Gossypol overcomes stroma-mediated resistance to the BCL2 inhibitor ABT-737 in chronic lymphocytic leukemia cells ex vivo

Chronic lymphocytic leukemia (CLL) is characterized by the deregulated accumulation and persistence of B lymphocytes in the blood. Although the exact causes of CLL are unknown, the evasion of apoptosis through aberrant expression of BCL2-family proteins is a common feature.¹ A class of compounds, termed BH3 mimetics, has been developed to directly inhibit BCL2 proteins and selectively kill tumor cells. To date, the most successful of these compounds are the BCL2/BCLXL inhibitors ABT-737² and ABT-263 (navitoclax),³ as well as the BCL2-specific inhibitor ABT-199.⁴ Results from early clinical trials with navitoclax have demonstrated single-agent efficacy in patients with relapsed or refractory CLL.⁵ However, there was heterogeneity in response rates between patients, and dose-limiting toxicities including thrombocytopenia and neutropenia which prevented further dose-escalation. In addition, CLL cells residing within various microenvironments (e.g. lymph nodes and bone marrow) are resistant to BCL2 inhibitors. This resistance results from the upregulation of additional BCL2-proteins, such as BCLXL, MCL1 and BFL1, the latter two of which are not inhibited by navitoclax, and therefore protect the leukemia cells from apoptosis.⁶ Additional drugs are needed to enhance the efficacy of navitoclax. Here, we demonstrate that gossypol overcomes stroma-mediated resistance to ABT-737 without enhancing the sensitivity of normal lymphocytes and platelets.

The BH3-only protein, NOXA, is a potent inhibitor of MCL1 and BFL1,⁷ but has recently been recognized to inhibit BCLXL with lower affinity.⁸ Therefore, compounds which induce NOXA may inhibit MCL1, BFL1 and BCLXL, thus overcoming resistance to navitoclax. We previously reported that six putative BH3 mimetics do not directly inhibit BCL2 in cells, but instead activate the integrated stress response and induce NOXA.⁹ Of these six compounds, gossypol has advanced into clinical trials in a racemically purified form (AT-101).¹⁰ We hypothesized that gossypol, through induction of NOXA, would sensitize CLL cells to ABT-737.

CLL cells from consented patients were incubated ex vivo with up to 10 μM ABT-737, which is comparable to the peak plasma concentration of navitoclax in a Phase I trial.⁵ Apoptosis, as assessed by chromatin condensation, (or caspase cleavage of PARP; data not shown) was induced within 6 h as previously noted.⁹ The CLL samples were highly sensitive to ABT-737 as a single agent, with most cells undergoing apoptosis between 10 and 100 nM (Figure 1A). Gossypol alone (5 – 20 μM) induced NOXA but had minimal impact on apoptosis within this time frame. Importantly, gossypol sensitized the cells to ABT-737, with most cells now undergoing apoptosis between 1 and 10 nM.

Figure 1.

Gossypol overcomes stroma-mediated resistance to ABT-737 in CLL cells treated ex vivo at concentrations minimally toxic to normal lymphocytes and platelets. (a) Gossypol induces NOXA and sensitizes CLL cells to ABT-737 ex vivo. CLL cells from consenting patients were isolated using a ficol extraction protocol and incubated with ABT-737 and gossypol as indicated for 6 h ex vivo. Cells were then incubated with Hoechst 33342 and scored for condensed chromatin via microscopy. Survival is expressed as the percentage of cells which do not exhibit condensed chromatin. Error bars represent one S.E.M. (n=5–11). NOXA induction following ex vivo incubation with gossypol in a representative CLL patient was detected by western blotting, with actin as a loading control (inset). (b) Co-culture of CLL cells with CD154-expressing stroma upregulates BCLXL and BFL1. CLL cells isolated from five patients were co-cultured on CD154-expressing stromal cells for 24 h¹² CLL cells were then gently removed from stroma and passed through an 8 micron filter to remove any residual stromal cells. Purified CLL cells were lysed (urea lysis buffer) and analyzed for MCL1, BCLXL, BFL1, BCL2 and actin (loading control) via western blotting. (c) Gossypol resensitizes CLL cells to ABT-737 in a stroma co-culture model. Isolated CLL cells were co-cultured on CD154-expressing stroma for 24 h, and then incubated for 6 h as indicated and scored for % survival (n=5–11). (d) Normal lymphocytes are resistant to ABT-737 and gossypol combinations. Lymphocytes were isolated from healthy individuals via ficol extraction, treated for 6 h as indicated and scored for % survival (n=2). (e) Gossypol does not increase platelet death induced by ABT-737. Platelets from a healthy donor were isolated via centrifugation and treated for 6 h as indicated. Caspase-dependent cleavage of gelsolin was used as a marker of apoptosis in platelets. DMSO (0.1%) was used as a vehicle control.

Co-culture of CLL cells on stromal cells can mimic the protective microenvironment seen in patients, and promote resistance to ABT-737 and navitoclax. Different stroma have variously been reported to increase the expression of MCL1, BFL1 and BCLXL.^{6, 11} We hypothesized that gossypol might overcome stroma-mediated resistance to ABT-737 by inducing NOXA. We used a co-culture system whereby CLL cells were incubated with fibroblasts expressing CD154 (CD40L).¹² Following 24 h co-culture, BCLXL was markedly increased in all CLL samples; there was variable increase in BFL1 and no significant increase in MCL1 (Figure 1B). BCL2 expression was unchanged. In agreement with a previous report,⁶ we observed at least 100-fold resistance, with most cells requiring 1–10 μM ABT-737 to induce apoptosis (Figure 1C). Importantly, we found that gossypol resensitized the CLL cells to ABT-737 in a concentration-dependent manner, with apoptosis again occurring at 10 – 100 nM ABT-737.

To determine whether this sensitization was unique to CLL cells, we incubated normal lymphocytes with ABT-737 and gossypol ex vivo. Normal lymphocytes were significantly resistant to ABT-737 alone and in combination with gossypol (Figure 1D), with even less apoptosis occurring than in the CLL cells incubated on stroma. Since thrombocytopenia was the major dose-limiting toxicity of navitoclax in clinical trials,⁵ we determined whether gossypol increases ABT-737-mediated apoptosis in platelets. Caspase-mediated cleavage of gelsolin, a marker for apoptosis, was only observed when platelets were incubated with 10 μM ABT-737 (Figure 1E). Importantly, the addition of gossypol did not increase gelsolin cleavage indicating that it did not sensitize the platelets to ABT-737.

It is interesting that CLL cells incubated on stroma are killed by 10 μM ABT-737 as a single agent, which is the same concentration that induces apoptosis in platelets. As platelets are dependent on BCLXL, and BCLXL appears to be the predominantly induced BCL2 protein in CLL cells on stroma, this suggests that this resistance in CLL cells is dependent on BCLXL. This would be consistent with recent data demonstrating that ABT-737 is less effective at inhibiting BCLXL than BCL2.¹³ In addition, the fact that gossypol does not kill platelets suggests that it is not acting as a direct BCLXL inhibitor, but may instead require NOXA-induction for BCLXL inhibition. A previous report that gossypol inhibits BCLXL may therefore be explained by the induction of NOXA, which was simultaneously observed but not considered the mechanism of BCLXL inhibition.¹⁴

It has been reported the NF-kB inhibitors can prevent the development of resistance to ABT-737 mediated by CD154.⁶ However, this approach would fail to kill CLL cells in the lymph node that are already resistant. Other drugs, such as dasatinib and fludarabine, have been shown to resensitize CLL cells to ABT-737 in this stroma co-culture model, but these combinations required 24–48 h and only partially overcame resistance.¹⁵ In contrast, the gossypol plus ABT-737 combination reported here overcame resistance almost completely in 6 h. To the best of our knowledge, the efficacy demonstrated here with this drug combination is far superior to any previously published.

Taken together, these findings identify a novel drug combination which is able to overcome resistance to a BCL2 inhibitor in a clinically relevant CLL model. Importantly, normal lymphocytes and platelets are resistant to combinations of ABT-737 and gossypol, which offers a therapeutic window to selectively kill CLL cells. It is therefore plausible that this combination may kill resistant CLL cells within the stromal niche in vivo, and may improve clinical outcome. Since both gossypol and navitoclax have been tested in humans, this data supports the notion of a combination trial in patients with CLL. Both drugs have demonstrated toxicity when administered on a chronic daily basis,^5,10 but the data presented here suggests acute treatment might be effective and thereby avoid the undesirable toxicities.