Administration of low-dose endotoxin to blinded healthy volunteers provoked symptoms of depression and anxiety and cognitive impairment in the hours following administration, and these mood changes were correlated with increased circulating concentrations of the cytokines interleukin 6 (IL6), tumour necrosis factor alpha (TNFα) and IL1 receptor antagonist (IL-1ra) (Reichenberg et al., 2001). Similarly, typhoid vaccination induced negative changes in mood that were accompanied by and correlated with increased circulating concentrations of IL6 (Harrison et al., 2009). Interestingly, vaccination-associated mood deterioration correlated with enhanced activity within subgenual anterior cingulate cortex (sgACC) during emotional face processing assessed using functional magnetic resonance imaging (fMRI). The fact that inflammation-related deterioration in mood is associated with increased sgACC activity is of significance considering the large body of data implicating sgACC hyperactivity in the pathophysiology of depression, and as an antidepressant target (Drevets et al., 2008).

Inflammatory stimuli such as the Toll-like receptor-4 (TLR4) agonist bacterial lipopolysaccharide (LPS), the TLR3 agonist and viral mimetic polyinosinic–polycytidylic acid (Poly I:C), or infection with Bacillus Calmette–Guérin (BCG), can provoke symptoms of anxiety and depression distinguishable from generalized sickness behaviour on a temporal profile in animal models (O’Connor et al., 2008b, 2009; Gibney et al., 2012). Specifically, it has been demonstrated in these animal models that anxiety and depressive-like behaviours are still evident when acute sickness has resolved. In the case of BCG and experimental autoimmune encephalomyelitis (EAE), the development of depressive behaviours lasts for several weeks and is associated with a persistent increase in circulating interferon gamma (IFNγ) and TNFα levels. Evidence supporting the role of cytokines in the brain responsible for depressive symptoms is reviewed elsewhere (Dunn et al., 2005; Dantzer et al., 2008). IL1β and TNFα are considered the main cytokines as there are numerous reports that systemic or central administration of these cytokines to laboratory rats and mice induces sickness behaviour in a dose- and time-dependent manner. It has been shown that IL1β can instigate anhedonia in laboratory rodents independent of effects on anorexia. Other cytokines too are considered to play a role, including IL6 which contributes to the expression of brain IL1β and TNFα and may play a role in LPS-induced hippocampal-mediated cognitive impairment (Sparkman et al., 2006). In other reports IL2, but not IL1β or IL6, has been reported to provoke long-lasting anhedonic effects in mice and rats (reviewed by Dunn et al., 2005).

By contrast to pro-inflammatory mediators, central administration of the anti-inflammatory cytokine IL10 attenuates the behavioural signs of sickness induced by centrally injected LPS (Bluthe et al., 1999). Some growth factors such as insulin-like growth factor-1 (IGF1) also antagonize pro-inflammatory cytokines in the brain and attenuate sickness behaviour following central administration of LPS. The interplay between pro- and anti-inflammatory mediators and associated signalling is suggestive of a balance between cytokines in the regulation of the sickness and depressive-like response to immune stimuli. Furthermore, it is of interest that pro-inflammatory cytokines including TNFα and IL1β can provoke a reduction in IGF1 sensitivity and promote resistance to this growth factor (reviewed by O’Connor et al., 2008a).

Increased expression of IL1β and TNFα observed in the circumventricular organs and blood–brain barrier is associated with the initial wave of LPS-induced sickness. The secondary wave of depression-like behaviour, which occurs 12 to 24 h later, may be associated with increased expression of cytokines within parenchymal regions. Mapping of neuronal activation in response to LPS administration has been carried out, and the peak of sickness behaviour is associated with increased expression of the cellular activation marker and immediate early gene product, c-fos, in the paraventricular nucleus and the bed nucleus of the stria terminalis (BNST), areas involved in endocrine and autonomic components of LPS-induced sickness. By contrast, immunostaining for FosB and its truncated splice variant δFosB, both of which have a longer half-life than c-fos and accumulate during repeated or long-lasting stimulation, are increased in several hypothalamic nuclei and extend to the amygdala and hippocampus. Such activation patterns point towards the involvement of these structures in cytokine-induced depression (reviewed by Dantzer et al., 2008).

By far, the largest body of data relating to the ability of cytokines to induce changes in mood stems from studies where cytokine immunotherapy for viral hepatitis or various forms of cancer provokes severe psychological disturbances, including depression. Numerous studies have reported that previously psychiatrically healthy individuals treated with high doses of exogenous cytokine IL2 or IFNα develop depressive-like symptoms such as depressed mood, increased somatic symptoms, stress reactions and cognitive impairment (Capuron et al., 2003, 2004; Valentine and Meyers, 2005). Literature on animal studies supports the view that peripherally administered IFNα can directly impact brain function, in that a robust induction of IFNα-inducible genes was observed in the CNS following systemic administration of IFNα to mice (Wang et al., 2008).

It is not entirely clear why only a proportion of individuals, typically less than 50%, that receive cytokine immunotherapy become depressed. However, it has been suggested that psychological stress may sensitize to the neurochemical and behavioural actions of cytokines, and therefore individuals with an anxious or stress-prone phenotype may be more susceptible to developing psychiatric sequelae in response to cytokine administration. In support of this idea, it was reported that patients with a higher cortisol response (stress response) to the initial injection of IFNα displayed a greater propensity to develop depression following treatment (Capuron et al., 2003). Further support for this notion stems from a pre-clinical study demonstrating that mice exposed to psychosocial stress showed exaggerated central monoamine changes, hypothalamic–pituitary–adrenal (HPA) axis reactivity and sickness behaviour to IFNα treatment (Anisman et al., 2007).

When one considers the literature on cytokines as a trigger for depressive illness, it is indisputable that treatment with IL2 or IFNα can induce depression; however, one must remember that the doses of cytokines administered to patients in these studies far exceed physiological concentrations. This is a factor that should be considered when implicating elevated endogenous cytokine secretion as a causal factor in the development of depressive symptoms. Thus, one cannot simply regard the psychiatric and biological sequelae that occur following exogenous administration of cytokines such as IFNα or IL2 as being akin to the biology of idiopathic depression where no obvious exogenous agent is driving changes in the psychiatric state.

Depressive symptoms appear to be due to the biochemical changes induced by cytokine treatment rather than psychological reactions to the illness for which the agents are being administered. For instance, evidence indicates that treatment with IFNα or IL2 induces metabolism of the essential amino acid tryptophan to kynurenine, which has the potential to limit tryptophan availability for serotonin synthesis (Capuron et al., 2002; Widner et al., 2002). It has been demonstrated that peripheral administration of IFNα to patients increases kynurenine concentrations in the plasma and also in the cerebrospinal fluid (CSF) (Raison et al., 2010). Furthermore, it was observed that the extent of tryptophan depletion in patients treated with IFNα was correlated with the incidence of depression (Capuron et al., 2002; Raison et al., 2010). However, despite the original hypothesis proposing that inflammatory cytokines could deplete tryptophan availability for serotonin synthesis, both clinical (Raison et al., 2010) and pre-clinical studies in animal models (O’Connor et al., 2009; Gibney et al., 2012) do not support this hypothesis. A number of studies have consistently reported increased (as opposed to decreased) tryptophan availability in the CNS following administration of LPS or other inflammogens. Moreover, there is no evidence of central serotonin depletion following administration of LPS or inflammatory cytokines to animals; in fact, an increase in serotonin release and metabolism coupled with increased activity of the rate-limiting enzyme for serotonin biosynthesis, tryptophan hydroxylase, have been observed in rats following a systemic inflammatory challenge with LPS (Nolan et al., 2000; Dunn et al., 2005). Clinical studies to date also argue against a kynurenine pathway–mediated depletion of serotonin synthesis in depressed patients (Wichers et al., 2005; Hughes et al., 2012). Of note is the fact that kynurenine itself has depressogenic effects in the forced-swimming and tail suspension tests, and it has therefore been proposed that kynurenine, or more likely one of its neuroactive pathway metabolites, mediates depressogenic behaviour in these animal models (O’Connor et al., 2009). Similarly, a role for neuroactive kynurenine metabolites has been implicated as mediators of IFNα-induced depression in humans (Wichers et al., 2005; Raison et al., 2010). It has been suggested that the action of kynurenine metabolites on the glutamatergic system may be involved in producing depressive symptomatology (Müller and Schwarz, 2007).

Some studies have demonstrated that inflammatory cytokines, including IL1β, TNFα and IFNα, increase expression of the serotonin transporter (SERT) and serotonin re-uptake in vitro (Tsao et al., 2008), and that a systemic inflammatory challenge with bacterial LPS increases SERT expression in rodent brain (Zhu et al., 2010). A single systemic injection of Poly I:C induces a persistent increase in IFNα expression in the CNS accompanied by increased SERT expression and reduced extracellular serotonin as quantified by in vivo microdialysis in the prefrontal cortex of rats (Katafuchi et al., 2005). Raised inflammatory cytokine expression is correlated with increased SERT expression on circulating leukocytes of depressed patients, and increased IFNα and SERT expression in depressed patients is restored to normal following chronic treatment with the selective serotonin reuptake inhibitor and antidepressant fluoxetine (Tsao et al., 2006). Moreover, we have recently demonstrated that activation of the BV2 mouse microglial cell line with LPS increases SERT mRNA expression in these cells. These findings suggest that central microglia may play a role in sequestering serotonin, particularly under inflammatory conditions. This finding has all the more significance as we have also demonstrated that microglia express monoamine oxidase A (MAO_A), the enzyme responsible for the metabolism of serotonin (Fagan et al., 2013).

Increased serum anti-serotonin antibody titres (Schott et al., 2003; Maes et al., 2012) and serum antibodies against the serotonin 5-HT_1A receptor (Tanaka et al., 2003) have been reported in depressed patients. The presence of such antibodies was associated with features of immunological activation indicated by increased plasma TNFα and IL1 concentrations. Autoimmune reactions to serotonin may play a role in the pathophysiology of depression as significant association between autoimmune activity to serotonin and the number of previous depressive episodes has been reported (Maes et al., 2012).

In addition to the serotonergic system as a biological target of cytokines, studies also indicate that inflammation can negatively impact brain-derived neurotrophic factor (BDNF). Specifically, a systemic inflammatory challenge with bacterial LPS reduces expression of BDNF in rat brain (Guan and Fang, 2006), and intra-hippocampal LPS administration has been shown to inhibit expression of BDNF and also its receptor, Trk_B (Tanaka et al., 2006). More recently, we demonstrated that systemic treatment with Poly I:C inhibited expression of both BDNF and Trk_B in hippocampus and cortex and that this occurred following a robust expression of the inflammatory cytokines IL1β, TNFα and IL6 in these brain regions (Gibney et al., 2012). IL1β is thought to be one of the main cytokines to compromise the BDNF pathway. Administration of IL1β has been shown to reduce hippocampal BDNF mRNA expression, and IL1ra has been shown to reverse stress-related reductions in BDNF (Barrientos et al., 2003). Alterations in BDNF expression and function are central to the ‘neurotrophin hypothesis of depression’, which suggests that reduced BDNF can lead to reduced neuronal protection and neurogenesis, and ultimately to the development of depressive symptoms (Martinowich et al., 2007). These are important findings considering the role that BDNF plays in driving neurogenesis, a process implicated in the pathogenesis of depression and in the therapeutic response to antidepressants.

Assessments of various psychological parameters that accompany the onset or recovery from infection consistently report depression as a psychological disturbance (Dantzer et al., 2008). Indeed, infection with Borna disease virus has been suggested as a risk for depressive illness in humans (Bode and Ludwig, 2003). There is also evidence that co-morbidity exists between a number of inflammatory disease states and depressive illness. For instance, depression and anxiety disorders represent a significant co-morbidity with inflammatory bowel disease (IBD) which adversely affects the quality of life of patients. In addition, individuals with IBD experience rates of depression that are triple those of the general population (Graff et al., 2009). Moreover, considering the emerging literature on the brain–gut axis, it is likely that stress-related psychiatric disorders such as anxiety and depression could exacerbate the severity of or slow recovery from the clinical symptoms of IBD (Hollander, 2003). Systemic lupus erythematous (SLE) is a progressive autoimmune disorder and is associated with chronic stimulation of various components of the immune system. Compared to control subjects, tryptophan was decreased and kynurenine was significantly increased in patients with SLE. The study also indicates that tryptophan depletion may be associated with neurologic and psychiatric disturbances in patients suffering from SLE (Widner et al., 1999; and see Chapter 10). Clinically significant depression can affect up to 50% of patients with multiple sclerosis over the course of their lifetime. In particular, an association between depression and structural brain abnormalities, including those derived from diffusion tensor imaging, was noted. Results from randomized controlled trials of antidepressant medication, cognitive behaviour therapy and mindfulness therapy reveal that depression in these patients can be successfully treated (Feinstein, 2011). Depression is 2–3 times more common in patients with rheumatoid arthritis when compared to the general population, and the degree of depression is associated with the level of pain experienced (Walker et al., 2011). It may be the case with immune-mediated inflammatory disease that depressive symptoms relate to factors such as changes in quality of life and the experience of pain and distress associated with the disease. Evidence is growing, however, for cytokines as a trigger for changes in affect as opposed to a consequence of incapacitation associated with the primary medical condition (reviewed by Gibney and Drexhage, 2013).

Depression is associated with increased circulating concentrations of pro-inflammatory cytokines, soluble cytokine receptors, chemokines and acute-phase proteins. Moreover, in the cases of IL6, IL1, TNFα, soluble IL2 receptor (sIL2R) and C-reactive protein (CRP), original findings have been supported by recent meta-analyses (Howren et al., 2009; Dowlati et al., 2010; Liu et al., 2012). An interesting feature of a study by Simon and co-workers (2007) is that serum concentrations of anti-inflammatory cytokines such as IL10 and IL4 were elevated in addition to pro-inflammatory cytokines, a feature that is shared with many of the studies from Maes and colleagues, who reported increased concentrations of anti-inflammatory cytokines such as IL1ra in depressive patients in parallel with increased concentrations of pro-inflammatory cytokines (Maes et al., 1997). Another interesting feature of a study by Simen and co-workers (2006) is that, despite a robust inflammatory profile observed in depressed patients, concentrations of the pro-inflammatory cytokine TNFα were not significantly altered between controls and depressives. This was somewhat surprising give the role that TNFα has in inducing a variety of molecules in the cytokine network, and also given the attention that TNFα has received as a potential mediator of depressive symptomatology and in response to antidepressant treatment (Simen et al., 2006; Powell et al., 2012; Raison et al., 2013). Similarly, in a recent study we failed to observe a significant increase in circulating TNFα concentrations in a group of treatment-resistant depressed patients, and whilst we observed a significant increase in plasma IL6 and IFNγ in depressives relative to controls, these increases were very modest (Hughes et al., 2012) compared to those reported in other studies (Simon et al., 2007; Cizza et al., 2008).

It should be noted that whilst statistically significant increased cytokine concentrations are consistently observed in the serum or plasma of depressed patients, the magnitude of the changes observed between patients and control subjects are very small, and therefore one has to question the ability of such small increases to impact brain function. Similarly, while the increase in CRP in depressed patients is statistically significant, it generally falls in the normal range (below 6 mg/L) and would not indicate the presence of overt inflammation per se (Hughes et al., 2012). Nonetheless, recent studies indicate that even mildly elevated IL6 and CRP concentrations independently predict the subsequent development of depression over a decade or more, even in individuals with no history of depression at the time of sampling (Gimeno et al., 2009; Pasco et al., 2010). Whilst further study is required to provide a mechanistic basis for these findings, it is noteworthy that we have recently observed that elevated circulating IL6 concentrations are associated with reduced hippocampal volumes in major depressive disorder (Frodl et al., 2012). These data may suggest that elevated IL6 has the potential to impact hippocampal neuroplasticity.

A high incidence of depression is evident in individuals with autoimmune diseases, including multiple sclerosis, SLE and RA where inflammatory cytokines are overexpressed (reviewed by Gibney and Drexhage, 2013), pointing to an association and a role for the adaptive immune system in depression. Reduced circulating T cell numbers and proliferation of peripheral blood mononuclear cells (PBMCs) in response to T cell mitogens has been reported in depressed patients. Moreover, depressed patients are reported to show fewer resting CD3⁺/CD25⁻ T cells with significantly more CD20⁺/CD5⁺ B cells when compared to healthy controls. Studies employing flow cytometry analysis have revealed that depressed patients have an increased number and percentage of T cells bearing activation markers, such as CD25 and HLA-DR, indicating the presence of acquired immunological activation in these patients (Irwin and Miller, 2007; Maes, 2011). The mechanisms of T cell alterations in depression are proposed to involve apoptosis, tryptophan depletion and changes in glucocorticoid and adrenergic receptor sensitivity (Miller, 2010).

By measuring stimulated cytokine production from PBMCs or diluted whole blood, investigators have examined the functional status of cytokine-producing cells in depressed patients. Mixed results have been observed using this approach. Seidel et al. (1995) reported a significant increase in mitogen-stimulated IFNγ and sIL2R production from PBMC cultures and elevated serum APP concentrations in depressed patients, which were maximal during the acute phase of the illness, and returned to control levels over a 6-week hospitalization period during which time a concomitant decrease in depression (HAMD) scores was apparent. However, in contrast to the studies conducted by Maes and colleagues (1993, 1995) that reported significant elevations in stimulated IL1β and IL6 production from PBMCs in depressed patients, the study by Seidel and co-workers (1995) reported only a slight but non-significant increase in mitogen-stimulated PBMC production of these cytokines in depressed patients. In stark contrast to the findings already outlined, Weizman and co-workers (1994) reported that IL1β, IL2 and IL3 production from mitogen-stimulated PBMC cultures was significantly reduced in depressed patients, when compared to age- and sex-matched controls.

Whilst the studies reporting increased stimulated cytokine production tally with the findings of increased plasma cytokines in depressed patients, the studies reporting reduced cytokine responses to immune stimulants tally with reports of reduced proliferative responses to immune stimulants (mitogens) which were published in the 1980s as one of the first pieces of evidence of an aberrant immune system in depression. It has been proposed that the hypo-responsiveness of PBMCs may be accounted for by the increase in circulating cytokines which suppress the ability of cells to respond to stimulation (see Maes et al., 2012) and that the inflammatory response may be related to impaired function of lymphocytes in depression through direct effects of cytokines on signalling through T cells (Blume et al., 2012; Haroon et al., 2012).

Levine and co-workers (1999) reported increased CSF IL1β concentrations in a group of depressives versus controls and reported a positive correlation between IL1β concentrations and the severity of depression. A more recent study reported increased CSF concentrations of IL6 in depressed patients. Patients who performed violent suicide attempts displayed the highest IL6 concentrations. IL6 and TNFα correlated significantly with the serotonergic metabolite 5-hydroxyindole acetic acid (5-HIAA) and the dopaminergic metabolite homovanillic acid (HVA) in CSF, but not the noradrenergic metabolite methoxyhydroxyphenylglycol (MHPG). Cytokine levels in plasma and CSF were not associated, and patients with increased blood–brain barrier permeability did not exhibit elevated cytokine levels. Thus, a role for CSF IL6 has been proposed in the symptomatology of suicidal behaviour, possibly through mechanisms involving alterations of dopamine and serotonin metabolism (Lindqvist et al., 2009). In a separate investigation, CSF levels of IL1, IL6 and TNFα were reported to be significantly correlated with depression severity in support of the presence of central inflammatory activation in depressed patients (Martinez et al. 2012).

Increased expression of IFNα and its receptor IFNα/βR1 were observed in post-mortem dorsolateral prefrontal cortex tissue from major depressives relative to a group of matched controls (Kang et al., 2007). This finding is of significance considering that interferons have been associated with major depressive disorder and in particular with depression associated with multiple sclerosis. Increased expression of HLA-DR has been reported in the hippocampus and prefrontal cortex of depressed patients. More recently, increased quinolinic acid, a tryptophan metabolite and end product of the kynurenine pathway, has been detected in ramified microglia within sub-regions of the anterior cingulate cortex of severely depressed patients (Steiner et al.