Skip main navigation

Common and Distinctive Pathogenetic Features of Arteriovenous Malformations in Hereditary Hemorrhagic Telangiectasia 1 and Hereditary Hemorrhagic Telangiectasia 2 Animal Models—Brief Report

Originally published, Thrombosis, and Vascular Biology. 2014;34:2232–2236



Hereditary hemorrhagic telangiectasia is a genetic disorder characterized by visceral and mucocutaneous arteriovenous malformations (AVMs). Clinically indistinguishable hereditary hemorrhagic telangiectasia 1 and hereditary hemorrhagic telangiectasia 2 are caused by mutations in ENG and ALK1, respectively. In this study, we have compared the development of visceral and mucocutaneous AVMs in adult stages between Eng- and Alk1-inducible knockout (iKO) models.

Approach and Results—

Eng or Alk1 were deleted from either vascular endothelial cells (ECs) or smooth muscle cells in adult stages using Scl-CreER and Myh11-CreER lines, respectively. Latex perfusion and intravital spectral imaging in a dorsal skinfold window chamber system were used to visualize remodeling vasculature during AVM formation. Global Eng deletion resulted in lethality with visceral AVMs and wound-induced skin AVMs. Deletion of Alk1 or Eng in ECs, but not in smooth muscle cells, resulted in wound-induced skin AVMs. Visceral AVMs were observed in EC-specific Alk1-iKO but not in Eng-iKO. Intravital spectral imaging revealed that Eng-iKO model exhibited more dynamic processes for AVM development when compared with Alk1-iKO model.


Both Alk1- and Eng-deficient models require a secondary insult, such as wounding, and ECs are the primary cell type responsible for the pathogenesis. However, Alk1 but not Eng deletion in ECs results in visceral AVMs.


Hereditary hemorrhagic telangiectasia (HHT) is an autosomal dominant vascular disorder with a prevalence of 1:5000.1 Several clinically indistinguishable forms of HHT are caused by mutations in ≥5 autosomal genes, among which endoglin (ENG; HHT1) and activin receptor-like kinase 1 (ACVRL1/ALK1; HHT2) constitute >80% of the cases.1 HHT is characterized by the presence of arteriovenous malformations (AVMs), abnormal direct connections between arteries and veins, in the skin, mucosa, and internal organs.1 It is generally thought that AVMs in the brain arise during developmental stages.2 But patients with HHT also develop small AVMs, called telangiectases in eyelids, lips, tongue, nasal cavity, and gastrointestinal tract, that are mostly acquired during postdevelopmental stages.3 The frequency and severity of intranasal and gastrointestinal telangiectases increase with the age, and their frequent ruptures can lead to severe anemia. Understanding the pathogenetic mechanisms that lead to the formation of telangiectases and identifying the cell types responsible for this process are essential for developing therapeutic strategies for epistaxis and gastrointestinal bleeding in patients with HHT.

We have previously shown that secondary events, such as wounding, are necessary for inducing de novo AVMs in adult stages of global Alk1-inducible knockout (iKO) model.4 Because both ALK1 and ENG are components of the transforming growth factor-β family signaling complex, it has been presumed that HHT1 and HHT2 share a common pathogenetic mechanism. However, it is not known whether Eng-deficient vessels also need an environmental insult in addition to the genetic predisposition to trigger the development of mucocutaneous AVMs in adult stages.

The cellular origin of the AVMs remains unresolved. We have previously shown that Alk1 deletion using an endothelial cell (EC)–specific Cre driver (L1Cre) resulted in AVMs in the brain, lungs, and gastrointestinal tract at embryonic and neonatal stages.4,5 However, it is unknown whether ECs are the primary cellular source of the wound-induced skin AVMs and visceral AVMs in adult Alk1-iKO model as well. Alk1 expression was detected beyond vascular ECs, such as the lymphatic ECs6 and monocytes/macrophages (Y.H. Kim, PhD, et al, 2014, unpublished data), and the Cre activity was detected in macrophages of L1Cre mice. These data raise a question whether ECs are the sole source of Alk1-deficienct effect for the AVM development in L1Cre;Alk12f/2f mice.4 In the case of Endoglin, the origin of the AVMs could reside in several cell sources, considering that it is expressed in a variety of cell types relevant for vascular functions, including ECs, smooth muscle cells (SMCs), endothelial precursors, and macrophages.7 Interestingly, deletion of either Alk1 or Eng by SM22α-Cre resulted in brain AVMs,8,9 implicating potential functions of ALK1 and ENG in vascular SMCs for establishing the proper arteriovenous network.

In this study, we examined the effect of wounding in the development of mucocutaneous telangiectases and investigated the cellular origin of de novo AVMs in both HHT1 and HHT2 mouse models.

Materials and Methods

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


To investigate the role of ENG in adult mice, Eng was globally deleted using the tamoxifen-inducible ROSA26CreER mouse strain (R26CreER).10 We found that 3 consecutive day injection of tamoxifen at 2.5 mg/25 g body weight was the most effective regimen for R26CreER/+;Eng2f/2f mice (Figure I in the online-only Data Supplement). In 3 to 4 days after the first tamoxifen injection, the R26CreER/+;Eng2f/2f mutant mice displayed signs of illness, such as slow movement, diarrhea, and dehydration, and died around day 4 to 10 (n>30). To analyze the subdermal vessels in the back skin, vascular casting with blue latex was performed at days 5 to 8 by infusing it into the left ventricle. Subdermal vessels were unaffected in the intact back skins of tamoxifen-injected adult R26CreER/+;Eng2f/2f mice (data not shown) and also wounded skin of tamoxifen-injected adult R26CreER/+;Eng2f/+ mice (Figure IIA in the online-only Data Supplement; n=6). However, areas of wounds in mid-dorsum and ear of tamoxifen-injected R26CreER/+;Eng2f/2f mice showed dilated and tortuous vessels, and the latex dye was found in both arteries and veins, indicating the presence of AV shunts (Figure IIB in the online-only Data Supplement; n=13). However, blood vessels away from the wound in Eng-iKOs had normal morphology and latex only in arterial branches.

To determine the vascular cell type where ENG plays a critical role for the development of the vascular network at adult stages, we used 2 cell-type–specific inducible Cre lines: Scl-CreER11 for ECs and Myh11-CreER12 for SMCs. Whole mount X-Gal staining of back skin tissues collected from tamoxifen-injected Scl-CreER(+);R26R(+) mice displayed Cre activity in the vascular network (Figure IIIAi and IIIAii in the online-only Data Supplement). Histological studies of the areas around the skin wound confirmed the presence of Cre recombinase activity in ECs of veins (Figure IIIAiii in the online-only Data Supplement), arteries (Figure IIIAiva in the online-only Data Supplement), and capillaries (Figure IIIAivb in the online-only Data Supplement), and vascular ECs in the intestines (Figure IIIE in the online-only Data Supplement), whereas no Cre activity was detectable in the lymphatic endothelium (Figure IIIC in the online-only Data Supplement) or in the macrophages (Figure IIID in the online-only Data Supplement). Tamoxifen-injected adult Scl-CreER(+);Eng2f/2f mice were subjected to wounding. Blood vessels seemed to be tortuous and enlarged and presented multiple AV shunts in the punched area of both the ear and the back skin (Figure 1Aiii and 1Aiv), whereas no such sign of AV shunts was visible in the controls (Figure 1Ai and 1Aii). Scl-CreER(+);Eng2f/2f mice did not exhibit visible signs of hemorrhages and AVMs in internal organs and survived longer than a month (Figure 1B). The Cre activity was found in the vascular SMCs of both arteries and veins of tamoxifen-treated Myh11-CreER(+);R26R(+) mice as expected (Figure IIIB and IIIF). Eng deletion induced in Myh11-CreER(+);Eng2f/2f mice by tamoxifen treatment in adult stages did not result in AV shunts in any areas of the skin including the wound areas (Figure 1C).

Figure 1.

Figure 1. De novo skin arteriovenous malformations (AVMs) in both endoglin (Eng)–deficient and activin receptor-like kinase 1 (Alk1)–deficient mice have an endothelial cell origin. A, Vasculature of tamoxifen-injected Scl-CreER(+);Eng2f/2f mutant mice (n=10) showing AVMs in the wounded area of ears (iii) and back skin (iv) vs normal blood vessels in the wounded area of ears (i) and back skin (ii) of the control Scl-CreER(+);Eng2f/+ mice (n=5), at day 8 after wounding. Magnified areas around the wound are shown below the panels ii and iv. B, Blood vessels of cecum, ileum, stomach, and uterus are visualized by blue latex perfusion in control (left) and mutant mice (right). No AVM was observed in either group. C, Vascular patterns observed by injection of blue latex into the left ventricle in control (Myh11-CreER(+);Eng2f/+; n=6), and smooth muscle-specific Eng-mutant mice (Myh11-CreER(+);Eng2f/2f, n=4) in the area around the wound of the ear and the back skin showed absence of AVMs 8 days after wounding. Magnified areas around the wounds are shown below. D, Blood vessels in the wounded ear and back skin of control (i and ii, n=8) and mutant (iii and iv; n=16) mice at day 8 after the first tamoxifen treatment and wounding. Magnified areas around the wound are shown below each panel. E, AVMs in visceras of Scl-CreER(+);Alk12f/2f mutant mice. Blood vessels perfused with latex dye in the cecum, ileum, stomach, and uterus of control (left) and mutant mice (right). Hemorrhage found in the cecum of the mutant mice is indicated with *. AVMs in gastrointestinal tract (small intestine and forestomach) and uterus were found in mutant mice at day 8 after tamoxifen treatment (n=16). F, Normal vessels present in the wounded area of the ear and the back skin of mutant Myh11-CreER(+);Alk12f/2f (n=8) when compared with the control Myh11-CreER(+);Alk12f/+ (n=8) mice. Magnified areas around the wound are shown below panels ii and iv. Arteries and veins are indicated by arrows and arrowheads, respectively (A and D). Characteristic loop-shaped AV shunts in A and D are indicated with red arrows. Wounds are indicated with *.Scale bars, 1 mm (i) and 4 mm (ii).

We have previously demonstrated that induced global deletion of Alk1 in adult stages (tamoxifen-treated R26CreER/+;Alk12f/2f) resulted in gastrointestinal AVMs and skin AVMs only in the wound area.4 We also investigated the cellular source of these malformations with the same inducible approaches used for Eng-iKOs. Tamoxifen-treated Scl-CreER(+);Alk12f/2f mice displayed multiple AV shunts in the ear and skin wound areas (Figure 1D). The vessels associated with AV shunts were tortuous, enlarged, and presented characteristic loops recapitulating those previously reported in R26CreER/+;Alk12f/2f mice.4 These Alk1-iKO mice exhibited AVMs in gastrointestinal tract and severe hemorrhages in the cecum (Figure 1E). The survival rate of the mutant mice was ≈2 weeks on average after first tamoxifen injection, and lethality is most likely associated with gastrointestinal hemorrhages. However, tamoxifen-treated Myh11-CreER(+);Alk12f/2f mice did not display AV shunts in any areas of the skin, including the wound areas (Figure 1F).

Defects in lymphangiogenesis have been reported under global Alk1 deletion.13 To investigate the association between vascular malformation and lymphangiogenesis defects, lymphatic vessels neighboring to the AVMs were analyzed. As shown in Figure IV in the online-only Data Supplement, deletion of Alk1 specifically in the vascular ECs did not affect the morphology, branch density, and caliber of lymphatic vessels. These data would indicate that the lymphatic phenotype previously observed in tamoxifen-treated R26CreER/+;Alk12f/2f mice13 is likely a cell-autonomous effect on lymphatic vasculature, and that vascular malformation occurs independent from lymphatic dysplasia.

A dorsal skinfold window chamber system coupled with hyperspectral imaging4,14 allows us to monitor the vascular remodeling and the AV shunts intravitally by mapping the oxygen saturation content in the blood vessels. We have tracked the process of wound-induced AVM formation in Scl-CreER(+);Eng2f/2f mice in parallel with Scl-CreER(+);Alk12f/2f mice. Consistent with the results shown in Figure 1, AVMs developed near the wound in both models (Figure 2; Figure V in the online-only Data Supplement). AV shunts began to appear around the day 3 as veins get dilated and arterial blood flows into veins, and characteristic looping shunts became apparent by days 8 to 9. There were individual variations in the process of AVM development in this wound-induced model. Nonetheless, the process of AVM formation in Eng model (Figure 2A) was generally much more dynamic than in the Alk1 model (Figure 2B), in which AVMs appeared at day 8 to 9 could be predicted from images in days 4 to 5. However, in the Eng model, the shunts formed around day 6 evolved and disappeared, and new shunts in different locations were formed and regressed in the following days 7 to 9 (Figure 2A). The reason for this difference is not clear. It might be rooted in the difference of the recombination efficiency of each gene by Cre.15 Alternatively, ECs deficient in each gene may respond differently to environmental cues generated by the wound.

Figure 2.

Figure 2. Real-time hyperspectral imaging of de novo arteriovenous malformation (AVM) in subdermal vessels of endothelial cell–specific Eng-inducible knockout in response to wound. Representative brightfield (left) and corresponding hyperspectral images (right) of the wounded area of the skin in a dorsal window chamber of Scl-CreER(+);Eng2f/2f (A) and Scl-CreER(+);Alk12f/2f (B) mice treated with tamoxifen. Images show days 1, 4, 7, 8, and 9 after first TM administration and wounding. Images of consecutive days 1 to 9 can be found in Figure V in the online-only Data Supplement. Wound sites are indicated by *. Arterial (A) and venous (V) vessels are indicated by red and blue dotted lines, respectively. AV shunts are indicated by yellow dotted lines and arrows. Notice that AV shunts subsequently regressed and formed new shunts between different vessels, in a highly dynamic process during AVM formation in Eng-inducible knockout mice. All images are shown at the same magnification. Scale bar, 0.5 mm. The color bar in the bottom right of B indicates hemoglobin saturation levels.


We showed here that mucocutaneous AVM formation in Eng-iKO (HHT1 model) and Alk1-iKO (HHT2 model) at adult stages share common features: both require environmental stresses, such as wounding, and the vascular EC is the primary cell type pertinent to the development of AVMs. Either ENG or ALK1 in SMCs is dispensable for maintaining normal vasculature and for the formation of a normal vascular network during wound healing in adult stages. We also found some differences in these 2 HHT models: EC-specific Alk1-iKO, but not Eng-iKO, exhibited gastrointestinal bleeding and visceral AVMs. This result may indicate that pathogenetic mechanisms for gastrointestinal AVM in HHT1 may involve non-ECs. In addition, intravital imaging of wound-induced AVM development guided by hyperspectral system revealed that Eng-iKO model involves more dynamic processes for de novo AVM development when compared with Alk1-iKO model.

This result further emphasizes the importance of secondary factors, such as wound, inflammation, infections, or trauma, in the development of mucocutaneous AVMs in adult stages for both HHT1 and HHT2 models. The wound-induced mucocutaneous AVM model shown here closely mimics telangiectases forming in the nasal mucosa of patients with HHT. Identification of ECs as the primary cell type pertinent to the development of AVMs provides an important platform for studying pathogenetic mechanisms of HHT and determining the cellular target of therapy. Further molecular and cellular mechanisms by which ALK1- or ENG-deficient ECs respond to various environmental cues would provide important insights for developing therapies directed to epistaxis in patients with HHT.

Nonstandard Abbreviations and Acronyms


arteriovenous malformation


endothelial cells


hereditary hemorrhagic telangiectasia


inducible knockout


smooth muscle cells


We thank Drs Su and Weiser-Evans for providing the Scl-CreER and Myh11-CreER transgenic mice, respectively. We thank Haeja Choi and Mi Jung Kim for technical assistance in histological analysis and Southern blot analysis, respectively.


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

Correspondence to S. Paul Oh, PhD, Department of Physiology and Functional Genomics, University of Florida, 1600 SW Archer Rd, CG20B, Gainesville, FL 32610. E-mail


  • 1. Govani FS, Shovlin CL. Hereditary haemorrhagic telangiectasia: a clinical and scientific review.Eur J Hum Genet. 2009; 17:860–871.CrossrefMedlineGoogle Scholar
  • 2. Fleetwood IG, Steinberg GK. Arteriovenous malformations.Lancet. 2002; 359:863–873.CrossrefMedlineGoogle Scholar
  • 3. Shovlin CL. Hereditary haemorrhagic telangiectasia: pathophysiology, diagnosis and treatment.Blood Rev. 2010; 24:203–219.CrossrefMedlineGoogle Scholar
  • 4. 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.MedlineGoogle Scholar
  • 5. 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.CrossrefMedlineGoogle Scholar
  • 6. Niessen K, Zhang G, Ridgway JB, Chen H, Yan M. ALK1 signaling regulates early postnatal lymphatic vessel development.Blood. 2010; 115:1654–1661.CrossrefMedlineGoogle Scholar
  • 7. Dallas NA, Samuel S, Xia L, Fan F, Gray MJ, Lim SJ, Ellis LM. Endoglin (CD105): a marker of tumor vasculature and potential target for therapy.Clin Cancer Res. 2008; 14:1931–1937.CrossrefMedlineGoogle Scholar
  • 8. Milton I, Ouyang D, Allen CJ, Yanasak NE, Gossage JR, Alleyne CH, Seki T. Age-dependent lethality in novel transgenic mouse models of central nervous system arteriovenous malformations.Stroke. 2012; 43:1432–1435.LinkGoogle Scholar
  • 9. Choi EJ, Chen W, Jun K, Arthur HM, Young WL, Su H. Novel brain arteriovenous malformation mouse models for type 1 hereditary hemorrhagic telangiectasia.PLoS One. 2014; 9:e88511.CrossrefMedlineGoogle Scholar
  • 10. Badea TC, Wang Y, Nathans J. A noninvasive genetic/pharmacologic strategy for visualizing cell morphology and clonal relationships in the mouse.J Neurosci. 2003; 23:2314–2322.CrossrefMedlineGoogle Scholar
  • 11. Göthert JR, Gustin SE, van Eekelen JA, Schmidt U, Hall MA, Jane SM, Green AR, Göttgens B, Izon DJ, Begley CG. Genetically tagging endothelial cells in vivo: bone marrow-derived cells do not contribute to tumor endothelium.Blood. 2004; 104:1769–1777.CrossrefMedlineGoogle Scholar
  • 12. Wirth A, Benyo Z, Lukasova M, Leutgeb B, Wettschureck N, Gorbey S, Orsy P, Horvath B, Maser-Gluth C, Greiner E, Lemmer B, Schuetz N, Gutkind S, Offermanns S. G(12)-G(13)-LARG-mediated signaling in vascular smooth muscle is required for salt-induced hypertension.Nature Medicine. 2008; 14:64–68.CrossrefMedlineGoogle Scholar
  • 13. Yoshimatsu Y, Lee YG, Akatsu Y, et al. Bone morphogenetic protein-9 inhibits lymphatic vessel formation via activin receptor-like kinase 1 during development and cancer progression.Proc Natl Acad Sci U S A. 2013; 110:18940–18945.CrossrefMedlineGoogle Scholar
  • 14. Sorg BS, Moeller BJ, Donovan O, Cao Y, Dewhirst MW. Hyperspectral imaging of hemoglobin saturation in tumor microvasculature and tumor hypoxia development.J Biomed Opt. 2005; 10:44004.CrossrefMedlineGoogle Scholar
  • 15. Choi EJ, Walker EJ, Shen F, Oh SP, Arthur HM, Young WL, Su H. Minimal homozygous endothelial deletion of Eng with VEGF stimulation is sufficient to cause cerebrovascular dysplasia in the adult mouse.Cerebrovasc Dis. 2012; 33:540–547.CrossrefMedlineGoogle Scholar


Most patients with HHT experience recurrent epistaxis and gastrointestinal bleeding that are a heavy burden to them because these inhibit social activities, cause anemia, and occur throughout their lifetime (it gets worse as one ages). This is the first side-by-side comparison data of HHT1 and HHT2 models for the development of gastrointestinal and mucocutaneous AVMs in adult stages. We demonstrate that for both models, a secondary factor represented by wound is required for Eng- or Alk1-deficient adult mice to develop de novo skin AVMs. These data infer 2 therapeutic axes for inhibiting de novo AVMs: overcoming ALK1 or ENG deficiency and blocking the secondary factor (eg, angiogenesis). Our results delineate endothelial cells as the primary cell type to scrutinize molecular and cellular mechanism of ALK1 or ENG deficiency further that leads to the development of abnormal arteriovenous connections in response to secondary factors.