Natural cryptic variation in epigenetic modulation of an embryonic gene regulatory network

Abstract

Gene regulatory networks (GRNs) that direct animal embryogenesis must respond to varying environmental and physiological conditions to ensure robust construction of organ systems. While GRNs are evolutionarily modified by natural genomic variation, the roles of epigenetic processes in shaping plasticity of GRN architecture are not well understood. The endoderm GRN in Caenorhabditis elegans is initiated by the maternally supplied SKN-1/Nrf2 bZIP transcription factor; however, the requirement for SKN-1 in endoderm specification varies widely among distinct C. elegans wild isotypes, owing to rapid developmental system drift driven by accumulation of cryptic genetic variants. We report here that heritable epigenetic factors that are stimulated by transient developmental diapause also underlie cryptic variation in the requirement for SKN-1 in endoderm development. This epigenetic memory is inherited from the maternal germline, apparently through a nuclear, rather than cytoplasmic, signal, resulting in a parent-of-origin effect (POE), in which the phenotype of the progeny resembles that of the maternal founder. The occurrence and persistence of POE varies between different parental pairs, perduring for at least 10 generations in one pair. This long-perduring POE requires piwi-interacting RNA (piRNA) function and the germline nuclear RNA interference (RNAi) pathway, as well as MET-2 and SET-32, which direct histone H3K9 trimethylation and drive heritable epigenetic modification. Such nongenetic cryptic variation may provide a resource of additional phenotypic diversity through which adaptation may facilitate evolutionary changes and shape developmental regulatory systems.

Keywords: SKN-1; endoderm; epigenetic inheritance; imprinting; parent-of-origin effect.

Fig. 1.

Dauer diapause-stimulated POE. (A) Schematic of the POE assay strategy. (A, Left) To test for a maternal effect, reciprocal crosses using well-fed or postdauer animals were performed on skn-1 RNAi plates, and the fraction of arrested F1 embryos with differentiated gut was examined. Blue and yellow colors represent different wild isotypes. (A, Right) To study the transgenerational phenotypes, reciprocal crosses were performed using different wild isotypes on NGM plates. Embryos were collected from mated hermaphrodites to establish the F1 population. Day 1 (D) or day 2 (C, E, and F and Figs. 2C and 4B and SI Appendix, Fig. S3) gravid worms were treated with alkaline hypochlorite solution to obtain the next generation embryos. L4 hermaphrodites were isolated for skn-1 RNAi assays at each generation (Materials and Methods). (B) Strong maternal effect in F1 embryos from mated skn-1(RNAi) mothers in N2 × JU1491 crosses. At least three independent crosses were performed. Arrowheads indicate phenotypes of N2 (29.2%) and JU1491 (1.2%). The phenotypes of the F1 progeny are not significantly different from those of their respective mother in all crosses, regardless of feeding status. One-sample t test (P > 0.05). (C) POE in skn-1(RNAi) embryos from four generations: F2, F3, F4, and F5 derived from reciprocal N2 × JU1491 crosses. At least five independent crosses were performed for each treatment. Arrowheads indicate the phenotypes of N2 (29.2%) and JU1491 (1.2%). (D) Dauer diapause-induced POE persists for at least 10 generations in JU1172 × MY16 descendants. Three independent crosses were performed. Alkaline hypochlorite treatment was performed on day 1 gravid adults at each generation. Error bars represent ±SEM. Two-way ANOVA: P = 0.08 and 9.14 × 10⁻¹¹ for the well-fed and postdauer experiments (F2 to F10), respectively. (E) Robust POE in skn-1(RNAi) embryos over four generations: F2, F3, F4, and F5 derived from reciprocal JU1172 × MY16 crosses. At least five independent crosses were performed for each treatment. Alkaline hypochlorite treatment was performed on day 2 gravid adults at each generation. Arrowheads indicate the phenotypes of JU1172 (40.0%) and MY16 (2.2%). (F) POE seen in skn-1(RNAi) F5 embryos from reciprocal crosses between MY16 and JU1172. Data points represent independent crosses. Arrowheads indicate the phenotypes of JU1172 (40.0%) and MY16 (2.2%). For B and F, blue text represents experiments with well-fed animals, while red text is for experiments with postdauer animals. Two-sample t test: not significant (NS), P > 0.05; *P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.001. Boxplots represent medians, with range bars showing upper and lower quartiles.

Fig. 2.

POE is not attributable to mitochondrial DNA or cytoplasmic inheritance. (A) Schematic of mitochondrial transfer experiment. Blue and yellow represent different wild isotypes. Five to 10 hermaphrodites were used to backcross to the paternal strain at every generation for 3 generations. (B) POE in skn-1(RNAi) embryos. Data points represent replicates from a single reciprocal cross using postdauer animals. Arrowheads indicate phenotypes of JU1172 (40.0%) and MY16 (2.2%). (C) POE shown by skn-1(RNAi) F5 embryos of reciprocal crosses performed between JU1491 and JU1172. Data points represent independent crosses. Arrowheads indicate phenotypes of original JU1172 (40.0%) and JU1491(1.2%) strains. (D) When PD2227 (N2^{GPR-1 OE}) hermaphrodites are mated to JU1491, non-Mendelian segregation of maternal and paternal chromosomes result in some F1 mosaic animals that express the maternally derived pharyngeal marker (myo-2::mCherry) but lose the body wall muscle (myo-3::mCherry) and germline (mex-5::GFP) markers. F2 self-progeny of the F1 mosaic animals contain only the JU1491 nuclear genome (SI Appendix, Fig. S5). (E) skn-1(RNAi) phenotype of N2/PD2227, JU1491, and F2 and F3 embryos from the mosaic animals. For C and E, blue represents experiments performed using well-fed animals, while red represents experiments performed using postdauer animals. Two-sample t test: not significant (NS), P > 0.05; *P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.001. Boxplots represent medians, with range bars showing upper and lower quartiles.

Fig. 3.

POE is not attributable to competitive fitness or maternal incompatibility. (A, Left) POE shown by skn-1(RNAi) F2 embryos of reciprocal crosses performed between MY16 and JU1172. (A, Right) Individual data points represent lines derived from late F2s of senescent F1 animals. Two independent crosses were performed for each direction. Two-sample t test: not significant (NS), P > 0.05; *P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.001. Boxplots represent medians, with range bars showing upper and lower quartiles. The arrowheads indicate the phenotypes of JU1172 (40.0%) and MY16 (2.2%). (B) Embryonic lethality in F2s. Two independent broods were examined. The total number of embryos scored is shown in brackets. (C, Right) Sterility and larval lethality of F2 progeny. (C, Left) Identity and the sex of the parents are indicated. Te total number of animals scored is indicated. Fisher’s exact test: NS, P > 0.05. WT, wild type. For A and C, blue text represents experiments performed using well-fed animal, while red text represents experiments performed using postdauer animals.

Fig. 4.

POE requires both the piRNA/nuclear RNAi pathway and factors required for H3K9me3 chromatin marks. (A) Schematic of RNAi experiments that test requirement for epigenetic regulators in POE. F3 L4 animals were treated with the indicated RNAis, and F4 L4s were used for the skn-1 RNAi assays. Animals in the control and treatment groups were siblings (Materials and Methods). (B) Knocking down prg-1, nrde-4, met-2, and set-32, but not ced-4, eliminates POE. Data points are replicates from at least two independent crosses. Boxplots represent medians, with range bars showing upper and lower quartiles. Arrowheads indicate phenotypes of JU1172 (40.0%) and MY16 (2.2%). Two-sample t test: not significant (NS), P > 0.05; ***P ≤ 0.001. (C) The effect of RNAi treatments on the skn-1(RNAi) phenotype. MY16 and JU1172 L4s were exposed to L4440 (control), prg-1, nrde-4, met-2, set-32, or ced-4 RNAi feeder strains, and the skn-1(RNAi) phenotype of the F1 were quantified. No difference between different RNAi treatments was detected (Fisher’s exact test, P > 0.05). The total number of embryos scored is indicated. (D) Model for POE (see Maintenance of POE Involves the Nuclear RNAi Pathway and Histone H3K9 Trimethylation).