Barrier 1: Susceptibility, establishment and early-cycle progression

To examine host susceptibility and early infection outcomes in the selected species [42], we performed single-molecule fluorescence in situ hybridization (smiFISH) targeting OrV positive-strand RNA. Synchronized animals inoculated at hatching were collected at 24 hpi for all species and additionally at 14 hpi for C. elegans and C. remanei, and the proportion of animals with a positive smiFISH signal was determined. Because this assay detects viral RNA after inoculation, smiFISH positivity does not by itself distinguish among viral entry, uncoating, initiation of replication, persistence of incoming viral RNA, or rapid clearance. Thus, we interpret these data as measuring detectable viral RNA associated with susceptibility, establishment, and early-cycle progression, rather than as direct evidence for viral entry alone.

Viral RNA was detected at 24 hpi in every host species except C. remanei, in which positive smiFISH signal was observed only at 14 hpi (Fig 2A and 2B), consistent with the viral accumulation dynamics shown in Fig 3E (see Barrier 2 below). Host species had a significant effect on the proportion of animals with detectable viral RNA (Fig 2C; 24 hpi: χ² = 308.571, 4 d.f., P < 0.0001; 14 hpi: χ² = 223.319, 1 d.f., P < 0.0001). At 24 hpi, C. elegans showed the highest proportion of smiFISH-positive animals (64.0%), whereas all other species exhibited significantly lower rates of detectable viral RNA: C. tropicalis 6.3%, C. wallacei 1.8%, C. macrosperma 4.9%, and C. sulstoni 9.3% (P < 0.0001 for all comparisons). At 14 hpi, viral RNA was detected in only 5.2% of C. remanei animals, again significantly lower than in C. elegans at the same time point (79.5%, P < 0.0001).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 2. Representative smiFISH images of OrV infection in alternative host species.

(A) C. elegans and the alternative host species at 24 hours post-inoculation (hpi). (B) C. elegans and C. remanei at 14 hpi. The brightness levels of all images were equalized to allow for comparisons. All images are displayed with the same brightness and contrast levels to allow for visual comparisons, except for the images that have insets. The inset displays same brightness and contrast as the other images, and the magnified images have increased brightness to visualize low signal. Infection was not detected in any C. remanei worms at 24 hpi. Top: differential interference contrast (DIC), OrV RNA1 (cyan), OrV RNA2 (magenta), DAPI (blue). (C) Percentage of infected worms at 24 and 14 hpi determined by counting the number of animals with a positive smiFISH signal (n = 3 replicates of 56 ± 4 animals). The legend showing colors used for each species also shows phylogenetic relationships adapted from Stevens and colleagues (2020). The data underlying this Figure can be found in https://doi.org/10.5281/zenodo.19472999 (Fig2C-FISH_infection_rate.xlsx).

https://doi.org/10.1371/journal.pbio.3003965.g002

Download:

• PNG
  larger image
• TIFF
  original image

Fig 3. OrV viral load dynamics and single worm viral load distributions across host species.

(A) Schematic explanation of the experimental design. Synchronized populations of ~500 worms were inoculated at hatching with 2.8 × 10⁹ copies of OrV RNA2 and sampled at the indicated timepoints for total RNA extraction or single worm lysis and OrV viral load by measured by absolute RT-qPCR. Viral load over time throughout the first 44 hpi in the different host species: (B) C. elegans ERT54 (gray), (C) C. tropicalis (yellow), (D) C. wallacei (blue), (E) C. remanei (red), (F) C. macrosperma (green), and (G) C. sulstoni (pink). Viral load was measured as the number of OrV RNA2 copies in 10 ng of total RNA; 3–16 replicates per strain per time point. (H) Violin plots of OrV viral load of individual animals (n = 24 for each species) at 24 hpi. Viral load was measured as the number of OrV RNA2 copies in 1/5 of the single worm lysate. Horizontal gray lines represent the technical detection limit (mean + 2 SD of negative controls). The data underlying this Figure can be found in https://doi.org/10.5281/zenodo.19472999 (Fig3B-G-population_viral_load_curves.xlsx, Fig3H_FigS4-individual_viral_load.xlsx).

https://doi.org/10.1371/journal.pbio.3003965.g003

These differences indicate that alternative host species vary strongly in the frequency with which OrV RNA is detected after exposure. However, reduced smiFISH positivity could reflect one or more early bottlenecks, including inefficient entry, impaired uncoating, failure to initiate or sustain replication, or rapid elimination of viral RNA by the host. Although smiFISH detects positive-strand viral RNA rather than active replication directly, previous work indicates that detectable signals require substantial viral RNA accumulation [39]. We therefore considered smiFISH positivity together with viral-load time courses, RNA1 and RNA2 co-detection, viral localization, and transmission assays to infer whether infection progressed beyond early viral RNA detection.

Consistent with impaired early-cycle progression in some hosts, we also observed substantial fractions of RNA1-positive cells or animals lacking detectable RNA2 at 14 hpi, particularly in C. remanei (see Barrier 3 below). This pattern is consistent with incomplete progression to balanced accumulation of both viral genome segments and may reflect abortive early infection and/or rapid host clearance. Together, these results support the existence of strong early barriers to productive OrV infection in several host species.

Barrier 2: Within-host replication dynamics

To better understand the replication dynamics of OrV in the alternative hosts, we measured viral load during the first 44 hpi. For this purpose, synchronized populations of ~500 animals were inoculated at hatching and sampled at nine time points (Fig 3A). Viral load was quantified as the number of OrV RNA2 copies per ng of total RNA using the absolute RT-qPCR method. The viral load trajectory observed in C. elegans (Fig 3B) matched our previous results [39]. Among the different host species, viral load dynamics differed significantly, as did the interaction between host species and time (Fig 3B–3G; gray horizontal line indicates the technical detection cutoff; χ² = 281.081, 4 d.f., P < 0.0001; χ² = 302.444, 48 d.f., P < 0.0001, respectively).

Averaged across all time points, C. elegans exhibited the highest viral load of all host species (C. tropicalis, −77.4%, P < 0.0001; C. wallacei, −97.0%, P < 0.0001; C. remanei, −98.9%, P < 0.0001; C. macrosperma, −82.7%, P < 0.0001; and C. sulstoni, −59.6%, P = 0.0008). Moreover, the alternative hosts showed differences in the duration of the eclipse phase of viral accumulation compared to the natural host. Slopes estimated from an exponential growth model using the first 10 hpi were lower for all alternative hosts relative to C. elegans (C. tropicalis, −86.9%, P = 0.0001; C. wallacei, −123.3%, P < 0.0001; C. remanei, −97.4%, P < 0.0001; C. macrosperma, −34.5%, P = 0.0317; and C. sulstoni, −56.0%, P = 0.0089). These results suggest a possible desynchronization of the viral replication cycle, with a prolonged lag phase and/or slower exponential growth rate in the alternative hosts. Viral load remained high throughout the entire 44 hpi observation period in all species except C. remanei, which showed only a short replication period followed by viral clearance at 24 hpi. Overall, these results indicate that OrV replication occurs with species-specific dynamics that are slower, reach lower peaks, and reflect a temporal shift in the viral cycle relative to the natural host, C. elegans.

To further examine inter-individual differences in the ability of C. elegans and alternative hosts to support viral replication and accumulation, we inoculated synchronized animals at hatching and measured the viral load of individual animals at 24 hpi. After discarding individuals with values below the cutoff threshold (mean + 2 SD of RT-qPCR values from noninfected animals), we found significant differences in viral load among host species (Fig 3H; χ² = 30.757, 4 d.f., P < 0.0001). C. elegans and C. tropicalis formed a homogeneous group with higher viral load than the remaining four species (C. wallacei, −97.7%, P = 0.0001; C. remanei, −96.7%, P = 0.0001; C. macrosperma, −88.5%, P = 0.0389; C. sulstoni, −97.1%, P = 0.0001).

Viral load among individuals of the same species showed significantly right-skewed distributions in all cases: C. elegans (range: 5.3 × 10²–1.4 × 10⁸; g₁ = 3.537 ± 0.481, P < 0.0001), C. tropicalis (3.2 × 10²–2.9 × 10⁷; g₁ = 1.611 ± 0.752, P = 0.0322), C. wallacei (3.0 × 10²–1.3 × 10³; g₁ = 2.368 ± 0.616, P = 0.0001), C. remanei (3.0 × 10²–2.4 × 10³; g₁ = 1.247 ± 0.501, P = 0.0129), C. macrosperma (3.0 × 10²–1.69 × 10⁶; g₁ = 2.552 ± 0.524, P < 0.0001), and C. sulstoni (3.1 × 10²–1.9 × 10³; g₁ = 1.857 ± 0.550, P = 0.0007). To test whether bimodal Lognormal models better fit our data than a unimodal single Lognormal distribution, we compared fits between the two approaches (Table 1, Fig A in S1 Appendix). Based on BIC evidence weights (>0.95), viral load data from C. wallacei and C. sulstoni were well explained by single Lognormal distributions (Fig A in S1 Appendix, panels C and F). In contrast, model fit improved significantly for C. elegans, C. tropicalis and C. macrosperma when using a mixture of two Lognormals (Fig A in S1 Appendix, panels A, B and E), despite the addition of three parameters (Table 1). Statistical power was not enough to decide among models in the case of C. remanei (Fig A in S1 Appendix, panel D). For C. elegans, 83.3% of individuals fell into a distribution with a mean of 4.0 × 10³, whereas the remaining high producers followed a distribution with a mean of 2.1 × 10⁷. For C. tropicalis, 87.5% belonged to a distribution with a mean of 2.1 × 10², and high producers followed a distribution with a mean of 5.9 × 10⁶. For C. macrosperma, 87.5% produced low viral loads (mean 5.2 × 10²), while high producers followed a distribution with a mean of 1.1 × 10⁶. These results suggest that the abundance of high producers varies across species, having a direct consequence on the differences in the prevalence of infections among host species.

Download:

• PNG
  larger image
• TIFF
  original image

Table 1. Fitting of viral loads (VL) to unimodal or bimodal Lognormal models.

https://doi.org/10.1371/journal.pbio.3003965.t001

Barrier 3: Viral genome stoichiometry and life-cycle progression

To clarify the stage of the viral cycle reached in each host species, we further analyzed our previously obtained smiFISH dataset by determining whether infected animals contained only RNA1 or both viral RNA segments. Overall, most infected animals contained both RNAs, and all animals with viral signal in the intestinal lumen carried both RNA segments. However, at 14 hpi, RNA2 was undetectable in a substantially higher fraction of infected C. remanei animals than of infected C. elegans animals. Specifically, 4/9 infected C. remanei animals contained only RNA1, compared with 4/138 infected C. elegans animals (Fig 4A; Fisher’s exact test, P = 0.0004; odds ratio = 0.050). This enrichment of RNA1-positive/RNA2-negative infections in C. remanei suggests that OrV infection in this species frequently fails to progress to balanced accumulation of both viral genome segments, consistent with early-cycle failure or rapid clearance. By 24 hpi, RNA2-negative infections were uncommon among the species in which infected animals were detected (C. elegans, 2/123; C. tropicalis, 1/9; C. wallacei, 0/3; C. macrosperma, 0/8; and C. sulstoni, 1/16), and their frequency did not differ significantly among species (Fisher’s exact test, P = 0.2634). Thus, the clearest species-specific pattern was the elevated frequency of RNA1-only infections in C. remanei at 14 hpi, supporting the idea that OrV infection is often abortive or transient in this host.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 4. Spatial localization and RNA ratios of OrV infection in alternative host species.

(A) Bar plot of the proportion of animals in which OrV RNA2 signal was detected. Error bars represent the standard error of the Binomial proportion (C. elegans, gray; C. tropicalis, yellow; C. wallacei, blue; C. remanei, red; C. macrosperma, green; C. sulstoni, pink). (B) OrV RNA2:RNA1 ratios at 24 and 14 hpi. OrV RNA2:RNA1 ratios were calculated based on the normalized median intensity values of RNA1 and RNA2 smiFISH signal. (C) Regression plots of RNA1 and RNA2 intensities in each species and time point. As a null model, the regression line of the corresponding C. elegans time point is shown as a gray line in all panels. Deviations from the null model were evaluated using BIC evidence weights (Table A in S1 Appendix). Intensities are displayed in arbitrary units. (D) OrV localization within cells, intestinal lumen, or cell and lumen at 24 and 14 hpi. Bar plot shows the proportion of infected worms in each category within each species and time point. The numbers inside the bar plot denote the number of worms in each category. (E) Infected area and position of infection per animal relative to intestine length at 24 and 14 hpi. At 14 hpi, C. elegans and C. remanei differ significantly in the size of the relative infected area (10.4% vs. 5.7%, P = 0.0014). (F) Number of infected cells per animal at 24 and 14 hpi. The histograms represent the relative abundance of each category within each species and time point. The data underlying this Figure can be found in https://doi.org/10.5281/zenodo.19472999 (Fig4A-presence_of_RNA2.xlsx, Fig4B-C-RNA2_RNA1_ratios.xlsx, Fig4D_F-cell_lumen_localization_and_number_of_cells.xlsx, Fig4E_relative_area_and_position.xlsx).

https://doi.org/10.1371/journal.pbio.3003965.g004

We previously showed that the RNA2:RNA1 ratio changes dynamically during infection in C. elegans [39]. As an additional measure of viral cycle progression, we quantified the RNA2:RNA1 ratio in each species by comparing their respective normalized smiFISH fluorescence intensities. RNA2:RNA1 ratio was significantly higher in C. elegans (1.1 ± 0.6) than in C. remanei (0.6 ± 0.2) at 14 hpi (Fig 4B; 46.3%, P = 0.0228). At 24 hpi, overall significant differences in the ratio were also found (Fig 4B; χ² = 16.423, 4 d.f., P = 0.0025) among species, with C. elegans still showing the highest mean ratio (0.92 ± 0.07). However, pairwise differences between C. elegans and C. macrosperma (0.3 ± 0.2) showed the only significant case (P = 0.0345). (In all other comparisons: C. tropicalis 0.8 ± 0.3, P = 1; C. wallacei 0.04 ± 0.37, P = 0.1652; and C. sulstoni 0.04 ± 0.37, P = 0.0512.) The measured RNA1 and RNA2 intensities were strongly correlated after controlling for host species and time post-inoculation (Fig 4C; r_p = 0.787, 232 d.f., P < 0.0001). Most of the cells of the alternate species fell below the C. elegans regression line (shown in gray), indicating that the ratio of RNA2:RNA1 in these hosts is skewed towards a greater amount of RNA1 compared to the C. elegans average (in all cases, BIC evidence weights ≥ 0.8495 for the fitting to a regression line different from the one observed for C. elegans; Table A in S1 Appendix). Only a few cells in C. macrosperma and C. tropicalis, the two species with high-viral load individuals, fell above the C. elegans line; however, all observed values were within the variability of observations in C. elegans. These results indicate that, depending on the host species, OrV is less efficient at replicating RNA2 relative to RNA1 and therefore may be impaired in producing new virions.

Barrier 4: Virion egress and tissue localization

To further assess how far OrV infection progressed in each host species, we examined the localization of viral RNA signal in the smiFISH dataset. Infected animals were classified according to whether viral RNAs were detected within intestinal cells, in the intestinal lumen or in both compartments (Fig B in S1 Appendix, panel A). Detection of viral RNA in the lumen was used as a proxy for progression beyond intracellular RNA accumulation toward a potentially transmissible stage. At 24 hpi, viral RNA localization differed significantly among species (Fig 4D; χ² = 10.073, 3 d.f., P = 0.0180). In C. elegans, many infected animals showed luminal viral RNA, either exclusively in the lumen (32/99) or in both cells and lumen (13/99). By contrast, infection in the other species was predominantly cell-restricted, with luminal viral RNA detected only in one C. macrosperma individual (cell + lumen: 1/8). In C. elegans, luminal localization was less frequent at 14 hpi than at 24 hpi (14 hpi: lumen, 15/133; cell + lumen, 14/133; Fig 4D; χ² = 16.552, 1 d.f., P < 0.0001), consistent with luminal viral RNA reflecting a later stage of infection. Although two C. remanei individuals showed luminal viral RNA at 14 hpi, this pattern was not observed at 24 hpi. Thus, rather than indicating desynchronization per se, these data suggest that OrV infection in most alternative hosts rarely progresses to detectable luminal viral RNA, consistent with impaired completion of later steps such as packaging, egress, or maintenance of transmissible infection.

To complement localization, we quantified infected area and position along the intestine and number of infected cells per animal. At 24 hpi, no significant differences were detected among species in relative infected area (Fig 4E; ~10.1% of the measured area infected, χ² = 3.079, 3 d.f., P = 0.3796), and no significant positional differences were observed either (χ² = 7.502, 3 d.f., P = 0.0575). At 14 hpi, the infected area in C. remanei (5.7%) was smaller than in C. elegans (10.4%) (P = 0.0143), although no significant positional differences were observed (P = 0.6222). The number of infected cells did not differ significantly among species at 24 hpi (Fig 4F; χ² = 0.680, 3 d.f., P = 0.8780) or between C. elegans and C. remanei at 14 hpi (Fig 4F; χ² = 0.065, 1 d.f., P = 0.7992). However, individuals containing two infected cells were most common, and animals with three or four infected cells were observed only in C. elegans, not in any alternative hosts (Fig B in S1 Appendix), mirroring the higher viral load outliers observed more frequently in C. elegans (Barrier 2).

Barrier 5: Transmission competence

To investigate how differences in OrV replication and viral cycle dynamics affect epidemiological parameters in alternative host species, we measured the probability of OrV transmission from each host species to C. elegans. To do this, we transferred 20 inoculated, synchronized animals at 14 or 24 hpi into populations of 111 ± 26 healthy C. elegans ERT54 animals (Fig 5A). This design imposed a standardized donor-population size across host species, allowing direct comparison under identical exposure conditions. The two time points were selected because they correspond to the initial release of new OrV virions (14 hpi) and a more advanced stage of the viral cycle near clearance (24 hpi) in C. elegans [39]. The ERT54 strain contains an integrated fluorescent infection reporter that enables real-time tracking of infections by fluorescence microscopy [61]. Twenty-four hours after transfer, we quantified the percentage of ERT54 animals expressing the infection reporter.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 5. OrV transmission rates are time- and species-specific.

(A) Experimental design. Synchronized populations of inoculated animals were washed and populations of 20 worms or individual worms were transferred to healthy C. elegans ERT54 animals which contain a GPF fluorescent marker of infection. Twenty-four hours after the transfer, the percentage of infected ERT54 animals was determined by counting the number of animals expressing the infection reporter. (B) Transmission of OrV from inoculated populations of 20 worms (n = 3) of different host species (C. elegans, gray; C. tropicalis, yellow; C. wallacei, blue; C. remanei, red; C. macrosperma, green; C. sulstoni, pink) to populations of 111 ± 26 healthy C. elegans ERT54 at 14 and 24 hpi. (C) Transmission of OrV from individual worms of different host species (n = 22 ± 3) to populations of 38 ± 15 healthy C. elegans ERT54 at 24 hpi. The values above each violin plot denotes the percentage of individual worms that successfully transmitted OrV. The data underlying this Figure can be found in https://doi.org/10.5281/zenodo.19472999 (Fig5B-population_transmission.xlsx, Fig5C-individual_transmission.xlsx).

https://doi.org/10.1371/journal.pbio.3003965.g005

Transmission from groups of 20 donors differed significantly among host species (Fig 5B; χ² = 80.061, 4 d.f., P < 0.0001). We also detected a significant interaction between species and the age of the infected donor animals (χ² = 24.886, 6 d.f., P = 0.0004), largely driven by a pronounced decrease in transmission between 14 and 24 hpi in C. elegans (18.7% versus 7.4%, P = 0.0003). At 14 hpi, C. elegans exhibited the highest transmission probability (18.7%; C. tropicalis, 2.2%, P < 0.0001; C. wallacei, 1.2%, P < 0.0001; C. remanei, 2.7%, P < 0.0001; C. macrosperma, 0.6%, P < 0.0001; and C. sulstoni, 0.8%, P < 0.0001). By 24 hpi, however, transmission rates across species were more uniform, with significant differences remaining only between C. elegans and C. wallacei (1.4%, P = 0.0122) and C. elegans and C. macrosperma (1.4%, P = 0.0062).

To complement the population-donor assay and directly test whether individual animals could transmit OrV, we then transferred single donors at 24 hpi into populations of 38 ± 15 healthy ERT54 animals (Fig 5A). This assay avoids uncertainty about how many infected animals were present within each 20-donor group, although it still captures transmission only under the specific time point and experimental conditions tested. The proportion of individuals successfully transmitting infection differed significantly among host species (Fig 5C; χ² = 13.321, 4 d.f., P = 0.0098). These differences reflected two statistically homogeneous groups: a higher-transmission group composed of C. elegans (55.0%), C. tropicalis (26.1%), C. remanei (50.0%), and C. sulstoni (20.0%) (P ≥ 0.1828 for all comparisons within the group), and a lower-transmission group composed of C. wallacei (12.0%) and C. macrosperma (15.0%) (P ≤ 0.0205). Host species also significantly influenced transmission outcomes (mean fraction of infected recipients; χ² = 37.272, 4 d.f., P < 0.0001). C. elegans showed significantly higher mean transmission (3.6%) compared to all alternate hosts, except for C. remanei (Fig 5C; C. tropicalis, 1.1%, P = 0.0286; C. wallacei, 0.3%, P = 0.0004; C. remanei, 2.9%, P = 1; C. macrosperma, 0.6%, P = 0.0031; C. sulstoni, 0.5%, P = 0.0017).

Together, these results show that OrV transmission, like viral load, is strongly species-specific and exhibits substantial within-species variability under the assay conditions used here. The 20-donor assay provides a standardized comparison across hosts but does not necessarily estimate the maximum transmission capacity of each species, particularly for low-prevalence hosts in which infected donors may be rare. The single-worm assay complements this approach by testing whether individual animals can transmit OrV. Combining these results with our viral load data indicates that RNA accumulation and transmission are not tightly correlated in all species, and that the relationship between viral accumulation and transmission varies in a species-specific way.

Barrier 6: Virulence and host fitness effects

Although infection rates and viral loads were significantly lower in the alternative hosts, we sought to determine whether the developmental impact of OrV infection was nonetheless species-dependent. To do this, we imaged healthy and inoculated populations from hatching through 72 h and measured body length as a proxy for developmental rate. GLMs were fitted separately for each species.

For the two hermaphroditic species, C. elegans (Fig 6A) showed highly significant effects of infection status (χ² = 86.535, 1 d.f., P < 0.0001) and the infection status-by-time interaction (χ² = 31.261, 4 d.f., P < 0.0001) consistent with previous results [51]. Inoculated animals were, on average, 5.2% smaller. C. tropicalis (Fig 6B) showed a similar pattern, with both infection status (χ² = 57.286, 1 d.f., P < 0.0001) and its interaction with time (χ² = 38.959, 3 d.f., P < 0.0001) having strong effects on development. In this species, animals from inoculated plates were on average 3.4% smaller.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 6. Virulence of OrV is species-specific and can change based on host sex.

Virulence measured as the difference in animal body length between inoculated and control populations from hatching to 72 hours. (A) C. elegans (gray), (B) C. tropicalis (yellow), (C) C. wallacei (blue), (D) C. remanei (red), (E) C. macrosperma (green), and (F) C. sulstoni (pink). Mean body length of control populations is represented by full lines and of inoculated populations by dashed lines. Populations of gonochoristic species were separated by sex once it was clearly discernable at 48 and 72 hpi. Males of each gonochoristic species are shown in a different shade color. The sex of each population is displayed next to the corresponding line on the graph. The data underlying this Figure can be found in https://doi.org/10.5281/zenodo.19472999 (Fig6-virulence_body_length.xlsx).

https://doi.org/10.1371/journal.pbio.3003965.g006

Because reproductive modes differ among species and sex can only be distinguished reliably from the L4 stage onward, males and females in the gonochoric species were separated at 48 and 72 h. We then tested whether sex contributed to variation in the developmental effects of OrV infection. In C. wallacei (Fig 6C), both infection status (infected animals 2.6% smaller; χ² = 8.125, 1 d.f., P = 0.0044) and sex (males 19.6% smaller; χ² = 416.124, 1 d.f., P < 0.0001) had significant effects. Infection effects remained constant across these time points (χ² = 0.665, 1 d.f., P = 0.4147). Notably, a significant infection status-by-sex interaction (χ² = 7.774, 1 d.f., P = 0.0053) indicated that OrV symptoms were more severe in females (4.6% reduction) than in males (0.1%).

Across the other three gonochoric species (C. remanei, C. macrosperma and C. sulstoni; Fig 6D–6F), sex was the primary determinant of body size. Males were consistently smaller, by 13.9% to 16.6%, and also showed slower growth, as indicated by significant sex (χ² ≥ 160.734, 1 d.f., P < 0.0001) and sex-by-hpi (χ² ≥ 21.787, 1 d.f., P < 0.0001) effects. In none of these species did infection status, alone or in interaction with other factors, have a detectable impact on development (χ² ≤ 3.076, 1 d.f., P ≥ 0.0795 in all cases).

Together, these findings indicate that despite low infection rates in non-natural hosts, OrV infection can still exert measurable virulence at the population level.

Barrier 7: Evolutionary sustainability

After characterizing OrV infection in the alternative host species, we tested whether OrV could be maintained through serial passage in each host under a standardized filtration-based passaging regime. Five independent viral lineages were initiated per host species by inoculating populations of 20 synchronized animals, which were allowed to grow and reproduce until plate starvation. Virus-containing material was then collected by filtration and used to inoculate the next passage. We performed this experiment using two versions of the protocol, differing in the developmental stage of the animals at inoculation, L1 or L4 (Fig C in S1 Appendix, panel A). OrV viral load in the inoculum was quantified after each passage by RT-qPCR.

When passages were initiated by inoculating L1 animals (Fig C in S1 Appendix, panel B), 4/5 OrV lineages passaged in C. elegans maintained detectable viral loads that fluctuated between approximately 10⁴ and 10⁶ OrV RNA2 copies/µL during passages 1–5, followed by an increase to approximately 10⁷ copies/µL at passage 6. The fifth C. elegans lineage showed a large decrease in viral load between passages 1 and 2, followed by persistence at low but detectable levels through passage 6. In contrast, viral loads in all lineages passaged in the alternative hosts declined sharply after passage 1 and reached levels comparable to negative controls, below approximately 10³ copies/µL, by passage 3 in C. macrosperma and by passage 4 in C. tropicalis and C. wallacei.

A similar pattern was observed when passages were initiated by inoculating L4 animals (Fig C in S1 Appendix, panel C). All OrV lineages passaged in C. elegans maintained high viral loads of approximately 10⁶ copies/µL across the three passages performed with this protocol. By contrast, all lineages passaged in the alternative hosts again declined to low viral-load levels, approximately 10³ copies/µL or below, by passage 2 and continued to decrease thereafter.

Together, these results indicate that, under our filtration-based serial passage conditions, OrV was sustainably maintained in C. elegans but not in the alternative host species tested. We do not interpret this outcome as demonstrating an absolute inability of OrV to adapt to, or persist in, alternative hosts. Rather, failure to maintain viral load in these species may reflect the combined effects of reduced infection prevalence, inefficient progression to transmissible stages, transmission bottlenecks, inoculum dose, the number of donor animals, timing of collection, or the selective features of the filtration-based passaging regime itself. In particular, because this protocol transfers virus recovered from filtered material, it likely favors viral genotypes that are efficiently released into the medium, and may underrepresent host-associated virus that remains within cells, within the intestinal lumen, or associated with animals at the time of collection.

We therefore conclude that the alternative host species functioned as evolutionary sinks for OrV under the specific experimental passaging conditions used here. This result is consistent with the earlier infection, viral load, localization, and transmission assays, which together suggest that infection and transmission in these hosts were insufficient to sustain OrV persistence during filtration-based passaging. However, alternative passage designs, such as transferring infected animals directly or using worm-picking regimes, could impose different bottlenecks and selective pressures by carrying host-associated virus, host traits, or different viral compartments across passages. Such approaches may therefore yield different outcomes and would be required to test whether OrV adaptation to alternative hosts is possible under less restrictive transmission regimes.

Collectively, mapping infection outcomes onto the barrier framework reveals species-specific restrictions at multiple stages, including susceptibility and establishment, viral accumulation, genome-segment balance, progression toward luminal localization, transmission, and maintenance during serial passage (Fig 7). These barriers help explain why OrV persists and can be serially maintained in its natural host C. elegans, whereas the alternative Caenorhabditis hosts tested here acted as evolutionary sinks under our experimental conditions.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 7. Quantitative summary of barrier strength by host species.

Colors indicate whether each barrier was weak or not apparent, partial or time-limited, strongly restrictive, or not detected/not directly interpretable as a barrier metric. Categories are intended as qualitative summaries of the experimental results rather than independent statistical classes. The dashed outline around C. remanei highlights that mechanistic interpretation should be cautious because the number of smiFISH-positive animals was low.

https://doi.org/10.1371/journal.pbio.3003965.g007

Worm strain maintenance

C. elegans strains ERT54 and JU2624 were obtained from E.R. Troemel and M.A. Félix, respectively. C. elegans N2, C. tropicalis JU1428, C. macrosperma JU1857, C. wallacei JU1873, and C. sulstoni SB454 were obtained from the Caenorhabditis Genetics Center (CGC, https://cgc.umn.edu). The genetically diverse C. remanei SP8 generated through artificial selection [84,85] was obtained from M. Lind. For C. elegans, the ERT54 strain was used in experiments unless stated otherwise. All strains were maintained under standard conditions [86,87] at 20 °C on NGM plates seeded with live Escherichia coli OP50. C. elegans, C. tropicalis, C. wallacei, and C. remanei belong to the Elegans group, while C. macrosperma and C. sulstoni belong to the Japonica group [53].

OrV stock generation and quantification

OrV stock was generated by inoculating two 6 cm plates of freshly starved JU2624 worms with the OrV strain JUv1580_vlc (GenBank: PP738529.1 for RNA1 and PP738530.1 for RNA2) and expanding them on 50 9 cm NGM plates until starvation. Plates were washed with M9 buffer (0.22 M KH₂PO₄, 0.42 M Na₂HPO₄, 0.85 M NaCl, 1 mM MgSO₄) and pooled into 15 mL tubes. Tubes were vortexed for 30 s, incubated at room temperature for 10 min, and centrifuged at 2,250g for 2 min. The worm pellet was discarded, and the supernatant was transferred to a new tube, centrifuged twice at 21,000g for 5 min at 4 °C, and filtered through a 0.22 µm syringe filter. The resulting viral preparation was stored at −80 °C.

Viral RNA was extracted using the Viral RNA Isolation Kit (NZYtech, Lisbon, Portugal) following the manufacturer’s instructions. OrV titers were quantified by standard curve RT-qPCR using the qPCRBIO SyGreen 1-Step Go Hi-ROX kit (PCR Biosystems, London, UK) on an ABI StepOnePlus Real-Time PCR System (Applied Biosystems, Foster City CA, USA), with primers targeting a region of OrV RNA2 (5′ACGAAGCAGTAGCCGTTAAG3′ and 5′GAGAACATCCTTCTCTGCGG3′).

The standard curve consisted of six serial dilutions ranging 1.25 × 10⁹–1.25 × 10⁴ copies of OrV RNA2/µL, generated from an in vitro transcribed RNA2 fragment. cDNA for in vitro transcription was synthesized using AccuScript High Fidelity Reverse Transcriptase (Agilent, Santa Clara CA, USA) and a reverse primer binding the 3′ end of RNA2 (5′ATAGCCGGGTATGGATAGCG3′). Approximately 1,000 bp from the 3′ region were amplified using a forward primer containing a 20 bp T7 promoter sequence (5′TAATACGACTCACTATAGGCCTGTCAGAGTTGAGAACA3′) and DreamTaq DNA Polymerase (Thermo Fisher Scientific, Waltham MA, USA). The PCR product was gel-purified using the MSB Spin PCRapace kit (Invitek Molecular GmbH, Berlin, Germany) and transcribed in vitro using T7 polymerase (Merck & Co., Rahway NJ, USA).

Worm synchronization and inoculation

Synchronized worm populations were obtained by washing plates with M9 buffer to remove adults and larvae while leaving eggs behind. Plates were incubated at 20 °C for 1 h, and newly hatched larvae were collected with M9 and transferred to new plates. Larvae were inoculated by applying OrV stock containing a total of 2.8 × 10⁹ copies of RNA2 onto the E. coli OP50 lawns. Control plates received an equivalent volume of M9 buffer.

Unless stated otherwise, each biological replicate corresponded to an independently inoculated worm population maintained on a separate NGM plate. For assays in which multiple animals were scored from the same plate, individual animals were treated as subsamples used to estimate the plate-level outcome, such as the proportion of smiFISH-positive animals, the distribution of localization classes, or the fraction of infected recipients in transmission assays. Thus, the experimental replicate was the independently inoculated plate/population, not the individual animal alone.

Total RNA extraction and RT-qPCR

Populations of ~500 inoculated or control worms were collected using PBS with 0.05% Tween-20. Samples were centrifuged at 400g for 2 min, washed three times, flash-frozen in liquid N₂, and stored at −80 °C overnight.

For lysis, 500 µL of TRIzol (Thermo Fisher Scientific) or EasyBlue (iNtRON Biotechnology, Kirkland WA, USA) were added, followed by five freeze-thaw cycles between liquid nitrogen and 37 °C. Samples were vortexed five times for 30 s with 30 s rests. Chloroform (100 µL) was added, samples were mixed for 15 s, incubated 3 min at room temperature, and centrifuged at 16,000g for 15 min at 4 °C. The aqueous phase was transferred to new tubes, mixed with an equal volume of ethanol, and RNA was purified using the RNA Clean & Concentrator-5 Kit (Zymo Research, Orange CA, USA).

RNA concentration was measured with a NanoDrop OneC spectrophotometer (Thermo Fisher Scientific), and samples were diluted to 10 ng/µL. RT-qPCR was performed using 10 ng of total RNA as described above.

Single-worm lysis and RT-qPCR

Single-worm lysis was adapted from [88]. Synchronized infected and control worms were washed three times with PBS + 0.05% Tween-20 and transferred to NGM plates without E. coli OP50. Individual worms were picked into 0.2 mL tubes containing 5 µL of lysis buffer (5 mM Tris pH 8.0, 0.25 mM EDTA, 0.5% Triton X-100, 0.5% Tween-20, 0.1% proteinase K). Lysis was performed in a thermocycler at 65 °C for 10 min and 85 °C for 1 min. Lysates were stored immediately at −80 °C. RT-qPCR was performed using 1 µL of lysate as template.

Transmission assays

Synchronized, inoculated worms were washed three times with PBS + 0.05% Tween-20 and once with M9, then transferred to fresh plates. Worms were picked onto plates containing healthy C. elegans ERT54 worms [61]. For C. elegans, the N2 strain was used as donor to avoid confusion with the recipient ERT54 animals because the two strains do not show differences in viral load (Fig D in S1 Appendix; Kruskal–Wallis test, χ² = 0.365, 1 d.f., P = 0.5457). After 24 h, ERT54 worms were examined for infection by GFP fluorescence using a MZ10F stereomicroscope (10 × /23B objective; mCherry M10F/MZ FLII and GFP3 MZ10F filters) (Leica Microsystems GmbH, Wetzlar, Germany). The percentage of GFP-positive ERT54 worms was calculated.

smiFISH

The smiFISH protocol was adapted from [89,90]. Populations of ~700 inoculated and control worms were washed using the same procedure as for RNA extraction, except with one additional wash and centrifugation at 400g for 1 min. After the final wash, samples were fixed with 400 µL Bouin’s fixative, 400 µL 100% methanol, and 1 µL β-mercaptoethanol, mixed for 30 min at room temperature on a rotary shaker, flash-frozen, and stored at −80 °C overnight or longer.

Samples were then thawed at 4 °C for 30 min and washed four times with 1 mL borate-Triton buffer (20 mM H₃BO₃, 10 mM NaOH, 0.5% Triton X-100) and five times with the same buffer containing 2% β-mercaptoethanol. The last three washes included 1 h incubations at room temperature.

Samples were incubated for 5 min in wash buffer A with formamide (20% Stellaris Wash Buffer A, 20% formamide, 60% water), centrifuged, and resuspended.

Twenty-two probes targeting OrV RNA1 and 18 targeting RNA2 were designed using Oligostan [89]. Probe sequences can be found elsewhere [39–41,50]. Probes (0.83 µM) were annealed with a FLAP-label (CAL Fluor 610 or Quasar 670) at 85 °C for 3 min, 65 °C for 3 min, and 25 °C for 5 min. Hybridization was performed overnight at 37 °C in the dark.

The next day, samples were washed with wash buffer A, incubated 30 min at 37 °C, stained with 25 ng DAPI, washed with Stellaris Wash Buffer B, resuspended in 70 µL of Buffer B with 0.1 ng DAPI, and mounted with N-propyl gallate. For each replicate, 56 ± 4 animals were randomly selected and imaged using a DMi8 microscope with a DFC9000 GTC sCMOS camera (Leica Microsystems GmbH). Images were processed in Fiji [91].

Animals were counted as infected if at least one of the OrV RNAs was detected. Infected animals were analyzed by measuring the distance from the pharyngeal-intestinal junction to the start and end of the infected region. Relative infected area was calculated as the infected length divided by the total intestinal length. The median fluorescence intensity of infected cells was measured for both RNA1 and RNA2. Background values were obtained from control worms. Median fluorescence intensities were background-corrected and normalized by the number of probes used for each RNA, and RNA2:RNA1 ratios were calculated.

Growth images and measurements

Populations of 80 ± 25 synchronized infected and control worms were transferred to NGM plates with E. coli OP50. Plates were imaged at designated time points using a MZ10F stereomicroscope and Flexacam C3 camera (Leica Microsystems GmbH). Worm length was measured in Fiji by tracing the midline from the anterior tip of the head to the posterior end of the body, excluding the tail.

Experimental evolution

The first passage was started with populations of 20 synchronized L1 or L4 worms which were inoculated with 2.8 × 10⁹ OrV RNA2 copies in NGM plates seeded with 200 µL of E. coli OP50. The populations were allowed to grow and reproduce until plate starvation at which point worms and virus from each plate were collected with M9 buffer. Each sample was vortexed 3 times for 30 s with 30 s rests and left to incubate at room temperature for 10 min. Samples were then centrifuged for 2 min at 400g and 4 °C. The supernatant was then centrifuged for 5 min at 12,000g and 4 °C, filtered through a 0.22 µm syringe filter, and stored at −80 °C. For each subsequent passage, worms and plates were prepared as described for the first passage, however, inoculation was done with 100 µL of OrV filtrate from the previous passage. The OrV filtrate from each passage was quantified by standard curve RT-qPCR as described for OrV stock quantification above.

Statistical analysis

Statistical analyses were performed in R version 4.5.2 using RStudio Desktop version 2026.01.0+ (Posit, Boston, MA, USA) or in SPSS version 30.0 (IBM, Armonk, NY, USA). Figure panels were organized in Inkscape version 1.2.

For population-level assays, independently inoculated plates were treated as biological replicates. When multiple animals were scored within a replicate, the data were analyzed as replicate-level counts or proportions, with the number of scored animals defining the binomial denominator where applicable. This structure avoids treating animals sampled from the same plate as independent experimental replicates.

Log₁₀-transformed viral load data from population experiments were analyzed using generalized linear models (GLMs) with a Normal distribution and identity-link function. Nematode species was included as a fixed factor nested within its corresponding taxonomic group (Elegans or Japonica). Time post-infection (hpi) was incorporated as an orthogonal random factor, and the species-by-time interaction was also evaluated. For single-animal viral load estimates, hpi was not included as a factor. To characterize the heavy-tailed viral load distributions, we fitted two alternative models: (i) a unimodal Lognormal distribution with two parameters, and (ii) a mixture of two Lognormal distributions with weights p₁ and 1 − p₁ and four additional parameters. Model fitting was performed using the Levenberg-Marquardt algorithm.

Transmission data from population experiments were analyzed as replicate-level binomial responses using GLMs with a Binomial distribution and logit-link function, with the number of infected recipient animals and the total number of recipient animals defining the response for each independently inoculated donor population. Species was treated as a fixed factor nested within taxonomic group, and host age at inoculation (14 or 24 h) was included as an orthogonal random factor. The interaction between species and age was also tested. For single-animal transmission assays, age was not included as a factor.

Infectivity data from smiFISH experiments at 14 or 24 hpi were analyzed as replicate-level binomial responses using GLMs with a Binomial distribution and logit-link function, with the number of smiFISH-positive animals and the total number of scored animals defining the response for each independently inoculated plate. Species was treated as a fixed factor nested within taxonomic group.

The effects of nematode species on the area and relative position of infected cells along the body axis at 24 hpi were analyzed using GLMs with a Normal distribution and identity-link function, with species nested within taxonomic group as a fixed factor. Comparisons between C. elegans and C. remanei at 14 hpi were conducted using Mann–Whitney U tests.

Counts of infected cells per animal at 14 and 24 hpi were analyzed using GLMs with a Poisson distribution and log-link function, with species nested within taxonomic group as a fixed factor.

The presence of virus in distinct anatomical locations (cells, lumen or both) was analyzed using GLMs with a Multinomial distribution and an accumulated logit-link function, with species treated as a fixed factor nested within taxonomic group.

Ratios of normalized RNA2 to RNA1 fluorescence at 24 hpi were analyzed using GLMs with a Normal distribution and identity-link function, with species nested within taxonomic group. Comparisons between C. elegans and C. remanei at 14 hpi were performed using Mann–Whitney U tests.

Finally, developmental size data were analyzed by fitting GLMs with a Normal distribution and identity-link function, including infection status and hpi as orthogonal factors. Biological replicate was nested within the interaction between the two orthogonal factors. For gonochoristic species (C. wallacei, C. remanei, C. macrosperma, and C. sulstoni), gender was included as an additional orthogonal factor.

For model selection analyses, we used the Bayesian information criterion (BIC) to compare competing models while accounting for differences in model complexity. Relative support for each candidate model was quantified using BIC evidence weights, which were calculated from ΔBIC values and interpreted as the probability that a given model was the best approximating model among those considered. BIC evidence weights were used to evaluate support for unimodal versus bimodal Lognormal fits of single-animal viral load distributions and to compare species-specific versus C. elegans reference regressions in the RNA2 to RNA1 intensity analyses.

In all analyses, post hoc pairwise comparisons were conducted using the Holm-Bonferroni method. Additional statistical tests are described where relevant in the text.