Impairment of the intestinal epithelium barrier through alteration of proliferation and differentiation balance in alcohol use disorder

Abstract

BACKGROUND

Alcohol use disorder (AUD) profoundly affects human health, yet its direct impact on the intestinal epithelium and stem cell niche remains insufficiently understood.

AIM

To characterize alcohol-induced alterations in intestinal epithelial morphology, proliferation, differentiation, and barrier-related functions in AUD patients.

METHODS

Duodenal biopsies from healthy and AUD patients were analyzed for epithelial architecture, proliferation, differentiation activity, and lineage composition using histological, immunofluorescence, proteomic, and transcriptomic approaches. Findings were complemented by studies in human-derived enteroids exposed to 40- or 70-mM ethanol to assess its direct effects on epithelial proliferation and signaling pathways.

RESULTS

A subset of AUD patients exhibited marked epithelial remodeling, including increased crypt depth, shortened villi, and expanded proliferation in the crypt base and transit-amplifying zone. Ki67⁺ cells and OLFM4⁺ stem-like compartment were increased, with CD44 identified as a potential driver of crypt-base proliferation. Moreover, a higher number of Ki67⁺ crypts co-expressed the enteroendocrine cells marker CHGA⁺ and the stem cell marker OLFM4. Ki67⁺ Kruppel-like factor 4⁺ cells were elevated in transit-amplifying zone, correlating with higher MUC2, MUC5AC, and anterior gradient-2 expression, alongside reduced antimicrobial peptides DEFA5, DEFA6, and REG3A. Nutrient absorption markers MTTP, FABP2, and SLC1A5 were dysregulated. Ethanol-exposed enteroids showed increased Ki67⁺ cells and cyclin B1 mRNA without changes in CD44 or Axin2 indicating a proliferative effect independent of canonical Wnt signaling.

CONCLUSION

Heavy alcohol consumption induces profound alterations in intestinal epithelial morphology, stem cell niche dynamics, and functional barrier properties. Together, these findings suggest that ethanol disturbs the intestinal proliferation-differentiation balance.

Key Words: Alcohol use disorder; Intestinal epithelium; Intestinal barrier; Stem cell niche; Intestinal organoids; Enteroids

Core Tip: The effects of alcohol on the human intestinal epithelium are poorly understood, as existing studies are limited, inconsistent, and largely based on animal models that incompletely reflect human pathology. Human-based analyses reveal that heavy alcohol consumption disrupts intestinal epithelial homeostasis by altering stem cell niche and lineage commitment. Duodenal crypts displayed increased depth and expanded proliferation driven by enlarged OLFM4⁺ stem-like compartments and elevated Ki67⁺ and CD44⁺ cells. Ethanol induced aberrant proliferation, impaired secretory lineage commitment, antimicrobial defense and nutrient transport. Using human intestinal organoids, we confirm a direct, Wnt-independent proliferative effect of ethanol on epithelial cells.

INTRODUCTION

Alcohol consumption has been an integral part of many cultures throughout history. However, like any psychoactive substance, its chronic use can lead to addiction and physical or psychological dependence. Various societal factors contribute to the development of an alcohol use disorder (AUD) and the profound impact of heavy alcohol consumption on human health[1-4].

The main point of entry of alcohol into the body is the gut. The human intestinal barrier is the center of several dynamic processes including nutrient absorption by enterocytes, barrier defense by Paneth, goblet and immune cells and mucus secretion by goblet cells[5,6]. Each of those processes plays an important role in the intestinal homeostasis and evidence suggests that this tightly regulated process is disturbed in AUD. As a part of the intestinal barrier, the gut epithelium comprises three essential compartments: The crypt where the stem cells are located, the transit-amplifying (TA) zone containing proliferating progenitor cells and the villi where terminal differentiated epithelial cells reside[7]. Most of the studies in humans and rodent models regarding the impact of alcohol on the gut epithelium predominantly highlight intestinal permeability driven by tight junction disruption in AUD related to alcohol-associated liver disease (ALD)[8-11]. Our group previously showed that tight junction disruption is only present in a subset of heavy drinkers with ALD suggesting that other mechanisms at the epithelium level besides increased paracellular permeability might contribute to gut barrier dysfunction and later the development of ALD[12].

Intestinal stem cells (ISCs) are the guardians of a good balance between the number of progenitor cells in the transit amplifying zone committed to differentiate and the optimal proportion of the various epithelial cell populations that comprise the differentiated cells. This highly controlled process is essential for intestinal homeostasis. Several studies reported deleterious effects of alcohol on stem cells in different tissues[13,14] and ISCs are not spared from those effects suggesting some kind of loss of stemness. Quiescent stem cells known as label-retaining cells are capable of replacing the stem cell pool upon tissue damage as are differentiated epithelial cells that might revert to stem cell-like cells[15,16].

Studies regarding the impact of alcohol on ISCs are scarce and somewhat contradictory. Using mouse models, some suggested a profound damage of ISCs upon ethanol use[17] whereas others postulate an increased cell proliferation through upregulation of stem cell markers[18] associated with upregulation of Wnt β-catenin target genes Axin2, Tcf4 and cyclinD1.

Here we describe the deleterious effects of alcohol on the intestinal epithelium and the stem cell niche in humans and the association with pathways implicated in finetuning the epithelial proliferation-differentiation balance. We suggest another potential mechanism of stem cell niche maintenance that drives alcohol induced proliferation on the epithelium based on the rescue of the crypt base population by a specific dedifferentiated cell type originating from enteroendocrine cells (EECs). We further demonstrate a shift of the differentiation program towards the secretory lineage and, in particular, goblet cells associated with over-expression of Kruppel-like factor 4 (KLF4) in proliferating cells in the transient amplifying zone. Using human-derived organoid models, we then studied in vitro the direct effect of ethanol on the epithelium and whether ethanol is the principal driver of the observed changes in vivo.

MATERIALS AND METHODS

Patients’ recruitment and sampling

AUD patients admitted for a highly standardized and controlled 3-week detoxification and rehabilitation program were enrolled. Fasting blood samples were collected at admission. All patients underwent routine upper gastrointestinal endoscopy on the second day of admission where biopsies from the distal duodenum were obtained. Healthy, age-and sex-matched subjects undergoing out-patient upper gastro-intestinal endoscopy for dyspepsia or reflux symptoms were recruited as controls. Healthy volunteers did not use medication or presented any clinically significant co-morbidities on history taking. Consequently, the study design did not include routine clinical laboratory testing in controls. In addition, control patients were only retained for the study if they had a normal gastro-intestinal endoscopy and no histological changes on the biopsy samples. The study protocol conforms to the ethical guidelines of the 1975 Declaration of Helsinki and was approved by the institution’s human research and Ethical Committee (Comité Ethique Hospitalo-facultaire, Cliniques Universitaires Saint Luc), No. B403201422657. Written informed consent was obtained from all patients and controls.

Patients’ classification according to the duodenal epithelium alteration

Based on histological evaluation of the epithelium, AUD patients were initially classified in two groups namely AUD patient with normal and those with an altered morphology (increased crypt hyperplasia and mild to moderate villi atrophy according to the modified Marsh-Oberhuber classification used in celiac disease). To more precisely quantitate this global histological evaluation, we measured the villi and crypt length in all the healthy subjects and calculated the crypt to villi ratios per subject. A valid analysis required the presence of at least two crypts connected to villi. We used the mean values of crypt and villi of the controls plus/minus SD of the mean (SEM) to further refine morphology alterations. In addition to the histological assessment, a minimum of two out of three criteria must be met to be classified into the altered AUD group: (1) Crypt/villi ratio > 0.653 (mean of the controls plus SEM); (2) Long crypt length > 143 μm (mean of the controls plus SEM); and/or (3) Short villi length < 309 μm (mean of controls minus SEM).

RNA isolation and quantitative reverse transcription-polymerase chain reaction

RNA was extracted from duodenal biopsies and enteroids using Promega Maxwell^® RSC miRNA Tissue kit (AS1460, Promega, Leiden, Netherlands). Reverse transcription was performed by High-Capacity complementary DNA Reverse Transcription Kit from Applied Biosystems™ (reference: 4368813, LIFE Technologies, Merelbeke, Belgium). Quantitative reverse transcription-polymerase chain reaction was performed using SYBR Green assays under standard conditions with Rotor gene Q series software version 2.3.1 (Version 2021). Samples were normalized to the housekeeping gene RPL19. Primers are depicted in Supplementary Table 1.

RNA sequencing

RNA sequencing has been performed as published recently by our group[19] and data has been re-analyzed for the current study. Briefly, RNA from duodenal biopsies was extracted using QIAGEN RNeasy Plus Mini kit (Cat. No. 74134, QIAGEN, Germany). High-throughput transcriptome profiling by RNA sequencing was performed on a NovaSeq S4 at the IGM Sequencing facility (University of California, San Diego, CA, United States), followed by differential expression and functional enrichment analysis. All sequencing data were analyzed using the Automated Reproducible MOdular workflow for preprocessing and differential analysis of RNA sequencing data (ARMOR v1.5.4) pipeline[20]. In this pipeline, reads underwent a quality check using FastQC (Babraham Bioinformatics). Quantification and quality control results were summarized in a MultiQC report before being mapped using Salmon[21] to the transcriptome index which was built using all Ensembl complementary DNA sequences obtained in the Homo_sapiens.GRCh38.cdna.all.fa file. Then, estimated transcript abundances from Salmon were imported into R using the tximeta package[22] and analysed for differential gene expression with edgeR[23].

Proteomics

Proteomic analysis was performed on duodenal biopsy samples by liquid chromatography-mass spectrometry essentially as previously described[24]. Prior to injection, the liquid chromatography and mass spectrometry samples were solubilized in 0.01% formic acid aqueous solution and MPDSmix was spiked. The samples (2.5 μg) were injected into a 1D Acquity UPLC M-Class (Waters, MA, United States) chromatography coupled online with a Q Exactive™ Plus Hybrid Quadrupole-Orbitrap™ mass spectrometer (Thermo Fisher Scientific, MA, United States), equipped with a nano-electrospray source operated in positive ion mode. To identify differentially expressed proteins, we used the limma v.3.60.4[25] R package to perform a linear model analysis on the label-free quantification intensities. The label-free quantification intensities were log2-transformed. Protein abundance data were filtered to remove proteins with missing values in more than 80% of the samples. A linear model was fitted to the data for each protein, and empirical Bayes moderation was applied to the standard errors to increase the power of the statistical tests. Proteins with an adjusted P value below 0.05 were considered statistically significant.

Immunohistofluorescence

Patient-derived duodenal biopsies were fixed in formaldehyde for 48 hours before being processed for paraffin inclusion. 4 μm sections were deparaffinized and rehydrated in successive baths of xylol and isopropanol. Endogenous peroxidase activity was blocked with 3% hydrogen peroxide in methanol for 15 minutes. Antigen retrieval with citrate buffer was performed followed by a blocking of non-specific antibody binding with Tris-buffered saline (TBS) containing 0.1% Tween20 (TBS-T) and 5% bovine serum albumin. The slides were then incubated with the following primary antibodies for 1 hour at RT or 4 °C overnight (Supplementary Table 2) in TBS-T containing 1% bovine serum albumin. Horseradish peroxidase-conjugated polymers were used as secondary antibodies (Agilent, CA, United States). Detection was performed by tyramide signal amplification using CF-conjugated tyramides (Biotium, CA, United States). Sequential labeling of up to five additional markers on the same tissue section was achieved by repeating heat-induced epitope retrieval and tyramide signal amplification cycles. Staining conditions are given in Supplementary Table 2 for the following markers: OLFM4, MUC2, LYZOZYME, CHGA, Ki67, pEGFR (Tyr1068), Axin2, pERK (Thr202/Tyr204), CD44, E-cadherin, KLF4. DAPI was used on the slides as counterstaining. The slides were mounted with Dako fluorescence mounting medium (Agilent, CA, United States) and scanned with an Axioscan.z1 slide scanner (Zeiss, Germany).

Computer-assisted image analysis

Image analyses were performed using QuPath (version 0.5.1) and Zen (version 3.5, Blue edition) software. Paneth cells were identified by lysozyme (LYZ) staining to locate crypt base columnar (CBC) cells and were excluded from stem cell counts. To delineate the crypt region from the TA zone, we applied the definition of a crypt which consists of cells from position 0 (bottom of the crypt) to position 4⁺[7,26]. The upper limit of the TA zone is defined by the last cell expressing the proliferation marker Ki67 where the villi start. For each patient, ten crypts were analyzed to assess crypt proliferation. Quantification of proliferating cells in the TA zone was performed using two to five villus-connected crypts. For EEC analysis, a total of twenty crypts were examined.

Crypt isolation, culture and treatment of enteroids

In culture, enteroids grow and form a round shape during the proliferation phase (day 0 to 6/7) followed by a differentiation phase (from day 6 or 7) characterized by the formation of buds. Three-dimensional (3D) enteroids were prepared as previously described in detail by Toulehohoun et al[27]. Briefly, duodenal biopsies were collected in calcium- and magnesium-free phosphate buffered saline (PBS) and washed three times with PBS containing 1% antibiotic-antimycotic [Gibco™ Antibiotic-Antimycotic (100 ×), ref: 15240-062, LIFE Technologies, CA, United States]. Crypts were released from the biopsies and the crypt pellet was resuspended in cold growth factor-reduced Matrigel (Corning^® Matrigel-Basement Membrane Matrix, phenol red free, ref: 356231, VWR, PA, United States) and plated in a 96-well plate. After polymerization at 37 °C for 10 minutes, wells were overlaid with pre-warmed IntestiCult™ organoid growth medium (#06010, STEMCELL^TM Technologies, Canada) supplemented with antibiotics to generate the enteroids. For passaging, Matrigel domes containing human enteroids were dissolved using cell recovery solution and incubated on ice. The collected enteroids were enzymatically dissociated using TrypLE Express [Gibco™ TrypLE™ Express Enzyme (1 ×), phenol red, #12605010, Thermo Fisher Scientific, MA, United States] at 37 °C for 25 minutes and washed. The resulting single cells were resuspended in cold growth factor-reduced Matrigel. Approximately 15000 cells were plated per well in a 96-well plate and incubated to allow the Matrigel to polymerize. Enteroids were cultured in IntestiCult™ human organoid growth medium supplemented with antibiotics. Medium was changed every 2 days until day 6, and daily from day 7 onward.

To culture the two-dimensional (2D) enteroid monolayer, 24 well-plate transwells (0.4 μm) were coated 20 minutes at 37 °C with 200 μL of diluted matrigel (1/50) in cold PBS. Enteroids that were first generated and multiplied in 3D, were dissociated into single cells as described above. The cell pellet was mixed with organoid growth medium. Then, the PBS was removed from the transwells which were thereafter seeded with 300 μL of cell suspension. The bottom well was filled with 300 μL of organoid growth medium.

Enteroids were treated with increasing ethanol concentrations of 40 mmol/L (0.2%) and 70 mmol/L (0.35%) commonly used in the literature[28-30] until day 13 for the differentiation phase. The 2D monolayer was treated for three days after full monolayer confluence. Cell culture medium was changed every day to maintain ethanol exposure constant. At the end of the treatment, enteroids were collected for RNA extraction and immunofluorescence analysis. Immunostaining was performed to identify proliferating cells using Ki67. Ki67⁺ cells were normalized per total nucleus stained with DAPI.

Enteroids immunostaining

3D enteroids were stained as described by Toulehohoun et al[27]. Briefly, enteroids were released from Matrigel using cell recovery solution and incubated on ice for 40 minutes. They were collected and transferred to 24 well-plates with treatment if applicable. Immunofluorescence against Ki67 marker was performed as previously described without antigen retrieval steps. For the 2D culture, the monolayer was washed twice with PBS after discarding the medium. Fixation and staining were performed similar to the 3D culture. Cleared enteroid images (using RapiClear 1.49) were acquired by a Zeiss LSM800 inverted confocal microscope equipped with 4 Lasers (405, 488, 561 and 640 nm), Variable Secondary Dichroics, GaAsp detectors and × 10/NA 0.3 Plan-Apochromat lens.

Statistical analysis

Statistical analysis was performed using Graph Pad Prism 10.4.1. The data is presented as mean ± SD unless otherwise stated. Shapiro-Wilk and Kolmogorov-Smirnov tests were used to assess normality prior to each analysis. T-test was used for comparison between two groups. One-way ANOVA followed by Dunnett’s post-hoc multiple comparisons test was applied for comparison among multiple groups. Data obtained from enteroids were mainly analyzed by paired one-way ANOVA with Greenhouse-Geisser to compare the same subjects across the different conditions. Gene set enrichment analysis was performed on transcriptomic and proteomic data on the Kyoto Encyclopedia of Genes and Genomes and Gene Ontology databases using the multiGSEA Bioconductor package[31].

RESULTS

Study population

Forty-six heavily drinking subjects undergoing elective in-patient alcohol withdrawal were recruited for the study and age, sex, and body mass index matched with fourteen healthy controls. All patients fulfilled the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition criteria for alcohol dependence with a self-reported alcohol consumption exceeding 60 g/day for more than a year. Demographic, biochemical and liver stiffness and controlled attenuation parameters are depicted in Table 1. For evident ethical reasons, only a limited number of duodenal biopsies could be taken per patient. Given the limited amount of material available, not all of the following analyses could be performed in the entire study populations. Care was taken to always include a representative number of patients in each study group.

Table 1 Clinical parameters healthy and alcohol use disorder patients, mean ± SD.

	Healthy	Alcohol use disorder
Demographics	n = 14	n = 46
Age (years)	48.4 ± 15.2	48.7 ± 12
Gender (female/male)	6/8	18/28
Body mass index	24.7 ± 2.5	24.8 ± 4.2
Alcohol (g/day)		182.7 ± 85.6
Biochemistry
AST (U/L)	ND	86 ± 106
ALT (U/L)	ND	63 ± 65
GGT (U/L)	ND	204 ± 318
ALP (U/L)	ND	90 ± 36
Bilirubin (mg/dL)	ND	0.6 ± 0.4
Albumin (g/L)	ND	47 ± 3
INR	ND	0.99 ± 0.09
Creatinine (mg/dL)	ND	0.77 ± 0.14
Fibroscan®
Liver stiffness (kPa)	ND	8.1 ± 6.5
Controlled attenuation parameter (dB/m), grading/staging	ND	276 ± 53
Steatosis (S0/S1/S2/S3)	ND	8/7/16/15
Fibrosis (F0-1/F2/F3/F4)	ND	31/8/4/3

ND: Not done; AST: Aspartate aminotransferase; ALT: Alanine aminotransferase; GGT: Gamma-glutamyl transferase; ALP: Alkaline phosphatase; INR: International ratio.

Morphology alteration

In 46 AUD patients with duodenal biopsies, intestinal morphology was considered normal in 52% (n = 24) while 48% (n = 22) showed an altered morphology (Figure 1A) based on the histological evaluation. Compared to the healthy subjects, crypt length was significantly increased (P = 0.0354) whereas villi length was decreased (P = 0.025) in the pooled group of AUD patients (Supplementary Figure 1). After applying our prespecified criteria, AUD patients with a histologically altered morphology presented reduction in the villi length (P = 0.0003) and elongation of crypts (P = 0.0002) with a crypt to villi ratio that doubled compared to healthy controls (Figure 1B). By contrast, villi and crypt length were similar to controls in AUD patients with a normal appearing morphology validating the initial classification of the patients. These results indicate that heavy alcohol drinking induces morphological changes in the intestinal epithelium in a subset of patients increasing the crypt length/depth and decreasing the villi length. For further analyses the more stringent definition combining histological changes with altered measurements were used to split AUD patients into a normal and an altered subgroup.

Figure 1 Crypt/villus morphology in alcohol use disorder patients. A: Representative immunofluorescence staining of duodenal biopsies showing morphology alterations in a subset of alcohol use disorder (AUD) patients (AUD altered) compared to healthy subjects and AUD patients with normal morphology (AUD normal). Lysozyme in purple stains Paneth cells to identify the crypt base. DAPI (blue) is used as counterstaining to show the nucleus; B: Quantification of the crypt (μm) and villi length (μm). Five to ten crypts/villi were measured per patient. Data are presented as mean ± SD. Crypt and villus length are significantly modified in the AUD altered group resulting in a higher crypt to villi ratio in these patients. The schematic illustration shows the boundaries of the crypt, the transit-amplifying zone, and the villus that defined the limits for the measurements. Healthy (n = 9); AUD normal (n = 10); AUD altered (n = 9). AUD: Alcohol use disorder.

Increased proliferation in crypts in AUD

To investigate the underlying mechanisms of crypt hyperplasia in AUD, we first assessed whether proliferation is altered in those crypts.

Increased proliferation of stem- and stem-like cells at the base of the crypts in AUD: Proliferation at the crypt base is a tightly regulated process through which the stem cells replenish the pool of differentiated epithelial cells. To determine the effect of chronic alcohol on these cells, we counted the proliferating cells stained by Ki67 at base of the crypts also referred to as the CBC in the literature[32]. LYZ, a marker of Paneth cells, was used to identify the CBC. AUD patients exhibited a significantly higher number of Ki67⁺ cells (P = 0.0087) compared to the healthy subjects. After splitting the population into the two prespecified groups, only AUD altered patients had a significantly higher Ki67⁺ cells count (P = 0.0182) in their crypt base (Figure 2A and B). Since Ki67⁺ cells were not found in all CBC, we also assessed the total number of crypts in proliferation per patient. The number of proliferating CBCs (containing at least one Ki67⁺ cell) was similar, indicating that increased cell proliferation is not due to an increased overall number of Ki67⁺ crypts (Supplementary Figure 2). Furthermore, Ki67⁺ cells were double positive for OLFM4, a stem cell marker (Supplementary Figure 2) ruling out the loss of stem cells at the CBC.

Figure 2 Cell proliferation in the crypt base and the transit amplifying zone. A: Immunohistofluorescence staining of duodenal biopsies showing Ki67⁺ proliferating cells (red), lysozyme⁺ Paneth cells (yellow) to identify the crypt base, and DAPI (blue) for the nuclei; B: Ki67⁺ cells were counted in 5-10 crypts per patient. Alcohol use disorder (AUD) patients exhibited a significant increase in proliferating cells within the crypt base. However, when split into 2 subgroups, only AUD-altered patients showed consistently higher proliferation within the crypt base. Healthy (n = 16); AUD normal (n = 18); AUD altered (n = 17); C: Multiplex immunofluorescence for OLFM4 (stem and stem-like cells), Ki67 (red), lysozyme (purple), and DAPI (blue). As indicated in the scheme, the transit amplifying (TA) zone begins at the end of the crypt base and extends to the upper limit of Ki67 staining; D: Ki67⁺ cell numbers in the TA zone were significantly increased in AUD patients, with no significant differences between AUD subgroups. Healthy (n = 11); AUD normal (n = 12); AUD altered (n = 5); E: Quantification of OLFM4⁺ cell distribution along the crypt axis to assess the spatial extension of stem-like cells. OLFM4⁺ distance is significantly increased in AUD altered patients in favor of an extension of stem-like cells into the TA zone. Data are presented as mean ± SD. Healthy (n = 9); AUD normal (n = 9); AUD altered (n = 10). AUD: Alcohol use disorder.

Increased proliferation of stem-like cells in the TA zone in AUD: Ki67 immunostaining also showed increased cell proliferation in the TA zone with the percentage of Ki67⁺ cells being significantly higher (P = 0.0035) in AUD patients (Figure 2C and D). Interestingly, all cells expressing OLFM4⁺ in the TA zone were double positive for Ki67 in AUD patients whereas only a few OLFM4⁺ cells in controls also expressed Ki67 (Figure 2C). In addition, in patients with epithelial damage (AUD altered), the length of the OLFM4⁺ zone was increased compared to controls and AUD normal patients (Figure 2E). Taken together, these observations indicate an extension of proliferating, undifferentiated stem-like cells in the TA zone in AUD patients which is particularly prominent in those with epithelial damage (AUD altered) given their extended lengths of the OLFM4⁺ zone.

Increased number of stem cell-like EECs in the base of the crypts in AUD: It has been reported that all differentiated cells are able to dedifferentiate and replenish the CBCs in case of tissue injury[15,33]. Therefore, we investigated whether the CBCs contained only a pure population of stem cells besides the Paneth cells using multiplex immunostaining to identify Paneth cells (LYZ⁺), EECs (CHGA⁺) and stem cells (OLFM4⁺). Whereas 25% of the crypts in controls contained at least one CHGA⁺ cell (named CHGA⁺ crypts), an average of 50% of crypts stained positive for CHGA⁺ in AUD patients (P = 0.0049) indicating an increased number of crypts containing EECs in AUD (Figure 3A and B). In addition, the number of CHGA cells positive in each crypt was also significantly higher in AUD patients with no differences between the subgroups (Figure 3C). The vast majority of these EECs was double positive for OLFM4 and CHGA (Figure 3A). These results indicate that the stem cell pool identified by the stem cell marker OLFM4 in AUD patients is partially composed of EECs.

Figure 3 Cell composition of the crypts. A: Multiplex immunofluorescence of duodenal biopsy from healthy and alcohol use disorder (AUD) patients. OLFM4 (yellow) marks stem and stem-like cells, CHGA (red) labels enteroendocrine cells, lysozyme (purple) identifies Paneth cells, DAPI (blue) marks nuclei; B: Quantification of the number of crypts containing at least one CHGA⁺ cell. The percentage of CHGA⁺ crypts was increased in both AUD subgroups compared with healthy controls. For each patient, 5-10 crypts were analyzed; C: Quantification of CHGA⁺ in each crypt. The number of CHGA⁺ cells per crypt was significantly higher in both AUD normal and AUD altered patients. Data are presented as mean ± SD. Each dot on the bar graphs represents one patient. Scale bar (200 μm). Healthy (n = 6); AUD normal (n = 6); AUD altered (n = 6). AUD: Alcohol use disorder.

Dysregulation of the Wnt pathway in AUD

To screen for factors/pathways that might be dysregulated between healthy subjects and AUD patients we re-analyzed our bulk RNA sequencing and additionally performed proteomics of duodenal biopsies. Details on differentially expressed genes have been published previously[19]. In addition, principal component analysis (PCA) was employed to assess the global patterns of variation between healthy individuals and AUD patients. PCA of transcriptomic data revealed a certain degree of separation between the groups, with healthy individuals forming a relatively tight cluster. In contrast, the AUD group exhibited considerable heterogeneity, resulting in a broader distribution that partially overlapped with the healthy control cluster (Figure 4A). Compared to the transcriptomic data, a substantially higher number of proteins (597) were differentially regulated (adjusted P < 0.05): 279 proteins were downregulated and 318 proteins upregulated. PCA demonstrated a clear separation between healthy and AUD subjects, indicating more pronounced divergence at the protein level (Figure 4B).

Figure 4 Dysregulation of the Wnt pathway in alcohol use disorder through CD44. A-D: Principal component analysis of bulk transcriptomic and proteomic data. At transcriptional level, a subset of alcohol use disorder (AUD) patients clusters closely with healthy subjects (A), whereas proteomic data (B) reveal a clear separation between both groups. Gene ontology dysregulation of the canonical Wnt signaling pathway on the transcriptomic (C) and protein (D) level. Transcriptomics, controls/healthy (n = 15); AUD (n = 43). Proteomics, controls/healthy (n = 7); AUD (n = 15); E: Multiplex immunofluorescence staining of duodenal biopsies using E-cadherin (green) for epithelium, Ki67 (red) for proliferating cells, CD44 (purple) as a Wnt/β-catenin target gene, lysozyme (yellow) for Paneth cells, and DAPI (blue) for nuclei; F: Quantification of the number of CD44⁺ cells at the crypt base. Overall CD44 expression increased in AUD patients, with proliferating, CD44⁺/Ki67⁺ double-positive cells, being significantly elevated in both AUD subgroups; G: Quantification of CD44⁺ cells in the transit amplifying zone. Only a very low number of cells (less than 10%) stained positive for CD44 with numbers tending to be slightly higher in AUD patients with an altered morphology (AUD altered). Data are presented as mean ± SD. Scale bar (200 μm). Healthy (n = 8); AUD normal (n = 8); AUD altered (n = 7). AUD: Alcohol use disorder.

Wnt canonical pathway: To determine the underlying mechanism of crypt base cell proliferation, the Wnt canonical pathway, a key regulator of stem cell proliferation at the CBC[34,35] was investigated. RNA sequencing and Bulk proteomic data pointed to a dysregulation of gene ontologies related to the canonical Wnt pathway in AUD (Figure 4C and D). Data in the literature indicates that CD44 is required for optimal Wnt signaling during intestinal epithelium repair[36]. Therefore, we assessed CD44 by immunofluorescence in our duodenal biopsies. Immunofluorescence staining localized CD44 in the lamina propria as well as the epithelium (Figure 4E). At the crypt base, the total number of cells expressing CD44 was increased in AUD including stem cells and a subset of Paneth cells (Figure 4E and F). However, Paneth cells did not express the proliferation marker Ki67 (Figure 4E). By contrast, the percentage of CD44 positive stem cells that also express Ki67 was significantly higher (P = 0.0006) in AUD patients compared to controls (9.7% vs 34.4%).

Wnt signaling pathway also plays a role in maintaining proliferation of transit amplifying progenitor cells derived from the stem cells[32]. Therefore, we also quantified CD44 positive cells in the TA zone of crypts connected to villi and normalized them to the total number of cells per TA zone. Although the number of CD44⁺ cells tended to be lower in controls, only less than 10% of all TA zone cells were positive for CD44 in AUD patients. While the proportion of CD44⁺ cells was higher in AUD altered patients compared to controls (9.07% vs 5.35%), the difference did not reach statistically significance (Figure 4E and G). In addition, only half of the CD44 positive cells co-expressed the proliferation marker Ki67 further indicating that a very low number of transit amplifying proliferating cells (< 5%) in the TA zone is associated with induction of Wnt/β-catenin signaling (Supplementary Figure 3).

Axin2 is a target gene of Wnt pathway that functions as a negative feedback regulator of Wnt pathway[37-39]. Axin2 was mainly restricted to the epithelium in the TA zone adjacent to the crypts whereas no cells at the crypt base expressed Axin2 (Figure 5A). Quantification of Axin2 immunofluorescence along the crypt-villus axis revealed no difference in AUD patients compared to controls (Figure 5B). Axin2⁺ cells co-expressed OLFM4 but a subset of Axin2⁺/OLFM4⁺ cells did not express Ki67 compatible with negative feedback on proliferation in this cell population (Figure 5A).

Figure 5 Wnt target gene Axin2 is not implicated in hyperproliferation in alcohol use disorder. A: Multiplex immunofluorescence staining of duodenal biopsies showing Axin2 (green), a Wnt target gene; Ki67 (red) for proliferating cells; OLFM4 (yellow) for stem and stem-like cells; and DAPI (blue) for nuclei; B: Quantification of Axin2 fluorescence intensity, normalized to the total crypt area and expressed as a percentage showing a non-significant trend towards decreased Axin2 fluorescence intensity in particularly in alcohol use disorder (AUD)-altered patients. Data are presented as mean ± SD. Scale bar (200 μm). Healthy (n = 5); AUD normal (n = 6); AUD altered (n = 5). AUD: Alcohol use disorder.

Wnt non-canonical pathways: To assess the non-canonical Wnt pathways, we examined the expression of their principal down-stream targets phospho c-Jun and NFATC1. Phosho c-Jun levels were upregulated in AUD (Figure 6A). However, phospho c-Jun expression was restricted to the tips of the villi while no expression was detected in the TA zone (Figure 6B). NFATC1 was detected in the lamina propria and at the base of the crypts (Figure 6C). In the epithelium, NFACT1 co-localized with LYZ indicating expression in Paneth cells at the crypt base (Figure 6C). These results indicate that non-canonical Wnt pathways are not directly involved in regulation of cell proliferation in the crypts.

Figure 6 Wnt non-canonical pathways are not involved in hyperproliferation of the crypts in alcohol use disorder. A: Schematic representation of the non-canonical Wnt/PCP pathway. Binding of Wnt ligands to frizzled receptors leads to downstream c-Jun phosphorylation and gene transactivation; B: Immunohistofluorescence showing phospho-c-Jun (Ser63) (red) localized at the villus tip. Quantification of pc-Jun (Ser63)⁺ cells normalized to total villus cells indicating increased phospho c-Jun expression in alcohol use disorder patients; C: Schematic of the Wnt/Ca²⁺ pathway, in which activation triggers NFAT dephosphorylation and nuclear translocation. Immunostaining of NFATC1+ (red) shows co-localization with Paneth cells (purple) stained by lysozyme within the epithelium marked by E-cadherin (green). Scale bar (200 μm). Healthy (n = 5); alcohol use disorder (n = 12). AUD: Alcohol use disorder; LYZ: Lysozyme.

TA zone hyperproliferation is independent of the epidermal growth factor receptor-mitogen-activated protein kinase pathway

The epidermal growth factor receptor (EGFR) pathway has been described as a regulator of TA zone proliferation and the transition from proliferation to differentiation[40]. Bulk proteomic and transcriptomic analysis provided a weak signal of dysregulation of EGFR signaling. To more precisely investigate the EGFR pathway, we performed immunolocalization of the active, phosphorylated form of EGFR. Phospho EGFR (Y1068) was primarily expressed in the epithelium in the villi but also in the TA zone (Figure 7A). Quantification of pEGFR positive cells in the TA zone did not show any difference between AUD patients and healthy controls (Figure 7B). Since the EGFR pathway signals down-stream through the mitogen-activated protein kinase (MAPK) pathways in cell proliferation[41], we investigated pERK expression in duodenal epithelial cells. Immunofluorescence localization of pERK shows expression in the lamina propria of the villi whereas expression in epithelial cells in the TA zone was absent (Figure 7C). Taken together, those results indicate that the EGFR-MAPK pathway is not a driver of hyperproliferation in the TA zone.

Figure 7 Epidermal growth factor receptor-mitogen-activated protein kinase signaling in duodenal biopsies. A: Immuhistofluorescence staining of pEGFR (red) which located in the crypt and villus in alcohol use disorder (AUD) and healthy subjects; B: Quantification of pEGFR⁺ cells in the transit-amplifying zone, normalized to the total number of transit-amplifying-zone cells, shows no significant difference between AUD patients and healthy subjects; C: Immunofluorescence staining of pERK, a downstream effector of the epidermal growth factor receptor-mitogen-activated protein kinase signaling pathway, reveals predominant expression in the lamina propria with only a very few pERK⁺ cells detected within the crypts. Scale bar (200 μm). Healthy (n = 3); AUD (n = 8). AUD: Alcohol use disorder; LYZ: Lysozyme; TA: Transit-amplifying.

A lineage shift towards the secretory type in AUD

To further determine how alcohol affects the proliferation and differentiation balance, we investigated pathways involved in the differentiation process. Bulk RNAseq and proteomics of duodenal biopsies showed an upregulation of gene ontologies implicated in epithelial cell differentiation both at transcriptional and translational levels (Figure 8A).

Figure 8 Transcriptomics and proteomics of factors associated with epithelium differentiation and function. A: Gene Ontology enrichment analysis showing dysregulation of pathways related to epithelial differentiation and intestinal epithelial cell differentiation in alcohol use disorder (AUD) patients; B: Proteomic data show that AUD patients exhibited significantly increased levels of the mucins MUC2 and MUC5AC, as well as AGR2, which supports MUC2 biosynthesis (orange boxes). Proteins involved in antimicrobial immune defense and barrier protection were impaired, with decreased production of Paneth cell-derived defensins DEFA5, DEFA6, and REG3A (green boxes). Dysregulation of absorption was reflected by reduced MTTP and FABP2 protein levels and increased expression of the amino acid transporter SLC1A5 (blue arrows). Finally, transcriptomics also reveals upregulation of genes implicated in terminal differentiation of goblet cells, including SPDEF and Kruppel-like factor 4 (KLF4) (red boxes). Transcriptomics, healthy (n = 15); AUD (n = 43). Proteomics, healthy (n = 7); AUD (n = 15); C: Immunofluorescence staining showing MUC2⁺ goblet cells (green), lysozyme⁺ Paneth cells (purple), and DAPI (blue). Quantification of MUC2⁺ cells reveals an increased number of goblet cells in both crypts and villi of AUD patients. Healthy (n = 6); AUD (n = 13); D: Immunofluorescence staining and quantification of Ki67 (green) for proliferating cells and KLF4 (red), a transcription factor involved in terminal differentiation of goblet cells. DAPI marks nuclei (blue). A significant increased number of Ki67⁺/KLF4⁺ double positive cells per crypt in AUD indicating a preferential commitment of cells towards the secretory goblet cells lineage. Scale bar (200 μm). Healthy (n = 4); AUD (n = 8). AUD: Alcohol use disorder; KLF4: Kruppel-like factor 4.

Proteomics data also showed that the Notch signaling pathway, implicated in driving the differentiation of progenitor cells towards enterocytes, was mainly downregulated (Supplementary Figure 4). By contrast, key transcription factors essential to terminal differentiation of goblet cells, SPDEF and KLF4, were increased in transcriptomics in AUD (Figure 8B) which was further confirmed by quantitative polymerase chain reaction (Supplementary Figure 4; Figure 8C). Subsequently, MUC2 staining confirmed that the total number of goblet cells along the crypt-villus axis was significantly higher (P = 0.0001) in AUD patients compared with healthy subjects (Figure 8C) as previously observed by our group in a separate cohort of patients[19]. In addition, the number of proliferating cells in the patients’ crypts committed to differentiate into goblet cells by co-expressing Ki67/KLF4 was significantly increased in AUD (Figure 8D). Together, these data support a shift in lineage commitment towards the secretory type in AUD patients primarily favoring the goblet cells pathway.

Epithelium function impairment in AUD

To link the morphological findings to functional roles of the epithelium, we focused on factors related to barrier function, antimicrobial defense and absorption in our proteomics analysis. Our data indicate an increased protein expression of several Mucins involved in epithelial barrier function. MUC2 protein levels were strongly and significantly elevated in AUD patients compared to healthy controls. This finding was supported by a significant increase of the anterior gradient-2 (AGR2), a key protein involved in the biosynthesis of Muc2 (Figure 8B). Strikingly, Muc5ac protein expression, a mucin known to be primarily restricted to the stomach and the respiratory tract, was also strongly up-regulated. These results were consistent with the upregulation of their corresponding gene entities in transcriptomics (Figure 8B). Together, these findings indicate a dysregulation of the function of goblet cells in the context of AUD.

Paneth cells are involved in the barrier defense against bacteria and viruses by producing several factors essential to the epithelium homeostasis. Although LYZ protein levels were increased, we found decreased protein levels of the defensins DEFA5, DEFA6 as well as REG3A (Figure 8B) all secreted into the mucus and involved in anti-bacterial defense. Absorption alteration mainly translated into a decrease of MTTP and FABP2 protein levels as well as dysregulation of several transporters such as, for example SLC1A5, involved in absorption of amino acids (Figure 8B).

AUD enteroids keep morphology alteration in vitro

To establish a potential cause-effect relationship between alcohol and the epithelium alterations observed in vivo in humans, we used human-derived intestinal organoids (enteroids). Enteroids grown from AUD patients presented a very distinctive morphology compared to controls either presenting no buds (representative of crypts in vivo) or a high number of buds thus mimicking the morphological phenotype observed in vivo (Figure 9A). After nine days in culture, mRNA level of the stem cell marker LGR5, the goblet cell marker MUC2, and the Paneth cell marker LYZ were quantified. Muc2 gene expression was upregulated whereas LGR5 and LYZ genes remained unchanged (Figure 9B).

Figure 9 Generation and alcohol stimulation of intestinal enteroids from healthy subjects and alcohol use disorder patients. A: Alcohol use disorder (AUD)-derived enteroids retained the morphological impairments observed in their host. A subset of AUD enteroids lacked budding structures, whereas another subset showed excessive budding. Images were acquired by optical microscopy at 10 × magnification; B: Gene expression analysis showing unchanged levels of LGR5 (stem cells) and lysozyme (Paneth cells) in AUD-derived enteroids, with a significant upregulation of MUC2, indicating increased goblet cell differentiation. Healthy (n = 3); AUD (n = 8); C: Experimental design of three-dimensional enteroids cultured in Matrigel and treated with 40 mmol/L or 70 mmol/L ethanol from day 2 after passage until full differentiation at day 13. Ethanol-containing medium was refreshed daily; D: Confocal microscopy of fixed and stained three-dimensional enteroids shows no significant differences in Ki67⁺ proliferating cells (green) across ethanol treatments (black: 70 mmol/L; gray: 40 mmol/L) as well as in the expression of cyclinB1 and Wnt target genes CD44 and Axin2; E: Experimental design of two-dimensional enteroids treated for three days after full monolayer confluence two-dimensional enteroids exposed to 70 mmol/L ethanol exhibited significantly increased proliferation; F: Further supported by upregulation of cyclin B1 mRNA. Expression of Wnt target genes (CD44, Axin2) was not changed upon ethanol exposure. Scale bar (50 μm). Healthy (n = 3). AUD: Alcohol use disorder; 3D: Three-dimensional; 2D: Two-dimensional.

Alcohol only weakly stimulates proliferation in enteroids derived from healthy patients

To mimic the initial alcohol insult, we exposed enteroids derived from healthy subjects to prolonged alcohol stimulation. In the 3D model (Figure 9C), quantification of Ki67 positive nuclei revealed no significant difference in proliferation at 40 mmol/L and 70 mmol/L of alcohol compared to no alcohol exposure (Figure 9C). This was further confirmed by similar cyclin B1 mRNA levels, an additional proliferation marker. Wnt β-catenin target gene expression Axin2 was significantly down-regulated whereas CD44 mRNA expression did not significantly change (Figure 9D). LGR5 and Muc2 mRNA expression was inconsistent (Supplementary Figure 5).

In 3D culture, the apical face, usually exposed to nutrients but also toxins in vivo, is located/hidden inside the enteroid. This might influence the impact of any treatment that is not directly in contact with the apical side, which is the case for alcohol in vivo. To better mimic this condition, we performed a 2D culture (Figure 9E) in transwells of the enteroids until full monolayer confluence which assures homogenous alcohol exposure also to the apical side[42]. In contrast to the 3D model, 70 mmol/L of ethanol increased Ki67 expression after already a relatively short time of stimulation (Figure 9E) which was mirrored by a significant increase in cyclin B1 mRNA levels (Figure 9F). CD44, Axin2, LGR5 and MUC2 mRNA levels did not significantly change after 3 days of ethanol exposure (Figure 9F; Supplementary Figure 5).

DISCUSSION

In this study, we show morphological and molecular changes in the duodenal epithelium of chronic heavy drinkers with deleterious effects on the intestinal villi-crypt architecture as well as the stem cell population. We further demonstrate an alteration of the intestinal proliferation-differentiation balance leading to a shift in the differentiation program towards a specific differentiated cell type, in particular the goblet cells. AUD patients displayed higher crypt length/depth associated with shorter villi supported by a higher crypt/villi ratio consistent with features of small intestinal enteropathy[43,44]. These findings are in line with recent studies in mice demonstrating an increased crypt depth in the chronic alcohol plus binge model[18] or reduction in villi length after alcohol binges[45]. Other studies also reported villi shrinkage and injury in AUD[12,46]. Interestingly, AUD patients exhibited an extension of the OLFM4⁺ zone, suggesting a sustained progenitor state and delayed differentiation, which may ultimately lead to a shortened villus. The increased length of the OLFM4⁺ zone may also reflect more abundant and active proliferation to rescue tissue damage and regenerate the intestinal epithelium in order to replenish the differentiated cells pool.

Intestinal CBC stem cells are the gate keeper for the rapid cell renewal in homeostatic conditions as well as in tissue injury[15]. Several studies reported the sensitivity of these cells when exposed to high amounts of alcohol[14,47]. Our data on duodenal biopsies showed an increased cell proliferation in the CBC in AUD patients associated with an upregulation of CD44 which links the crypt hyperproliferation with the canonical Wnt pathway. CD44 has been described as the stimulator of the Wnt pathway in regeneration[36] and has also been suggested as a potential stem cell marker[48].

In contrast, what stimulates hyperproliferation in and expansion of the TA zone remains unclear. Our observation that CD44⁺ cells constitute only a minor fraction of the progenitor population makes it unlikely that CD44 driven Wnt activation contributes to TA zone hyperproliferation. We did observe the presence of the phosphorylated form of the EGFR in the TA zone. By contrast, our observations did not reveal the presence of pERK in the TA zone providing an argument against downstream activation of the EGF-MAPK pathway as a contributor to TA zone cell proliferation. Additional evidence through modulation of the Wnt and EGF-EGFR pathways, for example in ex vivo systems, is required to formally rule out a role of those pathways in TA zone expansion in AUD. In addition, bone morphogenetic protein pathway was not investigated, which might also be involved in regulation of cell proliferation in the intestine[49].

Another intriguing finding was the presence of only one cell type next to Paneth cells. Those cells were EECs either double positive for OLFM4 or pure EECs positive for CHGA. It has been demonstrated that differentiated intestinal cells as enterocytes, Paneth cells, secretory lineage cells, and tuft cells are able to replenish the stem cell pool. This reprogramming of differentiated cells into stem cell state also called dedifferentiation is a well-regulated process[33]. Co-expression of CHGA and the stem cell marker OLFM4 as well as spatial redistribution indicate that EECs may rescue stem-like cells. These findings corroborate with a previous study combining fluorescence-activated cell sorting and single cell mRNA sequencing to explore the inter-relationship between putative mouse ISC populations. This study suggested mature EECs as a reservoir for ISCs[50]. However, our observations do not directly demonstrate lineage conversion or stem-like functionality. Single cell RNA sequencing of human samples might be useful to further address those issues by exploring pathways and markers related to stem cell conversion and functionality. Furthermore, chronic alcohol feeding and acute exposure of alcohol resulted in ISC dysregulation in mice indicating their particular sensitivity to an alcohol insult[17]. Overall, this is consistent with the loss of stemness in the intestinal epithelium under the influence of alcohol.

In addition to alteration in the proliferation program in AUD, we also confirmed hyperplasia of goblet cells[19] associated with upregulation of SPDEF and KLF4 involved in terminal differentiation of goblet cells. The increased commitment of cells in the TA zone towards differentiation into goblet cells indicates a preferential shift of the differentiation program towards the secretory lineage in AUD. This might be interpreted as a tentative to reinforce the intestinal barrier since the intestinal mucus layer secreted by goblet cells is the first line to fight against bacterial and derived products in the gut[51]. It has been demonstrated that bacterial overgrowth and microbial translocation is a signature of AUD. The MUC2 promoter has a binding site for lipopolysaccharide[52] which potentially links the overexpression of MUC2 observed in AUD patients to dysbiosis together with upregulation of AGR2, essential for driving MUC2 production. Remarkably, MUC5AC, the predominant mucin in the stomach, is overexpressed in the duodenum of AUD patients. MU5AC increase has also been reported in inflammatory bowel disease. In fact, inflammatory bowel disease patients exhibited a progressive replacement of MUC2 by MU5AC and MUC6[53]. Instead of being protective, these observations rather indicate a dysregulated mucus production by goblet cells in AUD, which is associated with less antigen presentation through goblet cells associated antigen passages[19]. Additional evidence for a defective physical intestinal barrier comes from our data showing decreased levels of defensin 5 and 6 as well as REG3A, anti-bacterial proteins secreted into the mucin layer by Paneth cells. Alcohol-exposed mice displayed similar Paneth cells dysfunction characterized by altered α-defensins (defensins 2, 4, 5 20, and 21) activity and deficiencies in REG3B and REG3G, leading to increased bacterial translocation[54,55].

The intestinal villi mostly composed of enterocytes measure 0.5 mm to 1.6 mm in length[56]. Enterocytes lining the villi are responsible of nutrients absorption. Our data showed a reduction of the villi in a subset of AUD patients which also translated into dysregulation of several transporter proteins. The reshaping of the villi might reflect an impairment of intestinal absorption processes as reported in several other studies[57]. Future studies are needed to evaluate how alcohol-induced changes in the differentiation process directly affect functional absorptive capacity of the intestine.

We used human-derived enteroids to investigate the effects of alcohol on the intestinal epithelium. These 3D organoid cultures retained the key architectural phenotype of their host, and we could also identify our specific differentiated cell types of interest with a shift towards MUC2 overexpression. However, alcohol did not reproduce hyperproliferation or dysregulation of the Wnt pathway in human 3D organoids. This is in contrary to mouse-derived organoids where alcohol stimulated proliferation through upregulation of the beta catenin target genes Tcf4 and cyclin D1[18]. However, alcohol stimulated cell proliferation in the enteroid 2D monolayer model without modulating Wnt/β-catenin target genes CD44 and AXIN2. Since ethanol did not increase cell proliferation in the 3D model, it might be possible that the contact of ethanol with the basal face of the organoid in the 3D model may not induce the same effect compared to contact with the apical face in 2D enteroids. This is supported by recent data in mice where successive alcohol binges with high luminal alcohol concentration in the proximal small gut promoted duodenal alterations rather than alcohol diffusing from the circulation via the basal cellular pole of the duodenal epithelium[58]. As we could not establish a direct cause-effect relationship between alcohol treatment and the Wnt pathway in the human-derived enteroids as shown with mouse derived enteroids, different pathways might be involved in alcohol-induced epithelium damage in humans. Additional studies are required using, for example, single cell RNA sequencing to unravel those mechanisms. Although we cannot formally rule out a contribution of alcohol to cell proliferation in enteroids, alcohol exposure alone does incompletely recapitulate phenotypes with substantial variations. Alternatively, the effect on the epithelium might be mediated by its metabolites like acetaldehyde or a combination of both.

The study presents some limitations. First, the results primarily reflect the intestinal status during the early stage of withdrawal and cannot be generalized to later stages of AUD recovery, where intestinal physiology and host-microbiota interactions may differ substantially. Examining the dynamics of the changes in the gut upon abstinence would be interesting. However, it is difficult to ethically justify an additional invasive procedure like a gastroscopy at later time points in abstinent patients purely for research purposes. Second, although healthy volunteers did not take any medication or present any clinically relevant co-morbidities, the lack of biochemical and liver-related parameters in the control group may limit comparability of systemic metabolic or hepatic indicators between AUD patients and controls. Additionally, results including proteomics and transcriptomics are primarily associations that do not formally prove causality which requires additional functional validation.

CONCLUSION

In conclusion, our findings demonstrate profound alterations of the physical barrier comprised by the intestinal epithelium and the mucus layer in AUD patients. These modifications extend far beyond the classically reported increased paracellular permeability and affect intestinal morphology, the proliferation-differentiation balance, as well as several principal epithelial cell functions. Many of alcohol-induced changes in the gut are independent of ALD severity and occur before the progression of liver disease. Our findings corroborate and indirectly prove the hypothesis on the existence of alcohol-induced intestinal disease, also referred to as alcohol-associated bowel disease (ABD)[59]. Future translational and multi-centric studies are warranted to establish a consensus definition of the different stages of ABD. In addition, those studies should include approaches that aim at restoring a functional intestinal barrier in AUD. They should not only focus on anti-inflammation, but also consider how to restore normal stem cell dynamics and balance of differentiation. Targeting CD44 or related signaling pathways might be one option to restore normal stem cell physiology. Since animal models only partially reproduce the alcohol-induced changes observed in humans, alternative models such as human-derived enteroids are likely inevitable to advance our knowledge in the field, which could enable the identification of novel therapeutic strategies, not only for the ABD per se, but also for brain and liver disorders influenced by gut dysfunction.

ACKNOWLEDGEMENTS

We thank Aurélie Daumerie of the 2IP platform of Institut de Recherche Expérimentale et Clinique for technical assistance. We also thank the CTMA platform of Institut de Recherche Expérimentale et Clinique for the Omics data analysis support.