Embryonic Morphogenesis in Caenorhabditis elegans Integrates the Activity of LET-502 Rho-Binding Kinase, MEL-11 Myosin Phosphatase, DAF-2 Insulin Receptor and FEM-2 PP2c Phosphatase

Corresponding author: Paul E. Mains, Department of Biochemistry and Molecular Biology, University of Calgary, 3330 Hospital Dr. NW, Calgary, AB T2N 4N1, Canada., mains{at}ucalgary.ca (E-mail)

Communicating editor: R. K. HERMAN

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

let-502 rho-binding kinase and mel-11 myosin phosphatase regulate Caenorhabditis elegans embryonic morphogenesis. Genetic analysis presented here establishes the following modes of let-502 action: (i) loss of only maternal let-502 results in abnormal early cleavages, (ii) loss of both zygotic and maternal let-502 causes elongation defects, and (iii) loss of only zygotic let-502 results in sterility. The morphogenetic function of let-502 and mel-11 is apparently redundant with another pathway since elimination of these two genes resulted in progeny that underwent near-normal elongation. Triple mutant analysis indicated that unc-73 (Rho/Rac guanine exchange factor) and mlc-4 (myosin light chain) act in parallel to or downstream of let-502/mel-11. In contrast mig-2 (Rho/Rac), daf-2 (insulin receptor), and age-1 (PI3 kinase) act within the let-502/mel-11 pathway. Mutations in the sex-determination gene fem-2, which encodes a PP2c phosphatase (unrelated to the MEL-11 phosphatase), enhanced mutations of let-502 and suppressed those of mel-11. fem-2's elongation function appears to be independent of its role in sexual identity since the sex-determination genes fem-1, fem-3, tra-1, and tra-3 had no effect on mel-11 or let-502. By itself, fem-2 affects morphogenesis with low penetrance. fem-2 blocked the near-normal elongation of let-502; mel-11 indicating that fem-2 acts in a parallel elongation pathway. The action of two redundant pathways likely ensures accurate elongation of the C. elegans embryo.

THE Rho family of Ras-like GTPases has been implicated in the regulation of the actin cytoskeleton, resulting in altered cell shapes, movements, and cytokinetic events (for reviews see VAN AELST and D'SOUZA-SCHOREY 1997 ; HALL 1998 ; MACKAY and HALL 1998 ; ASPENSTROM 1999 ; KAIBUCHI et al. 1999 ). Downstream effectors of Rho-GTPases include the Rho-binding kinases, which have been characterized by binding assays in vitro, by in vivo cell culture transfection experiments, and by mutational analyses in Caenorhabditis elegans. In mammalian fibroblast cells two Rho-binding kinases, ROK (Rho-associated kinase type ) and p160^ROCK (Rho-associated kinase), promote the formation of stress fibers and focal contacts, whereas dominant-negative forms cause disassembly of these structures (LEUNG et al. 1995 , LEUNG et al. 1996 ; FUJISAWA et al. 1996 ; AMANO et al. 1997 ; ISHIZAKI et al. 1997 ).

In addition to regulating actin organization within cells, Rho-binding kinases are involved in smooth muscle contraction. Actin-myosin contractions result when regulatory myosin light chains (MLC) are phosphorylated by myosin light chain kinase (MLCK). Myosin phosphatase PP1c holoenzyme blocks these contractions by dephosphorylating MLC, which thus antagonizes MLCK leading to muscle relaxation (ALLEN and WALSH 1994 ; HARTSHORNE 1998 ; HARTSHORNE et al. 1998 ). Phosphorylation of the regulatory targeting subunit of myosin phosphatase (M110, M130, and M133) by ROK and p160^ROCK decreases the activity of myosin phosphatase toward MLC, resulting in the accumulation of MLCK-phosphorylated MLC and a contractile response (SHIMIZU et al. 1994 ; ICHIKAWA et al. 1996 ; KIMURA et al. 1996 ). The Rho-binding kinases also phosphorylate MLC directly to induce contraction in vitro (AMANO et al. 1996 ; KUREISHI et al. 1997 ; FENG et al. 1999 ).

The ability of Rho-binding kinases to phosphorylate and thereby negatively regulate myosin phosphatase and hence to positively regulate MLC-mediated contraction, suggests a model where Rho-binding kinases can mediate cell shape changes by altering the contractile state of actin/myosin filaments. However, this model is based on evidence from cell culture, smooth muscle preparations, and purified proteins. We previously described a role for the C. elegans homologs of Rho-binding kinase, let-502, and the regulatory subunit of myosin phosphatase, mel-11, in regulating the epidermal cell shape changes that drive elongation of the embryo (WISSMANN et al. 1997 , WISSMANN et al. 1999 ). This provided in vivo evidence that Rho-binding kinases and myosin phosphatases interact to regulate actin-mediated cell shape changes. C. elegans is thus a useful system to study this pathway, particularly for using genetic methods to identify new components. Genetic methods can identify redundant pathways that would be difficult to find by biochemical approaches.

The C. elegans embryo undergoes a fourfold increase in length without additional cell proliferation or increase in cell volume, and this occurs through actin-mediated contractions (PRIESS and HIRSH 1986 ). let-502 mutations result in elongation defects or adult sterility, depending on the allele used (HOWELL and ROSE 1990 ; WISSMANN et al. 1997 , WISSMANN et al. 1999 ). Loss of mel-11 function results in embryonic arrest due to hypercontraction during elongation (KEMPHUES et al. 1988 ; WISSMANN et al. 1997 , WISSMANN et al. 1999 ). Mutations of let-502 and mel-11 suppress one another's elongation defects, implying that let-502 and mel-11 function antagonistically during elongation. mlc-4, a C. elegans MLC gene, is also expressed in the epidermis at this time, and mutants display elongation defects similar to those of let-502 mutants (SHELTON et al. 1999 ). Our genetic results and the analogies to smooth muscle contraction suggest that MEL-11 may prevent contraction of the epidermal cells prior to elongation by negatively regulating MLC-4. At the appropriate time, LET-502 negatively regulates MEL-11 to relieve the inhibition of MLC-4. This allows contraction of the circumferentially oriented microfilaments in the epidermal cells, causing the cells to change shape, driving elongation of the embryo (PRIESS and HIRSH 1986 ; WISSMANN et al. 1997 , WISSMANN et al. 1999 ).

Several issues were raised by our previous work. In smooth muscle, contraction can be induced by either MLCK or Rho-binding kinase. Therefore, let-502 Rho-binding kinase (or MLCK) might be redundant for elongation. Indeed, all previously identified let-502 alleles had gain-of-function (gf) properties (WISSMANN et al. 1997 ). While these likely represent dominant-negative (antimorphic) mutations, the phenotypes might instead reflect neomorphic activities. In the latter case, the let-502 loss-of-function (lf) phenotype would be wild type (i.e., the gene would be redundant or otherwise nonessential), explaining why only gf mutations were found in previous screens for visible phenotypes (lethality or sterility). Alternatively, the high frequency of gf mutations could reflect an underlying protein structure that was prone to mutating to gf products. In this article, we describe a series of novel let-502 mutations that were identified using a screen that could isolate lf alleles. Analysis of our new mutations, used in conjunction with RNA interference (RNAi), indicates that let-502 is an essential gene. The zygotic null phenotype is an adult sterile, but hypomorphic and dominant-negative alleles indicate that let-502 also has essential functions during the early embryonic cleavages and morphogenesis.

Another question left unanswered by previous work was the nature of the cosuppression between let-502 and mel-11. The let-502; mel-11 double mutants undergo near-normal elongation even though the individual mutations are elongation defective (WISSMANN et al. 1997 ). This could imply that the pathway as a whole is redundant; phenotypes arise when only one or the other gene is not functioning. In analogy with vertebrate contractile systems, elongation could be triggered, and properly regulated, entirely through an MLCK-dependent system when the Rho-binding kinase (let-502) and myosin phosphatase (mel-11) pathway is inoperative. Alternatively, the observed mutual suppression between let-502 and mel-11 could result from equally low (but nonzero) levels of the two products. The analysis reported here of let-502 and mel-11 double mutants coupled with RNAi indicates that the elongation pathway is redundant. Furthermore, the behavior of other mutations that genetically interact with mel-11 indicate that two genes (mig-2 Rho/Rac and daf-2 insulin receptor) function in the same elongation pathway as let-502 and mel-11 while others [unc-73 Rho/Rac guanine exchange factor (GEF) and mlc-4 MLC] act either downstream or in parallel to the let-502/mel-11 system.

A surprising result reported here is that both mel-11 and let-502 interact genetically with fem-2 (PP2c phosphatase). A role for fem-2 in embryonic elongation has not been previously described, even though its essential function in sex determination is well established. fem-2(+) leads to sperm production in both XX hermaphrodites and X0 males and male somatic development in X0 animals (KIMBLE et al. 1984 ; HODGKIN 1986 ; PILGRIM et al. 1995 ; CHIN-SANG and SPENCE 1996 ; for reviews, see MEYER 1997 ; SCHEDL 1997 ). We found that fem-2 mutations enhanced alleles of let-502 and suppressed those of mel-11. Our genetic results indicate that fem-2's activity during morphogenesis is independent of its role in sex determination and that fem-2's action during embryonic elongation is redundant with the let-502/mel-11 elongation pathway.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Strains and alleles:
C. elegans (N2 var. Bristol) were maintained under standard conditions (BRENNER 1974 ). Strains were constructed using standard procedures, and cis-linked morphological markers were often used to follow mutations of mel-11 and let-502 through crosses. Homozygous lethal or sterile mutations were maintained as heterozygous stocks balanced either with appropriate crossover suppressors or normal chromosomes with flanking morphological markers. In those cases where the phenotype being scored might be confused with the homozygous balancer (e.g., unc-11 unc-40 used to balance let-502 resembles let-502; mel-11), the gene of interest was maintained over a fully wild-type chromosome. To determine hatching rates of different genetic combinations, four or more L4 hermaphrodites were brooded at the appropriate temperatures until they ceased to lay fertilized embryos; a minimum of 400 progeny were scored unless otherwise noted (MAINS et al. 1990 ).

To avoid effects on the sexual identity of temperature-sensitive (ts) sex-determination mutations, the animals of interest were reared at the permissive temperature of 15° and upshifted to 20° or 25° as young adults (i.e., after sexual identity was established). These animals were then purged of embryos fertilized prior to upshift by incubation for 2 hr (25°) or 3 hr (20°) before brood collection commenced.

fem-3(e2006ts) hermaphrodites quickly stop laying fertilized embryos upon upshift to 25°. To quantify the effects of this mutation on embryonic viability, gravid animals that had been raised at 15° were transferred to a drop of room-temperature water, pregastrulation embryos (<2 hr postfertilization, fewer than 28 cells) were removed from the hermaphrodite by dissection with a scalpel, and embryos were placed on a petri dish preequilibrated to 25°. Since the temperature-sensitive period for mel-11 begins 6 hr postfertilization (WISSMANN et al. 1999 ), embryos shifted in this manner are likely comparable to those fertilized within hermaphrodites grown at 25°.

Nomenclature follows that of HORVITZ et al. 1979 . Genes, alleles, and balancer chromosomes listed below were used; descriptions can be found in HODGKIN 1997 and EDGLEY et al. 1995 :

• Linkage group I: let-502(ca201, ca201sb54, sb93, sb95, sb103, sb106, sb107, sb108, sb109), unc-73(rh40), dpy-5(e61), bli-4(e937), daf-16(mgDf50).

• Linkage group II: dpy-10(e128), mel-11(it26ts, sb55ts, sb56), unc-4(e120), sqt-1(sc13), age-1(mg44).

• Linkage group III: fem-2(b245ts, e2105), mlc-4(or253), daf-2(e1370ts, m212ts).

• Linkage group IV: fem-1(hc17ts, e1965), fem-3(e1996, e2006ts, q20gf,ts), dpy-20(e1282), tra-3(e1107).

• Linkage group V: tra-1(e1575gf).

• Linkage group X: lon-2(e678), mig-2(mu28).

• Balancer chromosomes: The crossover suppressor hT2 (I;III) was sometimes used to balance let-502 and mnC1 II was used to balance mel-11.

hDf6 is a deficiency on linkage group I that deletes the let-502 locus (MCKIM et al. 1988 ).

Isolation of mel-11 suppressors:
To obtain novel let-502 alleles, we exploited our observation that a deficiency that deletes the let-502 locus (hDf6) dominantly suppresses the ts maternal-effect lethality of mel-11(it26), resulting in 8% hatching at 20° vs. 0.4% for controls (WISSMANN et al. 1999 ). Therefore, a mel-11 suppressor screen could identify let-502(lf) alleles even if such mutations had no phenotype on their own. mel-11(it26) unc-4 sqt-1/mnC1 hermaphrodites were mutagenized with 25 mM ethyl methanesulfonate (BRENNER 1974 ), F₁ progeny were allowed to self at the permissive temperature of 15° for 3–4 days and animals were then upshifted to the restrictive temperature of 20° (where 1% normally hatch). A total of 8000 haploid genomes were screened for plates with substantial hatching. Twenty-seven independent suppressors were obtained in total. Mutations were outcrossed at least twice and mapped using standard genetic methods. Suppressors of mel-11(it26) that mapped to LGI were tested for complementation with let-502(ca201). The seven new let-502 alleles were cycle sequenced from single worms as outlined previously (WISSMANN et al. 1997 ).

RNAi:
RNAi was performed as previously described (FIRE et al. 1998 ). A full-length 4.3-kb cDNA clone of let-502 (pAW12.7) in pBluescript SK⁺ was used to make let-502 RNA. Primers with T3 and T7 promoter sequences were designed and used to amplify mel-11 cDNA, with mRNA collected from gravid hermaphrodites as a template (FastTrack 2.0, Invitrogen, San Diego; Titan, one-tube RT-PCR, Roche Diagnostics). Single-stranded RNA was made from the T3 and T7 promoters, respectively (Ambion system, Ambion, Austin, TX). RNA was precipitated and resuspended in RNase-free TE buffer (10 mM Tris, 1 mM EDTA pH 8.0) according to manufacturer's instructions. Double-stranded RNA was made by mixing equal amounts of the complementary strands together, heating at 70° for 15 min, and then incubating at 37° for 30 min. Injections with 10 ng/µl of let-502 dsRNA were performed in wild-type and let-502(sb108) animals as described by FIRE et al. 1998 , and similar concentrations of mel-11 dsRNA were injected into wild-type and mel-11(it26) animals. let-502 and mel-11 dsRNA were mixed with both at concentrations of 10 ng/µl and injected into wild-type and let-502(sb108); mel-11(it26) animals. Progeny of injected animals were observed by Nomarski optics on a Zeiss Axioplan microscope and flash-photographed using Kodak Techpan film (Rochester, NY) developed at ASA 100. The C. elegans myotonic dystrophy protein kinase (CeDMPK) homolog is the C. elegans gene most similar to let-502 at the protein level; however, at the DNA level the longest stretch of identity is only 80% over 50 nucleotides. As a control for possible cross-interference between the genes, dsRNA derived from CeDMPK (located on cosmid K08B12) was injected into wild-type animals. The cDNA was made by RT-PCR, according to manufacturer's instructions (FastTrack 2.0, Invitrogen; Titan RT-PCR kit, Roche Diagnostics), using primers specific to the CeDMPK gene (C. ELEGANS SEQUENCING CONSORTIUM 1998). The RNAi phenotype was viable with what appears to be weak body wall muscle and/or epidermal defects that differed from those seen with let-502, indicating that the results obtained from the let-502(RNAi) were specific.

Testing of feminizing mutations for interactions with let-502 and mel-11:
Since loss of tra-1 function causes transformation into males, we employed a gf allele that acts as a dominant feminizer (HODGKIN 1987 ) to investigate possible maternal effects on elongation. Because these animals are not self-fertile, we crossed mel-11(it26) unc-4; tra-1(gf)/+ females to mel-11(it26) unc-4/+ males. Since mel-11(it26) shows near-complete zygotic rescue (KEMPHUES et al. 1988 ), we can use the mel-11/+ (phenotypically wild type) outcross progeny as a control to determine the relative viability of their homozygous mel-11 unc-4 (phenotypically Unc) outcross sibs. Using the cross-assay, we found that tra-1(gf) resulted in 5.3% viability of mel-11 homozygous progeny compared to 2.8% mel-11 viability in the isogenic control without tra-1. In contrast, replacing tra-1(gf) with fem-2(b245) in the female parent increased the viability of mel-11(it26) progeny to 16%. Therefore, tra-1(gf) showed little or no interaction with mel-11. With similar crosses, we found that maternal homozygosity for null alleles of fem-1(e1965) or fem-3(e1996) that are also phenotypically female (DONIACH and HODGKIN 1984 ; HODGKIN 1986 ) also had little effect on mel-11 viability (data not shown).

Microscopy:
Animals were mounted on agarose pads and observed with a Zeiss Axioplan microscope using Nomarski optics. The embryos were flash-photographed with Kodak TechPan film, which was developed at 100 ASA.

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Establishing the null phenotype of let-502 and interactions with mel-11:
Our previous work on let-502 and mel-11 left several questions unanswered (WISSMANN et al. 1997 , WISSMANN et al. 1999 ). These included determining the let-502 null phenotype and whether let-502 and mel-11 function redundantly with another pathway to regulate embryonic elongation. In the following sections, we will first demonstrate that the let-502(null) is an adult sterile, but this gene also acts during the early embryonic cleavages and subsequently during embryonic elongation. We will also show that the let-502/mel-11 pathway is redundant during embryonic elongation. In later sections, we will examine mutations in genes that interact with let-502 and mel-11 and show that several are part of the redundant elongation pathway.

Isolation and phenotypes of new let-502 alleles: All previously identified let-502 alleles were isolated on the basis of their lethal or sterile phenotypes and they behave as gf mutations. When homozygous, genetically strong alleles cause arrest during embryogenesis (ca201, h783, h835) while weak alleles result in adult sterility (ca201sb54, h392, h509, h732). Although we use the terms "strong" and "weak," even the strongest gf mutations are only weakly dominant, producing only low-penetrance morphological defects in heterozygotes. To ascertain the let-502 null phenotype, new let-502 alleles were obtained from a suppressor screen of mel-11(it26) that could potentially isolate lf alleles, including those with no phenotypes on their own (see MATERIALS AND METHODS). The properties and our interpretation of the nature of each mutation (as described below) of the new let-502 alleles, along with the previously identified mutations, are presented in Table 1 (sb95 and sb108 have identical nucleotide changes, and so only sb108 underwent further genetic testing). While two of the new mutations (sb93 and sb109) are similar to previously identified let-502 alleles in showing elongation defects and adult sterility (Ste), the other alleles are homozygous viable in the absence of the mel-11 mutation. The latter alleles show variable penetrance and have associated morphological, Ste, and/or maternal-effect lethal (Mel) phenotypes.

In addition to the previously described elongation and sterile phenotypes, let-502(sb103) and let-502(sb106) displayed similar incompletely penetrant early cleavage defects. Fig 1 shows a wild-type C. elegans embryo at pronuclear fusion (A) and after the first two cell divisions (B and C). At 25°, 58% of the progeny from let-502(sb106) homozygotes either failed to complete any cleavages (D) or formed cleavage furrows that would then regress (E and F). All of the let-502(sb106) embryos that failed to hatch had cleavage defects and died prior to morphogenesis (n = 20). The let-502(sb106) larvae that hatched all showed defects ranging from arrest as unelongated larvae to adult morphogenetic phenotypes such as dumpy (Dpy) and rolling (Rol). let-502(sb103) behaved similarly, with incompletely penetrant cleavage defects, but all of the hatched embryos failed to elongate.

View larger version (95K): [in this window] [in a new window] [Download PPT slide]   Figure 1. Nomarski photomicrographs of wild-type and mutant embryos at 25°. Shown is a wild-type embryo at (A) pronuclear fusion, (B) the two-cell stage, and (C) the four-cell stage. Nomarski images of let-502(sb106) homozygous embryos show (D) an embryo that completely lacks cytokinesis and (E–F) an embryo that formed cleavage furrows, which later regressed. Bar, 10 µm.

let-502(sb106) and let-502(sb103) homozygous mutants had low brood sizes. At 20°, let-502(sb106) mutants produced 30 progeny/hermaphrodite, but brood size increased to 150 progeny/hermaphrodite by outcrossing to either wild-type or let-502(sb106) males. This suggests that let-502(sb106) sterility is due to the homozygous hermaphrodite having limited amounts of available sperm, likely related to our previously observed defects in the spermatheca, the sperm storage organ (WISSMANN et al. 1999 ).

A strict maternal requirement for let-502 was found in alleles that displayed embryonic lethality. When let-502(sb103) or let-502(sb106) hermaphrodites were outcrossed to wild-type males, the percentage of unhatched embryos was the same as when hermaphrodites were selfed (Table 2). However, most of the embryos that did hatch grew to morphologically normal adults. Thus, zygotic let-502(+) rescued the elongation defects of the hatched animals but not the lethality resulting in unhatched embryos. When let-502(sb103)/+ and let-502(sb106)/+ hermaphrodites were selfed, 99% hatching was observed, demonstrating that the failure of embryos to hatch can be also maternally rescued. Taken together, these data indicate that the embryonic viability has a strict maternal requirement for let-502(+) while the larval arrest and morphological defects can be rescued by either maternal or zygotic let-502(+).

Nature of the new let-502 mutations: We sequenced the new let-502 alleles and identified their molecular lesions. These mutations are shown in reference to the predicted protein structure in Fig 3, along with the mutations for the previously identified let-502 alleles. The new let-502 alleles, including the viable mutations, had missense mutations in conserved amino acids in the kinase domain (Fig 4).

View larger version (111K): [in this window] [in a new window] [Download PPT slide]   Figure 2. Nomarski photomicrographs of a wild-type embryo in varying stages of embryonic elongation in comparison with let-502 mutants. A wild-type embryo is shown at (A) early morphogenesis, (B) midmorphogenesis (comma stage), and (C) fully elongated (pretzel stage). (D) Also shown is a Nomarski image of the homozygous let-502(ca201) terminal phenotype characterized by the hatched larva arrested with little elongation. let-502(RNAi) injected into wild-type hermaphrodites resulted in hatched larva (E) with little elongation. let-502(RNAi) injected into let502(sb108) hermaphrodites resulted in hatched larva (F) with an elongation defect more severe than that in E. (G) Hatched let-502(ca201); fem-2(b245) larva has undergone even less elongation than the let-502(ca201) animal shown in D. (I) Hatched fem-2(e2105) ceased elongation at a stage similar to that of the let-502(ca201) homozygote shown in F. (J) Midstage fem-2(e2105) larva shows abnormal elongation in the pharyngeal region. Because of the maternal effect of fem-2(e2105), animals in I and J were the F2 progeny of heterozygous hermaphrodites. All fem-2 bearing strains were grown at 25°. Bar, 10 µm.

View larger version (26K): [in this window] [in a new window] [Download PPT slide]   Figure 3. A schematic representation of the let-502 mutations. The domains are indicated with the exception of the PH domain and Cys-rich region in the C terminus of the protein. The nonsense mutants (which lead to predicted truncations) are shown as truncated proteins. Also shown are the genetic properties of the alleles: null, hypomorph (hypo), dominant negative (dn) or weak dominant negative (w-dn), and their corresponding phenotypes. Alleles isolated in this study are designated in boldface type. The other alleles were identified by HOWELL and ROSE 1990 and WISSMANN et al. 1997 .

View larger version (90K): [in this window] [in a new window] [Download PPT slide]   Figure 4. Alignment of the protein kinase domain from LET-502 in comparison with the kinase domain from p160ROCK (human Rho-associated kinase; FUJISAWA et al. 1996 ), ROK (rat Rho-associated kinase type ; LEUNG et al. 1995 ), GEK (D. melanogaster Genghis Khan; LUO et al. 1997 ), MRCK (rat myotonic dystrophy kinase-related Cdc42-binding kinase; LEUNG et al. 1998 ), DMPK (human myotonic dystrophy kinase; BROOK et al. 1992 ; FU et al. 1992 ), Citron-K (mouse Citron kinase; DI CUNTO et al. 1998 ; MADAULE et al. 1998 ), and PKA (human cAMP-dependent kinase ß; BEEBE et al. 1990 ). Genetics Computer Group, Inc. (GCG) software, PileUp, and freeware Boxshade were used to perform the sequence analysis. Identical residues are shaded in black and conserved residues in gray. The location of mutations in LET-502 and the corresponding amino acid changes are shown. Also indicated is whether the allele behaves genetically as a hypomorph (hypo), dominant negative (dn), or weak dominant negative (w-dn). Previously identified alleles are designated with asterisks (HOWELL and ROSE 1990 ; WISSMANN et al. 1997 ). ca201sb54, h392, h509, and h732 are truncations due to nonsense or splice donor/acceptor mutations and are not shown.

Using genetic tests, we determined whether the new let-502 alleles behaved as antimorphs (dominant-negatives), nulls, or hypomorphs (MULLER 1932 ). The classifications were made by examining the progeny that resulted from crossing hermaphrodites heterozygous for a new let-502 allele to males carrying either a strong gf allele (ca201), a weak gf allele (ca201sb54), or a deficiency (hDf6) that deletes the let-502 locus. The most extreme phenotype is early larval arrest, while mid- to late arrest Ste and Mel represent progressively less severe phenotypes. If b/a arrests earlier than b/Df, then a must have gf properties (where a and b are different alleles and Df is the deficiency hDf6). We will assume that these gf properties are dominant negative; this will be confirmed using RNAi in a later section. If b/a resembles b/Df, then a is acting like a null, while a later lethal phase of b/a relative to b/Df reveals that a is hypomorphic since it retains some wild-type activity. If a/a is more severe than a/Df, then a has dominant-negative properties. A less severe a/a phenotype compared to a/Df indicates that a is hypomorphic. Since let-502 alleles show dominant maternal effects (WISSMANN et al. 1997 ), we performed reciprocal crosses. The results are listed in Table 3 and an overall summary is included in Table 1.

We showed previously that both ca201 and ca201sb54 have gf (likely dominant-negative) properties, with ca201 being relatively stronger (WISSMANN et al. 1997 ). This is seen in Table 3 where allelic combinations including ca201 always showed phenotypes that were equal to, and usually more severe than, those that included the Df. This was true whether ca201 was contributed maternally (column 3 vs. 5) or paternally (column 6 vs. 8). When maternally inherited (column 4 vs. 5), ca201sb54 was more severe than the Df when in trans to ca201, sb109, or sb106 (the latter at 20°), but ca201sb54 was equivalent to the Df for all other alleles. When ca201sb54 was inherited paternally, it always acted as a null (column 7 vs. 8). Thus, ca201sb54 is similar to a null in most respects, but it behaves as a dominant negative in a few heteroallelic combinations.

sb103, sb106, and sb107 each showed mixtures of dominant-negative, null, and hypomorphic properties depending upon the heteroallelic combination examined. For example, when sb106/+ males were crossed to ca201/+ hermaphrodites, the ca201/sb106 progeny arrested as early to midstage larvae, as compared to ca201/Df animals, which consistently arrested as early larvae (line 5 vs. 6). Thus, sb106 retains some wild-type activity and is a hypomorph. In contrast, animals with maternally inherited ca201sb54 demonstrated that sb106 has weak dominant-negative characteristics: while ca201sb54/Df were semisterile adults, ca201sb54/sb106 had a slightly more severe phenotype, arresting as mid- to late-stage larvae. Finally, sb106 behaved as a null when sb106/sb106 and sb106/Df were compared since both led to similar semisterile adult phenotypes. When tests were repeated at 25°, each sb106 heteroallelic combination showed a slightly earlier arrest (compare lines 6 and 7), indicating that this mutation loses wild-type activity and becomes more dominant negative with increasing temperature. Of the other new mutations, sb109 consistently acted as a dominant negative in heteroallelic combinations (Table 3, compare lines 2 and 5), while sb108 was the only allele that clearly behaved as a hypomorph in all tests (Table 3, line 5 vs. 10).

RNAi indicates that let-502 is essential: All of the let-502 alleles have been classified as dominant negatives, nulls, and hypomorphs as summarized in Table 1. The exact null phenotype remains difficult to assign since none of the alleles clearly mimic hDf6 in all assays. It is possible that removal of zygotic activity results in an adult Ste phenotype like ca201sb54, the mutation that most closely approximates the null. Larval arrest would then result from the removal of most maternal and all zygotic let-502(+) activity as seen with ca201, i.e., the dominant negative ca201 allele decreases the amount of let-502(+) present in the heterozygous hermaphrodite and totally eliminates it from the progeny. Alternatively, the null phenotype could be wild type, similar to sb108. If the latter were true, alleles classified as dominant negatives would in fact be neomorphs.

To unambiguously determine the consequences of loss of let-502 activity we used RNAi, which has been shown to mimic lf (often null) phenotypes in C. elegans by eliminating maternal and zygotic product (FIRE et al. 1998 ). Experiments were performed with double-stranded RNA (dsRNA) obtained from the full-length (4.3-kb) let-502 cDNA. Embryonic elongation in wild-type animals is shown in Fig 2A&NDASH;C, and the ca201 phenotype is shown in Fig 2D. let-502 dsRNA injected into wild-type hermaphrodites resulted in embryos that failed to elongate, although a range of phenotypes was observed from early (as shown in Fig 2E) to midlarval arrest, with 10% growing into Ste, Rol, Dpy adults. Therefore, the let-502(lf) phenotype is clearly not wild type. The observed phenotypic range could reflect the inefficiency of RNAi or partial let-502 redundancy. If the RNAi was inefficient, injection of let-502 dsRNA into hermaphrodites homozygous for the putative hypomorphic sb108 allele should shift the range of phenotypes to a higher percentage of early larval arrest. This was indeed the case, as dsRNA injected into let-502(sb108) animals resulted in all progeny displaying early larval arrest (Fig 2F). Therefore, decreasing maternal and zygotic let-502(+) activity results in the larval arrest similar to that seen with the strong dominant-negative alleles. Escapers were Ste and this could represent the zygotic null phenotype.

We occasionally observed injected hermaphrodites that laid unhatched embryos, but these were not examined for the cleavage defects observed in let-502(sb106) and let-502(sb103). The lack of consistent embryonic lethality likely indicates that RNAi does not eliminate all maternal let-502 activity. Embryos with sufficient maternal let-502(+) to prevent the early cleavage defects would later fail to elongate properly.

let-502 suppression of mel-11: The genetic properties of the let-502 alleles were tested further by their suppression of homozygous mel-11(it26) worms. At 20° hDf6/+; mel-11(it26) worms had a small percentage (8%) of progeny that survived compared to 0.4% for the control, indicating partial suppression of the elongation defects. However, as shown in Table 4, most let-502 alleles were better at suppressing mel-11(it26) than the deficiency, as expected for dominant-negative alleles. sb109, ca201, and sb103 had the highest levels of suppression as heterozygotes (63–68%). sb107, sb108, sb106, and ca201sb54 had lower levels of suppression as heterozygotes, but still had higher levels of suppression than hDf6 (26–37% vs. 8%, respectively). However, the comparison of let-502/+; mel-11 to hDf6/+; mel-11 is deceptive because while the let-502 homozygous segregants hatch, hDf6 homozygotes do not. Therefore, to directly compare the suppression of mel-11 by let-502/+ to hDf6/+, only the percentage of heterozygous progeny should be considered. Using this metric, sb108 (11%) and ca201sb54 (6.8%) are similar to hDf6 (8.0%) in their ability to dominantly suppress mel-11(it26). The interpretations of these results are summarized in Table 1.

In contrast to the elongation phenotypes, the let-502 adult sterile phenotype was not suppressed by mel-11(it26). let-502(ca201); mel-11(it26) and let-502(sb109); mel-11(it26) grew to adulthood due to mel-11's suppression of elongation defects, but the animals were sterile. Similarly, let-502(sb106); mel-11(it26) had low brood sizes similar to those seen with let-502(sb106) alone at both 20° and 25° (data not shown).

mel-11 can suppress the early cleavage defects of let-502(sb106). This is most clearly seen using the hypomorphic allele mel-11(sb55) (WISSMANN et al. 1999 ). By itself, let-502(sb106) showed 65% hatching at 20°, but inclusion of mel-11(sb55) increased the hatching rate to 91% (Table 4).

RNAi indicates that let-502/mel-11 are nonessential for elongation: The mutual suppression of mel-11 and let-502 elongation defects could be due to a balance of equally low residual amounts of let-502(+) and mel-11(+) activity. Alternatively, the let-502/mel-11 pathway could be redundant, and a parallel pathway might regulate embryonic elongation in the absence of both let-502 and mel-11. To distinguish between these possibilities, mel-11 and let-502 dsRNAs were coinjected into let-502(sb108); mel-11(it26) animals, which were then brooded at the nonpermissive temperature of 25°. Since let-502(sb108) is hypomorphic and mel-11(it26) behaves as a genetic null at 25° (WISSMANN et al. 1999 ), RNAi should further decrease the already limiting amounts of the two gene products. The hatching rate was the same as for uninjected animals (90%), suggesting that normal elongation in let-502; mel-11 is the double null elongation phenotype.

Characterization of the genetic pathways that contribute to embryonic elongation:
Based on analogies with vertebrate systems, there are a number of C. elegans candidate genes that are likely to mediate embryonic elongation in concert with let-502 and mel-11. We had previously demonstrated that mutations in unc-73 and mig-2 genetically interact with mel-11 (WISSMANN et al. 1999 ). Recently, SHELTON et al. 1999 showed that mlc-4 has elongation defects similar to those of strong let-502 alleles. In the following sections we will determine how these genes precisely fit into the elongation pathway. We will also identify additional genes that genetically interact with let-502 and/or mel-11.

Enhancers of mel-11 could mediate elongation by acting in the let-502/mel-11 pathway, either as an activator of mel-11(+) or as an inhibitor of let-502(+). Alternatively, these genes could act in a parallel elongation pathway. We distinguished these possibilities by building triple mutants with let-502(ca201) and mel-11(it26). Addition of a mutation in a gene acting upstream of let-502 or mel-11 should not affect the near-normal elongation of the let-502; mel-11 double. That is, in the absence of both let-502 and mel-11, the additional loss of an upstream gene would be irrelevant. In contrast, mutations acting downstream of the let-502/mel-11 pathway would block elongation while mutations in genes acting in parallel would prevent morphogenesis because both pathways would be compromised.

mlc-4 is the likely downstream target of let-502 and mel-11: The similar epidermal expression patterns and unelongated phenotypes of let-502 and mlc-4 suggest that MLC-4 is the target for the contraction regulated by the LET-502/MEL-11 pathway during elongation (SHELTON et al. 1999 ). mlc-4/+ hermaphrodites segregated one-fourth the unelongated progeny. While few mel-11(it26) progeny hatched at 20° due to hypercontraction, the viability was increased to 23% when the hermaphrodite parent was also heterozygous for mlc-4 (Table 5). All of the survivors exhibited the Mlc-4 unelongated larval arrest phenotype, indicating that the mlc-4 failure to elongate was epistatic to the mel-11 hypercontraction. That is, mlc-4(+) is required for mel-11(-) embryos to hypercontract. Furthermore, mlc-4 blocked the near-normal elongation characteristic of let-502; mel-11, which indicates that mlc-4 acts in parallel to or downstream of let-502 and mel-11. Based on analogies with other systems, mlc-4 is likely downstream.

unc-73 functions in a pathway parallel to let502/mel-11: We previously reported (WISSMANN et al. 1999 ) that unc-73 (Rho/Rac GEF, STEVEN et al. 1998 ) enhances the maternal-effect lethality of mel-11. Table 6 shows that unc-73 appears to act downstream or on a branch of the elongation pathway different from let-502 and mel-11. Even at the semipermissive temperature, unc-73/+; mel-11 segregated only 1.4% Unc progeny, 18-fold less than the expected 25%. Similarly, at the restrictive temperature, the viability of unc-73 let-502; mel-11 was only 0.4%, 30-fold less than the let-502; mel-11 control. In this case, the enhancement of mel-11 by unc-73 was epistatic to let-502's suppression of mel-11. Similar results were obtained using the hypomorphic allele let-502(sb108) where the hatching rate decreased over 25-fold when unc-73 was included with mel-11, with the few survivors arresting as early larvae (Table 6). Together, these data suggest that unc-73 acts downstream or in a pathway parallel to let-502/mel-11.

mig-2 acts upstream of let-502 and/or mel-11: We also previously reported (WISSMANN et al. 1999 ) that mig-2 (Rho/Rac-like, ZIPKIN et al. 1997 ) enhances the maternal-effect lethality of mel-11. Table 7 demonstrates that mig-2 acts in the same branch of the elongation pathway as do let-502 and mel-11. At the semipermissive temperature mig-2 causes a 20-fold enhancement of mel-11(it26). At the restrictive temperature, no progeny survived. However, even at the nonpermissive temperature, the viability of let-502(ca201); mel-11; mig-2 was decreased <2-fold compared to let-502(ca201); mel-11. Therefore, the suppression of mel-11 by let-502 was epistatic to the enhancement of mel-11 by mig-2, consistent with mig-2 acting in the let-502/mel-11 pathway. We performed similar experiments with the hypomorphic allele let-502(sb108). As shown in Table 7, let-502(sb108); mel-11; mig-2 showed only a small decrease in hatching compared to the corresponding strain lacking mig-2. Although the interpretation is clearer using the elongation null let-502(ca201), use of the hypomorphic mutation sb108 leads to the same conclusion that mig-2 is acting upstream of let-502 and/or mel-11.

daf-2 acts in the let-502/mel-11 pathway: In other systems, the insulin receptor pathway acts upstream of the Rac (NOBES et al. 1995 ) and Rho (OHAN et al. 1999 ) signal transduction cascades. To determine the effects of the C. elegans homologs of these genes on embryonic elongation, we built double mutants between mel-11 and the daf-2 gene that encodes a C. elegans insulin-like receptor (KIMURA et al. 1997 ). daf-2 acts in genetic pathways affecting dauer formation and life span (see KENYON 1997 and RIDDLE and ALBERT 1997 for reviews; GEMS et al. 1998 ). While daf-2(e1370) had modest effects on mel-11(it26), it strongly enhanced mel-11(sb55) hypomorphic mutations, decreasing the hatching rate nearly 50-fold at 25° (Table 8). daf-2(m212) strongly enhanced mel-11(it26), decreasing hatching nearly 30-fold at 15° (Table 8). daf-2 mutations did not genetically interact with let-502 alleles (data not shown).

To determine if daf-2 is functioning in the context of other parts of the dauer/longevity pathways when it enhances mel-11, we examined the effects of mutations of other genes in the daf pathway. The gene age-1, which encodes a phosphatidylinositol-3-OH kinase (PI3 kinase, MORRIS et al. 1996 ), ordinarily acts downstream of daf-2. As shown in Table 8, age-1 strongly enhanced mel-11(it26). daf-16 (OGG et al. 1997 ) suppresses the dauer and longevity phenotypes of daf-2, but showed only a weak genetic interaction with mel-11 alone (Table 8). However, daf-16 partially suppressed the enhancement of mel-11 by daf-2(m212). Therefore, mel-11 genetically interacts with several members of the dauer/longevity pathways.

To determine if the daf-2 genetic interaction occurs upstream of let-502/mel-11 or in a parallel pathway, we examined the let-502(ca201); mel-11; daf-2 triple. As shown in Table 8, daf-2 did not affect the viability of this genotype relative to let-502; mel-11, indicating that daf-2 acts upstream of let-502/mel-11. Similar results were obtained using the let-502(sb108) hypomorph, where viability showed a modest (twofold) decrease in the presence of daf-2 relative to let-502(sb108); mel-11.

fem-2 functions in a pathway parallel to let-502/mel-11: While constructing self-sterile strains for other purposes, we noted that fem-2 enhanced both dominant and recessive phenotypes of the strong dominant-negative allele let-502(ca201): all ca201/+; fem-2 progeny showed elongation defects, ranging from early arrest to Dpy, lumpy larva and adults, while let-502(ca201); fem-2 homozygotes underwent even less elongation than let-502(ca201) (Fig 2G). This was true for both a ts allele, fem-2(b245), and a genetic null allele, fem-2(e2105) (Table 9, lines 1–3; HODGKIN 1986 ; PILGRIM et al. 1995 ).

Other mutations of let-502 were similarly enhanced by fem-2, indicating a lack of allele specificity. In the presence of fem-2, the weak/null allele let-502(ca201sb54) arrest was shifted from adult Ste to early to midstage larvae (Table 9, lines 4–5). When the phenotypically wild-type hypomorph let-502(sb108) was combined with fem-2, all larvae arrested with elongation defects (lines 6–8).

The enhancement of let-502 by fem-2 involved both maternal and zygotic activities of the two genes. Mating let-502(sb108); fem-2(b245) hermaphrodites to either let-502(-); fem-2(+) or let-502(+); fem-2(-) males resulted in partial rescue of the elongation defects in the resulting progeny (Table 9, lines 9–10); complete rescue was seen upon mating to wild-type males (line 11).

Although sexual phenotypes of fem-2 are well characterized, defects in morphogenesis have not been described. We found that both of the fem-2 alleles examined resulted in low penetrant elongation defects. At 25°, 17% of fem-2(b245ts) larvae appeared slightly small (Sma), with rounded heads (Table 9, line 12), but this phenotype became less apparent as the animals matured. With the null allele fem-2(e2105), we observed 4% of progeny arresting with a let-502(ca201)-like phenotype (Fig 2H) and another 20% of animals arrested from L2 onward, often with a Sma-like phenotype (Fig 2I) or grew to Rol adults (line 13).

Since let-502 and mel-11 suppress one another's elongation defects, a strong enhancer of let-502 would decrease let-502 activity and should suppress mel-11. This was indeed the case. At 20°, fem-2(b245) suppressed the ts maternal-effect lethality caused by mel-11(it26) by over 50-fold (Table 10, lines 1–3). At 25°, 0/1777 mel-11(it26) embryos hatched, but the presence of fem-2 increased this value to 3.1% (61/1955). The genetic null allele fem-2(e2105) also suppressed mel-11 (lines 4–5). fem-2 suppression of mel-11 is maternal since there was little or no rescue when the hermaphrodite parent was heterozygous for fem-2 (line 6). Suppression of maternal-effect lethality was also observed for the hypomorphic mel-11(sb55) mutation (lines 7–8). The putative null allele mel-11(sb56) is an adult sterile due to spermathecal defects (WISSMANN et al. 1999 ). This phenotype was not suppressed by fem-2 (data not shown).

In the triple mutant let-502(ca201)/+; mel-11; fem-2 animals, we could assess whether let-502's suppression by mel-11 or the enhancement by fem-2 predominated. Among the progeny of these hermaphrodites, 27% arrested as unelongated larvae (Table 11). These likely represented all of the let-502(ca201) homozygotes, and most of these animals arrested with the characteristic very short let-502(ca201); fem-2 phenotype seen in Fig 2G. Thus, the enhancement of let-502 by fem-2 overcomes the suppression of let-502 by mel-11, suggesting fem-2 acts in parallel to the let-502/mel-11 pathway.

We also examined unc-73/+; mel-11; fem-2 triple mutants to determine if fem-2's suppression of mel-11 would overcome the enhancement of mel-11 by unc-73. While unc-73/+; mel-11 hermaphrodites segregated few Unc progeny at 15° (1.4%, 3/219), this value increased over fourfold, to 6.2% (12/192), when fem-2 was included (Table 12). (This was less than the theoretical maximum of 25%, likely because the fem-2 allele is non-null at the temperature used. There are insufficient surviving progeny at higher temperatures to conduct the experiment.) Thus, the ability of fem-2 to block the enhancement of mel-11 by unc-73 indicates that fem-2 probably acts downstream or in parallel to unc-73.

fem-2's role during elongation is independent of other sex-determination genes: fem-2 acts in concert with a number of genes to determine the sexual identity of the C. elegans soma and germline (reviewed in SCHEDL 1997 ; MEYER 1997 ). Because interactions between fem-2 and let-502 showed a strong maternal component, we focused mainly on feminizing (rather than masculinizing) mutations so that we could assess the effects of mutant product contributed by the oocyte. The genes fem-1 and fem-3, which act at the same step of sex determination as fem-2 (KIMBLE et al. 1984 ), showed little or no genetic interactions with let-502 or mel-11, nor did they modify interactions between fem-2 and elongation genes (Table 9, lines 14–21; Table 10, lines 9–15). Significantly, a fem-3(gf) mutation that suppresses the germline sexual phenotypes of fem-2 (BARTON et al. 1987 ) did not alter the fem-2 interactions with let-502 or mel-11 (Table 9, line 21; Table 10, line 15). tra-3, which acts upstream of fem-2 in the sex-determination pathway (HODGKIN and BRENNER 1977 ), showed no genetic interactions with let-502 or mel-11 (Table 9, lines 22–23; Table 10, line 16). We also tested a gf allele of the downstream sex-determination gene tra-1 (HODGKIN 1987 ), but again no genetic interactions were seen with mel-11 (data is in MATERIALS AND METHODS). Thus, fem-2 apparently acts independently of other sex-determination genes during embryonic elongation.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

C. elegans elongation, the transformation of the embryonic ball of cells into a vermiform larva, involves dramatic epidermal cell shape changes. Microfilaments become circumferentially aligned within epidermal cells just prior to elongation, and contraction of these microfilaments results in the dramatic cell shape changes that drive elongation of the embryo (PRIESS and HIRSH 1986 ). let-502 (Rho-binding kinase), mel-11 (regulatory subunit of myosin phosphatase), and mlc-4 (nonmuscle regulatory myosin light chain) encode proteins predicted to function in a biochemical pathway that regulates contractile and relaxation events (WISSMANN et al. 1997 , WISSMANN et al. 1999 ; SHELTON et al. 1999 ). In this report, we further examined let-502's role in elongation by analyzing mutants with genetic and phenotypic properties distinct from the previously identified let-502 mutants. This analysis uncovered a new phenotype, early cleavage defects. We also used the new let-502 alleles to investigate the nature of their interactions with mel-11 and a number of other genes thought to be involved in morphogenesis. A surprising result was that the fem-2 sex-determination gene has a previously unknown role in the process.

In vertebrate smooth muscle, contraction can be triggered by phosphorylation of MLC by either MLCK or Rho-binding kinase (SOMLYO and SOMLYO 1994 ; AMANO et al. 1996 ; KLEMKE et al. 1997 ; KUREISHI et al. 1997 ; STULL et al. 1998 ; FENG et al. 1999 ). Therefore, it is possible that embryonic elongation in C. elegans could be redundant at two levels. First, let-502 Rho-binding kinase might not be necessary for elongation if the process can be triggered solely by MLCK. Second, the entire let-502/mel-11 pathway might be redundant if appropriate regulation of MLCK can elongate the embryo. In that case, let-502 and mel-11 would be necessary only in the presence of the others' wild-type activity. In this work we show that let-502 is essential, but only in the presence of wild-type mel-11 activity.

let-502 is an essential gene:
The results obtained from different assays for genetic properties (summarized in Table 1) indicated that let-502 alleles often do not fall into discrete categories. Instead, many show mixtures of dominant-negative, null, and hypomorphic properties depending on the heteroallelic combination examined. A caveat of this analysis is that the deficiency used as a benchmark for the null could delete other genes that interact with either let-502 or mel-11. unc-73 is uncovered by hDf6 and is known to interact with mel-11 (but not let-502, WISSMANN et al. 1999 ). However, this was unlikely to have influenced our results since unc-73 shows no dominant interactions with either let-502 or mel-11. hDf6 is a reasonably small deficiency (MCKIM et al. 1988 ), so dominant interactions with other, unidentified genes are also unlikely.

Our collection of different let-502 alleles demonstrates that the gene has at least three essential and genetically distinct functions during the C. elegans life cycle. These functions differ in their requirements for maternal and zygotic gene activity. Animals homozygous for the weak dominant-negative/null mutation ca201sb54 are sterile, demonstrating a strict zygotic requirement for let-502(+), likely in the somatic gonad where let-502 reporters are expressed (WISSMANN et al. 1999 ). The alleles sb103 and sb106 retain sufficient wild-type zygotic activity for partial fertility, but the low levels of maternal let-502(+) activity in the embryos that they produce are often insufficient for the early cleavages. Some sb103 and sb106 embryos have adequate maternal let-502(+) function to escape these early defects, but they often arrest soon after hatching or show morphological defects. This reveals a second embryonic requirement for let-502 during elongation. This elongation function can be provided either maternally or zygotically since embryos can be rescued by either maternal heterozygosity for let-502(+) or by mating homozygous mothers to wild-type males. The elongation phenotype is also apparent for the homozygous progeny of hermaphrodites heterozygous for the stronger dominant-negative alleles like ca201. The maternal let-502(+) contribution from ca201/+ mothers, although high enough to rescue the early cleavage defects, is insufficient (or by midembryogenesis has decayed to levels that are too low) to provide the elongation function to their homozygous progeny [which, unlike sb106 embryos, totally lack zygotic let-502(+)]. In addition to varying in their maternal vs. zygotic requirements for let-502 expression, the elongation and cleavage phenotypes, but not the adult sterility, are suppressed by mel-11.

In higher eukaryotes, Rho-binding kinase and the related Citron kinase are implicated in cytokinesis (MADAULE et al. 1998 ; YASUI et al. 1998 ). In Drosophila, recent evidence supports roles for Rho and Rho GEF in cytokinesis and cleavage furrow formation, and in mammalian cell culture elevated levels of myosin-binding subunit phosphorylated by Rho-binding kinase are found at the cleavage furrow (KAWANO et al. 1999 ; PROKOPENKO et al. 1999 ). In C. elegans, let-502 could be the Rho-binding kinase that acts in early cleavage events. We are currently investigating the let-502 cleavage defects in more detail.

Structure of LET-502:
As shown in Fig 3 and Fig 4, the newly identified let-502 alleles have no obvious correlation between the molecular nature of their mutations (conservation or proximity to residues known to be critical for kinase function; JOHNSON et al. 1996 ) and their genetic properties and phenotypes. All of the mutations are within conserved residues. Four of the new mutations sb103, sb106, sb107, and sb108 have mixtures of dominant-negative and lf properties. These mutations likely decrease but do not eliminate wild-type activity, explaining their hypomorphic properties. However, multimerization of these products with LET-502(+) or other proteins (likely through the extensive LET-502 coiled-coil domain) could decrease the activity of the resulting complexes, explaining their partial dominant-negative behavior. Our different genetic assays could emphasize either the hypomorphic or dominant-negative aspects of a mutation because each test presents the mutant protein with different interaction partners and/or has different tissue-specific requirements for LET-502(+) activity. We likely failed to isolate true null alleles because if a mutation does not completely abolish both enzymatic activity and the ability to interact with itself or other proteins, it will have dominant-negative properties. Further interpretation of these results awaits more information on the three-dimensional structure of Rho-binding kinase.

The let-502 and mel-11 elongation pathway is redundant:
Coinjecting both let-502 and mel-11 dsRNAs into let-502(sb108); mel-11(it26) animals, which already have low levels of both gene activities, resulted in progeny that still elongated. This argues that the function of each gene is required only when the other is present and that there is another pathway that can compensate for the combined loss of let-502(+) and mel-11(+) activities. This parallel pathway likely involves MLCK, which is known to trigger actin-myosin contractile events independently of Rho-binding kinase (SOMLYO and SOMLYO 1994 ; AMANO et al. 1996 ; KLEMKE et al. 1997 ; KUREISHI et al. 1997 ; STULL et al. 1998 ; FENG et al. 1999 ).

Recently SANDERS et al. 1999 showed that p21-activated kinase (PAK) can inhibit MLCK, which could alleviate the need for myosin phosphatase (MEL-11) to negatively regulate MLCK-induced contraction. CHEN et al. 1996 showed that a C. elegans PAK localizes to the epidermal cell borders during elongation. However, others have reported that PAK can induce rather than inhibit contraction (VAN EYK et al. 1998 ; SELLS et al. 1999 ; see BAGRODIA and CERIONE 1999 for discussion). In addition, a number of other protein kinases inhibit MLCK by phosphorylation, but the in vivo significance of these results is not clear (STULL et al. 1993 ; VERIN et al. 1998 ).

Consistent with the idea of redundancy, we found that mutations that enhance mel-11 fall into two groups on the basis of their interactions with the let-502; mel-11 double. Since elongation is nearly normal in the double mutant, addition of a third mutation compromising the redundant elongation pathway would be lethal. This is what we observed with unc-73 and fem-2. In contrast, genes acting upstream of let-502 and/or mel-11 should not affect the double mutant phenotype; this was observed for mig-2 and daf-2. As expected, the gene predicted to act at the convergence of the two elongation pathways, mlc-4, blocked elongation when combined with let-502; mel-11.

Fig 5 shows a summary of our genetic interactions placed in the context of vertebrate smooth muscle contraction. One might have predicted that UNC-73, a Rho/Rac GEF (STEVEN et al. 1998 ), and MIG-2, a Rho/Rac-like protein (ZIPKIN et al. 1997 ), would function in a linear pathway. However, the results reported here place them on branches of different elongation pathways. Previous results also indicate that unc-73 and mig-2 do not act exclusively in a linear pathway. ZIPKIN et al. 1997 reported that unc-73; mig-2 has cell and nuclear migration phenotypes and fertility defects not seen in the two single mutants indicating that the genes function in different (redundant) pathways. We previously reported that unc-73(null) more strongly enhanced mel-11 mutations than did mig-2(null), which again implies that unc-73 cannot act solely through mig-2 (WISSMANN et al. 1999 ). The UNC-73 Rho/Rac GEF could interact with another Rac to activate PAK, as shown in Fig 5.

View larger version (20K): [in this window] [in a new window] [Download PPT slide]   Figure 5. Pathway of C. elegans elongation, based on our genetic results and analogies with other systems. Positive and negative interactions are indicated by pointed and barred arrows, respectively. Dashed arrows indicate that precise relationships are not yet clear. In smooth muscle, the demonstrated biochemical activities of Rho, Rho-binding kinase, myosin phosphatase, regulatory myosin light chain, PAK, and MLCK place them in the pathway as shown. The genetic interactions of corresponding C. elegans genes (let-502, mel-11, and mlc-4) are consistent with placing them in the same positions of the pathway as their smooth muscle homologs. Rho, PAK, and MLCK have not been tested for their ro