Dysregulated signal transducer and activator of transcription 3 drives intestinal permeability and may contribute to acute-on-chronic liver failure

Abstract

BACKGROUND

Enhanced intestinal epithelial permeability contributes to disease progression in advanced liver disease. Patients with acute-on-chronic liver failure (ACLF) frequently exhibit gut barrier dysfunction. However, mechanisms underlying the transition from cirrhosis to ACLF remain incompletely understood. Signal transducer and activator of transcription 3 (STAT3) is a key regulator of epithelial homeostasis, yet its role in intestinal barrier integrity during ACLF has not been explored.

AIM

To investigate whether imbalanced intestinal STAT3 expression drives gut barrier dysfunction during liver disease advancement, particularly during progression to ACLF, we analyzed its effects on tight junction architecture, lipid composition, and epithelial morphology.

METHODS

Morphology of intestinal tissues from mouse models of steatosis, cirrhosis, and ACLF was analyzed by hematoxylin and eosin staining while STAT3 expression and activation were determined by immunohistochemistry. Messenger RNA levels of Stat3, Stat1, and tight junction components [zonula occludens-1 (Tjp1), occludin (Ocln), claudin (Cldn) 1, Cldn2, and Cldn3] from intestinal mouse tissues were analyzed by reverse transcription-quantitative polymerase chain reaction. Additionally, stably differential STAT3-expressing epithelial Caco-2 or T84 monolayers were generated to assess barrier integrity using transepithelial electrical resistance measurements. Protein expression of STAT3, STAT1, zonula occludens-1, OCLN, claudin (CLDN) 1, CLDN2, and CLDN3 was determined by western blotting. Lipid composition was analyzed by lipidomics, and epithelial morphology was assessed through immunofluorescence and electron microscopy.

RESULTS

Liver disease progression was associated with pronounced intestinal morphological alterations accompanied by a marked increase in STAT3 expression during the transition from cirrhosis to ACLF. In vitro, elevated STAT3 levels induced tight junction remodeling, increased epithelial permeability, and significantly altered lipid composition. Lipidomic profiling revealed remodeling of major phospholipid, ether lipid, sphingolipid, lysophospholipid, and triglyceride classes, indicating disturbed epithelial membrane lipid homeostasis. These effects were paralleled by pronounced STAT3-dependent structural changes in epithelial monolayers.

CONCLUSION

Accumulation of unphosphorylated STAT3 may serve as a potential biomarker of gut barrier destabilization during the transition from cirrhosis to ACLF.

Key Words: Acute-on-chronic liver failure; Cirrhosis; Gut barrier dysfunction; Signal transducer and activator of transcription 3; Tight junctions

Core Tip: Intestinal barrier disfunction is frequently observed in acute-on-chronic liver failure (ACLF) secondary to chronic liver disease, yet the underlying molecular mechanisms remain largely unknown. This study identifies a disrupted expression pattern of the intestinal signal transducer and activator of transcription 3 (STAT3) as a central regulator of epithelial barrier integrity during the course of liver disease. Dysregulated STAT3 significantly impairs tight junction architecture and epithelial morphology, leading to increased intestinal permeability. The accumulation of unphosphorylated STAT3, independent of canonical STAT3 phosphorylation, may represent a potential biomarker linking cirrhosis to ACLF and could serve as a novel therapeutic target.

INTRODUCTION

Acute-on-chronic liver failure (ACLF) is a syndrome in patients with liver cirrhosis[1] characterized by acute liver decompensation[2] followed by multi organ failure and high short-term mortality rates[3]. The underlying molecular pathways leading to ACLF, however, are not fully understood. In the Western world, bacterial infections and increased alcohol consumption have been identified as the major trigger for ACLF[4]. So far, there are no specific treatment options apart from the possibility of liver transplantation in selected patients[5] or viral suppression therapy when ACLF is precipitated by viral infection, such as hepatitis B[6]. Currently, the most important forms of treatment for ACLF include early administration of a broad-spectrum antibiotic and symptomatic treatment of various organ failures in specialized intensive care units. There is an urgent need to improve treatment options and to understand the clinical picture in more detail.

ACLF progression is associated with a massive increase of systemic inflammation and many of the patients generate a leaky gut[7]. Enhanced intestinal permeability enables the translocation of bacteria or pro-inflammatory signals from the intestinal lumen, which, in turn, exacerbates disease progression and leads to an uncontrolled immune response in patients[8]. Three independent but interlinked layers prevent the transfer of harmful microorganisms, antigens, and toxins from the intestinal lumen into the bloodstream. Together, they form a physical barrier against bacterial intrusion from the gut lumen: The luminal mucus layer, the gut epithelial layer [formed by a continuous sheet of intestinal epithelial cells (IECs)], and the internal layer that forms the mucosal immune system[9]. The layer of IECs is held together by tight junctions and accompanied with mucin secreting goblet cells, whereas the mucus layer is primarily composed of mucin-2. The regulation of the intestinal permeability runs via signaling proteins whereas the Janus kinase-signal transducer and activator of transcription (STAT) pathway is involved in mediating the receptor signaling of numerous cytokines[10].

One major STAT protein, STAT3, plays a key role in safeguarding the intestinal epithelial barrier. It acts by modulating protein expression in response to external signals like microbes and cytokines. Through this, STAT3 supports essential biological processes including controlling inflammation, cell growth, programmed cell death, and the capacity for self-renewal[11,12]. This indicates that epithelial STAT3 regulates gut homeostasis. Accordingly, serious consequences for the intestinal epithelium have been described when STAT3 expression or activation is dysregulated. On the one hand, it has been demonstrated that development of colonic inflammation is associated with enhanced STAT3 activity in IECs[13], and that persistent activation of STAT3 is frequently observed in the IECs of human patients with ulcerative colitis and Crohn’s disease[14]. On the other hand, conditional knockout mice with an IEC-specific deletion of STAT3 activity have been described to be highly susceptible to experimental colitis and showed a striking defect of epithelial restitution. These observations are consistent with the role of STAT3 as a nuclear transcription factor that directly regulates the expression of epithelial tight junction genes, thereby shaping tight junction composition and barrier integrity through transcriptional control of core junctional components such as zonula occludens-1 (Tjp1) and occludin (Ocln)[15]. Functionally, the zonula occludens-1 (ZO-1) protein provides a cytoplasmic backbone linking junctional complexes to the actin cytoskeleton, while sealing proteins, such as claudins, including claudin (CLDN) 1, contribute to barrier integrity[16,17]. In contrast, CLDN2 forms paracellular pores and is associated with increased epithelial permeability[18].

Emerging evidence suggests that dysregulated epithelial STAT3 signaling, rather than merely altered STAT3 abundance, may critically disturb tight junction composition and barrier stability under inflammatory conditions. However, its role in chronic liver disease progression and the development of ACLF remains insufficiently defined.

Given that epithelial STAT3 is rapidly induced by cytokine signaling and mediates epithelial restitution and barrier function, and that cirrhosis/ACLF are accompanied by gut dysbiosis, increased intestinal permeability, and systemic inflammation, we hypothesized that intestinal STAT3 expression and activation might be altered during chronic liver disease progression and may significantly contribute to impaired barrier integrity.

To test this, we analyzed epithelial morphology, STAT3 protein levels, and the amount of activated STAT3 [phospho-STAT3 (pSTAT3)] in intestinal tissue from mice with steatosis, cirrhosis, or ACLF. In parallel, we performed reverse transcription-quantitative polymerase chain reaction (RT-qPCR) to assess the messenger RNA (mRNA) levels of Stat3, Stat1, and barrier-related genes such as Tjp1, Ocln, Cldn1, Cldn2, and Cldn3.

To directly examine the role of STAT3 in epithelial function, we additionally generated Caco-2 and T84 cells with stably enhanced or reduced STAT3 expression, evaluated barrier properties, determined protein levels of the already via RT-qPCR in vivo analyzed enzymes, and analyzed the lipid compositions as well as epithelial morphologies of corresponding cell monolayers.

MATERIALS AND METHODS

Animal models

For the induction of different stages of liver disease in mice, specific pathogen-free male wildtype (C57BL/6J) mice (10 weeks old; Charles River, Sulzfeld, Germany) were housed under standard conditions (22 ± 0.5 °C, 50% humidity, 12:12-hour light/dark cycle) in individually ventilated cages with ad libitum access to food and water. All experiments were conducted in accordance with the animal welfare guidelines and approved by the Regierungspräsidium Darmstadt, which is the responsible authority for animal studies in the federal state of Hessen, Germany (permit number FK/2003).

Liver steatosis (hereinafter referred to as steatosis) was induced by administration of ethanol in the drinking water (4% during week 1, 8% during week 2, and 16% thereafter until sacrifice) together with a normal chow diet (Ssniff, Soest, Germany). Liver fibrosis (hereinafter referred to as cirrhosis) was induced by intraperitoneal injection of carbon tetrachloride (CCl₄) (2 μL/g body weight, CCl₄: Corn oil = 1:2) twice per week for seven weeks to mimic alcoholic steatohepatitis as described before[19-22].

ACLF was precipitated based on established chronic-plus-binge ethanol models[23-27]. In brief, ACLF was generated by a single ethanol binge (100%, 6 g/kg, gavage) on top of seven weeks of ethanol feeding and CCl₄ injections. Mice were sacrificed 72 hours after ACLF induction.

Control groups consisted of untreated, age-matched mice. Prior to euthanasia, mice were anesthetized with an intraperitoneal injection of ketamine (100 mg/kg body weight) and xylazine (10 mg/kg body weight). Intestinal samples were either rapidly snap-frozen and stored at -80 °C, fixed in 4% formaldehyde followed by paraffin embedding, or embedded in Tissue-Tek OCT compound (Sakura Finetek, Staufen, Germany) after fixation, depending on the downstream analyses.

RT-qPCR

mRNA levels in gut tissue of the mouse models were determined by RT-qPCR. Total RNA was extracted from the cells using TRIzol reagent (Invitrogen; Thermo Fisher Scientific, Inc.) according to the manufacturer’s protocol. First-strand complementary DNA (cDNA) was prepared from 1 μg RNA with 0.5 μg/μL random hexamer primers using ImProm-II Reverse Transcription System (Promega Corporation) according to the manufacturer’s protocol. RT-qPCR was performed using GoTaq^® Probe qPCR Master Mix assay (Promega Corporation) for Stat1 (Mm01257286 m1), Stat3 (Mm01219775 m1), Ocln (Mm00500910 m1), Cldn1 (Mm00516701 m1), Cldn2 (Mm00516703 s1), Cldn3 (Mm00515499 s1), Tjp1 (Mm00493699 m1). 18s (Taqman Pre-Developed Assay Reagents Euk 18s rRNA, Applied Biosystems; Thermo Fisher Scientific) was used as the housekeeping gene. RT-qPCR reactions included 5 μL TaqMan Gene Expression Mastermix, 0.5 μL TaqMan assay, RNase-free water and 1 μL cDNA (100 ng) in a total volume of 10 μL. The thermocycling conditions were as follows: 50 °C for 2 minutes, 95 °C for 10 minutes; followed by 50 cycles of 95 °C for 15 seconds and 60 °C for 1 minute, in a StepOnePlus^TM real-time PCR system (Applied Biosystems; Thermo Fisher Scientific, Inc.). StepOne version 2.3 software was used to measure the RT-qPCR curves. Finally, Cq values were exported and analyzed in Microsoft Excel to determine the 2^-ΔΔCq values[28]. All experiments were performed at least five times, each with two technical replicates.

Cell lines

Colorectal adenocarcinoma Caco-2 cells (ATCC HTB-37) were from the American Type Culture Collection (Rockville, MD, United States). T84 cells (ATCC CCL-248) were purchased from Merck (Darmstadt, Germany). Caco-2 cells were grown in Dulbecco’s Modified Eagle Medium (DMEM) (Gibco; Thermo Fisher Scientific, Inc.) with 10% fetal bovine serum (FBS) (Sigma-Aldrich; Merck KGaA) and 1% penicillin-streptomycin (Sigma-Aldrich; Merck KGaA). T84 cells were cultivated in DMEM/F-12 (Gibco; Thermo Fisher Scientific, Inc.) with 10% FBS (Sigma-Aldrich; Merck KGaA) and 1% penicillin-streptomycin (Sigma-Aldrich; Merck KGaA). Additionally, the media for stably transduced cell lines contained 5 μg/mL puromycin. The cells were tested frequently for mycoplasma and characterized in May 2025 using short tandem repeats (STR) profiling, as indicated by the DSMZ online catalogue[29]. STR profiling of the eight STR loci was performed as described.

Transduction

Caco-2 as well as T84 cells express STAT3 endogenously. To reduce STAT3 expression in these cell lines, they were transduced with lentivirus encoding interfering MISSION^® short hairpin RNA (shRNA) nucleic acid molecules against STAT3, according to the manufacturer’s protocol (MISSION^® shRNA; Sigma-Aldrich; Merck KGaA). Briefly, cells were plated at a density of 3 × 10⁵ cells or 5 × 10⁵ cells per well and transduced with 3 μg of shRNA targeting STAT3 (MISSION^® shRNA TRNC0000329887) delivered through a viral vector (MISSION^® pLKO.1-puro). As a control, Caco-2 and T84 were transduced with the same amount of viral vector containing non-mammalian shRNA (MISSION^® pLKO.1-puro control plasmid DNA, SHC002). Transduced cells were selected using 5 μg/mL puromycin in the cell culture medium.

To obtain enhanced STAT3 expressing Caco-2 as well as enhanced STAT3 expressing T84 cell lines, cells were stably transduced with 3 μg of pLV[Exp]-Puro-CMV > hSTAT3 (VectorBuilder Inc., Chicago, IL, United States) as described above. Transduced cells were selected using 5 μg/mL puromycin in the cell culture medium.

Generation of the epithelial cell layer

For the generation of the epithelial monolayer, 1.5 × 10⁵ Caco-2 cells or 2.5 × 10⁵ T84 cells were seeded into polycarbonate transwell inserts (Tissue Culture-Insert, 12-well, polyethylene terephthalate, 0.4 μm, Transwell Plate, Sarstedt, Nürnberg, Germany) that were placed in 12-well culture plates (CELLSTAR multiwall plate, 12-well, standard, flat bottom, Greiner, Kremsmünster, Austria). The cells were cultured in DMEM (Gibco, United States) with 10% fetal calf serum (Sigma-Aldrich, United States), and 1% penicillin-streptomycin (Sigma-Aldrich, United States) for 28 days. Following seeding into transwell inserts, Caco-2 and T84 cells underwent a post-confluent differentiation process during which they form a continuous epithelial monolayer and establish functional tight junctions, leading to the gradual development of high transepithelial electrical resistance (TEER) and an effective epithelial barrier. Therefore, TEER measurements were also used as an indicator for barrier formation rather than depending only on cell number. Medium was changed every two to three days and the barrier integrity was assessed via measurement of TEER using Millicell ERS-2 volt/ohm meter (Merck, Germany). Resistance values were calculated for each well by subtracting values of a medium-only insert from the recorded resistance value, multiplied by the surface area of the membrane. All experiments were performed at least three times.

Western blotting

For western blotting, whole cell extracts (50 μg) were separated on 10% or 8% and 15% layered polyacrylamide gels, followed by transfer onto nitrocellulose membranes. Antibody detection was performed using standard procedures as previously described[30]. If indicated, the band intensity of protein expression from > 3 western blots was quantified using Fiji (ImageJ distribution; version 2.14.0/1.54f; National Institutes of Health, Bethesda, MD, United States). Antibodies used in this study are listed in Table 1. All experiments were performed at least three times.

Table 1 Primary antibodies, secondary antibodies, and target proteins.

Primary antibody (clone)	Target (molecular weight)	Dilution	Supplier
Anti-claudin 1	CLDN1 (approximately 22 kDa)	1:200	Invitrogen
Anti-claudin 2	CLDN2 (approximately 22 kDa)	1:300	Invitrogen
Anti-claudin 3	CLDN3 (approximately 20 kDa)	1:1000	Invitrogen
Anti-β-actin (AC-15)	β-actin (42 kDa)	1:5000	Sigma-Aldrich
Anti-occludin	OCLN (65 kDa)	1:500	Invitrogen
Anti-phospho-STAT1 (Tyr701) (58D6)	pSTAT1 (approximately 85 kDa)	1:500	Cell Signaling Technology
Anti-phospho-STAT3 (Tyr705) (D3A7)	pSTAT3 (approximately 80 kDa)	1:250	Cell Signaling Technology
Anti-STAT1 (42H3)	STAT1 (approximately 91 kDa)	1:1000	Cell Signaling Technology
Anti-STAT3 (124H6)	STAT3 (79-86 kDa)	1:1000	Cell Signaling Technology
Anti-ZO-1 monoclonal antibody (ZO1-1A12)	ZO-1 (225 kDa)	1:250	Invitrogen
Anti-ZO-1 monoclonal antibody (ZO-1A12, FITC)	ZO-1 (225 kDa)	1:500	Invitrogen
HRP-linked anti-mouse IgG	HRP-linked anti-mouse IgG	1:5000	Cell Signaling Technology
HRP-linked anti-rabbit IgG	HRP-linked anti-rabbit IgG	1:5000	Cell Signaling Technology

CLDN: Claudin; OCLN: Occludin; STAT: Signal transducer and activator of transcription; ZO-1: Zonula occludens-1; FITC: Fluorescein isothiocyanate; HRP: Horseradish peroxidase; IgG: Immunoglobulin G.

Immunofluorescence staining

Immunofluorescence staining of monolayers was performed by fixation of the cells in the inserts with 4% paraformaldehyde, permeabilization with 0.25% Triton X100, and blocking with 1% bovine serum albumin (BSA) for 1 hour. Then the cells were incubated with anti-ZO-1 (1:500) for 3 hours in the dark. Nuclei were counterstained by incubation with the intercalating agent 4’,6-diamidino-2-phenylindole (1:5000, 1 mg/mL) for 5 minutes. The inserts were thoroughly washed with phosphate-buffered saline (PBS) between these steps. Thereafter, the cells carrying filter membranes of the inserts were excised using a scalpel, placed on glass slides, and mounted with Fluormount Aqueous Mounting Medium (Sigma Aldrich). Fluorescence images were acquired using a Keyence BZ-X810 fluorescence microscope (Keyence, Osaka, Japan). All experiments were performed at least three times.

Immunohistochemical analysis

For immunohistochemistry, 3.5 μm sections of samples cut from formalin-fixed, paraffin-embedded colon tissue were used. First, sections were deparaffinized twice with xylene and rehydrated in a decreasing series of five alcohol solutions. Antigen retrieval was performed by heating the sections in ethylenediaminetetraacetic acid buffer [potential of hydrogen (pH) = 9.0] using a microwave oven and brought to boiling seven times for a total duration of approximately 20 minutes. This was followed by incubation for 10 minutes with 3% hydrogen peroxide to block endogenous peroxidase activity. Sections were washed with 1X PBS (Gibco; Thermo Fisher Scientific, Inc.) before and in between incubation steps. The primary antibodies STAT3 (RD Systems, Minneapolis, MN, United States; clone 124H6) and pSTAT3 (ThermoFisher Scientific, Waltham, MA, United States; cat. no. 44-384G) were diluted in PBS containing 1% BSA. Sections were incubated with the primary antibody at 4 °C overnight, followed by application of the mouse EnVision System (cat. no. K4000; Agilent Technologies, Inc.), which employs the enzyme horseradish peroxidase coupled to a secondary antibody and the chromogen 3,3’-diaminobenzidine. Sections were counterstained for 30 seconds using Gill’s hematoxylin solution (Sigma Aldrich; Merck KGaA) [hematoxylin and eosin (HE) staining]. Immunohistochemical staining was examined using a Keyence BZ-X810 fluorescence microscope (Keyence, Osaka, Japan). Negative controls were processed in parallel to exclude non-specific staining.

Transmission electron microscopy, sample preparation, and image generation

Cell monolayers of Caco-2 and T84 cells were grown on transmembrane inserts for 28 days, as described above. The integrity of the cell monolayers was assessed by TEER measurement. Thereafter, the monolayers were prepared for transmission electron microscopy (TEM) as follows: A fresh fixation solution was prepared by mixing 9 parts fixative solution with 1 part 25% glutaraldehyde solution. The glutaraldehyde was handled using a needle and syringe under a fume hood to ensure safety. The apical culture medium was carefully removed, and 500 μL of PBS was gently added to the apical side. Then, the basolateral medium was aspirated, and 1 mL PBS was added to the basolateral compartment. The apical PBS was subsequently removed to eliminate residual culture medium. The inserts were then placed on sterile glass slides, and the membrane was cut off from the rest of the transwell.

The excised membranes were transferred into a 12-well plate with the cell layer facing upward. The fixation of samples was carried out as described[31]. In brief, 2 mL of 2.5% glutaraldehyde (CARL ROTH, 4157.2) in 0.1 M cacodylate buffer pH = 7.2 (CARL ROTH, 5169.2) solution was gently added to each well, ensuring the membranes were fully immersed. The plate was incubated on a shaker at the lowest speed for 1 hour. The fixation solution was then replaced with fresh fixation solution for a second 1-hour incubation under the same conditions. After two washes in 0.1 M cacodylate buffer containing 2% sucrose, cells were post-fixed in 1% reduced osmium tetroxide dehydrated (CARL ROTH, 8088.1) and embedded in Epoxy embedding resin (Araldite CY212 Premix Kit; Agar Scientific, United Kingdom). 50 nm sections were cut with an ultramicrotome (Leica). Ribbons of sections were transferred on Pioloform-coated copper slot grids and contrast enhanced with 5% uranyl acetate in methanol/water and lead citrate Reynolds[32]. Micrographs were taken with a Zeiss TEM900 microscope operated at 80 keV in the bright-field mode and quipped with a Troendle 2K camera.

Liquid chromatography time-of-flight mass spectrometry

Cell monolayers of Caco-2 and T84 cells were grown on transmembrane inserts for 28 days, as described above. After washing with 500 μL PBS, cells were detached mechanically (scraping and pipetting, without trypsin), counted, and 2.5 × 10⁵ cells (according to analysis request) were transferred into 50 μL PBS. For each cell line and condition, five biological replicates (independent passages) were prepared. Each biological replicate was measured in three different inserts as triplicates (technical replicates). Samples were blinded using random alphanumeric IDs and stored at -80 °C until liquid chromatography high-resolution mass spectrometry analysis (LC-HRMS).

LC-HRMS measurement was carried out as described previously[33]. For the analysis, 2.5 × 10⁵ cells (according to analysis request) in 50 μL PBS were mixed with 75 μL of internal standards in methanol, along with 250 μL of methyl tert-butyl ether (MTBE) and 10 μL of 50 mmol/L ammonium formate. The mixture was vortexed and then subjected to centrifugation at 20000 g for 5 minutes at 4 °C. The upper organic phase was carefully transferred, while the aqueous phase underwent a re-extraction using 100 μL of a solvent mixture (MTBE/methanol/water in a ratio of 10:3:2.5, v/v/v). The combined organic phases were then evaporated at 45 °C under a nitrogen stream and stored at -80 °C. Prior to analysis, the samples were reconstituted in 100 μL of methanol.

Quality control involved pooling different cell lines to assess method precision and pooling human plasma samples to assess technical variability. Two quality control injections were performed at the start and end of each run, with an additional injection after every 10 samples.

The measurements were carried out using a Vanquish Horizon system (Thermo Fisher), equipped with a Zorbax RRHD Eclipse Plus C8 1.8 μm, 50 mm × 2.1 mm ID column and a matching pre-column from Agilent Technologies. This setup was coupled to an Exploris 480 Orbitrap mass spectrometer (Thermo Fisher, Dreieich, Germany). Ionization was performed using Health and Environmental Sciences Institute in both positive and negative modes. The mass range was set from 180 m/z to 1500 m/z at a resolution of 120000, with data-dependent acquisition featuring a cycle time of 0.6 seconds and a resolution of 15000 for enhanced identification (± 5 ppm). Data acquisition was conducted using Xcalibur 4.4, and data evaluation was performed with TraceFinder 5.1 (both Thermo Fisher, Dreieich, Germany).

Statistical analysis

Ct values of target genes were normalized to 18S rRNA by calculating ΔCt values (Ct target – Ct 18S) for each sample. ΔCt values were used for all statistical analyses. Group differences were assessed using a one-way analysis of variance. Relative gene expression is presented as 2^-ΔΔCt, representing the x-fold change (FC) compared with the reference group. Data are presented as mean ± SEM, and a P value < 0.05 was considered statistically significant. Statistical analyses were performed using GraphPad Prism 10 for macOS [version 10.1.1 (270); GraphPad Software, San Diego, CA, United States] and Microsoft Excel (Microsoft Corporation, Redmond, WA, United States).

For statistical analysis of lipidomics, data were unblinded, and technical replicates were averaged per biological replicate to avoid pseudo-replication. Raw lipid intensities were log₂-transformed prior to analysis. Pairwise group comparisons were performed separately for each cell line using two-sided Welch’s t-tests (unequal variances assumed). Specifically, enhanced STAT3 expressing Caco-2 or T84 vs lipids of control cells and reduced STAT3 expressing Caco-2 or T84 vs lipids of control cells with P < 0.05 in each contrast were considered significantly altered. To identify robust and biologically reproducible changes, only lipids that were significant in both cell lines (Caco-2 and T84) and showed the same direction of regulation were retained (cross-validation across datasets).

FC calculation

Log₂FC = log₂ (mean of control/mean of group A) for each comparison. Bidirectional regulation between STAT3 overexpressing and reduced STAT3 expressing cells (opposite log₂FC signs) was interpreted as STAT3-dependent activation or repression of the corresponding lipid species. Mean values, SD, and SEM were calculated from five independent biological replicates using standard formulas (SD = sample standard deviation; SEM = SD/√n).

RESULTS

Intestinal tissue from mice with ACLF shows significantly increased STAT3 expression and marked changes in cell morphology

To analyze cell epithelial morphology as well as the expression and activation of STAT3 in intestinal tissue during the progression of liver disease from steatosis to ACLF, we first performed HE staining followed by immunohistochemical analysis of STAT3 as well as pSTAT3 in intestinal tissue of healthy mice, mice with steatosis, mice with cirrhosis, and those with ACLF (Figure 1). The epithelial structure of the intestinal tissue increasingly lost its cohesion during the progression of liver disease (Figure 1A). A significant increase of STAT3 expression could be detected in intestinal tissue of mice with cirrhosis and ACLF compared to the expression of STAT3 in intestinal tissue of healthy mice (Figure 1B and D). Also, STAT3 expression in intestinal tissue of mice with steatosis tended to be lower than in tissue of mice with ACLF (Figure 1B and D). However, the pSTAT3 levels did not increase in the same manner and were not significantly altered in the intestinal tissue of mice over the course of the disease from steatosis to cirrhosis to ACLF (Figure 1C and E). This highlights a clear in vivo disconnect between total STAT3 and its phosphorylation status.

Figure 1 Intestinal tissue of mouse models shows increased signal transducer and activator of transcription 3 expression and altered epithelial integrity during progression from liver steatosis to acute-on-chronic liver failure. A-C: Paraffin embedded intestinal tissue of mouse models generating steatosis, cirrhosis, and acute-on-chronic liver failure (ACLF) was used for: Hematoxylin and eosin (HE) staining (A); Immunohistochemical analysis of signal transducer and activator of transcription 3 (STAT3) and phospho-STAT3 (pSTAT3) and compared to tissue of healthy controls, respectively (B and C); D and E: STAT3 and pSTAT3 protein levels of immunohistochemical staining were quantified using Fiji (ImageJ) by color deconvolution, followed by signal quantification and graphical representation in GraphPad Prism; F: Disease stage-dependent changes of STAT3 messenger RNA levels from cirrhosis to ACLF were determined by reverse transcription-quantitative polymerase chain reaction (RT-qPCR) and compared. HE staining demonstrated a progressive disruption of intestinal epithelial architecture during disease progression. Immunohistochemistry revealed a significant upregulation of STAT3 expression in intestinal tissue from ACLF mice compared with healthy controls, which was confirmed by RT-qPCR. Statistical analysis of immunohistochemical staining was performed using one-way analysis of variance followed by Tukey’s multiple comparisons test. Differences in STAT3 messenger RNA expression were assessed by unpaired Student’s t-test. P values are two-sided (n = 3-6). ^aP < 0.05. ^bP < 0.01. ^cP < 0.001. ACLF: Acute-on-chronic liver failure; STAT3: Signal transducer and activator of transcription 3; pSTAT3: Phospho-signal transducer and activator of transcription 3; mRNA: Messenger RNA; NS: Not significant.

STAT3 and tight-junction mRNA levels are altered in intestinal tissue during steatosis, cirrhosis, and ACLF

To better understand a possible link between increased STAT3 expression and proteins associated with increased intestinal permeability and progression of liver disease, we also conducted a comparative analysis of the intestinal tissue of our different mouse models focusing on the mRNA levels of Stat3, Stat1, Tjp1, Ocln, Cldn1, Cldn2 and Cldn3. RT-qPCR results, summarized in a heatmap normalized to healthy controls (Supplementary Figure 1, blue boxes), revealed disease-stage-dependent transcriptional alterations of barrier- and STAT-related genes. While most transcripts displayed consistent trends across disease models, statistically significant changes were observed for Stat3 in cirrhosis and ACLF. The direction and magnitude of these transcriptional changes differed between disease stages.

In steatosis, Tjp1 and Cldn3 were upregulated, whereas cirrhosis was characterized by a pronounced downregulation of Cldn2, Ocln, Stat1, and Stat3, accompanied by increased expression of Tjp1 and Cldn3. In intestinal tissue of mice with ACLF, Cldn2, Stat1, and Ocln showed a mild upregulation, while Stat3 was lower in intestinal tissue of mice with cirrhosis but still significantly enhanced compared to the healthy control.

Successful generation of stably differential STAT3 expressing cells

To analyze the effect of enhanced expression of STAT3 on gut permeability in vitro we generated differential STAT3 expressing Caco-2 cells as well as T84 cells by lentiviral transduction (both cell lines are able to form epithelial monolayer). The establishment of these cell lines was intended to provide a model system that may later allow us to mimic aspects of the in vivo context and to explore potential therapeutic interventions. Successful modulation of STAT3 expression was verified at the protein level.

STAT3 protein levels were verified by western blotting (Figure 2A; corresponding original uncropped western blots shown in Supplementary Figure 2) comparing decreased and enhanced STAT3 expressing cell lines to endogenous STAT3 expressing controls. Quantification of four independent experiments demonstrated the successful generation of reduced and enhanced STAT3 expressing Caco-2 and T84 cells (Figure 2B).

Figure 2 Successful generation of differential signal transducer and activator of transcription 3 expressing cell lines. Signal transducer and activator of transcription 3 (STAT3) expression was modified by generation of differential, stable, lentiviral transduced Caco-2 or T84 cell lines. Reduction of STAT3 (STAT3 -) was produced via short hairpin RNA (shRNA) targeting STAT3. Enhanced STAT3 expression (STAT3 +) was induced by transduction of STAT3 expression plasmids. A non-mammalian shRNA expressing pLKO.1 vector was used as a control. Differential STAT3 expression was validated frequently from the start to the end of the experiments. Total protein was extracted, and proteins of interest were analyzed. A: Western blot; B: Quantified using Fiji (ImageJ distribution; version 2.14.0/1.54f). Values were normalized to Caco-2 or T84 control cell lines. Bars indicate mean ± SD. P values were calculated using one-way analysis of variance and post hoc Tukey analysis (n = 4). ^bP < 0.01. ^dP < 0.0001. STAT3: Signal transducer and activator of transcription 3.

Changes in STAT3 expression causes enhanced epithelial permeability

To analyze the influence of enhanced or reduced STAT3 expression on epithelial permeability, monolayers of differential STAT3 expressing Caco-2 and T84 cells were cultivated over a period of 28 days and TEER measurements were performed. As shown in Figure 3, both increased and reduced STAT3 expression led to significantly lower monolayer resistance values for Caco-2 and T84 cells compared to stably transduced mock control cells. Hereby, reduced STAT3 expressing Caco-2 as well as reduced STAT3 expressing T84 monolayer (Figure 3C and D) showed a slightly stronger reduction of the epithelial barrier than enhanced STAT3 expressing Caco-2 or enhanced STAT3 expressing T84 monolayer compared to the controls, respectively (Figure 3C and D). In addition, Caco-2 monolayers showed generally lower resistance values (Figure 3A and C) than T84 monolayers (Figure 3B and D).

Figure 3 Signal transducer and activator of transcription 3 reduction as well as enhanced signal transducer and activator of transcription 3 expression impairs permeability of epithelial cell layers. To determine the influence of signal transducer and activator of transcription 3 (STAT3) on epithelial permeability, monolayers consisting of Caco-2 or T84 cells with different levels of STAT3 expression were generated and epithelial permeability was analyzed by transepithelial electrical resistance (TEER) measurements over 28 days and compared to controls. A and B: Graphs show the differences of TEER values after 4 days, 7 days, 11 days, 14 days, 18 days, 21 days, 25 days and 28 days compared with endogenous STAT3 expressing (pLKO.1 transduced) controls; C and D: TEER values at day 28 were calculated in relation to controls in percent and shown as bar diagrams. Reduced (STAT3 -) as well as enhanced STAT3 (STAT3 +) expression resulted in a significant reduction in monolayer resistance values of Caco-2 and T84 monolayers. Data are expressed as mean ± SEM as appropriate. Differences between mean resistance values were assessed for statistical significance using one-way analysis of variance followed by Dunnett multiple comparisons test. P values are two-sided (n = 3-5). ^dP < 0.0001. STAT3: Signal transducer and activator of transcription 3; TEER: Transepithelial electrical resistance.

Altered STAT3 expression has significant impact on STAT signaling and tight-junction proteins in Caco-2 and T84 monolayers

To assess how differential STAT3 expression influences STAT signaling and the expression of key tight junction proteins, we analyzed Caco-2 and T84 monolayers with reduced or increased STAT3 levels after 28 days using western blotting. As expected and demonstrated in Figure 4, pSTAT3 levels were substantially decreased in Caco-2 and T84 monolayers with reduced STAT3 expression. Notably, even increased STAT3 expression led to an approximately 50% reduction in pSTAT3 in both cell lines. In contrast, STAT1 expression remained unchanged in both cell lines, whereas pSTAT1 was reduced in Caco-2 monolayers with reduced STAT3 expression and significantly decreased when STAT3 was increased (Figure 4A and C). A similar pattern was observed in T84 monolayers, where pSTAT1 was significantly diminished under both conditions, with the strongest reduction seen in cells with reduced STAT3 expression (Figure 4B and D). To determine whether these alterations in STAT signaling were accompanied by structural changes at the tight junction level, we next examined key junctional proteins.

Figure 4 Expression of relevant tight junction proteins is significantly modified by differential signal transducer and activator of transcription 3 expression in Caco-2 and T84 monolayers. Monolayers consisting of Caco-2 or T84 cells with different levels of signal transducer and activator of transcription 3 (STAT3) expression were generated and epithelial permeability was determined by transepithelial electrical resistance measurements over 28 days. A and B: Whole protein extracts were analyzed via western blot (an exemplary blot is shown on which the protein extracts from three independently performed test series were applied next to each other; C and D: Quantified using Fiji (ImageJ distribution; version 2.14.0/1.54f). Values were normalized to Caco-2 or T84 control cell lines. Bars indicate mean ± SD. Western blots were performed with specific primary antibodies against STAT3, phospho-STAT3, STAT1, phospho-STAT1, Ocln, Cldn1, Cldn2, Cldn3 and zonulin 1, while β-actin served as a housekeeping protein. As shown, reduced (STAT3 -) as well as enhanced STAT3 (STAT3 +) expression had an important impact on the activation of STAT1 as well as the expression of tight junction proteins. P values were calculated using two-way analysis of variance followed by Dunnett’s multiple comparisons test. The data shown are mean ± SD, and the following P values were considered statistically significant (n ≥ 3). ^aP < 0.05. ^bP < 0.01. ^cP < 0.001. ^dP < 0.0001. STAT3: Signal transducer and activator of transcription 3; pSTAT3: Phospho-signal transducer and activator of transcription 3; STAT1: Signal transducer and activator of transcription 1; pSTAT1: Phospho-signal transducer and activator of transcription 1; ZO-1: Zonula occludens-1.

Looking at tight junction proteins, we detected that they were differentially affected in the analyzed cell lines. Ocln was significantly reduced in Caco-2 as well as T84 monolayers with reduced STAT3 expression (Figure 4). In contrast, Ocln levels showed a trend to be enhanced in Caco-2 monolayers with increased STAT3 levels (Figure 4A and C; corresponding original uncropped western blots see Supplementary Figure 3A). Meanwhile, increased STAT3 expressing T84 monolayers showed significantly increased Ocln levels (Figure 4B and D; corresponding original uncropped western blots in Supplementary Figure 3B). Cldn1 was significantly decreased in Caco-2 cells with reduced STAT3 and showed no change in cells with increased STAT3 levels. In contrast, T84 monolayers with reduced and enhanced STAT3 expression both led to a significant reduction of Cldn1. Given the functional importance of claudin isoforms in regulating paracellular permeability, we subsequently analyzed Cldn2 and Cldn3 expression patterns.

Cldn2 expression was significantly increased only in Caco-2 monolayers with elevated STAT3 expression, in contrast to T84 monolayers, where it was significantly reduced under increased STAT3 conditions. Cldn3 was significantly elevated in Caco-2 monolayers under both reduced and increased STAT3 expression, while T84 monolayers showed a significant reduction when STAT3 was reduced. ZO-1 was significantly enhanced in Caco-2 monolayers with increased STAT3 levels. However, increased STAT3 expressing T84 monolayers showed only a trend of enhanced ZO-1 levels.

Enhanced as well as reduced STAT3 expression induced major architectural changes in Caco-2 and T84 monolayers

To demonstrate the effects of differential STAT3 expression on cell organization and morphology, both immunofluorescence staining (Figure 5A and B) as well as TEM (Figure 5C and D) were performed. Using immunofluorescence, we found that the uniform lattice structure visible in the monolayers of the control cells was significantly disturbed in reduced STAT3 as well as increased STAT3 expressing Caco-2 and T84 monolayers (Figure 5A and B). Moreover, ZO-1 seemed to be no longer clearly arranged at the cell edges but rather diffusely distributed over the cell in the differential STAT3 expressing cell monolayers (Figure 5A and B). To further characterize these structural alterations at ultrastructural resolution, we next performed TEM.

Figure 5 Immunofluorescence staining and transmission electron microscopy verify deconstructed monolayer architecture in enhanced as well as reduced signal transducer and activator of transcription 3 expressing Caco-2 and T84 monolayers. Differential signal transducer and activator of transcription 3 (STAT3) expressing Caco-2 and T84 cell lines were grown over a period of 28 days and the formation of Caco-2 and T84 monolayers was verified during this period. A and B: Caco-2 (A) T84 (B) monolayers were fixed, permeabilized, incubated with anti zonula occludens-1, and subsequently stained with a fluorescent-labeled antibody. Nuclei were counterstained with 4’,6-diamidino-2-phenylindole. Imaging was performed using a Keyence BZ X810 fluorescence microscope; C and D: In parallel, Caco-2 (C) and T84 (D) monolayers were prepared for transmission electron microscopy (TEM). Immunofluorescence as well as TEM showed the deconstructed architecture in enhanced STAT3 (STAT3 +) as well as reduced STAT3 (STAT3 -) expressing monolayers. Orange arrows = gaps at cell boundaries, black arrows = location of tight junctions. STAT3: Signal transducer and activator of transcription 3; DAPI: 4’,6-diamidino-2-phenylindole; ZO-1: Zonula occludens-1.

By TEM, it could be detected that tight junction proteins were no longer arranged in a structured manner, that cell-cell contacts were significantly destroyed, and that the morphology of the layers was disrupted. Decreased STAT3 as well as increased STAT3 expressing Caco-2 and T84 monolayers showed gaps at the cell boundaries and cells appeared to form disorganized convolutions resembling the structure of apical microvilli, respectively (Figure 5C and D, orange arrows). Looking at the localization of the tight junctions (Figure 5C and D, black arrows), those of decreased STAT3 as well as increased STAT3 expressing Caco-2 and T84 monolayers showed significant alterations in localization. Tight junctions of differential STAT3 expressing monolayers were positioned deeper within the cells and no longer formed a tight seal between adjacent cells (Figure 5C and D, black arrows). Both decreased STAT3 as well as increased STAT3 expressing Caco-2 and T84 monolayers showed a visibly looser cell connection compared to endogenous STAT3 expressing monolayers which were characterized by dense cell-cell contacts and tight junctions located at the very apical part of the cells and formed a tight seal (Figure 5C and D).

Differential expression of STAT3 is correlated with significant changes in lipid species

TEER values reflect both paracellular and transcellular permeability. To specifically assess transcellular changes and investigate whether STAT3 expression influences epithelial lipid composition, LC-HRMS analysis was performed using enhanced STAT3 as well as reduced STAT3 Caco-2 and T84 cell monolayers and compared to endogenous STAT3 expressing control cells. LC-HRMS analysis revealed significant STAT3-dependent alterations in membrane-associated lipid species, which were partially altered in both cell lines (Figure 6A and B, Supplementary Tables 1 and 2). In decreased STAT3 expressing Caco-2 or T84 cells (Figure 6C), several phospholipid classes were significantly down-regulated, including phosphatidylglycerol (40:8) and lysophosphatidylcholines (LPC) (LPC 18:2, LPC 20:3), while ether-linked phosphatidylcholine (PC) (PC O-36:1) and multiple triacylglycerol (TG) species [TG 52:1; ether-linked TG (TG O-50:1/-50:2/-50:3/-52:3)] were increased. Conversely, enhanced STAT3 expressing Caco-2 or T84 cell monolayers showed up-regulation of ether-linked phosphatidylcholine (PC O-18:1, 18:1, PC O-34:1), and sphingomyelin (SM) (SM 36:0; dihydroxy sphingoid base (O2) (Figure 6D).

Figure 6 Identification of signal transducer and activator of transcription 3-dependent differences of lipid species by liquid chromatography high-resolution mass spectrometry analysis. Caco-2 and T84 cell monolayers were generated over a period of 28 days. Thereafter, cells were harvested and lipidomic analysis was performed. A and B: Venn diagrams illustrate the number of significantly altered lipid species in decreased signal transducer and activator of transcription 3 (STAT3) (STAT3 -) and enhanced STAT3 (STAT3 +) expressing Caco-2 and T84 cell monolayers compared to the controls. The overlapping area represents lipids significantly regulated in both epithelial models, indicating STAT3-dependent lipid remodeling; C and D: Bars show all lipid species that were significantly changed in Caco-2 as well as T84 monolayers (see also A and B), respectively. Bars represent mean log₂ fold change ± SEM. Cross-validation across two independent intestinal epithelial models ensures robustness and minimizes false-positive discovery. P < 0.05, two-sided Welch’s t-test (n = 5). STAT3: Signal transducer and activator of transcription 3; PG: Phosphatidylglycerol; LPC: Lysophosphatidylcholines; PC: Phosphatidylcholine; TG: Triacylglycerol; SM: Sphingomyelin; PC O: Ether-linked phosphatidylcholine; TG O: Ether-linked triacylglycerol.

DISCUSSION

The present study aimed to determine the possible drivers for frequently occurring increased permeability of the intestinal epithelium in the course of disease progression from liver cirrhosis to ACLF[34]. Since STAT3 has been described to play a crucial role in epithelial homeostasis[35], immunohistochemical staining from intestinal tissues of different mouse models, representing various stages of liver disease, were performed at the beginning of this study. The results demonstrated a clear increase in STAT3 protein concentrations during the progression of liver disease, with a significant elevation in ACLF compared to healthy controls. Notably, this increase was not paralleled by a proportional rise in phosphorylated STAT3, suggesting accumulation of unphosphorylated STAT3 (uSTAT3). This also indicates that disruption of STAT3 homeostasis, rather than enhanced canonical signaling, may contribute to intestinal barrier dysfunction. This interpretation is consistent with previous studies demonstrating that STAT3 exerts context- and dose-dependent effects on epithelial integrity, is essential for mucosal repair, and becomes maladaptive when dysregulated[13,36].

The mRNA analysis of the intestinal tissue from our mouse models showed altered mRNA levels of key tight junction proteins compared to controls, indicating compromised epithelial barrier integrity. These findings support the idea of tight junction protein regulation via increased STAT3. To verify this hypothesis, we generated different STAT3 expressing colon cells lines and analyzed the effect of differential STAT3 expression on barrier function, expression level of associated tight junction proteins, morphological structure, and lipid composition using in vitro generated gut epithelium monolayers. Interestingly, we found that any alteration of STAT3 expression, both increased and decreased, was able to significantly change various structural and functional determinants of epithelial barrier homeostasis, including its permeability, the expression and organization of tight junction proteins, the cellular morphology, and the lipid composition of epithelial cells.

The detected clear association between reduced STAT3 activation and barrier dysfunction in our study is consistent with previous findings of Pang et al[36]. Pang et al[36] demonstrated that the reduction of STAT3 activity significantly impairs the intestinal barrier and demonstrated that it increases the severity of colitis and significantly impairs epithelial proliferation. Furthermore, it has been shown that the increase of activated STAT3 can enhance epithelial permeability. Mechanistically, long-chain fatty acids have been detected to promote STAT3 activation through palmitoylation, thereby enhancing its phosphorylation state, while genetic ablation of STAT3 effectively counteracts these pro-inflammatory effects[37]. In addition, increased STAT3 activity has been reported to drive interleukin-6 production in T helper cells, further amplifying mucosal inflammation[38]. However, no study has conclusively shown that changes in STAT3 expression alone, independent of phosphorylation-driven activation, are sufficient to disrupt epithelial barrier homeostasis. In this study, we demonstrate for the first time that increased STAT3 abundance, without proportional phosphorylation, is associated with impaired epithelial barrier integrity, indicating deregulated STAT3 levels in the absence of physiological activation. Notably, this pattern was most pronounced in advanced disease stages, particularly in ACLF. These findings suggest that accumulation of uSTAT3 may contribute to epithelial barrier vulnerability and could represent a molecular feature that precedes disease progression from cirrhosis, reflecting the culmination of STAT3 imbalance.

While we did not directly assess chromatin binding or transcriptional complex formation of uSTAT3 in IECs, prior studies have demonstrated that uSTAT3 can regulate gene expression through alternative transcriptional mechanisms[39-44], supporting the biological plausibility of our model. We therefore hypothesize that ACLF-associated inflammatory signaling may upregulate STAT3 expression to a level that enables its transcriptional activity in IECs independent of canonical phosphorylation-based activation and might affect key tight junction proteins. Indeed, we observed a clear STAT3-level-related regulation of Ocln and ZO-1. Furthermore, the tested Cldn also showed a pronounced, albeit heterogeneous, change in expression. An imbalance of STAT3 thus clearly disrupts the coordinated transcriptional regulation of tight junctions, with the results being influenced by the specific mutation status of each cell line. Moreover, we detected that essential epithelial characteristics such as morphology and lipid composition were strongly influenced by STAT3 expression levels. In agreement with this, Yang et al[45] described that elevated uSTAT3 can drive gene expression via heterodimerization with other transcription factors. Based on this, we speculate that increased STAT3 levels contribute substantially to altered tight junction composition and, consequently, impaired epithelial permeability.

Tight junction proteins critically regulate cell-cell contacts and ion exchange[46], and both their abundance and localization are essential for barrier function. To our surprise, we detected increased ZO-1 expression in enhanced STAT3 expressing cells, but immunofluorescence and TEM revealed markedly disorganized ZO-1 distribution compared with the well-defined junctional pattern in controls. Although, most reports associate ZO-1 with barrier protection[47-49], a correlation that would be expected due to increased ZO-1 expression, several other studies describe ZO-1 as an adaptor linking tight junction proteins to the actin cytoskeleton and modulating their ability to regulate barrier integrity[50-53]. This means that ZO-1 mislocalization, not only loss, can disrupt junctional organization[54]. Thus, the aberrant ZO-1 distribution observed in our models offers a plausible explanation for the STAT3 dependent increase in epithelial permeability.

While the data discussed above directly relate to changes in epithelial permeability, the detected lipidomic alterations did not fit with the classical lipid signatures typically associated with paracellular leakiness. Elevated LPCs are well known to induce permeability and junctional disassembly in epithelial and endothelial models via Ras homolog family member A/Rho-associated coiled-coil containing protein kinase activation and cytoskeletal stress fiber formation[55]. In rat small intestine, for example, LPC exposure has been shown to increase mucosal permeability and cause epithelial injury[56]. However, since LPC 18:2 and LPC 20:3 levels in our dataset were significantly reduced in cell layers with decreased STAT3 expression, we suggest that the observed increase in epithelial permeability occurs independently of LPC accumulation. Likewise, the higher abundance of ether-linked PCs (plasmalogens) and sphingomyelin species, which we detected in cells with elevated STAT3 expression, has been previously described as protective against oxidative[57] or inflammatory membrane stress[58] and therefore does not appear to be associated with increased epithelial permeability. The lipidomic alterations observed in cells with enhanced STAT3 expression likely reflect STAT3-dependent membrane remodeling and may represent adaptive or contributory responses. However, their direct causal involvement in barrier dysfunction remains to be determined by dedicated functional studies. Collectively, these data suggest that lipid alterations are unlikely to represent the primary driver of barrier dysfunction in our models. Rather, the observed permeability changes appear to be predominantly associated with tight junction disassembly, although context-dependent contributions of membrane lipid remodeling cannot be fully excluded.

Pharmacological strategies aiming to restore STAT3 homeostasis, rather than blanket inhibition, could be promising. This includes modulating upstream regulators to prevent STAT3 overexpression or specifically inhibiting the nuclear translocation or DNA-binding of pathologically accumulated uSTAT3[4,56,59,60]. However, our findings suggest that therapeutic benefit may not only derive from complete STAT3 blockade, but rather from restoration of physiological STAT3 balance. Future strategies may therefore require context-dependent normalization of STAT3 activity instead of indiscriminate inhibition. This concept is consistent with recent reviews proposing that therapeutic strategies targeting STATs should aim to modulate its activity and restore physiological homeostasis rather than pursue complete pathway abrogation[61].

In this context, given the pleiotropic and context-dependent functions of STAT signaling pathways, we assume that careful patient stratification and disease stage-specific modulation of STATs activity will be most essential in advanced liver disease. Therapeutic strategies that selectively attenuate pathological inflammatory signaling by preserving STAT3-mediated epithelial homeostatic functions may represent a more refined approach to restoring intestinal barrier integrity and potentially slowing cirrhosis progression.

It should be noted that the steatosis and cirrhosis models used in this study were exclusively based on chronic alcohol exposure or combined alcohol/CCl₄ administration and therefore may not fully recapitulate alternative etiologies, including diet-induced or metabolic liver disease. Future studies employing complementary disease models will be required to determine whether the STAT3-dependent mechanisms identified here represent a common feature across different etiologies of chronic liver disease[62]. In the broader context of ACLF pathophysiology, intestinal STAT3 imbalance and accumulation of uSTAT3 may exacerbate gut-liver axis dysfunction by facilitating bacterial translocation and systemic inflammatory amplification, thereby contributing to the progression from compensated cirrhosis to acute decompensation[63].

Although our experimental model reflects alcohol-associated ACLF, the observed STAT3 dysregulation is likely driven by inflammation-dependent mechanisms common to advanced liver failure. Nevertheless, validation in additional non-alcoholic models will be essential to determine the broader applicability of our findings.

CONCLUSION

In conclusion, our results demonstrate for the first time that STAT3 homeostasis, particularly increased STAT3 abundance beyond physical levels, is sufficient to induce markedly increased intestinal permeability and profound structural changes in epithelial cell monolayers, recapitulating central features of the intestinal phenotype observed in alcohol-associated ACLF mice. These findings suggest that preserving physiological STAT3 homeostasis in the intestinal epithelium represents a promising strategy to mitigate disease severity and potentially delay progression from liver cirrhosis to ACLF.

ACKNOWLEDGEMENTS

We thank Uschner F for his valuable initial contribution by providing rat models. We further acknowledge Lafferton B for her support in generating stable Caco-2 cell lines and Henle G for providing mouse intestinal tissue. We thank Mojaradfar R for his technical assistance in conducting the LC-HRMS measurements. Finally, we would like to thank Overby S for her careful proofreading. This manuscript is part of Leinz N’s PhD thesis.