Mechanisms of non-classical multidrug resistance in cancer

Division Of Basic Sciences - Nci

Investigators

Linked publications & trials

Abstract

Three major approaches have been taken to define non-classical multidrug resistance in cancer. In the first, we isolate KB cells and ovarian cancer cells resistant to increasing levels of cisplatin (CP-r) and demonstrate multidrug resistance to many other cytotoxic agents. In some cases, this cross-resistance pattern is due to reduced uptake of each of these agents because their receptors have been relocalized from the cell surface into the cytoplasm of the cell. We have undertaken a complete genomic analysis using RNA-seq, ATAC-seq and Pro-seq technologies to define the alterations in gene expression that accompany the development of drug resistance in cisplatin-selected cell lines and one cataloguing alterations in cisplatin-resistant cells that contribute to drug resistance. Comparing gene expression in cisplatin-sensitive 1A9 ovarian cancer cells, cisplatin-resistant 1A9CP80 cells and partially revertant 1A9CP80R cells, we noted increased expression of TPPP3 (tubulin polymerization promoting protein 3) in the resistant cells, with lower levels in the revertant cells and no expression observed in the parental line. Additionally, we noted that cisplatin treatment destabilizes microtubule ends and reduces microtubule length and hypothesized that TPPP3 might mitigate these effects. Interestingly, the ability of TPPP3 to counteract the effects of cisplatin treatment were most effective in tubulin purified from 1A9CP80 cells and least effective in 1A9 cells and appeared to correlate with changes in expression of tubulin isoforms in the cell lines. Deletion of TPPP3 via CRISPR knockout partially resensitized the 1A9CP80 cells to cisplatin. High expression of TPPP3 in tumors from patients treated with cisplatin correlated with worse survival probability, suggesting a possible clinical role for this protein. In addition, cells exposed to cisplatin (which we have shown destabilizes microtubules) are more resistant to paclitaxel, an antimicrobule drug that stabilizes microtubules. These results have implications both for the neurotoxicity of cisplatin, and for strategies that employ combinations of cisplatin and paclitaxel to treat ovarian cancer and other cancers. To understand more about non-classical mechanisms of multidrug resistance in cancer, we are undertaking CRISPR screens in cells exposed to various drugs including cisplatin and oxaliplatin. These screens involve using gRNAs in combination with CRISPR-cas constructs that can activate, inhibit, or knock out target genes. Cells exposed to platinum compounds or other drugs undergo cell death and surviving cells overexpress gRNAs which turn on genes which can independently confer resistance, or underexpress genes whose expression is needed for sensitivity to cisplatin. We are identifying genes whose over- or under-expression affects drug resistance with the goal of defining clinically relevant molecular changes. Recent studies using CRISPR screens to determine the basis of oxaliplatin resistance in colon cancer cells has identified amino acid transporters as playing an important role in sensitivity to this drug. Histone deacetylase inhibitors (HDIs) are used clinically to treat cutaneous and peripheral T-cell lymphomas, diseases for which 3 HDIs have been FDA approved as single-agent therapies. In the case of solid tumors, the HDIs have not been effective, suggesting intrinsic resistance mechanisms to these drugs. We found that synergistic killing can be achieved with HDIs and inhibitors of the MAPK and PI3K signaling pathways in cells that harbor Ras mutations. We also found that a dual ERK/PI3K inhibitor could take the place of separate MAPK and PI3K inhibitors when combined with an HDI. Further studies have shown that the dual BRD4/PI3K inhibitor SF2523 is synergistically toxic to Ras mutant cells when combined with an HDI. In collaboration with Dr. Mari Yohe, we demonstrated that SF2523 alone is particularly effective in childhood rhabdomyosarcoma cell line models and its efficacy can be increased by the addition of the HDI romidepsin. The Center for Advanced Preclinical Research (CAPR) has agreed to examine the efficacy of the SF2523/romidepsin combination in patient-derived xenograft models of rhabdomyosarcoma. Resistance to HDI's such as romidepsin can occur in cultured cells owing to overexpression of P-glycoprotein, but in clinical cancers resistance does not appear to be due to this mechanism. To identify non-P-gp mechanisms of resistance, we selected MCF-7 breast cancer cells with romidepsin and verapamil to yield the MCF-7 DpVp300 line which is about 200-fold more resistant to romidepsin than the parental cells. The cells are uniquely resistant to romidepsin, as the resistant line was only 3- to 5-fold more resistant to other HDIs such as vorinostat, belinostat, or panobinostat. RNA Seq analysis comparing the parental and resistant line identified the gene METTL7A, which codes for a poorly-described methyltransferase, as a potential resistance mechanism. METTL7B, a paralog of METTL7A, was recently determined to be an alkly thiol methyltransferase that is capable of methylating thiol groups. As the active form of romidepsin has a thiol in its active form, and as methylation of the thiol group would prevent coordination of the molecule with zinc in the HDAC binding pocket, we hypothesized that METTL7A might be able to inactivate romidepsin or other HDIs with a thiol as the zinc-binding group. In support of this hypothesis, knockout of METTL7A from DpVp300 cells resensitized the cells to romidepsin as well as other thiol-based HDIs such as KD5170 and largazole. Interestingly, HEK293 cells transfected with METTL7A were resistant to all of the thiol-based HDIs, but METTL7B overexpression conferred less resistance to largazole and KD5170 than METTL7A and no resistance to romidepsin. METTL7A and METTL7B thus appear to be methyltransferases with somewhat different specificity that confer resistance to thiol-based HDIs by inactivating these drugs. To determine if animal models could be used to elucidate the normal function of these methyltransferases, we examined homologs of METTL7A from different species. We found that METTL7A is conserved across vertebrates while METTL7B is not. To determine if the ability of METTL7A to methylate thiols is conserved, we transfected HEK293 cells to express mouse, rat, chicken, or zebrafish METTL7A. We found that expression of any of the METTL7A isoforms could confer resistance to all of the thiol-containing HDACis tested, suggesting that the function of METTL7A is conserved across species. These results have led us to create a transgenic zebrafish where METTL7A is deleted. We will characterize the knockout fish in hopes of finding a physiological role for METTL7A. Validation of these results, indicating that MDR is complex and multifactorial in clinical cancers, will require the development of reliable in vitro culture models. Towards this goal, we have developed a bioreactor that mimics capillary delivery (through silicon hydrogels and the polymer PTMS) of oxygen to cells grown in 3D suspension. We have demonstrated physiological oxygen gradients and altered growth of cancer cells more closely approximating in vivo phenotypes. Evidence that oxygen gradients substantially change gene expression patterns has been obtained by detailed RNAseq analysis. Delivery of physiological concentrations of 3% oxygen directly to cells via artificial capillaries mimics the gene expression patterns of 20% oxygen delivered via diffusion. The bioreactor can be scaled up for growth of multiple cultures of primary cancer cells or cultured cancer cells to determine whether growth conditions and mode of oxygen delivery play a primary role in affecting patterns of drug resistance.