of several proteins, we thus wondered whether UBE2O could mediate c-Maf ubiquitination given its interaction with c-Maf. To this end, Myc-UBE2O, Flag-Ub, and HA-c-Maf plasmids were co-transfected into HEK293T cells. After immunoprecipitation with an anti-Flag antibody (for Ub) followed by immunoblotting with an anti-HA antibody (for c-Maf), polyubiquitination level of c-Maf was found to be markedly increased by UBE2O (Fig. 2a). Next, we wondered c-Maf ubiquitination in MM cells in the presence of UBE2O. To this end, myeloma cell lines LP1 and KMS11 were infected with lentiviral UBE2O. Seventy-two hours later, cells were subjected to whole cell lysate preparation, followed by immunoprecipitation with a c-Maf specific antibody and immunoblotting with an anti-ubiquitin antibody. As shown in Fig. 2b, c, UBE2O markedly increased the polyubiquitination level of c-Maf in both cell lines. Because UBE2O is the only E2 that was reported to display ubiquitin ligase activity [1], we wondered whether UBE2O could mediate c-Maf polyubiquitination independent of an E3 ligase. To this end, an in-tube ubiquitination assay was performed. As shown in Fig. 2d, UBE2O markedly added ubiquitin molecules to c-Maf protein in the absence of an extra E3 ligase. Because E3 ligase is essential for the ubiquitination of a specific protein, this finding suggested that UBE2O possesses an E3 liage activity in c-Maf ubiquitination. There are various ubiquitination manners of which the most common ones are the K48-linked and the K63-linked ubiquitination, and each type of ubiquitination leads to unique functions and outcomes, and previous studies showed that UBE2O could lead to monoubiquitination [4] and K63-linked polyubiquitination [5]; we wondered which type of ubiquitination was mediated by UBE2O on c-Maf. To this end, the ubiquitin plasmids with K48R or K63R mutation were co-transfected into HEK293T cells along with c-Maf and UBE2O. The subsequent immunoprecipitation/immunoblotting assays showed that the K63R-Ub mediated c-Maf ubiquitination in the presence of UBE2O, but when K48 was mutated (K48R), c-Maf failed to be ubiquitinated by UBE2O (Fig. 2e). Consistent with this finding, K48- but not K63-Ub mediated ubiquitin conjugation to c-Maf (Fig. 2f). Therefore, UBE2O mediates the K48-linked polyubiquitination on c-Maf.

The K48-linked polyubiquitinaiton usually results in protein degradation. Because the above studies revealed that UBE2O mediated K48-linked polyubiquitination of c-Maf, we next investigated whether UBE2O modulated c-Maf degradation. To this end, a c-Maf plasmid was transfected into HEK293T cells along with increased Myc-UBE2O. Immunoblotting revealed c-Maf protein was decreased by UBE2O in a concentration- and time-dependent manner (Fig. 3a). However, UBE2O failed to induce degradation of MafA, another key member in the large Maf subfamily (Fig. 3b). Degradation of c-Maf by UBE2O was further demonstrated by the CHX chase assay from which UBE2O markedly shortened the time for c-Maf degradation (Fig. 3c). Moreover, UBE2O-mediated c-Maf protein degradation was abolished by MG132, a proven proteasomal inhibitor (Fig. 3d). All the above results thus led to a conclusion that UBE2O mediated c-Maf polyubiquitination and degradation in the proteasomes.

The above results clearly demonstrated that UBE2O as a ubiquitin-conjugating enzyme mediates c-Maf ubiquitination. It reported that there are 14 lysine residues in c-Maf protein, and our previous studies showed that each lysine residue was important for c-Maf ubiquitination [12, 15]. To find out which lysine residue was important in UBE2O-mediated c-Maf ubiquitination, we generated a series of lysine (K) to arginine (R) mutants of c-Maf, followed by co-transfection with UBE2O. The measurement of c-Maf protein levels by immunoblotting assay indicated that UBE2O led to degradation of the wild-type c-Maf and most K to R mutants including K85R, K283R, K290R, K297R, K320R, K347R, and K350R (Fig. 4a). However, it failed to decrease c-Maf proteins with several mutants, including K29R, K33/34R, K308R, K331R, and K345R (Fig. 4a), which was similar to K0, the all K to R mutant (Fig. 4a) and previous studies [15]. Among these mutants, K331R and K345R were further confirmed not to mediate c-Maf degradation. As shown in Fig. 4b, UBE2O decreased the wild-type c-Maf in a concentration-dependent manner, but had no effects on the stability of the K331R, K345R, and K0 mutants. Notably, UBE2O increased the protein levels of the K33/34R c-Maf mutant (Fig. 4b) which was consistent with the results observed in a single dose of UBE2O transfection (Fig. 4a). To further confirm this finding, K331R, K345R, and K33/34 mutant plasmids were individually transfected into HEK293T cells along with UBE2O, followed by immunoprecipitation with an HA antibody (for c-Maf) and immunoblotting using a Flag antibody (for Ub). As shown in Fig. 4c, UBE2O failed to induce polyubiquitination on c-Maf with K331R or K345R mutation; in contrast, the wild-type and K33/34R c-Maf proteins were mediated polyubiquitination by UBE2O (Fig. 4c). Therefore, the K331 and K345 residues were critical for UBE2O-mediated c-Maf degradation.

c-Maf is an oncogenic transcription factor that modulates the expression of several key genes including cyclin D2 involved in MM pathophysiology [10]. To explore the expression profile of UBE2O in MM cells, mRNA analysis was performed on a DNA microarray dataset from a panel of primary bone marrow samples of healthy donors, MGUS (monoclonal gammopathy of undetermined significance), SMM (smoldering MM), and MM patients [12, 22]. It showed that UBE2O was expressed in the healthy donors, but was significantly decreased in MGUS, SMM, and MM patients (Fig. 5a). To confirm this, UBE2O was measured in primary bone marrow cells from MM patients and healthy donors using RT-PCR. As shown in Fig. 5b, c, UBE2O in the bone marrow cells from MM patients were significantly lower than that from healthy donors, which was consistent with the DNA microarray assay (Fig. 5a). Therefore, UBE2O was downregulated in the process of myelomagenesis.

To further evaluate the effects of UBE2O on c-Maf activity, a luciferase reporter assay was performed after UBE2O was transfected into HEK293T cells along with pCCND2.Luci with or without c-Maf plasmids. The results showed that c-Maf markedly increased the luciferase activity under the control of CCND2 promoter containing a c-Maf recognition element (Fig. 5d) [15]. To confirm this conclusion, we next measured c-Maf and cyclin D2 levels in MM cells after MM cell lines LP1 and RPMI-8226 were infected with lentiviral UBE2O. As shown in Fig. 5e, both c-Maf and CCND2 were downregulated when UBE2O was re-expressed (Fig. 5e). However, when these MM cell lines infected with lentiviral UBE2O were subjected to cell cycle analysis, the results showed that UBE2O arrested both RPMI-8226 and LP1 cells at the S or S/M phases (Fig. 5f), which was out of expectation because cyclin D2 is a key player in control the progress of cell cycle from G1 to S phase.

The above studies demonstrated that UBE2O mediated c-Maf degradation and cell cycle arrest but it was downregulated in MM cells; we next evaluated the effects of UBE2O on proliferation and apoptosis in various MM cells with different levels of c-Maf. As shown in previous reports [21] and the current study, RPMI-8226 and LP1 cells expressed a high level of endogenous c-Maf (Fig. 5e), while NCI-H929 and KMS12 cells expressed a very low level of endogenous c-Maf (Fig. 6a); these cell lines were applied for this study. All cells were infected with lentiviral UBE2O followed by evaluation of cell viability and apoptosis. We first measured the cleavage level of PARP, a hallmark of apoptosis. As measurement by immunoblotting, UBE2O had no effect on PARP cleavage in NCI-H929 or KMS12 that lack c-Maf (Fig. 6a) but induced PARP cleavage in RPMI-8226 and LP1 that express c-Maf (Fig. 7a), indicating UBE2O probably induced MM cell apoptosis in association with c-Maf expression. To confirm this finding, MM cells after lentiviral UBE2O infections were subjected to apoptotic analysis by a flow cytometer after staining with Annexin V, a direct hallmark of cell apoptosis. As shown in Fig. 6b, UBE2O failed to increase the Annexin V+ fraction in NCI-H929 or KMS12 cells, but markedly raised this fraction in RPMI-8226 and LP1 cells (Fig. 7b). Because UBE2O arrested cell cycle of MM cells expressing c-Maf, we next assayed cell proliferation status after infection with lentiviral UBE2O by the MTT method. The results showed that UBE2O failed to inhibit the proliferation of NCI-H929 or KMS12 cells (Fig. 7c), but it inhibited the proliferation of RPMI-8226 and LP1 cells in a time-dependent manner (Fig. 7c). These findings were consistent with c-Maf expression levels and c-Maf degradation by UBE2O. Therefore, these results demonstrated that UBE2O induced MM cell apoptosis and suppressed MM cell proliferation in association with c-Maf expression.

Previous studies have demonstrated that degradation of c-Maf or interference of c-Maf leads to delayed MM tumor growth in nude mice [10, 12]; we wondered whether UBE2O generated similar effects on MM tumor growth in vivo. Myeloma xenografts with human MM cell line LP1 were established in immunodeficient nude mice, followed by intratumoral transfection of UBE2O plasmids delivered by in vivo jetPEI as demonstrated previously [17, 23, 24]. The results showed that re-expression of UBE2O decreased myeloma tumor growth (Fig. 8a). UBE2O also significantly extended survival time of myeloma-bearing mice. As shown in Fig. 8b, all mice carrying myeloma tumors and being transfected with control plasmids died within 32 days; in contrast, all mice survived when receiving UBE2O in an intratumoral manner. Notably, endogenous c-Maf protein in MM tumors was also decreased in the mice receiving UBE2O (Fig. 8c). To find out whether UBE2O induced apoptosis in MM tumor tissues, all tumor samples were subjected to immunoblotting assay against PARP and caspase-3. As shown in Fig. 8c, UBE2O induced markedly cleavage and activation of PARP and caspase-3, a signal of apoptosis. Therefore, UBE2O downregulated c-Maf and delayed MM tumor growth.