Cell motility is central for tumor invasion, as cells must transmigrate physiologic barriers such as the extracellular matrix, the basal membrane and in some cases enveloping muscular layers.1 Many peptide growth factors enhance cell motility, including the ligands acting through the epidermal growth factor receptor (EGFR). Overexpression of EGFR has been associated with a large number of human malignancies, including breast cancer, nonsmall cell lung cancer (NSCLC), brain tumors, prostate cancer and oral cancer.2, 3, 4, 5, 6 The mechanisms of how EGFR becomes oncogenic are various, including autocrine growth factor loops, amplification of the EGFR gene and deletions/mutations that render the receptor activity independent of ligand and thus constitutively active.2, 3, 4, 5, 6, 7 The mutant receptor most frequently expressed by human tumors is the class III mutant EGFR (EGFRvIII, de2-7 EGFR, Δ2-7 EGFR). This aberrant receptor has been found expressed in many different types of cancer, including 78% of breast carcinomas, 16% of nonsmall cell lung carcinomas, 57% of gliomas, 86% of medulloblastomas and 73% of ovarian carcinomas.8, 9 EGFRvIII results from an in-frame deletion of exons 2–7, giving rise to a mature mRNA lacking 801 nucleotides.10, 11 The resulting protein lacks 267 amino acids (aa 6–273) in the extracellular domain resulting in a 145 kDa truncated receptor with a distorted ligand binding area.11 Because the deletion occurs downstream of the sequence encoding the signal peptide, EGFRvIII is properly targeted to the cell membrane. A role for EGFRvIII in the genesis and progression of human cancers is now supported by a number of studies.12, 13, 14
The molecular mechanisms behind these cellular effects of EGFRvIII expression are not well characterized, even though a constitutively active intrinsic tyrosine kinase and lack of EGFRvIII downregulation appear important.7, 15
To enhance the understanding of the role of EGFRvIII in tumor development, we have investigated the effects of EGFRvIII and 2 downstream signaling molecules on cell motility of fibroblasts and compared the effect of EGFRvIII on cell motility to that of ligand-activated EGFR. We find that EGFRvIII induces protrusions, focal adhesion disassembly, loss of stress fibers and increases cell motility in a manner similar to ligand-activated EGFR. In addition, these effects are dependent on the receptor tyrosine kinase activity and MEK activity. Furthermore, EGFR-mediated motility was dependent on the amount of EGFR expressed by the cell.
MATERIAL AND METHODS
Wortmannin, AG1478 and recombinant human EGF were obtained from Calbiochem (San Diego, CA) and U0126 was from Promega (Madison, WI). Polyclonal antibodies to ERK1/ERK2, phosphorylated ERK1/ERK2, polyclonal antibodies to PLCγ1 and phosphorylated PLCγ1 (Tyr783) were from Cell Signaling Technology (Beverly, MA). Monoclonal antibodies to vinculin (Clone hVIN-1) were obtained from Sigma-Aldrich (Copenhagen, Denmark). Mouse monoclonal anti-EGFR/-EGFRvIII (Ab-5, clone H-11) was from NeoMarkers (Fremont, CA). Mouse monoclonal anti-phosphotyrosine antibody (Py20) was from Zymed (San Francisco, CA). HRP-conjugated mouse anti-rabbit and TRITC-conjugated rabbit anti-mouse antibodies were purchased from DAKO (Glostrup, Denmark). Alexa Fluor® 568 phalloidin was from Molecular Probes (Eugene, OR) and Matrigel was obtained from BD Biosciences (San Jose, CA).
The cell lines NR6, NR6M and NR6W have been described previously16 and were obtained from Dr. Darell Bigner, Duke University. Briefly, NR6M and NR6W were generated by transfection of NR6, a variant of the Swiss 3T3 murine fibroblast cell line that lacks expression of endogenous EGFR, with cDNA encoding full-length human EGFR (NR6W) or the type III mutant EGFR (NR6M). The NR6-wtEGFR cell line (which expresses a lower number of receptors as compared to NR6W) has also been described previously and is a generous gift from Dr. Allan Wells, Department of Pathology, University of Pittsburgh.17, 18 U87MG and U87MG-ΔEGFR were provided by Dr. Webster Cavenee, Ludwig Institute for Cancer Research, University of California, San Diego. Cells were maintained as monolayer cultures in Dulbecco's minimal essential medium (DMEM) (Invitrogen, Taastrup, Denmark) supplemented with 10% fetal calf serum (FCS) (Invitrogen) and 200 units/ml penicillin and 200 μg/ml streptomycin in a humidified chamber with 5% CO2 at 37°C for up to 15 passages.
Random (individual) cell motility.
NR6, NR6M, NR6W and NR6wtEGFR cells were seeded in 6-well Matrigel-coated plates (10 μg/ml) to approximately 20,000 cells per well on the day of motility measurements. Twenty-four hours after seeding, the cells were switched to low-serum media (0.5% FCS) and starved for 24 hr. The inhibitors 10 μM AG1478, 20 μM U0126 and 0.1 μM Wortmannin were added 5 hr prior to and EGF 1 hr prior to measurement of cell motility. As stock solutions of the compounds were prepared in DMSO, control cells were tested in the presence of DMSO. In all cases, the final concentration of DMSO was 0.2% (v/v). Time-lapse video recordings of live cells were performed on a microscope workstation consisting of a 120 × 100 mm joystick-controlled motorized stage mounted on of an Eclipse TE300 inverted microscope (Nikon, Kanagawa, Japan). Video recordings were performed using a Video 1100 analogue b/w CCD-camera (DFA, Copenhagen, Denmark) attached to the microscope. All recordings were performed at 37°C. Images from 15–25 microscopic fields per well were recorded at intervals of 15 min for a period of 4 hr using the PRIGRA software (Protein Laboratory, Copenhagen, Denmark).
Marking of the centers of cell nuclei was performed manually on computer images. Approximately 50 cells per experiment per cell line were marked and tracked. To exclude the possible influence of cell-cell interactions on cell behavior, cells touching each other were excluded.
The track of an individual cell was defined as a sequence of positions of the center of its nucleus over time. The dispersion of a cell is the Euclidean distance between two points on a plane measured in μm. The displacement of a single cell after a given time of observation, dtobs, was calculated as
where x(·) and y(·) are the x- and y-coordinates of the cell, respectively, and tobs is the time during which a given displacement dtobs takes place with starting time t0.
Root mean square speed (S) and persistence time, P, were estimated by plotting the mean-squared displacement, 〈d2〉 against time with subsequent curve fitting to the equation
where ti is the time interval of interest.19 The rate of diffusion, R, was calculated using the equation R = S2P.
Cells were seeded in triplicate in 6-well Matrigel-coated plates and allowed to grow until 80% confluent. Cells were then switched to serum starving conditions (0.5% FCS) overnight and treated with 10 μM AG1478 or DMSO for 4 hr or treated with 10 nM EGF 1 hr before wounding, after which a wound through the central axis of the plate was gently made using a pipette tip. Migration of cells into the wound was observed at 4 preselected time points, 0, 2, 4, 8 and 10 hr, and photographed at 100× magnification.
NR6W and NR6wtEGFR cells were seeded in 6-well plates either uncoated or coated with fibronectin (10 μg/ml) or Matrigel (10 μg/ml) and serum starved overnight. Varying concentrations of EGF (1–100 nM) was added and the cells incubated at 37°C for 24 hr. Photographs were taken after 1 hr, 2 hr, 5 hr, 10 hr and 24 hr.
Western blot analysis
For determination of phosphorylated proteins, serum-starved cells were lysed directly in 1× NuPAGE sample buffer (Invitrogen), immediately separated on precast 3–8% NuPAGE SDS-PAGE gels (Invitrogen) and electroblotted onto nitrocellulose membranes (Invitrogen). Inhibitors 10 μM AG1478, 20 μM U0126, 0.1 μM Wortmannin and the control DMSO were added for 3 hr and EGF for 0 min.
Membranes were blocked in 5% nonfat dry milk in 20 mM Tris pH 7.5, 150 mM NaCl, 0.1% Tween-20 for 1 hr followed by incubations in rabbit polyclonal antibodies to PLC-γ, phosphorylated PLC-γ, ERK, phosphorylated ERK, EGFR and monoclonal antibodies to phosphotyrosine. The membranes were washed in 10 mM Tris pH 7.5, 100 mM NaCl, 0.1% Tween-20 for 3 × 10 min and bound antibodies detected by incubation with appropriate HRP-conjugated secondary antibodies (DAKO) and visualized using ECL (Amersham Pharmacia Biotech, Buckinghamshire, UK).
Immunostaining and staining for F-actin
For vinculin and F-actin staining, cells were plated at a density of 104 cells per chamber on a Labtek 4 well chamber slides (NUNC) coated with Matrigel. Cells were grown in 10% FCS until 80% confluent and then switched to starvation medium with 10 μM AG1478 or DMSO and incubated overnight. An amount of 10 nM EGF was added to NR6W cells 5 hr prior to fixation. Staining for vinculin was performed by fixation in 3% (w/v) formaldehyde, permeabilization with 0.2% (v/v) Triton-X-100 and incubation with a monoclonal antibody to vinculin. After incubation with TRITC-coupled secondary antibodies, the stainings were evaluated using confocal laser-scanning microscopy (Radiance 2000, Bio-Rad, Hercules, CA). For F-actin staining, cells were washed in actin-stabilizing buffer [10 mM Tris-Cl pH 7.4, 0.15 M NaCl, 2 mM MgCl2, 10% (v/v) glycerol], fixed in 3% (w/v) formaldehyde and permeabilized with 0.2% (w/v) saponin and incubated with Alexa Fluor® 568 phalloidin and visualized as above.
Expression of EGFRvIII increases fibroblast cell motility
NR6 cells stably expressing EGFRvIII16 were examined for motility and compared to control and EGFR-expressing cells (Fig. 1). Individual cell motility was evaluated by plating cells as single cells on a substratum closely resembling the in vivo situation, in casu Matrigel®. Individual tracks of control cells, EGFRvIII and EGFR-expressing cells either with or without addition of EGF were determined using time-lapse video recording and computer-assisted image analysis (Fig. 1). Subsequently, the cell-displacement was expressed as the mean-square-displacement, d2, for τ = 0–240 min, using the method of overlapping intervals.20 Figure 1a shows representative displacement curves for the 4 cell lines: NR6, NR6M, NR6W and NR6wtEGFR in the presence or absence of EGF. Figure 1b shows “wind-rose” plots prepared by superimposing the starting points of tracks of individual cells. From these 2 figures it can be seen that expression of EGFRvIII resulted in a marked increase in cellular displacement compared to both control cells and unstimulated EGFR-expressing cells.
To quantify the motile behavior values for the root mean square speed, S (Fig. 1d), the cellular persistence time, P (Fig. 1e), and the rate of diffusion, R = 2S2P (Fig. 1c) were estimated by fitting the individual curves to the equation 〈d2〉 = 2S2P (τ − e -τ/p)) (for definitions, see Material and Methods). Expression of EGFRvIII caused a marked increase in the rate of diffusion from a mean R-value of 1.50 ± 0.13 μm2/min for the control cells to a mean R-value of 24.51 ± 3.41 μm2/min for cells expressing EGFRvIII. This increase in the rate of diffusion was due to a 350% increase in the root mean square cell speed, S, from 0.13 ± 0.01 μm/min of control cells to 0.46 ± 0.09 μm/min for EGFRvIII-expressing cells. The persistence times, P, were not significantly different (Fig. 1e). The motility of control and EGFRvIII-expressing cells were also measured in the presence of 10 nM EGF and values similar to those found without EGF were obtained (Fig. 1c–e) showing that EGFRvIII increases individual cell motility independent of EGF.
Overexpression of the EGFR induces cell de-adhesion rather than motility upon ligand stimulation
To compare the effect of EGFRvIII on cell motility to that of ligand-activated EGFR, we measured the motility of a cell line expressing high amounts of EGFR (NR6W) with or without the addition of EGF (Fig. 1). Surprisingly, we found a decrease in the rate of diffusion from 5.00 ± 1.80 to 1.81 ± 0.13 μm2/min (Fig. 1c) and a corresponding decrease in root mean square cell speed from 0.25 ± 0.09 to 0.14 ± 0.01 μm/min (Fig. 1d) of cells stimulated with EGF as compared to untreated cells. The persistence time was not significantly altered by EGF stimulation (Fig. 1e). The effect of EGF on the cells was visually apparent, as the cells lost their adhesive properties, acquiring a more rounded morphology.
The motility (parameters R and S) of NR6W cells in the absence of EGF were, however, significantly greater than the control cell line, indicating some level of spontaneous EGFR activation in these cells (Fig. 1). We thus hypothesized that a lower EGF concentration might induce motility. However, addition of EGF in the concentration range from 0.1 nM to 60 nM all had a similar effect: de-adhesion and rounding of cells (data not shown).
To evaluate the effect of EGFR expression level on cell motility, we included the cell line NR6wtEGFR, which expresses a lower number of receptors (1.4 × 105 receptors/cell) as compared to NR6W (1.5 × 106 receptors/cell), NR6wtEGFR and NR6M express comparable receptor amounts (Fig. 2a,b).16, 17 Basic motility in the absence of EGF was comparable for the 2 EGFR-expressing cell lines. However, in contrast to NR6W, the motility of NR6wtEGFR cells was increased upon addition of 10 nM EGF to a level similar to EGFRvIII: increased rate of diffusion (from 8.16 ± 1.92 to 55.35 ± 8.02 μm2/min) and increased root mean square speed (from 0.27 ± 0.02 to 0.83 ± 0.14 μm/min) but unaltered persistence time (53.13 ± 18.89 and 49.45 ± 11.08) (Fig. 1). We observed a dose-dependent increase in cell motility of NR6wtEGFR cells by EGF; for example, addition of 25 nM EGF increased the rate of diffusion to remarkable a 120.40 ± 17.44 μm2/min and a root mean square speed to 1.10 ± 0.16 μm/min (data not shown). Thus, the different outcome of EGF stimulation of NR6W and NR6wtEGFR cells is probably due to the difference in receptor number. It is also interesting to note that even though NR6M and NR6wtEGFR express similar amounts of either mutated or normal EGFR, respectively, ligand-activated EGFR seems to be a more potent activator of cell motility than EGFRvIII. This is probably due to a higher level of receptor activation as determined by total tyrosine phosphorylation of the receptors (Fig. 2c).
To investigate if these results obtained with single cells could be confirmed with a population of cells, wound-healing assays were performed (Fig. 3). Again NR6M and EGF-stimulated NR6wtEGFR cells were able to migrate into the wound at a rate greater than the parental cell NR6 and nonligand-stimulated NR6W and NR6wtEGFR cells.
The de-adhesion and rounding of NR6W cells is dependent on the substratum
To test if the de-adhesion and rounding of the NR6W cells after addition of EGF were dependent on the substratum, we investigated the visible phenotypic changes induced by different concentrations of EGF on different substrata. It was found that NR6W cells grown on fibronectin did not respond to 10 nM EGF by de-adhesion and rounding of cells, whereas they did on noncoated plates and plates coated with Matrigel (Fig. 4). No effects of substratum or varying concentrations of EGF were visibly apparent on NR6wtEGFR cells (data not shown). Concentrations as low as 1 nM EGF resulted in de-adhesion and rounding of NR6W cells on Matrigel and plastic, whereas concentrations of 100 nM did not have an effect on these cells growing on fibronectin (data not shown).
EGFRvIII induced motility is dependent on receptor tyrosine kinase activity and MEK activity but only weakly on PI3-kinase activity
To determine if the observed increase in motility of EGFRvIII-expressing cells was indeed mediated by EGFRvIII or a consequence of the clonal selection of cells, the effect of the EGFR/EGFRvIII tyrosine kinase inhibitor, AG1478, on cell motility was tested (Fig. 5). From the displacement curves (Fig. 5a) and the “windrose plots” (Fig. 5b), it can be seen that addition of AG1478 markedly decreased the motility of NR6M cells to a level equivalent to that of the control cell line. Both the rate of diffusion (from 24.51 ± 3.41 to 1.24 ± 0.87 μm2/min) and the mean root square speed (from 0.46 ± 0.09 to 0.10 ± 0.02 μm/min) decreased 80–95% by AG1478, whereas the persistence time was unaffected (Fig. 5c–e). A similar effect was observed on the NR6W cell line upon addition of AG1478, resulting in a decrease in motility to a level similar to NR6, again indicating that the EGF receptors on this cell line exist in a partially activated state (Fig. 5b–d). AG1478, in addition, inhibited the EGF-mediated de-adhesion and rounding of NR6W cells (data not shown). AG1478 had no significant effect on the motility of control cells. Results obtained using AG1478 were verified in a wound-healing assay and, as anticipated, AG1478 inhibited the migration of NR6M-, NR6W- and EGF-stimulated NR6wtEGFR cells to a level similar to that of the control cell line, NR6 (Fig. 3b).
The effects of the MEK inhibitor U0126 and the PI3K inhibitor Wortmannin on the EGFRvIII- and EGFR-mediated changes in cell motility were investigated (Fig. 5). As seen on the “windrose plots” U0126 decreased the motility of all cell lines tested (NR6, NR6M and NR6W), although the effect was most pronounced on NR6M cells (Fig. 5c). This indicates that MEK activity is important for both adhesion receptor-dependent motility (basal motility) and growth factor-mediated motility. Interestingly, the decrease in basal motility (rate of diffusion), as measured by the motility of NR6, was due to a decreased persistence time rather than a decrease in mean root square speed as observed for growth factor-mediated motility (Fig. 5c–e). As for AG1478, U0126 abrogated the de-adhesion and rounding of the NR6W cells after addition of EGF, demonstrating that MEK might be involved in mediating this signal. In contrast to U0126, Wortmannin only had a partial effect on motility but was unable to decrease motility mediated by EGFRvIII to more than 50%, even at doses as high as 5 μM (Fig. 5 and data not shown). Wortmannin did not have a significant effect on the motility of NR6 and NR6W cells.
EGFRvIII and ligand-activated EGFR induce cell polarization, focal adhesion disassembly and a reduction in the number of stress fibers via a receptor tyrosine kinase-dependent mechanism
Several parameters are important for cell motility, including generation of protrusions (lamellipodia) and strength of cell adhesion to substratum with maximal motility observed at intermediate attachment strength. We investigated the changes induced by EGFRvIII for these 2 parameters. Cell protrusions were visualized by actin staining of migrating cells. EGFRvIII-expressing cells had multiple prominent extending protrusions, whereas the control cell line and unstimulated EGFR-expressing cell lines did not (Fig. 6a). However, stimulation of NR6wtEGFR with EGF induced a phenotype similar to NR6M cells with the appearance of multiple extending protrusions in the majority of cells. EGF had no effect on morphology of NR6 and NR6M cells (data not shown). To demonstrate that the increase in protrusion was due to constitutively active EGFRvIII and ligand-activated EGFR, AG1478 were added prior to staining. AG1478 completely abolished the appearance of cellular protrusion on NR6M and EGF-stimulated NR6wtEGFR cells. In addition, treatment of these cells with AG1478 led to a marked increase in the number of stress fibers reaching a level similar to that of the control cell line. Again we saw that upon addition of EGF to NR6W cells, cells became rounded and lost their adherence to substratum. Immunostaining of the EGFRvIII-expressing cell line for the focal adhesion component vinculin revealed that these cells indeed displayed a marked decrease in focal adhesions compared to the control cell line (Fig. 6b). In addition, the few focal adhesions observed were mainly found in cellular protrusions consistent with findings in a breast carcinoma cell line.21 These contacts are presumably necessary for forward movement and stabilization of the extending lamellipodia. A similar loss of focal adhesions was seen upon EGF stimulation of NR6wtEGFR cells, indicating that EGFRvIII is able to signal focal adhesion disassembly in a manner similar to ligand-activated EGFR. Addition of AG1478 resulted in stabilization of focal adhesions, although the majority of these were located at the cell periphery. Interestingly, we observed that in the process of de-adhesion and rounding of NR6W cells upon EGF stimulation, cells leave behind a network of vinculin-containing adhesion plaques (Fig. 6b). Again this process was inhibited by AG1478.
Both EGFRvIII and ligand-activated EGFR induce PLC-γ and ERK phosphorylation
Previous studies have shown that both ERK and PLC-γ activity are necessary for EGFR-mediated cell motility.22 To confirm that ERK and PLC-γ are phosphorylated upon ligand stimulation in NR6W and NR6wtEGFR cells and to determine if EGFRvIII phosphorylated these molecules to a similar extent, western blot analyses were performed with antibodies specific for phosphorylated ERK and PLC-γ. EGF treatment of NR6W and NR6wtEGFR cells was found to increase the phosphorylation of ERK but to have no effect on NR6 and NR6M cells (Fig. 7). The amount of ERK phosphorylation after stimulation was, however, much higher in the NR6wtEGFR cell line. NR6M cells were found to have a high basal level of ERK phosphorylation comparable to that of EGF-stimulated NR6W cells. The phosphorylation of ERK was sensitive to AG1478 and U0126 in both the EGFR-expressing cell lines and the mutant EGFR-expressing cell line. In contrast, the PI3K inhibitor, Wortmannin, only had a partial effect on ERK phosphorylation and only on NR6M and NR6W cells.
The phosphorylation of PLC-γ in NR6M cells, on the other hand, was lower compared to EGF-stimulated NR6W and NR6wtEGFR cells, although higher than in NR6 cells. Surprisingly, NR6W cells had a very high level of PLC-γ phosphorylation in the absence of EGF stimulation. AG1478 inhibited the phosphorylation of PLC-γ completely in NR6M, NR6W and NR6wtEGFR cells, whereas Wortmannin had no effect. U0126 was found to have an effect on PLC-γ phosphorylation in the NR6wtEGFR cell line but did not show inhibition, although it increased the level of phosphorylation (Fig. 7b).
EGFRvIII also increases the motility of the glioblastoma cell line U87MG
To verify that EGFRvIII not only increases the motility of fibroblasts, the motile behavior of the glioblastoma cell line U87MG-ΔEGFR stably expressing EGFRvIII and in parental cell line were investigated in a wound-healing assay. As can be seen in Figure 8a, expression of EGFRvIII also increases the motility of this cell line.
As observed for NR6M, U87MG-ΔEGFR motility was inhibited by AG1478 to a level similar to the parental cell line (Fig. 8b). The small increase in motility after addition of EGF to both U87MG and U87MG-ΔEGFR is most likely due to the low level of endogenous expression of EGFR.
The EGFR status is often altered in human tumors originating from many different tissues and EGFR has been associated with increased proliferation, invasion, angiogenesis and survival of these tumors.4, 23, 24 However, it is becoming increasingly clear that it is not only the level of EGFR that contributes to the transforming and tumor-promoting effect of EGFR. Autocrine ligand production and deletions/mutations in the EGFR gene that render the receptor constitutively active seem equally important.9, 25, 26, 27, 28 The role of the constitutively active type III mutant EGF receptor EGFRvIII in cancer progression and development has been studied extensively over the past several years, and the transforming and tumor-promoting properties of EGFRvIII is supported by numerous studies.8, 13, 14, 16, 29, 30, 31 More recently, EGFRvIII has specifically been implicated in tumor cell invasion. Expression of EGFRvIII enhanced in vitro invasion of both a glioblastoma cell line and a small cell lung cancer cell line.14, 32
In our study, the effect of EGFRvIII expression on individual fibroblast motility was evaluated by time-lapse video recordings and compared to the effect of expression and activation of EGFR. We, for the first time to our knowledge, demonstrate that EGFRvIII expression has major effects on the actin cytoskeleton and focal adhesions leading to increased membrane protrusive activity and augmented motility as compared to a control cell line. The effect was indistinguishable from that observed for a cell line expressing similar amounts of EGFR and stimulated with 10 nM EGF. In contrast, a cell line expressing 10 times the amount of receptors reacted by detachment from the extracellular matrix and decreased motility upon stimulation with EGF. All results could be confirmed in a wound-healing assay. Furthermore, we found that motility enhancement by EGFRvIII was sensitive to pharmacologic inhibitors of EGFR-, MEK- and PI3-Kinase-mediated signaling pathways; pathways that have previously been reported to be involved in EGFRvIII-mediated cellular transformation.33, 34
To gain mechanistic insight into EGFRvIII-mediated cell motility, the rate of diffusion was calculated based on the persistent random walk model. Rate of diffusion (R) can be visualized as the area of a circle around the cell in which the cell can reach any point within a minute. Rate of diffusion is a function of the 2 basic parameters: root mean square cell speed (S) and persistence time (in direction, P) (R = 2S2 × P). Here it was found that EGFRvIII increased root mean square speed but not persistence time of cells, resulting in a squared increase in rate of diffusion. Rate of diffusion has been shown to correlate well with the invasive potential of cells, indicating that expression of EGFRvIII may have major implications for the malignant behavior of cancer cells in vivo,35 resulting in tumor progression and development of an invasive phenotype.
Immunostaining revealed that EGFRvIII expression caused a redistribution of F-actin, resulting in a loss of stress fibers. A loss of focal adhesions was also observed, indicating that the effect of EGFRvIII on cell motility may be related to decreased cell substratum interactions and/or alterations in the organization or dynamics of the actin cytoskeleton.
It was found that the EGFR kinase inhibitor AG1478 reduced the motility of NR6M cells to a level comparable to that of the control cell line NR6. Thus, kinase activity of EGFRvIII is necessary for motility enhancement by this EGF receptor mutant, just as the kinase activity has been shown to be essential for movement elicited by ligand-activated EGFR.17 AG1478 had no effect on persistence time but decreased the root mean square speed. A pathway involving the dual specificity kinase, MEK, has been reported to be central for EGFR-mediated motility by inducing focal adhesion disassembly.36, 37 It has been shown that this pathway activates the proteinase calpain II, which is required for fibroblast motility.37 We speculate that a similar pathway could be active downstream of EGFRvIII, explaining the motility inhibition of the NR6M cell line seen with U0126. Interestingly, inhibition of MEK not only resulted in decreased mean root square cell speed but also persistence time. This suggests that MEK regulates persistence time independently of EGFRvIII.
A role for PI3K in EGFR-mediated motility is uncertain, although PI3K plays an important role in induction of cell motility by the platelet-derived growth factor receptor (PDGFR).38 However, the motility of NR6M cells as evaluated by the rate of diffusion was partially inhibited by the PI3K inhibitor, Wortmannin, indicating that a Wortmannin-sensitive event plays a role in EGFRvIII-induced motility. Interestingly, the main effect of Wortmannin was, in contrast to AG1478 and U0126, solely on the persistence time and not mean root square speed. This suggests that the effect of Wortmannin on cell motility is mediated by a signaling mechanism separate from EGFRvIII and that this signaling might be permissive rather than active.
We also evaluated and compared the motility of NR6M cells to that of the 2 cell lines, NR6W and NR6wtEGFR, expressing different levels of EGFR. Interestingly, we found that these 2 cell lines responded very differently to EGF stimulation. In our system, NR6wtEGFR responded in a way similar to that previously reported, by increased motility.17 The increase in motility of NR6wtEGFR cells by stimulation with 10 nM EGF was higher than the basal level of NR6M cell motility, although the cell lines express similar levels of receptors, which suggests that it is the level of phosphorylation that determines the increase in motility. This is further supported by the fact that the motility of NR6wtEGFR could be induced further by addition of 25 nM EGF. In contrast to previous studies using the NR6wtEGFR cell line, we were not able to see a significant decrease in persistence time upon stimulation with EGF.39 Generally expression and activation of EGFRvIII and EGFR did not have an effect on persistence time in our model system. Rather, the increased motility mediated by EGFRvIII and ligand-activated EGFR was due to enhanced cell speed. It is reasonable to assume that the discrepancy in the effect of EGF on persistence time is due to differences in the extracellular matrix constituents as the study by Ware et al.39 used Amgel, whereas Matrigel was used in our study.
In contrast to NR6M and NR6wtEGFR, the NR6W cell line, which expresses a very high number of EGFR, reacted upon EGF addition by detachment from the extracellular matrix. This phenomenon has previously been demonstrated for the A431 human epidermoid carcinoma cell line, which also has overexpression of EGFR.40 Lu et al. found using the A431 cell line that the EGF-induced detachment was mediated by focal adhesion kinase dephosphorylation, leading to a rapid destruction of focal adhesions. It is plausible that a similar mechanism causes the detachment of NR6W cells and that it is a general phenomenon of cells expressing very high levels of EGFR. Fibronectin is known to activate integrins resulting in the phosphorylation of FAK, counteracting the dephosphorylation caused by EGFR activation.40 To test this hypothesis in our system, NR6W and NR6wtEGFR cells were grown on plastic, Matrigel or fibronectin. When NR6W cells grown on fibronectin were stimulated with EGF, they did not de-adhere or become rounded, as they did on Matrigel and plastic. Thus the signal mediated by fibronectin must be dominant over the EGF-mediated signal. No effect of EGF stimulation on the adhesive properties of NR6wtEGFR was visually apparent even at EGF concentrations as high as 100 nM.
There does not seem to be any obvious advantage for cells to move into a state of weak or absent adherence except possibly during cytokinesis.41 What might be the significance of this de-adhesion and does it relate to invasive and metastatic properties of cells?
Lu et al. found that the motility and invasive capacity of A431 increased upon EGF stimulation, whereas our study clearly demonstrated that NR6W cells had decreased motility when stimulated with EGF, as they simply did not adhere to substratum despite similar numbers of EGF.40 The difference is most likely due to the different methods of measuring cell motility used in these 2 studies as Lu et al. used a chamber mobility assay (Transwells), whereas our study used time-lapse video recording to track individually moving cells and a wound-healing assay. Time-lapse video recording, although much more sensitive and quantitative, is not suitable for measuring motility of cells floating in the media. The chamber mobility assay may be more useful for cells constantly adhering and de-adhering. However, how does motility in vitro relate to tumour invasion and metastasis in vivo? Is detachment from the extracellular matrix more related to these processes than a fast migration on a substratum? Lu et al. suggest that downregulation of FAK might be essential and required for early metastatic spreading, enabling vascular circulation of tumour cells without adhesion. Once tumour cells reattach to the extracellular matrix, integrin stimulation of FAK promotes adhesion and the growth of metastatic tumours.40
Thus, detachment from the extracellular matrix may be a step further in the malignant transformation of cells going from motile, but anchorage dependent, to anchorage independent cells capable of metastasizing. However, additional studies are needed to address this important question.
ERK and PLC-γ are among the main downstream effectors of EGFR-mediated signaling and have been reported to regulate cell motility.37 Western blot analyses using phosphospecific antibodies showed that ERK was phosphorylated to a great extent upon EGF stimulation in NR6wtEGFR cells and moderately in EGF-stimulated NR6W cells. In addition, it was confirmed that expression of EGFRvIII leads to a moderate level of constitutive ERK phosphorylation as previously suggested.33 Treatment with AG1478 and U0126 completely blocked all ERK phosphorylations, demonstrating that the ERK phosphorylation observed in these cells is mediated by receptor signaling to ERK via MEK. It was quite unexpected that the low EGFR-expressing cell line NR6wtEGFR had the most pronounced ERK phosphorylation upon EGF stimulation of the 2 EGFR-expressing cell lines. Considerable evidence point to the fact that different kinetics of ERK activation can result in completely different cellular responses and that different signaling pathways are used to elicit these different kinetics of activation.42, 43 Thus the difference in the level of ERK phosphorylation might be a critical determinative factor of the different effects of EGF stimulation on the phenotype of NR6wtEGFR and NR6W cells, respectively.
Interestingly, the level of PLC-γ phosphorylation was reciprocal in the 2 EGFR-expressing cell lines as compared to that of ERK. NR6W cells had a high constitutive level of PLC-γ activation mediated by the receptor, as it was completely abolished by AG1478. No increase in PLC-γ phosphorylation could be found in this cell line upon addition of EGF. A similar pattern was seen in the NR6M cell line, which for the first time demonstrates that EGFRvIII mediates PLC-γ phosphorylation, although the level of phosphorylation was weaker compared to ligand-activated EGFR. PLC-γ phosphorylation was induced in the NR6wtEGFR cell line upon EGF stimulation but was undetectable in the absence of ligand. Thus, a low to moderate phosphorylation of PLC-γ correlates with increased motility, as observed for the NR6M and NR6wtEGFR cell lines, whereas a high constitutive level of phosphorylation correlates with decreased motility and rounding of cells. Although the involvement of ERK in EGFR-mediated motility has been well documented, the actual role of ERK activity in motility is not well described.22 It has been shown that ERK activates the protease M-calpain leading to de-adhesion and disruption of focal adhesions.37 This, however, does not correlate well with our results, as the cell line with the lowest ERK activity is actually the least adhesive one. ERK has also been involved in activation of myosin light chain kinase (MLCK), resulting in phosphorylation of myosin light chain (MLC).44 This pathway increased cell migration and the authors suggest that this could be due to increased actin-myosin motor function mediated by phosphorylation of MLC.44 This is in agreement with our results, showing that the EGFR kinase inhibitor AG1478 and the MEK inhibitor U0126 had a pronounced inhibitory effect on cellular speed. PLC-γ activity, on the other hand, has been involved in mobilization of actin-modifying proteins such as gelsolin, profilin and cofilin, leading to cytoskeletal reorganization.45 In addition, it has been suggested that PLC-γ is involved in the activation of PKC, which attenuates EGFR mitogenic signaling.46 Thus, it is possible that the high level of PLC-γ activity observed in the NR6W cell line leads to extensive cytoskeletal reorganization, resulting in de-adhesion and rounding of cells. In addition, the high level of PLC-γ activity may participate in downregulating ERK activity.
Based on the experiments described herein and the work by others, it is possible to depict a model describing the different kinetics of signal pathway activation induced by ligand-activated EGFR, overexpressed EGFR and EGFRvIII (Fig. 9). This model suggests that overexpressed EGFR, when activated by ligand, induces a pronounced activation of PLC-γ and FAK dephosphorylation, resulting in cellular detachment from the extracellular matrix.40 More moderate levels of EGFR, when induced by ligand, in contrast leads to a manifest increase in ERK activation and a more moderate activation of PLC-γ and dephosphorylation of FAK. EGFRvIII induces moderate activation of ERK and a weak activation of PLC-γ. Interestingly, EGFRvIII activates ERK not only through Ras/Raf but also through PI-3K. Overall, it seems as if a high ERK activity and a moderate PLC-γ activity are ideal for EGFR and EGFRvIII-mediated motility (Fig. 9).
To verify that EGFRvIII-induced motility is a general phenomenon, the wound-healing assay was also performed on the human glioblastoma cell lines U87MG and U87MG-ΔEGFR, which clearly demonstrates that expression of the EGFRvIII increases cell motility.
In conclusion these data demonstrate that expression of EGFRvIII strongly enhances fibroblast migration. This effect seems to be mediated mainly by an increase in cell speed, whereas the persistence time remains unaltered. Furthermore, the increase in motility by EGFRvIII was sensitive to inhibition of the EGFRvIII tyrosine kinase, MEK and partially PI3K. The effect on cell motility was similar to that observed for EGFR, although the level of EGFR expression and activation appeared to be important for the cellular outcome. A low to moderate expression and activation of EGFR resulted in increased motility after ligand stimulation, whereas overexpression and activation caused cell detachment from the extracellular matrix and hence decreased motility. Overall based on these results, tumor-associated EGFR mutations might be critical for tumor cell motility, invasion and thus establishment of metastatic growth. Furthermore, the difference in the cellular response to different levels of EGFR expression and activation may have important effects on the behavior of tumors cells and implications for treatments of tumors expressing EGFR.
We thank Dr. D. Bigner, Dr. W. Cavenee and Dr. A. Wells for their generous gifts of cell lines.