INTRODUCTION

Schizophrenia (SCZ) and bipolar disorder (BD) share some genetic risk factors and abnormalities in brain structure and function (Craddock et al, 2005; Lichtenstein et al, 2009; Purcell et al, 2009), the latter are also evident to some degree in unaffected relatives (Whalley et al, 2005; Fusar-Poli et al, 2007). To understand the differential pathophysiologies of these conditions and potentially identify specific markers for vulnerability, it is important to identify genetic factors related to each disorder and any associated brain differences.

Recently, the largest genome-wide association study (GWAS) to date, which included over 40 000 individuals, identified one of the strongest novel genetic associations in SCZ (Ripke et al, 2011). This occurred on chromosome 1p21.3 at rs1625579, within an intron of a primary transcript for a microRNA (miRNA), MIR137. MicroRNAs are short, noncoding RNA molecules that bind to target sequences on mRNA transcripts and act as post-transcription regulators. They are critical in brain development and function, and are consequently important in neurodevelopmental disorders such as SCZ and BD (Perkins et al, 2007). Typically, individual miRNAs are involved in regulating the expression of multiple genes and therefore could explain the complex phenotypic expression associated with these conditions. Indeed, postmortem studies have indicated differential expression levels of several miRNAs in prefrontal regions in SCZ and BD vs healthy controls (Perkins et al, 2007; Forero et al, 2010; Kim et al, 2010; Beveridge et al, 2011). In terms of MIR137 specifically, normal levels of expression are required for regulating the balance of neuronal proliferation and differentiation (Sun et al, 2011; Shi et al, 2010; Smrt et al, 2010; Szulwach et al, 2010). Although the expression pattern has yet to be fully established, it is enriched in neurons, particularly in the dentate gyrus, a region implicated in psychiatric disorders and known to undergo neurogenesis into adulthood (Sempere et al, 2004; Smrt et al, 2010). It is also notable that genes associated with SCZ, including TCF4, CACNA1C, and c10orf26, are also targets of MIR137 (Ripke et al, 2011).

Studying the effect of MIR137 on traits associated with SCZ and BD in unaffected individuals at increased genetic risk for the illnesses may help elucidate the pathological mechanisms of the polymorphism in the absence of psychotropic medication and other illness associated factors, which could confound the results. Examining two high genetic risk populations, one for SCZ and one for BD, also allows examination of whether MIR137 has a specific effect in the SCZ high-risk sample, or whether it has a more general effect on other psychiatric disorders. We examined the effect of variation in MIR137 (rs1625579) on brain activation in (i) those at high genetic risk of SCZ (n=44), (ii) those at high genetic risk of BD (n=90), and (iii) healthy controls (n=81), during performance of a sentence completion task. This task differentiates patients with SCZ, BD, and those at increased familial risk for both disorders, from healthy controls (Whalley et al, 2004, 2011; McIntosh et al, 2008b). It activates language-related regions involved in the pathophysiology of both disorders (Lawrie and Abukmeil, 1998; Shenton et al, 2001; Glahn et al, 2008; Minzenberg et al, 2009). Moreover, it probes neuropsychological deficits in executive function, verbal initiation, and verbal fluency commonly observed in patients (Gourovitch and Goldberg, 1996; Heinrichs and Zakzanis, 1998; Bokat and Goldberg, 2003; Arts et al, 2008).

Mounting evidence implicating prefrontal and medial temporal regions in both SCZ and BD, and expression of MIR137 in medial temporal lobe regions, led us to focus on these regions of interest (the prefrontal cortex, amygdala, and hippocampus). We also report other brain regions that reached significance at the whole brain level (Phillips et al, 2003; Smrt et al, 2010; Szulwach et al, 2010). We examined whether there were differences in activation according to MIR137 genotype across all groups, and whether there were diagnosis-specific differences in genotype between the groups. We hypothesized that there would be a greater influence of the genotype in the SCZ high-risk group, given the genetic association with phenotype in that group alone (Ripke et al, 2011).

MATERIALS AND METHODS

Study Population

The study examined healthy subjects with no personal or family history of a psychiatric disorder, a group at familial risk of BD, and a group at familial risk of SCZ. These are referred to as ‘diagnostic’ groups to avoid confusion with genotype categorization. No individuals met criteria for a psychiatric disorder and none had received psychotropic medication for a psychiatric disorder. The controls and bipolar high-risk group were recruited as part of the Scottish Bipolar Family Study (Whalley et al, 2011). The SCZ high-risk group originated from the Edinburgh High-Risk Study (Hodges et al, 1999; Johnstone et al, 2000).

Individuals with a diagnosis of BDI or SCZ were identified across Scotland. Diagnosis of affected subjects was confirmed with the Operational Criteria symptom checklist (OPCRIT; McGuffin et al, 1991) using data from clinical case notes and the structured clinical interview for DSM-IV disorders (SCID; First et al, 2002). To address issues relating to the substantial overlap in inheritance between the conditions, we excluded unaffected family members where there was a relative with the alternative disorder in equal or closer proximity to the original proband. Unaffected relatives of these individuals between the ages 16–25 years with at least one first-degree or two second-degree relatives with BDI or SCZ were invited to participate in the study. Unaffected, unrelated control subjects were identified from the social networks of the bipolar high-risk subjects.

Only unrelated individuals were included in the current analysis. Exclusion criteria for all groups included a personal history of major depression, mania or hypomania, psychosis, substance dependence, an IQ <70 or clinical diagnosis of learning disability, or any major neurological disorder or history of head injury that included loss of consciousness, and any contraindications to MRI. A total of 81 controls, 90 unrelated bipolar high-risk, and 44 SCZ high-risk participants provided suitable fMRI data along with genetic information, see Table 1. Written informed consent was provided by all participants and studies were approved by the Multicentre Research Ethics Committee for Scotland.

Table 1 Demographics and Behavioral Measures for Groups Separated into miR137 Subgroups
Full size table

Genotyping

Genomic DNA was extracted from venous blood. Genotyping was conducted by the Wellcome Trust Clinical Research Facility, Edinburgh, United Kingdom (http://www.wtcrf.ed.ac.uk) and used a standard TaqMan/PCR-based assay (TaqMan, AssayByDesign, Applied Biosystems, Foster City, California). Subjects were typed for the G/T alleles at SNPs rs1625579 (Ripke et al, 2011). As it is the common variant that is reported to be the risk allele (T), the groups were split into those carrying two copies of the risk allele (TT, referred to as ‘RISK+’), and those carrying only one or no copies (GG or GT, referred to as ‘RISK−’). Tests of Hardy–Weinberg equilibrium (HWE) were performed using the genetics package in ‘R’ (version 2.12.1, R Foundation for Statistical Computing).

Experimental Paradigm

Participants performed the Hayling sentence completion paradigm in the scanner (see Burgess and Shallice, 1997; Whalley et al, 2004). Subjects were shown sentences with the last word missing and asked to silently think of an appropriate word to complete the sentence and press a button. This was a blocked design with four levels of difficulty according to sentence context. Sentences were assigned difficulty levels based on the frequency of the most typically presented word in the list of norms (Bloom and Fischler, 1980). The baseline condition consisted of viewing a screen of white circles on a black background. Each block lasted 40 s. The order of the blocks was pseudo-random. Each block was repeated four times using different sentences. This design allowed a standard subtraction analysis (sentence completion vs baseline) and the more tightly constrained parametric analysis (examining areas of increasing activation with increasing task difficulty).

Immediately after scanning, subjects were given the same sequence of sentences on paper and requested to complete each sentence with the word they first thought of in the scanner. ‘Word appropriateness’ scores were determined from the word frequency list of sentence completion norms (Bloom and Fischler, 1980). A score of one was given to the most frequently produced word in the word frequency list, a score of two for the next most frequently produced word, etc.

Analysis of demographic data used two sample t-tests or ANOVAs and χ2 tests for continuous and categorical data as appropriate. Between-group comparisons for the behavioral data including mean reaction time and word appropriateness were also conducted using two sample t-tests or ANOVAs. All analyses were conducted using SPSS for Windows (version 14.0, SPSS, USA).

Scanning Procedure

Imaging was carried out at the Brain Imaging Research Centre of Scotland on a GE 1.5 T Signa scanner (GE Medical, Milwaukee, WI). For the bipolar family study, the functional imaging protocol consisted of axial gradient-echo planar images (EPIs; TR/TE=2000/40 ms; matrix=64 × 64; field of view (FOV)=240 × 240 mm2) acquired continually during the experimental paradigm. Twenty-seven contiguous 5 mm slices were acquired within each TR. Each EPI acquisition was run for 404 volumes. The first four were discarded. Data for the SCZ high-risk family study was collected non-concurrently and hence there were unavoidable technical differences to the above (TR/TE=4000/40 ms; matrix=64 × 128; FOV=220 × 440 mm2). Thirty-eight contiguous 5 mm slices was acquired within each TR and each EPI acquisition was run for 204 volumes, of which the first four were discarded. The T1 sequence yielded 180 contiguous 1.2 mm coronal slices (matrix=192 × 192; FOV=24 cm; flip angle 8°).

Image Processing and Analysis

EPI and T1 images were reconstructed into nifti format (Mayo Foundation, Rochester, MN) in SPM5 (Statistical Parametric Mapping: The Wellcome Department of Cognitive Neurology and collaborators, Institute of Neurology, London) running in Matlab (The MathWorks, Natick, MA). Images were pre-processed using standard protocols available in SPM5. EPI images were realigned to the mean volume in the series. The functional images were then normalized according to standard co-registration procedures using each individual's structural scan. All realigned and normalized images were smoothed with an 8 × 8 × 8 mm3 full-width half-maximum Gaussian filter.

First level statistical analysis was performed using the general linear model approach. At the individual subject level, the data was modeled with four conditions corresponding to the four difficulty levels each modeled by a boxcar convolved with a synthetic hemodynamic response function. Estimates of the subject's movement during the scan were entered as ‘covariates of no interest’. The participants data were filtered in the time domain using high pass filter (128 s cutoff) and serial correlations were accounted for by using the autoregressive (AR(1)) model. Contrasts were constructed to examine all four sentence completion conditions vs baseline, and areas of increasing activation with increasing task difficulty (parametric contrast).

Second Level Analysis

For each contrast of interest (sentence completion vs baseline and parametric effects), one contrast image per subject was entered into a second-level random effects analysis. A full factorial ANOVA model was used to compare the three groups. F tests determining the main effect of genotype were examined, followed by tests of genotype*group interactions. Genotype and group were entered as two factors with two levels of genotype, (RISK− (GG and GT) and RISK+ (TT)) and three ‘diagnostic’ groups (healthy controls, individuals at high-risk of BD, and individuals at high-risk of SCZ). This analysis was conducted controlling for age and verbal intelligence as measured with the National Adult Reading Test (Nelson, 1982), as analysis of demographic data indicated differences between the three groups (see Table 1). Pairwise t-tests were conducted to determine the direction of any interaction effects. For all analyses, statistical maps were thresholded at a level of p<0.005 (uncorrected) and regions were considered significant at a cluster level of p<0.05, corrected for multiple comparisons across the whole brain. Based on prior hypothesis, small volume corrections were also applied for the prefrontal cortex, amygdala, and hippocampus created using the WFU PickAtlas (Tzourio-Mazoyer et al, 2002; Maldjian et al, 2003). All coordinates are quoted in Montreal Neurological Institute (MNI) convention (http://www.mni.mcgill.ca). Images are overlaid onto standard brain in MNI space using Mango software package (http://ric.uthscsa.edu/mango).

RESULTS

Demographic, Clinical, and Behavioral Measures

Of the 81 controls, 50 subjects were RISK+ and 31 subjects were RISK− (two GG homozygotes). Of the 90 bipolar high-risk subjects, 66 were RISK+ and 24 were RISK− (four GG homozygotes). Of the 44 SCZ high-risk subjects, 26 were RISK+ and 18 were RISK− (one GG homozygote). Allele frequencies did not differ from the HWE across all groups (p=0.99), or within each of the diagnostic groups (controls p=0.50, bipolar high risk p=0.21, SCZ high risk p=0.66).

Demographic details are presented in Table 1. Within each diagnostic group there were no significant differences in age, gender, handedness, or IQ between the genotype groups, with the exception of a trend for a difference in handedness within controls between the genotype groups. There were no significant differences between the diagnostic groups in terms of gender or handedness, however, there were differences in mean age and mean IQ. These were entered as covariates into the statistical model used to assess the imaging data.

Statistical analyses of the behavioral data within the diagnostic groups revealed no significant differences in within-scanner reaction times or word appropriateness scores between the genotype groups. There were, however, differences in mean word appropriateness scores between the diagnostic groups. This is explored further in terms of brain activation below.

Task-Related Brain Activation Patterns

All subjects demonstrated the expected patterns of brain activation and behavioral responses, indicating subjects were performing the task appropriately in the scanner (Whalley et al, 2004, 2011; McIntosh et al, 2008a). Regions activated across the groups for the sentence completion vs baseline contrast included the left medial prefrontal regions, including supplementary motor area (BA 6), lateral prefrontal regions, left lateral temporal cortex, sub-cortical structures including the lentiform and thalamus extending to hippocampal regions, left lateral parietal cortex, occipital lobes bilaterally, and right cerebellum. For the parametric contrast, all groups demonstrated activation in similar areas including left lateral and medial prefrontal cortex, left lateral temporal cortex, and right cerebellum, for further details see Supplementary Figure 1.

Genotype-Dependent Differences in Brain Activation

Sentence completion vs baseline

For sentence completion vs baseline, there were no significant main effects of genotype across the groups. There were, however, significant genotype*group interactions in the left amygdala (p<0.01, KE=109, F=10.25, x=−24, y=6, z=−16, restricted to region of interest as defined by the WFU PickAtlas tool; Tzourio-Mazoyer et al, 2002) and left pre/postcentral gyrus (p=0.03, KE=459, F=9.97, x=−44, y=−10, z=30, corrected at the whole brain level; Table 2, Figure 1a). There were no significant differences in the other pre-specified regions of interest.

Table 2 Genotype*group Interactions and Main Effect of Genotype
Full size table
Figure 1
figure 1

Genotype*group interaction for MIR137. (a) Depicts genotype by group interactions in left amygdala and left pre/postcentral gyrus for sentence completion vs baseline. Images are overlaid onto standard brain in MNI space using Mango software package (http://ric.uthscsa.edu/mango). Map represents F-statistic images thresholded equivalent to p uncorrected=0.005, see Materials and Methods for further details. (b, c) Depicts graphs of extracted data for the clusters of significant interaction, see text for further details.

PowerPoint slide

Full size image

Further t-tests examining the direction of interaction revealed this was significantly different between the controls (in the direction of RISK− greater activation than RISK+) vs the SCZ high-risk group (RISK− less activation than RISK+) for both the amygdala (p=0.030, restricted to regions of interest as above) and pre/postcentral gyrus (p=0.005, corrected at the whole brain level). These regions did not demonstrate significant interactions for the controls and bipolar high-risk comparison, nor between the SCZ high-risk and bipolar high-risk subjects. Graphs of the extracted data are presented in Figure 1b and c. For the amygdala, there was a general decreased activation across all groups for this contrast, with the exception of the SCZ high-risk RISK+ group.

There was no significant association between the extracted data from either of these clusters and behavioral measure of word appropriateness score, either within or across groups.

Parametric contrast

For the parametric contrast, there was a significant main effect of genotype across all groups in the right posterior medial frontal gyrus (p<0.01, KE=831, F=13.04, x=12, y=−14, z=62, corrected at the whole brain level), extending to superior frontal gyrus and to paracentral lobule, see Figure 2a and b. Further t-tests examining the direction of effect indicated this was in the direction of RISK− greater activation than RISK+ (p<0.001). There were no significant genotype*group interactions for this contrast.

Figure 2
figure 2

Main effect of MIR137 genotype across groups. (a) Depicts effects of genotype across all three groups in the left posterior medial frontal gyrus for the parametric contrast. Images are overlaid onto standard brain in MNI space using Mango software package (http://ric.uthscsa.edu/mango). Map represents F-statistic images thresholded equivalent to p uncorrected=0.005, see materials and methods for further details. (b) Depicts graphs of extracted data for the cluster of significant main effect of genotype, see text for further details. For groups additionally separated for ‘diagnostic’ grouping, see Supplementary Figure 2.

PowerPoint slide

Full size image

There were also no significant associations between the extracted data from this cluster and behavioral performance as determined by the word appropriateness scores, either within or across groups.

DISCUSSION

We have demonstrated a common effect of the MIR137 genotype across controls and two groups of individuals at high risk of major mental illness on activation in the right posterior medial frontal gyrus. We have also shown significant differential effects of the MIR137 genotype across groups in the left amygdala and left pre/postcentral gyrus. In regions where a significant interaction was found, differences occurred between controls and SCZ high-risk sample, not between the controls and bipolar high-risk group, nor between the familial groups themselves. These findings suggest, therefore, that there is a specific effect of MIR137 on the SCZ high-risk sample, with additional general effects across the groups. This pattern is compatible with the GWAS mega-analysis, in which the association was primarily seen with the diagnosis of SCZ, rather than BD (Ripke et al, 2011). As previously described, the task was found to engage typical language-related regions and behavioral performance measures indicating that subjects were performing the task appropriately in the scanner (Whalley et al, 2004, 2011). Importantly, we were able to show that the above effects of genotype in the absence of the confounding disease and medication effects that compromise studies of patient populations. Also, there were no significant relationships between the main findings of interest and measures of behavioral performance on the task.

One of our main findings was an overall effect of the MIR137 genotype across the diagnostic groups for the parametric contrast in the right posterior medial frontal gyrus, encompassing BA6 and supplementary motor areas, SMA and preSMA. This region demonstrated a reduced gradient of brain response to increasing task difficulty in those homozygous for the risk allele (RISK+) vs those heterozygous or homozygous for the ‘protective’ allele (RISK−), irrespective of diagnostic group (Figure 2b, and see Supplementary Figure 2). The parametric contrast examines levels of difficulty based on sentence constraint. This finding is therefore unlikely to be purely attributable to the simple motor response of button pressing. Indeed, this brain area is engaged in response to increasing task load (Whalley et al, 2004), and is activated in other executive and language-processing tasks (Rypma et al, 1999; Jansma et al, 2000; Bitan et al, 2007). This region has also been repeatedly shown to be abnormal in SCZ (McGuire et al, 1995; Honey et al, 2005; Exner et al, 2006). Although this region is in general implicated in the planning of motor actions, recent research indicates more diverse roles, eg, linking cognition to action (Nachev et al, 2008). It also forms part of the network of regions in the cognitive dysmetria model of SCZ where dysfunction, or disconnection, between cortico-thalamic–cerebellar regions are postulated to contribute to the abnormal coordination of mental processing underlying features of the disorder (Andreasen et al, 1998).

In addition to diagnostic independent effects, we also report significant genotype*group interactions in the amygdala and pre/postcentral gyrus. In particular, the amygdala has been repeatedly implicated in neurobiological models of SCZ (Phillips et al, 2003; Strakowski et al, 2005). In general, we found decreased activation of the amygdala during sentence completion vs rest, as we have reported previously (Whalley et al, 2011). The SCZ high-risk RISK+ group, however, failed to demonstrate this normal task-related deactivation. In SCZ, abnormally reduced responses of the amygdala to emotional vs neutral stimuli have previously been reported (Schneider et al, 1998; Phillips et al, 1999; Gur et al, 2002; Williams et al, 2004; Romaniuk et al, 2010), proposed to derive from an increased response to neutral stimuli (Schwartz et al, 2003; Holt et al, 2006; Surguladze et al, 2006; Hall et al, 2008; Romaniuk et al, 2010). Together with behavioral studies showing deficits in emotion recognition, it is suggested there is an overactivation and misinterpretation of ambiguous stimuli as being emotionally salient. This response is considered to contribute to clinical symptoms such as delusions, where neutral stimuli may be incorrectly perceived as threatening (Phillips et al, 2003). With regard to the current task, the paradigm is primarily an executive, non-emotional task. We speculate therefore that this finding may reflect an abnormal amygdala response in the SCZ high-risk RISK+ group to non-emotional stimuli. It should be considered, however, that since the common variant is reported to be the risk allele for MIR137, those individuals in the RISK− group also carry a copy of the risk allele. Another interpretation of the data, therefore, is that the comparison of the RISK−/+ group represents a comparison of those with/without the ‘protective’ allele. In this case, the interpretation would be that the presence of the ‘protective’ allele is associated with normal amygdala function, and that those in the SCZ high-risk RISK+ group without the protective allele do not demonstrate the expected deactivation during the task. It is notable that emotional processing bias and amygdala dysfunction has previously been reported in BD (Wessa and Linke 2009) and in their unaffected relatives (Whalley et al, 2011). In apparent contrast to these findings, we found no association between MIR137 genotype and amygdala dysfunction in the bipolar high-risk sample. It is possible that the effect of MIR137 risk genotype on brain activation in the bipolar high-risk sample is negligible against a background of presumed multiple genetic risk variants for BD and amygdala dysfunction.

The other cluster showing an interaction effect was in the pre/postcentral gyrus. The pattern of response in this region was different from that of the amygdala cluster above, reflecting that this is a task-activated region. The graph of extracted values demonstrated greater activation in the control RISK− group vs the control RISK+ group, whereas the reverse was seen for the SCZ group (RISK−<RISK+). Indeed, the only group to demonstrate reduced activation during the task was the SCZ high-risk RISK− group. The pre/postcentral gyrus is primarily considered to be involved in somatosensory processing and motor functions, but its role has more recently been shown to also include higher cognitive tasks including executive functioning and linguistic processing (Whalley et al, 2004; Minzenberg et al, 2009). It is also reported to be affected in SCZ (Tanskanen et al, 2010; Job et al, 2002; Zhou et al, 2005; Glahn et al, 2008), and there are reports of associations of abnormalities in this region with clinical symptoms seen in the disorder (Kircher et al, 2001; Zhou et al, 2005; Nenadic et al, 2010; Jardri et al, 2011) together with neurological soft signs such as sensory integration deficits (Dazzan et al, 2004).

The importance of miRNAs have only relatively recently been realized. MicroRNAs are involved in many developmental processes spanning developmental timing, apoptosis, differentiation, and morphogenesis (Ambros 2004). As such, abnormalities seen in neurodevelopmental disorders could originate from disruption of post-translational processes under control by miRNAs (Sempere et al, 2004). The functional role and expression pattern of MIR137 has yet to be fully determined. This variant was chosen because it was identified by one of the largest GWAS to date (Ripke et al, 2011). At present there is a rather limited literature. It is considered to be enriched in neurones, particularly, in the dentate gyrus and it is thought to have multiple functions during different stages of neurodevelopment (Shi et al, 2010; Smrt et al, 2010; Szulwach et al, 2010). Two recent molecular studies indicate MIR137 is involved in multiple epigenetic mechanisms regulating the balance of neuronal proliferation and differentiation (Sun et al, 2011; Smrt et al, 2010; Szulwach et al, 2010). In one study, overexpression in the dentate gyrus resulted in inhibition of dendritic morphogenesis and maturation of neurones, whereas reduced expression had the opposite effect (Smrt et al, 2010). In another study, overexpression promoted proliferation of adult neural stem cells and reduced expression resulted in enhanced differentiation (Szulwach et al, 2010). Deficits in expression may therefore interfere with normal synaptic morphogenesis and signal transduction, thereby affecting aspects of cognitive function as seen in SCZ (Sempere et al, 2004). Effects have also been indicated out-with hippocampal regions, eg, reduced expression of MIR137 in the frontal cortex has been reported in a small number of sporadic Alzheimer's Disease patients (Geekiyanage and Chan 2011).

One limitation of the study was that the two high-risk samples were collected separately, with consequent unavoidable minor differences in the samples. The main demographic differences are controlled for using standard statistical approaches. The main methodological difference was that a TR of 4 was used for the SCZ high-risk study and a TR of 2 was used for the bipolar family study. It should be stressed, however, that both studies were conducted on the same scanner using the same paradigm design. Differences in TR are reported to primarily effect the number of data points collected and therefore efficiency of the experiment, with minimal impact on noise and smoothness of the data (Smith et al, 2007). For the current study with relatively long blocks optimized for the longer TR, we considered the impact of the different TRs to be minimal, especially given quantitative analysis of activation maps suggesting the location and extent of activation was consistent across the studies (see Supplementary Figure 1). We have previously reported acceptable reproducibility both between and within scanning sites at the group and subject level, suggesting the viability of combining fMRI data sets (Gountouna et al, 2011; Suckling et al, 2012). We cannot completely exclude the possibility that these differences may impact on the current findings. However, such confounding effects are most likely to be evident at the between diagnostic-group level. Genotype effects and interactions between genotype and diagnostic group are less susceptible to the effects of differences in imaging prescription, as individuals of each genotype were scanned using both versions and the interaction effects were estimated using a model that took account of these differences. There was also a significant difference in one of the measures of behavioral performance, word appropriateness, between the diagnostic groups. Examining associations between these measures and the extracted data, however, revealed no significant relationships, indicating that these differences were not impacting on the main findings. It is also the case that although there were differences between diagnostic groups, there were no significant differences in either measures of performance between the genotype groups. Another limitation is that this is a relatively small sample size in terms of genetic studies, however, we would like to note that it is a relatively large sample size given other imaging studies, particularly, given the difficulties in recruiting and practicalities of scanning individuals at high familial risk of mental disorder. Finally, another issue is that some of the high-risk subjects have not yet passed the maximal risk period for development of the disorder. It is possible that some of these familial individuals may develop either of these disorders. Conversely, it is also true that the majority of individuals will remain well, therefore, markers for resilience will also be seen in the high-risk groups. Clinical follow-up of these individuals will clarify whether these differences are RISK− related or protective. Also, due to the diversity of functional imaging paradigms applied across the field and due to practicalities in recruiting such a large familial sample, we were not able to perform a replication analysis of these findings. The findings should therefore be viewed considering these limitations.

In summary, mounting evidence suggests that variation in MIR137 could contribute to neural abnormalities seen in the major psychiatric disorders and potentially offers novel etiological pathways and therapeutic opportunities. As yet knowledge of these underlying pathways are only just beginning to be established, however, the current study indicates diagnosis independent and differential effects of MIR137 genotype in regions previously implicated in these psychiatric disorders.