The C. elegans intestine^*

Midway through embryogenesis, at the point when the developing intestine has 20 cells, three of the “int” rings, int2, int3 and int4, undergo a clockwise rotation (as seen facing the embryo) of approximately 90° (Sulston et al., 1983); this rotation increases to approximately 180° by the time of hatching (Sulston and Horvitz, 1977) but may be variable (See WormAtlas). Presumably, the resulting twist allows the intestine to intertwine with the asymmetrically disposed hermaphrodite gonad. The cellular and molecular basis of this morphogenetic twist has been investigated by Hermann et al. (2000) and by Neves and Priess (2005), who have shown that the twist is the result of repeated application of the LIN-12/Notch pathway (see Notch signaling in the C. elegans embryo).

When the intestine primordium has four cells, the LIN-12 receptor can be detected on the surface of all four cells; no left/right asymmetry can be detected and this early phase of lin-12 gene expression appears to be controlled by the END-1 GATA factor. By the 8E cell stage, however, LIN-12 is markedly downregulated in cells of the left half of the primordium because particular MS-lineage cells expressing the LIN-12 ligand LAG-2 contact the intestine primordium on the left side, not on the right side. Thus, the intestine asymmetry can be traced to MS-lineage asymmetry, which in turn can be traced to the fundamental left/right asymmetry apparent at the six cell stage of the embyo (Wood, 1991; Wood et al., 1996). The LIN-12 asymmetry within the early intestine primordium sets the stage for a second phase in which the actual rotation takes place; this second phase also involves asymmetric LIN-12 receptor but now with the alternative ligand APX-1. Both of these interactions, the LAG-2 dependent imposition of LIN-12 asymmetry at the 4E stage and the subsequent APX-1 dependent process of cell rotation, involve the bHLH protein REF-1 as a direct downstream target of Notch signaling (see Notch signaling in the C. elegans embryo; Neves and Priess, 2005). The REF-1 effect is ascribed to a repression of downstream genes (in particular, the lin-12 gene itself in the first interaction) and involves the corepressor UNC-37/Groucho. The decision of which cells will rotate (i.e., what sets the anterior/posterior boundary of the twist) is controlled by the Wnt pathway implicated in the anterior-posterior patterning of gene expression, as will be described in the following section.

Fascinating questions of cell biology remain to be addressed. For example, how do these twisting cells break contact with their neighbours, rotate through this relatively precise angle, rearrange the complex three-dimensional scaffold of apical junctions with their neighbours, and yet still manage to exert the correct degree of rotational torque necessary to impart the helical form to the overall intestine?

A significant number of genes are expressed in specific non-uniform patterns along the anterior-posterior axis of the intestine: for example, genes associated with the defecation pacemaker are expressed towards the intestine posterior (Dal Santo et al., 1999), the ges-1ΔB transgene is expressed in the anterior int1 and int2 cells (Schroeder and McGhee, 1998) and the pho-1 acid phosphatase gene is expressed everywhere in the intestine except in int1 and int2 (Beh et al., 1991; Fukushige et al., 2005). Indeed, in situ hybridization (see Protocols for large scale in situ hybridization on C. elegans larvae, http://nematode.lab.nig.ac.jp/db2/index.php) reveals a number of clear examples where endogenous transcripts, not just reporter gene expression patterns, are expressed either in the intestine anterior, like ges-1ΔB, or everywhere but in the intestine anterior, like pho-1. It is well to keep in mind that reporter genes driven by incomplete promoters often express in the posterior intestine and, to a lesser extent, in the anterior intestine (Ruvinsky and Ruvkun, 2003; see Transcriptional regulation); thus, such expression patterns should be treated with caution.

The anterior-posterior patterning of gene expression is autonomous within the intestine lineage and reflects zygotic (i.e., non-maternal) application of the Wnt pathway (see Wnt signaling, Schroeder and McGhee, 1998), at least as worked out for the case of the ges-1ΔB transgene. At least the major components (WRM-1, LIT-1, and the downstream effector POP-1) of this pathway are the same molecules studied so intensely in the earlier P₂-EMS contact that specifies the intestine. Overall, the ges-1ΔB transgene behaves as if its transcription is activated in the intestine anterior by high levels of the HMG protein POP-1 (Schroeder and McGhee, 1998). Conversely, expression of the pho-1 acid phosphatase gene behaves as if it is repressed in the intestine anterior by high levels of POP-1 (Fukushige et al., 2005). Because there are no obvious POP-1 binding sites in either promoter, POP-1 and the Wnt pathway may not act directly on the ges-1ΔB or pho-1 promoters but rather may exert their effects via fate changes earlier in the E lineage. As noted in the previous section, the same Wnt pathway also appears to limit the anterior-posterior extent of the intestinal twist (Hermann et al., 2000).

A further example of patterning within the intestine lineage is provided by the germ cell precursors extending lobes into the int5 cells of the intestine, apparently to obtain nutrition (Sulston et al., 1983). Edgar et al. (2001) found that the pal-1 homeobox gene (whose maternal phase specifies the fate of the C and D blastomeres (Hunter and Kenyon, 1996) is expressed in int5 beginning from approximately 350 cells until at least hatching and perhaps this is what makes int5 different from the other ints. However, int5-specific expression of pal-1 may not reflect patterning pathways within the intestine but rather regulatory control from within the primordial germ cells: the pie-1 gene appears to promote this germ cell-intestine interaction and pal-1 expression within the intestine might then be a downstream consequence (Tenenhaus et al., 2001).

At the time of hatching, the L1 intestine has 20 cells, each with a single diploid nucleus (Hedgecock and White, 1985). At the beginning of the L1 lethargus, intestinal nuclei replicate their DNA and divide in a spatially patterned but slightly variable manner: the six nuclei of int1 and int2 neither replicate their DNA nor divide; the four nuclei of int8 and int9 may or may not replicate and divide; the ten nuclei of int 3 to int7 always replicate and divide. As a result, the late L1 intestine retains 20 cells but now has 30 to 34 nuclei, all still diploid (Hedgecock and White, 1985). These nuclear divisions do not occur in null mutants of the heterochronic gene lin-14 (Hong et al., 2000) but the intestine appears to function nonetheless.

All intestinal nuclei endoreduplicate their DNA prior to the L1-to-L2 moult, whether or not they have just undergone one of the nuclear divisions described in the last paragraph. These endoreduplications occur without obvious chromosome condensations and are repeated before each of the three remaining moults, thereby producing the large 32C nuclei of the adult intestine (Hedgecock and White, 1985). Presumably, the basic C. elegans cell cycle machinery must be involved or must somehow be modified to produce these correctly-timed endoreduplications. For example, it has been shown that CKI-1 (a member of a family of cyclin dependent kinase inhibitors) is involved in limiting the number of cell divisions in the intestine. cki-1 knockouts, either by RNAi or by chromosomal deletions, lead to >30 nuclei, with the extra nuclei being produced by an extra round of division shortly before hatching (Hong et al., 1998; Kostic and Roy, 2002; Fukuyama et al., 2003). Simultaneous knockout of cki-1,2 and lin-35/Rb leads to an average of 84 intestinal nuclei (Boxem and van den Heuvel, 2001). The effects of cki-1 knockdowns are seen in many embryonic lineages, not just the intestine, and this holds true for mutations/RNAi in a number of other cell cycle control genes (see reviews by, Kipreos, 2005; Koreth and van den Heuvel, 2005; Boxem, 2006) or even in a calmodulin gene (Karabinos et al., 2003a). In contrast, two independent genetic screens identified mutations that affect only the intestine; both mutations result in an approximate doubling of both intestinal nuclei and intestinal cells, caused by an interpolated round of cell division at the 8E cell stage (Clucas et al., 2002; Kostic and Roy, 2002). In spite of this intestinal phenotype, both mutants are viable and fertile. The two mutations are both maternal dominant (or semi-dominant) gain-of-function, and are caused by single amino acid changes in adjacent residues in a CDC-25.1-like cell cycle phosphatase. Although these two specific gain-of-function alleles affect only the intestine, RNAi-induced loss of function of the same gene clearly affects many lineages and indeed the protein is widespread in the early embryo.

Finally, we recall the observation that expression of biochemical markers in the early E lineage depends upon a critical prior round of DNA synthesis occuring in the E cell, the clonal progenitor of the intestine (Edgar and McGhee, 1988). Perhaps it would be worthwhile to re-visit this observation in the light of the current detailed understanding of endoderm specification.

In forming the dauer larva, the alternative long-surviving third larval stage, the intestine becomes radially constricted and the lumenal circumference decreases by several-fold, to the point where the microvilli can become indistinct (Popham and Webster, 1979; Albert and Riddle, 1988). In the L2d stage, the L2 larval stage leading up to dauer formation, the intestinal cytoplasm can be packed with electron dense “cytosomes” (Popham and Webster, 1979; Albert and Riddle, 1988). How loss of intestinal volume and increase in density happens is not understood but autophagy could be involved (Melendez et al., 2003). It is also not known how the remodeling or the recovery pathways of the dauer intestine relate to the developmental pathway of the non-dauer intestine. The behaviour of particular mutants in daf-9 and daf-15 show that remodeling in the intestine can be separated from the remodeling of other tissues (Albert and Riddle, 1988).

The most prominent feature of the intestinal apical domain is the microvillar brush border (see Figure 3) and a number of genes integral to C. elegans microvilli have now been identified. The following summarizes the various points of interest.

Although it had previously been known that a YFP/GFP-tagged ACT-5 actin protein produced in the intestine would localize to the apical surface (Bossinger et al., 2004; Gobel et al., 2004). MacQueen et al. (2005) have now shown clearly that, at the ultrastructural level, ACT-5 resides within the microvilli. Knockouts of act-5 cause lethality shortly after hatching. The mutant intestinal cells retain normal polarity and have apical junctions but are completely devoid of microvilli. The intestinal lumen is now circular, rather than the usual squashed ellipse. Expression of the act-5 gene is limited to the intestine and several other cells with microvilli. Experiments to replace act-5 with other C. elegans actin genes were not successful, suggesting that ACT-5 performs a unique function in microvilli (MacQueen et al., 2005).

Proteins belonging to the ERM family (ezrin-radixin-moesin) are usually regarded as connecting the actin cytoskeleton to the plasma membrane and that is where ERM-1, the C. elegans homolog, is placed on Figure 3. Both antibody staining and GFP reporters place ERM-1 at or closely beneath the intestinal apical surface, certainly more apically disposed than are the apical junctions (Gobel et al., 2004; Segbert et al., 2004). RNAi to erm-1 is lethal, producing distorted intestines that appear to have adhesions across the lumen. These defects could arise from a failure in a general process of lumen formation (defects are also present in the excretory canal cell), from aberrant intercalation of cells during intestinal morphogenesis (see Figure 2 above) or from defects in the maturation of the apical junctions. As in the knockouts of act-5, the intestinal lumen of erm-1 knockouts is now circular rather than ellipsoidal and the microvilli are replaced by a disorganized tangle of fibrils, which may well be actin (Gobel et al., 2004).

The EPS-8 protein has been located by immunoelectron microscopy in the microvillar tips and, to a lesser extent, in the microvillar bases (Croce et al., 2004). Loss of EPS-8 function can be lethal, with obvious intestine defects: EPS-8 deficient microvilli appear less dense, less rigid and generally more irregular than wildtype microvilli. EPS-8 was also shown to have an actin-capping function in vitro and might function in the process of quality control, somehow regulating the remarkable uniformity of microvillar length.

The intermediate filament protein IFB-2 is the epitope reacting with the monoclonal antibody MH33 (Francis and Waterston, 1985; Karabinos et al., 2001; Bossinger et al., 2004) and is located within the “endotube” (Bossinger et al., 2004), a well-defined sub-apical layer within the terminal web of nematode intestines (Munn and Greenwood, 1984). This layer abuts directly onto the inter-enterocyte apical junctions, as seen by immunoelectron microscopy. Presumably, the function of the endotube is to provide structural strength and to anchor the overlying microvilli. The C. elegans genome contains 11 genes coding for intermediate filaments (Dodemont et al., 1990; Dodemont et al., 1994; Karabinos et al., 2001; Karabinos et al., 2002; Karabinos et al., 2003b; Karabinos et al., 2002; Karabinos et al., 2004) and six are expressed in the intestine (Karabinos et al., 2004); of these, three (in addition to IFB-2) are expressed exclusively in the intestine and produce proteins that, at least from light microscopy, are also located in the terminal web. These intestinal intermediate filaments probably function redundantly; intestinal malformations can be observed when several of the intermediate filament genes are simultaneously targeted by RNAi (Karabinos et al., 2004).

Electron micrographs show a fuzzy layer of material encasing the microvilli and extending slightly into the intestinal lumen (see Figure 3). Presumably this layer corresponds to the C. elegans glycocalyx or “peritrophic matrix” (Lehane, 1997), a complex, relatively unstructured collection of highly modified glycoproteins that serves several critical functions, including: (i) to protect the microvillar surfaces from mechanical injury or from pathogenic attack (see below); (ii) to serve as a filter allowing only digestion products to reach the absorptive surfaces, and; (iii) to provide a scaffold for presenting digestive enzymes to the contents of the intestinal lumen. The ultrastructure of this layer has been investigated in several nematodes, including C. elegans (Borgonie et al., 1995) and chemical stains reveal both heparin sulfate and dermatan sulfate in the region of the terminal web and brush border (Schimpf et al., 1999). Several lectins have been shown to bind to the edge of the intestinal lumen, some in an anterior-posteriorly patterned manner (Borgonie et al., 1994; Borgonie et al., 1997).

A number of further proteins show distinct apical localization within the intestine but may not play a primary role in enterocyte structure. Many of these proteins could be equally well discussed in later sections on digestion, metabolism, channels and transporters but are described here because what is presently known about their subcellular distribution appears more striking than what is known about their function:

The basement membrane surrounding the intestine (see Figure 3) has three of the four major constitutents found in other C. elegans basement membranes: laminin α and β chains (see Basement membranes; Huang et al., 2003; Kao et al., 2006), nidogen/entactin (Kang and Kramer, 2000) and type IV collagen (Graham et al., 1997), all detected on the surface of the intestine beginning in early morphogenesis phase. Interestingly, type IV collagen is not synthesized in the intestine but rather in muscles and in somatic gonads (Graham et al., 1997), from which the product must be able to diffuse through the pseudocoelom to assemble into the basement membrane covering the intestine as well as the pharynx. In contrast, perlecan assembly appears to be cell-autonomous (Moerman et al., 1996) but is not found in the basement membrane of the intestine (Norman and Moerman, 2000). Longitudinal tracts of hemicentin, a large member of the immunoglobulin superfamily, are produced by muscles and gonads and apparently mechanically anchor the intestine, probably via the intestinal basement membrane, to the regions between muscles and hypodermis (Vogel and Hedgecock, 2001).

The following proteins have been associated with the basolateral domain of intestine cells:

Discrete electron dense regions can be detected just beneath the apical, lumenal surface of the intestinal cells, joining each cell to its neighbour (Legouis et al., 2000). The nomenclature of these structures has changed over the years but they are now generally referred to as “apical junctions”. The C. elegans apical junctions are regarded as the counterpart of the bipartite cell junctions seen in Drosophila epithelia (zonula adherens and septate junctions) and the tripartite structures seen in vertebrate epithelia (tight junctions, zonula adherens and desmosomes). Indeed, many of the proteins found in fly and vertebrate cell junctions have clear homologs in C. elegans and this is how many of the C. elegans components were identified. However, the function of each component and its position in the assembly pathway in C. elegans epithelia can differ markedly from other systems. As particular examples, depletion of cadherins, catenins and proteins such as crumbs (CRB-1) has disasterous consequences for fly epithelia but relatively minor consequences in the C. elegans intestine (Costa et al., 1998; Bossinger et al., 2001). One particular view of the relative positioning of proteins and protein complexes within the apical junctions, is shown in Figure 3. The WormBook chapter by Labouesse (see Epithelial junctions and attachments) provides a much more detailed discussion of the important topic of apical-basal polarity in C. elegans epithelia, including the intestine.

Two other interesting genes are highly expressed in the intestine, with the encoded proteins apparently located in the apical junctions:

The well-known birefringent “gut granules” that have served so well as markers of intestinal differentiation have been shown to be lysosome-related organelles (Clokey and Jacobson, 1986; Hermann et al., 2005). The evidence derives from staining with a number of lysosome-specific dyes and protein markers, as well as showing that gut granule formation is prevented by ablating genes whose homologs are known to function in lysosome biogenesis in other experimental systems. (Hermann et al., 2005) have screened for a gut granule loss (Glo) phenotype in order to identify genes involved in lysosome formation. In some of these Glo mutants, the birefringent material is no longer retained within intestinal cells but rather is mis-routed into the intestinal lumen. GLO-1 is a RAB GTPase expressed only in the intestine and is indeed found to be associated with the granules. The identity of other genes in this pathway have suggested that gut granules may be similar to pigment granules in Drosophila and melanosomes in humans (Hermann et al., 2005).

Several other proteins have been associated with gut granules or with related intestinal organelles:

It is difficult to disagree with the lament that “unfortunately, the biology of digestion (in C. elegans) represents something of a blind spot in this otherwise remarkably well-characterised organism” (Walker et al., 2005). However, digestion and metabolism have been studied in other nematodes (Behm, 2002; Munn and Munn, 2002) and there is increasing understanding of these processes in C. elegans as well. For example, analysis of a SAGE library prepared from dissected adult intestines identifies transcripts from over four thousand genes (McGhee et al., 2007) and this list of genes must hold many keys to understanding digestion, metabolism and other aspects of intestinal physiology. In this section, we will highlight genes that are expressed highly, often exclusively, in the intestine and for which a functional role in digestion and metabolism is either known or plausible.

The general steps by which the animal digests its bacterial food can be surmised and are fully supported by analysis of the most highly expressed, most intestine-restricted set of genes identified in our SAGE analysis (McGhee et al., 2007). Presumably, the first act of digestion is physical damage to the bacteria inflicted by the pharyngeal grinder (Avery and Thomas, 1997). A small number of lysozymes (Mallo et al., 2002; McGhee et al., 2007) are highly expressed in the intestine, are probably secreted and are the likely perpetrators of the next step in digestion, degradation of the bacterial cell wall. Several saposins/amoebapores are also highly expressed in the intestine and some appear to be secreted; presumably these small pore-forming proteins are able to puncture the bacterial membranes to release contents that can now be degraded by lumenal hydrolases (Banyai and Patthy, 1998).

What is known about the general properties of the lumenal environment in which digestive hydrolases must act? For example, what is the lumenal pH? Most protease activity in crude extracts likely derives from intestinal proteases (see below) and the pH optimum of this overall protease activity is approximately 5 (Sarkis et al., 1988; Geier et al., 1999). The pH optimum of the PHO-1 acid phosphatase, which is GPI-anchored in the intestinal lumen, is approximately 4 (Beh et al., 1991). These two observations suggest that the intestinal lumen must be mildly acidic. Considering that roughly half of the intestinal contents are ejected with each defecation cycle of less than one minute (Avery and Shtonda, 2003; Ghafouri and McGhee, 2007) and considering that C. elegans can live in highly buffered media, the acidification capacity of the intestine must be enormous. Indeed, several vacuolar ATPase subunits are highly expressed in the intestine and, in particular, the VHA-6 protein is localized to the apical surface of the intestine, where it is suggested to acidify the intestinal lumen as part of the mechanism of nutrient uptake (Oka et al., 2001).

Proteases/peptidases appear to be more prominent than any other class of lumenal digestive hydrolases. For example, 18 of the top 80 genes expressed largely (or exclusively) in the adult intestine are proteases/peptidases and five of these are aspartic proteases (McGhee et al., 2007). Indeed, >90% of the protease activity in crude C. elegans extracts can be inhibited by pepstatin, a specific inhibitor of aspartic proteases (Geier et al., 1999). Probably the most thoroughly characterized digestive enzyme in C. elegans is the aspartic protease ASP-1, which was shown to be expressed exclusively in the intestine; ASP-1 appears late in embryogenesis, reaches peak levels during the first two larval stages and declines thereafter (Tcherepanova et al., 2000). Antibody staining indicated that most ASP-1 protein is in the intestinal lumen, although some protein could also be detected in lysosome-like cytoplasmic punctae (Tcherepanova et al., 2000). Other peptidases, some apparently quite specialized, are also highly expressed in the intestine: the PAM-1 puromycin sensitive aminopeptidase (Brooks et al., 2003), the APP-1 aminopeptidase (Laurent et al., 2001) and the LAP-1 leucine aminopeptidase (Joshua, 2001).

Nematodes rely heavily on lipids for energy storage and energy metabolism (Behm, 2002) and the intestine appears to be a major site of these biochemical activities. Lipases are highly represented on the SAGE list of intestinally-expressed genes and a number of these are likely to be secreted and function in lumenal digestion of lipids (McGhee et al., 2007). It was mentioned in the previous section that peroxisomes are one of the two sites of fatty acid oxidation in C. elegans and most peroxisomes are in the intestine. A number of enzymes involved in lipid metabolism are highly expressed in the intestine: for example, cln-3.1 is homologous to a human gene involved in the degenerative disorder neuronal ceroid lipofuscinosis (de Voer et al., 2005) and the spl-1 gene encodes sphingosine phosphate lyase (Mendel et al., 2003; Oskouian et al., 2005). The ELO-2, ELO-5 and ELO-6 enzymes have fatty acid elongation activities and are involved in the synthesis of polyunsaturated and branched fatty acids (Kniazeva et al., 2003; Kniazeva et al., 2004). C. elegans has three Δ-9-desaturase enzymes, expressed primarily (fat-5 and fat-6) or exlusively (fat-7) in the intestine (Brock et al., 2006); single knockouts have little phenotype but the triple knockout is lethal, demonstrating an overall requirement of C. elegans for adequate levels of unsaturated fatty acids (Brock et al., 2006). Finally, the GES-1 nonspecific carboxylesterase resides in the intestinal endoplasmic reticulum (McGhee, 1987; Kennedy et al., 1993) and could function in lipid metabolism and/or detoxification.

The following list contains a brief description of other genes that are all expressed highly and often exclusively in the intestine. The list provides a hint of the enormous metabolic diversity of the C. elegans intestine:

Combined with the C. elegans strengths of classical genetics and RNAi, recent technical advances (e.g., the ability to image intracellular fluorescent dyes, fluorescent substrates and environmentally responsive GFP variants and the ability to directly investigate the electrophysiology of dissected intestines or intestinal cells) hold great promise for understanding nutrient transport, ionic homeostasis and complex behaviours such as defecation.

The C. elegans genome contains nine genes encoding sodium-hydrogen exchangers and three of the genes are expressed exclusively in the intestine (Nehrke and Melvin, 2002). Translational fusions with GFP show that NHX-2 is located on the apical membrane, NHX-7 is located in the basolateral membrane and NHX-6 appears to be a mixture, basolateral at the ends of the intestine and apical in the central. Perfusion experiments, combined with intestine-specific expression of pH-sensitive GFP variants, were used to investigate how C. elegans controls the intracellular pH of the intestine and, more particularly, how the NHX-2 exchanger compensates for the acidification caused by uptake of food-derived peptides (Nehrke, 2003). The OPT-2/PEP-2 protein is the major C. elegans dipeptide transporter and is also located on the lumenal surface of the intestine (Nehrke, 2003; Meissner et al., 2004).

The intestine is the most likely site for the receptor/transporter responsible for uptake of double-stranded RNA from the environment when dsRNA mediated interference (RNAi) is performed by feeding or soaking. Several candidate genes have been described. SID-1 is a multi-transmembrane protein located at the periphery of most C. elegans cells but appears especially enriched in cells, such as the intestine, exposed to the environment (Winston et al., 2002a). SID-1 functions cell autonomously, is necessary for systemic RNAi and can confer dsRNA uptake capabilities on heterologous cells (Winston et al., 2002b; Feinberg and Hunter, 2003). SID-2 is less well characterized but is also required for effective RNAi following feeding or soaking and is strongly expressed in the intestine (personal communication from Craig Hunter, cited in Britton and Murray (2006)). The haf-6 gene encodes a half molecule ABC transporter expressed predominantly in the intestine and germline; haf-6 mutants are especially defective in RNAi administered by feeding (Sundaram et al., 2006). A further eight ABC transporter genes (of the approximately 60 such genes in C. elegans) show effects in RNAi uptake or spreading and several of these are expressed in the intestine (Sundaram et al., 2006).

Brief descriptions of other intestinal transporters are provided in the following list:

The C. elegans defecation motor program coordinates events that occur in the intestine, in muscles (especially enteric muscles) and in a small number of neurons (Avery and Thomas, 1997). A recent review (Branicky and Hekimi, 2006) lists >50 genes acting in one phase or another of the pathway and approximately 20 of these genes act in the intestine.

An important and striking finding was that the basic 45-50 second interval of the defecation cycle is associated with cyclic fluctuation in calcium ion levels in the posterior intestine; moreover, these calcium fluctuations reflect the behaviour of the inositol trisphosphate receptor ITR-1 (Dal Santo et al., 1999). Recently, powerful new techniques (e.g., electrophysiology, loading with calcium sensitive dyes, expression of calcium-sensitive GFP variants, combined with RNAi) have been applied to understand the molecular basis of these calcium fluctuations. Distinct calcium conductances have been identified by patch-clamping isolated intestinal cells (Estevez et al., 2003; Estevez and Strange, 2005). Dissected intestines (as long as they remain attached at the posterior to the carcass) are capable of sustained rhythmic oscillations in calcium ion levels (Espelt et al., 2005; Teramoto and Iwasaki, 2006). The rise in calcium ion levels closely corresponds to the pBoc phase of the defecation motor program (Espelt et al., 2005; Teramoto and Iwasaki, 2006). Waves of fluorescence reflecting transient increases in calcium ions can be observed to propagate from the posterior to the anterior intestine (Espelt et al., 2005; Teramoto and Iwasaki, 2006); the oscillations in fluorescence appear tightly focused on the apical region of the intestinal cell (Espelt et al., 2005). An example of this remarkable behaviour is shown in Movie 1 (Teramoto and Iwasaki, 2006). Blocking propagation of the calcium wave half-way down the intestine halts the defecation motor program, suggesting that the calcium wave is responsible for triggering the later muscle and neuronal phases (Teramoto and Iwasaki, 2006). Espelt et al. (2005) have extended such assays to show that only two of the six phospholipase C genes in C. elegans, egl-8 and plc-3, influence the defecation behaviour, either cycle time or cycle regularity. It was mentioned earlier that egl-8 is expressed highly in the intestine, with the EGL-8 enzyme concentrated near the apical junctions (Miller et al., 1999); plc-3 is also expressed in the intestine (Espelt et al., 2005). Intestine-specific RNAi was used to demonstrate that egl-8, plc-3 and itr-1 do indeed function in the intestine. It was further suggested that PLC-3 is responsible for generating the inositol triphosphates that control ITR-1 activity; in contrast, EGL-8 seems to function in a separate pathway to maintain the rhythmic oscillations in calcium ion (Espelt et al., 2005).

flr-1 mutants were originally isolated on the basis of resistance to fluoride (Katsura et al., 1994) but turned out to have a short defecation interval. FLR-1 is a degenerin/epithelial sodium channel type of protein, expressed only in the intestine and possibly concentrated at apical membranes (Take-Uchi et al., 1998). The flr-4 gene encodes an apparently membrane-associated kinase that is expressed in the intestine (+ pharynx and several neurons) beginning from the comma stage; only intestinal expression is necessary for mutant rescue. GFP translational fusions to FLR-4 appear localized to the enterocyte membranes, especially the lateral membranes (Take-Uchi et al., 2005). Proteins binding to the ITR-1 protein in a yeast two-hybrid assay have also been identified in the intestine (and other tissues; Walker et al., 2004). All of the above genes are candidates for participating in signals that originate from the oscillating calcium levels in the posterior intestine and that result in contraction of the posterior body wall muscles (the pBoc phase of defecation). Cycle times are influenced by additional intestinally-expressed genes, for example, genes that influence fatty acid composition (Branicky and Hekimi, 2006).

The C. elegans intestine becomes active in endocytosis at least by the 8E cell stage, when there is transfer of yolk into the intestine primordium from the remainder of the embryo (Bossinger and Schierenberg, 1996; Yu et al., 2006). Once the animal hatches and begins to feed, nutrient uptake and trafficking must become an even larger fraction of intestinal activity, one that is just beginning to be investigated. Digestive and protective molecules must be secreted into the intestinal lumen and displayed on the apical plasma membranes. Nutrients must be taken up, either by endocytosis or by dedicated transporters. Processed food must be passed on to the rest of the animal through basolateral domains, either as building blocks or as synthesized products such as the vitellogenins. Similarly, there must be a variety of molecules, ranging from waste products to signals, passing in the opposite direction, back across the intestine basal surface from the rest of the animal. Thus one would expect that the C. elegans intestine is a hotbed of endocytosis, exocytosis and trafficking in general. Consistent with such a view, the intestine cell cytoplasm is filled with rough endoplasmic reticulum but smooth ER is either rare or absent (Rolls et al., 2002).

The process of endocytosis has been most productively studied in oocytes and in coelomocytes (Fares and Grant, 2002) but the identified genes are probably common to most C. elegans cells and indeed are now beginning to be identified in the intestine (Sato et al., 2005; Chen et al., 2006). See also the WormBook chapter “Intracellular trafficking” which discusses the general topic of intracellular trafficking in C. elegans in the detail it deserves.

Transport of a toxic derivative of linolenic acid from the intestine to the gonad may involve proteins such as NRF-5, NRF-6 and NDG-4 (Watts and Browse, 2006). The nrf genes were originally identified because of an altered response to Prozac (fluoxetine) (Choy and Thomas, 1999) and one class of these genes is proposed to be involved in transport of hydrophobic molecules from the intestine to the rest of the animal (Choy et al., 2006). In particular, NRF-5 is a secreted lipid binding protein similar to the cholesterol ester binding protein of higher organisms (Choy et al., 2006).

Considering the C. elegans habitat and dietary practices, it is not surprising that the intestine is a major site of response to toxins and chemical stressors in the environment. The C. elegans genome has two metallothionein genes (mtl-1 and mtl-2), both encoding proteins that are cysteine-rich but otherwise not closely related to metallothioneins described in other organisms (Slice et al., 1990; Freedman et al., 1993). Both genes are induced at the transcriptional level by exposure to low levels of cadmium (in the dose range of 1-100 micromolar), with a response time on the order of hours. The mtl-2 gene is expressed exclusively in the intestine, from L1 larvae through to adults, and either metal exposure or heat shock are required for detectable expression. The mtl-1 gene is also inducible in the intestine but has a minor constitutive component in several cells of the pharynx (Freedman et al., 1993). Gene knockouts show that MTL-1 and MTL-2 do indeed protect against cadmium toxicity (Swain et al., 2004).

There are other pathways besides metallothioneins that protect C. elegans against metal toxicity and several dozen genes responding to cadmium exposure have been identified (Liao and Freedman, 1998). One gene in particular (called cdr-1 for cadmium responsive gene) has been studied in greater detail (Liao et al., 2002). Like mtl-1 and mtl-2, cdr-1 is induced at the transcriptional level with a response time of hours but induction is highly specific for cadmium; expression of cdr-1 is exclusively in the intestine in all stages post-hatching. The CDR-1 protein is highly hydrophobic, appears novel and has been located in intestinal lysosomes (Liao et al., 2002). RNAi-induced loss-of-function shows that CDR-1 does indeed protect against cadmium toxicity. A series of related genes (cdr-2 through cdr-7) have now been identified (Dong et al., 2005) but it is not known whether these genes are also expressed in the intestine.

The maternally-provided SKN-1 transcription factor has been well studied for its role in specification of EMS cell fate in the early embryo but SKN-1 has long been suspected of having an additional role in intestinal differentiation and/or intestinal function (Bowerman et al., 1992; Bowerman et al., 1993). The proposed somatic function of SKN-1 is to regulate a bank of genes, possibly all intestinal, that are involved in “phase II” detoxification and that provide protection against reactive oxygen species and other environmental toxins (An and Blackwell, 2003). One example of such a gene is gcs-1, encoding the enzyme glutamylcysteine synthetase involved in the production of glutathione. A gcs-1::GFP reporter construct is expressed in the intestine, pharynx and ASI neurons; intestinal expression is induced by heat or by oxidative stress and is abolished either by mutating a SKN-1 binding site in the gcs-1 promoter or by abolishing SKN-1 activity (An and Blackwell, 2003). Thermal or oxidative stress causes a SKN-1::GFP reporter protein to relocate from intestinal cytoplasm to intestinal nuclei within minutes (see Figure 4). The distribution of SKN-1 protein between nucleus and cytoplasm reflects competition between two kinases: glycogen synthase kinase-3 and p38 MAP kinase (An et al., 2005; Inoue et al., 2005). Consistent with a function of SKN-1 in regulating intestinal detoxification genes, skn-1 mutants show shorter lifespans and are more sensitive to agents such as paraquat (An and Blackwell, 2003).

The C. elegans genome contains approximately 80 genes encoding cytochrome P450 proteins, among whose roles is the metabolism of “xenobiotics”. A number of C. elegans P450 genes respond to compounds that induce mammalian cytochrome P450 genes (Menzel et al., 2001). Transcription of one gene in particular, CYP35A2, is strongly induced by exposure to ß-naphthoflavone and expression is exclusively in the intestine.

The four C. elegans P-glycoproteins (PGP-1 to PGP-4) belong to a family of membrane transporters containing an ATP-binding cassette. P-glycoproteins are often found on absorptive/excretory epithelial surfaces in higher organisms, where they are involved in the excretion of large hydrophobic toxins encountered in the diet or, for humans, encountered during chemotherapy. The pgp-1 and pgp-3 genes are expressed heavily in the C. elegans intestine throughout the lifespan (Lincke et al., 1993); antibody staining reveals that the proteins are located in the apical membranes facing the intestinal lumen (Broeks et al., 1995). pgp-3 is also expressed highly in the excretory cell. A deletion of the pgp-3 gene increases sensitivity to colchicines and to chloroquine (Broeks et al., 1995). The pgp genes, together with the related mdr genes, also contribute to protection against heavy metals (Broeks et al., 1996).

During the past decade, an entirely new field has developed in which C. elegans is used as a host to study the effects of pathogenic micro-organisms. The response of C. elegans to such pathogens is varied, ranging from no apparent effect up to killing within hours (in this case, usually because of a secreted toxin; see below). Probably the most common observation is that pathogenic bacteria are able to visibly and persistently colonize the intestine; indeed, a summary of responses of C. elegans to a series of pathogenic microbes (28 bacteria and two fungi) lists intestine as the main target (Schulenburg et al., 2004). This infectious interaction between microbes and C. elegans is beyond the scope of the present chapter and here we focus only on the best understood example of how the C. elegans intestine responds to a secreted bacterial toxin.

Aroian and coworkers have used C. elegans to study interactions with a set of crystal toxins produced by Bacillus thuringiensis (of agricultural fame) and the intestine appears to be the major target; see also the earlier work of Borgonie and coworkers (Borgonie et al., 1995; Borgonie et al., 1996). Exposure of C. elegans to purified toxin causes the intestine to become vacuolated within hours, followed ultimately by death (Marroquin et al., 2000). A series of BT-toxin-resistant mutants (bre-1 to bre-5) were isolated and the corresponding genes subsequently cloned. The major conclusion is that toxin resistance is associated with a series of glycosyltransferases, some of which are highly related to Drosophila genes with apparently similar function (Marroquin et al., 2000; Griffitts et al., 2001; Griffitts et al., 2003; Griffitts et al., 2005). The “