INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

IN THE ABSENCE OF LIVER transplantation, fulminant hepatic failure (FHF) is associated with a 50-90% mortality rate and accounts for ~6% of all liver-related deaths in the United States (12, 16, 20). This poor prognosis is due, at least in part, to the absence of an appropriate animal model, which would allow for a greater understanding of the pathophysiology of this syndrome. Essential criteria for an animal model of FHF include reproducibility, death from liver failure, and a long therapeutic window (28). Because hepatic encephalopathy (HE) is invariably associated with FHF in humans (6, 24), altered mental status also should be considered an essential criterion for animal models of this syndrome. Yet no currently used animal model satisfies all these different criteria (Table 1). Equally important, not all toxins currently used in the study of FHF in animals have been reported to cause injury in humans.

A report from Guam in the early 1960s noted that cycad palm nuts induced a variety of cancers of the gastrointestinal tract (15). The active ingredient, azoxymethane (AOM), has since been used for the study of colon cancer in laboratory animals (37). Interestingly, this report described in anecdotal fashion that cycad palm nuts also caused liver injury in humans, rats, and livestock fed meal derived from this nut (15). We were therefore interested to know if, in addition to its established role in causing colon cancers, AOM also could be used as an FHF-inducing hepatotoxin.

In this study we investigated the effects of AOM as a hepatotoxin in C57BL/6J mice. We restricted this investigation to this species because whole animal genetic manipulations (i.e., transgenics, knockouts) can only be performed in mice, and this particular strain is the best characterized (27). We herein demonstrate that AOM induces a dose-dependent FHF in mice that is highly reproducible, causes death from liver failure in a dose-dependent manner due to hepatic mitochondrial injury, has a long therapeutic window, and generates an associated encephalopathy with evidence of cerebral edema in end-stage disease. AOM is thus the first toxin to satisfy all the criteria previously identified as essential for an animal model of FHF (28) that also is associated with the development of hepatic encephalopathy.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Mouse care and maintenance. Male C57BL/6J mice (25-30 g body wt) were obtained from Jackson Laboratory (Bar Harbor, ME) and used at 20 wk of age. Mice were fed ad libitum with Lab Diet Formulab 5008 rodent chow (PMI Feeds, St. Louis, MO) and were allowed free access to water. The mice were housed in microisolator cages with GreenTru laboratory bedding (Green Products, Conrad, IO) and exposed to a controlled light cycle of 10:14 h (light:dark) at 25°C. To accommodate video recording, 48 h before AOM (Sigma, St. Louis, MO) administration, mice were exposed to continual light. For most experiments, animals received 100 µg/g AOM dissolved in 100 µl sterile saline (Fujisawa USA, Deerfield, IL) administered by intraperitoneal injection. For all experiments, animals were matched with a saline-injected control group.

Clinical and biochemical evaluation. Mice were monitored continuously with an AG-180 video recording system (Panasonic, Secaucus, NJ) from 48 h before AOM administration until they were killed or died, allowing us to quantify individual eating and drinking events. Specifically, no more than three animals, each in separate cages, were subjected to continuous video monitoring after receiving saline or AOM. The number of times each animal drank or ate per hour was determined when a reviewer, blinded as to whether saline or AOM had been provided, reviewed the videotape. Animals also were assessed each hour by one of three investigators (K. A. Matkowskyj, J. A. Marrero, or R. E. Carroll) for activity level (asleep or moving about in the cage spontaneously) and reflexes (scattering, ataxia, righting, corneal) as previously described (3, 38). Mice were killed by CO₂ asphyxiation at the indicated time points. Freshly killed animals were immediately subjected to cardiac puncture to obtain serum for alanine aminotransferase (ALT), ammonia, glucose, bilirubin, alkaline phosphate, and creatinine. Automated chemistries were determined by the Clinical Pathology Laboratory of the University of Illinois at Chicago Medical Center, with the use of a CX7 Synchron automated analyzer (Beckman Instruments, Fullerton, CA). All data are reported as means ± SE for a minimum of three separate experiments.

Tissue collection. All organs, including the liver, brain, heart, lungs, pancreas, spleen, kidney, stomach, and intestine, were harvested and placed immediately in 10% Formalin. The tissues were processed and embedded in paraffin according to standard Armed Forces Institute of Pathology (AFIP) protocol (1). For each tissue, 5-µm sections were prepared with the use of a Spencer model 820 microtome (American Optical, Buffalo, NY), heat fixed at 70°C for 20 min, and then stained with hematoxylin and eosin according to standard AFIP protocol (1).

Immunohistochemistry. A three-stage indirect immunoperoxidase technique was used to label the primary antibody. Heat-fixed tissue sections were rehydrated in graded alcohols and then rinsed in a running water bath. The sections were incubated for 5 min at room temperature in 3% hydrogen peroxide in a light-impermeable chamber to quench endogenous activity. After the sections were rinsed with 1× PBS, slides were incubated for 20 min in blocking solution (90% water, 5% skim milk, and 5% H₂O₂). The excess solution was removed, and sections were incubated with 1:250 diluted glial fibrillary acidic protein (GFAP) primary antibody (Sigma) for 1 h in a humidity chamber. The sections were washed in PBS buffer and then incubated with biotinylated anti-rabbit IgG (DAKO, Carpinteria, CA) for 15 min. The slides were rinsed with PBS and incubated with streptavidin-conjugated horseradish peroxidase (DAKO) for 15 min. After the slides were rinsed with PBS, protein identification was performed by incubating slides with Liquid DAB substrate-chromogen system (DAKO) for 2 min. After a final wash in PBS followed by water, the slides were counterstained with Gills' hematoxylin for 1 min, dehydrated in graded alcohols, and mounted with a coverslip with the use of Permount.

Microscopy. Images were acquired with a MicroLumina digital scanning camera (Leaf, Westborough, MA) connected to a Nikon E600 microscope (Tokyo, Japan) with PlanApo objectives. Transmission electron microscopy was performed on freshly resected liver tissue immediately placed in 4% glutaraldehyde in 0.1 M cacodylate buffer (CB) overnight, then postfixed for 2 h in 2% aqueous osmium tetroxide in CB. The tissue was dehydrated in graded alcohols, and the samples were embedded in pure Epon resin (EMS, Fort Washington, PA) and were polymerized at 60°C for 12 h. Ultrathin sections were stained with uranyl acetate and Reynolds' lead citrate and were examined with a Phillips 410 transmission electron microscope.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Clinical findings. Animals treated with 100 µg/g AOM developed evidence of HE that was preceded by a novel prodromal phase. The prodromal phase was characterized by decreased spontaneous activity and decreased food and water intake (Fig. 1). This decrease in activity and feeding occurred before the loss of the animals' scatter reflex, which has previously been defined as characteristic of stage I HE (38). This prodromal phase of decreased activity reproducibly occurred 5-7 h after AOM administration and lasted for ~15 h, at which time the animals lost their scatter reflex. Thus lethargic animals in the prodromal period can be differentiated from those in stage I HE by virtue of the presence or absence of the scatter reflex.

View larger version (23K): [in this window] [in a new window]   Fig. 1.   Alterations in animal activity after intraperitoneal injection with saline () or 100 µg/g azoxymethane (AOM, ) until time of killing or death. Inactivity of the prodromal phase was differentiated from the lethargy of stage I hepatic encephalopathy (HE) during hourly examinations by observing animal response to manual stimulation. Data represent means. Error bars are not shown in order to retain graphical clarity.

Histological findings. AOM caused a progressive liver injury that preceded alteration in clinical behavior. The earliest histopathological alteration observed was the presence of microvesicular steatosis 2 h after AOM administration (Fig. 2B; normal mouse liver is shown in A). Four hours after AOM injection sinusoidal dilatation was apparent, predominantly in the area around the central vein (Fig. 2C). By the time stage I HE had developed ~20 h after AOM exposure, profound centrilobular necrosis was evident (Fig. 2D). In the preterminal stage IV HE animal, this necrosis was primarily hemorrhagic in nature (Fig. 2E). Given the early observation of microvesicular steatosis, we wondered if this was associated with mitochondrial injury. Consistent with AOM acting as a mitochondrial toxin, transmission electron microscopy revealed profound damage to the cristae (Fig. 2F).

View larger version (126K): [in this window] [in a new window]   Fig. 2.   Histopathological changes observed in liver of AOM-treated C57BL/6J mice. A: normal mouse liver. hematoxilyn and eosin (H&E) staining; magnification = ×400. B: liver section of a mouse exposed to AOM for 2 h showing presence of microvesicular steatosis, demonstrated by the translucent blue globules. Toluidine blue staining; magnification = ×1,000. C: liver section of a mouse 4 h after exposure to AOM showing centrilobular sinusoidal dilatation. H&E staining; magnification = ×400. D: section obtained 20 h after AOM administration showing centrilobular necrosis. Some hemorrhage can be observed at periphery in midzonal region. H&E staining; magnification = ×400. E: liver section from a preterminal stage IV encephalopathic mouse. Central veins have been obliterated 35 h after exposure to AOM by severe hemorrhage, with remaining viable hepatocytes present only in portal areas. H&E staining; magnification = ×100. F: transmission electron microscopic study of a liver section obtained 8 h after AOM administration. Yellow arrow identifies a mitochondrion. The double walls of mitochondria can be seen, but those of cristae contained within are not discernable. White arrows indicate fat. Osmium tetroxide, uranyl acetate, and Reynolds' lead citrate staining; magnification = ×18,000.

View larger version (55K): [in this window] [in a new window]   Fig. 3.   Histopathological changes observed in brain of AOM-treated C57BL/6J mice. A: brain section of a healthy control mouse. Arrows identify normal-appearing astrocytes. H&E staining; magnification = ×400. B: brain section of a mouse exposed to AOM for 35 h (stage IV HE) showing astrocyte ballooning and twinning (arrow), as well as chromatin displacement. H&E staining; magnification = ×400. C: immunohistochemistry for glial fibrillary acidic protein (GFAP) on a brain section from a mouse in stage IV HE. Note the absence of astrocyte staining for GFAP in the gray matter (arrow). Magnification = ×400. Inset: immunohistochemistry for GFAP remains positive when performed on a stage III mouse (arrow).

Biochemical alterations. We evaluated serum ALT, arterial ammonia, glucose, alkaline phosphate, bilirubin, and creatinine in saline-injected control animals and in animals at varying time points after AOM administration. In addition to each encephalopathic stage, we evaluated animals 4 and 8 h after drug delivery, because at the former time there is evidence of liver damage but no encephalopathy, whereas at the latter time subtle alterations in mental status can be detected.

Effect of AOM dose and lot. All results reported to this point were obtained with the use of 100 µg/g AOM injected intraperitoneally. To determine if AOM-induced liver injury was dose dependent, we also studied animals treated with 20 µg/g, 50 µg/g, and 200 µg/g ip AOM. At the lowest dose used, a dose commonly employed for the induction of colon cancer in mice (37), AOM was not hepatotoxic. In contrast, 50 µg/g and 200 µg/g caused identical changes in liver histology as reported for 100 µg/g but altered the rate at which FHF occurred and altered the time to death (Table 3).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

FHF is a life-threatening condition resulting in a 50-90% mortality rate in the absence of liver transplantation (5, 8, 12, 16, 20). In the past 10 years little progress has been made in improving patient outcome, in part because of the lack of a suitable animal model. Other commonly used drugs or toxins do not satisfy all the necessary criteria for an ideal model of FHF (28), in part because they are not reproducible, produce inconsistent toxicity, fail to generate a clinically significant lesion, or require supportive therapy (18, 38). The three most commonly studied toxins include carbon tetrachloride, acetaminophen, and galactosamine. Use of carbon tetrachloride as a model is problematic because of the inconsistent results between experiments and across species and the need for concomitant enzyme induction or partial hepatectomy depending on the species studied because this drug by itself does not induce deep coma (17, 32, 34). In contrast, acetaminophen and paracetamol produce inconsistent toxicity from animal to animal and between experiments (9-11, 14, 19). Similarly, galactosamine is associated with lot-to-lot variability and has no known human correlate as a hepatotoxin (3, 26, 33). Furthermore, whereas most toxins cause a centrilobular pattern of injury, galactosamine produces diffuse rather than zonal injury (3, 26, 33). Thus the three most commonly used animal models of FHF are far from ideal, with all models currently used possessing significant limitations (Table 1).

In this study we demonstrate that AOM is the first toxin to satisfy all the essential criteria for an animal model insofar as it is reproducible, causes death from liver failure, has a long therapeutic window, and poses minimal hazard to personnel when handled properly. In addition, AOM-induced liver injury generates all four previously characterized stages of HE (38), which is especially important because encephalopathy is invariably associated with FHF in humans (6, 12, 16, 20). One of the most important features of the murine AOM model of FHF is the identification of a prodromal phase separate and distinct from the first stage of HE (Fig. 1). The encephalopathic phase is ~19 h long and is detectable by the loss of scatter reflex when liver injury is already extensive (Fig. 2D). In contrast, the prodromal phase lasts a similar length of time (Fig. 1) but occurs when liver injury is relatively modest (Fig. 2, B and C). Awareness of this prodromal phase may be critical for future studies attempting to identify factors causing HE.

Although we did not specifically study the pathogenesis of HE caused by AOM-induced FHF, our findings do provide some insight into factors that cannot be implicated in altered mental status. Arterial ammonia has been proposed as a cause of HE (2, 6, 29, 32, 34), yet no increase in arterial ammonia was observed in the prodromal phase, and increases were not statistically significant until the development of stage III HE (Table 2). In contrast, significant decreases in serum glucose were detected as early as 4 h after AOM administration, with the decrease in serum glucose preceding the onset of even the prodromal phase. However, administration of exogenous glucose and maintenance of euglycemia did not improve the mental status of the AOM-treated mice; therefore, hypoglycemia per se does not appear to be responsible for the alterations in mental status. This is similar to what has been previously reported in other species, where hypoglycemia is commonly observed to complicate FHF in humans (5, 8, 23) and other animal models of FHF, including the dog, pig, and rabbit (3, 14, 17, 26). In no instance does correction of the hypoglycemia have any impact on improving the mental status of humans and other species suffering from FHF. Finally, edema is well known to complicate the progression of FHF (20), with up to 25% of humans reported to suffer this as a late complication (7). We specifically used astrocyte swelling as a measure of cerebral edema because the gravimetric method, widely used in rats, has not been validated in mice and because morphological changes consistent with brain edema are the most sensitive marker of this condition (22). Using this technique, we observed astrocyte swelling only in the brains of animals that had progressed to stage IV HE. The inability to observe these findings earlier, however, indicates that, similar to humans, CNS edema is a late finding and may not be responsible for the observed alterations in mental status.

Finally, a major advantage of AOM-induced FHF is its apparent ability to cause liver injury and failure in multiple species, including humans. AOM is the active metabolite of cycasin (36), found in cycad palm nuts only on the island of Guam. In a single anecdotal report from that island in the 1960s (15), pyknotic nuclei and focal necrosis in the centrilobular region were identified in humans, livestock, and rats as early as 48 h after ingesting unwashed cycad nuts. In retrospect, Guamanian FHF in humans, rodents, and livestock all appear to be due to AOM hepatotoxicity.

In conclusion, AOM causes a dose-dependent centrilobular necrosis of the liver, possibly by acting as a mitochondrial toxin, that progresses to liver failure and death. This agent causes both a neurological prodrome as well as frank HE. It is the first agent with a human correlate that may be used across species that is potentially reversible, is reproducible, and causes death from liver failure.