Anaphylactic shock depends on endothelial G_q/G₁₁

To analyze the role of G_q/G₁₁- and G₁₂/G₁₃-mediated signaling in endothelial responses to vasoactive mediators, we generated ECs lacking the α subunits of G_q/G₁₁ or G₁₂/G₁₃. We have previously generated floxed alleles of the genes encoding Gα_q (Gnaq) and Gα₁₃ (Gna13) which allow the conditional inactivation of these genes in Gα₁₁- or Gα₁₂-deficient backgrounds (29, 30). To induce Gα_q/Gα₁₁ or Gα₁₂/Gα₁₃ double deficiency, we prepared pulmonary microvascular ECs from WT, Gnaq^flox/flox;Gna11^−/−, and Gna12^−/−;Gna13^flox/flox mice and infected them with an adenovirus transducing the recombinase Cre. As shown in Fig. 1 A, expression of Cre recombinase in Gnaq^flox/flox;Gna11^−/− or Gna12^−/−;Gna13^flox/flox ECs resulted in Gα_q/Gα₁₁ and Gα₁₂/Gα₁₃ deficiency, respectively.

We then analyzed the role of G_q/G₁₁- and G₁₂/G₁₃-mediated signaling in the regulation of endothelial NO formation by known endothelial stimuli acting via GPCRs. To determine NO-dependent activation of guanylyl cyclase, we performed a transfer bioassay in which (cyclic guanosine monophosphate) cGMP levels were determined in RFL6 fibroblasts incubated with WT, Gα_q/Gα₁₁-deficient, or Gα₁₂/Gα₁₃-deficient lung ECs treated without or with thrombin, PAF, or ionomycin (Fig. 1 B). Although thrombin and PAF induced a significant increase in cGMP levels in cocultures containing WT and Gα₁₂/Gα₁₃-deficient ECs, the effects in cocultures containing Gα_q/Gα₁₁-deficient ECs were strongly reduced. None of the stimuli induced guanylyl cyclase activation when added to RFL6 fibroblasts or ECs alone (unpublished data). The effect of ionomycin was not affected by Gα_q/Gα₁₁ or Gα₁₂/Gα₁₃ deficiency in ECs. This indicates that G_q/G₁₁, but not G₁₂/G₁₃, are critically involved in thrombin- and PAF-induced NO-dependent stimulation of guanylyl cyclase activity.

Because the phosphorylation state of the myosin light chain (MLC) is a critical determinant of endothelial contractility, we analyzed the effect of thrombin on MLC phosphorylation in WT, Gα_q/Gα₁₁-deficient, and Gα₁₂/Gα₁₃-deficient ECs. As shown in Fig. 1 (C and E), thrombin induced a rapid increase in MLC phosphorylation that was maximal after ∼3 min, whereas thrombin had no effect on MLC phosphorylation in ECs lacking Gα_q/Gα₁₁. The defect of thrombin-induced MLC phosphorylation in Gα_q/Gα₁₁-deficient cells could be rescued by adenovirus-mediated expression of Gα_q (Fig. 1 D). Lack of Gα₁₂/Gα₁₃ did not completely block thrombin-induced MLC phosphorylation but led to a reduced and more transient response to thrombin. Interestingly, the abrogation of thrombin-induced MLC phosphorylation in cells lacking Gα_q/Gα₁₁ was not accompanied by any defect in thrombin-induced RhoA activation, whereas thrombin-induced RhoA activation was abrogated in ECs lacking Gα₁₂/Gα₁₃ (Fig. 1 F).

For in vivo experiments, we restricted Gα_q/Gα₁₁ and Gα₁₂/Gα₁₃ double deficiency to ECs by using a bacterial artificial chromosome (BAC) transgenic mouse line that expresses a fusion protein of the Cre recombinase with the modified estrogen receptor binding domain (CreER^T2) (31) under the control of the tie2 promoter (see Materials and methods). The inducible endothelium-specific Cre transgenic mouse line (tie2-CreER^T2) did not show any Cre activity in the absence of tamoxifen when crossed with the Gt(ROSA)26Sor Cre reporter mouse line (Fig. 2 A). However, after treatment of animals with tamoxifen, ECs showed Cre-mediated recombination, indicating that Cre had been activated with high efficacy. Cre-mediated recombination was exclusively observed in ECs of various organs (Fig. 2 A). The lack of Gα_q/Gα₁₁ and Gα₁₂/Gα₁₃ in ECs of tamoxifen-treated tie2-CreER^T2;Gnaq^flox/flox;Gna11^−/− (EC-Gα_q/Gα₁₁-KO) and tie2-CreER^T2;Gna12^−/−;Gna13^flox/flox (EC-Gα₁₂/Gα₁₃-KO) mice was verified by Western blotting of pulmonary EC lysates from the respective mouse lines (Fig. 2 B). Western blot analysis of platelets, leukocytes, and vascular smooth muscle cells showed no difference between WT and EC-Gα_q/Gα₁₁-KO mice with regard to Gα_q/Gα₁₁ expression (Fig. S1, available).

We then analyzed the effect of various vasoactive substances on the vascular permeability in EC-Gα_q/Gα₁₁-KO and EC-Gα₁₂/Gα₁₃-KO mice. In the absence of any intradermal injection, the vascular leakage of Evans blue given i.v. was negligible (unpublished data). Intradermal injection of lysophosphatidic acid (LPA), the protease-activated receptor 1 (PAR-1)–activating peptide SFLLRN-NH₂, histamine, PAF, and leukotriene C₄ each induced a dose-dependent increase in the leakage of Evans blue dye (Fig. 3, A and B; Fig. S2, available). In addition, intradermal injection of control buffer resulted in a small extravasation of Evans blue that was significantly smaller than the one seen in response to the vasoactive stimuli, suggesting that the manipulation resulted in the local release or production of some active mediators. Both basal vascular permeability and stimulus-induced increases in vascular permeability were severely reduced in mice with endothelial-specific Gα_q/Gα₁₁ deficiency but not in mice lacking Gα₁₂/Gα₁₃ in ECs. The small remaining response to the PAR1-activating peptide observed in EC-Gα_q/Gα₁₁-KO mice was not further reduced in mice lacking both Gα_q/Gα₁₁ and Gα₁₂/Gα₁₃ in ECs (Fig. 3 B). To test the regulation of the endothelial barrier in a more complex model of local anaphylaxis, we sensitized mice by intradermal injection of anti-DNP IgE antibodies and subsequently injected DNP–human serum albumin (HSA) systemically. In addition, in this IgE-mediated model of local anaphylaxis opening of the endothelial barrier was not significantly affected in EC-Gα₁₂/Gα₁₃-KO mice, whereas mice with endothelium-specific Gα_q/Gα₁₁ deficiency showed strongly reduced vascular permeability (Fig. 3 C). Thus, local regulation of vascular permeability requires G_q/G₁₁-mediated signaling in ECs but not G₁₂/G₁₃.

i.v. injection of histamine induced a rapid and transient drop in the systolic blood pressure to levels of ∼50 mmHg in WT mice (Fig. 4 A). Normal values were restored ∼90 min after the application of histamine. In EC-Gα_q/Gα₁₁-KO mice, the same dose decreased blood pressure for only ∼20 min with maximal hypotensive values of ∼90 mmHg, whereas mice with endothelium-specific Gα₁₂/Gα₁₃ deficiency responded comparable to WT mice (Fig. 4 A). The strongly reduced hypotensive response of EC-Gα_q/Gα₁₁-KO mice to histamine was not caused by a general defect in the regulation of the vascular tone, as is indicated by the indistinguishable response of WT, EC-Gα_q/Gα₁₁-KO, and EC-Gα₁₂/Gα₁₃-KO mice to the NO-donor sodium nitroprusside as well as to the NO synthase (NOS) inhibitor N^ω-nitro-l-arginine methyl ester (l-NAME; Fig. 4, B and C).

We then tested the effect of endothelium-specific Gα_q/Gα₁₁ and Gα₁₂/Gα₁₃ deficiency on the systemic response to PAF, which is thought to be a critical mediator of anaphylactic shock (32–34). i.v. injection of PAF induced severe hypothermia (Fig. 4 D) and resulted in the death of WT and EC-Gα₁₂/Gα₁₃-KO mice within 20 min (Fig. 4 E). However, mice with endothelial Gα_q/Gα₁₁ deficiency were protected from PAF-induced shock, and all of the animals assessed survived the injection of PAF with only a transient drop in body temperature (Fig. 4, D and E). Mice lacking only Gα₁₁ demonstrated an intermediate phenotype with more severe hypothermia than EC-Gα_q/Gα₁₁-KO mice and a survival rate of only 25% (two of eight tested animals; unpublished data). Interestingly, the intraperitoneal injection of the endotoxin LPS induced a severe hypotension and eventual lethality in WT and EC-Gα₁₂/Gα₁₃-KO as well as in EC-Gα_q/Gα₁₁-KO mice (Fig. 4 and not depicted). Thus, endothelial Gα_q/Gα₁₁ deficiency does not protect from endotoxic shock.

To further evaluate the role of endothelial G protein–mediated signaling pathways under pathophysiologically more relevant conditions, we set up models for passive and active systemic anaphylaxis. To test the role of endothelial G_q/G₁₁ and G₁₂/G₁₃ in passive systemic IgE-dependent anaphylaxis, we injected WT and EC-Gα_q/Gα₁₁-KO and EC-Gα₁₂/Gα₁₃-KO mice i.v. with anti-DNP IgE and challenged them 24 h later with DNP-HSA. As shown in Fig. 5 A, WT and EC-Gα₁₂/Gα₁₃-KO mice responded with a rapid drop in systolic blood pressure down to values of ∼35 mmHg. After a few minutes, the blood pressure started to slowly rise but remained hypotensive for more than 90 min. Both lines also showed a strong increase in their hematocrit when determined 10 min after application of the allergen as an indicator of severe extravasation of plasma (Fig. 5 B). Under the same conditions, mice with endothelial lack of Gα_q/Gα₁₁ showed only a small and very transient reduction in blood pressure, and the hematocrit of EC-Gα_q/Gα₁₁-KO mice remained unchanged after allergen administration (Fig. 5, A and B).

We then actively sensitized mice with BSA together with adjuvant. 2 wk later, mice were challenged with an i.v. injection of the same allergen. Within minutes after this challenge, all mice developed severe hypothermia (Fig. 5 C), and WT and EC-Gα₁₂/Gα₁₃-KO mice died within 20 min (Fig. 5 D). However, mice with endothelium-specific Gα_q/Gα₁₁ deficiency recovered from hypothermia after ∼1 h, and all of the tested animals (n = 5) survived the anaphylactic challenge (Fig. 5, C and D). Mice lacking only Gα₁₁ exhibited an intermediate phenotype in the active systemic anaphylaxis model showing a survival rate of 20% (2 of 10 animals; unpublished data).

The pathological processes induced by mediators of anaphylaxis involve diverse organs such as the bronchial and immune systems, blood vessels, or the heart and require complex cell–cell and mediator–mediator interactions which involve various signaling pathways (5, 35, 36). In this study, we addressed the role of defined endothelial G protein–mediated signaling pathways in the pathomechanism of systemic anaphylaxis. We report here that the endothelium-specific ablation of G_q/G₁₁ prevents the loss of the endothelial barrier function induced by various inflammatory mediators as well as by local anaphylaxis. The systemic effects of anaphylactic mediators like histamine and PAF as well as of IgE-mediated passive anaphylaxis were blunted in EC-Gα_q/Gα₁₁-KO mice, and mice with endothelium-specific Gα_q/Gα₁₁ deficiency, but not with Gα₁₂/Gα₁₃ deficiency, were protected against the fatal consequences of active systemic anaphylaxis. Thus, the blockade of endothelial G_q/G₁₁ signaling is sufficient to protect against fatal anaphylactic shock, indicating that endothelial G_q/G₁₁-mediated signaling is critically involved in local and systemic anaphylactic reactions. In contrast, endothelial G_q/G₁₁ does not appear to play a role in septic shock as the degree of hypotension and the lethality after systemic administration of LPS was indistinguishable between WT and EC-Gα_q/Gα₁₁-KO mice.

The analysis of the role of G_q/G₁₁- and G₁₂/G₁₃-mediated signaling pathways in the adult endothelium under in vivo conditions has been hampered by the fact that mice lacking the α subunits of these G proteins are embryonic lethal (37–39). By crossing a newly generated inducible and endothelium-specific Cre transgenic mouse line with conditional and null alleles of the genes encoding Gα_q/Gα₁₁ and Gα₁₂/Gα₁₃, we were able to study the role of G_q/G₁₁ and G₁₂/G₁₃ in the endothelium of adult animals in which lack of Gα_q/Gα₁₁ or Gα₁₂/Gα₁₃ did not lead to any obvious defects. There was also no acute or delayed change in the systemic blood pressure after induction of endothelial Gα_q/Gα₁₁ or Gα₁₂/Gα₁₃ deficiency (unpublished data). At the same time, the short and transient drop in blood pressure induced by i.v. injection of histamine was strongly reduced in EC-Gα_q/Gα₁₁-KO mice, indicating that pharmacological responses were affected. Thus, although endothelial G_q/G₁₁ and G₁₂/G₁₃ are obviously not critically involved in the regulation of vascular functions under basal physiological conditions, G_q/G₁₁-mediated signaling plays a crucial role in the regulation of endothelial functions under inflammatory and anaphylactic conditions. Studies in mice lacking Gα₁₃ have indicated a critical role of endothelial G₁₃ in embryonic angiogenesis (38, 40). Female EC-Gα₁₂/Gα₁₃-KO mice are fertile, and we have not observed any defects in wound healing suggesting that endothelial G₁₂/G₁₃ are not required for adult angiogenesis in the female reproductive system or during wound healing. However, the potential role of G₁₃ in tumor angiogenesis remains to be evaluated.

The stimulation of endothelial permeability by inflammatory and anaphylactic mediators like thrombin, bradykinin, histamine, PAF, etc. requires the retraction of ECs as a result of increased actomyosin-mediated contraction as well as the disruption of cell–cell contacts (41, 42). Endothelial contraction is regulated by the phosphorylation state of the MLC which in its phosphorylated form allows myosin to interact with actin and to generate contractile forces (43, 44). Analogous to the situation in smooth muscle cells (45–47), the dual regulation of MLC phosphorylation in ECs via the Ca²⁺-dependent MLC kinase activation and the Rho/Rho kinase–mediated myosin phosphatase inhibition is believed to be initiated by the dual coupling of receptors to G_q/G₁₁ and G₁₂/G₁₃, respectively (44). Our in vitro studies using Gα_q/Gα₁₁- and Gα₁₂/Gα₁₃-deficient pulmonary ECs indicate that thrombin-induced MLC phosphorylation is abrogated in the absence of G_q/G₁₁, a defect which can be rescued by transfection of cells with Gα_q, whereas RhoA activation by thrombin was not affected in Gα_q/Gα₁₁-deficient ECs. In cells lacking G₁₂/G₁₃, MLC phosphorylation in response to thrombin was only reduced and RhoA activation was blocked. This indicates that the G₁₂/G₁₃-RhoA–mediated signaling pathway plays only a minor role in thrombin-induced MLC phosphorylation in primary pulmonary ECs. This is consistent with our in vivo data, which show that endothelial Gα₁₂/Gα₁₃ deficiency has no effect on vascular leakage induced by thrombin, PAF, histamine, or anaphylactic reactions, whereas Gα_q/Gα₁₁ deficiency blocked these effects. A predominant role of G_q/G₁₁ compared with G₁₂/G₁₃ was recently also demonstrated for thrombin-induced increase in endothelial permeability analyzed in vitro (48). Thus, G_q/G₁₁-mediated signaling, rather than G₁₂/G₁₃, is critically involved in the regulation of endothelial barrier function by inflammatory mediators acting via GPCRs.

The role of NO in systemic anaphylaxis has been controversial (49, 50). Recently, it was shown that the systemic inhibition of NOSs prevented mortality in various models of anaphylaxis in mice (51). This effect could also be seen in mice lacking the endothelial NOS (eNOS) but not the inducible NOS (iNOS). Although eNOS is expressed in ECs, it can also be found in various other tissues, and it has been suggested that it is the NO production in non-ECs which is involved in anaphylaxis (52). Our data indicate that the stimulation of NO formation in isolated ECs depends on G_q/G₁₁ but not on G₁₂/G₁₃. In addition, endothelium-specific lack of G_q/G₁₁ results in a strong reduction in histamine-induced hypotension and various anaphylactic reactions very similar to the effects seen in mice lacking eNOS (51, 53). Thus, our data are consistent with a primary role of endothelial NOS in systemic anaphylaxis.

Using conditional mutagenesis, we have generated mice with inducible endothelium-specific Gα_q/Gα₁₁ or Gα₁₂/Gα₁₃ deficiency. When challenged with anaphylactic mediators or subjected to systemic anaphylaxis, EC-Gα_q/Gα₁₁-KO mice were protected, whereas mice with endothelium-specific Gα₁₂/Gα₁₃ deficiency responded like WT animals. Endothelial Gα_q/Gα₁₁ deficiency blocked MLC phosphorylation and NO formation as well as increases in vascular permeability induced by various inflammatory and anaphylactic mediators. This study identifies endothelial G_q/G₁₁-mediated signaling as a critical process in the pathophysiology of systemic anaphylaxis. Because lack of G_q/G₁₁-mediated signaling does not affect basal physiological regulation of endothelial function, it may be an interesting target to treat systemic anaphylaxis.

For Western blotting, the following antibodies were used: anti-Gα_q/11 and anti-Gα₁₃ (Santa Cruz Biotechnology, Inc.), anti–α-tubulin and anti-MLC (Sigma-Aldrich), and anti-pMLC (Cell Signaling Technology). Histamine, thrombin, PAF, LPA, PAR-1 peptide (SFLLRN-NH₂), Evans blue, anti–DNP-IgE, DNP-HSA, and BSA were obtained from Sigma-Aldrich. Ionomycin was obtained from Invitrogen.

All procedures of animal care and use in this study were approved by the local animal ethics committee (Regierungspräsidium Karlsruhe, Germany). The generation of floxed alleles of the genes encoding Gα_q (Gnaq) and Gα₁₃ (Gna13) and of null alleles of the genes encoding Gα₁₁ (Gna11) and Gα₁₂ (Gna12) have been described previously (29, 30, 37, 39).

To generate an inducible EC-specific Cre transgenic mouse line, a cassette consisting of the CreER^T2 followed by a polyadenylation signal from bovine growth hormone and a module containing the β-lactamase gene flanked by frt sites was introduced into the coding ATG of the mouse tie2 gene carried by a BAC using ET recombination as previously described (54–56). Correct recombinants were verified by Southern blotting. After FLPe-mediated recombination, the recombined BAC was injected into male pronuclei derived from fertilized FvB/N oozytes. Transgenic offspring were analyzed for BAC insertion by genomic PCR amplification. To verify inducibility and activity of the Cre fusion protein, tie2-CreER^T2 mice were mated with animals of the Cre reporter transgenic line Gt(ROSA)26Sortm1sor (ROSA26-LacZ). Cotransgenic progeny from these matings were treated i.p. with 5 × 1 mg/d tamoxifen or vehicle alone and were killed 14 d after induction. For histological analysis of β-galactosidase activity, staining was performed on 10–12-µm cryosections followed by eosin counterstaining.

Mouse lung ECs were isolated as described previously (57). Lungs were minced and digested in 50 U/ml dispase for 1 h at 37°C with shaking (350 rpm). After filtration, the cells were washed in PBS containing 0.5% BSA. Cells were incubated with anti-CD144 antibody-coated (BD) magnetic beads (Invitrogen) for 1 h at room temperature, washed, and isolated with a magnet (Invitrogen). Cells were grown in DMEM/F12 (Invitrogen) supplemented with 10% FBS, penicillin/streptomycin, and EC growth supplement with heparin (PromoCell) on fibronectin-coated wells. To induce Cre-mediated recombination or to express Gα_q, the cells were infected with 5 × 10⁷ PFU of Adeno-Cre-GFP virus (Vector Laboratories) or Adeno-Gα_q virus (58, 59) 72 h before the experiments.

RhoA activation in primary ECs was detected by a luminescence-based G-LISA RhoA activation assay kit (tebu-bio) according to the manufacturer’s instructions. Briefly, mouse primary lung ECs were grown on 12-well plates and stimulated with 1 U/ml thrombin for 1 min, washed with 1.5 ml of ice-cold PBS, and lysed in 150 µl of lysis buffer on ice. Protein concentrations were measured and equalized with lysis buffer if necessary.

For detection of MLC phosphorylation, mouse primary ECs were cultured on 24-well plates. The cells were stimulated with 1 U/ml thrombin for the indicated time periods and lysed in 2× Laemmli buffer, incubated for 10 min at 100°C, and then loaded on 12% SDS PAGE gels. MLC phosphorylation was detected by Western blotting using an anti-pMLC antibody (1:1,000).

NO formation was determined as previously described (60). Lung ECs from WT, Gnaq^flox/flox;Gna11^−/−, or Gna12^−/−;Gna13^flox/flox mice were cultured and treated with Cre transducing adenovirus as described. Cells were then suspended by treatment with accutase (PAA Laboratories) and washed in Hepes-buffered Tyrode solution containing 0.1 mM of the phosphodiesterase inhibitor 3-isobutyl-1-methylxanthine and 100 U/ml of superoxide dismutase. Approximately 5 × 10⁴ cells were added to RFL6 fibroblasts cultured in 24-well plates and incubated (37°C) for 5 min in the absence or presence of the indicated stimuli. Thereafter, the incubation was stopped by the addition of trichloroacetic acid (6%), and the concentration of cyclic GMP was determined by a radioimmunoassay (GE Healthcare).

We determined agonist-induced vascular permeability changes using Evans blue dye. Mice were anaesthetized with 50 mg/kg pentobarbital sodium and injected i.v. with 0.04 µg/g Evans Blue in saline. After 1 min, different doses of agonists (histamine, PAF, PAR-1 peptide, and LPA) in 20 µl of PBS were injected into the shaved back skin. PBS was injected as control. Mice were killed after 10 min, and ∼1 cm² of skin containing the site of injection was removed. The skin punches were incubated in 500 µl of formamide at 55°C for 48 h, and the Evans blue content was determined by absorption at 595 nm.

30 ng of anti-DNP IgE in 20 µl of sterile 0.9% NaCl was injected into the dorsal skin of the right ear. The left ear of mice received an equal volume of saline and served as control. After 24 h, we challenged the passively immunized mice by an i.v. injection of 0.5 mg of DNP-HSA together with 0.08 µg/g Evans blue in saline. Mice were killed 30 min after the challenge by cervical dislocation, and ear biopsies were collected. Evans blue was extracted in 400 µl of formamide at 55°C for 24 h and quantified by measuring light absorption at 595 nm.

We used a radiotelemetry system (PA-C10; Data Sciences International) to monitor blood pressure in conscious unrestrained mice, as described previously (46). Pressure sensing catheters were implanted into the left carotid artery, and the transducer unit was inserted into a subcutaneous pouch along the right flank. After a recovery period of at least 1 wk, arterial pressure recordings were collected, stored, and analyzed with Dataquest A.R.T. software (version 4.0; Data Sciences International). We collected data for basal blood pressure measurements with a 10-s scheduled sampling every 5 min and used the 24-h mean values for analysis. For analyzing the acute effects of agonists, we collected data continuously in 5-s intervals for different time periods as indicated in the figures. The body temperature was measured with a temperature control module (TKM-0902; Föhr Medical Instruments GmbH).

To induce passive systemic anaphylaxis, we injected mice i.v. with 20 µg of anti-DNP IgE. After 24 h, we challenged these passively immunized mice by an i.v. injection of 1 mg DNP-HSA. Control mice were injected with saline and challenged as described for immunized mice. For determining hematocrit, blood samples were collected before and 10 min after the challenge. Blood pressure measurements were done using the telemetric system.

For inducing active systemic anaphylaxis, we first immunized mice with i.p. injection of 1 mg BSA and 300 ng pertussis toxin as adjuvant in pyrogen-free 0.9% NaCl. After 14 d, mice were challenged with i.v. injection of 2 mg BSA. We monitored the body temperature and survival of the mice for 120 min after the challenge.

Fig. S1 shows that the expression of Gα_q/Gα₁₁ did not differ between various non-ECs prepared from WT and EC-Gα_q/Gα₁₁-KO mice. Fig. S2 shows the effect of leukotriene C₄ on the extravasation of Evan’s blue in WT and the absence of this effect in EC-Gα_q/Gα₁₁-KO.

We thank Manuela Ritzal, Melanie Bernhard, and Isabel Winter for expert technical assistance and Rose LeFaucheur for excellent secretarial help.

T. Wieland was supported by the German Research Foundation (TR/SFB 23), and N. Wettschureck and S. Offermanns were supported by the German Research Foundation (SFB 405).

The authors have no conflicting financial interests.