Molecular requirements for C. elegans transgenerational epigenetic inheritance of pathogen avoidance

Abstract

Bacteria are Caenorhabditis elegans' food, and worms are naturally attracted to many bacteria, including pathogenic Pseudomonas, preferring PA14 over laboratory Escherichia coli (OP50). Despite this natural attraction to PA14, prior PA14 exposure causes the worms to instead avoid PA14. This behavioral switch can happen quickly - even within the duration of the choice assay. We show that accurate assessment of the animals' true first choice requires the use of a paralytic (azide) to trap the worms at their initial choice, preventing the switch from attraction to avoidance of PA14 within the assay period. We previously discovered that exposure of C. elegans to 25°C plate-grown PA14 at 20°C for 24 hr not only leads to PA14 avoidance, but also to four generations of naïve progeny avoiding PA14, while other PA14 paradigms only cause P0 and/or F1 avoidance. We also showed that the transgenerational (P0-F4) epigenetic avoidance is mediated by P11, a small RNA produced by PA14. P11 is both necessary and sufficient for TEI of learned avoidance. P11 is highly expressed in our standard growth conditions (25°C on surfaces), but not in other conditions, suggesting that the reported failure to observe F2-F4 avoidance is likely due to the absence of P11 expression in PA14 in the experimenters' growth conditions. Additionally, we tested ~35 genes for involvement in TEI of learned pathogen avoidance. The conservation of multiple components of this sRNA TEI mechanism across C. elegans strains and in multiple Pseudomonas species suggests that this TEI behavior is likely to be physiologically important in wild conditions.

Keywords: C. elegans; RNA interference; behavior; genetics; genomics; inheritance; sid-1; small RNA; transgenerational.

Figure 1.. The omission of the paralytic sodium azide from choice assays has significant effects on naïve choice assays.

(A) Schematic of PA14 and worm growth conditions, worm training, and choice assay parameters. Critical factors affecting assay performance are highlighted. (B) Scatter dot plots of representative replicate choice assays (OP50 vs PA14 bacteria) from naïve day 1 adult worms (from Moore et al., 2019; Kaletsky et al., 2020; Moore et al., 2021). Each dot represents an individual choice assay plate (n=10) containing ~50–100 worms per plate (average ~80). Choice index = (number of worms on OP50 − number of worms on PA14)/(total number of worms). (C) Individual replicates of naïve choice assays from Moore et al., 2019; Kaletsky et al., 2020; Moore et al., 2021, are shown as violin plots. Each replicate is a separate experiment made up of 5–10 plates. The median is shown in red. (D) Individual replicates from (C) were pooled. Box plots: center line, median; box range, 25–75th percentiles; whiskers denote minimum–maximum values. (E) Naïve worms were placed on choice assay plates with spots of PA14 and OP50 with or without sodium azide on the bacterial spots, then allowed to choose for 1 hr, followed by 1 hr at 4°C. No azide: CI = 0.4 ± 0.06, p<0.0001. (F) Estimation plot of difference in CI between azide and no azide plates; mean difference = 0.482 [95.0%CI 0.369, 0.64]. (G) After 1 hr of the choice assay at room temperature, choice assay plates were placed at 4°C. Worms were counted at each spot at the indicated time points. Mean ± SD from three choice assay plates. Time point 0=immediately before 4°C incubation. (H) Worms were placed on choice assay plates with spots of Pseudomonas fluorescens 15 (PF15) and OP50 with or without sodium azide on the bacterial spots, then allowed to choose for 1 hr, followed by 1 hr at 4°C. No azide CI = 0.26 ± 0.05, p<0.0001. (i) Estimation plot of difference in CI between azide and no azide plates; mean difference = 0.547 [95.0%CI 0.348, 0.737].

Figure 2.. P0 and F2 worms trained on PA14 bacteria reproducibly learn to avoid PA14 in choice assays.

(A) P0 replicates: A representative experiment (replicate) showing the choice index and effect size of 24 hr PA14-trained worms compared to the OP50-trained control. (B) Mean differences of individual replicates from Moore et al., 2019; Kaletsky et al., 2020; Moore et al., 2021, are shown for PA14- vs OP50-trained mothers. (C) Individual replicates from (B) were pooled. (D–E) F2 animals from P0-control or PA14-trained worms. (D) F2 replicates: A representative experiment (replicate) showing the choice index and effect size of F2 worms. (A, D) The mean difference is shown as a Gardner–Altman estimation plot. Both groups are plotted on the left axes; the mean difference is plotted on floating axes on the right as a bootstrap sampling distribution. The mean difference is depicted as a dot; the 95% confidence interval is indicated by the ends of the vertical error bar. (E) Mean differences of individual replicates from Moore et al., 2019; Kaletsky et al., 2020; Moore et al., 2021, are shown for F2s from PA14- vs OP50-trained grandmothers. (B, E) Box plots: center line, median; box range, 25–75th percentiles; whiskers denote minimum–maximum values. Unpaired, two-tailed Student’s t-test. ****p<0.0001. (F) Individual replicates from (E) were pooled. (C, F) Each replicate is a separate experiment made up of 5–10 plates (average ~80 worms per plate). The mean difference (effect size) is shown as a Cumming estimation plot. Each mean difference is plotted as a bootstrap sampling distribution. Mean differences are depicted as dots; 95% confidence intervals are indicated by the ends of the vertical error bars. Estimation graphics generated as described in Ho et al., 2019.

Figure 3.. P11 sRNA and components of germline and RNAi function are required for transgenerational inheritance.

(A) P0 worms trained on E. coli expressing the P11 small RNA reproducibly avoid PA14 in choice assays. Individual replicates from Kaletsky et al., 2020, Moore et al., 2021, are shown for P0 worms obtained from P11 vs control training. (B) Individual replicates from (A) were pooled. (C) F2 animals from P0 worms trained on E. coli expressing the P11 small RNA reproducibly avoid PA14 in choice assays. Individual replicates from Kaletsky et al., 2020, Moore et al., 2021, are shown for F2 worms obtained from P11- vs control-trained grandmothers. (D) Individual replicates from (A) were pooled. (A, C) Each replicate is a separate experiment made up of 5–10 plates (average ~80 worms per plate). The mean difference (effect size) is shown as a Cumming estimation plot. Each mean difference is plotted as a bootstrap sampling distribution. Mean differences are depicted as dots; 95% confidence intervals are indicated by the ends of the vertical error bars. (E) Normalized RNA counts for P11 expression from PA14 under different bacterial growth conditions. Sequencing data and figure adapted from Kaletsky et al., 2020. Adjusted p-values from DESeq2. (F) P0 meg-1;meg-3 mutants exhibit normal learned PA14 avoidance when trained on PA14 lawns, but (G) P0 meg-1;meg-3 mutants do not exhibit learned PA14 avoidance when trained with PA14 sRNA. (H) P0 nrde-1(gg66) mutants exhibit normal learned PA14 avoidance when trained on PA14 lawns, (I) but are defective for PA14 avoidance learning when exposed to PA14 sRNA. (J) F2 nrde-3 mutants from P0-PA14 lawn-trained grandmothers do not avoid PA14 compared to wild-type controls. (K) Wild-type P0 worms exposed to E. coli-expressing P11 avoid PA14, but rde-3 mutants do not. (B, D–K) Box plots: center line, median; box range, 25–75th percentiles; whiskers denote minimum–maximum values. (B, D) Unpaired, two-tailed Student’s t-test. (F–K) Two-way ANOVA with Tukey’s multiple comparison’s test. *p≤0.05, ***p≤0.001, ****p<0.0001, NS, not significant. Estimation graphics generated as described in Ho et al., 2019.

Figure 4.. Summary of genes tested for small RNA-mediated pathogen avoidance and characterization of PA14 avoidance-based assays.

(A) Summary of genes tested for naïve Pseudomonas preference behavior and small RNA-based learning and inheritance. (B) Table describing different assays measuring Pseudomonas learning and avoidance. Pseudomonas avoidance learning in the parental generation (P0), intergenerationally (F1), and transgenerationally (F2 through F4) is noted. References: 1, Meisel et al., 2014; 2, Singh and Aballay, 2019; 3, Zhang et al., 2005; 4, Pereira et al., 2020; 5, Moore et al., 2019; 6, Kaletsky et al., 2020; 7, Moore et al., 2021; 8, Sengupta et al., 2024; 9, Seto et al., 2025.