Pharmacological implications from the adhesion‐induced signaling profiles in cultured human retinal pigment epithelial cells

Abstract

Extracellular matrix (ECM) plays an active and complex role in regulating cellular behaviors, including proliferation and adhesion. This study aimed at delineating the adhesion‐induced signaling profiles in cultured human retinal pigment epithelium (RPE) cells and investigating the antiadhesion effect of antiproliferative drugs in this context. RPE R‐50 cells grown on various ECM molecules, such as type I and IV collagens, fibronectin, and laminin, were used for adhesion assay and for examining the phosphorylation profiles of signaling mediators including Akt, extracellular signal‐regulated kinase (ERK) 1/2, and integrin‐linked kinase (ILK) using Western blotting. The cells receiving antiproliferative drug treatment at subtoxic doses were used to evaluate their antiadhesive and suppressive effects on kinase activities. ECM coating enhanced adhesion and spreading of RPE cells significantly. The cellular attachment onto ECM‐coated surfaces differentially induced Akt, ERK1/2, and ILK phosphorylation, and concomitantly increased p53 phosphorylation and cyclin D1 expression, but decreased Bcl‐2/Bax ratios. Treatment with antiproliferative agents, including 5‐fluorouracil, mitomycin C, and daunomycin, at subtoxic doses suppressed the ability of RPE cells to adhere to ECM substratum significantly. This suppression was in part mediated through reduction of integrin β1 and β3 expressions and interfering Akt‐ILK signaling activity. Mechanistically, blockade of PI3K/Akt signaling resulted in the suppressed adhesion of RPE cells to ECM. These findings support the hypothesis that, in addition to their antimitogenic effect, antiproliferative agents also exhibit suppressive effect on the adhesiveness of cultured RPE cells. Moreover, inhibitors of the PI3K/Akt signaling mediator can potentially be used as therapeutic agents for proliferative vitreoretinopathy.

Introduction

Proliferative vitreoretinopathy (PVR) is the most common cause for failure of retinal detachment surgery [1], and involves uncontrolled proliferation of non‐neoplastic cells capable of forming cellular membranes on either side of the retina or along the detached surface of vitreous gel [2]. Etiologically, either previous retinal surgery, vitrectomy, preoperative PVR, large retinal tears with exposed retinal pigment epithelium (RPE), or preoperative choroidal detachment are the major causes of surgical failure [3]. Histological observations on surgically excised membranes reveal that RPE cells are one of the major cellular components involved in the pathogenesis of PVR [[4], [5]]. RPE cells activated by proinflammatory cytokines at sites of injured tissues migrate into the vitreous cavity and generate fibrogenic mediators that remodel biosynthesis of the extracellular matrix (ECM) [6].

In addition to the cellular elements, PVR membranes consist of abundant ECM molecules, primarily collagen type I (CoI), collagen type IV (CoIV), fibronectin (FN), laminin (LM), and various glycosaminoglycans [7]. ECM plays an active role in regulating morphogenesis of cells, and multifunctionally influences their migration, proliferation, and metabolic function [[8], [9]]. Contraction of PVR membranes creates a tractional force, which can distort or detach the retina [10]. Cell–ECM interaction has been demonstrated to be mediated by the cell surface adhesion receptor integrin. Integrin molecules are a group of highly homologous transmembrane proteins, consisting of a large extracellular domain, a hydrophobic membrane‐spanning domain, and a smaller cytoplasmic domain. Each integrin contains one α and one β subunit. Upon ligation to ECM, the intracellular domain of integrin β subunit recruits an integrin‐linked kinase (ILK) that conveys external stimuli into nuclei through activation of downstream kinase substrates, including Akt [11]. Tuning of the integrin–ILK axis signal cascade by phosphoinositide 3‐kinase (PI3K) has been emphasized in the epithelial mesenchymal transdifferentiation of tumors [12] and RPE cells [13]. In fact, integrin immunoactivity has long been noted in PVR membranes [14]. Recent evidence reveals that integrin α5β1 mediates adhesion, migration, and proliferation of RPE cells [15], and that functional inhibition of integrin suppresses dedifferentiation of tumor cells [16] and adhesion of RPE cells to ECM substratum [17]. Similarly, targeting of ILK with a small interfering RNA suppresses PVR development through inhibition of attachment, spreading, migration, and proliferation of human RPE cells [13], highlighting the significance of integrin‐related signaling in regulating RPE cell behaviors.

Seeding of human RPE cells onto ECM represents a suitable system to study the mechanisms underlying cell–ECM interactions. This study aimed at delineating the profiles of Akt, ERK1/2, and ILK phosphorylation in cultured human RPE cells seeded onto various ECM‐coated surfaces. We also found that antiproliferative drugs, including 5‐fluorouracil (5‐FU), mitomycin C (MMC), and daunomycin (DM), prevented the adhesion of RPE cells via suppression of the critical signaling pathway.

Materials and methods

Reagents

Antiproliferative drugs including 5‐FU, MMC, and DM were purchased from Sigma‐Aldrich (St. Louis, MO, USA). Ham's F12 nutrient mixture (F12 medium), fetal bovine serum, L‐glutamine, trypsin–EDTA, and antibiotics were obtained from Invitrogen/Gibco BRL (Gaithersburg, MD, USA). The materials used for coating ECM were bovine serum albumin (BSA, Sigma‐Aldrich), CoI (extracted from rat tail tendon as previously described) [18], CoIV (Southern Biotech, Birmingham, AL, USA), FN (Biomedical Technologies, Stoughton, MA, USA), and LM nonapeptide (Merck Calbiochem, San Diego, LA, USA). Kinase‐selective inhibitors, including wortmannin for PI3K, SB203580 for p38 mitogen‐activated protein kinase (p38 MAPK), PD98058 for MAPK/extracellular signal‐regulated kinase (ERK) kinase 1 (MEK1), and SP600125 for c‐Jun N‐terminal kinase (JNK), were purchased from Sigma‐Aldrich. All kinase inhibitors were stocked in Dimethyl sulfoxide (DMSO) at a concentration of 10 mM, and stored at –20°C. Antibodies against β‐Actin, P53, phospho‐P53 (Ser15), and integrin β1 and β3 were purchased from Santa Cruz Biotechnology (Santa Cruz, CA, USA), those against P21 and Bcl‐2 from Millipore (Billerica, MA, USA), and antiproliferating cell nuclear antigen (anti‐PCNA) from BD Biosciences (San Jose, CA, USA). Antibodies against Bax, cyclin D1, the phosphorylated form of Akt (Ser473), ILK (Ser246), and ERK1/2 (Thr185/Tyr187) were obtained from Cell Signaling Technology (Danvers, MA, USA).

Cell culture and treatment

A human RPE cell line (clone R‐50, a gift from Dr Dan‐Ning Hu, New York Eye and Ear Infirmary, New York Medical College, New York) was primarily isolated from a donor eye using the previously described method [19]. R‐50 cells were grown in an F‐12 medium with supplements as previously described. The subtoxic doses used for 5‐FU, MMC, and DM treatment were 100 ng/mL, 10 ng/mL, and 10 ng/mL, respectively [20].

ECM coating and cell adhesion assay

Well surfaces of cell culture plates (Corning Inc., Corning, NY, USA) were coated with CoI, CoIV, FN, or LM at a final density of 10 μg/cm² in phosphate‐buffered saline (PBS) for 16 hours at 4°C. The plates were further masked with 1% BSA/PBS for 2 hours at 4°C prior to plating the cells. To examine the effects of antiproliferative drugs and kinase inhibitors on RPE adhesiveness, the adhesion assay was performed by seeding 1 × 10⁵ cells/well into 24‐well plates, as described previously [20]. Negative controls (cells without drug treatment or all seeded cells without removal) were regarded as 100%.

Western blotting

After drug treatment, cells were washed once with ice‐cold PBS and lysed in prechilled Radio‐Immunoprecipitation Assay (RIPA) buffer (25 mM Tris‐HCl, pH 7.6; 150 mM NaCl; 1% NP‐40; 1% sodium deoxycholate; and 0.1% SDS). Equal amounts (40 μg) of protein extracts were loaded and separated by sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS‐PAGE), followed by electrotransferring, blocking, and immunodetecting procedures as described previously [20]. Densitometrical analyses were performed using ImageJ image analytical software (National Institutes of Health, Bethesda, MD, USA). Relative protein expression levels were shown by the density ratios of the protein of interest to β‐Actin in the same sample. All results were from three independent experiments and normalized as percentage or induction fold of (negative or BSA) controls.

Statistics

All results are expressed as the mean ± standard deviation (SD). Two‐sample Student t test and ANOVA were used for statistical analysis. A p value of <0.05 was considered statistically significant.

Results

Attachment kinetics of RPE cells

To compare the differential kinetics of RPE cell adhesion to surfaces coated with ECM molecules, i.e., CoI, CoIV, FN, and LM, an adhesion assay was performed and attachment kinetic curves were plotted. Not surprisingly, all ECM coatings, including FN, LM, CoIV, and CoI, strengthened the adhering ability of RPE cells significantly compared to BSA control (Fig. 1A). However, among these coating materials, CoI was a relatively weak enhancer for RPE cell adhesion. Based on our previous observation [20], 30 minutes after seeding was suggested to be the optimal time point for observing adhesion of normal RPE cells to ECM‐coated dishes at around 50% adhesion rate normalized to the total seeded cell number. In parallel, microscopic observation of cell morphology at 60 minutes after seeding revealed that the cells seeded on FN exhibited the best spreading morphology compared to those seeded on other ECM coatings (Fig. 1B).

Figure 1.

Adhesion kinetics and morphology of cultured human RPE cells seeded on different ECMs. (A) An adhesion assay was performed by seeding human RPE R‐50 cells on the 24‐well plates precoated with CoI, CoIV, FN, LM, or BSA (as a negative coating control). The percentage of cell attachment was shown as mean ± SD from three independent experiments. (B) The morphology of the RPE R‐50 cells adhered to the ECM‐coated surfaces was documented at 60 minutes after cell seeding. Bar = 50 μm. *A p value of <0.05, compared with respective BSA control (Fig. 1A). BSA = bovine serum albumin; CoI = collagen type I; CoIV = collagen type IV; ECM = extracellular matrix; FN = fibronectin; LM = laminin; RPE = retinal pigment epithelium; SD = standard deviation.

Signaling profiles in the adherent RPE cells

Because the activities of Akt and ILK have been implicated in the regulation of RPE cell adhesion [21], we next delineated the profiles of Akt, ERK, and ILK phosphorylation during the adhering process. RPE R‐50 cells were seeded onto different ECM‐coated surfaces at different time points for observing the kinetics of Akt, ERK, and ILK phosphorylation (Fig. 2A and B). Western blotting showed that Akt phosphorylation was elevated significantly in the cells seeded onto CoIV‐, FN‐, and LM‐coated dishes at 1 hour and diminished rapidly at 3 hours thereafter; however, this elevation was unseen in CoI‐coated group (Fig. 2C). ERK1/2 phosphorylation was induced prominently at 1 hour after seeding onto all ECM‐coated surfaces and abolished totally at 3 hours (Fig. 2D). Similarly, ILK phosphorylation increased significantly at 3 hours in the adherent cells of all ECM‐coated groups and declined gradually after 6 hours of attachment (Fig. 2E).

Figure 2.

Kinetics of Akt, ERK1/2, and ILK phosphorylation in cultured human RPE cells seeded on different ECMs. (A, B) Human RPE R‐50 cells were seeded onto the surfaces precoated with CoI, CoIV, FN, LM, or BSA (as a negative coating control). Protein lysates of S and adherent cells were subjected to Western blotting detection for Akt, ERK1/2, and ILK phosphorylation. Phosphorylation densities of (C) Akt, (D) ERK1/2, and (E) ILK were expressed as mean ± SD from three independent experiments. *A p value of <0.05, compared with respective BSA control. BSA = bovine serum albumin; CoI = collagen type I; CoIV = collagen type IV; ECM = extracellular matrix; ERK = extracellular signal‐regulated kinase; FN = fibronectin; ILK = integrin‐linked kinase; LM = laminin; RPE = retinal pigment epithelium; S = suspension; SD = standard deviation.

ECM modulates cell cycle and apoptotic regulators in RPE cells

To examine whether binding of cells to ECM affects the cell cycle progression and apoptotic program in RPE cells, R‐50 cells seeded and grown on various ECM‐coated surfaces for 24 hours were subjected to Western blotting for measuring cell cycle and apoptotic regulators. The tumor suppressor P53 is known to mediate apoptotic induction and senescence of RPE cells [22]. As shown in Fig. 3A, phosphorylation of P53 at serine 15 elevated significantly in RPE cells adhered to the surfaces coated with CoIV, FN, and LM, compared with those adhered to BSA and suspension controls (Fig. 3B). Total P53 levels in the cells grown on CoI, CoIV, FN, or LM did not change compared with BSA control. The CDK inhibitor P21, a negative regulator of G1/S phase transition, is a transcriptional target gene of P53 [23]. P21 protein levels increased significantly in RPE cells adhered to FN‐ and LM‐coated (but not CoI‐ and CoIV‐coated) surfaces compared to BSA control (Fig. 3A and C). Notably, the content of cyclin D1, whose activity is required for cell cycle transition and RPE cell proliferation [24], also increased prominently in the cells seeded onto CoIV‐, FN‐, and LM‐coated dishes (Fig. 3D). Attachment to CoIV, FN, and LM increased intracellular PCNA levels remarkably (Fig. 3E). In the aspect of apoptotic regulation, levels of the antiapoptotic Bcl‐2 protein increased in the CoI‐ and CoIV‐coated groups, while that of the proapoptotic Bax protein increased in the FN‐ and LM‐coated groups (Fig. 3F). As a consequence, Bcl‐2‐to‐Bax ratios were upregulated remarkably in the cells of CoI‐ and CoIV‐coated groups, but downregulated in those grown on FN and LM (Fig. 3G).

Figure 3.

The effect of ECM on the expression of cell cycle and apoptotic regulators in cultured human RPE cells. (A, F) Human RPE R‐50 cells were seeded onto the surfaces precoated with CoI, CoIV, FN, LM, or BSA (as a negative coating control). Protein lysates of S and adherent cells collected at 24 hours after seeding were subjected to Western blotting detection for cell cycle and apoptotic regulators. Band densities of (C) P21, (D) cyclin D1, and (E) PCNA as well as ratios of (B) phospho‐P53 to total P53 and (G) Bcl‐2 to Bax were plotted and expressed as mean ± SD from three independent experiments. *A p value of <0.05, compared with respective BSA control, which were normalized as 100%. BSA = bovine serum albumin; CoI = collagen type I; CoIV = collagen type IV; ECM = extracellular matrix; FN = fibronectin; LM = laminin; RPE = retinal pigment epithelium; PCNA = proliferating cell nuclear antigen; S = suspension; SD = standard deviation.

Antiproliferative agents suppress RPE cell adhesion

Our previous study had characterized the IC50 values of antiproliferative drugs in cultured human RPE cells [25]. To further determine whether these antiproliferative drugs, including 5‐FU, MMC, and DM, influence the adhesion of R‐50 cells to ECM, the cells were exposed to the drugs at subtoxic doses for 24 hours prior to adhesion assay. Results showed that 5‐FU exhibited the weakest efficacy in inhibiting RPE cell adhesion because it reduced significantly the adhesiveness of cultured RPE cells to CoIV‐coated (Fig. 4B), but not to CoI‐ (Fig. 4A), FN‐ (Fig. 4C), and LM‐ coated (Fig. 4D) surfaces. In comparison, MMC suppressed RPE cell adhesion prominently under all circumstances, except under LM‐coated microenvironment (Fig. 4D). Notably, DM pretreatment suppressed the adhesiveness of RPE cells to all ECM coatings significantly.

Figure 4.

The effect of antiproliferative drugs on the adhesiveness of cultured human RPE cells grown on different ECMs. Human RPE R‐50 cells were pretreated for 24 hours with various antiproliferative drugs (i.e., 5‐FU at 100 ng/mL, MMC at 10 ng/mL, and DM at 10 ng/mL) and subjected to adhesion assay by seeding on the dishes precoated with (A) CoI, (B) CoIV, (C) FN, or (D) LM for 30 minutes. An NC without drug treatment was considered as 100%. Data were expressed as mean ± SD from three independent experiments. *A p value of <0.05, compared with NC. CoI = collagen type I; CoIV = collagen type IV; DM = daunomycin; ECM = extracellular matrix; FN = fibronectin; 5‐FU = 5‐fluorouracil; LM = laminin; MMC = mitomycin C; NC = negative control; RPE = retinal pigment epithelium; SD = standard deviation.

Antiproliferative agents affect cell cycle and apoptotic machineries

Because recruitment of ILK to the cytoplasmic domain of receptor integrin β, including β1 and β3 subunits, has been demonstrated to be one of the critical events in the regulation of RPE cell adhesion [[15], [26], [27]], we examined the expression levels of integrin β1 and β3 in the R‐50 cells grown on different ECM substrata. Western blotting data showed that the expression of integrin β1 and β3 was induced significantly after 24 hours of adherence to the CoIV‐, LM‐, and FN‐coated surfaces compared to suspension control (Fig. 5A). To determine whether antiproliferative agents influence the integrin β1 and β3 expression levels directly, the cells grown on either BSA or FN were treated with drugs at subtoxic doses for 24 hours. Western blotting indicated that MMC and DM reduced the amount of integrin β3 significantly, whereas only DM suppressed integrin β1 abundance in both BSA and FN groups (Fig. 5B). To further determine whether these agents affect cell cycle and apoptotic machineries, Western blotting again demonstrated that, under both coating conditions, MMC and DM induced remarkable phosphorylation of P53, although P21 levels in all drug‐treated groups were elevated slightly (Fig. 5C). Moreover, it is noteworthy that the phosphorylated P53 level in the cells of negative control with FN coating was the same as in BSA control, implicating that the adhesion‐induced P53 phosphorylation shown in Fig. 3 might decline to constitutive level after 48 hours of cultivation. In terms of apoptotic regulation, only DM treatment reduced Bcl‐2 contents prominently and, overall, the Bcl‐2‐to‐Bax ratios in all groups were suppressed significantly under both coating conditions (Fig. 5D).

Figure 5.

The effect of ECM on integrin molecule expression and the modulatory effects of antiproliferative agents on apoptotic regulators in cultured human RPE cells. (A) Human RPE R‐50 cells were seeded onto the surfaces precoated with CoI, CoIV, FN, LM, or BSA (as a negative coating control). Protein lysates at 24 hours after seeding were subjected to Western blotting detection for integrin β1 and β3 expressions. (B) R‐50 cells were grown on either BSA‐ or FN‐coated surfaces for 24 hours, followed by treatment with various antiproliferative drugs (i.e., 5‐FU at 100 ng/mL, MMC at 10 ng/mL, and DM at 10 ng/mL) for another 24 hours. (C) Collected protein lysates were subjected to Western blotting detection for the expression of integrin β1 and β3 as well as that of cell cycle and apoptotic regulators. (D) Bcl‐2‐to‐Bax density ratios were shown as mean ± SD from three independent experiments. *A p value of <0.05, compared with NC. ^#A p value of <0.05 between groups. BSA = bovine serum albumin; CoI = collagen type I; CoIV = collagen type IV; DM = daunomycin; ECM = extracellular matrix; FN = fibronectin; 5‐FU = 5‐fluorouracil; LM = laminin; MMC = mitomycin C; NC = negative control; RPE = retinal pigment epithelium; SD = standard deviation.

Antiproliferative agents modulate integrin‐related signaling activities

Because the integrin signaling has been shown to convey the FN‐driven inward stimulus into RPE cells [21], we next sought to answer whether antiproliferative drugs also affect the integrin‐mediated signaling activity. The R‐50 cells grown on either BSA or FN received a short‐term drug treatment, followed by examination of the phosphorylation of Akt, ILK, and MAPKs (Fig. 6A). Western blotting and subsequent densitometrical analyses indicated that all three antiproliferative drugs significantly downregulated constitutive Akt phosphorylation under BSA‐coated conditions, whereas 5‐FU and MMC, but not DM, suppressed Akt phosphorylation levels under the FN‐coated environment after receiving 1 hour of drug treatment (Fig. 6B). Interestingly, 5‐FU suppressed ERK1/2 phosphorylation, while MMC and DM enhanced ERK1/2 phosphorylation dramatically under both coating circumstances (Fig. 6C). Moreover, MMC and DM decreased ILK phosphorylation levels significantly in the R‐50 cells of BSA control, but none of the drugs influenced ILK phosphorylation levels in those of FN‐coated groups (Fig. 6D).

Figure 6.

Effects of antiproliferative agents on constitutive phosphorylation of Akt, ERK1/2, and ILK. (A) Human RPE R‐50 cells grown on either BSA‐ or FN‐coated surfaces for 24 hours were treated with various antiproliferative drugs (i.e., 5‐FU at 100 ng/mL, MMC at 10 ng/mL, and DM at 10 ng/mL) for 1 hour, followed by Western blotting detection. Band densities of phosphorylated (B) Akt, (C) ERK1/2, and (D) ILK were normalized to corresponding Actin levels. Data were shown as mean ± SD from three independent experiments. *A p value of <0.05, compared with drug treatment NC. BSA = bovine serum albumin; DM = daunomycin; ECM = extracellular matrix; ERK = extracellular signal‐regulated kinase; FN = fibronectin; 5‐FU = 5‐fluorouracil; ILK = integrin‐linked kinase; MMC = mitomycin C; NC = negative control; RPE = retinal pigment epithelium; SD = standard deviation.

Blockade of PI3K/Akt signaling reduces RPE cell adhesion

To specify the interconnectionship among signaling mediators including Akt, ILK, and MAPKs, RPE R‐50 cells grown on either BSA or FN received a short‐term treatment of various kinase inhibitors, followed by examination of the phosphorylation of each kinase (Fig. 7A). Western blotting analysis showed that a blockade of p38 MAPK activity suppressed Akt phosphorylation significantly but enhanced ERK1/2 phosphorylation under BSA‐coated condition (Fig. 7B). FN coating appeared to potentiate the enhancement of ERK1/2 phosphorylation through p38 MAPK inhibition. Conversely, FN reversed the SP600125‐, PD98059‐, and wortmannin‐suppressed ERK1/2 phosphorylation significantly (Fig. 7C). Although treatment with all kinase inhibitors suppressed ILK phosphorylation in the presence of FN effectively, suppression of MEK activity using PD98059 upregulated ILK phosphorylation dramatically under a non‐FN‐coating condition (Fig. 7D). To further examine the effect of kinase suppression on cell adhesion, RPE R‐50 cells grown on either BSA or FN were treated with various kinase‐selective inhibitors and subjected to adhesion assay. Adhesiveness of RPE cells under both scenarios was effectively suppressed only by PI3K activity blockade (Fig. 7E and F).

Figure 7.

Effects of kinase inhibitors on constitutive levels of Akt, ERK1/2, and ILK phosphorylation in cultured human RPE cells and their impact on cell adhesion. (A) Human RPE R‐50 cells grown on either BSA‐ or FN‐coated surfaces were treated with various kinase inhibitors for 1 hour, including 10 μM WM, PD, SB, SP, or equivalent DMSO as a solvent control. Protein lysates were collected for Western blotting. (B) Alternatively, R‐50 suspension cells were pretreated with various kinase inhibitors for 1 hour and subjected to adhesion assay under BSA‐ or FN‐coated conditions after 30 minutes of attachment. A control without cell removal (all cells seeded) was considered as 100%. Data were expressed as mean ± SD of adhesion suppression from three independent experiments. *A p value of <0.05, compared with NC. ^#A p value of <0.05 between groups. BSA = bovine serum albumin; ERK = extracellular signal‐regulated kinase; FN = fibronectin; ILK = integrin‐linked kinase; PD = PD98059; RPE = retinal pigment epithelium; SB = SB203580; SD = standard deviation; SP = SP600125; WM = wortmannin.

Discussion

The present study is the first to delineate closely the profiles of attachment‐induced Akt, ERK, and ILK phosphorylation in the RPE cells seeded on different ECM substrata. The constitutively low phosphorylation of Akt, ERK, and ILK in suspended RPE cells was found to be upregulated dramatically on being adhered to the ECM‐coated surfaces (Fig. 2). Kinetic profiles of signaling mediators strongly suggest that the environment enriched with pathogenic ECM molecules, including CoIV, FN, and LM, is responsible for the activation of RPE cells. In this regard, FN and LM have been found to be the most prominent noncollagenous ECM overproduced in diseased status [28]. In the pathogenic context of PVR, vitreous FN has been shown to initiate epiretinal membrane formation [26], to enhance TGF‐β2‐induced α‐smooth muscle actin expression in the transdifferentiated RPE cells [29] and to stimulate RPE cell migration [30]. In conjunction with LM, FN also plays a pathogenetic role in the structural arrangement of newly formed capillaries in Proliferative diabetic retinopathy (PDR) [27]. In line with those findings in tumor cells [[11], [12]], the present study demonstrated that both ECMs induce ILK activity in cultured human RPE cells (Fig. 2).

The significance of integrin upregulation in PVR pathogenesis has long been addressed. Contraction of PVR membranes is known to involve the intercellular tractional force initially derived from upregulated integrin molecules, whereby mobilization of intracellular cytoskeletons is increased through activation of the signaling cascades downstream of integrins, including attachment‐induced Akt and ILK activities [14]. In addition, the regulatory role of integrin‐mediated Akt and ILK signaling in cell proliferation has been noted in ovarian cancer [16] and RPE cells [13]. More recently, a P53‐P21 axis signaling activity has been implicated in the resistance of tumor cells to the integrin β1 gene silencing‐induced apoptosis [31]. Intriguingly, the anchorage‐dependent P53 phosphorylation evidenced in the adherent RPE cells (Fig. 3B) may underlie the higher resistance of RPE cells to the stress‐induced apoptosis. On the other hand, a P53‐dependent G1 growth arrest can be induced, irrespective of P21‐mediated cyclin D1 induction [32]. Because the production of cyclin D1 in proliferative RPE cells is known to be regulated under the control of Ras/Raf/MEK/ERK pathway [24], upregulation of cyclin D1 (Fig. 3D) and PCNA (Fig. 3E) seen in the ECM‐coated groups (e.g., CoIV, FN, and LM) also suggested that the attachment of RPE cells to ECMs might trigger relevant signaling cascades, thereby giving rise to a high propensity to proliferate or, at least, to better resistance against external stimuli. Due to the well‐known and opposite functions of P21 and PCNA in cell cycle regulation, a seeming discrepancy may arise from simultaneous upregulation of P21 and PCNA in the RPE cells grown on ECM‐coated surfaces (Fig. 3). In fact, the CDK inhibitor P21 is known to bind physically to PCNA, thereby interfering with the complex formation of PCNA with DNA polymerase δ and inhibiting DNA replication [[33], [34]]. Conversely, PCNA also drives P21 degradation via a PCNA‐dependent E3 ubiquitin ligase activity [[34], [35]]. Moreover, PCNA plays roles not only in DNA replication, but also in DNA damage repair in a P53‐dependent manner [36]. Because P53 phosphorylation was also seen in those cells seeded on ECMs, the simultaneously upregulated levels of both P21 and PCNA raised the possibility that the DNA repairing machinery was alerted therein. However, this speculation awaits further investigation.

Therapeutic mechanisms of antiproliferative agents in terms of PVR pathogenesis have widely been evaluated in cultured human RPE cells. Our previous studies documented that 5‐FU, MMC, and DM exhibited prominent cytotoxicity in cultured RPE cells [25], and that 5‐FU and MMC induced a G1 phase arrest, while DM retarded RPE cell growth at G2/M phase [37]. Another report indicated that, in addition to their cytotoxic and cytostatic effects, these agents also gave rise to other pharmacological effects such as anticontraction [38]. To extend the evaluation to their antiadhesion effect, this study provides evidence showing that the treatment with antiproliferative agents at subtoxic doses inhibited the adhesiveness of cultured RPE cells to ECM substratum (Fig. 4). Among these drugs, DM appeared to be the best for interfering cell adhesion, because 5‐FU and MMC did not suppress the adhesiveness of the cells seeded onto LM‐coated surface. The pharmacological efficacy might be reflected by the fact that DM suppressed integrin β1 and β3 expression, reduced intracellular Bcl‐2 levels, and downregulated Akt and ILK phosphorylation significantly (5, 6), while 5‐FU and MMC also exhibited similar effects but to a lesser extent. Although DM did not suppress the constitutive Akt phosphorylation effectively in the RPE cells grown on FN (Fig. 6A and B), the discrepancy between the inability of DM to suppress Akt phosphorylation and the antiadhesive effect (Fig. 4C) may be mainly due to the different observation time points in two experimental settings. Besides, DM was suggested to be a weak Akt phosphorylation suppressor under the FN‐coated scenario, as compared to other antiproliferative drugs. Supporting our findings, functional inhibition of integrin receptors by a neutralizing antibody has been shown to block RPE cell adhesion [17], and even cellular proliferation and migration [15]. Moreover, ERK hyperphosphorylation was observed in the MMC‐ and DM‐treated RPE cells (Fig. 6). Although the MMC‐induced ERK phosphorylation is reportedly responsible for the chemokine overexpression in corneal fibroblasts [39], the underlying mechanism for the drug‐activated ERK signaling in RPE cells awaits further investigation. Taken together, antiproliferative agents may contribute to alleviating cellular adhesive tension through downregulation of the expression of surface integrin receptors, thereby desensitizing the responsiveness of downstream signaling cascades.

In the context of the signaling mediators involved in RPE cell–matrix adhesion, the detailed mechanism is obscure and rarely discussed. The ERK1/2 MAPK cascade has been implicated in the attachment, spreading, and migration of TNF‐α‐enhanced RPE cells on FN and CoI matrix [40]. Moreover, the PI3K/Akt signaling cascade is also involved in the RPE cellular contractility in collagen gels [41]. This study further provides a spectral screening of the signaling pathways that initiate RPE cell adhesion. Kinetic profiles of attachment‐induced phosphorylation of kinase mediators in the functionality of RPE adhesion may shed a new light on the development of antiadhesion therapeutic modalities for PVR and relevant ocular diseases. Our data strongly suggest that the pharmacological intervention in PI3K/Akt activity may result in a substantial suppression of RPE cell adhesion (Fig. 7), implicating that the inhibition of PI3K/Akt signaling activity may also be a useful strategy for the PVR treatment. Nevertheless, we cannot rule out the risks derived from this kinase intervention, such as induction of RPE cell apoptosis and early retinal degeneration, because Akt confers survival signal through regulation of antiapoptotic members of the Bcl‐2 protein family [21]. Additional experiments are required to better understand the pathophysiological role of integrin‐mediated signaling in the pathogenesis of PVR.

In summary, this study demonstrated that ECM coating enhanced RPE cell adhesion and spreading significantly. The attachment‐induced phosphorylation of Akt and ILK, anchorage‐dependent p53 phosphorylation, induced cyclin D1 upregulation, and decreased Bcl‐2‐to‐Bax ratios were all demonstrated in the adherent RPE cells. In addition to their antimitogenic effect, antiproliferative agents suppressed cell adhesion to ECM substratum significantly; the suppression was derived not only from the reduction of integrin β1 and β3 receptors, but also from their interference in Akt‐ILK signaling cascade. Mechanistically, the PI3K‐Akt signaling axis was shown to be one of the major determinants in the RPE cell adhesiveness. We thus conclude that, these antiproliferative agents and PI3K/Akt‐specific inhibitors can be potentially used for treating PVR.

Acknowledgments

This study was supported in part by grants from the National Science Council, Executive Yuan, Taiwan (NSC93‐2314‐B‐037‐031), Kaohsiung Medical University Hospital (KMUH100‐0R33), Kaohsiung Municipal Hsiao‐Kang Hospital (KMHK‐97‐011), and E‐Da Hospital, Kaohsiung, Taiwan (EDAHT‐100022).