12 May 2013
13 May 2013. MicroRNAs (and one not-so-microRNA) took center stage in two symposiums at the International Congress for Schizophrenia Research held in Orlando, Florida, 22-25 April 2013. Though these molecules do not code for proteins, recent events have boosted their stature. First, the ENCODE project has emphasized that these non-coding RNA transcripts act as important control knobs on gene expression (see SRF related news story): for example, microRNAs work at multiple points along the gene-to-protein pipeline, typically reducing gene expression, and a single microRNA can target multiple genes. Second, the largest genomewide association study (GWAS) of schizophrenia to date has catapulted a particular microRNA, miR-137, to prominence (see SRF related news story), and researchers have begun to explore miR-137’s effects on the brain (see SRF related news story).
“Not too long ago, the only thing we were interested in were coding proteins,” said Steven Potkin of the University of California, Irvine, who chaired a session on Monday, 22 April, devoted to miR-137. This microRNA holds additional interest for schizophrenia because it controls expression of five other schizophrenia risk genes (Kwon et al., 2013), and because, even before the schizophrenia GWAS of 2011, Potkin and colleagues identified it as a molecule of interest in a GWAS of brain activation (Potkin et al., 2010).
Theo van Erp of the University of California, Irvine, explored how brain activation varied with miR-137 genotype. Using functional magnetic resonance imaging (fMRI) to monitor activation of the dorsolateral prefrontal cortex (DLPFC), a region implicated in schizophrenia, van Erp and colleagues found that participants with schizophrenia had higher DLPFC activation than did controls during a working memory task performed with similar success. In addition, van Erp reported that this activation differed between people carrying the risk genotype for miRNA-137, which constitutes two copies of the T risk allele for the single nucleotide polymorphism (SNP) fingered by the schizophrenia GWAS, rs1625579. For both the control and schizophrenia groups, homozygous carriers of the T allele showed greater activation than carriers of the non-risk G allele (either GG or GT), with TT carriers with schizophrenia showing the highest level of activation—a sign of brain inefficiency. Though intriguing, the origins of the link between miR-137 genotype and brain activity remain obscure, and van Erp concluded by outlining experiments in stem cells designed to elucidate how exactly the TT risk genotype influences gene expression and neuronal features.
Stephen Lawrie of Edinburgh University, U.K., presented data published last year that explored effects of the same miR-137 risk genotype on brain activation (Whalley et al., 2012). To focus on “trait” effects related to a genetic predisposition for the disorder, rather than “state” effects that come from living with the disorder, the study examined individuals related to someone with schizophrenia or bipolar disorder, but who themselves were not ill. This highlighted different patterns of activation in TT risk allele carriers in the medial frontal gyrus. Also, among those related to someone with schizophrenia, but not bipolar relatives or controls, an abnormal pattern of activation in the amygdala and pre-/post-central gyrus was found. This kind of disorder-related specificity could mean that the ultimate effect of a single risk allele will depend on the genetic context in which it finds itself—for example, among a collection of other risk alleles associated with schizophrenia. As the number of risk alleles rises with larger GWAS samples (see SRF related conference story), Lawrie said that imagers will need to find ways of simultaneously examining the effects of 60-80 risk genotypes.
Turning to the venerable electroencephalogram (EEG) technology, Jeroen Decoster of Maastricht University in the Netherlands reported that the same miR-137 genotype varied with the size of the P300 component of evoked brain activity waves. P300 reflects aspects of attention and working memory, and is reduced, on average, in schizophrenia. In his study of 336 people with schizophrenia, Decoster and colleagues found that TT risk genotype carriers had lower P300 amplitudes than the non-risk, genotype carriers, and that miR-137 could account for 2 percent of the variability in P300. Other significant associations with P300 were found for SNPs in ABCB1, BDNF, and DISC1.
Moving out of the living brain, Steven Potkin presented data from a postmortem exploration of miR-137 regulation. Using qPCR to measure levels of miR-137 expression in DLPFC, no significant differences emerged among schizophrenia, bipolar disorder, and control brains. Subdividing the brains according to miR-137 genotype of the donor, he reported that among controls, TT carriers had less miR-137 expression than did non-risk G carriers. A non-significant trend for this was seen among schizophrenia and bipolar disorder brains, too, and suggests that TT genotype translates into lower levels of miR-137. Potkin then turned to a nanostring technology that allowed his team to survey more than 800 microRNAs in the brain and found one that was differentially expressed in schizophrenia: miR-1253 was decreased four fold in DLPFC in schizophrenia compared to controls, whereas it was increased nearly fivefold in bipolar disorder. “My apologies to those in the audience who think schizophrenia and bipolar are the same disorder,” he said.
Other members of the non-coding RNA club
On Tuesday, 23 April, symposium talks suggested that other non-coding RNAs besides miR-137 may be at work in schizophrenia. Claes Wahlestedt of the University of Miami, Florida, presented work on miR-132, a microRNA that is reduced in schizophrenia postmortem brain (Miller et al., 2012). Because miR-132 regulates many of the enzymes involved in adding, reading, or removing epigenetic tags that influence gene expression, Wahlestedt and colleagues set out to directly survey epigenetic enzymes in schizophrenia. Applying a nanostring technology to two separate cohorts of postmortem brain, they found two interesting genes: EZH1 and SETD5—both targets of miR-132. Wahlestedt said that miR-132 is not affected by antipsychotic medication, yet because there are no genetic data implicating it in schizophrenia, he suggested that miR-132 might reflect something downstream from the original events that put the brain on the path to the disorder.
Leaving short RNAs behind, Guy Barry of the University of Queensland, St. Lucia, Australia, focused on a longer type of non-coding transcript, descriptively named “long non-coding RNA” (lncRNA) (Guttman and Rinn, 2012). Measuring over 10 kb long and highly tissue specific, lncRNAs are murky in terms of their function, but could somehow contribute to the human brain’s complexity. Barry and colleagues found that one, called Gomafu, was downregulated in mouse cortical neurons after stimulation (Barry et al., 2013). Using a protein microarray, they then found that Gomafu binds to splicing factors encoded by QK1 and SRSF1. Alternative splicing in general (Glatt et al., 2011), and QK1 specifically (Aberg et al., 2006), have been associated with schizophrenia before, suggesting that Gomafu is involved in schizophrenia-related alternative splicing. To further explore this idea, Barry and colleagues overexpressed Gomafu in human induced pluripotent stem cells (iPSCs) and found that this suppressed splice variants of DISC1 and ErbB4; underexpressing Gomafu produced the opposite effect. In postmortem brain samples of superior temporal gyrus, Barry reported about half as much Gomafu expression in schizophrenia compared to controls and proposed that ensuing problems with activity-dependent alternative splicing may contribute to the disorder.—Michele Solis.