Filaggrin and Non-Coding Variation in Distinct Cohorts of Atopic Dermatitis

Atopic dermatitis (AD) is a chronic, inflammatory skin disorder characterized by scaly rash flares and pruritis. This disease is highly heritable, affecting up to 11% and 10% of pediatric and adult populations in the United States, respectively. Loss of function mutations in the epidermal barrier gene, filaggrin (FLG), represent the most widely replicated genetic risk factor for AD. Additionally, the total number of intragenic FLG repeats affects AD risk in a dose-dependent manner, with addition of each repeat reducing the odds ratio for disease. While FLG-null mutations account for up to 50% of moderate-to-severe cases, the remaining heritable factors affecting AD disease risk are not well understood. Here, I further our understanding of the role of heritable factors for this complex disease by investigating the underlying genetics that define distinct clinical cohorts for AD. African-American (AA) pediatric patients are disproportionately affected with AD yet rarely harbor FLG-null mutations. The genetic epidemiology with respect to AD severity for this health disparities group is unknown. I sought to identify the severity and clinical features of AD specific to a group of AA pediatric patients and further assess the genotypes that represent the total number of FLG monomer repeats (10, 11, or 12 monomers), referred to as intragenic filaggrin copy number variants (CNV). A total of 39 African-American pediatric patients were recruited through the St. Louis Children’s Hospital pediatric dermatology clinic and provided written informed assent or consent after meeting the UK Working Party’s Diagnostic Criteria for Atopic Dermatitis and moderate to severe AD (SCORAD ≥25). This study was approved by the Institutional Review Board at Washington University School of Medicine. Recruited patients exhibited severe and chronic AD (mean SCORAD = 58.5) with severe lichenification and skin dryness (mean score 2.2 each, scale [0-3]). The FLG CNV 10 allele represented the majority of the 3 genotyped FLG CNV alleles (63.5%) with only two 2 FLG-null mutation carriers (R501X) identified. Patients with a total of either 20 or 21 FLG CNV exhibited higher SCORAD (mean, 63.7) indicative of severe AD compared to the patients with 22, 23, and 24 FLG CNV (mean SCORAD, 48.5) (P = 0.01). Thus, I identified low FLG CNV (total of 20 or 21 FLG repeats) associated with severe AD. This health disparities study is the first to report a link between low FLG CNV and AD disease severity. I next hypothesized a role for regulatory variants in a unique adult cohort with severe AD. The patients are characterized by a distinct molecular phenotype in their baseline (unaffected) skin: downregulation of many EDC genes in the absence of FLG R501X and 2282del4 common mutations. I aimed to discover AD-related regulatory variants and assess their functional impact using targeted sequencing and cell-based reporter assays. Targeted sequencing of 9 functional regulatory elements (REGs 180, 184, 189, 346, 7153, 7553, 7953, 2153, and 4353) that mapped within linkage disequilibrium blocks with AD-GWAS tagging SNPs in 1q21 and 11q13 (rs877776 and rs7927894, respectively) were performed in pooled cases (n = 27) and controls (n = 27). I identified 17 regulatory variants, including two novel variants: 180NOV (hg19, chr1:152,135,930, T to G) and 184NOV (hg19, chr1:152,140,669, G to A.) The novel variants were genotyped in 2 patients with severe AD (SCORAD = 59 and 56) and were found to be in cis with rs877776, an AD-tagging SNP on 1q21. Only the AD-specific variants, two novel variants (180NOV and 184NOV) and a 6-SNP haplotype allele in REG180, exhibited significant increased effect sizes in keratinocyte functional assays. Chromatin immunoprecipitation demonstrated SOX9 and NRF2 occupancies in primary human keratinocytes at location of the 180NOV and 184NOV variants, respectively, suggesting disruption of known epidermal regulator SOX9 and NRF2 binding in these enhancers. A separate AD cohort was additionally sequenced to validate the variants found in the discovery cohort and identified the 180NOV variant in one additional AD patient and the REG180 6-SNP haplotype allele in two additional patients with AD. Thus, I reveal a functional impact for AD-specific regulatory variants in two independent adult cohorts for severe AD. Motivated by these results, I wanted to further investigate the impact of regulatory variants in the 923 EDC enhancer in the context of AD. 923 is an epidermal-specific enhancer within the EDC with known functional activity observed during epidermal barrier development. The 923 enhancer forms physical interactions with many EDC genes whose expression is critical for the skin barrier. I hypothesized variants in the 923 enhancer to affect enhancer function in the context of AD. Targeted amplicon sequencing of the 923 enhancer in 117 AD patients and 27 controls identified 7 variants, 5 of which are AD-specific (including 1 novel variant) and 3 significantly affecting regulatory activity in keratinocytes. The functional variants in the 923 enhancer converge on Kruppel-like, AP-1, and SMAD family transcription factors with established epidermal development function. Together, my data for functional regulatory variants in the context of AD suggests a role for epidermal development TFs in the homeostasis of the skin barrier via enhancer-mediated regulation of EDC genes. In summary, I have expanded our understanding of genetic risk factors aside from FLG-null mutations in moderate to severe AD cases: low FLG CNV in AD disease severity in a health disparities group and a functional role for AD-associated regulatory variants predicted to impact TFs involved in epidermal differentiation