Activated Neutrophils Injure the Isolated, Radical ... - NCBI - NIH

American Journal of Pathology, Vol. 139, No. 5, November 1991 Copyright © American Association of Pathologist

Activated Neutrophils Injure the Isolated, Perfused Rat Liver by an Oxygen Radical-dependent Mechanism Lawrence J. Dahm,* A. Eric Schultze,#f and Robert A. Roth*t From the Departments of Pharmacology and Toxicology,* and Pathology,t and the Institute for Environmental Toxicology,* Michigan State University, East Lansing, Michigan Under certain circumstances, segmented neutrophils (PMNs) injure extrabepatic tissue by releasing toxic oxygen species and degradative enzymes. The authors used an isolated, perfused rat liver preparation to determine whetherPMNs might injure the liver. Livers from fasted rats were perfused with Krebs-Ringer bicarbonate buffer (pH 7.4) containing 3% bovine serum albumin (BSA) in a recirculating system. Rat peritoneal PMNs (4 X 108) or vehicle (Hank's balanced salt solution [HBSS], pH 7.35) were added, and liver injury was assessed 90 minutes later by release of alanine aminotransferase (ALT) into the perfusion medium and histopathologic analysis of liver sections. Perfusion of livers receiving only HBSS for 90 minutes resulted in a small increase in ALT activity in the perfusion medium but did not significantly alter histologic features of liver sections. Addition of unstimulatedPMNs did not increase further the ALT activity and, with the exception of vascular neutrophilia did not significantly change the histomorphology compared with controls. When PMNs activated with a combination ofphorbol myristate acetate (PMA, 31 ng/ml) and lithocholate (100 pLmol/l [micromolar]) were added to the perfusion system, however, livers released greater amounts ofALT than those perfused with PMA, lithocholate, and HBSS. Activated PMNs caused a transient reduction inflow of perfusion medium that lasted approximately 5 to 15 minute& Liver sections had multifocal to coalescing foci of moderate to severe, acute hepatocellular necrosis associated with the areas of intense sinusoidal neutrophilia In addition a second type of lesion was observed and was characterized by triangularfoci of necrosis located adjacent to periportal regions of sinusoids or portal veins containing neutrophilic thrombi These lesions were void of PMNs and were consistent with infarcts. A combination of superox-

ide dismutase and catalase added to the perfusion medium (500 U/ml each) prevented the elevation in ALT activity but not the transient reduction in flow. These results indicate that activated PMNs may cause liver injury by an oxygen radical-dependent mechanism. It is unclear whether PMN-derived oxygen radicals, hepatocellular-derived oxygen species resulting from reduced tissue perfusion and reperfusion, or both are involved in the pathogenesis. (Am JPathol 1991, 139:1009-1020)

Segmented neutrophils (PMNs) are phagocytic cells that play a major role in defense against microorganisms. When PMNs come in contact with bacteria, they undergo a respiratory burst and generate superoxide anion (02-) by a membrane-bound nicotinamide-adenine dinucleotide phosphate (NADPH) oxidase.' 02-7 other reactive oxygen species derived from it, and degradative enzymes located within cytosolic granules aid in destruction of bacteria. Although PMNs are undoubtedly beneficial in host defense, they may, under certain circumstances, cause injury to host tissue by the same mechanisms used to kill invading pathogens. For example, PMNs become activated after thermal injury to rat skin, and they cause lung damage by a mechanism dependent on toxic oxygen species.2 Several other models of PMN-dependent tissue injury include endotoxin-induced lung injury,3 ischemia/reperfusion injury in the heart,4 and immune complex-mediated kidney injury.5 The possibility that PMNs cause liver injury has not been generally explored as a mechanism of hepatotoxicity. We have reported recently that PMNs play a causal role in chemically induced liver injury caused by ot-naphThis work was supported by USPHS Grant ES04139. LJD was supported in part by a Procter and Gamble Fellowship administered by the Society of Toxicology. His present address is Department of Biochemistry, Emory University School of Medicine, Atlanta, GA 30322. AES was supported by NIH Training Grant ES07146. Accepted for publication June 20, 1991. Address reprint requests to Dr. Robert A. Roth, Department of Pharmacology and Toxicology, Michigan State University, East Lansing, Ml 48824.

1009

1010

Dahm, Schultze, and Roth

AJP November 1991, Vol. 139, No. 5

thylisothiocyanate (ANIT).6 In rats, ANIT causes cholestasis, injury to bile duct epithelial cells and periportal hepatocytes, and an intense infiltration of PMNs in liver.7 Administration of a polyclonal antibody to rat PMNs markedly reduced numbers of circulating PMNs, prevented hepatic PMN accumulation, and afforded protection against the liver injury associated with ANIT.6 Other investigators have used co-culture systems of PMNs and hepatocytes to assess PMN-mediated injury to liver. When co-cultured with hepatocytes, PMNs activated with phorbol myristate acetate (PMA) or other stimuli cause release of hepatocellular enzymes indicative of injury.810 With PMA and opsonized zymosan as stimuli, the mechanism of hepatocellular injury is dependent on proteinases but not on toxic oxygen species.9'10 In these studies, hepatocytes appeared to be insensitive to toxic oxygen species, presumably because of the presence of protective systems such as catalase and glutathione.9 Although co-culture systems provide valuable information regarding PM-hepatocyte interactions, it is difficult to extrapolate these data to conditions in vivo. One concern is that hepatocytes in culture become resistant to the toxic effects of oxygen radicals, presumably because of elevations in glutathione or vitamin E content.11 Second, co-cultures do not mimic normal liver architecture of intact animals, which eliminates the possibility that nonparenchymal cells or factors present in intact liver may be necessary for oxygen-radical-mediated liver injury to be manifested. To circumvent these problems, we have used an isolated, perfused rat liver preparation to determine whether activated PMNs cause liver injury. With this preparation, liver architecture is maintained, and the contents of the perfusion medium are easily manipulated.

Methods Materials Lithocholic acid, glycogen (type II from oyster), catalase, and kit 505 for alanine aminotransferase (ALT) activity were purchased from Sigma Chemical Co. (St. Louis, MO). Superoxide dismutase was obtained from Diagnostic Data, Inc. (Mountainview, CA). Phorbol myristate acetate was purchased from LC Services (Woburn, MA). Bovine serum albumin (BSA, fraction V) was purchased from ICN Immunobiologicals (Lisle, IL). All other reagents were of the highest grade commercially available. Polyethylene (PE) tubing was obtained from Clay Adams

(Parsippany, NJ).

Animals Male Sprague-Dawley rats (CF:CD[SD]BR) (Charles River, Portage, Ml), weighing 220 to 330 g, were housed

in plastic cages on aspen chip bedding under conditions of controlled temperature (1 80C to 21 OC) and humidity (55 ± 5%). A 12-hour light/i 2-hour dark cycle was maintained. Rats were allowed tap water and rat chow (Wayne Lab Blox, Allied Mills, Chicago, IL) ad libitum before experimentation.

PMN Isolation from Rat Peritoneum Male, retired breeder, Sprague-Dawley rats (Charles River) received 25 to 35 ml 1% glycogen solution in 0.9% sterile saline, intraperitoneally. Four hours later, rats were lightly anesthetized with diethyl ether, decapitated, and exsanguinated. Thirty milliliters heparin-treated (1 U/ml), phosphate-buffered saline (PBS, pH 7.4) were injected into the peritoneum. The abdominal wall was opened, and the contents were poured through layered gauze. The peritoneum was washed with another 30 ml heparintreated PBS, and the combined contents were spun in a centrifuge at 5OOg for 7 minutes. The supernatant was discarded, and the pellet was resuspended in 0.15 mol/A (molar) NH4CI with 0.01 mol/l NaHCO3 and 0.001 mol/l ethylenediaminetetraacetic acid (disodium salt). After 2 minutes in this red blood cell-lysing solution, PBS was added, and the contents were spun in a centrifuge at 320g for 7 minutes. The supernatant was discarded, and the cells were resuspended in Hank's balanced salt solution (HBSS, pH 7.35). Smears of the cell suspension were prepared and immersed in Wright-Giemsa stain. The percentage of PMNs in the preparation as well as cell viability were more than 95%.

Protocol for Perfusion of Isolated Rat Liver Rats were fasted for 24 to 30 hours and were anesthetized with sodium pentobarbital (50 mg/kg, intraperitoneally) in preparation for surgical removal of the liver. Livers were isolated and perfused essentially by the method of Miller.12 After a midline laparotomy was performed, the liver was exposed, and the common bile duct was cannulated with PE 10 tubing. Heparin (500 U) was injected into the inferior vena cava, and a blood sample was withdrawn from the same vessel for plasma determination of ALT activity as described below. The portal vein was cannulated with PE 240 tubing, and the liver was perfused with ice-cold, heparin-treated (1 U/ml), oxygenated saline. After ligation of the hepatic artery, the thoracic vena cava was cannulated with PE 260 tubing. The inferior vena cava was ligated anterior to the right renal vein. The liver was excised and transferred to a perfusion chamber maintained at 370C. It was perfused at an inflow pressure

Neutrophils and Liver Injury

1011

AJP Noeinber 1991, Vol. 139, No. 5

of 16 cm H20 with an oxygenated (95% 02/5% C02) Krebs-Ringer bicarbonate buffer (KRBB, pH 7.4) containing 1% BSA in a recirculating system. Medium was continuously oxygenated before entry into the liver. In the present studies, inflow oxygen content (PO2) averaged 429 + 5 mm Hg (n = 48). After a 30- to 50-minute preperfusion period (stabilization period), a sample of perfusion medium was removed from the reservoir and analyzed for ALT activity as described below. The medium was drained from the apparatus except for a residual amount (26 ml), and 85-92 ml KRBB containing 4% BSA was added. Approximately 22 cm upstream from the liver, 4 x 1 08 PMNs in a 7- to 15-ml volume were added to the perfusion system as a bolus. The final perfusion volume was 126 ml and contained 3% BSA. In one experiment, superoxide dismutase (SOD) and catalase were included in the perfusion medium (500 U/ml each), and activities during the perfusion were measured by the methods of Misra13 and Beers and Sizer,14 respectively. Livers were perfused for 90 minutes after addition of PMNs, and the pH of the perfusion medium was kept between 7.3 and 7.4 by adding NaHCO3. Control livers received an equivalent volume of HBSS vehicle in lieu of PMNs. In some experiments, PMNs were stimulated by treatment with PMA (31 ng/ml) for 5 minutes at 370C, followed by lithocholate (LC, 100 pumol/l [micromolar]) for an additional 5 minutes before addition to the perfusion system. Activated PMNs were added to the perfusion system when maximal release Of 02- from PMNs occurred.15 In these studies, PMA was added to the perfusion medium after addition of PMNs such that the final concentration was 31 ng/ml. Control livers received PMA, LC, and HBSS (in lieu of PMNs).

At 90 minutes, various indicators of liver viability were measured. A sample of perfusion medium was withdrawn from the system for determination of ALT activity. Aminotransferase activity (Sigma kit 505) was measured spectrophotometrically by the method of Reitman and Frankel,16 and it was used as a marker of liver injury because PMNs contain negligible ALT activity.9 Bile flow was measured and expressed per gram of tissue. At 80 to 90 minutes, oxygen consumption was calculated by multiplying the perfusate flow by the inflow-outflow difference in concentration and was expressed per gram of tissue. Inflow and outflow oxygen concentrations were calculated from P02 values measured with an Instrumentation Laboratory model 113-01 blood gas analyzer. At the end of the perfusion, the liver was removed from the apparatus, blotted with gauze, and weighed. Liver weight (wt) was expressed as a percentage of body weight. 02

Histopathologic Evaluations After the perfusions, liver samples were fixed in 10% buffered formalin and were embedded in paraffin. Sections were cut at 6 ,u and were stained with hematoxylin and eosin. Slides were randomized, coded, and evaluated with light microscopy. The pathologist was unaware of treatment groups. Parenchymal and stromal lesions were assigned a numerical score for severity and distribution. Severity of necrosis was quantified as follows: no lesion (0), mild (1), moderate (2), marked (3), severe (4). Distribution of lesions was assigned a score as follows: no lesion (0), focal (1), multifocal (2), coalescing (3), or diffuse (4). Frequency of thrombosis was recorded as fol-

Table 1. Effects of Unstimulated PMNs on Isolated, Perfused RatLivers Treatment* Liver wt/body wt x 100 Bile flow (,ul/hr/g) Perfusate flow (ml/min/g)

02 consumption (ml/hr/g)

ALT activity (SF units/ml) Donor rat plasma Preperfusion medium 90-min perfusion medium Histopathologic evaluationt Necrosis severity Lesion distribution Thrombosis frequency

HBSS ± 0.08

3.28 74 8.2 2.5

±4 ± 0.4 ± 0.2

PMN 3.20 66 8.9 2.0

± ± ± ±

0.16 3 0.8 0.2

18 ± 3 3± 1 10 ± 3t

17 ± 2 2± 1 11 ± 3t

0.4 ± 0.2 0.8 ± 0.5 0 0

0.3 ± 0.2 0.5 ± 0.4 0.4 ± 0.1

* Livers were isolated from rats fasted for 24 to 30 hr and were perfused with KRBB containing 1 % BSA for a 30 to 50 min preperfusion period. ALT activity was measured in medium at the end of the preperfusion period. The medium was changed to KRBB containing 4% BSA, and either PMNs (4 x 1 O8) or HBSS were added to the perfusion medium 22 cm upstream from the liver. Markers of liver viability were measured 90 min later. N = 4-5. t Significantly different from respective, preperfusion value. t Severity of necrosis, distribution of lesions and frequency of thrombosis were assigned scores from 0 to 4 as described in the Methods section. See text for histopathologic diagnoses and more detailed descriptions of the lesions.

1012


AJP Novemnber 1991, Vol. 139, No. 5

I Figure 1. Photomicrographs of a section of isolated rat liver after perfusion with HBSS (A) or unstimulated PMNs (B) (see methods for additional details regarding perfusions). A: Left, notice the multfocal areas of mild sinusoidal dilitation and slight atrophy of accompanying hepatic cords; right, perfusion did not significantly alter the morphology of hepatocytes. B: Left, with the exception of the minor vascular lesions noted in (A), the morphology is essentially unchanged with addition of unstimulated PMINs; right, PMNs within sinusoids caused no evidence of necrosis or cellular atypia. H&E: left, x64; right, x640.

lows: thrombi not seen (0), seen rarely (0.5), seen occasionally (1), few/several (2), multifocal (3), or many/diffuse (4). In addition, a detailed description of histologic lesions was provided for each slide.

Statistical Analysis Results are expressed as mean ± standard error (SE). Homogeneity of variance was tested using the F-max test. Log transformations were performed on nonhomogeneous data. If the variances were homogeneous, data were analyzed with a mixed-design analysis of variance (ANOVA), completely randomized factorial ANOVA, or Student's t-test, as appropriate. Individual comparisons between treatment means were made with Tukey's w test.17 Histologic data were analyzed by a one-tailed Wilcoxon's signed rank test.17 Correlation of histologic lesions with ALT release was performed using Spearman's rank correlation coefficient. The criterion for significance was P < 0.05 for all comparisons.

Results Isolated Rat Liver Perfusions and Effects of Unstimulated PMNs A 90-minute perfusion of livers receiving HBSS resulted in a small increase in ALT activity in the perfusion medium (Table 1). Liver sections were adequately preserved and, with the exception of few, minor, vascular lesions, they resembled closely normal rat liver (Figure 1 a). There were multifocal areas of mild sinusoidal dilatation. The dilated sinusoids radiated from central veins, and the accompanying hepatic cords were slightly atrophied. There were multifocal areas of mild periportal edema, and the endothelium lining the central veins, hepatic arteries, and portal veins had mild degenerate changes, including cytoplasmic vacuolation, change in nuclear polarity, and in some instances, sloughing of endothelium. The addition of unstimulated PMNs did not further increase ALT activity or change the liver weight/body


1013


Figure 1. (Continued).

weight ratio, bile flow, perfusate flow, or oxygen consumption (Table 1). Segmented neutrophils caused a transient (ie, 1- to 4-minute) decrease in perfusate flow when added to the system (data not shown) but did not

significantly alter the histomorphology of the sections from that of controls. With the exception of myriads of PMNs within portal veins and hepatic sinusoids, no other major lesions were identified (Figure 1 b).

Table 2. Effects of PMNs Activated with PMA and LC on Isolated, Perfused Rat Livers

Treatment*

HBSS/PMAILC Liver wt/body wt x 100 Bile flow (pil/hr/g) Perfusate flow (ml/min/g)


ALT activity (SF units/ml) Donor rat plasma Preperfusion medium 90 min perfusion medium Histopathologic evaluationil Necrosis severity Lesion distribution Thrombosis frequency

2.76 ± 58 ± 7.0 ± 2.1 ±

0.08 4 0.4 0.2

PMN/PMA/LC 3.08 ± 0.11t 50 ± 5 6.7 ± 0.9 2.1 ± 0.2

15 ± 1 2± 1 15 ± 2t

11 ± 1 1 ±0 56 ± 11tt

0.8 ± 0.3 1.3 ± 0.3 0±0

2.7 ± 0.4t 2.5 ± 0.2t 2.5 ± 0.3t

* Livers were isolated from rats fasted for 24 to 30 hr and were perfused with KRBB containing 1 % BSA for a 30 to 50 min preperfusion period. ALT activity was measured in medium at the end of the preperfusion period. The medium was changed to KRBB containing 4% BSA, and activated PMNs were added to the perfusion medium 22 cm upstream from the liver. PMNs were activated by treatment with PMA (31 ng/ml) for 5 min at 370C followed by LC (100 ,uM) for an additional 5 min. Control livers received PMA, LC, and HBSS instead of PMNs. PMA was added to the medium to keep the concentration at 31 ng/ml. Markers of liver viability were measured 90 min later. N = 5-7. t Significantly different from respective, preperfusion value. t Significantly different from HBSS/PMA/LC group. 11 Severity of necrosis, distribution of lesions and frequency of thrombosis were assigned scores from 0 to 4 as described in the Methods section. See text for histopathologic diagnoses and more detailed descriptions of the lesions.

1014



Figure 2. Photomicrographs of a section of isolated rat liver perfused with PMA and LC (A) or PMNs activated with PMA and LC (B) (see methodsfor additional details regarding perfusions). A: Left, histomorphology is consistent with that of control rat liverperfusions (see Figure 1A); right, hepatocellular cords remain intact and are free of necrosis. B: Left, notice the multifocal, irregularly shaped, coalescing areas of moderate to severe, acute hepatocellular necrosis; right, the foci of necrosis had swollen, pale, and extremely vacuolated hepatocytes which were associated with areas of sinusoidal neutrophilia. Some hepatocytes had enlarged, hyperchromatic or small, shrunken pyknotic nuclei. Normal architecture was lost in the most severely affected areas. Numerous neutrophils (arrows) filled the sinusoids. H&E, left, x 64; nght, x640.

Effects of Activated PMNs on Isolated, Perfused Rat Livers Neutrophils were activated in vitro by sequential treatment with PMA and LC, because this procedure has been shown to cause a more than eightfold release of 02- from PMNs compared with PMA treatment alone.15 The addition of activated PMNs to the perfusate reduced flow through livers almost completely for approximately 5 to 15 minutes (data not shown). Flow returned gradually, and the return appeared to be complete within 10 to 40 minutes after addition of PMNs. Activated PMNs caused injury to the liver, as indicated by an increase in ALT activity in the medium after 90 minutes of perfusion (Table 2). They also slightly increased the liver weight/body weight ratio but did not change bile flow or 02 consumption. Sections of livers that received PMA and LC but no

PMNs were essentially normal or had multifocal, thin rims of extremely mild centrilobular necrosis. No significant foci of parenchymal necrosis were observed (Figure 2a). Sections of perfused livers that received activated PMNs had multifocal to coalescing foci of moderate to severe, acute hepatocellular necrosis associated with areas of sinusoidal neutrophilia (Figure 2b). These lesions occurred in a panlobular distribution but were identified most often in periportal and midzonal regions. They were characterized by multifocal, irregularly shaped areas of swollen, pale, extremely vacuolated hepatocytes with either slightly enlarged, hyperchromatic nuclei or small pyknotic nuclei. An occasional hepatocyte was shrunken and anucleate. In some foci, necrosis was pronounced, hepatocytes were individualized, and there was a complete loss of normal hepatic architecture. In addition, a second, distinct type of lesion was apparent. These lesions were characterized by multifocal, roughly triangular areas of necrosis that frequently were located immedi-

-


Figure 3. Photomicrograph of a section of isolated rat liter perfused uith PlNs activated uith PMA and LC (see methodsfor additional details regarding perfusions). The infarct was characterized &' a triangularfocus of necrosis that was located immediately adjacent to a portal vein that contained a neutrophilic thrombu4s (arrou). H&E, x208.

1016



Figure 4. Photomicrographs of a section of isolated rat liver perfused with PMNs activated with PMA and LC (A) or activated PMNs plus SOD and catalase (B) (see methods for details regarding perfusions). A: Left, liver sections had multifocal, coalescing areas of marked to severe, acute hepatocellular necrosis associated with areas of sinusoidal neutrophilia; right, foci of necrosis with loss of normal hepatocellular architecture were numerous. Hepatocytes were swollen and had marked cytoplasmic vacuolation. Hepatocyte nuclei were pyknotic. B: Left, liver sections had multifocal, coalescing areas of moderate to marked hepatocellular necrosis associated with areas of sinusoidal neutrophilia; right, foci of necrosis had slightly less hepatocellular cytoplasmic vacuolation and had normal hepatocellular cording. H&E, left, x64;

right, x640.

ately adjacent to portal veins or periportal regions of hepatic sinusoids containing neutrophilic thrombi. They had necrotic centers with no or minimal neutrophilic infiltrates and were highly suggestive of infarcts (Figure 3). The correlation coefficient for severity of hepatocellular necrosis and ALT release from livers was statistically significant (r = 0.799, P < 0.001). When PMNs were stimulated with PMA alone (31 ng/ ml) and added to the perfusion system, they reduced perfusate flow to nearly the same extent as did PMNs stimulated with PMA and LC. They did not elevate ALT activity in the perfusion medium, however, or increase the liver weight/body weight ratio (data not shown). For example, perfusate ALT values from livers receiving PMAactivated PMNs were 18 + 6 SF units/ml, whereas ALT values from control livers receiving only PMA were 16 + 7 Sigma-Frankel units/ml. Livers that received only PMA were essentially normal in histologic appearance or had

multifocal, extremely mild, acute centrilobular necrosis similar to that described previously (data not shown). Sections from livers that received PMA-activated PMNs varied somewhat in histologic appearance. Approximately one half of the livers were essentially normal histologically, with the exception of vascular neutrophilia. The remaining livers had multifocal to coalescing, moderate to marked, acute hepatocellular necrosis (data not shown). In general, livers perfused with PMNs activated with PMA alone had similar but less severe lesions (necrosis severity 1.1 ± 0.5, lesion distribution 1.3 ± 0.5) than those perfused with PMNs activated with PMA and LC (necrosis severity, 2.7 ± 0.4; lesion distribution, 2.5 ± 0.2). Neutrophilic thrombi within sinusoids and portal veins were observed less frequently in livers given PMAactivated PMNs (thrombosis frequency, 1.1 ± 0.2) than those receiving PMNs activated with PMA and LC (thrombosis frequency, 2.5 ± 0.3).


1017



Effects of SOD and Catalase on PMN-mediated Liver Injury To determine whether toxic oxygen species released from activated PMNs mediate the liver injury, a combination of SOD and catalase was added to the perfusion medium before addition of PMNs. Superoxide dismutase and catalase activities were elevated in the perfusion medium during the 90-minute perfusion, and these agents attenuated the severity of necrosis and the elevation of perfusate ALT activity caused by activated PMNs (Figure 4 and Table 3). The correlation coefficient for ALT release and severity of hepatocellular necrosis was statistically significant (r = 0.688, P < 0.05). The combination of SOD and catalase did not affect the distribution of lesions, thrombosis frequency, or alterations in perfusate flow caused by activated PMNs (Table 3).

Discussion Toxic oxygen species released from phagocytic cells (ie, macrophages and Kupffer cells) have been implicated in

liver injury. In one model, intravenous treatment of rats with Corynebacterium parvum recruits macrophages into the liver, and subsequent administration of endotoxin 6 days later results in coagulative hepatic necrosis.18'19 02- appears to be involved in the pathogenesis, because administration of SOD before endotoxin reduces the hepatotoxicity.19 Others have provided preliminary evidence that macrophages and Kupffer cells may be involved in chemically induced liver injury by oxygen radical-dependent mechanisms.20-22 The role of PMNs as effector cells in liver injury has not been studied extensively, although we and others have evidence to support their involvement via oxygen radicalindependent mechanisms.6'9'10 In the present study, we have shown that activated PMNs cause modest injury to the isolated, perfused rat liver as indicated by leakage of ALT into the perfusion medium and histologic evidence of necrosis. The mechanism of injury may involve toxic oxygen species, because a combination of SOD and catalase added to the perfusion medium affords protection. It is unclear which toxic oxygen species are involved in the pathogenesis, because the combination of SOD

1018


AJP Novernber 1991, Vol. 139, No. 5

Table 3. Effects of SOD and Catalase on PMN-mediated Liver Injury

Addition* None

Liver wt/body wt x 100 Bile flow (,ul/hr/g) Perfusate flow (ml/min/g)


ALT activity (SF units/ml) Donor rat plasma Preperfusion medium 90 min perfusion medium Catalase activity (U/ml) 0 min 90 min SOD activity (U/ml) 0 min 90 min Histopathologic evaluationil Necrosis severity Lesion distribution Thrombosis frequency

3.17 51 7.0 2.1

± ± ± ±

0.09 5 0.7 0.2

15 ± 1 1 ±0 100 ± 19t

Catalase + SOD 2.93 51 7.8 2.1

± ± ± ±

0.13 3 0.8 0.2

19 ± 2 2± 1 35 ± 10tt

2± 1 6±2

455 ± 33t 160 ± 20t

0±0 4±4

603 ± 76t 492 ± 48t

2.8 ± 0.4 2.6 ± 0.2 3.2 ± 0.4

1.9 ± 0.4t 2.4 ± 0.2 2.7 ± 0.5

* Livers were isolated from rats fasted for 24 to 30 hr. They were perfused with PMNs, PMA and LC as described in the footnote to Table 2 except that SOD and catalase were included in the KRBB containing 4% BSA. SOD and catalase activities were measured in the perfusion medium immediately after addition of PMNs (0 min) and at the end of the perfusion (90 min). Markers of liver viability were measured after 90 min of perfusion. N = 4-5.

t Significantly different from respective, preperfusion value. t Significantly different from group not receiving SOD and catalase. 1 Severity of necrosis, distribution of lesions and frequency of thrombosis were assigned scores from 0 to 4 as described in the Methods section. See text for histopathologic diagnoses and more detailed descriptions of the lesions.

and catalase would be expected to degrade °2- and H202 as well as to prevent OH formation by the ironcatalyzed Haber-Weiss reaction. These results supporting a role of toxic oxygen species in hepatocellular injury are in contrast to those obtained in co-cultures of hepatocytes and PMNs, in which PMNs activated with PMA or opsonized zymosan caused hepatocellular injury by a proteinase-dependent mechanism.9'10 Differences may relate to the particular model employed. For example, hepatocytes in culture become resistant to the toxic effects of oxygen radicals, presumably because of elevations in glutathione or vitamin E content.1 1 Alternatively, factors or cell types found in the perfused liver preparation but not in co-cultures may be required for oxygen-radical-dependent injury to be manifested. Our results suggest a role for toxic oxygen species in the pathogenesis, although they do not rule out an additional proteinase-dependent component. The source of the toxic oxygen species is likely PMNs, because they were activated with PMA and the bile salt lithocholate (LC), a procedure that causes much greater release Of 2- from rat PMNs than other commonly used stimuli.1523 In addition, activated PMNs were added to the perfusion system when maximal 02- release from PMNs occurred.15 Because activated PMNs transiently reduced perfusate flow through liver, however, the possibility must be considered that toxic oxygen species resulted from hypoxia/reoxygenation. In other models employing perfused livers from fasted rats, a prolonged pe-

riod of low flow or hypoxia/reoxygenation causes liver injury by a mechanism that appears to be mediated by toxic oxygen species generated intracellularly in hepatocytes by xanthine oxidase.2626 Presumably the release of toxic oxygen species occurs as a result of conversion of xanthine dehydrogenase to xanthine oxidase, which occurs during hypoxia/ischemia.27 The protection afforded by SOD and catalase in our study is consistent with this alternative mechanism, because either enzyme prevents hypoxia/reoxygenation injury to the isolated, perfused rat liver.26 The mechanism of protection is not well understood, although it may reflect degradation of intracellular oxygen radicals, because SOD can enter hepatocytes by pinocytosis.28 Thus the data support a role for toxic oxygen species, although it is not known whether the protection afforded by SOD and catalase reflects degradation of toxic oxygen species generated intracellularly, extracellularly, or both. Alternatively the protective effect of SOD and catalase may be related to other unidentified effects of these agents, because the role of reactive oxygen species in hypoxia/reoxygenation and ischemia/reperfusion injury is controversial.`9 The cause for the reduction in perfusate flow by activated PMNs is unclear. It does not appear to be mediated by PMN-derived oxygen species, because the combination of SOD and catalase did not alter the reduction in flow. Activated PMNs release vasoconstrictor agents30 that might have caused the reduction in perfusate flow. For example, activated rat PMNs release thiol


1019


ether leukotrienes,31 which cause vasoconstriction in rat liver.32 Alternatively activated PMNs aggregate and adhere to endothelium, and it is possible that aggregates of PMNs occluded sinusoids and transiently blocked perfusate flow. Neutrophilic thrombi were observed in hepatic sinusoids and portal veins in livers receiving activated PMNs. Histopathologic analysis of rat livers receiving activated PMNs showed two distinct lesions that likely contributed to the increase in perfusate ALT activity. These lesions differed in morphology and lobular distribution, and they are consistent with the contention that PMNderived oxygen species as well as hepatocellularderived oxygen species are involved in the pathogenesis. The first type of lesion was identified in a panlobular distribution, although it occurred more often in periportal and midzonal regions. Lesions were characterized by multifocal, irregularly shaped, coalescing foci of acute hepatocellular necrosis associated with areas of sinusoidal neutrophilia. This lesion was consistent with the proposal that PMN-derived oxygen radicals caused hepatocellular necrosis and ALT leakage. Presumably oxygen radicals released in close proximity to hepatocytes caused injury by mechanisms discussed below. A second type of lesion was evident in sections of livers perfused with activated PMNs. These lesions consisted of roughly triangular foci of acute hepatocellular necrosis and were located adjacent to periportal regions of sinusoids or portal veins containing neutrophilic thrombi. These foci were essentially devoid of PMNs and appeared to be infarcts. This lesion was consistent with the proposal that toxic oxygen species generated by hepatocytes during localized hypoxia/reoxygenation caused hepatocellular necrosis. In vivo, the occurrence of true infarcts in the livers of animals is uncommon.' 34 The dual blood supply and lattice structure of hepatic sinusoids effectively decreases the probability of significant ischemic necrosis due to infarction. Reactive oxygen species are toxic to a wide variety of eukaryotic cells by mechanisms that may involve peroxidation of membrane lipids and damage to intracellular organelles.35 In hepatocytes, menadione and diquat undergo redox cycling and generate toxic oxygen species, which apparently lead to injury by causing protein thiol oxidation and lipid peroxidation, respectively.3638 Oxygen radicals generated intracellularly during hypoxia! reoxygenation might cause hepatocellular injury by similar mechanisms. Alternatively oxygen radicals released from PMNs in close proximity to hepatocytes might cause peroxidation of plasma membrane lipids or oxidize protein thiols and thereby alter the structural integrity of the plasma membrane. In addition, because certain toxic oxygen species cross membranes,35 PMN-derived oxygen species might injure intracellular organelles or oxidize

critical protein thiols. Segmented neutrophil-derived oxygen radicals also might cause hepatocellular injury indirectly by interacting with nonparenchymal cells or other factors in liver. In summary, we have shown that PMNs activated with PMA and LC cause injury to the isolated, perfused rat liver by a mechanism involving toxic oxygen species. It is not clear whether the injurious oxygen species are released from activated PMNs, from hepatocytes that were reoxygenated after an hypoxic episode, or both. Further work is required to define the relative contribution of these two sources of toxic oxygen species in PMN-mediated injury to the liver. The present study also demonstrates the potential advantages of the isolated liver preparation compared with PMN-hepatocyte co-cultures in studying PMN-mediated liver injury, because an intact liver architecture allows for better extrapolation to conditions in vivo.

Acknowledgments The authors thank Robert A. Burgess, Kathleen M. Vorick, Eric M. Shobe, and James G. Wagner for valuable technical assistance.

References 1. Babior BM: Oxygen-dependent microbial killing by phagocytes. N Engl J Med 1978, 298:659-668 2. Till GO, Beauchamp C, Menapace S, Tourtellote W, Kunkel R, Johnson KJ, Ward PA: Oxygen radical dependent lung damage following thermal injury of rat skin. J Trauma 1983, 23:269-277 3. Brigham KL, Meyrick B: Interactions of granulocytes with the lungs. Circ Res 1984, 54:623-635 4. Romson JL, Hook BG, Kunkel SL, Abrams GD, Schork MA, Lucchesi BR: Reduction of the extent of ischemic myocardial injury by neutrophil depletion in the dog. Circulation 1983, 67:1016-1023 5. Johnson, KJ, Ward PA: Biology of disease: Newer concepts in the pathogenesis of immune complex-induced tissue in-

6.

7.

8.

9.

jury. Lab Invest 1982, 47:218-226 Dahm W, Schultze AE, Roth RA: An antibody to neutrophils attenuates a-naphthylisothiocyanate-induced liver injury in the rat. J Pharmacol Exp Ther 1991, 256:412-420 Plaa GL, Priestly BG: Intrahepatic cholestasis induced by drugs and chemicals. Pharmacol Rev 1977, 28:207-273 Holman JM, Saba TM: Hepatocyte injury during postoperative sepsis: Activated neutrophils as potential mediators. J Leuk Biol 1988, 43:193-203 Guigui B, Rosenbaum J, Preaux A, Martin N, Zafrani ES, Dhumeaux D, Mavier P: Toxicity of phorbol myristate acetate-stimulated polymorphonuclear neutrophils against rat

1020



10.

11.

12. 13.

14.

15. 16. 17.

18. 19.

20.

21.

22.

23.

hepatocytes. Demonstration and mechanism. Lab Invest 1988, 59:831-837 Mavier P, Preaux AM, Guigui B, Lescs MC, Zafrani ES, Dhumeaux D: In vitro toxicity of polymorphonuclear neutrophils to rat hepatocytes: Evidence for a proteinasemediated mechanism. Hepatology 1988, 8:254-258 Ruch RJ, Crist KA, Klaunig JE: Effects of culture duration on hydrogen peroxide-induced hepatocyte toxicity. Toxicol Appi Pharmacol 1989,100:451-464 Miller LL: Technique of isolated rat liver perfusion, Isolated Liver Perfusion and its Applications. Edited by Bartosek, A Guaitani, LL Miller. New York, Raven Press, 1973, pp 1 1 -52 Misra HP: Quantitation of superoxide dismutase, Handbook of Methods for Oxygen Radical Research. Edited by RA Greenwald. Boca Raton, FL, CRC Press, 1985, pp 237-241 Beers RF, Sizer IW: A spectrophotometric method for measuring the breakdown of hydrogen peroxide by catalase. J Biol Chem 1952,195:133-140 Dahm W, Hewett JA, Roth RA: Bile and bile salts potentiate superoxide anion release from activated, rat peritoneal neutrophils. Toxicol Appl Pharmacol 1988, 95:82-92 Reitman S, Frankel S: A colorimetric method for the determination of serum glutamic oxaloacetic acid and glutamic pyruvic transaminases. Am J Clin Pathol 1957, 28:56-63 Steel RGD, Torrie JH: Principles and Procedures of Statistics. A Biometrical Approach. 2nd edition. New York, McGraw Hill, 1980, pp 173-175, 185-186, 195-238, 336-347, 390-393, 539-540 Ferluga J, Allison AC: Role of mononuclear infiltrating cells in pathogenesis of hepatitis. Lancet 1978, 2:610-611 Arthur MJP, Bentley IS, Tanner AR, Kolwoski-Saunders P, Millward-Sadler GH, Wright R: Oxygen-derived free radicals promote hepatic injury in the rat. Gastroenterology 1985, 89:1114-1122 Laskin DL, Pilaro AM: Potential role of activated macrophages in acetaminophen hepatotoxicity: I. Isolation and characterization of activated macrophages from rat liver. Toxicol Appl Pharmacol 1986, 86:204-215 Laskin DL, Pilaro AM, Ji S: Potential role of activated macrophages in acetaminophen hepatotoxicity. II. Mechanism of macrophage accumulation and activation. Toxicol Appl Pharmacol 1986, 86:216-226 ElSisi AE, Earnest DC, Sipes IG: Vitamin A potentiation of carbon tetrachloride hepatotoxicity. The Toxicologist 1987, 7:130 Dahm U, Roth RA: Differential effects of lithocholate on rat neutrophil activation. J Leuk Biol 1990, 47:551-560

24. Zhong Z, Lemasters JJ, Thurman, RG: Role of purines and xanthine oxidase in reperfusion injury in perfused rat liver. J Pharmacol Exp Ther 1989, 250:470-475 25. Marotto ME, Thurman RG, Lemasters JJ: Early midzonal cell death during low-flow hypoxia in the isolated, perfused rat liver: Protection by allopurinol. Hepatology 1988, 8:585-590 26. Younes M, Strubelt 0: The involvement of reactive oxygen species in hypoxic injury to rat liver. Res Commun Chem Pathol Pharmacol 1988, 59:369-381 27. Engerson TD, McKelvey TG, Rhyne DB, Boggio EB, Snyder SJ, Jones HP: Conversion of xanthine dehydrogenase to oxidase in ischemic rat tissues. J Clin Invest 1987, 79:15641570 28. Kyle ME, Nakae D, Sakaida I, Miccadei S, Farber JL: Endocytosis of superoxide dismutase is required in order for the enzyme to protect hepatocytes from the cytotoxicity of hydrogen peroxide. J Biol Chem 1988, 263:3784-3789 29. Jaeschke H, Smith CV, Mitchell JR: Reactive oxygen species during ischemia-reflow injury in isolated perfused rat liver. J Clin Invest 1988, 81:1240-1246 30. Bach MK: The Leukotrienes: Their Structure, Actions, and Role in Diseases. Kalamazoo, The Upjohn Company, 1983, pp 15-16 31. Orange RP, Valentine MD, Austen KF: Release of slow reacting substance of anaphylaxis in the rat: Polymorphonuclear leukocyte. Science 1967, 157:318-319 32. Krell H, Dietze E: Hemodynamic and metabolic responses to leukotriene C4 in isolated perfused rat liver. Hepatology 1989,10:300-305 33. Jubb KVF, Kennedy PC: Pathology of Domestic Animals. Vol 2. New York, Academic Press, 1970, pp 196-199 34. Rubin E, Farber JL: Pathology. Philadelphia, JB Lippincott, 1988, p 787 35. Fantone JC, Ward PA: Oxygen-Derived Radicals and Their Metabolites: Relationship to Tissue Injury. Kalamazoo, The Upjohn Company, 1985, pp 21-26 36. Di Monte D, Ross G, Bellomo G, Elkow L, Orrhenius S: Alterations in intracellular thiol homeostasis during the metabolism of menadione by isolated rat hepatocytes. Arch Biochem Biophys 1984, 235:334-342 37. Di Monte D, Bellomo G, Thor H, Nicotera P, Orrhenius S: Menadione-induced cytotoxicity is associated with protein thiol oxidation and alteration in intracellular Ca2 + homeostasis. Arch Biochem Biophys 1984, 235:343-350 38. Smith CV: Evidence for the participation of lipid peroxidation and iron in diquat-induced hepatic necrosis in vivo. Mol Pharmacol 1987, 32:417-422