Ponatinib sensitizes myeloma cells to MEK inhibition in the high-risk VQ model

Multiple myeloma (MM) is a malignant plasma cell cancer. Mutations in RAS pathway genes are prevalent in advanced and proteasome inhibitor (PI) refractory MM. As such, we recently developed a VQ MM mouse model recapitulating human advanced/high-risk MM. Using VQ MM cell lines we conducted a repurposing screen of 147 FDA-approved anti-cancer drugs with or without trametinib (Tra), a MEK inhibitor. Consistent with its high-risk molecular feature, VQ MM displayed reduced responses to PIs and de novo resistance to the BCL2 inhibitor, venetoclax. Ponatinib (Pon) is the only tyrosine kinase inhibitor that showed moderate MM killing activity as a single agent and strong synergism with Tra in vitro. Combined Tra and Pon treatment significantly prolonged the survival of VQ MM mice regardless of treatment schemes. However, this survival benefit was moderate compared to that of Tra alone. Further testing of Tra and Pon on cytotoxic CD8+ T cells showed that Pon, but not Tra, blocked T cell function in vitro, suggesting that the negative impact of Pon on T cells may partially counteract its MM-killing synergism with Tra in vivo. Our study provides strong rational to comprehensively evaluate agents on both MM cells and anti-MM immune cells during therapy development.


Re-purposing screen identifies de novo resistance of VQ MM cells to the BCL-2 inhibitor venetoclax.
We previously showed that an FDA-approved MEK inhibitor, trametinib (Tra), killed MM cells in a dose-dependent manner and downregulated surface PD-L1 expression in vitro 3 . In VQ-D1 MM recipient mice, Tra reversed exhausted cytotoxic CD8 + T cell phenotypes ( Figure S1) and prolonged their survival 3 . Therefore, we sought to identify new MEK inhibition-based combination therapies utilizing two VQ myeloma cell lines. A high-throughput screening assay was developed in which VQ 4938 cells were cultured in 384-well plates for 48 h and cell viability was measured using the CellTiter-Glo luminescence assay. Cells treated with Tra served as a positive control for the assay, with DMSO treated cells as the negative control. Z' factor for the assay was consistently greater than 0.50, indicating assay reproducibility and consistency 4 .
To expediate clinical testing, we initially focused on combining Tra with a library of 147 FDA-approved anti-cancer drugs provided by the National Cancer Institute (AOD IX panel). VQ 4938 cells were treated with the AOD IX panel drugs at concentrations of 100 nM and 1000 nM in the presence or absence of 10 nM Tra (Fig. 1A). Viability was measured as the relative change in luminescence compared to DMSO treated wells. Of note, 10 nM Tra alone led to ~ 50% viability relative to DMSO control. Viability fold change of anti-cancer drug alone (X axis) versus viability fold change when the drug was combined with 10 nM Tra (Y axis) was then plotted and analyzed via linear regression (Fig. 1B). Area under the curve represents increased efficacy of compounds when combined with Tra.
The AOD IX panel includes many drugs approved for MM treatment. Based on their initial screening results as single agents (Table S1) and the knowledge of drug actions, these compounds were classified as positive, false negative, or negative (Fig. 1C). False negative group included cyclophosphamide, a pro-drug that needs to be metabolized by the liver to be active in vivo 5 , and IMiDs (e.g. lenalidomide), which are known to be ineffective against murine cells due to the species difference at the cereblon (Crbn) codon 391 6 . Our in vitro validation of VQ response to lenalidomide ( Figure S2) is consistent with its in vivo testing in Vk*MYC mice 7 .
Interestingly, VQ 4938 cells showed de novo resistance to venetoclax, with an IC 50 > 1000 nM in the primary screen (Table S1). We subsequently validated this result using two VQ cell lines and a broad range of drug concentrations (Fig. 1D). Consistent with our primary screen result, the IC 50 was not reached at 16 µM in both VQ cell lines and estimated to be ~ 20 µM. Venetoclax is considered one of the few targeted therapies for MM patients with t(11;14) translocations and/or high BCL2:BCL2L1 and BCL2:MCL1 gene expression ratios 8 . 1C and Table S1). Again, these results were validated using both VQ 4935 and 4938 cell lines in dose response tests ( Figures S3, S4A, and S4D). In comparison to human myeloma cell lines 9 , both VQ MM cell lines displayed increased resistance to Btz and Cfz based on their IC 50 values (~ 9 nM and ~ 60-70 nM respectively, Figures S4A and S4D). This is in line with clinical data showing patients with NRAS mutations have reduced Btz sensitivity 10 . Because PIs are used in all lines of MM treatment, we further explored them in vivo. In our previous study 3 , we used Btz in the VQ model as a single agent following a treatment scheme established with the Vk*Myc model 7 . However, a significant proportion of treated mice died soon after the treatment, suggesting that VQ MM mice may not tolerate this treatment scheme very well. Therefore, we adjusted it based on the current clinical practice in human patients and found that this revised scheme showed transient effectiveness in controlling VQ growth in vivo ( Figure S4B) and provided a moderate but significant increase in survival ( Figure S4C).
To further boost the survival benefit, we used Cfz as part of a combination therapy regimen with dexamethasone (Dex), Tra, and GSK525762 (GSK), a pan-BET inhibitor 11 . We previously showed that combined Tra and GSK prolonged the survival of VQ MM mice better than single agents alone 3 . In this new combination treatment, Cfz and Dex were administered once a week for two weeks, followed by one week of daily treatment with Tra and GSK. Although combo therapy slowed VQ growth after the first treatment cycle ( Figure S4E), it did not significantly prolong the survival of VQ-bearing mice ( Figure S4F). Overall, our data show that PIs only provide short-term disease control in the VQ model.

Combination trametinib and ponatinib treatment are synergistic against VQ myeloma cells in vitro.
Screening of the AOD IX panel identified 1000 nM ponatinib (Pon) as having high synergy with Tra against VQ myeloma cells ( Fig. 2A,B). Pon is a multi-tyrosine kinase inhibitor (TKI) currently approved for second-line treatment of chronic myeloid leukemia and Philadelphia chromosome-positive acute lymphoblastic leukemia 12 . Interestingly, no significant efficacy was observed for other TKIs as single agents or in combination with Tra ( Fig. 2A,B). Although dose-response testing of VQ 4935 and 4938 cell lines confirmed that Pon had limited effect as a single agent (Fig. 2C), it showed strong synergy with Tra against both VQ cell lines based on  Figure S5). Of note, this synergy appeared to be more prominent at higher concentrations of Pon (> 250 nM). We noticed that unlike other TKIs in the AOD IX panel, Pon is a potent pan-fibroblast growth factor receptor (FGFR) inhibitor with IC 50 values < 100 nM 15 . In human MM, FGFR3 is overexpressed in the t(4;14) high-risk subset as the translocation places FGFR3 expression under the control of the heavy chain immunoglobulin promoter on chromosome 14 16 . However, parental VQ-D2 cells show little to no expression of FGFR1-4 at the mRNA level, comparable to control plasma cells ( Figure S6A) and do not harbor any mutations in FGFR1-4 either (our unpublished results). To determine if FGFR inhibition plays a role in Pon efficacy, we next tested the FGFR1 inhibitor sorafenib 17 , the FGFR1-3 inhibitor pazopanib 18 , and the pan-FGFR inhibitors dovitinib and lenvatinib 19,20 in our VQ MM cell lines. Similar as other TKIs tested in the AOD IX panel, none of the FGFR inhibitors displayed single agent efficacy at 100 nM and 1000 nM or in combination with Tra ( Figure S6B), suggesting that the efficacy of Pon in VQ MM cells is not primarily mediated through FGFR signaling.
To further explore the efficacy of Pon and the synergy between Tra and Pon in vitro, we extended the drug treatment to several human myeloma cell lines (HMCLs), including OPM2 with t(4;14) translocation, MM.1S with oncogenic KRAS mutation, H929 with t(4;14) translocation and oncogenic NRAS mutation, and Delta47 without either event [21][22][23] . Interestingly, regardless of t(4;14) status, OPM2 and MM.1S were more responsive to Pon with an IC 50 at ~ 1.4-2 µM (Fig. S7A). These results are consistent with those from VQ MM cell lines.
When HMCLs were treated with combined Tra and Pon using the same range of concentrations as in VQ MM cell lines, we only observed a moderate additive effect in OPM2 and MM.1S cell lines (Fig. S7B), which may result from the minimal response of these HMCLs to 10-50 nM Tra.
Combination trametinib and ponatinib significantly prolongs survival of VQ mice in two different treatment regimens. Because Pon is clinically available as an oral agent and has not been evaluated in MM before, it was of interest to determine its in vivo efficacy alone and in combination with Tra. VQ-D1 MM cells were transplanted into sub-lethally irradiated recipient mice as previously described 3 . Once MM was established, recipients were divided into 4 groups with comparable gamma-globulin to albumin (G/A) ratios and similar complete blood count (CBC) parameters and treated with vehicle, Tra, Pon, and combined Tra and Pon (Figs. 3A and S8). Twenty-one days after treatment, all four groups of mice showed increased but indistinguishable G/A ratios (Fig. 3B) and the overall CBC results were unchanged ( Figure S8). Consistent with our in vitro (E) Serum protein electrophoresis was performed to quantify the G/A ratios in VQ recipient mice before treatment and at day 21 of treatment. Note: One Vehicletreated recipient was found dead and unable to be analyzed. (F) Kaplan-Meier survival curves were plotted against days after treatment. Log-rank test was performed. Note: One vehicle-treated animal was euthanized for reasons unrelated to treatment study and was excluded from analysis. *p < 0.05; **p < 0.01; ***p < 0.001. www.nature.com/scientificreports/ analysis, Pon treatment did not prolong the survival of VQ-D1 MM mice, while both Tra and combined Tra and Pon treatments did (Fig. 3C). Although combination treated mice had the longest overall survival, we did not observe significant difference between Tra-and Tra/Pon-treatment groups.

Scientific
We subsequently sought to determine if increasing the Tra dosage would significantly prolong the survival of Tra/Pon treated mice. To combat against the potential cumulative toxicity associated with higher Tra dose, we also took the 3-week on and 1-week off schedule as followed in a typical myeloma treatment regimen 1 . In a second in vivo experiment, recipients were divided into 3 groups with comparable G/A and CBC parameters and then treated with vehicle, Tra alone, or combination Tra/Pon in 28-day cycles (Figs. 3D and S9). Once again, no significant difference was observed in G/A ratios between groups after one treatment cycle (Figs. 3E and S9). Interestingly, although no survival benefit was observed with single agent Tra treatment, combo treated mice had significantly prolonged survival compared to the vehicle-treated group (Fig. 3F).
Ponatinib, but not trametinib, inhibits CD8 + T cell proliferation and activation in vitro. We investigated if the discrepancy between in vitro and in vivo combo treatment outcomes results from the drug effects on cytotoxic CD8 + T cells, which play an important role in anti-MM immunity 24 . To test this idea, CD8 + T cells were isolated from spleens of wildtype B6 mice, stained with CFSE dye, and activated via α-CD3/α-CD28 antibodies in the presence of Tra or Pon for 48 h (Fig. 4A). Tra treatment did not cause significant reduction in T cell proliferation as measured by CFSE tracing (Fig. 4B) and T cell activation as demonstrated by surface expression of CD69, DNAM-1, and PD-1 (Fig. 4C-E). By contrast, Pon treatment completely inhibited T cell proliferation and activation (Fig. 4B, C). Our results are consistent with prior studies showing that ponatinib and related BCR-ABL TKIs (e.g. dasatinib, imatinib) impair T cell function and viability in a dose-dependent manner [25][26][27] .

Discussion
While the introduction of IMiDs, PIs and monoclonal antibody treatments has revolutionized MM therapy, most patients still develop drug-refractory disease and eventually die of myeloma 1 . RAS pathway hyperactivation is a common molecular event in progressive myeloma, with almost 75% of drug-refractory myeloma patients harboring mutations in NRAS, KRAS, or BRAF 2 . In this study, our group used the recently developed VQ model of high-risk myeloma as a platform to develop new treatment regiments. To assess the effectiveness of existing MM therapies against VQ myeloma and expedite clinical testing, we carried out a re-purposing screen of 147 FDA-approved anti-cancer compounds against VQ cells in vitro (Fig. 1A). We found that VQ cells showed de novo resistance to venetoclax (Fig. 1D), likely owing to low Bcl2:Bcl2l1 and Bcl2:Mcl1 gene expression ratios (Fig. 1F) 8 . In addition, VQ cells showed increased resistance to Btz ( Figure S4A) and Cfz ( Figure S4D) compared to human MM cell lines in vitro 9 . Limited response to PI treatment was also observed in vivo, either as single agent ( Figure S4C) or as part of multi-drug treatment regimen ( Figure S4F). This is not altogether unexpected, as resistance to single agent Btz has been observed in patients harboring NRAS mutations 10 .
As a single agent, Tra displays dual effects on MM cells and T cells. Tra kills MM cells in a dose-dependent manner and downregulates surface PD-L1 expression in vitro 3 . Tra treatment of purified splenic CD8 + T cells in vitro did not significantly impact their proliferation (Fig. 4B) and activation (Fig. 4C-E). These in vitro T cell results are consistent with our prior study that downregulation of RAS/ERK signaling in Kras -/-T cells does not affect CD8 + T cell-mediated anti-leukemia activity in vivo 28 . In VQ-D1 MM recipients, Tra reversed exhausted cytotoxic CD8 + T cell phenotypes ( Figure S1) and prolonged their survival 3 . MEK inhibition has previously been shown to have pro-CD8 + T cell effects in oncogenic Kras-driven colorectal cancer 29 . In the context of the Kras G12D CT26 model, multiple groups found that MEK inhibition reduces PD-1 expression and prevents apoptosis of antigen-experienced CD8 + T cells 29,30 . Of note, low dose, continuous Tra treatment (Fig. 3C) worked better than high dose, on/off Tra treatment in vivo (Fig. 3F). This observation is consistent with cancer patients treated with Tra 31 .
Interestingly, among all the TKIs in the AOD IX panel and other FGFR inhibitors, only Pon showed moderate cell killing at 1 µM as single agent (Fig. 2B) and strong synergism with Tra at concentrations above 250 nM (Fig. 2D, E). Our data suggest that the efficacy of Pon in VQ MM cells does not primarily come from its inhibition of FGFRs. Rather, Pon may be targeting other tyrosine kinases due to its multi-RTK inhibitory function. This possibility is supported by our results from HMCLs, which responded to Pon regardless of their t(4,14) status (Fig. S7). Alternatively, Pon is a relatively more potent TKI compared to others; the clinical dose of Pon is indeed lower than those of other TKIs 32 . However, considering 1 µM is a fairly high concentration in vivo, we did not pursue other TKIs at higher concentrations.
Synergism between MEK and FGFR inhibition has previously been established in the context of KRAS mutant lung cancer: a shRNA screen of Tra-treated H23 KRAS G12C lung cancer cells identified FGFR1 signaling as compensatory for MEK inhibition 33 . H23 cells were highly susceptible to Tra and Pon combination treatment 33 , similar as what we observed in the VQ model. However, in a phase I clinical of KRAS mutant non-small cell lung cancer patients, it was found that combination Tra/Pon treatment was associated with cardiovascular (CV) and bleeding toxicities 34 . CV side effects of this combo treatment are not unsurprising, as Pon was temporarily withdrawn by the US Food and Drug Administration in 2013 for heart failure and other CV side effects before being returned with additional safety warnings and restrictions 35 .
Despite the strong synergistic effects of Tra + Pon in vitro, the in vivo effect was only moderately better than Tra alone (Fig. 3). We postulated that the anti-myeloma effects of combination Tra/Pon treatment were abrogated by their unfavorable impacts on non-myeloma cells, such as myeloma killing cytotoxic CD8 + T cells. Indeed, Pon, but not Tra, prevented CD8 + T cell proliferation and activation (Fig. 4B-E). Our data suggest that the negative impact of Pon treatment on T cells may counteract its MM-killing synergism with Tra in vivo. It is possible that other yet unidentified negative effects of these drugs exist. The complexity of MM in vivo warrants   Transplantation of myeloma cells. Donor cells from two moribund VQ-D1 MM bearing mice were pooled equally and resuspended in 100 μl of PBS containing 2% mouse serum (Jackson ImmunoResearch, 015-000-120). Eight-to fourteen-week-old CD45.1 + recipient mice were sub-lethally irradiated at 4.0 Gy using an X-RAD 320 Irradiator (Precision X-Ray Inc.) and transplanted with 5 × 10 5 of donor cells via intracardiac injection.

Serum protein electrophoresis (SPEP).
Mice were retro-orbitally bled with plain micro hematocrit tubes (Bris, ISO12772). Blood samples were spun in microtainer tubes (BD, 365967) at 2000×g for 10 min to collect serum. Serum was loaded into Hydragel agarose gel (Sebia, 4140) and processed using the Hydrasys instrument (Sebia) following the manufacturer's instruction. The processed film was scanned and pixel density of Albumin and γ-globulin bands were quantified using Adobe Photoshop.
Complete blood count. Peripheral blood samples were collected via retro-orbital bleeding and analyzed with a Hemavet 950FS (Drew Scientific).
Small compound treatment. For in vivo bortezomib treatment, bortezomib (Selleck) was dissolved in sterile PBS and administered at 0.5 mg/kg twice a week for four weeks via intra-peritoneal (IP) injection. For in vivo treatment of carfilzomib, dexamethasone, trametinib, and GSK525762, carfilzomib (Selleck) was dissolved in sterile PBS and administered at 16 mg/kg once a week via IP injection for two weeks. Dexamethasone (Selleck) was dissolved in sterile PBS and administered at 1 mg/kg once a week via IP injection for two weeks. Trametinib (Chemitek) was dissolved in 0.5% hydroxypropylmethylcellulose (Sigma) and 0.2% Tween-80 (Sigma) in distilled water (pH 8.0) and given orally at 0.5 mg/kg every morning for one week. GSK525762 (Chemitek) was dissolved in 1% methylcellulose (Sigma) containing 0.2% SDS and given orally at 15 mg/kg every evening for one week.
For in vivo treatment of trametinib and ponatinib, both compounds were dissolved in 0.5% hydroxypropylmethylcellulose (Sigma) and 0.2% Tween-80 (Sigma) in distilled water (pH 8.0) and administered at 0.2 mg/kg and 10.0 mg/kg respectively, via oral gavage daily. In a second treatment experiment, trametinib and ponatinib were dissolved in 0.5% hydroxypropylmethylcellulose (Sigma) and 0.2% Tween-80 (Sigma) in distilled water (pH 8.0) and administered at 0.5 mg/kg and 10.0 mg/kg, respectively, in 28 day cycles with 21 days of treatment followed by 7 days of rest.
Mice were not allocated to treatment groups in a blinded manner but were instead allocated so that G/A and CBC parameters were statistically similar between each group (One-way Analysis of Variance with Tukey-Kramer test). Small compounds were not administered to animals in a blinded manner due to necessary daily preparation of working concentrations for treatment. Animal care staff were blinded to experimental groups during animal assessment. Post-experiment data analysis was not blinded.
Flow cytometric analysis of hematopoietic tissues. Flow cytometric analysis of surface antigens on hematopoietic cells was performed as previously described 37 . Stained cells were analyzed on a LSRII Fortessa (BD Biosciences). Directly conjugated antibodies specific for the following mouse surface antigens were purchased from Biolegend unless specified: CD3(17A2), CD4 (eBioscience, GK1.5), CD8 (eBioscience, 53-6. Statistics. For Kaplan-Meier survival curves, survival differences between groups were assessed with the log-rank test, assuming significance at p < 0.05. Unpaired, two-way t Test was used to determine significant differences between two groups unless specified. One-way Analysis of Variance with Tukey-Kramer test was used to determine the significance between multiple data sets simultaneously unless specified, assuming significance at p < 0.05. Statistical analysis was carried out using GraphPad Prism v9.2.0.

Data availability
The datasets generated and/or analyzed during the current study are available from the corresponding author on reasonable request.