METHODS

Subjects

Eleven families with 97 subjects, of whom 70 were available for thorough testing for dyslexia, were studied (fig 1). Nine families with 70 members altogether (28 dyslexic, 28 non-dyslexic, and 14 not tested) were recruited from the Department of Paediatric Neurology at the Hospital for Children and Adolescents (formerly Children’s Castle Hospital), University of Helsinki, Finland. These families were selected from about 140¹⁷ based on their informativeness for linkage. Two families with 27 members altogether (12 dyslexic, 11 non-dyslexic, and four uncertain) were recruited from the Central Hospital of Central Finland, Jyväskylä, Finland. In these two families, dyslexia testing, with normal results, was performed in three of the non-dyslexic subjects, and a further eight subjects reporting normal reading performance were also classified as unaffected.

The diagnostic criteria for dyslexia included remarkable deviation (depending on the age, at least two years) in reading skills compared to chronological age and normal performance intelligence quotient (IQ >85). The diagnosis of dyslexia was determined by Finnish reading and spelling tests designed for children under 13 years of age¹⁸ and adults,¹⁹ described elsewhere in detail.²⁰ The IQ was determined by WAIS-R²¹ or WISC-R²² and subjects with an IQ below 85 were excluded from this study. In order to determine whether dyslexia was the result of a deficit in one or more of phonological awareness, rapid naming, or verbal short term memory, reading related neurocognitive skills were assessed by neuropsychological tests.^23–26 The study was approved by the appropriate ethical committee and all family members participated with informed consent.

Genotyping

Twenty ml of EDTA blood was collected from each subject and DNA was extracted by a standard non-enzymatic method.²⁷ Genome wide scan was carried out at the Finnish Genome Centre, University of Helsinki, using 376 microsatellite markers from the Applied Biosystems Linkage Mapping Set MD-10. The average distance between markers was 10 cM. DNA (20 ng) was dried on microtitre plates for each PCR assay. The PCRs were performed in 5 μl volumes in conditions recommended by the reagent manufacturer (Applied Biosystems). The fluorescence labelled PCR products were pooled (10–20 markers/pool), and separated on a MegaBace 1000 capillary electrophoresis instrument (Molecular Dynamics, Sunnyvale, CA). The alleles were visualised using Genetic Profiler 1.5 software (MolecularDynamics).

Linkage analysis

The genome scan data were analysed by non-parametric and parametric linkage analysis using Genehunter and MLINK, respectively. Genehunter performs reconstruction of haplotypes and complete multipoint analysis of allele sharing identical by descent (IBD) among all affected family members at each location in the genome.²⁸

For parametric linkage analysis, a genetic model with a disease allele frequency of 0.0001, autosomal dominant inheritance (based on pedigree information, fig 1), and equal female and male recombination rates was specified. The penetrances for homozygous normal, heterozygous, and homozygous affected were 0.06, 0.95, and 0.95, respectively. In addition to dominant inheritance, parametric linkage analysis was also performed using recessive inheritance with a disease allele frequency of 0.1 and equal female and male recombination rates. The penetrances for homozygous normal, heterozygous, and homozygous affected were 0.02, 0.04, and 0.8, respectively. Both inheritance models were analysed because common recessive alleles in the population can cause inheritance patterns that are reminiscent of autosomal dominant heritance. These particular models were chosen to reflect similar prevalence figures under both inheritance patterns. Subjects with uncertain phenotypes were not included in the analysis.

Candidate gene FOXP2

The whole coding region, consisting of 17 exons of FOXP2 and including exon 1 and alternatively spliced exons 3a, 3b, and 4a, was sequenced from six subjects with dyslexia and three non-dyslexic subjects as controls (fig 2). One control was selected from the family members in this study and all had been tested negative for dyslexia by the neuropsychological tests (fig 1). One patient and two controls were selected from tested dyslexic families not included in this linkage analysis.

Primers flanking each exon were designed using the Primer3 program (http://www-genome.wi.mit.edu/cgi-bin/primer/primer3).²⁹ PCRs were carried out in 50 μl reactions containing 30–50 ng of genomic DNA, 1 × DyNAzyme II buffer, 1.5 mmol/l MgCl₂, 160 μmol/l dNTPs, 0.6 μmol/l of each primer, 0.6 U of DNA polymerase (DyNAzyme II, Finnzymes, Espoo, Finland), and 0–4% DMSO. In case of poor amplification, the DNA polymerases AmpliTaq Gold (Perkin Elmer, Roche Molecular Systems Inc) or DyNAzyme EXT (Finnzymes) were used in similar conditions. Amplifications were performed with an initial denaturation at 94°C for two minutes, followed by 35–40 cycles each of 35 seconds at 94°C, 35 seconds at 55–62°C, and one minute at 72°, with a final elongation at 72°C for eight minutes. Purified PCR products (PCR purification kit, Gel extraction kit, Qiagen) were either directly sequenced or cloned before sequencing (TOPO TA Cloning Kit, pCR 2.1-TOPO vector, Invitrogen). Sequencing was performed with ABI 377 and ABI 3100. To find sequence variants, all sequence reads were inspected by two investigators independently.

RESULTS

A total of 88 subjects from 11 families with 38 dyslexic members were genotyped with microsatellite markers spanning the whole genome. Genome scan data were analysed by non-parametric multipoint and parametric two point linkage analysis. For non-parametric analysis, the affected only mode of analysis was used. Two loci linked to dyslexia were found: the previously defined region on 2p corresponding to DYX3 and a novel locus on 7q32 corresponding to SPCH1 (fig 3). On chromosome 2, the highest NPL score was 2.55 for marker D2S2216 on 2p11 (p=0.004), with a single high NPL 3.02 (p=0.03) in family 2 (fig 1). Marker D2S2216 is approximately 34 cM centromeric from the DYX3 locus, based on marker locus information from the Marshfield and DeCode maps. On chromosome 7, the highest NPL score was 2.77 (p=0.003) for marker D7S530 on 7q32, with one family (No 10 in fig 1) showing an NPL of 4.21 (p=0.03). Parametric analysis showed a significant two point lod score of 3.01 for marker D2S286 (2p12) in the autosomal dominant model, whereas no significant lod scores were seen for chromosome 7 (table 1). Parametric analysis with the autosomal recessive model showed no significant lod scores. We observed no evidence for linkage above background to the previously reported dyslexia loci on chromosomes 15q21, 6p21.3, 3p12-q13, and 18p11.2.

The FOXP2 gene is located within the peak of our novel linkage region, approximately 15 Mb centromeric from marker D7S530 (fig 4) and became thus a positional candidate gene. The entire coding sequence of the candidate gene FOXP2 was sequenced from three controls and six dyslexic subjects from different families, including the two highest scoring pedigrees (10 and 5, fig 1). No mutations or SNPs were found; specifically, the G to A nucleotide transition in exon 14 was not detected in our samples.

DISCUSSION

Linkage of dyslexia to at least five chromosomal loci has been verified in previous studies and is consistent with a multifactorial phenotype. We conducted a genome wide scan in 11 families ascertained on the basis of dyslexia by commonly accepted criteria. In the genome scan, we detected novel linkage of dyslexia to 7q32 and further confirmed linkage to 2p11.^30,31 No other previously linked loci showed significant NPL scores. Intriguingly, the same 7q region had previously been linked to autism and the first gene implicated in a speech and language development, FOXP2, resides in the same region.^9,13 Our linkage peak at 7q32, at marker D7S530, maps approximately 15 Mb from FOXP2. FOXP2 appears to have a role in several facets of language processing and grammatical skills, which are also deficient in dyslexic subjects.¹⁵ Although the FOXP2 gene appears to have no role in autism or SLI, it may be a plausible candidate gene for other more specific language disorders such as dyslexia.

Our linkage peak 2p11 maps approximately 34 cM from the linkage peak reported originally by Fagerheim et al.³⁰ We suggest that this difference may be caused by dissimilar sample sets, diagnostic criteria, or markers used, and the results may in fact reflect the presence of one and the same locus. Alternatively, it is possible that there are indeed two different but closely located genes for dyslexia. A similar discrepancy has previously arisen for the mapping of susceptibility genes in pre-eclampsia with three different localisations in chromosome 2.³² Remarkably, both the non-parametric and parametric analysis modes in our material coincided for the peak of linkage. It is not possible to resolve the number of loci at present, but the question will need to be addressed with additional genetic studies.

To evaluate the role of FOXP2 in dyslexia, we screened the entire coding region in six dyslexic subjects, but found no mutations or polymorphisms. Obviously, possible mutations might hide in promoter regions or introns, but our results do not support a role for FOXP2 as a dyslexia candidate gene. Recently, six novel exons have been discovered, of which only one is included in the coding region.¹¹ It is possible that as yet undetected coding regions exist, with undetected mutations, but currently any such data remain beyond reach.

In addition to FOXP2, chromosome 7q31-q32 contains several genes that might be considered as candidates for dyslexia. Among genes that are expressed in brain are the G protein coupled receptor 37 (GPR37)³³ and the actin regulation protein (WASL).³⁴ Other interesting candidates include PTPRZ1, a protein tyrosine phosphatase receptor type zeta-1 that is expressed only in the central nervous system,³⁵ WNT2, a putative signalling molecule involved in CNS development,³⁶ and NRCAM, a neuronal cell adhesion molecule³⁷ (fig 4). In the absence of a more exact genetic localisation of our linkage peak, we did not sequence any of these genes.

Although FOXP2 appears not to be a dyslexia gene, the numerous potential genes in the area make our new linkage region at 7q32 extremely interesting. Even more, the linkage of autism to the same area suggests that genes other than only FOXP2 involved in language development might reside there. The roles of the remaining candidate genes in dyslexia and also other forms of language disorders, as well as fine mapping of DYX3, can now be targeted with samples from the families described here.