Vascular Morphogenesis: A Human Genetics Approach
Advances in our understanding of fundamental biological events can often be made by the analysis of defects manifested in inherited diseases. The genes responsible for these genetic syndromes often encode proteins that act at critical points of the pathways that control fundamental biological processes such as cell division, differentiation, and cell death. This approach has lead to the discovery of novel gene products and/or biochemical pathways involved in disease, genes that in turn play a fundamental role in normal biological processes.
My laboratory has taken this genetic approach for the study of angiogenesis, beginning with Mendelian disorders of vascular dysplasia and progressing to more complex vascular phenotypes. Our objectives are twofold: (1) to gain specific knowledge of the pathology observed in these disorders, and (2) to provide basic knowledge on the role of these genes and gene products in angiogenesis. The first step is to identify the genetic loci that underlie these syndromes. These mapping and positional cloning endeavors become the basis for future molecular biological studies on the role of the mutant gene products in the pathology of the disease, and the role of the normal proteins in vascular development. In order to investigate disease mechanism and pathogenesis, we then create an appropriate transgenic or knockout mouse as an animal model of the disease. The animal model serves both as a tool to further understand the pathophysiology of the disease, and a more tractable system to begin to identify other factors (genetic and environmental) that may alter the phenotype. Coming full circle, we can determine if the factors identified in the animal model also modify the clinical phenotype in the human disease. We currently have four projects that utilize this approach, and a new initiative that begins with a mouse model of hypertrophic cardiomyopathy.
Project 1. Molecular Genetics of Arterio-Venous Communication: Hereditary Hemorrhagic Telangiectasia: Hereditary Hemorrhagic Telangiectasia (HHT or Osler-Weber-Rendu disease) is an autosomal dominant disorder characterized by hemorrhagic stroke, gastrointestinal bleeding, and other vascular pathology. The clinical features result from the development of focal vascular malformations characterized by direct arteriovenous shunts with a loss of the capillary beds. The pathology of this disorder suggests a critical role for the HHT gene(s) in vascular development and angiogenesis of the capillary bed. However, until our work on this disorder, the nature of the molecular defect remained unknown. We have shown that HHT is actually a group of related disorders with overlapping but distinct phenotypes and genetic etiologies (McAllister et al., 1994a; Porteous et al., 1994; Berg et al., 1996). We established genetic linkage at two distinct loci for HHT, one on chromosome 9q33 (McDonald et al., 1994) and the other on 12q13 (Johnson DW, et al., 1995). We subsequently identified the gene for HHT 1 as endoglin (McAllister et al., 1994b), a transforming growth factor-ÿ(TGF-ÿÿ) binding protein of endothelial cells, and the HHT 2 locus as the activin-like kinase receptor, ALK-1, which has sequence homology to type I TGF-ÿÿ receptors (Johnson et al., 1996). Expression analyses of the mutant allele for endoglin (McAllister et al., 1995; Gallione et al., 1998) and for ALK-1 (Berg et al., 1997; Klaus et al., 1998) has shown that most alleles lead to unstable message or reduced levels of protein, suggesting inherited haploinsufficiency. These data suggest a critical role for the TGF-ÿÿ signal transduction pathway in the pathology of this disease, and more significantly, in the development and/or repair of blood vessels.
Our research has more recently involved a combination of genetic and molecular biological approaches to further study the role of ALK-1 and endoglin in cardiovascular disease and angiogenesis. We are investigating whether sequence polymorphisms in the endoglin and/or ALK-1 genes are risk factors for cerebrovascular disease in the general population with retrospective case-control association studies. We have already shown that one sequence polymorphism in endoglin is associated with increased risk for hemorrhagic stroke (Alberts et al., 1997), an observation which has now been replicated by others in the Japanese population (Takenaka et al., 1999). We are currently attempting to understand how this intronic polymorphism may affect RNA splicing or message stability. Three endoglin coding-polymorphisms are also under investigation, as the intronic polymorphism may be in linkage disequilibrium with one of these.
Although both endoglin and ALK-1 share sequence homology to the TGF-ÿ family of receptors, little is known about their precise role in signaling. Due to the lack of a signaling assay for ALK-1, there is no evidence other than sequence homology that ALK-1 is a TGF-ÿ receptor. We created a signaling assay for ALK-1 by constructing a chimeric receptor consisting of the extracellular ligand-binding domain of ALK-1 fused to the kinase domain of the TGF-ÿ receptor TÿRI, which can activate a reporter gene driven by the PAI-1 promoter. Using this chimera, we have demonstrated ligand specific activation for TGF-ÿ1, -ÿÿ3 (but not -ÿÿ2), and an additional uncharacterized ligand present in serum (Lux et al., 1999). Significantly, this ligand specificity for the TGF-ÿ isoforms parallels that of endoglin. We have also shown that ALK-1 and endoglin can be immunoprecipitated together in a receptor complex (Lux et al., 1999), suggesting that HHT is a result of altered signaling via an endothelial-specific TGF-ÿ receptor complex.
We have used this signaling assay to attempt to identify the novel ligand present in serum. Results with all available mammalian TGF-ÿ family members were negative (Lux et al., 1999). The Drosophila SAX gene encodes a receptor which may be the most similar in this species to mammalian ALK-1. We have now shown that the Drosophila SAX ligand screw (SCW) activates the ALK-1 chimera in our assay (Lux, Arora, and Marchuk, unpublished). The reciprocal experiment, performed in the laboratory of our collaborator Kavita Arora (UC, Irvine) suggests that mammalian ALK-1 can partially substitute for SCW in Drosophila development. The combined data suggest that ALK-1 signaling may involve a novel TGF-ÿ related ligand, and that a novel biochemical pathway involved in fundamental angiogenic processes awaits identification. Attempts to clone the mammalian homologue of the SCW ligand are in progress, using a variety of bioinformatic, molecular and biochemical approaches. Using our signaling assay, we are attempting to identify the downstream effectors of this pathway, as well as characterize the changes in gene expression due to signaling through these receptors.
We also wish to create animal models for HHT1 and 2. We now have ALK-1 and endoglin knockout mice. Although in both cases the mutant homozygotes show embryonic lethality, we have recently identified gastrointestinal vascular malformations (telangiectasias) in the ALK-1 heterozygotes (Yu, McLendon, and Marchuk, unpublished). The mice have become a critical resource for future studies on HHT, as will be outlined in the section on Future Directions.
Project 2. Molecular Genetics of Venous Angiogenesis: Familial Venous Malformations: Venous malformations are composed of large vascular lumens lined by a single-layered flat mature endothelium, which are thus prone to hemorrhage. The cutaneous and visceral lesions occur as single or multiple lesions with a nodular or tumor-like appearance. The presence of a few families exhibiting autosomal dominant inheritance of these vascular lesions suggested that genetic linkage analysis and positional cloning could be used to identify the responsible gene(s). We confirmed an earlier report by the laboratories of Bjorn Olsen and Matt Warman of linkage to chromosome 9p21 for autosomal dominant venous malformations (Gallione et al., 1995). We subsequently mapped the gene for the tie-2 receptor kinase to the candidate interval, and in collaboration with this group, identified an identical mutation in the kinase domain of tie-2 for two large kindreds (Vikkula et al., 1996). In our laboratory, we subsequently identified a second missense mutations in tie-2 which causes FVM, and have shown that both known mutations constitutively activate the receptor (Calvert et al., 1999). Although the tie-2/angiopoietin pathway has been implicated in vasculogenesis and embryonic angiogenesis from knockout studies in the mouse, our work suggests that this pathway has a role in the maintenance of proper vascular structure extending into adulthood. We speculate that this pathway is involved in endothelial-smooth muscle association based on the anatomy of the vascular lesion.
We are now creating a transgenic mouse model of this disorder, by expressing both tie-2 mutations under the control of its own promoter. This mouse model will be most useful for our understanding of the initiating event in lesion formation in these families, since although the phenotype is expressed in the heterozygous state, there appears to be no global vascular pathology outside of the vascular lesions themselves.
We have also identified 5 additional families with a clinical presentation similar to our tie-2 venous malformation families, but which exclude the tie-2 locus by linkage analysis. In some families the lesions were classified as glomangiomas, a benign tumor derived from the glomus cells (smooth muscle derivative) of the vasculature. We have recently shown that these families map to chromosome 1p (Calvert et al. submitted), and are actively searching for mutations in genes mapping to the critical region.
Project 3. Molecular Genetics of Cerebrovascular Angiogenesis: Cerebral Cavernous Malformations: Cerebral cavernous malformations (CCM) are congenital vascular anomalies of the brain comprising focal, thin-walled, grossly dilated vascular spaces. The lesions are responsible for significant neurologic disability, in particular, intractable migraine, seizures, and hemorrhagic stroke. Autosomal dominant forms of CCM have been described and we and others have shown that a gene for CCM (CCM1) maps to chromosome 7q (Marchuk et al., 1995). Taking advantage of a shared disease haplotype in Mexican-American families, we were able to narrow the critical region to under 500 kb (Johnson et al., 1995c; and unpublished data). We have now identified the CCM1 gene as KRIT1, a recently discovered binding partner of the Krev-1/rap1a tumor suppressor gene (Sahoo et al., in press). A common mutation in most Hispanic families (16 of 21 families analyzed) confirms the founder effect in this population. Other Hispanic and non-Hispanic families harbor different mutations, all of which appear to be null alleles. These data show the strength of this genetic approach to cardiovascular disease, as the Krev-1/rap1a signaling pathway, although implicated in cancer (e.g. Tuberous sclerosis), has not previously been shown to be required for angiogenesis, nor has it been previously implicated in any cardiovascular pathology. We are currently working to create a mouse model of CCM1, as well as to characterize the role of this biochemical pathway in the pathology of hemorrhagic stroke associated with CCM. Mutation analysis is already underway to determine the involvement of KRIT1 mutations/polymorphisms in common (non-hereditary) cerebral cavernous malformations.
Project 4. Molecular Genetics of Capillary Angiogenesis: Hemangiomas: Our work on the autosomal dominant vascular disorders has identified genes involved in the regulation of vascular growth. However, all of these germline mutations are compatible with normal vascular development. We also wish to identify novel genes that might be essential for vasculogenesis and embryonic angiogenesis. Such genes can in principle be identified using human phenotypes, by searching for somatic mutations underlying non-inherited vascular anomalies. We are using this approach to identify the gene(s) underlying the most common tumor (of any kind) in infancy - hemangiomas.
Hemangiomas are benign tumors consisting primarily of proliferating capillaries, which often occur as an elevated purple or red spot on the skin. Hemangiomas usually develop shortly after birth, but are self-limiting and go through a characteristic two-staged process of growth and regression. The rapid proliferation stage suggests an uncontrolled stage of angiogenesis. Our hypothesis is that hemangiomas arise from an early somatic mutation within a critical gene for capillary angiogenesis. Due to the localized nature of the tumor, our prediction is that the mutated gene products will be acting in a cell-autonomous nature, with the tumor resulting from a clonal expansion cell containing the original mutation. Using a clonality assay based on non-random X-chromosome inactivation, we have shown that most proliferative hemangiomas appear mono-clonal (Walter and Marchuk, in preparation). We have also shown significant loss of heterozygosity for markers on chromosome 5q (Berg et al., in press), in the same region that we have identified a genetic susceptibility locus (see below). We are currently searching for mutations in the genes involved in the vascular endothelial growth factor (VEGF) pathway.
Although the great majority of hemangiomas occur sporadically, we have identified a number of families displaying autosomal dominant segregation of childhood hemangiomas (Blei et al., 1998). Using these families to identify genetic loci that may predispose children to hemangioma development, we have found linkage to a region on distal chromosome 5q (Walter et al., 1999). We hope that these independent lines of analysis will converge as we discover that some of these genetic loci are involved in both familial and sporadic cases, the mutations being inherited in the familial cases and acquired somatically in the sporadic cases.
Project 5. Molecular Genetics of Hypertrophic Cardiomyopathy: A Mouse Model of Heart Failure: Heart failure is a common cause of mortality in the United States. It is the final outcome of a variety of conditions, both primary and secondary, that affect the heart. Myocardial hypertrophy and the progression to heart failure are highly heterogeneous and are the result of a combination of environmental and genetic factors. The genetic factors in particular have been recalcitrant to identification.
Mice with cardiac-specific over-expression of the calsequestrin (CSQ) gene develop a severe form of hypertrophic cardiomyopathy that serves as an appropriate mouse model of heart failure. Intriguingly, the extent of cardiomyopathy and the progression to heart failure are highly strain dependent, suggesting the existence of modifying genes in each strain. Using a backcross strategy and quantitative trait locus (QTL) mapping, we have identified three genetic loci that affect the progression of disease in this mouse model (Carlson, Suzuki, Marchuk and Rockman, in preparation). A gene on mouse chromosome 2 strongly affects the progression to heart failure, accounting for 37% of the phenotypic variation in the disease outcome. A locus on chromosome 4 shows a slightly lower effect on disease outcome. In addition, another locus on chromosome 3 strongly affects the extent of cardiomyopathy, as measured by % fractional shortening of the heart. We are in the process of refining the map positions of these three loci, with the goal of the identification of these genes affecting cardiomyopathy and heart failure in this mouse model. Polymorphic variants in these same genes in the human population may be the elusive genetic risk factors for heart failure. Once identified, these will be tested in case-control association studies with appropriate patient populations.