A Role for Thrombin Receptor Signaling in Endothelial Cells During Embryonic Development

See allHide authors and affiliations

Science  31 Aug 2001:
Vol. 293, Issue 5535, pp. 1666-1670
DOI: 10.1126/science.1061259


The coagulation protease thrombin triggers fibrin formation, platelet activation, and other cellular responses at sites of tissue injury. We report a role for PAR1, a protease-activated G protein–coupled receptor for thrombin, in embryonic development. Approximately half of Par1 –/– mouse embryos died at midgestation with bleeding from multiple sites. PAR1 is expressed in endothelial cells, and a PAR1 transgene driven by an endothelial-specific promoter prevented death ofPar1 –/– embryos. Our results suggest that the coagulation cascade and PAR1 modulate endothelial cell function in developing blood vessels and that thrombin's actions on endothelial cells—rather than on platelets, mesenchymal cells, or fibrinogen—contribute to vascular development and hemostasis in the mouse embryo.

The serine protease thrombin is the product of a highly regulated cascade of zymogen activation that is triggered when plasma coagulation factors meet tissue factor. Tissue factor is expressed by cells that do not directly contact blood and by cytokine-activated leukocytes and endothelial cells. In the adult, thrombin is generated in the setting of tissue injury or inflammation. Thrombin cleaves fibrinogen to fibrin monomer, which polymerizes to form fibrin matrices. Thrombin also activates platelets and endothelial cells. These observations and the phenotypes of gain-of-function and partial loss-of-function mutations in coagulation factor genes in humans cast the coagulation cascade and its effector protease thrombin in the role of orchestrating hemostatic and inflammatory responses to tissue injury (1– 3).

Cellular responses to thrombin are mediated, at least in part, by protease-activated G protein–coupled receptors (PARs) (3). Mouse embryos lacking PAR1 or coagulation factors die with varying frequency at midgestation, often with signs of bleeding (4–10). We sought to define the cellular basis for death of Par1 –/– embryos and the relation between the phenotypes caused by deficiencies in coagulation factors and PAR1.

Matings between Par1 +/– andPar1 –/– mice in a C57BL/6J background (≥97%) yielded 135 Par1 +/– and 73Par1 –/– progeny (number ofPar1 –/– offspring was 54% of that expected by Mendelian inheritance; P < 0.001 by chi-square test). Of the Par1 –/– offspring, 36 were females, and 37 were males. These results confirmed the partial embryonic lethality initially reported in mixed 50%C57BL/6J-50%129/Sv and in pure 129/Sv backgrounds (4, 11).Par1 –/– ×Par1 –/– matings continued to yield approximately 50% embryonic loss over ∼15 generations. This observation and the similar frequencies of embryonic death in the C57BL/6J and 129/Sv inbred backgrounds suggest that the partial penetrance of the Par1 –/– phenotype is not due to a modifier gene.

Characterization of Par1 –/– embryos in the C57BL/6J (≥97%) background revealed hemorrhage and cardiovascular failure at midgestation (12) (Fig. 1). At embryonic day 8.75 (E8.75), Par1 –/– and wild-type embryos were indistinguishable by gross appearance, and somite counts were present in equal numbers inPar1 +/– × Par1 +/–litters. At E9.5, gross examination revealed blood in the exocoelomic and/or pericardial cavities in 22% of Par1 –/–embryos. These embryos were pale, and their yolk sacs lacked blood-filled vessels. Embryos with gross bleeding usually had a dilated pericardial sac, a sign of cardiovascular failure. By E10.5, bleeding was seen in 35% of Par1 –/– embryos; pericardial bleeding was especially prominent (Fig. 1D). By E12.5, 52% of Par1 –/– embryos were dead with evident bleeding. The remaining E12.5 Par1 –/– embryos were alive and appeared normal. Analysis of histological sections of gravid uterine segments containing Par1 –/–embryos revealed that bleeding occurred earlier and more frequently than was apparent by gross examination (Fig. 1A). Collections of extravasated blood cells were seen as early as E9.0, before gross bleeding was detected, and by E9.5, blood cells were detected in the exocoelomic, amniotic, and/or pericardial and peritoneal cavities in 19 of 29 Par1 –/– embryos (66%). No bleeding was detected in 16 wild-type E9.5 embryos examined in parallel. Of the 19 E9.5 Par1 –/– embryos with microscopic bleeding, four had not developed dilated pericardial sacs or other apparent abnormalities, and live E9.5 to E10.5 embryos with both gross bleeding and vigorous heart beats were occasionally seen. Dead embryos always had evidence of bleeding. At face value, these observations suggest that bleeding preceded and probably caused rather than resulted from cardiovascular failure.

Figure 1

Phenotype ofPar1 –/– embryos. (A) Frequency of abnormalities at various gestational ages (dpc, days post coitum). Gross bleeding indicates the presence of pooled blood in the pericardial or exocoelomic cavities detected with a dissecting microscope. Microscopic bleeding indicates the presence of collections of embryonic blood cells in the pericardial and/or peritoneal, amniotic, and/or exocoelomic cavity as detected in serial sections of gravid uterine segments embedded en bloc (i.e., minimally manipulated). Abnormal yolk sac vasculature indicates disorganization or delayed remodeling based on PECAM-1 immunostaining and TIE2p/e-LacZ staining at E8.75 to E10.5 and lack of blood-filled yolk sac vessels at E12.5. (B and D) Par1 +/–(B) and Par1 –/– (D) littermate E10.5 embryos. Note developmental delay and dilated, blood-filled pericardial cavity (arrow) in the Par1 –/– embryo. (Cand E) Yolk sacs from the embryos shown in (B) and (D), respectively. These embryos carried a TIE2p/e-LacZ transgene to allow visualization of the vascular endothelium. β-Galactosidase–stained yolks sacs are shown. Scale bars, 1 mm.

Why do Par1 –/– embryos bleed? Blood in the exocoelomic and pericardial cavities ofPar1 –/– embryos implies bleeding from extraembryonic vessels and from the embryo proper, suggesting a general defect in hemostatic mechanisms or in the vasculature itself. It is unlikely that defective thrombin signaling in platelets accounts for bleeding in Par1 –/– embryos because (i)Par1 –/– platelets respond normally to thrombin (4), (ii) mice that do have platelets with defective thrombin signaling (13, 14) or that lack platelets (15) develop normally, and (iii) platelets are not present at E9.5 (16). PAR1 is highly expressed in endocardium and endothelium in E9.5 embryos by RNA in situ hybridization (17). However, other cell types in the embryo also express PAR1 (17), and fibroblasts cultured from mouse embryos express sufficient PAR1 to mediate thrombin signaling (4). To better characterize PAR1 expression, we generated PAR1-LacZ “knock-in” mice (18). β-Galactosidase staining of E9.5 PAR1-LacZ knock-in embryos (12, 18) revealed PAR1 expression in endocardium and vascular endothelium in great and small vessels, in occasional round cells adhering to the vessel wall, and in a subset of mesenchymal-appearing cells in the septum transversum and elsewhere (Fig. 2, A to D). Assessment of vascular and cardiac anatomy by light microscopy, whole-mount staining for platelet–endothelial cell adhesion molecule (PECAM-1) (Fig. 2, E and F), and β-galactosidase staining for a LacZ transgene driven by the endothelial-specific TIE2 promoter-enhancer (TIE2p/e-LacZ; see Fig. 1) (19) revealed variable developmental delay in Par1 –/–embryos but no gross vascular malformation. Similarly, the fraction of cells positive for the endothelial marker intercellular cell-adhesion molecule–type 2 (ICAM2) by flow cytometry was not different between E9.5 Par1 +/+ andPar1 –/– embryos, and transmission electron microscopy, immunostaining for smooth muscle α-actin, and β-galactosidase staining for an smooth muscle α-actin promoter-LacZ transgene did not reveal consistent differences. However, analysis of complete serial sections of threePar1 –/– E9.5 embryos with hemopericardium did reveal a breach in the wall of the sinus venosus in two embryos (Fig. 2G). These openings appeared large enough to allow blood cells to enter the pericardial cavity from the intravascular space, and they were unlikely to be artifacts of handling because similar defects were seen in serial sections of uterine segments containing E9.5Par1 –/– embryos embedded en bloc before sectioning. In the extraembryonic vasculature, β-galactosidase staining of yolk sacs from E8.5 Par1–/– embryos carrying the TIE2p/e-LacZ transgene revealed a normal-appearing primary vascular plexus. However, by E9.5, embryos with gross bleeding showed an abnormal yolk sac vascular pattern—usually delayed vascular remodeling (Fig. 1, A, C, and E). Yolk sac vascular abnormalities were seen only in embryos with gross bleeding, and it is possible that they were secondary to bleeding and cardiovascular collapse. However, the defects noted in the walls of great vessels in the embryo proper were almost certainly the source of, rather than the result of, bleeding. Taken together, these observations raised the possibility that loss of PAR1 signaling in endocardial and/or endothelial cells might be the primary defect inPar1 –/– embryos and the direct or indirect cause of bleeding and death.

Figure 2

PAR1 expression andPar1 –/– vessels at E9.5. (A toD) β-Galactosidase stain of unaffected homozygous PAR1-LacZ knock-in embryos. (A and B) Whole-mount β-galactosidase staining. BA, branchial arch; EC, endocardium; ST, septum transversum; IV, intersomitic vessel. (C andD) Histological sections. Note staining of endocardium (EC) and endothelium (EN) of great vessels. EB, embryonic blood. (E and F) PECAM-1 staining of wild-type (E) andPar1 –/– (F) embryos. (G) Section from a Par1 –/– embryo like that in Fig. 1D showing a defect (arrow) in the wall of the sinus venosus (SV) large enough to allow blood cells into the pericardial cavity (PC). Scale bars, 50 μm in (C) and (G); 12 μm in (D).

If loss of PAR1 function in endothelial cells were the primary defect in Par1 –/– embryos, endothelial-specific expression of PAR1 should prevent death ofPar1 –/– embryos. To test this prediction, we generated transgenic mouse lines from two independent founders in which the endothelial-specific TIE2 promoter-enhancer (TIE2p/e) (19) drove mPAR1 expression (18). Transgene expression could be detected as early as E8.0 by the reverse transcriptase–polymerase chain reaction (RT-PCR), and in situ hybridization of E9.5 and 13.5 embryos demonstrated selective expression of the transgene in endothelium and endocardium (Fig. 3A). Fibroblasts fromPar1 –/– mice bearing the transgene did not show thrombin-triggered increases in cytosolic calcium (Fig. 3B)—additional evidence for cell type–specific transgene expression. In crosses designed to yield equal numbers of transgenic and nontransgenic Par1 –/– andPar1 +/– mice, transgene-negativePar1 –/– offspring were generated at approximately half the expected frequency, but transgene-positivePar1 –/– pups were born at a frequency indistinguishable from Par1 heterozygotes (Fig. 3C). Similar results were obtained using the transgenic line from an independent founder. To directly examine the effect of the transgene onPar1 –/– embryos,Par1 +/– mice hemizygous for the transgene were crossed with Par1 –/– mice, and E11.5 to E12.5 embryos were scored as alive or dead and genotyped (Fig. 3D). The death rate associated with PAR1 deficiency was 39% in transgene-negative embryos but only 14% in transgene-positive embryos (P< 0.001 by chi-square test). The live transgene-positivePar1 –/– embryos appeared normal. Thus, in three separate experiments, the TIE2p/e-PAR1 transgene greatly reduced or prevented embryonic death due to PAR1 deficiency. In both TIE2p/e-LacZ mice and TIE2p/e-PAR1 E9.5 embryos, transgene expression was detected only in endothelium and endocardium and in a small number of circulating cells [Fig. 3A and (19)]. Taken together, our results strongly suggest that the death ofPar1 –/– embryos is due to a lack of PAR1 in endothelial cells.

Figure 3

Endothelial expression of PAR1 rescuesPar1 –/– animals from death. (A) In situ hybridization for the expression of the TIE2p/e-PAR1 transgene in an E13.5 hemizygous (Tg+/0) embryo using probe to an SV40 sequence tag (12). Dark-field photomicrograph; scale bar, 250 μm. Note silver grains indicating expression of the transgene overlying endocardium (E) and endothelium of the great vessels (arrows). Similar results were obtained at E9.5. (B) Cytosolic calcium transients in fibroblasts from wild-type,Par1 –/–, and Tg+/0:Par1 –/– mice were measured using the calcium-sensitive fluor Fura-2. Thrombin (αTh; 30 nM) or lysophosphatidic acid (LPA; 5 μM) were added at the indicated times. (C) Genotypes of offspring alive at weaning from Tg+/0:Par1 –/– ×Par1 +/– and Tg+/0:Par1 +/–×Par1 –/– matings. (D)Par1 –/– females were mated with Tg+/0:Par1 +/– males; embryos were scored at E11.5 to E12.5 as dead or alive on the basis of the absence or presence of a heartbeat, then genotyped by Southern analysis. Five resorbed embryos could not be genotyped and were not included in the analysis; they probably belonged to the Tg0/0:Par1 –/– group.

Is thrombin the endogenous activator of endothelial PAR1 during embryonic development, and is PAR1 the major effector of the coagulation cascade in this setting? There are similarities between the phenotypes of Par1 –/– embryos and coagulation factor knockout embryos. Most of tissue factor–deficient embryos (85 to 100%) die at mid-gestation with yolk sac vascular defects (6) and bleeding (5,7). These observations suggested that tissue factor might contribute to blood vessel development and/or hemostasis in the embryo via its ability to activate the coagulation cascade, by direct signaling or, by other unknown mechanisms (5–7). About 50% of embryos deficient in factor V (FV –/–), a cofactor necessary for thrombin production, die between E9.5 and E10.5 with multiple abnormalities, including hemorrhage (8). Likewise, ∼50% of prothrombin-deficient embryos die between E9.5 and E11.5; bleeding was described in one report (9) but not in another (10). Whether the decreased penetrance of the prothrombin and factor V versus the tissue factor phenotype is due to maternal coagulation factors reaching the embryonic circulation or to a thrombin-independent role for tissue factor is unknown. The similar phenotypes of Par1 –/–,FV –/–, and prothrombin-deficient embryos suggested a simple model in which factor V mediates conversion of prothrombin to thrombin that then signals via PAR1 (Fig. 4A). If this model were true, the phenotype resulting from combined deficiency of PAR1 and factor V should be identical to that of either deficiency alone. This was not the case. Comparison of embryos deficient in PAR1 or factor V to those lacking both genes revealed synthetic lethality (Fig. 4B). Only 4% (1/23) of FV –/– Par1 –/– embryos survived to E12.5, a rate markedly less than that of either Par1 –/– orFV –/– embryos (52% or 67% at E12.5, respectively) and significantly less than the 35% survival rate predicted if PAR1 and factor V deficiencies caused death independently. The 22 dead FV –/– Par1 –/– embryos were pale, runted, and severely necrotic with pericardial effusions and blood pooled in their pericardial and exocoelomic cavities. Of note, this phenotype and degree of lethality is similar to that reported for the tissue factor–knockout mouse (5– 7). The synthetic lethality of combined factor V and PAR1 deficiency suggests that the factor V and PAR1 pathways interact and raises intriguing possibilities. The observation that PAR1 deficiency enhanced death ofFV –/– embryos suggests that PAR1 can be activated in the absence of embryonic factor V. Thus, molecules other than factor V might mediate activation of prothrombin and/or molecules other than thrombin might mediate activation of PAR1 during development (Fig. 4C). We cannot conclude this rigorously because factor V from theFV +/– mothers [used by necessity in these experiments (8)] may enter FV –/–embryos and support some thrombin signaling via PAR1. Loss of such signaling would explain how PAR1 deficiency exacerbates the phenotype of FV –/– embryos. Looking at the data from another perspective, the observation that factor V deficiency enhanced death of Par1 –/– embryos clearly implies that factor V has actions beyond activating PAR1 during development (Fig. 4C). Our observations suggest that thrombin and/or another factor V–dependent agonist act on PAR1 and on other target(s) with interacting functions during embryonic development.

Figure 4

Interaction of factor V and PAR1 deficiencies. (A) Role of factor V and PAR1 in development: initial model. A simple linear pathway in which activated factors V and X (Va/Xa) convert prothrombin to thrombin that then activates PAR1. (B)FV +/– Par1 +/– females were mated with FV +/– Par1 –/– males. E11.5 to E12.5 embryos were scored dead or alive and genotyped by Southern analysis. Note virtually complete lethality for doubly deficient embryos (P < 0.001 by chi-square test versus any other group). (C) Revised model. The data shown in (B) suggest PAR1-independent actions of factor V and, possibly, alternative routes to PAR1 activation (see text).

Fibrinogen was one potential other target. Although not necessary for development in otherwise normal embryos (20), fibrinogen might become important in the context of the bleeding diathesis seen inPar1 –/– embryos. However, in a study analogous to that in Fig. 4B, fibrinogen deficiency did not enhance death ofPar1 –/– embryos at E12.5. It is possible that fibrinogen from the Fib +/– mothers [used by necessity in these studies (20)] entered the embryonic circulation in sufficient quantity to prevent a phenotype, but, at face value, this negative result suggests that still other coagulation protease substrates are important for embryonic development. Other PARs are appealing candidates. Indeed, the notion that other PARs might in part compensate for PAR1 deficiency offers one explanation for both the partiality of the PAR1 phenotype and the synthetic lethality seen with PAR1 and factor V deficiencies. This hypothesis can be tested by using mice with multiple PAR deficiencies.

Our results show that PAR1 function in endothelium and/or endocardium contributes to mouse embryonic development and highlight a role for the coagulation system in directly regulating endothelial cell function during blood-vessel formation. Loss of PAR1 signaling in endothelial cells probably explains, at least in part, death at midgestation caused by knockout of tissue factor (5–7), factor V (8), and prothrombin (9, 10). Bleeding in Par1 –/– embryos might reflect a role for the coagulation cascade and endothelial PAR1 in sensing occasional “spontaneous” breaks in developing blood vessels and mounting an acute hemostatic response. Because platelets are not present at midgestation and because fibrinogen deficiency did not exacerbate bleeding in Par1 –/– embryos, this model raises the possibility of an endothelial-dependent, platelet- and fibrinogen-independent mechanism for effecting hemostatic responses and deserves further exploration. Our results are equally consistent with models in which the coagulation cascade and endothelial PAR1 signaling contribute to vascular development and integrity in ways unrelated to acute hemostatic responses. The coagulation cascade and PARs may allow nascent vessels to sense permeability to plasma proteins or allow endothelial and/or endocardial cells to monitor their connection to the circulation; roles in regulation of vascular remodeling and/or endothelial-mesenchymal transformation are easily imagined. In addition, ischemia can increase endothelial permeability and can induce parenchymal tissue factor expression in the adult (21), raising the possibility that the coagulation cascade and PARs might provide one means for developing vessels to sense and respond to inadequate tissue perfusion. In culture, thrombin causes cultured endothelial cells to change shape, disrupts cell-cell junctions (22), and increases the permeability of endothelial cell monolayers (23). Thrombin also enhances endothelial responsiveness to vascular endothelial growth factor (24) and fibroblast growth factor (25) and triggers secretion of extracellular matrix proteins (26) and other growth factors such as platelet-derived growth factor (27). Thus PAR1 activation might contribute to vascular development by regulating endothelial cell shape, migration, proliferation, and/or interactions with perivascular cells and extracellular matrix.

  • * These authors contributed equally to this work.

  • To whom correspondence should be addressed. E-mail: coughlin{at}


Stay Connected to Science

Navigate This Article