|Home | About | Journals | Submit | Contact Us | Français|
Mammalian sterile 20-like kinases 1 and 2 (Mst1 and Mst2, respectively) are potent serine/threonine kinases that are involved in cell proliferation and cell death. To investigate the physiological functions of Mst1 and Mst2, we generated Mst1 and Mst2 mutant mice. Mst1−/− and Mst2−/− mice were viable and fertile and developed normally, suggesting possible functional overlaps between the two genes. A characterization of heterozygous and homozygous combinations of Mst1 and Mst2 mutant mice showed that mice containing a single copy of either gene underwent normal organ development; however, Mst1−/−; Mst2−/− mice lacking both Mst1 and Mst2 genes started dying in utero at approximately embryonic day 8.5. Mst1−/−; Mst2−/− mice exhibited severe growth retardation, failed placental development, impaired yolk sac/embryo vascular patterning and primitive hematopoiesis, increased apoptosis in placentas and embryos, and disorganized proliferating cells in the embryo proper. These findings indicate that both Mst1 and Mst2 kinases play essential roles in early mouse development, regulating placental development, vascular patterning, primitive hematopoiesis, and cell proliferation and survival.
Mammalian sterile 20-like kinases 1 and 2 (Mst1 and Mst2, respectively) belong to the germinal center kinase group II family of mitogen-activated protein kinase-related kinases (4). The amino acid sequences of Mst1 and Mst2 are highly conserved in humans and mice; both kinases contain an N-terminal kinase domain and a C-terminal SARAH domain (17). Mst1/2 play a role in mediating the apoptosis induced by various types of apoptotic stress in vitro (1, 6, 17, 22). The kinase domain of Mst1, when activated by caspase-mediated cleavage, phosphorylates histone H2B, promoting the chromatin condensation that precedes apoptotic cell death (2, 32). Mst1/2 promote neuronal cell death by phosphorylating and activating forkhead box O (FOXO) transcription factors in response to oxidative stress (18, 36). AKT and Mst1 are mutually antagonistic in mammalian cells and in the zebrafish (3, 8).
In addition to their roles in mammalian apoptosis, Mst1/2 are also key components in the Hippo signaling pathway, which is a key developmental pathway involved in restricting cell proliferation and promoting apoptosis in Drosophila melanogaster (7, 9, 24, 31, 33). This pathway involves Hippo (Hpo) kinase (Drosophila homolog of Mst1 and Mst2) and its scaffold, Salvador (Sav) (Drosophila homolog of WW45); Warts kinase (Drosophila homolog of LATS kinases) and its scaffold, Mats (Drosophila homolog of Mob); and the transcriptional coactivator, Yorkie (Yki) (Drosophila homolog of YAP). In flies, deletion of any Hippo pathway component or overexpression of Yki leads to increased cell proliferation and decreased cell death in developing epithelial tissues (5, 7, 9, 11, 12, 24, 26, 30, 31, 33, 34). In contrast to these fly phenotypes, mice with a deletion of one of these Hippo components show more complicated phenotypes during embryogenesis. Lats2-null embryos die at approximately embryonic day 12.5 (E12.5) with pleiotropic defects, including neural tube and pericardium defects (19). Yap knockout embryos die at about E8.5 with defects in chorioallantoic fusion and yolk sac vasculogenesis (20). WW45-null embryos also die at about E18.5 in utero due to placental defects (16).
Several studies of flies have supported the importance of the Hippo pathway in mammalian tumorigenesis. Consistent with the overexpression of Yorkie in flies, overexpression of YAP also causes increased cell proliferation, expansion of progenitor cells, and reduced cell death (16, 23, 38). In fact, amplification of Yap has frequently been observed in liver and breast tumor models (23, 37), and mice with a homozygous deletion of LATS1 develop tumors (27). We have recently shown that loss of WW45 affects developing epithelial tissues, producing precancerous characteristics (16). Consistent with these findings, loss of LATS1 and LATS2 expression and mutations of WW45 and MATS1 are found in cancer cell lines (10, 14, 28, 30).
Two recent studies have showed that mice lacking Mst1 display reduced numbers of naïve T cells in secondary lymphoid organs and peripheral blood (13, 39). However, the physiological functions of Mst1 and Mst2 kinases in mice are not fully understood. To reveal the functions of these two proteins, we disrupted Mst1 and Mst2 genes by gene targeting in mice. Mst1-Mst2 double-knockout embryos die early in gestation and show defects in placental development, vascular patterning, primitive hematopoiesis, and regulation of cell proliferation and survival. These data suggest that Mst kinases are essential for the early developmental program in mice.
We isolated the mouse Mst1 genomic locus encompassing exons 2 to 5 and the Mst2 locus encompassing exon 1 from a 129/SvJ mouse bacterial artificial chromosome library. The murine Mst1-targeting vector was constructed by sequentially subcloning a 2.3-kb BamHI fragment (right arm) obtained by PCR (forward primer, 5′-AAT TGG ATC CCT TCT TCT GGC CTT TTG TGC-3′; reverse primer, 5′-AAT TGG ATC CCA AAA TAT TTC CCG CCT TGA-3′) into the BamHI site of a PGK-puro targeting vector (PGK stands for phosphoglycerokinase, and puro stands for the 1.5-kb puromycin resistance gene) and then a 3.7-kb HindIII fragment (left arm) into the HindIII site of the vector. An EcoRI site at the 5′ end of the PGK-puro cassette was retained in this targeting vector to facilitate genotyping of the targeted allele. The murine Mst2-targeting vector was constructed by subcloning a 4.4-kb BamHI-SpeI fragment into a BamHI-XbaI site of the PGK-puro targeting vector, and a 4.0-kb EcoRV fragment into an EcoRI site (blunted). The diphtheria toxin A chain gene was then inserted into a unique NotI site at the 3′ end of the right arm in each targeting vector for negative selection. In this strategy, an Mst1 genomic region spanning exons 3 and 4 (3.3 kb) and an Mst2 genomic region coding exon 1 (3.8 kb) were replaced with a 1.5-kb puromycin resistance gene (puro), introduced as a positive-selection marker.
After electroporation, embryonic stem (ES) cell clones were grown in the presence of 3 μg/ml puromycin and isolated after culturing for 8 days. Homologous recombination was confirmed by Southern blot analysis of genomic DNA using 3′ external probes (Fig. (Fig.1B).1B). Embryos and mice were genotyped by PCR using primers M-8 (5′-AGC ATG TTT GGG AAA TTT AAA AGA-3′), M-9 (5′-AAT CTG GCC AGT CTC TTT ATG AAT-3′), and PGK-3 (5′-GCA CGA GAC TAG TGA GAC GTG CTA C-3′) for Mst1, and M2gF-01 (5′-AAA GCT GGT ACT GGG GTT CA-3′), M2gR-04 (5′-CCC CAA GAA TAC ATG CCA AT-3′), and PGK-3 (5′-GCA CGA GAC TAG TGA GAC GTG CTA C-3) for Mst2. The resulting PCR products were 679 bp (wild-type allele) and 556 bp (puro allele) for Mst1 and 1,027 bp (wild-type allele) and 675 bp (puro allele) for Mst2 (Fig. (Fig.1C).1C). To generate Mst1−/−; Mst2−/− embryos, Mst1+/−; Mst2−/− males were mated with Mst1−/−; Mst2+/− females (Fig. (Fig.1D).1D). The embryos were dissected from the yolk sacs on E8.0 to E9.5 and genotyped using genomic DNA from yolk sacs. Embryo morphology was analyzed after dissection, and digital photographs were obtained using stereomicroscopy (Olympus).
For 5-bromo-2-deoxyuridine (BrdU; Sigma) treatment of embryos, pregnant mice from Mst1+/−; Mst2−/− and Mst1−/−; Mst2+/− matings at specific times were injected intraperitoneally with BrdU (50 μg/g of body weight) on E8.5 to E9.5. Pregnant mice were sacrificed 30 min after injection, and the uteri were removed. For histological analysis, embryos and tissues were fixed in 4% paraformaldehyde overnight at 4°C and embedded in paraffin wax for sectioning. Generally, 5-μm sections were cut and stained with hematoxylin and eosin (H&E).
The embryos were immunostained as described previously (16). Briefly, the embryos were fixed in 4% paraformaldehyde, bleached with 3% H2O2, blocked in 5% bovine serum albumin, incubated sequentially with primary and second antibodies, and stained with the EnVision dual-link system-horseradish peroxidase (DAB+) (DakoCytomation). Immunohistochemical analyses were performed using standard protocols with primary antibodies against PECAM-1 (BD), BrdU (BD), and Ki67 (Novocastra) and secondary horseradish peroxidase-conjugated goat anti-rat immunoglobulin G (Jackson ImmunoResearch), Alexa Fluor 488- or Alexa Fluor 594-conjugated antibodies (Molecular Probes), respectively. TUNEL assays were performed on sections using an in situ cell death detection kit (Roche). TUNEL/BrdU/Ki67/phosphorylated histone H3-positive cells were counted, and the mean number was used to obtain the relative ratio to Mst1+/−; Mst2+/− embryos.
Guinea pig polyclonal antibodies to Mst2 (GP 2) were generated by injecting guinea pigs with a keyhole-limpet-hemocyanin-conjugated peptide corresponding to the N-terminal amino acids of Mst2 (EQPPAPKSKLKKLSCys). Specific antibodies were affinity purified with the appropriate antigens.
Mst1 and Mst2 knockout mouse embryonic fibroblasts (MEFs) and embryos were lysed in lysis buffer (25 mM Tris-HCl [pH 7.4], 150 mM NaCl, 1 mM EDTA, 1 mM MgCl2, 0.2% Triton X-100, 0.3% NP-40, leupeptin, pepstatin A, phenylmethylsulfonyl fluoride, NaF, Na3OVO3, and beta-glycerophosphate). Successful ablation of Mst1 and Mst2 was confirmed by Western blot analysis using antibodies against Mst1 (Cell Signaling) and Mst2 (Cell Signaling [N-terminal region], Santa Cruz [C-terminal region], and GP 2 [N-terminal region]). Western blot analysis from E8.5 embryos were performed using antibodies against YAP, phosphorylated YAP, Mst1, Mst2, FOXO3A (Cell Signaling), WW45 (16), LATS1, LATS2 (Bethyl Laboratories Inc.), and glyceraldehyde-3-phosphate dehydrogenase (Abcam). Densitometry was performed using the software Multi Gauge V3 (Fujifilm).
Pregnant mice at specific times were injected intraperitoneally on E8.5 with rhodamine 123 (1 μg/g of body weight; Sigma) (29). The mice were sacrificed 2 h after injection, and the embryos were removed from the uteri and analyzed with a fluorescence microscope.
RNA preparation and cDNA synthesis were previously described (15). Briefly, total RNAs were extracted from Mst1 and Mst2 MEFs, E8.5 Mst1+/−; Mst2+/− and Mst1−/−; Mst2−/− embryos and placentas using Trizol reagent (Intron) according to the manufacturer's instructions. Aliquots of 1 μg of RNA were used as template for reverse transcription with oligo(dT) primers (Intron). Real-time PCR was performed with iQ SYBR green reagent kits (Genet Bio) on an iQ5 PCR machine (Bio-Rad). For quantitative PCR, SYBR fluorescence was utilized to generate a threshold cycle (CT) value for each transcript. These values were normalized for absolute quantity against Gapdh. The normalized CT values for Mst1+/−; Mst2+/− embryos were compared to the CT values for Mst1−/−; Mst2−/− embryos using the calculation method. The sequences of specific primers are available upon request.
Hematopoietic colonies were generated as described previously (15). Briefly, E8.5 yolk sacs were isolated and dissociated in collagenase with 20% fetal bovine serum, and the resulting cells were replated in 1% methylcellulose containing hematopoietic cytokines. Hematopoietic colonies were counted 4 to 7 days later.
Whole-mount in situ hybridization was carried out as previously described (15). Briefly, antisense digoxigenin-labeled riboprobes were generated from the pBluescript(+) KS vector containing amplified Mst1 and Mst2 cDNA fragments (Mst1 locus, positions 889 to 1118 in cDNA, and Mst2 locus, positions 1 to 314 in cDNA) using the digoxigenin RNA labeling kit (Roche) according to the manufacturer's instructions. The sequences of specific primers are listed in Table Table11.
Mst kinases are key regulators of the apoptotic response to various cell death signals. In addition, loss of Hippo kinase (Mst1/2 homolog) in flies results in uncontrolled cell proliferation and reduced cell death. To further explore the physiological functions of Mst kinases in mice, we targeted the murine Mst1 and Mst2 genes by homologous recombination in ES cells, generating Mst1 and Mst2 mutant mice (Fig. 1A, B, and C), as described in Materials and Methods. Western blot analysis (Fig. (Fig.1E)1E) and RT-PCR analysis (data not shown) confirmed the absence of Mst1 and Mst2 protein in mutant MEFs, demonstrating that each mutant was null for the respective allele. Mice homozygous null for the Mst1 allele (Mst1−/−) developed normally and were viable but had markedly reduced numbers of peripheral naïve T cells and exhibited dysregulated lymphocyte trafficking as described previously (13, 39; S. Oh and D.-S. Lim, personal observations). In contrast, mice homozygous for the mutant Mst2 allele (Mst2−/−) were viable, fertile, and phenotypically indistinguishable from wild-type littermates over a 24-month observation period (data not shown). These observations were in apparent contradiction to in vitro data suggesting their prominent role in apoptosis.
The lack of obvious developmental defects in Mst1−/− or Mst2−/− mice prompted us to suggest that Mst1 and Mst2 can functionally compensate for each other's deficiency. Thus, we first examined whether spatiotemporal regulation of Mst1 and Mst2 expression overlap by analyzing Mst1 and Mst2 protein levels in various adult tissues and at different developmental stages (Fig. (Fig.2A).2A). Both Mst1 and Mst2 proteins were expressed ubiquitously at particularly high levels in the lymphoid organs for Mst1 and in the brain for Mst2. During development, Mst1 and Mst2 were detected as early as E8.5, and their expression appeared to increase to a maximum on E12.5 and decrease thereafter, suggesting their possibly important roles in early embryonic development. Next, we performed a whole-mount in situ hybridization on E8.5 to determine the expression profile of Mst1 and Mst2 transcripts. The embryos had ubiquitous and largely overlapping expression of Mst1 and Mst2 with particularly high expression along the neural fold (Fig. (Fig.2B).2B). Given the spatiotemporal overlap in embryonic and adult tissues and extensive homology of Mst1 and Mst2 proteins, it is likely that functional redundancy may account for the lack of an overt phenotype in Mst1−/− and Mst2−/− single-knockout mice.
Next, we crossed Mst1−/− and Mst2−/− mice to generate Mst1+/− Mst2+/− mice, and these mice are consequently intercrossed to generate Mst1+/−; Mst2−/− mice, Mst1−/−; Mst2+/− mice, and Mst1−/−; Mst2−/− mice (Fig. (Fig.1D).1D). We first found that Mst1+/−; Mst2+/− mice Mst1+/−; Mst2−/− mice, and Mst1−/−; Mst2+/− mice were viable and fertile. We then crossed Mst1+/−; Mst2−/− animals with Mst1−/−; Mst2+/− animals to obtain Mst1−/−; Mst2−/− double mutant mice. Notably, we did not observe any Mst1−/−; Mst2−/− mice at birth, indicating that these double-null mice were embryonic lethal. Next, we determined the developmental stage at which lethality occurs and found viable Mst1−/−; Mst2−/− embryos on E8.0 to E9.5 with distorted Mendelian frequency (Table (Table2).2). This is consistent with our suggestion that Mst1 and Mst2 functionally overlap at least during early development.
By E8.5, the Mst1−/−; Mst2−/− embryos were deformed and were significantly smaller in size than Mst1+/−; Mst2+/− littermates (Fig. 3A to D). Although some Mst1−/−; Mst2−/− embryos were occasionally recovered on E9.0 or E9.5, many were undergoing resorption by E9.5 (Fig. 3H and L). Mst1−/−; Mst2−/− embryos on E9.5 showed overt developmental delay, as evidenced by incomplete embryo turning, fusion of cranial neural plates, and growth retardation (Fig. (Fig.3L).3L). No Mst1−/−; Mst2−/− embryos were recovered beyond E9.5 (data not shown).
Since the growth retardation characteristic of Mst1−/−; Mst2−/− embryos manifested on approximately E8.5, we further confirmed these phenotypes in serial histological sections of embryos on E8.0 to E9.5. Histological sections revealed that E8.0 Mst1−/−; Mst2−/− embryos were in the process of gastrulation and had normally formed all three germ layers and extraembryonic tissues (Fig. (Fig.4A,4A, panels a and b). We also analyzed the levels of expression of endodermal genes (Foxa2, Sox17, Afp, and Cerl), ectodermal genes (Fgf5, NeuroD, Sox2, and Otx2), and mesodermal genes (Tpbp, Gsc, Fgf8, and Tbx6) by quantitative RT-PCR (Fig. (Fig.4B).4B). Expression levels of these marker genes expressed in the endoderm, ectoderm, and mesoderm had no significant difference between Mst1+/−; Mst2+/− embryos and Mst1−/−; Mst2−/− embryos, indicating normal germ layer formation in Mst1−/−; Mst2−/− embryos. Of interest, transverse sections from E8.5 to 9.0 showed that the Mst1−/−; Mst2−/− neural plate failed to fuse to form a neural tube and was kinked dorsoventrally (Fig. (Fig.4A,4A, panels c to f). These observations indicate that even Mst1 and Mst2 are required for mouse development in the early postimplantation period, but not for axial or germ layer patterning.
Examination of organogenesis in Mst1−/−; Mst2−/− embryos on E9.5 revealed a severe developmental retardation of early endocardium (Fig. (Fig.4C,4C, panel b). Because Mst1 is a key kinase in cardiomyocyte apoptosis (35), we stained for smooth muscle α-actin, a marker for cardiomyocytes, in Mst1+/−; Mst2+/− embryos and Mst1−/−; Mst2−/− embryos. Unexpectedly, the smooth muscle α-actin immunohistochemical staining pattern was similar in the Mst1+/−; Mst2+/− embryos and Mst1−/−; Mst2−/− embryos (Fig. (Fig.4C,4C, panels c and d), suggesting that cardiomyocytes developed normally. Interestingly, Mst1−/−; Mst2−/− embryos also exhibited marked disorganization of the neuroectoderm: they lacked any obvious boundary of the neuroectoderm in the forebrain region and had a disorganized accumulation of neuroepithelial cells (Fig. (Fig.4C,4C, panel f). These findings indicate that Mst1 and Mst2 are dispensable for differentiation of progenitor cells that constitute vital organs, such as the heart and brain, but are actively involved in organogenesis.
The initial step in placenta development involves attachment of the allantois to the chorionic plate on about E8.5, followed by vascular invasion of the labyrinth layer. On E8.0, we observed normal fusion of the chorionic plate and the allantois in both Mst1+/−; Mst2+/− and Mst1−/−; Mst2−/− embryos (Fig. (Fig.4A,4A, panel b). From E8.5 to E9.5, Mst1+/−; Mst2+/− embryos showed normal morphogenesis of the labyrinth layer, including invasion of fetal blood vessels from the allantois into the chorionic plate, and differentiation of trophoblast cells. In contrast, formation of the labyrinth layer and all other subsequent steps in the development of the placenta were severely disrupted in Mst1−/−; Mst2−/− embryos (Fig. (Fig.4D).4D). Notably, whereas the labyrinth layer of Mst1+/−; Mst2+/− placentas contained blood vessels with embryo-derived nucleated erythrocytes, that of the Mst1−/−; Mst2−/− placenta contained sparse blood vessels with few nucleated erythrocytes (boxed areas in Fig. Fig.4D4D).
To further characterize placental defects, we analyzed cell proliferation and cell death in the placenta. TUNEL staining of E8.5 placentas revealed a significant increase in the number of dying cells in Mst1−/−; Mst2−/− placentas (Fig. (Fig.4E,4E, panels a and b). Immunostaining of Mst1−/−; Mst2−/− placentas with anti-BrdU and anti-Ki-67 antibodies revealed that Ki-67 nuclear staining was apparently reduced in the ectoplacental cone and was restricted to the chorion (Fig. (Fig.4E,4E, panels c and d). However, we observed no detectable difference in BrdU incorporation between Mst1+/−; Mst2+/− placentas and Mst1−/−; Mst2−/− placentas (Fig. (Fig.4E,4E, panels c and d). To further examine the origin of placental defect, we isolated RNA from E8.5 placentas and performed quantitative RT-PCR for transcriptional factor Gcm1 (for labyrinthine trophoblasts), Mash2 (for ectoplacental cone and the chorion), Tpbp (for spongiotrophoblast), and Pl1 (for trophoblast giant cell) (29). In Mst1−/−; Mst2−/− placentas, the Pl1 and Mash2 mRNA levels were significantly decreased, whereas the mRNA levels of other markers, such as Gcm1 and Tpbp, showed no alteration (Fig. (Fig.4F).4F). On the basis of these results, ablation of Mst1 and Mst2 leads to disruption of the generation of specific placental lineages.
Next, we assessed the functional capacity of the placenta by injecting rhodamine 123 and measuring the efficiency of transport from the placenta to the fetus (29). Rhodamine 123 was readily detectable within 2 h after intraperitoneal injection into the pregnant female in Mst1+/−; Mst2+/− embryos, but significantly less rhodamine 123 was detected in Mst1−/−; Mst2−/− littermate embryos (Fig. (Fig.4G).4G). Therefore, placental vascular defects, disruption in specific placental lineages, and the resulting deficit in placental function contribute to the death of Mst1−/−; Mst2−/− embryos.
Since deletion of Mst kinases decreased cell proliferation and increased cell death in the placenta, we evaluated Mst1+/−; Mst2+/− embryos and Mst1−/−; Mst2−/− embryos for similar effects. We first pulse-labeled E9.0 embryos with BrdU for 30 min to examine embryonic cell proliferation (Fig. (Fig.5A,5A, panels a and b). However, whereas BrdU-positive cells were readily detected in Mst1+/−; Mst2+/− embryos, BrdU-positive cells were largely absent in Mst1−/−; Mst2−/− embryos, likely reflecting limited availability of BrdU due to defects in placental function. Therefore, we measured embryonic cell proliferation by staining for Ki-67 (Fig. (Fig.5A,5A, panels c and d). Ki-67-positive cells were densely distributed in highly proliferative regions of Mst1+/−; Mst2+/− embryos, particularly in the neural tube. The numbers of proliferating cells in E9.0 and E9.5 Mst1−/−; Mst2−/− embryos, as detected by anti-Ki-67 staining, were not higher than those in Mst1+/−; Mst2+/− embryos. Of interest, these anti-Ki-67-positive proliferating cells were restricted to the proliferative neural tube but were rather disorganized (Fig. (Fig.5A).5A). This indicates that ablation of Mst1 and Mst2 does not significantly increase cell proliferation in the embryo proper. However, consistent with observations related to the organogenesis of heart and neuroectoderm, ablation of Mst1 and Mst2 severely disrupted the spatiotemporal patterning of proliferating cells within the embryo.
We next analyzed cell death in Mst1−/−; Mst2−/− embryos by TUNEL staining of E8.5 and E9.5 embryo sections (Fig. (Fig.5B).5B). Mst1−/−; Mst2−/− embryos had greater numbers of apoptotic cells on both E8.5 and E9.5than Mst1+/−; Mst2+/− embryos did. Since Mst kinases are involved in cell cycle progression (25), we further examined the number and distribution of mitotic cells in double mutant embryos. Although we have observed the severe disorganization in distribution of mitotic cells due to overall disorganization of neural tissue, we found no difference in the number of mitotic cells in Mst1+/−; Mst2+/− embryos and Mst1−/−; Mst2−/− embryos (Fig. (Fig.5C).5C). Collectively, these data suggest that the loss of Mst kinases increases cell death during embryonic development. This is unlike the case in cancer cells, where loss of Mst function disrupts the normal apoptotic process, consistent with the role of these kinases as tumor suppressors. The increased apoptosis, growth and developmental retardation, and eventual death of Mst1−/−; Mst2−/− embryos could be a secondary consequence of the observed placental functional defects.
Even though placental functional defects in Mst1−/−; Mst2−/− embryos could cause the embryonic defects, including increased apoptosis and growth and developmental retardation, there is a possibility that Mst1−/−; Mst2−/− embryos also have the intrinsic embryonic functional defects. In fact, whereas histological examination of Mst1−/−; Mst2−/− embryos from E8.0 to E9.0 revealed apparently normal yolk sac development (Fig. (Fig.6A),6A), by E9.5 we observed a collapse in the yolk sac vasculature. Thus, we further examined the vascular architecture of embryos by immunostaining whole-mounted E9.5 Mst1−/−; Mst2−/− embryos with anti-PECAM-1. As shown in Fig. Fig.6B,6B, panels b and c, differentiated endothelial cells were strikingly abnormal in the Mst1−/−; Mst2−/− embryos, and total vascularity was decreased. On E9.5, PECAM-1 staining readily detected well-organized blood vessels in the dorsal aorta and intersomitic regions of Mst1+/−; Mst2+/− embryos. However, the dorsal aorta of Mst1−/−; Mst2−/− embryos had only a primitive vasculature and the somites of Mst1−/−; Mst2−/− embryos did not have small sprouting blood vessels. These results indicate that disruption of Mst1 and Mst2 genes affects vascular patterning and vascular branching in the yolk sac and embryonic tissues.
Mst1 protein is highly expressed in lymphoid tissues (Fig. (Fig.2A)2A) and Mst1−/−; Mst2−/− yolk sacs on E9.5 appeared to have reduced numbers of primitive hematopoietic cells (Fig. (Fig.6A).6A). Therefore, we determined primitive hematopoietic capacity of Mst1−/−; Mst2−/− yolk sacs by replating the E8.5 yolk sacs into methylcellulose containing hematopoietic cytokines (Fig. (Fig.6C).6C). The numbers of hematopoietic colonies that were generated from Mst1+/−; Mst2+/− yolks sacs, Mst1+/−; Mst2−/− yolk sacs, and Mst1−/−; Mst2+/− yolk sacs were not significantly different (Fig. (Fig.6C).6C). In contrast, E8.5 Mst1−/−; Mst2−/− yolk sacs had significantly decreased number of colonies formed when subjected to hematopoietic replating. These results suggest decreased numbers and/or differentiation capacities of intrinsic hematopoietic progenitors in Mst1−/−; Mst2−/− yolk sacs.
Because the phenotypic differences between Mst1+/−; Mst2+/− embryos and Mst1−/−; Mst2−/− embryos during the course of development were very dramatic, we examined the development of Mst1+/−; Mst2−/− embryos and Mst1−/−; Mst2+/− embryos. As noted above, Mst1+/−; Mst2−/− mice and Mst1−/−; Mst2+/− mice were viable and fertile. On E8.0 to E9.0, Mst1+/−; Mst2−/− embryos and Mst1−/−; Mst2+/− embryos had phenotypically normal yolk sacs and placentas (Fig. (Fig.7A7A for Mst1+/−; Mst2−/− embryos and 7B for Mst1−/−; Mst2+/− embryos). PECAM-1 staining showed that neither Mst1+/−; Mst2−/− embryos nor Mst1−/−; Mst2+/− embryos had vascular defects (Fig. (Fig.7C).7C). Even at the later E12.5 stage, Mst1+/−; Mst2−/− embryos were indistinguishable from the normally developing Mst1+/−; Mst2+/− embryos (Fig. (Fig.7D,7D, panel a for Mst1+/−; Mst2+/− embryos and panel b for Mst1+/−; Mst2−/− embryos). Collectively, these data indicate that a haplosufficiency of either Mst1 or Mst2 can compensate for the loss of the other's function; only complete ablation of both Mst1 and Mst2 leads to developmental defects.
Finally, we determined whether components in Mst signaling pathway were dysregulated in Mst1−/−; Mst2−/− embryos on E8.5 by Western blot analysis (Fig. (Fig.8A).8A). Phosphorylation of FOXO3A on serine 207 (18) and LATS1/2, known substrates of Mst1/2, as judged by phospho-specific antibody or shift in electrophoretic mobility, were not observed in both Mst1+/−; Mst2+/− embryos and Mst1−/−; Mst2−/− embryos. We observed unaltered phosphorylation of YAP by LATS1/2 in both Mst1+/−; Mst2+/− embryos and Mst1−/−; Mst2−/− embryos. Since YAP is the principal regulator of TEAD4 transcription factor for trophectoderm lineage development (21), we determined the levels of expression of Cdx2, a major transcription target of TEAD4 (Fig. (Fig.8C).8C). However, expression of Cdx2 was not significantly changed in Mst1−/−; Mst2−/− embryos. Of interest, ablation of Mst1 and Mst2 significantly decreased the protein level of WW45, suggesting that Mst kinases are required for maintaining WW45 stability. Overall, Mst1−/−; Mst2−/− embryos did not exhibit any dysregulation in phosphorylation or levels of components in the Mst signaling pathway compared to Mst1+/−; Mst2+/− embryos on E8.5 (Fig. (Fig.8B).8B). WW45 is required for Mst1 activation and consequent phosphorylation of LATS1/2 and YAP to sequester YAP in the cytoplasm during epithelial differentiation (16). Comparing this previous report with these data in present study, the effects of ablation of the Mst signaling pathway likely vary between early embryonic development and epithelial differentiation.
Mst1 and Mst2 are best known for their originally identified function as key apoptotic kinases that can be activated by various cell death signals in in vitro culture systems. Hippo (Hpo), the Drosophila homolog of Mst kinases, is involved in restricting cell proliferation and promoting apoptosis during development of the fly. However, the roles of Mst kinases in mammalian development have not been fully investigated.
This study reveals that Mst kinases are essential for normal mouse development. Although both Mst1−/− and Mst2−/− single-knockout mice were outwardly normal and lacked severe morphological or histological defects, Mst1−/−; Mst2−/− double mutant embryos died in utero with multiple developmental defects. First, Mst1−/−; Mst2−/− embryos showed functional placental defects that could be expected to restrict the availability of nutrients to the embryo. These placental defects could account for the observed growth retardation and decreased rate of cell proliferation in Mst1−/−; Mst2−/− embryos, and ultimately for inducing cell death in the embryo proper. Second, Mst1−/−; Mst2−/− embryos showed circulation defects, including abnormal vascular patterning in the yolk sac and embryo proper, as well as turning failure, cellular disorganization, and failure of neural ectodermal cells to fuse to form a neural tube. In particular, Mst1−/−; Mst2−/− embryos showed decreased numbers and/or differentiation capacities of hematopoietic progenitors in the yolk sacs. These data suggest that Mst1 and Mst2 are essential for regulating primitive hematopoiesis intrinsically. As previously noted, defects of many tissues in double mutant embryos, including placental and/or intrinsic embryonic defects could be secondary to the observed developmental retardation. Taken together with the high sequence homology between Mst kinases, the phenotypic differences between single and double knockouts suggest that Mst1 and Mst2 kinases are functionally redundant. Collectively, these results indicate that Mst1 and Mst2 are both essential for early developmental processes, including placental development, vascular patterning, and primitive hematopoiesis.
In Drosophila and mammals, the Hippo pathway restricts cell proliferation and promotes apoptosis in differentiating epithelia (7, 9, 11, 12, 16, 24, 30, 31, 33, 34). We have recently shown that Mst1 kinase phosphorylates LATS1/2 and consequent phosphorylation of YAP to sequester YAP in cytoplasm (16) during epithelial differentiation in mammals. In this study, however, we could not observe any phosphorylation of known substrate proteins even in Mst1+/−; Mst2+/− embryos on E8.5. Moreover, transcriptional coactivation by YAP was not significantly altered as judged by Cdx2 transcription. Instead, loss of Mst1/2 decreased the level of the WW45 protein, suggesting possible regulation of stability. This finding suggests that although the Hippo pathway conserved from Drosophila is important for epithelial proliferation and differentiation, Mst kinases may be involved in other pathways crucial for early mammalian development. It could also be possible that spatiotemporal regulation affects either the composition of Mst signaling components or substrate specificities during embryogenesis.
In mammals, dysfunctions in the Hippo pathway lead to inappropriate proliferation and expansion of cell compartment and subsequent induction of tumor formation. Unexpectedly, we found no evidence for tumor development over a 24-month observation period in mice lacking either Mst1 or Mst2 (data not shown), a result that may indicate that that Mst1 and Mst2 functionally compensate for each other to prevent tumorigenesis. Instead, we observed that ablation of Mst kinases increased cell death and decreased proliferation. This observation is not in accord with in vitro data from mammalian cells. Thus, it is possible that these defects are not direct consequences of ablation of Mst kinases in the embryo proper but are secondary to placental defects and insufficient nutrient transport. To exclude this possibility and investigate a possible role for Mst kinases in cell proliferation and cell death, we attempted to isolate and culture MEFs from Mst1−/−; Mst2−/− embryos on E8.5 to E9.5. However, we were unable to propagate fibroblasts from embryos on E8.5 or E9.5 under standard growth conditions. Interestingly, in contrast to previous reports that depletion of Mst1 reduced the incidence of Fas- and oxidative stress-induced apoptosis in cancer cell lines (18, 22), we saw no significant decrease in the incidence of Fas- or H2O2-induced apoptosis in Mst1 or Mst2 single-null MEFs (Oh and Lim, personal observations), again suggesting the functional redundancy of Mst1 and Mst2. The reason for the discrepancy between our results and previous reports is not clear but may reflect differences in cellular context between primary MEFs and cancer cells.
We also examined possible haploinsufficiency effects of Mst1 in an Mst2-deficient background and vice versa. However, Mst1+/−; Mst2−/− mice and Mst1−/−; Mst2+/− mice were developmentally normal and viable and did not develop tumors over a 10-month observation period (data not shown). Further investigation of tumorigenesis in Mst1+/−; Mst2−/− mice and Mst1−/−; Mst2+/− mice will be required to understand the gene dosage-dependent functions of Mst genes. Because Mst1−/−; Mst2−/− embryos have severe developmental defects at an early embryonic stage, the use of conditional knockout mice to exclude developmental defects is warranted in future studies of the molecular mechanisms of Mst proteins in cell proliferation and apoptosis.
This work was supported by the National Research Laboratory Program, the South Korea National Cancer Center Control Program, and the Nuclear Research Program of South Korea. This work was also partly supported by a grant from the KOSEF (M10503010002) to K.-S.K.
Published ahead of print on 28 September 2009.