H2-Rich Medium Improves Barrier Dysfunction in CACO-2 CellsScientific Research

INTRODUCTION

Defined as the presence of infection together with systemic manifestations, sepsis is a deleterious host response ⁽¹⁾. Sepsis, as well as its adverse consequences, remains to be the primary cause of death in the noncoronary intensive care unit (ICU) ^{(1, 2)}. The intestine has widely been recognized as a crucial organ during the genesis and development of sepsis, as well as multiple organ dysfunction syndrome (MODS) ⁽³⁾. The gastrointestinal epithelium is frequently in touch with xenobiotics and various microfloras, and represents a defensive barrier between the organism and the external environment ⁽⁴⁾.

An integrated gut barrier is composed of well differentiated intestinal epithelial cells, which are tightly connected as a monolayer by the apical junctional complex (AJC). The AJC contains tight junctions (TJs) and adherens junctions (AJs), and encircles the apical ends of the lateral membranes of the epithelial cells, determining the selective paracellular permeability of the intestinal barrier ⁽⁵⁾. Both TJs and AJs are dynamically modulated in physiological and pathological conditions.

Among a number of signaling molecules that have been demonstrated to regulate the AJC, Ras homolog gene family (Rho) of small GTPases stands out ⁽⁶⁾. RhoA ranks the most studied molecule among the Rho family, regulating the assembly, maintenance, and disassembly of AJC in the intestinal epithelium ⁽⁷⁾. Mammalian diaphanous-related formin 1 (mDia1), or interchangeably called DIAPH 1, is a main RhoA effector that modulates barrier via actin polymerization, as well as microtubule organization ^{(8, 9)}. Hence, the RhoA-mDia1 signaling is supposed as a potential vital target for the treatment of lipopolysaccharide (LPS)-related hyperpermeability in the intestine.

Molecular hydrogen (H₂), considered as a physiological noble gas before, has currently been identified as a promising medical gas for many diseases ^{(10, 11)}. Our previous studies have revealed that H₂ inhalation at a concentration of 2% or 4% exerts significant curative effects on sepsis and sepsis-related multiple organ damage through reducing uncontrolled inflammation, excessive oxidative stress, irregular apoptosis, and regulating abnormally activated signal pathways ^(12–14). Coincidentally, H₂ treatment can attenuate intestine injury ^(15–17). However, whether the protective effects of H₂ on sepsis is via regulating intestinal barrier integrity or not is still largely unknown.

On the basis of the studies mentioned above, we utilized an in vitro model of intestinal epithelium, Caco-2 cell monolayers, to hypothesize the potential beneficial role of H₂-rich medium in alleviating LPS-caused gut barrier dysfunction, likely through the RhoA-mDia1 signal pathway.

MATERIALS AND METHODS

Ethical approval

This research was carried out according to the Institutional Animal Care and Use Committee Guide in General Hospital of Tianjin Medical University.

Reagents

The whole cell culture reagents and supplements were from GIBCO (Grand Island, NY). Fluorescein-isothiocyanate (FITC)-labeled dextran 4 kDa (FD4) and Escherichia coli LPS 0111:B4 (Product number: L4391) were from Sigma-Aldrich (St Louis, MO). Rho inhibitor C3 exoenzyme and RhoA activator CN03 were from Cytoskeleton (Denver, CO).

Cell culture

The human colon cancer cell line Caco-2 were obtained from ATCC (Manassas, VA), and maintained in complete Dulbecco modified Eagle medium (DMEM) with 20% fetal bovine serum, 100 U/mL penicillin and 100 g/mL streptomycin, 4 mM glutamine, 10 mM (4-(2-hydroxyethyl)-1-piperazine ethanesulfonic acid), 1 mM sodium pyruvate, and 1% nonessential amino acids, and kept at 37^°C in a humidified atmosphere with 5% CO₂. Cells were subcultured by 0.25% trypsin solution (containing ethylene diamine tetraacetic acid [EDTA]) every 3 days. Cell culture medium was replaced everyday. Cells from passages 22 to 60 were used for experiments.

Hydrogen treatment and measurement

The preparation and storage method for H₂-rich medium is in accordance with previous studies ^{(10, 18, 19)}. Briefly, H₂ was produced by a H₂ generator, and then was dissolved into DMEM by a special catheter for 4 h under high pressure (0.4 MPa) to a supersaturated level to generate H₂-rich medium. After production, H₂-rich medium was filtered to remove bacteria, and then immediately used for cell culture. The concentrations of H₂ in medium were detected by a needle-type Hydrogen Sensor (Unisense A/S, Aarhus, Denmark) immediately after H₂-rich medium preparation was finished (0.60 ± 0.03 mM). After 24 h incubation, H₂ concentration was again detected (0.23 ± 0.01 mM). For treatment, Caco-2 cells were cultured in H₂-saturated medium under a humidified condition of 75% H₂, 20% O₂, and 5% CO₂. The pH of the culture medium with or without H₂ were, respectively, measured (7.31 ± 0.03 and 7.44 ± 0.01, separately).

Detection of transepithelial electrical resistance (TER) and paracellular permeability

The Caco-2 cells were plated on transwell inserts (Millipore, Billerica, MA) with polyethylene terephthalate membrane (0.33 cm², 0.4 mm pores) at a density of 1 × 10⁵ cells/mL cultured for about 21 days when cells reached confluence and completely differentiated. A transepithelial voltohmeter (World Precision Instruments, FL) was utilized to measure TER according to previous description ⁽²⁰⁾. Both apical and basolateral sides of the transwell were thrice washed by Hank balanced salt solution (HBSS) before measurement. TER was measured until similar values were recorded for three consecutive measurements. The electrical resistance values were expressed as Ohm·cm² (Ω·cm²). Determinations were repeatedly operated on three different sites per transwell insert, respectively.

Caco-2 monolayer permeability was measured using FD4, an established paracellular marker. After washed with HBSS, the well-grown Caco-2 monolayers were incubated with serum-free DMEM without phenol red containing 10 mg/mL FD4 in the presence or absence of different mediators as indicated below. After 2 h incubation of FD4, paracellular flux was assessed by taking 100-mL aliquots from the outer chamber. Fluorescence was measured using fluorescence spectrophotometer (Hitachi, Tokyo, Japan), with excitation and emission at 485 and 535 nm, respectively. For all experimental conditions, the permeability coefficient (P_E) was calculated by the following formula: P_E = [(ΔC_A/Δt) × V_A]/(S × ΔC_L), where P_E = diffusive permeability (cm/s), ΔC_A = change of FD4 concentration, Δt = change of time, V_A = volume of the abluminal medium, S = surface area, and ΔC_L = constant luminal concentration ⁽⁷⁾.

Measurement of cell viability

Cell Counting Kit-8 (CCK8) (Dojindo, Tokyo, Japan) assay was introduced to detect Caco-2 cell viability according to manufacturer’s instructions. This assay is based on conversion of Dojindo highly water-soluble tetrazolium salt-8 (WST-8) to formazan by dehydrogenases. Water solubletetrazolium salts-8 is reduced by dehydrogenases in cells to give a yellow product (formazan), which is soluble in tissue culture medium. The amount of formazan dye generated by the activity of dehydrogenases in cells is directly proportional to the number of live cells. In brief, Caco-2 cells (1 × 10⁵ cells) were seeded on a 96-well plate. After 24 h LPS or H₂ stimulation, 10 μL cell counting kit-8 (CCK-8) solution was added to each well, and the reaction was allowed to occur in standard culture condition for 1 h. Absorbance was measured at 450 nm using a precision microplate reader (Ke Hua Biology, Shanghai, China).

Detection of intracellular reactive oxygen species (ROS)

Intracellular levels of ROS were detected by using a commercial kit (Beyotime, Shanghai, China) according to the manufacturer’s instructions. In brief, after treatment, Caco-2 cells cultured in 96 well plates were stained by 2′,7′-dichlorodihydrofluorescein diacetate, a ROS-sensitive probe, for 30 min at 37°C, after which a precision microplate reader (Ke Hua Biology, Shanghai, China) was utilized to measure intracellular ROS level at 488 nm (excitation) and at 525 nm (emission). A positive control was conducted in per experiment via stimulating Caco-2 cells with H₂O₂ before the addition of 2′,7′-dichlorodihydrofluorescein diacetate. The results were expressed as fluorescence intensity per mg protein and compared to the relative controls.

Western blot analysis

Caco-2 cells (1 × 10⁵ cells) were plated in 6-well plates. Once the experiment was finished, cells were rapidly rinsed with ice-cold PBS, and then were lyzed with radio immunoprecipitation assay (RIPA) lysis buffer (Solarbio, Beijing, China), and scraped. Cell lysates were centrifuged in a Centrifuge (Kubota, Osaka, Japan) to yield clear lysate, and then the supernatant was collected. The protein was quantified using a BCA Protein Assay kit (Beyotime, Shanghai, China). The equal amounts of protein samples were separated by SDS-PAGE, and then transferred to polyvinylidene fluoride (PVDF) membranes (Millipore, Billerica, MA). The membranes were incubated for 2 h in Tris-buffered saline with Tween 20 (TBST) buffer containing 5% skim milk for blocking, and then incubated overnight at 4^°C with appropriate primary antibodies: a 1:500 dilution of a rabbit monoclonal anti-occludin antibody (Invitrogen, Carlsbad, CA), a 1:1,000 dilution of a rabbit monoclonal anti-E-cadherin antibody (Cell Signaling Technology, Boston, MA), a 1:1,000 dilution of a rabbit monoclonal anti-mDia1 antibody (Abcam, Cambridge, UK), and a 1:2,000 dilution of a mouse monoclonal anti-β-actin antibody (Boster, Wuhan, China). After being rinsed in TBST buffer, the membrane was incubated for 1 h at room temperature with horseradish peroxidase (HRP)-conjugated goat antirabbit or antimouse secondary antibodies (Boster, Wuhan, China). Protein bands were detected using enhanced chemiluminescence (ECL) reagent (Millipore, Billerica, MA), and then visualized and photographed using Gel quantitative Quantity One system (BIO-RAD, Tokyo, Japan). The whole western blot analyses were performed at least three times. Each protein level was normalized to β-actin, respectively.

Immunofluorescence

Caco-2 cells (1 × 10⁴ cells) were plated on coverslips and cultured. Once the experimental period was completed, cells were washed three times by cold PBS, and then were fixed (4% formaldehyde in PBS) for 15 min and permeabilized (0.5% Triton X-100 in PBS) for 10 min at room temperature. Caco-2 cells were blocked in blocking solution composed of 10% normal goat serum (Solarbio, Beijing, China) for 1 h, and then were incubated at 4^°C overnight with appropriate primary antibodies: 1:250 antirabbit occludin (Invitrogen, Carlsbad, CA), 1:500 antirabbit E-cadherin (Cell Signaling Technology, Boston, MA), and 1:500 anti-mDia1 (Abcam, Cambridge, UK), followed by FITC or (tetramethyl rhodamin isothiocyanate)–TRITC conjugated secondary antibodies (Boster, Wuhan, China) incubation and DAPI (4′,6-diaminidino-2-phenylindole; Sigma-Aldrich, St Louis, Mo) counterstaining. The stained proteins were visualized and images were obtained under a confocal fluorescence microscope (Leica, Wetzlar, Germany).

RhoA activity detection

For measurement of RhoA activation, the respective G-Lisa Activity Assay Biochem Kit (Cytoskeleton, Denver, CO) was used according to manufacturer’s recommendations as described previously ⁽²¹⁾. Briefly, after incubation in presence or absence of different mediators, Caco-2 cells were washed with cold PBS. Ice-cold cell lysis buffer was added, and cell lysates were harvested by centrifugation at 14,000 rpm at 4^°C for 2 min. After rewarming, 50 μL lysate were added to wells of the GTPase binding plate coated with RhoA-GTP–binding domain. Other wells were added using lysis buffer or nonhydrolyzable RhoA as a negative or positive control, respectively. The plate was shook by a cold orbital shaker (Thermo Scientific, Rockford, IL) at 400 rpm at 4^°C for 30 min. After the plate was washed thrice, RhoA primary antibody diluted at 1:200 was added for 45 min incubation. Then secondary HRP-labeled antibody at 1:100 dilution was added for 45 min, after which HRP detection reagent were added and incubated for 15 min at 37^°C. Then HRP stop buffer was added. Finally, the signaling was immediately detected at 490 nm using a precision microplate reader (Ke Hua Biology, Shanghai, China).

RNA interference

Lentiviral vectors containing the human-specific shRNA against mDia1 sequence or the nonsilencing shRNA sequence were purchased from GeneCopoeia (Rockville, MD). Virus was produced using Lenti-Pac lentivirus packaging kit (GeneCopoeia), according to the manufacturer’s instructions. Caco-2 cells were transfected 24 h after seeding, after which the infected cells were selected using puromycin (Solarbio, Beijing, China) for 48 h. Individual cells still growing in the presence of puromycin were isolated, and then were cultured and selected using puromycin to build a stable-transfected cell line.

Statistical analysis

SPSS statistical software 21.0 (IBM, Armonk, NY) was used. All data were expressed as means ± standard deviation (SD). The statistical significance of differences between groups was determined by one-factor analysis of variance (ANOVA), followed by the least significant difference (LSD) t test for multiple comparisons. A two-tailed P value of less than 0.05 was considered statistically significant.

RESULTS

Effects of LPS at different concentrations on Caco-2 intestinal epithelial permeability

In the following studies, the effects of different concentrations of LPS (0, 1, 10, and 100 μg/mL, and 1 mg/mL) on gut barrier permeability was determined by measuring TER and mucosal-to-serosal flux rates of FD4 in filter-grown Caco-2 intestinal monolayers for up to 48 h. Increasing concentrations above the 100 μg/mL (including 100 μg/mL) caused a dose-dependent drop in TER, whereas concentrations below 100 μg/mL produced no significant decrease (Fig. 1A). Exposure of Caco-2 monolayer to 100 μg/mL or 1 mg/mL LPS resulted in a sharp time-dependent decrease in TER between 3 and 24 h, and did not induced significant drop between 24 and 48 h (Fig. 1A). Accordingly, 100 μg/mL and 1 mg/mL LPS also, respectively, caused a time-dependent increase in the flux of FD4 (Fig. 1B). These results indicated that LPS at concentrations of 100 μg/mL or above induce Caco-2 intestinal barrier dysfunction.

Previous studies showed that an integrated intestinal barrier function can be destroyed either by leading to enterocyte death or simply by breaking AJC ^(22–24). To exclude the possibility that epithelial barrier breakdown in response to LPS resulted from major cell death instead of specific modulation of intercellular junction properties, we next investigated whether 100 μg/mL or 1 mg/mL LPS had a cytotoxic effect on Caco-2 cells. The effect of LPS on Caco-2 cell viability was determined over a 24-h experimental period by CCK8 assay. The results showed that 100 μg/mL LPS did not affect cell viability of Caco-2 cells, in contrast to which LPS at 1 mg/mL was high enough to have a cytotoxic effect on Caco-2 cells (Fig. 1C). Together, these results demonstrated that LPS at 100 μg/mL did not result in cytotoxicity and that the abnormal Caco-2 monolayer permeability is not due to cell death.

H₂ prevents the TER decline and FD4 flux increase induced by LPS in Caco-2 monolayers

In consideration of the pathological effects of LPS on intestinal epithelial barrier without Caco-2 cytotoxicity, 100 μg/mL LPS was applied for the following studies. To determine the impacts of H₂ on barrier-disrupting effects of LPS, Caco-2 monolayer models were incubated with both H₂-rich medium and LPS simultaneously for 24 h—a time point when 100 μg/mL LPS could extremely cause barrier breakdown. As shown in Figure 2, A and B, in comparison with the group exposed to LPS alone, H₂-rich medium markedly alleviated pathological TER decrease and FD4 flux increase, suggesting a definite role of H₂ in Caco-2 gut barrier protection.

Effects of H₂ on oxidative stress in LPS-stimulated Caco-2 cells

Since oxidative stress acts as an important factor for barrier injury, we next investigated the levels of ROS in Caco-2 cells after the administration of LPS and H₂-rich medium. Results in Figure 3 showed that 100 μg/mL LPS up-regulated the intracellular levels of ROS. H₂-rich medium partly reduced intracellular ROS production (Fig. 3), suggesting the benefit of H₂ for Caco-2 barrier is associated with its antioxidant effect.

H₂ relieves expression and structure changes in epithelial apical junctions induced by LPS

The gastrointestinal paracellular barrier is achieved by intercellular junctional structures such as TJ and AJ proteins. Changes in TJ and AJ can lead to perturbations of paracellular permeability. In particular, TJ proteins such as occludin, and AJ proteins such as E-cadherin, play a major role in barrier regulation. Alterations in the expression of these proteins have been constantly discovered in gut barrier dysfunction. We therefore evaluated the effects of LPS and H₂ on expression levels and structures of these junctional proteins. Well grown Caco-2 cells were incubated for three periods (6, 12, and 24 h) with LPS. Western blot revealed that, in line with the time-dependent changes in TER and FITC-dextran flux, the expression levels of occludin and E-cadherin were all down-regulated upon exposure to LPS at these three time points, and the most remarkable changes occurred at 24 h incubation (Fig. 4, A–I). Simultaneous cotreatment with H₂-rich medium partly restored alterations of these TJ and AJ proteins, whereas H₂ exposure alone did not affect their abundance (Fig. 4, A–I).

Next we investigated whether the structures of these AJCs suffered the parallel alterations or not. Immunofluorescence staining of normal Caco-2 monolayers showed that both occludin and E-cadherin were regularly distributed along the cell borders, presenting a typical “chicken wire” labeling pattern (Fig. 5). In line with the aggravated decline in expression levels, stimulation of Caco-2 with LPS for 24 h profoundly disrupted architectures of both occludin and E-cadherin by interrupting continuous band pattern and reorganizing the distribution of these proteins (Fig. 5). Concurrent administration of H₂-rich medium substantially reversed these LPS-induced alterations (Fig. 5).

H₂ alleviates LPS-caused Caco-2 barrier dysfunction by mediating the activation of RhoA

On the basis of accumulating evidences that RhoA GTPase acted as a major regulator for the AJC formation and disassembly, impacts of both LPS and H₂ on the RhoA activation were examined by G-Lisa assay ^(25–27). Results shown in Figure 6A revealed that LPS administration for 24 h dramatically activated GTP-binding RhoA in Caco-2 cells, whereas simultaneous addition of H₂-rich medium effectively calmed down GTP-RhoA activity and kept it at a lower but still activated level, implying a potential role of RhoA in H₂-mediated intestinal barrier protection.

To further evaluate involvements of RhoA in H₂-induced barrier benefits, both RhoA activator CN03 and Rho inhibitor C3 exoenzyme were applied, respectively. Caco-2 monolayers were preincubated with 1 μg/mL CN03 for 3 h before cotreatment of LPS and H₂-rich medium. Figure 6B and C demonstrated that CN03 addition remarkably counteracted the beneficial effects of H₂ on TER and FD4 flux. In addition, pretreatment with 2.5 μg/mL C3 exoenzyme for 1 h before LPS stimulation mitigated the pathological permeability disrupted by LPS (Fig. 6, B and C). Thus, these observations suggested that H₂ may exert protective effects on LPS-induced barrier dysfunction of Caco-2 monolayers via down-regulating RhoA activation.

H₂ attenuated LPS-induced disruptions of the epithelial AJC by suppressing RhoA activity

Since RhoA was identified as playing a pivotal role in H₂-induced restoration of Caco-2 barrier function, we further investigated involvements of RhoA in H₂-mediated protection on TJ and AJ. Western blot analysis showed that, in agreement with functional changes in barrier permeability, robust RhoA activation by CN03 largely abrogated benefits for occludin and E-cadherin expression exerted by H₂ upon LPS stimulation, and RhoA inhibition by C3 exoenzyme alleviated LPS-caused down-regulated expressions of TJ and AJ (Fig. 7, A–C). Likewise, immunofluorescence in Figure 8 revealed that CN03 preadministration observably eliminated H₂-modulated reconstruction of continuous and integrated band pattern of occludin and E-cadherin, and C3 exoenzyme preaddition reversed LPS-induced abnormal alterations in structures of occludin and E-cadherin, hinting that RhoA is necessary for the AJC protection by H₂. On the basis of the above discussion, a perspective was upheld that H₂ down-regulates RhoA activity to a mild level, therefore providing antihyperpermeability effects in the presence of LPS.

mDia1, a downstream of RhoA, is required for H₂ to counteract LPS-induced barrier dysfunction and the AJC disruptions

In search for downstream effectors of H₂/RhoA, we next investigated the potential role of a typical RhoA target mDia1, which assembles actin filaments and modulates microtubule dynamics to establish and maintain intercellular junctional complex ^{(8, 9)}. First we noticed that LPS stimulation alone decreased mDia1 expression in Caco-2 cells (Fig. 9, A and B). The concurrent treatment with LPS and H₂-rich medium, which maintained mild RhoA activation, distinctly improved the abundance of mDia1 (Fig. 9, A and B). Then we investigated whether mDia1 participated in H₂-exerted interference with intestinal barrier dysfunction caused by LPS. As is shown in Figure 9D and E, knocking down of mDia1 by siRNA restored the ability of LPS to wreck the normal TER and FD4 flux of the Caco-2 monolayers, which were previously blocked by H₂. Caco-2 barrier function was also impaired by mDia1 knockdown without LPS stimulation, suggesting the necessity of mDia1 for barrier maintenance. Furthermore, the down-expression of mDia1 eliminated the effects of H₂-rich medium on expression levels of occludin and E-cadherin (Fig. 10, A–C). Accordingly, mDia1 knockdown counteracted the beneficial effects of H₂ on structures of TJ and AJ (Fig. 11). Single mDia1 knockdown without LPS treatment destroyed both TJ and AJ (Fig. 10, A–C and Fig. 11). Together, these findings support an emerging notion that, in an in vitro Caco-2 monolayer barrier, H₂ can modulate LPS-stimulated robust RhoA activation to a down-regulated but still moderately activated level, which next increases mDia1 expression level, thereby preventing the destruction of both TJ and AJ, and enhancing the intestinal epithelial barrier function.

DISCUSSION

The gut epithelium offers a crucial physical barrier against the access of hostile substances from the external environment. TJs and AJs, collectively known as the AJC, seal the paracellular gaps between adjacent enterocytes, and thereby form multifunctional structures, guarding the internal environment against luminal components ^{(3, 5)}. Intestinal epithelial barrier dysfunction is widely accepted as a major origin or final pathway contributing to sepsis, septic shock, and MODS ⁽²⁸⁾.

Lipopolysaccharide—a main component from cell wall of Gram-negative bacteria—is one of the most potent innate oxidative stress-activating stimuli ⁽²²⁾. Previous experiments have discovered that, when infected by LPS,, the intestinal barrier becomes pathologically hyperpermeable ^{(22, 23)}. The human intestinal Caco-2 cell line, which is originally obtained from human adenocarcinoma and expresses typical morphological and functional features of mature enterocyte after differentiation, has been widely utilized as a validated in vitro model for the gut barrier ⁽²⁹⁾. Previous experiments have revealed that, different doses of LPS show distinct effects on Caco-2 monolayers ^{(22, 28)}. In the present study, we reconfirmed that LPS at a concentration of 100 μg/mL observably showed its barrier-breaker role by disturbing the expressions and distributions of TJ and AJ, without affecting Caco-2 viability. Actually, the physiologically and clinically relevant concentration of LPS is about 0 to 10 ng/mL, and LPS at 1 μg/mL can markedly activate cultured cells ⁽²²⁾. However, the present study supported that LPS at lower concentrations (0–10 μg/mL) hardly affected barrier function.

Nowadays, H₂ has been recognized as a medical gas with promising potentials to prevent or cure a series of diseases, such as ischemia/reperfusion injury, neurodegeneration, metabolic syndrome, inflammation, mitochondrial diseases, and even cancers ^{(11, 30, 31)}. Our researches published previously have revealed that 2% or 4% H₂ inhalation effectively improves survival rates of septic animals and relieves sepsis-related organ injuries ^{(12, 13)}. However, studies about the hypothesis whether these mentioned protections against sepsis induced by H₂ is in relation to the intestine have seldom been reported. In fact, H₂ administration with different methods has been proved definitely effective for several intestinal diseases ^{(15, 16, 32)}. Consistent with these benefits for the intestine, in the present study, we proved the protective role of H₂-rich medium in LPS-disrupted gut barrier model in vitro. The excessive oxidative stress has been reported to be associated with gut barrier dysfunction ⁽³³⁾. In consideration of antioxidant effects of H₂, we examined the oxidative stress, and confirmed that H₂ could lower intracellular ROS production, hinting that the advantages for gut barrier by H₂ may be related to antioxidation.

Hydrogen gas can also protect against sepsis by modulating many signal pathways ^{(14, 20, 30)}. Therefore, experiments were conducted to elucidate mechanisms by which H₂ modulated Caco-2 barrier function. Since the present results revealed involvements of RhoA-mDia1 in LPS-stimulated gut barrier breakdown, H₂ effects on this signal pathway were explored. Our current results found that H₂ suppressed RhoA activity. Furthermore, RhoA activator eliminated the protections of H₂ on barrier integrity and AJC, and RhoA inhibitor reversed LPS-disrupted gut barrier, suggesting the protective effects of H₂ against barrier dysfunction is dependent on the decrease of RhoA activity. In physiological conditions, TJ and AJ undergo dynamic regulations between assembly and disassembly, and appropriate Rho activity plays a crucial role in this dynamics, thereby maintaining the AJC at a stable state ^{(8, 34)}. Instead of suppressing RhoA activity below a physiological degree, H₂ kept it at a moderately activated level in LPS-stimulated Caco-2 cells, which may explain how H₂ served as a rescuer for gut permeability.

Previous investigations have indicated that the complicated connections between RhoA and the AJC can be elucidated by two different downstream effectors, Rho-associated coiled-coil protein kinase (ROCK) and mDia1 ⁽⁶⁾. Under physiological circumstance, ROCK promotes actomyosin contraction and actin polymerization, whereas mDia1 facilitates actin polymerization and microtubule organization ^{(6, 9)}. Sahai et al. confirmed that RhoA activation in low degree selectively transmits signals through mDia to stabilize the AJC, whereas robust RhoA activity tends to favor ROCK-dependent AJ destruction ⁽⁸⁾. Gavard et al. also found that both intensity and space distribution of active RhoA impact options of downstream signaling ⁽³⁵⁾. Additionally, our previous findings have demonstrated that H₂-rich medium decreases expression of ROCK to attenuate LPS-caused vascular endothelial hyperpermeability and vascular endothelial-cadherin disruption ⁽²⁰⁾. Thus, in the present study, we supposed whether an alternative downstream target, mDia1, participated in mild RhoA activation-dependent protection of intestinal barrier by H₂-rich medium. Results showed that, H₂ could effectively increase mDia1 expression in LPS-exposed Caco-2 cells, whereas H₂ treatment alone did not affect mDia1, conforming mDia1 as a downstream of H₂-modulated RhoA. Further, in mDia1-interfered Caco-2 monolayers, H₂ lost its protective effects on gut barrier permeability, and both expressions and structures of occludin and E-cadherin were again disrupted by endotoxin, suggesting that H₂-induced advantages for intestine was mDia1-dependent.

However, there are two limitations in this study. First of all, 100 μg/mL LPS used in the present study is so high that this concentration could be unlikely achieved in vivo. In addition, the employment of immortalized cancer cell line as an in vitro gut model is another limitation in this study. On the basis of these two limitations, an in vivo model of the LPS-induced increase in mouse intestinal permeability as previously described should be introduced for further investigation ⁽²²⁾.

To sum up, our researches illuminated that, in Caco-2 monolayers, 100 μg/mL LPS caused pathological RhoA signal activation which then damaged barrier permeability, and H₂-rich medium could stabilize RhoA activity at a mild but still active level to increase mDia1 expression, therefore mitigating disruptions of cell–cell junctions and enhancing the intestinal epithelial barrier function. These findings may provide a possible mechanism for curative effects of H₂ on sepsis.