Hypoxia induces Bortezomib resistance in MM cells
Hypoxia has been known as a very important reason of drug resistance in ample solid tumors, therefore we investigated whether hypoxia influences the sensitivities of MM cells to drugs that have been shown effective in the clinic, such as Bortezomib (BTZ), carfizomib (CFZ), and melphalan (Mel). The sensitivity to BTZ of two MM cells, MM.1S and LP-1, were all obviously receded under hypoxia stimulus compare to the normoxia control (Fig 1A, 1B), as also shown by the significantly augmented half-maximal inhibitory concentration (IC50) values (Fig 1C), the distinct abrogated cleavage of PARP as a marker of cell apoptosis (Fig 1D), and the remarkable suppressed apoptotic cell rates (Fig 1E). Intriguingly, hypoxia yielded the similar chemosensitivity alteration when treated another proteasome inhibitor Carfizomib (Fig 1F-J), but failed to induce conspicuous consequences of Melphalan in MM cells (Fig 1K-O). Thus, these results indicate that hypoxia more easily induces chemoresistance to protease inhibitor of MM cells.
Hypoxia upregulates SRC-3 level in MM cells
As our previous study has demonstrated high SRC-3 is associated with poor prognosis in MM, and overexpression of SRC-3 promotes BTZ resistance.19 To dissert the correlation between SRC-3 and hypoxia, we assessed whether hypoxia affects SRC-3 expression in MM cells at mRNA or protein levels. We confirmed that hypoxia induced the augmentation at SRC-3 protein level in a time-dependent manner, neither coordinate with the alteration trend of HIF-1a, nor at mRNA level (Fig 2A, 2B). Immunofluorescence assay also validated that hypoxia promoted the accumulation of SRC-3 in the nucleus (Fig 2A, 2B). Similarly, high SRC-3 level was also found in our previously established bortezomib-resistant (BR) MM.1S and LP-1 cells compared with their parental cells (Fig 2C, 2D),19,22 and administration of a newly developed small molecule targeting SRC-3, the SI-2,23 partially rescued the sensitivity to BTZ of MM cells under hypoxia condition (Fig 2E, 2F). Importantly, combination of SI-2 and BTZ only showed synergistic anti-MM effect under hypoxia condition, but not in the normoxia controls, as evidenced by the remarkably augmented apoptotic cell rate and cleavage of PARP as a marker of cell apoptosis (Fig 2G, 2H). These data suggest that high SRC-3 protein provoked by hypoxia might be a critical regulator for sensitivity to PIs of MM cells.
Degradation of SRC-3 is ubiquitin-proteasome dependent
Since the augmentation of SRC-3 protein under hypoxia stimulus neither coordinate with HIF-1a level, nor occurred at transcriptional level, we speculate that hypoxia may enhance the stability of SRC-3 through another regulator. We assessed three major pathways governing protein degradation in MM cells, including the ubiquitin-proteasome pathway, lysosomal proteolysis and the calpain system. Therefore, we treated MM cells with different doses of MG132, Baf-A1, and MG101 for 24 hours, that are specific inhibitors for the above three pathways, respectively. The results showed that SRC-3 degraded mainly in a proteasome-dependent manner (Fig S 1A-C), which is similar with the alteration of SRC-3 in BTZ-treated MM cells (Fig 3A). We further used co-immunoprecipitation to confirm a physical association of SRC-3 and ubiquitin, and identified that SRC-3 was degraded mainly through lysine 11-(K11), K6-, and K29-linked polyubiquitination (Fig 3C). However, when the K6, K11, and K29 lysine residues of ubiquitin overexpressed in MM cells together with SRC-3, we only confirmed that SRC-3 protein was modified through K11-linked polyubiquitination, since we only elicited a dose-dependent modification of K11-linked polyubiquitination (Fig 3D, 3E) on SRC-3, but failed to observe the K6- and K29-linked polyubiquitination (Fig 3F, 3G). Moreover, when the K11 lysine was mutate to arginine, polyubiquitination of SRC-3 barely detected (Fig 3H). Thus, these results indicate that SRC-3 protein is mainly modified through K11-linked ubiquitination for degradation by the 26S proteasome.
SENP1 stabilizes SRC-3 through deSUMOylation
Next, we sought to deduct the key regulators for SRC-3 protein degradation through dissert the interactome of stably expressing flag-tagged SRC-3 MM.1S cells under hypoxia stimulation for 12 hours using mass spectrometry. Intriguingly, we did not find any E3 ligase in the components of SRC-3 interactome, on the contrary, we discovered a cysteine protease of the sentrin-specific protease (SENP) family, SENP1, which reverses SUMO conjugation on target proteins (Fig 4A). The interaction between SRC-3 and SENP1 was confirmed by co-immunoprecipitation in the HEK293T cells, either using HA-SENP1 as bait protein to pulldown the SRC-3, or using Flag-SRC-3 as bait protein to pulldown SENP1, respectively (Fig 4B). Moreover, the endogenous interaction of SRC-3 and SENP1 was also validated in MM cells using anti-SRC-3 or anti-SENP1 antibodies, respectively (Fig 4C). In addition, protein level of SRC-3 was positively related to the exogenously expressed SENP1 in MM cells (Fig S 2A), and overexpression of SENP1 obviously prolonged the half-life of SRC-3 protein in MM cells (Fig 4D). These data suggested that degradation of SRC-3 might be regulated through coordination of SUMOylation and ubiquitination.
Previous studies have shown that transactivation activity of SRC-3 could be regulated by SUMOylation.20 Actually, convergence of ubiquitination and SUMOylation in modulation of protein functions has emerged as a crucial cellular mechanism in regulating pathogenesis.24 Our result showed that SRC-3 physically interacted with small ubiquitin-related modifier (SUMO)1, SUMO2, and SUMO3, but the interaction with SUMO1 was dominant (Fig S 2B). When SUMO1 was gradually overexpressed in MM cells, SRC-3 degradation was enhanced in a dose-dependent manner (Fig S 2C); on the contrary, SUMO1 depletion using lentivirus carrying shRNA resulted in attenuation of the SRC-3 ubiquitination (Fig S 2D). Thus, we confirmed that SUMOylation was involved in SRC-3 protein degradation. To further investigate the role of SUMOylation in SRC-3 stability in vivo, we generated a SUMO1 knockout mouse, and confirmed that SUMO1 has been completely depleted in B cells (Fig S 2E). In addition, the half-life of SRC-3 protein was markedly increased in the SUMO1-/- B cells (Fig S 2F), and ubiquitination of SRC-3 was barely detected in the SUMO1-/- B cells (Fig 4H). Taken together, these results suggest that SUMOylation was involved in proteasome-dependent degradation of SRC-3.
To further explore the role of SENP1 in SRC-3 modification, we knockdown SENP1 in the HEK293T cells that have stably expressed SUMO1 and flag-tagged SRC-3, and observed that SUMOylation of SRC-3 was markedly elevated (Fig S 3A); on the contrary, overexpression of SENP1 (SENP1-OE) remarkably attenuated the ubiquitination modification of SRC-3 compared with the vector controls (Fig S 3B). Importantly, overexpression of SENP1 dramatically relieved the K11-linked ubiquitination, but not K6- and K29-linked ubiquitination of SRC-3 (Fig 4E). To confirm the phenotype of SENP1 in regulating SRC-3 stability in vivo, we generated a SENP1 conditional knockout mouse model using SENP1-floxp and CD19-Cre crossed mice, and efficiently deleted SENP1 in the B cells (Fig S 3C). As a consequence, protein level of SRC-3 in the SENP1fl/fl; CD19-Cre B cells was obviously downregulated, and administration of MG132 failed to restore the protein level of SRC-3 compared with the B cells from wild type control mice (Fig S 3D). Moreover, the half-life of SRC-3 protein was markedly shortened in the SENP1fl/fl; CD19-Cre B cells compared with that in the wild type B cells (Fig 4F). As expected, endogenous ubiquitination and SUMOylation of SRC-3 was more readily detected in the SENP1fl/fl; CD19-Cre B cells but not in wild type B cells (Fig 4G). Collectively, our results indicate that SENP1 protects SRC-3 against degradation by SUMOylation-associated ubiquitination.
SENP1 is a downstream target of HIF-1a in MM cells
We next assessed how SENP1 was regulated by hypoxia in MM cells. Firstly, we tested expressions of the SENP family genes with HIF-1a overexpression (HIF-1a OE) in MM cells, and identified that hypoxia dominantly induced the expression of SENP1 in protein and mRNA level (Fig 5A, 5B), the results were similar under hypoxia condition in MM cells (Fig 5C, 5D). On the contrary, when HIF-1a was knockdown by lentivirus carrying shRNAs in MM cells (Fig 5E), SENP1 expression was dramatically suppressed both at mRNA and at protein levels, together with the downregulated SRC-3 only at protein level (Fig 5F). Interestingly, we verified the similar expression pattern of SENP1 in the BR-MM cells (Fig 5G, 5H). To determine whether HIF-1a is a transcriptional factor for SENP1 promoter, we constructed a 2Kb SENP1 promoter, which contains two cis-acting elements of HIF-1a at -102~-111bp and -1684~-1691bp (Fig 5I), and ectopically expression of HIF-1a resulted in over 10 folds activation of the SENP1-luciferase reporter (Fig 5J). Importantly, hypoxia triggered significant enrichment of HIF-1a on SENP1 promoter in MM cells when detected by chromatin immunoprecipitation assay (Fig 5K). These results uncover that SENP1 is a direct target of HIF-1a under hypoxia condition.
SENP1 plays critical role in regulating chemosensitivity to PIs in MM cells
To further validate whether SENP1 is an important regulator of chemosensitivity in MM cells, we ectopically overexpressed (OE) or depleted the expression (KD) of SENP1 in MM.1S and LP-1 cells using lentivirus carrying expressing or shRNAs vectors, respectively (Fig S 4A, 4B). As expected, knockdown of SENP1 expression increased sensitivity to BTZ treatment in a dose-dependent manner, with a significantly reduced IC50 in both MM.1S and LP-1 cells (Fig 6A, 6B). Meanwhile, increased cell apoptosis due to BTZ treatment were observed in the SENP1-KD MM cells compared with the pared controls, as evidenced by augmented cleavage of PARP protein (Fig 6C), and apoptotic cell rates (Fig 6D). In contrast, overexpression of SENP1 in MM cells led to indolent response to BTZ treatment and inhibited cell apoptosis (Fig 6E-H). Importantly, suppression of SENP1 in the SRC-3 stably expressing MM cells using a specific inhibitor, Momordin Ιc (Mc), remarkably sensitized the anti-MM efficacy of BTZ, but the synergetic effect of Mc failed to elicited in the SRC-3-KD MM cells (Fig 6I, S 4C). In addition, rescue of SRC-3 expression in the SENP1-KD MM cells largely reinstated the resistance to BTZ treatment compared with its paired vector control (Fig 6J, S 4D). Clinically, we observed that protein levels of HIF-1a, SENP1 and SRC-3 were mutually correlated, and the expressions were all significantly augmented in the refractory/relapsed MM patients (Fig 6K, S 4E-H), and combination of SENP1 inhibitor Mc with BTZ dramatically augmented cell apoptosis in primary CD138+ plasma cells that have been resistant to BTZ-based regiments (Fig 6L). In conclusion, these data suggest that SENP1 plays a critical role in regulating chemosensitivity to PIs in MM cells.
Targeting SENP1 suppresses hypoxia-induced SRC-3 and drug resistance in vivo
To assess the effect of targeting SENP1 in overcoming drug resistance in vivo, we established BTZ-resistant MM cell derived xenograft models, RRMM patient CD138+ plasma cell derived intra-bone growth MM model, as well as the Vk*Myc transgenic and transplant mouse models of MM. The SENP1 inhibitor Mc alone administration had no obvious inhibitory effects on tumor growth, nor did BTZ alone administration, however, the combination of BTZ and Mc considerably suppressed the tumor growth of MM (Fig 7A), prolonged the survival rate of mice (Fig 7B), and importantly, obviously more cell apoptosis (Fig 7C) and downregulated expression of SENP1 and SRC-3 were observed in the combined treatment groups (Fig 7D). Moreover, in three relapsed MM patients’ bone marrow unsorted bone marrow mononuclear cells derived xenograft model, tumor burdens were also noticeably extenuated in the combination treatment groups, as evidenced by significantly suppressed M-protein levels (Fig 7E), and CD138+ cells left in the bone marrow microenvironment (Fig 7F). Furthermore, in the successfully constructed Vk*Myc transplant mouse models of MM (Fig S 5A-D), we further validated the synergistic anti-MM effect of combination of SENP1 inhibitor and BTZ, as shown by the conspicuously reduced M-protein level (Fig 7G), increased survival rate of mice (Fig 7H), and nearly eradicated proportion of CD138+ plasma cells in the bone marrow microenvironment (Fig 7I). Taken together, these in vivo data strongly suggest that pharmacologically targeting SENP1 abrogates chemoresistance to PIs in MM cells.