Morselli M et al. (2023), Age-associated DNA methylation changes in Xenop...

XB-ART-59756

Epigenetics 2023 Dec 01;181:2201517. doi: 10.1080/15592294.2023.2201517.

Show Gene links Show Anatomy links

Age-associated DNA methylation changes in Xenopus frogs.

Morselli M , Bennett R , Shaidani NI , Horb M , Peshkin L , Pellegrini M .

???displayArticle.abstract???
Age-associated changes in DNA methylation have been characterized across various animals, but not yet in amphibians, which are of particular interest because they include widely studied model organisms. In this study, we present clear evidence that the aquatic vertebrate species Xenopus tropicalis displays patterns of age-associated changes in DNA methylation. We have generated whole-genome bisulfite sequencing (WGBS) profiles from skin samples of nine frogs representing young, mature, and old adults and characterized the gene- and chromosome-scale DNA methylation changes with age. Many of the methylation features and changes we observe are consistent with what is known in mammalian species, suggesting that the mechanism of age-related changes is conserved. Moreover, we selected a few thousand age-associated CpG sites to build an assay based on targeted DNA methylation analysis (TBSseq) to expand our findings in future studies involving larger cohorts of individuals. Preliminary results of a pilot TBSeq experiment recapitulate the findings obtained with WGBS setting the basis for the development of an epigenetic clock assay. The results of this study will allow us to leverage the unique resources available for Xenopus to study how DNA methylation relates to other hallmarks of ageing.

???displayArticle.pubmedLink??? 37092296
???displayArticle.pmcLink??? PMC10128463
???displayArticle.link??? Epigenetics
???displayArticle.grants??? [+]

Species referenced: Xenopus tropicalis
Genes referenced: atrx clock dmap1 dnmt1 dnmt3a pcna
GO keywords: DNA methylation

???displayArticle.gses??? GSE221656: NCBI

GSE222107: NCBI
GSE222108: NCBI

???attribute.lit??? ???displayArticles.show???

	Figure 1. Cytosine methylation levels in CpG context. (A) Distribution for X. tropicalis of three different age groups and human mammary epithelial tissue. For each X. tropicalis age group the average of three different individuals. Common sites with at least 5× coverage were used. The human mammary epithelial dataset is from ENCODE (doi:10.17989/ENCSR656TQD, file ENCFF699GKH). (B) CpG methylation distribution over chromosomes for each age group (average of all samples within the same group). Vertical gold bars represent centromere positions. Common sites for all samples with at least 3× coverage were used. Resolution 500Kb.
	Figure 2. CpG methylation distribution over genomic features. (A) Gene metaplot with upstream and downstream 2Kb. In the inset, a zoom in of the central part of the gene body (from the second decile to the TTS). (B) Gene body CpG methylation in repeated elements bigger than 1Kb [not distinguished by class nor family]. (C) Principal Component Analysis (showing PC1 vs. PC2) using CpG methylation levels for the nine frog samples. CpG sites with at least 5× coverage were used and filtered to remove the variables with low variance (PCAtools, removeVar = 0.1). (D) CpG methylation preliminary clock leave-one-out predictions. Each point represents the age prediction on one sample from an elastic net model which was trained on the other eight samples. For all panels, common sites for all samples with at least 5× coverage were used. The boxplots in the insets show the mean of the distribution.
	Figure 3. Selection of CpG sites correlated with age. (A) Pearson correlation (r) distribution of the approximately 3400 CpG sites selected. (B) CpG methylation levels of the selected sites (same as in A). Sites are grouped in five clusters, based on the clustering (method = ‘complete’). (C) CpG methylation levels of the top 100 sites correlated with age (right, gold), and anti-correlated (left, blue) (top). Distribution of the closest genomic elements (both genes and repeats) for each site (bottom).

	Supplementary Figure 2: Predictive CpG sites. (A) Number of CpG sites used in N models. E.g. The CpG sites in the category “Number of models = 9” correspond to the sites employed by all 9 models; the CpG sites in the category “Number of models = 5” correspond to the sites employed in 5 models only (9 models total). (B) Distribution of the closest genomic elements for all CpG sites used in at least one model (left, corresponding to all the 331 CpG sites) and for the CpG sites common to all 9 models (right, corresponding to the black bar in Supplementary Figure 2A, for a total of 57 CpG sites). Refer to Supplementary Table 2 for the list of the predictive CpG sites.
	Supplementary Figure 3: (A) Distribution of distances between selected regions (refer to Supplementary Table 3) and genome elements: genes (left) and repeats (right). (B) Distribution of the closest genomic elements (both genes and repeats) for each selected region.
	Supplementary Figure 4: (A) Overlap between the 331 predictive CpG sites (divided by the number of models they appear in) and the 3441 selected regions (light blue). (B) DNA methylation levels of the predictive CpG sites not overlapping with the selected regions (green gradient). (C) DNA methylation levels of the predictive CpG sites overlapping with the selected regions (purple gradient).
	Supplementary Figure 5: (A) Correlation (Pearson) between samples’ age and Principal Components (PC), p-value and adjusted p-value (Holm method). (B) Scatterplots between PCs and samples’ age.
	Supplementary Figure 6: Autocorrelation of CpG DNA methylation for the closest 20 CpG sites (Lag). Lag =0 (the site itself) has been omitted for all samples (r =1).
	Supplementary Figure 7: Pilot Targeted Bisulfite Sequencing (TBSeq) on 16 X. tropicalis samples using the selected age-associated CpG sites. (A) Principal component analysis using common CpG sites covered by at least 100 reads filtered to remove the variables with low variance (PCAtools, removeVar = 0.1). Samples are colored based on age. (B) Pearson correlation followed by hierarchical clustering using all 100x covered common CpG sites. (C) Heatmap showing the DNA methylation levels of the most variable CpG sites (approximately 500 sites).

References [+] :

Anastasiadi, A clockwork fish: Age prediction using DNA methylation-based biomarkers in the European seabass. 2020, Pubmed