A Role for Phospholipase C-γ-mediated Signaling in Tumor Cell Invasion¹

Cancers of the breast and prostate are among the leading causes of death in adults, collectively responsible for over 80,000 deaths annually in the United States (1). Although localized tumors are curable by excision and have good prognoses, invasive and metastatic tumors are often nonresponsive to treatments and present significant morbidity and mortality. Many different therapeutic treatments have been used in the past years with only limited success because the exact mechanism of disease progression that ultimately causes death remains unclear. Therefore, elucidation of the mechanism of breast and prostate tumor invasion and metastasis would provide for future rational therapeutic approaches.

EGFR³ family members are significantly amplified and/or overexpressed in a wide variety of tumors. Specifically, both EGFR and its relative, ErbB2 (also referred to as neu and HER2), have been shown to be overexpressed in breast and prostate cancers. These cancers often express autocrine stimulatory loops involving EGFR and its ligands (2, 3, 4, 5). Increased expression of EGFR and/or ErbB2 correlates with greater tumor invasion and metastasis and poorer prognosis in many cancers, including those of the breast and prostate (6, 7, 8, 9). This observation invites the hypothesis that EGFR and ErbB2 play significant roles in tumor progression to the invasive and metastatic state.

There are a number of cellular mechanisms associated with tumor invasion and metastasis, including recognition of the extracellular matrix, protease activity that creates a defect through the matrix, and tumor cell migration. Among these, tumor cell migration is promoted by EGFR signaling and subsequent activation of PLCγ (10). Our previous studies have defined an EGFR-PLCγ signaling pathway that is required for enhanced cell motility but not mitogenesis (10, 11). In one prostate cancer cell line, DU-145 (12), we demonstrated that pharmacological or molecular inhibition of PLCγ signaling reduced cell invasiveness both in vitro and in vivo (13, 14, 15). The diminished invasiveness was attributed to decreased motility, based on biophysical studies in fibroblasts and the fact that abrogation of PLCγ signaling did not reduce cell proliferation or increase tumor cell apoptosis. Interestingly, PLCγ, which is activated by many growth factor receptors, including ErbB2 (16, 17, 18, 19, 20), has been shown to be required for motility signaling from a number of growth factor receptors, including those for platelet-derived growth factor and insulin-like growth factor-I (21, 22). We propose that this key molecule serves as a point of convergence that is used by diverse growth factors and, thus, may be a common target for tumor progression promoted by amplification of receptors in addition to EGFR, such as ErbB2. Thus, it is important to determine whether the concept of PLCγ-mediated migration being required for invasion is universal, which in itself would be an important contribution to the study of invasion and metastasis in malignancies.

To determine whether the role of EGFR as observed in DU-145 is generalizable, we investigated the PLCγ motility pathway in other cells derived from prostate and breast carcinomas. The prostate line chosen is from a mouse prostate tumor that occurs de novo in the TRAMP model (23, 24). These mice express the SV40 large T-antigen driven by a probasin promoter. Furthermore, because these mice develop invasive prostate carcinomas, using these cells would enable the extension of any finding to in vivo settings in the future. To determine whether PLCγ-mediated signaling was important in the progression of other tumors arising in sex hormone-responsive tissue, breast cancer cell lines were examined. Because EGFR overexpression has been correlated with tumor progression and poor metastasis in large series of breast carcinomas (8, 25, 26), the EGFR-overexpressing lines MDA-468 and MDA-231 were chosen. However, amplification of c-erbB2/neu also is frequent in metastatic breast carcinomas (9, 27). Because this closely related EGFR family member also presents PLCγ docking sites and activates PLCγ (28, 29, 30, 31, 32), we postulated that erbB2 signaling may subsume a similar role as EGFR signaling and requires the proposed convergent PLCγ signaling pathway for increased tumor invasion. Thus, the role of PLCγ signaling in the invasiveness of an erbB2-overexpressing line, MDA-361, was investigated. In short, we found that pharmacological and/or molecular inhibition of PLCγ reduced invasion of these cell types through Matrigel and that inhibition of EGFR kinase disrupted an autocrine stimulatory loop driving this invasion. These findings suggest that growth factor-induced motility involving PLCγ signaling is generalizable to a variety of tumor types and may serve as a target to limit tumor progression.

TRAMP C1 and TRAMP C2 cells were derived from prostate tumors that developed in the TRAMP model (23, 24). The cells were maintained in high-glucose DMEM containing 10% FCS and supplemented with insulin (5 ng/ml), penicillin (25 units/ml), streptomycin (25 μg/ml), and dihydrotestosterone (10⁻⁸ m). MDA-MB-468 (MDA-468) were a kind gift from Dr. Jeffrey Kudlow (University of Alabama at Birmingham, Birmingham, AL), whereas MDA-231 and MDA-361 cells were obtained from American Type Culture Collection. The breast cancer MDA cell lines were maintained in DMEM containing 10% FCS supplemented with l-glutamine (2 mm), penicillin/streptomycin (100 u/ml), nonessential amino acids (0.1 mm), and sodium pyruvate (1 mm). Media for the MDA-361 cells also contained 10 ng/ml insulin, as recommended previously (33).

RNA was isolated from both TRAMP C1 and TRAMP C2 by the Trizol method (Life Technologies, Inc.; catalog #15596-018). Suitable primers were designed in accordance to DNA maps of EGFR, EGF, and TGFα in which ∼50 bp of each map was selected based on minimal looping, suitable G-C:A-T ratio, and the presence of one restriction site within the fragment for confirmation. RT-PCR was performed on RNA isolated from both TRAMP C1 and TRAMP C2 (primers for EGFR: 5′-AGACCATCCAGGAGGTGGC and 3′-GATGGCTCTGTAAGTCCATTG; for EGF: 5′-TTGCCCTGACTCTACCGCAC and 3′-CCACCATGATGTCATGCTTCTG; and for TGFα: 5′-GGTGCAGGAAGAGAAGCCAG and 3′-GCACGGCACCACTCACAGTG) for 30 cycles. PCR products were run on 1% agarose gels both intact and cut with the appropriate restriction enzymes to confirm the identity of the bands. For immunoblots, cell lysates were obtained from confluent cells, size-fractionated, and analyzed as below. These were then probed with antibodies that recognize the murine EGFR (Transduction Laboratories; catalog #12020) and murine pro-TGFα (a kind gift from Dr. Jeffrey Kudlow; Ref. 34). Membranes exposed to the pro-TGFα antibody were blocked in 5% fat-free milk containing 100 μm DTT, and the antibody itself was diluted in TBS (10 μg/ml) with 0.05% Tween 20 and containing 5 μm DTT.

Introduction of exogenously encoded PLCz was accomplished using the lipofectin method. PLCz is a dominant-negative fragment of PLCγ and consists of the SH2 and SH3 domains and an inhibitory (I) domain (35). The constitutively expressed PLCz (pXf PLCz), as well as its control (pXfcontrol), was transcribed from an SV40 early promoter. Briefly, the DNA (5–10 μg) was precipitated in ethanol and 0.25 M ammonium acetate and then mixed with 30 μl of lipofectin (Life Technologies, Inc.) before introduction into 6-well plates containing MDA-468 cells. Media was replaced after 5 h, and transfected cells were selected in media containing 1200 nm methotrexate. Expression was verified by immunoblotting with an anti-PLCγ antibody.

Protein expression and phosphorylation were determined, as described previously (10). Cells were tested under conditions that either enable or minimize paracrine/autocrine signaling. To observe the autocrine signaling, cells were grown to confluence and then maintained at maximal confluence for at least 48 h in media containing10% FCS in a 10-cm dish, after which cells were either lysed or incubated for an additional 24 h under serum-free conditions before lysing. To minimize autocrine signaling, subconfluent cells were quiesced in 1% dialyzed FCS (dFCS) or serum-free media for 24 h, as determined for each cell line by basal thymidine incorporation and cell viability, followed by stimulation with EGF for 5 min. Cells were washed with PBS and lysed in lysis buffer [10% glycerol, 1% Triton X-100, 100 nm NaCl, 20 mm HEPES (pH 7.4), and 1 mm sodium vanadate] for 1 h at 4°C. After clarification by microcentrifugation, the lysate was incubated for 1.5 h with a specific antibody-agarose bead mixture that had been incubated overnight at 4°C. Primary antibodies included antihuman EGFR (Oncogene Science; catalog #GR01), antihuman ErbB2 (Calbiochem; catalog #OP39), and anti-PLCγ antibody (Upstate Biotechnologies, catalogue #05-163). The lysate-antibody-agarose bead mixture was then washed five times with lysis buffer before analysis by reducing SDS-PAGE and immunoblotting.

Cell invasiveness in vitro was determined by the ability to transmigrate a layer of extracellular matrix, Matrigel, in a modified Boyden chamber assay (36, 37). Matrigel invasion chamber plates were obtained from Becton Dickinson/Biocoat (catalog #40480 and #40481). For each individual cell line, cells were plated randomly and distributed among plates with different lot numbers, with each experiment performed in triplicate. Despite possible variances in EGFR ligand concentrations in Matrigel, these concentrations are saturating because even “reduced growth factor Matrigel” contain relatively high amounts of EGF (up 0.5 ng/ml compared with 0.5–1.3 ng/ml in regular matrigel; Becton Dickinson Labware 1997/98 catalog description, page 128). Cells were kept in serum-free media containing 1% BSA for the first 24 h and then replaced with only serum-free media for the remaining 48 h. Enumeration of the cells that invaded through the matrix over a 72-h period was accomplished by two different methods. Initially, cells were metabolically labeled in the presence of 5 μCi/ml [methyl-[³H]] thymidine and the acid-precipitable counts on the bottom of the filter and in the targeting well measured; this procedure eliminates cell proliferation as a confounding variable (15). The radioactivity associated with the well and on the bottom of the filter were consistent so that in later experiments only the filter-associated label was measured. In later experiments, we visually counted cells on the bottom of the filter, as per routine procedures (36), after we verified that cell number corresponded with transmigrated radioactivity. In all cases, individual experiments were performed in triplicate.

Cells were plated into 12-well plates, allowed to grow to confluence, and then placed in serum-free media for 48 h. After this, appropriate wells were treated with 10 nm EGF and/or appropriate concentrations of U73122 for an additional 18 h. [methyl-[³H]] thymidine (5 μCi/ml) was added to each well, and incubation resumed for another 8 h. Wells were then washed with PBS, followed by acid precipitation (5% TCA, 4°C for 30 min). The wells were washed with PBS and then treated with 0.2 N NaOH to solubilize the incorporated radiolabel. Scintillation counting quantitated the amount of incorporated thymidine. Each experiment was performed in duplicate.

PLC activity was determined by measuring accumulation of IPs (38, 39). Cells were labeled by culturing in 5 μCi/ml myo [1,2-[³H](N)] inositol for 24 h, after which the plates were washed twice with 37°C PBS. To limit degradation of IPs, 10 mm lithium chloride was added to the cell media and incubated for 10 min before proceeding. The plates were then treated with 10 nm EGF for 30 min. Cells were lysed with 1 ml of boiling water, and lysates were separated with a Dowex (AG1-X8 100–200 mesh) anion-exchange mini column containing 1 ml of 1:1 ionized sephadex resin:water to bind charged phosphate groups. The contents were eluted with buffers in the following sequence: water to release free inositol, sodium borate/sodium formate (5 mm/60 mm) to release glycero-phosphoinositol), and ammonium formate/formic acid (200 mm/100 mm) to release IP. The amount of IP was measured by obtaining radioactive counts of the final eluate. We have previously determined that this assay correlates closely with measuring IP3 production directly by more time-consuming high-performance liquid chromatography analysis (11). Furthermore, the extended time period of 30 min allows for significant accumulation of IP (in the presence of LiCl) and, thus, measures PLC-mediated turnover as a sensitive readout instead of steady-state IP3 (38, 39).

All analyses were performed as paired Student’s t tests, with a level of significance assigned at <0.05.

The TRAMP cell lines were derived recently, and their EGFR expression status was unknown. Using RT-PCR, we detected expression of EGFR and its primary ligands EGF and TGFα, indicating the possible presence of an autocrine loop in these cells (Fig. 1). Protein lysates of TRAMP C1 and TRAMP C2 cells were immunoblotted with antibodies generated against murine EGFR or murine pro-TGFα (because the cell lysate and not supernatant is probed). Both of these were recognized in TRAMP C1 and TRAMP C2 cells. Thus, the TRAMP prostate cell lines present a potential autocrine stimulatory loop present in prostate epithelial and carcinoma cells. (We could not detect EGF due to the inability to obtain a suitable antibody against this ligand.)

To determine whether the cells under investigation in this study possess functional EGFR, cells were plated under conditions that minimize paracrine/released autocrine signaling. Cells were then treated with 10 nm EGF, and the breast cell lines cell lysates were immunoprecipitated with anti-EGFR antibody and immunoblotted with antiphosphotyrosine, whereas cell lysates from the TRAMP cell lines were directly size-fractionated by SDS-PAGE. EGFR was markedly phosphorylated due to EGF stimulation in the cell lines with up-regulated EGFR (Fig. 2). These findings are consistent with previously published studies (40, 41) and serve to demonstrate that the lines we are investigating behave as reported.

In the breast cell line MDA-361, which presents increased levels of the EGFR relative ErbB2, ErbB2 was found to be constitutively phosphorylated ErbB2 independent of the presence of heregulin or EGF (Fig. 2). MDA-361 cells also contain low levels of EGFR that are phosphorylated on EGF stimulation (data not shown). The slight decrease in phosphorylation of ErbB2 in MDA-361 cells on ligand stimulation, which is a previously observed event (42, 43), is postulated to be due to ErbB2 redistribution in the plasma membrane, as well as increased degradation on transphosphorylation or complexing with other ErbB-family members.

A central postulate of this study is that PLCγ-mediated signaling is operational in diverse tumor cell lines. Because the TRAMP model was only recently derived and characterized, we examined the responsiveness of TRAMP C1 and TRAMP C2 cells to EGF (10 nm) stimulation. Both TRAMP cell lines responded with a marked phosphorylation of PLCγ in response to EGF stimulation under conditions that minimize autocrine signaling (Fig. 3 A). To demonstrate that this phosphorylation coincided with increased PLC activity, the PLC inhibitor U73122 significantly reduced EGF-induced IP production in TRAMP C1 cells by 44% (data not shown).

To confirm that PLCγ is activated on ligand stimulation in all cell types, we also immunoprecipitated lysates of MDA-468 and MDA-231 cells treated or untreated with 10 nm EGF under autocrine signaling-limiting conditions with antiphosphotyrosine and blotted with anti-PLCγ antibody. These determinations, some of which confirm previously reported results (40, 41), serve as a necessary positive control for experiments to follow. Acute EGF treatment causes an increase in phosphorylation of PLCγ in these cells (Fig. 3 A). PLC activity assays also were performed on MDA-468 cells. EGF treatment increased IP production significantly (1.8-fold), and pretreatment with U73122 decreased IP production by up to 76% (data not shown).

Increased PLCγ phosphorylation also was observed after EGF stimulation in the ErbB2-overexpressing MDA-361 cells (Fig. 3 A). MDA-361 cells exhibited constitutively active PLCγ, which is expected due to its overexpression of constitutively active ErbB2. PLCγ phosphorylation increased with EGF treatment, presumably due to low EGFR levels in these cells that cross-phosphorylate ErbB2. Interestingly, previous studies indicate modulation of downstream phosphorylation by cytoplasmic (nonextracellular domain-containing) EGFR (and possibly ErbB2; Refs. 44 and 45). The significance of these results is simply that there is an inducible PLCγ pool in these cells.

The above manipulations demonstrated that the motility-associated PLCγ signaling can be induced in these cells lines. However, a central point of our model posits activation of this motility pathway be dysregulated autocrine signaling in the case of up-regulated EGFR or constitutive signaling in the case of overexpressed ErbB2. As such, we tested whether PLCγ signaling was operative under autocrine signal-permitting conditions in the four EGFR-expressing cell lines. To address this question, cells plated under autocrine-promoting conditions were compared with those under autocrine-limiting conditions. Two of the cell lines (TRAMP C1 and MDA-468) were challenged for 6 h by EGF (Fig. 3, B and C). This latter condition was chosen to both represent the time period at which EGF-induced motility is maximal after an initial lag phase ⁴ and to allow for cellular adaptation with regulated down-regulation that would be operational during autocrine signaling. In this analysis, cells under autocrine-promoting conditions presented enhanced PLCγ phosphorylation comparable with the same cells under autocrine-restricting conditions or after extended EGFR signaling. Further support for constitutive activation of PLCγ signaling was noted by decreases in IP production in the presence of U73122 in TRAMP C1 (down to 57–66% in cells treated with or without EGF in the presence of U73122, when compared with cells treated with or without EGF in the absence of U73122).

Our previous studies using DU-145 prostate cancer cells implicated the EGFR-stimulated PLCγ motility pathway in tumor cell invasion (13, 14). We hypothesized that this role of the PLCγ is not unique to DU-145 cells. We predicted that this is especially true in breast and prostate that exhibit a high incidence of EGFR overexpression when transformed and, consequently, higher invasion rates and poorer prognoses. To this end, we assessed the in vitro invasiveness of the TRAMP C1, TRAMP C2, MDA-468, and MDA-231 cell lines. Cells were serum-starved and labeled with tritiated thymidine for 24 h, after which they were plated onto EGFR ligand-rich Matrigel-coated invasion chambers and allowed to invade for 72 h in the presence or absence of a pan-PLC inhibitor, U73122 (46, 47). The upper chamber contained media with 1% BSA for the first 24 h and was replaced with serum-free media for the remaining 48 h, whereas the bottom (collecting) chamber contained media with 10% FCS throughout. In both prostate and breast cell types, those treated with U73122 invaded to a significantly lesser extent than nontreated cells (Fig. 4). Invasiveness of TRAMP C1 and TRAMP C2 cells treated with U73122 were reduced by 47% and 48%, respectively, compared with their nontreated counterparts, and invasiveness of MDA-468 and MDA-231 cells were reduced by 29% and 36%, respectively. These reductions in PLCγ motility pathway-modulated invasiveness compare favorably with those seen in DU-145 cells treated similarly; importantly, this was reflected in vivo by near complete abrogation of invasion (13, 14, 15). There were no observed differences in cell morphology or adherence to Matrigel at the concentrations of U73122 used when compared with nontreated cells. The reduced invasion of these cells as a result of PLCγ inhibition with U73122 emphasizes the important role of growth factor-induced PLCγ motility pathway in cell invasion, as shown using EGFR-expressing cells. Furthermore, the MDA-361 cells that overexpress a constitutively active ErbB2 likewise demonstrated reduced invasion when treated with U73122 (52%).

The observed inhibition of cell invasiveness by U73122 might be due to nonspecific drug toxicity. To test for such toxicity by examining a PLCγ-independent response, mitogenesis (11, 48), cells were grown on Matrigel-coated 24-well plates and treated with U73122 concentrations used during the invasion assays for each cell type. The cells were under identical conditions as in invasion assays (except that tritiated thymidine was added 8 h before harvesting) for 72 h, after which incorporated thymidine counts were obtained. The cells treated with U73122 did not show any reduced thymidine incorporation when compared with nontreated cells (Fig. 4). Mitogenesis assays were also performed on cells in regular non-Matrigel-coated plates and showed similar results of not affecting cell proliferation (data not shown). These results demonstrate that U73122 does not affect invasion by hindering cell mitogenesis or proliferative functions.

To further explore the role of the PLCγ motility pathway in tumor cell invasion, MDA-468 cells were engineered to express the dominant-negative PLCγ fragment, PLCz. Cells were transfected with a constitutively transcribed PLCz (pXf PLCz) plasmid or a similar vector containing an irrelevant peptide (pXfcontrol). We used pooled clones to avoid cell-cell microheterogeneity. These cells were subjected to PLC activity assays, as described above, and cells expressing PLCz exhibited a reduced IP yield compared with untransfected cells (Fig. 5), whereas cells expressing pXfcontrol did not. MDA-468 cells expressing PLCz were assayed for Matrigel invasion. Cells expressing PLCz showed reduced invasion (by ∼50%) on the order of that noted in the presence of U73122.

The Matrigel invasion experiments were performed in the presence of EGFR ligands derived from cell autocrine signaling and present in Matrigel. Autocrine activation of PLCγ signaling was demonstrated by continuous phosphorylation of PLCγ despite the absence of exogenous EGF (Figs. 3,B and 4,A). This suggested that the PLCγ-dependent aspect of invasiveness is secondary to EGFR signaling based on the presence of a potential autocrine stimulatory loop. However, to determine whether EGFR signaling was actually involved in invasion, we inhibited EGFR signaling and thereby the presumed downstream activation of PLCγ using the EGFR kinase-specific inhibitor PD153035 (49). TRAMP C1 and MDA-468 cells were treated with PD153035, and invasion assays were performed as described above. Invasion wells containing 100,000 TRAMP C1 cells or 200,000 MDA-468 cells each were treated with PD153035 (500 nm, determined empirically as the concentration needed to inhibit EGFR-induced phosphorylation), and the invaded cells were quantitated (Fig. 6). Invasiveness of TRAMP cells treated with PD153035 was reduced by 32% compared with their nontreated counterparts, whereas invasion of MDA-468 cells was reduced to 45%. These reductions in invasiveness are similar to those seen with the pharmacological inhibitor of PLC. One concern is that inhibition of EGFR signaling reduces cell proliferation and this may be seen as fewer cells transmigrating the Matrigel barrier; however, as the TRAMP C1 cells were metabolically labeled and enumerated, this assay is independent of cell proliferation. Furthermore, there were no observed differences in cell morphology, adherence to Matrigel, or cell survival at the concentrations of PD153035 used when compared with nontreated cells.

Up-regulated signaling from EGFR or erbB2 is strongly correlated with tumor invasion and metastasis, although the responsible intracellular events remain undefined. Taking into account the sequence of events that are involved in tumorigenesis and progression, our studies have focused on cell motility under the hypothesis that this event modulates tumor invasion and metastasis (50, 51). In this study, we demonstrate that disrupting the motility-associated PLCγ-mediated signaling pathway inhibits in vitro invasion in multiple cell types.

This signaling contributing to invasion was first described by our laboratory in the human prostate cancer cell line DU-145 (13, 14, 15). These prior studies were vital for establishing proof of the concept that EGFR-enhanced migration can be a rate-limiting step in the invasion of a tumor. However, questions remained as to whether this critical role of the motility pathway in invasion is unique to DU-145 cells only, whether it is prostate cancer-specific, or whether it is a more universal concept that holds in a variety of different tumor types. The studies presented here investigated the role of the EGFR-induced PLCγ-mediated motility pathway in cell invasion in a different, de novo-occurring mouse prostate carcinoma (the TRAMP model) and also extended this concept to other tumors of the steroid-responsive tissue, namely breast (MDA-468 and MDA-231). Lastly, because PLCγ is a point of convergent signaling from multiple growth factor receptors (21, 22), we determined whether PLCγ-mediated signaling was also required for cell invasiveness in a breast cancer line, MDA-361, overexpressing the closely related ErbB2 receptor. In these experiments, ErbB2 phosphorylation levels decreased slightly on stimulation with EGF. This is a previously observed occurrence that is due to EGF modulating increased internalization and down-regulation rates of EGFR-ErbB2 heteromer aggregates (42, 43).

Our hypothesis postulates that EGFR-induced cell migration is a major regulatory step in tumor progression. Disruption of this pathway, through specific inhibition of PLCγ, should, therefore, reduce cell invasiveness. We treated cells with U73122, a pharmacological agent that specifically inhibits PLC and, as such, inhibits cell motility but not mitogenesis both in fibroblasts (11, 48) and prostate epithelial cells (13). Ancillary studies showed that cells treated with U73122 demonstrated reduced IP production on EGF stimulation in a drug dose-dependent manner (data not shown), indicating that U73122 inhibited PLCγ in cells regardless of tissue of origin. In our earlier studies in fibroblasts (11), we found that PLCγ signaling levels controlled the extent of migration and that even partial inhibition of enzymatic activity was linearly related to extent of cell motility. Thus, if motility was rate-limiting for invasion, the relative decreases in IP production in the presence of U73122 should be reflected by similar decreases in Matrigel transmigration. It should be noted that although significant, inhibition is not complete. Cells usually exhibit some degree of baseline nonligand-induced motility, which likely contributes to some invasiveness in this in vitro assay. Interestingly, when examining invasiveness of DU-145 variants, we found that the cells expressing the nonmotility inducing c’973 EGFR presented 60% of the invasiveness of parental cells in vitro, but were almost completely noninvasive in vivo (Xie, 1995; Turner, 1996). Our more recent preliminary data suggest that PLCγ signaling is also important in invasiveness of cells from bladder carcinoma and glioblastomas, two other tumors in which up-regulated EGFR correlates with progression (52, 53, 54).

These studies strongly suggest that PLCγ-mediated signaling is a generalizable property for tumor cell invasiveness induced by growth factor receptors and not unique to DU-145 prostate carcinoma cells. The possibility that this decrease in invasion may be due to a toxic effect of U73122 was rendered unlikely due to the apparent nonsensitivity of cells to U73122, in terms of mitogenesis. U73122 is also unlikely to be affecting other signaling pathways because it is, by all accounts, targeted toward PLC and not even PLD or PLA₂ (47). However, to enhance specificity, we performed a similar set of analyses in which PLCγ signaling was abrogated by expression of an exogenously encoded dominant-negative fragment, PLCz (14, 35, 55). At the concentration and in the manner used, neither U73122 nor PLCz completely inhibit PLC activity (11), and, as such, we have not considered the difference in inhibition among the two methods as biologically significant. However, invasiveness was reduced to levels similar to or lower than those obtained in the earlier DU-145 studies using either method of inhibition of PLC; these partial levels of in vitro inhibition corresponded to almost complete inhibition of invasion in vivo (13, 14, 15). The results of both the PLC activity and Matrigel transmigration assays mirrored those obtained using U73122, which strongly supports a specific role for the PLCγ-mediated motility pathway in tumor invasion.

It may be noted that the various tumor cell lines transmigrated the Matrigel barrier in the absence of exogenously added EGFR ligands. The addition of EGF was deemed unnecessary for two reasons. First, prostate and breast carcinoma cell lines often present autocrine stimulatory loops involving the EGFR and its ligands (2, 3, 4, 5). The TRAMP cells present both EGFR and EGF and TGFα (Fig. 1) and are dependent, at least in part, on EGFR signaling for mitogenesis. The EGFR-expressing breast cell lines MDA-468 and MDA-231 have been reported also to possess an autocrine stimulatory loop and, thus, present up-regulated EGFR signaling (56, 57). Our experimental data also point to autocrine activation of EGFR signaling because we have determined that the EGFR-specific inhibitor PD153035 limits TRAMP cell line mitogenesis even in the absence of added exogenous ligand or matrix (data not shown). We do not imply that invasion and motility are signaled solely via PLCγ; other intermediary signaling pathways are involved. However, our findings simply show that invasiveness seen in these cells occurs secondary to PLCγ- and EGFR-mediated signaling.

It was difficult to demonstrate autocrine signaling at the level of EGFR autophosphorylation because ligand-induced down-regulation under autocrine-promoting conditions resulted in constitutively low levels of EGFR, similar to 6 h treatment with EGF (data not shown). This result was expected from previous detailed analyses of EGFR trafficking in response to autocrine stimulation in which ligand is in excess (58, 59). Thus, we inferred the autocrine loop by the expression of both ligand and receptor and the effects of PD153035 on downstream signaling. This autocrine signaling is likely sufficient in itself because we found earlier that DU-145 cells transmigrated a human matrix barrier in an EGFR-dependent manner, although the Amgel does not contain detectable levels of EGFR ligands (15, 60). Second, Matrigel contains high levels of EGFR ligands (61). Although the presence of copious growth factors, including platelet-derived growth factor receptor and TGFβ receptor ligands in Matrigel confounds in vitro analyses, it may be physiologically representative of the organismal situation in which stromal cells produce TGFα and other EGFR ligands (notably amphiregulin and heparin-binding EGF) that are often present in extracellular matrices (61, 62). Thus, it is likely that in the presence of Matrigel both the matrix-associated ligands and the autocrine signaling sufficiently activate the overexpressed EGFR. On the other hand, it is possible that we disrupted a signaling pathway that is not actually up-regulated in these tumor cells, but rather the basal signaling of which is required for active cell motility. As such, PLCγ signaling would be considered as permissive for invasiveness rather than as a consequence of up-regulated EGFR or ErbB2 signaling. We favor the former possibility because disruption of EGFR signaling with the EGFR kinase-specific inhibitor PD153035 reduced invasiveness of TRAMP C1 and MDA-468 cells similarly to PLCγ abrogation. In either case, whether PLCγ signaling is actively regulated or simply permissive, this signaling pathway would still be a target for rationale therapeutic intervention to limit tumor invasion.

It is important to state that enhanced cell motility is not the only rate-limiting step or cell behavior with regard to tumor progression and invasiveness. Tumor progression can be limited by blocking other required cell events, such as cell proliferation and production of proteolytic enzymes (63, 64). In many cases, it is not clear whether these cell events are promoted at the transition to invasiveness or are present in a permissive manner at this stage from earlier cell alterations (65). It also is likely that other signaling pathways, such as those involving PI-3 kinase and mitogen-activated protein kinase (22, 66, 67), are important for signaling motility, and abrogation of those would similarly limit invasiveness. We did not pursue these because PI-3 kinase phosphorylation was not induced by EGF in our cells (data not shown) and mitogen-activated protein kinase was considered too broad a target, which would also interrupt cell proliferation (67, 68). Although there is evidence that the up-regulated EGFR and ErbB2 receptors signal increase proliferation (69) and production of select proteases (70), future studies are required to determine whether these particular events are linked to receptor up-regulation and its correlation to tumor invasion.

In summary, these data strongly suggest that the role of the EGFR- PLCγ cell motility pathway in tumor invasion encompasses a number of different cell types and suggest that the requirement of cell migration for invasion may be a universal concept. Ultimately, whether this is actually the case will require studying the invasiveness of these cells, along with inhibiting the PLCγ motility pathway in an in vivo environment. Such experiments would require suitable cell lines, such as MDA-231, that are relatively invasive in vivo compared with MDA-468 (71). Also of significance is that these results suggest that ErbB2 can subsume the role of EGFR in as far as activating PLCγ signaling. The results presented here also provide for a possible therapeutic approach in which specific targeting of the motility-inducing mechanisms of the cell may provide a treatment against tumor progression.

We thank Angela Glading, Mark Van Epps-Fung, Heng Xie, and Kiran Gupta for discussion and technical assistance.