October 23, 2013. Over 900 researchers gathered in Boston, Massachusetts, to attend the 21st World Congress of Psychiatric Genetics. Organizers Jordan Smoller and Lynn DeLisi, both of Harvard Medical School, welcomed people from all over the world to the city, which offered up bright autumn days to offset dimly lit conference rooms. On Friday, October 18, a schizophrenia genetics symposium emphasized that the way forward lies in collaborations among researchers from all over the world.
The PGC takes Manhattan
On behalf of the Psychiatric Genomics Consortium (PGC), Stephan Ripke of the Broad Institute in Cambridge, Massachusetts, gave an update on their efforts to find common variants contributing to schizophrenia through genomewide association studies (GWAS) (see SRF related news story). Since 2009, the PGC has been amassing ever larger sample sizes with increasing returns, and this year was no different: Comparing single nucleotide polymorphisms (SNPs) between 35,476 cases of schizophrenia and 46,839 controls revealed a whopping 97 genomewide-significant hits. “This is one of the most thrilling analyses of psychiatric genetics ever,” Ripke said.
These hits replicated in an independent sample and together implicated 108 distinct regions, including three on the X chromosome. Plotted across the genome, these signals crossed the very high bar for genomewide significance (p < 5 x 10-8), reaching the skyscraper-like heights looked for in these so-called Manhattan plots. Together these signals accounted for 13 percent of liability for schizophrenia. These hits shine spotlights on a daunting 672 genes, and Ripke highlighted a selection that neuroscientists will recognize: DRD2, which encodes the D2 dopamine receptor, the common target of antipsychotic drugs; various genes encoding glutamate receptors (e.g., GRM3, GRIN2A); and genes for voltage-gated calcium channels (CACNAC1C, CACNB2, CACNA1L). These results mesh with current ideas about overactive dopamine and underactive glutamate signaling in schizophrenia (see SRF hypotheses by Anissa Abi-Dargham and Bita Moghaddam).
Ripke was also this year’s winner of the Theodore Reich Young Investigator Award, and in his celebratory talk on Sunday morning, he described some of the genes that harbored GWA-significant variants in more detail. One was NLGN4, an autism-related gene that encodes a protein involved in synaptic wiring. Another was KCTD13, which lies within the large region at 16p11.2 that, when duplicated, increases risk for schizophrenia. Here, the GWAS signals narrowed in on KCTD13, which controls head size in zebrafish (see SRF related news story). Further analyses showed that the signals are reassuringly enriched in brain-related genes as well as immune system genes (even when signals in the major histocompatibility region were excluded). Ripke argued that even things with very small effect sizes give a clearer picture of biological mechanisms, and he emphasized that finding all of them will require further collaboration.
Consortiums (and acronyms) prevail
Steven McCarroll of the Broad Institute described a newly developing resource called the Genomic Psychiatry Cohort (GPC; Pato et al., 2013), which aims to sequence the entire genome in over 30,000 people, including a subset with schizophrenia. So far, 759 whole genomes are done, including 500 schizophrenia cases. This has picked up 28 million variants with high resolution. For example, McCarroll highlighted the crisp boundaries of the well-known 22q11 deletions found in schizophrenia, which give more accurate measures of the size and locations of these deletions. In addition, whole-genome sequencing detected other types of gene-disrupting variations: smaller deletions that interrupted single exons within a gene; extreme instances of CNVs in which a person carried two to 12 copies of a DNA segment; and insertions of short, mobile pieces of DNA known as Alu elements that can move throughout the genome.
McCarroll also made the point that sequencing turns up many stretches of DNA that don’t align to the human reference sequence. That’s because this venerable reference is, in fact, incomplete, particularly around the centromeres. Studying people of mixed ethnicity has helped map these regions (Genovese et al., 2013), which include one—1q21—hit by deletions in schizophrenia. McCarroll mentioned that things that looked like fairly identical 1q21 deletions are, with sequencing, turning out to be different in size and relative location—something that may help explain the variable expressivity and penetrance associated with such deletions.
Looking for clues among the subset of genes expressed by the brain, Menachem Fromer of the Mount Sinai School of Medicine in New York City outlined the CommonMind Consortium (CMC), a new, collaborative effort to sequence the RNA from large numbers of brain samples. Consisting of five academic groups, two pharmaceutical companies, and one nonprofit, the consortium has already sequenced the RNA from postmortem samples of dorsolateral prefrontal cortex from 228 people with schizophrenia and 240 controls. Unlike microarrays that probe expression of select transcripts, RNA sequencing can pick up a more comprehensive collection of transcripts, including rare or unknown splice variants. Preliminary results point to 27 genes upregulated and 63 genes downregulated in schizophrenia compared to controls. Finally, the CMC plans to eventually make its data available to other researchers to promote further analysis by using the Synapse platform.
Other penetrating results
In the same session, George Kirov of Cardiff University in the United Kingdom evaluated just how harmful deletions or duplications of segments of DNA—called copy number variants (CNVs)—are. Though rare, many CNVs scattered across the genome increase risk for schizophrenia (see SRF related news story) as well as for other neural disorders. According to Kirov’s newly published analysis (Kirov et al., 2013), a given CNV is not equally potent across disorders, however. For 70 specific CNVs, a lower frequency was found in cases of schizophrenia than in cases of autism, developmental delay, or congenital malformations lumped together. Based on these frequencies, plus the occurrence of these CNVs in controls, Kirov estimated the penetrance for each CNV, which was the probability of developing schizophrenia for people carrying a particular CNV. For schizophrenia, this revealed a high penetrance, ranging between 10 to 100 percent for each CNV. But their penetrance was even higher for the developmental delay category. This indicates that these CNVs are highly pathogenic, but more likely to produce an early-onset disorder than schizophrenia, and suggests that some other factors interact with CNVs to buffer, or worsen, their effects.
Karolina Aberg of Virginia Commonwealth University in Richmond reminded the audience of yet another kind of genetic variation, found within the complicated patterns of DNA methylation across the genome (the “methylome”). Methyl groups added to stretches of DNA suppress the expression of the genes underneath, and Aberg asked whether these patterns differed in schizophrenia. Starting with blood cells from 750 people with schizophrenia and 750 controls, Aberg extracted the methylated parts of the genome and then sequenced it. This revealed 141 differently methylated regions, implicating 139 regions within or near genes. Several of these were replicated in a second dataset, including those pointing to FAM63B—a gene linked to neuronal differentiation—and CREB1, SMAD3, and ARNT—genes with roles in hypoxia and which suggest lasting marks of environmental mishaps. Preliminary results in postmortem brain tissue suggested a similar, disease-specific pattern of overmethylation in gene pathways related to hypoxia and the immune system. Still, one audience member worried that studying the methylome derived from blood rather than brain may not be optimal. Aberg countered that blood-based information may offer a reliable biomarker for, if not causal insight into, schizophrenia.—Michele Solis.