HO-1/CO Maintains Intestinal Barrier Integrity through NF-κB/MLCK Pathway in Intestinal HO-1−/− Mice

Background Intestinal barrier injury is an important contributor to many diseases. We previously found that heme oxygenase-1 (HO-1) and carbon monoxide (CO) protect the intestinal barrier. This study is aimed at elucidating the molecular mechanisms of HO-1/CO in barrier loss. Materials and Methods We induced gut leakiness by injecting carbon tetrachloride (CCl4) to wildtype or intestinal HO-1-deficient mice. In addition, we administrated tumor necrosis factor-α (TNF-α) to cells with gain- or loss-of-HO-1 function. The effects of HO-1/CO maintaining intestinal barrier integrity were investigated in vivo and in vitro. Results Cobalt protoporphyrin and CO-releasing molecule-2 alleviated colonic mucosal injury and TNF-α levels; upregulated tight junction (TJ) expression; and inhibited epithelial IκB-α degradation and phosphorylation, NF-κB p65 phosphorylation, long MLCK expression, and MLC-2 phosphorylation after administration of CCl4. Zinc protoporphyrin completely reversed these effects. These findings were further confirmed in vitro, using Caco-2 cells with gain- or loss-of-HO-1-function after TNF-α. Pretreated with JSH-23 (NF-κB inhibitor) or ML-7 (long MLCK inhibitor), HO-1 overexpression prevented TNF-α-induced TJ disruption, while HO-1 shRNA promoted TJ damage even in the presence of JSH-23 or ML-7, thus suggesting that HO-1 dependently protected intestinal barrier via the NF-κB p65/MLCK/p-MLC-2 pathway. Intestinal HO-1-deficient mice further demonstrated the effects of HO-1 in maintaining intestinal barrier integrity and its relative mechanisms. Alleviated hepatic fibrogenesis and serum ALT levels finally confirmed the clinical significance of HO-1/CO repairing barrier loss in liver injury. Conclusion HO-1/CO maintains intestinal barrier integrity through the NF-κB/MLCK pathway. Therefore, the intestinal HO-1/CO-NF-κB/MLCK system is a potential therapeutic target for diseases with a leaky gut.


Introduction
In the intestine, the epithelial barrier that regulates the interaction between the luminal material (e.g., gut microbiome) and the interstitium (e.g., mucosal immune cells) is crucial for maintaining homeostasis. The intestinal epithelial barrier function is critical for selective gut permeability and limits the entry of bacteria and pathological bacterial components like lipopolysaccharide (LPS) from the intestinal lumen to the body [1]. If the epithelium is intact, the intestinal barrier function is largely defined by tight epithelial junction (TJ) proteins, such as the transmembrane protein occludin [2] and the peripheral membrane protein zonula occludens 1 (ZO-1) [3]. Mild or severe disruption of the intestinal epithelial barrier can enhance or directly trigger inflammatory bowel disease (IBD) [4], colorectal carcinoma [5], or liver diseases [6]. Therefore, repairing intestinal barrier loss is essential in preventing or delaying the progression of such diseases.
Heme oxygenase-1 (HO-1), a stress-inducible enzyme, catalyzes the initial and rate-limiting step in the oxidative degradation of heme, yielding equimolar amounts of biliverdin IXα (BV), carbon monoxide (CO), and free iron [7]. COreleasing molecule 2 (CORM-2) can spontaneously transfer CO and exert typical CO-mediated pharmacological effects [8]. Recent in vivo and in vitro studies have demonstrated that the HO-1-CO axis prevents intestinal barrier dysfunction [9,10]. HO-1 and CO can prevent intestinal inflammation in mice by promoting bacterial clearance [11]. In our previous study, we identified that HO-1 dependently preserves the intestinal mucosal barrier integrity by abrogating TJ dysregulation and epithelial cell damage [12]. We also found that HO-1 elevation ameliorates intestinal barrier function in bile duct ligation-(BDL-) induced cholestatic liver injury by inhibiting NF-κB p65 [13]. However, whether NF-κB p65 directly mediates the intestinal TJ protein dysregulation still remains unclear.
The myosin light-chain kinase (MLCK) participates in intestinal barrier dysfunction [14]. MLCK has two splice variants derived from the same gene using different promoters. Short or smooth muscle MLCK is not expressed in the intestinal epithelium, whereas long MLCK is highly expressed in intestinal epithelial cells and regulates TJ permeability by inducing phosphorylation of myosin light-chain 2 (MLC-2) [14][15][16], which, in turn, leads to remodeling of the TJ structure. Tumor necrosis factor α (TNF-α) has been shown to promote TJ dysregulation and induce epithelial barrier loss by elevating the expression and activity of long MLCK [17,18]. This elevation is in part mediated by NF-κB [19,20], and a few κB sites have been identified in the upstream promoter region that specifically drives long MLCK activation [21,22].
Hence, based on our previous findings, in this study, we induced gut dysfunction (leakage) by injecting carbon tetrachloride (CCl 4 ) to wildtype (WT) or intestinal HO-1deficient mice or by administrating TNF-α to HO-1 overexpression or knockdown cells. The effects of HO-1/CO maintaining intestinal barrier integrity were examined in vivo and in vitro. These data may provide new ideas for the targeted regulation of intestinal epithelial barrier integrity.

Materials and Methods
2.1. Animal Experiments. C57BL/6 male WT mice (6-8 weeks of age and weighing 20-25 g) were obtained from the Laboratory Animal Center of Dalian Medical University (Liaoning, China). The intestinal HO-1 conditional knockout (HO-1 -/-) mice were constructed using C57BL/6 mice by the Beijing Viewsolid Biotechnology Co. Ltd. (Beijing, China). All the animals were housed in an environment with a temperature of 22 ± 1°C, a relative humidity of 50 ± 1%, and a light/dark cycle of 12/12 hr and fed with food and water ad libitum. All animal studies (including the mice euthanasia procedure) were done in compliance with the regulations and guidelines of Dalian Medical University institutional animal care and conducted according to the AAALAC and the IACUC guidelines (approval No. AEE18006).
At the end of the experiment, mice were sacrificed by cervical dislocation, and blood, colon, and liver samples were collected. The serum samples were obtained by centrifugation of the blood at 2,500 × g for 10 min. The serum levels of alanine aminotransferase (ALT) were determined using a commercial kit (Nanjing Jiancheng Biotechnology Institute, Nanjing, China), according to the manufacturer's instructions. The isolated colon tissue concentrations of TNF-α were measured using the ELISA kits (Wuhan USCN Business Co., Ltd., Wuhan, China) following the manufacturer's protocols. The colon and liver tissues in each group were fixed with 10% paraformaldehyde for histopathological staining, and the remnants of colon tissues were stored at -80°C for later use.  . Membranes were then washed in TBST for 30 min, exposed to the secondary antibody linked to horseradish peroxidase for 1 h, and washed for 30 min in TBST before being developed using ECL detection reagents (Millipore Corp., Billerica, MA, USA).

Hematoxylin and Eosin, Mayer-Sirius Red Staining, and
Immunohistochemistry. The paraffin-embedded colon and liver samples were used to prepare 5 μm thick sections with a microtome. The sections were stained with hematoxylin and eosin (H&E) using standard methods. For Mayer-Sirius red collagen staining, the liver sections were deparaffinized and stained with Sirius red buffer for 1 h at room temperature. After washing, the sections on the slides were stained with Mayer solution and mounted. Immunohistochemistry (IHC) staining was performed according to standard methods. All of these slides were examined and read by an experienced pathologist who was blinded to the study design. The Image J software was used to analyze the images of Sirius red and IHC staining. Colonic epithelial injury from H&E staining was scored according to the inflammatory manifestations and lesion depths of the colon [26]. Inflammatory manifestations (1-4 points) are as follows: 1 = mild inflammation and scattered mononuclear cells can be seen focally; 2 = moderate inflammation with scattered mononuclear cells in many places; 3 = severe inflammation, accompanied by increased vascular density and significantly thickened intestinal wall; and 4 = extreme inflammation, accompanied by a full layer of leukocyte infiltration of the intestinal wall and disappearance of goblet cells. Lesion depths (0-3 points) are as follows: 0 = none; 1 = submucosa; 2 = muscle layer; and 3 = serious film layer.

Data Analysis.
All presented data were representative of three or more independent experiments, each with similar results. The continuous data are shown as mean ± standard deviation. Comparisons between the two groups were performed using Student's t-test. Comparisons among multiple groups were made using ANOVA of Tukey's post hoc test. P values ≤ 0.05 were considered statistically significant.

Damage to the Intestinal Mucosal Barrier and Activation
of the NF-κB p65/MLCK Pathway Is Abolished after CoPP and CORM-2 Treatment following CCl 4 Injection. To define the role of HO-1/CO in repairing intestinal barrier loss, C57BL/6 WT mice were subjected to 12 weeks of CCl 4 injection. All surviving mice (about 75% survival rate) were then administrated with CoPP (an HO-1 inducer), ZnPP (an HO-1 inhibitor), CORM-2, or iCORM-2 for the last 2 weeks. Western blot and qRT-PCR were used to confirm that HO-1 protein and mRNA were upregulated in colonic epithelia after applying CoPP, but not with ZnPP (Supplementary Figure S1a and S1b). Next, we investigated whether HO-1 and CO are involved in regulating intestinal epithelial barrier integrity following CCl 4 injection. The increased pathological score and TNF-α levels of the colon (Figures 1(a)-1(c)), the reduced length of the colon (Figure 1(d)), and the disrupted proteins of TJs such as ZO-1 (Figures 1(e) and 1(k)) and occludin (Figures 1(e) and 1(l)) were observed in the CCl 4 -treated group (all P < 0:001). CoPP treatment attenuated the CCl 4induced colon pathological changes (P < 0:05) and TNF-α levels (P < 0:001) but did not affect the colon length (P > 0:05) (Figures 1(a)-1(d)). Importantly, CoPP and CORM-2 administration significantly increased the expression of colonic epithelial ZO-1 (both P < 0:001) and occludin (both P < 0:01) proteins as compared to the CCl 4treated group (Figures 1(e), 1(k), and 1(l)). ZnPP treatment significantly promoted colonic TNF-α levels ( Figure 1 On the other hand, CCl 4 induced a significant degradation of the inhibitor of nuclear factor-κB α (IκB-α), which was consistent with a marked upregulation of the phosphorylation levels of IκB-α and NF-κB p65 in colonic 3 Oxidative Medicine and Cellular Longevity   j)). In addition, iCORM-2 significantly upregulated NF-κB p65 phosphorylation, but it had no effects on the degradation and phosphorylation of IκB-α and on the expression of long MLCK and phospho-MLC-2 after CCl 4 administration (Figures 1(e)-1(j)). Taken together, these findings indicated that the NF-κB p65/MLCK-p-MLC-2 pathway might be the crucial downstream molecular mechanism of the HO-1-CO axis on protecting against intestinal barrier loss after CCl 4 injection.

HO-1 Overexpression and NF-κB p65 Signaling in
Intestinal Epithelial Cells Are Required for Regulating Barrier Loss following TNF-α Stimulation. To investigate whether NF-κB p65 mediated the downstream MLCK-p-MLC-2 signaling in intestinal epithelial cells and its relationship with HO-1 and barrier loss, we used JSH-23, a specific NF-κB inhibitor, to pretreat Caco-2 cells in which HO-1 function was either increased or decreased. The results showed that in the presence of JSH-23, HO-1 overexpression significantly reduced NF-κB p65 phosphorylation, long MLCK expression, and MLC-2 phosphorylation and markedly increased ZO-1 and occludin expression in Caco-2 cells transfected with the FUGW-HO-1 plasmid after TNF-α stimulation, compared to the scrambled control group (Figures 3(a) and 3(b)). In contrast, even in the presence of JSH-23, Caco-2 cells transfected with the pLKO.1-sh-HO-1 plasmid showed higher phosphorylation of NF-κB p65, expression of long MLCK, and phosphorylation of MLC-2 and lower expression of ZO-1 and occludin after TNF-α stimulation, compared to the scrambled control group (Figures 3(c) and 3(d)). These findings indicated that NF-κB p65 might mediate the activation of epithelial long

11
Oxidative Medicine and Cellular Longevity contributes to barrier loss, and HO-1 suppresses the activation of MLCK/p-MLC-2 signaling in a dependent manner.

Intestinal HO-1-Deficient Mice Contribute to Increased
Intestinal Permeability after CCl 4 Injection. To further confirm the intestinal epithelial cell-specific function of HO-1 in mediating barrier loss, we used VillinCre Hmox1 floxp/floxp mice whose hmox1 genes were conditionally knocked out in intestinal epithelial cells. WT and Hmox1 floxp/floxp mice were used as the controls for the experiments involving VillinCre Hmox1 floxp/floxp mice. HO-1 -/mice showed more serious colonic mucosal injury, which was characterized by infiltration of inflammatory cells in colonic serosa and thickening of the colon wall ( Figure 5(a)). Yet, no significant difference was observed in pathological colon scores between VillinCre Hmox1 floxp/floxp mice and WT or Hmox1 floxp/floxp mice after CCl 4 challenge ( Figure 5(b)). The TNF-α levels of the colon were increased (Figure 5 Table 1 shows liver fibrosis grades in each group. All control mice were Grade 0; Grade 3 fibrosis was only observed in the CCl 4 +ZnPP and CCl 4 +iCORM-2 groups. The distinct pathological changes in the liver, including the proliferation of fibrous tissue around the portal area (Figure 6(a)), the formation of a fibrous septum (Figure 6(a)), and the elevation of serum ALT levels ( Figure 6 There was no significant difference in hepatic fibrosis between the CCl 4 and iCORM-2 groups (all P > 0:05) ( Figure 6). In addition, VillinCre Hmox1 floxp/floxp mice showed disordered hepatic lobular structures, more fibrous tissue proliferation and collagen deposits, higher α-SMA expression in the liver, and higher serum ALT levels than WT and Hmox1 floxp/floxp mice after CCl 4 exposure (Supplementary Table 2 and Figure 7). In summary, these data were consistent with the vital contributions of HO-1/CO to the intestinal barrier restoration pathway.

Discussion
Our data suggested that colonic mucosal injury, TNF-α production, TJ disruption, and epithelial NF-κB p65/MLCK/p-MLC-2 signaling pathway activation are markedly decreased by exogenous upregulation of HO-1 or endogenous supplementation of CO after chronic CCl 4 injection. The effects of TNF-α on TJ permeability and epithelial NF-κB p65/MLCK/p-MLC-2 signaling pathway activation are attenuated in an HO-1-dependent fashion. Using intestinal HO-1-deficient mice further demonstrates the crucial contributions of HO-1 in maintaining intestinal barrier integrity and the relative mechanisms in these processes. Consistent with the above conclusions, alleviated hepatic  fibrogenesis and serum ALT levels confirm the clinical significance of HO-1/CO repairing intestinal barrier injury. To the best of our knowledge, this is the first study on the functional linking of the intestinal HO-1/CO-NF-κB/MLCK system to gut leakiness at different levels.
HO-1 and CO might be possible candidates to initiate intestinal barrier-restorative effects due to their antiinflammation and antioxidative damage properties [27,28]. However, they often used a kind of middle mechanism (e.g., the nuclear factor erythroid-2-related factor 2 (Nrf 2 )/HO-1/CO pathway) for regulating the intestinal barrier dysfunction [29,30], and the direct effect of the HO-1-CO axis on intestinal barrier injury is poorly understood. Our data indicated that intestinal mucosal injury, TNF-α production, and TJ disruption are markedly attenuated by exogenous upregulation of HO-1 (CoPP) or endogenous supplement CO (CORM-2) after chronic CCl 4 injection. As the major connection between intestinal epithelial cells, the TJ proteins in intestinal mucosa have an important role in maintaining the intestinal mucosa's mechanical barrier integrity and functions [1]. Decreased expression of TJ proteins leads to the increase of intestinal permeability, thus facilitating the entry of pathogens and toxic substances into the body [31][32][33]. TNF-α may have a central role in the complex chain reaction of cytokine-mediated intestinal mucosal injury [21,22,34]. Consistent with in vivo studies, our in vitro data showed that HO-1 dependently attenuates TJ disruption in the cell monolayers after TNF-α stimulation. Moreover, using intestinal HO-1 -/mice, we further confirmed the vital role of intestinal-specific HO-1 in mediating barrier loss.
Previous studies reported that intestinal damage induces long MLCK expression by activating the NF-κB signaling pathway [21,35]. We demonstrated that intestinal IκB-α, as an upstream inhibitor of NF-κB, is degraded and phosphorylated, after which the NF-κB p65 is activated and has a central role as a key transcription factor of barrier loss. We previously demonstrated that the inhibition of NF-κB p65 contributes to stabilizing the intestinal barrier [13]. However, we provided no evidence that NF-κB p65 directly mediates the intestinal TJ protein dysregulation. The presence of increased epithelial long MLCK expression and activity, which is mediated by NF-κB p65, contributes to TJ dysregulation [14][15][16]. MLC-2 has a central role as a common final pathway of barrier disruption, and the phosphorylation of MLC-2 is the molecular basis for the increase of permeability of the intestinal barrier [1,36]. This study shows that TNF-α strongly induces epithelial NF-κB p65 phosphorylation, long MLCK expression, and MLC-2 phosphorylation and, consequently, increases the TJ disruption in the cell monolayers. Intestinal barrier disruption is facilitated by NF-κB p65 expressed on intestinal epithelial cells. Moreover, NF-κB   Table S1. Pathological grading of liver fibrosis in each proups. Table S2. Pathological grading of liver fibrosis in each groups (Supplementary Materials)