Fig 2: ATRX and the ZNF274/TRIM28/SETDB1 complex bind to ZNF genes with an atypical chromatin signature and distinctive genomic and epigenetic features. (A) Average K562 enrichment ChIP-seq profiles of ATRX, H3K9me3 and H3K36me3 at ZNFs classified by their ATRX content. Class I contains high levels of ATRX (n = 91), Class II contains medium to low levels of ATRX (n = 303) and Class III is devoid of ATRX enrichment (n = 342). (B) Spearman correlation heatmap of K562 ChIP-seq signal at ZNF Class I genes (left) and ZNF Class III genes (right). (C) Left: Distribution of genetic features among ZNF classes (sorted from high to low ATRX enrichment from top to bottom). Dashed lines show separation of the 3 classes. Colors represent presence of KRAB domains (black), number of zinc finger motifs (pink), G content at the C-terminal ZNF region (last 3 kb of the gene) (gray) and presence of sequences predicted to form G-quadruplexes (brown). RNA-seq bar shows the Z score of the normalized RPKM signal (log2(RPKM+1)) in K562; red: high expression signal and blue = low expression signal. For statistical tests between the classes see Table S3. Right: Box plots displaying the number of ZNF motifs, G-content at the ZNF region and RNA-seq values in K562 per ZNF Class. Asterisks show significant differences (P-value < 1 × 10-4). (D) Metagene analysis of ChIP-seq enrichment over input profiles at ZNF gene bodies ± 1 kb. (E) Spearman correlation heatmaps between ChIP-seq profiles genome-wide (left) and at ZNF genes (right). Black boxes indicate the significant correlations. (F) Immunoblots for endogenous ATRX Co-IP of chromatin bound proteins in K562 cells after pulldown with IgG or ATRX antibody. DAXX used as a positive control for the ATRX IP.

Fig 3: ZNF274 KO reduces ATRX and H3K9me3 levels at ZNFs. (A) ZNF274, (B) ATRX and (C) H3K9me3 ChIP-qPCR in K562 ZNF274 KO and K562 double ZNF274/ATRX KO at ZNF genes. Single ATRX KO2 cells are used for the H3K9me3 ChIP for comparison. In all graphs, the bars represent the average of at least 2 independent biological replicates. Error bars depict SEM. Results of statistical comparisons in Table S3. A non-specific sgRNA (random) used as control (see Table S7). (D) Chromatin immunoblot of ?H2A.X in control (Rnd) and ZNF KO K562 cells. Histones used as loading control. (E) Representative K562 cell cycle profiles of control (random), ATRX and ZNF274 single and double KO assessed by BrdU/PI staining. n = 6 biological replicates. (F) Graph depicting quantifications of (E). The bars show the average % of cells in each phase, error bars depict SEM. Asterisks show significant changes compared to the control (*P-value < 0.05; **P-value < 0.01).

Fig 4: Suppressed innate immune response to HCMV infection in ATRX knockdown HFFs.(A) ATRX knockdown efficiency in siATRX cells was assessed by Western blotting; ß-actin served as an internal loading control. (B) For further characterization of the siATRX cells, ATRX and its interaction partner Daxx were immunostained. 4’,6-diamidino-2-phenylindole (DAPI) was counterstained to visualize cellular nuclei. (C) To determine ATRX mRNA levels, total RNAs were isolated from ATRX knockdown HFFs (siATRX) and respective control HFFs (siC) and RT-qPCR was performed. (D+E) ATRX knockdown HFFs and respective control HFFs were either mock-infected or infected with the laboratory HCMV strain AD169 (MOI of 1). (D) At 6 and 8 hpi, cells were harvested for Western blot analyses to determine IRF3 phosphorylation; ß-actin served as an internal loading control. Signal intensities were quantified relative to infected siC cells at 6 hpi (lane 2). (E) Additionally, total RNAs were isolated at 8 hpi and RT-qPCR was performed to determine IFNB1 induction. (C+E) Depicted values were calculated from triplicates relative to (C) untreated or (E) mock-infected siC cells using GAPDH as a housekeeping gene and are shown as mean ± SD. One out of (C) four or (E) two independent experiments is shown. Statistical analysis was performed with respective ?Cq-values using a student’s t-test (unpaired, two-tailed); *p<0.05, **p<0.01. (A+B+D) The following antibodies were used: (A+D) anti-ATRX (39-f), anti-ß-actin (AC-15); (D) anti-phospho-IRF3 (Ser386) (EPR2346), anti-IRF3 (D6I4C); (B) anti-ATRX (D-5), anti-Daxx (M112).

Fig 5: ATRX status correlates with PML expression levels. (A) ATRX and PML protein expression was analyzed by means of western blot in the indicated panel of cell lines. (B) PML NBs per cell (left plot) and PML fluorescence intensity within PML NBs (center plot) were quantitated using automated imaging. PML protein expression was also quantitated by western blot and plotted relative to actin and normalized to the JFCF-6/T.1M cell line (right plot). Each data point represents a cell line, >200 nuclei counted per cell line, mean±s.e.m., n=3 independent experiments. Cell lines are grouped by ATRX status; **P<0.01, Mann–Whitney test. (C) HT1080 cells were treated with control siRNA (siNC) or siATRX for the time indicated, and western blotting performed at the indicated timepoints to evaluate expression levels of ATRX and PML. (D) Quantitation of ATRX and PML expression subsequent to siNC or siATRX transfection, as shown in C. Expression was normalized to actin, and then to the siNC-treated sample at each time point. Data are expressed as the mean±s.e.m. of three biologic replicates; *P<0.05, **P<0.01, paired two-tailed t-test.