Cerebral cavernous malformations form an anticoagulant vascular domain in humans and mice

Cerebral cavernous malformations (CCMs) are common central nervous system (CNS) vascular anomalies that occur sporadically or heritably and affect ∼1 in 200 humans.¹  Mutations of 3 genes, KRIT1 (CCM1), CCM2, or PDCD10 (CCM3), are associated with the development of CCMs, cerebral venous capillary dysplasias with clusters of endothelium filled with blood and prone to hemorrhage.^2,3  The KRIT1^+/− genotype is the most common cause of the familial form of CCM, whereas the PDCD10^+/− genotype often results in a more severe clinical manifestation of the disease.^4,5  Unlike sporadic CCM, in which a single lesion is typically observed, familial CCMs are characterized by multiple vascular lesions that are believed to be due to random inactivation of the normal allele.^1,4,6  Moreover, a combination of chronic bleeding and episodes of acute hemorrhagic stroke, associated with CCM, causes focal neurological deficits, headaches, epileptic seizures, and, occasionally, death.⁷  In addition, patients with PDCD10 mutations are more likely to present with significant CCM hemorrhages earlier in life.⁴  Although the annual symptomatic hemorrhage rate varies largely among studies from 0.25 to 22.9% per patient-year,^2,8  it is thought that all CCMs harbor occult bleeding.^7,8  Natural history studies and magnetic resonance imaging (MRI) analysis have identified prior CCM bleeding within a year as a predictor of repeated hemorrhage and subsequent clinical sequelae.^7,9,10  However, the detailed molecular mechanisms underlying the pathogenesis of CNS hemorrhage in CCM remain elusive.

Recently, we performed genome-wide transcriptome analysis of the acute effects of inactivation of Krit1 in murine brain microvascular endothelial cells (BMECs) and found increased levels of Thbd messenger RNA (mRNA), which encodes the natural anticoagulant receptor thrombomodulin (TM).¹¹  Although TM levels are notably low in normal young brain endothelium, TM plays a role in the thromboresistant properties of the brain.^12,13  TM binds thrombin and, while bound, thrombin fails to convert fibrinogen into insoluble fibrin; instead, it catalyzes the formation of activated protein C (APC). APC generation is enhanced by the presence of the endothelial cell protein receptor (EPCR),^13,14  the mRNA of which is also increased with loss of Krit1 in BMECs.¹¹  Because APC is a potent natural anticoagulant, and high levels of TM have been associated with a bleeding disorder^15,16  and used as a biomarker of endothelial cell dysfunction,¹⁷  we hypothesized that increased TM and EPCR could create a local bleeding diathesis in CCM.

Here, we show that TM is upregulated in CCM lesions, as well as in the plasma of individuals with the sporadic or familial form of the disease. In 2 acute models of CCM, we report marked increases in brain endothelial TM and EPCR following genetic inactivation of Krit1 or Pdcd10. Increased expression of TM, but not EPCR, was ascribable to elevation of the transcription factors KLF2 and KLF4 in CCM. Increased TM expression contributes to CCM hemorrhage, because genetic inactivation of 1 or 2 copies of the Thbd gene decreases brain bleeding in Pdcd10^ECKO mice. Moreover, administration of blocking antibodies against TM and EPCR reduced CNS bleeding in Pdcd10^ECKO mice. Thus, we propose that upregulation of TM and EPCR in CCM endothelium forms an anticoagulant vascular domain that contributes to the bleeding-induced morbidity in CCM.

From May of 2015 to November of 2017, 77 patients (mean age, 34.26 ± 18.53 years; range, 5.18-76.02) with a confirmed diagnosis of CCM (38 sporadic, 21 CCM1, and 18 CCM3) and 10 healthy subjects (mean age, 33.26 ± 7.49 years; range, 23.0-43.75) were enrolled in this study. The recruitment of patients was performed in conjunction with their routine clinical evaluations or follow-up. All participants provided informed consent to participate in this research in accordance with the Declaration of Helsinki and according to guidelines approved by The University of Chicago Institutional Review Board. The ethical principles guiding the Institutional Review Board are consistent with The Belmont Report and comply with the rules and regulations of The Federal Policy for the Protection of Human Subjects (56 FR 28003).

As per currently accepted disease categorization, CCM cases were classified as sporadic if they harbored a solitary lesion on the most sensitive susceptibility weighted imaging MRI sequences or a cluster of lesions associated with a developmental venous anomaly.^3,10  They were classified as familial if they harbored multifocal CCM lesions, a family history of CCM in a first-degree blood relative, or a mutation genotyped at a CCM gene locus.^3,10  Patients with CCM lesion resection or any prior brain irradiation were excluded.

The endothelial-specific conditional Krit1- or Pdcd10-null mice were generated by breeding transgenic mice expressing endothelial-specific Pdgfb promoter–driven tamoxifen-regulated Cre recombinase, iCreERT2,¹⁸  with loxP-flanked Krit1 exon 5 (Krit1^fl/fl) mice (Pdgfb-iCreERT2; Krit1^fl/fl)¹¹  or with loxP-flanked Pdcd10 exon 4 and 5 mice (Pdcd10^fl/f; a generous gift from Wang Min, Yale University) (Pdgfb-iCreERT2; Pdcd10^fl/fl). All experiments were performed using age-matched Krit1^fl/fl or Pdcd10^fl/fl littermates on the same C57BL/6 background. Thbd^fl/fl mice (a generous gift from Hartmut Weiler, Medical College of Wisconsin) were crossed with Pdgfb-iCreERT2; Pdcd10^fl/fl mice to generate Pdgfb-iCreERT2; Pdcd10^fl/fl; Thbd^fl/wt and Pdgfb-iCreERT2; Pdcd10^fl/fl; Thbd^fl/fl mice. Mice were administered 50 μg of 4-hydroxy-tamoxifen (Sigma-Aldrich; H7904) by intragastric injection on postnatal day (P)1 to induce Cre activity and endothelial Krit1 or Pdcd10 gene inactivation in the littermates bearing iCreERT2 (Krit1^ECKO or Pdcd10^ECKO mice). These mice and control Krit1^fl/fl or Pdcd10^fl/fl mice were euthanized on the indicated days. All animal experiments were carried out in compliance with animal procedure protocols approved by the University of California, San Diego (UCSD) Institutional Animal Care and Use Committee.

Adult mice (2-4 months) were euthanized, and their brains were removed, placed into ice-cold buffer A (10 mM N-2-hydroxyethylpiperazine-N′-2-ethanesulfonic acid, 1× penicillin-streptomycin, 0.5% bovine serum albumin in Dulbecco’s modified Eagle medium), and processed as previously described.¹¹ 

BMECs at passages 1 through 3 were maintained at 37°C in 95% air and 5% CO₂, grown to 85% confluence, and treated for 48 hours with 5 μM 4-hydroxy-tamoxifen (Sigma-Aldrich; product number H7904). Thereafter, the medium was replaced with medium lacking 4-hydroxy-tamoxifen,¹¹  and cells were harvested at the time indicated.

Pdcd10^ECKO mice and littermate control Pdcd10^fl/fl mice were treated with inhibitory antibodies (4.6 mg/kg) against mouse TM (MTM-1701)¹⁹  and (2.6 mg/kg) mouse EPCR (rcr-16)²⁰  at P7 and euthanized at P10. Pdcd10^ECKO;Thbd^ECKO/wt, Pdcd10^ECKO;Thbd^ECKO, Pdcd10^ECKO, and littermate control Pdcd10^fl/fl mice were euthanized at P10. Brains were harvested and transferred into 500 μL of 1× PBS (sample A) and incubated for 3 minutes with gentle agitation. Retinas were transferred into 1 mL of 4% PFA for 20 minutes and incubated at room temperature. After 3 washes with 1× PBS, retinas were dissected to remove cornea and lens and to expose extravascular blood from the retina using 100 μL of 1× PBS (sample B). Then extravascular bleedings from brains (sample A) and retinas (sample B) were processed. For brains, we used 200 μL of sample A, whereas we used 50 μL of sample B plus 150 μL of PBS for retinas. Acetic acid was added to a final concentration of 1% (v/v) to lyse the extravasated red blood cells, and hemoglobin was subsequently measured using a microplate reader (VersaMax; Molecular Devices) set to 405 nm.

Data are expressed as mean ± standard error of the mean (SEM). For all experiments, the number of independent experiments (n) is indicated. The sample sizes were estimated with a 2-sample t test (2 tailed). A 2-tailed unpaired Student t test was used to determine statistical significance. For multiple comparisons, 1-way analysis of variance (ANOVA), followed by the Tukey post hoc test, was used.

Supplemental Methods and a table of oligonucleotides used (supplemental Table 1) are available on the Blood Web site.

We recently performed genome-wide transcriptome analysis of the effects of acute inactivation of Krit1 on murine BMECs and observed increased Thbd mRNA that encodes a major anticoagulant endothelial receptor, TM.¹¹  TM expression is modest in normal brain microvasculature relative to systemic blood vessels, suggesting that the brain is a hemostatically “privileged” environment to avert hemorrhage.^13,21,22  To examine whether TM expression is increased in human CCM, we stained lesions obtained at surgery. TM staining was increased in the endothelium of human CCM lesions in comparison with lesion-free brain tissue from a CCM patient or nonneurological disease control (Figure 1A). Furthermore, increased expression of TM in CCM endothelial cells was confirmed by an ∼2.9-fold increase in THBD mRNA in CCM lesion endothelial cells isolated by laser capture microdissection (Figure 1B).

We next examined venous plasma levels of soluble TM in familial and sporadic forms of CCM. Significantly increased soluble TM was detected in patients with familial CCM due to mutations in KRIT1 (CCM1) or PDCD10 (CCM3) (3.57 ± 0.64 ng/mL [mean ± SEM]) compared with healthy subjects (2.78 ± 0.11 ng/mL [mean ± SEM]). There was also significantly increased soluble TM in patients with sporadic CCM (3.06 ± 0.03 ng/mL [mean ± SEM]) (Figure 1C). Thus, CCM lesions arising from loss of function of KRIT1 (CCM1) or PDCD10 (CCM3) led to an increase in TM protein and mRNA levels in CCM lesions and an increase in TM in the plasma.

To confirm that reduced expression of KRIT1 causes increased TM expression in human brain endothelium, we analyzed the effect of silencing of KRIT1 on TM expression in a human brain endothelial cell line. Infection of hCMEC/D3 cells with a virus encoding short hairpin RNA for KRIT1 (shKRIT1)¹¹  (Figure 2A) significantly increased the expression of TM mRNA and protein (Figure 2B-C). Immunofluorescence also revealed elevated levels of membrane-bound TM protein in KRIT1-depleted human brain endothelial cells (Figure 2D). Flow cytometry analysis showed that surface TM protein increased ∼1.7-fold in KRIT1-depleted cells (Figure 2E-F). An increase in human brain endothelial TM following loss of KRIT1 was accompanied by an increased capacity of brain endothelial cells to generate APC (Figure 2G). The increased APC generation was specific, because blocking antibodies against TM or EPCR reduced APC generation (Figure 2G).

We used BMECs isolated from mice bearing floxed alleles of Krit1 (Krit1^fl/fl) or Pdcd10 (Pdcd10^fl/fl) and an endothelial-specific tamoxifen-regulated Cre recombinase (Pdgfb-iCreERT2¹⁸ ) to investigate the effect of acute genetic inactivation of CCM genes¹¹  on brain endothelial TM expression. Consistent with results observed in human brain endothelial cells, TM protein levels were significantly increased in 4-hydroxy-tamoxifen–treated Pdgfb-iCreERT2; Krit1^fl/fl(Krit1^ECKO) BMECs compared with control Krit1^fl/fl BMECs (Figure 3A). Increased APC generation was also observed following genetic inactivation of Krit1 in BMECs (data not shown). In addition, acute genetic inactivation of Pdcd10 in 4-hydroxy-tamoxifen–treated Pdgfb-iCreERT2;Pdcd10^fl/fl(Pdcd10^ECKO) BMECs caused an ∼2.3-fold increase in TM protein levels, as assessed by western blotting (Figure 3B). Consistent with increased protein abundance, RT-qPCR analysis confirmed a dramatic increase in Thbd mRNA levels in Krit1^ECKO BMECs (∼8-fold) and Pdcd10^ECKO (∼8.8-fold) BMECs compared with control Krit1^fl/fl or Pdcd10^fl/fl BMECs, respectively (Figure 3C).

Transcription factors KLF2 and KLF4 are regulators of endothelial anticoagulant gene expression.^23-26  We confirmed that the changes in brain endothelial TM, at the protein and mRNA levels, were associated with a large increase in transcription factors KLF2 and KLF4 in Krit1^ECKO and Pdcd10^ECKO BMECs (Figure 3D-E). Therefore, to test whether an increase in KLF2 and KLF4 induces TM during CCM, we used lentivirus-mediated transduction to ectopically express KLF2 and KLF4 in human endothelial cells (Figure 4A) at levels that we previously showed to suppress a target gene, Thbs1.¹¹  Overexpression of KLF2 and KLF4 led to a significant increase in TM mRNA levels compared with control GFP-infected cells (Figure 4B). We also observed that ectopic expression of KLF2 or KLF4 resulted in ∼4.2-fold or ∼5.8-fold increases in TM protein, respectively, as determined by western blot analysis (Figure 4C). Flow cytometry analysis showed that membrane-bound endothelial TM was increased ∼4-fold in KLF4-overexpressing cells compared with control infected cells (Figure 4D-E). Moreover, we observed that reducing the expression of KLF2 and KLF4 (Figure 4F-G) prevented increased TM expression in KRIT1-depleted human brain endothelial cells (Figure 4H). Together, these observations suggest that the increase in KLF2 and KLF4 led to increased brain endothelial TM in CCMs.

We used Pdcd10^ECKO mice to investigate the distribution and expression of TM protein in experimental murine CCM. At P9, Pdcd10^ECKO hindbrains exhibited visible hemorrhagic CCM lesions (Figure 5A); bleeding was confirmed in hematoxylin and eosin–stained sections (Figure 5B) and immunofluorescence staining of sections for erythrocytes using anti-TER119 (Figure 5C). Double staining of P9 Pdcd10^ECKO cerebellums for TM and isolectin B4 showed increased TM staining of the luminal surface of CCMs (Figure 5D). In contrast, P9 Pdcd10^fl/fl cerebellum showed minimal expression of TM (Figure 5D). To confirm this increase in TM expression, we isolated Pdcd10^ECKO hindbrains and quantified TM protein and mRNA levels. Consistent with results observed in vitro and in situ, Pdcd10^ECKO hindbrains showed ∼10-fold and ∼3.3-fold increases in TM protein and mRNA levels, respectively, as assessed by western blot or RT-qPCR analysis, respectively (Figure 5E-F). There was no increase in PECAM1 abundance, as assessed by western blot, confirming that the increased TM was not due to increased density of blood vessels (data not shown). Moreover, we observed an increase in TM levels in heart tissue, but not in lung or intestine tissue, from Pdcd10^ECKO mice (Figure 5F). We also noted that expression of KLF2 and KLF4 was markedly increased in Pdcd10^ECKO hindbrain tissue (Figure 5G), corroborating previous findings.²⁷ Pdcd10^ECKO mice developed retinal vascular lesions in the leading front of the vascular plexus with extensive hemorrhages (Figure 5H-I) associated with a dramatic increase in TM staining; the increased bleeding was confirmed by increased hemoglobin (Figure 5I-J). A loss-of-function mutation in KRIT1 is the most common cause of the familial form of CCM. P11 Krit1^ECKO hindbrains exhibited an increase in TM staining of the luminal isolectin B4⁺ CCM endothelial cells (supplemental Figure 1); however, TM staining was clearly less intense than that observed in Pdcd10^ECKO CCMs. In contrast with Pdcd10^ECKO mice, Krit1^ECKO mice exhibited much fewer hemorrhagic CCM lesions (supplemental Figure 1). Collectively, these findings suggested that CCMs in Pdcd10^ECKO and Krit1^ECKO mice, like humans, express increased levels of TM in lesions that contribute to brain hemorrhage.

EPCR, an endothelial transmembrane protein that presents protein C to the thrombin–TM complex¹⁴  is expressed in the brain vasculature and plays a crucial role in the vascular anticoagulant pathway^28,29  by facilitating protein C activation.³⁰  We next investigated whether EPCR levels were also altered in CCMs. PROCR (EPCR) mRNA and protein expression was increased in human CCMs in comparison with control brain tissue (Figure 6A-B). RT-qPCR and western blot analysis confirmed the upregulation of EPCR mRNA and protein in Krit1^ECKO and Pdcd10^ECKO BMECs (Figure 6C-E). Consistent with results observed in vitro, Pdcd10^ECKO hindbrains showed an ∼2-fold increase in Procr (Epcr) mRNA abundance (Figure 6F). Increased Procr mRNA was specific to lesion areas, because we did not observe an increase in other vascular beds of Pdcd10^ECKO mice analyzed (Figure 6F). In contrast to TM, ectopic expression of KLF2 or KLF4 did not change EPCR protein or mRNA levels in endothelial cells, as assessed by western blot or RT-qPCR analysis, respectively (Figure 6G-H).

To assess the role of endogenous endothelial TM in CCM hemorrhages, we examined the impact of genetic inactivation of endothelial Thbd in Pdcd10^ECKO mice. Pdcd10^ECKO;Thbd^ECKO/wt mice exhibited a significant reduction in brain bleeding (∼1.6-fold decrease), as did Pdcd10^ECKO;Thbd^ECKO mice (∼1.8-fold decrease), compared with Pdcd10^ECKO littermates (Figure 7A). Moreover, the upregulation of TM and EPCR contributed to brain hemorrhage in Pdcd10^ECKO mice; blocking antibodies against mouse TM and mouse EPCR reduced bleeding, as signaled by an ∼2-fold reduction in extractable hemoglobin in Pdcd10^ECKO mice (Figure 7B). These data suggest that CCMs caused by mutations of KRIT1 or PDCD10 lead to the formation of an anticoagulant vascular domain due to increased expression of TM and EPCR that is driven, in part, by transcription factors KLF2 and KLF4, thereby contributing to increased hemorrhage (Figure 7C).

CNS bleeding is a primary source of morbidity and mortality in CCM patients. We report that CCM endothelial cells manifest elevated expression of TM and EPCR mRNA and protein. This cell-autonomous increased expression of TM and EPCR is a direct result of a decrease in the products of KRIT1 or PCDC10, 2 genes whose loss of function leads to CCM formation. An increase in the expression of KLF2 and KLF4, 2 transcription factors already implicated in CCM pathogenesis, can account for increased TM expression; however, these transcription factors did not account for the upregulation of EPCR expression. The increased abundance of TM and EPCR on the endothelial cell surface results in an increased capacity to support the generation of APC, a potent endogenous anticoagulant that inactivates the coagulation cofactors factor Va and factor VIIa.²²  Increased TM expression levels contributed to CCM hemorrhage, because genetic inactivation of 1 or 2 copies of the Thbd gene in endothelial cells decreases brain bleeding in Pdcd10^ECKO mice. Moreover, administration of blocking antibodies against TM and EPCR significantly reduced CCM hemorrhage in Pdcd10^ECKO mice. These studies identify that an increase in expression of 2 endothelial cofactors generates anticoagulant APC in CCM and indicate that the formation of an anticoagulant endothelial domain in CCM lesions can contribute to bleeding. Thus, manipulation of the balance between anticoagulant and cytoprotective effects of APC could serve to reduce the incidence and morbidity of CCM bleeding.

The elevation in endothelial surface TM and EPCR can contribute to bleeding in CCM. Cerebral capillary dysplasias with hemosiderin deposition detected by MRI^10,31  or histology^3,10,32  are hallmarks of CCM lesions. Electron microscopic studies have documented loss of tight junctions in some CCMs,³³  and KRIT1 enables Rap1 stabilization of endothelial cell–cell junctions.^34,35  Indeed, we have recently found that loss of tight junctions is 1 of the earliest manifestation of inactivation of Krit1 or Pcdc10 in BMECs.¹¹  Disruption of brain interendothelial junctions in CCMs^11,36,37  could trigger the recurring microhemorrhages that cause hemosiderin deposition. Importantly, bleeding is much less evident in other cerebrovascular diseases that exhibit loss of blood–brain barrier integrity due to disruption in brain endothelial tight and adherens junctions.³⁸  Normal brain vasculature maintains restricted levels of TM and EPCR compared with systemic blood vessels,^21,39  suggesting that the brain is a prohemostatic environment. Our finding that loss of CCM1 or CCM3 causes a marked increase in EPCR and TM on the surface of CCM endothelial cells suggests that CCMs form an anticoagulant vascular domain within the brain that causes a local bleeding diathesis.

Familial and sporadic forms of CCMs exhibited significant increases in soluble TM in plasma, suggesting that this may be a biomarker for patients who are at elevated risk for bleeding. Elevated levels of TM in plasma have been documented and are associated with endothelial dysfunction^17,40  and bleeding disorders.^15-17  As discussed above, the increased TM can reflect an increased formation of APC in lesions which, in turn, leads to anticoagulant activity¹³  that facilitates bleeding. As the number of potentially new therapeutic interventions in CCMs grows,^11,41-44  the identification of markers to identify patients with an elevated risk for hemorrhage becomes imperative. Indeed, recent studies^45,46  showed that levels of inflammatory and angiogenic molecules in peripheral blood plasma could predict CCM clinical behavior, and this study now identifies soluble TM as a new putative biomarker candidate in this disease.

Sangwung et al⁴⁷  showed that endothelial cell loss of KLF2 and KLF4 results in substantial loss of TM, and we report here that overexpression of these transcription factors can drive TM expression, as would be predicted from previous studies.^23-25  There is now compelling evidence that elevation of KLF2 and KLF4 makes important contributions to the pathogenesis of CCM.^11,27,41,48  Endothelial-specific deletion of KLF2 and KLF4 in adults leads to lethality,⁴⁷  indicating that inhibition of these transcription factors is unlikely to be a suitable therapeutic strategy. Here, we add upregulation of TM to the growing list¹¹  of KLF2 and KLF4 targets that can contribute to the pathogenesis of CCM disease and might serve as therapeutic targets. In contrast, we could not ascribe the increase in EPCR expression to KLF2 and KLF4. In particular, depleting KRIT1 increased EPCR and TM expression in cultured endothelial cells, whereas overexpression of KLF2 and KLF4 only increased TM. Thus, the cell autonomous increase in EPCR does not appear to be a consequence of increased expression of KLF2 and KLF4 or of TM. The links between loss of CCM genes and increased EPCR expression could be a fruitful area for future work.

This study suggests new potential avenues for the management of CNS hemorrhage in CCM. Insights gleaned from bleeding in hemophilia increased our awareness that excessive or unbalanced APC generation can actively contribute to bleeding and that the anticoagulant activity of APC is a valid target for drug development.^49,50  However, APC is a multifunctional protease and, independently of its anticoagulant activity, APC induces cytoprotective effects on endothelial cells and neurons that are dependent on protease-activated receptor 1 and 3, as well as on EPCR. These cell-signaling activities of APC are central to the primary mechanism of action for APC’s neuroprotective effects in a variety of acute and chronic neuropathologies, including ischemic stroke, traumatic brain injury, chronic cerebral ischemia, amyotrophic lateral sclerosis, and multiple sclerosis.⁵¹  Therefore, strategies targeting the protein C system and aimed at mitigating or preventing bleeding in CCMs need to be specific for APC anticoagulant activity while leaving APC’s neuroprotective activities unaltered. This was also illustrated by our experiments in which inhibition of all APC activities (anticoagulant and cytoprotective), using the anti-mouse APC antibody SPC-54,⁵⁰  caused premature death and cerebral thrombosis in Pdcd10^ECKO mice (data not shown). Several anticoagulant-specific strategies targeting the protein C pathway have been proposed for control of bleeding in hemophilia patients with inhibitors, including anti-APC and anti-protein S antibodies specifically targeting APC’s anticoagulant activity and APC-resistant factor Va.^49,50  Whether these strategies can be safely adapted to CCM will have to be determined, because the safety window in hemophilia and CCM is likely quite different; targeting of acute CNS bleeds vs the risk of rebleeding in the 1-year window of increased vulnerability that follows such bleeds will be an important consideration for this.^7,8 

Another consideration is that the high level of APC in CCM lesions could diminish morbidity by means of its cytoprotective effects, which include protection of vascular and blood–brain barrier integrity.^14,28  Indeed, signaling-selective 3K3A-APC with minimal anticoagulant activity reduced tissue-type plasminogen activator–induced bleeding in a murine ischemic stroke model,⁵²  indicating that the consequences of enhanced APC generation in CCM might be more complex. New studies are needed to appreciate these complexities and to evaluate which strategies targeting the different components of the protein C system will be most advantageous in CCM. This study motivates the development of such novel strategies for the management of CCM hemorrhage and suggests new biomarkers that could serve to identify those patients with an elevated risk for bleeding as potential subjects for such new therapeutic strategies.

The authors thank Angioma Alliance for providing human CCM specimens; Selina Chang, Riya Verma, Tiffany S. Shamas, and Shady Soliman for technical assistance; Jennifer Santini for microscopy technical assistance; and the UCSD School of Medicine Microscopy Core. The anti-EPCR antibodies (rcr-252, rcr-2, and rcr-16) were generously made available by Kenji Fukudome (Saga Medical School, Saga, Japan). Thbd^fl/fl mice were the generous gift of Hartmut Weiler (Blood Research Institute, Milwaukee, WI).

This work was supported by National Institutes of Health, National Institute of Neurological Disorders and Stroke grant NS092521 (M.H.G. and I.A.A.) and National Institutes of Health, National Heart, Lung, and Blood institute grants K01 HL133530 (M.A.L.-R.), HL104165, HL130678, and HL142975 (L.O.M.), and HL078784 (M.H.G.), as well as by the UCSD School of Medicine Microscopy Core (P30 NS047101).

Contribution: M.A.L.-R. designed and performed the experiments, analyzed and interpreted the data, generated the figures, and wrote the manuscript; A.P. and P.H. performed and analyzed histological and gene expression experiments in culture and conditional knockout mice; R.G., J.K., L.S., S.P., and I.A.A. performed, analyzed, and interpreted the RNA transcriptome of human CCM lesions and human TM plasma levels; T.W., A.Y., and L.O.M. performed, analyzed, and interpreted protein C activation and bleeding studies; R.G., A.P., P.H., T.W., A.Y., I.A.A., and L.O.M. helped to write the manuscript; F.L., I.A.R., and L.O.M. provided reagents and helped to interpret the data; C.T.E. provided critical reagents; and M.H.G. conceived the project, designed the overall study, analyzed and interpreted data, and wrote the manuscript.

Conflict-of-interest disclosure: UCSD and The Scripps Research Institute hold intellectual property rights related to APC-resistant factor Va and some uses of APC mutants for which L.O.M. is listed as inventor. L.O.M. is a founder and a member of the Board of Directors of Hematherix LLC. The remaining authors declare no competing financial interests.

Correspondence: Mark H. Ginsberg, Department of Medicine, Leichtag Biomedical Research Building, University of California San Diego, La Jolla, CA 92093; e-mail: mhginsberg@ucsd.edu.