RARE DE NOVO AND TRANSMITTED COPY NUMBER VARIATIONS IN AUTISM SPECTRUM DISORDERS: IMPLICATIONS FOR FUNCTIONAL NETWORKS OF GENES INVOLVED IN NEUROGENESIS, NEURONAL METABOLISM, SYNAPTIC FUNCTION, NEUROIMMUNITY, INTRACELLULAR SIGNALING AND CHROMATIN REMODELING

Autistic behaviors were independently identified as recognizable syndromes in the early 20th century by Heller and, subsequently, by Kanner and Asperger. The autism diagnosis spans a broad continuum of what are collectively known as Autism Spectrum Disorders (ASDs) or Pervasive Developmental Disorders (PDDs). ASDs include several conditions, namely full-syndrome autism (Autistic Disorder or Idiopathic Autism), Childhood Disintegrative Disorder, Asperger Syndrome (AS) and Pervasive Developmental Disorder Not Otherwise Specified (PDD-NOS). The latest estimates put the population prevalence of ASDs at approximately 1 in 110. Furthermore, ASDs show a 4:1 male to female gender bias, which may rise to 11:1 when considering Asperger disorder. Although heterogeneous, ASDs are united by a combination of three core behavior symptoms: a) impaired language and communication; b) deficiencies in social interaction; c) restricted interest and repetitive stereotypic behavior. Symptoms of ASDs usually begin in early childhood with evidence of delayed development before age 3 years. The etiology of ASDs is complex and encompasses the roles of genes, the mitochondria, the environment, and the immune system. However, there is strong evidence for the importance of complex genetic factors comprised of different form of genetic variations in the etiology of ASDs. Twin and family studies have, indeed, established the preponderant genetic basis of autism and indicate that the heritability of autism is over 90%, which is the highest heritability value so far associated with a neuropsychiatric disorder. The genetic causes of ASDs can be classified as: (1) ASD-related monogenic syndromes; (2) rare chromosomal abnormalities; (3) rare copy number variations (CNVs); (4) rare gene mutations; (5) common genetic variants. Recently, submicroscopic CNVs, de novo and inherited, are emerging as an important category of genetic risk for ASDs with a different impact depending on the type of CNV identified. Screening for CNVs by means of array CGH and SNP array technologies has proven to be one of the more successful strategies for the discovery of ASD candidate loci over the past five years. Furthermore, the same or overlapping CNVs are being identified as risk factors across a few neurodevelopmental disorders, indicating that some ASD loci are likely pleiotropic with variable expressivity. The increased resolution of the array-based approaches suggests that the proportion of ASD cases (both idiopathic and syndromic) that may be ultimately attributed to a rare structural variant is around 10-20%. The de novo CNV rate in ASDs is roughly three to seven times than that in controls and has been reported to be higher in simplex (low-risk) compared to multiplex families (high-risk). Furthermore, CNV screening gave rise to a paradigm shift away from a common variant model of ASD genetic architecture (based on low penetrating variants) to one suggesting a role for multiple rare and distinct genetic risk factors (with a higher penetrance), known as oligogenic heterozygosity model, which does not exclude, however, a modulation of the phenotype by common susceptibility genetic variants. We collected a series of 115 patients (92 males and 23 females) who were diagnosed with ASDs: 41 with idiopathic Autism, 46 with PDD-NOS, 15 with Syndromic Autism, 8 with High-Functioning Autism, and 4 with Asperger syndrome. Genomic DNA from all patients was used to perform array CGH analysis (Agilent Technology) in order to detect CNVs. In particular, 18 patients were analyzed at a lower resolution (44K or 60K Kits) and 97 at a higher resolution (244K Kit). In the event of detection of rare CNVs, if possible, patients’ parents were analyzed to characterize the origin of the unbalanced microrearrangement (de novo or inherited). In 63 of 115 patients (55%), rare CNVs (one or more) were detected that were not already reported in healthy subjects according to the Database of Genomic Variants (DGV). This group comprised 49 patients analyzed by the Agilent 244K Kit (detection rate ~50.5%) and 14 by the 44K or the 60K Kits (detection rate 77.8%); 51 sporadic and 12 familial cases. Overall, 120 rare CNVs were detected, 73 gains (60.8%) and 47 losses (39.2%), ranging from 10 kb to 11 Mb in size. Furthermore, inheritance is unknown for 13 CNVs (10.8%). Twenty of the remaining 107 CNVs were de novo (16.7%), and 87 were inherited (72.5%), 50 from the mother (57.5%) and 37 from the father (42.5%). The ad hoc analysis of the rare CNV gene content using databases revealed a total of 276 genes that were considered good candidates for ASDs. A small percentage of the selected genes (~11%) have been previously reported as causative genes based on mutations and/or CNVs, often de novo, that have been described in autistic patients. Moreover, SNPs in a very few genes (~5%) have been significantly associated with ASDs. However, most of the proposed candidate genes (54%) have not been previously reported in association with ASDs, and a significant percentage (30.5%) have been involved in some microdeletion/microduplication syndromes that are comorbid with ASDs. On the basis of gene expression data, function, and pathway of action, all the selected genes contribute to CNS neurodevelopment and maintenance, acting during embryonic and foetal development as well as in the early postnatal period and, in some cases, in adult life. Sixty-six of the 276 selected genes are implicated in neurogenesis and neurodevelopment (24%), 27 in CNS metabolism (10%), 29 in synaptogenesis and synaptic plasticity (10.5%), 19 in CNS development, homeostasis, and immunosurveillance mediated by the immune system (7%), 81 in intracellular signaling and trafficking (29%), and 51 in transcriptional and translational regulation and chromatin remodeling (18.5%). For 3 of the 276 genes, the function is still unknown (1%). Of note, 89 of 276 selected genes contribute to CNS development and maintenance acting in concert with the immune system. On the basis of the rare CNV gene content, it is possible to confirm in the present cohort the wide genetic heterogeneity associated with ASDs. Indeed, the same affected loci were detected in only a few unrelated patients (LCLAT1 in patients 7 and 33, and MACROD2 in patients 3, 33, and the siblings 40–41), whereas most patients demonstrated specific subsets of rare CNVs in combinations characteristic for each patient, thus supporting the genetic complexity of ASDs. Moreover, most of the selected genes have never been reported in association with ASDs, and thus they are suggested as new candidate loci and have been grouped into six functional networks that all contribute to CNS development and maintenance. Although a subgroup of the selected genes contribute to synaptic function, thus confirming the neurological interpretation of ASD as a “synaptopathy”, our findings support the existence of dozens of non-synaptic genes that may be implicated in ASDs, which encompass a wide range of biological function and cellular processes, such as intracellular signaling and chromatin-mediated transcription. In particular, it may be speculated that anomalies in the interplay between the nervous and the immune systems, due to genetic causes, may be responsible for autism development in specific subsets of autistic patients. Finally, the genotype-phenotype correlation in the reported ASD series seems rather complex. First, rare CNVs were found in only 55% of the patients, thus emphasizing the need to analyze large cohorts of autistic patients by means of different high-throughput genome-wide approaches in order to increase the detection rate for these disorders. In addition, most patients positive in the array CGH analysis (~76%) were found to be carriers of more than one CNV, which were present in different combinations, thus supporting the existence of a genetic model characterized by oligogenic heterozygosity. Furthermore, when more rare CNVs are found in a single patient, it is difficult to attribute a “major” causative role to one of them, even in case of de novo CNVs, which are generally considered high-penetrance variants. Indeed, the more severe clinical pictures in our cohort (i.e., syndromic autism) do not always correlate with the finding of rare CNVs, neither with the CNV size nor with the presence of de novo CNVs in the subset of patients bearing rare CNVs.