1. Trapping and Transport of Bacteria in the Pharyngeal Lumen

Caenorhabditis elegans is a filter-feeder: The worms take in liquid with suspended particles (bacteria) and then spit out the liquid while retaining the particles. This separation, accomplished by the corpus and anterior isthmus, is poorly understood. In a microscopic system such as the pharynx, fluid motions are reversible and linearly related to force (Purcell 1977). If the motions of the pharyngeal muscles during relaxation are the reverse of those during contraction, fluid and particle motions should also be reversed. Thus, if corpus and isthmus muscles contracted and relaxed in synchrony (Fig. 3a), bacteria that came in through the mouth should exit through the mouth. In fact, they are trapped. This implies a complexity in the muscle motions not obvious to real-time visual inspection.

Relaxation is the key step in filtering. During contraction, bacteria move posteriorly with the fluid as one would expect. Relaxation expels the fluid, but somehow the bacteria remain trapped (Avery 1993b). The mechanism of trapping is not known, because the relaxation is so fast (<17 msec; Avery 1993a) that the motions cannot be seen in videotapes. The speed of relaxation is likely to be important, since the pharyngeal motor neuron M3, which regulates relaxation, is important for effective trapping of bacteria (Avery 1993b).

Seymour et al. (1983), analyzing pharyngeal muscle motions in ciné films, reported that the motions of the anterior isthmus are delayed compared to the corpus. Subsequent analysis of videotapes with better time resolution confirmed that isthmus contraction begins when corpus contraction is well advanced and that isthmus relaxation takes place well after corpus relaxation has ended (L. Avery, unpubl.). Seymour et al. (1983) also reported that procorpus motions precede those of metacorpus, but this observation could not be confirmed on videotapes (L. Avery, unpubl.; L. Philipson, pers. comm.). This delay of the isthmus with respect to the corpus is interesting because it means that the motions of the pharyngeal muscles during relaxation are not the reverse of those during contraction, a theoretical requirement for trapping. Seymour et al. (1983) also described perplexing motions of the metastomal flaps. These flaps, operated by the pharyngeal pm1 and pm2 muscles, prevent bacteria from entering the pharynx during most of the corpus contraction. What function this serves is unknown.

There is clearly a great deal left to be learned about how these worms catch food. Laser ablation of the pm1 and pm2 muscles might reveal the role of the metastomal flaps in this process. High-speed video recordings of the motions of individual particles in the lumen would be useful for understanding trapping. Computer or physical models of fluid flow within the pharyngeal lumen could test the importance of the metastomal flaps and the timing of muscle motions.

2. Terminal Bulb Function

The terminal bulb grinds up food and passes the debris to the intestine. The pm6 and pm7 muscle cells secrete a thick, ridged cuticle called the grinder on their lumenal surfaces. The three segments of the grinder (made by the three pairs of muscle cells) engage one another. Contraction of the muscles rotates the segments (usually all three together, although occasionally individual segments turn); food caught between them is ground up and passed back to the intestine through the pharyngeal-intestinal valve, which is opened at this time (Doncaster 1962), probably by contraction of pm8. Relaxation of the terminal bulb has no active purpose—it merely returns the grinder to its resting position.

3. Isthmus Function

We believe that the isthmus acts as a double valve which separates high- and low-pressure regions of the pharynx, allowing food to be forced into the intestine against the pressure gradient. Our model is shown in Figure 4. At rest, the pharyngeal lumen is closed except for a small region around the grinder, which is at high pressure (Fig. 4a). During a pump (Fig. 4b), the lumen of the corpus and anterior isthmus are at ambient pressure, open to the outside fluid. The lumen around the grinder is at high pressure, connected to the intestinal lumen by the open pharyngeal-intestinal valve. These two regions are separated by the closed posterior isthmus. Near the end of the peristalsis (Fig. 4d), the isthmus lumen is connected to the terminal bulb lumen and is at high pressure.

Figure 4

Pressure differences during feeding. At rest, the pharyngeal lumen is closed except for a small region around the grinder, which is at high pressure (a). During a pump (more...)

The movement of the bacteria occurs within isobaric regions. The region from the mouth to the middle of the isthmus is at low pressure during the pump, the region from the isthmus and terminal bulb is at high pressure during isthmus peristalsis, and the region from the terminal bulb to intestine is at high pressure during the pump. Food is transferred from low to high pressure between the end of the pump and the beginning of the isthmus peristalsis (i.e., between b and c in Fig. 4), as the anterior isthmus closes. This transfer occurs without motion of the food and therefore requires little energy. Most of the work done to move the bacteria against the pressure gradient is done when the corpus lumen opens against the internal/external pressure difference. This is analogous to the way a diver leaves a submarine through a lock. The diver moves from the submarine into the lock at low pressure, closes the hatch to the submarine, stands still while the hatch to the ocean is opened and the pressure rises, and then moves from the lock to the ocean at high pressure. The work necessary to move the diver from low to high pressure is done when the water is pumped out of the lock.

This model is supported by observations of air bubbles sucked into the pharynx. They remain unchanged in size in the corpus and anterior isthmus, showing that these regions are at ambient pressure, then shrink in the isthmus and vanish into solution in the terminal bulb, showing that the pressure is high there (Doncaster 1962; L. Avery, unpubl.). It is not known what determines when an isthmus peristalsis occurs, although it is likely to be controlled by the M4 neuron (see below).

1. The Pharyngeal Muscle Action Potential

Pharyngeal muscle was the first C. elegans cell type from which intracellular electrical recordings were achieved (S. Lockery, pers. comm.). Figure 5a shows a recording of the membrane potential of an active pharyngeal muscle cell (Davis et al. 1995). The action potentials are very similar to those recorded from Ascaris pharyngeal muscle (del Castillo and Morales 1967; Byerly and Masuda 1979) and resemble action potentials of vertebrate heart muscle cells (particularly cells from the ventricles) in that they are long-lasting and have three phases, which Raizen and Avery (1994) called E, P, and R for excitation, plateau, and repolarization. The action potential begins with excitation: a rapid rise in the membrane potential from the resting value of −40 to −50 mV to a depolarized value of 30 to 40 mV. This is followed by a plateau phase during which the membrane remains depolarized. The length of the plateau phase can vary from 50 to 500 msec, but during normal wild-type pumping, it is typically about 150 msec. The plateau phase ends with an abrupt drop in membrane potential (repolarization) during which membrane potential becomes more negative than resting values. After repolarization, the membrane potential slowly rebounds. During slow pumping, as in Figure 5a, it relaxes back toward the resting potential, but during rapid pumping, the rebound continues right past the resting potential to trigger another action potential, typically 100 msec after repolarization. Each contraction-relaxation cycle corresponds to a single muscle action potential (Raizen and Avery 1994; Davis et al. 1995). Contraction is first visible about 30 msec after excitation and proceeds during the plateau of the action potential. Relaxation immediately follows repolarization.

Intracellular recording methods have only recently been developed for C. elegans. Most of our knowledge of electrical events in the pharynx comes from a simpler recording method in which the electrical currents that flow in and out of the worm's mouth are measured (Raizen and Avery 1994). One of these recordings, called electropharyngeograms or EPGs, is shown in Figure 5b. The EPG is the time derivative of the membrane potential summed over all pharyngeal muscles. Thus, each rapid positive change in membrane potential is seen as a positive spike in the EPG, and each rapid negative potential change is seen as a negative spike. The largest features in the EPG, which correspond to the excitation and repolarization of the corpus muscles (Raizen and Avery 1994; T.A. Starich et al., in prep.; M.W. Davis, pers. comm.), are labeled E and R in Figure 5b.

The nervous system is not necessary for generation of these potential changes—The capacity is probably intrinsic to the muscle cells. Avery and Horvitz (1989) showed that pumping continues even after the entire pharyngeal nervous system is killed. Raizen and Avery (1994) showed that E and R spikes are not greatly altered even after all pharyngeal neurons except M4 are killed. (The essential neuron M4 was spared in order to allow the worms to reach adulthood. M4 is believed not to affect pumping [Raizen et al. 1995].)

2. Ionic Basis of the Action Potential

Although little is known for certain about the ionic basis of the pharyngeal muscle action potential, it is likely to be mediated by voltage-activated calcium channels. Except in chordates, muscle action potentials have been reported to be mostly or entirely calcium-mediated (Hagiwara and Byerly 1981), and there is no reason to suppose that nematode pharyngeal muscle is an exception. Thus, the sustained high potential during the plateau phase and some part of the rise in membrane potential during excitation are probably caused by influx of Ca⁺⁺ through plasma membrane calcium channels. Recent genetic studies have identified a gene that probably encodes a subunit of the pharyngeal muscle Ca⁺⁺ channel (R.Y.N. Lee et al., in prep.).

Muscle contraction results from a rise in the cytoplasmic Ca⁺⁺ concentration. There are two potential sources for this Ca⁺⁺: the Ca⁺⁺ that enters through the plasma membrane channel, and Ca⁺⁺ release from intracellular stores. On excitation, most muscle cells release Ca⁺⁺ from the endoplasmic reticulum (or sarcoplasmic reticulum, a specialized endoplasmic reticulum) into the cytoplasm through a calcium channel called the ryanodine receptor, which opens in response to the opening of the plasma membrane channel.

C. elegans possesses a ryanodine receptor (Kim et al. 1992; Y. Sakube and H. Kagawa, pers. comm.) encoded by unc−68 (E. Maryon and P. Anderson, pers. comm.). Ryanodine, which opens the ryanodine receptor channel, causes body muscle contraction in wild-type (Kim et al. 1992) but not unc−68 worms (E. Maryon and P. Anderson, pers. comm.). Surprisingly, unc−68 is not necessary for body or pharyngeal muscle contraction (E. Maryon and P. Anderson, pers. comm.). This result suggests that enough Ca⁺⁺ may enter through the plasma membrane to cause contraction. (Although Ca⁺⁺ release through a different intracellular channel such as the IP3 receptor [H. Baylis et al., pers. comm.] cannot be excluded, such a channel would not be expected to respond rapidly enough to plasma membrane potential changes to account for pharyngeal muscle motions.) unc−68 mutants are uncoordinated (Brenner 1974) and have abnormal pharyngeal muscle motions (R.Y.N. Lee, pers. comm.). Furthermore, they do not respond as strongly as wild type to excitation of body muscles by the acetylcholine agonist levamisole (Lewis et al. 1980a). Thus, the ryanodine receptor does have a role in body wall muscle contraction. Perhaps it allows Ca⁺⁺ concentration to rise more rapidly in response to membrane excitation than the plasma membrane channel alone could accomplish. This hypothesis could be tested by measuring the onset of contraction and the rise of Ca⁺⁺ concentration in response to voltage pulses in unc−68 and wild type.

If the action potential is sustained by a voltage-activated calcium channel, some other event at the beginning of the action potential must produce the initial rise in membrane potential that opens the calcium channel. The control of this event is a key behavioral function, since the frequency of such excitations determines the frequency of pumping, which is the major behavioral response of the pharynx. During rapid pumping, the action potential is likely to be triggered by an excitatory postsynaptic potential from the motor neuron MC (see below). However, the muscle can also be excited in the absence of the nervous system. Nothing is known about the muscle-intrinsic excitation.

The best-understood part of the action potential is the repolarization, thanks to voltage-clamp studies on Ascaris pharyngeal muscle by Byerly and Masuda (1979). They identified a voltage-gated potassium channel called the negative spike channel that closes when the membrane is depolarized but is opened by rapid negative potential changes. Its opening allows K⁺ to leave the cell, causing the potential to become more negative, which causes more negative spike channels to open, resulting in a fast regenerative negative-going spike. The effects of membrane potential on the rate of repolarization are consistent with the existence of such a channel in C. elegans pharyngeal muscle (Davis et al. 1995).

3. Synchronization of Pharyngeal Muscles

During a pump, corpus and terminal bulb muscles contract together. EPG recordings show that this synchronization of contraction is accomplished by synchronization of the action potentials. In the wild type, the excitation of the isthmus and corpus happens within a few milliseconds of each other, resulting in one large E spike (Fig. 5b). Repolarization is less tightly synchronized, but corpus and terminal bulb are still coupled: The terminal bulb R spike usually occurs less than 50 msec after the corpus R spike (Raizen and Avery 1994). Within the corpus or within the terminal bulb, synchronization is nearly perfect. In mutants homozygous for the eat−5 mutation (only one allele is known), corpus and terminal bulb contractions often occur separately (Avery 1993a), and when this happens, separate excitation spikes are seen: Small spikes correlate with terminal bulb contraction and large spikes correlate with corpus contraction (Starich et al. 1995).

Figure 5

Electrical recordings from pharyngeal muscle. (a) Intracellular recording from a terminal bulb muscle cell. (E) Excitation, the sharp rise in membrane potential that begins the (more...)

Avery and Horvitz (1989) proposed that pharyngeal contractions are synchronized by electrical coupling of pharyngeal muscles through gap junctions. Synchronization does not require the pharyngeal nervous system (Avery and Horvitz 1989), and it is difficult to imagine any nervous-system-independent mechanism other than electrical coupling that could synchronize muscle action potentials with millisecond precision.

The eat−5 gene, which is necessary to synchronize the corpus and terminal bulb, encodes a member of the OPUS family of membrane proteins (T.A. Starich et al., in prep.), including C. elegans UNC−7 and Drosophila Ogre and Passover. On the basis of predicted protein structure and the unc−7 and Passover mutant phenotypes, Barnes (1994) proposed that OPUS is a family of invertebrate gap junction proteins. If this proposal could be verified by functional expression of OPUS proteins, it would greatly strengthen the case for electrical coupling of pharyngeal muscles.

The purpose of corpus and terminal bulb synchronization is unknown. eat−5 worms grow more slowly than wild type, but this may be because the terminal bulb pumps more slowly, perhaps because the pacemaker is in the corpus (see below). Nematodes of the genus Panagrellus feed efficiently with unsynchronized corpus and terminal bulb contractions (Mapes 1965). Panagrellus silusiae and Panagrellus redivivus are bacteria-eating soil nematodes with a pharynx similar in overall form to that of C. elegans. The terminal bulb pumps more rapidly than the corpus in both species (Mapes 1965; L. Avery, unpubl.). Perhaps Panagrellus and C. elegans use different mechanisms to generate terminal bulb rhythm.

The isthmus does not contract in tight synchrony with the corpus and terminal bulb. In fact, the anterior and posterior isthmus do not contract at the same time, even though each muscle cell runs the entire length of the isthmus. Within the anterior or posterior isthmus, contraction occurs as a wave that propagates from anterior to posterior, instead of simultaneously along the length as in the corpus or terminal bulb. Furthermore, no EPG signal has ever been detected from the isthmus. This suggests that depolarization of the isthmus, like its contraction, is spread out in time, so that no single discrete excitation spike results. We believe that the capacity of the isthmus for unsynchronized contraction is functionally important. Asynchrony of the anterior and posterior halves allows the terminal bulb and corpus lumen to be at different pressures (see above), and the anterior to posterior waves of contraction appear to be important for transporting bacteria posteriorly.

The lack of synchrony in isthmus contraction can be explained by proposing that isthmus muscle, unlike terminal bulb or corpus muscle, is incapable of regenerative action potentials. In this case, local excitation of the muscle would produce a local depolarization and local contraction, both of which would tend to spread slowly from the site of excitation and decrease with distance. Thus, the delayed contraction of the anterior isthmus during a pump would be explained by excitation at the anterior end through electrical coupling to corpus muscle cells. Posterior isthmus peristalsis could be a result of excitation by the motor neuron M4 (see below), which synapses on the posterior end of the isthmus muscle cells (Albertson and Thomson 1976). This idea could be tested by intracellular recording from the isthmus at various points along its length.

1. M3: A Single Neuron Proprioceptive Loop?

The M3s are inhibitory motor neurons that control the timing of pharyngeal relaxation. They are a bilaterally symmetric pair of motor neurons with output to the metacorpus and perhaps the isthmus (Albertson and Thomson 1976). When they fire, they produce fast negative changes in muscle membrane potential (inhibitory postsynaptic potentials) that can trigger repolarization and therefore relaxation (Avery 1993b; Raizen and Avery 1994). This regulation of the timing of relaxation seems to be important for effective transport of bacteria within the pharyngeal lumen (Avery 1993b).

Each M3 may constitute a single-neuron proprioceptive loop, firing in response to corpus muscle contraction and causing relaxation. Albertson and Thomson (1976) first proposed that M3 might be a sensorimotor neuron because it has free endings in the metacorpus. Consistent with this idea, Raizen and Avery (1994) saw M3 inhibitory postsynaptic potentials even when other pharyngeal neurons had been killed (showing that it is a motor neuron) and only when corpus muscle was contracted (suggesting it is sensory). The evidence still falls short of proof (for discussion, see Raizen and Avery 1994).

J.A. Dent et al. (in prep.) have tentatively identified the M3 neurotransmitter as glutamate. The following are the key results: (1) Pulses of glutamate applied to the pharyngeal muscle mimic the effect of M3; (2) avr−15 (avermectin-resistant) mutants, whose pharyngeal muscle does not respond to glutamate pulses, also lack M3 transmission (see below). Immunocytochemical detection of glutamate in the M3 neurons has not yet been successful. Glutamate probably acts by opening an avermectin-sensitive glutamate-gated chloride channel in the pharyngeal muscle (see Rand and Nonet, this volume). The avermectins are broad-spectrum nematocidal drugs that paralyze Ascaris body muscle by irreversibly opening Cl⁻ channels (Martin 1993). To identify avermectin targets, Arena et al. (1991) injected C. elegans mRNA into Xenopus oocytes and found a Cl⁻ channel that was irreversibly opened by avermectin. They subsequently cloned two cDNAs that encode such a channel (Cully et al. 1994). On the basis of the finding that expression of these cDNAs resulted in a glutamate-gated Cl⁻ conductance, they proposed that this channel might mediate fast inhibitory glutamatergic transmission (Arena et al. 1992; Cully et al. 1994).

C. elegans pharyngeal muscle is paralyzed by low concentrations of avermectin (Avery and Horvitz 1990; M. Chalfie, pers. comm.), and this paralysis can be reversed by lowering the extracellular Cl⁻ concentration (J.A. Dent, pers. comm.), suggesting that in C. elegans pharyngeal muscle as in Ascaris body muscle (Martin 1993), avermectins open Cl⁻ channels. To test the hypothesis that M3 transmission is mediated by an avermectin-sensitive glutamate-gated Cl⁻ channel, J.A. Dent et al. (in prep.) recorded EPGs from avermectin-resistant mutants isolated by C.D. Johnson (Rand and Johnson 1995; C.D. Johnson, unpubl., cited by Anderson 1995). One gene that can confer avermectin sensitivity, avr−15, was necessary for M3 transmission. Furthermore, pulses of glutamate applied to wild-type but not avr−15 mutant pharyngeal muscle mimic the effects of M3 (J.A. Dent et al., in prep.; H. Li et al., unpubl.). Cloning of avr−15 showed that it encodes a new member of the family of avermectin-sensitive glutamate-gated chloride channel subunits (J.A. Dent et al., in prep.).

2. MC Controls the Rate of Pumping

The MCs are excitatory motor neurons that control the initiation of pharyngeal muscle action potentials and therefore the frequency of pumping (Avery and Horvitz 1989; Raizen et al. 1995). When they fire, they produce a fast positive change in muscle membrane potential (an excitatory postsynaptic potential), which usually triggers a muscle action potential. Raizen et al. (1995) proposed that MC is the pacemaker for rapid pharyngeal pumping; it is the sole neuron type necessary for this behavior. The rate of pharyngeal pumping is regulated by the presence of food, the nutritional state of the worm, and neurotransmitters such as serotonin (Horvitz et al. 1982; Avery and Horvitz 1990; Raizen et al. 1995). Most of this regulation requires MC (Avery and Horvitz 1989; Raizen et al. 1995), so it is likely that MC is the major target of regulation of the rate of pumping. However, there is also a nervous-system-independent muscle response to serotonin (D.M. Raizen, pers. comm.).

MC appears to function as a motor neuron, because its ability to produce postsynaptic potentials in pharyngeal muscle does not depend on other neurons (Raizen et al. 1995). Albertson and Thomson (1976) found that the MCs synapse not on muscle, but on the marginal cells, which are structural cells located between the muscle cells (see Fig. 2a). Since nematodes lack obvious postsynaptic specializations, it cannot be excluded on the basis of electron microscopy that pharyngeal muscle is a postsynaptic partner. However, a more interesting possibility is that MC in fact excites the marginal cells, which then serve as a conduction pathway to excite electrically coupled muscle cells, somewhat like the Purkinje fibers of the vertebrate heart. In fact, gap junctions between marginal cells and muscle cells have been observed in electron micrographs (D. Hall, pers. comm.).

Like M3, MC may be a sensorimotor neuron. It has a putative mechanosensory ending at the boundary between the procorpus and metacorpus (Albertson and Thomson 1976). Laser microsurgery and electrophysiological recordings suggest that MC is stimulated by bacteria in the pharynx and that the precise timing of its firing is adjusted in response to muscle motions (Raizen et al. 1995).

The identity of the MC neurotransmitter is not clear. Pharmacological and genetic data point to acetylcholine, but these are contradicted by immunocytochemical experiments. Acetylcholine agonists are excitatory to pharyngeal muscle (Avery and Horvitz 1990; Raizen et al. 1995). This effect is probably mediated in part by a nicotinic receptor distinct from the levamisole receptor (Avery and Horvitz 1990), the major body muscle nicotinic receptor. The nicotinic blocker curare blocks MC neuromuscular transmission (Raizen et al. 1995). Furthermore, the eat−18 gene is necessary both for pharyngeal muscle to respond to nicotine and for MC neuromuscular transmission. The phenotype of eat~-18 mutants is indistinguishable from that of worms in which MC has been killed. Worms that carry partial loss-of-function mutations in the cha−1 and unc−17 genes, which are necessary for cholinergic transmission (see Rand and Nonet, this volume), pump slowly (Avery 1993a), and cha−1 null mutants do not pump (Avery and Horvitz 1990). However, MC does not stain with antibodies against CHA−1 or UNC−17 protein (J. Duerr and J. Rand, pers. comm.). These antibodies stain extrapharyngeal motor neurons previously known to be cholinergic, and they stain certain pharyngeal neuron types other than MC (J. Duerr and J. Rand, pers. comm.). Molecular characterization of eat−18 and mosaic analysis to determine the focus of the cha−1/unc−17 slow-pumping defect might resolve this question.

Figure 6 summarizes the proposed actions of M3 and MC on the pharyngeal muscle action potential. Parts of the model shown in this figure are speculative. In particular, since there is presently no way of recording directly from pharyngeal neurons, the patterns of M3 and MC firing are uncertain.

3. M4 Controls Isthmus Peristalsis

The M4 motor neuron synapses on the posterior half of the isthmus muscles (Albertson and Thomson 1976) and is necessary for posterior isthmus peristalsis (Avery and Horvitz 1987). Worms lacking M4 swallow little or no food and therefore fail to grow. (Like intact worms deprived of food, M4⁻ worms may survive up to 2 weeks.) They continue to pump, and the pumping appears to be functional in that bacteria are concentrated in the anterior isthmus and corpus, which become stuffed (Avery and Horvitz 1987). Although M4⁻ worms pump slowly, this is probably a consequence of the stuffing, since if the corpus is cleared by placing the M4⁻ worms on drugs that excite isthmus muscle, they will pump rapidly until the lumen is again full (see Raizen et al. 1995).

Figure 6

Model for the function of MC and M3. The figure shows four schematic traces. The top three represent hypothesized intracellular recordings from an MC neuron, an M3 neuron, and a (more...)

Little is known about M4 activity or communication between M4 and isthmus muscle. Isthmus peristalsis occurs after some pumps but not others (Avery and Horvitz 1989) and appears to be an all-or-none event. All pumping cycles in an M4⁻ worm are similar to those cycles in intact worms in which isthmus peristalsis does not occur. This suggests that isthmus peristalses may be signaled by single M4 action potentials. However, no M4 postsynaptic potentials are detected in the EPG, suggesting that its direct effect is not fast electrical excitation. It might, for instance, activate intracellular signals that allow the muscle to contract in response to depolarization from the terminal bulb. However, there are several alternative explanations: Perhaps M4 is merely permissive for posterior isthmus contraction, and the muscle has an intrinsic mechanism for making an all-or-none decision to contract, or perhaps postsynaptic potentials in the isthmus cannot be detected in the EPG.

M4 stains with antibodies against UNC−17, suggesting that it may use acetylcholine as its transmitter (J. Duerr and J. Rand, pers. comm.). However, Albertson and Thomson (1976) saw large dense-core vesicles at M4 neuromuscular junctions, so if acetylcholine is an M4 neurotransmitter, it is probably not the only one. Furthermore, mutations that weaken cholinergic transmission have no striking effect on isthmus peristalsis (L. Avery, unpubl.).

4. NSM: A Signal of the Presence of Food?

Although they are not as important as MC, M3, and M4 in the control of feeding, the NSMs have received as much attention. This pair of neurons has synapses in the isthmus not only on the muscle, but also on the basal lamina that bounds the pharynx (Albertson and Thomson 1976). Because the extrapharyngeal nerve ring (the closest thing a nematode has to a brain) lies on the other side of this basal lamina and because the NSMs have putative sensory endings at the boundary between the corpus and isthmus (where bacteria accumulate), Albertson and Thomson (1976) proposed that they might secrete something into the pseudocoelom when their endings detected food in the pharyngeal lumen. This proposal was supported by the discovery that the NSMs contain serotonin (Albertson and Thomson 1976). Serotonin has three obvious effects on hermaphrodites: It depresses locomotion, stimulates egg laying, and stimulates pumping (Croll 1975a; Horvitz et al. 1982). These are all sensible responses to the presence of food and are in fact all seen in the presence of bacteria (Croll 1975b; Croll and Smith 1978).

Unfortunately, proof of a neurohumoral function of the NSMs has been hard to come by. Killing them has almost no effect on behavior, although subtle effects in the expected direction have been seen (Avery et al. 1993; E. Sawin; D.M. Raizen; both pers. comm.). The meagerness of these effects can be reconciled with the evidence that NSM signals the presence of food by proposing that NSM is redundant with other neurons that detect bacteria, depress locomotion, and stimulate pumping and egg laying. Proof will require identifying the redundant neurons. Unfortunately, the best candidate for the redundant neuron that stimulates feeding is MC, which is probably also the major target through which the NSMs stimulate feeding; i.e., serotonin probably stimulates pumping largely by increasing the firing frequency of MC, but MC also accelerates its firing in response to bacteria in the absence of serotonin (D.M. Raizen, pers. comm.). Thus, the usual sign of redundancy—a synergistic effect when both redundant neurons are killed—cannot be tested in this case because killing MC effectively eliminates both pathways. If this model is correct, demonstration of a substantial effect of NSM on feeding will require specific elimination of the NSM-independent sensory function.

5. Other Neurons

We have discussed 4 of the 14 pharyngeal neuron types in detail. For most of the remaining 10, there is even less functional information than for NSM. What are these remaining 10 neuron types doing? Some of them unquestionably regulate MC, M3, and M4. The I1s, for instance, receive the gap junctions from the RIPs, which connect the pharyngeal nervous system to the extrapharyngeal nervous system. They synapse on MC (Albertson and Thomson 1976) and affect the rate of pumping in the absence of bacteria (D.M. Raizen, pers. comm.). Similarly, I5 is a sensory neuron that synapses on M3 (Albertson and Thomson 1976). Killing I5 hastens relaxation, but killing both I5 and M3 results in a delay in relaxation indistinguishable from that caused by killing M3 alone. Thus, it is likely that I5 inhibits M3. (I5 also has M3-independent effects on the isthmus.) Other neurons might have small or redundant effects. Another possibility is that some of these neurons are evolutionary detritus. Nematodes have extraordinarily varied pharyngeal morphology and feeding patterns. Perhaps some pharyngeal neurons serve a purpose in other nematodes that is redundant or unnecessary in C. elegans.