Recurrent Rearrangements in Synaptic and Neurodevelopmental Genes and Shared Biologic Pathways in Schizophrenia, Autism, and Mental Retardation

Context  Results of comparative genomic hybridization studies have suggested that rare copy number variations (CNVs) at numerous loci are involved in the cause of mental retardation, autism spectrum disorders, and schizophrenia.

Objectives  To provide an estimate of the collective frequency of a set of recurrent or overlapping CNVs in 3 different groups of cases compared with healthy control subjects and to assess whether each CNV is present in more than 1 clinical category.

Design  Case-control study.

Setting  Academic research.

Participants  We investigated 28 candidate loci previously identified by comparative genomic hybridization studies for gene dosage alteration in 247 cases with mental retardation, in 260 cases with autism spectrum disorders, in 236 cases with schizophrenia or schizoaffective disorder, and in 236 controls.

Main Outcome Measures  Collective and individual frequencies of the analyzed CNVs in cases compared with controls.

Results  Recurrent or overlapping CNVs were found in cases at 39.3% of the selected loci. The collective frequency of CNVs at these loci is significantly increased in cases with autism, in cases with schizophrenia, and in cases with mental retardation compared with controls (P < .001, P = .01, and P = .001, respectively, Fisher exact test). Individual significance (P = .02 without correction for multiple testing) was reached for the association between autism and a 350-kilobase deletion located at 22q11 and spanning the PRODH and DGCR6 genes.

Conclusions  Weakly to moderately recurrent CNVs (transmitted or occurring de novo) seem to be causative or contributory factors for these diseases. Most of these CNVs (which contain genes involved in neurotransmission or in synapse formation and maintenance) are present in the 3 pathologic conditions (schizophrenia, autism, and mental retardation), supporting the existence of shared biologic pathways in these neurodevelopmental disorders.

The development of microarray-based technologies for comparative genomic hybridization (array-CGH) analysis has enabled the detection of submicroscopic microdeletions or microduplications, also referred to as copy number variations (CNVs). Recently, this approach has been widely used in neurologic and psychiatric disorders, including mental retardation (MR),^1-3 autism spectrum disorders (ASDs),^4-7 and schizophrenia.^8-11 Findings from these studies suggested that several genes involved in similar neurodevelopmental pathways may be associated with these conditions. However, so far only rare structural variants, sometimes present in a single case, have been identified. Therefore, it is difficult to decipher which of these variations are causative, which are risk factors, and which are only rare polymorphisms unrelated to any pathologic phenotype. De novo rearrangements are usually considered pathogenic, but this argument (which is acceptable for rare large rearrangements detectable by conventional cytogenetics) should be considered with caution for smaller CNVs, for which a high mutation rate is expected. Indeed, it has been estimated that a de novo segmental deletion occurs in 1 per 8 newborns and a de novo segmental duplication in 1 per 50 newborns, with most of these rearrangements being benign polymorphic variants.¹² Therefore, the disease association of CNVs has to be tested systematically by comparing the frequency of each candidate CNV in cases and in healthy control subjects. Given the low frequency of each CNV, this would require the study of huge series, achievable only in the context of forthcoming meta-analyses. Other problems arise because ascertainment of most of the published samples, initially recruited for linkage studies, is biased toward multiplex cases and because control samples, when present, are generally composed of subjects not screened for the studied pathologic conditions. The objectives of the present study were (1) to provide an estimate of the collective frequency of a set of recurrent or overlapping CNVs in 3 different groups of cases compared with controls and (2) to assess whether each CNV is present in more than 1 clinical category.

Cases with schizophrenia and cases with MR were ascertained at University Hospital, Rouen, France, from consecutive hospitalizations in patients with schizophrenia or from consecutive referrals for phenotypic and genetic investigations in patients with intellectual disability. The ASD sample included cases ascertained from consecutive consultations at 4 units specializing in autism diagnosis and evaluation located in Rouen, Tours, and Dijon (France) and in Messina (Italy), as well as cases directly referred by the French Autism Foundation. Controls, all ascertained at University Hospital, Rouen, were screened using a standardized data sheet derived from the Schedule for Affective Disorders and Schizophrenia¹³ and were required to be free of any psychotic disorder or MR themselves or in their first-degree relatives. All psychiatric diagnoses were established according to DSM-IV criteria following review of case notes and direct examination of cases. The Schedule for Affective Disorders and Schizophrenia¹³ was used for the clinical assessment of all cases with schizophrenia or schizoaffective disorder. The Autism Diagnostic Interview–Revised,¹⁴ the Autism Diagnostic Observation Schedule–Generic,¹⁵ or the Childhood Autism Rating Scale¹⁶ was used for 83.0% of cases with ASDs (100.0% of cases having ASDs with CNVs). Evaluation of IQs was performed using standardized neuropsychological tests (ie, validated mental age–appropriate Weschler scales [Wechsler Preschool and Primary Scale of Intelligence, Wechsler Intelligence Scale for Children, or Wechsler Adult Intelligence Scale]).

The schizophrenia group included 189 cases with schizophrenia and 47 cases with schizoaffective disorder. Postmorbid IQs were available for two-thirds of cases with schizophrenia; 18.0% of these cases had an IQ lower than 70. The ASD group included 257 cases with autism and 3 cases with Asperger syndrome. The MR group included 235 cases with MR and 12 cases with developmental language disorder. All cases with MR and two-thirds of cases with ASDs were examined by an experienced clinical geneticist (A.G., V.D.-G., V.L., F.B.-B., S.O., L.F., or G.D.R.) and were screened for fragile X mutation and karyotype abnormalities. Cases with large chromosomal anomalies, fragile X syndrome, or other established syndromes were excluded. Cases with common environmental causes of MR such as fetal alcohol syndrome or birth complications were also excluded. Additional clinical features, including intrauterine or postnatal growth retardation and dysmorphic features or malformations, were present in 8.5% of cases with ASDs and in 62.0% of cases with MR. Demographic characteristics of the sample, including 979 unrelated white non-Hispanic subjects from France or Italy, are summarized in Table 1.

After written informed consent, blood samples were drawn from all included participants and whenever possible from parents and affected relatives of cases. Ethics committee approval was obtained from all regions where families were recruited.

A MEDLINE search using the terms CNV, schizophrenia, autism, and mental retardation allowed us to select nonexhaustively a set of 28 loci with microrearrangements characterized by prior array-CGH analyses, often in a single case. This set included major candidate CNV loci identified in cases with ASDs and schizophrenia before April 2008, as well as 8 functionally related CNV loci identified in MR (Table 2). Each locus generally contained a single disease–associated CNV, but in some cases, overlapping CNVs with different boundaries had been described in cases. The gene content of these loci ranged from 1 to 28. At each locus, at least 1 candidate gene had been previously suggested in the seminal studies and was retained for the present analysis. Functionally, most of these candidate genes can be classified in 2 main categories related to synapse formation and maintenance or to neurotransmission.

Copy number variation at each locus was assessed by quantitative multiplex polymerase chain reaction (PCR) of short fluorescent fragments (QMPSF), a method based on the simultaneous amplification of several short genomic fragments under quantitative conditions.⁴⁴ For each locus, amplicons were designed in the coding sequence of selected candidate genes. All assays were grouped in 3 multiplex PCR experiments that included 10 short genomic fragments (range, 100-301 base pair) each. Primer sequences and PCR conditions are summarized in eTable 1(http://www.archgenpsychiatry.com. DNA fragments generated by QMPSF were separated on a sequencer (ABI Prism 3100; Applied Biosystems, Norwalk, Connecticut), and the resulting fluorescence profiles were analyzed using commercially available software (Gene Scan 3.7 software; Applied Biosystems). For each case, the QMPSF profile was superimposed on that generated from a control by adjusting the same level that the peak obtained for a control amplicon corresponding to a short exonic fragment of the PBGD gene. When a CNV was detected, further analyses aiming to confirm and delineate the size of the rearrangements were performed using additional dedicated QMPSF assays (eFigure 1), array-CGH, or fluorescence in situ hybridization analyses.

Oligonucleotide array-CGH was performed using a commercially available array (human genome CGH microarray 4 × 44 K) (Agilent Technologies, Santa Clara, California). This array contains 60-mer oligonucleotide probes (n = 44 290) covering the whole genome, with a mean spatial resolution of approximately 30 to 35 kilobases (kb). Eighty-four percent of the probes reside in intragenic regions, and more than 30 000 genes are each represented by at least 1 probe. All experiments were performed using the June 2006 version of the protocol (version 4.0, Agilent Oligonucleotide Array-Based CGH for Genomic DNA Analysis; Agilent Technologies).

Fluorescence in situ hybridization analyses were performed on metaphase spreads obtained from peripheral lymphocytes from the cases. Selected human genomic bacterial artificial chromosome clones were obtained from a distribution center (BACPAC Resources Center, Oakland, California [http://bacpac.chori.org]).

Sequence analysis of the coding exons of the PRODH (OMIM_606810)) gene was performed using primers and PCR conditions previously described⁴⁵ via an automated sequencer (model 3100; Applied Biosystems). Paternity was checked by microsatellite typing.

Plasma proline levels in cases were determined after overnight fasting. All samples were analyzed using ion exchange chromatography (LC 3000 system; Biotronik, Maintal, Germany).

Categorical variables were compared using the Fisher exact test. Two hypotheses were tested. First, the distribution of the collective set of recurrent or overlapping CNVs found in each disease group of cases was compared with that found in controls (3 tests). Second, the distribution of each recurrent or overlapping CNV present in our population was compared between each disease group of cases and controls (33 tests). P values are reported without Bonferroni correction.

Among 743 cases, the proportion of recurrent or overlapping CNVs identified among 28 selected loci (Table 2) was 11 of 28 (39.3%). Their collective frequency was 10 per 236 cases with schizophrenia (4.2%), 16 per 260 cases with ASDs (6.2%), and 13 per 247 cases with MR (5.3%) vs 1 per 236 controls (0.4%), demonstrating a significant excess of these CNVs in each disease group compared with controls (P = .01, P < .001, and P = .001, respectively, Fisher exact test) (Table3, eTable 2, and eTable 3). None of the cases had more than 1 of the 28 CNVs. Only 1 CNV identical to a previously described disease-associated CNV (ie, a 350-kb deletion located at 22q11 spanning the PRODH and DGCR6 (OMIM_601279)) genes)¹⁷ was detected in the control group. This deletion, present in a single control, had a low frequency (1 per 236 controls) similar to that previously reported in Japanese⁴⁶ and Canadian⁴⁷ populations. Individual significance for the association with ASDs was reached for this PRODH/DGCR6 deletion (9 per 260 cases with ASDs vs 1 per 236 controls [P = .02]).

Among the 4 most prevalent CNVs, 3 (located at 22q11, 16p11, and 15q13) were flanked by known regions of segmental duplication and resulted most likely from a nonallelic homologous recombination mechanism. At the 2p16 locus, the NRXN1 gene was recurrently disrupted by a set of partially overlapping deletions spanning the promoter and first exons of neurexin 1α or the exons coding for the middle section of this protein, as well as for the proximal region of neurexin 1β. These rearrangements occurred in a region devoid of any segmental duplication and resulted from another mechanism distinct from a nonallelic homologous recombination mechanism.

Among 27 families in which transmission was tested (69.2% of families with CNVs), 8 CNVs (located at 8p23, 15q11-q13, 15q13, 16p11, and 22q13) had occurred de novo (Table 3). The mean (SD) paternal age was not significantly different between families with de novo and inherited CNVs (27.2 [4.7] vs 30.7 [4.6] years; P = .19, Mann-Whitney test). In most families, CNVs were transmitted from an apparently nonaffected (although not clinically or neuropsychologically assessed) parent. This includes a partial duplication of the X-linked GRIA3 gene, present in a young male case with autism, which was inherited from the nonaffected mother. The 350-kb deletion located at 22q11 spanning the PRODH/DGCR6 locus was also transmitted in 11 of 11 tested cases. PRODH encodes for proline dehydrogenase, and PRODH deficiency is responsible for type 1 hyperprolinemia, a condition often associated with cognitive impairment and with psychotic symptoms.⁴⁵ However, hemizygous deletion of the PRODH gene is insufficient per se to result in hyperprolinemia, as only 35% to 50% of cases with velocardiofacial syndrome, all bearing a single copy of PRODH, exhibit hyperprolinemia.^45,48 Indeed, a reduction of more than 50% of the enzymatic activity is generally required to produce hyperprolinemia.⁴⁵ Therefore, the presence of a mutation affecting enzyme activity⁴⁹ on the second allele is necessary. To examine this issue, the remaining PRODH allele was sequenced in all cases bearing the 350-kb deletion, and the plasma proline level was assessed whenever possible. As summarized in Table 4, 14 of 15 cases harbored a genotype predicted to result in a reduction of at least 70% of enzymatic activity. Among 12 cases from whom the plasma proline level was assessed, 9 had mild to severe hyperprolinemia, and 3 had plasma proline levels at the upper boundary of normal values.

Cosegregation of the CNV with pathologic conditions was examined in 4 multiplex families in which DNA from affected siblings was available. In family 144 (Tables 3 and 4), the 2 sibs with schizophrenia or schizoaffective disorder harbored the PRODH/DGCR6 deletion. In family 11 695 (Table 3), the 2 MR sibs harbored the 2p16.3 deletion. In family 14 390 (Table 3), the 16p11 deletion was present in the proband with developmental language disorder but not in his sib with MR. In family 33 (Table 3), the 16p11 duplication was present in 2 sibs with schizophrenia and in a nonaffected sibling but was absent in a third sibling with schizoaffective disorder (eFigure 2).

Combining the results of the present study and previous findings, none of the observed rearrangements were disease specific, with the possible exception of the maternally derived 15q13 duplication associated with ASDs. The 22q11¹⁷ and 2p16¹ deletions were found in the 3 conditions, whereas the 22q13 deletion in 2 cases with ASDs had already been described in ASDs and in MR.^4,18-20 The 16p11 and 8p23 rearrangements previously described in ASDs^4,5 were found in cases with schizophrenia, and the 16p11 rearrangement was found in cases with MR. The Xp11.4 duplication spanning the TSPAN7 gene, previously described in MR and in ASDs,^4,21 was found in a case with schizophrenia, as well as the 17q21 duplication previously described in a patient with MR.²² Two different-sized 15q13 duplications that included APBA2 were found in 1 case with ASDs and in 1 case with schizophrenia. A partial duplication of the GRIA3 gene, including the promoter region and the exons coding for the proximal region of GRIA3, was detected in a single case with autism. Although slightly different based on its size and the number of duplicated exons, this partial duplication is reminiscent of that recently reported in a patient with MR.²³ At the 8p23 locus, gain and loss of material were found, as well as at the 16p11 and 17q21 loci, as recently described.^22,24,25 This suggests that dosage-sensitive genes, whose expression is finely tuned, are located within these rearranged segments.

From a phenotypic viewpoint, 3 of 9 cases with schizophrenia bearing a candidate CNV had mild MR in an IQ assessment obtained after the onset of their psychotic symptoms (Table 3). No premorbid IQ was available for any case. Although postmorbid IQ likely constitutes an underestimation of the premorbid level of cognitive functioning in cases with schizophrenia, cognitive deficits manifest by severe learning disorders and failure to follow normal schooling were already noted in these 3 cases during childhood before the onset of their psychotic symptoms, supporting the comorbid diagnosis of MR and schizophrenia. In cases with potential MR (case 144.1 [Table 3] and cases 313 and 33.1 [Table 3]), psychotic features appeared at ages 25, 18, and 27 years, respectively, and included prominent positive symptoms such as persecutory delusions, thought insertions, delusions of being controlled, and auditory hallucinations with voices making insulting statements. The 3 cases had marked behavioral disorders such as aggressiveness and psychomotor agitation. Mood instability and suicidal ideation were present in case 144.1. The 3 cases were considered good responders to atypical neuroleptic drug therapy. Initial symptoms gradually declined, and the course of the disease was marked by social isolation, blunted affects, and loosening of associations. These cases are living in long-term institutions.

Except for 2 cases who had normal cognitive functioning (high-functioning autism), all tested CNV-bearing cases with autism had IQs in the range of MR, although case T35 had only mild cognitive dysfunction (Table 3). In 2 cases with high-functioning autism (cases 12 746 [Tables 3 and 4] and 44 813 [Table 3], aged 6 and 8 years, respectively), onset was in the first year of life, when the parents noticed few gestures, almost no babbling, and poor shifting gazes. Subsequently, the cases acquired spoken language, although with significant delay. They are able to carry on conversation, carry out commands, imitate, and dress and groom themselves. They participate in public school with adapted educative programs. However, they remain impaired in their communicative and social skills, and their use of language is often inappropriate. Both cases have developed ritualistic behaviors and show restricted patterns of interest.

In addition to the set of CNVs tested for recurrence in this study, 2 reciprocal rearrangements, previously unassociated with any psychiatric condition, were incidentally observed at 2 loci. Both were benign polymorphisms. A common CNV reciprocal to the expected one (ie, a 350-kb duplication) was found at the 22 q11 locus in 6 of 236 controls, as well as in 7 of 236 cases with schizophrenia, 4 of 260 cases with ASDs, and 9 of 247 cases with MR. Another overlapping reciprocal CNV (ie, a 490-kb duplication) was detected at the CHRNA7 locus located on chromosome 15q13.3. This CNV was present in 2 of 236 controls, in 1 of 236 cases with schizophrenia, in 1 of 260 cases with ASDs, and in 1 of 247 cases with MR and was unrelated to any pathologic condition (eTable 2).

After a first wave of CNV discovery by array-CGH analyses in neuropsychiatric disorders, this study for the first time (to our knowledge) examines the involvement of a limited number of candidate loci in large samples of cases with different clinical diagnoses. Two strengths of our study design are (1) the inclusion of controls carefully screened for the studied pathologic conditions and of series of cases mostly ascertained through consecutive admissions or consultations and, therefore, (2) the inclusion of cases belonging to simplex or multiplex families. Given the expected rarity of each variant, our first goal was not to test the association of every individual CNV with schizophrenia, ASDs, or MR but to determine whether these variants were collectively more frequent in cases with these diseases than among controls. This aim was successfully achieved, and we were able to obtain suggestive statistical significance for the association between the 350-kb deletion located at 22q11 and ASDs. This deleted segment, located within the chromosomal region deleted in velocardiofacial syndrome (a contiguous gene syndrome known to be associated with a high frequency of MR, ASDs, and psychosis), contained the 2 genes PRODH and DGCR6. Although we cannot exclude an involvement of DGCR6 in the neuropsychiatric phenotype of the cases bearing this CNV, previous work from our group strongly suggests that PRODH is the prime candidate.⁴⁵ It was previously shown that hyperprolinemia, resulting from partial or total inactivation of this enzyme, (1) may lead to MR and autism in patients with type 1 hyperprolinemia,⁴⁵ (2) is a risk factor for schizoaffective disorder,⁵⁰ and (3) is inversely correlated with IQ in velocardiofacial syndrome.⁴⁵ Herein, we show that all cases except 1 harboring this deletion were compound heterozygotes, also bearing mutations affecting enzymatic activity on the second allele. This resulted in a loss of at least 70% of the predicted PRODH residual activity in 14 of 15 assessed cases and resulted in hyperprolinemia in 9 of 12 assessed cases.

Second, we show that de novo CNVs and CNVs inherited from an apparently healthy parent can be found in cases. For transmitted CNVs, the mode of inheritance of the disease was recessive in some cases (eg, hyperprolinemia related to the 22q11 deletion) or implied the transmission of an X-linked gene (GRIA3) by a woman to her son. Consistent with findings in previous studies,^4,8,26 the 16p11 rearrangements were inherited from an apparently nonaffected parent in 3 families. These CNVs, whose estimated frequency in the Icelandic population was 5 per 18 834 (0.03%) for the duplication and 2 per 18 834 (0.01%) for the deletion,²⁵ should be considered risk factors rather than fully causative variations. The presence of affected siblings that do not share the CNV, already noted in a previous study,²⁶ does not necessarily rule out the causative implication of these CNVs but raises the question of intrafamilial genetic heterogeneity. This hypothesis, which is plausible for these frequent disorders that are often characterized by assortative mating, remains speculative because the parents and their relatives were not psychiatrically or cognitively assessed in these families.

Third and most important, our study confirms and extends recent evidence suggesting that many candidate CNVs are not disease specific but are involved in the expression of different behavioral phenotypes, including MR, ASDs, and schizophrenia. This implies the existence of shared biologic pathways in these 3 neurodevelopmental conditions. These pathways chiefly affect synapse formation and maintenance, as well as neurotransmission (with a special emphasis on glutamate and γ-aminobutyric acid). The dysfunction of specific neuronal networks underlying the particular symptoms of each clinical condition most likely depends on additional genetics, epigenetics, and environmental factors that remain to be characterized. From a clinical point of view, despite the diversity of categorical diagnoses, many cases harboring these CNVs shared some clinical features: one-third of cases with schizophrenia and 83.3% of cases with autism having CNVs had a level of cognitive functioning in the range of MR. This is in accord with previous studies showing that point prevalence of schizophrenia is increased by a factor of 3 in cases with intellectual disabilities⁵¹ and that 50% of cases with autism have MR.⁵² However, the following 2 caveats should be noted: (1) because attention and communication are markedly impaired in children with autism, assessment of their IQs (even performance IQs in nonverbal cases) is unreliable, and (2) these results were not obtained in a single community–based population but in 3 disease groups ascertained according to different schemes, a factor whose effect is difficult to appreciate but which is likely to have implications related to the phenotypic severity in these cases.

Fourth, targeted procedures for CNV analysis such as the QMPSF method is a cost-effective alternative to array-CGH for the screening of candidate loci in large case-control cohorts. We plan to conduct extensive resequencing of these candidate genes to further validate their role in these conditions.

Since our submission of this article for publication, additional studies^9,11,53,54 have been published documenting shared CNVs between MR, ASDs, and schizophrenia.

Correspondence: Dominique Campion, MD, PhD, Institut National de la Santé et de la Recherche Médicale, Unité 614, Institut Hospitalo-Universitaire de Recherche Biomédicale, 22 Blvd Gambetta, 76000 Rouen, France (dominique.campion@univ-rouen.fr).

Submitted for Publication: August 5, 2008; final revision received February 6, 2009; accepted February 23, 2009.

Author Contributions: Dr Campion had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Financial Disclosure: None reported.

Funding/Support: This study was supported by the Fondation de France, by Collaborative Biologic Resources From the Autism Foundation (RBCFA), and by the French Autism Foundation. Dr Guilmatre received a fellowship from Region Haute Normandie, and Dr Mosca received a fellowship from the Académie Nationale de Médecine.

Role of the Sponsors: We acknowledge the usefulness of the available resources from the RBCFA and from the families involved in the RBCFA. RBCFA is a program of the French Autism Foundation, which is recognized to be of public usefulness, was created by parents of autistic patients, and whose scientific director is Sylvain Briault, MD, PhD, researcher at the INSERM.

Additional Contributions: Technical support for the transcriptomic platform was provided by OUEST Genopole and by Genethon. We thank our colleagues who helped to identify patients, as well as the patients and their families for their participation.