DNA to RNA to Protein

The central dogma of genetics has been that DNA is transcribed into RNA that is than translated into protein—the “business” end of the process. So, following its transcription from DNA in the nucleus, mRNA is transported out of the nucleus to the cytoplasm, and possibly to a specific subcellular location depending on the mRNA, where it can be deciphered by the cell. This takes place via interaction with a complex known as the ribosome, which binds the mRNA and converts its genetic information into protein via the process of translation. The ribosome initiates translation at a pre-encoded start site and converts the mRNA sequence into protein until a designated termination site is reached. Sequence information is read in three-nucleotide groups called codons, each of which specifies an individual amino acid. With the four distinct bases, there are mathematically 64 possible codons, but these have an element of redundancy and code for only 20 different amino acids and 3 termination signals (UAG, UGA, and UAA), also called stop codons. The start codon is ATG and codes for methionine. These amino acids are joined by the ribosome to synthesize a protein. This protein, which may undergo further modification, will ultimately carry out a programmed biological function in the cell. Regulation of this process is highly coordinated and important in learning, for example, where activity-dependent translation at the synapse underlies some aspects of synaptic plasticity, which may go awry in certain disorders such as fragile X syndrome and autism (Morrow et al., 2008).

Over the last decade, the discovery of several classes of functional non–protein coding RNAs has added additional complexity to our understanding of how the genetic code is manifest at the level of cellular function. Of these, microRNAs (miRNAs) are increasingly being recognized as vital players in gene regulation and neurological disease (Weinberg and Wood, 2009). Nascent miRNA molecules are processed to form short (approximately 22-nucleotide) RNA duplexes that target endogenous cellular machinery to specific coding RNAs and induce posttranscriptional gene silencing through a diverse repertoire including RNA cleavage, translational blocking, transport to inactive cell sites, or promotion of RNA decay (Filipowicz et al., 2008; Weinberg and Wood, 2009). Depending on the cell and the context, miRNA activity can result in specific gene inactivation, functional repression, or more subtle regulatory effects and may involve multiple RNAs in a given biological pathway (Flynt and Lai, 2008). Estimates suggest that miRNAs may regulate 30% of protein-coding genes, implicating these molecules as important targets for future research into the biology of neurological disease (Filipowicz et al., 2008; Weinberg and Wood, 2009).

For a specific disease-related gene, the DNA sequence present within an individual is referred to as their genotype, and the expression of that code often results in a feature (or features) that can be observed or measured, known as the phenotype. Genes are further organized into higher-order structures termed chromosomes, which together comprise the entire set of DNA, or genome, of the individual. The human genome is diploid, meaning we possess 23 pairs of chromosomes, 22 autosomes and 1 sex chromosome. Consequently, normal individuals possess two copies (or alleles) of every autosomal gene, one from the mother and one from the father. Because there are two distinct sex chromosomes, X and Y, genes on these chromosomes are expressed in a slightly different manner, discussed in more detail later for the sex-linked disorders.

It is important to emphasize that most genes are not simply “on” or “off.” In reality, cells maintain strict regulatory control over their genes. Some genes, such as those required for cell structure or maintenance, must be expressed constitutively, but genes with specific precise functions may only be needed in certain cells at certain times under certain conditions. Potential levels of regulation are depicted in Fig. 40.2 and include virtually every stage of gene expression. Initially, genes can be regulated at the level of transcription, ranging from the regulated binding of histone proteins, which leads to chromosome condensation, inactivating genes, to the coordinated activity of protein factors that activate or repress gene transcription in response to cell state, environmental conditions, or other factors. Once expressed, the RNA is subject to processing regulation, particularly through alternative splicing as already discussed. Transport of the mRNA and its translation provide additional steps for cellular regulation. Lastly, the final protein can be subject to control via posttranslational modifications or interactions with other proteins. To operate, all these levels of regulation require trans-acting factors, such as proteins, which stimulate or repress a particular step, as well as cis-acting elements, sequences recognized and bound by the regulatory factors.

These detailed levels of regulation provide a dynamic and expansive capability to precisely control cellular function, essential for growth, development, and survival in an unpredictable environment. However, this also provides many potential points at which disease can arise from disrupted regulation. Consequently, a defective gene could cause disease directly through its own action or indirectly by disrupting regulation of other cellular pathways. For example, the forkhead box P2 (FOXP2) transcription factor regulates the expression of genes thought to be important for the development of spoken language (Konopka et al., 2009). Mutations in this gene cause an autosomal dominant disorder characterized by impairment of speech articulation and language processing (Lai et al., 2001). However, other mutations in this gene are responsible for approximately 1% to 2% of sporadic developmental verbal dyspraxia (MacDermot et al., 2005), likely via downstream effects. Mutation of the methyl-CpG-binding protein 2 (MECP2), which regulates chromatin structure, causes the neurodevelopmental disorder, Rett syndrome, but other mutations in this gene can cause intellectual disability or autism (Gonzales and LaSalle, 2010). Similarly, the FOX1 protein (also called ataxin 2 binding protein 1, or A2BP1), a neuron-specific RNA splicing factor (Underwood et al., 2005) predicted to regulate a large network of genes important to neurodevelopment (Yeo et al., 2009; Zhang et al., 2008), causes autistic spectrum disorder when disrupted (Martin et al., 2007) but has also been implicated as a susceptibility gene associated with both primary biliary cirrhosis (Joshita et al., 2010) and hand osteoarthritis (Zhai et al., 2009), presumably due to downstream effects or specific effects in non-neural tissues. This concept of genes acting on other genes will be further explored later (see Common Neurological Disorders and Complex Disease Genetics).

In addition to the complexity of regulatory mutations that affect gene expression by altering RNA or protein levels or by disrupting RNA splicing, there are certain mutations that do not cause protein dysfunction, but instead have effects restricted to the RNA itself. For example, RNA inclusions are found in several forms of triplet repeat disorders (see Repeat Expansion Disorders) including myotonic dystrophy type 1 and the fragile X–associated tremor/ataxia syndrome (FXTAS) (Garcia-Arocena and Hagerman, 2010; Orr and Zoghbi, 2007). The latter is particularly interesting from a genetic standpoint, because a disorder of late-onset progressive ataxia, tremor, and cognitive impairment occurs in carriers of FMR1 alleles of intermediate sizes, which are not full fragile X–causing mutations (Garcia-Arocena and Hagerman, 2010). FXTAS is a dominant gain-of-function disease that occurs via an entirely different mechanism than the recessive loss-of-function disease, fragile X syndrome (Garcia-Arocena and Hagerman, 2010; Penagarikano et al., 2007). Primary disorders of RNA still represent relatively uncharted territory, and it is likely that more RNA-specific diseases will be identified. This is particularly exciting for many reasons, not the least of which is that certain classes of these disorders may be amendable to therapy (Nakamori and Thornton, 2010; Wheeler et al., 2009).

Rare versus Common Variation

As dictated by the principles of natural selection, most genetic variation is not deleterious, and the induced phenotypic variability can be beneficial as a source on which evolution may act. From a clinical standpoint, it is helpful to dichotomize genetic variation into common and rare variation, while accepting that genetic variation is likely a continuum, and the choice of cutoff could be considered arbitrary. Rare genetic variants are of low frequency in the population (<1% frequency), either because they are deleterious and selected against, or because they are new and most often benign. Common genetic variation (>1% to 5% population frequency), on the other hand, is either adaptive, neutral, or not deleterious enough to be subject to strong negative selection; such variants are referred to as polymorphisms. The preeminent genetic model has been that common disease susceptibility is related to common genetic variation, and more rare forms of disease are caused by rare genetic variants, so-called mutations, which act in a Mendelian fashion. In contrast, common variants or polymorphisms may increase susceptibility for disease, but alone are not sufficient to cause disease (see Common Neurological Disorders and Complex Disease Genetics).

Polymorphisms and Point Mutations

The most prevalent form of genetic polymorphism is the single nucleotide polymorphism (SNP), which occurs on average every 300 to 1000 base pairs in the human genome. Most of these SNPs are relatively benign on their own and do not directly cause disease, so for the purposes of this initial discussion, we will concern ourselves primarily with mutations: rare genetic variants sufficient to cause disease. Pathogenic mutations can occur in numerous ways and vary from single nucleotide changes to gross rearrangements of chromosomes (Fig. 40.4). Owing to the large volume of DNA in the human genome, heritable mutations can arise spontaneously in the germline over time through errors in DNA replication or from DNA damage by metabolic or environmental sources despite the constant surveillance of extensive cellular preventive proofreading and repair mechanisms. Thus, mutations can be inherited from the parent or occur de novo in the germline. An example of a common de novo variant is trisomy 21, which causes Down syndrome (discussed further in Chromosomal Analysis and Abnormalities). The smallest pathogenic alterations, termed point mutations, involve a change in a single nucleotide within a DNA sequence. A point mutation can result in one of three possible effects with respect to protein: (1) a change to a different amino acid, called a missense mutation, (2) a change to a termination codon, called a nonsense mutation, or (3) creation of a new sequence that is silent with regard to protein sequence but alters some aspect of gene regulation, such as RNA splicing or transcriptional expression levels. Nonsense mutations can cause premature truncation of a protein, whereas a missense mutation can affect a protein in different ways depending on the chemical properties of the new amino acid and whether the change is located in a region of functional importance.

Fig. 40.4 Genetic mutations. A, Categories of chromosomal aberrations. Paired homologous chromosomes are shown, with various anomalies indicated. An insertional translocation is depicted; other common types include reciprocal translocations and centric fusions (Robertsonian translocations). B, Types of point mutations. A generic DNA sequence is shown (boxed) along with its corresponding mRNA sequence. Codons are indicated, as are their translation into protein (designed by the standard three-letter code). Mutations are in purple, as are the corresponding alterations in the mRNA and protein if present. Note that silent point mutations do not alter the protein sequence. C, Repeat expansion disorders. An example mRNA is shown with a CAG-codon (polyglutamine) repeat region indicated. In the expanded form, an additional number of repeats are present which may perturb the function of the protein produced and/or lead to cell damage via the expanded polyglutamine region (see text for details).

It should be emphasized that not all point mutations are disease-causing variants, although until recently, many considered that a premature stop codon was a “smoking gun.” Whole genome sequencing demonstrates that more than 100 such nonsense mutations may exist per genome, and the vast majority are expected to be relatively benign (Lupski et al., 2010; see Whole Genome/Exome Sequencing in Disease Gene Discovery). So in many cases, the pathogenicity of rare variants is not immediately discernable, and without strong statistical or functional evidence, labeling such genetic variation a mutation is premature and may be misleading. It is likely that most of these, including some variants thought previously to cause rare Mendelian diseases, may simply be benign genetic variation. This is because even a complete knockout of one allele caused by a premature stop codon (haploinsufficiency) may have no discernable effect on gene function for a majority of genes in the human genome (Lupski et al., 2010; Ng et al., 2009; Yngvadottir et al., 2009).

Occasionally, silent coding mutations or point mutations in noncoding regions may be significant for disease if they damage sequences important for gene expression (e.g., transcriptional and/or RNA processing regulatory elements). It has been estimated that up to half of all disease-causing mutations impact RNA splicing, which can have dire consequences given the importance of splicing to regulated gene expression. Such is the case for frontotemporal dementia with parkinsonism linked to chromosome 17 (FTDP-17), where in some populations, the most common mutations disrupt splicing, causing a pathogenic imbalance in tau isoforms (D’Souza and Schellenberg, 2005). As for noncoding mutations, given the large volume of such sequences in the human genome—perhaps up to 96%—and our still imprecise ability to predict sequences required for regulation or to interpret identified sequence changes without direct experimentation, the majority of these mutations likely go unrecognized. It is hoped that the next generation of sequencing and bioinformatic technologies will allow for a better understanding of the role of these types of mutation in human disease.

Structural Chromosomal Abnormalities and Copy Number Variation

Small deletions and insertions can occur through slippage and strand mispairing at regions of short, tandem DNA repeats during replication. If the deletion or insertion is not a multiple of three, a frameshift will result, which leads to the translation of an altered protein sequence from the site of the mutation. On a larger scale, errors of chromosomal replication or recombination can result in inversions, translocations, deletions, duplications, or insertions (Stankiewicz and Lupski, 2010). When the region of deletion or duplication is greater than 1 kb, this is referred to as a copy number variation (CNV). Copy number variation is far more common than previously suspected, and it is estimated that at least 4% of the human genome varies in copy number (Conrad et al., 2010; Redon et al., 2006), much of which is commonly observed in the population and benign (Conrad et al., 2010). However, some rare CNVs such as the recurrent chromosome 17p12 duplication underlying most cases of Charcot-Marie-Tooth type 1A (Shchelochkov et al., 2010) or the alpha-synuclein triplication that can cause Parkinson disease (PD)(Singleton et al., 2003) are pathogenic and act in a Mendelian fashion. Even though such changes may be extensive, they may not be pathogenic if they do not disrupt expression of any key genes. This is particularly true for balanced translocations where genetic material is rearranged between chromosomes, yet no significant portion is actually lost. Although an individual with such a condition may be normal, if the germline is affected, their offspring may receive unbalanced chromosomal material and consequently develop a clinical phenotype (Kovaleva and Shaffer, 2003). CNVs will be discussed in greater detail when we consider common and complex disease genetics (see Copy Number Variation and Comparative Genomic Hybridization).

Repeat Expansion Disorders

Most mutations thus far discussed pass from parent to offspring unaltered, and in large affected families, the identical mutation can potentially be traced back generations. In contrast, there is a specific class of mutation, the repeat expansion (Orr and Zoghbi, 2007) (Table 40.2), which is unstable and can present with earlier onset and increasing severity in successive generations, a process known as anticipation. There are several examples of diseases caused by expanded repeats in coding sequence (e.g., most spinocerebellar ataxias, HD), as well as examples in noncoding sequence (e.g., fragile X syndrome, myotonic dystrophy) and within an intron (e.g., Friedreich ataxia). Interestingly, virtually all these disorders show neurological symptoms that can include such features as ataxia, intellectual disability, dementia, myotonia, or epilepsy, depending on the disease. The most common repeated sequence seen in these diseases is the CAG triplet, which codes for glutamine and expansion of which is seen in a variety of the spinocerebellar ataxias (SCAs) including SCA types 1, 2, 3, 6, 7, 17, and dentatorubropallidoluysian atrophy (DRPLA). In addition to protein-specific effects, these disorders likely share a common pathogenesis due to the presence of the polyglutamine repeat regions. In some disorders, the phenotype can be quite different depending on the number of repeats, such as in the FMR1 gene, where more than 200 CCG repeats causes fragile X syndrome, but repeats in the premutation range of 60 to 200, from which fully expanded alleles arise, can result in FXTAS or premature ovarian failure (Oostra and Willemsen, 2009). Although in general, the underlying mutation is similar, each specific repeat expansion has distinct effects on its corresponding gene, and thus in addition to varying phenotypes, they may also show very different inheritance patterns as illustrated later (see Disorders of Mendelian Inheritance).

Autosomal Dominant Disorders

Diseases involving autosomal genes that require mutation of only one allele are defined as dominant. In most cases, the affected individual has two distinct alleles of a gene (in this case, one normal and one pathogenic) and is described as being heterozygous. Often these pathogenic mutations impart new functionality, referred to as a toxic gain of function, meaning that the phenotype is produced as a result of the expression of the mutated protein. Other disease mechanisms in dominantly inherited conditions include: (1) haploinsufficiency, where inactivation of a single allele is sufficient to produce disease despite the presence of another normal copy, and (2) dominant negative effects, where a mutated protein disrupts function of the normal protein transcribed from the other nonmutant allele.

Autosomal dominant inheritance is characterized by direct transmission of the disorder from parent to child (Fig. 40.6). Affected individuals are seen in all generations, and a vertical line can be drawn on the pedigree to illustrate the passage of the disorder. Since only one deleterious copy of the disease gene is necessary, risk of transmission from an affected parent is 50%. Since the disorder is autosomal, there is no sex preference, and both males and females can present with the disease. One caveat involves the concept of penetrance, or the percent likelihood that a trait will manifest in a person with a specific genotype. A dominant gene is considered to have complete penetrance if all individuals with a given mutation develop disease. In practice, however, many autosomal dominant genes show varying degrees of penetrance or expressivity, most likely due to the influence of other genes and environmental factors.

Fig. 40.6 Autosomal dominant inheritance. A pedigree diagram is shown, using standard nomenclature. Generations are numbered consecutively on the left, and individuals are numbered within each generation. Males are depicted as squares and females as circles. Affected persons are indicated by filled icons. Death is indicated by a diagonal line. A union producing offspring is indicated by horizontal lines. A diamond represents individuals (n) of unknown sex. A triangle represents a spontaneous abortion. Individuals V-2 and V-3 illustrate the diagramming of dizygotic twins. The proband of the pedigree is indicated by an arrow. An autosomal dominant pedigree demonstrates vertical transmission of disease without a sex preference. On average, 50% of offspring are affected. Individual III-4 represents a case of incomplete penetrance (dark circle) where the individual carries the mutation but does not manifest disease. Anticipation (see text) would be illustrated by increasing severity/onset in patients III-1, IV-2, and V-4.

There are over 400 examples of diseases with neurological phenotypes that show autosomal dominant inheritance (OMIM, 2010). These conditions include hyperkalemic periodic paralysis (voltage-gated sodium channel Na_V1.4 on chromosome 17, often caused by missense mutations), HD (Huntington on chromosome 4, caused by CAG repeat expansion), SCA type 3 (ataxin-3 on chromosome 14, caused by CAG repeat expansion), Charcot-Marie-Tooth type 1B (myelin protein zero on chromosome 1, often caused by missense mutations), early-onset familial Alzheimer disease (AD)(presenilin-1, often caused by missense mutations), frontotemporal dementia with parkinsonism (microtubule-associated protein tau on chromosome 17, often caused by missense or splicing mutations), tuberous sclerosis type 1 (hamartin on chromosome 9, often caused by nonsense mutations and frameshifts), neurofibromatosis type 1 (neurofibromin on chromosome 17, caused by point mutations, frameshifts, and splicing mutations), and familial amyotrophic lateral sclerosis (ALS) (superoxide dismutase-1 on chromosome 21, caused by missense mutations), to name a few. Even rare Mendelian forms of more common syndromes such as epilepsy or sleep disorders (e.g., familial advanced sleep-phase syndrome) have been identified. More detailed lists can be found using the recommended online resources (Table 40.3).

Table 40.3 Selected Online Clinical Neurogenetics Resources

Autosomal Recessive Disorders

Disease involving autosomal genes that require mutation of both alleles is defined as recessive. An unaffected individual who harbors one disease-causing allele is referred to as a carrier of that allele. For some disorders, a mild phenotype can be seen in these individuals, who are then described as symptomatic carriers. An individual with two identical alleles (in this case both pathogenic) is described as being homozygous. Alternatively, if they possess two different pathogenic alleles this is described as being compound heterozygous. In general, autosomal recessive mutations modify the function of the protein in a negative way, meaning that the phenotype is produced because of the absence of the mutated protein. This is referred to as a loss of function.

Autosomal recessive inheritance is characterized by lack of intergenerational transmission, in contrast to dominantly inherited disorders (Fig. 40.7). Affected individuals are seen in single generations, often separated by one or more unaffected generations. Because two deleterious copies of the disease gene are necessary, transmission requires both parents to be either affected or carriers. In the most common scenario when both parents are carriers, the risk of an affected child is 25% (50% from each parent). As with all autosomal disorders, there is no sex preference, and both males and females can present with the disease. In families showing this mode of inheritance, it is important to ask about consanguinity. In rare cases of families with considerable inbreeding, recessive alleles may be so common as to cause disease in successive generations, creating a pseudodominant pattern of inheritance.

Fig. 40.7 Autosomal recessive inheritance. A pedigree diagram is shown, using standard nomenclature as described in Fig. 40.6. Carriers of disease are indicated by half-filled icons. Individuals V-2 and V-3 illustrate the diagramming of monozygotic twins. Consanguineous mating is indicated by a doubled line. An autosomal recessive pedigree demonstrates indirect transmission of disease without a sex preference, often in a single generation (occasionally described as horizontal). On average, 25% of offspring of two carriers are affected.

As mentioned for the autosomal dominant disorders, diseases that share this mode of inheritance may have very distinct types of underlying mutations. Upwards of 600 disorders with autosomal recessive inheritance show neurological symptoms (OMIM, 2010). Examples include Friedreich ataxia (frataxin on chromosome 9, caused by intronic GAA repeat expansion), spinal muscular atrophy type 1 (survival of motor neuron 1 on chromosome 5, caused by deletion of exon 7), Wilson disease (ATPase, Cu⁺⁺ transporting, beta-polypeptide on chromosome 13, often caused by missense mutations), Tay-Sachs disease (hexosaminidase A on chromosome 15, commonly caused by frameshift, splicing, or nonsense mutations), glycogen storage type II or Pompe disease (acid alpha-glucosidase gene on chromosome 17, often caused by point mutations, splicing mutations, and exon deletions), phenylketonuria (phenylalanine hydroxylase on chromosome 12, often caused by missense mutations), and ataxia-telangiectasia (ataxia-telangiectasia mutated on chromosome 11, often caused by point mutations and splicing mutations). More detailed lists can be found using the recommended online resources (see Table 40.3).

Sex-Linked (X-Linked) Disorders

The sex chromosomes in humans are referred to as the X and Y chromosomes, the latter of which programs the individual to be male. There are as yet no known Y-linked diseases, so we will focus on the X chromosome. As males only possess a single X chromosome, they are hemizygous for all its genes, and consequently any pathogenic mutation is expressed by default. Because of this, dominance of X-linked genes applies with respect to whether female carriers express disease. This is complicated by the observation that although females possess two X chromosomes, no single cell expresses genes from both; instead, one chromosome is randomly and permanently inactivated during development via a process known as lyonization. Therefore, all women inherently possess cells of two different genotypes, or are mosaic, for the X chromosome. This can be clinically relevant insofar as disproportionate activation of an abnormal X chromosome could potentially lead to clinical phenotypes in female carriers of recessive X-linked disorders. Usually though, skewing occurs, so that the pathogenic allele is less expressed than the other normal allele.

Recessive X-linked transmission is characterized by the presence of disease in males only (Fig. 40.8). Affected males cannot pass the disease on to their sons, but all their daughters must inherent the abnormal X chromosome and are, therefore, obligate carriers. A carrier female has a 50% chance of passing the disease allele to a child, but all males receiving it will be affected. Dominant X-linked transmission (see Fig. 40.8) is similar, except carrier females are affected and transmit the disease to 50% of their children irrespective of their sex. Affected males usually show a more severe phenotype, or may even exhibit lethality, and transmit the disease to all of their daughters and none of their sons.

Fig. 40.8 X-linked inheritance. A, X-linked recessive disease. A pedigree diagram is shown using standard nomenclature as described in Fig. 40.6. Carriers of disease are indicated by half-filled icons. Disease manifests only in hemizygous males. Fathers cannot pass the disease to their sons, but all daughters of an affected male are obligate carriers of disease. Carrier females have a 50% chance to pass on the disease gene and can have affected sons. In some cases, a female carrier can be mildly symptomatic, usually due to non-random lyonization. B, X-linked dominant disease. A pedigree diagram is shown using standard nomenclature as described in Fig. 40.6. Disease manifests in heterozygous females (although severity may be affected by lyonization). The mutant gene is either lethal in males (as shown here) or has a much more severe phenotype. Affected females pass on the disease 50% of the time.

Over 100 X-linked disorders with neurological phenotypes are known (OMIM, 2010). The majority of these X-linked disorders are recessive, and as seen for the autosomal diseases, mutation type varies widely among the different disorders. Some examples include X-linked adrenoleukodystrophy (ATP-binding cassette subfamily D member 1, commonly caused by missense and frameshift mutations), Duchenne muscular dystrophy (dystrophin, commonly caused by deletions), Emery-Dreifuss muscular dystrophy-1 (emerin, often caused by nonsense mutations), Menkes disease (ATPase, Cu⁺⁺-transporting, alpha-polypeptide, commonly caused by frameshifts, nonsense mutations, and splicing mutations), Fabry disease (alpha-galactosidase A, commonly caused by point mutations, gene rearrangements, and splicing mutations), and Pelizaeus-Merzbacher disease (proteolipid protein-1, often caused by duplications and missense mutations). X-linked dominant disorders include Rett syndrome (methyl-CpG-binding protein-2, often due to missense and nonsense mutations), incontinentia pigmenti (inhibitor of kappa light polypeptide gene enhancer in B cells, kinase gamma [IKBKG], often due to deletions), and Aicardi syndrome (gene unknown). More detailed lists can be found using the recommended online resources (see Table 40.3).

Mitochondrial Disorders

Mitochondria are double-membraned organelles responsible for energy production within the cell via the process of oxidative phosphorylation, which relies on the transfer of electrons through a chain of protein complexes within the inner mitochondrial membrane. Disruption of mitochondrial function can lead to a variety of diseases with multisystem involvement, including prominent neurological symptoms (DiMauro and Hirano, 2009; Zeviani and Carelli, 2007). Mitochondria possess their own genome with 37 genes. Because mitochondria are cytoplasmic and the majority of cytoplasm within the zygote is derived from the egg and not the sperm, disorders involving mitochondrial DNA are inherited through the maternal line (Fig. 40.10). A single cell contains many mitochondria which all replicate independently of the nuclear DNA, so it is possible that a mutation in the mitochondrial genome may be present in some of the mitochondria but not others, a condition termed heteroplasmy. This proportion can affect whether a disease is expressed and, if so, what tissues are affected if a minimum threshold of abnormal mitochondria is reached. Heteroplasmy may also change over time as cells divide and the mitochondria are redistributed. Some examples of such disorders include MELAS (mitochondrial encephalomyopathy, lactic acidosis, and stroke-like episodes, caused by point mutations within the gene encoding mitochondrial tRNA^LEU), MERRF (myoclonic epilepsy with ragged red fibers, caused by point mutations within the gene encoding mitochondrial tRNA^LYS), and LHON (Leber hereditary optic neuropathy, most often caused by point mutations in either of two mitochondrial genes encoding complex I subunits, ND4 or ND6).

Fig. 40.10 Mitochondrial (maternal) inheritance. A pedigree diagram is shown using standard nomenclature as described in Fig. 40.6. As the mutant gene is carried in the mitochondrial genome, disease is passed on to all the offspring of affected females (see text). Males can be affected but cannot pass on disease. Severity and onset of the disease may be affected by heteroplasmy, the proportion of abnormal mitochondria per cell, as illustrated by a severe phenotype seen in patient IV-1.

Because the mitochondria themselves contain only a few genes, the majority of mitochondrial proteins, including the machinery responsible for the replication and repair of the mitochondrial genome, are all encoded by nuclear genes. Since these genes are located within the nuclear genome, despite the fact that their mutation gives rise to dysfunctional mitochondria, the disease will show a Mendelian pattern of inheritance. Some examples include infantile-onset SCA (twinkle on chromosome 10, autosomal recessive, caused by missense mutations), progressive external ophthalmoplegia A2 (adenine nucleotide translocator 1 on chromosome 4, autosomal dominant, caused by missense mutations), and Charcot-Marie-Tooth type 2A2 (mitofusin-2 on chromosome 1, autosomal dominant, often caused by missense mutations). Interestingly, various mutations, commonly missense, of the nuclear gene DNA polymerase gamma (POLG) on chromosome 15, which encodes the polymerase responsible for both replication and repair of the mitochondrial genome, cause a wide variety of diverse phenotypes with different modes of inheritance (Hudson and Chinnery, 2006). These include the autosomal recessive Alpers syndrome of encephalopathy, seizures, and liver failure, an autosomal dominant form of chronic progressive external ophthalmoplegia, and autosomal recessive phenotypes of cerebellar ataxia and peripheral neuropathy, among others.

Imprinting

For most genes, expression is controlled by distinct cellular processes that operate irrespective of the gene’s parental origin. However, for some genes, expression in the offspring differs depending on whether the allele was maternally or paternally inherited, and such genes are described as being imprinted (Spencer, 2009). Imprinting arises from epigenetic modifications such as DNA or histone methylation, which are parent-specific alterations that do not change the actual DNA sequence (Fig. 40.11). One example of this is sex-specific DNA methylation that occurs for some genes during the formation of gametes. In the offspring, the methylated gene is bound by histone proteins forming transcriptionally inactive heterochromatin. This allows all gene expression to be driven by the allele derived from the other parent. This can be dynamic depending on the gene, and the magnitude of differential expression between the alleles can vary based on stage of development, tissue type, and possibly other factors. Deletion of an imprinted region or defective imprinting in gametogenesis can lead to disease as illustrated by observations involving chromosome 15q (Lalande and Calciano, 2007). In this example, differential methylation affects the expression of multiple genes, and loss of maternal patterning can lead to Angelman syndrome, characterized by intellectual impairment, epilepsy, ataxia, and inappropriate laughter, while loss of the paternal pattern causes Prader-Willi syndrome, associated with intellectual impairment, obesity, and behavioral problems. The most common mechanism involves de novo deletion of the imprinted region from one parent, although in some cases, defective imprinting can also occur during gametogenesis. In the majority of cases, defective imprinting occurs spontaneously and is therefore unlikely to recur in families; however, imprinting defects can rarely be due to small deletions involving sequences important for regulating parent-specific methylation.

Fig. 40.11 Epigenetics in human disease. A, Imprinting. Gene expression on human Chromosome 15q11-q13 is subject to epigenetic regulation via imprinting. The region contains the loci for two neurological diseases, Prader-Willi syndrome and Angelman syndrome (see text). When inherited from the father, gene expression occurs from the Prader-Willi locus (blue arrow), and this also inactivates genes at the Angelman locus via a presumed antisense-RNA mechanism (dashed arrow). In contrast, when inherited from the mother, a specific site on the chromosome called the imprinting center (circle) becomes methylated (Me). This methylation causes transcriptional inactivation of the genes within the Prader-Willi locus (X), which correspondingly allows transcription from genes at the Angelman locus (pink arrow). If imprinting does not properly occur, either Angelman or Prader-Willi syndrome will arise depending on whether the maternal or paternal expression pattern is absent. B, Uniparental disomy. During gamete/zygote formation, errors in chromosomal segregation or chromosomal rearrangement can result in retention of all or part of a chromosome inherited from the same parent. Although there is no loss of genetic information, the epigenetic imprinting pattern is lost, and therefore correct gene expression patterns are not retained. For chromosome 15q11-q13, for example, this can give rise to Angelman or Prader-Willi syndrome depending on whether the duplicated chromosome is that of the father or the mother, respectively.

Uniparental Disomy

Uniparental disomy arises when pairs of chromosomes are inherited from the same parent, either in their entirety or in large segments due to segregation errors or chromosomal rearrangement (Kotzot, 2008) (see Fig. 40.11). The uniparentally inherited chromosomes can be identical (isodisomic) or different (heterodisomic). In families where the parents lack underlying chromosomal abnormalities, these events usually occur spontaneously and are unlikely to recur. Disease can result from effects related to loss of chromosomal imprinting, pairing of an autosomal recessive mutation, pairing of an X-linked recessive mutation in a female child, or from the generation of a mosaic trisomy. The disorders most commonly associated with this mechanism are the Prader-Willi and Angelman syndromes, discussed previously for imprinting disorders, which can arise from maternal and paternal uniparental disomy, respectively, due to a loss of the imprinting pattern from the missing parental allele. Down syndrome can also rarely result from a mosaic trisomy. There are several examples in the literature of single cases where an autosomal recessive disease arose in a child from uniparental disomy pairing an abnormal allele from a carrier parent, including disorders such as abetalipoproteinemia, Bloom syndrome, autosomal recessive deafness-1A, spinal muscular atrophy, cystic fibrosis, and others (Zlotogora, 2004).

Common Variants and Genome-Wide Association Studies

As already discussed, genetic linkage provides a means of localizing a disease gene to a specific region of a chromosome by using a DNA marker that tracks with affected individuals within families. Linkage analysis, while not without value in genetically complex disease, is less powered than genetic association studies for identification of common variation in complex genetic disease. Genetic association studies assess whether one or more of a defined set of genetic variants are increased or decreased in a disease versus a control population. If a genetic variant is observed in individuals with disease significantly more often or less than expected by chance, that variant is said to be associated with the disease. When one or a few genes are studied, this is a candidate gene association study. When common variants from across the entire genome are studied in this manner, the result is a genome-wide association study, or GWAS (Mullen et al., 2009; Simon-Sanchez and Singleton, 2008) (Fig. 40.12). Original genetic association studies were conducted with a small number of candidate genes, but advances in technology have permitted GWAS in thousands of subjects in a wide variety of human diseases, including dozens of neurological conditions (see Table 40.5). Although the SNPs themselves may directly influence the disease under study, most often this is not the case, and SNPs are best thought of as markers for the location of a gene(s) or region relevant to the disease. In fact, most alleles of the second major type of common genetic variation, CNVs, are mostly captured by SNPs (Conrad et al., 2010) and can be identified by the common SNP genotyping platforms, allowing GWASs to evaluate the contribution of common inherited CNVs as well as SNPs.

Fig. 40.12 Genome-wide association study (GWAS). A GWAS for disease is performed by genotyping a selected population of cases and controls using microarray or other technology for single nucleotide polymorphisms (SNPs) across the genome. In this example, a sample SNP is depicted, with major and minor alleles illustrated as green or red, respectively. Detailed computational analysis is performed to determine whether any individual SNPs are associated with the disease state greater than by chance. In this example, the major allele (green) is associated with the disease and more likely to be present in cases than controls, reflected in an odds ratio above 1.0. Note that while the SNP in question may be involved in the disease, it may also be a marker near an involved gene.

The model that underlies the value of GWAS is based on the concept that common disease is predicted to arise from the interplay of effects caused by common polymorphisms in multiple genes, as well as environmental and other factors. The aim of a GWAS is to identify these common variants that correlate to risk for the disease in question but do not alone cause the disease. Because the effect size, or increase in odds for a disease, is expected to be small (negative selection would have removed strongly deleterious variants from the population), and many independent genetic markers are tested, large sample sizes are needed to have power to detect genome-wide association. This is further compounded by two of the many major factors challenging GWASs of common neurological diseases, phenotypic and genetic heterogeneity. Phenotypic heterogeneity describes the wide and variable clinical spectrum patients with a particular neurological disorder or syndrome (e.g., frontotemporal dementia, epilepsy, multiple sclerosis, autism) manifest. Genetic heterogeneity refers to the notion that even in those with a relatively homogeneous phenotype, many different genetic factors may be contributing in different individuals to lead to the same phenotype. Both of these forms of heterogeneity require large samples to have adequate power to detect genetic risk factors of even moderate size. The smaller the effect of any given genetic variant, the larger the sample size needed to detect that variant. One strategy that may increase power is to study intermediate phenotypes, or endophenotypes, that may be more related to individual genetic risk factors than the broad clinical diagnosis of a disorder, such as specific measures of language or social behavior in autism (Abrahams and Geschwind, 2008; Alarcon et al., 2008; Vernes et al., 2008). Alternatively, such phenotypes can be used to identify more homogeneous subgroups of patients, such as those with specific forms of pathology, as in TAR DNA-binding protein (TDP-43) inclusion-positive frontotemporal dementia (FTD), which may have improved power in a recent FTD GWAS by reducing heterogeneity (Van Deerlin et al., 2010).

Efficiently generating the extensive genotype data necessary for a GWAS has been made possible using microarray technology (Coppola and Geschwind, 2006; Geschwind, 2003). In this type of experiment, specific fragments of DNA corresponding to the sequences of the target SNPs are immobilized in a grid pattern across a glass slide, termed the array. Genomic DNA from individual cases and controls is fluorescently labeled, hybridized to the slide, and the signals from laser-induced dye excitation are collected. The readout will be a map of the SNP pattern for each patient. Data cleaning, quality control, and statistical analysis are performed to determine whether any SNPs are associated with patients more than controls. Given the large number of independent tests performed in a GWAS, statistical significance is commonly set at 5 × 10⁻⁸ (McCarthy et al., 2008; Wellcome Trust Case Control Consortium, 2007) to correct for multiple comparisons. It is now also considered standard to demonstrate that any statistically significant association identified is present in more than one study population, providing an independent replication of the initial finding. Study power and replication may also both be aided by the availability of shared GWAS data (Box 40.1).

Recently published genome-wide association studies of interest involving neurological disease are shown in Table 40.5. One example illustrating the use of a GWAS in complex disease is from a 2009 study by Ikram and colleagues, who performed a GWAS using a population of 19,602 white persons, of whom 1544 had strokes (Ikram et al., 2009). They identified two intergenic SNPs on chromosome 12p13 with significant genome-wide association for total stroke implicating the NINJ2 gene, which is a cell-adhesion molecule found in radial glia (Ikram et al., 2009). Replication in an independent cohort confirmed the association of one SNP with a combined hazard ratio of 1.29 for ischemic stroke in white persons (Ikram et al., 2009). The mechanism of how NINJ2 increases risk for ischemic stroke is unclear at this time, but the results of this GWAS open up a new avenue of research by highlighting it as a candidate for future molecular and cellular studies into stroke etiology.

GWASs can also contribute to the discovery of biological pathways relevant to disease, as seen in a recent study of FTD patients grouped pathologically by the presence of TDP-43 inclusions. The study identified a susceptibility locus on chromosome 7p21.3 that contained a previously uncharacterized transmembrane protein, TMEM106B (Van Deerlin et al., 2010).

Similarly, for the most common neurodegenerative dementia, Alzheimer dementia, recent GWASs have benefited from large numbers of available cases and expanded the loci known to be associated with disease beyond the apolipoprotein E locus to include other neuronal molecules such as BIN1 and PICALM, which are involved in clathrin-mediated endocytosis and intracellular trafficking, and the apolipoprotein, CLU (Harold et al., 2009; Lambert et al., 2009; Seshadri et al., 2010). In Parkinson disease, recent GWASs in large cohorts of European and Japanese patients identified alpha-synuclein (SNCA) and LRRK2 as susceptibility loci (Edwards et al., 2010; Satake et al., 2009; Simon-Sanchez et al., 2009), which is notable because both genes also give rise to autosomal dominant forms of parkinsonism. The tau protein (MAPT), another gene responsible for autosomal dominant forms of parkinsonism, was also found to be associated with disease in European populations (Edwards et al., 2010; Simon-Sanchez et al., 2009). Together these results suggest a commonality between Mendelian and sporadic forms of this disorder.

It is important that physicians have a clear understanding of the meaning of GWAS results so as to be able to differentiate common variants associated with disease from disease-causing mutations. A potential error to be avoided in the clinical interpretation of GWAS data is directly equating the findings to the future development of the disease. It must be reiterated that the finding of an association with a common variant does not equal the finding of a disease gene. By definition, these common variants must have low penetrance, otherwise they would not be so common in normal individuals, and they would likely act in a more Mendelian way. Furthermore, such variants might be associated with disease modifiers—for example, genes acting either upstream or downstream in pathways where disruption or dysregulation can lead to the disease, or perhaps genes involved in the production or regulation of factors involved in such pathways. Instead of directly causing disease, such modifier genes confer a risk of disease, the magnitude of which is sometimes not directly quantifiable because it involves interaction with other genes and the environment. Therefore, for most conditions, reported GWAS information cannot be directly translated into a clinical setting, because the presence of the variant does not necessarily lead to the disease in most cases, particularly for the more rare disorders. As an example, one of the strongest and best-known identified associations, the apolipoprotein E ε4 allele detected in sporadic AD, with an odds ratio of 4 (Coon et al., 2007), has such an inconsistent predictive value that it is not recommended for routine use in disease prediction nor as a typical part of most clinical dementia evaluations (Knopman et al., 2001). Despite this, some commercial organizations have begun to create direct-to-consumer tests for genetic variation associated with disease. As the public has become more aware of the impact of genetics on health and disease, there has been a growing desire for preemptive screening, particularly for individuals with family members afflicted with common disease. In response to this need, genetic variation screening tests are often marketed as a means of assessing the potential for future development of disease. Given the caveats discussed, there is no definitive means at present to accurately define an individual’s risk of disease based on the presence of one or more associated common variants. It is important for the physician to be aware of this insofar as patients may contact them regarding such testing, and it should be emphasized that any positive results would have unclear predictive value.

Perhaps even more vexing is the fact that most genetic risk factors identified in GWASs are of very small effect sizes, ranging from a relative risk of 1.1 to 1.3, so the clinical consequences of having such a single variant in an individual patient is likely minor. The hope is that by combining many such variants, a clinically relevant risk profile can be developed.

There are examples of clinically important allelic variants identified by other methods, so such expectations for GWAS in neurological disease are not unfounded. One such illustration is the variation seen in the cytochrome P450 isoenzyme, CYP2C9, which is responsible for the metabolism of a number of clinically relevant pharmaceutical agents, in particular the anticoagulant, warfarin (Sanderson et al., 2005). The major allele, CYP2C9*1, is seen in more than 95% of Asian and African populations, but multiple variants commonly exist in European and Caucasian populations, including CYP2C9*2 and CYP2C9*3, both of which reduce warfarin metabolism (Sanderson et al., 2005). In one study, 20% of patients carried either CYP2C9*2 or CYP2C9*3 and required a mean reduction of their warfarin dosage by 27% to maintain an optimal therapeutic range, reflected by an increased relative risk of bleeding of about 2.3 (Sanderson et al., 2005). Although the relative risk in this example is still greater than typically seen in most GWASs, it demonstrates how common variant risk information can potentially affect the care of an individual patient. As we discover more regarding the nature of complex genetic disease, new ways of utilizing this information clinically will likely be determined. In the meantime, the value of GWAS data, especially from a pharmacogenomic research perspective, is significant; it can help identify new genes, pathways, and biological networks related to disease that may have therapeutic benefit (Box 40.2).

Rare Variants and Candidate Gene Resequencing

So far, common variation is only able to explain a small percentage of genetic risk for common neurological disease. The other major model that attempts to explain what is currently referred to as the missing heritability (Manolio et al., 2009) in complex genetic disease implicates rare variants with medium to high penetrance instead of more common ones with low penetrance (Schork et al., 2009) (Fig. 40.13). Rare variants are defined as DNA alterations that are found in less than 1% of most populations or, in some cases, are “private” and only seen in specific affected families. In this model, one or more rare variants, alone or in combination with common variants, produce the disease in question. A GWAS is not well suited to detect these variants, because they are rare and most likely to be relatively recent mutations that do not segregate on common haplotypes measured in these studies. Even when they do, they do not occur in high enough frequency in the general population to provide statistical power for their detection using current sample sizes. Detection generally requires resequencing of potentially involved candidate genes in a defined population of patients and controls. One major difficulty of such investigations is that the baseline level of rare variation among normal humans is not clearly established. Studies such as the 1000 Genomes Project (http://www.1000genomes.org/) are attempting to catalog normal human variation within the 0.1% to 1% range, so researchers will be able to better define this class of rare variants and develop more effective strategies for their detection.

Fig. 40.13 Models of causal variants in complex disease. In the common disease–common variant model, risk of disease is imparted by the presence of one or more gene variants present in 5% or more of the population (red). Such variants are amendable to detection by genome-wide association studies (GWAS). Conversely, in the common disease–rare variant model, disease is caused by rare genetic variants present in less than 1% of the population or only in specific families (various colors). Such variants would not be amendable to detection by GWAS, since they would not be represented in large enough numbers to generate statistical significance. Note that both models are not mutually exclusive, and both may contribute to common disease.

An example of this approach involves the developmental disorder, autism, where sequencing of the gene, contactin-associated protein-like 2 (CNTNAP2), in 635 patients with autism spectrum disorder (ASD) and 942 controls found 13 rare variants unique to patients, including one which was seen in 4 patients from 3 unrelated families (Bakkaloglu et al., 2008). Recessively inherited mutations in CNTNAP2 in an Amish family with a syndromic form of autism with epilepsy provided the most convincing evidence for the causal role of mutations in this gene (Strauss et al., 2006). Interestingly, this same gene illustrates that the common disease–rare variant and common disease–common variant hypotheses are not mutually exclusive, since common variants in this gene modulate language function in ASD and other conditions (Alarcon et al., 2008; Vernes et al., 2008). Exciting advances in DNA sequencing (see Whole Genome/Exome Sequencing in Disease Gene Discovery) will allow us to finally analyze many whole genomes and understand to what extent common and/or rare variants contribute to many common neurological diseases.

Copy Number Variation and Comparative Genomic Hybridization

The majority of variation and disease-causing mutations discussed to this point have centered around single base pair changes in DNA sequence. However, as previously described in Structural Chromosomal Abnormalities and Copy Number Variation, the CNV (Beckmann et al., 2007; Stankiewicz and Lupski, 2010; Wain et al., 2009; Zhang et al., 2009) (Fig. 40.14) actually represents more total real estate in our genome. Advances in methods such as the advent of the microarray indicate that such changes occur quite commonly (at 10⁻⁴ to 10⁻⁶ per locus per generation) compared to single nucleotide changes (10⁻⁸ per base pair per generation on average) (Lupski, 2007). Overall, CNVs are estimated to represent at least 4% (Conrad et al., 2010) and potentially up to 13% of the total human genome (Redon et al., 2006; Stankiewicz and Lupski, 2010). The high frequency of these events may reflect an evolutionary advantage of CNVs as a mechanism for producing genetic diversity (Zhang et al., 2009) but also implies that clinically relevant CNVs are quite likely to occur de novo more frequently than point mutations (Table 40.6). CNVs can potentially cause disease in numerous ways, including disruption of a gene’s coding region (which could cause a dominant effect or release a recessive effect on the homologous allele) or by altering regulated gene expression via positive or negative dosage effects. If the CNV itself results in the disease phenotype, it could be transmitted as a Mendelian disorder, as is the case for Charcot-Marie-Tooth type 1A. Such CNVs may be examples of rare variants in the common disease model. Alternatively, their contribution may be more subtle and insidious, with low penetrance and variable expressivity contributing to the risk of a complex genetic disease, such as in autism (Bucan et al., 2009).

Fig. 40.14 Copy number variation (CNV). A, Copy number variation can be detected via comparative genomic hybridization or chromosomal microarray analysis, shown here. In this example, patient genomic DNA and an equal amount of control DNA is hybridized to a microarray platform containing representative probes spanning the genome at a specified resolution, usually at the kilobase level. In the illustration, patient DNA is fluorescently labeled green, and control DNA is labeled red. Following hybridization, regions present in equal amounts are yellow, whereas regions duplicated in the patient are green, and deletions are red. In this example, the patient possesses two CNVs, a duplication on chromosome 7 (illustrated by the increased green signal at that locus on the array) and a deletion on chromosome 16 (with corresponding increased red signal at the locus). The patient also has Turner syndrome (monosomy X) reflected by the increased red signal across the entire chromosome. Chromosome 10 is shown as an example of a chromosome that does not differ between the samples (yellow). B, Introduction of CNV by the nonallelic homologous recombination (NAHR) mechanism. NAHR occurs when genomic instability is introduced by the presence of low copy repeat (LCR) regions greater than 1 kilobase in size with more than 90% homology. Pairing of nearby regions during DNA replication can lead to deletions, duplications, or inversions as illustrated. C, Introduction of CNVs by the fork stalling and template switching (FoSTeS) mechanism. FoSTeS occurs when replication on the lagging strand stalls during DNA replication and resumes at an adjacent replication fork. The structural variation introduced depends on whether the reinitiation occurs upstream or downstream of the original fork and whether it occurs on the lagging or leading strand. Examples of how deletions, duplications, or inversions might result are shown (orange arrows). Furthermore, if more than one FoSTeS event occurs (purple arrow), a complex structural rearrangement could result.

CNVs can be detected via essentially the same microarray technology used to detect SNPs, with only a few minor adjustments. In this case, DNA probes corresponding to specific chromosomal regions are placed on an array and hybridized with differentially fluorescent-labeled genomic DNA from the individual being studied and from a reference genomic DNA sample, a technique termed array comparative genomic hybridization (CGH) (also called chromosomal microarray analysis) (see Fig. 40.14, A). The average ratio of fluorescence is normalized across the array and then evaluated for each probe. If both samples hybridize to a given probe equally, the corresponding DNA region is present equally in both samples. However, if the DNA sample being studied hybridizes more or less intensely than the reference sample, it must contain either more or less of the chromosomal region in question, thus indicating a copy number variation at that location. The minimum size of a CNV that can be detected by this method is limited to the genomic distance between the minimum number of probes needed to observe a statistically significant signal change, but is usually on the order of kilobases for the highest resolution arrays. The same microarrays used to genotype SNPs in GWASs may also be used to detect CNVs, incorporating both intensity and inheritance data. Array CGH essentially produces a molecular karyotype capable of detecting genomic structural changes with much finer detail than routine microscopic methods. In most major diagnostic labs, this method has replaced microscopic karyotyping and FISH, the latter of which is now used for confirmation.

Some examples of clinically relevant copy number variations are seen in Mendelian disorders including adult-onset autosomal dominant leukodystrophy (autosomal dominant, caused by duplication of the lamin B1 gene on chromosome 5), Charcot-Marie-Tooth type 1A (autosomal dominant, most frequently caused by duplication of the peripheral myelin protein 22 on chromosome 17), hereditary liability to pressure palsies (autosomal dominant, most commonly due to deletion of the peripheral myelin protein 22 on chromosome 17), spastic paraplegia type 4 (autosomal dominant, occasionally caused by deletion of the spastin gene on chromosome 2), juvenile PD 2 (autosomal recessive, occasionally caused by deletions or duplications in parkin on chromosome 6), and Williams syndrome (autosomal dominant, caused by deletion of several contiguous genes on chromosome 7).

CNVs are also particularly important for neurodevelopmental disorders, with de novo CNVs present in more than 5% of patients with intellectual disability (ID) (Koolen et al., 2009) or ASD (Bucan et al., 2009; Marshall et al., 2008; Pinto et al., 2010; Sebat et al., 2007). Based on these findings, array CGH is now clinically indicated in children with a wide range of neurodevelopmental disabilities including ID and ASD (Miller DT et al., 2010). These studies also revealed several potential new autism candidate genes as well as novel biological pathways for future study of disease pathogenesis (Bucan et al., 2009; Pinto et al., 2010). Remarkably, de novo CNVs are also associated with schizophrenia (Stefansson et al., 2008; Walsh et al., 2008), especially childhood-onset forms, and some of the same CNVs observed in ASD are also observed in schizophrenia (Cantor and Geschwind, 2008), suggesting some shared liability between what were previously considered clinically distinct conditions.

Evaluation and Diagnosis

Evaluation and diagnosis benefit from the arsenal of genetic testing available for single gene disorders and for genomic variation. Many commercial laboratories offer testing for Mendelian disease genes, and in some settings, genetic testing has become as routine as other common blood tests. However, because genetic testing carries additional implications for a patient and their family, particularly with regard to heritability of disease, it is important that it be used appropriately and that patients be fully educated prior to such testing. Important points to consider for genetic testing are summarized in Table 40.8. Although how the testing is incorporated into a clinical evaluation strategy will vary by disease, a general principle is that most genetic disease is diagnosed clinically via a thorough history (including family history) and physical examination. A complete evaluation for nongenetic causes should be performed as appropriate prior to any genetic testing so that possible treatments can be initiated in a timely manner. Genetic testing should only be used to confirm a clinical suspicion, not for screening purposes, because currently this is low yield and not cost-effective in the majority of cases. Specialist referral to a tertiary center is appropriate for all cases where a diagnostically useful clinical phenotype cannot be established. Genetic counseling (see later) should be provided, either by a physician or a licensed genetic counselor, prior to testing to ensure that patients understand the nature of the test and the possible results. When testing is ordered, it should be based on phenotype and supported by mode of inheritance if this can be determined. Testing of an asymptomatic minor is never indicated for a genetic disease where there is no treatment or cure. Knowledge of the disease status without chance for treatment may have many negative consequences.

Table 40.8 The Neurogenetic Evaluation and the Clinical Utilization of Genetic Testing

Many companies now offer broad genetic panels based on general phenotypes or modes of inheritance for a particular symptom, which have appeal because they are simple to order and often advertised as a molecular means of differentiating between overlapping phenotypes. Unfortunately this does a disservice to the patient, since these panels can be quite costly (up to $15,000 or more) and despite being billed as complete, often test disorders with such diverse phenotypes as to make it impossible to consider both in the same individual, or they test genes so rare that only a few families are even known to possess them. The clinical examination should be used to precisely define the patient’s phenotype, which will in turn suggest the most high-yield conditions for genetic testing. This systematic approach is of immense benefit in resource management.

The types of single-gene testing available vary per laboratory and gene (Table 40.9). The most comprehensive (and expensive) testing type is full gene sequencing, where all coding regions, as well as approximately 50 bases in each intron/exon junction, are sequenced for the presence of mutation. This will detect all coding point mutations and splice-site mutations as well as small insertions and deletions but will miss more detailed structural variation. Importantly, novel coding mutations can be detected in this way. Targeted sequence analysis (also called select exon testing) consists of specific sequencing reactions designed to only detect one or a few previously identified mutations. This will not detect any sequence variations outside of the limited region of the gene being searched. For repeat disorders, there are specific tests to identify the relevant expansions using either polymerase chain reaction (PCR) or Southern blotting, a hybridization-based DNA sizing technique. Larger deletions or duplications (e.g., copy number variations) can be detected by quantitative PCR methods or by comparative genomic hybridization. It is important to be aware of the type of testing being ordered; in some cases, such as select exon testing, a negative result does not exclude mutations elsewhere in the gene being tested. Interpretation of these genetic results may be straightforward, for example, if no mutations are present or if known pathogenic changes are found. In contrast, interpretation may be complicated if novel sequence variants of unknown pathological significance are identified. Inconclusive results may require interpretation by a specialist and/or further testing to determine the likelihood of pathogenicity.

Table 40.9 Types of Genetic Testing

* Includes missense, nonsense, and silent mutations.

† Includes only those involving splice sites and exonic splicing regulatory sequences.

‡ Includes promoter mutations and noncoding splicing regulatory elements.

§ Arbitrarily defined here as any deletion/duplication/insertion larger than detectable by Sanger sequencing.

‖ Targeted mutation analysis using either polymerase chain reaction (PCR) and/or Southern blot is preferred, as sequencing may be inaccurate due to the large size of many repeat regions.

¶ Only detectable if sequencing is performed as opposed to individual mutation-specific detection methods.

# Size and number of region(s) targeted varies per individual test.

** Genome-wide testing method.

†† Minimum size of CNVs detected and density of genomic coverage varies per test.

‡‡ Not yet available clinically.

Common diseases must be approached in a different manner, because detailed phenotype alone cannot always predict the mutation to test for, particularly when assessing genomic variation. Still, the goal remains to develop strategies incorporating known genetic information into a systematic protocol designed to maximize diagnostic capability while minimizing cost and unnecessary testing (Lintas and Persico, 2009). Tests such as chromosomal microarray analysis are clinically available to search genome-wide for disease-causing CNVs and are recommended for sporadic causes in disorders such as intellectual disability or autism where CNVs have been found responsible for a reasonable percentage of disease (Geschwind and Spence, 2008; Miller DT et al., 2010). Use of such testing in sporadic adult-onset disease is less clear, so the physician is advised to refer to current published guidelines for the disease in question before ordering. For more specific phenotypes, other available tests include those assessing for CNVs (often called simply deletions/duplications) involving individual genes or specific chromosomal regions. Overall, interpretation of CNV results can be challenging, particularly if the CNV was previously unreported. Here, the parents will often need to be evaluated to determine whether the CNV in question is inherited or de novo. As already discussed for DNA sequence changes, such findings may require interpretation by a specialist and/or further testing to determine the likelihood of pathogenicity.

Whole genome and/or whole exome sequencing are not yet routinely available in the clinic but are predicted to arrive within the next 5 years or sooner. Whole genome sequencing is the more comprehensive of the two and capable of detecting more types of mutation, as well as structural variation, but its use will hinge on the development of accurate and efficient bioinformatic techniques for translating the expected massive genomic variation per patient (millions of SNPs and hundreds of CNVs across the whole genome) into clinically meaningful results. How such a pipeline would operate has not yet been established, but we expect that the cost should be equivalent to that of an MRI study within 5 years. Incorporation of such testing into a clinical evaluation will also depend on other elements such as cost of testing and time of analysis, but these factors are not expected to vary much from methods of genetic testing currently in use.

Genetic Counseling

Establishing a precise genetic diagnosis will definitively establish the means of inheritance of a disorder and is extremely useful in genetic counseling and family planning, particularly for disorders that show incomplete penetrance. However, unlike other tests typically ordered by physicians, a positive diagnosis carries implications not only for individual patients but for the entire family. Genetic counseling, therefore, should be provided in all cases where genetic testing is recommended, either by an experienced neurologist, a geneticist, or a licensed genetic counselor. Follow-up counseling should also be provided to all patients with a positive test result and, in many cases, offered to other family members who may be at risk for disease or as carriers. Physicians must be aware of the various ethical implications involved in such testing (Ensenauer et al., 2005). One area of particular importance in this regard involves considerations of genetic testing in asymptomatic individuals, especially minors. This stems in part from concerns that have been raised regarding risks of depression and suicide in asymptomatic individuals diagnosed with fatal genetic disease, although this is not well established, and further study will be important for determining best practices. For minors, standard practice dictates that unless there is disease-modifying therapy available for them, they should not be tested if asymptomatic until they reach an age to consent to such testing and are properly counseled as to the implications. Counseling regarding prenatal testing and assisted reproduction are other topics of relevance to patients of reproductive age. Current reproductive medicine techniques such as in vitro fertilization and preimplantation genetic testing, by assuring that offspring will not harbor the mutation in question, can aid couples concerned about the risk for passing on inherited conditions. Other ethical considerations may also apply, depending on the disease and specific family/patient circumstances.

Prognosis and Treatment

A confirmed genetic diagnosis can contribute clinically useful data concerning patient prognosis, as it allows information from published case studies to be utilized in the care of an individual patient. This can aid in the identification of specific clinical features to focus on for surveillance in the development of a particular genetic disorder, such as cognitive decline in a patient with isolated chorea found to have HD or cardiac testing in an autistic patient with chromosome 15q duplication. A genetic diagnosis may also alert the clinician to potential life-threatening comorbidities such as adrenal insufficiency in X-linked adrenoleukodystrophy or cardiomyopathy in Friedreich ataxia. Review of case studies in a particular disorder may help answer questions regarding life expectancy or future disability, such as years of disease prior to loss of ambulation in the various SCAs Lastly, there are important positive psychological aspects to establishing a definitive diagnosis, particularly for patients who have undergone many fruitless clinical evaluations.

Although the majority of genetic diseases are not curable, therapies do exist for many of them. Defining the genetic etiology of a patient’s disease allows for utilization of the published literature on symptomatic treatments and pharmacotherapy that may benefit a specific condition. Phenylketonuria is an excellent example of this, since dietary restriction of phenylalanine initiated soon after birth will prevent cognitive impairment and enable virtually normal development (Burgard et al., 1999). More importantly, new clinical trials are being developed frequently and can be offered to patients with an established diagnosis. Many disease-based patient registries exist to facilitate this.

The ultimate goal of translational neuroscience is to utilize advances in our understanding of disease at the molecular level to aid in the treatment of patients in the clinic. Recent new treatments, which take advantage of the molecular aspects of these disorders, show promise in the clinic and the laboratory. Such treatments include enzyme replacement therapy for metabolic disorders such as the severe fatal glycogen storage disorder Pompe disease, where use of recombinant acid α-glucosidase in 18 infants prior to 6 months of age enabled all to live to the age of 18 months, a 99% reduction in death, as well as reduced their risk of death or invasive ventilation by 92% compared to historical controls (Kishnani et al., 2007). Work in animal models has suggested potential new pharmacological treatments, such as a recent research study which demonstrated that the use of histone-deacetylase inhibitors can unsilence expanded frataxin alleles in a Friedreich ataxia mouse model, restoring wild-type gene expression levels and reversing cellular transcription changes associated with frataxin deficiency (Rai et al., 2008). Targeted molecules have been designed to correct specific disease-causing biological defects, as shown by recent work where antisense oligonucleotides were used to block mutations that promote splicing defects in the ataxia-telangiectasia mutated (ATM) gene in cell lines from patients with ataxia-telangiectasia, leading to restoration of functional protein (Du et al., 2007). Such newer techniques may markedly exceed the therapeutic benefit of current options, such as in Duchenne muscular dystrophy where patients can expect only moderate short-term benefit (up to 2 years) from the gold standard, glucocorticosteroid treatment (Manzur et al., 2008; Wood et al., 2010). Newer molecular strategies such as dystrophin splice-modulation, which promotes exon skipping via antisense oligonucleotides to bypass point mutations or frameshifts, may potentially resolve the primary defect and has shown promising results in early clinical trials (Wood et al., 2010). Novel treatments aimed at genetic modification of disease are also in development, as was seen in a recent study where investigators used RNA interference techniques to specifically degrade and thus silence the disease allele in a rat model of SCA type 3, resulting in a reduction in neuropathological changes in the brain (Alves et al., 2008). New technologies such as next-generation sequencing and the use of systems biology approaches to disease are expected to lead to additional new innovations. With these advances, the future of clinical neurogenetics is full of promise and stands poised to answer the challenge stated most eloquently by Bernard Baruch (1870-1965): “There are no such things as incurables; there are only things for which [medicine] has not found a cure.”