Association between FOXP2 gene and speech sound disorder in Chinese population


Hongwei Ma, MD, PhD, Department of Developmental Pediatrics, Shengjing Hospital, China Medical University, 36, Sanhao St, Shenyang 110004, China. Email:


Aim:  FOXP2 was described as the first gene relevant to human speech and language disorders. The main objective of this study was to compare the distribution of FOXP2 gene polymorphisms between patients with speech sound disorder and healthy controls.

Methods:  Five FOXP2 polymorphisms, rs923875, rs2396722, rs1852469, rs17137124 and rs1456031, were analyzed in 150 patients with speech sound disorder according to DSM-IV, as well as in 140 healthy controls. Coding exons for key domains of FOXP2 were also sequenced in all the patients.

Results:  Significant differences in the genotype (P = 0.001) and allele (P = 0.0025) frequencies of rs1852469 (located 5′ upstream of the ATG initiator codon) were found between patients and controls. The excess of the T allele in the patients group remained significant after Bonferroni correction (P = 0.0126). Further investigations revealed a risk haplotype: rs2396722T/+rs1852469T. Our screening of key domains did not detect any point mutations in this sample. But we detected heterozygous triplet deletion of the glutamine-encoding region of exon 5 that alter FOXP2 protein sequence in five probands. These changes are predicted to yield a polyglutamine tract reduction from 40 to 39 consecutive glutamines.

Conclusions:  Our data support a possible role of FOXP2 in the vulnerability to speech sound disorder, which adds further evidence to implicate this gene in speech and language functions.

SPEECH SOUND DISORDER (SSD) is also known as developmental phonological disorder. Children with SSD have difficulties in correctly producing the speech sounds appropriate for their age and dialect. It is different from the larger DSM-IV category of phonological disorder.1 For a diagnosis of SSD, cases that arise from hearing impairment, structural abnormalities of the speech mechanism, or known neurological conditions must be excluded. It is the most common speech disorder in children. The estimated prevalence is 15.6% in children at age 3 and 3.8% at age 6.2,3 In China, about 2.14% of preschool children suffer from SSD. Although the cause of SSD is unknown, there is a lot of literature suggesting that susceptibility to SSD is genetic, including familial aggregation studies4–6 and some twin studies.5,7,8

Forkhead-box P2 (FOXP2) gene (MIM 605317), located on 7q31, encodes a transcription factor containing a polyglutamine tract and a forkhead DNA-binding domain. It was the first gene discovered to be involved in speech and language disorder.9 In 1990, Hurst et al.10 described a rare pedigree known as KE family in which approximately half the members showed a speech and language disorder that is primarily characterized by problems coordinating sequences of mouth movements during speech (known as developmental verbal dyspraxia), accompanied by wide-ranging deficits in expressive and receptive language, both oral and written. In 2001, Lai et al.9 discovered that all 15 affected members of this family carried an causative mutation in FOXP2. This mutation involves a single base transition (G/A) in exon 14 of FOXP2 gene that results in an arginine to histidine substitution (R553H) in the DNA-binding domain of the encoded protein. Furthermore, the FOXP2 gene was directly disrupted by a translocation breakpoint of a patient with a similar clinical presentation, which increased further support.9 Structural and functional neuroimaging with gene expression studies supported a significant relationship between FOXP2 and the language neural system.11,12

In addition, researchers reported severe motor impairments, premature death, and absence of ultrasonic vocalization in response to stressors in Foxp2 knockout mice.13 However, Groszer et al.14 demonstrated that mice lacking function Foxp2 can produce structured ultrasonic vocalizations, but only do so under higher levels of stress. They further discovered that mice with a half-dosage of functional Foxp2 display abnormal synaptic plasticity and impaired motor-skill learning.14 Studies of songbirds have shown that reduced FoxP2 dosage can disrupt vocal learning performance in zebra finches.15

All these studies suggest that the FOXP2 gene is involved in the development of the neural system that mediates the specific motor coordination necessary for speech. As a consequence, FOXP2 has been considered as a candidate for a range of different language-related disorders, but findings have varied between studies.

Autism is a severe neurodevelopmental condition characterized by disturbances in social interaction and communication by repetitive body movements and restricted interests, and by atypical language development. Newbury et al.16 carried out association screening of FOXP2 as well as mutation screening in all coding exons, and concluded that variants of this gene are unlikely to play a role in autism or common forms of language impairment. Several other groups tested for the presence of linkage or linkage disequilibrium between FOXP2 and autistic disorder. The coding sequence of FOXP2 was screened and several polymorphisms were identified in multiplex autism families and normal controls. But they found no linkage or association between the FOXP2 gene and autism.17,18 Gong et al. reported that the FOXP2 gene may be related to the pathogenesis of autism in Chinese Han population,19 although their findings do not meet significance thresholds if adjusted for multiple testing.

The language receptive and expressive disorders are the main characteristics of specific language impairment. O'Brien et al.20investigated 96 probands with SLI and their family members by linkage and association to markers within and around FOXP2, and samples from 96 probands with SLI were also directly sequenced for the mutation in exon 14 of FOXP2. No mutations were found in exon 14 of FOXP2, but strong association was found to a marker within the CFTR gene and another marker on 7q31.

Language impairment is also one of the symptoms of schizophrenia. Up to now, only one article has reported a significant association between the FOXP2 polymorphism and schizophrenia with auditory hallucinations.21

In contrast to the generally negative results in autism and specific language impairment, MacDermot et al.22 detected novel variants that alter FOXP2 protein sequence in three probands with developmental speech problems. They investigated the entire coding region of FOXP2, including alternatively spliced exons, in 49 probands affected with developmental verbal dyspraxia. One of the coding changes was a heterozygous C/T transition in exon 7, yielding a stop codon at position 328 of the FOXP2 protein (R328X). This nonsense mutation was also found in the affecting sibling of the proband and their mother, who had a history of speech problems. The other two coding changes were a heterozygous A/T transition in exon 2 in one proband, and a heterozygous insertion of the sequence CAGCAGCAACAA into the polyglutamine-encoding region of exon 5 in another. In addition, Zeesman et al.23 found a child with developmental apraxia of speech and mild cognitive delay who had a deletion of 7q31 that included FOXP2 gene.

All these results supported the possibility that the FOXP2 gene might be involved in the pathogenesis of speech-related disease. SSD is a speech disorder with similarities to developmental verbal dyspraxia, and may involve shared risk factors. So we attempted to investigate the association between the FOXP2 gene and SSD in the Chinese Han population using the analyses of association and haplotype. We also examined the coding exons for key domains by polymerase chain reaction (PCR) and sequencing analysis. For five probands, all the coding exons were tested, including the alternatively spliced exons, exon3a and 3b.



A total of 150 patients and 140 healthy unrelated controls of similar ethnic background were recruited from Shenyang, Shengjing Hospital, China Medical University. Of the 150 cases, 109 were male and 41 were female. The mean age of the children at the time of testing was 6.5 years (range 4–11 years). All the patients fulfilled the DSM-IV criteria for speech sound disorders (phonological disorder). The code number of DSM-IV is 315.39. Physical examinations were performed and patients were excluded if they had any medical or genetic conditions that could be contributing, such as prematurity, birth complications, mental retardation, autism spectrum disorder, sensorineural hearing loss, craniofacial anomaly (e.g. cleft palate), or any other known or acquired neurological condition. All subjects provided written, informed consent for participation in this study. The study was approved by the ethics committee of Shengjing Hospital, China Medical University. Of the 140 controls, 100 were male and 40 were female. The mean age was 6.8 years (range 4.5–12 years).


Genomic DNA was obtained from peripheral blood leukocytes using the standard phenol-chloroform method. We selected five single nucleotide polymorphisms (SNP) in the FOXP2 gene from the dbSNP database ( These SNP included rs923875, rs2396722, rs1852469, rs17137124 and rs1456031. Three SNP, rs923875, rs23967222 and rs1852469, are located in 5′ upstream of the ATG initiator codon. SNP rs17137124 is located in intron 3, and rs1456031 is situated in intron 10. All polymorphisms were genotyped by polymerase chain reaction-restriction fragment length polymorphism (PCR-RFLP) analysis and direct DNA sequencing was used in some random select samples to test the results of PCR-RFLP. The information of primers and restriction enzymes for each polymorphism is shown in Table 1.

Table 1.  Detailed information of the polymerase chain reaction-restriction fragment length polymorphism analysis for the five SNP and exons
SNP and exonsPrimer sequences (5′-3′)Product (bp)RFLP
  1. NA, not applicable; RFLP, restriction fragment length polymorphism; SNP, single nucleotide polymorphisms.

rs923875F: 5′-CTTGGGAAACTGAAGCCAG-3′625ApalI
rs2396722F: 5′-AAGGCATGAGTTCTTTGG-3′275VspI
rs17137124F: 5′-TACAAGTGGGAAGTAAGG-3′437AflII

The PCR amplification was performed in a 25-µL volume containing 50 ng of genomic DNA, 0.3 µm each primer, 2 µL of dNTP, 1U of Taq DNA polymerase, and 2.5 µL of 10 × PCR buffer. Samples were initially denatured at 94°C or 4 min, followed by 30–35 cycles of 94°C for 30 or 40 sec, 50–60°C for 40−60 sec, 72°C for 50 sec and a final extension step at 72°C for 7 min. In brief, restriction fragment length polymorphisms were digested with the appropriate restriction enzymes. The samples were loaded on 2% or 2.5% agarose gels containing ethidium bromide for electrophoresis at 100 V for 50 min. Gels were read blindly by two independent raters with discrepancies resolved by re-genotyping. Six random select samples for each SNP were tested again by direct DNA sequencing. PCR products were purified using a QIAQuick PCR purification kit (Qiagen, Germany). Direct sequencing of the samples was performed on an ABI 3730 DNA sequencer (Perkin Elmer, Foster city, California, USA). Sequencing results were compared with the reference human FOXP2 sequence and the results of RFLP.

A number of sample genotypes could not be assigned due to repeated PCR failure or unclear genotype results. These consist of two genotypes for SNP rs2396722; seven for SNP rs1852469; two genotypes for SNP rs17137124; and two genotypes for SNP rs1456031.


All the patients were also tested for the coding exons for key domains, including exon 5, 6, 8, 9, 10, 12, 13, 14 by PCR and direct sequencing. Also, exon 7 in which MacDermot found nonsense mutation was also examined. The entire coding region, including 3a and 3b, were examined in five probands with abnormal sequencing results. The exon 5 in 120 normal control chromosomes were also screened. The information of these primers is also shown in Table 1. The PCR amplification was performed in a 40-µL volume. Samples were initially denatured at 94°C for 3 min, followed by 35 cycles of 94°C for 40 sec, 55–65°C for 60 sec, 72°C for 60 sec and a final extension step at 72°C for 7 min. PCR products were purified using a QIAQuick PCR purification kit. Direct sequencing of the samples was performed on an ABI 3730 DNA sequencer. Sequencing results were compared with the reference human FOXP2 sequence. Clone sequencing was also used in five probands with abnormal direct sequencing results.

Statistical analysis

Significance level was previously established at 0.05. The Hardy–Weinberg equilibrium for genotype frequencies was evaluated by the χ2-test. The χ2-test was also used to compare the distribution of FOXP2 polymorphisms between patients and controls. The comparisons of allelic frequencies and genotype analyses between patients and controls were performed using the SHEsis program online24 ( The pairwise linkage disequilibrium (LD) between the FOXP2 SNP was also estimated using the SHEsis program. The normalized LD statistic, D′, was calculated to measure LD between markers. To estimate haplotype frequencies from the genotype data and to check for statistical significance of haplotype frequencies between patients and controls, the SHEsis program was used, too. The Bonferroni test was applied to correct for multiple comparisons.


We have genotyped five SNP of the FOXP2 gene in 150 speech sound disorder patients and 140 controls. Five SNP were selected, which covered from 8.5 kb to 571 kb of the FOXP2 gene. Three of them are located at 5′ upstream of the ATG initiator codon. The genotype and allelic frequencies of all five SNP are presented in Table 2. All of them were found to be in Hardy–Weinberg equilibrium in both patient and control samples. Statistical analyses of the SNP showed that neither genotype nor allele frequency distributions were different between SSD patients and control subjects with the exception of the SNP rs1852469. In this case, there were significant differences in the genotype (χ2 = 13.772, d.f. = 2, P = 0.001) and allele (χ2 = 9.129, d.f. = 1, P = 0.0025) frequencies (Table 2). These P-values remained significant after Bonferroni correction (P = 0.005; P = 0.0126, respectively).

Table 2.  Genotype and allele distribution of the FOXP2 gene in speech sound disorder patients and controls
  1. *Significant difference. SNP, single nucleotide polymorphism.

Patients567024  182118  
Controls537116  177103  
Patients347640  156144  
Controls267240  152124  
Patients205967  99193  
Controls247934  127147  
Patients428224  166130  
Controls566222  174106  
Patients427830  162138  
Controls397326  151125  

To test for LD between the FOXP2 SNP, D′ and r2, values were calculated for all pairs of SNP on patients and controls. The D′ values are indicated in Table 3. Moderate LD between markers rs923875 and rs2396722 (D′ = 0.579), and between rs2396722 and rs1852469 (D′ = 0.506) were seen in patients. A weak LD is detectable between rs923875 and rs1852469, and between rs17137124 and rs1456031. There is also a weak LD between rs923875 and rs1456031.

Table 3.  Pairwise LD between 5 SNP of the FOXP2 gene
  1. Lower diagonal, D′ values for pairwise LD in the control group. Upper diagonal, D′ values for pairwise LD in the patient group. *D′ values > 0.50. LD, linkage disequilibrium; SNP, single nucleotide polymorphism.


We considered the SNP with D′ > 0.5 to perform the haplotype analysis. Table 4 shows haplotypes carrying two or three SNP, with frequencies higher than 3% in both patients and controls. A risk haplotype was detected, rs2396722T/+rs1852469T (P = 0.0047, Global χ2 = 15.507, d.f. = 3, Globe P = 0.0014), which was significantly associated with SSD. After correcting the P-value by Bonferroni test to avoid false-positive results because of multiple testing, the association remained significant. In addition, when three SNP were considered, a statistically significant difference in the haplotype frequency was observed between patients with SSD and controls for the haplotype rs923875A/+rs2396722T/+rs1852469T (P = 0.0103, Global χ2 = 36.678, d.f. = 7, Globe P = 5.68e-006). However, the P-value failed to reach significant association when the Bonferroni test was applied. Alternatively, a protected haplotype rs2396722T/+ rs1852469A was also identified (P = 0.0009).

Table 4.  Frequencies of the major observed haplotypes
Haplotypesrs923875rs2396722rs1852469PatientsControlsχ2P-valueOR95%CIGlobe P
  • *

    Significant difference. CI, confidence interval; OR, odds ratio.

 1AC 0.0860.1343.3830.06590.6090.358∼1.0370.055826
 2AT 0.3070.2324.1110.0427*1.4681.012∼2.129
 3CC 0.4340.4170.1710.67891.0720.770∼1.493
 4CT 0.1730.2171.8170.17770.7530.497∼1.139
 5 CA0.2590.2910.6890.4066890.8550.590∼1.2390.001446
 6 CT0.2650.2540.0850.7707191.0580.725∼1.543
 7 TA0.0800.17211.0680.000884*0.4160.245∼0.705
 8 TT0.3960.2837.9800.004749*1.6621.167∼2.367

Then we examined the coding exons for key domains and exon 7. No mutations were found in exon 6, 7, 8, 9, 10, 12, 13 and 14. Exon 5 encodes 40 polyglutamine tracts by (CAG)4 CAA (CAG)4 (CAA)2 (CAG)2 (CAA)2 (CAG)3 (CAA)5 (CAG)2 (CAA)2 (CAG)5 CAA (CAG)5 CAA CAG. On direct sequencing, overlapping peaks in exon 5 sequencing readouts were observed for five probands, which made the bases difficult to call (see Fig. 1). A potential explanation was that these probands might carry heterozygous changes in the numbers of CAG/CAA repeats. We tested this hypothesis by clone sequencing. The five probands were detected as the same heterozygous deletion of a triplet codon CAA into the polyglutamine-encoding region of exon 5. The sequence is as follows: (CAG)4 CAA (CAG)4 (CAA)2 (CAG)2 (CAA)2 (CAG)3 (CAA)4 (CAG)2 (CAA)2 (CAG)5 CAA (CAG)5 CAA CAG.

Figure 1.

Overlapping peaks of exon 5 sequencing.

But for the 120 normal control chromosomes, no deletions were detected in exon 5. After that, we examined all the other coding exons for the five probands, but no other mutations were detected.


In the present study, we investigated the association of the FOXP2 gene and speech sound disorder in the Chinese Han population. The main finding of this study is that SNP rs1852469 of the FOXP2 gene showed significant association with SSD. In addition, the haplotype analyses support the association between the haplotypes rs2396722T/+rs1852469T, and rs923875A/+rs2396722T/+rs1852469T and SSD. The haplotype rs2396722T/+rs1852469T maintained a significant association with SSD when the Bonferroni test was performed.

A significant association between FOXP2 polymorphisms and schizophrenia with auditory hallucinations has been reported.21 But the T allele frequencies of SNP rs1852469 are very low in both patients and controls, and no association was found between rs1852469 and schizophrenia. Different ethnic distributions could contribute to a different genetic association.

Previous studies have identified missense (the R553H mutation of the KE family) and nonsense (R328X mutation in a second small family) mutations of FOXP2, causing speech and language deficits. We screened FOXP2 exons that code for important functional domains of the protein in all our SSD patients, but did not detect any missense or nonsense mutations. Nevertheless, we did find a heterozygous triplet codon deletion of exon 5 in five probands. At the amino acid level, this change is predicted to yield a polyglutamine tract reduction from 40 to 39 consecutive glutamines. No similar changes were identified in control groups, and no evidence for expansion instability of this polyglutamine stretch was found. Wassink et al.17 (2002) previously detected small internal deletions in the Q40 tract in two families with autism. MacDermot et al.22 reported a heterozygous insertion of the sequence CAGCAGCAACAA into the polyglutamine-encoding region of exon 5 in a proband with developmental verbal dyspraxia. Bruce and Margolis25 investigated the length of FOXP2 CAG/CAA repeat in 142 individuals with progressive movement disorders of unknown cause. They detected two individuals who had a single additional triplet in one allele. Based on Butland's analysis,26 the Q-tract length variation is 34–40. But neither Wassink nor MacDermot detected deletion or expansion of the longer FOXP2 polyglutamine tract in their control chromosomes. The different results are perhaps due to different ethnic distribution. Sobczak and Krzyzosiak27 illustrated the hairpin structure of poly-Q of FOXP2 transcripts. According to the structure, the mutating number of CAG/CAA repeat could change the hairpin of poly-Q, which perhaps influences the transcription of the FOXP2 gene. The functional study of poly-Q will be important to assess whether it has any impact on the behavior of FOXP2 or not, such as the transcription of FOXP2.

Newbury et al.16 (2002) identified a potential insertion of two CAG within the smaller (Q10) tract in a family with SLI, but this did not cosegregate with the disorder, and its location in a repetitive region at the intron/exon border of exon 6 suggested that it would not have any impact on the FOXP2 protein sequence. In our study, we haven't found any deletion or expansion of exon 6.

The FOXP2 gene is an important regulating gene. There are many FOXP2 targets in human basal ganglia and the inferior frontal cortex, and these target genes are known to play critical roles in specific aspects of central nervous system patterning or development, such as neurite outgrowth, as well as plasticity.28,29 Continually, Vernes et al.30found that FOXP2 binds to and dramatically downregulates CNTNAP2, a gene that encodes a neurexin and is expressed in the developing human cortex. CNTNAP2 is implicated in language deficits in both specific language impairment and autism. Recently, Konopka et al.31 identified 61 genes significantly upregulated and 55 genes downregulated by FOXP2. As a regulating gene, FOXP2 itself is regulated tightly. Schroeder and Myers32 identified four transcriptional start sites for human FOXP2 gene with varying cellular specificities. Also, the FOXP2 gene has alternative splicing exons at the 5′ end of the gene. SNP rs1852469 is located in 5′ upstream of the ATG initiator codon, which perhaps influences the transcription of the FOXP2 gene and the target genes of FOXP2 further. At a key stage of embryogenesis, FOXP2 and its target genes perhaps result in the abnormal development of neural structures that are important for speech and language. Our study gives another support to the role of FOXP2 in human cognition. But the function of rs1852469 is unknown, which is the obvious limitation of the study. Also, we have not sequenced all exons in the patient sample, so we cannot rule out the presence of mutations in the untested exons.

Taken together, the results in the present study indicate that the FOXP2 gene may confer vulnerability to speech sound disorder. Given the complex nature of SSD, further replications on different ethnic samples are warranted and other polymorphisms of the FOXP2 gene should be investigated. The FOXP2 gene is a regulating gene; further research on other genes, especially the downstream genes of FOXP2, may be more helpful.


This study was supported by the Section of Scientific Research, Shengjing Hospital, China Medical University and Science and Technology Research Fund of Educational Office, Liaoning Province (Grant No. L2010614). Special thanks to the patients for their support and participation.