Garfield DA et al. (2013), The impact of gene expression variation on the ...

ECB-ART-43107

PLoS Biol 2013 Oct 29;1110:e1001696. doi: 10.1371/journal.pbio.1001696.

Show Gene links Show Anatomy links

The impact of gene expression variation on the robustness and evolvability of a developmental gene regulatory network.

Garfield DA , Runcie DE , Babbitt CC , Haygood R , Nielsen WJ , Wray GA .

Abstract
Regulatory interactions buffer development against genetic and environmental perturbations, but adaptation requires phenotypes to change. We investigated the relationship between robustness and evolvability within the gene regulatory network underlying development of the larval skeleton in the sea urchin Strongylocentrotus purpuratus. We find extensive variation in gene expression in this network throughout development in a natural population, some of which has a heritable genetic basis. Switch-like regulatory interactions predominate during early development, buffer expression variation, and may promote the accumulation of cryptic genetic variation affecting early stages. Regulatory interactions during later development are typically more sensitive (linear), allowing variation in expression to affect downstream target genes. Variation in skeletal morphology is associated primarily with expression variation of a few, primarily structural, genes at terminal positions within the network. These results indicate that the position and properties of gene interactions within a network can have important evolutionary consequences independent of their immediate regulatory role.

PubMed ID: 24204211
PMC ID: PMC3812118
Article link: PLoS Biol
Grant support: [+]

Species referenced: Echinodermata
Genes referenced: arid3a erg foxb1 gata6 impact LOC100887844 LOC100893907 LOC105438357 LOC115919910 LOC582915 tgif2l

Article Images: [+] show captions

	Figure 1. Developmental gene regulatory network of S. purpuratus.(A) Development progresses from the egg (top), through cleavage and gastrulation (middle), to a free living larva capable of feeding (bottom). Skeletogenic cell lineage indicated in red, skeleton in solid black. The seven post-fertilization stages and times (hours) shown correspond to time points 1–7 discussed throughout this article. (B) The gene regulatory network is initiated by maternal transcripts and proteins (top) that activate a cascade of subsequent gene regulatory interactions (see text for references). Names of genes assayed in this study are shown in black, other genes in gray. Solid lines, colored to allow visual separation, denote experimentally verified direct molecular interactions among genes: transcription factor:DNA binding (arrows = activators, bars = repressors) or ligand:receptor interactions (nested arrowheads pointing to receptor). Distinct spatial territories of cell fates specific in the embryo are indicated by colored background and name. Based on [13]–[24].
	Figure 2. Parent-of-origin effects on gene expression profiles.Changes in transcript abundance across seven developmental stages are plotted for each family for four representative genes. Families are color-coded by parent of origin: dam in (A, C, D) and sire in (B). In each case, gene expression profiles in the families derived from one parent stand out as distinct from all the other families (color versus grey in magnified time segments to the right of each plot; yellow rectangles indicate magnified portion).
	Figure 3. Parental components to gene expression variation.Median female and male parental contributions to scaled expression variation (variance/mean2; see Methods) are plotted for each time point on a log scale to facilitate visualization (analyses were carried out on untransformed values using non-parametric tests); whiskers on each bar indicate quartiles. Maternal effects significantly exceed paternal effects only at the first time point and paternal effects are relatively uniform across development.
	Figure 4. Changes in regulatory interactions across development.(A) Mean correlation (r2) in expression between pairs of genes are plotted for known direct regulator-target interactions (red) and random pairs of genes not thought to interact (purple). Error bars for random interactions are based on boot strap replicates carried out independently for each stage; values for known interactors are counts at each stage and thus no error bars are shown. Asterisks denote significant differences between interactors and random pairs of genes (p<0.01, by permutation, for all but time point 3, for which p<0.05). Note that known interactors are no more correlated than random pairs of genes during early development, but become more highly correlated from time point 3 onwards. (B) The proportion of sensitive regulatory interactions among known interactors is plotted. Many regulatory interactions among zygotically expressed genes are insensitive (i.e., switch-like) during embryogenesis, with an increasing proportion of sensitive (i.e., quantitative) interactions at later time points.
	Figure 5. Correlations in gene expression levels among direct interactors.Scatter plots of expression levels for pairs of upstream regulators (x-axis) and direct targets (y-axis). Expression levels and regulatory interactions are color-coded by developmental time point (bottom). Regression lines are shown at active time points. Regulatory inputs to the downstream gene at active time points are drawn to the right of each plot using the same color-coding for time points. Information about active/inactive edges is not available for time point 7, which is therefore omitted. Note that gene regulatory interactions differ qualitatively. (A) Some are roughly linear: increased expression of the upstream gene (GataE) is correlated with increased expression of its direct target (Fmo1,2,3). (B) Other interactions are more switch-like: beyond a certain level, differences in expression of the upstream gene (Dri) has little impact on the expression level of its direct target (CyP). (C) Regulatory interactions can also change during development. Expression of Hex is sensitive to Tgif expression levels during all four active time points, but is more sensitive (steeper slope) when also receiving input from Erg (time points 3 and 4) than when it is not (time points 5 and 6).
	Figure 6. Correlations between gene expression and larval morphology.(A) Gene regulatory sub-network in skeletogenic cells (see Figure 1 for broader network context). Yellow boxes indicate genes encoding regulatory proteins; purple boxes indicate genes encoding structural proteins of the skeleton and surrounding matrix. These boxes correspond to rectangles in the remaining panels, with a horizontal line separating the two classes of genes. (B–E) Results of tests for correlations between variation in gene expression and variation in skeletal morphology. Gray indicates no correlation; color indicates correlation with expression from a single time point; black indicates a correlation based on multiple time points (see Text S1). (B) Morphological associations with expression based on PCA. SM30-E is related to PC I (primarily length), FoxB and Hex with PC III (primarily aspect ratio). (C) Morphological associations based on partial least squares analysis. Very early effects (time point 1) operate through regulators high in the network. (D) Morphological associations based on weighted contributions by partial least squares analysis. Four genes show associations from early stages and three from late stages. (E) Morphological associations that are conservatively based solely on male genetic contributions. The three strongest associations come from late expression. Note that SM30-E is identified in all four analyses.

References [+] :

Balhoff, Evolutionary analysis of the well characterized endo16 promoter reveals substantial variation within functional sites. 2005, Pubmed, Echinobase