Chronic gut-derived systemic inflammation is able to take different anatomical routes connecting the gastrointestinal tract and the bloodstream with the central nervous system (CNS) in order to provoke an inflammatory response within the brain (Figure 1).
1.1. Enteric, Autonomic and Sensory Nervous System Signaling
The peripheral (i.e., enteric, autonomic) and central nervous systems are tightly connected to form a bidirectional neural communication highway between the gastrointestinal tract and the brain. This route can be employed by inflammatory signaling towards the CNS directly from the gut, without requiring transition into the bloodstream
[1]. A major transit route of gut-derived signals is the vagal nerve. Under homeostasis, vagal nerve afferents in the gut monitor key physiological parameters
[2] and microbial metabolites like the SCFA butyrate, a surrogate for nutritional state
[3]. These afferents project to the nucleus tractus solitarii (NTS) in the brainstem, directly activating neurons in the efferent dorsal motor nucleus of the vagal nerve to signal back to the gut, but also project to many other brain regions
[4]. Under inflammatory conditions, vagal sensory neurons are able to sense Tnf and Il-1β, and signal inflammatory cues to the NTS by cytokine-specific electrophysiological patterns
[5]. Moreover, early studies implicated the vagal nerve in CNS transmission of peripherally derived Il-1β in fever induction that was abrogated in rats by vagotomy
[6]. The role of the vagal nerve in transmitting chronic gut inflammation in IBD towards the brain is not well elucidated and requires further investigation. In chronic mild gut inflammation caused by infection of mice with the parasite
Trichuris muris (
T. muris), anxiety-like behavior was reduced by systemic anti-inflammatory treatment, but not altered by vagotomy
[7]. Besides a potential role of the vagal nerve in signaling of inflammatory cues to the brain, cholinergic efferent vagal nerve fibers can ameliorate gastrointestinal inflammation in TNBS-induced colitis
[8]. Moreover, vagal nerve stimulation was reported to be beneficial both in depression and IBD, again pointing to the 10th cranial nerve as a major regulator of the gut–immune–brain axis
[1][9].
A second neural route potentially involved in immune communication from the gut to the brain is the nociceptive system. IBD is associated with severe abdominal pain. In vitro data suggested that TNF in the supernatant of colonic biopsies from UC patients activated nociceptors on dorsal root ganglia neurons
[10]. In line with this, IBD patients receiving TNF antagonists show a rapid and profound reduction in blood oxygen level dependent (BOLD) brain activity in functional magnetic resonance imaging (fMRI) after application of painful stimuli
[11]. Strikingly, this CNS-mediated treatment response was present prior to the resolution of gastrointestinal inflammation
[11]. These data suggest TNF-mediated upward signaling of gastrointestinal inflammation via the nociceptive afferent system contributes to functional CNS alterations in IBD. Therefore, neural communication pathways like the vagal nerve and the nociceptive system transmit gut inflammation to the brain.
1.2. Blood–Brain Barrier
The blood–brain barrier (BBB) controls the passage of circulating molecules and cells from the blood into the brain parenchyma. It consists of vascular endothelial cells connected by tight junctions and situated on a vascular basement membrane. The brain parenchyma is bordered by a layer of astrocytic end feet and their corresponding basement membrane. The perivascular space is located between the two basement membranes and contains perivascular macrophages and pericytes
[12][13]. In most brain regions, the BBB prevents passive diffusion of hydrophilic molecules larger than 500 Da into the brain
[14]. Thus, the BBB is by far less permeable than the GVB. However, several small regions located around the third and fourth ventricle, the so-called circumventricular organs (CVOs), exhibit more permeable endothelia to allow bidirectional interaction between the brain and the periphery, e.g., for endocrine signaling.
Systemic inflammation can modulate BBB function by different mechanisms
[15]. Disruption of BBB tight junctions facilitates paracellular influx of inflammatory molecules and immune cells. In animal models of IBD, BBB disruption was investigated based on expression of tight junction markers and dye permeability assays, leading to heterogenous results. Reduction in the tight junction markers occludin and claudin-5 in the brains of mice with acute DSS-induced colitis was reported in one study
[16], while this reduction required additional hypoxia in a second study
[17]. Similarly, the permeability of Evans Blue-binding albumin (molecular weight 69 kDa) into the brain parenchyma of mice with acute DSS-induced colitis was increased
[18] or unchanged
[16], respectively. A recent time course analysis in mice treated with DSS for 3 days followed by 2 days of drinking water revealed a reduction in vascular zonula occludens-1 (Zo-1) after 3 days and a normalization after 5 days
[19]. Strikingly, these mice showed an initial increase followed by a significant decrease in permeability for systemically administered fluorescent dextran (molecular weight 70 kDa), which also normalized after 5 days
[19], indicating rapid dynamics of BBB regulation including a transient closure. Transient alterations of BBB permeability were also observed in TNBS-induced colitis showing enhanced permeation of Evans Blue-albumin after 1 day and a normalization at later time points
[20]. Moreover, TNBS-induced colitis caused higher leakage of fluorescein (molecular weight 376 Da) into the brain parenchyma and different circumventricular organs
[21][22]. Overall, these data indicate that disruptive changes in the BBB during the course of colitis may be of transient and highly dynamic nature. While changes in BBB permeability during the phase of acute colitis were demonstrated, data on chronic preclinical IBD are largely lacking.
Besides disruption and increased permeability of the BBB, systemic inflammation is able to induce so-called non-disruptive BBB alterations. This includes the upregulation of cellular adhesion molecules like intercellular adhesion molecule-1 (Icam-1) and vascular adhesion molecule-1 (Vcam-1) on BBB endothelia, facilitating the evasion of immune cells via the intact BBB
[15]. Such mechanisms were barely addressed in experimental colitis models, except for one study reporting increased Vcam-1 mRNA in the brains of DSS-treated mice
[23]. Infiltration of blood-derived immune cells contributes to parenchymal CNS inflammation due to either disruptive or non-disruptive BBB changes in distinct experimental models of IBD. Increased numbers of blood-derived monocytes and macrophages were observed in acute DSS-induced colitis
[17][23][24], and to an even larger extent in chronic DSS-induced colitis
[23]. Brain infiltration of neutrophils in acute DSS-induced colitis was reported either increased
[23][24] or reduced
[17]. In T cell transfer colitis, brain infiltration of monocytes and T cells was proposed
[23]. However, many of these data must be interpreted with caution, as they are solely based on flow cytometry experiments and might be confounded by cells residing in the vascular lumen, the perivascular space, the meninges or the choroid plexus (CP). Future detailed structural studies must decipher the precise path of peripheral-derived immune cell localization and entry route.
Finally, systemic inflammation may result in direct inflammatory activation of BBB endothelial cells, which are able to secrete inflammatory mediators to modulate microglial and neuronal function
[25]. During aging, upregulation of Vcam-1 on BBB endothelia inhibits neural progenitor cell (NPC) activity, mediates microglial activation, and induces an age-related impairment in hippocampal-dependent learning and memory
[26]. These alterations were reversed by anti-Vcam1 antibody treatment
[26]. The role of endothelia, rather as active neuroimmune players than a passive bystander, has not been addressed in the context of mucosal inflammation or experimental colitis.
Collectively, the permeability of the BBB during acute colitis is dynamically regulated, while insights on the BBB in chronic colitis as well as non-disruptive changes and immune activation of endothelial cells and perivascular macrophages during IBD need to be further analyzed in future studies.
1.3. Choroid Plexus and Blood–CSF Barrier
Besides alterations in BBB integrity, the CP has been identified as a gateway for pathogens, cells, and molecule transport into the CNS via the cerebrospinal fluid (CSF)
[19][27][28]. The CP contains blood vessels with fenestrated endothelia as well as a layer of epithelial cells connected by tight junctions separating the CP from the CSF. Together, CP endothelia and epithelia form the blood–cerebrospinal fluid barrier (BCSFB). In addition, the CP contains a heterogeneous pool of immune cells, including T cells, B cells, dendritic cells, natural killer cells, lymphocytes, and two types of resident macrophages, one of which resembles brain parenchymal microglia
[29].
In recent years, several findings shed light on the CP and the BCSFB as an important immune interface between the systemic circulation and the CNS. During aging and systemic inflammation, type I-interferon signaling in the CP becomes upregulated, which mediates glial cell activation and cognitive impairment
[25][30]. Moreover, lipopolysaccharide (LPS)-induced systemic inflammation has the potential to trigger the release of extracellular vesicles by the CP, which propagate inflammation towards the brain
[28].
Recently, the BCSFB has been thoroughly characterized in acute colitis induced by application of DSS for 3 days
[19]. Strikingly, increased permeation of the CP vasculature after 1 day was followed by a transient closure of the vascular CP barrier after 3 days, whereas the tight junction marker Zo-1 was reduced between CP epithelial cells
[19]. Together, these changes lead to a decrease in fluorescent tracer permeation into the CSF. Transient closure of the CP vascular barrier was orchestrated by the Wnt-β-catenin-pathway in endothelial cells and contributed to anxiety-like behavior and memory impairment
[19]. These findings challenge the concept of general vascular barrier disruption by systemic inflammation, and suggest a contribution of the open CP vascular barrier to cognitive processing. Apart from the BCSFB, the role of the diverse CP immune cells during IBD and related CNS comorbidity is not well understood. The closure of CP vasculature in DSS-treated mice was accompanied by reduced numbers of CP macrophages and a rapid reduction in neutrophil infiltration
[19]. Overall, acute DSS-induced colitis led to increased numbers of CD45+ leukocytes in the CP, but a further characterization of cell types was not performed
[31]. Future studies focusing on the contribution of CP immune cell subsets and the BCSFB, especially in chronic experimental colitis models, are necessary.
In summary, the CP and the BCSFB may display a promising target for the treatment of IBD-associated neuropsychiatric comorbidity. This is further suggested by recent imaging data drawing a possible link between CP volume, BBB and BCSFB closure, and neuroinflammation in patients with depression
[27].
1.4. Meninges
The meninges, covering the CNS surface, are divided into the leptomeninges (pia mater and arachnoidea mater) and the dura mater, both of which contain resident macrophage populations as well as diverse other immune cells under homeostasis
[29]. A growing body of evidence underpins the relevance for meningeal immune signaling for CNS homeostasis. T cells in the meninges regulate neural activity and social behavior through IFN-γ that directly activates γ-aminobutyric-acid (GABA)-ergic neurons
[32]. Additionally, meningeal γδT cells were recently shown to signal to glutamatergic neurons via Il-17a inducing anxiety-like behavior
[33]. Interestingly, immunological interaction between the gut and the meninges was observed in stroke, where gut-derived CD11c+ myeloid cells were found to migrate to the meninges and CNS
[34]. Moreover, T cells expressing gut-homing receptors have been shown to circulate in the CSF of patients with non-inflammatory neurological disease
[35]. These cells might enter the CSF via the meninges or the CP.
In mice with chronic DSS-induced colitis, activation of the NACHT, LRR, and pyrin domain containing protein 3 (NLRP3) inflammasome in the meninges was proposed and linked to an increased infiltration of gut-derived T cells
[36]. These findings indicate that the meninges may act as a gut-to-brain immune interface in IBD.
Overall, distinct anatomical interaction pathways may contribute to the propagation of gut-derived systemic inflammation towards the CNS, subsequently triggering neuroinflammation.
2. How Neuroinflammation Is Linked to Depression and Anxiety
CNS immune activation is a well-known phenomenon in depression and anxiety. Recent findings indicate that immune processes represent a key pathogenic driver rather than a pure epi-phenomenon in both conditions. This is supported by the notion that brain-resident immune cells including microglia and brain-resident CD4+ T cells are essential for neuronal homeostasis and physiological behavior
[37][38][39]. Moreover, microglia were reported to mediate behavioral deficits in models of depression and anxiety induced by chronic unpredictable stress or early-life inflammation induced by intraperitoneal LPS injection
[40][41][42]. Involved inflammatory pathways being potential therapeutic targets include microglia–astrocyte crosstalk via glutaminase-1
[43], the NLRP3 inflammasome
[44][45] and the clearance of reactive oxygen species (ROS) via silent information regulator 2 homolog 1 (Sirt1)—nuclear factor erythroid 2-related factor 2 (Nrf2)—hemoxygenase 1 (Ho-1)—signaling
[46][47].
Interestingly, modulation of the BBB is also implicated in the etiology of stress-induced depression and anxiety in chronic social defeat stress. In particular, stress-susceptibility and depression-like behaviors were dependent on downregulation of the tight junction marker claudin-5 (Cldn5) in the hippocampus and nucleus accumbens, which was orchestrated via Tnf and histone deacetylase 1 (Hdac1) and facilitated vascular influx of Il-6 into the brain parenchyma
[48][49]. As chronic gastrointestinal inflammation was reported to cause BBB tight junction downregulation comparable to social stress, it might also represent a trigger for BBB-mediated depressive-like behavior. Furthermore, non-disruptive BBB alterations may contribute to anxiety-like behavior. Social stress-induced anxiety was mediated by Il1-receptor 1 producing endothelial cells, which were activated by Il-1β-expressing monocytes attracted to the BBB by microglia
[50]. These findings suggest that endothelia closely interact with myeloid cells and may act as a gatekeeper to further develop neuropsychiatric symptoms. A better understanding of non-disruptive BBB changes in IBD is necessary to explore if such mechanisms are present during gastrointestinal inflammation.
Though these and other studies strongly imply the relevance of neuroinflammation in depression and anxiety, they do not precisely delineate how neuroinflammation alter the function of neuronal circuits involved in behavioral and emotional regulation to ultimately trigger psychiatric symptoms. In the context of IBD, this might be mediated by several pathways (Figure 2).
Figure 2. Impaired neuronal functions in inflammatory gut-to-brain communication. Microglia, peripheral immune cells and endothelia contribute to immune-mediated impairment of neuronal functions in IBD by different mechanisms. First, reduced neurotrophic signaling by BDNF is induced by neuroinflammation and observed in IBD. Second, inflammation interferes with neurotransmitter metabolism, e.g., resulting in reduced availability of serotonin. Third, electrophysiological properties of distinct neuronal populations are modified, and synaptic plasticity is reduced. Fourth, adult hippocampal neurogenesis is impaired. Fifth, increased microglial engulfment of synapses may lead to aberrant synaptic degradation. Finally, inflammatory signaling could induce neuronal cell death. Intestinal microbiota essentially contribute to neuronal alterations in IBD, either by augmenting inflammatory activation of immune cells or by direct influence on neurons. Disturbed neuronal function is the pathological correlate of behavioral changes and neuropsychiatric comorbidity in IBD. BDNF: brain derived neurotrophic factor; ↑: increased; ↓: decreased. Figure created with
BioRender.com (accessed on 18 August 2022).
First, CNS immune activation may compromise neuronal activity and synaptic transmission in key regions involved in anxiety and depression. Microglial activation was recently shown to reduce neuronal excitability in the dorsal striatum in a prostaglandin-dependent manner. Thus, targeting cyclooxygenase-1 (Cox-1)-mediated prostaglandin synthesis in microglia may alleviate depressive symptoms
[40]. In the basolateral amygdala, inflammation induced by peripheral LPS application leads to increased glutamate release and projection neuron excitability, which was linked to depressive and anxiety-like behavior
[51]. In line with these findings, TNBS-induced colitis reduced synaptic plasticity and elicited enhanced synaptic transmission in hippocampal glutamatergic neurons
[52]. Interestingly, manganese-enhanced MRI indicated reduced hippocampal activity in chronic DSS-induced colitis
[18]. Recently, inflammation-induced dysregulation of neuronal circuits was proposed to diminish inhibitory input from the prefrontal cortex and hippocampus on hypothalamic corticotropin releasing hormone (CRH) secretion. This, in turn, might aggravate both colitis and depressive-like behavior
[53]. Together, electrophysiological properties and synaptic transmission of defined neuronal subtypes may be impaired during IBD-related neuroinflammation.
Additionally, CNS immune activation can shift neurotransmitter metabolism, in particular resulting in reduced availability of serotonin. Impaired serotoninergic signaling is involved in depression and is a major target for antidepressant treatment. In IBD, peripheral serotonin may act as a double-edged sword by augmenting mucosal inflammation
[54], but protecting the enteric nervous system
[55]. Of note, antidepressant serotonergic treatment positively influenced the disease course among patients with CD and UC, decreasing systemic pro-inflammatory cytokine levels
[56]. Activated microglia express indoleamine 2,3-dioxygenase (IDO) to catabolize tryptophan into kynurenine instead of serotonin. Kynurenine is further processed into excitotoxic metabolites
[57]. In acute DSS-induced colitis, IDO expression was increased in the prefrontal cortex
[58]. Moreover, chronic colitis induced by infection with
T. muris was accompanied by increased serum kynurenine levels
[7]. These data suggest impaired tryptophan-serotonin metabolism in IBD, but a direct link between IBD-related neuroinflammation, impaired serotonergic signaling, and behavioral deficits has not yet been drawn.
Additionally, microglia can engulf and prune synapses. Microglial synaptic pruning is essential for proper brain development, but is aberrantly upregulated during neurodegeneration
[59]. Intriguingly, the complement system, which is essentially involved in synaptic pruning, was recently implicated in stress-induced depressive-like behavior
[60]. Moreover, microglia-synapse interactions were altered in models of depression in a spatiotemporally distinct manner. Early-life LPS-induced inflammation enhanced microglial engulfment of glutamatergic neuronal spines in the anterior cingulate cortex and thereby elicited depressive-like behavior in adolescence
[41]. Depression and anxiety provoked by chronic unpredictable stress were linked to upregulation of microglial phagocytosis by neuronal colony-stimulating factor (CSF) 1, leading to reduced dendritic spine density on pyramidal neurons in the medial prefrontal cortex
[42]. In contrast, early-life stress disturbed microglial engulfment of excitatory synapses in stress-sensitive CRH-expressing neurons in the paraventricular nucleus (PVN) of the hypothalamus
[61]. The resulting activation of the hypothalamo–pituitary–adrenal axis impaired behavioral stress response. In IBD, synaptic clearance and involved pathways like the complement system are yet to be investigated. First insights indicate loss of Map2-positive dendritic nerve fibers in the cortex and hippocampus during chronic DSS-induced colitis in aged mice
[36].
Besides structural dynamics of synapses, neuroinflammation in IBD might cause neuronal cell death. In line with this, the number of total neurons in the cortex and hippocampus of aged mice with chronic DSS-induced colitis was reduced
[36]. Correspondingly, elevated expression of caspase 3 indicated increased apoptotic cell death in the brain during acute colitis, although this was not yet assigned to particular cell types
[16]. Besides apoptosis, other kinds of cell death were not addressed in the CNS during IBD. Altogether, neuroinflammation in IBD could contribute to psychiatric symptoms by inducing structural alterations or degradation of synapses or cell death of neurons.
Impairing neuronal plasticity is another pivotal mechanism by which neuroinflammation might be able to mediate neuropsychiatric symptoms. Adult neurogenesis, the generation of new neurons in the brain throughout adult life, only occurs in few niches including the subgranular zone of the hippocampal dentate gyrus. Adult hippocampal neurogenesis is involved in learning, memory, and pattern separation
[62]. However, impaired adult hippocampal neurogenesis is also linked to depression
[63][64]. Importantly, there is broad evidence for impaired adult hippocampal neurogenesis during neuroinflammation. While homeostatic microglia maintain adult hippocampal neurogenesis
[65], inflammatory cytokines like Tnf
[66], as well as Il-1β
[67] and peripheral inflammation induced by LPS administration
[68], inhibit NPC proliferation and maturation. Intriguingly, elevated blood levels of Ccl11, a chemokine also observed in the serum of CD and UC patients
[69][70], decreased adult hippocampal neurogenesis during aging-related peripheral inflammation
[71]. In the context of IBD, a direct mechanistic link between neuroinflammation and adult hippocampal neurogenesis was not yet shown, but several studies in acute and chronic DSS-colitis show impaired adult neurogenesis in the hippocampus
[72][73][74][75][76]. Noteworthy, acute and chronic DSS-induced colitis altered distinct aspects of adult hippocampal neurogenesis. Acute colitis increased progenitor cell proliferation, but dysregulated cell cycle kinetics, while chronic colitis led to reduced migration and functional integration of newly generated neurons
[72]. Collectively, adult hippocampal neurogenesis is vulnerable towards peripheral and cerebral inflammation and may contribute to IBD-linked neuropsychiatric symptoms.
Homeostatic brain functions, including adult neurogenesis, are governed by several trophic factors. One pivotal factor is the brain derived neurotrophic factor (BDNF), which signals via its receptor tyrosine receptor kinase b (Trkb). BDNF supports the release of neurotransmitters as well as the expression and function of neurotransmitter receptors and ion channels
[77]. Moreover, BDNF augments synaptic plasticity and adult neurogenesis
[77]. Interestingly, reduced BDNF levels were linked to depression
[44], and a major mode of action of antidepressant drugs was recently revealed to be the amplification of BDNF-Trkb-signaling
[78]. Of note, there is evidence for reduced BDNF signaling during neuroinflammation. The expression of BDNF is suppressed by Il-1β
[79]. Furthermore, astrocyte-derived Il-33 was reported to reduce BDNF levels in the amygdala, which was linked to impaired signaling of GABAergic neurons
[80]. In line with this, BDNF levels in the brain were reduced in acute and chronic DSS-induced as well as in DNBS-induced colitis
[24][74][75][81]. Treatment with liver hydrolysate rescued neuroinflammation and depressive-like behavior in acute DSS-induced colitis, putatively by inducing BDNF expression via adenosine monophosphate-activated protein kinase (AMPK)
[75]. Though neuroinflammation and reduced BDNF levels were only coincident and not causally linked in IBD models, impaired BDNF signaling might be triggered by neuroinflammation and contribute to depression and anxiety in IBD.
In summary, neuroinflammation can trigger neuronal dysfunction via a plethora of distinct mechanisms, thereby mediating neuropsychiatric comorbidity in IBD. Though many of these potential mechanisms were described in animal models for IBD, a major limitation of most studies is the lack of a causal relation between coinciding neuroinflammation and depressive-like behavior. Only marginal data supporting this causality were generated in pharmacological studies. Inhibition of the DAMP S100a9 alleviated DSS-induced colitis, neuroinflammation, and behavioral impairment
[24]. However, these effects might be explained solely by the reduction in colitis rather than interference with immune gut-to-brain communication. Interestingly, systemic inhibition of RNS did not affect TNBS-induced colitis but diminished hippocampal Tnf levels and reversed depressive-like behavior
[82]. Moreover, local intracerebroventricular administration of the antibiotic and immune modulatory drug minocycline reduced microglial activation and normalized synaptic plasticity in TNBS-induced colitis
[52]. Though these findings link neuroinflammation and neuronal dysfunction in the context of IBD, the applied treatment paradigms are unspecific. Thus, more specific approaches are required to investigate the causal link between individual immune cell types, neuronal dysfunction and neuropsychiatric symptoms in IBD.