Background

Psoriasis is an immune-mediated multigenic skin disease characterized by symmetrically well-defined erythema, covered with silvery scales involving elbows, knees, torso, and scalp [1]. The histopathological traits of psoriasis are diverse, typically including hyperkeratosis with parakeratosis and immune cell infiltration, Munro micro-abscess, acanthosis thickening, vascular dilatation congestion, elongation of rete pegs, and granulosa thinning. Psoriasis patients are usually accompanied by mental and physical burden because of its high incidence, chronic course, disability, malformation, and comorbidities, such as metabolic syndrome and cardiovascular diseases [2, 3]. To address this burden, scientists have tried to unravel the etiology and pathogenesis of psoriasis with various strategies and have made significant progress, but there are still many mysteries. More than 80 susceptibility genes were recently identified [4], and some cytokines, such as TNF-α, IL-17, and IL-23, have been developed for biological agents [5]. However, the exact pathogenesis of psoriasis was still not fully revealed.

The histopathological features of psoriasis indicate a critical alteration in disease progression, but the molecular mechanism under these features is largely unknown. Few studies aimed to link epigenetic modifications with the histopathological characteristics in psoriasis. For example, Chandra et al. carried out genome-wide DNA methylation to figure out which epigenetic loci are associated with Munro micro-abscess in psoriasis [6]. Nevertheless, chromatin accessibility and its potential regulatory roles in histopathological changes are still missing.

Assay for targeting accessible-chromatin with high-throughput sequencing (ATAC-seq), based on Tn5 transposase hyperactivity, helps us investigate genome chromatin accessibility and reveals multiple aspects of transcriptional regulation [7]. It provides the whole open chromatin across the entire genome at one time, exploring transcription factor binding and gene expression regulation, which has been widely used in various diseases [8], including psoriasis. Utilizing ATAC-seq on 15 psoriatic lesions, 9 non-psoriatic lesions, and 19 normal healthy skin tissues, we previously identified 4,195 differentially accessible regions [9]. Further analysis showed that the sequence of differentially accessible regions was enriched in the FRA1/AP-1 transcription factor binding region [9]. Upregulation of AP-1 family members has been shown in psoriatic skins, but the exact mechanism is not precise [10]. Currently, we tried to perform an integrative analysis of RNA-seq and ATAC-seq data, exploring the potential network of AP-1 regulating psoriasis and aiming to find out whether some genes are implicated in histopathological alterations.

Several methods have been used to integrate transcription and chromatin accessibility datasets by directly overlap** relevant genes, which might underestimate the potential roles of TF binding targets [7, 11,12,13]. The BETA algorithm is an efficient web tool to identify motifs of transcription factors, infer their target genes, and explore these factors' activating or repressive status [14]. To search the direct targets of differentially accessible regions, we attempted to perform BETA-plus (Version 1.0.7) to integrate ATAC-seq accessible peak data with differential expression data. Interestingly, we identified 408 upregulated genes and 133 downregulated genes (rank product < 0.01) and depicted significant binding motifs and putative collaborating factors. These upregulated genes were strongly targeted by AP-1 family transcription factors. Then, we found that their gene expression differences were related to the different pathological manifestations of psoriasis. It provides us with a novel insight into the potential regulatory mechanism and therapy of psoriasis and a future direction for us to deeply explain the mechanism of AP-1 in psoriatic lesions.

Results

The prediction of TFs' function and direct targets

Based on our RNA-seq and ATAC-seq data, we used BETA-plus to integrate differentially expressed genes and open accessible peaks [9, BETA analysis

BETA (version 1.0.7) is an exposed source at http://customer.org/ap/ [14]. Utilized the information of binding site and differential expression, BETA predicts the function of activation or inhibition of TFs, infers the target genes, and identifies the motif and its binders of TFs. All BETA calculations are based on the Regulatory Potential Scores of each target gene, which is the likelihood that a gene is regulated by a factor, and each gene estimates its regulatory potential. The regulatory potential is calculated as Sg = \({\sum }_{i=1}^{k}{e}^{-(0.5+4\Delta i)}\) [14, 60]. k is all binding sites near the TSS of gene (g) within the specified range (default 100 kb) of the different peaks. Δ is the exact distance between k and TSS, proportional to 100 KB, Δ = 0.1 means exact distance = 10 kb. The possibility of gene regulation by factors depends on the number of binding sites in the TSS region and the distance between the binding site and TSS. Data input takes a set of peaks as BED (tissue-specific open chromatin regions from ATAC-seq), and the differential gene expression results from RNA-seq. Analysis was worked using default parameters on the Galaxy Cistrome platform, apart from the significance threshold p < 0.05.

KEGG pathway analysis

Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis, annotation, visualization was carried out using the R (https://www.r-project.org/) package clusterProfiler. The enriched pathway was plotted by package Pathview.

Histopathological section

The collection of skin tissues came from the Department of Dermatology, the First Affiliated Hospital, Anhui Medical University. The overall histopathological abnormalities, including rete peg elongation, the presence of hyperkeratosis with parakeratosis, parakeratosis, Munro micro-abscess, elongation of rete pegs, focal hypergranulosis, granulosa thinning, acanthosis thickening, vascular dilatation congestion, and lymphocyte infiltration, were evaluated by H&E-stained slides. Finally, all the histopathological changes were collected in the form of presence or absence.

Statistical analysis

GSE80047, GSE53552, GSE41662, GSE30999, and GSE14905 expression data matrix was downloaded from GEO database (https://www.ncbi.nlm.nih.gov/geo/). Limma package (version: 3.40.2) was used to identify the differentially expressed genes. All statistical analyses were performed on the R (version 4.1.1) platform. T test, Pearson correlation coefficient, and chi-square test were used to evaluate the significance via SPSS. P < 0.05 was considered statistically significant.