A brain arteriovenous malformation (AVM) is an abnormal tangle of arteries and veins that lacks intervening capillaries and thereby forms a low-resistance, high-flow nidus prone to life-threatening intracerebral hemorrhage and seizures. AVMs have long been considered congenital lesions that are present at birth. AVM pathogenesis might be better understood if an animal model replicated the pathologic anatomy and hemodynamics of human AVMs, behaved similarly with microhemorrhage and rupture, and allowed for experimental manipulation. However, animal models have been elusive. Fistulas created microsurgically, like the carotid artery–jugular vein anastomosis model, replicate arteriovenous hemodynamics and elicit dural arteriovenous fistula (DAVF) formation by causing venous hypertension and angiogenesis activity, but DAVFs and brain AVMs are not the same diseases.1,2 Angiogenic stimulation of blood vessels in the brains of quail embryos synchronized experimental manipulation with vascular development but resulted in a modest expansion of arteries without the formation of arteriovenous pathology.3
A genetics-based approach to developing an animal model recognized hereditary hemorrhagic telangiectasia (HHT, or Osler-Weber-Rendu syndrome) as the only condition with multiple inherited brain AVMs, as well as AVMs in the lungs and liver. Genetic studies have identified mutations in endoglin (ENG) and activin receptor-like kinase 1 (ACVRL1 or ALK1) genes as the cause of this disease.4,5 Both gene products are located on the cell surface and are involved in signaling transforming growth factor–β/bone morphogenetic protein family ligands.6 Mouse models that deleted Alk1 or Eng genes with conventional knockout systems were impractical because widespread gene deletion caused embryonic lethality.7–10 However, conditional gene deletion can be focal and can recreate the genetic environment for AVM formation in mature animals. The first successful genetics-based brain AVM model used an Alk1 conditional knockout mouse in combination with a stereotactic injection of a Cre-expressing adenoviral vector and an adenovirus-associated viral vector expressing vascular endothelial growth factor (VEGF).11 However, this combination of Cre-induced conditional Alk1 deletion and VEGF-induced angiogenesis stimulation resulted in cerebrovascular dysplasia more than arteriovenous pathology resembling the human disease. Conditional gene deletion with viral vectors was avoided with a simpler model that crossed Alk1-floxed mice with deleter mice to generate offspring in which both copies of the Alk1 gene were deleted by Cre recombinase expressed from the transgene SM22α/Tagln-Cre.12
In this study, we developed a longitudinal brain AVM model. We found that it generates brain AVMs in mice that replicate the human disease better than any previous model system. We characterized these AVMs radiographically with MRI, MRA, and DSA, as is done clinically in human patients. We correlated these radiographic findings with histopathological findings. AVM models require extended animal survival to examine the longitudinal effects of experimental manipulation. Therefore, we examined AVM stability and hemorrhagic behavior in mice over time. This study offers a novel experimental model of human brain AVM for future studies of AVM pathogenesis, pathophysiology, growth, rupture, and therapeutic regression.
Methods
Transgenic Mice
All animal procedures were conducted according to the Institutional Animal Care and Use Committee guidelines at Barrow Neurological Institute and St. Joseph’s Hospital and Medical Center. The Alk12f alleles were established in laboratory mice using techniques described elsewhere.13 In brief, mutant mice were created with a Cre-Lox recombination system that flanks the DNA sequence of the Alk1 gene containing exon 5 encoding the transmembrane domain with loxP sites. Recombination of these loxP sequences results in excision of exon 5 and leads to elimination of ALK1 production from the mutant allele (conditional gene deletion). Conventional knockout of the Alk1 gene results in early embryonic lethality, whereas conditional knockout with the Cre-Lox system allows Alk1 genes to be deleted in only those cells in which Cre recombinase is expressed. Tagln-Cre+;Alk12f/2f mice were created by crossing Tagln-Cre mice (Jackson Laboratory, stock #004746) with Alk12f/2f mice. Mutant mice were crossed with R26mTmG/+ reporter mice (Jackson Laboratory, stock #007576) to identify Cre-positive cells. Mice used in this study were of a mixed (129Sv/C57BL6) background.
MRI
AVMs in mice were studied with brain MRI and 3D time of flight or MRA. Each mouse was induced with and maintained under isoflurane anesthesia (3% induction, 1%–2% maintenance) in medical air. Respiration was continually monitored via a pillow sensor positioned under the abdomen (SA Instruments), and normal body temperature (36°C–37°C) was maintained with a circulating warm water blanket (Thermo Fisher Scientific).
MRI was performed using a 7T small-animal, 30-cm horizontal-bore magnet and a BioSpec Avance III spectrometer (Bruker) with a 116-mm high-power gradient set (600 mT/m) and a 30-mm whole-body mouse quadrature coil. Fast spin-echo scout images were acquired in three orthogonal planes covering the brain (repetition time [TR] 1000 msec, echo time [TE] 12.5 msec, effective TE [TEeff] 50 msec, 128 × 128 matrix, 0.234 × 0.234 × 1.5–mm voxels, number of acquisitions [NEX] 1). The scout images were then used to determine the placement of T2-weighted and MRA images. Coronal T2-weighted rapid acquisition with relaxation enhancement images spanning the volume of brain were acquired to identify AVM lesions (TR 5000 msec, TE 12 msec, TEeff 60 msec, Rare Factor 8, 180 × 180 matrix, 0.1 × 0.1 × 0.5–mm voxels, 30 slices, NEX 4). The 3D time-of-flight MRA images were acquired using a flow-compensated gradient echo method (TR 30 msec, TE 3.5 msec, α = 30°, 180 × 180 × 200 matrix, 0.1 × 0.1 × 0.1–mm voxels, NEX 4). Magnetization transfer pulses were added to the MRA acquisition to provide additional background tissue suppression (offset frequency = 1500 Hz, radiofrequency amplitude = 0.5 μT).
DSA
DSA was performed on mice after induction of anesthesia using a mixture of ketamine hydrochloride (100 mg/kg body weight) and xylazine (10 mg/kg body weight) injected intraperitoneally. Respiration was monitored via a pillow sensor positioned under the abdomen, and normal body temperature was maintained with a circulating warm water blanket. The carotid artery was exposed through a midline incision in the neck. The common carotid artery (CCA), internal carotid artery, and external carotid artery were occluded with temporary aneurysm clips. An arteriotomy in the CCA was made with a 25-gauge needle and microarteriotomy scissors. The artery was then cannulated with a 1-Fr or 2-Fr polyurethane catheter. The catheter was advanced to the carotid bifurcation and secured with 5-0 silk sutures, and the distal temporary clip was removed. IsoVue 300 iodinated contrast (1 mL) was injected into the catheter, and high-resolution DSA images were obtained with C-arm image intensifier–based fluoroscopy.
Latex Dye Perfusion
Blue latex dye (Connecticut Valley Biological Supply Co.) was injected through the left ventricle of the heart after sequential perfusion as previously described elsewhere (5 µL/g body weight).14 After overnight fixation, the brains were isolated and cleared by organic solvents as described elsewhere.15 Cleared brains were sectioned coronally in 1-mm-thick slices using a brain slicer matrix (Zivic Instruments). Blood vessels and AVMs containing the latex dye in the cleared brain and sectioned brains were imaged with a charge-coupled device camera (Leica).
Histopathological Analysis and Brain Endothelial Cell Isolation
Detailed methods for histopathological analyses and brain endothelial cell isolation can be found in Supplementary Methods.
Results
Tagln-Cre;Alk12f/2f Transgenic Mice
Intercross between Tagln-Cre+;Alk12f/2f and Tagln-Cre−;Alk12f/2f mice generated 189 mice in 28 litters (mean litter size 6.6 mice). Genotype analysis of these mice at the weaning age (3–4 weeks after birth) identified 105 Tagln-Cre− and 84 Tagln-Cre+ mice, revealing that approximately 20% of Tagln-Cre+;Alk12f/2f mice died at embryonic or neonatal stages. An additional survival study conducted with 40 Cre− and 40 Cre+ mice from 1 month to 6 months of age showed that 35% of mice (14 of 40) died by 6 months of age, most likely due to bleeding from the gastrointestinal tract, lungs, and brain.12
Brain AVMs on MRI and MRA
The brains of control mice were normal on 7T high-resolution T2-weighted MRI in axial, coronal, and sagittal planes, without evidence of arteriovenous lesions or hemorrhage. Similarly, 2D MRA and 3D reconstructed MRA showed normal arterial anatomy of the anterior and posterior circulations and the circle of Willis, without evidence of arteriovenous lesions (Fig. 1A).
Representative 3D MR gradient echo flow compensation angiograms of control (TaglnCre−;Alk12f/2f) and mutant (TaglnCre+;Alk12f/2f) mice in axial, sagittal, and coronal views. A: MR angiogram obtained in a control mouse, showing normal arteries in the brain. B: MR angiogram obtained in a mutant mouse with no detectable arteriovenous lesion (group 1). C and D: MR angiograms obtained in mutant mice with parietal AVFs (group 2) with paramedian (C) and lateral (D) locations. E and F: MR angiograms obtained in mutant mice with nidal AVMs (group 3) with deep locations in the thalamus.
The brains of 55 Tagln-Cre+;Alk12f/2f mutant mice were studied with MRI and MRA and categorized into three groups. Group 1 had no detectable vascular lesions (23 of 55, 42%; Fig. 1B); group 2 had arteriovenous fistulas (AVFs) with no nidus (10 of 55, 18%; Fig. 1C and D); and group 3 had AVMs with a true nidus consisting of a stereotypical tangle of arteries and dilated veins within the brain parenchyma (22 of 55, 40%; Fig. 1E and F).
Table 1 reports the characteristics of these AVFs and AVMs with a nidus. AVMs in group 3 were observed in 9 female and 13 male mice and were more common in the left side (14 of 22, 64%). AVMs were located most commonly in the parietooccipital region (15 of 22, 68%), followed by deep regions (thalamus and basal ganglia; 4 of 22, 18%), the frontal lobe (2 of 22, 9%), and the brainstem (1 of 22, 5%). No AVMs were observed in the temporal lobe or cerebellum. The mean nidus size was 2.94 mm (range 1.12–6.40 mm). Venous drainage was superficial in 13 AVMs (59%), deep in 6 (27%), and both superficial and deep in 3 (14%). Microhemorrhage, detected on both MRI and 2D MRA as discrete areas of low signal on the gradient echo flow-compensated images, was observed in 11 AVMs (50%). Most brain AVMs were compact (19 of 22, 86%), and only 3 had a diffuse nidus (14%).
Detailed imaging findings of Tagln-Cre+;Alk12f/2f mutant mice with brain AVFs (group 2) and nidal AVMs (group 3)
| Mouse No. | Age (mos) | Sex | Side | Location | Subtype | Size (mm)* | Venous Drainage | Compactness |
|---|---|---|---|---|---|---|---|---|
| Group 2 | ||||||||
| 1 | 1.5 | M | Right | Frontal | Basal | 1.05 | Superficial | Compact |
| 2 | 1.7 | F | Right | Parietooccipital | Paramedian | 1.40 | Superficial | Compact |
| 3 | 3.8 | F | Right | Frontal | Lateral | 0.70 | Superficial | Compact |
| 4 | 5.4 | M | Right | Deep | Thalamic | 0.62 | Deep | Compact |
| 5 | 6.3 | F | Left | Frontal | Lateral | 0.80 | Superficial | Compact |
| 6 | 6.3 | F | Left | Parietooccipital | Paramedian | 0.62 | Superficial | Compact |
| 7 | 6.3 | M | Left | Frontal | Lateral | 0.35 | Superficial | Compact |
| 8 | 10.3 | F | Right | Frontal | Lateral | 0.93 | Superficial | Compact |
| 9 | 11 | F | Left | Parietooccipital | Lateral | 1.75 | Superficial | Compact |
| 10 | 13.8 | M | Left | Parietooccipital | Lateral | 0.62 | Superficial | Compact |
| Group 3 | ||||||||
| 1 | 1 | M | Right | Parietooccipital | Lateral | 2.45 | Superficial | Compact |
| 2 | 1 | F | Left | Parietooccipital | Paramedian | 4.67 | Superficial, deep | Compact |
| 3 | 1.5 | M | Left | Deep | Basal ganglia | 2.49 | Superficial, deep | Compact |
| 4 | 2 | M | Left | Parietooccipital | Lateral | 2.45 | Superficial | Compact |
| 5 | 2.3 | M | Left | Parietooccipital | Lateral | 2.10 | Superficial | Compact |
| 6 | 3.1 | M | Left | Parietooccipital | Lateral | 1.12 | Superficial | Compact |
| 7 | 3.3 | F | Middle | Deep, brainstem | Thalamic | 6.40 | Deep | Compact |
| 8 | 3.5 | F | Right | Parietooccipital | Basal | 2.80 | Deep | Compact |
| 9 | 3.5 | F | Right | Parietooccipital | Medial | 4.20 | Deep | Compact |
| 10 | 4.3 | M | Left | Frontal | Medial | 1.87 | Superficial | Compact |
| 11 | 4.3 | F | Left | Parietooccipital | Lateral | 2.00 | Superficial | Compact |
| 12 | 4.3 | M | Left | Deep | Insular | 2.18 | Deep | Diffuse |
| 13 | 4.5 | M | Left | Parietooccipital | Lateral | 3.42 | Superficial | Diffuse |
| 14 | 4.8 | F | Right | Parietooccipital | Paramedian | 2.80 | Superficial | Compact |
| 15 | 5.5 | M | Right | Frontal | Paramedian | 3.85 | Superficial | Compact |
| 16 | 7.6 | F | Left | Parietooccipital | Basal | 4.98 | Superficial, deep | Compact |
| 17 | 8 | M | Right | Parietooccipital | Lateral | 2.80 | Superficial | Compact |
| 18 | 8.2 | M | Left | Parietooccipital | Lateral | 2.80 | Superficial | Compact |
| 19 | 9.6 | M | Left | Brainstem | Pontine | 2.00 | Deep | Compact |
| 20 | 10.2 | F | Left | Parietooccipital | Lateral | 2.80 | Superficial | Diffuse |
| 21 | 14.7 | F | Middle | Deep | Thalamic | 1.96 | Deep | Compact |
| 22 | 15.5 | M | Left | Parietooccipital | Lateral | 2.49 | Superficial | Compact |
AVF size for group 2 and AVM nidus size for group 3.
Brain AVMs on DSA
DSA, the gold-standard imaging study for diagnosing brain AVMs clinically, validated the MRI and MRA findings in 3 control and 3 mutant mice. DSA in control mice demonstrated normal arterial anatomy of the anterior circulation and the circle of Willis without evidence of arteriovenous lesions. Mutant mice with AVMs fed by the carotid circulation were studied with contrast injection in one CCA on the side ipsilateral to the AVM, which limited imaging to the anterior circulation. Anteroposterior and lateral views visualized AVMs in all 3 mutant mice (2 of them are shown in Fig. 2). AVMs had the angiographic hallmarks of this lesion, with early opacification of the nidus, a tangle of arteries and dilated draining veins, and rapid shunting of blood flow to the draining vein in the arterial and arteriolar phase of the angiogram. In addition, AVM morphology visualized on DSA correlated precisely with AVM morphology visualized on MRI and MRA.
MR and digital subtraction angiograms obtained in 2 mutant mice, demonstrating nidal AVMs and arteriovenous shunting. A and B: Sagittal MR angiograms showing parietal (A) and frontal (B) AVMs with feeders (orange arrows) from the posterior cerebral artery (PCA) and middle cerebral artery (MCA), respectively. Lateral digital subtraction angiograms in sequential views (A1–5) show filling of the AVM nidus from PCA feeders and early opacification (orange arrows) of the draining vein, emptying into the superior sagittal sinus. Similarly, lateral digital subtraction angiograms in sequential views (B1–5) show filling of the AVM nidus from MCA feeders and early opacification (orange arrows) of the draining vein, emptying into the superior sagittal sinus. The time sequence of angiographic images is depicted in seconds after injection of the contrast.
Histological Analyses of Brain AVMs
Latex dye perfused through the left heart normally remains in arteries but is transmitted through AVFs and AVMs into the draining veins, which shows the arteriovenous pathology. This method validated the MRA findings in AVFs (group 2 [Fig. 3A]) and AVMs (group 3, small [Fig. 3B] and large [Fig. 3C] nidi) in Tagln-Cre+;Alk12f/2f mutants. Latex dye perfusion correlated with the MRA findings for location and size with AVFs and AVM nidi, thereby validating the accuracy of MRA.
Correlation of MRA (upper rows) and latex dye perfusion studies (lower rows). A left lateral parietal AVF (A; group 2), a right lateral parietal AVM (B; group 3), and a left large, deep AVM (C, group 3) are shown on axial, sagittal, and coronal MR angiograms. Latex dye–perfused brain in the same mice confirms the MRA findings, as seen in a top-down view of the whole brain (left) and low-magnification (center; orange outlines indicate area of magnification) and high-magnification (right) views of the coronal section. In all cases, latex dye perfusion studies validated the findings on MRA. White arrows indicate arteriovenous lesions. The planes of coronal sections are indicated by vertical dotted lines in the top views.
Histological analysis also validated findings on MRI. AVFs and AVMs observed on 3D MRA (Fig. 4A) and cross-sectional MRI (Fig. 4B) corresponded to a pathologic AVM nidus as well as dilated and tortuous draining veins in the brain parenchyma on histological sections (Fig. 4C). Microhemorrhage was detected adjacent to AVMs, as shown by iron deposition and Prussian blue staining of brain sections (Fig. 4C). There were no examples of AVM rupture with hematoma in the acute or subacute phase or encephalomalacia in the chronic phase. Immunostaining for macrophage infiltration using CD68 showed that macrophages were attracted to and accumulated in the brain adjacent to the AVMs in areas of microhemorrhage (Fig. 4D).
Histological validation of MRA. A: Three-dimensional MR angiogram showing a left thalamic AVM (orange arrows indicate nidus) in axial, sagittal, and coronal views. B: Two-dimensional MRI sections demonstrate this AVM (orange arrows indicate nidus) in corresponding axial, sagittal, and coronal views. White arrows indicate the microhemorrhagic lesion adjacent to the nidus. C: Coronal section of the brain showing the AVM in the thalamus in low-magnification (left; orange outline indicates area of magnification; original magnification ×10) and high-magnification (center; original magnification, ×50) views. White arrows indicate microhemorrhagic signs in the perivascular parenchyma of AVMs as hemosiderin staining, which was confirmed by Prussian blue staining (right; original magnification ×50). D: CD68-positive macrophage/monocytes (yellow outlines indicate areas of magnification; left, original magnification ×40; center and right, original magnification, ×100) were detected in the perivascular parenchyma, indicating an inflammatory response to AVM microhemorrhage.
Cellular Source of Brain AVMs
We previously used L1Cre mice to delete the Alk1 gene in endothelial cells.15 L1Cre+;Alk12f/2f mice developed widespread AVMs in multiple areas of the brain, resulting in lethality within 1 week after birth. In contrast, Tagln-Cre+;Alk12f/2f mice develop brain AVMs in focal areas, similar to human clinical cases. The Tagln promoter is known to drive expression in smooth muscle cells.16 If Alk1 deletion in vascular smooth muscle cells is the cause of AVMs, we would expect much more widespread AVMs. Focal AVMs were thought to be due to mosaic expression of Cre recombinase in a small population of endothelial cells.12 To test whether Cre was active in a subset of brain endothelial cells in Tagln-Cre mice, we used the mTmG reporter mice. Cre-negative cells were marked by red fluorescence (tdTomato or mT), and Cre-positive cells were marked by green fluorescence (mG). Brain endothelial cells were isolated from Tagln-Cre;mTmG mice. CD31 staining confirmed that the majority of these cells were endothelial cells (Fig. 5A). Fluorescence imaging showed that approximately 37% of these endothelial cells were GFP positive (mean [SD] 37% [10%], n = 5; Fig. 5B–D), indicating that Cre recombinase expression and subsequent gene deletion are present in limited endothelial cells.
Identification of Cre-positive endothelial cells in the brains of Tagln-Cre mice (original magnification ×100). A: Endothelial cells isolated from the brain of Tagln-Cre;mTmG mice are CD31 positive. B–D: Mixed population of tdTomato red-positive (Cre-negative) and (C) GFP-positive (Cre-positive) cells were found in brain endothelial cells, indicating that Cre is active in a limited population of brain endothelial cells of Tagln-Cre mice, as shown in the merged view (D).
Brain AVM Stability Over Time
Brain AVM stability was assessed using serial MRI and MRA in all 22 mutant mice in group 3. Initial imaging was performed at 3 months of age (mean 2.9 months, range 1–10 months) and every 1–4 months thereafter. The last imaging sessions were performed at a mean age of 9.5 months (range 3–17 months). Two examples demonstrate a large right parietal AVM draining into the superior sagittal sinus (Fig. 6) and a large right paramedian frontal AVM draining into the superior sagittal sinus (Fig. 7). The location and general morphology of these AVMs were stable over time.
Longitudinal stability of a nidal AVM. This right frontal AVM is shown on MR angiograms in axial, sagittal, and coronal views at 6, 8, and 10 months.
Longitudinal stability of a nidal AVM. This right frontal AVM is shown on MR angiograms in axial, sagittal, and coronal views at 2, 7, 10, and 12 months.
Discussion
A High-Fidelity Experimental AVM Model, Finally
The lack of an animal model has limited our basic understanding of brain AVMs. This pathology is not seen in animals, and the arteriovenous angioarchitecture is not easily recreated in an animal model. In this study, we developed a model of brain AVM that uses transgenic mice and conditional gene deletion of Alk1 to generate AVMs with the essential features of the human disease. Alk1-floxed mice were crossed with SM22α/Tagln-Cre transgenic mice. Although the Cre recombinase is primarily expressed in smooth muscle cells, our studies demonstrated Cre recombinase expression in a subset of endothelial cells in the brain, enabling Alk1 deletion in the cell population responsible for AVM formation. Expression of Cre recombinase and the deletion of both copies of the Alk1 gene in endothelial cells are thought to lead to the formation of arteriovenous pathology, as observed in patients with HHT.15,17 HHT is an autosomal dominant disorder caused by the mutation of ALK1 (HHT2) on 12q13; however, it is also caused by mutations of ENG (HHT1) on 9q34 or SMAD4 on 18q21 (juvenile polyposis and HHT overlap syndrome).4,5,18 Patients with this familial syndrome have mucocutaneous telangiectasias and multiple AVMs in the lungs, liver, or brain.19,20 HHT has been used as a window of opportunity to discover the underlying genetics and pathobiology of brain AVMs.
The brain vascular malformations in patients with HHT were recently categorized by Krings et al.21 into three distinct lesions based on the following neurovascular imaging features: 1) capillary vascular malformations, which are small, superficial, nonshunting conglomerates of vessels without enlarged feeding arteries or draining veins; 2) AVFs, which are high-flow shunts with enlarged feeding arteries and draining veins but without a nidus; and 3) nidal AVMs, which are high-flow shunts with a tangle of arteries and veins and enlarged feeding arteries and draining veins. The vascular malformations observed in our transgenic mouse model resembled those observed in patients with HHT, with group 2 mice having AVFs and group 3 mice having nidal AVMs. The 40% incidence of nidal AVMs in our experiments is similar to the 43% incidence of nidal AVMs in patients with HHT, and the 18% incidence of AVFs in mice is similar to the 12% incidence in humans. Capillary vascular malformations were not imaged in our study, but these tiny lesions may have been present but undetectable in group 1 mice.
Nidal AVMs in our model had radiographic characteristics on MRI, MRA, and DSA identical to AVMs in patients observed clinically. AVMs appeared as a nidus of vessels with flow signal voids on MRI, and DSA demonstrated early opacification of the nidus with visualization of draining veins in the arterial or arteriolar phase on angiography. Even though the size of the mouse brain is a fraction of that of the human brain, nidal AVMs could be identified, localized anatomically, and measured. The venous drainage pattern and compactness could have been characterized using the Spetzler-Martin and Lawton-Young supplementary grades if analogous categories for size and age were appropriately downscaled for mice for assigning points in these grading systems.22,23 Radiographic findings of AVMs in our model were confirmed by latex dye perfusion, pathological staining, and immunohistochemical analysis. Therefore, approximately 40% of transgenic mice (group 3) satisfied the following requirements of an experimental AVM model: 1) They replicated AVM anatomy with a nidus of abnormal arteries connected directly to veins without intervening capillaries. 2) They replicated AVM physiology and hemodynamics with arteriovenous shunting. 3) They behaved biologically like human brain AVMs, as evidenced histologically by microhemorrhages. 4) They also remained stable in a viable host for a longitudinal study. Transgenic mice with AVFs (group 2) had a recognizable HHT phenotype. However, these mice might not be as suitable for experimental modeling as those with nidal AVMs.
AVM Pathogenesis
AVMs have long been understood as congenital lesions resulting from aberrant angiogenesis in the developing fetus, consistent with their description as "malformations." During embryogenesis, developing arteries and veins are in direct contact, and miscues or miscommunications could lead to persistent arteriovenous architecture and AVM formation. Our experimental results align with this congenital pathogenesis. Systemic inactivation of Alk1 or Eng in transgenic mice results in AVM formation, but the embryos do not survive. Regional or tissue-specific conditional gene deletion in our mouse model prevents embryonic death. AVMs are observed in young mice on their initial imaging as early as 1 month after birth. The critical pathogenetic events appear to be Alk1 deletion with loss of gene function in the setting of the increased angiogenesis activity associated with embryogenesis.
Angiogenesis has been implicated in other experimental models of arteriovenous pathology. DAVF pathogenesis is understood as an acquired lesion following sinus thrombosis after trauma, surgery, infection, or spontaneous causes.24 Experimentally, venous hypertension, produced by a combination of microsurgical common carotid artery–external jugular vein anastomosis, induced superior sagittal sinus thrombosis, and occlusion of mastoid emissary draining veins elicits DAVF formation in rats25,26 due to overexpression of VEGF and hypoxia-inducible factor-1.27 Translating these findings to brain AVMs, angiogenesis stimulation combined with conditional gene deletion of Alk1 led to brain AVM formation in adult Alk1-floxed transgenic mice treated with stereotactic injections of adenoviral vector with genes expressing Cre recombinase and adenovirus-associated viral vector with genes expressing VEGF.28 This model led to the "response-to-injury" hypothesis that brain AVMs result from an underlying genetic mutation, often a heritable allele loss plus a “second hit” with loss of heterozygosity, and overlying angiogenesis stimulation, most commonly due to injury and endogenous inflammation.29 The healing response normally produces stable neovasculature, but aberrant signaling in the ALK1 or ENG pathways produces dysplasia followed by arteriovenous pathology. Our model supports this response-to-injury hypothesis because angiogenesis is hyperactive in developing and newborn mice. Conditional Alk1 gene deletion occurs during embryogenesis in our transgenic mice. The overlying angiogenesis elicits AVFs and nidal AVMs in the absence of specific injury. The response-to-injury hypothesis allows us to understand AVM pathogenesis as congenital in most patients but acquired in those unusual patients with a confluence of genetic and angiogenic events later in life.
Applications
In addition to contributing to the study of pathogenesis, this novel experimental brain AVM model facilitates studies of hemorrhagic behavior and response to therapeutic and pharmacological interventions. The AVM stability and extended mouse survival confirmed in this study are needed to examine the longitudinal effects of experimental manipulation. AVM microhemorrhage was consistently observed adjacent to AVMs in mice, but frank rupture was not. AVMs in this mouse model share the genetics of patients with HHT, and AVMs in these patients have a lower annual hemorrhage rate than those in patients with sporadic AVMs. Kim et al.30 and the Brain Vascular Malformation Consortium found an annual rupture rate of 1.02% in 153 patients with HHT and brain AVMs. This rate is half the annual rupture rate of 2% reported in the ARUBA (A Randomized Trial of Unruptured Brain Arteriovenous Malformations) trial characterizing sporadic AVMs.30,31 The lack of ruptured AVMs in group 3 mice is consistent with this low hemorrhage rate observed clinically. The absence of rupture may be a limitation of this model, but microhemorrhage predicts future rupture and, therefore, may be a meaningful measure of hemorrhage in the model.32,33 Alternatively, provocative interventions, such as induced hypertension, may increase the rupture rate. Therapeutic AVM regression after pharmacological intervention is another important objective of AVM research, and this transgenic mouse model will facilitate preclinical evaluations of novel therapies for brain AVMs.2,34,35 Suppression of transgene expression has been shown to cause regression of enlarged, tortuous vessels in an AVM mouse model.36
Only approximately 5%–10% of brain AVMs are familial and related to HHT;37 the rest are sporadic with a different genetic etiology. A recent study identified somatic activating KRAS mutations in brain AVMs in 45 (63%) of 72 patients.38 Several independent studies have validated this finding in sporadic brain AVM specimens.39–42 There may be common downstream targets between ENG-ALK1 and KRAS signaling pathways, but these mechanisms are unclear. Recently, Fish et al.43 demonstrated that expression of mutant KRAS protein in brain endothelial cells could induce AVMs. Interestingly, they reported that the expression of mutant KRAS protein in the adult brain vasculature without angiogenic stimulation was sufficient to induce AVMs. These findings contrast with the model of familial brain AVMs that require angiogenic stimulation plus the deletion of Alk1 or Eng to induce brain AVMs.28,44–47 Transgenic mouse models based on the genetics of familial AVMs may not be suited to studying the etiology of sporadic AVMs. Still, they should be suited to studying AVM hemodynamics and the biology of rupture more generally. Sporadic and familial brain AVMs have clinical differences that may be maintained in mouse models, such as their size and superficiality as well as their hemorrhagic behavior. Better characterization and comparison studies are needed. We now have robust experimental systems that should advance our study of the biology and management of brain AVMs.
Conclusions
Approximately 40% of transgenic mice satisfied the requirements of a stable experimental AVM model by replicating nidal anatomy, arteriovenous hemodynamics, and microhemorrhagic behavior. Transgenic mice with AVFs had a recognizable HHT phenotype but were less suitable for experimental modeling. AVM pathogenesis can be understood as the combination of conditional Alk1 gene deletion during embryogenesis and angiogenesis that is hyperactive in developing and newborn mice. This study offers a novel experimental brain AVM animal model for future studies of AVM pathophysiology, growth, rupture, and therapeutic regression.
Acknowledgments
This work was supported by grants from the Barrow Neurological Foundation (BNF), Leducq Foundation (ATTRACT), the US Department of Defense (PR161205), and the National Institutes of Health (HL128525) to S.P.O. and a BNF postdoctoral fellowship to C.H.
We thank S. C. Ma, X. Zhang, and the Barrow Neurological Institute–Arizona State University Center for Preclinical Imaging for technical assistance with MRI. We thank the staff of Neuroscience Publications at Barrow Neurological Institute for assistance with manuscript preparation.
Disclosures
The authors report no conflict of interest concerning the materials or methods used in this study or the findings specified in this paper.
Author Contributions
Conception and design: Lawton, Han, Oh. Acquisition of data: all authors. Analysis and interpretation of data: all authors. Drafting the article: Lawton, Han, Oh. Critically revising the article: Lawton, Han, Oh. Reviewed submitted version of manuscript: Lawton, Han, Oh. Statistical analysis: Lawton, Han, Oh. Administrative/technical/material support: Lawton, Han, Oh. Study supervision: Lawton, Han, Oh.
Supplemental Information
Online-Only Content
Supplemental material is available with the online version of the article.
Supplementary Methods. https://thejns.org/doi/suppl/10.3171/2021.6.JNS21717.
Previous Presentations
This work was presented in part at the American Academy of Neurological Surgery 81st Annual Scientific Meeting, Rome, Italy, September 19, 2019.
References
- 1↑
Tu J, Karunanayaka A, Windsor A, Stoodley MA. Comparison of an animal model of arteriovenous malformation with human arteriovenous malformation. J Clin Neurosci. 2010;17(1):96–102.
- 2↑
Lawton MT, Jacobowitz R, Spetzler RF. Redefined role of angiogenesis in the pathogenesis of dural arteriovenous malformations. J Neurosurg. 1997;87(2):267–274.
- 3↑
Vates GE, Hashimoto T, Young WL, Lawton MT. Angiogenesis in the brain during development: the effects of vascular endothelial growth factor and angiopoietin-2 in an animal model. J Neurosurg. 2005;103(1):136–145.
- 4↑
McAllister KA, Grogg KM, Johnson DW, Gallione CJ, Baldwin MA, Jackson CE, et al. Endoglin, a TGF-β binding protein of endothelial cells, is the gene for hereditary haemorrhagic telangiectasia type 1. Nat Genet. 1994;8(4):345–351.
- 5↑
Johnson DW, Berg JN, Baldwin MA, Gallione CJ, Marondel I, Yoon SJ, et al. Mutations in the activin receptor-like kinase 1 gene in hereditary haemorrhagic telangiectasia type 2. Nat Genet. 1996;13(2):189–195.
- 6↑
Goumans MJ, Liu Z, ten Dijke P. TGF-beta signaling in vascular biology and dysfunction. Cell Res. 2009;19(1):116–127.
- 7
Oh SP, Seki T, Goss KA, Imamura T, Yi Y, Donahoe PK, et al. Activin receptor-like kinase 1 modulates transforming growth factor-β1 signaling in the regulation of angiogenesis. Proc Natl Acad Sci U S A. 2000;97(6):2626–2631.
- 8
Bourdeau A, Dumont DJ, Letarte M. A murine model of hereditary hemorrhagic telangiectasia. J Clin Invest. 1999;104(10):1343–1351.
- 9
Li DY, Sorensen LK, Brooke BS, Urness LD, Davis EC, Taylor DG, et al. Defective angiogenesis in mice lacking endoglin. Science. 1999;284(5419):1534–1537.
- 10
Arthur HM, Ure J, Smith AJ, Renforth G, Wilson DI, Torsney E, et al. Endoglin, an ancillary TGFβ receptor, is required for extraembryonic angiogenesis and plays a key role in heart development. Dev Biol. 2000;217(1):42–53.
- 11↑
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(3):197–201.
- 12↑
Milton I, Ouyang D, Allen CJ, Yanasak NE, Gossage JR, Alleyne CH Jr, Seki T. Age-dependent lethality in novel transgenic mouse models of central nervous system arteriovenous malformations. Stroke. 2012;43(5):1432–1435.
- 13↑
Park SO, Lee YJ, Seki T, Hong KH, Fliess N, Jiang Z, et al. ALK5- and TGFBR2-independent role of ALK1 in the pathogenesis of hereditary hemorrhagic telangiectasia type 2. Blood. 2008;111(2):633–642.
- 14↑
Han C, Choe SW, Kim YH, Acharya AP, Keselowsky BG, Sorg BS, et al. VEGF neutralization can prevent and normalize arteriovenous malformations in an animal model for hereditary hemorrhagic telangiectasia 2. Angiogenesis. 2014;17(4):823–830.
- 15↑
Park SO, Wankhede M, Lee YJ, Choi EJ, Fliess N, Choe SW, et al. Real-time imaging of de novo arteriovenous malformation in a mouse model of hereditary hemorrhagic telangiectasia. J Clin Invest. 2009;119(11):3487–3496.
- 16↑
Holtwick R, Gotthardt M, Skryabin B, Steinmetz M, Potthast R, Zetsche B, et al. Smooth muscle-selective deletion of guanylyl cyclase-A prevents the acute but not chronic effects of ANP on blood pressure. Proc Natl Acad Sci U S A. 2002;99(10):7142–7147.
- 17↑
Garrido-Martin EM, Nguyen HL, Cunningham TA, Choe SW, Jiang Z, Arthur HM, et al. 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(10):2232–2236.
- 18↑
Gallione CJ, Repetto GM, Legius E, Rustgi AK, Schelley SL, Tejpar S, et al. A combined syndrome of juvenile polyposis and hereditary haemorrhagic telangiectasia associated with mutations in MADH4 (SMAD4). Lancet. 2004;363(9412):852–859.
- 19↑
Guttmacher AE, Marchuk DA, White RI Jr. Hereditary hemorrhagic telangiectasia. N Engl J Med. 1995;333(14):918–924.
- 20↑
Govani FS, Shovlin CL. Hereditary haemorrhagic telangiectasia: a clinical and scientific review. Eur J Hum Genet. 2009;17(7):860–871.
- 21↑
Krings T, Kim H, Power S, Nelson J, Faughnan ME, Young WL, terBrugge KG. Neurovascular manifestations in hereditary hemorrhagic telangiectasia: imaging features and genotype-phenotype correlations. AJNR Am J Neuroradiol. 2015;36(5):863–870.
- 22↑
Spetzler RF, Martin NA. A proposed grading system for arteriovenous malformations. J Neurosurg. 1986;65(4):476–483.
- 23↑
Lawton MT, Kim H, McCulloch CE, Mikhak B, Young WL. A supplementary grading scale for selecting patients with brain arteriovenous malformations for surgery. Neurosurgery. 2010;66(4):702–713.
- 24↑
Gupta A, Periakaruppan A. Intracranial dural arteriovenous fistulas: a review. Indian J Radiol Imaging. 2009;19(1):43–48.
- 25↑
Terada T, Higashida RT, Halbach VV, Dowd CF, Tsuura M, Komai N, et al. Development of acquired arteriovenous fistulas in rats due to venous hypertension. J Neurosurg. 1994;80(5):884–889.
- 26↑
Herman JM, Spetzler RF, Bederson JB, Kurbat JM, Zabramski JM. Genesis of a dural arteriovenous malformation in a rat model. J Neurosurg. 1995;83(3):539–545.
- 27↑
Zhu Y, Lawton MT, Du R, Shwe Y, Chen Y, Shen F, et al. Expression of hypoxia-inducible factor-1 and vascular endothelial growth factor in response to venous hypertension. Neurosurgery. 2006;59(3):687–696.
- 28↑
Walker EJ, Su H, Shen F, Choi EJ, Oh SP, Chen G, et al. Arteriovenous malformation in the adult mouse brain resembling the human disease. Ann Neurol. 2011;69(6):954–962.
- 29↑
Kim H, Su H, Weinsheimer S, Pawlikowska L, Young WL. Brain arteriovenous malformation pathogenesis: a response-to-injury paradigm. Acta Neurochir Suppl (Wien).2011;111:83–92.
- 30↑
Kim H, Nelson J, Krings T, terBrugge KG, McCulloch CE, Lawton MT, et al. Hemorrhage rates from brain arteriovenous malformation in patients with hereditary hemorrhagic telangiectasia. Stroke. 2015;46(5):1362–1364.
- 31↑
Mohr JP, Parides MK, Stapf C, Moquete E, Moy CS, Overbey JR, et al. Medical management with or without interventional therapy for unruptured brain arteriovenous malformations (ARUBA): a multicentre, non-blinded, randomised trial. Lancet. 2014;383(9917):614–621.
- 32↑
Guo Y, Saunders T, Su H, Kim H, Akkoc D, Saloner DA, et al. Silent intralesional microhemorrhage as a risk factor for brain arteriovenous malformation rupture. Stroke. 2012;43(5):1240–1246.
- 33↑
Abla AA, Nelson J, Kim H, Hess CP, Tihan T, Lawton MT. Silent arteriovenous malformation hemorrhage and the recognition of "unruptured" arteriovenous malformation patients who benefit from surgical intervention. Neurosurgery. 2015;76(5):592–600.
- 34↑
Du R, Hashimoto T, Tihan T, Young WL, Perry V, Lawton MT. Growth and regression of arteriovenous malformations in a patient with hereditary hemorrhagic telangiectasia. Case report. J Neurosurg. 2007;106(3):470–477.
- 35↑
Lawton MT, Stewart CL, Wulfstat AA, Derugin N, Hashimoto T, Young WL. The transgenic arteriovenous fistula in the rat: an experimental model of gene therapy for brain arteriovenous malformations. Neurosurgery. 2004;54(6):1463–1471.
- 36↑
Murphy PA, Kim TN, Lu G, Bollen AW, Schaffer CB, Wang RA. Notch4 normalization reduces blood vessel size in arteriovenous malformations. Sci Transl Med. 2012;4(117):117ra8.
- 37↑
Brinjikji W, Iyer VN, Wood CP, Lanzino G. Prevalence and characteristics of brain arteriovenous malformations in hereditary hemorrhagic telangiectasia: a systematic review and meta-analysis. J Neurosurg. 2017;127(2):302–310.
- 38↑
Nikolaev SI, Vetiska S, Bonilla X, Boudreau E, Jauhiainen S, Rezai Jahromi B, et al. Somatic activating KRAS mutations in arteriovenous malformations of the brain. N Engl J Med. 2018;378(3):250–261.
- 39
Hong T, Yan Y, Li J, Radovanovic I, Ma X, Shao YW, et al. High prevalence of KRAS/BRAF somatic mutations in brain and spinal cord arteriovenous malformations. Brain. 2019;142(1):23–34.
- 40
Oka M, Kushamae M, Aoki T, Yamaguchi T, Kitazato K, Abekura Y, et al. KRAS G12D or G12V mutation in human brain arteriovenous malformations. World Neurosurg. 2019;126:e1365–e1373.
- 41
Goss JA, Huang AY, Smith E, Konczyk DJ, Smits PJ, Sudduth CL, et al. Somatic mutations in intracranial arteriovenous malformations. PLoS One. 2019;14(12):e0226852.
- 42
Priemer DS, Vortmeyer AO, Zhang S, Chang HY, Curless KL, Cheng L. Activating KRAS mutations in arteriovenous malformations of the brain: frequency and clinicopathologic correlation. Hum Pathol. 2019;89:33–39.
- 43↑
Fish JE, Flores Suarez CP, Boudreau E, Herman AM, Gutierrez MC, Gustafson D, et al. Somatic gain of KRAS function in the endothelium is sufficient to cause vascular malformations that require MEK but not PI3K signaling. Circ Res. 2020;127(6):727–743.
- 44
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(2):e88511.
- 45
Chen W, Guo Y, Walker EJ, Shen F, Jun K, Oh SP, et al. Reduced mural cell coverage and impaired vessel integrity after angiogenic stimulation in the Alk1-deficient brain. Arterioscler Thromb Vasc Biol. 2013;33(2):305–310.
- 46
Chen W, Sun Z, Han Z, Jun K, Camus M, Wankhede M, et al. De novo cerebrovascular malformation in the adult mouse after endothelial Alk1 deletion and angiogenic stimulation. Stroke. 2014;45(3):900–902.
- 47
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(6):540–547.
