Cell adhesion to the extracellular matrix protein fibronectin modulates radiation-dependent G2 phase arrest involving integrin-linked kinase (ILK) and glycogen synthase kinase-3β (GSK-3β) in vitro

Cell adhesion to extracellular matrix (ECM) is thought to confer resistance against cell-damaging agents, that is, drugs and radiation, in tumour and normal cells in vitro. The dependence of cell survival on β1-integrin-linked kinase (ILK), protein kinase Bα/Akt (PKBα/Akt) and glycogen synthase kinase-3β (GSK-3β) activity, which participate in β1-integrin signalling and cell cycle progression was investigated as a function of radiation exposure. Colony-formation assays on polystyrene, fibronectin (FN), laminin (LA), bovine serum albumin (BSA) or poly-L-lysine (poly-L) (0–8 Gy), kinase assays, flow cytometric DNA and annexin-V analysis and immunoblotting were performed in nonirradiated and irradiated (2 or 6 Gy) A549 human lung cancer cells and CCD32 normal human lung fibroblasts. Cell contact to FN in contrast to polystyrene elevated basal ILK, PKBα/Akt and GSK-3β kinase activities in A549 and CCD32 cells, as well as the basal amount of A549 G2 phase cells. Irradiation on FN or LA as compared to polystyrene, BSA or poly-L significantly improved cell survival. Following irradiation, kinase activities were stimulated strongly on polystyrene but showed to be less prominent on FN, which was because of the FN-related basal induction. Following irradiation, FN compared to polystyrene enlarged and prolonged G2 arrest in both the cell lines. For the analysis of phosphatidylinositol-3 kinase (PI3-K) dependence of protein kinases and cell cycle transition, the PI3-K inhibitors LY294002 and wortmannin were used showing decreased kinase activities, antiproliferative and radiation-dependent G2 accumulation-abrogating effects accompanied by downregulation of cyclin D1 and phospho-pRb in cells attached to polystyrene. Fibronectin partly abrogated these effects PI3-K-independently. These findings suggest a novel pathway that makes direct phosphorylation of GSK-3β by ILK feasible after irradiation. Conclusively, the data indicate that ILK, PKBα/Akt and GSK-3β are involved in modulations of the cell cycle after irradiation. These interactions are strictly dependent on ECM components in a cell line-specific manner. Our findings provide molecular insights into mechanisms likely to be important for ECM-dependent cell survival and cellular radioresistance as well as tumour growth.

The presence of extracellular matrix (ECM) increases cellular resistance to cell-damaging agents such as drugs or ionising radiation (Damiano et al, 1999;Rose et al, 1999;Hazlehurst et al, 2000;Cordes and Meineke, in press;Cordes et al, in press). For the treatment of cells with cytotoxic agents in the presence of ECM, the convergence of integrin-related and cell cycle-related signal transduction pathways could be of great importance. Possible interactions of these pathways are likely to be modulated by the ECM and, thus, may play an important role in resistance-mediating mechanisms (Sethi et al, 1999;Hazlehurst et al, 2000), anchorage-independent cell growth (Shin et al, 1975;Radeva et al, 1997), oncogenic transformation (White et al, 2001), the assessment of in vitro cytotoxicity studies, and new strategies for controlling and healing cancer.
Signals generated by cell -ECM contact to critical cellular processes, that is, cell cycle transition, survival, differentiation, migration or adhesion, are transmitted in particular by the integrin receptor family (Hynes, 1992;Giancotti and Ruoslahti, 1999). Among at least 22 different a/b-heterodimers, the b1-integrin is the most widespread receptor subunit existing in combination, for example, with a5 the fibronectin receptor (Argraves et al, 1987) and with a1, a3 or a6 receptors for laminin or collagen (Hynes, 1992). Initiated integrin clustering after cell -ECM attachment activates intracellular effectors that are capable of coupling integrins and growth factor receptors to specific downstream targets such as b1-integrin-linked kinase (ILK) (Hannigan et al, 1996) or focal adhesion kinase (Schaller and Parsons, 1994). b1integrin-linked kinase functioning in a phosphatidylinositol-3 kinase (PI3-K)-dependent manner (Delcommenne et al, 1998) is able to stimulate protein kinase Ba/Akt (PKBa/Akt) and glycogen synthase kinase-3b (GSK-3b) (Troussard et al, 1999). These events proceed in inhibition of apoptosis by phosphorylating and inactivating apoptotic factors such as caspase-9 and Bad through PKBa/Akt (Datta et al, 1997;Khwaja et al, 1997;Tian et al, 2002) and cell cycle progression via prevention of cyclin D1 proteolysis (Diehl et al, 1998) and gene expression via AP-1 or NF-kB through GSK-3b (Troussard et al, 1999;Hoeflich et al, 2000).
Concerning the coactivation of in-membrane complexes-associated integrin and growth factor receptors, identical intracellular biochemical pathways are used by the different receptors, but in many cases they do it at different steps (Pawson and Saxton, 1999;Schwartz and Baron, 1999). These coactivations are thought to be necessary in certain cases for optimal signal transduction to regulate diverse cellular functions (Schwartz and Ginsberg, 2002;Yamada and Even-Ram, 2002). Thus, the treatment of ECMattached cells with external cytotoxic stimuli is likely to show a modulated responsive pattern in comparison with artificial substrates, that is, polystyrene (Meredith et al, 1993;Giancotti and Ruoslahti, 1999).
An integral part of the cellular radiation response pattern upon DNA damage is the blockage of the cell cycle. Recent findings also reported intense alterations of cell cycle regulatory proteins through cell detachment presenting similar radiation-related DNA damages (Kang and Krauss, 1996;Zhu et al, 1996;Wu and Schönthal, 1997). The question that still remains unclear is what are the acute consequences on the cell cycle after irradiation of monolayer cultures in an optimised in vitro microenvironment.
To elucidate the molecular mechanisms involved in improved clonogenic cell survival after irradiation by ECM proteins, we analysed the acute changes of the activities of the upstream-located b1-integrin signalling pathway protein kinases ILK, PKBa/Akt and GSK-3b, as well as cell cycle progression in human lung tumour cells and normal human lung fibroblasts in vitro.

Cells
The lung carcinoma cell line A549 was purchased from the American Type Culture Collection (ATCC, Rockville, MD, USA) and the normal human lung fibroblastic stem line (CCD32) was a generous gift from Professor HP Rodemann (Section of Radiobiology and Molecular Environmental Research, University Tuebingen). Dulbecco's modified Eagle's medium (DMEM) (PAA, Linz, Austria) supplemented with 1% nonessential amino acids and 10% foetal bovine serum (FBS; PAA, Linz, Austria) was applied to culture the cells. Routinely, cells were incubated at 371C in a humidified atmosphere containing 10% CO 2 buffered at pH 7.35. Where indicated, serum starvation of cells was always performed using DMEM supplemented with 1% nonessential amino acids without FBS. All experiments using CCD32 cells were performed between passages 5 and 15. Furthermore, for all experiments only asynchronous, exponentially growing cell cultures were employed.

Radiation exposure
Irradiation was delivered at room temperature using single doses of 240 kV X-rays (Isovolt 320/10; Seifert, Ahrensburg, Germany) filtered with 3 mm Be. The absorbed dose was measured using a Duplex dosimeter (PTW, Freiburg, Germany). The dose rate was approximately 1 Gy min À1 at 13 mA. The applied doses ranged from 0 to 8 Gy.

Colony-formation assay
The colony-formation assay was applied for the measurement of clonogenic cell survival. Exponentially growing A549 or CCD32 cells were plated onto noncoated or fibronectin (FN; 1 mg cm À2 ; Becton Dickinson, Heidelberg, Germany), laminin (LA; 1 mg cm À2 ; Sigma-Aldrich GmbH, Taufkirchen, Germany), bovine serum albumin (BSA; 1 mg cm À2 ; GIBCO, Karlsruhe, Germany) or poly-L-lysin (poly-L; 1 mg cm À2 ; Calbiochem-Novabiochem GmbH, Bad Soden, Germany) precoated six-well dishes (Becton Dickinson, Heidelberg, Germany) 36 h prior to irradiation. To examine the effect of the specific PI3-K inhibitor LY294002 (Sigma-Aldrich GmbH, Taufkirchen, Germany) on clonogenic survival, cells were seeded on polystyrene or FN for 24 h, serum-starved and incubated with concentrations of 0.1 -100 mM of the inhibitor for 18 h. In combination with irradiation, cells were prepared as described and withdrawal of the inhibitor was performed immediately prior to radiation. At 8 -10 days after irradiation, grown colonies were stained with Coomassie blue. Colonies greater than 50 cells were counted. All experiments were repeated three times.

b1-integrin-linked kinase assay and Western blotting
As reported, ILK (Delcommenne et al, 1998) and PKBa/Akt (Cross et al, 1995) are able to phosphorylate GSK-3b in vitro. Cells were plated onto polystyrene or FN for 24 h, subsequently serumstarvation 750 mM LY294002 or 100 nM wortmannin occurred for 18 h prior to irradiation (6 Gy). Cell lysis subsequent to irradiation was performed for 10 min on ice at 5 and 60 min. Untreated and treated cells were scrapped off and 500 mg of the total protein extracts were incubated for 3 h, firstly, with monoclonal ILK antibodies clone 65.1.9 (Upstate, Hamburg, Germany) at 41C and, secondary, with protein-A -agarose beads (Sigma-Aldrich GmbH, Taufkirchen, Germany) at room temperature. Immunoprecipitated ILK was then employed for protein kinase activity measurements using kinase buffer plus GSK-fusion protein and 200 mM ATP (New England Biolabs, GmbH, Frankfurt a.M., Germany). After a 30 min incubation at 301C, the reaction was terminated with 4 Â SDSsample buffer. Immunoprecipitates were separated by SDSpolyacrylamide electrophoresis, transferred onto a nitrocellulose membrane (Schleicher and Schuell GmbH, Dassel, Germany), blocked using 5% nonfat dry milk powder in PBS and probed overnight at 41C with rabbit antiphospho-GSK-3b antibodies specific for Ser9 (New England Biolabs, GmbH, Frankfurt a.M., Germany). Three independent experiments were performed. The protein detection was accomplished using specific HRP-conjugated goat anti-rabbit antibodies in combination with the enhanced chemiluminescence detection systems (ECL; Amersham, Freiburg, Germany). Measurements of protein band density were carried out using ImageQuant version 5.0 software (Molecular Dynamics, Krefeld, Germany).

PKBa/Akt kinase assay
The nonradioactive Akt kinase assay kit purchased from New England Biolabs, GmbH, Frankfurt a.M., Germany, was used. Cells were plated onto polystyrene or FN for 24 h, subsequently serumstarvation 750 mM LY294002 or 100 nM wortmannin occurred for 18 h prior to irradiation (6 Gy). Then, cells were processed as recommended by the manufacturer's device. Immobilised Akt antibody agarose beads were used for immunoprecipitation. Subsequently, kinase buffer, 200 mM ATP and GSK-fusion protein were added to the PKB -agarose bead complex. After a 30 min incubation at 301C, the reaction was terminated and SDSpolyacrylamide electrophoresis and Western blotting were performed as described above. Experiments were repeated three times.

Flow cytometric analysis of apoptosis by annexin-V staining
The effect of LY294002 on the induction of apoptosis was analysed by flow cytometry. Treatment of 1 Â 10 6 cells grown on polystyrene or FN with 50 mM LY294002 for 18 h in combination with serum deprivation was followed by detachment of cells using trypsin/ EDTA, washing with PBS and preparation for analysis following the instructions of the annexin-V-FLUOS staining Kit (Roche, Germany). After simultaneous staining of cells with annexin-VFfluorescein plus propidium iodide solution for 15 min, acquisition of data for 10 000 events was performed using a Becton Dickinson Fluorescence-activated Cell Sorter (FACScan). The distribution and differentiation of vital, apoptotic and necrotic cells were analysed from the dot plots using CELLQuest software. Experiments were repeated two times.

Cell cycle analysis
Nonirradiated or irradiated cells (1 Â 10 6 ) grown on FN or polystyrene were detached at 0, 4, 8, 12 and 24 h, washed twice with PBS and fixed in 80% ethanol until use. Where indicated, cells were incubated with 50 mM LY294002 and serum starved from 18 h prior to irradiation up to 24 h after irradiation. Cells were prepared for analysis following the instructions of the CycleTESTt PLUS DNA Reagent Kit (Becton Dickinson, Heidelberg, Germany). After staining of cells with propidium iodide solution for 30 min, acquisition of data for 10 000 events was performed with the means of a FACScan. The distribution of cells in the different phases of the cell cycle was analysed from the DNA-histograms using CELLQuest software. Experiments were repeated three times.

Data analysis
Means7s.d. of surviving fractions, protein kinase activities, induction of apoptosis and cell cycle data were calculated with reference to untreated controls defined as 1.0 or in a percentage scale. To test statistical significance, analysis of variance was performed by means of ANOVA with a software package (Microsoft, Exel 97) on IBM computer systems. Results were considered statistically significant if a P-value of less than 0.05 was reached. The fit of the dose -effect curves was calculated by means of the linear-quadratic model (log S ¼ ÀaDÀbD 2 ).

Extracellular matrix-dependent colony formation
To test whether FN or LA influences the clonogenic survival of A549 and CCD32 cells after irradiation, cells were plated onto FN-, LA-, BSA-or poly-L-coated or noncoated surfaces. The plating efficiency of cells was B70% (A549) or B10% (CCD32) independent of the substrate on which the cells were plated. Significantly improved survival could be detected for irradiated cells grown on FN or LA as compared to culture polystyrene, BSA or poly-L ( Figure 1). The P-values of radiation doses X4 Gy for A549 cells on FN or LA and X2 Gy for CCD32 cells on FN or LA were calculated to be less than 0.05. Cell -fibronectin contact modulates G2 phase arrest N Cordes and D van Beuningen b1-integrin-linked kinase, PKBa/Akt and GSK-3b kinase activities We analysed ILK, PKBa/Akt and GSK-3b activities to evaluate their role within the acute cellular radiation response (Figure 2A -C).
Concerning ILK, we could show a fast and transient radiationdependent induction in both A549 and CCD32 cells (Figure 2A). Radiation-induced ILK activity in cells grown on FN demonstrated to be less prominent compared to cells attached to polystyrene. This difference in induction might be because of cell adhesionmediated elevation of basal ILK activity on FN compared to polystyrene. Additionally, PKBa/Akt and GSK-3b also demonstrated elevated basal kinase activity or Ser9 phosphorylation, respectively (Figure 2A -C). The downstream target of ILK PKBa/ Akt showed a fast activation within 5 min after irradiation ( Figure 2B). Subsequently, in contrast to ILK, PKBa/Akt remained on a high level up to 60 min post irradiation. For GSK-3b, we detected a strong increase in phosphorylation at the amino-acid residue Ser9 by irradiation in cells grown on FN or polystyrene ( Figure 2C).
To investigate PI3-K-dependence of ILK, PKBa/Akt and GSK-3b (Delcommenne et al, 1998), cells were exposed to 50 mM LY294002 or 100 nM wortmannin. Both A549 and CCD32 cells grown on Protein kinase activities of ILK (A) and PKBa/Akt (B) and GSK-3b phosphorylation at the amino-acid residue Ser9 (C) were examined in A549 and CCD32 cells attached to polystyrene (P) or FN at 5 or 60 min after irradiation with 6 Gy (right panels). Basal kinase activities were strongly stimulated by FN in both tumour and normal cells (left panels). Radiation-dependent increases of ILK and PKBa/Akt activity and GSK-3b phosphorylation demonstrated to be pronounced on polystyrene and less prominent on FN. Additionally, ILK, PKBa/Akt and GSK-3b protein were detected to exclude changes in total protein amounts and b-actin served as loading control (data not shown). Each data point shown represents the fold induction (means7s.d.) of protein kinase activity or Ser9 phosphorylation of GSK-3b from the densitometric protein band analysis of three independent experiments in relation to untreated controls (mock). Inset, photographic demonstration of one exemplary protein kinase assay used for densitometric analysis. To examine the dependence of ILK, PKBa/Akt activity and GSK-3b phosphorylation on the PI3-K pathway, cells attached to FN or polystyrene were incubated with the PI3-K-specific inhibitors LY294002 (50 mM) (D) or wortmannin (100 nM) (E) in serum-free medium and then irradiated with 6 Gy (IR). The data uncovered PI3-Kdependent decrease of basal and radiation-induced kinase activities (ILK and PKBa/Akt at 5 min; GSK-3b at 60 min) in cells grown on FN or polystyrene. However, irradiation on FN was able to stimulate ILK and GSK-3b independent of PI3-K indicating the involvement of yet unknown signalling pathways.
Cell -fibronectin contact modulates G2 phase arrest N Cordes and D van Beuningen polystyrene or FN demonstrated pronounced decreases in basal ILK and PKBa/Akt activity as well as GSK-3b phosphorylation after incubation with these inhibitors indicating a PI3-K-dependent mechanism ( Figure 2D and E). Most interestingly, in contrast to PKBa/Akt, the inhibitory effect of LY294002 or wortmannin was partly abolished by irradiation in ILK and GSK-3b when cells had FN contact.

Dependence of colony formation and induction of apoptosis on PI3-K
Single 18-h LY294002 treatment using various concentrations in combination with serumdeprivation resulted in a nonlinear and matrix-independent dose -response relation for A549 and CCD32 cells ( Figure 3A and D). Choosing a 50 mM LY294002 concentration Cell -fibronectin contact modulates G2 phase arrest N Cordes and D van Beuningen for further experiments was based on an acute, noncytotoxic effect of the agent, which was determined by measurement of apoptosis using annexin-V staining ( Figure 3B and E), as well as a strong inhibition of PI3-K-dependent protein kinases ILK, PKBa/Akt and GSK-3b. Investigating the influence of PI3-K inhibition in combination with irradiation on the clonogenic survival of A549 and CCD32 cells was performed to assess the consequences seen in the acute alterations of ILK, PKBa/Akt and GSK-3b ( Figure 3C and F). Fibronectin counteracted significantly (Po0.01) the cytotoxic effect of LY294002 in both irradiated A549 and CCD32 cells.

Cell cycle analysis
Next, the influence of FN on the cell cycle response of irradiated A549 and CCD32 cells was examined (Figure 4) with regard to ILK and GSK-3b involvement in the regulation of cell cycle events (Radeva et al, 1997). As indicated by flow cytometry, nonirradiated A549 cells grown on FN demonstrated an increase of G2 phase cells by a two-fold and a concomitant decrease of S phase cells 24 h after plating in comparison with cells plated onto polystyrene. In contrast, nonirradiated CCD32 cells grown on FN or polystyrene did not show any differences in cell cycle distributions. A549 cells on FN presented a dose-dependent G2 block, which showed to be increased and prolonged compared to the G2 phase block seen in cells on polystyrene (Figure 4). Cells attached to FN accumulated significantly (Po0.01) in G2 from 36% in nonirradiated controls to 56% (2 Gy) or 75% (6 Gy) and on polystyrene from 14 to 27% (2 Gy) or 50% (6 Gy) (Figure 4). Cells in the G0/G1 or S phase shifted concomitantly to the changes detected in G2. Concerning CCD32 cells, irradiation induced a dose-dependent G2 phase arrest. Similar to A549 cells, FN markedly altered the cell cycle distribution presenting 43 or 60% G2 phase cells after 2 or 6 Gy compared to 23% in the untreated control ( Figure 4). Irradiation on polystyrene resulted in an accumulation of G2 phase cells from 19% in the untreated control to 27% (2 Gy) or 30% (6 Gy) (Figure 4). Cells in the G0/G1 or S phase shifted concomitantly to changes detected in G2. In both A549 and CCD32 cells, the normal cell cycle distribution similar to untreated controls was reached 24 h after irradiation.
In spite of G2 phase arrest in both the cell lines on polystyrene and FN observed by flow cytometry (see Figure 4), A549 and CCD32 cells showed pronounced differences in cyclin D1, pRb and phospho-pRb expression patterns comparing polystyrene to FN ( Figure 5A and B). A549 cells grown on both FN and polystyrene demonstrated decreases in cyclin D1 at 2 h on polystyrene and at 12 h on FN. In parallel, pRb protein and Ser795 phosphorylation diminished almost completely after irradiation on polystyrene, whereas on FN the strong Ser795 phosphorylation remained unchanged. At 24 h postirradiation, pRb Ser795 phosphorylation was undetectable on both the substrates ( Figure 5A).
Concerning CCD32 cells, cyclin D1 decreased at 6 h on polystyrene and continuously on FN ( Figure 5B). The pattern of pRb protein and pRb phosphorylation showed similarity to A549 cells with regard to the dramatic reduction at 24 h postirradiation. On polystyrene, pRb expression was induced after irradiation and reduced on FN. The phosphorylation of pRb at Ser795 was changed in parallel.
With regard to the convergence of integrin and growth factor signalling cascades, we analysed whether the detected PI3-K- Cell -fibronectin contact modulates G2 phase arrest N Cordes and D van Beuningen dependent and -independent changes in ILK, PKBa/Akt activity and GSK-3b phosphorylation (see Figure 2D) could be responsible for alterations in cell cycle progression, cyclin D1 expression and pRb phosphorylation (Figures 6 and 7). LY294002 exposure showed marked decreases in the G2 cell fraction of A549 cells on polystyrene (2.2%) or FN (12%) and of CCD32 cells on polystyrene (1%) or FN (5.2%) compared to noninhibited controls (see Figure 4). The radiation-dependent G2 cell accumulation was completely impaired in cells grown on polystyrene and decreased in cells grown on FN ( Figure 6). Concomitantly to changes detected in G2, cells in the G0/G1 and more prominent in the S phase increased. In parallel, cyclin D1 and phospho-pRb expression were downregulated after a 12-h incubation with LY294002 on polystyrene but, most interestingly, not on FN (Figure 7).

DISCUSSION
Physiologically normal and malignant cells are imbedded into an ECM, which is a requirement for cell proliferation and survival for a wide variety of cell types (Meredith et al, 1993;Frisch and Francis, 1994). Furthermore, the ECM might have a great impact on the cellular responsiveness to external stimuli. Published in vitro and in vivo findings showed the importance of the extracellular environment on drug (Sethi et al, 1999;Hazlehurst et al, 2000) and radiation sensitivity (Vlodavsky et al, 1980;Cordes and Meineke, in press;Cordes et al, in press), conferring resistance to DNA-damaging agents. Since integrin and growth factor receptors are colocalised at the source of integrin signalling, which are called focal adhesions, convergence and mutual modification between these networks are highly likely to occur. This crosstalk could result in differentiated regulative schemes of cellular functions such as cell growth and survival and, thus, may be central to our understanding of resistance-mediating mechanisms. These mechanisms are not only important for single tumour or normal cells, but are possibly of outstanding interest for a growing tumour that consists of organ-specific, already transformed epithelial cells and the adjacent stroma containing normal cells such as fibroblasts, endothelial cells and leucocytes (Seljedid et al, 1999). Focusing on tumour and fibroblastic cells of, for example, a human lung cancer, the cells correspond intensely via cell -cell contact and soluble signalling molecules that modulate the composition of the extracellular environment. These interactions have been reported to be able to support strongly the growth of the tumour (Lukashev and Werb, 1998;Kunz-Schughart and Knuechel, 2002). On the one hand, every single cell responds on its own upon treatment with a cytotoxic agent like ionising radiation, but, on the other, there occur multiple interactions between these two neighbours, which might modulate the cellular radiation response in terms of improved radioresistance. In this study, our in vitro results show, for the first time, a participation of the widespread b1-integrin subunit-associated protein kinases, ILK, PKBa/Akt, GSK-3b within the radiation response of a human lung cancer cell line as well as normal human lung primary fibroblasts. These data indicate that these specific kinases are likely to be involved in improved radioresistance at ECM presence, in ECM-modulated cell cycle progression and the expression of cyclin D1 and pRb (Figure 8).
The major findings of this study include, firstly, a significantly improved clonogenic survival of A549 and CCD32 cells after irradiation in the presence of FN or LA in contrast to polystyrene, BSA or poly-L; secondly, ILK and PKBa/Akt stimulation and increased GSK-3b phosphorylation by ionising radiation in a matrix-dependent manner; thirdly, a PI3-K-independent ILK stimulation and GSK-3b inhibition by irradiation at FN presence in combination with undisturbed cyclin D1 expression and pRb phosphorylation; fourthly, a prolongation and increase of the radiation-induced G2 phase arrest by FN which could be correlated with the expression patterns of cell cycle proteins and which could be impaired by PI3-K inhibition in a matrix-dependent manner. Additionally, basal protein kinase Cell cycle alterations by the PI3-K-specific inhibitor LY294002 (LY) were analysed in the presence or absence of FN using nonirradiated A549 and CCD32 controls in comparison with irradiated cells. Irradiation was delivered following an 18-h incubation with 50 mM of LY294002 in serum-free medium. Each data point represents the mean7s.d. of three independent experiments. LY294002 led to a strong decrease in G2 phase cells on polystyrene, which was partly prevented by FN. Radiationdependent accumulation of cells in the G2 phase was completely prevented on polystyrene and markedly decreased on FN in comparison with irradiated cells not exposed to LY294002 as shown in Figure 4.  Figure 7 Cyclin D1 expression and pRb phosphorylation were investigated after a 12-h LY294002 (50 mM) incubation in A549 and CCD32 cells grown on polystyrene or FN in comparison with untreated controls (mock). Correlating the decreases of cyclin D1 expression and pRb phosphorylation with the DNA analysis shown in Figure 6 supported the important role of PI3-K-dependent growth factor pathways for the regulation of cell division and associated cell cycle proteins at FN presence and absence. Most interestingly, cells attached to FN did not demonstrate a downregulation of cyclin D1 expression or pRb phosphorylation, which indicates a strong influence of ECM on the cell cycle regulation via ILK-GSK-3b.

Mock
activities are shown to be markedly elevated when cells had FN contact.
Cell survival and cell growth regulated by growth factor signalling (Moustakas et al, 2002;Neve et al, 2002) have been shown by Wu et al (1998) to be also substantially affected by ILK signalling upon ligand binding to b1-integrin receptors. The ILK signalling cascade seems to be involved in tumour growth, tumorigenesis and in the development of a metastatic phenotype. Recently, White et al (2001) provided direct evidence for the oncogenic potential of ILK in the induction of mammary gland hyperplasia and formation of tumours in transgenic mice in vivo. The observed ILK overexpression was accompanied by a constitutive phosphorylation of the ILK downstream targets PKBa/Akt and GSK-3b. To date, investigating these molecular factors under cell treatment with cytotoxic agents such as ionising radiation or drugs has only uncovered PKBa/Akt but not ILK or GSK-3b involvement. PKBa/Akt affects cellular radiosensitivity (Akimoto et al, 2001) in a PI3-K- (Enns et al, 1999;Brognard et al, 2001) or protein kinase C-dependent manner (Tenzer et al, 2001) leading to changes in GSK-3a phosphorylation and alterations of cellular radiosensitivity. Concerning DNA damage caused by chemical agents such as camptothecin, Watcharasit et al (2002) were able to show GSK-3b participation within the cellular response of DNA repair. However, changes in basal kinase activities because of cell -ECM contact have not been taken into account. As shown here, cell growth on FN, which characterises a more physiologic experimental condition, activated the b1-integrin pathway per se in a cell type-independent manner. With regard to the improved clonogenic survival, we hypothesise that this basal activation provides an optimised physiologic status for the cell to counteract external stimuli. The radiation-mediated stimulation (ILK, PKBa/Akt) and inhibition (GSK-3b) indicates the involvement of the tested protein kinases in the acute radiation response mechanisms of A549 and CCD32 cells.
Elucidating the process of cell cycle transition that is markedly influenced by irradiation in that cells respond by delaying the cell cycle at checkpoints in the G1-or G2-phase (Lane, 1992;Gadbois et al, 1997;Dimitrijevic-Bussod et al, 1999), additionally showed a regulative role for cell adhesion to ECM (Shin et al, 1975;Hannigan et al, 1996;Kang and Krauss, 1996;Zhu et al, 1996). These checkpoint delays are believed to provide the cell with more time to repair the damage-preserving genome integrity in daughter cells before they continue the division cycle (Little and Nagasawa, 1985;Lane, 1992). Since there is a huge gap in our knowledge concerning the effects of the ECM on radiation-dependent cell cycle arrest, the presented data give first evidence that the ECM protein FN enables tumour and, especially, normal cells to delay for a longer period of time in the G2 phase possibly promoting DNA repair and thus survival. However, comparing basal and radiation-altered cell cycle distributions, the proportional differences of the G2 cell amounts on polystyrene or FN suggest a pronounced FN-related change in the cellular responsiveness of the lung fibroblasts rather than the lung cancer cells to irradiation. This effect resulted in an increase of survival that is likely to be in part because of an optimised DNA repair machinery. In contrast, A549 cells possibly posses specific genomic mutations that provide a survival advantage independent of an FN-mediated overproportional G2/DNA repair phase.
While the mitogenesis signal involves the activation of the rasraf-mitogen-activated protein kinase (MAPK) and PI3-K signalling pathways (Chen et al, 1994;Dedhar and Hannigan, 1997), the pathways participating in integrin-mediated regulation of survival and differentiation are not well characterised but have also implicated PI3-K, MAPK and PKB/Akt (Khwaja et al, 1997). Especially, the cell cycle seems to be proceeded by a PKB/Aktindependent mechanism involving GSK-3, b-catenin or AP-1 transcription factor (Radeva et al, 1997;Troussard et al, 1999). Using the specific PI3-K inhibitors LY294002 and wortmannin, we were able to show that both radiation-and adhesion-mediated activation of the ILK-PKBa/Akt-GSK-3b pathway were partly PI3-K-dependent. In fact, we provide evidence for a radiationinducible, PI3-K-independent ILK-GSK-3b signalling pathway in the case of cell -FN contact. With regard to cyclin D1 expression and pRb phosphorylation at PI3-K inhibition, the protein patterns showed only slight changes when cells were plated on FN. In contrast, the PKBa/Akt cascades indicate that not they are susceptible to either ILK or radiation without PI3-K support. These findings strongly suggest that only the combination of cell anchorage to ECM plus growth factors is able to modulate cellular responsiveness to ionising radiation most effectively and, thus, suggesting a tight convergence between these pathways already taking place upstream at the cytoplasmic face of the cell membrane.
The results of this study identify ILK, PKBa/Akt and GSK-3b within the critical b1-integrin pathway as important molecular factors to regulate individual cellular radiosensitivity in the presence of an ECM. We could uncover a novel radiationinducible, PI3-K-independent and membrane-located pathway via ILK and GSK-3b. Basal cell cycle transition as well as the radiation-induced G2 arrest was intensively altered by FN in a PI3-K-dependent manner. On the basis of great similarity of the results generated in the human lung cancer cells A549 and the normal human lung fibroblasts CCD32, most of the cell adhesiontransduced resistance mechanisms seem to work independent of the genetic and differentiation status of the cell. The detailed identification of the molecular mechanisms will possibly provide Schematic diagram of how integrin-linked kinase (ILK) and glycogen synthase kinase-3b (GSK-3b) might be involved in the acute radiation response with regard to cell survival, cell cycle progression and initiation of G2 phase arrest. Binding of cells to extracellular matrix (ECM) components via b1-integrins stimulates ILK and downstream targets protein kinase Ba/Akt (PKBa/Akt) (phosphorylation at amino-acid residues Ser473 and Thr308) and GSK-3b (phosphorylation at amino-acid residue Ser9). These events suppress apoptosis and promote survival by inhibiting Bad and caspase-9 and cell cycle transition by blocking proteolysis of cyclin D1. Facilitating growth factor binding to growth factor receptors (GFR) activates similar pathways downstream of the central regulator phosphatidylinositol-3 kinase (PI3-K). In bold letters, arrows and circles, we suggest and thereby support the hypothesis of direct ILK phosphorylation of GSK-3b when PI3-K is inhibited and cells are attached to ECM. Irradiation (IR) is able to activate this pathway, which then does not stimulate proliferation but rather blocks cells in the G2 phase possibly allowing damage repair.
considerable insight into the understanding of cell adhesionmediated drug and radioresistance, cell -ECM-interactions and tumour growth with respect to the modulation of multiple cellular network convergence by the microenvironment.