L-selectin and ICAM-1 mediate reperfusion injury and neutrophil adhesion in the warm ischemic mouse liver

Surinder S. Yadav, David N. Howell, Wenshi Gao, Douglas A. Steeber, Robert C. Harland, Pierre-Alain Clavien


Leukocytes recruited during ischemia-reperfusion to the liver are important mediators of injury. However, the mechanisms of leukocyte adhesion and the role of adhesion receptors in hepatic vasculature remain elusive. L-selectin may critically contribute to injury, priming adhesion for later action of intercellular adhesion molecule-1 (ICAM-1). Paired experiments were performed using mutant mice (L-selectin −/−, ICAM-1 −/−, and L-selectin/ICAM-1 −/−) and wild-type mice (C57BL/6) to investigate leukocyte adhesion in the ischemic liver. Leukocyte adhesion and infiltration were assessed histologically. Aspartate aminotransferase levels were significantly reduced (2- to 3-fold) in mutant vs. wild-type mice in most groups but most significantly after 90 min of partial hepatic ischemia. Leukocyte adhesion was significantly reduced in all mutant mice. Areas of microcirculatory failure, visualized by intravital microscopy, were prevalent in wild-type but virtually absent in L-selectin-deficient mice. After total hepatic ischemia for 75 or 90 min, survival was better in mutant L-selectin and L-selectin/ICAM-1 mice vs. wild-type mice and ICAM-1 mutants. In conclusion, L-selectin is critical in the pathogenesis of hepatic ischemia-reperfusion injury. Poor sinusoidal perfusion due to leukocyte adhesion and clot formation is a factor of injury and appears to involve L-selectin and ICAM-1 receptors.

  • gene-targeted deficient mice
  • hepatic ischemia-reperfusion injury
  • survival
  • no reflow

ischemia-reperfusion (I/R) injury occurs when blood flow to an organ or tissue is interrupted for some period and subsequently reestablished. The sequence of events that leads to tissue injury in this situation is incompletely understood but involves adhesion and infiltration of polymorphonuclear leukocytes (PMN) as an early step (1, 9, 22, 34, 35, 41, 47). Subsequently, the PMN may elicit tissue damage by diverse processes ranging from direct elaboration of cytotoxic mediators (17, 44) to mechanical obstruction of blood flow (10). In extreme cases, thrombosis of vascular channels may totally abrogate perfusion of the affected tissue, a condition that has been termed the “no-reflow” phenomenon (25).

Adhesion of PMN is enabled by a multistep process mediated by a variety of cell surface molecules. The selectin family of adhesion receptors (P-, E-, and L-selectin) supports rapid capture of leukocytes and subsequent rolling along the vascular endothelium before firm adhesion and migration. L-selectin, a receptor expressed constitutively on quiescent PMN, is involved in the very initial process of rolling. L-selectin has been shown to solely mediate cell arrest under conditions of shear flow (14). Once the slowly rolling PMN becomes activated, β2-integrin receptors are upregulated, and L-selectin is shed from the PMN membrane (23). Firm adhesion then occurs through interaction with intercellular adhesion molecule-1 (ICAM-1), a major counterreceptor for β2-integrins (39). ICAM-1, a member of the immunoglobulin superfamily, is constitutively present at low levels on most endothelial cells (3) and is upregulated maximally over a 8- to 24-h period at sites of inflammation. Thus L-selectin and its counterendothelial receptor act in an integrated fashion with the β2-integrin/ICAM-1 complex, a pathway thought to be an essential step for flattening and migration of PMN into the extracellular matrix (15).

In I/R injury, PMN adhesion is accompanied and followed by a complex sequence of hemodynamic events. Pressures and flow may vary dramatically once reperfusion occurs in the injured liver. Alterations in the hemodynamics of the hepatic microvasculature are also influenced by PMN adhesion to the sinusoidal lining. L-selectin receptors play a role in propagating adhesion and PMN plugging in the microvasculature through interaction between previously adherent and flowing PMN (27). Platelets and other blood substances may also adhere to stationary PMN, producing partial or complete plugging that leads to a reduction or cessation of blood flow.

The liver is particularly susceptible to I/R injury, which is evident after conditions such as shock, trauma, transplantation, and surgical hepatectomy. As in other organs, PMN contribute significantly to liver injury as shown in experimental models of inflammation (24, 28, 48) and some models of I/R injury (21, 45). The liver provides an ideal environment for upregulation of adhesion receptors with an enormous capillary surface, the sinusoidal endothelium. ICAM-1 expression has been shown to be constitutively present in normal human livers on the sinusoidal lining cells, including endothelial and Kupffer cells, with strong inducibility under appropriate stimuli such as acute rejection (43). The liver also contains the largest population of macrophages in the body, providing a large source of mediators, including reactive oxygen intermediates (2, 7, 19), eicosanoids (40), and acute reactant cytokines (6, 20), able to readily induce expression of numerous adhesion molecules.

The role of ICAM-1 in hepatic I/R injury has been examined in several studies, most of which employed monoclonal antibodies against ICAM-1 as a means of blocking PMN adhesion (11, 12, 26, 29). However, the contribution of L-selectin to I/R injury in the liver and its possible interactions with ICAM-1 in this process have not been explored in detail. A powerful tool for analyzing these complex events is provided by the recent development of knockout mice lacking one or both of these adhesion molecules (1, 42).

Our objective was to assess the independent role as well as the additive effects of deficits for L-selectin and ICAM-1 receptors in the reperfused ischemic liver. We initially developed models of hepatic ischemia in the mouse to evaluate neutrophil-mediated injury. Because the degree of PMN adhesion and reperfusion injury is strongly dependent on the duration of ischemia, various periods of ischemia were studied. The degree of PMN infiltration and microvascular failure was evaluated by histological examination of tissue samples and by intravital microscopy. Liver injury was quantified by release of transaminases, tissue necrosis, and animal survival.



Gene-targeted mice deficient in L-selectin were generated as previously described (1) and backcrossed with C57BL/6 mice for at least seven generations. This mutation resulted in mice completely deficient in L-selectin cell surface expression but not in the expression of other selectins (P- or E-selectin). ICAM-1-deficient mice were generated as described elsewhere (42). Mice deficient in the expression of both L-selectin and ICAM-1 were generated by crossing the single-deficient animals. There were no obvious indications of pathology or disease susceptibility for any of the mice up to 1 yr of age. The absence of both L-selectin and ICAM-1 led to elevated numbers of circulating neutrophils (580% of wild type, P < 0.001). Wild-type mice (C57BL/6; Jackson Laboratories, Bar Harbor, ME) were used as control animals. All animals were maintained on 12:12-h dark-light cycles and were given food and water ad libitum. All procedures were performed according to the Duke University Institutional Animal Care and Use Committee guidelines.

Model of Partial Hepatic Ischemia

Mice were anesthetized by inhalation of metofane. A midline incision was made extending from the xiphisternum to the pubis. The liver was exposed completely with retractors placed in the flanks and a clamp was attached to the xiphisternum, which was elevated. The ligamentous attachments of the lateral left lobe were carefully divided, and the lobe was mobilized and freed. The blood supply to this lobe was interrupted by applying a microclamp (FD562; Aesculap, South San Francisco, CA) to the vascular pedicle. The ischemic lobe corresponded to 39.0 ± 1.0% (mean ± SE,n = 5) of the total hepatic mass. Using an intravenous dye (trypan blue), we confirmed that the lateral left lobe was completely ischemic, whereas the other lobes were homogeneously perfused.

Model of Total Hepatic Ischemia

To evaluate animal survival, we developed a new and simple model of total hepatic ischemia in the mouse (50). Mice were anesthetized with metofane. A midline incision was made, and bowel loops were exteriorized, placed on saline-soaked gauze, and covered with plastic wrap/clear film. The quadrate and the papillary process of the liver were ligated with 6.0 silk suture and resected followed by the lateral right and the caudate lobes. The amount of tissue resected corresponded to 29.2 ± 1.2% (mean ± SE,n = 4) of the total liver mass (Fig.1). Total ischemia of the remaining liver tissue, consisting of the median and lateral left lobes, was achieved with the use of vascular microclamps. In preliminary studies, total ischemia was confirmed by infusion of trypan blue into the portal vein. None of the dye was able to enter the liver. Congestion of the gut was completely absent in this model due to collateral vessels providing a shunt from the portal vein to the inferior vena cava. Survival was assumed to be permanent if animals were alive 30 days after surgery.

Fig. 1.

New model of total hepatic ischemia in the mouse showing the resected parts of the liver after ligation of the respective pedicles. Resected tissue represented <30% of the total hepatic mass. Magnification: ×5.

Serum Markers of Reperfusion Injury

Serum levels of aspartate aminotransferase (AST) and alanine aminotransferase (ALT) were used as established markers of I/R injury (18). Blood samples were obtained from the inferior vena cava using a 24-gauge catheter via a midline incision. The samples were stored in serum separator tubes (Becton-Dickinson, Franklin Lakes, NJ) and immediately centrifuged at 11,000 rpm for 5 min. The serum transaminases analyzed at each I/R time point were determined from individual samples (1 ml/mouse). A 10-μl sample of serum was diluted with 0.9% saline and was analyzed using the serum multiple biochemical analyzer (Ektachem DTSCII; Johnson & Johnson, Rochester, NY). In preliminary experiments with wild-type mice subjected to varying periods of partial ischemia (30, 60, 90, and 120 min), peak AST and ALT levels occurred after ∼6 h of reperfusion, and both transaminase levels plateaued after 24 h of reperfusion (data not shown). Blood samples were collected in separate experiments so as not to interfere with other studies.

Histological Assessment of Liver Tissue Injury After Reperfusion

Ischemic and nonischemic liver tissue from each animal were fixed in buffered Formalin and processed to paraffin. Sections were stained with hematoxylin and eosin and examined by a pathologist (D. N. Howell) unaware of the duration of warm ischemia and genotype of the mouse. The degree of vascular congestion/thrombosis and hepatocyte death was used as a marker of tissue injury. Vascular thrombosis/congestion was defined as engorgement of portal venules, sinusoids, or terminal hepatic venules with erythrocytes, platelets, and/or fibrin material. Hepatocyte death was assessed by loss of nuclear detail and well-defined cellular borders. A semiquantitative scale from zero to four according to the percent of tissue involved was developed for each respective hepatic zone (zone 1: periportal; zone 2: intermediate; zone 3: around terminal hepatic venules). The scale was defined as grade 0 (<1%), grade 1(1–10%), grade 2 (11–25%),grade 3 (26–50%), andgrade 4 (>50%). For the assessment of PMN adhesion and infiltration to the hepatic vasculature, a semiquantitative scale for each hepatic zone from zero to four was developed (Table 1). The vast majority (>95%) of inflammatory cells identified in all cases were PMN.

View this table:
Table 1.

Grading scale for PMN adhesion and infiltration in zones 1 and 3 (periportal and pericentral) and zone 2 (intermediate) in liver tissue stained with H&E

Intravital Microscopy

Animals were anesthetized with metofane, and hepatic ischemia was achieved using a microvascular clamp placed on the vascular pedicle of the lateral left lobe. The ischemic liver lobe was placed on a table and fixed to the surface with glue (Krazy glue). The animal was kept anesthetized and placed on an adjustable Plexiglas stage. The liver surface was kept continuously moist with warm saline. Observations were made using an inverted intravital microscope (K2SBio/Optiphot) with a ×20 objective lens (Olympus), and, for epi-illumination, a 100-W mercury lamp (Nikon) was used. Images were recorded using a silicone intensified camera (Dage-MTI, Michigan City, IN) and were recorded on videotape using a video recorder (JVC) for evaluation of dynamic images. Observations were made for 30 min after removal of the microvascular clamp.

The effect of ischemia on microcirculatory integrity was assessed by examination of 10 random epi-illuminated microscopic fields in each animal. Sinusoidal areas with dark, irregular outline were considered to be early indicators of microcirculatory breakdown, as described previously (31). Data are expressed as number of failing areas/microscopic field.

Statistical Analysis

Data are expressed as means ± SE. Paired Student’st-test was performed, and ANOVA was performed using the Statistical Analysis System to compare serum transaminase levels and the Mann-Whitney test to compare histological data. The Fischer’s exact test was performed to compare survival. AP value of <0.05 was considered statistically significant.


Effects of Hepatic Ischemia on Animal Survival

Assessment of survival after prolonged periods of hepatic ischemia requires a technique of total hepatic ischemia. However, complete occlusion of the liver is rapidly lethal due to massive mesenteric congestion unless a shunt is created between the portal and the systemic circulations. Because these shunts are difficult to perform in small animals and have not been reported in the mouse, we used our new technique of total hepatic ischemia in the mouse (50).

All wild-type animals survived >30 days after being subjected to 60 min of total hepatic ischemia while all died within 24 h of surgery after 90 min of total ischemia (n = 7 in each group, Table2). An intermediate time period of 75 min of ischemia was associated with partial survival (2 out of 7). Survival was significantly improved in L-selectin-deficient mice (L-selectin −/−) and the combined L-selectin and ICAM-1-deficient mice (L-selectin/ICAM-1 −/−); after 75 min of ischemia, all of these groups survived permanently (Table2). Improved survival was also observed after 90 min of warm ischemia in the same deficient mice, although it did not reach significance. Survival in the ICAM-1 −/− mice was comparable to wild-type animals at all ischemia times studied.

View this table:
Table 2.

Animal survival after varying periods of total hepatic ischemia

L-Selectin and ICAM-1 in Hepatic Reperfusion Injury

Serum AST and ALT levels were measured as markers of hepatic injury. The baseline serum AST values in untreated mice were as follows: L-selectin −/− = 147 ± 13 IU/l; L-selectin/ICAM-1 −/− = 141 ± 25 IU/l; ICAM-1 −/− = 157 ± 12 IU/l (n = 3 for each deficient mouse); and wild-type mice = 135 ± 13 IU/l. Sham experiments were performed to assess the effect of surgery on serum AST levels. Levels at 6 and 24 h after surgery were 188 ± 26 IU/l, which remained within baseline normal levels (baseline AST levels = 59–247 IU/l). Animals were studied after varying periods of partial hepatic ischemia (30, 60, and 90 min) and at different times of reperfusion (15 min, 1 h, 6 h, and 24 h). A model of partial ischemia was developed to assess ischemic injury and PMN adhesion to the liver without influencing animal survival. In each paired study, deficient and wild-type animals (n = 5 in each respective group) were subjected to one specific ischemia and reperfusion time (e.g., 60 min of ischemia/6 h of reperfusion). Because serum AST and ALT levels yielded comparable results, only AST levels, a sensitive marker of hepatic I/R injury (18), are shown (Table3).

View this table:
Table 3.

Transaminase levels in wild-type mice and deficient mice

Wild-type mice. In each ischemia group, transaminase levels increased up to 6 h after reperfusion and then declined (Table 3). In animals subjected to 30 or 60 min of ischemia, transaminase levels were comparable at 6 h of reperfusion but were significantly higher after 24 h of reperfusion in the 60-min group (P < 0.005, unpaired Student’s t-test). The greatest increase in transaminase levels was observed after 90 min of ischemia followed by 6 h of reperfusion. In this ischemia group, transaminase levels remained well above baseline after 24 h of reperfusion.

L-selectin −/−. At most ischemia times, transaminase levels were lower in L-selectin −/− mice compared with wild-type mice (Table 3). The most striking reduction was seen in L-selectin −/− mice subjected to 90 min of ischemia. In this group, AST levels were two- to threefold lower in the L-selectin −/− mice after 15 min and 6 h of reperfusion compared with wild-type mice, whereas no difference was observed at 24 h. Of note, AST levels were also significantly lower in L-selectin −/− mice after 15 min of reperfusion compared with ICAM-1 or even L-selectin/ICAM-1-deficient mice.

ICAM-1 −/−. Serum transaminase levels were significantly lower than paired wild-type controls after 6 h of reperfusion in most ischemia groups but not at earlier or later stages of reperfusion (Table 3). The most significant differences were observed in livers subjected to 90 min of ischemia and reperfused for 6 h. In fact, ICAM-1 −/− mice had the lowest levels of all animals in this cohort.

L-selectin/ICAM-1 −/−. Serum transaminase levels were significantly lower at all I/R time points compared with wild-type mice (Table 3). Although levels were usually comparable to those found in the L-selectin −/− mice, they were significantly lower (2- to 3-fold) in L-selectin/ICAM-1 −/− livers subjected to 90 min of ischemia at 1 and 24 h of reperfusion (P < 0.05).

L-Selectin and ICAM-1 in Hepatic Tissue Injury

The extent of PMN adhesion, infiltration, and tissue damage in ischemic livers was assessed using a semiquantitative grading scale (Table 1). The nonischemic lobes in all animals that underwent the surgery were histologically unremarkable and showed no detectable signs of neutrophil adhesion, infiltration, or congestion and necrosis. In addition, sham-operated animals showed no abnormal histology. Comparisons were made between wild-type and L-selectin/ICAM-1 −/− mice at various times (Table4). In livers rendered ischemic for 30 or 60 min, minimal histological alterations were observed at any time after reperfusion in either the wild-type groups or the L-selectin/ICAM-1 −/− animals, a finding consistent with the survival studies in which all types of animals survived 60 min of ischemia.

View this table:
Table 4.

Histological analysis of L-selectin/ICAM-1 −/− and wild-type animals subjected to 90 min of warm ischemia followed by 1 and 6 h of reperfusion

In livers subjected to 90 min of ischemia followed by 1 h of reperfusion, there was a significant difference in infiltration of PMN between the L-selectin/ICAM-1 −/− and wild-type mice. This difference was most striking inzones 2 and3. In the wild-type mice, numerous PMN were seen adherent to the endothelium of terminal hepatic venules and within adjacent sinusoids, whereas PMN adhesion was minimal in the L-selectin/ICAM-1 −/− group (Table 4 and Fig.2). A similar difference in PMN adhesion was seen in the L-selectin −/− (Fig. 2) and ICAM-1 −/− (data not shown) animals after the same period of ischemia and reperfusion. After 6 h of reperfusion in the same ischemia group (90 min ischemia), there was variable, often extensive infiltration of PMN in all three hepatic zones, and no differences were detected between the mutant and wild-type groups (Table 4).

Fig. 2.

Hematoxylin and eosin (H&E) sections of livers subjected to 90 min of ischemia and reperfused for 1 h from paired experimental and control groups. Numerous neutrophils adherent to the endothelium and infiltrating the tissue are visible in the wild-type mouse (A) and were markedly reduced in the L-selectin/intercellular adhesion molecule-1 (ICAM-1) −/− (B) and L-selectin −/− (C) livers. Magnification: ×315.

We also looked at the degree of tissue congestion after prolonged ischemia as a factor of tissue injury and vascular plugging. Although there were no significant differences (P < 0.08), there was a trend toward reduced sinusoidal congestion in the L-selectin/ICAM-1 −/− mice subjected to 90 min of ischemia compared with the wild-type mice (Table 4 and Fig. 3).

Fig. 3.

H&E sections from an ischemic lobe in a wild-type mouse (A) and an ischemic lobe in an L-selectin/ICAM-1 mutant mouse (B), both subjected to 90 min of partial ischemia and reperfused for 1 h. Degree of vascular thrombosis/congestion was greater in the wild-type mouse. Magnification: ×170.

Extensive histological tissue/hepatocyte necrosis, evidenced by nuclear degeneration and loss of distinct cellular borders, was present in animals subjected to 90 min of ischemia followed by 6 h of reperfusion. In this ischemia group, hepatocyte necrosis inzone 1 (periportal) was significantly less in the L-selectin/ICAM-1 −/− mice than in the wild-type group (Table 4), whereas, in zones 2 (intermediate) and 3(around terminal hepatic venules), no significant differences were found between the L-selectin/ICAM-1 −/− and wild-type mice. Evidence of tissue necrosis was not detectable at any earlier time points of ischemia.

L-Selectin and ICAM-1 in Microvascular Injury Leading to Sinusoidal Perfusion Failure

The macroscopic and microscopic examination of control vs. L-selectin/ICAM-1 −/− livers subjected to prolonged periods of ischemia differed significantly for evidence of obstructed vasculature. Macroscopically diffuse dark patches were seen on the surface of ischemic livers in wild-type mice subjected to 90 min of ischemia followed by 6 h of reperfusion, whereas they were absent in the L-selectin/ICAM-1 −/− mice (data not shown). Histological analysis as described earlier indicated a trend toward reduced congestion although no statistical significance (P < 0.08) was achieved in the L-selectin/ICAM-1 −/− mice compared with wild-type mice subjected to the same conditions.

A method of in vivo intravital microscopy was developed to further evaluate sinusoidal areas with extensive microvascular breakdown. Dark irregular areas correspond to petechial bleeding and perfusion failure as established elsewhere (31; as shown in Fig.4). Animals were subjected to 60 and 90 min of ischemia, and analysis was performed after 30 min of reperfusion. In wild-type mice, ischemia of 60 min produced a significant increase in the number of sinusoidal areas with microcirculatory failure compared with the L-selectin −/− mice and L-selectin/ICAM-1 −/− mice (Fig.5). Hepatic ischemia of 90 min produced a significant increase in the number of nonperfusing areas with persistent difference between wild-type mice and L-selectin −/− livers (Fig. 5).

Fig. 4.

Petechial bleeding as an early predictor of microcirculatory failure was markedly evident in wild-type animals. Magnification: ×20.

Fig. 5.

Intravital microscopy was used to detect sinusoidal areas with microcirculatory failure in livers subjected to 60 min (A) and livers subjected to 90 min (B) of partial hepatic ischemia. Increased numbers of areas were present in the wild-type animals compared with the mutant mice (n = 20 random fields; * P < 0.05, ANOVA and unpaired Student’s t-test).

In a group of animals subjected to 120 min of ischemia with 6 h of reperfusion, serum transaminase levels were paradoxically higher in the L-selectin/ICAM-1 −/− mice (26,268 ± 769 IU/l) compared with the wild-type mice (18,077 ± 684 IU/l;P < 0.05, paired Student’st-test) and ICAM-1 −/− mice (data not shown). In two out of seven wild-type mice subjected to these extreme ischemia times (90 and 120 min), there was evidence of extensive clot formation and vascular occlusion (Fig.6), whereas this was not seen in the L-selectin/ICAM −/− mice. The lower transaminase levels in the controls may reflect the poor release of enzymes into the circulation due to occluded vessels.

Fig. 6.

H&E sections from ischemic lobes in a wild-type (A) and L-selectin/ICAM-1 −/− (B) mouse subjected to 120 min of partial hepatic ischemia and reperfused for 6 h. Engorgement and total occlusion of the central veins are present in the wild-type liver but not the L-selectin/ICAM-1 −/− liver. Magnification: ×68.


In this study using mutant mice, we provide direct evidence indicating a pivotal role for L-selectin and, to a lesser degree, ICAM-1 in the mechanisms of hepatic I/R injury. This study also introduces novel models of partial and total hepatic ischemia specifically designed to study varying periods of hepatic ischemia and reperfusion in the mouse. In wild-type mice, manifestations of I/R injury included early infiltration of PMN and extensive tissue injury as assessed by histologic examination, measurement of serum transaminase levels, and microcirculatory collapse visualized by intravital microscopy. In mice deficient in L-selectin alone, a wide range of protective effects on I/R injury was observed, including decreases in tissue PMN infiltration, prevention of microcirculatory failure, lower transaminase levels at a wide range of time points, and better animal survival after total hepatic ischemia. The latter was particularly impressive after 75 and 90 min of ischemia; 90 min of ischemia was a duration associated with 100% death in the wild-type animals. Protective effects of ICAM-1 deficiency, although significant, were limited to decreases in PMN infiltration and decreases in transaminase levels at a single time point of reperfusion (6 h). Survival in ICAM-1 −/− mice after total hepatic ischemia was similar to that of wild-type animals. L-selectin/ICAM-1 −/− mice, with a small number of exceptions, showed a pattern of response to I/R injury that was similar to L-selectin −/− mice. Taken as a whole, these experiments support a pathogenetic sequence in which PMN adhesion promotes hepatocellular injury in a complex manner, with one final pathway of injury involving a decrease or cessation in blood flow.

Our data demonstrate a primary role for L-selectin in mediating injury in the reperfused warm ischemic liver. Loss of the L-selectin receptor caused reduction in PMN accumulation early in reperfusion. L-selectin receptors were also essential in mediating hepatocyte injury, as documented by significant reduction in transaminase levels in most ischemia groups for the deficient mice. These beneficial effects were ultimately reflected in improved survival for the L-selectin-deficient mice compared with wild-type animals.

The beneficial effects of blocking L-selectin have been observed in animal models of inflammatory injury such as the heart (5, 32), peritoneum (38), and lung (33). However, evidence supporting a role for L-selectin receptors in the pathogenesis of liver injury has been limited to a study on cold ischemia in the rat that employed blocking antibodies (16) and one other that used a nonspecific selectin blocker, also in the rat (37). In fact, studies in nonischemic models of inflammation in the liver have suggested little or no role for selectins in mediating neutrophil interaction with the endothelium (48). The cause of this discrepancy is unclear; it may reflect differences in PMN recruitment in response to different proinflammatory stimuli.

In contrast to L-selectin, ICAM-1 was less crucial in mediating tissue damage in our model. Deficiency in ICAM-1 alone significantly reduced injury only after 6 h of reperfusion and had no effects on early (1 h) or late (24 h) reperfusion injury. This time course is consistent with the known pattern of ICAM-1 upregulation after inflammatory stimuli in which peak expression occurs 6–8 h after stimulation (8). The absence of ICAM-1 also had no effect on animal survival after prolonged ischemia.

In the livers of mice lacking ICAM-1 receptors, after 90 min of ischemia and 1 h of reperfusion, there was a significant reduction in infiltrating PMN, similar to that seen in L-selectin −/− mice. At this time point, however, ICAM-1 −/− mice showed a rise in serum transaminase levels comparable to wild-type controls, whereas L-selectin −/− mice had significantly lower transaminase levels. This disparity suggests that the early rise in transaminase levels seen in ICAM-1 −/− mice (and, by extension, wild-type mice) may involve factors not directly related to PMN infiltration or alternatively may be mediated by small numbers of infiltrating PMN below the threshold of histological detection. Discrepancies between PMN infiltration and tissue damage have been noted in a small number of studies in I/R (4,12, 49). By the same token, the reduced transaminase levels seen in L-selectin −/− mice may not be a simple result of decreased PMN infiltration. Because L-selectin, unlike ICAM-1, is expressed on PMN themselves, it is possible that the protective effect of L-selectin deficiency involves aspects of PMN function other than or in addition to adhesion (e.g., release of cytokines, receptor signaling events). Indeed, recent studies have shown an increased tendency for PMN to undergo programmed cell death in mice that were deficient for L-selectin receptors (30).

Blocking the function of ICAM-1 adhesion receptors with monoclonal antibody has proven to be protective in the ischemic liver, with documented decreases in infiltration of PMN, oxygen radical formation, and transaminase levels (11, 12, 26, 29). These studies, although strongly suggestive of a role for ICAM-1 in I/R injury, are hampered by difficulties inherent in the monoclonal antibody blocking approach. The efficacy of blocking antibodies may be diminished when there is substantial injury to the sinusoidal lining (12, 36), and attenuation of injury may be unrelated to the blocking function of the antibody (46). In spite of these observations, our studies with ICAM-1 −/− mice, which circumvent the complexity of monoclonal antibody blocking experiments, support and extend the findings of the previous studies. However, from a therapeutic point of view, our data suggest that blocking ICAM-1 alone is unlikely to provide significant beneficial effects in the ischemic liver.

Combined deficiencies for L-selectin and ICAM-1 receptors also markedly diminished many of the detrimental effects of hepatic I/R injury. In many respects, including survival, histologic manifestations of tissue damage, and PMN infiltration, these animals resembled L-selectin −/− mice. Unlike either L-selectin −/− or ICAM-1 −/− mice, the L-selectin/ICAM-1 −/− mice enjoyed a statistically significant reduction in transaminase levels after 24 h of reperfusion, suggesting a possible additive protective effect. However, the discrepancy in the site of early PMN margination (zone 3) and the eventual pattern of tissue injury/protection seen in these mice suggest that the PMN may be mediating damage by some mechanism other than direct cytotoxicity. One possible mechanism is an impedance of hepatic blood flow potentiated by the early margination of the PMN, leading to subsequent exacerbation (through poor perfusion) of ischemic damage.

In all animals, both wild-type and mutant mice, there was extensive diffuse PMN infiltration and adhesion after 6 h of reperfusion; the abrogation of differences between groups at this time point reflects alternative pathways of PMN recruitment. These data suggest a complex role for PMN-mediated injury. It seems that the prevention of early interaction between PMN, adhesion receptors, and the endothelium is essential for an impact on persistent attenuation of injury. Our results indicate that L-selectin and ICAM-1 receptor-mediated leukocyte adhesion is one critical factor in a complex pathway for ischemic injury to the liver. However, it also seems likely that the impact of the loss of these receptors is not solely mediated through leukocyte adhesion.

Our studies did not examine the possible contribution of adhesion molecules other than L-selectin and ICAM-1 in hepatic I/R injury. Expression of E- and P-selectin is reportedly lacking in the sinusoidal endothelium of human livers, even after severe inflammatory injuries such as chronic rejection and sepsis (43). Other endothelial surfaces in the liver, however, including those of hepatic arteries and portal and central veins, express these selectins in both a basal and inducible fashion (43). We are currently investigating a possible role for P-selectin in hepatic I/R injury using gene-targeted deficient mice.

Another important finding in our studies was the identification of a critical ischemia time resulting in obstruction of the hepatic sinusoids by blood elements, a phenomenon mediated by L-selectin and ICAM-1 receptors. The involvement of adhesion molecules in mediating microcirculatory disturbances due to leukostasis has remained controversial. Some authors have proposed that the development of microcirculatory occlusion is solely dependent on factors such as endothelial swelling, protrusion of blebs, and hemoconcentration (31), whereas others have suggested that PMN plugging is an important event in the occlusion of microvasculature after reperfusion (13, 25). Koo et al. (25) have described this phenomenon of no-reflow to occur as early as after 30 min of hepatic ischemia.

In our studies, gross examination of livers in wild-type mice subjected to prolonged ischemia (90 and 120 min) showed the presence of widespread areas of patchy mottling, a finding not observed in L-selectin −/− or L-selectin/ICAM-1 −/− mice (data not shown). Using intravital microscopy, this phenomenon of microcirculatory failure was consistently observed after 60 min of hepatic ischemia as areas of petechial bleeding. Microcirculatory failure was almost absent in the L-selectin −/− and L-selectin/ICAM-1 −/− livers. Under extreme conditions, a few (2 out of 7) histological sections of control livers subjected to 90 and 120 min ischemia, when compared with paired experiments in L-selectin/ICAM-1 −/− mice, showed extensive clot formation and vascular plugging. The lower levels of transaminases observed in wild-type mice compared with mutant mice after 120 min ischemia and 6 h of reperfusion may represent poor perfusion of liver tissue with the inability to release enzymes in the circulation.

The breakdown of vascular integrity with subsequent occlusion (the no-reflow phenomenon) occurring in livers subjected to 60 min or more of ischemia may be critical for animal survival. Survival was improved significantly in both L-selectin −/− and L-selectin/ICAM-1 −/− mice after 75 min of ischemia. After 90 min of ischemia, survival was achieved in the L-selectin −/− and L-selectin/ICAM-1 −/− mice, whereas none of the wild-type mice survived. These findings may have important implications, not only regarding mechanisms of injury but also from a therapeutic perspective. Blocking either L-selectin alone or both L-selectin and ICAM-1 may provide significant benefits in liver function after prolonged periods of ischemia.

In summary, this study demonstrates that L-selectin mediates microcirculatory disturbances and hepatocyte injury in the ischemic mouse liver. ICAM-1, although involved independently to a lesser degree, is important in mediating injury and microcirculatory disturbances in conjunction with L-selectin. The demonstration that deficits in these adhesion receptors attenuate injury and prevent microcirculatory compromise in the liver provides opportunity for interventions to minimize hepatic injury and extend the permissible duration of ischemia. Currently, a variety of agents are available, such as antibodies, chimeric molecules, peptides or glycan moieties, and other pharmacophores, to act against selectin-mediated leukocyte adhesion. These findings open new avenues for therapeutic interventions and provide a rationale to study the exact timing of interventions on specific adhesion molecules.


We thank Dr. J. J. Lemasters and Robert T. Currin for help with the intravital studies and Dr. T. F. Tedder for invaluable discussion and provision of deficient mice.


  • Address for reprint requests: P.-A. Clavien, Dept. of Surgery, Duke Univ. Medical Center, PO Box 3247, Durham NC 27710.

  • A preliminary report of this work was presented at the American Society of Transplant Surgeons, Chicago, IL, May 1997.

  • The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.


View Abstract