Genes to Mental Health Network Open Session Meeting

Welcome and Opening Remarks

Joshua A. Gordon, M.D., Ph.D., Director, National Institute of Mental Health

Shelli Avenevoli, Ph.D., Deputy Director, National Institute of Mental Health

Geetha Senthil, Ph.D., Deputy Director, National Center for Advancing Translational Sciences

Jonathan Pevsner, Ph.D., Chief, Genomics Branch, National Institute of Mental Health

Co-Chairs:

Donna McDonald-McGinn, M.S., C.G.C., Children’s Hospital of Philadelphia; University of Philadelphia

Jonathan Sebat, M.S., Ph.D., Rady Children’s Hospital; University of California, San Diego

National Institute of Mental Health (NIMH) leadership welcomed participants to the Genes 2 Mental Health (G2MH) Network Investigators Meeting. Rare genetic variants can significantly increase the risk of psychiatric disorders. Because one genetic variant could increase risk across multiple disorders, there was a need to better understand the dimensionality of psychopathology across diverse populations including the relationship between genotype and phenotype, the trajectory of mental illness, the mechanisms that underlie mental illness, and the interplay of environmental and genetic risk factors. NIMH established G2MH to promote a collective, harmonized, gene-first approach to understanding psychopathology and neuropsychiatric symptoms in people with rare diseases. Now in its fifth year, G2MH investigators gathered to review what has been learned about genetic risk in rare psychiatric disorders and future directions for research.

Project Presentations

Integrating insights from the gene first approach in 22q11.2 and 16p11.2 deletions and duplications to neuropsychiatric disorders

Raquel Gur, M.D., Ph.D., University of Pennsylvania

Dr. Raquel Gur explained the rationale of a gene-first approach. It is known that there are copy number variations (CNV) within specific genomic loci that are associated with highly prominent neuropsychiatric features such as developmental delay, cognitive impairment, intellectual disability, attention deficit hyperactivity disorder (ADHD), and features of autism spectrum disorder (ASD) and psychosis spectrum disorder. A gene-first approach therefore provided an opportunity to prospectively study multiple disorders across the lifespan, focus on mechanisms found in animal models, and generalize findings across psychiatric presentation. Dr. Gur’s team aimed to 1) apply common dimensional phenotypic measures across CNVs relevant to neuropsychiatric disorders, 2) investigate genetic determinants of psychopathology in CNV carriers, 3), examine the effects of genetic and environmental factors, and 4) establish common phenotypic and genomic resources for the public. She emphasized the importance of including family perspectives in research.

Dr. Gur talked about the challenges they faced in their research workflow. Their phenotypic assessment included neurocognitive screening, psychopathology screening with an emphasis on psychosis, ASD family assessment, and life events. Their participants were children aged 7 and older and their parents. They had seven sites across Europe and North America that coordinated the collection and analysis of phenotype and genotype data. This coordination required multiple steps including memorandums of understanding, training across sites, language translation, data entry and linkages, cross-site standardization, and quality control processes—all of which occurred during the pandemic.

Their study sample included 1,927 individuals, of which 1,095 had a deletion or duplication at a specific loci. The majority of the variations were 22q deletion (638), followed by 22q duplication (184), 16p deletion (184), and 16p duplication (89). The 22q deletion was associated with the most pathology in depression, anxiety, ADHD, ASD, and psychosis. The 16p deletion was associated with slightly higher pathology than the duplications, but much less than the 22q deletion. This pattern was similar to findings from a larger study from the Psychiatric Genomic Consortium. Going forward, Dr. Gur’s team will continue to collect data, refine quality control, and improve site integration. They have submitted multiple manuscripts and are considering their lessons learned in terms of feasibility. They are also engaging with clinicians, families, and other stakeholders to enhance clinical relevance with the aim of preparing for the rigorous standards of a clinical trial.

Leveraging rare genetic diseases to advance knowledge and treatment of brain disorders

Christa Lese Martin, Ph.D., F.A.C.M.G., Geisinger Health System

Dr. Christa Lese Martin provided an overview of her team’s study, which had three primary aims. Aim 1 was to leverage existing electronic health record (EHR) data and other routine evaluation data, with a goal of recruiting more than 1,000 individuals with any rare genetic disease known to be causative of brain disorders. Aim 2 was to assess phenotype in families living with rare genetic diseases associated with brain disorders, with a goal of recruiting 720 individuals with specific CNVs and single gene disorders. Aim 3 was to assess other genetic contributions associated with brain disorder severity, with goals to calculate polygenic risk and resilience scores and to identify second hits across the genome.

Their research was embedded within clinical care, in which diagnosed patients had an opportunity to participate in research and provided consent for access to existing and prospective data collection. Dr. Martin remarked on how the use of clinical data reduced the burden on participants and allowed grant funds to be applied toward more detailed assessment. Their expanded assessment domains included cognition, social skills, adaptive skills, language, executive function, visual motor skills, schizoptypy, neuropsychiatric history, and additional medical history. These domains were aligned with the domains collected by Dr. Gur’s team. To date, Dr. Martin’s team has recruited 1,714 individuals with any rare genetic disorder, as well as 363 probands and 400 family members.

Dr. Martin highlighted two of her team’s studies. The first was a literature review of CHD8 studies that summarized known clinical features associated with this gene, which her team described as a “neurodevelopmental disorder with overgrowth.” Previously, CHD8 was considered a gene associated with ASD, but the team found that other developmental disorders were just as commonly associated, as were clinical phenotypes such as sleep disturbance and gastrointestinal issues. The second was a study exploring phenotypic shifts in CNVs within families. Comparing children with the same pathogenic CNV from two different families, they found that the magnitude of impact on the child’s IQ was the same at around two standard deviations from their parents. However, one family had a lower IQ starting range than the other, which shifted their child’s IQ into the range of intellectual disability. Dr. Martin’s team replicated these findings in 16p11.2 duplication syndrome with social responsiveness. These findings demonstrated the contribution of family background to phenotype expression.

Going forward, Dr. Martin and her team aim to use polygenic risk scores to narrow down prognostic diagnoses so that clinicians can be more proactive with interventions. They will continue to explore the impact of environmental effects on phenotypes to better quantify the type and magnitude of risk across other genes associated with neuropsychiatric disorders. In addition, they are collaborating with other G2MH teams to harmonize data and combine analyses.

Accidental investigations of 16p and 22q

Elise Robinson, Sc.D., Massachusetts General Hospital, Harvard Medical School

Dr. Elise Robinson talked about an anomaly associated with ASD risk. She noted that while most neuropsychiatric disorders have both common and rare genetic variations that often influence similar suites of phenotypic correlations, common and rare genetic variations in ASD appear to influence different phenotypes. On average, rare variations in ASD are associated with intellectual disability while common variations are associated with intelligence. To better understand the variability and regional relevance of the autism polygenic score, Dr. Robinson and her team used portions of the polygenic score to compare diagnosed children with their undiagnosed parents. They noticed one area of the genome—the p-arm of chromosome 16p—that was consistently more associated with risk than it should be, given its size and properties. 

Further investigation indicated that the p-arm of 16p had reduced expression of certain genes in the presence of the 0.5 Mb 16p11.2 deletion, which they referred to as an extended cis effect. Using postmortem tissue, they found that a higher polygenic score for ASD on chromosome 16 was also associated with reduced expression of the same set of genes on the p-arm. They then considered contact patterns across the p-arm and used High-C to determine that there was a high degree of chromatin contact within 16p, suggesting that this variable might contribute to the complexity of genetics in ASD. Dr. Robinson and her team found a similar pattern in 22q, particularly in the neuronal progenitor cells, but also in neurons and stem cells from induced 22q cell lines.

Going forward, her team has been developing models to better understand how reduced expression of 22q increased the risk of psychopathology by examining megabase “chunks” across the genome and aims to generate new High-C data to further examine variability. Other ongoing projects include their Transcriptome-wide Analysis of Differential Expression (TRADE), which is a model to measure total transcriptome change that allows comparisons between perturbation conditions, ongoing studies on the 16p11.2 deletion, mutation models to predict the amount of certain variations, NeuroDev projects in South Africa and Kenya to develop constraint models for diverse ancestry populations, and 3D genome characterization to better understand transcriptome-wide effects of mutations.

Large-Scale Evaluation of the effect of rare genetic variants on psychiatric symptoms and cognitive ability 

David Glahn, Ph.D., Boston Children’s Hospital

Dr. David Glahn talked about his team’s work in understanding CNVs of Major Psychopathology (CAMP); in other words, how CNVs work in very large populations. They had four study aims: 1) acquire a very large cohort of archival data and harmonize those phenotypes, 2) characterize a set of recurrent CNVs in large general populations, 3) examine the contributions of common CNVs using polygenic risk scores and CNVs in humans, and 4) model the effect sizes of rare CNVs in dimensional phenotypes.

To date, Dr. Glahn and his team have identified 31 cohorts of about 1.5 million people, which have varying levels of cognitive, symptom, and diagnostic data. They then harmonized phenotypes and aggregated the datasets. To characterize recurrence, they looked for classic CNVs such as 22q11 or 16p that show a change in DNA between two low-copy repeats. They searched the literature to identify individual CNVs contributing to psychiatric symptoms and cognitive ability and the overlap of those CNVs across groups. From that list of CNVs, they looked at individual groups of symptomatology such as depression and IQ. They found larger effects from depletions than duplications at the population level, a finding that they replicated across different measures. They also replicated these findings across the prevalence of CNVs by ancestry.

Using an independent sample from the SPARK cohort, Dr. Glahn and his team found evidence that polygenic scores could help explain variable expressivity, especially when looking at specific mutations. To explore the stability of their polygenic results, they looked at the overlap between two discovery samples and a third independent test sample. They found a low correlation in depression, but a high correlation in schizophrenia. However, most of the people in the first discovery sample for schizophrenia were also in the second discovery sample. A polygenic score was therefore dependent on the discovery sample used. The team then modeled the relationship between a depression measure and CNVs at aggregate and found that there was no interaction, suggesting that variation may be an additive rather than an interactive effect. Finally, the team modeled CNVs either in the same locus or randomly spread through the genome and found a model to estimate the likelihood that a gene will be a deletion or duplication. This model could be important for considering the aggregate effect of a CNV on the likelihood that an individual will have a psychiatric diagnosis.

Affiliate Presentations

The 3q29 deletion and duplication: Adventures in remote phenotyping

Jennifer Mulle, Ph.D., Rutgers University

Dr. Jennifer Mulle started her presentation by explaining the historical importance of rare disease research in transforming the understanding of important discoveries, for instance, understanding human metabolism, developing statin drugs, and understanding mutations involved in cancer. Because of these contributions, she suggested that rare diseases should instead be called fundamental diseases.

The 3q29 deletion is an extremely rare syndrome that has only recently been definitively associated with schizophrenia. As such, it took a long time to amass a sample size large enough to understand the syndrome better. With their initial funding, Dr. Mulle and her team conducted deep phenotyping to better describe the cognitive, social, behavioral, motor, and other deficits involved. With additional funding, they expanded into neuroimaging and found that 3q29 deletion was associated with substantial deviations in the posterior fossa and the cerebellum. The team has explored mechanisms through in silico analyses to understand the impact of the genes in the interval and identified cellular phenotype and specifically, mitochondrial dysregulation. They have also engaged with caregivers to gain their perspectives.

During the pandemic, Dr. Mulle and her team developed a remote phenotyping protocol that they decided to pilot as a method for expanding data collection. They aimed to enroll and collect data from 200 individuals with 3q29 deletion and their parents. Since launching earlier this year, they have enrolled 21 families with 3q29 deletion, two families with 3q29 duplication, and a group of controls. They have measured social responsiveness and executive functioning in both the individuals with 3q29 deletion and their families and compared these scores to controls to obtain a clinically significant score. Interestingly, they found that parents tend to score a bit higher than controls—suggesting that there may be ascertainment issues in selecting those who were more severely affected. They also found unexpected results in executive functioning scores and plan to compare scores in participants included in a previous study to determine if scores improve over time. Additionally, the team gathered data on medication use and found that children with 3q29 deletion began receiving antidepressant medication at a very young age (as young as age 3). 

Dr. Mulle summarized that there was much more to understand about 3q29 deletion and that a remote phenotyping protocol was a viable data collection tool that helped them increase their study population.

Sex chromosome dosage effects on human brain and behavior

Armin Raznahan, M.D., Ph.D., Intramural Research Program, NIMH

Dr. Armin Raznahan reviewed his team’s research on sex chromosome aneuploidies, which are particularly valuable to study because of their collective commonality, very wide dosage ranges, and relationship to the larger topic of biological sex differences—a potent modifier in neuropsychiatric outcomes. There is evidence from population registries that sex chromosome aneuploidies have pronounced penetrance in psychiatric disorders along a similar magnitude of more classically associated CNVs.

Dr. Raznahan reviewed outcomes from an XYY cohort, which showed that some individuals who do not fall within a diagnostic category do fall within substantial clinically relevant sub-thresholds. This was important as researchers designed their studies to minimize ascertainment bias and maximize their ability to obtain dense phenotypic data across multiple groups. Dr. Raznahan and his team then measured individuals across 66 dimensional psychopathology measures of clinical relevance, ranked the domains, then mapped the behavioral phenotype for XYY. They found that XYY most prominently impacted social impairment domains, but proactive aggression was hardly impacted—a significant finding considering the persistent idea that XYY is associated with aggression. Then, the team compressed the 66 domains into 10 partially dissociable axes of clinical covariation to identify how the different clinical covariations relate to other phenotypes. For example, they found that externalizing symptoms were correlated with caregiver strain, and early social impairments were correlated with adaptive function impairment. When comparing XYY and XXY, they found that social impairment was more sensitive to an extra Y than an extra X, but that both X and Y were equally potent in internalizing symptoms.

Dr. Raznahan and his team then used neuroimaging to compare the effects of an added X, an added Y, and a control of an added chromosome 21 (i.e., Down syndrome) on the brain. They used 15 measures of cortical structure and function from three neuroimaging modalities to derive effect sizes for 300 brain regions. They found that each aneuploidy created a very different fingerprint on the brain, but there was a correlation between the three at the first principal component. To explore the mechanism for this positive correlation, the team decoded their map against publicly available gene expression maps in the human brain. They found that the regions of vulnerability fell within the ventral attentional system, the anterior cingulate, and the rostral insular—areas that are enriched for gene expression signatures that tag serotonergic signaling in some cell types. Further, the map of human cortical vulnerability in aneuploidy was correlated with the principal component across multiple psychiatric conditions, suggesting that there could be an underpinning of regional brain vulnerability. Dr. Raznahan and his team aim to explore this through postmortem data, mouse models, and human-derived cell lines.

Setting the ground for precision medicine in 17q12 CNVs and 15q13.3 deletions

Daniel Moreno DeLuca, M.D., M.Sc., University of Alberta, Canada

Dr. Daniel Moreno DeLuca talked about the increasing population of people with ASD in the U.S., 40 percent of whom have an identified genetic cause, representing more than 1.8 million people. Despite this large population, there remains a lack of genetic testing and awareness. The 17q12 deletion is one of the recurrent CNVs that is now associated with ASD but was historically known to be associated with diabetes, kidney complications, and uterine complications. The deletion is very rare—out of a cohort of 15,000, Dr. DeLuca’s team identified only 18 people with the 17q12 deletion. In this sample, they found a strong association with ASD, schizophrenia, and psychopathology in general, as well as renal cysts and diabetes. To go beyond the dichotomy of autism or no autism, the team sought to understand not only behavioral domains but also somatic clinical features. They also sought to understand the contribution of background genetic variation, which they hope will inform the basis of genetically-informed clinical trials.

Their current study has a cohort of 12 people with 17q12 deletion and 15 with 17q12 duplication. In addition to ASD, they also found common co-occurring conditions such as general developmental disorders and ADHD (which tended to be more frequent in 17q12 duplication). They did not find any cases of schizophrenia but did find seizures in those with 17q12 duplication more than those with 17q12 deletion.

Dr. DeLuca also talked about his team’s work in 15q13.3, which they aim to translate into clinical trials. There are several entry points into the actionable genome (i.e., genes with defined medical recommendations), such as susceptibility to medication side effects, pharmacological treatment based on underlying biology, or prophylaxis of concurrent conditions. Dr. DeLuca reviewed two case studies. One case was a male adult with psychosis, rage outbursts, and epilepsy who was being treated with several medications. The other was a child in an inpatient hospital with significant behavioral outbursts and ASD who was also on several medications. Both cases had 15q13.3 deletion, which is known to include the cholinergic alpha-7 receptor. Leveraging this knowledge, Dr. DeLuca and his team tested galantamine to inhibit the breakdown of acetylcholine. This led to decreased hospitalizations in the first case and greatly reduced polypharmacy in the second case. The medication was well tolerated with no major side effects in this context. The team is currently recruiting for galantamine treatment, and preliminary data show that the drug is associated with fewer aggressive episodes over time.