To analyze mechanisms that modulate serotonin signaling, we investigated how Caenorhabditis elegans regulates the function of serotonergic motor neurons that stimulate egg-laying behavior. Egg laying is inhibited by the G protein Gαo and activated by the G protein Gαq. We found that Gαo and Gαq act directly in the serotonergic HSN motor neurons to control egg laying. There, the G proteins had opposing effects on transcription of the tryptophan hydroxylase gene tph-1, which encodes the rate-limiting enzyme for serotonin biosynthesis. Antiserotonin staining confirmed that Gαo and Gαq antagonistically affect serotonin levels. Altering tph-1 gene dosage showed that small changes in tph-1 expression were sufficient to affect egg-laying behavior. Epistasis experiments showed that signaling through the G proteins has additional tph-1-independent effects. Our results indicate that (1) serotonin signaling is regulated by modulating serotonin biosynthesis and (2) Gαo and Gαq act in the same neurons to have opposing effects on behavior, in part, by antagonistically regulating transcription of specific genes. Gαo and Gαq have opposing effects on many behaviors in addition to egg laying and may generally act, as they do in the egg-laying system, to integrate multiple signals and consequently set levels of transcription of genes that affect neurotransmitter release.
DEPRESSION is linked to a reduction in serotonin signaling since it is alleviated by selective serotonin reuptake inhibitors (SSRIs), which increase serotonin signaling at synapses (Lucki 1998). A better understanding of depression requires knowledge of the basic mechanisms that regulate serotonin signaling. The best-characterized example of serotonin signaling in a genetically tractable model organism occurs in the Caenorhabditis elegans egg-laying system (Schafer 2005). Egg-laying results from contraction of egg-laying muscles (ELMs), which are stimulated by serotonin released from the hermaphrodite-specific motor neurons (HSNs) (Trent et al. 1983). Egg-laying rate is strongly regulated (Schafer 2005) and thus provides a model for analyzing mechanisms that modulate serotonin signaling. Mutants for the neural G proteins Gαo and Gαq lay eggs too frequently and too rarely, respectively (Mendel et al. 1995; Ségalat et al. 1995; Brundage et al. 1996), and thus may be defective in mechanisms that adjust serotonin signaling in the egg-laying system.
Gαo is the most abundant G protein in the human brain and mediates signaling by many neurotransmitters (Sternweis and Robishaw 1984; Jiang et al. 2001), but the mechanism of Gαo signaling remains unclear. Gαo and Gαq have opposing effects on many C. elegans behaviors, apparently through opposing effects of these G proteins on neurotransmitter release (Lackner et al. 1999; Miller et al. 1999; Nurrish et al. 1999). Gαo activity decreases the abundance of the synaptic vesicle priming protein UNC-13S at presynaptic termini of C. elegans ventral cord neurons (Nurrish et al. 1999) and negatively regulates synaptic active zone size in C. elegans GABAergic neurons (Yeh et al. 2005). Although these results potentially reveal mechanisms by which Gαo acts to inhibit neurotransmitter release, it is unknown to what extent either of the aforementioned changes is actually responsible for the effects of Gαo on behavior.
It is also unknown whether Gαo and Gαq antagonize each other by functioning in the same cells or by acting in different cells. Each behavior affected by the G proteins is controlled by multiple neurons. Gαo and Gαq are expressed in every C. elegans neuron as well as in some muscles, including all cells of the egg-laying system (Mendel et al. 1995; Ségalat et al. 1995; Bastiani et al. 2003). Although some data suggest that the G proteins may act in the HSNs and/or ELMs (Ségalat et al. 1995; Bastiani et al. 2003; Shyn et al. 2003; Moresco and Koelle 2004), their site(s) of action in the egg-laying system remain unclear because this issue has not been systematically analyzed.
Here we have manipulated G-protein signaling in specific cells of the egg-laying system to determine where Gαo and Gαq act to regulate egg laying and have also visualized functional consequences of G-protein signaling using fluorescent reporters. Our results show that Gαo and Gαq function in the HSNs to have antagonistic effects on transcription of the tryptophan hydroxylase gene tph-1. This alters serotonin level in the HSNs and, ultimately, the rate of egg-laying behavior, apparently by setting the level of serotonin release from the HSNs.
MATERIALS AND METHODS
C. elegans strains were maintained at 20° under standard conditions, and double- and triple-mutant strains were generated using standard genetic techniques (Brenner 1974). The wild-type strain was Bristol N2; a list of all other strains used in this study can be found in the supplemental Materials and Methods at http://www.genetics.org/supplemental/. Mutations used were as follows: egl-30(n686), goa-1(n1134), goa-1(sa734), lin-15(n765ts), ser-4(ok512), tph-1(mg280), unc-4(e120), and arDp2.
Transgenes for cell-specific expression:
Cell-specific expression transgenes were based on a set of vectors that each contained a cell-specific promoter, a polylinker into which a cDNA of interest can be inserted, and the unc-54 3′ untranslated region. To determine the specificity of the cell-specific promoters, we inserted cDNAs for DsRed2, GFP, and cyan fluorescent protein (CFP) into the ventral cord type C neuron (HSN, VC) and ELM expression vectors, respectively, and generated a separate chromosomally integrated transgene for each construct. Strains carrying these transgenes were outcrossed and analyzed independently to show that each construct used drove expression in only one cell type of the egg-laying system. The transgenes were crossed together to obtain the triple-labeled strain LX975. In separate experiments, when we generated transgenic strains carrying the same plasmids as in LX975, co-injected and integrated as a single transgene, we observed a significant loss of specificity (data not shown). It is possible that enhancers from one promoter drive expression of fluorescent genes in other plasmids incorporated in the same transgenic array. Thus, for all experiments involving expression in more than one cell type, we generated separate integrated transgenes and crossed them together to avoid this problem.
cDNAs for signaling proteins were inserted into the vectors at exactly the same sites as the cDNAs for the fluorescent reporter proteins described above to minimize factors that might alter expression specificity. Each cell-specific promoter was used to drive cDNAs encoding GOA-1, GOA-1Q205L, the S1 subunit of PTX, EGL-30, and EGL-30Q205L. The HSN promoter was also used to drive expression of SNB-1∷GFP or UNC-13S∷GFP, cotransforming with a construct to express DsRed2 in the HSNs. Our UNC-13S∷GFP cassette was modified from that of Nurrish et al. (1999) to remove sequences 5′ of the transcription start site and the 5th through 10th introns (to remove the promoter at the 5′-end of the gene and an internal promoter). All integrated transgenes used in this article were generated by co-injection with a lin-15 rescuing marker plasmid into lin-15(n765ts) animals and integrated using psoralen/UV mutagenesis. Integrated strains were outcrossed at least four times to lin-15(n765ts). Supplemental Tables S1 and S2 (http://www.genetics.org/supplemental/) contain information on plasmids and integrated transgenes.
Triple-labeled C. elegans were immobilized with 10 mm levamisole, and a Zeiss LSM 510 confocal microscope was used to obtain a Z-stack image. Channel unmixing (LSM 510 software) was used to eliminate bleed-through observed between the CFP/GFP channels. The three-dimensional reconstruction in Figure 1 and supplemental Movie S1 (http://www.genetics.org/supplemental/) was created using Volocity software (Improvision).
Quantitative fluorescence microscopy was performed on a Zeiss LSM 510 confocal microscope (×63 objective with ×4 zoom), and image analysis was carried out using Volocity software. For imaging, late fourth larval stage (L4) animals were isolated, cultured 40 hr at 20° to produce staged adults, and immobilized with 3 mm levamisole. The left HSN (HSNL) cell-body images were single optical sections through the center of the cell body. Mean DsRed2 intensity in the HSNL cell body minus the nuclear area was measured using Volocity. All significant differences described were also significant without subtraction of the nuclear region. GOA-1 and EGL-30 effects on tph-1 reporter transgene expression levels were reproduced with five independent integrated transgenes (P < 0.001 for each; Student's t-test was used for all statistical comparisons in this work), although the fold changes seen varied between transgenes used, as GOA-1 increased expression 1.5- to 2.1-fold and EGL-30 decreased expression 1.8- to 6.3-fold. Since the transgene vsIs108 integrated on chromosome III, which allowed us to easily cross it with goa-1, egl-30, and all other integrated transgenes, vsIs108 was used for experiments in Figure 5 and in supplemental Figure S4 (http://www.genetics.org/supplemental/). Antiserotonin immunofluorescence staining was carried out on staged adult worms (see supplemental Materials and Methods) and imaging of the stained HSNLs was performed as described above for the fluorescent reporters. Quantitative analysis of the antiserotonin staining was performed by taking the mean fluorescence intensity from the entire cell body. Further details about image acquisition and SNB-1∷GFP and UNC-13S∷GFP analysis can be found in the supplemental Materials and Methods.
Unlaid eggs and early-stage laid eggs were quantitated as described (Chase and Koelle 2004). Staged adults were obtained by picking late L4 animals and culturing them for 40 hr at 20°.
tph-1 gene dosage:
Animals carrying different numbers of functional copies of the tph-1 gene were isolated by picking non-Unc cross progeny from the following genetic crosses: (i) tph-1(mg280) males × tph-1(mg280); unc-4(e120) hermaphrodites (zero functional copies); (ii) tph-1(mg280) males × unc-4(e120) hermaphrodites (one functional copy); (iii) N2 males × unc-4(e120) hermaphrodites (two functional copies); (iv) arDp2 males × tph-1(mg280); unc-4(e120) hermaphrodites (two functional copies); and (v) arDp2 males × unc-4(e120) hermaphrodites (three functional copies). This experiment was carried out twice (n = 60/genotype each time). Similar significant results were observed each time; Figure 7 contains data from one experiment.
Cell-specific promoters for the egg-laying system:
We set out to genetically manipulate G-protein signaling in the three individual cell types of the egg-laying system to determine in which cells Gαo and Gαq act to control egg laying. The egg-laying system consists of two HSN and six VC motor neurons, which synapse onto the ELM cells (White et al. 1986). We and others have developed a set of three cell-specific promoters that can be used to express proteins in the HSNs, VCs, or ELMs (Harfe and Fire 1998; Bany et al. 2003; Moresco and Koelle 2004). To analyze promoter specificity, we used the three promoters to transgenically express three different fluorescent proteins (Figure 1A). Confocal images of the transgenic animals demonstrated that each fluorescent protein was expressed in one cell type and was not detectable in the other two cell types of the egg-laying system (Figure 1, B–D). Although each promoter also directed expression in one or two additional cell types not related to egg laying (see supplemental Materials and Methods at http://www.genetics.org/supplemental/), their specificity within the egg-laying system allowed us to express signaling proteins in individual cell types of this system and to examine the consequences on behavior.
In the course of this experiment, we were able for the first time to simultaneously image the cells of the egg-laying system in living animals and visualize spatial relationships among these cells. Figure 1A is a two-dimensional representation of a three-dimensional confocal image of the egg-laying system shown in a rotation in supplemental Movie S1 (http://www.genetics.org/supplemental/). Our fluorescence imaging showed that the HSN and VC processes formed varicosities that contacted each other and the ELMs (boxed regions in Figure 1, B and C). The HSN processes were otherwise <1.0 μm in diameter, and the varicosities ranged from 3.6 to 6.4 μm across their largest dimension. These varicosities appeared to be the sites of synapses, as confirmed below. A number of features of the VCs and HSNs were variable between animals, including the precise paths of the processes and the number, size, and shape of the varicosities. However, the presence of the varicosities and their spatial relationships to each other and to the ELMs were consistent.
Gαo functions in the HSNs to inhibit egg laying:
In which cell(s) of the egg-laying system is Gαo signaling sufficient to have inhibitory effects on behavior? To determine this, we generated chromosomally integrated transgenes to express a constitutively active mutant form of GOA-1, the C. elegans Gαo ortholog, in the HSNs, VCs, or ELMs. The mutant protein used, GOA-1Q205L, is analogous to mutant forms of mammalian Gαs and Gαi, which cannot hydrolyze GTP and thus are locked in the active GTP-bound state (Mendel et al. 1995). We determined where GOA-1 signaling was sufficient to rescue the hyperactive egg-laying behavior of a goa-1 mutant by crossing the chromosomally integrated GOA-1Q205L transgenes into the mutant and counting the number of eggs retained in utero. For these and subsequent experiments, we used the goa-1(n1134) mutant (Ségalat et al. 1995), which lacks the N-terminal myristoylation site in GOA-1, preventing membrane localization and thus blocking most or all neurotransmitter signaling by GOA-1. While the wild type retained 15.5 ± 0.4 eggs in utero, the goa-1 mutant retained only 4.5 ± 0.4 unlaid eggs because it laid eggs too often (Figure 2, A–C). Expression of GOA-1Q205L in the HSNs of the goa-1 mutant rescued the hyperactive egg-laying defect and actually caused the retention of even more eggs (24.6 ± 1.4) than in the wild type, presumably due to excess GOA-1 signaling (Figure 2, A and D). An even greater gain-of-function effect was observed when the same transgene was used to express GOA-1Q205L in the HSNs of wild-type animals, resulting in accumulation of 48.5 ± 1.6 eggs/animal (Figure 2A). These results are consistent with those observed when GOA-1Q205L was expressed from extrachromosmal transgenes in the HSNs of wild-type animals (Moresco and Koelle 2004). By coexpressing a fluorescent protein with GOA-1Q205L in the HSNs and examining the cells, we determined that the accumulation of unlaid eggs in these animals was not the result of detectable defects in HSN development or morphology (data not shown). Neither rescue of the goa-1 mutant egg-laying behavior nor a gain-of-function effect in wild-type animals was detected when chromosomally integrated transgenes were used to express GOA-1Q205L in the VCs or ELMs (Figure 2A). Additional rescue results and controls can be found in supplemental Figure S1A (http://www.genetics.org/supplemental/), including experiments expressing wild-type GOA-1 rather than the activated GOA-1Q205L. Our results suggest that GOA-1 signaling in the HSNs, but not in the VCs or ELMs, is sufficient to inhibit egg laying.
To determine in which cell(s) of the egg-laying system GOA-1 signaling is necessary to inhibit egg laying, we used the cell-specific promoters to express the catalytic subunit of pertussis toxin (PTX), which inactivates Gαo proteins by ADP ribosylation. Ubiquitous expression of an S1 subunit of the PTX transgene in C. elegans results in a phenotype indistinguishable from that of goa-1 null mutants and suppresses the effects of GOA-1Q205L, indicating that GOA-1 is inactivated by PTX (Darby and Falkow 2001). Four other C. elegans Gα subunits (GPA-1, GPA-3, GPA-4, and GPA-16) are distantly related to Gαo and contain the conserved cysteine that is covalently modified by PTX activity. However, none of these proteins are expressed in the HSNs or the VCs, and only GPA-16 is expressed in the ELMs (Jansen et al. 1999). Thus, PTX can be used to specifically inactivate GOA-1 in the HSNs and VCs, but likely acts on both GOA-1 and GPA-16 in the ELMs. Expression of PTX in the HSNs of wild-type animals led to the retention of only 4.1 ± 0.3 eggs, compared to 15.5 ± 0.4 in the wild-type control, and thus phenocopied the hyperactive egg laying of the goa-1 mutant (Figure 2E). Expression of PTX in the VCs or ELMs did not cause hyperactive egg-laying behavior (supplemental Figure S1B at http://www.genetics.org/supplemental/). These results demonstrate that GOA-1 signaling in the HSN neurons is necessary for GOA-1 to inhibit egg-laying behavior.
Our results show that the HSNs are the principal sites where GOA-1 acts to inhibit egg laying. Since the HSN neurons release the neurotransmitter serotonin to stimulate egg laying (Trent et al. 1983; Waggoner et al. 1998), it is possible that GOA-1 inhibits egg-laying behavior by preventing serotonin release from the HSNs. Although we did not observe any evidence of GOA-1 function in the VCs or ELMs in our rescue experiments, we cannot exclude the possibility that GOA-1 may have some minor function in these cells, as has been previously suggested (Ségalat et al. 1995; Shyn et al. 2003). Shyn et al. (2003) saw effects of GOA-1 on the ELMs in animals lacking HSNs and VCs and thus suggested that GOA-1 functions in ELMs. However, these effects could also be the result of GOA-1 signaling in a cell outside the three cell types (HSN, VC, and ELM) that have been thought to compose the egg-laying system. Recently, GOA-1 was shown to act in the uv1/utse neuroendocrine cells to inhibit egg laying (Jose et al. 2007), and the observations of Shyn et al. (2003) could be explained by GOA-1 function in the uv1/utse rather than in the ELMs.
Gαq functions in the HSNs and ELMs to stimulate egg laying:
Analogous experiments were used to determine in which cell(s) signaling by the C. elegans Gαq ortholog EGL-30 is sufficient to stimulate egg laying. egl-30 mutants are defective in egg-laying behavior and, as a result, 50.0 ± 1.3 eggs accumulated in each animal, compared to 15.5 ± 0.4 eggs in the wild-type animals (Figure 3, A and B). We crossed chromosomally integrated transgenes that expressed constitutively active EGL-30Q205L in the HSNs, VCs, or ELMs in an egl-30 mutant strain and determined where such expression was sufficient to rescue the egl-30 egg-laying defective phenotype. The egl-30(n686) partial loss-of-function mutant was used for these experiments because complete loss of EGL-30 results in lethality (Brundage et al. 1996). Expression of EGL-30Q205L in the HSNs led to partial rescue (29.3 ± 1.8 eggs), while expression in the ELMs was sufficient for full rescue (13.2 ± 1.0 eggs; Figure 3, A, C, and D). Rescue effects were also observed when wild-type EGL-30 was expressed in the HSNs or ELMs of the egl-30 mutant (supplemental Figure S1C at http://www.genetics.org/supplemental/). It had earlier been suggested that EGL-30 acts in the ELMs to transduce signaling by serotonin released from the HSNs (Bastiani et al. 2003). Our results are consistent with this idea and also indicate that EGL-30 functions in the HSNs to allow these neurons to stimulate the ELMs.
The effects of EGL-30 signaling in the HSNs cannot be fully assessed only by expressing EGL-30Q205L in the HSNs of the egl-30 partial loss-of-function mutant, since the reduced EGL-30 function in the ELMs of these mutants prevents the animals from responding fully to HSN function. Thus, we crossed the transgene expressing EGL-30Q205L in the HSNs into a wild-type background, and this resulted in strong hyperactive egg laying, as these animals accumulated only 4.9 ± 0.3 eggs compared to the 15.5 ± 0.4 eggs retained by the nontransgenic control (Figure 3A). This gain-of-function effect demonstrates that, when the ELMs are fully functional, EGL-30 signaling in the HSNs is sufficient to strongly stimulate egg laying. Expression of EGL-30Q205L in the ELMs, but not in the VCs, of wild-type animals was also sufficient to induce a gain-of-function hyperactive egg-laying phenotype (Figure 3A). These results suggest that EGL-30 acts in the HSNs to stimulate release of serotonin and/or other neurotransmitters onto the ELMs, and also in the ELMs, to transduce signaling by neurotransmitters released from the HSNs. Furthermore, these data demonstrate that GOA-1 and EGL-30 antagonize each other by acting in the same neurons, the HSNs. This establishes the HSNs as model neurons for studying the mechanism by which GOA-1 and EGL-30 have antagonistic effects on serotonin release.
No detectable changes in HSN synapse morphology or presynaptic markers in Gαo mutants:
We used the HSN-specific promoter to express fluorescently tagged proteins to visualize potential changes at the HSN synapses that might reveal the mechanism by which GOA-1 affects neural function. We generated an integrated transgene to coexpress the red fluorescent protein DsRed2, to fill out the HSNs, with GFP-tagged synaptobrevin (SNB-1), which localizes to synaptic vesicles and labels presynaptic termini (Nonet 1999). HSN function was not significantly altered by expression of these fluorescent proteins since animals carrying the transgene did not exhibit egg-laying defects (supplemental Figure S2 at http://www.genetics.org/supplemental/). The transgene was crossed into wild-type and goa-1 mutant backgrounds, and we obtained three-dimensional confocal images of the synaptic region of the HSNL (Figure 4, A–I). We quantitatively analyzed SNB-1∷GFP fluorescence to look for effects of GOA-1 on presynaptic termini and DsRed2 fluorescence to look for effects on HSNL morphology (see supplemental Materials and Methods).
The SNB-1∷GFP fluorescence perfectly coincided with the varicosities in the HSNL process, indicating that the HSN synapses are located in the varicosities (Figure 4, A–C). Considerable variability was observed in the shape of the synaptic varicosities in wild-type HSNLs; however, there were consistently between two and four varicosities per process within 10 μm of the vulval slit (Figure 4, D–F). Analysis of SNB-1∷GFP indicated that the overall number of synaptic varicosities as well as total synaptic volume was not altered in the HSNs of the goa-1 mutant compared to the wild type (Figure 4, J and K). Analysis of DsRed2 fluorescence showed no difference between wild-type and goa-1 mutant animals in any parameter of HSNL morphology examined, including average width of the HSN process in nonsynaptic regions, mean distance across the largest dimension of the synaptic varicosities, or occurrence of branches emanating from a subset of varicosities (data not shown).
A previous study in ventral cord motor neurons showed that a GFP-tagged version of the synaptic vesicle priming protein UNC-13S accumulated at presynaptic termini at higher levels in goa-1 mutants than in wild-type animals (Nurrish et al. 1999). To determine if GOA-1 regulates egg laying by having this effect in the HSNL, we generated a chromosomally integrated transgene to coexpress both DsRed2 and the same UNC-13S∷GFP protein used by Nurrish et al. (1999) in the HSNs. Using three-dimensional confocal images, we quantitated UNC-13S∷GFP fluorescence in the HSNL synaptic region. UNC-13S∷GFP was concentrated 1.7-fold in the synaptic vs. nonsynaptic regions of the HSNL in both wild-type and goa-1 mutant animals (supplemental Figure S3 at http://www.genetics.org/supplemental/). The 1.7-fold synaptic enrichment of this diffuse UNC-13 isoform at wild-type HSN synapses is likely similar to the modest UNC-13S∷GFP enrichment at ventral nerve cord synapses, although UNC-13S∷GFP enrichment at ventral cord synapses could not be quantitated since it was visualized within a bundle of many neural processes (Nurrish et al. 1999). Our inability to detect an effect of GOA-1 on UNC-13S∷GFP in the HSNL similar to the effect seen in ventral cord neurons by Nurrish et al. (1999) could be due to differences between the cell types or imaging methodologies used. The lack of any detectable change in synaptic morphology or in presynaptic markers in the HSNL of goa-1 mutants led us to look for other mechanisms by which GOA-1 might affect serotonin release from the HSN neurons.
Gαo and Gαq regulate expression of the tryptophan hydroxylase gene:
In the course of the synaptic protein localization experiments, we came across an unanticipated and striking effect of GOA-1 and EGL-30 signaling on the HSN neurons. As detailed below, mutations in goa-1 and egl-30 had opposite effects on the levels of fluorescent proteins expressed using the HSN-specific promoter, suggesting that expression from this promoter is regulated by GOA-1 and EGL-30 signaling. This phenomenon is not likely to have affected the cell-specific rescue experiments using this promoter presented in Figures 1 and 2, as expression from the promoter remains specific to the HSNs (plus one other cell type outside the egg-laying system) in all genotypes examined. However, the different expression levels observed in the goa-1 and egl-30 mutants are of great interest because the promoter used in these experiments is that of tph-1, the sole C. elegans gene encoding the tryptophan hydroxylase enzyme that acts at the rate-limiting step in serotonin biosynthesis (Sze et al. 2000).
Mutations in goa-1 caused a twofold increase in expression of fluorescent transgenes driven by the tph-1 promoter in the HSNL (Figure 5, A–C), suggesting that GOA-1 signaling reduces expression from the tph-1 promoter. A sixfold decrease in fluorescence intensity in the HSNL was observed in egl-30 mutants, indicating that EGL-30 signaling stimulates expression from the tph-1 promoter (Figure 5, A, B, D, and E). The opposing effects of GOA-1 and EGL-30 on tph-1 expression were reproduced using five independent chromosomally integrated transgenes (data not shown). Each consisted of the tph-1 promoter and 23 bp of the tph-1 5′ untranslated region, followed by coding sequences for a variety of DsRed2 or GFP proteins and the unc-54 3′ untranslated region. An analogous GFP transgene in which the tph-1 promoter was replaced by the unc-86 promoter was also expressed in the HSNs (Adler et al. 2006) but no changes in expression levels were observed in goa-1 and egl-30 mutant animals compared to the wild type (Figure 5A). These results show that GOA-1 and EGL-30 antagonistically regulate expression from the tph-1 promoter, presumably by regulating transcription.
We tested whether GOA-1 and EGL-30 act cell autonomously in the HSNs to have their opposing effects on tph-1 expression. We crossed the chromosomally integrated transgenes that had previously been used to manipulate GOA-1 and EGL-30 function in the HSNs into one of the tph-1 promoter∷DsRed2 transcriptional reporter strains. Expression of PTX specifically in the HSNs to inactivate GOA-1 caused a threefold increase in fluorescence intensity (Figure 5F). This effect was greater than that observed using the goa-1 mutation (Figure 5A), likely because PTX causes more complete inactivation of GOA-1 than the partial loss-of-function goa-1 mutation used. Expression of GOA-1Q205L in the HSNs of the goa-1 mutant caused a 2.5-fold reduction in expression levels of the tph-1 reporter transgene (Figure 5G). EGL-30Q205L expression in the HSNs of the egl-30 mutant caused a 2.3-fold increase in expression levels of the tph-1 reporter transgene (Figure 5H). A 1.5-fold increase in expression levels of the tph-1 reporter transgene was also observed when wild-type EGL-30 was expressed in the HSNs of the egl-30 mutant (1311 ± 90 fluorescence intensity units compared to 888 ± 83 fluorescence intensity units in the egl-30 mutant control). These results show that GOA-1 and EGL-30 act cell autonomously to regulate tph-1 expression in the HSNs.
We considered the possibility that serotonin released from the HSNs signals through GOA-1-coupled autoreceptors, as suggested by Shyn et al. (2003), to feedback inhibit tph-1 expression and serotonin production in the HSNs. However, we saw no effects on tph-1 reporter expression in tph-1 mutants (supplemental Figure S4 at http://www.genetics.org/supplemental/), which lack serotonin (Sze et al. 2000), or in mutants for the SER-4 G protein-coupled serotonin receptor (supplemental Figure S4), which inhibits egg laying by an unknown mechanism (Dempsey et al. 2005). We also saw no decrease in tph-1 reporter expression in the HSNs when we chronically exposed animals to 7.5 mm serotonin (data not shown). Thus, we were unable to detect evidence for serotonin signaling causing feedback inhibition of tph-1 expression.
Gαo and Gαq regulate serotonin levels in the HSNs:
Since tryptophan hydroxylase functions at the rate-limiting step in serotonin biosynthesis (Jéquier et al. 1967), changes in tph-1 transcription could affect serotonin levels. To analyze serotonin levels in the HSNs, animals were fixed and stained with an antiserotonin primary antibody, followed by a fluorescent secondary antibody. The HSN cell body, process, and synaptic varicosities stain visibly using this method (Figure 6A). This staining is specific to serotonin since it was absent in the tph-1 null mutant (data not shown). We quantitated antiserotonin stain intensity to determine relative serotonin levels in the HSNs of wild-type, goa-1, and egl-30 animals. The goa-1 mutant showed a 1.3-fold increase in antiserotonin staining in the HSNs (Figure 6, B, C, and E), suggesting that the increase in tph-1 transcription seen in the goa-1 mutant caused an increase in serotonin synthesis. A 1.3-fold increase in antiserotonin staining was also observed in animals in which GOA-1 was inactivated by PTX expression in the HSNs (data not shown). A 1.3-fold decrease in antiserotonin staining was observed in the HSNs of the egl-30 mutant (Figure 6, D and E), indicating that the decrease in tph-1 transcription seen in the egl-30 mutant resulted in a decrease in serotonin synthesis. The intensity of the antiserotonin staining is not likely a linear reflection of serotonin levels, and the 30% changes in staining in the G-protein mutants may not indicate the actual magnitude of changes in serotonin levels. However, our results do show that these changes are significant.
GOA-1 and EGL-30 signaling may affect serotonin levels in all serotonergic neurons, or this phenomenon could be HSN specific. Thus, we analyzed the effects of G-protein mutations on serotonin levels in two other pairs of C. elegans serotonergic neurons, the ADFs and NSMs. In the ADFs, the G proteins affected serotonin levels and tph-1 transcription as observed in the HSNs. Antiserotonin staining increased 1.3-fold in the goa-1 mutant and decreased 1.6-fold in the egl-30 mutant (Figure 6F), and analogous changes in tph-1 reporter transgene expression were also seen in the ADFs (data not shown). However, the G proteins caused different effects in the NSM neurons compared to those seen in the HSNs. Antiserotonin staining decreased 1.4- to 1.6-fold in the NSMs of both the goa-1 and the egl-30 mutants (data not shown). These results indicate that there is cell-type specificity in G-protein regulation of tph-1 expression, demonstrating that this phenomenon must be studied in specific identified neurons.
We determined whether GOA-1 and EGL-30 act cell autonomously to have their opposing effects on serotonin levels in the HSNs. The chromosomally integrated transgenes used to manipulate GOA-1 and EGL-30 function in the HSNs were crossed into the goa-1 and egl-30 mutants, respectively, and we quantitated antiserotonin stain intensity in these strains. Compared to the goa-1 mutant with no transgene, expression of GOA-1Q205L in the HSNs of the goa-1 mutant caused a 1.6-fold reduction in HSN serotonin staining (Figure 6G). EGL-30Q205L expression in the HSNs of the egl-30 mutant caused a 1.5-fold increase in HSN serotonin staining compared to the nontransgenic egl-30 mutant (Figure 6H). These results show that GOA-1 and EGL-30 act cell autonomously to regulate serotonin levels in the HSNs.
Small changes in tph-1 expression levels significantly alter egg-laying behavior:
Thus far we have established that GOA-1 and EGL-30 function antagonistically in the HSNs to regulate transcription from the tph-1 promoter, which affects serotonin biosynthesis in the HSNs. Our results suggest a model in which changes in serotonin levels in the HSNs lead to changes in the amount of serotonin released from the HSNs, which could account for effects of GOA-1 and EGL-30 signaling on egg-laying behavior (Figure 7A). To determine if modest changes in tph-1 expression actually affect egg laying, we analyzed the egg-laying behavior of animals carrying zero, one, two, or three functional copies of the tph-1 gene. Each wild-type chromosome II carries one functional copy of tph-1, while a chromosome II with a null mutation in tph-1 carries zero functional copies. A free duplication of a portion of chromosome II containing tph-1 was used to add an extra functional copy of the gene. We performed genetic crosses to isolate animals carrying different combinations of these chromosomes. Animals assayed for the dosage experiments were precisely genotypically matched to the controls to which they are compared, such that the only difference was in the number of functional copies of tph-1. Altering the dosage of tph-1 in this way allowed us to generate animals that should have ∼50% changes in tph-1 expression.
We found that these small changes in tph-1 cause significant changes in egg-laying behavior. Animals carrying one functional copy of tph-1 retained significantly more eggs in utero (18.9 ± 0.4 eggs) than control animals that carried two functional copies of tph-1 (16.1 ± 0.4 eggs), while complete loss of tph-1 (zero functional copies) led to the retention of many unlaid eggs (28.7 ± 1.0 eggs; Figure 7B). Increasing the number of functional copies of tph-1 to three led to the retention of significantly fewer eggs (17.5 ± 0.5 eggs) than the respective control with only two functional copies (20.5 ± 0.6 eggs; Figure 7C). Since even small changes in tph-1 expression are sufficient to cause significant changes in egg laying, the changes in tph-1 expression caused by GOA-1 and EGL-30 signaling appear to be a physiologically significant mechanism for altering egg-laying behavior.
Since tph-1 is expressed in multiple neurons, we tested whether TPH-1 acts in the HSNs to regulate egg-laying behavior. We determined in which cells TPH-1 signaling was sufficient to rescue the egg-laying-deficient behavior of the tph-1 mutant by generating extrachromosomal transgenes to express the tph-1 cDNA in different serotonergic neurons. Expression of the tph-1 cDNA plus GFP from the HSN promoter caused the tph-1 null mutant to retain significantly fewer eggs (22.7 ± 1.1 eggs) than the tph-1 null mutant expressing only GFP from the HSN promoter (29.5 ± 1.2 eggs; Figure 7D). It is likely that only partial rescue of the tph-1 egg-laying defect was observed because of the mosaic expression associated with extrachromosomal transgenes. Since the HSN promoter also results in expression in the serotonergic NSM neurons, we expressed the tph-1 cDNA from a control NSM-specific promoter (Moresco and Koelle 2004) and did not see rescue of the tph-1 null mutant (29.4 ± 1.2 eggs; Figure 7D). This indicates that the rescue effects observed with the HSN promoter are the result of tph-1 expression in the HSNs, not in the NSMs.
To determine the extent to which regulation of tph-1 transcription in the HSNs accounts for G-protein regulation of egg laying, we crossed together the tph-1 null mutation with the chromosomally integrated transgenes that express PTX or EGL-30Q205L in the HSNs. Transgenic expression of PTX in the HSNs of the tph-1 null mutant produced animals that retained significantly fewer eggs (6.7 ± 0.5 eggs) than the tph-1 null mutant with no transgene (30.6 ± 1.2 eggs; Figure 7E). Expression of EGL-30Q205L in the HSNs of the tph-1 mutant also led to the retention of significantly fewer unlaid eggs (17.0 ± 1.4 eggs) compared to the nontransgenic tph-1 control (Figure 7E). Since inactivation of GOA-1 and increased EGL-30 signaling still have effects in the absence of tph-1 function, GOA-1 and EGL-30 signaling must have additional consequences beyond regulation of tph-1 transcription. The HSNs apparently release other neurotransmitters in addition to serotonin to stimulate egg laying, since animals lacking HSNs (Trent et al. 1983) have a more severe egg-laying defect than do tph-1 mutants (Sze et al. 2000). Thus, the G proteins may regulate synthesis and/or release of these other neurotransmitters to have their additional affects on egg laying.
We used C. elegans egg-laying behavior to investigate the basic mechanisms that modulate serotonin signaling. Our results show that antagonistic signaling through Gαo and Gαq in the HSN motor neurons alters transcription of the tryptophan hydroxylase gene tph-1, altering serotonin biosynthesis to change serotonin content in the HSNs. We showed that even small changes in tph-1 expression levels significantly alter egg-laying behavior, demonstrating that regulation of tph-1 is a significant (albeit not the only) mechanism by which Gαo and Gαq control behavior.
Serotonin signaling is regulated at the level of serotonin biosynthesis:
How does altering serotonin content alter the ability of the HSNs to stimulate the postsynaptic egg-laying muscles? We suggest that increased cytosolic serotonin leads to increased loading into synaptic vesicles and thus to greater release of serotonin and greater postsynaptic response. The amount of neurotransmitter released by a single depolarization event of a presynaptic neuron dictates the magnitude of postsynaptic response, as postsynaptic and extrasynaptic receptors are not saturated by such events (Bunin and Wightman 1998; Liu et al. 1999; Ishikawa et al. 2002). Regulating the content of synaptic vesicles is a potential mechanism for regulating neurotransmitter release, as the amount of neurotransmitter loaded into a single vesicle is not invariant (Pothos et al. 1998; Gong et al. 2003; Wilson et al. 2005). The amount of neurotransmitter inside synaptic vesicles depends on the cytosolic concentration of neurotransmitter and the number and activity of vesicular transporters (Pothos et al. 1998; Wilson et al. 2005). For example, rats treated with 5-hydroxytryptophan, the serotonin precursor synthesized by tryptophan hydroxylase, exhibited an increase in calcium-dependent serotonin release in the hypothalamus (Gartside et al. 1992). Thus, an increase in presynaptic serotonin levels increases serotonin release. Our results suggest that this is a normal physiological mechanism used to modulate serotonin signaling.
Tryptophan hydroxylase controls serotonin signaling in C. elegans and humans:
Regulation of tryptophan hydroxylase expression may generally be used to modulate serotonin signaling. We found that Gαo and Gαq antagonistically regulate transcription of tph-1 in the HSN motor neurons and that the G proteins similarly affect tph-1 expression in the ADF sensory neurons. Previous studies also analyzed changes in tph-1 expression in the ADFs. Mutations in the transient receptor potential vanilloid-type (TRPV) ion channels OSM-9 and OCR-2 decrease tph-1 transcription in the ADFs (Zhang et al. 2004). In addition, regulation of TGF-β by the calcium channel UNC-2 and signaling through the DAF-2/insulin pathway control the transcription factor DAF-16 to regulate tph-1 expression in the ADF neurons, and this may also involve UNC-43/CaMKII (Estevez et al. 2004, 2006). It was not possible to assess the physiological significance of the tph-1 regulation in these studies because the function of the serotonin released from ADF was unknown. However, a recent study showed that exposure to pathogenic bacteria increases tph-1 expression and serotonin levels in the ADF neurons as part of a mechanism by which animals avoid subsequent encounters with such bacteria (Zhang et al. 2005). The relationships between G-protein control of tph-1 expression and factors that affect tph-1 expression in the ADFs, as seen in other studies, remain to be determined. The TRPV channels that affect tph-1 expression in the ADFs are not expressed in the HSNs (Jose et al. 2007). In addition, we observed differences in regulation of serotonin signaling by G proteins in comparing the HSN and NSM neurons. Thus, regulation of tph-1 is cell-type specific and must be analyzed in individual identified serotonergic neurons.
Understanding regulation of tryptophan hydroxylase expression is important because a decrease in this expression is a possible cause of depression. In humans, single nucleotide polymorphisms identified in tph-1 and tph-2 have been associated with major depression (Zill et al. 2004; Gizatullin et al. 2006). Functional polymorphisms in the promoter and 3′-UTR of tph-1 lead to significant decreases in tph-1 expression and serotonin levels in the developing mouse brain, causing animals to engage in depression-related behaviors (Nakamura et al. 2006). Antidepressant drugs may act to affect tph-1 transcription and serotonin biosynthesis, since chronic treatment of rats with the SSRI sertraline causes upregulation of tph mRNA and serotonin levels (Kim et al. 2002).
Gαo and Gαq act in the same neurons to have opposite effects on neurotransmitter release:
Beyond its focus on serotonin signaling, this study provides a greater understanding of neural G-protein signaling. Gαo is ubiquitously expressed in the nervous systems of higher organisms and mediates much of the neurotransmitter signaling in the mammalian brain (Jiang et al. 2001), but the mechanism of Gαo signaling and its ultimate effect on neurons remain unclear. Genetic studies in C. elegans show that the effects of Gαo on behavior are precisely opposed by those of another ubiquitously expressed neural G protein, Gαq. Mutations in Gαo and Gαq have opposite effects on egg laying, locomotion, pharyngeal pumping, and olfactory adaptation (Mendel et al. 1995; Ségalat et al. 1995; Brundage et al. 1996; Matsuki et al. 2006). Until this point, it was not known whether the G proteins opposed each other directly by acting in the same neurons or indirectly by acting in different neurons. This study clarifies that Gαo and Gαq oppose each other directly in the HSN neurons. Our results are consistent with previous studies that used indirect methods to infer possible sites of action of the G proteins in the egg-laying system (Trent et al. 1983; Mendel et al. 1995; Shyn et al. 2003). Gαo and Gαq may generally oppose each other by acting directly in the same neurons to have opposite effects on neurotransmitter release. This is consistent with studies of locomotion behavior, in which indirect methods suggest that Gαo and Gαq oppose each other by acting in the same ventral cord motor neurons to alter acetylcholine release (Lackner et al. 1999; Nurrish et al. 1999; Miller et al. 1999; Chase et al. 2004).
What is the biological purpose of the antagonism between Gαo and Gαq? Analysis of the egg-laying system provides some insight. Egg laying is under complex control, as it occurs in timed clusters and is regulated by the presence of food and the availability of eggs (Waggoner et al. 1998; Dong et al. 2000; Jose et al. 2007). Many neurotransmitters appear to signal through receptors on the HSNs to control egg laying (Bany et al. 2003; Shyn et al. 2003; Moresco and Koelle 2004; Jose et al. 2007). A model thus emerges in which Gαo and Gαq integrate information from multiple signals to set levels of neurotransmitter release from the HSNs and thus set levels of egg laying appropriate to the complex set of conditions that a worm may experience.
Gαo and Gαq affect both transcriptional and nontranscriptional targets:
We have for the first time determined that antagonistic Gαo/Gαq signaling affects behavior by modulating transcription. The mechanism by which G-protein signaling alters transcription of tph-1 remains to be determined. A transcription factor required for tph-1 expression in the HSNs, UNC-86, has been identified (Sze et al. 2002) and is thus a possible target of Gαo/Gαq signaling. The regulation might also occur through the CREB transcription factor, since Gαo/Gαq signaling can control expression of an artificial reporter gene regulated by CREB (Suo et al. 2006). Regardless of the specific mechanism, our results show that Gαo/Gαq regulate egg laying in part through control of tph-1 transcription and suggest that the G proteins may more widely regulate neural gene expression to have their antagonistic effects on neural function.
Our epistasis results show that regulation of tph-1 is not the only mechanism by which Gαo/Gαq signaling affects egg laying. In addition to effects on transcription, Gαo/Gαq signaling must also have fast-acting, nontranscriptional effects on neural activity. Application of exogenous serotonin or dopamine to worms causes paralysis within minutes as a result of rapid signaling through Gαo (Ségalat et al. 1995; Chase et al. 2004). These rapid effects may be the result of altering diacylglycerol levels to relocalize UNC-13S and affect synaptic vesicle priming, as suggested by Nurrish et al. (1999), although we failed to see evidence for such an effect on the HSN neurons. The functional significance of diacylglycerol-induced UNC-13S relocalization in Gαo/Gαq signaling remains to be assessed, since the lethality of unc-13 mutations has prevented them from being used in genetic epistasis experiments. However, other experiments show that mechanisms that alter diacylglycerol levels cannot account for all, or perhaps even most, of the effects of Gαo/Gαq signaling (Bastiani et al. 2003; Jose and Koelle 2005). Regulation of tph-1 expression thus constitutes the downstream effect of Gαo/Gαq signaling that, at this point, has been best validated as being physiologically responsible for altering behavior.
The picture that emerges of Gαo/Gαq signaling in C. elegans parallels results from studies of G-protein signaling in Aplysia learning, which suggest that signaling through the G protein Gαs alters neural function through both fast-acting mechanisms and slower changes in transcription of neuronal genes (Kandel 2001). The HSN neurons in C. elegans serve as a model in which both the transcriptional and the nontranscriptional consequences of neural G-protein signaling can be studied.
We thank the Caenorhabditis Genetics Center and Iva Greenwald for strains; Adam Hartley for help with image analysis; and Tony Koleske, Thomas Biederer, and members of the Koelle lab for critical reading of the manuscript. This work was supported by National Institutes of Health grant NS036918 and a National Science Foundation graduate fellowship (to J.E.T.).
- Received August 3, 2007.
- Accepted October 26, 2007.
- Copyright © 2008 by the Genetics Society of America