| |
Liver-induced inflammation hurts the brain
|
| |
| |
Download the PDF here
Download the PDF here
Journal of Hepatology March 2012
Rita Garcia-Martinez1, Juan Cordoba2,3,4,
1Liver Failure Group, University College London Institute of Hepatology, The Royal Free Hospital, Pond Street, London NW3 2PF, UK; 2Servei de
Medicina Interna-Hepatologia, Hospital Vall d'Hebron. Barcelona, Spain; 3Departament de Medicina, Universitat Autonoma de Barcelona, Spain;
4Centro de Investigacion Biomedica en Red de Enfermedades Hepaticas y Digestivas (CIBEREHD), Instituto de Salud Carlos III, Madrid, Spain
The immune system is activated following injury or infection. The local response can be accompanied by a systemic response, which includes the synthesis and release of different mediators by innate immune cells. The liver is not an exception and when exposed to an acute or chronic insult generates an inflammatory response that may affect other organs. Liver-induced inflammation is able to cause disturbances in the central nervous system (CNS) including metabolic (hyperthermia, somnolence, loss of body weight) and behavioural manifestations (lethargy, anhedonia, decreased social interaction). These manifestations are collectively termed "sickness behaviour" [1], and are attributed to dysfunction of the CNS.
The underlying mechanisms responsible for this periphery-to-brain inflammatory communication are not fully understood, but include both neural and humoral pathways that act in parallel (Fig. 1) [2]. The neural pathway involves afferent nerves (vagal and trigeminal nerves) that can be locally activated by cytokines present at the site of injury and project to different areas of the brain. The humoral route involves cytokines that interact with the brain diffusing freely from the blood in areas lacking the blood-brain-barrier (the circumventricular organs), or communicate with the brain by activating the endothelium and transmitting signals to brain parenchyma. A key component of this signalling process is the activation of the immune cells resident in the brain parenchyma (microglia), which are usually in a quiescent state. The activation of microglia is considered essential in the innate immune response of the brain and leads to the release of molecules, such as neurosteroids and/or prostaglandins that affect neurons.
Neuroinflammation is a process that has gained increased attention in neurodegenerative diseases, such as Alzheimer's and Parkinson's disease, and has also been described in hepatic encephalopathy [3]. According to the data published in this issue of the Journal of Hepatology, neuroinflammation may also participate in more subtle neurological manifestations of liver disease. The article by Nguyen et al. investigates the underlying mechanisms of sickness behaviour in a murine model of cholestasic liver injury (bile duct ligation, BDL). In agreement with other studies [4], sickness behaviour is related to an increase in serum interleukin-6 (IL-6). In the current model, the generation of IL-6 was clearly shown to originate in the cholestasic liver and induced IL-6 signalling in the brain via endothelial activation, as suggested by an increase in the expression of p-STAT3 in endothelial cells of the hippocampus. These observations support the notion that systemic symptoms such as fatigue, frequently present in chronic cholestasic diseases, may originate from the effects of liver-induced inflammation on the brain. This is an interesting concept that may initiate a new approach for treating the symptoms associated with chronic liver disorders.
The authors were able to modify the IL-6 immune response by manipulating regulatory T cells (Treg). Sickness behaviour in BDL mice was enhanced by depletion of Treg and diminished following Treg infusion. Treg manipulation did not alter liver injury, but was associated with changes in hepatic levels of IL-6 mRNA, plasma levels of IL-6 and peripheral blood mononuclear cell expression of IL-6. No remarkable changes were found in TNFα or IL-1ß levels. The potential role of IL-6 in sickness behaviour development was further investigated in IL-6 knockout BDL mice. These animals showed similar baseline behaviours and a resistance to developing neurological sickness behaviour after BDL. However, they exhibited typical sickness behaviours when injected with IL-6. Previously, there have been attempts to modify the course of primary biliary cirrhosis (PBC) using immunomodulatory therapies [5] that may yet be shown to be beneficial. Infusion of Treg's has been used to hasten immune reconstitution in bone marrow transplantation [6]. However, Treg therapy is not currently understood as a treatment to be applied to PBC patients, but rather as a paradigm to investigate its pathogenesis. The current study validates the BDL mice to investigate the systemic manifestations of cholestasis and supports the neurological origin of sickness behaviour. This interpretation is in accordance with data from PBC patients, in whom an increased plasma concentration of neurosteroids, which activate the GABA-A receptor and result in neuroinhibition, has been found [7]. Similarly, abnormalities in central activation and cortical inhibitory and excitatory circuits have been recently described in fatigued PBC patients in a neurophysiology study [8].
Recently, inflammation has been identified as a factor with important systemic repercussions on liver diseases [9]. Cirrhotic patients are particularly susceptible to bacterial infections which may be followed by systemic complications [10] that can be responsible for patient's demise, in spite of apparent control of the bacterial infection. The systemic inflammatory response appears to be an important mediator that leads to vasodilatation of splanchnic arterial vessels, circulatory impairment, and hepatorenal syndrome [11]. The brain has long been considered an immune-privileged organ protected from systemic inflammation by the blood-brain-barrier, a concept that has recently changed. Systemic infection can induce delirium without direct bacterial infestation of the CNS and without signs of systemic sepsis [12]. Activation of the peripheral immune system has strong effect on the brain through pro-inflammatory cytokines, which can induce delirium in susceptible subjects. Older people appear to be more sensitive due to a more permeable blood-brain-barrier, over-activation of microglia or insufficient cholinergic inhibition in the CNS. Similarly, systemic inflammation can precipitate hepatic encephalopathy in predisposed patients, due to previous small-vessel cerebrovascular disease [13], activated microglia [14] or brain atrophy [15].
Continuous activation of peripheral inflammation can have long-lasting consequences on the brain, as it has been proposed for chronic hepatitis C [16]. The occurrence of infections does increase the risk of Alzheimer's disease or accelerate the progression of established dementia [17], probably because peripheral inflammation causes a continuous activation of microglia. The administration of non-steroidal anti-inflammatory drugs (NSAID) may become a new strategy to prevent dementia; in rheumatoid arthritis NSAID protect against the subsequent development of Alzheimer's disease [18]. In experimental models of liver failure, NSAID ameliorate cognitive disturbances [19]. However, NSAID use is not promoted in cirrhosis, because they may be dangerous due to their effects on renal function and enhanced bleeding risk. The explanation for the beneficial effects of NSAID on cognitive function is that they interfere with mediators that can cause persistent damage through bystander injury to neighbouring neurons. Activation of microglia (the "macrophages" of the brain) appears to play a pivotal role in this process. In patients with cirrhosis, infection, and inflammation are frequent precipitating factors for hepatic encephalopathy that may activate microglia [20]. Cognitive decline has been documented following an episode of encephalopathy [21], even after complete recovery of liver function following liver transplant [22]. In addition, patients with history of hepatic encephalopathy that have undergone liver transplant show permanent cognitive impairment and smaller normalised brain volume with reduced N-acetyl aspartate (an indication of neurone loss) [23].
There are multiple data that show that activation of peripheral inflammation may cause injury to the brain in patients with liver disease. The extent and consequences of the CNS damage depends on multiple factors, including the characteristics of the inflammatory response. The work by Nguyen et al. shows that this inflammatory process can be initiated in the liver and can cause sickness behaviour associated with cholestasic diseases. Although many questions remain unclear and further investigations are required, this study supports the notion that liver-induced inflammation hurts the brain and justifies the evaluation of new immunomodulatory therapies in liver diseases.
--------------------
Regulatory T cells suppress sickness behaviour development without altering liver injury in cholestatic mice
Introduction
Patients with cholestatic liver diseases commonly exhibit a number of associated symptoms including fatigue, malaise, and a loss of interest in engaging in social activity; collectively termed sickness behaviours [1], [5], [11], [30], [34], [42]. Patients with cholestasis (e.g. PBC, PSC, obstructive cholestasis) often describe these symptoms as having the greatest detrimental impact upon their quality of life [5], [11], [30], [40]. These symptoms are major causes of disability and work absenteeism, not to mention the high associated social and economic costs [8], [17], [24]. Moreover, currently available therapies for most cholestatic liver diseases have minimal, if any, impact on most of these sickness behaviours [2], [21].
It is widely accepted that the brain can orchestrate changes in behaviour upon receiving signals which originate outside of the brain, within an inflamed tissue [26], [42]. However, how the periphery and brain communicate in this fashion, and how this process is regulated, are poorly understood. Cytokines, released into the circulation during peripheral immune responses signal the brain to elicit sickness behaviours [15], [16], [42]. In particular, three cytokines have most commonly been implicated in the induction of sickness behaviours; namely, tumour necrosis factor (TNF)-α, interleukin (IL)-1ß, and IL-6 [15], [23]. In the setting of cholestatic liver injury, TNF-α, and IL-6 are most commonly measurable in the peripheral circulation of patients, as well as in animal models of cholestasis [4], [14], [22], [50]. The activation of various immune cells is a tightly regulated process that is orchestrated in part by regulatory T lymphocytes (Tregs) [38], [47]. Tregs are able to control the production of pro-inflammatory cytokines by activated immune cells during peripheral inflammation, and are being investigated clinically as potential therapeutic agents for the treatment of numerous immune-mediated diseases [35]. Therefore, we speculated that Tregs may also be able to modulate periphery-to-CNS immune signalling in the setting of cholestatic liver injury, to alter sickness behaviour development. Through a series of loss and gain of function experiments, we identify an important role for Tregs in the development of sickness behaviours in mice with cholestatic liver injury due to bile duct ligation (BDL), in part, via the regulation of IL-6 production in the liver and by peripheral blood monocytes and subsequent IL-6-driven signalling to the brain.
Discussion
In this study we have identified a novel role for Tregs in suppressing the development of sickness behaviours in BDL mice; independent of overt changes in the severity of liver injury. Our findings suggest a novel pathway whereby Tregs can regulate the development of sickness behaviours by inhibiting the production and release of IL-6 from the liver.
Tregs regulate numerous inflammatory diseases, and have been considered as having potential therapeutic benefit for treating these diseases [38], [47], [53]. Importantly, Treg manipulation may have a potential role in the treatment of liver diseases [53]. BDL surgery in mice results in the development of significant cholestatic liver injury and sickness behaviours, coupled with an innate immune cell driven inflammatory response [6], [48]; associated with a significant reduction in circulating, but no change in liver recruited, Treg numbers compared to sham controls. Somewhat surprisingly, neither depletion nor augmentation of Treg numbers in BDL mice altered the degree of cholestatic liver injury. Therefore, Tregs appear to be relatively ineffective in regulating overt hepatic inflammatory injury in BDL mice, consistent with findings often indicating a failure of increased numbers of Tregs to suppress hepatic inflammation in the clinical setting [9], [41]. In contrast, the development of sickness behaviours in BDL mice was significantly augmented by Treg depletion, and suppressed by Treg infusion. These findings strongly suggest that Tregs are capable of regulating the development of sickness behaviours in the setting of cholestatic liver injury, independent of overt changes in the degree of liver injury.
Tregs are capable of regulating the innate immune response, including the activation of macrophages and their production of cytokines, including IL-6 [46], [52]. Cytokines have been implicated historically in sickness behaviour development, both in humans with inflammatory disease and in animal model disease correlates [15], [16]. Moreover, anti-TNF, and more recently anti-IL-6, neutralizing antibodies have been used clinically to treat inflammatory diseases, and are typically associated with improvements in sickness behaviours (e.g. fatigue, malaise) in these patients; often well before changes in local inflammation severity have been noted [13], [32], [33]. Similarly, inhibition of these cytokines can also improve sickness behaviours in animal models of inflammatory disease [7], [15]. We could not detect circulating IL-1ß in BDL mice, and only a minimal elevation in circulating TNF-α levels were noted in BDL mice which were unchanged with Treg depletion or augmentation. In contrast, BDL mice exhibited a striking increase in hepatic IL-6 mRNA and circulating IL-6 levels compared to shams. Moreover, depletion of Tregs in BDL mice further increased IL-6 levels (both circulating protein, hepatic mRNA), and increasing Treg numbers significantly reduced IL-6 levels; in parallel to changes observed in sickness behaviour development. Interestingly, the effect of Treg manipulation did not appear to be strictly limited to the liver, as depletion of Tregs was also associated with increased circulating monocyte IL-6 production. These observations suggested that Treg-mediated effects on BDL-associated sickness behaviour development are driven by alterations in circulating IL-6 levels. Therefore, we next examined how circulating IL-6 might drive alterations in behaviour in BDL mice.
To explore whether IL-6 is a mediator of sickness behaviour development in BDL mice, we employed IL-6 KO mice. The degree of liver injury was similar in IL-6 KO BDL and wildtype BDL mice, as reflected by serum ALT and bilirubin levels and histology; similar to reports by others [39]. This observation is consistent with our findings that changes in Treg numbers in BDL mice are associated with changes in IL-6 protein and mRNA levels, but not in the degree of liver injury. Furthermore, IL-6 KO BDL mice demonstrated reduced sickness behaviours compared to wildtype BDL mice. These observations suggest that IL-6 is a critical signalling molecule from the liver to the brain in inducing sickness behaviours in BDL mice. IL-6 within the circulation would typically be excluded from the CNS by the blood-brain barrier; however, IL-6 can activate cerebral endothelial cells which could in turn generate secondary signals driving changes in behaviour [37], [49]. Activation of cells by IL-6 leads to activation of STAT3, as reflected by the phosphorylation of STAT3 (i.e. p-STAT3) [28], [37], [49]. Therefore, endothelial cells activated by IL-6 may be quantified by their expression of p-STAT3 [28], [37]. No p-STAT3 expressing endothelial cells were observed in blood vessels within the hippocampus (an area of the brain commonly implicated in the genesis of sickness behaviours) [12], [18] of sham mice. In contrast, BDL mice demonstrated a marked increase in hippocampal endothelial cell p-STAT3 expression; an increase significantly blunted in IL-6 KO BDL mice. Importantly, IL-6 KO BDL mice injected with rmIL-6 demonstrated a p-STAT3 expression profile in hippocampal cerebral endothelial cells similar to that observed in wildtype BDL mice at 4h post-injection [19], [51]. Furthermore, overt sickness behaviours developed in IL-6 KO BDL mice injected with rmIL-6, whereas this did not occur in saline injected IL-6 KO BDL mice. These observations strongly suggest that IL-6 released into the circulation in BDL mice is a critical step in signalling the brain, at least in part, by activating cerebral endothelium, ultimately leading to the development of sickness behaviours.
In summary, the present study identifies Tregs as a novel modulator of sickness behaviour development during experimental cholestatic liver injury; an effect mediated mainly by Treg-driven suppression of hepatic IL-6 production and release, and not through the attenuation of liver inflammation or injury. Moreover, our findings suggest that Treg administration, or possibly inhibition of IL-6 signalling, could potentially be used for the treatment of severe or refractory sickness behaviours in patients with cholestatic liver disease.
Background & Aims
Cholestatic liver diseases are commonly accompanied by debilitating symptoms, collectively termed sickness behaviours. Regulatory T cells (Tregs) can suppress inflammation; however, a role for Tregs in modulating sickness behaviours has not been evaluated.
Methods
A mouse model of cholestatic liver injury due to bile duct ligation (BDL) was used to study the role of Tregs in sickness behaviour development.
Results
BDL mice developed reproducible sickness behaviours, as assessed in a social investigation paradigm, characterized by decreased social investigative behaviour and increased immobility. Depletion of peripheral Tregs in BDL mice worsened BDL-associated sickness behaviours, whereas infusion of Tregs improved these behaviours; however, liver injury severity was not altered by Treg manipulation. Hepatic IL-6 mRNA and circulating IL-6 levels were elevated in BDL vs. control mice, and were elevated further in Treg-depleted BDL mice, but were decreased after infusion of Tregs in BDL mice. IL-6 knock out (KO) BDL mice exhibited a marked reduction in sickness behaviours, compared to wildtype BDL mice. Furthermore, IL-6 KO BDL mice injected with rmIL-6 displayed sickness behaviours similar to wildtype BDL mice, whereas saline injection did not alter behaviour in IL-6 KO BDL mice. BDL was associated with increased hippocampal cerebral endothelial cell p-STAT3 expression, which was significantly reduced in IL-6 KO BDL mice.
Conclusions
Tregs modulate sickness behaviour development in the setting of cholestatic liver injury, driven mainly through Treg inhibition of circulating monocyte and hepatic IL-6 production, and subsequent signalling via circulating IL-6 acting at the level of the cerebral endothelium.
Results
Characterization of mouse model of cholestatic liver injury
BDL mice had increased serum ALT (sham 16.5±1.7U/L vs. BDL 419.8±58.1U/L; n=4/group, p<0.001) and total bilirubin (sham 1.8±0.5μmol/L vs. BDL: 230.3±14.7μmol/L; n=4 mice/group, p<0.001) levels compared to shams. Hematoxylin and eosin (H&E) stained liver sections demonstrated portal based inflammatory cell infiltrates in BDL mice (absent in sham mice; Supplementary Fig. 1A).
BDL mice exhibit reproducible sickness behaviours
BDL mice displayed overt sickness behaviours as reflected by a significant reduction in time spent in social interaction (Fig. 1A); however, total number of social interaction attempts were similar in both groups (Fig. 1B). In addition, BDL mice were more immobile than sham mice (Fig. 1C).
Sickness behaviours in BDL mice are enhanced by Treg depletion, and improved with Treg infusion
Treg depletion in peripheral blood was confirmed by FACS (IgG-treated BDL: 1.2±0.09% or 1.4x103±0.1x103cells/ml of blood vs. anti-CD25-treated BDL: 0.6±0.04% or 0.6x103±0.07x103cells/ml of blood; n=4-5mice/group, p<0.01). Treg-depleted BDL mice exhibited increased sickness behaviours; decrease in the total time spent in social interactive behaviours, fewer social interactions, and increased immobility (Fig. 1). Treg depletion did not alter the severity of BDL-induced cholestatic liver injury (Supplementary Fig. 1B and E).
In contrast, Treg-infused BDL mice demonstrated less sickness behaviours; increase in the total time spent in social exploration behaviours, and a decrease in immobility time, compared to control cell infused BDL mice (Fig. 1A and C). In contrast, no significant difference was observed in the total number of social interactions (Fig. 1B). The severity of cholestatic liver injury was similar in Treg and control cell infused BDL mice (Supplementary Fig. 1C and E).
Manipulations of Treg numbers are paralleled by changes in serum and hepatic IL-6 levels in BDL mice
Serum levels of TNF-α, IL-1ß, and IL-6 were assessed in sham, BDL, Treg-depleted, and Treg-infused BDL mice. Serum TNF-α levels were detectable at the lower limit of assay detectability and were slightly increased in BDL vs. sham mice; however, serum TNF-α levels were unchanged in Treg-depleted and non-depleted BDL mice (data not shown). In contrast, serum IL-1ß was undetectable in all groups of mice. However, serum IL-6 levels were ~20-fold higher in BDL compared to sham mice (p<0.01) (Fig. 2A). Treg-depleted BDL mice demonstrated a ~2-fold increase in serum IL-6 levels compared to non-Treg depleted BDL mice, and infusion of Tregs into BDL mice resulted in a significant decrease in serum IL-6 levels, to levels below those documented in BDL mice and in BDL mice depleted of Tregs (Fig. 2A).
The liver is a main source of circulating IL-6, and hepatic IL-6 expression is increased in BDL mice [6]. Moreover, hepatic IL-6 mRNA and circulating IL-6 protein levels are increased in patients with the cholestatic liver disease PBC [25], [29]. BDL mice demonstrated a significant ~4-fold increase in hepatic IL-6 mRNA expression compared to sham mice (p<0.001). Changes in hepatic IL-6 mRNA levels paralleled changes in serum IL-6 concentrations in response to decreasing or increasing relative numbers of peripheral Tregs in BDL mice (Fig. 2B). In addition to increased hepatic IL-6 mRNA expression, depletion of peripheral Tregs in BDL mice also resulted in an increase peripheral blood mononuclear cell expression of IL-6 when compared to non-Treg depleted BDL mice (Supplementary Results).
IL-6 knockout BDL mice exhibit a striking reduction in the development of sickness behaviours
The potential role of IL-6 in sickness behaviour development in BDL mice was examined through social exploration studies involving IL-6 KO mice, compared to wildtype BDL mice. Baseline behavioural assessments were similar in un-operated IL-6 KO and wildtype mice. IL-6-deficient BDL mice spent more time in social investigation than wildtype BDL mice (Fig. 3A); however, the total number of social interactions in these two groups were similar (Fig. 3B). In addition, IL-6 KO BDL mice exhibited less immobility than wildtype BDL mice (Fig. 3C). The severity of cholestatic liver injury was similar in IL-6 KO and wildtype BDL mice (Supplementary Fig. 1D and E).
rmIL-6 Injection induces sickness behaviours in IL-6 KO BDL mice
IL-6 KO BDL mice injected with rmIL-6 developed significant sickness behaviours at 2h post-injection, as reflected by decreased total time of social investigation (Fig. 3A) and number of social interactions (Fig. 3B), as well as by an increase in total time of immobility (Fig. 3C) (similar to observations in wildtype BDL mice), in comparison to IL-6 KO BDL mice that were injected with saline (exhibited sickness behaviours similar to non-injected IL-6 KO BDL mice).
IL-6 mediates activation of STAT3 in brains of BDL mice
We addressed the possibility that circulating IL-6 signalled the brain via the cerebral endothelium, to bring about changes in behaviour. IL-6 activates endothelium through stimulation of STAT3, generating p-STAT3 [36], [37], [49]. Therefore, we examined p-STAT3 expression within the hippocampus in general (by Western blot), and within cerebral endothelium of hippocampal blood vessels (by immunohistochemistry) of BDL vs. sham mice. The hippocampus is an area of the brain commonly implicated in sickness behaviour regulation [12], [18], [27]. Western blotting of total hippocampal protein homogenates revealed a significant increase in the p-STAT3/STAT3 ratio in BDL vs. sham mice (Supplementary Fig. 3B). Hippocampal STAT3/actin (loading control) ratios were similar in sham and BDL mice (Supplementary Fig. 3A).
In addition, we determined endothelial p-STAT3 expression (by immunohistochemistry) in hippocampal brain sections of BDL and sham mice. Numerous p-STAT3 positive staining endothelial cells were identified within hippocampal blood vessels in BDL mice, whereas none were evident in sham mice (Fig. 4A). The percentage of endothelial p-STAT3 positive staining hippocampal blood vessels was significantly reduced in IL-6 KO BDL (and similarly in saline injected IL-6 KO BDL mice) compared to wildtype BDL mice, but the percentage of p-STAT3 positive staining hippocampal blood vessels was increased in rmIL-6 injected IL-6 KO BDL to levels similar to those observed in wildtype BDL mice (Fig. 4B).
|
|
| |
| |
|
|
|
