Reduced gut microbiome protects from alcohol-induced neuroinflammation and alters intestinal and brain inflammasome expression

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.

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].

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.

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 ΔCt was calculated using the average Ct value of each sample duplicate and subtracting the average ΔCt 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.

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.

While the modulating effects of chronic alcohol administration have been studied in the gut microbiome, alcoholic liver disease, and neuroinflammation, it is unclear how shorter alcohol use and/or alcohol binge affect inflammation signaling in the CNS and what role the gut microbiome plays in this process. In this study, mice received 5% alcohol (EtOH) in a liquid diet for 10 days (after a 5-day liquid diet acclimation period), followed by a one-time alcohol binge or a calorie-matched pair-fed (PF) diet [29]. Female mice were chosen because they have greater sensitivity to alcohol, and previous studies have focused on female animals [26‐28]. To elucidate the importance of the gut microbiome in the translocation of pathogen-associated molecular patterns (PAMPs) from the intestine to extra-enteric organs, we used oral administration of a cocktail of antibiotics (ampicillin, neomycin, vancomycin, and metronidazole) to drastically reduce the bacterial load in the gut (Fig. 1a). Oral antibiotic treatment (Abx) caused a significant reduction in endotoxin in the circulation at the time of sacrifice both in pair-fed and alcohol-fed mice (Fig. 1b). The expression of 16S bacterial DNA, measured from mice stools collected immediately prior to sacrifice, was dramatically reduced by antibiotic treatment (Fig. 1c, d). Stool bacteria cultured on non-selective agar plates also revealed almost complete elimination of culturable colonies after 5 days of antibiotic treatment (Fig. 1e). Some recovery of bacteria in the stool was observed by the conclusion of the 15-day study, likely due to the development of antibiotic resistance (Fig.1e). However, bacterial colony-forming units (CFUs) were dramatically reduced in the stool obtained on the day of sacrifice in antibiotic-treated animals compared with untreated mice (Fig.1f). Together, these data indicate that antibiotic treatment successfully suppressed gut bacterial load and reduced circulating endotoxin in both pair- and alcohol-fed mice.

Chronic alcohol induces circulating proinflammatory cytokines in both animal models and in human patients [30, 31]. To determine whether this systemic cytokine induction also occurs in the acute-on-chronic model in mice, we measured circulating TNFα and IL-6 in the serum (Fig. 2a). While alcohol did not induce statistically significant increases in either cytokine, antibiotic treatment significantly reduced circulating TNFα in both pair-fed and alcohol-fed mice (Fig. 2a).

Chronic alcohol use results in neuroinflammation both in humans and in mice [7, 12]. We found that 10 days of chronic alcohol feeding followed by a one-time binge in mice, a model of acute-on-chronic alcohol consumption not previously used to study neuroinflammation, induced significantly higher expression of proinflammatory cytokine genes including Mcp-1, Hmgb1, and Il-17 and non-significant trends toward increased expression of Tnfα and Il-23 in the brain cortex (Fig. 2b). Tnfα, Mcp-1, Hmgb1, Il-17, and Il-23 are proinflammatory cytokines that can be released by multiple cell types, and each has previously been associated with alcohol-induced neuroinflammation [9, 14, 32, 33]. Alcohol did not induce expression of Il-6 or Cox2. Interestingly, Il-6 was induced in pair-fed antibiotic-treated mice compared to non-treated mice, and alcohol feeding reduced this induction (Fig. 2b).

Previous studies indicate that antibiotic treatment that reduces intestinal bacterial load also reduces alcohol-induced inflammation in the liver [23]. Here, we hypothesized that translocation of gut bacterial products to the CNS contributes to alcohol-induced neuroinflammation and that this process is regulated by the gut microbial load. Therefore, we sought to investigate whether gut decontamination could protect from neuroinflammation associated with alcohol consumption. We observed that the proinflammatory cytokine expression increase in the cortex in alcohol-fed mice compared to PF controls was markedly reduced in mice treated with Abx (Fig. 2b). Antibiotic treatment fully prevented alcohol-related induction of Mcp1, Il-17, and Il-23 mRNA expression in the cortex. Tnfα was induced in antibiotic-treated, alcohol-fed mice compared to antibiotic-treated pair-fed mice, but its expression was still significantly lower compared to alcohol-fed mice without antibiotic treatment. Expression of Tnfα, Mcp1, Il-17, and Il-23 was also reduced in the cortex of antibiotic-treated pair-fed mice compared to those without antibiotic treatment. These results indicate that acute-on-chronic alcohol feeding in mice increases proinflammatory cytokine induction that is prevented by the reduction in gut-derived PAMPs and the gut microbiome.

Because we found that multiple proinflammatory cytokines were reduced in the cortex of antibiotic-treated mice (Fig. 2), we next measured inflammasome-related transcripts to elucidate if alcohol or antibiotics influenced inflammasome-mediated cytokine expression. The inflammasome is a multiprotein complex containing NOD-like receptors (NLRs, including NLRP3) that can sense pathogens and danger signals, the adaptor molecule, ASC, and the effector molecule, caspase-1. Inflammasome activation leads to cleavage of pro-IL-1β and pro-IL-18 to their respective bioactive forms, IL-1β and IL-18 [9]. We found that although alcohol did not induce IL-1β, antibiotic treatment increased circulating serum IL-1β in pair-fed mice (p < 0.05) and trended toward an increase in alcohol-fed mice (p = 0.055) (Fig. 3a). Interestingly, although chronic alcohol consumption models have led to increased expression of inflammasome components and Il-1β [9], we found no significant increase in alcohol-induced Il-1β mRNA expression in this acute-on-chronic alcohol model (Fig. 3b). However, cortical Il-1β mRNA expression in antibiotic-treated pair-fed mice was significantly increased and we observed an increasing trend in Il-1β in antibiotic-treated alcohol-fed compared to untreated mice. Interestingly, in antibiotic-treated mice, alcohol administration significantly increased Il-1β mRNA expression compared to pair-fed mice. Expression of Il-18 was induced in alcohol-fed mice in the cortex, and similar to the increase in pair-fed Il-1β, we also found that Il-18 and Asc were elevated in antibiotic-treated, PF mice compared to untreated PF mice (Fig. 3b). Acute-on-chronic alcohol administration reduced the expression of Nlrp3 and Asc and increased the expression of Il-18 in untreated alcohol-fed mice compared to untreated PF controls. Asc and Il-18 mRNA expression were reduced in antibiotic-treated compared to untreated alcohol-fed mice (Fig. 3b). Caspase-1 mRNA levels did not change significantly in any of the treatment groups (Fig. 3b). These observations suggest that regulation of the inflammasome and IL-1β depends on the gut microbiome and is minimally influenced in the acute-on-chronic alcohol model in mice.

To characterize the effects of the acute-on-chronic alcohol model in the CNS, we next examined the microglia activation. Microglia are the resident macrophages of the CNS capable of expressing proinflammatory cytokines in response to an insult, such as alcohol [34]. Activated microglia are characterized by altered cell morphology, taking on an amoeboid shape with enlarged cell bodies (soma) and shortened peripheral processes [35]. We used immunohistochemistry to identify IBA1-positive microglia (representative images shown in Fig. 4a, b). The soma size and length of cell extensions off the soma were measured in the cortical and hippocampal microglia in all treatment groups and normalized to PF mice. No significant differences in soma size were observed in the cortex (Fig. 4c). Investigation of the sub-regions of the hippocampus, such as the CA1, CA3, and dentate gyrus (DG) areas, revealed that alcohol increased the soma area only in the microglia of the CA3 region. There was no change in the soma area for CA3 microglia in EtOH-fed mice compared to PF controls that were both treated with antibiotics (Fig. 4d). Importantly, we found that alcohol reduced the total process length compared to pair-fed mice in the cortex (Fig. 4e), consistent with the condensed cell morphology characteristic of microglial activation [35]. Antibiotic treatment eliminated this alcohol-induced reduction in process length in cortical microglia. Hippocampal microglia process length in alcohol-fed mice was significantly reduced compared to pair-fed controls in all regions investigated, and as in the cortex, antibiotic treatment eliminated this morphological change (Fig. 4f). The number of microglia in the cortex was not changed in EtOH-fed compared to PF mice in either treatment group, although antibiotic treatment in PF mice modestly reduced the number of cortical microglia compared to untreated PF mice (Fig. 4g). There was no change in microglial numbers in the hippocampus (Fig. 4h).

The alcohol-induced changes we observed in the brain could be due to a loss of integrity of the gut barrier. Previous studies have shown that intestinal cytokine expression can reduce gut barrier integrity and may allow leakage of pathogen-associated molecules from the intestinal lumen into the systemic circulation [36]. Therefore, we measured the intestinal expression of various proinflammatory cytokines and found that they were increased after the acute-on-chronic alcohol administration compared to calorie-matched pair-fed mice (Fig. 5a). The expression of Tnfα, Mcp1, and Hmgb1 mRNA was significantly increased in the small intestine following alcohol consumption, and Il-17 and Il-23 expression also showed an increasing trend in EtOH mice. Treatment with the antibiotic cocktail reduced the bacterial load in the intestine (Fig. 1) and led to significantly attenuated alcohol-induced Mcp1 and Hmgb1 mRNA levels. Antibiotic treatment reduced the baseline expression of the inflammatory cytokines including Tnfα, Il-17, and Il-23 in PF mice compared to untreated PF mice (Fig. 5a). Interestingly, even with antibiotic treatment, alcohol feeding still increased the expression of Tnfα, Il-17, and Il-23 in the small intestine of antibiotic-treated alcohol-fed mice compared to antibiotic-treated pair-fed mice (Fig. 5a).

Recent research has highlighted an important connection between the intestinal microbiome and inflammasomes [37], particularly the NLRP3 inflammasome [38]. Therefore, we investigated whether antibiotic decontamination of the gut impacted the expression of inflammasome components in the small intestine. Alcohol induced expression of Il-1β, Nlrp3, and Asc compared to pair-fed controls (Fig. 5b). Antibiotic treatment abrogated the alcohol induction of Il-1β, Nlrp3, and Asc, and antibiotics also increased the baseline expression in pair-fed mice of Il-1β, Il-18, Asc, and Casp1 (Fig. 5b).