Cognitive state and connectivity effects of the genome-wide significant psychosis variant in ZNF804A

Abstract

Alterations of connectivity are central to the systems-level pathophysiology of schizophrenia. One of the best-established genome-wide significant risk variants for this highly heritable disorder, the rs1344706 single nucleotide polymorphism in ZNF804A, was recently shown to modulate connectivity in healthy carriers during working memory (WM) in a pattern mirroring that which was found in overt disease. However, it was unclear whether this finding is specific to WM or if it is present regardless of cognitive state. Therefore, we examined genotype effects on connectivity in healthy carriers during rest and an emotion processing task without WM component.

111 healthy German subjects performed a battery of functional imaging tasks. Functional connectivity with the right dorsolateral prefrontal cortex during rest and an implicit emotion recognition task was determined using the seed voxel method and compared to results during WM.

During rest and during the emotional task, a pattern of reduced interhemispheric prefrontal connectivity with increasing number of rs1344706 risk alleles could be seen that was close to identical to that during WM, suggesting a state-independent influence of the genetic variant on interhemispheric processing, possibly through structural effects. By contrast, the abnormal prefronto-hippocampal connectivity was only seen during the WM task, indicating a degree of task specificity in agreement with prior results in patients with schizophrenia. Our findings confirm a key role for disturbed functional connectivity in the genetic risk architecture of schizophrenia and identify cognitive state-dependent and independent components with regard to WM function.

Introduction

Since Wernicke, it has been suspected that the presence of dramatic behavioral impairments despite only subtle regional brain pathology that characterizes schizophrenia may indicate disturbances in connectivity, i.e. the interactions of brain regions during distributed processing of information. This has mainly been studied in the context of working memory (WM), since cognitive impairment, and especially impairment of WM function, is a core feature of schizophrenia (Vaz and Heinrichs, 2006). Abnormal connectivity during WM has indeed been identified consistently in schizophrenia patients (Crossley et al., 2009, Ragland et al., 2007, Spoletini et al., 2009, Whitfield-Gabrieli et al., 2009, Wolf et al., 2009). One node specifically highlighted in these network analyses is lateral prefrontal cortex. Here, a well-replicated finding is impaired interhemispheric prefrontal connectivity during working memory in patients (Schlosser et al., 2003a, Schlosser et al., 2003b) and also in patients and unaffected siblings during a choice reaction task, pointing towards heritability of the effect (Woodward et al., 2009). These functional findings are supported by several studies using structural imaging or diffusion tensor imaging (Carpenter et al., 2008, Kubicki et al., 2008, Price et al., 2007). In these studies, reduced volume or tract integrity of the corpus callosum, especially in the genu, was consistently found.

Another way in which connectivity may influence WM in schizophrenia is through altered prefronto-temporal interactions. During verbal encoding (episodic memory), a reduction of functional connectivity was seen between the dorsolateral prefrontal cortex (DLPFC) and parahippocampal/superior temporal gyrus (STG) regions (Wolf et al., 2007), while during WM, an increase in functional coupling between frontal and temporal regions has been observed (Crossley et al., 2009). Specifically, schizophrenia patients failed to uncouple right DLPFC from left hippocampal activation during WM, but not a control condition (Meyer-Lindenberg et al., 2005b). Typically, the hippocampus is deactivated during WM tasks, at least in tasks with non excessive WM load and non familiar stimuli (Esposito et al., 2006, Zarahn et al., 2005). Therefore, this persistent prefrontal hippocampal coupling in schizophrenia could be interpreted as aberrant recruitment of associative memory capacities during a WM task. These data suggested that schizophrenia, at least functionally, may not be mainly a “dis”connection syndrome (where connectivity is uniformly reduced) but rather a “dys”connection syndrome, where both abnormally increased and decreased coupling may be present during some cognitive states (Stephan et al., 2006).

Since schizophrenia is highly heritable, with estimates as high as 81% (Sullivan et al., 2003), it is reasonable to ask whether these findings in manifest disease extend to genetic risk. Extensive studies in healthy candidate gene variant carriers provide broad support for an impact of these variants on connectivity (Bertolino et al., 2006, Buckholtz et al., 2007, Kempf et al., 2008, McIntosh et al., 2008, Meyer-Lindenberg et al., 2007, Tan et al., 2008). However, candidate gene studies are open to the objection that they study genetic variants that have not been unambiguously linked to the schizophrenia disease phenotype. An opportunity to further establish a role for connectivity in mediating genetic risk was therefore presented by the recent discovery, through genome-wide association (GWA) and follow-up, of a genome-wide significant risk single nucleotide polymorphism for psychosis in the gene ZNF804A (rs1344706) (O'Donovan et al., 2008). This variant has been confirmed in subsequent independent GWA study (Purcell et al., 2009) and large case–control datasets (Riley et al., 2009, Steinberg et al., 2010). Although not every dataset surveyed has shown positive association, a recently published meta-analysis including 23 studies (Williams et al., 2010) showed that the evidence for association between rs1344706 and schizophrenia surpasses widely accepted benchmarks of significance by several orders of magnitude (p = 2.5E−11). Therefore, this variant near ZNF804A represents one of the best-established genetic entry points into studying systems-level mechanisms in schizophrenia genetics.

In a previous analysis (Esslinger et al., 2009) we found that healthy subjects carrying the ZNF804A schizophrenia risk allele showed changes in functional connectivity of right DLPFC during WM that resembled those discussed earlier for schizophrenia, namely a gene dose dependent reduction in interhemispheric prefrontal connectivity and an increase in connectivity of DLPFC with contralateral hippocampus. However, the question remained whether this result is specific for WM (highlighting cognitive state-dependent mechanisms) or broadly present (therefore arguing for state-independent, for example structural, changes in connections between brain areas). To answer this question, we studied healthy subjects from our ongoing study (Esslinger et al., 2009) stratified for ZNF804A rs1344706 genotype during rest and during a face emotion processing task that places no demands on WM. Based on the literature discussed earlier, we hypothesized that diminished interhemispheric connectivity in the ZNF804A risk allele would be task independent, i.e. found during all three conditions. In contrast, we expected abnormal prefronto-hippocampal connectivity to be present only during WM, following our previous findings in patients with schizophrenia (Meyer-Lindenberg et al., 2005b).