Search

Additional file 1: Figure S1. Geographical distribution and genetic diversity of Japanese medaka strains, as confirmed by genome resequencing. Figure S2. Quantification of phenotypic variation, its read-depth dependency and performance to classify different samples. Figure S3. Selecting genes with deviations significantly higher than technical errors. Figure S4. Whole embryonic phenotypic variations evaluated in various categories of gene sets. Figure S5. Expression-level differences of each gene in wild strains and inbred twins and correction for potential bias in gene expression variation. Figure S6. Representative developmental genes in the 10% of those with the highest or the lowest stability in gene expression levels. Figure S7. Genes with pleiotropic expression tend to have greater stability and higher conservation in microevolution. Figure S8. Features of the potential regulatory region did not significantly correlate with either gene expression stability or microevolutionary conservation. Figure S9. GO slim terms enriched in the 10% of genes with the least expression variation.

Additional file 1 Table S1. Information on whole-embryo ATAC-seq samples. Table S2. Numbers of ACRs at each developmental stage. Table S3. FRiP score of whole-embryo ATAC-seq data. Table S4. The numbers of ACRs at promoters and the two-sided Fisher’s exact test. Table S5. Biological reproducibility of whole-embryo ATAC-seq signal intensities. Table S6. Information on whole-genome pairwise alignment data. Tables S7 and S8. Statistical information in Figs. 3 and 4. Table S9. Detailed information on representative enhancers from the VISTA Enhancer Database. Tables S10–S22. Statistical information in Figures S7–S13.

Additional file 3 Text S1. The phylotypic period in medaka embryogenesis. Text S2. Testing the recapitulative pattern with the different datasets.

Additional file 2 Figure S1. Identification of conserved mid-embryonic stages during medaka embryogenesis by transcriptome similarity. Figure S2. Genomic distribution of mouse, chicken, and medaka ACRs with respect to genome annotations. Figure S3. Enhancers that drive wider expression tend to have higher ATAC-seq signals. Figure S4. Four methods for estimating evolutionary ages of ACRs. Figure S5. Categorization of evolutionary ages of ACRs and expressed protein-coding genes did not differ among different methods. Figure S6. The recapitulative pattern was also observed for relative sequence length of evolutionarily categorized ACRs within the genome. Figure S7. Developmental gene expression levels did not show a recapitulative pattern in the analysis including genes lost secondarily, related to Fig. 4. Figure S8. The recapitulative pattern of whole-embryo chromatin accessibility was consistent between methods I, II, III, and IV, related to Fig. 3. Figure S9. Essentially the same recapitulative pattern was observed in the analysis using different sets of species. Figure S10. Essentially the same recapitulative pattern was observed for the analyses with different criteria in filtering ATAC-seq reads. Figure S11. Chromatin accessibility of mouse early stages did not show the recapitulative pattern. Figure S12. Exonic ACRs did not follow a recapitulative pattern. Figure S13. Similar recapitulative pattern observed for the chromatin accessibility under strong negative selection. Figure S14. No ACR could be detected at three of five representative regulatory regions associated with taxon-specific features.

The relationship between development and evolution is a central topic in evolutionary biology1,2. Recent transcriptome-based studies support the developmental hourglass model, which predicts that the animal embryogenetic program is most strongly conserved at mid-embryonic stages3-9. This model does not necessarily contradict the classical hypothesis10,11that animal development recapitulates its evolutionary history after the mid-embryonic stages2,12. However, to date there is no molecular evidence supporting the hypothesis that gene-expression profiles that are more evolutionarily derived appear sequentially in late development. Here, by estimating activated genomic regions and their evolutionary origins, we show that the recapitulative pattern appears during late embryonic stages. We made a genome-wide assessment of accessible chromatin regions throughout embryogenesis in three vertebrate species (mouse, chicken, and medaka) and determined the phylogenetic range at which these regions were shared. In all three species, sequential activation of putative regulatory regions that were more derived occurred later in embryogenesis, whereas ancestral ones tended to be activated early. Our results clarify the chronologic changes in accessible chromatin landscapes and reveal a phylogenetic hierarchy in the evolutionary origins of putative regulatory regions that parallels developmental stages of activation. This relationship may explain, at least in part, the background for morphological observations of recapitulative events during embryogenesis.

Additional file: Figure S4. Survival rate did not decrease after UV irradiation in the pharyngula period. Survival curves of zebrafish embryos after UV irradiation in the pharyngula period (Kaplan–Meier method). The horizontal axis is developmental stage and does not reflect actual time length. Orange arrowhead, most conserved developmental period in vertebrates [18, 20]. Black line, control (same group depicted in Fig. 2a); orange line, embryos UV irradiated in the pharyngula period (24 hpf = Prim-5); shaded area, 95% CI; vertical dotted line, UV irradiation. Numbers of embryos used in this analysis: control group, n = 72; treated group n = 50.

Additional file 3: Figure S3. Windowing did not affect lethality in chicken embryos. Rate of lethal phenotypes of chicken embryos treated with TSA. Eggs were windowed and embryos were treated during the gastrula (Gst), pharyngula (Pha), and late (Lat) periods, or eggs were windowed but embryos not treated (Wnd). These are the same data shown in Fig. 1d. Additional data is the lethality of embryos that were not windowed or treated (Intct). Data are means; error bars are SD. *P

Additional file 2: Figure S2. Efficient penetration of TSA in African clawed frog late embryos. Immunohistochemistry was performed with the anti-histone H3 acetyl K27 antibody. (a-1)–(f-1) Lateral views of whole st. 40 embryos. Arrowheads indicate the position of the cross section and the observed plane. (a-2)–(f-2) Transverse section through the trunk of st. 40 embryos. Areas included in the magenta and green squares are shown at higher magnification in (a-3)–(f-3) and (a-4)–(f-4), respectively. nt, neural tube; nc, notochord. (a-3)–(f-3) Neural tube and notochord of st. 40 embryos. (a-4)–(f-4) Trunk of st. 40 embryos. The experiment was performed two times (n = 5 in each replicate). The results of each experiment are shown separately, as DAB staining reaction proceeds rapidly and staining intensity is variable among experiments.

Additional file 1: Figure S1. Developmental dynamics of zebrafish embryos showing normal development and examples of embryonic death. After UV irradiation, embryonic survival was tracked by time-lapse imaging in 1-h intervals until the hatch period (at least 60Â h). The elapsed time from the start of recording is indicated in the upper left corner in each panel. (a) Entire time-lapse sequences of the typical normal development of a zebrafish embryo. (b, c) Examples of embryonic death; in these cases, we determined that embryonic death occurred at 6Â h (b) and 19Â h (c), respectively. In both frames, critical deformation and cessation of development were observed.