BCL6 maintains survival and self-renewal of primary human acute myeloid leukemia cells

Abstract

B-cell lymphoma 6 (BCL6) is a transcription repressor and proto-oncogene that plays a crucial role in the innate and adaptive immune system and lymphoid neoplasms. However, its role in myeloid malignancies remains unclear. Here, we explored the role of BCL6 in acute myeloid leukemia (AML). BCL6 was expressed at variable and often high levels in AML cell lines and primary AML samples. AMLs with higher levels of BCL6 were generally sensitive to treatment with BCL6 inhibitors, with the exception of those with monocytic differentiation. Gene expression profiling of AML cells treated with a BCL6 inhibitor revealed induction of BCL6-repressed target genes and transcriptional programs linked to DNA damage checkpoints and downregulation of stem cell genes. Ex vivo treatment of primary AML cells with BCL6 inhibitors induced apoptosis and decreased colony-forming capacity, which correlated with the levels of BCL6 expression. Importantly, inhibition or knockdown of BCL6 in primary AML cells resulted in a significant reduction of leukemia-initiating capacity in mice, suggesting ablation of leukemia repopulating cell functionality. In contrast, BCL6 knockout or inhibition did not suppress the function of normal hematopoietic stem cells. Treatment with cytarabine further induced BCL6 expression, and the levels of BCL6 induction were correlated with resistance to cytarabine. Treatment of AML patient-derived xenografts with BCL6 inhibitor plus cytarabine suggested enhanced antileukemia activity with this combination. Hence, pharmacologic inhibition of BCL6 might provide a novel therapeutic strategy for ablation of leukemia-repopulating cells and increased responsiveness to chemotherapy.

Graphical abstract

Figure 1.

BCL6 expression in AML subclasses is heterogeneous. (A) BCL6 expression was compared between normal human hematopoietic cells (GSE42519 and human primary cell atlas) and human AML cells using 2 different AML cohorts. Using 2 independent cohorts of AML from Oregon Health & Science University (OHSU; 531 cases) and Erasmus (520 cases), BCL6 expression level normalized by housekeeping genes (RRN18S, ACTB, GAPDH, PGK1, PPIA, RPL13A, RPLP0, ARBP, B2M, YWHAZ, SDHA, TFRC, GUSB, HMBS, HPRT1, and TBP) was compared between molecular markers. Trends for biomarkers and BCL6 expression level were compared between 2 cohorts, each of which was compared with normal hematopoietic cells. Red and blue dotted lines are placed as a reference for the levels of expression for HSC and GMP, respectively. Statistical differences for BCL6 expression were calculated between each of the AML categories vs median values of either normal HSCs or GMPs. Asterisks represent the significance in either red or blue in comparisons vs HSCs or GMPs, respectively. (B) Expression level of BCL6 was compared in the same AML cohorts between different FAB subclasses. (C) Primary AML cells were subjected to quantitative reverse transcription polymerase chain reaction to compare expression levels of BCL6 either among primary AML cells or to reference normal donor blood cells. For primary AML specimens, bulk AML cells were subjected to blast purification (MACS sorting to exclude CD3 positive and CD19 positive cells), and BM cells were sorted for CD34 positive cells (using MACS). Relative expression of BCL6 was calculated by normalization to GAPDH (glyceraldehyde-3-phosphate dehydrogenase). Histograms represent averages of duplicates, and error bars are standard deviations. (D) Expression levels of BCL6 were analyzed by intracellular staining and flow cytometry. Expression levels of AML cells were compared vs those of normal BM cells. Mean fluorescence intensity (MFI) values were normalized to isotype control. Experiments were done in triplicates, and the mean from 3 independent experiments was plotted. Error bars indicate standard deviations. BC, band cell; CMP, common myeloid progenitor cell; MEP, megakaryocyte-erythroid progenitor cell; MM, metamyelocytes; MPP, multipotential progenitors; MY, myelocyte; nBMC (CD34⁺), CD34 positive normal bone marrow cells; PM, promyelocyte; PMN, polymorphonuclear cells.

Figure 2.

BCL6-targeted peptide inhibitors induced significant apoptosis associated with differential expression of BCL6 genes in primary AML cells. (A) Fifty-five different AML cells were subjected to ex vivo treatment with 10 μM RI-BPI for 48 hours. The relative percentage of live cells (annexin V negative/7-AAD negative) were plotted and sorted from lowest (left) to highest (right) sensitivity to RI-BPI. Dotted line indicates 50% of relative live cell percentage with RI-BPI treatment. Profiles of each of the primary cells used in the analysis are provided in supplemental Table 1. (B) Representative flow cytometry data to show how dead cells were detected, with a comparison between untreated and RI-BPI–treated cells. (C) Based on mutational profiles of the 55 primary AML cells, normalized percent live cells after RI-BPI (10 mM, 48 hours) treatment was compared between subgroups with different molecular markers. Linear regression analysis showing a correlation between sensitivity to RI-BPI and expression level of BCL6 in 6 different primary AML cells (D) and in 11 different AML cell lines (E). MFI, mean fluorescence intensity.

Figure 3.

Inhibition of BCL6 in AML cells resulted in regulations of multiple numbers of biologically important pathways. Three primary AML cells (AML37, AML95, and AML98) were treated with FX1 using the concentration for EC₅₀ in pilot ex vivo treatment experiments, for 12 hours. RNA was extracted from dimethyl sulfoxide (DMSO)- or FX1-treated cells and subjected to RNA-sequencing. (A) After a consensus clustering of gene expression profiles, GSEA was performed between FX1-treated and DMSO-treated AML cells. Representative gene signatures are highlighted and sorted with biological functions shown at the right. (B-G) Enrichment plots of regulated gene signatures of interest obtained by GSEA analysis showing enrichments in FX1-treated vs DMSO. NES and FDR values are attached for each of the gene sets.

Figure 4.

BCL6 inhibition ablates LSCs. (A) Measurement of expression levels of BCL6 in LSCs (CD45^dim, CD34⁺, and CD38^–) and blast cells (CD45^dim, CD34⁺, and CD38⁺) were measured by intracellular staining and flow cytometry. Averages of absolute mean fluorescence intensity (MFI) values of PE-BCL6 are shown from triplicates, with standard deviations as error bars. (B) Representative flow cytometry data of the analysis done in (A). (C) Three different primary AML cells were treated with 10 μM RI-BPI for 48 hours, and percentages of dead cells were measured by using annexin V–fluorescein isothiocyanate (FITC) staining and flow cytometry. P values indicate significance of correlations between dose and percent live cells for each of the primary AML cells. (D) Representative flow cytometry density plots to compare untreated and RI-BPI treatment. (E) Primary AML cells were treated with 10 μM RI-BPI for 24 hours in duplicates, and colony numbers were counted after 14 days. Values were normalized to vehicle-treated controls. The data are plotted as mean of duplicates. Error bars indicate standard deviation. (F) Schematic illustration and (G) comparison of results in 2 different in vivo behaviors of AML cells treated with a BCL6 inhibitor. Two primary AML cells (AML2 and AML9) were subjected to 24 hours of ex vivo pretreatment with either RI-BPI (5 μM), control peptide inhibitor (CPI), or were untreated (UT). The treated cells were then injected into patient-derived xenograft recipient mice. Six to 8 weeks after the injection, percentages of human AML cells were monitored by measuring mCD45^–, hCD45⁺, and hCD33⁺ cells in the BM aspirates. Schematic illustration (H) and comparison of results (I) of in vivo engraftment of AML cells transfected with BCL6 siRNA. BCL6 or nonspecific control siRNA transfected AML cells (AML33) were injected in mice. The plot shows the percentages of human AML cells in the presence or absence of BCL6 knockdown. Rx, radiation.

Figure 5.

BCL6 is induced in AML cells treated with chemotherapeutic agents. (A) Linear regression analysis showing correlation between basal BCL6 expression and sensitivity to AraC treatment in 11 different AML cell lines. (B) Linear regression analysis showing correlation between fold induction of BCL6 expression and sensitivity to AraC treatment (1.25 μM) using 11 different AML cell lines. Fold inductions in each of the cells were calculated based on mean fluorescence intensity (MFI) values by comparing treated vs untreated cells. (C-D) Induction of BCL6 in 4 different primary AML cells (AML54, AML74, AML95, and AML109) treated with AraC (111, 333, 1000, and 3000 nM, for 24 hours) in duplicates. Relative MFI of PE-BCL6 in each of the cells. The fold induction was calculated by normalization to untreated samples, and fractions of increase/decrease are presented. The data are plotted as averages of duplicates. Error bars indicate standard deviation. (E) Schematic illustration of BCL6 expression analysis after Ara C in vivo treatment of AML patient-derived xenografts. (F) Plot showing induction of BCL6 expression following engraftment of AML cells pretreated with AraC. (G) Analysis of BCL6 protein levels in AML cell lines treated with 6 different therapeutic drugs. BCL6 expression was evaluated by using flow cytometry. Rx, radiation.

Figure 6.

Combination of BCL6 inhibitor and AraC mediates synergistic effects. (A) Five primary AML specimens (AML54, AML74, AML95, AML98, and AML109) were subjected to ex vivo combination therapy using AraC and FX1 with various combinations of limiting dilution assay. Matrices indicate calculated synergy scores obtained from each pair of combination doses. Synergy scores were calculated by using SynergyFinder software. The highest single agent (HSA) algorithm was used to calculate synergy scores, and color keys indicate the scores in heatmaps. (B) Colony formation assay in primary AML cells treated with AraC, BPI, or combination of both. Statistical significance between each of the conditions was calculated by one-way analysis of variance. (C) Evaluation of cell deaths in primary AML cells (AML33) treated with AraC in the presence or absence of siRNA-induced BCL6 knockdown. Primary AML cells (AML33) were transiently transfected with control or BCL6 siRNA; the cells were then treated with AraC in duplicates. Relative percent dead cells were calculated by normalization to untreated cells. Comparing the relative percent dead cells vs that of parental AML33 cells treated with AraC, the increase of dead cells with each siRNA is presented as relative values. Averages were obtained from duplicates, and error bars represent standard deviations. (D) The same series of primary AML33 cells as (C) were subjected to colony formation assays after 24 hours of ex vivo treatment with AraC. The number of colonies was evaluated after 14 days. Colony counts were compared between untreated and AraC treatment to obtain relative colony numbers, and relative colony number reductions were obtained by comparing the relative values vs parental AML cells. In (C) and (D), statistical significance was evaluated by using an unpaired Student t test. (E) Representative photographs of colony formation assays. Thirty-five (7 × 5) pictures of single fields were tiled to show entire views of each plate. (F) Representative fields marked by the box in (E) were presented in zoomed views. CFU, colony-forming unit; NS, not significant; NTC, nontargeting control.

Figure 7.

Combination therapy using BCL inhibitor and AraC. (A) Schematic illustration of serial BM transplantation (BMT) assay. (B) Primary AML cells were transplanted in recipient mice and treated with AraC (10 mg/kg), BPI (25 mg/ kg), or in combination for 3 weeks after confirming engraftment. Percent human AML cells in each arm was compared. (C) BM cells were harvested and transplanted into new recipient mice (secondary transplant). NS, not significant; PDX, patient-derived xenograft.