Abstract
Heterotrimeric GTP-binding proteins (G proteins) participate in cellular signaling and regulate a variety of physiological processes. Disruption of the gene encoding the G protein subunit α13 (Gα13) in mice impaired the ability of endothelial cells to develop into an organized vascular system, resulting in intrauterine death. In addition, Gα13 (−/−) embryonic fibroblasts showed greatly impaired migratory responses to thrombin. These results demonstrate that Gα13 participates in the regulation of cell movement in response to specific ligands, as well as in developmental angiogenesis.
G proteins are signal transducers that couple a variety of receptors to effectors such as enzymes and ion channels (1). Heterotrimeric G proteins consist of an α, β, and γ subunit and are defined by the identity of their α subunit. Sixteen α-subunit genes have been cloned, and on the basis of sequence similarities they can be divided into four families (Gαs, Gαi/o, Gαq, and Gα12) (2). Whereas considerable knowledge exists about the cellular function of most members of the Gαs, Gαi/o, and Gαq families, the cellular and biological functions of G12 and G13 (which constitute the G12 family) are not known. The α subunit of G13 is expressed ubiquitously and shows 67% amino acid identity to Gα12 (3). Studies with constitutively activated forms of Gα13 have provided evidence that it may participate in the regulation of the Jun kinase/stress-activated protein kinase pathway (4), the Na+/H+ exchanger, and the Rho-dependent formation of stress fibers (5, 6).
To elucidate its biological function, we have disrupted the gene encoding Gα13 (Gna-13) by homologous recombination in embryonic stem (ES) cells (Fig. 1). Germline-transmitting chimeric mice were generated with two independently targeted ES cell clones, and the targeted mutation of the Gα13 gene was crossed into inbred (129/Sv) and outbred backgrounds (C57BL/6J). Mice heterozygous for the targeted mutation appeared normal and showed no obvious defects over an 18-month period. However, no homozygous Gα13 mutant mice were found among 451 newborn animals from intercrosses between heterozygous mice. The ratio of wild-type to heterozygous offspring was 1:1.93, indicating a recessive lethal phenotype.
Targeted disruption of the murine Gα13 gene. (A) Part of the wild-type Gα13 locus containing the first and second exon (WT), the targeting construct (TC), and the targeted locus (Mut.) are shown. The Gα13 gene was disrupted by means of a targeting vector in which a Bam HI-Xho I fragment containing the first exon was replaced by the neomycin resistance (neor) gene (22). The sizes of the Sma I fragments predicted to hybridize to the indicated diagnostic probe are shown. Restriction endonucleases: B, Bam HI; E, Eco RI; K, Kpn I; S, Sma I; and X, Xho I. (B) RT-PCR analysis of Gα expression. Total RNA was prepared from E8.5 mutant and wild-type yolk sacs (YS) and from the embryo proper (EP) and used for RT-PCR analysis with Gα13-, Gα12-, and Gαq-specific primers hybridizing to regions encoded by different exons (23).
To determine the time at which the Gα13 deficiency was manifest, we examined embryos from heterozygous intercrosses at different developmental stages. Homozygous embryos were mostly resorbed by embryonic day 10.5 (E10.5), but could be recovered at E9.5 with hearts still beating. E9.5 homozygous mutants [detected by Southern (DNA) blotting or polymerase chain reaction] were poorly developed and misshapen compared to wild-type embryos (Fig. 2, C and D) and represented about 25% of all embryos at this stage. E8.5 Gα13 (−/−) embryos, in contrast, appeared to be normal (Fig. 2, A and B). The phenotypes of homozygous Gα13 mutant embryos were indistinguishable in inbred and outbred genetic backgrounds as well as in mouse lines derived from two independent ES cell clones.
(left). Gα13 mutant phenotype. (A and B) Side view of wild-type (A) and mutant E8.5 embryos (B). (C and D) Side view of wild-type (C) and mutant E9.5 embryos (D). (E to H) Whole mount immunohistostaining of wild-type (E and G) and mutant yolk sacs (F and H) from E8.5 (E and F) and E9.5 embryos (G and H) with anti-PECAM-1 (24). Bars: 250 μm (A and B), 500 μm (C and D), 350 μm (E and F), and 650 μm (G and H).
(right). Vascular system defects in Gα13 mutant embryos. (A and B) Hematoxylin-eosin-stained sections through E9.0 wild-type (A) and mutant yolk sac (B). Am, amnion; Bc, blood cell; En, extraembryonic endoderm; Me, extraembryonic mesoderm. (C and D) Toluidin blue-stained transverse Epon sections through heads of wild-type (C) and mutant E8.5 embryos (D) showing enlarged blood vessels in the head mesenchyme of Gα13-deficient embryos (asterisks). (E and F) Toluidin blue-stained transverse Epon sections through heads of wild-type (E) and mutant E9.5 embryos (F). (G and H) Magnified areas from transverse section through head of mutant E9.5 embryo showing abnormal location of blood vessel between embryonic ectoderm and neuroepithelium of the mesencephalon (G) as well as defective blood vessels (H). Arrows point to endothelial cells; arrowheads mark degenerated endothelial-like cells. (I) Ultrastructural analysis of endothelial cells from E9.5 mutant embryos. Arrowheads point to junctions between individual endothelial cells (25). Bars: 60 μm (A and B), 120 μm (C and D), 170 μm (E and F), 40 μm (G and H), and 0.9 μm (I).
We examined the expression of the Gα13 gene by means of reverse transcription-polymerase chain reaction (RT-PCR) with total RNA prepared from wild-type and mutant E8.5 embryos. The results (Fig. 1B) demonstrate the expression of Gα13 both in the embryo proper and in the yolk sac of wild-type embryos but not in mutant embryos. Two other G protein α subunits, Gα12 and Gαq, were expressed both in mutant and wild-type embryos (Fig. 1B). Whole-mount in situ hybridization indicated that Gα13 was expressed ubiquitously in E8.5 and E9.5 wild-type embryos, whereas no expression was detected in homozygous mutant embryos. Gα12, similarly, showed ubiquitous expression in E8.5 and E9.5 wild-type embryos but was also expressed in Gα13 homozygous mutant embryos (7). Ubiquitous expression suggests that Gα13 might play a role in basic cellular function, whereas the apparently normal development of the embryo until E8 to E8.5 suggests that the absence of Gα13 gene function did not have an acute effect until a specific developmental stage was induced.
The first obvious morphological defect in Gα13 homozygous mutant embryos was present in the yolk sac, which starting from E8.5 appeared opaque and roughened compared to wild-type yolk sacs and developed no visible blood vessels. We visualized the vascular endothelial cells in wild-type and mutant yolk sacs in whole mount with monoclonal antibodies to PECAM-1 (anti-PECAM-1), which serve as a marker for differentiated endothelial cells (8). In E8.5 mutant yolk sacs, differentiated endothelial cells were present but did not form the organized honeycombed array seen in wild-type yolk sacs (Fig. 2, E and F). In normal development, between E8.5 and E9.5 endothelial cells reorganize into a vascular network that feeds into the embryo proper. E9.5 mutant yolk sacs, however, showed no vascular structures and appeared to contain fewer endothelial cells (Fig. 2, G and H). The degeneration of the yolk sac vascular system resulted in the separation of the mesodermal and endodermal layers of the yolk sac with the occasional presence of hematopoietic cells, whereas blood islands containing hematopoietic cells were readily apparent at this stage between the closely attached layers in the wild-type yolk sac (Fig. 3, A and B).
Similar defects in organization of the vascular system were observed in the embryo proper. The head mesenchyme of E8.5 mutant embryos contained enlarged blood vessels instead of the small vessels found in wild-type or heterozygous embryos (Fig. 3, C and D). The endothelial cell layer was at some places defective, resulting in leakage of blood cells into the mesenchyme. The heart and dorsal aortas appeared to be normal. In contrast, the E9.5 mutant embryos were dystrophic and cell death was observed throughout the embryo, especially in the neuroepithelium. Blood vessels of the head mesenchyme were extremely enlarged and highly disorganized; they were lined by endothelial cells and contained some nucleated blood cells (Fig. 3F). Blood vessels were also found at abnormal locations, such as the space between the surface ectoderm and the neuroepithelium (Fig. 3G). Sporadic discontinuities of the endothelial layer resulted from a lack of cell integrity rather than from defective cell junctions that appeared to be normal at high magnification (Fig. 3, H and I). We conclude that Gα13 homozygous mutant embryos die because of a failure to develop a functional vascular system.
Blood vessel formation during embryonic development is mediated by two processes: vasculogenesis (the differentiation of endothelial cells from progenitor cells) and angiogenesis (the subsequent growth and sprouting of endothelial cells, which gives rise to the organized vascular system) (9). The lack of Gα13 did not affect the differentiation of progenitor cells into endothelial cells, which were apparent throughout the embryo. Rather, defects occurred during the subsequent process of angiogenesis, which includes sprouting, growth, migration, and remodeling of endothelial cells. The vascular system defects at E8.5 were most apparent in the extraembryonic vessels and in the vessels of the head mesenchyme. The head mesenchyme is probably vascularized by angiogenesis through migration of angioblasts rather than by in situ differentiation of endothelial cells (10). Examination of the molecular signaling processes that regulate vascular system development has so far focused mainly on receptor tyrosine kinases and cell adhesion molecules. Both systems play important roles in vasculogenic and angiogenic processes (9, 11, 12, 13). The phenotype of Gα13-deficient mice demonstrates that G protein-mediated signal transduction processes are also involved in angiogenesis.
To examine possible defects in responses to extracellular stimuli, we cultured fibroblastlike cells from E8.5 embryos as described (13). Gα13 (−/−) cells were morphologically indistinguishable from wild-type cells and attached well to fibronectin and polylysine. Thrombin, whose receptor couples to Gα13 (14), bradykinin, and lysophosphatidic acid (LPA), increased the production of inositol phosphates in wild-type and mutant cells, indicating that both cell types expressed appropriate G protein-coupled receptors (Fig. 4A). Similarly, mitogenic effects of thrombin, LPA, and fetal calf serum (FCS) were not significantly different in wild-type and Gα13-deficient cells (Fig. 4B). Whereas in wild-type cells, thrombin, fibronectin, and FCS (7) caused an increase in cell migration, in Gα13-deficient cells, the effect of thrombin was almost completely abrogated (Fig. 4C). Cell migration is a complex integrated process that involves the organized polymerization of actin filaments and the regulated formation of adhesive complexes (15). Cdc42, Rac, and Rho, all members of the Rho family of small molecular weight G proteins, participate in the regulation of these processes by extracellular factors (16). An activated form of Gα13 induces cytoskeletal changes by way of Rho and activates Na+/H+ exchange in a RhoA- and Cdc42-dependent manner (5, 6). However, other G proteins—including Gi and G12, which are expressed in embryonic fibroblasts (7) and are activated by thrombin receptors (14)—have been implicated in the regulation of cytoskeletal organization or cell motility (6, 17). We have shown that preincubation of embryonic fibroblasts with pertussis toxin, which uncouples Gi-type G proteins from receptors, resulted in a 50% reduction in migration of wild-type cells in response to thrombin (7). Thus, our data suggest that Gα13 is required for the full migratory response to certain stimuli, but is probably not the only G protein involved in G protein-mediated regulation of cell movement.
Phenotype of Gα13-deficient embryonic fibroblasts. (A) Accumulation of inositol phosphates in the absence (open bars) and presence of thrombin (Thr., 1 U/ml), bradykinin (Bk, 100 nM), and LPA (100 nM) in cells from wild-type animals (left) and from Gα13-deficient animals (right). Shown are mean values of triplicates ± SD (26). (B) Proliferation of wild-type and Gα13 (−/−) cells in response to thrombin (1 U/ml), LPA (100 nM), and FCS (5% v/v). Shown are mean values of triplicates ± SD (26). (C) Migration of serum-starved embryonic fibroblasts in response to thrombin (0.1 U/ml), LPA (50 nM), and fibronectin (Fn, 1 μg/ml). Cell migration was examined with a blindwell microchamber (27). Numbers represent mean ± SD (n = 6) of total cells per high-power field (200×).
We speculate that a failure of local cell movement and orientation in response to specific extracellular stimuli is a cellular mechanism underlying the observed vascular system defects in Gα13 (−/−) embryos. The observed defects suggest that Gα13 and its closest relative, Gα12, fulfill at least partially nonoverlapping cellular and biological functions. In the adult organism, new vessels are formed only through angiogenesis, and adult angiogenesis with few exceptions is confined to pathological situations like wound healing and tumor angiogenesis, which is important for the growth of solid tumors (18). The continued expression of Gα13 and its role in developmental angiogenesis suggest that G protein-mediated signaling events involving Gα13 may also be important in these pathological processes.
