FLR-2, the glycoprotein hormone alpha subunit, is involved in the neural control of intestinal functions in Caenorhabditis elegans

Authors

  • Akane Oishi,

    1. Structural Biology Center, National Institute of Genetics, Mishima 411-8540, Japan
    2. Technical Section, National Institute of Genetics, Mishima 411-8540, Japan
    Search for more papers by this author
  • Keiko Gengyo-Ando,

    1. Department of Physiology, Tokyo Women’s Medical University School of Medicine, 8-1 Kawada-cho, Shinjuku-ku, Tokyo 162-8666, Japan
    2. Core Research of Evolutional Science & Technology (CREST), Japan Science and Technology Agency (JST), Tokyo, Japan
    Search for more papers by this author
  • Shohei Mitani,

    1. Department of Physiology, Tokyo Women’s Medical University School of Medicine, 8-1 Kawada-cho, Shinjuku-ku, Tokyo 162-8666, Japan
    2. Core Research of Evolutional Science & Technology (CREST), Japan Science and Technology Agency (JST), Tokyo, Japan
    Search for more papers by this author
  • Akiko Mohri-Shiomi,

    1. Structural Biology Center, National Institute of Genetics, Mishima 411-8540, Japan
    Search for more papers by this author
    • a

      Present address: Division of Brain Function, National Institute of Genetics, Mishima 411-8540, Japan.

  • Koutarou D Kimura,

    1. Structural Biology Center, National Institute of Genetics, Mishima 411-8540, Japan
    2. Department of Genetics, The Graduate University for Advanced Studies (Sokendai), Mishima 411-8540, Japan
    Search for more papers by this author
    • b

      Present address: Department of Biological Sciences, Graduate School of Science, Osaka University, 1-1 Machikaneyama, Toyonaka, Osaka 560-0043, Japan.

  • Takeshi Ishihara,

    1. Structural Biology Center, National Institute of Genetics, Mishima 411-8540, Japan
    2. Department of Genetics, The Graduate University for Advanced Studies (Sokendai), Mishima 411-8540, Japan
    Search for more papers by this author
    • c

      Present address: Department of Biology, Graduate School of Science, Kyushu University, 6-10-1 Hakozaki, Higashi-ku, Fukuoka 812-8581, Japan.

  • Isao Katsura

    Corresponding author
    1. Structural Biology Center, National Institute of Genetics, Mishima 411-8540, Japan
    2. Department of Genetics, The Graduate University for Advanced Studies (Sokendai), Mishima 411-8540, Japan
    Search for more papers by this author

  • Communicated by: Eisuke Nishida

* Correspondence: ikatsura@lab.nig.ac.jp

Abstract

The intestine plays an essential role in organism-wide regulatory networks in both vertebrates and invertebrates. In Caenorhabditis elegans, class 1 flr genes (flr-1, flr-3 and flr-4) act in the intestine and control growth rates and defecation cycle periods, while class 2 flr genes (flr-2, flr-5, flr-6 and flr-7) are characterized by mutations that suppress the slow growth of class 1 flr mutants. This study revealed that flr-2 gene controls antibacterial defense and intestinal color, confirming that flr-2 regulates intestinal functions. flr-2 encoded the only glycoprotein hormone alpha subunit in C. elegans and was expressed in certain neurons. Furthermore, FLR-2 bound to another secretory protein GHI-1, which belongs to a family of lipid- and lipopolysaccharide-binding proteins. A ghi-1 deletion mutation partially suppressed the short defecation cycle periods of class 1 flr mutants, and this effect was enhanced by flr-2 mutations. Thus, FLR-2 acts as a signaling molecule for the neural control of intestinal functions, which is achieved in a functional network involving class 1 and class 2 flr genes as well as ghi-1. These results are informative to studies of glycoprotein hormone signaling in higher animals.

Introduction

The intestine is not only an organ for food digestion and nutrition uptake, but it also plays an important role in the regulatory network of the whole body (Bueno & Fioramonti 1994; Migrenne et al. 2006). It receives many kinds of environmental cues from the food that animals have ingested, and sends the information to other parts of the body including the nervous system. Conversely, the nervous system regulates intestinal functions depending on the information received by the intestine as well as many sensory organs. Such interaction between the intestine and the nervous system is important for balanced responses as a whole organism and for defense against various stresses. Impairment of the interaction in humans results in difficulties, such as irritable bowel syndrome (Mulak & Bonaz 2004; Wood 2007) and obesity (Badman & Flier 2005). Research in this field would be facilitated by introducing a simple model-organism as the experimental material, which would also give insight into the evolutionary origin of those regulatory systems.

In the model-organism Caenorhabditis elegans, the intestine is also known to perform various organism-wide regulatory functions, such as defecation behavior, stress responses, interaction with pathogens and regulation of aging (McGhee 2007). Some of these functions are controlled by class 1 flr genes (flr-1, flr-3 and flr-4), which encode a Na+ channel of the DEG/ENaC family, a kinase-like molecule and a putative Ser/Thr protein kinase respectively (Katsura et al. 1994; Iwasaki et al. 1995; Take-uchi et al. 1998, 2005; Kawakami et al., unpublished data). Mutants in these genes exhibit slow larval growth, small brood sizes, strong resistance to fluoride ions, short defecation cycle periods, frequent skip of the expulsion step of defecation motor program, and tendency to form the dauer larva, a special third stage larva that appears under harsh conditions. Expression of these genes in the intestine is essential for the functions. Considering the molecular nature of the class 1 flr gene products and the results that defecation cycle is controlled by calcium ions in intestinal cells (Dal Santo et al. 1999), class 1 flr gene products seem to control various functions by regulating the ion concentrations and/or membrane potential of intestinal cells.

Many regulatory systems consist of two groups of genes that have opposing functions, and this may be the case also for the regulatory system involving class 1 flr genes. Class 1 flr genes genetically interact with another class of genes, called class 2 flr genes (flr-2, flr-5, flr-6, and flr-7), whose mutations confer weak resistance to fluoride ions (Katsura et al. 1994; I. Katsura, unpublished data). Mutations in these genes suppress some of the class 1 flr phenotypes, including slow larval growth, small brood sizes, and dauer larva formation abnormality (Katsura et al. 1994; Take-uchi et al. 1998). How class 2 flr genes interact with class 1 flr genes and control these functions is an important issue in the regulation of intestinal functions. To elucidate the mechanism at the cellular and molecular level, however, the following problems must be solved. (i) Although the genetic interaction with class 1 flr genes suggests that class 2 flr genes may control intestinal functions, it has not been shown directly. (ii) None of the class 2 flr genes has been cloned.

In this study, we therefore investigated the effect of flr-2 mutations on intestinal functions/properties and discovered that flr-2 controls antibacterial defense and intestinal color. We then cloned flr-2 gene and found that it encodes the only alpha subunit of the glycoprotein hormone in the C. elegans genome, which shows strong homology to the mammalian glycoprotein hormone alpha 2 (GPA2). flr-2 gene was expressed in some neurons in the pharynx, head and tail. Furthermore, we identified a secretory protein that physically and functionally interacts with FLR-2. This protein, which we named GHI-1, belongs to a family of lipid- and lipopolysaccharide-binding proteins (LBP) and was expressed in the body wall muscles and posterior intestine. Although a deletion mutation in ghi-1 gene did not suppress the slow growth of class 1 flr mutants, unlike flr-2 mutations, it partially suppressed the short defecation cycle periods of flr-4 mutants, and this effect was enhanced by flr-2 mutations. Thus, the glycoprotein hormone in C. elegans acts as a signaling molecule for the neural control of intestinal functions, which is achieved in a functional network involving class 1 and class 2 flr genes as well as ghi-1.

Results

flr-2 controls antibacterial defense

Previous studies revealed that mutations in class 2 flr genes (flr-2, flr-5, flr-6 and flr-7) suppress some phenotypes of mutations in class 1 flr genes (flr-1, flr-3 and flr-4), which are expressed in the intestine and considered to control the physiology of intestinal cells (Katsura et al. 1994; Take-uchi et al. 1998; I. Katsura, unpublished results). Furthermore, the phenotypes of the class 2 flr genes, especially the suppression of slow larval growth and small brood sizes suggest suppression of malnutrition. These results indicate that class 2 flr genes are likely to control intestinal functions, like class 1 flr genes. However, we considered that the evidence was not enough.

We therefore investigated antibacterial defense against Escherichia coli OP50 as an intestinal function. It is known that live OP50 shortens the lifespan of C. elegans by proliferating in the intestine of aged animals (Garigan et al. 2002). This effect is much more prominent for mutants defective in antibacterial defense (e.g. dbl-1 mutants) than the wild type (Mallo et al. 2002). Furthermore, antibacterial defense can be distinguished from longevity itself by measuring lifespan on dead OP50, such as OP50 on plates containing the DNA synthesis inhibitor 5-fluoro-2′-deoxyuridine (FUdR), because dead OP50 does not shorten the lifespan of antibacterial defense mutants specifically.

Based on these ideas, we measured the lifespan of the flr-2(ut5) mutant on NGM plates with live OP50. As shown in Fig. 1a, the flr-2 mutant showed a shorter lifespan than the wild-type N2 strain (Fig. 1a). The flr-5(ut73) mutant also showed a reduced lifespan, suggesting that this phenotype is probably common to all the class 2 flr mutants. On the other hand, the flr-1(ut11) mutant showed essentially a normal lifespan, and the flr-1 mutation did not suppress the short lifespan of flr-2(ut5). Then, to determine whether the short lifespan of the flr-2 mutant is due to defects in antibacterial defense or reduction in longevity itself, we measured the lifespan of the flr-2 mutant and wild-type animals on plates containing FUdR. In this assay, the flr-2 mutant as well as the flr-2;flr-1 double mutant greatly increased its lifespan and lived as long as the wild type, which only slightly increased its lifespan by FUdR (Fig. 1b). These results indicate that the short lifespan of flr-2 is due to defects in antibacterial defense rather than reduction of longevity itself.

Figure 1.

 Lifespan of the flr-2(ut5) mutant cultured with Escherichia coli OP50 on plates without or with 5-fluoro-2′-deoxyuridine (FUdR) and cultured with the pathogenic bacteria Enterococcus faecalis. (a) Lifespan of wild type, flr-2(ut5), flr-5(ut73), flr-2(ut5);flr-1(ut11) and flr-1(ut11) on NGM plates without FUdR. The average lifespan ± SEM (no. of animals tested) were 17.83 ± 1.01 (N = 24), 8.96 ± 0.49 (N = 28), 9.96 ± 0.69 (N = 25), 7.21 ± 0.48 (N = 24) and 16.95 ± 1.27 days (N = 22) for wild type, flr-2(ut5), flr-2(ut5); flr-1(ut11), and flr-1(ut11) respectively. (b) Lifespan of wild type, flr-2(ut5), and flr-2(ut5); flr-1(ut11) on NGM plates containing 1.6 mm FUdR. The average lifespan ± SEM (no. of animals tested) were 21.62 ± 1.33 (N = 21), 23.60 ± 1.50 (N = 20) and 22.71 ± 1.41 days (N = 21) for wild type, flr-2(ut5), and flr-2(ut5); flr-1(ut11) respectively. (c) Lifespan of wild type and flr-2(ut5) on NGM plates without FUdR. Those animals that did not form bags of worms for 9 days after birth were used for the measurements. The average lifespan ± SEM (no. of animals tested) were 19.23 ± 0.60 (N = 39), and 13.73 ± 0.47 (N = 37) for wild type and flr-2(ut5) respectively. (d) Lifespan of flr-2(ut5) animals that carried the wild-type flr-2 transgene (+Ex) and that had lost the transgene (−Ex). Those animals that did not form bags of worms for 9 days after birth were used for the measurements. The average lifespan ± SEM (no. of animals tested) were 15.74 ± 0.64 (N = 35) and 11.93 ± 0.41 days (N = 15) for flr-2+Ex and flr-2-Ex respectively. (e) Lifespan of wild type and flr-2(ut5) on plates with the pathogenic bacterium E. faecalis (killing assay). In these experiments the temperature was exceptionally 25 °C, at which temperature flr-2(ut5) animals lived also as long as the wild type on FUdR plates with OP50 (data not shown). The day when L4 animals were placed on the plates to test the killing activity of E. faecalis was defined as day 2, so that the age of the animals is the same as the number on the abscissa like in other figures. The average lifespan ± SEM (no. of animals tested) were 8.83 ± 0.31 (N = 58) and 5.60 ± 0.13 (N = 62) for wild type and flr-2(ut5) respectively.

Nonetheless, we were worried about another possibility. During the measurements of lifespan on plates without FUdR, we noticed that the flr-2, flr-5 and flr-2;flr-1 mutants frequently formed bags of worms caused by hatching of progeny worms in the uterus, unlike the wild type or flr-1 mutants. Because the larvae in the uterus eat and kill the parent animals, this may be the cause of the short lifespan of flr-2, flr-5 and flr-2;flr-1. Furthermore, FUdR suppressed the formation of bags of worms by inhibiting the development of embryos in the uterus, which may be the reason why flr-2 lived as long as the wild type on plates containing FUdR. To reject this possibility, we measured the lifespan of 10-day-old animals that had finished laying eggs and that did not form bags of worms, using NGM plates without FUdR. In this assay, the flr-2 mutant still showed a shorter lifespan than the wild type (Fig. 1c). We therefore concluded that the short lifespan of flr-2 mutants is not due to the formation of bag of worms, but to defects in antibacterial defense against OP50.

To obtain further evidence that flr-2 controls intestinal functions, we conducted a killing assay using the Gram-positive pathogenic bacterium Enterococcus faecalis OG1RF, which is known to proliferate in the worm intestine and kills it by the action of a quorum-sensing system and cytolysin (Garsin et al. 2001). The results showed that flr-2 mutants are hypersensitive not only to E. coli OP50 but also to E. faecalis OG1RF (Fig. 1e). Thus, flr-2 gene may play a role in innate immunity against a certain range of bacteria including both Gram-negative and Gram-positive, confirming that flr-2 controls intestinal functions.

Next, to learn whether flr-2 acts through the dbl-1 pathway for antibacterial defense, we examined the antibacterial defense of flr-2, sma-3, and sma-3;flr-2 mutants by the measurements of lifespan on NGM plates with live OP50. sma-3 encodes a Smad protein that acts downstream of DBL-1, one of the TGF-β’s in C. elegans (Savage-Dunn et al. 2000), where both dbl-1 and sma-3 mutants show defects in antibacterial defense (Mallo et al. 2002; Kurz & Tan 2004). The results showed that sma-3 mutants exhibited shorter lifespan than the wild type and flr-2 mutants on live OP50 (Fig. S2 in Supporting Information). Furthermore, sma-3;flr-2 double mutants lived as long as sma-3 mutants. This suggests that flr-2 may control antibacterial defense through the dbl-1 pathway, although more experiments are required to prove it unambiguously.

flr-2 controls intestinal color

We then took notice of intestinal color. By observation of worms under a dissecting microscope, we knew that class 1 flr mutants show a pale intestine and that class 2 flr mutations seem to suppress this phenotype. Although the functions determining intestinal color remain to be studied, we considered that quantitative measurements of intestinal color would demonstrate directly that flr-2 controls a property of the intestine.

The darkness of intestinal color was measured quantitatively as described in Experimental procedures. The images of wild type, flr-1, flr-2, and flr-2;flr-1 animals were taken with a dissecting microscope equipped with a digital camera, and the darkness of the intestinal color in the images was analyzed quantitatively with compensation for the thickness of the intestine. As shown in Fig. 2 and Fig. S1 in Supporting Information, two flr-2 mutations tested, ut5 and ut71, suppressed the pale intestine phenotype of the flr-1 mutant. While ut5 had a slightly darker intestine, ut71 did not show a significantly (P < 0.05) darker intestine than the wild type, which may suggest that ut5 is a stronger allele than ut71. Those results demonstrate that flr-2 mutations have an effect on the intestine.

Figure 2.

 The darkness of intestinal color in wild type (N2), flr-1(ut11), flr-2:flr-1(ut11) and flr-2 animals. (a) The allele ut5 was used as the flr-2 mutation. The average darkness ± SEM (no. of animals tested) was 52.6 ± 1.4 (N = 51), 46.9 ± 1.6 (N = 51), 60.0 ± 2.1 (N = 47) and 57.6 ± 1.3 (N = 64) in arbitrary units for wild type, flr-1(ut11), flr-2(ut5);flr-1(ut11), and flr-2(ut5) respectively. (b) The allele ut71 was use as the flr-2 mutation. The average darkness ± SEM (no. of animals tested) was 58.8 ± 0.9 (N = 53), 44.2 ± 1.4 (N = 51), 56.6 ± 1.0 (N = 62) and 60.2 ± 1.1 (N = 48) in arbitrary units for wild type, flr-1(ut11), flr-2(ut71);flr-1(ut11), and flr-2(ut71) respectively. The results of t-test are also shown.

flr-2 gene encodes a glycoprotein alpha 2 homolog

To elucidate the molecular mechanism of flr-2 functions, we cloned flr-2 gene by positional cloning (Fig. 3; See Experimental procedures for details). A 2.3-kb genomic PCR fragment containing only F58H1.4 rescued the phenotype of flr-2 in the flr-2(ut5);flr-1(ut11) double mutant: animals carrying the transgene grew slowly and showed the generation time of 5–8 days or longer, like the flr-1(ut11) mutant, whereas those without the transgene showed the generation time of 3–4 days (Fig. S3 in Supporting Information). The 2.3-kb fragment also rescued two other flr-2 phenotypes: a short lifespan on live OP50 (Fig. 1d) and the suppression of the dauer-constitutive phenotype of the flr-1(ut11) unc-3(e151) double mutant (Table S1 in Supporting Information). We therefore concluded that flr-2 is identical to F58H1.4.

Figure 3.

 Positional cloning of flr-2 gene. (a) (From top to bottom) Genetic map, cosmid clones in the region between daf-11 and vab-8, ORFs in the region shared by R31 and F58H1, and DNA fragments used for rescue experiments are shown together with the results of rescue experiments. (b) The structure of flr-2 gene. Solid and open boxes represent the coding sequence and the 5′- and 3′-untranslated sequences respectively. Splice junctions are indicated by Vs between adjacent exons.

The nucleotide sequence of flr-2 cDNA, which was obtained by RT-PCR, revealed that flr-2 encodes a secretory protein of 122 amino acid residues with a signal peptide sequence at the amino terminus. It was identical to the protein reported by Park et al. (2005) as a homolog of GPA2, the alpha subunit of thyrostimulin. In fact, proteins that were most similar to FLR-2 by BLAST search (except for FLR-2 homologs in other nematodes) were insect glycoprotein alpha 2 homologs, of which the Drosophila GPA2 was shown to activate glycoprotein hormone receptors (Sudo et al. 2005). FLR-2 also contained the Pfam-A DAN domain (http://pfam.sanger.ac.uk/family/DAN) and showed homology to some TGF-β antagonists such as DAN, Gremlin and Cerberus. However, FLR-2 resembled GPA2s more than the TGF-β antagonists in the following aspects, besides the BLAST scores (Fig. 4). The 5th and 6th conserved cysteine residues are adjacent in FLR-2 and GPA2s, whereas they are separated by two other amino acid residues in the TGF-β antagonists (indicated by $ in Fig. 4). Furthermore, some amino acid residues were conserved among GPA2s from various species (including FLR-2) or among the TGF-β antagonists but not among all of them (indicated by # in Fig. 4).

Figure 4.

 Comparison of the amino acid sequences of FLR-2, Drosophila glycoprotein hormone alpha 2, human glycoprotein hormone alpha 2, F35B12.10 (C. elegans homolog of DAN/Gremlin/Cerberus), mouse DAN, Xenopus Gremlin and Xenopus Cerberus. All the cysteine residues are shown by white letters in black boxes, whereas those residues that show difference between GPH2s and the TGF-β antagonists are indicated by ‘$’. Boxed residues indicate the amino acids conserved among GPH2s (top three lines), among the TGF-β antagonists (bottom four lines) or among both of them, where the former two cases are indicated by ‘#’. The amino acid substitutions in flr-2(ut5) and flr-2(ut71) are indicated by asterisks. Numbers in parentheses indicate the numbers of amino acid residues that are not shown. ‘-’ indicates a gap. The DDBJ/EMBL/GenBank accession numbers of the sequences are AB491765 (FLR-2), AY940435 (GPA2(Dm)), NM130769 (GPHA2(Hs)), NM001028705 (F35B12.10), D50263 (DAN), BC169983 (gremlin) and NM001088331 (cerberus).

Sequencing of the two flr-2 mutants ut5 and ut71 revealed that they are missense mutants in conserved amino acid residues: Gly63 to Glu for ut5 and Cys114 to Tyr for ut71. We tentatively think that these mutations are hypomorphic or null, because they are recessive for all the phenotypes tested and because there is so far no evidence indicating that they are antimorphic, neomorphic or hypermorphic. It is consistent with our unpublished results that deletion mutations in the beta subunit and the receptor, respectively, of the glycoprotein hormone show essentially the same phenotypes as these flr-2 mutations.

flr-2 is expressed in neurons

The expression of flr-2 was examined by using flr-2p::GFP, flr-2p::VENUS (improved YFP; Nagai et al. 2002) and specific polyclonal antibodies. flr-2p::GFP and flr-2p::VENUS were expressed in some neurons in the pharynx, head and tail from L1 larvae to adults (Fig. 5). These neurons were tentatively identified as pharyngeal motor neurons M1, M2L/R, M5, NSML/R, head neurons AIAL/R and tail neurons LUAL/R. Double staining of worms carrying flr-2p::GFP with anti-GFP and anti-FLR-2 antibodies revealed that the cells stained with anti-FLR-2 were the same as those stained with anti-GFP. Thus, all the cells that we detected the expression of FLR-2 were neurons. This observation is consistent with our results that the feeding RNAi of flr-2, unlike mutations in flr-2 or the feeding RNAi of the glycoprotein hormone receptor gene fshr-1, did not suppress the slow growth of flr-1 mutants (data not shown). In C. elegans, it is known that feeding RNAi of neuronal genes does not work efficiently, unless the animal has a special mutation that enhances RNAi in neurons (Schmitz et al. 2007).

Figure 5.

 Expression of FLR-2. Micrographs of wild-type animals carrying flr-2p::venus. (a) (b): head, (c), (d): tail. (a), (c): DIC contrast micrographs, (b), (d): fluorescence micrographs. The scale bar indicates 50 μm.

To identify the cells in which flr-2 gene expression is sufficient for the wild-type phenotype, flr-2 cDNA was expressed with the extrinsic tissue-specific promoters pH20 (neuron), pges-1 (intestine), and pmyo-3 (muscle) in flr-2(ut5); flr-1(ut11) double mutant animals. The expression with each of these promoters increased the percentages of slow-growing animals: those animals that did not reach the L4 stage within 3 days increased from 0% (no transgene) to 46% (pH20), 57% (pges-1), and 68% (pmyo-3), respectively, indicating that the flr-2 phenotype (suppression of the slow growth of flr-1) was rescued. The results show that the expression of flr-2 in neurons, intestinal cells, or muscle cells is sufficient for the function of flr-2, although flr-2 is expressed only in some neurons in wild-type animals. This is consistent with the idea that FLR-2 is secreted from those cells into the pseudocoelom before it performs its function.

FLR-2 physically interacts with GHI-1 protein

Identification of proteins that interact with FLR-2 may reveal how FLR-2 performs its function and/or how the activity of FLR-2 is controlled. To look for such proteins, we conducted expression cloning by using cultured cells (Flanagan & Cheng 2000). For this purpose, a cDNA library was prepared with RNA from C. elegans primary cultured cells and transfected on COS cells. The cells were then fixed with paraformaldehyde to allow identification of FLR-2-binding soluble proteins in addition to membrane proteins. Finally, the cells that bound FLR-2::AP (FLR-2-alkaline phosphatase fusion protein heterologously expressed in HEK293 cells) were identified by detection of alkaline phosphatase activity. By screening 192 000 clones, one cDNA clone that produced a FLR-2::AP-binding protein was identified. Sequencing of the cDNA clone revealed that it corresponded to ZK20.1, which we named ghi-1 after glycoprotein hormone-interacting protein-1.

The sequences of the cDNA clone obtained by the expression cloning and the cDNA donated by Yuji Kohara (yk1207.a08, with the SL1 trans-splicing leader sequence at the 5′-end) revealed that ghi-1 encodes a secretory protein of 434 amino acids with a signal peptide. Although simple BLAST search of GHI-1 did not yield meaningful results, PSI-BLAST search starting from its C. briggsae homolog (CBP21173) revealed that GHI-1 has weak overall homology to proteins of the LBP/BPI/CETP family (Bingle & Craven 2004), which include mammalian lipopolysaccharide-binding and lipid transport proteins, such as LBP, bactericidal permeability-increasing protein (BPI), cholesteryl ester transfer protein (CETP) and phospholipid transfer protein (PLTP) (Fig. 6 and data not shown). Of these proteins, LBP and BPI bind to lipopolysaccharides on some Gram-negative bacteria and kills them directly (BPI) or by mediating attachment to monocytes and macrophages (LBP). CETP and PLTP function in the transfer of lipids between plasma lipoprotein particles. The C. elegans genome encodes more than 10 proteins of this family, but most of them have not been studied except NRF-5, which is involved in nose resistance to fluoxetine, coelomocyte uptake, and transport of a toxic derivative of linolenic acid (Choy et al. 2006; Watts & Browse 2006).

Figure 6.

 Comparison of the amino acid sequences of Caenorhabditis elegans GHI-1, C. briggsae CBP21173 (GHI-1 homolog), human LBP and human PLTP. Boxed residues indicate the same or similar amino acids in at least three of the four proteins, where similar amino acids are (K, R), (D, E), (I, L, M, V) and (F, W, Y). Numbers in parentheses indicate the number of amino acid residues that are not shown. ‘-’ indicates a gap. The arrow on the top represents the region corresponding to the ghi-1 deletion mutation ut293, which extends to the first letter of the Ala32 codon. The DDBJ/EMBL/GenBank accession numbers of the sequences are NM064085 (GHI-1), XM001679378 (CBP21173), NM004139 (LBP) and AB076694 (PLTP).

The expression of ghi-1 was investigated by using a GFP-fusion gene. Animals carrying an extrachromosomal array of ghi-1p::GFP showed fluorescence in body wall muscles and one or two cells at the posterior end of the intestine (Fig. 7).

Figure 7.

 Expression of ghi-1 gene. Micrographs of a wild-type animal carrying ghi-1p::GFP. (a) DIC contrast micrograph, (b) fluorescence micrograph. GFP fluorescence was detected in body wall muscles and one or two cells in the posterior intestine. The latter expression may be an artifact, which is occasionally found for GFP fusion genes in Caenorhabditis elegans. The scale bar indicates 50 μm.

ghi-1 regulates defecation cycle periods in collaboration with flr-2

GHI-1 may have functional as well as physical interaction with FLR-2. To investigate the function of ghi-1, we isolated a deletion mutant, ut293, which had a deletion of 496 nucleotides including the SL1 trans-splice site and the initiation codon (402 bp in 5′-UTR and 94 bp in the coding region) of ghi-1 gene. This mutant, which is a putative null allele, showed roughly the same shape and movement as the wild type. It was clear that ghi-1 does not belong to class 2 flr genes; the ut293 mutation neither conferred resistance to fluoride ions, nor suppressed the slow growth of class 1 flr mutants (data not shown).

To learn whether ghi-1 gene controls intestinal functions, we examined the effect of the ghi-1 mutation on defecation cycle periods in some mutant backgrounds. As already reported (Take-uchi et al. 1998, 2005), flr-1 and flr-4 mutants as well as flr-2;flr-1 and flr-2;flr-4 double mutants showed short defecation cycle periods, while flr-2 mutants showed normal defecation cycle periods. The ghi-1 mutation significantly increased the defecation cycle periods of the weak flr-4 allele n2259 and those of the double mutant flr-2(ut5); flr-4(ut7), although it did not change the defecation cycle periods of the wild type, flr-2(ut5), flr-4(ut7), and flr-4(sa201) single mutant animals (Fig. 8). To demonstrate that such effect is due to ut293 itself and not due to a possible side mutation, we conducted RNAi experiments. The results showed that ghi-1(RNAi) increased the defecation cycle periods of flr-2(ut5);flr-4(n2259). ghi-1(RNAi) might also increase the defecation cycle periods of flr-4(n2259) and flr-2(ut5);flr-1(ut11), but the effect was not very significant (0.05 < P < 0.1). It was reported that class 2 flr mutations essentially do not suppress the short defecation cycle periods of strong class 1 flr alleles (Take-uchi et al. 1998). In contrast, the experiments of this study revealed that the flr-2(ut5) mutation significantly increases defecation cycle periods in ghi-1(ut293);flr-4(n2259) and ghi-1(ut293);flr-4(sa201) backgrounds. Thus, ghi-1 and flr-2 mutations showed synergistic effects on the suppression of the short defecation cycle periods of class 1 flr mutants.

Figure 8.

 Defecation cycle periods of the ghi-1(ut293) deletion mutant and ghi-1(RNAi) in various genetic backgrounds. In RNAi experiments (right side), animals without RNAi were grown on Escherichia coli carrying the feeding RNAi vector without inserts. The error bars show SEM. The results of t-test are shown by (+) and (±), where (+) means statistically significant difference (P < 0.02) and (±) means difference that may not be significant (0.05 < P < 0.1).

Discussion

flr-2 gene encodes the only alpha subunit of the glycoprotein hormone in the C. elegans genome

The flr-2 gene was identified as one of the class 2 flr genes (flr-2, flr-5, flr-6 and flr-7), whose mutations confer weak resistance to fluoride ions and suppress the slow larval growth of class 1 flr mutants (flr-1, flr-3 and flr-4) (Katsura et al. 1994; Take-uchi et al. 1998, 2005). These phenotypes suggest that the class 2 flr genes constitute a regulatory system that acts in opposite direction to the class 1 flr regulatory system, which controls diverse functions in the intestine.

This study revealed that flr-2 encodes a homolog of GPA2, the alpha subunit of thyrostimulin, a glycoprotein hormone discovered most recently (Nakabayashi et al. 2002). Mammals have five glycoprotein hormones: thyrotropin (TSH), follitropin (FSH), lutropin (LH), chorionic gonadotropin (CG) and thyrostimulin. Of these hormones, thyrostimulin consists of unique alpha and beta subunits, while the other hormones consist of a common alpha subunit, distinct from GPA2, and a hormone-specific beta subunit. Thyrostimulin, like TSH, FSH, and LH, is synthesized in the anterior pituitary, but also in other peripheral tissues such as the eye and the testis. Although thyrostimulin is known to activate the TSH receptor, its exact role remains to be elucidated.

Caenorhabditis elegans has each one gene for the alpha and beta subunit as well as the receptor of the glycoprotein hormone. Comparative studies on the alpha and beta subunits from nematodes to humans revealed that those in C. elegans are the ancient forms (Park et al. 2005). Studies on the function of these molecules, therefore, are expected to reveal the evolutionary origin or a prototype of glycoprotein hormone signaling. This study demonstrated that the glycoprotein hormone in C. elegans acts as a signaling molecule for the neural control of intestinal functions and regulates antibacterial defense, growth rates, intestinal color, defecation behavior, dauer larva formation, etc. which is unexpected from the functions of glycoprotein hormones in mammals. In C. elegans, the intestine has to perform many regulatory functions due to the lack of liver, pancreas, thyroid, etc. Moreover, it is not innervated. Therefore, the intestine must be an appropriate and indispensable target of hormones, and possibly the same was true for ancient, primitive metazoans.

FLR-2 acts as a signaling molecule for the neural control of intestinal functions

This study revealed a role of the glycoprotein hormone in C. elegans: it acts as a signaling molecule for the neural control of intestinal functions. FLR-2, the C. elegans GPA2, is expressed in neurons, as shown by GFP fusion genes and by staining with specific antibodies. On the other hand, flr-2 gene controls intestinal functions. This was already suggested by the results that flr-2 mutations suppress some phenotypes of class 1 flr genes, which regulate the physiology of intestinal cells. This study provided more direct evidence. First, flr-2 mutants showed defects in antibacterial defense against E. coli OP50 and E. faecalis OG1RF. Second, flr-2 mutations suppressed the pale intestinal color of flr-1 animals. The former evidence is consistent with the results of a recent paper showing that the glycoprotein hormone receptor FSHR-1 acts in intestinal cells for antibacterial defense against Pseudomonas aeruginosa (Powell et al. 2009). We also found that the expression of FSHR-1 in the intestine is essential for defense against E. faecalis and for the suppression of the slow growth of class 1 flr mutants (A. Mohri-Shiomi et al., manuscript in preparation).

The neural control of intestinal functions by flr-2 may be a part of a wider regulatory network in the organism. DNA microarray experiments revealed that the expression of flr-2/F58H1.4 is upregulated by the pathogenic bacteria E. faecalis and E. carotovora (Wong et al. 2007). On the other hand, we demonstrated in this study that flr-2 mutants are hypersensitive to E. faecalis. These results indicate that the enhanced expression of flr-2 gene by pathogenic bacteria may be part of the inducible antibacterial defense response reported by Mallo et al. (2002). Considering the results of this study, this response seems to involve signals from the intestine to neurons and back to the intestine. Recently, Styer et al. (2008) reported that neurons expressing NPR-1 control antibacterial defense. Relationship between this regulatory system and the flr-2 regulatory system remains to be studied, although there seems to be no overlap between neurons expressing NPR-1 and FLR-2.

FLR-2 may also control nonintestinal functions by acting directly on the glycoprotein hormone receptor FSHR-1 on cells other than intestinal cells. It is known that FSHR-1 is expressed in many types of cells and controls presynaptic function and germline differentiation/survival (Sieburth et al. 2005; Cho et al. 2007). In the latter case fshr-1 is known to act in the somatic gonad, but in the former case whether fshr-1 acts cell autonomously has not been determined. Although it is likely, it has not been demonstrated in either case that FLR-2 regulates FSHR-1.

FLR-2 physically and functionally interacts with another secretory protein, GHI-1

GHI-1 was identified in this study as a protein that physically interacts with FLR-2. It was expressed in body wall muscles and the posterior intestine, had a signal sequence, and showed weak but overall homology to proteins in the LBP/BPI/CETP family. These properties suggest that it may be secreted into the pseudocoelom and act in antibacterial defense like BPI and LBP or in the transfer of lipids like CETP and PLTP. We examined antibacterial defense defects in the ghi-1(ut293) mutant but could not detect it (our unpublished results). However, this does not necessarily mean that GHI-1 does not function in antibacterial defense, because it may be functionally redundant with some other proteins in the LBP/BPI/CETP family in C. elegans.

This study revealed that ghi-1 controls defecation cycle periods in collaboration with flr-2. The ghi-1 mutation partially suppressed the short defecation cycle periods of class 1 flr mutants, and this effect was enhanced by flr-2 mutations. In contrast to the similarity between ghi-1 and flr-2 in this function, the ghi-1 mutation neither conferred resistance to fluoride ions nor suppressed the slow growth of class 1 flr mutants, unlike flr-2 mutants.

The relationship between ghi-1 and flr-2 in regulatory pathways remains to be studied. FLR-2 most probably acts on the receptor FSHR-1 to control intestinal functions. GHI-1 may control intestinal functions by regulating the activity or stability of FLR-2. Although the ghi-1 mutation and the flr-2 mutation showed synergetic effects in defecation cycle periods, it does not necessarily contradict with this hypothesis, if the flr-2(ut5) mutation is a hypomorph. However, it is also possible that GHI-1 and FLR-2 may act through distinct pathways, which crosstalk each other by interaction between these two molecules.

Future prospects

This study gives a prospect of analyzing the molecular mechanism of glycoprotein hormone signaling in the regulatory network in the C. elegans intestine. Genes that act downstream of the glycoprotein hormone receptor or that modify the components of the signaling pathway can be identified by testing the phenotypes of their mutants, for instance, growth rates in the class 1 flr mutant background. Such studies will reveal how flr-2 gene interacts with class 1 flr genes. This line of experiments is in progress in our laboratory. We recently found that flr-5 encodes the C. elegans glycoprotein hormone beta subunit, while flr-6, a novel class 2 flr gene identified by forward genetics, encodes the C. elegans glycoprotein hormone receptor FSHR-1 (A. Mohri-Shiomi, A. Oishi, T. Ishihara, K. D. Kimura and I. Katsura, unpublished data). Thus, the glycoprotein hormone signaling seems to be carried out by the class 2 flr genes in C. elegans. Molecules downstream of FLR-2 in the glycoprotein signaling pathway are considered to antagonize class 1 flr functions in the intestine. How this is achieved is an important issue for elucidating the mechanism of regulation by class 2 flr genes.

This study revealed the roles of glycoprotein hormone in the neural control of intestinal functions and discovered a glycoprotein hormone-interacting protein. These results are informative to studies on human hormones as well as the evolution of hormones.

Experimental procedures

General methods and strains

Worm cultures and genetic studies were performed essentially as described by Brenner (1974). NGM plates and E. coli OP50 were used as the culture medium and the food, respectively. The worm strains used in this study were C. elegans var. Bristol N2 (wild type) and strains containing the following mutations: flr-1(ut11) X, flr-2(ut5) V, flr-2(ut71) V, flr-4(n2259) X, flr-4(sa201) X, flr-4(ut7) X, flr-5(ut73) V, ghi-1(ut293) II, sma-3(e491) III, unc-3(e151) X, unc-42(e270) arDf1 V/nT1[unc(n754) let](IV,V).

Measurements of lifespan

Two or three gravid hermaphrodites were placed on NGM plates seeded with OP50 for 6 h and then removed from the plates (0th day). The progeny worms were transferred to new plates every day while they laid eggs. The worms were checked for their death, which was judged by the absence of response to touch with a platinum wormpick. To test the effect of FUdR, NGM plates containing 1.6 mm FUdR were used from the 3rd to 10th days, while plates without FUdR were used before and after these days. The measurements of lifespan in the presence of E. faecalis OG1RF (killing assays) were performed as described by Garsin et al. (2001), except that 20 °C grown worms were used for the assays, which were conducted at 25 °C.

Measurements of the darkness of intestinal color

L4 hermaphrodite animals, cultured at 20 °C, were placed on agar pads and anesthetized with a drop of 50 mm sodium azide. Their images were taken with an Olympus SZH-131 dissecting microscope equipped with a digital camera Nicon COOLPIX 4500 (4.0 Megapixels) and analyzed with the software imagej (http://rsb.info.nih.gov/ij/). The optical density of the intestine in the images was measured at 150–250 points, depending on the length of the intestine from the anterior to the posterior end. These data were averaged, and the background density was subtracted. The averaged net optical density of the image of the intestine was divided by the body width of the worm to estimate the averaged darkness of the intestine per unit thickness, assuming that the thickness of the intestine is proportional to the body width.

Cloning of flr-2 gene

flr-2 was already mapped to the center-right region of LGV, a position near sma-1 and deleted by the deficiency ctDf1 (Katsura et al. 1994). In this study, we found that the deficiency arDf1 does not delete flr-2, which narrowed down the position of flr-2 within 18 cosmid clones. These clones were injected into flr-2(ut5); flr-1(ut11) double mutant animals to examine whether the transgene cancels the suppression of the slow growth phenotype of flr-1 by the flr-2 mutation. The cosmids R07A10 and R31 rescued the phenotype of flr-2 in the flr-2(ut5);flr-1(ut11) double mutant. Furthermore, a 2.3-kb genomic PCR fragment containing only F58H1.4 gene, which is present on both of the cosmids, rescued this phenotype as well as antibacterial defense defect of flr-2(ut5), and the suppression of the dauer-constitutive phenotype of the flr-1(ut11) unc-3(e151) double mutant by flr-2(ut5). In these experiments, the test DNA was injected with the marker myo-3::GFP into the gonad of corresponding mutants. The DNA concentrations were 1 ng/μL for cosmid clones and 10 ng/μL for PCR fragments amplified from N2 genomic DNA. The following primers were used to amplify the 2.3-kb genomic DNA containing F58H1.4: R31b-1 (GCAGATGATCAAACTGACGGTGTCC) and R31b-3 (GCGATAGCGAACAAGTACCATGAGG).

Isolation of flr-2 cDNA

The 5′ half of flr-2 cDNA was obtained by RT-PCR using the primers SL1 (GGTTTAATTACCCAAGTTTGAG) and s-2 (TGTTGCAGATGGTATTCG), which yielded a 0.4-kb fragment, while no DNA fragment was amplified with the primers SL2 (GGTTTTAACCCAGTTACTCAAG) and s-2. The 3′ half of the cDNA was obtained, using the primers s-3 (ACAGTACTGCACAGCAGG), n-s-3 (TTCGAATCAATCGATGC), and dT adaptor1 (TAGCACTAGTAACTCACGTCG). The cDNA sequence has been submitted to the DDBJ/EMBL/GenBank database (accession number AB491765).

DNA constructs for expression studies

flr-2p::GFP was made by inserting the two DNA fragments into the promoterless GFP vector pPD95.67 as follows. (i) The PCR fragment containing the 3.9-kb upstream sequence of flr-2, amplified with the primers f2gfp-b1 (CTCGGATCCGGTCTCTAGTACAGTTGACC) and f2gfp-b2 (CTCGGATCCGAGCCCATCATTTGAGAAGG), was inserted into pPD95.67 in frame with the GFP coding sequence, using the BamHI sites generated by the primers. (ii) The 1.4-kb flr-2 coding sequence, amplified with the primers f2gfp-a1 (TCTGGGCCCGCACGACGACGTTTAAGTTG) and f2gfp-a2 (TCTGGGCCCCAGATGATCAAACTGACGG) was inserted into the cloning site downstream of the GFP coding sequence, using the ApaI sites generated by the primers. The latter DNA sequence was required for easily detectable expression. flr-2p::VENUS (improved YFP; Nagai et al. 2002) was made by replacing the GFP part (KpnI-EagI fragment) of the flr-2p::GFP construct with the VENUS part (KpnI-EagI fragment) of pPD95.79-venus. For expression studies, these constructs were injected into wild-type worms at a concentration of 33 ng/μL, together with the injection marker rol-6(su1006dom) (33 ng/μL).

Expression studies with antibodies

Recombinant FLR-2 protein was synthesized using the NOVAGEN pET System in E. coli BL21 carrying the plasmid pET21b-flr-2cDNA, which contained the flr-2 cDNA without the signal peptide sequence. The His-tag::FLR-2 fusion protein, which formed inclusion bodies, was solubilized with 5M guanidine hydrochloride, dialyzed against 1× PBS, and purified. The anti-FLR-2 antibody was made by injecting the purified protein into rats by Hokudo Co., Ltd (Sapporo, Japan).

For antibody staining, worms were fixed with a Bowin’s fixative (0.75 mL saturated picric acid, 0.25 mL formalin, 0.05 mL glacial acetic acid and 0.01 mL distilled water) for 30 min, followed by 3× freeze-thawing. After washing the fixative, worms were stained with primary antibodies (anti-FLR-2 made as above and anti-GFP obtained from Q-BIO-gene, now MP Biomedicals, Solon, OH, USA) and Cy3- or Cy5- labeled (Jackson ImmunoResearch Laboratories, Inc., West Grove, PA, USA) secondary antibodies.

Rescue of the flr-2 phenotype by tissue-specific expression

For neuron-specific expression, pH20::FLR-2 was made by inserting the H20 promoter (2.4-kb PstI fragment) into the promoterless GFP vector pPD95.69 and by replacing its GFP part (KpnI-ApaI) with the PCR product amplified from flr-2 cDNA with the primers f2-kpn (CTCGGTACCTATACCTTCTCAAATGATG) and f2-apa (TCTGGGCCCGGATGAATGTGGCACTGG). pges-1::FLR-2 for intestine-specific expression and pmyo-3:: FLR-2 for muscle-specific expression were made in a similar way by replacing the GFP part of myo-3::GFP (pPD115.02) and ges-1::GFP (pPD95.79-ges-1) with the flr-2 cDNA sequence. These DNA constructs were injected into flr-2(ut5); flr-1(ut11) worms together with the marker myo-3::GFP. The transformants were tested for their growth rates: Worms that grew to the L4 larva or adult on the third day of egg lay were regarded as normal growers, while those that were at the stage of the L3 or earlier larva were regarded as slow growers. pges-1::FLR-2 and pmyo-3::FLR-2 rescued the flr-2 phenotype [suppression of the slow growth of the flr-1(ut11) mutant] when they were injected at the concentration of 10 ng/μL, while pH20::FLR-2 rescued it at the concentrations of 0.5 and 2 ng/μL, but not 10 ng/μL.

Expression cloning of FLR-2-binding proteins

Expression cloning was conducted essentially according to Flanagan & Cheng (2000). The DNA construct for producing FLR-2::AP was made as follows: the flr-2 cDNA with NheI and BamHI sites at the ends were amplified using the primers AP5-Nhe (TCTGCTAGCCACCATGATGGGCTCCAAAGCACG) and AP5-Bam (TCTGGATCCTCCTCGAACAAGACAATCAAAAC), and this fragment was inserted into the NheI and BglII sites of the APtag-5 vector (GenHunter, Nashville, TN, USA). The DNA construct was transfected into HEK293 cells to yield a cell line that stably produced FLR-2::AP. The supernatant of the cell culture, which contained 2–7 nm FLR-2::AP, was kept and used as the source of FLR-2::AP in the expression cloning experiments.

For expression cloning, a cDNA library was made from C. elegans primary cultured cells (Christensen et al. 2002). Each 1000 clones of the cDNA library were transfected into 192 cultures of COS cells (50–80% confluent). The cells were cultured further for 2 days to reach confluence, fixed with 3% paraformaldehyde in Hank’s buffered saline (HBS) (150 mm NaCl, 10 mm HEPES, pH 7.0), treated with 0.1% Triton X-100 in HB (HBS with 0.5% BSA), incubated with FLR-2::AP, washed thoroughly, and tested for the binding of FLR-2::AP using alkaline phosphatase detection kit (Nacalai, Kyoto, Japan). The fixation procedure is considered to allow detection of secretory proteins. Alkaline phosphatase activity was detected in one of the 192 cell cultures only after incubation with FLR-2::AP. Of the 1000 cDNA clones used for the transfection of this cell culture, only one clone conferred the FLR-2::AP-binding activity to COS cells by transfection, which we named ghi-1. Sequencing of the cDNA clone revealed that it corresponded to ZK20.1.

ghi-1p::GFP fusion gene

The 5-kb upstream sequence together with the 23 base 5′ coding sequence of ghi-1 gene was amplified with the primers ZK20.1GFP-1 (TCTGGATCCGTTGATAGTCAGTCACTGCC) and ZK20.1GFP-2 (TCTGGATCCAGCATGTGGAGGAGTGAGAAC), which were designed so that the PCR product had BamHI sites at both ends. The PCR product was cleaved with BamHI and inserted into the BamHI site of the promoterless GFP vector pPD95.67. The ghi-1p::GFP construct (50 ng/μL) was injected together with the marker rol-6(su1006dom) (50 ng/μL) into wild-type animals to obtain transformants.

Isolation of a deletion mutant in ghi-1 gene

The deletion mutant ghi-1(ut293) was isolated according to Gengyo-Ando & Mitani (2000). About 200 000 genomes of TMP/UV-treated worms were screened for mutants in ghi-1 by using the primers zk20EL (CTGCAAATTGCCGATCGTTTGCCGC) and zk20ER-2 (TTTCCGCTCGCCAATTCCTGGTTGC) for the first PCR, and zk20IL (TCAGGTGCCTTTAACCTGCCATGCC) and zk20IR-2 (GCTTATCACTAACTAACCTCGTACG) for the second PCR. The ut293 mutant was outcrossed five times to wild-type animals before use.

Measurements of defecation cycle periods

Defecation cycle periods were measured at room temperature (23 °C) except for the temperature-sensitive mutants flr-4(sa201), flr-4(n2259), and double- or triple-mutants containing these mutations, which were examined at 25 °C. Ten periods were measured for each of ten animals for each strain. The average was calculated for each animal, and the average and standard error of the averages were calculated for each strain. The defecation motor steps of class 1 flr mutants and class 2 class 1 flr double mutants sometimes became too weak to observe even the pBoc step, and the animals looked as if they ceased the defecation behavior for a long time. Hence, we did not include defecation interval lengths longer than 100 s in the statistics (Take-uchi et al. 1998).

RNAi of ghi-1

The DNA construct for the feeding RNAi was made with the vector pPD129.36 (L4440) by inserting the PCR fragment amplified from ghi-1 cDNA (yk1207.a08) with the primers yk1207-1 (TCTGGTACCTCGCCGTTCCCTCAAATG) and yk1207-2 (TCTGGTACCCAGATAAGTGGTTGTGGG), which added KpnI sites at both ends. The construct was transfected into E. coli HT115(DE3). The transformant was cultured, subjected to induction, and seeded onto NGM plates containing Amp and Tet essentially according to Protocol 2/4 of Kamath et al. (2001) except that induction was performed for 4 h. Adult hermaphrodites having various genotypes were allowed to lay eggs on these plates, and the defecation cycle periods of the progeny were measured on the same plates when they grew to young adults. It was confirmed that the results were the same when the defecation cycle periods were measured on NGM plates with OP50. E. coli HT115(DE3) carrying the vector pPD129.36 was used for control experiments.

Acknowledgements

We thank Masaya Take-uchi for the mapping of flr-2 with arDf1, Hitoshi Makino, Tomoko Motohashi, Tomonori Iwama, Shinobu Tayama for technical assistance, members of our laboratory for discussions, A. Garsin for E. faecalis OG1RF, Andy Fire for pPD vectors, Jyunichi Miyazaki for pCAGGS2and Yuji Kohara for yk1207.a08. Some nematode strains used in this work were provided by the Caenorhabditis Genetics Center, which is funded by the NIH National Center for Research Resources (NCRR). This work was supported by Grants-in-Aid for Scientific Research from the Ministry of Education, Culture, Sports, Science and Technology of Japan to I.K.

Ancillary