693 - DNA methylation changes in survivors and non-survivors in preterm infants with surgical necrotizing enterocolitis

Publication Number: 693.4204

Nicole G. Hall, Wake Forest School of Medicine, Winston-Salem, NC, United States; Jessica K. Jackson, Wake Forest School of Medicine of Wake Forest Baptist Medical Center, Winston Salem, NC, United States; Padma Garg, University of Mississippi School of Medicine, Jackson, MS, United States; Timothy Howard, Wake Forest School of Medicine of Wake Forest Baptist Medical Center, Winston-Salem, NC, United States; Parvesh Mohan Garg, Wake Forest School of Medicine of Wake Forest Baptist Medical Center, winston salem, NC, United States

Background: Surgical necrotizing enterocolitis (NEC) is a common cause of mortality in preterm infants, and the pathogenesis remains unclear.

Objective: We aim to determine the epigenetic differences in DNA methylation between survivors and non-survivors in preterm infants with surgical NEC.

Design/Methods: DNA was isolated from 10μm sections of NEC formalin-fixed, paraffin-embedded (FFPE) human intestine specimens (N=46) obtained following laparotomy. DNA was extracted using the QIAmp DNA FFPE Tissue Kit (Qiagen), purified using the ZR-96 Genomic DNA Clean and Concentrator-5 Kit (Zymo), and restored using the Infinium HD FFPE Restore Kit (Illumina). DNA was bisulfite converted using the EZ-96 DNA Methylation Kit (Zymo), and DNA methylation data was generated using Illumina’s Infinium EPIC v2.0 BeadChip. Data analysis was performed with the Chip Analysis Methylation Pipeline (ChAMP), a software package available in R. Molecular Complex Detection (MCODE) and the Search Tool for the Retrieval of Interacting Genes/Proteins (STRING) were used to discern relationships between the genes of interest and to conduct functional enrichment analyses of the identified clusters. Both tools were used as plug-ins in Cytoscape (version 3.10.1).

Results: DNA methylation differences were compared from tissue from survivors (n=12) and non-survivors (n=36) with NEC. After adjusting for standard batch effects (i.e., different BeadChips, position on chip, etc.), we identified 4,570 CpG sites, within or near 3,279 genes, that were significantly differentially methylated (FDR p< 0.01) between survivors and non-survivors. The mean difference in methylation for significant CpG sites for survival was 9.7 percent, and nearly half of these sites (2,196) differed by at least 15 percent. The number of significant sites per gene ranged from one to 10, with 54 genes containing four or more sites. Two of the top genes were SEPTIN9, which contained eight CpG sites significant for survival and PRDM16, which contained 10 sites.

The genes that were significant with regard to survival were functionally enriched in 13 biological processes, including “Anatomical structure development” (p=5.77x10-7) and “Embryo development” (p=0.0094), and one pathway (KEGG; Rap1 signaling pathway; 6 genes/201 background genes; FDR value: 0.0187) (Figure 1).

Conclusion(s): We have identified novel epigenetic and transcription signatures in the survivors and non-survivors in preterm infants with surgical NEC. These findings will advance our understanding of disease pathogenesis, prediction, diagnosis and phenotyping.

Table 1. Gene ontology biological processes enrichment analysis of genes significantly differentially methylated between intestinal tissue samples with low and high levels of necrosis. Enrichment analysis of 1,688 genes with three or more methylation sites that were significantly (Benjamini-Hochberg adjusted p ≤ 0.01) different between intestinal tissue samples of necrotizing enterocolitis (NEC) patients with low necrosis (n=18) and high necrosis (n=28). A total of 232 biological processes were identified, and those with the 20 lowest false discovery rate (FDR) values are shown. Redundant terms were removed at a Jaccard similarity cutoff of 0.5.
Table 1. Gene ontology biological processes enrichment analysis of genes significantly differentially methylated between intestinal tissue samples with low and high levels of necrosis. Enrichment analysis of 1,688 genes with three or more methylation sites that were significantly (Benjamini-Hochberg adjusted p ≤ 0.01) different between intestinal tissue samples of necrotizing enterocolitis (NEC) patients with low necrosis (n=18) and high necrosis (n=28). A total of 232 biological processes were identified, and those with the 20 lowest false discovery rate (FDR) values are shown. Redundant terms were removed at a Jaccard similarity cutoff of 0.5.

Figure 1. Cluster and pathway analysis of genes differentially methylated between intestinal tissue samples of survivors and non-survivors. Cluster and pathway analysis of genes found to have significantly (Benjamini-Hochberg adjusted p ≤ 0.01) different methylation levels when comparing intestinal tissue samples of survivors (n=34) and non-survivors (n=12) of necrotizing enterocolitis (NEC). A total of 3,205 genes were found to be significantly different, and 150 with the greatest total difference in methylation were included in the analysis. Gene symbols and the number of significant sites within each gene are listed. Lines represent protein-protein interactions. Line width represents the confidence of each interaction. Clustering was performed by Molecular Complex Detection (MCODE). Pathway analysis was performed by Search Tool for the Retrieval of Interacting Genes/Proteins (STRING). FDR = false discovery rate.
Figure 1 nec epigenetics survivors vs non survivour pas25.jpegFigure 1. Cluster and pathway analysis of genes differentially methylated between intestinal tissue samples of survivors and non-survivors. Cluster and pathway analysis of genes found to have significantly (Benjamini-Hochberg adjusted p ≤ 0.01) different methylation levels when comparing intestinal tissue samples of survivors (n=34) and non-survivors (n=12) of necrotizing enterocolitis (NEC). A total of 3,205 genes were found to be significantly different, and 150 with the greatest total difference in methylation were included in the analysis. Gene symbols and the number of significant sites within each gene are listed. Lines represent protein-protein interactions. Line width represents the confidence of each interaction. Clustering was performed by Molecular Complex Detection (MCODE). Pathway analysis was performed by Search Tool for the Retrieval of Interacting Genes/Proteins (STRING). FDR = false discovery rate.