Rho-associated kinase (ROCK) is a serine/threonine kinase working in the Rho signaling to actin cytoskeleton. We previously reported that loss of ROCK-I results in the eyelid open at birth (EOB) and omphalocele phenotype in mice, while loss of ROCK-II results in placental dysfunction leading to intrauterine growth retardation and fetal death. Here, we report that after backcross to the C57BL/6 N genetic background, ROCK-II knockout (KO) neonates are born also with open eyelid and umbilical hernia, a phenotype similar to that of ROCK-I KO mice. ROCK-II KO embryos show impaired extension of the eyelid epithelial sheet with disorganized actin bundles in the leading edge of the sheet. These results suggest that ROCK-I and ROCK-II cooperatively regulates the assembly of actin bundles essential for closure of the eyelid and ventral body wall in mouse embryos. Consistently, ROCK-I+/–ROCK-II+/– double heterozygous mice also show the EOB and omphalocele phenotype.
Actin cytoskeleton is a highly dynamic and versatile structure providing mechanical force and support for cell movement. The small GTPase Rho regulates several steps of actin dynamics and contributes to a variety of cellular processes including cell migration and cytokinesis (Etienne-Manneville & Hall 2002). Rho acts on various downstream effector molecules and exerts such functions. ROCK, a serine/threonine kinase, is one of the major Rho effectors involved, for example, in regulation of myosin light chain (MLC) phosphorylation, actin severing by cofilin and activation of type I sodium proton exchanger (Tominaga et al. 1998; Maekawa et al. 1999; Amano et al. 2000). Through these actions, ROCK plays important roles in the control of cell morphology, behavior, growth and differentiation (Riento & Ridley 2003). There are two ROCK isoforms in mammals, ROCK-I and ROCK-II. The N-terminal serine/threonine kinase domain is highly conserved with more than 90% homology between the two isoforms, but the carboxyl terminal is substantially different (Nakagawa et al. 1996). Consistently, over-expression of C-terminal-truncated mutants of the two ROCK isoforms produces similar phenotype in cultured cells. However, whether each ROCK isoform exerts identical or distinct functions in the cell still remains unknown. Functions in vivo of each isoform are more obscure, because they show partially overlapping distribution in the body. We previously reported generation of mice deficient in each ROCK isoform. Our analyses demonstrated that loss of ROCK-I results in the eyelid open at birth (EOB) and omphalocele phenotype (Shimizu et al. 2005), while loss of ROCK-II results in placental dysfunction, intrauterine growth retardation and fetal death (Thumkeo et al. 2003). These results suggest that although structurally similar, ROCK-I and ROCK-II cannot compensate for loss of each other in some systems during embryonic development.
Eyelid closure in mice starts at 15.5 days postcoitum (dpc), when the outermost epithelial layer of the eyelid extends as a sheet toward the center of the eye, and is complete at 16.5 dpc, when the opposite epithelial sheets fuse in the center to cover the entire ocular surface (Kaufman 2002). It is suggested that eyelid closure in mammals and dorsal closure in fly are highly homologous developmental processes (Xia & Karin 2004). In the Drosophila dorsal closure, actin cables assemble in the leading edge cells of the epithelium and provide the driving force for epithelial sheet movement (Kiehart et al. 2000; Hutson et al. 2003). It was also reported that the Rho signaling and the JNK/c-Jun pathway are implicated in this process (Jacinto et al. 2002; Mizuno et al. 2002). Studies in the past few years revealed the role of JNK/c-Jun pathway also in the eyelid closure in mammals (Li et al. 2003; Zenz et al. 2003; Zhang et al. 2003; Weston et al. 2004). However, the role of Rho signaling during this process was not shown until our recent analysis of ROCK-I knockout (KO) mice. We found that ROCK-I−/– mice have impaired eyelid closure due to disorganization of actin cables in the eyelid epithelial sheet, and thus verified a role of Rho signaling evolutionarily conserved in fly and mice (Shimizu et al. 2005). Furthermore, we also found that ROCK-I−/– mice are impaired in another type of tissue closure, i.e. the ventral body wall closure. During midgestation, visceral organs rapidly expand in volume and exceed the space of the peritoneal cavity, resulting in protrusion of the midgut loop through the umbilical ring, a phenomenon called physiological umbilical herniation. At 15.5 dpc the loop begins to return to the peritoneal cavity and physiological herniation ceases at 16.5 dpc when the ventral body wall closes completely (Kaufman 2002). Failure of this closure leads to an omphalocele, in which the midgut loop does not return to the peritoneal cavity. Unlike eyelid closure, there are only a few mouse mutants showing the omphalocele phenotype, and the mechanism underlying the ventral body wall closure is still poorly understood (Eggenschwiler et al. 1997; Brewer & Williams 2004). Our previous results indicate that ROCK-I-mediated actomyosin contraction in the epithelium works critically also in closure of the ventral body wall (Shimizu et al. 2005).
In this work, we show that ROCK-II KO mice backcrossed in the C57BL/6 N genetic background exhibit not only the placental phenotype as previously reported but also show both EOB and omphalocele. Disorganization of actin cables similar to that found in ROCK-I KO mice is also found in the eyelid epithelium of ROCK-II-deficient mice, indicating that either of the two isoforms alone is not sufficient for the tissue closure. Consistently, mice heterozygous for both ROCK-I and ROCK-II showed also EOB and omphalocele. These results suggest that ROCK-I and ROCK-II cooperatively regulate actin bundle assembly required for eyelid and ventral body wall closure.
ROCK-II KO neonates display the EOB and omphalocele phenotype
Mice deficient in ROCK-II were backcrossed for more than eight generations to the C57BL/6 N genetic background and subjected to heterozygous mating. Analysis of the genotype distribution in offspring revealed that the number of homozygous ROCK-II−/– mice (1/428, Table 1) were not only significantly underrepresented from the value predicted by Mendelian inheritance but also significantly less than that we found previously in the same genotype of mice in the mixed genetic background between 129/SvJ and C57BL/6 N (6/122). As those in the mixed genetic background (Thumkeo et al. 2003), ROCK-II KO mice in the C57BL/6 N genetic background exhibited the thrombus formation in labyrinth layer (data not shown) and were born runts (Fig. 1A). They also showed a sign of thrombus in their hind limb. In addition to these phenotypes found in those in the mixed genetic background, ROCK-II−/– neonates in the C57BL/6 N genetic background exhibited the eyelid open at birth (EOB) and omphalocele phenotype (Fig. 1A). The severity of EOB (Fig. 1B) and that of omphalocele vary among the ROCK-II neonates, and are not correlated to each other (Fig. 1A). Analysis of 18.5 dpc embryos in utero revealed that the number of viable KO embryos were comparable to that found in the previous study (Thumkeo et al. 2003), indicating that the presence of an omphalocele itself did not affect the survival (Table 1). This together with the finding that a part of abdominal organs, notably the liver and intestine, was gnawed out in dead ROCK-II KO neonates (Fig. 1C) indicates that the omphalocele is cannibalized after birth by the mother in the process of removing placenta to cause neonatal death. Heterozygous mice never display these phenotypes and are comparable to their wild-type littermates in growth and appearance.
Table 1. Genotypes of offspring obtained by cross-mating of ROCK-II+/− heterozygous mice
Total no. of offspring
No. of mice with genotype
Male;(+/−) × female;(+/−)
ROCK-II is required for embryonic eyelid closure
To examine details of the EOB phenotype in ROCK-II KO embryos, we performed histological analysis on the extending eyelid of wild-type and ROCK-II−/– embryos. The eyelid of ROCK-II−/– embryos showed morphology similar to that of wild-type littermates at 15.5 dpc but extended poorly at 16.5 dpc when the eyelid already fused in wild-type embryos (Fig. 2A). Analysis by scanning electron microscopy (SEM) on the 16.5 dpc eyelid clearly demonstrates the delay in eyelid extension in ROCK-II−/– mice (Fig. 2B). Wild-type embryos at this stage had eyelids united with a midline seam but none of the KO embryos examined (9/9) exhibited such fusion. These results suggest that the extension of the eyelid epithelial sheet is significantly impaired and delayed in ROCK-II KO mice, causing the EOB phenotype.
ROCK-II mediates assembly of actin bundles in extending eyelid epithelium
We previously reported the existence of actin bundles in the leading edge of the extending eyelid epithelium, and proposed it provides the driving force necessary for eyelid closure (Shimizu et al. 2005). This actin structure was impaired by loss of ROCK-I. We therefore speculated that loss of ROCK-II might also affect formation of this actin structure. To examine this possibility, we dissected the eyelid at late 15.5 dpc and subjected it to whole-mount phalloidin staining that specifically detects filamentous actin. As previously reported (Shimizu et al. 2005), strong staining was observed in the outermost part of eyelid epithelial sheet surrounding the eye of wild-type embryo (data not shown). When this structure was examined in detail with higher resolution and magnification by confocal microscopy, we found that actin bundles extend beyond cells and formed several cables in a well-organized manner in wild-type embryo (Fig. 3A, left panel). On the other hand, although accumulation of actin filaments was also apparent in the edge of eyelid epithelial sheet surrounding the eye in ROCK-II−/– embryo, they were fewer in number, not continuous and not organized into cables (Fig. 3A, right panel). Consistently, while phalloidin staining of transverse frozen sections revealed strong signals of bundled filamentous actin in epithelial cells of the eyelid sheet of wild-type 15.5 dpc embryo, only weak signals were observed in the ROCK-II−/– embryo (Fig. 3B).
Expression of ROCK-II in epithelial cells that constitute extending eyelid
We next examined expression of ROCK-II in the extending eyelid of mouse embryos by using a novel polyclonal antibody to ROCK-II we developed in this study. The specificity of the antibody was examined first by Western blot analysis using brain lysates from wild-type and ROCK-II−/– mice (Fig. 4A). The antibody detected a single 160-kDa band corresponding to ROCK-II in wild-type brain lysates but not in the ROCK-II−/– lysates. Consistently, immunostaining with this antibody revealed strong ROCK-II signal throughout the cerebral cortex of wild-type but not ROCK-II−/– mice (Fig. 4B). Furthermore, this antibody also detected ROCK-II expression in the labyrinth layer of the placenta of wild-type but not ROCK-II−/– mice (Fig. 4C). The ROCK-II expression in the latter tissue was previously determined by whole-mount staining for β-galactosidase inserted in the targeting vector (Thumkeo et al. 2003). Signals detected by this antibody are not affected in ROCK-I−/– mice (data not shown), further verifying that the antibody does not crossreact with ROCK-I. We next used this anti-ROCK-II antibody and performed immunofluorescence in the embryonic eyelid. ROCK-II was uniformly expressed throughout the epithelium of the 15.5 dpc embryonic eyelid (Fig. 5A). The ROCK-II signals were also detected in the skin (Fig. 5B) and in the cytoplasm of primary keratinocytes in culture (Fig. 5C). Thus, the ROCK-II signal was not limited to the cells containing the actin cables seen in Fig. 3A but rather was ubiquitously present throughout the eyelid epithelial sheet. These results suggest that local activation of ROCK-II rather than its expression regulates the assembly of actin cables during eyelid closure. In fact, ROCK-I which cooperatively regulate eyelid closure with ROCK-II also localized ubiquitously in the eyelid epithelial sheet in similar manner to that of ROCK-II (Shimizu et al. 2005).
EOB and omphalocele phenotype in ROCK-I/ROCK-II double heterozygous mice
The above results together with our previous findings have thus demonstrated that both ROCK isoforms are present in the epithelium of the extending eyelid and ventral body wall of mouse embryo, and that the presence of either isoform alone cannot support their closure. Because one ROCK isoform does not increase in response to loss of the other (Thumkeo et al. 2003; Shimizu et al. 2005; this study), the above results raise two possibilities. One is that the two isofoms work at different but interdependent steps of actin bundle assembly, and the other is that they work at the same step(s) and the amount of either isoform alone is not sufficient for the closure. To examine these two possibilities, we generated ROCK-I+/–ROCK-II+/– double heterozygous mice by mating heterozygous ROCK-I+/– mice with ROCK-II+/– mice. Resultant double heterozygous ROCK-I+/–ROCK-II+/– mice were then mated with wild-type mice. Analysis of the genotype distribution in offspring from this mating revealed that the number of double heterozygous ROCK mice under the C57BL/6 N genetic background were significantly underrepresented from the value predicted by Mendelian inheritance (Table 2). Notably, a significant number of the newborn double heterozygotes were born with the EOB and omphalocele phenotype (Fig. 6A,B). However, the phenotype of the double heterozygous neonates was milder than ROCK-I−/– or ROCK-II−/– mice, as roughly 30% of double heterozygous neonates survive, while about 10% of ROCK-I−/– or less than 1% of ROCK-II−/– mice can survive (Tables 1, 2; Shimizu et al. 2005). Moreover, some of the double heterozygous neonates did not display any apparent phenotype at all. Surviving double heterozygous neonates subsequently developed normally and the adult mutants were fertile and apparently healthy (data not shown).
Table 2. Genotypes of offspring obtained by mating of double heterozygous ROCK mutant mice with wild-type mice
No. of mice with genotype
Male;(R1+/−)(R2+/−) × female;(R1+/+)(R2+/+)
In the present study, we backcrossed ROCK-II KO mice in the C57BL/6 N genetic background and found unexpectedly that the neonates generated exhibited EOB and omphalocele in addition to the placental dysfunction phenotype. As found in ROCK-I KO mice, the omphalocele itself does not affect the survival of embryos but strongly affects the survival of neonates due to cannibalization by the mother and the resultant intraperitoneal bleeding. It is therefore assumed that only neonates with mild omphalocele and placental phenotype could survive into adulthood. Indeed, the survival rate of ROCK-II KO mice, with combined omphalocele and placental defect, drastically decreased to less than 1% in comparison with roughly 10% in ROCK-I−/– mice that do not have placental dysfunction. The reason that genetic background influences expression of the EOB-omphalocele phenotype in ROCK-II KO mice is unknown. Previously, EGFR KO mice that show the EOB phenotype were also reported to exhibit different phenotypes in different genetic backgrounds; for example, they die at mid-gestation in a 129/Sv background, and some survive in either a 129/Sv × C57BL/6 genetic background or a CD-1 background until birth and show open eye (Sibilia & Wagner 1995; Threadgill et al. 1995). The authors proposed that EGFR activation could be modified differently in different genetic backgrounds. The mechanism by which the genetic background affects ROCK-II KO mice phenotype will be examined in future studies.
Analyzing the EOB phenotype in ROCK-II embryos, we have found that the final outcome of the loss of ROCK-II is quite similar to that in ROCK-I KO mice. As demonstrated by histology and SEM, we found that extension of eyelid epithelial sheet of ROCK-II KO mice was impaired with full penetrance at 16.5 dpc. Phalliodin staining revealed the formation of actin cable in the leading cells of eyelid epithelial sheet of wild-type mice while the structure of these cables were severely disrupted in ROCK-II KO embryos. We also used a novel antibody to ROCK-II and detected the ROCK-II signal ubiquitously in the eyelid epithelial sheet. All of these findings are similar, if not identical, to those of ROCK-I KO mice we previously reported (Shimizu et al. 2005). It is therefore likely that ROCK-I and ROCK-II work in the same pathway required for actin cable assembly that is necessary for force generation for the extension of eyelid epithelial sheet in the eyelid closure. To determine whether the two isoforms work at the same steps in a redundant manner, we have generated ROCK-I/-II double heterozygous mice under C57BL/6 N genetic background and examined their phenotype. While the ROCK-I+/–ROCK-II+/– neonates also display EOB and omphalocele, their phenotype was milder than ROCK-I KO neonates, resulting in survival of about 30% of the double heterozygotes compared to about 10% of ROCK-I KO mice. These results indicate that, although ROCK-I and ROCK-II share the highly conserved kinase domain (Nakagawa et al. 1996), they may not work exactly in the same manner in eyelid epithelial cells during the developmental process of tissue closure. Whether this difference comes from different or preferential use of different substrates by either isoform remains to be examined.
Eyelid closure in mammals and dorsal closure in fly are highly homologous developmental processes (Xia & Karin 2004). Both processes are driven by actin cable assembly in the leading edge cells of the epithelium, and requires not only the Rho signaling but also the JNK/c-Jun pathway (Jacinto et al. 2002; Mizuno et al. 2002; this study). However, how the Rho signaling and the JNK/c-Jun pathway are linked is not known. One link may be indirect. JNK1+/–/JNK2−/– mice (Weston et al. 2004) as well as mice with keratinocyte-specific knockout of c-Jun (Li et al. 2003; Zenz et al. 2003) that are defective in eyelid closure showed reduced expression of the EGFR ligands EGF or heparin-binding EGF (HB-EGF), indicating the presence of a positive feedback loop. This together with our previous finding that Rho is activated downstream of EGF in primary cultured keratinocytes (Shimizu et al. 2005) indicate that Rho activation is indirectly regulated by JNK in a positive feedback loop of JNK-EGF. On the other hand, given accumulating evidence that JNK can directly regulate cytoskeleton assembly (Otto et al. 2000; Chang et al. 2003; Huang et al. 2003; Kawauchi et al. 2005) suggest that the Rho signaling and the JNK pathway may interact also at the level of actomyosin assembly. This issue should be examined by identifying the cytoskeleton structure in the eyelid epithelial cells during embryonic eyelid closure of JNK1/2 mutant mice.
In conclusion, we have shown the importance of ROCK-II during eyelid closure and ventral body wall closure. This study combined with our previous study on ROCK-I KO and findings of ROCK-I/-II double heterozygous mice strongly suggest that ROCK-I and ROCK-II cooperatively regulate eyelid closure. The lack of phenotype in other tissues of ROCK-I−/–, ROCK-II−/– and ROCK-I/-II double heterozygous mice suggests that one ROCK isoform is able to compensate functionally for the loss of the other for development of most tissues and organs. The generation of double knockout mice and conditional KO mice will provide further insight into specific roles and functions of ROCK isoforms.
ROCK-II KO mice of a mixed 129/Sv and C57BL/6 N genetic background (Thumkeo et al. 2003) were backcrossed to the C57BL/6 N strain for more than 8 generations. Double heterozygous mice were generated by cross-mating of ROCK-II mutant mice with ROCK-I mutant mice (Shimizu et al. 2005) under C57BL/6 N genetic background. We designated the morning of the day on which a vaginal plug was detected as 0.5 dpc. All animal experiments were approved by the Committee on Animal Research of Kyoto University Faculty of Medicine and were performed according to the guidelines for the protection of experimental animals of Kyoto University.
Production of antibodies to mouse ROCK-II and Western blot analysis
A GST fusion protein of fragment R2A of mouse ROCK-II (amino acids 1120–1388) was produced in Escherichia coli BL21 cells, and purified with the use of glutathione-Sepharose 4B beads (Amersham Biosciences). The GST-fused ROCK-II-R2A was used as the antigen and injected into rabbits. Specific antibodies to mouse ROCK-II were then affinity-purified from serum with the use of ROCK-II-R2A immobilized on CNBr beads (Amersham Biosciences). The whole brain of adult mice were homogenized, and used for Western blot analysis as described (Thumkeo et al. 2003) with antibodies to either ROCK-I (Ishizaki et al. 1996), ROCK-II or β-tubulin (TUB 2.1, Sigma).
Histology and immunofluorescence
Embryos were fixed in 10% formaldehyde, dehydrated with a graded series of ethanol solutions, and embedded in paraffin. Sections of 8-µm thickness were prepared, subjected to removal of paraffin by immersion in xylene, rehydrated, and stained with hematoxylin-eosin. Alternatively, the heads of the embryos were embedded in Tissue Tek OCT compound (Sakura, Tokyo, Japan) and frozen on dry ice. Cryostat sections (16 µm thickness) were prepared and subjected to indirect immunofluorescence with an antibody to ROCK-II (this study). Immune complexes were detected with Alexa Fluor 488-conjugated donkey antibodies to rabbit IgG (Molecular Probes). Phalloidin labeled with Oregon green (Molecular Probes) were used for staining of filamentous actin. Primary keratinocytes were prepared as previously described (Sano et al. 1999), fixed with 4% paraformaldehyde for 15 min at RT, permeabilized for 5 min with PBS containing 0.1% Triton X-100, and then stained for ROCK-II. Specimens were examined with the LSM 510 confocal microscope, at RT, with either 40X (Plan-Neofluor; NA 1.3) or 63X (Plan-Apochromat; NA 1.4) objective lens. Images were acquired with LSM 510 AIM software (Carl Zeiss). All pseudo-color representations were assembled using Adobe Photoshop version 6.0 for illustrative purpose only.
Whole-mount phalloidin staining
Eyelids isolated together with the skin around the eye from embryos at 15.5 dpc were processed for whole-mount phalloidin staining as previously described (Shimizu et al. 2005). Specimens were examined under the LSM 510 confocal imaging system (Carl Zeiss).
Scanning electron microscopy
The head of embryos at 16.5 dpc was fixed in 0.1 m sodium phosphate buffer (pH 7.4) containing 1% glutaraldehyde and 4% formaldehyde, washed with 0.1 m sodium phosphate buffer containing 0.1 m sucrose, and dehydrated with a graded series of ethanol solutions. Specimens were then dried and sputter-coated according to standard procedures before examination with an S4700 scanning electron microscope (Hitachi).
This work was supported in part by a grant-in-aid for Specially Promoted Research from the Ministry of Education, Culture, Sports, Science and Technology of Japan and a grant from Mitsubishi Pharma Corporation. We thank M. Fujioka for advice on scanning electron microscopy; Y. Andou for help with preparation of the antibodies to ROCK-II; T. Fujiwara, K. Nonomura and K. Hamajima for technical assistance; and T. Arai and Y. Kitagawa for secretarial assistance.