Introduction

A variety of scientific studies indicate a relationship between the pathophysiology of affective disorders and the immune system (Lichtblau et al. 2013; Lotrich 2012; Kenis and Maes 2002). Moreover, psychopharmacological agents for the treatment of affective disorders have immunological properties. These agents include antidepressants (Hinze-Selch et al. 2000; Kraus et al. 2002; Munzer et al. 2013) and mood stabilizers (Himmerich et al. 2013), which have been hypothesized to restore the immunological imbalance associated with the affective disorder or lead to immunologically mediated side effects. Lithium is another treatment for affective disorders, recommended for acute as well as long-term treatment for bipolar disorder (Müller-Oerlinghausen et al. 2002; Geoffroy et al. 2014; Geddes and Miklowitz 2013). Furthermore, augmentation with lithium is currently the best-evidenced therapy for depressed patients who do not respond to monotherapy with an antidepressant (Bschor et al. 2003; Bauer et al. 2010; Baumann et al. 1996).

However, lithium has a number of side effects such as hypothyroidism, hyperparathyroidism, gastrointestinal and renal problems, diarrhoea, tremor, polyuria, nocturnal urination, weight gain, oedema, flattening of affect and cutaneous side effects such as psoriasis (McKnight et al. 2012; Grandjean and Aubry 2009; Beyaert et al. 1992; Paquet et al. 2005). These side effects are of specific concern for individualizing psychiatric therapy, for example, if a patient with known psoriasis or autoimmune thyroid disease is in line for lithium therapy, because lithium increases the propensity to autoimmunity at least in susceptible individuals (Kibirige et al. 2013).

However, not only certain side effects, but also antidepressant effects of lithium may be, at least in part, mediated by cytokines. Cytokines mediate the most important signalling pathway from the immune system to the brain and back (Capuron and Miller 2011). On a molecular level, pro-inflammatory cytokines such as interleukin (IL)-1ß, IL-2, IL-6 or tumour necrosis factor (TNF)-α can influence neurotransmitter metabolism (Seidel et al. 1995), for example the metabolism of the monoamines dopamine, noradrenalin and serotonin (Capuron and Miller 2011; Himmerich et al. 2009; Berthold-Losleben and Himmerich 2008). Immune activation with increased production of pro-inflammatory cytokines, such as interferon (IFN)-γ, leads to the production and activation of indolamine-2,3-dioxygenase (IDO) (Brandacher et al. 2007). Increased IDO activity, in turn, leads to tryptophan deficiency by metabolizing tryptophan to kynurenic acid, which can cause depressive symptoms due to the resulting reduction in serotonin synthesis in the CNS (Müller 2014; Catena-Dell’Osso et al. 2013; Müller and Schwarz 2007). Furthermore, increased levels of pro-inflammatory cytokines may activate the hypothalamic–pituitary–adrenal (HPA) axis and increase the release of corticoid-releasing factor (CRF), adrenocorticotropic hormone (ACTH) and glucocorticoids; and overactivity of the HPA axis has been repeatedly reported to be associated with major depression (Besedovsky et al. 1991; Ising et al. 2005).

As some antidepressants have been shown to reduce the production of cytokines, such as IL-1ß (Himmerich et al. 2010b) and IFN-γ (Himmerich et al. 2010a) it was hypothesized that antidepressants exert their therapeutic effects by reducing the activation of pro-inflammatory cytokines, in addition to their inhibitory action on monoamine reuptake transporters as specified in Schildkraut’s monoamine deficiency hypothesis of depression (Schildkraut 1965). With regard to the therapy of manic episodes, lithium’s therapeutic ability to dampen the mood may also be related to its influence on cytokine signalling and may specifically be a consequence of an activation of pro-inflammatory cytokine secretion.

Lithium’s activation of immunological processes has also been shown to have beneficial effects in diseases and syndromes that are associated with attenuated global immune response or specific immunological deficits that can be compensated by lithium. Examples of immunological diseases that might benefit from lithium therapy are infection with the human immunodeficiency virus (HIV) (Crews et al. 2009; Letendre et al. 2006; Harvey et al. 2002) and systemic lupus erythematosus (SLE) (Kucharz et al. 1993). Patients suffering from one of these diseases and a coincident affective disorder may specifically benefit from lithium therapy.

In depression, lithium is usually given in combination with antidepressants. However, the influence of the combination of antidepressants and lithium on cytokine production has not yet been systematically investigated. Therefore, we sought to explore the effect of lithium, of antidepressants, and of antidepressants in combination with lithium, in vitro. We measured the levels of the cytokines IL-1ß, IL-2, IL-4, IL-6, IL-22, IL-17 and TNF-α in the stimulated blood of fifteen healthy female and fifteen healthy male subjects supplemented with various antidepressants (citalopram, escitalopram and mirtazapine) alone, with lithium alone and with a combination of the mentioned antidepressants with lithium in a whole blood assay.

As a stimulant we used OKT3/5C3 and phytohemagglutinin (PHA). OKT3/5C3 consists of the murine anti-human CD3 monoclonal antibody OKT3 (Muromonab-CD3) which binds to the T cell receptor CD3 complex and is an established T cell activator (Adair et al. 1994) and the 5C3 monoclonal antibody, which reacts with human CD40 and is reported to be used for activation of B cells in functional in vitro assays (Pound et al. 1999).

PHA reacts with certain oligosaccharides on the cell surface, leading to erythrocyte agglutination. It is an established stimulatory molecule to investigate cytokine production under the influence of antidepressants (Kubera et al. 2004).

Although IL-17 is implicated in numerous immune and inflammatory processes (Hemdan et al. 2010; Park et al. 2005; Murdaca et al. 2011) not only in the body’s periphery but also in autoimmune diseases of the brain such as multiple sclerosis (Park et al. 2005), it has not yet gained much attention regarding the immunological properties of lithium. In previous studies, we demonstrated that other mood stabilizing substances such as antipsychotics (Himmerich et al. 2011) and valproic acid (Himmerich et al. 2014) influence IL-17 production. Therefore, we included IL-17 in the present experiment to test the influence of lithium on IL-17 production under different conditions and in combination with widely prescribed antidepressants.

Methods and materials

Subjects

The sample comprised 15 healthy female and 15 healthy male subjects, between 22 and 52 years of age without a history of any psychiatric disorder. Exclusion criteria were current psychotropic medication, use of illegal drugs or regular alcohol consumption. All participants gave written informed consent. The study was approved by the Ethics Committee of the Medical Faculty, University of Leipzig.

Experimental procedure

The whole blood assay was performed as described previously (Kirchner et al. 1982; Seidel et al. 1996; Himmerich et al. 2010a; Himmerich et al. 2011). Blood was collected once from all subjects using a citratmonovette (Sarstedt, Nürtingen, Germany) and cultured in a whole blood assay within 1–2 h after blood collection. Cell concentration was adjusted at 3.5 × 109 cells/l using RPMI 1,640 medium (Biochrom, Berlin, Germany). Subsequently, 100 μl of this blood plus RPMI solution was introduced into a tube and mixed with 100 μl of pure psychopharmacological substance plus RPMI, resulting in a final cell concentration of 1.5 × 109 cells/l. The intended final concentrations of psychotropic drugs in this mixture were chosen according to the Association of Neuropsychopharmacology and Pharmacopsychiatry-Therapeutic Drug Monitoring (AGNP-TDM) Expert Group Consensus Guidelines: Therapeutic Drug Monitoring in Psychiatry (Baumann et al. 2004). We used the maximum therapeutic concentration (onefold concentration) for citalopram (130 ng/ml), escitalopram (130 ng/ml), mirtazapine (80 ng/ml), lithium carbonate (1.2 mmol/l), and combinations of lithium and citalopram (1.2 mmol/l lithium and 130 ng/ml citalopram), lithium and escitalopram (1.2 mmol/l lithium and 130 ng/ml escitalopram) and lithium and mirtazapine (1.2 mmol/l lithium and 80 ng/ml mirtazapine). We additionally applied the same drugs and combinations at twofold maximal therapeutic concentration. We refer to these as onefold and twofold concentrations. As a no drug control condition we used a charging stock with whole blood and one of the two immune stimulators, with no psychotropic drugs added.

The design tested the effects of each drug or combination on cytokine concentrations under three conditions of immune stimulation: PHA stimulated, OKT3/5C3 stimulated and unstimulated. We tested cytokine concentrations in 1,350 samples comprising 4 different psychotropic drugs plus 3 combinations of psychotropic drugs, each tested at one- and twofold concentrations, under 3 conditions of immune stimulation, plus 3 control charging stocks, for each of 30 patients.

Citalopram, lithium and mirtazapine were obtained from Sigma-Aldrich Laborchemikalien GmbH (Seelze, Germany), escitalopram from Lundbeck (Denmark).

For induction of cytokines, OKT3/5C3 at a final concentration of 100 ng/ml or PHA at a final concentration of 10 μg/ml were used. All tubes were covered and samples were incubated in an atmosphere of 5 % CO2 at 36 °C for 48 h. Cell-free supernatants were harvested after incubation and stored at −80 °C.

For quantification of cytokines IL-1ß, IL-2, IL-4, IL-6, IL-17 and TNF-α, bead array flow cytometry (FACSArray Bioanalyzer, BD Biosciences, Franklin Lakes, NJ, USA) was used. IL-22 was determined using a human IL-22 DuoSet Elisa (R&D Systems Europe, Abingdon,UK).

Statistical analysis

We compared eight levels of drug supplementation (no drug, lithium alone, each of the three antidepressants alone, each of the three antidepressants combined with lithium) at two levels of drug concentrations (onefold and twofold therapeutic concentration). The levels of cytokines under each drug or drug combination were compared to the no drug control as tabulated in Tables 1, 2 and 3. The comparisons were carried out separately for each immune stimulation condition. As described above, we used three conditions of immune stimulation (unstimulated, PHA stimulation, OKT3/5C3 stimulation).

Table 1 Median (M), first (Q1) and third (Q3) quartile cytokine concentrations (ng/ml) in unstimulated blood
Full size table
Table 2 Median (M), first (Q1) and third (Q3) quartile cytokine concentrations (ng/ml) in OKT3/5C3-stimulated blood. Control refers to OKT3/5C3-stimulated blood without drug
Full size table
Table 3 Median (M), first (Q1) and third (Q3) quartile cytokine concentrations (ng/ml) in PHA-stimulated blood
Full size table

In view of the non-normal distribution of results (Kolmogorov–Smirnov test: p < 0.05) we used non-parametric paired Wilcoxon tests for all between-group comparisons which provide suitable results also in the case of low to modest sample sizes. Due to the exploratory nature of this study, an uncorrected p value (p < 0.05) was considered significant.

To illustrate, we wanted to examine the influence of lithium alone at a concentration of 1.2 mmol/l on IL-1ß production. Therefore, we first compared IL-1ß levels in the unstimulated blood with lithium (see column “lithium, 1.2 mmol/l” of Table 1; median = 14.07 ng/ml) with the IL-1ß levels in the unstimulated blood without lithium (see “control” column of Table 1; median = 0.09 ng/ml). Second, we compared IL-1ß levels in the OKT3/5C3-stimulated blood with lithium (see column “lithium, 1.2 mmol/l” of Table 2; median = 19.29 ng/ml) with the IL-1ß levels in the OKT3/5C3-stimulated blood without lithium (see “control” column of Table 2; median = 0.00 ng/ml). Third, we compared IL-1ß levels in the PHA-stimulated blood with lithium (see column “lithium, 1.2 mmol/l” of Table 3; median = 11.39 ng/ml) with the IL-1ß levels in the PHA-stimulated blood without lithium (see “control” column of Table 3; median = 0.00 ng/ml). In this case, lithium led to a significant increase of IL-1ß levels across all three test conditions as is explicated in the results section of this article.

Results

Details of median, first (Q1) and third quartile (Q3) cytokine concentrations, in unstimulated, OKT3/5C3- and PHA-stimulated blood, with no drug, lithium and antidepressants, alone and in combination, are shown in Tables 1, 2 and 3.

In unstimulated blood (Table 1), IL-1ß levels were increased by lithium, alone and in combination with each antidepressant. However, IL-1ß concentrations did not increase under the influence of antidepressants (citalopram, escitalopram or mirtazapine) alone. IL-2 and IL-4 were significantly decreased by citalopram and mirtazapine at onefold concentration or in combination with lithium. However, lithium alone did not have any significant influence on IL-2 or IL-4. IL-6 increased under all applied drugs in all combinations and concentrations tested. In contrast, IL-17 levels did not show any response to drugs. Lithium increased TNF-α concentrations, alone or in combination with citalopram, escitalopram or mirtazapine at onefold concentration.

In OKT3/5C3-stimulated blood (Table 2), all applied drugs increased IL-1ß and IL-6 levels. IL-2 and IL-4 concentrations were not consistently affected by any of the applied substances or combinations. IL-17 levels increased under nearly all applied substances and combinations under stimulation with OKT3/5C3. As in the unstimulated blood, lithium applied at onefold concentration, alone or in combination with antidepressant, led to an increase in TNF-α concentration.

In PHA-stimulated blood (Table 3), only lithium, alone and in combination, increased IL-1ß and TNF-α concentrations. Lithium also significantly increased IL-2, IL-6 and IL-17 concentrations.

Table 4 shows differences in cytokine levels with and without lithium, combined with citalopram, escitalopram or mirtazapine, under the three immune stimulation conditions. It can be seen that lithium increases IL-1ß (see also Fig. 1), IL-6 (see also Fig. 2) and TNF-α (see also Fig. 3), when added to these antidepressants. It additionally increased IL-17 concentrations in the OKT3/5C3-stimulated blood when combined with citalopram.

Table 4 Cytokine levels under antidepressant alone versus antidepressant combined with lithium
Full size table
Fig. 1
figure 1

IL-1ß concentrations in unstimulated blood, comparing no drug control, lithium or antidepressant alone, and combination of each antidepressant with lithium, all at onefold concentration. Values are depicted as mean ± SEM. Values are depicted as mean ± SEM. Asterisk significant difference, uncorrected p < 0.05, non-parametric Wilcoxon test

Full size image
Fig. 2
figure 2

IL-6 concentrations in unstimulated blood, comparing no drug control, lithium or antidepressant alone, and combination of each antidepressant with lithium, all at onefold concentration. Values are depicted as mean ± SEM

Full size image
Fig. 3
figure 3

TNF-α concentrations in unstimulated blood, comparing no drug control, lithium or antidepressant alone, and combination of each antidepressant with lithium, all at onefold concentration. Values are depicted as mean ± SEM

Full size image

In summary, lithium was the most cytokine-modulating drug of those tested. Lithium consistently increased IL-1ß, IL-6 and TNF-α in unstimulated and stimulated blood, alone and in combination with an antidepressant.

Discussion

The main finding of this in vitro study is that lithium has a consistent activating effect on pro-inflammatory cytokine production under different conditions of immunological stimulation and in combination with different antidepressants. The latter has not been reported previously and is the novel finding of this in vitro study.

It is difficult to compare these results with previous studies, because the literature regarding the modulation of cytokine production by lithium reveals conflicting results. Some studies describe an increase of pro-inflammatory cytokines (Himmerich et al. 2005; Merendino et al. 1994; Maes et al. 1999; Beyaert et al. 1991, 1992), others report that lithium decreases (Himmerich et al. 2013; Arena et al. 1997) or does not affect production of pro-inflammatory cytokines (Himmerich et al. 2014). Interpreting the results of these studies and experiments, one has to take into account that various in vivo or in vitro approaches with disparate conditions have been used. For example, in vitro studies have used a variety of stimulants such as lipopolysaccharide (LPS), PHA, OKT3/5C3 or toxic shock syndrome toxin-1 (TSST-1) (Maes et al. 1999; Himmerich et al. 2013, 2014). However, due to the fact that in vivo studies in animals (Beyaert et al. 1992) and humans (Himmerich et al. 2005) have mainly found that lithium increases cytokine production, we think that we managed to simulate the peripheral effects of lithium using the present approach.

The results may be compared with two previous in vitro studies by the authors, which included lithium and immune stimulators.

Himmerich et al. (2013) investigated the effects of mood stabilizers, including lithium, and antiepileptic drugs on cytokine production in vitro, using TSST-1 as the immunological stimulant. Lithium was found to decrease IL-1ß and TNF-α production (Himmerich et al. 2013). TSST-1 is a staphylococcal-secreted exotoxin which leads to nonspecific binding of major histocompatibility complex class II with T cell receptors, resulting in T as well as B cell activation, stimulation of mononuclear cells and increased cytokine production (Kum et al. 1993; Dinges et al. 2000). Thus, TSST-1 is a reliable but supraphysiological immunological stimulator, which may be too strong to simulate blood cells in a clinically relevant manner. TSST-1 stimulation may also have produced a ceiling effect that prevented lithium inducing pro-inflammatory cytokine production any further. Therefore, the results of this previous study may not be indicative of the effects of lithium treatment in vivo.

Himmerich et al. (2014) investigated the impact of mood stabilizers, including lithium and antiepileptic drugs on cytokine production using OKT3/5C3-stimulated blood of 14 healthy female subjects. Lithium in the applied concentrations led to a nominal but not statistically significant increase of IL-1ß, IL-2 and TNF-α production (Himmerich et al. 2014). Therefore, the present study increased the sample size to 30 healthy participants and compared an unstimulated and two stimulated experimental conditions, more closely approximating the natural clinically relevant state.

However, it is noted that these in vitro effects might be indicative of peripheral effects in blood that differ from the effects of lithium in the central nervous system.

In the present study we used lithium in a maximum therapeutic onefold (1.2 mmol/l) and twofold (2.4 mmol/l) concentration (Baumann et al. 2004). Our finding that lithium activates pro-inflammatory cytokines, when given in doses usually chosen for their antimanic effect, supports the hypotheses that an increased amount of cytokine signalling may contribute to the therapeutic dampening of mood, and the affective flattening effect of lithium.

Somewhat in contrast with these results, activation of pro-inflammatory cytokines have been suggested as a biomarker for depression (Lichtblau et al. 2013), since pro-inflammatory cytokines have been shown to be elevated in depressed patients (Himmerich et al. 2008; Schmidt et al. 2014) and to decrease during successful antidepressant therapy (Himmerich et al. 2010b). Moreover, blockers of these cytokines, for example the TNF-α blocker etanercept, demonstrate antidepressant effects in animals (Krügel et al. 2013) and humans (Tyring et al. 2006). The fact that lithium treatment of depressed patients increases pro-inflammatory cytokine plasma levels (Himmerich et al. 2005) and activates pro-inflammatory cytokine production in vitro questions the general applicability of these cytokines as biomarkers for depression, because lithium augmentation has a reliable antidepressant effect even in therapy resistant depressed patients (Bschor et al. 2003) despite its activating effect on cytokine production. The finding that citalopram also increased some pro-inflammatory cytokines, yet is an effective antidepressant, additionally points to caution in interpreting. A further consideration is that the value of cytokines as depressive biomarkers might also differ between patients with unipolar depression and bipolar disorder.

The inflammatory effects of lithium may be relevant for its specific immunologically induced side effects. As noted, lithium is frequently associated with cutaneous side effects such as psoriasis (Paquet et al. 2005). It is well established that psoriasis results from a local increase in cytokine or growth factor production such as IL-1ß, IL-6 and TNF-α as well as activated inflammatory cells (Grossman et al. 1989; Gomi et al. 1991). Activation of these cytokines by lithium may, therefore, be a good explanation for the appearance or exacerbation of psoriasis during lithium therapy. This view has been substantiated in animal studies using lithium to promote psoriatic symptoms while measuring local cytokine concentrations (Beyaert et al. 1992).

Apart from the skin, lithium’s side effects often find their expression in the thyroid gland. Recently, the thyroid side effects of lithium therapy regained substantial scientific interest following a meta-analysis of the toxicity of lithium (McKnight et al. 2012), leading to an intense scientific discussion regarding its toxicity and specifically its toxicity for the thyroid gland (Wells et al. 2012; Müller-Oerlinghausen et al. 2012; Bschor and Bauer 2013). Lithium affects thyroid functioning through multiple mechanisms. These include decreasing thyroid hormone synthesis and release, peripheral deiodination of thyroxine, and an increased propensity to thyroid autoimmunity in susceptible individuals leading to Hashimoto thyreoiditis and hypothyroidism (Kibirige et al. 2013). Therefore, hypothyroidism is a clinically relevant thyroid abnormality in patients on long-term lithium therapy. Possible autoimmune mechanisms include that lithium augments the activity of B lymphocytes leading amongst other things to increased IgG and IgM production, reduction of the ratio of circulating suppressor to cytotoxic T cells (Wilson et al. 1989; Kibirige et al. 2013) and its effect of inducing pro-inflammatory cytokine production. The latter assumption is substantiated by studies which describe the role of IL-1ß (Paolieri et al. 1999), IL-6 (Simons et al. 1998) and TNF-α (Tarhan et al. 2013) in autoimmune thyroiditis. Moreover, anti-TNF-α-medication seems to be associated with a decreased risk for autoimmune thyroiditis (Tarhan et al. 2013).

As mentioned in the introduction, lithium’s pro-inflammatory properties may also have benefits for those patients who suffer from a loss of function within the immune system. For example, lithium has been described as having immunoenhancing properties, and improving neuropsychological performance in HIV-infected patients (Sztein et al. 1987; Crews et al. 2009). Moreover, the published literature suggests that lithium may exert these effects by inhibiting neuronal glycogen synthase kinase (GSK)3ß, by granulopoietic actions, by regulatory effects on cytokines, and by increased synthesis of neuroprotective proteins in the human brain (Letendre et al. 2006; Harvey et al. 2002). Moreover, case reports in acquired immunodeficiency syndrome (AIDS) patients, as well as animal studies using related immunodeficiency viruses, provide support for a possible therapeutic role for lithium in the treatment of HIV infection and specifically for the therapy of AIDS-related neurological deficits (Harvey et al. 2002).

SLE is a chronic autoimmune disease accompanied by the activation and proliferation of T cells and B cells, involving increased activity of the GSK3ß signalling pathway (Tang et al. 2009). Decreased production of IL-2 by T cells has also been found in patients with SLE, and it was shown that incubation in vitro of T cells with lithium augmented IL-2 production (Kucharz et al. 1993). In the PHA-stimulated blood, we could replicate the finding of an increased IL-2 production under the influence of lithium (see Table 3). However, we did not consistently find this effect through all applied stimulation methods and the potential relevance of this effect to SLE requires clarification.

In summary, the activation of pro-inflammatory cytokines by lithium may contribute to the intended therapeutic effect of dampening mood in manic patients, but may also cause some of the side effects of lithium, seen in patients suffering from certain autoimmune diseases such as autoimmune thyroiditis and psoriasis. On the other hand, this immunomodulatory property may be beneficial for patients suffering certain deficits within the immune systems. Thus, patients with HIV infection or SLE could benefit from lithium therapy. In patients with bipolar disorder, the existence of one of these diseases should influence clinical reasoning for or against the use of lithium.

A novel finding of our study was the increase in IL-17 levels under nearly all applied substances and combinations, under stimulation with OKT3/5C3 (see Table 2). When the blood was stimulated with PHA, IL-17 levels were similarly raised with lithium alone and the combinations of 1.2 mmol/l lithium plus 130 ng/ml citalopram, 2.4 mmol/l lithium plus 260 ng/ml citalopram and both applied doses of lithium (see Table 3). As IL-17 also promotes immune responses, our finding fits very well with the induction of other pro-inflammatory cytokines such as IL-1ß, IL-2, IL-6 and TNF-α by lithium. As antipsychotics, which are also used as antimanic agents, have also been shown to increase IL-17 levels, IL-17 may be an interesting novel research target for bipolar disorder.

To our knowledge, this is the first study investigating differences in cytokine production between lithium alone and in combination with specific antidepressants. Tables 1, 2, 3, 4 and Figs. 1, 2, 3 show medians and means of the cytokine concentrations under the influence of lithium alone or in combination with a specific antidepressant. For example, in the unstimulated blood, the median of the TNF-α concentration was 0.21 ng/ml in the control condition without drug, 1.67 ng/ml under the influence of 1.2 mmol/l lithium alone and 3.67 ng/ml under the combination of 1.2 mmol/l lithium with 80 ng/ml mirtazapine (see Table 1). The means of the TNF-α concentration (see Fig. 3) for the control condition, lithium alone and lithium in combination with mirtazapine draw a similar picture. This is indicative of additive effects of psychopharmacological combinations on cytokine production. Given that in vivo studies have shown that plasma TNF-α increases in patients treated with mirtazapine (Kraus et al. 2002) and lithium (Himmerich et al. 2005), our results may be clinically relevant. However, to determine this an in vivo study of cytokine plasma levels in combination therapy with antidepressant and lithium in patients would be required. Such a clinical study would offer the opportunity to explore the relevance of cytokine levels to therapeutic as well as side effects.

The present in vitro study only included healthy subjects and does not permit an uncritical generalization to patients suffering from unipolar or bipolar affective disorders. Another limitation of our study is that the reported effects shown in this in vitro experiment may not be therapeutically relevant due to the relatively high serum concentrations studied. Therefore, it would be advisable for further studies to include lower drug doses. However, the onefold dose is in principle therapeutically relevant for patients suffering from manic episodes.

Apart from IL-1ß, IL-2, IL-4, IL-6, IL-17 and TNF-α, several other cytokines such as IL-10, interferon-γ (IFN-γ), transforming growth factor (TGF)-β, erythropoietin (EPO), cytokine receptors such as the TNF-α receptors, TNF-R p55 and TNF-R p75 and cytokine receptor antagonists such as the IL-1 receptor antagonist (IL-1ra) have been implicated in the pathophysiology of mood disorders (Lichtblau et al. 2013). Therefore, we may have missed effects of lithium on any of these important cytokines. We did not measure markers of cell death in the reported experiments. Therefore, we can not rule out that cytotoxicity may have contributed to the observed effects of the tested drugs on cytokine production. In the statistical analysis we have reported all significant effects at a p level of less than 0.05. We did not use a correction for multiple tests such as a Bonferroni correction in view of the exploratory nature of the study.

In conclusion, we found that lithium activates the production of cytokines that promote inflammatory processes such as IL-1ß, IL-2, IL-6, IL-17 and TNF-α. These lithium-induced changes of the cytokine system might contribute to the therapeutic effects, but also the inflammatory side effects of lithium, as seen in autoimmune thyroiditis or psoriasis. The inflammatory properties of lithium may also be beneficial for patients with certain immunological disorders such as HIV infection and SLE. Several questions remain unexplained such as conflicting results from studies using different methods, contradictory effects between different stimulation methods, and the possible role of IL-17 in bipolar disorder and treatment with lithium.