Children with Early-Onset Psychosis Have Increased Burden of Rare GRIN2A Variants

Background: Children and adolescents with early-onset psychosis (EOP) have more rare genetic variants than individuals with adult-onset forms of the illness, implying that fewer EOP participants are needed for genetic discovery. The Schizophrenia Exome Sequencing Meta-analysis (SCHEMA) study predicted that 10 genes with ultra-rare variation were linked to adult-onset schizophrenia. We hypothesized that rare variants predicted “High” and “Moderate” by the Variant Effect Predictor Algorithm (abbreviated as VEPHMI) in these 10 genes would be enriched in our EOP cohort. Methods: We compared rare VEPHMI variants in individuals with EOP (N = 34) with race- and sex-matched controls (N = 34) using the sequence kernel association test (SKAT). Results: GRIN2A variants were significantly increased in the EOP cohort (p = 0.004), with seven individuals (20% of the EOP cohort) carrying a rare VEPHMI variant. The EOP cohort was then compared to three additional control cohorts. GRIN2A variants were significantly increased in the EOP cohort for two of the additional control sets (p = 0.02 and p = 0.02), and trending towards significance for the third (p = 0.06). Conclusion: Despite a small sample size, GRIN2A VEPHMI variant burden was increased in a cohort of individuals with EOP in comparison to controls. GRIN2A variants have been associated with a range of neuropsychiatric disorders including adult-onset psychotic spectrum disorder and childhood-onset schizophrenia. This study supports the role of GRIN2A in EOP and emphasizes its role in neuropsychiatric disorders.


Introduction
Schizophrenia is a complex neurodevelopmental condition with a lifetime prevalence estimated to be just under 1% worldwide [1][2][3]. Schizophrenia symptoms include a combination of positive symptoms such as hallucinations, delusions, and disorganized behavior; negative symptoms such as social withdrawal and diminished speech; and cognitive impairments [1]. Symptoms typically emerge in late adolescence and early adulthood, but early-onset and childhood-onset cases, while much more rare, also occur. Individuals with early-onset and childhood-onset schizophrenia experience psychotic symptoms before ages 18 and 13 years, respectively [4]. Childhood-onset schizophrenia is considered to be an especially severe form of psychosis, with a higher rate of brain abnormalities as well as greater environmental and genetic risk factors [5,6].
Individuals with childhood-onset schizophrenia (COS) have a nearly three-fold increase of recurrent genomic copy number variations (CNVs) relative to those with adultonset schizophrenia [6,7]. This disparity suggests an increased contribution of rare genetic variation to the etiopathology of childhood-onset schizophrenia. However, fewer than half of children and adolescents with a psychotic spectrum disorder meet the strict criteria for schizophrenia [8] and childhood psychiatric diagnoses often change over the course of development [9,10]. Consequently, there is considerable interest in understanding the genetic underpinnings of the more inclusive early-onset psychosis (EOP) categorization, which is defined as any psychiatric diagnosis with pronounced psychotic symptoms that is onset prior to 18 years of age [11]. The EOP categorization captures psychotic symptomatology across the spectrum of diagnostic criteria and is therefore more representative of the phenotypic variability in children and adolescents with psychotic disorders. Existing research has demonstrated that, compared to adult-onset illness, EOP may be associated with lower cognitive and psychosocial functioning, longer duration of untreated illness, more hospitalizations, and less functional improvement over time [8,12].
We recently documented that similarly to childhood-onset schizophrenia, children and adolescents with the broader EOP diagnosis have a two-to three-fold increase in recurrent CNVs relative to individuals with adult-onset schizophrenia [13]. This suggests that genetic burden, particularly for rare high penetrant mutations, is higher for individuals with EOP than it is for those with typical adult-onset psychotic spectrum disorders. Thus, EOP cohorts that are much smaller than those required to study adult-onset schizophrenia can provide important insights into the rare genetic underpinnings of psychosis.
The recent report of the Schizophrenia Exome Sequencing Meta-analysis (SCHEMA) consortia [14] represents a major advance in our understanding of the genetic architecture of schizophrenia. Through meta-analysis of whole-exomes from 24,248 cases and 97,322 controls, the SCHEMA consortia reported that ultra-rare coding variants in 10 genes (TRIO, CUL1, XPO7, SP4, SETD1A, CACNA1G, GRIA3, HERC1, RB1CC1, and GRIN2A) confer substantial risk for schizophrenia (odds ratios 3-50, p < 2.14 × 10 −6 ) [14]. Many of the genes included in the SCHEMA consortia list have also emerged as strong candidates for other neuropsychiatric conditions, such as bipolar disorder, major depressive disorder, and autism spectrum disorder [15][16][17][18][19][20][21][22], indicating that these genes have a broad continuum of influence on psychosis spectrum, mood, and neurodevelopmental disorders. The findings of the SCHEMA study have exciting potential to significantly impact the development of targeted therapies for schizophrenia and related disorders, as well as to enhance our understanding of the underlying biological mechanisms of these conditions.
Despite the importance of this herculean effort, the observation of SCHEMA genes in an independent sample has yet to be replicated. Early-onset and severe phenotypes often have a higher genetic component than later-onset or milder forms of disorders, and therefore provide a more efficient route for identifying genetic factors [23][24][25]. Identifying genetic factors involved in early-onset and severe forms of a disease can also provide important insights into the underlying biological mechanisms of the disorder. Therefore, we would expect the 10 genes named by SCHEMA to have rare and damaging variation that is overrepresented in early-onset presentations, and for this overrepresentation to be an important validation of the genes' involvement in psychosis spectrum disorders. The goal of this study was to attempt such a SCHEMA replication in a cohort of individuals with EOP (N = 34) and matched unaffected controls (N = 34). More specifically, we tested the hypothesis that the burden of rare (MAF < 0.05) variants predicted to be of "High" or "Moderate" impact on the Variant Effect Predictor Algorithm [26] in the 10 SCHEMA genes would be enriched in the EOP group relative to the controls.

Materials and Methods
EOP Samples: Unrelated children, adolescents, and young adults with EOP were referred to the Developmental Neuropsychiatry Program in the Department of Psychiatry at Boston Children's Hospital. Prior to enrollment into the cohort, clinical diagnoses were ascertained by a board-certified child psychiatrist (JGH) who specializes in EOP. Diagnoses were subsequently confirmed via medical record review with a Diagnostic and Statistical Manual of Mental Disorders (Fifth Edition) (DSM-5) checklist. Age of onset of psychotic symptoms and demographic information (sex, race, and ethnicity) were also collected from the medical record. Ethnicity categories reflect preset options in the medical record.
Inclusion criteria were having a DSM-5 diagnosis for current or lifetime Axis I schizophrenia spectrum or other psychotic disorder with onset prior to age 18. Exclusion criteria were (1) substance-or medication-induced psychosis; (2) psychosis secondary to a brain infection (e.g., encephalitis); and (3) psychosis due to a neurodegenerative disorder such as Wilson's disease, Dystonia muscular deformations, Huntington's disease, Friedrich's ataxia, Ataxia Telangiectasia, or Parkinson's disorder. Diagnoses included schizophrenia, schizoaffective disorders, and unspecified and other specified schizophrenia spectrum and other psychotic disorders.
All EOP participants or caregivers/legal guardians provided written informed consent (and participants <18 years provided assent when possible) on forms approved by the Boston Children's Hospital Institutional Review Board as part of the Manton Center for Orphan Disease Research protocol. After providing written informed consent/assent, each participant provided peripheral blood samples for DNA extraction.
Control Samples: Comparison samples consisted of sex-and racially-matched unaffected biological parents of individuals with suspected Mendelian disorders who are enrolled in the Manton Center for Orphan Disease Research protocol.
Sequencing: Whole exome sequencing on Illumina short read platforms for the EOP cohort was conducted by the Yale Center for Mendelian Genomics (New Haven, CT, USA) and controls were sequenced through the Broad Institute Center for Mendelian Genomics (Cambridge, MA, USA). Analysis was performed using Codified Genomics (San Diego, CA, USA) and Genuity Science (Dublin, Ireland). Variants in the analysis included those predicted "Moderate" or "High" impact by Ensembl's Variant Effect Predictor Algorithm (abbreviated as VEPHMI) and include transcript ablation, splice acceptor variant, splice donor variant, stop gained, frameshift variant, stop lost, start lost, transcript amplification, inframe insertion, inframe deletion, missense variant, protein altering variant, splice region variant, and/or incomplete terminal codon variant [26].
Data Processing: We created a "freeze", which is a proprietary data model developed by Genuity Science that is similar to a joint-called variant call format, or VCF, within the Genuity Science platform [27]. Recognizing that the cases and controls were sequenced at different sequencing facilities, we were particularly concerned to make sure that the coverage within the 10 genes of interest was robust. The overall coverage of the sequencing reads for Variant Effect Predictor (VEP) High or Moderate variants across each of the cases and controls was high, despite the mixed sequencing facilities, with more than 94% of the VEP High or Moderate variants included in the freeze being assigned to either homozygous reference, heterozygous call, or homozygous call in each individual. This 94% quality metric for the freeze ensured the validity of comparing the two cohorts.
A set of 138 housekeeping genes (see Supplementary Table S1) was used as a control to ensure that there were no systematic differences between case and control sequences (e.g., increased rare variant calls in the EOP cohort). The housekeeping genes were selected by a pseudorandom number generator from a list of 3803 housekeeping genes from Eisenberg et al. [28].
Statistical Methods: The sequence kernel association test (SKAT) was used to test for group differences in VEPHMI burden between the EOP and control samples for each of the 10 SCHEMA genes separately. SKAT is a supervised, flexible, computationally efficient regression method to test for association between genetic variants (common and rare) in a region and a continuous or dichotomous trait [29]. SKAT analysis [30] was performed in Genuity Science (CSA: 5.16.2, Sequence miner: 13.2.9, GOR: 12.6.2) [31] using GRCh37. Allele frequency upper threshold was 0.05, with a Harvey-Weinberg equilibrium threshold of 1e-6. Bonferroni correction for the 10 genes tested adjusted the α level to 0.005.
Additional Control Analyses: After a gene with an increased incidence in the EOP cohort was identified, three additional race-matched control cohorts were created to further verify the finding. Each additional control cohort consisted of unaffected biological parents of individuals with suspected Mendelian disorders who were enrolled in the Manton Center for Orphan Disease Research protocol. These additional control cohorts were also compared to the EOP cohort using SKAT.
Genetic Burden: Using SKAT, VEPHMI rare variant burden was first compared between individuals with EOP (N = 34) and an initial set of controls (N = 34) (Tables 2 and S2).
GRIN2A had a significantly increased burden of VEPHMI variation in the EOP cohort (p = 0.004). Seven individuals with EOP were carriers of a rare VEPHMI GRIN2A variant (20% of the cohort, see Table 3). No VEPHMI rare variants in GRIN2A were identified in the first comparison cohort. Among EOP participants, the average age of symptom onset for VEPHMI variant carriers (9.9 years, SD: 2.8) ( Table 4) was very similar to the average age of onset for the entire cohort. GRIN2A was the only SCHEMA gene that showed a statistically significant increased burden in the EOP group relative to the controls after controlling for multiple comparisons. However, the burden of VEPHMI variants in SRCAP genes was nominally increased in the EOP group (p = 0.04).  Additional Control Analyses: To confirm the finding, GRIN2A variants were compared against three additional race-matched control cohorts (N = 34 each) using SKAT. A significantly increased burden of GRIN2A VEPHMI variation was found in the EOP cohort compared to two of the additional control cohorts (p = 0.02 for both). An increased burden of GRIN2A in the EOP cohort was trending towards significance in comparison to the third additional control cohort (p = 0.06) ( Supplementary Tables S3 and S4). Analysis of a set of 138 housekeeping genes revealed no significant differences in variant burden between cases and all four sets of controls.
Interpretation: While the identified GRIN2A VEPHMI variants were low allele frequency and missense or splice site variants, they were all predicted to be benign with the exception of one variant of unknown significance (VUS, chr16-9934641 G>T, p.Ala505Glu) using the American College of Medical Genetics (ACMG) criteria ( Table 5). The individual with EOP and this variant has a diagnosis of EOP with co-occurring epilepsy and macrocephaly. The variant is inherited from the individual's mother who does not have seizures or epilepsy but has a diagnosis of ADHD ( Table 6). The Combined Annotation Dependent Depletion (CADD) score for this variant was 25, indicating it was one of the top 1% of deleterious variants in the human genome [32].  Note: ASD = autism spectrum disorder; OCD = obsessive compulsive disorder; ADHD = attentiondeficit/hyperactivity disorder; PTSD = post-traumatic stress disorder.
Variants were inherited in all individuals where it was possible to be assessed (N = 4; see Table 6). Positive family histories of neuropsychiatric disease were noted in all families. In particular, out of four GRIN2A variant carrier parents, three had a history of neuropsychiatric disease.

Discussion
Despite a relatively small sample size, GRIN2A rare variant burden was increased in our cohort of individuals with EOP in comparison to a total of four control cohorts. This finding is in accordance with past research that identified an association between GRIN2A common polymorphisms (rs7206256 and rs11644461) and childhood-onset schizophrenia [33]. While the variants are not pathogenic according to ACMG criteria, and we are not able to confirm the impact, if any, of these variants on the psychosis pathology of the children and adolescents in the EOP cohort, our findings provide additional support for a role of GRIN2A in EOP and neuropsychiatric disease and represent a partial replication of the SCHEMA consortium.
Multiple lines of evidence suggest that the Glutamate Ionotropic Receptor N-methyl-Daspartate (NMDA) Type Subunit 2A (GRIN2A) gene plays an important role in the pathobiology of psychotic disorders. In addition to the SCHEMA study [14], rare de novo GRIN2A mutations are believed to contribute to sporadic cases of adult-onset schizophrenia [34], and common variation in GRIN2A (polymorphisms rs7206256 and rs11644461) appears to influence the risk of childhood-onset schizophrenia [33]. These findings generally support the glutamatergic hypothesis of psychosis [35], which postulates that glutamatergic hypofunction in NMDA receptors is a mechanism of disease [36][37][38].
Located at 16p13.2, GRIN2A is a GluN2A subunit in the NMDA receptor. NMDA receptors are reliant on the binding of D-Serine to the GluN1 subunit and the binding of Glutamate to the GluN2 subunit [39]. NMDA receptor hypofunction is associated with memory and learning impairments as well as psychosis [40]. Mouse models have determined that dysfunction of NMDA receptors influences symptoms associated with schizophrenia, such as anxiety and memory loss [41]. Further, knockout mouse models of GRIN2A show similarities to schizophrenia models [42]. The rat model has shown that GRIN2A expression is altered in the prefrontal cortex during earlier stages of development, which could impact symptom severity [43]. Literature associating GRIN2A common polymorphisms with schizophrenia implies that more mild changes in expression also contribute to neuropsychiatric disorder etiology [33,44,45].
GRIN2A is a highly sequence-constrained gene, with a pLi of 1 and a loss-of-function observed/expected upper bound fraction (LOEUF) score of 0.188 (GnomAD, accessed 1 December 2022) [46]. Five of the seven GRIN2A variants had a CADD score of 20 or above, indicating they were amongst the top 1% of deleterious variants in the human genome [32]. While the ACMG scoring of these variants was unimpressive with only one reaching the criteria of VUS, this is to be expected for variants with low penetrance that influence the risk for common disease, as these guidelines were not developed for the interpretation of variants in genes associated with complex disorders. While we and others have identified Mendelian forms of EOP [7,13,24], it is clear that not all cases appear to be Mendelian in nature. It is possible that these variants are contributing to a background risk of neuropsychiatric disorders and additional genetic burden is necessary for the EOP phenotype to occur. After all, all individuals with EOP in our cohort had a family history of neuropsychiatric disorders, such as bipolar disorder, depression, and ADHD. In EOP cohort participants 2 and 5, the father was the carrier of the GRIN2A variant, but the maternal side had a history of schizophrenia. The interpretation of the functional impact of these variants is unclear.
Our study has several strengths, such as the use of a unique cohort of individuals with EOP and a range of co-occurring conditions, which is highly representative of children and adolescents with psychotic spectrum disorders. Further, while our EOP sample was small, we were still able to gain important new insight into the potential role that a SCHEMA gene plays in the etiopathology of EOP. Nonetheless, it will be important to replicate this finding in additional, larger cohorts.
One limitation of this study is the utilization of parents of children with Mendelian disorders as a comparison cohort, as these individuals may themselves have an increased burden of damaging variants. However, the inclusion of these participants would arguably bias the analysis against significance. Additionally, we do not know the functional impact of any of the GRIN2A variants, and, while matched on race, subpopulations can have extremely varying allele frequencies. For example, the frequency of rs7193290 is elevated in some African populations. It is possible that the statistical effect that we found is due to, or impacted by, racial subpopulation frequencies.
Relatively high impact genes such as GRIN2A are promising targets for elucidating disease mechanisms. As common polymorphisms of GRIN2A have also been found to be associated with early-onset schizophrenia, adult-onset schizophrenia, and schizotypy [33,44,45], investigations into the role of GRIN2A in early-onset psychotic spectrum disorders deserves further study. Since EOP diagnoses may shift as affected children and adolescents develop [9,10], it will be especially important for future research to follow this cohort and report on the correlation of GRIN2A with different psychosis spectrum phenotypes over time. Continuing to understand the impact of rare GRIN2A variants may lead to faster diagnostics and create opportunities to develop tailored therapeutics for individuals with EOP and other complex neuropsychiatric conditions. Supplementary Materials: The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/genes14040779/s1, Table S1: Housekeeping gene list; Table S2: SKAT markers and allele frequencies of 10 SCHEMA genes; Table S3: Repeated SKAT GRIN2A analysis of cases vs. 3 control cohorts; Table S4: SKAT markers and allele frequencies of GRIN2A variants in cases vs. 3 control cohorts. Institutional Review Board Statement: This study was conducted in accordance with the Declaration of Helsinki, and approved by the Institutional Review Board of Boston Children's Hospital (protocol code 10-02-0053). This protocol was most recently re-approved on 8 March 2023.

Informed Consent Statement:
Informed consent for all participants involved in the study was obtained from the participants themselves and/or from their parent/legal guardian.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author. The data are not publicly available due to privacy concerns.