Introduction Birthweight is a known risk factor for diseases in adult life however; the biomolecular mechanisms leading to disease processes remain poorly understood. DNA modifications, transcripts, proteins and metabolites have the potential to imprint biological responses during early life and drive adult risk profiles to chronic conditions. The aim of this study is to enhance the understanding of the complex pathways related to birthweight through integration of multiple OMIC techniques. Methods This study combines methylation (n = 468,408 CpGs), transcriptomics (n = 29,164 transcripts), inflammatory proteins (n = 16) and metabolomics (n = 4712 metabolic features) profiles from cord blood of participants in the EXPOsOMICS birth cohort consortium including ENVIRONAGE (n = 200), INMA (n = 85), Rhea (n = 99) and Piccolipiu (n = 99). We computed Pearson correlation coefficients between pairs of OMICs. Specifically, using a metabolites and methylation driven approaches, we respectively correlated metabolic (n = 68) and epigenetic (n = 8696) signatures of birthweight identified in previous EXPOsOMICS birth cohort study with the other omics platforms. The OMIC platforms have been denoised using technical variables, pairs were considered significant after Bonferroni correction for multiple testing. Significant correlations were visualized in a globe plot with the Circos visualization toolkit version 0.69–6. Transcripts that were significantly correlated to metabolites and to CpGs sites were studied in an overrepresentation analysis (ORA) using DAVID software ( https://www.david.ncifcrf.gov ). Results In the metabolites driven approach, of 33,835,984 correlations between metabolomics and the other levels, 20,097 (0.06%) were significant, mainly inverse (94%) and with a mean absolute correlation coefficient of 0.30 (range = 0.19–0.57). Over the 68 metabolites under study, 13 were significantly correlated with inflammatory proteins, 9 with the transcriptome and 20 with the methylome for a total of 28 unique metabolites. A globe plot of significant correlations showed that all the 9 metabolites were positively correlated with the transcriptome and had an inverse correlation with methylome. Few overlaps (n = 5) were detected in metabolites correlated with both methylome and proteins; the correlations in both pairs went in the same direction (mainly positive). ORA of the 38 genes associated to the significant 41 transcripts resulted in 8 significant pathways, including immune response (such as NK mediated cell cytotoxicity), signaling (MAPK signaling pathway) and also diseases-related (diabetes mellitus and thyroid autoimmunity) pathways. In the methylation driven approach, 0.02% (n = 66,719) of a total 293,606,396 correlations were identified as significant with correlation coefficients having absolute mean of 0.42 (range 0.23–0.69), and mostly negative (67%). Over the 8,663 CpG sites under study, 57 were significantly correlated with proteins, 1,467 with the transcriptome and 2174with the methylome. The globe plot of the significant correlations showed that 13 CpGs appeared in both transcripts and proteins correlation sets, 27 CpGs were common to metabolites and proteins correlation sets, and 791 CpGs were common to metabolites and transcripts correlation sets. Significant correlations in the metabolites and methylation driven approaches shared 34 transcripts, 14 metabolites, 529 CpG sites and 3 proteins. ORA among the 333 genes associated to the 391 unique transcripts significantly correlated to the methylome identified 61 pathways including 4 over the 9 pathways detected in the metabolites driven analysis (including NK mediated cell cytotoxicity and antigen processing and presentation), and also hormone (progesterone-mediated oocyte maturation) and nitrogen metabolism pathways. Conclusion This study provides initial insights into the molecular mechanisms related to birthweight and the biological pathways potentially leading to adverse health effects.
Read full abstract