We recently showed that P. pacificus bites C. elegans adults even though it is difficult to consume them. This biting of adult C. elegans prey forces these animals to leave the bacterial lawn, resulting in more exclusive access to the lawn by the predator (Quach and Chalasani, 2022). Using a modified version of the protocol in our previous study (Quach and Chalasani, 2022), we placed three predators and three C. elegans on an assay plate containing a small, dense bacterial lawn. Animals were restricted to an arena that included the lawn and a small area of empty agar (see Materials and methods). Control plates (‘mock’) had six C. elegans to maintain a consistent number of animals between plates with and without predators. These behavioral arenas were imaged under various experimental conditions, and coordinates of the eggs in arenas were determined. These coordinates were used to compute the distances of individual eggs from lawn center as well as their position relative to the lawn edge. Since Pristionchus also lay eggs, we used a C. elegans strain that expresses the GFP fluorophore in all of its eggs (Pelt-2::GFP) (Figure 1a).

To observe whether predator biting affects C. elegans prey behavior, we chose several different types of predators: P. pacificus strains PS312 and RS5194, a St-only P. pacificus mutant TU445 eud-1(tu445) (Ragsdale et al., 2013), and an isolate of P. uniformis, JU1051. P. pacificus strain RS5194 is more aggressive than PS312 as characterized by an increased probability of bite per encounter (Quach and Chalasani, 2022) so both strains were included in this analysis. The St-only (non-predative) mutant was included to demonstrate whether biting was required for predators to alter C. elegans behavior. We also included the cohabiting predator P. uniformis. As a more naturalistic predator which has coevolved with C. elegans, we wondered how prey response to this predator may differ from P. pacificus. P. uniformis males and females were considered separately, while only hermaphrodite P. pacificus were used. We first tested if short-term predator exposure could alter where eggs were laid by determining the numbers of eggs on and off bacterial lawns in our experimental arenas. These tabulations allowed us to fit a logistic regression model (Materials and methods, Equations 1 and 2) that estimated the probability of off lawn egg laying (‘P(off)’) as a function of time and in interaction with various predators or other conditions. To prevent eggs hatching into L1s, which secrete pheromones that promote lawn leaving (Scott et al., 2017) this assay only ran for 6 hr. C. elegans in general showed an increase to P(off) over time regardless of predator condition although animals exposed to the aggressive strain P. pacificus RS5194 showed slightly higher P(off) at 6 hr compared to mock (C. elegans only 0.22, RS5194-exposed animals 0.29, p=0.043) (Figure 1—figure supplement 2). We also observed an increase to P(off) between 3 and 5 hr when exposed to P. uniformis females but by 6 hr P(off) in both mock control and P. uniformis female-exposed conditions appeared comparable.

We hypothesized that increasing predator exposure time would more greatly increase the probability of off lawn egg laying in predator-exposed animals. We conducted a long-term assay with L4 C. elegans and J4 Pristionchus instead of adults and stopped the assay after 20 hr of exposure. Juveniles develop into adulthood over the course of the assay (C. elegans starts laying eggs approximately 8–10 hr after the L4 stage; Brenner, 1974). Thus, as eggs were laid primarily in the latter portion of the 20 hr time period, this limited L1 hatching during the assay. Arenas with P. pacificus eud-1 mutants showed similar P(off) compared to mock (C. elegans only)-exposed animals, while all other Pristionchus predators showed pronounced increases to the probability of off lawn egg laying (Figure 1b–c). These data indicate that interactions between eud-1 mutants and prey (secretions, contacts, and others) are unable to alter the locations of C. elegans eggs. We confirmed that this change was primarily due to altered egg laying location and not overall changes to the number of eggs (no significant change in egg numbers after predator exposure, Figure 1—figure supplement 3). While P. uniformis males triggered a similar proportion of C. elegans eggs to be laid off lawn (P(off)=0.72) compared to both strains of P. pacificus (RS5194 0.74, PS312 0.73), P. uniformis females had an intermediate effect (P(off)=0.62). Taken together, these experiments show that C. elegans change their location of egg laying away from a lawn occupied by primarily Eu Pristionchus capable of biting.

Next, we tested whether Pristionchus biting-induced injury was required for the change in C. elegans egg location. We used a C. elegans reporter strain expressing GFP (green fluorescent protein) under the control of an nlp-29 promoter. This strain (Pnlp-29::GFP) has been shown to increase GFP expression upon wounding the cuticle with a microinjection needle, a laser beam, or fungal infection (Pujol et al., 2008a; Pujol et al., 2008b). We paired each of the predators tested in our egg location assay with this reporter strain and monitored GFP fluorescence relative to the co-injection marker (Pcol-12::dsRED) (Figure 1c). We found that both isolates of P. pacificus (PS312 and RS5194) were able to increase reporter fluorescence in this reporter strain within 4 hr (Figure 1—figure supplement 4). In the 20 hr assay, C. elegans exposed to P. pacificus RS5194 were killed and could not be measured, but animals exposed to P. pacificus PS312 adults showed increased reporter fluorescence (Figure 1d). In contrast, the stenostomatus eud-1 mutant was unable to increase GFP fluorescence even after 20 hr. Notably, neither P. uniformis males nor females were able to increase GFP fluorescence in this Pnlp-29::GFP reporter strain. However, while it is difficult to confirm biting when the bites are relatively ineffective, C. elegans do appear to sense putative bites from P. uniformis by exhibiting escape response typical of other aversive stimuli (Quach and Chalasani, 2022; Hilliard et al., 2002; Video 1). It is possible that these bites are causing low level of harm without damaging the cuticle sufficiently to increase expression from the Pnlp-29::GFP reporter. We planned to use the C. elegans egg location assay for the remainder of our studies in non-fluorescent wildtype (WT) C. elegans and so chose a predator that does not lay eggs (P. uniformis males) in our assays (Figure 1e). Furthermore, failure to elicit a change in Pnlp-29::GFP fluorescence also indicated that changes to P(off) when exposed to P. uniformis animals in our egg location assay was not the result of extensive injury.

We next tested how the ratio of predators (P. uniformis males) to prey (C. elegans) affected the location of prey eggs and the expected value of P(off). While maintaining the same arena size and total number of animals (six), we altered the ratio of predators and prey. We found that the presence of even a single predator was able to increase the P(off) and adding additional predators resulted in greater increase to P(off), appearing to asymptote after ≥2 predators in the arena (Figure 1—figure supplement 5a). These changes to predator:prey ratio did not alter the overall abundance of C. elegans eggs (Figure 1—figure supplement 5b). These data are consistent with results in our previous study using P. pacificus (Quach and Chalasani, 2022).

As exposure to P. uniformis males did not result in strong injury to C. elegans but nevertheless was associated with changes to off lawn egg laying, we wondered whether, rather than biting itself, this phenomenon was due to compounds secreted by the predator. We have previously shown that P. pacificus secretions are aversive to C. elegans (Liu et al., 2018). We tested whether P. uniformis was secreting an aversive chemical that drives C. elegans away from the bacterial lawn. We conditioned lawns with P. uniformis males or sterile C. elegans as a control (to simulate changes in lawn caused by animal movement) and tested whether naïve C. elegans would alter their egg location on these lawns. We were unable to detect a shift in P(off) as the result of exposure to P. uniformis-conditioned lawns (Figure 1—figure supplement 6a). We did detect, curiously, an increase to the overall number of C. elegans eggs, though this was likely driven by an outlier effect (Figure 1—figure supplement 6b). In these assays, C. elegans was allowed to lay eggs in the arena for 2 hr. These data suggest that P. uniformis males either do not secrete a C. elegans aversive signal that can account for the observed predator-induced change to egg location or that C. elegans requires substantially longer exposure to such a signal compared to P. pacificus.

While C. elegans exhibits increased P(off) when occupying a lawn with predators, it may be that C. elegans is not truly avoiding the lawn in general, but simply altering its decision about where to lay its eggs. To determine where the prey themselves were located throughout the course of a predator exposure experiment, we used an imaging setup (WormWatcher) to monitor the locations of mScarlet-expressing C. elegans over 20 hr with images of acquired every 4 min (Figure 2a). We found that, when exposed to P. uniformis, C. elegans exhibited a shift in location to just outside the lawn boundary, starting at approximately 5–6 hr (Figure 2b). This shift in location was sustained in predator-containing arenas through the remainder of the 20 hr assay, while mock controls exposed only to other C. elegans remained mainly within the lawn. Thus, we infer that changes to P(off) observed in our egg location assays is likely a consequence of this sustained avoidance.

We observed that arenas containing C. elegans hermaphrodites and P. uniformis males, but not controls, had streaks of bacteria outside the main lawn (Figure 1e). Given the duration of our assay, these streaks might represent bacteria that sticks on the C. elegans body and gets deposited onto the agar as it exits the lawn. Over the duration of the assay, these streaks grow and are visible to the naked eye by the end of the 20 hr period. We tested whether the presence of streaks outside the main lawn alone could account for the change in egg location. We artificially streaked bacterial lawns at the beginning of the assay and monitored the location of the eggs in these predator-free arenas (Figure 3b). Indeed, artificial streaking was able to induce an increase in P(off) nearly threefold, however this response was greater in arenas containing P. uniformis (Figure 3c). These data showed that the presence of bacteria outside the main lawn can drive egg location change but may not be the only contributor to the decision of where to lay eggs when exposed to predator.

Next, we tested whether changes to egg location persist even in the absence of predators. We ‘trained’ C. elegans prey in our egg location assay setup with P. uniformis males for 20 hr and transferred only the prey to a test arena. The position of eggs laid in the test arena was quantified over 6 hr and subjected to the same analyses as our other egg location assays, allowing us to test more nuanced hypotheses about the effect of recent exposure to predator.

We tested animals recently exposed to P. uniformis or mock (C. elegans only) controls in three environments: completely filled arenas, normal small (~1.5 mm radius) lawn arenas, and arenas with artificial streaks as in Figure 3; Figure 4a. In a completely filled arena, there is no detectable lawn boundary. Rather than computing a P(off) statistic, we were able to use this arena to estimate predator-induced changes to overall distributional properties of eggs in the absence of a lawn boundary. We looked at the average distance from center eggs laid over 6 hr (Figure 4b) as an estimate of the prey’s central tendencies, as well as the coefficient of variation of the egg distribution (Figure 4c) which estimates changes to the width of the egg distribution that may have been brought about by recent predator exposure. We were unable to detect significant differences due to predator exposure, though we did detect a significant main effect of time on each metric. The average distance of eggs from center decreased over the course of 6 hr, while the coefficient of variation of these distributions tended to increase (Figure 4c). The estimated slopes for these effects over time are shown in Figure 4—source data 3.

When we tested artificial streaking (Figure 3), results suggested both effects of changes to bacterial topology and an interaction with the presence of P. uniformis males. We were curious about dynamics of this interaction in the absence of predator. We tested C. elegans recently exposed to predator or non-exposed controls in our learning paradigm in arenas containing either a small main lawn or a lawn with artificial streaks, and determined the number of eggs laid at 1–6 hr in independent arenas. We found a significant three-way interaction between time, recent predator exposure, and bacterial topology (Figure 4d). Animals exposed only to other C. elegans and then subsequently laying eggs in test arenas with small unperturbed lawns tended to have very low values of P(off) in general, which decayed negatively over time. By contrast, when tested in arenas with artificial streaks, not only was P(off) increased generally, but showed a positive relationship with off lawn laying increasing over time. When exposed to P. uniformis males and tested in arenas with unperturbed lawns, as expected animals did show a potentiation of P(off) and this potentiation to P(off) was comparable to that exhibited by C. elegans in the artificially streaked arenas at the early time points. However, in contrast to the temporal dynamics shown by changes to bacterial topology, P(off) was flatter with recent predator exposure across all time points. Finally, combining recent predator exposure and testing on lawns with artificial streaks showed the greatest potentiation to P(off) in general, with a similarly flat response over time. These results suggest that for at least 6 hr there are two separate phenomena: egg laying off the lawn driven by the presence of low concentrations of bacteria at a distance from the main lawn, and egg laying off the lawn driven by recent predator exposure. Predator exposure and artificial streaks together exhibit a combined effect on potentiating P(off) overall which is more than additive. With respect to the time evolution over 6 hr, recent predator exposure appears to trump the effects of bacterial topology, indicated by the relatively flat slopes in predator-exposed animals in either bacterial topological condition. The estimated slopes for these effects over time are shown in Figure 4—source data 4.

We wondered whether this elevation to P(off) would persist at even greater periods of time away from predator exposure. We transferred 20 hr exposed C. elegans to a rest plate completely filled with food for 1, 2, or 24 hr (Figure 5a). We then quantified eggs laid on a test plate containing artificial streaks, as it appeared that artificial streaking of the bacteria was likely to bring about the greatest potentiation of predator-induced changes to P(off). Consistent with the positive slope conditions observed over 6 hr in artificially streaked test arenas (Figure 4d), we saw a significant elevation of the baseline level of P(off) at 24 hr in the mock control condition where animals were not exposed to predator (Figure 5b). Predator-exposed animals showed elevation to P(off) at all time points including 24 hr, with a flatter relationship over time. This indicates that C. elegans are able to ‘remember’ their recent predator experience for at least 24 hr. However, it is also clear that the baseline probability of off lawn egg laying increases by 24 hr regardless of predator exposure, as exhibited in the mock condition. Thus, we computed changes to the fold change between predator and mock observed at each time point, and defined this as the predator response (see Materials and methods, Equation 3). This difference of differences captures the overall magnitude of observed shifts in egg laying behavior associated with the presence of predator. Although our data are not paired, the generalized linear modeling frameworks allows us to compute estimated confidence intervals on this fold change for performing statistical inference. We found that this response was significantly lower at 24 hr than at 1 or 2 hr, as a result of the increase in the baseline P(off) in the mock condition (Figure 5c). This indicates that while P(off) remains elevated, C. elegans may be beginning to extinguish its ‘memory’ of recent exposure by 24 hr.

Biogenic amines already are well established as modulators of egg laying behavior in general as well as egg laying during different locomotor modes (Alkema et al., 2005; Cermak et al., 2020; Horvitz et al., 1982). Additionally, biogenic amines are known to modulate behaviors over long time scales (Chase and Koelle, 2007), and the change in egg location behavior upon predator exposure appears to last for many hours. We hypothesized that egg laying behavior in response to predator might be subject to modulation by biogenic amines, and therefore tested mutants in genes required for their synthesis. We observed variable changes both to the baseline P(off) probabilities in animals not exposed to predator and to the magnitude of predator exposure response (Figure 6a–b). This is consistent with previous studies showing that dopamine and serotonin signaling is required for overall locomotion (Sawin et al., 2000; Flavell et al., 2013). In general, all mutants were able to show potentiation in P(off) when exposed to predator (Figure 6a). Mutants in the C. elegans homolog of the mammalian vesicular monoamine transporter, cat-1 (Duerr et al., 1999) exhibited a lower P(off) in the absence of predators. Although cat-1 mutant animals showed an increase to P(off) with predator exposure, the magnitude of predator response (as in Figure 5c) was lower than WT animals (Figure 6b). We also found that mutants in cat-2 (which encodes tyrosine hydroxylase for dopamine synthesis; Sulston et al., 1975; Lints and Emmons, 1999) and tph-1 (tryptophan hydroxylase for serotonin synthesis; Sze et al., 2000) had similar changes to baseline off lawn egg laying, but nevertheless increase P(off) in the presence of predator. This magnitude of increase was greater than WT in tph-1 mutants given the very low baseline P(off) in these animals in mock conditions, and the increase in cat-2(e1112) was similar in fold change magnitude compared to WT, again given their low baseline of P(off) in non-exposed conditions. Mutants in tbh-1 (which encodes tyramine beta-hydroxylase which converts tyramine into octopamine; Alkema et al., 2005) showed a similar baseline of P(off) to WT animals and a greater potentiation with predator. Mutants in tdc-1 (tyrosine decarboxylase, which converts tyrosine into tyramine) showed an elevation of P(off) in mock controls and a slight decrease in fold potentiation in the presence of predator compared to WT animals. Tyramine is known to inhibit egg laying (Alkema et al., 2005), however we did not detect significant changes to the number of eggs laid per C. elegans animals in tdc-1 mutants (not shown). Taken together, these data show that loss of biogenic amine neurotransmitters can modify off lawn egg laying behavior, attenuating or even increasing the observed response to predator, though these two phenomena were not so clearly separable. Loss of both dopamine and serotonin neurotransmitters in cat-1 mutants, however, not only reduced the general probability of off lawn laying but also contributed to the largest blunting of the predator response. We focused our remaining studies on dopaminergic signaling, but future work will investigate the role of serotonin signaling as serotonin has been previously shown to modify egg laying behavior (Schafer, 2005; Schafer, 2006).

We continued to investigate the consequence of loss of dopamine synthesis by testing a second mutant allele of cat-2, n4547. Both cat-2 mutants showed a similar reduction to baseline P(off) in mock conditions (Figure 6c), and a similar magnitude of predator response (Figure 6d). In C. elegans adult hermaphrodites, CAT-2 protein is expressed by eight neurons (four CEPs, two ADEs and PDEs), and dopamine signaling has been previously shown to affect modulation of locomotion as well as learning (Chase and Koelle, 2007). Additionally, analysis of the dopamine transporter promoter has identified specific elements that drive expression of transgenes in subsets of these dopaminergic neurons (Flames and Hobert, 2009). Using these cell-selective promoter elements, we expressed full-length coding sequence of the cat-2 cDNA under either CEP- or ADE/PDE-specific promoters. Transgenic rescue in ADE/PDE partially restored baseline P(off) in mock controls (Figure 6e), with rescue in CEP neurons resulting in the greatest increase to baseline P(off), even greater than WT levels. In this particular experiment, cat-2(e1112) mutants did in fact show a blunted predator response even though this metric accounts for the reduced levels of baseline P(off) in the mock condition, and both rescues also show significantly lower predator response compared to WT (Figure 6f). This indicates some variability in absolute loss of dopamine synthesis on modulating predator response vs. modulating off lawn egg laying in general. The cohorts of cat-2 mutants used in Figure 6a–b, Figure 6c–d, as well as the results shown in Figure 7 described below, indicate that changes to the underlying probability of laying eggs off the lawn is likely driving any observed effects to predator response. Additionally, differences in promoter strength used to drive expression of cat-2 may explain why dopaminergic cell types show differing ability to restore baseline P(off). Nevertheless, it is clear that re-expression of CAT-2 protein in either ADE/PDE or CEP is only sufficient to at least partially restore baseline off lawn egg laying behavior.

We also monitored locomotor activity of cat-2(e1112) animals over the course of 20 hr using the WormWatcher imager. Mutants were still capable of elevating distance from center upon predator exposure. However, there were approximately 40% fewer time points at which mutants differed between mock and predator-exposed conditions as compared to controls (Figure 6—figure supplement 1a–b). When computing confidence intervals for the fold increase (change between predator and mock conditions), both cat-2(e1112) mutants and WT exhibited similar response, though cat-2 mutants did show lower magnitudes of change at a few time points (Figure 6—figure supplement 1c). A mutant in the dopamine reuptake transporter gene dat-1, which has increased amounts of dopamine at synapses (Nass et al., 2005; Carvelli et al., 2004), showed a nearly identical response to WT animals (Figure 6—figure supplement 1d–f). Toward the end of the 20 hr observation period, however, dat-1 mutants in the mock condition began to move away from the lawn, consistent with the role of excess dopamine in altering locomotion (Hills et al., 2004; Chase and Koelle, 2007; Calhoun et al., 2015). These results suggest that dopamine signaling is required for off lawn exploration and changes in this pathway likely affects both animal position and egg laying distribution.

Next, we hypothesized that adding exogenous dopamine would restore normal egg laying to cat-2 dopamine deficient mutants. To test our hypothesis, we first exposed WT and cat-2 mutant C. elegans to P. uniformis males for 20 hr (training) and then transferred them to a plate with a lawn and artificial streaks (as in Figures 5–6) with and without exogenous dopamine (Figure 7a). This assay setup avoids exogenous dopamine from altering P. uniformis behavior, and leverages our data that prey responses persist for 24 hr even without predators. Previously 2 mM exogenous dopamine has been shown to rescue basal slowing upon encountering food (Sawin et al., 2000) and density pattern discrimination of PDMS pillars (Han et al., 2017) in cat-2 mutants. Consistent with our previous results, cat-2 mutants exhibited reduced off lawn egg laying in both control and predator-exposed conditions (Figure 7b). We found that adding 2 mM dopamine restored normal off lawn egg laying in both of these conditions. In the case of cat-2 mutants, addition of exogenous dopamine restored baseline P(off) to significantly greater levels than in WT, and thus exhibited a net reduction the predator response (Figure 7c). Together, these data suggest that dopamine signaling is required for off lawn egg laying in both control and predator-exposed conditions.

Complete loss of dopamine synthesis appeared to primarily affect baseline levels of egg laying activity off the bacterial lawn, so we next explored the roles of the cognate dopamine receptors in modifying this behavior. The C. elegans genome encodes at least four dopamine receptors (dop-1, -2, -3, and -4) with viable mutants in each (Chase and Koelle, 2007). These receptors each have multiple protein isoforms whose sequence alignments are depicted in Figure 8a. C. elegans DOP-1 is a homolog of the mammalian D1-like receptors and DOP-2/3 are homologs of mammalian D2-like receptors (Chase and Koelle, 2007). DOP-4 is also D1-like, however this receptor belongs to a unique invertebrate family of D1-like including receptors found in Drosophila melanogaster and Apis mellifera (Sugiura et al., 2005). We tested single mutants in each of these four receptors in our egg location assay along with a quadruple mutant that lacked all four receptors. P(off) was increased with exposure to predator in all cases (Figure 8b). Complete loss of all four receptors was associated with a trend to reduce the baseline of P(off) in mock controls (p=0.08 after multiple testing correction) and did not show a significant change to the predator response compared to WT (Figure 8c), which were similar effects observed when removing dopamine synthesis. Loss of individual receptors had varying results. Loss of dop-1, dop-2, dop-3 all elevated baseline P(off) to varying degrees (Figure 8b) and showed concomitant reductions to the magnitude of predator-induced fold increases (Figure 8c). Thus, loss of single receptors, though able to modulate overall fold change in P(off) when predator was present, still appeared to do so as a consequence of changes to background. Only dop-4 single mutants show mock condition P(off) not significantly distinct from WT and also showed comparable predator-evoked response.

Since dopamine receptors are known to exist as heteromers (Perreault et al., 2014), we analyzed mutants in pairwise combinations. Again, all combinations of two dop- mutants showed an elevation of P(off) when predator was present (Figure 9a). These combinations also had differing effects on baseline P(off) in mock controls. dop-1;dop-4 mutants were the most similar to WT. dop-1;dop-2, dop-2;dop-4, and dop-2;dop-3 all showed elevation of baseline off lawn egg laying activity relative to WT, and dop-1;dop-3 and dop-3;dop-4 showed reductions to baseline P(off). The magnitude of predator response in these mutant combinations is shown ordered from highest to lowest in Figure 9b. WT and dop-3;dop-4 double mutants show the highest fold change increase in P(off) relative to their respective mock controls. All combinations containing dop-4 rank intermediate with dop-2;dop-3 and dop-1;dop-3 ranking lowest. Other than dop-3;dop-4, all other combinations showed reduction to predator response relative to WT.

The ranked magnitudes of fold change in predator-exposed conditions suggested that combinations with dop-4 mutants were intermediate or closer to WT response level, regardless of changes to baseline P(off). To test the hypothesis of the presence or absence of just dop-4 influencing predator-evoked behavior, we performed an experiment comparing triple mutant animals in dop-1;dop-2;dop-3 to quadruple mutants of all four receptors (Figure 9c). Once again, the quadruple mutant showed reduction to the baseline P(off) in the mock control as in Figure 8, however, the triple mutant showed a comparable level of off lawn laying in the mock condition relative to WT. This enabled us to more easily interpret the significant reduction to predator response observed when comparing the triple mutant to the quadruple mutant or WT (Figure 9f). Taken together, dopaminergic receptor signaling can affect both baseline off lawn laying activity and predator response, and the specific exclusion of dop-1, dop-2, and dop-3 from the assembly of available receptors modulates predator response while maintaining otherwise normal levels of egg laying activity.

We speculate that responses by C. elegans to predator exposure fit within the broader context of prey refuge, wherein a prey adopts a strategy to reduce predation risk. The prey refuge brings with it the potential cost of decreased feeding opportunities, which is weighed against the benefit of minimal harm induced by the predator (Sih, 1987). This theoretical framework is consistent with the interactions we find between predator and changes to bacterial topology. Predator-exposed C. elegans shift egg to streaks away from a central lawn, and this strategy may lower the encounter probability with Pristionchus thus minimizing risk to the prey. This is especially so given the observations we have previously made that Pristionchus predators prefer to patrol a main lawn when it is available, thus leaving refuges open for exploitation by C. elegans (Quach and Chalasani, 2022). However, such refuges afforded by these streaks may have detriment to C. elegans fitness due to their lower density of available food, and longer term monitoring of health and fitness of prey in these conditions has yet to be tested. Variations to the number of available refuges for fleeing prey, as well as their local food density and quality, can be modified in the future to gain a better appreciation for this intriguing model of prey risk minimization strategy in C. elegans.

Dopamine has been shown to affect multiple C. elegans behaviors including locomotion, foraging, and learning (Hills et al., 2004; Calhoun et al., 2015). For example, we previously showed that this pathway is required for learning the size of a bacterial lawn and then driving a search strategy when removed from that lawn (Calhoun et al., 2015). Furthermore, dopamine has been shown to promote egg laying when animals roam (Cermak et al., 2020). This may explain the interaction effects we observe between predator exposure and artificially streaking bacteria. The combination of these inputs may motivate a roaming program, which continues to promote egg laying at a distance, explaining the large boost in P(off) observed in Figures 3c and 4d. Our video tracking data (Figure 6—figure supplement 1) suggests that cat-2(e1112) mutants are able to avoid the predator at least some of the time at perhaps an attenuated magnitude of response. However, despite this, they very rarely lay eggs off the lawn at all with P(off) values as low as 0.004 and as high as 0.13 across all experiments. Given that loss of dopamine synthesis appears to suppress P(off) and addition of exogenous dopamine restores this baseline (Figures 6–7), this is consistent with the hypothesis that dopamine is important in modulating egg laying while roaming. Even when cat-2 mutants are straying from the lawn, they are still by and large laying eggs on the lawn.

It is curious that in combination, loss of signaling via the DOP-1;DOP-2;DOP-3 receptors modulates predator response without modulating baseline P(off), while additional loss of DOP-4 modulates the baseline. This suggests that potentially the route through which egg laying while roaming is altered requires DOP-4, while predator response proceeds through signaling via the other receptors. Double mutant combinations in our data however are complex with both effects to magnitude of predator response and baseline. These data nevertheless stratify combinations with DOP-3 and either DOP-1 or DOP-2 as showing the most attenuated predator responses (Figure 9b). In Chase and colleagues’ work identifying DOP-3, triple mutants in dop-1;dop-2;dop-3 show attenuated basal slowing response in the presence of food, but nevertheless show normal dopamine-dependent paralysis, and this is also exhibited by dop-1;dop-3 double mutants (Chase et al., 2004). It may be that predator-evoked changes to shifting egg locations is linked to lawn edge detection. As dop-1;dop-2;dop-3 mutants in our work here show similar background levels of P(off) in the mock control condition, this may indicate that lawn edge detection is not the driving force in basal off lawn egg laying. However, when C. elegans learns to associate the main lawn with an aversive stimulus such as predator threat, then proper detection of the lawn edge would be crucial to avoiding it. However, when dop-4 is also mutated, it may be that this mimics loss of dopaminergic signaling observed in dopamine synthesis mutants, which in turn serves to modulate the baseline off lawn laying rate.

In the future, the role of serotonin should be further investigated. Serotonin has been shown to modulate dopamine-dependent behaviors. For example, while dopamine signaling is required for basal slowing when encountering a lawn of food, serotonin can enhance the slowing response if the animal is starved (Sawin et al., 2000). Thus, dopamine modulates basal behavior while serotonin modulates it in an experience-dependent manner. Whether serotonin acts in a similar manner in this assay is yet to be investigated.

In summary, after predator exposure, C. elegans lays eggs in areas of high food variability that still have some food, rather than laying eggs in a dense food patch inhabited and preferred by predators. Loss of dopamine synthesis alters baseline egg laying activity restored by exogenous dopamine, while nuanced combinations of dopaminergic receptors exert effects on specific predator-evoked response. This study lays the foundation for studying prey behavior in C. elegans. Future studies can use this system to interrogate the impact of various neurotransmitter signaling pathways on C. elegans feeding, reproductive, and general exploration strategies modified by experience.

Nematode strains used in this study are shown in the following table. Mutant crosses generated for this study in the table below available upon request (CGC = Caenorhabditis Genetics Center).

Nematode strains were maintained at 20°C on 6 cm Petri plates containing Nematode Growth Medium (NGM) seeded with Escherichia coli OP50 bacteria (CGC) as food (Brenner, 1974).

Assay plates are created by spotting 0.5 µl of OP50 liquid culture (OD600=0.5) on 35 mm standard NGM plates (Brenner, 1974). The bacterial lawns are allowed to grow at 20°C for 30 hr, then stored for up to 1 month at 4°C. Whatman filter paper with ¼" punch forms the ‘corral’ and encircles the lawn, allowing approximately 1.5 mm of clean agar in between the lawn edge and the corral edge. All animals are allowed to crawl on a clean section of agar to clean them of bacteria and picked to the assay plate using a sanitized eyelash, placed next to the lawn on a clean area of agar. Three predators are picked first, staged by overall size and pigment development as J4s. Then, three C. elegans L4s are picked to the assay plate. The animals are allowed to interact for a determined amount of time, 20 hr for an overnight assay, at 20°C. For short-term exposure (6 hr and under), gravid C. elegans adults and adult predators are used by picking L4s or J4s the day before to plates with plenty of food. The juveniles are allowed to grow overnight into adulthood and then used in the same assay setup. After their interaction, corrals and all adults are removed from the plate and the area inside the corral is imaged using an AxioZoom V16 (ZEISS).

For the streaked lawn variation, streaks are formed by gently dragging a sanitized eyelash through the center of the lawn in radial streaks 10 times, followed by two concentric circular streaks halfway between the lawn and the corral edge. The streaked lawn is then used immediately.

Injury assays are set up in the same way as the egg location assays, using a C. elegans strain containing the array frIs7 [Pnlp-29::GFP+Pcol-12::DsRed]. After the set interaction time, worms are immobilized by placing the plates on ice and imaged on an AxioZoom V16 (ZEISS) within 1 hr, with exposure times kept constant for fluorescence imaging (25 ms).

C. elegans are trained using the 20 hr egg location assay. At the same time as the animals used for training are transferred to their assay plates, test plates are set up. Three types of test plates are used: a filled lawn (10 µl of OP50 [OD600=0.5]), a streaked lawn (same as the streaked lawn variant of the egg location assay), and a small lawn (same as the original assay plate). The training plates with animals on them and the test plates are incubated at 20°C for 20 hr, during which the C. elegans is exposed to JU1051 males and the smears on the test plates are allowed to grow. (The bacteria on the other test plates are also allowed to grow at this time so that the bacteria are at a similar metabolic state and density across test plates, and streaks are already present.) Filter paper corrals like those used in the egg location assay are centered over the test plate lawns.

After the C. elegans are incubated in their training conditions for 20 hr, they are carefully removed with an eyelash pick from their training plates to a clean section of an NGM plate. The animal is allowed to crawl for a few seconds to remove bacteria and then picked to a test plate halfway between the central lawn and the corral edge. For the filled test lawns, the animals are placed in an equivalent position relative to the corral edge. The test plates are then imaged every hour on an AxioZoom V16 for 6 hr. For the variant including rest plates, the C. elegans are picked from their training plate to rest plates (‘filled lawn’ plates) for the set rest time. They are then transferred to a streaked lawn test plate and egg locations are observed after 2 hr. In learning experiments, all three C. elegans in predator-exposed conditions are transferred to a rest plate or test arena. In mock controls, where there are six C. elegans present, three C. elegans are selected randomly for transfer.

When adding exogenous dopamine to the learning assay, a 200 mM stock of dopamine hydrochloride (Code 122000100 Lot: A0427132, CAS: 62-31-7, Acros Organics) in water was prepared. Two hours before the trained worms needed to be transferred to the test plates, 50 µl of the dopamine stock or water as a control was gently applied onto the streaked lawn test plate. The plates were allowed to diffuse and dry with the lids off for 2 hr, at which time the trained worms were transferred to the test plates. The trained worms were allowed to lay eggs for 2 hr before their plates were imaged.

Egg location images are quantified in FIJI with the experimenter blinded to the condition by randomizing the file order and obscuring the filenames (using the Filename_Randomizer macro found at https://imagej.nih.gov/ij/macros/Filename_Randomizer.txt). Eggs are manually selected with the multipoint tool and lawns are selected as circles. Distances from each egg from to the lawn edge are calculated in Python. All assays are performed with their relevant controls over at least 3 separate days.

Assays conducted in the WormWatcher (Tau Scientific Instruments, West Berlin, NJ, USA) were performed on a single 6 cm 2.5% agar NGM plate in a 12-arena setup. The 12-well corral was created by cutting a 3×4 array of ¼" circles into a plastic sheet using a Cricut machine. OP50 was spotted in the 3×4 pattern using the same concentration and allowed to grow for the same amount of time as in the egg location assay although in this setup lawn radii averaged approximately 1.2 mm as compared to 1.5 mm in other assays. The increased agar percentage on the WormWatcher plates helped prevent worms from escaping under the plastic edges of the corral.

The assays were set up like the egg location assays, with three L4 C. elegans and three J4 JU1051 males or six L4 C. elegans in control arenas. The positions of predator-containing and/or mutant-containing arenas were alternated on different assay days. The WormWatcher was set to acquire fluorescent frames with a green LED excitation light every 4 min. A reference darkfield image was acquired before and after every experiment to reference the positions of the arenas and the size and positions of the lawns. After the experiment was completed, each arena was inspected and imaged to determine whether any worms escaped away or into it. Custom code was written to segment the C. elegans and arenas in each position and the median distance from the lawn edge to the midpoint of each worm body per well was recorded. Data from arenas were discarded if two worms had escaped from an arena, or if a P. uniformis was seen in a control arena.

Pristionchus mouthform analysis was performed as reported in Werner et al., 2017. Briefly, Pristionchus were egg-prepped via bleaching and eggs were cultured either on standard solid NGM plates or in liquid culture. After eggs reached adulthood, they were immobilized on agarose slides with sodium azide. The slides of different strains from different culture conditions were mixed and their labels obscured while they were observed. The slides were scored as either Eu (wide mouth, two teeth) or St (narrow mouth, one tooth) while the experimenter was blinded to strain.