Fig 1.
Most naturally associated bacteria reduce C. elegans susceptibility to viral infection compared to E. coli OP50.
For each bacterial environment, three replicates of ca. 100 ERT54 animals were challenged with the Orsay virus JUv1580. The proportion of animals activating the pals-5p::GFP reporter was measured at 72 hpi (raw data are in S2 File). Data are plotted here as mean ± standard deviation between the three replicates, relative to the mean proportion on Escherichia OP50 measured on the same day. Significance was calculated using a general linear model with bacteria as a factor and Dunnett’s contrasts to compare all conditions against the Escherichia OP50 reference (bar highlighted in yellow). Bacterial strains that induce a significant (P < 0.05) reduction of pals-5p::GFP reporter activation are colored in red (darker red for those having a lower than 0.5 relative activation of pals-5), those with no significant differences with OP50 in grey, and those with significantly higher activation in blue. Bacteria are arranged according to their phylogenetic relationships, with taxonomic classifications provided in the top rows and a phylogenetic tree based on their 16S sequences in the bottom row. Bacterial strain names in blue indicate a significant phylogenetic signal for these strains. Bacterial strains marked with a grey dot were selected for further investigation.
Fig 2.
Focus on five bacterial strains confirms their modulation of C. elegans susceptibility to viral infection.
(A-B) Proportion of ERT54 animals showing pals-5p::GFP activation (left panel) or infected intestinal cells (stained by FISH against the virus; right panel), after OrV inoculation on bacteria that induced (A) strong activation or (B) no activation at 72 hpi of pals-5p::GFP in the initial screen, with Escherichia OP50 as a reference. Each bacterial strain is color-coded throughout this and following figures. The proportion of GFP positive animals was assayed on at least 100 animals in 2 independent experiments (different days) with 3 biological replicates per condition each. The two experiments are represented by triangles and circles, respectively. (C) Proportion of GFP-positive animals for three transcriptional reporters (pals-5, F26F2.1, and sdz-6), 72 hours post inoculation of Orsay virus in Acinetobacter BIGb0102 or Comamonas BIGb0172, assayed in three independent experiments, with three replicates of 100 animals each. Each experiment is represented by a distinct shape of the datapoints. (D) Viral load at 72 hpi, measured using RT-qPCR of viral RNA1, in parallel with pals-5p::GFP reporter activation measured in six biological replicates of 100 animals each (left panel). From these six populations, three pooled samples were created by combining two populations each. Data normalization was achieved using the copy number of the gene eft-2 as an endogenous reference (right). (E) Proportion of ERT54 animals at 72 hpi, showing pals-5p::GFP activation when challenged with different Orsay virus strains on different bacteria, done in three replicates of 100 animals each. (F) Proportion of infected C. briggsae JU1264 animals at 72 hpi inoculated with Santeuil virus strain JUv1264 on different bacteria, assayed using FISH against the virus. Three biological replicates were evaluated per experiment, with animals per datapoint. Data are presented as mean ± standard error. Black symbols indicate the significance of the difference between the labeled bacteria and the Escherichia OP50 reference: *** P < 0.001; ** P < 0.01; * 0.01 < P < 0.05; P values higher than 0.05 are labeled as “ns” (same symbols for all the figures of this study). Significance was calculated for panels 2A-B using a general linear-mixed model where the bacteria was the fixed factor and experiments a random effect; for panel 2C we used a general linear-mixed model with a Gamma distribution and a log-link function in which the bacteria was the fixed factor; Tukey contrasts were used for post hoc analyses. For panels 2D-F we used an analysis of variance with bacteria as a factor and Dunnett’s test for post hoc analyses.
Fig 3.
Impact of selected bacteria on the fitness and developmental rates of C. elegans in the absence of viral infection.
(A) Upper panel shows total brood size of non-infected ERT54 animals when placed on each bacterial strain. Lower panels represent the daily production of viable progeny of non-infected ERT54 animals. Two separate experiments were conducted, in which the viable progeny of individual animals was monitored daily, with at least 5 individuals observed per bacterial type in each experiment. The upper panel represents the total progeny of those shown in the lower panels; the two experiments are represented by triangles and circles, respectively. (B) Proportion of adults assayed after exposing arrested axenic L1 larvae of the ERT54 strain to each bacterium, after 46 h (left panel) and after 62 h (right panel), in a population of 100 animals in each of three replicate populations per bacterial strain. Data are presented as mean ± standard error. Black symbols on the graphs indicate the statistical significance of differences when compared to the Escherichia OP50 reference: *** P < 0.001; ** P < 0.01; * 0.01 < P < 0.05. P-values exceeding 0.05 are labeled as “ns”. Significance was calculated for panel 3A using a linear-mixed model where the bacteria was the fixed factor with experiment as random effect; Tukey contrasts were used for post hoc analyses. For panels 3B we used an analysis of variance with bacteria as a factor and Dunnett’s test for post hoc analyses.
Fig 4.
Prevalence in mixed bacterial environments of the suppressive effect of viral infection.
In all panels the infection was evaluated at 72 hpi. (A-F) Activation of pals-5p::GFP reporter or FISH staining of viral RNA2 (panel B) after OrV inoculation of ERT54: (A) on bacterial lawns seeded with dual combinations of resistance-inducing bacteria and OP50 in the indicated proportions (assayed in four replicates of 100 animals each); (B) on bacterial lawns seeded with a 80–20% combination of resistance-inducing bacteria and BIGb102, a bacterial strain permissive for viral infection (assayed in three replicates of 100 animals each); (C) on a lawn of resistance-inducing bacteria supplemented with OP50 (assayed in three replicates of 100 animals each); (D) on a lawn of resistance-inducing bacteria supplemented with BIGb102; (E) on a lawn of Escherichia OP50 supplemented with live or heat-killed cultures of resistance-inducing natural bacteria (assayed in two independent experiments, one with three replicates and the other with four replicates of 100 animals each—each experiment is represented by a distinct shape of the datapoints); (F) on Escherichia OP50 supplemented with resuspended pellet or filtered supernatant of resistance-inducing natural bacteria (assayed in four replicates of 100 animals each). In all panels, the top row shows a schematic representation of the experimental design. Data are presented as mean ± standard error. Black symbols indicate the significance of the difference between the labeled bacteria and the Escherichia OP50 reference: *** P < 0.001; ** P < 0.01; * 0.01 < P < 0.05; P values higher than 0.05 are labeled as “ns” (same symbols for all the figures of this study). In panels A-D and panel F significance was calculated using a general linear model where the factors were bacteria and treatment. In panel E significance was calculated using a general linear-mixed model, where the factors were the bacteria and the treatment, and the experiment was considered a random effect. In both cases Tukey contrasts were used for post hoc analyses.
Fig 5.
Aversion to the bacterial lawn does not explain the suppression of viral infection.
(A) Around 100 axenic arrested L1 larvae of the ERT54 strain were placed around different bacterial lawns that where mock-inoculated with M9 (upper panel) or inoculated with OrV JUv1580 (lower panel). The proportion of individuals occupying the bacterial lawn was visually counted at different timepoints. (B) Proportion of animals showing a transcriptional IPR response (pals-5p::GFP activation; left panel) or infected intestinal cells (stained by FISH against the virus; right panel) at 72 hpi in plates fully covered by bacteria and the JUv1580 virus inoculum. Three biological replicates were tested per bacteria and 100 animals from each were assayed. Data are presented as mean ± standard error. Black symbols indicate the significance of the difference between the labeled bacteria and the Escherichia OP50 reference: *** P < 0.001; ** P < 0.01; * 0.01 < P < 0.05. P-values greater than 0.05 are labeled as “ns”. Significance was calculated using an analysis of variance with bacteria as a factor and Dunnett’s test for post hoc analyses.
Fig 6.
Bacterial environments can suppress OrV persistence over generations.
(A) Schematic representation of the experimental design (detailed in methods). ERT54 animals carrying the pals-5p::GFP reporter were inoculated with OrV JUv1580 on E. coli OP50 and transferred to selected bacteria. (B) Activation of the pals-5p::GFP reporter (upper panel) or proportion of infected animals (stained using FISH; lower panel) at two successive generations. Datapoints represent 100 animals, with three biological replicates per bacterial strain. The bar represents the mean ± standard error among replicates. Black symbols indicate the significance of the difference between the labeled bacteria and the Escherichia OP50 reference: *** P < 0.001; ** P < 0.01; * 0.01 < P < 0.05. P-values greater than 0.05 are labeled as “ns”. Significance was calculated using a general linear model where bacteria and generation were the factors. Tukey contrasts were used for post hoc analyses.
Fig 7.
Suppressive bacterial environments act after ingestion of virus.
(A) Activation of the pals-5p::GFP reporter in ERT54 animals on Escherichia OP50, challenged with viruses previously incubated with various bacterial cultures (detailed in Methods section “Coincubation of bacterial culture and virus inoculum”). Four replicate populations were evaluated per condition, with at least 100 animals assayed per population. Upper panel shows a schematic representation of the experimental design (detailed in Methods). (B) Proportion of GFP-positive ERT54 animals after initial exposure to the virus on Escherichia OP50 and subsequent transfer to different, non-virus-inoculated, bacteria. The upper panel shows a schematic representation of the experimental design (detailed in Methods section “Common garden inoculation experiment”). For the transcriptional response, three replicate populations were evaluated per condition and experiment. These three populations were pooled and the vRNA FISH stained to quantify the proportion of infected animals. Each data point represents a biological replicate, with at least 100 animals assayed per population. Independent experiments performed on different days are represented by the shape of the data point. Data are presented as mean ± standard error. "dpi" = days post-inoculation. Black symbols indicate the significance of the difference between the labeled bacteria and the Escherichia OP50 reference: *** P < 0.001; ** P < 0.01; * 0.01 < P < 0.05; P values higher than 0.05 are labeled as “ns”. Significance was calculated for panel A using an analysis of variance with bacteria as a factor and Dunnett’s test for post hoc analyses and for panel B using a general linear-mixed model where bacteria was the fixed factor with experiments as random effect and Tukey contrasts for post hoc analyses.
Fig 8.
Bacterial environments also suppress OrV infection in animals with hampered antiviral immune responses.
(A) Proportion of drh-1 animals that activate the lys-3p::GFP reporter after OrV inoculation on different viral resistance-inducing bacteria. (B-C) Proportion of infected animals, as assayed by RNA2 FISH, on two bacteria that enable slight activation of the reporter in panel A, for (B) wild-type and drh-1; (B) wild-type and rde-1 animals. (D-F) Proportion of infected animals, as assayed by RNA2 FISH, on the three selected suppressive bacteria for (D) wild-type, drh-1 and rde-1; (E) zip-1; (F) cde-1 animals. (G) Number of OrV-infected intestinal rings, as assayed by RNA2 FISH, in WT and drh-1 animals on the. In panels (A-F), three replicate populations were evaluated per condition. Each data point represents a replicate of at least 100 animals. Data are presented as mean ± standard error. In panel G three populations were pooled and 100 animals were evaluated. Black symbols indicate the significance of the difference between the labeled bacteria and the Escherichia OP50 reference meanwhile red symbols indicate the significance of differences between genotypes: *** P < 0.001; ** P < 0.01; * 0.01 < P < 0.05; P values higher than 0.05 are labeled as “ns”. Significance was calculated using an analysis of variance with bacteria as a factor (Fig 8A, 8E, and 8F) or a general linear model where the factors were bacteria and host genotype (Fig 8B, 8C, and 8D). In both cases Tukey contrasts were used for post hoc analyses.