Characterization of Heparanase-induced Phosphatidylinositol 3-Kinase-AKT Activation and Its Integrin Dependence*

Background: Heparanase levels are associated with tumor metastasis. Latent heparanase induces the prosurvival PI3K-AKT pathway via uncharacterized mechanisms. Results: AKT activation by heparanase involved PI3K p110α, RAS, RICTOR, a PP2-sensitive kinase, FAK or PYK2, and β1 or β3 integrin-mediated adhesion. Conclusion: The heparanase receptor(s) intimately cooperates with integrins during induction of the PI3K-AKT pathway. Significance: Key steps in the heparanase-PI3K-AKT axis were identified. Heparanase functions as a heparan sulfate-degrading enzyme and as a ligand for an unidentified signaling receptor(s). Here, several reactions involved in the activation of the PI3K-AKT pathway by latent heparanase were characterized. Protein suppression using specific siRNAs revealed that heparanase-induced phosphorylation of AKT at Ser-473 was RICTOR-mTOR-dependent, whereas ILK and PAK1/2 were dispensable. p110α was the PI3K catalytic isoform preferred by heparanase for AKT activation and cell proliferation because the p110α inhibitor YM024 blocked these processes. Heparanase-induced AKT phosphorylation was low in mouse embryonic fibroblast cells expressing a RAS interaction-defective p110α compared with wild type cells, indicating that RAS has an important role in the PI3K-AKT activation. The response to heparanase was also inefficient in suspension cultures of several cell lines, suggesting a requirement of integrins in this pathway. Adhesion via either αVβ3 or α5β1 promoted heparanase-induced AKT phosphorylation, and a stronger effect was seen when both integrins were engaged. Simultaneous inhibition of FAK and PYK2 using a chemical inhibitor, or suppression of their expression, inhibited heparanase-induced AKT activation and cell proliferation. Stimulation of cells with heparanase enhanced their resistance against oxidative stress- or growth factor starvation-induced apoptosis. These results demonstrate that there is an intimate cross-talk between the heparanase receptor(s) and integrins during induction of the prosurvival PI3K-AKT pathway by heparanase.

Heparanase functions as a heparan sulfate-degrading enzyme and as a ligand for an unidentified signaling receptor(s). Here, several reactions involved in the activation of the PI3K-AKT pathway by latent heparanase were characterized. Protein suppression using specific siRNAs revealed that heparanase-induced phosphorylation of AKT at Ser-473 was RICTOR-mTORdependent, whereas ILK and PAK1/2 were dispensable. p110␣ was the PI3K catalytic isoform preferred by heparanase for AKT activation and cell proliferation because the p110␣ inhibitor YM024 blocked these processes. Heparanase-induced AKT phosphorylation was low in mouse embryonic fibroblast cells expressing a RAS interaction-defective p110␣ compared with wild type cells, indicating that RAS has an important role in the PI3K-AKT activation. The response to heparanase was also inefficient in suspension cultures of several cell lines, suggesting a requirement of integrins in this pathway. Adhesion via either ␣V␤3 or ␣5␤1 promoted heparanase-induced AKT phosphorylation, and a stronger effect was seen when both integrins were engaged. Simultaneous inhibition of FAK and PYK2 using a chemical inhibitor, or suppression of their expression, inhibited heparanase-induced AKT activation and cell proliferation. Stimulation of cells with heparanase enhanced their resistance against oxidative stress-or growth factor starvation-induced apoptosis. These results demonstrate that there is an intimate crosstalk between the heparanase receptor(s) and integrins during induction of the prosurvival PI3K-AKT pathway by heparanase.
Heparanase is a unique protein that possesses both enzymatic activity and signaling functions (1). Heparanase activity is correlated with cancer metastasis and angiogenesis, and the heparanase protein is up-regulated in essentially all types of human cancer, namely carcinomas, sarcomas, and hematological malignancies (1)(2)(3). Increased heparanase levels are often associated with increased tumor metastasis, higher microvessel density, and reduced postoperative patient survival (3)(4)(5)(6)(7), thus critically supporting the intimate involvement of heparanase in tumor progression.
Heparanase is secreted in a latent form, which after activation can cleave heparan sulfate chains at selected sites and thereby modify the extracellular matrix and cell surface proteoglycans. The activation process involves endocytosis of the latent protein, intracellular proteolytic cleavage, and release of the protein out of the cell (1,8). Independent of its enzymatic function, heparanase possesses signaling activity (9,10). This activity has been mapped to the C-terminal part outside of the catalytic domain (11), and both the enzymatically latent and active forms of the protein can trigger intracellular signals. The first identified heparanase-induced signaling event was the activation of AKT. Exogenous addition of enzyme-inactive heparanase can potently activate AKT in many cell types, in a heparan sulfate-independent manner (1,9). Heparanase overexpression in transfected cells or addition of the purified protein to adherent cells also induces activation of SRC and p38 MAPK (12). As one of the SRC-dependent effects, heparanase can enhance EGF receptor phosphorylation at specific sites (Tyr-1173 and Tyr-845) (10). Two downstream mediators of this EGF receptor phosphorylation are STAT3 and STAT5b (13).
The signaling is believed to be mediated by a cell surface heparanase receptor(s). Evidence indicates that this receptor(s) is localized to lipid rafts, based on the inhibitory effect of cholesterol-depleting agents on heparanase signaling (14). Different proteins that have been suggested as heparanase receptors include mannose 6-phosphate receptor and low density lipoprotein receptor-related protein (15). However, heparanase efficiently activates AKT in cells where these proteins have been knocked out, and therefore the identity of the receptor(s) remains unknown (14). Furthermore, the mechanisms by which the heparanase receptor(s) activates its downstream mediators have not been fully characterized.
In this study we aimed to characterize key steps of the heparanase-induced PI3K-AKT activation pathway. In addition to the identification of the PI3-kinase catalytic isoform and the PDK2 that mediates AKT activation in response to heparanase, a close dependence on focal adhesion kinase (FAK) 3 or prolinerich tyrosine kinase 2 (PYK2) and integrin-mediated adhesion is reported.

Heparanase Stimulation Experiments
Adherent Cells-Recombinant latent heparanase (1 g/ml) was added to cells previously cultured in serum-free medium for 18 h. Fatty acid-free BSA was used as a carrier and was added to the control cells. Cells were stimulated for 10 min except for time-response experiments as indicated. After stimulation, culture medium was removed, and the cells were directly lysed in SDS-PAGE buffer.
Cells in Suspension-MCF7 and Tet-FAK cells were harvested by trypsin-EDTA treatment, whereas CHO-K1 cells were harvested by 10 mM EDTA. Trypsin was inactivated by addition of soybean trypsin inhibitor (1 mg/ml), cells were washed with serum-free medium, adjusted to 5 ϫ 10 5 /ml, and incubated (1 h, 37°C) to recover from the effects of the detachment treatments. Subsequently, the cell were stimulated as described above, collected by centrifugation, and lysed in SDS-PAGE buffer.
Cells on Defined Substrates-Six-well cell culture plates were coated with vitronectin or fibronectin (20 g/ml in PBS) overnight at 4°C. The uncoated surfaces were blocked with Pluronic (a nonadhesive polymer). GD25T␤1A cells cultured in serum-free medium overnight were harvested by trypsin-EDTA treatment, seeded at a density of 5 ϫ 10 5 /well, and allowed to adhere to the vitronectin-or fibronectin-coated wells in serum-free medium for 3 h before stimulation with heparanase. Where indicated, adhesion to fibronectin-coated wells was performed in the presence of 10 g/ml cyclic RGD peptide (fibronectin-R).

Cell Surface Integrin Pulldown
WGA-binding glycoproteins, including maturely glycosylated integrins, of adherent or suspended MCF7, Tet-FAK, and CHO-K1 cells were pulled down using WGA-Sepharose 4B. Cell lysates were prepared in Triton lysis buffer (1% Triton X-100, 20 mM Tris, pH 7.5, 150 mM NaCl, 1 mM EGTA, protease inhibitor mixture (Roche Applied Science)) from serumstarved adherent cells or cells detached from tissue culture flasks as described above. Total protein levels were determined using a Bradford assay kit (Bio-Rad), and equal amounts of protein from attached and suspended cells (MCF7 600 g; Tet-FAK and CHO-K1 520 g) were subjected to the WGA-Sepharose pulldown as described previously (17).

Inhibitor Treatments
Stock solutions of all inhibitors were prepared in DMSO. To inhibit PAK, cells were treated with IPA3 (30 M) for 20 min before and during stimulation with heparanase. The cells were similarly pretreated with the SRC family kinase inhibitor PP2 (10 or 20 M), PI3K inhibitor LY294002 (10 M), PI3K catalytic isoform inhibitors YM024 (0.1-1 M), TGX221 (0.1-1 M) and AS252424 (0.05-0.5 M), and the FAK/PYK2 inhibitor PF562271 (1-4 M). The effects of the PI3K and FAK/PYK2 inhibitors on cell growth of U87 Hepa cells were analyzed using the MTT assay (see below). DMSO-treated cells were used as controls in all these experiments.

MTT Assays
The MTT assay to determine cell number was performed essentially as described by Edmondson et al. (25).

TUNEL Assay
GD25T␤1A cells were serum-starved for 18 h, and 1 ϫ 10 5 cells were allowed to adhere to fibronectin-coated 24-well cell culture plates. After 3 h, cells were either stimulated with heparanase for 10 min or left untreated as control. Subsequently, the cells were exposed to 1 mM H 2 O 2 for 2 h, and cell death was determined by performing TUNEL staining as recommended by the manufacturer (Invitrogen). Images were taken using an Axiovert 200M inverted microscope (Carl Zeiss).

Mitochondrial Membrane Potential Assay
U87 V 0 and Hepa cells growing in complete medium were harvested by trypsin-EDTA treatment, and 5 ϫ 10 5 cells were seeded on fibronectin-coated cell culture dishes (3.5 cm) in the presence or absence of serum. After 4 h, changes in mitochondrial membrane potential caused by serum deprivation were analyzed by staining the cells with JC1 dye following the protocol suggested by the kit manufacturer (Sigma). An Axiovert 200M inverted microscope was used to take photographs.

SDS-PAGE and Western Blotting
Cell lysates were resolved by SDS-PAGE on Bio-Rad pre-cast mini-gels. The separated proteins were transferred to Hybond-ECL nitrocellulose membrane (GE Healthcare) and probed for different antigens as indicated. Signals were developed using ECL and captured either on an x-ray film or with the Chemi-Doc TM MP System (Bio-Rad).

Statistical Analysis
Statistical analyses were performed using Student's t test. p values Ͻ 0.05 were considered significant. All quantifications provided in the form of graphs are mean values Ϯ S.E.

Heparanase-mediated AKT Ser-473 Phosphorylation Is
RICTOR-mTOR-dependent-The RICTOR-mTOR kinase complex (TORC2) has been shown to mediate phosphorylation of AKT Ser-473 in response to stimulation of growth factor receptors as well as ␤1 integrins (24,26). However, other kinases may also perform this function in different contexts (27)(28)(29). We therefore investigated whether RICTOR-mTOR was required for heparanase-induced AKT Ser-473 phosphorylation. The RICTOR protein level in MCF7 cells was suppressed using siRNA, and the cells were exposed to heparanase. Whereas phosphorylation at AKT Ser-473 was increased ϳ2.5fold in response to heparanase addition in cells transfected with a nontarget siRNA, it was inhibited in cells transfected with RICTOR-directed siRNA (Fig. 1A). Moreover, the heparanaseinduced phosphorylation of FOXO1 Thr-24, a downstream AKT target, was also inhibited following RICTOR suppression. In contrast, suppression of ILK, a protein mediating integrininduced signaling and suggested to participate in AKT Ser-473 phosphorylation (30,31), had no effect on heparanase-induced phosphorylation at AKT Ser-473 or FOXO1 Thr-24 (Fig. 1B). Notably, heparanase exposure did not stimulate ERK phosphorylation.
PAK1/2 function as scaffolding proteins in the PI3K-AKT pathway downstream of some growth factors (32). In MCF7 cells, PAK is required for AKT Ser-473 phosphorylation induced by PDGF or lysophosphatidic acid, but not by EGF (24). Heparanase-induced AKT Ser-473 phosphorylation in MCF7 cells was not affected by simultaneous suppression of PAK1 and PAK2 (Fig. 1C) or by treatment with the PAK inhibitor IPA3 (Fig. 1D). The effect of IPA3 on lysophosphatidic acid-induced AKT Ser-473 phosphorylation was used as a positive indicator of inhibitor efficiency.
Heparanase-induced AKT Activation Is Blocked by PP2 but Involves a Kinase Other Than the SRC Family Kinases SRC, YES, and FYN-It has been shown that the SRC family kinase inhibitor PP2 completely blocks heparanase-induced AKT activation in LNCaP cells (10). Here, this was confirmed in MCF7 cells ( Fig. 2A). To test whether heparanase-induced AKT activation was mediated by one of the three ubiquitously expressed SRC family kinases (SRC, YES, and FYN), we added heparanase to MEFs with a targeted deletion of these proteins individually. Heparanase efficiently stimulated phosphorylation of AKT at Ser-473 in each of these cells (data not shown). Also, in the SYF cell line where all three kinases are knocked out, heparanase was able to induce efficient phosphorylation of AKT at Ser-473 within 5 min of addition to these cells (Fig. 2B). Thus, the inhibitory effect of PP2 on the PI3K-AKT pathway is not mediated by the inactivation of SRC, YES, or FYN.
AKT Activation upon Heparanase Stimulation Requires PI3K Catalytic Subunit p110␣-It has been reported that different receptors prefer different PI3-kinase catalytic subunit isoforms to activate the AKT pathway. For example, whereas the EGF receptor and ␤1 integrins utilize the p110␣ isoform (19,33), G protein-coupled lysophosphatidic acid receptors prefer the p110␤ catalytic subunit (34). Because the receptor(s) that mediates heparanase signaling is not known, identification of the PI3K catalytic subunit(s) involved in heparanase-mediated AKT activation may help in identifying this elusive receptor(s). To address this question, we used isoform-selective chemical inhibitors for the catalytic subunits p110␣, p110␤, or p110␥ and examined AKT phosphorylation following addition of heparanase. As expected, heparanase was unable to induce AKT acti-vation in MCF7 cells treated with the general PI3K inhibitor LY294002 (Fig. 3A). The p110␣ inhibitor YM024 reduced AKT Ser-473 phosphorylation by heparanase in magnitude similar to LY294002. In contrast, the p110␤ and p110␥ inhibitors TGX221 and AS252424 only moderately affected heparanaseinduced AKT Ser(P)-473 (Fig. 3A), suggesting that the p110␣ subunit mediates heparanase signaling to AKT.
To characterize the effect of p110␣ inhibition on heparanase-stimulated cell growth, we used U87 glioma cells cultured under serum-free conditions. Overexpression of latent heparanase stimulated U87 cell proliferation compared with control cells (Hepa DMSO and V 0 DMSO; Fig. 3B). Proliferation of U87 Hepa cells was reduced by YM024 (0.5 M) and LY294002 (10 M) to a similar magnitude (Fig. 3B), suggesting that p110␣ mediates the proliferative effect of heparanase.
The activation of PI3K is known to involve GTP-RAS in a stimuli-dependent manner (19). The role of RAS in heparanase-induced AKT activation was studied by incubating the p110␣ wt and p110␣ mut cells with latent heparanase. Timeresponse curves revealed that AKT phosphorylation at both activation sites was markedly reduced in p110␣ mut cells compared with their wild type counterparts (Fig. 3C).
Integrins Promote Heparanase-induced PI3K-AKT Pathway Activation-Given the similarities between heparanase-(Figs. 1-3) and integrin-induced activation of the PI3K-AKT pathway (24,33,35) and the ability of heparanase to activate integrins (16,36), we examined whether increased AKT phosphorylation by heparanase involved integrins. Addition of heparanase to adherent MCF7 cells resulted in ϳ5-fold increased AKT phosphorylation (Fig. 4A, top panel, Ad) in agreement with previous results (Fig. 1). In striking contrast, heparanase failed to enhance AKT phosphorylation when added to MCF7 cells in suspension (Fig. 4A, top panel, Sp). A lower magnitude of AKT phosphorylation was similarly observed following addition of heparanase to Tet-FAK and EDTA-detached CHO-K1 cells kept in suspension compared with adherent cells (Fig. 4A, middle and bottom panels, respectively). To confirm that this difference in AKT activation was not due to variable integrin levels in adherent and suspended cells, cell surface glycoproteins were isolated from the cells, and the amount of integrin subunit ␤1 was determined as described under "Materials and Methods." Similar levels of WGA-binding ␤1 integrins (i.e. maturely glycosylated) were available on these cells during adherent and suspension culture (Fig. 4B).
Next, we asked whether heparanase-induced AKT phosphorylation levels were different in cells attached via ␤3 and/or ␤1 integrins. We utilized a GD25-derived cell line where the expression of the ␤1 integrin subunit is turned off in the pres-  ence of tetracycline (GD25T␤1A; Fig. 4C). Addition of heparanase to these cells in the presence of tetracycline resulted in increased AKT phosphorylation after 5 min, followed by a rapid decline. Notably, addition of heparanase to GD25T␤1A cells in the absence of tetracycline resulted in higher and prolonged AKT phosphorylation (Fig. 4C). Similarly, cells where the ␤1 integrin gene was inactivated using a cre/LoxP system showed lower levels of AKT phosphorylation in response to heparanase addition than the ␤1 integrin-expressing control cells (Fig. 4D). To analyze directly the ability of ␣V␤3 and ␣5␤1 to promote heparanase signaling, we added heparanase to ␤1 integrin-expressing GD25T␤1A cells plated on vitronectin, or on fibronectin with or without ␣V␤3-blocking cyclic RGD peptide in the medium. Heparanase stimulated AKT Ser-473 phosphorylation under all of these adhesion conditions, but Ser-473 phosphorylation was notably higher in cells plated on fibronectin without RGD (Fig. 4E). Taken together, these results show that both ␣V␤3 and ␣5␤1 can promote heparanase signaling, and the signaling response is stronger when both integrins are engaged in adhesion.
Syndecans act as co-receptors to integrins (37), and latent heparanase can bind to syndecan1 and syndecan4 (38). Whereas syndecan1 is expressed mainly in epithelial cells and transiently in condensing mesenchyme during embryogenesis (37,39), syndecan4 is widely expressed and co-localizes with ␤1 and ␤3 integrins in focal adhesion sites (37). To investigate the possible involvement of syndecan4 in heparanase-induced AKT activation syndecan4-expressing wild type and syndecan4 knock-out MEFs were analyzed. There was no significant difference between these cells in PI3K-AKT pathway induction upon heparanase exposure (Fig. 4F).
Heparanase Stimulation Increases FAK/PYK2 Autophosphorylation-FAK is a mediator of several integrin signaling pathways. Autophosphorylation of FAK at Tyr-397 in response to integrin engagement results in binding and activation of PI3K and subsequently AKT (40). The FAK-related enzyme PYK2 is autophosphorylated at Tyr-402 in response to integrin stimulation (41) or exposure to cytokines such as angiotensin-II and PDGF (42). Addition of heparanase to MCF7 and U87 cells increased PYK2 phosphorylation at Tyr-402 in dose-and timedependent manners. A moderate increase in FAK Tyr-397 phosphorylation was also seen in these cells (Fig. 5, A and B). Furthermore, latent heparanase-secreting U87 Hepa cells had higher PYK2 Tyr(P)-402 and AKT Ser(P)-473 levels compared with the U87 V 0 control cells (Fig. 5C). In some other cell lines, which maintained high endogenous phosphorylation levels at FAK Tyr-397 or PYK2 Tyr-402 after overnight culture in serumfree medium, the heparanase-induced increase in FAK or PYK2 autophosphorylation was less prominent (data not shown).
Heparanase-induced AKT Activation Is Dependent on Active FAK or PYK2-To investigate whether FAK or PYK2 was involved in AKT activation by heparanase we utilized PF562271, an inhibitor of both kinases. PF562271 efficiently blocked heparanase-induced phosphorylation of AKT as well as of the AKT target FOXO1 Thr-24 (Fig. 6A). The individual contribution of FAK and PYK2 to this pathway was analyzed in the Tet-FAK cell line where FAK expression can be regulated by tetracycline. Heparanase strongly induced autophosphorylation of both FAK and PYK2 in these cells (Fig. 6B). Turning off FAK expression did not affect heparanase-induced AKT phosphorylation. Conversely, when PYK2 was knocked down while FAK was expressed, only a minor decrease in AKT phosphorylation was obtained (Fig. 6C). Simultaneous suppression of FAK and PYK2 protein levels resulted in a significantly weaker AKT activation after heparanase treatment compared with control cells (Fig. 6D). The phosphorylation of FOXO1 Thr-24 was also inhibited under this condition. These results indicate that both FAK and PYK2 could mediate activation of the PI3K-AKT pathway upon heparanase exposure.  ). B, WGA-binding proteins were isolated from lysates prepared from adherent and suspension cultures of MCF7, Tet-FAK, and CHO-K1 cells and subjected to SDS-PAGE under nonreducing conditions. The figure shows an immunoblot for the integrin subunit ␤1. All samples were run on the same SDS-polyacrylamide gel. The figure has been made by combining a long exposure (MCF7) and a short exposure (Tet-FAK, CHO-K1). C, a representative Western blot (n ϭ 2) and quantification graph show kinetics of AKT Ser-473 and Thr-308 phosphorylation after addition of heparanase to GD25T␤1A cells cultured in the presence or absence of tetracycline to regulate the ␤1 integrin expression (*, p Ͻ 0.05). D, AKT phosphorylation kinetics of ␤1 integrin-null (␤1 Ϫ/Ϫ ) and ␤1 wt (␤1 fl/fl ) MEF cells stimulated with 1 g/ml heparanase for different time periods as indicated are shown. E, serum-starved GD25T␤1A cells (cultured in the absence of tetracycline) were allowed to adhere to matrix proteins as indicated for 3 h. The cells were then stimulated with heparanase and AKT Ser(P)-473 levels were analyzed in the cell lysates. The quantification given below shows -fold increase in AKT Ser(P)-473 in response to heparanase treatment (mean Ϯ S.E., n ϭ 3; *, p Ͻ 0.05; n.s., nonsignificant). F, AKT phosphorylation kinetics of heparanase-treated, syndecan4-expressing wild type and syndecan4 knock-out cells are shown. Next, proliferation of U87 Hepa cells in serum-free medium containing different concentrations of PF562271 was studied by the MTT assay. PF562271 (1 or 2 M) blocked the growth of these cells to the level of untreated U87 V 0 control cells, suggesting that the proliferative function of heparanase is mediated by the FAK-PYK2-AKT axis (Fig. 6D).
Heparanase Increases Resistance to Stress-induced Apoptosis-The ability of heparanase to enhance cell survival during cell stress was analyzed by two methods. GD25T␤1A cells adhering to fibronectin were stimulated by heparanase for 10 min and then subjected to oxidative stress by exposure to 1 mM H 2 O 2 for 2 h. Heparanase stimulation reduced the number of TUNELpositive cells compared with unstimulated cells (Fig. 7A). The resistance to serum starvation was analyzed by monitoring changes in the mitochondrial membrane potential, an early event in apoptosis, using the JC1 dye. Heparanase-overexpressing U87 Hepa cells maintained mitochondrial membrane potential to a higher extent, as shown by less JC1 monomers (green) and more J aggregates (red), compared with U87 V 0 cells (Fig. 7B).

DISCUSSION
Heparanase possesses several activities that may be utilized by cancer cells to promote their expansion. The heparan sulfate-degrading activity can facilitate cell migration and invasion by disrupting basement membranes and other barriers; it can also release chemotactic and angiogenic factors from the extracellular matrix. The most prominent enzyme-independent signaling function of heparanase observed so far is the activation of AKT (9), but it can also induce phosphorylations on SRC, p38 MAPK, and EGF receptor and induce VEGF gene expression (1,10,12). Elevated heparanase signaling may thus result in enhanced cell survival, proliferation, and angiogenesis.
Binding studies with heparanase have suggested that a high affinity binding site for heparanase exists on many cell types and strongly support the existence of a specific heparanase receptor(s) (11,14). Several key features of the mechanism of AKT activation triggered by the putative receptor were identified in the present study. (i) The heparanase-induced pathway to AKT used RICTOR-mTOR for the AKT Ser-473 phosphorylation, and this reaction did not involve ILK or PAK (Fig. 1). (ii) PI3K p110␣ was the main catalytic isoform utilized by the heparanase receptor(s) for AKT Ser-473 phosphorylation in MCF7 cells, as well as for heparanase-induced proliferation of U87 Hepa cells (Fig. 3, A and B). The essential role of p110␣ in the heparanase-AKT pathway provides further molecular evidence for the tumorigenic properties of heparanase, because activating mutations in this isoform are oncogenic and have been found in many tumors (43,44). (iii) The activation of PI3K/AKT by heparanase was inefficient if the PI3K-RAS interaction was prevented (Fig. 3C) as shown previously for ␤1 integrin ligands, EGF and FGF2, whereas the interaction was dispensable for PDGF stimulation of AKT (19,33). (iv) Integrin-mediated adhesion strongly promoted the heparanase response (Fig. 4).
(v) The activation of PI3K-AKT by heparanase required active FAK or PYK2 (Fig. 6), and (vi) it was PP2-sensitive but did not involve SRC, YES, or FYN (Fig. 2). Exposure to heparanase led to enhanced resistance to stress-induced apoptosis (Fig. 7), presumably due to activation of the PI3K-AKT pathway.
Integrins have been reported to promote responses from many growth factors by several different mechanisms (for excellent review, see Ref. 45). Two main mechanisms for the cooperation are (i) synergistic signaling from the two receptors by converging pathways, as described for the ERK pathway after EGF stimulation (46,47); and (ii) regulation of growth factor receptors by direct interaction with integrins, e.g. VEGFR2 and ␤1 (48). The mechanism by which the heparanase receptor(s) and integrins cooperate cannot be ascertained until the receptor(s) is identified. However, some conclusions regarding the mechanisms can be drawn from our present data and are summarized in Fig. 8. First, although AKT Ser-473 phosphorylation in response to heparanase stimulation was markedly reduced in suspension cultures of MCF7, Tet-FAK, and CHO-K1 cells, the reaction did occur to some extent in the detached Tet-FAK and CHO-K1 cells (Fig. 4A). This indicates that the heparanase receptor(s), albeit inefficiently, could activate the PI3K-AKT pathway by itself. Second, the components of integrin-mediated AKT activation and the heparanase-induced pathway were strikingly similar, but a notable difference in the receptor activities was the absence of ERK activation upon heparanase stimulation (Ref. 12 and Fig. 1, A-C). Because ␤1 and ␤3 integrins are potent activators of ERK, heparanase was clearly not exerting its signaling function through a general engagement of these integrins. Instead, ␤1 and ␤3 integrin-mediated adhesion appeared to facilitate the endogenous activity of the heparanase receptor(s).
FAK is autophosphorylated at Tyr-397 in response to cell attachment on extracellular matrix proteins, and via this site FAK can interact with the regulatory subunit of PI3K (40). An increase in FAK autophosphorylation in glioma cells overexpressing heparanase has been shown previously (16,36). We found that latent heparanase increased both FAK and PYK2 autophosphorylation and that these reactions have central roles in the PI3K-AKT pathway. Blocking FAK/PYK2 autophosphorylation with the inhibitor PF562271 completely abrogated heparanase-induced AKT activation and phosphorylation of the downstream AKT target protein FOXO1 in MCF7 cells (Fig. 6A), and it reduced heparanase-mediated proliferation of U87 Hepa cells (Fig. 6D). The results from individual and simultaneous suppression of FAK or PYK2 (Fig. 6, B and C) indicate that they could compensate for each other in the heparanase-integrin-AKT pathway.
Our data suggest that the heparanase receptor(s) engages integrins for efficient PI3K-AKT pathway induction through a mechanism that requires FAK or PYK2 (Fig. 7). Generally, the activation of FAK requires clustering of integrins by extracellular ligands and binding of the C-terminal domain of FAK to FIGURE 8. Schematic model for the mechanism of cooperation between integrins and the heparanase receptor. Heparanase stimulation leads to activation of PI3K and phosphorylation of the downstream kinase AKT at Ser-473 and Thr-308. Integrins promote the heparanase-induced PI3K-AKT pathway activation by a mechanism that involves FAK/PYK2. The pathway is promoted by the PI3K-RAS interaction. Important missing steps and remaining questions in the heparanase receptor(s)-integrins cross-talk are indicated with question marks: What is the adhesion-independent link between heparanase receptor and PI3K? Does the heparanase receptor activate FAK/PYK2, or does FAK/PYK2 activate the heparanase receptor? What is the identity of the PP2-sensitive kinase? talin (49). PYK2 is not localized to adhesion sites in fibroblasts (50,51), and the mechanism by which it is activated by ␤1 and ␤3 integrins in MEFs is therefore unclear.
Several growth factors have been reported to induce FAK and PYK2 activation (42,52,53) although their receptors have not been shown to interact with talin. In the case of EGF and PDFG, the activation was achieved through the bridging of FAK and PYK2 between the receptors and integrin-associated proteins (presumably the talin-paxillin complex) (52,54). Whether a similar mechanism underlies the observed interplay between heparanase receptor, integrins, FAK/PYK2, and PI3K remains to be determined. The mechanism of possible direct induction of PI3K by heparanase receptor is another open question. To provide satisfactory answers to these questions, it is imperative to identify the protein(s) that serves as the heparanase receptor(s).