Identification of blood biomarkers for psychosis using convergent functional genomics

Article metrics

Abstract

There are to date no objective clinical laboratory blood tests for psychotic disease states. We provide proof of principle for a convergent functional genomics (CFG) approach to help identify and prioritize blood biomarkers for two key psychotic symptoms, one sensory (hallucinations) and one cognitive (delusions). We used gene expression profiling in whole blood samples from patients with schizophrenia and related disorders, with phenotypic information collected at the time of blood draw, then cross-matched the data with other human and animal model lines of evidence. Topping our list of candidate blood biomarkers for hallucinations, we have four genes decreased in expression in high hallucinations states (Fn1, Rhobtb3, Aldh1l1, Mpp3), and three genes increased in high hallucinations states (Arhgef9, Phlda1, S100a6). All of these genes have prior evidence of differential expression in schizophrenia patients. At the top of our list of candidate blood biomarkers for delusions, we have 15 genes decreased in expression in high delusions states (such as Drd2, Apoe, Scamp1, Fn1, Idh1, Aldh1l1), and 16 genes increased in high delusions states (such as Nrg1, Egr1, Pvalb, Dctn1, Nmt1, Tob2). Twenty-five of these genes have prior evidence of differential expression in schizophrenia patients. Predictive scores, based on panels of top candidate biomarkers, show good sensitivity and negative predictive value for detecting high psychosis states in the original cohort as well as in three additional cohorts. These results have implications for the development of objective laboratory tests to measure illness severity and response to treatment in devastating disorders such as schizophrenia.

Introduction

Our group has developed a powerful combined approach for extracting signal from noise in genetic and gene expression studies, termed convergent functional genomics (CFG). CFG translationally integrates multiple independent lines of evidence-genetic and functional genomic data, from human studies and animal models, as a Bayesian strategy for identifying and prioritizing findings, reducing the false-positives and false-negatives inherent in each individual approach. The CFG methodology has already been applied with some success to help identify and prioritize candidate genes, pathways and mechanisms for neuropsychiatric disorders such as bipolar disorder,1, 2 alcoholism3 and schizophrenia,4 as well as blood biomarker discovery for mood disorders.5 We have now applied this approach (Figures 1 and 2) to blood biomarker discovery efforts for hallucinations and delusions, core symptoms of psychotic disorders. Objective blood biomarkers for illness state and treatment response would make a significant difference in our ability to assess and treat patients with psychotic disorders, eliminating subjectivity and reliance on patient's self-report of symptoms.

Figure 1
figure1

Convergent functional genomics approach for candidate biomarker prioritization. Scoring of independent lines of evidence (maximum score=9 points).

Figure 2
figure2

Top blood candidate biomarker genes for (a) hallucinations and (b) delusions. The CFG lines of evidence scoring are depicted on the right side of the pyramid.

Materials and methods

Human subjects

We present data from four cohorts (Table 1). One cohort consisted of 31 subjects with psychotic disorders (schizophrenia, schizoaffective disorder and substance-induced psychosis), from which the primary biomarker data was derived, from testing done at their first visit (v1). A second cohort consisted of 17 subjects from the first cohort that had a change in psychotic symptom (hallucinations or delusions) Positive and Negative Symptom Scale (PANSS) scores at follow-up testing 3 months (v2) or 6 months (v3) later. A third cohort consisted of 10 new subjects with psychotic disorders, and the fourth cohort consisted of 9 subjects from the third cohort that had a change in symptom scores at follow-up testing 3 months (v2) later.

Table 1 Demographics

Subjects consisted primarily of men (and one woman) over 18 years of age. Subjects were recruited from the patient population at the Indianapolis VA Medical Center and the Indiana University School of Medicine. A demographic breakdown is shown in Table 1. We focused in our initial studies primarily on an age-matched male population, due to the demographics of our catchment area (primarily male in a VA Medical Center), and to minimize any potential gender-related state effects on gene expression, which would have decreased the discriminative power of our analysis given our relatively small sample size. The subjects were recruited largely through referrals from care providers, the use of brochures left in plain sight in public places and mental health clinics, and through word of mouth. Subjects were excluded if they had significant acute medical or neurological illnesses, or had evidence of active substance abuse or dependence. All subjects understood and signed informed consent forms detailing the research goals, procedure, caveats and safeguards. Subjects completed diagnostic assessments by an extensive structured clinical interview—Diagnostic Interview for Genetic Studies—at a baseline visit, followed by up to three testing visits, each three months apart. At each testing visit, they received a psychosis rating scale (PANSS), which includes items that score symptoms of hallucinations and delusions (see Table 2), and blood was drawn. Whole blood (10 ml) was collected in two RNA-stabilizing PAXgene tubes, labeled with an anonymized ID number, and stored at −80 °C in a locked freezer until the time of future processing. Whole blood (predominantly lymphocyte) RNA was extracted for microarray gene expression studies from the PAXgene tubes blood, as detailed below.

Table 2 Hallucinations and Delusions scoring as part of administration of the Positive and Negative Symptom Scale (PANSS)

Human blood gene expression experiments and analysis

RNA extraction

2.5–5 ml of whole blood was collected into each PAXgene tube by routine venipuncture. PAXgene tubes contain proprietary reagents for the stabilization of RNA. The cells from whole blood were concentrated by centrifugation, the pellet washed, resuspended and incubated in buffers containing Proteinase K for protein digestion. A second centrifugation step was done to remove residual cell debris. After the addition of ethanol for an optimal binding condition the lysate was applied to a silica-gel membrane/column. The RNA bound to the membrane as the column was centrifuged and contaminants were removed in three wash steps. The RNA was then eluted using diethylpyrocarbonate-treated water.

Globin reduction

To remove globin messenger RNA (mRNA), total RNA from whole blood was mixed with a biotinylated Capture Oligo Mix that is specific for human globin mRNA. The mixture was then incubated for 15 min to allow the biotinylated oligonucleotides to hybridize with the globin mRNA. Streptavidin magnetic beads were then added, and the mixture was incubated for 30 min. During this incubation, streptavidin binds the biotinylated oligonucleotides, thereby capturing the globin mRNA on the magnetic beads. The streptavidin magnetic beads were then pulled to the side of the tube with a magnet, and the RNA, depleted of the globin mRNA, was transferred to a fresh tube. The treated RNA was further purified using a rapid magnetic bead-based purification method. This consists of adding an RNA binding bead suspension to the samples, and using magnetic capture to wash and elute the GLOBINclear RNA.

Sample labeling

Sample labeling was performed using the Ambion MessageAmp II-BiotinEnhanced amplified RNA (aRNA) amplification (Ambion Inc., Austin, TX, USA). The procedure is briefly outlined below and involves the following steps:

  1. 1

    Reverse transcription to synthesize first strand complementary DNA (cDNA) is primed with the T7 Oligo(dT) Primer to synthesize cDNA containing a T7 promoter sequence.

  2. 2

    Second strand cDNA synthesis converts the single-stranded cDNA into a double-stranded DNA template for transcription. The reaction employs DNA Polymerase and RNase H to simultaneously degrade the RNA and synthesize second strand cDNA.

  3. 3

    cDNA purification removes RNA, primers, enzymes and salts that would inhibit in vitro transcription.

  4. 4

    In vitro transcription to synthesize aRNA with Biotin-NTP Mix generates multiple copies of biotin-modified aRNA from the double-stranded cDNA templates; this is the amplification step.

  5. 5

    aRNA purification removes unincorporated NTPs, salts, enzymes and inorganic phosphate to improve the stability of the biotin-modified aRNA.

Microarrays

Biotin-labeled aRNAs were hybridized to Affymetrix HG-U133 Plus 2.0 GeneChips (Affymetrix, Santa Clara, CA, USA; with over 40 000 genes and expressed sequence tags (ESTs)), according to the manufacturer's protocols. http://www.affymetrix.com/support/technical/manual/expression_manual.affx. Arrays were stained using standard Affymetrix protocols for antibody signal amplification and scanned on an Affymetrix GeneArray 2500 scanner with a target intensity set at 250. Present/absent calls were determined using GCOS software with thresholds set at default values. Quality control measures including 3′/5′ ratios for glyceraldehyde 3-phosphate dehydrogenase and β-actin, scale factors, background and Q values were within acceptable limits.

Analysis

We have used the subject's psychosis scores at time of blood collection, specifically the scores for hallucinations (from 1—no symptoms to 7—extreme symptoms) and the scores for delusions (1–7), obtained from a PANSS scale (Table 2). We looked only at all or nothing gene expression differences that are identified by Absent (A) vs Present (P) Calls in the Affymetrix MAS software. We classified genes whose expression is detected as Absent in the asymptomatic subjects (no hallucinations or no delusions, scores of 1) and detected as Present in the highly symptomatic subjects (high hallucinations or high delusions, scores of 4 and above), as being candidate biomarker genes for high hallucinations or high delusions states, respectively. Conversely, genes whose expression are detected as Present in the asymptomatic subjects and Absent in the highly symptomatic subjects are being classified as candidate biomarker genes for no hallucinations or no delusions states, respectively.

We employed two thresholds for analysis of gene expression differences between no symptoms and high symptoms (Table 3). First we used a high threshold, with at least 75% of subjects in the cohort showing a change in expression from Absent to Present between no symptoms and high symptoms (reflecting an at least threefold psychosis state related enrichment of the genes thus filtered). We also used a low threshold, with at least 60% of subjects in the cohort showing a change in expression from Absent to Present between no symptoms and high symptoms (reflecting an at least 1.5-fold psychosis state related enrichment of the genes thus filtered).

Table 3 High- and low-threshold analyses in the primary psychosis cohort (n=31)

The higher threshold would identify candidate biomarker genes that are more common for all subjects, with a lower risk of false positives, whereas the lower threshold will identify genes that are present in more restricted subgroups of subjects, with a lower risk of false negatives. The high threshold candidate biomarker genes have, as an advantage, a higher degree of reliability, as they are present in at least 75% of all subjects with a certain hallucinations state (high symptoms or no symptoms) tested. They may reflect common aspects related to psychosis across a diverse subject population, but may also be a reflection of the effects of common medications used in the population tested, such as antipsychotics. The low threshold genes may have lower reliability, being present in at least 60% of the subject population tested, but may capture more of the diversity of genes and biological mechanisms present in a genetically diverse human subject population.

Animal model gene expression studies

Our schizophrenia pharmacogenomic model consists of phencyclidine (PCP) and clozapine treatments in mice (see Le-Niculescu et al.4 for experimental details and analysis/categorization of brain gene expression data).

For the current work, we repeated that series of experiments, to obtain blood gene expression data. All experiments were performed with male C57/BL6 mice, 8–12 weeks of age, obtained from Jackson Laboratories (Bar Harbor, ME, USA), and acclimated for at least 2 weeks in our animal facility prior to any experimental manipulation.

Mice were treated by intraperitoneal injection with single-dose saline, PCP (7.5 mg kg−1), clozapine (2.5 mg kg−1), or a combination of PCP and clozapine (7.5 and 2.5 mg kg−1). Three independent de novo biological experiments were performed at different times. Each experiment consisted of three mice per treatment condition, for a total of nine mice per condition across the three experiments.

Mouse blood collection

Twenty-four hours after drug administration, following behavioral testing, the mice were decapitated to harvest blood. The headless mouse body was put over a glass funnel coated with heparin and approximately 1 ml of blood/mouse was collected into a PAXgene blood RNA collection tubes (Qiagen/BD Diagnostics, Valencia, CA, USA). Blood samples from three mice per treatment condition were pooled. The PAXgene blood vials were stored in −4 °C overnight, and then at −80 °C until future processing for RNA extraction.

RNA extraction and microarray work

For the whole mouse blood RNA extraction, PAXgene blood RNA extraction kit (PreAnalytiX, a Qiagen/BD Company) was used, followed by GLOBINclear–Mouse/Rat (Ambion Inc.) to remove the globin mRNA. All the methods and procedures were carried out as per manufacturer's instructions. The quality of the total RNA was confirmed using an Agilent 2100 Bioanalyzer (Agilent Technologies, Palo Alto, CA, USA). The quantity and quality of total RNA was also independently assessed by 260 nm ultraviolet absorption and by 260/280 ratios, respectively with a Nanodrop spectrophotometer (Thermo Scientific, Wilmington, DE, USA). Starting material of total RNA labeling reactions was kept consistent within each independent microarray experiment. Equal amounts of total RNA extracted from pooled blood samples was used for labeling and microarray assays. We used Mouse Genome 430 2.0 arrays (Affymetrix). The GeneChip Mouse Genome 430 2.0 Array contain over 45 000 probe sets that analyze the expression level of transcripts and variants from over 34 000 well-characterized mouse genes. Standard Affymetrix protocols were used to reverse transcribe the mRNA and generate biotinlylate cRNA (http://www.affymetrix.com/support/downloads/manuals/expression_s2_manual.pdf). The amount of cRNA used to prepare the hybridization cocktail was kept constant intra-experiment. Samples were hybridized at 45 °C for 17 h under constant rotation. Arrays were washed and stained using the Affymetrix Fluidics Station 400 and scanned using the Affymetrix Model 3000 Scanner controlled by GCOS software. All sample labeling, hybridization, staining and scanning procedures were carried out as per manufacturer's recommendations. All arrays were scaled to a target intensity of 1000 using Affymetrix MASv 5.0 array analysis software. Quality control measures including 3′/5′ ratios for glyceraldehyde 3-phosphate dehydrogenase and β-actin, scaling factors, background, and Q values were within acceptable limits.

Microarray data analysis

Data analysis was performed using Affymetrix Microarray Suite 5.0 software (MAS v5.0). Default settings were used to define transcripts as present (P), marginal (M) or absent (A). A comparison analysis was performed for each drug treatment, using its corresponding saline treatment as the baseline. ‘Signal,’ ‘Detection,’ ‘Signal Log Ratio,’ ‘Change’ and ‘Change P-value’ were obtained from this analysis. Only transcripts that were called Present in at least one of the two samples (saline or drug) intra-experiment, and that were reproducibly changed in the same direction in at least two out of three independent experiments, were analyzed further.

Cross-validation and integration: CFG

Gene identification

The identities of transcripts were established using NetAffx (Affymetrix), and confirmed by cross-checking the target mRNA sequences that had been used for probe design in the Mouse Genome 430 2.0 Array GeneChip or the Affymetrix Human Genome U133 Plus 2.0 GeneChip with the GenBank database. Where possible, identities of ESTs were established by BLAST searches of the nucleotide database. A National Center for Biotechnology Information (NCBI) (Bethesda, MD, USA) BLAST analysis of the accession number of each probe-set was done to identify each gene name. BLAST analysis identified the closest known gene existing in the database (the highest known gene at the top of the BLAST list of homologues) which then could be used to search the GeneCards database (Weizmann Institute, Rehovot, Israel). Probe sets that did not have a known gene were labeled ‘EST’ and their accession numbers kept as identifiers.

Human postmortem brain convergence

Information about our candidate genes was obtained using GeneCards, the Online Mendelian Inheritance of Man database (http://ncbi.nlm.nih.gov/entrez/query.fcgi?db=OMIM), as well as database searches using PubMed (http://ncbi.nlm.nih.gov/PubMed) and various combinations of keywords (gene name, psychosis, schizophrenia, schizoaffective, human, brain, postmortem). Postmortem convergence was deemed to occur for a gene if there were published reports of human postmortem data showing changes in expression of that gene in brains from patients with psychotic disorders (schizophrenia, schizoaffective d/o). In terms of concordance of direction of change in expression between published postmortem brain data and our human blood data, we made the assumption that schizophrenia postmortem brain data reflected a highly symptomatic phase of the illness. While this may arguably be the case, it is nevertheless an assumption, as no consistent objective data exists regarding the phase of the illness when the subjects deceased, which is one of the limitations of human postmortem brain data to date.

Human genetic data convergence

To designate convergence for a particular gene, the gene had to have published positive reports from candidate gene association studies, or map within 10 cM of a microsatellite marker for which at least one published study showed evidence for genetic linkage to psychotic disorders (schizophrenia or schizoaffective disorder). The University of Southampton's sequence-based integrated map of the human genome (The Genetic Epidemiological Group, Human Genetics Division, University of Southampton: http://cedar.genetics.soton.ac.uk/public_html/) was used to obtain cM locations for both genes and markers. The sex-averaged cM value was calculated and used to determine convergence to a particular marker. For markers that were not present in the Southampton database, the Marshfield database (Center for Medical Genetics, Marshfield, WI, USA: http://research.marshfieldclinic.org/genetics) was used with the NCBI Map Viewer web-site to evaluate linkage convergence.

CFG analysis scoring

Genes were given the maximum score of 2 points if changed in our human blood samples with high threshold analysis, and only 1 point if changed with low threshold (see Figure 1). They received 1 point for each external cross-validating line of evidence: other human tissue data, human genetic data (1 point for assoc., 0.5 point for linkage), animal model brain data, and animal model blood data. Genes received additional bonus points if changed in other human tissue and our blood data, as follows: for brain-2 points if changed in the same direction, 1 point if changed in opposite direction; for lymphoblastoid cell lines and fibroblasts, 1 point if changed in same direction, 0.5 point if changed in opposite directions. Genes also received additional bonus points if changed in brain and blood of the animal model, as follows: 1 point if changed in the same direction in the brain and blood, and 0.5 points if changed in opposite direction. Thus the total maximum CFG score that a candidate biomarker gene can have is 9 (2+4+2+1). As we are interested in discovering blood biomarkers, and because of caveats discussed above, we weighted more heavily our own live subject human blood data (if it made the high threshold cut) than literature-derived human postmortem brain data, human genetic data, or our own animal model data. We also weighted more heavily the human blood–brain concordance than the animal model blood–brain concordance. Other ways of weighing the scores of line of evidence may give slightly different results in terms of prioritization, if not in terms of the list of genes per se. Nevertheless, we feel that this empirical scoring system provides a good separation of genes based on our focus on identifying human blood candidate biomarkers.

Pathway analysis

Ingenuity Pathway Analysis 7.0 (Ingenuity Systems, Redwood City, CA, USA) was used to analyze the biological roles (molecular and cellular functions) categories of the top candidate genes resulting from our CFG analysis.

Results

Hallucinations biomarkers

Using our approach, out of over 40 000 genes and ESTs on the Affymetrix Human Genome U133 Plus 2.0 GeneChip, we have ended up with 50 candidate biomarker genes (Supplementary Table S1) which had a CFG score of 2 or above, meaning either maximal score from the A/P analysis or at least one other line of prior independent evidence for potential involvement in psychotic disorders. Of interest, one of our candidate biomarker genes (Phlda16) had been previously reported to be changed in expression in the same direction, in lymphoblastoid cell lines from schizophrenia subjects. Another one, Adrbk2 (adrenergic receptor kinase, beta 2), also known as Grk3, has been previously reported by us to be decreased at a protein level in lymphoblastoid cell lines from bipolar patients.1

Selecting the top CFG scoring candidate biomarkers for hallucinations (CFG score of 3 and above, meaning, for example, a maximal score from the A/P analysis and at least one other line of prior independent evidence for potential involvement in psychotic disorders), we generated a panel of seven biomarkers for hallucinations (Table 4). To test the predictive value of our panel (to be called the BioM-7 hallucinations panel), we have looked in the cohort of 31 psychotic disorders subjects, containing the 23 subjects (12 no hallucinations, 11 high hallucinations) from which the candidate biomarker data was derived, as well as 8 additional subjects with hallucinations symptoms in the intermediate range (PANSS hallucination scores of 2 or 3). We derived a prediction score for each subject, based on the presence or absence of the biomarkers of the panel in their blood GeneChip data. Each of the biomarkers gets a score of 1 if it is detected as Present (P) in the blood from that subject, 0.5 if it is detected as Marginally Present (M), and 0 if it is called Absent (A). The ratio of the average of the high hallucinations biomarker scores divided by the average of the no hallucinations biomarker scores is multiplied by 100, and provides a prediction score. If the ratio of high hallucinations biomarkers to no hallucinations biomarkers is 1, that is, the two sets of genes are equally represented, the prediction score is 1 × 100=100. The higher this score, the higher the predicted likelihood that the subject will have high hallucinations. We then compared the predictive score with actual PANSS hallucination scores. A prediction score of above 100 had an 80% sensitivity and a 70% specificity for predicting high hallucinations (Table 6).

Table 4 Top candidate biomarker genes for hallucinations prioritized by CFG score for multiple independent lines of evidence

Additionally, we have also conducted human blood gene expression analysis in three other cohorts, subsequently collected. Cohort 2 consisted of 17 subjects from the first cohort that had a change in psychotic symptom (hallucinations and/or delusions) scores at follow-up testing 3 months (v2) or 6 months (v3) later. Cohort 3 consisted of 10 new subjects with psychotic disorders, and Cohort 4 consisted of 9 subjects from Cohort 3 that had a change in symptom scores at follow-up testing 3 months (v2) later.

These cohorts were used as replication cohorts, to verify the predictive power of the hallucinations state biomarker panel identified by analysis of data from the primary psychosis cohort. Overall, the BioM-7 panel had good sensitivity and negative predictive value for high hallucinations state across the different cohorts (Figure 3 and Table 6). Detecting and not missing patients who have high symptom levels is arguably the critical clinical issue, as well as a potential practical application. As such, the sensitivity of the tests for high symptoms (high hallucinations), as well as its negative predictive value, is the most important parameter in that regard.

Figure 3
figure3

Comparison of BioM-7 hallucinations prediction scores and Positive and Negative Symptom Scale (PANSS) hallucinations scores. For hallucinations scores: blue—no hallucinations; red—high hallucinations; white—intermediate hallucinations. Hallucinations scores are based on PANSS scale administered at the time of blood draw. For biomarkers: A (blue)—called Absent by MAS5 analysis; P (red)—called Present by MAS5 analysis; M (yellow)—called Marginally Present by MAS5 analysis. A is scored as 0, M as 0.5 and P as 1. BioM Hallucinations Prediction Score is based on the ratio of the sum of the scores for high mood biomarkers and sum of scores for low mood biomarkers, multiplied by 100. We have used a cutoff score of above 100 for high hallucinations. Infinity—denominator is 0. ND—not determined.

Delusions biomarkers

Using our approach, we have identified 107 candidate biomarker genes (Supplementary Table S2) which had a CFG score of 2 or above, meaning either maximal score from the A/P analysis or at least one other line of prior independent evidence for potential involvement in psychotic disorders.

Selecting the top CFG scoring candidate biomarkers for delusions (CFG score of 3 and above), we generated a panel of 31 biomarkers (Table 5). To test the predictive value of our panel (to be called the BioM-31 delusions panel), we have looked in the cohort of 31 psychotic disorders subjects, containing the 23 subjects (9 no delusions, 13 high delusions) from which the candidate biomarker data was derived, as well as 9 additional subjects with delusions symptoms in the intermediate range (PANSS delusions scores of 2 or 3). We derived a prediction score for each subject, based on the presence or absence of the biomarkers of the panel in their blood GeneChip data. As for hallucinations, each of the biomarkers gets a score of 1 if it is detected as Present (P) in the blood form that subject, 0.5 if it is detected as Marginally Present (M), and 0 if it is called Absent (A). The ratio of the average of the high delusions biomarker scores divided by the average of the no delusions biomarker scores is multiplied by 100, and provides a prediction score. If the ratio of high delusions biomarkers to no delusions biomarkers is 1, that is, the two sets of genes are equally represented, the prediction score is 1 × 100=100. The higher this score, the higher the predicted likelihood that the subject will have high delusions. We then compared the predictive score with actual PANSS delusions scores. A prediction score of above 100 had a 92.3% sensitivity and a 61.1% specificity for predicting high delusions (Figure 4 and Table 6).

Table 5 Top candidate biomarker genes for delusions prioritized by CFG score for multiple independent lines of evidence
Figure 4
figure4figure4

Comparison of BioM-31 delusions prediction scores and actual Positive and Negative Symptom Scale (PANSS) delusions scores. For delusion scores: blue—no delusion; red—high delusion; white—intermediate delusion. Delusion scores are based on PANSS scale administered at the time of blood draw. For biomarkers: A (blue)—called Absent by MAS5 analysis. P (red)—called Present by MAS5 analysis. M (yellow)—called Marginally Present by MAS5 analysis. A is scored as 0, M as 0.5 and P as 1. BioM delusions Prediction Score is based on the ratio of the sum of the scores for high mood biomarkers and sum of scores for low mood biomarkers, multiplied by 100. We have used a cutoff score of above 100 for high delusion. Infinity-denominator is 0. ND—not determined.

Table 6 Psychosis biomarkers panels: sensitivity for predicting high hallucination and high delusion states

Additionally, we also tested our BioM-31 delusions panel in the three other cohorts subsequently collected, used as replication cohorts, to verify the predictive power of the delusions state biomarker panel identified by analysis of data from the primary psychosis cohort. Overall, the BioM-31 panel had good sensitivity and negative predictive value for high delusions state, with the exception of one of the cohorts—Cohort 2 (Table 6). It may be that delusions are more private, diverse and ambiguous to assess by PANSS than hallucinations. If not asked specifically about a particular delusion, a subject may not endorse it. As some of our PANSS testing was done by testers who were not familiar clinically with the subject (that is, different testers had performed the Diagnostic Interview for Genetic Studies in those subjects), that could potentially have contributed to false negatives on the PANSS scoring for delusions, and as a consequence resulted in the apparent lower sensitivity of our test in Cohort 2. Regardless if that was the case or not, the reluctance of patients to report psychiatric symptoms underscores the necessity of developing objective tests such as the blood biomarker ones described in this paper, and the need to validate them in multiple cohorts.

Discussion

Strengths and limitations of our work

As a way of identifying biomarkers, we initially conducted gene expression profiling studies in peripheral whole blood from a primary cohort of 31 human subjects with psychotic disorders (30 males, 1 female) (see Table 1). We measured their psychological testing (PANSS) assessed hallucinations scores, respectively delusions scores (on a scale of 1 to 7) at the time of blood collection. We then looked at gene expression differences between the no symptoms of hallucinations, respectively delusions vs high symptoms of hallucinations, respectively delusions, groups. As in our previous work to identify mood biomarkers,5 we have used an all or nothing Absent (A) vs. Present (P) Calls in the Affymetrix MAS software.

Given the genetic heterogeneity and variable environmental exposure, it is possible, indeed likely, that not all subjects will show changes in all the biomarker genes. Hence we have used two stringency thresholds: changes in 75% of subjects, and in 60% of subjects with no symptoms vs high symptoms. Moreover, our approach, as described above, is predicated on the existence of multiple cross-validators for each gene that is called a candidate biomarker (Figure 1): (1) is it changed in human blood, (2) is it changed in animal model brain, (3) is it changed in animal model blood, (4) is it changed in postmortem human brain, and (5) is there any human genetic data (linkage, association) implicating the gene in psychosis. All these lines of evidence are the result of independent experiments. The virtues of this networked Bayesian approach are that, if one or another of the nodes (lines of evidence) becomes questionable/non-functional upon further evidence in the field, the network is resilient and maintains functionality. The prioritization of candidates is similar conceptually to the Google PageRank algorithm7—the more links (lines of evidence) to a candidate, the higher it will come up on our priority list. As more evidence emerges in the field for some of these genes, they will move up in the prioritization scoring.8 Using such an approach, we were able to identify and focus on a small number of genes as likely candidate biomarkers, out of the over 40 000 transcripts (about half of which are detected as Present in each subject) measured by the microarrays we used.

By cross-validating with other human datasets and with animal model data using CFG (Figure 1), we were able to extract a shorter list of genes for which there are external corroborating line of evidence (human genetic evidence, human postmortem brain data, animal model brain and blood data) linking them to psychotic disorders, thus reducing the risk of false positives. This cross-validation identifies candidate biomarkers that are more likely directly related to the relevant disease neuropathology, as opposed to being potential artifactual effects related to a particular cohort or indirect effects of lifestyle and environment. The power of our CFG approach is exemplified in the fact that our biomarker panels had good predictive power in independent cohorts, a key litmus test in our view, and one that needs to be applied more systematically in this nascent field.

All subjects recruited were on prior prescribed medications. We cannot exclude, and in fact would anticipate that medications may have an effect on biomarker expression levels. However, of note, the patients were on a very diverse list of antipsychotics, mood stabilizers, and other psychotropic medications (Supplementary Table S4). While that makes pinpointing a particular medication effect not feasible with our current design (clinical trials with specific medications are a better setting for identifying such effects), it is re-assuring that we are obtaining with our CFG approach consistent findings that show predictive power in independent cohorts, despite this diversity of medications and of a variety of other environmental effects.

Clozapine, modeled in the pharmacogenomic animal model work, is a broad-spectrum drug, one of the current gold standards, and encompasses many of the actions of some of the other antipsychotics currently used in schizophrenia. The premise of using it, along with PCP, in a pharmacogenomic animal model of schizophrenia,4 was that they may modulate the expression of genes involved in the pathogenesis of schizophrenia. The findings in that model, cross-validated with other independent approaches and lines of evidence, support its validity.4 Comparisons with the non-medicated normal control group will in the future permit additional distinctions regarding medication effects, as will systematic large-scale within-subject comparisons of subjects whose medications remain constant but symptoms state and markers change from one visit to the next.

Moreover, psychosis state and blood gene expression changes may be influenced not only by the presence or absence of medications, but also of drugs of abuse. While we had access to the subject's medical records and clinical information as part of the informed consent for the study, medication non-compliance, on the one hand, and substance abuse, on the other hand, are not infrequent occurrences in psychiatric patients.

More extensive follow-up studies may benefit from the prospective inclusion of toxicology and medication levels testing. That medications and drugs of abuse may have effects on psychosis state and gene expression is in fact being partially modeled in the mouse pharmacogenomic model, with clozapine and PCP treatments respectively. In the end, it is the association of blood biomarkers with psychosis state that has been the primary goal of the work reported in this paper, regardless of the proximal causes, which could be diverse and will need to be the subject of subsequent hypothesis-driven studies beyond the scope of this initial work.

Our sample size for human subjects (n=31 for the primary cohort; n=17, n=10, n=9 for the other three cohorts) is relatively small, but comparable to the size of cohorts for human postmortem brain gene expression studies.9, 10, 11 We have in essence studied live donor blood samples instead of postmortem donor brains, with the advantage of better phenotypic characterization, more quantitative state information, and less technical variability. Our approach also permits repeated intra-subject measures when the subject is in different psychosis states, which is an area of future interest and work. In fact, two of our psychosis cohorts are composed of a subset of subjects from the primary and secondary psychosis cohorts, that displayed a different psychosis state (no symptoms vs. intermediate vs. high symptoms) when they were re-tested at a later time point, 3 or 6 months later. Overall, our design was geared towards validating state biomarkers for psychosis while minimizing the noise of genetic and environmental background differences. For trait biomarkers, larger population studies and comparisons with normal controls may be needed. Of note, we have studied almost exclusively male subjects, which means our results might be male-specific. Future studies looking at potential gender differences are warranted.

Overall, our approach of: (1) using individual phenes12 reflecting internal subjective experiences (hallucinations or delusions), which are the hallmark of psychosis (as opposed to more complex and disease specific state/trait clinical instruments or DSM categorical diagnosis); (2) looking at extremes of state; combined with (3) robust differential expression based on A/P calls, and (4) CFG prioritization, seems to be able to identify state biomarkers for psychosis that may be, at least in part, generalizable to independent cohorts.

In the work reported here, similar to our previously published mood biomarker work,5 we decided to focus on using CFG scoring as a cut-off to decide which biomarkers to include in panels, rather than find best panel sizes by fit-to-data and receiver operating characteristic curves. We reasoned that an objective CFG scoring cut-off would pick up signal relevant to illness and increase generalizability of our panels across independent cohorts, while a fit-to-data receiver operating characteristic approach, while it might achieve excellent results in the primary cohort, driven at least in part by the noise particular to that cohort, would have poorer results in independent cohorts. In fact, CFG prioritization has been shown to lead to generalizabilty across cohorts not only in our previous5 and current biomarker work, but also when we applied it to genome-wide association studies data,13 where P-value criteria are the equivalent of fit-to-data analyses.

While it appears that panels of biomarkers chosen by CFG scoring criteria are the way to go due to population heterogeneity and impact of environmental factors on gene expression, it remains an open empirical question for future work as to how large the panels should be, and whether it may be possible to identify particular single biomarkers that have almost as good a predictive power as that of a larger panel. Ongoing studies are also examining the issue of using incremental differential expression comparisons as opposed to all or nothing A/P calls to identify biomarkers, and are expected to yield an expanded repertoire of biomarkers.

Finally, some of the top candidate biomarker genes identified by our human blood work reported here have no previous evidence for involvement in psychotic disorders other than our mapping them to schizophrenia genetic linkage loci (Supplementary Tables S1 and S2), and thus constitute novel candidate genes for schizophrenia and related disorders. They merit further exploration in genetic candidate gene association studies, as well as comparison with emerging results from whole-genome association studies of schizophrenia and related disorders. Moreover, as more evidence accumulates in the field, all grist for the mill for our CFG approach, and as prospective studies are done, it is possible that the composition of top biomarker panels for hallucinations and for delusions will be refined or changed for different sub-populations. That being said, it is likely that a large number of the biomarkers that would be of use in different panels and permutations are already present in our lists of candidate biomarker genes (n=50 for hallucinations—Supplementary Table S1; n=107 for delusions—Supplementary Table S2).

Hallucinations and delusions: similarities and differences

There are more genes with high CFG scores for delusions than for hallucinations, reflecting the fact that more prior evidence exists for them in terms of involvement in schizophrenia and related disorders, and perhaps there is a higher degree of diversity in the genetic architecture of delusions, a more evolved cognitive phene, compared to that of hallucinations, a more primitive sensory phene. As a consequence, using the same CFG cut-off, the panel size for delusions was larger than that for hallucinations. Of note, there is co-directional overlap between the candidate biomarkers for delusions (Supplementary Table S2) and hallucinations (Supplementary Table S1) identified by us, which is reassuring in terms of the technical reliability of our assessments and findings, as these symptoms are often co-morbid clinically. More interestingly, there is some overlap between candidate biomarkers for hallucinations, delusions and mood state previously identified by us5 (Supplementary Figure S1), with the mood markers being generally contra-directional to the psychosis markers. Taken together with the heterogeneity of biomarker expression seen in patients that have a similar psychiatric diagnosis (Figures 3 and 4), our work is consistent with an emerging Lego-like model of complexity, heterogeneity, overlap and interdependence of major psychiatric disorders.4, 14 Practical implications and predictions of this view would be the relative lack of specificity of single genes and biomarkers for a particular disorder, and the need to use profiling with panels of markers to achieve some disease specificity.

From biomarkers to biology

Remarkably, among our candidate blood biomarker genes for delusions (Table 5) are key genes with extensive evidence in brain pathophysiology in psychotic disorders (dopamine receptor 2Drd2,15 neuroregulin 1Nrg116, 17) and neurodegenerative disorders (apolipoprotein EApoE). A polymorphism in Drd2 was reported to be associated specifically with delusions and disorganization symptomatology in major psychoses.18 Of interest, delusion symptoms were reported to be associated with ApoE epsilon4 allelic variant in late-onset Alzheimer's disease.19 Moreover, plasma ApoE has been reported to be significantly decreased in treatment-free subjects with schizophrenia spectrum disorders and bipolar disorder,20 consistent with our findings of ApoE being decreased in expression in high delusion states. Recently, variations in levels of expression of ApoE have also been tied by us to the risk and progression of Alzheimer's disease (AD) irrespective of ε4 status.21 Overall, the ApoE connection warrants future empirical work as a possible molecular underpinning of the Kraepelinian view of schizophrenia as dementia praecox.

At the top of our list of candidate biomarker genes for hallucinations (Table 4), we have four genes decreased in expression in high hallucinations states (Rhobtb3, Aldh1l1, Mpp3, Fn1), and three genes increased in high hallucinations states (Arhgef9, Phlda1, S100a6). Although all of these genes have prior evidence of differential expression in schizophrenia patients, they are less well known than the candidate biomarker genes for delusions discussed above. A non-obvious top candidate biomarker for hallucinations, increased in high hallucinations state, is Arhgef9 (Cdc42 guanine nucleotide exchange factor 9, also known as collybistin). Arhgef9 can regulate actin cytoskeletal dynamics and may also modulate GABAergic neurotransmission through binding of a scaffolding protein, gephyrin, at the synapse.22 Interestingly, it has also been implicated in X-linked mental retardation with sensory hyperarousal.23 Aldh1l1, another non-obvious candidate, is a folate metabolic enzyme with antiproliferative effects, expressed in astrocytes.24

Fn1 (Fibronectin 1), one of our top scoring candidate biomarkers for hallucinations and for delusions (Figure 2), is decreased in high hallucinations states and high delusions states, was also previously reported to be decreased in fibroblasts from schizophrenia patients.25, 26 It has also been identified as a top candidate gene for alcoholism in previous work from our group.3 This raises interesting issues about the psychosis-modulating properties of alcohol, specifically hallucinations and delusions symptoms in alcoholism, as well as the more general issue of clinical co-morbidity between schizophrenia and alcoholism.

Overall, the top candidate biomarker genes results discussed above and the results of a biological functions analyses (Tables 6 and 7) suggest that genes involved in cancer, plasticity and connectivity (cell morphology, cell-to-cell signaling and interaction) are prominent players in psychotic disorders, and are reflected in the blood profile, consistent with previous work in the field implicating developmental and connectivity mechanisms in schizophrenia.4, 27, 28 Unlike for our mood biomarker work,29 we did not find myelin genes prominently represented among our top psychosis biomarkers. Interestingly, the top canonical pathways for both hallucinations and delusions had to do with interleukin signaling, consistent with previous work in the field implicating immune and inflammatory mechanisms in schizophrenia pathophysiology.30 For example, IL-8 signaling, which was identified as the top canonical pathway in hallucinations, has been previously implicated as a maternal risk factor for schizophrenia in the offsprings,31 and IL-8 levels have been reported to be elevated in neuroleptic-free schizophrenia patients compared to normal controls.32

Table 7 Biological roles

The model that is emergent is that of increased plasticity and decreased connectivity4 in high psychosis states compared to no psychosis states. This perspective is of speculative evolutionary interest and pragmatic clinical importance. Speculatively, nature may have selected primitive cellular mechanisms involved in the response to damage, insults and stressors for analogous higher organism level-functions (Figure 5). In this view, psychosis is the higher organismal/brain equivalent of cellular de-differentiation and disconnection such as occurs in early stages of inflammation33, tissue re-modeling34 and cancer metastasis.35 Specifically, the decrease in FN1 expression and increase in NRG1 expression in high delusions states, as well as decrease in fibronectin expression and increase in calcyclin (S100A6) in high hallucination states, are consistent with increased metastatic potential, though not necessarily increased tumorigenesis/cellular proliferation. Indeed, there seems to be a decrease incidence of respiratory cancers in schizophrenia patients, despite the high incidence of smoking in that population. Pragmatically, the psychotic episodes may be correlated with metastasis in cancers.36 Typical antipsychotic medications may have protective effects against cancer,37 consistent also with our connectivity map results identifying fluphenazine as having an opposite gene expression profile to that of high delusions (Table 8). Lastly, the involvement of interleukin signaling canonical pathways suggests that anti-inflammatory and immune-modulating medications should to be more systematically evaluated for prevention and early intervention in psychotic disorders, consistent with some emerging clinical data.38, 39 In particular, omega-3 fatty acids may have a favorable effects to side-effects ratio and multiple whole-body health benefits in this patient population.40

Figure 5
figure5

Psychosis: disconnection and de-differentiation.

Table 8 Connectivity map interrogation of drugs that have similar gene expression signatures to that of (A) high hallucinations and (B) high delusions

Conclusions

We propose, and provide proof of principle for, a translational convergent approach to help identify and prioritize blood biomarkers for psychosis states, specifically for hallucinations and for delusions. A validation of our approach is the fact that our primary cohort-derived biomarker panels showed not only good sensitivity and specificity in the primary cohort, but also predictive ability in three other cohorts. Finally, a data-derived model for whole-body biological mechanisms associated with psychosis is proposed.

Biomarker-based tests may help with early detection, intervention and prevention efforts in schizophrenia41, 42 and related disorders,43 as well as monitoring response to various treatments. In conjunction with other clinical information, such tests may come to play an important part in personalizing treatment to increase precision, effectiveness and avoid adverse reactions. Last but not least, new drug development efforts would particularly benefit from the use of such markers.

References

  1. 1

    Niculescu A, Segal D, Kuczenski R, Barrett T, Hauger R, Kelsoe J . Identifying a series of candidate genes for mania and psychosis: a convergent functional genomics approach. Physiol Genomics 2000; 4: 83–91.

  2. 2

    Ogden CA, Rich ME, Schork NJ, Paulus MP, Geyer MA, Lohr JB et al. Candidate genes, pathways and mechanisms for bipolar (manic-depressive) and related disorders: an expanded convergent functional genomics approach. Mol Psychiatry 2004; 9: 1007–1029.

  3. 3

    Rodd ZA, Bertsch BA, Strother WN, Le-Niculescu H, Balaraman Y, Hayden E et al. Candidate genes, pathways and mechanisms for alcoholism: an expanded convergent functional genomics approach. Pharmacogenomics J 2007; 7: 222–256.

  4. 4

    Le-Niculescu H, Balaraman Y, Patel S, Tan J, Sidhu K, Jerome RE et al. Towards understanding the schizophrenia code: an expanded convergent functional genomics approach. Am J Med Genet B Neuropsychiatr Genet 2007; 144B: 129–158.

  5. 5

    Le-Niculescu H, Kurian SM, Yehyawi N, Dike C, Patel SD, Edenberg HJ et al. Identifying blood biomarkers for mood disorders using convergent functional genomics. Mol Psychiatry 2009; 14: 156–174.

  6. 6

    Middleton FA, Pato CN, Gentile KL, McGann L, Brown AM, Trauzzi M et al. Gene expression analysis of peripheral blood leukocytes from discordant sib-pairs with schizophrenia and bipolar disorder reveals points of convergence between genetic and functional genomic approaches. Am J Med Genet B Neuropsychiatr Genet 2005; 136: 12–25.

  7. 7

    Morrison JL, Breitling R, Higham DJ, Gilbert DR . GeneRank: using search engine technology for the analysis of microarray experiments. BMC Bioinformatics 2005; 6: 233.

  8. 8

    Le-Niculescu H, McFarland MJ, Mamidipalli S, Ogden CA, Kuczenski R, Kurian SM et al. Convergent functional genomics of bipolar disorder: from animal model pharmacogenomics to human genetics and biomarkers. Neurosci Biobehav Rev 2007; 31: 897–903.

  9. 9

    Vawter MP, Crook JM, Hyde TM, Kleinman JE, Weinberger DR, Becker KG et al. Microarray analysis of gene expression in the prefrontal cortex in schizophrenia: a preliminary study. Schizophr Res 2002; 58: 11–20.

  10. 10

    Choudary PV, Molnar M, Evans SJ, Tomita H, Li JZ, Vawter MP et al. Altered cortical glutamatergic and GABAergic signal transmission with glial involvement in depression. Proc Natl Acad Sci USA 2005; 102: 15653–15658.

  11. 11

    Vawter MP, Tomita H, Meng F, Bolstad B, Li J, Evans S et al. Mitochondrial-related gene expression changes are sensitive to agonal-pH state: implications for brain disorders. Mol Psychiatry 2006; 11, 615, 663–679.

  12. 12

    Niculescu AB, Lulow LL, Ogden CA, Le-Niculescu H, Salomon DR, Schork NJ et al. PhenoChipping of psychotic disorders: a novel approach for deconstructing and quantitating psychiatric phenotypes. Am J Med Genet B Neuropsychiatr Genet 2006; 141: 653–662.

  13. 13

    Le-Niculescu H, Patel SD, Bhat M, Kuczenski R, Faraone SV, Tsuang MT et al. Convergent functional genomics of genome-wide association data for bipolar disorder: comprehensive identification of candidate genes, pathways and mechanisms. Am J Med Genet B Neuropsychiatr Genet 2009; 150B: 155–181.

  14. 14

    Niculescu 3rd AB . Polypharmacy in oligopopulations: what psychiatric genetics can teach biological psychiatry. Psychiatr Genet 2006; 16: 241–244.

  15. 15

    Glatt SJ, Faraone SV, Lasky-Su JA, Kanazawa T, Hwu HG, Tsuang MT . Family-based association testing strongly implicates DRD2 as a risk gene for schizophrenia in Han Chinese from Taiwan. Mol Psychiatry 2008; 14: 885–893.

  16. 16

    Georgieva L, Dimitrova A, Ivanov D, Nikolov I, Williams NM, Grozeva D et al. Support for neuregulin 1 as a susceptibility gene for bipolar disorder and schizophrenia. Biol Psychiatry 2008; 64: 419–427.

  17. 17

    Goes FS, Willour VL, Zandi PP, Belmonte PL, Mackinnon DF, Mondimore FM et al. Family-based association study of Neuregulin 1 with psychotic bipolar disorder. Am J Med Genet B Neuropsychiatr Genet 2009; 150B: 693–702.

  18. 18

    Serretti A, Lattuada E, Lorenzi C, Lilli R, Smeraldi E . Dopamine receptor D2 Ser/Cys 311 variant is associated with delusion and disorganization symptomatology in major psychoses. Mol Psychiatry 2000; 5: 270–274.

  19. 19

    Spalletta G, Bernardini S, Bellincampi L, Federici G, Trequattrini A, Caltagirone C . Delusion symptoms are associated with ApoE epsilon4 allelic variant at the early stage of Alzheimer's disease with late onset. Eur J Neurol 2006; 13: 176–182.

  20. 20

    Dean B, Digney A, Sundram S, Thomas E, Scarr E . Plasma apolipoprotein E is decreased in schizophrenia spectrum and bipolar disorder. Psychiatry Res 2008; 158: 75–78.

  21. 21

    Maloney B, Ge YW, Petersen RC, Hardy J, Rogers JT, Perez-Tur J et al. Functional characterization of three single-nucleotide polymorphisms present in the human APOE promoter sequence: differential effects in neuronal cells and on DNA-protein interactions. Am J Med Genet B Neuropsychiatr Genet, (2011) 16, –;, 5 June 2009; e-pub ahead of print.

  22. 22

    Papadopoulos T, Korte M, Eulenburg V, Kubota H, Retiounskaia M, Harvey RJ et al. Impaired GABAergic transmission and altered hippocampal synaptic plasticity in collybistin-deficient mice. Embo J 2007; 26: 3888–3899.

  23. 23

    Marco EJ, Abidi FE, Bristow J, Dean WB, Cotter P, Jeremy RJ et al. ARHGEF9 disruption in a female patient is associated with X linked mental retardation and sensory hyperarousal. J Med Genet 2008; 45: 100–105.

  24. 24

    Anthony TE, Heintz N . The folate metabolic enzyme ALDH1L1 is restricted to the midline of the early CNS, suggesting a role in human neural tube defects. J Comp Neurol 2007; 500: 368–383.

  25. 25

    Mahadik SP, Mukherjee S, Wakade CG, Laev H, Reddy RR, Schnur DB . Decreased adhesiveness and altered cellular distribution of fibronectin in fibroblasts from schizophrenic patients. Psychiatry Res 1994; 53: 87–97.

  26. 26

    Miyamae Y, Nakamura Y, Kashiwagi Y, Tanaka T, Kudo T, Takeda M . Altered adhesion efficiency and fibronectin content in fibroblasts from schizophrenic patients. Psychiatry Clin Neurosci 1998; 52: 345–352.

  27. 27

    Bassett DS, Bullmore E, Verchinski BA, Mattay VS, Weinberger DR, Meyer-Lindenberg A . Hierarchical organization of human cortical networks in health and schizophrenia. J Neurosci 2008; 28: 9239–9248.

  28. 28

    Sun D, Stuart GW, Jenkinson M, Wood SJ, McGorry PD, Velakoulis D et al. Brain surface contraction mapped in first-episode schizophrenia: a longitudinal magnetic resonance imaging study. Mol Psychiatry 2008; 14: 976–986.

  29. 29

    Le-Niculescu H, Kurian SM, Yehyawi N, Dike C, Patel SD, Edenberg HJ et al. Identifying blood biomarkers for mood disorders using convergent functional genomics. Mol Psychiatry 2008; 14: 156–174.

  30. 30

    Shirts BH, Wood J, Yolken RH, Nimgaonkar VL . Comprehensive evaluation of positional candidates in the IL-18 pathway reveals suggestive associations with schizophrenia and herpes virus seropositivity. Am J Med Genet B Neuropsychiatr Genet 2008; 147: 343–350.

  31. 31

    Brown AS, Hooton J, Schaefer CA, Zhang H, Petkova E, Babulas V et al. Elevated maternal interleukin-8 levels and risk of schizophrenia in adult offspring. Am J Psychiatry 2004; 161: 889–895.

  32. 32

    Zhang XY, Zhou DF, Zhang PY, Wu GY, Cao LY, Shen YC . Elevated interleukin-2, interleukin-6 and interleukin-8 serum levels in neuroleptic-free schizophrenia: association with psychopathology. Schizophrenia Res 2002; 57: 247–258.

  33. 33

    Garcia-Bueno B, Caso JR, Leza JC . Stress as a neuroinflammatory condition in brain: damaging and protective mechanisms. Neurosci Biobehav Rev 2008; 32: 1136–1151.

  34. 34

    Lau CG, Zukin RS . NMDA receptor trafficking in synaptic plasticity and neuropsychiatric disorders. Nat Rev 2007; 8: 413–426.

  35. 35

    Kanakry CG, Li Z, Nakai Y, Sei Y, Weinberger DR . Neuregulin-1 regulates cell adhesion via an ErbB2/phosphoinositide-3 kinase/Akt-dependent pathway: potential implications for schizophrenia and cancer. PLoS ONE 2007; 2: e1369.

  36. 36

    Schonfeldt-Lecuona C, Freudenmann RW, Tumani H, Kassubek J, Connemann BJ . Acute psychosis with a mediastinal carcinoma metastasis. Med Sci Monit 2005; 11: CS6–CS8.

  37. 37

    Wei Z, Qi J, Dai Y, Bowen WD, Mousseau DD . Haloperidol disrupts Akt signalling to reveal a phosphorylation-dependent regulation of pro-apoptotic Bcl-XS function. Cell Signalling 2009; 21: 161–168.

  38. 38

    Riedel M, Strassnig M, Schwarz MJ, Muller N . COX-2 inhibitors as adjunctive therapy in schizophrenia: rationale for use and evidence to date. CNS Drugs 2005; 19: 805–819.

  39. 39

    Laan W, Smeets H, de Wit NJ, Kahn RS, Grobbee DE, Burger H . Glucocorticosteroids associated with a decreased risk of psychosis. J Clin Psychopharmacol 2009; 29: 288–290.

  40. 40

    Peet M . Omega-3 polyunsaturated fatty acids in the treatment of schizophrenia. Israel J Psychiatry Relat Sci 2008; 45: 19–25.

  41. 41

    Huang JT, Wang L, Prabakaran S, Wengenroth M, Lockstone HE, Koethe D et al. Independent protein-profiling studies show a decrease in apolipoprotein A1 levels in schizophrenia CSF, brain and peripheral tissues. Mol Psychiatry 2008; 13: 1118–1128.

  42. 42

    Sawa A, Cascella NG . Peripheral olfactory system for clinical and basic psychiatry: a promising entry point to the mystery of brain mechanism and biomarker identification in schizophrenia. Am J Psychiatry 2009; 166: 137–139.

  43. 43

    Kato T, Iwayama Y, Kakiuchi C, Iwamoto K, Yamada K, Minabe Y et al. Gene expression and association analyses of LIM (PDLIM5) in bipolar disorder and schizophrenia. Mol Psychiatry 2005; 10: 1045–1055.

  44. 44

    Le-Niculescu H, Balaraman Y, Patel S, Tan J, Sidhu K, Jerome RE et al. Towards understanding the schizophrenia code: an expanded convergent functional genomics approach. Am J Med Genet B Neuropsychiatr Genet 2007; 144: 129–158.

  45. 45

    Paunio T, Tuulio-Henriksson A, Hiekkalinna T, Perola M, Varilo T, Partonen T et al. Search for cognitive trait components of schizophrenia reveals a locus for verbal learning and memory on 4q and for visual working memory on 2q. Hum Mol Genet 2004; 13: 1693–1702.

  46. 46

    Kim S, Choi KH, Baykiz AF, Gershenfeld HK . Suicide candidate genes associated with bipolar disorder and schizophrenia: an exploratory gene expression profiling analysis of post-mortem prefrontal cortex. BMC Genomics 2007; 8: 413.

  47. 47

    Glatt SJ, Everall IP, Kremen WS, Corbeil J, Sasik R, Khanlou N et al. Comparative gene expression analysis of blood and brain provides concurrent validation of SELENBP1 up-regulation in schizophrenia. Proc Natl Acad Sci USA 2005; 102: 15533–15538.

  48. 48

    Vawter MP, Ferran E, Galke B, Cooper K, Bunney WE, Byerley W . Microarray screening of lymphocyte gene expression differences in a multiplex schizophrenia pedigree. Schizophr Res 2004; 67: 41–52.

  49. 49

    Bowden NA, Weidenhofer J, Scott RJ, Schall U, Todd J, Michie PT et al. Preliminary investigation of gene expression profiles in peripheral blood lymphocytes in schizophrenia. Schizophrenia Res 2006; 82: 175–183.

  50. 50

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

  51. 51

    Dean B, Pavey G, Scarr E, Goeringer K, Copolov DL . Measurement of dopamine D2-like receptors in postmortem CNS and pituitary: differential regional changes in schizophrenia. Life Sci 2004; 74: 3115–3131.

  52. 52

    Seeman P, Guan HC, Nobrega J, Jiwa D, Markstein R, Balk JH et al. Dopamine D2-like sites in schizophrenia, but not in Alzheimer's, Huntington's, or control brains, for [3 H]benzquinoline. Synapse 1997; 25: 137–146.

  53. 53

    Torrey EF, Barci BM, Webster MJ, Bartko JJ, Meador-Woodruff JH, Knable MB . Neurochemical markers for schizophrenia, bipolar disorder, and major depression in postmortem brains. Biol Psychiatry 2005; 57: 252–260.

  54. 54

    Zvara A, Szekeres G, Janka Z, Kelemen JZ, Cimmer C, Santha M et al. Over-expression of dopamine D2 receptor and inwardly rectifying potassium channel genes in drug-naive schizophrenic peripheral blood lymphocytes as potential diagnostic markers. Dis Markers 2005; 21: 61–69.

  55. 55

    Sun J, Kuo PH, Riley BP, Kendler KS, Zhao Z . Candidate genes for schizophrenia: a survey of association studies and gene ranking. Am J Med Genet B Neuropsychiatr Genet 2008; 147B: 1173–1181.

  56. 56

    Allen NC, Bagade S, McQueen MB, Ioannidis JP, Kavvoura FK, Khoury MJ et al. Systematic meta-analyses and field synopsis of genetic association studies in schizophrenia: the SzGene database. Nat Genet 2008; 40: 827–834.

  57. 57

    Yamada K, Gerber DJ, Iwayama Y, Ohnishi T, Ohba H, Toyota T et al. Genetic analysis of the calcineurin pathway identifies members of the EGR gene family, specifically EGR3, as potential susceptibility candidates in schizophrenia. Proc Natl Acad Sci USA 2007; 104: 2815–2820.

  58. 58

    Straub RE, MacLean CJ, O'Neill FA, Walsh D, Kendler KS . Support for a possible schizophrenia vulnerability locus in region 5q22-31 in Irish families. Mol Psychiatry 1997; 2: 148–155.

  59. 59

    Devlin B, Bacanu SA, Roeder K, Reimherr F, Wender P, Galke B et al. Genome-wide multipoint linkage analyses of multiplex schizophrenia pedigrees from the oceanic nation of Palau. Mol Psychiatry 2002; 7: 689–694.

  60. 60

    Smalla KH, Mikhaylova M, Sahin J, Bernstein HG, Bogerts B, Schmitt A et al. A comparison of the synaptic proteome in human chronic schizophrenia and rat ketamine psychosis suggest that prohibitin is involved in the synaptic pathology of schizophrenia. Mol Psychiatry 2008; 13: 878–896.

  61. 61

    Faraone SV, Lasky-Su J, Glatt SJ, Van Eerdewegh P, Tsuang MT . Early onset bipolar disorder: possible linkage to chromosome 9q34. Bipolar Disord 2006; 8: 144–151.

  62. 62

    Kampman O, Anttila S, Illi A, Mattila KM, Rontu R, Leinonen E et al. Apolipoprotein E polymorphism is associated with age of onset in schizophrenia. J Hum Genet 2004; 49: 355–359.

  63. 63

    Hahn CG, Wang HY, Cho DS, Talbot K, Gur RE, Berrettini WH et al. Altered neuregulin 1-erbB4 signaling contributes to NMDA receptor hypofunction in schizophrenia. Nat Med 2006; 12: 824–828.

  64. 64

    Petryshen TL, Middleton FA, Kirby A, Aldinger KA, Purcell S, Tahl AR et al. Support for involvement of neuregulin 1 in schizophrenia pathophysiology. Mol Psychiatry 2005; 10: 366–374, 328.

  65. 65

    Chagnon YC, Roy MA, Bureau A, Merette C, Maziade M . Differential RNA expression between schizophrenic patients and controls of the dystrobrevin binding protein 1 and neuregulin 1 genes in immortalized lymphocytes. Schizophrenia Res 2008; 100: 281–290.

  66. 66

    Zhang HX, Zhao JP, Lv LX, Li WQ, Xu L, Ouyang X et al. Explorative study on the expression of neuregulin-1 gene in peripheral blood of schizophrenia. Neurosci Lett 2008; 438: 1–5.

  67. 67

    Blouin JL, Dombroski BA, Nath SK, Lasseter VK, Wolyniec PS, Nestadt G et al. Schizophrenia susceptibility loci on chromosomes 13q32 and 8p21. Nat Genet 1998; 20: 70–73.

  68. 68

    Chiu YF, McGrath JA, Thornquist MH, Wolyniec PS, Nestadt G, Swartz KL et al. Genetic heterogeneity in schizophrenia II: conditional analyses of affected schizophrenia sibling pairs provide evidence for an interaction between markers on chromosome 8p and 14q. Mol Psychiatry 2002; 7: 658–664.

  69. 69

    Gurling HM, Kalsi G, Brynjolfson J, Sigmundsson T, Sherrington R, Mankoo BS et al. Genomewide genetic linkage analysis confirms the presence of susceptibility loci for schizophrenia, on chromosomes 1q32.2, 5q33.2, and 8p21-22 and provides support for linkage to schizophrenia, on chromosomes 11q23.3-24 and 20q12.1-11.23. Am J Hum Genet 2001; 68: 661–673.

  70. 70

    Pulver AE, Mulle J, Nestadt G, Swartz KL, Blouin JL, Dombroski B et al. Genetic heterogeneity in schizophrenia: stratification of genome scan data using co-segregating related phenotypes. Mol Psychiatry 2000; 5: 650–653.

  71. 71

    Suarez BK, Duan J, Sanders AR, Hinrichs AL, Jin CH, Hou C et al. Genomewide linkage scan of 409 European-Ancestry and African American families with schizophrenia: suggestive evidence of linkage at 8p23.3-p21.2 and 11p13.1-q14.1 in the combined sample. Am J Hum Genet 2006; 78: 315–333.

  72. 72

    Kaufmann CA, Suarez B, Malaspina D, Pepple J, Svrakic D, Markel PD et al. NIMH Genetics Initiative Millenium Schizophrenia Consortium: linkage analysis of African-American pedigrees. Am J Med Genet 1998; 81: 282–289.

  73. 73

    Arion D, Unger T, Lewis DA, Levitt P, Mirnics K . Molecular evidence for increased expression of genes related to immune and chaperone function in the prefrontal cortex in schizophrenia. Biol Psychiatry 2007; 62: 711–721.

  74. 74

    DeLisi LE, Shaw SH, Crow TJ, Shields G, Smith AB, Larach VW et al. A genome-wide scan for linkage to chromosomal regions in 382 sibling pairs with schizophrenia or schizoaffective disorder. Am J Psychiatry 2002; 159: 803–812.

  75. 75

    Straub RE, MacLean CJ, Ma Y, Webb BT, Myakishev MV, Harris-Kerr C et al. Genome-wide scans of three independent sets of 90 Irish multiplex schizophrenia families and follow-up of selected regions in all families provides evidence for multiple susceptibility genes. Mol Psychiatry 2002; 7: 542–559.

  76. 76

    Clark D, Dedova I, Cordwell S, Matsumoto I . A proteome analysis of the anterior cingulate cortex gray matter in schizophrenia. Mol Psychiatry 2006; 11: 459–470, 423.

  77. 77

    Takahashi S, Faraone SV, Lasky-Su J, Tsuang MT . Genome-wide scan of homogeneous subtypes of NIMH genetics initiative schizophrenia families. Psychiatry Res 2005; 133: 111–122.

  78. 78

    Cardno AG, Holmans PA, Rees MI, Jones LA, McCarthy GM, Hamshere ML et al. A genomewide linkage study of age at onset in schizophrenia. Am J Med Genet 2001; 105: 439–445.

  79. 79

    Vawter MP, Barrett T, Cheadle C, Sokolov BP, Wood 3rd WH, Donovan DM et al. Application of cDNA microarrays to examine gene expression differences in schizophrenia. Brain Res Bull 2001; 55: 641–650.

  80. 80

    Mudge J, Miller NA, Khrebtukova I, Lindquist IE, May GD, Huntley JJ et al. Genomic convergence analysis of schizophrenia: mRNA sequencing reveals altered synaptic vesicular transport in post-mortem cerebellum. PLoS ONE 2008; 3: e3625.

  81. 81

    Aston C, Jiang L, Sokolov BP . Microarray analysis of postmortem temporal cortex from patients with schizophrenia. J Neurosci Res 2004; 77: 858–866.

  82. 82

    McInnes LA, Lauriat TL . RNA metabolism and dysmyelination in schizophrenia. Neurosci Biobehav Rev 2006; 30: 551–561.

  83. 83

    Fallin MD, Lasseter VK, Wolyniec PS, McGrath JA, Nestadt G, Valle D et al. Genomewide linkage scan for schizophrenia susceptibility loci among Ashkenazi Jewish families shows evidence of linkage on chromosome 10q22. Am J Hum Genet 2003; 73: 601–611.

  84. 84

    Sullivan PF, Lin D, Tzeng JY, van den Oord E, Perkins D, Stroup TS et al. Genomewide association for schizophrenia in the CATIE study: results of stage 1. Mol Psychiatry 2008; 13: 570–584.

  85. 85

    Benes FM, Lim B, Matzilevich D, Subburaju S, Walsh JP . Circuitry-based gene expression profiles in GABA cells of the trisynaptic pathway in schizophrenics versus bipolars. Proc Natl Acad Sci USA 2008; 105: 20935–20940.

  86. 86

    Walsh T, McClellan JM, McCarthy SE, Addington AM, Pierce SB, Cooper GM et al. Rare structural variants disrupt multiple genes in neurodevelopmental pathways in schizophrenia. Science (New York, NY) 2008; 320: 539–543.

Download references

Acknowledgements

This work was supported by funds from INGEN (Indiana Genomics Initiative of Indiana University), INBRAIN (Indiana Center for Biomarker Research In Neuropsychiatry) and NARSAD Young Investigator Award to ABN, as well as NIMH R01 MH071912–01 to MTT and ABN. ABN would like to thank Howard Edenberg for excellent help and advice with animal model microarray data, as well as Sudharani Mamidipalli, Griffin Fitzgerald and Jesse Townes for their precise work with database maintenance and data analysis. Most importantly, we would like to thank the subjects who participated in these studies, their families and their caregivers. Without their generous participation, such work to advance the understanding of mental illness would not be possible.

Author information

Correspondence to A B Niculescu.

Ethics declarations

Competing interests

ABN and DRS are founders and hold an equity interest in Mindscape Diagnostics, Inc. MAG holds an equity interest in San Diego Instruments, Inc.

Additional information

Supplementary Information accompanies the paper on the Molecular Psychiatry website

Supplementary information

Rights and permissions

Reprints and Permissions

About this article

Keywords

  • convergent functional genomics
  • blood
  • schizophrenia
  • hallucinations
  • delusions
  • biomarkers

Further reading