Reciprocal Alterations in Regulator of G Protein Signaling 4 and microRNA16 in Schizophrenia

Abstract

N-methyl-d-aspartate receptor (NMDAR) hypofunction in the dorsolateral prefrontal cortex (DLPFC) has been implicated in the pathology of schizophrenia. NMDAR activity is negatively regulated by some G protein–coupled receptors (GPCRs). Signaling through these GPCRs is reduced by Regulator of G protein Signaling 4 (RGS4). Thus, lower levels of RGS4 would enhance GPCR-mediated reductions in NMDAR activity and could contribute to NMDAR hypofunction in schizophrenia. In this study, we quantified RGS4 mRNA and protein levels at several levels of resolution in the DLPFC from subjects with schizophrenia and matched healthy comparison subjects. To investigate molecular mechanisms that could contribute to altered RGS4 levels, we quantified levels of small noncoding RNAs, known as microRNAs (miRs), which regulate RGS4 mRNA integrity after transcription. RGS4 mRNA and protein levels were significantly lower in schizophrenia subjects and were positively correlated across all subjects. The RGS4 mRNA deficit was present in pyramidal neurons of DLPFC layers 3 and 5 of the schizophrenia subjects. In contrast, levels of miR16 were significantly higher in the DLPFC of schizophrenia subjects, and higher miR16 levels predicted lower RGS4 mRNA levels. These findings provide convergent evidence of lower RGS4 mRNA and protein levels in schizophrenia that may result from increased expression of miR16. Given the role of RGS4 in regulating GPCRs, and consequently the strength of NMDAR signaling, these findings could contribute to the molecular substrate for NMDAR hypofunction in DLPFC pyramidal cells in schizophrenia.

Introduction

N-methyl-d-aspartate receptor (NMDAR) hypofunction¹ within the circuitry of the dorsolateral prefrontal cortex (DLPFC) has been implicated in the pathology contributing to cognitive impairments in schizophrenia.² NMDAR-mediated neuronal activity is reduced, in part, by activation of certain G protein–coupled receptors (GPCRs) associated with G_αi/o and G_αq.^3–6 Signaling of these GPCRs is arrested by Regulator of G protein Signaling 4 (RGS4),^7–9 a GTPase-activating protein that accelerates the hydrolysis of GTP to GDP.^8,9 Accordingly, reductions in RGS4 augment the negative regulation of NMDAR signaling.^4,5 Hence, lower RGS4 levels could represent a molecular mechanism contributing to the hypofunctional NMDAR state thought to exist in schizophrenia.

Studies of RGS4 mRNA or protein levels in DLPFC gray matter from schizophrenia subjects have produced mixed results, with some finding lower levels^10–12 and others finding no differences relative to comparison subjects.^13–15 These apparent discrepancies could reflect differences in cohort characteristics or quantification methods across studies. In addition, since RGS4 exhibits prominent laminar and cellular patterns of expression in primate DLPFC,¹⁶ such that most labeled neurons are pyramidal cells in layers 2, 3, and 5, cellular level analyses might have greater sensitivity for the detection of a disease effect on RGS4 levels.

The molecular mechanisms that might lead to dysregulation of RGS4 expression in schizophrenia have also not been examined. For example, small noncoding RNAs, known as microRNAs (miRs), regulate the levels of hundreds of gene products.¹⁷ The activity of miRs have been implicated in brain development,^18,19 neurocognitive function,^20,21 and the pathogenesis of psychiatric disorders.^22–24 Indeed, by binding to and destabilizing their target transcripts, miRNAs can produce target degradation and indirectly reduce cognate protein levels by making less transcript available for translation.²⁵ Thus, several lines of evidence have suggested that RGS4 mRNA is a target of multiple miRs,²⁶ and some of these miRs (eg, miR15a, 15b, 16, 107, and 195) have been reported to be altered in the DLPFC from subjects with schizophrenia.^26–29

Consequently, in this study, we quantified RGS4 mRNA and protein levels in the DLPFC from matched pairs of schizophrenia and healthy comparison subjects, and RGS4 mRNA levels specifically in pyramidal cells located in DLPFC layers 3 or 5. In addition, we explored whether altered RGS4 mRNA and protein levels in schizophrenia could be due to altered expression of miRs by quantifying RGS4-targeting miRs within the DLPFC of the same schizophrenia and matched healthy comparison subjects.

Materials and Methods

Human Subjects

Brain specimens (n = 124) were obtained during autopsies conducted at the Allegheny County Medical Examiner’s Office (Pittsburgh, PA) after consent for donation was obtained from the next of kin. As previously described,³⁰ an independent committee of experienced research clinicians made consensus DSM-IV diagnoses³¹ for each subject. The same approach was used to confirm the absence of a psychiatric diagnosis in healthy comparison subjects. All procedures were approved by the University of Pittsburgh’s Committee for the Oversight of Research and Clinical Trials Involving the Dead and Institutional Review Board for Biomedical Research.

In order to control for experimental variance and to reduce biological variance between groups, each subject with schizophrenia (n = 39) or schizoaffective disorder (n = 23) was matched with 1 healthy comparison subject for sex and as closely possible for age, and tissue samples from both members in a pair were processed together in all experiments. Subject groups (n = 62 pairs) did not differ in mean age, postmortem interval (PMI), RNA integrity number (RIN; Agilent Bioanalyzer), or tissue storage time at −80°C (all t ₁₂₂ < 0.95, P > .34; table 1). Although brain pH significantly differed between diagnostic groups (t ₁₂₂ = 2.5, P = .01), the mean difference was 0.1 pH unit and of uncertain biological significance. Each subject had a RIN ≥7.0, indicating excellent RNA quality. For measures of RGS4 protein, only subjects with a PMI ≤20 hours were used due to prior immunoblot evidence of a PMI effect on RGS4 immunoreactivity.³² These subject groups (n = 23 pairs) did not differ in mean age, PMI, or tissue storage time at −30°C (table 1; all t ₄₄ < 1.6, P > .1). The subject groups used for the microarray analyses (n = 36 pairs) did not differ in mean age, PMI, pH, RIN, or tissue storage time at −80°C (table 1; all t ₇₁ < 3.36; all P > .07). Demographic details on individual subjects are presented in supplementary table S1.

Table 1.

Characteristics of the Subject Groups for Each Measure of RGS4

Characteristic	Quantitative PCR		Immunofluorescence		Microarray
	Comparison	Schizophrenia	Comparison	Schizophrenia	Comparison	Schizophrenia
Number	62	62	23	23	36	36
Sex	47M, 15 F	47M, 15 F	18M, 5 F	18M, 5 F	27M, 9F	27M, 9 F
Race	52W, 10 B	46W, 16 B	17W, 6 B	17W, 6 B	30W, 6 B	24W, 12 B
Age (y)	48.7±13.8	47.7±12.7	47.8±13.3	46.5±11.1	48.1±13.0	46.9±12.4
PMI (h)	18.8±5.5	19.2±8.5	13.6±4.4	11.4±4.8	17.6±6.1	18.0±8.8
Brain pH	6.7±0.2	6.6±0.3	6.8±0.2	6.5±0.3	6.7±0.2	6.6±0.4
RIN	8.2±0.6	8.1±0.6	8.4±0.6	8.2±0.6	8.3±0.6	8.2±0.6
Storage time (mo)	124.9±55.3	121.1±59.7	152.0±53.6	147.5±56.4	122.2±49.8	125.7±53.1

Note: Values are mean ± SD. PMI, postmortem interval; RIN, RNA integrity number.

Antipsychotic-Exposed Monkeys

Experimentally naive, male, young adult, long-tailed macaque monkeys (Macaca fascicularis) received oral doses of olanzapine, haloperidol, or placebo (n = 6 monkeys per group) twice daily for 17–27 months.³³ The dosage of antipsychotic medications administered to the animals produced trough serum levels in the therapeutic range for the treatment of schizophrenia,^33,34 and the tissue was processed as described previously.³⁵ All studies were carried out in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals and were approved by the University of Pittsburgh Institutional Animal Care and Use Committee.

Quantitative PCR

RGS4 mRNA and selected miRs were quantified in total gray matter from the right hemisphere DLPFC area 9 using quantitative PCR (qPCR) (supplementary methods). RGS4 mRNA was quantified as previously described.^36,37 Briefly, qPCR was performed to determine the cycle threshold (CT) for RGS4 and 3 internal reference transcripts³⁶ with Power SYBR Green master mix and ViiA 7 Real-Time PCR system (Life Technologies). For quantification of miRs, cDNA was synthesized using the Universal cDNA synthesis kit (Exiqon) and processed using microRNA LNA^TM primer set (Exiqon). qPCR was performed to determine CTs for all miRs and 2 internal reference transcripts (supplementary methods and supplementary figure 1) using Power SYBR Green (Exiqon) and StepOnePlus Real-Time PCR System (Life Technologies). Levels of miRs were first determined in 22 pairs of subjects, and then validation was conducted in an additional 38 pairs; 2 subject pairs were not available for the miR analyses due to the limited tissue section availability (supplementary table S1).

In the analyses of both mRNA and miRs by qPCR, the difference in CT for each target transcript was calculated by subtracting the geometric mean CT for the relevant internal reference transcripts from the mean CT of the target transcript. Because the difference in cycle threshold (dCT) represents the log₂-transformed expression ratio of each target transcript to the reference transcripts, the relative expression level of the target transcript is determined as 2^−dCT.^38,39

Microarray Analyses

RGS4 transcript levels were quantified within pyramidal cells as previously described.⁴⁰ Briefly, 200 pyramidal cells from each of DLPFC layers 3 and 5 were dissected (LMD 6500; Leica Microsystems) and then pooled together per layer per subject. Synthesized cDNA was loaded on Affymetrix GeneChip HT HG-U133⁺ PM Array (Affymetrix), which was designed to assess transcript levels from the human genome. Array results were not readable for layer 5 samples from 2 subjects, and thus those 2 subject pairs were excluded as previously described.⁴⁰ From antipsychotic-exposed monkeys, 150 pyramidal cells from each of DLPFC layers 3 and 5 were dissected. Microarray analysis was conducted using GeneChip® Rhesus Macaque Genome Array (Affymetrix).³⁵ In human and monkey samples, the microarray signals for all 3 probe sets targeting RGS4 were averaged within each sample.

Confocal Immunofluorescence Microscopy

Confocal immunofluorescence microscopy was performed to quantify RGS4 relative protein levels in the left hemisphere of DLPFC area 9 as previously described⁴¹ (supplementary methods) with minor modifications. Briefly, 2 sections per subject were incubated in rabbit anti-RGS4 (1:500, AbCam, Ab9964), then donkey anti-rabbit conjugated to Alexa 568 (1:500; Invitrogen). Images were collected on an Olympus BX51 microscope using a 10×0.40 NA. objective. On each section, two 660 µm-wide traverses extending from the pial surface to the layer 6-white matter border were imaged in a location where area 9 was cut perpendicular to the cortical surface. Coordinate points were set using a motorized stage such that the entire traverse was imaged. To avoid edge effects, the middle 330-µm-wide portion of each traverse was analyzed. The average traverse area for comparison (201.4mm² ± 6.3) and schizophrenia (203.4mm² ± 4.4) subjects did not differ (t ₄₄ = −1.3, P = .2). Channel exposure time was constant for each traverse such that no pixels were saturated and the dynamic range of the camera was filled.

Because channel exposure time was held constant for all samples, a minimum intensity thresholding method was used to mask RGS4-immunoreactive objects (SlideBook 5.0, Intelligent Imaging Innovations). Mean RGS4 fluorescence intensity was calculated by dividing the total RGS4 fluorescence intensity in all masked objects by the total sampled area. Lipofuscin autofluorescence was also quantified as previously described⁴¹ (supplementary methods).

Data Analysis and Statistics

We performed 2 ANCOVA models. For the first model, because subjects were selected and processed as pairs, we conducted a paired ANCOVA which included the relevant dependent variable, diagnostic group as the main effect, subject pair as a blocking factor, and tissue storage time and PMI as covariates. Subject pairing may be considered an attempt to balance diagnostic groups for sex and age, and to account for the parallel processing of tissue samples from a pair, and thus to not be a true statistical paired design. Thus, a second unpaired ANCOVA model was performed without subject pair as a blocking factor that included age, sex, tissue storage time, and PMI as covariates. For RNA-dependent measures, both ANCOVA models also included pH and RIN as covariates since these factors are known to reflect RNA quality.⁴²

The potential influence of other factors that are frequently comorbid with schizophrenia on each dependent measure was assessed using unpaired ANCOVA models. For these analyses, each comorbid factor (sex, diagnosis of schizoaffective disorder, death by suicide, history of substance dependence or abuse, tobacco use at the time of death, or use of antidepressants, benzodiazepines or sodium valproate, or antipsychotics at the time of death) was used as the main effect and age, tissue storage time, PMI, brain pH and RIN (RNA analyses only) as covariates.

For the antipsychotic-exposed monkey study, we employed ANOVA models with mRNA level as the dependent variable, treatment group as the main effect, and triad as a blocking factor.

Statistical tests were conducted with α = .05. The reported P values for correlations of RGS4 mRNA and protein levels are 1 tailed based on the hypothesis that protein abundance is regulated by mRNA levels. All other P values are 2 tailed.

Results

qPCR Analysis of RGS4 mRNA Levels in Schizophrenia

We previously reported that mean RGS4 mRNA levels were lower in the schizophrenia subjects from the first 42 subject pairs of our cohort.¹⁰ In the remaining 20 pairs, mean levels of RGS4 mRNA were also lower (figure 1) in the schizophrenia subjects (−11.2%, paired: F _1,15 = 4.1, P = .06; unpaired: F _1,32 = 5.2, P = .03). Across all 62 pairs, mean RGS4 levels were significantly 14.5% lower (paired: F _1,57 = 29.4, P < .001; unpaired: F _1,116 = 32.8, P < .001) in the schizophrenia subjects; RGS4 mRNA levels were lower in 51 of the 62 schizophrenia subjects relative to the matched comparison subject. Across all 62 schizophrenia subjects, RGS4 mRNA levels did not differ as a function of factors that are frequently comorbid with the disease (supplementary figure 2A).

Fig. 1.

Regulator of G protein Signaling 4 (RGS4) mRNA levels in dorsolateral prefrontal cortex area 9 for each matched pair of schizophrenia and subject in the new cohort of 20 pairs (gray triangles), and the previously reported cohort of 42 pairs (black triangles). Markers below the diagonal unity line indicate lower RGS4 mRNA expression in the subject with schizophrenia relative to the matched comparison subject.

Confocal Microscopy Analysis of RGS4 Protein Levels in Schizophrenia

To determine whether lower mean RGS4 transcript levels resulted in less RGS4 protein, the fluorescence intensity of RGS4 immunoreactivity was quantified by confocal microscopy. The robustness and laminar pattern of RGS4 labeling in human DLPFC (figure 2A) was consistent with a previous report from monkey DLPFC.¹⁶ Though RGS4-immunoreactive pyramidal or nonpyramidal neurons could be identified in nearly all layers, most RGS4 neurons were pyramidal cells in layers 2, 3, and 5 (figure 2B). Mean RGS4 relative protein levels were significantly 10.7% lower (paired: F _1,20 = 6.9, P = .02; unpaired: F _1,40 = 5.4, P = .03) in the schizophrenia subjects (figure 2C); RGS4 protein levels were lower in 15 of the 23 schizophrenia subjects relative to the matched comparison subject. Among the schizophrenia subjects, RGS4 relative protein levels did not differ as a function of factors that are frequently comorbid with the disease (supplementary figure 2B). RGS4 mRNA and protein levels were significantly positively correlated (r = .28, P = .03) across all subjects (figure 2D) despite being quantified in different hemispheres (ie, right and left, respectively) with different tissue storage methods (ie, frozen and paraformaldehyde-fixation, respectively) and different quantification techniques. RGS4 mRNA and protein levels were not correlated within healthy comparison (r = .12, P = .3) or schizophrenia (r = .2, P = .2) subjects when considered separately, probably due to the smaller range of values within each subject group.

Fig. 2.

(A) A ×10 magnification image showing the predominant pyramidal cell labeling of regulator of G protein Signaling 4 (RGS4). (B) A ×60 magnification image showing a layer 3 pyramidal cell labeled for RGS4. Note the labeling of apical and basilar dendrites, including primary and secondary branches. As expected, lipofuscin autofluorescence (blue) was predominately identified in and near cell bodies. (C) Relative RGS4 protein levels in dorsolateral prefrontal cortex area 9. Markers below the diagonal unity line indicate lower RGS4 protein expression in the subject with schizophrenia relative to the matched comparison subject. (D) Correlation of RGS4 mRNA and protein levels in healthy comparison (filled) and schizophrenia (open) subjects. Scale bar is 50 µm in A and 10 µm in B.

Microarray Analysis of RGS4 mRNA Levels in Pyramidal Cells in Schizophrenia

Using data derived from our previous microarray study,⁴⁰ 2 of 3 RGS4 probes were significantly reduced in layer 3 and layer 5 pyramidal cells in schizophrenia subjects at a false discovery rate of 0.05. We also averaged the expression of the 3 RGS4 probes for a composite value for each subject. Mean levels of RGS4 mRNA were significantly lower in the schizophrenia subjects in pyramidal cells from layer 3 (−14.7%; paired: F _1,31 = 7.7, P = .009; unpaired: F _1,64 = 10.0, P = .002), and layer 5 (−17.4%; paired: F _1,29 = 14.0, P = .001; unpaired: F _1,60 = 17.2, P < .0001). Levels of RGS4 mRNA were lower for 28 of the 36 schizophrenia subjects in layer 3 (figure 3A) and for 25 of the 34 schizophrenia subjects in layer 5 (figure 3B) relative to the matched comparison subject. These findings are similar to the 16% reduction in mean RGS4 mRNA levels in total gray matter as determined by qPCR for the same 36 pairs of subjects. RGS4 mRNA levels quantified by qPCR in total gray matter were significantly correlated with levels in pyramidal cells of layer 3 (r = .65, P < .001) and layer 5 (r = .57, P < .001).

Fig. 3.

RGS4 mRNA levels in dorsolateral prefrontal cortex area 9 pyramidal cells of layer 3 (A) and layer 5 (B). Markers below the diagonal unity line indicate lower pyramidal cell RGS4 mRNA expression in the subject with schizophrenia relative to the matched comparison subject.

Analysis of RGS4 mRNA Levels in PFC Gray Matter and Pyramidal Cells in Antipsychotic-Exposed Monkeys

RGS4 mRNA levels quantified by qPCR analysis in DLPFC gray matter did not differ among monkeys chronically exposed to haloperidol, olanzapine, or placebo (F _2,10 = 0.31, P = .7). In gray matter, RGS4 mRNA levels were nonsignificantly 2.6% higher in haloperidol-treated monkeys and nonsignificantly 0.5% higher in olanzapine-treated monkeys relative to placebo-treated monkeys. RGS4 mRNA levels quantified by microarray analysis in DLPFC pyramidal cells did not differ among monkeys chronically exposed to haloperidol, olanzapine, or placebo in either layer 3 (F _2,10 = 3.3, P = .08) or layer 5 (F _2,10 = 0.45, P = .7). In layer 3 pyramidal cells, RGS4 mRNA levels were nonsignificantly 4.8% lower in haloperidol-treated monkeys and nonsignificantly 13% lower in olanzapine-treated monkeys relative to placebo-treated monkeys. In layer 5 pyramidal cells, RGS4 mRNA levels were nonsignificantly 5.2% lower in haloperidol-treated monkeys and nonsignificantly 0.7% lower in olanzapine-treated monkeys relative to placebo-treated monkeys.

qPCR Analysis of miR Expression in Schizophrenia

We quantified the levels of 5 miRs (15a, 15b, 16, 107, and 195) previously implicated in the regulation of RGS4 transcript levels.²⁶ In an initial analysis of 22 subject pairs (which had 17.8% lower RGS4 mRNA levels in the schizophrenia subjects; paired: F _1,17 = 11.5, P = .003; unpaired: F _1,36 = 16.9, P < .001), miR16 was 14.7% higher in the schizophrenia subjects (paired: F _1,17 = 5.2, P = .04; unpaired: F _1,36 = 4.3, P = .04). In contrast, levels of the other 4 miRs did not differ between subject groups (supplementary figure 3; paired: all F _1,107 < 1.5, P > .2; unpaired: all F _1,36 < 1.0, P > .3). In an additional 38 subject pairs, miR16 was also significantly 12% higher (paired: F _1,33 = 17.0, P < .001; unpaired: F _1,68 = 7.2, P = .009). Across all 60 pairs, the mean level of miR16 was 13.1% higher in the schizophrenia subjects (paired: F _1,55 = 21.1, P < .001; unpaired: F _1,112 = 9.4, P = .003); miR16 levels were higher in 44 of the 60 schizophrenia subjects relative to the matched comparison subject (figure 4A). Levels of miR16 did not differ as a function of other factors that are frequently comorbid with the disease (supplementary figure 2C).

Fig. 4.

(A) miR16 levels in dorsolateral prefrontal cortex area 9. Markers below the diagonal unity line indicate lower miR16 expression in the subject with schizophrenia relative to the matched comparison subject. (B) Correlation of miR16 and RGS4 mRNA levels in healthy comparison subjects. (C) Correlation of miR16 and RGS4 mRNA levels in schizophrenia subjects.

In order to exclude false negative findings in the initial 22 subject pair analysis, we also measured miR195 in the additional 38 subject pairs, as previous studies have reported both higher and lower miR195 expression levels in schizophrenia.^26,28,29 Across all 60 pairs, mean levels of miR195 expression did not differ (paired: F _1,55 = 0.04, P = .84; unpaired: F _1,112 = 0.006, P = .94) between subject groups.

Given this evidence that higher levels of miR16 might be specific to the disease process of schizophrenia, we examined the relationship between miR16 and RGS4 mRNA (figures 4B and 4C). Levels of miR16 were significantly negatively correlated (r = −.45, P < .001) with levels of RGS4 mRNA across all subjects examined. This correlation was driven largely by the relationship in the schizophrenia subjects (r = −.53, P < .001), as miR16 and RGS4 mRNA levels were not correlated in healthy comparison subjects (r = −.14, P = .3).

Analysis of miR16 Levels in PFC Gray Matter in Antipsychotic-Exposed Monkeys

Levels of miR16 in DLPFC gray matter did not differ among monkeys chronically exposed to haloperidol, olanzapine, or placebo (F _2,10 = 0.11, P = .9). In gray matter, miR16 levels were nonsignificantly 1.6% lower in haloperidol-treated monkeys, and nonsignificantly 1.8% lower in olanzapine-treated monkeys relative to placebo-treated monkeys.

Discussion

In this study, we utilized complementary approaches to address several questions regarding the potential role of altered RGS4 signaling in DLPFC dysfunction in schizophrenia. First, we found that levels of RGS4 mRNA were significantly lower in the right DLPFC of a large cohort of subjects with schizophrenia, suggesting that lower RGS4 mRNA expression might be a common feature of the illness. Indeed, RGS4 mRNA levels were lower in 82% of the 62 subjects with schizophrenia studied. Second, consistent with these transcript findings, measures of RGS4 protein were also significantly lower in the left DLPFC of the same subjects with schizophrenia. The presence of abnormalities in both hemispheres suggests that a deficit in RGS4 is a conserved feature of the DLPFC in schizophrenia. Third, RGS4 mRNA levels were lower in both layer 3 and layer 5 pyramidal cells from schizophrenia subjects, and the magnitude of this reduction was similar to that seen in DLPFC gray matter. Fourth, levels of miR16 were higher in subjects with schizophrenia, and negatively correlated with RGS4 mRNA levels in these subjects.

Our findings of increased miR16 in the DLPFC of subjects with schizophrenia is in agreement with a previous study.²⁶ Although the present and previous^10–12 studies have found lower RGS4 mRNA and protein expression in the DLPFC of schizophrenia, other studies have found no differences between healthy comparison and schizophrenia subjects in the DLPFC.^13–15 Methodological and subject characteristics may contribute to these disparate findings. For example, some protein studies of RGS4 protein used antibodies that detect a band of higher molecular weight than is predicted for RGS4.^14,15 Subject characteristics that reflect both protein and mRNA quality were not reported, differed between diagnostic groups, or had values suggesting that the preservation of mRNA or protein was less than ideal in some previous studies.^13–15 However, given the present findings, additional investigations of RGS4 mRNA and protein and relevant miRs in new cohorts of subjects using robust methods are needed.

Several observations suggest that lower levels of RGS4 mRNA and protein, and higher levels of miR16, in schizophrenia are likely to reflect the disease process rather than to be a consequence of factors frequently associated with the illness. First, none of the comorbid factors examined in the schizophrenia subjects accounted for the differences in levels of RGS4 or miR16. Second, relative to placebo-exposed animals, RGS4 mRNA expression was unaltered in DLPFC gray matter and layers 3 and 5 pyramidal cells from monkeys chronically exposed to oral haloperidol or olanzapine at doses that produced serum levels in the therapeutic range for the treatment of schizophrenia and were associated with changes in brain volume similar to those seen in schizophrenia.^33,43,44 These findings are consistent with a prior report that RGS4 mRNA expression was unaltered in DLPFC gray matter in monkeys chronically exposed to intramuscular injections of haloperidol decanoate that produced high serum levels of the drug and extrapyramidal signs requiring pharmacological treatment.¹¹ Third, miR16 expression in DLPFC gray matter did not differ among monkeys chronically exposed to haloperidol, olanzapine, or placebo. Fourth, the disease specificity of the miR16 finding was also suggested by the absence of alterations in other miRs that regulate RGS4.

Together, our data suggest that higher expression of miR16 contributes to lower levels of RGS4 mRNA, and consequently protein, in the DLPFC of subjects with schizophrenia. By binding to the AU-rich elements in the 3’ untranslated region of RGS4,^45,46 miR16 destabilizes RGS4 mRNA, leading to its degradation.^26,47 Thus, with less mRNA available, less RGS4 protein is translated.²⁵ While specific upstream regulators of cortical miR16 expression have yet to be investigated, damaging DNA increases miR16 expression in human fibroblasts^48,49 and maternal deprivation increases miR16 in the hippocampus.⁵⁰ Although our findings are consistent with miR16 as an upstream factor accounting for lower RGS4 levels in schizophrenia, RGS4 levels can be lowered through other mechanisms. For example, genetic ablation of brain-derived neurotrophic factor (BDNF) during adulthood reduces RGS4 levels,⁵¹ and findings consistent with lower BDNF signaling in the DLPFC have been reported in schizophrenia.^52–55 Moreover, as any given miR is predicted to regulate hundreds of gene products,¹⁷ increased miR16 in the PFC almost certainly affects numerous other transcripts. Application of recently developed algorithms for the in silico prediction of miR targets⁵⁶ to specific tests of human PFC tissue will provide essential data for interpreting the functional consequences of alterations in miRs in schizophrenia.

The impact of lower RGS4 on GPCR signaling in the DLPFC of schizophrenia subjects is likely to be strongest in pyramidal cells, where RGS4 is predominately expressed.¹⁶ Because GPCRs have distinct cellular and subcellular localizations in the DLPFC,^57–61 lower RGS4 expression could be a key factor in altering the signaling of GPCRs that are localized to pyramidal cell dendrites. Moreover, RGS4 is located postsynaptic to Type 1, presumably glutamatergic, axodendritic synapses in DLPFC pyramidal neurons.¹⁶ Thus, RGS4 is specifically positioned to affect GPCR-mediated regulation of NMDAR signaling,⁶² given the preferential location of NMDARs at glutamatergic inputs. The GPCRs known to regulate NMDAR signaling via RGS4 are α1- and α2-adrenergic receptors (α1/α2-ARs)⁵ and the serotonin 1A receptor (5HT_1AR),⁴ both of which are present in PFC pyramidal cell dendrites. Activation of these GPCRs in PFC pyramidal cells reduces NMDAR-mediated currents,^4,5 and lowering RGS4 availability enhances their ability to further reduce NMDAR signaling.^4,5

Thus, the present findings provide convergent evidence suggesting that increased expression of miR16 leads to lower RGS4 mRNA levels and less translation of RGS4 protein which could contribute, in addition to other mechanisms,⁶³ to NMDAR hypofunction in schizophrenia in the absence of changes in NMDAR levels.^64–70 The preferential localization of RGS4 to pyramidal cells is also consistent with recent studies indicating that NMDAR hypofunction is more likely to be mediated by NMDAR signaling on pyramidal cells than on the parvalbumin class of interneurons.^71,72 However, it should be noted that some studies suggest that schizophrenia is associated with direct, endogenous NMDAR antagonism,^73–75 and chronic administration of NMDAR antagonists can directly lower RGS4 levels in the rodent PFC.^4,76 Thus, we cannot exclude the possibility that lower RGS4 levels are a consequence of NMDAR antagonism in the disease process. Finally, as experimental reductions in cortical RGS4 expression can enhance signaling of some GPCRs,⁴ we cannot exclude the possibility that lower RGS4 levels are a compensatory mechanism to restore disease-associated reductions in GPCR signaling.

Importantly, the current findings inform future studies in experimental models that can determine whether higher miR16 and lower RGS4 levels in schizophrenia are likely to reflect cause, consequence, or compensation in the disease process. For example, determining the disease-relevant upstream mechanisms that could lead to increased miR16 levels, the degree to which higher miR16 can lower mRNA and protein levels of RGS4 and other signaling molecules, and whether and how these manipulations affect NMDAR-mediated signaling in PFC pyramidal cells will be essential for interpreting the roles of altered miR16 and RGS4 expression in schizophrenia.

Supplementary Material

Supplementary material is available at http://schizophreniabulletin.oxfordjournals.org.

Funding

This work was supported by the National Institute of Mental Health (MH043784 and MH084053 to D.A.L., MH096985 to K.N.F., and MH100066 to D.W.V.) and by Nara Medical University (S.K.). D.L. currently receives investigator-initiated research support from Pfizer