Dipyridamole-like compounds potentiate fluvastatin-induced cancer cell death
Computational pharmacogenomic screen identifies drugs that potentiate the anti-breast cancer activity of statins
Statins, a family of FDA-approved cholesterol-lowering drugs that inhibit the rate-limiting enzyme of the mevalonate metabolic pathway, have demonstrated anticancer activity. Evidence shows that dipyridamole potentiates statin-induced cancer cell death by blocking a restorative feedback loop triggered by statin treatment. Leveraging this knowledge, we develop an integrative pharmacogenomics pipeline to identify compounds similar to dipyridamole at the level of drug structure, cell sensitivity and molecular perturbation. To overcome the complex polypharmacology of dipyridamole, we focus our pharmacogenomics pipeline on mevalonate pathway genes, which we name mevalonate drug-network fusion (MVA-DNF). We validate top-ranked compounds, nelfinavir and honokiol, and identify that low expression of the canonical epithelial cell marker, E-cadherin, is associated with statin-compound synergy. Analysis of remaining prioritized hits led to the validation of additional compounds, clotrimazole and vemurafenib. Thus, our computational pharmacogenomic approach identifies actionable compounds with pathway-specific activities.
Dipyridamole potentiates statin efficacy to drive tumor cell death by blocking the statin-induced restorative feedback response of the MVA pathway23,25. Statins are readily available, safe, approved and manufactured as inexpensive generic drugs. Our goal was to expand the number of agents that potentiate the pro-apoptotic activity of statins to ultimately better use statins as anticancer agents at clinically achievable doses. Systematic and targeted efforts have been made in the past to identify drugs that potentiate statins’ anticancer effects47,48,49,50,51,52. To this end, we used a computational pharmacogenomics pipeline to enrich for compounds with similar properties to our prototypic compound, dipyridamole, at the level of structure, anti-proliferative activity and MVA pathway gene expression perturbation. We identified 19 potential compounds and evaluated several for their ability to potentiate statin-induced apoptosis by blocking the restorative feedback loop of the MVA pathway. By this approach, we first validated nelfinavir and honokiol from the prioritized hits by combining fluvastatin with each of these agents, whereby the concentration of fluvastatin needed to induce cell death was then lowered to a clinically-achievable range53,54. Analysis of basal RNA and protein expression identified the epithelial cell marker, CDH1 (E-cadherin) as a biomarker of the synergistic response to both fluvastatin-nelfinavir and fluvastatin-honokiol treatment. From the remaining hits, we identified another 2 compounds, clotrimazole and vemurafenib, that potentiate fluvastatin-induced cell death and block SREBP2 activity.
In addition to blocking statin-induced SREBP2 activity, several other activities of dipyridamole have been described, including inhibition of phosphodiesterases55, nucleoside transport56 and glucose uptake57. By restricting the gene perturbation layer of the pharmacogenomics pipeline to MVA pathway genes, our rationale was to bypass these extraneous activities and focus our analysis on identifying drugs whose mechanisms centered on the MVA pathway. This highlights that the computational pharmacogenomics pipeline described here is likely tunable to drug-specific structural features, activities and signaling pathways. Indeed, as the two pharmacogenomic data sets used here continue to increase in size, these additional drugs and genes can be leveraged to further customize the analysis.
The first two statin-sensitizing agents identified using MVA-DNF include nelfinavir and honokiol, which we demonstrate here inhibit statin-induced SREBP2 cleavage and activation similar to dipyridamole23,25. To date, a number of agents have been identified that block SREBP2 activation, including fatostatin, betulin, and xanthohumal, which block ER-Golgi translocation. Additional SREBP2 inhibitors include BF175 and tocotrienols that target SREBP2 transcriptional activity and protein stability, respectively. However, other than nelfinavir, these agents are either under development for clinical application or are only used for research purposes and are not likely to be advanced to patient care. The S2P protease inhibitor nelfinavir (marketed as Viracept) was approved for use in 1997 as an antiviral for the treatment of HIV, and is under evaluation for its utility as an anticancer agent58,59,60,61,62,63. This further reinforces that the combination of statin-nelfinavir is immediately actionable and should be further evaluated at the clinical level without delay. We suggest the fluvastatin-nelfinavir combination is preferred compared to other statins, as distinct cytochrome P450 enzymes are used to process these agents, thereby reducing the potential for adverse drug-drug interactions64.
To our knowledge, honokiol in combination with statins in the context of cancer has not been well investigated. Honokiol is a natural product commonly used in traditional medicine and has a number of reported mechanisms of action. Here we show that honokiol inhibits SREBP2 translocation and induction of gene expression in combination with fluvastatin. As honokiol and its derivatives are presently under investigation, discovering this activity for honokiol can be incorporated into future analyses of structure-activity relationships of this agent.
We also observed this fluvastatin-nelfinavir and fluvastatin-honokiol synergistic response to the combination therapies across multiple subtypes of BC. Previously, we and others had identified the basal-like BC subtype as more sensitive to statins alone and identified a mesenchymal-enriched gene expression profile as highly predictive of statin sensitivity42,65,66. Here, we have expanded the scope of statin treatment to encompass a wider spectrum of BCs when used as combination therapy. Moreover, analyses of gene and protein expression data across a large collection of BC cell lines identified canonical epithelial cell marker CDH1 as predictive of synergy to all three statin-compound (dipyridamole, nelfinavir or honokiol) combinations. We further showed that low CDH1 expression levels served as a biomarker of synergistic response in BC cell lines67. Other groups have independently correlated E-cadherin expression with statin resistance and suggested its use as a biomarker of statin sensitivity67. This suggests further evaluation of CDH1 as a biomarker is warranted in addition to previously published gene expression signatures for statin sensitivity48.
To expand upon the 2D cell culture findings, we screened four 3D primary BC patient-derived tumor organoids to evaluate fluvastatin-nelfinavir activity, alone and in combination. All four patient-derived organoid models were synergistic to fluvastatin-nelfinavir. Synergy was observed at physiologically achievable23,45,53 concentrations of fluvastatin and low nanomolar concentrations of nelfinavir.
To further validate the remaining MVA-DNF compounds, we used two high-throughput screening assays to identify compounds that could first potentiate fluvastatin-induced cell death and second block the SREBP2 mediated feedback response. First, we leveraged a live-cell imaging assay that integrates the results from three independent live-cell dyes to determine cell death. Second, we developed a live-cell imaging assay to quantify SREBP2 cytoplasmic-to-nuclear translocation. Using this approach, we identified two additional SREBP2 inhibitors, clotrimazole and vemurafenib, as statin-sensitizers. Clotrimazole is a topical antifungal agent and vemurafenib is an oral V600E BRAF inhibitor. The mechanism of both these compounds as SREBP2 inhibitors remains unclear and warrants further investigation.
Additional compounds tested in this study include selumetinib, baccatin III and mitoxantrone; the former two were observed to sensitize BC cells to statin-induced apoptosis, but the latter did not. Our data suggest that selumetinib and baccatin III function through a SREBP2-independent mechanism. Drugs that function through alternative mechanisms of statin potentiation identified using our method were anticipated due to the selection criteria in the MVA-DNF pipeline. The identification of such compounds is potentially advantageous as some multiple myeloma and prostate cancer cell lines have been shown to lack statin-induced SREBP2 activity22,23,45. Future investigation to assess the effectiveness of the drug combinations in ex-vivo and murine models of primary tumor growth and metastasis is warranted. Testing of drug combinations using additional models can provide insights that are currently lacking from testing drug combinations in vitro. These include the tumor microenvironment and immune cell interactions.
The data presented here have important clinical implications for statins as anticancer agents. Despite encouraging results from window-of-opportunity clinical trials in BC using statins as a single-agent, only a modest effect was observed with some but not all patients20,21. Accordingly, discovery of therapeutic combinations is necessary to achieve significant clinical impact. Our study provides a strong preclinical rationale to warrant further investigation of the fluvastatin-nelfinavir, fluvastatin-honokiol, fluvastatin-clotrimazole and fluvastatin-vemurafenib combinations, as well as the utility of CDH1 as a biomarker of response. Since nelfinavir and vemurafenib are poised for repurposing, and statins have demonstrated anticancer activity in early-phase clinical trials20,21,53,68,69,70,71,72, clinical studies to further evaluate the therapeutic benefit of these combinations can proceed swiftly. We validated the pharmacogenomic pipeline using breast cancer as a model system, however these statin-compound treatment options may also be effective for additional cancers in which the mevalonate pathway is contributing to disease. The availability of these approved, well-tolerated, oral drugs as well as simple methods for assessing CDH1 expression could enable rapid translation of these findings to improve cancer patient outcomes.