Skip main navigation

Interaction Between ALK1 Signaling and Connexin40 in the Development of Arteriovenous Malformations

Originally published, Thrombosis, and Vascular Biology. 2016;36:707–717



To determine the role of Gja5 that encodes for the gap junction protein connexin40 in the generation of arteriovenous malformations in the hereditary hemorrhagic telangiectasia type 2 (HHT2) mouse model.

Approach and Results—

We identified GJA5 as a target gene of the bone morphogenetic protein-9/activin receptor-like kinase 1 signaling pathway in human aortic endothelial cells and importantly found that connexin40 levels were particularly low in a small group of patients with HHT2. We next took advantage of the Acvrl1+/− mutant mice that develop lesions similar to those in patients with HHT2 and generated Acvrl1+/−; Gja5EGFP/+ mice. Gja5 haploinsufficiency led to vasodilation of the arteries and rarefaction of the capillary bed in Acvrl1+/− mice. At the molecular level, we found that reduced Gja5 in Acvrl1+/− mice stimulated the production of reactive oxygen species, an important mediator of vessel remodeling. To normalize the altered hemodynamic forces in Acvrl1+/−; Gja5EGFP/+ mice, capillaries formed transient arteriovenous shunts that could develop into large malformations when exposed to environmental insults.


We identified GJA5 as a potential modifier gene for HHT2. Our findings demonstrate that Acvrl1 haploinsufficiency combined with the effects of modifier genes that regulate vessel caliber is responsible for the heterogeneity and severity of the disease. The mouse models of HHT have led to the proposal that 3 events—heterozygosity, loss of heterozygosity, and angiogenic stimulation—are necessary for arteriovenous malformation formation. Here, we present a novel 3-step model in which pathological vessel caliber and consequent altered blood flow are necessary events for arteriovenous malformation development.


Hereditary hemorrhagic telangiectasia (HHT) is an autosomal dominant genetic vascular disease that affects 1 in 5000 individuals worldwide.1 The abnormal vascular structures in HHT result predominantly from mutations in ENG (HHT1)2 or ACVRL1 (HHT2).3 The protein products of these genes are receptors for transforming growth factor-β and bone morphogenetic protein (BMP) expressed in endothelial cells that share functions in signaling.4,5 Mutations identified to date represent null alleles that lead to reduced levels of receptors at the cell surface, indicating that haploinsufficiency is the predominant underlying mechanism of HHT.6 Both receptors signal to the downstream effectors Smad1/5/8, second messengers that translocate to the nucleus after activation.6 Therefore, the primary cause of HHT is considered defective transforming growth factor-β/BMP signaling in endothelial cells that may lead to the abnormal vasculature.1

Clinical manifestations of HHT are evident as multiple vessel abnormalities known as telangiectases in the nose, mouth, and gastrointestinal tract. These lesions exhibit focal dilation of postcapillary veins that are susceptible to rupture and hemorrhage because of weak vessel walls and high perfusion pressure. As consequence, recurrent and severe epistaxis and gastrointestinal bleeding are common presentation of the disease; this leads to severe anemia requiring iron supplementation and blood transfusions.7 Large arteriovenous malformations (AVMs) are also found in major organs. They can cause life-threatening complications, although the majority of AVMs remain asymptomatic.7 AVMs are arteries and veins that seem to fuse without intervening capillaries to form a network of direct high-flow arteriovenous shunts.8 They form at the interface between arteries and veins where the capillary bed normally lies and are thought to arise from smaller lesions, such as telangiectasia by progressive vascular remodeling.9 Typically present at birth in the brain, they may develop and grow over time in the lung and liver, although there is still little direct evidence to support this idea.7 Nevertheless, the recent development of mouse models for HHT and intravital imaging technologies have provided important insights into the mechanisms of AVM formation.10 Heterozygous Eng+/− and Acvrl1+/− mice, which are the closest animal models of HHT in terms of genotype, surprisingly develop a relatively normal vasculature with no major defects during developmental angiogenesis. However, some vascular lesions appear in these mice but only at low frequency and in an unpredictable manner. This suggests that additional triggers are needed for AVM development.1113 Local homozygous loss of Eng or Acvrl1 gene expression, neoangiogenesis, inflammation, and wounding have been implicated triggering arteriovenous shunt formation, in accordance with secondary triggers acting as underlying mechanisms. Ectopic expression of vascular endothelial growth factor, the prime angiogenic growth factor, using adeno-associated viruses has been shown to induce cerebrovascular dysplasia in both HHT1 and HHT2 mouse models.14,15 Inflammation induces endoglin protein null locally16 that may increase the risk of vascular abnormalities in Eng+/− mice.13 Postnatal homologous loss of the Eng or Acvrl1 gene in endothelial cells leads to the formation of arteriovenous shunts resembling those seen in HHT individuals only in sites where angiogenesis is active, supporting an hypothesis that at least 3 hits—the loss of both Eng and Acvrl1 alleles combined with environmental proangiogenic triggers—are necessary for AVM development.10

How mutations in the Eng and Acvrl1 genes lead to AVM formation is still poorly understood, although recent findings indicate that aberrant angiogenesis may account for the development of such vessel abnormalities.10 Although the initial stages of AVM formation occur irrespective of blood flow, this process is further exacerbated by flow.1719 High-velocity turbulent arterial blood flow results in dilatation and tortuosity of the downstream veins in the skin of mice harboring homologous deletion of Acvrl117 and promotes mural cell coverage of AVMs in Eng-iKO mice.18 In zebrafish, activin receptor-like kinase 1 (ALK1) acts downstream of blood flow to limit the number of endothelial cells maintaining the vessel caliber. In agreement, arteries of zebrafish harboring alk1 mutations deliver a greater blood volume to the downstream vessels that in turn adapt by enlarging and retaining arteriovenous connections that are normally transient during angiogenesis to normalize hemodynamic forces. This vessel remodeling seems to represent a normal adaptive response to increased blood flow.19 It is not known how ALK1 regulates arterial vessel caliber. Two flow responsive genes, cxcr4a and edn1, have been proposed to act downstream of ALK1 to control vessel diameter. These genes encode a proangiogenic chemokine receptor and a vasoconstrictive peptide, respectively, although additional experiments are required to establish their functions during AVM development in HHT.19,20 Here, we describe cooperation between ALK1 and connexin40 (Cx40) in the regulation of blood vessel caliber. The study revealed that reduced expression of Cx40 results in enlargement of the arterial vessels in HHT2 mice and that consequent altered blood flow precipitates flow-dependent adaptive responses involving rarefaction of the capillary network and the formation of direct arteriovenous connections. This cooperation is sufficient to trigger arteriovenous shunt formation during active angiogenesis and on additional environmental insult, which resembles vascular lesions seen in patients with HHT. Our data suggest that GJA5 might be a genetic modifier in HHT2.

Materials and Methods

Materials and Methods are available in the online-only Data Supplement.


BMP9/ALK1 Regulates Endothelial Cx40 Expression

Both circumferential strain and wall shear stress affect endothelial gene expression, so that these mechanical forces can be transduced to biochemical signals that facilitate adaptation to changes in blood flow. As ALK1 expression requires blood flow, it is reasonable to assume that this receptor might lie in a mechanotransduction pathway either upstream or downstream of known mechanoresponsive genes. However, these genes need to be identified. Here, we revealed that BMP9 stimulation of human arterial endothelial cells not only induced expression of ID1 and HEY2 (Figure 1A and 1B), common downstream targets of ALK1 and Notch signaling pathways,21 but also strongly stimulated the expression of GJA5 to levels ≈20-fold higher than the untreated cells after 24 hours of growth factor addition (Figure 1C). BMP9 has been shown to activate ALK1-inducing Smad1/5 phosphorylation in endothelial cells.5 To examine whether ALK1 controls GJA5 expression, we analyzed the effect of siRNA-mediated knockdown of ALK1 on BMP9-induced GJA5 in human arterial endothelial cells. siRNA-mediated downregulation of ALK1 expression was confirmed by quantitative polymerase chain reaction (Figure 1D). We validated that Smad1/5 phosphorylation was reduced (Figure 1E) and importantly found that GJA5 mRNA expression was blocked (Figure 1F) when ALK1 was decreased in endothelial cells stimulated by BMP9. Among the endothelial connexins, Cx40, which is highly expressed in arterial vessels, is essential for the effective transduction of vasodilatation.2225 In sections of human skin biopsies stained for Cx40 and platelet-endothelial cell adhesion molecule-1 as a marker of endothelial cells (Figure 1G), we compared 5 control samples with samples isolated from 4 individuals with HHT2. All of these patients had severe HHT-related nosebleeds: the recurrent epistaxis (often 4 incidents per day) did not improve with the regular argon plasma treatment or medication-like tranexamine acid and N-acetylcysteine. Hemoglobin levels were generally low: for 1 patient, for example, they varied from 4.4 to 8.0 mmol/L for a 5-year period, and 56 blood transfusions with monthly iron transfusions were necessary to maintain normal blood levels. This patient was also treated with thalidomide with benefit but stopped because of neuropathy side effects.26 Another had recurrent epistaxis particularly at night and was also diagnosed with atrial fibrillation. All 4 patients underwent a Saunders procedure in which their nasal epithelium was partially replaced by skin of their upper arm because of the severity of their symptoms. The surplus skin from the Saunders surgery was frozen and sectioned for analysis of Cx40 expression by immunofluorescence. Cx40 intensity levels were defined as the ratio of the Cx40-integrated intensity to the vessel surface measured as platelet-endothelial cell adhesion molecule-1–positive pixels. Cx40 protein levels were particularly low in HHT2 individuals compared with control biopsies (Figure 1H), suggesting that the downregulation of Cx40 expression levels in HHT2 individuals is most probably caused by the fact that the remaining wild-type ACVRL1 allele is unable to contribute protein for normal vascular functions. Our data support an association between ALK1 signaling and Cx40 expression in vivo.

Figure 1.

Figure 1. Connexin40 (Cx40) is a target gene of the bone morphogenetic protein-9 (BMP9)/activin receptor-like kinase 1 (ALK1) signaling pathway. AC, Effect of BMP9 stimulation on ID1 (A), HEY2 (B), and GJA5 (C) mRNA expression in human arterial endothelial cells as determined by quantitative polymerase chain reaction (PCR). Results are representative of 4 independent experiments. D, Expression of ALK1 in human arterial endothelial cells transfected with siRNA scramble or siRNA ACVRL1 was analyzed by quantitative PCR. E, ALK1 downregulation inhibits Smad1/5 phosphorylation induced by BMP9. Human arterial endothelial cells transfected with siRNA scramble or siRNA ACVRL1 were stimulated with 1 ng/mL of BMP9 for 45 minutes at 37°C before lysis. Whole-cell extracts were fractionated by sulfate polyacrylamide gel electrophoresis (SDS-PAGE) and blotted. The filter was incubated with pSMAD1/5, SMAD1/5, or GAPDH antibody. F, SiRNA ACVRL1 leads to a reduced BMP9-induced GJA5 transcriptional activity. G, Confocal imaging showing Cx40 protein expression in arterial endothelial cells in skin sections of 1 representative healthy donor and 1 representative hemorrhagic telangiectasia type 2 (HHT2) individual. Endothelial cells are stained for platelet-endothelial cell adhesion molecule-1 (PECAM-1; red) and Cx40 (white), therefore representing arterial vessels. Scale bars, 50 µm. H, Quantification of Cx40-positive surface intensity normalized to PECAM-1 expression in the skin of 5 healthy donors and 4 HHT2 individuals. ****P<0.0001 results from 1-way ANOVA that compares means of multiple groups. NS indicates not significant.

Reduced Levels of Cx40 Affect Angiogenesis in Acvrl1+/− Mice

To investigate whether Acvrl1 and Gja5 function in the same pathway, we took advantage of the Acvrl1+/− mutant mice that develop vascular lesions similar to those in patients with HHT2 only at low frequency and in an unpredictable, age-dependent manner.12 We generated Acvrl1+/−; Gja5EGFP/+ mice that have reduced expression of both genes as validated by quantitative polymerase chain reaction (Figure IA and IB in the online-only Data Supplement) and first examined the neonatal retina. Gja5EGFP/+ mice are normally viable and fertile without cardiovascular abnormalities.27 The retinal vasculature of the single or double heterozygotes at postnatal day 7 (P7) showed a regular alternating pattern of arteries and veins with an intervening capillary network as in control mice (Figure 2A). Endothelial tip cells formed filopodial protrusions at the sprouting front of the plexus (Figure 2A). However, Acvrl1+/− retinas showed excessive angiogenesis with a denser and more highly branched vascular plexus at the front, as previously reported in Acvrl1-iKO mutants28 (Figure 2A–2C). By contrast, Acvrl1+/−; Gja5EGFP/+ mutant mice reproducibly showed reduced angiogenesis with much less dense postarteriolar capillary plexus (Figure 2B) that had fewer branch points (Figure 2C). By comparison, capillaries in other regions seemed unaffected (Figure 2A–2C; Figure IIA–IIC in the online-only Data Supplement). Changes in blood flow are known to drive vessel pruning, leading to maturation of the vasculature. Blood flow is generally low during vascular development in the neonatal retina excepted for the arterial segments close to the optic disc and for some of the first arterial branches29 (Figure IIA in the online-only Data Supplement). Interestingly, the reduced number of capillaries was in areas where the blood flow is estimated to be relatively high (Figure 2A–2C; Figure IIA in the online-only Data Supplement).29 Moreover, mural cell coverage of the arteries was enhanced in Acvrl1+/−; Gja5EGFP/+ mutant mice at P7 compared with single heterozygous and wild-type littermates as revealed by staining for α-smooth muscle actin (α-SMA). Arterioles where only few α-SMA–positive cells would be normally found at P7 were covered by regular layers of smooth muscle cells, especially near their branch points with the arteries (Figure 2D–2F). In agreement, Gja5 expression that correlates with the arterial flow pattern was strongly increased in the arterial vessels of Acvrl1+/−; Gja5EGFP/+ P7 mutant mice compared with Gja5EGFP/+ heterozygous mice as revealed by enhanced green fluorescent protein (EGFP), suggesting that the blood flow is perturbed in the Acvrl1+/−; Gja5EGFP/+ mutant mice (Figure 2D). To eliminate adaptive processes that may occur during embryonic development, in particular those related to blood flow regulation, confounding the analysis, we generated Acvrl1Flox/+; cdh5 (PAC)-CreERT2 (Acvrl1-iHET); Gja5EGFP/+ mice in which tamoxifen injection of neonatal mice led to efficient reduction of Acvrl1 mRNA expression to generate Acvrl1+/−; Gja5EGFP/+ mice (Figure IIIA and IIIB in the online-only Data Supplement).30,31 Impaired angiogenesis was much more severe in Acvrl1-iHET; Gja5EGFP/+ mice than in Acvrl1+/−; Gja5EGFP/+ mice, with particularly strong inhibition of postarterial capillary plexus density (Figure IIID and IIIE in the online-only Data Supplement). This confirmed that ALK1 signaling regulates angiogenesis by directly cooperating with Cx40. Thus, reduced Cx40 in HHT2 mice disrupts proper formation of the capillary bed connecting the artery and vein.

Figure 2.

Figure 2. Effect of Acvrl1 and Gja5 haploinsufficiency in vivo. A, Isolectin B4–stained endothelial cells in retinal vessels in control (n=11), Gja5EGFP/+ (n=10), Acvrl1+/ (n=9), and Acvrl1+/; Gja5EGFP/+ (n=8) mice at postnatal day 7 (P7). The outlined red boxes indicate the areas in which vascular parameters were quantified. Quantification of postarterial capillary density (B) and the number of vessel branch points (C) per field. D, Confocal imaging of retinas from control (n=6) Gja5EGFP/+ (n=4), Acvrl1+/ (n=5), and Acvrl1+/; Gja5EGFP/+ (n=7) mice at P7 stained for isolectin-B4 that marks endothelial cells (blue) and for α-smooth muscle actin (α-SMA) that marks vascular smooth muscle cells (red). Enhanced green fluorescent protein (EGFP) expression reveals arteries. E and F, Quantification of the α-SMA (+) vessel length from the first arterial branch and the number of arterial branch points. All error bars represent SEM. *P<0.05, **P<0.01, and ****P<0.0001 results from unpaired t test. Scale bars, 200 μm. a indicates arteries; NS, not significant; and v, veins.

Low Levels of Cx40 Leads to Arteriovenous Shunts in Acvrl1+/− Retinas

Because arteriovenous shunts are thought to arise from an abnormal capillary bed, we next explored the possibility that reduced levels of Cx40 promote AVM development in Acvrl1+/− retinas. We defined vessels ≥12.5 μm as arteriovenous shunts because arteriovenous connections of this diameter were not observed in control or single heterozygous mice at P7 (Figure 3A–3D). Arteriovenous shunts occurred in 71% (n=14) of the Acvrl1+/−; Gja5EGFP/+ P7 mice but were completely absent in control, Acvrl1+/−, or Gja5EGFP/+ mice. The arteriovenous shunts were found to arise from the capillary bed starting at the postarterial capillary vessels to form enlarged vessels that connected directly to the veins (Figure 3E). Similar phenotypes were observed in Acvrl1-iHET; Gja5EGFP/+ mice with a prevalence of 61% (n=13; Figure IVA in the online-only Data Supplement). Acvrl1-iHET mice did not develop any arteriovenous shunts (Figure IVA in the online-only Data Supplement). To determine whether the increase in AVM diameter was attributable to the increased endothelial cell number, we performed bromodeoxyuridine analysis of isolectin b4–stained vessels. In Acvrl1-iHET; Gja5EGFP/+ mice, the number of bromodeoxyuridine-labeled endothelial cells was significantly higher than in the other genotypes, particularly in the capillary plexus where arteriovenous connections were found (Figure IVB in the online-only Data Supplement). Moreover, these arteriovenous shunts lacked smooth muscle cell coverage (not shown). Because increased mural cell coverage of the AVM has been proposed to be a secondary response to increased blood flow, our data indicated that the arteriovenous shunts found in the Acvrl1+/−; Gja5EGFP/+ mutant mice at P7 might represent an early stage of AVM formation. Thus, our data are consistent with a primary abnormality at the capillary level and point to vessel enlargement promoting the development of AVM.

Figure 3.

Figure 3. Development of multiple transient arteriovenous connections in Acvrl1+/; Gja5EGFP/+ neonatal retina. AD, Confocal images of flat-mounted retinas labeled with isolectin-B4 that reveals the vascular plexus from control, Gja5EGFP/+, Acvrl1+/, and Acvrl1+/; Gja5EGFP/+ mice at postnatal day 7 (P7). The red arrow indicates a direct connection between an artery and a vein. E, Higher magnification of a typical microshunt found in Acvrl1+/; Gja5EGFP/+ P7 retinas. Yellow dotted lines delimit the shunt. Scale bars, 200 μm. a indicates arteries; and v, veins.

Low Levels of Cx40 Promote the Production of Reactive Oxygen Species and Lead to Arterial Dilation in Acvrl1+/− Mice

To explore the possibility that the enlargement of capillary-like vessels plays a causal role in the development of AVM, we examined the vasculature in the dorsal ear skin of Acvrl1+/−; Gja5EGFP/+ adult mice. This area of skin has recently proven extremely useful for intravital vascular imaging and is widely used to follow AVM development in real time, particularly in wound healing.17 We stained whole-mounts of 3-month-old mouse ears for platelet-endothelial cell adhesion molecule-1 and α-SMA (Figure 4A). This staining showed that the overall vessel patterning in Acvrl1+/−; Gja5EGFP/+ mice was similar to that in control, Gja5EGFP/+, or Acvrl1+/− mice (Figure 4A). The vascular network forming a finger-like architecture of larger veins that localize together with arteries that were stained for α-SMA (Figure 4A). However, the main arteries in the ear skin were reproducibly enlarged in Acvrl1+/−; Gja5EGFP/+ adult mice compared with control, Acvrl1+/−, or Gja5EGFP/+ mice (Figure 4A and 4B). Interestingly, the latter seems to occur from early development as seen in the mesencephalic artery of embryonic day 12.5 embryos (Figure VA–VC in the online-only Data Supplement) and in the lung and intestine tissue of embryonic day 17.5 embryos (Figure VD in the online-only Data Supplement). We next investigated the Gja5 expression pattern by following EGFP expression as an indicator of hemodynamic changes and capillary arterialization. In the Gja5EGFP/+ mice, EGFP expression was restricted to the main arteries and first arteriole branches (data not shown) copying the α-SMA staining (Figure 4A). By contrast, Acvrl1+/−; Gja5EGFP/+ mice showed numerous enlarged prearteriolar capillaries that expressed EGFP (not shown) or α-SMA (Figure 4A) compared with Gja5EGFP/+ mice. Quantification of the number of branch points and the length of the EGFP (+) vascular networks revealed an arterialization of the blood capillary bed in Acvrl1+/−; Gja5EGFP/+ mice (Figure 4C and 4D). We explored how reduced Gja5 expression affects the arterial vessel functionalities in Acvrl1+/− mice. We examined the spontaneous oscillation in tone of the skin arterial vessels (Figure 4E) and the ability of the retinal arteries to constrict after an electric stimulation (Figure 4F). The arterial responses of Acvrl1+/−, Gja5EGFP/+ mice did not differ from the control, Gja5EGFP/+, or Acvrl1+/− mice (Figure 4E and 4F), suggesting that the functionalities of the arteries are not defective per se. Both in vitro and in vivo biochemical data have suggested that reduced expression of Acvrl1 or Gja5 altered reactive oxygen species production.32,33 To access the potential role of oxidative stress, reactive oxygen species production was measured by identifying dihydroethidium-positive nuclei in arteries of lung sections of 1-month-old control, Gja5EGFP/+, Acvrl1+/−, or Acvrl1+/−; Gja5EGFP/+ mice. Interestingly, greater dihydroethidium staining was observed in Acvrl1+/−; Gja5EGFP/+ mice, suggesting higher O2 production (Figure 4G).

Figure 4.

Figure 4. Increased reactive oxygen species production and arterial dilation in Acvrl1+/; Gja5EGFP/+ mice. A, Images of whole-mounted ears of control (n=4), Acvrl1+/− (n=5), Gja5EGFP/+ (n=8), and Acvrl1+/−; Gja5EGFP/+ (n=12) mice stained with antibodies against platelet-endothelial cell adhesion molecule-1 (PECAM-1; endothelial cells in red) and α-smooth muscle actin (α-SMA; vascular smooth muscle cells in green) at postnatal day 60 (P60) to P90. α-SMA reveals arteries. Left, Main arteries and veins. Scale bars, 500 μm. Right, Higher magnifications of the whole-mount ears reveal an extension of the arterial network in Acvrl1+/−; Gja5EGFP/+ mice evident as increase numbers of α-SMA (+) vessels. White asterisks indicate arterial branch points. Scale bars, 200 μm. B, Quantification of the arterial diameters. C and D, Quantification of the number of branch points and vessel length of the enhanced green fluorescent protein (EGFP) (+) vascular network. E, Relationship between vessel diameter and a vasomotion index defined as the area under the curve for percent spontaneous changes in vessel diameter with a ±5% cutoff threshold; at least n=21 vessel segments from 3 mice per genotype were analyzed. F, Mean intensities that are able to induce the first retinal arterial constriction in control (narteries=32), Gja5EGFP/+ (narteries=28), Acvrl1+/− (narteries=32), or Acvrl1+/−; Gja5EGFP/+ (narteries=30) mice. At least 4 mice (8 weeks old) were analyzed per group. G, Increased dihydroethidium (DHE; red fluorescence) staining in lung sections of 8-week-old Acvrl1+/−; Gja5EGFP/+ (n=3) mice vs age-matched control (n=3), Gja5EGFP/+ (n=3), and Acvrl1+/− (n=3) mice. Arterial vessels show more positive nuclei with more intense red fluorescence in Acvrl1+/−; Gja5EGFP/+ mice. Scale bars, 50 µm. Quantification of the DHE red fluorescence intensity as determined by subtracting the image background from the average gray value within the α-SMA (+) vessels in control (nartery=40), Gja5EGFP/+ (nartery=62), Acvrl1+/ (nartery=35), and Acvrl1+/; Gja5EGFP/+ (nartery=53) lung sections. All error bars represent SEM. *P<0.05, **P<0.01, ***P<0.001 results from unpaired t test or from 1-way ANOVA test for multiple group comparison. a indicates arteries; l, lymphatic vessels; NS, not significant; and v, veins.

Basal Red Blood Cell Flow in Individual Capillaries of Acvrl1+/−; Gja5EGFP/+ Mice

Reactive oxygen species production and reduced nitric oxide bioavailability have been reported to reduce endothelial cell survival leading to pruning of the microvasculature and contributing to the muscularization of the small arteries.34 We observed a strong reduction in the density of the capillary bed (Figure 5A and 5B) associated with a slight increase in capillary diameter (Figure 5C) in Acvrl1+/−; Gja5EGFP/+ mice compared with control, Acvrl1+/−, and Gja5EGFP/+ mice. Capillary rarefaction is associated with local blood flow deregulation. To examine whether this contributed to the vascular phenotype here, we used 2-photon microscopy to measure red blood cell (RBC) flow with micrometer spatial and millisecond temporal resolution in individual capillaries.35 Retroorbital injection of red dextran revealed the vascular architecture of the ear skin and individual RBCs, which appeared as shadows flowing in the fluorescent plasma (Figure 5D). We used rapid line scans along the capillary axis to determine the instantaneous RBC flow (Figure 5E). We analyzed basal RBC flow in at least 29 capillaries that were located in a postarteriolar position in each genotype and detected all passing RBCs for 20 s (Figure 5F and 5G). Acvrl1+/−; Gja5EGFP/+ mice showed an increase in basal RBC flow compared with control, Acvrl1+/−, and Gja5EGFP/+ mice, most likely reflecting an increase in blood flow (Figure 5F and 5G).

Figure 5.

Figure 5. Measurements of red blood cell (RBC) flow in skin capillaries of Acvrl1+/; Gja5EGFP/+ mice. A, Images of whole-mounted ears of control (n=4), Acvrl1+/− (n=5), Gja5EGFP/+ (n=8), and Acvrl1+/−; Gja5EGFP/+ (n=12) mice stained with antibodies against platelet-endothelial cell adhesion molecule-1 (PECAM-1; endothelial cells in red) and α-smooth muscle actin (α-SMA; vascular smooth muscle cells in green) at postnatal day 60 (P60) to P90. α-SMA reveals arteries. The capillary network in whole-mount images of ears reveal rarefaction of the capillary network and increased capillary diameter in Acvrl1+/−; Gja5EGFP/+ mice compared with control, Acvrl1+/−, or Gja5EGFP/+ mice. White arrows indicate increased capillary diameter. Scale bars, 200 μm. B and C, Quantification of the number of capillary branch points and vessel diameters. D, Injection of Rhodamine B isothiocyanate–dextran reveals capillaries in skin. The white arrow indicates an enlarged capillary directly connected to a small arteriole labeled with enhanced green fluorescent protein (EGFP) in Acvrl1+/; Gja5EGFP/+ mice. Scale bars, 200 μm. E, The fluorescence of Rhodamine B isothiocyanate–dextran in the plasma is shadowed by passing RBCs during the excitation gate. Bottom, Red arrows correspond to individual RBCs. F, Automatic detection of RBC transients allows local extraction of blood flow rates. G, Quantification of RBC per second in capillaries of adult ears in control (ncapillary=29), Gja5EGFP/+ (ncapillary=29), Acvrl1+/ (ncapillary=45), and Acvrl1+/; Gja5EGFP/+ (ncapillary=34) mice. All error bars represent SEM. *P<0.05, **P<0.01, ***P<0.001, and ****P<0.0001 results from unpaired t test. a indicates arteries; l, lymphatic vessels; NS, not significant; and v, veins.

Reduced Cx40 Promotes Wound-Induced AVM Formation in Acvrl1+/−

To explore the possibility that reduced expression of Cx40 might predispose to AVM formation in adult Acvrl1+/− mice, we generated punch wounds in the ears of 3-month-old control, Acvrl1+/−, Gja5EGFP/+, and Acvrl1+/−; Gja5EGFP/+ mice. This type of wound induces environmental stress that triggers AVM formation in homozygous Acvrl1-iKO mice.17,36 Left ventricular injection of latex blue into the heart was performed 30 days after wounding to visualize arteriovenous connections in the skin. As expected, the blue latex did not cross the capillary bed and was retained within the arterial branches in control mice or Gja5EGFP/+ mice (Figure 6A). Moreover, Acvrl1+/− also showed normal morphology and latex only in arterial branches confirming that 3 events are required for AVM formation (Figure 6A).10 By contrast, Acvrl1+/−; Gja5EGFP/+ mice showed dilated and tortuous vessels, and the latex dye was found in both arteries and veins, indicating the presence of arteriovenous shunts (57%; n=7; Figure 6A). Blood vessels away from the wound in Acvrl1+/−; Gja5EGFP/+ mice had normal morphology and no arteriovenous shunts. We also stained whole mounts of mouse ears for platelet-endothelial cell adhesion molecule-1 and α-SMA 14 days after wounding. This staining showed that even at this stage, Acvrl1+/−; Gja5EGFP/+ mice developed abnormal connections between arterioles and enlarged capillary-like vessels that were EGFP (+) forming a nidus (Figure 6B). Thus, our data identify Gja5 as a possible genetic modifier of HHT2 and provide proof-of-concept that genes implicated in blood flow regulation might have important functions during AVM formation.

Figure 6.

Figure 6. Wounding can induce de novo arteriovenous (AV) malformation (AVM) formation in Acvrl1+/−; Gja5EGFP/+ mice. A, Skin vasculature of control (n=8), Gja5EGFP/+ (n=5), Acvrl1+/ (n=9), and Acvrl1+/; Gja5EGFP/+ (n=7) ears shown by latex dye injection after 1 month of wounding. Scale bars, 1 mm. Right, Higher magnifications of the vascular network around the wound. Scale bars, 200 μm. The development of AVM is indicated by a white arrow. Note that only Acvrl1+/; Gja5EGFP/+ mutant mice developed AV shunts, revealed by the presence of blue latex in both arteries and veins as indicated in (white arrowhead). The presence of AVMs was only found in the wound area. Asterisks indicate the center of the wound. B, Confocal images of whole-mounted ears of 3 month-old control (n=8), Gja5aEGFP (n=5), Acvrl1+/− (n=9), and Acvrl1+/−; Gja5aEGFP (n=7) mice stained 2 weeks after wounding for platelet-endothelial cell adhesion molecule-1 (PECAM-1) marking endothelial cells in red and for α-smooth muscle actin (α-SMA; vascular smooth muscle cells in green) or labeled in enhanced green fluorescent protein (EGFP) to identity the arteries. The white arrow indicates the formation of a system of multiple feeding arteries, the tangle or nidus and enlarged draining veins in Acvrl1+/−; Gja5aEGFP. C, Working model for AVM formation in hereditary hemorrhagic telangiectasia type 2 (HHT2). Heterozygosity of Acvrl1 represents the baseline situation in HHT2. The vascular network shows distinguishable arteries and veins separated by a highly branched vascular plexus. Pathological enlargement of the arterial vessels results in the delivery of more blood volume to the downstream capillaries that adapt by enlarging and by forming transient AV shunts. Sustained angiogenesis promotes further the enlargement of these AV shunts that may form large AVMs. a indicates arteries; and v, veins.


In this study, we identified Gja5 as a potential genetic modifier of AVMs in HHT2. Our work revealed that the BMP9/ALK1 signaling pathway targets Gja5 and that reduced expression of Gja5 in Acvrl1 heterozygous mutants leads to arterial vasodilation and rarefaction of the postarterial capillary network. To normalize the changes in hemodynamic forces and drain the engorged arterial system, capillaries become enlarged and form transient arteriovenous connections that can develop into AVMs when exposed to environmental insults (Figure 6C). Our data suggest that Acvrl1 haploinsufficiency combined with the effects of modifier genes that regulate vessel caliber and blood flow is responsible for the heterogeneity and severity of the clinical manifestations in individuals with HHT2 and provide a novel 3-hit hypothesis model for AVM development.

Our data provide the first demonstration that changes in arterial vessel precede AVM formation in HHT mouse models, and importantly, we identify Gja5 as a potential modifier gene for HHT2. We report that the ALK1 signaling pathway stimulates the expression of Gja5, most importantly though BMP9. Reduced expression of Gja5 in Acvrl1 heterozygous mice results in enlarged arterial vessels, altered blood flow, and the formation of transient arteriovenous shunts in the capillary bed; these can remodel into large AVMs where there is a proangiogenic and proinflammatory environment. Our findings are consistent with recent work showing that loss of Acvrl1 in zebrafish embryos leads to pathological arterial enlargement and consequently altered blood flow to induce lethal AVM formation.19Acvrl1+/−; Gja5EGFP/+ double heterozygous mice provide a novel genetic model in which AVMs develop consistently and robustly. Most importantly, the AVMs resemble those seen in patients with HHT, making this model an invaluable tool for uncovering the molecular and cellular defects that lead to vascular malformations in HHT, in particular those related to blood flow alterations. The Gja5 gene encodes for Cx40, a gap junction protein expressed in the developing arterial network, starting at the onset of perfusion.37 In the vascular system, endothelial cells predominantly express Cx37 (Gja4) and Cx40, whereas vascular smooth muscle cells mostly express Cx43 (Gja1) and Cx45 (Gjc1). Gap junction proteins form channels between neighboring cells to allow direct intercellular exchanges of ions and small metabolites, which are needed to coordinate vasoconstriction and vasodilation along the vessels.38,39 Genetic studies show that mice lacking Cx40 develop hypertension because of increased secretion of renin and reduced relaxation of peripheral vessels.40 These defects have recently been shown to be at least partially dependent on endothelial Cx40 function indicating that Cx40 expression levels regulate blood pressure. How endothelial Cx40 controls blood pressure remains poorly understood but might be attributable to endothelial nitric oxide synthase activity, an important modulator of vascular tone.40 Interestingly, both ALK1 and Cx40 interact and regulate the activity of endothelial nitric oxide synthase,33,41 suggesting that the phenotype of the Acvrl1+/−; Gja5EGFP/+ mice may be at least partially attributable to uncoupled endothelial nitric oxide synthase activity. To support this hypothesis, we have found an increase production of reactive oxygen species in the lung arteries of Acvrl1+/−; Gja5EGFP/+ mice. Mice carrying deletions in Gja5 also have fewer collateral arterioles,37 whereas Gja4−/−; Gja5−/− double-knockout mice die in utero, showing angiogenic remodeling defects with dilated blood vessels and hemorrhages,42 suggesting that Cx40 might regulate angiogenesis. Loss of Acvrl1 expression results in excessive endothelial cell proliferation that precedes the development of arteriovenous shunts in zebrafish19 and in Acvrl1-iKO mice.17 Here, we show that reduced expression of Gja5 in Acvrl1 heterozygous embryos and in neonatal retinas of postnatal day 7 mice leads to significant vasodilation of the arteries and markedly reduced numbers of postarterial branches. In contrast to Acvrl1+/ mutant mice, we did not observe excessive endothelial proliferation in Acvrl1+/−; Gja5EGFP/+ mice. Defective angiogenesis may, therefore, be the primary event responsible for this arterial enlargement, although the mechanisms that account for this observation—decreased arterial growth, retarded migration, and defective endothelial sprouting—remain to be elucidated. The presence of dilated arteries that deliver more blood induces the enlargement of the downstream Cx40-independent capillary-like vessels to drain this system and the formation of arteriovenous connections. This enlargement is accompanied by increased endothelial proliferation suggesting that the presence of flow stimulates further the remodeling of the capillary bed, a mechanism that might be independent of Cx40. Furthermore, loss of Acvrl1 in zebrafish has previously been shown to result in increased expression of cxcr4a and decreased expression of edn1, suggesting that ALK1 might promote the quiescence of nascent arteries by alternative mechanisms.19 Given the fact that these genes encode a proangiogenic chemokine receptor and a vasoconstrictive peptide, respectively, it is logical to consider that they may regulate arterial caliber downstream of ALK1. Nevertheless, concomitant increase in cxcr4a and loss of edn1 in zebrafish embryos did not copy the lack of Acvrl1 and were not sufficient to generate AVMs.19

Genetic polymorphisms have been detected in both promoter regions of GJA5 and suspected to be associated with risk of cardiovascular diseases, including hypertension.39 These polymorphisms have been shown to affect GJA5 promoter activity by reducing gene expression by approximately half with interassay variations ranging from 20% to 65% reduction. By comparing the expression levels of Cx40 in sections of human skin biopsies isolated from 5 healthy donors and 4 patients with HHT2, we have confirmed that important differences exist between individuals, and more importantly, we reveal that Cx40 protein levels are particularly low in the majority of a small selection of patients with HHT2 compared with the healthy donors supporting an association between ALK1 signaling and Cx40 expression.

In conclusion, our findings provide novel insights into the mechanisms underlying AVM pathogenesis in HHT2 elicited by increased arterial caliber that might ultimately be used for drug development for HHT. Moreover, we identity GJA5 as a potential modifier gene for HHT2 in which genetic variations, such as polymorphisms affecting normal expression levels, are associated with disease progression.

Nonstandard Abbreviations and Acronyms


α-smooth muscle actin


arteriovenous malformation


bone morphogenetic protein




enhanced green fluorescent protein


hereditary hemorrhagic telangiectasia


red blood cell


The online-only Data Supplement is available with this article at

Correspondence to Franck Lebrin, PhD, UMR CNRS 7241/INSERM U1050, Center for Interdisciplinary Research in Biology, Collège de France, 11 Pl Marcelin Berthelot, 75231 Paris cedex 05, France. E-mail


  • 1. Shovlin CL. Hereditary haemorrhagic telangiectasia: pathophysiology, diagnosis and treatment.Blood Rev. 2010; 24:203–219. doi: 10.1016/j.blre.2010.07.001.CrossrefMedlineGoogle Scholar
  • 2. McAllister KA, Grogg KM, Johnson DW, Gallione CJ, Baldwin MA, Jackson CE, Helmbold EA, Markel DS, McKinnon WC, Murrell J. Endoglin, a TGF-beta binding protein of endothelial cells, is the gene for hereditary haemorrhagic telangiectasia type 1.Nat Genet. 1994; 8:345–351. doi: 10.1038/ng1294-345.CrossrefMedlineGoogle Scholar
  • 3. Johnson DW, Berg JN, Baldwin MA, et al. Mutations in the activin receptor-like kinase 1 gene in hereditary haemorrhagic telangiectasia type 2.Nat Genet. 1996; 13:189–195. doi: 10.1038/ng0696-189.CrossrefMedlineGoogle Scholar
  • 4. Lebrin F, Goumans MJ, Jonker L, Carvalho RL, Valdimarsdottir G, Thorikay M, Mummery C, Arthur HM, ten Dijke P. Endoglin promotes endothelial cell proliferation and TGF-beta/ALK1 signal transduction.EMBO J. 2004; 23:4018–4028. doi: 10.1038/sj.emboj.7600386.CrossrefMedlineGoogle Scholar
  • 5. David L, Mallet C, Mazerbourg S, Feige JJ, Bailly S. Identification of BMP9 and BMP10 as functional activators of the orphan activin receptor-like kinase 1 (ALK1) in endothelial cells.Blood. 2007; 109:1953–1961. doi: 10.1182/blood-2006-07-034124.CrossrefMedlineGoogle Scholar
  • 6. Lebrin F, Deckers M, Bertolino P, Ten Dijke P. TGF-beta receptor function in the endothelium.Cardiovasc Res. 2005; 65:599–608. doi: 10.1016/j.cardiores.2004.10.036.CrossrefMedlineGoogle Scholar
  • 7. McDonald J, Bayrak-Toydemir P, Pyeritz RE. Hereditary hemorrhagic telangiectasia: an overview of diagnosis, management, and pathogenesis.Genet Med. 2011; 13:607–616. doi: 10.1097/GIM.0b013e3182136d32.CrossrefMedlineGoogle Scholar
  • 8. Leblanc GG, Golanov E, Awad IA, Young WL; Biology of Vascular Malformations of the Brain NINDS Workshop Collaborators. Biology of vascular malformations of the brain.Stroke. 2009; 40:e694–e702. doi: 10.1161/STROKEAHA.109.563692.LinkGoogle Scholar
  • 9. Braverman IM, Keh A, Jacobson BS. Ultrastructure and three-dimensional organization of the telangiectases of hereditary hemorrhagic telangiectasia.J Invest Dermatol. 1990; 95:422–427.CrossrefMedlineGoogle Scholar
  • 10. Tual-Chalot S, Oh SP, Arthur HM. Mouse models of hereditary hemorrhagic telangiectasia: recent advances and future challenges.Front Genet. 2015; 6:25. doi: 10.3389/fgene.2015.00025.CrossrefMedlineGoogle Scholar
  • 11. Bourdeau A, Dumont DJ, Letarte M. A murine model of hereditary hemorrhagic telangiectasia.J Clin Invest. 1999; 104:1343–1351. doi: 10.1172/JCI8088.CrossrefMedlineGoogle Scholar
  • 12. Srinivasan S, Hanes MA, Dickens T, Porteous ME, Oh SP, Hale LP, Marchuk DA. A mouse model for hereditary hemorrhagic telangiectasia (HHT) type 2.Hum Mol Genet. 2003; 12:473–482.CrossrefMedlineGoogle Scholar
  • 13. Torsney E, Charlton R, Diamond AG, Burn J, Soames JV, Arthur HM. Mouse model for hereditary hemorrhagic telangiectasia has a generalized vascular abnormality.Circulation. 2003; 107:1653–1657. doi: 10.1161/01.CIR.0000058170.92267.00.LinkGoogle Scholar
  • 14. Xu B, Wu YQ, Huey M, Arthur HM, Marchuk DA, Hashimoto T, Young WL, Yang GY. Vascular endothelial growth factor induces abnormal microvasculature in the endoglin heterozygous mouse brain.J Cereb Blood Flow Metab. 2004; 24:237–244. doi: 10.1097/01.WCB.0000107730.66603.51.CrossrefMedlineGoogle Scholar
  • 15. Hao Q, Zhu Y, Su H, Shen F, Yang GY, Kim H, Young WL. VEGF induces more severe cerebrovascular dysplasia in endoglin than in Alk1 mice.Transl Stroke.Res. 2010; 1:197–201. doi: 10.1007/s12975-010-0020-x.CrossrefMedlineGoogle Scholar
  • 16. Li C, Guo B, Ding S, Rius C, Langa C, Kumar P, Bernabeu C, Kumar S. TNF alpha down-regulates CD105 expression in vascular endothelial cells: a comparative study with TGF beta 1.Anticancer Res. 2003; 23(2B):1189–1196.MedlineGoogle Scholar
  • 17. Park SO, Wankhede M, Lee YJ, Choi EJ, Fliess N, Choe SW, Oh SH, Walter G, Raizada MK, Sorg BS, Oh SP. Real-time imaging of de novo arteriovenous malformation in a mouse model of hereditary hemorrhagic telangiectasia.J Clin Invest. 2009; 119:3487–3496. doi: 10.1172/JCI39482.MedlineGoogle Scholar
  • 18. Mahmoud M, Allinson KR, Zhai Z, Oakenfull R, Ghandi P, Adams RH, Fruttiger M, Arthur HM. Pathogenesis of arteriovenous malformations in the absence of endoglin.Circ Res. 2010; 106:1425–1433. doi: 10.1161/CIRCRESAHA.109.211037.LinkGoogle Scholar
  • 19. Corti P, Young S, Chen CY, Patrick MJ, Rochon ER, Pekkan K, Roman BL. Interaction between alk1 and blood flow in the development of arteriovenous malformations.Development. 2011; 138:1573–1582. doi: 10.1242/dev.060467.CrossrefMedlineGoogle Scholar
  • 20. Laux DW, Young S, Donovan JP, Mansfield CJ, Upton PD, Roman BL. Circulating Bmp10 acts through endothelial Alk1 to mediate flow-dependent arterial quiescence.Development. 2013; 140:3403–3412. doi: 10.1242/dev.095307.CrossrefMedlineGoogle Scholar
  • 21. Larrivée B, Prahst C, Gordon E, del Toro R, Mathivet T, Duarte A, Simons M, Eichmann A. ALK1 signaling inhibits angiogenesis by cooperating with the Notch pathway.Dev Cell. 2012; 22:489–500. doi: 10.1016/j.devcel.2012.02.005.CrossrefMedlineGoogle Scholar
  • 22. de Wit C, Roos F, Bolz SS, Kirchhoff S, Krüger O, Willecke K, Pohl U. Impaired conduction of vasodilation along arterioles in connexin40-deficient mice.Circ Res. 2000; 86:649–655.LinkGoogle Scholar
  • 23. Figueroa XF, Paul DL, Simon AM, Goodenough DA, Day KH, Damon DN, Duling BR. Central role of connexin40 in the propagation of electrically activated vasodilation in mouse cremasteric arterioles in vivo.Circ Res. 2003; 92:793–800. doi: 10.1161/01.RES.0000065918.90271.9A.LinkGoogle Scholar
  • 24. Milkau M, Köhler R, de Wit C. Crucial importance of the endothelial K+ channel SK3 and connexin40 in arteriolar dilations during skeletal muscle contraction.FASEB J. 2010; 24:3572–3579. doi: 10.1096/fj.10-158956.CrossrefMedlineGoogle Scholar
  • 25. Vorderwülbecke BJ, Maroski J, Fiedorowicz K, Da Silva-Azevedo L, Marki A, Pries AR, Zakrzewicz A. Regulation of endothelial connexin40 expression by shear stress.Am J Physiol Heart Circ Physiol. 2012; 302:H143–H152. doi: 10.1152/ajpheart.00634.2011.CrossrefMedlineGoogle Scholar
  • 26. Lebrin F, Srun S, Raymond K, et al. Thalidomide stimulates vessel maturation and reduces epistaxis in individuals with hereditary hemorrhagic telangiectasia.Nat Med. 2010; 16:420–428. doi: 10.1038/nm.2131.CrossrefMedlineGoogle Scholar
  • 27. Meysen S, Marger L, Hewett KW, Jarry-Guichard T, Agarkova I, Chauvin JP, Perriard JC, Izumo S, Gourdie RG, Mangoni ME, Nargeot J, Gros D, Miquerol L. Nkx2.5 cell-autonomous gene function is required for the postnatal formation of the peripheral ventricular conduction system.Dev Biol. 2007; 303:740–753. doi: 10.1016/j.ydbio.2006.12.044.CrossrefMedlineGoogle Scholar
  • 28. Tual-Chalot S, Mahmoud M, Allinson KR, Redgrave RE, Zhai Z, Oh SP, Fruttiger M, Arthur HM. Endothelial depletion of Acvrl1 in mice leads to arteriovenous malformations associated with reduced endoglin expression.PLoS One. 2014; 9:e98646. doi: 10.1371/journal.pone.0098646.CrossrefMedlineGoogle Scholar
  • 29. Bernabeu MO, Jones ML, Nielsen JH, Krüger T, Nash RW, Groen D, Schmieschek S, Hetherington J, Gerhardt H, Franco CA, Coveney PV. Computer simulations reveal complex distribution of haemodynamic forces in a mouse retina model of angiogenesis.J R Soc Interface. 2014; 11. doi: 10.1098/rsif.2014.0543.CrossrefMedlineGoogle Scholar
  • 30. Park SO, Lee YJ, Seki T, Hong KH, Fliess N, Jiang Z, Park A, Wu X, Kaartinen V, Roman BL, Oh SP. ALK5- and TGFBR2-independent role of ALK1 in the pathogenesis of hereditary hemorrhagic telangiectasia type 2.Blood. 2008; 111:633–642. doi: 10.1182/blood-2007-08-107359.CrossrefMedlineGoogle Scholar
  • 31. Wang Y, Nakayama M, Pitulescu ME, Schmidt TS, Bochenek ML, Sakakibara A, Adams S, Davy A, Deutsch U, Lüthi U, Barberis A, Benjamin LE, Mäkinen T, Nobes CD, Adams RH. Ephrin-B2 controls VEGF-induced angiogenesis and lymphangiogenesis.Nature. 2010; 465:483–486. doi: 10.1038/nature09002.CrossrefMedlineGoogle Scholar
  • 32. Looft-Wilson RC, Billaud M, Johnstone SR, Straub AC, Isakson BE. Interaction between nitric oxide signaling and gap junctions: effects on vascular function.Biochim Biophys Acta. 2012; 1818:1895–1902. doi: 10.1016/j.bbamem.2011.07.031.CrossrefMedlineGoogle Scholar
  • 33. Jerkic M, Kabir MG, Davies A, Yu LX, McIntyre BA, Husain NW, Enomoto M, Sotov V, Husain M, Henkelman M, Belik J, Letarte M. Pulmonary hypertension in adult ALK1 heterozygous mice due to oxidative stress.Cardiovasc Res. 2011; 92:375–384. doi: 10.1093/cvr/cvr232.CrossrefMedlineGoogle Scholar
  • 34. Jerkic M, Letarte M. Contribution of oxidative stress to endothelial dysfunction in hereditary hemorrhagic telangiectasia.Front Genet. 2015; 6:34. doi: 10.3389/fgene.2015.00034.CrossrefMedlineGoogle Scholar
  • 35. Lecoq J, Parpaleix A, Roussakis E, Ducros M, Goulam Houssen Y, Vinogradov SA, Charpak S. Simultaneous two-photon imaging of oxygen and blood flow in deep cerebral vessels.Nat Med. 2011; 17:893–898. doi: 10.1038/nm.2394.CrossrefMedlineGoogle Scholar
  • 36. Garrido-Martin EM, Nguyen HL, Cunningham TA, Choe SW, Jiang Z, Arthur HM, Lee YJ, Oh SP. Common and distinctive pathogenetic features of arteriovenous malformations in hereditary hemorrhagic telangiectasia 1 and hereditary hemorrhagic telangiectasia 2 animal models–brief report.Arterioscler Thromb Vasc Biol. 2014; 34:2232–2236. doi: 10.1161/ATVBAHA.114.303984.LinkGoogle Scholar
  • 37. Buschmann I, Pries A, Styp-Rekowska B, et al. Pulsatile shear and Gja5 modulate arterial identity and remodeling events during flow-driven arteriogenesis.Development. 2010; 137:2187–2196. doi: 10.1242/dev.045351.CrossrefMedlineGoogle Scholar
  • 38. Haefliger JA, Nicod P, Meda P. Contribution of connexins to the function of the vascular wall.Cardiovasc Res. 2004; 62:345–356. doi: 10.1016/j.cardiores.2003.11.015.CrossrefMedlineGoogle Scholar
  • 39. Molica F, Meens MJ, Morel S, Kwak BR. Mutations in cardiovascular connexin genes.Biol Cell. 2014; 106:269–293. doi: 10.1111/boc.201400038.CrossrefMedlineGoogle Scholar
  • 40. Morton SK, Chaston DJ, Howitt L, Heisler J, Nicholson BJ, Fairweather S, Bröer S, Ashton AW, Matthaei KI, Hill CE. Loss of functional endothelial connexin40 results in exercise-induced hypertension in mice.Hypertension. 2015; 65:662–669. doi: 10.1161/HYPERTENSIONAHA.114.04578.LinkGoogle Scholar
  • 41. Le Gal L, Alonso F, Mazzolai L, Meda P, Haefliger JA. Interplay between connexin40 and nitric oxide signaling during hypertension.Hypertension. 2015; 65:910–915. doi: 10.1161/HYPERTENSIONAHA.114.04775.LinkGoogle Scholar
  • 42. Simon AM, McWhorter AR. Vascular abnormalities in mice lacking the endothelial gap junction proteins connexin37 and connexin40.Dev Biol. 2002; 251:206–220.CrossrefMedlineGoogle Scholar


Hereditary hemorrhagic telangiectasia is a rare inherited vascular disease characterized by weak vessel walls that hemorrhage easily. Profound nosebleeds and enlarged vessels in the major organs are principle features of the disease. Why some patients with hereditary hemorrhagic telangiectasia have more severe symptoms than others or it affects them earlier is unknown. Using both mutant mice and tissue from patients with hereditary hemorrhagic telangiectasia, we identify, in this study, 3 novel mechanisms that may underlie 1 particular form of the disease: polymorphisms or mutations in a junctional protein expressed in endothelial cells that regulates vascular tone; blood flow in capillaries which regulates blood vessel caliber and inflammatory triggers to which patients may be exposed. The study provides an inroad that could be used by clinicians to assess risk of severe vascular abnormalities in patients with hereditary hemorrhagic telangiectasia becoming life threatening.