Microenvironmental support protects malignant cells from apoptosis and from the effects of therapy. We hypothesized that in chronic lymphocytic leukemia (CLL) cells, the non-malignant bystander cells induce a transcriptional signature that prevents apoptosis. We aimed to use this transcriptional signature to identify compounds that target this protective microenvironmental effect. If these compounds are effective against protected CLL cells, we assume that understanding the underlying molecular pathways will open novel therapeutic options for therapy-resistant CLL.
In primary CD19+ CLL cells cultured without support after isolation from peripheral blood, apoptosis can already be detected after 8 hrs. These cells can however be rescued by coculture with either human or murine bone-marrow derived stromal cell lines HS-5 and M210B4. In order to identify clinically relevant substances that counteract this pro-survival microenvironmental support, we determined the transcriptional signature that is induced in primary CLL cells by protective feeder cells at 8 timepoints before apoptosis could be detected and compared it to the connectivity map database of transcriptomes induced by 1300 compounds (cMAP). We identified 7 compounds with a transcriptome signature that is negatively correlated to the protective signature induced in CLL cells by in-vitro coculture. Of note, 5 of these substances were more potent than fludararabine to induce apoptosis in protected primary CLL cells. In contrast, none of the 10 control compounds with non-correlated signatures had an effect after 48h of coculture. Most potent were the isoquinoline alkaloids emetine and cephaeline and the cardiac glycoside ouabain that displayed IC50s of 34nM, 190nM and 287nM, while a concentration of 5.1µM of fludarabine was required for comparable induction of apoptosis.
To test the efficacy of emetine against CLL in vivo, TCL1-induced murine CLL-like tumors were transplanted into immunocompetent syngeneic mice and treated with 3 high doses of emetine (24mg/kg) after CD5+CD19+ CLL cells reached 40% of leukocytes (CD45+ cells). Treated animals displayed a significant reduction in body weight compared to untreated (26.0g vs 29.1g, p<0.0001). However, treatment of leukemic mice with emetine reduced total lymphocyte counts in the peripheral blood compared to untreated mice (19.3 vs 48.1x10^3/mm^3, p<0.0001) and also the proportion of CLL cells in the bone marrow (4.1% vs 13.0%, p=0.0041). Most importantly, treatment increased survival of mice transplanted with syngeneic murine CLL tumors (median not reached in the treatment arm vs 15 days for untreated mice, n=7 and n=9, p=0.0012, censored for different start of treatment, panel A). This shows that emetine effectively targets murine TCL1-driven CLL even if they are microenvironmentally protected in vivo.
Next we aimed to understand the molecular mode of action of therapeutic targeting of the microenvironmental support. Comparing transcriptomes of primary CLL cells cocultured with and without murine or human stromal cell support, 6 of the 10 most deregulated pathways involved redox homeostasis. Downregulation of the Von-Hippel-Lindau (VHL) gene led to an increase in hypoxia-inducible factor HIF1a activity (panel B). In fact, hypoxic in-vitro incubation (3% O2) of primary CLL cells significantly prevented apoptosis between 24h and 72h (n=8, p<0.05). Functionally we found in cell lines stably expressing redox reporter fusion proteins that emetine and ouabain impair recovery from mitochondrial but not cytosolic oxidative stress (Figure, panel B). In fact, the observed effect was stronger than for 13 other redox-active compounds. In addition we could show that emetine and ouabain overcome the anti-apoptotic microenvironmental downregulation of VHL in CLL cells by direct reduction of HIF1a protein and also HIF1a activity, shedding light on their molecular mode of action (panel B).
In conclusion, we identified modulation of redox homeostasis to play the most important role in microenvironmental protection of CLL cells by bone-marrow stromal cells in-vitro. In addition we could show that this opens a new target of the malignant CLL cells that reside in the stromal niche protecting them from chemotherapy. Even though unexpected for a chronic leukemia, hypoxia-regulated pathways seem to be a novel therapeutic option for resistant CLL.
No relevant conflicts of interest to declare.
Asterisk with author names denotes non-ASH members.