1. Mutants

A large number of mutations that disturb the DMP have been isolated. Most were identified by screening directly for constipated mutants (Thomas 1990; E. Jorgensen, pers. comm.). This method of isolation biases toward mutations that cause a sharp reduction in the amount of feces expelled per time. Mutants with less severe defects have been isolated in smaller mutagenesis screens by direct observation of defecation cycles (Iwasaki et al. 1995). In addition, observation of behavioral mutants isolated on the basis of other mutant defects has identified several mutants with altered defecation (Thomas 1990; Reiner et al. 1995). Summaries of mutants with altered motor programs, cycle period abnormalities, or muscle excitation defects have been published previously (Thomas 1990; Iwasaki et al. 1995; Reiner and Thomas 1995). The most severely constipated mutants probably have a degree of constipation that approximates the complete elimination of the defecation motor program. Mutants such as aex-1 and exp-2 never produce active enteric-muscle contractions, and they expel gut contents about ten times less often than the wild type. Release of gut contents in these mutants often does not coincide with the remaining parts of the DMP. Following such a release (which is explosive and often clears nearly the entire gut lumen), the animal continues to feed normally and gradually becomes more constipated, as each defecation cycle fails to expel any gut contents. Eventually the animal becomes severely constipated but continues to feed well, suggesting little or no feedback regulation of feeding. It is a tribute to the power of the pharynx that eventually it fills the gut to the bursting point, the anus is forced open, and a new constipation cycle begins. The release of gut contents in such mutants can occur at any time during the defecation cycle and is not associated with any visible muscle contractions, suggesting that the release is caused by internal pressure. This pattern is phenocopied by killing the enteric muscles in the wild type, indicating that the pattern does not require enteric muscles and is not a mutant artifact. Severely constipated mutants and animals missing the enteric muscles are viable and fertile and appear behaviorally normal, although they mature slowly and are scrawny and small, characteristics of malnourishment (Avery 1993a). On the basis of these observations, it seems unlikely that overt expression of the defecation motor program is essential for viability or fertility. Despite this, large screens for constipated mutants have failed to identify mutations that eliminate the motor program (Thomas 1990; E. Jorgensen et al., unpubl.), suggesting that such mutations are rare.

In addition to providing the raw material for more detailed investigations of defecation behavior, the pattern of mutant defects has some interesting properties. In theory, a series of stereotyped motor steps might be controlled in various ways. In a dependent-pathway model, each step depends on execution of the previous step. In this case, the periodicity of defecation and the timing of motor steps would be determined by simple delays after the execution of each dependent step. In an extreme alternative model, a cycle controller exists that is independent of motor steps and this controller sequentially activates motor steps at appropriate times. A hybrid model could involve a cycle controller that receives checkpoint feedback about the execution of motor steps, in a manner similar to mechanisms proposed for the cell cycle (Hartwell and Kastan 1994). The phenotypes of defecation mutants support a fairly strict controller model: Most defecation mutants affect only one part of the motor program, leaving periodicity and other parts of the motor program relatively unaffected (Thomas 1990). Similarly, most periodicity mutants leave the motor program unaffected (Iwasaki et al. 1995).

2. Motor Neurons

The precision of the defecation motor program demands exact spatial and temporal coordination. It is presumed that this coordination is mediated by the nervous system, but there is direct evidence for only two of the motor steps, aBoc and Exp. unc-25 mutants, which lack detectable amounts of the neurotransmitter GABA, are deficient in activation of the E.p step (Thomas 1990; McIntire et al. 1993a). Antibodies to GABA stain the neurons AVL and DVB, and killing both AVL and DVB with a laser microbeam eliminates enteric-muscle contractions (McIntire 1993b). DVB makes a neuromuscular junction with the anal depressor (White et al. 1986; E. Jorgensen, pers. comm.), and AVL forms a process with varicosities adjacent to the anal depressor muscle, although no clear neuromuscular junction is seen (E. Jorgensen, pers. comm.). No other neurons that have cell bodies or processes in the anal region are required for EMCs (McIntire 1993b). These data indicate that AVL and DVB are excitatory GABAergic motor neurons for the enteric muscles. It is also known that exogenous serotonin inhibits EMCs (Ségalat et al. 1995), although it is unclear how this functions in vivo.

Studies of AVL and DVB raise interesting points that may be generally significant for nematode neurons. First, AVL and DVB are partially redundant in activating EMCs. When either neuron alone is killed, EMC frequency is nearly normal, but when both are killed, EMCs are eliminated (McIntire et al. 1993b). We discuss neuronal redundancy near the end of this chapter. A second interesting point is that the EMC defect caused by killing AVL and DVB is more severe (∼0% EMC) than that caused by the absence of GABA. In unc-25 mutants, about 15% of cycles have an EMC (Thomas 1990; McIntire 1993b). Neuron kills in an unc-25 mutant show that the residual EMCs in the absence of GABA require DVB but not AVL (E. Jorgensen, pers. comm.; J.H. Thomas, unpubl.), despite the fact that DVB is slightly less important than AVL for EMC activation in the wild type (McIntire et al. 1993b). These findings suggest that DVB employs a second EMC-activating neurotransmitter that is absent from AVL. A plausible candidate transmitter is the peptide FLRFamide (Schinkmann and Li 1992; see Rand and Nonet, this volume), since it is present in DVB but not AVL, and it is known that peptides commonly accompany small molecular transmitters in other organisms.

A third interesting point is that GABA appears to be an excitatory transmitter for the enteric muscles. Usually in mammals, and in all other characterized cases in C. elegans, GABA is an inhibitory transmitter (see, e.g., Tobin 1991; McIntire et al. 1993a). It is unlikely that GABA indirectly excites enteric muscles by inhibiting the action of an inhibitory neuron: Both AVL and DVB make output to the enteric muscles, and no other neuron in the region is required for normal EMCs (White et al. 1986; McIntire et al. 1993b). If GABA is indeed excitatory, it might act on a novel ionotropic GABA receptor or it might act via a metabotropic receptor and a second-messenger system. A G-protein G_o is expressed in the enteric muscles, but it is not known whether this G-protein is a target of GABA (Mendel et al. 1995; Ségalat et al. 1995). The exp-1 gene, mutations in which are unusual in producing a defecation phenotype identical to that of unc-25 , is a plausible candidate for encoding this novel GABA receptor (Thomas 1990; E. Jorgensen, pers. comm.; J.H. Thomas, unpubl.).

A final point is that AVL functions redundantly with DVB in activating EMCs, but killing AVL alone causes a strong aBoc-defective phenotype (McIntire 1993b). Mutants that lack GABA function have a normal aBoc step (Thomas 1990). These facts indicate that AVL has a nonredundant role in activating aBoc and that this function does not require GABA. AVL probably does not directly activate the head muscles that contract during aBoc, because no process from AVL passes anywhere near those muscles (White et al. 1986). By the conventional definition of neuron types, AVL is an interneuron using one transmitter for one muscle contraction (aBoc), and it is a motor neuron using a second transmitter (GABA) for a second muscle contraction (EMC). Functional complexity has also been described for the sensory neuron ASH, which mediates response to both nose touch and osmotic stimuli, possibly using different neurotransmitters (Bargmann et al. 1990; Kaplan and Horvitz 1993; Hart et al. 1995; Maricq et al. 1995). We speculate that such complexity of neuron function is common in nematodes, perhaps as a result of the limited repertoire of available neurons.

Ten genes have been identified that mutate to an aBoc and Exp-defective (Aex) phenotype reminiscent of killing AVL and DVB (Thomas 1990; J.H. Thomas, unpubl.; E. Jorgensen, pers. comm.). We think it likely that these genes are required for the function or activation of AVL and DVB, rather than any of their specific transmitter systems, since we expect that elimination of any one transmitter would not produce an Aex phenotype. The strongest aex mutations cause a phenotype nearly identical to killing AVL and DVB, but some other aex mutations result in a weaker phenotype, in which aBoc and E.p are more frequently present. It is unknown whether or not these aex mutations are null. In addition to the tight connection between the aBoc and E.p steps, there is also some indication that pBoc and E.p are connected in some manner not yet understood. Although many mutants that are deficient in pBoc have normal Exps, egl-8 mutants (Trent et al. 1983) have a very weak pBoc and a variably reduced EMC frequency, as do some other less-characterized mutants (K. Iwasaki and J.H. Thomas, unpubl.; E. Jorgensen, pers. comm.).

3. pBoc Activation

The body-wall muscles used for pBoc and aBoc are the same as those used for locomotion. During locomotion, the dorsal and ventral body muscles are reciprocally contracted and relaxed to generate bends in the body, and the motor neurons that control these movements have been identified (Stretton et al. 1985; White et al. 1986; McIntire et al. 1993b). In contrast, during pBoc and aBoc, dorsal and ventral body muscles are contracted simultaneously, causing the body to shorten locally (Fig. 7). A large number of mutants have been identified that profoundly affect locomotion, including many that severely perturb the locomotory motor neurons (see Hedgecock and Garriga, this volume). None of these mutants is defective in pBoc (Thomas 1990; J.H. Thomas, unpubl.; E. Jorgensen, pers. comm.), indicating that activation of pBoc occurs by a distinct pathway. (This may be true for aBoc as well, but the aBoc contraction is less robust and has not been analyzed in as much detail.) The source of pBoc activation is mysterious, since laser kills suggest that no neuron in the region of these muscles is required for their contraction (E. Jorgensen, pers. comm.). It is possible that some neuron acts at a distance to activate pBoc, perhaps through the pseudocoelomic space, or pBoc may be activated by a nonneuronal pathway, for example, by the hypodermal syncytium.

4. Excitation of the Enteric and Egg-laying Muscles

Many of the mutations that affect enteric muscle contraction share two properties: They are semi-dominant and they also affect egg-laying muscle contraction (Greenwald and Horvitz 1980, 1986; Trent et al. 1983; Park and Horvitz 1986a; Levin and Horvitz 1993; Reiner et al. 1995; Weinshenker et al. 1995). Many of these mutations also affect additional muscle groups and were originally identified by these other muscle defects (Brenner 1974; Greenwald and Horvitz 1980, 1986; Trent et al. 1983; Park and Horvitz 1986a; Avery 1993b; Levin and Horvitz 1993; R. Waterston, pers. comm.). The muscle specificity of these mutations is summarized in Table 1. For each of these mutants, the muscle myofilaments appear to be relatively unaffected. Polarized-light microscopy showed that their muscle organization is normal, and each mutant can contract its anal-depressor muscle when it is shot with a laser (Reiner et al. 1995). The mutants do not lay eggs in response to excitatory transmitters that are thought to act directly on egg-laying muscle (Trent et al. 1983). These results suggest that these genes affect muscle excitation but not contractile function.

Table 1

Summary of egl/exp genes and affected muscle groups.

An extraordinary feature of the egl/exp mutations is that every one of them is dominant, despite the fact that nearly all were isolated in standard F₂ screens designed to detect recessive mutants. Most mutations causing abnormal myofilament structure are recessive (Waterston 1988; Fire and Moerman, this volume) as is characteristic of mutations affecting most processes. Deficiencies eliminating 8 of the 12 genes exist, and none of these cause dominant muscle defects, indicating that the dominance of the egl/exp mutation is due to a gain of function. Putative loss-of-function mutations have been identified for eight of the egl/exp genes and seven of these produce no obvious phenotype. It will be interesting to learn why so many muscle excitation genes are identified by dominant mutations. We speculate that excitation of these muscles involves a number of negative regulatory pathways, which can mutate to the easily detected Egl or E.p phenotypes only by gain of function. Loss-of-function mutations in such genes would cause the relatively subtle phenotype of stronger muscle contraction. An example of such a negative regulator of neuronal and muscle excitation is K⁺ channels, which function to shape and terminate action potentials.

1. Mutants

Circadian rhythms, with a period of about 1 day (circa dia), are being analyzed genetically and molecularly in several organisms, including cyanobacterium, plants, Drosophila, Neurospora, and mouse (see, e.g., Konopka and Benzer 1971; Hall 1990; Kondo et al. 1994; McClung et al. 1989; Vitaterna et al. 1994). Ultradian rhythms, with a period of less than 24 hours, are also widespread but have been the subject of relatively little genetic analysis (see, e.g., Edmunds 1988). Its high frequency, tight periodicity, and ease of observation make defecation periodicity amenable to genetic analysis.

A number of defecation cycle period (Dec) mutants have been identified, genetically mapped, and characterized phenotypically (Thomas 1990; Iwasaki et al. 1995). Most such mutations do not cause substantial constipation and are identified and analyzed by direct observation of defecation cycles. A total of 12 genes have been identified that can mutate to affect the defecation cycle period specifically (Figs. 9 and 10). For each mutation, feeding and defecation motor steps appear to be grossly normal. Mutations in these genes fall into two major groups: short cycle (Dec-s) and long cycle (Dec-L; we use the capital L to distinguish it from the numeral “1”). Mutations in seven genes cause a short cycle period. These mutants can be divided into three subclasses based on detailed phenotype. Mutations in flr-1 , flr-3 , and flr-4 are recessive and cause a very short mean cycle period, especially at 25°C where cycles are often less than 20 seconds. Mutations in these three genes were first identified and named for their fluoride resistance (Flr) phenotype (Katsura et al. 1994), but it is unclear how the Dec-s and Flr phenotypes are related. Mutations in two other genes, dec-7 and unc-16 , are recessive and cause moderately short cycle periods. Finally, Dec-s mutations in two genes, dec-9 and dec-10 , are semidominant, and genetic deficiency tests suggest that both mutations are gain-of-function. None of these last four Dec-s genes and none of the Dec-L genes confer fluoride resistance. All five identified Dec-L mutations are recessive and each Dec-L gene is defined by a single mutation. dec-2 and dec-4 alone among all of the Dec mutations cause an altered motor program: The interval between the pBoc and E.p steps is slightly longer than normal (Iwasaki et al. 1995). The recently described clk-1 mutant also has a Dec-L phenotype and, unlike other Dec mutants, clk-1 affects the timing of many other behaviors and developmental events (Wong et al. 1995).

Figure 10

Defecation cycle periods of the wild-type and dec mutants at 20°C and 25°C. (Solid bars) 20°C; (hatched bars) 25°C. Genes (more...)

The phenotypes of the clock mutants and the motor program mutants are parsimoniously explained by the model in Figure 11. In this model, a clock runs separately from the motor program and periodically initiates a motor program. Each step in the motor program can be individually affected by mutation without perturbing the clock, so the motor steps are depicted as branching from each other after the program initiation step. The aex genes are depicted as affecting a step common to aBoc and E.p, probably the activation or function of AVL and DVB. One of the chief challenges of future research will be to move beyond this formal model by deciphering the cellular and molecular pathways that underlie the genetic pathway.

2. Temperature Compensation

In general, biochemical reactions occur faster at higher temperatures. The fact that the defecation cycle period remains nearly constant over a range of temperatures implies that a specific mechanism compensates the cycle period at different temperatures. In contrast to the wild type, most dec mutants have temperature-dependent cycle periods (Fig. 10). Temperature-dependent phenotypes can be caused by heat- or cold-sensitive gene products, but only a small subset of mutations produce such thermolability. The high frequency of temperature-dependent dec mutants implies that they do not result from thermolabile gene products. Supporting this interpretation, some dec mutations confer pleiotropic phenotypes that are not temperature-dependent (Iwasaki et al. 1995). The simple interpretation is that temperature compensation is an intrinsic feature of the clock mechanism, rather than a separate circuit that regulates cycle periodicity. In Drosophila and Neurospora, mutations in per and frq cause temperature-compensation defects in addition to abnormal circadian rhythms (Loros and Feldman 1986; Konopka et al. 1989), suggesting that this clock is also integrated with temperature compensation.

Figure 11

A genetic model of defecation. A cycle generator controls the timing of defecation. A branching pathway controls the individual motor steps. Genes affect various steps as indicated by lines (more...)

All of this begs the question of why the defecation cycle should be temperature-compensated. Presumably, the rate of defecation is the major determinant of the residence time of food in the gut, and this residence time must influence the thoroughness of digestion and absorption of nutrients. Assuming that the rate of digestive processes increases with temperature, the temperature compensation of the defecation cycle means that nutrient extraction is less complete at lower temperatures. We have no idea why this is adaptive.