We used a translational genetics approach to investigate the relevance of DARPP-32 for aspects of human brain morphology and function. Resequencing of the gene revealed common single-nucleotide variants mostly within a single LD block, which we summarized for subsequent multilevel association analyses in a 7-SNP haplotype. These 7 SNPs were selected because they showed strong association to cognitive phenotypes implicated in cortical-striatal function and were then used as a haplotype to test other biologic associations in independent datasets. This strategy reduces the number of association tests in a complex phenotype dataset while at the same time increasing the likelihood that chromosomes with causative mutations/haplotypes are contrasted with other chromosomes. All uncovered variants were noncoding, in agreement with recent findings from a limited resequencing study in 50 Chinese probands (35
). While it cannot be excluded that the effects observed in this study are due to several rare mutations that are coding, arose on the background of a common (ancestral) haplotype, and are in LD with the analyzed SNPs, we propose that the associations with cognition and brain structure and function observed in our dataset may reflect complex patterns of allelic heterogeneity that have an impact on expression. This is supported by the analysis of mRNA expression, which showed an impact of PPP1R1B
genetic variation on the abundance of all identified full-length PPP1R1B
isoforms, compatible with the assumption of polymorphic functional regulatory elements. Expression was highest for subjects homozygous for the frequent (CGCACTC) haplotype and lowest for carriers of the GATGTCA haplotype. This provided a plausible mechanism for an impact of the genetic variation characterized here on neuronal function, similar to observations in other psychiatric risk genes of noncoding marker haplotypes having effects on mRNA expression (36
In an effort to bridge the gap from gene transcription to cognitive behavior, we studied the effects of the associated haplotype at the level of neural systems in another independent sample of healthy controls using neurobiological phenotypes plausibly connected to the cognitive phenotypes and to variation in gene expression. As reviewed, DARPP-32 is unambiguously linked to neostriatal function (11
), which is, via interaction with prefrontal cortex, prominently implicated in a variety of cognitive domains (13
). In agreement with the neurochemical anatomy of DARPP expression, we observed a substantial impact of genetic variation on neostriatal volume in structural neuroimaging. Volume was relatively reduced in dorsal putamen in carriers of the frequent haplotype. The dorsal putamen participates in interactions with the prefrontal cortex (13
); in primates, this has been labeled the associative loop
and includes the DLPFC, much of the caudate nucleus, and the precommisural putamen as well as the ventral anterior thalamus (15
). Perturbing this circuit at any point may affect its function, and indeed, there are studies in primates and rats showing that dysfunction of the striatum produces behavioral deficits similar to those found in dysfunction of the prefrontal cortex (37
). A genetic effect on striatal-prefrontal interaction was confirmed by an analysis of structural connectivity of the striatum, which showed a strong increase in its correlation with prefrontal cortex in the frequent haplotype. While it is important to bear in mind that this structural connectivity reflects covariation of regional volumes across subjects and does not directly quantify anatomical connections in white matter, we (32
) and others (38
) have previously observed patterns of structural connectivity that agree with known anatomical connectivity and are sensitive to genetic variation (32
), and studies of conditions with known impact on anatomical connectivity have shown that structural covariance characterizes these processes in aging (39
) and in schizophrenia (42
). Furthermore, in the human optic system, it has been demonstrated directly that anatomically connected structures (i.e., optic tract, lateral geniculate nucleus, and primary visual cortex) covary in volume across individuals (43
The structural findings were confirmed and extended by the results in functional imaging, where functional connectivity between striatum and prefrontal cortex was again shown to be increased in carriers of the frequent haplotype in both fMRI paradigms. Regarding striatal activation, the striatum showed less responsiveness to the environmental stimuli in carriers of the frequent haplotype in both tasks, suggesting that this finding indicates more efficient, intrastriatal processing. Indeed, the relatively smaller putaminal volume associated with this haplotype may also reflect more efficient circuitry. Since regional volume changes are taken into account when functional imaging data are spatially normalized and the main effect of task is a deactivation during working memory but an activation during face processing, these functional imaging results are not likely to be partial volume artifacts. Again, the localization of statistical maxima was in lateral putamen (also reaching significance in caudate for the face-processing task), in agreement with the structural findings and highlighting those regions receiving input from the DLPFC and premotor prefrontal cortex. It is important to note that while our tasks are cognitive and engage associative prefrontal-striatal circuitry broadly, our results highlight putamen as the component of the structural and functional circuit DARPP most has an impact on but not necessarily as the main striatal processing station related to task performance. Taken together, the neuroimaging data therefore show that the frequent haplotype is associated in healthy individuals with more efficient intrastriatal processing combined with an increase of prefrontal cortical input onto a smaller striatum. We note that the size reduction may contribute to striatal processing efficiency as it has been proposed that increasing overlap of functional projection fields by compression of pathways into successively smaller striatal structures is a key mechanism for the integration of information streams in the basal ganglia (15
). Since our analyses of function concerned 2 tasks that were designed to assess prefrontal function during emotional and working memory processing, it would be useful to extend the present study by tasks that are specifically intended to activate striatal areas engaged in other cognitive functions, such as implicit memory or reward.
These systems-level findings provide a neural mechanism for the observed pronounced impact of genetic variation in PPP1R1B
on a wide range of cognitive domains during neuropsychological testing. The imaging findings confirm the behavioral importance of DARPP-32 effects on frontostriatal processing since working memory and response alternation and attention critically depend on both prefrontal and striatal function and are disturbed in diseases thought to have an impact on these structures, such as schizophrenia and Parkinson disease (18
). Improved working memory and executive capacity may also underlie the association with general intelligence (44
). The range of cognitive tests affected by variation in this gene likely reflects the critical importance of frontostriatal function for core aspects of cognition (18
). As an important negative, 3 memory tasks depending on hippocampal but not striatal function, the California Verbal Learning Test and Wechsler Memory Scale–Revised logical memory 1 and 2 tests, were not affected by PPP1R1B
variation, suggesting a relatively specific impact on the frontostriatal circuit as well as a limited role of DARPP-32–dependent processing in neural systems for episodic memory in humans.
Our control analysis using the functional common Val108/158 Met genetic variation in COMT
showed no effect on PPP1R1B
mRNA expression, volume, or functional activation of the striatum, in good agreement with human (45
) and animal knockout (26
) data showing that this variant has an impact on dopamine turnover in prefrontal cortex but not striatum, where dopamine flux is mainly dependent on the dopamine transporter. These negative control findings, therefore, suggest that the biological associations with PPP1R1B
found in the present study are specifically related to the integrative function of DARPP-32 in dopaminoceptive striatal neurons and not to a general effect on dopaminergic neurotransmission elsewhere in this circuit.
Finally, we conducted a family-based association study to investigate an impact of the observed association with risk for schizophrenia. Since DARPP-32 has an impact on cortical plasticity, striatal function, dopaminergic neurotransmission, and response to psychotomimetics and neuroleptic treatment (3
), it has long been viewed as an attractive candidate molecule for this disease. Its candidacy is further supported by our findings on frontostriatal processing in healthy controls since disinhibited dopaminergic neostriatal neurotransmission is associated with schizophrenia and has been linked to deficiencies in prefrontal function (27
) and abnormal gating, proposed as a central mechanism in the disorder (16
). Located at 17q21, PPP1R1B
, the gene encoding DARPP-32, is also in or near a region implicated in risk for schizophrenia by a recent metaanalysis of whole genome linkage (46
). Indeed, we found positive associations for the M04 G allele as well as for the frequent (CGCACTC) haplotype, the same alleles that had an impact on cognitive, expression, and imaging phenotypes.
This finding, however, must be regarded as preliminary since we did not have an adequately powered family-based replication sample, and it is therefore difficult to know whether this is a true association or a false positive. However, it is noteworthy that the alleles and haplotypes implicated in disease risk also predicted changes in controls mirroring observations in manifest schizophrenia. Striatal volumes in drug-naive, first-episode patients have been found to be decreased (47
), an effect confounded later in the illness by neuroleptic treatment, which increases striatal volume (48
). Imaging studies of patients with schizophrenia also show increased prefrontal-subcortical connectivity (49
). This supports the tentative clinical association by showing that the directionality of the observed genetic effects agreed with the disease phenotype.
To the degree that this genetic association with schizophrenia can be confirmed, our data lead to the provocative observation that a frequent haplotype in PPP1R1B
predicts increased frontostriatal interactions that appeared beneficial (as evidenced by relatively better performance on a wide range of cognitive tasks) yet contributed to risk for schizophrenia. This raises the question of whether a genetic advantage in normal subjects may translate into a disadvantage in the context of other functional impairments also associated with schizophrenia, such as abnormal function of the prefrontal cortex. To the degree that the striatum acts as an active gating station (50
), increased information flow through this regulatory cognitive subsystem is predicted to contribute to increased flexibility, working memory capacity, and control capabilities, as observed in our normal subjects (51
); on the other hand, in manifest schizophrenia, which is characterized by an inefficient, fractionated pattern of activation in prefrontal cortex (33
), the same information-processing constellation could facilitate the persistence of disorganized cortical information and contribute to an escape of dysfunctional, unmodulated information from frontostriatal loops, leading to deficient cognition and inappropriate behavioral responses (16
). Stated another way, the common DARPP-32 haplotype appears associated with optimized frontal-striatal function regardless of the specific information being processed through the system. The molecular processes by which genetic and environmental information regulate the development and modification of prefrontal-striatal circuitry involve synaptic plasticity, and DARPP-32 is a key molecule for shaping plasticity in striatal neurons receiving frontal afferents (52
). Further work is necessary to confirm or refute this speculation.
In summary, we present convergent evidence in 3 independent datasets implicating DARPP-32 in a frontostriatal neural system for executive cognition and response selection in humans. We hope that this genetic identification of a molecular target at a critical nexus of dopaminergic neurotransmission and synaptic plasticity gives renewed impetus to the pursuit of therapeutic strategies aimed at postsynaptic signal integration in dopaminoceptive neurons that might benefit a diverse range of psychiatric disorders, notably addiction and schizophrenia.