Mechanisms of Acquired Drug Resistance to the Isoform Selective HDAC6 Inhibitor Ricolinostat Reveals Markedly Upregulated Elements of the BTK Pathway Revealing Rational Drug : Drug Combinations

HDAC6 is cytoplasmic with non-histone substrates and binds poly-ubiquinated misfolded proteins facilitating transport to the aggresome for proteasome-independent degradation. The aggresome is a key outlet for the unfolded protein response (UPR), a quality control mechanism that promotes survival during endoplasmic reticulum stress and signals CHOP (C/EBP-homologous protein) mediated apoptosis when protein homeostasis cannot be reestablished. To better understand the discrete function of HDAC6 and its role in lymphoma, we developed lymphoma cell lines resistant to the selective HDAC6 inhibitor ricolinostat (Acetylon Pharmaceuticals, Inc). Ricolinostat has demonstrated activity in preclinical models of multiple myeloma and lymphoma and is currently being studied in the clinical setting for both of these diseases (NCT02091063, NCT01323751, NCT01997840, NCT01583283). 

The diffuse large B-cell lymphoma cell line OCI-Ly10 was exposed to increasing concentrations of ricolinostat over 1 yr.  Systematic incremental increase in drug exposure led to the development of distinct cell lines with IC50 values 10-20 fold greater than that for parental lines (P).  The resistant R10-OCI-LY10 (R10) achieved an IC50 of 10 uM and the R20-OCI-LY10 (R20) achieved an IC50 of 20 uM compared to P’s IC50 of 0.9 uM at 48 hours.  Resistance was maintained after repeated passages for ≥1 month and was not overcome by inhibition of efflux pumps as determined with verapamil co-exposure. Compared to P, R10 and R20 did not depolarize the mitochondrial membrane following treatment with ricolinostat 2.5 µM. However, there was no difference in the BH3 profiles of the R- and P-cell lines.  Interestingly both R10 and R20 were also resistant to vorinostat which strongly inhibits HDAC6 (R20 IC50=7 uM vs P IC50=2 uM), and sensitive to romidepsin which is known to predominantly inhibit HDAC1, 2 and 3 with minimal activity against HDAC6 (R20 IC50=5.2 uM vs P IC50=4.5 uM).  In addition, treatment of R20 with bortezomib demonstrated intermediate activity underscoring the idea that although the drug targets are vastly different, both drug mechanisms converge on the processing of misfolded proteins (R20 IC50=8 nM vs P IC50=3.3 nM).  Accordingly, R10 and R20 displayed down-regulation of GRP78, the master regulator of the UPR, with concomitant up-regulation of both the PERK and IRE-1 pathways leading to up-regulation of p-eIF2a, ATF4, MAPK10 and XBP-1.  Additionally, AKT was more highly phosphorylated in R10 and R20 compared to P.

Gene expression profiling was performed on R10 and R20 independently and compared to P. Results were analyzed by GSEA and gene cluster analysis.  There were 1363 genes in common from R10 and R20 2-log fold up-regulated as compared to P and 1825 genes in common from R10 and R20 2-log fold down-regulated as compared to P (p < 0.05).  In addition to changes in the UPR, differentially expressed genes in R10 and R20 included up-regulation of MAPK10, HELIOS, HDAC9 and FYN, as well as down-regulation of SH3BP5 and LCK.  The change in expression was confirmed by PCR and western blot analysis.  Given the up-regulation of FYN, a tyrosine kinase in the B-cell receptor pathway, and down-regulation of SH3BP5 a negative regulator of the Bruton’s Tyrosine Kinase, R10 and R20 cells were treated with ibrutinib 0.1 uM which was able to overcome resistance with 70% viability in R10 compared to 64% in P at 72 hrs.  P cells treated with ibrutinib in combination with ricolinostat demonstrated synergy with RRR=0.46 (where RRR<1 connotes synergy) after 96 hour exposure to ibrutinib and ricolinostat. The combination led to modulation of HELIOS, p-BTK, p-AKT, SH3BP5, and FYN.  These findings were confirmed in 2 additional ABC-DLBCL and MCL cell lines as well as in 2 primary patient samples of lymphoma: 1 CLL and 1 LPL with RRR of 0.54, and 0.52 respectively.  In vivo confirmation of synergy is now underway in mouse models of Ly10 and R20-Ly10.

The development of these ricolinostat drug resistant cell lines has proven to be a powerful tool to better understand the mechanistic role of HDAC6 in lymphoma and may potentially lead to the identification of biomarkers of response or resistance.  In addition, we believe this strategy may offer a new way to identify rational and synergistic drug : drug combinations.   Translation of these findings to the clinic are now under consideration.