Thalidomide Reduces Hemorrhage of Brain Arteriovenous Malformations in a Mouse Model
Background and Purpose—Brain arteriovenous malformation (bAVM) is an important risk factor for intracranial hemorrhage. Current treatments for bAVM are all associated with considerable risks. There is no safe method to prevent bAVM hemorrhage. Thalidomide reduces nose bleeding in patients with hereditary hemorrhagic telangiectasia, an inherited disorder characterized by vascular malformations. In this study, we tested whether thalidomide and its less toxic analog, lenalidomide, reduce bAVM hemorrhage using a mouse model.
Methods—bAVMs were induced through induction of brain focal activin-like kinase 1 (Alk1, an AVM causative gene) gene deletion and angiogenesis in adult Alk1-floxed mice. Thalidomide was injected intraperitoneally twice per week for 6 weeks, starting either 2 or 8 weeks after AVM induction. Lenalidomide was injected intraperitoneally daily starting 8 weeks after AVM induction for 6 weeks. Brain samples were collected at the end of the treatments for morphology, mRNA, and protein analyses. The influence of Alk1 downregulation on PDGFB (platelet-derived growth factor B) expression was also studied on cultured human brain microvascular endothelial cells. The effect of PDGFB in mural cell recruitment in bAVM was explored by injection of a PDGFB overexpressing lentiviral vector to the mouse brain.
Results—Thalidomide or lenalidomide treatment reduced the number of dysplastic vessels and hemorrhage and increased mural cell (vascular smooth muscle cells and pericytes) coverage in the bAVM lesion. Thalidomide reduced the burden of CD68+ cells and the expression of inflammatory cytokines in the bAVM lesions. PDGFB expression was reduced in ALK1-knockdown human brain microvascular endothelial cells and in mouse bAVM lesion. Thalidomide increased Pdgfb expression in bAVM lesion. Overexpression of PDGFB mimicked the effect of thalidomide.
Conclusions—Thalidomide and lenalidomide improve mural cell coverage of bAVM vessels and reduce bAVM hemorrhage, which is likely through upregulation of Pdgfb expression.
Brain arteriovenous malformation (bAVM) is an important risk factor for intracranial hemorrhage.1 Many patients are not treated because of high risks associated with currently available interventions.2 There is no specific medical treatment available for bAVM patients.
The causative gene for sporadic bAVM is still unknown. Approximately 5% of the bAVM patients have a genetic disorder called hereditary hemorrhagic telangiectasia (HHT). The 2 main subtypes of HHT (1 and 2) are caused by mutations in the endoglin (ENG) gene or the activin receptor-like kinase 1 (ALK1 or ACVLR1) gene.1 The familial forms of the more common sporadic disorders have been used to study the disease mechanisms of sporadic cerebrovascular diseases.3 We took this concept to consider HHT as a familial form of the more common sporadic bAVM, as HHT bAVM possesses a phenotype that is similar to sporadic bAVM. Knowledge of the inherited disease can shed light on sporadic bAVM pathogenesis.
We have established several adult-onset bAVM mouse models through conditional knockout of either Eng or Alk1 gene and focal brain angiogenic stimulation.4–7 These bAVM models have several key phenotypes resembling those of human bAVMs, such as dilated tortuous vessels, direct arteriovenous shunts,4–7 hemorrhage, macrophage infiltration, and reduced mural cell coverage.8,9
In the Alk1-deficient bAVM mouse model, many AVM vessels do not have vascular smooth muscle cells (vSMCs) and have fewer pericytes than normal brain vessels, which are associated with vascular leakage and hemorrhage.4,5 PDGFB (platelet-derived growth factor B) and PDGFR-β (platelet-derived growth factor receptor-β) signaling plays an important role in promoting pericyte and vSMC recruitment to endothelial tubes during vascular maturation. We found that Pdgfrβ expression in the bAVM is reduced,5 suggesting that Pdgfb/pdgfrβ signaling might be impaired in bAVM and may be a potential therapeutic target.
Thalidomide belongs to a class termed immunomodulatory drugs. As a result of thalidomide’s well-known adverse effects, for example, peripheral neuropathy and drowsiness,10 less-toxic second-generation immunomodulatory drugs, such as lenalidomide, have been identified.11 Thalidomide inhibits gastrointestinal bleeding and stabilizes telangiectasia vessels in HHT patients, increases Pdgfb expression, and improves mural cell recruitment in the retina of Eng+/− mice.12 However, it is not clear whether these drugs can stabilize bAVM vessels.
In this study, we tested whether thalidomide and lenalidomide can reduce bAVM hemorrhage using an Alk1 bAVM model and found that both agents increased vascular pericyte and vSMC coverage and reduced bAVM hemorrhage. In addition, overexpression of PDGFB in the bAVM showed a similar effect to thalidomide, suggesting that PDGFB is the critical factor responsible for the beneficial effects of thalidomide.
All data and supporting materials are available with the article and its online-only Data Supplement.
A total of 148, 8- to 10-week-old Alk12f/2f mice13 in C57BL background with locus of crossover in P1 sites flanking exons 4 to 6 and 4 wild-type C57BL mice (The Jackson Laboratory, Bar Harbor, ME) were used. Equal numbers of male and female mice were included. Experimental procedures for using laboratory animals were approved by the Institution of Animal Care and Use Committee of the University of California, San Francisco.
Injection of Viral Vectors Into Mouse Brain
Mice were anesthetized through inhalation of 4% isoflurane and placed in a stereotactic frame with a holder (David Kopf Instruments, Tujunga, CA). A burr hole was drilled in the pericranium 2 mm lateral to the sagittal suture and 1 mm posterior to the coronal suture. A total of 2 μL viral vectors were stereotactically injected into the basal ganglia4 3 mm beneath the brain surface. Mice were randomly assigned to each treatment groups.
For mice with thalidomide treatment starting 2 weeks after model induction, 2×109 genome copies (gc) of AAV-VEGF (an adeno-associated viral vector expressing human vascular endothelial growth factor) and 1×107 plaque-forming units of Ad-Cre (an adenoviral vector expressing Cre recombinase) were injected. Because the level of VEGF has been shown to be correlated with bAVM hemorrhage,14 to induce significant hemorrhage in established bAVM, we used a higher AAV1-VEGF dose (5×109 gc) to induce bAVM in the mice that have thalidomide and lenalidomide treatment started 8 weeks after model induction. For overexpression of human PDGFB, 2×109 gc of AAV1-VEGF, 1×107 plaque-forming units of Ad-Cre or Ad-GFP (an adenovirus encoding green fluorescent protein; control), and 1.5×109 gc of lentivirus (Lenti)-PDGFB or Lenti-GFP (control) were used. Ad-Cre and Ad-GFP were purchased from Vector Biolabs (catalog number 1700 and catalog number 1060, respectively; Malvern, PA), and AAV-VEGF was made by our laboratory using method described previously.15 Lenti-PDGFB and Lenti-GFP were purchased from GeneCopoeia (catalog number LP-EGFP-Lv105 and LP-A0380-Lv105-0200-S; Rockville, MD).
Thalidomide (75 mg/kg of body weight; Sigma-Aldrich, St Louis, MO) or dimethyl sulfoxide (DMSO; vehicle) was administered intraperitoneally twice a week for 6 weeks starting 2 or 8 weeks after model induction (Figure IA and IB in the online-only Data Supplement). Lenalidomide (50 mg/kg of body weight; Hangzhou ICH Biopham Co, Ltd, Hangzhou, China) or DMSO was administered intraperitoneally daily for 6 weeks starting 8 weeks after model induction (Figure IC in the online-only Data Supplement). Body weight was measured weekly or biweekly as indicated.
Dissection of Basal Ganglia Microvessels
After the mice were anesthetized with isoflurane, the brain samples were collected and immersed immediately in Hanks buffer supplemented with 15 mmol/L HEPES. Brain tissue around vector injection site was dissected and homogenized using a Dounce grinder in Hanks buffer supplemented with 15 mmol/L HEPES, 0.5% (wt/vol) BSA, 10 mmol/L glucose, 20 mmol/L sodium bicarbonate, and 1 mmol/L sodium pyruvate (pH 7.4). Homogenates were centrifuged in 15% dextran at 6000g for 20 minutes at 4°C. Supernatant was removed. Pellets (microvessels) were resuspended in RNAzol (Molecular Research Center, Cincinnati, OH) for RNA extraction. Because of the limited amount of RNA that can be extracted from the microvessels collected from the viral injection region of individual mouse, samples from 4 to 9 similar treated animals were pooled.
The quantification of vessel densities, dysplastic vessels, pericytes and vSMC coverage, and Prussian blue positive areas was done by researchers who were blinded to the treatment groups on sections with section numbers scrambled. Data are presented as mean±SD. Because of the skewed nature of the Prussian blue–positive area, the data observations were log-transformed before the analysis. All data were analyzed using 1-way ANOVA followed by Sidak multiple comparisons or Student t test to compare the means of 2 groups. A P value <0.05 was considered to be significant. Sample sizes were indicated in the Figure legends.
Additional methodologies are described in the online-only Data Supplement.
Thalidomide Treatment Inhibits bAVM Development
The AVMs are formed 8 weeks after induction of Alk1 deletion and angiogenesis.4 To test whether thalidomide inhibits bAVM formation, thalidomide treatment was started 2 weeks after AVM induction at the onset of AAV-VEGF–induced brain angiogenesis16 and continued for 6 weeks. The bAVM phenotypes were analyzed after the completion of the treatment (Figure IA in the online-only Data Supplement). The thalidomide and DMSO groups have similar vascular densities (P=0.11). However, the thalidomide group had fewer dysplasia vessels (3.38±0.83/200 vessels) than that of DMSO group (7.74±1.85; P=0.004; Figure 1). In addition, the thalidomide group has more vessels completely covered by vSMCs (39.1±13.1%) than DMSO group (23±6.5%; P=0.02) and fewer vSMC-negative vessels (11.6±6%) than that of DMSO group (47.1±11.6%; P=0.007; Figure 1A and 1D). These data suggest that thalidomide reduces the formation of dysplastic vessels and may promote vSMC recruitment.
Thalidomide and Lenalidomide Increase Mural Cell Coverage and Reduce Hemorrhage in Established bAVMs
To test whether thalidomide and lenalidomide reduce hemorrhage of established bAVM, we treated mice 8 weeks after the model induction when bAVMs have fully developed (Figure IB and IC in the online-only Data Supplement).4 Because abnormally high level of VEGF has been shown to be associated with bAVM hemorrhage,14 to increase the severity of bAVM hemorrhage, we used a higher dose of AAV-VEGF (5×109 gc) to induce the bAVM phenotype.
Thalidomide or lenalidomide did not alter vessel density (P=0.61) and the number of abnormal vessels (P=0.1; Figure II in the online-only Data Supplement) in the established bAVMs. However, the treatments improved vSMC coverage in the bAVMs. Compared with DMSO groups, the thalidomide group has more vessels that are completely covered by vSMCs (DMSO versus thalidomide: 9.5±6.3% versus 22.9±8.6%; P=0.03), and the lenalidomide group showed a trend of increase of vSMCs coverage (DMSO versus lenalidomide: 8.4±7.7 versus 22.1±9.8; P=0.06). Thalidomide- and lenalidomide-treated groups also have fewer vSMC-negative vessels than DMSO groups (DMSO versus thalidomide: 71.6±12.3% versus 49.9±12.3%; P=0.007; DMSO versus lenalidomide: 73.6±24.1% versus 49.9±10.7%; P=0.004; Figure 2A and 2B; Figure III in the online-only Data Supplement).
We have also analyzed pericyte coverage. In the hemisphere contralateral to the bAVM lesion, the pericyte coverage was 77±5%, which is similar to that reported for normal brain.17 In bAVMs, pericyte coverage was reduced (59±4.6%, DMSO group; P=0.004). Thalidomide treatment restored pericyte coverage (77±2.2%; P=0.004; Figure 2C and 2D).
bAVM hemorrhage was measured using Prussian blue staining of iron deposition. Compared with DMSO controls (3.13±0.93 pixel/mm2 of Prussian blue–positive area), thalidomide group showed a trend of reduction of Prussian blue–positive area (2.28±1.38 pixel/mm2; P=0.08). Lenalidomide-treated group had smaller Prussian blue–positive area (1.98±1.47 pixel/mm2) than its DMSO control group (3.5±1.06 pixel/mm2; P=0.04; Figure 3). These data suggest that thalidomide and lenalidomide treatments improve mural cell coverage and reduce hemorrhage in established bAVMs.
Thalidomide Restored Pdgfb and Pdgfrβ Expression in bAVM Vasculature
To explore mechanisms that might mediate the reduction of mural cell coverage in bAVM vessels, we performed in vitro studies first using human brain microvascular endothelial cells. When ALK1 expression was knocked down to 30% of its normal level by lenti-shALK1 (a lentiviral vector expressing human ALK1 shRNA; Figure IV and Table II in the online-only Data Supplement), PDGFB expression was reduced to 44% in the absence of VEGF (P=0.001) and to 64% in the presence of VEGF (P=0.006; Figure IVA and Table III in the online-only Data Supplement). We then analyzed gene expression in the bAVM vessels. Compared with the microvessels isolated from wild-type mouse brain, the microvessels isolated from the bAVMs showed a ≈50% reduction of Alk1 expression in both DMSO- and thalidomide-treated mice. This was associated with an 18% reduction of Pdgfb expression in DMSO-treated mice. Thalidomide treatment restored Pdgfb expression (Figure 4B). These data suggest that a reduction of PDGFB expression in ALK1-mutated endothelial cells could be responsible for the loss of mural cell coverage in the bAVM vessels, which may be restored with thalidomide treatment.
We next quantified the expression Pdfgrβ and Tek, which are 2 important factors in mural cells recruitment, in both bAVM vessels and the surrounding brain tissue by Western blot analysis (Figure V in the online-only Data Supplement). Compared with normal mice, DMSO-treated bAVM mice express 56% lower Pdfgrβ (P=0.01) and 38% lower Tek (P=0.01). Thalidomide treatment increased Pdfgrβ expression (P=0.035 versus DMSO group; Figure 4C) but did not alter Tek expression in bAVM vessels and their surrounding tissue (Figure 4D). Therefore, it is likely that thalidomide improves mural cell coverage of bAVM vessels through an upregulation of Pdgfb/Pdgfrβ signaling pathway.
Thalidomide Reduced Inflammation in bAVM Lesion
Studies on the genetics and cytokine expressions suggest that inflammation may contribute to AVM progression and rupture.9 To test whether thalidomide reduces bAVM inflammation, the expression of inflammatory cytokines, C-X-C motif chemokine receptor 4 (Cxcr4), Il1b (interleukin 1b), and TNF-α (tumor necrosis factor-α) were analyzed by quantitative real-time polymerase chain reaction. We found that levels of Cxcr4, Il1b, and TNF-α were higher in bAVM microvessels than that in the wild-type microvessels. Thalidomide reduced the levels of these cytokines in bAVM microvessels (Figure VIA in the online-only Data Supplement). Thalidomide has also reduced CD68+ burden in bAVM. Compared with DMSO-treated mice (513±270 cells/mm2), thalidomide-treated mice have fewer CD68+ cells in bAVM (210±188 cells/mm2; P=0.02). The number of CD68+ cells is positively correlated with the number of vSMC-negative vessels (r=0.71; P<0.0001) and Prussian blue–positive area (r=0.46; P=0.012; Figure VII in the online-only Data Supplement).
Overexpression of PDGFB Mimicked Thalidomide Effect
To test whether upregulation of Pdgfb mimics thalidomide’s therapeutic effect, we overexpressed human PDGFB in mouse bAVM lesion through lentiviral vector-mediated gene transfer, because protein sequences of human and mouse PDGFB are at 96% homology (Figure VIII in the online-only Data Supplement). Lenti-PDGFB or Lenti-GFP (control) was injected to the brain at the time of bAVM induction. PDGFB expression was detected at the injection site 8 weeks after viral injection, and GFP expression was detected 3 and 7 days after injection (Figure IX in the online-only Data Supplement). Overexpression of PDGFB did not alter the vessel densities (Figure IX in the online-only Data Supplement) but reduced the number of dysplastic vessels (dysplasia index) in bAVM lesion (Lenti-PDGFB versus Lenti-GFP: 5.25±1.75 versus 8.35±3.03; P=0.04; Figure X in the online-only Data Supplement) and increased pericyte coverage of bAVM vessels (Lenti-PDGFB versus Lenti-GFP: 78±2.9% versus 62±13.1%; P=0.017; Figure 5A and 5B). The Prussian blue–positive area in mice coinjected Lenti-PDGFB with Ad-Cre and AAV-VEGF (1.56±1.23 pixels/mm2) was smaller than that of mice received coinjection of Lenti-GFP with Ad-Cre and AAV-VEGF (3.03±0.51 pixel/mm2; P=0.001) and was similar to that of the 2 control groups that injected with lenti-GFP, Ad-GFP, and AAV-VEGF or Lenti-PDGFB, Ad-GFP, and AAV-VEGF (Ps>0.05; Figure 6A and 6B). Therefore, overexpression of PDGFB mimicked thalidomide’s effect, which suggests that upregulation of Pdgfb expression contributes to thalidomide’s beneficial effect.
Thalidomide and Lenalidomide Did Not Alter Mouse Body Weight
No obvious abnormal behavior was observed in the thalidomide- or lenalidomide-treated mice. The body weights are similar among all groups throughout the treatment period (Figure XII in the online-only Data Supplement). Mortality was reported in (Table IV in the online-only Data Supplement).
In this study, using an Alk1-deficient bAVM mouse model,4 we showed that thalidomide inhibits bAVM formation and improves mural cell coverage and reduces hemorrhage in established bAVMs. Pdgfb and Pdgfrβ expression are reduced in the Alk1-deficient bAVM. Thalidomide restores their expression.
PDGFB/PDGFRβ signaling is indispensable for mural cell recruitment during angiogenesis.18 In the brain, disruption of PDGFB or its receptor PDGFRβ leads to a reduction of vascular mural cells, causing various vascular abnormalities, such as microaneurysm and chronic microhemorrhage.17,19 The role of PDGFB/PDGFRβ signaling in human bAVM pathogenesis is largely unknown. Increased PDGFB expression has been detected in a subset of surgically resected human sporadic bAVMs.14,20 However, studies including ours demonstrated that vessels in bAVMs and their surroundings have fewer mural cell coverage than normal cerebrovasculature.21 We showed in this study that lentivirus-mediated overexpression of PDGFB mimicked the effects of thalidomide, which suggest that upregulation of Pdgfb/Pdgfrβ pathway could be a underlying mechanism of thalidomide effect. However, more studies will be needed to confirm the role of PDGFB in bAVM pathogenesis.
In prior reports, thalidomide seemed to improve mural cell recruitment in the retina of Eng+/− mice.12 Thalidomide stabilizes small capillaries in the telangiectasia and reduces nose bleeds and the requirement of blood transfusions in HHT patients.12 We showed previously that bAVMs in Alk1-deficient model have fewer mural cells, which is associated with a reduction of Pdgfrβ expression.5 We showed in the present study that both thalidomide and lenalidomide reduce hemorrhage and improve mural cell coverage in bAVMs, which is most likely through upregulation of PDGFB/PDGFRβ signaling.
Antiangiogenesis has been shown to be effective in treating bAVM.22,23 Although previous studies suggested that thalidomide has antiangiogenic function, we found neither thalidomide nor lenalidomide altered vessel densities in bAVMs, suggesting that the beneficial effects shown in this study are not through antiangiogenesis.
Interestingly, studies discovered CRBN (cereblon), a substrate receptor of the CUL4–RBX1–DDB1 (cullin4-ring-box 1-damage specific DNA binding protein 1) ubiquitin ligase complex (CRL4), is a direct and primary target of thalidomide and lenalidomide in in vitro cell models and in zebrafish embryos.24,25 The association between CRBN and PDGFB is unclear. Thalidomide has been suggested to indirectly reduce the expression of certain cytokines, such as Il6, Il1b, and TNF-α.26 We showed that thalidomide reduced Cxcr4, Il1b, and TNF-α in mouse bAVM microvessels. In addition, elevation of PDGFB in human bAVM may reflect a potential compensatory mechanism in response to the reduction of mural cells. Further characterization of endothelial-to-mural cell signaling and quantitative studies of mural cell recruitment and turnover are needed to better delineate the role of mural cells in human bAVM pathogenesis.
Because both thalidomide and lenalidomide are currently used in clinical practice, the doses and side effects of these drugs in human have mostly been defined. In this study, our goal is to test their therapeutic efficacy in bAVM. The doses we used were selected based on previous studies. For thalidomide, both 75 and 150 mg/kg (IP) had probably near-maximal therapeutic effects12; however, 150 mg/kg retarded mouse growth. For lenalidomide, based on available rodent studies,27 we have empirically chosen a lower dose compared with thalidomide and administrated the drug daily based on its pharmacokinetics.28
Thalidomide has well-known adverse effects that limit its applicability in treating bAVM patients.10 We showed that lenalidomide, one of the newer derivatives of thalidomide, has similar effects to that of thalidomide. Therefore, the newer and safer derivatives of thalidomide could be better options for treating bAVM patients.
Our study had several limitations: (1) Ad-Cre that expresses Cre in all cell types has been used to induce focal Alk1 deletion during model induction. However, our previous studies showed that this model has many phenotypes that resemble human bAVM.4,5 (2) Different viral vectors have been used in experimental mice, which may lead to more severe inflammation than those occurred in human bAVMs. However, we found that thalidomide treatment reduces inflammation in this model, suggesting that thalidomide could reduce inflammation in human bAVMs.
In summary, our study is the first to show that thalidomide and lenalidomide inhibit bAVM development and improve vascular integrity of existing bAVMs. Our data suggest that thalidomide and its safer derivatives should be further explored as therapeutic options to reduce bAVM hemorrhage in patients.
We thank Dr S. Paul Oh at Barrow Neurological Institute, Phoenix, Arizona, for providing us the Alk1-floxed mice. Dr Zhu, Dr Chen, Dr Zou, Dr Wang, Mr Bao, Mr Saw, Dr Wang, Ms Zhan, Dr Lawton, Dr Winkler, Dr Li, Dr Zhang, Dr Shen, and Dr Shaligram performed the experiments and collected samples. Dr Zhu, Dr Chen, Dr Wang, Mr Bao, Dr Wang, Dr Lawton, Dr Winkler, and Dr Li, performed data analyses. Drs Zhu and Su designed the study and wrote the article.
Sources of Funding
This study was supported by grants to Dr Su from the National Institutes of Health (R01 NS027713, R01 HL122774, and R21 NS083788) and from the Michael Ryan Zodda Foundation and by a grant from the University of California San Francisco Research Evaluation and Allocation Committee.
The online-only Data Supplement is available with this article at http://stroke.ahajournals.org/lookup/suppl/doi:10.1161/STROKEAHA.117.020356/-/DC1.
- Received December 5, 2017.
- Revision received February 13, 2018.
- Accepted February 16, 2018.
- © 2018 American Heart Association, Inc.
- Mohr JP,
- Parides MK,
- Stapf C,
- Moquete E,
- Moy CS,
- Overbey JR,
- et al
- Sanders RD,
- Avidan MS
- Chen W,
- Guo Y,
- Walker EJ,
- Shen F,
- Jun K,
- Oh SP,
- et al
- Chen W,
- Sun Z,
- Han Z,
- Jun K,
- Camus M,
- Wankhede M,
- et al
- Guo Y,
- Saunders T,
- Su H,
- Kim H,
- Akkoc D,
- Saloner DA,
- et al
- Park SO,
- Lee YJ,
- Seki T,
- Hong KH,
- Fliess N,
- Jiang Z,
- et al
- Su H,
- Lu R,
- Kan YW
- Hao Q,
- Liu J,
- Pappu R,
- Su H,
- Rola R,
- Gabriel RA,
- et al
- Bjarnegård M,
- Enge M,
- Norlin J,
- Gustafsdottir S,
- Fredriksson S,
- Abramsson A,
- et al
- Walker EJ,
- Su H,
- Shen F,
- Degos V,
- Amend G,
- Jun K,
- et al
- Zhu W,
- Shen F,
- Mao L,
- Zhan L,
- Kang S,
- Sun Z,
- et al
- Mercurio A,
- Adriani G,
- Catalano A,
- Carocci A,
- Rao L,
- Lentini G,
- et al
- Pather K,
- Helsby NA,
- Palmer BD,
- Ching L-M