The neuronal genome of Caenorhabditis elegans^*

Calcium is a broadly used signaling molecule, but it also has several specialized functions in the nervous system, e.g., in synaptic vesicle release, in modulation of ion channel activity and, of course, as an ion that is itself involved in generating currents across excitable membranes in neurons. This is an absolutely critical feature of calcium since C. elegans does not generate sodium-based action potentials (Goodman et al., 1998). In this section, I will not only summarize calcium channels but also cover other genes related to “neuronal calcium”.

The molecular biology of neuronal calcium is briefly summarized in Figure 3 (Grienberger and Konnerth, 2012). Some calcium-permeant channels, namely nAChR-type receptors and glutamate receptors (NMDA, Kainate and AMPA-type), are discussed in an ensuing section (Section 2.5) and so are metabotropic receptors that signal to mobilize intracellular calcium stores (Section 5.1).

The TRP (Transient Receptor Potential) superfamily of cation channels are evolutionarily related to voltage-gated ion channels: they contain six transmembrane domains and a pore loop between the fifth and sixth transmembrane domains. However, TRP channels are generally not activated by voltage, but rather by a remarkable diversity of ligands or sensory inputs (Kahn-Kirby and Bargmann, 2006; Venkatachalam and Montell, 2007).

TRP channels fall into distinct classes based on overall sequence features (Yu et al., 2005) (Figure 2). The C. elegans genome contains 23 genes that display similarities to TRP channels. 17 of them are canonical TRP channels which fall into the TRPA, TRPC, TRPM, TRPML, TRPN, TRPV and TRPP subfamilies (Table 7) (Kahn-Kirby and Bargmann, 2006; Xiao and Xu, 2009) and one is a TRP-related TRPP1-type protein, LOV-1 (an 11-transmembrane domain protein). Five genes code for uncharacterized, multipass-transmembrane paralogs of a nematode-specific expansion (named trpl for TRP-like) (Table 7). trpl genes show relatively little sequence similarity to TRP channels, but do contain sequence signature motifs found in TRPM channels (Panther domain PTHR13800 “TRP, SUBFAMILY M”). The human genome encodes 28 TRP channel genes. Many of the C. elegans genes have been functionally analyzed and most are expressed in the nervous system. The so-far-characterized neuronally expressed TRP channels function as thermosensors, mechanoreceptors, proprioceptors or transduce signals in olfaction (Xiao and Xu, 2009).

Cyclic nucleotide gated (CNG) ion channels are signal-transducing cation-selective ion channels that form tetramers using specific combinations of α and β-type subunits. Even though they are not voltage-gated, they are members of the superfamily of voltage-gated ion channels (Yu et al., 2005) (Figure 2). C. elegans contains a total of six CNGs (Table 8). One, tax-4, encodes a canonical α subunit, whereas tax-2 encodes a canonical β subunit and both have been involved in various sensory paradigms (see Chemosensation in C. elegans). Vertebrates also contain six CNG channels, and all are α- or β-type. Four additional C. elegans CNGs, cng-1, cng-2, cng-3 and che-6, encode neither clear α or β subunits but display a somewhat higher sequence affinity to α subunits. The expression of most of the six CNGs has been investigated, revealing expression in partially overlapping subsets of sensory neurons.

As mentioned above, hyperpolarization-activated channels (HCNs) are related in sequence to the CNGs but, in contrast to flies and vertebrates, the C. elegans genome contains no HCN orthologs (Figure 2).

Neurotransmitters signal via two types of receptors: ion channels, also called ionotropic receptors (this section), and G-protein-coupled receptors (GPCRs), also called metabotropic receptors (Section 5.1 below).

Most ligand-gated ion channels (LGICs) in the C. elegans genome fall into the cysteine-loop superfamily of ion channels, which are characterized by the presence of a disulfide bond between two invariant cysteine resides in an extracellular loop region (Figure 1). Cys-loop LGICs consist of five homologous subunits arranged in a homomeric or heteromeric manner around a central pore (Sine and Engel, 2006). In mammals, the LGIC superfamily consists of about 45 genes, insects have just over 20 such genes, but the C. elegans genome contains 102 LGIC subunit-encoding genes (Jones and Sattelle, 2008) (Figure 4). Members of this C. elegans gene family include cation-permeable acetylcholine receptors related to vertebrate nicotinic acetylcholine receptors (nAChRs, see Section 2.5.1), anion-permeable GABA receptors related to vertebrate GABA_A receptors (Section 2.5.2), and glutamate-gated anion channels (Section 2.5.2) related to channels found widely in invertebrate species (Jones and Sattelle, 2008). In addition, C. elegans contains LGICs not yet identified in vertebrates and insects including anion channels gated by acetylcholine or biogenic amines (serotonin, tyramine, dopamine) (Ringstad et al., 2009), and possibly other ligands. Of the many additional orphan LGICs (all termed lgc genes) several fall into broad families, but it is unknown how they are gated (Jones and Sattelle, 2008).

A phylogenetic analysis of the LGIC superfamily from various nematode and non-nematode species (Rufener et al., 2010) reveals that the above-mentioned groups fall into two large blocks, as illustrated in Figure 4: a very large group of the nAChR-related genes (including vertebrate and C. elegans bona-fide nAChRs, as well as many “orphan” genes, Section 2.5.1) and a block of non-nAChR-type genes (Section 2.5.2). Characterized members of the former block are cation channels and characterized members of the latter block (with the exception of exp-1) are anion channels.

New types of ion channels are still being discovered. Two entirely new cation non-selective, plasma membrane channels with a >30 transmembrane topology were identified in 2010 in vertebrates, called Piezo1 and Piezo2 (Coste et al., 2010). More recently Piezo proteins were shown to be the pore forming unit of a new type of mechanoreceptor (Coste et al., 2012). These channels bear no obvious homology to any other type of ion channels. C. elegans contains a single ortholog of the Piezo family (T20D3.11 fused to C10C5.1). Given the Piezo family precedent it would not be surprising if more ion channels remain to be identified.

In summary, genome sequence analysis shows that the following types of ion channels are notably absent in C.elegans: voltage-gated sodium channels, glycine-gated ion channels, P2X channels, and HCN channels. Note that the absence of voltage-gated sodium channel is not generally indicative of an absence of classic action potential in C. elegans since all-or-none action potentials can, at least in C. elegans muscle, be generated by voltage-gated calcium channels as well (Gao and Zhen, 2011; Liu et al., 2011). In most cases the absence of the channel is considered to be a loss since it is paralleled by the absence of the channel in some but not all invertebrates (P2X channels, HCN channels, voltage-gated sodium channels). In one case (glycine-gated LGIC) the channel may have only originated in the vertebrate lineage (Tsang et al., 2007).

Acetylcholine (ACh) is synthesized from choline by the enzyme choline acetyltransferase (cha-1, Figure 6A). In vertebrates, ATP citrate lyase, which generates the CoA cofactor for the acetyl transfer reaction (Figure 6B), is broadly expressed but becomes largely restricted to cholinergic neurons in the mature nervous system. Two ATP citrate lyase orthologs (acly-1, acly-2) are encoded in the worm genome, both uncharacterized; perhaps one of them has specialized for its role in cholinergic neurotransmission, while the other may perform a more general metabolic function. In vertebrates, choline is not generated de novo in neurons but synthesized in the liver by a specific biosynthetic pathway and then taken up by neurons through the choline transporter ChT. Curiously, C. elegans, like plants and fungi, has a distinct pathway for choline synthesis, the PEAMT pathway, which allows neurons to generate choline cell-autonomously from phospho-ethanolamine (thereby lessening the importance of the ChT homolog cho-1 in C. elegans) (Brendza et al., 2007; Mullen et al., 2007). The key enzymes in this alternative pathway are pmt-1 and pmt-2 (Brendza et al., 2007).

GABA is synthesized from the amino acid glutamic acid by the enzyme glutamic acid decarboxylase (GAD), encoded by unc-25 (Figure 6A). Vertebrates contain several GAD isozymes, but C. elegans only contains one.

Glutamate metabolism in the vertebrate nervous system is remarkably complex and not well explored in C. elegans. Vertebrate neurons do not express a pyruvate carboxylase for de novo synthesis of glutamate but are rather provided with glutamate from support cells (astrocytes), which convert glutamate to glutamine via glutamine synthetase and then provide glutamine to neurons. Neurons then synthesize glutamate from glutamine using glutaminase. In C. elegans, there is one pyruvate carboxylase (pyr-1), 4 glutamine synthetases (gln-1, gln-2, gln-3, gln-5—a nematode-specific expansion) and 3 glutaminases (glna-1, glna-2, glna-3—again a nematode-specific expansion). Whether these biosynthetic enzymes are also differentially expressed in C. elegans neurons and putative support cells is not known.

Most of the synthesis pathways of monoamine neurotransmitters are multistep processes (summarized in Figure 7). For dopamine biosynthesis, tyrosine is first hydroxylated by tyrosine hydroxylase (TH, one gene in C. elegans, cat-2) to produce L-Dopa and for 5-HT synthesis, tryptophan is hydroxylated by tryptophan hydroxylase (TPH, one gene in C. elegans, tph-1) to produce 5-hydroxytryptophan (Figure 6A, Figure 7). Both TH and TPH require a cofactor, tetrahydrobiopterin (BH₄), which is generated through a multistep biosynthetic process (Figure 6C). Of the enzymes involved in this process (Figure 6C, Table 14), only the GTP cyclohydrolase encoded by the cat-4 locus has been analyzed to date in worms and is expressed exclusively in serotonergic and dopaminergic neurons, as expected. After their generation by TH and TPH, both L-Dopa and 5-hydroxytryptophan are then decarboxylated by the same amino acid decarboxylase (AAAD), encoded by bas-1, to produce dopamine and serotonin, respectively. Besides bas-1, there are four more genes in the genome that code for AAADs (Hare and Loer, 2004) (Table 14): one of them likely an inactive enzyme, two of them with unknown substrates and the last one being tyrosine decarboxylase (tdc-1) which is utilized to generate tyramine from tyrosine. In one of the two neuron classes that synthesize tyramine (i.e., express tdc-1), tyramine is converted by tyramine β-hydroxylase (tbh-1) to octopamine (Figure 6A, Figure 7).

Melatonin is a biogenic amine that can act as a neuromodulator in various species (Hardeland and Poeggeler, 2003). Melatonin has been detected in worms and is involved in regulating locomotory behavior (Tanaka et al., 2007). Melatonin is synthesized through the N-acetylation of serotonin by serotonin N-acetyltransferase (called AANAT for arylalkylamine N-acetyltransferase) (Figure 7). There are many N-acetyltransferases encoded in the worm genome and one of them, anat-1, is most closely related to AANAT (Migliori et al., 2011). anat-1 is expressed in several uncharacterized neuron types and is functionally also uncharacterized. N-acetylated serotonin is then converted into melatonin by hydroxyindole-O-methyltransferase (HIOMT, homt-1 in C. elegans), which is also neuronally expressed, but functionally uncharacterized (Tanaka et al., 2007).

Vesicular transporters for small-molecule neurotransmitters fall into the phylogenetically conserved SLC superfamily of solute carriers. The SLC18 family is called the “vesicular amine transporter family” (He et al., 2009) and contains the vesicular transporter for biogenic amines (dopamine, serotonin, tyramine, octopamine), encoded by cat-1, and the acetylcholine vesicular transporter, encoded by unc-17 (Table 5).

The SLC32 family (“Vesicular inhibitory amino acid transporter family”) contains as its sole C. elegans member the vesicular GABA transporter, encoded by unc-47. To be localized appropriately within GABA neurons, the UNC-47 protein requires the phylogenetically conserved LAMP (lysosome associated membrane proteins)-like protein UNC-46, which is exclusively expressed in GABA neurons (Schuske et al., 2007). Vertebrate UNC-46 homologs are also expressed in a neuron-type specific manner (David et al., 2007). UNC-46 is distantly related to the more canonical C. elegans LAMP proteins lmp-1 and lmp-2. There are no other obvious paralogs of UNC-46.

The SLC17 family of transporters contains several bona fide neurotransmitter transporters (Reimer and Edwards, 2004) and has significantly expanded in worms. The SLC17 family is subdivided into several subfamilies (Table 5). The vertebrate SLC17A6-8 subfamily is composed of vesicular glutamate transporters (VGluTs). C. elegans has three members of this subfamily: the well characterized eat-4 gene and two closely related, likely VGluTs, called vglu-2 and vglu-3. Mammalian genomes also encode three VGluTs, but the C. elegans genes represent an independent expansion. The vertebrate SLCC17A1-5 subfamily contains vesicular aspartate/glutamate transporters and C. elegans contains one uncharacterized homolog of this subfamily, C38C10.2 (Table 5). The SLC17A9 subfamily contains vesicular nucleotide transporters and C. elegans contains one homolog of this subfamily, vnut-1, which is also uncharacterized. Notably, however, there are no obvious homologs of ionotropic (P2X) or metabotropic (P2Y) purinergic neurotransmitter receptors and as such the substrate for vnut-1 is not clear. In addition, C. elegans contains nine genes (eight of which constitute a C. elegans specific expansion) that are clear SLC17 superfamily members, but show no homology to any specific SLC17 subfamily (Table 5).

Casting the web even wider and considering more divergent SLC17 family members, the expansion of the C. elegans family is even more obvious—there are at least 51 SLC17 members in worms and only nine in humans (Hoglund et al., 2011). Most of this expansion appears to be nematode-specific (see the 43 genes in TreeFam tree TF315412). The role of these additional members in the nervous system, if any, is unknown.

Recent work demonstrated the localization of the concentrative nucleoside transporter CNT2 (an SLC28 family member) on synaptic vesicle membranes in rat (Melani et al., 2012). It is therefore possible that the two worm CNT2 homologs slc-28.1 and slc-28.2 may be involved in vesicular uptake of adenosine (discussed more in Section 3.5).

Another substance present in neurotransmitter-containing vesicles is zinc. Zinc is used in a variety of distinct processes in many cell types but particularly notable is its presence in many glutamatergic vesicles in the mammalian nervous system (Bitanihirwe and Cunningham, 2009). Neurons with zinc-containing glutamatergic synaptic vesicles have been termed “gluzinergic” (Bitanihirwe and Cunningham, 2009). Synaptic zinc modulates the overall excitability of the brain through effects on voltage-gated calcium channels, glutamatergic, GABAergic, dopaminergic and nicotinic receptors (Bitanihirwe and Cunningham, 2009). Zinc is transported into synaptic vesicles through members of the SLC30 family of SLC carriers (10 in humans) (Lichten and Cousins, 2009). There are 12 SLC30 members encoded in the worm genome, six show sequence affinity with specific human SLC30 subtypes and six of them are diverse (Table 5). At least one of them is expressed in a subset of neurons (toc-1). There are two worm orthologs (cdf-2, ttm-1) of the human SLC30A2/3/4/8 subfamily which contains SLC30A3/ZnT3, the best characterized zinc synaptic vesicular transporter.

Members of the SLC transporter superfamily mediate the reuptake of a neurotransmitter once it has been released at the synapse (He et al., 2009). Depending on the neurotransmitter system, reuptake of the neurotransmitter (or a break-down product such as choline) occurs exclusively by the presynaptic cell, by adjacent cells, or by a combination of both.

The C. elegans genome contains homologs for all canonical reuptake transporters, based on both sequence and functional analysis (Table 11): one transporter for serotonin (mod-5, SLC6 family); one for dopamine (dat-1, SLC6 family); one for GABA (snf-11, SLC6 family); one for choline, the breakdown product of acetylcholine (cho-1, SLC5 family); and six glutamate plasma membrane transporters (glt genes, SLC1 family). The glutamate transporters are expressed in multiple distinct cell types (similar to the situation in vertebrates where glutamate is taken up not by the neuron but by surrounding tissue) (Mano et al., 2007), while the other transporters are mostly restricted in their expression to the neurons that have produced the transmitter (with the exception of the snf-11 GABA transporter which may be expressed only in a subset of GABAergic neurons) (Mullen et al., 2006).

However, this is not the full story. The SLC6 family, which is called “K⁺/Cl^- dependent neurotransmitter transporter family”, contains the serotonin, dopamine and GABA reuptake transporters as well as 14 additional members, three of them with low sequence similarity and perhaps not acting as transporters (Table 5). One of them, snf-6, codes for a muscle-expressed acetylcholine/choline transporter (Kim et al., 2004) and another, snf-12, is expressed in the hypodermis and involved in an immunity response (its cargo is unknown) (Dierking et al., 2011). The remaining genes are completely uncharacterized, but two of them are expressed in neurons and six of them represent a nematode-specific gene expansion (Table 5). Any of these genes may encode reuptake transporters for known neurotransmitters systems for which no plasma membrane reuptake transporter has yet been identified (e.g., biogenic amines like tyramine, octopamine or trace amines), or for uncharacterized neurotransmitters.

Aside from the Na⁺/Cl^- dependent neurotransmitter reuptake of monoamines by the SLC6 family, alternative monoamine uptake mechanisms are known to exist. The recently identified human plasma membrane monoamine transporter PMAT is a low-affinity, high capacity and Na⁺/Cl^- independent monoamine transporter (Engel et al., 2004) and is a member of the small family of SLC29 transporters (four mammalian members). Another SLC29 family member is also selectively expressed in the mammalian brain (Dahlin et al., 2009) and a knockout of a fly SLC29 family member shows various neurophysiological defects (Knight et al., 2010). There are seven members of the SLC29 family in the C. elegans genome (ent-1 through ent-7) (Table 5). ent-1 and ent-2 are expressed in non-neuronal cells and the remaining genes have not yet been characterized (Table 5).

Either before or after reuptake several neurotransmitters are degraded (schematically shown in Figure 6A). Acetylcholine is already broken down in the synaptic cleft by acetylcholinesterases (AChE, ace genes) before the breakdown product, choline, is taken up by the presynaptic, cholinergic neuron. While mammals only contain a single AChE gene, C. elegans contains four (Table 14). The four ace genes are expressed in cholinergic neurons as well as other cell types.

Monoamine neurotransmitters such as dopamine and serotonin are removed from the synapse by specific plasma membrane reuptake transporters, as mentioned above. In mammals, two monoamine oxidases (MAO-A and MAO-B) are important for subsequent degradation of monoamine neurotransmitters. The C. elegans genome encodes several proteins with homologies to MAO, with AMX-2 being the most similar to mammalian MAO-A and MAO-B (Table 14). In an alternative degradation pathway the catechol-O-methyltransferase COMT degrades monoamines. In contrast to the single enzyme in humans, there are 5 COMT-like proteins encoded in the C. elegans genome, all uncharacterized (Table 14). None of these genes have been functionally analyzed to date.

In insects MAO activity is weak and instead the major enzyme for monoamine breakdown is serotonin N-acetyltransferase AANAT (anat-1 in worms) (Tsugehara et al., 2007). Whether anat-1, which is involved in melatonin synthesis in worms (Migliori et al., 2011), is also employed for serotonin degradation is not known. anat-1 is expressed in multiple unidentified neurons.

The GABA degradation pathway consists of the enzymes GABA transaminase (GABA-T in vertebrates, gta-1 in worms) and succinic semialdehyde dehydrogenase (SSADH, alh-7 in worms) (Table 14). The genes encoding these enzymes have not been functionally analyzed to date in worms.

There are several neurotransmitters in other organisms whose existence in C. elegans is unclear, such as glycine, purines (mainly ATP and ADP), histamine and trace amines. Since electrophysiological methods for measuring neurotransmitter-induced or -modulated currents, are not readily applicable to C. elegans neurons, the only readily available route to identify neurotransmitter systems is to seek homologs of neurotransmitter pathway genes in the genome.

At present 122 neuropeptide genes encoding over 250 distinct neuropeptides have been identified in the C. elegans genome (Li and Kim, 2010). Of these, 40 genes encode insulin-like peptides (ins genes), 31 genes encode FMRFamide-related peptides (flp genes), and 51 genes encode non-insulin, non-FMRFamide-related neuropeptides (most encoded by the so-called nlp genes) (Li and Kim, 2010) (Table 15). Some of the nlp genes show similarities to neuropeptides of other species (Nathoo et al., 2001). There are likely more neuropeptides awaiting discovery, as evidenced by the slow trickle of newly discovered neuropeptides that initial analyses did not uncover. For example, recent proteomic analysis has identified a PDF peptide homolog (Janssen et al., 2009), genetic screens for behavioral mutants identified a previously uncharacterized neuropeptide, snet-1 (Yamada et al., 2010), and recent searches for C. elegans homologs of neuropeptides for which receptors are predicted in the worm genome (see Section 5.2 below) revealed a putative oxytocin homolog, ntc-1 (Beets et al., 2012; Garrison et al., 2012).

Expression patterns have been analyzed for more than 60 of the neuropeptide-encoding genes using reporter gene fusions. As summarized by Li and Kim, these studies show that all ins and flp genes examined and most of the nlp genes are expressed in the nervous system, with each gene having a unique expression pattern (Li and Kim, 2010) (Table 15). So far, more than half the cells in the nervous system express at least one neuropeptide gene but this number may increase because genes involved in neuropeptide release (e.g., unc-31) appear to be very broadly if not ubiquitously expressed in the nervous system. Furthermore, many individual neurons clearly express multiple neuropeptide genes.

Neuropeptides are produced as proproteins that contain several individual neuropeptides. These precursors are first cleaved by proprotein convertases, then the C-terminal basic residue is cleaved by a carboxypeptidase (Li and Kim, 2010). Afterwards most neuropeptides become further modified through amidation. There are four Kex2/subtilisin-like proprotein convertases (PCs) in the C. elegans genome (Table 16), one uncharacterized ortholog of a chaperone for the protease (sbt-1) and one carboxypeptidase E (egl-21) which has been explicitly linked to neuropeptidergic signaling (Li and Kim, 2010). In addition, the genome contains two, as yet uncharacterized homologs of the carboxypeptidase D family (cpd-1 and cpd-2), which are functionally related to the E family (Dong et al., 1999). These genes are the only representatives of the neuropeptide-processing carboxypeptidases in C. elegans, formerly called “regulatory carboxypeptidases” and now called M14B subfamily of metallocarboxypeptidases (TF315592) (see Merops database). C. elegans also contains a number of additional M14A-type carboxypeptidases (Merops database) but, in light of the function of their orthologues, they all likely act in digestion.

While egl-3 and egl-21 are well characterized in terms of function and expression (Kass et al., 2001; Jacob and Kaplan, 2003), much less is known about the amidation process. Studies in other organisms have shown that amidation involves the modification of a peptide-glycine processing intermediate by two enzymatic activities: a mono-oxygenase (called PHM) that associates with copper and molecular oxygen to hydroxylate glycine, and an enzyme (PAL) which cleaves the hydroxyglycine moiety to produce the peptide-amide (Prigge et al., 2000). In humans, both enzymatic activities reside in a single protein called PAM. In flies these two activities have split into two proteins (PHM and PAL). Curiously, the C. elegans genome contains genes for both versions: there is an as-yet-uncharacterized homolog of the multi-enzyme PAM protein encoded by pamn-1, as well as PHM and PAL orthologs, encoded by pghm-1 and pgal-1, respectively (Table 16). pamn-1 is expressed in the nervous system; the other genes have not been studied yet.

Analogous to classic neurotransmitter removal by reuptake or degradation, the activity of neuropeptides is also terminated through specific mechanisms. No reuptake mechanisms have yet been discovered for neuropeptides, rather signal-termination mechanisms work through the degradation of the neuropeptide by specific proteases. A number of different protease families have been implicated in this signal termination process, with the neprilysins being the best studied (Turner et al., 2001). Neprilysins are zinc metallopeptidases present on the outer surface of cells. The neprilysin family has significantly expanded in C. elegans, containing 27 members (Table 16), while there are only five neprilysin-like proteases in mammals. Two of the nep genes have been implicated in the execution of several distinct C. elegans behaviors (Spanier et al., 2005; Yamada et al., 2010).

Other proteases implicated in neuropeptide degradation (Isaac et al., 2009) are class III and class IV dipeptidyl peptidases (of which there are eight encoded in the C. elegans genome), the angiotensin-converting enzyme family (of which there is only one C. elegans homolog encoded by acn-1; the protein product is, however, predicted to be catalytically inactive), and tripeptidyl peptidase II (one C.elegans gene, tpp-2) (Table 16).

The most prevalent class of neuropeptides receptors are 7TM G-protein-coupled receptors (GPCRs). These types of receptors and their relationship to neuropeptide signaling will be discussed in the ensuing GPCR section (Section 5). However, GPCRs are not the only neuropeptide receptors, as exemplified by the neuromodulatory insulin peptides. The insulin-like peptide encoded by ins-1 acts in a neuromodulatory manner to affect the local activity in a chemosensory circuit (Tomioka et al., 2006). As mentioned above, there are 40 insulin-encoding genes, and all genes examined so far are expressed in restricted domains within the nervous system, suggesting neuronal functions for many other insulins as well. DAF-2 is the only insulin/IGF receptor-like tyrosine kinase in the C. elegans genome and several insulin-like peptides, such as ins-1 are known to signal through DAF-2 (Tomioka et al., 2006). Intriguingly, the C. elegans genome also contains an unusual types of insulin receptor-related proteins (Dlakic, 2002), encoded by the irld genes (for insulin/EGF receptor L domain). Unlike DAF-2, these proteins do not contain tyrosine kinase domains but only contain extracellular Cys-rich L domains (related to LRR domains), found in insulin and EGF receptors (IPR000494). Like the insulin-encoding ins genes, genes for these type of putative insulin binding proteins have vastly expanded in the worm genome (69 genes) (Table 17). Many of these genes are recent duplicates. One third of these proteins contain transmembrane domains or membrane anchors and may be part of atypical insulin receptor complexes. L domain-only proteins cannot be found in Drosophila or vertebrate genomes (Dlakic, 2002).

Other types of neuropeptide receptors exist. The currents of ENaC/DEG/ASIC channels from snails and vertebrates are modulated by FMRFamide and related neuropeptides (Askwith et al., 2000; Jeziorski et al., 2000). Whether some C. elegans ENaC/DEG/ASIC-channels are gated by FMRFamides is not known. In any case, it should be kept in mind that non-GPCRs may act as receptors for other neuropeptides as well.

The C. elegans genome contains three GPCRs for ACh (gar-1, gar-2, gar-3), two for GABA (gbb-1, gbb-2), and three for glutamate (mgl-1, mgl-2, mgl-3). These receptors can be clearly identified based on sequence features (Table 19). The metabotropic glutamate receptors mgl-1, mgl-2 and mgl-3 fall into the three ancestral groups of mGluRs (Kuang et al., 2006). In addition, there are two more, as-yet-uncharacterized GPCRs in the genome, C30A5.10 and F35H10.10, that show significant similarities to mGluRs and display an annotated “Ligand-binding domain of metabotropic glutamate receptor” annotated by the CCD domain database. F35H10.10 contains signature motifs of the GPCR family 3 (also called family C, IPR000337).

The ACh GPCRs fall into the into the Rhodopsin class of GPCRs (like most other GPCRs), while the GABA and Glu GPCRs fall into the Glutamate receptor class (class C) of GPCRs (IPR000337). Aside from the above-mentioned genes there are no other Class C family members in the worm genome.

C. elegans contains 16 GPCRs for monoaminergic neurotransmitters (Table 19). These receptors fall into the Rhodopsin or “Class A” family of GPCRs. Four biogenic amines have been identified as neurotransmitters in C. elegans so far: serotonin, dopamine, tyramine and octopamine (as mentioned above, there may be more). Several but not all of these receptors can be assigned to specific ligands by sequence alone. Biochemical and functional analysis has assigned most of the receptors to specific ligands as shown in Table 19. Reporter-based expression patterns exist for all 16 families members, revealing their expression in a restricted set of neurons. Several of the receptors are also expressed in muscle cells. In addition to the 16 genes encoding aminergic GPCRs, there is also another gene, C24A8.6, that likely arose through a local inversion of the neighboring dop-6 gene. Parts of this gene are identical to dop-6, but as the identity is limited to one restricted part of the protein, C24A8.6 is unlikely to form a functional GPCR. More information on biogenic amines and their receptors can be found in Biogenic amine neurotransmitters in C. elegans.

As mentioned above, even though it is not clear whether adenosine can generally be classified as a neurotransmitter in any system, a clear ortholog of a GPCR-type adenosine receptor, ador-1, is encoded in the worm genome but has not yet been characterized in terms of expression or function.

Apart from some known exceptions described in Section 4.3, neuropeptides generally signal through GPCRs. Neuropeptide GPCRs are mostly of the Rhodopsin class, but also the Secretin class (Table 18). The ability to predict neurotransmitter receptors by sequence (i.e., to distinguish them from other types of GPCRs, such as olfactory GPCRs) is not as straight-forward as it is for metabotropic receptors of classic neurotransmitters. Nevertheless, some basic sequence features are conserved enough in neuropeptide receptors to assess the number of neuropeptide receptors in the worm genome as was done in several previous studies (Keating et al., 2003; Wenick and Hobert, 2004; Janssen et al., 2010). By revisiting these previous datasets and supplementing them by additional analysis, a list of 153 genes can be assembled whose products share significant homology to known neuropeptide receptors (Table 20, see legend for methodology). This list includes the 14 C. elegans neuropeptides receptors that have been biochemically deorphanized as FLP or NLP receptors (Li and Kim, 2010). The cutoff for significance was set to BLAST scores of 1e-04 to exclude known non-neuropeptide receptor GPCRs (three deorphanized true sensory receptors, odr-10, srbc-64, and srg-37, and one likely sensory receptor, str-2, were used to set this threshold as they all have e values larger than 1e-04).

About one quarter of the 153 putative neuropetide receptors stand out in their obvious relation to vertebrate neuropeptide receptor families, such as the neuropeptide Y receptor family (at least 12 members in C.elegans), the neuromedin receptor family (at least 6 C. elegans genes), the neurokinin/neuropeptide FF/orexin receptor family (6 genes), the somatostatin family (6 genes), the galannin receptor family (4 genes), and the gonadotropin-relasing hormone receptor family (8 genes) (Cardoso et al., 2012) (Table 20). Most of the remaining C. elegans neuropeptide receptors cluster into related and likely paraologous families, some of them quite large. For example, one family contains 20 genes, termed frpr genes, and clusters with the Drosophila FMRFamide receptor FR (TF316702). Two of the family members have been shown biochemically to be FLP receptors (Li and Kim, 2010). However, in spite of its similarity with other members of this family, one family member (daf-37) was recently reported to be a sensory receptor for ascarosides, glycolipids that worms use to communicate with one another (Park et al., 2012). Another family with 17 genes, the dmsr family, clusters with Drosophila dromyosuppressin receptor Dms-R2 (TF315509), which serves as a receptor for a FMRFamide peptide (Klose et al., 2010). Indeed one of the worm family members, EGL-6, has been shown to be a receptor for FLP-10 and FLP-17 (Ringstad and Horvitz, 2008). Members of another family (TF315321) contain low similarity to fly FMRFamide receptor, but the degree of similarity is consistent across most family members. Many of the remaining putative neuropeptide receptors also fall into smaller families and two of them are known to be activated by NLP peptides (Table 20). Some of them are obvious sequence orthologs to fly or vertebrate receptors (Table 20). Three of them are secretin-type (formerly class B-type) neurotransmitter receptors, pdfr-1, seb-2 and seb-3 (for secretin/class B-type receptor). A recent analysis of GPCRs in the C. elegans genome comes to a similar conclusion (Frooninckx et al., 2012).

Almost all of the >20 candidate neuropeptide receptor-encoding genes for which an expression pattern has been examined show expression in a restricted number of neurons (Table 20). Given that many neuronally expressed peptides modulate transmission at neuromuscular junctions or contraction of other organ types (e.g., the gut (Nichols et al., 2002)) it is likely that many of the receptors will also be expressed in muscle or other non-neuronal cell types.

However, there are possibly many more than the 153 neuropeptide receptors described above. As Robertson and Thomas note in The putative chemoreceptor families of C. elegans, the divergent srw gene family, which consists of 100 members, is related to neuropeptide receptors. A BLAST analysis of representative genes from individual srw subfamily branches illustrates this notion (Table 21). Since almost 90% of srw genes reside in large clusters on the arms of chromosome V, they likely represent a nematode-specific expansion. Whether they serve to monitor internal signals or serve as receptors for environmental peptides, as suggested by Robertson and Thomas, remains to be seen. In addition, several members of the srsx family of GPCRs (37 genes) show significant BLAST hits to neuropeptide receptors. Any of those GPCRs could of course also be ligands for other chemical substances such as lipids or other small organic molecules. That this is more than just a mere possibility is illustrated by daf-37 and daf-38, two GPCRs recently reported to be receptors for ascarosides (as mentioned above, these are glycolipids that worms use to communicate with one another) (Park et al., 2012). Both daf-37 and daf-38 display significant homology to neuropeptide receptors (Table 21), illustrating the diversity and unpredictability of ligands for GPCRs.

Robertson and Thomas identified ~1,280 likely chemoreceptor genes in the genome, which they classify into several groups in Table 1 of The putative chemoreceptor families of C. elegans. The evidence for the function of these GPCRs as chemoreceptors is very strong: First, for the many dozen genes for which expression profiles have been examined, expression is clearly observed in sensory neurons (Troemel et al., 1995; Colosimo et al., 2004; Chen et al., 2005). Second, several of the receptors are localized to the sensory dendritic endings of chemosensory neurons where the chemosensory apparatus is thought to be concentrated (Dwyer et al., 1998; Colosimo et al., 2004). Third, since the discovery of odr-10 as a receptor for the odorant diacetyl (Sengupta et al., 1996; Zhang et al., 1997), four additional genes (srbc-64, srbc-66, srg-36, srg-37) have been shown to code for receptors for ascarosides (Kim et al., 2009; McGrath et al., 2011). Fourth, as pointed out by Robertson and Thomas (The putative chemoreceptor families of C. elegans), it is hard to imagine what else other than sensory receptors could be encoded by such a large gene family.

The sensory modalities of the sensory GPCRs may be diverse. Apart from small chemicals (e.g., diacetyl for ODR-10), some of the GPCRs may also be light detectors (Edwards et al., 2008; Liu et al., 2010). Sensory modalities are difficult to predict, for example, the UV receptor lite-1 is a member of a small subfamily of worm GPCRs (five gur genes) that were initially recognized based on their similarity to Drosophila gustatory receptors (Robertson et al., 2003).

It is noteworthy that a substantial number of the chemosensory GPCR family members are not only expressed in sensory neurons, but also in distinct sets of interneurons and motorneurons (a few samples are shown in The putative chemoreceptor families of C. elegans). In fact, some are only expressed in interneurons, e.g., sra-11 (Troemel et al., 1995). Since more than one third of so-far-analyzed chemosensory GPCRs are expressed in non-sensory neurons (Troemel et al., 1995; Chen et al., 2005), it is to be expected that hundreds of the total of ~1,280 receptors may be expressed in many different parts of the nervous system. These genes may code for receptors that monitor neuronal or non-neuronally-derived internal signals. These signals could range in their composition from peptides to small organic substances to lipids, all of which are recognized by GPCRs in other systems. These signals may be highly species-specific. For example, there are no worm homologs for a good number of vertebrate GPCRs known to sense different types of lipids (Kostenis, 2004). Additionally, there are no homologs of cannabinoid GPCRs (key enzymes in endocannabinoid synthesis are also missing), no homologs of the LPA receptor, and no homologs of the FFA (free fatty acid receptor) GPCRs. Worms may use different types of lipids for signaling.

Together with a small number of secretin-type hormone receptors (three in C. elegans), this family used to be part of the class B family of GPCRs, but is now recognized as its own family (Lagerstrom and Schioth, 2008) (Table 18). The signature features of adhesion GPCRs are (1) an extended N-terminus that contains distinct sets of domains involved in protein-protein interactions (hence their presumed role in cell adhesion), (2) a signature GPS domain involved in autoproteolytic cleavage of the N-terminal extracellular domain, and (3) the 7TM part which is related to the secretin/hormone-type GPCRs.

While humans contain diverse set of proteins in the adhesion group, each with distinct extracellular domains (33 in total) (Lagerstrom and Schioth, 2008) and diverse function (Yona et al., 2008), C. elegans only contains five members of this group (Table 22). Three of them are obvious orthologs of well-characterized fly and human proteins: fmi-1 (Flamingo ortholog), lat-1, and lat-2 (latrophilin orthologs). fmi-1 function in neuronal development and synapse formation (Steimel et al., 2010; Najarro et al., 2012) and lat-1 has a role in synaptic transmission (Willson et al., 2004). C. elegans contains single homologs of proteins that Latrophilins are thought to interact with, neurexin (nrx-1 in worms) and teneurin-1 (ten-1 in worms), both with presumed functions in synapse formation.

Methuselah is the founding member of a GPCR subfamily that expanded specifically in Drosophila and is considered a third subgroup of the former “class B” class of GPCRs (Harmar, 2001). Methuselah has documented functions in synaptic transmission (Song et al., 2002) and can bind various neuropeptides (Ja et al., 2009). Based on similarity identified in the Panther database (PTHR12011), there are two genes in C. elegans related to Methuselah's, mth-1 and mth-2. In contrast to fly Methuselah proteins, the two worm proteins contain the signature GPS domain which clearly places them in the adhesion group of class B GPCRs. The function or expression of mth-1 and mth-2 have not yet been examined.

There are no Taste2 GPCRs in the C. elegans genome, but there are four Frizzled-type GPCRs (Table 18), which are receptors for Wingless-type signaling molecules (of which there are five in the C. elegans genome). In both C. elegans and other organisms, there is abundant evidence for the role of Frizzled signaling in neuronal development (for a review of the C. elegans work, see Wnt signaling. In flies and vertebrate, there is evidence for a role of Frizzled in synapse function, particularly in the form of anterograde and retrograde feedback signals at the synapse (Speese and Budnik, 2007). In worms, the evidence for adult neuronal functions of Frizzleds is the dependence of AMPA receptor localization on a Frizzled effector gene, β-catenin (Dreier et al., 2005) and the observation that reporters for the two Frizzled receptors mom-5 and cfz-2 are expressed in a subset adult neurons. The expression or function of other Frizzleds in the adult nervous system has not yet been reported.

GPCRs usually signal through heteromeric G-protein complexes by acting as nucleotide exchange factors (signaling downstream of Frizzled GPCRs tends to be more complex (Speese and Budnik, 2007)). There are 21 Gα–, two Gβ–, and two Gγ-encoding genes in the C. elegans genome (see Heterotrimeric G proteins in C. elegans), compared to 21, 5 and 19 genes in humans (Table 23). Three Gα-(gsa-1, egl-30, goa-1), both Gβ and one Gγ-encoding gene (gpc-2) are very broadly expressed, most of the remaining 18 Gα and one Gγ-encoding gene (gpc-1) are expressed in a highly neuron-type specific manner, most of them in sensory neurons (see Heterotrimeric G proteins in C. elegans).

The activity of GPCRs is controlled by a number of regulatory factors, including G protein coupled receptor kinases (GRKs), of which there are two in the worm genome (grk-1 and grk-2) (Table 23), and arrestin. Arrestins regulate the inactivation, internalization and trafficking of GPCRs (Gurevich and Gurevich, 2006). Mammalian genomes encode four conventional arrestins, which are either specifically expressed in visual systems (vertebrate cone arrestin and rod arrestin) or are broadly expressed (arrestins 2 and 3). There is one conventional arrestin in the C. elegans genome, called arr-1 (Table 23).

Intriguingly, a functionally uncharacterized family of arrestin-related proteins (now called α-arrestins) has been recognized in genomic sequences across phylogeny (Alvarez, 2008). Their overall sequence homology to classic arrestin is relatively low, but their predicted overall structure appears highly related. Classic arrestin is composed of two modules with antiparallel β-sheets (Arrestin-N and Arrestin-C domain) which are similar to an Fn3 module (Aubry et al., 2009). The arrestin-related proteins (called ADCs or ARRDCs for “Arrestin domain containing”) share this structure (Aubry et al., 2009). There are five of these ARRDC proteins in humans, but C. elegans has vastly expanded its repertoire of this type of protein—it contains 31 arrd genes coding for ARRDC proteins. As listed in Table 23, expression of the three arrd genes whose expression has been analyzed to date was detected in subsets of neurons. It is intriguing to think about this family expansion in the context of expansion of the worm repertoire of GPCRs, as well as G proteins (above). The limited number of neurons in the worm means that individual neurons likely express dozens of GPCRs. To distinguish between activation of different GPCRs (and to therefore permit discrimination between distinct inputs), GPCR subfamilies may be able to hook up with distinct downstream signaling molecules to produce distinct signaling outputs.

GPCR function is also regulated by recently discovered transmembrane proteins of the RAMP family and by the GASP proteins (Magalhaes et al., 2012); there are no obvious homologs of either type of protein encoded in the worm genome.

Heterotrimeric G-protein signaling is controlled by various regulatory factors, including guanine nucleotide exchange factors (GEFs) and guanine nucleotide dissociation inhibitors (GDIs) proteins. Apart from the GPCRs themselves, there is at least one other dedicated GEF encoded in the worm genome, ric-8, and there are three GDI proteins each of which contain “GoLoco” domains (Table 23). A large family of G-protein regulators are encoded by the rgs genes (“regulator of G protein signaling”) (Porter and Koelle, 2009). There are 21 members of this family encoded in the worm genome (compared to 38 in humans), many with either broad or cell-type specific expression in the nervous system and specific roles in nervous system function (Porter and Koelle, 2009) (Table 23).

Guanylyl cyclases generate a second messenger molecule, cGMP, which is predominantly used in the nervous system of C. elegans, as assessed by the neuron-type specific expression of most guanylyl cyclases (Table 24). Generally, guanylyl cyclases exist either as soluble, cytoplasmic versions (sGCs) or in single pass transmembrane versions with large extracellular domains (rGCs) (Potter, 2011). There are 27 gcy genes in the C. elegans genome coding for rGCs and seven gcy genes coding for sGCs (Table 24), compared to five and four genes in human, respectively (Fitzpatrick et al., 2006; Ortiz et al., 2006).

Members of both families are thought to be receptors for small ligands In C. elegans. Molecular oxygen is the ligand for sGCs, with different sGCs being tuned to detect different oxygen concentrations (Gray et al., 2004; Zimmer et al., 2009). sGCs are exclusively expressed in sensory neurons (all likely oxygen-sensory neurons).

rGCs are expressed in many sensory neuron (where several of them localize to sensory dendrites), but are also expressed in non-sensory neurons. Specific functions have been identified in non-sensory neurons (Shinkai et al., 2011), suggesting that rGC function is not restricted to sensing environmental cues. No specific ligands have yet been identified for any C. elegans rGC proteins, but there are many diverse candidates: (1) small peptides, in analogy to the ligands of mammalian GCY proteins (Potter, 2011); (2) salt ions, based on their role sensory taste perception (Ortiz et al., 2009); and (3) other small organic molecules (based on the presence of the “extracellular ligand-binding receptor domain”, IPR001828, which can also be found in metazoan glutamate receptors and bacterial amino acid transporters).

rGCs may also be activated by CO₂ (Hallem et al., 2011; Brandt et al., 2012). In contrast to the other putative ligands of rGCs which likely act through the extracellular domain, CO₂ sensing may work through the intracellular catalytic domain, in analogy to vertebrates rGCs (Potter, 2011). This process may entail the conversion of CO₂ into bicarbonate, catalyzed by carbonic anhydrases (see below, Section 7).

GCAPs (guanylate cyclase-activating proteins) and GCIPs (guanylate cyclase-inhibitory proteins) are calcium-binding, EF hand proteins that control GCY activity in the phototransduction pathway in the vertebrate retina (Palczewski et al., 2004). The closest relatives to the GCAP and GCIP proteins in the C. elegans genome are the NCS proteins NCS-1, NCS-2 and NCS-3 (Table 6). The expression of ncs-1 in many sensory neurons matches the expression of receptor-type gcy genes.

Aside from cGMP-gated ion channels (CNG) discussed above, other critical neuronal effectors of cGMP are cGMP-dependent protein kinases, of which there are two encoded in the worm genome: egl-4, which is expressed in the nervous system and has various nervous-system-associated functions (see Chemosensation in C. elegans); and pkg-2, which is presently uncharacterized.

cGMP levels are controlled by phosphodiesterases. The C. elegans genome encodes six of these enzymes (pde-1 through pde-6), each of which has a closely-related human ortholog (Conti and Beavo, 2007) (Table 24). Based on this orthology, pde-4 and pde-6 are cAMP-specific while pde-1, pde-2, pde-3 and pde-5 control cGMP levels. pde-1, pde-2 and pde-3 are expressed in the nervous system; the expression of other genes has not been investigated. Six pde genes do not reflect a nematode expansion as there are more than 20 human pde genes; the clear one-to-one orthology observed for human and worm PDEs (Conti and Beavo, 2007) is rarely seen in other gene families analyzed in this paper.

Besides several broad cellular roles, members of the kinesin, dynein, and myosin superfamily molecular motors have specific roles in neuronal function, plasticity, and morphogenesis (Hirokawa et al., 2010). Kinesins have been divided into specific subfamilies, several of which carry out essential functions in all dividing cells, while the most relevant for nervous system development are the subfamilies involved in synaptic vesicle transport and intraflagellar transport (see below and also The sensory cilia of Caenorhabditis elegans). 21 kinesin-like proteins are encoded in the C. elegans genome (Table 29), some characterized (e.g., unc-104, osm-3, klp-6), some completely uncharacterized. In addition, C. elegans contains three atypical kinesins, one with a well-documented function in axon pathfinding (vab-8).

Cytoplasmic dyneins consist of heavy chains, light intermediate chains, intermediate chains, and light chains (which fall into three further subfamilies). C. elegans contains representatives for each family, with some of the families being notably expanded (Table 30).

Lastly, the C. elegans genome contains 18 genes coding for proteins with myosin motor domains, including classic muscle myosin, but also several genes that are expressed in neurons (Table 31).

Cilia are microtubule-containing organelles that emanate from the surfaces of most animal cells. There are two types of cilia: motile cilia used for locomotion or for the generation of fluid flow, and non-motile (primary) cilia, which are implicated in sensing the environment (see The sensory cilia of Caenorhabditis elegans). Unlike many organisms, including humans, sensory neurons are the only ciliated cell types in C. elegans. 60 of the 302 hermaphroditic neurons are sensory neurons that possess non-motile, primary cilia, many with a striking morphological diversity (http://www.wormatlas.org).

In order to build ciliated structures and transport specialized functional components (such as receptors and ion channels) into this structure, a sophisticated anterograde and retrograde transport machinery exists that traffics substrates from the so-called transition zone to the distal ends of the ciliated dendrites (The sensory cilia of Caenorhabditis elegans). Specific types of kinesins provide the anterograde motor activity, and specific dyneins the retrograde motor activity (Table 32). The intraflagellar transport complexes that contain the motor proteins contain a substantial number of additional proteins that fall into three separate modules, the IFT-A, IFT-B, and BBSome modules (amounting together to a total of 27 proteins). A “parts list” of these complexes was recently put together by Inglis et al. (Inglis et al., 2009) and is summarized in Table 32. Differential regulation of individual components of this core set of ciliary genes is thought to be at least in part responsible for building distinct types of cilia (Mukhopadhyay et al., 2007; Silverman and Leroux, 2009). More genes involved in building specific types of ciliated endings likely remain to be identified.

Members of the immunoglobulin superfamily are involved in axon pathfinding (e.g., unc-40/DCC, sax-3/Robo), synapse formation (syg-1, syg-2), neuronal axon and soma adhesion (sax-7), axonal maintenance (zig genes) and neurotransmitter receptor clustering (oig-4). Essentially all immunoglobulin superfamily members examined so far are expressed in a subset of neurons (Aurelio et al., 2002; Schwarz et al., 2009). The immunoglobulin superfamily of C. elegans has been described previously (Vogel et al., 2003; Hobert et al., 2004). Excluding intracellular Ig domain proteins (such as those encoded by unc-22, unc-89, dim-1, or unc-73), there are a total of 64 proteins that contain one to many copies of easily recognizable subtypes of Ig domains (Table 33). 18 of these proteins contain no other obvious protein domains, are not necessarily closely related to one another and are small secreted or transmembrane proteins of the zig and oig type. Many of the remaining proteins contain additional fibronectin-III domains, which are related to Ig domains, and some contain distinct complements of domains, some of which are implicated in cell adhesion, others in signaling (Table 33). In contrast to many other gene families discussed here, the Ig domain family has not expanded in worms, but has expanded in humans which have many hundreds of immunoglobulin superfamily members.

Notably absent from the C. elegans genome is a homolog of the immunoglobulin superfamily member DSCAM, a neuronal recognition protein with remarkable isoform diversity in flies (Hattori et al., 2008). There are also no obvious orthologs of mouse SynCAM proteins, although proteins with similar domain architecture exist (igcm-3, igcm-4). There is a worm homolog of vertebrate Sidekick (rig-4), implicated in synaptic targeting in vertebrates (Schwarz et al., 2009). Receptor tyrosine phosphatases (all containing extracellular Ig domains) have been implicated in various aspects of neuronal development (Paul and Lombroso, 2003). There are three of these genes in the C. elegans genome (Table 33), one of them the LAR ortholog ptp-3 in worms, which has been implicated in synapse maturation in several different species including worms (Stryker and Johnson, 2007).

Members of the LRR family include the axon guidance cue Slit/slt-1 and several vertebrate proteins involved in synapse formation and function (de Wit et al., 2011). Extracellular LRR (eLRR) proteins (i.e., either secreted or transmembrane) in worms have been analyzed in silico (Dolan et al., 2007). 29 eLRR protein-encoding genes can be found in the worm genome (Table 33). Most are exclusively composed of LRR repeats, some also contain Ig domains. Some of the proteins are secreted, but most C. elegans eLRR proteins have transmembrane or GPI anchors. A number of them have highly conserved vertebrate orthologs, such as the guidance cue slt-1, the Toll-like receptor protein tol-1 or neuropeptide receptor fshr-1. Even though there are no obvious orthologs of the vertebrate LRR synaptic adhesion molecules (LRRTMs, SALMs and NGLs), there are C. elegans proteins with similar domain architectures (multiple LRR domains and a transmembrane domain, or multiple LRR and Ig domains) (Table 33).

Many of the eLRR family members have been analyzed for expression in the nervous system, revealing neuronal expression for many of them (Liu and Shen, 2011) (Table 34). Notably, in contrast to many other gene families discussed here, the LRR family has, like the Ig domain family, not expanded in worms. There are almost five times as many eLRR proteins in mouse and humans (Dolan et al., 2007).

The C. elegans genome contains 13 genes that code for proteins with cadherin domains, one more than previously noted (see The cadherin superfamily) (Table 34). This is significantly less than the more than 100 cadherin and cadherin-related genes in mammals. Representatives of several ancient cadherin subgroups can be found in C. elegans, including Flamingo, FAT and Dachous-type cadherins as well as more classic cadherins, yet there are also a number of nematode-specific cadherins, which are mostly uncharacterized (The cadherin superfamily) (Table 34).

There are no protocadherins encoded in the fly or worm genome (The cadherin superfamily). DSCAM in Drosophila and the protocadherins in mammals are the two classes of diversely spliced cell-cell recognition molecules in metazoan nervous systems (Zipursky and Sanes, 2010). The absence of both types of isoform-rich molecules in the worm genome may be a testament to the reduced morphological complexity of its nervous system.

Vertebrate neurexins are synaptic proteins that interact with a set of distinct partners to help synapses mature and function appropriately. C. elegans contains a single neurexin gene, nrx-1, that is broadly expressed in the nervous system but not yet functionally characterized (Table 35). There is presently no evidence in transcriptome datasets that the nrx-1 locus produces anything close to the tremendous amount of alternatively spliced isoforms that are characteristic of vertebrate neurexins (Missler and Sudhof, 1998). Neurexin-like genes of the CASPR family can also be found in the worm genome (Haklai-Topper et al., 2011) (Table 35).

The C. elegans genome contains orthologs of several neurexin binding proteins, including neuroligin (nlg-1), a neuroligin-related protein (glit-1) and two latrophilin genes (lat-1 and lat-2). There are no obvious orthologs of LRRTM1/2, synaptic adhesion molecules that interact with Neurexin, but proteins with similar domain architectures can be found, as mentioned above in the LRR section (Table 34). Clear orthologs of other neurexin binding partners (LRRTM1/2, Cbln1, Neurexophilins) cannot be found in the worm genome (Table 35).