Main

Hepatic ischemia reperfusion (IR) injury is a major clinical problem during the perioperative period. Hepatic IR injury is a major cause of acute liver failure (ALF) and is a frequent complication of major hepatic resection, liver transplantation or septic shock.1, 2, 3 Acute kidney injury (AKI) occurs frequently in patients with ALF and is a serious clinical problem during the perioperative period.4, 5 For example, the incidence of AKI in patients with ALF ranges from 40 to 85% depending on the diagnostic criteria and etiology.4, 6 In addition, AKI is a common complication after orthotopic liver transplantation and it exponentially increases patient mortality.4 Unfortunately, the pathophysiology of AKI associated with ALF is poorly understood in part due to a lack of diagnostic histological changes or an adequate animal model for this critical clinical syndrome. Moreover, development of AKI after liver injury in mice has never been described in detail.

Previous animal models for AKI associated with liver failure include carbon tetrachloride (CCl4) injection after bile duct ligation in mice7 and partial hepatic IR in rats.8 The CCl4 injection model is limited by the chronic nature of the protocol requiring several8, 9, 10, 11, 12 weeks (whereas human renal failure associated with ALF can develop acutely). The rat model of AKI after liver IR is limited by the modest nature of renal dysfunction after liver IR injury, which, therefore, fails to mimic the severity of human disease.8, 9

In this study, we tested the hypothesis that mice subjected to severe liver IR develop AKI resembling human AKI associated with ALF. We demonstrate that mice subjected to hepatic IR develop AKI in less than 24 h characterized by proximal tubular necrosis and inflammation. Moreover, renal peritubular endothelial cell apoptosis, peritubular/interstitial neutrophil infiltration, JGA hyperplasia as well as marked disruptions of the kidney tubular actin cytoskeletal architecture were observed. Our results show that mice subjected to liver IR develop AKI rapidly, closely mimicking the time course of this complication in humans.

MATERIALS AND METHODS

Murine Model of Hepatic IR

After Columbia University IACUC approval, male C57BL/6 mice (25–30 g; Harlan, Indianapolis, IN, USA) were anesthetized with intraperitoneal pentobarbital, 50 mg/kg or to effect. Mice were placed under a heating lamp and on a 37°C heating pad. After a midline laparotomy and intraperitoneal application of 20 U heparin, left lateral and median lobes of the liver were subjected to 60 min of ischemia, with a microaneurysm clip occluding the hepatic triad above the bifurcation. This method of partial hepatic ischemia results in a segmental (∼70%) hepatic ischemia but spares the right lobe of the liver and prevents mesenteric venous congestion by allowing portal decompression through the right and caudate lobes of the liver.10, 11 Preliminary studies showed that 30, 60 or 90 min of liver ischemia causes mild, moderate or severe liver injury, respectively. The liver was kept moist with gauze soaked in 0.9% normal saline. The body temperature was monitored by an infrared temperature sensor (Linear Laboratories, Fremont, CA, USA) and maintained at 37°C using a heating lamp and a heating pad. After 60 min, the liver was reperfused and the wound closed. Sham-operated mice were subjected to laparotomy and identical liver manipulations without the vascular occlusion. Twenty-four hours after reperfusion, the liver tissue subjected to IR and both kidneys were collected to measure necrosis, neutrophil infiltration (with immunohistochemistry) and apoptosis (with terminal deoxynucleotidyl transferase-mediated dUTP nick-end labeling staining). In some mice, liver and kidneys were collected 4 h after reperfusion to detect renal inflammation by RT–PCR for proinflammatory cytokine mRNAs. We also collected plasma for the measurement of alanine aminotransferase (ALT) and creatinine (Cr) 4 and 24 h after reperfusion.

Assessment of Liver and Kidney Dysfunction

Plasma ALT was measured by using a Prep-Profile II kit and a VetScan VS2 Point-of-Care Analyzer (Union City, CA, USA) and expressed as Units/liter (U/l). Renal dysfunction after liver IR injury was assessed by measurement of plasma creatinine by a colorimetric method based on the Jaffe reaction.12

Histology and Quantification of Hepatic and Renal Injury after Hepatic IR Injury

Twenty-four hours after reperfusion, the liver tissue subjected to IR and both kidneys were collected. For histological preparations, explanted murine livers or kidneys were fixed in 10% formalin solution overnight. After automated dehydration through a graded alcohol series, transverse liver slices were embedded in paraffin, sectioned at 4 μm, and stained with hematoxylin-eosin (H&E). To quantify the degree (%) of hepatic necrosis, liver H&E sections were digitally photographed and the percent of necrotic area was quantified with NIH IMAGE (Image-J, 1.37v) software by a person who was unaware of the treatment each sample received. Renal H&E sections were evaluated for renal proximal tubule necrosis and apoptosis, endothelial cell apoptosis, tubular simplification, juxtaglomerular apparatus hypertrophy and cell number by an experienced renal pathologist (VDD) who was blinded to the treatment each animal had received.

Assessment of Renal Inflammation

Kidney inflammation after hepatic ischemia was determined by the detection of neutrophil infiltration by immunohistochemistry 24 h after hepatic IR as described previously13, 14, 15 and by measuring mRNA encoding markers of inflammation, including keratinocyte derived cytokine (KC), intercellular adhesion molecule-1 (ICAM-1), monocyte chemoattractive protein-1 (MCP-1), macrophage inflammatory protein-2 (MIP-2), and tumor necrosis factor-alpha (TNF-α) 4 h after liver IR. Both semiquantitative13, 14, 15 and quantitative real-time RT-PCR (Q-RT-PCR) were performed. Q-RT-PCR was performed with the MyiQ Real Time Detection System (Bio-Rad, Hercules, CA, USA) using SYBR Green I Brilliant Mastermix (Stratagene, La Jolla, CA, USA). The cDNA template was synthesized using Omniscript Reverse Transcriptase and oligo-dT primer (Qiagen, Valencia, CA, USA). Specificity of the amplification was checked by melting curve analysis and by agarose gel electrophoresis. All reactions were performed in duplicate with appropriate negative controls. The Ct values were determined by using Mx3000P software. Values were normalized for GAPDH mRNA and relative expression of proinflammatory mRNA was calculated with the ΔΔCt method.

Detection of Renal Apoptosis with In Situ Terminal Deoxynucleotidyl Transferase Biotin-dUTP Nick-End Labeling Assay

Twenty-four hours after reperfusion, the liver tissue subjected to IR and both kidneys were collected. We used in situ terminal deoxynucleotidyl transferase biotin-dUTP nick-end labeling (TUNEL) staining to detect DNA fragmentation in apoptosis as described previously.16, 17

Filamentous-Actin Staining of Liver and Kidney Sections after Liver IR Injury

Twenty-four hours after reperfusion, the liver tissue subjected to IR and both kidneys were collected. As breakdown of filamentous (F)-actin occurs early after IR, we visualized the F-actin cytoskeleton by staining with phalloidin as an early index of liver as well as renal injury.18, 19 We visualized the F-actin cytoskeleton by staining with phalloidin. Twenty-four hours after liver IR, liver and kidney tissue were embedded in Tissue-Tek oxytetracycline compound (Fisher Scientific, Pittsburgh, PA, USA) and cut into 5 μm sections. To reduce background staining, the sections were incubated in 1% FBS dissolved in PBS for 10 min at room temperature. The sections were then stained with Alexafluor 594 (Red)-labeled phalloidin (Invitrogen, Carlsbad, CA, USA) for 30 min at 37°C in a humidified chamber in the dark. Sections were then washed twice in PBS and mounted with Vectashield (Vector Laboratories, Burlingame, CA, USA). F-actin images were visualized with an Olympus IX81 epifluorescence microscope (Tokyo, Japan) and captured and stored using SlideBook 4.2 software (Intelligent Imaging Innovations Inc., Denver, CO, USA) on a personal computer.

Analysis of Liver and Kidney Vascular Permeability

Twenty-four hours after reperfusion, the liver tissue subjected to IR and both kidneys were collected. Changes in liver and kidney vascular permeability were assessed by quantitating extravasations of Evans blue dye (EBD) into the tissue as described by Awad et al20 with some modifications. Briefly, 2% EBD (Sigma Biosciences, St Louis, MO, USA) was administered at a dose of 20 mg/kg i.v. 24 h after liver injury. One hour later, mice were killed and perfused through the heart with PBS and EDTA with 10 cc cold saline with heparin (100 U/ml). Liver and kidneys were then removed, allowed to dry overnight at 60°C, and the dry weights were determined. EBD was extracted in formamide (20 ml/g dry tissue; Sigma Biosciences), homogenized, and incubated at 60°C overnight. Homogenized samples were centrifuged at 5000 g for 30 min and the supernatants were measured at 620 and 740 nm in a spectrophotometer. The extravasated EBD concentration was calculated against a standard curve and the data expressed as micrograms of EBD per gram of dry tissue weight.

Protein Determination

Protein contents were determined with a bicinchoninic acid protein assay kit (Pierce Chemical Co., Rockford, IL, USA), using BSA as a standard.

Statistical Analyses

The data were analyzed with Student's t-test when comparing means between two groups or with one-way analysis of variance plus Tukey's post hoc multiple comparison test to compare mean values across multiple treatment groups. In all cases, significance was assumed at a probability statistic of <0.05. All data are expressed throughout the text as mean±s.e.

Reagents

Unless otherwise specified, all reagents were purchased from Sigma.

RESULTS

Acute Hepatic and Renal Dysfunction after Liver IR

The survival rates for sham-operated animals and animals subjected to 60 min of liver ischemia and 24 h reperfusion were 100% (5/5) and 94% (15/16), respectively. Sham-operated C57BL/6 mice had normal plasma ALT and Cr at 4 h (ALT=61±12 mg/100 ml, N=4 and Cr=0.28±0.09 mg/100 ml, N=4) and 24 h after surgery (ALT=58±11 U/l, N=5 and Cr=0.31±0.11 mg/100 ml, N=5). However, C57BL/6 mice subjected to liver IR developed severe liver dysfunction at 4 and 24 h after hepatic ischemic injury with significantly higher plasma ALT levels (20 097±1434 U/l, N=6, P<0.0001 and 14 560+2275 U/l, N=10, P<0.0001, respectively, compared to sham-operated mice). Moreover, C57BL/6 mice subjected to liver IR also developed AKI with significant rises in plasma Cr 4 h (Cr=0.60±0.09 mg/100 ml, N=6, P<0.05 vs sham) and 24 h (Cr=0.91±0.15 mg/100 ml, N=10, P<0.05 vs sham) after liver IR. There was a direct relationship between the severity of liver dysfunction (ALT) and the degree of AKI (Cr) 24 h after IR (P<0.0001 and r2=0.8925; Figure 1). The BUN values also significantly increased in mice subjected to liver ischemia and 24 h reperfusion (128±10 mg/100 ml, N=10, P<0.0001) compared to the mice subjected to sham surgery (12±2 mg/100 ml, N=4).

Figure 1
figure 1

Correlation between plasma ALT and creatinine (Cr) values (P<0.0001 and r2=0.8925). C57BL/6 mice were subjected to 60 min liver ischemia and 24 h reperfusion, and plasma collected for ALT and creatinine assays (N=10).

Full size image

Hepatic IR Injury Results in Histological Changes in the Kidney

As we observed previously, our model of 60 min partial hepatic IR injury produced moderate to severe liver necrosis (76±9%, N=10) in all samples examined 24 h after IR with some sparing of the hepatocytes near the portal triad (Figure 2). We failed to detect liver necrosis in sham-operated animals. When we examined the kidneys from the mice subjected to liver IR, we observed multifocal acute tubular injury including individual cell necrosis involving the juxtamedullary proximal tubules of the S3 segment. In addition the cortical tubules exhibited focal tubular simplification, cytoplasmic vacuolization, dilated lumina and focal granular bile/heme casts, accompanied by mild interstitial edema. Endothelial apoptosis of peritubular capillaries was also visible in H&E sections as linear arrays of apoptotic bodies corresponding to interstitial capillary lining endothelium (Figure 3). Manual counting of cells in the S3 segment of proximal tubules showed significantly higher number of necrotic cells compared to the sham-operated mice (Figure 3). The number of glomeruli with detectable juxtaglomerular apparatus in H&E sections and the number of cells per juxtaglomerular apparatus were both increased after liver IR, indicating hyperplasia of the juxtaglomerular apparatus (Figures 3 and 4).

Figure 2
figure 2

Representative photomicrographs of five experiments (hematoxylin and eosin staining, magnification × 200 (a and c) and × 400 (b and d) of mice subjected to sham operation or to liver IR injury. Sham-operated animals show normal-appearing hepatocyte parenchyma (a and b). Following liver IR, there is severe hepatocyte necrosis and loss of normal liver architecture with some sparing of the hepatocytes near the portal triad. PV=portal vein. BD=bile duct.

Full size image
Figure 3
figure 3figure 3

Representative photomicrographs of six experiments (hematoxylin and eosin staining, magnification × 600) of mice subjected to sham operation or to liver IR injury. Sham-operated animals show normal-appearing glomeruli (a) and tubules (b). Following liver IR, many glomeruli display prominent hyperplasia of the juxtaglomerular apparatus (arrow) located at the glomerular hilus (c). There is focal coarse clear cytoplasmic vacuolization (arrow) of proximal tubular epithelial cells (d). Some cortical tubular cells display acute epithelial injury with cellular condensation and cytoplasmic hypereosinophilia (long arrow) and shrunken, hyperchromatic nuclei (insert expanded; e). There is focal apoptosis of interstitial capillary endothelial cells (insert expanded) with marked vascular stasis within the interstitial capillary lumina (long arrow; f, g). We also observed focal tubular dilatation with intraluminal shedding of cytoplasmic debris and formation of granular eosinophilic casts (arrow; h).

Full size image
Figure 4
figure 4

Renal histological assessment after liver IR in C57BL/6 mice. Mice were subjected to sham operation (N=5) or to liver IR (N=8). Twenty-four hours later, mice subjected to liver IR showed increased number of necrotic S3 proximal tubules per 400 × field (a), increased number of juxtaglomerular apparatus (JGA) per 10 glomeruli counted (b) and increased number of cells per JGA (c). *P<0.01 vs sham-operated mice. #P<0.05 vs sham-operated mice. Data presented as mean±s.e.m.

Full size image

Hepatic IR Injury Increases Proinflammatory mRNA Expression in the Liver as well as in the Kidney

With RT–PCR, we measured the expression of proinflammatory cytokine mRNAs in the liver and the kidney 4 h after liver IR. We showed previously that 60 min of hepatic ischemia and 4 h reperfusion resulted in significant upregulation of proinflammatory cytokine mRNAs in the liver compared to the sham-operated mice21 and again demonstrate this finding here (Figure 5a and c). We now show that hepatic IR injury was associated with significantly increased proinflammatory mRNA expression (ICAM-1, TNF-α, KC, MCP-1 and MIP-2) in the kidney (Figure 5b and c) compared to the sham-operated mice.

Figure 5
figure 5

Representative gel images of semiquantitative RT-PCR of the pro-nflammatory markers ICAM-1, KC, MCP-1, MIP-2 and TNF-α from liver tissues (a) or from renal cortices (b) of C57BL/6 mice subjected to sham operation or to liver IR (60 min ischemia and 4 h reperfusion). Fold increases in proinflammatory mRNAs (in log10 scale) normalized to GAPDH from Q-RT-PCR reactions for each indicated mRNA (N=4 for sham and N=4 for liver IR) are shown (c). *P<0.05 vs sham-operated mice. Error bars represent 1 s.e.m.

Full size image

Hepatic IR Injury Increases Renal Neutrophil Infiltration 24 h after IR Injury

We showed previously that 60 min of hepatic ischemia and 24 h of reperfusion resulted in a significant influx of neutrophils into the liver compared to the sham-operated mice.21 Neutrophil infiltration was not detected in the kidneys from sham-operated mice, however, infiltration was easily visible in the kidneys from mice subjected to liver IR (Figure 6). Neutrophil infiltration was seen in both capillaries and in the interstitial space. Capillary infiltrations were seen in juxtamedullary cortex with extravasations occurring in small foci around peritubular capillaries. Interstitial infiltrations were more prominent involving the subcapsular and outer cortical regions.

Figure 6
figure 6

Representative photomicrographs (100 × ) of four experiments of immunohistochemistry for neutrophil infiltration (arrows) in the outer medulla/inner cortex of the kidneys of C57BL/6 mice subjected to sham operation or to 60 min liver ischemia and 24 h reperfusion.

Full size image

Hepatic IR Injury Increases Renal Apoptosis 24 h after IR Injury

When the kidneys isolated from mice subjected to liver IR were examined for apoptosis, we observed significantly increased number of apoptotic nuclei (TUNEL-positive cells; Figure 7). TUNEL-positive cells were heavily localized to the interstitial capillary area rather than to the renal proximal tubule cells (cells typically showing apoptotic changes after renal IR). To clarify the location of TUNEL-positive cells, we double stained the kidney sections with hematoxylin and TUNEL stain. Figure 8 shows again that the TUNEL-positive cells are localized mostly in the interstitial perivascular area. We further confirmed that the TUNEL-positive cells are interstitial capillary endothelial cells by staining parallel kidney sections with TUNEL and CD34 (an endothelial cell marker; Abcam Inc., Cambridge, MA, USA) and confirmed that TUNEL-positive cells also stained for CD34 (Figure 9; representative of four representative serial sections).

Figure 7
figure 7

Representative fluorescence photomicrographs (of four experiments) of kidney sections illustrating apoptotic nuclei (terminal deoxynucleotidyl transferase biotin-dUTP nick-end labeling (TUNEL) fluorescence staining, 100 × ). C57BL/6 mice were subjected to sham-operation or to 60 min liver ischemia and 24 h reperfusion.

Full size image
Figure 8
figure 8

Renal endothelial cell apoptosis after liver IR. C57BL/6 mice were subjected to sham operation or to 60 min liver ischemia and 24 h reperfusion. Top panel: Hematoxylin stain (400 × ) of kidneys from sham-operated mouse (a) or mouse subjected to liver IR (b). Middle panel: fluorescence photomicrographs of TUNEL staining from kidneys of mice subjected to sham operation (c) or to liver IR (d). Bottom panel: Merge of a and c (Sham, e) revealed no evidence of apoptotic endothelial cells. In contrast, images from b and d (Liver IR) were merged (f) demonstrating apoptotic nuclei (represent in pink) within the renal endothelial cells in the kidney of the mouse subjected to liver IR. Some of the endothelial cells are labeled (a–e) showing overlapping nuclear stain and apoptotic stain. Representative of six experiments.

Full size image
Figure 9
figure 9

Further confirmation of renal endothelial cell apoptosis after liver IR. C57BL/6 mice were subjected to 60 min liver ischemia and 24 h reperfusion and serial kidney sections were stained with TUNEL (a) or with CD34 (an endothelial cell marker, b). TUNEL-positive endothelial cells are labeled (a–k). Representative of 4 serial sections stained.

Full size image

Hepatic IR Disrupts Both Hepatocyte and Renal Tubule F-Actin Architecture

Liver F-actin staining in sham-operated mice shows localization to the hepatocyte periphery and around the bile canaliculus. As expected, 60 min of liver ischemia and 24 h of reperfusion resulted in severe disruption of liver parenchymal F-actin compared to the sham-operated mice (Figure 10; representative of six experiments). Surprisingly, liver IR injury also resulted in a severe loss of F-actin in renal proximal tubules. In Figure 10, 24 h posthepatic IR induced disruptions of the F-actin cytoskeleton in renal proximal tubular epithelial cells are shown. C57BL/6 mice were subjected to sham surgery showed intense stain in the tubular epithelial and in the basal plasma membrane. In contrast, kidneys from mice subjected to liver IR showed loss of F-actin staining in the tubular epithelial cells.

Figure 10
figure 10

Representative fluorescent photomicrographs of phalloidin labeling to visualize F-actin in the liver (a and c) and in the renal proximal tubules (b and d) from C57BL/6 mice subjected to sham operation (a and b) or to 60 min liver ischemia and 24 h reperfusion (c and d). Representative of six independent experiments. Arrows represent some of the intact F-actin structure in the liver or in the renal proximal tubular epithelial cells.

Full size image

Hepatic IR Increases Vascular Permeability in the Liver as well as in the Kidneys

We measured liver and kidney vascular permeability after liver IR with EBD injection. EBD binds to plasma proteins and its appearance in extravascular tissues reflects an increase in vascular permeability. Analysis of EBD extravasations in sham-operated mice and mice subjected to liver IR is shown in Figure 11. Liver IR not only increased the EBD content in the liver (337±43 μg EBD/g dry liver, N=10) compared to sham-operated mice (45±8 μg EBD/g dry liver, N=6, P<0.001) but also increased the EBD content in the kidney as well (liver IR EBD: 77±6 μg EBD/g dry kidney, N=10 vs sham EBD: 46±9 μg EBD/g dry kidney, N=6, P<0.02).

Figure 11
figure 11

Liver IR increases vascular permeability in the liver as well as in the kidney. C57BL/6 mice were subjected to sham operation (N=6) or to 60 min liver ischemia and 24 h reperfusion (N=10). EBD was extracted in formamide and the amount of extravasated EBD concentration in liver and kidney was calculated against a standard curve. *P<0.05 vs sham-operated mice. Error bars represent 1 s.e.m.

Full size image

DISCUSSION

The major new finding of this study is that mice subjected to a clinically relevant model of liver IR injury developed severe AKI. Liver IR-induced AKI in mice shared similar histological changes observed in human AKI associated with liver failure including JGA hyperplasia, renal tubule necrosis, proximal tubule simplification and vacuolization. In addition, we showed that mice with AKI displayed marked renal endothelial apoptosis and F-actin disruption in the kidney. Moreover, liver IR also caused renal cortical proinflammatory mRNA expression and infiltration of PMNs.

Hepatic IR induced ALF is a very common clinical problem and frequently leads to dysfunction of remote organs including the lung, heart and kidney.6, 22, 23, 24 AKI associated with liver failure is a grave clinical problem with high mortality rate. In fact, the diagnosis of AKI in patients with liver failure implies a significant increase in mortality and morbidity.4, 25, 26, 27 Supportive therapies for AKI associated with liver failure are limited to hydration, blood pressure support and short-term hemodialysis until renal function recovers. In fact, for hepatorenal syndrome (a type of AKI associated with ALF), liver transplantation is the only definitive treatment.6 Unfortunately, the pathogenesis of AKI associated with liver failure is poorly understood, largely due to a lack of a reproducible and reliable animal model. A reproducible animal model would allow preclinical testing of therapeutic approaches that could lead to human therapy. In this study, we describe a model of AKI in mice induced with liver IR that is reproducible, rapid to develop (in less than 24 h) and mimics the histological (renal tubular damage, apoptosis, necrosis and JGA hyperplasia) as well as biochemical (plasma creatinine, inflammatory markers) changes observed with human AKI associated with ALF.6

Although the pathophysiological and biochemical mechanisms of AKI associated with liver failure are not fully understood, the initiating event appears to be portal hypertension-induced splanchnic vasodilatation with subsequent intrarenal vasoconstriction.6, 25, 26 Patients with ALF develop hypotension secondary to splanchnic and peripheral vasodilation and hypovolemia.6 The renin–angiotensin system is activated due to a reduced systemic blood pressure and splanchnic vasodilation.6, 7 Upregulation of renin–angiotensin system leads to severe reduction glomerular filtration rate, urinary sodium excretion and free water excretion. It has been proposed that intense intra-renal ischemia subsequent to renin–angiotensin activation leads to renal tubular necrosis and renal dysfunction. In this study, we show significant hyperplasia of juxtaglomerular apparatus with increased cellularity in mice subjected to liver IR. These changes are similar to the changes seen in human AKI associated with ALF reflecting intense upregulation of the renin–angiotensin system.6, 25 We also show several hallmark signs of renal damage due to intense intrarenal vasoconstriction including renal tubular necrosis, tubular simplification and vacuolization.

However, other mechanisms may play important roles in the pathogenesis of renal dysfunction with ALF. Systemic inflammatory response after liver IR may induce renal injury.6, 22 After liver IR injury, circulating proinflammatory cytokines and transcription factors including TNF-α and HMGB1 increase significantly.23, 28, 29 Release of proinflammatory cytokines such as TNF-α and HMGB1 from the liver may promote inflammatory changes in the kidney after liver IR.22, 28, 29 These proinflammatory factors can upregulate endothelial adhesion molecules in distant organs including the kidney. Upregulation of renal endothelial adhesion molecules including E-selectin, P-selectin and ICAM-1 promotes leukocyte recruitment and extravasations to the renal interstitial space.30, 31 In addition, circulating bile acids, endotoxin and circulating immune complexes may contribute to the development of AKI.6 Combined, these contributing factors may potentiate systemic inflammation leading to renal injury and renal PMN infiltration.

The endothelial barrier serves to protect against AKI by regulating leukocyte recruitment to the areas of injury.1, 30 In this study, we were surprised to observe a marked endothelial cell apoptosis in the kidney after liver IR (Figures 7 and 8). Apoptosis is an important contributor in the development of renal failure.32, 33 Apoptotic cell death represents the execution of an ATP-dependent death program often initiated by death ligand/death receptor interactions, such as Fas ligand with Fas, which leads to a caspase activation and cleavage of nuclear materials such as DNA and PARP.34, 35 After liver IR, we show that interstitial capillary endothelial cell apoptosis in the kidney was much more prominent than the proximal tubule cell apoptosis. This is in contrast to the findings after renal IR where proximal tubule cell apoptosis predominates. Endothelial cell death due to apoptosis would impair defense against leukocyte invasion into the kidney leading to a further exacerbation of renal injury. The results of our study suggest that reducing endothelial cell death may limit renal injury associated with ALF.

In this study, we demonstrate not only significant endothelial cell apoptosis but also recruitment of PMNs to the kidney after liver IR. Neutrophils are activated during and after liver ischemia and activated neutrophils attach to and then transverse the renal capillary endothelium into the subendothelial space, where they release enzymes and cytokines, causing direct renal injury and the recruitment of other injurious cells, such as monocytes and macrophages.36, 37 Weakened endothelial defenses after liver IR may potentiate the PMN infiltration and conversely, increasing the survival of endothelial cells after liver IR may limit PMN infiltration into the kidney and improve renal function. Future studies with endothelial stabilizing agents (such as activated protein C or sphingosine-1-phosphate) will determine whether protecting endothelial integrity will reduce renal injury after liver IR injury.

IR injury in vivo results in degradation of F-actin. Breakdown of F-actin contributes to the pathophysiology of organ dysfunction after IR injury or sepsis.38 Whereas we were not surprised to see changes in liver F-actin content after liver IR (Figure 9), we were surprised to demonstrate that renal F-actin breakdown occurs with liver IR. Actin cytoskeletal damage contributes significantly to the development of AKI.38, 39 Moreover, F-actin disruption promotes apoptosis in several cell lines.40 Therefore, loss of actin cytoskeleton may have contributed to the development of renal tubular and endothelial apoptosis after liver IR.

We demonstrate in this study that plasma ALT and Cr have a direct and linear relationship: mice with higher ALT showed increased renal dysfunction after liver IR. However, we also noted that the degree of renal dysfunction was mild when the plasma ALT levels were below 10 000 U/l. Behrends et al8 recently studied the renal dysfunction after extensive liver IR injury in rats. In their study, rats subjected to 75 min of partial liver IR developed acute liver dysfunction at 24 h. However, the degree of renal dysfunction in their rat model was mild with modest increases in plasma creatinine. They also noted upregulation of ICAM-1 mRNA expression over control animals but no changes in renal MCP-1, MIP2 and iNOS expression were noted. Correspondingly, the kidneys from rats subjected to liver IR showed little morphological damage and they failed to see increases in apoptosis (caspase-3) and neutrophil infiltration (myeloperoxidase activity). Their study confirms previous observations in rats that renal dysfunction was mild with little or no histological changes after liver IR.9, 41 Polat et al9 also reported that 45 min of liver ischemia resulted in mild renal dysfunction without detectable changes in renal morphology. The reason for this discrepancy between our mouse model and the previous rat model is unclear. However, in the study by Behrends et al,8 plasma ALT rose only to ∼6800 U/l whereas in our mouse model, the plasma ALT rose much higher (∼15–20 000 U/l). Longer liver ischemic period (eg, 90 min) in a rat is usually lethal (unpublished observations). Therefore, it appears that the murine model is better suited as model of liver failure-induced AKI than the rat model.

Some investigators utilized CCl4-induced cirrhosis as a model of liver failure-induced renal injury.7 Our model has several advantages over this model as our model is more clinically relevant, does not require administration of a tubular toxin and results in rapid and severe AKI. The CCl4 injection model is limited by the chronic nature of the protocol requiring several weeks (whereas human AKI can develop in less than a week). Moreover, full-blown renal dysfunction does not usually occur until day 3 with the CCl4 model whereas we can produce liver failure-induced AKI in less than 24 h.

To our knowledge, this is the first study utilizing mice to develop a model of AKI after acute liver dysfunction. Our model of AKI after liver IR closely mimics human AKI associated with ALF in terms of severity and the acute nature of renal dysfunction as well as similarities in histological changes. Utilization of mice has further advantages in that genetically modified mice are available to complement pharmacological approaches in studying the signaling mechanisms as well as examining potential therapies for AKI associated with ALF.

Serum creatinine is the most commonly used clinical indicator of glomerular filtration rate but its reliability as a measurement of glomerular filtration rate may be limited by several factors after acute liver injury. One of the main functional consequences of acute liver injury is a dramatic reduction in serum protein level, which influences the intravascular volume and glomerular filtration rate. Acute liver injury actually can lead to low serum creatinine due to decreased hepatic conversion of creatine to creatinine, decreased dietary protein intake and increased renal tubular secretion of creatinine. The increased fraction of creatinine excretion by tubular secretion in the setting low GFR also diminishes the reliability of creatinine as a marker of renal dysfunction. Consequently, the plasma creatinine values reported in this study in mice (0.91±0.15 mg/100 ml) may actually underestimate the degree of AKI associated with liver IR. Our measurements of plasma BUN correlated less well with renal function, probably because of interference with the urea cycle and hepatic metabolism after liver IR.

In summary, we have developed a novel murine model of AKI after liver IR, which closely approximates the human disease, characterized by renal tubular necrosis, endothelial cell apoptosis, PMN infiltration and JGA hyperplasia. Our model should provide a useful experimental context to delineate the mechanisms of AKI after liver IR and evaluate potential pharmacological therapies. Our study is the first to characterize the histological changes in detail and to demonstrate marked endothelial apoptosis and F-actin destruction in AKI associated with acute liver injury. It remains to be determined in future studies whether stabilization of endothelial cells would protect against AKI after liver IR injury.