Click
here to close Hello! We notice that
you are using Internet Explorer, which is not supported by Echinobase
and may cause the site to display incorrectly. We suggest using a
current version of Chrome,
FireFox,
or Safari.
Front Cell Dev Biol
2021 May 28;9:749963. doi: 10.3389/fcell.2021.749963.
Show Gene links
Show Anatomy links
Derivedness Index for Estimating Degree of Phenotypic Evolution of Embryos: A Study of Comparative Transcriptomic Analyses of Chordates and Echinoderms.
Leong JCK
,
Li Y
,
Uesaka M
,
Uchida Y
,
Omori A
,
Hao M
,
Wan W
,
Dong Y
,
Ren Y
,
Zhang S
,
Zeng T
,
Wang F
,
Chen L
,
Wessel G
,
Livingston BT
,
Bradham C
,
Wang W
,
Irie N
.
???displayArticle.abstract???
Species retaining ancestral features, such as species called living fossils, are often regarded as less derived than their sister groups, but such discussions are usually based on qualitative enumeration of conserved traits. This approach creates a major barrier, especially when quantifying the degree of phenotypic evolution or degree of derivedness, since it focuses only on commonly shared traits, and newly acquired or lost traits are often overlooked. To provide a potential solution to this problem, especially for inter-species comparison of gene expression profiles, we propose a new method named "derivedness index" to quantify the degree of derivedness. In contrast to the conservation-based approach, which deals with expressions of commonly shared genes among species being compared, the derivedness index also considers those that were potentially lost or duplicated during evolution. By applying our method, we found that the gene expression profiles of penta-radial phases in echinoderm tended to be more highly derived than those of the bilateral phase. However, our results suggest that echinoderms may not have experienced much larger modifications to their developmental systems than chordates, at least at the transcriptomic level. In vertebrates, we found that the mid-embryonic and organogenesis stages were generally less derived than the earlier or later stages, indicating that the conserved phylotypic period is also less derived. We also found genes that potentially explain less derivedness, such as Hox genes. Finally, we highlight technical concerns that may influence the measured transcriptomic derivedness, such as read depth and library preparation protocols, for further improvement of our method through future studies. We anticipate that this index will serve as a quantitative guide in the search for constrained developmental phases or processes.
FIGURE 1. Tree based on phenotypic derivedness of embryos. (A) Schematic illustration of a phylogenetic tree drawn by classic viewpoint (left) and that by transcriptomic derivedness (right). Phenotypic derivedness aims to introduce quantitative evaluation of how derived each sample or embryo became from their common ancestor. (B) In contrast to evaluating conservation with commonly shared genes, such as 1:1 orthologs, our method takes advantage of ortholog group-based expression table, which considers the expression of paralogs and potential lost genes.
FIGURE 2. Proposed workflow to estimate phenotypic derivedness of embryos using gene expression profiles and evaluation of methods. (A) Outline of measuring transcriptomic derivedness. Whole embryonic transcriptomes from different species were compared using the expression levels of ortholog groups, which consider not only 1:1 orthologs but also paralogs and genes that are presumably lost in specific lineages. Derivedness index of each embryo was defined as the total branch length from the putative common ancestral node on the inferred tree. (B) Criteria for selecting the suitable method for quantifying derivedness of embryos. These include: [1] clusters samples by species; [2] topology of derivedness tree consistent with known phylogeny estimated from genomic sequences (Kalinka et al., 2010; Hu et al., 2017), with support from biological replicates; [3] Transcriptomic similarities show gradual change along developmental stages. (C) Rank (involved in calculating Spearman distance) and logarithmic normalizations of expression data tend to meet the criteria of clustering stages by species. Shaded boxes represent normalization and distance methods showing this topology in the inferred tree. (D) Spearman distance scores the highest in smoothness analysis. (E) Summary of the most suitable method among the combinations tested.
FIGURE 3. Derivedness tree of developmental stages of echinoderms and chordates based on transcriptomic profiles. (A) Tree showing transcriptomic derivedness of chordate and echinoderm embryos. Embryos of each species are denoted by the same color. Support values are consensus from 100 random biological replicates-included (BRI) trees (see also Methods). Shaded stages in echinoderm species are developmental phases with penta-radial structures. (B) The range of derivedness indices of embryos for each species. Vertebrate species and the two sea urchin species have fairly high derivedness indices whereas C. intestinalis has the lowest (blue: chordate species; purple: echinoderm species; Kruskal-Wallis p < 2e-16) (C) The range of derivedness indices of echinoderm and chordate embryos. This shows that echinoderms do not appear to have much higher derivedness index than chordates. (Mann-Whitney p = 0.75) (D) The range of derivedness indices of the bilateral (green) and the pentaradial (yellow) phases of echinoderm development. The penta-radial phase is more derived in feather star (Oj), green sea urchin (Lv), and purple sea urchin (Sp). (Mann-Whitney U test, **: p < 0.01; ***: p < 0.001; ****: p < 0.0001; ns: p > 0.05). Species abbreviations: Gg, chicken; Ps, soft-shelled turtle; Mm, mouse; Xl, clawed frog; Ol, medaka; Dr, zebrafish; Ci, ascidian; Bf, amphioxus; Oj, feather star; Aj, sea cucumber; Lv, green sea urchin; Sp, purple sea urchin.
FIGURE 4. Derivedness index of embryos of chordates and echinoderms estimated from their respective common ancestors. The range of derivedness indices of each embryo, from the common ancestor of chordates and echinoderms, respectively, in 100 biological replicates-included (BRI) trees. (A–F) The least derived stages in vertebrate species (Gg, Ps, Mm, Xl, Dr, and Ol) are mid-embryonic and organogenesis stages (Gg: HH14-19, Ps: TK11, Mm: E7.5*, Xl: stage 31, Dr: 14-somites, and Ol: stage 23). *: E9.5, in mouse when the Quartz-Seq samples were removed (see Figures 6B–6ii). (G–H) The tunicate C. intestinalis shows relatively lower derivedness indices in stage 1 to stage 10 embryos, and the least derived stage in the amphioxus B. floridae is around the L2 (open-mouth larva) stage. (I–L) In echinoderm species, the least derived stage is around the gastrula in sea cucumber and sea urchins, whereas the 32-cell stage is the least derived in feather star. (Differences in derivedness index for each developmental stage are statistically significant; Friedman test).
FIGURE 5. Characterization of least derived stages by ortholog-groups. (A) Heatmap showing Spearman’s correlation coefficient between expression levels and derivedness indices for each ortholog-group for each species (blue: negative correlation; red: positive correlation). Selected areas in the heatmap are zoomed in to show clusters of ortholog-groups with (from left to right) negative correlation in most species; negative correlation in most vertebrate species; negative correlation only in the two sea urchins; positive correlation in amniotes (mouse, chicken, and turtle). (B) Ortholog-groups showing negative correlation (negative DCOs) in vertebrates. 7,775 ortholog-groups were analyzed for vertebrates, and were further classified into seven categories based on the number of species in which they show negative correlation. 695 ortholog-groups in category 6 (showing negative correlations in all six vertebrate species analyzed) were further analyzed in (C–E). (C) Category 6 negative DCOs, including 18 HOX ortholog-groups (left) and 201 development-related ortholog-groups (right). (D) The putative phenotype of the least derived/conserved mid-embryonic stage of vertebrates. Genes known to express in homologous anatomical structures during this developmental phase in mouse, chicken, frog and zebrafish are highlighted. (E) DCOs with negative correlations in more species tended to show higher degree of temporal pleiotropy, estimated by the ratios of stages detected (TPM ≥ 1). (F) The number of DCOs with negative correlations in 6 vertebrate species (red), C. intestinalis (tunicate; green), and B. floridae (amphioxus; blue). 230 ortholog-groups showed negative correlations in all three groups, suggesting that they might be involved in ancestral functions retained in chordates. (G) 2,414 negative DCOs were detected across echinoderm species (left); however, the functions of most of these genes remain unknown. Well-studied gastrulation-related genes, such as Ets1 and HesC, were among the groups showing positive correlation in feather star (middle) and the sea urchin-specific gene group (right), respectively.
FIGURE 6. Technical concerns in derivedness tree inference. (A) Read depth control. (i) In the tree inferred from 10 M-controlled expression data, the range of derivedness indices significantly increased in some species (Gg, Mm, Ol, Ci, Bf, and Oj). (B) Comparison of TruSeq and Quartz-Seq datasets. (i) The samples of early development in mouse (shaded in blue) and all samples of sea cucumber (not shown) were collected by Quartz-seq. (ii) Removing the Quartz-seq samples changed the tree topology inside the mouse clade. The least derived stage became E9.5. (iii) After removing the Quartz-seq samples, the range of derivedness indices for most species were not greatly affected, not influencing the conclusions we drew in the previous sessions. (iv–v) Tree with all original samples and the new samples to compare TruSeq & Quartz-seq datasets. Unexpectedly, the samples tend to cluster by protocols rather than by stages or biological replicates. This tendency is stronger in (v) where the E9.0 TruSeq samples were omitted, while the E15.5 Quartz-seq and Tru-Seq samples still did not cluster together. Samples from the same starting total RNA but by different protocols (such as E15.5_Tru_1 and E15.5_Qua_1) did not cluster together either. (Shaded in orange: E9.0 by Quartz-seq; green: E9.0 by TruSeq; pink: E15.5 by Quartz-seq; purple: E15.5 by TruSeq; yellow: E9.0 and E15.5 by TruSeq from the published dataset). For (A-i), (A-ii), and (B-iii), Mann–Whitney–Wilcoxon test was performed (*: p < 0.05; **: p < 0.01; ***: p < 0.001; ****: p < 0.0001; ns: p > 0.05).