30 September 2011. In Monday morning’s plenary talk, Aravinda Chakravarti of Johns Hopkins University in Baltimore, Maryland, exhorted the audience to embrace all genetic leads when searching for the answers to complex diseases like psychiatric disorders. Drawing lessons from his research on Hirschsprung's disease and autism—both developmental disorders of the nervous system with complex patterns of inheritance, variable phenotypes, and involving multiple genes—he argued for an appreciation for both rare and common variants, coding and non-coding mutations, and the genetic context in which these mutations land. As a case in point, he described an interaction between RET and EDNRB genes in Hirschsprung's disease discovered in a Mennonite population (Carrasquillo et al., 2002), but which isn’t found in the general population. This genetic interaction, called “epistasis,” highlighted a disrupted interaction between neuroblasts and the gut mesenchyme during development that is central to disease etiology. “We should not expect any one kind of family, or any one kind of genetic analysis to uncover everything,” Chakravarti said.
Kenneth Kendler of Virginia Commonwealth University in Richmond echoed this sentiment in a talk marking his receipt of the Snow and Ming Tsang Lifetime Achievement Award. After detailing fascinating highlights of his career, he argued that understanding the causes of psychiatric disease would be best served by taking a broad, yet rigorous view of multiple factors that contribute at multiple levels. “We are too flighty,” he said, referring to his view that the field tends to spring from one new approach or technology to the next without buckling down to struggle with the inherent complexity of the problem.
No matter how you pronounce it—short or long "i"—dysbindin-1 was the focus of an afternoon symposium devoted to probing its function. Originally linked to schizophrenia in a family-based association study in the pre-GWAS era (Straub et al., 2002), the gene encoding dysbindin (DTNBP1) has lost its luster as a candidate gene for schizophrenia in recent years because follow-up studies, including GWAS, have not consistently linked variants in the gene to the disorder (see SRF related news story). Still, the possibility that the gene may play more of a role in certain subpopulations, combined with dysbindin-1 reductions found in postmortem brain tissue from schizophrenia cases (Tang et al., 2009), has attracted a number of researchers to pursuing dysbindin’s function. To wit: Dysbindin-1 mutants show abnormalities in relevant neurotransmitter systems (see SRF related news story), and a naturally occurring dysbindin mutant mouse exhibits behavior proposed to model aspects of schizophrenia (Talbot, 2009).
Victor Faundez of Emory University in Atlanta, Georgia, plunged into the cell biology of dysbindin-1, describing its place within an eight-protein BLOC-1 complex, which in turn interacts with another complex called AP-3. Together, these protein complexes are involved in vesicle trafficking, a key process in getting proteins to their appropriate places in a cell. Faundez showed how perturbing dysbindin or other proteins in these complexes could derail this process in cultured cortical neurons, leaving a membrane-bound enzyme, PI4KIIα, stranded in the cell body rather than at its normal axon terminal target. Faundez suggested that dysbindin’s contribution to schizophrenia susceptibility might lie in this type of protein mistargeting—something which could potentially disturb synapse composition and function.
Daniel Weinberger, now at the new Lieber Institute in Baltimore, Maryland, analyzed the combined effects of DTNBP1 and COMT, a gene that influences dopamine synthesis in the cortex, on cognition. He presented both mouse and human data showing that considering gene-gene interactions can produce surprising results. “This is how epistasis throws single locus analyses into a tailspin,” he said. For example, a new mouse model of dysbindin-1 deficiency performed better than controls in a working memory task, as did mice lacking only COMT. But mice lacking both of these genes performed worse than controls. This counter-intuitive scenario also played out in a human fMRI experiment, in which the brains of people with the DA-enriching Met version of COMT operated more efficiently than those with the Val allele when subjects carried a non-risk-associated DTNBP1 haplotype (Bray et al., 2005). However, this relationship was turned on its head among those carrying the putative risk DTNBP1 haplotype; the brains of those with the Met COMT allele operated less efficiently than those with the Val COMT allele. Drawing from dysbindin’s role in trafficking dopamine receptors to the membrane (Ji et al., 2009), Weinberger proposed that D2 receptors abnormally accumulate in those with the putative risk DTNBP1 haplotype, and coupling this with the extra dopamine from the Met COMT allele leads to an imbalance in dopamine signaling that hampers cognition.
Delving into the synapse, David Jentsch of University of California in Los Angeles described abnormalities in glutamatergic synapses found in mice lacking one or both copies of the dysbindin-1 gene. In the prefrontal cortex and hippocampus, excitatory signals were reduced, and this stemmed both from pre- and post-synaptic changes, including a decrease in expression of NR1, a subunit of the NMDA receptor (Karlsgodt et al., 2011). Decreasing NR1 expression also correlated with poorer performance on a working memory task. To get a brainwide view, Jentsch also described manganese-enhanced microMRI brain scans of these mice, finding deformations in their brain structure, including sensory regions of temporal cortex and distorted activity in dopaminergic networks and the hippocampus. In line with this, mice lacking dysbindin exhibited abnormal contextual fear conditioning, a test of hippocampal function.
Disrupting dysbindin also had repercussions for interneurons, according to Konrad Talbot of the University of Pennsylvania in Philadelphia. Mice lacking dysbindin showed a substantial reduction in inhibitory signaling within the hippocampus, as well as a decrease in parvalbumin (PV) staining within interneurons—whose numbers seem unaffected—in the hippocampus and auditory cortex. Oscillatory activity in their hippocampi reflected impaired inhibition, with reduced auditory evoked gamma oscillations. These mice also exhibited reduced auditory gating and pre-pulse inhibition, consistent with electrophysiological reports in schizophrenia. In human postmortem brain tissue, Talbot reported decreased PV cell density in schizophrenia compared to controls; whether this reflects a loss of PV content or a loss of PV-containing interneurons is unclear, but Talbot argued that PV loss in an otherwise sound neuron could still limit the fast-spiking of these interneurons. Because compromised interneuron function has been proposed as a contributing factor to schizophrenia (see SRF hypothesis), Talbot suggested that these PV interneurons could represent “a final common pathway” affected by other risk variants.
Sorting Swedish CNVs
Sarah Bergen of the Broad Institute of Harvard and MIT rounded out the afternoon with another exploration of CNV occurrence in schizophrenia and bipolar disorder. Drawing from the new, unpublished Swedish sample mentioned the previous day (see SRF related news story), Bergen used a microarray (Affymetrix 6.0) to detect CNVs in some 1,500 individuals with schizophrenia, 800 with bipolar disorder, and more than 2,000 controls. In all, 4,438 large (>100 kb), rare, autosomal CNVs were found. “Singleton” (those observed in only one individual) deletions were enriched in schizophrenia and bipolar cases compared to controls, as were medium-sized CNVs ranging from 200-500 kb. The two disorders diverged in some ways, with a greater incidence of large deletions (>500 kb) and a greater number of genes impacted by CNVs in schizophrenia. In schizophrenia, CNVs landed in previously implicated loci like 22q11 and 16p11.2, as well as in new places like 9q12, a relatively uncharacterized stretch of chromosome, and 9q34.3, a region linked to Joubert syndrome, which is marked by cerebellar malformation. Bergen noted that although there seemed to be a greater involvement of CNVs in schizophrenia than bipolar disorder, this may reflect the greater number of schizophrenia cases in the study.—Michele Solis.