Nonmuscle myosin II isoform and domain specificity during early mouse development

An Essential but Isoform-Independent Role for NM II in Visceral Endoderm Formation.

The generation of four mutant knock-in mouse lines is described in SI Materials and Methods and shown in Fig. S1. All heterozygotes from the different lines are indistinguishable from their wild-type littermates. They were crossed to produce homozygous embryos. Control A^mCh/A^mCh mice are born at the expected Mendelian frequency and are normal, demonstrating that the phenotypes observed in the mutant mice are not a result of genetic manipulations of the Myh9 locus (Table S1). The ratio of the GFP-NM II-B to the endogenous II-B in MEF cells is 2.8:1, indicative that GFP-NM II-B is expressed under control of the NMHC II-A promoter (Fig. S1D, right-most lane). Fig. S1E indicates that expression of the chimeric NM IIs are similar to GFP-NM II-B in A^b*/A^b* mice.

We first determined whether knock-in NM II-B or chimeric NM IIs could functionally replace NM II-A, and rescue the cell-cell adhesion defects of the visceral endoderm associated with II-A deficiency at E6.5. A^b*/A^b*, A^ab/A^ab, A^ba/A^ba (collectively referred to as “mutant”) and A⁺/A⁺ embryo sections were stained with antibodies to NMHC II-A or II-B together with E-cadherin (Fig. 1 and Fig. S2). In contrast to A⁻/A⁻ embryos, which have unidentifiable cell layers and a disorganized visceral endoderm marked by GATA4 staining of the nuclei (compare Fig. 1 J with I), all mutant embryos (Fig. 1 E and G; Fig. S2 E, G, I, K) appear normal, with a polarized columnar visceral endoderm similar to A⁺/A⁺ embryos (Fig. 1 A and C and Fig. S2 A and C). Moreover, the cell-cell borders of the visceral endoderm of A⁺/A⁺ and mutant embryos contain E-cadherin (arrows in Fig. 1I and Fig. S2), which is not present in the A⁻/A⁻ GATA4-positive cells (arrows, Fig. 1J). As noted previously (7), wild-type visceral endoderm expresses NM II-A at the cell-cell boundaries but has no detectable II-B (Fig. 1 B and D, arrows), whereas in mutant visceral endoderm, NM II-B or chimeric NM IIs are expressed in place of NM II-A (Fig. 1 F and H and Fig. S2 F, H, J, L, arrows). The presence of NM II-B or the chimeric NM IIs restores normal cell-cell adhesion to the visceral endoderm. Furthermore, unlike A⁻/A⁻ mice, all of the mutant mice undergo gastrulation, confirming that the early lethality of A⁻/A⁻ mice is the result of a failure in formation of a functional visceral endoderm. This finding is consistent with a role for NM IIs in cell-cell adhesion, a process that requires the cross-linking properties of NM II, therefore allowing substitution between the different NM II isoforms because of their similar structural properties. These results also confirm that the E6.5 lethality of A⁻/A⁻ embryos is caused by the absence of both NM II-B and II-C from the visceral endoderm, rendering it particularly vulnerable to NM II-A ablation.

Substitution for NM II-A by II-B or Chimeric NM IIs is Lethal.

No mutant homozygous mice were found at weaning, indicating that the development of the mutant mice is arrested at earlier stages (Table S1). A^b*/A^b* and A^ba/A^ba embryos die between E9.5 and E10.5, whereas A^ab/A^ab embryos die between E11.5 and E12.5. Despite the differences in life span, the mutant embryos exhibit a similar phenotype based on their appearance, as exemplified by A^b*/A^b* embryos. These embryos display pale yolk sacs with fewer visible vessels compared with the A⁺/A⁺ yolk sacs (Fig. 2A). Whole-mount PECAM-1 staining also reveals that A^b*/A^b* embryos have a less intricate vascular network in the head and trunk regions compared with the well-developed and hierarchically organized vascular architecture of A⁺/A⁺ counterparts (Fig. 2B). These findings suggest that although vasculogenesis occurs in the A^b*/A^b* yolk sacs, there is a defect in subsequent angiogenesis, which results in the impairment of the blood supply to the embryo. In addition, A^b*/A^b* embryos show growth retardation, which becomes more pronounced with age (Fig. S3). However, this defect in growth is not a result of abnormalities in cell proliferation or apoptosis, as shown in Fig. S4A, which shows no difference in the BrdU and TUNEL staining between A^b*/A^b* and wild-type embryos at E9.5.

Specific Requirement for NM II-A in Placenta Development.

The defects found in the yolk sac suggested that similar abnormalities in vascular development might be observed in the placenta, which could be the cause of lethality. Mouse placental development involves at least three steps: formation of the allantois and the chorion, chorioallantoic fusion, and vascularization (17). We therefore examined the placenta of mutant mice between E8.5 and E10.5. Similar to A⁺/A⁺ mice, all of the mutant embryos form proper allantoic structures (Fig. S3A, arrows); the fusion of the allantois to the chorionic plate around E8.5 appears to be normal too. However, subsequent vascularization of these mutant placentas was abnormal. At E10.5, A⁺/A⁺ placentas acquire the typical trilaminar structure composed of a distal circumferential giant cell layer, a middle spongiotrophoblast layer, and a proximal labyrinthine layer with a network of fetal capillaries and maternal blood sinuses (Fig. 3A, enlarged in C). In contrast, embryonic blood vessels of A^b*/A^b* or A^ba/A^ba mice fail to invade the labyrinthine layer and remain restricted to the chorioallantoic region. Consistent with this, A^b*/A^b* and A^ba/A^ba placentas are thinner and more compact, composed predominantly of clusters of cuboidal trophoblasts, making the labyrinthine and spongiotrophoblast layers difficult to discern (Fig. 3 B and D and Fig. S5 C and F). Interestingly, invasion of embryonic blood vessels was observed in A^ab/A^ab placentas, but this process was compromised, as shown by the reduced thickness of the placenta and decreased internal space of fetal and maternal vessels (Fig. S5 B and E). This finding is consistent with the increased survival of A^ab/A^ab mice compared with A^b*/A^b* or A^ba/A^ba mice, indicating that the presence of the N-terminal domain of NM II-A supports increased placental maturity. Staining with CD34, a marker for endothelial cells, shows the blood vessels in the labyrinthine layer of the A⁺/A⁺ placenta (Fig. 3E). However, very few CD34-positive cells are found in the labyrinthine layer of E10.5 mutant mice (Fig. 3F and Fig. S5 H and I, compare with G). Although formation of the labyrinthine layer fails and the trophoblast compartment appears thinner in mutant placentas, no abnormality in cell proliferation and apoptosis is found in the mutant extraembryonic tissues (Fig. S4B).

To further support the idea that defects in the mutant placentas are a result of the specific loss of NM II-A, which is not rescued by expression of II-B (or chimeric NM IIs) from the Myh9 locus, we analyzed the expression of NM IIs in the placenta by immunofluorescence staining, which reveals an enriched and uniform expression of NM II-A at the fetal side of the A⁺/A⁺ E9.5 placenta (Fig. S6A), consistent with the importance of NM II-A in placenta development. Despite tracking the NM II-A tissue expression pattern in the mutant placenta (Fig. S6 A and D), NM II-B or chimeric NM IIs cannot adequately substitute for II-A functions. Importantly, the placentas of E11.5 NM II-B-ablated (B⁻/B⁻) embryos do not show any obvious defects (Fig. S5 K and M), providing further evidence of the specific requirement for NM II-A. Collectively, these results suggest that NM II-A plays a unique role in placenta development, a role which is different from its functions in cell-cell adhesion in the visceral endoderm and, as shown by the failure of the chimeric NM IIs to rescue this defect, requires the presence of both functional domains: a specific motor domain and a specific C-terminal domain.

Mutant Mouse Embryonic Fibroblasts Exhibit Cell Migration Defects.

The failure of the mutant blood vessels to invade the labyrinthine layer implies that the impaired vascularization might result from a defect in cell migration. We therefore examined the migration of mouse embryonic fibroblast (MEF) cells derived from wild-type and mutant embryos. In a wound-healing assay to evaluate the speed of cell migration, A^b*/A^b* and A^ab/A^ab MEFs closed the wound more rapidly than A⁺/A⁺ MEFs (A^b*/A^b*, 33.9 ± 2.8 μm/h; A^ab/A^ab, 36.7 ± 3.9 μm/h; A⁺/A⁺ 24.2 ± 2.6 μm/h) (Fig. S7A). Interestingly, A^ba/A^ba cells migrate at 20.5 ± 4.8 μm/h, a speed similar to A⁺/A⁺ cells. Previous studies reported that the absence of NM II-A increases the speed of migration (3, 18). Our results reveal that the C-terminal rod domain but not the N-terminal motor domain of II-A has an important effect on cell migratory speed, because its absence in A^b*/A^b* and A^ab/A^ab cells correlates with the increased speed of wound closure where the cells migrate as a sheet. On the other hand, using a transwell assay to examine directional migration of single cells toward serum, the number of MEFs from all three mutant lines migrating across the transwell pores after 16 h was significantly less than that of the A⁺/A⁺ MEFs (Fig. S7B). There is no significant difference among the three mutant MEFs, suggesting that both domains contribute to single-cell migration through the pores of the chamber. The directionality of individual cells was further evaluated by time-lapse microscopy in 2D cell-culture dishes and persistence expressed as the ratio of net to total migrated distance (N/T). The migration speed calculated from individual cells (total distance/time) agrees with that derived from the wound-healing assay. Interestingly, whether speed of migration increases in A^b*/A^b* and A^ab/A^ab cells or is unchanged, as in A^ba/A^ba cells, their migratory persistence decreases significantly (Fig. S7C), suggesting that these mutant cells have a defect in directed cell migration. Thus, our results indicate that either the absence of the motor domain of NM II-A in A^ba/A^ba cells or the lack of the NM II-A rod domain in A^ab/A^ab cells causes abnormalities in cell migration, demonstrating that the two functional domains collectively determine the specific function of NM II.

Mutant MEFs Display Reduced Focal Adhesions and Abnormal Stress Fibers.

To investigate the mechanism underlying the abnormalities of cell migration in the mutant cells, we examined actin cytoskeletal structure and focal adhesion formation, which are essential for cell migration. The distribution of actin stress fibers and focal-adhesion contacts was visualized by staining with phalloidin and antibodies to focal-adhesion proteins. Typical, robust and parallel stress fibers are observed in the A⁺/A⁺ cells (Fig. 4A), although mutant cells show thin actin filaments that are disorganized and often fail to align with the direction of migration (Fig. 4 D, G, and J). Interestingly, the actin cytoskeletal structure of A^ba/A^ba cells appears more like the wild-type cells. The staining for paxillin, a component of the focal-adhesion contacts, which anchors actin stress fibers, reveals abundant and well-spread focal adhesions throughout the A⁺/A⁺ cells (Fig. 4B). In contrast, fewer and smaller focal adhesions are detected in the mutant cells (Fig. 4 E, H, and K), especially in the A^b*/A^b* and A^ab/A^ab cells. This decrease in both number and size of focal adhesions is quantified in Table S2. The merge images indicate that the stress fibers are anchored at both ends by focal adhesions in A⁺/A⁺ cells (Fig. 4C), whereas in the mutant cells, these structures are in disarray and one or both ends of actin filaments are missing large focal-adhesion contacts, particularly in the A^b*/A^b* and A^ab/A^ab cells (Fig. 4 F, I, and L). In addition to reduced paxillin staining in focal adhesions, A^b*/A^b* cells exhibit almost no visible phospho-Tyr118-paxillin immunofluorescence staining compared with A⁺/A⁺ cells (Fig. 5A). Immunoblots demonstrate that the total amount of focal-adhesion proteins, including paxillin and vinculin, are unchanged, whereas the level of the phospho-Tyr118-paxillin, indicative of paxillin activation and focal adhesion formation is undetectable in the A^b*/A^b* cells (Fig.5B). Of note, when a plasmid encoding wild-type mCh-NMHC II-A is transfected into A^b*/A^b* cells, the focal adhesions are restored, confirming that the reduction in focal adhesions is associated with the loss of NM II-A (Fig. 5C). The above finding is consistent with a specific role for NM II-A in cell migration, as reflected by the requirement for II-A for actin cytoskeletal organization and focal adhesion formation. During these processes, both the N-terminal and C-terminal domains of NM II-A are required. Neither the N-terminal motor domain of NM II-A in the A^ab/A^ab cells or the C-terminal rod domain of II-A in the A^ba/A^ba cells can rescue the defects caused by the loss of intact II-A.

Subcellular Localization of Knock-in NM II in Mutant MEF Cells.

To address the question of whether a decrease of focal adhesions in the mutant MEF cells is related to the alterations in the cellular distribution of NM IIs, we studied the in vitro subcellular localization of knock-in GFP-NM II-B or chimeric NM IIs in A⁺/A^b*, A⁺/A^ab, and A⁺/A^ba heterozygous cells in which both the endogenous NMHC II-A and knock-in NMHC IIs are under control of the NMHC II-A promoter. GFP-NM II-BA like NM II-A, and GFP-NM II-AB like NM II-B, display distinct subcellular localizations: GFP-NM II-BA colocalizes to regions of the MEFs with endogenous II-A (Fig. 6, compare O to R), whereas GFP-NM II-B or GFP-NM II-AB colocalize to regions of the MEFs with endogenous II-B (Fig. 6 compare F and L to C and I). These results agree with previous reports (19–21) and confirm, using an endogenous promoter, the observation that subcellular localization is determined by the C-terminal domain of NMHC II. These results also indicate the important effects of the C-terminal domain of NM II on focal adhesion formation and help explain our observation of an increase in focal adhesions in A^ba/A^ba cells compared with the other mutant cell types.

This study supports the idea that, when the cross-linking function of NM II is required, the isoforms are interchangeable in vivo, such as during the process of cell-cell adhesion when the actomyosin complex interacts with cell-cell adhesion proteins. However, during cell migration and focal adhesion formation, dynamic processes, in which the actomyosin complex interacts with cell matrix adhesion proteins and which require not only motor activity but also correct subcellular localization, substitution in vivo is unsuccessful (Table S3). Differences in these dynamic properties may reflect differences in the kinetic properties between NM II-A and II-B. These include significant differences in the rate of ATP hydrolysis when myosin is bound to actin, which is approximately 3-fold greater in the case of NM II-A. In contrast, NM II-B has a higher duty ratio and affinity for ADP, which makes it particularly well suited to exert tension on actin filaments for longer periods of time (22–24). Importantly, our results reveal that the essential role of NM II in visceral endoderm formation and function is isoform-independent, whereas placenta development depends on the specific properties of NM II-A.

Targeting Constructs.

All mouse procedures were carried out in accordance with National Heart, Lung, and Blood Institute Animal Care and Use Committee guidelines. To create the knock-in/knockout constructs for homologous recombination, DNA fragments flanking exon2, the first coding exon of the mouse Myh9 gene, were amplified from a 129/Sv genomic BAC clone harboring the complete Myh9 locus (25). The arms consisted of a 4-kb fragment 5′ of the initiating ATG and a 1.7-kb fragment 3′ of the ATG codon. The targeting constructs are depicted in Fig. S1A and consist of the 5′ arm, a cDNA cassette encoding mCherry (mCh)-human NMHC II-A or GFP-human NMHC II-B or chimeric GFP-human NMHC II-AB or GFP-human NMHC II-BA, followed by SV40 polyA, a Neo^r cassette flanked by two loxP sites, the 3′ arm, and the thymidine kinase cassette. Chimeric GFP- NMHC II-AB including amino acids 1 to 836 of II-A and 844 to 1,977 of II-B, and II-BA containing amino acids 1 to 843 of II-B and 837 to 1,961 of II-A were generated by PCR with the templates GFP-NMHC II-A and GFP-NMHC II-B, described previously (26). Nucleotide sequences of the cloned DNA fragments were confirmed in all cases by sequencing.

Histology and Immunofluorescence Staining.

Embryos and placentas were dissected, fixed in 4% PFA overnight at 4 °C, dehydrated by methanol series, and stored in 100% methanol at −20 °C. Embryos and placentas were embedded in paraffin, sectioned at 5 μm on silanized slides, and stained with H&E (Histoserv). Alternatively, for immunofluorescence staining, antigen retrieval was performed as described before (27). These sections were then blocked with 10% normal goat serum and probed overnight at 4 °C with primary antibodies: NMHC II-A, II-B (28), E-cadherin (Sigma, 1:100 or BD-Biosciences, 1:250), GATA4 (Santa Cruz, 1:50), CD34 (BD-Biosciences, 1:200), BrdU (Sigma, 1:500) or p-Histone-H3 (Santa Cruz, 1:200). Where appropriate, an N-terminal NMHC II-A or II-B antibody was used and is noted in the figure legends. Fluorescently labeled secondary antibodies (Molecular Probes) were used and the slides were mounted with coverslips using Prolong Gold antifade reagent (Molecular Probes).