Click here to close
Hello! We notice that you are using Internet Explorer, which is not supported by Xenbase and may cause the site to display incorrectly.
We suggest using a current version of Chrome,
FireFox, or Safari.
Developmental and Injury-induced Changes in DNA Methylation in Regenerative versus Non-regenerative Regions of the Vertebrate Central Nervous System.
Reverdatto S
,
Prasad A
,
Belrose JL
,
Zhang X
,
Sammons MA
,
Gibbs KM
,
Szaro BG
.
Abstract
BACKGROUND: Because some of its CNS neurons (e.g., retinal ganglion cells after optic nerve crush (ONC)) regenerate axons throughout life, whereas others (e.g., hindbrain neurons after spinal cord injury (SCI)) lose this capacity as tadpoles metamorphose into frogs, the South African claw-toed frog, Xenopus laevis, offers unique opportunities for exploring differences between regenerative and non-regenerative responses to CNS injury within the same organism. An earlier, three-way RNA-seq study (frog ONC eye, tadpole SCI hindbrain, frog SCI hindbrain) identified genes that regulate chromatin accessibility among those that were differentially expressed in regenerative vs non-regenerative CNS [11]. The current study used whole genome bisulfite sequencing (WGBS) of DNA collected from these same animals at the peak period of axon regeneration to study the extent to which DNA methylation could potentially underlie differences in chromatin accessibility between regenerative and non-regenerative CNS.
RESULTS: Consistent with the hypothesis that DNA of regenerative CNS is more accessible than that of non-regenerative CNS, DNA from both the regenerative tadpolehindbrain and frog eye was less methylated than that of the non-regenerative frog hindbrain. Also, consistent with observations of CNS injury in mammals, DNA methylation in non-regenerative frog hindbrain decreased after SCI. However, contrary to expectations that the level of DNA methylation would decrease even further with axotomy in regenerative CNS, DNA methylation in these regions instead increased with injury. Injury-induced differences in CpG methylation in regenerative CNS became especially enriched in gene promoter regions, whereas non-CpG methylation differences were more evenly distributed across promoter regions, intergenic, and intragenic regions. In non-regenerative CNS, tissue-related (i.e., regenerative vs. non-regenerative CNS) and injury-induced decreases in promoter region CpG methylation were significantly correlated with increased RNA expression, but the injury-induced, increased CpG methylation seen in regenerative CNS across promoter regions was not, suggesting it was associated with increased rather than decreased chromatin accessibility. This hypothesis received support from observations that in regenerative CNS, many genes exhibiting increased, injury-induced, promoter-associated CpG-methylation also exhibited increased RNA expression and association with histone markers for active promoters and enhancers. DNA immunoprecipitation for 5hmC in optic nerve regeneration found that the promoter-associated increases seen in CpG methylation were distinct from those exhibiting changes in 5hmC.
CONCLUSIONS: Although seemingly paradoxical, the increased injury-associated DNA methylation seen in regenerative CNS has many parallels in stem cells and cancer. Thus, these axotomy-induced changes in DNA methylation in regenerative CNS provide evidence for a novel epigenetic state favoring successful over unsuccessful CNSaxon regeneration. The datasets described in this study should help lay the foundations for future studies of the molecular and cellular mechanisms involved. The insights gained should, in turn, help point the way to novel therapeutic approaches for treating CNS injury in mammals.
Figure 1. Chromosome-wide overview of injury-induced changes in DNA methylation for regenerative (tadpole) vs. non-regenerative (post-metamorphic frog) hindbrain after spinal cord injury (SCI) for two representative chromosomes (180 Mb of Chr 2L, top; 104 Mb of Chr 9_10S, bottom) as revealed by whole genome bisulfite sequencing (WGBS). Tracks for regenerative tadpole and non-regenerative frog hindbrain are grouped separately (Tad, top; Frog, bottom). For each chromosome, the vertical scales, which indicate the level of methylation (5mC) in each methylation context (CpG, dark green; CHG, olive green; CHH, navy blue), were group-autoscaled across tadpole and frog SCI and controls to facilitate comparisons between injury conditions (SCI vs. control) and developmental stage (tadpole vs. frog). Methylation differences between SCI and control (ΔCpG, ΔCHG, ΔCHH) indicate log2(SCI 5mC/control 5mC). The resulting increased (>0) and decreased methylation (<0) levels are shown in light green vs. blue, above and below the horizontal axes, respectively. Changes in RNA expression between SCI and control are also indicated (ΔRNA-Seq Tad and Frog; log2(SCI/control), with red and blue indicating increased and decreased expression, respectively [11]; note, RNA-Seq and WGBS were performed on RNA and DNA, respectively, isolated from the very same animals. H3K4me3 peaks at gastrulation (st. 10.5, [41]) and the locations of annotated genes (gene models: Mayball [21; 88; 89]; X. laevis v. 9.1 [122]) are also indicated. For all three DNA methylation contexts (CpG, CHG, CHH), methylation levels increased between tadpole and frog stages pervasively across the entire chromosome, and SCI induced opposite, pervasive methylation responses (ΔCpG, ΔCHG, ΔCHH) in tadpole vs. frog [increased (light green) vs. decreased (light blue) methylation, respectively]. As illustrated in these two representative examples, similar patterns were seen for all chromosomes, with no overall differences between L and S homeologous chromosomes
Figure 2. Quantitation of DNA methylation across the genome. Fraction of C’s exhibiting methylation marks in each context (A, % total 5mC; B, %CpG; C, %CHG; D, %CHH), as determined by WGBS, were averaged (±SE) across three biological replicates (5 pooled tadpole and frog hindbrains, 6 pooled frog eyes). One way ANOVA indicated that methylation differed significantly across all groups compared (P < 0.002). Results of post hoc comparisons are indicated by the brackets above (Fisher LSD; *, P < 0.05). See the text for further details concerning differences among conditions
Fig. 3. View of DNA methylation tracks (CpG, CHH, CHG) in hindbrain after SCI for representative genes known to be differentially expressed in successful vs. unsuccessful CNSaxon regeneration [11]. Description of tracks and abbreviations are as in Fig. 1. Whereas changes in CHH and CHG methylation are pervasive across the genome, spanning both inter and intra genic regions, changes in CpG methylation are primarily confined to regions spanning the transcriptional start site (red boxes), where CpG methylation levels are generally lower than elsewhere.
Fig. 4. Degree of CpG methylation surrounding the transcriptional start site (TSS; ±2.0 kb) correlated with RNA expression for regenerative (tadpole) and non-regenerative (frog) hindbrain before and after SCI. Top, average level of CpG-methylation for different levels of gene expression (Q1 to Q4, representing the 25% most highly to least expressed genes, respectively from a total of 45,099 gene models in X. laevis v.9.1 [122]). (+ or -) Δ, indicates the injury-induced changes for genes in Q1. x-axis units, distance from the predicted transcription start site (TSS) in kilobases (kb); y-axis units (Methylation Density), number of 5mCs in a 50 bp bin x 1 million/total number of Cs. Bottom, heatmaps of CpG methylation for each quartile, clustered from highest to lowest. The degree of CpG-methylation exhibited the expected negative correlation with RNA expression, but for the top two quartiles, it increased in regenerative hindbrain and decreased in non-regenerative hindbrain after SCI. The degree of CpG methylation is indicated by the intensity of the color, as indicated to the right of each heatmap (Methylation Density)
Fig. 5. Numbers of genes having promoters (defined as 750 bp upstream to 250 bp downstream of the TSS) harboring differentially CpG-methylated regions (CpG-DMRs) for the various tissue- and injury-related comparisons. Bars indicate the total number of genes (from a total of 45,099 gene models in X. laevis v.9.1 [122]) that harbored such CpG-DMRs. Numbers above each bar indicate the fraction of such genes with DMRs >0 between the first vs. the second listed condition (hyper-methylated CpG DMR; black). CpG DMRs between uninjured regenerative vs. non-regenerative tissues (i.e., tadpolehindbrain and frog eye vs. frog hindbrain, respectively) were predominantly hypomethylated (CpG DMR < 0). With CNS injury (SCI or optic nerve crush (ONC)), CpG DMRs were predominantly hyper-methylated between injury vs. control conditions in regenerative CNS (i.e., tadpole SCI hindbrain and frog ONC eye), and hypo-methylated in non-regenerative CNS (i.e., frog SCI hindbrain)
Fig. 6. Tissue- and Injury-related changes in CpG methylation across the TSS (-2 kb – +2 kb) for all genes in the genome (N = 45,099 gene models), correlated with differential gene expression. Top and bottom panels illustrate average density and heatmaps of CpG methylation across the TSS for genes in each differential expression grouping. Whether a gene fell into the RNA Up, RNA Down, or no significant change category was determined by RNA-seq from RNA samples previously collected [11] from the same animals analyzed here by WGBS. Comparisons indicate changes between the first vs. second conditions (e.g., negative differential methylation and RNA Up indicate hypo-methylation and significantly increased RNA expression, respectively, in the first vs. the second condition, etc.). Injury vs. control heatmaps for non-regenerative CNS (frog hindbrain) resembled those generated by comparisons between regenerative vs. non-regenerative CNS (tadpole and frog eye vs. frog hindbrain, respectively) and the opposite of what was seen with CNS injury in the two regenerative CNS regions (i.e., tadpole SCI hindbrain and frog ONC eye). x-axis units, as in Fig. 4; y-axis units, Δ5mC within a 50 bp bin x 1 million/total number of Cs
Fig. 7. Heatmaps of active histone marks (H3K4me3 and H3K27ac) across the TSS, relative to quartiles of RNA expression (A) and Promoter CpG-DMR’s (B) in regenerative CNS tissues after SCI (left) and ONC (right). A, the presence of active marks declined with decreasing levels of RNA expression (Q1 to Q4, most to least). B, active histone marks were more highly represented among DMRs exhibiting increased methylation with injury (DMR Up) than among those exhibiting decreased methylation (DMR DN). See Fig. 5 for total numbers of DMR-bearing genes in each category. x-axis units as in Fig. 4; y-axis units, number of mapped reads in a 50 bp bin x 1 million/total number of mapped reads
Fig. 8. Gene Ontologies (Metascape [116]) for genes exhibiting increased CpG-DMRs after SCI in tadpole (CNS axon regenerative) hindbrain. The twenty highest ranking categories (i.e., lowest probability (P) of arising by chance) are listed separately for up-regulated (top) and down-regulated (bottom) genes. Up-regulated genes were highly enriched for categories representing genes associated with DNA replication, repair, and covalent modification (red boxes). Injury-induced genes that decreased in expression had many genes associated with physiological functions and developmental functions (see text). Total numbers of genes in the top and bottom panels were 738 and 460, respectively
Fig. 9. In optic nerve injury, changes in 5-hydroxymethylcytosine are distinct from those of CpG 5mC. Chr 2L and 9_10S: as was seen for SCI in tadpolehindbrain (Fig. 1, illustrated for the same two chromosomes), ONC-induced increased CpG methylation (ΔCpG MS ONC: log2(MS-CpG ONC/MS-CpG control eye)) pervasively across the entire chromosome (light green vs light blue for increased vs. decreased CpG MS, respectively). In contrast, 5hmC (Δ5hmC ONC) exhibited both increases and decreases spread across the chromosome. ina.s and nefm.s, two neuronal intermediate filament genes that increase with ONC in retinal ganglion cells, illustrate the complexity of 5-hydroxymethylation. Red boxes, regions encompassing the TSS marked by active histone marks (H3K4me3, H3K27ac). Green box, region encompassing the TSS, that exhibited increased CpG methylation (ΔCpG MS ONC, light green). 1, blue box, in ina.s, region near the TSS exhibiting increased 5hmC with injury (Δ5hmC ONC, olive green), which mostly flanked that marked for increased CpG 5mC. Note, nefm.s, exhibited little to no 5hmC marks in the corresponding region. Blue ellipses: 2, 5hmC marks at the borders of intron 2 of ina.s. 3, 5hmC marking integrated adenoviral retroviral sequences in ina.s. 4, 5hmC marking the template strand of a region of exon 3 of nefm that is highly enriched for repetitive glutamates. See text for more details
Fig. 10. Additional examples of 5hmC vs CpG 5mC for representative genes changing in expression during ONC. Track labels, as well as red, green, and blue boxes, are as in Fig. 10. Magenta boxes, examples of increased H3K27ac marks induced by ONC. All genes exhibited the same behavior with respect to CpG as illustrated previously for ezh2.L, vim.L, idh1.L and jarid2.S in tadpolehindbrain after SCI (Fig. 3). Changes in 5hmC (Δ5hmC ONC: olive green, up; blue green, down) across the TSS were generally not congruent with those of CpG 5mC marks (blue boxes vs. green boxes, respectively). Blue ellipse, vim.L, an integrated adenoviral sequence marked by 5hmC
Fig. 11. Distributions of CpG 5mC surrounding the transcriptional start site (TSS; ± 2.0 kb) to ONC differed markedly from that of 5hmC. A, Although less pronounced than in tadpole SCI, the density of CpG DMRs across the TSS follows the same pattern as for tadpole SCI, with CpG 5mC DMRs increasing for the top two quartiles of RNA expression with injury. Units and labels are as in Fig. 4. B, In contrast, 5hmC showed no such response, and overall was markedly different from the pattern of CpG 5mC (see text for details). Units and labels are as in Fig. 7
