Pigment Epithelium-derived Factor and Its Phosphomimetic Mutant Induce JNK-dependent Apoptosis and p38-mediated Migration Arrest*

Pigment epithelium-derived factor (PEDF) is a potent endogenous inhibitor of angiogenesis and a promising anticancer agent. We have previously shown that PEDF can be phosphorylated and that distinct phosphorylations differentially regulate its physiological functions. We also demonstrated that triple phosphomimetic mutant (EEE-PEDF), has significantly increased antiangiogenic activity and is much more efficient than WT-PEDF in inhibiting neovascularization and tumor growth. The enhanced antiangiogenic effect was associated with a direct ability to facilitate apoptosis of tumor-residing endothelial cells (ECs), and subsequently, disruption of intratumoral vascularization. In the present report, we elucidated the molecular mechanism by which EEE-PEDF exerts more profound effects at the cellular level. We found that EEE-PEDF suppresses EC proliferation due to caspase-3-dependent apoptosis and also inhibits migration of the EC much better than WT-PEDF. Although WT-PEDF and EEE-PEDF did not affect proliferation and did not induce apoptosis of cancer cells, these agents efficiently inhibited cancer cell motility, with EEE-PEDF showing a stronger effect. The stronger activity of EEE-PEDF was correlated with a better binding to laminin receptors. Furthermore, the proapoptotic and antimigratory activities of WT-PEDF and EEE-PEDF were found regulated by differential activation of two distinct MAPK pathways, namely JNK and p38, respectively. We show that JNK and p38 phosphorylation is much higher in cells treated with EEE-PEDF. JNK leads to apoptosis of ECs, whereas p38 leads to anti-migratory effect in both EC and cancer cells. These results reveal the molecular signaling mechanism by which the phosphorylated PEDF exerts its stronger antiangiogenic, antitumor activities.

tified in the retina; however, it has now become evident that it is expressed throughout the human body and is constantly present in the systemic circulation (3,4). Studies of the past decade have implicated PEDF in the pathophysiology of a wide range of angiogenesis-associated disorders, including diabetic complications (5), macular degeneration (6), and the growth of many types of solid tumors (7)(8)(9)(10). Furthermore, the natural antiangiogenic activity of PEDF, which is far greater than that of any other known endogenously produced factor (2,11), has turned it into a highly potent therapeutic agent for the inhibition of pathological neovascularization and, therefore, a promising tumor suppressor. To date, a number of studies have shown that exogenous administration of PEDF results in tumor regression, decreased intratumoral microvessel density, and prolonged survival in various animal cancer models (12)(13)(14)(15). The underlying molecular mechanism of anticancer activity of PEDF has been under extensive investigation recently (13,16). Yet, the signaling pathways that mediate the effects of PEDF are not fully delineated and its exact mechanism of action remains to be elucidated.
Previous studies by our group revealed that plasma PEDF is a phosphoprotein that can be phosphorylated by protein kinase CK2 on Ser 24 and Ser 114 and by protein kinase A on Ser 227 (17). Using double and triple phosphorylation site mutants, we have identified the physiological significance of these phosphorylations by demonstrating that variable phosphorylation states of PEDF differentially regulate its biological mode of activity (18). In these studies, mutants mimicking either protein kinase CK2, or the accumulative protein kinase CK2 and protein kinase A phosphorylation (S24E114E227A, EEA-PEDF and S24E114E227E, EEE-PEDF) exhibited much stronger antiangiogenic activity than their WT counterpart (17,18). More recently, we have also shown that phosphomimetic mutants of PEDF, particularly EEE-PEDF, are significantly more efficient than WT-PEDF in inhibiting tumor growth and neovascularization in human breast, colon cancer, and glioblastoma xenograft models (19). Immunohistochemical analysis revealed that PEDF and its mutants affect mainly tumor-residing ECs and prevent the formation of intratumoral vascular network by facilitating EC apoptosis. On the other hand, PEDF and its mutants did not affect survival of cancer cells, indicating that the antiangiogenic activity of these agents is the foremost element of the observed antitumor effect.
In the current study, we aimed to uncover mechanistic aspects of the enhanced antiangiogenic and anticancer activity of the phosphomimetic PEDF, by examining the physiological effects and signaling cascades induced by WT-PEDF and EEE-PEDF. We found that EEE-PEDF functions more potently than WT-PEDF in suppressing EC proliferation due to induction of caspase-3-dependent apoptosis and also in inhibiting migration of these cells. In addition, we demonstrate that despite having no effect on cancer cell proliferation, WT-PEDF and more so EEE-PEDF efficiently inhibit cancer cell migration. The enhanced EEE-PEDF effects are correlated to a better binding to laminin receptor (LR). The proapoptotic and antimigratory activities of WT-PEDF and EEE-PEDF are further shown to be differentially and independently regulated by JNK and p38 MAPKs. As compared with WT-PEDF, EEE-PEDF induces much stronger signaling via p38 and JNK, and we propose that this stronger signaling effect is the reason for the enhanced antiangiogenic activity of this PEDF mutant.

PEDF Production
WT-PEDF and EEE-PEDF (from human origin) were cloned from pBlueScriptSKII(ϩ)/PEDF, and pcDNA3/ EEE-PEDF (17,18), respectively into pRSET(A) (Invitrogen), and expressed in Escherichia coli BL21. Bacterial cells were grown at 30°C to A 600 ϭ 0.5-0.6, and the expression of re-combinant proteins was induced by 0.5 mM isopropyl 1-thio-␤-D-galactopyranoside for 4 -5 h. Pelleted bacterial cells were lysed in ice-cold nickel-nitrilotriacetic acid binding buffer (containing 300 mM NaCl, 50 mM NaH 2 PO 4 , 20 mM imidazole, pH 8.0) supplemented with 10 g/ml leupeptin, 1 mM PMSF, 10 mM ␤-mercaptoethanol, and 1 mg/ml lysozyme followed by sonication. Lysates were then cleared by centrifugation at 12,000 ϫ g at 4°C for 15 min. Batch purification of recombinant proteins was performed using ion metal affinity chromatography with nickel-nitrilotriacetic acid His-Bind resin (Novagen) according to the manufacturer's protocol. Elution fractions were resolved on SDS-PAGE followed by silver staining and immunoblotting with anti-PEDF antibody. The identity of recombinant WT-PEDF and its mutants was verified my mass spectroscopy. To remove the excess of imidazole, eluates that exhibited Ͼ 90% purity were dialyzed overnight at 4°C against PBS. Endotoxin content in final protein formulations was evaluated by LAL turbidimetric assay and was Ͻ2 enzyme units/g protein.

Immunoblotting (Western blotting)
Cells were grown to subconfluency and then serum-starved (0.1% FCS and 0.5% FCS for HUVEC) for 16 h. After incubation with indicated treatments, cells were rinsed twice with ice-cold PBS and scraped into radioimmune precipitation assay buffer (0.2 ml/plate). Extracts were then obtained by centrifugation at 15,000 ϫ g for 15 min at 4°C. Aliquots of cellular extracts were subjected to SDS-PAGE and transferred onto nitrocellulose membranes (Tamar, Israel) by electroblotting. Membranes were incubated overnight at 4°C with the corresponding primary antibody, followed by incubation with either horseradish peroxidase or alkaline phosphataseconjugated secondary antibody (Jackson ImmunoResearch laboratories). Membranes were developed either using EZ-ECL kit (Biological Industries, Kibbutz Beit Haemek, Israel) in Bio-Rad's ChemiDoc XRS imaging station or with nitro blue tetrazolium/5-bromo-4-chloro-3-indolyl phosphate solution (Roche Applied Science). Each experiment was performed at least three times to test for reproducibility and obtain statistically significant data. Quantification of the band intensities was performed using QuantityOne software (Bio-Rad).

Proliferation Assay
The effect of WT-PEDF and EEE-PEDF on cell proliferation was determined using methylene blue assay. For this purpose, cells were incubated for 48 h in 1% FCS-containing medium with the indicated amounts of recombinant proteins. Following treatment, cells were fixed in 4% paraformaldehyde at room temperature. Cells were then washed once in 0.1 M sodium borate buffer, pH 8.5, and, thereafter, incubated with 1% methylene blue in 0.1 M sodium borate buffer, pH 8.5, for 10 min. Excess of stain was washed out with double distilled water, and the stain was extracted with 0.1 M HCl (0.4 ml/well) for 2 h at room temperature with shaking. Thereafter, aliquots of each sample were transferred into 96-well plate, and A 595 was determined using an ELISA reader.

In Vitro Migration Assays
Transwell Assay-The migration of BAEC, HUVEC, and MDA-MB-231 cells was assayed in 24-well Transwell plates (Corning, NY). For ECs, the upper surface of the polycarbonate filters with 8-m pores was coated with 0.2% gelatin in PBS. A suspension of cells (5-8 ϫ 10 4 cells/100 l) was placed in the upper chambers. The lower chambers were filled with 600 l of the corresponding medium containing 20 ng/ml bFGF. Cells were allowed to migrate for 16 -20 h in the presence of different treatments. Thereafter, cells were removed from the upper compartment of the filter with a cotton swab. Cells that reached the lower surface of the filter were fixed with 3.5% paraformaldehyde, stained with 0.1% crystal violet, and images were then captured using a digital camera coupled with a microscope. Relative migration was quantified following stain extraction with methanol and A 540 measurement in an ELISA reader.
"Wound Healing" Assay-BAEC and MDA-MB-231 cells were grown to 95-100% confluence. Thereafter, a cell scratch spatula was used to create a fixed-width wound in a cell monolayer, followed by incubation with medium containing 1% FCS in the presence of bFGF and/or other treatments for 12-16 h. Wound closure was monitored by capturing the images using digital camera coupled with a microscope and following the advancement of the migrating front at defined time intervals.

Apoptosis Assay
Apoptosis was evaluated using a TUNEL kit (Roche Applied Science). After indicated treatments, cells grown on 18-mm coverslips were washed and fixed with 3% paraformaldehyde in PBS. Thereafter, apoptotic cells were identified using manufacturer's protocol (Roche Applied Science) and quantified. A DeltaVision OMX fluorescent microscope station supplied with a digital camera and SoftWorx software (Applied Precision) was used to process the slides.

In Vitro Binding Assay
LR and patatin-like phospholipase domain-containing protein 2 (PNPLA2) sequences were amplified from HUVECs cDNA using specific primers flanked by SalI (forward) and BamHI (reverse) restriction sites. To create GFP-LR and GFP-PNPLA2 constructs, amplicons were respectively cloned between SalI and BamHI sites of pEGFP-C1 (Clontech). Thereafter, COS-7 cells were transiently transfected with 0.5 g of either GFP-LR or GFP-PNPLA2 plasmids, using poly ethyl inositolamine. Forty eight hours after transfection, cells were lysed, and GFP-LR or GFP-PNPLA2 proteins were immunoprecipitated using anti-GFP antibody prelinked to A/G-agarose beads (Santa Cruz Biotechnology). Thereafter, immunoprecipitate-bound A/G beads were washed twice with ice-cold washing buffer containing 10 mM Tris, pH 7.4, 1 mM EDTA, 1 mM EGTA, pH 8.0, 150 mM NaCl, and 0.5% Triton X-100. Following the second wash, beads were resuspended in PBS and aliquoted. Each aliquot was combined with 4 g of recombinant GST-WT-PEDF or GST-EEE-PEDF in PBS and incubated for 4 h at 4°C with rotation. Thereafter, beads were washed three times with washing buffer, resuspended in 1ϫ sample buffer and boiled. Resolved proteins were analyzed by immunoblotting with anti-PEDF antibody. To confirm equal amounts of GFP-LR or GFP-PNPLA2, membrane was then reprobed with anti-GFP antibody.

Statistical Analysis
Data are expressed as mean Ϯ S.E. Statistical evaluation was carried out using functional analysis and Student's t test (two-tailed) to test for differences between the control and experimental results. Values of p Ͻ 0.05 were considered statistically significant.

Effect of WT-PEDF and EEE-PEDF on Endothelial and
Cancer Cell Proliferation-We have previously shown that when administered to mice bearing MDA-MB-231, HCT116, or U87-MG xenografts, WT-PEDF and its phosphomimetic mutant EEE-PEDF predominantly affect ECs of the tumor vasculature, rather than cancer cells themselves (19). Thus, we undertook to further study the mechanisms of action of WT-PEDF and of EEE-PEDF, which caused the most profound inhibition of xenograft growth, in cultured endothelial and cancer cells. First, we examined the effect of WT-PEDF and EEE-PEDF on proliferation of BAECs by methylene blue assay. We found that nanomolar concentrations of both PEDF constructs suppress their proliferation in a dose-dependent manner, where EEE-PEDF had a significantly stronger effect than WT-PEDF (Fig. 1A). On the other hand, the same concentrations of the PEDF constructs did not affect proliferation of MDA-MB-231 (Fig. 1B), HCT116 (Fig. 1C), and U87-MG (Fig. 1D), as well as HeLa, DU145, and MCF-7 (data not shown) cancer cells. These results suggest that the inhibitory effect of PEDF constructs on tumor growth is mostly indirect.
Effect of WT-PEDF and EEE-PEDF on Endothelial Cell Apoptosis-According to our previous results (19), the antiproliferative effect of PEDF toward ECs was associated with an enhanced EC apoptosis. To verify this, we first compared the rate of WT-PEDF-and EEE-PEDF-induced apoptosis by TUNEL assay. In these experiments, we observed a time-and dose-dependent induction of apoptosis in BAEC by both WT-PEDF and EEE-PEDF, where the apoptotic effect of EEE-PEDF was almost 3-fold higher than that of WT-PEDF in all tested concentrations (Fig. 2, A and B). These results were supported by a dose-dependent activation of caspase-3 upon treatment with both WT-PEDF and EEE-PEDF, as manifested by the appearance of 17-and 19-kDa caspase-3 cleavage products (Fig. 2C). In contrast to BAEC, WT-PEDF as well as EEE-PEDF failed to induce apoptosis of cultured MDA-MB-231 cells as evaluated by TUNEL assay over a 48-h period (19). For further information on the apoptotic effect, we next used a FACS-based cell cycle analysis. By this method, we found that a large fraction of cells treated with WT-PEDF and more so EEE-PEDF are in sub-G 1 phase (hypodiploid nuclei), without any significant arrest in G 1 or G 2 phases of the cell cycle (supplemental Fig. S1). This effect was completely abolished by the general caspase inhibitor Z-VAD, clearly confirming the apoptotic effect of the PEDFs.
Effect of WT-PEDF and EEE-PEDF on Endothelial and Cancer Cell Motility-It is well known that the establishment of tumor vascular network depends on the motility of actively proliferating ECs, and the migration of cancer cells generally serves as a parameter for the degree of their invasive behavior (20). Thus, the effect of the PEDF constructs on cell migration was examined in both BAEC and MDA-MB-231 cells. For this purpose, we first employed an in vitro Transwell assay, where chemoattractant-driven cell migration through the polycarbonate membranes in the presence of either WT-PEDF or EEE-PEDF was followed over 16 h. Both PEDF constructs dose-dependently inhibited bFGF-induced BAEC migration, and the ability of EEE-PEDF to attenuate cell migration was more substantial than that of WT-PEDF (Fig. 3, A and B). Notably, unlike their lack of effect on apoptosis, WT-PEDF and more so EEE-PEDF did affect the migration of MDA-MB-231 cells, although this effect was not as marked as in BAEC (Fig.  3, A and B). These results were corroborated using wound healing assay, in which a fixed-width scratch in a cell monolayer was created and the advancement of the migrating front was followed in the presence of either WT-PEDF or EEE-PEDF (Fig. 3C). Wound healing experiments performed in HCT116 and U87-MG cells showed similar results as with MDA-MB-231 (data not shown), confirming that the effect on cancer cells is general. Taken together, these observations indicate that EEE-PEDF is more efficient than WT-PEDF in inhibiting different angiogenic activities of cultured ECs, such as proliferation and migration, and also, to some extent, in the inhibition of cancer cell migration.
Enhanced EEE-PEDF Interaction with LR but Not with PNPLA2-Next, we studied the molecular signaling mechanism(s) by which EEE-PEDF exerts its stronger antiangiogenic activity. As a first step, we examined the ability of the PEDF constructs to bind to two putative receptors that were recently published, namely LR (21) and PNPLA2 (22). For this purpose, we transfected COS7 cells with either GFP-LR or GFP-PNPLA2. Forty eight hours after transfection, cells were lysed, and co-immunoprecipitation was performed with anti-GFP antibody. The immunoprecipitated proteins were briefly washed, incubated with recombinant WT-PEDF or EEE-PEDF, and subjected to immunoblotting that revealed that although EEE-PEDF and WT-PEDF interacted with PNPLA2 equally well, the binding of EEE-PEDF to LR was much stronger than that of WT-PEDF (Fig. 4). This result was confirmed using a GST pulldown that showed again a much better binding of EEE-PEDF to LR (supplemental Fig. S2). Interestingly, in a previous study (17), we found that WT-PEDF and EEE-PEDF bind to plasma membrane in similar affinities. In view of the relatively weak binding of PNPLA2 and the changes in binding to LR, this similar binding to membranes may indicate that the PEDF and its phosphomimetic mutants bind to additional proteins on the cell surface, which may lead to distinct functions. However, because LR seems to be the main PEDF receptor that leads to its antiangiogenic effects (21), it is likely that the effect of EEE-PEDF is mediated by this particular receptor and not by PNPLA2 or any of the other putative binding proteins.

Time-dependent Activation of Distinct MAPK Cascades by WT-PEDF and EEE-PEDF-Because
LR is probably not the only PEDF receptor and the mechanism of its action is not clear yet, it was important to elucidate the intracellular regulatory mechanisms that mediate the effects of WT-PEDF and EEE-PEDF in endothelial and cancer cells. Hence, we examined whether the PEDF constructs affect the activity of MAPK cascades, ERK, p38, and JNK, which are known to play an important role in survival, apoptosis, and cellular response to stress (23). In BAEC, EEE-PEDF induced marked and sustained elevation in p38␣ and JNK1/2 phosphorylation up to 270 min of incubation, whereas the activation of ERK1/2 peaked at 15 min and declined back to basal level within 120 min (Fig. 5, A-C). WT-PEDF had a similar trend of activation of all three MAPKs, although the fold changes of p38␣ and JNK1/2 phosphorylation were significantly lower than that observed with EEE-PEDF. The changes in p38␣ and JNK1/2 activity were preceded by the activation of the corresponding kinases at the MAP2K level (Fig. 5, D and E). WT-PEDF and more so EEE-PEDF induced noticeable phosphorylation of MKK3/6 and MKK4, known to be the respective activators of p38 and JNK (24) but almost did not affect phosphorylation of MKK7, which also acts upstream of JNK, though is not critical when MKK4 is activated (25). Interestingly, both WT-PEDF and EEE-PEDF did not affect JNK1/2, but did elevate to some extent p38␣ activity in MDA-MB-231 cells, with EEE-PEDF showing stronger effect than WT-PEDF (Fig. 5, F and H). In this cell line, known to harbor K-Ras activating mutation (26) and high basal levels of ERK1/2 activity, both PEDF constructs failed to induce further increase in ERK1/2 phosphorylation (Fig. 5, G and H). Addition of bFGF simultaneously with the PEDF treatment did not significantly change the phosphorylation of p38 and JNK in BAEC cells (supplemental Fig. S3). Furthermore, the short term AKT phosphorylation was affected neither by WT-PEDF nor by EEE-PEDF in the cell lines examined (supplemental Fig. S4). These results indicate that PEDF-induced signaling is not likely affected by, or interfering with angiogenic/proliferation/survival-regulating signal transduction pathways. Overall, these results indicate that WT-PEDF, and more so its phosphomimetic mutant, induces diverse intracellular signaling events in different cell lines and further support our previous findings on preferential susceptibility of ECs to PEDF action.
Differential Regulation of Apoptotic and Antimigratory Activities of WT-PEDF and EEE-PEDF by JNK and p38 MAPKs-Next, we undertook to examine whether the signaling events described above are involved in the apoptotic and the antimigratory activities of WT-PEDF and EEE-PEDF. For this purpose, we first followed the effects of the PEDF constructs in BAEC and MDA-MB-231 cells pretreated with selective pharmacological inhibitors of JNK1-3 (SP600125), p38␣/␤ (SB203580), or MEK1/2 (U0126). The specificity of p38␣/␤ and JNK1-3 inhibitors was validated in BAEC, where these agents, correspondingly and without interfering with effects of each other, inhibited WT-PEDF-and EEE-PEDF-induced phosphorylation of the respective p38␣ and JNK1/2 substrates MAPKAPK2 and c-Jun (supplemental Fig. S5A). Similarly, MEK1/2 inhibitor prevented WT-PEDF-and EEE-PEDF-induced ERK1/2, but not p38␣ or JNK1/2 phosphorylation (supplemental Fig. S5, B and C). We found that JNK1/2 appear to be the main executors of PEDF-induced apoptosis in BAEC, as pretreatment with JNK1-3 inhibitor almost completely attenuated the apoptotic effect of both PEDF constructs in these cells, whereas pretreatment with the p38␣/␤ inhibitor or the MEK1/2 inhibitor did not (Fig. 6, A and B). Also, JNK1-3 but not p38␣/␤ inhibition resulted in significant decline of WT-PEDF-and EEE-PEDFinduced caspase-3 cleavage (Fig. 6C). Conversely, the antimigratory activity of the PEDF constructs toward BAEC appears to be mediated by p38␣ because the p38␣/␤ inhibitor but not the inhibitors of the JNK1-3 and ERK1/2 cascades significantly reduced the ability of the PEDF constructs to inhibit BAEC migration (Fig. 7, A and B, and supplemental Fig. S6). Inhibition of p38␣/␤ produced a similar outcome in MDA-MB-231 cells, where only pretreatment with the p38 ␣/␤ inhibitor prevented the antimigratory effects of WT-PEDF and EEE-PEDF (Fig. 7, A and C, and supplemental Fig. S6). The results obtained above were confirmed by additional selective JNK (BI78D3) and p38 (SB202190 and RWJ67657) inhibitors, which gave very similar results to that obtained with SP600125 and SB203580, respectively (supplemental Figs. S7-S9) Thus, p38␣ and JNK1/2 appear to have distinct regulatory functions in controlling the antiangiogenic and antitumor activities of WT-PEDF and EEE-PEDF, where JNK1/2 executes EC apoptosis, whereas p38␣ is involved in the inhibition of endothelial and to some extent also of cancer cell migration.
PEDF has been previously reported to exert opposite effects on endothelial cells of distinct phenotypes cultured under different conditions (27). Therefore, we validated the observed effects in HUVECs. The HUVEC culturing capacity is known to be different from that of BAEC, and these cells represent a suitable model for young endothelial cells. In low passage HUVEC, WT-PEDF and EEE-PEDF induced signaling events similar to those observed in BAEC, i.e. activation of ERK1/2, p38␣, and JNK1/2, with EEE-PEDF causing much stronger effect on p38␣ and JNK1/2 than WT-PEDF (Fig. 8A). Pharmacological inhibition of WT-PEDF-and EEE-PEDFinduced JNK1/2 and p38␣ activities in HUVEC confirmed their respective roles in the induction of EC apoptosis (Fig. 8,  A and B) and the inhibition of EC migration (Fig. 8, C and D). These results clearly show that WT-PEDF-and EEE-PEDFinduced effects persist in EC populations of different phenotypes and, therefore, substantiate the generality of the observed effects toward ECs.

DISCUSSION
The initial isolation of PEDF was based on its ability to promote and support the growth of neuronal cells, and therefore, it was first described as a neurotrophic factor (1,28). Later on, it was found that PEDF is also a very potent endogenous inhibitor of angiogenesis, turning it into a promising agent for the inhibition of neovascularization-dependent tumor growth (2,12,29). The anticancer activity of PEDF has been suggested to involve both indirect and direct antitumor effects. Indirectly, reduction in tumor growth is achieved through the antiangiogenic action of PEDF, in which the selective targeting of newly-formed vasculature without harming the existing blood vessels is of a particular importance (30). This effect was suggested to involve inhibition of activity and/or expression of VEGF (4, 31) as well as the proapoptotic activity of PEDF toward immature and migrating ECs (4,31,32). Yet, in our previous study, which employed three different xenograft models to test the anticancer effect of WT-PEDF and its mutants, these agents did not affect VEGF expression and activity in vivo, implying direct action on tumor ECs as a major determinant of the observed anticancer activity (19). Though still controversial, it has been reported that PEDF also exerts a direct antitumor effect, possibly by inducing either antiproliferative or prodifferentiation activities toward cancer cells (31,33).
We have previously shown that physiological functions of PEDF are differentially regulated by phosphorylation and mimicking the fully phosphorylated state of PEDF significantly enhances its antiangiogenic activity (17,18). In a recent study, we have examined the antitumor activity of the phosphomimetic mutants of PEDF and found that EEE-PEDF mutant can serve much more potent anticancer agent than WT-PEDF (19). The observed antitumor effect of the phosphomimetic PEDF was associated with a more profound reduction in intratumoral microvessel density than that caused by its wild-type form. We have also found that inhibition of tumor neovascularization by WT-PEDF and its mutants occurs independently of VEGF down-regulation and is rather a result of their direct effect toward tumor ECs (19). Our present study shows that this direct effect involves inhibition of EC migration and induction of EC apoptosis, and that EEE-PEDF is much more efficient than WT-PEDF in inducing both these activities. In addition, we demonstrate that WT-PEDF and EEE-PEDF do not affect cancer cell proliferation but nevertheless inhibit cancer cell motility (see Fig. 9 for a model). Thus, we provide here an answer to the widely debatable question of whether PEDF exerts a direct effect on cancer cells and the nature of this effect. Our findings are in agreement with another study that showed lack of PEDF effect on proliferation of cultured HeLa cells but efficient eradication of HeLa xenografts in vivo (34). The lack of the proapoptotic effect of PEDF toward cancer cells could be explained by the higher basal activity of survival pathways or the presence of a distinct, less-specific receptor. However, existence of a putative PEDF receptor is still obscure. Overall, our results indicate that the antiangiogenic activity of PEDF  and its mutants, e.g. their ability to affect actively proliferating and migrating ECs is the main component of their antitumor action, although the antimigratory activity toward cancer cells is likely to contribute to such profound anticancer effect as well. Furthermore, the latter suggests that PEDF and more so, phosphomimetic PEDF may be beneficial for the inhibition of tumor invasiveness and metastatic potential. Thus, in contrast to exclusively antiangiogenic agents, PEDF exerts a combined antitumor effect, which may have important implications for its further development as an anticancer drug.
Our results indicate that PEDFs inhibit the proliferation of ECs but not cancer cells. As discussed in our previous article (19), the reason for this intriguing effect is not clear yet. One possibility to explain it, is the expression of different PEDF receptors, with distinct signaling machinery, in the different cells. However, since LR seems to be expressed in both types of cells (21), this is probably not the main differential mechanism. Nonetheless, it is still possible that different co-receptors are expressed in the different cells and those modulate the effects of PEDF. Another possibility is that cancer cells are resistant to PEDF-induced apoptosis by the induction of a higher basal activity of survival factors. One such a survival factor might be AKT, which exhibits much higher basal activity in MDA-MB-231 than in BAEC (supplemental Fig. S4 and data not shown). However, additional studies are needed to further clarify these points.
Detailed examination of the molecular mechanism by which WT-PEDF and EEE-PEDF exert their effects in endothelial and cancer cells revealed a regulation by differential activation of two distinct MAPK cascades. We found that JNK1/2 and p38␣ independently of each other, respectively, regulate apoptotic and antimigratory activities of WT-PEDF and its phosphomimetic mutant (Fig. 9), where a much stronger activation of p38␣ and JNK1/2 by EEE-PEDF results in the enhanced antiangiogenic and, therefore, anticancer properties of this mutant. The involvement of JNK1-3 in apoptosis has been extensively elaborated in a variety of cell systems (35), including apoptosis of ECs in response to PEDF (36) or other stimuli (37,38). Accordingly, p38 MAPKs have been implicated in the regulation of cell migration (39). However, our study here is unique in showing such distinct and independent activities of these two stress-related signaling cascades when simultaneously activated by the same agent.
We further show that WT-PEDF and EEE-PEDF also cause transient increase in ERK1/2 phosphorylation, which is not required for their apoptotic and antimigratory activities and does not interfere with these effects. Because the ERK1/2 cascade is known to induce survival in various cells (40), the molecular mechanisms that prevent such an effect in ECs treated with PEDF constructs are not clear but may be dependent on the decline of ERK1/2 phosphorylation before JNK1/2 and p38␣ reach their peak of activity. PEDF-induced ERK1/2 phosphorylation has been previously reported by us (18) and others that linked it to the cytoprotective properties of PEDF (41,42). Indeed, our previous report showed that WT-PEDF and EEE-PEDF exhibit the same degree of neurotrophic and neuroprotective activity (18). Accordingly, our present results indicate that WT-PEDF and EEE-PEDF induce similar levels of ERK1/2 phosphorylation.
A number of independent studies have demonstrated the involvement of different signaling pathways in the biological activity of PEDF. It has been shown, for example, that PEDFinduced apoptosis of ECs can be mediated by the activation of Fas-FasL system (32) and by peroxisome proliferator-activated receptor ␥ signaling to p53 (43). Furthermore, PEDF has been reported to block mitogenic effects of growth factors on ECs through the modulation of PI3K/AKT pathway (44), which is known to be critical for EC survival. Modulation of MAPK signaling module by PEDF has been also suggested to play an important role in its physiological activities (44). PEDF-induced ERK1/2 phosphorylation has been largely implicated in the prosurvival effects of PEDF toward neural cells (41,42). On the other hand, the ability of PEDF to block growth of ECs and to reduce angiogenesis has been associated with up-regulation of p38 and JNK MAPKs (45,46) and p38dependent cleavage of multiple caspases (47). In addition, recent work by Biyashev et al. (48) demonstrated the involvement of ERK5 phosphorylation and the subsequent displacement of SMRT co-repressor in PEDF-induced activation of peroxisome proliferator-activated receptor ␥. In our experiments, we did not observe any ERK5 activation by PEDF, whereas JNK and p38 were found to mediate respectively PEDF-induced EC apoptosis through caspase-3 activation and both endothelial and cancer cell migration arrest. We have also shown herein that ERK1/2 activation by PEDF and its mutant does not interfere with the above effects. These evidences suggest that antiangiogenic signaling by PEDF occurs mainly through the activation of p38 and JNK stress-related kinases, but the overall downstream outcome can be also influenced by additional signaling pathways that vary in different cell systems. Therefore, additional studies are needed to fully delineate the exact subsequence of intracellular signaling events that mediate the effects of PEDF.
In summary, our study unveils important mechanistic insights of the biological activity of PEDF and its phosphomimetic mutant, EEE-PEDF. In continuation to our previous studies, we are now showing that the enhanced antiangiogenic and anticancer properties of the phosphomimetic PEDF are achieved through significantly stronger binding to LR and stronger intracellular signaling effects, as compared with WT-PEDF. Additionally, we demonstrate that JNK and p38 MAPKs differentially regulate proapoptotic and antimigratory activities of PEDF. We believe that our findings will contribute to better understanding of the complexity of PEDF effects on endothelial and cancer cells.