Role of IL-10 in hepatocyte tight junction alteration in mouse model of experimental colitis.

BACKGROUND
A variety of hepatobiliary abnormalities have been described in patients with chronic inflammatory bowel diseases (IBDs). The purpose of this study was to investigate the role of endogenous IL-10 in alteration of hepatocyte TJ paracellular barrier and in the rapid transcytotic vesicular pathway modification associated with intestinal inflammation.


MATERIALS AND METHODS
To address this question, we used an experimental model of colitis, induced by dinitrobenzene sulfonic acid (DNBS). When compared to DNBS-treated IL-10 wild-type (IL-10WT) mice, DNBS-treated IL-10 knock-out mice (IL-10KO) mice experienced a higher rate of the extent and severity of the histological signs of colon injury.


RESULTS
Colon and liver levels of the pro-inflammatory cytokines tumour necrosis factor, interleukin-1 beta and interleukin-6 were also greatly enhanced in IL-10KO mice in comparison to wild-type mice. Liver histology from IL-10KO and IL-10WT did not show any parenchymal and portal tract inflammation at 4 days after DNBS administration. Serum total bilirubin and Alanine aminotransferase, were significantly increased in DNBS-IL-10KO mice vs. DNBS-IL-10KO mice. Therefore, we found an increase of tight junctional permeability to lanthanum nitrate (molecular weight, 433) in the livers from DNBS-treated IL-10WT mice; lanthanum accumulated throughout the junctional area up to the most apical region bordering the lumen. Absence of a functional IL-10 gene in IL-10KO mice resulted in a significant augmentation of apical diffusion of lanthanum after DNBS-induced colitis. Immunofluorescent labelling of frozen liver sections from DNBS-IL10KO mice, immunolocalization for and claudin-1 and ZO-1 resulted in a significant alteration in the localization of the immunosignals for claudin-1 and ZO-1 after DNBS administration in comparison with DNBS-IL10WT.


CONCLUSION
In conclusion, we suggest that the absence of IL-10 may represent an important pathophysiological mechanism of hepatobiliary injuries and cholestasis observed in patients with IBD.


Introduction
A variety of hepatobiliary abnormalities have been described in patients with chronic inflammatory bowel diseases (IBDs). The prevalence of abnormal liver function tests in patients with ulcerative colitis and Crohn's disease may be as high as 17% (1). It has long been known that serum total bilirubin and aminotransferase and alkaline phosphatase activities can be increased in patients with ulcerative colitis. (2). Histologically, analysis of liver biopsy specimens from patients with IBD can show inflammation of portal areas, cholangitis, and primary sclerosing colangitis. (3) The pathogenetic mechanisms responsible for hepatobiliary alterations in patients with IBD are not known. Several hypotheses have been proposed, including autoimmunity, genetic factors, and gut-derived bacteria antigens or toxins.
A number of animal models for IBD have been developed. The colitis can be observed in several genetically engineered animal models that are created through targeted-gene deletion (4,5).
During the inflammatory reaction, antiinflammatory cytokines are also produced and tend to modulate the inflammatory process. Interleukin-10 (IL-10) is a potent anti-inflammatory cytokine that inhibits the synthesis of the major pro-inflammatory cytokines and chemokines, upregulates humoral immune responses and attenuates cell-mediated immune reactions (27). Because of its properties, IL-10 has the ability to modulate several infectious, immune and inflammatory diseases. It has been demonstrated that genetic deficiency of IL-10 renders mice prone to spontaneous colitis (28) and more susceptible to develop a severe bacterial granulomatous pneumonitis (29). Recently it has been proposed that a breakdown of the hepatocyte TJ paracellular barrier is associated with intestinal inflammation and that it contributes to the pathogenesis of liver disease (30).
The purpose of this study was to investigate the role of endogenous IL-10 in alteration of hepatocyte TJ paracellular barrier and in the rapid transcytotic vesicular pathway modification associated with intestinal inflammation. To address this question, we used an experimental model of colitis, induced by dinitrobenzene sulfonic acid (DNBS), which has been widely used to reproduce and study intestinal inflammation (31,32). After induction of colitis in mice IL-10KO, liver function was assessed by measurement of aminotransferase and alkaline phosphatase activities. The permeability of TJs in vivo was determined by measuring the biliary output of horseradish peroxidase (HRP) in perfused livers under single-pass conditions (33). In addition, semiquantitative analysis of the penetration of lanthanum nitrate into the junctional complexes and bile canalicular lumen (32) provided a morphological evaluation of the permselectivity of TJs. TJ structure was further assessed by immunofluorescent localization of the TJ proteins Claudin-1 and ZO-1. In addition the rapid transcytotic vesicular pathway was quantified observing the HRP positive area over the total pericanalicular or perisinusoidal area as well as bile output. The results show that the genetic absence of IL-10 is associated with an significant increase in some serum markers of cholestasis and with a significant increase in hepatocyte TJ permeability. These observations suggest that endogenous IL-10 may play an antiinflammatory role in the hepatobiliary injury associated with IBD.

Materials and Methods
Animals C57BI-JlO mice (4-5 wk old, 2022 g) with a targeted disruption of the IL-10 gene (IL-10 KO) and littermate wild-type controls (IL-10WT) were purchased from Jackson Laboratories (Bar Harbor, ME, USA). All animals were allowed access to food and water ad libitum. Animal care was in compliance with Italian regulations on protection of animals used for experimental and other scientific purpose (D.M. 116192) as well as with the EEC regulations (O.J. of E.C. L 358/1 12/18/1986)

Induction of Experimental Colitis
Colitis was induced with a very low dose of DNBS (4 mg per mouse) by using a modification (34) of the method first described in rats (31). Mice were anaesthetized by Enflurane. DNBS (4 mg in 100 l of 50% ethanol) was injected into the rectum through a catheter-inserted 4.5cm proximally to the anus. Carrier alone (100 l of 50% ethanol) was administered in control experiments. At day 4 after the beginning of DNBS treatment mice were anaesthetized by pentothal sodium administration (50 mg/kg body wt intraperitoneally), and livers were perfused in situ (flow rate, 16-18 mL/min) with perfusion medium (Krebs'-Henseleit buffer, pH 7.4) containing 1% (wt/vol) bovine serum albumin, 5 mmol/L D-glucose, 1.2 mmol/L CaCl 2 , amino acid mixture, and 20% (vol/vol) washed human erythrocytes (35). The perfusate was maintained at 37'C in a thermostatically controlled cabinet and was continuously gassed with O 2 /CO 2 (19:1, vol/vol). The perfusion pressure was between 6 and 8 cm H20, which has been shown not to damage liver tissue. (36) The viability of perfused livers was assessed on the basis of the marked difference in colour between the afferent and efferent perfusates, the homogenity of liver surface, and the initial bile flow rates that were Ͼ4 L-min'-100 g body wt and aspartate aminotransferase activity in the perfusate (36).

Measurement of Cytokines
TNF␣, IL-6 and IL-1␤ levels were evaluate in colon and in the liver samples. The assay was carried out by using a colorimetric commercial kit (Calbiochem-Novabiochem, La Jolla, CA, USA). The ELISA has a lower detection of 10 pg/ml.

Ultrastructural Study
Electron Microscopy was carried out for evaluation of paracellular transport. Briefly, Livers were flushed with physiological solution before lanthanum perfusion fixation to clear liver tissue from blood and to prevent lanthanum precipitation. The perfusion/fixation solution was 2.5% glutaraldehyde/4% lanthanum nitrate in 0.1 mol/L cacodylate buffer, pH 7.8. After perfusion, livers were surgically in liver samples using the 3,3'-diaminobenzidine technique of Graham and Karnovsky (20). The specimens were incubated in a saturated solution of 3,3-diaminobenzidine in 0.05 M Tris HCl pH 7.6, containing 0.01% (v/v) H 2 O 2 . The liver samples were washed in 0.1 M Tris HCl buffer, pH 7.6, then in 0.1M cacodylate buffer, pH 7.8, and post fixed for 45 min at 4'C in 1% (w/v) osmium tetroxide containing 1% (w/v) potassium ferrocyanide. Finally, the fragments were dehydrated and processed for embedding in Epon 812. Ultrathin sections were stained with lead citrate and examined in a Hitachi H-600 transmission electron microscope at 75 W. After 15 min of recycling perfusion, taurodehydrocholate (TDHQ) was coinfused into the portal vein at a constant rate of 1 mol/min until the end of the experiment. TDHC has been shown to restore the bile flow rate during the assessment of hepatic tight junction permeability (40). Furthermore, TDHC does not affect the transcytotic vesicular pathway (39).
Morphometric analysis were carried out on ten electron micrographs obtained from the pericanalicular (PC) and perisinusoidal (PS) areas of each liver at an initial magnification of 7500X and were printed at a final magnification of 15,0000X. On each electron micrograph, a 1-gm' area around the canalicular and basolateral membrane was delimited. The bile canalicular space and microvilli were excluded. The total pericanalicular and perisinusoidal area of each HRP-containing structure were quantified with the Ibas Kontron semiautomatic analyser. The results are expressed as the percentage of HRP positive area over the total pericanalicular or perisinusoidal area.

Immunofluorescence
Indirect immunofluorescence staining was performed on 7 m-thick sections of unfixed mice liver. Sections were cut in with a cryostat at Ϫ27ЊC, excised, cut into fragments, and incubated in fixation solution (2.5% glutaraldehyde in 0.1 mol/L cacodylate buffer, pH 7.8). Tissue samples were rinsed for 30 minutes in 0.1 mol/L cacodylate buffer and post fixed in 1% osmium tetroxide for 1 hr at 4ЊC. Samples were then dehydrated by graded ethanol and embedded in Agar 100. Ten fragments from each animal (CON-mice, n ϭ 3; DNBS-IL-10WT mice, n ϭ 3, DNBS-IL-10KO mice, n ϭ 3) were sectioned with an Ultramicrotome System 2128 (Ultratome, Bromme) ultramicrotome. Sections (100 nm) were examined at a magnification of 35,000X using a Hitachi H-600 electron microscope (Tokyo, Japan) operated at 50 kV. For morphological evaluation, bile canaliculi were scored as permeable to lanthanum when several particles were observed in the bile canaliculus lumen. Canaliculi in CON-livers were scored as negative when they contained no particles. The second evaluation (apical lanthanum) consisted of determining the number of cell-cell junctions where lanthanum was accumulated in the most apical region of the junction, e.g., infiltrating the junction to the canalicular lumen. This type of scoring was performed on junctions in which the plane of section was such that the accumulation of tracer was most confidently occurring apically with respect to the position of TJs. Observations were performed on number-coded specimens to avoid observer bias.

Measurement of Biliary HRP
Measurement of Biliary HRP output was carried out for evaluation of trancellular transport. To study the pattern of biliary HRP excretion, livers were perfused for an initial 30min under recycling conditions. Livers were then converted to single-pass conditions and 25 mg of HRP (dissolved in 1 ml of BSA-and erythrocyte-free perfusion medium) were infused from a syringe pump over a 1 min period into the perfusion line just prior to the portal cannula. After a further 5 min, recycling perfusion was restored until the end of the experiment. Bile was collected as six 5-min samples during the 30-min stabilization period and then as five 2-min samples followed by eight 5-min samples. Bile flow rates were determined gravimetrically assuming that bile has a density of 1 g/ml.
To study rapid transcytotic vesicular pathways, Imin pulses of HRP (500 mg) was given after 30 min of recycling perfusion according to Hayakawa et al (39). After 2 min of single-pass perfusion, the livers were fixed by perfusion for 2 min with 2.5% (w/v) glutaraldehyde, 0.8% (w/v) paraformaldehyde in 0.1 M cacodylate buffer, pH 7.8. After the perfusion fixation, liver tissue was cut into 1mm 3 fragments and further immersed in the same fixative for 2 hr. Subsequently, fragments were rinsed for 30 min in 0.1 M cacodylate buffer, pH 7.8, containing 0.1M sucrose. HRP activity was examined transferred onto clean glass slides, and dried overnight at room temperature. Sections were fixed with acetone, and rehydrated in phosphatebuffered saline (PBS) and incubated with rabbit polyclonal anti-claudin-1 antibody (1:100 in PBS, vol/vol) (Zimed, Milan, Italy) or with rabbit polyclonal anti-ZO-1 antibody (1:500 in PBS, v/v; Zymed Laboratories) in a humidified chamber for 1 hr at 37ЊC. Sections were washed with PBS and incubated with secondary antibody (anti-rabbit Texas-red conjugate antibody (1:100 in PBS, vol/vol) for 1 hr at 37ЊC. Sections were washed as before, mounted with 90% glycerol in PBS, and observed with a LSM 510 Zeiss laser confocal

Effects of Endogenous IL-10 on the Degree of Colitis (Histology and General Assessment)
Four days after intra-colonic administration of DNBS, the colon appeared flaccid and filled with liquid stool. The macroscopic inspection of cecum, colon and rectum showed presence of mucosal congestion, erosion and hemorrhagic ulcerations (see damage score Fig. 1). The histopathological features included a transmural necrosis and oedema and a diffuse leukocyte cellular infiltrate in the submucosa of colon section from DNBS-treated IL-10WT mice (Fig. 2, C). The absence of IL-10 significantly increases the extent and severity of the histological microscope equipped with a 4OX oil objective. Images were recorded on disk, color-reversed so that fluorescence (white) appeared black, transferred onto slide film and printed from the slide.

Metabolic Changes
Plasma levels of bilirubine and alanine aminotransferase were determined by a clinical laboratory.

Data Analysis
All values in the figures and text are expressed as mean Ϯ standard error (sem.) of the mean of n observations. For the in vivo studies n represents the number of animals studied. In the experiments involving histology or immunohistochemistry, the figures shown are representative of at least three experiments performed on different experimental days. The results were analysed by one-way ANOVA followed by a Bonferroni post-hoc test for multiple comparisons. A p-value less of than 0.05 was considered significant.  signs of colon injury (Figs. 1, 2-D). Four days after colitis induced by DNBS treatment, all mice had diarrhoea and a significant reduction in body weight (compared with the control groups of mice) (Fig. 3). Absence of a functional IL-10 gene in IL-10KO mice resulted in a significant augmentation of lost of body weight (Fig. 3). No histological alteration was observed in the colon tissue from vehicletreated IL-10WT and IL-10KO mice (Fig. 2, A-B, respectively).

Endogenous IL-10 Modulates Production of TNF-␣, IL-6 and IL-1␤ after DNBS Administration
To test whether endogenous IL-10 may modulate the inflammatory process through the regulation of the secretion of other cytokines, we analysed the colon and liver levels of pro-inflammatory cytokines TNF-␣, IL-6 and IL-1␤ in IL-10KO and wild-type mice. A substantial increase of TNF-␣, IL-1␤ and IL-6 formation was found in colon and liver samples collected from IL-10WT mice at 4 d after DNBS administration (Figs. 4, 5). Colon and liver levels of TNF-␣, IL-6 and IL-1␤ were significantly higher in IL-10 deficient mice in comparison to those of IL-10WT animals (Figs. 4, 5).

Liver Histology and Biochemical Parameters
Liver histology from IL-10KO and wild-type mice IL-10WT did not show any parenchymal and portal tract inflammation at 4 d after DNBS administration (Fig. 6, C-D). Serum total bilirubin and Alanine aminotransferase, were significantly increased in DNBS-IL-10WT mice vs. vehicle-mice ( Table 1). Absence of a functional IL-10 gene in IL-10KO mice resulted in a significant augmentation of serum total bilirubin total and alanine aminotransferase production after DNBS-induced colitis.
No histological alteration was observed in the liver tissue from vehicle-treated IL-10WT and IL-10KO mice (Fig. 6, A-B, respectively)

Lanthanum Permeability
Electron microscopic evaluation of the penetration of lanthanum, an electron-dense heavy metal tracer, across the junction has been used to assess the permeability of hepatocyte TJs. (41). We determined the accumulation of lanthanum in the most apical area of the junction. In livers from vehicle-treated WT mice (Fig. 7, A) or from vehicle-treated IL-10KO mice (Fig. 7, B), lanthanum accumulation is arrested at a position at some distance from the bile canaliculus lumen, suggesting that normal TJs prevent apical diffusion of lanthanum. Instead, in livers from DNBStreated IL-10WT mice, lanthanum accumulated throughout the junctional area up to the most apical region bordering the lumen (Fig. 7, C). Absence of a functional IL-10 gene in IL-10KO mice resulted in a significant augmentation of apical diffusion of lanthanum (Fig. 7, D) after DNBSinduced colitis. Quantitative analysis of lanthanum accumulation in the junctions showed that apical accumulation occurred in 59.78% of junctions of livers from DNBS-treated IL-10WT mice and in 6.3% of sham IL-10WT livers (see Fig. 8, A). A more pronounced lanthanum accumulated throughout the junctional area up to the most apical region bordering the lumen was observed in the liver from DNBS-treaded IL-10KO mice (92.8%; Fig. 8, A). Lanthanum permeability was also assessed by determining the percentage of bile canaliculi that contained lanthanum particles (41; see Fig. 8, B). Using this criterium, sham IL-10WT livers showed 1.92% of bile canaliculi morphological evaluation of hepatocyte TJs using two distinct parameters of scoring indicated an increased permeability of hepatocyte TJs. Increased permeability that was more pronounced in IL10KO mice.

HRP Excretion Pattern
To test whether endogenous IL-10 may modulate the paracellular pathway and rapid transcellular pathway during inflammatory, we analysed the first HRP peak. A significantly (P Ͻ 0.05) increased of the HRP output was observed in the liver from DNBS-treated IL-10WT mice. Levels of HRP output were significantly higher in IL-10 deficient mice in comparison to those of IL-10WT animals (Fig. 9). Therefore a high dose of HRP (500 mg) was infused as a 1-min pulse, the percentage area of the pericanalicular and perisinusoidal HRP containing structure obtained at 2 min was not significantly in livers from the different experimental groups (Fig. 10).  from DNBS-treated WT mice. In liver fragment from DNBS-treated IL-10KO mice was found a more significant irregular distribution pattern of ZO-1 (D) and ␤-catenin (C-2). Immunofluorescent labelling of frozen liver section from vehicletreated WT mice (Fig. 12, A) or from vehicletreated IL-10KO mice (Fig. 12, B) with anti-ZO-1 antibody showed that the protein is distributed uniformly and continuously.

Discussion
The objective of the present study was to investigate the role of IL-10 on the hepatic TJ permeability alteration during DNBS-induced experimental colitis.
In the present study we have observed that the increase in serum bilirubin (bilirubin monoglucuronide) and bile acids, markers of cholestasis during DNBS-induced colitis was significantly higher in IL-10KO mice in comparison to the correspondent WT mice. Base on these results we hypothesized that IL-10 plays a role in the modulation of hepatic TJs perselectivity. Cholestasis has been observed in a rat model of IBD induced by intracolonic injection of TNB (30). Our data demonstrate that the increase of the first peak of biliary HRP is significantly higher in the IL-10KO mice in comparison to the IL-10WT mice indicating that tight junctions are more permeable in the absence of IL-10. The results were confirmed by the quantitative analysis of lanthanum accumulation in the TJs of the bile canaliculi. In addition, HRP excretion reflects functional permeation rate, whereas lanthanum indicates structural permeation sites. The increase in the permeability of TJs well correlated with disruption of the junctional complexes, as shown by the immunofluorescent localization of the TJ proteins Claudin-1, which in DNBS liver from IL-10KO

Claudin-1 and ZO-1Immunolocalization
Immunofluorescent labelling of frozen liver section from vehicle-treated WT mice (Fig. 11, A) or from vehicle-treated IL-10KO mice (Fig. 11, B) with anti-Claudin-1 antibody showed that, in longitudinal sections of canaliculi, the protein is distributed uniformly and continuously along two parallel lines bordering the canaliculus. In DNBS-IL10WT mice, immunolocalization for and claudin-1 (Fig. 11, C) was modified in comparison with sham-IL10WT mice. Absence of a functional IL-10 gene in IL-10KO mice resulted in a significant alteration in the localization of the immunosignals for claudin-1 (Fig. 11, D) after DNBS administration.
Similarly, as shown in Figure 12, a significant disruption of immunofluorescence signal for ZO-1 (C) was observed in the liver section obtained  mice was significantly damaged in comparison to the IL-10WT mice. Recently studies have demonstrated that immunostaining for the TJ-associated proteins, ZO-1 (30,42) and cingulin (30) is not changed in hepatocytes of TNB rats.
In contrast, recently it has been found that the immunoistochemistry for 7H6 diminished at the border of the bile canaliculi and immunolocalization was spread to the submembranous cytoplasm showing dissociation in localization of 7H6 and ZO-1 at hepatocyte (42).
Recent studies have demonstrated an important role of ZO-1 in structural and functional organization of tight junctions. In addition Fanning and colleagues (43) have demonstrated that ZO-1 bind to carboxy-tail of occludin, via an NH 2 terminal domain, and actin via COOH-terminal domain, thus acting as a bridge, between plasma membrane and cytoskeleton proteins. These studies suggest that the targeting of claudin to tight junctions is dependent on an interaction with ZO-1 and that ZO-1 appears to play an active role in maintaining the proper subcellular localization of claudin to the tight junctions (44). This allows us to suggest that ZO-1 reduction in DNBS-treated WT mice may impair claudin binding at the membrane with a resultant decrease in membranous claudin staining. Therefore, we have demonstrated that this alteration is significantly more evident in the hepatocyte from DNBS-treated IL-10KO mice. Thus, in the present study the redistribution of TJ protein(s) seems to be correlated to the presence of IL-10. The possibility that TJ alterations may be caused by a toxic effect of DNBS on hepatocytes is unlikely because endoportal administration of DNBS or ethanol produced minor changes in lanthanum permeability of junctions (30). First, chronic portal bacteraemia observed in rats with experimental colitis could act by stimulating macrophages and Kuppfer's cells in the liver to produce a wide range of cytokines. Cytokines are known to modulate TJ function (30,(45)(46)(47). Indeed, endotoxinemia and pro-inflammatory cytokines have been known to affect the function of TJ (48)(49)(50)(51). We confirm here that the inflammatory process (DNBS-induced colitis) leads to a substantial increase in the levels of pro-inflammatory cytokines in the colon and in the liver (Figs. 4, 5). Originally, interleukin 10 was described as a cytokine synthesis inhibitory factor produced by murine Th2 cell clones that could IL-10KO mice are used as a model of spontaneous inflammatory bowel disease (59), in the young IL-10KO mice used for our experiments, the damage appeared to be secondary to the DNBS intracolonic administration since sham mice did not exhibit any histological or clinical signs of colon inflammation. In agreement with our data, it is well known that deficiency of IL-10KO mice developed intestinal diseases, such as colitis, in conditions of age related-stress and changes in bacterial colonization (59,65).
In conclusion, we documented that IL-10 modulate during experimental bowel disease: (1) the functional alteration of permselectivity of hepatocyte TJ; (2) the alteration of tranmembrane dominie of hepatocyte TJ; (3) pro-inflammatory cytokine production and (4) development of cholestasis. We suggest that this may represent an important pathophysiological mechanism of hepatobiliary injuries and cholestasis observed in patients with IBD.
inhibit the synthesis of interferon-g by Th, clones (52)(53)(54)(55). Recent studies have demonstrated that IL-10 can inhibit the synthesis of the major proinflammatory cytokines and chemokines, upregulates humoral immune responses and attenuates cell-mediated immune reactions (56)(57)(58)(59)(60). Interestingly, because of its properties, IL-10 has the ability to modulate several infectious, immune and inflammatory diseases. To prove the crucial role of IL-10 in controlling the inflammatory responses during experimental colitis, previous experimental studies have depended on in vivo administration of exogenous IL-10 and using IL-10KO mice (61)(62)(63)(64). In our study, using genetically engineered mice, we have demonstrated that the endogenous production of IL-10 has a significant role in determining the outcome of colitis, as well as hepatic function. Thus, our findings confirm that the small intestine of young IL-10KO mice is more susceptible to inflammatory stimuli such as experimental colitis. In fact Fig. 11. Confocal laser scanning microscopic images. Claudin-1 in sham-IL10WT mice (A), sham-IL10KO mice (B), DNBS-IL10WT mice (C) and DNBS-IL10KO mice (D). Immunostaining for Claudin-1 was localized bordering the bile canaliculi (A), appearing as two parallel thin continuous lines. In DNBS-IL10WT mice (C) the distribution of the protein was modified in comparison with sham-IL10WT mice. In contrast, the immunohistochemical localization of the claudin-1 in section of liver from DNBS-IL10KO mice was irregular and discontinuous (D).