The last 10 years evidence has accumulated that dysfunctions of T cells, and monocytes/macrophages are important factors in the development of bipolar disorder (BD). The majority of studies focused on cytokine levels in serum, and a recent meta-analysis confirmed the elevation of pro-inflammatory, anti-inflammatory, and regulatory cytokines in BD (1).

Cellular studies have accumulated too. Our team has studied in the past 10 years the percentages of circulating subsets of T cells and the monocyte inflammatory gene expression in three patient cohorts of BD patients. We have studied (1) a cohort of 56 BD patients from the Dutch site of the Stanley Foundation Bipolar Network (D-SFBN) cohort, mean age 42 years (2, 3), (2) a cohort of 90 BD patients, aged 43 years (mean) from the MOODINFLAME-Groningen-Leuven site (4), and (3) a cohort of 140 children of a bipolar parent in follow-up, from 16 through 29 years, the so-called Dutch bipolar offspring (DBO) study (5). In these studies, we detected that the monocytes of subjects were characterized by the abnormal expression of two coherent clusters of inflammation-related genes. One cluster of monocyte genes represented pro-inflammatory cytokines and key regulating compounds for the production of these cytokines (cluster 1 with genes, such as IL1B, IL6, TNF, PDE4B, and ATF3). The other cluster represents monocyte genes playing a role in chemotaxis, motility, nuclear signaling, and metabolism (cluster 2 with genes, such as CCL2, CCL7, NAB2, PTPN7, and DHRS3). The two clusters were found upregulated in monocytes particularly during a manic/depressive episode (4, 6), in longstanding disease with an early onset (6), and in 16-year-old DBO subject irrespective of the presence of mood symptoms (5).

The pro-inflammatory state of the monocytes was also found in yet another cohort of patients (a fourth cohort), namely, patients with postpartum psychosis, using not only gene expression but also microRNA profiling (7, 8). Postpartum psychosis is a pathologic condition related to BD. In postpartum disorder, monocytes not only showed increased levels of cluster 1 and 2 genes (7) but also decreased levels of miR-146a (8). MiR-146a is a well-established inflammation-regulating microRNA, and its decrease underscored the pro-inflammatory state of the patient monocytes. In addition, in silico studies showed monocyte miR-146a to target genes of the pro-inflammatory monocyte signature, such as PTGS2 and IRAK2.

Indeed, microRNA studies form an entire new field to investigate a new layer of epigenetic regulation of cellular function. While the study of Weigelt et al. (8) and those of others (9) target circulating leukocytes, most studies in psychiatry have focused on abnormal microRNA expression in the brain investigating microRNAs playing a role in brain development [reviewed in Ref. (10)].

Apart from the pro-inflammatory state of monocytes, we also detected in these four cohort studies that the T cell system was set at another equilibrium. In all cohorts, there was a deficit in the number of T cells, while within the T cell population there were abnormal fluctuations over time in the circulating levels of anti-inflammatory T regulatory (Treg) cells and T-helper 2 (Th2) cells and of pro-inflammatory T-helper 1 (Th1) and T-helper 17 (Th17) cells (7, 11, 12).

These observations enforced our idea that the immune system of individuals at high risk to develop BD (the DBO subjects) and of BD patients was abnormal and dynamically activated and deactivated over time showing episodes of inflammatory activation, particularly in early stages of the disease process in adolescence and the absence of mood symptoms and during active disease.

Recently, we collected circulating leukocytes of another cohort of BD patients. These patients belonged to the so-called Bipolar Stress Study (BiSS) cohort and had a mean age of 52 years (significantly older than the Stanley Foundation, the MOODINFLAME, and the DBO cohort). The BiSS study is a 2-year longitudinal study, designed to identify risk factors that have an impact on the clinical course and the treatment of outpatients with BD (13). The cohort has been studied in depth clinically (14–17) and is special in that the subjects had also been studied for the presence of the metabolic syndrome (MetS) (18). BiSS subjects showed a significantly higher prevalence of MetS when compared to subjects with MDD and non-psychiatric controls of the same age (28.4 versus 20.2 and 16.5%, respectively, p < 0.001), also when adjusted for sociodemographic and lifestyle factors. It is known that patients with severe mental illness do have a higher risk for cardiovascular disease (19).

Also special is that BiSS subjects had been studied for their hypothalamic–pituitary–adrenal (HPA) axis activity over a longer period of time via measuring their hair cortisol level (16). The hair cortisol test is relatively novel and reflects the mean HPA-axis activity over a period of time, in our case over the last 3 months before the leukocyte collection. Staufenbiel et al. (17) reported an increase of hair cortisol levels in the BD patients of the BiSS cohort, but only in those with negative life events within 3 months before the cortisol determination. Hair cortisol was found normal across the entire BiSS cohort.

The MetS, the HPA axis, and the immune system have strong mutual influences (20–22). The collection of circulating leukocytes in the BiSS cohort at the end of the 2-year follow-up period enabled us to study the monocyte gene expression and T cell subset distribution in the BiSS subjects in relation to the mean HPA-axis activity in the 3-month period before blood collection and in relation to the presence/absence of the MetS.

We therefore not only determined the earlier reported cluster 1 and 2 gene sets in the monocytes of the BiSS subject but also the gene expression levels of the genes for the glucocorticoid receptor α (GRα, activating) and GRβ (blocking receptor). We hypothesized that there would a relationship between the expression of the GR receptor in monocytes, the cluster 1 and 2 gene expression, and the hair cortisol levels.

Not only did we add the genes for the GR but also additionally determined the expression of the gene for hepatocyte growth factor (HGF). In a recent study on type 2 diabetic (T2D) and MetS patients (23), we found HGF gene expression raised in the circulating monocytes of the T2D patients. HGF is an important vascular repair factor (24) and monocyte-derived circulating angiogenic cells (CACs), capable of repairing endothelial damage, are characterized by the expression of HGF (25). We hypothesized that the HGF expression would be raised in particularly BiSS subjects with the MetS.

In addition to monocyte gene expression studies, fluorescence-activated cell sorting (FACS) analysis was performed on the collected leukocytes of the BiSS cohort to determine the percentages of monocytes, lymphocytes, B cells, NK cells, T cells, as well as the distribution of the T cells into the cytotoxic T cells and the helper T cells. The latter population was further quantified for its subpopulations, i.e., the regulatory T cells, the Th1 cells, the Th2 cells, and the Th17 cells. We hypothesized that we would detect again the T cell deficits and an abnormal distribution of T-helper subsets within the circulating T cell population of the BiSS subjects, similar as was found before.

Thus, our prime objective of this study was to determine the monocyte gene activation patterns and leukocyte distribution abnormalities in this new BiSS cohort of BD patients, and to compare outcomes to the outcomes of the previous studied BD cohorts, which had been investigated with the same techniques. Our secondary objective was to study correlations between the immune outcomes and the hair cortisol level and the presence/absence of the MetS.

This study shows that the pro-inflammatory gene cluster 1 expression was downregulated in the monocytes of the BD patients of the here studied BiSS cohort, indicating a reduced inflammatory state of the circulating monocytes as compared to that of HCs.

This reduced inflammatory state of the circulating monocytes of the BiSS patients is in contrast to the pro-inflammatory state observed in previously studied cohorts of BD patients while in all cohorts the same collection and laboratory determination techniques were used. In the BD patients of the D-SFBN cohort, an increased inflammatory gene expression was found that correlated to the activity of disease, particularly to manic symptoms (2, 3, 6). The patients in the D-SFBN cohort were indeed special in that a relative high percentage had active severe disease (38%). In the MOODINFLAME Leuven–Groningen cohort, the raised pro-inflammatory gene expression in monocytes was again only found in patients with active mania or depression, these patients constituted around 10% of the cohort (4). Our data thus support a concept that in BD patients over expression of monocyte inflammatory genes can only be found during active episodes of the disease, in particular during mania. The BiSS cohort only had six patients with severe depression.

Our observations on monocytes are by and large in accord with serum cytokine/inflammatory factor studies that also found that BD patients with an active episode and particularly those who have or develop manic symptoms have an increased inflammatory activity, especially high CRP, compared with individuals who do not develop manic symptoms (34).

However, before to conclude that the pro-inflammatory state in BD is only linked to (manic) mood derailment in BD, it is relevant that in the studies on the DBO children over expression of monocyte inflammatory genes was also found in adolescents at 16 years of age, but in this case irrespective of present or later mood symptoms (5). Five years later, in early adulthood (at 21 years of age), the pattern of monocyte gene expression had changed in the DBO children to a reduced expression of cluster 1 genes, but still increased cluster 2 gene upregulation. At 28 years, monocyte gene expression had virtually normalized. Neither increases nor decreases in monocyte inflammatory gene expression were linked to mood symptomatology in the DBO children in the various stages of investigation.

Collectively, these findings on the inflammatory state of monocytes point in the direction of dynamic changes of inflammatory gene expression in monocytes over time in subjects at risk to develop BD and in BD patients. Apparently the inflammatory set point of monocytes is dependent on the (environmental or hormonal) state of the organism, an observation that is supported by twin studies in BD, showing that common endogenous or environmental factors play a prime role in the pro-inflammatory set point of the monocytes (35). In sum, only episodes of pro-inflammatory activation occur in circulating monocytes of individuals at risk to develop BD and in BD patients, and these episodes are probably environmentally and endogenously precipitated and only in part linked to the presence and activity of (manic) mood symptoms.

The present study also shows that the gene expression of the vascular repair factor HGF was significantly raised in the monocytes of BD patients of the BiSS cohort as compared to HCs. We previously found a similar gene expression profile, i.e., a reduced pro-inflammatory cluster 1 and raised HGF expression, in the circulating monocytes of 64 T2D patients (23, 36). These T2D patients were on average 62 years old.

We took the observation in the studies of Baldóon Rojas et al. of a high HGF expression and a low expression of inflammatory signs in the circulating monocytes of T2D patients as indicating that many of the circulating monocytes in the T2D cases were differentiating into CACs and involved in the repair of endothelium damaged by the diabetic process. CACs are bone marrow-derived monocytes involved in vascular regeneration and angiogenesis, are anti-inflammatory, and exert their repair function by secreting vascular growth factors, such as HGF, which is considered as a marker of CACs (25). Healthy angiogenesis is crucial for vascular regeneration (24).

We therefore like to hypothesize that the elevated gene expression of HGF in the anti-inflammatory monocytes of the patients of the BiSS cohort probably reflects that a high percentage of the monocytes had developed into vessel-repair-supporting CACs to counteract a putative endothelial damage in the BD patients of the BiSS cohort. Indeed, the MetS and dyslipidemia were prevalent in the BD patients of the BiSS cohort (18), yet there were no correlations between the HGF expression and the MetS. Therefore, formal studies in older BD patients focusing on monocyte-derived CAC development in relation to parameters of vascular damage need to be performed to verify or refute the here proposed idea of the development of anti-inflammatory CACs as overruling mechanism preventing a pro-inflammatory state of monocytes in BD patients. Also endothelial damage in atherosclerosis is considered a pro-inflammatory environment (37), and it is therefore at first sight puzzling that the CACs are anti-inflammatory. However, there is literature that it is particularly in situations of chronic mild hyperglycemia (38) and in situations of unstable plaque formation (39) that CACs and circulating monocytes become pro-inflammatory and dysfunctional. Therefore, special care should be taken in follow-up studies whether such conditions exist in bipolar patients next to a putative heightened endothelial damage.

The GRα was also higher expressed in the monocytes of the BD patients of the BiSS cohort as compared to those of HCs, and the expression of the GRα expression correlated to the expression of HGF. Since the GRα is the activating receptor signaling downregulation of pro-inflammatory molecules during glucocorticoid exposure, an upregulation of the GRα suggests a higher sensitivity for glucocorticoids to downregulate the inflammatory state of monocytes.

These data suggest that in principle a raised cortisol level might be instrumental in downregulating the inflammatory state of the circulating monocytes in BD patients. Such a raised cortisol level was reported in BiSS BD patients who experienced recent negative life events (17). However, although we were able to verify the previous correlation between high hair cortisol levels and negative life events, we were unable to find in this study a correlation between both high hair cortisol levels/negative life events on the one hand and a reduced inflammatory state of the circulating monocytes of the BiSS cohort on the other hand.

We therefore assume that not negative life events and a higher HPA-axis activity, but the earlier mentioned putative endothelial damage and consequent transition of the circulating monocytes into anti-inflammatory vessel repair cells CACs plays a overruling and prime role in the here found characteristic profile (high HGF, reduced pro-inflammatory compounds) of the circulating monocytes in the BD patients of the BiSS cohort. But as stated before, this needs formal verification, also with regard to an enhanced monocyte sensitivity to dexamethasone in functional in vitro exposure tests.

With regard to the T cell findings in the BiSS cohort the following remarks. Snijders et al. (12) took the monocyte data on the DBO adolescents of Mesman et al. (5) further and showed that the DBO 16-year-old adolescents had deficits in the number of circulating lymphocytes, particularly of total CD3⁺ T cells and within this population particularly in the Treg cell subset. Interestingly, she found that the lower the level of the anti-inflammatory Treg cells in the DBO adolescents, the higher the monocyte inflammatory gene activation. These data corroborate earlier findings that Treg cells are capable of downregulating monocyte inflammatory activation and that in particular a defective Treg cell system might form the underlying condition allowing endogenous or environmental factors to induce an early raised pro-inflammatory state of monocytes in the 16-year-old DBO subjects at risk to develop BD.

During further aging of the DBO children, the aberrant T cell state showed a dynamic course over time. During young adulthood (mean age 21 years), the partial T cell deficit, i.e., the reduction in the total number of CD3⁺ T cell was still detectable, but not of the Treg cell subset anymore, but of the pro-inflammatory Th1 and Th17 cell subsets, suggesting a phase of relative anti-inflammation at young adulthood, but in the presence of deficits of total T cells. Also the upregulation of cluster 1 inflammatory genes in monocytes had disappeared and even a downregulation of such genes as compared to HC monocytes was found (see before), enforcing the idea that young adulthood at 21 years of age was a phase of relative anti-inflammation.

In late adulthood (mean age 28 years), the various subsets of T cells (Th1, Th2, Th17, and Treg cells) and the abnormal monocyte gene expression had largely normalized as compared to the controls, only the deficit of lymphocytes and total CD3⁺ T cells could still be detected.

Collectively, these DBO data show that T cell deficits deregulate the inflammatory response in adolescence and young adulthood in individuals at risk for BD, and we assume that the immune disequilibria represent vulnerability factors for later BD development by influencing morphological brain and hippocampus development during adolescence and early adulthood, since T cells and non-inflammatory microglia are essential support cells for proper brain development (see Figure 1).

Although the T cell and monocyte inflammatory disequilibria were more or less normalized in adulthood in the DBO subjects, a partial T cell defect (a deficit in lymphocyte and total T cell numbers) still existed. This report on the BiSS BD patients with a mean age of 52 years still finds the apparently over time consistent partial T cell deficit in showing reduced levels of lymphocytes and total CD3⁺ T cells also in the relatively old BiSS cohort, while the restoration within the CD4⁺ T-helper cells subset had further continued into an overshoot, showing significantly raised levels of Th2 and Th17 subpopulations, while Treg cells were reduced.

A composite of the dynamics over time of T cells and monocytes as found in the various cohorts studied by us is shown in Figure 1. In this figure, it is also hypothesized that particularly in adolescence and early adulthood changes in brain structure and physiology are induced by the T cell deficits and inflammatory influences. In later phases of life, partial corrections of the immune aberrancies occur, and only during an active episode signs of a peripheral monocyte inflammatory state can be found.

Despite the limitations, we are confident that this study shows that lymphopenia, and in particular a reduced percentage of T cells, is a characteristic of relatively old BD patients. Since lymphopenia and a reduced percentage of T cells are phenomena also found in children at risk for BD and younger BD patients, a deficit of T cells might be a trait phenomenon of BD.

The monocyte gene expression profile of the relatively old BD patients showed a downregulation of inflammatory genes and an upregulation of HGF, suggesting CAC activity of the cells. This profile is different from the pro-inflammatory profile found in 16-year-old offspring of a bipolar parent and in BD patients with active disease. Therefore, the changes in monocyte profiles suggest dynamic changes over time in the inflammatory state of monocytes of individuals at risk for manic-depressive episodes. We therefore assume that the monocyte inflammatory set point is a state rather that a trait phenomenon of BD.

All the authors collected part of the study cohort, evaluated the data, particularly the hair cortisol and metabolic parameters, and assisted in writing of the manuscript.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.