Fig 2: STAG1 D797 residue is essential for binding to the cohesin ring, chromatin localization, sister chromatid cohesion, and cell viability in STAG2-deficient cells.(A) HCT 116 STAG2-502c4 cells were stably transduced with FLAG-tagged, siRNA-resistant wild-type (wt) and mutant STAG1 or empty vector control transgenes, and nuclear expression was assessed by FLAG immunofluorescence. Scale bar: 20 µm. (B) Capillary immunodetection assay was used to detect FLAG–STAG1 protein levels (see Fig S6A for quantifications). (C) Protein extracts from stably expressing FLAG–STAG1 wild-type and mutant cells were subjected to FLAG immunoprecipitation and analyzed for co-precipitation of cohesin ring members by capillary immunodetection assay (see Fig S6B for quantifications). (D) Cytoplasmic extract (CE), nuclear-soluble (NS), and chromatin-bound (CB) fractions were obtained to determine the subcellular distribution of FLAG–STAG1 wild-type and mutant protein by capillary immunodetection assay (for quantifications, see Fig S6C). (E) HCT 116 cells were transfected with nontarget control (NTC) and STAG1 siRNA duplexes, and immunofluorescence analysis was performed 72 h after transfection to determine the mitotic index by scoring the fraction of histone H3 phosphoSer10-positive (H3pS10+) cells (n = 437 cells, triangles denote values of two independent experiments and error bars denote SD). (F) Prometaphase chromosome spreads were prepared 72 h after transfection of cells with NTC control or STAG1 siRNA duplexes. The status of sister chromatid cohesion of individual Giemsa spreads was categorized into normal, partial loss of cohesion, or single chromatid phenotypes (n = 100 spreads, error bars denote SD of two independent experiments with each two technical replicates). Scale bar: 10 µm. (G) Cell viability was assessed 5–6 d after siRNA transfection using a metabolic assay, and viability was normalized to NTC control (n = 4 biological repeats, error bars denote SD).Source data are available for this figure.

Fig 3: STAG1 and STAG2 display partially distinct and overlapping roles in gene expression. a Overlap of differentially expressed genes (DEGs) in Stag1-/- cells compared to wild-type cells and Stag2-/- cells compared to wild-type cells. Cells were treated with siGLO as a transfection control. b Correlation plot of log2 fold changes of DEGs. DEGs specific to Stag1-/- are shown in red, DEGS specific to Stag2-/- are shown in blue, and DEGs sensitive to the loss of either STAG (common) are shown in purple. c Heatmap of log2 fold changes for a combined list of DEGs in Stag1-/- and Stag2-/- cells. Heatmap of STAG1 and STAG2 ChIP-seq signal in wild-type cells at the promoters of the combined list of DEGs. d Average signal plots for STAG1, STAG2, and RAD21 ChIP-seq signal in wild-type cells at Stag1-/- specific, common, and Stag2-/- specific DEG promoters. e Gene Ontology (GO) terms for biological processes that are Stag1-/- specific, Stag2-/- specific, and common to both knockouts. f Violin plot depicting log2 fold changes for all DEGs, and those within Super-enhancer Domains and Polycomb Domains for Stag1-/- and Stag2-/- cells. Dotted lines indicate the mean. Asterisks indicate significant differences between groups (****p < 0.0001, **p < 0.01). g Bar graphs with log2 fold change of expression of cell identity genes, including those that represent pluripotency (Pou5f1, Sox2, Nanog), ectoderm (Pax6 and Nestin), and endoderm lineages (Gata6 and Sox17) in Stag1-/- and Stag2-/- cells. Asterisks indicate significant differences from wild-type cells (padj < 0.01)

Fig 4: Structure of STAG1 hotspot regions in complex with RAD21 peptides.(A) Cartoon diagram of the overall structure of STAG1 N-terminal region (pink), and central region (cyan) superposed on to the STAG2 structure (gray with semi transparency), with RAD21 peptide regions in green and heat repeats numbered. (B) Close-up view of the interaction between RAD21 and STAG1 N-terminal region (sites I and II), with hydrogen bonds shown as dashed lines and key interacting residues shown in stick format. (B, C) Schematic view of the interaction between RAD21 residues 322-221 and the N-terminal region of STAG1; the orientation and coloring correspond roughly to what is shown in panel (B). Hydrogen bonds are shown as dashed lines and van der Waals contacts are shown as red-spoked arcs. (D) Close-up view of the interaction between RAD21 and STAG1 central region (site IV), with hydrogen bonds shown as dashed lines and key interacting residues shown in stick format. The first 12 residues of the RAD21 peptide, which form a further short section of a-helix in the STAG2 structure are disordered in the electron density and have not been included in the model. (D, E) Schematic view of the key interactions between RAD21 and STAG1 central region, highlighting the interacting region spanning RAD21 residues 375–382, the orientation and coloring correspond roughly to what is shown in panel (D). Hydrogen bonds are shown as dashed lines, and van der Waals contacts are shown as red-spoked arcs. (F) Comparison of STAG1 central region WT versus D797A mutant binding to a RAD21356–395 peptide by surface plasmon resonance. Biotinylated RAD21356–395 was immobilized on the chip, and increasing concentrations of STAG1 were injected in a single cycle kinetic experiment, a clear dose-dependent binding response is observed for the wild-type protein (red curve, with fit to a 1:1 binding model shown as black line), whereas only small nonspecific responses are seen for the D797A variant (blue curve). (G) Analysis of the STAG1 central region–RAD21356–395 interaction on surface plasmon resonance using equilibrium analysis. Increasing concentrations of STAG1 were injected over the sensor surface, and the response at equilibrium is fit to a dose–response curve (inset) with an apparent Kd of 148 ± 22 nM.

Fig 5: ChIP-seq analysis of the genomic distribution of STAG1, STAG2 and SMC3ac.(A) Coverage-corrected Hi-C contact matrix of the 95.2–95.7 Mb region of HeLa chromosome 1. ChIP-seq signals for EGFP-STAG1, STAG2-EGFP, SMC3, SMC3ac and CTCF are shown below the Hi-C contact matrix. STAG1 and STAG2 immunoprecipitation was performed using anti-GFP antibodies. Note that the SMC3ac signal is decreased following ESCO1 depletion, confirming the specificity of the anti-SMC3ac antibody. (B) Venn diagram illustrating the genome-wide co-localization between STAG1 and STAG2 ChIP-seq signals. STAG1/2 immunoprecipitations from EGFP-STAG1 and STAG2-EGFP cells were performed using anti-GFP antibodies. Average read density plots for EGFP-STAG1 and STAG2-EGFP at SMC3ac and SMC3 sites is shown on the right. (C) Pie charts showing the distribution of STAG1-only (left), STAG2-only (middle) and STAG1/2 common ChIP-seq sites (right) in G1 relative to TSS and active enhancers. (D) Venn diagrams showing the distribution of STAG1-only (left), STAG2-only (middle) and STAG1/2 common ChIP-seq sites (right) relative to CTCF ChIP-seq sites. (E) Venn diagrams illustrating the genome-wide overlap between SMC3ac and STAG1/2 common sites, SMC3 and CTCF ChIP-seq sites respectively. (F) Venn diagrams showing the distribution of STAG1-only (left), STAG2-only (middle) and STAG1/2 common ChIP-seq sites (right) relative to SMC3ac ChIP-seq sites. (G) Venn diagrams illustrating the genome-wide overlap between the loop anchors determined using the Hi-C map shown in (A) and SMC3 (left panel) and SMC3Ac (right panel) ChIP-seq binding sites.