OUP user menu

Gene expression profiling in treatment-naive schizophrenia patients identifies abnormalities in biological pathways involving AKT1 that are corrected by antipsychotic medication

Nishantha Kumarasinghe, Natalie J. Beveridge, Erin Gardiner, Rodney J. Scott, Surangi Yasawardene, Antoinette Perera, Jayan Mendis, Kanishka Suriyakumara, Ulrich Schall, Paul A. Tooney
DOI: http://dx.doi.org/10.1017/S1461145713000035 1483-1503 First published online: 1 August 2013


Distinct gene expression profiles can be detected in peripheral blood mononuclear cells (PBMCs) in patients with schizophrenia; however, little is known about the effects of antipsychotic medication. This study compared gene expression profiles in PMBCs from treatment-naive patients with schizophrenia before and after antipsychotic drug treatment. PBMCs were obtained from 10 treatment-naive schizophrenia patients before and 6 wk after initiating antipsychotic drug treatment and compared to PMBCs collected from 11 healthy community volunteers. Genome-wide expression profiling was conducted using Illumina HumanHT-12 expression bead arrays and analysed using significance analysis of microarrays. This analysis identified 624 genes with altered expression (208 up-regulated, 416 down-regulated) prior to antipsychotic treatment (p < 0.05) including schizophrenia-associated genes AKT1, DISC1 and DGCR6. After 6–8 wk treatment of patients with risperidone or risperidone in combination with haloperidol, only 106 genes were altered, suggesting that the treatment corrected the expression of a large proportion of genes back to control levels. However, 67 genes continued to show the same directional change in expression after treatment. Ingenuity® pathway analysis and gene set enrichment analysis implicated dysregulation of biological functions and pathways related to inflammation and immunity in patients with schizophrenia. A number of the top canonical pathways dysregulated in treatment-naive patients signal through AKT1 that was up-regulated. After treatment, AKT1 returned to control levels and less dysregulation of these canonical pathways was observed. This study supports immune dysfunction and pathways involving AKT1 in the aetiopathophysiology of schizophrenia and their response to antipsychotic medication.

Key words
  • AKT1 signalling
  • antipsychotic drugs
  • gene expression
  • peripheral blood mononuclear cells
  • schizophrenia


Schizophrenia is a mental illness with a high heritability contributing up to 83% of the risk for developing this disorder (Cannon et al., 1998; Jablensky and Kalaydjieva, 2003; Lewis et al., 2003; Sivagnansundaram et al., 2003). For many decades, researchers have strived to identify genes or loci that provide clues to the nature of this genetic risk; however, many of these findings have not been consistently replicated and the nature of the genetic risk remains elusive (Shi et al., 2009; Stefansson et al., 2009). Given the heterogeneity in the presentation and symptom profiles of patients diagnosed with schizophrenia and their overlap with other forms of psychiatric illness (Sivagnansundaram et al., 2003; Jablensky, 2006), it is not surprising that the genetic underpinning of the disorder has been difficult to identify.

To compensate for this, genetic studies of increasing magnitude have been conducted over the past decade, culminating in the largest genome-wide association study of schizophrenia reported in 2011. This study used 21 856 individuals of European ancestry in a discovery phase and 29 839 independent individuals in the replication phase (Ripke et al., 2011). Genome-wide significance was shown for the association of seven loci in schizophrenia, five of which were new (1p21.3, 2q32.3, 8p23.2, 8q21.3 and 10q24.32-q24.33) and two that had been previously reported at loci 6p21.32-p22.1 and 18q21.2 (Ripke et al., 2011). One outcome from the genetic association studies is that many genes of small effect size are likely to be involved, each contributing to disease risk. In support of this, many candidate genes have been identified, but their association remains inconsistent with regard to their disease contribution. Indeed, one genetic association study of a large schizophrenia cohort with European ancestry was unable to demonstrate support for the top 14 candidate genes, RGS4, DISC1, DTNBP1, STX7, TAAR6, PPP3CC, NRG1, DRD2, HTR2A, DAOA, AKT1, CHRNA7, COMT and ARVCF (Sanders et al., 2008).

At the molecular level, the transcriptome sits between environmental influence and the genetic susceptibility to schizophrenia and thus may provide a bridge between endophenotypes and the genetic changes that lead to the disorder. Indeed, many studies have reported gene expression changes in several brain regions in subjects with schizophrenia (Mirnics et al., 2000; Hakak et al., 2001; Hemby et al., 2002; Weidenhofer et al., 2006; Bowden et al., 2008). However, it is impractical to use biopsied brain tissue from schizophrenia patients for the development of a molecular signature that may assist with unravelling disease heterogeneity. In this regard, peripheral blood mononuclear cells (PBMCs) can be easily collected from patients and followed longitudinally with gene expression analyses, assisting in the identification of the signatures of clinical subtypes, their prognosis and perhaps treatment response.

A number of gene expression profiling studies using PBMCs have been conducted to date (see recent review, Kumarasinghe et al., 2012). Early studies showed that gene expression profiles can be detected in PBMCs that do indeed have some potential for the development of a diagnostic tool (Tsuang et al., 2005; Middleton et al., 2006) or for subtype classification (Bowden et al., 2006). A more recent study used a convergent functional genomics approach to identify blood biomarkers for hallucinations and delusions (Kurian et al., 2011). Whether gene expression signatures from PBMCs can be consistently detected that relate to abnormal brain function remains to be determined. Indeed, Sullivan et al. (2006) showed there was significant overlap in gene expression in the blood and brain and in our own study of PMBCs we demonstrated that 18 brain-related genes had altered expression in PMBCs in schizophrenia patients compared to non-psychiatric controls (Bowden et al., 2006).

Most schizophrenia studies are conducted while patients are on medication or have a history of pharmacotherapy. Gene expression profiling studies in PBMCs are no different; however, a number of studies have been conducted on PBMCs in treatment-naive participants. In 2005, gene expression profiling of PBMCs from 13 drug treatment-naive schizophrenia patients showed an increase in expression of the inwardly rectifying potassium channel (Kir2.3) and dopamine-3 receptor genes compared to matched healthy control volunteers (Zvara et al., 2005). Craddock et al. (2007) also used freshly isolated T cell-derived RNA and compared the gene expression from six minimally treated or first-episode schizophrenia patients with controls and identified changes in the expression of genes involved with cell cycle machinery, intracellular signalling, oxidative stress and metabolism. More recently, Takahashi et al. (2010) used neural network analysis of PBMCs from neuroleptic naive patients with schizophrenia and identified a 14 gene probe set as predictors of diagnosis, indicating that it might be possible to use this technology to diagnose and/or subtype patients with schizophrenia. The current study investigated gene expression in patients with schizophrenia before and 6–8 wk into antipsychotic pharmacotherapy.


Participant recruitment and cohort characterization

Ethical and safety clearance for the study was obtained from human research ethics committees of University of Newcastle (Australia), University of Sri Jayewardenepura Sri Lanka and National Institute of Mental Health (NIMH) Angoda, Sri Lanka. Patients gave informed written consent prior to participation in this study.

Ten treatment-naive schizophrenia patients of Sinhalese ancestry meeting DSM-IV criteria for schizophrenia (Castle et al., 2006) were recruited from the out-patient department of the NIMH between January 2007 and July 2009 (Table 1). Previous antipsychotic pharmacotherapies, low global IQ (< 70), a history of alcohol or illicit drug use, a neurological (e.g. epilepsy, traumatic brain injury) or chronic medical condition (e.g. diabetes) or pregnancy precluded participants from the study. Eleven pair-wise age-matched healthy volunteers of Sinhalese ancestry were recruited through the Family Practice Centre of the University of Sri Jayewardenepura (Sri Lanka; Table 1). Gender group differences were examined with χ2 test (Yates-corrected), age group differences with Mann–Whitney U test and symptom change pre vs. post treatment in patients with Wilcoxon's signed rank test. Healthy volunteers were screened for a history of mental illness (including their first-degree biological relatives), a history of alcohol or illicit drug use, a neurological or chronic medical condition, low global IQ (< 70) or pregnancy and excluded from participation accordingly. General examination (history/physical examination) and routine full blood counts completed at the out-patients department did not reveal any infection or other systemic causes for altered immune state in any of the participants.

View this table:
Table 1

Characteristics of participants in the study

PatientsAge (yr)GenderFamily history of schizophrenia*Duration of illness at inclusion (months)Antipsychotic medicationDuration of medication (wk)BPRS total score (before medication)BPRS total score (after medication)Healthy controlsAge (yr)Gender
SZ430MaleNot present8Risp/Halo65344C135Female
SZ630MaleNot present6Risp64836C538Male
SZ722MaleNot present6Risp74939C619Male
SZ947FemaleNot present11Risp65944C848Female
  • Risp, Risperidone; Halo, haloperidol.

  • * Determined by case records/family history/social workers records.

  • Age of patients vs. controls: Z = 0 p = 1.0.

  • Gender group differences: Yates χ2 = 1.4 (d.f. = 1) p = 0.24.

  • Brief Psychiatric Rating Scale (BPRS) total score of patients pre vs. post treatment: Z = −2.7 p = 0.007.

Patients were rated on the Brief Psychiatric Rating Scale (BPRS) at study inclusion and referred for standard clinical care as determined and supervised by the treating physician. Seven patients were commenced on 4 mg/d risperidone and three patients received a combination of 2 mg/d risperidone and 2 mg/d haloperidol, thus equalling a total daily dose of 200 mg chlorpromazine equivalents (Woods, 2003) in both instances. Six to eight weeks into pharmacotherapy, patients underwent another clinical rating using BPRS.

Blood was collected between 07:00 and 09:00 hours from patients and controls into PAXgene Blood RNA tubes (approx. 2.5 ml; PreAnalytiX GmbH, Switzerland) then frozen at −80°C. A second blood sample was collected from patients after antipsychotic treatment into PAXgene Blood RNA tubes and frozen at −80°C. Samples were transported to the University of Newcastle on dry ice and stored briefly at −80°C until processed.

RNA purification from PBMC

RNA was purified using the PAXgene blood RNA Kit (manual extraction method) as per manufacturer's instructions (Qiagen, USA). Briefly, PAXgene tubes were incubated for 2 h then centrifuged for 10 min/4000 g. Then 4 ml RNase-free water was added to the pellet and centrifuged for 10 min/4000 g. The pellet was dissolved in 350 µl BR1 suspension buffer and 300 µl binding buffer and 40 µl proteinase K was added, mixed and incubated for 10 min at 55°C. The lysate was then passed through a PAXgene Shredder spin column for 3 min at maximum speed in a microfuge. The flow through supernatant was collected and added to 350 µl 100% ethanol and vortexed. Samples were transferred to the PAXgene RNA spin column and centrifuged for 1 min at full speed, after which the column was washed with 350 µl BR3 wash buffer. The DNA bound to the column was digested with 80 µl DNase I mixture at room temperature for 15 min and the column washed with BR3 wash buffer and then twice with buffer BR4. The PAXgene RNA spin column was placed into a fresh 1.5 ml tube and the RNA eluted with 40 µl elution buffer BR5. The elution step was repeated and the sample incubated for 5 min at 65°C, chilled and then stored at −20°C until required for microarray analysis. RNA quantification was carried out by using the Quant-iT RNA assay kit with the Qubit fluorometer (Life Technologies, USA). RNA integrity was assessed using an Experion bioanalyser (Bio-Rad, USA) according to manufacturer's instructions. All samples had RNA quality indicator values that were considered to be in the acceptable range according to the manufacturer's guidelines.

Microarray analysis

Purified RNA was amplified and biotinylated using the Illumina TotalPrep amplification kit (Life Technologies) according to the manufacturer's protocol. Briefly, 500 ng RNA was reversed transcribed in a two-stage process using an oligo-dT primer that bears a T7 promoter. The resulting cDNA was column purified and used as a template for in vitro transcription with T7 RNA polymerase and biotin-UTP. The amplified and biotinylated RNA product was likewise column purified prior to hybridization to the array. Then 750 ng labelled RNA was hybridized to Illumina HT-12_V3 beadchips (∼48 000 probes) according to the manufacturer's protocol. Expression data underwent quality control analysis and normalization using the GenomeStudio data analysis software v2009.1 (Illumina, USA). Briefly, quality control assessed the Direct Hyb control plots within the GenomeStudio software. All control plots displayed expected values as per the Illumina specifications. Control measures included hybridization controls, negative and background controls, biotin-, low- and high-stringency controls, housekeeping gene intensities and average gene intensities. Data were background subtracted and normalized using the cubic spline method within GenomeStudio according to Illumina's recommendations. Genes were considered expressed if fluorescence intensity was twice that of background. Differential expression analysis was performed using the significance analysis of microarrays (SAM) v2.23 (Tusher et al., 2001). Initially, data analysis was performed on probe lists. Upon the generation of gene lists for pathways analysis (described below), the probe lists were collapsed into gene lists.

Quantitate real-time reverse transcription PCR (qPCR) validation

Validation of differentially expressed genes was performed by qPCR as previously described (Beveridge et al., 2010). Briefly, 500 ng sample RNA was treated with DNAse I (Invitrogen; Life Technologies) and reverse transcription performed with Superscript II reverse transcriptase (Invitrogen; for primer/probe sequences, see Supplementary Table S1). Seven genes were chosen for validation on the basis of significant differential expression on the microarray and/or biological relevance to schizophrenia. Triplicate reactions were set up in a 96-well format using the epMotion 5070 automated pipetting system (Eppendorf, Germany) and performed using the Applied Biosystems 7500 real-time PCR machine. Serial dilutions of PBMC cDNA were used as standards and data were analysed with the relative quantitation method with efficiency correction. Relative gene expression was calculated as the ratio of the gene and the geometric mean of controls hydroxymethylbilane synthase and β-glucuronidase. These housekeeping genes were chosen since they were stably expressed in a separate study of PBMCs in a cohort of 83 participants (Gardiner et al., 2012). qPCR data were not normally distributed (Kolmogorov–Smirnov test); therefore, the significance of differential expression was determined using the Mann–Whitney U-test (GraphPad Prism v5.0).

Pathways and gene set enrichment analysis

Lists of differentially expressed genes (with corresponding fold-changes/p values) were analysed using Ingenuity Pathway Analysis (IPA) v6.3 (Ingenuity Systems, USA). A gene set analysis was conducted on the entire gene expression data set using the Gene Set Enrichment Analysis software (GSEA; http://www.broadinstitute.org/gsea/index.jsp; Mootha et al., 2003; Subramanian et al., 2005). The gene expression data sets for the controls and patients were submitted to the GSEA software and analysed against the BioCarta-derived (http://www.biocarta.com/genes/index.asp) gene sets using 1000 permutations. The software generated lists of gene sets enriched by phenotype (either schizophrenia before or after treatment against control) and those lists with a normalized p value < 0.05 and a false discovery rate measured by q < 0.1 are reported in Results.


Changes in symptom ratings after antipsychotic drug (APD) treatment

All patients met DSM-IV criteria for schizophrenia. The patient and control groups did not differ significantly in age and gender, while patients' psychopathology significantly improved with antipsychotic pharmacotherapy as rated with the BPRS [total score: 46.3 (s.d. = 9.1) at study entry compared to 39.5 (s.d. = 5.6; z=−2.7; p = 0.007) 6–8 wk into antipsychotic pharmacotherapy; Table 1].

Gene expression changes before and after APD treatment

The genome-wide expression profiling of 48 803 transcripts (comprising 25 202 genes) from PBMC revealed 10 207 genes as being expressed. SAM analysis identified 208 up-regulated and 416 down-regulated genes in patients prior to antipsychotic treatment when compared to controls (i.e. control vs. schizophrenia before treatment analysis; p < 0.05; Supplementary Table S2). This list included the schizophrenia-associated genes AKT1 (v-akt murine thymoma viral oncogene homologue 1), DISC1 (disrupted in schizophrenia 1; both up-regulated) and DGCR6 (DiGeorge syndrome critical region gene 6; down-regulated).

Six to eight weeks into antipsychotic drug treatment, gene expression in patients was compared to the control sample and only 106 genes were identified as dysregulated (p < 0.05), with six up-regulated and 102 down-regulated (Supplementary Table S3). When these genes were compared to the gene expression data prior to treatment (Supplementary Table S1), the expression of 11 down-regulated genes in schizophrenia was corrected after treatment, returning them to control levels, suggesting an expression change in response to antipsychotic medication (Table 2). In addition, 63 genes remained down-regulated before and after treatment when compared to control levels (Supplementary Table S4), whereas four genes (i.e. G6PD, F5, RNF144B and TIMP2) remained significantly up-regulated in schizophrenia 6–8 wk into pharmacotherapy (1.4–2.8 fold-change range; p < 0.03 for all genes). Gene expression was then directly compared in the schizophrenia patients before and then 6–8 wk into antipsychotic treatment. No gene was significantly down-regulated by treatment and 28 genes were significantly up-regulated (p < 0.05; Table 3).

View this table:
Table 2

Genes down-regulated in SZ patients before treatment and significantly up-regulated in response to antipsychotic drug treatment by array analysis

Illumina IDGene symbolEntrez gene nameCon vs. SZ BTCon vs. SZ AT
aFold change (SZ/Con)p value*Fold change (SZ/Con)p value*
ILMN_2225887ATP5EP2ATP synthase, H+ transporting, mitochondrial F1 complex, epsilon subunit pseudogene 2−1.780.0012.050.020
ILMN_1687279DHPSDeoxyhypusine synthase−1.450.0481.300.001
ILMN_2404625LATLinker for activation of T cells−1.730.0012.220.001
ILMN_1813175LPHN1Latrophilin 1−1.820.0141.980.001
ILMN_1754303RPL30Ribosomal protein L30−1.870.0011.710.017
ILMN_1746516RPS25Ribosomal protein S25−1.950.0011.680.001
ILMN_1736135TYW1tRNA-yW synthesizing protein 1 homologue (S. cerevisiae)−1.630.0482.030.020
ILMN_2110281UFC1Ubiquitin-fold modifier conjugating enzyme 1−1.670.0201.450.015
ILMN_2232936UQCRHUbiquinol-cytochrome c reductase hinge protein−1.790.0011.480.015
  • SZ, Schizophrenia; Con, control; BT, before treatment; AT, after treatment.

  • * p values determined using significance analysis of microarrays.

View this table:
Table 3

Direct comparison of gene expression in PBMCs taken from patients with schizophrenia before and then after antipsychotic drug treatment

Illumina IDGene symbolEntrez gene nameFold change (SZ/Con)p value*
ILMN_2344956ACP1Acid phosphatase 1, soluble2.320.001
ILMN_1733581C16orf35Nitrogen permease regulator-like 3 (S. cerevisiae)1.890.001
ILMN_1707356CFL2cofilin 2 (muscle)1.920.019
ILMN_1687279DHPSDeoxyhypusine synthase1.300.001
ILMN_1715636EIF3BEukaryotic translation initiation factor 3, subunit B1.360.014
ILMN_1678618ELAVL3ELAV (embryonic lethal, abnormal vision, Drosophila)-like 3 (Hu antigen C)2.200.022
ILMN_1771185EME2Essential meiotic endonuclease 1 homologue 2 (S. pombe)2.190.022
ILMN_1677483EXOSC1Exosome component 11.510.015
ILMN_1804938GPR175Transmembrane protein, adipocyte associated 12.080.001
ILMN_1776412KRTAP10−11Keratin associated protein 10–113.430.034
ILMN_2404625LATLinker for activation of T cells2.220.001
ILMN_1776283LGALS12Lectin, galactoside-binding, soluble, 122.190.022
ILMN_1813175LPHN1Latrophilin 11.980.001
ILMN_1765499NTAN1N-terminal asparagine amidase5.710.057
ILMN_1652161PNKDParoxysmal nonkinesigenic dyskinesia2.750.028
ILMN_1767766PRDX2Peroxiredoxin 21.990.020
ILMN_1765518RNASEH2CRibonuclease H2, subunit C2.130.021
ILMN_1754303RPL30Ribosomal protein L301.710.017
ILMN_1746516RPS25Ribosomal protein S251.680.001
ILMN_1736135TYW1tRNA-yW synthesizing protein 1 homologue (S. cerevisiae)2.030.020
ILMN_2110281UFC1Ubiquitin-fold modifier conjugating enzyme 11.450.015
ILMN_2232936UQCRHUbiquinol-cytochrome c reductase hinge protein1.480.015
  • PBMC, peripheral blood mononuclear cells; SZ, schizophrenia, Con, control;

  • * p values determined using significance analysis of microarrays.

qPCR validation of differentially expressed genes

A search in the Genetic Association Database (geneticassociationdb.nih.gov) and SzGene (schizophrenia gene database; www.schizophreniaforum.org) identified several differentially expressed genes that have previously been associated with schizophrenia. Six of these genes, AKT1, DISC1, DGCR6, RXRA (retinoid X receptor, α), MMP9 (matrix metallopeptidase 9) and MAL (mal, T cell differentiation protein) were selected for qPCR validation. An additional gene RPS25 (ribosomal protein S25) was also selected on the basis of it being significantly down-regulated before treatment (1.95 fold down-regulation; p < 0.001) and up-regulated in response to treatment (1.68 fold up-regulation; p < 0.001), suggesting a medication effect. AKT1, RXRA and MMP9 were significantly up-regulated in schizophrenia patients and then returned to control levels after treatment on the array. qPCR confirmed that AKT1 (1.6 fold up; p = 0.028), RXRA (2.2 fold up; p = 0.002) and MMP9 (1.8 fold up; p = 0.008) were significantly up-regulated in schizophrenia patients before treatment (Fig. 1a). After treatment, these three genes showed a strong response to medication and returned to control levels (Fig. 1a). DISC1 was up-regulated in patients on the array before treatment and this was confirmed by qPCR (6 fold up; p = 0.047). Interestingly, the qPCR suggested that DISC1 remained up-regulated after the antipsychotic drug treatment in patients with schizophrenia (p = 0.022; Fig. 1a).

Fig. 1

Gene expression changes in schizophrenia patients before and after antipsychotic drug treatment by quantitate real-time reverse transcription PCR (qPCR). (a) AKT1, RXRA and MMP9 each displayed increased expression in schizophrenia patients, which returned to control levels upon drug treatment (p = 0.028, p = 0.002 and p = 0.008 respectively). By qPCR, DISC1 also remained significantly elevated after drug treatment compared to controls (before treatment p = 0.047, after p = 0.022). (b) RPS25 displayed decreased gene expression in patients before treatment (p = 0.013) then returned to control levels. DGCR6 was not significantly reduced before treatment (−1.31 fold down p = 0.171) but expression was significantly higher after drug treatment, similar to control levels (p = 0.049). (c) MAL displayed decreased expression in patients before treatment (p = 0.0002) and while remaining lower than control levels (−1.24 fold) this was not below the threshold for significance after treatment (p = 0.111). Bars are mean+s.e.m. Statistics are Mann–Whitney one-tailed tests. (d) Fold-changes obtained by microarray and qPCR are highly correlated (Pearson's r = 0.933, r2 = 0.8712, p = 0.0002).

RPS25 and DGCR6 were significantly down-regulated in patients and then returned to control levels after treatment on the array. qPCR confirmed that RPS25 was significantly down-regulated in patients before treatment (1.8 fold down, p = 0.013), which brought the expression back to control levels (Fig. 1b). DGCR6 showed a 1.3 fold down-regulation that was not significant (p = 0.171), followed by a significant (p = 0.049) increase in expression after treatment (Fig. 1b). MAL was significantly down-regulated in patients and did not respond to treatment on the array analysis. qPCR confirmed that MAL was down-regulated prior to treatment (1.8 fold down, p = 0.0002) and while remaining lower than control levels, the level of expression was not significantly different after treatment (−1.24 fold down; p = 0.111; Fig. 1c). Pearson's correlation analysis confirmed that the array fold changes were highly correlated to the qPCR fold changes (r = 0.93, R2 = 0.87, p = 0.0002; Fig. 1d).

IPA of differentially expressed genes

IPA of the differentially expressed genes in patients vs. controls prior to treatment identified a variety of functional changes associated with inflammation, immune cell trafficking, infection and respiratory disease, haematological system development and function, cell–cell signalling, gene expression, protein synthesis, cell development and movement (see Table 4 and Supplementary Table S5). Moreover, IPA also detected alterations to biological functions related to inflammation and immune function after treatment of patients when compared to controls (see Table 4 for a list of biological functions related to inflammation and immunity and Table 5 for a selection of genes that were altered prior to antipsychotic drug treatment, which map to these biological functions). In addition, Table 4 shows a comparison of these inflammatory and immune functions, suggesting that the treatment has stabilized gene expression and reduced the number of genes represented in each functional category. This accounts for a reduced range of p values from between 10−2 to 10−8 before treatment to 10−2 to 10−4 after treatment.

View this table:
Table 4

Top ranked biological functions over-represented by genes dysregulated in SZ before and after antipsychotic drug treatment

Functional category Diseases and disordersCon vs. SZ BTFunctional category Diseases and disordersCon vs. SZ AT
p valueMolecules (n)p valueMolecules (n)
Inflammatory response* 2.61E-11–1.99E-0292Inflammatory disease5.50E-04–3.98E-02 7
Infectious disease* 3.49E-11–9.54E-0398Inflammatory response* 5.50E-04– 4.97E-0213
Respiratory disease* 3.49E-11–1.84E-0258Skeletal and muscular disorders5.50E-04–2.67E-02 2
Cancer5.10E-05–2.01E-02156Infectious disease* 5.50E-04–2.67E-0214
Connective tissue disorder3.44E-04–4.00E-0382Respiratory disease* 5.50E-04–2.67E-02 5
Molecular and cellular functionsMolecular and cellular functions
Cell-to-cell signalling and interaction9.96E-08–2.13E-0272Carbohydrate metabolism5.50E-04–3.55E-02 4
Gene expression4.06E-07–2.20E-0320Cellular growth and proliferation5.50E-04–4.93E-02 7
Protein synthesis4.06E-07–3.53E-0354Nucleic acid metabolism5.50E-04–3.55E-02 5
Cellular development7.01E-07–2.01E-0280Small molecule biochemistry5.50E-04–4.41E-02 9
Cellular movement1.08E-06–2.10E-0282Antigen presentation1.76E-03–4.84E-02 5
Physiological system development and functionPhysiological system development and function
Haematological system development/function* 9.96E-08–2.13E-0294Haematological system development/function* 5.50E-04–4.93E-0216
Immune cell trafficking* 9.96E-08–1.99E-0273Immune cell trafficking* 1.76E-03– 4.84E-0210
Haematopoiesis* 5.44E-06–1.70E-0249Cell-mediated immune response2.92E-03–3.11E-02 5
Lymphoid tissue structure and development* 7.70E-06–1.70E-0235Haematopoiesis* 2.92E-03–4.84E-02 9
Tissue development5.50E-05–1.99E-0244Lymphoid tissue structure and development* 2.92E-03–3.55E-02 6
  • SZ, Schizophrenia; Con, control; BT, before treatment; AT, after treatment.

  • * Functional categories assigned by ingenuity pathways analysis that are in common in both conditions.

View this table:
Table 5

Selection of genes with altered expression prior to antipsychotic drug treatment that are over-represented in biological functions related to the immune and inflammatory pathways from the ingenuity pathways analysis in Table 4

Down-regulated genes
Illumina IDGene symbolEntrez gene nameFold change (SZ/Con)p value*
ILMN_1763875ABCF1ATP-binding cassette, sub-family F (GCN20), member 1−1.350.023
ILMN_2103841AIPAryl hydrocarbon receptor interacting protein−1.280.029
ILMN_1722491APRTAdenine phosphoribosyltransferase−1.30.012
ILMN_2179837BANF1Barrier to autointegration factor 1−1.40.023
ILMN_1665761BCL11BB-cell CLL/lymphoma 11B (zinc finger protein)−1.910.020
ILMN_1669663BCRBreakpoint cluster region−1.760.036
ILMN_1668996C1QBPComplement component 1, q subcomponent binding protein−1.570.023
ILMN_1715131CCR7Chemokine (C–C motif) receptor 7−1.530.048
ILMN_1695025CD2CD2 molecule−1.70.001
ILMN_2208903CD52CD52 molecule−2.420.012
ILMN_1746565CD6CD6 molecule−1.440.023
ILMN_1710017CD79BCD79b molecule, immunoglobulin-associated β−1.60.048
ILMN_2328666CD83CD83 molecule−2.630.025
ILMN_2354191CD8BCD8b molecule−2.320.009
ILMN_1711573CD96CD96 molecule−1.610.048
ILMN_1702105EFSEmbryonal Fyn-associated substrate−2.360.013
ILMN_2091412FLT3LGFms-related tyrosine kinase 3 ligand−1.750.001
ILMN_1769383GIMAP5GTPase, IMAP family member 5−1.450.001
ILMN_1779324GZMAGranzyme A (granzyme 1, cytotoxic T-lymphocyte-associated serine esterase 3)−1.60.041
ILMN_1689655HLA-DRAMajor histocompatibility complex, class II, DR α−1.270.041
ILMN_1766713HSPD1Heat shock 60 kDa protein 1 (chaperonin)−1.950.029
ILMN_1786823ICAM2Intercellular adhesion molecule 2−1.280.001
ILMN_1778010IL32Interleukin (IL) 32−1.440.025
ILMN_1699160ITKIL2-inducible T-cell kinase−1.750.029
ILMN_2404625LATLinker for activation of T cells−1.450.009
ILMN_1679185LEF1Lymphoid enhancer-binding factor 1−2.110.001
ILMN_1795464LTALymphotoxin α (TNF superfamily, member 1)−1.540.034
ILMN_2366391PRDX1peroxiredoxin 1−1.560.009
ILMN_1701613RARRES3Retinoic acid receptor responder (tazarotene induced) 3−1.630.001
ILMN_2322498RORARAR-related orphan receptor A−1.450.034
ILMN_1787949RPS15ARibosomal protein S15a−1.90.029
ILMN_1784717RPS19Ribosomal protein S19−1.570.013
ILMN_1771801SIRPGSignal-regulatory protein γ−1.680.020
ILMN_1760109SLAMF6SLAM family member 6−1.570.029
ILMN_1662438SOD1Superoxide dismutase 1, soluble−2.170.023
ILMN_2414325TMSB10/TMSB4XThymosin β 4, X-linked−1.650.009
ILMN_1764851TP53RKTP53 regulating kinase−1.640.020
Up-regulated genes
ILMN_1766054ABCA1ATP-binding cassette, sub-family A (ABC1), member 12.070.048
ILMN_1708348ADAM8ADAM metallopeptidase domain 81.520.025
ILMN_2388507AKT1v-akt murine thymoma viral oncogene homologue 11.380.010
ILMN_1680996ALOX5Arachidonate 5-lipoxygenase1.680.023
ILMN_1778723AMICA1Adhesion molecule, interacts with CXADR antigen 11.470.006
ILMN_1763837ANPEPAlanyl (membrane) aminopeptidase1.950.041
ILMN_1695157CA4Carbonic anhydrase IV1.990.029
ILMN_1722622CD163CD163 molecule1.840.041
ILMN_1673363CD97CD97 molecule1.530.034
ILMN_1743570CEACAM3Carcinoembryonic antigen-related cell adhesion molecule 31.710.048
ILMN_2376455CSF2RAColony stimulating factor 2 receptor, α, low-affinity (granulocyte-macrophage)1.860.006
ILMN_2371280CSF3RColony stimulating factor 3 receptor (granulocyte)1.580.006
ILMN_1743032CTSScathepsin S1.940.048
ILMN_1728478CXCL16Chemokine (C-X-C motif) ligand 161.610.036
ILMN_1781285DUSP1Dual specificity phosphatase 11.440.048
ILMN_1720158ETS2v-ets erythroblastosis virus E26 oncogene homologue 2 (avian)1.470.041
ILMN_1709233F5Coagulation factor V (proaccelerin, labile factor)2.810.010
ILMN_2365091FCARFc fragment of IgA, receptor for2.350.036
ILMN_1705302FCGRTFc fragment of IgG, receptor, transporter, α1.630.041
ILMN_2368318FGRGardner-Rasheed feline sarcoma viral (v-fgr) oncogene homologue1.340.048
ILMN_1844692FOXO3Forkhead box O31.440.029
ILMN_2092118FPR1Formyl peptide receptor 11.390.034
ILMN_2364110GBAGlucosidase, β, acid1.470.029
ILMN_1735155GLB1Galactosidase, β 11.450.048
ILMN_1791771HCKHaemopoietic cell kinase1.550.014
ILMN_2212763ICAM3Intercellular adhesion molecule 31.490.006
ILMN_1676515IMPDH1IMP (inosine 5′-monophosphate) dehydrogenase 11.530.001
ILMN_1808299IQSEC1IQ motif and Sec7 domain 11.610.034
ILMN_1792679ITGA5Integrin, α 5 (fibronectin receptor, α polypeptide)1.410.048
ILMN_2254635ITGAXIntegrin, α X (complement component 3 receptor 4 subunit)1.950.001
ILMN_2175912ITGB2Integrin, β 2 (complement component 3 receptor 3 and 4 subunit)1.270.048
ILMN_1673282LAMP2Lysosomal-associated membrane protein 21.370.048
ILMN_1716983LILRA2Leucocyte immunoglobulin-like receptor, subfamily A (with TM domain), member 21.870.048
ILMN_1667476LTBRLymphotoxin β receptor (TNFR superfamily, member 3)1.610.041
ILMN_1781155LYNv-yes-1 Yamaguchi sarcoma viral related oncogene homologue1.410.041
ILMN_1779010MAP3K3Mitogen-activated protein kinase kinase kinase 31.90.006
ILMN_2331544MBPMyelin basic protein1.440.023
ILMN_1678170MMEMembrane metallo-endopeptidase1.840.034
ILMN_1796316MMP9Matrix metallopeptidase 93.110.006
ILMN_1792910MNTMAX binding protein1.520.025
ILMN_1738523MYD88Myeloid differentiation primary response gene (88)1.570.048
ILMN_1681239MYO1FMyosin IF2.140.001
ILMN_1758735NLRP12NLR family, pyrin domain containing 122.050.023
ILMN_1801105PRKCDProtein kinase C, δ1.360.034
ILMN_1687315RXRARetinoid X receptor, α1.530.029
ILMN_2183409SCARB1Scavenger receptor class B, member 11.80.041
ILMN_1702787SEMA4ASemaphorin 4A1.870.023
ILMN_1741165SLC11A1Solute carrier family 11, member 120.023
ILMN_2410986STAT3Signal transducer and activator of transcription 31.60.048
ILMN_1684034STAT5BSignal transducer and activator of transcription 5B1.40.048
ILMN_1763198STAT6Signal transducer and activator of transcription 6, IL-4 induced1.280.041
ILMN_1711994TCIRG1T cell, immune regulator 1, ATPase, H+ Transporting, lysosomal V0 subunit A31.830.014
ILMN_1749078TIMP2TIMP metallopeptidase inhibitor 22.330.029
ILMN_1722981TLR5Toll-like receptor 51.950.048
  • SZ, Schizophrenia; Con, control.

  • * p values determined using significance analysis of microarrays.

IPA canonical pathways analysis showed a significant representation of differentially expressed genes in the eukaryotic initiation factor-2 (EIF2) signalling, regulation of eIF4 and p70S6K, oxidative phosphorylation, mTOR signalling and mitochondrial dysfunction pathways in patients prior to treatment (Table 6). These pathways also showed a significant representation of differentially expressed genes after treatment, but once again there were fewer genes represented and the p values were lower (Table 6). This is illustrated in Fig. 2, which shows the EIF2 signalling pathway that is involved in the initiation of protein synthesis. In Fig. 2a, 16 genes including a number of EIF genes in this pathway were down-regulated (green) and AKT1 was up-regulated (red) in patients, whereas after treatment only four genes remained down-regulated and AKT1 had returned to control levels, as confirmed above in the qPCR analysis (Fig. 2b).

Fig 2

Changes to the EIF2 signalling pathway in peripheral blood mononuclear cells (PMBCs) from patients with schizophrenia ingenuity pathway analysis of differentially expressed genes in PMBCs from schizophrenia patients before (a) and after (b) treatment with antipsychotic drugs, compared to controls identified the EIF2 signalling pathway as the top canonical pathway containing the most significant representation of differentially expressed genes. Before treatment, the schizophrenia-associated gene AKT1 was markedly up-regulated (red) while many of the EIF genes were down-regulated (green) suggesting disruption to the control of protein synthesis. The antipsychotic drug treatment reduced this widespread dysregulation in the pathway in the patients and returned the expression of a number of the genes including AKT1 back to control levels.

View this table:
Table 6

Ingenuity canonical pathway analysis

Ingenuity canonical pathwaysCon vs. SZ BTCon vs. SZ AT
p valueRatiop valueRatio
EIF2 signalling9.26 × 10 − 2640/199 (0.201)5.27 × 10 − 1011/199 (0.055)
Regulation of eIF4 and p70S6K signalling3.81 × 10 − 1425/170 (0.147)4.65 × 10 − 045/170 (0.029)
Oxidative phosphorylation2.90 × 10 − 1223/160 (0.144)1.38 × 10 − 045/160 (0.031)
mTOR signalling2.17 × 10 − 0922/201 (0.109)1.38 × 10 − 035/201 (0.025)
  • Con, Control; SZ, schizophrenia; BT, before treatment; AT, after treatment.

Interestingly, AKT1 is also a member of the mTOR signalling pathway and the eIF4/p70S6K pathways, which show the same pattern of less dysregulation of gene expression and no change in AKT1 expression after treatment (data not shown). These three pathways that contain AKT1 were also highlighted in the IPA of differentially expressed genes from the direct comparison of patient samples before and after treatment. But once again there were fewer genes represented and the p values were lower (data not shown).

The IPA was repeated by restricting the gene lists to genes that were changed in the same direction both before and after treatment (i.e. 64 down-regulated and four up-regulated). This restricted list still detected over-representation of genes with roles in biological functions involved in inflammation, immune cell trafficking, infectious and respiratory disease, haematological system development and function, cell–cell signalling, gene expression, protein synthesis and cell development and movement (Supplementary Table 5). It also predicted dysfunction of the EIF2 signalling, regulation of eIF4 and p70S6K, oxidative phosphorylation and mTOR signalling canonical pathways (not shown).

Gene set enrichment analysis (GSEA)

GSEA was conducted on the total gene expression data set using the BioCarta-derived gene sets. An analysis of the gene expression in the patients prior to treatment vs. controls identified 20 gene sets enriched in the patients before treatment, with one gene set (CSK_Pathway) enriched in the controls (Table 7). One of the most significant findings was the enrichment of genes involved in the mitogen-activated protein kinase (MAPK) signalling pathway in the patients prior to treatment (Table 7). In agreement with the IPA, a number of the enriched gene sets prior to treatment belong to pathways with roles in regulating the immune system and inflammation (Table 7). In addition, three gene sets (GSK3_Pathway, IL2RB_Pathway and PTDINS_Pathway) have AKT1 as a member of their pathway. Furthermore, GSEA of the gene expression data after treatment (i.e. patients after treatment vs. controls) with antipsychotic medication did not identify any significantly enriched gene sets, providing further evidence that the treatment has stabilized gene expression.

View this table:
Table 7

Gene Set Enrichment Analysis in schizophrenia patients before treatment with antipsychotic medication

BioCarta pathway nameDescriptionSize (genes)NESN p valueFDR q-value
MAPK_PATHWAY* MAPK signalling601.84100.048
SPPA_PATHWAYSignalling pathway in platelet activation151.75300.037
TOLL_PATHWAY* Toll-like receptor signalling251.8870.0040.049
IL10_PATHWAY* IL-10 anti-inflammatory signalling151.7680.0060.047
KERATINOCYTE_PATHWAYKeratinocyte differentiation351.7630.0060.040
EPO_PATHWAYErythropoietin signalling151.8100.0080.041
TPO_PATHWAYThrombopoietin signalling181.6760.0140.066
MET_PATHWAYSignalling of hepatocyte growth factor receptor221.6560.0160.066
IL2RB_PATHWAY* IL-2 receptor β chain in T cell activation241.6050.0160.070
GSK3_PATHWAY* Inactivation of Gsk3 by AKT201.6760.0170.059
INTEGRIN_PATHWAYIntegrin signalling261.6360.0210.063
PTDINS_PATHWAYPhosphoinositides and their downstream targets161.6460.0220.063
IL2_PATHWAY* IL 2 signalling171.5980.0230.062
IL6_PATHWAY* IL 6 signalling171.6130.0230.070
EGF_PATHWAYEpidermal growth factor signalling231.6020.0260.068
PDGF_PATHWAYPlatelet derived growth factor signalling231.6020.0260.063
IL1R_PATHWAY* Signal transduction through IL1R201.6930.0290.064
BIOPEPTIDES_PATHWAYBioactive peptide induced signalling331.5470.0290.087
CSK_PATHWAY* Activation of Csk by cAMP-dependent protein kinase18−1.4930.0460.464
  • NES, normalized enrichment score; N p value, normalized p value; FDR q-value, false discovery rate q value; MAPK, mitogen-activated protein kinase; IL, interleukin.

  • * Denotes pathways involved in immune function and inflammation.

  • A positive NES represents gene sets enriched in patients before treatment, a negative NES represents gene sets enriched in controls.


This study investigated gene expression changes in PBMCs in schizophrenia patients before and then 6–8 wk into antipsychotic pharmacotherapy of 200 mg/d chlorpromazine equivalents. In the treatment-naive state, patients exhibited greater dysregulation of gene expression (⩾600 genes) than after treatment.

The RXRA, MAL, DISC1 and RPS25 genes were significantly altered in PBMCs in patients prior to treatment. RPS25 a gene coding for a ribosomal protein has not been reported to be associated with schizophrenia before. In contrast, RXRA has been implicated in sib pairs with schizophrenia where a copy number variation disrupted the RXRA gene (Lee et al., 2010). MAL, a gene coding for a myelin and lymphocyte protein was down-regulated in the patients with schizophrenia, which is similar to that observed in the prefrontal cortex (BA46) in a post mortem study of schizophrenia (Hakak et al., 2001). MAL has also been reported to be dysregulated in major depressive disorder (Aston et al., 2005) suggestive of a broader phenotype. Detection of myelin-associated gene changes in PBMCs in schizophrenia was also reported in a study comparing schizophrenia patients and their unaffected and unmedicated siblings to controls (Glatt et al., 2011), where myelin basic protein (MBP) was shown to be up-regulated in schizophrenia compared to controls. In this study, MBP was also up-regulated on the array prior to treatment of the patients but returned to control levels.

DISC1 (disrupted in schizophrenia 1), a known genetic candidate for schizophrenia has been implicated in neurodevelopment, cognition and other psychiatric disorders (reviewed in Brandon and Sawa, 2011). Kamiya et al. (2005) emphasized the role of DISC1 in cerebral cortical development and others have described this gene's functions in synaptogenesis and sensory perception (Hennah and Porteous, 2009). Interestingly, a post mortem study reported increased expression of specific isoforms of DISC1 in the hippocampus in schizophrenia (Nakata et al., 2009), which is consistent with the increased expression of DISC1 prior to and after treatment, in this study. This provides additional evidence to our previous study showing that brain-related gene changes can be detected in PBMCs in schizophrenia (Bowden et al., 2006).

In a comparison of all genes dysregulated in this study by the microarray analysis, there was no significant difference in the expression of 14 genes identified by Takahashi et al. (2010) in treatment-naive patients nor did we detect changes to the DRD2 and Kir3.2 genes highlighted in the study of treatment-naive patients by Zvara et al. (2005). However, three genes down-regulated prior to treatment in this study, ABCF1 (ATP-binding cassette sub-family F member 1), BTBD11 (BTB/POZ domain-containing protein 11) and BCL11B (B-cell CLL/lymphoma 11B zinc finger protein), were also down-regulated in PBMCs from schizophrenia patients in two other studies, with one additional report also showing down-regulation of ABCF1. The first study showed down-regulation of ABCF1 in recent onset schizophrenia patients that were medicated compared to healthy controls (van Beveren et al., 2012). The second study of 92 medicated schizophrenia patients and 111 healthy controls reported down-regulation of ABCF1, BTBD11 and BCL11B, with all three genes confirmed to be down-regulated in a second cohort consisting of 29 treatment-naive schizophrenia patients (de Jong et al., 2012). Finally, we recently completed a PBMC study from 114 medicated patients with schizophrenia/schizoaffective disorder and 80 non-psychiatric controls from the Australian Schizophrenia Research Bank (Loughland et al., 2010) and showed that ABCF1, BTBD11 and BCL11B were down-regulated by microarray analysis (Gardiner et al., 2012). Whilst the SAM analysis of the gene expression data from the current study suggested that these three genes returned to control levels after treatment, individual students t tests conducted on the microarray expression data confirmed a significant reduction for ABCF1 and a trend towards down-regulation for BCL11B after treatment, suggesting these two genes are down-regulated in PBMCs in schizophrenia prior to treatment and remain that way post treatment.

The BTBD11 gene, which appears to respond to antipsychotic drug treatment, is part of the BTB/POZ domain-containing protein family thought to function as transcription factors in different signalling pathways (Liu et al., 2011). BCL11B, a transcription factor with roles in driving commitment to the T cell lineage during haematopoiesis (Rothenberg, 2012), is known to be involved in the development of layer 5 cortical neurons (Chen et al., 2008) and was recently linked to brain-derived neurotrophic factor (BDNF) signalling (Tang et al., 2011). Interestingly, ABCF1 is located in the 6p21.32-p22.1 major histocompatibility complex locus reported to be associated with schizophrenia (Purcell et al., 2009; Shi et al., 2009; Stefansson et al., 2009; Ripke et al., 2011). ABCF1 was discovered in experiments that identified it as a gene that responded to tumour necrosis factor treatment of synoviocytes isolated from healthy controls and patients with rheumatoid arthritis and expressed in the brain (Richard et al., 1998). It was also reported to be associated with autoimmune pancreatitis (Ota et al., 2007) and thus may play a role in inflammation, a process gaining more attention in schizophrenia research.

The IPA and GSEA conducted in this study identified enrichment of genes with altered expression in schizophrenia that contribute to biological functions related to immune function, inflammation and infectious diseases. These findings are in accordance with the study by de Jong et al. (2012) mentioned above, which also identified enrichment of genes with altered expression in schizophrenia to pathways such as cell-mediated immune response, antigen presentation, haematological system development and function, inflammatory response, infectious disease and immune cell trafficking (de Jong et al., 2012). What is pertinent to this current study is that these pathways were also identified as dysregulated in the follow-up cohort of antipsychotic-free schizophrenia patients (de Jong et al., 2012).

Epidemiological studies have analysed data from collections dating back many decades and provided evidence for increased risk of developing schizophrenia in people exposed to infectious agents, such as rubella and influenza, either in utero or during early life (reviewed in Brown 2011). The 6p21.32-p22.1 major histocompatibility complex locus (Ripke et al., 2011) and the interleukin (IL)-1 gene complex (Xu and He, 2010) are associated with schizophrenia. Furthermore, studies have shown increased levels of soluble IL-1β, IL-2R, IL-6, IL-8 and IL-18 protein in serum (Zhang et al., 2004; Schmitt et al., 2005; Potvin et al., 2008; Bresee and Rapaport, 2009; Garcia-Miss Mdel et al., 2010; Kunz et al., 2011; Garcia-Rizo et al., 2012; Palladino et al., 2012; Xiu et al., 2012) and IL-1β, IL-6 and IL-8 mRNA in the dorsolateral prefrontal cortex (Fillman et al., 2012) in people with schizophrenia. The GSEA reported here shows enrichment for genes in IL pathways, IL-1R, IL-2, IL-2Rβ, IL-6 and IL-10, in schizophrenia patients prior to treatment. Furthermore, evidence suggests IL-6 may have a neuroprotective effect by activating the MAPK pathway (Wang et al., 2009), which was one of the top ranked gene sets enriched in the patients prior to treatment identified by GSEA. The MAPK pathway is linked to the activity of the glutamate NMDA receptor, which is implicated in schizophrenia and both have roles in the brain in long-term potentiation, learning and memory, neurotoxicity and oxidative stress (for review, see Haddad, 2005). In addition, antipsychotic drug treatment is known to affect the MAPK signalling pathway (for review, see Molteni et al., 2009). However, the exact role of the MAPK signalling pathway in the development of schizophrenia remains to be determined.

The data presented herein suggest that the biological functions related to immunity, inflammation and infectious diseases are present in treatment-naive patients with schizophrenia and that antipsychotic pharmacotherapy can partially compensate for these changes in these pathways by ‘normalizing’ gene expression. This would suggest that the signatures of immune dysfunction generated in PBMCs and brain tissue are likely to be reflective of the actual pre-treatment states, providing evidence for the infection/immune hypothesis of schizophrenia. To this end, IL32 was down-regulated in this study of PBMCs both prior to and after treatment and, interestingly, IL32 was also shown to be down-regulated in the left superior temporal cortex in post mortem brains from schizophrenia subjects who received treatment (Schmitt et al., 2011). Whether this pro-inflammatory cytokine has a role in the development of schizophrenia remains to be determined.

Further evidence for immune/infection hypothesis for schizophrenia was recently provided by the large study of PBMCs isolated from 112 patients with schizophrenia and 76 non-psychiatric controls from the Australian Schizophrenia Research Bank (Loughland et al., 2010), which was conducted by this laboratory (Gardiner et al., 2011). This study showed that 33 miRNA were down-regulated in schizophrenia patients and that some of the functions of the genes targeted by these miRNAs are related to the immune system and inflammation (Gardiner et al., 2011). Furthermore, 17 of the down-regulated miRNAs are transcribed from a single imprinted locus at the maternally expressed DLK1-DIO3 region on chromosome 14q32. Interestingly, BCL11B mentioned above and AKT1 (up-regulated in this current study then normalizes with treatment) are both located in the 14q32 chromosomal region. Indeed, AKT1 is predicted to be targeted according to miRGen (http://www.diana.pcbi.upenn.edu/cgi-bin/miRGen/v3/Targets.cgi) by four miRNAs that were down-regulated in PMBCs in schizophrenia, including one located in the imprinted region. Whether these miRNA are down-regulated prior to treatment is yet to be determined, but one could speculate that this might contribute to the up-regulation of AKT1 observed in schizophrenia patients before treatment.

The evidence for a role of AKT1 in schizophrenia has been increasing, but there are also several reports of non-association (for review, see Balu and Coyle 2011). Post mortem studies suggest a reduction in the levels of AKT1 protein in the prefrontal cortex, hippocampus and lymphoblast cell lines in schizophrenia (Emamian et al., 2004), whereas reductions of AKT1 mRNA levels in the prefrontal cortex have also been reported (Thiselton et al., 2008). AKT1 deficiency in schizophrenia has also been associated with impairment of hippocampal plasticity and function (Balu and Coyle, 2011). Interestingly, in models of maternal infection, mice displayed reductions in AKT1 positive cells in the prefrontal cortex and altered cognitive behaviours (Bitanihirwe et al., 2010). Beaulieu et al. (2004) reported, a link between the Akt/GSK3 signalling pathway, a pathway identified by GSEA in this study as enriched in patients prior to treatment, and dopamine neurotransmission in mice (Beaulieu et al., 2004). Another study of AKT1 deficient mice suggested that AKT1 function is critical for dopaminergic neurotransmission and dopamine-dependent behaviours (Emamian et al., 2004).

While the literature suggests there is AKT1-deficiency in the brain and lymphoblasts in schizophrenia, our data suggest that in PBMCs that are not EBV transformed and were isolated prior to treatment with antipsychotic medication have an increase in AKT1 expression. In addition, the correction of up-regulated AKT1 expression provides additional evidence for a link between dopamine D2 receptors and AKT1. What is striking about the data presented here is that three of the top five ranked canonical pathways identified by IPA all contain AKT1 and remained significantly over-represented after treatment. Each of these pathways, EIF2 signalling, regulation of eIF4 and p70S6K and mTOR signalling, have roles in regulating protein synthesis required for cell growth, cell survival and development (Carter, 2007). In the brain, these signalling cascades respond to neuregulin and growth factors such as BDNF (Carter, 2007) and respond to glutamate via NMDA receptors, all of which have been implicated in schizophrenia. They also respond to hormones, cytokines and stressors such as viral infections (Carter, 2007). There is also evidence that the AKT1/mTOR signalling pathway is a target of DISC1 (Kim et al., 2009), which was also shown to be up-regulated here.

Are these pathways altered as a ‘state’ effect rather than a ‘trait’ effect? Mental stress associated with psychosis may also have induced widespread shutdown of the EIF2 pathway and the increase of AKT1 expression may represent an attempt to correct this situation. As each of the patients did show improvement with pharmacotherapy, lower stress levels may have led to a return to control levels for many of the EIF genes dysregulated in the EIF2 pathway (Fig. 2), thus resulting in a reduction of AKT1 activation. In addition, although this study was conducted on a small sample it was pair-wise matched for age, but not gender. There were more females in the control group and it is difficult to determine if this gender difference may have influenced the gene expression group differences reported in this study. Furthermore, three patients also received haloperidol treatment. Since the typical and atypical antipsychotics have differences in their pharmacological effects, whether this impacted on the gene expression changes in the patients observed in this study requires further investigation.

Overall, there were 67 genes remaining dysregulated in the same direction after treatment with medication and these genes were still significantly over-represented in the EIF2, the regulation of eIF4 and p70S6K and the mTOR signalling pathways. While preliminary due to a relatively small sample size, our data suggest changes to gene expression that highlight particular biological pathways relevant to schizophrenia pathology, including those related to inflammation and immune function, can be detected in PBMCs and that these pathways are affected by antipsychotic medication, thus potentially serving as biomarkers for the disorder and its treatment.

Supplementary material

Supplementary material accompanies this paper on the Journal's website.

Supplementary information supplied by authors.

Supplementary information supplied by authors.

Supplementary information supplied by authors.

Supplementary information supplied by authors.

Supplementary information supplied by authors.

Statement of Interest



Nishantha Kumarasinghe was supported by a doctoral scholarship from the World Health Organization. The project received financial and in-kind support from the Schizophrenia Research Institute utilizing funding from the NSW Ministry of Health and the University of Newcastle, Australia and the University of Sri Jayewardenepura, Sri Lanka.


View Abstract