Introduction

Schizophrenia is a complex mental disorder, one of the leading causes of disease burden in the world, notably in young adults [1]. Its phenotype is highly heterogeneous, possibly due to different clinical subgroups [2] and staging characteristics [3]. It has an important genetic component, with a high heritability, estimated at up to 80 % [4], in which common genetic variants with a small effect and/or rare genetic variants with a large effect interact with environmental factors [5]. Immune-inflammatory processes are involved in disorder onset and progression of schizophrenia and are associated with both genetic and environmental risk factors [6, 7].

Different pathological processes triggered by multiple hits, such as infections or psychosocial trauma, play a role in the pathophysiology of schizophrenia. These pathways include the dopaminergic system, immune-inflammatory and neuroprogressive pathways, neurodevelopmental processes, messenger RNA (mRNA) degradation or translational inhibition as well as epigenetic factors [3, 7]. The largest genome-wide association studies (GWASs) have consistently provided support for a link between the immune system, neuroprogression, and schizophrenia [6, 8]. These pathways contribute to the pathophysiology of schizophrenia, and their interaction may explain the multiple triggers that induce the disorder [7, 9].

Many different immune abnormalities are found in schizophrenia, including increased levels of pro-inflammatory cytokines, such as interleukin (IL)-6 and tumor necrosis factor (TNF)-α, and the negative immune regulatory cytokine IL-10 [10]. IL-6 and TNF-α are involved in both neurodevelopmental and neuroprogressive processes and may cause dysfunctions in neuroplasticity and neurogenesis, including growth, differentiation, myelination, apoptosis, synaptic branching, and neurotrophin regulation [11, 12]. They also activate neurotransmitter signaling, e.g., catecholamines [13]. IL-10, on the other hand, has anti-inflammatory properties, while negatively modulating the immune response and having neuroprotective effects [14].

Increased levels of pro-inflammatory cytokines regulate gene expression and protein synthesis. For example, the administration of pro-inflammatory cytokines or lipopolysaccharide (LPS) provokes a significant reduction of BDNF expression [15]. The activities of pro-inflammatory cytokines also activate nuclear factor-κB, a transcription factor, responsible for the regulation of several genes, including cytokines [16, 17]. Moreover, IL-6 may mediate epigenetic alterations in cells via the regulation of the DNA methyltransferase enzyme gene (DNMT-1) [18], which has also been implicated in schizophrenia [19, 20].

Complex effects of antipsychotic drugs on the immuno-inflammatory, oxidative stress, and neuroprogressive pathways have been described showing that these drugs may modulate immune and oxidative pathways, synaptic plasticity, and myelin generation [7, 21, 22]. Risperidone treatment has immunoregulatory effects on drug naïve first episode psychosis (FEP) patients [21] and may regulate mRNA expression of GABRR2 and GCH1 [23–25].

Clinical variables may contribute to the regulation of genetic and immune pathways. Depression is a prevalent comorbidity in schizophrenia [26], reaching up to 80 % in FEP patients [27]. Patients with concomitant depressive symptoms have more dysfunctions in immune-inflammatory pathways [28]. In this study, we selected 11 targeted genes in schizophrenia (AKT1, DICER1, NDEL1, TNF, UFD1L, DROSHA, COMT, DISC1, MBP, IL2, and IL6) and analyzed their expression in FEP patients in relation to depression and circulating levels of cytokines both before and after treatment with risperidone. Genes selection was based on their function as follows: dopamine neurotransmission (COMT); inflammation and immune system (TNF, IL2, and IL6); neurodevelopment (DISC1 and NDEL1); myelination (MBP); cell signaling (AKT1); microRNA machinery (DICER1 and DROSHA); and degradation of proteins (UFD1L); and based on their mRNA expression previously described in whole blood (according to information available in http://www.genecards.org/).

Subjects and Methods

The present study is part of a prospective cohort study performed in Sao Paulo, Brazil, which examines different characteristics of antipsychotic naïve FEP patients, including gene expression and their immune-inflammatory profile. Patients were followed during the treatment with risperidone for delineating the effects of this antipsychotic. All participants provided written informed consent prior to enrollment in this study. The study was approved by the Research Ethics Committee of UNIFESP (Sao Paulo, Brazil) and carried out in accordance with the Declaration of Helsinki.

Study Population

Antipsychotic-naïve FEP patients (N = 174) were recruited from a psychiatric emergency unit in São Paulo, Brazil. The diagnosis of a psychotic disorder was established by trained psychiatrists, according to the criteria of the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM-IV), using the Structured Clinical Interview of the DSM-IV (SCID-I). Only individuals between 16 and 40 years of age with no prior history of antipsychotic medication exposure were selected. Individuals with psychotic episodes due to a general medical condition, substance-induced psychotic disorder, or intellectual disability were excluded. All patients fulfilled the criteria for one of the following diagnoses: schizophrenia (53.2 %), schizophreniform disorders (17.4 %), brief psychotic disorders (12.0 %), psychotic disorders not otherwise specified (10.9 %), and manic episode with psychotic symptoms (6.5 %). None of the patients had previously received any psychiatric diagnosis. The comparison group consisted of 77 healthy volunteers. Controls and their first-degree family members had a negative lifetime history of major psychiatric disorder, according to the criteria of the DSM-IV. We excluded patients and controls with acute and chronic medical conditions associated with an imbalance in the immune system, including infections (e.g., HIV), allergic reactions, pregnancy, the postpartum period, rheumatic disorders, and using medications with immunomodulatory effects such as nonsteroidal anti-inflammatory drugs, corticosteroids, and immunosuppressants.

All patients received risperidone as standard treatment, with mean dose of 4.0 mg/day (SD = 2.1). A subgroup of 72 patients treated with risperidone was followed for 64 days (q25 = 56, q75 = 78) and reassessed after this period. All patients were evaluated using the (a) PANSS (Positive and Negative Syndrome Scale) [29] and (b) CDSS (Calgary Depression Scale for Schizophrenia) [30]. In the subgroup that had a follow-up after treatment, the patients underwent interviews by the same experienced psychiatrist.

Analyses

A total of 15 mL of whole blood was collected in PAXgene® RNA (PreAnalytix, Hombrechtikon, Switzerland) and SST tubes (Becton Dickinson, Plymouth, UK), and from all patients at admission, before the first dose of risperidone, and after treatment, and from 77 healthy controls after the SCID-I interview. Blood was centrifuged and RNA was isolated using a PAXgene® Blood RNA kit (Qiagen, Germantown, MD) according to the manufacturer’s instructions. The serum was stored at −80 °C until thawed for assays of cytokines.

mRNA Expression

The RNA integrity was determined through electrophoresis on a 1.0 % agarose gel, and the quality and quantity of the RNA samples were determined using a NanoDrop® ND-1000 spectrophotometer (Nanodrop, Wilmington, DE). Approximately 400 ng of each RNA sample was reverse-transcribed using the High-Capacity cDNA Reverse Transcription Kit (Life Technologies, Foster City, CA), and 20–100 ng were diluted in H2O, mixed with TaqMan® Universal PCR Master Mix (Life Technologies) and loaded on the Taqman Low-Density Array (TLDA) microfluidic cards (Life Technologies). Probes and primers of 11 target genes and two housekeeping genes (HKGs: ACTB and GAPDH) were preloaded in the 384 wells of each TLDA card (Life Technologies), and the experiments were performed in accordance with the manufacturer’s instructions using the ViiA™ 7 Real-Time PCR System (Life Technologies). Assays and exons and transcripts that they recognize are described in Supplementary Table 1. Gene expression was quantified using the relative threshold method (Crt) with the geometric mean (GM) between ACTB and GAPDH as the endogenous control. ΔCrt values (ΔCrt = Crt target gene − CrtGM) were calculated for each sample and included in the SPSS (version 19.0) dataset. There is an inverse correlation between ΔCrt and gene expression. Gene expression data were measured in 72 patients and 73 controls. Power analysis showed that for the primary outcome data (mRNA gene expression), the sample size should be around 140 when considering 2 groups, an effect size of 0.24, α = 0.05, and power = 0.8.

Cytokines

Serum IL-6, IL-10, and TNF-α were measured by flow cytometry using the Cytometric Bead Array Human Th1/Th2/Th17 Kit (BD Biosciences). Acquisition was performed with a FACS Canto II flow cytometer (BD Biosciences). The instrument was checked for sensitivity and overall performance with Cytometer Setup and Tracking beads (BD Biosciences) prior to data acquisition. Quantitative results were generated using FCAP Array v1.0.1 software (Soft Flow Inc.). The cytokine data were entered as continuous variables (in Ln transformation) and as groups that is lower versus higher concentration levels, i.e., using cutoff values of IL-6 2.2 pg/mL; TNF-α 1.5 pg/mL and IL-10 1.3 pg/mL. Cytokines were measured in 157 patients and in 58 controls.

Statistical Analyses

We used analysis of contingency tables (χ 2 test) or simple analysis of variance to assess the associations between diagnosis and clinical and sociodemographic data. Multiple post-hoc differences among treatment means were examined using protected Bonferroni-corrected comparisons. Automatic binary logistic regression analysis was employed to delineate the significant gene expression predictors of the FEP group. Multivariate and univariate general linear model (GLM) analyses were used to assess the associations between the gene expression data (dependent variables) and explanatory variables, including diagnosis, age, gender, and sociodemographic data. Regression analysis was used to examine the relationships between gene expression data (dependent variable) and explanatory variables, including IL-6 levels. Repeated measurements (RM) design GLM analyses were used to assess within-subject effects, i.e., time (treatment with risperidone) and the interaction time × group (e.g., diagnosis or IL-6 groups). The Levene test was used to test for homogeneity of variance. All tests were two-tailed and a p value of 0.05 was used for statistical significance.

Results

Table 1 shows the demographic and biomarker data of the subjects in this study. There were no significant differences in age or gender ratio between cases and controls. FEP subjects had a significantly lower personal and familial income than the controls. There were significantly more smokers in FEP patients than in controls. Therefore, we adjusted our results for these differences in these sociodemographic data.

Table 1 Sociodemographic, clinical characteristics, and biomarker measures of (FEP) patients and healthy controls (HC)
Full size table

The mRNA expression data for IL-2 and IL-6 were not detectable in many subjects, and therefore, these two cytokine expression data could not be included in the computations. Among the 72 FEP and 73 controls, DICER1 mRNA was measurable in all subjects, and mRNA expression of AKT1 (n = 1), COMT (n = 3), DROSHA (n = 9), NDEL1 (n = 1), TNF (n = 12), and UFD1L (n = 1) was not measurable in few subjects. There were no significant differences in the number of undetectable mRNA expression data between controls and patients for any of these genes.

The multivariate GLM analysis showed a significant effect of diagnosis (FEP patients versus controls) on the ΔCrt values of the nine genes (F = 2.68, df = 9/51, p = 0.012). Forced entry of age, gender, and smoking status showed that these variables were not associated with the gene expression data and that the effects of diagnosis remained significant. Consequently, we have performed between-subject and univariate GLM analyses on each of the genes separately because the number of subjects with detectable mRNA levels differs between the nine genes included in this study (thus lowering the degrees of freedom due to an increased number of missing values when the genes are combined in multivariate analyses). There were significant differences in NDEL1, DROSHA, COMT, DISC1, and MBP ΔCrt values between FEP patients and controls. There were also significant differences in serum levels of IL-10, IL-6, and TNF-α between both study groups.

Table 2 shows the results of two binary logistic regression analyses with the FEP diagnosis as dependent variable (healthy controls as reference group) and the significant variables (see Table 1), i.e., NDEL1, DROSHA, COMT, DISC1, and MBP with or without cytokines (IL-6, IL-10, and TNF-α), as explanatory variables. We found that the NDEL1 (negatively) and DISC1 (positively) ΔCrt values were significantly associated with the diagnosis of FEP (χ 2 = 12.36, df = 2, p = 0.002). Of all patients, 69.0 % were correctly classified with a sensitivity of 53.3 % and specificity of 81.8 % (Nagelkerke = 0.156). We also found that the DROSHA ΔCrt values, IL-6, and TNF-α groups were significantly positively associated with the diagnosis of FEP (χ 2 = 22.24, df = 3, p < 0.001). 73.5 % of all patients were correctly classified with a sensitivity of 63.3 % and specificity of 83.7 % (Nagelkerke = 0.271).

Table 2 Results of two binary logistic regression analyses with the ΔCrt values of NDEL1, DROSHA, COMT, DISC1, and MBP with or without cytokines, i.e., interleukin (IL)-6, IL-10, and tumor necrosis factor (TNF)-α as explanatory variables and the diagnosis of first episode psychosis as dependent variable (the controls as reference group)
Full size table

Table 3 shows the results of automatic univariate GLM analyses with the gene expression data as dependent variables and diagnosis (FEP patients versus controls) and cytokine groups as explanatory variables. AKT1 expression was significantly related to IL-6 groups. Diagnosis and IL-10 groups were significant predictors of NDEL1 and DISC1 expression. DROSHA was significantly related to the diagnosis and IL-6 groups. COMT expression was predicted by the diagnosis, and MBP was associated with IL-10 and TNF-α groups.

Table 3 Results of univariate GLM analyses with the gene expression data as dependent variables and diagnosis (FEP versus controls) and interleukin (IL)-6, tumor necrosis factor (TNF)-α, and IL-10 groups (high versus low levels) as explanatory variables
Full size table

Table 4 shows the results of univariate GLM analyses with diagnostic groups, i.e., controls versus FEP patients with and without depression, as explanatory variables, and the genes that were significant in Table 1 as dependent variables. There were significant differences in NDEL1, DROSHA, COMT, and MBP, but not DISC1, between the three groups. There were additional differences in NDEL1 and COMT expression between patients with and without depression.

Table 4 ΔCrt values of five genes in healthy controls (HC) and first episode psychosis (FEP) patients divided into those with or without depression according to the cutoff score of 7 on the Calgary Depression Scale (CDSS)
Full size table

Table 5 shows the results of repeated measurement GLM analysis performed on the pre- and post-treatment gene expression data and considering the interaction time × IL-6 groups. We found significant time × IL-6 group interaction patterns for four genes, i.e., AKT1, DICER1, DROSHA, and COMT. Thus, the ΔCrt values of these four genes were decreased by treatment with risperidone in FEP patients with higher IL-6, but not in those with lower IL-6. There was additionally a significant effect of time on NDEL1 showing that the ΔCrt values were increased after treatment. The same table also shows the effects of treatment on the rating scale scores. All the psychopathological scale scores were significantly lowered by treatment with risperidone, while there was no significant interaction time × IL-6 groups.

Table 5 Results of repeated measurements on the mRNA and clinical data before and after treatment with risperidone in first episode psychosis (FEP) individuals divided or not into those with or without increased serum interleukin (IL)-6 levels
Full size table

Discussion

The first major finding of this investigation is that genes, which were associated with schizophrenia and/or bipolar disorder in previous studies, are significantly altered in blood of FEP patients. We found that NDEL1 and MBP genes were upregulated and DROSHA, COMT, and DISC1 were downregulated in FEP patients compared to controls. FEP patients also showed significantly increased serum levels of IL-10, IL-6, and TNF-α.

Interestingly, the NDEL1 and DISC1 were the most significant genes predicting FEP, and their combined use resulted in a significant diagnostic performance for FEP. NDEL1 encodes a protein involved in neuron outgrowth and migration, and cell signaling and is highly expressed in the brain [31]. A decrease in blood NDEL1 mRNA levels [32] and plasma NDEL1 enzyme activity [33] was reported in schizophrenia patients. In contrast to Kumarasinghe et al., who observed an upregulation of DISC1 in the schizophrenia patient group before and after antipsychotic treatment [34], we found a downregulation of this gene. DISC1 protein interacts with NDEL1, and this interaction mediates neurite outgrowth and neuronal migration [35–37]. The differences between our results and those in previous studies may be explained by differences in study sample selection, i.e., FEP patients versus schizophrenia patients in later stages and drug-free versus medicated patients.

The second major finding of this study is that FEP patients with depressive symptoms showed a different gene expression pattern compared to those without depressive symptoms. We found that FEP patients with more depressive symptoms (CDSS > 7) had an increased expression of COMT and decreased expression of NDEL1. Phrased differently, increased expression of NDEL1 and a lowered expression of COMT are hallmarks of FEP without depression. COMT is one of the enzymes responsible for the catabolism of dopamine in the brain, and the association between COMT Val158Met polymorphism and schizophrenia has been largely investigated [38]. Depression is a common and harmful dimension of schizophrenia, particularly in FEP [27, 39], and presents specific alterations in immune-inflammatory biomarkers [40, 41]. Recently, we have demonstrated that depression in FEP has a distinct cytokine profile, indicating that specific biological pathways may underpin this association [42]. Moreover, by investigating the co-expression of genes, Kim et al. (2015) observed a differential activation of the immune-inflammatory response across major psychiatric disorders, including schizophrenia and major depression [43]. Thus, the present results further underscore that FEP patients with depression are a biologically different phenotype than FEP patients without depression.

The third major finding of this study is that the circulating levels of cytokines, including IL-6, IL-10, and TNF-α, were significantly associated with the expression of several genes. As described in the Introduction, these cytokines modulate the expression of many different genes [44, 45]. We found that increased IL-6 was associated with lowered expression of AKT1 and DROSHA, while increased IL-10 was associated with increased NDEL1, DISC1, and MBP expression, all of them seem to play a role in neuronal processes and were significantly altered in FEP patients compared to controls. AKT1 signaling is involved in neuroplasticity and dopaminergic neurotransmission [46]. In addition, lithium, antidepressants, antipsychotics, and other mood stabilizers may increase phosphorylation of AKT1 [47–49]. Interestingly, there is also a connection between AKT1 and DISC1 signaling pathways [50]. DROSHA encodes a protein that is involved in the biogenesis of microRNA (miRNA), which is a short RNA with the capacity to target hundreds of genes, negatively regulating their expression in processes such as development and differentiation [51]. MBP is a component of the oligodendrocyte myelin and has a key role in myelin membrane biogenesis [52]. Genetic studies have described the involvement of myelin-related genes, including MBP, in schizophrenia [53, 54] and in FEP patients [55]. Thus, while increased IL-6 levels in FEP patients may downregulate cellular functions that are AKT1-mediated and dysregulate genes involved in miRNA machinery, increased IL-10 may show neuroprotective properties. Changes in the equilibrium between IL-6 and IL-10 production may therefore have consequences for neuronal functions. By inference, one possible pathway through which circulating cytokines may cause neuroprogression is regulation of the expression of genes related to neurogenesis, neuroplasticity, and neurodegeneration.

The fourth major finding is that treatment with risperidone has a significant effect on the expression of different genes, including AKT1, DROSHA, COMT, and NDEL1. Therefore, risperidone appears to normalize the initial disorders in gene expression by lowering NDEL1 and increasing COMT, DROSHA, and AKT1 expression. Most importantly, we found that the baseline circulating levels of IL-6 modulated the gene expression response of AKT1, DICER1, DROSHA, and COMT to treatment with risperidone. An upregulation of DICER1 was shown in the dorsolateral prefrontal cortex [56, 57] and lymphoblastoid cell lines [58] of schizophrenia cases. DICER1 is involved in the generation of miRNAs, similarly to DROSHA. Mutations in DICER1 abolish the generation of mature miRNAs and can induce deleterious developmental consequences, including malformations of the nervous system [59]. Thus, increased IL-6 levels appear to prime these genes to an enhanced response to risperidone, suggesting that cytokine × treatment interactions may improve cell function profiles (i.e., normalized DROSHA, COMT, and AKT1).

Our risperidone findings have important consequences for the interpretation of gene expression data in schizophrenia: (a) gene expression is only interpretable when fully controlled for the drug state of the patients and (b) baseline levels of circulating IL-6 should be taken into account for the interpretation of some gene expression data.

Conclusion

This study presents important data, shedding some light on gene expression in FEP and its interaction with depression and cytokine levels. We observed that gene expression and immune-inflammatory biomarkers are altered in FEP patients, as described previously. We also showed that FEP patients with depression present different expression of COMT and NDEL1 genes, strengthening the theory that distinct biological pathways underpin depression in FEP. Our results also suggest that circulating levels of IL-6 and IL-10 may regulate the expression of AKT1, DROSHA, NDEL1, DISC1, and MBP genes. We demonstrated that risperidone treatment may modulate gene expression and that these effects are partially regulated by cytokine levels, extending the comprehension of the role of the immune system in the pathophysiology of schizophrenia. The understanding of genetic and immune mechanisms in FEP opens new perspectives in the quest of biomarkers for the disorder and its treatment.