BCL6 Antagonizes NOTCH2 to Maintain Survival of Human Follicular Lymphoma Cells

Abstract

Although the BCL6 transcriptional repressor is frequently expressed in human follicular lymphomas (FL), its biological role in this disease remains unknown. Herein, we comprehensively identify the set of gene promoters directly targeted by BCL6 in primary human FLs. We noted that BCL6 binds and represses NOTCH2 and NOTCH pathway genes. Moreover, BCL6 and NOTCH2 pathway gene expression is inversely correlated in FL. Notably, BCL6 upregulation is associated with repression of NOTCH2 and its target genes in primary human and murine germinal center (GC) cells. Repression of NOTCH2 is an essential function of BCL6 in FL and GC B cells because inducible expression of Notch2 abrogated GC formation in mice and killed FL cells. Indeed, BCL6-targeting compounds or gene silencing leads to the induction of NOTCH2 activity and compromises survival of FL cells, whereas NOTCH2 depletion or pathway antagonists rescue FL cells from such effects. Moreover, BCL6 inhibitors induced NOTCH2 expression and suppressed growth of human FL xenografts in vivo and primary human FL specimens ex vivo These studies suggest that established FLs are thus dependent on BCL6 through its suppression of NOTCH2Significance: We show that human FLs are dependent on BCL6, and primary human FLs can be killed using specific BCL6 inhibitors. Integrative genomics and functional studies of BCL6 in primary FL cells point toward a novel mechanism whereby BCL6 repression of NOTCH2 drives the survival and growth of FL cells as well as GC B cells, which are the FL cell of origin. Cancer Discov; 7(5); 506-21. ©2017 AACR.This article is highlighted in the In This Issue feature, p. 443.

Figure 1. BCL6 displays a specific genomic localization pattern in FL

(A) The relative enrichment of specific gene signatures on FL BCL6 target gene sets summarized in a heat map. The statistical significance (BH adjusted p values) is provided in color key. (B) A heatmap representation of the relative transcript abundance of BCL6 target genes in FLs that display inverse correlation (p<0.05, Spearman correlation) with BCL6 expression, from a publicly available dataset of 191 primary FL expression profiles. The color key indicates the relative expression values. (C) Primary FL gene expression profiles were sorted by BCL6 expression from low to high (top row of heatmap), and the relative expression values of a set of Notch complex and target genes displayed in subsequent rows, indicating their degree of inverse correlation (p values are all p<0.05, Spearman correlation) with BCL6. Details are provided in Supplementary Table S4. (D) BCL6 binding represented for NOTCH2 and HPRT genes (negative control), in red binding of BCL6 on 4 independent FL patient samples. Y-axis represents read densities normalized to total number of reads. Threshold setting is explained in methods section. Promoter expands to −1000 base pairs (bp) downstream of TSS. (E) Cartoon representation of the RBP-Jk, HES1, MAML1, MAML2 and NOTCH2 promoter regions indicating BCL6 DNA binding motifs (orange dots) and QChIP amplicon location (arrows). (F) QChIP assays were performed in DoHH2 and Sc-1 FL cells using BCL6 antibody (black bars) and IgG (negative control, gray bars) for the genes shown in B and a negative control (NEG). The X-axis represents percent enrichment of BCL6 antibody vs. input DNA. See additional data in Supplementary Figure S1.

Figure 2. Inverse correlation between BCL6 and NOTCH2 complex genes in primary GC B-cells

(A) QPCR was performed in purified human tonsillar naïve B-cells (black) and GC B-cells (gray) to measure the relative transcript abundance of the indicated genes. The Y-axis represents mRNA expression levels normalized to HPRT. (B) A heatmap representation of BCL6, NOTCH2 and Notch complex and target gene expression levels in five human naïve B-cells and five GC B-cell specimens. The color key shows relative expression values. (C) Expression values (FPKM) of BCL6 from Naïve B-cells (purple, n=5), Centroblast (yellow, n=7) and Centrocytes (orange, n=7) from independent specimens each. (D) Expression values of NOTCH2 as in panel C. (E) Human naïve B cells were cultured with OP9 stromal monolayer and stimulated with IL4 plus IL21 or left untreated. QPCR was performed for the indicated genes. The Y-axis represents the fold change, normalized to HPRT, and relative to vehicle (control) at day 4 when maximum levels of BCL6 were reached. (F) Mouse resting B220⁺ cells were isolated and activated with mouse cytokines IL4 and IL21 for 24h. The same rationale as for panel D was followed. For panels A, D and E, the mean of three independent experiments is represented along with the SEM. For panels D and E p values are based on unpaired two-tailed t test. See Supplementary Figure S2 for additional information.

Figure 3. GC reaction is impaired in Notch2 knock-in mice

(A) Representative images of spleen sections from WT control and ICN2 knock-in mice stained for GC markers PNA (left) and BCL6⁺/B220⁺ (right) 14 days after immunization with NP₆₅-CGG. The black squares on left column (4X) highlight GCs, which are shown at 20X amplification in the right column. Scale bars are 40 μm (10X) and 20 μm (20X). (B) The number of GCs per spleen (Y-axis) from immunohistochemistry shown in panel A. PNA⁺ clusters (left) and BCL6⁺B220⁺ clusters (right) are shown. The range bars represent the mean values and SEM and the p values are shown on top. (C) The surface area occupied by GCs in the spleens of immunized and induced ICN2 and control mice is shown for PNA and BCL6/B220 staining respectively, and is represented by their area in μm² (Y-axis). The average of the means for each group is shown. Each column corresponds to an individual mouse; each point is an individual GC; p values are shown on top. SEM and p values are shown. (D) Average of GC area (μm²) of WT (black) and ICN2 (green) mice from IHC of paraffin-embedded spleen slides stained with PNA⁺ and BCL6⁺/B220⁺ antibodies. The mean values are 261.7 +/−36.48 vs. 91.05 +/−10.22 μm² for PNA⁺ GC, and 177.1 +/−30.22 vs. 83.93 +/−13.13 μm². The statistical significance of this difference is shown based on unpaired two-tailed t test (P= 0.0015 and P= 0.0336 respectively). Data shown for immunized WT (n=6) and ICN2 (n=6) mice. (E) Flow-cytometry analysis of B220⁺GL7⁺CD95⁺ labeled splenic GC B-cell populations. Cells were gated for B220⁺, GL7 is on the Y-axis and CD95(Fas) is on the X-axis. The percentage of double positive cells corresponds to GC B-cells (black box). (F) The percentage of B220⁺ gated GL7⁺/CD95⁺ GC B-cells among total splenocytes is shown from the spleens of immunized WT (n=6) and ICN2 induced (n=6) mice. In panels B, C and E the statistical values are based on unpaired two-tailed t test. (G) Average of GC area (μm²) of WT (black) and ICN2 (green) mice from IHC of paraffin-embedded spleen slides stained with PNA⁺ and BCL6⁺/B220⁺ antibodies. The mean values are 262+/−36 vs. 91+/−10 μm² for PNA⁺ GC, and 177+/−30 vs. 84+/−13 μm². Data shown for immunized CγCre-WT (n=5) and CγCre-ICN2 (n=8) mice. The statistical significance of this difference is shown based on unpaired two-tailed t test (P= 0.0015 and P= 0.0336 respectively). (H) Flow-cytometry analysis of B220⁺GL7⁺CD95⁺ labeled splenic GC B-cell populations. Cells were gated for B220⁺, GL7 is on the Y-axis and CD95(Fas) is on the X-axis. The percentage of double positive cells corresponds to GC B-cells (black box). (I) The percentage of B220⁺ gated GL7⁺/CD95⁺ GC B-cells among total splenocytes is shown from the spleens of immunized WT (n=5) and ICN2 induced (n=8) mice. In the presence of ICN2, the abundance of B220⁺GL7⁺CD95⁺ GC B-cells was reduced three-fold as compared to WT animals (1.7% vs. 0.6% mean GC B-cells vs. total splenocytes, p=0.0003 unpaired two-tailed t test. See Supplementary Figures S3 and S4 for additional data.

Figure 4. BCL6 represses NOTCH2 complex genes and Notch activity

(A) The relative transcript abundance of NOTCH2, MAML1, MAML2 and RBPJ-k was examined by QPCR in DoHH2 (black bars) and Sc-1 (gray bars) 72 h after BCL6 siRNA depletion vs. control siRNA. Values were normalized to HPRT and fold change (Y-axis) is represented over a scrambled siRNA control. (B) The relative transcript abundance of NOTCH2, MAML1, MAML2 and RBP-Jk was examined by QPCR in FL cell lines DoHH2 (black bars) and Sc-1 (gray bars) 72 h after RI-BPI treatment (15 μM RI-BPI) in the right panel. Values were normalized to HPRT and fold change (Y-axis) is represented over Control Peptide (CP). (C) Reporter assays performed in DoHH2 (black bars) and Sc-1 (gray bars) cells transfected with pGL2-HESAB (Notch reporter) or pGL2 control vector, and with BCL6 (siBCL6) or control siRNA (siC). The Y-axis shows the luciferase activity relative to renilla (internal control). All panels represent the mean of three independent experiments, each performed in triplicate, and the error bars represent the standard error of the mean (SEM). The statistical values are based on unpaired two-tailed t test. (D) Expression values (FPKM) of NOTCH ligands DLL1, DLL3, DLL4, JAG1 and JAG2 on DoHH2 (blue), Sc-1 (yellow), OCI-Ly1 (purple), SU-DHL-4 (green). (E) As in panel D, expression values (FPKM) of metalloproteases ADAM10 and ADAM17 on the same cell lines. (F) Expression values (FPKM) of DLL1 (upper panel), DLL3 (middle panel) and DLL4 (bottom panel) from left to right Naïve B-cells (n=5, NB), Germinal Center B cells (n=4, GCB), Centroblast (n=7, CB), Centrocytes (n=.7, CC), Bone Marrow Plasma Cells (n=3, BMPC), Tonsilar plasma cells (n=5, TPC), Memory B cells (n=8, MB), Follicular lymphoma (n=77, FL) from independent specimens each. Star means p-value <= 0.05. Square dots are outliers (below 1st quartile or above 4th quartile). (G) As in panel F, expression values (FPKM) of JAG1 (upper panel) and JAG2 (bottom panel). (H) As in panel G, expression values (FPKM) of ADAM10 (upper panel) and ADAM 17 (bottom panel). (I) Flow cytometry to assess apoptosis of FL cell lines driven by DLL1 ligand. The % of apoptotic cells is observed on the upper-right quadrant of double positive labeled cells for propidium iodide (Y-axis) and Annexin V (X-axis) on DoHH2 (top) and Sc-1 (bottom) cell lines co-cultured with HS5-Control (right graph) or HS5-DLL1 (triplicates in left graphs) stromal cell line for 48h. See Supplementary Figure S5 for additional data.

Figure 5. FL cells are dependent on BCL6 in a NOTCH2-dependent manner

(A) DoHH2, Sc-1 and WSU-DLCL2 FL cell lines were exposed to six concentrations of RI-BPI (from 1 to 40 μM) or vehicle (water) for 48 h. The X-axis shows the dose of RI-BPI. The Y-axis shows the fractional effect of RI-BPI vs. control on cell viability. The experiment was done in triplicates. The dose (in μM) that inhibited cell growth by 50% (GI₅₀) is shown next to each cell line. (B) Luminescent Caspase 7 and 3 activity assays were performed in FL cell lines (X-axis) exposed to vehicle (gray columns) or RI-BPI 10 μM (black columns) for 24 h. Results are expressed in percent of RLU to control (Y-axis). (C) DoHH2, Sc-1, SU-DHL-4 and OCI-Ly1 cells were transfected NOTCH2 siRNA, BCL6 siRNA or both as indicated and cell viability measured at 48 and 72 hours. The Y-axis represents percent cell viability, normalized to control siRNA (siC, dotted line). (D) DoHH2 cells were transfected with NOTCH2 siRNA as indicated and control siRNA as indicated, 24 hours post-transfection cells were treated with 10 μM RI-BPI and 24 hours post-treatment cell viability was measured as described before. The figure shows the mean of 3 experiments with SEM. (E) DoHH2 and SU-DHL-4 cells were pre-treated with anti-IgG1 or NOTCH2 antagonist antibody NRR2 for 24h and then exposed to 10 μM RI-BPI or control peptide (CP). Y-axis represents viability (At 24 and 48h) relative to control antibody (anti-IgG1) treated with CP. All experiments, unless otherwise indicated, were performed in triplicates and the error bars represent the SEM. Statistical significance is shown (p value one-tailed t test): * p<0.05; **p<0.005. See Supplementary Figures S6 and S7 for additional data.

Figure 6. RI-BPI suppresses FL tumors in vivo and ex vivo

(A) Single cell suspensions of 17 confirmed FL specimens were exposed to vehicle (Control line) or 20 μM of RI-BPI (except case 14 that was treated with 5 μM) for 48 h. Seven samples were BCL6 negative (gray columns) and 10 samples were BCL6 positive (black columns). Cell viability (represented as percent of control treated cells) is shown on the Y-axis. Individual cases as well as the average for all the cases (m) are shown on the X-axis. Statistical significance (unpaired t test) was determined for the average of BCL6 positive vs. BCL6 negative cases. The experiment was carried out in duplicates. (B) An FL specimen exposed to 10 μM RI-BPI was harvested 48h post-treatment and mRNA abundance examined by QPCR for NOTCH2, HES6 and HES1 and normalized to HPRT. Results are expressed as fold induction compared to control (vehicle). (C) Tumor growth plots in DoHH2 (left) and Sc-1 (right) xenografted mice treated with vehicle (PBS, n=5, gray lines) or RI-BPI 25 mg/kg/day (n=5, black lines) for 10 consecutive days. The Y-axis indicates tumor volume (in mm³) and X-axis days of treatment. The p values represent the comparison of tumor volumes in treated to control mice at day 10 by Student’s t test. (D) Representative immunohistochemistry images from DoHH2 and Sc-1 tumors after treatment with control or RI-BPI assayed for apoptosis by TUNEL and caspase 3 staining (top and bottom panels respectively). Red bar represents 50 μm. (E) QPCR was performed in triplicate from the DoHH2 FL xenografts of mice treated with vehicle (n=4) or RI-BPI 25 mg/kg/day for seven days (n=4) to assess transcript abundance of NOTCH2, MAML1, MAML2 and HES1, normalized to HPRT. Statistical significance was determined by Mann-Whitney test. See Supplementary Figure S8 for additional data.