Talking genes the molecular basis of language impairment

Transcription

1 the molecular basis of language impairment Dianne F Newbury and Anthony P Monaco University of Oxford, UK Many children acquire language so smoothly that it appears to be an innate ability. If this is true, then it should be possible to identify genes that underlie variations in linguistic abilities. For most children the process of understanding and using language seems to occur with little effort and minimal instruction. The typical five-year-old has a vocabulary of words a feat that requires an average lexical growth of nine words per day. Furthermore, in parallel to the completion of this task, they are also required to master other, more complex, aspects of language (e.g., grammar, pragmatics). The ease with which most children fulfil these expectations has led to the proposal that language is an innate instinct, enabling the acquisition of language in the absence of explicit tuition, and must, at least in part, be genetically determined (Pinker, 1994). Specific Language Impairment (SLI) For some children, the acquisition of language can prove to be a much more difficult task. In many cases these difficulties are attributed to additional developmental or medical problems (e.g., cleft lip, autism, cerebral palsy). However, there remain a small, but significant, number of children who appear to be developing normally in all other respects, but yet encounter an inexplicable disorder in the acquisition of language. These children are said to have Specific Language Impairment or SLI. Complexities of the SLI phenotype When presented in this manner the definition of SLI appears relatively straightforward. However, a diagnosis is often applied in terms of exclusionary criteria rather than by a defined set of distinct clinical characteristics (see Table 1). A diagnosis implies that a child s language skills are disproportionately worse than their non-verbal abilities and that they are not affected by any additional disorders that might affect language development (e.g., deafness, autism). However, the exact nature of the language problem often shows great variation from child to child with differences both in the range of linguistic domains involved (phonology, morphology, syntax, semantics and pragmatics), and in the language modality affected (expressive or receptive). Furthermore, affected individuals may present with associated developmental delays (e.g., dyslexia and attention deficit hyperactivity disorder), and additional non-verbal deficits (e.g., in coordination) are not uncommon. Although it has been proposed that the language impairments may underlie many of these additional problems, the exact relationship between the factors remains unclear, and their importance in diagnosis continues to be a matter of debate. Genetics of SLI Lenneberg (1967) was perhaps the first to document the familial nature of language impairments, an observation that has since been consistently endorsed by many independent studies, suggesting that SLI may be genetic in nature. First-degree relatives of language-impaired individuals are at an increased risk of developing SLI them- Biologist (2002) 49 (6) 1

2 Table 1. Current diagnostic research criteria for Specific Language Impairment (SLI) ICD-10 Research Diagnostic Criteria for Specific Developmental Disorders of Speech and Language (World Health Organisation, 1993): 1) language skills that fall outside the two standard deviation limit for the child's age (although this may not be the case in older subjects); 2) the language delay should not be directly attributable to neurological or speech mechanism abnormalities, sensory impairments, mental retardation or environmental factors; 3) the language delay should not be directly attributable to hearing loss in childhood. Cases of partial hearing loss may be included if the hearing loss is considered to be a complicating factor but not a sufficient direct cause for the language delay; 4) the language delay should not form part of a pervasive mental retardation or global developmental delay; 5) the language delay may be accompanied by associated problems and is often followed by difficulties in reading and spelling, abnormalities in interpersonal relationships, and emotional and behavioural disorders. DSM-IV Criteria for Mixed Expressive-Receptive Language Disorder (American Psychiatric Association, 1994): 1) receptive and expressive language skills that are substantially below those obtained from standardised measures of nonverbal intellectual capacity; 2) language difficulties interfere with academic or occupational achievement or with social communication; 3) the language delay should not be directly attributable to mental retardation, a speech-motor sensory deficit, or environmental deprivation; 4) the language delay should not form part of a pervasive mental retardation; 5) the language delay may co-occur with attention-deficit/ hyperactivity disorder, developmental coordination disorder and enuresis. NOTE: Both the ICD-10 and DSM-IV make a distinction between expressive language disorder (in which a child's ability to use expressive spoken language is delayed, but language comprehension is within the normal limits) and receptive language disorder (in which the child's understanding of language is below the appropriate level for their mental age). Figure 1. The genetics of simple and complex disorders A. Simple genetic disorders ñ e.g., the KE family Father's pair of chromosome 7 Mother's pair of chromosome 7 Affected Unaffected B. Complex genetic disorders ñ e.g., SLI Mother Father Possible combinations of parental chromosome 7 in children XX } } Affected Unaffected Figure 1. The genetics of simple and complex disorders A. Simple Monogenic Disorders e.g., the KE family. Monogenic disorders are caused by a mutation in a single gene (which lies on chromosome 7 in the above diagram). Each individual carries two variants (or alleles) of each gene, one on each chromosome 7. Children may inherit a normal allele (yellow on diagram above) or a mutated allele (red on diagram above). Children who inherit a mutated allele always develop the disorder. For monogenic disorders, it is possible to predict the probability that two individuals will have an affected child. The couple shown above have a 50% chance of having an affected child. B. Complex Genetic Disorders e.g., SLI. The above diagram represents a hypothetical model of a complex disorder (such as SLI), which is caused by interactions between several genes and the environment. In the hypothetical situation depicted above, it is assumed that there are eight genes (i.e., 16 alleles) that interact to produce a risk factor for SLI (Red lines = alleles which make you susceptible to SLI, yellow lines = alleles which have no impact upon language status). If the hypothetical model assumes that, out of the 16 alleles, an individual needs to inherit more than four risk alleles to have a high risk of developing SLI. Then in the above diagram, the mother will probably have language problems because she has seven susceptibility variants whereas the father is unlikely to develop SLI because he only has three susceptibility alleles. Because of the number of genes involved, it is difficult to predict the probability that this couple will have a child with SLI. In the above hypothetical model any child with more than four risk alleles will have a high chance of developing the disorder (whether they inherit the risk alleles on chromosome 1, 3, 6, 11, 12, 16 or 19). XY selves, and often report a history of other related disorders such as dyslexia and hyperactivity. Moreover, investigations of twin pairs affected by SLI indicate that monozygotic twins, who are genetically identical, are more similar in terms of their language profile than dizygotic (non-identical) twins, who share only half their genes. These investigations provide strong evidence for the role of genes in the aetiology of SLI (Bishop, 2001). However, for generalised cases of language impairment (as opposed to specific impairments in particular components of language e.g., Van der Lely and Stollwerck, 1996), no clear inheritance pattern has ever been observed. It is therefore considered likely that the onset of these forms of SLI involves several genes that interact, both with each other and the linguistic environment, to produce an overall susceptibility to the development of the disorder (i.e., a complex disease see Figure 1). The KE family Historically, there is only one case of a speech and language disorder for which the above statement has been proved to be untrue. Since the early 1990s, scientists interested in language impairment have been studying a special three-generation pedigree known as the KE family (see Figure 2). 2 Biologist (2002) 49 (6)

3 Figure 2. The KE pedigree Squares represent males. Circles represent females. Black symbols represent individuals affected by a severe speech and language disorder. Symbols with lines through them represent those individuals are deceased. This family are affected by a severe form of a speech and language impairment that is not inherited like a complex disease, but instead involves a change (or mutation) within a single gene (i.e., a monogenic disorder see figure 1). The speech and language impairment in the KE family affects similar numbers of males and females, indicating that the gene involved does not lie on a sex chromosome (X or Y). It affects approximately 50% of the family members, indicating that it follows a dominant pattern of inheritance (i.e., a single mutated copy of the gene is sufficient to cause the disorder). Furthermore, there is no evidence to suggest the presence of complicating factors such as decreased penetrance (i.e., when individuals with the mutation do not develop the disorder) and phenocopies (i.e., when individuals with the disorder do not possess the mutation). One or both of these are generally expected to be widespread in more common and genetically complex forms of SLI. The KE phenotype It should be noted that many researchers consider certain aspects of the KE phenotype to be atypical of specific language impairment and would therefore exclude them from a strict diagnosis of SLI. Affected individuals present with a severe orofacial dyspraxia, which impedes both verbal and non-verbal articulatory movement, and thus might be predicted to directly underlie, or at least contribute to, their speech impediment. However, studies of mute children and children affected by cerebral palsy have indicated that even a complete absence of expressive linguistic abilities does not necessarily result in an impairment in the receptive domain. Thus, since the KE deficits are not limited to the expressive domain, we might expect the aetiology to be more complex than that of a speech impairment caused by orofacial dyspraxia. An additional problem is found when one considers the non-verbal abilities of the KE family. A number of affected members have non-verbal IQ scores that are low enough to place them outside of the normal range. Since these deficits preclude the existence of a discrepancy between verbal and non-verbal abilities some researchers argue that the KE language problems may form part of more-general learning difficulties. However, it has been found that these non-verbal difficulties do not co-segregate with the phenotype (i.e., some affected individuals have normal non-verbal IQs and some unaffected individuals have non-verbal IQs outside the normal range). Therefore, no clear association can be made between the deficit and the speech and language impairment. Thus, although the affected members of the KE family members clearly have speech and language difficulties, it remains a matter of debate whether these difficulties have a similar, or different, aetiology to those found in SLI. Linkage analysis For simple monogenic disorders, genes are traditionally mapped using the positional cloning approach. This Figure 3. Recombination and linkage A given chromosome from child 1 Biologist (2002) 49 (6) 3 a b c The corresponding chromosome from child 2 (sibling of child 1) Figure 3. Recombination and linkage The above diagram shows a comparison of a single chromosome between two siblings. The parent carried one light blue and one dark blue chromosome, and recombination between these chromosomes resulted in the mosaic patterns seen in each of the siblings (see text). Because DNA is inherited in these mosaic blocks, if it is known that the two siblings share the same DNA at point a then it is likely that they will also be sharing at point b. However, we are unable to infer anything about the sharing at point c from the information at a because it is likely that a recombination will have occurred between these two points.

4 Figure 4. Linkage analysis using microsatellite markers Microsatellite markers are amplified, tagged with fluorescent labels and separated on polyacrylamide gels, according to size. In the gel above each lane represents a single individual. The blue, green and yellow markers represent different microsatellites within each individual. The use of three different colours allows the running of similar sized microsatellites upon the same gel. The differences in sizes between individuals allows us to follow the way in which chromosomal segments are inherited within families. The red bands are a size standard, which allow the sizing of each microsatellite. process basically involves the reconstruction of inheritance patterns for the entire genome within families affected by the disorder, thus allowing the identification of regions that are passed onto affected children more often than would be expected by chance alone. Such an approach allows the mapping of disease genes to specific chromosomes when little, or nothing, is known about the biochemical basis of the disorder and is very powerful for simple disease, when the pattern of inheritance is clear. Although positional cloning theoretically involves scanning of the entire genome, it does not require the sequencing of every base or the study of every gene. Instead it exploits the fact that DNA is inherited as mosaic blocks from parents to offspring. During the production of gametes, chromosomes become entangled with each other and can cross over swapping parts of genetic material between chromosome pairs (i.e., recombination). Recombination is a highly random process and therefore, even two cells that have inherited the same array of parental chromosomes are likely to differ. As a consequence of the recombination process, the closer together that two genes are on a chromosome, the more likely that they will be inherited within the same block unit of DNA (see Figure 3). Thus by sampling regions of DNA along each chromosome, it is possible to infer the origin of the majority of the genome without the need to sequence every base. This process is known as a genome screen. A genome screen typically involves the genotyping of hundreds of microsatellite markers spread evenly across the genome (Figure 4). These are selectively neutral, noncoding regions of DNA that consist of repeat stretches of sequence (e.g., CACACACA). The repetitive nature of these sequences means that they show great variability between individuals. For example, one individual may have five repeat units (CACACACACA) where another may have nine (CACACACACACACA- CACA). Using microsatellite information it is possible to identify regions where affected children are more similar at the DNA level than would be expected by chance alone. These are the regions where genes involved in the disorder are likely to lie. The probability that a given region is involved in a given disorder is measured by the logarithmic odds ratio (LOD score). This is the probability of the data arising if the given region is linked to a disorder, against the probability of the data arising if the given region is not linked to a disorder. As a general rule, any region with a LOD of above 3.0 (i.e., a ratio of 1000:1 in favour of linkage) is considered worthy of further investigation. In some cases, these regions may contain many genes, all of which need to be studied in detail to allow the evaluation of their involvement in the disorder under study. In cases where the biochemical basis of the disorder is known, it is possible to prioritise the study of genes, first investigating those which represent the best candidates (e.g., for language, a good candidate would be expressed in the brain during development). FOXP2 In October 2001, scientists in Oxford used a positional cloning approach to isolate the gene mutated in the KE family (Lai et al., 2001). The gene is found on chromosome 7 and is called FOXP2. This is a member of a large family of genes, all of which possess a conserved sequence known as a winged-helix or forkhead box (fox) domain. There are over 100 fox genes, all of which act as transcription factors, i.e., they are responsible for regulating the expression of other genes. FOXP2 has been shown to be expressed in many cell types but has a particularly strong expression in foetal brain. The KE mutation involves a change at a single DNA nucleotide from guanine (G) to adenine (A), which co-segregates with the speech and language phenotype (i.e., all affected individuals carry the mutation, but no unaffected individuals have the change). This single base change results in an alteration within the protein amino acid sequence (arginine to histidine). Although such a small change may not always be expected to alter the function of a protein, Lai et al. showed that this particular base falls within the fox domain and is conserved in the sequence of all FOX genes, not only in humans but also across species such as mice, fruit flies and yeast, indicating that it has an important function. Lai et al. provided further evidence for the involvement of FOXP2 in speech and language impairments by identifying another patient (known as CS) who is unrelated to the KE family, but has a similar deficit and carries a disruption in the FOXP2 gene. The CS patient did not have 4 Biologist (2002) 49 (6)

5 a mutation in the FOXP2 gene but instead had a translocation (i.e., a transfer of DNA between two chromosomes) involving chromosomes 5 and 7, in which the chromosome 7 breakpoint fell directly within the FOXP2 sequence and thereby disrupts the gene. FOXP2 in generalised SLI The evidence for the involvement of FOXP2 in these two cases of language impairment is clear. However, given the dissent regarding the KE phenotype, questions remained regarding its relevance in more common and genetically complex forms of SLI. A recent study screened the entire FOXP2 coding sequence within 43 individuals with SLI and 48 individuals with autism (which often presents with associated language delays; Newbury et al. 2002). All the variants found within these individuals were in non-coding sequence and were found to occur at equal frequencies within non-language-impaired controls. Thus, it appears that the role of FOXP2 may not generalise to SLI, or may only be relevant to limited numbers of individuals with a specific subtype of speech and language impairment. However, it remains possible that the characterisation of the FOXP2 pathway may reveal alternative upstream or downstream genes that do play a role in more common forms of SLI. Furthermore, a recent investigation indicated that the sequence of the FOXP2 gene is likely to extend beyond that originally characterised (Bruce & Margolis, 2002), and therefore additional mutation screens will be required before the gene can be fully excluded as a candidate in SLI. Complex SLI In contrast to the KE family, the study of genetically complex SLI continues to pose more of a challenge. Because these forms of language impairment are expected to involve several genes, the effect of each gene variant is reduced over that in a single-gene disorder and the individual genes are therefore harder to detect. Moreover, complex diseases are not necessarily caused by mutations in genes but may arise as a result of common genetic variations that even perfectly healthy people carry. The more of these susceptibility variants an individual carries the greater their risk of developing SLI. Thus, because the gene variations do not directly cause the language problems, the exact combination of genes underlying the disorder may vary from person to person (i.e., SLI is a heterogeneous disorder). Whilst the positional cloning approach can be applied to the study of complex diseases, heterogeneity necessitates larger sample sizes and often results in a loss of power, meaning that less clearly defined chromosomal regions may be identified, which tend to be much larger in size. Because of these complications the significance threshold for genome screens of complex diseases differs from that for a monogenic disorder. For complex disease, a LOD of above 2.2 is considered to be suggestive of linkage and a LOD of above 3.6 is considered a significant linkage. Genome screens for loci involved in SLI Recently, two studies have used the genome screen method to investigate which chromosomal regions might be involved in complex forms of SLI (SLIC, 2002; Bartlett et al., 2002). The SLI Consortium collected DNA from 98 families, all of which contained at least one individual affected by SLI. These samples were taken from two sources: clinical samples of severely affected children were collected from special language schools in the London area and more mainstream samples were taken from an epidemiological study in Cambridge. Using these cohorts, SLIC researchers performed a systematic genome screen using 500 microsatellite markers spaced evenly across the genome (approximately one every eight million base-pairs (Mb)). They investigated three language-related measures within these families: a composite score of expressive language abilities, a composite score of receptive language abilities and a putative measure of phonological short-term memory (which has been proposed to be a core deficit in the SLI phenotype). They identified a region on the long arm of chromosome 16 that was linked to the short-term memory score with a LOD of 3.55, and an area on the long arm of chromosome 19 linked to the expressive language score, also with a LOD of Bartlett et al. studied 86 individuals selected from branches of five large Canadian families. In contrast to the SLIC study, this investigation used a binary system under which family members were classified as affected or unaffected according to three alternative diagnostic schedules (language impairment, reading impairment and clinical impairment). Using this approach, two regions of linkage were found, one on chromosome 13q (Maximum LOD Score = 3.92) under a recessive model of reading impairment and a second on chromosome 2p under a recessive model of language impairment (Maximum LOD Score = 2.79). Note that neither of the above studies found evidence for linkage to chromosome 7 the site of the FOXP2 gene. The apparent dissociation between the results of these two genome screens is not unusual in the study of complex disorders and may have arisen for several reasons. Each study employed very different methods with respect to the selection of subjects, the measurement of language impairment and the detection of linkage within their samples. The divergence between the methods used and the results obtained further highlights the challenges involved in the genetic investigation of a complex disorder such as SLI. It should also be noted that the genome screen approach only highlights regions that are likely to be involved in the disorder under study and, given the complexity of SLI, it remains possible that these results may represent false positives. Thus replication within other groups of language-impaired children will be required to validate each of these regions of linkage. Furthermore, it is likely that all of the areas identified by the above studies contain hundreds of genes, each of which will need to be screened in order to evaluate their involvement in SLI. Conclusion In conclusion, this has been an exciting year in linguistic genetics. We have witnessed the identification of the first gene to be implicated in a speech and language disorder, and have taken the first step forward in the process of identifying gene variants that may affect language variation within a more general population. It is hoped that over the next few years the extension of these studies will provide us with a better understanding of the processes involved in language acquisition, and will pave the way for better diagnostic tests and potential therapies for disorders such as SLI. References Pinker S (1994) The Language Instinct. The New Science of Language and Mind. Penguin Books Biologist (2002) 49 (6) 5

6 Bartlett CW et al (2002) A major susceptibility locus for Specific Language Impairment is located on chromosome 13q21. American Journal of Human Genetics, 71, Bishop DVM (2001) Genetic and environmental risks for specific language impairment in children. Philosophical transactions of the Royal Society of London. Series B: Biological sciences, 356, Bruce HA & Margolis RL (2002) FOXP2: Novel exons, splice variants, and CAG repeat length stability. Human Genetics, 111, Lai CSL et al. (2001) A forkhead-domain gene is mutated in a severe speech and language disorder. Nature, 413, Newbury DF et al. (2002) FOXP2 is not a major susceptibility gene for autism or Specific Language Impairment (SLI) The American Journal of Human Genetics, 70, The SLI Consortium (2002) A genome-wide scan identifies two novel loci involved in specific language impairment (SLI). The American Journal of Human Genetics, 70, Van der Lely HK and Stollwerck L (1996) A grammatical specific language impairment in children: an autosomal dominant inheritance? Brain and Language, 52, Further reading For a review of the SLI phenotype please see: Bishop DVM (1997) Uncommon Understanding: Development and disorders of language comprehension in children. Hove: Psychology Press For a review of the genetic analysis of complex traits please see: Lander ES and Schork NJ (1994) Genetic dissection of complex traits. Science, 265, For a review of the molecular genetics of speech and language disorders please see: Fisher SE (2002) Isolation of the genetic factors underlying speech and language disorders. In: Behavioural Genetics in the Postgenomic Era Plomin R, DeFries JC, Craig IW, McGuffin P (Eds) APA Books: Washington DC For a review of the genetics of language please see: Stromswold K (2001) The heritability of language: A review and metaanalysis of twin, adoption and linkage studies. Language, 77, Website The website for the Wellcome Trust Centre for Human Genetics in Oxford, with details of the on-going research. Anthony Monaco received his undergraduate degree from Princeton University and his MD-PhD from Harvard Medical School. He was a post doctoral fellow in Hans Lehrach's laboratory at the ICRF followed by four years as an ICRF senior scientist and head of the Human Genetics Laboratory at the Institute of Molecular Medicine, Oxford. In 1995, he was awarded a Wellcome Trust Principal Research Fellowship and joined the Wellcome Trust Centre for Human Genetics in Oxford working on the genetic basis of neurological and psychiatric disorders including autism and developmental language and reading disorders. In 1998 he was appointed as Director of the Wellcome Trust Centre for Human Genetics. Dianne Newbury received her undergraduate degree from Nottingham University and her PhD from Oxford University. She has worked in Anthony Monaco s laboratory at the Wellcome Trust Centre for Human Genetics since 1997and heads the SLI genome screen project. The Wellcome Trust Centre for Human Genetics University of Oxford Roosevelt Drive Oxford OX3 7BN, UK anthony.monaco@well.ox.ac.uk dianne@well.ox.ac.uk 6 Biologist (2002) 49 (6)