Rationally designed BCL6 inhibitors target activated B cell diffuse large B cell lymphoma

Abstract

Diffuse large B cell lymphomas (DLBCLs) arise from proliferating B cells transiting different stages of the germinal center reaction. In activated B cell DLBCLs (ABC-DLBCLs), a class of DLBCLs that respond poorly to current therapies, chromosomal translocations and amplification lead to constitutive expression of the B cell lymphoma 6 (BCL6) oncogene. The role of BCL6 in maintaining these lymphomas has not been investigated. Here, we designed small-molecule inhibitors that display higher affinity for BCL6 than its endogenous corepressor ligands to evaluate their therapeutic efficacy for targeting ABC-DLBCL. We used an in silico drug design functional-group mapping approach called SILCS to create a specific BCL6 inhibitor called FX1 that has 10-fold greater potency than endogenous corepressors and binds an essential region of the BCL6 lateral groove. FX1 disrupted formation of the BCL6 repression complex, reactivated BCL6 target genes, and mimicked the phenotype of mice engineered to express BCL6 with corepressor binding site mutations. Low doses of FX1 induced regression of established tumors in mice bearing DLBCL xenografts. Furthermore, FX1 suppressed ABC-DLBCL cells in vitro and in vivo, as well as primary human ABC-DLBCL specimens ex vivo. These findings indicate that ABC-DLBCL is a BCL6-dependent disease that can be targeted by rationally designed inhibitors that exceed the binding affinity of natural BCL6 ligands.

Figure 1. Identification and characterization of FX1 as a BCL6 BTB inhibitor.

(A) SILCS FragMaps overlaid on the apo BCL6 BTB domain with the aromatic (purple), aliphatic (green), hydrogen bond donor (blue) and acceptor (red), and charged acceptor (orange) maps. (B and C) Superposition of the SILCS FragMaps with 79-6 alone (B) and with FX1 and 79-6 in complex with BCL6 BTB (C). (D) Comparison of the affinity of the natural ligand SMRT peptide and the small molecules 79-6 and FX1 to BCL6 BTB domains determined by microscale thermophoresis (MST; n = 3 independent experiments; error bars represent the SD). (E) Comparison of the MST results of SMRT and FX1 interaction with the BTB domains of BCL6 or LRF. Results represent mean ± SD of 3 independent experiments. (F) Luciferase reporter assays showing activity of 79-6 or FX1 as compared with vehicle against the repressor activity of a GAL4^DBD-BCL6^BTB fusion construct compared with the GAL4^DBD alone. The y axis represents the relative percentage of repression mediated by the fusion protein in the presence of vehicle, set as 100%. Bars represent mean ± SD of 3 independent experiments. (*P < 0.05, **P < 0.005, t test.) (G) Heteronuclear single quantum coherence spectrum of 250 μM BCL6 BTB with 5% DMSO (red) is superimposed onto the spectrum of BTB with 500 μM FX1 (blue). Residues that experience the most significant chemical shift perturbation are labeled. (H) A graphical representation of the BCL6 BTB domain homodimer based on Protein Data Bank (PDB) structure 1R2B is shown, indicating residues perturbed upon binding of FX1 in magenta.

Figure 2. FX1 disrupts BCL6 repression complexes and induces derepression of BCL6 target genes.

(A) Quantitative ChIP was performed in SUDHL-6 cells exposed to FX1 (black bars) or vehicle (white bars) in DLBCL cells using antibodies for BCL6, SMRT, BCOR, or IgG control to enrich for known BCL6 binding sites in the CD69, CXCR4, and DUSP5 loci, or a negative control region. The y axis represents fold enrichment of binding versus input, as compared with IgG control (**P < 0.005, t test). (B) Quantitative PCR was performed in SUDHL-6 and SUDHL-4 cells after exposure to FX1 or vehicle to measure abundance of the BCL6 target genes CASP8, CD69, CXCR4, CDKN1A, and DUSP5. The y axis shows fold enrichment versus HPRT mRNA based on the ΔΔ^Ct values (*P < 0.05, **P < 0.005, t test). (C) Gene set enrichment analysis was performed using gene expression profiles obtained by RNAseq after exposure to FX1 as compared with vehicle in the indicated cell lines, against the ranked list of genes induced by BCL6 siRNA in DLBCL cells. Bars in A and B represent mean ± SEM of 3 independent experiments. NES, normalized enrichment score.

Figure 3. FX1 phenocopies the BCL6 mutant phenotype.

Ten C57BL/6 mice were immunized with sheep red blood cells and then treated i.p. with 80 mg/kg/d FX1 or vehicle alone daily for 8 days starting 48 hours after immunization. (A) Spleen weights from mice treated with FX1 or vehicle (2-tailed Mann-Whitney unpaired test). (B) Flow cytometry quantification of total B cells (B220⁺) (t test). (C) Flow cytometry quantification of GC B cells (B220⁺DAPI^–GL7⁺FAS⁺, t test). (D) IHC of spleens from mice treated with FX1 or vehicle and stained with peanut agglutinin (PNA), Ki-67, and B220. Scale bars: 500 μm. Quantification of number and area of GCs was performed by ImageJ software. The y axis shows number of positive cells/total cell number of different sections of spleens (n = 10) (2-tailed Mann-Whitney unpaired test). Values in A–D are shown as mean ± SEM.

Figure 4. FX1 selectively suppresses BCL6-dependent GCB-DLBCL growth in vitro and in vivo.

(A) Viability of BCL6-dependent and -independent DLBCL cell lines after 48 hours of treatment with different concentrations of FX1 based on resazurin reduction. The y axis shows percentage of growth-suppressing effect of the compound compared with vehicle-treated cells. Effect = 100% – 100% × (fluorescence of FX1-treated cells/fluorescence of vehicle-treated cells). The graph represents average of 3 independent experiments, and the GI₅₀ values represent mean ± SD of 3 biological replicates. (B) Tumor volume of established OCI-Ly7 xenografts implanted in SCID mice treated with daily injections of 25 or 50 mg/kg 79-6 or FX1 versus vehicle for 10 days (n = 10 mice per group, 2-tailed Mann-Whitney unpaired test). (C) Tumor burden AUC was calculated from the same mice as in B between the initial tumor volume (t₀: 100 mm³) and the final volume at day 9 (n = 10 mice per group, 2-tailed Mann-Whitney unpaired test). Values in B and C are shown as mean ± SEM.

Figure 5. FX1 suppresses the growth of ABC-DLBCLs.

(A) ABC-DLBCL cells were treated with increasing doses of FX1 versus vehicle. Cell viability was measured after 48 hours of treatment by resazurin reduction. The y axis shows percentage of growth-suppressive effect of the compound compared with vehicle-treated cells. Effect = 100% – 100% × (fluorescence of FX1-treated cells/fluorescence of vehicle-treated cells). The graph represents an average of 3 independent experiments. (B) Tumor volume of established HBL-1 xenografts implanted in NOD/SCID mice during treatment with daily 50 mg/kg FX1 versus vehicle for 10 days (n = 10, 2-tailed Mann-Whitney unpaired test). (C) Tumor burden is shown (AUC) for the same mice as in B and calculated between the initial tumor volume (t₀: 100 mm³) and the final volume at day 9 (n = 10, 2-tailed Mann-Whitney unpaired test). (D) TUNEL IHC was performed in the same mice as in B. Scale bars: 100 μm. (E) Three primary DLBCL specimens were maintained in a coculture system and treated with FX1 50 μM versus vehicle. The y axis represents viability of 2 biological replicates (represented as percentage of vehicle-treated cells) determined using annexin V/DAPI flow cytometry after 48 hours. Live cells are defined as CD20⁺CD3^– cells that are annexin V/DAPI double negative (unpaired t test, **P < 0.005). Hans and gene expression classification is shown below each sample number. (F) Combinatorial dosing of FX1 and doxorubicin is shown in the indicated cell lines, in cells treated with both agents simultaneously. The y axis represents the dose-reduction index (DRI) of 3 independent experiments, based on the DRI for both doxorubicin and FX1. (DRI > 1 = favorable combination.) Values in A and D represent mean ± SD. Values in B, C, E, and F are shown as mean ± SEM.