C. elegans transgenerational avoidance of P. fluorescens is mediated by the Pfs1 sRNA and vab-1

Abstract

In its natural habitat, Caenorhabditis elegans must distinguish friend from foe. Pseudomonas are abundant in the worm's environment and can be nutritious or pathogenic. Previously, we found that worms learn to avoid Pseudomonas aeruginosa and Pseudomonas vranovensis through a small RNA (sRNA)-mediated pathway targeting the C. elegans gene maco-1, and this behavior is inherited for four generations. Here, we show that C. elegans learns to transgenerationally avoid another pathogenic bacteria Pseudomonas fluorescens 15 (PF15). The PF15 sRNA, Pfs1, targets the VAB-1 ephrin receptor through 16 nt of perfect match, suggesting the evolution of a distinct bacterial sRNA/C. elegans gene target pair. Knockdown of both maco-1 and vab-1 induce PF15 avoidance, and vab-1 loss reduces maco-1 expression, placing both genes in the sRNA-targeted pathogenic avoidance pathway. Thus, multiple genes in this avoidance pathway can act as targets for bacterial sRNAs, expanding the possibilities for evolution of trans-kingdom regulation of C. elegans behavior.

Fig. 1.. PF15 is pathogenic and induces learned avoidance in C. elegans.

(A) PF15 is fluorescent in ultraviolet (UV) light. Choice plates with spots of OP50 and PF15 bacteria. Left: Image in white light. Right: Image in UV light. (B) Representative images acquired after exposing day 1 adults to OP50 (left) or PF15 (right) for 24 hours. PF15 is pathogenic and 24-hour exposure makes worms sick. Scale bars, 100 μm. (C) C. elegans on PF15 lawn have reduced survival compared to survival on an OP50 lawn (25°C) (median survival = 40 hours, P < 0.0001). One replicate shown, representative of two biological replicates. (D) Representative images of irg-1p::gfp expression in day 1 adults exposed to OP50 (left) or PF15 (right) for 24 hours starting at L4. Scale bars, 100 μm. (E) Quantification of irg-1p::gfp in P0 PF15-trained worms. Expression of irg-1p::gfp is induced by PF15 bacterial lawn exposure. Data pooled from three biological replicates. (F) Adult C. elegans trained for 24 hours on PF15 avoid PF15 compared to OP50-trained control. Choice index = (# of worms on OP50 − # of worms on PF15)/(total # worms). (G) pmk-1(km25) mutants can learn PF15 avoidance from PF15 lawn training. Each dot represents an individual worm (E) or an individual choice assay plate [(F) and (G)]. Boxplots: Center line, median; box range, 25th to 75th percentiles; whiskers denote minimum-maximum values. Biological replicates: two (C), three (E), seven (F), and four (G). For the survival assay in (C), P < 0.0001 by log-rank (Mantel-Cox) test for survival. Unpaired, two-tailed Student’s t test [(E) and (F)]; two-way ANOVA with Tukey’s multiple comparison’s test (G). ****P < 0.0001.

Fig. 2.. PF15 exposure induces TEI of PF15 avoidance.

(A) TEI of pathogen avoidance: PF15-trained P0 worms learn to avoid PF15 compared to OP50 control, and their F1-F4 progeny retain PF15 avoidance behavior before behavior resets in F5. (B) Learning index [average trained choice index − average naïve (OP50-trained) choice index for each biological replicate] of generation P0 through F5. Significance comparison to F5. Error bars represent means ± SEM. (C) Representative images of 24 hours OP50-trained or PF15-trained worm expressing daf-7p::gfp in P0 (left) or F1 (right). P0 worms (left) trained on PF15 (bottom) show increased daf-7p::gfp in the ASI (blue arrowheads) and ASJ sensory neurons (orange arrowhead) compared to OP50-trained controls (top). F1 progeny (right) of PF15-trained P0 (bottom) also show increased daf-7p::gfp only in the ASI compared to F1 progeny of OP50-trained P0 (top). Scale bars, 10 μm. Quantification of daf-7p::gfp in P0 24-hour OP50- and PF15-trained worms and their F1 progeny. PF15 training increases daf-7p::gfp expression in the ASI neuron pair in both P0 and F1. Quantification representative of three (P0) or two (F1) independent biological replicates. (D) ASI-ablated worms (PY7505) show high naïve PF15 avoidance. (E) Training naïve worms with conditioned media collected from F2 progeny of OP50- or PF15-trained grandmothers (left) show those trained with conditioned media from PF15-trained F2 learn to avoid PF15 through horizontal transfer (right). This figure was created using BioRender. Each dot represents an individual neuron (C) or an individual choice assay plate [(A), (D), and (E)]. Boxplots: Center line, median; box range, 25th to 75th percentiles; whiskers denote minimum-maximum values. Biological replicates: four to seven [(A) and (B)], three [P0; (C)], and two [F1; (C) to (E)]. One-way ANOVA with Tukey’s multiple comparison’s test [(A) and (B)]; unpaired, two-tailed Student’s t test [(C) to (E)]. ****P < 0.0001, ***P < 0.001, **P < 0.01, and *P > 0.05. ns, not significant; a.u., arbitrary unit.

Fig. 3.. PF15 sRNAs are sufficient to cause transgenerational PF15 avoidance and Cer1 virus-like particles are required for sRNA-mediated learning.

(A) Representative images of day 1 worms on OP50 sRNAs (left) or PF15 sRNAs (right) added to OP50 lawns for 24 hours. Worms treated with sRNAs isolated from PF15 do not appear sick, unlike lawn-trained worms (see Fig. 1B). Scale bars, 50 μm. (B) Training with PF15 sRNAs for 24 hours causes PF15 avoidance compared to OP50 sRNA-trained control. (C) Comparative magnitude (learning index) of PF15 lawn versus PF15 sRNA-only P0 avoidance suggests that P0 avoidance is the cumulative effect of sRNA-dependent and sRNA-independent avoidance pathways. (D) Cer1(gk870313) mutants are unable to avoid PF15 with PF15 sRNA training. (E) Transgenerational inheritance of pathogen avoidance: P0 worms trained on PF15 sRNAs for 24 hours learn to avoid PF15 compared to OP50 control, and their F1-F4 progeny retain PF15 avoidance behavior until it returns to baseline in F5. (F) Learning index of PF15 sRNA-trained worms generations P0–F5. Magnitude of learning is consistent from P0 to F4. Significance comparisons to F5. Error bars represent means ± SEM. Each dot represents an individual choice assay plate [(B), (D), and (E)] or biological replicate (C). Boxplots: Center line, median; box range, 25th to 75th percentiles; whiskers denote minimum-maximum values. Biological replicates: four (B), four to seven (C), three (D), two to four [(E) and (F)]. Unpaired, two-tailed Student’s t test [(B) and (C)]; two-way ANOVA with Tukey’s multiple comparison’s test (D); one-way ANOVA with Tukey’s multiple comparison’s test [(E) and (F)]. ****P < 0.0001, ***P > 0.001, **P > 0.01, and *P > 0.05.

Fig. 4.. Sequencing the PF15 genome and PF15 sRNAs identifies the C. elegans gene vab-1 as a regulator of PF15 avoidance.

(A) Representation of the PF15 genome. Outer: Segments represent individual contigs; middle: GC content; inner: genome length. Generated by Proksee. (B) sRNAs were collected from PF15 in three growth conditions—25°C plate, 37°C plate, and 37°C liquid—and used for training alongside OP50 sRNA control. PF15 sRNAs from all three conditions cause PF15 avoidance. (C) Differential mRNA expression of P0 C. elegans trained on OP50 versus PF15 lawn for 24 hours. (D) PF15 genome contains a 16-nt perfect match to the C. elegans vab-1 gene in exon 6. This figure was created using BioRender. (E) Worms exposed to adult-only vab-1 RNAi naively avoid PF15 compared to control. See fig. S2A. (F) Differential expression analysis of PF15-P0 versus OP50-P0 C. elegans adult mRNA seq (C) reveals vab-1 expression is decreased in PF15-trained P0. Log2FC = −0.4363014, adjusted P value = 0.0189176. (G) vab-1(e2) mutant exhibits morphological head defects (white arrow) compared to N2. (H) N2 worm treated with adult-only vab-1 RNAi (bottom) shows no morphological defects compared to RNAi control (top). (I) vab-1(e2) mutants have high naïve avoidance of PF15 compared to N2. (J) vab-1(e2) mutants do not learn further avoidance with PF15 training. Each dot represents an individual choice assay plate [(B), (E), and (I) to (J)], gene (C), or sequencing replicate (F). Scale bars, 10 μm [(G) and (H)]. Boxplots: Center line, median; box range, 25th to 75th percentiles; whiskers denote minimum-maximum values. Biological replicates: four (B), three [(E) and (F)], two [(I) and (J)]. One-way ANOVA with Tukey’s multiple comparison’s test (B); unpaired, two-tailed Student’s t test [(E) and (I)]; two-way ANOVA with Tukey’s multiple comparison’s test (J). ****P < 0.0001 and **P < 0.01

Fig. 5.. Training with Pfs1 sRNA causes transgenerational avoidance of PF15.

(A) PF15 intergenic region (IntReg, includes 16-nt perfect vab-1 match) training causes transgenerational PF15 avoidance P0-F4, resetting in F5. (B) sRNA sequencing reveals a sRNA within IntReg (A, S3F), Pfs1, containing the 16-nt perfect vab-1 match. RACE defines a 207-nt sRNA, Pfs1. (C) mFold secondary structure prediction shows the 16-nt match to vab-1 (green) in a stem-loop structure. (D) Training with Pfs1 causes PF15 avoidance compared to control. (E) Pfs1-trained worms transgenerationally avoid PF15 from P0-F4, resuming naïve behavior in F5. (F) sid-1(qt9) mutants fail to avoid PF15 after Pfs1 training (controls shared with Fig. 6F). (G and H) Representative images and quantification show increased daf-7p::gfp fluorescence in ASI sensory neurons (blue arrowheads, left) in Pfs1 versus control-trained P0 worms (G) or in F2 grandprogeny of Pfs1 versus control-trained worms (H). Representative of three (G) or two (H) independent biological replicates. Scale bars, 100 μm. (I) ASI-ablated worms (PY7505) and (J) worms treated with adult-only vab-1 RNAi display more short reversals in minutes 1 to 5 of starvation compared to N2 (I) or control (J). (K) Pfs1-trained P0 worms display more short reversals in minutes 1 to 5 of starvation compared to those trained on control bacteria, as do their (L) F1 progeny and (M) F2 progeny. Each dot represents an individual choice assay plate [(A) and (D) to (F)], neuron [(G) and (H)], or worm [(I) to (M)]. Boxplots: Center line, median; box range, 25th to 75th percentiles; whiskers denote minimum-maximum. Biological replicates: three to four (A), 3 [(D) to (G) and (K) to (M)] and two [(H) to (J)]. One-way ANOVA with Tukey’s multiple comparison’s test [(A) and (E)]; unpaired, two-tailed Student’s t test [(D) and (G) to (M)]; two-way ANOVA with Tukey’s multiple comparison’s test (F). ****P < 0.0001, ***P < 0.001, **P < 0.01, and *P < 0.05.

Fig. 6.. Pfs1 vab-1 homology is sufficient and necessary for transgenerational PF15 avoidance.

(A) Control-trained vab-1(e2) mutants have high naïve (untrained) avoidance to PF15 compared to N2 and do not further avoid PF15 with Pfs1 training. (B) A PF15 mutant was generated (see Materials and Methods) with five mismatches in the Pfs1 sequence (red) in the 16 nt of perfect homology between Pfs1(bottom) and vab-1 exon 6 (top). (C) Secondary structure of PF15 Pfs1 vab-1 mismatch mutant, with mutations shown in red. Two additional mutations were made outside of the vab-1 homology region to preserve secondary structure of the mutated Pfs1 sRNA. We called this mutant PF15 “Pfs1 vab-1 mismatch”. (D) P0 N2 worms trained on both wild-type PF15 and PF15 Pfs1vab-1 mismatch strains learn to avoid wild-type PF15 with the same magnitude after training. (E) F1 progeny of wild-type PF15-trained mothers have inherited PF15 avoidance; however, F1 progeny of the vab-1 mismatch PF15-trained mothers do not avoid PF15; PF15 vab-1 mismatch training results in only P0 avoidance and no inheritance of learned avoidance. (F) Cer1(gk870313) mutants are unable to learn PF15 avoidance from Pfs1 training in P0. (G) vab-1 sequence containing the perfect match to PF15 is preserved in several C. elegans wild isolate strains. Each dot represents an individual choice assay plate [(A) and (D) to (F)]. Boxplots: Center line, median; box range, 25th to 75th percentiles; whiskers denote minimum-maximum values. Biological replicates: two (A), four [(D) and (E)], and three (F). Two-way ANOVA with Tukey’s multiple comparison’s test [(A) and (F)]; one-way ANOVA with Tukey’s multiple comparison’s test [(D) and (E)]. ****P < 0.0001, ***P < 0.001, and **P < 0.01.

Fig. 7.. vab-1 acts upstream of maco-1 to induce PF15 avoidance.

(A) maco-1(ok3165) loss-of-function mutant worms naively avoid PF15 relative to wild-type worms. (B) Worms exposed to adult-only maco-1 RNAi (L4 to day 1 adult) naively avoid PF15 compared to control. (C) maco-1(ok3165) loss-of-function mutant worms do not show increased PF15 avoidance learning with Pfs1 training. (D) maco-1 expression is decreased in PF15-trained P0 worms, as shown by differential mRNA expression (Fig. 4C). Adjusted P value = 0.0001186023. (E) qPCR of maco-1 levels show reduced maco-1 expression in F2 grandprogeny of Pfs1-trained worms compared to F2 control-trained worms. Fold change [2^(−ΔΔCt)] of maco-1 transcript levels in F2 progeny of control or Pfs1-trained worms relative to the respective OP50-trained control (act-1 was used as the housekeeping gene for reference). Each data point represents an independent biological replicate, and three technical replicates were performed for each biological replicate (see Materials and Methods). (G) Worms trained on Pfs1 sRNA-expressing bacteria avoid P. aeruginosa (PA14) compared to control-trained worms. (H) Worms trained on Pfs1 sRNA-expressing bacteria avoid P. vranovensis (GRb0427) compared to control-trained worms. (I) Worms trained on Pfs1 sRNA-expressing bacteria avoid P. mendocina (MSPm1) compared to control-trained worms. Each dot represents an individual choice assay plate [(A) to (C) and (G) to (I)] or a biological replicate [(D) and (E)]. Boxplots: Center line, median; box range, 25th to 75th percentiles; whiskers denote minimum-maximum values. Biological replicates: three [(A to (D) and (G) to (I)] and five (E). Unpaired, two-tailed Student’s t test [(A) and (B) and (G) to (I)]; two-way ANOVA with Tukey’s multiple comparison’s test (C). ****P < 0.0001, ***P < 0.001, **P < 0.01, and *P > 0.05.

Fig. 8.. Model of vab-1/maco-1/daf-7 axis of pathogen avoidance.

(A) Worms exposed to adult-only vab-2 RNAi (L4- day 1 adult) naively avoid PF15 compared to control. (B) Worms exposed to adult-only efn-3, efn-4, nck-1, and wsp-1 RNAi (L4- day 1 adult) naively avoid PF15 compared to control. (C) Model of vab-1 epistasis and mechanism of Pfs1-mediated PF15 avoidance in comparison to previous findings of P11 and Pv1 mediated avoidance (8, 10). Small RNAs P11 (PA14), Pv1 (P. vranovensis), and Pfs1 (PF15) cause shared avoidance by converging on maco-1 down-regulation and daf-7 up-regulation. This figure was created using BioRender. Each dot represents an individual choice assay plate [(A) and (B)]. Biological replicates: 3 [(A) and (B)]. Unpaired, two-tailed Student’s t test (A); one-way ANOVA with Tukey’s multiple comparison’s test (B). ****P < 0.0001, ***P < 0.001, and *P< 0.05.