Despite many recent successes in the treatment of cancer, the development of chemoresistance in many of the initially responding patients, and primary resistance in others, remains a major impediment in therapy development. Our studies provide evidence for a novel mechanism underlying drug resistance: Gli1 dependent drug glucuronidation.
While carrying out a Phase II clinical trial of targeting the eukaryotic translation initiation factor eIF4E with ribavirin in M4/M5 subtypes of AML, we observed that all responding patients eventually became clinically and molecularly resistant. To understand the cause of this resistance, we generated ribavirin resistant cell lines. In these models, ribavirin no longer targeted eIF4E activity or impaired growth, and importantly, the ability of ribavirin to bind eIF4E was severely impaired. However, the eIF4E gene was not mutated and its protein levels were not altered. The cell lines could be divided into two groups: type I with a defect in drug uptake and type II with a normal uptake. In type I resistant cells, we observed a substantial reduction in levels of Adenosine Kinase (ADK) an enzyme that catalyzes the rate limiting step in the metabolic activation of ribavirin allowing its retention in the cells.
We used RNA Sequencing to examine the molecular underpinnings of type II resistance. Our data revealed a drastic increase in the levels of Gli1. In stably overexpressing cells, Gli1 was sufficient to produce the same resistance phenotype that we observed for type II cell models, both molecularly and at the level of cell growth. In addition, Gli1 overexpression correlated with the loss of drug-to-target interaction, as observed by our eIF4E immunoprecipitation studies using 3H-Ribavirin, similarly to the resistant cell lines. Conversely, Gli1 knockdown in type II cells or its pharmacological inhibition with the FDA approved Gli1 inhibitor GDC0449/Vismodegib, restored the eIF4E-ribavirin interaction and re-sensitized these cells to ribavirin.
Our subsequent studies revealed a close correlation between Gli1 expression and the protein levels of the UGT1A glucuronosyl transferase enzymes involved in phase II drug metabolism whereby xenobiotics or metabolites are modified by the addition of a sugar, glucuronic acid. Given these findings, we examined whether the loss of the eIF4E interaction in resistant cells was due to the glucuronidation of ribavirin. Using 13C/12C ribavirin and mass spectrometry, we observed glucuronidated forms of ribavirin in resistant cells and cells overexpressing Gli1 but not in parental cells and that ribavirin is glucuronidated on its triazole ring which binds eIF4E. Treatment of cells with the Gli1 inhibitor GDC0449 reduces UGT1A levels, and correlates with reduced levels of ribavirin-glucuronides and the re-emergence of ribavirin-eIF4E complexes.
We further hypothesized that the type II resistant cells could be resistant to other drugs. We observe that our ribavirin resistant cell lines are also resistant to the cornerstone of AML therapy, cytarabine. GDC0449/Vismodegib treatment reverts resistance to cytarabine in these cells. Preliminary studies indicate that these cells are also resistant to azacytidine and cisplatin. This is particularly striking as these cells were never exposed to these compounds. Thus, this could represent a novel form of multi-drug resistance.
To establish the clinical relevance of our findings to patients in our AML ribavirin trial, we examined features of type I and type II resistance. Out of 10 patient samples available for evaluation, all six responding patient specimens showed elevated Gli-1 mRNA levels, up to 26 fold, upon relapse relative to levels during response. For most, the ratio of Gli1 during response relative to at relapse was about 2-4 fold with some patients up to 10 fold. For the two patients examined that did not respond, both had highly elevated Gli-1 levels prior to treatment relative to healthy individuals, and this was not lowered after 28 days of ribavirin treatment. We also noted elevated UGT1A protein levels upon relapse in our patient population. Type I resistance was observed in only two patients whereas Gli1 and UGT1A were dysregulated at relapse in all patients examined.
In summary, we identified a novel form of drug resistance: Gli1 dependent drug glucuronidation. Treatment with Gli1 inhibitors appears to be a promising avenue for overcoming this form of drug resistance.
No relevant conflicts of interest to declare.