22 March 2011. On 10-11 March 2011, over 200 researchers gathered at the New York Academy of Sciences for a meeting entitled Advancing Drug Discovery for Schizophrenia. With a lofty view of Manhattan from the fortieth floor, researchers sifted through clues—from dopaminergic and glutamatergic to genetic and epigenetic, not to mention stem cell, imaging, and postmortem—that could be leveraged into treatment strategies.
Since the 1950s, schizophrenia treatment has relied solely on antipsychotic drugs that block dopamine receptors, specifically the D2 subtype, to dampen the hallucinations and psychosis ("positive symptoms") of schizophrenia. Newer antipsychotics appear to have avoided the motor system side effects of the older drugs, while producing modest gains in antipsychotic efficacy. However, most carry their own side effect of significant metabolic syndrome. Worse still, no drug has been found to treat the anhedonia, blunted affect, and lack of motivation ("negative symptoms"), or cognitive deficits that are also debilitating. This state of affairs, coupled with the exit of major drug companies from research into psychiatric disorders, makes finding new approaches to treating schizophrenia especially urgent.
To expedite this, meeting organizers Stephen Marder of UCLA, Bita Moghaddam of the University of Pittsburgh, and Bryan Roth of the University of North Carolina assembled a line-up of speakers who brought diverse perspectives to the question. With a measure of optimism, and caution, three keynote speakers kicked off Day 1 of the meeting by outlining the current genetic landscape of schizophrenia, which has been propelled by mapping of the human genome, but which remains limited by animal models.
A route to rational drug discovery
Narrowing in on the genetic basis of schizophrenia should uncover core problems giving rise to the disorder, and suggest novel targets for treatment. Patrick Sullivan of the University of North Carolina argued that schizophrenia is a polygenic disorder, characterized by variation in multiple genes. He enumerated new schizophrenia-related variants discovered through genomewide association studies (GWAS), including the stretch along chromosome 6 that is home to major histocompatibility complex (MHC) genes (see SRF related news story), and miR137, a microRNA that targets other schizophrenia-associated genes, such as c10orf26, TCF4, and CACNA1C. GWAS are just reaching the critical size for producing hits, and even larger GWAS (on the order of 100,000 subjects) are needed to uncover more variants, he argued.
Edward Scolnick of the Broad Institute of Massachusetts Institute of Technology and Harvard University described how the bottom-up approach of using genetics to find targets for drugs has paid dividends in treating diseases like cardiovascular disease and breast cancer. Though psychiatry lags behind, he insisted that the genetic results just coming in—even those of small effect—are setting up the pathway to discovery. "Nothing, except money and time, stands in the way of really figuring this out," he said.
Though genetic findings can point to new understandings of basic brain biology, research has struggled to "close the loop" in turning these insights into something for the clinic, said Eric Nestler of Mount Sinai Medical Center. This is partly due to the unique challenges of modeling a psychiatric disorder in animal models (Nestler and Hyman, 2010). Although no one expects to come up with a mouse model that copies all aspects of schizophrenia, efforts to model components of the disorder are hampered by the fact that it is defined entirely in terms of behavior, rather than with measurable biomarkers. To help researchers evaluate their animal models, Nestler categorized them as those with: construct validity, where a mouse carries a genetic variant linked to a disorder; face validity, where a mouse behavior is equivalent to a human symptom; and predictive validity, where both are treated by the same interventions. Those with construct validity might be on firmer ground, but the complicated genetics of schizophrenia—with different patients potentially harboring different complements of disease variants—hinder more accurate construct validity. "We as a field should be more honest and self-reflective about what we are able to do in a rodent," Nestler said.
Despite these difficulties, Nestler has developed a mouse model of social defeat that has some features of depression (see SRF related news story). Interestingly, one-third of the mice subjected to the paradigm do not develop the symptoms of anhedonia and anxiety. Gene expression patterns distinguish resilient from susceptible mice, and suggest that the pathway to resilience is independent from that for susceptibility. This means that ways to promote resilience to stress, rather than dampening the effects of stress, could be a viable strategy.
Genetic approaches before GWAS
The next two speakers explored genetic clues that came to light independently from GWAS. Daniel Weinberger of the National Institute of Mental Health described his work on a brain-specific potassium channel called KCNH2 3.1 that is overexpressed in schizophrenia brains relative to controls, and which produces hyperexcitable cells (see SRF related news story). Antipsychotics bind to this channel, raising the question of whether subtle variations in the channel could predict antipsychotic response.
Weinberger presented new results from a placebo-controlled study of antipsychotic therapy in two patient groups. Those homozygous for single nucleotide polymorphisms (SNPs) in KCNH2 3.1 that are associated with elevated levels of KCNH2 3.1 had significant improvements in their positive symptoms scores, and were five times more likely to continue on the antipsychotic olanzapine than those who were not homozygous for these alleles. The next steps for the Weinberger group involve expressing the novel isoform in cell lines to screen compounds that could stem the hyperexcitability. A recent study suggests that simply adding back the missing intracellular piece of protein missing from this isoform may do the trick (Gustina et al., 2009).
Maria Karayiorgou of Columbia University made the case for using rare, highly penetrant variants to understand schizophrenia. She looks for copy number variations (CNVs)—the loss or gain of an extra segment of DNA—that occur in people with schizophrenia more often than in controls. Though rare, these discrete anomalies in the genome offer clear genes for follow-up and lend themselves to animal modeling. In collaboration with others, her group has recently linked a duplication containing the vasoactive intestinal peptide receptor gene (VIPR2) to schizophrenia (see SRF related news story); however, her talk focused on deletions at 22q11, which comprise the strongest known genetic risk factor for schizophrenia.
With 30 percent of people with these deletions developing schizophrenia, Karayiorgou and her team have been studying mice engineered to carry the equivalent deletion. They find working memory deficits, and disrupted signaling between the prefrontal cortex and hippocampus (see SRF related news story). A suspected risk-conferring gene in the deleted region is a microRNA-processing gene called Dgcr8 (see SRF related news story). This gene controls miRNA expression in the brain, and when deleted in mice, this produces some of the cognitive defects observed in the full deletion, and perturbs cortical circuitry (Fénelon et al., 2011).
The epigenome cometh
The last three talks of the day focused on epigenetic mechanisms involved in schizophrenia. Because the development of schizophrenia involves environmental factors, researchers have been interested in finding signatures of environmental hits, like stress, on the genome. For example, chemical add-ons to the DNA sequence of the genome, like methyl groups, can obstruct transcription, thus dictating which genes are turned on or off. These epigenetic marks do not distinguish cause from consequence, however; they may reflect environmental factors that tilt someone towards schizophrenia, or they may arise from disease progression or treatment. Still, many researchers mentioned that any epigenetic signature of schizophrenia may be a useful biomarker for the disorder.
Alessandro Guidotti of the University of Illinois at Chicago explored the idea that epigenetic marks that repress several different brain genes are involved in schizophrenia (see SRF hypothesis). Postmortem studies have found decreased levels of mRNA for the GABA-synthesizing enzyme GAD-67, the neural migration player called reelin (RELN), and other neurotransmitter receptors and transporters, and have suggested that epigenetic marks inhibit the expression of these proteins. Guidotti presented results showing that inhibiting histone deacetylase, which normally facilitates methylation of DNA, with valproate may be a way to rewrite the patterns of epigenetic marks on the genome. He finds that antipsychotic drugs like clozapine interact synergistically with valproate to promote demethylation—a state that is permissive for expression—at the promoter of RELN in the frontal cortex of mice. Other dibenzepine compounds, like olanzapine, share this ability, but haloperidol and risperidone do not.
Prompted by epidemiology that shows that advanced paternal age increases risk for schizophrenia in offspring (see SRF hypothesis), Jay Gingrich of Columbia University talked about epigenetic mechanisms behind this effect. With each round of sperm production, epigenetic marks are erased then reprogrammed—a process which may become degraded with advanced paternal age. To explore this, Gingrich and colleagues compared the offspring from older mouse fathers to those from young mouse fathers. Several behaviors differed between the two, including measures of an open field ambulatory test, startle responses, and paired pulse inhibition. These mice showed different methylation patterns on their genomes at 0.4 percent of methylation sites. At these sites, the offspring of older males were less methylated than those of the younger males in introns, exons, and promoters. This suggests that differences in promoter activity and in alternate splicing may drive some of the changes observed in the mice. Notably, hotspots of altered DNA methylation could be found in schizophrenia-related genes, though the researchers didn't select the mice for disease-related behaviors.
Schahram Akbarian of the University of Massachusetts rounded out the epigenetics discussion with a talk focused on an epigenetic mark—called histone H3-trimethyl-lysine 4 (H3K4me3)—placed on proteins rather than DNA. H3K4me3 marks are found on the histone protein spools around which DNA wraps, near transcriptional start sites, and are thought to make the nearby DNA open for transcription. Akbarian finds large-scale remodeling of chromatin that occurs over lifespan (Cheung et al., 2010), as well as subtle changes in H3K4me3 patterns in disease. Using postmortem tissue from prefrontal cortex taken from 15 control brains and 30 from people with autism and schizophrenia, neuron-specific chromatin can be isolated and screened for H3K4me3 marks by ChIP-Seq (Robertson et al., 2007). For example, a subset of autism cases had H3K4me3 marks that occurred at transcriptional start sites, but which were more variable, spreading abnormally away from the site rather than being concentrated in a tight peak. A similar kind of spreading was noted in three schizophrenia brains at the start sites for genes encoding GABA receptors, and one exhibited a loss of the marker at this site; neither type of change was observed in controls.
These subtle differences suggest that the precise pattern of these marks matter. If they turn out to be therapeutically relevant, procedures such as simply promoting or inhibiting methylation will be too coarse a manipulation to normalize these patterns. Furthermore, not much is known about the mechanisms that govern where these marks are put down in the first place. Though it is uncertain whether epigenetic marks hold clues to the origins of the disease or instead reflect things like subsequent drug treatment history, stress, or smoking, they might still form part of a molecular phenotype of schizophrenia. Several researchers mentioned the need to check whether these epigenetic patterns hold true in readily available peripheral tissues like blood cells, which could potentially make them a much-needed biomarker.—Michele Solis.