FORMATION OF MATURE NEUROPEPTIDES

Neuropeptides are derived from larger precursor molecules, which must be cleaved to yield individual, active neuropeptides. Precursor molecules may contain multiple neuropeptides that are distinct from each other, a single neuropeptide, multiple copies of the same neuropeptide or a combination of these possibilities. Precursor molecules that encode multiple, distinct neuropeptides may undergo differential cleavage patterns among different cell types, thereby changing the composition of neuropeptides available in a specific cell.

The typical cleavage site in C. elegans is C-terminal to dibasic residues, which flank the peptide sequence (Rosoff et al. 1993; Marks et al. 1995, 1997, 1998, 1999a, 2001); however, cleavage after mono- or tribasic residues also occurs (Rosoff et al. 1993; Marks et al. 1997, 2001; Husson et al. 2005). These basic site cleavages are performed by kex2/subtilisin-like proprotein convertases (PC). C. elegans contains four PCs, but whether all four of the PCs cleave neuropeptide precursors is still unclear. Loss of kpc-1 and egl-3/kpc-2 results in behavioural deficits, such as defects in locomotion (Thacker & Rose, 2000; Kass et al. 2001; Jacob & Kaplan, 2003), egg-laying (Kass et al. 2001; Jacob & Kaplan, 2003), mechanosensation (Kass et al. 2001; Jacob & Kaplan, 2003), and growth (Thacker & Rose, 2000), suggesting that these two PCs are involved in neuropeptide processing. egl-3/kpc-2 is expressed in many, but not all of the neurons in the nervous system (Kass et al. 2001). Because neuropeptide immunoreactivity is decreased but not eliminated in egl-3/kpc-2 mutants, multiple PCs are likely to function in single neurons (Kass et al. 2001). Complete loss of function alleles in bli-4/kpc-4 result in lethality (Thacker et al. 1995), so the role of bli-4/kpc-4 in the nervous system has been difficult to assess. However, like egl-3/kpc-2 (Kass et al. 2001), bli-4/kpc-4 is expressed in the nervous system (Thacker et al. 1995; Thacker & Rose, 2000), suggesting that bli-4/kpc-4 may also function as a PC cleavage enzyme.

Further processing of precursor molecules by carboxypeptidases E (CPEs) removes the flanking basic residues from the neuropeptide sequences (Steiner, 1998). Three CPEs are present in C. elegans (Jacob & Kaplan, 2003). egl-21 CPE is expressed in roughly 60% of the neurons. Loss of egl-21 shows similar, but more severe phenotypes than loss of the PCs. The level of FLP neuropeptide immunoreactivity is also greatly diminished in egl-21 CPE mutants, suggesting that the remaining two CPEs are responsible for cleaving the remaining FLP precursors, as well as other neuropeptides (Jacob & Kaplan, 2003).

As with the mammalian neuropeptides, many C. elegans neuropeptides are modified at their N- or C-termini, presumably to protect against degradation and/or to generate an active form of the peptide (Schinkmann & Li, 1992; Steiner, 1998; Husson et al. 2005). The most common known modification in C. elegans is amidation. Peptides are presumably amidated if they contain a C-terminal glycine, which donates an amino group in the amidation process. By definition, all predicted FLPs are amidated, and the presence of a C-terminal glycine indicates that many of the NLPs are also likely to be amidated. The enzyme(s) involved in C. elegans amidation is(are) unknown. Based on homology to the mammalian amidation enzymes, peptidylglycine-alpha-hydroxylating monooxygenase (PHM), peptidyl-alpha-hydroxyglycine alpha-amidating lyase and the bifunctional peptidylglycine α-amidating monooxygenase (PAM) (Eipper et al. 1993), C. elegans contains at least one PAM-like and one PHM molecule (Han et al. 2004).

IDENTIFICATION OF PUTATIVE NEUROPEPTIDE GENES

The complete DNA sequence of the C. elegans genome (C. elegans Sequencing Consortium, 1998) allowed scanning of the genome for candidate neuropeptide genes. The sheer number of identified neuropeptide genes, particularly within certain families, was surprising. For instance, thirty-eight genes that encode insulin-like molecules were identified (Table 1; Duret et al. 1998; Gregoire et al. 1998; Kawano et al. 2000; Pierce et al. 2001; Li, Kennedy & Ruvkun, 2003), but at the time, only one insulin-like receptor had been identified (Kimura et al. 1997). Our laboratory identified twenty-four flp genes (flp-1 to flp-23 and flp-28) that encode peptides sharing a common C-terminal RF-amide moiety (Li, Kim & Nelson, 1998; Kim & Li, 2004; C. Li, unpublished observations); recently, McVeigh and co-workers (2005) reported the identification of five other flp genes, flp-24 to flp-27 and flp-32, by EST data mining and BLAST searches (Table 2). Hart and co-workers used similarity and pattern-based scans in more general BLAST screens to identify other neuropeptide genes. Using the characteristics of neuropeptide precursor processing, pattern-based scans designed to search for peptide sequences that were flanked by mono- or dibasic sequences revealed 34 non-insulin-like, non-FLP-like nlp genes (Table 3; Li et al. 1999; Nathoo et al. 2001; A. Hart, personal communication); nlp-33 was recently identified by Couillault and co-workers (2004; Table 3) and nlp-36 to nlp-42 by Husson and co-workers (2005; Table 3). A total of 109 neuropeptide genes encoding over 250 putative neuropeptides have now been identified (Tables 1–3), but this number is likely to be an underestimate of the total number of C. elegans neuropeptide genes (see below).

CONFIRMATION OF NEUROPEPTIDE GENES

The large number of neuropeptide genes and predicted peptides was surprising, particularly because so many of the predicted neuropeptides have similar sequences. This multitude of genes immediately raised two questions: (1) are all the genes actually transcribed and (2) if the genes are transcribed, are all the predicted peptides produced? To determine whether the genes are transcribed, two basic approaches have been used. The first is to take advantage of the ongoing C. elegans EST and ORFeome projects and the second is to isolate cDNAs by reverse transcription/polymerase chain reaction. Based on the isolated cDNAs, 36 of the 38 insulin-encoding genes (Table 1; Gregoire et al. 1998; Kawano et al. 2000; Pierce et al. 2001; Li et al. 2003), 28 of the 29 flp genes (Table 2; Rosoff, Burglin & Li, 1992; Nelson et al. 1998; Kim & Li, 2004; McVeigh et al. 2005; I. Miskelly, N. J. Marks and A. G. Maule, personal communication; unpublished observations), and 38 of the 42 nlp genes (Table 3; Nathoo et al. 2001; Couillault et al. 2004) are expressed. These data suggest that most, if not all, of the predicted genes are transcribed.

The more difficult problem has been to determine whether the neuropeptides are produced and the actual sequences of the peptides. Most of the biochemical isolations initially centered on the FLPs. Thirteen FLPs encoded by seven flp genes (flp-1, 6, 8, 9, 13, 14 and 18) were isolated by high pressure liquid chromatographic fractionation of C. elegans extracts (Table 2; Rosoff et al. 1993; Marks et al. 1995, 1997, 1998, 1999a, 2001). More recently, Husson and co-workers (2005) have used more sophisticated peptidomic analysis involving two-dimensional nanoscale liquid chromatography in tandem with mass spectrometry to isolate a large number of FLP peptides, including many that were previously isolated; in total, 32 FLPs encoded by 12 genes were isolated, bringing the total number of isolated FLPs to 35. A few of the FLPs isolated by Husson and coworkers (2005) are longer peptides than predicted; these peptides may represent either incompletely cleaved peptides or longer peptides that are also functional. Two of the isolated FLPs are shorter than predicted (Husson et al. 2005). One of these peptides suggests that the initial prediction (SDRPTRAMSPLIRFamide) was incorrect and a monobasic cleavage site was used (to yield AMDSPLIRFamide). In the second example, a truncated peptide, GAMPGVLRFamide, was isolated in addition to the larger predicted peptide, DFDGAMPGVLRFamide; either the truncated peptide is a degradation product or a novel cleavage site was used in the larger peptide to produce the shorter one (Husson et al. 2005). Two FLPs that are not encoded by any of the identified flp genes have also been isolated (N. J. Marks and A. O. W. Stretton, personal communication), underscoring the difficulty of identifying small neuropeptide genes with BLAST searches and suggesting that the roughly 70 FLPs encoded by 29 genes is an underestimate.

Concurrent with the biochemical analysis of the FLPs, Husson and co-workers (2005) also isolated the first NLPs from C. elegans. Twenty nine NLPs encoded by 19 genes were isolated. As with the FLPs, some of the isolated NLPs are truncated or there are N-terminal extensions of the predicted NLPs, again suggesting that incomplete cleavages, novel cleavages, and/or degradation products are among the isolated peptides. No biochemical isolations have been performed for the insulin-like peptides.

Peptides identical to some of the predicted FLPs and NLPs have also been isolated from related nematodes, such as Ascaris suum (Cowden, Stretton & Davis, 1989; Cowden & Stretton, 1993, 1995; Yew et al. 2005), Haemonchus contortus (Keating et al. 1995; Marks et al. 1999b), and Panagrellus redivivus (Geary et al. 1992; Maule et al. 1994a,b, 1995). In particular, of the 16 C. elegans FLPs and NLPs isolated from Ascaris, five peptides encoded by flp-3, 11, 12, 18 and 21 and two peptides from nlp-12 have not been isolated from C. elegans thus far (Yew et al. 2005), suggesting that these peptides are also produced in C. elegans. However, it is the ongoing projects to isolate ESTs from various nematode species that highlight how prevalent the flp genes are in Nematoda (McVeigh et al. 2005; Yew et al. 2005). Indeed, the recent database mining by McVeigh and co-workers (2005) indicates that 33 species of nematodes, including Ancylostoma caninum, Globodera pallida and G. rostochiensis, Heterodera glycines, Meloidogyne arenaria, M. chitwoodi, and M. incognita and Onchocerca volvulus, have sequelogs to the C. elegans flp genes. These data suggest that many of the predicted neuropeptides are indeed produced and highlight the rich diversity of neuropeptides not only in C. elegans but in other nematode species as well.

EXPRESSION PATTERN OF NEUROPEPTIDE GENES

The difficulty in generating specific antisera against peptides that have very similar sequences (for instance, the insulin-like peptides share common A and B domains and the FLPs all share a common C-terminal Arg-Phe-NH₂), have led researchers to find other ways to determine the expression pattern of the different neuropeptide genes. Because transgenic animals are relatively easy to generate in C. elegans, the most frequently used approach to determine the expression patterns is to construct transcriptional fusions between the promoter region of the neuropeptide gene with the coding region of a reporter gene, of which green fluorescent protein (GFP) is the most commonly used, for microinjection. This relatively simple method has been used to determine the expression pattern of 60 neuropeptide genes (Tables 1–3), including 15 insulin-like genes (Pierce et al. 2001; Li et al. 2003), 19 flp genes (Kim & Li, 2004) and 26 nlp genes (Nathoo et al. 2001).

Although there are inherent caveats to using reporter constructs, the expression data illuminate several points about neuropeptides. First, the expression of neuropeptides is widespread in C. elegans; not only is expression seen throughout the nervous system, but expression is also seen in non-neuronal tissues (Nathoo et al. 2001; Pierce et al. 2001; Li et al. 2003; Kim & Li, 2004; Tables 1–3). The FLPs alone are expressed in over 160 neurons, greater than half the number of cells in the C. elegans nervous system (Kim & Li, 2004). Fifteen of the ins genes and daf-28 are expressed in neurons, including some of the amphidial chemosensory neurons (Pierce et al. 2001; Li et al. 2003). In addition, several ins, flp and nlp genes are expressed outside the nervous system, such as in intestine, gonad, muscle, and hypodermal cells (Nathoo et al. 2001; Pierce et al. 2001; Li et al. 2003; Kim & Li, 2004), where they presumably function in a more endocrine fashion. Secondly, as in other vertebrate and invertebrate systems, there is considerable overlap in the expression patterns. For instance, each flp gene examined thus far is expressed in a unique set of cells, although a specific cell may express several flp genes (Kim & Li, 2004). Similarly, expression of the ins (Pierce et al. 2001), daf-28 (Li et al. 2003), and nlp (Nathoo et al. 2001) genes is widespread and multiple peptide families can be expressed in the same cell. The chemosensory neuron ASI, for example, has an extremely rich neuropeptide repertoire. ASI expresses daf-28, ins-1 and 9, nlp-1, 5, 6, 9, 14, 18, 24 and 27, and flp-2, 10 and 21 (Nathoo et al. 2001; Pierce et al. 2001; Li et al. 2003; Kim & Li, 2004), suggesting that ASI has the potential to release a diverse range of neuropeptides to modulate neuronal activity.

NEUROPEPTIDE FUNCTION

Bioinformatic tools have allowed the identification of the neuropeptide genes, but the more daunting challenge for researchers is to unravel the function of the different genes and the individual peptides. The functional characterization of the genes is complicated by several factors. First, many of the peptides have highly similar sequences, particularly with peptides encoded by the same gene, but even among peptides encoded by different genes. These peptides may bind to the same receptor, making it difficult to tease apart the functions of the individual peptides. Indeed, EMPGVLRFamide encoded by flp-18 and GLGPRPLRFamide encoded by flp-21 bind to a common receptor, NPR-1 (Rogers et al. 2003). The widespread expression patterns of the neuropeptide genes suggest that the peptides are involved in a multitude of behaviours in C. elegans. However, the extensive overlap among the expression patterns of different genes suggests that many of the peptides will also have functional overlap. Despite these caveats, inactivation of several neuropeptide genes indicates that at least some of the genes have unique functions.

NEUROPEPTIDE RECEPTORS

A parallel method towards determining the function of the different neuropeptides is to inactivate the receptor(s) through which the peptides signal. This strategy circumvents the problem when multiple peptides bind to the same receptor and have functional overlap. Inactivation of a specific peptide, for instance, may not reveal a phenotype, whereas inactivation of the receptor may give insights into the function of the peptides.

DAF-2, a tyrosine kinase receptor, is most similar to the mammalian insulin-like receptor and was the only insulin-like receptor identified for several years (Kimura et al. 1997). As discussed above, inactivation of daf-2 leads to constitutive dauer formation (Riddle & Albert, 1997), increased longevity (Kenyon et al. 1993), and increased body size (McCulloch & Gems, 2003). The primary ligand of DAF-2 appears to be DAF-28 (Li et al. 2003), although INS-4 and INS-6 can also activate DAF-2 if expressed at high levels (Pierce et al. 2001). Given the plethora of insulin-like peptides, however, the dearth of insulin-like receptors was somewhat surprising. More recently, Dlakic (2002) used different search paradigms to uncover a family of 56 divergent insulin-like, tyrosine kinase receptors. Many of the receptors are clustered and appear to result from recent gene duplications (Dlakic, 2002). Presumably, some of the insulin-like peptides signal through these receptors to affect dauer formation and other processes.

As with the insulin receptors, many FLP receptors have been isolated from other systems. By contrast to the insulin receptors, however, the FLP receptors are G protein-coupled receptors (Tensen et al. 1998; Bonini et al. 2000; Cazzamali & Grimmelikhuijzen, 2002; Meeusen et al. 2002; Duttlinger, Mispelon & Nichols, 2003), with the exception of a molluscan FMRFamide-gated amiloride-sensitive channel, which has homology to the MEC-4 and MEC-6 mechanoreceptors (Lingueglia et al. 1995). Of the 1000 G protein-coupled receptors in C. elegans, over 50 are candidate neuropeptide receptors (Bargmann, 1998). Keating and co-workers (2003) used RNAi to inactivate sixty G protein receptors that were predicted to bind either a small molecule transmitter or a neuropeptide. Inactivation of six receptors, C16D6.2, C25G6.5, C26F1.6, F35G8.1, F41E7.3, and F59C12.2, affected brood size by either increasing or decreasing the number of progeny (Keating et al. 2003). Disruption of eight receptors, AC7.1 (tachykinin-like), C15B12.5, C10C6.2, C24A8.4, F15A8.5, F59D12.1, T02E9.1, and T05A1.1, affected the movement of the animals (Keating et al. 2003). The phenotypes from the RNAi data were confirmed in two cases by the isolation of deletion mutants for T05A1.1 and F35G8.1 (Keating et al. 2003). Several of the ligands for these receptors have now been identified (see below). No NLP receptors have been identified thus far.

Given the large number of G protein-coupled receptors and the even larger set of FLP ligands, several groups have developed high throughput methods to match FLP ligands to specific G protein-coupled receptor binding partners. Specifically, candidate receptors are expressed in either heterologous cells or Xenopus oocytes, FLP ligands are applied singly or in combination, and different assays are used as the readout (Kubiak et al. 2003a,b; Lowery et al. 2003; Rogers et al. 2003; Mertens et al. 2004, 2005; Table 2). Using Chinese hamster ovary (CHO) cells, the Upjohn/Pharmacia group (Kubiak et al. 2003a,b; Lowery et al. 2003) transfected candidate receptors and chimeric G proteins and screened for ligand-induced GTPγS binding to membranes of transfected cells. By this method, flp-15 peptides were matched to C10C6.2 (Kubiak et al. 2003b), flp-18 peptides to C16D6.2, F41E7.3, Y58G8a.1, C53C7.1, and C25G6.5 (Lowery et al. 2003), flp-3 peptides to C53C7.1 (Lowery et al. 2003), and FLP-21 to C25G6.5 (Lowery et al. 2003). FLPs encoded by other genes also bind to C16D6.2 and C10C6.2 (Lowery et al. 2003). Mertens and coworkers (2004, 2005) expressed candidate receptors and G_α16 in human embryonic kidney (HEK) or CHO cells and screened for an increased calcium response, as monitored by an increase in fluorescence. Their group identified that FLPs encoded by flp-7 and flp-11 bind to C26F1.6 (Mertens et al. 2004), and FLP-2 peptides bind to two isoforms of T19F4.1 (Mertens et al. 2005). Given that a receptor can bind to multiple FLPs encoded by distinct genes and a single FLP can bind to multiple receptors, the potential complexity of peptide actions in C. elegans is enormous.

The NPR-1 receptor is a G protein-coupled receptor that is homologous to the mammalian neuropeptide Y receptor (de Bono & Bargmann, 1998). Disruption of the NPR-1 receptor affects aggregation behaviour (de Bono & Bargmann, 1998) and tolerance to alcohol (Davies et al. 2004). In the presence of ample food, wild-type animals feed alone (referred to as solitary feeding); a specific amino acid change in NPR-1 causes the animals to aggregate during feeding (referred to as social feeding) and accumulate at the edges of the bacteria (referred to as bordering behaviour; de Bono & Bargmann, 1998). The aggregation behaviour of npr-1 mutants can be suppressed by mutations in gcy-35 or gcy-36 (Cheung et al. 2004), both of which encode soluble guanylate cyclases (Morton et al. 1999). Interestingly, GCY-35 guanylate cyclase binds oxygen, suggesting that the aggregation behaviour of npr-1 mutants is related to oxygen levels in the local environment of the animals (Gray et al. 2004). No neuropeptide Y (NPY) homologue is present in the C. elegans genome. Because NPY has some sequence similarity to the FLPs, the de Bono group hypothesized that the NPR-1 ligand was a FLP and injected constructs for NPR-1 and an inwardly rectifying potassium channel into Xenopus oocytes; different FLPs were applied and receptor activation of the potassium channels was used as the readout (Rogers et al. 2003). Both Rogers et al. (2003) and Kubiak et al. (2003a) determined that NPR-1 binds to FLP-21; in addition, Rogers et al. (2003) found that peptides encoded by flp-18 also activated NPR-1. Animals carrying mutations in flp-21 display only mild aggregation defects compared to npr-1 mutants (Rogers et al. 2003; M. de Bono, personal communication; C. Li unpublished observations), presumably because FLP-18 ligands can functionally substitute for loss of FLP-21 (Rogers et al. 2003). As with the other FLP receptors, NPR-1 is promiscuous in its binding to multiple FLP ligands produced by different flp genes.

CONCLUSIONS

The number of potential neuropeptides in C. elegans is immense and rivals the numbers found in mammals thus far. Although some of the peptides may be invertebrate specific, the insulin-like and FMRFamide-related peptides have counterparts in mammals. If the current EST databases are any guide, parasitic nematodes are likely to share the enormous neuropeptide diversity of C. elegans. In contrast to the mammalian systems, the diversity of the neuropeptides comes from utilizing similar motifs repeatedly. The ~70 FLPs, for instance, all share a common C-terminal Arg-Phe-amide and the N-terminal sequences are often also very similar. Many challenging tasks await future studies. Specifically, the problem of determining the functions of individual peptides is as overwhelming as matching the different peptide ligands to specific receptors. Nonetheless, it is apparent that despite the relatively ‘simple’ behaviours of nematodes, there is a large collection of neuropeptides available to mediate or modulate these behaviours.