Epigenomics and the Kidney

Abstract

Purpose of review:

Epigenetic modifications are reversible changes to a cell’s DNA or histones that alter gene expression but not DNA sequence. This review will explore epigenomic profiling and bioinformatics techniques for the study of kidney development and disease.

Recent findings:

Reversible DNA and histone modifications influence chromatin accessibility and can be measured by a variety of recent techniques including DNase-seq, ATAC-seq and single cell ATAC-seq (scATAC-seq). These approaches have been used to demonstrate that DNA methylation is critical for nephron progenitor maturation, for example. New bioinformatics techniques allow the prediction of chromatin loops that connect regulatory elements to target genes. Recent studies have demonstrated that DNA elements regulate transcription in the kidney via long-range physical interactions and create a new framework for understanding how GWAS risk loci contribute to kidney disease. Increasingly, epigenomic approaches are being combined with transcriptomic analyses to generate multimodal datasets.

Summary:

Epigenomics has expanded our knowledge of gene architecture and regulation. Novel tools and techniques have led to the emergence of ‘multi-omics’ in which epigenomic profiling, transcriptomics, and additional methods complement each other to improve our understanding of kidney disease and development.

Introduction

Next generation sequencing (NGS) has profoundly impacted our ability to understand the epigenome. Epigenomic features include reversible DNA and histone modifications that regulate chromatin architecture and transcription¹. Recent studies have applied a variety of methods to characterize DNA methylation², chromatin accessibility³ and chromatin-chromatin interactions⁴ in the kidney. This review will encompass recent advances in epigenomics and highlight studies that have contributed to our understanding of kidney disease and development.

Genome Wide Association Studies and Risk Variants in Kidney Disease

Large genome wide association studies (GWAS) of estimated glomerular filtration rate (eGFR) have demonstrated that risk loci are enriched in regions with kidney-specific disease associations^5–7. These studies have identified novel risk loci that influence the progression of common kidney diseases across multiple ethnic groups. A majority of candidate risk loci are in intronic and intergenic regions and the mechanism by which such variants affect disease pathogenesis remains unclear⁸. Single cell RNA sequencing has been helpful in determining how risk alleles influence gene expression in a cell type specific manner⁹. Expression quantitative trait loci (eQTL) are loci that explain a portion of the genetic variance for gene expression. Recent eQTL studies have better defined the role of genetic variation in nephrotic syndrome⁹, chronic kidney disease¹⁰, and low eGFR⁶. These data will help us better understand how genetic risk predicts kidney disease outcomes and may lead to the development of novel therapeutics. However, genetic variants can do much more than modify gene expression. Genetic variants may influence the probability that a chromatin region is open or closed and future studies may benefit from exploring how genetic variation affects chromatin accessibility (caQTL)¹¹ and chromatin-chromatin interactions⁴ in kidney disease.

Histone modifications and transcriptional regulation

Histones compartmentalize the genome by packaging DNA into nucleosomes. Histones undergo post-translational methylation, acetylation, phosphorylation, ubiquitylation and sumoylation that can either promote or repress gene expression by altering chromatin accessibility and recruiting histone modifiers¹². There is a growing list of modifications that have been associated with increased (eg. H3K4me3, H3K27ac) or decreased (eg. H3K27me3) expression of nearby genes. Recent studies have examined the role of histone modifications in unilateral ureteral obstruction¹³ and diabetic nephropathy models¹⁴. Hewitson et al. used mass spectrometry to assess histone modifications in a mouse model of unilateral ureteral obstruction (UUO)¹³. Within the healthy kidney, they reported compartment-specific histone marks that likely determine lineage-specific gene expression. UUO led to relatively few histone modifications, although they did detect enrichment for H3K79me2 in a subset of proximal tubules that had increased basolateral expression of GLUT1. Their data suggests that H3K79 methylation may promote transcription of GLUT1 as part of a cellular response to hypoxia. Notably, this study measured histone modifications in aggregate as opposed to individual cell types, which may limit sensitivity for detecting differences in less abundant cell types. Jia et al. explored the role of transforming growth factor-β1 in mesangial cell cultures and diabetic rodent models¹⁴. They demonstrated that the repressive histone mark, H3K27me3, was decreased in the promoters of pro-fibrotic genes following TGF-β stimulation. In particular, increased transcription of murine Ctgf, Serpine1, and Ccl2 was associated with decreased H3K27me3 marks. Additionally, they observed downregulation of the suppressive histone methyltransferase EZH2, which is a core subunit of the polycomb repressive complex. These additive effects resulted in increased expression of pro-fibrotic genes and was reproducible in streptozotocin-induced diabetic nephropathy and human mesangial cells. Together these data create a model for increased expression of diabetic kidney disease-related genes driven by a reduction in repressive histone marks.

DNA methylation in ageing and allografts

DNA methylation is a dynamic process that regulates transcription by placing a methyl group on cytosine bases that are in CpG islands – regions of DNA with a high number of cytosine – guanine repeats. CpG islands are typically clustered in gene regulatory regions, for example in the upstream regions of a gene transcriptional start site¹⁵. Increased methylation in the vicinity of transcriptional start sites has been associated with gene silencing [5]. Recent studies have demonstrated that changes in DNA methylation patterns are associated with age-related kidney dysfunction². Heylen et al. showed that increased DNA methylation is associated with ageing and that differentially methylated regions are commonly located in genes involved in Wnt/beta-catenin signaling². In an independent cohort of 67 biopsies obtained after reperfusion, these epigenetic marks were associated with reduced kidney allograft function one year after transplant. These findings suggest that methylation profiling can predict future kidney allograft injury.

DNA methyltransferases in kidney development

DNMT1 is a DNA methyltransferase that catalyzes the transfer of methyl groups to newly synthesized DNA during replication. Li et al. performed genome-wide methylation profiling to demonstrate that promoters and enhancers undergo dynamic changes in methylation status during nephron development¹⁶. Developing kidneys had elevated expression of DNA methyltransferases and increased methylation in promoter and enhancer regions compared to adult mice. Nephron progenitor-specific deletion of Dnmt1 (Six2^CreDnmt1^flox/flox) resulted in a severe kidney developmental defect that was not observed in newborn mice lacking Dnmt3a (Six2^CreDnmt3a^flox/flox) or Dnmt3b (Six2^CreDnmt3b^flox/flox). By contrast, deletion of Dnmt1 in developing podocytes (Pod^CreDnmt1^flox/flox) did not affect kidney development. These data indicate that DNMT1 is critical for development of the functional nephron epithelium, most likely by acting to regulate methylation status in Six2⁺ metanephric mesenchyme. Mice lacking DNMT1 in SIX2+ progenitor cells had a genome-wide decrease in CpG methylation that was predominantly located in intergenic regions coinciding with transposable elements. Endogenous retroviruses (ERVs), long interspersed elements (LINEs), and short interspersed elements (SINEs) are families of transposable elements that are silenced by hypermethylation in mature cells¹⁷. DNMT1 knockout mice had decreased methylation in ERVs, increased ERV transcription, and increased P53 activation, suggesting that loss of hypermethylation in these regions leads to their developmental defect. Another group led by Wanner et al. investigated the role of DNMT1 in prenatal nephron progenitor programming¹⁸. They demonstrated that high glucose stimulation of E12.5 metanephric organs leads to a reduction in nephron number and decreased methylation of repetitive DNA sequences. They identified Dnmt1 and Dnmt3a expression in the nephrogenic niche, ureteric bud, and early stages of developing nephrons, but not in mature cell types. Wanner et al. generated conditional knockout mice lacking Dnmt1, Dnmt3a, or Dnmt3b in Six2+ metanephric mesenchyme and examined them at E19.5. Kidneys lacking Dnmt1 showed a reduction in kidney weight and nephron quantity, which they attributed to a loss of proliferative capacity in the nephrogenic niche between E15.5 and E19.5. Similar to Li et al, Wanner et al. reported an upregulation of endogenous retroviral transcripts (ERVs) and pathways involved in the response to interferon type 1. These two papers collectively enhance our understanding of the role of DNA methylation in kidney development and provide insights into the role of specific transcription factors in nephron progenitor maturation.

Chromatin Accessibility as a Marker of Gene Activity

The Assay for Transposase-Accessible Chromatin (ATAC-seq) identifies open chromatin by probing DNA accessibility with hyperactive Tn5 transposase (Figure 1)¹⁹. ATAC-seq requires fewer cells and reduced sample processing time than DNase-seq and has been used to generate an atlas of chromatin accessibility in adult mouse kidney³, mouse nephron progenitors^2,20, and mouse nephron segments²¹. Hilliard et al. used single cell RNA sequencing (scRNAseq), ChIP-seq, and ATAC-seq to investigate the chromatin landscape of Six2+ nephron progenitors²⁰. They demonstrated that open chromatin areas are increased in promoter and enhancer regions of younger nephron progenitors and that their location changes with time. In particular, they detected an increase in open chromatin for Six2 in E16 nephron progenitors compared to perinatal day 2 mice. Nephron progenitor maturation was associated with a decrease in open chromatin regions in Six2 and an increase in open chromatin regions in the pro-differentiation gene Hnf1b. The distal regions of Six2 showed enhanced accessibility for Bach2/AP1 transcription factor binding sites that promote differentiation. ChIP-seq of cultured nephron progenitors demonstrated an increase in H3K27me3/H3K4me1 bivalent marks, which is an indication that cells are poised for differentiation. These marks were in genes that regulate cell junction assembly, and inactivation of the MAPK pathway. Furthermore, the open chromatin regions detected in their ATAC-seq studies coincided with active enhancers (H3K27ac/H3K4me1) and binding sites for core transcription factors. These data show that histone modification and chromatin accessibility coordinate the differentiation of nephron progenitors by regulating the activity of key transcription factors that balance self-renewal and differentiation.

Figure 1: Assay for Transposase-Accessible Chromatin by Sequencing (ATAC-seq) –

ATAC-seq is a method for measuring genome-wide chromatin accessibility. We have displayed a representative gene within an open chromatin region in green and a closed chromatin region in red. The gene contains a promoter region (Pr) and several exons (Ex). Hyperactive mutant Tn5 Transposase (Tn5) cleaves double-stranded DNA and inserts sequencing adapters into open regions of chromatin in a process called “tagmentation”. Tn5 cannot access areas of closed chromatin. Tagged DNA fragments are purified, PCR-amplified, and sequenced. Sequencing reads are used to infer regions of increased chromatin accessibility displayed as ATAC peaks. An increase in the number or amplitude of ATAC peaks is a measure of how open the chromatin is and roughly correlates with gene expression. Tissue or cell type-specific transcription factor (TF) activity can be predicted by searching for transcription factor binding motifs within open promoter (Pr) regions. Hi-C, ChIA-PET, or bioinformatics tools can be used to predict cis-regulatory interactions between promoters and distal enhancers that promote or repress gene expression.

Single cell methods for measuring chromatin accessibility and predicting chromatin interactions

Single cell ATAC-seq (scATAC-seq) is an extension of bulk ATAC-seq that measures chromatin accessibility in hundreds to thousands of individual cells²². Cao et al. successfully performed scATAC-seq in combination with single cell RNA sequencing (scRNA-seq) to simultaneously interrogate chromatin accessibility and transcription in the adult mouse kidney²³. This approach improves cell type identification by leveraging the transcriptional profiles obtained by scRNA-seq to help overcome the sparsity of scATAC-seq data. An interesting finding was that differential chromatin accessibility in a gene promoter is not necessarily associated with differential gene expression. Thus promoter accessibility may have an alternate role in gene regulation through its interaction with cis-regulatory elements. Candidate cis-regulatory elements can be predicted from scATAC-seq data using a bioinformatics tool called Cicero²⁴. Cicero calculates the covariance of accessibility between chromatin regions. Open chromatin areas that covary are more likely to represent loops of chromatin that link regulatory DNA elements to their target genes. Cao et al. linked distal regulatory regions with the transcriptional start site of nearby genes and demonstrated that these interactions are cell type specific. The link with the highest correlation was between SLC12A3 (Sodium Chloride Co-transporter, NCC) and a candidate enhancer 36kb downstream of its transcriptional start site that was specific to the distal convoluted tubule. This candidate enhancer may drive convoluted tubule-specific expression of NCC. Similar approaches could be used to predict additional cell type specific chromatin-chromatin interactions that drive differentiation.

Techniques for measuring chromatin interaction

Increasing attention is being paid by investigators on the role of 3-dimensional genome architecture in regulating gene expression. Hi-C and ChIA-PET are two techniques that capture the 3-dimensional architecture of the genome^25–27. Hi-C utilizes formaldehyde to crosslink chromatin regions followed by digestion, biotin labeling, and re-ligation of covalently-bound DNA fragments. Next generation sequencing of purified chimeric DNA ligation junctions enables the characterization of numerous chromatin-chromatin interactions. Chromatin Interaction Analysis using Paired-End Tag sequencing (ChIA-PET) also uses formaldehyde cross-linking, but incorporates antibody-based chromatin immunoprecipitation prior to proximity-based DNA ligation, digestion and sequencing. In general, Hi-C measures protein-independent chromatin folding and ChIA-PET measures protein-mediated chromatin folding (for example, transcription factors). These technologies have enhanced our understanding of gene organization and the relationship between promoters, enhancers and insulators.

Chromatin interaction and regulation of transcription

Hi-C has been used together with DNase-seq and RNA-seq to investigate how chromatin-chromatin interactions affect gene expression⁴. The study by Sieber et al. is a new addition to the rapidly evolving field of multi-omics⁴. This study demonstrated that GWAS risk loci for kidney disease are located in cell type specific open chromatin areas and may participate in long-range interactions with distant chromatin regions. In particular, the single nucleotide polymorphisms (SNPs) rs219779 and rs219780 are candidate risk loci for kidney stones located in the last coding exon of CLDN14. Sieber et al. showed that although these variants are closest to CLDN14, they may regulate expression of a nearby gene (SIM2) via long-range physical interaction. Similarly, a SNP associated with eGFR (rs84178) in an intronic region of KCNQ1 interacts with CDKN1C, which is an important regulator of podocyte differentiation. The striking finding from this study is that GWAS risk loci do not necessarily regulate expression of nearby genes.

Sequence variation influences chromatin interactions

Large scale efforts have used Hi-C to identify long-range chromatin interactions between promoters and regulatory regions in multiple tissue types²⁸. The vast majority of promoters had at least one significant chromatin interaction and the median distance between the promoter and the interacting region was 158kb. Interestingly, approximately 13 to 45% of these interactions were unique to the tissue type. Jung et al. hypothesized that promoter-specific chromatin interactions are a regulatory mechanism through which eQTL influence transcription of target genes. In fact, they demonstrated that tissue-specific eQTL were highly enriched in chromatin interaction domains. These data strongly suggest that cis-regulatory relationships between distant eQTL and target genes may underlie physiological traits and disease pathogenesis. Similar approaches have been used to investigate the role of allele-specific SNPs (AS-SNPs) in susceptibility to immune and B cell related diseases²⁹. AS-SNPs can change transcription factor binding affinity and are one potential explanation for how common genetic variants affect cell-specific gene expression (and thereby confer disease risk). Cavalli et al. used long-range chromatin interactions and transcription factor motif predictions to prioritize transcription factors affected by AS-SNPs. They predict that AS-SNPs can influence chromatin folding by affecting the activity of the architectural proteins CTCF and SA.1. CTCF binding sites define the boundaries of topologically associated domains (TADs) where the majority of promoter-enhancer interactions occur. They hypothesized that AS-SNPs in interaction domains influence the formation of chromatin loops and by extension the expression of genes within the loop. Future studies are likely to implement these approaches in the kidney and newer methods will be able to examine these regulatory interactions on a single cell basis²⁷.

Conclusion

Epigenomics will enhance our understanding of kidney injury, development, and regeneration. Future studies could apply this knowledge to improve kidney organoid differentiation protocols or therapies for acute tubular injury. Epigenetic features are potentially reversible and represent an opportunity to develop novel therapies. Reversal of epigenetic marks may be useful in treating conditions like diabetic nephropathy, which is characterized by metabolic memory³⁰. Histone deacetylase and DNA methylation inhibitors are used as anticancer agents^31,32 and similar approaches could be applicable to kidney disease.

Key Points.

• DNA methylation and histone modification regulate chromatin accessibility

• Chromatin accessibility influences gene expression and transcription factor binding in kidney disease and development

• Gene expression is regulated by long-range interactions with DNA elements that contribute to cell type specificity

• Epigenomic profiling helps predict the function of GWAS risk loci for kidney disease