Polygenicity and Epistasis Underlie Fitness-Proximal Traits in the Caenorhabditis elegans Multiparental Experimental Evolution (CeMEE) Panel

Abstract

Understanding the genetic basis of complex traits remains a major challenge in biology. Polygenicity, phenotypic plasticity, and epistasis contribute to phenotypic variance in ways that are rarely clear. This uncertainty can be problematic for estimating heritability, for predicting individual phenotypes from genomic data, and for parameterizing models of phenotypic evolution. Here, we report an advanced recombinant inbred line (RIL) quantitative trait locus mapping panel for the hermaphroditic nematode Caenorhabditis elegans, the C. elegans multiparental experimental evolution (CeMEE) panel. The CeMEE panel, comprising 507 RILs at present, was created by hybridization of 16 wild isolates, experimental evolution for 140-190 generations, and inbreeding by selfing for 13-16 generations. The panel contains 22% of single-nucleotide polymorphisms known to segregate in natural populations, and complements existing C. elegans mapping resources by providing fine resolution and high nucleotide diversity across > 95% of the genome. We apply it to study the genetic basis of two fitness components, fertility and hermaphrodite body size at time of reproduction, with high broad-sense heritability in the CeMEE. While simulations show that we should detect common alleles with additive effects as small as 5%, at gene-level resolution, the genetic architectures of these traits do not feature such alleles. We instead find that a significant fraction of trait variance, approaching 40% for fertility, can be explained by sign epistasis with main effects below the detection limit. In congruence, phenotype prediction from genomic similarity, while generally poor ([Formula: see text]), requires modeling epistasis for optimal accuracy, with most variance attributed to the rapidly evolving chromosome arms.

Keywords: GWAS; MPP; Multiparent Advanced Generation Inter-Cross (MAGIC); QTL; body size; complex trait; epistasis; experimental evolution; fertility; genetic architecture; heritability; multiparental populations; polygenicity; quantitative trait; selfing.

Figure 1

CeMEE derivation. The multiparental intercross funnel phase comprised four stages of pairwise crosses and progeny mixing, carried out in parallel at controlled population sizes. One hybridization cycle for a single-founder cross is inset at left: in each cycle, multiple reciprocal crosses were initiated, increasing in replicate number and census size each filial generation. $F_1$ and $F_2$ progeny were first sib-mated, then reciprocal lines were merged by intercrossing the $F_3$ and expanding the pooled $G_4$ (for three to four generations) before commencing the next reduction cycle. The resulting multiparental hybrid population was archived by freezing, and samples were thawed and maintained for 140 nonoverlapping generations of mixed selfing and outcrossing under standard laboratory conditions to generate the A140 population. Hermaphrodites were then sampled from the A140 and selfed to generated the A140 RILs. Additionally, the outbred A140 population was evolved for a further 50 generations under the same conditions (CA) or under adaptation to a salt gradient with varying sex ratios (GT, GM, and GA lines; Theologidis et al. 2014). See Materials and Methods for description of subpanels, and Teotónio et al. (2012) for details of replicate numbers and population sizes for each funnel generation. CA, control adapted lines; CeMEE, C. elegans multiparental experimental evolution panel; GA, gradual adaptation androdioecious; GM, gradual adaptation monoecious; GT, gradual adaptation trioecious; RILs, recombinant inbred lines.

Figure 2

MAFs of founders and the outbred A140 population (A), A140 and RILs [inbreeding only for the A140 RILs, further adaptation then inbreeding for G50 RILs; (B)], and founders against all RILs (C). Insets show frequency quantiles. Changes in major/minor class across contrasts are ignored (among these cases, unfolded frequencies for just 3699 sites differ by > 50% for founders vs. RILs). (D) Change in allele frequency (absolute log ratios) for the same contrasts by functional class: intronic, synonymous and nonsynonymous, putative regulatory variation (US/DS; ≤ 200 bp from an annotated transcript or N2 pseudogene), or intergenic (none of the above). Points are mean values (diameter exceeds the SE). DS, downstream; MAF, minor allele frequency; RILs, recombinant inbred lines; US, upstream.

Figure 3

Linkage disequilibrium in founders (A) and all CeMEE RILs [(B); F${}_2$ genetic map distance, LOESS fit to mean $r^2$]. (C) Interchromosomal structure is weak but significant. Observed mean $r^2$ between all chromosomes (red vertical bar) plotted against the null distribution from permutations randomizing lines across chromosomes (within subpanels to exclude effects of population structure). (D) Permutations dropping pairs of chromosomes implicate X–autosome interactions. Color and size are (redundantly) scaled by enrichment over the null distribution (95% percentile), relative to the genome-wide mean value. CeMEE, C. elegans multiparental experimental evolution panel; LOESS, LOcal regrESSion; RILs, recombinant inbred lines.

Figure 4

A140 RIL founder haplotype reconstruction and structure for chromosomes I (A), IV (B), and X (C). Founder haplotypes in physical and genetic distances (in subpanels labeled “a”). Each plotted point is a marker, with its size scaled by posterior probability (minimum 0.2; regions of low marker density are visible as vertical white swathes through RIL haplotypes). Founder contributions are summarized below (in subpanels labeled “b”). Loci discussed in the text are indicated: the zeel-1/peel-1 incompatibility on the left arm of chromosome I (haplotype compatibility group), either experimentally tested in Seidel et al. (2008) or determined here from genotype data, is indicated below as an arrowhead for Bristol (N2) or an “x” for Hawaii (CB4856); extreme haplotype differentiation within a piRNA cluster on the right arm and tip of chromosome IV; and the fixation of N2/CB4507 haplotypes over a large region of the X chromosome left arm spanning npr-1, alleles of which have pleiotropic effects on behavior and laboratory adaptation (de Bono and Bargmann 1998; Gloria-Soria and Azevedo 2008; McGrath et al. 2009; Andersen et al. 2014). Subpanels “c–g” show summary statistics evaluated at 5 kb or 0.01 cM resolution, with vertical scales for each metric fixed across chromosomes, and the positions of recombination rate boundaries inferred for the N2 $\times$ CB4856 RIAILs (Rockman and Kruglyak 2009) indicated with shaded bars. Haplotype length: mean length extending from the focal position (in subpanels labeled “c”). P (haplo.): test of reconstructed founder haplotype proportions, relative to expectation based on reconstruction frequency from G${}_{150}$ simulations ($-lo{g}_{10}\left(P\right)$ from a ${\chi }^2$ goodness-of-fit test) (in subpanels labeled “d”). t (geno.): change in allele frequency from the founders (absolute value of Welch’s t statistic for founder vs. RIL genotype counts) (in subpanels labeled “e”). N haplo.: the number of unique founder haplotypes detected at each position, with the maximum value of 16 indicated (in subpanels labeled “f”). N RILs: the number of RIL haplotypes reconstructed at each interval ($>0.2$ posterior probability), with the maximum value of 178 indicated (in subpanels labeled “g”). RIAIL, recombinant inbred advanced intercross line; RIL, recombinant inbred line.

Figure 4

A140 RIL founder haplotype reconstruction and structure for chromosomes I (A), IV (B), and X (C). Founder haplotypes in physical and genetic distances (in subpanels labeled “a”). Each plotted point is a marker, with its size scaled by posterior probability (minimum 0.2; regions of low marker density are visible as vertical white swathes through RIL haplotypes). Founder contributions are summarized below (in subpanels labeled “b”). Loci discussed in the text are indicated: the zeel-1/peel-1 incompatibility on the left arm of chromosome I (haplotype compatibility group), either experimentally tested in Seidel et al. (2008) or determined here from genotype data, is indicated below as an arrowhead for Bristol (N2) or an “x” for Hawaii (CB4856); extreme haplotype differentiation within a piRNA cluster on the right arm and tip of chromosome IV; and the fixation of N2/CB4507 haplotypes over a large region of the X chromosome left arm spanning npr-1, alleles of which have pleiotropic effects on behavior and laboratory adaptation (de Bono and Bargmann 1998; Gloria-Soria and Azevedo 2008; McGrath et al. 2009; Andersen et al. 2014). Subpanels “c–g” show summary statistics evaluated at 5 kb or 0.01 cM resolution, with vertical scales for each metric fixed across chromosomes, and the positions of recombination rate boundaries inferred for the N2 $\times$ CB4856 RIAILs (Rockman and Kruglyak 2009) indicated with shaded bars. Haplotype length: mean length extending from the focal position (in subpanels labeled “c”). P (haplo.): test of reconstructed founder haplotype proportions, relative to expectation based on reconstruction frequency from G${}_{150}$ simulations ($-lo{g}_{10}\left(P\right)$ from a ${\chi }^2$ goodness-of-fit test) (in subpanels labeled “d”). t (geno.): change in allele frequency from the founders (absolute value of Welch’s t statistic for founder vs. RIL genotype counts) (in subpanels labeled “e”). N haplo.: the number of unique founder haplotypes detected at each position, with the maximum value of 16 indicated (in subpanels labeled “f”). N RILs: the number of RIL haplotypes reconstructed at each interval ($>0.2$ posterior probability), with the maximum value of 178 indicated (in subpanels labeled “g”). RIAIL, recombinant inbred advanced intercross line; RIL, recombinant inbred line.

Figure 4

A140 RIL founder haplotype reconstruction and structure for chromosomes I (A), IV (B), and X (C). Founder haplotypes in physical and genetic distances (in subpanels labeled “a”). Each plotted point is a marker, with its size scaled by posterior probability (minimum 0.2; regions of low marker density are visible as vertical white swathes through RIL haplotypes). Founder contributions are summarized below (in subpanels labeled “b”). Loci discussed in the text are indicated: the zeel-1/peel-1 incompatibility on the left arm of chromosome I (haplotype compatibility group), either experimentally tested in Seidel et al. (2008) or determined here from genotype data, is indicated below as an arrowhead for Bristol (N2) or an “x” for Hawaii (CB4856); extreme haplotype differentiation within a piRNA cluster on the right arm and tip of chromosome IV; and the fixation of N2/CB4507 haplotypes over a large region of the X chromosome left arm spanning npr-1, alleles of which have pleiotropic effects on behavior and laboratory adaptation (de Bono and Bargmann 1998; Gloria-Soria and Azevedo 2008; McGrath et al. 2009; Andersen et al. 2014). Subpanels “c–g” show summary statistics evaluated at 5 kb or 0.01 cM resolution, with vertical scales for each metric fixed across chromosomes, and the positions of recombination rate boundaries inferred for the N2 $\times$ CB4856 RIAILs (Rockman and Kruglyak 2009) indicated with shaded bars. Haplotype length: mean length extending from the focal position (in subpanels labeled “c”). P (haplo.): test of reconstructed founder haplotype proportions, relative to expectation based on reconstruction frequency from G${}_{150}$ simulations ($-lo{g}_{10}\left(P\right)$ from a ${\chi }^2$ goodness-of-fit test) (in subpanels labeled “d”). t (geno.): change in allele frequency from the founders (absolute value of Welch’s t statistic for founder vs. RIL genotype counts) (in subpanels labeled “e”). N haplo.: the number of unique founder haplotypes detected at each position, with the maximum value of 16 indicated (in subpanels labeled “f”). N RILs: the number of RIL haplotypes reconstructed at each interval ($>0.2$ posterior probability), with the maximum value of 178 indicated (in subpanels labeled “g”). RIAIL, recombinant inbred advanced intercross line; RIL, recombinant inbred line.

Figure 5

Additive QTL mapping simulations. Detection power (A), precision (B), and resolution (C) (2-LOD drop interval size for detected QTL) from single QTL simulations for the full mapping panel of 507 lines, as a function of detection threshold (significance at 0.01, 0.05, and 0.1) and phenotypic variance explained by the simulated QTL. Total heritability of simulated phenotypes is twice that of the focal QTL, with the polygenic contribution spread over 10, 100, or 1000 background markers [plotted in (A) and combined in (B and C)]. In (B) points are mean $\pm$ SE. Mean precision declines with SNP variance at high levels as chance associations reach significance, although the median value (+ symbols) is 1.0 at 5% significance for variants that explain ≥ 7.5% of trait variance. In (C) boxes span the interquartile range, with the median value indicated with a black bar.

Figure 6

One-dimensional GWAS. (A and B) Trait value distributions across RILs (replicate means; bars show data range or the SE for samples with > 2 replicates). Values for the reference N2 strain are shown for comparison. Note that values are raw replicate means on the original scale, and so include all sources of technical variation (unlike model coefficients used for mapping). (C) Single-marker association results for fertility and adult body size (colors as above). α = 0.1 thresholds = 4.38$\times {10}^{-6}$ and 5.57$\times {10}^{-7}$ for size and fertility, with minimum observed P-values of 2.8$\times {10}^{-5}$ and 7.23$\times {10}^{-5}$, respectively. GWAS, genome-wide association study; RIL, recombinant inbred line.

Figure 7

Strong-sign epistasis and highly polygenic interactions contribute to trait variance. (A) The distribution of significant interactions for fertility and size plotted by genetic distance. Pairwise interactions are linked over one-dimensional genome-wide association study test statistics ($-log10\left(P\right)>1$) for each trait (size in blue and fertility in red). Markers with a significant excess of polygenic interactions are indicated with black points. These two-dimensional (2D) sum tests are one-to-many interactions between a single focal marker and all other markers on one other chromosome, with the sum of likelihood ratios significant under a null permutation model. (B and C) and (D and E) show genotype class means for significant size (and fertility) interactions: pairwise tests are in B and D (mean $\pm$ SE), and the individual pairwise tests that contribute to 2D sum tests are in C and E (mean values only). The marker with the highest summed likelihood ratio for each significant chromosome combination is shown. In C and D, line color and intensity is scaled by the interaction F statistic for each interaction according to the scale shown in (E).