Rho-binding kinase and myosin phosphatase regulate the contraction of actomyosin filaments in non-muscle and smooth muscle cells. Previously, we described the role of C. elegans genes encoding Rho-binding kinase(let-502) and myosin phosphatase targeting subunit (mel-11)in epidermal cell-shape changes that drive morphogenesis and in spermathecal contraction. Here we analyze their roles in a third contractile event,cytokinesis within early embryos. We demonstrate that these genes function together to regulate the rate of cleavage furrow contraction, with Rho-binding kinase/LET-502 mediating contraction, whereas myosin phosphatase/MEL-11 acts as a brake to contraction: early embryonic cleavage often fails or is slowed when let-502 is mutated, whereas mel-11 mutations result in ectopic furrowing and faster furrow ingression. These phenotypes correspond to changes in the levels of phosphorylated regulatory non-muscle myosin light chain (rMLC). The gene products of let-502 and mel-11colocalize at cleavage furrows, and their mutations alleviate one another's defects. rMLC is phosphorylated in let-502; mel-11 double mutants,indicating that a kinase is able to phosphorylate rMLC in the absence of both LET-502 and MEL-11. Genetic and molecular epistasis experiments place LET-502 and MEL-11 in a cytokinetic pathway. LET-502 and MEL-11 regulate the activity of non-muscle myosin after actin, non-muscle myosin heavy chain/NMY-2,regulatory non-muscle myosin light chain/MLC-4 and early formin/CYK-1 have formed a contractile ring. Proteins including Rho GTPase activating protein/CYK-4 and late CYK-1, which are required for late stages of cytokinesis, function downstream of LET-502 and MEL-11.
Cytokinesis, the final step of mitosis, is the separation of the cytoplasm and replicated nuclear contents into daughter cells through actomyosin ring contraction. The ring forms during anaphase and contracts during telophase to form two daughter cells. Furrowing terminates after the ring fully ingresses and contacts an interzone bundle of microtubules, the central spindle(Raich et al., 1998;Field et al., 1999;Jantsch-Plunger et al., 2000;Robinson and Spudich, 2000;Glotzer, 2001). This paper focuses on genes, namely C. elegans Rho-binding kinase(let-502) and myosin phosphatase (mel-11), which have antagonistic activities to regulate contraction of the actomyosin ring.
Rho GTPases have been implicated in regulating the various stages of cytokinesis in eukaryotes (Drechsel et al.,1996; Mackay and Hall,1998; Prokopenko et al.,1999; Jantsch-Plunger et al.,2000; Prokopenko et al.,2000; Glotzer,2001). For example, inhibiting Rho activity blocks cytokinesis in both Xenopus and C. elegans embryos, suggesting that Rho is required for furrow formation (Drechsel et al., 1996; Jantsch-Plunger et al., 2000). Similarly in Drosophila, furrow formation fails in mutations of pebble, a Rho guanine exchange factor (GEF)(Prokopenko et al., 1999).
Rho-binding kinase (ROK) is a downstream Rho effector that is a candidate for modulating myosin activity during furrow ingression. Non-muscle myosin forms fibres at the onset of cytokinesis, which is concurrent with the localization of active myosin light chain kinase (MLCK) to the cleavage furrow(DeBiasio et al., 1996;Totsukawa et al., 1999;Poperechnaya et al., 2000). The regulatory non-muscle myosin light chain (rMLC) homologues (Drosophila spaghetti-squash and C. elegans mlc-4), the non-muscle myosin heavy chain homologue (C. elegans nmy-2) and the MLCK homologue(Dictyostelium MLCK-A) are all required for cytokinesis(Karess et al., 1991;Guo and Kemphues, 1996;Smith et al., 1996;Shelton et al., 1999). In smooth muscle and during stress fibre formation, contraction is induced by MLCK phosphorylation of rMLC (Gallagher et al., 1997; Totsukawa et al.,2000; Somlyo and Somlyo,2000; Katoh et al.,2001). Myosin phosphatase blocks contraction by dephosphorylating rMLC to counteract MLCK. Rho activates ROK, which then releases the brake to contraction by phosphorylating and inhibiting the myosin phosphatase targeting subunit (MYPT) (Kimura et al.,1996; Kawano et al.,1999; Somlyo and Somlyo,2000). ROK also directly phosphorylates rMLC in vitro(Amano et al., 1996), but it is not clear if this occurs in all in vivo systems(Sward et al., 2000). In Drosophila, planar cell polarity signaling activates ROK (Drok) to regulate the actin cytoskeleton through the regulation of MLC(Spaghettisquash) activity (Winter et al.,2001).
The precise roles that ROK and myosin phosphatase play during cytokinesis are not clear. Injection of dominant-negative ROK constructs resulted in multinuclear cells in Xenopus embryos and cultured mammalian cells owing to failed glial fibrillary acid protein (GFAP) disassembly, which is required for proper cell separation following cleavage(Yasui et al., 1998). However,the use of ROK inhibitors in cultured mammalian cells instead indicated its requirement for furrow contraction. Another ROK-related Rho effector,citron-k, localized to the cleavage furrow and midbody in HeLa cells, and transfection with dominant-negative constructs caused abnormal furrow contractions (Madaule et al.,1998; Yasui et al.,1998; Kosako et al.,1999; Kosako et al.,2000). However, in vivo studies indicate that citron-k is not the global regulator of cytokinesis and probably functions redundantly(Di Cunto et al., 2000). Myosin phosphatase activity is downregulated following mitosis, but the physiological relevance of this has not been shown(Totsukawa et al., 1999).
There is a need for determining ROK's role in cytokinesis using endogenous loss-of-function mutations rather than transfected dominant-negative mutations or chemical inhibitors (Madaule et al.,1998; Yasui et al.,1998; Kosako et al.,1999; Kosako et al.,2000), both of which could have effects on other (unknown)proteins. Using endogenous let-502 and mel-11 mutations, we previously described in vivo roles for the C. elegans ROK (LET-502)and MYPT (MEL-11) in morphogenesis and spermathecal function, two different contractile events in the worm (Wissmann et al., 1997; Wissmann et al.,1999; Piekny et al.,2000). In morphogenesis, LET-502 and MEL-11 together regulate the actin-mediated epidermal cell shape changes that drive elongation of the embryo (Wissmann et al., 1997;Piekny et al., 2000). During oocyte fertilization, LET-502 and MEL-11 each independently regulate the contraction of different tissues within the spermatheca(Wissmann et al., 1999). Here we use let-502 and mel-11 mutations to demonstrate in vivo roles for both genes in another contractile event, cytokinesis. We show that LET-502 and MEL-11 together control cytokinesis in a manner similar, but not identical to, their regulation of epidermal cell shape changes during morphogenesis. This suggests that LET-502 and MEL-11 are utilized at different stages of the life cycle for various contractile events, but each event is genetically and biochemically distinct.
Materials and Methods
Strains and alleles
C. elegans strains var. Bristol were maintained according to standard procedures (Brenner,1974). The following alleles and chromosome rearrangements were used in this study (Hodgkin et al.,1988; Hodgkin,1997): let-502(sb106, sb108), dpy-5(e61), mel-11(it26),unc-4(el120), mnCl [dpy-10(e128) unc-52(e444)] chromosome II balancer (Edgley et al., 1995), cyk-1(t1568, t1611, s2833, or36), unc-32(e189), unc-36(e251),ncl-1(e1865), qCl [glp-1(q339) dpy-19(e1259)] chromosome III balancer (Edgley et al., 1995), him-3(e1147), him-8(e1489) and cyk-4(t1689).
Microscopy and immunofluorescence
Early embryos were dissected from gravid adult hermaphrodites. For temperature-sensitive (ts) alleles, hermaphrodites were upshifted to the non-permissive temperature for 1-2 hours prior to dissection. Embryos were mounted on 3% agarose pads in M9 solution(Sulston and Hodgkin, 1988)and examined by Nomarski optics on a Zeiss Axioplan microscope. Images were photographed using either Techpan film (Kodak) developed at ASA 100 or using videotape recording.
Early embryos were collected for staining by dissecting hermaphrodites on polylysine-coated slides in M9 solution(Sulston and Hodgkin, 1988). A coverslip was placed on each slide, and the embryos were frozen on dry ice for a minimum of 30 minutes. The coverslips were removed and slides were placed in-20°C MeOH for 15 minutes, followed by 5 minutes in -20°C acetone. For actin staining only, freeze-cracked slides were placed in a 3.7% formaldehyde solution (w/v) in 75% MeOH at room temperature for 10 minutes and then placed in 100% MeOH at room temperature for 5 minutes(Waddle et al., 1994). Actin staining could be performed only with the formaldehyde-based fixation procedure; however, LET-502 and MEL-11 staining could be performed only with the methanol fixation procedure. For all methods, the slides were then placed directly into 1× phosphate-buffered saline (PBS) with 0.1% Triton X-100(PBT) buffer for a minimum of 1 hour, then incubated with appropriate dilutions of antisera in PBT with 20% normal goat or donkey serum (Jackson Immunoresearch Laboratories). Mouse anti-actin monoclonal antibodies (Sigma)were used at a 1:50 dilution, rabbit anti-NMY-2 polyclonal antibodies (K. Kemphues, Cornell University, Ithaca, NY) were used at a 1:50 dilution, rabbit anti-CYK-1 polyclonal antibodies (B. Bowerman, University of Oregon, Eugene,OR) were used at a 1:20 dilution, rabbit anti-rMLC phosphoserine-19 polyclonal antibodies (M. Walsh, University of Calgary, Calgary, AB) were used at a 1:50 dilution, P-granules were stained with mouse OIC1D4 monoclonal antibodies (S. Strome, Bloomington, IN) at a 1:1 dilution, rat anti-LET-502 polyclonal antibodies (see below) were used at a 1:100 dilution and rabbit anti-MEL-11 polyclonal antibodies (see below) were used at a 1:50 dilution. All slides were incubated with primary antibodies overnight at room temperature and washed three times with PBT prior to adding the appropriate secondary antibody. Anti-rat IgG conjugated to indocarbocyanine (Cy-3; Jackson Immunoresearch Laboratories), anti-mouse IgG conjugated to fluorescein isothiocyanate (FITC; Jackson Immunoresearch Laboratories) or Alexa 488(Molecular Probes) and anti-rabbit IgG conjugated to cyanine (Cy-2; Jackson Immunoresearch Laboratories) or Alexa 488 (Molecular Probes) were diluted 1:100 in PBT, and slides were incubated with the appropriate secondary antibody at room temperature for one hour. Slides were washed three times with PBT prior to being incubated with 1 μg/mL DAPI (Roche) for 10 minutes at room temperature. After one wash with PBT, any remaining liquid was removed with a kimwipe. A drop of Slowfade Light Antifade solution (Molecular Probes)and a coverslip were added to each slide and sealed with nail polish. Some images were photographed using a 35mm camera on a Zeiss AxioPlan fluorescence microscope with a 63× oil objective (embryos) or a 40× objective(gonads). Other images were collected for each fluorochrome as stacks of∼12-15× 1 μm from a Leica DM-R fluorescence microscope using either a 40× oil objective (embryos) or a 20× oil objective(gonads) with an Optivar tube set at a distance of 2 and images were collected with a Princeton Instruments 14-bit cooled charge-coupled device (CCD). The collected stacks were digitally deconvoluted using the nearest neighbor algorithm of Autoquant Autodeblur version 5.1 for Windows, and all images were analyzed using similar parameters. These images were then individually imported into Adobe Photoshop version 4.0 for Windows to add colour and merge images.
RNA-mediated interference (RNAi)
Double-stranded RNA was generated for both let-502 and mlc-4 as previously described(Piekny et al., 2000;Shelton et al., 1999). RNA for the citron genes was made by oligo(dT)-primed synthesis (GibcoBRL)followed by PCR of the pooled cDNA. The primers used were specific to W02B8.2:the forward primer included the T3 promoter binding sequence 5′AATTAACCCTCACTAAAGGGATGAACGAATCAATATATATAC 3′ and the reverse primer including the T7 promoter binding sequence 5′TAATACGACTCACTATAGGGTTAGTTTTTGGATCTTTTCA 3′. Primers specific to F59A6.5 were: the forward primer, which included the T3 promoter binding sequence 5′ AATTAACCCTCACTAAAGGGATGTGTGACTCTGTTTAC 3′ and reverse primer including the T7 promoter binding sequence 5′TAATACGACTCACTATAGGGCCCACGAAGCAATCCAAG 3′. The PCR product was then used for in vitro transcription (Megascript T3 and T7, Ambion). Concentrations of∼500-1000 μg/ml dsRNA were used, diluted in diethylpyrocarbonate-treated water. Wild-type, let-502(sb106) or mel-11(it26) unc-4 L3 and L4 stage larvae were collected(∼100-200) and placed into microfuge tubes with 20-40 μl of dsRNA solution and left overnight at 20°C. The soaked worms were then pipetted onto fresh Escherichia coli seeded plates and allowed to recover for 12-24 hours. Worms of similar stages were then placed into groups of between three and five on seeded plates and allowed to lay eggs overnight. Worms were collected from plates with low hatching rates (0-15%) and dissected as described above to observe their embryos. (Worms were soaked rather than injected as let-502(sb106) hermaphrodites could not recover well from injection and to allow for the collection of large numbers of embryos.)
LET-502, MEL-11 and rMLC phosphoserine-19 antisera
Rat polyclonal antibodies were raised against a HIS-LET-502 fusion using the pQE30 vector and the QiaExpression kit (Qiagen) with 103 amino acids corresponding to a region between the kinase domain and the coiled-coil region(encoded by 309 bp from the start of exon 5, using the following primers:forward 5′ GGTGGATCCAAATCGGACGATGAC 3′ and reverse 5′GGGGTCGACTTCTCGGTTTTTCGA 3′). Antisera were affinity purified with a GST-LET-502 fusion using the same LET-502 fragment as described above cloned into the pGEX-3X vector and glutathione S-transferase (GST) expression system(Pharmacia) coupled to a cyanogen-bromide-activated Sepharose column(Pharmacia). Western blot analysis showed that the affinity-purified antiserum recognized one band at ∼130 kDa (expected Mr 129 kDa)with gravid adult hermaphrodite extracts solubilized with 1 M NaCl, implying that a LET-502 isoform is preferentially associated with the cytoskeleton. This band also was detected in 1× PBS extracts, suggesting that some LET-502 isoforms are in the cytoplasm. Two other bands that differ in size by only a few kDa (∼130-140 kDa) were also seen in the 1× PBS extracts,implying that several LET-502 isoforms exist (owing to alternative splicing)and/or some of the isoforms are phosphorylated or partially degraded. All bands were blocked by adding excess GST-LET-502 to the antisera. Decreased immunostaining in the let-502 mutants further supported the conclusion that the antisera are specific (see Results).
Rabbit polyclonal antibodies were raised against a GST-MEL-11 fusion containing 64 amino acids from a portion 3′ to the ankyrin repeats(encoded by 192 bp from the start of exon 13, using forward primer 5′CAGGATCCGCAGTCGTCCAACAGAAC 3′ and reverse primer 5′GCGAATTCCGGCAACCGATAAAT 3′). The antisera were affinity purified using a cyanogen-bromide-activated Sepharose column coupled to the MEL-11 fragment cleaved from the GST-MEL-11 protein. Western blot analysis showed that the antisera recognize five bands in the ∼110-120 kDa size region (which is consistent with the expected size range for the five splice variants described by Wissmann et al., with extracts from gravid hermaphrodites in 1× PBS(Wissman et al., 1999). Some of these bands were also visualized when solubilized in 1 M NaCl, suggesting that some of the MEL-11 isoforms are preferentially associated with the cytoskeleton, whereas others are in the cytoplasm. Multiple MEL-11 isoforms exist because of alternative splicing, as observed by Wissmann et al., but in addition it is likely that some MEL-11 isoforms are phosphorylated or degraded (Wissman et al., 1999). All bands were blocked by incubating the antisera with excess GST-MEL-11 protein. Decreased immunostaining in mel-11 mutants provide further evidence that the antisera are specific (see Results).
Polyclonal antibodies that specifically recognize rMLC (20 kDa light chain of myosin) phosphorylated at serine 19 were raised in rabbits by injection of a peptide (KKRPQRATS(P)NVFC) corresponding to residues 11-22 of the chicken light chain with a phosphoserine at position 19 and a C-terminal Cys coupled to KLH. Antibodies were purified from the IgG fraction by peptide affinity column chromatography and shown to be specific for rMLC phosphorylated at serine 19 by western blotting. The antibodies were generously donated by Michael P. Walsh (University of Calgary, AB) and his laboratory.
let-502 and mel-11 control cleavage furrow contraction
We previously described a role for let-502 and mel-11 in morphogenesis and oocyte fertilization, two different contractile events in the worm (Wissmann et al.,1997; Wissmann et al.,1999; Piekny et al.,2000). More recently, we identified two let-502 alleles(sb103 and sb106) that show strict maternal-effect embryonic lethality owing to defects in another contractile event, cytokinesis(Piekny et al., 2000). One of these alleles, sb106, behaves as a weak dominant-negative for the abnormal morphogenesis phenotype but acts as a hypomorph for the cytokinetic defect (Piekny et al., 2000). Here we describe let-502's role in cleavage in detail by further analyzing let-502(sb106)'s cytokinetic phenotype. We previously described only morphogenetic and sterility phenotypes for mel-11(Wissmann et al., 1997;Wissmann et al., 1999;Piekny et al., 2000), and here we show that mel-11 embryos also have non-lethal cleavage defects.
Embryos from hermaphrodites mutant for mel-11(it26) showed ectopic furrowing during both pseudocleavage (the incomplete cleavage that occurs during pronuclear migration after fertilization) and during early cell divisions (Fig. 1, compare wild-type A-F with mutant G-L). mel-11(it26) is a temperature-sensitive allele that behaves as a null at the restrictive temperature. Cells in mel-11(it26) embryos initiated their nuclear divisions and cleavage furrow ingressions at the correct time, but furrow ingression was completed approximately twice as fast as in wild-type embryos(Fig. 2). For example in the mel-11(it26) embryo in Fig. 1T, the AB cell (the anterior blastomere) completed division prior to the initiation of the division of the P1 cell (the posterior blastomere). In wild-type embryos (Fig. 1S),AB completed its cleavage only after cleavage was initiated in P1. mel-11's maternal-effect lethality is rescued by mating to wild-type males (Kemphues, 1988; Wissmann et al.,1999). Because the observed cytokinetic defects occur too early to be rescued by a paternally contributed wild-type allele, mel-11's maternal-effect lethality must primarily stem from defects in elongation rather than cytokinesis.
let-502(sb106) embryos displayed phenotypes opposite to those seen in mel-11(it26), with embryos showing little or no pseudocleavage and slowed or failed cytokinesis (Fig. 1M-R). In let-502(sb106) embryos with successful first divisions, the AB cell furrow initiated at the correct time but ingressed at only half the wild-type speed (Fig. 2). For example, in the let-502(sb106) embryo inFig. 1U, P1 completed its division prior to AB. In let-502(sb106) cells that failed to complete cytokinesis, furrows initiated at the correct time and ingressed slightly but then regressed (Fig. 1O,P). The ability of furrows to ingress even slightly could be because of residual LET-502 activity or may reflect the activity of a partially redundant pathway.
Failed cell divisions occur (apparently) at random in let-502(sb106) embryos, with many divisions being normal. Again, this could be the result of either residual LET-502 activity or a partially redundant pathway. We examined let-502(sb106) embryos treated with dsRNA (used for RNAi) as they presumably would have LET-502 levels depleted to a greater extent. Indeed, let-502(sb106RNAi) embryos had decreased embryonic viability (i.e. at 25°C, let-502(sb106) had 41%hatching and let-502(sb106RNAi) had 0%). However, divisions still failed at random with the number of successful cell divisions at 25°C as follows: six had no divisions, two had one division, four had two divisions,five had four divisions and the remaining 35 embryos all arrested prior to morphogenesis (probably indicating that at least one fatal division error occurred prior to that time). Therefore let-502(+) is required for the embryo to complete all of its cell divisions correctly but is not required for the successful completion of every cell division.
The above results suggest that let-502 and mel-11 have antagonistic activities during cleavage furrow ingression. We predict that mutations in let-502 and mel-11 should alleviate one another's cleavage defects, as previously we had reported that both let-502's early cleavage defects and elongation phenotype (which are genetically separable) are suppressed by mel-11. At 25°C, let-502(sb106) had 41% hatching, mel-11(sb55) had 25.2%hatching and let-502(sb106); mel-11(sb55) had 90% hatching(Wissmann et al., 1997;Piekny et al., 2000). Indeed,combining a hypomorphic let-502 allele, sb106 or sb108 (the latter displays no cleavage defects) with mel-11(it26) resulted in furrow ingression times near to wild-type rates, with few failed divisions (Fig. 2).
In summary, let-502 mutants have slow furrow ingression, implying that LET-502(+) is required for contraction to proceed. In contrast, mel-11 mutants show faster furrow ingression and ectopic furrows,suggesting that MEL-11(+) acts as a brake to contraction and prevents furrow formation at inappropriate locations. The genetic interactions between let-502 and mel-11 imply that they influence each other's activities in the furrow. Together, LET-502 and MEL-11 regulate the rate or force of cleavage furrow contraction, which is essential for a high fidelity of successful cell divisions.
LET-502 and MEL-11 proteins are enriched in cleavage furrows
Our evidence indicates that let-502 and mel-11 regulate cleavage furrow ingression. Therefore, we examined LET-502 and MEL-11 localization at furrows using immunofluorescence (IF) with antibodies raised against each protein (see the Materials and Methods). In wild-type embryos,LET-502 and MEL-11 both localized to cleavage furrows at early stages of furrow ingression and remained there during later stages of ingression(Fig. 3iA-F). Anti-MEL-11 also faintly stained the central spindle during late anaphase, but this structure was not noticeably altered in mel-11(it26) mutant embryos stained with anti α-tubulin (data not shown).
We examined LET-502 and MEL-11 localization in mutant embryos to verify that their molecular and genetic phenotypes are consistent and to determine whether they are co-dependent on one another for their localization. Cytoplasmic LET-502 levels were decreased but not eliminated in let-502(sb106) embryos, consistent with the interpretation that let-502(sb106) is a hypomorph(Fig. 3iG,I). In let-502(sb106RNAi) embryos, LET-502 levels were depleted to a greater extent than in let-502(sb106) embryos, and indeed LET-502 was often undetectable (see Fig. 5F,Fig. 7D below). In addition,MEL-11's intensity at furrows appeared to weaken in let-502(sb106)embryos (Fig. 3iH). Although MEL-11 staining was severely depleted in mel-11(it26) embryos raised at the non-permissive temperature, LET-502 localization was unaffected(Fig. 3iJ-L). Therefore,LET-502 is not dependent on MEL-11 for its localization to cleavage furrows,but MEL-11 may be partially dependent on LET-502.
We used deconvolution microscopy to determine the extent of overlap between LET-502 and MEL-11 at cleavage furrows. Initially we used wild-type embryos(Fig. 3iiA-C); however, high cytoplasmic LET-502 often obscured our visualization of LET-502 at the cleavage furrow. Using the hypomorph let-502(sb106), which has decreased LET-502 in the cytoplasm (see above), LET-502 clearly overlapped with MEL-11 at the furrow (Fig. 3iiD-F). Embryos from hermaphrodites heterozygous for the dominant-negative mutation let-502(ca201) gave similar patterns of LET-502 and MEL-11 colocalization (Fig. 3iiG-I). Kosako et al. found similar effects in cell culture experiments where the presence of ROK inhibitors decreased levels of ROK in the cytoplasm, more clearly revealing its presence at the furrow(Kosako et al., 2000).
We conclude that LET-502 and MEL-11 colocalize to ingressing cleavage furrows, implying that LET-502 and MEL-11 together regulate cleavage furrow ingression.
LET-502 and MEL-11 are not required to form cleavage furrows
The let-502 and mel-11 phenotypes suggest that they regulate the rate of cleavage furrow ingression and are not involved in furrow formation. LET-502 colocalized with NMY-2 myosin thick filaments at the cleavage furrow (Fig. 4A-C),implying that LET-502 (and by implication MEL-11) is associated with the contractile ring. [We were not able to examine the localization of LET-502 and MEL-11 with actin because of incompatible fixation requirements (see Materials and Methods)]. LET-502's association with the thick filaments of the contractile ring is consistent with LET-502 targeting either rMLC or MEL-11(and MEL-11 targeting rMLC) to regulate its activity during furrow contraction.
rho-1(RNAi) leads to failure of furrow formation in C. elegans embryos (Jantsch-Plunger et al., 2000). Since LET-502/ROK is a Rho effector, we determined whether LET-502 and MEL-11 also are involved in the initial formation of the actomyosin ring. NMY-2 and actin retained their localization at the site of furrow formation in all let-502(sb106) (n=16) and let-502(sb106RNAi) (n=16) embryos examined at the one, two and four cell stages (Fig. 4D-I), despite the fact that 33% of let-502(sb106)embryos have visible cleavage defects at these times. NMY-2 and actin similarly retained their location in mel-11(it26) embryos(Fig. 4J-O; n=29). Therefore, let-502 and mel-11 probably are not required to properly localize components of the cleavage furrow to its site of formation. Interestingly NMY-2 localized to the mitotic spindle during metaphase but disappeared as the cell cycle progressed(Fig. 4G,I).
In summary, LET-502 colocalizes with NMY-2 in the contractile ring, but LET-502 and MEL-11 are not required for initial furrow formation and probably are recruited to the furrow after the actomyosin ring is formed.
LET-502 and MEL-11 regulate cleavage furrow contraction by altering the levels of active rMLC
The let-502 and mel-11 phenotypes imply that they regulate the rate of cleavage furrow contraction, most probably by regulating rMLC, which is activated by phosphorylation of serine 19 in higher eukaryotes(which corresponds to serine 18 in C. elegans). We found that an antibody raised against rMLC phosphoserine 19 in chickens (M. Walsh,University of Calgary, personal communication) specifically crossreacted with MLC-4 in C. elegans. Using this reagent, we observed that high amounts of rMLC phosphoserine 19/18 were detectable in the cytoplasm and the cortex and in the cleavage furrow of wild-type embryos, and rMLC phosphoserine 19/18 colocalized with LET-502 at all locations(Fig. 5A). Furrow and cytoplasmic rMLC phosphoserine 19/18 levels were elevated in mel-11(it26) mutant embryos (Fig. 5B), whereas these levels were greatly decreased in let-502(sb106) mutant embryos(Fig. 5C). The antisera were specific for the rMLC encoded by mlc-4(Shelton et al., 1999), which is required for early C. elegans cleavages, since rMLC phosphoserine 19/18 was either non-detectable (Fig. 5D) or present a very low levels(Fig. 5E) in mlc-4(RNAi) embryos. In addition, the antisera recognized a single band at the appropriate molecular weight on western blots (data not shown). rMLC phosphoserine 19/18 levels were restored close to wild-type levels in let-502(sb106RNAi) mel-11(it26) embryos(Fig. 5F).
In summary, LET-502 appears to positively regulate contraction either by direct phosphorylation of MLC-4 or indirectly by inhibiting MLC-4 dephosphorylation by MEL-11. Probably, there is a threshold level of active rMLC required for successful cytokinesis, and cells that lack this level fail to undergo cleavage. The double mutants restored sufficient amounts of active rMLC in the furrow for successful cleavages to occur. This suggests that another kinase, other than LET-502/ROK, phosphorylates rMLC in these circumstances.
Where do LET-502 and MEL-11 fit into the cytokinetic pathway?
Our evidence suggests that LET-502 and MEL-11 regulate contraction of the actomyosin ring after furrow formation but prior to the end of cleavage furrow ingression. We performed molecular and genetic epistasis experiments and placed LET-502 and MEL-11 in a pathway with respect to other proteins known to be involved in furrow progression. We tested genes encoding MLC-4 (rMLC),early and late CYK-1 (formin) activity and CYK-4 [Rho GTPase activating protein (GAP)]. Early CYK-1 is required for actin polymerization and the earliest stages of furrow formation (A. Severson and B. Bowerman, University of Oregon, Eugene, personal communication). Actin polymerization then is followed by or occurs in conjunction with the incorporation of NMY-2 and MLC-4 to form an actomyosin ring prior to furrow ingression(Shelton et al., 1999). Late CYK-1 and CYK-4 are required for the final stages of furrow ingression and/or termination (Swan et al.,1998; Jantsch-Plunger et al.,2000).
We examined LET-502 and MEL-11 localization in embryos defective for early cyk-1 function using embryos from cyk-1 heteroallelic hermaphrodites carrying a weak allele (showing late cytokinetic defects on its own) and a strong (sterile) allele (A. Severson and B. Bowerman, University of Oregon, Eugene, personal communication). LET-502 and MEL-11 failed to localize in this genetic background (Fig. 6A-C). Therefore, since early CYK-1 activity is involved in actin polymerization and contractile ring formation, we conclude that LET-502 and MEL-11 require a properly formed contractile ring for their localization.
As described earlier, MLC-4 is probably a substrate for MEL-11 and/or LET-502 (Shelton et al., 1999;Piekny et al., 2000), and indeed LET-502 and MEL-11 failed to localize in mlc-4(RNAi) embryos(Fig. 6D-F). It is likely that LET-502 and MEL-11 are recruited to the furrow to act on MLC-4 as a target. Consistent with both let-502 and mlc-4 acting during cytokinesis, decreased activity of these genes enhance (exacerbate) one another's defects. Weak mlc-4(RNAi) (using lower concentrations of dsRNA) results in 43% hatching, while in let-502(sb106) embryos show∼85% hatching at 20°C. If there were no genetic interactions between let-502 and mlc-4, the predicted hatching rate would be∼37%, but the observed value was 13%, implying that the genes are genetically influencing the same process.
CYK-4 is a Rho GAP required for a late step in cytokinesis(Jantsch-Plunger et al.,2000). We examined LET-502 and MEL-11 localization in cyk-4 mutant embryos, which form furrows that ingress substantially but then regress (Jantsch-Plunger et al.,2000). LET-502 and MEL-11 retained their location in cyk-4 mutant embryos, implying that LET-502 and MEL-11 are not dependent on CYK-4 (Fig. 6G-I). However, let-502 and cyk-4 do interact genetically: at 15°C, let-502(sb106 or sb108) cyk-4(t1689) animals were sterile owing to germline proliferation defects (see next section). cyk-4 also was epistatic to mel-11, with embryos displaying the cyk-4 phenotype (arresting at the one cell stage).
CYK-1's late role in cytokinesis may be to stabilize ingressed cleavage furrows until they can be disassembled, since weak cyk-1 mutations result in embryos that form cleavage furrows that ingress substantially but then regress (Swan et al.,1998). LET-502 and MEL-11 distributions were not altered in weak cyk-1 mutant embryos, suggesting that LET-502 and MEL-11 are not dependent on late CYK-1 function (Fig. 6J-L). A CYK-1 antibody was used for the reciprocal experiment to determine whether LET-502 and MEL-11 are required to localize CYK-1 for its late role. CYK-1 colocalized with LET-502 in wild-type embryos(Fig. 7A-C), and it was still present at furrows in let-502(sb106RNAi) embryos, although at less intense levels than in wild-type embryos(Fig. 7D-F). Therefore, CYK-1's late role in cytokinesis may be partly dependent on LET-502. Genetically, cyk-1 and let-502 mutants strongly enhanced one another: at 20°C let-502(sb106 or sb108) cyk-1(t1568 or t1611 or or36) animals were sterile owing to germline proliferation defects (see next section). CYK-1's location at furrows was not altered in mel-11(it26) embryos(Fig. 7G-I), but cyk-1also was epistatic to mel-11, with embryos displaying the cyk-1 phenotype (arresting at the one cell stage). Therefore, CYK-1 is not dependent on MEL-11, consistent with `normal', albeit rapid, furrow ingression in mel-11.
Function of let-502 in oocyte cellularization
The C. elegans hermaphrodite gonad contains mitotically proliferating nuclei within a syncytium(Fig. 8i). These nuclei progress through meiosis and become fully cellularized into oocytes as they move through the gonad in a distal to proximal direction (relative to the vulva) (Kimble and Ward,1988). Membranes invaginate into the cytoplasm around each nucleus in the distal mitotic region but do not fully enclose the nuclei. The membranes eventually pinch off to cellularize nuclei into separate oocytes,which may occur by a mechanism similar to cytokinesis.
Genes that are required for cytokinesis, such as mlc-4 and cyk-1, also affect oocyte formation within the hermaphrodite gonad,and mutations result in varying levels of sterility(Swan et al., 1998;Shelton et al., 1999) (A. Severson and B. Bowerman, University of Oregon, Eugene, personal communication). This effect raises the question as to whether let-502regulates oocyte cellularization, which would be similar to its role in cytokinesis. let-502(sb106) hermaphrodites laid eggs of noticeably variable sizes in comparison to the uniform range seen in wildtypes. Some embryos (although not obviously polyploid) were more than twice the normal length, whereas others were smaller than the wildtype(Fig. 8ii). This variation was also reflected in the size of oocytes within the gonad where the most mature oocyte was not always the largest (as is seen in wildtype), indicating that the membranes are probably not forming at proper distances from one another(data not shown). Some oocytes were multinuclear, indicating a failure to cellularize properly. Embryo sizes were normal in mel-11 mutants, and mel-11 mutants suppressed the size defect of let-502 mutants(data not shown).
LET-502's role in cellularization is consistent with its localization in the gonad, where LET-502 and MEL-11 colocalized at the membrane invaginations surrounding each nucleus within the mitotically dividing syncitium(Fig. 8iiiA-F). The membrane invaginations still formed in let-502(sb106) gonad arms; however, the gonad arms were much thinner, and some of the membrane invaginations appeared to be disrupted, with some surrounding multiple nuclei(Fig. 8iiiG-L). In addition,LET-502 localized to oocyte cell boundaries as they became cellularized in a pattern similar to the MEL-11 and NMY-2 distribution(Fig. 8ivM-Q). rMLC phosphoserine 19/18 also localized to oocyte cell boundaries (data not shown). Therefore, LET-502's expression pattern within the gonad combined with its phenotype is consistent with let-502 being involved in oocyte cellularization. Despite the lack of an obvious gonad phenotype in mel-11 mutants, MEL-11 was present at cell boundaries, suggesting that mel-11 may also play a non-essential in oocyte formation.
LET-502 and MEL-11 are not required for early embryonic polarity
Rho GTPases (Rho and Cdc42) and non-muscle myosin (NMY-2 and MLC-4) are also involved in cell polarity (Guo and Kemphues, 1996; Shelton et al., 1999; Prokopenko et al.,2000; Kay and Hunter,2001). However, let-502 and mel-11 mutant embryos did not appear to have any polarity defects, that is, the asymmetrical events of pronuclear fusion, placement of the mitotic spindle and segregation of P granules all were normal in their cellular position, structure and timing(A.P., unpublished).
The function of ROK during cytokinesis has been investigated in vitro;however, its precise role is not clear. One report suggested that ROK is required late, for cell separation via targeting GFAP for disassembly(Yasui et al., 1998), while another indicated that ROK regulated cleavage furrow contraction by altering myosin activity (Kosako et al.,2000). These experiments employed dominant-negative mutations or drugs that could have unknown effects on other proteins. In addition, no experiments have examined the function of MYPT in cytokinesis. We show here,through loss-of-function studies of the endogenous genes, that C. elegans ROK (let-502) and MYPT (mel-11) together regulate cleavage furrow contraction during cytokinesis. This assertion is supported by the let-502 and mel-11 mutant phenotypes coupled with altered levels of phosphorylated (i.e. active) rMLC within these mutants: let-502 mutants have slow or failed furrow ingression and lower levels of phosphorylated rMLC, whereas mel-11 mutants have faster furrow contraction and higher levels of phosphorylated rMLC. Simultaneous loss of let-502 and mel-11 restores normal cleavage and close to wild-type levels of phosphorylated rMLC. This indicates that let-502 and mel-11 function antagonistically and probably redundantly with another pathway to regulate furrow contraction, and a kinase other than ROK/LET-502 can phosphorylate rMLC. let-502 and mel-11 probably are not required for cell separation since in let-502 and in let-502; mel-11 double mutant embryos, cells that successfully complete cleavage appear to separate as in wild-type embryos.
To define further the roles of let-502 and mel-11 during furrow contraction, we examined the molecular epistasis between let-502,mel-11 and other gene products known to regulate cytokinesis in C. elegans. We propose that let-502 and mel-11 function downstream of actin, myosin and early formin (CYK-1) activity to regulate cytokinesis (Fig. 9). Actin and myosin (NMY-2) retain their localization at furrows in the absence of LET-502 and MEL-11, and LET-502 and MEL-11 localize to furrows only after ingression is initiated. In addition, LET-502 and MEL-11 no longer localize in embryos mutant for early cyk-1 function. CYK-1/formin may regulate actin polymerization and functions early during contractile ring assembly (A. Severson and B. Bowerman, University of Oregon, Eugene, personal communication). Actin and NMY-2 are then incorporated into the contractile ring, followed by recruitment of LET-502 and MEL-11.
The likely target of LET-502/ROK and MEL-11/MYPT is MLC-4/rMLC, whose phosphorylation triggers contraction in smooth muscle and stress fibre formation (Amano et al., 1996;Kimura et al., 1996;Kawano et al., 1999;Somlyo and Somlyo, 2000;Sward et al., 2000). In higher eukaryotes, ROK induces contraction by either activating rMLC by direct phosphorylation (Amano et al.,1996) (see also Sward et al.,2000) or indirectly by downregulating MYPT, which itself inhibits rMLC by dephosphorylation (Kimura et al.,1996; Kawano et al.,1999). Similarly, our previous genetic studies showed that MLC-4 is probably a target of LET-502 and MEL-11 during morphogenesis, the contractile event that drives elongation(Piekny et al., 2000). In this report we have shown that levels of phosphorylated (active) rMLC are dependent on LET-502 and MEL-11 and that LET-502 and MEL-11 localizations are disrupted when MLC-4 is removed. Therefore, we propose that LET-502 and MEL-11 regulate cytokinesis in a manner similar to their homologues in smooth muscle contraction and stress fibre formation in higher eukaryotes. LET-502 and MEL-11 regulate MLC-4/rMLC activity to control contraction. The balance of LET-502 and MEL-11 is crucial; when this balance is disrupted, as in the single mutants, cleavage fails to occur properly.
Furrow ingression is terminated through the downregulation of Rho by CYK-4/Rho GAP when the contractile ring contacts the central spindle(Jantsch-Plunger et al.,2000). Consistent with this, LET-502 and MEL-11 retain their localization in cyk-4 mutant embryos, which have furrows that ingress extensively but then regress. Therefore, CYK-4 may function after LET-502. CYK-1/formin also functions late in cleavage, possibly by regulating the stability of the ingressed furrow until it can be disassembled(Swan et al., 1998). LET-502 and MEL-11 are still localized to furrows in embryos mutant for late cyk-1 function. LET-502 could be involved in regulating CYK-1's localization for its late function, as CYK-1 localization is partially disrupted in let-502 mutant embryos.
A caveat in our work is that the mutations and RNAi may not completely eliminate LET-502 activity in early embryos. LET-502 is involved in oocyte cellularization, and completely depleting LET-502 might prevent proper oocyte partitioning, precluding the examination the role of LET-502 during the subsequent mitotic cleavages. Therefore, low levels of LET-502 sufficient for oocyte formation could be present in the embryos and also may be responsible for the cleavage furrows we sometimes observed in those embryos. However, we favour the possibility that residual oocyte formation and the sometimes normal mitotic cleavages may reflect the existence of a parallel pathway that can regulate cleavage furrow contraction in the absence of let-502.
In both cytokinesis and morphogenesis, simultaneous loss of mel-11and let-502 resembles wildtype(Wissmann et al., 1997;Wissmann et al., 1999;Piekny et al., 2000). This suggests that redundant pathways regulate contraction in the absence of both let-502 and mel-11. During elongation, this redundant pathway includes fem-2 [PP2c phosphatase(Pilgrim et al., 1995;Chin-Sang and Spence, 1996;Piekny et al., 2000]; however,genetic experiments indicate that fem-2 is not involved in the redundant pathway during cytokinesis (P.M., unpublished). In higher eukaryotes, the ROK-related kinase, citron-k, is involved in cytokinesis(Madaule et al., 1998). The C. elegans citron-like genes (F59A6.5 and W02B8.2) do not encode kinase domains, and RNAi to either gene in wild-type or let-502(sb106) worms has no effect on cytokinesis (A.P.,unpublished). Future work will be needed to uncover the elements of the cytokinetic pathway acting in parallel to let-502 and mel-11.
C. elegans ROK (LET-502) and MYPT (MEL-11) are used repeatedly for different contractile events in the worm. They regulate morphogenesis, the process whereby actin-mediated contractions within epidermal cells cause the cells to change shape to drive elongation of the embryo(Priess and Hirsh, 1986;Wissmann et al., 1997;Piekny et al., 2000). In let-502 mutants, the epidermal cells fail to contract, whereas in mel-11 mutants, the epidermal cells hypercontract. let-502and mel-11 mutants suppress one another's defects, implying that they function together (and antagonistically) to regulate the cell shape changes(Wissmann et al., 1997;Piekny et al., 2000). let-502 and mel-11 also regulate contraction of the spermatheca, an organ used for oocyte fertilization(Wissmann et al., 1999), and weak alleles of either gene cause sterility owing to failed oocyte fertilization. However, unlike elongation, let-502 and mel-11 mutants are not able to suppress one another's sterility and indeed are expressed in different regions of the spermatheca(Wissmann et al., 1999). In the present work, we describe the role of let-502 and mel-11in another contractile event, cytokinesis. As in morphogenesis, let-502 and mel-11 function together to regulate cleavage furrow ingression. However, genes involved in cytokinesis, such as cyk-1,nmy-2 and cyk-4 have no apparent role in morphogenesis(Swan et al., 1998;Guo and Kemphues, 1996;Janstch-Plunger et al., 2000),whereas genes involved in morphogenesis, such as fem-2, unc-73 and mig-2 do not appear to function in cytokinesis(Pilgrim et al., 1995;Steven et al., 1998;Zipkin et al., 1997;Wissmann et al., 1999;Piekny et al., 2000) (P.M.,unpublished). Therefore, let-502 and mel-11 act in concert with different genes during elongation and cytokinesis, suggesting that the`contractile cassette' functions with different genes in different tissues.
We thank B. Bowerman and A. Severson for the cyk-1 strains, CYK-1 antibody, mlc-4 cDNA and for communicating their unpublished results and helpful discussions. We thank K. Kemphues for the NMY-2 antibody, M. Walsh for the rMLC phosphoserine-19 antibody and S. Strome for the OIC1D4 P-granule antibody. We thank members of the McGhee and Mains laboratories and M. Walsh for discussion. This work was supported by grants from the Canadian Institutes of Health Research and the Alberta Heritage Foundation for Medical Research to P.E.M. We received nematode strains from T. Steirnagle at the Caenorhabditis Genetics Center, funded by the National Institutes of Health Center for Research Resources.