Decreased mitochondrial apoptotic priming underlies stroma-mediated treatment resistance in chronic lymphocytic leukemia

Abstract

Stroma induces treatment resistance in chronic lymphocytic leukemia (CLL), possibly because of alterations in the BCL-2 family of proteins, which are key regulators of apoptosis. We previously developed BH3 profiling, a functional assay that assesses mitochondrial depolarization in response to BH3-only peptides, to measure "apoptotic priming," the proximity of a cell to the apoptotic threshold. In the present study, we use BH3 profiling to show that CLL cells from the PB are highly primed. Increased priming is associated with improved clinical response and, unexpectedly, with unmutated IGHV status. Coculturing CLL cells in vitro with stroma decreases priming. Using matched PB, BM, and lymph node compartment samples, we found in vivo that BM-derived CLL cells are the least primed. CLL cells cocultured with stroma were treated with the PI3K δ-isoform inhibitor CAL-101 (GS1101). CAL-101 caused CLL cell de-adhesion, leading to increased CLL cell priming. Stimulation of CLL cells with anti-IgM or CXCL12 caused decreased priming that could be reversed by CAL-101. Our results show that inhibition of stromal interactions leading to displacement of CLL cells into the blood by CAL-101 in vivo may increase CLL cell priming, suggesting a mechanism by which agents inducing lymphocyte redistribution might facilitate improved clinical response when used in combination with other therapies.

Figure 1

CLL cells are highly primed for apoptosis and usually BCL-2 dependent, and increased priming is associated with improved clinical response and unmutated IGHV. (A) PB CLL cells were BH3 profiled by FACS (n = 30) and the level of apoptotic priming for each sample was measured by quantifying the mitochondrial depolarization induced by the BIM BH3 peptide at a 0.03μM final concentration. Patient number, as detailed in supplemental Table 1, is depicted above each corresponding level of priming. (B) Examples of BH3 profiles from 3 individual patients showing pattern of relative dependence on BCL-2, MCL-1, and BCL-XL (see supplemental Table 1 for complete depolarization data for each patient by peptide). (C) Pretreatment samples from treatment-naive patients achieving a partial response (PR) or complete response (CR) by International Workshop on Chronic Lymphocytic Leukemia 2008 criteria are more primed than samples from patients with progressive disease during or within 6 months of completing frontline CLL therapy (P = .0333). (D) BH3 profiling shows that patients with unmutated IGHV status (n = 11) are significantly more primed than patients with mutated IGHV status (n = 12; P = .0074). (E) Percentage of VH homology to germline is positively correlated with level of priming (P = .0020; Spearman r: 0.480).

Figure 2

Primary CLL cells cocultured for 24 hours with mouse or human stroma are resistant to the BH3-mimetic ABT-737. (A) Representative primary FACS data of PB-derived CLL cells cocultured with either CD154 ⁺ or CD154⁻ stroma shows that the CD154⁺ cells significantly maintain CLL cell viability (annexin V and propidium iodide negative) in response to 100nM ABT-737. (B) Aggregate data for CLL cells cocultured ± 100nM ABT-737 with CD154 stroma confirms stromal protection (n = 4), with means depicted as horizontal bars ± SEM; P ≤ .05 between each group. (C) Aggregate data for CLL cells cocultured ± 100nM ABT-737 with or without the human stromal cell line StromaNKTert confirms the strongly protective effect of human stroma (n = 7). (D) Aggregate data from dose-response curves on 5 CLL patient samples cocultured ± ABT-737 with or without StromaNKTert. Both drugs kill consistently in a dose-dependent manner in the absence of stroma; induction of apoptosis by both drugs is significantly inhibited in the presence of stroma.

Figure 3

In vitro, BH3 profiling demonstrates that primary CLL cells cocultured for 24 hours with a variety of stroma are less primed to undergo apoptosis. Priming of CLL cells from individual patients cocultured without and with CD154⁺ L-cell (A; n = 5), StromaNKTert (B; n = 4), or primary human nurse-like cells (C; n = 5) stroma for 24 hours, measured by 0.03μM BIM BH3 peptide (complete BH3 profiling data from individual patients are provided in supplemental Figure 2). All BH3 profiles are via the FACS-based method, except for 2 samples in panel B and 3 samples in panel C, which were via the plate-based method (see Methods). (D) Aggregate data from all 3 stroma (n = 14) reveals significantly decreased priming in stroma-exposed CLL cells (P = .020 by paired t test).

Figure 4

In vivo, priming is decreased in BM-derived CLL cells compared with PB-derived CLL cells. BM-derived CLL cells (n = 6) were found to be significantly less primed than their PB (n = 6) counterparts using BIM BH3 peptide at 0.01μM (P = .011), with means depicted as horizontal bars ± SEM and P ≤ .05 as indicated by asterisks. LN-derived (n = 4) CLL cells trended toward having increased priming compared with their BM-derived counterparts (P = .11).

Figure 5

CAL-101 releases CLL cells sequestered in stroma to overcome stroma-mediated resistance. PB-derived CLL cells were labeled with Calcein-AM and cocultured on StromaNKTert for 24 hours ± 10μM CAL-101, rinsed by gentle pipetting, and quantified by whole-well fluorometry. The number of adherent cells is proportional to the fluorescence units on the y-axis. Significantly more CLL cells were adherent at 24 hours (A) than at 1 hour (B), and significantly decreased CLL cell adherence was observed in the presence of 10μM CAL-101 at both 24 hours (A) and at 1 hour (B; 1-tailed P = .045 and 0.032, respectively). (C) PB-derived CLL cells were cocultured in the presence or absence of StromaNKTert for 24 hours with drug treatments as depicted in the graph, and viability assessed by annexin V/propidium iodide. The mean percent viability for 6 patients is depicted along with SEM. Patients demonstrated stroma-mediated resistance to either 100nM ABT-737 or 10μM CAL-101 alone, but this resistance was overcome by the combination of the 2 drugs in all patients. (D) Dose-response curves for CLL cells from 4 individual patients cultured in the presence of ABT-737 for 24 hours ± StromaNKTert and ± CAL-101. ABT-737 alone or in combination with CAL-101 showed dose-dependent killing in the absence of stroma. Resistance to ABT-737 alone and CAL-101 alone was observed in the presence of StromaNKTert, but this could be overcome by adding CAL-101 at 10μM to concentrations of ABT-737 as low as 10nM.

Figure 6

PI3K inhibition may help overcome stroma-mediated resistance by increasing CLL cell priming. PB-derived CLL cells from 8 individual patients were cocultured for 24 hours ± StromaNKTert and ± 10μM CAL-101. (A) Mean CLL cell percent apoptosis as measured by annexin V/propidium iodide is depicted along with SEM for all 8 patient samples. In 2-way ANOVA analysis, stroma provided significant protection from spontaneous apoptosis in the absence of CAL-101. In the absence of stroma, CAL-101 induced significantly more apoptosis than control. In the presence of stroma, CAL-101 was also able to induce significantly more apoptosis than control. (B) A trend toward increased priming by BIM BH3 peptide was observed in stroma-exposed CLL cells treated with CAL-101 compared with controls (1-tailed P = .0749). (C-D) BAD BH3 peptide and ABT-737 used as a peptide both induced significantly increased mitochondrial depolarization in stroma-exposed CLL cells treated with CAL-101 compared with control (we predicted this result based on the results shown in panel A, and therefore 1-tailed P values were used: P = .0462 and P = .0468 for BAD and ABT-737, respectively).

Figure 7

BH3 profiling demonstrates that primary CLL cells stimulated in vitro with anti-IgM or CXCL12 are less primed to undergo apoptosis and that priming can be restored with CAL-101. (A-B) PB-derived CLL cells from 4 individual patients were stimulated with 10 μg/mL of anti-IgM and BH3 profiling was performed at 18 hours. As measured by BAD peptide at 100μM and ABT-737 used like a peptide at 1μM in permeabilized cells, priming was decreased in the presence of anti-IgM, but could be restored in the presence of 5μM CAL-101. (C-D) Analogous experiments revealed that priming was also decreased in the presence of stimulation with 100 ng/mL of CXCL12, but that priming could be restored in the presence of CAL-101.