¹Department of Neurobiology University of Pittsburgh School of Medicine, Pittsburgh, PA 15261, USA

²Department of Psychiatry University of Pittsburgh School of Medicine, Pittsburgh, PA 15261, USA

³Department of Neuroscience University of Pittsburgh School of Medicine, Pittsburgh, PA 15261, USA

⁴Department of PittArray University of Pittsburgh School of Medicine, Pittsburgh, PA 15261, USA

Correspondence to: K Mirnics MD, Dept of Psychiatry, University of Pittsburgh, E1602 BST, Pittsburgh, PA 15261, USA. E-mail: karoly+@pitt.edu

Complex defects in neuronal signaling may underlie the dysfunctions that characterize schizophrenia. Using cDNA microarrays, we discovered that the transcript encoding regulator of G-protein signaling 4 (RGS4) was the most consistently and significantly decreased in the prefrontal cortex of all schizophrenic subjects examined. The expression levels of ten other RGS family members represented on the microarrays were unchanged and hierarchical data analysis revealed that as a group, 274 genes associated with G-protein signaling were unchanged. Quantitative in situ hybridization verified the microarray RGS4 data, and demonstrated highly correlated decreases in RGS4 expression across three cortical areas of ten subjects with schizophrenia. RGS4 expression was not altered in the prefrontal cortex of subjects with major depressive disorder or in monkeys treated chronically with haloperidol. Interestingly, targets for 70 genes mapped to the major schizophrenia susceptibility locus 1q21-22 were present on the microarrays, of which only RGS4 gene expression was consistently altered. The combined data indicate that a decrease in RGS4 expression may be a common and specific feature of schizophrenia, which could be due either to genetic factors or a disease- specific adaptation, both of which could affect neuronal signaling. Molecular Psychiatry (2001) 6, 293-301.

schizophrenia; major depression; antipsychotic; haloperidol; gene expression; regulator of G-protein signaling; RGS4; microarray; prefrontal; cerebral cortex; susceptibility gene; 1q21-22

Introduction

Schizophrenia is a devastating brain disorder that affects ~1% of the population.^1,2 The etiology of the disease remains elusive, with studies implicating the effects of both genetic and environmental factors on neurodevelopmental processes.^3,4,5 The dorsal prefrontal cortex (PFC) is a prominent site of dysfunction in schizophrenia,^6,7,8 where convergent lines of evidence have reported alterations in both pre- and postsynaptic elements.^{6,7,9,10,11,12,13}

Regulators of G-protein signaling (RGS), a recently discovered family of proteins,^14,15 play a crucial role in modulating signaling through G-protein pathways.^16,17 RGS proteins attach to GTP-bound G-alpha proteins with variable substrate specificity. By acting as GTP-ase activating proteins (GAPs), RGS proteins shorten the duration of G-protein mediated intracellular signaling. At least 20 RGS family members have been identified, and initial expression data indicate that subsets of RGS proteins are expressed in virtually all cells.^16,17 In the brain, distinct anatomical distributions of different family members have been observed,¹⁸ although RGS expression patterns have not been linked to specific neurotransmitter systems.¹⁹ In addition, G-protein coupled neurotransmitter receptors (GPCRs) that signal through RGS-modulated G-alpha complexes, such as the dopamine D₂ or metabotropic glutamate receptors, are also expressed in restricted, though overlapping, neuronal subpopulations.^20,21,22,23 Many of these RGS-modulated GPCRs are the targets of antipsychotic drugs.^24,25,26

We used cDNA microarrays to investigate potential alterations in transcript expression of six pairs of schizophrenic subjects and non-affected controls. We determined that RGS4 is the most consistently changed individual transcript of over 7800 genes represented on the microarray, whereas other RGS family members and G-protein signaling components, as a group, were unchanged. In situ hybridization was used to verify the RGS4 microarray data, to include additional subject comparisons, and to examine the regional and disease-related specificity of this change. RGS4 maps to locus 1q21-22 (NCBI accession number U27768, UniGene cluster Hs.227571), a chromosome region strongly linked to schizophrenia,²⁷ and thus is a candidate for a major schizophrenia susceptibility gene on this locus.

Material and methods

Characteristics of subjects

Two groups of schizophrenic and matched control subjects were used in these studies. Except for pair 557c (male)-537s (female), subject pairs were matched for sex (17 males and five females). The mean (± SD) difference within pairs was 4.6 ± 3.5 years for age and 4.4 ± 2.7 h for postmortem interval (PMI). The entire group of schizophrenic and control subjects did not differ in mean (± SD) age at time of death (46.5 ± 10.7 and 45.1 ± 11.5 years, respectively), PMI (19.4 ± 7.1 and 17.7 ± 5.0 h, respectively), brain pH (6.85 ± 0.29 and 6.81 ± 0.15, respectively), or tissue storage time at -80°C (45.4 ± 12.3 and 37.7 ± 13.1 months, respectively) when the studies were initiated. Nine of 11 schizophrenic subjects were receiving antipsychotic medications at the time of death, five had a history of alcohol abuse or dependence, and one died by suicide. We also studied 10 subjects with major depressive disorder (MDD) without psychosis, each matched to one control subject. These subject pairs were also matched for sex (18 males and two females). The mean (± SD) difference within pairs was 1.2 ± 1.4 years for age and 2.5 ± 2.1 h for PMI. The subject groups did not differ in mean (± SD) age at time of death (52.7 ± 13.1 and 52.1 ± 13.1 years, respectively), PMI (14.9 ± 5.3 and 15.7 ± 5.5 h, respectively), brain pH (6.81 ± 0.17 and 6.72 ± 0.30), or tissue storage time at -80°C (39.0 ± 17.4 and 39.9 ± 13.2 months, respectively). Two of the depressed subjects had a history of alcohol dependence, and six died by suicide. Two of the control subjects had also been matched to the subjects with schizophrenia (685c, 604c). Consensus DSM-IIIR diagnoses were made for all subjects using data from clinical records, toxicology studies and structured interviews with surviving relatives, as described in detail previously.²⁸

Microarray experiments

Methods of tissue preparation, nucleic acid isolation, sample labeling, microarray hybridization and data analysis were reported previously.¹⁰ Briefly, 200 ng mRNA was reverse transcribed using Cy3- or Cy5-labeled fluorescent primers; appropriate matched control and schizophrenic sample pairs were combined, and hybridized onto the same UniGEM-V cDNA microarray (Incyte Genomics, Fremont, CA, USA). Each UniGEM-V array contained over 7800 unique and sequence verified cDNA elements. If a gene or EST was differentially expressed, the cDNA feature on the array bound more of the labeled probe from one sample than the other, producing either a greater cy3 or Cy5 signal intensity. Microarrays were scanned under Cy3 and Cy5 fluorescence, and the resulting images were independently analyzed for signal intensity. The signal intensities were balanced to standards and compared using Incyte's proprietary software, GemTools. The operators performing the labeling, hybridization, scanning and signal analysis were blind to the specific category to which each sample belonged.

Gene expression criteria: Based on Incyte's control hybridization studies and our control experiments,¹⁰ array data reliability and reproducibility cutoffs were established as follows:

Gene group analysis: Of genes represented on the microarray, a G-protein group of 274 genes was created¹⁰ for data analysis, and included transcripts for GPCRs, heterotrimeric G-protein subunits, RAS proteins, RGS family members, and G-protein dependent inward rectifying K⁺ channels. The 1999 NCBI database human 1q21-22 map is represented by 70 genes on the microarray, although some of them are not expressed in the CNS. These genes (including RGS4 and RGS5) were also analyzed together as a gene group.¹⁰

RGS4 sequences: The RGS4 microarray bait sequence matched the entry in the NCBI database (accession number U27768, UniGene cluster Hs.227571). Of the 800-bp mRNA, the double-stranded DNA microarray bait was complementary to the 3' region of 571 nucleotides. The antisense, in situ hybridization probe was derived from the mRNA region spanning nucleotides 39-739, resulting in a 700-nucleotide long cRNA probe (see below). Northern hybridization experiments confirmed the specificity of the sequence-verified antisense riboprobe across multiple brain regions.

In situ hybridization

Tissue blocks containing the regions of interest (PFC area 9, motor cortex (MC) and visual cortex (VC)) were identified using surface landmarks and sulci (the superior frontal gyrus, the central sulcus and precentral gyrus, and the calcarine sulcus, respectively). After histological verification of the regions, 20-m sections containing these regions were cut with a cryostat at -20°C, mounted onto gelatin-coated glass slides, and stored at -80°C until use.

To generate cRNA probes, double stranded cDNA containing the RGS4 sequence was first amplified from normal human brain cDNA using custom-designed primers (Forward primer sequence: CCGAAGCCACA GCTCCTC; Reverse primer sequence: CATCCC TCTCCCTTCAGGTG) in a standard PCR reaction. Following cloning of the PCR product and sequence verification,¹⁰ [³⁵S]-labeled riboprobes were synthesized. During hybridization, approximately 3 ng of probe (~2 ´ 10⁶ DPM) were used per slide in a total volume of 90 l. All other methods used were described previously.^10,29 Following hybridization and washing, slides were air dried and exposed to BioMax MR film (Kodak) for 8-22 h and then dipped in emulsion (NTB-2, Kodak), and exposed for 3-5 days at 4°C. High-resolution scans of each film image were used for quantification of signal with Scion Image (version 4.0 b), and darkfield images were captured from the developed slides. Slides were coded so that the investigator performing the analysis was blinded to the diagnosis of the subjects, with subject pairs always processed in parallel. Three sections per subject per area were analyzed. Hybridization of sections with sense RGS4 riboprobe did not result in detectable signal.

RGS4 hybridization signal was virtually absent in the white matter of human and monkey sections, with both cortical and subcortical labeling patterns comparable to that shown in rat previously.¹⁸ Thus, quantification was performed on film autoradiograms by subtracting the OD of the white matter from the average signal OD measured in five non-overlapping rectangular regions on each section (three sections per tissue block). In PFC and MC, these rectangular regions spanned cortical layers II-VI. Due to the lack of RGS4 signal in layer IV throughout the neocortex, and the great expansion of this layer in VC, we analyzed the supragranular and infragranular signal ODs separately in VC. There were no significant differences in the levels of signal contained in the supra- and infragranular layers, so they were combined as a measure of overall VC signal intensity. The ODs from matched schizophrenic and control subjects were used to calculate relative expression differences between subject pairs.

Monkey experiments: Four pairs of male cynomolgus (Macaca fascicularis) monkeys, matched for age and weight, were studied. Frontal cortex from one pair was compared on cDNA microarray and all four pairs were analyzed by in situ hybridization for RGS4 expression in areas 9 and 46. In each pair, one animal was treated for 9-12 months with the antipsychotic medication haloperidol decanoate. Serum levels were in the therapeutic range for the treatment of schizophrenia.³⁰ Extrapyramidal symptoms were effectively managed by maintenance administration of benztropine mesylate.

Comparison strategies and statistical analyses

Microarray analysis: Experimental subjects were compared to control subjects in a pairwise design to control for the effects of age, race, sex and PMI on gene expression. A gene group was changed between the control and schizophrenic subjects if it reported differential expression by both Chi-square and t-test compared to the mean and distribution of all expressed genes on the six microarrays.¹⁰

In situ hybridization analysis: In situ hybridization data were analyzed using an analysis of covariance (ANCOVA) with diagnosis as the main effect and brain pH and tissue storage time as covariates. The analysis was performed both with and without subject pair as a blocking factor. Since both models produced similar results, only the values from the ANCOVA with subject pair as a blocking factor are reported. Changes between groups were also analyzed by descriptive statistics, Pearson correlation and Factor analysis. We elected to use semi-quantitative analysis of tissue sections processed for in situ hybridization, rather than other methods, in order to gain important anatomical information on the cellular and regional expression patterns of the RGS4 transcript in different human cortical regions. In several subject comparisons that had sufficient tissue for additional harvesting, RGS4 transcript levels assessed by quantitative RT-PCR showed the same robust decreases as the microarray and in situ hybridization (data not shown).

Results

We used high-density cDNA microarrays (UniGEM-V, Incyte Genomics) to examine the expression patterns of over 7800 genes and expressed sequence tags in postmortem samples of PFC area 9 from six matched pairs of schizophrenic and control subjects. Our initial microarray findings¹⁰ focusing on hierarchical data analysis reported that the group of genes encoding proteins involved in presynaptic secretory function were consistently changed across schizophrenic subjects.

RGS expression studies in the PFC of schizophrenic subjects

Of the 4.8% of differentially expressed genes, RGS4 transcript was the only gene decreased at the 99% CL in all schizophrenic subjects. Across the six microarray comparisons these decreases ranged from 50% to 84% (Figure 1).

To verify the microarray findings for the decrease in RGS4 expression, we performed in situ hybridization on the PFC from the same five subject pairs used for the microarray experiments (for pair 794c/665s, sections were not available from the same block of tissue used in the microarray experiment). As a further test of the robustness of our microarray data, we added five subject pairs to the in situ hybridization analysis. In control subjects, RGS4 labeling was heavy in the PFC (Figure 2a), mimicking previously described labeling in the rat.¹⁸ The labeling of large and small cells was most prominent in layers III and V, with sparse labeling in the intervening granular layer IV. White matter labeling was virtually indistinguishable from off-section background. Based on optical density analysis, 9/10 subject pairs exhibited a 10.2-74.3% decrease in PFC RGS4 expression (Figure 2b). Across all ten subject pairs, the mean level of RGS4 expression was significantly decreased (-34.4%) in the schizophrenic subjects (F_1,15 = 6.95; P = 0.019). This decrease was comparable in schizophrenic subjects with and without any history of alcohol abuse (t-test; P = 0.125).

Previous studies have shown that significant cell loss does not occur in the cerebral cortex or PFC of schizophrenic subjects.^31,32 Thus, the reduced RGS4 labeling that we report here is almost certainly due to decreased transcript levels in individual neurons.

RGS4 expression in the PFC of subjects with major depressive disorder (MDD)

We next examined whether the decreased expression of RGS4 transcript was associated with another psychiatric disorder. In situ hybridization analysis of PFC area 9 in ten subjects with MDD revealed no difference in mean levels of RGS4 (+3.5%) expression (F_1,15 = 0.27; P = 0.61) compared to matched controls (Figure 2c).

Changes in RGS family members, RGS substrates and G-proteins

To investigate whether the decrease in RGS4 transcript reflects a component-specific change in G-protein signaling in schizophrenia, we analyzed the microarray data for consistent gene expression changes across other RGS-family members (Figure 1). Nine of the 11 RGS family members represented on the microarrays were expressed in four or more of the microarray comparisons. Other than RGS4, none reported consistent expression changes across schizophrenic subjects.

Several reports suggest that changes in heterotrimeric G-proteins, the main substrates for RGS family members,^16,33 are associated with schizophrenia^34,35,36 and D₂ radioligand binding is consistently higher in schizophrenic patients.^37,38,39 Of the eight G-alpha RGS substrates represented on the microarrays, only G-olf expression was changed beyond the 95% CL in three or more pairwise comparisons (data not shown). Interestingly, these three subjects with increased G-olf levels (317s, 547s and 622s) showed the most robust decrease in RGS4 expression in both the PFC microarray and in situ hybridization analyses.

Using gene group comparison methods, we examined the expression of 274 genes known to be involved in G-protein signaling cascades. A mean of 105 genes belonging to this group were expressed per subject pair. At the 99% CL, 2.8% of G-protein transcripts were decreased, while 2.8% were increased in the PFC of schizophrenic subjects (Figure 3a). Gene group analysis by Chi-square test (P = 0.12) and t-test (P = 0.62) closely matched the distribution of all expressed genes, confirming that the different expression levels in the G-protein group are attributable to normal human variability, and not to the disease.

Expression changes of microarray-represented genes located on chromosome 1q21-22

The RGS4 gene maps to the schizophrenia susceptibility locus 1q21-22.²⁷ To address whether any other genes at this locus have altered expression in the PFC of schizophrenic subjects, we analyzed 70 microarray-represented transcripts that map to this cytogenetic region (Figure 3b). An average of 40 genes were expressed in each comparison, and at the 99% CL, 0.4% of 1q21-22 genes were increased, and 5.9% were decreased in the schizophrenic subjects. Of all genes on the 1q21-22 locus, only RGS4 showed a consistent expression change across all the pairwise comparisons over the 99% CL. Of the remaining genes on this locus, only the ALL1-FUSED gene reported, in three pairwise comparisons, a consistent expression change over the 95% CL in the schizophrenic subjects. The expression of the remaining genes on locus 1q21-22 showed the same overall pattern as genes located on non-schizophrenia loci (data not shown).

Influence of chronic neuroleptic exposure on RGS4 expression

In our sample of 10 subjects with schizophrenia, two were not receiving antipsychotic medications at the time of death (622s and 537s). Both of these subjects still showed decreased expression of RGS4. Using an animal model, we probed the potential of haloperidol to directly modulate RGS4 transcript levels (Figure 4). In situ hybridization analysis in four matched pairs of chronically treated and control cynomolgus monkeys³⁰ showed no difference in PFC RGS4 expression (mean = +5.3%; P = 0.26). Similarly, a cDNA microarray comparison in one monkey pair confirmed that haloperidol treatment did not alter RGS4 gene expression in the frontal cortex (0% change).

Regional RGS4 gene expression changes

Recent studies suggest that changes in schizophrenia involve functional cortical deficiencies in multiple brain areas (for review, see Harrison⁴⁰). To examine whether there is a more widespread cortical deficiency in RGS4 transcript, we assessed by in situ hybridization RGS4 expression in visual (VC) and motor (MC) cortex from the same 10 pairs of control and schizophrenic subjects (for pair 558c/317s MC material was not available, and this pair was substituted with pair 794c/665s).

In VC, RGS4 showed heavy labeling in the gray matter, with a very prominent bilaminar pattern in the supragranular and infragranular layers (Figure 5a), but sparse labeling in layer IV. Combined RGS4 expression in the supragranular and infragranular layers of VC was decreased by 32.8% (F_1,15 = 8.24; P = 0.012) (Figure 5b).

In MC, RGS4 expression was concentrated over cell-rich layers II-III and V-VI of both control and schizophrenic subjects (Figure 6a). Layer IV is very attenuated in MC, and exhibited virtually no RGS4 labeling. The mean RGS4 expression (Figure 6b) was decreased by 34.2% across the 10 schizophrenic subjects (F_1,15 = 10.18; P = 0.006).

Within the subjects with schizophrenia RGS4 expression was consistently decreased across the PFC, VC and MC (Figure 7). Factor analysis of the pairwise differences in RGS4 gene expression across the three different cortical areas for all nine common schizophrenic and control subject pairs revealed that over 84% of the total variance in expression was accounted for by diagnosis (variance proportion = 0.848, eigenvalue = 2.544, P = 0.001).

Discussion

Gene microarray analysis of complex brain diseases provides an opportunity to discover specific alterations in gene expression among thousands of transcripts that may not yet be associated with a particular disorder.⁴¹ In combination with in situ hybridization analysis, the present study reveals a robust and consistent change in levels of the RGS4 transcript in schizophrenia, and the absence of significant transcript changes in other RGS or G-protein signaling components represented on the microarrays. The analysis of another psychiatric disorder, MDD, showed that the cortical RGS4 transcript decrease is not associated with all psychiatric disorders. Given the role of RGS4 in regulating signaling through the G-protein cascade,¹⁶ the present findings, together with our recent finding of decreases in the expression of genes regulating presynaptic secretory function,¹⁰ suggest that schizophrenic subjects may have pronounced alterations in synaptic responsiveness in multiple brain areas.

Microarray technology is by design unbiased,^42,43 with no requirement for a specific null-hypothesis, but with the ability to rapidly generate massive datasets. This data-driven approach revealed consistent changes in RGS4 mRNA that we did not predict to be associated with schizophrenia a priori. The unchanged G-protein and 1q21-22 gene groups on the microarrays also highlight the specificity and robustness of the RGS4 findings, not easily assessed by conventional techniques. While it is likely that the robust changes in RGS4 transcript expression reflect an altered functional state, direct analysis of protein levels and activity is warranted. In our preliminary studies in cortical tissue, all currently available RGS4 antibodies failed to show requisite specificity against RGS4 on Western blots or in immunohistochemistry experiments, using human or monkey cortical tissue (data not shown). We are in the process of producing new antibodies.

RGS4 modulation of neural networks

Neurons co-express many different GPCRs and receptor signaling is coupled via multiple G-alpha proteins. RGS4 is situated at the point of convergence of many Gi, G-olf and Gq-coupled signaling pathways, potentially regulating the activity of multiple neurotransmitter systems.^44,45,46,47 Consequently, the specific biological effects of a widespread RGS4 decrease are difficult to predict. First, functional outcome will be influenced both by downstream gene expression changes and adaptations at multiple transcript levels. Second, functional changes will depend upon intrinsic differences among affected cell populations. For example, RGS family members are co-expressed in certain neurons, but in different overall patterns.^16,18 Thus, other RGS family members may compensate for decreased RGS4 function in some neurons, but not in others. Moreover, cells expressing RGS4 may exhibit different patterns of putative substrate G-alpha proteins.^48,49 Third, RGS4 is heterogeneously expressed in the CNS,¹⁸ so that deficits in RGS4 expression could result in region-specific effects. Fourth, the effect of RGS4 deficiency on specific neural circuitry also will depend on the balance between altered inhibitory and excitatory signaling.

The known pharmacological effects of antipsychotic drugs are consistent with RGS4 serving as a regulator of GPCR-mediated signaling in the neocortex, and more specifically with D₂ receptor overactivity.³⁹ Typical antipsychotic agents, including haloperidol, are potent and relatively selective antagonists for Gi-coupled D₂-like receptors.^23,25,50,51 Atypical antipsychotic compounds (eg clozapine, olanzapine, risperidone), in addition to D₂-like receptor binding, have antagonistic effects on multiple neurotransmitter receptors.^26,52 As a result, atypical antipsychotics may inhibit Gi and Gq-dependent signaling in a much larger neuronal pool than typical antipsychotic agents. We hypothesize that blockade of multiple GPCRs could compensate for the reduced ability of RGS4 to act as a GAP in schizophrenia, and thus inhibit putatively hyperactive Gi/Gq-mediated signaling.

Potential mechanisms leading to altered RGS4 expression

We suggest two potential mechanisms to explain the consistent RGS4 expression decrease in schizophrenia. First, since schizophrenia is a multifactorial disease with multiple gene loci implicated in the inheritance pattern,^27,53 the decreased RGS4 expression across various cortical areas may reflect an inherited susceptibility factor for schizophrenia. This view, to some degree, is supported by our finding that of the 70 genes mapped to the 1q21-22 locus that were investigated on the microarrays, only RGS4 exhibited altered expression across multiple schizophrenic subjects. However, sequence-based analysis of the coding region of RGS4 in these schizophrenic subjects did not report informative polymorphisms or mutations (data not shown). Nonetheless, to establish if RGS4 is a susceptibility gene for schizophrenia, a more extensive genomic analysis will be required.

Second, the RGS4 decrease could be a schizophrenia-associated adaptive response, perhaps to compensate for a reduction in synapse number or function. Such an epigenetic influence is consistent with the second decade onset of the disease.¹ In childhood, high synaptic density could compensate for impaired synaptic efficacy.^54,55 In the cortex, normal synaptic pruning ends in late adolescence,^56,57 eliminating exuberant numbers of synapses. In schizophrenia, neurons could down-regulate RGS4 in an attempt to restore pre-pruning levels of GPCR signaling, resulting in less RGS4-dependent GAP activity and prolonged signaling through the G-protein pathways. Furthermore, in the neonatal ventral hippocampal lesion model of schizophrenia,⁵⁸ RGS4 expression in the PFC of adult rats was prominently decreased.⁵⁹

Regardless of the specific mechanisms of the decrease in RGS4 gene expression in schizophrenia, the consistency and robustness of our findings suggest that deficits in RGS4 expression are directly related to the more conserved symptoms of the disease, and thus represent a novel and promising target for therapeutic intervention.

Acknowledgements

We are grateful to the colleagues who read and commented on earlier versions of this manuscript, as well as Dr J Pierri for his involvement in the chronic haloperidol treatment of monkeys. The research was supported by projects 1 (DAL) and 2 (PL, KM) of NIMH Center Grant MH45156 (DAL), an endowment fund from the RK Mellon Foundation (PL) and NIMH training grant T32 MH18273 (FM).

1 Lewis D, Lieberman J. Catching up on schizophrenia. Neuron 2000; 28: 325-334, MEDLINE

2 Hyman SE. The NIMH perspective: next steps in schizophrenia research. Biol Psychiatry 2000; 47: 1-7, MEDLINE

3 Lewis DA. Is there a neuropathology of schizophrenia? The Neuroscientist 2000; 6: 208-218,

4 Pilowsky LS, Kerwin RW, Murray RM. Schizophrenia: a neurodevelopmental perspective. Neuropsychopharmacology 1993; 9: 83-91, MEDLINE

5 Weinberger DR. From neuropathology to neurodevelopment. Lancet 1995; 346: 552-557, MEDLINE

6 Weinberger DR, Aloia MS, Goldberg TE, Berman KF. The frontal lobes and schizophrenia. J Neuropsychiatry Clin Neurosci 1994; 6: 419-427, MEDLINE

7 Goldman-Rakic PS, Selemon LD. Functional and anatomical aspects of prefrontal pathology in schizophrenia. Schizophr Bull 1997; 23: 437-458, MEDLINE

8 Glantz LA, Lewis DA. Decreased dendritic spine density of prefrontal cortical pyramidal neurons in schizophrenia. Arch Gen Psychiatry 2000; 57: 65-73, MEDLINE

9 Pettegrew JW, Keshavan MS, Panchalingam K et al. Alterations in brain high-energy phosphate and membrane phospholipid metabolism in first-episode, drug-naive schizophrenics. A pilot study of the dorsal prefrontal cortex by in vivo phosphorus 31 nuclear magnetic resonance spectroscopy. Arch Gen Psychiatry 1991; 48: 563-568, MEDLINE

10 Mirnics K, Middleton F, Marquez A, Lewis D, Levitt P. Molecular characterization of schizophrenia viewed by microarray analysis of gene expression in prefrontal cortex. Neuron 2000; 28: 53-67, MEDLINE

11 Knable MB, Weinberger DR. Dopamine, the prefrontal cortex and schizophrenia. J Psychopharmacol 1997; 11: 123-131, MEDLINE

12 Bunney WE, Bunney BG. Evidence for a compromised dorsolateral prefrontal cortical parallel circuit in schizophrenia. Brain Res Rev 2000; 31: 138-146, MEDLINE

13 Aghajanian GK, Marek GJ. Serotonin model of schizophrenia: emerging role of glutamate mechanisms. Brain Res Rev 2000; 31: 302-312, MEDLINE

14 Berman DM, Kozasa T, Gilman AG. The GTPase-activating protein RGS4 stabilizes the transition state for nucleotide hydrolysis. J Biol Chem 1996; 271: 27209-27212, MEDLINE

15 Berman DM, Wilkie TM, Gilman AG. GAIP and RGS4 are GTPase-activating proteins for the Gi subfamily of G protein alpha subunits. Cell 1996; 86: 445-452, MEDLINE

16 De Vries L, Zheng B, Fischer T, Elenko E, Farquhar MG. The regulator of G protein signaling family. Annu Rev Pharmacol Toxicol 2000; 40: 235-271, MEDLINE

17 Zheng B, De Vries L, Gist Farquhar M. Divergence of RGS proteins: evidence for the existence of six mammalian RGS subfamilies. Trends Biochem Sci 1999; 24: 411-414, MEDLINE

18 Gold SJ, Ni YG, Dohlman HG, Nestler EJ. Regulators of G-protein signaling (RGS) proteins: region-specific expression of nine subtypes in rat brain. J Neurosci 1997; 17: 8024-8037, MEDLINE

19 De Vries L, Gist Farquhar M. RGS proteins: more than just GAPs for heterotrimeric G proteins. Trends Cell Biol 1999; 9: 138-144, Article MEDLINE

20 Hoyer D, Clarke DE, Fozard JR et al. International Union of Pharmacology classification of receptors for 5-hydroxytryptamine (serotonin). Pharmacol Rev 1994; 46: 157-203, MEDLINE

21 Westbrook GL. Glutamate receptor update. Curr Opin Neurobiol 1994; 4: 337-346, MEDLINE

22 Kerr DI, Ong J. GABAB receptors. Pharmacol Ther 1995; 67: 187-246, Article MEDLINE

23 Missale C, Nash SR, Robinson SW, Jaber M, Caron MG. Dopamine receptors: from structure to function. Physiol Rev 1998; 78: 189-225, MEDLINE

24 Seeman P, Chau-Wong M, Tedesco J, Wong K. Brain receptors for antipsychotic drugs and dopamine: direct binding assays. Proc Natl Acad Sci U S A 1975; 72: 4376-4380, MEDLINE

25 Creese I, Burt DR, Snyder SH. Dopamine receptor binding predicts clinical and pharmacological potencies of antischizophrenic drugs. Science 1976; 192: 481-483, MEDLINE

26 Lieberman JA, Mailman RB, Duncan G et al. Serotonergic basis of antipsychotic drug effects in schizophrenia. Biol Psychiatry 1998; 44: 1099-1117, MEDLINE

27 Brzustowicz LM, Hodgkinson KA, Chow EW, Honer WG, Bassett AS. Location of a major susceptibility locus for familial schizophrenia on chromosome 1q21-q22. Science 2000; 288: 678-682, Article MEDLINE

28 Volk DV, Austin MC, Pierri JN, Sampson RA, Lewis DA. Decreased GAD67 mRNA expression in a subset of prefrontal cortical GABA neurons in schizophrenia. Arch Gen Psychiatry 2000; 57: 237-245, MEDLINE

29 Campbell DB, North JB, Hess E. Tottering mouse motor dysfunction is abolished on the Purkinje cell degeneration (pcd) mutant background. Exp Neurol 1999; 160: 268-278, Article MEDLINE

30 Pierri JN, Chaudry BS, Woo TUW, Lewis DA. Alterations in chandelier neuron axon terminals in the prefrontal cortex of schizophrenic subjects. Am J Psychiatry 1999; 156: 1709-1719, MEDLINE

31 Pakkenberg B. Total nerve cell number in neocortex in chronic schizophrenics and controls estimated using optical disectors. Biol Psychiatry 1993; 11: 768-772,

32 Thune JJ, Pakkenberg B. Stereological studies of the schizophrenic brain. Brain Res Brain Res Rev 2000; 31: 200-204, MEDLINE

33 Guan KL, Han M. A G-protein signaling network mediated by an RGS protein. Genes Dev 1999; 13: 1763-1767, MEDLINE

34 Nishino N, Kitamura N, Hashimoto T et al. Increase in [3H]cAMP binding sites and decrease in Gi alpha and Go alpha immunoreactivities in left temporal cortices from patients with schizophrenia. Brain Res 1993; 615: 41-49, MEDLINE

35 Yang CQ, Kitamura N, Nishino N, Shirakawa O, Nakai H. Isotype-specific G protein abnormalities in the left superior temporal cortex and limbic structures of patients with chronic schizophrenia. Biol Psychiatry 1998; 43: 12-19, MEDLINE

36 Okada F, Tokumitsu Y, Takahashi N, Crow TJ, Roberts GW. Reduced concentrations of the alpha-subunit of GTP-binding protein Go in schizophrenic brain. J Neural Transm Gen Sect 1994; 95: 95-104, MEDLINE

37 Wong DF, Wagner HN Jr, Tune LE et al. Positron emission tomography reveals elevated D2 dopamine receptors in drug-naive schizophrenics [published erratum appears in Science 1987 Feb 6; 235: 623]. Science 1986; 234: 1558-1563, MEDLINE

38 Seeman P, Ulpian C, Bergeron C et al. Bimodal distribution of dopamine receptor densities in brains of schizophrenics. Science 1984; 225: 728-731, MEDLINE

39 Abi-Dargham A, Rodenhiser J, Printz D et al. From the cover: increased baseline occupancy of D2 receptors by dopamine in schizophrenia. Proc Natl Acad Sci U S A 2000; 97: 8104-8109, MEDLINE

40 Harrison PJ. The neuropathology of schizophrenia. A critical review of the data and their interpretation. Brain 1999; 122: 593-624, Article MEDLINE

41 Whitney LW, Becker KG, Tresser NJ et al. Analysis of gene expression in multiple sclerosis lesions using cDNA microarrays. Ann Neurol 1999; 46: 425-428, MEDLINE

42 Brown PO, Botstein D. Exploring the new world of the genome with DNA microarrays. Nat Genet Suppl 1999; 21: 33-37,

43 Geschwind DH. Mice, microarrays, and the genetic diversity of the brain. Proc Natl Acad Sci U S A 2000; 97: 10676-10678, MEDLINE

44 Popov SG, Krishna UM, Falk JR, Wilkie TM. Ca²⁺/calmodulin reverses PIP3-dependent inhibition of RGS GAP activity. J Biol Chem 2000; 25: 18962-18968,

45 Cavalli A, Druey KM, Milligan G. The regulator of G protein signaling RGS4 selectively enhances alpha2A adrenoceptor stimulation of the GTPase activity of Go1alpha and Gi2alpha. J Biol Chem 2000; 275: 23693-23699, MEDLINE

46 Xu X, Zeng W, Popov S et al. RGS proteins determine signaling specificity of Gq-coupled receptors. J Biol Chem 1999; 274: 3549-3556, MEDLINE

47 Greengard P, Allen PB, Nairn AC. Beyond the dopamine receptor: the DARPP-32/protein phosphatase-1 cascade. Neuron 1999; 23: 435-447, MEDLINE

48 Friberg IK, Young AB, Standaert DG. Differential localization of the mRNAs for the pertussis toxin insensitive G-protein alpha sub-units Gq, G11, and Gz in the rat brain, and regulation of their expression after striatal deafferentation. Mol Brain Res 1998; 54: 298-310, MEDLINE

49 Aoki C, Go CG, Wu K, Siekevitz P. Light and electron microscopic localization of alpha subunits of GTP-binding proteins, G(o) and Gi, in the cerebral cortex and hippocampus of rat brain. Brain Res 1992; 596: 189-201, MEDLINE

50 Stockmeier CA, DiCarlo JJ, Zhang Y, Thompson P, Meltzer HY. Characterization of typical and atypical antipsychotic drugs based on in vivo occupancy of serotonin2 and dopamine2 receptors. J Pharmacol Exp Ther 1993; 266: 1374-1384, MEDLINE

51 Seeman P, Tallerico T. Rapid release of antipsychotic drugs from dopamine D2 receptors: an explanation for low receptor occupancy and early clinical relapse upon withdrawal of clozapine or quetiapine. Am J Psychiatry 1999; 156: 876-884, MEDLINE

52 Meltzer HY. The role of serotonin in antipsychotic drug action. Neuropsychopharmacology 1999; 21: 106S-115S, Article MEDLINE

53 Pulver AE. Search for schizophrenia succeptability genes. Biol Psychiatry 2000; 47: 221-230, Article MEDLINE

54 Pettegrew JW, Keshavan MS, Minshew NJ. 31P nuclear magnetic resonance spectroscopy: neurodevelopment and schizophrenia. Schizophr Bull 1993; 19: 35-53, MEDLINE

55 Feinberg I. Schizophrenia: caused by a fault in programmed synaptic elimination during adolescence? J Psych Res 1982; 17: 317-334,

56 Huttenlocher P. Synaptic density in human frontal cortex¾developmental changes and effects of aging. Brain Res 1979; 163: 195-205, MEDLINE

57 Bourgeois JP, Goldman-Rakic PS, Rakic P. Formation, elimination and stabilization of synapses in the primate cerebral cortex. In: Gazzaniga M (ed). The Cognitive Neurosciences. Cambridge University Press: Cambridge, 2001 (in press,

58 Sams-Dodd F, Lipska BK, Weinberger DR. Neonatal lesions of the rat ventral hippocampus result in hyperlocomotion and deficits in social behaviour in adulthood. Psychopharmacol (Berl) 1997; 132: 303-310,

59 Marcotte ER, Quirion R, Srivastava LK. Gene expression changes in adult prefrontal cortex and nucleus accumbens following neonatal ventral hippocampal lesions. SFN Abstr 2000; 385: 14,

Figure 1 RGS changes in the PFC across six cDNA microarray pairwise comparisons between matched schizophrenic and control subjects. The x axis reports subject pairs (boxes), the y axis reports percent change between schizophrenic (S) and control (C) subjects. Individual symbols represent a gene expression difference between a schizophrenic and control subject in a single pairwise comparison. Black dashed line denotes similar expression between schizophrenic and control subjects (0% change), green dashed line denotes the 95% CL (37.5% change), red dashed line represents 99% CL (47.5% change). Missing symbols in some pairwise comparisons indicate that those genes did not meet expression criteria in that experiment.

Figure 2 PFC RGS4 expression levels are decreased in nine of ten subjects with schizophrenia (SCH), but not in subjects with major depressive disorder (MDD). (a) Emulsion-dipped in situ hybridization micrographs of PFC tissue sections from a schizophrenic (622s) and matched control (685c) subject viewed under darkfield illumination. Roman numbers denote cortical layers. Note the strong labeling across all cortical layers except lamina IV, and the diminished labeling is absent. Scale bar = 400 m. (b) The in situ hybridization data from 10 PFC pairwise comparisons were quantified using film densitometry. The x axis represents subject classes, the y axis reports average film optical density from three repeated hybridizations, measured across all layers. Lines connecting symbols indicate a matched subject pair. Hash marks indicate group means. Note that in 10 PFC pairwise comparisons, nine schizophrenic subjects showed RGS4 transcript reduction (mean = -34.5%; F_1,15 = 6.95; P = 0.019). (c) Subjects with MDD exhibit comparable RGS4 expression levels as their matched control subjects (mean = +3.5%; F_1,15 = 0.27; P 0.61).

Figure 3 G-protein signaling and 1q21-22 locus-related expression differences in the PFC of schizophrenic and control subjects as revealed by microarray analysis. For both gene groups, all expressed genes were classified into signal intensity difference intervals (0.1 bins) according to their cy5/cy3 signal ratio. Transcripts in a '1' bin had identical Cy5 vs Cy3 signal intensities. Positive values (to the right) on the x axis denote higher Cy5 signal in schizophrenic subjects (S > C), negative values (to the left) correspond to higher Cy3 signal intensity in the control subjects (C > S). The y axis reports % of expressed genes across the six subject pairs per bin for each gene group. In both panels, the white bars (All genes) denote distribution of all expressed genes across the six PFC pairwise comparisons (n = 22 408). (a) Each microarray contained baits for 274 G-protein signaling related genes. Across the six microarray comparisons, an average of 105 G-protein signaling related genes were expressed. (b) Out of 70 microarray-represented 1q21-22 locus-related genes, 239 were detected across the six microarray comparisons. RGS4 contribution to the transcript distribution is denoted by a hatched bar. Note that in both (a) and (b), the Cy3/Cy5 signal distributions of G-protein and 1q21-22 gene groups were comparable to the distribution of all expressed genes across the six microarray comparisons.

Figure 4 PFC RGS4 expression is unchanged in four matched pairs of haloperidol-treated and control cynomolgus monkeys. Figure layout is the same as in Figure 2. (a) Emulsion-dipped in situ hybridization micrographs from the PFC of a haloperidol-treated (208 h) and matched control (207c) monkey viewed under darkfield microscopy. The PFC labeling pattern of RGS4 mRNA mimics human PFC labeling. Roman numbers denote cortical layers. Scale bar = 400 m. (b) PFC expression changes for four in situ pairwise comparisons of chronic haloperidol-treated and control monkeys. RGS4 OD measurements were almost identical across all four monkey comparisons, regardless of drug or placebo treatment (mean = +5.3%, P = 0.26).

Figure 5 RGS4 levels are significantly decreased in VC of the schizophrenic subjects. Figure layout as in Figure 2. (a) Emulsion-dipped in situ hybridization micrographs of VC tissue sections from the same matched pair of schizophrenic and control subjects represented in Figure 2a. Scale bar = 400 m. The weak labeling corresponds to layer IV, which is quite extensive in visual cortex. (b) In ten VC SCH pairwise comparisons, schizophrenic subjects also showed a significant reduction of RGS4 expression (mean = -32.8%; F_1,15 = 8.24; P = 0.012).

Figure 6 RGS4 levels are significantly decreased in MC of the schizophrenic subjects. Figure layout as in Figure 2. (a) Emulsion-dipped in situ hybridization micrographs of MC tissue sections from the same matched pair of schizophrenic and control subjects represented in Figure 2a and 6a, scale bar = 400 m. Because of the attenuated layer IV in motor cortex, the RGS4 labeling is almost uniform across all layers. (b) Schizophrenic subjects across the MC had significantly decreased RGS4 expression levels (-34.2%, F_1,15 = 10.18; P = 0.006).

Figure 7 Relative RGS4 expression changes across the experimental groups. The x axis represents experimental groups, the y axis reports percent RGS4 expression changes in the PFC, VC and MC of SCH subjects, as well as the PFC of subjects with MDD viewed by in situ hybridization. Each symbol represents percent of change in a single pairwise comparison; the same symbols represent the same subject pairs. Green lines denote mean expression difference for each group. The same schizophrenic subjects showed a comparable and highly correlated decrease in RGS4 expression across all three cortical regions (PFC-VC: r = 0.88, P = 0.0003; PFC-MC: r = 0.69, P = 0.0384; VC-MC: r = 0.76, P = 0.0144). In contrast, subjects with MDD reported variable RGS4 expression changes when compared to their matched controls.