Screening a library of approved drugs reveals that prednisolone synergizes with pitavastatin to induce ovarian cancer cell death

The survival rate for patients with ovarian cancer has changed little in the past three decades since the introduction of platinum-based chemotherapy and new drugs are needed. Statins are drugs used for the treatment and prevention of cardiovascular diseases. Recent work from our laboratory has shown that pitavastatin has potential as a treatment for ovarian cancer if dietary geranylgeraniol is controlled. However, relatively high doses of statins are required to induce apoptosis in cancer cells, increasing the risk of myopathy, the most common adverse effect associated with statins. This makes it desirable to identify drugs which reduce the dose of pitavastatin necessary to treat cancer. A drug-repositioning strategy was employed to identify suitable candidates. Screening a custom library of 100 off-patent drugs for synergistic activity with pitavastatin identified prednisolone as the most prominent hit. Prednisolone potentiated the activity of pitavastatin in several assays measuring the growth, survival or apoptosis in several ovarian cancer cells lines. Prednisolone, alone or in some cases in combination with pitavastatin, reduced the expression of genes encoding enzymes in the mevalonate pathway, providing a mechanistic explanation for the synergy.

www.nature.com/scientificreports www.nature.com/scientificreports/ in vitro 11 . Secondly, many of the statins have a relatively short half-life, and once-daily dosing is inadequate to maintain the continual inhibition of HMGCR that we have shown to be necessary to induce apoptosis 12,13 . The choice of statin tested in clinical trials has been, in our opinion, uniformly inappropriate 10 . Hydrophilic statins are less potent in a cancer setting, while lipophilic statins are the most potent anti-cancer agents 12 but they are generally associated with short metabolic half-lives 13 . Pitavastatin is the only lipophilic statin with an adequate half-life to maintain continual inhibition of HMGCR using a practical dosing schedule. Consequently, we consider that high doses of pitavastatin, administered twice daily, are the most likely to succeed. Lastly, dietary sources of geranylgeraniol can interfere with the anti-tumour activity of pitavastatin, suggesting diet should be controlled during therapy 9 . These observations suggest clinical trials of pitavastatin are warranted.
The use of statins at a high dose and with continuous exposure, as we propose, brings with it an increased risk of myopathy, the most common adverse effect associated with statin use. In some rare cases this can lead to rhabdomyolysis and the incidence of this is likely to increase if high statin doses are used 14 . It is, therefore, desirable to identify compounds which synergize with the anti-tumour activity of pitavastatin in order to reduce the dose needed and potentially reduce the incidence of myopathy. Combination therapies are among the most successful forms of treatment of cancer. Tumours, especially in adults, are associated with multiple mutations and intratumoral clonal heterogeneity is often observed as a result of several different pathological mechanisms participating in their evolution 15 . Thus, drug combinations can be more successful than single agents 16 . Drug combinations can also simultaneously affect different signalling pathways in individual cancer cells, potentially leading to synergistic activity. Drug combinations may also reduce the emergence of drug-resistant subpopulations. Lastly, there is an historical precedent for the use of drug combinations because many chemotherapeutic regimens incorporate several different drugs. Therefore, combining drugs offers the prospect of obtaining a more sustained clinical response. We, and others, have already shown that bisphosphonates such as zoledronatate potentiate the activity of statins 11,[17][18][19][20][21] and dipyridamole has been shown to potentiate the activity of atorvastatin 22 .
To identify additional drugs which might be synergistic with pitavastatin, we screened a library of 100 off-patent clinically approved drugs in combination with pitavastatin using cell growth assays. This library was designed to allow testing of the drugs at clinically achievable concentrations 23 . This library has previously been screened to identify niclosamide, an anti-helminthic drug, as a potential therapy for multiple myeloma as it killed several cell lines at clinically achievable non-toxic concentrations 23 . In this study, we show that prednisolone potentiates the activity of pitavastatin against a panel of ovarian cancer cell lines.

Material and Methods
Compounds. Pitavastatin (Livalo, Adooq), Prednisolone, farnesol, geranylgeraniol and mevalonate (Sigma-Aldrich) were prepared as 20 mM stock solution in DMSO. The custom-made drug repurposing library (FMC1) was provided by Dr. Farhat Khanim, School of Biosciences, University of Birmingham and is comprised of off-patent, mainly orally bioavailable drugs, at a concentration which is a multiple of each drug's plasma C max observed in patients 23 .
screening the drug library with pitavastatin. Ovcar-4 cells were seeded (5000 cells/well) in a 96-well plate. The next day, cells were exposed to vehicle (DMSO), pitavastatin (10 µM), a library compound, or a combination of pitavastatin and a compound from the library. The experimenters were blind to the identity of the drugs which were each given an anonymized code. Each drug was tested in triplicate in two independent experiments. After 72-hours incubation, cells were fixed and relative surviving cells were estimated by staining with SRB and measuring A 570 as described previously 24 . The Bliss independence criterion was used to estimate the expected effect of the drug combination if the drugs interacted additively 25 . The "Bliss excess" was calculated from the difference between the observed effect and expected additive effect.
Cell growth assays. Human ovarian cancer cells (5000 cells/well) were plated in 96-well plates. The following day, cells were exposed 18 different concentrations of the drugs for 72-hour, with the exception of Cov-318 and Cov-362 cells which were incubated for 120-hour because of their slower growth rate. Cells were stained with sulforhodamine B as previously described 24 . IC 50 values and Hill coefficients were determined using Graphpad Prism 6.
To evaluate synergy, complete concentration-response curves for pitavastatin were measured in the absence or presence of a fixed concentration of prednisolone. Combination indices were calculated as described 26 at fraction affected = 0.5. Some cells were also exposed to 20 μM mevalonate, 10 μM farnesol (FOH), or 10 μM geranylgeraniol (GGOH) as indicated. spheroid cultures. GravityTRAP ULA Plates (InSphero) were pre-wet with 40 µl of medium before seeding 500 cells in 70 µl growth medium per well. Following centrifugation (ALC PK120 Centrifuge, 1 min at 900 rpm), the plates were returned to the incubator. After 3-5 days, spheroids were observed. Thereafter, drugs were added in 30 µL of growth medium. Ovcar-4 or Cov-362 cells were incubated for 72 or 120 hours, respectively before relative ATP levels were measured by addition of 25 µL of cell Titer-Glo Luminescent assay reagent (Promega, www.nature.com/scientificreports www.nature.com/scientificreports/ Madison, WI, USA). The effect of the combination was compared to that expected for an additive interaction using the Bliss independence criterion as described above.
Caspase-Glo3/7 assays. For each experiment two 96 well plates containing 5000 cells per well in 80 µl of growth medium were prepared. After 48 hours, caspase activity was measured in one plate by addition of 20 µL of Caspase-Glo 3/7 reagent (Promega, Madison, WI, USA). The second plate was stained with SRB as described above and the caspase activity normalised to the SRB staining.
Annexin V/propidium iodide staining. Ovcar-4 or Cov-362 cells were seeded at density of 2 × 10 6 cells per well of a 6 well plate in 2 mL of growth medium. The following day, drugs were added in 20 μL of growth medium to the indicated final concentration. Ovcar-4 and Cov-362 cells were incubated with drugs for 48 and 72 hours, respectively. The cells were labelled using an annexin-V FITC kit (Miltenyi biotech). Cells were trypsinized, washed in ice-cold PBS and centrifuged at 300 × g for 5 minutes. The pellets were re-suspended in 1 ml of binding buffer, and centrifuged for 10 minutes at 300 × g. The pellets were re-suspended in 100 µl of annexin V binding buffer and 10 µl of annexin V fluorochrome were added to each sample and incubated for 10 minutes in dark at room temperature. The washing step were repeated with 1 ml of annexin V binding buffer. Lastly, the cells were re-suspended in 500 µl annexin V Binding Buffer and 5 µl of propidium iodide (1 µg/ml) added before analysis by flow cytometry. The viability of cells was defined as alive (annexin V-negative and PI-negative), early apoptotic cells (annexin V-positive and PI-negative), late apoptotic cells or dead cells (annexin V-positive and PI-positive) and necrotic cells (annexin V-negative and PI-positive).

Results testing a library of compounds in combination with pitavastatin.
Ovcar-4 cells, which are considered representative of high grade serous ovarian cancer 28 , were used to test the effect in cell growth assays of pitavastatin alone and in combination with individual compounds from a library (FMC1) of off-patent, licensed drugs 23 . Six compound showed significant growth inhibitory activity against Ovcar-4 cells when they were tested as single agents at a concentration similar to their C max achieved in patients. single agent activity in panel of ovarian cancer cell line. The single agent activity of prednisolone was determined using a panel of ovarian cancer cells considered representative of high grade serous ovarian cancer, namely Ovcar-4, Ovcar-3, Ovsaho, Cov-318 and Cov-362 cells 28 . Prednisolone, as a single agent, showed weak growth inhibitory activity in all ovarian cancer cell lines at concentrations up to 500 μM and an accurate IC 50 could not be determined using pharmaceutically-relevant concentrations. This agrees with other studies that report that corticosteroids have insignificant growth inhibitory activity against solid tumours 29 . In contrast, and as we have previously reported 9,17 , pitavastatin inhibited the growth of tested cell lines with an IC 50 s ranging from 1.1 to 4.8 µM (Table 1). www.nature.com/scientificreports www.nature.com/scientificreports/ pitavastatin combination with prednisolone. To confirm the results of the screen, a range of concentrations of pitavastatin were combined with a fixed concentration of prednisolone (70 µM) and their activity assessed in cell growth assays. At this concentration, prednisolone has no measureable effect as a single agent, so any change in the apparent potency of pitavastatin must reflect a drug interaction. Prednisolone potentiated the activity of pitavastatin against all the ovarian cancer cell lines that were tested (Ovsaho, Cov-318, Cov-362, Ovcar-3 and Ovcar-4), with significant reduction in pitavastatin IC 50 s (Fig. 1A). To confirm this formally, combination indices were calculated. Significant synergy between prednisolone and pitavastatin was observed in all the cell lines (Fig. 1B).
Effect of mevalonate pathway intermediate metabolites on the combination. We 9 , and others [30][31][32] , have previously shown that some mevalonate pathway metabolites downstream of HMGCR including mevalonate and geranylgeraniol reduce the cytotoxic effect of statins. To determine if the anti-proliferative activity of the pitavastatin and prednisolone combination resulted from inhibition of mevalonate pathway, Ovcar-4 and Cov-362 cells were exposed to the drug combination and further supplemented with mevalonate, farnesol or geranylgeraniol. The addition of mevalonate to cells significantly reduced the growth inhibitory activity of drug combination. Furthermore, supplementing the combination with geranylgeraniol but not farnesol also significantly prevent growth inhibition (Fig. 2). These results suggested that activity of combination is mediated mainly through inhibition of mevalonate pathway and most likely through inhibition of geranylgeranylation.
Atp assay in spheroid cultures. To recapitulate the 3D architecture of tumours more closely in vitro, Ovcar-4 and Cov-362 spheroids were prepared and the effect of the pitavastatin and prednisolone combination was evaluated. ATP was measured as a surrogate of surviving cell number. The combination of prednisolone and pitavastatin reduced ATP significantly more than would have been anticipated if the drugs acted additively (estimated from the Bliss independence criterion; Fig. 3), thereby confirming a synergistic interaction between the two drugs.
prednisolone and pitavastatin synergistically induce apoptosis. When Ovcar-4 cells were treated with solvent or with prednisolone and viewed by phase contrast microscopy they retained their original morphology. In contrast, cells exposed to pitavastatin alone detached from the plate surface ( Supplementary Fig. 1) were round, shrunken and with blebs and this was more pronounced in cells treated with the drug combination. To determine if these morphological changes resulted from apoptosis, Ovcar-4 and Cov-362 cell lines were exposed to pitavastatin, prednisolone or the combination of the two agents before annexin V and propidium www.nature.com/scientificreports www.nature.com/scientificreports/ iodide staining was measured by flow cytometry. There were significantly more early apoptotic or late apoptotic dead cells in samples treated with the drug combination than in cells treated with pitavastatin alone (Fig. 4) suggesting that the drug combination synergistically induced apoptosis. To confirm this, caspase-3/7 activity and PARP cleavage were assessed. Although prednisolone had negligible effect on caspase 3/7 activity on its own, the combination of pitavastatin with prednisolone caused significantly more caspase activation than that caused by pitavastatin alone. Consistent with this, immunoblot analysis demonstrated that the prednisolone and pitavastatin combination caused significant accumulation of cleaved PARP that was greater than that observed with each single agent (Fig. 5).
Identification of a potential mechanism underlying synergy between pitavastatin and prednisolone. Prednisolone regulates the expression of genes by binding to the glucocorticoid receptor, a ligand-dependant transcription factor. This raised the possibility that the synergy between pitavastatin and prednisolone occurred as a result of prednisolone-induced changes in gene expression. Previous work 33 has identified genes whose expression is altered in 3T3-L cells exposed to prednisolone, including some which form part of the mevalonate pathway. The expression of the genes encoding HMGCR, geranylgeranyl transferase I and II (GGTI, GGTII), isopentenyl diphosphate isomerase (IDI1), mevalonate decarboxylase (MVD) and farnesyl diphosphate synthase (FDPS) were reported to be decreased in cells exposed to prednisolone. Consequently, the effect of prednisolone, alone and in combination with pitavastatin, on these genes products was assessed in ovarian cancer cells. Neither pitavastatin nor prednisolone when used as single agents notably altered the levels of HMGCR, FDPS, IDI1, MVD, GGTI-β. However, GGTII-β was reduced upon exposure to either pitavastatin or prednisolone as single agents as well as in cells exposed to the combination of these two drugs. The combination of pitavastatin and prednisolone, but not the single agents alone, caused significant reduction in levels of HMGCR and FDPS (Fig. 6).

Discussion
Statins in general are well tolerated when used at recommended doses as anti-hypercholesterolemia agent in clinic. However, to cause apoptosis in cancer cells, high doses are likely to be required, which increases the risk of myopathy, creating a challenge for redeployment of statins as chemotherapeutic agent. One strategy to potentially minimize the adverse effects is to identify drugs which synergize with the anti-cancer activity of statins, thereby reducing the dose of statin required. Our screen of approved drugs to discover those which potentiate the activity of pitavastatin identified prednisolone. The synergy between pitavastatin and prednisolone was confirmed in several assays using a panel of ovarian cancer cell and led to the decreased expression of mevalonate pathway genes, providing a potential mechanistic explanation for the synergy. Ovcar-4 and Cov-362 cell lines were exposed to serial dilution of pitavastatin in combination with prednisolone (70 µM) for 72 and 120 hours, respectively. The data is presented as a fraction of the top of the curve calculated by curve fitting (mean ± S.D., n = 3). "C" represents the control cells exposed to solvent alone. (2019) 9:9632 | https://doi.org/10.1038/s41598-019-46102-1 www.nature.com/scientificreports www.nature.com/scientificreports/ The synergy between the pitavastatin and prednisolone combination identified in the screen was verified in several assays. This included cell growth assays in monolayers and in 3D cell culture. The increase in apparent potency of pitavastatin in the presence of prednisolone, at a concentration at which itself has minimal effect, provides unequivocal evidence of synergy between the two drugs. Cell death was mediated, at least in part through, induction of apoptosis. A synergistic increase in apoptosis was observed in three separate apoptosis assays (Annexin V labelling, caspase 3/7 activity and PARP cleavage). These observations provides robust evidence that pitavastatin and prednisolone can interact synergistically.
We have previously shown that pitavastatin causes cell death through inhibition of the mevalonate pathway even when pitavastatin is used at relatively high concentration 9,17 . Inhibition of the mevalonate pathway causes disruption of several GTPases which are involved in cell signalling, regulating cell cycle progression and cell survival 34 . Consequently, statins induce apoptosis including an increase in release of mitochondrial cytochrome C to cytosol, and activation of caspases 3, 8 and 9 35,36 . In this study we also found that that the effects of the combination of prednisolone and pitavastatin was also dependant on inhibition of the mevalonate pathway, because geranylgeraniol and mevalonate both reduced the activity of the drug combination. Reminiscent of the effects of statins alone 12,[30][31][32] , farnesol failed to block the activity of the combination against the cancer cells. This suggests that the effects of the drug combination are primarily mediated through inhibition HMGCR and the consequent inhibition of geranylgeranylation. This is also consistent with our previous observation that inhibition of both GGTI and GGTII potentiates the activity of pitavastatin as a single agent 17 .
The foregoing discussion, particularly the effects of geranylgeraniol, strongly argues that the effect of the combination of pitavastatin and prednisolone depends upon its effects on the mevalonate pathway. We have also previously shown that that dual inhibition of the mevalonate pathway, using bisphosphonates to inhibit FDPS, is also synergistic with pitavastatin 17 . To explore the mechanism by which prednisolone was synergistic with pitavastatin in more detail, levels of mevalonate pathway enzymes were investigated by immunoblotting. An earlier study found that prednisolone altered the expression of genes encoding several mevalonate pathway enzymes including HMGCR, GGTI, GGTII, IDI1, MVD and FDPS 33 . Inhibiting the mevalonate pathway by two separate mechanisms provides a potential rationale to explain the synergy we observed between pitavastatin and prednisolone. We found that the prednisolone-pitavastatin combination cause significant reduction in level of HMGCR and FDPS enzymes. It was striking that the combination, but not the single agents, affected the level of these enzymes, consistent with this contributing to the synergy observed between the drugs. A reduction in GGTII-β was observed following exposure to either of the drugs as a single agents, as well as in the combination.  (Cov-362). The relative viabilities were then measured using CellTiter-Glo assay to measure relative ATP and expressed as fraction of that measured in control samples treated with solvent (mean ± SD; n = 3). The observed effect of the drug combination were compared to the effect expected if the drugs had additive effects which was calculated using the Bliss independence criterion (shown with a line for each drug combination) from the measured effect of the individual drugs in each individual experiment. The results were significantly different from the Bliss expected effect where shown (*P < 0.05; paired t-test).  www.nature.com/scientificreports www.nature.com/scientificreports/ Taken together, this strongly suggests that synergy between prednisolone and pitavastatin may result from inhibiting multiple points on the mevalonate pathway. The effect of related steroids on mevalonate pathway enzymes has also been reported by others. Investigation of the short term effects of dexamethasone in rat hepatocytes revealed a reduction in cholesterol synthesis 37 . Dexamethasone also causes down regulation of HMGCR and FTase enzymes activity in rat AR 4-2J cells 38 . Specifically, the authors found that there is significant reduction in FT-α subunit upon treatment of the cells with dexamethasone for 48 hours. In contrast, the β-subunit of the enzyme was either unchanged or slightly reduced. However, it was claimed that even a 50% reduction of FT activity is not sufficient to prevent Ras isoprenylation and Ras protein were even found to accumulate during dexamethasone treatment. Therefore, it is plausible that a relatively small amounts of an active prenyl transferase is sufficient to maintain prenylation process and support cell survival. We have also previously shown that inhibition of geranylgeranyl transferase I and II simultaneously is necessary to synergize with pitavastatin 17 , suggesting that one prenyltransferase may compensate for the reduced activity of one of the other enzymes. These observations suggest that for efficacy in cancer, robust inhibition of the mevalonate pathway is required and may explain why relatively high concentrations of pitavastatin are required to induce apoptosis. In turn, this may also further explain why combinatorial inhibition of the mevalonate pathway potentiates cell death. Although we have shown the combination induces apoptosis, it is also plausible that the combination affects the autophagy pathway which we 12 and others 39 have already shown to be affected by statins.
The most straightforward explanation for the changes in the abundance of mevalonate pathway enzymes in cells exposed to prednisolone is that their expression is controlled by the glucocorticoid receptor, a transcription factor to which prednisolone binds. However, the mevalonate pathway is subject to a complex set of regulatory mechanisms which may provide an alternative explanation for the activity of prednisolone. Mevalonate pathway enzymes, particularly HMGCR, are regulated by sterol and non-sterol products of the pathway 40 . The HMGCR enzyme itself is regulated at several levels including regulation of its catalytic activity, its rate of degradation and its rate of synthesis 41 . In particular, sterols and oxysterols, which are product of the mevalonate pathway, play a role in feedback regulation of the mevalonate pathway. Sterols and oxysterols inhibit transcription of HMGCR and other mevalonate pathway genes. They bind to the regulatory proteins SCAP and Insig and prevent the translocation of SREBP to the Golgi complex where it otherwise undergoes activation by proteolytic cleavage to allow transcription of mevalonate pathway genes 42,43 . Oxysterols can also directly affects the activity of HMGCR, squalene monooxygenase, FDPS and several enzymes in cholesterol biosynthetic pathway 44 . Oxysterols also accelerate the degradation of HMGCR through sterol-sensing domain in a fashion that depends on the mevalonate pathway regulator Insig 40 . Therefore, it is reasonable to speculate that prednisolone, which also has a sterol ring structure, may mimic sterols and oxysterols and binds to some of these sterol regulatory binding sites. In this manner, prednisolone may reduce the levels of HMGCR and FDPS enzymes either by decreasing transcription or increasing the degradation of the enzymes, or a combination of both mechanisms. In this scenario, adding prednisolone to pitavastatin may prevent reactivation of the mevalonate pathway which would otherwise occur as a result of reduced cholesterol synthesis following inhibition of HMGCR by pitavastatin. In other words, maintenance of feedback inhibition by prednisolone provides one explanation for the reduction in mevalonate pathway enzymes observed in this study. Further work will be necessary to uncover the detailed basis of the regulation of the mevalonate pathway by prednisolone.
Our previous observations 9, 10,12,17 suggest that pitavastatin warrants clinical evaluation in ovarian cancer. The current work suggests that it may be appropriate to evaluate the combination of prednisolone and pitavastatin in clinical trials. The concentration of prednisolone we have used in these studies, although relatively high, is comparable to those clinically achievable (C max ) using a relatively high dose of prednisolone 45 . We do not consider, however, that prednisolone warrants exploration as a single agent in ovarian cancer. Although steroids can induce Figure 6. Effect of prednisolone and pitavastatin on mevalonate pathway enzymes. Ovcar-4 and Cov-362 cell line were exposed for 48 hours or 72 hours, respectively, to prednisolone (70 μM) and or pitavastatin (10 μM). The level of mevalonate enzymes was detected by immunoblotting for HMGCR, GGTI-β, GGTII-β, FDPS, MVD, IDI1 and GAPDH, (n = 3).
www.nature.com/scientificreports www.nature.com/scientificreports/ apoptosis in lymphoid cells 46 steroids as monotherapy show only limited activity in breast and prostate cancers but not in other cancer types 47,48 . In agreement, we found a very limited effect of prednisolone as a single agent on ovarian cancer cell lines. A further possibility is to consider more complex combinations. We have previously shown synergy between pitavastatin and bisphosphonates 17 and other workers have also reported activity of bisphosphonates as single agents against ovarian cancer xenografts in mice 39 . Thus, it may be worthwhile considering the clinical use of a combination of pitavastatin with both a bisphosphonate and with prednisolone. All these drugs are approved for clinical use so there is no regulatory barrier preventing this in principle. It is also worth considering whether statins should be combined with chemotherapy. Although some workers have reported mild synergy between lovastatin and either carboplatin or paclitaxel 39 , we previously mostly observed additivity 12 . Furthermore, we observed profound antagonism if cells were exposed to simvastatin prior to carboplatin 12 , possibly reflecting cell cycle arrest reducing the activity of the chemotherapeutic agents. Thus, we currently do not favour combining pitavastatin with a chemotherapeutic agent.
In conclusion, drug repositioning provides a great opportunity to find new indications for existing drugs. The anti-cancer activity of pitavastatin is potentiated significantly by prednisolone by augmenting inhibition of the mevalonate pathway. Clinical trials of prednisolone with pitavastatin in patients with ovarian cancer may be warranted.