Frequent mutation in cancer. STAG2 encodes a subunit of the cohesin complex and is among the most mutated genes in human cancer, with bladder cancer showing one of the highest frequencies of STAG2 inactivation, occurring in up to 40% of papillary non-muscle-invasive tumors.
The cohesin complex exists in two versions: one containing STAG1 and one containing STAG2. These two variants appear to have distinct biological functions, and the mechanisms by which STAG2 loss contributes to tumor development have remained poorly understood.
Beyond chromosome segregation. Early hypotheses linked STAG2 loss to aneuploidy, but genetic analyses of bladder cancer and acute myeloid leukemia showed that STAG2-mutant tumors are actually genomically stable, pointing toward other molecular mechanisms such as chromatin organization and gene regulation.
STAG2 mutations in bladder cancer are particularly associated with activating FGFR3 mutations and low genomic instability, and among muscle-invasive tumors they tend to cluster in those with luminal differentiation, raising questions about the mechanistic link between STAG2 and urothelial identity.
Model system and experimental design. The study used RT112, a well-characterized luminal-type bladder cancer cell line with wild-type STAG2 and mutant FGFR3. STAG2 was knocked down using two independent shRNAs delivered via lentiviral infection, enabling controlled loss-of-function experiments.
Chromatin immunoprecipitation followed by deep sequencing (ChIP-Seq) was performed for STAG1, STAG2, SMC1 (a shared cohesin subunit), and multiple histone modifications (H3K4me1, H3K4me3, H3K27me3, H3K27Ac, H3K9Ac, H3K9me3) as well as CTCF to map cohesin binding and chromatin states genome-wide.
Hi-C for 3D chromatin architecture. Hi-C libraries were generated from G1-arrested RT112 cells to capture genome-wide chromatin contacts at 20 kb and 100 kb resolution, enabling analysis of compartments, topologically associating domains (TADs), and DNA loops under control and STAG2-depleted conditions.
RNA-Seq in triplicate was performed on control and STAG2-silenced cells to measure transcriptional changes. Data integration across ChIP-Seq, Hi-C, and RNA-Seq datasets allowed researchers to link cohesin binding, chromatin looping, and gene expression in a unified analysis framework.
Chromatin states were defined using ChromHMM with 10 states based on six histone marks and CTCF, and differential cohesin binding sites were categorized as common, STAG1-enriched, or STAG2-enriched using the DiffBind R package with FDR thresholds.
Three categories of cohesin binding. ChIP-Seq analysis identified 35,321 common cohesin positions, 5,007 STAG1-enriched sites, and 2,330 STAG2-enriched sites in RT112 cells. STAG2-enriched sites displayed broader peaks, suggesting higher cell-to-cell variability or greater dynamic binding at these locations.
While STAG1-enriched positions were more prevalent in intergenic regions and transcriptionally inactive chromatin marked by H3K27me3 or H3K9me3, STAG2-enriched positions showed strong overlap with promoters, exons, and 5' UTRs, and were predominantly found in transcriptionally active chromatin states including active enhancers and promoters.
Independence from CTCF. STAG2-enriched binding sites were notably depleted of CTCF occupancy compared to common and STAG1-enriched sites. Motif analysis revealed enrichment for tissue-specific transcription factor binding motifs including ASCL1 and KLF5 at STAG2-enriched sites, rather than the CTCF motif that dominated STAG1-enriched sites.
This distribution indicates functional compartmentalization between the two cohesin variants: cohesin-STAG1 predominantly occupies CTCF-anchored boundary elements, while cohesin-STAG2 preferentially localizes to active gene regulatory elements independent of CTCF, suggesting distinct roles in gene regulation versus chromatin boundary maintenance.
Stable large-scale organization. Despite efficient STAG2 protein depletion confirmed by western blot, Hi-C analysis showed that global A/B compartment structure was largely unaffected. Only 0.6-0.8% of the genome switched from A to B compartments and 0.75-1.2% switched from B to A, with compartment strength remaining comparable between control and STAG2-depleted cells.
TAD boundaries were similarly resilient to STAG2 loss. The number of TAD borders remained stable (approximately 2,400 per condition), with average conservation rates of 90.8% and 91.5% for the two shRNAs relative to control cells. TAD strength and insulation scores were also unchanged.
Ploidy unaffected. STAG2 depletion did not cause gross changes in genome ploidy, as assessed by read-count analysis of genome-wide contact matrices, consistent with the known genomic stability of STAG2-mutant bladder cancers observed in clinical samples.
These findings confirm that megabase-scale chromatin organization, including chromosomal compartments and TAD boundaries, can be maintained by cohesin-STAG1 alone, and that the functional impact of STAG2 loss is confined primarily to shorter-range chromatin interactions within TADs rather than large-scale architectural features.
Shift in interaction distances. STAG2 knockdown resulted in a significant overall increase in the genomic distances spanned by chromatin contacts. Specifically, there was a loss of short-range interactions (under 250 kb) and a compensatory increase in long-range interactions (over 1 Mb), reflecting rewiring of the chromatin looping landscape.
A total of 1,686 lost interactions and 1,881 gained interactions were identified consistently across both STAG2-targeting shRNAs. Lost interactions preferentially spanned shorter genomic distances, were depleted of CTCF motifs, and showed higher overlap with the B compartment and heterochromatin states compared to gained interactions.
Assortativity reveals spatial clustering. Chromatin assortativity analysis showed that loci involved in lost interactions were non-randomly clustered in 3D space and tended to interact with each other, reaching maximum assortativity for interactions spanning less than 1 Mb (intra-TAD scale). Gained interactions showed no such spatial clustering.
This pattern suggests that lost interactions reflect direct loss of STAG2-cohesin-mediated short-range loops, while gained interactions likely arise from redistribution of remaining STAG1-cohesin complexes to pre-existing binding sites, extending their reach to longer genomic distances in the absence of STAG2 competition.
Pairwise enrichment analysis demonstrated that STAG2-enriched cohesin binding sites preferentially interacted with other STAG2-enriched sites rather than with other cohesin complex types, indicating a degree of spatial segregation between the DNA loop networks established by different cohesin variants.
Modest but specific transcriptional changes. STAG2 knockdown resulted in statistically significant changes in a subset of genes (510 with sh1 and 438 with sh2), with roughly equal numbers of up- and down-regulated transcripts. Gene expression changes were positively correlated between the two shRNAs, and 20-32% of significantly altered genes were shared between both knockdown conditions.
Gene Set Enrichment Analysis (GSEA) revealed consistent and significant up-regulation of basal and squamous gene signatures, and significant down-regulation of luminal papillary gene signatures, in STAG2-silenced cells. This mirrors the transcriptional shift observed when comparing STAG2-low to STAG2-high human bladder tumor samples from the UROMOL cohort of 476 non-muscle-invasive cases.
Clinical validation. Genes down-regulated in STAG2-silenced RT112 cells significantly overlapped with genes differentially expressed in STAG2-low versus STAG2-high tumors from the UROMOL cohort, strongly validating the RT112 cell model as an appropriate system to study STAG2 function in urothelial biology.
Specific basal marker genes including TNC and COL17A1 showed loss of promoter-associated chromatin contacts concomitant with their transcriptional up-regulation upon STAG2 silencing, providing mechanistic evidence linking contact rewiring to gene derepression.
Promoter contacts control silencing. Integration of Hi-C and RNA-Seq data revealed that lost interactions predominantly overlapped promoters and gene bodies of genes expressed at low levels in control cells. Critically, loss of DNA contacts at promoters (but not gene bodies) was associated with statistically significant up-regulation of gene expression.
Genes whose promoters lost contacts upon STAG2 depletion tended to be low-expressed Polycomb target genes. Two exemplar basal genes, TNC and COL17A1, showed clear loss of promoter-associated chromatin interactions alongside transcriptional derepression, linking the structural change directly to gene activation.
Polycomb domain compaction. The contacts lost upon STAG2 depletion preferentially overlapped the B compartment and constitutive or facultative heterochromatin domains, suggesting that STAG2-cohesin normally contributes to compaction of silent chromatin regions through DNA looping, helping to maintain Polycomb-mediated repression of lineage-specifying genes.
Motif analysis identified a collection of transcription factor binding motifs enriched in both differentially looping regions and up-regulated genes, suggesting partial mechanistic coupling between chromatin loop rewiring and transcriptional derepression, although some expression changes likely reflect secondary epigenetic events rather than direct loop loss.
Proposed tumor suppressor mechanism. The findings support a model in which STAG2-cohesin maintains DNA looping within silent chromatin domains, particularly at Polycomb-repressed lineage-specifying loci. Loss of STAG2 leads to decompaction of these domains and derepression of basal differentiation genes, destabilizing luminal urothelial identity.
Two transcription factors enriched at STAG2-only binding sites, TFAP2A and KLF5, show tissue-restricted expression in squamous epithelia including skin and esophagus. TFAP2A is repressed by PPARg in luminal bladder tumors and up-regulated in basal tumors, while KLF4-driven regulons are selectively activated in basal-squamous subtypes, providing mechanistic links between STAG2 occupancy and basal program control.
Context-dependent tumor suppression. Although STAG2 mutations in bladder cancer are most common in papillary, luminal-type tumors, STAG2 knockdown activates the basal program. This apparent paradox is consistent with clinical observations that STAG2-mutant muscle-invasive tumors and certain non-muscle-invasive class 3 tumors harbor enriched basal marker expression despite overall luminal features.
The study proposes that STAG2 acts as a tumor suppressor in the urothelium by establishing and maintaining Polycomb-mediated repression of basal differentiation genes through DNA looping, and that loss of this function shifts cells toward a basal transcriptional state that may facilitate tumor progression in specific contexts.
New understanding of STAG2 function. This study provides the first genome-wide analysis of STAG2 in urothelial cells, filling a major gap given that bladder cancer has the highest frequency of STAG2 mutations among human cancers. The results demonstrate that cohesin-STAG2 has distinct genomic distributions and functional roles compared to cohesin-STAG1.
The functional compartmentalization between STAG1 and STAG2 means that STAG1 can maintain large-scale architecture (compartments and TAD boundaries) but cannot compensate for STAG2-mediated short-range promoter interactions within silent chromatin domains, explaining why STAG2 loss has specific transcriptional consequences despite preserved global architecture.
Therapeutic relevance. Understanding that STAG2 loss specifically derepresses basal lineage-specifying genes through chromatin loop disruption opens potential therapeutic avenues. Restoring or mimicking STAG2-dependent chromatin compaction, or targeting transcriptional effectors of the basal program activated by STAG2 loss, could represent strategies for treating STAG2-mutant bladder cancers.
The validation of RT112 cells as a faithful model of STAG2 function in bladder cancer, and the concordance with clinical UROMOL cohort data, provides a robust experimental framework for future mechanistic and therapeutic studies targeting the cohesin-STAG2 axis in urothelial malignancy.