Abstract
Schizophrenia (SZ) is a devastating mental condition with onset in young adulthood. The identification of molecular biomarkers that reflect illness pathology is crucial. Recent evidence suggested immune and inflammatory cascades in conjunction with infection may play a role in the pathology. To address this question, we investigated molecular changes in cerebrospinal fluid (CSF) from antipsychotic-naïve patients with SZ and at risk mental status for psychosis (ARMS), in comparison with healthy controls (HCs). We measured 90 analytes using a broad multiplex platform focusing on immune and inflammatory cascades then selected 35 with our quality reporting criteria for further analysis. We also examined Toxoplasma gondii (TG) and herpes simplex virus 1 antibody levels in CSF. We report that expression of 15 molecules was significantly altered in the patient groups (SZ and ARMS) compared with HCs. The majority of these molecular changes (alpha-2-macroglobulin [α2M], fibrinogen, interleukin-6 receptor [IL-6R], stem cell factor [SCF], transforming growth factor alpha [TGFα], tumor necrosis factor receptor 2 [TNFR2], IL-8, monocyte chemotactic protein 2 [MCP-2/CCL8], testosterone [for males], angiotensin converting enzyme [ACE], and epidermal growth factor receptor) were consistent between SZ and ARMS patients, suggesting these may represent trait changes associated with psychotic conditions in general. Interestingly, many of these analytes (α2M, fibrinogen, IL-6R, SCF, TGFα, TNFR2, IL-8, MCP-2/CCL8, and testosterone [for males]) were exacerbated in subjects with ARMS compared with subjects with SZ. Although further studies are needed, we optimistically propose that these molecules may be good candidates for predictive markers for psychosis from an early stage. Lastly, reduction of IL-6R, TGFα, and ACE was correlated with positivity of TG antibody in the CSF, suggesting possible involvement of TG infection in the pathology.
Key words: schizophrenia, at risk mental status, inflammation, cerebrospinal fluid, biomarker, Toxoplasma gondii
Introduction
Schizophrenia (SZ) is a mental condition with many genetic and environmental risk factors.1 While individual genetic factors have a low effect size, environmental factors can also impact disease risk. These environmental stressors include obstetric complications, developmental nutritional deprivation, as well as possible infection with Toxoplasma gondii (TG) and other viruses, which are all risk factors for disease.2–6 Furthermore, an animal model for SZ includes developmental immune activation that leads to behavioral abnormalities in the offspring.7–9
Recent evidence suggested that immune and inflammatory cascades in conjunction with infection may play a role in the pathology of SZ.6,10–12 Patients with SZ displayed increased levels of TG antibodies compared with healthy controls (HCs).13,14 Furthermore, exposure to herpes simplex virus 1 (HSV1) was linked to poorer performance in cognitive tasks in SZ patients.15–17 Nonetheless, the mechanism of how such environmental stressors affect the pathology of the disease is unclear.
Molecular analysis may be a useful approach to address this question. Several studies utilized the plasma or serum of SZ patients with broad multiplex screens and found changes in many molecules.18–22 While there have been outstanding biomarker studies published using cerebrospinal fluid (CSF) from SZ patients, these studies applied candidate molecular approaches in which only a small number of molecules were tested.23–26 As far as we are aware, broad multiplex screens have not been applied to studies using CSF from SZ patients, in particular of antipsychotic-naïve patients.
In this study, we examined expression changes in dozens of molecules in the CSF from antipsychotic-naïve patients with SZ and at risk mental status for psychosis (ARMS), compared with HCs, by utilizing a broad multiplex screen. Furthermore, we explored how disease-associated infectious agents (TG and HSV1) affect these molecular changes.
Methods
Study Participants and Sample Collection
Participants were recruited from 2 clinical sites in Germany: 46 from University of Cologne and 50 from the Central Institute of Mental Health, Mannheim. All study participants provided written informed consent, and the respective institutes’ ethical committees approved the study protocol. Procedures were performed in accordance with the code of ethics of the world medical association for experiments involving humans (Declaration of Helsinki). HCs had no history of mental illness or psychotic episodes and were recruited within the same geographic area and population. Patients defined as having SZ met criteria of the DSM-IV. ARMS patients met the following 3 alternative criteria: (1) attenuated positive symptoms, (2) brief limited intermittent psychotic symptoms that spontaneously resolved within 1 week, and (3) a recent decline in function for at least 1 month, in combination with the existence of a first or second degree relative with a history of any DSM-IV psychotic disorder, or met criteria for schizotypal personality disorder.27–29 All patients were antipsychotic naïve, but some ARMS patients received low-dose benzodiazepines. Demographic information for age, gender, body mass index (BMI), ethnicity, years of education, and smoking status is given in table 1. Patients and HCs were demographically similar for age, gender, BMI, and smoking status as described in Methods and table 1. Demographic information for each cohort is provided in supplementary table 1. CSF samples were collected and stored at −80°C or lower, and samples were thawed for aliquoting before each analysis.5
Table 1.
Clinical and Demographic Information for Patients and HCs
| HCs | SZ | ARMS | Test | P Value | |
|---|---|---|---|---|---|
| N | 35 | 46 | 15 | ||
| Agea | 26.4±0.6 | 25.8±0.8 | 24.9±1.0 | ANOVA | 0.531 |
| Gender (female/male) | 14/21 | 10/36 | 7/8 | Fisher’s Exact | 0.088 |
| BMIa | 23.3±0.5 | 23.0±0.4 | 23.1±1.1 | ANOVA | 0.912 |
| Ethnicityb | 34/1/0 | 46/0/0 | 13/1/1 | Fisher’s Exact | 0.040 |
| Educationc (≥13/<13 years) | 33/1 | 21/25 | 4/11 | Fisher’s Exact | 1.04E−08 |
| Smoking (yes/no) | 17/18 | 30/16 | 5/10 | Fisher’s Exact | 0.074 |
| Antipsychotic medication | 0% | 0% | 0% |
Note: Demographic information is reported for all participants within each diagnostic group (healthy controls [HCs], schizophrenia [SZ], and at risk mental status for psychosis [ARMS]). We report information on the number of participants (N), participants’ age, gender, body mass index (BMI), ethnicity, years of education (percentage that completed at least 13 years, a high-school diploma equivalent), smoking, and antipsychotic treatment (all patients were antipsychotic naïve). HCs and patients were statistically similar for age, gender, BMI, and smoking. Significantly fewer patients finished their education during illness. The majority of participants were Caucasian, but 3 individuals were non-Caucasian, 1 HC and 2 patients.
aMean ± SEM.
bCaucasian/Asian/Other.
cIndividuals who completed at least 13 years of education or a high-school diploma equivalent.
Analyte Measurements
Sample aliquots were shipped to Myriad-Rules Based Medicine (RBM) for evaluation using the multianalyte profiling based on multiplex immunoassay platforms to simultaneously quantify panels of analytes. Sample processing at RBM facilities were described in detail previously.26 We designed a custom analyte panel of 90 markers including inflammatory cytokines, chemokines, metabolic factors, and growth factors. Assays were developed, validated, and raw intensities converted to absolute concentrations by comparison with a standard curve by the Clinical Laboratory Improvement Amendment certified laboratories at Myriad-RBM. Final data were reported as the absolute concentrations in the CSF, less than the least detectable dose (LDD), or quantity not sufficient (QNS).
Infectious Antibody Measurements
TG immunoglobulin G (IgG) antibody levels were measured using a commercially available ELISA kit from IBL America. HSV1 IgG antibody levels were measured using a commercially available ELISA kit from Focus Diagnostics. All assays were performed according to the manufacture instructions. The IgG levels for these infectious agents were recorded as absorbance values and LOG normalized before further analysis. Kit standards were run with each plate to control for interplate variation and identify antibody positive and negative individuals.
Statistical Analysis
All statistical analyses were performed using the R statistical analysis software (version 3.0.2). Reproducible R code for each analysis is provided in the supplementary methods.
Patients and HCs Demographic Characteristics.
HCs and patients with SZ, and ARMS across both cohorts (Cologne and Mannheim) were comparable for age (P = 0.53) and BMI (P = 0.91) using analysis of variance (ANOVA) across each diagnosis (table 1). Demographic similarity for gender (female or male, P = 0.088), education duration (≥13 years or <13 years, P = 1.04E−08), and smoking (yes or no, P = 0.074) was analyzed using a 2 × 3 Fisher’s exact test (table 1). Ethnicity (Caucasian, Asian, or Other, P = 0.040) was analyzed using a 3 × 3 Fisher’s exact test (table 1). The patients completed significantly fewer years of education than the HCs possibly due to a need for treatment during the illness. Ethnicity was not statistically similar, but the vast majority of patients were Caucasian with only 3 non-Caucasian individuals, 1 HC and 2 patients (table 1).
Analyte Expression Analysis.
Among the 90 analytes we examined, we specifically selected 35 for further analysis based on having less than 5% QNS values and at least 75% of values reported: 77% (27 analytes) required no imputing, 11% (4 analytes) required less than 15% imputation, and 11% (the last 4 analytes) required more than 15% imputation (supplementary table 2). All data were LOG normalized before further analysis. We fit a parametric Gaussian survival regression model relating each analyte to diagnosis, cohort, and smoking status. Removing smoking status from our model did not affect the result outcomes; thus, smoking status does not significantly contribute to the results (see supplementary methods for R code). The survival model treated the analytes as left censored if the values were below the LDD. The model was fit using the survival package in R, see supplementary material R code and documentation for exact details of the model.30–32 The QNS values were not included in the analysis. An ANOVA (lambda = 0.7) among the diagnoses was performed on the survival model. For testosterone, the females were below the LDD; thus, the analyses were computed on only the male participants. Fold change was calculated as the ratio of the median expression of patients to HCs. Correction for multiple comparisons was evaluated using a q-value conversion to control the false discovery rate; thus, a q value below 0.05 is considered significant.33 Median was calculated from the untransformed data and reported as median and interquartile range (IQR). Box and whisker plots were used to graphically represent the transformed data providing the median, first and third quartile, range, and outliers. We performed a correlation between each analyte to identify related analytes and represented the data in a heatmap with the strongest correlations listed in a table; missing data were imputed to the LDD or the analyte average for values below the LDD and QNS, respectively (see supplementary methods for R code and supplementary figure 1).
Comparison of Infectious Agents in Patients vs HCs.
To compare the level of infectious agents between each diagnostic group, we performed an ANOVA for the level of infectious antibodies (HSV1 or TG) among the diagnoses (HCs and patients with SZ and ARMS) using site as a covariate. Next, to further investigate significant associations, we performed a Fisher’s exact test by assigning the participants into dichotomous positive or negative groups for HSV1 and TG based on the negative control values in each ELISA assay.
Infectious Agent and Analyte Correlation Analysis.
Similar to above, parametric survival analysis was performed on normalized data, and an ANOVA was performed between the infectious agents (HSV1 and TG) and each analyte. Correction for multiple comparisons was performed using q value conversion.33 In addition, a linear regression was performed on the significant analytes (TG q < 0.05, HSV1 p < 0.05), and the correlation coefficient (r) and adjusted R 2 are reported. Next, the participants were categorized as HSV1 positive, HSV1 negative, TG positive, and TG negative depending on whether the antibody level was above or below the negative control, and a Student’s t-test was performed on the analyte expression level of the identified molecules between those positive and negative for each infectious agent.
Results
Differential Molecular Expression Between Patients With Psychosis and HCs
First, we compared the HC group with the patient group (SZ and ARMS together), which resulted in 5 differentially expressed analytes (table 2, supplementary figures 2–3): Angiotensin converting enzyme (ACE) and fibrinogen were significantly reduced in the patient group (supplementary figure S2A-B), whereas interleukin-8 (IL-8), monocyte chemotactic protein 2 (MCP-2/CCL8), and testosterone were increased (supplementary figure S2C-E). Testosterone was compared only in male participants.
Table 2.
Differential Analyte Expression Between all Patients and HCs
| Protein | Units | HCa | Psya | Fold Change | q Value |
|---|---|---|---|---|---|
| ACE | ng/ml | 1.60 (0.75) | 1.30 (0.50) | 0.81 | 0.035 |
| Fibrinogen | µg/ml | 0.170 (0.14) | 0.120 (0.16) | 0.71 | 0.011 |
| IL-8 | pg/ml | 21.0 (9.5) | 27.0 (14) | 1.29 | 0.011 |
| MCP-2/CCL8 | pg/ml | 2.15 (0.50) | 3.10 (1.5) | 1.44 | 0.011 |
| Testosterone | ng/ml | 0.085 (0.088) | 0.185 (0.11) | 2.18 | 0.011 |
Note: Analyte expression from all patients (Psy) compared with healthy controls (HCs). Angiotensin converting enzyme (ACE) and fibrinogen were decreased in the patient population, whereas interleukin-8 (IL-8), monocyte chemoattractant protein 2 (MCP-2/CCL8), and testosterone were increased in the patient population. Testosterone was only compared in male participants. IQR = interquartile range, Fold change = Psy/HCs.
aMedian (IQR).
Next, we compared HC, SZ, and ARMS groups separately. Fifteen analytes were significantly different among these groups (table 3). Depending on the pattern of expression change, we divided these 15 analytes into 5 subgroups (figure 1, supplementary figure 4). Six analytes in group 1A (alpha-2-macroglobulin [α2M], fibrinogen, interleukin-6 receptor [IL-6R], stem cell factor [SCF], transforming growth factor alpha [TGFα], and tumor necrosis factor receptor 2 [TNFR2]) were decreased in the SZ group with further reduction in the ARMS group (figures 1A–F). Three analytes in Group 1B (IL-8, MCP-2/CCL8, and testosterone [males]) were increased in the SZ group with further augmentation in the ARMS group (figures 1G–I). We speculate that these 2 groups represent molecular changes associated with psychosis, but the extent of the changes (regardless of increase or decrease) is greater in the risk stage. Group 2 analytes (ACE and epidermal growth factor receptor [EGFR]) were decreased in patients with SZ and ARMS similarly, possibly representing traits associated with psychosis in general (figures 1J and K). In contrast, group 3 analytes (adiponectin and matrix metalloproteinase 3 [MMP3]) were uniquely decreased only in ARMS patients (figures 1L and M). Finally, group 4 (complement 3 [C3] and interleukin-6 [IL-6]) showed an increase in SZ patients and decrease in subjects with ARMS (figures 1N and O): C3 was only slightly increased in SZ patients with a more prominent decrease in the ARMS group, whereas IL-6 showed a substantial increase in SZ patients and a slight decrease in patients with ARMS.
Table 3.
Differential Analyte Expression Among HCs and SZ and ARMS Patients
| Analyte | Units | HCsa | SZa | ARMSa | FC, SZ/ARMS | q Value | Group |
|---|---|---|---|---|---|---|---|
| α2M | µg/ml | 3.30 (0.80) | 3.05 (0.80) | 2.80 (0.80) | 0.92/0.85 | 0.023 | 1A |
| ACE | ng/ml | 1.60 (0.75) | 1.35 (0.50) | 1.30 (0.71) | 0.84/0.81 | 0.020 | 2 |
| Adiponectin | ng/ml | 3.60 (2.25) | 3.30 (2.2) | 1.50 (2.1) | 0.92/0.42 | 0.026 | 3 |
| C3 | µg/ml | 1.90 (1.10) | 2.00 (0.88) | 1.70 (0.7) | 1.05/0.89 | 0.037 | 4 |
| EGFR | ng/ml | 0.140 (0.04) | 0.125 (0.052) | 0.120 (0.058) | 0.89/0.86 | 0.027 | 2 |
| Fibrinogen | µg/ml | 0.170 (0.14) | 0.130 (0.22) | 0.064 (0.13) | 0.76/0.38 | 0.020 | 1A |
| IL-6 | pg/ml | 0.920 (0.58) | 1.400 (1.07) | 0.690 (0.77) | 1.52/0.75 | 0.020 | 4 |
| IL-6R | ng/ml | 0.970 (0.43) | 0.855 (0.34) | 0.750 (0.39) | 0.88/0.77 | 0.023 | 1A |
| IL-8 | pg/ml | 21.00 (9.5) | 25.00 (15.0) | 27.00 (9.0) | 1.19/1.29 | 0.022 | 1B |
| MMP3 | ng/ml | 0.160 (0.09) | 0.155 (0.088) | 0.110 (0.059) | 0.97/0.69 | 0.045 | 3 |
| MCP-2/CCL8 | pg/ml | 2.15 (0.50) | 2.90 (1.65) | 3.40 (0.95) | 1.35/1.58 | 0.020 | 1B |
| SCF | pg/ml | 49.00 (17.5) | 44.00 (17.5) | 38.00 (18.5) | 0.90/0.78 | 0.026 | 1A |
| TGFα | pg/ml | 22.00 (7.0) | 20.00 (9.0) | 19.00 (0.15) | 0.91/0.86 | 0.020 | 1A |
| TNFR2 | ng/ml | 0.490 (0.16) | 0.440 (0.23) | 0.380 (0.26) | 0.90/0.78 | 0.041 | 1A |
| Testosterone | ng/ml | 0.085 (0.088) | 0.150 (0.10) | 0.235 (0.068) | 1.76/2.76 | 0.020 | 1B |
Note: Expression changes between healthy controls (HCs), schizophrenia (SZ), and at risk mental status for psychosis (ARMS) were the following: Alpha-2 macroglobulin (α2M), angiotensin converting enzyme (ACE), complement 3 (C3), epidermal growth factor receptor (EGFR), fibrinogen, interleukin-6 (IL-6), interleukin-6 receptor (IL-6R), interleukin-8 (IL-8), matrix metalloproteinase 3 (MMP3), monocyte chemoattractant protein 2 (MCP-2/CCL8), stem cell factor (SCF), transforming growth factor alpha (TGFα), tumor necrosis factor receptor 2 (TNFR2), and testosterone. Fold change (FC) reported as (SZ/HCs)/(ARMS/HCs). Bold values were listed in table 2. Testosterone analyzed only in males. IQR = interquartile range.
aMedian (IQR).
Fig. 1.
Box and whisker plots of differentially expressed analytes between patients with schizophrenia (SZ), at risk mental status for psychosis (ARMS), and healthy controls (HCs). (A–F) (Group 1A) Analytes with decreased expression in SZ patients with ARMS showing the greatest decrease. Group 1A included α-2 macroglobulin (α2M), fibrinogen, interleukin-6 receptor (IL-6R), stem cell factor (SCF), transforming growth factor α (TGFα), and tumor necrosis factor receptor 2 (TNFR2). (G–I) (Group 1B) Analytes with increased expression in SZ patients and further increase in ARMS patients, including IL-8, monocyte chemotactic protein 2 (MCP-2/CCL8), and testosterone. (J and K) (Group 2) Angiotensin converting enzyme (ACE) and epidermal growth factor receptor (EGFR) were decreased comparably in SZ and ARMS patients. (L and M) (Group 3) Adiponectin and matrix metalloproteinase 3 (MMP3) were only decreased in ARMS patients. (N and O) (Group 4) Complement 3 (C3) and IL-6 in SZ and ARMS patients were changed in opposite directions. C3 showed a slight increase in SZ patients with a more pronounced decrease in ARMS patients. IL-6 showed a substantial increase in SZ patients and a slight decrease in ARMS patients. Testosterone was only analyzed in males. The analyte concentrations were LOG transformed before analysis; thus, the axis displays the transformed concentration (see table 3 for values). Box and whisker plots display the median (bold line), first and third quartile (box), range (dashed lines), and outliers (circles).
Infectious Agents Effect on Analyte Expression
Given that psychotic disorders, such as SZ, occur as the outcome of gene and environmental interactions, it is very important to clarify how environmental stressors, such as infectious agents, may be related to the molecular signatures we uncovered above. To address this question, we measured the level of antibodies for HSV1 and TG in the CSF and correlated the antibody levels with the changes in analyte expression studied above.
First, we failed to observe a significant change in the antibody levels between groups of HC and patients (SZ and ARMS together) for HSV1 or TG. Thus, we next addressed the correlation between each analyte expression and antibody level evaluating all groups together by an unbiased parametric survival analysis. As for HSV1, the antibody levels were correlated with α2M, ACE, TGFα, and vitamin D binding protein (VDBP) (table 4A). However, these did not remain significant after multiple comparison correction. In contrast, the levels of an anti-TG antibody were significantly correlated with α-1 antitrypsin (AAT), ACE, β-2 microglobulin (β2M), EGFR, heparin binding-epidermal growth factor (HB-EGF), IL-6R, macrophage inflammatory protein β (CCL4), SCF, and TGFα, all of which survive multiple comparison correction and remain statistically significant (table 4A).
Table 4.
Correlation of Analyte Expression With Each Infectious Agent
| (A) | ||||
|---|---|---|---|---|
| HSV1 | r | R 2 | P Value | q Value |
| α2M* | 1.1414 | 0.0485 | 0.0161 | 0.0871 |
| ACE* | 0.6908 | 0.0371 | 0.0312 | 0.0871 |
| TGFα* | 0.7246 | 0.0352 | 0.0348 | 0.0871 |
| VDBP | 0.4227 | 0.0389 | 0.0281 | 0.0871 |
| TG | ||||
| AAT | −0.5461 | 0.0343 | 0.0367 | 0.0329 |
| ACE* | −0.6836 | 0.0429 | 0.0222 | 0.0329 |
| β2M | −0.4763 | 0.0282 | 0.0524 | 0.0329 |
| EGFR* | −0.7412 | 0.0630 | 0.0070 | 0.0329 |
| HB-EGF | −2.0272 | 0.0498 | 0.0149 | 0.0329 |
| IL-6R* | −0.6377 | 0.0352 | 0.0348 | 0.0329 |
| MIP-1β/CCL4 | −0.4779 | 0.0336 | 0.0382 | 0.0329 |
| SCF* | −0.6802 | 0.0330 | 0.0395 | 0.0329 |
| TGFα* | −0.6816 | 0.0358 | 0.0335 | 0.0329 |
| (B) | ||||
| HSV1 | Positivea, N = 34 | Negativea, N = 62 | P Value | |
| α2M (µg/ml) | 3.42±0.15 | 2.96±0.08 | 0.016 | |
| ACE (ng/ml) | 1.67±0.10 | 1.40±0.06 | 0.019 | |
| TGFα (pg/ml) | 24.12±1.22 | 20.62±0.78 | 0.016 | |
| VDBP (µg/ml) | 1.34±0.11 | 1.08±0.08 | 0.023 | |
| TG | Positivea, N = 18 | Negativea, N = 78 | P Value | |
| AAT (µg/ml) | 5.08±0.57 | 6.19±0.22 | 0.064 | |
| ACE (ng/ml) | 1.26±0.13 | 1.55±0.06 | 0.043 | |
| β2M (µg/ml) | 0.666±0.06 | 0.851±0.047 | 0.055 | |
| EGFR (ng/ml) | 0.112±0.01 | 0.135±0.005 | 0.072 | |
| HB-EGF (pg/ml) | 152.4±4.96 | 164.2±2.35 | 0.049 | |
| IL-6R (ng/ml) | 0.74±0.06 | 0.92±0.03 | 0.040 | |
| MIP-1β/CCL4 (pg/ml) | 16.6±1.40 | 21.5±1.32 | 0.067 | |
| SCF (pg/ml) | 41.2±3.60 | 48.8±1.80 | 0.067 | |
| TGFα (pg/ml) | 18.7±1.38 | 22.6±0.75 | 0.040 | |
Note: (A) Correlation of each infectious agent antibody levels with the CSF analyte expressions. The top correlations are listed. HSV1 correlations with α2M, ACE, TGFα, and vitamin D binding protein (VDBP) had a P < 0.05 but did not maintain significance after multiple comparison correction (q > 0.05). TG was significantly correlated with α-1 antitrypsin (AAT), ACE, β-2-microglobulin (β2M), EGFR, heparin binding epidermal growth factor (HB-EGF), IL-6R, macrophage inflammatory protein 1β (MIP-1β/CCL4), SCF, and TGFα (q < 0.05). r = correlation coefficient. R 2 = adjusted R 2. Asterisks indicate analytes differentially expressed in patients from table 3. (B) HSV1-positive participants expressed significantly more α2M, ACE, TGFα, and VDBP. TG-positive participants expressed significantly less ACE, HB-EGF, IL-6R, and TGFα. Analyte expression is reported as mean ± standard error of the mean (SEM). Bold values indicate significance. N = number of participants in each group.
aMean ± SEM.
Next, we divided the subjects into 4 groups: HSV1 positive, HSV1 negative, TG positive, and TG negative (as described in the Methods section, the threshold was determined by the kit standards). Then we tested whether the protein expression, listed in table 4A, was different between these positive and negative groups. We found that the levels of α2M, ACE, TGFα, and VDBP were significantly higher in the CSF of HSV1-positive subjects compared with HSV1-negative subjects (table 4B). Interestingly, for the TG positive subjects, we found significantly less expression of ACE, HB-EGF, IL-6R, and TGFα (table 4B).
Discussion
In this study, we reported 15 analytes that were differentially expressed among HC, SZ, and ARMS subjects. These analytes were subdivided into 5 groups, depending on the pattern of expression change. We highlighted sets of analytes that are changed consistently in SZ and ARMS patients compared with HCs. Moreover, the extent of change in some analytes was greater in the ARMS group compared with the SZ group. Furthermore, we showed how the change of these molecules is related to disease-associated infectious agents, HSV1 and TG.
We believe this study has strengths in the following 2 points. First, all subjects with SZ and ARMS were antipsychotic naïve and only a small number of subjects received low doses of benzodiazepine. Thus, this sample set is very useful to address the molecular signature directly associated with psychotic disorders, particularly in the early phase of psychosis. Furthermore, in contrast to most studies with CSF that utilized candidate molecular approaches in which only a small number of target molecules were evaluated per study, we used a broad multiplex platform to test dozens of molecules simultaneously. Second, in addition to addressing molecular changes in patients with psychosis, we explored the possible relationship of such changes in the CSF to disease-associated infectious agents by noting the CSF antibody levels against HSV1 and TG.
Our study highlighted important molecular markers for psychosis but was limited by our small sample size and difference in educational level. The difference in education maybe due to disrupted education or cognitive dysfunction in patients but not in HCs. To accurately represent the HCs and avoid introducing a bias, we did not restrict our HCs to a less educated population. However, the patients and controls were similar for the other demographic characteristics.
Some studies, but not all, demonstrated an increased level of TG antibodies in patients with SZ. Our results do not negate these findings but merely were not replicated in our subset of participants. We further analyzed how infectious agents might affect the CSF analytes beyond the classification of diagnosis, based on the idea that the infectious agents affect host factors that underlie the pathophysiology of the cases. Our findings are further validated because previous studies found that HSV1 binding to α2M promoted infectivity in a neuronal cell line.34 In addition, a previous study showed infection of macrophages, in vitro, with TG blocked production of proinflammatory cytokines, such as IL-6, and blocked the increase of over 50 LPS-induced cytokines,35 providing evidence for broad protein suppression upon TG infection.
A number of the differentially expressed proteins uncovered in this study directly impact EGF signaling. Overall, we found decreases in EGFR, fibrinogen, and TGFα in patients compared with HCs. We also found a negative correlation of TG with EGFR, HB-EGF, and TGFα. EGFR (ErbB1) is also relevant in SZ because EGFR forms heterodimers with other ErbB receptors, such as ErbB4, a known genetic risk factor for SZ36. Thus, EGF signaling may be a candidate to investigate gene and environmental interactions for psychotic conditions. Furthermore, previous studies of cardiovascular and renal disease identified that ACE can affect the activation of EGF signaling by modulating the phosphorylation status of the EGF receptors or the downstream EGFR signaling intermediates.37,38
It is also interesting that the IL-6 ligand and receptor were significantly changed in opposite directions in SZ patients. Coughlin et al26 also reported this observation of IL-6/IL-6R in the CSF from recent-onset SZ patients. Conceivably, an increase in the level of IL-6 ligand in the context of immune activation drives a subsequent decrease in IL-6R as a means of compensating for the overproduced ligand. In addition, the idea of compensation is observed by the increase in testosterone and decrease in ACE. A human study of multiple sclerosis showed treatment with testosterone provided immune protection and symptom improvement.39 Furthermore, ACE produces angiotensin II that binds to the angiotensin receptor 1 to elicit a proinflammatory immune response in microglia.40 Taken together, increased testosterone, decreased ACE, and increased IL-6R may be comprehensively interpreted as homeostatic compensation for a baseline increase in inflammation in patients.
It is noteworthy that the majority of the molecular changes (11 out of 15, α2M, fibrinogen, IL-6R, SCF, TGFα, TNFR2, IL-8, MCP-2/CCL8, testosterone [for males], ACE, and EGFR] were consistent between SZ and ARMS groups compared with HCs. Thus, these likely represent trait changes associated with psychotic conditions in general. Among these, disease-associated reduction of IL-6R, TGFα, and ACE was correlated with positivity of TG antibody.
Identifying early molecular changes is important for early detection and treatment often leading to better outcomes. Of clinical interest in this study, the ARMS subjects exhibited greater changes in most of these analytes (9 out of 11) compared with the SZ patients. Given that ARMS included subjects at risk for or in early stage of psychosis and SZ, the intermediate molecular profile observed in SZ may be the result of compensatory and homeostatic feedback mechanisms initiated after the early stage. There is precedence that disease signatures become less prominent during the course of disease progression as some viral infections, such as human immunodeficiency virus (HIV) or syphilis, can have dormant phases after a more severe initial infection. Although further studies with much larger samples are needed, we optimistically suggest that these 9 molecules (α2M, fibrinogen, IL-6R, SCF, TGFα, TNFR2, IL-8, MCP-2/CCL8, and testosterone [for males]) may be good candidates for predictive markers for psychosis from an early stage.
Supplementary Material
Supplementary material is available at http://schizophreniabulletin.oxfordjournals.org.
Funding
National Institutes of Health (MH-084018 to A.S., MH-94268 Silvo O. Conte center to A.S., J.T.L., E.G.S., and R.H.Y., MH-069853 to A.S., MH-085226 to A.S., MH-088753 to A.S., MH-092443 to A.S.); grants from Stanley (A.S.); S-R (A.S.); RUSK (A.S.); NARSAD (A.S., J.M.C); MSCRF (A.S.); the Stanley Medical Research Institute (E.G.S., K.L.G., F.M.L., and R.H.Y.).
Supplementary Material
Acknowledgments
We thank the individuals who participated in our study. We thank the Sawa, Leweke, and Yolken lab members for both technical assistance in sample preparation and handling and critical comments on the project. We thank Anna Chang for critical reading of the manuscript. We also thank William Eaton for guidance on statistical analyses. The Authors have declared that there are no conflicts of interest in relation to the subject of this study.
References
- 1. McGrath JJ, Mortensen PB, Visscher PM, Wray NR. Where GWAS and epidemiology meet: opportunities for the simultaneous study of genetic and environmental risk factors in schizophrenia. Schizophr Bull. 2013;39:955–959 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Brown AS. Epidemiologic studies of exposure to prenatal infection and risk of schizophrenia and autism. Dev Neurobiol. 2012;72:1272–1276 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Husted JA, Ahmed R, Chow EW, Brzustowicz LM, Bassett AS. Early environmental exposures influence schizophrenia expression even in the presence of strong genetic predisposition. Schizophr Res. 2012;137:166–168 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Singh S, Kumar A, Agarwal S, Phadke SR. Genetic insight of schizophrenia: past and future perspectives. Gene. 2013;535 :97–100 [DOI] [PubMed] [Google Scholar]
- 5. Leweke FM, Gerth CW, Koethe D, et al. Antibodies to infectious agents in individuals with recent onset schizophrenia. Eur Arch Psychiatry Clin Neurosci. 2004;254:4–8 [DOI] [PubMed] [Google Scholar]
- 6. Yolken RH, Torrey EF. Are some cases of psychosis caused by microbial agents? A review of the evidence. Mol Psychiatry. 2008;13:470–479 [DOI] [PubMed] [Google Scholar]
- 7. Meyer U, Feldon J. Neural basis of psychosis-related behaviour in the infection model of schizophrenia. Behav Brain Res. 2009;204:322–334 [DOI] [PubMed] [Google Scholar]
- 8. Giovanoli S, Engler H, Engler A, et al. Stress in puberty unmasks latent neuropathological consequences of prenatal immune activation in mice. Science. 2013;339:1095–1099 [DOI] [PubMed] [Google Scholar]
- 9. Canetta SE, Brown AS. Prenatal infection, maternal immune activation, and risk for schizophrenia. Transl Neurosci. 2012;3:320–327 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Kirkpatrick B, Miller BJ. Inflammation and schizophrenia. Schizophr Bull. 2013;39:1174–1179 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Benros ME, Eaton WW, Mortensen PB. The epidemiologic evidence linking autoimmune diseases and psychosis. Biol Psychiatry. 2014;75:300–306 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Brown AS. The environment and susceptibility to schizophrenia. Prog Neurobiol. 2011;93:23–58 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Torrey EF, Bartko JJ, Yolken RH. Toxoplasma gondii and other risk factors for schizophrenia: an update. Schizophr Bull. 2012;38:642–647 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Arias I, Sorlozano A, Villegas E, et al. Infectious agents associated with schizophrenia: a meta-analysis. Schizophr Res. 2012;136:128–136 [DOI] [PubMed] [Google Scholar]
- 15. Yolken RH, Torrey EF, Lieberman JA, Yang S, Dickerson FB. Serological evidence of exposure to Herpes Simplex Virus type 1 is associated with cognitive deficits in the CATIE schizophrenia sample. Schizophr Res. 2011;128:61–65 [DOI] [PubMed] [Google Scholar]
- 16. Prasad KM, Eack SM, Goradia D, et al. Progressive gray matter loss and changes in cognitive functioning associated with exposure to herpes simplex virus 1 in schizophrenia: a longitudinal study. Am J Psychiatry. 2011;168:822–830 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Shirts BH, Prasad KM, Pogue-Geile MF, Dickerson F, Yolken RH, Nimgaonkar VL. Antibodies to cytomegalovirus and Herpes Simplex Virus 1 associated with cognitive function in schizophrenia. Schizophr Res. 2008;106:268–274 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. García-Bueno B, Bioque M, Mac-Dowell KS, et al. Pro-/Anti-inflammatory dysregulation in patients with first episode of psychosis: Toward an integrative inflammatory hypothesis of schizophrenia. Schizophr Bull. 2014;40:376–387 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Ramsey JM, Schwarz E, Guest PC, et al. Distinct molecular phenotypes in male and female schizophrenia patients. PLoS One. 2013;8:e78729. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Schwarz E, van Beveren NJ, Ramsey J, et al. Identification of subgroups of schizophrenia patients with changes in either immune or growth factor and hormonal pathways. Schizophr Bull. 2013. Sep 11. [Epub ahead of print]. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Schwarz E, Guest PC, Rahmoune H, et al. Identification of a biological signature for schizophrenia in serum. Mol Psychiatry. 2012;17:494–502 [DOI] [PubMed] [Google Scholar]
- 22. Domenici E, Willé DR, Tozzi F, et al. Plasma protein biomarkers for depression and schizophrenia by multi analyte profiling of case-control collections. PLoS One. 2010;5:e9166. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Huang JT, Leweke FM, Tsang TM, et al. CSF metabolic and proteomic profiles in patients prodromal for psychosis. PLoS One. 2007;2:e756. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Holmes E, Tsang TM, Huang JT, et al. Metabolic profiling of CSF: evidence that early intervention may impact on disease progression and outcome in schizophrenia. PLoS Med. 2006;3:e327. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Miller BJ, Buckley P, Seabolt W, Mellor A, Kirkpatrick B. Meta-analysis of cytokine alterations in schizophrenia: clinical status and antipsychotic effects. Biol Psychiatry. 2011;70:663–671 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Coughlin JM, Ishizuka K, Kano SI, et al. Marked reduction of soluble superoxide dismutase-1 (SOD1) in cerebrospinal fluid of patients with recent-onset schizophrenia. Mol Psychiatry. 2013;18:10–11 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Ruhrmann S, Schultze-Lutter F, Salokangas RK, et al. Prediction of psychosis in adolescents and young adults at high risk: results from the prospective European prediction of psychosis study. Arch Gen Psychiatry. 2010;67:241–251 [DOI] [PubMed] [Google Scholar]
- 28. Ruhrmann S, Schultze-Lutter F, Klosterkötter J. Probably at-risk, but certainly ill–advocating the introduction of a psychosis spectrum disorder in DSM-V. Schizophr Res. 2010;120:23–37 [DOI] [PubMed] [Google Scholar]
- 29. McGlashan TH. Psychosis treatment prior to psychosis onset: ethical issues. Schizophr Res. 2001;51:47–54 [DOI] [PubMed] [Google Scholar]
- 30. Therneau TM. A Package for Survival Analysis in S. R Package Version 2.37–4; 27 March 2013. http://CRAN.R-project.org/package=survival
- 31. Klein JP, Moeschberger ML. Survival Analysis: Techniques for Censored and Truncated Data. New York, NY: Springer; 2003 [Google Scholar]
- 32. Therneau TM, Grambsch PM. Modeling Survival Data: Extending the Cox Model. New York, NY: Springer; 2000 [Google Scholar]
- 33. Storey JD, Tibshirani R. Statistical significance for genomewide studies. Proc Natl Acad Sci USA. 2003;100:9440–9445 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34. Alonso M, Dimitrijevic A, Recuero M, Serrano E, Valdivieso F, López-Guerrero JA. Interaction of alpha-2-macroglobulin and HSV-1 during infection of neuronal cells. J Neurovirol. 2001;7:556–563 [DOI] [PubMed] [Google Scholar]
- 35. Lee CW, Bennouna S, Denkers EY. Screening for Toxoplasma gondii-regulated transcriptional responses in lipopolysaccharide-activated macrophages. Infect Immun. 2006;74:1916–1923 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36. Walsh T, McClellan JM, McCarthy SE, et al. Rare structural variants disrupt multiple genes in neurodevelopmental pathways in schizophrenia. Science. 2008;320:539–543 [DOI] [PubMed] [Google Scholar]
- 37. Wolf G. “As time goes by”: angiotensin II-mediated transactivation of the EGF receptor comes of age. Nephrol Dial Transplant. 2005;20:2050–2053 [DOI] [PubMed] [Google Scholar]
- 38. Nakano S, Kobayashi N, Yoshida K, Ohno T, Matsuoka H. Cardioprotective mechanisms of spironolactone associated with the angiotensin-converting enzyme/epidermal growth factor receptor/extracellular signal-regulated kinases, NAD(P)H oxidase/lectin-like oxidized low-density lipoprotein receptor-1, and Rho-kinase pathways in aldosterone/salt-induced hypertensive rats. Hypertens Res. 2005;28:925–936 [DOI] [PubMed] [Google Scholar]
- 39. Gold SM, Chalifoux S, Giesser BS, Voskuhl RR. Immune modulation and increased neurotrophic factor production in multiple sclerosis patients treated with testosterone. J Neuroinflammation. 2008;5:32. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40. Saavedra JM. Angiotensin II AT(1) receptor blockers as treatments for inflammatory brain disorders. Clin Sci (Lond). 2012;123:567–590 [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.