9 Oct 2017
by Hakon Heimer
The two-day biennial symposium hosted by the Stanley Center for Psychiatric Research at the Broad Institute began on September 25 in Cambridge, Massachusetts. As in previous years, the Stanley Center invited a range of pioneering researchers from the border zones of mental illness, genetics, and neuroscience, and there were many presentations of direct or indirect relevance to psychotic disorders.
In his welcoming remarks, Stanley Center Director Steve Hyman, who organized the meeting with Stanley Center researchers Guoping Feng of the Massachusetts Institute of Technology and Ben Neale of Harvard University, said, "This field is very old—it probably started with Hippocrates—but very young, rejuvenated by genomic technologies.” To kick things off the morning genetics session, Neale reviewed some data from recent genetic studies of early childhood mental illnesses. He said that the Psychiatric Genomics Consortium (PGC) has been able to increase its sample sizes for sequencing of common variants contributing to autism spectrum disorder (ASD) and attention-deficit/hyperactivity disorder (ADHD) thanks to contributions from the iPSYCH program in Denmark. Surprisingly, the genetic material extracted from bloodspots collected from newborns, and maintained in freezers for many decades, can be used to sequence exomes. These new samples should allow for the identification of many more variants contributing to these disorders.
Neale also discussed recent data on the contribution of de novo mutations to ASD. He said that some 20 to 30 genes can clearly be termed major causal factors in ASD. It has also been found that some genes are less tolerant to disruption than others, suggesting that a high level of the risk is driven by a small number of genes. Preliminary results suggest that the genes that predispose to ASD overlap substantially with ADHD risk genes.
In the second half of his talk, Neale took a look forward to the analyses that he and his colleagues will do on samples from among the 500,000 that are part of the United Kingdom's Biobank. Given that participation is voluntary, the cohort is, not unexpectedly, biased toward those groups more likely to participate—women, the well elderly, and the more affluent. Indeed, there are only about 225 schizophrenia cases. Still, Neale believes that the data will be useful for looking at genes that turn up in psychiatric diagnosis studies to determine what other effects they have on health.
Neale was followed by Mick O'Donovan of Cardiff University in the United Kingdom, who leads the schizophrenia portion of the PGC. The number of hits in the schizophrenia GWAS project continues to rise, with about 150 from the dataset that is currently under review at a journal. O'Donovan sees no reason that the next data freeze shouldn't contain about 250 risk loci.
One thing that is encouraging is that the results from Asian cohorts are congruent with those from Europeans, so the PGC can combine these. He notes, however, that African genomes are still underrepresented. O'Donovan then mentioned a study recently posted on the bioRxiv preprint server (Pardiñas et al., 2016), wherein Cardiff researchers found 50 loci not previously seen in PGC publications. Within these, he said, they have already found more than 10 genes that are credible as the source of the signal, and for which the markers index functional variations. In other work, he mentioned that pathway analysis is showing that there is some overlap between common variant and CNV schizophrenia gene sets.
Tarjinder Singh of Harvard Medical School and the Broad Institute then discussed research into rare variants in schizophrenia, which might contribute to the "missing heritability" of the disorder and, indeed, might confer greater risk than common variants. A number of papers have pointed to a greater burden of rare variants in people with schizophrenia, as well as with intellectual disability. Under the more recent Schizophrenia Exome Meta-analysis (SCHEMA) consortium, Singh and colleagues are using the exomes of some 20,000 cases and 40,000 controls to search for rare variants that contribute to disease risk. Although he presented some fresh data pointing to specific genes, he asked that they not be reported in the Twittersphere or elsewhere, as they were analyzed just in the previous couple of days.
The next speaker, Lea Davis of Vanderbilt University, said that she and her colleagues are interested in adding to the knowledge about psychiatric disease, but they approach the work from a curiosity-driven angle. In particular, her group has an interest in how psychiatric diseases can persist in the population despite the low fecundity of people with severe mental illness. Citing Theodosius Dobzhansky's thesis that "Nothing in biology makes sense except in the light of evolution," Davis noted that the mutation that gives rise to sickle cell anemia also has a plus: heterozygotes have some resistance to malaria. She said that her group has searched for more complex evolutionary genetic forces—"polygenic selection"—and found sets of common genetic variants associated with complex traits that also confer evolutionary advantages or disadvantages.
Looking for "signatures of selection" in data from the PGC and the Genetics of Personality Consortium (see SRF related news story), Davis' group found that schizophrenia was the only psychiatric disorder or trait that showed recent adaptation, favoring schizophrenia-protective alleles. However, several of the other phenotypes being studied remain underpowered.
David Goldstein of Columbia University finished the morning session with a talk on epilepsy genetics, which he said holds lessons for psychiatric disorders as well. Trios—family groups consisting of a patient and his/her parents—have been "spectacularly good" at finding mutations that contribute strongly to rare versions of epilepsy. In some cases, it looks like these could lead to therapeutics as simple as amino acid supplements. However, these are the minority of cases, and have not as yet led to treatments for the bulk of epilepsy sufferers.
In one of the short talks just before lunch, Guus Smit of Vrije Universiteit in Amsterdam introduced the SYNGO project. It was initiated by the Stanley Center and is being developed by the Center for Neurogenomics and Cognitive Research at VU, with annotation of pre- and postsynaptic genes including localization of proteins, processes in which they are involved, and their functions.
The challenge of modeling brain disorders
The first afternoon session focused on in vitro research, specifically, the creation of organoids and other multicellular models of mental illnesses derived from human induced pluripotent stem cells (hiPSCs).
Paola Arlotta of Harvard University described her group's use of droplet sequencing to phenotype tens of thousands of brain organoid cells by their gene expression profiles (Quadrato et al., 2017). Arlotta said they were able to distinguish neuronal from glial cells, but even more surprisingly, they identified cells with forebrain versus retinal phenotypes. Within the "forebrain" cells, there were subtypes that corresponded to radial glia, interneurons, callosal neurons, and corticofugal neurons. All of this, she said, is self-organizing and takes place without "outside supervision"; hence, they may be very good models for in vivo neural cells.
Arlotta and her colleagues have managed to grow these organoids for over a year now, and they see synapses by about three months, electrically active cells after four months, and spines and synaptic vesicles by eight months. And in an experiment that one might term the "primordial brain" (with apologies to Oparin and Haldane), they found that when they applied light—sensory information—to the advanced organoids, their cells responded with organized spiking. Finally, she described work with an autism model in which organoids with mutations in the gene CHD8 fail to recapitulate all of the cell clusters found in control brain organoids.
Another achievement that would have been impossible only a decade ago—forebrain organoids—was described by Guo-Li Ming, who recently moved to the University of Pennsylvania. Though there was reduced heterogeneity in cell types compared with in vivo brains, they are seeing layers that look intriguingly like cortical layers, perhaps even a subventricular zone with radial glial cells. In the more superficial layers, they see functional neuronal cells expressing GABA and glutamate, and possessing spontaneous synaptic currents. The GABAergic cells even have subtypes found in vivo positive for parvalbumin and other markers. Ming presented some data from studies of Zika virus, but she and her colleagues are also creating organoids intended to recapitulate specific cortical areas such as hippocampus to study other disorders.
Sergiu Pasca of Stanford University described three-dimensional models of specific brain areas, termed "spheroids," created in his lab (Paşca et al., 2015) based on hiPSCs. By 20 weeks, they are able to get "corticogenesis" with deep and superficial layers. Eventually, the researchers see glutamatergic neurons and radial glia-like cells, and there is no apparent "developmental" time limit: They have some organoids going as long as 850 days. In collaboration with Ben Barres' group at Stanford, which has a particular interest in astrocytes, they have shown that the "astrocytes" from organoids show adult-like phenotypes after approximately 300 days (about nine months), similar to what is seen in fetal development (see Alzforum news story).
Taking things to yet another level of complexity, Pasca's group has combined several hiPSC-derived spheroids with phenotypes of different brain regions—for example, pallium and subpallium—into "assembloids," where they see interneurons migrating from one spheroid to the other, just as they do in cultured fetal cells, forming dendrites and synapses. Pasca and colleagues have also used hiPSCs from the neurodevelopmental Timothy syndrome to show irregularities in interneuron migration in assembloids.
The second afternoon session moved to very different models: primates, both human and nonhuman. Alvaro Pascual-Leone of Harvard University described data generated from transcranial magnetic stimulation (TMS), which shows some evidence of being effective in improving symptoms in psychiatric disorders. He and his colleagues look for changes in intracortical inhibition on EEG in different disease states. Small studies have shown differences among control, bipolar disorder, depression, and schizophrenia subjects in terms of responses to TMS, but larger numbers and ways to control for medication are needed. Because similar experiments can be done with nonhuman primates, there is the potential for probing the effects of TMS at the cellular or synaptic level.
Rogier Landman of the Broad Institute followed, with preliminary data from a few Shank 3 transgenic marmosets mimicking Phelan-McDermid syndrome, in which 80 percent of people develop autism. He stressed that the data from the small sample varied widely but noted possible differences in behavior, eye tracking, and pupil responses to faces.
In his concluding remarks, Randy Buckner of Harvard offered some big-picture thoughts, pointing out that even if there are similar sensory and motor circuits in humans and other primates, there have been immense expansions of cortical association areas in humans, even in the second-level "sensory association" regions. These areas are wired very differently from primary sensory cortical regions—interconnected rather than wired from the bottom up. Learning more about the wiring and activity of these higher-order circuits seems to be an endeavor that could lead to targets for treatment of psychiatric disorders such as schizophrenia.