An association study was performed to elucidate the role of the serotonin transporter (5-HTT) gene as a susceptibility factor for autism as treatment of patients with antidepressant drugs which selectively target 5-HTT reduced autistic or concomitant symptoms, such as repetitive behavior and aggression, and ameliorate language use. Using the transmission/ disequilibrium test (TDT) an analysis was done for a common polymorphism in the upstream regulatory region (5-HTTLPR), a VNTR in intron 2 of the gene and a haplotype of both loci in 52 trios fulfilling stringent criteria for autism and an extended group of 65 trios including patients showing no language delay in their first 3 years of life. A higher frequency and preferential transmission of the long allele of the 5-HTTLPR was observed, but the TDT gave a statistically significant value (P = 0.032) only for the extended patient group. This result is in contrast to a recent study by a US group presenting preliminary evidence for preferential transmission of the short allele of 5-HTTLPR in 86 trios. Both studies failed to reveal significant linkage disequilibrium between the VNTR in intron 2 of the gene and autism. In our study haplotype analysis of the 5-HTTLPR and the VNTR in intron 2 supplied evidence for an association of 5-HTT and autism in the stringent (P = 0.069) and extended patient group (P = 0.049). Overall, we were not able to replicate the findings of the first study on 5-HTT and autism and instead observed a tendency for association of the opposite genetic variant of the gene with the disorder. The implications for genetic variants of the serotonin transporter in the etiology of autism and possible subgroups of patients, therefore, needs clarification in further studies with other and larger patient samples.
Infantile autism (McKusick no. 209850) is a severe developmental disorder characterized by marked social deficits, deviant language and a restricted range of stereotyped repetitive behaviors, usually occurring within the first 3 years of life. The prevalence in autism is ∼4/10 000 births and the ratio of affected males to females is ∼3:1 (1). A genetic etiology for autism is now strongly supported by family and twin studies (2–6). The recurrence rate among siblings of autistic individuals is ∼3%, which is 50–100 times higher than the risk in the general population (1,7). Together these data suggest that several susceptibility factors contribute to autism. Molecular genetic studies aimed at finding a gene or genes responsible for causing this complex disorder are therefore necessary.
Serotonin (5-hydroxytryptamine, 5-HT) is involved in a range of behaviors and psychological processes, including mood, anxiety, obsessive-compulsive symptoms, aggression, impulsivity, sleep and social interaction and affiliation. A series of studies have shown elevated whole blood 5-HT in children with autism (8–11). The potential role of the serotonergic system in the etiology of autism is still under investigation but several serotonin transporter (5-HTT) re-uptake inhibitors have been demonstrated to be partially successful in treatment of autistic symptoms such as repetitive behavior, aggression and language use (12,13). Therefore, 5-HTT seems to be a good candidate for association studies using two known common polymorphisms in the serotonin transporter gene. A VNTR in intron 2 has been described (14) and already shown to be associated with affective disorders (15,16). A deletion/insertion polymorphism in the transcriptional control region (5-HTTLPR) of 5-HTT displays different transcriptional efficiencies in transfection assays and in lymphoblastoid cell lines (17,18). In a recent study the short variant of the 5-HTTLPR was found to be associated with autism and haplotype analysis of both polymorphisms in a multiallelic TDT underlined the significance of the locus (19). The current study replicating these analyses, though this was not the primary purpose, in a German patient sample shows opposite findings and enables us to reflect on the importance of the serotonin transporter as a genetic risk factor in autism.
The characteristics of the sample of 109 patients from 88 singleton and 10 multiplex families are outlined in Table 1. Patients were subdivided into groups A and B, the latter containing 25 patients showing no language delay. Although language delay is one stringent criterion for the diagnosis of infantile autism, the patients in group B nevertheless fulfilled the criteria in all four domains of the ADI-R (qualitative impairment in social interaction; qualitative impairments in communication; restricted, repetitive and stereotyped patterns of behavior, interests and activities; onset prior to age 3). Patient groups did not differ significantly in age and IQ. The ratio of male to female was 3:1, as previously reported by other studies. A total of 52 complete trios in patient group A and 65 trios in the combined patient groups A and B was finally used for genotyping and subsequent association studies, because blood was not available for one or both parents of 36 patients (Table 1) and incomplete trios would bias the results of association analyses. Affected siblings from multiplex families were eliminated primarily because the TDT assumes independent observations.
Genotype and allele frequencies of the patient and parent group samples used for the TDT for the 5-HTT gene-linked polymorphic region (5-HTTLPR) and for the intron 2 VNTR of the 5-HTT gene are shown in Tables 2 and 3respectively. In patient group A and the combined sample A and B the frequency of the long allele of the 5-HTTLPR was higher than in the parent sample (Table 2). For the intron 2 VNTR of 5-HTT the 12 copy allele occurred at a slightly higher frequency in the patient sample when compared with the parent control group (Table 3). For both loci the differences between the male and female patients and the patient and parent groups respectively did not reach statistical significance at the 5% level using Fisher's exact test (P = 0.41–1.00). Overall, the parent allele frequencies of both polymorphic loci are in agreement with previously published data from several control populations (15–18). No significant difference was identified between the observed 5-HTTLPR genotype distribution of a German control sample of 301 individuals (19) and the parents of patient group A (P = 0.734) or A and B (P = 0.840) (Table 2). Comparison of the allele frequencies between these two groups showed a P value of 0.651. For the 5-HTT intron 2 VNTR the parent groups were compared with a German control sample of 218 individuals (20; Table 3). Allele frequencies did not differ significantly in both parent groups (P = 0.723 and P = 0.769) in comparison with the control sample, but the genotype distribution of the parents of group A showed a significant difference, with a P value of 0.031 compared with the control. This difference was still apparent in the parents of the broader phenotype patients in the combined groups A and B (P = 0.138).
TDT analysis with 65 trios for the 5-HTTLPR and 5-HTT intron 2 VNTR alone and a combined constructed haplotype is shown in Table 4. Haplotyping was not possible in four trios because of an irresolvable phase. In the TDT analysis these were treated as not genotyped. Using patient sample A, marginal association was only evident with haplotypes of the two allelic loci of the 5-HTT gene (P = 0.069). We found significant preferential transmission of the long allele of the 5-HTTLPR in the combined German patient sample (P = 0.032). This is in contrast to the study of the US group using 86 trios for association analysis (21), who presented preferential transmission of the short allele of the 5-HTTLPR. In agreement with the haplotype data of the US group, haplotype analysis of the combined patient sample in our case emphasized the significance of the 5-HTT gene region (P = 0.049), but with another combination of alleles of the two marker loci. Here, preferential transmission of the long/12 copy and long/10 copy haplotypes was demonstrated, reflecting the observed preferential transmission of the long allele in the single marker analysis. Therefore, association studies in the German patient sample also suggest the serotonin transporter gene as a possible susceptibility factor in autism. In contrast to the US patient sample, however, the long allele of the 5-HTT gene-linked polymorphic region is the preferentially transmitted allele. Inclusion of affected siblings of multiplex families in the sample, treating them as independent trios, did not significantly change the overall results. In the combined patient sample a slightly less significant P value of 0.059 for the 5-HTTLPR was determined, but a more significant P value of 0.014 was obtained for the haplotype analysis.
In the present study we analyzed the possible involvement of 5-HTT in the etiology of autism by transmission/disequilibrium testing of two 5-HTT gene-specific polymorphisms in a German sample of 65 trios. To define marked phenotypic differences in cases the patient sample was subdivided into two groups, A and B. Probands who showed no obvious language delay in their first 33 months of life were not considered as patients in sample A, which were required to fulfill stringent inclusion criteria (narrow phenotype also displaying language delay). Impairment of communication is one key factor in the description of autism and there is a possibility that language use has its own valence in the disorder. Therefore, patients little impaired in their verbal communicative ability could represent a specific subgroup of autism which might disturb any findings related to the classical phenotype of autism.
The results of our study were marginally supportive of association between haplotypes of the two markers in the 5-HTT gene and autism for the patient group A. In contrast, each marker on its own did not reach statistical significance for preferential transmission of one of the alleles in the patient group required to fulfill stringent diagnostic criteria, but a tendency for transmission of the long allele of 5-HTTLPR and the VNTR 12 copy allele of 5-HTT was evident. Using the combined patient sample with a slightly broader phenotype we discovered preferential transmission of the long allele of the 5-HTTLPR and significant distortion of transmission for the haplotypes, especially the long/12 copy and short/10 copy variants, of the two markers. To our knowledge this is the first time that preferential transmission of the long allele of the 5-HTT promoter variant has been observed in a psychiatric disorder.
Recently the short variant has been reported to be a quantitative trait locus for anxiety (18) and bipolar affective disorder (16). But most remarkable is a recent study by a US group demonstrating preliminary evidence for the short variant of 5-HTTLPR being preferentially transmitted from parents to autistic patients (21). For this and our group patient ascertainment relied on the widely accepted instruments AdI-R and ADOS and patients with an IQ below 35 and known medical conditions for a developmental disorder were excluded. Inclusion of patients who showed no language delay strengthened the significance of the distortion of transmission for the long variant of the 5-HTTLPR and the haplotype of the two markers in our study. It is speculative to think of a different genetic background for patients of this phenotype in comparison with the classically defined phenotype of autism from a smaller patient sample. On the other hand, it is possible that the more significant value of the TDT analysis for the promoter variant and the haplotype stems from the larger patient sample used and that the specific lack of communication deficits had no specific impact for the study. Contrarily, the tendency to preferential transmission of the VNTR 12 copy allele was lowered in the combined patient sample in comparison with patient sample A, overall showing no statistically evident distortion. Also, the US group did not find linkage or a linkage disequilibrium between autism and the VNTR of 5-HTT (21). In this case both studies were comparable, with a slight preference for the 12 copy allele. We cannot make any assumptions if the observed significant difference in the genetic distribution of the VNTR alleles of the parent group A in comparison with the German control sample (Table 3), which was mainly due to an increased number of the 12/10 copy alleles and a decreased number of the 10/10 copy alleles, has an influence on the association results. Analysis of the haplotypes in both studies, with the difference that contrary allele combinations were preferentially transmitted, leads to the conclusion that the VNTR is in linkage disequilibrium with another variation in the gene rather than having an effect on gene expression itself (19). One reason we can imagine leading to the discrepancy between the outcome of our study and the results of the US group is that subjects were from different geographical areas and ethnic backgrounds. The proportion of non-Caucasian probands was relatively high in the US study group in comparison with only one Asian patient in our sample, in which all others were of Caucasian background. We can speculate about different founder effects of an associated susceptibility factor for autism in populations of different ethnic background if the discovered association results can each be replicated in a separate patient sample collected from the same geographical region. In the light of the conflicting findings of the two studies, each only testing a relatively small patient sample, involvement of a specific allelic variant of the serotonin transporter as a genetic risk factor in autism is not yet clear. The results of the German study can therefore be interpreted as a non-replication, especially as the patient group fulfilling stringent diagnostic criteria did not reach statistical significance for either marker of the 5-HTT gene and only marginal statistical significance for the haplotype. Other studies using the VNTR in intron 2 of 5-HTT were also contradictory in showing association of the rare 9 copy allele with unipolar but not bipolar affective disorder in Scottish patients (15), the 12 copy allele associated with English and Welsh but not Chinese (16), Scottish (15) or German subjects (20). Hence, this would suggest a population-specific association of the 5-HTT gene rather than an effect of the gene itself for the disorder under investigation.
Nevertheless, we believe that more frequent inheritance of the long allele of 5-HTTLPR is in better agreement with known biochemical studies of the serotonergic system in patients with autism. The finding of the US group has already been discussed as counterintuitive (22). Hyperserotonemia was found in several studies in autistic individuals (8–11). The raised whole blood levels of serotonin detected are mostly contained in the platelets and not in the plasma (23,24). It was shown that the elevated serotonin level was a consequence of increased active serotonin uptake into platelets (25). Functional gene expression studies of 5-HTT clearly demonstrated that the long variant of the 5-HTTLPR is responsible for increased expression and uptake of 5-HT into JAR human placental choriocarcinoma cells and human lymphoblastoid cell lines (17–19,26). Increased 5-HT re-uptake into neurons of the serotonergic system in the brain would therefore lead to a depletion of 5-HT and decreased signal transduction in the post-synaptic neuron (24). Antidepressants, such as the selective 5-HT re-uptake inhibitors clomipramine and fluvoxamine, have been shown to successfully lower autistic symptoms or symptoms frequently seen in probands with autism, such as repetitive and ritualistic behavior and aggression, and also ameliorate language use (12,13). Hence, blocking of the re-uptake mechanism for 5-HTT would then lead to higher steady-state levels of 5-HT for neurotransmission.
In summary, we show preliminary evidence that in a German study the long variant of the 5-HTT promoter locus was preferentially transmitted from parents to patients with autism. We found high transcriptional activity of 5-HTT as likely to be a possible genetic susceptibility factor for autism. Whether the conflicting findings concerning the preferentially transmitted alleles of 5-HTT between the German and US patient samples (21) reflect etiological heterogeneity, differences in the selection of patients, a low power of the tests due to a small sample size or true non-replication needs careful replication in other and larger independent patient samples of the same and different ethnic and geographical backgrounds to clarify the implications for 5-HTT in the etiology of autism. Clinical studies should explore the relationship between genotypes of 5-HTT in autistic patients with the response to potent selective 5-HTT re-uptake inhibitors. Integration of results from molecular, clinical and biological psychiatry should further support the understanding, treatment and finally genetic background of autism in the future.
Materials and Methods
Subjects with autism were recruited in collaboration with a parents' association (2018;Helping the Autistic Child2019;) in several regions of Germany, in the Child Psychiatry Clinics of Frankfurt and Munich and in Austria. This ongoing study was designed to collect families with one or more autistic offspring for molecular genetic research. Patients were assessed using the audiotaped German version of the Autism Diagnostic Interview-Revised (ADI-R) (27,28) according to the DSM-IV (29) and ICD-10 (30) research guidelines and the videotaped Autism Diagnostic Observation Schedule (ADOS) (31), psychologically tested for an IQ > 35 and neurologically examined. For the psychometric tests the German version of the Wechsler Intelligence Scale for Children (WISC/WISE; HAWIK-R/HAWIE-R) was used or, if not otherwise testable, the Columbia Mental Maturity Scale (CMM) (32). Subjects with identified causes of autistic symptoms (e.g. fragile X syndrome) were excluded from the patient sample (33), as were patients having a record of infantile spasms in their first years of life, a possible medical condition leading to a phenotypical condition resembling autism. After diagnosis 109 patients from 88 singleton and 10 multiplex families respectively were included in the study. One family was Asian, all others were Caucasian. Patients or legal representatives gave their informed consent prior to inclusion in the study.
Blood DNA was extracted by standard methods and diluted to 20–100 ng/µl for genotyping. Amplification of the short (484 bp) and long (528 bp) variants of the 5-HTT gene-linked polymorphic region (5-HTTLPR) was accomplished with oligonucleotides stpr5, corresponding to nucleotide positions-1416 to–1397 (52032;-GGCGTTGCCGCTCTGAATGC), and stpr3, positions-910 to-888 (52032;-GAGGGACTGAGCTGGACAACCAC), as previously described (17). Briefly, PCR was performed in a final volume of 30 µl containing 20–100 ng DNA, 100 µm dNTPs (dGTP/7-deaza-22032;-dGTP = 1:1), 0.1 µg each primer, 1.5 mM MgCl2, 20 mM Tris-HCl, pH 8.6, 50 mM KCl, 0.2% (w/v) bovine serum albumin, 5% DMSO and 0.5 U Taq polymerase (Ampli-Taq; Perkin-Elmer). Initial denaturation was at 95°C for 3 min, followed by two cycles at 95 (30 s), 63 (30 s) and 72°C (1 min), two cycles with annealing at 62°C (30 s) and 35 cycles with annealing at 61°C (30 s), with a final extension of 10 min at 72°C. Primers HTT2X (52032;-TGGATTTCCTTCTCTCAGT-GATTGG) and HTT2Y (52032;-TCATGTTCCTAGTCTTACGC-CAGTG) amplified 390 (12 copy), 360 (10 copy) or 345 bp (9 copy) fragments of the 5-HTT intron 2 variant (21). PCR was performed in a 50 µl volume with 20–100 ng DNA, 100 µm dNTPs, 20 pmol each primer, 1.0 mM MgCl2, 20 mM Tris-HCl, pH 8.6, 50 mM KCl, 0.2% (w/v) bovine serum albumin and 0.5 U Taq polymerase. PCR conditions were 2 min initial denaturation at 94°C, 40 cycles at 94 (1 min), 57 (1 min) and 72°C (2 min), followed by 10 min at 72°C final extension. Alleles of both polymorphic loci were resolved with 2% agarose gel electro-phoresis next to a DNA molecular weight standard.
Fisher0027;s exact test for testing independence in contingency tables was used for comparison of genotype distribution and allele frequencies in the samples (34). The TDT (35) was used for two allelic marker loci. The TDT statistic for testing the null hypothesis of no association is given by the classical McNemar test statistic for pairs of transmitted and non-transmitted parent alleles from only heterozygous parents. Under the null hypothesis the test statistic is approximately χ2 distributed with 1 degree of freedom. If multiallelic marker loci were tested for disease marker association the maxima of the TDT statistics for the individual alleles were used. A randomization procedure was applied and an empirical P value was computed based on the number of 2000 permutations used in the procedure (36). Individual differences were judged as statistically significant if the corresponding P value of the test was <5%. In addition, a haplotype was constructed of both loci in the 5-HTT gene and used in further analyses, assuming that recombination frequency between the two loci is negligible because of close proximity («15 kb) of the markers (26). For extended analysis including the affected siblings of multiplex families these were split into single trios.
We express our gratitude to all families for providing blood samples, for their patience during interviews and all psychiatrists and physicians involved in the study for their support in collecting data. We thank A.Spieler, E.Dittert and S.Epp for excellent technical assistance. We are grateful to members of the International Molecular Genetic Study of the Autism Consortium for their helpful discussions concerning inclusion criteria of patients and D.Weeks and N.Heiss for their valuable discussions during revision of this paper. This work was supported by grants of the Deutsche Forschungsgemeinschaft to A.Poustka and F.Poustka. K.-P.L. is supported by the Hermann and Lilly Schilling Foundation.
Autism Diagnostic Interview-Revised
5-HT transporter gene
5-HTT gene-linked polymorphic region