We have identified the Caenorhabditis eleganshomolog of the mammalian vesicular monoamine transporters (VMATs); it is 47% identical to human VMAT1 and 49% identical to human VMAT2.C. elegans VMAT is associated with synaptic vesicles in ∼25 neurons, including all of the cells reported to contain dopamine and serotonin, plus a few others. When C. elegans VMAT is expressed in mammalian cells, it has serotonin and dopamine transport activity; norepinephrine, tyramine, octopamine, and histamine also have high affinity for the transporter. The pharmacological profile of C. elegans VMAT is closer to mammalian VMAT2 than VMAT1. The C. elegans VMAT gene iscat-1; cat-1 knock-outs are totally deficient for VMAT immunostaining and for dopamine-mediated sensory behaviors, yet they are viable and grow relatively well. Thecat-1 mutant phenotypes can be rescued by C. elegans VMAT constructs and also (at least partially) by human VMAT1 or VMAT2 transgenes. It therefore appears that the function of amine neurotransmitters can be completely dependent on their loading into synaptic vesicles.
The loading of catecholamines and other biogenic amines into synaptic vesicles (and other types of release vesicles) is mediated by specific vesicular monoamine transporters (VMATs). Amine transport requires a pH gradient and is coupled to proton antiport (for review, see Schuldiner et al., 1995). In mammals, two related transport proteins (and genes) have been identified: VMAT1 is often found in neuroendocrine cells, and VMAT2 is primarily neuronal (Erickson et al., 1992, 1996; Liu et al., 1992;Weihe et al., 1994). Recombinant VMATs have been shown to mediate the transport of dopamine, norepinephrine, epinephrine, serotonin, and histamine (VMAT2 only) in vitro, as expected from previous biochemical studies on chromaffin granules and brain synaptic vesicles (Schuldiner et al., 1995; Erickson et al., 1995, 1996). The proteins have been used as markers of particular cell types, and specific antibodies raised against these transporters have been of use in studies of vesicular localization and maturation (Liu et al., 1994;Weihe et al., 1994; Nirenberg et al., 1995).
To study the role(s) of VMAT proteins in neuronal and behavioral function, we have been using an experimental system amenable to gene knock-out and transgenic technology in which specific biogenic amines are used by identified cells involved in particular behaviors. The simple soil nematode Caenorhabditis elegans contains and uses several biogenic amines (for review, see Rand and Nonet, 1997a). Dopamine has been identified biochemically, and the technique of formaldehyde-induced fluorescence (FIF) was used to localize dopamine to particular C. elegans neurons (Sulston et al., 1975). Two behaviors, locomotion and egg laying, are transiently inhibited by exogenous dopamine (Schafer and Kenyon, 1995). Serotonin has also been identified in C. elegans neurons by FIF (Horvitz et al., 1982) and by anti-serotonin immunostaining (Desai et al., 1988;McIntire et al., 1992). Exogenous serotonin stimulates egg laying and pharyngeal pumping and inhibits locomotion and defecation (Horvitz et al., 1982; Ségalat et al., 1995). Serotonin is also required for male mating behaviors (Loer and Kenyon, 1993).
Octopamine (p-hydroxyphenylethanolamine) has been detected in C. elegans extracts (Horvitz et al., 1982), although it is not yet known which cells contain this compound. Exogenous octopamine stimulates movement and inhibits egg laying; its biological actions thus appear to antagonize those of serotonin (Horvitz et al., 1982). Other biogenic amines, such as epinephrine, norepinephrine, and histamine, have not yet been identified as putative transmitters in C. elegans.
We now report the presence of a VMAT homolog in C. elegans. We demonstrate that it is associated with synaptic vesicles of known biogenic amine-containing neurons, and that it can function as an amine transporter when expressed in mammalian cells. We also describe VMAT-deficient mutants and their phenotypes, and we show that expression of the vesicular transporter is required for proper function of identified dopamine-containing neurons. Finally, we present evidence for partial phenotypic rescue of C. elegans VMAT mutants by transgenic expression of human VMAT1 or VMAT2.
MATERIALS AND METHODS
Growth and culture
C. elegans were grown on nematode growth medium (NGM) as described by Brenner (1974), modified by the addition of streptomycin and mycostatin to reduce contamination and the use of the streptomycin-resistant bacterial strain OP50/1 (Johnson et al., 1988). Strains containing cat-1(e1111), egl-6(n592),egl-13(n483), egl-14(n549), dpy-1(e1),sup-5(e1464), sup-5(e1877),tra-2(e2046), unc-86(e1416), andunc-104(e1265) were obtained from theCaenorhabditis Genetics Center (University of Minnesota, Minneapolis, MN); cat-1(n733) was a gift from Beth Sawin and Bob Horvitz (Massachusetts Institute of Technology, Cambridge, MA);pha-1(e2123) was a gift from Heinke and Ralf Schnabel (Max-Plank-Institute fur Biochemie, Martinsried, Germany); and strains containing mgIs18 [Pttx-3-green fluorescent protein (GFP)] and mgIs21 (Plin-11-GFP) were gifts from Oliver Hobert (Harvard University, Cambridge, MA).
Oligonucleotide primers were designed from the predicted open reading frame of the partial VMAT homolog on cosmid W01C8 and were used to amplify a predicted 720-bp fragment from a C. eleganscDNA library prepared in λ-Zap II (a gift from Bob Barstead, Oklahoma Medical Research Foundation). This fragment was then used to probe the same cDNA library (to obtain a full-length cDNA) as well as a C. elegans genomic library prepared in λ-Dash (gift from Heidi Browning and Tom Blumenthal, Indiana University, Bloomington, IN). DNA sequence was determined using the fmol DNA cycle sequencing system (Promega, Madison, WI). The cDNA was completely sequenced on both strands. Sequence alignment and analysis were performed with the Genetics Computer Group (Madison, WI) Wisconsin package (version 8, September 1994). Mutations were analyzed by amplification of specificcat-1 genomic regions using direct “single-worm PCR” from individual mutant animals (Barstead and Waterston, 1991), followed by sequencing of the purified PCR product with nested primers. The altered sequence associated with the n733 missense mutation was engineered into the wild-type cDNA using the QuickChange system (Stratagene, La Jolla, CA). The resulting mutant cDNA (referred to as CelVMAT/n733 in this study) was then sequenced to assure that no other nucleotide changes were present.
Histofluorescence of monoamines was performed with a modified version of the sucrose-potassium phosphate-glyoxylic acid (SPG) reaction of de la Torre (1980). Nematodes were collected in drops of water on poly-l-lysine-coated slides and slightly compressed by a second slide. The slide sandwich was immediately placed on dry ice for 10–60 min. The frozen slides were separated, and the bottom slide with adherent nematodes was immediately placed in ice-cold SPG solution (0.2 m sucrose, 235 mmKH2PO4, and 1% glyoxylic acid, pH 7.4) for 7.5 min. Slides were dried under a cool hair dryer for 10–30 min. Light mineral oil was placed on the slide, and it was incubated at 95°C for 2.5 min. Coverslipped slides were observed with a Zeiss (Thornwood, NY) Axiophot microscope using 4′,6-diamidino-2-phenylindole (DAPI) and fluorescein isothiocyanate (FITC) filters.
VMAT. Antisera were raised against two synthetic peptides. PEP1 (VELRQNGDSRVTNEN) was derived from the C-terminal region of C. elegans VMAT (amino acids 514–528). The peptide was manufactured either as a multiple antigenic peptide (MAP) (Posmett et al., 1988) or with an N-terminal cysteine, which was used to couple the peptide to maleimide-activated keyhole limpet hemocyanin (KLH; Imject, Pierce, Rockford, IL). Goats were immunized with the peptide in MAP form, the peptide coupled to KLH, or a combination of the two forms of peptide; rabbits were immunized with KLH-coupled peptide. Three goats and two of three rabbits yielded specific antisera.
A second peptide, PEP2 (KIDRGEPEGSSIKQ), derived from a region between putative transmembrane domains 6 and 7 (amino acids 302–315 ofC. elegans VMAT), was synthesized with an N-terminal cysteine. PEP2 was coupled to KLH and used to immunize two goats and two rabbits. One rabbit yielded specific antisera.
To purify the sera, glutaraldehyde was used to cross-link the peptides to goat serum albumin or rabbit serum albumin (Harlow and Lane, 1988). Cross-linked peptide was bound to nitrocellulose membranes with methanol (Smith and Fisher, 1984). Antisera from goats were incubated with peptide cross-linked to goat serum albumin; antisera from rabbits were incubated with peptide cross-linked to rabbit serum albumin. After incubation and rinsing, bound antibody was eluted with a low-pH, high-salt wash (5 mm glycine, 0.5 m NaCl, pH 2.3) followed by a high-pH wash (50 mm TEA, pH 11.5). After neutralization with Tris buffer, the sera were exchanged into PBS and concentrated using Centriprep 30 ultrafiltration (Amicon, Beverly, MA).
The reported staining pattern was obtained with antisera generated against both PEP1 and PEP2. Most cell identification was done with antisera against PEP1-MAP (goat 258), because this serum gave the most specific signal with indirect immunofluorescence. The specific staining described below was eliminated by preincubation of the sera with the appropriate uncoupled peptide.
Serotonin. Rabbit antibody to formaldehyde-conjugated serotonin was purchased from Dr. Harry Steinbusch (Free University, Amsterdam, The Netherlands).
Anti-human VMATs. Rabbit antibodies to human VMAT1 and VMAT2 were generated against specific C-terminal peptides, as previously described (Erickson et al., 1996).
Anti-GFP. Rabbit antibody to GFP was purchased from Molecular Probes (Eugene, OR).
Nematodes were prepared with a variation of the freeze–crack method of Albertson (1984). Mixed populations of nematodes were rinsed and placed in a water drop on a poly-l-lysine-coated slide (made by incubating acid-cleaned slides for 5 min in 1–2 mg/ml poly-l-lysine). A second poly-l-lysine-coated slide was placed on top of the nematodes so that the nematodes were compressed. The slide sandwich was immediately placed on a piece of dry ice for at least 20 min. The slides were separated, and the bottom slide was immediately placed in ice-cold fixative. For VMAT (or VMAT and GFP), the fixation consisted of 2 min in methanol followed by 4 min in acetone. For anti-serotonin, fixation was done in 4% formaldehyde in 0.1 m phosphate for 24 hr.
After fixation, slides were rinsed in PBS. All slides were incubated in 10% donkey serum in antibody buffer (0.5% Triton X-100, 1 mm EDTA, and 0.1% BSA in PBS with 0.05% sodium azide) for 1 hr. Primary antibody incubations (1:50–1:200) were done overnight. After thorough rinsing with antibody buffer, slides were incubated in secondary antibody for 4 hr. Unlabeled and indocarbocyanine (Cy3)-labeled secondary antibodies were obtained from Jackson ImmunoResearch (West Grove, PA); Oregon Green 488 was coupled to secondary antibodies using the Oregon Green labeling kit from Molecular Probes. After rinsing, slides were mounted in antibleaching medium (Finney and Ruvkun, 1990).
Functional expression of VMAT cDNAs was obtained using the vaccinia virus/T7 hybrid system (Fuerst et al., 1986). The transport of [3H]serotonin and [3H]dopamine was measured in digitonin-permeabilized CV-1 cells expressing the cDNAs essentially as described by Erickson and Eiden (1993). Labeled substrates were purchased from DuPont NEN (Boston, MA); final concentrations in the assay were 63 nm for dopamine and 130 nm for serotonin. Assays were performed at 37°C for consistency with previously published studies; however, assays performed at 25°C (the maximum permissible growth temperature for C. elegans) gave qualitatively similar results (data not shown).
Body movement, pharyngeal pumping, and grazing were all measured on young adult hermaphrodites raised at 20–25°C. At these temperatures, only individuals of the transgenic lines that carry the extrachromosomal array in the cells of the pharynx during midembryogenesis will survive.
For reserpine treatment, 5 μl of reserpine (50 mm in acetic acid) was diluted in 395 μl of M9 buffer and was poured over the surface of a 6 cm NGM plate (with or without food). Plates were used after at least 15 min, when the fluid had been absorbed by the agar and/or had evaporated; plates remained potent for several days. For behavioral assays, nematodes were raised and monitored on reserpine treated plates.
Movement on and off food was assayed as described (Sawin, 1996), except that nematodes were rinsed by transferring them to a thin layer of S-basal buffer on an NGM plate for 1 min before transferring them to the test plates. Pharyngeal pumping rates on and off food were quantified as previously described (Miller et al., 1996) for 100 hermaphrodites of each phenotype. Thrashing assays were performed as described by Miller et al. (1996), except that counting was done at room temperature (∼23°C) in M9 on 6 cm NGM plates.
“Grazing” behavior was evaluated as follows. Groups of 10–30 nematodes were transferred with a metal pick to a new plate with a thin central streak of bacterial lawn. The nematodes were placed at the edge of the plate, ∼1.5 cm from the lawn. As the nematodes moved outward from the point of transfer, they spontaneously encountered the edge of the lawn. We measured the time between when the tip of the snout entered the food and the tip of the tail entered the food. Wild-type hermaphrodites generally slow their forward progression significantly when moving from a region of no visible bacteria into the bacterial lawn, so that many individuals take more than 1 min to fully enter the lawn. Generally, cat-1 animals do not change their speed when entering food. Spontaneous locomotion off food generally propels wild-type worms forward one body length every 3–5 sec. The percentage of individuals of a given phenotype that took ≥1 min to enter the lawn (grazers) was determined for each phenotype. Individuals of any phenotype taking >10 sec to move forward one body length after stimulation with a pick were excluded from the data. Individuals encountering the lawn <1 min after transfer to the assay plate were also excluded from the data, because the transfer itself could cause a temporary increase in locomotion.
To evaluate male mating behavior, individual males were put with individual immature tra-2 females onto 6 cm agar plates with an ∼1 cm central dot of bacteria. After maturation of thetra-2 female, the pairs were kept on the plate for 24–36 hr at room temperature. tra-2 females make no sperm (Hodgkin and Brenner, 1977), so all progeny are cross-progeny from successful mating(s) by the tested male. Wild-type males generally mate multiple times and produce >100 progeny, whereas fewer than half ofcat-1 males mate successfully, and almost none of them were able to sire >50 cross-progeny.
The genomic phage RM#424L was isolated as described above and contained the complete VMAT gene plus ∼3 kb of upstream sequence. The cDNA plasmids for transformation used the pPD49.26 vector (a gift from Andy Fire, Carnegie Institution, Washington, DC) with the C. elegans VAMP (synaptobrevin) promoter cloned into the first multiple cloning site and one of three cDNAs cloned into the second multiple cloning site. The VAMP gene is expressed in all neurons (Nonet et al., 1998); a 3 kb genomic clone containing the VAMP promoter was obtained from Mike Nonet (Washington University, St. Louis, MO) and was slightly modified to add some restriction sites at the 3′ end. The cDNAs used were derived from either the C. elegans VMAT (described above), the human VMAT1 (Erickson et al., 1996), or the human VMAT2 (Erickson and Eiden, 1993). DNA transformation methods forC. elegans were essentially those of Mello et al. (1991), except that a plasmid containing the wild-type pha-1 cDNA (a gift from Heinke and Ralf Schnabel, Max-Plank-Institute fur Biochemie) was used as a transformation marker. The pha-1(e2123) mutant is temperature-sensitive for embryogenesis; animals homozygous for this mutation will not hatch at 25°C but can grow normally at 16°C (Granato et al., 1994). The recipient strain for transformation had the genotype cat-1(e1111); pha-1(e2123) and was constructed in our laboratory. After injection, the recipient animals were transferred to 25°C to select for those progeny expressing the wild-type PHA-1 protein.
The C. elegans VMAT gene
The C. elegans Genome Sequencing Project reported the sequence of a cosmid (W01C8) containing part of a gene with similarity to mammalian VMATs. Using standard methods we isolated a genomic λ phage, RM#424L, which included this gene, and we isolated and sequenced the corresponding cDNA. This 1.8 kb cDNA was apparently full-length, based on the presence of a polyA tail at the 3′ end, and part of the SL1 trans-spliced leader sequence (Krause and Hirsh, 1987; Bektesh et al., 1988) at the 5′ end. Subsequent to our cDNA analysis, the C. elegans Genome Sequencing Project reported the sequence of cosmid E03E2, which overlaps W01C8 and contains the entire VMAT gene.
The deduced open reading frame of C. elegans VMAT encodes 553 amino acids with a predicted 12-transmembrane domain structure (Figs. 1 and2). There does not seem to be any obvious correlation between the exon organization of the gene and the domain structure of the protein. The deduced protein is 47% identical to human VMAT1 and 49% identical to human VMAT2 (Fig. 2) (Erickson and Eiden, 1993; Erickson et al., 1996). Inspection of the highly conserved region (which includes transmembrane domains 2–12) reveals 78 residues (of 344) where the human VMAT1 and VMAT2 sequences differ from each other; at 19 of those sites the C. elegans VMAT has the residue present in VMAT1, whereas at 26 of the sites the C. elegans VMAT has the residue present in VMAT2 (and at the remaining 33 sites all three proteins differ). The C. elegans protein thus has molecular features of both mammalian VMATs. This is consistent with the dendrogram of the members of the VMAT/VAChT protein family (Fig. 3), indicating that the C. elegans VMAT is comparably distant from the mammalian VMAT1 and VMAT2 classes.
The expressed C. elegans protein has VMAT activity
The C. elegans VMAT was expressed in CV-1 cells (see Materials and Methods), and a permeablized cell assay (Erickson and Eiden, 1993) was used to measure transport activity. The C. elegans VMAT mediated uptake of [3H]dopamine and [3H]serotonin that was time-dependent and saturable (Fig. 4) and was inhibited by an excess of unlabeled substrate. The uptake of both substrates was inhibited by the known VMAT inhibitors tetrabenazine and reserpine and was blocked by carbonyl cyanidep-trifluoromethoxyphenylhydrazone (FCCP), which disrupts transmembrane pH gradients (Fig. 4 A, inset). This inhibitor profile is a similar to that of mammalian VMATs (Erickson and Eiden, 1993).
Kinetic analyses indicated competitive inhibition of [3H]dopamine uptake by dopamine, norepinephrine, serotonin, histamine, tyramine, and octopamine, suggesting that all of these compounds were potential substrates for VMAT (Fig.5, Table1). Experiments monitoring the inhibition of [3H]serotonin uptake gave similar results (Table 1): for both substrates, the rank order of affinity was dopamine ∼ tyramine > serotonin > norepinephrine ∼ octopamine > histamine. The major difference between the C. elegans protein and the two human proteins is in the affinity for dopamine, which is ∼20- to 100-fold higher for the nematode protein (Table 1). Mammalian VMAT1 and VMAT2 differ significantly from each other in their affinity for histamine (Erickson et al., 1996); in this respect, C. elegans VMAT is more like VMAT2 than VMAT1 (Table 1). To our knowledge, tyramine and octopamine have not previously been reported as potential substrates for any VMAT.
CeVMAT is associated with synaptic vesicles
Antibodies to VMAT peptides (see Materials and Methods) were used for indirect immunofluorescence staining of wild-type C. elegans. VMAT-specific staining in the nervous system was punctate and was observed primarily in synaptic regions (Fig.6). In general, neuronal processes and cell bodies were poorly stained or not stained at all, although the trajectories of many processes could be inferred from the punctate immunofluorescence presumably associated with the en passant synapses made by the processes.
Analysis of unc-104 mutants suggested that the VMAT immunoreactivity was associated with synaptic vesicles. Theunc-104 gene encodes a kinesin-related protein, which is required for the transport of synaptic vesicles from neuronal cell bodies along the axons to synapses (Hall and Hedgecock, 1991; Otsuka et al., 1991). In unc-104 mutants, synaptic vesicles are not found at synapses but, rather, are found in large clusters in cell bodies (Hall and Hedgecock, 1991); in addition, a number of synaptic vesicle-associated proteins, such as synaptotagmin and the vesicular acetylcholine transporter (UNC-17/VAChT), are mislocalized to neuronal cell bodies (Nonet et al., 1993; Alfonso et al., 1993). We find that VMAT staining is similarly mislocalized to cell bodies inunc-104 mutants (Fig. 6), consistent with a synaptic vesicle association of this protein.
In addition, punctate non-neuronal VMAT staining was observed in three somatic cells in the male gonad (data not shown). Interestingly, synaptotagmin has also been found in a different set of cells in the male vas deferens (Nonet et al., 1993).
Identification of VMAT-containing cells
Most C. elegans neurons may be identified on the basis of cell body position, size, and process morphology (White et al., 1986). As indicated above, VMAT immunostaining in unc-104mutants was present in neuronal cell bodies, which helped in the identification of specific neurons expressing the protein. Many of these same cells were also visualized by an induced fluorescence technique or with anti-serotonin antibodies. We observed strong and reproducible VMAT-specific immunoreactivity in 20 neurons and weak and variable staining in 5 additional neurons, all of which are listed below.
ADE, PDE, and CEP
The two ADE neurons, two PDE neurons, and four CEP neurons are sensory cells previously reported to contain dopamine-like FIF (Sulston et al., 1975). We have also observed dopamine-like induced fluorescence using a slightly different method (Fig.7). These neurons have ciliated endings in the deirids (ADE and PDE) or the cephalic sensilla (CEP). VMAT-positive regions in the PDE cells usually extend anteriorly in the ventral nerve cord to at least the retrovesicular ganglion (as inHedgecock et al., 1985; but unlike White et al., 1986).
The two NSM neurons are prominent cells of the pharynx. FIF studies (Horvitz et al., 1982) and anti-serotonin immunohistochemistry (Desai et al., 1988) have shown that these are the “juiciest” serotonin-containing neurons in C. elegans hermaphrodites. By ultrastructural analysis, they have sensory endings and neuromuscular output, as well as varicosities, fine branches, and endings on the surface of the pharynx (Albertson and Thomson, 1976), suggesting that serotonin might be released into the pseudocoelom and have a humoral function.
The two HSN neurons have been previously reported to contain serotonin (Desai et al., 1988). These are motor neurons with cell bodies in unique positions in the lateral midbody; they receive input from several interneurons, and their predominant morphological outputs are to the vulval muscles and the VC neurons (White et al., 1986). In addition, their axons run anteriorly in the ventral nerve cord to the nerve ring, with minor synapses onto a number of motor neurons and interneurons (White et al., 1986).
VC4 and VC5
The VC cells are a set of six postembryonically derived motor neurons with cell bodies in the ventral nerve cord (Sulston, 1976;White et al., 1976). VC4 and VC5, the members of the class closest to the vulva, make numerous neuromuscular synapses onto the vulval muscles; the other VC cells are reported to have less extensive output to the same muscles (White et al., 1976, 1986). In addition, all of the VC cells have sparse output to the ventral body muscles and other motor neuron classes (White et al., 1976, 1986). We have confirmed unpublished reports (see Rand and Nonet, 1997b) that VC4 and VC5 contain weak and variable serotonin immunoreactivity and serotonin-like induced fluorescence (data not shown). Interestingly, although the VC cells arise during the molting period between the L1 and L2 stages, VMAT immunoreactivity is not acquired by VC4 and VC5 until the L4 stage; this is approximately the same time that the extensive innervation of the vulval muscles occurs (Li and Chalfie, 1990). In contrast, other VMAT-positive cells acquire immunoreactivity within a short time after birth.
ADF and RIH
There have been reports of serotonin staining in the two ADF neurons and the RIH neuron (Sawin, 1996 and see Rand and Nonet, 1997b). The ADF cells are sensory neurons involved in chemotaxis (Bargmann and Horvitz, 1991a) and dauer larva development (Bargmann and Horvitz, 1991b). The function of the RIH neuron is unknown. We have confirmed that ADF and RIH contain weak, variable, serotonin-like induced fluorescence and serotonin immunoreactivity in unc-104mutants (data not shown). The ADF cells contain significant amounts of VMAT immunoreactivity, whereas RIH is weakly positive for VMAT immunoreactivity in unc-104 mutants.
AIM and RIC
Previous unpublished reports noted a pair of serotonin-positive cells in the head; these were tentatively identified as the RIG neurons (see Rand and Nonet, 1997b). Using unc-104 mutants, we saw two pairs of VMAT-positive cells in the region of, but anterior to, the RIG cells. To aid in the identification of these cells, we stained two strains of transgenic animals (obtained from Oliver Hobert) which express GFP in the AIY neurons (Hobert et al., 1997) or the AIZ and RIC neurons (Hobert et al., 1998). Using these local landmarks, we identified one pair of VMAT-positive cells as the (adjacent) AIM neurons. In unc-104 mutants, these VMAT-positive cells were occasionally weakly positive for serotonin immunoreactivity or serotonin-like induced fluorescence. Using a different set of criteria,Sawin (1996) also concluded that the AIM cells are serotonin-positive. Using unc-86 mutants (Finney and Ruvkun, 1990) and these GFP landmarks, the second pair of VMAT-positive cells were identified as the RIC cells. These cells were never positive for serotonin immunoreactivity or for serotonin- or dopamine-like induced fluorescence. Therefore, the RIC neurons may use a different amine neurotransmitter.
The two CAN cells have cell bodies and processes along the excretory canals, which extend laterally along the body (White et al., 1986). They are immunopositive for VMAT but appear to be negative for serotonin immunoreactivity and induced fluorescence. Only a single synapse onto an epidermal cell was reported by White et al. (1986), but in wild-type animals, we observe a few spots of VMAT immunoreactivity.
We sometimes observe an additional pair of VMAT-positive cells in the lateral ganglia. The staining is weak and variable and is only seen in unc-104 mutants; we are not yet certain about the identity of these two cells. In addition, we observe a large number of male-specific VMAT-positive cells in the ventral nerve cord and the tail; the identity and properties of these cells will be reported in the future.
Does cat-1 encode VMAT?
Based on the position of the VMAT gene on the physical map, it was possible to assign it an approximate location on the C. elegans genetic map, and this in turn suggested that it might correspond to a previously identified mutant locus. The most likely candidates, based on map position and mutant phenotype, werecat-1 (see phenotypic description below) andegl-6, egl-13, and egl-14, three egg-laying defective mutants that were considered because of the involvement of serotonin in egg laying (Horvitz et al., 1982). However, mutants in all three of these Egl genes had wild-type patterns of induced fluorescence and VMAT immunoreactivity in the head, suggesting that they did not encode VMAT (data not shown).
The cat-1 gene was first identified by Sulston et al. (1975)in a screen for mutants with abnormal patterns of FIF. cat-1mutants lack (dopamine-specific) FIF in neuronal processes, although some FIF is still present in cell bodies (Sulston et al., 1975). In addition, serotonin immunoreactivity is variably reduced in serotonin-containing neurons (Loer and Kenyon, 1993). These are the same phenotypes caused by treatment of wild-type C. eleganswith reserpine, which led to the suggestion more than 20 years ago that this gene might encode a synaptic vesicle neurotransmitter transporter (Sulston et al., 1975).
There are two alleles of cat-1 currently available:e1111 and n733. Both mutations lead to loss of dopamine- and serotonin-specific induced fluorescence (Fig. 7). Analysis of cat-1(e1111) mutants shows that they are completely deficient for VMAT immunoreactivity (using antiserum specific for peptide 1; see Materials and Methods), whereascat-1(n733) animals have slightly less VMAT immunoreactivity than wild type (Fig. 7).
Sequence of cat-1 mutant alleles
Both cat-1 mutant alleles were associated with mutations in the VMAT coding sequence (Fig. 1). The cat-1allele e1111 contains a G to A transition, which changes a tryptophan to an amber termination codon (located between transmembrane domains 3 and 4); this is consistent with the complete lack of immunoreactivity in e1111 homozygotes. This allele has been reported to be suppressible by the amber-suppressor genesup-5 (Waterston and Brenner, 1978). We have determined thate1111 homozygotes regain a low level of VMAT immunoreactivity in a sup-5 mutant background (data not shown). Furthermore, sup-5 partially suppressed the mating defect of cat-1(e1111) males (Table2).
The cat-1(n733) mutation is a glycine to arginine missense mutation in the middle of transmembrane domain 5; this is consistent with presence of VMAT immunoreactivity in n733 homozygotes. In addition, the mutant protein appears to be correctly localized (Fig.7), which suggests that the mutant phenotype derives from impaired protein function. Consistent with this hypothesis, the VMAT protein corresponding to the n733 mutation, CelVMAT/n733, has only 14% of the activity of wild-type CelVMAT for [3H]serotonin uptake in the mammalian cell-based transport assay (Fig. 4B inset), although the residual transport activity of the mutant was inhibited by reserpine to the same degree as the wild-type transporter; 100 nm reserpine decreases wild-type and mutant transport to 6.3 ± 2.4 and 5.7 ± 2.8% of control transport, respectively (mean ± SEM of three separate experiments).
Behavioral phenotypes of cat-1 mutants
cat-1 mutants have several behavioral deficits, which appear to reflect deficient function of biogenic amine-containing cells. Using laser ablation, Sawin (1996) demonstrated that the eight dopamine-containing cells (ADE, PDE, and CEP) are collectively required for a wild-type response to bacteria. Wild-type hermaphrodites move significantly slower when they are in a bacterial lawn than when they are on clean agar. When all of the dopamine-containing cells are ablated, locomotion is the same on or off the lawn (Sawin, 1996). Using the same paradigm, we found that the locomotion of cat-1mutants is also insensitive to the presence of bacteria:cat-1 mutants move at the same rate on or off the bacterial lawn (Table 2).
Wild-type animals display two other responses to the presence of food: they slow down dramatically when first encountering a bacterial lawn and graze for a while before resuming movement, and the rate of pharyngeal pumping is increased in the presence of the lawn. We have shown that these responses to the bacterial lawn are also deficient incat-1 animals (Fig. 8), suggesting that these paradigms are also mediated through aminergic neurons.
cat-1 mutants also have deficits in serotonin-regulated behaviors. The easiest such behavior to measure is egg laying. Exogenous serotonin stimulates egg laying (Horvitz et al., 1982), whereas ablation of the serotonin-containing HSN cells leads to a profound decrease in egg laying such that the animals become quite bloated with unlaid eggs (Trent et al., 1983; Desai et al., 1988; Desai and Horvitz, 1989). We have found a mild but reproducible temperature-sensitive reduction in the rate of egg laying bycat-1 mutants (Table 2), although we have not yet determined the identity of the cells that mediate this effect. It is noteworthy that cat-1 mutants, as well as other serotonin-deficient mutants (e.g., cat-4 andbas-1), clearly do not display the severe egg-laying defect associated with ablation of the HSN cells; this has led to the proposal that the HSNs mediate egg laying by using another neurotransmitter in addition to serotonin (Weinshenker et al., 1995).
In addition, as previously described, cat-1 males are deficient in mating performance (Table 2); this behavior has been shown to require the male-specific serotonin-containing CP motor neurons (Loer and Kenyon, 1993), as well as dopamine- and serotonin-containing sensory rays in the male tail (Liu and Sternberg, 1995).
Rescue of cat-1 phenotypes by transgenic VMAT expression
Using several different assays, we found that the genomic phage RM#424L was able to rescue several cat-1 mutant phenotypes in some animals (Figs. 7, 8). However, experiments with the genomic phage were complicated by an apparent dosage-sensitive toxicity: many of the transgenic animals were small, and some were dumpy and uncoordinated. In these individuals, VMAT was expressed in some or all of the hypodermal seam cells. It is possible that this was attributable to sequences unrelated to VMAT function, although we have not explored this in any detail.
To eliminate any potential problems caused by such genomic sequences, we prepared rescue constructs in which the C. elegans VMAT cDNA was driven by the C. elegans VAMP (synaptobrevin) promoter; the VAMP gene is expressed in all neurons (Nonet et al., 1998). When introduced into cat-1 animals, such constructs led to VMAT immunoreactivity in most neurons and provided behavioral rescue (Fig. 8) with only minimal behavioral abnormalities. These results provide confirmation that C. elegans VMAT is encoded by cat-1.
Rescue of cat-1 mutants with human VMATs
We then prepared similar rescue constructs containing either the human VMAT1 or VMAT2 cDNA driven by the C. elegans VAMP promoter and introduced them into cat-1 mutants. Using antibodies specific for the human proteins, we found significant levels of expression of the heterologous proteins, as well as correct localization of the human VMATs to synaptic regions (Fig.9). However, the transgenic lines containing the human cDNA constructs showed nonuniform expression of the transgenes: often the normally VMAT-positive cells contained little or no heterologous protein. We estimate that only perhaps 10% of the transgenic animals had significant transgene expression in most of the normally VMAT-positive cells. Although we do not yet understand the biological basis for the variability of expression, both of the human VMAT constructs nevertheless provided significant rescue ofcat-1 behavioral defects in pharyngeal pumping (Fig.10) and in grazing (data not shown). This rescue was similar for human VMAT1 and VMAT2 cDNAs (Fig. 10).
C. elegans VMAT
The gene we have characterized appears to be a structural and functional homolog of the mammalian VMAT genes. This conclusion is based on the similarity of the respective protein sequences, the localization of the C. elegans protein to amine-containing neurons, the association of the protein with synaptic vesicles, the ability of the C. elegans protein to transport amines in anin vitro assay, and the ability of mammalian VMATs to restore partial function in C. elegans cat-1 mutants.
The sequence in Figure 2 and the dendrogram in Figure 3 show thatC. elegans VMAT, although clearly a close relative of the mammalian VMATs, is neither a VMAT1 nor a VMAT2. Furthermore, theC. elegans VMAT protein is expressed in all of the cells previously shown to contain biogenic amines. Taken together, these results suggest that the gene we have described is the only VMAT gene in C. elegans (and no additional VMAT homologs are present in the >85% of the C. elegans genome sequenced to date). We conclude that the two mammalian VMAT genes diverged from each other subsequent to the divergence of nematode and mammalian ancestors.
Biochemical properties of the C. elegans VMAT
The properties of C. elegans VMAT appear more like those of the mammalian VMAT2 (“neuronal”) isoform than the VMAT1 (“neuroendocrine”) isoform: C. elegans VMAT has an affinity for histamine in the (mammalian) physiological range and is inhibited by tetrabenazine at <1 μm. Various laboratories (Peter et al., 1996; Varoqui and Erickson, 1997) have suggested that recognition of histamine and tetrabenazine may be structurally related properties of VMAT2. The ability of C. elegans VMAT to recognize histamine suggests that the mammalian VMAT1 isoform may have evolved in part to provide a carrier thatfails to recognize histamine, rather than the mammalian VMAT2 isoform evolving to provide a carrier that hasacquired affinity for histamine.
A significant difference between C. elegans and mammalian VMATs is that the nematode transporter appears to have a higher affinity for dopamine than for serotonin. It is also noteworthy that octopamine, a major invertebrate neurotransmitter also found in the mammalian nervous system, and tyramine, the precursor of octopamine, have high affinities for C. elegans VMAT. The affinities of octopamine and tyramine for mammalian VMATs have not yet been reported.
cat-1 is the VMAT structural gene
This assignment is based on sequencing and antibody staining of mutant alleles and on transgenic rescue experiments. We have shown that both cat-1 mutations are associated with point mutations in the VMAT coding sequence. We have also described four different phenotypes associated with cat-1 mutants: altered induced fluorescence pattern, altered serotonin immunoreactivity pattern, lack of VMAT immunoreactivity (in e1111), and defective behavioral responses to bacterial lawns. All of these phenotypes are rescued by both a genomic phage containing the complete VMAT gene and a VMAT cDNA under the control of the C. elegans VAMP promoter.
So far, two biogenic amine neurotransmitters have been identified in specific C. elegans neurons. Serotonin immunoreactivity or serotonin-like immunofluorescence has been described in 11 cells in hermaphrodites (Horvitz et al., 1982; Desai et al., 1988; McIntire et al., 1992; Sawin, 1996; Rand and Nonet, 1997a); we have shown that these cells are immunopositive for VMAT, supporting the hypothesis that these cells are serotonergic. Similarly, the identification of VMAT in the eight neurons reported to contain dopamine-like immunofluorescence (Sulston et al., 1975) provides additional evidence that these cells are dopaminergic. We also find VMAT in at least six additional cells, suggesting that these cells may use a different biogenic amine transmitter. It is likely that at least some of these cells use octopamine as a neurotransmitter: octopamine is present in C. elegans homogenates; exogenous octopamine has distinct behavioral effects (Horvitz et al., 1982); and we have shown it to be a substrate for VMAT.
There are reported differences among nematode species in the pattern of serotonin immunoreactivity. In Ascaris suum, only one pair of serotonin-immunopositive cells, putative NSM homologs, has been observed in females (Johnson et al., 1996). It is likely thatAscaris females do not have HSN homologs, but they appear to have homologs of ADF, AIM, RIH, and probably VC cells, and none of these neurons contains serotonin immunoreactivity (Johnson et al., 1996). The ADF, AIM, RIH, and VC cells of C. elegans have significantly weaker staining for serotonin than the NSM cells; it is therefore unclear whether the discrepancies between the two species merely represent differences in relative abundance and/or assay sensitivity or fundamental differences in differentiated neurotransmitter phenotypes.
VMAT is required for proper function of dopamine-containing neurons
Laser ablation studies have demonstrated that the dopamine-containing ADE, PDE, and CEP neurons are collectively required for the sensing of a bacterial lawn and/or the resulting slowing of locomotion (Sawin, 1996). This behavior and related bacterial sensing behaviors are deficient in cat-1 mutants. These neurons are thus apparently unable to function properly in the absence of the vesicular transporter. We believe that this is the first demonstration that vesicular amine transport is essential for the function of specific (aminergic) neurons.
Not all behaviors mediated by VMAT-positive cells are mediated by VMAT
In contrast to the results just cited for the eight dopamine-containing neurons, laser ablation studies of other VMAT-positive neurons are not in good agreement with thecat-1 phenotype. Loss of the HSN cells by ablation or programmed cell death (Trent et al., 1983; Ellis and Horvitz, 1986;Desai and Horvitz, 1989) leads to a dramatic decrease in egg laying; this is a far stronger phenotype than the mild egg-laying defect incat-1 mutants (Table 2). Ablation of the VMAT-containing CAN cells causes the animals to wither and die (J. Sulston, cited in White et al., 1986), yet cat-1 mutants are relatively healthy. Thus, for these neurons, cell ablation has more severe consequences than elimination of VMAT function in putative null cat-1mutants. It is possible that another vesicular transporter exists in some C. elegans neurons, which can transport serotonin as well as whatever amine transmitter is used by CAN (a multiple transporter model). Alternatively, the HSN and CAN cells might use a nonamine neurotransmitter in addition to the amine transported by VMAT; thus elimination of amine release leads to only partial compromise of cellular function (a multiple transmitter model).
For HSN cells, we favor the second model. Using both antibody staining and induced fluorescence, we see no serotonin present in HSN processes and synaptic regions of cat-1 mutants; this suggests a total lack of vesicular serotonin transport (and degradation of the transmitter in the cytoplasm) rather than the presence of another transporter for serotonin. In addition, genetic and pharmacological studies have led Weinshenker at al. (1995) to suggest that HSN cells use a second neurotransmitter, probably acetylcholine, as well as serotonin. With respect to CAN, until we have information about the putative neurotransmitter(s) used by this pair of cells, we cannot decide between multiple transporters and multiple transmitters. In fact, a third possibility for CAN is that VMAT expression is unrelated to neurotransmitter release or function.
Functional roles of monoamines in C. elegans
Previously published reports of deficits in cat-1mutants include reduced male mating efficiency (Sulston et al., 1975), slight hyperactivity (Loer and Kenyon, 1993), slightly smaller size, and slightly reduced feeding (Avery and Horvitz, 1990). We have now shown defects in some sensory responses and a mild defect in egg laying. It is striking that apparent elimination of all synaptic biogenic amine function in cat-1 mutants does not lead to profound behavioral deficits. There are several possible explanations for this. Although specific cells and functions may be “dispensable” under laboratory conditions, the evolutionary persistence of such cells and functions suggests a relative selective advantage to the animal in a “normal,” i.e., soil, environment. For example, the ability to sense and respond to food is expected to be far more important in the soil than in a slurry of bacteria. Another possible explanation for the mild phenotype of cat-1 mutants is redundancy of genes, neuronal pathways and/or neurotransmitters. As discussed above, we consider the possibility of an additional VMAT gene in C. elegans to be remote, but it cannot be completely excluded until the sequencing of the genome is complete. Nevertheless, the observation that ablation of the eight dopamine-containing cells leads to the same phenotype as loss of VMAT activity within those cells (Sawin, 1996) argues that there are no other transporters or transmitters for this cellular function.
cat-1 rescue studies and VMAT function
We have demonstrated rescue of several phenotypes withcat-1-containing transgenic arrays, including induced fluorescence (in both putative dopamine and putative serotonin containing cells), serotonin immunoreactivity pattern, and two behavioral responses to bacteria (grazing and pharyngeal pumping). We were also able to demonstrate partial phenotypic rescue using human VMAT1 and VMAT2 transgenes. We do not understand the reason for the variable cellular expression of the human VMATs, but the level of rescue was comparable to the degree of cellular expression of the transgenes. It therefore appears that when either human VMAT1 or VMAT2 is expressed in the proper cells, it can substitute for C. elegans VMAT.
Knock-out of the VMAT2 gene in mice has recently been reported (Fon et al., 1997; Wang et al., 1997). The null phenotype is lethal, which is not surprising in view of the many complex monoamine-dependent functions required for early postembryonic life (Thomas et al., 1995;Zhou et al., 1995) and the probability that impairment of at least one monoaminergic function might have dramatic biological consequences. Is VMAT absolutely required for all monoaminergic systems, or is VMAT merely necessary for optimal neurotransmission? The present report indicates that on both an organismal and a cell biological level, VMAT is required for a broad range of monoamine-dependent behaviors and functions in C. elegans. The utility of in vivoanalysis in C. elegans, however, is that behaviorally impaired mutants can be sustained in an artificial environment, which allows VMAT function to be linked to behavior in the context of individual, defined circuits that include monoaminergic neurons.
The cat-1 gene and mutants, together with cell-specific promoters and other tools now being developed, will allow us to address how deficits in the metabolism and release and reuptake of specific monoamine transmitters in specific cells can directly affect complex behaviors, and how these deficits might be specifically ameliorated with pharmacological maneuvers. This has implications for gene therapeutic, grafting, and other approaches to human monoamine-linked diseases.
This work was funded by Grant GM38679 from the National Institute of General Medical Sciences to J.B.R., a grant from the Oklahoma Center for the Advancement of Science and Technology to J.S.D., a National Research Service Award from the National Institute of Neurological Disorders and Stroke to D.L.F., and a Pharmacology Research Associate Trainee Fellowship to K.A. We thank Dr. Jeffrey Erickson for assistance with the transport assay for CelVMAT and advice and assistance in preparation of hVMAT-reconstituted C. elegans strains. We also acknowledge our gratitude to the community of C. elegans researchers for their generous sharing of reagents and information: Beth Sawin provided advice on behavioral assays; Bob Barstead, Heidi Browning, and Tom Blumenthal provided libraries; Cori Bargmann helped with cell identification; Andy Fire provided transformation vectors; Beth Sawin and Bob Horvitz provided thecat-1(n733) allele; Mike Nonet provided the C. elegans VAMP promoter; Heinke and Ralf Schnabel provided thepha-1 plasmid; Oliver Hobert provided the Pttx-3-GFP and Plin-11-GFP transgenic strains; Gian Garriga and Curtis Loer shared unpublished data; and the C. elegans Genome Sequencing Project made our research much easier. Oligonucleotide primers and peptide immunogens were synthesized by the Molecular Biology Resource Facility of the University of Oklahoma Health Sciences Center. Some nematode strains used in this work were provided by the Caenorhabditis Genetics Center, which is funded by the National Institutes of Health National Center for Research Resources.
Correspondence should be addressed to Dr. James B. Rand, Program in Molecular and Cell Biology, Oklahoma Medical Research Foundation, 825 Northeast 13th Street, Oklahoma City, OK 73104.