NRF2 at the Interface of Oxidative and Endoplasmic Reticulum Stress Pathways Is a Critical Integrator of Bortezomib Response in Mantle Cell Lymphoma in Vitro and In Vivo.

Abstract 727

The proteasome inhibitor bortezomib (BZM) is effective as single-agent in relapsed and refractory mantle cell lymphoma (MCL) but more than half of patients remain insensitive to BZM. Suggested mechanisms of action include activation of Noxa, p53, oxidative, and endoplasmic reticulum (ER) stress. To define mechanisms relevant for BZM-induced cytotoxicity we pursued two approaches: first, we characterized gene expression changes in 10 MCL cell lines exposed for 24h to 10nM BZM, a concentration that kills >50% of sensitive, but <20% of resistant cells. Secondly, we analyzed gene expression changes in tumor cells of patients with leukemic MCL undergoing BZM treatment in a clinical trial. RNA was extracted from MCL cell lines at early (1h and 3h), intermediate (6h) and late (24h) time points. Virtually no changes in gene expression were detectable at 1 and 3h of drug exposure and only about 100 genes changed by 6h. After 24h of treatment 524 genes were significantly changed in sensitive and 271 genes in resistant cell lines respectively. The delayed onset of gene expression changes is consistent with the reversibility of BZM toxicity for up to 8 hours. Using Ingenuity pathway analysis (IPA) and gene set enrichment analysis (GSEA) we identified two dominant responses induced by BZM: 1) an oxidative stress response mediated by NRF2 and related transcription factors, and 2) an ER stress/ubiquitin proteasome response (FDR by GSEA <0.1). Both responses were primarily apparent in sensitive cell lines. A set of 20 experimentally validated NRF2 target genes was used as a core NRF2 signature: this signature was increased 15-fold on average in sensitive cell lines but only 2-fold in resistant cell lines (P=0.006). Similarly, an XBP1 and ATF6 signature, reflecting activation of the ER stress response, was stronger induced in sensitive than in resistant cell lines (average 1.9-fold vs 1.3-fold; P=0.003). Activation of these stress pathways upon BZM treatment was confirmed by demonstrating accumulation of nuclear NRF2 in sensitive Jeko1 but not in resistant Mino cells. Also markers of ER stress such as phosphorylation of ER resident nuclease Ire1 that splices the transcription factor XBP1 and activation of ATF3, ATF4, and CHOP downstream of PERK were readily detected in Jeko1 but not in Mino cells. Finally, Noxa, the BH3-only protein primarily responsible for BZM-induced apoptosis, was only induced in sensitive Jeko1 cells. We next analyzed the effect of BZM on purified tumor cells from five patients with leukemic MCL treated with BZM (1.5mg/m², day 1, 4, 8 and 11). Two patients showed a >50% reduction in circulating tumor cells after 2 injections of drug (day 8) and >75% reduction after 4 injections (day 2, sensitive), while in three patients there was no change or an intermediate response (resistant). Western blotting demonstrated Noxa up-regulation in circulating tumor cells of sensitive but not resistant samples. This is consistent with the demonstrated importance of Noxa for induction of apoptosis in response to BZM in cell line studies. Next we performed gene expression profiling immediately before, at 6h, and 24h after the first and 24h after the second dose of BZM. Using IPA and GSEA up-regulation of the ubiquitin/proteasome pathway and the NRF2-mediated oxidative stress response was again prominent, but an ER stress response was less apparent. XBP1 splicing was not detected in tumor cells from sensitive samples indicating that an ER stress response was not fully activated by BZM in vivo. Consistent with in vitro data the NRF2 signature was induced 2.3-fold on average in sensitive but not in resistant samples (P<0.05). Intriguingly, baseline expression of the NRF2 signature genes was significantly higher in resistant than in sensitive cells (P=0.0007). In summary, we identify NRF2 as critical integrator of different stress pathways in response to BZM in MCL. Thus, rapid induction of NRF2 target genes might be a useful biomarker of BZM-induced cellular stress and predict clinical response. Our data suggest a more complex function of NRF2 than previously appreciated. NRF2-regulated genes serve primarily homeostatic roles and enable cells to deal with oxidative and xenobiotic insults; a function that may come to play in BZM resistant cells with higher baseline expression of NRF2 target genes. On the other hand, our data suggest a possible pro-apoptotic role of acute induction of high levels of NRF2 that is currently under investigation.