Specific requirement for two ADF/cofilin isoforms in distinct actin-dependent processes in Caenorhabditis elegans

Actin depolymerizing factor (ADF)/cofilin is an essential enhancer of actin turnover. Multicellular organisms express multiple ADF/cofilin isoforms in different patterns of tissue distribution. However, the functional significance of different ADF/cofilin isoforms is not understood. The Caenorhabditis elegans unc-60 gene generates two ADF/cofilins, UNC-60A and UNC-60B, by alternative splicing. These two ADF/cofilin proteins have different effects on actin dynamics in vitro, but their functional difference in vivo remains unclear. Here, we demonstrate that the two isoforms are expressed in different tissues and are required for distinct morphogenetic processes. UNC-60A was ubiquitously expressed in most embryonic cells and enriched in adult gonads, intestine and oocytes. In contrast, UNC-60B was specifically expressed in the body wall muscle, vulva and spermatheca. RNA interference of UNC-60A caused embryonic lethality with variable defects in cytokinesis and developmental patterning. In severely affected embryos, a cleavage furrow was formed and progressed but reversed before completion of the cleavage. Also, in some affected embryos, positioning of the blastomeres became abnormal, which resulted in embryonic arrest. In contrast, an unc-60B-null mutant was homozygous viable, underwent normal early embryogenesis and caused disorganization of actin filaments specifically in body wall muscle. These results suggest that the ADF/cofilin isoforms play distinct roles in specific aspects of actin reorganization in vivo.

ADF/cofilin changes the twist of the actin filaments (McGough et al., 1997) and weakens lateral contacts in the filaments (McGough and Chiu, 1999). ADF/cofilin binds to F-actin in a cooperative manner and generates an unstable population of filaments (McGough et al., 1997;Ressad et al., 1998).
Multicellular organisms have multiple ADF/cofilin genes that are often expressed in tissue-specific patterns, whereas simple organisms, such as yeast, have a single essential gene for cofilin (Iida et al., 1993;Moon et al., 1993). Vertebrates have two or three ADF/cofilin genes that are designated as ADF (also known as destrin) (Abe et al., 1990;Adams et al., 1990;Moriyama et al., 1990), non-muscle-type cofilin/cofilin-1 (Matsuzaki et al., 1988) and muscle-type cofilin/cofilin-2 (Gillett et al., 1996;Ono et al., 1994;Thirion et al., 2001). In mice, cofilin-1 is expressed in most cell types, whereas cofilin-2 is predominant in muscle cells and ADF is specific for epithelia and endothelia (Mohri et al., 2000;Ono et al., 1994;Vartiainen et al., 2002). These ADF/cofilin isoforms exhibit different actin-regulatory activities. ADF is more potent in depolymerization than cofilin Giuliano et al., 1988;Nishida et al., 1985), whereas cofilin apparently remains bound to filaments Nishida et al., 1984). Detailed comparisons of ADF and cofilin revealed that they have nearly identical activities to accelerate depolymerization and filament severing, but are different in their nucleating activities to initiate polymerization when they make complexes with actin (Yeoh et al., 2002). The cofilinactin complex initiates spontaneous polymerization more efficiently than the ADF-actin complex. The difference is significant at a high pH and confers their pH-sensitive activities (Yeoh et al., 2002;Yonezawa et al., 1985). Cofilin-2 is more effective in enhancing polymerization than ADF and cofilin-1 (Vartiainen et al., 2002), suggesting that the critical concentration of the cofilin-2-actin complex is very low.
In cells where multiple ADF/cofilin isoforms are expressed, the isoforms behave very similarly in localizing to stressinduced intranuclear actin rods  and in translocating to the lamellipodia after growth factor stimulation (Meberg et al., 1998). Also, many extracellular stimuli regulate phosphorylation levels of both ADF and cofilin in common pathways (Kanamori et al., 1995;Meberg et al., 1998;Saito et al., 1994). However, ADF and cofilin respond differently to changes in some of the cellular states. Colocalization of ADF with monomeric actin is enhanced upon an increase in intracellular pH, whereas that of cofilin is less sensitive to pH changes (Bernstein et al., 2000), which is consistent with their in vitro properties (Yeoh et al., 2002). Expression of ADF, but not cofilin, is downregulated by an increase in the actin monomer pool (Minamide et al., 1997). In contrast, expression of cofilin, but not ADF, is upregulated in dystrophic muscles (Hayakawa et al., 1993;Nagaoka et al., 1996). Thus, ADF and cofilin might have both redundant and non-redundant functions.
There are some functional differences among ADF/cofilin isoforms from studies using model organisms in which mutations in ADF/cofilin genes have been isolated. The slime mold Dictyostelium discoideum has two cofilin genes, cofilin-1 and cofilin-2 (Aizawa et al., 2001). A null mutant of cofilin-1 has not been isolated in previous attempts of gene knockout, suggesting that cofilin-1 is essential for cell viability (Aizawa et al., 1995), whereas cofilin-2-null cells develop normally (Aizawa et al., 2001). The nematode Caenorhabditis elegans expresses two ADF/cofilins from the unc-60 gene (McKim et al., 1994;Ono and Benian, 1998). Point mutations of UNC-60B, a muscle-specific isoform, cause defects in actin organization specifically in body wall muscle . A lethal unc-60 mutation has been isolated, but the responsible isoform for lethality has not been identified (McKim et al., 1994). Multiple ADF/cofilin isoforms have also been identified in Drosophila (Edwards et al., 1994;Goldstein and Gunawardena, 2000;Gunsalus et al., 1995) and plants (Dong et al., 2001a;Dong et al., 2001b;Lopez et al., 1996), but functional characterization and comparison of the isoforms are still limited. In this study, we investigated the function of the two ADF/cofilin isoforms in C. elegans using isoformspecific RNA interference and gene knockout. We found that each isoform was required for distinct actin-dependent processes. The results suggest that multicellular organisms have functionally different ADF/cofilin isoforms to support the complexity of tissue organization.

Northern blot
A northern blot containing total RNA from different developmental stages was kindly provided by Edward Kipreos (University of Georgia, Athens, Georgia). This blot was first hybridized with a probe for 18S rRNA as described previously (Kipreos et al., 1996), then stripped and re-hybridized with a probe specific for unc-60A, stripped and re-hybridized with a probe specific for unc-60B. unc-60 probes were created by PCR from a cDNA library (kindly provided by Robert Barstead, Oklahoma Medical Research Foundation, Oklahoma City, OK), and 50 ng was labeled with 32 P by random primer method. Hybridization was conducted overnight at 42°C in 150 µg/ml singlestranded salmon sperm DNA, 5×SSC (0.75 M NaCl, 0.075 M trisodium citrate, pH 7.0), 1×Blocking Quencher (Molecular Research Center, Inc.), 0.2% SDS and 50% formamide. Final washes were at 65°C in 0.5× SSC, 1% SDS and 2×Denhardts solution (0.04% Ficoll 400, 0.04% polyvinylpyrrolidone, 0.04% bovine serum albumin).
Immunofluorescence microscopy Worm embryos were obtained by cutting gravid adults on poly-lysinecoated slides, freeze-cracked as described previously (Epstein et al., 1993) and fixed with methanol at -20°C for 5 minutes. They were washed with phosphate-buffered saline (PBS) for 10 minutes and stained with antibodies diluted in 1% bovine serum albumin in PBS. Gonads were dissected by cutting adult worms at the level of the pharynx on poly-lysine-coated slides, freeze-cracked, fixed and stained in the same manner as embryo staining. Antibody staining of adult worms was performed as described elsewhere (Finney and Ruvkun, 1990).

RNA interference experiments
The cDNAs containing entire coding regions of UNC-60A or UNC-Functions of ADF/cofilin isoforms 60B were cut out of bacterial protein expression vectors pET-UNC-60A or pET-UNC-60B (Ono and Benian, 1998) and cloned into L4440 (provided by Andrew Fire, Carnegie Institution of Washington, Baltimore, Maryland) at the cloning site between two oppositely oriented T7 promoters (Timmons and Fire, 1998). E. coli HT115 (DE3), an RNase III-deficient strain (Timmons et al., 2001), was transformed with the plasmids and used for RNA interference (RNAi) by feeding worms essentially as described previously (Ono and Ono, 2002). For control experiments, HT115 (DE3) was transformed with L4440 with no insert and used to feed worms. For scoring brood size and phenotypes, a single L4 larva was placed in an RNAi-feeding plate (60 mm diameter) and transferred to a new plate after 24 hours when they became an adult. The adult worm was transferred to a new plate every 24 hours and the progeny was examined 1 day after the adult had been removed. For time-lapse recording of embryogenesis, 20 L4 larvae were cultured in an RNAi-feeding plate (100 mm diameter) for 24 hours, transferred to a new plate and cultured for 24 hours. Then, the adults were dissected by a pair of 26G needles in egg salts (118 mM NaCl, 48 mM KCl, 2 mM MgCl2, 2 mM CaCl2, 10 mM HEPES-NaOH, pH 7.4) and the released embryos subjected to time-lapse differential interference contrast (DIC) microscopy.

Time-lapse DIC microscopy
Embryos were mounted on a 2% agarose pad with egg salts and covered with a glass coverslip. They were set on a Nikon Eclipse TE2000 inverted microscope and observed with a 40× CFI Plan Fluor objective with DIC optics. Images were captured by a SPOT RT Monochrome CCD camera (Diagnostic Instruments) and recorded every 15 seconds using the IPLab imaging software (Scanalytics, Inc.).

Sequencing of genomic DNA
Total DNA was prepared from unc-60 homozygotes as described previously (Sulston and Brenner, 1974) and used as a template for PCR reactions. Several primer sets were used to amplify 1 to 2 kb genomic DNA fragments from different regions of the unc-60 gene by REDTaq DNA polymerase (Sigma-Aldrich) to determine the location of deletions in the unc-60 alleles. DNA fragments with deletions were sequenced with an ABI PRIZM dye terminator cycle sequencing kit and an ABI310 genetic analyzer (Applied Biosystems) to determine precise deletion sites.

Western blot
Total worm lysates were prepared as described previously (Ono and Ono, 2002). Protein concentrations of the lysates were determined by a filter paper dye-binding assay . 10 µg of proteins from each lysate was separated on a 15% SDSpolyacrylamide gel and transferred onto a polyvinylidene difluoride membrane (Immobilon-P: Millipore) using a Genie Blotter (Idea Scientific). The membranes were blocked in 5% non-fat milk in PBS containing 0.1% Tween 20 and incubated for 1 hour with anti-UNC-60A, anti-UNC-60B or anti-actin (C4; ICN Biomedicals) antibodies followed by treatment with peroxidase-labeled goat anti-rabbit IgG or goat anti-mouse IgG (Pierce Chemical Co.). The reactivities were detected with a SuperSignal chemiluminescence reagent (Pierce Chemical Co.).

Motility assay
A motility assay was performed as described previously (Epstein and Thomson, 1974). Briefly, adult worms were placed in M9 buffer (22 mM KH2PO4, 42 mM Na2HPO4, 85.5 mM NaCl, 1 mM MgSO4). Then, one beat was counted when a worm swung its head to either left or right. The total number of beats in 30 seconds was recorded.

Results
Two ADF/cofilins are expressed in different patterns during development and among tissues Expression of mRNAs for unc-60A and unc-60B was examined during C. elegans development by using a northern blot (Fig.  1). The unc-60A mRNA was abundant in the embryos (Fig. 1, lane 1), declined at the L1 stage (Fig. 1, lane 2) and gradually increased during larval development ( Fig. 1, lanes 2-5). Gravid adults had more unc-60A mRNA than young adults (Fig. 1, compare lanes 6 and 7), which was probably because early embryos were carried by the uterus of the gravid adults. Expression of the unc-60B mRNA was highest in the embryos (Fig. 1, lane 1) and gradually decreased during later development ( Fig. 1, lanes 2-7). There was no increase in the unc-60B mRNA in the gravid adults compared with the young adults ( Fig. 1, compare lanes 6 and 7), suggesting that unc-60B is not expressed in the early embryos. These results indicate that the mRNA levels of the ADF/cofilin isoforms are differentially regulated during development.
Expression and localization of UNC-60A and UNC-60B were characterized by immunofluorescent staining of wild-type C. elegans with specific antibodies. UNC-60A was maternally expressed and present in most of embryonic cells throughout embryogenesis ( Fig. 2). At the one-cell stage, UNC-60A was diffuse in the cytoplasm and localized to no particular structure (Fig. 2a,d). Actin was localized throughout the cortex as well as in the diffuse cytoplasm ( Fig. 2b) and assembled at the early cleavage site in the telophase of the first mitosis ( Fig. 2e), whereas UNC-60A did not significantly accumulate in these structures (Fig. 2c,f). During progression of cytokinesis and after the two-cell stage, association of UNC-60A with the cortical regions, especially at the cell-cell boundaries became evident (Fig. 2g). In addition, UNC-60A was colocalized with actin to dot-like structures at the anterior cortex of each blastomere ( Fig. 2g-i). These are believed to be remnants of cell division (Waddle et al., 1994). In the later stages, staining of UNC-60A in the nerve ring became intense (Fig. 2j). In contrast, we have previously demonstrated that UNC-60B is not expressed in early embryos and become detectable from the comma to 1.5-fold stages (approximately 290 minutes after first cell division) only in the body wall muscle cells Ono and Ono, 2002). UNC-60B is diffusely localized in the cytoplasm of the embryonic body wall muscle in the later stages as demonstrated previously Ono and Ono, 2002).
In adult worms, UNC-60A was highly expressed in the intestine (data not shown), germ cells in the distal gonad, oocytes and spermatheca ( Fig. 3a) but not in body wall muscle (data not shown). By contrast, UNC-60B was expressed in body wall muscle , vulva (data not shown) and spermatheca (Fig. 3b). Both ADF/cofilin isoforms were not detected in the smooth-muscle-like myoepithelial sheath cells (Fig. 3c,d). Localization of the nuclei of the gonad by DAPI indicated that both UNC-60A (Fig. 3a,e,g) and UNC-60B (Fig.  3b,f,h) were mostly localized to the cytoplasm. Also, they were undetectable in sperm, which does not have an actomyosin system (Nelson et al., 1982) (data not shown).
UNC-60A is essential for early embryonic development To characterize the in vivo roles of UNC-60A and UNC-60B, we suppressed expression of each ADF/cofilin isoform by RNAi and characterized the resultant phenotypes. When worms were fed with bacteria expressing dsRNA for unc-60A, they produced many dead embryos and a greatly reduced number of developed progeny (Table 1). However, treatment with dsRNA for unc-60B caused did not alter brood size (Table  1), morphology or motility (data not shown). Immunostaining of the unc-60A (RNAi) embryos showed remarkable reduction of the UNC-60A protein to somewhat variable extents (Fig. 4). In control embryos (Fig. 4ac), UNC-60A was distributed in the same patterns as in wild-type embryos under standard culture conditions (Fig. 2). The unc-60A (RNAi) treatment significantly reduced the cytoplasmic staining of UNC-60A ( Fig. 4d-f), whereas staining of some structures that resembled cell-cell boundaries or premature cleavage furrows was observed (Fig. 4d, arrowheads), suggesting that UNC-60A was not completely depleted. Western blot analysis indicated that the UNC-60B protein was not decreased by feeding with the unc-60B dsRNA (data not shown). Therefore, we concluded that RNAi was successful in demonstrating the requirement of UNC-60A but not UNC-60B in embryonic development.
Time-lapse recording of the developmental process of the unc-60A (RNAi) embryos showed that there were variable defects in their early development (Fig. 5). Among 36 embryos examined, 39% had a defect in cytokinesis, 28% showed abnormal positioning of the blastomeres and 33% were apparently normal up to the eight-cell stage. There was a tendency for the cytokinesisdefective embryos to have a greater reduction in UNC-60A immunostaining than the positioning-defective or apparently normal embryos (data not shown), suggesting that the cytokinesis defect is a stronger phenotype than the positioning defect. The cytokinesis-and patterning-defective embryos were also defective in pseudocleavage (Fig. 5a-c). In the cytokinesis-defective embryos (Fig. 5b,e,h,k,n), a cleavage furrow was formed and progressed ( Fig. 5h) but regressed or disassembled before completion of cell division (Fig. 5k, Movie 1). The furrow in this embryo progressed at ~1 µm/minutes (Movie 2), which is one tenth of the speed of furrow progression in control embryos (~10 µm/minute) (Movie 1). However, regression of the furrow (~6 µm/minute) Journal of Cell Science 116 (10)  Bar, 10 µm. Functions of ADF/cofilin isoforms was relatively rapid (Movie 2). Embryos still underwent multiple rounds of nuclear division and attempts at cytokinesis, which resulted in multinucleated cells or irregularly compartmentalized embryos (Fig. 5n). In a representative positioning-defective embryo (Fig.  5c,f,i,l,o), first and second rounds of cytokinesis were apparently normal except that the cleavage of the P1 blastomere occurred slightly earlier than the control (Fig. 5l). However, subsequent positioning of the blastomeres at the four-cell stage became abnormal (compare Fig. 5m with o). In control embryos, the cleavage plane of the AB blastomere occurs parallel to the long axis of the embryo, whereas that of the P1 blastomere is set perpendicular to the long axis ( Fig. 5j, Movie 1). The resultant ABp and EMS blastomeres slide to the sides, and ABa and P2 are located at anterior and posterior poles, respectively (Fig. 5m). However, in some of the unc-60A (RNAi) embryos, although the orientation of the spindles and cleavage planes was normal (Fig. 5l), both ABa and ABp remained at the anterior side and EMS was squeezed into the center of the embryos (Fig. 5o, Movie 3). These embryos continued to divide but were not successful at completing the following morphogenesis.
In addition, we observed a defect in extrusion of the polar body and abnormal cortical activity in the unc-60A (RNAi) embryos (Fig. 6a,b). In 71% (n=17) of the unc-60A (RNAi) embryos, the polar body was not extruded after meiosis, which resulted in the appearance of the third pronucleus during pronuclear migration (Fig.  6a, arrow). This is likely to  Micrographs of control and unc-60A (RNAi) were taken at the same exposure settings. Staining of UNC-60A was greatly reduced by the RNAi treatment (compare a with d), whereas weak UNC-60A staining was still observed in some structures (d, arrowheads). Staining with the anti-tubulin antibody (b,e) ensured that these embryos were properly permeabilized. Bar, 40 µm. be due to a cytokinesis defect in meiosis. Also, 41% of the unc-60A (RNAi) embryos showed abnormally active membrane protrusion and retraction after the two-cell stage (Fig. 6b, arrowheads, Movie 3, see the lower side of cell-cell boundary at the two-cell stage), suggesting that the cell cortex became unstable, as observed with RNAi of the Arp2/3 complex (Severson et al., 2002).
Actin filaments are important for establishment of the anterior-posterior (A-P) polarity at the one-cell stage (Hill and Strome, 1988;Hill and Strome, 1990). However, the A-P polarity is not severely disturbed by the unc-60A (RNAi) treatment (Fig. 6c-e). We measured positions of pronuclear meeting (Fig. 6c) and the first mitotic spindle (Fig. 6d), which are normally posteriorly localized (Golden, 2000). The position of pronuclear meeting was more variable in unc-60A (RNAi) embryos than in control embryos (Fig. 6e). In some unc-60A (RNAi) embryos, the pronuclei met near the center of the embryos (Fig. 5c), whereas in other embryos, it occurred at a posterior region (Fig. 5b). However, the average value of the meeting position in unc-60A (RNAi) embryos (65±8.9% egg length from the anterior pole) was not significantly different from that of control (67±4.1%) (P=0.54 by a t-test). Similarly, positions of the anterior and posterior spindle poles were slightly more variable in unc-60A (RNAi) embryos than control embryos, but the differences were not statistically significant (Fig. 6e). These results suggest that the A-P polarity Journal of Cell Science 116 (10) A cleavage furrow appeared after nuclear division (h) but regressed before completing the cleavage (k, arrows). The attempts at cleavage continued and resulted in irregular compartmentalization (n). A representative patterning defect is shown in c, f, i, l, o and Movie 3. Spindle orientation was normal at the second division (l), but the subsequent positioning of the four blastomeres was abnormal (o) compared with the control embryo (m). Bar, 10 µm. might be slightly unstable but not severely disturbed when UNC-60A was suppressed.
In control embryos, actin was uniformly localized to the cortex (Fig. 7a). However, in the unc-60A (RNAi) embryos with the multinucleated phenotype, cortical actin was unevenly distributed and enriched at one end of the cell cortex (Fig. 7d,g) and sometimes accumulated at a cleavage-furrowlike structure ( Fig. 7g; indicated by an arrow). These phenotypes suggest that UNC-60A is required for even distribution of cortical actin and proper progression of a cleavage furrow. Mitotic spindles were often found around nuclei (Fig. 7e), indicating that separation of the spindle poles and spindle assembly were not significantly affected by depletion of UNC-60A.
A null mutation of unc-60B causes specific defects in actin assembly in body wall muscle Previously characterized mutations of unc-60B were all point mutations that resulted in expression of mutant UNC-60B proteins . Therefore, although these mutants were homozygous viable and showed disorganization of actin filaments in body wall muscle, we were not able to exclude the possibility that the mutant UNC-60B proteins function sufficiently in some cellular activities other than myofibril assembly. To clearly determine the requirement of unc-60B in vivo, we need to characterize a null phenotype of unc-60B.
We collected unc-60 mutant strains from other researchers, determined their sequence alterations and found that two unc-60 alleles contain deletions in the unc-60B region (Fig. 8A). unc-60 (s1586) is homozygous lethal at a late larval stage (McKim et al., 1994). We found that this allele had a deletion that completely removes exon 2B (Fig. 8A). However, the deletion also extends to the unc-60A region and removes approximately 250 bp of the 3′-untranslated region in exon 5A that contains a putative polyadenylation signal for unc-60A. Therefore, this deletion is likely to affect expression of both ADF/cofilin isoforms, which may be a cause of the lethal phenotype. In contrast, we found that unc-60 (su158) (Zengel and Epstein, 1980) had a deletion of 600 bp that completely removed exons 3B and 4B and did not disturb the unc-60A region (Fig. 8A). Even if exons 2B and 5B are connected by an aberrant splicing event, exons 1 and 2B encodes only 30 amino acids from the N-terminus of UNC-60B and the coding region in exon 5B will be out of frame. The unc-60 (su158) mutant is homozygous viable and shows a more severe motility defect than a strong loss-of-function mutant unc-60 (e677) (Fig. 8B). In the unc-60 (su158) homozygotes, the UNC-60B protein was not detected by western blotting, whereas UNC-60A was present at an equivalent amount to that found in wildtype worms (Fig. 8C). Thus, unc-60 (su158) is a null allele of unc-60B that does not interfere with unc-60A.
The unc-60B null mutants were homozygous viable and showed a muscle-specific defect in actin organization. During embryonic development, morphological observation by DIC microscopy revealed no abnormalities (data not shown). However, staining with an anti-actin antibody revealed that, although early embryogenesis from the one-cell to 1.75-fold stages appeared to be normal (Fig. 9Ad), actin became discontinuous in the body wall muscle after the two-fold stage (Fig. 9Bd,Cd, indicated by arrows). The defect in actin organization was specifically detected in the body wall muscle cells. In addition, other myofibrillar components, myoA myosin heavy chain (Fig. 9A-Cb,e) and vinculin ( Fig. 9A-Cc,f), appeared undisturbed, suggesting that initial assembly of the myosin thick filaments and adhesion structures are independent of UNC-60B-mediated actin dynamics. Nonetheless, as a result of the muscle defects, movement of the mutant embryos in the egg shell was significantly slower than wild-type embryos (data not shown). In addition, RNAi of unc-60A on the unc-60B-null mutant resulted in similar cytokinesis and patterning defects (40% and 27%, respectively, n=15) to wildtype, and no enhancement of the phenotype was observed, suggesting that UNC-60B does not function in early embryogenesis.
When the unc-60B-null mutants grew into adult worms, they were nearly paralyzed and large actin aggregates were formed in the body wall muscle (Fig. 10). The extent of actin disorganization was slightly more severe than a strong loss-of-function mutant unc-60 (e677) . However, unlike the embryos, the organization of the myosin filaments and dense bodies in the adults was disturbed (Fig. 10). In wildtype, myosin was clearly arranged in a striated pattern (Fig. 10Aa). In the unc-60B-null mutant, myosin was assembled into wide and uneven bands with very obscure striation (Fig. 10Ad). These myosin bands were spatially segregated from the actin aggregates (Fig. 10Af). Dense bodies are adhesion structures that appear as discrete spots in the center of the I-bands, as shown by staining for α-actinin and vinculin (Fig. 10B,C). In the unc-60B null mutant, both αactinin and vinculin were aligned in striation, but the staining patterns were often continuous as if several dense bodies were fused together (Fig. 10Bd,Cd). In addition, vinculin was somewhat diffuse in some regions (Fig. 10Cd). Double staining of the dense body components with actin showed that α-actinin and vinculin were not the components of actin aggregates in the mutant (Fig. 10Bf,Cf). These results suggest that UNC-60B-dependent actin dynamics are required for proper alignment of thick filaments and dense bodies during postembryonic development, and that the actin-bundling activity by myosin, α-actinin or vinculin does not contribute to formation of the actin aggregates in the unc-60B mutants. In addition, the expression pattern of UNC-60A in the unc-60 (su158) mutant Journal of Cell Science 116 (10) Fig. 9. Assembly of myofibrillar components in the unc-60B-null mutant embryos. Wild-type (a-c) or unc-60 (su158) (d-f) embryos at the 1.75-fold stage (430 minutes after first cleavage) (A), two-fold stage (450 minutes) (B) and three-fold stage (520 minutes) (C) were stained with anti-actin (a,d), anti-myoA (b,e) and anti-vinculin (c,f) antibodies. Actin and myoA were stained by double-labeling of the same embryo. In Bd and Cd, arrows indicate the regions where actin is markedly disorganized. The age of the embryos was determined from their morphology as described previously (Epstein et al., 1993;Hresko et al., 1994). Bar, 20 µm.
was not significantly different from wild-type (data not shown), suggesting that there is no compensatory mechanism to upregulate UNC-60A in the absence of UNC-60B.

Discussion
Our study demonstrated that the two ADF/cofilin isoforms in C. elegans are required for distinct actin-dependent processes. The two ADF/cofilins, UNC-60A and UNC-60B, have different activities to enhance actin dynamics (Ono and Benian, 1998) and are expressed from the unc-60 gene by alternative splicing in developmentally regulated and tissue-specific manners. RNAi suppression of UNC-60A causes embryonic lethality with cytokinesis or patterning defects during early embryogenesis. In contrast, a null mutation of unc-60B specifically impairs myofibril assembly in body wall muscle without affecting early embryogenesis. The biochemical difference between the two ADF/cofilin isoforms are consistent with the cell biological observations that UNC-60A functions in a highly dynamic actin rearrangement during embryogenesis, whereas UNC-60B is involved in a less dynamic event in muscle cells.

Role of UNC-60A in embryogenesis
We found that UNC-60A was required for cytokinesis in early C. elegans embryos. Similarly, ADF/cofilin has been reported to be required for cytokinesis in Drosophila (Gunsalus et al., 1995;Somma et al., 2002) and Xenopus eggs , indicating that ADF/cofilin is a conserved regulator of cell division. Our immunostaining indicated that UNC-60A was not concentrated in the early accumulation of actin at the cleavage site. Rather, it was localized to progressing cleavage furrows in the late phase of cytokinesis and remained associated with cortical actin at cell-cell contacts. This is also consistent with the observations that cofilin is localized to the cleavage furrow in cultured cells (Nagaoka et al., 1995) and Xenopus eggs . This localization pattern suggests that the role of UNC-60A is in late cytokinesis and explains the cytokinesis defect by RNAi of unc-60A. In the unc-60A (RNAi) embryos, a cleavage furrow was often formed and progressed, but prematurely disassembled, suggesting that UNC-60A-mediated actin dynamics are required for efficient completion of cytokinesis. Also intriguingly, progression of a cleavage furrow in cytokinesis-defective embryos was much slower than that in control embryos, suggesting that UNC-60A-dependent actin dynamics might be a driving force for furrow progression. The contractile ring in dividing cells has been shown to be an active site of actin polymerization (Noguchi and Mabuchi, 2001;Pelham and Chang, 2002). Therefore, actin depolymerization and severing by ADF/cofilin would be required to maintain dynamic actin turnover during cytokinesis.
Similar phenotypes in late cytokinesis have been described for mutants or RNAi suppression of the formin-like protein CYK-1 (Swan et al., 1998), the kinesin-like protein ZEN-4/CeMKLP1 (Powers et al., 1998;Raich et al., 1998;Severson et al., 2000), CYK-4 Rho-GAP (Jantsch-Plunger et al., 2000), the aurora-related kinase AIR-2 (Kaitna et al., 2000;Schumacher et al., 1998;Severson et al., 2000) and the syntaxin SYN-4 (Jantsch-Plunger and Glotzer, 1999). Formin has been shown to nucleate actin polymerization (Pruyne et al., 2002;Sagot et al., 2002), but it is not known how ADF/cofilin affects this process. ZEN-4 and CYK-4 physically interact and are required for assembly of the central spindle, which is implicated in completion of cytokinesis (Mishima et al., 2002). Syntaxin mediates vesicle fusion and may be required for membrane addition at the cleavage furrow (Skop et al., 2001). However, the mechanism by which these proteins promote the completion of cytokinesis is poorly understood. How UNC-60A might be involved in these processes remains to be determined.
Defects in early embryonic patterning in the unc-60A (RNAi) embryos are probably weaker phenotypes than the cytokinesis defects, because embryos with patterning defects were successful in multiple rounds of cell division. Nonetheless, the observed phenotype is novel. Embryonic polarity and asymmetric cell division are essential for determining cell fate and embryonic patterning. Actin and myosin are required for polarized localization of the products of polarity genes (Rose and Kemphues, 1998). However, the unc-60A (RNAi) embryos showed an apparently normal pattern of spindle orientation and cleavage pattern at the second cell division, whereas subsequent positioning of the blastomeres was defective. It is still possible that the embryonic polarity is partially disrupted so that the blastomeres were not able to locate at appropriate positions. Alternatively, cortical rigidity or activity might be disturbed by unc-60A (RNAi). UNC-60A is colocalized with cortical actin at the cell-cell contacts, and unc-60A (RNAi) alters cortical distribution of actin and induces irregular membrane activity. Therefore, the cell cortex might have lost its rigidity and, thus, might be unable to push or to be pushed by the neighboring cells.
Specific requirement of UNC-60B in myofibril assembly and development Phenotypic analysis of the unc-60B-null mutant confirms our previous observations that UNC-60B is specifically required for assembly of actin . In addition, our extended analysis demonstrates that UNC-60B is not required for early development and that it is required for organized assembly of other myofibrillar components during postembryonic development. This indicates that the presence of the UNC-60A isoform is sufficient for viability of worms. UNC-60A is expressed in every embryonic cell but not in adult body wall muscle cells. The expression of UNC-60A in muscle was already low in larval stages, and so we were not able to determine when UNC-60A is downregulated (K.O. and S.O., unpublished). Although the antibody failed to detect UNC-60A in adult muscle, it is possible that a low level of UNC-60A is expressed in adult muscle and supports cell viability.
Phenotypic characterization of the unc-60B-null mutant showed that actin was disorganized from the embryonic stages onwards, whereas myosin thick filaments and dense bodies were disorganized only in adults. This suggests that UNC-60B is primarily required for actin assembly and that the effects on myosin and dense bodies are secondary and are due to disrupted thin filament organization. Therefore, the initial assembly of thick filaments and dense bodies may not require UNC-60B-mediated actin dynamics. This supports early observations that myosin and actin are initially assembled into nascent structures that are distinct from myofibrils (Epstein et al., 1993), suggesting that the assembly processes of myosin and actin are independent. However, the subsequent development and maintenance of these structures may need an organized myofibril structure that can generate contractile forces. In vertebrate striated muscle cells, inhibition of muscle contraction causes disorganization of myofibrils (De Deyne, 2000;Soeno et al., 1999). These observations suggest that a regulated actin-myosin interaction facilitates proper alignment of other myofibrillar components.
Functional significance of ADF/cofilin isoforms We demonstrated that the two C. elegans ADF/cofilin isoforms have different functions in vivo. This is consistent with the different actin-regulating activities of UNC-60A and UNC-60B (Ono and Benian, 1998). We also observed that UNC-60A accelerates subunit dissociation from F-actin more rapidly than UNC-60B (S. Yeoh, S.O. and A. Weeds, unpublished). Our results are consistent with biochemical studies using different methods showing that UNC-60A is ADF-like, whereas UNC-60B is cofilin-like (H. Chen and J. Bamburg, personal communication). UNC-60A is a more potent depolymerizing agent than UNC-60B, suggesting that UNC-60A is suitable in cells where actin filaments are dynamic, whereas UNC-60B is adapted to cells, such as muscle, in which actin filaments are relatively stable. In addition, we have recently found that UNC-78/actin-interacting protein 1 (Ono, 2001) disassembles actin filaments more efficiently in the presence of UNC-60B than UNC-60A (K. Mohri, A. G. Weeds and S. Ono, unpublished), indicating that isoform-specific interactions with other cytoskeletal proteins are also important determinants of actin dynamics.
The other mechanism that differentiates these isoforms is their differential expression in different cell types. RNAi suppression or mutations of each isoform causes a specific phenotype in which the expression of the target isoform is predominant. However, it is still possible that the two ADF/cofilin isoforms have a redundant function in some biological aspects when both isoforms co-exist. UNC-60A and UNC-60B are produced by alternative splicing of the unc-60 gene (McKim et al., 1994). Therefore, the tissue-specific splicing machinery seems to be involved in the regulation of expression of the ADF/cofilin isoforms. It would be interesting to examine if misexpression of the ADF/cofilin isoforms causes a dominant phenotype or if forced expression of UNC-60B in early embryos or UNC-60A in muscle suppresses the RNAi or mutant phenotypes. The lethal phenotype of the yeast cofilin-null mutant can be rescued by expression of vertebrate ADF/cofilins (Iida et al., 1993) or Dictyostelium cofilin-1, but not cofilin-2 (Aizawa et al., 2001). Thus, further functional characterization of the ADF/coflin isoforms is important to understand how evolution of actin-regulatory proteins has contributed to increasing the diversity in actin cytoskeletal structure and function.
We thank S. Langley for technical help in DNA sequencing, F. Piano for communication of preliminary results on RNAi, R. Aroian for preliminary characterization of the unc-60A (RNAi) phenotype and comments on the results, E. Kipreos for a northern blot, D. Baillie for the worm strain, H. Epstein for the worm strain and antibodies, M. Hresko for antibodies, and S. Yeoh and A. Weeds for critical comments on the manuscript. Some strains were provided by the Caenorhabditis Genetics Center, which is funded by the National Institute of Health National Center for Research Resources. This work was supported by grants from the National Science Foundation