Kidney Week

Abstract: TH-PO0118

Single-Cell Multi-Omic Analysis for Building Cell-Type-Specific Gene-Enhancer Maps in the Kidneys

Session Information

• Integrative Genomics and Multi-Omics in Noncystic Genetic Kidney Disease
  October 22, 2026 | Location: Exhibit Hall A, Convention Center
  Abstract Time: 10:00 AM - 12:00 PM

Category: Genetic Diseases of the Kidneys

• 1202 Genetic Diseases of the Kidneys: Non-Cystic (Complex and Non-Cystic Monogenic)

Authors

• Nguyen, Ngoc Duc, Boston Children's Hospital, Boston, Massachusetts, United States

Ngoc Duc Nguyen

• Han, Seong Kyu, Inha University, Michuhol-gu, Incheon, Korea (the Republic of)

Seong Kyu Han, PhD

• Lee, Chan-Kyu, Inha University, Michuhol-gu, Incheon, Korea (the Republic of)

Chan-Kyu Lee

• McNulty, Michelle, Boston Children's Hospital, Boston, Massachusetts, United States

Michelle McNulty, MS

• Benway, Christopher, Boston Children's Hospital, Boston, Massachusetts, United States

Christopher Benway, PhD

• Lemaire, Mathieu JM, The Hospital for Sick Children Division of Nephrology, Toronto, Ontario, Canada

Mathieu JM Lemaire, MD, MS, PhD, FASN

• Sampson, Matt G., Boston Children's Hospital, Boston, Massachusetts, United States

Matt G. Sampson, MD

• Lee, Dongwon, Boston Children's Hospital, Boston, Massachusetts, United States

Dongwon Lee, PhD

Background

>90% of GWAS loci for kidney traits lie in non-coding regions, yet the specific regulatory mechanisms and DNA elements driving these variants remain unclear. Single-cell epigenomic profiling enabled precise annotation of candidate cis-regulatory elements (CREs) in specific cell types, but element annotation alone neither links CREs to target genes nor reveals the mechanisms. Distinguishing chromatin-mediated transcriptional regulation from other regulatory modes therefore requires cell-type-resolved CRE-gene maps that can be tested directly against variant-level evidence. To address this, we employed a single-cell multimodal framework that constructs cell-type-specific CRE-gene links.

Methods

We analyzed paired single-cell RNA-seq and ATAC-seq data on ~100,000 nuclei from 23 non-tumor kidney cortex samples spanning pediatric to adult stages and diverse kidney functions. We developed a tailored workflow using (1) sequence-assisted multi-sample CRE identification that recovers regulatory elements using sequence-based machine learning, and (2) cell-type-specific statistical modeling with scMultiMap to quantify chromatin accessibility-gene expression associations. To validate mapping accuracy, we benchmarked CRE-gene links against eQTL signals stratified by genomic elements from 50 tissues and partitioned heritability for kidney phenotypes.

Results

We identified 630,000 candidate CREs across 14 cell types. Correlation-based analysis of >12 million candidate CRE-gene pairs identified 422,000 CRE-gene links which are highly cell-type-specific. In proximal tubule cells, candidate CREs harboring eQTL variants were more likely to form significant CRE–gene links in non-coding regions (intronic, promoter, intergenic; OR=1.2-17.3, p<0.001), but not exonic regions (CDS, 5'-UTR, 3'-UTR; p>0.05). This suggests our approach validates eQTL signals mediated through transcriptional regulation, while exonic eQTL variants likely operate via post-transcriptional mechanisms independent of chromatin accessibility. Consistently, in PT-VCAM1+ cells, top 20% connection-ranked CREs explained 18% of eGFRcrea heritability vs. 6% for accessibility-ranked CREs (3.3-fold), placing phenotype-relevant heritability at the CRE-gene connection rather than at the individual CRE.

Conclusion

We demonstrated that single-nucleus multimodal analysis enables high-precision, robust mapping of CRE-gene links across diverse kidney cell types.

Funding

• NIDDK Support