In alcoholic steatohepatitis, hepatic microvascular changes have pathogenic significance for hepatocellular function, perisinusoidal fibrosis, and portal hypertension. It is unclear whether similar changes occur in other forms of steatohepatitis. We therefore examined whether hepatic microvascular dysfunction occurs in fibrosing steatohepatitis induced by feeding mice a high-fat methionine- and choline-deficient (MCD) diet. Using in vivo microscopic—as well as histological and electron microscopic—methods, together with measurements of alanine aminotransferase (ALT), lipid content, and oxidative stress, hepatic microvascular structure and function were studied in relation to inflammatory and fibrotic changes during evolution of steatohepatitis. At 3 weeks of MCD diet intake, serum ALT was elevated and hepatic steatosis was pronounced. By 5 weeks, necroinflammatory change was noteworthy, and by 8 weeks perisinusoidal fibrosis was established. Compared with mice receiving the high-fat diet supplemented with methionine and choline (controls), levels of hepatic lipid and lipoperoxides were elevated at 3 weeks and beyond. The numbers of perfused sinusoids were significantly reduced at each time point. Enlarged, fat-laden hepatocytes together with perivascular fibrosis narrowed sinusoidal lumens, making vessels tortuous and impairing sinusoidal perfusion. At 3 and 5 weeks, MCD diet caused significant increases in phagocytic activity of macrophages in centrilobular regions. By 8 weeks, macrophage activity was less striking, but the number of leukocytes adherent to the sinusoidal lining had increased 5-fold compared with controls. In conclusion, these results are consistent with a dysfunctional hepatic microvasculature. Thus, microvascular changes may contribute to progressive liver injury in metabolic and toxic forms of steatohepatitis. (HEPATOLOGY 2004;40:386–393.)
Steatohepatitis is a form of liver pathology in which steatosis is accompanied by liver cell injury, predominantly lobular inflammation of mixed cell type, and resultant pericentral and perisinusoidal fibrosis. The archetypical example is alcoholic hepatitis (ASH), but metabolic and toxic etiologies cause steatohepatitis that is histologically remarkably similar. Among the clinical syndromes of steatohepatitis, the most common is that associated with central obesity, insulin resistance, type 2 diabetes, hypertriglyceridemia and other features of the insulin resistance (metabolic) syndrome.1–8 This syndrome, nonalcoholic steatohepatitis (NASH), is the most common liver disorder in the United States, Australia, and most other countries. Histological similarities between NASH and ASH are best explained by the two diseases sharing pathophysiological processes. Understanding such disease mechanisms, and particularly those that result in progressive fibrogenesis and propagation of liver injury, are pivotal to devise treatments that prevent portal hypertension, cirrhosis, and liver failure.
In alcoholic liver disease, accumulation of lipid droplets in the hepatic parenchymal cells causes these cells to enlarge. It has been suggested that this enlargement of parenchymal cells may affect the circulation of blood through the hepatic sinusoids, which in turn may exacerbate hepatic injury by depriving centrilobular regions of oxygen and nutrients. Indeed, this has been reported in genetically obese Zucker rats.9–13 However, several other factors also may contribute to the alteration of blood flow through the hepatic microcirculatory unit in ASH. These include activation of Kupffer cells with release of vasoactive intermediates (endothelins, reactive oxygen species [ROS], nitric oxide, prostanoids) as well as proinflammatory and profibrogenic cytokines, changes to the endothelial cells, activation and contraction of the perisinusoidal stellate cells, and deposition of collagen and other extracellular matrix proteins in the space of Disse.14, 15 The latter change encases the sinusoids with a connective tissue envelope, a process referred to as capillarization. This interferes with the diffusion of oxygen and impairs exchange of nutrients and waste products between hepatocytes and the intravascular compartment.
At present, it is unclear whether any of the above changes occurs in other forms of steatohepatitis, such as NASH. However, in this metabolic type of steatohepatitis, the pattern of hepatic fibrosis resembles ASH, but the inflammatory reaction is generally less exuberant. If microvascular changes do occur in other forms of steatohepatitis, capillarization of the sinusoid and its effects on parenchymal cells could explain why liver failure can eventually ensue, despite apparent resolution of steatosis and hepatitis.
Animal models of human diseases are a useful tool to study pathogenic processes, particularly those that require the intact organ for physiologically meaningful interpretation. There are no entirely satisfactory animal models of steatohepatitis with fibrosis that can be attributed to insulin resistance. However, typical steatohepatitis resulting in perisinusoidal fibrosis, together with at least transient insulin resistance, does occur when rodents are exposed to a high-fat diet and also rendered deficient in hepatic methionine. The latter can be achieved either by deletion of methionine adenosyltransferase type 1A16 or by feeding a diet deficient in both choline and methionine. In the nutritional model, rats or mice develop steatosis with severe oxidative stress and hepatocellular injury during the first 10 days of dietary feeding. This is followed by diffuse and focal lobular inflammation with perisinusoidal and pericentral fibrosis,17–19 thus strikingly resembling the liver pathology of human NASH.
In the present study, we tested the hypothesis that hepatic microvascular dysfunction is a feature of steatohepatitis. To do this, we examined the dynamic and spatial development of microvascular alterations in mice during the evolution of fibrosing steatohepatitis induced by a high-fat, methionine- and choline-deficient (MCD) diet. This was accomplished by examining the livers of anesthetized mice using established in vivo microscopic methods to determine: (1) dimensional changes in hepatic sinusoids, including their functional obliteration; (2) alterations in the patterns and rates of sinusoidal blood flow; (3) leukocyte adherence to the sinusoidal endothelium; and (4) numbers and phagocytic function of Kupffer cells and other hepatic macrophages. The changes were studied first at an early phase (3 weeks), when the predominant pattern of injury is steatosis, then at an intermediate time (5 weeks), when lobular inflammation is fully developed. Finally, we examined livers at 8 weeks of dietary feeding, when steatohepatitis and perisinusoidal fibrosis are fully established. At each of these times, we compared results of microvascular studies with measures of liver injury (serum alanine aminotransferase [ALT]) and pathology, and with biochemical measures of hepatic lipid content and oxidative stress. We also studied the ultrastructural concomitants of microvascular alterations by electron microscopy.
ASH, alcoholic hepatitis; NASH, nonalcoholic steatohepatitis; ROS, reactive oxygen species; MCD, methionine- and choline-deficient; ALT, alanine aminotransferase; SEC, sinusoidal endothelial cell; TEM, transmission electron microscopy; TBARS, thiobarbituric acid-reactive substances.
Materials and Methods
All experiments were performed in adherence to the National Institutes of Health guidelines for the use of experimental animals and followed protocols approved by the Animal Care and Ethics Committee of the Western Sydney Area Health Service.
Animals and Experimental Protocols.
Male C57Bl/6 mice initially weighing approximately 25 g were used as experimental animals and fed a 10% lipogenic MCD diet, as reported elsewhere.18, 19 Mice receiving the MCD diet supplemented with DL-methionine (3 g/kg) and choline bitartrate (2 g/kg) served as controls. Both the MCD (ICN 960439) and control (ICN 960441) diet were obtained commercially from ICN (Aurora, OH). The mice were studied 3, 5, and 8 weeks after being placed on the diet, at which times steatosis, necroinflammatory changes, and then fibrosis are evident. There were at least 5 mice in each experimental group.
In Vivo Microscopy.
Hepatic microvascular alterations were examined using established high resolution in vivo microscopic methods under ketamine/xylazine anesthesia (100/10 mg/kg body weight, intraperitoneal), as reported previously.20 Briefly, a compound binocular microscope (Olympus BH-2, Olympus Australia, North Ryde, Victoria, Australia) adapted for in vivo microscopy was equipped to provide either transillumination or epi-illumination as well as video microscopy using a JVC color camera (TK-870E), video monitor (TM-210PS), and S-VHS video recorder (BR-S600E) (JVC-Australia, Kingsgrove, New South Wales, Australia). The liver was exteriorized through a left subcostal incision and positioned over a window of optical-grade mica in a specially designed tray mounted on a microscopic stage. The tray provided for the drainage of irrigating fluids, and the window overlaid a long working distance condenser. The liver was covered by a piece of Saran Wrap (Dow Chemical Co., Midland, MI), which held it in position and limited movement. Homeostasis was insured by a constant suffusion of the organ with Ringer's solution maintained at body temperature. With the ×80 water immersion objective (Leitz, Wetzlar, Germany) employed for these studies, the resolution was 0.3 to 0.5 μm. Microvascular events were observed and recorded for at least 30 seconds for subsequent off-line analysis.
The phagocytic function of hepatic macrophages was assessed by measuring the uptake of fluorescent 1.0 μm latex particles (Fluoresbrite-fluorescent monodispersed polystyrene microspheres; Polysciences, Warrington, PA: excitation 490 nm/emission 530 nm) by individual cells. The latex particles were diluted 1:10 with sterile saline and injected into a mesenteric vein using a 30-gauge lymphangiography needle (Becton Dickinson, Franklin Lakes, NJ). The distribution and relative number of macrophages was measured by counting the number of cells that phagocytosed latex particles in a standardized microscopic field (4,125 μm2) 15 minutes after injection of latex. Phagocytic cells firmly affixed to the sinusoidal wall, including recruited monocytes/macrophages and resident macrophages (Kupffer cells), were considered macrophages. To assess regional distribution, the number of phagocytic macrophages per microscopic field was counted in 10 periportal and 10 centrilobular regions in each animal. The relative adequacy of blood perfusion through the sinusoids was evaluated by counting the number of the sinusoids containing blood flow in the same microscopic fields in which the numbers of phagocytic macrophages were determined. Transmitted brightfield and epi-illuminated fluorescence images were obtained simultaneously to permit imaging of the phagocytosis of the fluorescent latex beads and images of the sinusoidal wall and blood flow at the same time. Because reduced perfusion of individual sinusoids can limit the delivery of the latex particles to phagocytic cells within those vessels, the ratio of macrophages that phagocytosed latex particles to sinusoids containing flow was used as an overall mean index of phagocytic activity of macrophages per microscopic field.
To examine the interaction of leukocytes with the sinusoidal wall, quantification of leukocytes adhering to the endothelial lining of sinusoids was calculated by counting the number of leukocytes per 100-μm length of sinusoid in the same microscopic fields. A leukocyte was defined as adhering to the sinusoidal wall if it remained stationary for at least 30 seconds. The swelling of hepatic sinusoidal endothelial cells (SECs), which is an indication of activation and/or injury, was assessed by counting the numbers of swollen cells whose nuclear regions protruded across one-third or more of the lumen in the same microscopic fields.14, 21–24 The results were averaged and the data were represented as the average numbers of the parameters in each animal.
Following high-resolution in vivo microscopic studies, the hepatic microcirculation was visualized under lower magnification (×10 water immersion objective; Leitz, Wetzlar, Germany) to gain an overview of the hepatic microvasculature. In particular, the overall perfusion was assessed to determine if it was relatively uniform or if it was patchy. The flow and dimensional characteristics in portal and central venules were also evaluated.
Liver samples of the animals studied were collected following euthanasia and fixed in 10% formalin-buffered solution. Paraffin sections were stained with hematoxylin-eosin for histopathological evaluation and Sirius Red for evaluating the presence of connective tissue. The tissue sections were examined and photographed at ×100 to ×400 using a Leica photomicroscope (Leica, Deerfield, IL).
In at least 3 of the mice from each experimental group, routine methods were used to prepare liver specimens for transmission electron microscopy (TEM).20 Livers were fixed by perfusion of portal veins with 0.1 mol cacodylate buffer to wash out blood, and subsequently, to obtain fixation with 1.5% glutaraldehyde in 0.1 mol cacodylate buffer, pH 7.4. For TEM, small pieces of liver were washed in buffer, postfixed with 1% osmium tetroxide in 0.1 mol cacodylate buffer at 4°C, dehydrated through a graded series of ethanol solutions, briefly rinsed in propylene oxide, and embedded in epoxy resin. Thin sections were cut on a Reichert Ultracut microtome (Leica, Deerfield, IL) and examined and photographed using a Philips CM-12S electron microscope (Philips Electronic Instruments, Mahwah, NJ).
Plasma ALT and Hepatic Content of Total Lipid and Lipoperoxides.
Following in vivo microscopic studies, blood samples were collected from the inferior vena cava and stored at −70°C until required for determination of ALT activity. ALT levels were determined by the Department of Clinical Chemistry, Westmead Hospital, using automated procedures. Samples of liver were stored at −70°C until used for measurement of total lipid content and total lipoperoxides, the latter being a robust marker of oxidative stress. Liver triglycerides were assayed using triglyceride GPO-Trinder reagents (Sigma Diagnostics, St. Louis, MO). Liver lipoperoxides levels were estimated as thiobarbituric reactive substances (TBARS) using 1,1,3,3-tetramethopropane as a standard (Sigma) as previously described.18
Numerical data are expressed as mean ± SEM. Statistical analyses were performed using ANOVA followed by a Student-Newman-Keuls test for multiple groups. The 95% confidence level (P < .05) was considered significant.
Development of Steatohepatitis and Oxidative Stress in Mice Fed the MCD Diet.
As previously reported, mice fed the MCD diet lost weight compared to those fed the control high-fat diet; however, they otherwise remained active throughout the course of dietary feeding. Liver histology at 3, 5, and 8 weeks confirmed the early and continued presence of steatosis and necroinflammatory changes, while established fibrosis was evident at week 8 (Figs. 1 and 2). At 3 weeks, hepatic parenchymal cells were filled with multivesicular fat, and focal hepatocyte necrosis was seen in the centrilobular region (Figs. 1A and 2A). At 5 weeks, there was a marked increase in unilocular fat droplets, centrilobular necrosis was more conspicuous, and there were occasional fibrotic deposits (Figs. 1B and 2B). By 8 weeks, unilocular fat predominated, and significant centrilobular necrosis was accompanied by infiltration of leukocytes (Figs. 1C and E) and scattered fibrosis (Fig. 2C). ALT levels in MCD-fed mice were significantly increased at all experimental time points compared with controls (Fig. 3A). Total hepatic lipid content increased significantly at 3 weeks and beyond (Fig. 3B). Total lipid peroxides (TBARs; Fig. 3C) were significantly increased at 3 and 5 weeks, and especially at 8 weeks, indicating high levels of oxidative stress.
Changes in Sinusoidal Caliber, Orientation, and Patency.
At all experimental times, in vivo microscopy showed dramatically enlarged lipid-laden hepatocytes compressing and narrowing the lumen of sinusoids, particularly in the centrilobular regions (Figs. 4B and 5A). This gave these vessels a tortuous configuration and appeared to greatly impair sinusoidal perfusion. Flow rates were frequently slow, with cells appearing distinct in contrast to controls, in which flowing cells were blurred. The sluggish sinusoidal perfusion in MCD-fed mice may not be solely attributable to sinusoidal compression because portal and central venules also appeared to be narrowed.
Quantitative assessment of in vivo microscopy revealed that a significant proportion of sinusoids in MCD-fed mouse livers contained no blood flow, a change not seen in livers of animals fed the control diet. Thus, in this dietary steatohepatitis model, numbers of perfused sinusoids were significantly reduced, by 13%, 19%, and 10% at 3, 5, and 8 weeks, respectively (Fig. 5B).
Adherent Leukocytes and Phagocytically Active Hepatic Macrophages.
During the initial 5 weeks, there were no significant differences in the numbers of leukocytes adhering to sinusoids between MCD- and control diet-fed mice. However, at 8 weeks the number of adherent leukocytes in MCD-fed mouse liver was increased 5-fold compared with control liver (Fig. 6A). There was no evidence of swelling of the sinusoidal endothelium during the 8 weeks of MCD (Fig. 6B).
The MCD diet caused 109% and 121% increases in the phagocytic activity of macrophages in centrilobular regions at 3 and 5 weeks, respectively (Figs. 4E and 7). In contrast, dietary feeding reduced this activity in both periportal and centrilobular regions at 8 weeks by 31% and 43%, respectively (Figs. 4F and 7).
Electron microscopy demonstrated loss of fat droplets in stellate cells associated with the appearance of collagen bundles in the perisinusoidal space by 5 weeks of the MCD diet (Fig. 8A); collagen deposition became more prominent after 8 weeks (Fig. 8B). There was the impression that there may be a loss of fenestrae, but this is difficult to quantify and confirmation of this awaits examination by SEM of another series of experimental animals.
The enhanced macrophage activity evident in centrilobular zones by in vivo microscopy was confirmed by TEM at 3 and 5 weeks (Fig. 8C) but was again shown to have diminished by week 8 (Fig. 8D). During these ultrastructural studies of the hepatic architecture, it was also confirmed that hepatic parenchymal cells distended with fat droplets narrowed the sinusoid lumens to make them tortuous (Fig. 8D).
Presence of Microvascular Changes in Dietary Steatohepatitis.
The findings of this study confirm our hypothesis that substantial microvascular changes are present in steatohepatitis caused by factors other than ethanol toxicity. These changes appear to arise initially from the effects of enlarged hepatocytes, swollen with accumulated lipid, compressing sinusoidal lumens, as reported previously in genetically obese Zucker rats.9–13 This, together with collagen deposits in the space of Disse, renders sinusoids tortuous and they become inefficient conduits of blood (or at least of erythrocytes), with resulting impairment of tissue perfusion. That the microvascular changes were secondary to lipid accumulation and oxidative stress in parenchymal cells is suggested by the lack of swollen SECs, which is an early marker of SEC injury14, 22, 23 at any time point (Fig. 6B), and a lack of inflammation, as indicated by leukocyte adhesion to the SEC until significant parenchymal changes already occurred (Fig. 6A).
By in vivo microscopy, the most dramatic change seen was in the number of sinusoids that completely failed to allow passage of erythrocytes. While the in vivo microscopic observations were limited to a sample of the surface of the liver, they are thought to be representative of the entire organ, which is composed of multiple similar functional units (lobules) whose histopathology throughout the liver under these conditions appeared similar. No differences in the responses of the MCD+ or MCD− mice to the anesthetic were seen.
Causes and Consequences of Ineffective Sinusoidal Perfusion.
Dysfunctional (or effectively obliterated) sinusoids were rarely present in control livers but increased significantly from 13% at 3 weeks of dietary feeding to 19% at 5 weeks. Between 5 and 8 weeks, the number of leukocytes adhering to the sinusoidal lining increased substantially, consistent with the appearances on light microscopy of exuberant parenchymal mixed cellular inflammation. Such adherence of leukocytes to endothelial cells is characteristic of an hepatic microvascular inflammatory response.14, 15 In other work, we have shown that hepatic expression of the intercellular adhesion molecules Intercellular Adhesion Molecule-1 (ICAM-1) and Vascular Cell Adhesion Molecule (VCAM) also are increased in MCD diet-induced steatohepatitis.25 We have attributed such changes to up-regulation of nuclear factor κB (NF-κB), in turn activated in the liver by the presence of severe oxidative stress. The demonstration that this secondary inflammatory response is focused, at least partially, on the microvasculature of the liver indicates one possible mechanism whereby metabolic oxidative stress attracts an inflammatory response with the potential to perpetuate liver injury by means of changes in liver perfusion. Although we did not measure oxygen delivery in the present study, others have demonstrated that analogous microvascular changes affect tissue oxygen concentration in the centrilobular region of livers of obese Zucker rats.12
Macrophage Activity in Experimental Steatohepatitis.
Concomitant with the alterations in sinusoidal blood flow observed in the present study, significant increases in phagocytic activity of hepatic macrophages were evident in the centrilobular region at 3 and 5 weeks of the MCD diet. Since these cells in their activated state are known to release ROS and nitroradicals as well as cytokines and vasoactive prostanoids they are also likely to contribute importantly to parenchymal injury, the microvascular inflammatory response, and the activation of adjacent stellate cells with resultant loss of lipid droplets and increased production of collagen. Further studies are clearly indicated to elucidate the extent to which products released by Kupffer cells and SECs could mediate hemodynamic changes in the liver of steatohepatitis, and the present model may be a suitable one in which to conduct such experiments. By 8 weeks, however, phagocytic activity was diminished throughout the lobule. This reduced activity is consistent with the report of diminished Kupffer cell phagocytic activity in the livers of obese Zucker rats.26 Whether this leads to spill-over of gut-derived endotoxin into the systemic circulation and stimulation of increased cytokine production in peripheral tissues, as suggested for the Zucker rat,26 is unknown.
Role of Stellate Cells and Relevance to NASH.
The etiopathogenic factors that perpetuate steatohepatitis in the MCD dietary model seem unlikely to involve insulin resistance because the animals are lean after several weeks of dietary feeding and have low rather than high blood insulin levels27; in this respect, the dietary model differs from NASH. Despite this, the liver pathology is remarkably similar to other “florid” forms of steatohepatitis, such as that found in alcoholic liver disease and in toxin- or drug-induced steatohepatitis or jejuno-ileal bypass steatohepatitis, as well as in cases of class 3 or 4 nonalcoholic fatty liver disease or NASH. It is therefore of interest that by week 8 of the MCD diet, stellate cells adjacent to activated Kupffer cells appeared activated, as indicated by loss of storage (vitamin A) lipid droplets. It was also clear from the associated collagen bundles that, by this time, stellate cells had elaborated and deposited abundant extracellular matrix as part of an ineffective wound healing response. This appeared to continue in the later phase of dietary feeding despite the diminished phagocytic activity of hepatic macrophages by 8 weeks.
Activation of stellate cells also may play a role in modifying hepatic microvascular dynamics because activation increases their contractility and these cells are thought to play a role in the regulation of sinusoidal caliber and blood flow.28–29 It is possible that impairment of blood flow, together with deposition of collagen and other extracellular matrix proteins in the space of Disse, contributes to the continuing liver injury that is a feature of all forms of steatohepatitis. In particular, impaired oxygenation of centrilobular hepatocytes or ROS generated during spontaneous fluctuations in blood flow that cause ischemia followed by reperfusion would explain, in part, the predominant distribution of liver injury and fibrosis in the centrilobular region. It also should be noted that TEM suggested defenestration of the sinusoidal endothelium; this would be consistent with “capillarization” of the sinusoidal wall that is reported to occur during fibrosis and cirrhosis. Confirmation of this loss of fenestrae awaits examination by scanning electron microscopy of another series of experimental animals currently being fed the MCD+ and MCD− diets.
In summary, taken together, liver injury during the development of fibrosing steatohepatitis appears to be associated with, and could result from, a combination of impaired sinusoidal blood flow due to lipid accumulation in parenchymal cells and collagen in the space of Disse narrowing sinusoid lumens. This is exacerbated by leukocytes either mechanically trapped in the narrowed sinusoids or adhering as a result of activation of an hepatic microvascular inflammatory response. The latter involves the activation of Kupffer cells and other hepatic macrophages with the release of proinflammatory cytokines and free radicals that contribute to the inflammatory response, stimulate stellate cells, and exacerbate the oxidative stress (TBARs) in the liver, which progressively increased during the 8-week period of study.
The authors thank Jayshree Seska and Ross Boadle for their technical assistance.