Neuroimmunology of depression

Abstract

Depression is one of the leading causes of disability worldwide and a major contributor to the global burden of disease, yet the cellular and molecular etiology of depression remain largely unknown. Major Depressive Disorder (MDD) is associated with a variety of chronic physical inflammatory and autoimmune disorders, and mood disorders may act synergistically with other medical disorders to worsen patient outcomes. Here, we outline the neuroimmune complement, explore the evidence for altered immune system function in MDD, and present some of the potential mechanisms by which immune cells and molecules may drive the onset and course of MDD. These include pro-inflammatory signaling, alterations in the hypothalamic-pituitary-adrenal axis, dysregulation of the serotonergic and noradrenergic neurotransmitter systems, neuroinflammation, and meningeal immune dysfunction. Finally, we discuss the interactions between current antidepressants and the immune system and propose the possibility of immunomodulatory drugs as potential novel antidepressant treatments.

1. Introduction

More than 264 million people worldwide suffer from depression (James et al., 2018). For nearly 3 decades, depressive disorders have been one of the leading causes of disability worldwide and are a major contributor to the global burden of diseases (Shim et al., 2012). However, only half of workers with MDD receive timely treatment, and fewer than half of treated patients receive therapy consistent with published treatment guidelines (Kessler, Merikangas, & Wang, 2008). Depression is also the most common psychiatric disorder diagnosed in people who die by suicide (Hawton, Casañas i Comabella, Haw, & Saunders, 2013), another major health problem responsible for about 1.4% of all worldwide deaths (Brådvik, 2018). In fact, the severity of depression is a significant correlate of suicidality in both women and men (Hawton et al., 2013). While current antidepressants help many individuals and save lives, they are not effective for 30–50% of patients. In these patients, depression is categorized as “treatment-resistant,” and novel treatment options are a critical need.

The National Institute of Mental Health defines depression (major depressive disorder or clinical depression) as a common but serious mood disorder that involves some part of a constellation of persistent symptoms including sadness, loss of interest or pleasure in hobbies and activities, fatigue, impaired cognition or memory, altered sleep or appetite, suicidality, and pain (American Psychiatric Association, 2013), though the specific symptoms and their severity and frequency will vary from individual to individual. Moreover, the most effective treatments also vary by individual, and about 10–30% of patients do not improve or show partial improvement with current treatments (Al-Harbi, 2012).

1.1. Disparities in depression prevalence and treatment

The prevalence of major depression is 5.5% in women but only 3.2% in men (Whiteford et al., 2013). A similar ratio is consistently observed across many developed countries suggesting that the differential risk stems from biological sex and not simply culture, race, diet, education and other social or economic factors (Rai, Zitko, Jones, Lynch, & Araya, 2013). Even in countries where women have a markedly lower socioeconomic status than men, there is no clear evidence of a change in the rate of depression (Rai et al., 2013). The incidence of depression in girls remains the same as boys until puberty, and young women are the most at risk group for development of depression (Whiteford et al., 2013). In fact, twice as many young women as men develop depression, but this number decreases with age (Patten et al., 2006).

The reason for this disparity is not completely established. Exploring the mechanisms underlying this disparity in animal models has been challenging, in part because many preclinical behavior models of depression rely of social aggression models that work in male mice but are difficult to recreate in females mice (Berton et al., 2006; Sial, Warren, Alcantara, Parise, & Bolaños-Guzmán, 2016). However, sex differences in key depression-associated brain regions have been observed. For example, sex differences in hippocampal spine morphology and response to stressful events exist in rodent models (Shors, Chua, & Falduto, 2001). Additionally, there are sex differences in hippocampal long-term potentiation and hippocampus-dependent contextual learning (Maren, De Oca, & Fanselow, 1994). More recently, the mouse subchronic variable stress (SCVS) paradigm (Hodes et al., 2015) has been developed to model female-specific depressive-like stress responses, including anhedonia and behavioral despair, similar to those seen in human patients with depression. Estrogen receptor α in the nucleus acumens (NAc) has been identified as a modulator of female susceptibility to SCVS (Lorsch et al., 2018). More recently, a testosterone-dependent, sex-specific difference in ventral hippocampus (vHPC)—NAc neurons drives the female anhedonic response to SCVS (Williams et al., 2020).

These data link gonadal hormones to cellular excitability and key features of depressive states and support observations made in human MDD patients. For example, changes in gonadal hormones contribute to specific forms of depression-related illness in women including premenstrual dysphoric disorder, postpartum depression, and postmenopausal depression and anxiety (Pinkerton, Guico-Pabia, & Taylor, 2010). Additionally, meta-analyses of studies including nearly 1000 subjects demonstrate that adult (18–60 years old) but not elderly men with reduced testosterone are more likely to experience depression, and testosterone supplementation can relieve anhedonia and other depression symptoms (Amanatkar, Chibnall, Seo, Manepalli, & Grossberg, 2014). These data underscore the importance of the study gonadal hormones in MDD research.

While the prevalence of depression remains relatively the same across races and cultures, African Americans are more likely to experience chronic or persistent MDD than white Americans (Williams et al., 2007). This difference highlights healthcare disparities that often prevent minority groups in the United States from accessing sustained and successful treatment (Bailey, Mokonogho, & Kumar, 2019). For example, minority patients are far less likely to be seen in psychiatry than in primary care settings. This leaves the primary care physician to diagnose and treat with pharmacologic medication, which is not always the most effective treatment for the patient (Stockdale, Lagomasino, Siddique, McGuire, & Miranda, 2008). Lack of health insurance and access to proper resources in minority communities acts as another barrier to treatment. As a result, Hispanics and African Americans are less likely to have access to antidepressant medication (Jung, Lim, & Shi, 2014). Low socioeconomic positioning, poverty status, and unemployment, which are more common in minority groups due to systemic disadvantages, are also recognized as key risk factors for depression (Shim et al., 2012).

1.2. Stress and depression

Decades of research have established a robust and causal association between stressful life events and major depressive episodes (Hammen, 2005). One study concluded that as many as 80% of depressed cases were preceded by major life events (Brown, 1989; Mazure, 1998). Similarly, stress can impact the immune system in various ways. While acute stress enhances innate and adaptive immune responses, chronic stress can suppress immunity by decreasing immune cell numbers, function, and increasing immunosuppressive mechanisms (Dhabhar, 2009).

2. Immune disorders comorbid with depression

MDD is associated with a variety of chronic physical inflammatory and autoimmune disorders (Whitehouse et al., 2019). There is a particularly high association with inflammatory bowel disorders (IBD), and one study found the rates of depression in patients with IBD to be 16%, much higher than the prevalence in the general population (5.6%) (Fuller-Thomson & Sulman, 2006). However, the majority of these IBD patients were in remission. In a smaller study with more acute patients, the prevalence of MDD in IBD patients was as high as 35% (Walker, Gelfand, Gelfand, Creed, & Katon, 1996). In keeping with the general population, the rate of MDD was higher in women. Other immune disorders with high rates of depression include type 1 diabetes (15.2%) (Engum, Mykletun, Midthjell, Holen, & Dahl, 2005), rheumatoid arthritis (15%) (Pincus, Griffith, Pearce, & Isenberg, 1996), multiple sclerosis (18%) (Zorzon et al., 2001), and Guillain-Barre syndrome (GBS; 6.7% and 67% of most severe cases) (Kuitwaard, Bos-Eyssen, Blomkwist-Markens, & van Doorn, 2009). In addition, mood disorders may act synergistically with other medical disorders to worsen patient outcomes through effects on pro-inflammatory factors, hypothalamic-pituitary axis, autonomic nervous system, and metabolic factors.

3. CNS immunity

The immune complement of the central nervous system (CNS) is composed of both physical barriers and various immune cell types. The physical component includes 3 layers of collagenous tissue that surround the brain and spinal cord known collectively as the meninges (Fig. 1). Most CNS immune responses begin in the meninges before moving to the brain parenchyma, and meningeal immunity can profoundly influence brain and spinal cord homeostasis and even contribute to neurological disorders (Rua & McGavern, 2018). The first layer, in direct contact with the parenchyma, is the pia mater, and it is semi-permeable to cerebrospinal fluid (CSF) due to penetrating vessels (Hartman, 2009). The next layer, the arachnoid mater, has tight junctions that separate the outermost meningeal layer from the subarachnoid space. The subarachnoid space contains trabeculae and collagen bundles which connect the arachnoid to the pia mater, while also allowing for the movement of cerebrospinal fluid (CSF) which flows around the brain and spinal cord. CSF is produced as a filtrate from fenestrated blood vessels in the choroid plexus and provides transport for cytokines, neurotransmitters and hormones (Whedon & Glassey, 2009).

Fig. 1.

Immune complement of the meninges. The three layers of the meninges, the pia mater, arachnoid mater, and dura mater, make up the physical barrier between the brain/central nervous system and the outside environment. Blood and lymphatic vessels move through these layers, and the passage of immunoreactive substances into the brain is tightly regulated by a variety of immune cells: mast cells, T cells, B cells, macrophages, astrocytes, and dendritic cells. This meningeal immune complement regulates, and is modulated by, the brain and brain immune cells. Figure made in part with Biorender.com.

The outermost layer, or dura mater, is a thick, collagenous layer containing a high concentration of blood vessels (Rua & McGavern, 2018), and unlike cerebral vessels, these blood vessels are fenestrated and open to the passage of relatively large molecules (Balin, Broadwell, Salcman, & el-Kalli, 1986). The large network of fenestrated blood vessels within the meninges is accompanied by a separate network of lymphatic vessels. These intracranial lymphatic vessels run along the dural sinuses and middle meningeal artery and drain largely to the deep cervical lymph vessels. Dural lymphatic vessels traffic immune cells from the meninges back to the bloodstream via the CSF (Prinz & Priller, 2017), providing immune access to the brain. Because of their close proximity to the brain and ability to influence the brain’s physiology, the understanding the role of immune cells within the meninges is critical to uncovering the relationship between the immune system and depression.

4. Immune cells in the meninges

A large population of immune cells including macrophages, dendritic cells, innate lymphoid cells, mast cells, and neutrophils, B cells, and T cells exist within the meninges, especially the dura. These cells traffic from the blood into the meninges before entering lymphatic vessels and travel to the deep cervical lymph nodes (Radjavi, Smirnov, Derecki, & Kipnis, 2014).

4.1. T cells in the meninges

In addition to their role in immunity, emerging evidence suggests T cells are important for brain function, including learning and memory. Mice lacking T cells have an impaired performance in spatial memory tasks such as the Morris water maze, and when these mice were reconstituted with wild type CD4+ T cells from naїve spleens, Morris water maze performance was restored. Moreover, after 4 days of training in this test, wild type mice have increased levels of meningeal IL-4 producing CD4+ T cells (Brynskikh, Warren, Zhu, & Kipnis, 2008; Zarif et al., 2018, p. 8), indicating that learning induces T cell proliferation. Mice given a T cell-inhibiting antibody also exhibited decreased performance in Morris water maze along with decreased BDNF and hippocampal neurogenesis (Wolf et al., 2009). Additionally, mouse models of severe combined immunodeficiency (SCID), which lack B and T cells, showed a similar learning deficit that was reversed with intraperitoneal injection of wild type but not IL-4 deficient CD4+ T cells, suggesting that IL-4 released by innate immune cells is important to learning and memory as well (Derecki et al., 2010). In support of this notion, astrocytes treated with IL-4 in vitro express more BDNF, and mice with impaired IL-4 expression produced less BDNF after the Morris water maze than control mice. However, it still is unclear whether Cd4+ T cells are responsible for this IL-4 production (Derecki et al., 2010). Interestingly, elevation of IL-4 alone actually inhibited proliferation of hippocampal stem cell neurons in vitro, while mice with a restricted Cd4+ T cell repertoire had impaired hippocampal neurogenesis and reduced performance in memory tasks (Jeon et al., 2016). Thus, T cells appear critical for proper CNS function, memory, and cognition, but the mechanisms of their contributions remain incompletely defined.

T and B cells also seem to play an important role in social behavior. Mouse models for SCID that are deficient in B and T cells show reduced preference for social interaction with another mouse when given the choice between a novel mouse and a novel object, suggesting a social deficit (Filiano et al., 2016) and this was reversed by adoptive transfer of T cells. A similar deficit in social behavior was observed in interferon gamma (IFN-γ) knockout mice and also when the IFN-γ receptor was deleted from neurons in the prefrontal cortex (Filiano et al., 2016). It seems that IFN-γ induces activation of inhibitory circuits in the prefrontal cortex that prevent the hyper-connectivity observed in Autism spectrum disorder (Filiano et al., 2016). Together, these data suggest that IFN-γ immune defense evolved in some connection with social behavior, perhaps due to increased demand that is placed on the immune system when animals live in groups due to increased diversity and spread of pathogens (Rua & McGavern, 2018).

4.2. Macrophages in the meninges

Intravital imaging studies reveal large populations of macrophages in the dura as well as the pia mater(Morse & Low, 1972), and electron microscopy has also detected macrophages in the subarachnoid space. These macrophages, thought to be derived from embryonic precursors during development, are localized along meningeal blood vessels and do not appear to migrate. Meningeal macrophages promote tissue homeostasis, debris clearance, and protection from infections. Under basal conditions, these macrophages maintain an anti-inflammatory state that is regulated by IL-4 from CD4+ T cells (Derecki et al., 2010; Derecki, Quinnies, & Kipnis, 2011). Spatial learning in the Morris water maze causes myeloid cell activation in the meninges, and the water maze deficit in mice lacking T cells was partially improved by injecting IL-4 stimulated macrophages directly into the ventricular system (Derecki et al., 2011). Intravenous injection of anti-inflammatory macrophages had a similar effect. Therefore, it is unclear whether macrophages act primarily in the meninges or the periphery to exert their influence on meningeal homeostasis and cognitive function.

4.3. Mast cells in the meninges

A large population of mast cells exists in the meninges, primarily in the dura mater, and like macrophages, mast cells are localized along blood vessels (Rua & McGavern, 2018). These tissue resident cells have a variety of pro-inflammatory and immunoregulatory functions and play key roles in both innate and adaptive immune responses (Arac, Grimbaldeston, Galli, Bliss, & Steinberg, 2019; Tsai, Grimbaldeston, & Galli, 2011). They also release many cytokines, chemokines, proteases, and growth factors (Mukai, Tsai, Saito, & Galli, 2018), and their large impact on the meningeal environment and brain immunity make them an important potential therapeutic target. Mast cells are also implicated in many brain pathologies, including stroke (Arac et al., 2019; Rua & McGavern, 2018). Mast cell deficient mice are partially resistant to stroke pathology with reduced recruitment of myeloid cells to brain parenchyma, reduced brain swelling, and reduced infarct size (Arac et al., 2014). Injection of intravenous or intracranial wild type (but not IL-6 deficient) mast cells promoted neutrophil and monocyte recruitment to brain parenchyma, suggesting that mast cells, through the release mediators such as IL-6, play a role in orchestrating the inflammatory response post-stroke (Arac et al., 2014). In a murine model of EAE, meningeal mast cells are activated prior to clinical onset of disease and, through release of TNF, recruit neutrophils into the meninges (Christy, Walker, Hessner, & Brown, 2013). Thus, mast cells, particularly those highly concentrated in the meninges, play an important role in regulating neuroinflammation in the meninges associated with neurologic disorders.

5. Connecting meningeal immunity to depression

While meningeal inflammation has not been implicated as a direct cause of MDD, many patients with conditions associated with meningeal inflammation also experience comorbid depression (Amoozegar, 2017; Yang et al., 2016). In particular, epidemiologic findings indicate that there is a strong, bidirectional relationship between migraine, a meningeal-associated condition, and depression (Yang, Ligthart, et al., 2016). Experiencing chronic migraines increases the likelihood of having MDD and vice versa. Twin and family studies indicate that there is likely a significant genetic component to this disease profile (Yang et al., 2016a, 2016b). Stroke occurs when blood vessels that are part of the meninges rupture or become blocked, and depression is also particularly prevalent in patients with history of stroke (Kang et al., 2015). A 3-year longitudinal study found that the prevalence of MDD in the acute stage of stroke was 25%, and decreased to 16% at 12 months post-stroke (Astrӧm, Adolfsson, & Asplund, 1993). Depression is also seen with sufficient frequency to be considered a significant consequence after traumatic brain injury (TBI) (Rosenthal, Christensen, & Ross, 1998), which is often driven in part by swelling and damage to the meninges. A recent study found that patients who suffered from TBI and repeated head injury (RHI) independently had an worse mid- to later-life neuropsychiatric and cognitive functioning and depression symptoms (Bendlin, 2020). Thus, though direct connections between meningeal inflammation or immune activation and depression have not yet been explored, it is possible to connect the two states through comorbid conditions. Moreover, the meningeal immune complement directly affects brain immune activity, and, as we will explore below, neuroinflammation in the brain is strongly linked to depression.

6. Immune cells in the brain

Glia, the collective term for microglia and astrocytes, make up the brain’s innate immune system. These cells have many different functions including responding to pathogens and injury, gathering in regions of degeneration, and producing a wide variety of pro-inflammatory molecules. It is still unclear whether these roles are beneficial or detrimental to the overall brain health: it is likely that these cells may serve as a target in CNS trauma and disease states, but they also play critical roles in neuronal homeostasis, proliferation, and signaling. This section will outline the role of these cells in health, infection, injury and disease.

6.1. Microglia

Microglia are the main cell type in the brain’s innate immune system. While they are present throughout the CNS, they tend to be more highly concentrated in gray matter than white matter (Rivest, 2009). They have many projections which frequently grow and change even under normal healthy conditions. Microglia can be activated by systemic infection due to their strategic position in brain regions that have no blood-brain barrier. These regions include the circumventricular organs, which have a rich vascular plexus with, and somewhat open junctions between, capillary endothelial cells allowing for the diffusion of large molecules such as pathogen associated molecular patterns (PAMPs) (Rivest, 2009). The choroid plexus and leptomeninges are also highly vascularized and have a population of microglia that are rapidly activated by circulating pathogens. Activation of microglia at these sites leads to progressive activation of resident microglia of the parenchyma (Nadeau & Rivest, 2000; Nguyen, Julien, & Rivest, 2002). Like macrophages, microglia express Toll-like receptors (TLRs) which respond to TLR ligands and produce pro-inflammatory mediators (Rivest, 2003), and these receptors are implicated in many diseases, from Alzheimer’s disease to depression (Figueroa-Hall, Paulus, & Savitz, 2020; Richard, Filali, Préfontaine, & Rivest, 2008; Schmuck, Ullmer, Kalkman, Probst, & Lubbert, 1996).

Brain injury also activates microglia. Marked recruitment, proliferation and activation of microglia are detected in affected areas of the brain post injury (Roth et al., 2014), and suppression of this process provides a potential intervention for pathology associated with TBI.

6.2. Astrocytes

Astrocytes play a role in a variety of physiologic brain processes including maintenance of the BBB, modulation of blood flow, synaptic plasticity, and of energy homeostasis (Abbott, Rönnbäck, & Hansson, 2006; Bélanger, Allaman, & Magistretti, 2011; Perea, Navarrete, & Araque, 2009). The diversity of astrocyte function is reflected in the heterogeneity of the astrocyte population. For example, astrocytes in the hippocampus differ functionally from astrocytes in other brain locations (Chai et al., 2017). Even within the same brain region, astrocytes express molecular differences, and this variation allows for specificity when communicating with neurons and delivers a fine-tuned response to a broad spectrum of injuries (John Lin et al., 2017). Astrocytes rapidly activate in response to a variety of insults including primary brain tumors, brain metastasis, ischemia, epilepsy, multiple sclerosis, and neurodegeneration (Priego & Valiente, 2019). They also participate in cross talk with many cell types including T cells, microglia, and monocytes (Priego & Valiente, 2019), and thus are master regulators of immune function in the brain.

7. Glial degeneration and depression

Postmortem studies of brains of depressed patients show a decrease in density of glial cells in cortical regions of the brain, and this was particularly pronounced and consistent in the prefrontal cortex (Ongür, Drevets, & Price, 1998; Sanacora & Banasr, 2013). Astrocyte pathology in particular seems to play a role in depression. In depressed human subjects and animal models, decreased levels of glial fibrillary acidic protein (GFAP), an astrocyte marker, were found in the cortical and limbic structures of the brain (Gosselin, Gibney, O’Malley, Dinan, & Cryan, 2009; Miguel-Hidalgo et al., 2000; Si, Miguel-Hidalgo, O’Dwyer, Stockmeier, & Rajkowska, 2004). Human postmortem gene expression data also indicate astrocyte involvement in depression (Sanacora & Banasr, 2013). In addition, exacerbation of symptoms in depressed patients was correlated with an increase in S100B, a marker of glial degeneration in blood serum, and effect that was ameliorated by antidepressants (Schroeter, Sacher, Steiner, Schoenknecht, & Mueller, 2013).

Astrocytes are responsible for over 90% of the uptake and metabolism of glutamate in the brain (Bechtholt-Gompf et al., 2010). Loss of astrocytes, as seen in depression, therefore, results in accumulation of excess glutamate in the synaptic cleft. This is significant because glutamate driving excitatory-inhibitory imbalance and excitotoxicity is one of the leading theories in depression pathogenesis. Supporting this idea, an increase in glutamate was observed in the brains and cerebrospinal fluid of depressed patients (Hashimoto, Sawa, & Iyo, 2007). Furthermore, inhibition of glutamatergic function has an antidepressant effects (Zarate et al., 2006), with ketamine, an NMDA-type glutamate receptor antagonist, emerging as a new front-line fast-acting antidepressant. Moreover, some of ketamine’s antidepressant effects may be mediated by astrocytes (Ramadan & Mansour, 2020; Stenovec, Li, Verkhratsky, & Zorec, 2020). In both depressed patients and rat models, researchers observed a decreased number of GABAergic interneurons in the prefrontal cortex (Banasr, Dwyer, & Duman, 2011; Maciag et al., 2010; Zadrożna et al., 2011). This reduced population of interneurons could be driven by the loss of glial cells which normally support and prevent atrophy of neurons. GABAergic interneurons normally maintain tonic inhibitory control over the firing of excitatory glutamatergic neurons. Therefore, loss of interneurons, resulting from loss of glia, further enhances excess glutamate and excitatory/inhibitory imbalance. Magnetic resonance imaging studies showed that functional disruption of GABA transmission and decreased GABA levels were observed in depressed patients, providing evidence for this hypothesis (Sanacora & Saricicek, 2007).

8. Neuroinflammation and proposed mechanisms of depression

MDD is associated with increased serum chemokines, cytokines, and other indicators of peripheral immune activation (Köhler et al., 2017; Eyre, Air, et al., 2016; Eyre, Lavretsky, Kartika, Qassim, & Baune, 2016; Dowlati et al., 2010; Haapakoski, Mathieu, Ebmeier, Alenius, & Kivim€aki, 2015). In an analysis of two randomized control studies, multiple inflammatory markers proved useful biomarkers in predicting efficacy of pharmacologic antidepressant treatments in certain patients (Carboni et al., 2019). Interleukin 6 (IL-6), in particular, is consistently elevated across studies (Baker et al., 2001; Dowlati et al., 2010) and causes depressive symptoms associated with “sickness behavior” when injected systemically (Dantzer, O’Connor, Freund, Johnson, & Kelley, 2008). While the association between depression and elevation of IL-6 has long been studied, until recently, a causal relationship between this cytokine and depression had not been established. Using chronic social stress paradigms in mice, Hodes et al. demonstrated that mice susceptible to social withdrawal and anhedonia following stress had a higher pre-stress levels of IL-6 and circulating leukocytes, and these cells produced more IL-6 when exposed to acute stress ex vivo (Hodes et al., 2014). Chronic social stress can also lead to a loss of tight junctions in the blood-brain barrier increasing passage of IL-6 and depressive phenotypes (Menard et al., 2017). In fact, both in vivo and ex vivo increased IL-6 levels in response to acute immune stimulation were the strongest predictors of susceptibility to depression-related behaviors after social stress (Hodes et al., 2014). In addition, transplantation of bone-marrow-derived hematopoietic progenitor cells from susceptible mice promoted susceptibility, while transplantation of IL-6-deficient bone-marrow-derived hematopoietic progenitors promoted resilience, to both physical and emotional stressors in host mice (Hodes et al., 2014). This finding helps support previous evidence that IL-6 knockouts exhibit an antidepressant–like response to acute stress (Chourbaji et al., 2006). Furthermore, sequestration and elimination of peripheral IL-6 using IL-6 neutralizing monoclonal antibodies mitigates the maladaptive behavioral phenotypes seen in susceptible mice post-stress (Hodes et al., 2014). Finally, the same study demonstrated that MDD patients have higher circulating levels of IL-6 than controls, demonstrating the translational relevance of the studies (Hodes et al., 2014). These studies implicate the immune system in the pathophysiology of depression. However, future studies are needed to determine a clear mechanistic explanation that can be exploited to develop new treatments. In the following section, we will explore current proposed mechanisms of depression and the role of the immune system within these mechanisms.

8.1. Serotonin neurotransmission and the serotonin-kynurenine hypothesis of depression

Serotonin metabolism is reduced in depression patients (Hirschfeld, 2000) and depressed suicide subjects (Åsberg, 1997), and this is one of the most leading theories for a depression mechanism in the current literature. Current first line antidepressant drugs can combat this imbalance by enhancing the bio-availability of cerebral serotonin (Blier, de Montigny, & Chaput, 1990). The most commonly prescribed antidepressants are selective serotonin reuptake inhibitors, or SSRIs, like fluoxetine/Prozac©. Although these drugs alter serotonin levels at the synapse within minutes, they require weeks of administration to relieve symptoms of depression in humans (Harmer & Cowen, 2013) or reverse anhedonic or social withdrawal behaviors in mice (Berton et al., 2006; Krishnan et al., 2007). Although many potential mechanisms for this delay have been advanced, one possibility is that long-term changes in serotonergic signaling and metabolism driven in part by glia may be required to achieve functional behavioral effects.

Serotonin, also known as 5-hydroxytrptamine (5-HT), is synthesized from the amino acid tryptophan by the enzyme tryptophan hydroxylase. Acute tryptophan depletion in patients recovered from depression induces mild depressive symptoms (Moreno et al., 2010). In an alternative pathway, tryptophan can be catabolized by idoleamine 2,3-dioxygenase (IDO) and tryptophan 2,3 dioxygenase (TDO) into kynurenine (Badawy, 2017), reducing the serotonin levels. Inflammation seems to be one of the main catalysts for the switch from serotonin synthesis to the kynurenine pathway, as pro-inflammatory cytokines induce IDO activity, which is usually low under basal conditions (O’Connor et al., 2009). Kynurenine can then be catabolized into neurotoxic compounds 3-hydroxy-kynurenine, 3-hydroxy-anthralinic acid and quinolinic acid (Guillemin, 2012; Schwarcz, Bruno, Muchowski, & Wu, 2012). The conversion of kynurenine to neurotoxic 3-hydroxy-kynurenine is induced under inflammatory conditions (Schwarcz & Stone, 2017), and this predominantly occurs in immunocompetent cells including macrophages, monocytes and microglial cells (Guillemin, Smythe, Takikawa, & Brew, 2005). Kynurenine can also be metabolized via a “neuroprotective pathway” to kynurenic acid, a neuroprotective compound, via kynurenic-amino-transferase, an enzyme mainly expressed in astrocytes (Schwarcz et al., 2012). Moreover, it has been proposed that the homeostasis of glutamatergic neurotransmission can be regulated by the quinolinic acid/kynurenine ratio within microglia and astrocytes (Barone, 2019; Schwarcz & Stone, 2017).

Both clinical human and preclinical animal studies have shown that inflammation activates the kynurenine pathway. Excitotoxic metabolites and/or reductions in neuroprotective metabolites have been found in CSF or plasma samples in MDD patients (Ogawa et al., 2014) and suicide attempters (Messaoud et al., 2019; Sublette et al., 2011). A long-term dysregulation of the kynurenine pathway, with more excitotoxic metabolite production, was also observed in suicide attempters, and severity of symptoms was associated with increased inflammatory load (Bay-Richter et al., 2015). Various depression subtypes such as immunotherapy-related depression (Capuron et al., 2002), postpartum depression (Kohl et al., 2005), and depressive episodes in bipolar patients (Wurfel et al., 2017) have also shown activation of the kynurenine pathway. Furthermore, post mortem studies have revealed decreased quinolinic acid in the in the hippocampus of depressed subjects and increased microglial quinolinic acid in the anterior cingulate gyrus (Busse et al., 2015), and kynurenine metabolites and morphological changes may also be associated with prefrontal cortical thickness in MDD patients (Meier et al., 2016). Thus, a connection between depression and serotonin transmission/metabolism has been firmly established and many current therapies are based on manipulation of serotonin signaling, but a stronger understanding of the role of immune cells like glia in this signaling may uncover novel targets for therapeutic intervention in depression.

In addition to serotonin, norepinephrine dysregulation has also been implicated in depression pathophysiology as described in many reviews (Delgado & Moreno, 2000; Moret & Briley, 2011; Nutt et al., 2007). This neurotransmitter plays important roles in cognition, motivation, intellect, and social relationships (Ranjbar-Slamloo & Fazlali, 2020; Terbeck, Savulescu, Chesterman, & Cowen, 2016), and it is also the target of the depression treatment class known as serotonin norepinephrine reuptake inhibitors (SNRIs) (Sansone & Sansone, 2014).

8.2. The hypothalamic/pituitary/adrenal (HPA) axis

Though HPA axis (Fig. 2) dysregulation is significantly and consistently correlated with MDD, this dysregulation may become better understood through examination of the complex interplay between the HPA axis, inflammation, and the immune system.

Fig. 2.

The HPA axis in stress, immune signaling, and depression. In response to acute stress, the hypothalamus in the brain secretes corticotropin-releasing hormone (CRH), which causes the nearby anterior pituitary to release adrenocorticotropic hormone (ACTH) into the circulatory system such that it reaches the adrenal cortex near the kidneys, causing the release of glucocorticoids like cortisol, which drives the stress response. Glucocorticoids can drive both pro- and anti-inflammatory pathways, modulating brain function and mood. Under normal conditions, cortisol also activates glucocorticoid receptors (GR) throughout the HPA axis to dampen HPA signaling in a classical negative feedback mechanism. However, chronic stress reduces GR function in the HPA axis, removing inhibition and causing increased glucocorticoid signaling, favoring pro-inflammatory pathways that drive depression. Figure made in part with Biorender.com.

The HPA axis is the main regulatory system of glucocorticoids released by the adrenal glands in response to stressors. Glucocorticoid release has a multitude of effects throughout the body including metabolic, cardiovascular, immunological, and cognitive functions (Troubat et al., 2020). HPA axis dysfunction occurs in 35%–65% of MDD patients and usually results in increased expression of the adrenocorticotropic hormone (ACTH) secretagogue corticotropin-releasing hormone (Bao & Swaab, 2019; Holsboer, 2000). Higher levels of glucorticoids have been observed in plasma, saliva, and urine of depressed patients (Gillespie & Nemeroff, 2005), and MDD patients exhibit a reduced capacity to decrease glucocorticoid secretion in response to dexamethasone treatment, a glucocorticoid receptor agonist, that was restored after antidepressant therapies (Ising et al., 2007).

Glucocorticoids have well-studied anti-inflammatory and immunosuppressive effects and synthetic glucocorticoids have been widely used to treat chronic inflammatory conditions (Becker, 2013). Glucocorticoid receptor activation represses key inflammatory transcription factors such as the NF-kB/activator protein 1 (AP-1) pathway (De Bosscher, Vanden Berghe, & Haegeman, 2003) resulting in decreased pro-inflammatory gene expression (Cruz-Topete & Cidlowski, 2015). In addition to other immunosuppressive effects, glucocorticoids can also induce apoptosis of immune cells such as T lymphocytes, neutrophils, basophils, and eosinophils (Sorrells & Sapolsky, 2007).

However, glucocorticoids also have pro-inflammatory effects, and these are key in understanding the relationship between HPA axis dysfunction and MDD. Glucorticoids promote inflammation in three main ways: (1) engage in crosstalk to enhance activation of TLR pathways (Chinenov & Rogatsky, 2007); (2) promotion of P2Y2 purigenic receptors (Busillo, Azzam, & Cidlowski, 2011); and (3) regulation of the inflammasome via activation of the glucocorticoid receptor (Busillo et al., 2011). Within the context of stress, the pro-inflammatory effects of glucocorticoids may surpass their anti-inflammatory effects (Dantzer, 2018). In fact, in animals exposed to acute or chronic stress, the pro-inflammatory actions of glucocorticoids appear exacerbated, suggesting that stress and other factors may influence the effects of glucocorticoids toward the pro-inflammatory side (Bellavance & Rivest, 2014).

Inflammation directly impacts the HPA axis, as pro-inflammatory cytokines promote activation of the HPA axis and glucocorticoid release (Rivest, 2010). This is the case for both cytokines produced at sites of inflammation within the brain and production of peripheral circulating inflammatory mediators (Bellavance & Rivest, 2014). In turn, circulating inflammatory mediators can also influence permeability of brain capillaries, Cox-1, Cox-2, and prostaglandin E2 release, and stimulate the HPA axis (Elander et al., 2009; García-Bueno, Serrats, & Sawchenko, 2009; Serrats et al., 2010). Thus, immune activity may drive the HPA axis to create a state of chronic stress, a state directly linked to increased risk for depression (Cathomas, Murrough, Nestler, Han, & Russo, 2019).

Glucocorticoids can also have somewhat paradoxical impacts depending on a variety of factors including length of exposure to stress (i.e., acute vs chronic). The complexities of the influence of glucocorticoids and the HPA axis in depression and other psychiatric diseases have been explored in depth in the literature (Goudochnikov, 2018; Holsboer, 2001; Perrin, Horowitz, Roelofs, Zunszain, & Pariante, 2019).

8.3. Neurogenesis and neuroplasticity

Neurogenesis persists into adulthood in certain niches of the brain including the subventricular zone of the lateral ventricles and the sub-granular zone of the dentate gyrus of the hippocampus (Altman, 1962; Troubat et al., 2020). Hippocampal neurogenesis in particular plays an important role in the pathophysiology of depression and the effects of antidepressant drugs (Eisch & Petrik, 2012; Eliwa, Belzung, & Surget, 2017). In both animal models for the study of depression and human subjects undergoing a depressive episode, hippocampal neurogenesis is decreased (Eisch & Petrik, 2012; Troubat et al., 2020). Conversely, chronic antidepressant exposure induces an increase in hippocampal neurogenesis (Malberg, Eisch, Nestler, & Duman, 2000). Similarly, an artificial increase in adult hippocampal neurogenesis increases resilience to stress in mice (Hill, Sahay, & Hen, 2015), while ablation of hippocampal neurogenesis circumvents the effectiveness of antidepressants to cause remission of symptoms (Tsai, Tsai, Arnold, & Huang, 2015). Critically, neuroinflammation impacts adult hippocampal neurogenesis, as Inflammatory factors like LPS and IFN-alpha attenuate hippocampal neurogenesis in rodents (Cai et al., 2019; Ekdahl, Claasen, Bonde, Kokaia, & Lindvall, 2003; Fujioka & Akema, 2010; Kaneko et al., 2006). Moreover, as neuroinflammatory cells, microglia play a key role in modulating adult neurogenesis and eliminating apoptotic adult-born neurons through phagocytosis (Rogers et al., 2011; Sierra et al., 2010). In keeping with this, activation of microglia by LPS inhibits hippocampal neurogenesis (Ekdahl et al., 2003). It must be noted, however, that the existence and importance of adult neurogenesis in humans remains controversial (Kempermann et al., 2018), and although many existing antidepressants affect neurogenesis in animal models, direct targeting of neurogenesis in humans for therapeutic intervention in depression has not yet been established.

9. Antidepressants and the immune system

Current pharmacologic treatments for depression, including SSRIs, tricyclic antidepressants (TCA), and antiepileptic drugs (e.g., lamotrigine and valproic acid) and antipsychotics (including quetiapine or olanzapine) primarily modulate the level of neurotransmitters in the synaptic cleft, but also contribute to neurogenesis and neuromaturation (Andrade & Rao, 2010; Segi-Nishida, 2017). Moreover, new evidence suggests that SSRIs may act through receptors in addition to their blockade of reuptake through transporters (Vahid-Ansari, Zhang, Zahrai, & Albert, 2019), and new antidepressants directly target many serotonin (Amidfar & Kim, 2018). Further, considering that many patients do not respond to SSRIs, dysfunction of serotonergic signaling itself may not be the sole or even central explanation for depression. Growing evidence suggests that the lesser understood immunomodulatory effects of SSRIs and other antidepressant drugs may contribute to their therapeutic effects (Table 1).

Table 1.

Connecting antidepressant drugs to neuroimmune modulation.

Antidepressant	CNS effect	Immunomodulatory effect
Selective Serotonin Reuptake Inhibitors (SSRIs) ex. fluoxetine, paroxetine, escitalopram	Decrease reuptake of serotonin at presynaptic terminals resulting in increased serotonin in the synapse	Decrease cytokines Decrease T cell proliferation Increase regulatory T cells Increased serum BDNF
Tricyclic Antidepressants (TCA)	Bock the reuptake of serotonin and norepinephrine in presynaptic terminals, which leads to increased concentration of these neurotransmitters in the synaptic cleft	Decreased TNF-α in responders (Eyre et al., 2016) IL-6 may be useful in predicting responders vs non-responders (Eyre et al., 2016)
Ketamine	Glutamate receptor agonist Increased synaptic plasticity and strengthening of excitatory synapses (Zanos & Gould, 2018)	Reduces TNF-alpha and IL-6 (Beilin et al., 2007) Increase CD4+/CD8+ ratio and percentage of T-reg cells (Hou et al., 2016) Inhibits differentiation of Th2 cells (Gao et al., 2017) Reduces pro-inflammatory cytokine synthesis (Braun et al., 2010) Induces apoptosis of T lymphocytes (Braun et al., 2010) Inhibits maturation of dendritic cells (Zeng, Xia, Zhong, Li, & Lin, 2011)
Antiepileptic drugs (AEDs)	Increase sodium channel activation which can increase GABA levels in the CNS	Effects of AEDs on the immune system remain somewhat inconclusive Immune reactions such as Steven Johnson Syndrome, DRESS syndrome, and other IgE hypersensitivity reactions occur with increased frequency in patients taking AEDs (Godhwani & Bahna, 2016) AEDs can have varying effects on patient immunity including cytokine dysregulation, cytopenias, and various degrees of hypogammaglobulinemia (Godhwani & Bahna, 2016)

Depressed patients participating in clinical trials over the course of a 20 year period showed a decrease in serum TNF-alpha and IL-6 after antidepressant therapy (Strawbridge et al., 2015). When peripheral cytokines were recorded before and after a 12-week antidepressant therapy in 50 depressed patients, a significant reduction in IL-6, IL-7, IL-8, IL-10, G-CSF, IFN-y and IL-1 receptor antagonist was observed—but only in patients who responded to the antidepressant treatment (Dahl et al., 2014). A decrease in cytokines was also seen after treatment with the SSRI sertraline in a randomized trial of 73 patients (Brunoni et al., 2014), though cytokine decrease also occurred in patients taking placebo and was not correlated with antidepressant effect (Brunoni et al., 2014).

Antidepressants can also influence the numbers and activity of immune cells, and the number of T lymphocytes in the peripheral blood of mice was decreased after fluoxetine treatment (Di Rosso, Palumbo, & Genaro, 2016). In vitro T lymphocytes from human patients treated with the SSRI fluoxetine showed a decrease in proliferative activity (Fazzino, Urbina, Cedeño, & Lima, 2009). Moreover, 6 weeks of fluoxetine treatment caused a decrease in concentration of pro-inflammatory cytokines and an increase in number of regulatory T cells in depressed patients (Grosse et al., 2016; Himmerich et al., 2010). Antidepressant therapy also increases the number and cytotoxic activity of NK cells (Mizruchin et al., 1999; Park, Lee, Jeong, Han, & Jeon, 2015), and bipolar patients treated with lithium or valproate also showed decreased T lymphocyte proliferative activity and increased susceptibility of T lymphocytes to apoptosis (Pietruczuk, Lisowska, Grabowski, Landowski, & Witkowski, 2018). In vitro, these drugs both displayed anti-apoptotic effects, but no effect on the ability of lymphocytes to proliferate (Pietruczuk et al., 2018).

Antidepressants decrease expression of inflammasome genes and inflammasome activity in mononuclear cells taken from depression patients (Alcocer-Gómez et al., 2017). Brain derived neurotrophic factor (BDNF), a signaling protein associated with the formation of new neurons and synapses and with anti-inflammatory properties, was increased in patients’ serum after 5 weeks of SSRI treatment (Matrisciano et al., 2009). Ketamine, the antidepressant glutamate receptor antagonist described above, also modulates the immune system. Low doses of ketamine administered as anesthetic temporarily reduce patient blood concentrations of TNF-alpha and IL-6 in the postoperative period (Beilin et al., 2007). Ketamine also increases the CD4+/CD8+ ratio as well as the percentage of T-reg cells (Hou et al., 2016), inhibits the differentiation of Th2 cells (Gao et al., 2017), reduces pro-inflammatory cytokine synthesis, induces apoptosis of T lymphocytes (Braun et al., 2010), and inhibits the maturation of dendritic cells responsible for antigen presentation (Zeng et al., 2011). Thus, existing antidepressants directly affect immune cells and immune signaling in the brain and periphery. It will be critical to determine which, if any, of these immune effects are causal in the antidepressant properties of these drugs in order to determine whether specific immune cells or signaling pathways could be targeted to produce novel antidepressant medications.

10. Conclusion: Immunomodulatory drugs, a new class of antidepressant therapies?

Just as the immunomodulatory effects of antidepressants are being explored, the application of existing immunomodulators, such as nonsteroidal anti-inflammatory drugs (NSAIDs), cytokine inhibitors, polyunsaturated fatty acids, and curcumin are being considered for the treatment of MDD (Köhler-Forsberg et al., 2019; Szałach, Lisowska, & Cubała, 2019). NSAIDS, especially the cyclooxygenase-2 inhibitor celeoxib, can reduce the symptoms of depression (Köhler et al., 2014). The soluble TNF-alpha receptor, etanercept, also exhibited anti-depressive effects, while Infliximab, an anti-TNF antibody (Raison et al., 2013), and curcumin, a compound found in plants of the ginger family (Menon & Sudheer, 2007; Yu, Pei, Zhang, Wen, & Yang, 2015), have also shown anti-depressive potential. This repurposing of existing immunomodulatory drugs to treat depression or other mood disorders offers an exciting and potentially rapid pathway to improve depression outcome, particularly in currently treatment-refractive patients.