Serum levels of interleukin 1β (IL-1β) (A) and the expression of the chemokine ligand 2 receptor (CCR2) on monocytes (B) of patients with bipolar disorder (bipolar patients) and healthy control participants.
A, Box plot of log-transformed IL-1β is given. The serum IL-1β
level was determined on the serum samples of the same 42 Dutch patients and 25 healthy controls used for quantitative polymerase chain reaction analysis via a commercially available enzyme-linked immunosorbent assay (R&D Systems, Minneapolis, Minnesota) according to the manufacturer's protocol. The box indicates the lower and upper quartiles. The line within the box represents the median. The whiskers extend to the 2.5
and 97.5 percentiles. The outliers are characterized by the filled dots. Data were log-transformed to obtain a normal distribution of the cytokines/chemokines. Because this was not possible for IL-1β
(and IL-6 and chemokine ligand 7), the nonparametric Mann-Whitney test was used for statistical evaluation of IL-1β levels. B,
The expression of CCR2 on monocytes in mean fluorescence intensity levels as determined by flow cytometry analysis.
Means and standard deviations are given of CD14+CD16− monocytes and CD14+CD16+ monocytes.
The Mann-Whitney test was used to compare groups. For the test we used the monocytes of 12 bipolar patients (7 men and 5 women; mean age, 47 [range, 38-56] years), of whom 7 were euthymic; 5, depressed;
1, manic; and all taking lithium carbonate, and 9 healthy controls.
In the test, peripheral blood mononuclear cells (5 × 105 cells/mL) were stained (10 minutes at room temperature) with 25 μL of mouse antihuman fluorescein isothiocyanate (FITC)–
or phycoerythrin (PE)-conjugated monoclonal antibodies and washed afterward. The following monoclonal antibodies were used: anti-IgG1
FITC (Becton Dickinson, San Jose, California), anti-IgG1 PE (Becton Dickinson), anti-CD14 PE (Beckman Coulter, Hialeah, Florida), anti-CD16
FITC (1:10; Sanquin, Amsterdam, the Netherlands), anti-CD16 PE (1:10;
Becton Dickinson), and anti-CCR2 PE (1:5; R&D Systems). Immediately after staining, cells were measured using a flow cytometer (FACSCalibur;
Becton Dickinson) gating out debris and dead cells via light-scatter properties. Data were analyzed by means of commercially available software (CellQuestPro; BD Pharmingen, Alphen a/d Rijn, the Netherlands),
and monocytes were gated out by means of cell size (forward scatter)
and irregular shape (side scatter).
The hypothetical relationship of the 19 molecules (encircled) of the proinflammatory chemokinesis signature. The mitogen-activated protein kinases (MAPKs) are fundamental regulators of immune cell function.26 The 3 main classes are the extracellular signal-regulated kinases (ERKs), the p38 proteins, and the Jun N-terminal kinases (Jnks), which shuttle after an activating phosphorylation into the nucleus and initiate the transcription of immediate early genes of preexisting transcription factors.26 A preexisting transcription factor is Elk-1, in addition to activating transcription factor 3
(ATF3),26 which is bidirectionally regulated by the Jnk (positive) and ERK (negative) pathways27 and is induced during cellular stress. Dual-specificity phosphatases (DUSPs) regulate MAPK activity through dephosphorylation and also anchor or shuttle MAPKs.28,29DUSP2 (encoding PAC1) localizes to the nucleus and is one of the most highly induced transcripts in activated immune cells. A recent report shows that DUSP2 has a positive function in macrophage-mediated inflammatory responses via a lowering of the level of expression of Jnks and a compensatory rise in that of ERKs and p38.28 Of the actual MAPKs, MAPK6/ERK3 is overexpressed.
CDC42 acts as a Rho glutamyl transpeptidase–signaling molecule upstream from the MAPKs,31 but also as a molecule related to the cytoskeletal organization of the cell,
its motility, and chemotactic potential.30 Also, tumor necrosis factor α (TNF-α) feeds into the MAPK system and activates the Jnk cascade and ATF3 expression.27 Not only activators of the MAPK signaling pathway, but also repressors are part of the signature. NAB2 is a corepressor molecule that directly binds Egr-1 and inhibits its transactivating potential.32 Egr-1 is induced by Elk-1
and the Egr family plays a key role in coordinating subsequent waves of gene expression after the immediate early gene response induced by MAPKs.26,32 Another molecule with repressor activity, but part of the signaling cascade downstream from the TNF-α receptor (TNF-R), is TNFAIP3, which blocks the activation of the Jnk cascade by TNF-α33 and in this way demonstrates antiapoptotic and anti-inflammatory behavior.34 Another layer of negative regulation is seen at the level of the proinflammatory chemokine CCL2 and its receptor CCR2: an upregulation of the messenger RNA (mRNA) for CCL2 was opposed by downregulation of the mRNA of CCR2.
EMP1 is a tetraspan transmembrane protein playing a role in cell-cell adhesion and interactions with the extracellular membrane.35 Although the function of the molecule is not entirely known, it appears to be involved in cell survival and growth.36 BCL2A1 is a well-known apoptotic molecule.37 The proinflammatory cytokines and components are discussed in the text. Protein kinase A (PKA) and protein kinase C (PKC) are well-described second messengers. Nuclear factor kappa B (NF-κB) is a transcription factor known to be important in the inflammatory response. cAMP indicates cyclic adenosine monophosphate; ECM, extracellular matrix; Gαs, Gs protein; and Gq, Gq protein. Both Gαs and Gq are G proteins, which mediate the effects of G protein–coupled receptors.
Customize your JAMA Network experience by selecting one or more topics from the list below.
Padmos RC, Hillegers MHJ, Knijff EM, et al. A Discriminating Messenger RNA Signature for Bipolar Disorder Formed by an Aberrant Expression of Inflammatory Genes in Monocytes. Arch Gen Psychiatry. 2008;65(4):395–407. doi:10.1001/archpsyc.65.4.395
Mood disturbances are associated with an activated inflammatory response system.
To identify a discriminating and coherent expression pattern of proinflammatory genes in monocytes of patients with bipolar disorder.
A quantitative polymerase chain reaction (Q-PCR) case-control gene expression study on purified monocytes of bipolar patients, the offspring of bipolar patients, and healthy control participants after having selected 22 discriminating inflammatory genes using whole genome analyses.
Academic research setting in the Netherlands.
Forty-two bipolar patients with 25 healthy controls, 54 offspring of a bipolar parent (13 had a mood disorder and 3 developed one during follow-up), and 70 healthy children underwent Q-PCR.
Main Outcome Measure
Inflammatory gene expression levels in monocytes.
We detected in the monocytes of bipolar patients a coherent mutually correlating set (signature) of 19 aberrantly expressed (P < .01) messenger RNAs of inflammatory (PDE4B, IL1B, IL6, TNF, TNFAIP3, PTGS2, and PTX3), trafficking (CCL2, CCL7, CCL20, CXCL2, CCR2, and CDC42), survival (BCL2A1 and EMP1), and mitogen-activated protein kinase pathway (MAPK6, DUSP2, NAB2, and ATF3) genes. Twenty-three of 42 bipolar patients (55%) had a positive signature test result vs 7 of 38 healthy controls (18%)
(positive test result: positivity for PDE4B, ie, a messenger RNA 1 SD higher than the mean level found in healthy controls, plus 25% of the other genes with similar positive findings).
Positive signature test results were also present in 11 of 13 offspring with a mood disorder (85%), 3 of 3 offspring developing a mood disorder (100%), and 17 of 38 euthymic offspring (45%) vs 13 of 70 healthy children (19%). Lithium carbonate and antipsychotic treatment downregulated the gene expression of most inflammatory genes.
The monocytes of a large proportion of bipolar patients and offspring of bipolar parents showed an inflammatory gene expression signature. This coherent set of genes opens new avenues for biomarker development with possibilities for disease prediction in individuals genetically at risk and for the subclassification of bipolar patients who could possibly benefit from anti-inflammatory treatment.
The macrophage–T-cell theory of depression1,2 postulates an activated inflammatory response system in mood disorders and considers this activated inflammatory response system to be a driving force behind the illness, because proinflammatory cytokines are capable of destabilizing brain function.
This makes the brain vulnerable to stress and unknown endogenous factors such that major mood disturbances are the consequence. In animal models,
the behavioral changes induced by proinflammatory cytokines (sickness behavior) are comparable to the symptoms of depression as seen in humans.3Also in humans, low intravenous doses of endotoxin increase the level of these cytokines and induce depressive symptoms.4,5 Furthermore, depressive symptoms and mania can be precipitated by interferon alpha treatment,6 whereas anti–tumor necrosis factor (TNF) therapy given for psoriasis resulted in a markedly improved mood.7 Last, in patients with mood disorders, increased serum, saliva,
and cerebral spinal fluid levels of several proinflammatory compounds have been found, such as interleukin 6 (IL-6), IL-1β, TNF-α,
prostaglandin E2 (PGE2), and chemokine ligand 2 (CCL2),5,8-11 although other studies have reported negative findings.12 In addition, there have been reports on high serum levels of positive acute-phase proteins (eg, haptoglobin, α1-antitrypsin, ceruloplasmin, and C-reactive protein) and low levels of negative acute-phase proteins (eg, albumin and retinal-binding protein) in patients with mood disorders, indicating that mood disorders are accompanied by an acute-phase response.13-15
We herein report outcomes of a quantitative polymerase chain reaction (Q-PCR) study on monocytes of patients with bipolar disorder (hereinafter referred to as bipolar patients) in whom we found 19
aberrantly expressed genes involved in inflammation and inflammation-related processes after having identified and selected 21 such proinflammatory genes in whole-genome gene expression profiling (Affymetrix, Santa Clara, California) on purified CD14+ monocytes of a limited set of bipolar patients. The expression levels of the inflammatory and inflammation-related genes were mutually strongly correlated in 4 functional pathways, forming a monocyte gene expression signature.
To investigate whether this set of aberrantly expressed genes exists before disease, we also tested the monocytes of 54 children of bipolar parents; of these 54 children, 13 had a mood disorder and 3 were found to have a mood disorder within 2 to 3 years after blood sample collection.16
Outpatients with DSM-IV bipolar I or II disorder were recruited from the following 2 studies: the Dutch site of the Stanley Foundation Bipolar Network (SFBN), an international multicenter research program described elsewhere in detail17,18 (n = 19 for the gene profiling analysis studies and n = 19 for Q-PCR verification),
and an ongoing Dutch twin study on bipolar disorder described in detail in Vonk et al19 (n = 23
index cases for Q-PCR studies). Clinical characteristics of the bipolar patients used for Q-PCR are shown in Table 1.
The offspring of bipolar patients described herein belong to an ongoing prospective study among the adolescent offspring of a bipolar patient in the Netherlands. The children were not related to the bipolar patient group mentioned in the previous subsection. The study design,
recruitment procedure, and study population have been described elsewhere in detail.16,20,21 In brief,
86 bipolar patients and their spouses and 140 offspring aged 12 to 21 years were examined between November 1, 1997, and March 31, 1999
(time 1). Fourteen months after the first assessment, 132 offspring aged 13 to 23 years were available for reassessment (time 2). At time 2, we were able to collect immune cells for our study from 54 offspring (26 male and 28 female subjects). At the third assessment (time 3),
41 months after time 2, 129 offspring (aged 16-26 years) belonging to 80 families were still participating.16 Further characteristics of the bipolar offspring are given in Table 2.
A DSM-IV diagnosis of bipolar disorder was made by means of the Structured Clinical Interview for DSM-IV Axis I in the bipolar patients of the SFBN,
in the bipolar patients of the twin study, and in the bipolar offspring study at time 3 or by means of the Schedule for Affective Disorders and Schizophrenia, Children's Version in the bipolar offspring study at times 1 and 2. The bipolar patients of the SFBN underwent assessment at monthly follow-up visits by means of the Young Mania Rating Scale,
the Inventory of Depressive Symptomatology, and the Clinical Global Impressions Scale–Bipolar Version.
Adult healthy controls were recruited via the enrolling laboratory,
medical staff, and medical students. The offspring study controls were healthy Dutch high school students. The inclusion criteria for the control groups were an absence of psychiatric disorders such as psychosis, mood disorders, and anxiety disorders, and also of the medical disorders chronic fatigue syndrome and fibromyalgia, and a history free of these disorders in first-degree family members. Healthy controls had to be in self-proclaimed good health and free of any obvious medical illness for at least 2 weeks before the blood withdrawal,
including acute infections and allergic reactions. Healthy controls did not use any psychotropic or other medication.
The Medical Ethical Review Committee of the University Medical Center Utrecht approved the studies. Written informed consent was obtained from all participants after a complete description of the study was given.
Blood was collected in a clotting tube for serum preparation (frozen and stored at −80°C) and in tubes containing heparin sodium for immune cell preparation. From the heparinized blood samples,
peripheral blood mononuclear cell (PBMC) suspensions were prepared via low-density gradient centrifugation as described in detail elsewhere22 within 5 hours for all samples to avoid activation of the monocytes. The PBMCs were frozen in 10% dimethyl sulfoxide and stored in liquid nitrogen. This enabled us to test patient and control immune cells in the same series of experiments at appropriate times.
On the day of testing, stored PBMC suspensions were quickly thawed and directly diluted in RPMI 1640 medium with Ultraglutamine 1 and 25mM Hepes (BioWhittaker brand; Lonza, Verviers, Belgium) to wash out the dimethyl sulfoxide. The viability of the cells was invariably more than 90% (trypan blue exclusion staining). Monocytes were prepared from the PBMC suspensions by a magnetic cell sorting system for CD14
separation (MACS; Miltenyi Biotec, Auburn, California) according to the manufacturer's recommendations. Purity of the monocytes was routinely determined for each sample by morphological screening of the wet preparation after trypan blue staining. In addition, purity was determined by routine flow cytometry analysis (FACSCalibur [Becton Dickinson, Amsterdam,
the Netherlands] and antibody CD14 APC [BD Pharmingen, San Diego,
California]) on randomly chosen samples.
The RNA was isolated from the purified monocytes using minipreparation columns (RNeasy; Qiagen, Hilden, Germany) as described by the manufacturer and previously.23 The RNA was first converted into complementary DNA (cDNA) and subsequently into complementary RNA (cRNA). Fragmented cRNA was hybridized to U95Av2 microarrays (Affymetrix,
according to the manufacturer's protocol). For all experiments reported herein, the 5′ to 3′ ratios of glyceraldehyde-3-phosphate dehydrogenase (GAPDH) were 2 or less (usually 0.9-1.1).
All raw data obtained via whole genome gene expression profiling are available as MIAMExpress submission E-MEXP-1275 (http://www.ebi.ac.uk/miamexpress/).
The RNA was isolated from monocytes using minipreparation columns (RNeasy) per the manufacturer's protocol. After extraction, the RNA concentration was determined and RNA was stored at 80°C until use.
To obtain cDNA for the Q-PCR, we used the optimized extensively described Biomed-1 protocol.24 One microgram of RNA was reverse transcribed using a commercial reverse transcriptase kit (SuperscriptII; Invitrogen, Carlsbad, California)
and random hexamers (Amersham Biosciences, Roosendaal, the Netherlands)
for 50 minutes at 42°C.
Quantitative PCR was performed with a commercially available mix (TaqMan Universal PCR Master Mix; Applied Biosystems, Foster City, California). All TaqMan probes and consensus primers were preformulated and designed by the manufacturer (Assays-on-Demand; Applied Biosystems) on the basis of the detected transcript clusters of the bipolar discriminating genes detected in the Affymetrix analysis (available in eTable 1 and the “Results” section). The PCR conditions were 2 minutes at 50°C and 10 minutes at 95°C, followed by 40 cycles of 15 seconds at 95°C, and finally 1 minute at 60°C.
The PCR amplification of the housekeeping gene ABL (see eTable 1 for all gene accession numbers) was performed for each sample to allow normalization between the samples. We chose ABL as the housekeeping gene because it was previously shown that ABL was the most consistently expressed housekeeping gene in hematopoietic cells.25 Data were expressed as cycle threshold (CT)
values corrected to ABL (ΔCT values,
described in eTable 2 and eTable 3) and as fold change values determined via the ΔΔCT method (User Bulletin 2; Applied Biosystems) (Table 3 and Table 4). To correct for interassay variance we set the mean of the studied genes found in the healthy control groups in the same assay for each gene to 1, and the fold change values of the genes in patient monocytes were expressed relative to this set mean of 1 of the healthy controls for the given values.
Using the aberrantly expressed genes in the monocytes of the patients, various signatures could be defined. The definitions depend on the percentage of positive genes, with positivity defined as a messenger RNA (mRNA) expression 1 SD higher than the mean level found in the healthy controls.
Scanned microarray images were analyzed using commercially available software (Microarray Suite 4.2; Affymetrix). Further analysis was performed using Rosetta Resolver (http://www.rosettabio.com) biosoftware. The genes were classified with Ingenuity Systems software (http://www.ingenuity.com). Statistical analysis of the data was performed using the SPSS 11.0 statistical software package for Windows (SPSS Inc, Chicago, Illinois). Data were tested for normal distribution using the Kolmogorov-Smirnov test. Depending on the distribution pattern and the total number of subjects, parametric (normal distribution,
≥ 50 subjects) or nonparametric tests (skewed distribution,
< 50 subjects) were used. Differences in means between groups were determined using the Mann-Whitney test, or the Kruskal-Wallis test for more than 2 groups. Correlations were determined via Pearson or Spearman correlation coefficients, linear regression, and univariate analyses of variance and covariance. Frequencies were compared using the χ2 test. The specific tests used are indicated in the legends of the figures and tables.
Affymetrix analysis was performed on the MACS-purified pooled monocytes from 5 non–lithium carbonate–treated bipolar patients (aged 12-36 years; mean age, 22 years; 3 female and 2 male patients) and 6 healthy controls (aged 12-39 years; mean age, 20 years;
3 female and 3 male patients) with the aim to search in particular for inflammation-related genes. The 11 cell files are available as MIAMExpress submission E-MEXP-1275 (http://www.ebi.ac.uk/miamexpress/). We analyzed the data using the Rosetta Resolver program and considered for further study only genes with a statistically significant differential expression of more than 2-fold (P < .01)
between the bipolar patients and control group. This resulted in 71
discriminating genes for non–lithium-treated bipolar patients (64 upregulated and 7 downregulated). Major functional networks found to be involved in Ingenuity Systems analysis were indeed inflammation and cell movement but also cellular growth and differentiation. Pathways involved were the mitogen-activated protein kinase (MAPK) pathway,
the IL-6 signaling pathway, and the apoptosis pathway. To select for genes that could serve as potential monocyte biomarkers for the bipolar inflammatory condition, we took the top 7 genes from the list (ie, PDE4B, ATF3, MAPK6, DUSP2, TNFAIP3, CXCL2, and BCL2A1) and 6 genes with a statistically significant differential expression of more than 2-fold with a well-known involvement in inflammation, the MAPK and IL-6 pathways, and cell movement (ie, IL1B, IL6, TNF, PTGS2, PTX3, and CCL20).
In addition, we analyzed the pooled monocytes, purified by the magnetic cell sorting system, of lithium-treated bipolar patients and healthy controls. Cell pools were mainly used to limit costs and to avoid interindividual differences. One pool consisted of monocytes of 7 lithium-treated bipolar patients (1 man and 6 women; mean age,
39 years; age range, 27-57 years) compared with a pool of healthy control monocytes (1 man and 6 women; mean age, 40 years; age range,
24-56 years); the other pool consisted of 7 lithium-treated bipolar patients (4 men and 3 women; mean age, 44 years; age range, 37-57
years) compared with a pool of healthy control monocytes (4 men and 3 women; mean age, 45 years; age range, 39-53 years). The raw data of the 4 pools are also available as MIAMExpress submission E-MEXP-1275
(http://www.ebi.ac.uk/miamexpress/). After Rosetta Resolver analysis, we found 187 discriminating genes (114 upregulated and 73
downregulated) between lithium-treated bipolar patients and healthy controls (2-fold difference). Ingenuity Systems analysis showed that lithium treatment downregulated genes involved in inflammation but induced the de novo expression of genes involved in cell growth, differentiation,
survival, and apoptosis. Because we were searching for potential inflammatory biomarkers, we selected in this set of genes those well known to be involved in inflammation, the MAPK and IL-6 pathways, and cell movement,
that is, HSPA1A, NAB2, CDC42, ADAM17, CCL2, CCR2, CCL7, CX3CR1, and EMP1. The 22 Affymetrix-selected genes were further validated using Q-PCR.
Table 3 (column A) shows that, of the 22 mRNAs tested, 19 could be verified in Q-PCR to be significantly differentially expressed (P ≤ .01) in the monocytes of a series of 42 newly selected bipolar patients compared with 25 healthy controls (see eTable 2 for ΔCT values). These were the inflammatory genes PDE4B, IL1B, IL6, TNF, TNFAIP3, PTGS2, and PTX3; the chemokinesis/motility genes CCL2, CCL7, CCL20, CXCL2, CCR2, and CDC42; the MAPK pathway genes MAPK6, DUSP2, NAB2, and ATF3; and the cell survival/apoptosis genes BCL2A1 and EMP1. All were overexpressed in the monocytes of bipolar patients, except for CCR2.
Table 5 shows that the actual mood status of the patients is related to the inflammatory gene expression. During a manic episode, the mRNA expression of MAPK6 and CCL2 was significantly increased in monocytes of manic vs euthymic bipolar patients; during depressive episodes, expression of these mRNAs was raised in addition to that of IL6, PTX3, EMP1, and BCL2A1. Although active disease thus is related to the mRNA expression of these molecules, almost all 19 mRNAs were still significantly higher in euthymic bipolar patients compared with healthy controls, except for CCL2 and EMP1.
Table 6 shows that lithium and antipsychotic treatment reduced the expression of PDE4B and TNF. We therefore analyzed the Q-PCR monocyte data for bipolar patients without lithium or antipsychotic treatment separately from the data of those receiving lithium and/or antipsychotic treatment (Table 3, columns B and C). Results of this analysis clearly indicated that,
in non–lithium- and non–antipsychotic-treated bipolar patients, the gene expressions of PDE4B, IL1B, IL6, TNF, TNFAIP3, PTGS2, PTX3, CCL20, CXCL2, BCL2A1, and DUSP2 were at particularly high levels, which were reduced by lithium and antipsychotic treatment. Expression of CCL2, CCL7, CDC42, CCR2, ATF3, NAB2, and MAPK6 were not affected by lithium and/or antipsychotic treatment. Expression of EMP1 was not raised in bipolar patients off lithium and/or antipsychotics, whereas lithium treatment had a significantly stimulating effect on EMP1 expression (Tables 3 and 6). The Q-PCR data thus verify the limited Affymetrix data, showing a decrease in inflammatory genes and a rise in apoptotic genes in monocytes of lithium-treated bipolar patients.
Expression of EMP1 was also influenced by other medications; a higher level of expression was found with carbamazepine,
whereas a lower expression was found with valproate sodium (Table 6).
The other variables for which we tested (age, sex, body mass index, age at disease onset, duration of illness, and duration of lithium use) (via linear regression) lacked any significant effect on the aberrant mRNA expressions.
The expression levels of the 19 mRNAs were virtually all mutually strongly correlated in expression (see the eFigure [http://archgenpsychiatry.com]). The strongest correlations were found between PDE4B and the other mRNAs,
and PDE4B was the gene most consistently overexpressed. There are reasons to consider PDE4B as a key molecule in the proinflammatory state of the bipolar monocytes (discussed in the “Comment” section). We therefore tried out various definitions for the presence or absence of a PDE4B-associated proinflammatory mRNA signature (Table 7 and Table 8). The definitions given depend on the percentage of positive findings among the 19 genes, with positivity defined as an mRNA expression 1 SD higher than the mean level found in the healthy controls (Tables 7 and 8). In all definitions, PDE4B demonstrated positive findings. Using these definitions, it is evident that a PDE4B positivity in monocytes combined with a positivity for the other signature genes of up to 50% discriminates bipolar patients from healthy controls and that a proportion of up to 60% of bipolar patients is characterized by a PDE4B-associated proinflammatory, chemokinesis,
cell survival mRNA signature (depending on the signature definition).
It is also clear in this analysis that lithium and/or antipsychotic treatment downregulated the signature expression (Tables 7 and 8).
Table 4 demonstrates that the 13 bipolar offspring with a lifetime diagnosis of a mood disorder showed a raised level of expression of the mRNAs for all 18 signature genes tested (CCR2 was not tested) at time 2 (the time of monocyte testing) (for ΔCT values, see eTable 3). When expressed as signature positivity (Table 9 and Table 10), 11 of the 13 offspring with a lifetime diagnosis of a mood disorder at time 2 were positive for the proinflammatory signature if it was defined as positive findings for PDE4B plus 25% of the 19 genes. Four of those 13 offspring had a lifetime diagnosis of bipolar disorder; all 4 were positive for the signature. The other 9 children had a lifetime diagnosis of depression, and 7 of them had a proinflammatory signature. There was no relationship of positivity and negativity for this signature and the precise psychiatric diagnosis of the unipolar mood disorder (major depressed, dysthymia, or depression not otherwise specified).
Three offspring who were healthy at the time of blood collection at time 2 developed depression 41 months later at time 3. All 3 of these subjects had monocytes with the proinflammatory signature at time 2 (Tables 4, 9, and 10).
One offspring who had a lifetime diagnosis of depression at time 2 underwent conversion to a lifetime diagnosis of bipolar disorder at time 3; this subject had a full-blown proinflammatory signature (defined as positivity for PDE4B plus 100%
of the 18 genes) at time 2 (data not shown).
Table 4 additionally shows that healthy offspring of a bipolar parent (ie, no lifetime mood disorder at any time point tested) demonstrated aberrancies in the expression of the 19 proinflammatory, chemokinesis, and cell survival genes in their monocytes. The inflammatory gene signature defined as positive for PDE4B plus 25% or greater of the genes was found in 45% of them vs 19% in healthy young adults (Tables 9 and 10). However, expression levels of the proinflammatory genes IL6, TNF, TNFAIP3, PTX3, PTGS2, CCL7, CCL20, and CXCL2, and the MAPK pathway genes DUSP2 and ATF3 in particular were significantly higher in offspring with a mood disorder than in healthy euthymic offspring (Table 4).
We measured (via standard commercially available enzyme-linked immunosorbent assays) the protein expression levels of the signature genes IL1B, IL6, CCL2, and CCL7 in bipolar patients and healthy controls described in Table 1. Although the serum levels of all 4 cytokines were higher in the bipolar patients, only that of IL-1β was statistically significantly raised (2 times higher in bipolar patients vs healthy controls, Figure 1A). The serum level of IL-1β was not influenced by mood (P = .58, Kruskal-Wallis test) or by treatment with lithium, antipsychotics, or any other drug.
There was no effect of age, sex, or body mass index on the levels of the cytokines.
We also measured the expression of CCR2, the receptor for CCL2, on monocytes via standard flow cytometry analysis and found that CCR2 was mainly expressed on the mature CD14+CD16+ set of monocytes (Figure 1B). Expression of CCR2 was not significantly different between monocytes of bipolar patients and healthy controls.
Levels of CCR2 expression were independent of mood status and use of medication.
We herein show that the monocytes of most bipolar patients and bipolar offspring (particularly those who later develop a mood disorder)
have an altered mRNA expression of genes involved in inflammation and inflammation-related processes.26-47 Owing to the coherent and mutually strongly correlating aberrant expression, the mRNAs form an mRNA signature representing a set of proinflammatory genes that discriminate bipolar patients from healthy controls.
Based on the literature, an interaction model of the aberrantly expressed genes can be constructed (Figure 2). This model illustrates the mutual interdependency of the molecules. The expression of positive and negative signals in the network is an indication that in the aberrant monocytes of bipolar patients regulation is operative. For inflammatory cytokines, mRNA and protein levels often do not correlate. Indeed, the approximately 6-fold- raised expression of IL1B at the monocyte mRNA level was reflected only in a 2-fold-raised protein level in the serum of bipolar patients, whereas CCL2, CCL7, IL-6,
and CCR2 protein expression were not significantly different. Clearly,
a regulation at the protein transcription level is operative in the monocytes of bipolar patients to ensure a close-to-normal function,
and the question thus arises concerning which environmental or endogenous condition will create a failure of the monocyte to keep control over its aberrant gene expression, avoiding a higher protein production of the proinflammatory compounds. Psychological stress (both acute and chronic) might be such a condition (via adrenaline and glucocorticoid signaling),48 and indeed stressors have an upregulating effect of IL-1β and IL-6.49 In addition, stressful life events triggered the onset of mood disorder episodes in the offspring of bipolar patients described herein,50 showing a possible interaction between the existence of an aberrant monocyte proinflammatory gene expression signature and environmental stress factors. A cofactor playing a role in this stress-induced immune activation might be a relative resistance of immune cells to glucocorticoids, which has been reported in patients with bipolar disorder.22 Such decreased glucocorticoid sensitivity has also been thought to be the mechanism behind the well-known hypothalamic-pituitary-adrenal axis disturbances in patients with mood disorders,51-53 and these disturbances have led to the concept that impaired glucocorticoid receptor signaling is a crucial factor in the pathogenesis of mood disorders.54
The overexpression of proinflammatory cytokines, mediators,
and chemokines in the monocytes of bipolar patients was in particular correlated with the overexpression of phosphodiesterase type 4B (PDE4B),
a cyclic adenosine monophosphate (cAMP)–degrading enzyme. It is tempting to speculate that this molecule acts as a key molecule in the proinflammatory state of bipolar monocytes apart from the fact that it was the most consistently overexpressed gene, because (1)
cAMP is known to be inhibitory for inflammatory cells55; (2) of the cAMP-specific isoenzymes, PDE4
is expressed in all the inflammatory and immunomodulating cells40; (3) high activity of PDE4B leads to a proinflammatory state40,41; (4) the PDE4B gene is crucial for the proinflammatory action of monocytes in gene knockout studies42; (5) the in vitro differentiation of monocytes to inflammatory macrophages leads to upregulation of PDE4B43; and (6) PDE4 inhibitors broadly inhibiting functions of inflammatory cells40,41 are in a far stage of development.
Using such medication not only results in anti-inflammatory effects such as reduction of TNF-α production42,45 but also in antidepressive effects when tested in animal models47 and clinical trials.56
Our findings also suggest that an assessment of the monocyte gene signature might be useful for prognostic purposes. In a previous report on the offspring of bipolar patients, we confirmed that bipolar disorder is more prevalent in such offspring and is preceded by 1
or more episodes of (until then unipolar) depressive episodes.20 Herein we showed that bipolar offspring with a lifetime diagnosis of bipolar or (still) unipolar depressive disorder were positive for the mRNA signature (similar to our adult bipolar patients). More importantly, all 3 offspring who were psychiatrically healthy at the time of first monocyte testing but had developed depression 3 years later had a positive monocyte gene signature. This suggests that a positive gene signature precedes the onset of the first mood episode in individuals at risk and is not the consequence of the psychiatric condition. Also, a large proportion of the still healthy euthymic offspring showed an aberrant gene signature. We hypothesize that these signature-positive offspring in particular are at risk for the development of depression and, eventually, possible bipolar mood disorder. We are currently planning a 10-year follow-up of our offspring study and hope to validate the signature test as a potential inflammatory biomarker set linked to mood disorders.
Our data on a raised expression of the genes for proinflammatory cytokines and related compounds support the macrophage–T-cell theory of mood disorders.57-59 However,
other reports refute a high level of cytokine production from immune cells in mood disorders,60,61 and our finding of a discrepancy between mRNA and protein levels of cytokines may in part explain this controversy. Our finding also underscores the superiority of gene expression over protein detection in diagnosing the activated inflammatory response system of bipolar patients.
We also found PTGS2 (cyclooxygenase 2 [COX-2], involved in the production of PGE262) and PTX3 to be part of the proinflammatory gene expression signature. Increased PGE2 levels have been described in the saliva,11 serum,63 and cerebrospinal fluid10 of depressed patients, and PGE2 seems to be directly involved in the sickness behavior of animals.64 An in vitro study reported an increased PGE2 secretion from blood cells of depressed patients,65 and a COX-2 inhibitor has been reported to enhance the therapeutic effects of reboxetine in patients with a major depressive disorder.66 Cells of the monocyte-macrophage lineage exposed to a variety of inflammatory signals, including IL-6,38 express and release PTX3, which has a nonredundant role in resistance to selected microbial agents,39 is also expressed in the brain and involved in neuronal plasticity and degeneration,67 and has been found to be overexpressed in the fibroblasts of melancholic patients.68
With regard to a genetic background for the aberrant proinflammatory gene signature in mood disorder, gene polymorphisms in important signature genes (ie, IL1B, TNF, and CCL2) have been described as being linked to the presence or treatment response of the disorders.69-71 However,
the inflammatory program is also activated in response to infection.
Because bipolar disorder has been linked to infection by viral and intracellular pathogens,72-74 a simple explanation for the inflammatory signature could be that the monocytes are reacting to infection.
Most of our bipolar patients used various psychotropic medications that influenced the expression level of the signature genes. The strongest effects were for lithium and antipsychotics. Extensive literature describes the immunomodulating effects of antipsychotics and lithium.75-77 With regard to proinflammatory cytokine production, reports are inconsistent.60,78-80 In our in vivo study, the mRNA levels for important inflammatory cytokines/compounds were reduced in monocytes by lithium and/or antipsychotic treatment but did not completely normalize to the level found in healthy controls.
Lithium also has profound effects on the MAPK system.81,82 In the rat brain, it stimulates phosphorylation of signaling molecules in the extracellular signal-regulated kinase (ERK) branch of the MAPK route81 and influences the expression of the ERK pathway–regulated apoptosis molecule B-cell leukemia/lymphoma 2–related protein A1 (BCL2A1).82 In our bipolar patients with already raised mRNA levels of the MAPK pathway genes and of BCL2A1, we could not find a stimulating effect of lithium on MAPK pathway genes or on BCL2A1 in the monocytes.
We herein described an inflammatory monocyte gene expression signature in patients with bipolar disorder reflecting their activated inflammatory response system. This signature is also present in bipolar offspring, particularly in those who later develop a mood disorder.
Correspondence: Hemmo A. Drexhage,
MD, PhD, Department of Immunology, Erasmus Medical Center, PO Box 1738, 3000 DR Rotterdam, the Netherlands (email@example.com).
Submitted for Publication: June 4, 2007; final revision received December 5, 2007; accepted December 5, 2007.
Author Contributions: Dr Drexhage had full access to all data in the study and takes responsibility for the integrity of the data and the accuracy of the analysis. Drs Padmos and Hillegers share equal first authorship.
Financial Disclosure: None reported.
Funding/Support: This study was supported in part by a grant from the Stanley Medical Research Institute.
Additional Contributions: Harm J. de Wit, BSc, provided technical assistance. Christa Walgaard, MD, ran the CCR2 expression levels on bipolar monocytes.
Tar van Os helped in the design of the Figures. Caspar Looman, MSc,
provided guidance for the statistical analyses.
Create a personal account or sign in to: