Bacterial dysbiosis and decrease in SCFA correlate with intestinal inflammation following alcohol intoxication and burn injury

Animals

A total of 10–12-week-old C57BL/6 male and female mice were obtained from Charles River Laboratories and maintained in animal housing facilities at Loyola University Chicago Health Sciences Division, Maywood, Illinois, USA. All animal experiments were conducted in accordance with the guidelines set forth in the Animal Welfare Act and were approved by the Institutional Animal Care and Use Committee at Loyola University Health Sciences Division.

Murine model of acute alcohol intoxication and burn injury

As previously described, a well-established model of acute alcohol intoxication and scald burn injury was used.10 11 14 27–29 Briefly, mice were randomly assigned to two experimental groups: Sham injury and vehicle (water) or burn injury and ethanol. On the day of injury, mice were gavaged with 400 μL of 25% ethanol in water (2.9 g/kg) or 400 μL of water. Mice were given 1 mg/kg buprenorphine subcutaneously. Four hours following the gavage, mice were anaesthetised with a ketamine hydrochloride/xylazine cocktail (~80 mg/kg and ~1.2 mg/kg, respectively) given by intraperitoneal injection. The dorsal surface of the mice was shaved before placing the mice in a prefabricated template exposing ~12.5% of total body surface area (TBSA) calculated using Meeh’s formula.30 Burn group animals were immersed in ~85°C water bath for ~7 s to induce a full-thickness scald burn injury. Sham animals were placed in a 37°C water bath for 7 s. Following burn or sham injury, animals were dried gently and given 1.0 mL normal saline resuscitation by intraperitoneal injection. Animals were returned to their cages and allowed food and water ad libitum and monitored.

16S bacterial DNA sequencing

One day after injury, mice were euthanised and distal SI and caecum lumen contents were collected. Faecal bacterial DNA was isolated using the PowerFecal Pro DNA Kit (Qiagen). DNA concentration and quality were assessed via a NanoDrop 2000 Spectrophotometer (Thermo Scientific). 16S rRNA PCR and sequencing were performed at the Loyola Genomics Facility. Briefly, the V4 region of the 16S rRNA gene was amplified with 27f and 1492r primers.31 Sequencing was performed on an Illumina MiSeq using a V3 600 bp flowcell, yielding 300 bp paired end reads. Raw FASTQ files were analysed using QIIME2 and publicly available plugins. In short, sequences were demultiplexed using the demux plugin and then denoised for quality control using DADA2. Alpha and beta diversity metrics were calculated using the core-metrics-phylogenetic pipeline of the q2-diversity plugin. Features were annotated for taxonomic classification using the classify-sklearn plugin with a naïve Bayes classifier trained on reference sequences from the SILVAv132 database. Subsequent bacterial taxa and their read counts were used for further analysis. Demultiplexed paired-end sequences were deposited in the NCBI Sequence Read Archive (BioProject accession ID: PRJNA1207030).32

Sequencing data analysis

Using QIIME2, within-sample variance (alpha diversity) was assessed by calculating Shannon’s diversity index for each sample, which considers both the number of taxa present (richness) and the distribution of their abundances (evenness). To evaluate the contributions of richness and evenness separately, Chao1 and Pielou’s were calculated, respectively. Differences between male versus female and ethanol burn (EB) versus sham vehicle (SV) were determined using the non-parametric Kruskal-Wallis significance test. Variance between samples (beta diversity) was measured by weighted UniFrac with principal coordinate analysis (PCoA) and assessed using PERMANOVA statistical analysis. Bacterial taxa were analysed by raw read counts (phyla and families) and relative abundance (phyla) via GraphPad Prism. Two samples were removed due to low total read counts (<100), then outliers were assessed and removed. Data are presented as mean raw reads±SEM or mean relative abundance and analysed via Kruskal-Wallis with Dunn’s multiple test or Mann-Whitney U test, where applicable.

SCFA quantification

A small portion of each caecal content sample was snap frozen and sent to the University of Illinois mass spectrometry core for SCFA quantification via high-performance liquid chromatography mass spectrometry (HPLC-MS). Concentrations of each SCFA (acetate, propionate and butyrate) and combined total SCFAs are assessed as nmol SCFA per mg faeces.

SCFA-bacterial correlation

Multiple regression analysis was performed using RStudio between bacterial family read counts averaging >500 in both the SI and caecum with caecum SCFA concentrations of individual samples for acetate, propionate, butyrate and combined SCFAs taking into consideration injury and sex. Adjusted R-squared values for each family and SCFA are represented in a heatmap with accompanying p-values overlayed.

Small IEC culture and faecal slurry treatment

Murine duodenal cell clone-K (MODE-K) cells (generously provided by Dr. Jin Mo Park, HMS, Mass General, Charlestown, Massachusetts, USA) were cultured at 37°C with 5% CO₂ in high-glucose Dulbecco’s Modified Eagle Medium (DMEM) supplemented with 1% Antibiotic-Antimycotic Solution (Hyclone) and 5% fetal bovine serum (Life-Technologies). SI faecal samples isolated 1 day following alcohol and burn injury were homogenised in 1 mL PBS per 100 mg faeces and centrifuged at 10 000 rpm for 10 min to generate an unfiltered ‘faecal slurry’. MODE-K cells were seeded into 6-well plates for 48 hours. In a subset of experiments, cells were treated with filter-sterilised Sodium Butyrate (Sigma Aldrich) in PBS at a concentration of 2 mM or vehicle (PBS) 1 hour before faecal slurry. Cells were treated with faecal slurries diluted at 1:500 or PBS alone into culture media for 24 hours before media supernatant and cells were collected.

Faecal bacterial qPCR

Genomic DNA was isolated from small intestinal faecal samples isolated 1 day following alcohol and burn injury using QIAamp PowerFecal Pro DNA Kit (Qiagen) according to manufacturer’s instructions. DNA concentration was measured using the NanoDrop 2000 Spectrophotometer (ThermoScientific) and equal amounts were loaded for quantitative polymerase chain reaction (qPCR) reaction using 1X iTaq Universal SYBR Green Supermix (BioRad) with 200 nM of forward and reverse primers. Primers—total bacteria: F: (ACTCCTACGGGAGGCAGCAGT), R: (ATTACCGCGGCTGCTGGC); and Enterobacteriaceae: F: (GTGCCAGCMGCCGCGGTAA), R: (GCCTCAAGGGCACAACCTCCAAG). Level of Enterobacteriaceae is expressed as a ratio of Enterobacteriaceae/total bacteria.

RNA isolation and RT-qPCR

RNA was isolated from collected MODE-K cells using the RNeasy Mini Kit (Qiagen). The concentration and quality of isolated RNA were determined by NanoDrop 2000 spectrophotometer (ThermoScientific), and cDNA synthesis was performed using High-Capacity cDNA Reverse Transcription Kit (Life-Technologies). Expression of IL-6 mRNA was analysed by reverse transcription qPCR (RT-qPCR) using TaqMan Fast Advanced Master Mix (Life Technologies) and predesigned TaqMan primer probes (Life Technologies). Data were calculated via the ΔΔCt method with target gene Ct values normalised to beta actin housekeeping control.

ELISA

Conditioned media supernatant interleukin-6 (IL-6) was assessed using a DuoSet ELISA kit (BD Biosciences). The data were quantified as pgIL-6/mL conditioned media. Data were normalised to the average concentration of conditioned media from vehicle treated (no Tx).

Statistics

The data, wherever applicable, were presented as means±SEM and were analysed using one-way analysis of variance (ANOVA) with Tukey’s post-hoc test, Student’s t-test, Mann-Whitney U test or χ² test (GraphPad Prism, V.10). Multiple regression was performed on RStudio (V. 2024.04.2+764, Posit Software, Boston, Massachusetts, USA). A p value of <0.05 was considered statistically significant. Unless noted otherwise, significance is represented as follows: *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001.

Ethanol and burn injury significantly alter faecal microbiome diversity and composition

Pairwise comparison of the microbiome from EB versus SV samples revealed an association with significantly reduced richness in the SI (Chao1 p<0.05) (figure 1A), but not the caecum following combined injury (figure 1B), while significantly reduced evenness was seen in both the SI (Pielou, p<0.05) and caecum (Pielou, p<0.01). Shannon’s diversity demonstrated reduced bacterial diversity in the SI (Shannon, p<0.05) but not caecum (figure 1A, B). No significant differences were found from pairwise analysis of male versus female samples from either the SI or caecum (online supplemental figure 1A, B). Weighted UniFrac PCoA was performed to further assess differences in the microbial composition of EB samples compared with SV. EB samples exhibited separate clustering from SV, assessed by Pairwise PERMANOVA analysis (p<0.01), indicating distinct microbial compositions following ethanol and burn injury in both the SI and caecum (figure 1C, D).

Taxonomic analysis of the six most abundant bacterial phyla in the SI and caecum, respectively, revealed no significant changes in total read counts from male or female samples following ethanol and burn injury (online supplemental figure 1C, D). Following ethanol and burn injury, in the SI, there was a significant reduction in the relative abundances of Actinobacteria, Bacteroidetes and Verrucomicrobiota, while Proteobacteria was significantly elevated (figure 2A, B). In the caecum, the relative abundance of Firmicutes was significantly reduced in EB caecum samples compared with SV, while Bacteroidetes and Proteobacteria were significantly increased after injury (figure 2C, D). Male and female mice showed consistent trends in the relative abundances of bacterial phyla following ethanol and burn injury, with Proteobacteria remaining the most significantly altered phylum in both the SI and caecum (figure 2B, D).

We have previously shown increased intestinal inflammation and overgrowth of pathobionts following combined alcohol and burn injury.10 Herein, we sought to characterise intestinal dysbiosis following alcohol and burn injury through 16S sequencing. We used taxonomic analysis to assess changes in bacterial families and genera. Enterobacteriaceae is one such family from the Proteobacteria phyla and was the most abundantly increased in both the SI (figure 3A) and the caecum (figure 3B) following ethanol and burn injury. The Enterococcaceae family, also from the Proteobacteria phyla, was significantly increased following ethanol and burn injury in both the SI (figure 3A) and the caecum (figure 3B).

Following analysis of bacterial populations known to contain pathobionts, we investigated changes in probiotic bacteria. Phyla-level analysis revealed a decrease in Verrucomicrobiota in the SI but not caecum following ethanol and burn injury (figure 2). Verrucomicrobiota contains Akkermansia, an SCFA-producing bacterium known to confer protective properties to the intestinal epithelium.33 Further investigation revealed decreases in multiple SCFA-producing bacteria following ethanol and burn injury, including the Lachnospiraceae family, in particular Roseburia and Lachnospiriaceae NK4A13, in addition to Akkermansia, Muribaculaceae and Prevotellaceae (figure 3C). There were no significant differences between SV and EB in the caecum for any of these probiotic bacteria (online supplemental figure 2).

Faecal content from ethanol and burn injury promotes epithelial cell release of IL-6

Due to elevated levels of pathobionts and decreases in potentially beneficial bacteria in the SI, we hypothesised that the SI faecal content from EB mice would be more pro-inflammatory than that of SV mice. Treatment of the murine small IEC cell line, MODE-K, with bacterial products, including lipopolysaccharide, has been shown to stimulate pro-inflammatory cytokine production via toll-like receptor signalling.34 The pro-inflammatory cytokine IL-6 has been demonstrated to promote intestinal inflammation, tissue damage and gut barrier disruption following alcohol and burn injury.7 8 To assess the inflammatory capacity of the faecal microbiome following ethanol and burn injury, MODE-K cells were treated with faecal slurries isolated from SI faeces of EB or SV mice. MODE-K cell gene expression of IL-6 was significantly increased by treatment with EB faecal slurries compared with treatment with SV faecal slurries (figure 4B). EB faecal slurries also promoted more IL-6 protein expression compared with SV faecal slurries (figure 4C). Bacterial qPCR was used to confirm a significant increase in SI Enterobacteriaceae levels relative to total bacteria following ethanol and burn injury (figure 4A), consistent with the 16S sequencing results (figure 3A). To further assess the relationship between Enterobacteriaceae and small IEC inflammation, Spearman correlation analyses were performed between the relative quantity of Enterobacteriaceae in a sample and the level of IL-6 induced by that sample at the mRNA or protein level. Both IL-6 gene expression (figure 4D) and protein (figure 4E) exhibited significant positive correlations with Enterobacteriaceae levels.

Reduced caecal SCFA content correlates with reduced levels of SCFA-producing bacteria in the gut

In addition to direct interactions of bacteria with the gut barrier, metabolites produced by bacteria, particularly SCFAs, maintain the gut barrier and provide a major source of energy for the intestinal epithelial barrier.35 36 To study how changes in bacterial composition after alcohol and burn injury may impact levels of these bacterial metabolites, we quantified SCFAs (acetate, butyrate and propionate) via HPLC-MS in caecal contents of both male and female mice subjected to SV or EB. Overall, there were non-significant decreasing trends in the concentration of total SCFAs and acetate in the caecum following ethanol and burn injury (figure 4F, G), which was similar in both male and female mice (online supplemental figure 3A, B). There were no significant changes observed in propionate concentration (figure 4H) (online supplemental figure 3C). The only significantly differential SCFA after ethanol and burn injury was butyrate, which exhibited significantly reduced concentrations (figure 4I) with non-significant trending decreases in both male and female mice (online supplemental figure 3D). Due to the possibility of type-II error caused by smaller sample sizes when being stratified into male and female mice, more robustly powered studies are warranted in order to confirm if the butyrate findings observed in pooled samples remain significantly different between SV and EB for both male and female mice.

To begin to elucidate how changes in bacterial composition after ethanol and burn injury may contribute to the changes in SCFA concentration observed, multiple variable correlation analysis considering injury and sex was performed between bacterial family read counts and SCFA concentrations of individual samples. Heatmaps depicting the corresponding adjusted R-squared values were then generated to visualise the significance and degree of correlation. To investigate correlations within the same sampling location, SCFA concentrations were first compared with caecal bacterial family read counts. Surprisingly, caecal bacterial levels showed few strong correlations with SCFA concentrations (figure 5A). Interestingly, small intestinal bacterial families demonstrated more significant correlations with caecal SCFA concentrations, particularly of butyrate, which was the most significantly altered SCFA (figure 5B). In both the SI and caecum, few bacterial families correlated positively with propionate concentrations (figure 5A, B). The majority of small intestinal bacterial families that correlated with downstream caecum SCFA levels exhibited positive correlations, for many as bacterial levels decreased in the SI, there was a concurrent decrease in SCFA levels in the caecum, especially of butyrate.

Butyrate treatment can limit IEC release of IL-6 induced by faecal content from ethanol and burn

Butyrate is known to act in an anti-inflammatory manner in the intestines and is an imperative bacterial metabolite for maintenance of the host intestinal epithelial barrier.25 Herein, we sought to investigate if butyrate can limit inflammation induced by the dysbiotic microbiome following ethanol and burn injury. To assess the impact of butyrate on the inflammatory capacity of the faecal microbiome, MODE-K cells were treated with butyrate and then faecal slurry isolated from small intestinal faeces of EB or SV mice. As expected, treatment with butyrate reduced IL-6 produced by IECs (figure 6A). Treatment with butyrate prevented the induction of IL-6 on treatment with both SV and more significantly EB faeces (figure 6B, C).

Strengths and limitations

The purpose of this study was to assess the impact of combined alcohol and moderately sized burn injury on gut bacterial composition. Our previous studies evaluating 12.5% TBSA burn alone, or a single dose of alcohol alone did not significantly influence microbial communities as assessed by qPCR, compared with sham, but significant changes were observed in combined alcohol and burn injury.7 14 Previous animal and patient studies have shown perturbations of the microbiome following large burn injury or in a moderate burn combined with an additional insult such as advanced age or, in this case, alcohol.7 8 14 48 49 Therefore, this study only evaluated differences in microbiome composition between SV mice and those subjected to combined alcohol and burn injury, but not either alone, which we recognise as a significant limitation.

While the in vitro data presented provide preliminary evidence that butyrate may limit MODE-K IL-6 production in response to faecal slurries, more studies are necessary to functionally parse out the components of faecal samples, including but not limited to bacteria, metabolites, proteins, microRNAs, and their respective impact on intestinal inflammation and if butyrate is protective to gut epithelium following alcohol and burn injury.

This study reveals perturbations in the microbiome and SCFAs they produce following alcohol and burn injury. While the microbiome in both the caecum and SI were affected by alcohol and burn injury, the SI microbiome was dramatically altered, and these changes correlated with decreases in SCFAs, especially butyrate. Future in vivo studies are warranted to assess if restoring butyrate levels and/or microbial composition can mitigate intestinal barrier disruption and subsequent development of pathology in those suffering from burn injury with or without alcohol exposure.