A Bivalent Molecular Glue Linking Lysine Acetyltransferases to Oncogene-induced Cell Death

Abstract

Developing cancer therapies that induce robust death of the malignant cell is critical to prevent relapse. Highly effective strategies, such as immunotherapy, exemplify this observation. Here we provide the structural and molecular underpinnings for an approach that leverages chemical induced proximity to produce specific cell killing of diffuse large B cell lymphoma, the most common non-Hodgkin's lymphoma. We develop KAT-TCIPs (lysine acetyltransferase transcriptional/epigenetic chemical inducers of proximity) that redirect p300 and CBP to activate programmed cell death genes normally repressed by the oncogenic driver, BCL6. Acute treatment rapidly reprograms the epigenome to initiate apoptosis and repress c-MYC. The crystal structure of the chemically induced p300-BCL6 complex reveals how chance interactions between the two proteins can be systematically exploited to produce the exquisite potency and selectivity of KAT-TCIPs. Thus, the malignant function of an oncogenic driver can be co-opted to activate robust cell death, with implications for precision epigenetic therapies.

Keywords: induced proximity; lymphoma; lysine acetyltransferases; transcription.

Figure 1.. Design and Activity of KAT-TCIPs

(A) KAT-TCIPs targeting p300/CBP are designed to de-repress cell death and cell cycle arrest pathways controlled by the transcription factor BCL6 in DLCBL cells. (B) Structure of TCIP3, NEG1, which retains binding to p300/CBP, and NEG2, which retains binding to BCL6. (C) Activation of a BCL6-repressed GFP reporter construct integrated into K422 cells after compound treatment for 24 h; 3 biological replicates, mean ± s.e.m. (D) Cell-killing potencies of compounds after 72 h of treatment in SUDHL5 cells; 3–5 biological replicates, mean ± s.e.m. (E) IC₅₀ values (nM) of antiproliferation after 7 days of compound treatment in DLBCL and leukemia cells plotted against BCL6 expression in transcripts/million (TPM); 3–4 biological replicates, mean ± s.d.; R of BCL6 TPM vs AUC computed by Pearson’s correlation and P-value computed by two-sided Student’s t-test.

Figure 2.. TCIP3 Kills Cells Via Chemically Induced Proximity (CIP)

(A) Measurement of cell viability after competitive titration of constant 1 nM TCIP3 with BCL6^BTB domain inhibitors (BI-3812, GSK137, or CCT373566a), or (B) the p300/CBP bromodomain inhibitor GNE-781; cells were treated simultaneously with TCIP3 and the inhibitor or DMSO for 72 h; 3 biological replicates, mean ± s.e.m. (C) p300 immunoprecipitation-mass spectrometry (IP-MS) from SUDHL5 cells treated for 2 h with 1 nM TCIP3; plotted with cutoffs of log₂(fold change)| ≥ 2 and P ≤ 0.01; 3 biological replicates. For (C) and (D), P-values computed by a moderated t-test. (D) FLAG IP-MS from genomic knock-in FLAG-tagged BCL6 SUDHL5 cells treated with 1 nM TCIP3 for 2 h plotted with cutoffs of log₂(fold change) ≥ 2 and P ≤ 0.01. Proteins that did not contain peptides for DMSO but contained peptides upon TCIP3 treatment were imputed; 3 biological replicates.

Figure 3.. TCIP3 Redistributes p300/CBP Activity to BCL6 and Proximal Chromatin

(A) Acetylated lysine (K-ac)- IP and western blot (WB) for BCL6 after 1 h of TCIP3 in SUDHL5 cells; representative of three biological replicates. (B) Changes in histone 3 lysine 27 and histone 2B lysine 20 acetylation (H3K27ac and H2BK20ac) as measured by chromatin immunoprecipitation sequencing (ChIP-seq) after 1 h of 1 nM TCIP3 in SUDHL5 cells; significant: adj. P ≤ 0.05 and up, log₂(fold change) ≥ 0.5; down, log₂(fold change) ≤ −0.5; 2 biological replicates, P-values computed by two-sided Wald test and adjusted by multiple comparisons by Benjamini-Hochberg. (C) Enrichment of predicted transcription factor (TF) binding in gained H3K27ac and H2BK20ac peaks calculated by overlap with public ChIP-seq datasets from blood-lineage cells; full enrichment data in Supplemental Table 2; P-values computed by two-sided Fisher’s exact test and adjusted for multiple comparisons by Benjamini-Hochberg. (D) Induction of H2BK20ac and H3K27ac with time at the promoter of the BCL6 target gene ARID3B; BCL6 track is CUT&RUN in untreated SUDHL5 cells, tracks merged from two biological replicates and sequence-depth normalized and, for histone acetylation ChIP-seq, also input-subtracted. (E) Overlap of gained and lost H3K27ac peaks after 1 h of 1 nM TCIP3 with annotated enhancers and super-enhancers in SUDHL5 cells. (F) Changes in H3K27ac at annotated enhancers and super-enhancers after 2 h of 1 nM TCIP3; significant: adj. P ≤ 0.05 and up, log₂(fold change) ≥ 0.5; down, log₂(fold change) ≤ −0.5; 2 biological replicates, P-values computed by two-sided Wald test and adjusted by multiple comparisons by Benjamini-Hochberg. (G) Western blot of SUDHL5 cells treated with 1 nM TCIP3 for the indicated time periods; representative of 3 biological replicates. (H) Comparison of H3K27ac loading at differential regions at 15 min, 1 h, and 2 h of 1 nM TCIP3; differential regions defined as in (A), P-values adjusted by Tukey’s test after type II analysis of variance (ANOVA). (I) Gene set enrichment analysis of ranked log₂(fold change) in gene expression measured by RNA-sequencing (RNA-seq) after 1 nM TCIP3 in SUDHL5 cells; only all gene sets adj. P ≤ 0.05 at all timepoints displayed, positive normalized enrichment scores (NES) indicate gene sets enriched in TCIP3- induced genes while negative NES scores indicate sets enriched in decreased genes, P-values computed by permutation and adjusted for multiple comparisons by Benjamini-Hochberg. (J) Model of how TCIP3 redistributes p300/CBP from super-enhancers to BCL6.

Figure 4.. Activation of Apoptotic Signaling by TCIP3

(A) Comparison of whole-proteome profiling of SUDHL5 cells treated with 10 nM TCIP3 or 250 nM of the p300/CBP degrader dCBP-1 for 24 h; proteins labeled change statistically significantly (adj. P ≤ 0.05); 3 biological replicates; P-values computed using a moderated t-test and adjusted by Benjamini-Hochberg; R computed by Pearson’s correlation and P-value computed by two-sided Student’s t-test. (B) Signaling pathways (MSigDB Hallmark 2020) enriched in significantly increased proteins (adj. P < 0.05, log₂(foldchange) > 1) after 24 h treatment of 10 nM TCIP3 in SUDHL5. (C) Correlation of whole-proteome profiling of SUDHL5 cells treated with 10 nM TCIP3 for 24 h to bulk transcriptomics (RNA-seq) performed on SUDHL5 cells treated with 1 nM TCIP3 for 4 h; only genes whose transcripts that change significantly (adj. P ≤ 0.05) were analyzed; 3 biological replicates; P-value for RNA-seq computed by two-sided Wald test and adjusted by Benjamini-Hochberg; R computed by Pearson’s correlation and P-value for Pearson’s correlation computed by two-sided Student’s t-test. (D) Western blot of pro-apoptotic proteins in SUDHL5 cells treated with indicated compounds and doses after 48 h; representative of 3 biological replicates. (E) Quantification of BBC3/PUMA protein levels from (D); P-values computed by Fisher’s LSD test after ANOVA; **: P < 0.01, *: P < 0.05; only the comparisons to DMSO were computed. 3 biological replicates, mean ± s.d. (F) Quantification of Annexin V-positive SUDHL5 cells treated with indicated compounds and doses at 24, 48, or 72 h; P-values adjusted by Tukey’s test after ANOVA. The following comparisons were significantly different from each other at 72 h (****: adj. P < 0.0001): 10 nM TCIP3 vs DMSO, 10 nM NEG1, and 10 nM NEG2 72 h; 10 nM TCIP1 vs DMSO, 10 nM NEG1, and 10 nM NEG2; 250 nM dCBP1 vs DMSO, 10 nm NEG1 (adj. P = 0.0023), and 10 nM NEG2 (adj. P = 0.0008). The following comparisons were significant at 48 h (****: adj. P < 0.0001): 10 nM TCIP3 vs DMSO, 10 nM TCIP1 vs DMSO. The following comparisons were significant at 24 h: 10 nM TCIP3 vs DMSO (****: adj. P < 0.0001), 10 nM TCIP1 vs DMSO (****: adj. P < 0.0001), 10 nM TCIP3 vs 10 nM TCIP1 (*: adj. P = 0.0187). There were no significant differences between any other comparisons (adj. P ≥ 0.05). 3–12 biological replicates, mean ± s.e.m. (G) Concurrent analysis of apoptosis and cell cycle effects of TCIPs by terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) and total DNA content co-staining of SUDHL5 cells treated with DMSO, 10 nM TCIP1, or 10 nM TCIP3 for 24, 48, or 72 h; mean ± s.e.m; 4 biological replicates; P-values computed by Fisher’s LSD test after analysis of variance (ANOVA); *:. P < 0.05; ns: not significant, P > 0.05; only the comparisons to DMSO at each timepoint were computed. (H) Quantification of percentage of fixed SUDHL5 cells in G1, S Phase, or G2/M phase after 24 h treatment with indicated compounds; 3 biological replicates, mean ± s.e.m; P-values adjusted by Tukey’s test after analysis of variance (ANOVA) on cells in either S, G0/G1, or G2/M phase. For G2/M cells, there were no significant differences (adj. P ≥ 0.05); for S and G0/G1 phase cells, TCIP3 was significantly different from all other treatments (adj. P < 0.05). No other comparisons were significant.

Figure 5.. Rapid and Potent Reduction in c-MYC is Sufficient for Cell Cycle Arrest

(A) RNA-seq of SUDHL5 cells treated with 1 nM TCIP3 for 2 h; colored are differential genes defined by adj. P ≤ 0.05 and |log₂(fold change)| ≥ 0.5, P-values computed by two-sided Wald test and adjusted for multiple comparisons by Benjamini-Hochberg; 3 biological replicates. (B) All peptides of c-MYC detected by global proteomics after 10 nM TCIP3, NEG1, and NEG2 treatment of SUDHL5 cells treated for 24 h; intensities are mean of 3 biological replicates. Lines represent median and interquartile range. (C) Time-course of c-MYC transcripts in SUDHL5 cells treated with 1 nM of compounds normalized to GAPDH and DMSO treatment as quantified through reverse transcription quantitative PCR (RT-qPCR); 3 biological replicates, mean ± s.e.m. (D) Western blot of c-MYC protein in SUDHL5 cells treated with TCIP3 for 4 h; blot representative of 2 biological replicates. (E) Quantification of c-MYC protein normalized to GAPDH and DMSO levels from (C). (F) Measurement of c-MYC mRNA in SUDHL5 cells (normalized to GAPDH and DMSO treatment) by RT-qPCR after competitive titration of constant 10 nM TCIP3 with 500 nM of the BCL6^BTB domain inhibitor BI-3812 or 500 nM of the p300/CBP bromodomain inhibitor GNE-781; cells were treated simultaneously with TCIP3 and the inhibitor or DMSO for 1 h; effects of co-treatment of DMSO and inhibitors shown for comparison; 3 biological replicates, mean ± s.e.m.; P-values adjusted by Tukey’s test after analysis of variance (ANOVA). Only comparisons of co-treatments to TCIP3 were computed. (G) c-MYC transcripts in DLBCL and leukemia cells with varying BCL6 expression (TPM: transcripts/million) treated with 1 nM of TCIP3 normalized to GAPDH and DMSO treatment as quantified through RT-qPCR; 3 biological replicates, mean ± s.e.m. SUDHL5 data is from panel B. (H) Doxycycline-inducible overexpression of c-MYC in SUDHL5 cells. (I) S phase cells of SUDHL5^{TRE−3xFLAG-MYC} cells treated with or without 1 μg/mL doxycycline dissolved in ethanol or vehicle 24 h prior to 24 h treatment with 10 nM TCIP3 or DMSO; 3 biological replicates, P-values computed by two-tailed ratio paired Students’ t-test. (J) Viability of SUDHL5^{TRE−3xFLAG-MYC} cells treated with 1 μg/mL doxycycline dissolved in ethanol or vehicle 24 h prior to 72 h treatment with TCIP3; mean ± s.e.m., 3 biological replicates.

Figure 6.. The Crystal Structure and Cooperativity of the Ternary Complex

(A) Domain structure of full-length p300/CBP and BCL6. The highlighted p300 bromodomain (BD) and BCL6 BTB domain (BTB) were crystallized in complex with MNN-02–155. (B) Co-crystal structure of the ternary complex formed by MNN-02–155, BCL6^BTB, and p300^BD. One dimer of BCL6 is bound to two molecules of MNN-02–155, each of which engages one protomer of p300^BD. (C) Neo-hydrogen bond formed between MNN-02–155 and the backbone of P300G¹⁰⁸⁵. (D) Neo-protein-protein interactions formed at the interface of p300^BD and BCL6^BTB mediated by MNN-02–155 binding. (E) Binary TR-FRET displacement assay assessing the binding of 6x-His-P300^BD to MNN-02–155. MNN-02–155 was titrated into a biochemical complex of terbium-labeled 6x-His-p300^BD pre-incubated with MNN-06–112. Parallel experiments were performed by preincubating with BCL6 at concentrations exceeding its dissociation constant (>>> Kd). Binding affinities of p300^WT, p300^Q1083A, and p300^Q1083R to MNN-02–155 were evaluated in the absence and presence of BCL6 and a cooperativity constant was calculated (α = K_d binary/K_d ternary). n = 3 independent experiments, mean. (F) K_d calculations for each binding event, based on n = 2–3 independent experiments, mean ± s.e.m. α was calculated for both p300 and BCL6 (α = K_d binary/K_d ternary).