Suppression of multiple myeloma by mitochondrial targeting

Treatment of multiple myeloma (MM) aims at inducing cell apoptosis by surpassing the limited capacity of MM cells to cope with oxidative stress. MM cell survival may further be suppressed by limiting cellular cholesterol. Long-chain fatty acid analogs of the MEDICA series promote mitochondrial stress and inhibit cholesterol biosynthesis, thus prompting us to verify their efficacy and mode-of-action in suppressing MM cell survival, in comparison to bortezomib. MEDICA analog is shown here to effectively suppress survival of MM cells, and to inhibit growth of MM xenograft. Suppression of MM cell survival by MEDICA is accompanied by inhibition of the STAT3, MAPK and the mTORC1 transduction pathways due to mitochondrial oxidative stress. MEDICA-induced oxidative stress is abrogated by added exogenous cholesterol. Suppression of MM cell survival by bortezomib is similarly driven by bortezomib-induced oxidative stress, being abrogated by added cholesterol. In line with that, the time-to-best-response of MM patients to bortezomib-based treatment protocols is shown to be positively correlated with their plasma cholesterol level. MEDICA profile may indicate novel therapeutic potential in the management of MM.

Immunohistochemistry. Formaldehyde-fixed paraffin-embedded sections were stained with antibodies and dyes as indicated and quantified by computerized microscopy (Ariol).

RT-PCR.
Total RNA of MM cells was prepared using the TRI reagent (Sigma Aldrich). First strand cDNA used as template was synthesized by reverse transcription using oligo(dT) or random hexamers mix as primer and the Reverse-iTMAX First Strand Kit (ABgene). Transcripts were quantified by real-time PCR (Rotor Gene RG-3000A) using SYBER green MasterMix (Absolute Syber Green ROX Mix, ABgene). Primer sequences were as follows: human Cyclin D1: forward (5-GTG CTG CGA AGT GGA AAC C-3) and reverse ( ROS production. ROS production was determined by 2,7 dichlorofluoresceine diacetate (DCFDA) (5 μM) added to respective cell cultures for the last 15 min of incubation. Cells were washed once with PBS and analyzed by FACScan. Glutathione (GSH)/Glutathione disulfide (GSSG) ratio was determined by the GSH/GSSG-Glo assay kit (Promega).   (Fig. 3A,B). Inhibition of STAT3(Tyr705) phosphorylation was already evident within the first 3 h of treatment, and was accompanied by decrease in STAT3 expression, due to the positive feedback of nuclear P-STAT3 in activating its own transcription 21 . Inhibition of STAT3(Tyr705) phosphorylation was accompanied by decrease in gp130 and IL-6Rβ cellular content (Fig. 3B), indicating that suppression of the IL-6/STAT3 transduction by MEDICA may partly be accounted for by abrogating the IL-6 receptor ligation. Of note, MEDICA inhibited the proliferation of U266 cells in which STAT3 was knocked out (not shown), implying additional MM oncogenic drivers beyond STAT3 that are suppressed by MEDICA. In line with that, MEDICA treatment resulted in suppressing Erk(Thr204) phosphorylation in U266 and RPMI8226 cells (Fig. 3C). In addition, MEDICA treatment resulted in suppressing mTORC1 activity as verified by its phospho-S6K1(Thr389) and phospho-S6(Ser240/244) downstream substrates (Fig. 3D).
Mitochondrial ROS production by MEDICA and its abrogation by mitochondrial cholesterol. MEDICA has recently been reported to inhibit mitochondrial complex I 39 , resulting in mitochondrial oxidative stress. MEDICA-induced mitochondrial superoxide and reactive oxygen species (ROS) production was studied in U266 MM cells by increase in cellular Mitosox and DCFDA fluorescence, with concomitant decrease in GSH/GSSG ratio (Fig. 4A). Increase in MEDICA-induced ROS was accompanied by mitochondrial www.nature.com/scientificreports/ depolarization as verified by decrease in TMRM (Fig. 4A). Mitochondrial superoxide production was similar to that induced by rotenone or antimycin. Mitochondrial depolarization by MEDICA was abrogated by the synthetic SOD/catalase EUK207 (Fig. 4A), resulting in abrogating MEDICA effects in activating PARP and in suppressing phospho-Rb and mTORC1 (Fig. 4B). Surprisingly, MEDICA effects were partially abrogated by increasing mitochondrial cholesterol content by added exogenous cholesterol. Thus, adding cholesterol to U266 MM cells for 24 h resulted in 2.5-fold increase in mitochondrial cholesterol content (3.0 and 7.0 µg cholesterol/ mg mitochondrial protein in the absence and presence of added 25 µg cholesterol/ml, respectively). Added cholesterol resulted in partially abrogating MEDICA-induced mitochondrial superoxide production (Fig. 4C), growth inhibition (growth, CD1, phospho-Rb) (Fig. 4D), IL-6/STAT3 inhibition (phospho-STAT3, gp130, IL-6R) (Fig. 4E), and apoptosis (PARP, mcl-1) (Fig. 4F). Of note, abrogation of MEDICA effects by cholesterol was specific. Thus, PPARα activation by MEDICA 22 was maintained in the presence of added cholesterol (not shown). Also, MEDICA activity in suppressing MM cell survival was not accounted for by lipid raft disruption as verified by plasma membrane caveolin-1 and GM1 content (not shown).

Suppression of MM cell survival by bortezomib is mediated by mitochondrial ROS production and is partially abrogated by increase in mitochondrial cholesterol. Combination of MEDICA
with bortezomib had no additive effect in suppressing MM cell proliferation (Suppl Fig. 1D), implying possible shared target(s). Indeed, similarly to MEDICA, and in line with previous reports 5, 6 , treatment of U266 MM cells with bortezomib resulted in mitochondrial superoxide production, being abrogated by added Tiron (ROS scavenger) or cholesterol (Fig. 5A). Added Tiron or cholesterol further resulted in abrogating growth inhibition (Fig. 5B), CD1 and IL-6/STAT3 inhibition (phospho-STAT3, gp130, IL-6R) (  www.nature.com/scientificreports/ to best-response of 66.5 days (range 25-185). The time to hematologic response to bortezomib was found to be significantly correlated to the respective plasma cholesterol and LDL-C levels. Thus, the median time to best-response was shorter for patients having lower plasma cholesterol (Fig. 6A) or LDL-C cholesterol levels (Fig. 6B). In line with that, plasma cholesterol and LDL-C levels were lower in patients achieving best-response to bortezomib in less than 66 days as compared with late responders (Fig. 6C,D). These results were maintained upon setting a different median cutoff time of 100 days (Suppl Fig. 3A,B). In contrast to time to best-response, the overall response or the primary resistance to bortezomib therapy were not correlated with patient cholesterol levels (not shown). Also, in contrast to LDL-C, HDL-C levels were found to have no correlation with the time to best-response to bortezomib (not shown). Of note, 29 patients of the bortezomib-responsive cohort received statin therapy, but the correlation between the time to best-response to bortezomib and LDL-C levels was maintained upon excluding statin-treated patients (not shown).  5,6 , suppression of MM cell survival by bortezomib is reported here to be accounted for by bortezomib-induced mitochondrial ROS, being abrogated by added anti-oxidants, or surprisingly, by increase in mitochondrial cholesterol. Indeed, suppression of MM cell survival by MEDICA was not additive with that of bortezomib, indicating that the two drugs may converge onto shared target(s) (Fig. 6E). Of note, bortezomib activity in inhibiting MM cell survival is primarily ascribed to proteasome inhibition, resulting in unfolded protein response, followed by endoplasmic 5 and/or mitochondrial oxidative stress 6,[24][25][26][27][28] . In contrast to bortezomib, MEDICA efficacy is primarily due to its mitochondrial activity, while avoiding endoplasmic stress in MM cells (not shown). Hence, MEDICA may avoid side effects and resistance to proteasome inhibition. www.nature.com/scientificreports/ In pursuing treatment modes for MM, high dose statins or bisphosphonates were previously reported to induce growth arrest and apoptosis of MM cells, being rescued by added mevalonate, geranyl-PP, farnesyl-PP or cholesterol [29][30][31][32] . Growth suppression by statins or bisphosphonates was ascribed to suppressing the farnesylation/ geranylation/prenylation of small GTPases (e.g., Ras, RohA), or the disruption of lipid rafts due to cholesterol limitation 33 . Abrogation of MEDICA activity by added cholesterol as reported here is in line with the above previous reports. However, our findings further indicate that cholesterol may interfere with mitochondrial ROS production, and that cholesterol limitation may sensitize MM cells to selective cell apoptosis due to mitochondrial oxidative stress. This observation conforms to previous findings whereby mitochondrial cholesterol is reported to contribute to chemotherapy resistance in hepatocellular carcinoma 34 , while disruption of cholesterol uptake synergizes with chemotherapy in pancreatic adenocarcinoma 35 . The mode of-action of cholesterol in abrogating mitochondrial oxidative stress remains to be investigated. www.nature.com/scientificreports/ The cholesterol/oxidative stress interplay appears to be of clinical relevance, as verified here by the time to best-response to bortezomib-based treatment protocols of MM patients. The time to best-response to bortezomib, but not the overall response rate, was found to be positively correlated with respective plasma LDL-C/cholesterol levels of MM patients. In addition, as most patients reported here were not treated with statins, the time to best-response to bortezomib in statin-treated patients may be ascribed to LDL-C lowering, rather than putative off-target activities of statins. Indeed, survival of myeloma cells is reported to be increased in the presence of LDL-C in the culture medium 36 . Of note, the clinical correlation reported here is based on a retrospective cohort and a univariate analysis of the data, thus requiring further prospective validation.

Conclusions
Due to current standard-of-care, MM patients live for years with the disease under control. However, ultimately all patients do relapse at some point with 5-year survival rate of ~ 50%. Hence, MM is still an unmet need. Mitochondrial targeting by MEDICA is shown here to induce mitochondrial oxidative stress 40 and to suppress MM oncogenic drivers, resulting in anti-tumor effect. MEDICA and bortezomib appear to share a common mode-ofaction, and the efficacy of both may be enhanced by lowering plasma cholesterol. Hence, the combined activity of MEDICA, in inducing mitochondrial oxidative stress while lowering blood cholesterol, may translate into a novel effective treatment for MM patients. www.nature.com/scientificreports/