The redundancy and diversity between two novel PKC isotypes that regulate learning in Caenorhabditis elegans

Significance

The nervous system can store an experience of sensory stimulus. This function (i.e., learning) requires a robust molecular mechanism because accurate readout of information is crucial in survival. In this study, we found that the gustatory learning of Caenorhabditis elegans implemented as the amount of diacylglycerol in the sensory neuron can be read out not only by the activity of PKC-1, a protein kinase C, but also by that of another PKC, TPA-1. Because of its low sensitivity to diacylglycerol, TPA-1 does not function in the conventional learning assay. However, under conditions that may impair the system, such as aging, TPA-1 contributes to the learning. Our study shows the robustness of the learning system achieved by the two PKCs.

Abstract

The nematode Caenorhabditis elegans learns the concentration of NaCl and moves toward the previously experienced concentration. In this behavior, the history of NaCl concentration change is reflected in the level of diacylglycerol and the activity of protein kinase C, PKC-1, in the gustatory sensory neuron ASER and determines the direction of migration. Here, through a genetic screen, we found that the activation of Gq protein compensates for the behavioral defect of the loss-of-function mutant of pkc-1. We found that Gq activation results in hyperproduction of diacylglycerol in ASER sensory neuron, which leads to recruitment of TPA-1, an nPKC isotype closely related to PKC-1. Unlike the pkc-1 mutants, loss of tpa-1 did not obviously affect migration directions in the conventional learning assay. This difference was suggested to be due to cooperative functions of the C1 and C2-like domains of the nPKC isotypes. Furthermore, we investigated how the compensatory capability of tpa-1 contributes to learning and found that learning was less robust in the context of cognitive decline or environmental perturbation in tpa-1 mutants. These results highlight how two nPKC isotypes contribute to the learning system.

Diacylglycerol (DAG) and its major target, protein kinase C (PKC), are key molecules in various biological processes (1, 2). In the nervous system, this signaling pathway functions in diverse events of information processing, including response to neurotransmitters (3), synaptic vesicle release (4), nociception (5, 6), and, most notably, learning (7–9).

PKC can be divided into three subclasses: conventional PKC (cPKC), novel PKC (nPKC), and atypical PKC (aPKC). Each PKC requires distinct upstream activators. Specifically, cPKC is activated by Ca²⁺ and DAG, nPKC is activated by DAG, and aPKC responds to neither DAG nor Ca²⁺. Furthermore, each subclass includes several isotypes: In mammals, PKC-α, PKC-β, and PKC-γ belong to cPKC. PKC-ε, PKC-η, PKC-δ, and PKC-θ belong to nPKC, while PKC-ζ and PKC-ι belong to aPKC (10). These diverse isotypes can function similarly (11, 12) or in a distinct manner (8, 10), suggesting that there are both differences and redundancies between PKC isotypes. However, how the redundancies and differences between these isotypes contribute to biological functions is an elusive question, especially in the nervous system. Previous studies using knockout mutants or pharmacological activation revealed the contribution of each PKC to behavior. Nevertheless, since the nervous system has a highly heterogeneous nature and PKCs are expressed in diversely overlapping patterns, it is difficult to attribute these behavioral consequences to the function of each isotype.

Functions of PKCs are well studied in the nematode Caenorhabditis elegans (13). Since the nervous system of C. elegans consists of a limited number of neurons (about 300), the function in certain neurons can be studied by cell-specific manipulation of each PKC (14–17). Several studies have shown that C. elegans uses DAG/PKC as a canonical signaling pathway to memorize environmental NaCl concentration in the gustatory sensory neuron ASER: when C. elegans is conditioned at a certain concentration of NaCl in the cultivation medium, it migrates toward that concentration on an NaCl gradient (18–20). In this type of learning, changes in ambient NaCl concentration are represented by the abundance of DAG in the gustatory sensory neuron ASER (19): decrease in NaCl concentration activates ASER and causes an increased [Ca²⁺] in ASER, which results in an increase of DAG through the activation of EGL-8/PLC-beta. Conversely, an increase in salt concentration causes the opposite event, leading to a decrease of DAG. Furthermore, when PKC-1, an isotype of nPKC, is activated in ASER by DAG, the animals climb up the NaCl gradient, and when PKC-1 is inactivated, they descend down the gradient.

To look more closely into the mechanism, we performed genetic screening to obtain a suppressor mutant of pkc-1(nj3lf, loss of function). Interestingly, we found that activation of the Gq protein, which is considered to act upstream of DAG production, compensates for loss of pkc-1. Further analysis revealed that Gq functions in ASER to elevate the amount of DAG and thereby activates TPA-1, an nPKC isotype closely related to PKC-1. As a result, TPA-1 compensates for PKC-1 downstream of Gq. However, despite the compensatory capability of TPA-1, the tpa-1 mutant did not show any obvious defect in learned behaviors. This difference in the phenotypes between the nPKC mutants was suggested to be due to differences in their sensitivity to DAG, because TPA-1 seemed to respond only to higher concentration of DAG to generate migration bias. The expression of domain-replaced tpa-1 suggested that this difference in sensitivity arose from the difference in the C1 and C2-like domains. Furthermore, we sought to reveal how compensatory capability and redundant function of tpa-1 contribute to learning. We found that tpa-1 mutants show an impaired learning in the contexts of aging and multisensory perturbation. The former phenotype was rescued by the expression of tpa-1 in ASER, while the latter was not.

These results suggest that tpa-1 is dispensable in the conventional chemotaxis assay due to its lower sensitivity, while it still has supportive function in various contexts including aging.

Results

Go Inactivation and Gq Activation Compensate for the Impaired Migrational Bias in NaCl Chemotaxis of the pkc-1 Mutant.

C. elegans is known to exhibit chemotaxis toward the salt concentration at which they were cultivated (18–20). To evaluate this type of learning ability, we used the established behavioral assay (Fig. 1A) (18); worms were conditioned at 25 mM, 50 mM, or 100 mM and placed on 9-cm plates with a salt gradient from ∼35 mM to ∼95 mM. Migration bias was quantified using the chemotaxis index (see Methods for definition). In this assay, as previously reported (18), pkc-1(nj3lf) mutants always migrated toward low salt concentrations regardless of the cultivation conditions (Fig. 1B). To identify mutations that suppress the abnormal migration bias of pkc-1, we mutagenized the pkc-1(nj3lf) mutant using ethyl methanesulfonate (EMS) and performed a forward genetic suppressor screening (see Methods). Among the obtained mutants, we found that a mutation in the N-terminal region of goa-1, His322Tyr, partially suppressed pkc-1(nj3) (Fig. 1B and SI Appendix, Fig. S1).

Fig 1.

Inactivation of Go and activation of Gq compensate for the migrational bias in NaCl chemotaxis of the pkc-1(lf) mutant. (A) Schematic diagram of salt-chemotaxis learning in C. elegans. Animals were conditioned on NGM plates with high (100 mM), intermediate (50 mM), or low (25 mM) concentration of NaCl. After 5 or fewer hours, they were placed on an assay plate with the NaCl gradient. (B) goa-1(pe3800[H322Y]) suppresses migration bias of pkc-1(nj3lf). Chemotaxis in wild type is shown in the left panel. The color of each bar represents the concentration of NaCl in conditioning plates (indicated on the right end of the graph). Error bars indicate SEM; n = 6 assays. P values were determined by ANOVA followed by Tukey’s test for comparison of each strain conditioned by the same concentration. ***P < 0.001 for the comparison to pkc-1(nj3). ###P < 0.001, ##P < 0.01 for the comparison to goa-1(pe3800);pkc-1(nj3). (C) egl-30(pe914[Y61N]gf) also suppresses pkc-1. n = 6 assays. **P < 0.01, ***P < 0.001 for the comparison to pkc-1(nj3) and ###P < 0.001 for the comparison to egl-30(pe914) in Tukey’s test.

It is known that goa-1 encodes the alpha subunit of the trimeric G protein, Go. It often functions by antagonizing the Gq pathway in C. elegans (21). Therefore, we examined whether the mutation pe914[Y61N]gf, a constitutively active mutation of the Gαq homolog egl-30 (22), would also compensate for pkc-1(nj3). The results showed that egl-30(pe914gf) can also suppress pkc-1(nj3lf) (Fig. 1C), indicating that activation of Gq compensates for the loss of pkc-1(lf) in the learned behavior.

Activation of Gq in ASER Compensates for the Chemotaxis Defect of pkc-1 Mutants.

Next, we sought neurons in which egl-30(gf) functions. Because pkc-1 functions in ASER (18), we first tested whether ASER-specific expression of egl-30(gf) would also compensate for pkc-1(nj3). We found that egl-30(gf) functions in ASER, the same site as pkc-1 (Fig. 2A). This was intriguing, since Gq activation generally transduces a signal by activating PLC-beta and thereby producing DAG. To further investigate the function of egl-30 in ASER, we performed DAG imaging using the inverse-type fluorescence probe Downward DAG2 (19, 23). Indeed, egl-30(gf) increased the basal (i.e., the absolute amount) DAG level in ASER (Fig. 2B). On the other hand, the proportion of increase in DAG amount induced by an NaCl decrease was not altered in egl-30(gf) (Fig. 2C), suggesting that the egl-30(gf); pkc-1(lf) mutant maintains a higher DAG amount compared to pkc-1(lf) even when they are subjected to any salt-concentration changes during the learning assay. This can explain the high-salt migration phenotype of egl-30(gf), which is similar to the behavior caused by chronic addition of DAG analog, phorbol ester, to wild type (ref. 18, Fig. 3D).

Fig. 2.

Activated Gq/egl-30 functions in ASER to promote DAG production. (A) Chemotaxis of pkc-1(nj3) mutants expressing egl-30(Y61Ngf) under the ASER-specific gcy-5 promoter. n = 6 assays (line #1) or 5 assays (line #2), respectively. P values were determined by Dunnet’s test. **P < 0.01, ***P < 0.001. Note that the transgene in line #2 was spontaneously integrated during isolation. (B) The effect of egl-30(pe914gf) on the basal DAG levels in ASER. Downward DAG2, an inverse-type fluorescent reporter of DAG, was expressed in ASER. Error bar indicates SEM. n = 13 animals (nj3) and 14 animals(pe914;nj3), respectively. P values were determined by Welch’s test. *P < 0.05. (C) The effect of egl-30(pe914gf) on DAG changes induced by NaCl decrease. Animals were conditioned with 100 mM NaCl and then transferred to microfluidic chip filled with imaging buffer containing 100 mM NaCl. A decrease in NaCl (100 to 50 mM, gray shed) was applied 50 s after the start of image acquisition. dF/F indicates change in fluorescence relative to intial fluorescence (F0, see Methods). Note that a Downward DAG2 is an inverse-type probe; therefore, a decrease in fluorescence is interpreted as an increase in the amount of DAG. n = 7 animals (nj3) and 8 animals (pe914;nj3), respectively. P values were determined by Welch’s test. n.s., not significant (P > 0.05).

Fig. 3.

TPA-1 functions downstream of egl-30(gf) to compensate for loss of pkc-1. (A) The tpa-1(fr1) mutation suppresses the compensatory effect of egl-30(pe914) in pkc-1(nj3) mutants. n = 6 assays. P values were determined by Tukey’s test. n.s., P > 0.05, ***P < 0.001 compared to N2. NS, P > 0.05, ###P < 0.001 compared to egl-30(pe914). +++P < 0.001 compared to egl-30(pe914);pkc-1(nj3). (B) Chemotaxis of tpa-1(fr1) mutants. n = 6 assays. P values were determined by Welch’s test. n.s., not significant (P > 0.05). (C) Suppression by tpa-1(fr1) was rescued by expression of wild-type tpa-1 under the gcy-5 promoter. P values were determined by Welch’s test. n = 7 to 8 assays. ***P < 0.001. (D) The dose-dependent effect of the DAG analog PMA in nPKC mutants. Animals were conditioned with NGM plates containing 50 mM NaCl and indicated concentrations of PMA. As described in the illustrations, pkc-1(nj3) mutants, which have functional tpa-1 but not pkc-1, showed chemotactic bias only when treated with high (>1.0 μg/mL) concentrations of PMA. P values were determined by Dunnett’s test. n = 6 assays. *P < 0.05,**P < 0.01, ***P < 0.001 for comparison to 0 mg/mL PMA.

These results suggest that Gq activation changes the salt-concentration preference through increase in the amount of DAG, which can activate PKC. Indeed, the phenotype of egl-30(pe914gf); pkc-1(lf) double mutant was significantly different from that of egl-30(pe914gf) (Fig. 1C), indicating that egl-30(pe914gf) is suppressed by pkc-1(lf) (mutual suppression). However, the suppression was partial, suggesting that factors other than PKC-1 function downstream of DAG.

TPA-1, a DAG-Dependent and Ca²⁺-Independent PKC, Acts downstream of Gq to Compensate for the Loss of PKC-1.

Next, we searched for other genes that function downstream of Gq. Apart from PKC-1, there is another PKC isotype called TPA-1 (13, 24). As predicted, the phenotype of the triple mutant demonstrated that tpa-1(fr1lf) strongly suppressed the effects of egl-30(gf) in the pkc-1(lf) background (Fig. 3A). Furthermore, the effects of tpa-1(fr1) were cancelled by the expression of wild-type tpa-1 in ASER (Fig. 3C). These results suggest that tpa-1 acts downstream of activated Gq in ASER and thus compensates for the low-salt chemotaxis bias caused by the loss of pkc-1. However, while tpa-1 seems to have a compensatory function in chemotaxis, the tpa-1(fr1) single mutant did not show any obvious abnormalities in chemotaxis (Fig. 3B). This suggests that tpa-1 has some characteristics that are different from pkc-1, which may prevent tpa-1 from affecting PKC-1–mediated transduction of DAG abundance.

Functional Diversity between PKC-1 and TPA-1.

Based on the known molecular mechanisms of Gq activation, we hypothesized that TPA-1 may only respond to high levels of DAG. To test this, we treated mutants with different concentrations of Phorbol 12-myristate 13-acetate (PMA), a potent DAG analog. When treated at 0.2 to 0.4 mg/mL, wild-type C. elegans showed a chemotaxis bias in the direction of high-salt concentration (Fig. 3D). As shown in previous studies (25, 26), motility was strongly inhibited by PMA at or above 1 mg/mL, preventing the measurement of chemotaxis. In turn, tpa-1(fr1) mutants did not show abnormalities in motility when exposed to high concentrations of PMA (25), whereas a chemotaxis bias to high-salt concentrations was induced by all concentrations of PMA at or above 0.2 mg/mL. Importantly, pkc-1(nj3) did not alter chemotaxis upon application of low concentrations of PMA, whereas chemotaxis shifted in the direction of high-salt concentrations when exposed to high concentrations of PMA at or above 1 mg/L. Furthermore, the effect of PMA was negligible in the pkc-1(nj3); tpa-1(nj3) double mutant. These results suggest that TPA-1 responds to a higher concentration of DAG than PKC-1 does.

To explore which molecular differences cause functional differences between TPA-1 and PKC-1, we examined whether we could rescue pkc-1(nj3) by expressing tpa-1, whose domain was replaced with that of pkc-1. Here, we focused on three major domains: C2-like, C1, and kinase domains. First, we expressed wild-type pkc-1 or tpa-1 in the ASER of the pkc-1(nj3) mutant. While expression of pkc-1 clearly rescued the phenotype, the expression of tpa-1 had little effect on chemotaxis (Fig. 4 A and B). This confirms the functional difference between the two nPKCs. Next, we expressed tpa-1 whose C1 or C2-like domain was replaced with the corresponding domain of pkc-1. No effect was observed in either case (Fig. 4 C and D). In contrast, when both C1 and C2-like domains were replaced, the chemotaxis of pkc-1 mutants was clearly rescued (Fig. 4E). This suggests that the cooperative activity of the C1 and C2-like domains is responsible for the difference in responsiveness between PKC-1 and TPA-1.

Fig. 4.

The effect of expression of chimeric genes of TPA-1 and PKC-1. (A) The effect of the expression of wild-type pkc-1 on chemotaxis in the pkc-1 mutant. n = 6 assays. (B) The effect of expression of wild-type tpa-1 in the pkc-1 mutant. Each transgenic line was generated by outcrossing egl-30(pe914);tpa-1(fr1);pkc-1(nj3);Ex[gcy-5p::tpa-1] in Fig. 3C; thus, the transgenic arrays must have an ability to rescue the phenotype of the triple mutant. n = 6 assays. (C) The effect of the expression of chimeric tpa-1 on chemotaxis, whose C2-containing sequence was replaced with that of pkc-1 in pkc-1(nj3) mutants. (D) Effect of the expression of chimeric tpa-1 on chemotaxis, whose C1-containing sequence was replaced with that of pkc-1 in pkc-1(nj3) mutants. n = 6. For A–D, P values were determined by Welch’s test. ***P < 0.001, **P < 0.01. (E) The effect of the expression of chimeric tpa-1 on chemotaxis, whose C1- and C2-containing sequences were replaced with those of pkc-1 in pkc-1(nj3) mutants. n = 7 assays. P values were determined by Dunnet’s test for comparison of each strain conditioned by the same concentration. ***P < 0.001 for the comparison to pkc-1(nj3).

These results suggest that differences in both C1 and C2-like domains lead to higher DAG sensitivity of PKC-1 compared with TPA-1; therefore, TPA-1 does not to show a significant function in our standard chemotaxis assay.

TPA-1 Contributes to Robustness of Learned Behaviors in the Context of Aging or Environmental Perturbation.

These results indicate that the nPKC isotypes PKC-1 and TPA-1 carry overlapping potentials, but in the absence of pkc-1, tpa-1 cannot replace pkc-1 under standard conditions. Then what is the biological significance of the compensatory capability of tpa-1 in chemotaxis? Because redundancy is generally considered to contribute to robustness of a system, we focused on the robustness of learning in the tpa-1 mutant. Inspired from a previous study, which reported that activation of Gq slows the aging-dependent decline of long-term memory in olfactory associative learning (27), we inspected whether loss of DAG/PKC component, TPA-1, might result in learning impairment in aged worms. Indeed, the ability of salt-concentration learning gradually declined in wild type with aging (SI Appendix, Fig. S3). In contrast to typical chemotaxis assays with young adult worms (days 0 to 1), aged (day 4) tpa-1 worms showed a significant impairment in learning when conditioned with high NaCl (Fig. 5A). This was rescued by the expression of tpa-1 in ASER, indicating that functional redundancy indeed contributes to the robustness of learning.

Fig. 5.

tpa-1 contributes to memory robustness. (A) Chemotaxis in the aged worms. tpa-1(fr1);Ex[gcy-5p::tpa-1] was generated by outcrossing egl-30(pe914);tpa-1(fr1);pkc-1(nj3);Ex[gcy-5p::tpa-1] line #1 in Fig. 3C. n = 6 assays. P values were determined by Dunnet’s test for comparison of each strain conditioned by the same concentration. n.s., P > 0.05, *P < 0.05 for the comparison to N2. (B) The effect of temperature transfer in wild type and tpa-1(fr1) mutants. tpa-1(fr1);Ex[gcy-5p::tpa-1] is the same transgenic line as panel A. n = 7 assays(N2), 8 assays (fr1, fr1;Ex[gcy-5p::tpa-1]). P values were determined by Welch’s test. n.s., P > 0.05, **P < 0.01 compared to N2 (left) or Ex(-) (right).

Furthermore, we tested whether tpa-1 contributes to robustness of learning under environmental perturbation. A recent study demonstrated that ASER is a polymodal neuron that perceives cold temperatures via goa-1 (28). This led us to investigate whether temperature change would perturb salt-concentration learning. To test this, we conditioned worms at room temperature and performed the chemotaxis assay in a cold environment (at 15 °C, Fig. 5B). Wild-type worms conditioned at 50 mM NaCl showed chemotaxis toward low concentration, showing remarkable impairment in learning. On the other hand, the learning at 100-mM conditioning showed a certain degree of tolerance. The tpa-1(fr1) mutant, in turn, was less tolerant to perturbation (Fig. 5 B, left). However, this defect was not rescued by the expression of tpa-1 in ASER (Fig. 5 B, right). As such, tpa-1 may function in cells other than ASER for this function. Likewise, we also tested the sustainability of memory by setting a prolonged posttraining period (SI Appendix, Fig. S2), showing that tpa-1 also contributes to the sustainability of memory by acting in neurons other than ASER.

Taken together, these results show that tpa-1 contributes to robustness of learning in ASER and other neurons.

Discussion

PKC-1 and TPA-1 Work Partially Redundantly.

In this study, we found that Gq activation and high levels of DAG abundance recruit tpa-1 to compensate for the loss of pkc-1 (Figs. 1B and 3A). Most previous reports have focused on different phenotypes for each mutant. For instance, tpa-1 mutants show impaired innate immune response, while pkc-1 mutants do not (29). Also, pkc-1 mutants show defects in tap response habituation, while tpa-1 mutants do not (30). Another study showed redundant functions of tpa-1 and pkc-1 in PMA-mediated enhancement of odor chemotaxis (16, 31). Our data indicate that pkc-1 and tpa-1 have redundant functions but have differential responsiveness to DAG.

The mammalian homologs of PKC-1 and TPA-1, PKC epsilon and PKC delta, are also known to function differently (32) or even oppositely (33), while another report suggests that they function redundantly (12). It will be worth examining whether these PKCs also have functional difference and redundancy in the nervous system, as observed in this study.

C1 and C2-like Domains Cause the Difference between PKC-1 and TPA-1.

Our data suggest that a DAG-binding domain, C1 domain, functions cooperatively with the C2-like domain to cause a difference between PKC-1 and TPA-1. Unlike the C2 domain of conventional PKCs, C2-like domains of nPKCs do not bind to Ca²⁺. On the other hand, C2-like domains play an important role in interactions with lipids or the C1 domain (34, 35). Furthermore, a large conformational difference is known to exist between the C2-like domain of PKC epsilon and that of PKC delta, which results in a difference in their binding capacities (34, 36). Our study suggests that the functional consequence of diversity in the C1 and C2-like domains may provide a strategy for designing specific inhibitors or activators of each PKC isotype.

tpa-1 Functions against Cognitive Decline.

Fig. 5A indicated the function of TPA-1 in aged worms. The DAG level might be high enough to at least partially activate TPA-1 when transferred from the 100-mM conditioning plate. This appears to be consistent with the phenotype of pkc-1(nj3), which occasionally shows a higher chemotaxis index in 100-mM conditioning compared to 25-mM conditioning, possibly by the function of TPA-1 (see Figs. 2A or 3 B and C, for example). In aged nematodes, the activity of PKC-1 is probably reduced, and TPA-1 likely compensates for the reduction of the activity. In mammals, the activities of PKCs are considered to decline with aging. Intriguingly, in the hippocampus, the function of PKC epsilon (37, 38), the PKC-1 homolog, is significantly reduced with aging, while that of PKC delta, the TPA-1 homolog, remains unchanged (37). Therefore, PKC delta might also function in cognitive decline in a similar way to TPA-1. If the function of PKC-1 declines with aging progression like PKC epsilon in mammals, TPA-1 might work instead of PKC-1.

Partial Redundancy Makes a Robust System.

In addition to the conditions we tested in this study, there could be multiple factors that lead to the modulation or dysfunction of PKC-1. This could be a general problem that signaling pathways such as DAG/PKC-1 may often face. In the case of a digital signaling system, in which there are quasibinary changes in the second messengers, such as the cellular response to rapid changes in the external ligand, the robustness of the system can be ensured simply by having multiple functionally equivalent kinases. On the other hand, in an analog system, such as the DAG/PKC coding system for salt chemotaxis, in which the strength of the signal must be precisely encoded, even if there are multiple kinases that are functionally identical, the total amount of phosphorylation as an output cannot be properly controlled; if the activity of one of the kinases is reduced, the total amount of output will change when the amount of DAG is not sufficient to fully phosphorylate the substrates. If there are two kinases, one of which having lower sensitivity, the dysfunction of the less sensitive kinase would not impact the total output because the phosphorylation of the substrates might be nearly saturated. In addition, when the other kinase is somewhat functionally declined, the less sensitive kinase can compensate for it if the amount of DAG is sufficient. Our results suggest that in such an analog system, the robustness of the system can be achieved by including redundant molecules with different sensitivities to the second messengers.

Methods

C. elegans Strains and Culture.

The list of strains used in the experiments is shown in SI Appendix, Table 1. All mutants were outcrossed at least five times. Wild-type animals corresponded to the Bristol strain N2. Animals were maintained at 20 °C on the nematode growth media (NGM) plate, unless otherwise noted (39). Day 4–aged worms were age-synchronized by bleaching. After they were sexually matured (4 d after bleaching, day-0 adults), we washed the worms with the wash buffer every other day to avoid contamination of progeny. We performed chemotaxis assays after 4 d.

For behavioral assays, the animals were fed with Escherichia coli NA22 bacteria. For imaging experiments, they were fed with OP50 to avoid noise from autofluorescence of the intestine.

Behavioral Tests.

Salt chemotaxis assays were performed as previously described with slight modifications (18). Briefly, NaCl gradients were generated on assay plates with 9 cm diameter by placing two agar plugs (cylinder, 14.5 mm in diameter, with 150 mM and 0 mM NaCl) 23 to 25 h before the assay. Adult animals were washed from the growth plate and conditioned on NGM plates at certain concentrations (25 mM, 50 mM, and 100 mM) of NaCl. After 5 to 7 h of conditioning, they were collected from conditioning plates, washed two times to remove bacteria, and then placed onto the assay plates with NaCl gradient. Just before the assay, agar plugs were removed and, in turn, 1 µL 0.5 M NaN₃ was spotted at the position of each agar plug. The animals were allowed to run on the plates for ∼60 min. Then, animals were immobilized by transferring plates to 4 °C. For the quantification of chemotaxis, we counted the number of worms in each region described in Fig. 1A. The chemotaxis index is calculated as (N_high-salt − N_low-salt)/(N_all − N_{start point}).

For PMA treatment, plates with each concentration of PMA were prepared by adding 10 mg/mL PMA in EtOH to NGM before pouring into plates.

Molecular Biology and Transformation.

Gateway technology (Invitrogen) was used to generate cell-specific expression plasmids; for ASER-specific expression, the gcy-5 promoter (3.2 kb upstream), which was cloned into the pENTR vector, was used. To clone complementary DNA (cDNA) or generate chimeric cDNA, we used the In-Fusion HD Cloning Kit (Takara).These were cloned into pDEST vectors. goa-1p::goa-1 and gcy-5p::egl-30(Y61Ngf) were the same plasmids as previously described (22, 40, 41). Germ-line transformation was performed as previously described (42). The total concentration of DNA was 100 to 110 ng/µL containing 5 to 20 ng/µL of an expression plasmid, 10 to 15 ng/µL of a fluorescent marker plasmid, and pPD49.26 as a carrier DNA.

DAG Imaging.

Imaging of Downward DAG2 was performed as previously described (19).

Animals were grown with OP50 for 4 d and transferred to 100-mM conditioning plates with OP50 as described in Behavioral Tests. The worms were then washed with imaging buffer from the 100-mM conditioning plates. The imaging buffer contains 100 mM NaCl, 25 mM potassium phosphate, 1 mM CaCl₂, 1 mM MgSO₄, and 0.05% gelatin; the osmolarity was adjusted to 350 mOsm with glycerol. Subsequently, the worms were loaded into a dimethylpolysiloxane (PDMS) chip attached to inlet tubes (41). NaCl step stimulus from 100 mM to 50 mM was delivered to the nose tip of the worms by switching the solutions. Images were captured at 1 frame per second by timelapse imaging using a LEICA DMI6000B inverted microscope. For quantification, the average fluorescence intensity over 50 frames (50 s) prior to stimulation was set as F0.

To measure the basal level of DAG, young adult animals fed OP50 were placed on 5% agar pads. The animals were paralyzed with 1 mM tetramisole hydrochloride. The images were captured using the experimental setup described in the previous paragraph.

Genetic Screens for pkc-1(nj3) Suppressors and Identification of goa-1(pe3800).

Mutagenesis with EMS was performed as previously described (39). The F1 progeny (150,000 worms in total) of mutagenized pkc-1(nj3) mutants were separated into 120 independent groups. The F2 animals (1,500 × 120 in total) were grown on NGM plates with 50 mM NaCl and placed onto the assay plates as described in Behavioral Tests. After 50 to 60 min, worms in the “High NaCl” area, described in Fig. 1A, were collected. We repeated this selection process four times and subsequently singled out a worm from each plate. Among 17 suppressor candidates obtained, only pe3800 showed strong suppression of pkc-1(nj3). pe3800 were backcrossed five times, and then we sequenced the whole genome of pkc-1(nj3) and a backcrossed strain (43). A high frequency of putative EMS-induced single-nucleotide variants (G to A or C to T) was observed on the left arm of the chromosome I (0 to 8 Mb). This region had 11 missense mutations, among which we found the goa-1 H322Y mutation. We validated H322Y as a causal mutation by a transgenic rescue experiment (Fig. 1B).

Statistical Analyses.

All statistical tests were performed in R scripts. Welch’s test was used for comparison of a pair of strains. For multiple comparisons, a one-way ANOVA followed by Dunnett’s tests or Tukey’s honestly significant difference tests were used. Strains conditioned with the same concentration are compared.

Data Availability

All study data are included in the article and/or SI Appendix. The raw dataset for all behavioral and imaging experiments and sequence information for domain-swapped transgenes are available on Figshare (DOI 10.6084/m9.figshare.17999759).