Inhibition of Mitochondrial Antioxidant Defense and CDK4/6 in Mesothelioma

Advanced mesothelioma is considered an incurable disease and new treatment strategies are needed. Previous studies have demonstrated that mitochondrial antioxidant defense proteins and the cell cycle may contribute to mesothelioma growth, and that the inhibition of these pathways may be effective against this cancer. We demonstrated that the antioxidant defense inhibitor auranofin and the cyclin-dependent kinase 4/6 inhibitor palbociclib could decrease mesothelioma cell proliferation alone or in combination. In addition, we determined the effects of these compounds on colony growth, cell cycle progression, and the expression of key antioxidant defense and cell cycle proteins. Auranofin and palbociclib were effective in decreasing cell growth and inhibiting the above-described activity across all assays. Further study of this drug combination will elucidate the contribution of these pathways to mesothelioma activity and may reveal a new treatment strategy.


Introduction
Mesothelioma is a devastating and recalcitrant cancer that arises from the pleura or lining of the peritoneum. Treatments for advanced mesothelioma in the first-line setting may include chemotherapy with cisplatin and pemetrexed or immunotherapy with checkpoint inhibitors, primarily nivolumab and ipilimumab [1,2]. However, mesothelioma is incurable in the majority of cases with improvement in median overall survival with chemotherapy or immunotherapy measured in months [1,2]. To date, clear mechanisms for lack of longer term benefit from current treatments have been elusive. In addition, biomarkers for treatment of mesothelioma have not yet been validated. New treatment strategies against mesothelioma are needed. Two pathways that show some promise but remain to be significantly studied include mitochondrial antioxidant defense and the cell cycle.

Interaction between Mitochondrial Antioxidant Defense and the Cell Cycle
Increased expression of peroxiredoxin 3 (Prx3) and manganese superoxide dismutase (MnSOD) has been demonstrated in mesothelioma tumors [14,15]. CDK4 has been demonstrated to increase MnSOD activity via phosphorylation, causing superoxide levels to decline in keratinocytes [18]. FOXM1-transcribed Cyclin D1 and CDK4 enhance activity of this pathway [17,18]. Based on these findings that potentially connect mitochondrial antioxidant defense and cell cycle pathways, we hypothesized that downregulation of Trx2 via auranofin combined with CDK4/6 inhibition via palbociclib would be more effective than either agent alone to decrease mesothelioma cell growth. In this paper, we describe in vitro activity of either agent alone or in combination against mesothelioma cells.

Inhibition of Cell Proliferation by Auranofin and Palbociclib
The growth-inhibitory effects of CDK4/6 inhibition were evaluated by determining the IC 50 for palbociclib and auranofin against mesothelioma cell lines H2052, H2373, H2452, and H2461. Met5A cells were utilized as a non-transformed control cell line. Palbociclib and auranofin both resulted in decreased cell proliferation at 72 h for all cell lines ( Figure 1A,B).
For comparison to a current standard of care treatment in mesothelioma, H2373 and H2452 cells were exposed to increasing amounts of pemetrexed, a folate pathway inhibitor in similar fashion to palbociclib and auranofin. Figure 2 shows the IC 50 curve for pemetrexed in comparison to palbociclib and auranofin. The IC 50 was 14.58 nM for pemetrexed in H2373 cells and 34.71 nM in H2452 cells. We also exposed control 3T3 cells to palbociclib and auranofin in separate experiments for 72 h to determine effects on a control line. The IC 50 was 2.71 µM for auranofin and 22.6 µM for palbociclib. Palbociclib and auranofin both resulted in decreased cell proliferation at 72 h for all cell lines ( Figure 1A,B).  For comparison to a current standard of care treatment in mesothelioma, H2373 and H2452 cells were exposed to increasing amounts of pemetrexed, a folate pathway inhibitor in similar fashion to palbociclib and auranofin. Figure 2 shows the IC50 curve for pemetrexed in comparison to palbociclib and auranofin. The IC50 was 14.58 nM for pemetrexed in H2373 cells and 34.71 nM in H2452 cells. We also exposed control 3T3 cells to palbociclib and auranofin in separate experiments for 72 h to determine effects on a control line. The IC50 was 2.71 µM for auranofin and 22.6 µM for palbociclib.

Inhibition of Cell Proliferation by Combination of Auranofin and Palbociclib
We hypothesized auranofin and palbociclib could act in an additive fashion or synergistic fashion; therefore, we conducted cell proliferation experiments with varying combinations of concentrations of each agent per the methods of Chou-Talalay [30].  Determination of IC 50 pemetrexed in mesothelioma cell lines. Mesothelioma cell lines H2373 and H2452 were exposed to increasing concentrations of pemetrexed for 72 h. Cell viability was measured by CCK8 assay. * Notes statistically significance.

Inhibition of Cell Proliferation by Combination of Auranofin and Palbociclib
We hypothesized auranofin and palbociclib could act in an additive fashion or synergistic fashion; therefore, we conducted cell proliferation experiments with varying combinations of concentrations of each agent per the methods of Chou-Talalay [30]. Figure 2 exhibits viability relative to control for each drug concentration combination for cell lines H2373 and H2452. For both cell lines, several combinations were consistent with a synergistic effect in decreasing cell proliferation (Figures 3 and 4). Table 1 exhibits the combination indices for each combination of palbociclib and auranofin. Numerous combinations indices were < 1, suggesting synergism. Figure 2. Determination of IC50 pemetrexed in mesothelioma cell lines. Mesothelioma cell lines H2373 and H2452 were exposed to increasing concentrations of pemetrexed for 72 h. Cell viability was measured by CCK8 assay. *Notes statistically significance.

Inhibition of Cell Proliferation by Combination of Auranofin and Palbociclib
We hypothesized auranofin and palbociclib could act in an additive fashion or synergistic fashion; therefore, we conducted cell proliferation experiments with varying combinations of concentrations of each agent per the methods of Chou-Talalay [30]. Figure 2 exhibits viability relative to control for each drug concentration combination for cell lines H2373 and H2452. For both cell lines, several combinations were consistent with a synergistic effect in decreasing cell proliferation (Figures 3 and 4). Table 1 exhibits the combination indices for each combination of palbociclib and auranofin. Numerous combinations indices were < 1, suggesting synergism. Results of CCK8 assay are expressed as viability relative to control for each treatment. **, ***, **** denotes a statistically significant difference between the combinations, signifying a synergistic effect of the combination over either drug administered individually. Results of CCK8 assay are expressed as viability relative to control for each treatment. **, ***, **** denotes a statistically significant difference between the combinations, signifying a synergistic effect of the combination over either drug administered individually.

Colony-Forming Assays
Palbociclib and auranofin each significantly decreased H2373 and H2452 colony formation ( Figure 5). Palbociclib resulted in nearly complete inhibition of colony formation

Colony-Forming Assays
Palbociclib and auranofin each significantly decreased H2373 and H2452 colony formation ( Figure 5). Palbociclib resulted in nearly complete inhibition of colony formation at a concentration of 50 µM in both cell lines. Auranofin resulted in nearly complete inhibition of colony formation at 5 µM and at 10 µM. These concentrations were very close to the relative IC 50 values for each drug against mesothelioma cell lines as in Figure 1.

Cell Cycle Analysis
Effects of palbociclib on the cell cycle following a 48 h treatment in mesothelioma cells demonstrated distinct inhibition of the cell cycle at G1/S with a significant increase in the percentage of cells in G1 as evaluated by flow cytometry. Treatment with auranofin in mesothelioma cell lines resulted in an increased proportion of cells in the sub-G1 fraction which may indicate an apoptotic response to the thioredoxin inhibitor. See Figure 6 for results for both experiments.

Cell Cycle Analysis
Effects of palbociclib on the cell cycle following a 48 h treatment in mesothelioma cells demonstrated distinct inhibition of the cell cycle at G1/S with a significant increase in the percentage of cells in G1 as evaluated by flow cytometry. Treatment with auranofin in mesothelioma cell lines resulted in an increased proportion of cells in the sub-G1 fraction which may indicate an apoptotic response to the thioredoxin inhibitor. See Figure 6 for results for both experiments.

Modulation of Gene Expression in Response to Treatment with Palbociclib or Auranofin
We examined the expression of the four predominant cell cycle cyclins and CDKs to establish if inhibition of CDK4/6 by palbociclib was associated with alteration of CDKs or cyclins that have been implicated in acquisition of resistance to palbociclib in luminal breast cancer [31]. Substantially enhanced expression of Cyclins D1 and E1 were observed in H2052, but not in either H2373 or H2452. In contrast, auranofin resulted in the loss of Cyclin D1 protein expression in two of the three cell lines but did affect Cyclin E1 expression (Figure 7). Both drugs resulted in the loss of Cyclin A and B1 gene expression by 72 h (Figure 7), although the loss was less pronounced in the H2452 cells being observed only at the higher concentrations employed for both drugs. Auranofin at 2.5 µM for 72 h was > 90% lethal for the two Bap1-expressing cell lines H2052 and H2373 and thus was not analyzed. oma cell lines. Auranofin treatment was variably associated with decreasing CDK1 le depending on the cell line and drug concentration which mirrored the pattern obse for Cyclin D1 loss (Figure 7). CDK2 levels increased in all palbociclib-treated c whereas CDK4 and CDK6 levels were only modestly increased or unchanged with pa ciclib treatment. Auranofin resulted in loss of CDK2, CDK4, and CDK6 expression in of the three cells lines (Figure 7, Supplementary Table S1). We examined both FOXM1 and Rb transcription factors as both are known tar for the Cyclin D1/CDK4 phosphorylation that promotes cell cycle progression. As sh in Figure 8, treatment with palbociclib substantially reduced the phosphorylation of R S780, a site acted on equally by both Cyclin D1/CDK4 and Cyclin D1/CDK6. The aur fin-treated cells had a much smaller loss of pRb S780 expression, but in contrast to palbociclib treatment, overall levels of Rb expression decreased as well, rather than shift from the phosphorylated to the non-phosphorylated form of Rb observed Palbociclib led to a substantial loss of CDK1 gene expression in all three mesothelioma cell lines. Auranofin treatment was variably associated with decreasing CDK1 levels, depending on the cell line and drug concentration which mirrored the pattern observed for Cyclin D1 loss (Figure 7). CDK2 levels increased in all palbociclib-treated cells, whereas CDK4 and CDK6 levels were only modestly increased or unchanged with palbociclib treatment. Auranofin resulted in loss of CDK2, CDK4, and CDK6 expression in two of the three cells lines (Figure 7, Supplementary Table S1).
We examined both FOXM1 and Rb transcription factors as both are known targets for the Cyclin D1/CDK4 phosphorylation that promotes cell cycle progression. As shown in Figure 8, treatment with palbociclib substantially reduced the phosphorylation of Rb at S780, a site acted on equally by both Cyclin D1/CDK4 and Cyclin D1/CDK6. The auranofin-treated cells had a much smaller loss of pRb S780 expression, but in contrast to the palbociclib treatment, overall levels of Rb expression decreased as well, rather than the shift from the phosphorylated to the non-phosphorylated form of Rb observed with palbociclib. FOXM1 expression was lost in response to either drug treatment, with palbociclib inducing a greater loss.
Cell cycle arrest is effective in reducing the rate of tumor growth; however, treatmentinduced cell death is required for tumor shrinkage and thus we examined the expression of pro-apoptotic and pro-survival genes whose levels are modulated in malignant mesothelioma [32][33][34]. The pro-apoptotic genes Bax and/or Bid were induced in response to both auranofin or palbociclib, although the effect was more dramatic in the H2052 and H2452 cell lines (Figure 8, Supplementary Table S1). Of the four pro-survival genes, survivin and/or Mcl-1 levels were dramatically reduced in response to auranofin and palbociclib treatment. Bcl-xL loss occurred in all three cell lines with auranofin treatment, while palbociclib only induced Bcl-xL loss in the H2052 and H2452 cells. XIAP (X-linked inhibitor of apoptosis) expression remained unchanged by either drug. both auranofin or palbociclib, although the effect was more dramatic in the H2052 a H2452 cell lines (Figure 8, Supplementary Table S1). Of the four pro-survival genes, s vivin and/or Mcl-1 levels were dramatically reduced in response to auranofin and pal ciclib treatment. Bcl-xL loss occurred in all three cell lines with auranofin treatment, wh palbociclib only induced Bcl-xL loss in the H2052 and H2452 cells. XIAP (X-linked inh itor of apoptosis) expression remained unchanged by either drug.

Cell Lines and Reagents
The mesothelioma cell lines used in this study (H2052, H2373, H2452, and H20 and the control cell lines, Met5A (we note as Met5A AD) and 3T3, were obtained fr either the ATCC (American Type Culture Collection) or in collaboration with Dr. Rob Kratzke (University of Minnesota). All mesothelioma cell lines lack functional p16INK The cells were grown in RPMI-1640 medium (Gibco BRL, Grand Island, NY, USA) supp mented with 10% newborn bovine serum (Sigma, St. Louis, MO, USA), and 1x concen tion of antibiotic/antimycotic reagent (Gibco BRL, Grand Island, NY, USA) at 37 °C a

Cell Proliferation Assays
Live cells, as determined by trypan blue dye exclusion assay, were counted on a Countess II FL Automated Cell Counter and plated on 96-well plates (3000 cells/100ul/well) in the RPMI media (including supplements). After 24 h, palbociclib or auranofin at varying concentrations was added to the wells and gently vortexed to mix. After a 72 h incubation at 37 • C and 5% CO 2 , 10 µL of solution from Cell Counting Kit-8 (Dojindo Laboratories, Kunamoto, Japan) was added to each well and the plates were incubated for 2 h. The plates were read at 450 nm using a SpectraMax M5 microplate reader (Molecular Devices, Sunnyvale, CA, USA). The reduction in optical density represented the reduction in mitochondrial succinate dehydrogenase activity, hence the reduction in surviving cell numbers. For drug combination studies, increasing amounts of one drug were added to H2373 or H2452 cells (0.5, 1, 2.5, 5, and 10 µM) while holding the other at a fixed concentration. Fixed concentrations were at the same 5 different concentrations (0.5, 1, 2.5, 5, and 10 µM). Combination studies were analyzed via the methods of Chou-Talalay, and Combination Indices (CI) were calculated via Compusyn [30]. In addition, ANOVA for multiple comparisons were utilized to evaluate for statistically significant differences amongst multiple groups. Experiments were conducted in triplicate.

Colony-Forming Assays
Cells were grown in 6-well plates for 24 h, then treated with palbociclib or auranofin for 72 h [35]. The cells were then trypsinized, counted (including all dead cells) and plated on 24-well plates (400-1000cells/well). Plates were then grown for 10-14 days allowing single live cells to grow into colonies. Colonies were then fixed and stained with crystal violet and then read on the Licor CLx at 700 nm wavelength. The intensity of the colony staining was quantified using the Licor and associated software (Image Studio Version 5.2).

Cell Cycle Analysis
After treatment with drug or dimethyl sulfoxide (DMSO) control for 48 h, mesothelioma cell lines were fixed in 70% cold ethanol by dropwise addition, and then stained with FxCycle TM PI/RNase Staining Solution (Catalog #F10797, Invitrogen, USA), following the manufacturer's protocol. Flow cytometry was performed using a FACSAria III Cell Sorter (BD Biosciences, San Jose, CA, USA), and results were analyzed by FACSDiva software Version 6.1.3 (BD Biosciences).

Immunoblotting
Cells were plated to reach a density of~40% 24 h after plating and treated with the indicated concentration of drugs 24 h after plating. At 72 h of treatment, cells were harvested by scraping, and following centrifugation and washing with 1X PBS, cell pellets were flash frozen in powdered dry ice and stored at -80 • C. Cells were lysed with RIPA buffer (Harlow and Lane, CSH protocol) containing 1X Halt TM Protease and Phosphatase inhibitor (Thermo Scientific, Waltham, MA, USA) following resuspension; lysates were flash frozen in powdered dry ice, thawed on ice, vortexed for 10 s, and centrifuged at 13,000 rpm at 4 • C for 10 min. The supernatant (lysate) was aliquoted, flash frozen in powdered dry ice, and stored at −80 • C. Protein content was determined using a Millipore (

Discussion
Mitochondrial antioxidant defense and the cell cycle are two areas that may be targeted in mesothelioma. Previously, our group demonstrated via analysis of The Cancer Genomic Atlas that some tumors have increased mRNA expression associated with genes involved in antioxidant defense, such as Trx2 and Prx3 [36]. Previously, Cunniff et al. have demonstrated that inhibition of Prx3 by thiostrepton increases mitochondrial H 2 O 2 , disrupts mitochondrial energetics, and is associated with cytotoxicity in mesothelioma cells [12]. In addition, they demonstrated that thiostrepton and gentian violet, which downregulate TXN2, were associated with decreased tumor growth in mice mesothelioma xenografts [12]. We chose to evaluate auranofin, as it is an FDA-approved drug for rheumatoid arthritis and is known to downregulate Txn2 [19]. In our work, we demonstrated that it has clinically relevant potency against mesothelioma cells in vitro in cell proliferation and colon-forming assays.
The tumor suppressor p16INK4a has been shown to be downregulated in up to 90-100% of mesothelioma tumors, depending on the histology (epithelioid, sarcomatoid, or biphasic) [25]. p16INK4a inhibits CDK4 and CDK6 (CDK4/6), which leads to cell cycle arrest [20][21][22][23][24]. The drug palbociclib is a selective CDK4/6 inhibitor and is one of three such inhibitors (along with abemaciclib and ribociclib) approved for use in advanced breast cancer, and, as such, are considered standard of care. However, up to now, there has been very little evaluation of CDK4/6 inhibition in mesothelioma. Two exceptions are the reports by Bonelli et al., where they evaluate palbociclib with or without PI3K/MTOR inhibition in mesothelioma cells [28,29]. They reported EC 50 values ranging from 0.28 to 1.2 µM, which are slightly lower than what we observed. Part of this difference could be a slight variation in the protocols for cell proliferation assays. However, we also demonstrated that palbociclib results in a decrease in colony formation, which is a complementary technique to assess growth inhibition characteristics of a particular drug candidate.
We subsequently conducted combination inhibitor studies of auranofin and palbociclib. CDK4 has been demonstrated to phosphorylate manganese superoxide dismutase (MnSOD), which is involved in superoxide processing to H 2 O 2 [18]. In addition, CDK4 enhances activity of FOXM1, which has been demonstrated to affect mitochondrial antioxidant protein expression [17]. For several dosing combinations, we demonstrated that the combination of auranofin and palbociclib exhibited synergy against mesothelioma cells in proliferation experiments.
The immunoblotting data suggest that in response to palbociclib inhibition of CDK4/6 activity, the loss of Cyclin A and B expression as well as CDK1 is consistent with cell cycle arrest. However, the increased expression of Cyclins D1 and E1 as well as CDK2, CDK4, and CDK6 suggests that compensation for CDK4/6 inhibition is being induced. Although auranofin also resulted in the loss of Cyclins A and B and CDK1 in H2052 and H2452, the effect was less pronounced, suggesting reduced cell cycle progression rather than arrest. This supposition is further supported by the cell cycle analysis data presented in Figure 6 and the observation that the palbociclib-treated cells used in the study showed substantial reduction in number relative to the untreated controls. Further, the enhanced expression of Cyclin D1 and CDKs 4 and 6 in response to palbociclib inhibition of CDK4/6 activity suggests that this is one possible mechanism by which the drug would lose effectiveness. Further, the dramatic upregulation of CDK2 activity and its partner Cyclin E is another mechanism that has been shown to promote resistance both acquired and intrinsic to CDK4/6 inhibitors [37,38].
In contrast, treatment by auranofin resulted in a gene expression profile that suggested overall loss of viability by multiple mechanisms, not just a predominant effect on cell cycle. Its mechanism of action (inhibition of cytosolic and mitochondrial thioredoxin reductase activity) would be expected primarily to induce ROS-driven cell death; however, it caused a substantial decrease in Cyclins A, B1, and D1, as well as all the CDKs and the cell cycle transcription factors FOXM1 and Rb, suggesting altered cell cycling as a contributing factor to loss of cell viability. However, in contrast to palbociclib, where arrest was induced at the G1/S transition, auranofin appears to be associated with a dramatic decrease in the percentage of cells at G1/S or G2/M checkpoints. Various studies have shown a significant role of mitochondrial-produced ROS in regulating the activation or degradation of the G1/S phase Cyclins D1, A, and E and G2/M Cyclin B1, implying that this is one potential method by which the reduced expression of the cyclins following auranofin treatment is occurring [39,40]. In one study of SOD1 inhibition with the compound called LD100 in HeLa and DU145 cells, exposure to LD100 resulted in reduced mRNA levels of CDK4, Cyclin D1, Cyclin E1, and Cyclin B1 and increased mRNA levels of CDKN1A and CDKN2D [41,42].
Transcriptional downregulation of the cyclins and CDKs due to loss of FOXM1 is another mechanism, as they are all know targets of FOXM1 [17]. However, the loss of FOXM1 was even more pronounced in the palbociclib-treated cells, yet Cyclin D1 and CDK expression was decreased only in the auranofin-treated cells. This is intriguing as it suggests the efficacy of the combination of the two drugs, with enhanced lethality at lower concentrations while in combination, which may be due to the inability of the cells in the presence of auranofin to upregulate the expression of the cyclins and CDKs to overcome the effects of CDK4/6 inhibition.
Induction of pro-apoptotic Bax and Bid, both associated with mitochondrial-driven cell death, is consistent with a potential ROS mechanism. Inhibition of CDK4/Cyclin D1 kinase activation of SOD2 in the mitochondria by palbociclib and direct inhibition of thioredoxin 2 reduction in the mitochondria, which drives the detoxification of mitochondrial-generated ROS, are potential pathways for the increased expression of Bax and Bid. Loss of prosurvival gene expression was observed in all of the cell lines, with survivin being the only one that was consistent across all the mesothelioma cell lines in response to either drug. Mcl-1 and Bcl-xL loss was more consistently observed in the auranofin-treated cells, driving cell death as would be expected with the loss of redox capacity of both the mitochondrial and cytosolic thioredoxin systems. Taken together, these data suggest that cell death is the predominate mechanism resulting in loss of viability in auranofin-treated cells, which is consistent with its lethality at 2.5 µM in two of the three mesothelioma cell lines.

Conclusions
In conclusion, based on the constellation of findings presented, including the potent effects of palbociclib and auranofin against mesothelioma cell proliferation, colony formation, and the cell cycle, as well as the effect on Cyclin/CDK and pro-apoptotic/anti-apoptotic protein expression, we propose a potential model for how mitochondrial antioxidant defense and CDK4/6 may interact. Combination therapy with inhibition of mitochondrial antioxidant defense and CDK4/6 holds promise as a potential therapeutic strategy in mesothelioma. Future studies will be needed to fill in gaps and add to the model. grants providing salary support. This material is based upon work supported by the Department of Veterans Affairs (specifically the Veterans Health Administration). We thank Robert Kratzke, University of Minnesota, for the generous gift of several mesothelioma cell lines.