Susceptibility of Murine Periportal Hepatocytes toHypoxia-Reoxygenation: Role for NO and Kupffer

Cell–Derived OxidantsHisashi Taniai,1 Ian N. Hines,1 Sulaiman Bharwani,2 Ronald E. Maloney,1 Yuji Nimura,4 Bifeng Gao,3 Sonia C. Flores,3

Joe M. McCord,3 Matthew B. Grisham,1 and Tak Yee Aw1

Ischemia/reperfusion (I/R) is an important problem in liver resection and transplantationthat is associated with hepatocellular dysfunction and injury. This study was designed toinvestigate whether a difference in hepatocyte susceptibility occurs in the periportal (PP)and/or perivenous (PV) zones in response to hypoxia/reoxygenation (H/R), and to delineatethe mechanisms underlying this susceptibility. H/R was induced in an in situ perfusedmouse liver model with deoxygenated Krebs-Henseleit buffer followed by oxygenatedbuffer. Selective destruction of PP or PV sites was achieved by digitonin perfusion into theportal or inferior vena cava , and was confirmed by histological evaluations and zone-specificenzymes. Hepatocellular injury was assessed by alanine aminotransferase (ALT) release. Inwhole liver, H/R significantly increased perfusate ALT. H/R of PP-enriched zones causedALT release that was similar to that of whole liver (80 � 10 vs. 70 � 12 U/mg protein),consistent with significant PP hepatocyte injury. Minimal ALT release occurred in PV zones(10 � 5 U/mg protein). Administration of N-acetyl L-cysteine or a chimeric superoxidedismutase (SOD)—SOD2/3, a genetically engineered SOD—abrogated ALT release in H/R-perfused PP zones, implicating a role for superoxide (O2

�). This elevated ALT release was atten-uated by gadolinium chloride pretreatment, indicating that Kupffer cells are the O2

� source.Enzymatic inhibition of cellular nitric oxide synthase (NOS) or genetic depletion of endothelialnitric oxide synthase (eNOS) aggravated hypoxia injury while exogenous NO and induciblenitric oxide synthase (iNOS) deficiency abolished reoxygenation injury. In conclusion, PP hepa-tocytes are more vulnerable to H/R; this injury is mediated directly or indirectly by Kupffer cellderived O2

� and is limited by eNOS-derived NO. (HEPATOLOGY 2004;39:1544–1552.)

Previous studies in animal models show that isch-emia/reperfusion (I/R) injury to the liver occurs in2 distinct phases. The early reperfusion injury re-

sponse occurs between 1 and 6 hours, and reactive oxygenspecies (ROS), such as superoxide (O2

�), hydrogen perox-ide, and/or hydroxyl radical produced during reperfu-sion,1–5 have been implicated in this acute hepatic injuryprocess that is independent of leukocyte involvement. Be-cause hepatic proliferation, which is an important deter-minant of a patient’s survival after major hepaticresection, occurs from the periportal (PP) to perivenous(PV) zones,6 an understanding of the differential vulner-ability of hepatic zones to postischemic injury would pro-vide an important basis for preventing liver failure causedby alcoholic addiction or after major hepatectomy andliver transplantation. Given that metabolic heterogeneityof hepatic parenchymal cells occurs along the sinusoids inthe acinus wherein zonal differences have been describedfor gradients of oxygen, hormones, and xenobiotic detox-

Abbreviations: I/R, ischemia/reperfusion; ROS, reactive oxygen species; PP, peri-portal; PV, perivenous; KC, Kupffer cell; ALT, alanine aminotransferase; H/R,hypoxia/reoxygenation; NO, nitric oxide; eNOS, endothelial nitric oxide synthase;iNOS, inducible nitric oxide synthase; NAC, N-acetyl-L-cysteine; SOD, superoxidedismutase; L-NAME, N-nitro-L-arginine methyl ester; NOS, nitric oxide synthase;PK, pyruvate kinase; HE, hematoxylin-eosin.

From the 1Department of Molecular and Cellular Physiology and 2Department ofPediatrics, Louisiana State University Health Sciences Center, Shreveport, LA;3Webb-Waring Institute for Biomedical Research, University of Colorado HealthSciences Center, Denver, CO; and 4Division of Surgical Oncology, Department ofSurgery, Nagoya University Graduate School of Medicine, Nagoya, Japan.

Received August 27, 2003; accepted February 13, 2004.Dr. Hines is currently affiliated with Bowles Center for Alcohol Studies, Univer-

sity of North Carolina at Chapel Hill, Chapel Hill, NC.Supported by NIH grant DK 43785.Address reprint requests to: Tak Yee Aw, PhD, Department of Molecular and

Cellular Physiology, Louisiana State University Health Sciences Center, 1501Kings Highway, Shreveport, LA 71130-3932. E-mail: taw@lsuhsc.edu; fax:318-675-4217.

Copyright © 2004 by the American Association for the Study of Liver Diseases.Published online in Wiley InterScience (www.interscience.wiley.com).DOI 10.1002/hep.20217

1544

ification enzymes,7–9 the PP and PV regions of the liverare likely to exhibit differential vulnerability to I/R. Pre-vious studies in rat liver showed that low-flow I/R in-duced oxidative stress and impaired midzonalregions.10,11 In these studies, I/R were induced by chang-ing flow rate that altered both perfusion rate and oxygenconcentration. In other studies, Kato et al. found thathepatocytes from PP zones but nonparenchymal cellsfrom PV zones were more susceptible to hepatic I/R in therat.12 However, the mechanisms underlying differences inzonal susceptibility and tolerance to hepatic I/R remainpoorly understood.

Kupffer cells (KCs) are favored as the critical cellularcomponent in hepatic I/R injury because of their ability torelease ROS, cytokines, and chemokines.13–15 Evidencethat KC-derived ROS contribute to hepatic I/R injurycomes from studies showing that I/R enhances KC phago-cytic activity16 and that inactivation of KCs with gadolin-ium chloride (GdCl3) attenuated liver oxidative stress andplasma alanine aminotransferase (ALT) levels.13 Of noteis a preferential localization of KCs to hepatic periportalzones, as well as a preferential activation of PP KCs duringoxidative challenges like I/R17 and endotoxemia.18 Giventhis distinct and predominant PP distribution and activa-tion of KCs along the sinusoid, a zonal sensitivity to I/Rcan be expected. The current study was designed to inves-tigate whether a difference in susceptibility occurs in thePP and PV zones in response to hepatic hypoxia/reoxy-genation (H/R) and to delineate the mechanisms under-lying this susceptibility. To address these questions, wehave adapted a previously established rat liver model ofdigitonin perfusion7 to develop a novel in situ mousemodel of H/R to assess the sensitivity of PP and PV zones.This model will also enable us to determine the contribu-tion of KC-derived ROS and nitric oxide (NO) to theinjury response. The results indicate that H/R injury tohepatocytes primarily occurs in the PP zones and that thishepatocellular injury response is mediated by O2

� derivedfrom KCs. Our data further support a role for endothelialnitric oxide synthase (eNOS)-derived endogenous NOand for exogenous NO supplementation in protectionagainst this H/R-induced PP injury.

Materials and Methods

Animal Preparation. Age-matched male wild-type(�/�) C57BL/6 mice and mice deficient in eNOS (-/-)and inducible nitric oxide synthase (iNOS [-/-]), 8 to 10weeks old, were obtained from Jackson Laboratories (BarHarbor, ME) and allowed free access to laboratory chowand tap water. The animals were anesthetized with anintraperitoneal injection of pentobarbital sodium (50 mg/

kg), and the livers were perfused with hemoglobin- andalbumin-free Krebs-Henseleit solution (pH 7.4, 37°C)and gassed with carbogen (95% O2, 5% CO2) via 22-gauge intravenous catheters cannulated into the portalvein. Venous perfusates were collected from the inferiorvena cava cannulated in the same manner.19,20 Normalgrade perfusion was conducted using a peristaltic pump ata constant rate of 3.0 mL/min/g liver in a single-passmode. All animal protocols have been reviewed and ap-proved by the institutional animal review committee inaccordance with the animal use guidelines of the NationalInstitutes of Health.

Digitonin Perfusion. To obtain PP or PV zone-spe-cific livers, a 3 mg/mL digitonin-Krebs Henseleit bufferwas introduced into the inferior vena cava or portal vein ata flow rate of 0.3 mL/min/g liver. Selective zonal celldestruction was achieved within 30 seconds of digitoninperfusion. Washout of digitonin was achieved by perfu-sion with digitonin-free buffer into the opposite end tothe initial digitonin perfusion. We have verified that thisprocedure provided efficient digitonin clearance andwashout. After a –20-minute stabilization time after dig-itonin washout, the PP- or PV-enriched zones were sub-jected to H/R as described in Hypoxia/ReoxygenationProtocol.

Hypoxia/Reoxygenation Protocol. For all experi-ments, the following standard H/R protocol was used:liver hypoxia was induced by perfusion with buffer equil-ibrated with 95% N2/5% CO2 mixture via the portal veinfor 40 minutes, followed by perfusion with oxygenatedbuffer for an additional 40 minutes for the reoxygenationperiod.19,21 Perfusates were sampled at times before hyp-oxia initiation, at the end of the hypoxic period, and atevery 10 minutes during reoxygenation for determinationof hepatic release of ALT. Animals were divided into con-trols and 10 experimental groups. In the control group,livers were from wild-type mice subjected to H/R to es-tablish the baseline hepatocellular injury response to H/Rwithout treatment with digitonin or other agents. In theexperimental groups, livers from mice in groups 1 and 2were perfused with digitonin to obtain enrichment of PPand PV zones and thereafter subjected to H/R to deter-mine zonal injury responses. In the remaining experimen-tal groups (3 to 10), H/R studies were performed in PP-enriched livers, and the roles of ROS and NO wereexamined. To test the role of ROS, livers from mice ingroups 3 and 4 were treated with 5 mmol/L of N-acetyl-L-cysteine (NAC) or 1 unit/g body weight of a geneticallyengineered chimeric superoxide dismutase (SOD2/3).NAC was present throughout the H/R periods, andSOD2/3 was administered at the time of reoxygenation.To determine whether KCs were the ROS source, mice

HEPATOLOGY, Vol. 39, No. 6, 2004 TANIAI ET AL. 1545

from group 5 were injected intraperitoneally with 10mg/kg body weight of GdCl3 for 24 hours before prep-ping the livers for PP enrichment and H/R. Mice fromgroups 6 to 10 were used to test the role of NO. In group6, livers were treated with 1 mmol/L N-nitro-L-argininemethyl ester (L-NAME), an inhibitor of cellular NOSactivity. To specifically inhibit iNOS, group 7 livers weretreated with 5 mmol/L aminoguanidine. These inhibitorswere administered at the time of hypoxia and were presentthroughout reoxygenation. To further explore the roles ofeNOS and iNOS, livers from eNOS-/- or iNOS-/- micewere prepped for digitonin perfusion to acquire PP-en-riched zones and then subjected to the standard H/R pro-tocol in groups 8 and 9. Livers from group 10 mice were

subjected to the standard H/R protocol in the presence ofspermine NO (10 �mol/L), an NO donor. In some ex-periments, the role of NO was also evaluated in liversenriched with PV hepatocytes.

Measurements of Zone-Specific Enzymes in VenousPerfusates. In each experiment, venous perfusate sam-ples were collected before hypoxia, after 40 minutes ofhypoxia, and at every 10 minutes during reoxygenation todetermine the activities of ALT as an index of overallhepatocellular injury. ALT was quantified using a com-mercially available assay kit (Sigma, St. Louis, MO).7,8 Atthe end of the H/R perfusion, livers were excised, andtissue ALT activities were determined. Pyruvate kinase(PK) activities were determined as previously described invenous perfusates and tissues as a PV zone-specific en-zyme.7,8,22

Histology. In some experiments, liver histology wasperformed to determine the hepatic microarchitecture inPP- and PV-enriched livers after digitonin perfusion. Tis-sues were fixed in formalin, embedded in JB-4 plastic

Fig. 1. Enrichment of periportal (PP) or perivenous (PV) hepatic zonesby digitonin perfusion. Livers were first perfused in situ with hemoglobin-and albumin-free Krebs-Henseleit buffer via the portal vein as describedin Materials and Methods. Uniform blood clearance was achieved withnormal grade perfusion at a constant flow rate of 3.0 mL/min/g liver ina single-pass mode: (A) no digitonin treatment. (B) PV or (C) PPzone-specific livers were obtained with perfusion of a 3 mg/mL digitonin-Krebs-Henseleit buffer into the portal vein or inferior vena cava until liversurfaces exhibited the unique reticular patterns characteristic of PV or PPcell destruction (� 30 seconds). Washout of digitonin was achieved byperfusion with digitonin-free buffer into the opposite end to the initialdigitonin perfusion.

Fig. 2. Histology of untreated livers and livers perfused with digitoninvia the portal vein or inferior vena cava. Formalin-fixed tissues wereembedded in JB-4 plastic, sectioned, and stained with hematocylin-eosin(HE). (A-C) In untreated livers, hepatic architecture is intact, and uniformHE-stained PP and PV hepatocytes are evident. (A) Magnification, �10.Scale bar, 150 �m. (B and C) Magnification of PP and PV zones, �40.Scale bar, 100 �m. (D-F) In portal vein perfused livers, hepatic archi-tecture remains well preserved, with demarcated regions of PP and PVhepatocytes and little evidence of necrosis. PP cells exhibited lighterstaining with a halo effect and blanching, and PV cells retain normal HEstain. (D) Magnification, �10. (E and F) Magnification of PP and PVzones, �40. (E) The PP area exhibited some sinusoidal dilatation. (G-I)Inferior vena cava-perfused livers exhibited well-preserved hepatic archi-tecture, demarcated PP and PV zones, and little necrosis. PV hepatocyteswere lightly stained and blanched, and PP cells retained normal HE stain.(G) Magnification, �10. (H and I) Magnification of PP and PV zones,�40. (I) The PV area exhibited some sinusoidal dilatation. Data shownare 1 set representative of 5 separate preparations for all groups. pv,portal vein; hv, hepatic vein.

1546 TANIAI ET AL. HEPATOLOGY, June 2004

(Polysciences, Warrington, PA), and 5 �m sections werecut and stained with hematoxylin-eosin (HE).

Statistical Analysis. All data were expressed as themean � SEM. Differences between the control and ex-perimental groups were determined using one-wayANOVA with Fisher multiple-comparison test. P lessthan .05 was considered significant.

Results

Establishment of PP- and PV-Enriched Zones byDigitonin Perfusion and Their Responses to H/R. Asshown in Fig. 1, complete and uniform clearance of bloodfrom all the hepatic lobes was achieved with normal gradeperfusion of the liver via the portal vein at a flow rate of 3mL/min/g (Fig. 1A). Subsequent perfusion with digito-nin via the portal vein (Fig. 1B) or the inferior vena cava(Fig. 1C) resulted in selective destruction of parenchymalcells uniformly throughout the PP and PV regions of theliver. Notably, the surfaces of the livers exhibited charac-teristic reticular patterns: destruction of PV zones appearas pale center cores of damaged PV cells surrounded bydark brown rings of undamaged PP cells; destruction ofPP zones appear as pale rings of damaged PP cells sur-rounding dark brown center cores of undamaged PVcells.7 To validate that digitonin treatment did not com-promise the hepatic microarchitecture, liver histology wasperformed. Control (nondigitonin-perfused) livers exhib-ited normal hepatic architecture with equally well-stainedPP and PV areas (Figs. 2A–C). In digitonin-perfused liv-ers, the hepatic architecture is well preserved, with demar-cated regions of PP and PV hepatocytes and minimalhepatocellular necrosis and alterations in hepatic sinu-soids (Figs. 2D–I). PP zones perfused by digitonin wereevidenced by blanching and some sinusoidal dilatation;intact PV zones retained normal HE stain and architec-ture (Figs. 2E and F). Retention of normal stain in PPhepatocytes and blanched PV hepatocytes and some sinu-soidal dilatation were evident in livers perfused via theinferior vena cava (Figs. 2H and I). Liver wet-to-dryweight ratios were determined to assess the effect of digi-tonin and H/R on tissue swelling. Livers perfused withdigitonin (PP or PV zones) without H/R exhibited a small(�10%) but significant increase in wet-to-dry weight ra-tios as compared to nondigitonin-treated (whole) livers(Fig. 3A), indicating that digitonin induced some fluidretention. However, the wet-to-dry weight ratios of sinu-soidal endothelial cells, KCs, and hepatocytes preparedfrom elutriation of PP-enriched livers were not differentfrom those isolated from untreated livers (Fig. 3B), show-ing a lack of cell swelling; thus, the digitonin-inducedincrease in hepatic fluid accumulation was largely extra-

cellular. H/R exposure caused a significantly greater in-crease in wet-to-dry weight ratio in whole and PP-enriched livers as compared to the corresponding liversperfused with normoxic buffer (Fig. 3C), consistent withH/R-induced edema. Taken together, these results dem-onstrate that the digitonin method does provide a reason-able experimental approach for PP and PV enrichment toinvestigate zonal tolerance to hepatic I/R.

To verify that the targeted digitonin perfusion did, infact, cause enrichment of the desired specific zones withinthe liver, we monitored the time course of release of PK, (aPV-zone–specific enzyme) with respect to ALT, an en-zyme that is present in both the PP and PV zones but thatis generally higher in PP regions.7,8,22 Digitonin adminis-tration into the inferior vena cava resulted in significantrelease of PK and ALT (Fig. 4A), consistent with selectivedestruction of PV cells. In contrast, digitonin perfusioninto the portal vein caused release of ALT, but not PK(Figure 4A), consistent with selective destruction of PPcells. To validate that the digitonin perfusion into theportal vein or inferior vena cava yielded PV- or PP-en-riched hepatocytes, we measured total hepatocellular ac-tivities of PK and ALT post–digitonin perfusion andwashout. The results (Fig. 4B) show that PP-enrichedhepatocytes exhibited lower PK and higher ALT enzymeactivities, and higher PK and lower ALT activities werefound in PV-enriched hepatocytes. Accordingly, the ratioof PK –to ALT—an index of zonal enrichment—revealeda significantly greater PK/ALT in PV-enriched zonescompared to PP-enriched zones. Collectively, these re-sults demonstrate that the procedure of using targeteddigitonin perfusion via the portal vein or inferior vena

Fig. 3. Wet-to-dry weight ratios of untreated and digitonin-perfusedlivers before and after H/R exposure. (A) Whole (untreated) livers andlivers perfused with digitonin (PP or PV zones) without H/R. *P � .05compared to whole liver. (B) Hepatocytes (Hep), Kupffer cells (KC), andsinusoidal endothelial cells (SEC) isolated by elutriation from untreatedor digitonin-perfused PP enriched livers without H/R. (C) Whole (un-treated) and digitonin-perfused PP-enriched livers perfused with oxygen-ated buffer (normoxia) or with 40 minutes of deoxygenated buffer and 40minutes of oxygenated buffer (H/R). *P � .05 compared to correspond-ing normoxic livers. Data represent mean � SEM: n � 5 mice each forwhole and PP- and PV-enriched livers; n � 3 separate elutriationpreparations for Hep, KC, and SEC without or with digitonin treatment.

HEPATOLOGY, Vol. 39, No. 6, 2004 TANIAI ET AL. 1547

cava provided satisfactory establishment of select hepaticzones enriched in PV or PP hepatocytes. In subsequentexperiments, we have utilized this digitonin-perfusion ap-proach to generate zone-specific hepatocytes to investi-gate the responses of PP or PV zones to H/R and to definethe determinants of the injury response, namely, ROSand NO.

Figure 5A illustrates the kinetics of ALT release fromuntreated whole livers after a 40-minute perfusion withdeoxygenated buffer followed by a 40-minute perfusionwith oxygenated buffer. In some experiments, we deter-mined the release of lactate dehydrogenase as an index ofcellular damage, and the results were similar to those ofALT release from both PP- and PV-enriched livers. Be-cause ALT is a more specific PP enzyme, we subsequentlyassayed perfusate ALT as the index of PP hepatocyte in-jury. There was no detectable increase in perfusate ALTactivity during the hypoxia period. However, reoxygen-ation induced a time dependent increase in ALT thatapproached approximately 20% of total hepatic ALT ac-tivity over the 40-minute reoxygenation period. The ef-fect of H/R on hepatocyte injury in PV- and PP-enriched

livers is shown in Fig. 5B. Exposure of PP livers to H/Rresulted in a time-dependent increase in ALT release thatreached a value 6- to 8-fold greater at 40 minutes than at0, indicating an enhanced vulnerability of PP hepatocytesto H/R. In comparison, H/R caused a small but insignif-icant increase in ALT in PV livers at 40 minutes of reoxy-genation. There was no detectable release of PK in PVlivers under these conditions (data not shown), indicatingthat the PV regions of the liver are relatively tolerant of theH/R insult. Notably, the magnitude of H/R-inducedALT release in PP livers was quantitatively similar to thatof whole liver (Fig. 5A), suggesting that the H/R-inducedhepatocellular injury in whole liver was predominantlyattributed to hepatocyte damage in the PP zones.

Antioxidants and Depletion of KCs Abolish H/R-Induced PP Injury. To determine a role for ROS inmediating PP injury, two strategies were used. First, ROSproduction was measured in perfusates using oxidation ofdihydrorhodamine. However, while there was a trend forincreased dihydrorhodamine oxidation with H/R (datanot shown), these quantitative measurements in the per-fusates (at 3ml/min/g single-pass mode) were at the limitof detection. Second, H/R studies were performed in thepresence of antioxidants, namely, NAC, a thiol antioxi-dant, or SOD2/3, a genetically engineered chimeric SODthat combines the mature human manganese SOD2 withthe heparin-binding C-terminal region of the human ex-tracellular SOD3 that has an affinity for endothelial cellsurfaces. Figure 6A shows that NAC treatment com-pletely prevented the time-dependent increase in ALTinduced by H/R in PP livers. Similarly, the presence of

Fig. 5. Effect of H/R on ALT release from whole (untreated) liver andlivers enriched in PP or PV zones. The H/R protocol and digitoninperfusion for PP- or PV-zone enrichment were as described in Materialsand Methods. Perfusates were collected before hypoxic perfusion, at 40minutes after hypoxia and at every 10 minutes during reoxygenation forALT determination. The shaded areas indicate 40-minute periods ofhypoxic perfusion (0-40 minutes); unshaded areas indicate 40-minuteperiods of reoxygenation (40–80 minutes). (A) ALT release from wholeliver. Data represent the mean � SEM: n � 10 mice. (B) ALT releasefrom PP-enriched livers (open circle) or PV-enriched livers (opensquare). Data represent the mean � SEM: n � 7 mice for each group.*P � .05 compared to the value at 0 and 40 minutes. #P � .05compared to the PP group at the corresponding time.

Fig. 4. Zone-specific PK and ALT in venous perfusates and liver tissue.(A) time course of release of PK (a PV-zone-specific enzyme) and ALT (anenzyme present in both the PP and PV zones) at times after digitoninperfusion (3 mg/mL) and washout. Open circle, PV zone (n � 6); opensquare, PP zone (n � 6). Data represent mean � SEM. (B) Tissueactivities of PK and ALT and the ratio of PK to ALT in whole and PP- orPV-enriched liver. PP- or PV-enriched livers were achieved by digitoninperfusion into the inferior vena cava or portal vein. Following digitoninwashout and stabilization of perfusate levels of PK and ALT (at 10minutes), the livers were harvested and processed for determination oftissue activity of PK and ALT and the PK/ALT ratio. Data represent mean� SEM: n � 6 mice for each group. *P � .05 compared to PP-enrichedzone.

1548 TANIAI ET AL. HEPATOLOGY, June 2004

SOD2/3 was equally effective in preventing reoxygen-ation-induced ALT release (Fig. 6B). To determinewhether KCs were the source of ROS, we pretreated micewith GdCl3 (10 mg/kg body weight) to inactivate KCs.Figure 7 shows that GdCl3 pretreatment completely ab-rogated the elevation in ALT caused by H/R. Taken to-gether, these results implicate a role for ROS in mediating

H/R-induced PP injury and are consistent with KCs be-ing the source of the ROS.

eNOSDerived NO and Exogenous NO AttenuateH/R-Induced PP Injury. Since NO is known to rapidlyinteract with and decompose O2

� to nitrate and is anendogenous anti-inflammatory mediator,23 we deter-mined whether H/R-induced PP injury is modulated byNO. Administration of L-NAME (1 mmol/L) to wild-type mice elicited marked PP hepatocellular injury duringthe hypoxia phase as evidenced by a significant increase inALT release after 40-minute perfusion with deoxygenatedbuffer prior to reoxygenation (Fig. 8A). This enhancedALT activity was maintained at an elevated level withoutfurther increases during the 40-minute reoxygenation pe-riod (open square) compared to that of wild-type mice(open circle). A role for eNOS was evaluated using PP-enriched livers from eNOS-deficient mice (eNOS-/-). ThePP injury response during hypoxia in eNOS-deficient an-imals was similar to those after L-NAME treatment; thiselevated ALT activity was likewise maintained through-out the 40-minute reoxygenation without further in-creases (Fig. 8B), indicating that a deficiency in the eNOSgene per se elicited marked PP injury during the hypoxiaphase. In comparison, eNOS deficiency was without in-jurious effects on PV-enriched livers during hypoxia orreoxygenation (Fig. 8B, open triangle). That inhibition ofcellular NOS or loss of eNOS is deleterious is consistentwith a protective effect of endogenously generated NO inmaintaining PP integrity during the hypoxia phase. Be-

Fig. 6. Effect of NAC or SOD2/3 administration on H/R-induced PPinjury. Livers were perfused with digitonin for PP-zone enrichment asdescribed in Materials and Methods. NAC (5 mmol/L) was administratedat 0 minutes at the time of hypoxia initiation and was present throughoutH/R. SOD2/3 (1 unit/g body weight) was administered as a bolus dosefor 1 minute at the time of reoxygenation. PP hepatocellular injury wasassessed by ALT release into perfusates collected during the hypoxicphase (shaded areas, 0-40 minutes) and at every 10 minutes during thereoxygenation phase (unshaded areas, 40–80 minutes). (A) ALT releasefrom PP-enriched livers without (open circle) or with (open square) NACpretreatment. Data represent the mean � SEM: n � 5 mice for eachgroup. (B) ALT release from PP-enriched livers without (open circle) orwith (open square) SOD2/3 treatment. Data represent the mean �SEM: n � 5 mice for untreated control; n � 4 mice for SOD2/3-treatedgroup. *P � .05 compared to the corresponding value without NAC orSOD2/3 treatment.

Fig. 7. Effect of KC depletion on H/R-induced PP injury. Mice werepretreated with GdCl3 (10 mg/kg body weight) for 24 hours to inhibit KCactivation before livers were perfused with digitonin for PP-zone enrich-ment and then subjected to H/R. PP hepatocellular injury was assessedby ALT release into perfusates collected during the hypoxic phase(shaded area, 0-40 minutes) and at every 10 minutes during thereoxygenation phase (unshaded area, 40–80 minutes). Data representmean � SEM: n � 5 mice for control (open circle) and GdCl3-treatedgroup (open square). *P � .05 compared to the untreated control.

Fig. 8. Effect of NO and eNOS-derived NO on H/R-induced PP injury.To investigate the role of NO on H/R-induced PP injury, wild-type micewere treated with L-NAME. PP-zone-enriched livers were prepared fromeNOS-deficient mice to determine a role for eNOS. Some experimentswere performed in PV-enriched zones from eNOS-deficient mice. PPhepatocellular injury was assessed by ALT release into perfusates col-lected during the hypoxic phase (shaded areas, 0-40 minutes) and atevery 10 minutes during the reoxygenation phase (unshaded areas,40–80 minutes). (A) Data represent mean � SEM: n � 5 for untreatedwild-type mice (open circle) and L-NAME-treated mice (open square)prepped for PP zone enrichment. *P � .05 compared to non–L-NAME-treated controls. (B) Data represent mean � SEM: n � 5 for wild-type(open circle) and eNOS-/--deficient (open square) mice prepped forPP-zone enrichment; n � 4 for eNOS-/- mice (open triangle) prepped forPV-zone enrichment. *P � .05 compared to wild-type mice. **P � .05compared to PP-enriched zone.

HEPATOLOGY, Vol. 39, No. 6, 2004 TANIAI ET AL. 1549

cause a deleterious role of iNOS has been implicated ininflammatory processes, we evaluated its role in H/R-induced PP injury using iNOS-deficient mice (iNOS-/-).Figure 9A shows that iNOS deficiency afforded completeprotection against H/R-induced ALT release, suggestingthat iNOS upregulation during I/R promotes tissue dam-age. Similarly, treatment with aminoguanidine (5 mmol/L), a specific iNOS inhibitor, resulted in completeprotection against H/R-induced PP injury (Fig. 9B), thusconfirming a deleterious role for iNOS in postischemicinjury. To investigate whether exogenously administeredNO afforded protection against reoxygenation injury,mice were given spermine NO (10 �mol/L), an NO do-nor that produces 2 moles of NO per mole of parentcompound. Administration of the NO donor was with-out effect in normoxic livers (Fig. 10A). However, theH/R-induced ALT release from PP livers was prevented inthe presence of spermine NO (Fig. 10B). Measurementsof nitrate and nitrite in liver perfusates were at their limitof detection; thus, the production of NO under the dif-ferent pharmacological and genetic conditions cannot bereadily assessed. Notwithstanding, the collective resultsdemonstrate that (1) eNOS-derived NO is critical formaintaining PP hepatocyte integrity during hypoxia, (2)iNOS-derived NO is deleterious, and (3) exogenous NOcontributes to protection against reoxygenation injury.

DiscussionThe purpose of the current study was to investigate

potential differences in hepatocellular susceptibility in the

different hepatic zones to I/R and the mechanisms under-lying this differential susceptibility. We adopted a normalflow perfusion system by means of anoxic gas saturation(i.e., H/R model) to mimic the I/R model—without thecomplications caused by changes of blood flow and sheerstress—to evaluate the direct effect of oxygen deficit andoxidative stress.19,21 To achieve enrichment of selectivehepatic zones in mouse liver, we adapted the previouslypublished in situ rat liver perfusion model, using digito-nin perfusion for specific destruction of upstream ordownstream regions of the liver along the hepatic sinu-soid, to obtain PV or PP zones.7 Our results demonstratedfor the first time the successful establishment of an in situmouse model to obtain PP- and PV-enriched hepaticzones by the digitonin perfusion protocol. Histologicalevaluations showed that digitonin did not compromisethe hepatic microarchitecture; clearly delineated PP andPV zones were evident after perfusion of digitonin intothe inferior vena cava and portal vein. Despite some ex-tracellular fluid accumulation and sinusoidal dilatation,there was no evidence of intracellular swelling in hepato-cytes, KCs, and sinusoidal endothelial cells as a result ofdigitonin treatment. Zone enrichments were validated byanalyses of the distribution of the PV-specific enzyme,PK, relative to ALT.7,8 We have capitalized on this newlyestablished experimental mouse model to investigate he-patic zonal susceptibility to H/R and to define the contri-butions of ROS and NO to the injury process.

Our data show that PP hepatocytes sustained greaterH/R-induced injury than PV hepatocytes, in agreementwith previous studies in rats.12 This enhanced susceptibil-

Fig. 10. Effect of exogenous NO on H/R-induced PP injury. Wild-typemice were given spermine NO (10 �mol/L), an NO donor at the start ofhypoxia initiation. PP hepatocellular injury was assessed by ALT release intoperfusates collected during the hypoxic phase (shaded area, 0–40 minutes)and at every 10 minutes during the reoxygenation phase (unshaded area,40–80 minutes). For normoxic controls, livers from wild type mice wereprepped for PP-zone enrichment and perfused with normoxic buffer for 80minutes without or with 10 �mol/L spermine NO. (A) ALT release undernormoxic conditions in the presence of spermine NO (open square). Datarepresent mean � SEM: n � 5 mice. (B) ALT release during H/R in theabsence (open circle) or presence (open square) of spermine NO. Datarepresent mean � SEM: n � 5 for untreated and spermine NO-treated mice.*P � .05 compared with untreated mice.

Fig. 9. Effect of iNOS on H/R-induced PP injury was determined usingPP-zone-enriched livers from (A) iNOS-deficient mice (iNOS-/-) or from (B)wild-type mice treated with aminoguanidine (AG), an inhibitor of iNOS.Livers were perfused with digitonin for PP-zone enrichment and thensubjected to H/R. PP hepatocellular injury was assessed by ALT releaseinto perfusates collected during the hypoxic phase (shaded area, 0–40minutes) and at every 10 minutes during the reoxygenation phase(unshaded area, 40–80 minutes). (A) Data represent mean � SEM:n � 5 for wild-type mice (open circle) and iNOS-deficient mice (opensquare). *P � .05 compared to wild-type mice. (B) Data representmean � SEM: n � 5 for untreated wild-type mice (open circle) andaminoguanidine-treated wild-type mice (open square). *P � .05 com-pared to untreated mice.

1550 TANIAI ET AL. HEPATOLOGY, June 2004

ity of PP hepatocytes to oxygen lack is consistent with asignificantly higher oxygen requirement for maximal mi-tochondrial function and maintenance of cellular ho-meostasis in zone 1 hepatocytes.8,24 Moreover, the resultsimplicate KC-derived O2

� in mediating this H/R-inducedreoxygenation injury. Our conclusion is supported by sev-eral lines of evidence. First, evidence for a role for ROScomes from findings that H/R-induced ALT release isabolished by the administration of NAC25 and SOD2/3;notably, the attenuation by SOD2/3 suggests an involve-ment for O2

�. Because SOD2/3 was engineered to attachto endothelial cell surfaces by means of its 26-amino-acidheparin-binding tail,26 the enzyme serves to dismutateO2

� generated from within the lumen of the sinusoids (asof KCs) or produced from endothelial cells which thendiffuses into the extracellular milieu. These results under-score the importance of the endothelial barrier in O2

� elimi-nation during reoxygenation and are in agreement withprevious studies, in which the extent of oxidative damage ofhepatic parenchymal cells was found to be a function of thebalance between oxidant production from KCs and the an-tioxidant defenses of endothelial cells.27 Moreover, the re-sults suggest that, clinically, exogenous administration ofSOD2/3 and/or NAC may prove to be useful strategiesagainst hepatic I/R injury in perfused storage livers for trans-plants. That KCs were the likely source of O2

� was evidencedby the finding that inactivation of KCs with GdCl3 pretreat-ment essentially abolished the reoxygenation injury. The ki-netics of reoxygenation injury (i.e., within 40 minutes) areconsistent with a rapid activation of KCs and with the pro-duction of ROS and with the current paradigm that KC-derived O2

� mediates an early reperfusion injury that isindependent of neutrophil involvement.1–5 Notably, thefinding that the H/R-induced hepatocyte injury was local-ized to the PP zones is consistent with the predominant lo-cation of KCs as well as the preferential activation of PP KCsduring oxidative challenge, including that of I/R17,18 in he-patic zone 1.

There are numerous reports that NO plays a protectiverole in hepatic I/R injury.28,29 Our results show that hyp-oxia alone mediates PP hepatocyte injury in livers defi-cient in eNOS expression, suggesting that eNOS-derivedNO is essential for PP hepatocyte survival during anoxia.It is unclear how eNOS affords protection against hy-poxia-induced PP injury. One possibility is that eNOS-derived NO mediates dilation of terminal portal venulesand sinusoids to maintain hepatic microcirculation afteran ischemic/anoxic perturbation.27,30 Since vascularsmooth muscle cells of terminal portal venule and Ito cellscontrol hepatic microcirculation, and Ito cells are distrib-uted to PP zones,8 NO effects on these vasoresponsivecells could ultimately determine PP oxygenation and the

extent of hypoxia-induced injury. The finding that hyp-oxia did not elicit PV injury provides additional support tothe suggestion that eNOS-derived NO, smooth muscle cells,and the PP distribution of Ito cells participate in the hypoxicphase of the PP injury response. Curiously, the enzymaticinhibition of cellular NOS with L-NAME elicited a greaterPP hypoxia-induced eNOS deficiency, presumably due toother vascular effects of L-NAME on blood flow.29,30 Be-cause neutrophils are not involved in early hypoxia/reoxy-genation injury, we can rule out an effect of L-NAME onpolymorphonuclear leukocytes. Protection of PP hepato-cytes by endogenous NO against hypoxic damage may berelated to the conservation of cellular bioenergetics for main-tenance of tissue homeostasis during hypoxia via NO-medi-ated inhibition of the use of mitochondrially derived ATPfor nonessential metabolic processes.8,19,31

It is notable that the hypoxia-induced elevation in ALTin eNOS-deficient mice was sustained throughout thereoxygenation period without additional enzyme release.This suggests that endogenously derived NO plays a min-imal role in reoxygenation injury; however, the findingthat an exogenous NO donor afforded protection againstPP injury during reoxygenation indicates that NO is, infact, cytoprotective in the reoxygenation phase. The en-zymatic or cellular source of NO during reoxygenation isunknown. A potential candidate is iNOS. Sonin et al.have shown that iNOS messenger RNA was induced six-fold following 1 hour of hepatic ischemia and 6 hours ofreperfusion.32 Other studies have demonstrated the in-duction of iNOS in KCs in a variety of inflammatoryprocesses, including I/R, endotoxemia, and multiple or-gan failure.33–36 In this study, we found no increases iniNOS message or protein expression over the short reoxy-genation times (40 minutes, data not shown); this is inagreement with our previous findings that iNOS induc-tion is not detectable following hepatic I/R.28 These re-sults are consistent with the paradigm that injury events inthe early phase of reoxygenation were transcriptionallyindependent. Notwithstanding, we found that the H/R-induced injury was ameliorated in iNOS-deficient ani-mals (Fig. 9). The reason for this curiosity is unclear; onepossibility may be that iNOS gene depletion results in acompensatory upregulation of eNOS. Although an inter-esting possibility, there is no evidence to support it. An-other possibility is that there may exist a baseline level ofiNOS that can participate in the early initiation of I/Rinjury that is independent of transcriptional control.28,36

The fact that NAC can inhibit NO production by iNOSsuggests that the protective effect of NAC against H/R-induced PP injury may, in part, be related to its suppres-sion of iNOS function,25 apart from its role as anantioxidant. The possibility that differential NO produc-

HEPATOLOGY, Vol. 39, No. 6, 2004 TANIAI ET AL. 1551

tion resulting from eNOS and iNOS inhibition may ex-plain our results cannot be satisfactorily resolved atpresent. The large perfusate volumes from single-pass per-fusion precluded quantitative measurements of NO me-tabolites and assessment of NO formation under theseconditions.

In summary, we have established an in situ mouse per-fusion model to evaluate potential differences in suscep-tibility of PP and PV hepatocytes to hepatic H/R. Ourresults demonstrated marked zonal differences in suscep-tibility to H/R-induced damage, with the PP hepatocytesbeing more vulnerable than the PV hepatocytes. H/R-induced PP hepatocellular injury was mediated by KC-derived O2

�, while endogenously generated NO by eNOSprovided a restraint to the development of the hypoxicinjury response.

Acknowledgment: The authors thank Drs. JohnFuseler and Wen Ma and Ms. Carla Brown for assistancewith the liver histology.

References1. Fan C, Zwacka RM, Engelhardt JF. Therapeutic approaches for ischemic/

reperfusion injury in the liver. J Mol Med 1999;77:577–592.2. Jaeschke H. Mechanisms of reperfusion injury after warm ischemia of the

liver. J Hepatobiliary Pancreat Surg. 1998;5:402–408.3. Muller MJ, Vollmar B, Friedl HP, Menger MD. Xanthine oxidase and

superoxide radicals in portal triad crossclamping-induced microvascularreperfusion injury of the liver. Free Radic Biol Med 1996;21:189–197.

4. Rauen U, Viebahn R, Lauchart W, de Groot H. The potential role ofreactive oxygen species in liver ischemia/reperfusion injury following liversurgery. Hepatogastroenterology. 1994;41:333–336.

5. Scales WE, Campbell DA, Green ME, Remick DG. Hepatic ischemia/reperfusion injury: importance of oxidant/tumor necrosis factor interac-tions. Am J Physiol 1994;267:G1122–G1127.

6. Lee VM, Cameron RG, Archer MC. Zonal location of compensatoryhepatocyte proliferation following chemically induced hepatotoxicity inrats and humans. Toxicol Pathol 1998;26:621–627.

7. Quistorff B. Gluconeogenesis in periportal and perivenous hepatocytes ofrat liver, isolated by a new high-yield digitonin/collagenase perfusion tech-nique. Biochem J 1985;229:221–226.

8. Jungermann K, Kietzmann T. Oxygen: modulator of metabolic zonationand disease of the liver. HEPATOLOGY 2000;32:255–260.

9. Lindros KO, Penttila KE. Digitonin-collagenase perfusion for efficientseparation of periportal or perivenous hepatocytes. Biochem J 1985;228:757–760.

10. Suematzu M, Suzuki H, Ishii H, Kato S, Yanagisawa T, Asako H, et al.Early midzonal oxidative stress preceding cell death in hypoperfused ratliver. Gastroenterology 1992;103:994–1001.

11. Suzuki H, Suematzu M, Ishii H, Kato S, Miki H, Mori M, et al. Prostag-ladin E1 abrogates early reductive stress and zone-specific paradoxical ox-idative injury in hypoperfused rat liver. J Clin Invest 1994;93:155–164.

12. Kato Y, Tanaka J, Koyama K. Intralobular heterogeneity of oxidative stressand cell death in ischemia-reperfused rat liver. J Surg Res 2001;95:99–106.

13. Jaeschke H, Farhood A. Neutrophil and Kupffer cell-induced oxidantstress and ischemia-reperfusion injury in rat liver. Am J Physiol 1991;260:G355–G362.

14. Colletti LM, Remick DG, Burtch GD, Kunkel SL, Strieter RM, CampbellDA. Role of tumor necrosis factor-alpha in the pathophysiological alter-ations after ischemia/reperfusion injury in the rat. J Clin Invest 1990;85:1936–1943.

15. Horie Y, Wolf R, Russell J, Shanley TP, Granger DN. Role of Kupffer cellsin gut ischemia/reperfusion-induced hepatic microvascular dysfunction inmice. HEPATOLOGY 1997;26:1499–1505.

16. Gao W, Yakei Y, Marzi I, Lindert KA, Caldwell-Kenkel JC, Currin RT, et al.Carolina rinse solution—a new strategy to increase survival time after ortho-topic liver transplantation in the rat. Transplantation 1991;52:417–424.

17. Jaeschke H, Baustista AP, Spolarics Z, Spitzer JJ. Superoxide generation byKupffer cells and priming of neutrophils during reperfusion after hepaticischemia. Free Radic Res Commun 1991;15:277–284.

18. Baustista AP, Meszaros K, Bojta J, Spitzer JJ. Superoxide anion generationin the liver during the early stage of endotoxemia in rats. J Leukoc Biol1990;49:123–128.

19. Taniai H, Suematsu M, Suzuki T, Norimizu S, Hori R, Ishimura Y, et al.Endothelin B receptor-mediated protection against anoxia-reoxygenationinjury in perfused rat liver: nitric oxide-dependent and -independentmechanisms. HEPATOLOGY 2001;33:894–901.

20. Suematsu M, Goda N, Sano T, Kashiwagi S, Egawa T, Shinoda Y, et al.Carbon monoxide: an endogenous modulator of sinusoidal tone in theperfused rat liver. J Clin Invest 1995;96:2431–2437.

21. Jaeschke H, Smith CV, Mitchell JR. Hypoxic damage generates reactiveoxygen species in isolated perfused rat liver. Biochem Biophys Res Com-mun 1988;150:568–574.

22. Imamura K, Tanaka T. Pyruvate kinase isozymes from rat. Meth Enzymol1982;90:150–165.

23. Kubes P, Suzuki M, Granger DN. Nitric oxide: an endogenous modulatorof leukocyte adhesion. Proc Natl Acad Sci USA 1991;88:4651–4655.

24. Jungermann K. Metabolic zonation of carbohydrate metabolism in theliver. In: Lemasters JJ, Hackenbrock CR, Thurman RG, eds. Integration ofMitochondrial Function. New York: Plenum Press, 1988:561–579.

25. Bergamini S, Rota C, Canali R, Staffieri M, Daneri F, Bini A, et al. N-ace-tylcysteine inhibits in vivo nitric oxide production by inducible nitric oxidesynthase. Nitric Oxide 2001;5:349–360.

26. Gao B, Flores SC, Leff JA, Bose SK, McCord JM. Synthesis and anti-inflammatory activity of a chimeric recombinant superoxide dismutase:SOD2/3. Am J Physiol Lung Cell Mol Physiol 2003;284:L917–L925.

27. Spolarics Z. Endotoxemia, pentose cycle, and the oxidant/antioxidant bal-ance in the hepatic sinusoid. J Leukoc Biol 1998;63:534–541.

28. Hines IN, Harada H, Bharwani S, Pavlick KP, Hoffman JM, GrishamMB. Enhanced post-ischemic liver injury in iNOS-deficient mice: a cau-tionary note. Biochem Biophys Res Commun 2001;284:972–976.

29. Kawachi S, Hines IN, Laroux FS, Hoffman J, Bharwani S, Gray L, et al.Nitric oxide synthase and postischemic liver injury. Biochem Biophys ResCommun 2000;276:851–854.

30. Wang Y, Lawson JA, Jaeschke H. Differential effect of 2-aminoethyl-isothiourea, an inhibitor of the inducible nitric oxide synthase, on micro-vascular blood flow and organ injury in models of hepatic ischemia-reperfusion and endotoxemia. Shock 1998;10:20–25.

31. Shiomi M, Wakabayashi Y, Sano T, Shinoda Y, Nimura Y, Ishimura Y, et al.Nitric oxide suppression reversibly attenuates mitochondrial dysfunction andcholestasis in endotoxemic rat liver. HEPATOLOGY 1998;27:108–115.

32. Sonin NV, Garcia-Pagan JC, Nakanishi K, Zhang JX, Clemens MG. Pat-terns of vasoregulatory expression in the liver response to ischemia/reper-fusion and endotoxemia. Shock 1999;11:175–179.

33. MacMicking JD, Nathan C, Hom G, Chartrain N, Fletcher DS, Trum-bauer M, et al. Altered responses to bacterial infection and endotoxic shockin mice lacking inducible nitric oxide synthase. Cell 1995;81:641–650.

34. Cuzzocrea S, Mazzon E, Dugo L, Barbera A, Centorrino T, Ciccolo A, etal. Inducible nitric oxide synthase knockout mice exhibit resistance to themultiple organ failure induced by zymosan. Shock 2001;16:51–58.

35. Lorente JA, Tejedor C, Delgado MA, Fernandez-Segoviano P, Jara N,Tobalina R, et al. Hemodynamic, biochemical and morphological changesinduced by aminoguanidine in normal and septic sheep. Intensive CareMed 2000;26:1670–1680.

36. Lee VM, Cameron RG, Archer MC. Zonal location of compensatoryhepatocyte proliferation following chemically induced hepatotoxicity inrats and humans. Toxicol Pathol 1998;26:621–627.

1552 TANIAI ET AL. HEPATOLOGY, June 2004