Endothelial Cells Lining Sporadic Cerebral Cavernous Malformation Cavernomas Undergo Endothelial-to-Mesenchymal Transition
Background and Purpose—Cerebral cavernous malformation (CCM) is characterized by multiple lumen vascular malformations in the central nervous system that can cause neurological symptoms and brain hemorrhages. About 20% of CCM patients have an inherited form of the disease with ubiquitous loss-of-function mutation in any one of 3 genes CCM1, CCM2, and CCM3. The rest of patients develop sporadic vascular lesions histologically similar to those of the inherited form and likely mediated by a biallelic acquired mutation of CCM genes in the brain vasculature. However, the molecular phenotypic features of endothelial cells in CCM lesions in sporadic patients are still poorly described. This information is crucial for a targeted therapy.
Methods—We used immunofluorescence microscopy and immunohistochemistry to analyze the expression of endothelial-to-mesenchymal transition markers in the cavernoma of sporadic CCM patients in parallel with human familial cavernoma as a reference control.
Results—We report here that endothelial cells, a cell type critically involved in CCM development, undergo endothelial-to-mesenchymal transition in the lesions of sporadic patients. This switch in endothelial phenotype has been described only in genetic CCM patients and in murine models of the disease. In addition, TGF-β/p-Smad- and β-catenin-dependent signaling pathways seem activated in sporadic cavernomas as in familial ones.
Conclusions—Our findings support the use of common therapeutic strategies for both sporadic and genetic CCM malformations.
Cerebral cavernous malformation (CCM) is a vascular disease predominantly affecting the microvasculature of brain and spine. About 20% of CCM patients present inherited germ-line heterozygous mutation in one of 3 distinct genes, CCM1, CCM2, and CCM3.1 The remaining patients develop sporadic CCM and do not bear systemic mutation of CCM genes.1 However, cavernoma-localized mutation of CCM genes has been recently reported2 in sporadic CCM, suggesting a common genetic basis for both types of disease. Patients with sporadic CCM usually show single lesions and are diagnosed by magnetic resonance imaging after clinical manifestations. In contrast, familial CCM is characterized by multiple lesions appearing over time.1
In murine models of the pathology,3–5 deletion of CCM genes in endothelial cells plays a crucial role in the development of the malformations.3–5 In addition, in familial patients, biallelic mutations, in one of the CCM genes, have been observed locally in the endothelial cells lining the vascular malformation.6 Rho kinase pathways is aberrantly activated after ablation of CCM genes in endothelial cells in culture, and it is enhanced in endothelial cells lining both familial and sporadic CCM lesions.2,7 However, the molecular phenotype of endothelial cells in CCM lesions in sporadic patients is still poorly described. This information is crucial for a targeted therapy.
We have recently reported that endothelial cells lining vascular malformations in familial patients, as well as in genetic murine models of the disease, undergo endothelial-to-mesenchymal transition (EndMT),8 with loss of VE-cadherin localization to cell-to-cell contacts, and expression of mesenchymal/stem cell markers, such as KLF4, αSMA, S100a4, ID1, Ly6a, and N-cadherin.5 In addition, we have observed that TGF-β/p-Smad- and β-catenin-dependent signaling pathways, 2 important regulators of EndMT,9,10 play a causal role in the EndMT that accompanies CCM malformations.5,11,12
Importantly, in murine models, treatments that prevent5 or induce the regression11 of EndMT reduce the number and size of lesions, suggesting that EndMT is an important process in the etiology and development of this pathology.
Here we report that endothelial cells lining sporadic cavernomas present EndMT features and elevated nuclear p-Smad3 and β-catenin supporting the concept that the triggering process in sporadic malformations is likely the same as in familial malformations and may respond to the same therapeutic intervention.11
Surgical samples from CCM patients were obtained at Fondazione IRCCS Ca’ Granda Ospedale Maggiore Policlinico of Milan, Università Cattolica Sacro Cuore, Roma, and at Angioma Alliance Tissue Bank. CCM human samples were excised from adults for surgical indications unrelated to this research and according to Italian or US legislation. All samples were analyzed at IFOM as approved by the IFOM Ethical Committee.
A table summarizing the characteristics of these samples is reported in Figure 1.13,14 Postmortem non-CCM brain tissues from normal individuals were obtained at Fondazione IRCCS Ca’ Granda Ospedale Maggiore Policlinico of Milan, Università Cattolica Sacro Cuore, Roma, and at BioChain (Newark, CA).
In addition, normal-size vessels (pseudonormal) present in samples from cavernoma patients were also used as internal controls as indicated in the figures.
Paraffin-embedded and frozen samples were processed as described in Maddaluno et al.5 See online-only Data Supplement for further details, image quantification, and list of antibodies used (Table I in the online-only Data Supplement).
Surgical samples from 12 sporadic and 5 familial patients were examined. Controls were both pseudonormal peri-lesion vessels within the same samples and samples from 3 normal individuals who died for reasons unrelated to brain pathologies (Figure 1A). Sporadic patients had single lesion and no family history of CCM.2 In addition, they were diagnosed CCM by magnetic resonance imaging as described in Materials and Methods in the online-only Data Supplement. Familial patients had family history of CCM and underwent genetic diagnosis.
As shown in Figure 1B, both sporadic and familial patients show VE-cadherin delocalized from cell-to-cell junctions, and even in apical and basal locations, in endothelial cells lining the cavernoma (compare VE-cadherin distributed to endothelial junctions in normal brain and pseudonormal peri-lesion vessels; Figure 1). We have reported that delocalization of VE-cadherin from endothelial cell-to-cell junctions is an early response to ablation of CCM genes both in cultured endothelial cells and in animal models of the disease. Treatments that reduce the number and size of lesions also induce reorganization of VE-cadherin to endothelial cell-to cell junctions.11
In animal models of CCM disease and in the familial form, dismantling of junctions is accompanied by expression of mesenchymal/stem cell markers.5,11 We tested, therefore, whether the expression of these markers was also present in the sporadic malformations. As reported in Figure 2, KLF4, αSMA, S100a4 were upregulated to a similar level in endothelial cells of sporadic and familial CCM, but they were absent in endothelial cells of the normal brain vasculature. The expression of such markers in the sporadic and familial patients as well as normal brain samples examined is reported in Figures I–IX in the online-only Data Supplement. Other mesenchymal markers, such as fibronectin, collagen1, N-cadherin, and ID1 were also expressed in endothelial cells in both sporadic and familial CCM lesions (Figure X in the online-only Data Supplement).
Importantly, we found that both in sporadic and familial lesions, EndMT markers KLF4 and p-Smad3 are highly expressed in the endothelium of the cavernoma, whereas they are low in the endothelium of peri-lesion vessels of apparent regular size (pseudonormal). KLF4, for instance, is expressed in the endothelium lining the cavernoma and large abnormal vessels, whereas it is extremely low in pseudonormal vessels of the same sample (Figure 3A). Quantification of KLF4-positive endothelial nuclei shows that endothelial cells in the cavernoma express the higher percentage of KLF4-positive nuclei (Figure 3C).
p-Smads contribute to signal EndMT after CCM ablation in endothelial cells.5,11 Consistently, we observed frequent (≈90%) p-Smad3-positive endothelial nuclei in the cavernomas of sporadic CCM patients (Figure 3B and Figures XI–XIII in the online-only Data Supplement, for all patients). As for KLF4, p-Smad3 was high in endothelial nuclei of both cavernomas and large abnormal vessels, whereas it was low or absent in endothelial nuclei of pseudonormal vessels of the same sample (quantified in Figure 3C in all samples).
β-Catenin-driven transcription is activated after ablation of CCM genes,11,15 and we have observed that it can control the early phases of EndMT switch11 in endothelial cells in brain malformations of endothelial-selective CCM3-ablated mice (bearing a reporter construct for β-catenin transcription activity).11 Here, using an antibody to total β-catenin, we could observe positive signal in the nuclei of the endothelial cells lining the cavernoma (≈20% positive nuclei) of sporadic, as well as familial patients, whereas endothelial nuclei in control normal vessels were negative (Figure XIV in the online-only Data Supplement).
Therefore, both p-Smad and β-catenin signaling seem to be active in endothelial cells of sporadic cavernomas, as it was shown for genetic cavernomas.11
In our previous work, we have show that in familial patients as well as in experimental murine models of CCM with endothelial-specific CCM-deletion, endothelial cells lining the cavernoma acquire mesenchymal/stem cell features and show activated TGF-β/p-Smad- and β-catenin-dependent signaling.5,11 In addition, drugs able to inhibit this EndMT switch by targeting these 2 signaling pathways could reduce the number and size of cavernomas in vivo in murine models of CCM.5,11
Here, we extend these observations on EndMT switch to sporadic CCM patients.
In addition, we found that p-Smad3 is frequently observed in the nuclei of endothelial cells of sporadic cavernomas (≈90% positive) as in familial cavernomas, whereas β-catenin could be observed only in ≈20% of endothelial nuclei of both sporadic and familial cavernomas.
This result is consistent with our observations in endothelial-specific CCM3-ablated mice where we found that β-catenin signaling represents an early event after mutation of CCM gene and tends to decline in time, whereas p-Smad3 appears later, but persists. In our present working model, β-catenin signaling is required for the initiation of the lesion, whereas Tgf-β/p-Smads signaling sustains its maintenance and further development.11 The surgical samples that we examined in this study represent established lesions where p-Smad signaling is expected to be generally active, whereas β-catenin signaling is low.
The identification of the EndMT switch and of signaling pathways activated in sporadic cavernomas as in familial patients is particularly relevant considering that sporadic patients represent the majority (80%)1 of CCM patients and that CCM has an estimated prevalence of 0.2% to 0.5% in the population. The pharmacological management of this disease is lacking, and therapy is limited to symptomatic treatment and surgery whenever applicable. Therefore, it may be significant for future pharmacological therapies to have defined shared molecular features as potential common targets in familial and sporadic CCM.
We thank the Angioma Alliance US and Amy Akers for providing samples of familial cerebral cavernous malformation (CCM) patients.
Sources of Funding
This study was supported by grants from TELETHON, contract No. GGP14149 to E. Dejana; Associazione Italiana per la Ricerca sul Cancro (AIRC) and Special Program Molecular Clinical Oncology 5×1000 to AGIMM (AIRC-Gruppo Italiano Malattie Mieloproliferative) to E. Dejana; European Research Council (ERC): WNT FOR BRAIN, contract No. 268870; European Commission for Initial Training Networks (ITN) VESSEL, contract No. 317250 to E. Dejana; and for Brain Barriers Training (BtRAIN), contract No. 675619 to E. Dejana; Fondazione CARIPLO (Cassa di Risparmio delle Provincie Lombarde) contract No. 2014-1038 to N. Rudini; R. Cuttano was supported by the FIRC fellowship No. 16617.
The online-only Data Supplement is available with this article at http://stroke.ahajournals.org/lookup/suppl/doi:10.1161/STROKEAHA.115.011867/-/DC1.
- Received October 19, 2015.
- Revision received December 10, 2015.
- Accepted December 29, 2015.
- © 2016 American Heart Association, Inc.
- McDonald DA,
- Shi C,
- Shenkar R,
- Gallione CJ,
- Akers AL,
- Li S,
- et al
- He Y,
- Zhang H,
- Yu L,
- Gunel M,
- Boggon TJ,
- Chen H,
- et al
- Boulday G,
- Rudini N,
- Maddaluno L,
- Blécon A,
- Arnould M,
- Gaudric A,
- et al
- Akers AL,
- Johnson E,
- Steinberg GK,
- Zabramski JM,
- Marchuk DA.
- Stockton RA,
- Shenkar R,
- Awad IA,
- Ginsberg MH.
- Liebner S,
- Cattelino A,
- Gallini R,
- Rudini N,
- Iurlaro M,
- Piccolo S,
- et al
- Bravi L,
- Rudini N,
- Cuttano R,
- Giampietro C,
- Maddaluno L,
- Ferrarini L,
- et al
- 13.↵Cavé-Riant F, Denier C, Labauge P, Cécillon M, Maciazek J, Joutel A, et al. Spectrum and expression analysis of KRIT1 mutations in 121 consecutive and unrelated patients with cerebral cavernous malformations. Eur J Hum Genet. 2002;10:733–740.
- 14.Liquori CL, Berg MJ, Squitieri F, Leedom TP, Ptacek L, Johnson EW, et al. Deletions in CCM2 are a common cause of cerebral cavernous malformations. Am J Hum Genet. 2007;80:69–75.
- Glading AJ,
- Ginsberg MH.