Background
Prolonged alcohol consumption leads to translocation of gut bacterial components, such as endotoxin, from the intestinal lumen into the circulation [
1‐
3]. Once absorbed, alcohol along with gut-derived endotoxin is delivered via the portal circulation to the liver where metabolism begins and an inflammatory cascade is initiated. However, endotoxin, unmetabolized alcohol, and alcohol metabolites also pass through the liver and reach the systemic circulation and other organs, including the peripheral immune system and the central nervous system (CNS). While previous studies have investigated the direct effects of alcohol on the brain [
4‐
6], little is known about the role of gut-derived microbial products and their impact on the nervous system and neuroinflammation.
Microglia play a critical role in sensing and responding to alcohol consumption and are involved in multiple immune signaling pathways [
7‐
10]. Microglia express Toll-like receptor 4 (TLR4), a pattern recognition receptor critical in alcohol-induced neuroinflammation [
11‐
13] as well as the NLR family pyrin domain containing 3 (NLRP3) inflammasome [
9]. Previous studies showed that TLR4 knockout mice are protected from increased cytokine expression in various regions of the brain and from increased activation of microglia [
14‐
16]. TLR4 recognizes endogenous danger signals such as HMGB1 [
17,
18] and is the major pattern recognition receptor of bacterial endotoxin (also known as lipopolysaccharide (LPS)) [
19]. Although endotoxin is not generally believed to cross the blood-brain barrier [
20], data from TLR4 knockout mice suggests that signaling through TLR4 is an important component influencing alcohol-induced neuroinflammation. Neuroinflammation is mediated by the inflammasome complex, a multiprotein complex that senses pathogens and danger signals leading to cleavage and release of proinflammatory IL-1β and IL-18 [
9].
LPS signaling is also a critical component of liver pathology associated with alcohol consumption. Alcohol metabolism leads to cell stress, hepatocyte damage, and release of sterile danger signals in the liver [
21,
22]. Endotoxins, derived from the intestinal microbiome into the portal circulation, are recognized by pattern recognition receptors such as TLR4 and initiate an inflammatory response secondary to the hepatocyte stress and damage caused by the release of reactive oxygen species and other cellular stresses induced by alcohol metabolism. Interestingly, we and others have shown that treating mice with antibiotics to reduce the bacterial load in the gastrointestinal tract (and thereby reducing endotoxin levels) attenuates liver inflammation and steatosis after alcohol use [
23‐
25]. This reduction in gut bacterial load could ameliorate the alcohol-induced changes in the brain.
To further explore the critical role of the gut microbiome in the gut-brain axis, we used antibiotics to reduce the intestinal bacterial load in mice. Following acute-on-chronic alcohol consumption in mice (10 days of alcohol followed by an acute alcohol binge), we show that alcohol induces neuroinflammation in the CNS and also increases cytokine expression in the small intestine. Inflammation in both organs was attenuated with antibiotic-induced microbiome reduction. Interestingly, although cytokine expression was reduced, antibiotic treatment induced the mRNA expression of inflammasome components and cytokines processed by the inflammasome in the CNS and intestine. These results show for the first time that manipulation of the gut microbiome via reduction of the microbial load protects from alcohol-induced CNS and intestinal inflammation. Our study provides important insights into the interactions of the intestinal microbiome and brain in the gut-brain axis induced by alcohol.
Methods
Mouse alcohol feeding
All animal studies were approved by the Institutional Animal Care and Use Committee at the University of Massachusetts Medical School (UMMS). Wild-type C57BL/6J 6- to 8-week-old female mice were purchased from Jackson Laboratories and co-housed in the UMMS Animal Medicine Facility. Female mice were chosen because they are more susceptible to alcohol-induced liver injury than male mice [
26‐
28]. Alcohol feeding followed the acute-on-chronic model previously described by Bertola et al. [
29]. Briefly, all mice received the pair-fed Lieber-DeCarli (Bio-Serv) liquid diet for 5 days. Some mice then received 5% alcohol and maltose dextran in a liquid diet while pair-fed mice remained on the control liquid diet. Pair-fed mice were calorie-matched with the alcohol-fed mice. Nine hours prior to sacrifice, alcohol-fed mice received alcohol via oral gavage (5 g kg
−1 body weight) and pair-fed mice received isocaloric maltose dextran.
Antibiotic treatment
Mice were either treated twice daily with an oral intragastric gavage of water or a broad spectrum antibiotic cocktail (Abx) containing ampicillin (100 mg/kg body weight (BW); Sigma), neomycin (100 mg/kg BW; Gibco), metronidazole (100 mg/kg BW; Sigma), and vancomycin (50 mg/kg BW; Sigma). Gavages began on the first day of liquid diet and continued daily until the completion of alcohol feeding. Significant reduction in bacterial load was confirmed by bacterial culture (described below) similar to previous reports [
23].
Bacterial culture
Mouse feces were collected directly from the anus and suspended in thioglycolate media. Suspensions were plated on non-selective LB agar plates (EMD Millipore) and incubated for 24 h at 37 °C for assessment of bacterial load reduction.
qPCR analysis
RNA extraction from the small intestine and brain cortical tissue was performed using miRNeasy Extraction Kit (Qiagen) according to the manufacturer’s instructions, including on-column DNase digestion (Zymo Research). Reverse transcription for cDNA was completed from 1 μg of RNA and subsequent 1:5 dilution in nuclease-free water. Real-time qPCR using SYBR Green (BioRad) was performed according to the manufacturer’s instructions. RT-qPCR primers are listed in Table
1, and
18S mRNA expression was used as a housekeeping gene for 2
−ΔΔCt method of RNA expression analysis. For 16S comparison between antibiotic-treated and non-treated animals, stool bacterial DNA was extracted using QIAamp DNA Stool Mini Kit (Qiagen) according to the manufacturer’s protocol. After running a qPCR reaction using 16S primers similar to described above, a Δ
Ct was calculated using the average
Ct value of each sample duplicate and subtracting the average Δ
Ct of untreated pair-fed mice. The bacterial 16S PCR product was run on a 1% agarose gel to visualize the relative reduction in bacterial load.
Table 1
Real-time PCR primers
18S
| GTAACCCGTTGAACCCCATT | CCATCCAATCGGTAGTAGCG |
16S
| TCCTACGGGAGGCAGCAGT | GGACTACCAGGGTATCTAATCCTGTT |
Tnfα
| GAAGTTCCCAAATGGCCTCC | GTGAGGGTCTGGGCCATAGA |
Mcp-1
| CAG GTC CCT GTC ATG CTT CT | TCTGGACCCATTCCTTCTTG |
Il-1β
| TCTTTGAAGTTGACGGACCC | TGAGTGATACTGCCTGCCTG |
Il-17
| CAGGGAGAGCTTCATCTGTGT | GCTGAGCTTTGAGGGATGAT |
Il-23
| AAGTTCTCTCCTCTTCCCTGTCGC | TCTTGTGGAGCAGCAGATGTGAG |
Hmgb1
| CGCGGAGGAAAATCAACTAA | TCATAACGAGCCTTGTCAGC |
Il-6
| ACAACCACGGCCTTCCCTACTT | CACGATTTCCCAGAGAACATGTG |
Cox2
| AACCGAGTCGTTCTGCCAAT | CTAGGGAGGGGACTGCTCAT |
Nlrp3
| AGCCTTCCAGGATCCTCTTC | CTTGGGCAGCAGTTTCTTTC |
Asc
| GAAGCTGCTGACAGTGCAAC | GCCACAGCTCCAGACTCTTC |
Casp1
| AGATGGCACATTTCCAGGAC | GATCCTCCAGCAGCAACTTC |
Il-18
| CAGGCCTGACATCTTCTGCAA | TCTGACATGGCAGCCATTGT |
Serum cytokine measurement
Mice were cheek-bled prior to sacrifice, and serum was isolated. TNFα and IL-6 (Biolegend, San Diego, CA, USA) and IL-1β (R&D Systems, Minneapolis, MN, USA) were measured by ELISA.
Immunohistochemistry
Following sacrifice, brain tissue was dissected and fixed in 10% formalin overnight before paraffin embedding. Immunohistochemical staining was completed at the UMMS Morphology Core using anti-ionized calcium-binding adapter molecule (IBA1) antibody (Wako; 1:1000) and subsequently labeled with streptavidin-biotin immunoenzymatic antigen for detection with 3,3′-diaminobenzidine (DAB) (UltraVision Mouse Tissue Detection System Anti-Mouse HRP/DAB; Lab Vision). Images were acquired from the described CNS areas by light microscopy (cortex; CA1, CA3, and DG of the hippocampus) at × 40 magnification for process length and cell body size measurements of microglia using ImageJ. Cell process length for each microglial cell was measured by tracing all extensions off of the soma to their distal termination using ImageJ’s freehand measuring tool. For each microglia, the length of all processes was summed to obtain the total cell process length. The soma area was measured by tracing the perimeter of the cell body and measuring the contained area using ImageJ’s freehand tracer and the area measurement function. Microglia were analyzed from five to nine images taken randomly from each CNS region from each mouse. The investigator was blinded to the sample groups during staining, image acquisition, and ImageJ analysis. IBA1 positivity was measured using the Color Deconvolution plug-in in ImageJ.
Statistical analysis
Statistical analysis was carried out using GraphPad Prism Version 7.0 using Mann-Whitney test. p < 0.05 was considered statistically significant. Outlier exclusion was calculated using Grubbs’ outlier test with alpha set to 0.05.
Discussion
In this study, we show that acute-on-chronic alcohol administration results in the central nervous system and small intestinal inflammation and that reducing the gut microbial load with antibiotics protects against alcohol-induced neuroinflammation. The cocktail of oral antibiotics dramatically reduced the gut bacterial load and circulating endotoxin levels. Alcohol-induced neuroinflammation, including microglial morphologic changes and proinflammatory gene expression, was significantly attenuated in oral antibiotic-treated mice, providing novel evidence for the importance of gut bacterial load and PAMPs in the gut-brain axis in alcohol use. We also describe increased proinflammatory cytokine expression in the small intestine after alcohol consumption that can be reduced by treatment with intragastric antibiotics that drastically reduced the bacterial load in the intestine. Interestingly, reduction in the gut microbiome was associated with increased expression of inflammasome components in both the CNS and intestine.
Previously, we have shown that antibiotic treatment in the acute-on-chronic alcohol model protects the liver from alcohol-induced inflammation (including cytokine expression), immune cell infiltration, and steatosis [
23]. In the present study, we found evidence of microglial activation by acute-on-chronic alcohol administration in mice. CNS proinflammatory cytokine expression was increased, and average cell process length was decreased in EtOH mice indicating microglia activation. Activated microglia take on an amoeboid-like morphology with reduced process length and, typically, an increased soma size [
35]. Acute-on-chronic alcohol reduced cell process length in both the cortex and hippocampus and significantly increased the soma size in part of the hippocampus. Interestingly, although acute-on-chronic alcohol-induced proinflammatory cytokine expression in the CNS, alcohol feeding did not increase circulating levels of TNFα, IL-6, and IL-1β. This indicates that alcohol-induced neuroinflammation may occur independent of systemic inflammation, although further investigation of other peripheral signals will be necessary to rule out contributions from circulating factors.
Similar to observations in the liver [
23], antibiotic gut decontamination protected the CNS from proinflammatory gene expression and changes in the resident macrophage population. Interestingly, germ-free mice do not show the same protection from alcohol-induced liver damage that we have previously described using antibiotic decontamination [
39]. A possible explanation for these different observations is that some baseline bacterial load and/or presence of bacteria during development is critical for the alcohol-induced response of the immune system as well as for organ-specific immunity. Indeed, previous research has highlighted a role for antibiotic treatment during development in affecting the function of adaptive immune cells [
40]. Although multiple studies demonstrated alcohol-induced neuroinflammation after chronic, prolonged alcohol administration in mice and rats, here, we show that a 10-day alcohol feeding followed by an acute binge also results in alcohol-related neuroinflammation. Furthermore, this NIAAA model of alcohol administration results in common end-organ effects of inflammation on the brain, small intestine and liver.
Our data are consistent with previous studies examining the role of TLR4 signaling in alcohol-related organ pathology. While some have suggested that alcohol may interact directly with TLR4 or affect lipid membrane interactions required for proper TLR4 signal transduction [
41,
42], TLR4 also recognizes endogenous (including HMGB1) [
17,
18] and exogenous (i.e., bacterial components such as LPS) [
19] danger signals. Studies show that TLR4 knockout and knockdown mice are protected from numerous inflammation-related sequelae of alcohol exposure in the liver [
43] and in the brain [
14‐
16]. Rather than focusing on TLR4 and its signaling pathway, we used antibiotics to reduce bacterial LPS, one of the prominent ligands of TLR4, and reveal a similar reduction in tissue inflammation from the gut to the brain. Our study adds critical evidence to the understanding of the gut-brain axis that relates multifocal pathology in the body after chronic alcohol exposure.
An important remaining question is whether gut bacteria or their products are primarily responsible for organ damage. A direct link between LPS and organ inflammation is possible; leakage of live or dead bacteria or bacterial-derived products into the systemic circulation has been documented in various alcohol administration settings [
1,
2,
44,
45]. These bacterial signals could be directly responsible for inducing inflammation in the gut and in the brain, as well as the associated organ damage. Although LPS does not cross the blood-brain barrier at significant levels [
20], it could be interacting with juxta-cerebrovascular cells to transmit an immune signal across the barrier. Evidence of blood-brain barrier disruption in alcohol models and human patients provides another explanation for a possible direct mechanism of LPS-induced neuroinflammation [
46]. Alternatively, gut-derived signals, such as LPS, bacterial metabolites, or other undescribed intestinal signals, could lead to a systemic reaction. This reaction could include inflammatory cytokines or activated immune cells in the liver or in the circulation that then induce organ-specific inflammation in the CNS and elsewhere in the body. In the present study, we did not detect alcohol-induced increases in circulating TNFα, IL-6, or IL-1β which suggests that alcohol-induced neuroinflammation can be induced by alcohol in the absence of systemic cytokine increases. Developing models to investigate possible peripheral signaling to the CNS leading to neuroinflammation will be a critical area of further study to explain inter-organ communication after alcohol consumption.
Our data supports previous studies showing that alcohol can induce inflammatory signaling in the intestine. This inflammation may be a key factor in the breakdown of the intestinal barrier integrity and ensuing leakage of bacterial products into the circulation associated with alcohol. Using both in vitro and in vivo models, Al-Sadi et al. have shown that proinflammatory cytokines are capable of reducing tight junctions and gut barrier integrity, leading to breakdown and molecule translocation across the gastrointestinal tract [
47‐
49]. Other mechanisms of alcohol-induced loss of intestinal barrier integrity have been explored and include bacterial dysbiosis [
50,
51], luminal homeostasis [
45,
52], enterocyte cellular stress, and dysregulation of structural proteins [
53]. Furthermore, the relationship between proinflammatory gene expression and gut barrier dysfunction appears to be critical [
36,
54], and our data further emphasize the role of alcohol and intestinal bacteria in regulating intestinal cytokine levels.
Acknowledgements
The authors thank Dr. Liu of the University of Massachusetts Morphology Core, Karen Kodys, Donna Catalano, and Jeeval Mehta for their technical assistance as well as Candice Dufour and Melanie Trombly for their assistance in preparing the manuscript.