Enhancer-focused high-throughput screening of genetic and epigenetic variants in CRC
Figure 1 – schematic representation of the pipeline (i) candidate selection, (ii) variants screening, (iii) functional validation. (Created with Biorender.com)
Abstract
Colorectal cancer (CRC) is one of the most prevalent malignancies worldwide, yet the functional impact of non-coding variants on enhancer activity remains largely underexplored. This review presents the integrated pipeline developed by Chen and colleagues to systematically screen genetic and epigenetic variants for their impact on enhancer activity in primary and metastatic CRC cell lines. The results obtained provide promising biomarkers and potential targets for new treatments.
Review
CRC as a global burden
Colorectal cancer (CRC) is one of the most prevalent cancers worldwide, with approximately 1.9 million new cases diagnosed in 2022, as reported by the World Health Organization1. Due to its significant clinical burden, considerable research effort has been directed toward the characterization of genetic and epigenetic variants in coding sequences2. However, genome-wide association studies (GWAS) have revealed that the majority of disease-associated variants map to non-coding regulatory regions, highlighting the critical role of enhancer dysregulation in CRC pathogenesis3. Although studies focusing on non-coding regions have increased in recent years4,5, a comprehensive high-throughput approach capable of systematically screening variants across enhancer regions, and particularly in the metastatic context, is still lacking. This study therefore presents a broad-spectrum pipeline designed to screen and functionally characterize genetic and epigenetic variants within enhancer elements in both primary and metastatic CRC6.
A new high-throughput approach
To investigate the functional impact of genetic and epigenetic variants in CRC, Chen et al.6 developed a comprehensive high-throughput pipeline comprising three sequential steps: candidate selection, variant screening, and functional validation (Figure 1).
Candidate selection was performed through in silico analysis. ATAC-seq (profiling accessible chromatin) and H3K27ac ChIP-seq (marking active enhancer) data from the ENCODE database were integrated to define the putative active enhancers in CRC cells. For SNP candidate selection on these regions, the 1000 Genomes Project dataset was analyzed, and only common variants present in more than 5% of both European and East Asian populations were considered. In parallel, CRC-associated loci retrieved from the GWAS Catalog were integrated with eQTL data to prioritize variants linked to gene expression changes. For CpG candidate selection, sites located within ±60 bp of H3K27ac peak summits were included to ensure systematic coverage of active enhancer regions, while the Illumina 450K BeadChip dataset was used to identify sites showing at least 30% of differential methylation between tumor and normal tissue (|Δβ| > 0.3, FDR < 0.05).
Variant screening was carried out using SNP-STARR-seq and Methyl-STARR-seq, to quantify the enhancer performance. For SNP-STARR-seq, each candidate SNP position was represented by all possible nucleotide substitutions, each synthesized as a separate oligonucleotide based on known human genetic variants. The library, containing around 90,000 variants, was then transfected into four cell lines: HEK293T as a non-CRC control, HCT116 and SW480 as primary CRC lines, and SW620 as a metastatic line. Notably, SW480 and SW620 originate from the same patient, offering a paired model where transcriptional and epigenetic differences can be attributed to metastatic progression rather than genetic background variation. RNA and DNA from cells was sequenced and the Enhancer Activity Score (EAS) calculated as . The Allele-Specific Effect (ASE), computed as the log2 ratio of alternative to reference EAS, was used to rank all SNPs, and those reaching statistical significance (adjusted P < 0.05) were defined as preferentially active SNPs (paSNPs). For Methyl-STARR-seq, the library was split into a fully methylated pool, and an unmethylated control leading to the analysis of more than 134,000 CpGs and estimation of their EAS. The Methylation Effect (ME), calculated as the log2 ratio of methylated to unmethylated EAS, was used to rank CpG sites, with statistically significant sites (adjusted P < 0.05) defined as preferentially active Methylsites (paMethylsites).
Functional validation was performed on selected variants through luciferase reporter assays, ChIP-qPCR and ChIP-sequencing, and CRISPR-based genome editing. The results confirmed the reliability of the high-throughput screening, established transcription factor binding patterns at key loci, and demonstrated the regulatory impact of selected variants on target gene expression and tumor-associated cellular phenotypes.
Noncoding variants with clinical impact
Within the primary CRC context, the authors further investigated the selected paSNPs, focusing on two loci in linkage disequilibrium: rs6061231 and rs67941642. Population analysis revealed only two common haplotypes: C-C (77%) and A-T (23%). TF binding motif analysis identified p53 and GFI1 as the transcription factors binding these loci, with p53 preferentially binding rs6061231-C and GFI1 preferentially binding rs67941642-T. Despite the contrasting individual contributions of the two SNPs to enhancer activity, rs67941642 emerged as the dominant regulatory driver, because it conferred the strongest overall enhancer output. This result was in contrast with the previous knowledge derived from GWAS analysis, where rs6061231 was identified as the lead SNP7. In addition, CRISPR-based genome editing identified TAF4 as the target gene regulated by this enhancer, with cells carrying the A-T haplotype showing higher TAF4 expression, reduced proliferation, and lower migratory capacity. Consistently, TCGA-COAD clinical data showed a significantly better 5-year survival rate in patients with higher TAF4 expression.
In the metastatic context, the authors focused on rs1962004 among the metastasis-specific paSNPs. This position carries the A allele exclusively in SW620, where it was likely somatically acquired during metastatic progression.
This allele creates a binding site for JUN and JUND, as confirmed by TF binding motif analysis and ChIP-qPCR. Genome editing demonstrated that the A allele drives higher LRRC61 gene expression, increased cell proliferation and migration. Concordantly, clinical data showed worse survival in patients with elevated LRRC61 expression6.
Among the paMethylsites identified in primary CRC, multi-omics integration led to the selection of six high-confidence functional CpG sites. Filtering TCGA dataset for |Δβ| > 0.25 and FDR < 0.05 further narrowed this selection to two CRC-specific hypomethylated loci: cg25982657 and cg08640619. Functional characterization of cg08640619 revealed its localization within the first intron of KIRREL1, where TF binding motif analysis predicted RUNX2 binding. Since RUNX2 preferentially binds unmethylated DNA, its reduced enrichment at the hypermethylated locus was experimentally confirmed. Consistently, RUNX2 knockdown resulted in decreased expression of KIRREL1. Clinical data further supported these findings, with cg08640619 hypomethylation associating with worse patient prognosis.
Discussion
These findings suggest that the identified SNPs and their associated gene expression changes could represent promising tools for risk stratification and prognostic prediction in CRC. However, the study does not provide direct clinical evidence for the applicability of these variants as validated biomarkers, and further investigation in larger cohorts is warranted. Additionally, as LRRC61 is characterized here for the first time as a metastasis driver, its molecular mechanisms and downstream targets remain to be explored.
The CRC-specific hypomethylation pattern observed at the two CpG sites showed remarkable diagnostic potential, with a very high sensitivity and specificity, outperforming conventional promoter-based methylation biomarkers. The methylation landscape in the metastatic context was also explored, although this aspect was less extensively characterized compared to the primary CRC analysis6.
Regarding diagnostic applicability, the potential use of cg08640619 and cg25982657 as liquid biopsy biomarkers through circulating tumor DNA (ctDNA) detection is an intriguing perspective currently under investigation. However, given the low abundance of ctDNA in blood, reliable detection of methylation patterns represents a technical challenge that requires dedicated validation on blood-based samples, as highlighted by a recent systematic review on methylated ctDNA-based assays for CRC early detection8. Additionally, while KIRREL1 expression changes were documented, the functional consequences of its upregulation in CRC remain unexplored.
Limitations of the study
Several limitations and open questions remain. The study was conducted exclusively on cell lines, and validation in in vivo models or patient-derived organoids would better capture tumor complexity.
A methodological limitation concerns the 120 bp fragment length used in both libraries, imposed by oligonucleotide synthesis requirements, which may incompletely capture the full regulatory architecture of enhancers. Although Hi-C data were integrated to map topological associating domains and connect enhancer regions to target genes, this does not fully resolve the intrinsic limitation of testing short sequences in the STARR-seq assay. While CRISPR-Cas9 validation confirmed the functional relevance of selected variants in their native chromatin context, the authors suggest that future studies employing in situ fragmentation strategies could generate more representative libraries6. However, it is not fully clear how the in situ fragmentation would include all the accessible chromatin regions that are active enhancers. A possible solution is represented by CapSTARR-seq, which combines H3K27ac-based capture with genomic fragmentation to generate libraries of active enhancer regions with fragment lengths of 200-500 bp9.
Finally, both SNP and CpG candidate selection were predominantly based on European and East Asian population data, and the inclusion of additional ethnicities in future studies would improve the generalizability of these findings.
Conclusions
The evidence presented demonstrates the potential of high-throughput enhancer-focused screening as a tool to systematically identify functional genetic and epigenetic variants with clinical relevance and biomarker potential. The integrated pipeline developed by the authors provides a solid methodological framework that, given its design around enhancer-centric variant selection, is readily adaptable to the study of other epithelial cancers, opening new avenues for broader pan-cancer applications.
References
- Colorectal Cancer. World Health Organization https://www.who.int/news-room/fact-sheets/detail/colorectal-cancer (2026).
- Vogelstein, B. et al. Cancer Genome Landscapes. Science 339, 1546–1558 (2013).
- Hindorff, L. A. et al. Potential etiologic and functional implications of genome-wide association loci for human diseases and traits. Proc. Natl. Acad. Sci. 106, 9362–9367 (2009).
- Tuupanen, S. et al. The common colorectal cancer predisposition SNP rs6983267 at chromosome 8q24 confers potential to enhanced Wnt signaling. Nat. Genet. 41, 885–890 (2009).
- Bell, R. E. et al. Enhancer methylation dynamics contribute to cancer plasticity and patient mortality. Genome Res. 26, 601–611 (2016).
- Chen, E. et al. Systematic analysis of functional genetic and epigenetic variants in colorectal cancer. Sci. Adv. (2026).
- Zeng, C. et al. Identification of Susceptibility Loci and Genes for Colorectal Cancer Risk. Gastroenterology 150, 1633–1645 (2016).
- Khabbazpour, M. et al. Advances in blood DNA methylation-based assay for colorectal cancer early detection: A systematic updated review. Gastroenterol. Hepatol. Bed Bench 17, (2024).
- Vanhille, L. et al. High-throughput and quantitative assessment of enhancer activity in mammals by CapStarr-seq. Nat. Commun. 6, 6905 (2015).