(C) PLOS One
This story was originally published by PLOS One and is unaltered.
. . . . . . . . . .



A natural bacterial pathogen of C. elegans uses a small RNA to induce transgenerational inheritance of learned avoidance [1]

['Titas Sengupta', 'Lewis Sigler Institute For Integrative Genomics', 'Princeton University', 'Princeton', 'New Jersey', 'United States Of America', 'Department Of Molecular Biology', 'Jonathan St. Ange', 'Rachel Kaletsky', 'Rebecca S. Moore']

Date: 2024-04

C. elegans can learn to avoid pathogenic bacteria through several mechanisms, including bacterial small RNA-induced learned avoidance behavior, which can be inherited transgenerationally. Previously, we discovered that a small RNA from a clinical isolate of Pseudomonas aeruginosa, PA14, induces learned avoidance and transgenerational inheritance of that avoidance in C. elegans. Pseudomonas aeruginosa is an important human pathogen, and there are other Pseudomonads in C. elegans’ natural habitat, but it is unclear whether C. elegans ever encounters PA14-like bacteria in the wild. Thus, it is not known if small RNAs from bacteria found in C. elegans’ natural habitat can also regulate host behavior and produce heritable behavioral effects. Here we screened a set of wild habitat bacteria, and found that a pathogenic Pseudomonas vranovensis strain isolated from the C. elegans microbiota, GRb0427, regulates worm behavior: worms learn to avoid this pathogenic bacterium following exposure, and this learned avoidance is inherited for four generations. The learned response is entirely mediated by bacterially-produced small RNAs, which induce avoidance and transgenerational inheritance, providing further support that such mechanisms of learning and inheritance exist in the wild. We identified Pv1, a small RNA expressed in P. vranovensis, that has a 16-nucleotide match to an exon of the C. elegans gene maco-1. Pv1 is both necessary and sufficient to induce learned avoidance of Grb0427. However, Pv1 also results in avoidance of a beneficial microbiome strain, P. mendocina. Our findings suggest that bacterial small RNA-mediated regulation of host behavior and its transgenerational inheritance may be functional in C. elegans’ natural environment, and that this potentially maladaptive response may favor reversal of the transgenerational memory after a few generations. Our data also suggest that different bacterial small RNA-mediated regulation systems evolved independently, but define shared molecular features of bacterial small RNAs that produce transgenerationally-inherited effects.

C. elegans can learn to avoid a pathogenic clinical isolate of Pseudomonas aeruginosa, PA14, for four generations after training, through ingestion and RNA interference processing of a bacterial small RNA, P11, that targets a C. elegans neuronal gene, maco-1, through a 17-nucleotide perfect match. We screened bacteria associated with C. elegans in the wild, and found that lab C. elegans as well as wild C. elegans strains can also learn (and remember) to avoid P. vranovensis, a wild Pseudomonas pathogen. P. vranovensis uses a different small RNA that we identified and named Pv1, which targets a different exon of maco-1 (through a different 16-nucleotide match) and downregulates maco-1 expression transgenerationally, resulting in transgenerational inheritance of learned P. vranovensis avoidance. These data suggest that this mechanism of learning and remembering pathogen avoidance likely happens in the wild. Furthermore, the similarity in the “smell” of pathogenic and nutritious Pseudomonas (P. mendocina) may exert evolutionary pressure to forget the learned avoidance by the fifth generation, to prevent the worms from missing out on good food sources while avoiding pathogens.

Funding: This work was supported by a Pioneer Award to CTM (NIGMS DP1GM119167), a Transformative R01 Award (1R01AT011963-01) to ZG & CTM., CDCP 75D30122C15113 to CM, T32GM007388 (NIGMS) support of RSM., a Damon Runyon Fellowship (DRG-2481-22) to TS, a Ford Foundation Predoctoral Fellowship to RS ( https://www.nationalacademies.org/our-work/ford-foundation-fellowships ), and an NSF GRFP (DGE-2039656) predoctoral award to JS. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Copyright: © 2024 Sengupta et al. This is an open access article distributed under the terms of the Creative Commons Attribution License , which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Pseudomonas is one of the largest among the bacterial genera that constitute C. elegans’ natural microbiome [ 47 ]. In this study, we examined C. elegans’ behavioral responses to a Pseudomonad species present in its natural microbiome. We found that an isolate of Pseudomonas vranovensis can elicit learned avoidance and its transgenerational inheritance through a single small RNA that is both necessary and sufficient. However, this learned response to P. vranovensis also leads to avoidance of a beneficial bacteria also found in C. elegans’ environment, P. mendocina. Our work reveals a transgenerational effect in response to bacteria in C. elegans’ natural microbiome, underscoring the physiological relevance of transgenerational inheritance and its significance in the wild. We also identified a new small RNA that can induce a learned behavior in C. elegans, therefore expanding the repertoire of bacterial small RNA-mediated regulation of the host nervous system and helping to identify characteristics of small RNAs necessary for trans-kingdom signaling. Finally, the induced avoidance of a beneficial bacteria after pathogen training suggests that “forgetting” learned pathogen avoidance after a few generations might benefit C. elegans, limiting maladaptive behaviors.

Bacterial species in the worm microbiome that induce stress and immune response reporters are categorized as pathogenic [ 47 ], while species that promote increased worm growth rates are categorized as beneficial [ 47 , 49 , 53 ], and some of these species confer protection against pathogenic species [ 44 , 54 ]. Other species are beneficial in some contexts and pathogenic in others [ 55 ]. Therefore, it may be evolutionarily favorable for worms to have plastic responses to different bacterial classes that they naturally encounter. Worms are naively attracted to specific beneficial and neutral bacterial species (e.g., Pseudomonas mendocina and Proteus mirabilis, respectively) when given a choice between these bacteria and their laboratory diet E. coli HB101 [ 56 ]. Similarly, worms grown on the beneficial bacterial strain Providentia alcalifaciens prefer this bacterial species over their laboratory diet E. coli OP50 in a behavioral choice assay [ 57 ]. These beneficial bacterial species modulate C. elegans’ attraction towards several chemicals. Naïve or learned attraction in response to beneficial bacteria that worms encounter in their natural environment may have evolved as an evolutionarily favorable strategy. However, it is not known if worms can learn to avoid the various pathogenic bacterial species in their environment, or if they can inherit this learned avoidance. Additionally, whether bacteria in C. elegans’ natural environment can modulate the host nervous system through small RNAs and whether they can induce transgenerationally inherited effects are not known.

Diverse bacterial species influence C. elegans physiology and life history traits [ 40 – 42 ]. Bacterial species from C. elegans’ natural environment have been systematically characterized [ 43 – 49 ]. These studies revealed multiple features of the microbiota in C. elegans natural habitat and their relationship to host physiology [ 50 , 51 ]. Studying bacterial species that are naturally associated with C. elegans might reveal processes that occur in the wild, and not in the laboratory, and vice versa; for example, bacteria from C. elegans’ natural environment suppress mortal germline phenotypes that wild worms exhibit on laboratory strains of E. coli [ 52 ]. Therefore, it is important to test the physiological relevance of laboratory experimental results under more natural conditions.

Over the past decade, instances of multigenerational inheritance have been reported in various organisms [ 37 – 39 ]. We previously characterized an example of epigenetic inheritance in response to a physiological stimulus, highlighting its adaptive benefits in C. elegans: upon exposure to the pathogenic Pseudomonas aeruginosa strain PA14, worms learn to subsequently avoid the bacteria, then pass on this learned avoidance to four generations of progeny [ 27 ]. A single small RNA from PA14, P11, mediates this avoidance and its transgenerational inheritance through downregulation of the worm neuronal gene maco-1, which results in a switch from attraction to avoidance behavior [ 23 , 26 ]. These studies provided the first example of bacterial small RNA-mediated regulation of a learned behavior and its transgenerational inheritance [ 23 , 26 , 27 ]. However, PA14 is a human clinical Pseudomonas isolate; whether bacteria in C. elegans’ natural environment elicit learned responses and multi-generational inheritance of learned responses through small RNAs is not known.

Plants and animals have evolved diverse mechanisms to adapt to constantly changing environmental stimuli. Some of these stimuli are encoded as molecular changes that do not involve changes in DNA sequence, but are instead epigenetic, that is, mediated through changes in non-coding RNAs, DNA modifications, histone modifications, and nucleosome positioning [ 1 – 6 ]. These changes can occasionally cross the germline and confer adaptive benefits to the first generation of progeny (intergenerational) [ 7 – 20 ] or more (transgenerational) [ 21 – 33 ]. Multigenerationally-inherited effects can provide adaptive advantages in changing environments, particularly in organisms with short generation times [ 34 – 36 ].

Results

Wild microbiome bacteria induce learned avoidance To examine if exposure to bacteria isolated from C. elegans’ natural environment can produce stereotypic behavioral responses and further, whether these could potentially be small RNA-mediated, we tested C. elegans’ response to strains that are present in its natural microbiome [47]. We chose nine different bacterial species, mostly from the CeMBio collection [50] to test. These include non-pathogenic bacteria that are beneficial, as they enhance worm growth rates or provide protection against pathogen infection, or have positive or neutral effects depending on the physiological context (Pseudomonas mendocina (MSPm1), Raoultella sp. (Jub38), Leliottia sp. (Jub66), and Acinetobacter guillouiae (Myb10), Ochrobactrum vermis (Myb71), Enterobacter hormaechi (CEN2ent1), as well as three bacteria that are pathogenic or impair worm growth and development (Stenotrophomonas maltophilia (Jub19), Sphingobacterium multivorum (Bigb0170), and Pseudomonas vranovensis (GRb0427) [46,47,49,50]. Starting as late L4 animals, we exposed C. elegans for 24hrs either to OP50 E. coli (the standard lab cultivation strain) or to the test bacteria, and then assayed their preference to OP50 vs. the test strain (Fig 1A). In general, C. elegans prefer the wild strains—both beneficial and pathogenic—relative to the laboratory food E. coli OP50 (OP50-trained (gray bars), Fig 1A–1J). After 24hr exposure to wild bacteria (training), six of the bacteria showed no significant change in preference, despite the fact that two of those strains, Jub19 and Bigb0170, have detrimental effects on worms [47,50]; however, 24hr of exposure to three strains (Ochrobactrum vermis (Myb71), Enterobacter hormaechi (CEN2ent1)), and Pseudomonas vranovensis (GRb0427)) induced significant avoidance in the trained mothers (P0) (Fig 1F, 1G and 1J). That is, upon cultivation on a bacterial lawn for 24 hours, worms learn to robustly avoid the bacteria, as shown in a choice assay between OP50 and the test strain. (This P0 learned avoidance has been previously reported for Ochrobactrum vermis (Myb71) [58]). Longer exposure (36 hr) to the detrimental bacteria Sphingobacterium multivorum (Bigb0170) does not induce avoidance (S1 Fig). Thus, it seems that most wild strains are inherently attractive to C. elegans, whether they are beneficial, neutral or pathogenic, only a subset of strains induce avoidance in the P0, and this first-generation avoidance does not correlate with pathogenicity of the strain. PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 1. Wild microbiome bacteria induce learned avoidance. (A) Worms trained for 24 hours on E. coli OP50 or a test wild microbiome bacterial strain are tested in a choice assay between OP50 and the test bacterial strain. (B-J) Choice assays before and after training on Pseudomonas mendocina MSPm1 (B), Raoultella sp. Jub38 (C), Leliottia sp. Jub66 (D), Acinetobacter guillouiae Myb10 (E), Ochrobactrum vermis Myb71 (F), Enterobacter hormaechei CEN2ent1 (G), Stenotrophomonas maltophilia Jub19 (H), Sphingobacterium multivorum Bigb0170 (I), Pseudomonas vranovensis GRb0427 (J). (K) Worms trained for 24 hours on E. coli OP50 or a microbiome bacterial strain are bleached to obtain eggs, which are allowed to grow to Day 1 adults on OP50 plates. These adult F1 progeny are tested in a choice assay between OP50 and the respective bacterial strain. (L-N) Choice assays with F1 progeny of OP50 and Myb71-trained (L), OP50 and CEN2ent1-trained (M), and OP50 and GRb0427-trained (N) animals. Each dot represents an individual choice assay plate. Boxplots: center line, median; box range, 25th–75th percentiles; whiskers denote minimum-maximum values. Unpaired, two-tailed Student’s t test, ****p < 0.0001, ***p < 0.001, and *p<0.05, ns, not significant. Schematic representation in (A) and (K) were created using Biorender. https://doi.org/10.1371/journal.pgen.1011178.g001 To determine whether this P0 learned avoidance is inherited by the next generation, trained mothers (P0) were bleached and their progeny (F1) were raised on OP50 E. coli until adulthood, then were tested for their choice (with no F1 training) (Fig 1K). We observed that although neither Ochrobactrum vermis (Myb71) nor Enterobacter hormaechi (CEN2ent1)) progeny inherited the learned avoidance from their mothers (Fig 1L and 1M), progeny of Pseudomonas vranovensis (GRb0427)-trained mothers also avoided Pseudomonas vranovensis (GRb0427) (Fig 1N). Of the various bacteria we tested, only the pathogenic Pseudomonas vranovensis (GRb0427) induces learned avoidance and inheritance of avoidance.

C. elegans learn to avoid the natural bacterial pathogen, P. vranovensis Despite worms’ naïve attraction to Pseudomonas vranovensis, this bacterium is pathogenic to C. elegans: adult exposure to P. vranovensis (“GRb0427” hereafter) causes severe illness (Fig 2A) and significantly reduces survival to less than 2–3 days (Fig 2B), in contrast to C. elegans’ normal lifespan of 2–3 weeks. PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 2. GRb0427, a natural Pseudomonad pathogen of C. elegans induces learned avoidance. (A) Representative images acquired after exposing Day 1 worms to OP50 (left) or GRb0427 (right) for 24 hours. GRb0427 is pathogenic and 24-hour exposure makes worms sick. (B) Worms have significantly lower survival on a GRb0427 lawn compared to an OP50 lawn (p<0.0001 by Log-rank (Mantel-Cox)). (C) Representative image of a 24-hour OP50-trained or GRb0427-trained worm expressing daf-7p::gfp, which is expressed in the ASI sensory neurons (blue arrowheads). The dashed line indicates the outline of the worm head. Scale bar = 10 μm. (D) Quantification of mean ASI daf-7p::gfp intensities from OP50 and GRb0427 animals shows higher expression in GRb0427-trained animals. (E) Representative image of a 24-hour PA14-trained (top) or GRb0427-trained (bottom) worm. daf-7p::GFP is expressed in the ASI (blue arrowheads) and ASJ (orange arrowheads) sensory neurons (right panel) in the PA14-trained worm, but only in the ASI (blue arrowheads) in the GRb0427-trained worm. The dashed line indicates the outline of the worm head. Scale bar = 10 μm. (F) Quantification of mean ASI daf-7p::gfp intensities from OP50, PA14, and GRb0427 animals shows similar increase in daf-7p expression in PA14 and GRb0427-trained animals. (G) Expression of an irg-1p::gfp reporter in representative OP50-trained, PA14-trained and GRb0427-trained animals. Images are merged confocal micrographs of brightfield and GFP channels. Scale bar = 100 μm. (H) Mean fluorescence intensity of an irg-1p::gfp innate immune response reporter in OP50 (gray), Pseudomonas aeruginosa PA14 (blue), and GRb0427 (purple)-trained worms. Mean irg-1p::gfp reporter intensity is significantly lower in GRb0427-trained worms compared to PA14-trained worms. Each dot represents an individual neuron (D, F) or an individual worm (H). Boxplots: center line, median; box range, 25th–75th percentiles; whiskers denote minimum-maximum values. Unpaired, two-tailed Student’s t test, ****p < 0.0001 (D, F); one-way ANOVA with Tukey’s multiple comparison’s test, ****p<0.0001, **p<0.01, *p < 0.05, ns, not significant (H). For the survival assay in (B), ****p<0.0001 (by Log-rank (Mantel-Cox) test for survival). https://doi.org/10.1371/journal.pgen.1011178.g002 Exposure to P. aeruginosa PA14 causes gene expression changes in particular C. elegans sensory neurons; specifically, a 24-hour exposure to PA14 results in the induction of expression of the TGF-beta ligand DAF-7, as indicated by daf-7p::gfp, in the ASJ neurons and an increase in daf-7p::gfp expression in the ASI neurons [27,59]. PA14 small RNAs induce expression of daf-7p::gfp in the ASI that persists in the F1-F4 progeny generations [23, 27], while the increase in ASJ daf-7 is caused by PA14 secondary metabolites phenazine-1-carboxamide and pyochelin [59], and does not persist beyond the P0 [27]. To determine whether altered daf-7 levels correlate with the learned avoidance response to P. vranovensis, we examined daf-7p::gfp expression in GRb0427-trained animals (P0). Upon GRb0427 exposure, daf-7p::gfp levels significantly increase in the ASI neurons (Fig 2C and 2D), but no expression was observed in the ASJ neurons, in contrast to the response to PA14 training [27] (Fig 2E). Although daf-7p:GFP levels increase only in the ASI neurons upon GRb0427 exposure, this increase is comparable to that observed upon PA14 exposure (Fig 2F). Unlike other pathogenic bacteria, exposure to GRb0427 triggers a significantly milder innate immune response, as indicated by low expression of the innate immune response irg-1 (Infection Response Gene) promoter-GFP reporter, which is induced upon exposure to PA14 but not GRb0427 (Fig 2G and 2H). Lack of induction of phenazine-mediated ASJ daf-7p::gfp expression and only a mild induction of the irg-1 dependent innate immune pathway suggest that these innate immune pathways might not play a significant role in the neuronal response to the wild bacteria P. vranovensis GRb0427, even in the P0 generation, unlike the response to the clinical isolate PA14.

The avoidance response to P. vranovensis is transmitted for four generations Since training on P. vranovensis resulted in an increase in P0 daf-7p::gfp levels in the ASI (Fig 2C and 2D), and the adult F1 progeny of GRb0427-trained mothers showed robust avoidance of P. vranovensis compared to the F1 progeny of the control (OP50-trained) mothers (Fig 1) we examined the expression of daf-7p::gfp in progeny of GRb0427-trained mothers: these F1 animals express higher levels of daf-7p::gfp in the ASI neurons (Fig 3A and 3B) compared to that in F1 animals from OP50-trained mothers. PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 3. GRb0427-mediated learned avoidance is inherited transgenerationally. (A) F1 progeny of GRb0427-trained animals have higher daf-7p::gfp expression in the ASI sensory neurons (blue arrowheads). Scale bar = 10 μm. (B) Quantification of mean ASI daf-7p::gfp intensities from F1 progeny of OP50-trained and GRb0427 animals shows higher expression in F1 progeny of GRb0427-trained animals. (C, D) Quantification of mean ASI daf-7p::gfp intensities shows higher expression in F2 (C), F4 (D), and F5 (D) progeny of GRb0427-trained mothers, compared to the respective OP50 controls. (E) Untrained F1-F4 progeny of GRb0427-trained P0 animals avoid GRb0427 relative to the progeny of OP50-trained control P0 animals. This avoidance is lost in the F5 generation. (F) Learning index (trained choice index - naive choice index) of generations P0–F5. Error bars represent mean ± SEM. (G-I) N2 (wild-type) worms trained on a GRb0427 bacterial lawn learn to avoid GRb0427, but Cer1(gk870313) mutant animals don’t exhibit learned avoidance (G). The learned avoidance is inherited by the F1 (H) and F2 progeny (I) of GRb0427-trained N2 mothers but not by the progeny of GRb0427-trained Cer1(gk870313) mothers (n = 1). (J) Worms exposed to conditioned medium from the F1 progeny of OP50- and GRb0427-trained mothers were tested in a choice assay between OP50 and GRb0427. Worms exposed to conditioned medium from the F1 progeny of GRb0427-trained mothers exhibit avoidance of GRb0427. Each dot represents an individual choice assay plate (E, G-J) or an individual neuron for fluorescence images (B-D). Boxplots: center line, median; box range, 25th–75th percentiles; whiskers denote minimum-maximum values. Unpaired, two-tailed Student’s t test, ****p < 0.0001, ***p<0.001, *p<0.05, ns, not significant (B-D, J); one-way ANOVA with Tukey’s multiple comparison’s test, ****p<0.0001, **p<0.01, *p<0.05, ns, not significant (E, F), Two-way ANOVA with Tukey’s multiple comparison’s test, *p<0.05, ****p<0.0001, ns, not significant (G-I). https://doi.org/10.1371/journal.pgen.1011178.g003 To examine if learned avoidance to GRb0427 is inherited transgenerationally (beyond the F1 generation), we first asked whether daf-7p::gfp expression in the ASI remains high in the grandprogeny and subsequent progeny of GRb0427-trained mothers; like F1, the F2 and F4 animals had higher levels of ASI daf-7p::gfp, and these levels return to baseline (similar to the OP50 control) in the F5 generation (Figs 3C, 3D and S2). We then tested the next generations of progeny for avoidance. The learned avoidance of GRb0427 lasts up to the F4 generation, but returns to naïve attraction to GRb0427 in the F5 generation (Fig 3E). Thus, GRb0427 training induces transgenerational inheritance of learned avoidance behavior, as we previously found for PA14. Notably, in contrast to the higher avoidance of PA14 in the P0 generation than in F1-F4 [27], the level of avoidance of P. vranovensis is constant across P0 through F4 (Fig 3F). This result is consistent with the ASI (but not ASJ) expression of daf-7p::gfp and lack of expression of the innate immunity reporter irg-1p::gfp (Fig 2G and 2H) suggesting that innate immunity pathways may not contribute significantly to C. elegans’ avoidance of P. vranovensis, but rather that the major pathway of avoidance of P. vranovensis even in the first generation is through the same pathway as in F1-F4, rather than through classical innate immune pathways. This lack of an innate immune response to GRb0427 is particularly notable since P. vranovensis is found in C. elegans’ natural habitat [47], while PA14, the standard pathogen used for worm host-pathogen studies, is a human clinical isolate of P. aeruginosa.

P. vranovensis avoidance requires the Cer1 retrotransposon and can be horizontally transferred We had previously shown that the Cer1 retrotransposon is required for the learned avoidance of PA14 and its transgenerational inheritance [26] and is proposed to be involved in the transmission of information from the germline to neurons [26]. Similarly, we found that learned avoidance of P. vranovensis and the transgenerational inheritance of this avoidance require Cer1 (Fig 3G–3I). We had also found that training worms on conditioned media from the progeny of PA14-trained mothers can induce avoidance [26]; similarly, conditioned media from progeny of GRb0427-trained mothers can also induce learned avoidance (Fig 3J), indicating that the learned information can be horizontally transferred.

Wild C. elegans strains can learn to avoid P. vranovensis and transgenerationally transmit this information Wild C. elegans strains have been isolated all over the world [60,61], and can be helpful in distinguishing lab N2-strain-specific phenomena from those that are likely to function in the wild. We tested JU1580, a wild strain, for its responses to P. vranovensis; we find that JU1580, like N2, is attracted to GRb0427, but learns to avoid it after 24hr of training (Fig 4A). Like N2, JU1580 worms inherit this learned avoidance through the F4 generation, then return to naïve attraction in the F5 (Fig 4B). We tested an additional wild strain, ED3040 [49], and found that it behaved similarly to N2 and JU1580 in its initial attraction to P. vranovensis and its learned avoidance after 24hr of exposure (Fig 4C). Thus, it is likely that many wild C. elegans strains are attracted to P. vranovensis, learn to avoid it after exposure, and can transmit this learned avoidance transgenerationally, as we have shown for the laboratory C. elegans strain, N2. PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 4. Wild C. elegans strains can learn to avoid P. vranovensis and transgenerationally transmit this information. (A) A wild strain of C. elegans, JU1580 (a natural C. elegans isolate), avoids GRb0427 following 24-hour exposure to a GRb0427 bacterial lawn. (B) Untrained F1-F4 progeny of GRb0427 bacterial lawn-trained P0 JU1580 animals avoid GRb0427 relative to the progeny of OP50 bacterial lawn-trained control P0 JU1580 animals. This avoidance is lost in the F5 generation. (C) ED3040 (another natural isolate of C. elegans) also learns to avoid GRb0427 after a 24-hour exposure to a GRb0427 lawn. Each dot represents an individual choice assay plate. Boxplots: center line, median; box range, 25th–75th percentiles; whiskers denote minimum-maximum values. Unpaired, two-tailed Student’s t-test (A, C), ****p<0.0001; one-way ANOVA with Tukey’s multiple comparison’s test, ****p<0.0001, **p<0.01, ns, not significant (B). https://doi.org/10.1371/journal.pgen.1011178.g004

P. vranovensis small RNAs drive learned avoidance Since learned avoidance to P. vranovensis is transgenerationally inherited, and transgenerational inheritance of avoidance of PA14 is driven by its small RNA, P11, we next asked if small RNAs made by P. vranovensis induce avoidance. Like PA14, when adult C. elegans were exposed for 24 hours to sRNAs isolated from P. vranovensis GRb0427, worms learned to avoid P. vranovensis (Fig 5A). Exposure to P. vranovensis sRNAs increased daf-7p::gfp expression in the ASI sensory neurons (Fig 5B and 5C). We next tested if P. vranovensis sRNA-induced learned avoidance is transgenerationally inherited. Indeed, as observed for P. vranovensis lawn exposure, P. vranovensis sRNA-induced avoidance is inherited up to the F4 generation and resets in the F5 (Fig 5D and 5E). Training on GRb0427 small RNAs also induces learned avoidance and transgenerational inheritance in JU1580 worms (Fig 5F and 5G). PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 5. GRb0427 small RNAs induce learned avoidance and its transgenerational inheritance. (A) Worms trained on GRb0427 small RNAs exhibit learned avoidance of GRb0427 in an OP50-GRb0427 choice assay. (B) GRb0427 sRNA-trained animals have higher daf-7p::gfp expression in the ASI sensory neurons (blue arrowheads). Scale bar = 10 μm. (C) Quantification of mean ASI daf-7p::gfp intensities from OP50 sRNA-trained and GRb0427 sRNA-trained animals shows higher expression in GRb0427 sRNA-trained animals. (D) Untrained F1-F4 progeny of GRb0427 sRNA-trained P0 animals avoid GRb0427 relative to the progeny of OP50 sRNA-trained control P0 animals. This avoidance is lost in the F5 generation. (E) Learning index (index - naive choice index) of generations P0–F5. Error bars represent mean ± SEM. (F) GRb0427 sRNA-trained JU1580 animals avoid GRb0427 relative to OP50 sRNA-trained JU1580 animals. (G) F1-F4 progeny of GRb0427 sRNA-trained JU1580 animals avoid GRb0427 relative to the respective controls. This avoidance is lost in the F5 generation. Each dot represents an individual choice assay plate (A, D, F, G) or an individual neuron for fluorescence images (C). Boxplots: center line, median; box range, 25th–75th percentiles; whiskers denote minimum-maximum values. Unpaired, two-tailed Student’s t test (A, C, F), ***p<0.001, ****p<0.0001; one-way ANOVA with Tukey’s multiple comparison’s test, ****p< 0.0001, ***p<0.001, **p<0.01, *p<0.05, ns, not significant (D, E, G). https://doi.org/10.1371/journal.pgen.1011178.g005

P. vranovensis sRNA treatment also induces avoidance of PA14 We next asked if learned avoidance induced by P. vranovensis small RNAs is species-specific. We trained worms on P. vranovensis sRNAs and tested avoidance of PA14; P. vranovensis sRNA training induces avoidance to PA14, similar to training on PA14 sRNAs (Fig 6A). Previously, we found that a specific PA14 small RNA, P11, with 17nt of perfect match to the C. elegans neuronal homolog of the human nervous-system-specific ER membrane protein Macoilin, maco-1 [62], is required for learned avoidance [23]. Intriguingly, worms exposed to bacteria expressing only the PA14 sRNA P11 also induces avoidance to P. vranovensis (Fig 6B), suggesting that the underlying mechanism in P. vranovensis sRNA-induced avoidance is like that of PA14-induced avoidance. PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 6. GRb0427 sRNA-induced avoidance requires maco-1 and GRb0427 training heritably reduces maco-1 transcripts. (A) Worms were trained on OP50 (gray), Pseudomonas aeruginosa PA14 (blue), or GRb0427 (purple) small RNAs, and tested for their PA14 preference in a bacterial choice assay between OP50 and PA14. Worms treated on GRb0427 small RNAs not only avoid GRb0427 (Fig 3A), but also avoid PA14 in an OP50-PA14 choice assay. (B) Worms trained on the PA14 small RNA P11 avoid GRb0427 in an OP50-GRb0427 choice assay. (C) N2 (wild-type) worms trained on GRb0427 small RNAs avoid GRb0427, relative to Worms trained on OP50 small RNAs. maco-1(ok3165) loss-of-function mutant worms naively avoid GRb0427, relative to wild type worms, and do not show increased avoidance upon exposure to GRb0427 sRNAs. (D) Upon downregulation of maco-1 by RNAi, wild type worms exhibit higher naïve avoidance of GRb0427 compared to control RNAi-treated wild-type worms. (E) Fold change (2^(-ΔΔC t ) of maco-1 transcript levels in GRb0472-trained P0 (E), F2, F4, and F5 animals (F) relative to the respective OP50-trained controls (act-1 was used as the housekeeping gene for reference). Each data point represents an independent biological replicate, and 3 technical replicates were performed for each biological replicate. Each dot represents an individual choice assay plate (A-D) or a biological replicate in a qPCR assay (E,F). Boxplots: center line, median; box range, 25th–75th percentiles; whiskers denote minimum-maximum values, One-way ANOVA with Tukey’s multiple comparison’s test, ****p<0.0001, ns, not significant (A,B); Two-way ANOVA with Tukey’s multiple comparison’s test, **p<0.01, ****p<0.0001, ns, not significant (C); Unpaired, two-tailed Student’s t test, ***p<0.001, **p<0.01, *p<0.05 (D-F). https://doi.org/10.1371/journal.pgen.1011178.g006

P. vranovensis treatment decreases maco-1 transcripts through F4 Consistent with the idea of a conserved mechanism between PA14 and P. vranovensis-induced avoidance, loss of maco-1, either by mutation (maco-1(ok3165)) (Fig 6C) or by RNAi treatment (Fig 6D) significantly reduces the strong naïve preference for GRb0427. Similarly, exposure to P. vranovensis small RNAs does not further increase maco-1 mutants’ avoidance of P. vranovensis (Fig 6C), indicating that maco-1 loss phenocopies GRb0427 sRNA treatment. Quantitative RT-PCR showed a decrease in relative maco-1 transcript abundance in GRb0427-trained animals (Fig 6E), as well as in the F2 and F4 progeny of GRb0427-trained mothers (Fig 6F), but levels of maco-1 return to the same levels as untrained animals in the F5 generation (Fig 6F). That is, the decrease in maco-1 transcript levels after P0 treatment on P. vranovensis persists from F0 through F4 generation, mirroring the change in avoidance. This observation indicates that learned avoidance induced by P. vranovensis targets maco-1, as PA14’s P11 small RNA does. We also examined differentially-expressed genes between P. vranovensis-treated and E. coli HB101-treated P0 adult worms in RNA sequencing data reported in Burton et al., 2020 [8]; consistent with our results, maco-1 expression is downregulated in P. vranovensis-treated adult animals (log 2 fold change = -0.2837998, padj = 0.027) in this independent analysis [8]. (Our previous experiments suggest that we should only see changes in adult animals with fully-developed germlines [27]; the Burton dataset only provided adult data for the P0 generation [7,8]) We next examined whether P. vranovensis might encode a P11-like small RNA. We first analyzed the recently-sequenced genome of P. vranovensis [8], but we did not find any genomic region with sequence homology to P11. In fact, there is no region analogous to the operon that contains P11 in the P. vranovensis genome. That is, while P11 can induce similar avoidance of GRb0427 as it does for PA14, and GRb0427 induces avoidance through a small RNA, GRb0427 does not appear to encode a small RNA with a P11-like sequence in its genome; therefore, we needed to determine whether GRb0427 expresses a different sRNA that induces learned avoidance and inheritance of this avoidance. While a P11-like sRNA cannot account for the learned avoidance of GRb0427, our small RNA maco-1 experiments suggested that an sRNA with similarity to maco-1 might be involved. Therefore, we searched the P. vranovensis genome for similarity to the maco-1 sequence; we found five perfect matches to the maco-1 coding region in the P. vranovensis genome, but only one of these, a 16nt match, lies in an intergenic region that would be likely to encode a small RNA (Fig 7A–7C). Interestingly, this sequence identity lies in a different exon of maco-1 (Exon 1) from P11’s 17nt perfect match (Exon 8; Fig 7A). PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 7. An intergenic region in the GRb0427 genome contains a 16-nucleotide perfect match to maco-1 and is sufficient for learned avoidance of GRb0427. (A) The PA14 small RNA P11 contains a 17-nucleotide perfect match to the worm neuronal gene maco-1 (Exon 8). We found an intergenic region in the GRb0427 genome with a 16-nucleotide perfect match to a stretch of Exon 1 of maco-1. (B, C) The GRb0427 genome has an intergenic region, flanked by iron metabolism operon and sugar transport operons, containing a 16-nucleotide perfect match to maco-1 (B). This region is represented as a schematic in (C). 347 bp of this intergenic region (“IntReg”, shown in green) was cloned into E. coli for testing. IntReg contains the 16-nt match to maco-1 (indicated in purple). (D) Training worms on E. coli expressing the intergenic region (IntReg) with the match to maco-1 induces avoidance of GRb0427. (E) Untrained F1-F4 progeny of worms trained on E. coli expressing the intergenic region with the match to maco-1 exhibit higher avoidance of GRb0427 compared to controls. This higher avoidance is lost in the F5 generation. (F) Training of JU1580 worms on E. coli expressing the intergenic region (with the match to maco-1) induces avoidance of GRb0427. Each dot represents an individual choice assay plate (D-F). Box plots: center line, median; box range, 25th–75th percentiles; whiskers denote minimum-maximum values. Unpaired, two-tailed Student’s t test, **p<0.01, ***p<0.001 (D, F); one-way ANOVA with Tukey’s multiple comparisons test, **p<0.01, *p<0.05 ns, not significant (E). https://doi.org/10.1371/journal.pgen.1011178.g007 The P. vranovensis intergenic region containing this 16-nucleotide sequence match to maco-1 is flanked by bacterial protein-coding genes with predicted functions in the iron metabolism and sugar transport pathways (Fig 7B). To test whether the region containing this match might mediate P. vranovensis-induced avoidance, we expressed a 347 bp region of this intergenic sequence (“IntReg”; Fig 7C) in E. coli, and found that training on IntReg induces avoidance to P. vranovensis (Fig 7D). The avoidance induced by E. coli expressing the intergenic region persists for four generations after parental exposure and is lost by the F5 generation (Fig 7E), similar to the transgenerational inheritance of learned avoidance induced by a P. vranovensis lawn or small RNA exposure. Like P. vranovensis treatment, exposure of the wild strain JU1580 to the IntReg clone also induced avoidance of P. vranovensis (Fig 7F).

A specific P. vranovensis sRNA induces learned avoidance We next investigated if the intergenic region encodes a small RNA. While the genome of P. vranovensis is published, no information on small RNAs were publicly available, so we sequenced the small RNAs produced by P. vranovensis (see Methods), that is, the total small RNA pool that induces learned avoidance and its transgenerational inheritance. Indeed, we detected a small RNA that maps within the intergenic region that contains the 16-nucleotide sequence match to maco-1 (Fig 8A). This small RNA, which we named “Pv1”, is 124 bp long, and its homology to maco-1, like P11, lies in a predicted stem loop (Fig 8B). PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 8. Pv1, a GRb0427 sRNA, is necessary and sufficient for transgenerational inheritance of learned avoidance. (A) Alignment of GRb0427 small RNA sequencing reads to the GRb0427 genome. The peaks obtained from this alignment and the Sapphire small RNA promoter prediction software indicated the presence of three small RNAs (shown in dark gray, green, and light gray) in the intergenic region between the two predicted operons described in Fig 7. Of the 3 predicted sRNAs, the sRNA marked in green contains the 16-nucleotide match to maco-1 (purple), and we named it Pv1. (B) mFold structure prediction for Pv1; the maco-1 region is in a predicted stem loop of Pv1 (the boxed region). (C) Training worms on E. coli expressing just the Pv1 small RNA induces avoidance of GRb0427 (compared to training on E. coli expressing a control empty vector). (D) E. coli-Pv1 induced learned avoidance of GRb0427 is transgenerationally inherited. Untrained F1 and F2 progeny of E. coli-Pv1 trained P0 worms also exhibit higher GRb0427 avoidance compared to controls. (E-G) Naïve worms do not exhibit a preference towards either wild type GRb0427 or GRbΔ16nt strain in a GRb0427-GRbΔ16nt choice assay (E), while preferring both GRb0427 and GRbΔ16nt with respect to OP50 (F, G). (H) Survival of worms on lawns of GRb0427 and GRbΔ16nt are not significantly different. (I) F2 progeny of GRbΔ16nt-trained worms do not exhibit learned avoidance of GRb0427. (J) Mean ASI daf-7p::GFP intensities in GRbΔ16nt-trained worms are comparable to that of OP50-trained worms, in contrast to worms trained on GRb0427 where the mean ASI daf-7p::GFP intensities are higher than that of OP50-trained worms. (K) An E. coli strain expressing Pv1 containing 7 mismatches (4 of which lie within the 16- nucleotide match to maco-1) is predicted to have identical secondary structure to wild type Pv1, but the mismatches disrupt the sequence homology to maco-1. (L) Worms trained on E. coli-Pv1 learn to avoid GRb0427, while worms trained on E. coli expressing the Pv1 with mismatches fail to learn avoidance. Each dot represents an individual choice assay plate (C-G, I, L), or an individual neuron (J). Boxplots: center line, median; box range, 25th–75th percentiles; whiskers denote minimum-maximum values. Unpaired, two-tailed Student’s t test, **p<0.01, ***p<0.001, (C, D); One-way ANOVA with Tukey’s multiple comparison’s test, **p<0.01, ***p<0.001, ****p<0.0001, ns, not significant (I, J, L). For the survival assay in (H), ns–not significant (by Log-rank (Mantel-Cox) test for survival). https://doi.org/10.1371/journal.pgen.1011178.g008 We next asked whether exposure to Pv1 expressed in E. coli would be sufficient to induce avoidance to P. vranovensis; indeed, training on E. coli-Pv1 induces avoidance in the mother generation (P0; Fig 8C), as well as the inter (F1)- and transgenerational (F2) inheritance of this learned avoidance (Fig 8D). To determine if Pv1’s sequence identity to maco-1 is necessary for the learned avoidance to P. vranovensis and its transgenerational inheritance, we constructed a mutant P. vranovensis bacterial strain lacking the 16-nucleotide match to maco-1, Δ16. This Δ16 mutant GRb0427 strain is equally attractive to worms as wild-type GRb0427, and C. elegans prefer Δ16 to OP50 just as they prefer GRb0427 to OP50 (Fig 8E–8G), suggesting that the bacteria do not “smell” different to the worms. Additionally, Δ16 is similarly pathogenic to worms as wild-type GRb0427 (Fig 8H). However, Δ16 is unable to induce transgenerational inheritance of learned avoidance (Fig 8I), nor can Δ16 induce daf-7p::gfp expression in ASI neurons (Fig 8J). Finally, mismatch of four of the nucleotides in the loop of Pv1 within the 16nt maco-1 match (but that still retains its predicted stem-loop structure, Fig 8K) removes the ability of Pv1 to induce avoidance (Fig 8L). Together, our data suggest that a specific P. vranovensis small RNA, Pv1, is sufficient to induce avoidance, and Pv1’s 16nt match to the maco-1 sequence is necessary for sRNA-mediated learned avoidance of GRb0427 and its transgenerational inheritance.

Mechanism: RNAi components are required for Pv1-induced avoidance Previously, we showed that components of the RNA interference pathway are required for PA14- and P11-induced learned avoidance and its transgenerational inheritance [23,27]. These components included the SID-2 dsRNA transporter, the DCR-1 (Dicer) endoribonuclease, and the SID-1 RNA transmembrane dsRNA transporter. To determine whether the P. vranovensis sRNA Pv1 is processed through a similar mechanism, we tested mutants of these components. We found that sid-2(qt42) [63,64], dcr-1(mg375) [65,66], or sid-1(qt9) [67] animals were unable to learn P. vranovensis avoidance after Pv1 training (Fig 9A–9C). Moreover, mutants of hrde-1(tm1200), the nuclear Argonaute that binds 22G RNAs, cannot learn to avoid P. vranovensis after training on Pv1 (Fig 9D), indicating that HRDE-1 is also required. Together, our data suggest that the mechanism that we had previously identified for PA14-P11 sRNA processing is shared with P. vranovensis-Pv1 processing, and involves bacterial sRNA uptake (sid-2), dsRNA processing (dcr-1), 22G sRNA binding (hrde-1), dsRNA transport (sid-1), maco-1 transcript reduction, Cer1-mediated germline-to-neuron communication, daf-7 expression increase in the ASI, and then behavioral switching from attraction to avoidance (Fig 9E). PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 9. Genes involved in sRNA processing are required for Pv1 sRNA-mediated learned avoidance. (A-D) sid-2(qt42) (A), dcr-1(mg375) (B), sid-1(qt9) (C), and hrde-1(tm1200) (D) do not learn to avoid GRb0427 in response to Pv1 training. (E) Model of Pv1 uptake, processing, and induced avoidance of P. vranovensis (created using Biorender). Boxplots: center line, median; box range, 25th–75th percentiles; whiskers denote minimum-maximum values. Two-way ANOVA with Tukey’s multiple comparison’s test, **p<0.01, ***p<0.001, ns, not significant (A-D). https://doi.org/10.1371/journal.pgen.1011178.g009

[END]
---
[1] Url: https://journals.plos.org/plosgenetics/article?id=10.1371/journal.pgen.1011178

Published and (C) by PLOS One
Content appears here under this condition or license: Creative Commons - Attribution BY 4.0.

via Magical.Fish Gopher News Feeds:
gopher://magical.fish/1/feeds/news/plosone/

You are viewing proxied material from magical.fish. The copyright of proxied material belongs to its original authors. Any comments or complaints in relation to proxied material should be directed to the original authors of the content concerned. Please see the disclaimer for more details.