Fig 2: Compounds 21 and 22 bind the CBX8-CDs and form a DNA-ternary complex(A) Full 1H-15N-HSQC overlays for 15N-CD upon addition of compound 21 (left) or 22 (right) in the absence and the presence of 11bp DNA.(B) Normalized CSP (?d) between 11bp DNA-bound (top), 11bp DNA and 21-bound (middle), and 11bp DNA and 22-bound (bottom) spectra are plotted against CBX8 residue number.(C) Residues with significant CSPs upon addition of 11bp DNA only, 11bp DNA with compound 21, and 11bp DNA with compound 22 plotted onto a cartoon and surface representation of the CDs and colored red.

Fig 3: Structure-based design of selective CBX8 compounds(A) Reported CBX7 compounds and their in vitro potency in an AlphaScreen competition assay.(B) Comparison of the alanine binding pocket size of CBX7 (top) and CBX8 (bottom).(C) Comparison of the aromatic cage region of CBX7 (top) and CBX8 (bottom).(D) The binding mode of the N-terminal tert-butyl benzoyl cap of UNC3866 in CBX8.(E) The binding mode of the UNC3866 leucine side chain in CBX8. UNC3866 is displayed in ball-and-stick with carbons colored green.(F) SAR strategy for new compounds (red, Ala position; blue, N-cap position; green, Leu position; magenta, lysine mimetics position).

Fig 4: UNC7040 impairs DLBCL cell proliferation by blocking cPRC1 targeting via CBX8(A) Effects of treatment with PRC2 (UNC1999, EED226) or CBX8 (UNC7040) antagonists on SUDHL4 cell proliferation relative to vehicle control. UNC2400, UNC5679, and UNC7263 served as negative controls for UNC1999, EED226, and UNC7040, respectively.(B) Comparison of individual and combined treatment with UNC7040 and EED226 on SUDHL4 cell proliferation. Data in (A) and (B) represent an average of four to six replicates ± SD.(C) Genomic screenshots of CBX8 and RING1B cChIP-seq signals at four selected PRC2 target genes in SUDHL4 cells treated for 6 h with DMSO, UNC7263, or UNC7040. Bottom track shows H3K27me3 signal in untreated cells.(D) Heatmaps display cChIP-seq enrichment of CBX8 and RING1B in SUDHL4 cells treated for 6 h with DMSO, UNC7263 at 40 µM, or UNC7040 at 40 µM. Also shown are signals of H3K27me3 and H3K27ac in untreated cells. The signals from two independent cChIP-seq experiments are centered around RING1B peaks ±5 kb and plotted as spike-in normalized mapped read counts. Genomic intervals were separated into 10 equal groups and sorted based on log2 fold change in CBX8 occupancy upon UNC7040 treatment relative to DMSO.(E) Pie charts display relative distribution of genomic features in selected cChIP-seq groups based on (D).(F) Boxplots compare cumulative CBX8 signal (±0.5 kb around RING1B peaks) for selected groups in untreated and treated SUDHL4 cells. Significance (p value) was calculated using Wilcoxon signed-rank test (*p = 0.05, **p = 0.01, ***p = 0.001, ****p = 0.0001; ns, not significant).(G) Volcano plots show gene expression changes of UNC7263-treated (left) or UNC7040-treated SUDHL4 cells relative to DMSO. Numbers in represent repressed (blue) and upregulated (red) genes. Differential gene expression (padj < 0.05; LFC = <–0.5 and >0.5, respectively) was calculated from three independent replicates. Highlighted are genes associated with B cell maturation.(H) For functional analysis of canonical CBX8-PRC1 targets, differential peaks resulting from UNC7040 treatment were analyzed by GREAT (v.4.0.4) using the nearest gene within 100 kb to generate enrichment of biological processes.

Fig 5: Analyses of probe specificity for CD proteins and effect on CBX8 DNA binding(A) The binding profile of 34 (UNC7040), 35 (UNC7042), 38 (UNC7263), and 39 (UNC7045) for CBXs, CDYL2, and MPP8 chromodomains. For CBX2 and CBX4, the top 2 concentrations were discarded for technical issues in TR-FRET assay. Data shown are mean ± SD, n = 6.(B) The side chain of Tyr39 in CBX8 faces out from the aromatic cage making it more expansive than that of CDYL2 (right). The side chain of Glu36 in CYDL2 facing into the aromatic cage makes it more compact (left).(C) ChIP-qPCR analysis shows relative enrichments of CBX proteins at endogenous Polycomb target genes in CBX8-expressing mESCs. IAP serves as negative control locus. Data are mean ± SD (error bars) of two independent experiments.(D) TR-FRET ternary complex assay results with WT CBX8, biotin-dsDNA, and compounds or DMSO. Data are mean ± SD (error bars) of two independent experiments, n = 6.(E) TR-FRET ternary complex assay results with CBX8 R19A mutant, biotin-dsDNA, and compounds or DMSO. Data are mean ± SD (error bars) of three independent experiments, n = 9.