Molecular Genetics and Neurobiology of Neurodegenerative and Neurodevelopmental Disorders

The major focus of the research in our laboratory is aimed at understanding brain biology by studying specific neurologic disorders and selected neuronal genes. Toward this goal, we have been studying a neurodegenerative disorder known as spinocerebellar ataxia type 1 (SCA1) and two neurodevelopmental disorders Rett syndrome and microphthalmia with linear skin defects (MLS). More recently, we began studies aimed at identifying mammalian homologs of genes known to be essential for normal development of the nervous system in Drosophila. Studies of the vertebrate homologs will allow us to gain insight into vertebrate neural development and will provide candidate genes for human neurodevelopmental disorders. The following presentation summarizes the research progress in studies aimed at understanding the pathogenesis of SCA1 and cloning the MLS gene.

STUDIES IN SPINOCEREBELLAR ATAXIA TYPE 1

The dominantly inherited spinocerebellar ataxias (SCAs) are a group of heterogeneous disorders characterized by progressive ataxia, amyotrophy, variable degrees of sensory loss, ophthalmoparesis, and extrapyramidal symptoms. Progressive neuronal degeneration in the cerebellum, spinal tracts, and brain stem leads to total loss of motor control and coordination, and eventually death. For decades, neurologists and neuropathologists have classified these diseases using clinical and pathologic criteria, but often the overlapping phenotypes and the intrafamilial variability rendered most of these classifications suboptimal^{(1, 2)}. Over the past two decades, genetic mapping and cloning allowed the classification of at least six SCA loci (SCA1, SCA2, SCA3, SCA4, SCA5, and SCA7) and underscored the importance of genetic studies in clearly delineating the various subtypes⁽³⁾.

SCA1 is the first dominant ataxia locus to be identified based on genetic mapping studies. The SCA1 gene was mapped to the short arm of human chromosome 6 (6p) through linkage to the human leukocyte antigen (HLA) complex^{(4, 5)}. The clinical features of patients with genetically proven SCA1 include progressive gait and limb ataxia, dysarthria, muscle atrophy, loss of proprioception and vibratory sense, and ophthalmoparesis. In the later stages of the disease, bulbar dysfunction becomes prominent as manifested by severe dysarthria and dysphagia. Frequent choking spells, breathing difficulty, and recurrent aspiration pneumonia lead to respiratory failure and death. Although SCA1 is typically an adult onset disease with symptoms developing in the third to fourth decade, juvenile onset with a more rapidly progressive course has been reported^{(6, 7)}. The primary sites of neurodegeneration in SCA1 include the cerebellar cortex where marked loss of Purkinje cells and dentate nucleus neurons is noted, the inferior olive, dorsal columns and spinocerebellar tracts, and cranial nerve nuclei III, IV, IX, X, and XII⁽⁸⁾.

Detailed genetic mapping studies using a large number of highly polymorphic DNA markers and families studied in our laboratory and in the laboratory of our collaborator, Dr. Harry T. Orr (University of Minnesota, MN), allowed mapping of the SCA1 locus distal to HLA in 6p22-p23, and eventually to a 2 centiMorgan region flanked by the DNA markers D6S89 and D6S274^(9–11). The DNA from this region was cloned in overlapping yeast artificial chromosome clones and was determined to span approximately 1.2 million base pairs⁽¹²⁾. Because of the observation of anticipation (earlier onset of disease and more severe course in successive generations) in SCA1 kindreds^{(6, 7)} and the finding that expansion of trinucleotide repeats was the molecular basis for anticipation in myotonic dystrophy and the fragile X syndrome, we hypothesized that SCA1 is most likely caused by expansion of an unstable trinucleotide repeat. To test this hypothesis, we searched the candidate SCA1 region for trinucleotide repeats and identified a highly polymorphic CAG repeat, which proved unstable and expanded in affected individuals⁽¹³⁾. Normal alleles contain 6-42 repeats, whereas SCA1 alleles contain 40-82 repeats. The CAG repeat tract is uninterrupted on SCA1 chromosomes, whereas 1-3 CAT interruptions are found on all normal alleles with 22 or more repeats⁽¹⁴⁾. This is very helpful for determining whether alleles containing 35-45 repeats are pathogenic or normal. Normal alleles that are interrupted with at least one CAT unit can be easily identified using the restriction enzyme SfaNI which cleaves GCATC(N)₅^{(14, 15)}. The difference in the configuration of the SCA1 trinucleotide repeat on stable normal alleles and unstable expanded alleles led us to propose that CAT interruptions confer stability to normal alleles. To test this hypothesis, the stability of the repeat was evaluated in somatic tissues in a normal allele containing 39 repeat with a CAT interruption and a disease allele containing 40 CAG repeats without interruption. Those two alleles which are nearly identical in size were quite different with respect to repeat stability in peripheral blood leukocytes. The allele with CAT interruption was stable whereas that without interruption was unstable showing both expansions and deletions of the CAG tract⁽¹⁶⁾. These data support the hypothesis that the variant CAT stabilizes the repetitive CAG tract.

Analysis of the SCA1 CAG repeat in all families known to have SCA1 based on linkage data showed that the expansion beyond 40 repeats is the mutational mechanism in all the families and suggested that other mutations in the SCA1 gene are unlikely to cause the phenotype. An inverse correlation between the number of repeat units and the age of onset of disease was demonstrated, providing an explanation for the clinical observation of anticipation (Fig. 1).

Figure 1

Relationship between the age of onset and number of CAG repeats on SCA1 alleles from 113 symptomatic individuals. An inverse correlation is observed, r = -0.814 (p = 0.0001). Reproduced from Ranum et al.⁽¹⁰⁾ with permission from the American Journal of Human Genetics.

The CAG repeat resides within the coding region of an 11-kb transcript, which is expressed in a wide variety of tissues⁽¹⁷⁾. The SCA1 gene product, ataxin-1, is a novel protein and is predicted to contain 792-829 amino acids, depending on the number of repeats. Ataxin-1 is localized in the nuclei of neuronal cells in various brain regions, but in cerebellar Purkinje cells it has both nuclear and cytoplasmic localization⁽¹⁸⁾. Western analysis of cellular and brain extracts prepared from SCA1 patients with differing number of CAG repeats identified two bands. One band, estimated at 100 kD, is found in both patients and controls. The second band is found only in extracts from SCA1 patients, and its size is increased proportional to the number of CAG repeats on the expanded allele (Fig. 2). These results confirmed that the CAG repeat is within the coding region of ataxin-1 and support the hypothesis that the pathogenic effect of the mutation is mediated at the protein level.

Figure 2

Expression analysis of ataxin-1, the SCA1 gene product, in lymphoblast extracts from normal (NL) and SCA1 individuals. The number of CAG repeats in both alleles is shown. Both normal alleles and expanded alleles are translated. The mobility of the protein encoded by the expanded allele varies according to the number of CAG repeats. Reproduced from Servadio et al.⁽¹⁸⁾ with permission from Nature Genetics.

SCA1 is one of a growing number of neurodegenerative disorders caused by expansion of a translated CAG trinucleotide repeat which codes for glutamine. This group of diseases includes spinobulbar muscular atrophy (SBMA), Huntington disease (HD), SCA1, dentatorubropallidoluysian atrophy (DRPLA), and type 3 spinocerebellar ataxia (SCA3). In each of these disorders the gene is expressed in a wide variety of tissues and brain regions, yet the neuronal loss is specific⁽¹⁹⁾. A major question relevant to the pathogenesis of these disorders is the mechanism by which modest glutamine expansion leads to slow and selective neuronal degeneration. We and others proposed a gain of function mechanism of pathogenesis whereby the mutant proteins interact with new targets which may be cell-specific. The finding of individuals who do not have symptoms of SBMA or Huntington disease in spite of hemizygous and heterozygous deletions of the androgen receptor and HD gene, respectively, support the gain of function hypothesis^{(20, 21)}. To address the mechanism of pathogenesis in SCA1, studies were pursued to generate transgenic animals that harbor the human mutation and to identify putative ataxin-1 interactors.

In collaboration with Dr. Harry T. Orr, transgenic animals harboring an SCA1 cDNA with either 30 CAG repeats or 82 repeats under the control of a Purkinje cell specific promotor (Pcp2) were generated. Transgenic mice carrying a gene with 30 repeats are normal; however, transgenic mice with 82 repeats develop ataxia and Purkinje cell degeneration⁽²²⁾. We analyzed transgene ataxin-1 expression in these animals and found that, by immunoblotting, human ataxin-1 is detected in animals with 30 repeats but not in animals with 82 repeats. However, it was possible to detect ataxin-1 in the Purkinje cells of animals harboring either the wild type or expanded alleles using immunohistochemical analysis and the same antisera used for Western analysis. This finding, which is quite intriguing, suggests that the product of the expanded SCA1 allele is altered in Purkinje cells in a manner that renders it undetectable by immunoblotting. This feature is specific to Purkinje cells, the primary site of pathology in SCA1, because the mutant protein is easily detected by Western analysis in a variety of tissues in humans⁽¹⁸⁾.

To identify which proteins might interact with mutant ataxin-1, we used ataxin-1 as bait in the yeast two-hybrid system and identified the glycolytic enzyme glyceraldehyde-3-phosphate dehydrogenase (GAPDH) as a putative interactor⁽²³⁾. After demonstrating that GAPDH interacted specifically with ataxin-1 but not with 10 different heterologous bait fusion proteins, we studied the interaction in vitro using commercially available rabbit muscle GAPDH and either wild-type or mutant ataxin-1 over expressed in COS7 cells or from SCA1 lymphoblast extracts. Both wild type and mutant ataxin-1 were found to bind GAPDH, and the binding persisted up to 1 M NaCl wash and could be eluted only by using 2% formic acid. The binding of ataxin-1 to GAPDH was found to involve the N-terminal region of ataxin-1 that contains the CAG repeat and the NAD binding domain of GAPDH⁽²³⁾. We also demonstrated similar interactions(using yeast two-hybrid and in vitro binding) between GAPDH and the androgen receptor, the protein implicated in spinobulbar muscular atrophy. Using a different strategy, Burke et al.⁽²⁴⁾ demonstrated that the Huntington disease and dentatorubropallidoluysian atrophy gene products bind GAPDH. The finding that GAPDH inter-acts with the polyglutamine tracts of four gene products mutated in neurodegenerative diseases is provocative and interesting. GAPDH is a key regulatory enzyme in glycolysis where it catalyzes the oxidation and phosphorylation of D-glyceraldehyde-3-phosphate to 1,3-diphosphoglycerate and is involved in the conversion of glucose to pyruvic acid. Impairment of this enzyme by aberrant interaction with polyglutamine tracts could be a common pathogenic mechanism by impairment of neuronal energy metabolism. Although this is an attractive hypothesis, the physiologic relevance of polyglutamine/GAPDH interactions to neurodegeneration needs to be carefully investigated in vivo. This is particularly important given the high abundance of GAPDH in comparison with the polyglutamine proteins and the selective neuronal loss in face of wide cellular distribution of GAPDH.

The identification of the SCA1 gene and mutation has both clinical and biologic implications. The availability of an accurate DNA-based diagnostic test will clearly facilitate the diagnosis of SCA1 and will provide individuals at risk the option of presymptomatic and prenatal diagnosis. It is important to emphasize, however, that presymptomatic testing should be limited to adults after proper counseling. The availability of anti-ataxin-1 antisera and a transgenic mouse model, manifesting ataxia and Purkinje cell degeneration, should facilitate studies aimed at understanding the pathogenesis of SCA1 and the biology of polyglutamine mediated neurodegeneration.

MOLECULAR STUDIES IN MICROPHTHALMIA WITH LINEAR SKIN DEFECTS

Microphthalmia with linear skin defects syndrome (MLS) is a developmental disorder characterized by linear skin hypoplasia involving the face and neck, microphthalmia, corneal opacity, agenesis of the corpus callosum, seizures, heart defects, and mental retardation (MIM 309801). The syndrome was first described in 1990, and it is found in the literature under several names. MLS results from chromosomal rearrangements involving human Xp22 and leading to terminal deletions of Xpter-Xp22. All of the MLS patients are female with the exception of two XX male patients who have unbalanced X/Y translocations resulting in monosomy for Xpter-Xp22 and disomy for the rest of the X chromosome⁽²⁵⁾ (our unpublished data). These data indicate that MLS is a dominant disorder with lethality in males and that the gene(s) deleted in MLS patients are essential for normal development. There are two additional male-lethal X-linked dominant disorders, which share one or more features with MLS. These are the neurodevelopmental disorder Aicardi syndrome and Goltz syndrome. Aicardi syndrome is characterized by agenesis of the corpus callosum, neuronal migration defects, seizures, mental retardation, microphthalmia, and chorioretinal abnormalities. Patients with Goltz syndrome have linear asymmetric areas of skin hypoplasia throughout the body and following the lines of Blashchko, microphthalmia or anophthalmia, aniridia, chorioretinal abnormalities, and skeletal defects of trunk and limb. Because of the similarities in the phenotype of all three disorders and because all three are X-linked male lethal, we proposed that the same gene or genes may be involved in all three disorders and that the variability of the phenotype depends on patterns of X-inactivation and the type of the molecular defect(deletion versus point mutation or even possibly duplication)^{(25, 26)}.

To define the candidate region for MLS we characterized 54 cell lines from MLS patients and non-MLS patients with rearrangements in Xpter-Xp22. This analysis led to a high resolution map of a 30-megabase region of Xpter-Xp22 and the identification of the MLS candidate region^{(26, 27)}. The characterization of the candidate region revealed that there are no viable male patients with deletions which extend into the MLS region, confirming that the MLS gene(s) is essential for survival. Furthermore, the sizes of the deletions in the different patients did not help delineate which genes may be responsible for various components of the phenotype as is typical for contiguous deletion syndrome. In fact the patients with the most severe phenotype (microphthalmia, skin defects, agenesis of the corpus callosum, seizures, and mental retardation) had the smallest deletion, whereas the patient with the mildest phenotype (skin defects only) had the largest deletion. These data indicate that larger deletions result in preferential survival of cells harboring the normal X chromosome active except in a few tissues, such as the skin, whereas cells harboring the X chromosome with smaller deletion have a better chance for survival and hence result in a more severe phenotype. Using long range restriction mapping and yeast artificial chromosome YAC clones isolated in our laboratory, we refined the candidate region to approximately 450 kb of DNA and cloned this region in overlapping cosmids⁽²⁸⁾. We focused on the isolation and characterization of candidate genes from the region using exon trapping and cross-species conservation strategy⁽²⁹⁾. Two genes have been isolated so far, and the genomic loci for these two genes have been found to span 150 kb, approximately one third the MLS candidate region (Fig. 3).

Figure 3

Physical mapping of the MLS candidate region. A schematic summary of the mapping of the MLS candidate region in Xp22 is shown. At the top a few selected patient break points (represented by BA numbers) are shown. Bars indicate the regions which are present in the abnormal X chromosomes. The white bars represent patients who do not show features of MLS(BA 95 and BA 333). The striped bars represent patients who do have features of MLS. Solid squares represent the Xp22 markers that were used to determine the location of the patient break points. The MLS critical region has been localized between the break points of BA 333 and BA 325. The patient with break point BA 325 has all of the features observed in MLS. The position of the two candidate genes HCCS and RhoGAPX-1 are shown at the bottom of the figure. tel, telomere; cen, centromere.

The first gene we identified is a putative holocytochrome c-type synthetase (HCCS)⁽²⁹⁾. We attributed this putative function to the gene based on the homology of the deduced open reading frame in the human gene we isolated to other HCCS proteins from Caenorhabditis elegans, Saccharomyces cerevisiae, and Neurospora crassa. The similarities ranged from 54 to 68% and identity from 36 to 52%. HCCS is required for the post translational modification of apocytochrome c and its transport into the mitochondria, where it functions as an essential component of the electron transport chain. The deletion of the human synthetase in the MLS region could contribute to the death of male embryos with deletions spanning this region, and HCCS does undergo X-inactivation, which would be predicted for the MLS gene(s). However, it is not ceratin at this time if deletions of HCCS account for any of the phenotypic features of MLS.

The second gene we identified proved to be a novel member of the rho GAP family of proteins based on sequence analysis. This gene which we termed rhoGAPX-1 contains a rho-type GTPase-activating protein domain and a putative proline-rich SH3-binding domain. GAP proteins stimulate the GTPase activity of their specific target GTPases by facilitating their conversion from an active GTP-bound form to an inactive GDP-bound form⁽³⁰⁾. The rhoGAP family of proteins are involved in a wide variety of biologic processes, and at the cellular level have been implicated in the regulation of the actin cytoskeleton. Defects in the regulation of rho GTPases can produce broad developmental phenotypes. The product of the rotund locus in Drosophila, which contains rhoGAP, SH3-binding, and cysteine-rich domains, is involved in the morphogenesis of appendages. Flies with mutant rotund fail to form appendages due to cell death in the imaginal discs in the developing larvae⁽³¹⁾. The gene for the developmental disorder Aarskog-Scott syndrome is highly homologous to rho- type guanine-exchange factors (GEFs), which perform the converse function of GAPs in converting GTPases from inactive to active form⁽³²⁾. In light of these data, rhoGAPX-1 is an attractive candidate for causing the wide range of developmental defects seen in MLS. Like rotund and the Aarskog-Scott gene, rhoGAPX-1 contains an SH3-binding domain in addition to the GAP domain, suggesting a possible role in regulating the interactions of signaling molecules with the actin cytoskeleton.

Two approaches will be taken to determine whether either of the above genes are involved in MLS. The first approach will rely on mutation analysis. Because all MLS patients have deletions, we cannot perform mutation analysis on these patients to prove the identity of any putative candidate. The phenotypic similarity between MLS and Aicardi and Goltz, and the inheritance pattern in all three, make the latter two disorders candidates for the search for mutations. One of the MLS patients for whom we demonstrated to have a deletion of the candidate region is reported in the literature as an Aicardi patient⁽³³⁾, emphasizing the phenotypic overlap. Large numbers of Aicardi and Goltz patients have been screened for chromosomal abnormalities, and none was found, suggesting that the molecular defect is either a submicroscopic rearrangement or involves small mutations.

The second approach will focus on generating null mouse mutants for either of these two genes, using homologous recombination in embryonic stem cells. This approach will provide insight about the function of these two genes and will determine whether either is responsible for any of the abnormalities seen in MLS. As additional genes are isolated from the MLS candidate region, these genes will be systematically characterized to determine their role in causing the phenotype.

The identification of the gene(s) causing MLS should provide insight in the pathogenesis of this developmental disorder and possibly the molecular basis of Aicardi and Goltz syndromes.