Promotion of Cancer Cell Migration

A family of six high affinity IGF-binding proteins (IGFBPs 1–6) plays an important role in modulating IGF activities. Recent studies suggest that some IGFBPs may have IGF-independent effects, including induction of apoptosis and modulation of cell migration. However, very little is known about possible IGF-independent actions of IGFBP-6. We have generated a non-IGF-binding IGFBP-6 mutant by substituting Ala for four amino acid residues (Pro93/Leu94/Leu97/Leu98) in its N-domain IGF-binding site. A >10,000-fold loss of binding affinity for IGF-I and IGF-II was observed using charcoal solution binding assay, BIAcore biosensor, and ligand blotting. Wild-type and mutant IGFBP-6, as well as IGF-II, induced cell migration in RD rhabdomyosarcoma and LIM 1215 colon cancer cells. Cell migration was mediated by the C-domain of IGFBP-6. Transient p38 phosphorylation was observed in RD cells after treatment with IGFBP-6, whereas no change was seen in phospho-ERK1/2 levels. Phospho-JNK was not detected. IGFBP-6-induced cell migration was inhibited by SB203580, an inhibitor of p38 MAPK, and PD98059, an inhibitor of ERK1/2 MAPK activation. In contrast, SP600125, a JNK MAPK inhibitor, had no effect on migration. Knockdown of p38 MAPK using short interfering RNA blocked IGFBP-6-induced migration of RD cells. These results indicate that p38 MAPK is involved in IGFBP-6-induced IGF-independent RD cell migration.

Insulin-like growth factors (IGF 2 -I and -II) are important regulators of cell proliferation, differentiation, and survival. Dysregulation of the IGF system is implicated in many diseases, including cancer, diabetes, and atherosclerosis. A family of six IGF-binding proteins (IGFBPs 1-6) modulates IGF activity through high affinity binding that sequesters IGFs from their receptors, thereby inhibiting their actions. In some cases, IGFBPs may also potentiate IGF action by regulating IGF delivery to its receptors. As well as directly binding to IGFs, IGFBPs also modulate their actions indirectly via interactions with a number of plasma, extracellular matrix, and cell surface molecules. Some IGFBPs have IGF-independent effects, including inhibition of growth, and modulation of cell migration and apoptosis (1).
IGFBPs share conserved N-and C-domains, each of which contain multiple conserved disulfide bonds that are essential for maintaining protein conformation and biological activity. A nonconserved and unstructured L-domain lies between these domains. Previous studies using NMR spectroscopy and crystallization of an N-terminal subdomain of IGFBP-5 with IGF-I demonstrated an IGF-binding site containing a number of hydrophobic residues (2,3). A recent crystal structure of a complex comprising the N-domain of IGFBP-4 and IGF-I confirmed this binding site (4). Substitution of these hydrophobic residues in IGFBP-3 and -5 decreased their IGF-I binding affinities Ͼ1000-fold and diminished their effects on IGF actions (5,6). However, another study showed residual IGF binding after similar mutations in IGFBP-3, suggesting other regions are involved in IGF binding (7). Indeed, C-domain residues of IGFBPs also play a role in high affinity IGF binding (8 -10). Most IGF-independent effects of IGFBPs have been localized to their C-domains. IGFBP-6 is unique among the IGFBPs as it has an ϳ50-fold higher binding affinity for IGF-II than IGF-I, making it a relatively specific IGF-II inhibitor (11). Because of its binding preference, IGFBP-6 inhibits IGF-II-induced cell proliferation, differentiation, and survival (12)(13)(14)(15)(16), but it has marginal or no effect on IGF-I actions. A few studies have suggested that IGFBP-6 may have IGF-independent effects, but they have not been definitively demonstrated (17)(18)(19)(20). In this study, we have generated a non-IGF-binding IGFBP-6 mutant (mIGFBP-6) and used it as a tool to study possible IGF-independent effects of IGFBP-6. We show that mIGFBP-6 and wild-type IGFBP-6 (wtIGFBP-6) both promote migration of cancer cells in an IGFindependent manner. Furthermore, the migratory effects of IGFBP-6 are mediated by its C-domain. We also present evidence suggesting that IGFBP-6 activates p38 MAPK, which in turn is involved in IGFBP-6-stimulated migration of human RD rhabdomyosarcoma cells.

Mammalian Cell Culture
RD cells, derived from a human pelvic embryonal rhabdomyosarcoma, were obtained from ATCC and grown in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% heat-inactivated fetal bovine serum. LIM 1215 colon cancer cells, kindly provided by Dr. Robert Whitehead (Ludwig Institute for Cancer Research, Melbourne, Australia), were grown in RPMI 1640 medium containing 10% fetal bovine serum. Both media were supplemented with 2 mM glutamine, 100 units/ml penicillin, and 100 g/ml streptomycin. Cells were cultured at 37°C in a humidified atmosphere of 5% CO 2 .

Expression of Wild-type, Mutant, N-domain, and C-domain of IGFBP-6
The preparation of pProExHTb-IGFBP-6 that expresses a 28-amino acid peptide containing a His 6 tag and a TEV protease cleavage site at the N terminus of full-length human IGFBP-6 (without signal peptide) has been described previously (8). The vector was used as a PCR template to generate a non-IGF-binding IGFBP-6 mutant. A sense primer complementary to the N-terminal untranslated region (5Ј-AGC GGA TAA CAA TTT CAC ACA GG-3Ј) and an antisense mutant primer complementary to amino acids 88 -102 (5Ј-G GAC GAC GAG GCG GCT GCG CGC GCG GCG GCG CTC GGC CGA GGC-3Ј) were used to generate the N-terminal region of the mutant. The PCR conditions were 95°C for 1 min, then 25 cycles at 95°C for 0.5 min, 55°C for 0.5 min, and 68°C for 1.5 min. The C terminus of IGFBP-6 was obtained by digesting the pPro-ExHTb-IGFBP-6 plasmid with Eco52I/EcoRI. The pPro-ExHTb-mIGFBP-6 plasmid with P93A/L94A/L97A/L98A ( Fig.  1) was then constructed by three-way ligation. BamHI/BssHII digestion was used to identify colonies containing the mutant sequence, which were further confirmed by DNA sequencing. Wild-type, mutant, N-domain (residues 28 -106), and C-domain (residues 160 -240) of IGFBP-6 ( Fig. 1) were expressed separately in Escherichia coli and purified by Ni-NTA chromatography (8). Recombinant glycosylated IGFBP-6 (gIGFBP-6) was expressed in HEK293 cells and purified as described previously (21).

Surface Plasmon Resonance
Surface plasmon resonance (SPR) analysis was performed using the BIAcore 2000 system (BIAcore AB, Uppsala, Sweden). wtIGFBP-6 and mIGFBP-6 were immobilized by amine coupling onto separate channels of a CM5 chip (BIAcore AB) as described previously (8). Following immobilization, the chip was washed for 30 min with HBS buffer. The net increases in signal for wtIGFBP-6 and mIGFBP-6 were 682 and 1642 resonance units, respectively, where 1000 resonance units is equivalent to ϳ1 ng of protein/mm 2 . IGF-I or IGF-II (10 -400 nM) in HBS buffer was passed over the chip at 10 l/min for 10 min at 20°C. Following IGF injection, dissociation was evaluated by passing HBS buffer alone over the chip at 10 l/min for 10 min. After each run, the chip was regenerated with 0.01 M HCl.

MTT Assay
RD cell number was measured after 72 h in serum-free medium Ϯ wtIGFBP-6 or mIGFBP-6 (1 g/ml) using an MTT assay as described previously (12). A linear relationship between cell number and absorbance in this assay has been shown previously (12).

Cell Migration Assays
Transwell Assay-Cell migration was assayed using 6.5-mm Transwell permeable supports (PerkinElmer Life Sciences) with 8-m pore polycarbonate membrane inserts. Transwell membranes were coated with 100 g/ml gelatin at room temperature for 30 min. 600 l of serum-free DMEM containing 0.05% BSA and mitomycin C (0.5 g/ml) to inhibit cell proliferation, plus wtIGFBP-6 or mIGFBP-6 (1 g/ml) or IGF-II (100 ng/ml), was added to the lower chamber. RD cells (1 ϫ 10 5 cells/100 l) or LIM 1215 cells (3 ϫ 10 5 cells/100 l) in serumfree DMEM containing 0.05% BSA and mitomycin C were then seeded in the upper chamber. Cells were incubated for 24 h at 37°C, 5% CO 2 . Nonmigrating cells on the upper surface of the membrane were removed with a cotton swab. Migrating cells on the lower surface of the membrane were fixed, stained with crystal violet, and photographed. Three photographs from each of duplicate membranes were used for each experimental group. Migrating cells were counted using ImageJ software (National Institutes of Health, Bethesda). At least three independent experiments were performed.
Chemotaxis Chamber Assay-Inhibitor studies were performed in a 48-well microchemotaxis chamber (Neuroprobe, Cabin John, MD). wtIGFBP-6 or mIGFBP-6 (1 g/ml) was added to serum-free DMEM with or without inhibitors (PD098059, SB203580, and SP600125, all 10 M) and placed in the bottom wells of the chamber. Polycarbonate filters (8-m pore size) were pre-coated with gelatin (100 g/ml) and placed over the bottom wells. RD cells (5.6 ϫ 10 4 cells/well) were resuspended in serum-free DMEM and placed into top wells. The chamber was incubated at 37°C in 5% CO 2 for 24 h, after which the filters were removed, fixed, stained, mounted, and photographed. Nonmigrating cells were removed and at least three random fields/well of triplicate wells were photographed, and migrating cells were counted using ImageJ software. At least three independent experiments were performed.

Silencing of the p38 Gene by siRNA
p38 and control siRNAs were used from the SignalSilence pool p38 MAPK siRNA kit. RD cells (2 ϫ 10 5 cells/well in 12-well plates) were grown for 24 h at 37°C. p38 MAPK or control siRNA (final concentration 25 nM) was incubated in 67 l of serum-free DMEM, 0.05% BSA, 2 l of Lipofectamine 2000 for 20 min and added to cells in fresh serum-containing DMEM (333 l). Cells were incubated for 48 h at 37°C, following which Western blotting and cell migration assays were carried out as described above.

Statistical Analysis
Results are shown as mean Ϯ S.E. of 3-6 independent experiments. Migration experiments were initially analyzed using one-way or two-way analysis of variance as appropriate. For IGF-II ϩ IGFBP-6 experiments, data were log-transformed prior to analysis to stabilize variance. Post-hoc analyses were performed using the Fisher's PLSD correction for multiple comparisons.
Peptide mapping using mass spectrometry indicated that both wtIGFBP-6 and mIGFBP-6 contain identical disulfide linkages (results not shown), suggesting that the loss of IGFbinding activity of mIGFBP-6 is not because of misfolding. These results suggest that the high affinity binding site of IGFBP-6 for IGFs is similar to those of IGFBP-3 and -5 (5,6,24), and that mIGFBP-6 can be used as a tool for studying IGF-independent actions of IGFBP-6.
IGFBP-6 Decreases RD Cell Number, Most Likely Because of IGF Inhibition-The effects of wtIGFBP-6 and mIGFBP-6 on RD cell number were assessed using the MTT assay. As shown previously (12), wtIGFBP-6 (1 g/ml) significantly decreased cell number to 74 Ϯ 8% of control (p Ͻ 0.01 versus control); in contrast, mIGFBP-6 had no effect (Fig. 4). These findings, together with the high levels of autocrine IGF-II secreted by these cells, suggest that IGFBP-6 decreases RD cell number by inhibiting autocrine IGF-II actions.
Cell Migration Is Mediated by the C-domain of IGFBP-6-We next examined which domain of IGFBP-6 is involved in the stimulation of cell migration. The N-and C-domains of IGFBP-6 ( Fig. 1A) were expressed separately in E. coli and purified by Ni-NTA affinity chromatography as described previously (8). As seen in Fig. 6B, the C-domain of IGFBP-6 increased RD cell migration to 184 Ϯ 30% of control (p Ͻ 0.05). In contrast, the N-domain of IGFBP-6 had no effect on migration.
IGFBP-6 Induces RD Cell Migration via the p38 MAPK Pathway-Activation of the MAPK has been implicated in many biological processes, including cell migration (26). To determine whether MAPK pathways are involved in IGFBP-6induced RD cell migration, we first examined their activation status in response to wtIGFBP-6 and mIGFBP-6 by analyzing  . IGFBP-6 inhibits RD rhabdomyosarcoma cell proliferation in an IGF-dependent manner. RD cells (5 ϫ 10 3 cells/well) in 96-well plates were exposed to wtIGFBP-6 (1 g/ml) or mIGFBP-6 (1 g/ml). After 72 h, viable cell number was measured by the MTT assay. Cell numbers are expressed as a percentage of control. Results are shown as the mean Ϯ S.E. of eight independent experiments. **, p Ͻ 0.01, wtIGFBP-6 versus control; #, p Ͻ 0.05, wtIGFBP-6 versus mIGFBP-6. Error bars are ϮS.E. ERK1/2, p38, and JNK MAPK phosphorylation. As shown in Fig. 7, A and B, IGFBP-6 transiently increased phosphorylation of p38 MAPK. We and others have shown previously that ERK1/2 is constitutively activated in RD cells, whereas IGF treatment has no further effect (23,27,28). The present study confirmed a high level of constitutive ERK1/2 phosphorylation (Fig. 7C). wtIGFBP-6 and mIGFBP-6 had no significant effect on ERK1/2 activation in RD cells (Fig. 7, C and D). Using the same conditions, we were unable to detect a signal for phospho-or total JNK in basal or IGFBP-6-stimulated cells (results not shown).
To further evaluate whether MAPK activation contributes to IGFBP-6-induced migration in RD cells, we blocked these path-ways using specific inhibitors of p38, JNK, and ERK1/2 MAPKs. As shown in Fig. 8A, SB203580, a specific inhibitor of p38, completely abolished mIGFBP-6-induced RD migration but did not affect basal migration. Interestingly, PD98059, a specific inhibitor of ERK1/2 activation, showed a similar effect (Fig. 8B). In contrast, SP600125, a JNK inhibitor, did not affect RD migration (Fig. 8C). These results further demonstrate that p38 MAPK is involved in mediating IGFBP-6-dependent migration of RD cells. In addition, we found that the p38 inhibitor SB203580 abrogated IGF-II-mediated migration from 173 Ϯ 34% to 91 Ϯ 18% of control (p Ͻ 0.05 versus IGF-II alone, n ϭ 3). These results are consistent with a previous report showing that IGF-II-induced cell migration is p38-dependent (29).
Knockdown of p38 MAPK Abolishes IGFBP-6-induced Cell Migration-To further examine the involvement of p38 in IGFBP-6-stimulated RD migration, we transfected cells with either p38 siRNA or control siRNA. Western blot analysis showed that transfection with p38 siRNA efficiently suppressed the expression of endogenous p38 protein after 48 h, whereas transfection with control siRNA did not alter p38 expression (Fig. 9A). RD cell number after 48 h was not affected by either p38 or control siRNA transfection, excluding cytotoxic effects  ( Fig. 9B). To investigate the effect of p38 siRNA on IGFBP-6induced RD migration, cells were transfected with p38 or control siRNA for 48 h, and migration was then studied in the presence or absence of mIGFBP-6. Knockdown of p38 MAPK completely abolished mIGFBP-6-induced cell migration, confirming the effect previously seen with the p38 MAPK inhibitor (Fig. 9C). In contrast, mIGFBP-6-induced cell migration was not affected by control siRNA (Fig. 9C). These results clearly demonstrate that p38 MAPK plays an important role in IGFBP-6-dependent cell migration.

DISCUSSION
The present study describes four novel findings as follows: (a) substitution of alanine for Pro 93 /Leu 94 /Leu 97 /Leu 98 within the putative IGF binding domain of IGFBP-6 results in a more than 10,000-fold reduction in binding affinity for IGFs; (b) IGFBP-6 promotes cancer cell migration in an IGF-independent manner; (c) the C-domain underlies the promigratory effect of IGFBP-6; (d) p38 MAPK is involved in IGFBP-6-dependent migration of RD cells.
The non-IGF-binding IGFBP-6 mutant was designed based on previous structural and mutation studies of IGFBP-3 and -5 (5,6,24) showing an N-terminal hydrophobic patch that was involved in IGF binding (3). These hydrophobic residues are highly conserved in all IGFBPs, including IGFBP-6. mIGFBP-6 did not bind IGFs in three different assays, which is somewhat surprising because we recently reported that both the N-and C-domains of IGFBP-6 contribute to high affinity IGF binding, and we defined the IGF-II-binding site of the C-domain of IGFBP-6 (8, 10). We confirmed correct disulfide linkages within mIGFBP-6, indicating that misfolding is unlikely to explain the loss of IGF binding. Based on the proximity of Nand C-domain-binding sites on the IGF molecule, we previously suggested that interactions between the N-and C-domains of IGFBPs upon IGF binding may be required for optimal complex formation (9,22). A recent crystallographic study confirmed interactions between the N-and C-domains of IGFBP-4 when they are bound to IGF-I (4). Therefore, mutation of key residues in the N-domain of IGFBP-6 may also impair the ability of the C-domain to bind IGFs, thereby completely abrogating IGF binding.
A few studies suggest possible IGF-independent actions of IGFBP-6, but they are not definitive. One study proposed IGF-independent effects of IGFBP-6 in neuroblastoma cells based on its similar effects in the presence and absence of IGFs and des-(1-3)IGF-I, to which it does not bind (17). However, another study by the same group did not confirm this finding (18). In osteoblasts, antisense suppression of IGFBP-6 decreased basal levels of alkaline phosphatase, a differentiation marker (20). Exogenous IGFBP-6 apparently abolished an IGF-II-induced increase in alkaline phosphatase, but it had no effect on basal alkaline phosphatase; these data were not shown. It was inferred that the IGFBP-6 effects might be due both to extracellular IGF-II inhibition and an IGF-independent intracrine action. IGFBP-6 overexpression increased non-small cell lung cancer cell apoptosis (19). Apparently, added IGFs had no effect, and added IGFBP-6 did not reproduce it, but again these data were not shown. Further supportive but not definitive evidence comes from ϳ40 microarray papers showing regulation of IGFBP-6 (11). In about half of these, IGFBP-6 was the only IGF family member regulated, suggesting possible IGF-independent actions, although it may also indicate fine regulation of IGF actions by modulating a single IGF system component.
IGFs stimulate migration and/or invasion of many cells (25,29). IGFBPs also modulate these processes by IGF-dependent and IGF-independent mechanisms. For example, IGFBP-1 promotes cell migration by its C-terminal Arg-Gly-Asp (RGD) sequence binding to the ␣5␤1 integrin and activating focal adhesion kinase and MAPK (30,31). The heparin binding domains of IGFBP-5 and IGFBP-2 play important roles in stimulating cell migration and invasion (32,33). However, IGFBP-2 also inhibits IGF-dependent migration by an RGD-independent interaction with the ␣V␤3 integrin (34). A short C-terminal peptide of IGFBP-5 stimulates mesangial cell migration by activating the Rho kinase, Cdc42 (35). These results suggest several potential mechanisms underlying IGFBP modulation of cell migration/invasion. IGFBP-6 does not have an RGD sequence, but it contains a heparin binding domain and so can bind to glycosaminoglycans (36). Interestingly, our previous studies investigating the role of IGFBP-6 in migration using a wound assay showed that IGFBP-6 inhibited migration in an IGF-II-dependent manner (13), which conflicts with the results presented here. However, dual IGF-dependent and IGF-independent actions have previously been showed for other IGFBPs. James et al. (37) showed that myoblasts transfected with a sense IGFBP-5 cDNA delayed IGF- FIGURE 7. IGFBP-6 activates p38 but not ERK1/2 MAPK in RD cells. A and C, RD cells were incubated with wtIGFBP-6 or mIGFBP-6 (1 g/ml) for 5-60 min and cell lysates prepared. Cell lysates (50 -60 g of protein) were analyzed by Western blotting with specific antibodies for the indicated proteins (pp38, phospho-p38; p38, total p38; pERK, phospho-ERK1/2; ERK, total ERK1/2). B and D, phospho-p38 (5 min stimulation), total p38, phospho-ERK1/2 (15 min stimulation), and total ERK1/2 bands were quantitated, and the ratios of phospho-to total p38 and phospho-to total ERK1/2 were compared with control ratios (designated 1.0). Results are shown as the mean Ϯ S.E. of 3 independent experiments (*, p Ͻ 0.05 versus control (Con)). Error bars are ϮS.E.
induced cell differentiation. In contrast, Ewton et al. (38) reported that IGFBP-5 enhanced IGF-I-stimulated myoblast differentiation. IGFBP-5 has been reported to potentiate or inhibit IGF-I-induced DNA synthesis in vascular smooth muscle cells (39,40). Silha et al. (41) reported that transgenic mice overexpressing IGFBP-3 attenuated prostate tumor growth via both IGF-dependent and IGF-independent mechanisms. These findings indicate that dual mechanisms may exist in some cell systems. Therefore, non-IGF-binding IGFBP mutants are important tools to identify IGF-independent effects in these systems. An alternative explanation is that the wound assay measures chemokinesis, and the Transwell assay measures chemotaxis. wtIGFBP-6 but not mIGFBP-6 decreased RD cell number, which is consistent with our previous study showing that glycosylated IGFBP-6 decreased cell number (12). These findings, together with the high levels of autocrine IGF-II secreted by these cells, suggest that IGFBP-6 decreases RD cell number by inhibiting autocrine IGF-II actions. In contrast, wtIGFBP-6, mIGFBP-6, and IGF-II each induced RD and LIM 1215 colon cancer cell migration. Furthermore, coincubation of either wtIGFBP-6 or mIGFBP-6 with IGF-II increased RD cell migration to a similar extent. These results indicate that IGFBP-6 promotes cancer cell migration in an IGF-independent manner.
We previously showed that the C-domain of IGFBP-6 had no effect on IGF-induced proliferation of LIM 1215 colon cancer cells (8). In contrast, this domain induced RD cell migration as potently as wtIGFBP-6 and mIGFBP-6, suggesting that it is  . Knockdown of p38 MAPK inhibits IGFBP-6-induced migration of RD cells. A, Western blot of p38 protein after transfection with p38 or control siRNA. ATP citrate lyase is shown as a loading and specificity control. B, RD cells were harvested 48 h after p38 and control siRNA transfection and trypsinized, and viable cells were counted after trypan blue dye exclusion. C, mIGFBP-6-induced migration was measured 48 h after transfection of p38 and control siRNA in the presence of mitomycin C. Results are shown as the mean Ϯ S.E. of four independent experiments (*, p Ͻ 0.05 versus control (Con)). Error bars are ϮS.E. responsible for mediating this IGF-independent action. This finding is consistent with the majority of reports that map other IGF-independent IGFBP activities to their C-domains. Further studies are needed to define the precise sequence within the C-domain of IGFBP-6 that promotes cancer cell migration.
Three major MAPK pathways (ERK1/2, p38, and JNK) have been implicated in many cellular processes, including regulation of cell migration (26). The pattern of MAPK activation related to migration is dependent both on the specific stimulus and the cell type. For example, activation of p38 MAPK was necessary for halofuginone-induced migration in rat hepatic stellate cells (42), whereas ERK1/2, but not p38 or JNK, is involved in transforming growth factor-␤ 1 -stimulated migration in the same cells (43). Epidermal growth factor-induced cell migration is mediated by p38 and ERK1/2 activation in human keratinocytes (44), whereas ERK1/2 pathway is involved in epidermal growth factor-stimulated migration in human lens epithelial cells (45). In this study, IGFBP-6 induced transient phosphorylation of p38 MAPK in RD cells. IGFBP-6-stimulated migration of RD cells was blocked by SB203580, a selective p38 MAPK inhibitor, and also by specific siRNA knockdown of p38 MAPK. These results indicate that the p38 pathway is involved in IGFBP-6-dependent migration of RD cells.
p38 is activated by stress responses, including oxidative stress (46). It is noteworthy that IGFBP-6 expression is increased by cellular stressors, including H 2 O 2 -induced oxidative stress (47). p38 activation has apparently contradictory effects. In some cancer cells, it is involved in migration induced by IGF-II (29) and other factors (48,49). Indeed, IGF-II-induced migration was p38-dependent in the present study. These "tumorigenic" findings contrast with the suggestion that p38 activation is required for chemotherapy-initiated apoptosis in some cancers (50). This apparent contradiction may be resolved by cross-talk between different MAPK pathways integrating cellular responses, so that p38 MAPK actions may depend on the activation state of other MAPKs (50). The dynamics of p38 activation are also important, as prolonged p38 activation induces terminal differentiation and growth arrest of RD cells, whereas transient p38 and JNK activations by stress or cytokines do not (51). Prolonged p38 activation may also be required to promote apoptosis (46).
We found significant basal ERK1/2 phosphorylation in RD cells that was not significantly altered by IGFBP-6. It was therefore surprising that PD98059, an inhibitor of MAPK kinase that activates ERK1/2, also inhibited mIGFBP-6-induced migration when ERK1/2 phosphorylation was unaffected by IGFBP-6. One explanation is that constitutive ERK1/2 activation in these cells is necessary but not sufficient for IGFBP-6-dependent migration as has been suggested previously for H-ras-induced migration (49). Constitutive ERK1/2 activation may result in downstream cross-talk with IGFBP-6-induced p38 MAPK activation thereby resulting in the migration response (52)(53)(54)(55). Alternatively, MAPK kinase may also activate kinases other than ERK1/2 and mediate alternative signaling pathways (56,57).
JNK appears to be involved in cell migration to a lesser extent than p38 and ERK1/2. JNK inhibition did not affect IGFBP-6-de-pendent migration in RD cells, consistent with other reports that this pathway is not involved in cell migration/invasion (49,58).
In conclusion, using a non-IGF binding mIGFBP-6, we demonstrate definitively for the first time that IGFBP-6 stimulates cancer cell migration in an IGF-independent manner. p38 MAPK pathway plays a critical role in this process. Further work is required to define upstream and downstream molecules that are involved in this pathway. Given the increasingly recognized importance of the IGF system in cancer, understanding the IGF-dependent and IGF-independent actions of IGFBP-6 in detail may lead to novel therapeutic approaches.