Autism (MIM 209850) is a neurodevelopmental disorder beginning before the age of 3 that is characterized by impaired reciprocal social interaction and communication as well as restricted and stereotyped patterns of interests and activities. The prevalence of autism is estimated to be at least 10 per 10,000, with a male to female ratio of approximately 4:1 (1). In the majority of cases, the cause of autism is unknown, but many studies, including correlation in twins and association of autism with Mendelian diseases, support a genetic etiology for this condition (2, 3). Autistic disorder appears to be clinically and genetically heterogeneous, suggesting that multiple loci are involved (4). It is therefore suggested that numerous chromosomal loci contribute to the genetic susceptibility to the disorder, since the identification of chromosomal abnormalities in autistic individuals and data from large-scale linkage studies suggest a linkage to several different chromosomal regions (2, 5). Recently, mutations in the NLGN3 and NLGN4 genes have been associated with autism, Asperger syndrome, and X-linked mental retardation (6, 7), which suggests, in these cases, that autistic spectrum disorders follow a recessive Mendelian mode of transmission.

The incidence of chromosomal aberrations in the population of patients with autism is much higher than would be expected relative to low incidence in the healthy population, suggesting that there is a causal relationship between the occurrence of autistic disorder and the chromosomal anomaly.

In this study, we describe a complete molecular genetic and functional analysis of an autistic disorder patient with a de novo balanced 9q23/10q22 translocation. We show that the disruption of the KCNMA1 gene leads to a haploinsufficiency and a decreased activity of the KCNMA1 (or BK_Ca) protein. We then demonstrate that the action of a BK_Ca channel opener is able to enhance a BK_Ca current. These findings support the hypothesis that autism is associated with defects in neuronal circuitry and excitability.

After a complete description of the study was given, written informed consent from the parents was obtained as well as assent from the subjects.

Clinical Data for Autistic Subjects

Patient carrying the de novo 9q23/10q22 translocation

The patient was a 6-year-old boy at the time of clinical evaluation. He had impairments of reciprocal social interactions and communication skills, lack of spoken language, and poor communicating gestures. He also displayed restricted stereotyped behaviors. The diagnosis of autistic disorder was confirmed by administration of the Autism Diagnostic Interview–Revised (8) and the Childhood Autism Rating Scale (9). The patient satisfied the prespecified cutoff scores in all three of the following symptom areas of the Autism Diagnostic Interview–Revised: qualitative impairment, with a reciprocal social interaction score of 30 (cutoff for autism=10); qualitative impairment, with a nonverbal communication score of 13 (cutoff for autism=7); and repetitive behaviors and stereotyped patterns score of 7 (cutoff for autism=3). In addition, he showed evidence of first symptoms at <36 months old. His score of 40 for the Childhood Autism Rating Scale evaluation fell within the severe range of autism. He met all DSM–IV criteria for autistic disorder. Moreover, the results of the Psychoeducational Profile–Revised indicated a developmental score equivalent to an age of 23 months (10) and confirmed the associated diagnosis of severe mental retardation according to DSM–IV criteria. Physical examinations revealed normal growth parameters, and there were no discernible dysmorphic features. Electroencephalogram and magnetic resonance imaging findings were normal. The patient’s parents were nonconsanguineous and free of any medical and neuropsychiatric disorders. His 8-year-old brother was positive for phenylketonuria and had been treated with a phenylalanine-restricted diet from birth and later developed specific language impairment without mental deficiency. Analysis of the patient’s karyotype revealed no chromosomal disorders. In addition, no abnormalities of phenylalanine or tetrahydrobiopterin levels (phenylalanine hydroxylase cofactor) were detected.

The analysis of the karyotype of the patient and his parents revealed a de novo balanced translocation (46, XY, t [9; 10] [q23;q22]) associated with autistic disorder and severe mental retardation.

Patient carrying the ALA138VAL substitution

The patient was the fourth-born boy of a Tunisian family. He had impairments of reciprocal social interactions and communication skills, lack of spoken language, and poor communicating gestures. He also displayed restricted stereotyped behaviors. The diagnosis of autistic disorder was confirmed by the following: administration of the Autism Diagnostic Interview–Revised (with qualitative impairment), with a reciprocal social interaction score of 34 (cutoff for autism=10); qualitative impairment, with a nonverbal communication score of 15 (cutoff for autism=7); and repetitive behaviors and stereotyped patterns score of 17 (cutoff for autism=3). The patient showed evidence of first symptoms at <36 months old. He met all DSM–IV criteria for autistic disorder, and he had severe mental retardation and seizures. In addition, he had discrete facial dysmorphia with long face and retrognathism, but parental description was not available. Computed tomography (CT) scan results were normal. The child’s parents were nonconsanguineous and free of any medical and neuropsychiatric disorders.

Fluorescence in Situ Hybridization

Fluorescence in situ hybridization analyses were performed on metaphase spreads that were obtained from peripheral white blood cells from an Epstein-Barr virus immortalized cell line from the patient. Selected bacterial artificial chromosome/P1 artificial chromosome clones were obtained from the Children’s Hospital Oakland Research Institute and were biotynilated by nick-translation using the BioNick labeling system (Life Technologies, Gaithersburg, Md.). To identify chromosomes 9 and 10 on the metaphases spreads, we used alpha-satellite or telomeric probes (Qbiogen, Illkirch, France).

Semiquantitative Reverse Transcription-Polymerase Chain Reaction

We used total ribonucleic acid (RNA) extracted from Epstein-Barr virus-transformed lymphoblastoid cells for reverse transcription–polymerase chain reaction according to standard procedures. We amplified a fragment of the KCNMA1 C-deoxyribonucleic acid (cDNA) with a forward primer in exon 9 (5′-GGCTCCTATAGTGCGGTTAGT-3′) and a reverse primer in exon 14 (5′-TGTTTAGCAGATGGGCCTTGTT-3′). In addition, we coamplified a cDNA fragment of the β-actin gene as an internal standard with the following primers: forward primer, 5′-TCATGCCATCCTGCGTCTGGACCT-3′ and reverse primer, 5′-CCGGACTCATCGTACTCCTGCTTG-3′, with an annealing temperature of 60 °C.

Electrophysiology

Whole-cell recordings of K⁺ currents were acquired in lymphoblastoid cell lines obtained from the autistic patients and comparison subjects. The cells were washed and spun three times and suspended in a physiological saline solution composed of the following (in mmol/liter): NaCl=140; KCl=4; CaCl₂=2; MgCl₂=1; NaH₂PO₄=0.33; glucose=11.1; Hepes buffer=10; pH balanced to 7.4 with NaOH. An aliquot of these cells was then placed in a Petri dish containing 2 ml physiological saline solution. Petri dishes with cells were placed on the stage of an Elipse TE-300 Nikon microscope. The cells were superfused with experimental solutions via a parallel pipe system lowered into the vicinity of the cells. Solution exchange around each patched cell was estimated to be less than 10s. Cells were intracellularly perfused with a 400 nM free Ca²⁺ pipette solution (to activate the BK_Ca) containing (in mmol/liter) K-glutamate=125; KCl=20; CaCl₂=0.7; Mg-ATP=1; Mg-GTP=1; EGTA=1; Hepes buffer=10; pH balanced to 7.2 with KOH. Pipette tip resistance ranged between 3 MΩ and 5MΩ.. Macroscopic K⁺ currents were generated by progressive 8mV depolarizing steps (500 msec duration, 5 second intervals) from a constant holding potential of –70 mV. BK_Ca currents were defined as the outward current inhibited by 100 nmol/liter iberiotoxin (IbTx), a blocker of BK_Ca channels (11). The capacitance of each cell was estimated by integrating the capacitive current that was generated by a 10 mV hyperpolarizing pulse after electronic cancellation of pipette-patch capacitance that was provided by the Axopatch 200B amplifier (Axon Instruments, Union City, Calif.). IbTx-sensitive currents were expressed as current density (pA/pF). Voltage clamp protocol and data acquisition were controlled with pClamp V 8.1 software (Axon Instruments). Data were digitized at 20 kHz and recorded with an 80486 computer after filtering with a five-pole Bessel filter at 2 kHz. All experiments were conducted at room temperature. All data are expressed as means and standard deviations. Statistical comparisons between groups were performed with two-way repeated measures analysis of variance with a subsequent Dunnett post hoc analysis test. Significance was set at p<0.05.

Mutation Analysis

Polymerase chain reactions were performed in 50-μL reaction volumes containing 100 ng of the genomic DNA of the patients (10 pM of each primer, 1.25 mM of dNTPs, 0.5 U Taq DNA polymerase [Promega], and 1.5 mM MgCl₂). An initial denaturation of 5 minutes at 94°C was followed by 30 1-minute cycles at 94°C, 1 minute of annealing, a 1-minute extension at 72°C, and a final extension step of 7 minutes at 72°C (see Table 1 for polymerase chain reaction conditions and primers sequence). The products were then checked on a 1.5% agarose gel to verify amplification prior to analysis by denaturing high-performance liquid chromatography using the WAVE 3500 HT system (Transgenomic, Inc., Omaha, Neb.). For each amplified exon, melting profiles and temperatures were predicted by the Transgenomic Navigator software version 1.5.1. Pairs of amplified fragments from patient and comparison subject polymerase chain reaction products were pooled, denatured at 95°C for 3 minutes, and cooled to 40°C in decreasing increments of 0.05°C per second. The products were then injected and eluted with an acetonitrile gradient at a flow rate of 1.5 ml/min, with a mobile phase composed of two buffers (buffer A: 0.1 M triethylammonium acetate; buffer B: 0.1 M triethylammonium acetate with 25% acetonitrile). For each sample pair that exhibited an abnormal elution profile, the polymerase chain reaction products were purified and sequenced on an ABI377 DNA sequencer (Perkin Elmer).

Molecular Analysis

Fluorescent in situ hybridization studies with bacterial artificial chromosome clones obtained from the Children’s Hospital Oakland Research Institute library enabled us to map the 9q23 breakpoint to RP11-85P23 (accession number AL161632) and the 10q22 breakpoint to RP11-619F23 (accession number AL627447) (part A of Figure 1). Sequence analysis of the two breakpoint areas revealed 1) the absence of a gene in the 9q23 region and 2) localization of the KCNMA1 gene (NCBI NM_002227, MIM number [*600150]), encoding for a large conductance calcium-activated K⁺ (BK_Ca) channel in the 10q22 region (part B of Figure 1) (12). Polymerase chain reaction analyses on the RP11-619F23 clone indicated that the 10q breakpoint was located in the first intron. Molecular analyses of the genomic sequence spanning the two breakpoints revealed a 14 bp deletion (part C of Figure 1) that is located in large noncoding regions.

Since the initiation codon (ATG) of the KCNMA1 gene is located in the first exon, we hypothesized that the physical separation could cause an altered level of KCNMA1 transcript. We then performed a semiquantitative reverse transcription-polymerase chain reaction on total RNA isolated from lymphoblastoid cell lines of both the patient and a healthy comparison subject to verify that the translocation led to a lower expression of the KCNMA1 gene. Comparative analysis of the bands intensity with Bioprofil software (Vilber Lourmat, Marne La Vallee, France) between β-actin and KCNMA1 showed a KCNMA1:β-actin ratio of 1.0 in the comparison subject and 0.55 in the patient (data not shown), suggesting the haploinsufficiency of the KCNMA1 gene.

Functional Studies

To verify that the reduced mRNA coding for KCNMA1 corresponded to similar reduced activity of the BK_Ca channels, patch-clamp studies were performed in the patient’s lymphoblastoid cell lines (Figure 2). Whole-cell currents were elicited in the presence of 400 nM Ca²⁺ inside the cells by 8 mV depolarizing steps from −70 to +42 mV before and after blocking BK_Ca channels with a specific blocker, Iberiotoxin. Whole cell recordings of K⁺ currents were acquired in lymphoblastoid cell lines obtained from the autistic patient and the comparison subjects. While membrane capacitance was similar in both cell groups (16.3 pF [SD=0.7] and 17 pF [SD=0.9] in the comparison subjects and the patient, respectively), resting membrane potential and input membrane resistance were significantly different in the two cell groups (p<0.05). The changes were in agreement with reduced activity of a K⁺ channel. Average resting membrane potential was –72.8 mV (SD=0.9) in the comparison subjects’ cells (N=12 cells) and –68.4 mV (SD=1.7) in the patient cells (N=12 cells), and average input resistance was 509 MΩ (SD=76) in the comparison subjects’ cells and 744 MΩ (SD=57) in the patient cells. The density of the IbTx-sensitive current in the patient’s cells was approximately one-half that of the comparison subjects’ cells, in accordance with reduced expression of mRNA coding for KCNMA1 (Figure 2).

These results suggested that one copy of the KCNMA1 gene was silenced in the autistic patient. Interestingly, the addition of the Ca²⁺-sensitive BK_Ca opener, BMS-204532, a drug previously tested in patients with ischemic stroke (13), significantly increased the activity of the remaining channels (Figure 2) in the patient’s cells. To study the effects of BMS-204532, cells were first submitted to 4-AP to block most of the Kv channels found in the cells and then to BMS-204532 and 4-AP to observe the specific activation of the BK_Ca channels.

Mutation Analysis

To verify that KCNMA1 is a possible candidate gene for autistic disorder, we investigated mutations in 116 Fragile X-negative patients with autism and a normal karyotype (46 Tunisian subjects, 40 Italian subjects, and 30 French subjects) by single-strand conformation polymorphism and denaturing high-performance liquid chromatography analyses. In one patient, we found one base-pair substitution in the second exon of KCNMA1, changing an alanine to a valine, located in the first loop of the protein (part A of Figure 3). This sequence variation was not found in 225 Tunisian comparison subjects (450 chromosomes), and the confirmation of its absence in the other 115 subjects was verified by further denaturing high-performance liquid chromatography analyses. This intracellular region is highly conserved from C.elegans to mammals (part B of Figure 3), suggesting an important function. Furthermore, analysis of sequence motifs revealed that the corresponding heterozygous substitution might create a cryptic splice donor site in the second exon. The C-T exchange corresponded to position ⁺2 of a newly predicted splicing signal AGgtggat that matched the consensus sequence almost perfectly, which was highly predicted (probability of 0.83) by the splice-prediction program Splice-View (http://125.itba.mi.cnr.it/~webgene/wwwspliceview.html). However, we were unable to test the functional consequences of the sequence variation in the KCNMA1 protein, since we had access to DNA only because of the intercurrent death (road accident) of the patient and his mother. Further functional analyses by site-directed mutagenesis would clarify the potential functional consequences of such a sequence variation.

There have been a number of reports in the literature of chromosome aberrations in autism covering a broad spectrum of anomalies, including terminal and interstitial deletions, balanced and unbalanced translocations, and inversions (2, 5). However, particularly in the case of balanced translocations, none of these studies provide evidence of a functional defect associated with autism. In this study, we have presented arguments suggesting that the KCNMA1 gene is a strong candidate for autism and mental retardation.

Indeed, the KCNMA1 gene was physically disrupted on chromosome 10 by a balanced reciprocal translocation in a patient with autistic disorder and severe mental deficiency. The physical separation of the KCNMA1 gene promoter of the first exon from the remainder of the gene leads to a nonfunctional KCNMA1 allele (as shown in semiquantitative reverse transcription-polymerase chain reaction experiments). Although we found an associated 14 bp deletion at the breakpoints located in large noncoding regions, it is unlikely to have an effect on the decreased expression of the KCNMA1 transcript.

The 10q22 region has previously been suspected of being involved in autism and language disorders, with weak possible linkage to D10S201 (14) and D10S2327 (15), located at 1.5 Mb and 1 Mb telomeric from KCNMA1, respectively. The mapping of an autism susceptibility locus and a gene physically disrupted in the same chromosome region in an autistic patient suggest that this gene may be involved in this condition.

Large conductance calcium-activated potassium channels (BK_Ca channels) are ubiquitously distributed among tissues, except in cardiac myocytes, and activated by membrane depolarization and increases of intracellular calcium concentration. They play a key role in the control of neuronal excitability, hormone secretion, and smooth muscle tone. Therefore, BK_Ca channels are considered to be molecular integrators of biochemical and electrical signals.

The KCNMA1 gene is expressed abundantly in various human brain structures (e.g., amygdala, caudate nucleus, cerebral cortex, hippocampus, hypothalamus, spinal cord) (12) as well as in the cochlea (16). In the vertebrate brain, the BK_Ca channels are mainly located in the glutamatergic presynaptic active zone and might regulate synaptic transmission and transmitter release (17). BK_Ca channels are known to associate with β-subunit through the N-terminus, leading to specific regulation of channel function (18). BK_Ca channels have also been shown to associate with other proteins through the intracellular C-terminus region. In the brain, β2 adrenergic receptors form a macromolecular complex with BK_Ca, specifically to regulate channel activity (19). Other yeast two-hybrid screens have shown an interaction between BK_Ca and 1) the light chain of the microtubule-associated protein 1A (MAP1A) which, with microtubules, is an important determinant of neuronal and nonneuronal cellular morphology (20); 2) β-catenin, which is involved in synapse organization and mediating cell adhesion (β-catenin associates with the synaptic protein Lin7/Velis/MALS, whose interaction partner, Lin2/CASK, also binds voltage-gated Ca²⁺ channels; thus, β-catenin could provide a physical link between Ca²⁺ and BK_Ca channels at the presynaptic active zone) (21); and 3) syntaxin 1A, which is a soluble N-ethylmaleimide sensitive fusion attachment receptor protein required for the regulation of the activity and/or distribution of synaptic proteins within specialized cellular compartments (22).

Mutations in the homologous D. melanogaster and C. elegans of the KCNMA1 gene (Slowpoke and Slo-1, respectively) have been shown to cause both electrophysiological defects and multiple behavioral abnormalities (23, 24). The recent report (25) on mice lacking BK_Ca channels (-/-) showed that they presented cerebellar dysfunction and a dramatic reduction in the activity of the cerebellar Purkinje neurons. These data are very interesting, since many neuropathological observations on autistic individuals have revealed specific alterations of the Purkinje cells (26).

Therefore, the finding in our study of a breakpoint in the KCNMA1 gene in the autistic patient with a de novo translocation, together with the predominant expression of KCNMA1 in the brain and during development, its chromosomal localization in a locus previously implicated in autism, and the finding of a sequence variation in an unrelated patient in a highly conserved region of the ion channel suggest that the KCNMA1 gene is a candidate for autistic disorder in humans. Furthermore, a recent report (27) described a Ca_V1.2 calcium channel dysfunction associated with autism, also suggesting the implication of Ca²⁺ signaling in autism. Finally, Du and colleagues (28) recently reported a gain-of-function mutation in KCNMA1 associated with generalized epilepsy and paroxysmal movement disorder in a large family. Thus, alterations in neuronal excitability may contribute to different neurological disorders. Many potassium channel dysfunctions are associated with a range of neurological disorders, and drugs that target these channels might hold promise for clinical applications.

In summary, we report a channelopathy associated with autism and mental retardation, and we propose that a defect in BK_Ca activity may abolish neuronal excitability. Future studies, including mutation analyses for KCNMA1 in large cohorts of autistic subjects, are needed as well as association studies with single nucleotide polymorphisms and screening for genomic rearrangements (i.e., deletions, duplications) located in the KCNMA1 genomic area. Finally, we aim to perform the functional study of the identified sequence variations (particularly the ALA138VAL substitution) with the reconstitution of the potentially mutant channel in appropriate heterologous cellular systems. This prospective work will evaluate a potential broader involvement of the KCNMA1 gene in autism and mental retardation and, more generally, help us to better understand the consequences of such synaptic defects associated with these psychiatric disorders.