Different Structures of Arabinoxylan Hydrolysates Alleviated Caco-2 Cell Barrier Damage by Regulating the TLRs/MyD88/NF-κB Pathway

Arabinoxylan (AX) has been associated with alleviating intestinal barrier damage, and different structures of AX give rise to different effects on the intestinal barrier. This study investigated the main structural characteristics of AX, whose functional properties are attributed to alleviating intestinal barrier damage, and clarified their underlying mechanisms. An in vitro Caco-2 cell model was established to investigate the intestinal barrier effects of AX with various degrees of substitution (Ds) and molecular weight (Mw), with an added MyD88 inhibitor to verify the signaling pathways. Arabinoxylan treated with endo-1,4-β-xylanase (AXX) with higher Ds and Mw showed stronger physiological activity, which might be correlated with the uronic acid and bound ferulic acid contents in AXX. Moreover, AXX alleviated the intestinal barrier damage by upregulating the transepithelial electrical resistance (TER) and alleviating the decrease of claudin-1 (p < 0.05). AXX regulated the expression of inflammatory factors IL-2, TNF-α, IL-6 and IL-10 (p < 0.05). In addition, AXX reduced the intestinal barrier damage induced via inhibiting the TLRs/MyD88/NF-κB pathway and activating the TLRs/PKC pathway. Thus, AX with higher Ds and Mw might be better in alleviating intestinal barrier damage, and MyD88 might be the key point of AXX to identify these signaling pathways.


Introduction
The integrity of epithelial cell and tight junction (TJ) proteins is often referred to as the "intestinal barrier" [1,2]. The intestinal epithelial barrier plays a key role in preventing the transfer of pathogenic microorganisms and toxic substances from the intestinal lumen to the systemic circulation [3]. The loss of barrier integrity can lead to the invasion of harmful substances in the intestinal lumen, which can cause different diseases such as inflammatory bowel disease and extraintestinal autoimmune diseases [4,5]. TJ connectivity plays an important role in maintaining the structural and functional integrity of the intestinal barrier by regulating paracellular transport. In recent studies, increased intestinal permeability due to decreased TJ-related protein (such as zonula occludens family proteins, occludin and claudins) expression and epithelial barrier function was observed [6].
Toll-like receptors (TLRs) are a class of transmembrane proteins that trigger cell signal transduction. Toll-like receptor 2 (TLR2) is a potential therapeutic target for gastrointestinal diseases, in which TJ-related intestinal epithelial barrier disruption might be the main feature [7,8]. Toll-like receptor 4 (TLR4) is the best characterization protein of pathogen recognition receptors and plays a homeostasis role in maintaining TJ protein expression and affecting intestinal permeability [9]. Myeloid differentiation factor 88 (MyD88) is the core of transduction of extracellular stimulation via TLRs. After specific recognition, intracellular signal transduction mainly depends on MyD88-dependent and MyD88-independent 2.3.3. 1

H NMR Analysis
The analysis was conducted in accordance with a previous article [23]. Here, after exchange with D 2 O, the samples were finally dissolved in pure D 2 O. A 400.00 MHz spectrometer (JNM-ECZ400S, JEOL, Tokyo, Japan) was used to obtain the 1 H spectra of samples. The 1 H chemical shifts (ppm) were referenced to a D 2 O signal at 4.790 ppm at 25 • C.

FT-IR Analysis
The analysis was conducted in accordance with a previous article [22]. Fourier transform infrared spectra of freeze-dried samples were recorded at room temperature with an FTIR spectrometer (Thermo Fisher Scientific, Waltham, MA, USA).

Ferulic Acid Determination
FA content was measured by spectrophotometry [24]. The sample was dissolved in 1 mL of ultrapure water. A quantity of 900 µL of glycine-NaOH buffer (pH 10, 0.04 M) was added to 100 µL of the sample solution. The absorbance values were at 345 and 375 nm.

Establishment of the Cell Model 2.4.1. Cell Culture
Caco-2 cell culture was performed based on Fang, et al. [25]. Human colon cells (Caco-2) were selected as the research object. The Caco-2 cell line (ATCC, Manassas, VA, USA), derived from human colon cancer cells, in DMEM containing 10% fetal bovine serum, 50 U/L penicillin-streptomycin and 1% non-essential amino acid at 37 • C, 5% CO 2 highsugar medium, was cultured in a 25 cm 2 cell culture flask, replacing the culture medium every 1-2 days. After 5-7 days, the cells had grown to confluence and passaged at 1:2 or 1:3.

Treatment of Cells with Arabinoxylan Hydrolysates
The treatment was performed based on Wu, et al. [26]. The cells were starved for 12 h before the experiment on serum-free media, and the treatment groups were added with AXHs (400 µg/mL) and different concentrations of lipopolysaccharide (LPS, µg/mL) after 1 h preincubation. The control group (without LPS and AXH), negative control group (with LPS) and treatment groups (with LPS and AXH) were set. The resistance values of each group were detected after 24 h of culture, and the experiment was repeated three times.

Treatment of Cells with MyD88 Inhibitor
The treatment was performed based on Song, et al. [27]. The cells were starved for 12 h before the experiment (adding serum-free medium), and all groups were added with MyD88 inhibitor. After 1 h, the AXHs with the best screening effect was added (µg/mL). After 2 h, LPS (µg/mL) was added. The control group (no LPS and no AXH), negative control group (only LPS), treatment group I (LPS and AXH, no MyD88 inhibitor) and treatment group II (LPS, the best effect AXH and MyD88 inhibitor) were set. The resistance values of each group were detected after 24 h of culture, and the experiment was repeated three times.

Determination of Epithelial Monolayer Resistance
Determination was conducted with the method illustrated by Wu, et al. [28]. The cells were starved for 12 h before the experiment (adding serum-free medium), pre-incubated for 1 h and then added with LPS (µg/mL). Solvent was used as the control group, and the resistance value of each group was detected after 24 h of culture. The short and long arms of the cell resistor electrode were inserted into the upper and lower chambers of the transwell chamber, respectively. Three different points of the chambers were measured, and the average value of the measured values was taken.

Assessment of Protein Expression by Western Blot
Assessment by Western blot was performed according to Wu, et al. [29]. The Caco-2 cell culture medium was aspirated and rinsed with PBS twice. The cells were digested by trypsin, and the pipetted cells were transferred into a 1.5 mL EP tube. A total of 150 µL of RIPA lysate containing PMSF was added to the centrifuge tube and centrifuged at 4 • C, and the supernatant after centrifugation was prepared with 10% polyacrylamide gel to prepare SDS-PAGE gel. About 50 µg of protein sample was prepared and loaded. Firstly, electrophoresis was performed at a constant voltage of 80 V and then 100 V. The protein in the gel was transferred to the PVDF membrane by transfer membrane and blocked with TBST buffer containing 5% skimmed milk powder. Then, the primary antibody was incubated overnight at 4 • C. The secondary antibody was incubated at room temperature for 1 h. ImageJ image processing software was used to analyze the Mw and net optical density value of the target band. The average of five experimental results was regarded as the relative protein content. β-actin was the internal reference gene.

Quantification of Gene Expression Using Real-Time PCR
The cells were collected and washed with PBS buffer three times. Then, 1 mL of Trizol reagent was added to each well to extract the total RNA of the cells. RNA concentration and purity were detected and reversed transcription was synthesized into cDNA. The target gene sequence was acquired from the NCBI biological information website, and Primer 5 primer design software was used to construct the gene primers. A 10 µL SYBR Green reaction system was used for RT-PCR detection, β-actin and GADPH were used as internal reference genes for calculation, and the target gene level was expressed in terms of relative expression. The calculation method of the results was reported in the study of Pfaffl, et al. [30].

Enzyme-Linked Immunosorbent Assay
The culture medium was discarded. The cells were collected and washed three times with PBS buffer. An ELISA kit was used to determine the cytokines IL-2, IL-6, IL-10 and TNF-α in the Caco-2 cell culture medium content. The determination method was carried out in accordance with the ELISA kit's instructions.

Statistical Analysis
All data were entered into Excel 2019. OriginPro 9.1 was used for mapping. IBM SPSS Statistics 26 mathematical software was used for statistical analysis of experimental data, and analysis of variance (ANOVA) was used to verify the statistical differences between groups. The difference was statistically significant with p < 0.05, and the results were expressed as mean ± standard deviation.  Table 1. The monosaccharide composition of AXH was mainly composed of xylose and arabinose, following with glucose and galactose. Different AXHs had a different ratio of arabinose to xylose and AX X and AX AE contained more arabinose. Furthermore, the Mw of AXH acquired by endo-1,4-β-xylanase treatment was higher than that acquired by α-L-arabinofuranosidase treatment. The Mw is mainly distributed between 2.67 × 10 3 Da to 6.43 × 10 5 Da. Furthermore, in the acid-treated AX C1 and AX H2 , the Mw of AX H2 was lower than that of AX C1 .

1 H NMR Analysis of Arabinoxylan Hydrolysates
The structural change of AXH was verified by 1 H NMR. The NMR spectra of all samples showed 3-6 main peaks in the chemical shift range of 5.15-5.40 ppm. The signal peak of the arabinose residue was connected to the C(O)-3 position of the xylose residue on the main chain at a chemical shift of 5.38 ppm. The signal peak around 5.27 and/or 5.21 ppm corresponding to xylose residues was monosubstituted by arabinose residues at the C(O)-2 and/or C(O)-3 position. The chemical shift of xylose residues was monosubstituted at the C(O)-3 and/or C(O)-2 position at 4.50 and/or 4.60 ppm. The chemical shift of xylose residues was disubstituted at 4.62 ppm. As shown in Figure 1, the contents of xylose residues in AX H2 and AX AE were higher than those in AX X and AX C1 .

FT-IR Spectrum Analysis of Arabinoxylan Hydrolysates
As shown in Figure 2, a broad and strong peak appeared near 3410 cm −1 . The range of 3000-2800 cm −1 were produced by C-H vibration. The band between 1700-1600 cm −1 was the C=O asymmetric stretching vibration in COO-. The absorption band between 1400-1300 cm −1 was the C=O symmetric stretching vibration in COO-. Frequency band signals between 1200 cm −1 and 800 cm −1 were associated with a specific polysaccharide structure composed of pyranose ring vibration, which overlapped with C-C and C-OH stretching vibration and glycosidic bond (C-O-C) vibration. AX AE had a relatively strong absorption peak at 1074.50 cm −1 . These were the typical structural features of AX.

Ferulic Acid Content Analysis
FA was a major bound phenolic acid linked to AX mainly with ester bounds, which affected its physical and chemical properties. Free FA content was higher than that of bound FA (Figure 3). Bound FA content in AX X was higher than that acquired by acid and α-L-arabinofuranosidase treatment (p < 0.05).

FT−IR Spectrum Analysis of Arabinoxylan Hydrolysates
As shown in Figure 2, a broad and strong peak appeared near 3410 cm −1 . Th of 3000-2800 cm −1 were produced by C−H vibration. The band between 1700-16 was the C=O asymmetric stretching vibration in COO-. The absorption band b 1400-1300 cm −1 was the C=O symmetric stretching vibration in COO-. Frequen signals between 1200 cm −1 and 800 cm −1 were associated with a specific polysac structure composed of pyranose ring vibration, which overlapped with C-C an stretching vibration and glycosidic bond (C-O-C) vibration. AXAE had a relativel absorption peak at 1074.50 cm −1 . These were the typical structural features of AX.

Ferulic Acid Content Analysis
FA was a major bound phenolic acid linked to AX mainly with ester bound affected its physical and chemical properties. Free FA content was higher than bound FA (Figure 3). Bound FA content in AXX was higher than that acquired by acid and α−L−arabinofuranosidase treatment (p < 0.05).

Transepithelial Electrical Resistance Measurement
About 1 μg/mL of LPS was used as the optimal LPS emergency concentration for ou study. The alleviating effect of each sample on LPS−induced intestinal epithelial barrie integrity destruction was monitored by transepithelial electrical resistivity (Figure 4). Ou research showed that after treatment with the LPS group, the resistivity of Caco-2 intesti nal epithelial cells decreased by 9%. The TER value of the LPS + AXX group was highe than the LPS + AXC1 group (407.12 ± 34.08 Ω/cm 2 ), and there was no significant differenc among other groups.

Transepithelial Electrical Resistance Measurement
About 1 µg/mL of LPS was used as the optimal LPS emergency concentration for our study. The alleviating effect of each sample on LPS-induced intestinal epithelial barrier integrity destruction was monitored by transepithelial electrical resistivity (Figure 4). Our research showed that after treatment with the LPS group, the resistivity of Caco-2 intestinal epithelial cells decreased by 9%. The TER value of the LPS + AX X group was higher than the LPS + AX C1 group (407.12 ± 34.08 Ω/cm 2 ), and there was no significant difference among other groups. bound FA (Figure 3). Bound FA content in AXX was higher than that acquired by acid and α−L−arabinofuranosidase treatment (p < 0.05).

Transepithelial Electrical Resistance Measurement
About 1 μg/mL of LPS was used as the optimal LPS emergency concentration for our study. The alleviating effect of each sample on LPS−induced intestinal epithelial barrier integrity destruction was monitored by transepithelial electrical resistivity (Figure 4). Our research showed that after treatment with the LPS group, the resistivity of Caco-2 intestinal epithelial cells decreased by 9%. The TER value of the LPS + AXX group was higher than the LPS + AXC1 group (407.12 ± 34.08 Ω/cm 2 ), and there was no significant difference among other groups.

Tight Junction Protein Expression in Caco-2 Cells
Claudin-1, occludin and zonula occludens (ZO)-1 were down-regulated under LPS conditions (p < 0.05), and this effect was changed by the different structures of AXHs ( Figure 5A). Compared with LPS, AX X increased the expression of claudin-1 (p < 0.05), and there was no significant expression of occludin and ZO-1 in the AX X group (p > 0.05). As shown in Figure 5B, the expression of claudin-1 was down-regulated in the LPS + AX X + MyD88 inhibitor group (p < 0.05), whereas there was no significant change in occludin and ZO-1 (p > 0.05).

Tight Junction Protein Expression in Caco-2 Cells
Claudin−1, occludin and zonula occludens (ZO)−1 were down−regulated under LPS conditions (p < 0.05), and this effect was changed by the different structures of AXHs (Figure 5A). Compared with LPS, AXX increased the expression of claudin−1 (p < 0.05), and there was no significant expression of occludin and ZO−1 in the AXX group (p > 0.05). As shown in Figure 5B, the expression of claudin−1 was down−regulated in the LPS + AXX + MyD88 inhibitor group (p < 0.05), whereas there was no significant change in occludin and ZO−1 (p > 0.05).

Signaling Pathway in Caco-2 Cells
As shown in Figure 6A,B, compared with the LPS group, the AX X and AX H2 groups up-regulated the expression of TLR2 and TLR4 (p < 0.05). After MyD88 inhibitor was added, the expression of TLR2 and TLR4 was down-regulated compared with the LPS group (p < 0.05). The LPS + AX X group up-regulated the expression of MyD88 compared with the LPS group, whereas this change decreased in the MyD88 inhibitor group (p < 0.05) ( Figure 6C). Figure 6D,E shows that LPS-induced PKC and NF-κB expression in Caco-2 cells was upregulated and inhibited, respectively, compared with that in the AX X group (p < 0.05). In the LPS + AX X + MyD88 inhibitor group, NF-κB expression was up-regulated (p < 0.05), whereas PKC expression was not significantly changed compared with that in the AX X group (p > 0.05) ( Figure 6D,E).

Inflammatory Cytokines
As shown in Figure 7A-D, the LPS + AX X group down-regulated TNF-α, IL-6 and IL-10 expression and up-regulated IL-2 expression compared with the LPS group (p < 0.05). In addition, compared with the LPS + AX X group, the addition of MyD88 inhibitor upregulated TNF-α, IL-6 and IL-10 expression and down-regulated IL-2 expression (p < 0.05) ( Figure 7A-D).

Discussion
Our previous studies marked the effect of different dietary fibers on improving the intestinal barrier, and found that wheat bran fiber had the best effect [31]. Further studies found that the intestinal epithelial barrier function improvement exerted by wheat bran was related to its AX content [32]. AX with different structural characteristics exhibited different biological activities, and there was a strong correlation between the molecular structure and physiological functions of AX. In our previous study, AXHs with different structures were prepared by different methods [21]. According to their substitution degree and molecular weight difference, four representative AXHs were identified, including AX X with high Ds and high Mw, AX AE with high Ds and low Mw, AX C1 with low Ds and high Mw and AX H2 with low Ds and low Mw. The arabinose residue of AX performed different structures along the xylose backbone [23]. The 1 H NMR results showed that AXHs except AX AE had disubstituted arabinose residues, which indicated the role of α-L-arabinofuranosidase, and the higher A/X ratio of AX AE might be related to the region of monosubstituted arabinose residue and disubstituted xylose residues. In addition, AXHs except AX X had mono-and/or disubstituted xylose residues, indicating that endo-1,4-β-xylanase selectively removes xylose residues, and the higher degree of AX X could be confirmed (Ara/Xyl ratio: 1.17). Although the A/X ratio of AX AE and AX X were similar, the substitution patterns of these polymers were not the same. The A/X ratio of AX C1 was higher than AX H2 due to the former monosubstituted xylose residue. FT-IR spectrum results showed that the high Ds of arabinose at xylose residue C3 was indicated by the low-intensity peaks at 988.18 and 1154.83 cm −1 , which proved that AX AE had more branched structures and was consistent with the high Ds of AX AE . A strong absorption peak of AX X at 1650 cm −1 was associated with water in the sample, which might be related to the fact that AX X had a highly monoand/or disubstituted arabinose. The solubility of AX was reported to be determined by its structure, with a higher A/X ratio promoting its solubility [17]. Moreover, the solubilization of arabinose substituents might be due to the prevention of intermolecular aggregation of unsubstituted xylose residues.
Uronic acid was observed in AX X and AX C1 . Studies had shown that uronic acid polysaccharides activated MAPK and NF-κB signaling pathways by inducing the release of inflammatory cytokines. Therefore, AX with uronic acid might have more physiological properties [33,34]. Hromádková, et al. [35] revealed that the aggregation formed through the crosslinking of FA substituents might be responsible for the high Mw of AX, which was verified by the higher proportion of FA binding of AX X in our work. FA played an important role in the physiological functions of AX, especially in the antioxidant capacity of AX, but it contributed less to the effect of AX on the Caco-2 cell barrier. Zhang, et al. [16] suggested that FA bound to arabinose in AX to regulate intestinal barrier damage through gut microbiota. We speculated that AX X was more biologically active in the intestinal epithelial cell due to its unique structure. The molecular mechanisms behind this effect remains to be elucidated.
A decrease in TER was generally considered a reference index for cell damage or death, whereas the TER values might be restored and increased by LPS co-culture with AX X , which might be better for improving the permeability of Caco-2 cells. TJs surrounded and anchored adjacent cells. The depletion of TJ protein expression impaired intestinal barrier permeability and the ability to maintain TER [36]. Claudin-1, occludin and ZO-1 were three important TJ proteins. Claudin-1 interacted with other connexins in neighboring cells and formed fences that regulated the permeability of TJ complexes. Occludin could bind with adjacent cells through the outer part of the cell membrane, which conduced to control the permeability between cells. ZO-1 was mainly located at the boundary of adjacent epithelial cells, which anchored occludin and claudin-1 to the cytoskeleton and cocontrolled intestinal barrier permeability [37,38]. In this study, LPS reduced the expression of ZO-1, occludin and claudin-1. Claudin was offset by AX X . Claudin has been reported to largely determine the paracellular ion permeability at TJs [39]. We suggested that AX X maintained the integrity of the intestinal epithelial barrier by promoting the expression of claudin-1.
Pro-inflammatory cytokines (IL-1, IL-2, IL-6, etc.) had been proven to be interfering factors of the intestinal barrier. They lead to the further deterioration of intestinal epithelial function by inhibiting TJ protein expression [40]. In this study, AX X induced the expression of interleukin-2 (IL-2), inhibited the expression of interleukin-6 (IL-6), interleukin-10 (IL-10) and tumor necrosis factor-α (TNF-α) to enhance the intestinal immune function of the host, which was consistent with the study of Mendis, et al. [18]. These results suggested that AX X production played an important role in improving intestinal barrier and permeability by regulating TJ protein expression and inflammatory cytokine secretion.
Next, we tried to reveal the mechanism of AX X in alleviating the intestinal barrier in Caco-2 cells. Our previous study found that the up-regulation of TJ protein gene expression in intestinal epithelial cells by wheat bran dietary fiber was accompanied by the upregulation of TLR2 expression [31]. Dietary fiber inulin-type fructans could modulate TLR2 to enhance the intestinal barrier and prevented pathogens from entering the host [41]. TLR4mediated intestinal barrier dysfunction was considered to be a key factor in the initiation and enhancement of gastrointestinal injury [42]. In this study, the expression of TLR2 and TLR4 was up-regulated in the AX X and AX H2 groups, which suggested that structurally different AX had similarities in regulating intestinal barrier damage. AX X and AX H2 might induce cytokine production in intestinal epithelial cells in a TLR2-and/or TLR4-dependent manner. The similarity of AXH biologically affected with different Mw and Ds might be due to the similarity of glycosidic bonds and changes in the structure and number of arabinose substitutions in the xylose backbone [43]. In addition, we found that the infrared spectra of AX H2 and AX X were similar. However, according to the data of this study, AX X with high Ds might exert steric hindrance on the formation of intermolecular crosslinks and thus be easily dispersed into the reactive mixture to obtain a better performance.
MyD88 was an important signaling pathway regulating LPS-induced intestinal epithelial TJ permeability [44]. It had been widely reported that polysaccharides exerted their regulatory role by acting on MyD88. Based on these studies, a model with additional MyD88 inhibitor in the LPS + AX X group was proposed to explore and verify the signaling pathways by which AX X alleviated LPS-induced changes in barrier permeability. Our results confirmed that the expression of claudin-1 was affected by the addition of MyD88 inhibitor. The altered expression of inflammatory factors after MyD88 inhibitor suggested that this response might be related to MyD88-dependent TLR signaling. NF-κB activation might be an important pathway of TJ barrier dysfunction induced by pro-inflammatory cytokines. Supplementation with AX X reduced the damage of intestinal mucosa primarily by inhibiting the MyD88-dependent NF-κB pathway and promoting TJ protein expression, which was demonstrated with the addition of MyD88 inhibitor. The phosphorylation of downstream PKC was associated with the activation of TLR2 [45]. Studies had shown that PKC activation enhanced the integrity of TJs, and the PKC pathway might enhance the intestinal TJ barrier [40]. PKC expression was not significantly changed in the LPS + AX X + MyD88 inhibitor group, which suggested that the function of PKC in a pathway was not dependent on MyD88. Given that the addition of AX H2 had no effect on MyD88 expression, we hypothesized that AX X alleviated intestinal barrier damage by stimulating the TLRs/PKC and inhibiting the TLRs/MyD88/NF-κB signaling pathways.

Conclusions
The results of this study demonstrated that the function of AXH to alleviate Caco-2 cell permeability was related to its fine structure and physicochemical properties. AX X with higher Mw, Ds and bound FA levels was more beneficial to the improvement of barrier function. Furthermore, AX X promoted epithelial barrier integrity by influencing TER, claudin-1 and inflammatory cytokines via the TLRs/MyD88/NF-κB and TLRs/PKC signaling pathways, which was verified by additional MyD88 inhibitor. Whether TLR2 and TLR4 are involved in the two signaling pathways remains unclear. As far as the current results are concerned, further research is required to obtain AX with detailed structure and characteristics, which will lay a solid foundation on the precise regulation of AX.