The first candidate gene for dyslexia: Turning the page of a new chapter of research

The subfield of research on both typical and atypical reading development is one of the most saturated with ideas, researchers, and financial support with respect to the fields of human development and education. The article by Taipale et al. (1) in this issue of PNAS is an outcome of a truly multidisciplinary field, attended by researchers from different areas. There are multiple reasons for such a high level of attention. First, reading is a distinctly human skill, the functional role of which in human society has been increasingly synchronous with society's modernization so that gross national products highly correlate with literacy rates (www.unesco.org/education). Second, the acquisition of this skill has to be performed anew by every individual child, with enormous amounts of interindividual variability existing in the development of the skill (2). Third, the emergence of this skill occurs at multiple levels engaging genes, brain, and behavior (3). Finally, the biological machinery underlying the construction of reading competency appears to be general for different linguistic systems (4). Thus, the development of the skill of reading presents a wonderful opportunity to study a specifically human skill that is biologically rooted but that cannot be acquired outside of a social context. Its acquisition is individualized and results in suboptimal performance in up to 5% of the general population (5). The prevalence of specific reading difficulties (dyslexia) and their lifelong persistence and effects place dyslexia among the most important public health problems (www.nichd.nih.gov and ref. 6). Early evidence from the turn of the 20th century showed that suboptimal performance on reading tasks could occur in the absence of severe neurological problems and in the context of adequate education (7). In the 1950s, dyslexia was shown to be familial (8). Extensive effort has been invested in understanding the mechanisms of the familiality of dyslexia. Dozens of research groups have investigated heritability, models of transmission, and genetic background of specific difficulties in reading (9). Multiple genetic regions of interest have been identified suggesting the involvement of a number of genes (see Fig. 1).

Fig. 1.

Positive regions of genetic linkage and association implicated in at least two independent studies (approximated by bars), which included different samples of individuals with specific reading disabilities. The arrow indicates the position of DYX1C1 (15q21.1), residing in one of the previously reported regions of interest. [Modified with permission from ref. 9 (Copyright 2002, Nature Publishing Group).]

In this issue of PNAS, Taipale et al. (1) present results of research that are both a logical outcome of the extensive effort of many research groups and an exciting beginning of a new stage of research into genetic pathways of dyslexia. Taipale et al. report results of the characterization of a novel gene, which they refer to as DYXC1. This gene is located on 15q21, near the locus previously reported in a number of linkage studies of dyslexia (10–13). DYXC1 encodes a 420-aa protein with three tetratricopeptide repeat (TPR) domains. The TPRs, functionally, are known to be protein interaction modules; otherwise the DYXC1-encoded protein is not characterized by any homology to known proteins. Its role in the human organism is yet to be determined. DYXC1 was identified through studies of a family cosegregating the translocation (2;15)(q11;q21) and dyslexia. In addition, Taipale et al. conducted detailed studies of the structure of the gene and identified two sequence changes in DYXC1 that appear, at least in a statistically distinguishable portion of individuals, to cosegregate with dyslexia.

Their article is published 20 years after the publication of the first molecular-genetic paper on dyslexia in 1983 (14). Remarkably, Taipale et al. use a methodology similar to that used by Smith et al. in 1983. This method is a combination of cytogenetic and family- and population-genetic approaches. Nevertheless, the technological aspect of the 2003 article fully demonstrates the technological advances experienced by the field in the last 20 years.

Although the article represents an important step forward, by no means do the identification and initial characterization of DYXC1 solve all unknowns in the genetic equation of dyslexia. Below I discuss four issues that appear to be relevant in the context of the publication of the article by Taipale et al.

The first issue is whether and how soon the field will see a confirmation from different laboratories and different samples of the association between variants of DYXC1 and dyslexia. At this point, the answer to this question is ambiguous. The disruption of DYXC1 cosegregates with dyslexia in the family with the translocation (13). However, it is noticeable that, even in relatively genetically homogeneous samples of Finnish individuals with dyslexia, the frequencies of the two functional single-nucleotide polymorphisms (SNPs) and the haplotype including both SNPs are relatively low. Yet the associations are both present and convincingly statistically significant; unlike in the case of FOXP2, the gene implicated in at least two instances in another group of common neuropsychiatric disturbances such as speech and language disorders (15). As with the identification of DYXC1, FOXP2 was initially characterized in a patient with a translocation and a family with severe speech and language impairment; however, the identified variants of FOXP2 failed to show associations with samples of patients with specific language impairment (16–18). The evidence on the rate of coding divergence across different species for DYXC1 presented by Taipale et al. suggest, although indirectly, that the DYXC1 might be less evolutionarily preserved than FOXP2 and might be characterized by the presence of a number of common variants. Subsequently, in view of recent metaanalytical insights into the role of common variants in genetic susceptibility to common diseases (19), there is a basis for expecting that these common variants will be associated with dyslexia in other samples.

The second issue pertains to the specificity of the impact of functional mutations in DYXC1. In other words, will DYXC1, initially identified through a family with dyslexia, remain a specific dyslexia gene? Discussing the location and structure of the dyslexia-associated SNPs in DYXC1, Taipale et al. refer to animal work that suggests that the function of DYXC1 might not be limited to reading only, but might relate to broader mechanisms of learning. In addition, it is of interest that the father in the family with the translocation is characterized by severe reading and writing problems, whereas the son is characterized by severe reading problems and lowered general intellectual performance. The issue of comorbidity among common neuropsychiatric disorders with onset in childhood, such as reading, language, writing, math, and attention disorders, has been prominent in the literature for a number of years (20). Epidemiological studies have indicated that the rates of co-occurrence of these disorders are high. It also has been stressed that the common feature of all of these childhood conditions is that they impact learning and acquisition of specific competencies. Therefore, collectively, these conditions are often referred to as disorders of learning. To illustrate, the correlation between reading comprehension and mathematical problem-solving difficulties in individuals has been estimated at up to 0.80 (21). In addition, it has been reported that 25–40% of children diagnosed with specific reading disabilities meet the criteria for attention-deficit hyperactivity disorder (ADHD) (22). A number of hypotheses have been proposed to explain an observed overlap between these conditions. These hypotheses addressed, among other things, (i) the role of assortative mating, assuming the mutual attraction of individuals with various conditions of challenged learning patterns (23); (ii) manifested phenocopies of the second condition in the presence of the true first condition (24); and (iii) the presence of etiologically distinct groups with multiple problems (25). Yet the leading hypothesis in the field today recognizes the overlapping genetic etiologies for these conditions and suggests the presence of pleiotropic influences exerted by at least some of the many genes influencing disorders of learning for more than one comorbid condition (26, 27)†. At this point, there have been no systematic studies addressing the issue of genetic bases of comorbidity between various disorders of learning, but the field is fully cognizant of the need for such studies. It might turn out that the pleiotropic network of DYXC1 includes more phenotypes than merely specific reading disability. This seems especially likely in view of the widespread pattern of expression of the protein produced by DYXC1, as reported by Taipale et al.

The dyslexia-associated gene, DYXC1, might relate to broader mechanisms of learning.

The third relevant issue is that of the balance between evidence-driven and hypothesis-driven approaches to identifying genes for common disorders of learning. Traditionally, molecular-genetic work in the field of dyslexia has been of the former type, where regions of interest on the human genome, identified either by means of linkage analyses or through analyses of cytogenetic abnormalities, have been examined with the goal of identifying the genes contributing to susceptibility for dyslexia (13, 14, 28). Clearly, the discovery of DYXC1 provides the field with a good candidate gene for association studies. However, it is important to consider other candidate genes and candidate gene systems emerging from research on learning in animals and in humans whose performance is challenged by various pharmacological agents. Two lines of research warrant mention here. First, there is abundant evidence in behavioral studies of dyslexia that individuals with specific learning disorders, as a group, are characterized by working-memory deficits (29). In addition, there is seminal work in primates linking workingmemory functioning with the efficiency of dopaminergic neurons in the prefrontal cortex (30), as well as evidence showing the association between genetic variants of dopamine-related genes and indicators of working memory (31). Moreover, there is convincing evidence indicating the involvement of the prefrontal cortex in a variety of linguistic functions (32), including reading (33). However, the role of genes involved in the syntheses, distribution, and degradation of dopamine has never been investigated in dyslexia. Second, there is overwhelming evidence in the literature indicating the role of various γ-aminobutyric acid (GABA)-related agents to performance in diverse kinds of learning tasks in rodents. Moreover, there is growing support for the hypothesis of the central role of the GABAergic system in neuronal plasticity required for repetition adaptivity (34, 35). Yet the genes involved in the GABA turnover in the brain have never been systematically investigated as candidate genes for dyslexia. Thus, although the discovery of DYXC1 will, no doubt, stimulate a number of association studies, other candidate genes whose relevance to reading might be hypothesized through a number of relevant literatures should not be neglected as plausible candidates for dyslexia.

My final remark here will relate to the most important issue, namely, that despite the caveats discussed above, there is now a first candidate gene for dyslexia. The very discovery of such a gene opens possibilities for new innovative studies and further exploration of the relative contribution of DYXC1 in the manifestation of dyslexia in the context of everything else known about dyslexia at the levels of behavioral and brain functioning. Further studies of DYXC1 will determine the frequencies of dyslexia-related variants in different populations speaking different languages. Assuming that these variants are relatively common, stratification of the patients with dyslexia into subgroups will be possible. Letting one's imagination run away, one can foresee deeper understanding of the subtypes of dyslexia by genotype-based group comparisons of cognitive-behavioral profiles and specifics of organization and functioning of cortical systems differentiating and integrating processing of orthographic, phonological, and lexical-semantic features of words.