Ethanol exposure is associated with spatial and taxon-specific gut microbiota remodeling characterized by ecological adaptation rather than broad activation of microbial ethanol metabolism

ABSTRACT

Alcohol-associated diseases are linked to gut microbiota disruption, but how ethanol intake is associated with microbial functional remodeling remains unclear. Here, using male C57BL/6J mice, we integrated spatiotemporal quantification of ethanol and acetaldehyde across the gastrointestinal tract with 16S rRNA sequencing, metagenomics, metaproteomics, and metabolomics. Our results showed that small-intestinal communities were enriched in taxa and functions related to bile acid tolerance/utilization, whereas cecal communities exhibited pronounced remodeling of multiple core functions associated with ecological adaptation. Although the abundance of several ethanol metabolism-related genes increased, this genetic potential was not broadly translated into detectable protein-level activation. Ethanol metabolism-related proteins were mainly derived from Lachnospiraceae, whose metabolic activity was suppressed. In contrast, Muribaculaceae, Desulfovibrionaceae, and Barnesiellaceae gained functional advantages in substrate acquisition, energy metabolism, oxidative stress defense, and proteostasis. These findings provide a region-resolved functional map indicating that ethanol-associated microbiota remodeling is characterized by ecological adaptation, rather than uniform activation of direct microbial ethanol metabolism.

Graphical abstract

1. Introduction

Global alcohol consumption remains high, and alcohol-attributable harm continues to represent a major public health concern worldwide. ¹ Excessive alcohol use is among the leading preventable causes of morbidity and premature mortality and has been causally associated with more than 200 disease and injury conditions. ² Growing evidence indicates that ethanol is an important exogenous disruptor of gut microbiota homeostasis. ³ Sustained or excessive ethanol exposure reshapes microbiota structure and promotes dysbiosis, which in turn contributes to alcohol-related tissue injury and disease progression. ^{4 , 5} Understanding how ethanol intake is linked to microbial functional remodeling is therefore important for clarifying alcohol-associated host-microbiota interactions. However, existing studies have predominantly focused on ethanol-induced changes in microbiota composition. ⁶ Mechanistic insight into how gut microbiota undergo metabolic reprogramming to adapt to ethanol exposure and reshape ecological competition remains limited.

A central unresolved issue is whether the gut microbiota directly contributes to host ethanol metabolism. Early in vitro studies have suggested that fecal microbiota can oxidize ethanol to acetaldehyde. ⁷ Studies using germ-free or antibiotic-treated animal models have shown that the gut microbiota may be a significant source of local acetaldehyde following ethanol exposure. ^8-10 However, omics studies have produced a more complex picture: metagenomic analysis of heavy drinkers has suggested that their gut microbiota exhibits enhanced genetic potential for ethanol metabolism, ¹¹ whereas metatranscriptomic evidence has indicated strengthened acetate- and acetaldehyde-associated pathways, implying that cecal microbiota may preferentially promote acetaldehyde detoxification and acetate dissimilation rather than direct ethanol oxidation. ¹² Moreover, one study suggests that the gut microbiota contributes only minimally to acetaldehyde clearance in the gut. ¹³

The features of the intestinal ecosystem complicate this question. Ethanol and acetaldehyde exposure are spatially heterogeneous because absorption and local metabolism differ along the gastrointestinal tract; furthermore, microbial distribution, abundance, and metabolic characteristics also differ. ^{14 , 15} The key limitations of in vitro culture systems include incomplete community reconstitution, ¹⁶ non-physiological nutrient conditions, and mismatched oxygen tension. ¹⁷ Metagenomic and metatranscriptomic profiles capture genetic potential and transcriptional activity, respectively, and cannot directly represent protein-level expression efficiency, stability, or functional output. ^{18 , 19} A region-resolved framework that connects ethanol and acetaldehyde kinetics with microbiota composition, genetic potential, and protein-level functional execution is therefore needed. ^{6 , 20}

We focus on two core questions: (i) whether ethanol exposure enriches ethanol metabolism-related microbial genetic potential, and whether this potential is translated into protein-level functional output; and (ii) if broad activation of microbial ethanol metabolism is not observed, which microbial functional modules are associated with ecological adaptation under ethanol exposure. This design was intended to construct a region-resolved functional map and generate testable hypotheses. Our results indicate that ethanol exposure was not associated with broad protein-level activation of microbial ethanol metabolism-related functions, but rather with region- and taxon-specific functional remodeling characterized by ecological adaptation.

2. Materials and methods

2.1. Animals and experimental design

Six-week-old SPF, wild-type, untreated male C57BL/6J mice (20 ± 2 g) were obtained from SPF (Beijing) Biotechnology Co., Ltd. Mice were group-housed at six animals per open-top solid-bottom cage under SPF barrier conditions maintained at 22 ± 2 °C and 50–60% relative humidity, with a 12-h light/12-h dark cycle. Standard laboratory chow and water were provided ad libitum, and cages contained autoclaved wood-shaving bedding. After a 1-week acclimation period, mice underwent the experimental procedures outlined in Figure 1. Animal health status and body weight were monitored daily throughout the study. Predefined exclusion criteria included severe dehydration, diarrhea, or sustained body-weight loss exceeding 20%. Animals reaching any of these endpoints were removed from the study and euthanized immediately under isoflurane anesthesia by cervical dislocation. Three animals assigned to metagenomic analysis were excluded because of their body weight. All animal procedures were approved by the Experimental Animal Ethics Committee of Jiangnan University (JN.No20241230c0960409[679]) and were reported in accordance with the ARRIVE 2.0 guidelines.

Figure 1.

Experimental design and overall workflow of the study. (A) To characterize the spatiotemporal dynamics of ethanol exposure, mice received a single gavage of 40% (v/v) food-grade ethanol at 8 mL/kg body weight. Blood and contents from the stomach, the front, middle, and back of the small intestine, the cecum, and the colon were collected at 5, 10, 20, 30, 45, 60, 90, 120, 150, 180, 240, 300, 360, and 480 min post-gavage (n = 6). To assess dose-dependent exposure, independent groups of mice received 40% (v/v) ethanol at 8, 4, 2, or 1 mL/kg body weight, and samples were collected 20 min after gavage (n = 6 per dose). Ethanol and acetaldehyde concentrations were quantified by gas chromatography-mass spectrometry (GC-MS). (B) Mice were assigned to a control group (Ctrl) or an ethanol-exposure group (Alc) (n = 8). Ctrl mice were given an equal volume of water by gavage, whereas Alc mice received 40% (v/v) food-grade ethanol at 8 mL/kg body weight once daily for 42 days. Six hours after the final water or ethanol gavage, contents were collected separately from the front, middle, and back of the small intestine, the cecum, and the colon for full-length 16S rRNA sequencing. (C) Mice were assigned to the Ctrl, acute ethanol (Alc-A), and long-term ethanol (Alc-L) groups. During the 42-day intervention period, Ctrl and Alc-A mice received an equal volume of water, whereas Alc-L mice received 40% (v/v) food-grade ethanol at 8 mL/kg body weight once daily. At the final gavage, Ctrl mice received water, whereas Alc-A and Alc-L mice received 40% (v/v) ethanol at 8 mL/kg body weight. Small-intestinal and cecal contents were collected 6 h after the final gavage. Metabolomics was performed using eight mice per group (n = 8), metaproteomics using three mice per group (n = 3), and metagenomics using six (cecum) to seven (small intestine) mice per group (n = 6/7). (D) To assess ethanol-associated butanol accumulation in vivo, mice received a single gavage of 40% (v/v) ethanol at 8 mL/kg body weight and were euthanized 20 min later; control mice received an equivalent volume of water. Small-intestinal and cecal contents were collected for butanol quantification by GC-MS (n = 6). For the in vitro experiment, small-intestinal and cecal contents collected from untreated mice were separately suspended in sterile physiological saline with or without ethanol and incubated anaerobically at 37 °C in an anaerobic workstation, followed by GC-MS measurement of butanol (n = 6). These experiments were designed to assess ethanol-associated butanol accumulation rather than to establish direct carbon flux from ethanol to butanol. (E) Mice received one of the three candidate strains once daily for 1 week (200 μL, 2 × 10⁹ cells); control mice received sterile physiological saline (n = 6). One hour after the final bacterial gavage, mice received 40% (v/v) food-grade ethanol at 8 mL/kg body weight. One hour later, blood and contents from the stomach, the front, middle, and back of the small intestine, the cecum, and the colon were collected. Ethanol and acetaldehyde concentrations were quantified by GC-MS. This short-term trial assessed the candidate strains’ basic ethanol-processing capacity and does not reflect the pattern or mechanism of ethanol processing by the endogenous gut microbiota.

For each experimental module, mice were weighed at baseline and allocated to the corresponding treatment groups by the investigators using body-weight matching, to achieve similar baseline body-weight distributions across groups. Sample sizes were selected based on previous studies, experimental feasibility, and the minimum number of independent biological replicates required for the planned analyzes. Each mouse was considered an independent experimental unit, and each reported n represents one biologically independent animal rather than a technical replicate. Each mouse contributed samples to only one analytical platform and was not reused across the metagenomic, metaproteomic, or metabolomic analyzes. A total of 193 mice were used across all experiments. To minimize potential bias, all animals and samples were handled according to predefined protocols, collected at prespecified, group-matched time points within each experiment, and analyzed using identical procedures. For long-term ethanol exposure models, gavage was performed at a fixed time each day to minimize circadian and handling-related variation.

At each designated sampling time point, mice were anesthetized with isoflurane, blood was collected from the orbital venous plexus, and animals were euthanized by cervical dislocation under anesthesia. The small intestine was divided into three equal-length segments (front, middle, and back), and contents were collected from each segment of the small intestine as well as from the cecum and colon. ²¹ Samples intended for ethanol/acetaldehyde quantification by gas chromatography-mass spectrometry (GC-MS) were analyzed immediately after collection. Samples intended for multi-omics analyzes were snap-frozen in liquid nitrogen and stored at −80 °C until processing.

2.2. Quantitative characterization of ethanol metabolism

Blood and gastrointestinal contents were placed in saturated sodium chloride solution before analysis. Ethanol and acetaldehyde were quantified by GC-MS (Trace 1300-ISQ, Thermo Scientific, Germany) operated in single-ion monitoring (SIM) mode. Specimens were equilibrated at 80 °C for 5 min with agitation, and 100 μL of headspace gas was injected. GC analysis was performed with helium as the carrier gas at a constant flow rate of 1.0 mL/min. Acetaldehyde was monitored at m/z 29, 43, and 44, and ethanol at m/z 31, 45, and 46. ²² Quantification was based on external calibration curves generated from ethanol and acetaldehyde standards (GCS; Sinopharm Chemical Reagent Co., Ltd., Shanghai, China). Concentrations are reported as mol/mL for blood and mol/g for gastrointestinal contents.

2.3. In vitro culture of gut contents

To evaluate the ethanol-processing capacity of the gut microbiota under in vitro conditions, cecal contents were collected and suspended in sterile physiological saline containing 36 mM ethanol (approximately the peak ethanol concentration measured in the cecum). Suspensions were incubated at 37 °C under anaerobic conditions in an anaerobic workstation with a three-gas mixture, without additional nutrient supplementation, ⁷ with or without nicotinic acid (NA, 3 mM) or hydrogen peroxide (H₂O₂, 0.1 mM). ²³ After 24 h of incubation, ethanol levels were quantified by GC-MS (n = 6).

To assess ethanol-associated butanol accumulation, mice received acute ethanol gavage (40% [v/v] ethanol, 8 mL/kg body weight) and were euthanized 20 min later; control mice received water. Contents from the small intestine and the cecum were collected for butanol quantification by GC-MS. In parallel, contents from the small intestine and cecum were collected from untreated mice, suspended in sterile physiological saline containing ethanol or ethanol-free saline, and incubated at 37 °C under anaerobic conditions in an anaerobic workstation before butanol measurement (n = 6). All GC-MS conditions were the same as those described above; quantification was based on external calibration curves generated from butanol standards (GCS; Sinopharm Chemical Reagent Co., Ltd., Shanghai, China).

2.4. Measurement of NADH levels

NADH levels were measured using the NADH Assay Kit with WST-8 (Beyotime Biotechnology, Shanghai, China). Briefly, after collecting gut contents as described above, NADH levels were measured following the kit protocol.

2.5. Full-length 16S rRNA sequencing

Genomic DNA was extracted from gut contents using a magnetic bead-based stool DNA extraction kit (Tiangen Biotech, Beijing, China). Full-length 16S rRNA genes were amplified with primers 27F (AGRGTTTGATYNTGGCTCAG) and 1492 R (TASGGHTACCTTGTTASGACTT), and the amplicons were purified, quantified, and pooled at equimolar concentrations. SMRTbell libraries were constructed and sequenced on the PacBio Sequel II platform. Raw reads were processed in SMRT Link v8.0 to obtain circular consensus sequencing (CCS) reads. High-quality sequences were clustered into operational taxonomic units (OTUs) at 97% similarity using USEARCH v10.0, and taxonomic annotation was performed in QIIME2 using a Naive Bayes classifier trained on the SILVA database (release 138.1).

2.6. Metagenomic sequencing

Metagenomic DNA was extracted from gut contents (Magen, Guangzhou, China) and used for library construction (~350 bp insert size), followed by paired-end 150-bp sequencing on the Illumina NovaSeq X Plus platform. Raw reads were quality-filtered using fastp (v0.20.0), and host contamination was removed by alignment to the Mus musculus reference genome (NCBI_GCF_000001635.27_GRCm39). Raw sequencing reads were quality-filtered according to the following criteria: (1) reads containing ≥ 10% undetermined nucleotides (N) were removed; (2) reads in which ≥ 50% of the bases had Phred quality scores ≤ 20 were discarded; (3) reads containing adapter sequences were removed; and (4) reads shorter than 50 bp were excluded. Clean reads were assembled de novo using MEGAHIT (v1.2.9), and contigs ≥ 500 bp were retained for gene prediction with MetaGeneMark (v3.38). Predicted genes were clustered using CD-HIT (v4.6; 95% identity and 90% coverage) to construct a non-redundant gene catalog. Reads were then mapped to the catalog with Bowtie2 (v2.3.5.1), reassigned with Pathoscope (v2.0.7), and normalized to TPM, excluding genes supported by fewer than 2 reads.

The non-redundant gene set was compared with the KEGG (Release 111) database using DIAMOND (version 0.9.25; e-value < 10⁻⁵). Differentially abundant genes were screened, and KEGG functional enrichment was analyzed. Low-abundance genes were filtered using counts per million reads (CPM), and genes with CPM ≥ 1 in at least three samples were retained. CPM values were used only for filtering, whereas the corresponding unnormalized integer counts were analyzed using the DESeq2 framework implemented in PyDESeq2 (v0.5.4). P values were adjusted using the Benjamini–Hochberg procedure, and genes with adjusted P value < 0.05 and |log₂ fold change| ≥ 1 were considered differentially abundant genes. Increased and decreased genes were subjected separately to one-sided hypergeometric KEGG pathway enrichment analysis using all CPM-filtered KEGG-annotated genes as the background set. KEGG pathways with a BH-adjusted P value < 0.05 were considered significantly enriched.

2.7. Metaproteomic analysis strategy

Small-intestinal and cecal contents (1–2 g) were lysed in buffer containing 8 M urea and 1% SDS, followed by centrifugation at 16,000 × g for 20 min at 4 °C. Protein concentrations were determined using the BCA method (Beyotime Biotechnology, China). Protein denaturation, reduction, alkylation, trypsin digestion, and peptide cleanup were performed using the iST sample preparation kit (PreOmics, Germany). Peptides were reconstituted in 0.1% formic acid, and 200 ng of each sample was analyzed using an UltiMate 3000 (Thermo Fisher Scientific, USA) liquid chromatography system coupled to a timsTOF Pro 2 mass spectrometer (Bruker Daltonics, Germany). Peptides were separated on an AUR3-15075C18 column (15 cm × 75 μm i.d., 1.7 μm particle size, 120 Å pore size; IonOpticks) at 300 nL/min and 50 °C using a 30-min gradient. Mobile phase A was 0.1% formic acid in water, and mobile phase B was 80% acetonitrile containing 0.1% formic acid. The gradient started at 2.2% B, increased to 28% over 16 min, then to 38% over 8 min, then to 90% over 3 min, and was maintained at 90% for 3 min. DIA data were acquired in diaPASEF mode over an m/z range of 349–1229 using 22 × 40 Th precursor isolation windows. The raw data were compared with a protein database derived from concurrently generated metagenomic data in this study using Spectronaut 19. Carbamidomethylation of cysteine was set as a fixed modification, whereas methionine oxidation and protein N-terminal acetylation were set as variable modifications. FDR was controlled at 1% at the precursor, peptide, and protein-group levels. Protein group quantification was conducted using local normalization and MaxLFQ. Before differential protein analysis, contaminant proteins were removed, and missing values were handled group-wise. Proteins with missing values in more than 50% of the biological replicates within a group were considered not reliably quantified in that group, and the corresponding processed abundance values were coded as zero. For proteins with less than 50% missingness within a group, missing values were imputed using the minimum non-zero abundance observed in the corresponding group. Proteins that were not reliably quantified in any of the groups being compared were excluded.

2.8. Untargeted metabolomics analysis

Gut contents (50 mg) were extracted with 1,000 μL of precooled methanol/acetonitrile/water (2:2:1, v/v/v) containing internal standards. After vortexing, bead-mill homogenization, and ultrasonication, the samples were centrifuged, and the supernatants were collected, vacuum-dried, and reconstituted in acetonitrile/water (1:1, v/v). After a second centrifugation, the resulting supernatants were analyzed using a Waters ACQUITY I-Class PLUS UHPLC system coupled to a Waters Xevo G2-XS QTof mass spectrometer (Waters Corporation, USA). Metabolites were separated on an ACQUITY UPLC HSS T3 column (1.8 um, 2.1 × 100 mm) using water and acetonitrile as the mobile phases, with an injection volume of 2 uL. Mobile phase A consisted of water containing 0.1% formic acid, and mobile phase B consisted of acetonitrile containing 0.1% formic acid. MS data were acquired using MassLynx V4.2 in MSe mode, an elevated-energy data-independent acquisition mode that simultaneously collects precursor and fragment ion information through alternating low- and high-collision-energy scans. The scan range was m/z 50–1,200. The low-energy function was acquired without applied collision energy, whereas the elevated-energy function used a collision-energy ramp of 10–40 V. The scan time was 0.2 s per spectrum. The electrospray ionization source parameters were as follows: capillary voltage, 2,500 V in positive-ion mode and −2,000 V in negative-ion mode; cone voltage, 30 V; source temperature, 100 °C; desolvation temperature, 500 °C; and desolvation gas flow, 800 L/h. Raw data acquired with MassLynx V4.2 were processed with Progenesis QI for peak extraction, alignment, normalization, and downstream analysis. Metabolite annotation was performed by matching features against the online METLIN database in Progenesis QI and a proprietary spectral library provided by Biomarker Technologies. Database matching was based on precursor ion mass error (ppm), isotope distribution, and MS/MS fragment matching score. Principal component analysis and Spearman correlation analysis were used to assess within-group sample repeatability and the stability of quality control samples. Pooled quality control samples were prepared by mixing equal aliquots of all sample extracts and were analyzed throughout the analytical workflow to monitor instrumental stability and analytical reproducibility. Blank samples were also included to monitor background signals. In the quality control samples, 88% of the detected metabolic features showed relative standard deviations (RSDs) ≤ 30%, indicating acceptable analytical reproducibility for the untargeted metabolomics analysis.

2.9. Short-term functional assessment of candidate strains

Functional validation was performed using commercially obtained strains corresponding to candidate taxa: L. murinus (CICC 23141, isolated from the intestines of mice, China Center of Industrial Culture Collection); R. gnavus (ATCC 29149, isolated from human feces, Ningbo Testobio Co., Ltd.); and D. piger (ATCC 29098, isolated from human feces, Ningbo Testobio Co., Ltd.). Strains were cultured at 37 °C in their recommended media. D. piger and R. gnavus were cultivated anaerobically in an anaerobic workstation with a three-gas mixture, whereas L. murinus was cultured aerobically. Cultures were harvested at approximately mid-log phase and resuspended in sterile physiological saline to 1 × 10 ¹⁰ cells/mL.

Mice were gavaged once daily for seven consecutive days with 200 μL of the bacterial suspension (2 × 10 ⁹ cells per dose). The control group received an equal volume of sterile physiological saline. One hour after the final bacterial gavage, mice received a single ethanol gavage (40% [v/v] ethanol, 8 mL/kg body weight). One hour later, mice were anesthetized with isoflurane, euthanized, and blood and gastrointestinal contents were collected for GC-MS analysis (n = 6). All gavage and sampling procedures were performed at matched time points across groups.

2.10. Microbial co-occurrence network analysis

Microbial co-occurrence networks were constructed from genus-level relative abundance profiles obtained by 16S rRNA sequencing. The same set of 100 genera with the highest relative abundances was initially considered for each group. Genera with zero abundance across all eight biological replicates were excluded on a group-specific basis. Pairwise two-sided Spearman's rank correlations were calculated using Python v3.12.13. Corresponding P values were obtained using the t-distribution approximation with n-2 degrees of freedom. Associations with |ρ| ≥ 0.60 and unadjusted P < 0.05 were retained as edges in the exploratory networks used for visualization and primary topological analysis. Networks were treated as undirected weighted graphs. The absolute correlation coefficient |ρ| was used as the non-negative edge weight. Network properties, including the numbers of nodes and edges, proportions of positive and negative edges, weighted degree, and modularity, were calculated in Python. Weighted degree was defined as the sum of |ρ| for all edges incident on a node. Community structure was identified using the weighted Louvain algorithm, with the resolution parameter set to 1.0. For each network, 1,000 randomized Louvain runs were performed, and the partition with the highest modularity value was retained. Visualization was performed using a customized force-directed layout based on the Fruchterman–Reingold principle.

2.11. Statistical analyzes

Differentially abundant taxa were identified using LEfSe, with an initial screening using the Kruskal-Wallis test followed by pairwise Wilcoxon rank-sum tests; effect sizes were estimated using linear discriminant analysis (LDA). The LDA score threshold was set to 3.0, and statistical significance was defined as nominal P < 0.05. Two-sided unpaired Student's t-tests were performed using the processed, untransformed protein-abundance values for each ethanol-exposed group versus the control group. Nominal P values were used for exploratory differential-protein screening and functional mapping. Proteins quantitatively detected in both groups were considered exploratory differentially expressed proteins when they met both |log₂ fold change| ≥ 1 and nominal P < 0.05 and were classified as increased or decreased according to the direction of the fold change. Group-specific proteins were defined as proteins that were undetected or could not be reliably quantified in more than 50% of the biological replicates in one group but were detected in at least two of the three biological replicates in the other group. Group-specific proteins with nominal P < 0.05 were retained as exploratory differentially expressed proteins. For the FDR-based sensitivity analysis, proteins quantitatively detected in both groups were included in the FDR-adjusted subset when they met both |log₂ fold change| ≥ 1 and FDR-adjusted P < 0.05. Group-specific proteins were included in the corresponding FDR-adjusted subset when their FDR-adjusted P values were < 0.05. These FDR-adjusted subsets were used to evaluate whether the principal functional categories and their major taxonomic contributors identified in the exploratory analysis were retained after multiple-testing correction. For the cecal metagenomic analysis, one outlier was excluded from each group. Unless otherwise specified, comparisons between two groups were performed using the Mann-Whitney U test. Nominal P < 0.05 was considered statistically significant for exploratory analysis. Statistical analyzes and data visualization were performed using GraphPad Prism version 10.0, Origin version 2025, and R version 4.5.1. Data were visualized as mean ± SEM.

3. Results

3.1. Gastrointestinal ethanol dynamics following acute ingestion

Following acute ethanol exposure, the stomach emerged as the predominant site of ethanol accumulation (Figure S1A). Ethanol subsequently entered the proximal small intestine in a pulsatile manner consistent with intermittent gastric emptying. During the early phase (5–20 min), rapid proximal small-intestinal absorption coincided with declining ethanol levels in the stomach and the front/middle small intestine. In contrast, ethanol concentrations increased in blood and distal intestinal segments, with blood ethanol reaching an initial peak at 20 min. During the intermediate phase, renewed ethanol delivery from the stomach led to secondary increases in the proximal small intestine at 20–30 min and again at 90–120 min, accompanied by rebounds in blood ethanol at 45 min and 150 min, respectively. Thereafter, clearance dominated (45–90 min and 180 min); ethanol became undetectable in the colon by 150–180 min, in the middle and back segments of the small intestine and cecum by 180–240 min, and only trace levels persisted in the stomach and proximal small intestine at 240 min. These patterns indicate that blood ethanol does not decrease in a strictly monotonic fashion after the initial peak. Ethanol levels in the distal small intestine, cecum, and colon were highly synchronized with, or only slightly lagged, blood ethanol concentration (Figure 2A-B). Importantly, ethanol remained detectable in the distal intestine even at a lower gavage dose (Figure S1B).

Figure 2.

The dynamic characteristics of ethanol in the gastrointestinal tract and the region-resolved 16S rRNA sequencing of the gut microbiota. (A) Time-course profiles of ethanol and acetaldehyde in blood and along the gastrointestinal tract following acute ethanol gavage (n = 6). (B) Directional trends in ethanol concentrations across different types of samples over time; blue and orange arrows indicate decreases and increases, respectively. (C) LEfSe analysis shows discriminative taxa of microbiota across intestinal sites before and after ethanol exposure; the phylogenetic cladogram displays only biomarkers with LDA > 3 (n = 8). (D) Relative abundances of bile acid metabolism pathways and associated genes in the small-intestinal microbiota (n = 7). *Nominal P < 0.05 versus the control group. (E) Region-resolved genus-level microbial co-occurrence networks in control and ethanol-exposed mice. Positive and negative correlations are shown in red and blue, respectively, and edge color and width represent correlation direction and magnitude. Node color denotes bacterial phylum, and node size represents group-mean relative abundance using a common scale across all networks. The eight genera with the highest weighted degree in each network are outlined and labeled. Edges represent co-variation patterns rather than direct ecological interactions (n = 8).

Blood acetaldehyde rose rapidly within 5–10 min, declined during 10–45 min, increased again from 45–120 min to a peak at 120 min, and then decreased monotonically to undetectable levels by 360 min. In the small intestine, acetaldehyde peaked at 90 min in the proximal segment (with secondary peaks at 30 and 180 min), at 150 min in the middle small intestine, followed by a rapid decline, and at 45 min in the distal segment with additional peaks at 120 and 180 min. The cecum showed a bimodal pattern with peaks at 20 and 45 min, whereas the colon peaked at 20, 60, and 150 min (Figure 2A).

Together, these results indicate that exposure to ethanol and acetaldehyde was spatially and temporally heterogeneous across the gastrointestinal tract. Thus, the compositional and functional responses of the microbiota should not be viewed as a uniform community-wide reaction to ethanol intake.

3.2. Ethanol-associated region-specific remodeling of the gut microbiota

Because ethanol exposure varies significantly along the gastrointestinal tract, we performed 16S rRNA sequencing of luminal contents collected from the entire intestine. The LEfSe analysis was used to identify region-specific biomarkers before and after ethanol exposure.

The microbiota in different regions of the small intestine exhibited broadly similar changes. Specifically, Lactobacillales were broadly reduced (mainly taxa related to Ligilactobacillus murinus and Streptococcus danieliae), while microbiota associated with bile acid tolerance and/or utilization, such as Erysipelotrichales (Erysipelotrichaceae, Dubosiella), ^{24 , 25} Coriobacteriia (Coriobacteriales, Eggerthellaceae, Enterorhabdus), ^{26 , 27} and Bifidobacterium (B. pseudolongum) ^{28 , 29} were enriched. However, Lactobacillus johnsonii and Lactobacillus taiwanensis were enriched in the middle segment of the small intestine. These two microorganisms are potential bile acid-metabolizing/tolerant lactobacilli ^{30 , 31} (Figure 2C). Consistent with these changes, several bile acid metabolism-related genes were significantly increased in the small-intestinal microbiota, as reported previously ³² (Figure 2D).

In contrast to the small intestine, the cecal and colonic microbiota displayed distinct remodeling patterns, with a shared significant reduction in the butyrate-associated genus Lachnoclostridium. Studies have suggested that a reduction in butyrate-producing genera is a key component of ethanol-induced dysbiosis. ³³ In addition, in the cecum, multiple Campylobacterota lineages, including members assigned to Helicobacter, were enriched, accompanied by a significant increase in Clostridium sensu stricto 1 (C. disporicum), [Eubacterium] ruminantium group, Lactobacillus, and Oscillospirales (UCG 010). In the colon, Staphylococcales, Aerococcaceae (Aerococcus, A. viridans), and Muribaculum (M. intestinale) were enriched. Meanwhile, Turicibacter (T. sanguinis), which possesses bile acid metabolic capabilities ^{34 , 35} (Figure 2C), was enriched in the distal small intestine, cecum, and colon. However, the relative abundances of bile acid metabolism-related genes did not increase significantly in the cecum (Figure S2).

Microbial co-occurrence networks exhibited pronounced intestinal segment-specific responses to ethanol exposure. The proximal small-intestinal network showed the largest reduction in associations, while its modularity increased. The overall connectivity of the cecal network changed little, but the relative contribution of negative associations increased markedly, indicating that the primary change involved the direction of associations. In contrast, the colonic network showed increased connectivity and a higher proportion of positive associations. Although genus composition within the microbial co-occurrence networks of each intestinal segment showed substantial overlap before and after ethanol exposure, relatively few network edges were shared, and overlap among the key genera with the highest weighted degrees was also limited. These findings indicate that, despite the overall stability of genus composition, ethanol exposure was primarily associated with extensive reorganization of intergenus association structures and redistribution of network centrality. Notably, these results reflect statistical co-variation patterns only (Figure 2E, Table S1).

3.3. Integrated multi-omics revealed the ethanol-metabolizing potential of the gut microbiota

To evaluate changes in the genetic potential for ethanol metabolism in the gut microbiota, we first characterized the relative abundance of genes encoding enzymes involved in ethanol, acetaldehyde, acetate, or acetyl-CoA metabolism. Long-term ethanol exposure enriched the abundance of genes annotated to EC:1.2.1.3 and EC:2.8.3.1, encoded by the small-intestinal microbiota; these genes are involved in acetaldehyde and acetate metabolism, respectively, with acetyl-CoA as the end product. In the cecal microbiota, long-term ethanol exposure enriched the abundance of genes annotated to EC:1.1.1.1, EC:1.2.7.5, and EC:6.2.1.1, which are involved in the direct oxidation of ethanol and acetaldehyde to acetate and in the conversion of acetate to acetyl-CoA (Figure 3A). These results indicate that the genetic potential associated with the ethanol metabolism pathway has been altered.

Figure 3.

Integrated multi-omics revealed the ethanol-metabolizing potential of the gut microbiota. (A) Relative abundances of ethanol metabolism-related genes encoded by the microbiota in the small intestine and the cecum (n = 6/7). (B) The summed relative abundances of species-resolved microorganisms that code for genes involved in ethanol metabolic pathways (n = 6/7). (C) Relative abundances of microbial proteins in the cecum annotated as EC:1.1.1.1 and bi-functional alcohol/aldehyde dehydrogenase (AdhE) (n = 3). (D) Relative abundances of microbial proteins in the cecum annotated as EC:2.7.2.1 and EC:2.3.1.8 (n = 3). (E) The summed relative abundances of species-resolved microorganisms expressing EC:1.1.1.1, AdhE, EC:2.7.2.1, and EC:2.3.1.8 in the cecum (n = 6). (F) Heatmap shows the relative abundances of microbial proteins annotated as EC:1.1.1.1 and EC:1.1.1.2 in the cecum (n = 3). (G) The heatmap shows the relative abundances of AdhE and other microbial aldehyde dehydrogenase enzymes in the cecum (n = 3). (H) Heatmap shows the relative abundances of microbial proteins annotated as EC:2.7.2.1 in the cecum (n = 3). (I) Heatmap shows the relative abundances of microbial proteins annotated as EC:2.3.1.8 and EC:2.8.3.1 in the cecum (n = 3). (J) Relative abundances of species-resolved microorganisms expressing proteins EC:1.1.1.1 or AdhE (n = 6). (K) Changes in ethanol concentration during in vitro cultivation of cecal contents with or without supplementation of niacin (NA) or hydrogen peroxide (H₂O₂) (n = 6). (L) Butanol concentrations in cecal contents following acute ethanol gavage or in vitro cultivation with ethanol (n = 6). (M) The hypothesized microbial pathway for ethanol-associated butanol production in the cecum, together with a heatmap showing the relative abundances of key enzymes (n = 3). (N-O) Ethanol (N) or acetaldehyde (O) concentrations in blood and along the gastrointestinal tract with or without gavage of specific strains (n = 6). # Differential proteins identified by screening. *Nominal P < 0.05 versus the control group.

We next analyzed the relative abundances of microorganisms that harbor these genes and are identifiable at the species level. In the small intestine, microorganisms encoding EC:1.2.1.3, EC:2.7.2.1, EC:2.3.1.8, and EC:2.8.3.1 were significantly increased under long-term ethanol exposure; in contrast, microorganisms encoding EC:1.2.1.10 and bifunctional alcohol/aldehyde dehydrogenase (AdhE) were significantly decreased. In the cecum, the relative abundance of microorganisms encoding ethanol metabolism–related genes did not change significantly (Figure 3B).

Metaproteomics was then used to characterize the detectable protein-level response of the microbiota. In the cecum, the main detectable microbial proteins annotated to ethanol metabolism pathways were EC:1.1.1.1, AdhE, EC:2.7.2.1, and EC:2.3.1.8. AdhE, EC:2.7.2.1, and EC:2.3.1.8 all decreased significantly following long-term ethanol exposure (Figure 3C-D); however, these proteins were not widely expressed in the small intestine. In addition, the relative abundance of microorganisms expressing EC:1.1.1.1 in the cecum was significantly reduced following long-term ethanol exposure (Figure 3E). Taxonomic annotation of the protein producers indicated that these proteins were primarily derived from Lachnospiraceae (Figure S3). Although no significant increase was observed in the total relative abundance of any individual protein category, a subset of individual proteins was up-regulated (Figure 3F–I). Notably, the relative abundances of microorganisms expressing proteins most directly related to ethanol metabolism (EC:1.1.1.1 or AdhE) did not show a widespread and significant increase (Figure 3J).

To functionally probe ethanol processing by the microbiota, we cultured cecal contents in vitro. Cecal contents showed a significant reduction in ethanol concentrations, and supplementation with nicotinic acid (NA, an NAD precursor) or hydrogen peroxide (H₂O_2, a potential electron acceptor promoting NADH reoxidation) led to a more pronounced decrease. However, this enhancement did not reach statistical significance (Figure 3K). Therefore, these results do not establish NAD⁺ regeneration as a major constraint on microbial ethanol processing under the tested conditions. ³⁶

Notably, butanol was significantly increased in the cecum following acute ethanol gavage, and butanol levels also increased significantly in cecal contents incubated with ethanol in vitro (Figure 3L); this phenomenon was not observed in small-intestinal contents. Because direct ethanol-to-butanol carbon flux was not assessed, these findings do not establish that the accumulated butanol was synthesized directly from ethanol. However, the increase in acetyl-CoA availability following acute ethanol exposure, ¹² together with the detection of proteins assigned to EC 1.1.1.157 and EC 1.3.8.1, prompted us to propose the testable hypothesis that the cecal microbiota may harbor metabolic potential for butanol production (Figure 3M). ³⁷ This hypothesis-generating model suggests that the cecal microbiota may exhibit metabolic flexibility. Proteins assigned to this proposed route were also predominantly attributed to Lachnospiraceae (Figure S4).

To assess the basal ethanol-processing capacity of selected candidate strains in vivo, we selected three commercially available bacteria annotated as expressing ethanol metabolism-related proteins: Desulfovibrio piger (EC:1.1.1.1), Ruminococcus gnavus (AdhE), and Ligilactobacillus murinus (AdhE and EC:2.7.2.1). All three strains significantly reduced gastric ethanol and acetaldehyde concentrations after acute ethanol gavage (Figure 3N-O).

Collectively, these findings indicate that the gut microbiota possesses basal and condition-dependent ethanol-processing capacity under specific experimental conditions. However, although ethanol exposure increased the abundance of certain genes associated with ethanol metabolism, this genetic potential was not translated into a significant increase in the corresponding proteins within the endogenous gut microbiota.

3.4. Integrated multi-omics revealed key features of ethanol-associated gut microbiota remodeling

The apparent dissociation between ethanol metabolism-related genetic potential and protein-level functional output prompted us to examine further which functional programs were most prominently associated with microbiota remodeling under ethanol exposure.

We first compared changes in KEGG Orthology (KO) abundance in the metagenome. We specifically labeled the KOs involved in ethanol metabolism; notably, these KOs were not among the features with the largest effect sizes or the strongest statistical evidence (Figure 4A–D). Trend analysis revealed that KOs co-clustered with those significantly increased and associated with ethanol metabolism in the small intestine (K00128 and K01026, corresponding to EC:1.2.1.3 and EC:2.8.3.1, respectively) were primarily involved in the two-component system, quorum sensing, ABC transport, central carbon metabolism, cofactor biosynthesis, and amino acid/nucleotide biosynthesis (Figure 4E). Similarly, KOs co-clustered with those significantly increased and associated with ethanol metabolism in the cecum (K13953, K03738, and K01895, corresponding to EC:1.1.1.1, EC:1.2.7.5, and EC:6.2.1.1, respectively) were primarily involved in pyruvate metabolism, carbon metabolism, cofactor biosynthesis, ABC transport, and ubiquinone biosynthesis (Figure 4F).

Figure 4.

Integrated multi-omics revealed key features of gut microbiota remodeling during ethanol exposure. Small-intestinal microbiota (A-B) or cecal microbiota (C-D): genes annotated to the same KEGG Orthology (KO) identifier were summed before DESeq2 analysis. Volcano plots show KO-level changes in gene abundance under acute (A, C) or long-term (B, D) ethanol intervention; only KOs involved in ethanol metabolism are labeled (n = 6/7). (E-F) Temporal trend analysis of microbiota-encoded genes in the small intestine (E) or the cecum (F) is shown using Mfuzz clustering, with ethanol metabolism-related KOs highlighted (n = 6/7). (G) Metabolomics of cofactors in the small intestine and the cecum, reflecting shifts in microbiota metabolic states (n = 8). (H-I) OPLS-DA of ethanol-associated changes in microbial protein expression profiles in the small intestine (H) or the cecum (I) (n = 3). (J) Relative abundances of Lachnospiraceae, Muribaculaceae, Barnesiellaceae, and Desulfovibrionaceae in the cecum (n = 6). (K) Relative abundance of F₀F₁-ATP synthase expressed by Lachnospiraceae, Muribaculaceae, Barnesiellaceae, and Desulfovibrionaceae in the cecum (n = 3). (L-M) Differentially expressed proteins in the cecum under acute (L) or long-term (M) ethanol intervention were subjected to KEGG pathway enrichment analysis. Enrichment was tested using a hypergeometric framework and evaluated by Fisher’s exact test; pathways with p < 0.05 were considered significant. Enriched pathways were visualized as a network after excluding ko01120, ko01110, and ko01100. Node size indicates the number of pathway-pathway connections, and node color denotes the number of differentially expressed proteins mapped to each pathway (n = 3). *Nominal P < 0.05 versus the control group.

KEGG enrichment analysis of differentially abundant genes revealed specific remodeling of microbial genetic potential following ethanol exposure. In the small intestine, acute exposure depleted genes involved in DNA replication/repair, transcription, galactose metabolism, glycan degradation, sphingolipid metabolism, flavonoid degradation, and cyanoamino acid metabolism, indicating a broad reduction in growth, genome maintenance, and substrate utilization potential. Long-term exposure, in contrast, favored carbohydrate and complex glycan utilization and functions related to cell-surface and membrane remodeling. In the cecum, acute exposure enhanced carbohydrate-utilization potential but reduced teichoic acid biosynthesis, two-component signaling, phosphonate and phosphinate metabolism, chemotaxis, and flagellar assembly. Long-term exposure further increased carbohydrate-uptake potential, whereas chemotaxis and two-component signaling remained reduced. Collectively, ethanol may induce broad acute functional restriction followed by selective long-term adaptation in the small intestine, while consistently promoting carbohydrate acquisition and suppressing environmental sensing and motility-related functions in the cecum (Figure S5).

Changes in cofactor-related metabolites further suggested remodeling of cofactor-associated metabolic networks. ³⁸ In the small intestine, the most prominent changes involved the riboflavin module, vitamin B6 metabolism, and quinone electron carriers. In the cecum, alterations were more pronounced in CoA biosynthesis, NAD salvage, quinone electron carriers, and riboflavin metabolism. From a pathway perspective, the observed metabolite changes were related to electron transport, flavin-dependent redox processes, PLP-dependent amino acid metabolism, and CoA-dependent lipid metabolism (Figure 4G).

OPLS-DA indicated a clear separation of microbial protein expression profiles between the control and ethanol-exposed groups (Figure 4H-I). Under exploratory screening criteria, the numbers of differentially expressed proteins (DEPs) identified in the small-intestinal microbiota under acute and long-term exposure were 48 and 131, respectively, whereas the corresponding numbers in the cecal microbiota were 587 and 1071 (Figure S6). In the cecal microbiota, Lachnospiraceae contributed the majority of down-regulated proteins (Supplementary Materials 1). Consistently, the relative abundances of Lachnospiraceae and F₀F₁ ATP synthase expressed by Lachnospiraceae were both significantly reduced (Figure 4J-K). By comparison, although there were no significant changes in the relative abundances of Muribaculaceae, Desulfovibrionaceae, and Barnesiellaceae, proteins of these families accounted for the vast majority of the up-regulated proteins (Figure 4J, Supplementary Materials 1). The relative abundances of F₀F₁ ATP synthase in both Barnesiellaceae and Desulfovibrionaceae increased significantly under long‑term ethanol exposure (Figure 4K). For Muribaculaceae, although the relative abundance of NADH: ubiquinone reductase (Na⁺-transporting) decreased significantly under long-term ethanol exposure (Figure S7A), suggesting reduced NADH reoxidation capacity and membrane electrochemical driving force, ^{39 , 40} the relative abundance of F₀F₁ ATP synthase did not change significantly (Figure 4K).

In the small-intestinal microbiota, the relative abundance of Lachnospiraceae decreased significantly under acute ethanol exposure, whereas Muribaculaceae, Desulfovibrionaceae, and Barnesiellaceae showed no significant changes (Figure S7B). Fewer DEPs were identified in the small intestine, and the patterns of taxonomic contribution differed: Muribaculaceae contributed both major up- and down-regulated proteins under acute exposure and contributed most down-regulated proteins under long-term exposure. Desulfovibrionaceae dominated the up-regulated proteins under long-term ethanol exposure (Supplementary Materials 2).

At the pathway level, in the cecal microbiota, DEPs were most frequently assigned to carbon metabolism, ribosome, and glycolysis/gluconeogenesis (Figure S7C-D). In the small-intestinal microbiota, ribosomal proteins contained the highest number of DEPs (Figure S7E-F). KEGG pathway interaction networks constructed from DEPs revealed that energy metabolism forms the core of metabolic reprogramming in the cecal microbiota (Figure 4L-M). In contrast, changes in ribosomal proteins were most significant in the small-intestinal microbiota (Figure S7G-H).

From a functional perspective, DEPs are primarily concentrated in four candidate core modules: (1) oxidative stress defense; (2) nutrient acquisition, particularly the utilization of complex carbohydrates; (3) energy metabolism and electron transfer; and (4) protein synthesis and homeostasis. Notably, this metabolic reprogramming of the gut microbiota is highly family-specific (Supplementary Materials 1, 2). FDR-adjusted P < 0.05 were further used to assess the robustness of the core functional modules. Although FDR correction reduced the number of DEPs identified in the cecal microbiota, the core functional categories and major contributing taxa remained highly consistent with those identified using nominal P-value-based screening. Notably, no small-intestinal proteins met the combined FDR-adjusted P-value and effect-size criteria (Figure S8-9, Supplementary Materials 1, 2).

3.5. Ethanol-associated remodeling of microbiota nutrient acquisition capabilities

The ability to acquire nutrients is crucial for shaping a microbial niche. ⁴¹ The SusC/SusD-dependent polysaccharide uptake system was significantly up-regulated in the cecal microbiota (Figure 5A), including the SusC/RagA family TonB-linked outer membrane protein and the RagB/SusD family nutrient uptake outer membrane protein, with Muribaculaceae and Barnesiellaceae being the major contributors (Figure 5G-H). In addition, the relative abundance of the biopolymer transporter ExbD, a key energy-supporting protein for the system, was also significantly elevated. Consistent with these changes, the relative abundance of glycoside hydrolases contributed by Bacteroidota increased significantly, whereas that contributed by Bacillota decreased significantly (Figure 5I). At the metabolite level, acute ethanol exposure was associated with a decreasing trend in carbohydrate levels (Figure S10A). Collectively, these findings suggest an enhanced capacity for polysaccharide/glycan acquisition in Bacteroidota.

Figure 5.

Integrated multi-omics revealed remodeling of nutrient acquisition and energy metabolism of the gut microbiota. (A-F) Heatmaps show the relative abundances of substrate-acquisition-related proteins expressed by the cecal microbiota (A-C) or the small-intestinal microbiota (D-F): polysaccharide uptake (A, D), other substrate uptake/outer membrane-associated functions (B, E), and amino acid/peptide uptake (C, F). (G-H) Relative abundances of proteins associated with the SusC/SusD-dependent polysaccharide uptake system expressed by Muribaculaceae (G) or Barnesiellaceae (H) in the cecum. (I) Relative abundance of glycoside hydrolases expressed by Bacteroidota and Bacillota in the cecum. (J) Relative abundance of TRAP transporter substrate/solute-binding proteins expressed by Desulfovibrionaceae in the cecum. (K) Relative abundances of ABC transporter proteins associated with amino acid/peptide uptake expressed by Lachnospiraceae in the cecum. (L) Relative abundance of glycoside hydrolases expressed by Lachnospiraceae in the cecum. (M) Relative abundances of ABC transporter proteins associated with carbohydrate substrate uptake expressed by Lachnospiraceae in the cecum. (N) Relative abundance of BMP family ABC transporter substrate-binding proteins expressed by Lachnospiraceae in the cecum. (O) Relative abundances of proteins associated with the SusC/SusD-dependent polysaccharide uptake system expressed by Muribaculaceae in the small intestine. (P) Relative abundance of glycoside hydrolases expressed by Bacteroidota in the small intestine. (Q) Relative abundances of fumarate respiration-related proteins (EC:1.1.1.37, EC:4.2.1.2, and EC:1.3.5.1) expressed by Muribaculaceae in the small intestine and the cecum. (R) Relative abundances of proteins involved in the dissimilatory sulfate reduction pathway (EC:2.7.7.4, EC:1.8.99.2, and EC:1.8.1.22) expressed by Desulfovibrionaceae in the small intestine and the cecum. n = 3; *Nominal P < 0.05 versus the control group.

Furthermore, the relative abundance of urease expressed by Muribaculaceae was also significantly increased (Figure S10B), suggesting increased potential for urea utilization. The significant increase in TonB-dependent receptors expressed by Bacteroidota further suggests enhanced active uptake of substrates such as siderophores and vitamins. ^{42 , 43} The increase in OmpA-family proteins may reflect remodeling of outer-membrane maintenance accompanying altered transport activity ^{44 , 45} (Figure 5B). TRAP transporters are often linked to the uptake of C4 dicarboxylates. ⁴⁶ Under long-term ethanol exposure, the total relative abundance of the TRAP transporter substrate/solute-binding protein increased significantly (Figure 5B), and its expression in Desulfovibrionaceae also increased significantly (Figure 5J).

In the cecum, ABC transporter proteins for carbohydrate uptake showed a downward trend (Figure 5A), while those for amino acid/peptide uptake decreased significantly (Figure 5C). Concurrently, total relative levels of amino acids and peptides exhibited an upward trend (Figure S10C). Notably, the expression of ABC transporter proteins associated with amino acid/peptide uptake expressed by Lachnospiraceae decreased significantly (Figure 5K). While the relative abundance of Lachnospiraceae-derived glycoside hydrolases was unaffected (Figure 5L), the expression of ABC transporter proteins mediating carbohydrate uptake declined to varying extents (Figure 5M). The BMP family ABC transporter substrate-binding protein, typically associated with nucleotide uptake, ⁴⁷ was also significantly reduced (Figure 5B), and this reduction was largely attributable to Lachnospiraceae (Figure 5N). These results indicate that ethanol exposure was associated with reduced representation of multiple nutrient-uptake proteins in Lachnospiraceae.

In contrast to the cecal microbiota, the small-intestinal microbiota showed significantly reduced expression of RagB/SusD family nutrient uptake outer membrane proteins (Figure 5D). Notably, Muribaculaceae exhibited significantly reduced expression of both SusC/RagA family TonB‑linked outer membrane proteins and RagB/SusD family nutrient uptake proteins (Figure 5O), and there was no significant change in the relative abundance of urease (Figure S10D). The relative abundance of glycoside hydrolases expressed by Bacteroidota in the small intestine was significantly reduced (Figure 5P), while the relative levels of carbohydrate substrates showed no significant change (Figure S10E). Additionally, under long-term ethanol exposure, TRAP transporter substrate-binding proteins increased significantly (Figure 5E); the ABC transporter proteins involved in amino acid uptake showed an upward trend (Figure 5F), accompanied by a gradual decline in the total relative levels of peptides and amino acids (Figure S10F); all these proteins were derived from Desulfovibrionaceae in the small intestine. These findings further highlight the spatial and taxon-specific remodeling of microbial nutrient acquisition.

3.6. Ethanol-associated remodeling of microbiota energy metabolism strategies

In the cecal microbiota, Muribaculaceae showed significantly elevated expression of fumarate respiration-related proteins, including EC:1.1.1.37, EC:4.2.1.2, and EC:1.3.5.1 (Figures 5Q, S11), highlighting the contribution of an efficient fumarate respiration pathway to Muribaculaceae metabolic reprogramming. Moreover, the significant increases in EC:2.7.7.4, EC:1.8.99.2, and EC:1.8.1.22 imply enhanced dissimilatory sulfate reduction in Desulfovibrionaceae (Figure 5R). However, within the small-intestinal microbiota, the relative abundances of EC:4.2.1.2 and EC:1.3.5.1 decreased significantly following long-term ethanol exposure (Figure 5Q); the relative abundances of EC:1.8.1.22 and F₀F₁ ATP synthase expressed by Desulfovibrionaceae showed no significant change (Figures 5R, S12).

We analyzed the intermediates of central carbon metabolism in the cecum. Under acute ethanol exposure, D-fructose 6-P was significantly decreased, whereas succinic acid was significantly increased (Figure 6A–H). Consistently, microbial proteins involved in the corresponding pathways were broadly remodeled (Figure 6I). At the community level, EC:2.4.1.21 and EC:2.7.7.27 were significantly decreased, indicating a systematic weakening of glycogen synthesis. In contrast, EC:3.1.3.11, EC:4.2.1.11, EC:4.1.2.4, and EC:2.7.1.2 increased significantly, indicating increased abundance of proteins assigned to glycolysis and associated bypass pathways. In Muribaculaceae, the relative abundance of EC:1.2.7.1 increased significantly following acute ethanol exposure, whereas EC:5.4.2.2 and EC:2.7.2.3 increased significantly following long-term ethanol exposure (Figure 6J–L). EC:1.2.7.1 and EC:5.4.2.2 are involved in the conversion of pyruvate to acetyl-CoA and the rearrangement of the glucose-6-P intermediate, respectively, whereas EC:2.7.2.3 catalyzes a key ATP-generating step, suggesting enhanced energy acquisition and metabolic adaptability in Muribaculaceae. In contrast, EC:1.2.7.1 expressed by Lachnospiraceae was significantly reduced under long-term ethanol exposure (Figure 6L). Notably, EC:4.1.2.4 and EC:2.7.1.90 in Muribaculaceae were significantly increased under both acute and long-term ethanol exposure (Figure 6M-N). EC:4.1.2.4 links the degradation of deoxyribose 5-P to glycolysis by producing D-glyceraldehyde 3-P, while EC:2.7.1.90 drives glycolysis in an ATP-independent manner, ^{48 , 49} further suggesting that Muribaculaceae may support increased glycolytic capacity and ATP generation.

Figure 6.

Integrated multi-omics revealed remodeling of the central carbon metabolism pathway in the gut microbiota. (A–H) Relative levels of metabolites in the cecum related to central carbon metabolism: D-glucose, D-fructose 6-P, pyruvate, and D-lactic acid (A–D); citric acid, malic acid, fumaric acid, and succinic acid (E–H) (n = 8). (I) Heatmaps show the relative abundances of proteins involved in central carbon metabolism and related pathways in the cecal microbiota (n = 3). (J–L) Relative abundances of EC:5.4.2.2 (J), EC:2.7.2.3 (K), and EC:1.2.7.1 (L) expressed by Muribaculaceae and Lachnospiraceae in the cecum (n = 3). (M-N) Relative abundance of EC:4.1.2.4 (M) or EC:2.7.1.90 (N) expressed by Muribaculaceae in the cecum (n = 3). *Nominal P < 0.05 versus the control group.

The relative abundances of EC:2.3.1.39, EC:2.3.1.179, EC:2.3.1.180, and EC:1.1.1.100 increased simultaneously, suggesting that the protein-level representation of de novo fatty acid synthesis increased. At the metabolic level, total fatty acid levels did not change significantly. However, acute ethanol exposure markedly increased straight-chain fatty acids, while long-term exposure significantly decreased branched-chain fatty acids (Figure S13A-C). The accumulation of straight-chain fatty acids may be associated with increased availability of acetyl-CoA, ¹² and the reduction in branched-chain fatty acids may be consistent with an impaired ability of the microbiota to uptake branched-chain amino acids, which are typically essential for the microbiota to synthesize branched-chain fatty acids. ⁵⁰ Proteins involved in succinyl-CoA synthesis were broadly increased, which may favor a more abundant carbon-skeleton entry point for biosynthetic reactions. The increase in succinate-related features was consistent with greater use of fumarate-associated anaerobic metabolism. In addition, the key PRPP synthase (EC:2.7.6.1) was significantly increased under long-term ethanol exposure, suggesting that the microbiota may be under increased demand for nucleotide biosynthesis or repair-related processes. ^{51 , 52}

However, glycolysis in the small-intestinal microbiota did not increase significantly, and the relative abundance of proteins associated with succinate respiration decreased significantly (Figure S13D-E); at the metabolite level, only D-glucose decreased significantly following acute ethanol exposure (Figure S13F-L). Together, these findings indicate region- and family-specific remodeling of microbial energy metabolism.

3.7. Ethanol-associated stress adaptations: oxidative stress defense and proteostasis maintenance

Ethanol exposure may be associated with microbial stress responses, particularly oxidative stress defense and proteostasis. Trend analysis showed that the rapid upregulation of antioxidant proteins after acute ethanol exposure in the small intestine and cecum was primarily driven by Muribaculaceae. Under long-term ethanol exposure, antioxidant proteins were persistently responsive mainly in Muribaculaceae and Desulfovibrionaceae, but persistently suppressed in Lachnospiraceae (Figures 7A-B, S14-15).

Figure 7.

Integrated multi-omics revealed ethanol-associated stress responses in the gut microbiota. (A-B) Temporal trend analysis based on Mfuzz, and a heatmap shows the relative abundances of antioxidant proteins in the small-intestinal microbiota (A) or the cecal microbiota (B); protein classes are indicated in the right-hand annotation (n = 3). (C-D) Heatmaps show the relative abundances of various antioxidant protein types in the small-intestinal microbiota (C) or the cecal microbiota (D) (n = 3). (E) Total relative abundances of microbial antioxidant proteins in the small intestine and the cecum (n = 3). (F-G) Relative abundances of antioxidant proteins expressed by Muribaculaceae, Barnesiellaceae, Desulfovibrionaceae, and Lachnospiraceae in the small intestine (F) or the cecum (G) (n = 3). (H-I) Relative abundance of reverse rubrerythrin (H) or NAD-reducing hydrogenase (I) expressed by Lachnospiraceae in the cecum (n = 3). (J) Relative abundances of NADH peroxidase expressed by Lachnospiraceae, Oscillospiraceae, and Muribaculaceae in the cecum (n = 3). (K) Relative abundance of NAD(P)⁺ transhydrogenase expressed by Desulfovibrionaceae in the cecum (n = 3). (L) NADH content in the cecum (n = 5). (M-N) Relative abundances of chaperonin GroEL (M) or chaperone DnaK (N) expressed by Lachnospiraceae, Oscillospiraceae, Muribaculaceae, and Desulfovibrionaceae in the cecum (n = 3). (O) Relative abundances of flagellin expressed by Desulfovibrionaceae in the small intestine and the cecum (n = 3). *Nominal P < 0.05 versus the control group.

Multiple antioxidant protein categories were significantly up-regulated in the cecum but not broadly altered in the small intestine (Figure 7C-D). Multiple oxidative stress-related metabolites were significantly altered in the cecum, consistent with metabolite signatures associated with lipid peroxidation and oxidative protein damage (Figure S16). The significant upregulation of catalase, superoxide dismutase, rubrerythrin family proteins, desulfoferrodoxin, and cytochrome-c peroxidase is consistent with the increased demand for peroxide- and superoxide-related defense ^53-55 (Figure 7D). The significant upregulation of thiol peroxidase and peroxiredoxin under long-term ethanol exposure, alongside the upward trend in thioredoxin family proteins, indicates an increased need to repair thiol groups and aberrant disulfide bonds ^{56 , 57} (Figure 7D), suggesting that proteostasis is under stress.

However, the total relative abundance of antioxidant proteins did not change significantly in either the small intestine or the cecum (Figure 7E), suggesting a reconfiguration rather than an overall enhancement of the antioxidant system. Long-term ethanol exposure significantly reduced antioxidant proteins from Muribaculaceae in the small intestine, while significantly increasing those from Barnesiellaceae (Figure 7F). In the cecum, antioxidant proteins from Muribaculaceae, Barnesiellaceae, and Desulfovibrionaceae increased significantly, whereas those from Lachnospiraceae remained significantly suppressed (Figure 7G).

In the cecum, despite the overall reduction in Lachnospiraceae-derived antioxidant proteins, reverse rubrerythrin increased (Figure 7H), accompanied by persistent decreases in NAD-reducing hydrogenase and NADH peroxidase (Figure 7I-J). Together, these changes suggest enhanced oxidative stress but weakened NAD(H) recycling, indicating impaired redox homeostasis and a potential constraint on NAD⁺-dependent energy metabolism. ^{58 , 59} In contrast, the relative abundance of NADH peroxidase in Muribaculaceae significantly increased (Figure 7J), suggesting that Muribaculaceae may support H₂O₂ detoxification and NAD⁺ regeneration. ^{60 , 61} In Desulfovibrionaceae, the relative abundance of NAD(P)⁺ transhydrogenase significantly increased under long-term ethanol exposure (Figure 7K), suggesting that it may enhance the transhydrogenation of reducing equivalents from NADH to NADPH, thereby boosting antioxidant defense. ^{62 , 63} Under ethanol exposure, cecal NADH peroxidase and community NADH levels were both significantly reduced (Figure 7D, 7L), consistent with altered NADH reoxidation and redistribution of reducing equivalents. ⁶⁴

Ribosomal proteins are central to protein synthesis and homeostasis. In the cecum, the relative abundance of ribosomal proteins decreased significantly in Lachnospiraceae, indicating suppressed metabolic activity. ^{65 , 66} In contrast, ribosomal proteins increased significantly in Barnesiellaceae and Desulfovibrionaceae (Figure S17A). Meanwhile, in the small-intestinal microbiota, only Desulfovibrionaceae showed a significant increase in ribosomal proteins after long-term ethanol exposure (Figure S17B).

GroEL and DnaK are crucial molecular chaperones that prevent protein misfolding during biosynthesis or under stress conditions. ^{67 , 68} In the small intestine, the relative abundances of GroEL and DnaK increased significantly under long-term ethanol exposure (Figure S17C-D). By contrast, the total relative abundance of GroEL was significantly reduced in the cecum (Figure S17C). In the cecum, GroEL and DnaK derived from Lachnospiraceae were significantly reduced under both acute and long-term ethanol exposure. In contrast, Muribaculaceae-derived GroEL increased after acute exposure, whereas Desulfovibrionaceae-derived GroEL increased after long-term exposure (Figure 7M). Muribaculaceae-derived DnaK increased under both exposure conditions, while Desulfovibrionaceae-derived DnaK increased only after long-term exposure (Figure 7N).

Flagella propel bacterial motility and facilitate movement toward favorable niches, and flagellin is the major structural component of the flagellum. ^{69 , 70} Although the total relative abundance of flagellin did not change significantly in either the small intestine or the cecum (Figure S17E), flagellin expressed by Desulfovibrionaceae was significantly increased (Figure 7O). In addition, although the total relative abundance of Lachnospiraceae-derived flagellin did not change significantly (Figure S17F), a substantial number of individual flagellin proteins from Lachnospiraceae were down-regulated in the cecum (Figure S18). Together, these patterns indicate a taxon-specific redistribution of protein synthesis, chaperone-mediated proteostasis, and motility-related functions in response to ethanol exposure.

4. Discussion

This study provides a region-resolved functional map of how ethanol intake is associated with gut microbiota remodeling. We show that ethanol-associated microbial responses are spatially and taxon-specific. Importantly, the current study identifies coordinated functional patterns and candidate mechanisms, but it does not establish direct causality between specific microbial changes and host physiology.

In this study, ethanol and acetaldehyde were quantified to establish the spatiotemporal exposure context for regional microbial functional analysis. Ethanol levels in blood and the gastrointestinal tract do not follow a simple monotonic rise-and-fall trajectory but instead exhibit oscillatory dynamics. This pattern is consistent with the coupled effects of intermittent gastric emptying, rapid proximal small-intestinal absorption, systemic redistribution, and metabolic clearance. Ethanol concentrations in the distal small intestine, cecum, and colon are highly synchronized with blood ethanol concentration, indicating that distal ethanol exposure is primarily governed by the systemic circulation, in line with the conventional understanding. ⁷¹ By contrast, acetaldehyde levels in distal intestinal regions are shaped by circulatory delivery as well as by local generation and elimination, which may arise from mucosal and microbial metabolic pathways. ^{8 , 13 , 72}

Although small-intestinal microbiota exhibit increased bile acid tolerance and/or utilization potential after ethanol exposure, the causal relationship between changes in bile acids and small-intestinal microbiota remodeling remains unclear because the concentration, composition, and metabolic flux of bile acids are not directly measured. Targeted bile acid profiling and metabolic flux analysis are needed. In contrast, the remodeling patterns of the cecal and colonic microbiota are more complex. Campylobacterota can perform anaerobic respiration using nitrate or fumarate as terminal electron acceptors, and increases in these organisms may relate to ethanol-induced redox shifts and altered terminal electron acceptor availability. ^{73 , 74} Eubacterium-related taxa, Clostridium, and Muribaculum are commonly linked to complex carbohydrate degradation and have mucin-degrading potential. ^75-78 These simultaneous increases may indicate increased availability of complex polysaccharides and glycans in the cecum and colon. ⁷⁹ Future research should focus on targeted assays to determine the contribution of host mucosal damage and the availability of dietary polysaccharides.

At the metagenome level, we observed enrichment of ethanol metabolism-related genes in the small intestine and cecum, with distinct enrichment patterns. However, this genetic potential does not translate into broad protein-level activation. Under long-term ethanol exposure, there was no significant increase in the total relative abundance of various proteins associated with ethanol metabolism. Within complex microbial communities, gene abundances, the taxa that harbor them, and the corresponding proteins are not necessarily correlated. ^{80 , 81} Because metagenomic, metatranscriptomic, and metaproteomic measurements represent distinct regulatory layers, ^{81 , 82} gene, transcript, and protein abundances may differ due to taxonomic factors, post-transcriptional regulation, translational efficiency, protein turnover and stability, and the limited detectability of low-abundance proteins by mass spectrometry. ^{80 , 83 , 84} Lachnospiraceae were the principal taxonomic contributors to the detected proteins involved in ethanol metabolism. Ethanol exposure significantly reduced functional modules involved in substrate acquisition, energy metabolism, oxidative stress defense, and proteostasis in Lachnospiraceae. This may further limit the extent to which community-level genetic potential for ethanol metabolism is translated into detectable protein-level expression and functional output.

Consistent with previous studies, we found that in vitro incubation of gut contents reduced ethanol levels, suggesting that the microbiota retains ethanol-metabolizing capacity, potentially through partial relief of host- and niche-imposed metabolic constraints under ex vivo conditions and/or the activity of low-abundance community members. ^{85 , 86} In addition, functional validation using three strains expressing ethanol-metabolism-related proteins showed that the reduction in ethanol and acetaldehyde was confined to the stomach. This effect may reflect transient gastric retention of viable cells and their basal enzymatic activity. Although this short-term strain-gavage experiment does not reflect the ethanol-processing pattern of the endogenous microbiota under long-term ecological conditions, it still supports the notion that the selected strains possess a basal ethanol-metabolizing capacity. Future studies incorporating strain-specific qPCR, fecal microbiota transplantation, and site-specific persistence analyzes will be required to establish whether these bacteria can stably colonize relevant intestinal regions and contribute to ethanol metabolism in a long-term ecological context.

The most consistent functional patterns identified in this study involved substrate acquisition, energy metabolism, oxidative stress defense, and proteostasis. These patterns suggest that ethanol exposure is associated with ecological adaptation of microbial functional modules. One possible explanation is that the oxidative stress associated with ethanol metabolism contributed to the reconfiguration of the microbial antioxidant defense system. ^{87 , 88} Meanwhile, oxidative damage to proteins may have been mitigated by the thioredoxin system and GroEL/DnaK, thereby supporting proteostasis. ^{56 , 89} The coordinated increase in proteins related to energy-generating pathways may have helped meet the increased demand for ATP during oxidative defense and protein repair, ^90-92 while enhanced nutrient-acquisition capacity may have provided additional substrates to support the energetic and biosynthetic demands of stress adaptation and homeostasis maintenance. ^{85 , 93} In the cecum, Muribaculaceae and Barnesiellaceae showed increased abundance of proteins involved in complex polysaccharides and glycans acquisition, including glycoside hydrolases and components of the SusC/SusD-dependent polysaccharide uptake system, which may have supported glycolysis and fumarate-associated energy metabolism. Desulfovibrionaceae showed increased representation of TRAP transporters involved in dicarboxylate and other organic acid uptake, along with proteins related to dissimilatory sulfate reduction, consistent with enhanced capacity for anaerobic energy conservation.

The concurrent upregulation of fumarate respiration-related proteins in Muribaculaceae and dissimilatory sulfate reduction-related proteins in Desulfovibrionaceae suggests coordinated remodeling of anaerobic electron-acceptor utilization in the cecum. Previous studies have shown that fumarate- and sulfate-reducing bacteria can participate in H₂-mediated electron transfer in anaerobic communities. Thus, the observed changes raise the possibility of a broader syntrophic electron-sharing network. ^{94 , 95} However, direct interactions between Muribaculaceae and Desulfovibrionaceae remain to be validated through H₂/formate measurements, isotope tracing, hydrogenase activity assays, and defined co-culture experiments. Furthermore, the enhanced dissimilatory sulfate reduction pathway in Desulfovibrionaceae provides a potential metabolic basis for coupling ethanol oxidation and sulfate reduction. ⁹⁶

Collectively, the study challenges the simplistic notion that “gut microbiota adapt to ethanol exposure through broad enhancement of ethanol metabolism”. Our data support a model in which ethanol intake is associated with region- and taxon-specific functional remodeling of the gut microbiota. These remodeling features encompassed substrate acquisition, energy metabolism, oxidative stress defense, and proteostasis, thereby providing a conceptual framework and testable hypotheses for future mechanistic studies.

Compared to metagenomics and metatranscriptomics, metaproteomics provides a downstream view of detectable protein-level functional output of the gut microbiota. ⁹⁷ However, metaproteomics technologies often lose information about low-abundance proteins and depend heavily on databases. ^{98 , 99} This study has additional limitations. First, the present study used a gavage-based ethanol exposure model in male C57BL/6J mice. Although this design enabled precise control of ethanol dose and sampling time, it may not fully reproduce voluntary or chronic drinking patterns, and the generalizability of the findings across sexes, strains, and exposure models remains to be determined. Second, although ethanol and acetaldehyde were quantified to define the regional exposure landscape, acetate concentrations, bile acid profiles, and fluxes, and host physiological outcomes were not directly assessed. Consequently, the observed microbiota remodeling cannot be attributed solely to ethanol, ethanol-derived metabolites, or secondary host physiological changes. Third, the metaproteomic analysis was based on a relatively limited sample size (n = 3); therefore, functional-module comparisons based on nominal P values should be regarded as exploratory. These analyzes were intended to construct a region-resolved functional map and generate testable hypotheses, not to establish causal mechanisms. Although the FDR-adjusted metaproteomic findings and the concordant directional patterns observed across independent omics layers increase our confidence in the biological consistency of the major functional categories and their principal taxonomic contributors, these data do not allow us to determine whether the observed changes represent direct effects of ethanol, consequences of ethanol-induced ecological stress, or indirect responses to community restructuring and altered host physiology. Future studies should include larger metaproteomic cohorts, together with targeted perturbation experiments, targeted proteomics, enzyme-activity assays, stable-isotope tracing, bile-acid profiling, germ-free or defined-microbiota animal models, and fecal microbiota transplantation. These approaches will be required to establish the mechanistic and causal roles of the candidate functional modules and to quantify the in vivo microbial metabolic flux of ethanol-derived carbon.

Supplementary Material

Supplementary materials Statistical analysis of differentially expressed proteins in the cecum FDR adjusted P.pdf

Supplementary materials Statistical analysis of differentially expressed proteins in the cecum FDR adjusted P.pdf

Supplementary materials1.pdf

Supplementary materials1.pdf

Supplementary materials2.pdf

Supplementary materials2.pdf

Supplementary Material

Supplement_figures (1)-clean source

Funding Statement

This study was funded by the National Natural Science Foundation of China, Grant/Award Numbers: 22138004 and 22422807.

Disclosure of potential conflicts of interest

No potential conflicts of interest were disclosed.

Data availability statement

The raw 16S rRNA sequencing and metagenomic data from this study have been uploaded to the China National Center for Bioinformation (https://ngdc.cncb.ac.cn/gsa/) under access numbers CRA038194, CRA038192, and CRA038191. Further data to support these findings are available from the corresponding author upon reasonable request.

Ethics approval

The animal experiment was approved by the Experimental Animal Ethics Committee of Jiangnan University (JN.No20241230c0960409[679]).

Supplementary material

Supplemental data for this article can be accessed at https://doi.org/10.1080/19490976.2026.2734657.