30 April 2007. The afternoon of Day 1 at the Mt. Sinai Conference featured the symposium "Cognitive Genomics, Cognitive Functioning, and Schizophrenia," organized by Robert Bilder and Tyrone Cannon, both of the University of California, Los Angeles.
In the first lecture, Anil Malhotra of Zucker Hillside Hospital, Glendale, New York, gave an "Introduction to Cognitive Genomics: From Single SNPs to Whole Genome Approaches." He illustrated the single nucleotide polymorphism (SNP) approach with the well-known studies of the polymorphism in codon 158 of the catechol-o-methyltransferase (COMT) gene that is presumed to result in higher or lower dopamine availability in prefrontal cortex. COMT genotype has been reported to be associated with alterations in cognitive function in both healthy subjects and those with schizophrenia (see SRF related news story).
He then described the strategy of drilling down from hypothesis-free genomewide linkage studies to SNPs and blocks of SNPs, or haplotypes, using the example of the dysbindin gene, DTNBP1 (Straub et al., 2002), located at chromosome 6p22.3. Linkage analysis had identified a chromosomal locus associated with IQ at 6p (Posthuma et al., 2005), which was also linked to cognitive dysfunction in a group of patients with schizophrenia (Hallmayer et al., 2005). Malhotra's group found that a specific haplotype in DTNBP1 was linked to schizophrenia risk (Funke et al., 2004) and to some measures of cognition in normal subjects (Burdick et al., 2006), and in subjects with schizophrenia (Burdick et al., 2007). As an example of how multiple genes can be combined in this approach, Malhotra noted that this decrement was only found in people with the valine/valine form of COMT.
Finally, Malhotra described the genomewide association study (GWAS), several of which have been reported in the past year and a half. He then focused on his own group's GWAS, which found an association between schizophrenia and two cytokine receptors (described in SRF related news story). Although the GWASs to date have used the disease phenotype, he predicted that studies using cognition as the phenotype would also be forthcoming.
Ty Cannon of the University of California, Los Angeles, in his talk entitled, " A Translational Genetics Approach to Schizophrenia: The Example of DISC1 and Memory-Related Endophenotypes," began with an overview of the contentious aspects of schizophrenia genetics. He ventured that the time has arrived to study schizophrenia endophenotypes in transgenic animal models, but he noted that not everyone would agree. While the glass-half-full camp points to well-replicated linkage data and association studies implicating biologically plausible genes in those regions (e.g., COMT, DTNBP1, NRG1, RGS4, DISC1, G72, DAAO, Akt1), the glass-half-empty camp, among them many geneticists, argue that functionally significant variants have not been identified and that the markers and haplotypes studied are not consistent across studies.
The identification of replicable causal variants has been difficult for a number of reasons, Cannon said, the first being the sheer complexity of the genetic architecture. Thus, studies powered to detect many genes, each of less than 10 percent effect, may be necessary. Furthermore, different genes (or, given that many of the current candidate genes are very large, different mutations of the same gene) may be at play in different families or gene pools. A third likely reason for the failure to find unimpeachable genes is that they are probably controlled by the environment; that is, they may remain silent unless environmental triggers expose them. Finally, he pointed out that different genes may impact different symptom domains, or dimensions, of the schizophrenia "syndrome,” a lead-in to his bigger message about the need to parse the syndrome and identify disease endophenotypes with homologues in mice.
Cannon presented an interesting model, using the structure of the various tributaries to a watershed, to conceptualize the contributions of both genes and environment to endophenotypes and the disease itself. Implicit in this model is the notion that certain critical endophenotypes or symptom domains—branch points "upstream" of the main disease river—should be a focus for translational genetics. The approach that Cannon proposed suggests that researchers link genes with validated schizophrenia endophenotypes and then evaluate homologous phenotypes in genetic mouse models. Rescue of such animal model phenotypes might then point the way to human therapies.
Cannon then turned his attention to DISC1 and his collaboration with the laboratory of Alcino Silva, also at UCLA. In work led by Weidong Li, they have shown that an inducible DISC1 fragment, when turned on in young mice but not adult mice, leads to behavioral abnormalities that may serve as models for schizophrenia endophenotypes, including working memory deficits and abnormalities in social behavior (see SRF related news story).
Robert Freedman of the University of Colorado reviewed the strategy that his group has employed to examine the effects of alterations in a gene—in this case CHRNA7, which codes for a subunit of the nicotinic acetylcholine receptor—on phenotypes ranging from gene expression through the diagnosis of schizophrenia itself. The pivotal work in this line of research was the Freedman group's work on the p50 auditory evoked potential, deficits in which initially helped to identify CHRNA7 as a disease candidate gene in the region of chromosome 15q4. A statistical association between variants in the promoter region of the gene and diminished p50 inhibition was found in controls and patients (Leonard et al., 2002).
How can this be linked to cognitive symptoms of schizophrenia? Both CHRNA7 variants and reduced p50 inhibition have been linked to impaired performance on neuropsychological tests of attention and other cognitive functions that have been reported to be impaired in schizophrenia. Finally, CHRNA7 variants have received some support in association studies.
Based on these and other data, Freedman's group has tried two therapeutic approaches. The first, nicotine, normalizes the p50 response but did not help to enhance cognition. More recently, the researchers have employed an α7 nicotinic receptor agonist—DMXB-A—that has weak activity at other nicotinic receptor subtypes. This drug also increases the p50 response in patients, and although patients report a qualitative improvement in symptoms, this is only weakly reflected in cognitive testing (see SRF related news story).
Terry Goldberg of Zucker Hillside Hospital then discussed the "Impact of Dopamine Regulating Genes on Cortical and Subcortical Information Processing: COMT and DAT1." In his work with the NIMH group, Goldberg and colleagues replicated the earlier work of Egan et al.—that met/met individuals outperform val/val subjects—on several memory and problem-solving paradigms. However, in more recent, larger samples of healthy volunteers, the results were somewhat different. While COMT genotype affected N-back, a test of simple working memory, the researchers were not able to replicate effects on the problem solving of the Wisconsin Card Sort Test (WCST). They did extend the findings into the area of attention, reporting that COMT variation affects simple target detection (continuous performance test and 0-back test).
In fMRI studies of brain regions involved in attentional control—cingulate cortex, dorsolateral prefrontal cortex (DLPFC), and parietal cortex—Goldberg and colleagues have also found an effect of COMT genotype. Normal subjects with val/val COMT showed greater activation in dorsal cingulate than those with met/met. This inefficiency of cortical processing is presumably traceable in part to less synaptic dopamine.
What happens if you increase dopamine signaling with tolcapone? Cognitively, this turned out to benefit those who are functioning with less dopamine (COMT val/val), Perhaps surprisingly, however, it did not negatively affect met/met COMT individuals, for whom the drug might have been expected to drop them out of the optimal range. There was a main effect of tolcapone (i.e., both groups improved) both in terms of activation of prefrontal cortex (PFC), and also on N-back, but on CANTAB-ID/ED, a computerized version of WCST, effects were mixed—in some cases m/m did worsen with tolcapone. Thus, COMT seems to have subtle but measurable effects on DA neurocognition in healthy controls.
Finally, the researchers extended their research to look at dopamine processing in the striatum, specifically the role of a polymorphism that alters expression of DAT1, the major determinant of striatal extracellular DA levels. The researchers designed a task that requires updating of information without a large working memory component, and found that this task activated the caudate nucleus on fMRI during the updating, but not "overwriting" of information.
Carol Tamminga, of the University of Texas Southwestern in Dallas, followed with a description of her research team's approach to "Cognitive Phenotyping Across the Schizophrenia-Bipolar Boundary." Citing the overlap in clinical features, endophenotypes, and major candidate genes for the two disorders (e.g., NRG1, DISC1, DAAO(G30)/G72), Tamminga and her colleagues suggest that endophenotyping strategies focusing on cognition may be useful in helping to categorize and understand psychotic disorders. To that end she presented the design, and some preliminary data, of a small study of psychotic disorders, with patients with schizophrenia or bipolar disorder and their family members.
Evidence for the overlap between the disorders is apparent even in basic clinical data from the groups in the study, where first-degree relatives of subjects with schizophrenia had a significantly increased risk of major depressive disorder. Tamminga then reviewed early data from cognitive tests that suggest that, whereas schizophrenia probands and their family members have deficits in some neuropsychological measures, as do subjects with bipolar disorder, this is not seen in family members of bipolar probands.
In the final talk of the Day 1 oral session, Robert Bilder used UCLA's Consortium for Neuropsychiatric Phenomics as an example of the integration of research approaches and disciplines in the service exploring "mechanistically relevant neurobehavioral phenotypes rather than conventional psychiatric syndromes," in particular informatics approaches to "cognitive phenomics" (see, e.g., Freimer and Sabatti, 2003), that serve both top-down approaches such as GWA of cognitive and neuropsychiatric phenotypes and bottom-up biological approaches.
Among the components of the consortium's approach to study endophenotypes of relevance to psychiatric disease are shared structures between different studies; e.g., human studies of cognition share the same study population of normal individuals and transgenic animal models are created to support these research aims.
In the service of this effort, the consortium is developing a number of Web-based bioinformatics tools, such as:
1. The Hypothesis Web —a knowledge management system.
2. PubGraph —a graphical display of scientific literature relationships.
3. PubBrain —an anatomical depiction of PubMed search terms.
These tools will be compatible with other systems, e.g., National Library of Medicine databases such as Online Mendelian Inheritance in Man (OMIM) and The Gene Ontology.—Hakon Heimer.