lncRNA CDKN2B-AS1 regulates collagen expression

Abstract

The long noncoding RNA CDKN2B-AS1 harbors a major coronary artery disease risk haplotype, which is also associated with progressive forms of the oral inflammatory disease periodontitis as well as myocardial infarction (MI). Despite extensive research, there is currently no broad consensus on the function of CDKN2B-AS1 that would explain a common molecular role of this lncRNA in these diseases. Our aim was to investigate the role of CDKN2B-AS1 in gingival cells to better understand the molecular mechanisms underlying the increased risk of progressive periodontitis. We downregulated CDKN2B-AS1 transcript levels in primary gingival fibroblasts with LNA GapmeRs. Following RNA-sequencing, we performed differential expression, gene set enrichment analyses and Western Blotting. Putative causal alleles were searched by analyzing associated DNA sequence variants for changes of predicted transcription factor binding sites. We functionally characterized putative functional alleles using luciferase-reporter and antibody electrophoretic mobility shift assays in gingival fibroblasts and HeLa cells. Of all gene sets analysed, collagen biosynthesis was most significantly upregulated (P_adj=9.7 × 10^{- 5} (AUC > 0.65) with the CAD and MI risk gene COL4A1 showing strongest upregulation of the enriched gene sets (Fold change = 12.13, P_adj = 4.9 × 10^{- 25}). The inflammatory "TNFA signaling via NFKB" gene set was downregulated the most (P_adj=1 × 10^{- 5} (AUC = 0.60). On the single gene level, CAPNS2, involved in extracellular matrix organization, was the top upregulated protein coding gene (Fold change = 48.5, P < 9 × 10^{- 24}). The risk variant rs10757278 altered a binding site of the pathogen responsive transcription factor STAT1 (P = 5.8 × 10^{- 6}). rs10757278-G allele reduced STAT1 binding 14.4% and rs10757278-A decreased luciferase activity in gingival fibroblasts 41.2% (P = 0.0056), corresponding with GTEx data. CDKN2B-AS1 represses collagen gene expression in gingival fibroblasts. Dysregulated collagen biosynthesis through allele-specific CDKN2B-AS1 expression in response to inflammatory factors may affect collagen synthesis, and in consequence tissue barrier and atherosclerotic plaque stability.

Fig. 1

Differentially expressed genes and enriched gene sets in gingival fibroblasts after CDKN2B-AS1 knockdown. (A) Volcano plot of LNA GapmeRs transfected gingival fibroblasts showing differential expression of protein coding genes and numerous lncRNAs and pseudogenes. The names of the most significant differentially expressed protein coding genes are shown. The names of the most significant differentially expressed lncRNAs and pseudogenes are not shown to highlight the observed prominent role of CDKN2B-AS1 interaction with non-protein coding genes. (B) Transfection of primary gingival fibroblasts with LNA GapmeRs induced significant reduction of CDKN2B-AS1 transcript levels (qRT-PCR). (C-D) Western blot analysis validated that reduced CDKN2B-AS1 transcript levels correlated with significantly reduced CAPNS2 protein levels. Western Blot band intensities are normalized to ACTB (*p < 0.05; **p < 0.01). (E-G) Gene set enrichment analysis of GapmeR transfected gingival fibroblasts. Shown are evidence plots (receiver operator characteristic curves) for the significant gene sets with an area under the curve (AUC) ≥ 0.6. (C) From Reactome database, REACTOME_COLLAGEN_BIOSYNTHESIS_AND_ MODIFYING_ENZYMES _ M26999, enriched 62 genes. (D) From Reactome database, REACTOME_COLLAGEN_ CHAIN_TRIMERIZATION_ M27812, enriched 40 genes; (E) From Hallmark database, HALLMARK_TNFA_SIGNALING_VIA_ NFKB_ M5890, enriched 191 genes; (F&G) From Tmod database, EXTRACELLULAR MATRIX (I)_ LI.M2.0 enriched 30 genes and COLLAGEN, TGFB FAMILY ET AL_ LI.M77 enriched 31 genes, respectively. The gray rug plot underneath each curve corresponds to genes sorted by P value, with the genes belonging to the corresponding gene sets highlighted in red (upregulated genes) or blue (downregulated genes). Bright red or bright blue indicates that the genes are significantly regulated. (H) Western blotting validation of LNA GapmeRs transfected gingival fibroblasts showing COL4A1 and COL6A1 upregulation after CDKN2B-AS1 knockdown (COL4A1 fold change = 1.74, P = 0.0004; COL6A1 fold change = 1.57, P = 0.0155 respectively)

Fig. 2

rs10757278 is a functional SNP within the chr9p21.3 risk haploblock and localizes to a STAT1 binding site. (A) 55 common SNPs (MAF ≤ 0.05) in CEU and GBR populations in strong LD with the GWAS lead SNP rs1333049 and 24 SNPs located within chromatin elements that correlate with regulatory functions of gene expression, which is indicated by chromatin state segmentation for 3 cell types (data from ENCODE; orange = predicted strong enhancer, yellow = weak enhancer, blue = insulator). Some proxy SNPs locate in H3K4me1 and H3K27ac methylation marks, which are often associated with regulation of gene transcription and within TFBS that were determined from ENCODE ChIP-Seq data. The position of rs10757278 is marked with a dashed line. (B) The DNA sequence at rs10757278-A allele shares a matrix similarity of 95% with the STAT1 transcription factor (TF) binding motif. (C) EMSA was performed with rs10757278 allele-specific oligonucleotide probes and nuclear protein extract from gingival fibroblast cells. Binding of STAT1 antibody to allele-specific probes is shown in lane 2 and 7. The supershift caused by STAT1 antibody binding to the DNA probe-protein complex is seen in lane 3 and 6. Unlabeled DNA was added in lanes 1 and 8 to verify that the band shift was antibody-specific. (D) Absolute value area of the antibody-specific bands. In the background of rs10757278-G allele, STAT1 binding to the allele-specific oligonucleotide probe was reduced 14.4% compared to the A-allele

Fig. 3

rs10757278 STAT1 binding A-allele reduces reporter gene activity. (A, B) rs10757278-A significantly reduced luciferase activity in gingival fibroblasts (fold change = 1.72-fold, P = 0.0056) and in HeLa cells (fold change = 1.83-fold, P = 0.0063). (Control = empty pGL4.24 plasmid, *p < 0.05; **p < 0.01). (C) GTEx data indicate a cis-eQTL effect for rs10757278 on CDKN2B-AS1 expression (tissue: pituitary gland) with rs10757278-A significantly reducing CDKN2B-AS1 expression compared to the G allele (p = 3.9 × 10^− 6)