Caenorhabditis elegans TRPV ion channel regulates 5HT biosynthesis in chemosensory neurons

The monoamine 5HT acts as a neurotransmitter and a hormone to induce behavioral and endocrine responses to changes in environmental and physiologic states. A deficit in 5HT is implicated in a broad spectrum of disorders such as depression, eating disorders and type II diabetes (reviewed by Lucki, 1998; Leibowitz and Alexander, 1998; Davidson et al., 2000). In both vertebrates and invertebrates, 5HT is synthesized in a set of neurons with diverse synaptic properties and connectivities. Activation of serotonergic neurons by sensory stimuli was first clearly demonstrated in 1976 through studies in Aplysia, where sensory stimuli increase 5HT signals to induce synaptic facilitation and behavioral sensitization(Brunelli et al., 1976). Long-standing questions are: how do serotonergic neurons integrate sensory signals into 5HT neurotransmission, and are different serotonergic neurons regulated by cell-specific molecular mechanisms?

TRP-related proteins are a superfamily of cation channels that share structural homology to the Drosophila transient receptor potential(TRP) protein (reviewed by Clapham et al.,2001; Montell et al.,2002). TRP channels act as molecular integrators of a wide range of chemical and physical stimuli to regulate behavior and physiological function in both vertebrates and invertebrates (reviewed by Scott and Zuker,1998; Minke and Cook, 2002). Five genes in the C. elegans genome, ocr-1, ocr-2, ocr-3,ocr-4 and osm-9, encode TRPV subfamily proteins characterized by cytoplasmic N-terminal multiple ankyrin repeats, six transmembrane segments and a non-conserved cytoplasmic C terminus(Harteneck et al., 2000; Tobin et al., 2002). These TRPV genes are expressed in the sensory endings of chemosensory neurons and have been shown to regulate sensory functions and social behavior(Colbert et al., 1997; Tobin et al., 2002; de Bono et al., 2002). In mammals, TRPV channels transmit nociceptive stimuli such as pain(Tominaga et al., 1998) and noxious heat (Caterina et al.,1997; Caterina et al.,1999). In addition, mammalian TRPV ion channels also mediate the response to growth factors (Kanzaki et al., 1999). Agonists to TRPV channels induce hypothermia(Meller et al., 1992; Szallasi and Blumberg, 1996)and modulate oxygen consumption (Colquhoun et al., 1995). Mammalian TRPV proteins are expressed in the sensory neurons, as well as in the CNS(Tominaga et al., 1998; Hayes et al., 2000; Mezey et al., 2000; Delany et al., 2001). It has been postulated that TRPV ion channels regulate the release of neural mediators to modulate endocrine activity(Szallasi and Blumberg, 1996),but in vivo evidence has not yet been reported.

We describe the effect of mutations in two TRPV ion channel genes, osm-9 and ocr-2, on biosynthesis of 5HT in C. elegans. Previously, we demonstrated that the tph-1 gene, which encodes the key 5HT biosynthesis enzyme tryptophan hydroxylase, is essential for 5HT biosynthesis and that tph-1 expression is regulated by cell-specific mechanisms (Sze et al.,2000; Sze et al.,2002). In this study we show that a signaling pathway involving the osm-9 and ocr-2 TRPV channel proteins, CaMKII specifically regulates tph-1 expression in the serotonergic chemosensory neurons ADF. Results from this study reveal a remarkably neuron-specific mechanism regulating 5HT biosynthesis and provide genetic insights into how a neuron transduces the events at the cell-surface to 5HT signaling.

The strains used in this study were wild-type C. elegans Bristol strain (N2) and mutants osm-9(yz6), osm-9(n2743), osm-9(ky10),osm-9(n1516), ocr-2(yz5), ocr-2(ak47), odr-10(ky225), osm-11(n1604),tph-1(mg280), daf-7(e1372), daf-16(mgDf50), nss-1(yz12), odr-3(n2150);gpa-3(pk35); gpa-13(pk330); unc-43(n498)gf and unc-43(e408). The worms were fed with E. coli OP50 as the food.

The yz6 and yz5 mutations were isolated based on reduction/absence of GFP in the ADF neurons after ethylmethane sulfonate mutagenesis of wild-type animals carrying an integrated tph-1::gfptransgene. The mutagenesis and mutant screens have been described previously(Sze et al., 2002). We screened about 6500 haploid genomes and isolated 24 mutants. None of the mutations completely eliminates tph-1::gfp expression in the ADF neurons, but yz12 (Sze et al.,2002), yz5, yz6 and other three mutants showed stronger effects. Genetic mapping and complementation analysis indicated all these six strong mutants as single alleles, thus the screen is probably unsaturated. We mapped yz6 between the polymorphisms in the clones C09G12 and M02B7,and yz5 between C49H3 and C01F6. A PCR fragment of the cosmid M57 containing 2.8 kb of the upstream sequence, exons/introns and 1.3 kb downstream sequence of osm-9 restores tph-1::gfp expression in ADF of yz6 animals. A PCR fragment of the cosmid T09A12 containing 2.5 kb upstream sequence, exons/introns and 1.8 kb downstream sequence of ocr-2 restores tph-1::gfp expression in yz5mutants. The molecular lesion of the mutations was determined by PCR amplification of the exons and exon/intron boundaries from the mutant strains and sequencing the PCR fragments.

The osm-9 and ocr-2 genes are co-expressed in six pairs of neurons: ADF, AWA, ADL, ASH, PHA and PHB(Colbert et al., 1997; Tobin et al., 2002). Chimeric constructs were generated by expressing wild-type osm-9- or ocr-2-coding regions under the control of heterologous promoters that are expressed in a subset of these six pairs of the neurons. The expression pattern of the heterologous promoters overlaps with osm-9 and ocr-2 in the following neurons: odr-7, AWA(Sengupta et al., 1994); osm-10, ASH, PHA and PHB (Hart et al., 1999); tax-2, PHA and PHB(Coburn and Bargmann, 1996); cat-1, ADF (Sze et al.,2002); lin-11, ADF, ADL(Hobert et al., 1998); tph-1(BC), ADF (this work). In each case, we first constructed a fusion of the promoter region to the GFP and unc-54 3′-uncoding sequences in the plasmid pPD97.75 (A. Fire) to confirm the expression pattern,then, we replaced the GFP sequence with a genomic sequence encompassing the entire intron and exon regions of osm-9 or ocr-2. Individual constructs were introduced into yz6 or yz5 mutants carrying the integrated tph-1::gfp reporter. The tph-1(BC) construct was generated by inserting the sequence from –132 to –377 upstream of the tph-1 translational start [the BC region as described previously (Sze et al., 2002)]to a minimal promoter of the pes-10 gene in the GFP vector pPD122.53(A. Fire). The plasmid pRF4 containing the dominant Rol-6 gene was co-injected as a transgenic marker for Promoter::gfp constructs, and a plasmid containing elt-2::gfp (a gift from J. McGhee) as the marker for Promoter::osm-9 and Promoter::ocr-2 constructs.

The expression pattern of these GFP transgenes in wild-type animals has been published: the tph-1::gfp, lin-11::gfp and cat-1::gfptransgenes were integrated into the chromosomes, and gtpch-1::gfp was carried as extra chromosomal arrays(Hobert et al., 1998; Sze et al., 2000; Sze et al., 2002). The individual transgenes were crossed into mutants. Thus, the expression of the same transgene in wild-type and mutant animals was compared.

To quantify GFP intensity of tph-1::gfp in the ADF neurons, the images were captured with a Zeiss AxioCam digital camera at a fixed exposure time, and the fluorescence within a 25×25 pixel area of the cell body was scored, hence the same dimension of the ADF neurons in different genetic background was compared.

The staining of anti-5HT antibody was performed using the McIntire-Horvitz whole-mount procedure with modifications as described(McIntire et al., 1992; Sze et al., 2000).

Feeding and egg-laying assays were conducted with young adult animals. Well-fed larval stage 4 animals (L4) were picked onto fresh plates seeded with bacteria as the food, and allowed to develop ∼20 hours at 20°C. Feeding behavior was assayed by measuring the rate of pharyngeal pumping,which was scored by counting pharynx terminal bulb contractions(Duerr et al., 1999; Sze et al., 2000). Egg-laying behavior was scored by counting the number of fertilized eggs accumulated inside of the uterus of adults, using DIC optics(Sze et al., 2000).

For Dauer assays, 10 young adult animals of a strain were transferred onto a fresh plate and allowed to lay eggs for overnight. The parents were then removed, progeny were allowed to develop at 15°C, and the number of Dauers and L4/adults was scored 5-6 days later. Notice a higher Dauer frequency of tph-1;daf-7 double mutant animals than we previously reported(Sze et al., 2000). This difference is due to different time points at which we scored Dauers. tph-1;daf-7 mutants grow slower, some of the animals were still at pre-Dauer stages at the earlier time point.

Of the 302 neurons present in an adult C. elegans hermaphrodite,nine neurons from five distinct classes are detected by antibodies raised against 5HT (Horvitz et al.,1982) (Fig. 1A);four classes exist as left-right symmetric pairs: the ADF chemosensory neurons, the NSM pharyngeal secretory neurons, the HSN motor neurons, and the AIM interneurons and RIH is a single interneuron. These neurons are generated from different lineages during embryogenesis(Sulston et al., 1983). 5HT immunoreactivity can be detected in the ADF, NSM, AIM and RIH neurons shortly after hatching, and in the HSN neurons only in adults. The ability to monitor identified serotonergic neurons permits the isolation and analysis of mutant genes affecting serotonergic phenotype in specific neurons.

Biosynthesis of 5HT in C. elegans requires the tph-1gene, which encodes the enzyme tryptophan hydroxylase catalyzing the rate-limiting first step of 5HT biosynthesis. tph-1 is expressed in serotonergic neurons, and tph-1 knockout animals have no detectable 5HT (Sze et al., 2000). Our previous study indicated that tph-1 expression in different serotonergic neurons is regulated by distinct transcription programs(Sze et al., 2002). To define genes underlying this neuron-specific regulation of 5HT synthesis, we conducted a genetic screen for neuron-specific serotonin defective(nss) mutants, using a green fluorescent protein (GFP) fusion to tph-1 (tph-1::gfp) as a reporter.

yz5 and yz6 are two of the nss mutations that specifically downregulate tph-1 expression in the ADF neurons. In wild-type animals, tph-1::gfp is highly expressed in the ADF, NSM and HSN neurons, but the GFP level in the ADF neurons is dramatically reduced or undetectable in yz5 and yz6 mutants(Fig. 1B, Table 1). Consistent with the essential role of tph-1 in 5HT biosynthesis, staining of the mutant animals with anti-5HT antibody shows reduced/absence of 5HT immunoreactivity in the ADF neurons (Fig. 1C). However, neither the mutation nor yz5;yz6 double mutation has a detectable effect on tph-1::gfp expression or 5HT immunoreactivity in other serotonergic neurons (Table 1; Fig. 1B,C). The ADF neurons are the only serotonergic sensory neurons in hermaphroditic C. elegans, the data suggest that yz5 and yz6 specifically regulate the serotonergic phenotype of sensory neurons.

Genetic mapping and transgene rescue of yz5 and yz6mutations revealed two TRPV channel proteins regulating tph-1expression in serotonergic chemosensory neurons. The TRPV subfamily is characterized by the cytoplasmic N-terminal multiple ankyrin repeats, six transmembrane segments and a cytoplasmic C terminus(Harteneck et al., 2000)(Fig. 1D). Our genetic mapping and sequencing of the mutant genomic DNA show that yz6 is a nonsense mutation that results in a stop codon before the transmembrane domain in the osm-9 TRPV gene, and yz5 is a missense mutation adjacent to the conserved ankyrin motifs in ocr-2(Fig. 1D). To confirm it is the mutation in the TRPV genes that downregulates tph-1 expression in the ADF neurons, we examined tph-1::gfp expression and 5HT immunoreactivity in the ocr-2 deletion mutant ak47 and three osm-9 alleles (ky10, n2743, n1516). In every of the mutant strains, tph-1::gfp expression in ADF is downregulated and 5HT immunoreactivity in ADF is reduced (Table 1; data not shown). Furthermore, yz6 and yz5mutant animals carrying a transgene containing the wild-type osm-9 or ocr-2 gene, respectively, restore tph-1::gfp expression and 5HT immunoreactivity (Fig. 1B,C; Table 1). We conclude that yz6 is an allele of the osm-9 gene, and yz5 is an allele of ocr-2.

Both osm-9 and ocr-2 are expressed in the ADF neurons(Colbert et al., 1997; Tobin et al., 2002), but not in other serotonergic neurons (Fig. 2A). Beside ADF, osm-9 and ocr-2 also are co-expressed in five pairs of non-serotonergic chemosensory neurons: the AWA,ADL, ASH neurons in the head, and the PHA and PHB neurons in the tail(Tobin et al., 2002). These head neurons, as well as ADF, are component neurons of the amphid sensory organ, and each class senses distinct signals(Bargmann and Horvitz, 1991a). The cell bodies of these head neurons are clustered together, and their processes run parallel and are interconnected directly or indirectly en passant (White et al., 1986)(Fig. 1A). The TRPV channels could act in the ADF neurons to control tph-1 expression, or they could function in the other chemosensory neurons that regulate ADF neural activity. The AWA neurons detect the attractive odorant diacetyl, and the ASH and ADL neurons sense aversive signals; ocr-2(ak47) and osm-9 mutant animals are defective in sensing these sensory signals(Colbert et al., 1997; Tobin et al., 2002). We tested whether disruption of these sensory signaling is a cause of reducing tph-1::gfp expression. No GFP reduction was observed in animals with defective diacetyl receptor (odr-10)(Sengupta et al., 1996) or with defective ASH, ADL (osm-11) function(Table 1). Thus, osm-9/ocr-2 function in these sensory signaling is not required to activate tph-1 expression in the ADF neurons.

To identify cells in which the TRPV proteins function to upregulate tph-1 expression, we constructed a series of chimeric genes with the osm-9- or ocr-2-coding regions under the control of heterologous promoters and generated transgenic animals expressing wild-type osm-9 or ocr-2 proteins in a subset of these six neuronal types. When osm-9(yz6) and ocr-2(yz5) mutant animals carry a transgene expressed in the ADF neurons, tph-1::gfp expression in the ADF neurons is restored, whereas the mutant animals carrying the transgene not expressed in the ADF neurons exhibit reduced tph-1::gfp expression similar to their non-transgenic siblings(Table 2). These observations collectively argue that the osm-9 and ocr-2 genes act cell-autonomously in the ADF neurons to control the tph-1expression.

Mutations in OSM-9 and OCR-2 channels may disrupt the regulatory pathway of tph-1 transcription, or they may induce ADF neural degeneration due to ion imbalances. To address these possibilities, we assessed if the yz6 and yz5 mutations result in a general morphological transformation of the ADF neurons. The LIM-homeodomain transcription factor lin-11 is co-expressed with osm-9 and ocr-2 in the ADF and ADL chemosensory neurons (Freyd et al., 1990; Hobert et al.,1998). We crossed an integrated lin-11::gfp transgene into osm-9(yz6) and ocr-2(yz5) mutants; no reduction of lin-11::gfp expression levels can be detected in the mutant backgrounds (Fig. 2B). Judged by fluorescence microscopy, the morphology of the cell body and the processes of the ADF neurons are indistinguishable in the mutant and wild-type animals at any developmental stage. Thus, these mutations do not cause the ADF neurons to die prematurely. Rather, these results point to a signaling pathway downstream of the OSM-9 and OCR-2 TRPV channels regulating the transcription of tph-1 in the ADF neurons.

Does this TRPV channel signaling specifically regulate tph-1expression, or does it regulate the expression of all genes involved in 5HT synthesis and neurotransmission? We have explored this question by examining GFP reporters of marker genes. In each case, the GFP reporter construct was first introduced into wild-type animals, and the resulting transgene was then crossed into mutants. Thus, the expression of the same transgene in different genetic backgrounds was compared. CAT-1/vesicular monoamine transporter is required for 5HT neurotransmission (Duerr et al., 1999; Nurish et al., 1999). cat-1::gfp is a functional fusion of GFP to the entire protein coding segment of the cat-1 gene and is localized to the synapses of the serotonergic neurons (Sze et al., 2002). The gene F32G8.6 encodes a probable GTP-cyclohydrolase I (gtpch-1), a co-factor of tryptophan hydroxylase for 5HT biosynthesis, and a gtpch-1::gfp fusion gene also is expressed in the serotonergic neurons (Sze et al., 2002). Unlike the dramatic reduction of tph-1::gfp in the ADF neurons, there is no significant reduction of cat-1::gfp or gtpch-1::gfp in yz6 and yz5 mutant backgrounds(Fig. 2C,D). These observations are consistent with our previous results that the expression of tph-1,cat-1 and gtpch-1 is differentially regulated in serotonergic neurons (Sze et al., 2002). These data demonstrate a great specificity of the TRPV channel signaling within the ADF neurons and suggest the transcriptional regulation of the tph-1 gene as a major target.

Analysis of the tph-1 promoter has revealed a discrete cis-regulatory region essential for tph-1 expression in the ADF neurons (Sze et al., 2002). We investigated whether the TRPV ion channel signaling acts through this neuron-specific transcriptional regulatory mechanism. A GFP reporter under the control of the sequence –132 bp to –377 bp of tph-1 and a minimal promoter from the pes-10 gene is specifically expressed in the ADF neurons of wild-type animals, but the ADF GFP intensity is significantly reduced in yz6 mutant background(Fig. 3A). Thus, the 246 bp cis-regulatory region is sufficient to activate tph-1 expression in the ADF neurons, and signaling from the OSM-9 and OCR-2 ion channels regulates the activity of this neuron-specific transcriptional regulatory program.

Unlike mutations in the POU-transcription factor UNC-86 that completely abolish tph-1::gfp expression in the NSM and HSN neurons(Sze et al., 2002), none of the mutations in osm-9 or ocr-2, nor the mutation of both eliminates tph-1::gfp expression in the ADF neurons(Table 1; Fig. 3B), indicating that the TRPV activity modulates tph-1 expression levels but is not essential for the transcription. In yz6 background, the GFP reporter under the control of the 246-bp tph-1 cis-regulatory sequence fused to a pes-10 minimal promoter is expressed in the ADF neurons at slightly higher levels than the tph-1::gfp reporter under the control of the 3.1 kb of the tph-1 promoter (Fig. 1B, Fig. 3A). This could reflect the difference in the basal expression level of the tph-1 and pes-10 promoter in the reporter constructs. Alternatively, it could be an indication of additional cis-regulatory elements mediating inhibition of tph-1 expression but the elements are not present in the 246 bp tph-1 sequence.

The odr-3 Gα protein is essential for osm-9 and ocr-2 function in olfactory, osmosensory and mechanosensory behaviors mediated by the AWA and ASH neurons(Roayaie et al., 1998; Colbert et al., 1997; Tobin et al., 2002). It has been proposed that ODR-3 activity modulates the outputs of the TRPV channel signaling (Roayaie et al.,1998). odr-3 also is expressed in the ADF neurons;however, the odr-3(n2150) deletion mutation does not cause a significant reduction of tph-1::gfp expression(Fig. 3B). Thus, OSM-9 and OCR-2 can activate tph-1 expression in the absence of odr-3activity. Beside odr-3, the ADF neurons express two other Gαproteins, gpa-3 and gpa-13(Jansen et al., 1999). tph-1::gfp expression is unaffected in gpa-3 or gpa-13 deletion mutants (Fig. 3B). However, animals bearing a double mutation of TRPV and Gα still exhibit reduced tph-1::gfp expression in the ADF neurons similar to the TRPV mutants. These data indicate that unlike the role of ODR-3 in the sensory behaviors, the Gα proteins do not play an essential role in OSM-9/OCR-2-dependent regulation of tph-1expression. Because the sensory behaviors and 5HT production are mediated by different neurons, our results suggest that OSM-9 and OCR-2 TRPV channels in different neurons may be regulated by different mechanisms.

CaMKII is a critical mediator of Ca²⁺ signaling. The unc-43 gene encodes the only C. elegans CaMKII(Reiner et al., 1999; Rongo and Kaplan, 1999). An unc-43 loss-of-function mutation causes a twofold reduction of tph-1::gfp expression in the ADF neurons(Fig. 3B), but has no effect on other serotonergic neurons (not shown). Conversely, the unc-43(n498)gain-of-function mutation partially blocks the downregulation of tph-1 expression in yz6(Fig. 3b) and yz5mutants (data not shown). One simple model to explain these data would be that activation of OSM-9 and OCR-2 channels increases ADF intracellular Ca²⁺ which stimulates UNC-43 to induce tph-1transcription, whereas the unc43(n498) mutation, which causes constitutive Ca²⁺-independent activity(Reiner et al., 1999),bypasses the need of the channel function. However, these results do not exclude the possibility that UNC-43 acts less directly in the TRPV channel signaling pathway. These results suggest that CaMKII acts downstream of or in parallel with the OSM-9 and OCR-2 TRPV channels to control 5HT production in the ADF chemosensory neurons.

We find that both osm-9 and ocr-2 mutants show developmental defects reminiscent of tph-1 deletion mutants(Fig. 4A). The DAF-2/insulin receptor and DAF-7/TGFβ signaling act in parallel to control whether an animal enters the reproductive lifecycle or developmentally arrests at the metabolically inactive Dauer larval stage. Disruption of either pathway causes conditional abnormal arrest at the Dauer stage, but disruption of both the pathways causes constitutive Dauer arrest(Ogg et al., 1997) (reviewed by Riddle, 1997). Similar to the tph-1 deletion mutation, the osm-9(yz6); osm-9(ky10) and ocr-2(yz5) mutations enhance the Dauer phenotype of daf-7(e1372) mutants(Fig. 4A). At the 15°C growth temperature, about 10% of daf-7 mutant animals arrest as Dauers, but more than 70% form Dauers when daf-7 mutants carry a mutation in osm-9 or ocr-2. Ninety-eight percent of daf-7;tph-1 double mutant animals form Dauers, whereas 10-15%tph-1 mutants form Dauers (Sze et al., 2000). None of the TRPV mutants on their own formed Dauers when assayed under the same condition, although they grow slower than wild-type animals. Because e1372 is a daf-7-null mutation,this enhanced Dauer phenotype of the double mutants implies that the TRPV-mutations affect a pathway parallel to daf-7.

Enhancement and suppression genetics and other molecular experiments implicate that the TRPV mutations affect 5HT inputs to the insulin pathway. The DAF-16/forkhead transcription factor is a negative target of DAF-2/insulin signaling, and the daf-16(mgDf50) mutation bypasses the need of DAF-2(Ogg et al., 1997). The Dauer phenotype of daf-7;tph-1 and daf-7;osm-9 can be suppressed by the daf-16(mgDf50) mutation(Fig. 4A), indicating a reduction of DAF-2/insulin signaling in the double mutants that promotes the Dauer formation. Dauer arrest is also enhanced in daf-7(e1372)mutants carrying a mutation in the nss-1 gene, which also specifically affects tph-1 expression in ADF(Sze et al., 2002). This raises the possibility that it is the reduction of ADF 5HT signals that downregulates the DAF-2/insulin pathway. However, expression of the wild-type osm-9-coding sequence under a lin-11 promoter only mildly suppresses the Dauer phenotype of osm-9(yz6);daf-7(e1372) mutants(Fig. 4A), indicating that other osm-9-expressing cells also contribute to the normal development.

osm-9 and ocr-2 mutant animals do not display every deficit observed in mutants with all the serotonergic neurons defective. 5HT regulates several C. elegans behaviors. For example, applying exogenous 5HT to C. elegans stimulates pharyngeal pumping and egg-laying (Avery and Horvitz,1990; Weinshenker et al.,1995), whereas 5HT-deficient mutants tph-1 and cat-1 exhibit a slower pumping rate and accumulate a large number of fertilized eggs in the uterus (Duerr et al., 1999; Sze et al.,2000). However, osm-9 and ocr-2 mutant animals do not accumulate excess eggs in the uterus and their pharyngeal pumping rates are equivalent to wild-type animals (Fig. 4B,C). Thus, 5HT signals from the other neurons are sufficient for these behaviors. But, we cannot exclude subtle behavioral changes that are difficult to detect visually.

The genetic evidence presented here has two major implications. The phenotype of the TRPV mutants represents exquisite specificity in the control of 5HT production, as exemplified by our demonstration that mutations of the osm-9 and ocr-2 TRPV genes specifically downregulate 5HT biosynthesis in the ADF neurons, and this TRPV ion channel regulation of 5HT production is mediated by a neuron-specific transcription program. Our finding that the osm-9 and ocr-2 TRPV channel genes act in the ADF neurons, function upstream of CaMKII to control the key 5HT biosynthesis gene tph-1 provides insights in elucidating the genetic pathway by which a serotonergic neuron couples the activity at the cell surface and 5HT signaling.

It has been demonstrated in many experimental systems that sensory stimuli induce 5HT signals to produce changes in behavior and physiology (e.g. Barzilai et al., 1989; Boadle-Biber, 1993; Milner et al., 1998). Until this study, no endogenous membrane protein has been shown to act in a serotonergic neuron to regulate 5HT signaling. One important finding from this study is the pronounced effect of osm-9 and ocr-2 mutations on the expression of the 5HT synthesis gene tph-1(Fig. 1). This indicates that the production of 5HT is a site where sensory information is integrated to 5HT signaling. 5HT can be released by controlled exocytosis at the synapses as well as via paracrine `volume transmission', and even during the controlled exocytosis it is newly synthesized 5HT preferentially released to induce changes in the postsynaptic targets(Attwell et al., 1993; Sanders-Bush, 1982). Hence,the level of 5HT production is one mechanism controlling both forms of 5HT neurotransmission.

Transcriptional regulation may represent a general principle of regulation of hormones and neuromodulators. For example, C. elegans' favorite growth environment upregulates the expression of the daf-7/TGFβand daf-28/insulin genes to induce C. elegans proceeding reproductive development (Schackwitz et al., 1998; Li et al.,2003); in rats, noxious sound stimuli may alter the transcription of their tryptophan hydroxylase gene in a neuron-specific manner (reviewed by Boadle-Biber, 1993); and the expression of tyrosine hydroxylase can be modulated by hormones in mice(Kumer and Vrana, 1996). Transcriptional regulation of these signaling molecules is likely a mechanism to exert a relatively slow but profound effect in the signaling pathways.

Our genetic results indicate that the osm-9 and ocr-2channel proteins interact with different signaling transduction pathways to induce different behavioral outputs. osm-9 and ocr-2 are co-expressed in four pairs of the amphid sensory neurons, and are required for a normal response to attractive and aversive odorants mediated respectively by AWA, and ASH and ADL, as well as for ASH-mediated mechanosensory and osmosensory function (Colbert et al.,1997). There are genetic evidences indicating that osm-9and ocr-2 function in these sensory behaviors requires the odr-3 Gα protein (Roayaie et al., 1998). However, tph-1 expression is unaffected by mutations in odr-3 or in other two Gα protein expressed in the ADF neurons (Fig. 3B). Although the exact mode of activation of OSM-9 and OCR-2 in any neuron has not yet been defined, our data indicate that the activity of the channels in the ADF neurons is regulated by different signaling molecules. It is interesting to note that osm-9 and ocr-2 also act in the ASH and ADL neurons to regulate social behavior independent of odr-3 activity (de Mono et al., 2002); hence, the OSM-9 and OCR-2 can induce specific behaviors by coupling distinct signaling systems.

The reciprocal effects of the unc-43 CaMKII loss- and gain-of-function mutations on tph-1 expression indicate that the amount of Ca²⁺ modulates 5HT biosynthesis(Fig. 3B). The TRP superfamily is Ca²⁺-permeable channels, and CaMKII is known as an important mediator of Ca²⁺ signaling (reviewed by Hanson and Schulman, 1992). Our genetic study shows that the constitutively active,Ca²⁺-independent unc-43(n498) CaMKII(Reiner et al., 1999) can partially activate tph-1 expression in osm-9 deletion mutant animals (Fig. 3B). These results support the model that UNC-43 is a downstream effector of the OSM-9 and OCR-2 channels: Ca²⁺ influx through OSM-9 and OCR-2 activates UNC-43 CaMKII, which induces phosphorylation cascades to activate tph-1 expression. However, unc-43(lf) mutants still express a substantial amount of tph-1, and the unc-43(gf) mutation does not completely bypass OSM-9 activity, indicating that other OSM-9/OCR-2 downstream signaling molecules may act in parallel with UNC-43 to regulate tph-1 expression.

The effect of OSM-9 and OCR-2 TRPV ion channels in the ADF neurons is strikingly specific, given the involvement of Ca²⁺ and CaMKII. The general architecture of the ADF neurons is unaffected, we could not detect a significant change in the expression levels of ADF marker genes or genes directly involved in the serotonergic phenotype, nor we could detect an effect of unc-43(lf) or (gf) mutations on tph-1 expression in other serotonergic neurons. This specificity demonstrates that `multi-functional,widespread' signaling molecules may play a refined role in a particular native cellular setting. It is conceivable that such differential regulation of the serotonergic phenotype genes would allow the ADF neurons to adjust 5HT neurotransmission in response to multiple sensory signals. As TRPV channels are expressed in the serotonergic locus in mammals(Tominaga et al., 1998; Mezey et al., 2000), it would be interesting to determine whether there is a link between TRPV mutations and 5HT deficiency in human.

The Dauer phenotype of the osm-9 and ocr-2 mutants demonstrates a genetic link between the TRPV ion channels and endocrine signaling (Fig. 4). C. elegans Dauer/non-Dauer development reflects two alternative metabolic states controlled by sensory inputs to neuroendocrine signaling pathways. The pathways from DAF-7/TGFβ and DAF-2/insulin receptor converge to stimulate reproductive growth; harsh environmental conditions transduced by the amphid chemosensory neurons suppress the endocrine signaling to induce Dauer arrest(reviewed by Riddle, 1997). Our enhancement and suppression genetics implies that osm-9 and ocr-2 regulate the DAF-2/insulin-receptor signaling pathway(Fig. 4). The TRPV channels are probably acting in the ADF neurons to modulate endocrine activity: worms bearing defective ADF neurons tend to form Dauers(Shakir et al., 1993), and laser ablation experiments implicate ADF but not the other osm-9 and ocr-2 co-expressing chemosensory neurons in Dauer formation(Bargmann and Horvitz, 1991b). However, expression of the wild-type osm-9-coding sequence under a lin-11 promoter only partially suppresses daf-7;osm-9 Dauer phenotype, indicating that osm-9 activity in other cells also modulates Dauer phenotype. Alternatively, the lin-11 promoter may not be able to express sufficient amount of osm-9 to induce a wild-type level of ADF 5HT signals (Table 2), or ectopic expression of osm-9 in other lin-11-expressing cells may interfere with the neuroendocrine signaling cascades for normal development. In mammals, 5HT regulates insulin synthesis, release and response (Breum et al., 1995; Peschke et al.,1997). It has been proposed that a feedback regulatory loop between hypothalamus 5HT and circulating hormones such as insulin, leptin and adipose tissue-derived hormone modulates the satiety and maintains metabolic and energy homeostasis (Leibowitz and Alexander, 1998). Interestingly, mouse TRPV-like channels can be activated by insulin-like growth factors(Kanzaki et al., 1999). We propose that 5HT is one mediator of TRPV channels and endocrine activity.

We thank Drs G. Ruvkun and J. Gargus, and members of the Sze laboratory for critical reading of the manuscript; Dr C. Dempsey for statistical analysis of GFP intensities; Dr G. Jansen for advice on G-protein strains; Drs C. Bargmann, O. Hobert, J. McGhee, S. Nurrish, J. Kaplan and A. Fire for strains and constructs; Sanger Center, UK for cosmid clones; and T. Stiernagle and the Caenorhabditis Genetics Center for worm strains. This work is supported by grants from Whitehall Foundation and National Institute of Mental Health to J.Y.S.; G.B. is supported by National Institute of Health minority biomedical research support program.