Targeting BTK through microRNA in chronic lymphocytic leukemia

Abstract

Bruton's tyrosine kinase (BTK) is a critical mediator of survival in B-cell neoplasms. Although BTK inhibitors have transformed therapy in chronic lymphocytic leukemia (CLL), patients with high-risk genetics are at risk for relapse and have a poor prognosis. Identification of novel therapeutic strategies for this group of patients is an urgent unmet clinical need, and therapies that target BTK via alternative mechanisms may fill this niche. Herein, we identify a set of microRNAs (miRs) that target BTK in primary CLL cells and show that the histone deacetylase (HDAC) repressor complex is recruited to these miR promoters to silence their expression. Targeting the HDACs by using either RNA interference against HDAC1 in CLL or a small molecule inhibitor (HDACi) in CLL and mantle cell lymphoma restored the expression of the BTK-targeting miRs with loss of BTK protein and downstream signaling and consequent cell death. We have also made the novel and clinically relevant discovery that inhibition of HDAC induces the BTK-targeting miRs in ibrutinib-sensitive and resistant CLL to effectively reduce both wild-type and C481S-mutant BTK. This finding identifies a novel strategy that may be promising as a therapeutic modality to eliminate the C481S-mutant BTK clone that drives resistance to ibrutinib and provides the rationale for a combination strategy that includes ibrutinib to dually target BTK to suppress its prosurvival signaling.

Figure 1.

An miR signature targets BTK in CLL. (A) Putative binding sites of 6 miRs in the 3′UTR of BTK. (B) Primary CLL cells were transiently transfected with scrambled (Scr) oligonucleotides or miR-147b, miR-210, miR-425, miR-1253, miR-4269, or miR-4667-3p for 48 hours and immunoblotted for BTK and GAPDH. The figure is representative of 7 to 10 independent CLL samples which are quantitated in (C) (P < .05 for miR-210 and miR-425, Wilcoxon signed rank test). (D) Mec2 cells transfected with mimics or inhibitors of miR-210 and miR-425 and analyzed for the levels of BTK and GAPDH (P < .05, one-sample Student t test; two-tailed P). (E-F) Reporter gene analyses using BTK 3′UTR constructs. Mec1 cells were transfected with luciferase reporter constructs expressing wt BTK or BTK with the binding sites for miR-147, miR-210, miR-425, miR-1253, miR-4269, and miR-4667-3p mutated. Cells were then transfected with an irrelevant miR (miR-181) or miR-147b, miR-210, miR-425, miR-1253, miR-4269, or miR-4667-3p, and the luciferase counts were quantitated. Data represent mean ± standard error of the mean (SEM) of 12 independent transfections (P < .05 for miR-4269 and P < .001 for miR-147b, miR-210, miR-425, miR-1253, and miR-4667-3p; paired Student t test). (G) Expression analysis of miR-210, miR-425, miR-1253, miR-4269, and miR-4667-3p in 83 CLL samples expressed as a fraction of the levels for these miRs from CD19⁺CD5⁺ B cells from healthy donors (set at 1) (P < .01 average reduction across all miRs, mixed effects model). *P ≤ .05; **P ≤ .01; ***P ≤ .001; **** P ≤ .0001.

Figure 2.

Recruitment of HDAC1, HDAC2, and KDM1 to the promoters of BTK-targeting miRs in CLL. (A) Co-immunoprecipitation (IP) of HDAC1, HDAC2, and KDM1 from the nuclei of 3 CLL samples in comparison with IgG. Figure is representative of 3 experiments. (B) Relative recruitment of HDAC1, HDAC2, and KDM1 in comparison with IgG at the miR-147b, miR-210, miR-425, miR-1253, miR-4269, miR-4667-3p, CDKN1a (gene regulated by HDACs used as a positive control), and miR-622 (gene predicted to target BTK but not regulated by HDACs used as negative control) promoters in CLL samples with adverse cytogenetics. Data representative of 3 individual CLL samples analyzed by ChIP-Seq. (C-E) Recruitment of HDAC1, HDAC2, and KDM1 at miR-147b, miR-210, miR-425, miR-1253, miR-4269, miR-4667-3p using primers specific or nonspecific (NS) to the promoters for these miRs. Data represent mean ± SEM of 6 CLL samples.

Figure 3.

HDAC inhibition increases H3K4me3 at the promoters of the BTK-targeting miRs, induces their expression, and results in a reciprocal decrease in BTK protein in CLL. (A) Accumulation of the transcriptionally permissive marks H3K4me2 and H3K4me3 after HDAC inhibition at the miR-147b, miR-210, miR-425, miR-1253, miR-4269, miR-4667-3p, CNDN1a (positive control), and miR-622 promoters (negative control) in CLL samples with adverse cytogenetics. Data are representative of 3 individual CLL samples analyzed by ChIP-Seq. (B) Accumulation of the transcriptionally permissive mark H3K9Ac3 after HDAC inhibition at the miR-147b, miR-210, miR-425, miR-1253, miR-4269, miR-4667-3p promoters using primers specific or nonspecific to the promoters for these miRs. Data represent mean ± SEM of 4 CLL samples. (C) Induction of miR-147b, miR-210, miR-425, miR-1253, miR-4269, miR-4667-3p in 79 CLL samples after HDAC inhibition. Expression of the BTK-targeting miRs were normalized to the expression of snRNA48, which did not change after HDAC inhibition (P < .001, mixed effects model). (D) Effect of abexinostat for 24 or 36 hours on BTK mRNA. Data represent mean ± SEM of 22 CLL samples (P < .001, paired Student t test). (E) Effect of abexinostat for 36 hours on BTK protein levels. Data represent mean ± SEM of 60 CLL samples (P < .001, Wilcoxon signed rank test). ***P ≤ .001.

Figure 4.

RNAi against HDAC1 induces BTK-targeting miRs to reduce BTK protein in primary CLL cells. (A-B) Primary CLL cells were nucleofected with small interferring RNA (siRNA) against HDAC1 for 48 hours before being harvested for RNA and protein. The top panel shows the effect of siRNA-HDAC1 on HDAC1 protein. GAPDH was used as a loading control (middle panel). The bottom panel shows the consequence of si-HDAC1 on BTK protein. The figure is representative of 10 independent transfections. (B) Quantitation of the effect of siHDAC1 on HDAC1 protein. Data represent mean ± SEM of 10 CLL samples (P < .01, paired Student t test). (C) The expression of miR-147b, miR-210, miR-425, miR-1253, miR-4269, miR-4667-3p was quantitated in the primary CLL cells nucleofected in (A). Data represent mean ± SEM of 7 CLL samples. (D) The expression of BTK mRNA in the primary CLL cells nucleofected in (A). Data represent mean ± SEM of 7 CLL samples. (E) The expression of BTK protein (P < .05, Wilcoxon signed rank test) in the primary CLL cells nucleofected in (A). Data represent mean ± SEM of 9 CLL samples. *P ≤ .05; **P ≤ .01.

Figure 5.

Effect of abexinostat combined with ibrutinib on BTK protein, signaling, and CLL survival. (A) Induction of miR-147b, miR-210, miR-425, miR-1253, miR-4269, and miR-4667-3p in CLL cells left untreated and exposed to abexinostat (Abex), ibrutinib (IB), or abexinostat and ibrutinib combined. Data represent mean ± SEM of 18 CLL samples. After abexinostat: P < .001 for miR-210 and miR-4667-3p; P < .05 for miR-147b, miR1253, and miR-4269; after abexinostat and ibrutinib combined: P = .001 for miR-210; P < .05 for miR-4269 and miR-4667-3p mixed effects model. (B) BTK mRNA in cells left untreated and exposed to abexinostat, ibrutinib, or abexinostat and ibrutinib combined. Data represent the mean ± SEM of 10 CLL samples (P < .01 for BTK expression after exposure to either abexinostat or abexinostat plus ibrutinib, mixed effects model). (C) Action of abexinostat alone or combined with ibrutinib on p-Y223-BTK, total BTK, p-PLCγ2, total PLCγ2, p-ERK, total ERK, p-AKT, total AKT, and H3K9Ac. GAPDH was assayed as a loading control, and H3 was assayed as a control for H3K9Ac. (D) Effect of abexinostat, ibrutinib, or combined abexinostat and ibrutinib on the induction of apoptosis (Annexin V–positive cells) in primary CLL samples. Data represent the mean ± SEM of 13 CLL samples (P < .05, interaction test of synergy from a mixed effects model). *P ≤ .05; **P ≤ .01; ***P ≤ .001.

Figure 6.

HDAC inhibition effectively targets the C481S-mutant BTK clone in ibrutinib-resistant CLL. (A) Induction of miR-147b, miR-210, miR-425, miR-1253, miR-4269, and miR-4667-3p in response to abexinostat in paired CLL samples obtained at baseline or after acquisition of the C481S BTK clone while the patient was receiving ibrutinib therapy. Data represent mean ± SEM of 4 CLL samples (P < .01 for miR-210 before and after ibrutinib resistance had developed, and P < .01 for miR-4667-3p after ibrutinib resistance had developed; mixed effects model). (B) Effect of ibrutinib or abexinostat on BTK mRNA in paired CLL samples obtained at baseline or after acquisition of the C481S BTK clone while the patient received ibrutinib therapy. Data represent mean ± SEM of 4 CLL samples (P < .01; mixed effects model). (C) Effect of ibrutinib or abexinostat at 18 and 36 hours on BTK phosphorylation, protein, and signaling in paired CLL samples obtained at baseline or after acquisition of the C481S BTK clone while the patient received ibrutinib therapy. Levels of H3K4Ac3 were evaluated as a measure of HDAC inhibition, and H3 was measured as a control for H3K9Ac. (D) The decrease in BTK protein was quantitated; data represent mean ± SEM of 4 CLL samples (P < .001; mixed effects model). (E) Effect of abexinostat on CLL survival at 48 hours as measured by increase in percentage of annexin V–positive cells in paired CLL samples obtained at baseline or after acquisition of the C481S BTK clone while the patient received ibrutinib therapy. ***P ≤ .001.

Figure 7.

Role of HDACs in silencing the BTK-targeting miR in the TCL1 model of CLL and consequence of the combination of abexinostat and ibrutinib on BTK protein, signaling, and leukemia cell survival in vivo. (A) CLL cells were isolated from the spleen of TCL1 transgenic mice before (CD19⁺CD5⁺ cells <10%) and after developing CLL (CD19⁺CD5⁺ cells >90%). RNA was extracted and used to quantitate the expression of miR-210 and miR-425 (P < .05 for miR-425, two-sample Student t test). (B) CLL cells from TCL1 mice with leukemia (CD19⁺CD5⁺ cells >90%) were exposed to either panobinostat or abexinostat for 18 hours and then evaluated for the expression of miR-210 and miR-425 (P < .05 for both, paired Student t test). (C-D) Mice were randomly assigned to vehicle/vehicle (control), abexinostat/vehicle, ibrutinib/vehicle, or a combination of abexinostat and ibrutinib for 2 weeks. Spleen CLL cells were isolated and used to quantitate miR-210 and miR-425 (P < .05 for miR-210 with the combination; analysis of variance). (E) CLL cells from TCL1 mice with leukemia (CD19⁺CD5⁺ cells >90%) were left untreated or exposed to abexinostat, ibrutinib, or the combination before being assayed for p-BTK and BTK. GAPDH was used as a loading control, and H3K9/14 was assayed to measure HDAC inhibition. (F) BTK protein was quantified in CLL cells from (E) (P < .05 for the combination vs vehicle; paired Student t test). (G) Mice were randomly assigned to vehicle/vehicle (control), abexinostat/vehicle, ibrutinib/vehicle, or a combination of abexinostat and ibrutinib for 2 weeks. Spleen CLL cells were isolated from each group, and the number of CD19⁺CD5⁺ CLL cells was quantitated (P < .05 for the combination over treatment with each individual drug; mixed effects model). *P ≤ .05.