- Email: [email protected]
Kupffer Cell Inactivation Delays Repair in a Rat Model of Reversible Biliary Obstruction
Journal of Surgical Research 90, 166 –173 (2000) doi:10.1006/jsre.2000.5879, available online at http://www.idealibrary.com on
Kupffer Cell Inactivation Delays Repair in a Rat Model of Reversible Biliary Obstruction 1 Kevin K. Roggin, M.D., Elaine F. Papa, B.S., Arlet G. Kurkchubasche, M.D., and Thomas F. Tracy, Jr., M.D. 2 Rhode Island and Hasbro Children’s Hospitals, Brown University School of Medicine, Providence, Rhode Island 02903 Presented at the Annual Meeting of the Association for Academic Surgery, Philadelphia, Pennsylvania, November 18 –20, 1999
Background. During cholestatic liver injury, Kupffer cells (KC) and activated macrophages modulate cell proliferation and subsequent matrix deposition. The role of KC in the restoration of cell architecture and matrix metabolism during repair following chronic cholestatic liver injury is unknown. Materials and methods. To determine the effect of KC inactivation, adult male Sprague-Dawley rats underwent bile duct suspension (BDS) for 5 days followed by reversal of the obstruction. Saline (control) and gadolinium chloride (10 mg/kg) were administered 1 day prior to BDS and 1 day prior to reversal, to inactivate KC during both injury and repair. Serum bilirubin and quantitative cell morphometry were compared to verify the reversibility of the model. Collagen content of the liver was measured in trichromestained paraffin sections using NIH imaging software. Results. Reversibility of the obstruction was verified by normalization of direct serum bilirubin, which peaked at 8.42 ⴞ 0.76 mg/dL following 5 days of BDS and returned to sham-operated levels 2 days after reversal, 0.36 ⴞ 0.15 mg/dL. Hematoxylin and eosin (H&E)-stained paraffin-embedded liver sections from gadolinium-treated animals at 4 and 7 days after reversal exhibited persistent bile duct proliferation, matrix deposition, and inflammation. Gadoliniumtreated animals had altered collagen metabolism compared to saline controls. Whereas the collagen content in the saline group slowly returned to shamoperated levels over time, the treatment group demonstrated progressive accumulation of collagen during repair which was statistically significant at 7 days following reversal (8.79%/mm 2 ⴞ 2.17 in gadolinium group vs 2.33%/mm 2 ⴞ 0.34 in saline group, P ⴝ 0.0003). 1
This work was supported by NIH Grant DK46831. To whom correspondence should be addressed at Hasbro Children’s Hospital, 593 Eddy Street, Room 147, Providence, RI 02903. Fax: (401) 444-7629. E-mail: [email protected]. 2
0022-4804/00 $35.00 Copyright © 2000 by Academic Press All rights of reproduction in any form reserved.
Conclusions. These results demonstrate that inactivation of resident hepatic macrophages during liver repair impairs collagen metabolism, inhibits the resolution of fibrosis, and allows the persistence of inflammatory cell infiltrates in the portal areas. This is the first evidence of profibrogenic responses in the absence of an intact KC compartment during repair after cholestatic injury. © 2000 Academic Press Key Words: Kupffer cell; gadolinium chloride; biliary obstruction; liver repair. INTRODUCTION
Cholestatic liver injury following extra- or intrahepatic bile duct obstruction causes a stereotypic pattern of nonparenchymal cell proliferation and matrix deposition leading to end-stage liver disease and cirrhosis [1]. Despite the ability to mechanically alleviate obstruction, liver damage often continues unabated with a persistent inflammatory and fibrogenic responses [2]. Current hypotheses of the cellular and molecular events during the course of injury have focused on the central role of the large population of resident hepatic macrophages, or Kupffer cells (KC). Activation of this macrophage population leads to invasion of polymorphonuclear leukocytes, secretion and upregulation of a complicated network of proinflammatory chemical mediators, such as TNF␣, IL-1, and IL-6, and recruitment of systemic macrophages to the site of injury [3]. Subsequent to KC activation, and possibly supported by it, is the activation of hepatic stellate cells [4] to myofibroblasts which elaborate matrix (specifically collagen types I and III) and perpetuate the continued proliferation of myofibroblasts and other fibrogenic nonparenchymal cells of the liver [5]. Important aspects of liver regeneration and repair are assumed to be similar and certain cellular and molecular mechanisms are shared by both [2]. How-
166
ROGGIN ET AL.: KUPFFER CELLS AND BILIARY OBSTRUCTION
ever, given the discordance of their starting points, each must possess its own idiosyncrasies. Whereas regeneration starts with normal remnants of the liver after acute injury, repair after chronic injury occurs in the milieu of a prolonged stimulation of inflammatory and fibrogenic elements that must first be inhibited and subsequently reversed. Unlike regeneration that primarily starts with hepatocyte proliferation, repair of cholestatic liver injury is linked to the simultaneous regression of both myofibroblasts and hyperplastic ductal epithelium [6]. Furthermore, excess types I, III, and IV collagen must be degraded to restore the threedimensional spatial relationships between parenchymal and nonparenchymal cells necessary to perform the normal physiologic and metabolic functions of the liver. In this regard, liver repair following chronic cholestatic injury is analogous to normal wound healing following complex injuries seen in the skin, bowel, lung, and other tissues of the body. We have previously demonstrated that experimental depletion of KC during injury with gadolinium chloride has been shown to exacerbate the fibroproliferative response to total biliary obstruction [7]. The absence of Kupffer cells also inhibits the stem cell compartment, which may act as a potential reservoir to replenish injured hepatocytes and ductular epithelium [7]. Therefore, we speculate that an intact population of differentiated, resident hepatic macrophages is necessary for successful and efficient repair following chronic injury. METHODS Animals. Adult male rats (250 –375 g, virus-free, HarlanSprague-Dawley, Indianapolis, IN) were housed three per cage in an artificial 12 h light/dark cycle with access to rat chow and water ad libitum according to the NIH Guide for the Care and Use of Laboratory Animals. Experiments were carried out under operative guidelines prescribed by the Institutional Animal Care and Use Committee (IACUC) at the Central Research Facility of Rhode Island Hospital and Brown University School of Medicine. Experimental design. The animals were divided into two groups. The study group received gadolinium (III) chloride hexahydrate (Sigma, St. Louis, MO) according to the protocol of Hardonk et al. [8]. To inactivate Kupffer cells during both injury and repair, a dose of 10 mg/kg was administered intravenously into the dorsal penile vein 24 h prior to both bile duct suspension and reversal surgery. The control group received an equivalent volume of saline at identical time points. Animals were randomized to receive either bile duct suspension [9] or manipulation (sham). To simulate chronic liver injury, bile duct suspension was perform for a period of 5 days. Our experience with this model has documented characteristic bridging fibrosis between adjacent portals after similar periods of injury [3]. Following reversal of the obstruction, animals were sacrificed at 0, 2, 4, and 7 days of repair. Disappearance of acholic stools, normalization of serum bilirubin, and restitution of normal hepatic histology confirmed reversibility of our model. In addition, body weights were recorded at each time point. Reversible bile duct suspension surgery. Using specific modifications of a procedure first presented by Posner et al. [9], rats were placed in an induction chamber insufflated with 100% O 2 at a flow
167
rate of 4 L/min and isoflurane USP anesthetic (Fort Dodge Animal Health, Fort Dodge, IA) via an Ohio-style vaporizer. Following induction, rats were weighed, marked, shaved, and transferred to an operating platform equipped with a nonrebreathing circuit face mask for delivery of a continuous mixture of oxygen and inhalational anesthetic. The abdomen was prepared with 70% ethyl alcohol (Pharmco, Brookfield, CT) and 10% providone-iodine solution (The Purdue Frederick Company, Norwalk, CT) and covered with a transparent sterile drape. Using aseptic technique, the peritoneal cavity was entered through a midline laparotomy incision. The liver was elevated cephalad and the duodenum was retracted caudad to expose the porta hepatis. The common bile duct was mobilized at its proximal extent and suspended by a silicone SURG-I LOOP (Scanlan International, Saint Paul, MN). The free ends of the loop were brought out through separate stab incisions in the skin 1 cm lateral to the midline at the apex of the wound. A 1-cm premarked segment of the loop was suture-ligated to each costochondral margin with 3-O polyglactin 910 suture (Ethicon, Somerville, NJ) and buried in the subcutaneous tissue with separate horizontal mattress sutures. The peritoneum and skin incisions were closed with running 4-O and 2-O polyglactin 910 sutures, respectively. Sham-operated animals underwent common bile duct manipulation without suspension; all other technical aspects of the procedure were identical to those of the obstructed animals. Postoperatively, the rats were transferred to clean cages under warming lights and observed. To minimize pain, chow was placed on the cage floor and water was administered through long sippers ad libitum. The rats were checked daily after surgery for stool characteristics and signs of morbidity. Decompression procedure. Rats were anesthetized with isoflurane, weighed, and prepared with 70% ethyl alcohol and 10% providone-iodine. Blood (1 cc) was collected via the tail vein and stored at ⫺4°C until ready for use. The paired stab incisions were sharply opened to expose the subcutaneous loop ends. Caudad retraction of the loop ends exposed the costochondral sutures, which were cut to free the loop from its attachments. The loop was carefully removed and the skin was closed with a horizontal mattress stitch using 2-O polyglactin 910 suture. Postoperative care was performed as previously described. Necropsy. Anesthetized rats were weighed and then sacrificed with an intraperitoneal injection of pentobarbital sodium USP 120 mg/kg (Abbott Laboratories, North Chicago, IL). The abdomen was widely opened and 3 cc of venous blood was collected from the inferior vena cava and stored at ⫺4°C until ready for use. Following hepatectomy, liver samples were fixed in 10% buffered formalin phosphate (Fischer Scientific, Fair Lawn, NJ) and subsequently mounted in paraffin and processed (RIH Central Research Labs) for immunohistochemistry. For frozen tissue sectioning, 0.5–1 cm 2 liver cubes were placed in plastic molds filled with Cyto-Tek tissue freezing media (Sakura, Torrance, CA). These sections were frozen in a hexane (Fischer Scientific) bath cooled with acetone (Fischer Scientific) and dry ice for 5 min. Liver sections (5 m) were cut with a Shur/Cut cryostat (Triangle Biomedical Sciences, Durham, NC), fixed in cold acetone for 10 min, and stored at ⫺80°C until ready for use. Staining and immunohistochemistry. Formalin-fixed, paraffin embedded, 5-m liver tissue sections were stained with H&E to identify bile duct hyperplasia, inflammatory infiltrate, and fibrosis. To demonstrate type I collagen deposition, serial sections were treated with trichrome stain, with the nuclear stain omitted for increased resolution of collagen fibrils. Monoclonal antibody anti-cytokeratin-19 (Amersham, Little Chalfont, Buckinghamshire, England), diluted 1:10 in phosphate buffered saline (PBS) with 10% goat serum (Sigma, St. Louis, MO), was applied to frozen liver sections. A FITC-conjugated IgG2b goat antimouse secondary antibody (Boehringer, Indianapolis, IN), diluted 1:100 in PBS, was used to label CK-19-positive bile duct epithelial cells under fluorescent microscopy (Nikon Microphot-FXA, Nikon, Tokyo, Japan).
168
JOURNAL OF SURGICAL RESEARCH: VOL. 90, NO. 2, MAY 15, 2000
Weight Change Induced by Reversible Biliary Obstruction
FIG. 1. Average serum direct bilirubin assay was performed to verify the reversibility of our model. Bile duct suspension increased the direct serum bilirubin in all experimental animals (*P ⱕ .05 vs sham). Following removal of the loop suspension (repair), the average serum bilirubin normalized to sham-operated levels.
Image analysis for collagen content. Trichrome-stained tissue sections were scanned (SprintScan 35 Plus, Polaroid, Cambridge, MA) into a PowerMac 9600 (Apple Computer, Inc., Cupertino, CA) using a PathScan Enabler (0.054) optical card (Meyer Instruments, Inc., Houston, TX). NIH Imaging Software Version 1.61 (NIH, Bethesda, MD) was used to create binary images from the red, green, and blue channels. An arithmetic function calculated the intersection between channels to generate a composite binary image. Up to 900,000 mm 2 of tissue per animal was analyzed, measured, and calibrated to a standardized scale. Results were expressed as average percentage collagen per square millimeter of liver. Quantification of serum direct bilirubin. Frozen serum samples were shielded from direct light, thawed to room temperature, and assayed for direct serum bilirubin using a commercial kit (Sigma Diagnostics, St. Louis, MO) according to the manufacturer’s instructions. Statistical analysis. The data were processed using superANOVA statistical software (Abacus Concepts Inc., Berkeley, CA). A one-way analysis of variance test (ANOVA) was used to compare serum bilirubin and collagen data. Post hoc analysis was performed using Scheffe’s and Fisher’s Protected LSD tests. The criterion for significance for all studies was P ⱕ 0.05. Results are expressed as mean ⫾ standard error of the mean.
RESULTS
Direct Serum Bilirubin Assay Compared to sham-operated animals (bilirubin ⫽ 0.15 mg/dL ⫾ 0.12, N ⫽ 4), 5 days of bile duct suspension (N ⫽ 25) increased direct serum bilirubin to 8.42 mg/dL ⫾ 0.77 (P ⱕ 0.05, Fig. 1). Removal of the obstructing loop normalized serum bilirubin at all time points during repair: 0.25 mg/dL ⫾ 0.13 at 2 days of repair (N ⫽ 6), 0.15 mg/dL ⫾ 0.13 at 4 days of repair (N ⫽ 6), and 0.36 mg/dL ⫾ 0.15 at 7 days of repair (N ⫽ 7); there was no statistically significant difference between sham serum bilirubin and the values from repairing animals (Fig. 1). In addition, gadolinium administration did not alter serum bilirubin from saline-treated animals at similar time points (data not shown).
In sham-operated animals (N ⫽ 4), average body weight increased 5 days after common bile duct manipulation: 287.1 ⫾ 7.0 g (preoperative weight) to 305.4 ⫾ 8.5 g. Saline-treated animals had an average preop weight of 300.8 ⫾ 10.0 g (N ⫽ 13) that decreased to 274.1 ⫾ 9.7 g (N ⫽ 13) following 5 days of common bile duct suspension. During repair the average body weight was 287.2 ⫾ 20.6 g at 2 days of repair (N ⫽ 3), 316.6 ⫾ 14.3 g at 4 days of repair (N ⫽ 3), and 290.36 ⫾ 11.78 g at 7 days of repair (N ⫽ 4). The weight change in gadolinium-treated animals was not statistically significant from the saline group at equivalent time points: 310.2 ⫾ 9.4 g (preop, N ⫽ 12), 278.7 ⫾ 8.0 g (reversal, N ⫽ 12), 289.6 ⫾ 9.8 g (Repair Day 2, N ⫽ 3), 297.5 ⫾ 11.3 g (Repair Day 4, N ⫽ 3), and 304.0 ⫾ 7.1 g (Repair Day 7, N ⫽ 3). Histologic Changes during Injury and Repair Sham-operated animals did not show evidence of increased fibrosis, inflammatory infiltrate, or bile duct hyperplasia 5 days after common bile duct manipulation (data not shown). Bile duct suspension altered the histologic hepatic architecture of both saline- and gadolinium-treated animals. Changes included bile duct hyperplasia, periportal bridging fibrosis surrounded by an intense inflammatory infiltrate, and a few scattered areas of hepatocellular necrosis with relative sparring of the pericentral areas (Fig. 2). Following removal of the obstructing loop, salinetreated animals showed gradual restitution of normal hepatic architecture (Fig. 2). By 7 days of repair, only scattered clusters of hyperplastic ducts could be identified; both the inflammatory infiltrate and fibrotic changes were comparable to those of sham-operated animals (data not shown). Repair in gadoliniumtreated animals was delayed as evidenced by the persistence of periportal extracellular matrix over time; even at 7 days of repair, bridging fibrosis and hyperplastic duct epithelium could still be identified throughout the hepatic lobule (Fig. 2). In addition, high-powered magnification of portal triads showed increased matrix distorting the normal spatial relationships among bile ducts, hepatic arterioles, and portal venules. Delayed Bile Duct Resorption Following Kupffer Cell Inactivation Periportal cytokeratin-19-positive bile ductal epithelium was increased in liver sections from both salineand gadolinium-treated animals following bile duct suspension (Fig. 3). Similar to that observed in H&E stained liver sections, resorption of hyperplastic epithelium was delayed in gadolinium-treated animals
ROGGIN ET AL.: KUPFFER CELLS AND BILIARY OBSTRUCTION
169
FIG. 2. Histological changes in saline- and gadolinium-treated rat liver sections in model of reversible biliary obstruction. Five days of bile duct suspension induced bile duct hyperplasia (white arrows), bridging fibrosis between adjacent portal tracts (PT), and an intense inflammatory infiltrate in both saline (A) and gadolinium (B) groups. Following 7 days of repair, saline-treated animals (C) exhibited normal appearing hepatic architecture with absence of fibrosis between PT and central veins (CV). The liver sections from the gadolinium group (D) showed persistent bile duct hyperplasia, increased extracellular matrix, and bridging fibrosis suggestive of delayed repair.
compared to saline-treated controls 7 days after removal of the loop (Fig. 3). Altered Collagen Metabolism during Repair Following Reversible Biliary Obstruction Common bile duct suspension increased type I collagen in trichrome-stained paraffin liver sections in both saline and gadolinium-treated rats (Fig. 4). Compared to saline-treated sham-operated animals, the percentage of collagen increased after bile duct suspension from 1.13%/mm 2 ⫾ 0.07 (N ⫽ 4) to 8.26%/mm 2 ⫾ 1.32 (N ⫽ 3, Fig. 5A). Following removal of the loop, the percentage of collagen in liver sections progressively decreased over time: 4.83%/mm 2 ⫾ 0.58 at 2 days of repair (N ⫽ 3), 3.412%/mm 2 ⫾ 0.40 at 4 days of repair (N ⫽ 3), and 2.33%/mm 2 ⫾ 0.34 at 7 days of repair (N ⫽ 4, Fig. 5A). Gadolinium chloride altered collagen metabolism resulting in progressive increase in collagen staining over time. Following bile duct suspension (5.867%/ mm 2 ⫾ 1.43, N ⫽ 3), the amount of collagen did not decrease compared to saline-matched controls: 6.44%/ mm 2 ⫾ 1.59 at 2 days of repair (N ⫽ 3), 6.27%/mm 2 ⫾ 1.85 at 4 days of repair (N ⫽ 3), 8.79%/mm 2 ⫾ 2.17 at 7 days of repair (N ⫽ 3, P ⫽ 0.0003, Fig. 5B).
DISCUSSION
Globally, the morbidity and mortality of congenital or acquired liver diseases that might benefit from repair far exceed those of acute injuries which require the powers of liver regeneration. Unchecked, chronic liver injuries lead to lethal cirrhosis. Although transplantation for end-stage liver disease has had significant success [10], direct attacks on the sources of liver injury and augmentation of repair will benefit patients awaiting transplantation and those in whom transplant is not feasible [11]. The spectrum of liver injury includes infectious, metabolic, inflammatory, toxic, and cholestatic insults. Regardless of the pathogenic stimulus, most forms lead to scar formation [12]. Repeated insults (e.g., chronic ethanol abuse or viral hepatitis) or inefficient repair (e.g., malnutrition or hepatocellular carcinoma) can also stimulate fibrosis as the liver attempts to limit hepatocellular damage. In fact, the liver’s response to chronic injury is similar to other models of wound healing [5]. Therapeutic strategies to reverse this process often fail to interrupt this progressive cycle leading to the sustained accumulation of extracellular matrix characteristic of cirrhosis [13]. Chronic cholestatic liver disease is a common, pri-
170
JOURNAL OF SURGICAL RESEARCH: VOL. 90, NO. 2, MAY 15, 2000
FIG. 3. Delayed resorption of CK-19-positive bile duct epithelium in gadolinium-treated animals. Loop suspension induced comparable bile duct hyperplasia in both saline (A) and gadolinium (B) groups. At 7 days of repair, while most of the hyperplastic bile ducts regressed in the saline group (C), the gadolinium-treated rats (D) had increased numbers of proliferating bile duct epithelium, indicating delayed repair.
mary or secondary injury that predisposes adults and children to the development of cirrhosis [14]. Postcanalicular biliary obstruction leads to wellcharacterized bile duct epithelial cell proliferation, collagen deposition, and progressive derangement of hepatic architecture [15]. Successful techniques of biliary decompression for those with mechanical forms of extrahepatic obstruction have increased survival from previously lethal cholestatic disease [16]. Current research efforts to elucidate the cellular and molecular mechanisms of repair have been hindered by a paucity of clinically relevant animal models of liver repair, the poor translation of in vitro results to human pathophysiology, and the logistics of controlling the complex intercellular interactions during repair. Nevertheless, we have modified a model of reversible biliary obstruction [9] to study repair after chronic cholestatic injury in the rat. This model is unique in that it allows us to simulate a predictable chronic liver injury and subsequent repair under controlled conditions, which have been reproducible in our lab. In addition, our model avoids the technical difficulties and morbidity associated with bilioenteric anastomoses and repeated laparotomies in small laboratory animals that have limited the utility of other models of liver repair. Following experimental bile duct suspension, the elevation of direct serum bilirubin and histologic evidence of bile duct proliferation, bridging fibrosis, and inflammatory infiltrate confirm that our model is ca-
pable of simulating progressive cholestatic liver injury [3]. In addition, the rapid normalization of serum bilirubin and restitution of normal hepatic architecture in saline controls indicate that our model was in fact reversible. Although cholestasis caused a transient and reversible loss of body weight, this is most likely related to the trauma of surgery. Metabolic studies have shown that weight loss induced by bile duct ligation is transient and reversible [17]. Repair after cholestatic injury requires the systematic reversal of a number of cellular and molecular alterations stemming from bile duct obstruction [2]. Impediments to successful repair include the sustained proinflammatory cytokine feedback loops, the proliferation of bile duct epithelium, and the resorption of abundant extracellular matrix, including collagen, which makeup scar. We and others speculate that the macrophage is integral to this process [18]. As a key component of the reticuloendothelial system, the resident hepatic macrophage can clear bacteria from the bloodstream, mobilize host defenses through the secretion of inflammatory mediators, and secrete matrix metalloproteinases, which degrade the extracellular matrix [19]. Since the activated macrophage is responsible for the secretion of these mediators, they represent a potential target for anti-fibrogenic therapy. In the liver, as in many other tissues including, but not limited to, the lung, skin, and bowel, both parenchymal and immune-mediated injuries lead to the ac-
ROGGIN ET AL.: KUPFFER CELLS AND BILIARY OBSTRUCTION
171
and efficient repair depends upon the presence of an intact resident macrophage population. To test our hypothesis that repair after cholestatic injury is dependent on an intact liver macrophage population, we used gadolinium chloride to selectively deplete Kupffer cells. Gadolinium chloride is a rare earthmetal salt, which is selectively, absorbed by large zone I hepatic macrophages (KC) [20]. Gadolinium chloride, which is soluble at pH 5 or less, forms insoluble colloid complexes in the neutral pH of blood and is readily phagocytosed by KC [21]. Once inside the cells, elemental gadolinium is liberated, which inhibits phagocytosis and causes cell death [8]. Critical analysis of this technique has been documented in numerous animal
FIG. 4. Type I collagen in trichrome-stained paraffin sections in sham (A), saline (C), and gadolinium-treated rats (E) was processed and digitally converted to composite binary images; collagen is displayed as white lines in the black images on panels B (sham), D (saline), and F (gadolinium). Five days after bile duct manipulation, this representative sham-operated animal (B) showed a paucity of type I collagen. By 7 days of repair, the representative binary image from a saline-treated rat (D) showed less overall type I collagen than in a gadolinium-treated animal (F) at the same point during repair.
tivation of macrophages [18]. Through both cellular interaction and secretion of intercellular chemical signals, macrophages transform quiescent stellate cells into their activated phenotype, the myofibroblast [5]. This process leads to the net deposition of collagen and other components of the extracellular matrix which makeup scar. Successful repair following injury is dependent on a coordinated series of events, which reverse this process. It is our hypothesis that successful
FIG. 5. (A) Average percentage collagen per square millimeter of liver tissue in saline-treated animals during both injury and repair versus sham-operated saline controls (N ⱖ 3 at each time point). Following removal of the loop, the average percentage collagen progressively diminished over time, approaching sham levels by 7 days of repair. Abbreviations: SHAM (saline shams), Nd0 (Necropsy Day 0/saline group at reversal), Nd2 (Necropsy Day 2/saline group at 2 days of repair), Nd4 (Necropsy Day 4/saline group at 4 days of repair), Nd7 (Necropsy Day 7/saline group at 7 days of repair). (B) Average percentage collagen per square millimeter of liver tissue in gadolinium-treated animals during both injury and repair versus sham-operated gadolinium controls (N ⱖ 3 at each time point). Gadolinium chloride-induced Kupffer cell depletion delayed liver repair after bile duct suspension. Following reversal of the loop suspension, the average collagen content in rat livers increased over time and approached statistical significance by 7 days of repair (#P ⱕ 0.05 versus the saline group). Abbreviations: SHAM (saline shams), Nd0 (Necropsy Day 0/gadolinium group at reversal), Nd2 (Necropsy Day 2/gadolinium group at 2 days of repair), Nd4 (Necropsy Day 4/gadolinium group at 4 days of repair), Nd7 (Necropsy Day 7/gadolinium group at 7 days of repair).
172
JOURNAL OF SURGICAL RESEARCH: VOL. 90, NO. 2, MAY 15, 2000
models and it is considered to be a suitable approach for studying the in vivo effect of KC depletion [8]. We used gadolinium according to the protocol of Hardonk; the drug was administered 1 day prior to bile duct suspension and again 1 day prior to reversal (repair). Since previous studies have shown that systemic monocytes repopulate the liver 4 days after treatment and that the new macrophages are also vulnerable to gadolinium-depletion [8], this dosing schedule allowed us to sequentially eliminate KC during both injury and repair. Gadolinium-induced elimination of KC resulted in delayed repair as evidenced by persistent bile duct hyperplasia, a profound inflammatory infiltrate, and accumulation of collagen and extracellular matrix. KC regulate the deposition of collagen both by the secretion of macrophage-specific matrix metalloproteinases and through paracrine or cell-to-cell stimulation of other collagen and matrix-producing cells [22]. Both activated stellate cell, or myofibroblasts, and bile duct epithelial cells are potential targets for KC modulation. The phenotypic changes in the macrophage population in the liver during both injury and repair may influence the quality and quantity of extracellular matrix observed in our model. In vitro experiments involving the isolation of KC from both normal and injured animals treated with gadolinium will be required to characterize the production of metalloproteinases and their inhibitors and their impact on collagenolytic or gelatinolytic activity. While the use of gadolinium chloride remains an effective strategy for in vivo KC depletion, its potential deleterious effects on other aspects of liver physiology have led to controversy over its use in experimental models. Several studies in animal sepsis models have described in vivo gadolinium chloride-induced alterations of cytochrome P450 activity [23], coagulation cascade [24], hepatic microvascular perfusion [25], and serum levels of proinflammatory cytokines [25]. In our model of reversible biliary obstruction, gadolinium did not appear to cause any of these untoward effects. There were no histologic alterations induced in gadolinium-treated, sham-operated animals compared to saline controls. In addition, gadolinium did not alter the normalization of serum bilirubin or increase mortality over matched saline-treated rats. Several sepsis studies have also demonstrated reduced mortality to a lethal LPS challenge following pretreatment with gadolinium chloride [24, 26]. Therefore, although cognizant of its potential as a confounding variable, we are confident that the toxic side effects of gadolinium were not responsible for the delayed repair observed in the treated group. Furthermore, a recent report of increased collagen deposition by dexamethasone-induced Kupffer cell depletion in a model of biliary obstruction [27] confirms that this observation cannot solely be
explained by the toxicity of gadolinium chloride and that KC may protect the liver from cholestatic injury. Therefore, it is our conclusion that not only is the Kupffer cell a positive modulator of repair, but the presence of an intact population of resident hepatic macrophages remains a potential target for antifibrotic therapies designed to augment repair after cholestatic injury. The speculative nature of Kupffer cells as the primary positive modulator of liver repair after cholestatic injury requires further experimental confirmation and validation. We propose a model that during injury and subsequent repair, a distinct macrophage population regulates the secretion of chemoattractants, proinflammatory, and inhibitory cytokines, which mobilize host defenses, limit parenchymal cell damage, and initiates repair. Elucidation of this cellular and molecular process will help design therapies designed to augment repair following biliary decompression for cholestatic injury in surgical patients. REFERENCES 1.
2.
3.
4. 5.
6.
7.
8.
9.
10. 11. 12.
13.
Tracy, T., Goerke, M., Bailey, P., Sotelo-Avila, C., and Weber, T. Growth-related gene expression in early cholestatic liver injury. Surgery 114: 532, 1993. Tracy, T., and Fox, E. Molecular and cellular control points in pediatric liver injury and repair. Semin. Pediatr. Surg. 5: 175, 1996. Aldana, P. R., Goerke, M. E., Carr, S. C., and Tracy, T. F. The expression of regenerative growth factors in chronic liver injury and repair. J. Surg. Res. 57: 711, 1994. Friedman, S. Stellate cell activation in alcoholic fibrosis—An overview. Alcohol. Clin. Exp. Res. 23: 904, 1999. Friedman, S. L. The cellular basis of hepatic fibrosis. Mechanisms and treatment strategies. N. Engl. J. Med. 328: 1828, 1993. Ramm, G. A., Carr, S. A., Li, L., Vogler, C. A., Britton, R. S., Bacon, B. R., and Tracy, T. F. Regression of myofibroblasts during hepatic repair following chronic cholestatic liver injury in rats. Liver, 2000, in press. Olynyk, J. K., Yeoh, G. C., Ramm, G. A., Clarke, S. L., Hall, P. M., Britton, R. S., Bacon, B. R., and Tracy, T. F. Gadolinium chloride suppresses hepatic oval cell proliferation in rats with biliary obstruction. Am. J. Pathol. 152: 347, 1998. Hardonk, M. J., Kijkhuis, F. W. J., Hulstaert, C. E., and Koudstaal, J. Heterogeneity of rat liver and spleen macrophages in gadolinium chloride-induced elimination and repopulation. J. Leukocyte Biol. 52: 296, 1992. Posner, M., Burt, M., Stone, M., Han, B., Warren, R., Vydelingum, N., and Brennan, M. A model of reversible obstructive jaundice in the rat. J. Surg. Res. 48: 214, 1990. Wall, W. J. Liver transplantation: Past accomplishments and future challenges. Can. J. Gastroenterol. 13: 257, 1999. Alonso, M. H., and Ryckman, F. C. Current concepts in pediatric liver transplant. Semin. Liver Dis. 18: 295, 1998. Li, D., and Friedman, S. L. Liver fibrogenesis and the role of hepatic stellate cells: New insights and prospects for therapy. J. Gastroenterol. Hepatol. 14: 618, 1999. Kovacs, E. J., and DiPetro, L. A. Fibrogenic cytokines and connective tissue production. FASEB J. 8: 854, 1994.
ROGGIN ET AL.: KUPFFER CELLS AND BILIARY OBSTRUCTION 14. 15.
16. 17.
18.
19.
20. 21.
McGill, J. M., and Kwiatkowski, A. P. Cholestatic liver diseases in adults. Am. J. Gastroenterol. 93: 684, 1998. Morris, J. S., Gallo, G. A., Scheuer, P. J., and Sherlock, S. Percutaneous liver biopsy in patients with large bile duct obstruction. Gastroenterology 68: 750, 1975. Ahrendt, S. A., and Pitt, H. A. A history of the bilioenteric anastomosis. Arch. Surg. 125: 1493, 1990. Gouma, D. J., Roughneen, P. T., Kumar, S., Moody, F. G., and Rowlands, B. J. Changes in nutritional status associated with obstructive jaundice and biliary drainage in rats. Am. J. Clin. Nutr. 44: 362, 1986. Tracy, T. F., Dillon, P., Fox, E. S., Minnick, K., and Vogler, C. The inflammatory response in pediatric biliary disease: Macrophage phenotype and distribution. J. Pediatr. Surg. 31: 121, 1996. Winwood, P. J., and Arthur, M. P. J. Kupffer cells: Their activation and role in animal models of liver injury and human liver disease. In Seminars in Liver Disease. New York: Thieme Med. Pub., 1993. Pp. 50 –59. Lazar, G. The reticuloendothelial blocking effect of rare earth metals in rats. J. Reticuloendothel. Soc. 13: 231, 1989. Spencer, A. J., Wilson, S. A., Batchelor, J., Reid, A., Rees, J.,
22. 23.
24.
25.
26.
27.
173
and Harpur, E. Gadolinium chloride toxicity in the rat. Toxicol. Pathol. 25: 245, 1997. Arthur, M. J. Degradation of matrix proteins in liver fibrosis. Pathol. Res. Pract. 190: 825, 1994. Badger, D. A., Kuester, R. K., Sauer, J.-M., and Sipes, I. G. Gadolinium chloride reduces cytochrome P450: Relevance to chemical-induced hepatotoxicity. Toxicology 121: 143, 1997. Ruttinger, D., Vollmar, B., Kempter, B., and Messmer, K. Failure of Kupffer cell blockade to prevent disseminated intravascular coagulation in endotoxemic rats despite improved survival. Langenbecks Arch. Chir. 383: 75, 1998. Ruttinger, D., Vollmar, B., Wanner, G. A., and Messmer, K. In vivo assessment of hepatic alterations following gadolinium chloride-induced Kupffer cell blockade. J. Hepatol. 25: 960, 1996. Vollmar, B., Ruttinger, D., Wanner, G. A., Leiderer, R., and Menger, M. D. Modulation of Kupffer cell activity by gadolinium chloride in endotoxemic rats. Shock 6: 434, 1996. Melgert, B. N., Weert, B., Molema, G., Meijer, D. K., and Poelstra, K. Chronic treatment of bile duct ligated rats with dexamethasone selectively delivered to Kupffer cells. Hepatology 30: 329A, 1999.
Kupffer Cell Inactivation Delays Repair in a Rat Model of Reversible Biliary Obstruction 1 Kevin K. Roggin, M.D., Elaine F. Papa, B.S., Arlet G. Kurkchubasche, M.D., and Thomas F. Tracy, Jr., M.D. 2 Rhode Island and Hasbro Children’s Hospitals, Brown University School of Medicine, Providence, Rhode Island 02903 Presented at the Annual Meeting of the Association for Academic Surgery, Philadelphia, Pennsylvania, November 18 –20, 1999
Background. During cholestatic liver injury, Kupffer cells (KC) and activated macrophages modulate cell proliferation and subsequent matrix deposition. The role of KC in the restoration of cell architecture and matrix metabolism during repair following chronic cholestatic liver injury is unknown. Materials and methods. To determine the effect of KC inactivation, adult male Sprague-Dawley rats underwent bile duct suspension (BDS) for 5 days followed by reversal of the obstruction. Saline (control) and gadolinium chloride (10 mg/kg) were administered 1 day prior to BDS and 1 day prior to reversal, to inactivate KC during both injury and repair. Serum bilirubin and quantitative cell morphometry were compared to verify the reversibility of the model. Collagen content of the liver was measured in trichromestained paraffin sections using NIH imaging software. Results. Reversibility of the obstruction was verified by normalization of direct serum bilirubin, which peaked at 8.42 ⴞ 0.76 mg/dL following 5 days of BDS and returned to sham-operated levels 2 days after reversal, 0.36 ⴞ 0.15 mg/dL. Hematoxylin and eosin (H&E)-stained paraffin-embedded liver sections from gadolinium-treated animals at 4 and 7 days after reversal exhibited persistent bile duct proliferation, matrix deposition, and inflammation. Gadoliniumtreated animals had altered collagen metabolism compared to saline controls. Whereas the collagen content in the saline group slowly returned to shamoperated levels over time, the treatment group demonstrated progressive accumulation of collagen during repair which was statistically significant at 7 days following reversal (8.79%/mm 2 ⴞ 2.17 in gadolinium group vs 2.33%/mm 2 ⴞ 0.34 in saline group, P ⴝ 0.0003). 1
This work was supported by NIH Grant DK46831. To whom correspondence should be addressed at Hasbro Children’s Hospital, 593 Eddy Street, Room 147, Providence, RI 02903. Fax: (401) 444-7629. E-mail: [email protected]. 2
0022-4804/00 $35.00 Copyright © 2000 by Academic Press All rights of reproduction in any form reserved.
Conclusions. These results demonstrate that inactivation of resident hepatic macrophages during liver repair impairs collagen metabolism, inhibits the resolution of fibrosis, and allows the persistence of inflammatory cell infiltrates in the portal areas. This is the first evidence of profibrogenic responses in the absence of an intact KC compartment during repair after cholestatic injury. © 2000 Academic Press Key Words: Kupffer cell; gadolinium chloride; biliary obstruction; liver repair. INTRODUCTION
Cholestatic liver injury following extra- or intrahepatic bile duct obstruction causes a stereotypic pattern of nonparenchymal cell proliferation and matrix deposition leading to end-stage liver disease and cirrhosis [1]. Despite the ability to mechanically alleviate obstruction, liver damage often continues unabated with a persistent inflammatory and fibrogenic responses [2]. Current hypotheses of the cellular and molecular events during the course of injury have focused on the central role of the large population of resident hepatic macrophages, or Kupffer cells (KC). Activation of this macrophage population leads to invasion of polymorphonuclear leukocytes, secretion and upregulation of a complicated network of proinflammatory chemical mediators, such as TNF␣, IL-1, and IL-6, and recruitment of systemic macrophages to the site of injury [3]. Subsequent to KC activation, and possibly supported by it, is the activation of hepatic stellate cells [4] to myofibroblasts which elaborate matrix (specifically collagen types I and III) and perpetuate the continued proliferation of myofibroblasts and other fibrogenic nonparenchymal cells of the liver [5]. Important aspects of liver regeneration and repair are assumed to be similar and certain cellular and molecular mechanisms are shared by both [2]. How-
166
ROGGIN ET AL.: KUPFFER CELLS AND BILIARY OBSTRUCTION
ever, given the discordance of their starting points, each must possess its own idiosyncrasies. Whereas regeneration starts with normal remnants of the liver after acute injury, repair after chronic injury occurs in the milieu of a prolonged stimulation of inflammatory and fibrogenic elements that must first be inhibited and subsequently reversed. Unlike regeneration that primarily starts with hepatocyte proliferation, repair of cholestatic liver injury is linked to the simultaneous regression of both myofibroblasts and hyperplastic ductal epithelium [6]. Furthermore, excess types I, III, and IV collagen must be degraded to restore the threedimensional spatial relationships between parenchymal and nonparenchymal cells necessary to perform the normal physiologic and metabolic functions of the liver. In this regard, liver repair following chronic cholestatic injury is analogous to normal wound healing following complex injuries seen in the skin, bowel, lung, and other tissues of the body. We have previously demonstrated that experimental depletion of KC during injury with gadolinium chloride has been shown to exacerbate the fibroproliferative response to total biliary obstruction [7]. The absence of Kupffer cells also inhibits the stem cell compartment, which may act as a potential reservoir to replenish injured hepatocytes and ductular epithelium [7]. Therefore, we speculate that an intact population of differentiated, resident hepatic macrophages is necessary for successful and efficient repair following chronic injury. METHODS Animals. Adult male rats (250 –375 g, virus-free, HarlanSprague-Dawley, Indianapolis, IN) were housed three per cage in an artificial 12 h light/dark cycle with access to rat chow and water ad libitum according to the NIH Guide for the Care and Use of Laboratory Animals. Experiments were carried out under operative guidelines prescribed by the Institutional Animal Care and Use Committee (IACUC) at the Central Research Facility of Rhode Island Hospital and Brown University School of Medicine. Experimental design. The animals were divided into two groups. The study group received gadolinium (III) chloride hexahydrate (Sigma, St. Louis, MO) according to the protocol of Hardonk et al. [8]. To inactivate Kupffer cells during both injury and repair, a dose of 10 mg/kg was administered intravenously into the dorsal penile vein 24 h prior to both bile duct suspension and reversal surgery. The control group received an equivalent volume of saline at identical time points. Animals were randomized to receive either bile duct suspension [9] or manipulation (sham). To simulate chronic liver injury, bile duct suspension was perform for a period of 5 days. Our experience with this model has documented characteristic bridging fibrosis between adjacent portals after similar periods of injury [3]. Following reversal of the obstruction, animals were sacrificed at 0, 2, 4, and 7 days of repair. Disappearance of acholic stools, normalization of serum bilirubin, and restitution of normal hepatic histology confirmed reversibility of our model. In addition, body weights were recorded at each time point. Reversible bile duct suspension surgery. Using specific modifications of a procedure first presented by Posner et al. [9], rats were placed in an induction chamber insufflated with 100% O 2 at a flow
167
rate of 4 L/min and isoflurane USP anesthetic (Fort Dodge Animal Health, Fort Dodge, IA) via an Ohio-style vaporizer. Following induction, rats were weighed, marked, shaved, and transferred to an operating platform equipped with a nonrebreathing circuit face mask for delivery of a continuous mixture of oxygen and inhalational anesthetic. The abdomen was prepared with 70% ethyl alcohol (Pharmco, Brookfield, CT) and 10% providone-iodine solution (The Purdue Frederick Company, Norwalk, CT) and covered with a transparent sterile drape. Using aseptic technique, the peritoneal cavity was entered through a midline laparotomy incision. The liver was elevated cephalad and the duodenum was retracted caudad to expose the porta hepatis. The common bile duct was mobilized at its proximal extent and suspended by a silicone SURG-I LOOP (Scanlan International, Saint Paul, MN). The free ends of the loop were brought out through separate stab incisions in the skin 1 cm lateral to the midline at the apex of the wound. A 1-cm premarked segment of the loop was suture-ligated to each costochondral margin with 3-O polyglactin 910 suture (Ethicon, Somerville, NJ) and buried in the subcutaneous tissue with separate horizontal mattress sutures. The peritoneum and skin incisions were closed with running 4-O and 2-O polyglactin 910 sutures, respectively. Sham-operated animals underwent common bile duct manipulation without suspension; all other technical aspects of the procedure were identical to those of the obstructed animals. Postoperatively, the rats were transferred to clean cages under warming lights and observed. To minimize pain, chow was placed on the cage floor and water was administered through long sippers ad libitum. The rats were checked daily after surgery for stool characteristics and signs of morbidity. Decompression procedure. Rats were anesthetized with isoflurane, weighed, and prepared with 70% ethyl alcohol and 10% providone-iodine. Blood (1 cc) was collected via the tail vein and stored at ⫺4°C until ready for use. The paired stab incisions were sharply opened to expose the subcutaneous loop ends. Caudad retraction of the loop ends exposed the costochondral sutures, which were cut to free the loop from its attachments. The loop was carefully removed and the skin was closed with a horizontal mattress stitch using 2-O polyglactin 910 suture. Postoperative care was performed as previously described. Necropsy. Anesthetized rats were weighed and then sacrificed with an intraperitoneal injection of pentobarbital sodium USP 120 mg/kg (Abbott Laboratories, North Chicago, IL). The abdomen was widely opened and 3 cc of venous blood was collected from the inferior vena cava and stored at ⫺4°C until ready for use. Following hepatectomy, liver samples were fixed in 10% buffered formalin phosphate (Fischer Scientific, Fair Lawn, NJ) and subsequently mounted in paraffin and processed (RIH Central Research Labs) for immunohistochemistry. For frozen tissue sectioning, 0.5–1 cm 2 liver cubes were placed in plastic molds filled with Cyto-Tek tissue freezing media (Sakura, Torrance, CA). These sections were frozen in a hexane (Fischer Scientific) bath cooled with acetone (Fischer Scientific) and dry ice for 5 min. Liver sections (5 m) were cut with a Shur/Cut cryostat (Triangle Biomedical Sciences, Durham, NC), fixed in cold acetone for 10 min, and stored at ⫺80°C until ready for use. Staining and immunohistochemistry. Formalin-fixed, paraffin embedded, 5-m liver tissue sections were stained with H&E to identify bile duct hyperplasia, inflammatory infiltrate, and fibrosis. To demonstrate type I collagen deposition, serial sections were treated with trichrome stain, with the nuclear stain omitted for increased resolution of collagen fibrils. Monoclonal antibody anti-cytokeratin-19 (Amersham, Little Chalfont, Buckinghamshire, England), diluted 1:10 in phosphate buffered saline (PBS) with 10% goat serum (Sigma, St. Louis, MO), was applied to frozen liver sections. A FITC-conjugated IgG2b goat antimouse secondary antibody (Boehringer, Indianapolis, IN), diluted 1:100 in PBS, was used to label CK-19-positive bile duct epithelial cells under fluorescent microscopy (Nikon Microphot-FXA, Nikon, Tokyo, Japan).
168
JOURNAL OF SURGICAL RESEARCH: VOL. 90, NO. 2, MAY 15, 2000
Weight Change Induced by Reversible Biliary Obstruction
FIG. 1. Average serum direct bilirubin assay was performed to verify the reversibility of our model. Bile duct suspension increased the direct serum bilirubin in all experimental animals (*P ⱕ .05 vs sham). Following removal of the loop suspension (repair), the average serum bilirubin normalized to sham-operated levels.
Image analysis for collagen content. Trichrome-stained tissue sections were scanned (SprintScan 35 Plus, Polaroid, Cambridge, MA) into a PowerMac 9600 (Apple Computer, Inc., Cupertino, CA) using a PathScan Enabler (0.054) optical card (Meyer Instruments, Inc., Houston, TX). NIH Imaging Software Version 1.61 (NIH, Bethesda, MD) was used to create binary images from the red, green, and blue channels. An arithmetic function calculated the intersection between channels to generate a composite binary image. Up to 900,000 mm 2 of tissue per animal was analyzed, measured, and calibrated to a standardized scale. Results were expressed as average percentage collagen per square millimeter of liver. Quantification of serum direct bilirubin. Frozen serum samples were shielded from direct light, thawed to room temperature, and assayed for direct serum bilirubin using a commercial kit (Sigma Diagnostics, St. Louis, MO) according to the manufacturer’s instructions. Statistical analysis. The data were processed using superANOVA statistical software (Abacus Concepts Inc., Berkeley, CA). A one-way analysis of variance test (ANOVA) was used to compare serum bilirubin and collagen data. Post hoc analysis was performed using Scheffe’s and Fisher’s Protected LSD tests. The criterion for significance for all studies was P ⱕ 0.05. Results are expressed as mean ⫾ standard error of the mean.
RESULTS
Direct Serum Bilirubin Assay Compared to sham-operated animals (bilirubin ⫽ 0.15 mg/dL ⫾ 0.12, N ⫽ 4), 5 days of bile duct suspension (N ⫽ 25) increased direct serum bilirubin to 8.42 mg/dL ⫾ 0.77 (P ⱕ 0.05, Fig. 1). Removal of the obstructing loop normalized serum bilirubin at all time points during repair: 0.25 mg/dL ⫾ 0.13 at 2 days of repair (N ⫽ 6), 0.15 mg/dL ⫾ 0.13 at 4 days of repair (N ⫽ 6), and 0.36 mg/dL ⫾ 0.15 at 7 days of repair (N ⫽ 7); there was no statistically significant difference between sham serum bilirubin and the values from repairing animals (Fig. 1). In addition, gadolinium administration did not alter serum bilirubin from saline-treated animals at similar time points (data not shown).
In sham-operated animals (N ⫽ 4), average body weight increased 5 days after common bile duct manipulation: 287.1 ⫾ 7.0 g (preoperative weight) to 305.4 ⫾ 8.5 g. Saline-treated animals had an average preop weight of 300.8 ⫾ 10.0 g (N ⫽ 13) that decreased to 274.1 ⫾ 9.7 g (N ⫽ 13) following 5 days of common bile duct suspension. During repair the average body weight was 287.2 ⫾ 20.6 g at 2 days of repair (N ⫽ 3), 316.6 ⫾ 14.3 g at 4 days of repair (N ⫽ 3), and 290.36 ⫾ 11.78 g at 7 days of repair (N ⫽ 4). The weight change in gadolinium-treated animals was not statistically significant from the saline group at equivalent time points: 310.2 ⫾ 9.4 g (preop, N ⫽ 12), 278.7 ⫾ 8.0 g (reversal, N ⫽ 12), 289.6 ⫾ 9.8 g (Repair Day 2, N ⫽ 3), 297.5 ⫾ 11.3 g (Repair Day 4, N ⫽ 3), and 304.0 ⫾ 7.1 g (Repair Day 7, N ⫽ 3). Histologic Changes during Injury and Repair Sham-operated animals did not show evidence of increased fibrosis, inflammatory infiltrate, or bile duct hyperplasia 5 days after common bile duct manipulation (data not shown). Bile duct suspension altered the histologic hepatic architecture of both saline- and gadolinium-treated animals. Changes included bile duct hyperplasia, periportal bridging fibrosis surrounded by an intense inflammatory infiltrate, and a few scattered areas of hepatocellular necrosis with relative sparring of the pericentral areas (Fig. 2). Following removal of the obstructing loop, salinetreated animals showed gradual restitution of normal hepatic architecture (Fig. 2). By 7 days of repair, only scattered clusters of hyperplastic ducts could be identified; both the inflammatory infiltrate and fibrotic changes were comparable to those of sham-operated animals (data not shown). Repair in gadoliniumtreated animals was delayed as evidenced by the persistence of periportal extracellular matrix over time; even at 7 days of repair, bridging fibrosis and hyperplastic duct epithelium could still be identified throughout the hepatic lobule (Fig. 2). In addition, high-powered magnification of portal triads showed increased matrix distorting the normal spatial relationships among bile ducts, hepatic arterioles, and portal venules. Delayed Bile Duct Resorption Following Kupffer Cell Inactivation Periportal cytokeratin-19-positive bile ductal epithelium was increased in liver sections from both salineand gadolinium-treated animals following bile duct suspension (Fig. 3). Similar to that observed in H&E stained liver sections, resorption of hyperplastic epithelium was delayed in gadolinium-treated animals
ROGGIN ET AL.: KUPFFER CELLS AND BILIARY OBSTRUCTION
169
FIG. 2. Histological changes in saline- and gadolinium-treated rat liver sections in model of reversible biliary obstruction. Five days of bile duct suspension induced bile duct hyperplasia (white arrows), bridging fibrosis between adjacent portal tracts (PT), and an intense inflammatory infiltrate in both saline (A) and gadolinium (B) groups. Following 7 days of repair, saline-treated animals (C) exhibited normal appearing hepatic architecture with absence of fibrosis between PT and central veins (CV). The liver sections from the gadolinium group (D) showed persistent bile duct hyperplasia, increased extracellular matrix, and bridging fibrosis suggestive of delayed repair.
compared to saline-treated controls 7 days after removal of the loop (Fig. 3). Altered Collagen Metabolism during Repair Following Reversible Biliary Obstruction Common bile duct suspension increased type I collagen in trichrome-stained paraffin liver sections in both saline and gadolinium-treated rats (Fig. 4). Compared to saline-treated sham-operated animals, the percentage of collagen increased after bile duct suspension from 1.13%/mm 2 ⫾ 0.07 (N ⫽ 4) to 8.26%/mm 2 ⫾ 1.32 (N ⫽ 3, Fig. 5A). Following removal of the loop, the percentage of collagen in liver sections progressively decreased over time: 4.83%/mm 2 ⫾ 0.58 at 2 days of repair (N ⫽ 3), 3.412%/mm 2 ⫾ 0.40 at 4 days of repair (N ⫽ 3), and 2.33%/mm 2 ⫾ 0.34 at 7 days of repair (N ⫽ 4, Fig. 5A). Gadolinium chloride altered collagen metabolism resulting in progressive increase in collagen staining over time. Following bile duct suspension (5.867%/ mm 2 ⫾ 1.43, N ⫽ 3), the amount of collagen did not decrease compared to saline-matched controls: 6.44%/ mm 2 ⫾ 1.59 at 2 days of repair (N ⫽ 3), 6.27%/mm 2 ⫾ 1.85 at 4 days of repair (N ⫽ 3), 8.79%/mm 2 ⫾ 2.17 at 7 days of repair (N ⫽ 3, P ⫽ 0.0003, Fig. 5B).
DISCUSSION
Globally, the morbidity and mortality of congenital or acquired liver diseases that might benefit from repair far exceed those of acute injuries which require the powers of liver regeneration. Unchecked, chronic liver injuries lead to lethal cirrhosis. Although transplantation for end-stage liver disease has had significant success [10], direct attacks on the sources of liver injury and augmentation of repair will benefit patients awaiting transplantation and those in whom transplant is not feasible [11]. The spectrum of liver injury includes infectious, metabolic, inflammatory, toxic, and cholestatic insults. Regardless of the pathogenic stimulus, most forms lead to scar formation [12]. Repeated insults (e.g., chronic ethanol abuse or viral hepatitis) or inefficient repair (e.g., malnutrition or hepatocellular carcinoma) can also stimulate fibrosis as the liver attempts to limit hepatocellular damage. In fact, the liver’s response to chronic injury is similar to other models of wound healing [5]. Therapeutic strategies to reverse this process often fail to interrupt this progressive cycle leading to the sustained accumulation of extracellular matrix characteristic of cirrhosis [13]. Chronic cholestatic liver disease is a common, pri-
170
JOURNAL OF SURGICAL RESEARCH: VOL. 90, NO. 2, MAY 15, 2000
FIG. 3. Delayed resorption of CK-19-positive bile duct epithelium in gadolinium-treated animals. Loop suspension induced comparable bile duct hyperplasia in both saline (A) and gadolinium (B) groups. At 7 days of repair, while most of the hyperplastic bile ducts regressed in the saline group (C), the gadolinium-treated rats (D) had increased numbers of proliferating bile duct epithelium, indicating delayed repair.
mary or secondary injury that predisposes adults and children to the development of cirrhosis [14]. Postcanalicular biliary obstruction leads to wellcharacterized bile duct epithelial cell proliferation, collagen deposition, and progressive derangement of hepatic architecture [15]. Successful techniques of biliary decompression for those with mechanical forms of extrahepatic obstruction have increased survival from previously lethal cholestatic disease [16]. Current research efforts to elucidate the cellular and molecular mechanisms of repair have been hindered by a paucity of clinically relevant animal models of liver repair, the poor translation of in vitro results to human pathophysiology, and the logistics of controlling the complex intercellular interactions during repair. Nevertheless, we have modified a model of reversible biliary obstruction [9] to study repair after chronic cholestatic injury in the rat. This model is unique in that it allows us to simulate a predictable chronic liver injury and subsequent repair under controlled conditions, which have been reproducible in our lab. In addition, our model avoids the technical difficulties and morbidity associated with bilioenteric anastomoses and repeated laparotomies in small laboratory animals that have limited the utility of other models of liver repair. Following experimental bile duct suspension, the elevation of direct serum bilirubin and histologic evidence of bile duct proliferation, bridging fibrosis, and inflammatory infiltrate confirm that our model is ca-
pable of simulating progressive cholestatic liver injury [3]. In addition, the rapid normalization of serum bilirubin and restitution of normal hepatic architecture in saline controls indicate that our model was in fact reversible. Although cholestasis caused a transient and reversible loss of body weight, this is most likely related to the trauma of surgery. Metabolic studies have shown that weight loss induced by bile duct ligation is transient and reversible [17]. Repair after cholestatic injury requires the systematic reversal of a number of cellular and molecular alterations stemming from bile duct obstruction [2]. Impediments to successful repair include the sustained proinflammatory cytokine feedback loops, the proliferation of bile duct epithelium, and the resorption of abundant extracellular matrix, including collagen, which makeup scar. We and others speculate that the macrophage is integral to this process [18]. As a key component of the reticuloendothelial system, the resident hepatic macrophage can clear bacteria from the bloodstream, mobilize host defenses through the secretion of inflammatory mediators, and secrete matrix metalloproteinases, which degrade the extracellular matrix [19]. Since the activated macrophage is responsible for the secretion of these mediators, they represent a potential target for anti-fibrogenic therapy. In the liver, as in many other tissues including, but not limited to, the lung, skin, and bowel, both parenchymal and immune-mediated injuries lead to the ac-
ROGGIN ET AL.: KUPFFER CELLS AND BILIARY OBSTRUCTION
171
and efficient repair depends upon the presence of an intact resident macrophage population. To test our hypothesis that repair after cholestatic injury is dependent on an intact liver macrophage population, we used gadolinium chloride to selectively deplete Kupffer cells. Gadolinium chloride is a rare earthmetal salt, which is selectively, absorbed by large zone I hepatic macrophages (KC) [20]. Gadolinium chloride, which is soluble at pH 5 or less, forms insoluble colloid complexes in the neutral pH of blood and is readily phagocytosed by KC [21]. Once inside the cells, elemental gadolinium is liberated, which inhibits phagocytosis and causes cell death [8]. Critical analysis of this technique has been documented in numerous animal
FIG. 4. Type I collagen in trichrome-stained paraffin sections in sham (A), saline (C), and gadolinium-treated rats (E) was processed and digitally converted to composite binary images; collagen is displayed as white lines in the black images on panels B (sham), D (saline), and F (gadolinium). Five days after bile duct manipulation, this representative sham-operated animal (B) showed a paucity of type I collagen. By 7 days of repair, the representative binary image from a saline-treated rat (D) showed less overall type I collagen than in a gadolinium-treated animal (F) at the same point during repair.
tivation of macrophages [18]. Through both cellular interaction and secretion of intercellular chemical signals, macrophages transform quiescent stellate cells into their activated phenotype, the myofibroblast [5]. This process leads to the net deposition of collagen and other components of the extracellular matrix which makeup scar. Successful repair following injury is dependent on a coordinated series of events, which reverse this process. It is our hypothesis that successful
FIG. 5. (A) Average percentage collagen per square millimeter of liver tissue in saline-treated animals during both injury and repair versus sham-operated saline controls (N ⱖ 3 at each time point). Following removal of the loop, the average percentage collagen progressively diminished over time, approaching sham levels by 7 days of repair. Abbreviations: SHAM (saline shams), Nd0 (Necropsy Day 0/saline group at reversal), Nd2 (Necropsy Day 2/saline group at 2 days of repair), Nd4 (Necropsy Day 4/saline group at 4 days of repair), Nd7 (Necropsy Day 7/saline group at 7 days of repair). (B) Average percentage collagen per square millimeter of liver tissue in gadolinium-treated animals during both injury and repair versus sham-operated gadolinium controls (N ⱖ 3 at each time point). Gadolinium chloride-induced Kupffer cell depletion delayed liver repair after bile duct suspension. Following reversal of the loop suspension, the average collagen content in rat livers increased over time and approached statistical significance by 7 days of repair (#P ⱕ 0.05 versus the saline group). Abbreviations: SHAM (saline shams), Nd0 (Necropsy Day 0/gadolinium group at reversal), Nd2 (Necropsy Day 2/gadolinium group at 2 days of repair), Nd4 (Necropsy Day 4/gadolinium group at 4 days of repair), Nd7 (Necropsy Day 7/gadolinium group at 7 days of repair).
172
JOURNAL OF SURGICAL RESEARCH: VOL. 90, NO. 2, MAY 15, 2000
models and it is considered to be a suitable approach for studying the in vivo effect of KC depletion [8]. We used gadolinium according to the protocol of Hardonk; the drug was administered 1 day prior to bile duct suspension and again 1 day prior to reversal (repair). Since previous studies have shown that systemic monocytes repopulate the liver 4 days after treatment and that the new macrophages are also vulnerable to gadolinium-depletion [8], this dosing schedule allowed us to sequentially eliminate KC during both injury and repair. Gadolinium-induced elimination of KC resulted in delayed repair as evidenced by persistent bile duct hyperplasia, a profound inflammatory infiltrate, and accumulation of collagen and extracellular matrix. KC regulate the deposition of collagen both by the secretion of macrophage-specific matrix metalloproteinases and through paracrine or cell-to-cell stimulation of other collagen and matrix-producing cells [22]. Both activated stellate cell, or myofibroblasts, and bile duct epithelial cells are potential targets for KC modulation. The phenotypic changes in the macrophage population in the liver during both injury and repair may influence the quality and quantity of extracellular matrix observed in our model. In vitro experiments involving the isolation of KC from both normal and injured animals treated with gadolinium will be required to characterize the production of metalloproteinases and their inhibitors and their impact on collagenolytic or gelatinolytic activity. While the use of gadolinium chloride remains an effective strategy for in vivo KC depletion, its potential deleterious effects on other aspects of liver physiology have led to controversy over its use in experimental models. Several studies in animal sepsis models have described in vivo gadolinium chloride-induced alterations of cytochrome P450 activity [23], coagulation cascade [24], hepatic microvascular perfusion [25], and serum levels of proinflammatory cytokines [25]. In our model of reversible biliary obstruction, gadolinium did not appear to cause any of these untoward effects. There were no histologic alterations induced in gadolinium-treated, sham-operated animals compared to saline controls. In addition, gadolinium did not alter the normalization of serum bilirubin or increase mortality over matched saline-treated rats. Several sepsis studies have also demonstrated reduced mortality to a lethal LPS challenge following pretreatment with gadolinium chloride [24, 26]. Therefore, although cognizant of its potential as a confounding variable, we are confident that the toxic side effects of gadolinium were not responsible for the delayed repair observed in the treated group. Furthermore, a recent report of increased collagen deposition by dexamethasone-induced Kupffer cell depletion in a model of biliary obstruction [27] confirms that this observation cannot solely be
explained by the toxicity of gadolinium chloride and that KC may protect the liver from cholestatic injury. Therefore, it is our conclusion that not only is the Kupffer cell a positive modulator of repair, but the presence of an intact population of resident hepatic macrophages remains a potential target for antifibrotic therapies designed to augment repair after cholestatic injury. The speculative nature of Kupffer cells as the primary positive modulator of liver repair after cholestatic injury requires further experimental confirmation and validation. We propose a model that during injury and subsequent repair, a distinct macrophage population regulates the secretion of chemoattractants, proinflammatory, and inhibitory cytokines, which mobilize host defenses, limit parenchymal cell damage, and initiates repair. Elucidation of this cellular and molecular process will help design therapies designed to augment repair following biliary decompression for cholestatic injury in surgical patients. REFERENCES 1.
2.
3.
4. 5.
6.
7.
8.
9.
10. 11. 12.
13.
Tracy, T., Goerke, M., Bailey, P., Sotelo-Avila, C., and Weber, T. Growth-related gene expression in early cholestatic liver injury. Surgery 114: 532, 1993. Tracy, T., and Fox, E. Molecular and cellular control points in pediatric liver injury and repair. Semin. Pediatr. Surg. 5: 175, 1996. Aldana, P. R., Goerke, M. E., Carr, S. C., and Tracy, T. F. The expression of regenerative growth factors in chronic liver injury and repair. J. Surg. Res. 57: 711, 1994. Friedman, S. Stellate cell activation in alcoholic fibrosis—An overview. Alcohol. Clin. Exp. Res. 23: 904, 1999. Friedman, S. L. The cellular basis of hepatic fibrosis. Mechanisms and treatment strategies. N. Engl. J. Med. 328: 1828, 1993. Ramm, G. A., Carr, S. A., Li, L., Vogler, C. A., Britton, R. S., Bacon, B. R., and Tracy, T. F. Regression of myofibroblasts during hepatic repair following chronic cholestatic liver injury in rats. Liver, 2000, in press. Olynyk, J. K., Yeoh, G. C., Ramm, G. A., Clarke, S. L., Hall, P. M., Britton, R. S., Bacon, B. R., and Tracy, T. F. Gadolinium chloride suppresses hepatic oval cell proliferation in rats with biliary obstruction. Am. J. Pathol. 152: 347, 1998. Hardonk, M. J., Kijkhuis, F. W. J., Hulstaert, C. E., and Koudstaal, J. Heterogeneity of rat liver and spleen macrophages in gadolinium chloride-induced elimination and repopulation. J. Leukocyte Biol. 52: 296, 1992. Posner, M., Burt, M., Stone, M., Han, B., Warren, R., Vydelingum, N., and Brennan, M. A model of reversible obstructive jaundice in the rat. J. Surg. Res. 48: 214, 1990. Wall, W. J. Liver transplantation: Past accomplishments and future challenges. Can. J. Gastroenterol. 13: 257, 1999. Alonso, M. H., and Ryckman, F. C. Current concepts in pediatric liver transplant. Semin. Liver Dis. 18: 295, 1998. Li, D., and Friedman, S. L. Liver fibrogenesis and the role of hepatic stellate cells: New insights and prospects for therapy. J. Gastroenterol. Hepatol. 14: 618, 1999. Kovacs, E. J., and DiPetro, L. A. Fibrogenic cytokines and connective tissue production. FASEB J. 8: 854, 1994.
ROGGIN ET AL.: KUPFFER CELLS AND BILIARY OBSTRUCTION 14. 15.
16. 17.
18.
19.
20. 21.
McGill, J. M., and Kwiatkowski, A. P. Cholestatic liver diseases in adults. Am. J. Gastroenterol. 93: 684, 1998. Morris, J. S., Gallo, G. A., Scheuer, P. J., and Sherlock, S. Percutaneous liver biopsy in patients with large bile duct obstruction. Gastroenterology 68: 750, 1975. Ahrendt, S. A., and Pitt, H. A. A history of the bilioenteric anastomosis. Arch. Surg. 125: 1493, 1990. Gouma, D. J., Roughneen, P. T., Kumar, S., Moody, F. G., and Rowlands, B. J. Changes in nutritional status associated with obstructive jaundice and biliary drainage in rats. Am. J. Clin. Nutr. 44: 362, 1986. Tracy, T. F., Dillon, P., Fox, E. S., Minnick, K., and Vogler, C. The inflammatory response in pediatric biliary disease: Macrophage phenotype and distribution. J. Pediatr. Surg. 31: 121, 1996. Winwood, P. J., and Arthur, M. P. J. Kupffer cells: Their activation and role in animal models of liver injury and human liver disease. In Seminars in Liver Disease. New York: Thieme Med. Pub., 1993. Pp. 50 –59. Lazar, G. The reticuloendothelial blocking effect of rare earth metals in rats. J. Reticuloendothel. Soc. 13: 231, 1989. Spencer, A. J., Wilson, S. A., Batchelor, J., Reid, A., Rees, J.,
22. 23.
24.
25.
26.
27.
173
and Harpur, E. Gadolinium chloride toxicity in the rat. Toxicol. Pathol. 25: 245, 1997. Arthur, M. J. Degradation of matrix proteins in liver fibrosis. Pathol. Res. Pract. 190: 825, 1994. Badger, D. A., Kuester, R. K., Sauer, J.-M., and Sipes, I. G. Gadolinium chloride reduces cytochrome P450: Relevance to chemical-induced hepatotoxicity. Toxicology 121: 143, 1997. Ruttinger, D., Vollmar, B., Kempter, B., and Messmer, K. Failure of Kupffer cell blockade to prevent disseminated intravascular coagulation in endotoxemic rats despite improved survival. Langenbecks Arch. Chir. 383: 75, 1998. Ruttinger, D., Vollmar, B., Wanner, G. A., and Messmer, K. In vivo assessment of hepatic alterations following gadolinium chloride-induced Kupffer cell blockade. J. Hepatol. 25: 960, 1996. Vollmar, B., Ruttinger, D., Wanner, G. A., Leiderer, R., and Menger, M. D. Modulation of Kupffer cell activity by gadolinium chloride in endotoxemic rats. Shock 6: 434, 1996. Melgert, B. N., Weert, B., Molema, G., Meijer, D. K., and Poelstra, K. Chronic treatment of bile duct ligated rats with dexamethasone selectively delivered to Kupffer cells. Hepatology 30: 329A, 1999.