Loss of integrin linked kinase from mouse hepatocytes in vitro and in vivo results in apoptosis and hepatitis


  • Potential conflict of interest: Nothing to report.


Extracellular matrix (ECM) is fundamental for the survival of cells within a tissue. Loss of contact with the surrounding ECM often causes altered cell differentiation or cell death. Hepatocytes cultured without matrix lose patterns of hepatocyte-specific gene expression and characteristic cellular micro-architecture. However, differentiation is restored after the addition of hydrated matrix preparations to dedifferentiated hepatocytes. Integrin-linked kinase (ILK) is an important component of cell–ECM adhesions transmitting integrin signaling to the interior of the cell. ILK has been implicated in many fundamental cellular processes such as differentiation, proliferation, and survival. In this study, we investigated the role of ILK in mouse hepatocytes in vitro as well as in vivo. Depletion of ILK from primary mouse hepatocytes resulted in enhanced apoptosis. This was accompanied by increased caspase 3 activity and a significant decrease in expression of PINCH and α-parvin, which, along with ILK, form a stable well-characterized ternary complex at cell–ECM adhesions. The induction of apoptosis caused by ILK depletion could be substantially reversed by simultaneous overexpression of ILK, indicating that apoptosis is indeed a consequence of ILK removal. These results were further corroborated via in vivo data showing that adenoviral delivery of Cre-recombinase in ILK-floxed animals by tail vein injection resulted in acute hepatitis, with a variety of pathological findings including inflammation, fatty change, and apoptosis, abnormal mitoses, hydropic degeneration, and necrosis. Conclusion: Our results demonstrate the importance of ILK and integrin signaling for the survival of hepatocytes and the maintenance of normal liver function. (HEPATOLOGY 2007.)

Extracellular matrix (ECM) is of great importance for the survival, differentiation, and normal function of cells within a tissue. This is particularly true for hepatocytes, the parenchymal cells of the liver.1–4 In fact, hepatocytes in culture maintained in the absence of matrix rapidly lose patterns of hepatocyte-specific gene expression and characteristic cellular micro-architecture. Interestingly, however, when hydrated complex matrix preparations [Matrigel: a matrix extract from Engelbreth-Holm-Swarm mouse sarcomas or Type I Collagen Gels]5 are added over de-differentiated hepatocytes, differentiation is restored within 3 days.1, 6 Moreover, ECM remodeling is an essential part of liver regeneration after partial hepatectomy,7–9 again highlighting the importance of ECM for the normal function of hepatocytes in liver.

Signals from the ECM are transmitted to the interior of the cell via integrins. Integrins not only interact with components of the ECM and physically link them to the cytoskeleton, but they also associate with multiple receptor-proximal proteins, which in turn serve as scaffolds for the attachment of enzymes that modify and regulate these complex interactions.10

Integrin-Linked Kinase (ILK) is a Ser/Thr kinase that is emerging as a key regulator of cell–ECM adhesions. Activation of ILK, either by integrin clustering or by growth factors, affects multiple cell signaling pathways that regulate different processes such as survival, differentiation, proliferation, migration, and angiogenesis.11–14 Additionally, there is evidence that ILK also plays a role in oncogenic transformation.15–17 From the signaling perspective, ILK activation results in the phosphorylation and subsequent activation of protein kinase B (PKB/Akt)18–20 and mitogen-activated protein kinase,20 as well as inhibition of glycogen synthase kinase 3 beta.21, 22 Through these pathways, ILK regulates survival, differentiation, cell cycle progression, and many other crucial processes.

As an adaptor protein, ILK mediates multiple protein–protein interactions.23, 24 Apart from interaction with the β1-and β3-subunits of integrin,15 ILK also binds to PINCH (Particularly Interesting Cys-His-rich protein)25, 26 and α-parvin, member of the CH-ILKBP/α-parvin/actopaxin/affixin protein family,27–30 resulting in the formation of a ternary complex within cells. This stable ternary complex of PINCH-ILK-parvin at the cell–ECM adhesion sites has been shown to be crucial for cell survival.26–28, 31

Although extensive research has been done on ILK signaling and functions in cells, little is known about the function of ILK in hepatocytes. Evidence suggests, however, that integrin signaling is important for hepatocyte survival.2, 32, 33

Our aim in this study was to elucidate the role of ILK in mouse hepatocytes both in vitro and in vivo. To pursue this, we used the loxP-Cre system to directly eliminate ILK from primary mouse hepatocytes in culture as well as from the whole animal.


β-gal, beta-galactosidase; ECM, extracellular matrix; HE, hematoxylin-eosin; ILK, integrin-linked kinase; mAb, monoclonal antibodies; pfu, plaque-forming units; PKB, protein kinase B, TUNEL, terminal deoxynucleotidyl transferase-mediated nick-end labeling.

Materials and Methods


The following primary antibodies were used in this study: mouse monoclonal (mAb) anti-ILK antibody (clone 69) (Santa Cruz Biotechnology, Santa Cruz, CA), mAb anti-β1-integrin (BD Biosciences, San Jose, CA), rabbit polyclonal anti-ILK antibody (Cell Signaling, Danvers, MA), rabbit polyclonal anti-PINCH1, mAb anti–α-parvin (clone3B5), mAb anti-Mig-2 antibody as described previously,28, 34 rabbit polyclonal anti-phospho-Akt (Ser 473), phosphor-MAPK, phosphor-JNK, and phosphor-p38 (Cell Signaling Technology, Danvers, MA), and anti–β-actin mAb (Chemicon, Temecula, CA). All secondary antibodies were from Jackson ImmunoResearch Laboratories (West Grove, PA).


The ILK-floxed animals that were used in the current study were generated as described previously.14 All animals were housed in the University of Pittsburgh animal facility, and they were treated in a humane manner according to the criteria outlined in the Guide for the Care and Use of Laboratory Animals prepared by the National Academy of Sciences and published by the National Institutes of Health. All experiments were performed under protocols approved by the Institutional Animal Care and Use Committee of the University of Pittsburgh.

Eight wild-type animals, purchased from the Jackson Laboratory (Bar Harbor, ME), were used to investigate the inherent toxicity of the injected Cre-expressing adenovirus. Because the ILK-floxed animals were on a mixed genetic background, four of the controls were of the 129X1/SvJ strain, and the remaining four were of the C57BL/6J strain.

Isolation of Hepatocytes from the Mouse Liver.

Mouse hepatocytes from the ILK-floxed animals were isolated by an adaptation of the calcium 2-step collagenase perfusion technique, as described previously.1 Hepatocytes were plated at a density of 300,000 cells per milliliter, in 6-well (35-mm) plates, and the medium was renewed 3 to 4 hours after plating. The medium used was minimum essential medium (Sigma, St. Louis, MO) supplemented with 10% fetal bovine serum, 1% glutamine, and 1% penicillin and streptomycin.

Adenoviral Vectors.

The recombinant adenoviruses used in this study were either control adenoviral expression vector encoding β-galactosidase (β-gal), Cre-recombinase–expressing adenovirus (Cre), or adenoviral expressing vector encoding FLAG-tagged full-length mouse ILK. The control β-gal adenovirus was provided by Drs. Tong-Chuan He and Bert Vogelstein (Howard Hughes Medical Institute, The Johns Hopkins Oncology Center, Baltimore, MD). The Cre-expressing adenovirus was provided by Dr. Jorge Sepulveda (Department of Pathology, University of Pittburgh School of Medicine). The FLAG-tagged full-length mouse ILK-expressing adenovirus was generated by cloning the cDNA encoding the FLAG-tagged full length mouse ILK into the SalI/XbaI sites of the pAdTrack-CMV shuttle vector and mixed with supercoiled pADEsay-1, as previously described.29

The amplified adenoviruses were subsequently purified by cesium chloride (CsCl) density gradient ultracentrifugation, dialyzed in sterile virus storage buffer, aliquoted, and stored at −80°C until use.35, 36 Viral titers were initially calculated by measuring the optical density of the viral DNA at 260 nm (OD260). Titer was considered equal to the optical density at 260 nm divided by 9.09 × 10−13 particles/ml. However, before using the adenovirus in our cultured cells, we also measured the viral titer by plaque formation assay (pfu/ml) performed according to the protocol provided by BD Biosciences.

Adenoviral Infections of ILK-Floxed Mouse Hepatocytes in Culture.

Twenty-four hours after liver perfusion (day 1), the cultured ILK-floxed mouse hepatocytes were infected with 107 pfu/well of either control β-gal or Cre adenovirus. The medium was changed the following day, and cells were kept in culture for a total of 7 days.

For the rescue experiments, hepatocytes were co-infected with either β-gal adenovirus and ILK adenovirus, or Cre adenovirus and ILK adenovirus, 1 day after liver isolation. The viral titer used for these experiments was 107 pfu/well for the β-gal and Cre adenoviruses, and 8 × 106 pfu/well for the ILK adenovirus.

Tail Vein Injections of Adenoviral Expressing Vectors in ILK-Floxed Animals.

PBS, 1010 particles/mouse of control β-gal adenovirus, Cre-expressing adenovirus, or vehicle only (phosphate-buffered saline) as an additional control, were administered into 78 ILK-floxed mice by tail vein injection. The same number of viral particles of Cre adenovirus was also injected by tail vein in eight wild-type animals to test for the inherent toxicity of the Cre-recombinase.

Immunofluorescent Staining.

Immunofluorescent staining was performed as described previously.25, 34 Briefly, cells were plated on collagen-coated coverslips, fixed with 4% paraformaldehyde, and double-stained with the anti-ILK rabbit polyclonal antibody and the anti-β1-integrin mAb. A rhodamine-red–conjugated goat anti-rabbit and an FITC-conjugated goat anti-mouse were used as secondary antibodies.

Histology Scoring.

Hematoxylin-eosin (H&E) staining was performed on both formalin-fixed paraffin-embedded sections and paraformaldehyde-fixed frozen sections of the livers from ILK-floxed mice injected with β-gal or Cre-adenovirus. The histology was evaluated in a blind fashion by the pathologist (G.K.M.) and scored according to the following characteristics: Score 0, normal liver; Score 1, inflammation or fatty change; Score 2, inflammation or fatty change, with very few apoptotic cells; Score 3, inflammation or fatty change, with apoptosis; Score 4, inflammation or fatty change, apoptosis, ballooning degeneration, and abnormal mitoses; Score 5, inflammation or fatty change, apoptosis, ballooning degeneration, abnormal mitoses, hydropic degeneration, and necrosis.

Polymerase Chain Reaction.

DNA was isolated from the ILK-floxed mouse hepatocytes using the Trizol reagent from Invitrogen (Carlsbad, CA). Subsequently, the DNA was tested by PCR using a set of primers designed to detect the deleted ILK gene; forward: 5′-CCA GGT GGC AGA GGT AAG TA-3′, reverse: 5′-CAA GGA ATA AGG TGA GCT TCA GAA-3′. The conditions of the PCR reaction were as follows: 94°C for 5 minutes, 94°C for 1 minute, 55°C for 1 minute, 72°C for 3 minutes (40 cycles for the last 3 steps), 72°C for 10 minutes, 55°C for 5 minutes, and hold at 4°C. The PCR products were separated through a 1% agarose gel at 80 V for 4 hours and visualized with ethidium bromide staining.

DNA was also isolated from livers of ILK-floxed mice injected with either PBS, 26β-gal adenovirus, or Cre-expressing adenovirus using again the Trizol reagent. Subsequently, the DNA was subject to PCR using the following primers for the amplification of the Cre gene; forward: 5′- GCG GTC TGG CAG TAA AAA CTA TC-3′, reverse: 5′- GTG AAA CAG CAT TGC TGT CAC TT -3′. The conditions of the PCR reaction were as follows: 94°C for 3 minutes, 94°C for 30 seconds, 68°C for 1 minute, 72°C for 1 minute (35 cycles for the last three steps), 72°C for 2 minutes, and hold at 10°C. The PCR products were separated using a 2% agarose gel at 80 V for 2 hours and visualized with ethidium bromide staining.

Protein Isolation for Immunoblotting.

Total protein was isolated from the ILK-floxed mouse hepatocytes or livers of the ILK-floxed mice injected with PBS, 26β-gal, or Cre-expressing adenovirus using 1% sodium dodecyl sulfate in RIPA buffer [20 mM Tris/Cl pH 7.5, 150 mM NaCl, 0.5% NP-40, 1% TX-100, 0.25% sodium deoxycholated, 0.6-2 μg/ml aprotinin, 10 μM leupeptin, 1 μM pepstatin].

Caspase-3 Activity Measurement.

Apoptosis was assessed by caspase-3 activity measurement using the fluorogenic caspase-3 substrate VII (Ac-DEVD-aminofluoromethylcoumarin) from Calbiochem (San Diego, CA) following the company's protocol.

Statistical Analyses.

For the rescue experiments (Fig. 4B), we used a t test to compare average standard error means. For the histology scoring (Fig. 8), we used the Mann-Whitney test. All statistical analyses were performed using the Statgraphics Plus statistics software and a P value of less than 0.05 was considered statistically significant.

Terminal Deoxynucleotidyl Transferase-Mediated Nick-End Labeling Assay.

Apoptosis in livers from adenovirus-injected animals was assessed by terminal deoxynucleotidyl transferase-mediated dUTP-biotin nick end-labeling (TUNEL) assay using the ApopTag In Situ Apoptosis Detection Kit from Chemicon (Temecula, CA).


ILK and β1-Integrin Are Localized in the Same Cell–ECM Adhesions of Primary Mouse Hepatocytes in Culture.

ILK has been shown to localize to the cell–ECM adhesions in a number of different cell types.25 To test whether the localization of ILK is the same in hepatocytes, we stained primary mouse hepatocytes with anti-ILK antibody. ILK is indeed present at the cell–ECM adhesion sites of primary mouse hepatocytes (Fig. 1A). Furthermore, double staining of mouse hepatocytes with anti-ILK and anti–β1-integrin antibodies showed that β1-integrin localizes in the same ILK-containing hepatocyte–ECM adhesion sites (Fig.1B).

Figure 1.

Localization of ILK to the cell-ECM adhesions of primary mouse hepatocytes and colocalization with β1-integrin. Primary mouse hepatocytes were double stained with anti-ILK rabbit polyclonal antibody (A) and mAb anti–β1-integrin. The white arrowheads show the sites of co-localization between ILK and β1-integrin at cell–ECM adhesion sites.

ILK Gene and Protein Are Removed by Infection of the ILK-Floxed Mouse Hepatocytes with Cre Adenovirus.

To test the significance of ILK in hepatocytes, we employed the loxP-Cre system approach. Specifically, primary mouse hepatocytes were isolated from ILK-floxed mice by liver perfusion (day 0) and 24 hours after plating (day 1), cells were subject to adenovirus infections to introduce either Cre-recombinase (Cre-adenovirus) or β-gal (β-gal adenovirus) as a control. DNA was isolated from the adenovirus-infected cells and subsequently tested by PCR using a set of primers designed to detect the excised ILK gene. As shown in Fig. 2A, the ILK gene was excised by the Cre-recombinase as early as 1 day after infection of the cells with Cre-adenovirus (day 2). The ILK protein, however, was not substantially depleted until day 4, as shown in Fig. 2B.

Figure 2.

Excision of the ILK gene and removal of the ILK protein from primary ILK-floxed mouse hepatocytes. (A) PCR products showing the excised ILK gene (arrow) in the cells infected with Cre-adenovirus 2, 3, and 4 days after infection (lanes 6-8). Note that the uninfected cells (lanes 2 and 3) and the β-gal–infected cells (lanes 4 and 5) do not have the excised ILK band. (B) Western blot showing that ILK protein is not removed from the uninfected cells (lanes 1, 2, 5, and 8), nor the β-gal–infected cells (lanes 4, 7, and 10), but it is removed from the Cre-infected cells 4 days after infection with Cre adenovirus (compare lane 9 with lanes 3 and 6). Actin is used as loading control.

ILK Is Crucial for Hepatocyte Survival.

The ILK-floxed mouse hepatocytes that were either not treated or infected with control β-gal adenovirus exhibited normal hepatocyte morphology for at least 5 days in culture. In contrast, the ILK-floxed hepatocytes that were infected with an equal number of plaque-forming units (pfu/ml) of the Cre-adenovirus showed abnormal morphology, and increasing numbers of dead cells were observed over time. The accumulation of dead cells in the Cre-infected ILK-floxed hepatocytes started on day 3 (2 days after infection with Cre-adenovirus), as shown in Fig. 3A.

Figure 3.

Cell morphology and caspase 3 activity in mouse hepatocytes after ILK knock-down. (A) Cell morphology of uninfected (parental), control-adenovirus (β-gal)-infected, or Cre-infected hepatocytes, 2, 3, 4, and 5 days after infection. Cells were observed under an Olympus IX70 fluorescence microscope, and images were captured with a digital camera. Note that the parental or β-gal–infected hepatocytes have normal morphology during days 2 to 5, whereas the Cre-infected hepatocytes show increasing numbers of dead cells over time. (B) Caspase 3 activity 2, 3, 4, and 5 days after infection. Cre-infected hepatocytes have significantly elevated caspase 3 activity compared with the parental and control infected cells. Compare the solid black bars (Cre-infected cells) with the solid white (uninfected cells) or striped bars (β-gal–infected cells). The results of the caspase 3 activity in B are the mean of 4 independent experiments.

In view of the cell death observed in Cre-infected hepatocytes, we measured caspase 3 activity at different postinfection times, as a means of quantifying the observed cell death. Starting on day 3 after adenoviral infections, caspase 3 activity was 4-fold to 9-fold increased in the Cre-infected cells compared with the uninfected cells or the cells infected with the control β-gal adenovirus (Fig. 3B, compare the solid black bars with the solid white or striped bars).

ILK Overexpression Substantially Rescues Apoptosis Induced by the Removal of ILK.

We next sought to find whether the effect of ILK depletion on apoptosis could be rescued via reintroduction of the protein. Cells were co-infected with either β-gal adenovirus and an adenovirus expressing ILK (ILK-adenovirus), or Cre-adenovirus and ILK-adenovirus, and harvested 4 days after infection. As shown in Fig. 4A, ILK was effectively removed by Cre-recombinase (Fig. 4A, lanes 3 and 5), and it was efficiently overexpressed in the cells that were co-infected with ILK-adenovirus (Fig. 4A, lanes 4 and 5). Interestingly, parallel caspase 3 activity measurement experiments for the same sets of cells showed a reduction of approximately 36% in the caspase 3 activity in the cells that were co-infected with Cre and also overexpressed ILK, compared with those that were solely infected with the Cre adenovirus, and the reduction was statistically significant (P < 0.05) (Fig. 4B). This indicates that the dramatic apoptotic effect in the Cre-infected, ILK-floxed hepatocytes results from the loss of ILK and can be substantially reversed by co-expression of ILK.

Figure 4.

ILK overexpression substantially rescues the effect on apoptosis induced by the loss of ILK. (A) Western blot showing ILK knock-down (lanes 3 and 5), and overexpression (lanes 4 and 5), 4 days after infection. (B) Caspase 3 activity 4 days after infections. Statistical analysis was performed with the Statgraphics Plus software using the Student t test to compare average standard error means. The P value was equal to 0.0139 (P < 0.05). The results shown are the average of 3 independent experiments. The same sets of cells were used for (A) and (B).

PINCH-1 and a-Parvin Expression Levels Are Reduced in the ILK-Depleted Hepatocytes.

To further investigate the mechanism by which ILK removal induces apoptosis in hepatocytes, we first sought to determine whether ILK-associated proteins are affected. To do so, we tested the levels of PINCH and α-parvin that, together with ILK, form a stable ternary complex at the cell–ECM adhesion sites.24, 26, 28, 30, 37 ILK-floxed hepatocytes that were infected with Cre-adenovirus and had reduced levels of ILK (Fig. 5A), also showed a dramatic reduction in both PINCH (Fig. 5B) and α-parvin protein expression levels (Fig. 5C). Interestingly, the level of mitogen-inducible gene-2 (Mig-2), another cell–matrix adhesion protein closely associated with ILK,34, 38 was not altered after the removal of ILK from hepatocytes (Fig. 5D).

Figure 5.

Changes in the expression level of other proteins as a result of removal of ILK from ILK-floxed mouse hepatocytes. Western blot analysis for (A) ILK, (B) PINCH-1, (C) α-parvin, (D) Mig-2, (E) phosphor-Akt (Ser473), (F) phosphor-MAPK (p44/42), (G) phosphor-JNK, (H) phosphor-p38, 5 days after infection with either β-gal or Cre-adenovirus. The Western blots shown here are representative of 3 independent experiments.

The Phosphorylation of PKB/Akt Is Not Affected by the Loss of ILK from Hepatocytes.

Previous studies have shown that inhibition of either ILK, PINCH, or α-parvin induces extensive apoptosis in HeLa cells, and that the activation of PKB/Akt, a key regulator of cell survival, is also impaired.27, 31 To test whether depletion of ILK from primary hepatocytes leads to apoptosis through the PKB/Akt pathway, we examined the expression level of phosphorylated-Akt at Ser473, the phosphorylation site that activates Akt.19, 27, 31 As shown in Fig. 5E, the level of phosphorylated Akt at Ser473 does not change after removal of ILK.

To further elucidate the mechanism underlying the apoptosis phenomenon observed after removal of ILK from hepatocytes, we tested the phosphorylation status of the main members of the MAPK signal transduction pathways, as potential mediators of the ILK signaling. Phopsho-MAPK (p44/42) (Fig. 5F), phopsho-JNK (Fig. 5G), and phospho-p38 (Fig. 5H) were examined in uninfected hepatocytes or hepatocytes infected with β-gal or Cre adenovirus, but no significant change was observed in any of them.

ILK Protein Is Significantly Reduced in Animals Injected with Cre-Adenovirus, In Vivo.

To gain further knowledge on the significance of ILK for hepatocytes, we employed an in vivo approach. PBS, β-gal, or Cre adenovirus were delivered into ILK-floxed animals via the tail vein.39 Injections were performed in 78 ILK-floxed mice (19-31 weeks old), and mice were killed 2, 4, 5, 7, 8, or 10 days later. Livers were harvested and analyzed by PCR and immunoblotting.

As shown in Fig. 6A, the ILK protein level was reduced in the Cre-adenovirus–injected animals compared with the β-gal adenovirus–injected animals in all animals injected with Cre adenovirus at all the time points tested. This correlates with specific expression of Cre recombinase in the Cre adenovirus–injected mice (Fig. 6B; compare lanes 2 and 3 with lanes 4 and 5. The band that corresponds to the Cre gene is shown with an arrow).

Figure 6.

Elimination of ILK from the mouse liver by administration of Cre-recombinase expressing adenovirus via tail vein. (A) Western blots showing the expression level of ILK in samples from untreated ILK-floxed mouse livers (lanes 1 and 2), livers from mice injected with either β-gal (lanes 3 and 5) or Cre adenovirus (lanes 4 and 6) at days 5 and 10 after injection. The mAb anti-ILK antibody was used for the Western blot analysis. Equal loading was confirmed by actin reprobing of the same membrane. (B) PCR using primers that detect the presence of Cre gene (arrow). Note that only the Cre-injected animals have the Cre gene (compare lanes 2 and 3 with 4 and 5).

Livers from the Cre-Adenovirus–Injected Animals Are Significantly Impaired and Show Signs of Hepatitis.

ILK-floxed mice were killed at different times after injection with either β-gal or Cre-adenovirus, and liver histology was examined. The gross appearance of the livers from most of the Cre-adenovirus–injected animals (35 of 45) was greatly altered. Livers were larger, with a coarse granular surface and very pale, compared with control livers from β-gal adenovirus–injected mice (Fig. 7, compare A and B).

Figure 7.

Gross appearance and HE staining of the livers from β-gal– or Cre-injected animals. (A) Gross appearance of the liver of a control β-gal adenovirus–injected mouse. (B) Gross appearance of the liver of a mouse injected with Cre-recombinase expressing adenovirus. (C) HE staining on frozen liver sections taken from a β-gal adenovirus–injected mouse. (D) HE staining on frozen liver sections taken from a Cre-adenovirus–injected mouse. Arrows point to apoptotic cells. (E) HE staining on formalin-fixed, paraffin-embedded sections from a β-gal adenovirus–injected mouse liver. (F) HE staining on formalin-fixed, paraffin-embedded sections from a Cre-adenovirus–injected mouse liver. Arrows point to apoptotic cells, whereas arrowheads show necrotic cells with ballooning degeneration. (G) TUNEL staining on formalin-fixed, paraffin-embedded sections from a β-gal adenovirus–injected mouse liver. (H) TUNEL staining on formalin-fixed, paraffin-embedded sections from a Cre-adenovirus–injected mouse liver.

At the microscopic level, the liver histology exhibited a variety of abnormal characteristics, and it was blindly scored on a scale of 0 to 5. The highest score (5) was assigned to livers with the worst histology exhibiting inflammation, or fatty change, apoptosis, abnormal mitoses, ballooning degeneration, hydropic degeneration, and necrosis. Collectively, 85% (23 of 27) of the β-gal adenovirus–injected animals had normal livers (score 0-1) whereas the remaining 15% exhibited signs of mild inflammation and had a few apoptotic cells (score 2). In the Cre adenovirus–injected animals, 78% (35 of 45) had abnormal liver histology (score 3-5), whereas 22% appeared to have relatively normal histology (score 1-2). That can be attributed to problems with the tail vein administration of the virus, which resulted in the actual delivery of only a portion of the intended amount of virus in some of the animals. The Cre-adenovirus–injected animals, at each one of the time points tested, had more severe pathology and thus higher scores than the control β-gal–injected ones, and the difference was statistically significant, as shown by Mann-Whitney test (P < 0.05) (Fig. 8). Interestingly, three animals from the Cre-injected group died 2, 4, and 5 days after being injected, and their livers showed the most severe pathology. Representative images of the histology of the control and Cre-adenovirus-injected animals are shown in Fig. 7 (compare C and E with D and F). Finally, confirming the apoptosis observed from the HE staining, the livers of the Cre-adenovirus–injected animals exhibited dramatically increased TUNEL staining (Fig.7, compare G and H).

Figure 8.

Statistical analysis of the differences in the histology between β-gal and Cre adenovirus–injected ILK-floxed mouse livers. The bar graph shows the differences in the liver histology scores between β-gal control and Cre adenovirus–injected ILK-floxed animals, at different times after injection. The day of injections is considered as day 0. The statistical analysis was conducted using the Mann-Whitney test and the Statgraphics Plus statistics software. A P value of less than 0.05 was considered statistically significant.

Lastly, to determine whether genetic susceptibility to Cre recombinase toxicity played a role in the obtained results, eight wild-type mice of the strains 129X1/SvJ and C57BL/6J (the background of the ILK-floxed mice) were injected with Cre adenovirus using the same exact titer as the one used in the ILK-floxed animals. Wild-type livers were normal, both macroscopically and microscopically, indicating that the severe phenotype observed in the Cre-injected animals was a response to loss of ILK and not a side effect of the Cre adenovirus.


Hepatocytes are the parenchymal cells of the liver and as such perform a number of complex functions to maintain homeostasis. Of particular importance for normal hepatocyte function, both in vivo and in vitro, is the surrounding matrix.1, 3, 40, 41

ILK is a cell–ECM adhesion protein that mediates signal transduction between integrins, ECM, and the interior of cells. It directly interacts with PINCH and α-parvin, and together they form a stable ternary complex at the cell–ECM adhesion sites.30 ILK, as well as PINCH and α-parvin, are widely expressed in the human tissues and well conserved among different species.20, 38, 42 As such, it is not surprising that ILK is critically involved in many fundamental cellular processes.43–45

In the current study, we investigated the role of ILK in mouse hepatocytes by specifically eliminating it in vitro from primary hepatocytes in culture and in vivo from the whole liver. In vitro, the ILK-depleted cells underwent dramatic apoptosis, as assessed by morphological changes and caspase 3 activity measurements. Interestingly, re-introduction of ILK substantially reversed the effect on apoptosis. All in all, the fact that elimination of ILK leads to cell death suggests that ILK is important for hepatocyte survival (Fig. 3). We have noticed, however, that there is also upregulation in the expression level of ILK in culture (Fig. 2, compare ILK in lanes 1 and 2 with 5 and 8). This may be related to altered differentiation of hepatocytes in primary culture, or it may account for a role of ILK in providing necessary anti-apoptotic signals for hepatocyte survival.

Cells lacking ILK also display reduced levels of PINCH and α-parvin, whereas Mig-2, another ILK-binding protein, is not affected. The reduction in the level of PINCH and α-parvin after ILK-depletion has been shown previously in other cell types,27, 31 whereas both PINCH and α-parvin have been implicated in cell survival.27, 31 Therefore, reduction of their level could contribute to or enhance the apoptosis induced by the decreased level of ILK. Hence, in cultured primary hepatocytes, the components of the PINCH–ILK–parvin complex appear to function in concert to mediate cellular signaling for survival.

Mig-2 is a cell–ECM adhesion protein closely associated with the PINCH–ILK–parvin complex and integrin signaling and has been shown to regulate cell shape and spreading.34, 43, 45 Mig-2 level does not change after ILK removal from hepatocytes, which indicates that not all the cell-matrix adhesion proteins are affected by the depletion of ILK, but rather the two proteins that form the ternary complex with ILK are the ones that are most dramatically affected.

Several studies in cancer cell lines or immortalized cells have shown that ILK regulates survival through the PKB/Akt pathway.18, 19, 21, 31, 46 In this study, we tested the expression level of active-phosphorylated-Akt (Ser473) following depletion of ILK from primary hepatocytes and found that it was not affected (Fig. 5E), indicating that ILK regulates primary hepatocyte survival through a pathway other than the PKB/Akt. Our finding is in accordance with other studies using ILK-deficient fibroblasts43, 47 that also showed no alterations in the phosphorylation status of Akt. The reason for the discrepancy is unclear but may be attributed to either cell-type–specific differences or possibly differences in the mechanisms that are activated in primary cells versus immortalized or cancerous cell lines. We also examined the phosphorylation level of the three main members of the MAPK family of proteins (phosphor-ERK, phosphor-JNK, phosphor-p38), but no significant change was revealed in any of them, indicating that the mechanism governing the apoptosis induced by loss of ILK is more complex.

Consistent with our in vitro data, in vivo elimination of ILK from ILK-floxed mice resulted in severe liver abnormalities ranging from inflammation, apoptosis, and fatty change to acute, full-scale hepatitis. The finding that the liver was the most obviously affected organ by the ILK elimination is explained by the fact that administration of Cre-adenovirus through the tail vein leads to higher infectivity of the adenovirus with higher recombination levels in the liver and spleen, as opposed to other organs.48

Furthermore, the adenoviral injection itself caused some inflammation as shown by the results of the β-gal–injected animals, 15% of which show signs of inflammation and small-scale apoptosis. This is consistent with previous studies demonstrating toxicity effects of adenovirus vectors in hepatocytes.49

Altogether, our results clearly show that elimination of integrin signaling by removal of ILK causes apoptosis in primary cultured hepatocytes, as well as in hepatocytes in the whole liver. Thus, ILK is of great importance for the liver in general, consistent with recent work by Shafiei and Rockey50 highlighting the importance of ILK in liver wound healing.

ECM signaling is mediated through integrins. ILK plays a major role in mediating integrin signaling. Disassociation between epithelial cells and surrounding matrix is known to result in apoptosis, a phenomenon known as “anoikis.” Our results suggest that ILK is a major component of apoptosis induced by removal of ECM signaling in hepatocytes.

Nevertheless, more research is needed to decipher the exact mechanisms involved. Understanding the mechanisms that regulate survival in hepatocytes in vitro and in vivo is of utmost importance because the balance between cell survival and death is delicate and can be easily disrupted in cases of liver injury.