Binding of activated alpha2-macroglobulin to its cell surface receptor GRP78 in 1-LN prostate cancer cells regulates PAK-2-dependent activation of LIMK.

Two characteristics of highly malignant cells are their increased motility and secretion of proteinases allowing these cells to penetrate surrounding basement membranes and metastasize. Activation of 21-kDa activated kinases (PAKs) is an important mechanism for increasing cell motility. Recently, we reported that binding of receptor-recognized forms of the proteinase inhibitor alpha2-macroglobulin (alpha2M*) to GRP78 on the cell surface of 1-LN human prostate cancer cells induces mitogenic signaling and cellular proliferation. In the current study, we have examined the ability of alpha2M* to activate PAK-1 and PAK-2. Exposure of 1-LN cells to alpha2M* caused a 2- to 3-fold increase in phosphorylated PAK-2 and a similar increase in its kinase activity toward myelin basic protein. By contrast, the phosphorylation of PAK-1 was only negligibly affected. Silencing the expression of the GRP78 gene, using either of two different mRNA sequences, greatly attenuated the appearance of phosphorylated PAK-2 in alpha2M*-stimulated cells. Treatment of 1-LN cells with alpha2M* caused translocation of PAK-2 in association with NCK to the cell surface as evidenced by the co-immunoprecipitation of PAK-2 and NCK in the GRP78 immunoprecipitate from plasma membranes. alpha2M*-induced activation of PAK-2 was inhibited by prior incubation of the cells with specific inhibitors of tyrosine kinases and phosphatidylinositol 3-kinase. PAK-2 activation was accompanied by significant increases in the levels of phosphorylated LIMK and phosphorylated cofilin. Silencing the expression of the PAK-2 gene greatly attenuated the phosphorylation of LIMK. In conclusion, we show for the first time the activation of PAK-2 in 1-LN prostate cancer cells by a proteinase inhibitor, alpha2-macroglobulin. These studies suggest a mechanism by which alpha2M* enhances the metastatic potential of these cells.

Cancer of the prostate is the most commonly diagnosed malignancy of men (1). In the development of prostate cancer, deregulation of cell growth control often is accompanied by acquisition of androgen independence, a poor prognostic indicator (2,3). Growth factors, including epidermal growth factor, insulin-like growth factor, and fibroblast growth factor play a role in the progression of androgen-independent prostate can-cer (2,3). These growth factors induce mitogenic cellular responses by activating their specific receptors. Ligand binding to these receptors induces the autophosphorylation of the receptor on specific tyrosine residues resulting in the assembly of multiprotein complexes, which activate the Ras/MAPK 1 and PI 3-kinase signaling pathways (4). In addition to increased activation of signaling pathways that promote cellular proliferation and/or suppression of apoptosis, increased motility is often seen in malignantly transformed cells. This increase in motility, along with increased secretion of proteinases, especially matrix metalloproteinases, enables highly metastatic cancer cells to penetrate surrounding basement membranes and invade blood vessels and lymphatics. One mechanism that promotes increased motility of malignant cells is activation of members of the 21-kDa activated kinase (PAK) family.
These proteins are Ser/Thr kinases that mediate Rac and Cdc42 GTPase-dependent signaling (see reviews in Refs. 5-8 and references therein). The mammalian PAK family consists of six members, including PAK-1 and PAK-2. PAK-1 is tissuespecific in its expression, whereas PAK-2 is ubiquitously expressed. The catalytic activity of PAKs is regulated by the binding of active GTPases to the conserved p21 binding motif in the NH 2 -terminal domain leading to the relief of autoinhibitory interactions with the COOH-terminal catalytic domain (5)(6)(7)(8). PAK-2 is also activated by caspase or caspase-like proteinases, which generate constitutively active p34 PAK-2, the COOHterminal catalytic domain (9). Activated full-length PAK-2 stimulates cell survival and growth in response to various stress stimulants, whereas its proteolytic fragment, p34 protein, stimulates cell death (9,10). Stimulation of cell survival by activated full-length PAK-2 is partly mediated by phosphorylation and inhibition of pro-apoptotic Bad (8 -10). The activation of PAK-2 in response to irradiation or cytosine ␤-Darabinoside is dependent on protein-tyrosine kinase and PI 3-kinase activity (11). PAK-1 mediates signals from the Ras/ MAPK and PI 3-kinase signaling pathways to promote cell transformation. PAKs also play important roles in modulating the ability of cancer cells to move and metastasize (5)(6)(7)(8). A number of highly metastatic human breast cancer lines exhibit constitutively elevated PAK-1 or PAK-2 activity (12). ␣ 2 -Macroglobulin (␣ 2 M) is a broad specificity proteinase inhibitor that binds to cell surface receptors when activated by proteinases (13). The activated form of ␣ 2 M (␣ 2 M*) is also produced by direct reaction of internal thiol esters present in each of its four identical subunits with small amines or ammonia (13). Binding of ␣ 2 M* to macrophages (14,15), rheumatoid synovial fibroblasts (16), and 1-LN prostate cancer cells triggers increases in the levels of intracellular inositol 1,3,4trisphosphate and cytosolic-free calcium [Ca 2ϩ ] i and is followed by activation of components of the Ras/MAPK and PI 3-kinase signaling cascades (17)(18)(19)(20). As a consequence of these events, ␣ 2 M* up-regulates DNA synthesis and cellular proliferation (17)(18)(19)(20). Based on these and other observations, we hypothesized that ␣ 2 M* functions like a growth factor, and its receptor functions as a growth factor receptor (13). Low density lipoprotein receptor-related protein-1 was identified in the 1990s as an ␣ 2 M* receptor. Subsequent studies in our laboratory suggested that a receptor distinct from low density lipoprotein receptorrelated protein-1 must account for ␣ 2 M*-dependent signal transduction (14 -20). These events require the presence of a small number of sites (ϳ1500/cells) demonstrating very high ligand affinity (K d 50 -100 pM) for cellular binding of ␣ 2 M* or its receptor binding domain. This second receptor, initially termed the ␣ 2 M* signaling receptor, was later isolated from murine peritoneal macrophages and 1-LN human prostate cancer cells and identified as cell surface-associated GPR78, a heat shock protein of the HSP70 family (20). This molecular chaperone has been highly characterized for its ability to promote cell survival during endoplasmic reticulum (ER) stress (see reviews in Refs. 21-26 and references therein). GRP78 is involved in many cellular processes, including antigen presentation, translocation of newly synthesized polypeptides across the ER membrane, and their subsequent folding, maturation, transport, or retrotranslocation (21)(22)(23)(24)(25)(26). An increased expression of GPR78 is a part of the unfolded protein response required to alleviate ER stress, maintain ER function, and protect cells against cell death (21)(22)(23)(24)(25)(26). GRP78 is constitutively expressed, but its synthesis can be up-regulated by a variety of stressful conditions (21-26, 28, 29) that perturb protein folding and assembly within the ER, including glucose deprivation, acidosis, and hypoxia (21)(22)(23)(24)(25)(26). Poorly vascularized solid tumors demonstrate both hypoxia and acidosis, and these cells may be viewed as highly stressed.
The presence of GRP78 on the cell surface has only recently been appreciated. Constitutive cell surface expression on normal cells is low, but various treatments up-regulate its cell surface expression (18). For example, we have demonstrated in vivo up-regulation of GRP78 on the surface of antigen-presenting cells when mice are exposed to various stimuli (27). These events appear to have consequences with respect to ␣ 2 M*mediated antigen presentation (28). Normal fibroblasts do not show cell surface expression of GRP78, and ␣ 2 M* treatment does not trigger signaling responses by these cells. Rheumatoid synovial fibroblasts, however, express GRP78 on their cell surface and signal briskly when exposed to ␣ 2 M* (16). GRP78 also is expressed to a high degree on the surface of a number of cancer cells, including the highly metastatic 1-LN human prostate cancer cell line (29). By contrast, GRP78 cell surface localization and ␣ 2 M*-dependent signal transduction do not occur with PC-3 cells, a cell line of low malignant potential, and the parent line for the 1-LN cell line (see, for example, Refs. [17][18][19]. Recent studies have demonstrated antibodies against GRP78 in the sera of prostate cancer patients, and the presence of these antibodies is highly correlated with increased metastatic potential and a poor prognosis (30 -31). The appearance of the normally cryptic GRP78 protein on the cell surface in high concentration may be a critical factor in the development of autoantibodies to GRP78.
The circulating concentration of ␣ 2 M is 1-5 M, and ␣ 2 M* comprises about 200 -500 nM of this pool (32). It has been estimated that about 1 g of ␣ 2 M turns over daily (32). Prostate cancer cells also produce prostate cancer-specific antigen, a proteinase that binds readily to ␣ 2 M (33, 34). Thus highly aggressive prostate cancer may secrete prostate cancer-specific antigen, which by binding to ␣ 2 M generates ␣ 2 M* further increasing the concentration of ␣ 2 M* in the tumor microenvironment. Furthermore, tumors may be viewed as existing under ER stress and tumors protect themselves from ER stress by expressing unfolded protein response, of which enhanced GRP78 synthesis is a biomarker (13)(14)(15)(16)(17)(18). A small pool of newly synthesized GRP78 translocates to cell surface from the ER in association with MTJ-1 (35). Therefore, it could be envisaged that, under these conditions, a substantial amount of ␣ 2 M* would be available to bind to cell surface GRP78 thus triggering the activation of mitogenic signaling-dependent cell proliferation. Because it is known that PAKs can be activated via PI 3-kinase signaling (5)(6)(7)(8)11) and membrane localization (5)(6)(7)(8)36), we suggest that activated PAKs may mediate ␣ 2 M*-induced affects on 1-LN prostate cancer cells. Here we demonstrated that ␣ 2 M* mediates PAK-2 activation in 1-LN cells, but PAK-1 is only negligibly affected. We then examined the effect of treating 1-LN cells with ␣ 2 M* on the mechanism of activation of PAK-2, Rac-1, LIMK, and cofilin. We report that exposure of 1-LN cells to ␣ 2 M* induces autophosphorylation of PAK-2, activation of the kinase activity of PAK-2 toward myelin basic protein in a tyrosine kinase and PI 3-kinase-dependent manner, and recruitment of PAK-2 to plasma membrane via the adaptor protein NCK. Rac-1 is also activated. We further demonstrate activation of LIMK and cofilin, which are essential for regulating cytoskeletal organization.

EXPERIMENTAL PROCEDURES
Materials-Culture media were purchased from Invitrogen. Antibodies against PAK-1, PAK-2, and Bad, as well as the phosphorylated forms of PAK-1, PAK-2, LIMK, cofilin, and Bad (Ser 112 or Ser 136 ), were procured from Cell Signaling Technology, Inc. (Beverly, MA). Myelin basic protein and actin antibodies were from Sigma. Anti-GRP78 antibodies were purchased from Stressgen (Victoria, BC, Canada). [␥-33 P]ATP (specific activity, 3000 Ci/mmol) was purchased from PerkinElmer Life Sciences. The sources for the inhibitors used have been described previously. ␣ 2 M* was prepared as described previously. Other reagents used in the study were of analytical reagent quality and were procured locally.
Effect of ␣ 2 M* Stimulation on Activation of PAK-1 and PAK-2 in 1-LN Cells-The highly metastatic human prostate carcinoma cell line 1-LN, derived from less metastatic PC-3 cells, was a kind gift from Dr. Philip Walther (Duke University Medical Center, Durham, NC). Confluent 1-LN cells obtained after overnight incubation in 6-well plates (4 ϫ 10 6 cells/well) were washed twice with HBSS, with a volume of the HBSS added to the monolayers. One set of cells was stimulated with different concentrations of ␣ 2 M*, and cells were incubated as above for different time periods. The other set of cells was stimulated with different concentrations of ␣ 2 M* for 10 min. At the end of incubation, medium was aspirated, and the cells were lysed in lysis buffer containing 20 mM Tris⅐HCl (pH 8.6), 0.1 M NaCl, 1 mM EDTA, 50 mM NaF, 30 mM sodium pyrophosphate, 1 mM sodium orthovanadate, 1 mM phenylmethylsulfonyl fluoride, 20 g/ml leupeptin, and 0.5% Nonidet P-40 for 10 min on ice. The DNA strands were broken by passing the lysates through a 27-gauge needle and syringe several times. The lysates were centrifuged at 800 ϫ g for 5 min at 4°C to remove cell debris. The supernatants were transferred to clean tubes, and their protein contents were determined (37). Equal amounts of lysate proteins were electrophoresed according to Laemmli (38). Proteins from gel (10%) were transferred to a Hybond P® membrane and immunoblotted with antibodies against phosphorylated and unphosphorylated PAK-2 and PAK-1, respectively, according to the manufacturer's instructions. Protein bands on the membrane were visualized by ECF (Amersham Biosciences) and quantified using a Storm 860 PhosphorImager® (Amersham Biosciences). The respective membranes were stripped and reprobed for actin according to the manufacturer's instructions.
␣ 2 M*-induced Autophosphorylation of PAK-1 and PAK-2 in 1-LN Prostate Cancer Cells-Autophosphorylation of PAK-1 and PAK-2 in ␣ 2 M*-stimulated cells was measured essentially as described (39). 1-LN cells were grown in RPMI 1640 medium containing 10 nM insulin, 2 mM glutamine, 10% fetal bovine serum, 12.5 units of penicillin/ml, and 6.5 g/ml streptomycin in 6-well plates at 37°C in a humidified CO 2 incubator till 80 -90% confluent (4 ϫ 10 6 cells/well). The medium was aspirated, the monolayers were washed with chilled HBSS buffer (pH 7.4), and a volume of HBSS was added to the monolayer. The cells were stimulated with ␣ 2 M* (50 pM/10 min) and incubated as above. The reaction was terminated by aspirating the medium. The cells were lysed in a volume of lysis buffer containing 40 mM HEPES (pH 7.4), 1% Nonidet P-40, 100 mM NaCl, 1 mM EDTA, 25 mM NaF, 1 mM sodium orthovanadate, 10 g/ml leupeptin, and 10 g/ml aprotinin over ice for 15 min. The lysates were pipetted into Eppendorf tubes, DNA strands were broken by passing the lysate through a 27-gauge needle several times, and lysates were centrifuged at 1000 rpm for 5 min at 4°C to remove cell debris. The supernatants were transferred to new tubes, and their protein contents were determined (37). To equal amounts of lysate proteins in respective tubes, antibodies against PAK-1 (1:50) and PAK-2 (1:50) were added followed by the addition of Protein A-agarose, and the tubes were incubated overnight at 4°C in a rotary shaker. The immunoprecipitates were recovered by centrifugation (2500 rpm/5 min) at 4°C and washed twice with lysis buffer and thrice with kinase buffer containing 50 mM HEPES (pH 7.5) 10 mM MgCl 2 , 2 mM MnCl 2 , and 0.2 mM dithiothreitol. Autophosphorylation was measured in 50 l of kinase buffer containing 10 Ci of [␥-33 P]ATP (specific activity, 3000 Ci/mM) for 20 min at 30°C. The reaction was stopped by adding a volume of 4ϫ sample buffer and heating the tubes for 3 min at 90°C. The tubes were centrifuged, and protein was fractionated on a 10% gel according to Laemmli (38). The protein bands on the gel were transferred to membranes, the membranes were dried, and 33 P-labeled PAK-1 and PAK 2 were detected and quantified by autoradiography in the Storm 860 PhosphorImager®.
PAK-2 Kinase Activity in PAK-1 and PAK-2 Immunoprecipitates from ␣ 2 M*-stimulated 1-LN Cells-The kinase activities of PAK-2 toward myelin basic protein were determined essentially as described before (39). Briefly, 1-LN lysates from respective groups were immunoprecipitated with PAK-2 antibodies as above. The immunoprecipitates were washed thrice with lysis buffer and then thrice with 2ϫ kinase buffer as above. To respective immunoprecipitates in 50 l of kinase buffer were incubated with 5 g of myelin basic protein (MBP) for 5 min over ice. The kinase reaction of immunoprecipitates was initiated by the addition of 10 Ci of [␥-33 P]ATP (specific activity, 3000 Ci/mM) followed by the addition of ATP (20 M final concentration). The samples were incubated for 10 min at 25°C. The reaction was terminated by the addition of 1 volume of 4ϫ sample buffer. In experiments where the effects of tyrosine kinase and PI 3-kinase inhibitors were examined on PAK kinase activity, these were added before stimulation with ␣ 2 M*. Other details were identical to those described above. The samples were heated for 3 min at 90°C, electrophoresed on 12.5% gel, transferred to membrane and 33 P-labeled MBP, and visualized and quantitated by autoradiography in the Storm 860 PhosphorImager®.

Modulation of ␣ 2 M*-induced PAK-2 Activation in 1-LN Cells by Protein Kinases-In
experiments where the effects of inhibiting tyrosine kinases and of PI 3-kinase was studied on expression of activated PAK-2 by Western blotting, the specific inhibitors of these kinases were added to respective wells (4 ϫ 10 6 /cells/well) as described above, and cells were incubated for a specific time period before adding ␣ 2 M* (50 pM) and incubating them further for 10 more min. The reaction was stopped by aspirating the medium and lysing the cells in lysis buffer as described above. Equal amounts of lysate proteins were electrophoresed according to Laemmli (38). Proteins from gel (10%) were transferred to Hybond P® membrane and immunoblotted with antibodies against phosphorylated and unphosphorylated PAK-2, respectively, according to the manufacturer's instructions. Protein bands on the membrane were visualized by ECF and quantified using the PhosphorImager®. The respective membranes were stripped and reprobed for actin according to the manufacturer's instructions.
Assay for Rac⅐GTP-Active GTP-bound Rac was precipitated employing a PAK-PBD-based assay kit (Upstate Cell Signaling, Lake Placid, NY) (40). Confluent 1-LN cells (4 ϫ 10 6 well) in 6-well plates in RPMI 1640 medium containing 10% FBS, penicillin (12.5 units/ml), streptomycin (6.5 g/ml), 2 mM glutamine, and 10 nM insulin, were stimulated with ␣ 2 M* (50 pM/10 min) at 37°C in a humidified CO 2 (5%) incubator. The reaction was stopped by aspirating the medium and washing cells twice with ice-cold HBSS buffer (pH 7.4). The cells were lysed by adding a buffer containing 50 mM Tris⅐HCl (pH 7.5), 120 mM NaCl, 25 nM NaF, 1 mM sodium orthovanadate, 1 mM phenylmethylsulfonyl fluoride, 1 mM benzamidine, 10 g of leupeptin/ml, and 1% Nonidet P-40 on ice for 10 min. The lysates were transferred into Eppendorf tubes, DNA strands were broken by passing the lysate through a 27-gauge needle several times, and the lysates were centrifuged at 1000 rpm for 5 min at 4°C to remove cell debris. The supernatants were transferred to new tubes, and their protein contents were determined (37). To equal amounts of lysate proteins in the respective tubes, 40 l of PAK-PBD agarose was added, and the tubes incubated for 1 h at 4°C with gentle rotation. The tubes were centrifuged at 3500 rpm for 10 min at 4°C. The agarose pellet was washed thrice with the above lysate buffer. To the agarose pellets 40 l of reducing sample buffer was added, the tubes were heated at 90°C for 5 min, then centrifuged briefly, and the supernatant was processed for protein fractionation on a 10% gel according to Laemmli (38). The protein bands on the gel were transferred to Hybond P® membranes and immunoblotted with antibodies against Rac-1 (Santa Cruz Biotechnology, Santa Cruz, CA). Protein bands on the membranes were visualized by ECF and quantified using the PhosphorImager®. An aliquot of lysate was similarly processed for total Rac quantification.
Translocation of PAK-2 to Plasma Membrane in 1-LN Cells Stimulated with ␣ 2 M*-Confluent monolayers of 1-LN cells grown as above in RPMI medium in 4-well plates (12 ϫ 10 4 cells/well) were washed with HBSS twice, and a volume of HBSS was added to the monolayers. The cells were treated with ␣ 2 M* (50 pM) and incubated for 10 min as above. At the end of incubation, the medium was aspirated, and to the cells was added a volume of chilled HBSS buffer containing 10 mM Tris⅐HCl, pH 7.5, 10 mM NaCl, 1 mM phenylmethylsulfonyl fluoride, 10 M benzamidine, and 10 M leupeptin. The cells were scraped into a chilled glass homogenizing tube, and the membrane fractions highly enriched in plasma membrane were isolated as described previously. Briefly, the cells were homogenized by 30 up and down strokes with a Teflon pestle at 4°C. The homogenate was centrifuged at 600 ϫ g for 5 min at 4°C and the pellet discarded. The supernatant was layered onto a sucrose step gradient of 50 and 30% (3 ml each) and centrifuged at 200,000 ϫ g for 75 min in a Beckman Coulter Ultracentrifuge (Optima LE80) at 4°C. The membrane fraction at the interface between the sucrose layers was removed and suspended in a volume of incubation buffer containing 25 mM HEPES, pH 7.4, 10 mM KCl, 3 mM NaCl, 5 mM MgCl 2 , 2 M leupeptin, and 1 M Ca 2ϩ . The suspension was centrifuged at 400,000 ϫ g for 90 min as above. The pellet was suspended in a volume of incubation buffer. The enrichment of membrane preparation for plasma membrane was assessed as described previously, including by electron microscopy (41). These analyses showed this membrane fraction was highly enriched in plasma membranes (92-95%), and hence we designated this "membrane fraction" as the "plasma membrane fraction." The membrane pellet was lysed in lysis buffer, and the lysate was immunoprecipitated with antibodies against GRP78 (1:100) and GRP78 in the membranes quantified as above. The membranes on which these preparations were transferred after electrophoresis were reprobed and quantified for phosphorylated PAK-2 and NCK according to the manufacturer's instructions.
Chemical Synthesis of dsRNA Homologous in Sequence to the Target GRP78 Gene-The chemical synthesis of dsRNA homologous in sequences to the target GRP78 were as follows: 1) 370 KIQQLVK 376 , mRNA sequence 5Ј-AAA ATA CAG CAA TTA GTA AAG-3Ј and 2) 521 KNK-ITIT 527 , mRNA sequence 5Ј-AAG AAT AAA ATA ACA ATA ACA-3Ј peptides (Swiss-Prot GRP primary sequence accession number P11021). These were performed by Ambion (Austin, TX). For making dsRNA of the first mRNA sequence (sense (5Ј-AAU ACA GCA AUU AGU AAA GTT-3Ј) and antisense (5Ј-CUU UAC UAA UUG CUG UAU UTT-3Ј)) oligonucleotides and the second mRNA sequence (sense (5Ј-GAA UAA AAU AAC AAU AAC ATT-3Ј) and antisense (5Ј-UGU UAU UGU UAU UUU AUU CTT)) were annealed according to the manufacturer's instructions. Throughout the entire period of experimentation, handling of reagents was performed in an RNase-free environment. Briefly, equal amounts of sense and antisense oligonucleotides were mixed in annealing buffer and heated at 90°C for 1 min then maintained for 1 h at 37°C in an incubator. The dsRNA preparation was stored at Ϫ20°C before use.
Transfection of 1-LN Cells with dsRNA Homologous in Sequence to GRP78 Gene and Effect of ␣ 2 M* on PAK-2 and NCK-Confluent 1-LN cell monolayers (1.5 ϫ 10 6 /well in 6-well plates) incubated as described above were washed twice with HBSS and 2 ml of DMEM containing 10% of FBS and the above mentioned antibiotics added, and cells were incubated as above for 16 h. Just before each transfection, 25 g of both GRP78 dsRNA was diluted to 100 l of serum-and antibiotic-free DMEM in a tube. In another tube, 10 l of Lipofectamine was diluted into 100 l of serum-and antibiotic-free medium. The two solutions were combined, mixed gently, and incubated for 45 min at room temperature followed by the addition of 800 l of serum-and antibiotic-free medium to each tube in separate experiments. The monolayers were washed twice with serum-antibiotic-free DMEM, layered in each well with 1 ml of Lipofectamine-DMEM or lipid dsRNA mixtures containing 25 g of dsRNA of each GRP78 target mRNA gently mixed and incubated for 5 h at 37°C in a humidified CO 2 incubator in separate experiments. At the end of incubation, 1 ml of antibiotic-free DMEM containing 10% FBS was added to each well, and cells incubated for 16 h as above. Microscopic observations of the monolayers did not show evidence of toxicity consistent with previous studies. The medium was replaced with DMEM containing antibiotics and 10% FBS 24 h following the start of the transfection. The monolayers were incubated for a further 24 h as above. At the end of incubation, medium was aspirated and monolayers were washed with the above medium once, a volume of the same medium was added, and the cells were used for the experiment outlined below. To demonstrate that the transfection of 1-LN prostate cancer cells with dsRNA homologous in sequence to target GRP78 gene does not produce any nonspecific effects on target gene expression, the 1-LN cells were transfected with equimolar concentrations of scrambled small interference RNA (Silencer TM negative control, catalog number 4610, Ambion) under identical conditions as described above for transfection with GRP78 dsRNA. At the end of transfection period (48 h), the medium was aspirated and a volume of DMEM was added, and the cells were either stimulated with buffer or ␣ 2 M* (50 pM) for 10 min. The reaction was stopped by aspirating the medium and adding a volume of lysis buffer as described above. Equal amounts of lysate proteins (37) were electrophoresed according to Laemmli (38). Proteins from gel (10%) were transferred to a Hybond P membrane and immunoblotted with antibodies against GRP78, unphosphorylated PAK-2, and NCK, respectively, according to the manufacturer's instructions. Protein bands on the membrane were visualized by ECF (Amersham Biosciences) and quantified using the PhosphorImager®. The respective membranes were stripped and reprobed for actin according to the manufacturer's instructions.
Chemical Synthesis of dsRNA Homologous in Sequence to the Target PAK-2 Gene-The chemical synthesis of dsRNA homologous in sequence to the target PAK-2 ( 382 KLTDFGF 388 , mRNA sequence, 5Ј-AAA TTA ACA GAT TTT GGA TTT-3 peptide; Swiss-Prot primary accession number Q8CIN4) was performed by Ambion. For making dsRNA, the sense (5Ј-AUU AAC AGA UUU UGG AUU UTT-3) and antisense (5Ј-AAA UCC AAA AUC UGU UAA UTT-3) oligonucleotides were annealed according to the manufacturer's instructions as described above for GRP78 above.
Transfection of 1-LN Cells with dsRNA Homologous in Sequence to PAK-2 Gene and Effect of ␣ 2 M* on Phosphorylated LIMK-Confluent 1-LN cells monolayers (1.5 ϫ 10 6 /well in 6-well plates) incubated as described above were washed twice with HBSS and 2 ml of DMEM containing 10% FBS, the above mentioned antibiotics were added, and the cells were incubated as above for 16 h. For each transfection, 25 g of PAK-2 dsRNA was used and cells were transfected as described above for dsGRP78 dsRNA with Lipofectamine with or without dsPAK-2. The medium was replaced with DMEM containing antibiotics and 10% FBS 24 h following the start of the transfection. The monolayers were incubated for a further 24 h as above. Microscopic observation of the transfected monolayers did not show evidence of toxicity expect that 1-LN cells transfected with dsPAK-2 RNA showed impairments in the uniformity of spreading of monolayers. At the end of incubation, medium was aspirated, monolayers were washed with the above medium once, a volume of same medium was added, the cells were stimulated with ␣ 2 M* then lysed, and lysates were processed for phosphorylated LIMK, PAK-2, and LIMK protein assay by Western blotting as described above. Protein bands on the membrane were visualized by ECF (Amersham Biosciences) and quantified using the PhosphorImager®. To demonstrate that the transfection of 1-LN prostate cancer cells with dsRNA homologous in sequence to target PAK-2 gene does not produce any nonspecific effects on target gene expression, the 1-LN cells were transfected with equimolar concentrations of scrambled small interference RNA, under identical conditions as described above. At the end of transfection (48 h), the medium was aspirated, a volume of DMEM was added, and the cells were stimulated with either buffer or ␣ 2 M* and processed as above for the quantification of phosphorylated LIMK, PAK-2, and LIMK protein by Western blotting.

RESULTS
Binding of ␣ 2 M* to Cell Surface-associated GPR78 Enhances Phosphorylation of PAK-2 in 1-LN Cells-PAK activity is stim-ulated in response to a variety of extracellular stimuli, including chemoattractants acting on G protein-coupled receptors, growth factors interacting with receptor tyrosine kinases, cytokines, and extracellular matrix molecules binding to integrins (5)(6)(7)(8). We have previously shown that ␣ 2 M* promotes cellular growth of 1-LN prostate cancer cells (17,18). To understand the possible involvement of PAK-1 and PAK-2 in ␣ 2 M*-stimulated cellular growth, we first examined the levels of phosphorylated PAK-1 and PAK-2 in 1-LN cells treated with ␣ 2 M* for different periods of time and with varying concentrations of ␣ 2 M* ( Fig. 1 and Table I). In the Western blot, phosphorylated PAK-2 is seen as a doublet of 58 -60 kDa (Fig. 1). This appears to represent differential phosphorylation of PAK-2 as previously reported. The maximal increase in phosphorylated PAK-2 occurred at about 10 min of incubation and declined slowly thereafter, whereas the maximal increase in levels of phosphorylated-PAK-2 occurred at about 50 -100 pM ␣ 2 M*. Surprisingly, incubation of 1-LN cells with ␣ 2 M* under these conditions only showed a negligible effect on phosphorylation of PAK-1. Similar to the results described above, stimulation of 1-LN cells with ␣ 2 M* predominantly caused autophosphorylation of PAK-2, but negligible autophosphorylation of PAK-1 ( Fig. 2A).
␣ 2 M* Stimulates the Kinase Activity of PAK-2 in 1-LN Prostate Cancer Cells-In the next series of experiments we studied the activation of only PAK-2 by measuring its kinase activity toward MBP (Fig. 2B). Treatment of 1-LN prostate cancer cells with ␣ 2 M* increased PAK-2-dependent phosphorylation of MBP by 2-to 3-fold (Fig. 2B).
␣ 2 M* Activates Rac-1 in 1-LN Cells-p21-activated protein kinases are phosphorylated by small G proteins such as Rac-1 and CdC42 in the presence of GTP, which binds to the G protein binding site in the NH 2 -terminal regulatory domain. In the preceding section we showed that stimulation of 1-LN cells with ␣ 2 M* induces phosphorylation and the kinase activity of PAK-2 (Fig. 2, A and B). We next determined the activation of Rac-1, by quantifying the levels of Rac-1⅐GTP by Western blotting (Fig. 2C). Indeed ␣ 2 M* treatment of 1-LN cells elevated the levels of Rac-1⅐GTP by about 2-fold compared with unstimulated cells (Fig. 2C). The results show the importance of Recruitment of PAK-2 to Plasma Membranes via Interaction with the Adaptor Protein NCK in ␣ 2 M*-stimulated 1-LN Prostate Cancer Cells-In response to external stimuli, PAKs, which contain variable SH3 binding motifs/sites, can interact with the SH3-containing adaptor protein NCK (5)(6)(7)(8). This protein is involved in the recruitment of PAKs to activated tyrosine kinase receptors in plasma membranes. NCK is either constitutively associated with PAK-2 or its association is induced (5-8, 36, 42). Interaction of the adaptor protein NCK and PAKs has been implicated in its translocation and stimulation of PAK activity by growth factors (see Refs. 5-8, 36, and 42). We have evaluated the involvement of the adaptor protein NCK in the translocation of PAK-2 to plasma membrane GRP78 in 1-LN cells upon stimulation of cells with ␣ 2 M* (Fig.  3, A-C). GRP78 immunoprecipitate of plasma membrane lysates showed that GRP78, NCK, and PAK-2 are coimmunoprecipitated in 1-LN cells stimulated with ␣ 2 M* (Fig. 4, A-C). These results demonstrate that, in plasma membranes isolated from 1-LN cells stimulated with ␣ 2 M*, PAK-2 exists in complex with the adaptor protein NCK and the receptor GRP78.
Silencing of GRP78 Gene Expression with RNA Interference Attenuates the Association of PAK-2 and NCK in ␣ 2 M*-stimulated Cells-In the preceding section, we demonstrated the presence of PAK-2 in plasma membranes in association with the adaptor protein NCK and GPR78 after ␣ 2 M* stimulation (Fig. 3, A-C). Ligand-induced tyrosine phosphorylation of the receptor causes the recruitment of SH3-and SH2-containing adaptor proteins, and forms a multiprotein complex responsible for generating and propagating intracellular signaling responsible for cellular responses. To assess the role of ligandactivated GRP78 in the recruitment of NCK-PAK-2 complex to the plasma membrane, we silenced the expression of GRP78 gene with dsRNA homologous in sequence to the target gene mRNA sequence (5Ј-AAA ATA CAG CAA TTA GTA AAG-3Ј) and assayed the association of PAK-2 and NCK in cell lysates in ␣ 2 M*-stimulated 1-LN cells (Fig. 3, D and E, and Table II). Silencing GRP78 gene expression with this target sequence greatly attenuated the association of PAK-2 and NCK in these cells. To further ascertain that these effects are specific to GRP78 gene silencing, we next employed dsRNA of a second target gene sequence, 5Ј-AAG AAT AAA ATA ACA ATA ACA-3Ј, to silence the expression of GRP78 gene. This approach also profoundly reduced the levels of GRP78 protein (Fig. 3G) compared with cells treated with ␣ 2 M* alone (Fig. 3G). If GRP78 is in complex with NCK and PAK-2, then limitations imposed on the availability of GRP78, would also limit the levels of associated NCK and PAK-2. Indeed this was found in studies where silencing of GRP78 gene expression with dsRNA greatly attenuated the levels of GRP78 (Fig. 3I), PAK-2 (Fig. 3I), and NCK (Fig. 3K) employing the immunoprecipitation protocol described above. The specificity of silencing GPR78 gene expression as a cause of these effects is further supported by the use of a scramble dsRNA in control experiments (Fig. 3) under identical conditions. Scrambled dsRNA showed none or negligible effects on GRP78, PAK-2, or NCK (Fig. 4, D, E, G, I, J, and  K, and Table II). The results presented confirm that ligandinduced activation of GRP78 is required for recruitment of PAK-2 in association with NCK to the plasma membrane. The mechanism by which PAK-2 is activated at the membrane is not clearly understood; however, it has been reported that mere membrane localization of PAK-2 is sufficient for its activation.
␣ 2 M* Up-regulates Levels of Phosphorylated LIMK and Cofilin in 1-LN Prostate Cancer Cells-Reorganization of the cytoskeleton is an essential feature of motility, detachment, and invasion by cancer cells. Formation and stabilization of actin filaments provide the protrusive force for cellular extension of the leading edge of migratory cells (43)(44)(45). The Rho family of GTPases, Rho, Rac, and Cdc42 regulate cell morphology, cytokinesis, and cell motility through reorganization of actin filaments. The interplay between these GTPases is critically involved in the regulation of cell morphology and motility. Activation of Rac enhances cell spreading and migration by stimulation of actin polymerization at the plasma membrane and promoting lamellipodia formation. Rho stimulates contractibility and adhesion by inducing the formation of actin stress fibers and focal adhesions. Rho-associated protein kinase and Rho-associated protein kinase are downstream effectors of Rho to form stress fibers and focal adhesions. Rho kinases also activate LIM kinase, which phosphorylates cofilin and thereby stabilizes actin stress fibers. LIM kinases are dual specificity (Ser/Thr and Tyr) kinases that contain two NH 2 -terminal LIM domains, which are commonly associated with the actin cytoskeleton (46,47). LIMK mediates specifically Rac-induced actin cytoskeleton reorganization and focal adhesion complexes. Rac-induced activation of LIMK is mediated by PAK-1, which phosphorylates LIMK at its Thr 508 residue. Activated LIMK specifically phosphorylates cofilin, an actin-binding protein that promotes the disassembly of actin filaments (5-8, 42-48). Phosphorylation of cofilin inhibits its actin depolymerizing action and hence provides a mechanism by which LIMK could regulate the assembly of actin (49). LIMK mRNA is up-regulated in prostate adenocarcinomas and is correlated with the aggressiveness of these cells (48). Active PAKs greatly enhanced phosphorylation of both LIMK and cofilin in vitro (20 -23, 48, 49). In the proceeding sections we demonstrated that treatment of 1-LN prostate cancer cells with ␣ 2 M* activates PAK-2. 1-LN cells are highly malignant, motile, and invasive; therefore, to understand the involvement of LIMK in cytoskeleton reorganization in 1-LN cells upon treatment of ␣ 2 M*, we have determined the levels of phosphorylated LIMK (Fig. 4A and Table III) and phosphorylated cofilin (Fig. 4D and Table III) by Western blotting. A 2-to 3-fold increase in the levels of phosphorylated LIMK and phosphorylated cofilin was observed in 1-LN cells treated with ␣ 2 M*. The results indicate the involvement of phosphorylated-PAK-2 in cytoskeleton reorganization and possible protrusive behavior of 1-LN prostate cancer cells treated with ␣ 2 M* via phosphorylation of LIMK and cofilin. The results indicate that, upon treatment of 1-LN cells with ␣ 2 M*, the phosphorylation of LIMK and cofilin is markedly increased and suggests that these effects are possibly mediated by LIMK under our experimental conditions. Because LIMK is also activated by Rho kinases, to further understand the role of PAK-2 in LIMK activation, we silenced the expression of the PAK-2 gene by RNA interference (Fig. 4). Silencing of PAK-2 gene expression (Fig. 4G) attenuated the levels of phosphorylated LIMK (Fig. 4H) without affecting the levels of LIMK protein (Fig. 4I) compared with cells treated with ␣ 2 M* alone (Fig. 4, G-I) or cells transfected with scramble dsRNA and treated with ␣ 2 M* (Fig. 4 , G-I). The reductions in the protein levels of PAK-2 and phosphorylated LIMK were nearly comparable, which suggest that, in 1-LN prostate cancer cells, ␣ 2 M*-induced PAK-2 is involved in reorganization of the cytoskeleton via LIMK and cofilin. However, the role of Rho kinase in phosphorylation of LIMK in 1-LN cells is not ruled out.
␣ 2 M* Treatment of 1-LN Prostate Cancer Cells Up-regulate the Increased Expression of Phosphorylated Bad (Ser 112 and Ser 136 )-Different PAK family members regulate the balance between pro-survival and pro-apoptotic proteins of the Bcl-2 family (8). As noted above, caspase-induced cleavage of PAK-2 releases the constitutively active 34-kDa catalytic COOH-terminal fragment in response to multiple stimuli that induce apoptosis; however, full-length activated PAK-2 is anti-apoptotic (5)(6)(7)(8)11). Members of the Bcl-2 family are intracellular proteins that can either promote survival (Bcl-2 and Bcl XL ) or augment cell death (Bad and Bax) (50 -53). Bad binds to Bcl-2 and neutralizes the anti-apoptotic effects of Bcl-2 and promotes cell death. Phosphorylation of Bad causes its interaction with 14-3-3 protein and prevents its binding to Bcl-2, which then interacts with Bax to inhibit apoptosis. Bad is phosphorylated on Ser 112 , Ser 136 , and Ser 155 by protein kinases, including MAPKs and Akt (54,55), the latter of which is the effector of PI 3-kinase, and PAK-1 phosphorylates Bad at Ser 136 either directly or indirectly through PAK-2. Treatment of 1-LN prostate cancer cells with ␣ 2 M* profoundly elevated the levels of Bad phosphorylated at Ser 112 and Ser 136 (Fig. 5, A and B, and Table  III) with kinetics similar to that PAK-2 activation (Fig. 1). These studies suggest that PAK-2 is involved in the cellular proliferative effects of ␣ 2 M* observed in 1-LN prostate cancer cells by up-regulating the levels of antiapoptotic proteins. these pathways (16 -20). Activation of Ras and PI 3-kinase in macrophages stimulated with ␣ 2 M* induced increase expression of mitogenic signaling culminating in an increased DNA synthesis and cellular proliferation (17,19). In view of the functional dependence of PAK-2 activation on receptor tyrosine phosphorylation, and PI 3-kinase activation, we examined the effect of inhibiting tyrosine phosphorylation of GPR78 and PI 3-kinase activation consequent to ␣ 2 M* binding to 1-LN cells on the activation of PAK-2 by measuring its kinase activity toward MBP (Fig. 6A and Table IV) and then expression of phosphorylated PAK-2 ( Fig. 6B and Tables IV and V). Pretreatment of cells with genistein, a specific inhibitor of tyrosine kinases greatly attenuated the ␣ 2 M*-induced MBP phosphorylating activity of PAK-2 ( Fig. 6A and Table IV) as well as the levels of activated PAK-2 by Western blotting. Likewise, inhibition of PI 3-kinase with its specific inhibitor LY294004 profoundly inhibited the MBP-phosphorylating activity of PAK-2 ( Fig. 6A and Table IV) as well as the elevated levels of phosphorylated PAK-2 ( Fig. 6B and Tables IV and V) in 1-LN cells. The data presented show that ␣ 2 M*-induced increased activation of PAK-2 in 1-LN cells requires tyrosine phosphorylation of the receptor and activation of downstream PI 3-kinase signaling. Prior treatment of cells with genistein drastically attenuated kinase activity of PAK-2. Likewise, LY294002 also inhib-ited kinase activity of PAK-2, but the magnitude of inhibition was smaller than that of tyrosine kinase inhibition, which suggests that PI 3-kinase-independent mechanism may be involved in PAK-2 activation. DISCUSSION The development, progression, and metastasis of prostate cancer is a multistage phenomenon where the role of cellular metabolism, environment, and intracellular signaling play crucial roles. In this study, we have examined PAK activation in 1-LN cells treated with ␣ 2 M* and various PAK-induced downstream signaling events. The salient observations of this study are that ␣ 2 M* binding to 1-LN cells: 1) up-regulates activation of PAK-2; 2) induces autophosphorylation of PAK-2, and the tyrosine kinase-and PI 3-kinase-dependent kinase activity of PAK-2; 3) induces activation of Rac-1; 4) increases phosphorylation of LIMK and cofilin, which are greatly reduced upon silencing of PAK-2 gene expression; 5) promotes NCK-mediated   recruitment of PAK-2 to plasma membrane-localized GPR78; and 6) regulates phosphorylation of Bad at Ser 112 and Ser 136 . Finally, silencing expression of the GRP78 gene nearly abolishes the association of PAK-2 and NCK in the plasma membranes of ␣ 2 M*-stimulated 1-LN cells. The data presented suggest that PAK-2 is involved in ␣ 2 M*-induced cellular growth, migration, and the survival of 1-LN prostate cancer cells. The mechanism by which ␣ 2 M* activates PAK-2 appears to be dependent on receptor tyrosine phosphorylation and PI 3-kinase activation, because pretreatment of cells with genistein, LY294002, and wortmannin greatly attenuated the activation of PAK-2. Because silencing of the expression of the GRP78 gene also greatly attenuated the activation of PAK-2 in ␣ 2 M*treated 1-LN cells, ligand-induced activation of GRP78 is a prerequisite for PAK-2 activation in these cells. That the activation of PAK-2 in 1-LN cells under these conditions was not completely abolished also suggests the involvement of other mechanisms in PAK-2 activation.
Highly malignant and invasive androgen-independent human 1-LN prostate cancer cells were derived from less malignant and noninvasive PC-3 cells at this University a number of years ago. Unlike PC-3 prostate cancer cells, 1-LN cells in nude mice migrate and metastasize. 1-LN prostate cancer cells also differ from PC-3, LnCap, and DU145 cancer cells in one very important respect; namely when 1-LN, cells but not the other cell lines, are exposed to picomolar concentrations of ␣ 2 M*, cell proliferation increases substantially (16,18,19). The expression of GRP78 on the surface of PC3, DU145, and LnCap cells was either absent or minimal (17,18). By contrast, the highest levels that we have detected are on the surface of 1-LN cells, which were employed to purify ␣ 2 M* signaling receptor and identify it as GRP78 (20). Preincubation of 1-LN cells with antibodies against GRP78, silencing of the expression of the GRP78 gene and silencing of the expression of MTJ-1, with which GRP78 associates, greatly attenuated ␣ 2 M*-induced mitogenic signaling and cellular responses (16,17,19,35). These observations suggest a functionally close relationship between cell surface-associated GRP78 receptor activation and growth, metastasizing, and the invasive potential of 1-LN prostate cancer cells. Ligation of this cell surface-associated GRP78 on 1-LN cells with ␣ 2 M* triggers the activation of mitogenesis and cell proliferative secondary to inositol 1,3,4-trisphosphate/Ca 2ϩ signaling, Ras/MAPK signaling, and PI 3-kinase signaling (17,18). In our previous reports we have shown that cellular responses elicited post ␣ 2 M* binding to cell surface-associated GRP78 are analogous to various growth factors and its receptor behaves like a growth factor receptor. In androgen-independent 1-LN prostate cancer cells, ␣ 2 M* could function as a growth factor (16 -20). This is evidenced by increased DNA synthesis and cellular proliferation observed in 1-LN cells treated with ␣ 2 M* (16). Both these events are dependent upon tyrosine phosphorylation of GRP78, and receptor downstream Ras/MAPK and PI 3-kinase signaling. Based on positive correlation between circulating GRP78 antibodies and prostate cancer, GRP78 has been suggested as a diagnostic biomarker of prostate cancer (30,31). Prostate cancer cells also produce prostate cancer-specific antigen and matrix metalloproteinases (33,34), which bind readily to ␣ 2 M whose daily turnover has been estimated to be about 1 g (32). Thus highly aggressive prostate cancer may produce prostate cancer-specific antigen   Fig. 6 (A and B). converting ␣ 2 M to ␣ 2 M*, which would then be available to bind to cell surface-associated GRP78 on prostate cancer cells promoting their growth, metastasizing, and invasive potential mediated by mitogenic signaling elicited consequent to ␣ 2 M* binding to GRP78. Taken together with the clinical data cited above, it could be argued that up-regulation of ␣ 2 M* binding receptor ␣ 2 M* signaling receptor (GRP78) is part of the aggressive phenotype in prostate cancer. PAK-2 is ubiquitously expressed, whereas the expression of PAK-1 is tissue-specific (5)(6)(7)(8). In dividing cells PAK-2 is inactive but is transiently activated when cells are subjected to moderate stress conditions such as hyperosmolarity, ionizing radiations, and DNA-damaging drugs (5)(6)(7)(8). Under these conditions, PAK-2 activation requires upstream tyrosine kinase and PI 3-kinase activities. GRP78 is an ER-resident protein that is involved in many cellular processes, including antigen presentation, translocation of newly synthesized polypeptides across the ER membrane, and their subsequent folding, maturation, transport, or retrotranslocation (21)(22)(23)(24)(25)(26). An increased expression of GRP78 protein is a part of unfolded protein response required to alleviate ER stress, maintain ER function, and protect cells against cell death. GRP78 is constitutively expressed, but its synthesis can be induced by a variety of stressful conditions that perturb protein folding and assembly within the ER, including glucose deprivation, acidosis, and hypoxia, conditions that are generally present in poorly vascularized solid tumors (21)(22)(23)(24)(25)(26). The kinase activity of PAKs has been implicated in proliferative signaling by growth factor receptor kinases that in turn regulate cell survival, programmed cell death, and malignant transformation (5)(6)(7)(8). The dominant role of PI 3-kinase signaling in cell survival and proliferation is well documented (53). Akt, a downstream effector of PI 3-kinase, attenuates the apoptotic events by phosphorylating apoptotic proteases, apoptotic protein Bad, and transcription factor FOX01/FOX02/FOX03 and thus promotes cell survival (53). In 1-LN cells ␣ 2 M* promotes cellular proliferation by activating Ras/MAPK and PI 3-kinase signaling and protects cells from cell death by phosphorylating Bad at Ser 112 and Ser 136 . Reorganizing of cytoskeleton is an essential feature of motility, detachment, and invasion of cancer cells. Formation and stabilization of actin filaments provide the protrusive force for cellular extension of the leading edge of migratory cells (43)(44)(45). PAKs effect cytoskeletal reorganization via phosphorylating LIMK, which phosphorylates cofilin. These events are essential in the dynamics of actin assembly and disassembly (5)(6)(7)(8). It can be inferred from these studies that ␣ 2 M*-mediated activation of PAK-2 confers migratory properties to these cells thus promoting their invasive and metastatic properties. PI 3-kinases and their products have also been implicated in the regulation of the cytoskeleton (5)(6)(7)(8).
In conclusion we show here for the first time that binding of ␣ 2 M* to cell surface-associated GRP78 activates PAK-2 in highly metastatic and invasive prostate cancer cells. ␣ 2 M*induced activation of PAK-2 requires tyrosine phosphorylation of GRP78, activation of Ras/MAPK and PI 3-kinase downstream signaling, and recruitment of PAK-2 via the adaptor protein NCK to the plasma membrane. The increased expression of phosphorylated LIMK and cofilin confers motility, necessary for metastasis, in 1-LN cells. ␣ 2 M* treatment of 1-LN cells also protects these cell from cell death by inhibiting the pro-apoptotic protein Bad. The results of this study are summarized in Fig. 7.