EGF-receptor-mediated mammary epithelial cell migration is driven by ...

3688

Research Article

EGF-receptor-mediated mammary epithelial cell migration is driven by sustained ERK signaling from autocrine stimulation Elizabeth J. Joslin1, Lee K. Opresko2, Alan Wells3, H. Steven Wiley2 and Douglas A. Lauffenburger1,4,* 1

Department of Biological Engineering, MIT, Cambridge, MA 02139, USA Systems Biology Program, Pacific Northwest National Laboratory, PO Box 999, Richland, WA 99352, USA 3 Department of Pathology, University of Pittsburgh Medical Center, Pittsburgh, PA 15261, USA 4 Center for Cancer Research, MIT, Cambridge, MA 02139, USA 2

*Author for correspondence (e-mail: [email protected])

Journal of Cell Science

Accepted 21 August 2007 Journal of Cell Science 120, 3688-3699 Published by The Company of Biologists 2007 doi:10.1242/jcs.010488

Summary EGF family ligands are synthesized as membraneanchored precursors whose proteolytic release yields mature diffusible factors that can activate cell surface receptors in autocrine or paracrine mode. Expression of these ligands is altered in pathological states and in physiological processes, such as development and tissue regeneration. Despite the widely documented biological importance of autocrine EGF signaling, quantitative relationships between protease-mediated ligand release and consequent cell behavior have not been rigorously investigated. We thus explored the relationship between autocrine EGF release rates and cell behavioral responses along with activation of ERK, a key downstream signal, by expressing chimeric ligand precursors and modulating their proteolytic shedding using a metalloprotease inhibitor in human mammary epithelial cells. We found that ERK activation increased monotonically with increasing ligand release rate despite concomitant downregulation of EGF

Introduction The EGF receptor (EGFR, also known as erbB1 and HER1) tyrosine kinase system is vitally involved in normal physiological processes such as tissue morphogenesis and wound repair, because these events require the key cell responses of survival, proliferation and migration (Yarden and Sliwkowski, 2001). The family of ligands that bind specifically to the EGFR includes EGF, TGF-alpha (TGF-␣), amphiregulin (AR, AREG), heparin-binding EGF (HB-EGF, HBEGF), betacellulin, epiregulin and epigen (EPGN) (Harris et al., 2003). These ligands are synthesized as membrane-anchored precursors containing an N-terminal extension, an EGF-like receptor-binding domain, a membrane proximal region, a transmembrane region and a cytoplasmic tail. Ligands such as HB-EGF and AR might be capable of efficiently activating the EGFR in a juxtacrine mode at points of cell-cell contact, whereas others, including EGF and TGF-␣, require proteolytic cleavage from the cell surface to form mature soluble growth factors that act in a local autocrine fashion on the releasing cell or in a more distal paracrine manner on the same cell type or other cell types (Borrell-Pages et al., 2003; Dong et al., 2005; Singh and Harris, 2005). Autocrine or paracrine stimulation of

receptor levels. Cell migration speed was directly related to ligand release rate and proportional to steady-state phospho-ERK levels. Moreover, migration speed was significantly greater for autocrine stimulation compared with exogenous stimulation, even at comparable phosphoERK levels. By contrast, cell proliferation rates were approximately equivalent at all ligand release rates and were similar regardless of whether the ligand was presented endogenously or exogenously. Thus, in our mammary epithelial cell system, migration and proliferation are differentially sensitive to the mode of EGF ligand presentation. Supplementary material available online at http://jcs.biologists.org/cgi/content/full/120/20/3688/DC1 Key words: Cell motility, Cell proliferation, Metalloproteases, Transactivation, Growth factors

EGFR signaling governing cell functions is strongly implicated in a wide variety of adult tissue regulatory processes, including ovarian follicle remodeling (Conti et al., 2006), kidney repair after injury (El Hader et al., 2005), asthmatic bronchorestriction (Chu et al., 2005), neuronal plasticity (Gu et al., 2007), vascular morphogenesis (Semino et al., 2006), corneal wound healing (Xu et al., 2007), liver regeneration (Ross et al., 2001) and mammary gland development (Rosen, 2003). Various members of the EGF ligand family are involved in one or more of these processes, including EGF, TGF␣, AR and HB-EGF, depending on the tissue type and the developmental state. However, there is little understanding at the present time concerning the quantitative dependence of the cell functions underlying these processes on the extent of autocrine/paracrine EGF ligand-induced EGFR activation. The ADAM family metalloproteases appear to be central actors in the release of EGF family ligands. ADAM17 (also known as TACE) has been implicated in the proteolytic processing of TGF-␣, AR and HB-EGF, whereas ADAM10 is thought to be involved in EGF cleavage (Borrell-Pages et al., 2003; Sahin et al., 2004). Protease regulation of ligand release is under active investigation, especially because of the growing


Autocrine EGFR-mediated cell migration appreciation of this mechanism as underlying transactivation of EGFR family receptors by a variety of stimuli, such as G protein-coupled receptor ligands and mechanical stress (Eguchi et al., 2003; Fischer et al., 2003; Ohtsu et al., 2006). Evidence exists that signaling pathway activation can stimulate autocrine ligand release (Baselga et al., 1996; Fan and Derynck, 1999; Pandiella and Massague, 1991) and that this effect can be enhanced by constitutively active pathway components (Seton-Rogers et al., 2004; Wilsbacher et al., 2006). Although there is little sequence homology between the various ligands outside of the core EGF-like domain, the membrane-proximal region is thought to be a key determinant in ectodomain shedding and protease specificity (Arribas et al., 1997; Harris et al., 2003). Autocrine ligand signaling was initially identified by the ability of transformed cells to grow in culture in the absence of exogenous growth factor supplements (Sporn and Todaro, 1980), although this mode of signaling was subsequently found to be common in normal epithelial cells (Bates et al., 1990; Markowitz et al., 1990; Tsao et al., 1996). Because of the intrinsic ‘closed-loop’ nature of autocrine systems, however, identification and experimental analysis of autocrine effects on cellular behaviors are unusually challenging (Wiley et al., 2003). Cells expressing a form of pro-EGF that required shedding have shown increased directional persistence of migration compared with exogenously stimulated cells, suggesting that autocrine presentation might conceivably induce distinctive cell behavior due to spatial restriction of EGFR activation (Maheshwari et al., 2001). Both computational and experimental approaches have estimated that EGFR autocrine loops can operate on a scale less than a cell diameter and that several parameters – such as ligand release rate, receptor number, ligand affinity and ligand diffusivity – affect spatial localization (DeWitt et al., 2002; DeWitt et al., 2001; Forsten and Lauffenburger, 1992; Lauffenburger et al., 1998; Maly et al., 2004; Shvartsman et al., 2001). In addition, experiments have demonstrated that the ratio of ligand release to receptor production correlates with the fraction of ligand that is captured, suggesting that these parameters might determine whether cells operate in an autocrine or paracrine mode (DeWitt et al., 2001). Despite the strong belief that the EGFR system in normal and transformed epithelial cells is autocrine (Condeelis et al., 2005; Wells, 2000), it is rarely investigated in context of this ligandpresentation mode. Instead, investigators typically add a bolus of exogenous EGF that causes a wave of receptor occupancy. The time-dependent cell responses are then used to infer receptor regulatory mechanisms. Although such an approach can reveal important information about the dynamics of EGFR signaling pathways, it is difficult to know the physiological significance of the results. For example, the phenomenon of receptor downregulation, which is evidenced by a loss of receptor levels following bolus treatment with high ligand concentrations, has long been proposed to be an attenuation mechanism to prevent cells from over-responding in the face of ligand excess (Wells et al., 1990). Would such a process be physiologically important in vivo? Because EGFR activation in situ is controlled by the proteolysis of membrane-anchored ligand precursors, and proteolysis, in turn, is regulated by the activity of a variety of upstream stimuli such as cytokine receptors (Chen et al., 2004; Janes et al., 2006), the increase in

3689

receptor activity governed by modulation of ligand proteolysis should be far more gradual than that induced by a ligand bolus. Manipulating ligand release rates as a way to understand the regulation of the EGFR pathway has not found widespread use in experimental studies to date because of its severe technical challenge. We have previously described a cell system that allows the manipulation of EGFR ligand production by cells and have used it to explore some biophysical properties of autocrine signaling (DeWitt et al., 2001; Dong et al., 2005). This system employed a combination of tetracycline-regulated promoters and receptor-blocking antibodies to control the relative extent of ligand production and capture. Using this approach, we found that cells could recapture shed ligands with high efficiency, and that receptor occupancy was directly proportional to the rate of ligand shedding and influenced by ligand-binding affinity (DeWitt et al., 2001; DeWitt et al., 2002). Although these results were important to establish important biophysical parameters of autocrine signaling, they did not address the downstream consequence of modulating ligand shedding. This was because the cells used in those studies, highly aggressive transformed fibrosarcoma cells, were chosen for their suitability for manipulating gene expression rather than their response to EGF treatment. Highly responsive cells, such as human mammary epithelial cells (HMEC), typically require the use of retroviral vectors for gene expression, which prevents the use of regulated promoters. However, to circumvent these problems, we have devised a new methodology that relies on a combination of engineered ligands and protease inhibitors to control steady-state ligand production across a range from very low to very high occupancy of HMEC surface receptors. We now use this new experimental system to investigate how protease-mediated ligand release rates quantitatively govern receptor-mediated signaling and consequent cell proliferation and migration behavior. A foundational first question is can a ‘dose-response’ relationship be obtained across a range of ligand release rates, in a manner analogous to the wellestablished approach used to ascertain effects of exogenous ligand-receptor interactions? A consequent second major question, then, is do significant differences exist between cell responses to ‘chronic’ (e.g. as in persistent autocrine ligand stimulation) as opposed to ‘acute’ (e.g. the typical experimental method of bolus exogenous ligand addition in culture) treatment under conditions of comparable receptor activation? One might anticipate that cell behavioral responses to a high autocrine ligand release rate could be similar to those for exogenous ligand treatment, but this issue has not previously been rigorously explored. To address these questions we used the 184A1 HMEC line, which is highly dependent on EGFR activation for both proliferation and migration (Dong et al., 1999; Stampfer et al., 1993). Low and high release rates of EGF were achieved using chimeric transmembrane EGF ligand precursors, with EGF itself as the receptor-binding mature form because it is the most completely characterized and thus most suitable for an initial basic study of fundamental effects of ligandpresentation mode. Ligand shedding is then ‘tuned’ downwards using the metalloprotease inhibitor batimastat, to obtain a quantitative spectrum of ligand release rates. We measured surface EGFR downregulation, ERK phosphorylation, cell proliferation and cell migration under

3690

Journal of Cell Science 120 (20)


conditions of low to high ligand release. We found that cell migration speed is directly proportional to the steady-state phospho-ERK level driven by autocrine ligand release, in a clear demonstration of a ‘dose-response’ relationship. Surprisingly, we found that migration speed is greater under ‘chronic’ autocrine ligand control than under ‘acute’ exogenous ligand treatment, even though the latter generates greater peak and integrated levels of phosphorylated ERK as well as equivalent cell proliferation responses. Thus, cell migration – which is understood to be driven crucially by spatially localized signals (e.g. Condeelis et al., 2005) – appears to be remarkably sensitive to the mode of ligand presentation, especially in comparison to a more ‘global’ response such as cell proliferation. Results EGF chimeras and metalloprotease inhibition can quantitatively tune autocrine ligand release rates In this study we chose two EGF chimeras that encode differently regulated proteolytic target sequences at the cell surface to investigate the effects of ligand release on cell signaling and migration. In addition to the previously described membrane-bound EGF ligand named ECT [termed EGF cytoplasmic tail (EGF-ct) in previous reports (e.g. Dong et al., 2005)], we constructed an EGF chimera that contained the transmembrane and cytoplasmic domains of TGF-␣, named TCT (TGF-␣ cytoplasmic tail) (see Fig. 1). We verified that TCT was correctly processed into EGF at the correct cleavage sites by mass spectrometry of immuno-precipitated ligand (data not shown). For both ECT and TCT, the core ligand that was released was EGF and is thus comparable to commercially available EGF. When we measured the fractional release (the fraction of cell-associated ligand released per hour) of the ECT and TCT constructs, we found that TCT was shed more efficiently than ECT (~0.6 hr–1 versus ~0.3 hr–1, respectively; data not shown). To demonstrate that the HMECs bind the ligand produced by both constructs equivalently, we modulated the amount of each expressed ligand using adenovirus titration. As shown in Fig. 3A (closed symbols), for the ECT and TCT constructs,

A Signal

Prepro

Core

TM Cytoplasmic domain

increasing amounts of adenovirus resulted in an increasing amount of ligand released by the cells. A significant proportion of the released ligand was taken up by cell surface EGFR, because the amount of ligand accumulation in the medium was increased by blocking the receptor with a saturating concentration of antagonistic antibody (Fig. 3A, open symbols). The amount of accumulated ligand in the presence of saturating antibody can be considered to be the overall production rate (ligand/cell/time), whereas the difference between the accumulated ligand in the presence versus absence of antibody can be defined as the cellular consumption rate (ligand/cell/time) – i.e. the amount taken up by the cell receptors upon release. Fig. 3B shows that ECT and TCT exhibit similar linear ratios of consumption to production across their respective dose ranges, illustrating constant and equivalent receptor-binding behavior following chimera release across the entire range of production. Although using adenovirus is an effective way to express different levels of ligands, it is only a transient expression system and thus cannot be used to understand the long-term biological consequence of ligand shedding. Thus, we used the more suitable retroviral expression system that we have previously used successfully to study the behavior of ligand mutants (Wiley et al., 1998; Dong et al., 2005). In order to facilitate release rate ‘tuning’ capability across the widest possible dynamic range, we generated cells expressing high levels of ECT and TCT using retrovirus transduction and then used an inhibitor of zinc-containing metalloproteases, batimastat, which we have previously demonstrated to be highly efficient in blocking the release of EGF ligands from the cell surface (Dong et al., 1999; Dong et al., 2005). Fig. 4A shows that we can obtain a roughly 300-fold dynamic range between ECT and TCT release rates across an 8-hour timeperiod. Accordingly, these two EGF constructs in themselves offer a ‘coarse-grained tuning’ between very high and very low autocrine ligand release rates. Next, Fig. 4B,C demonstrate our ability to further ‘fine-tune’ ligand release within this range via a graded titration across batimastat concentrations for the TCT construct, with an IC50 value of approximately 0.33 ␮M batimastat. Thus, we can use both EGF constructs as well as the addition of batimastat to systematically vary EGF release in a quantitative manner across a wide spectrum of rates.

pEGF Native Ligands

pTGFα Membrane anchoring domain ECT

Ligand Chimeras

TCT

B Ligand

Increasing autocrine release rates downregulates surface EGFR levels but nonetheless increases the levels of active EGFR The addition of exogenous EGF has been shown to downregulate receptor levels on the

Stem

Transmembrane domain

Cytoplasmic domain

ECT

…Y R DL KW W E L R HAGHG Q Q Q K V I V V AV C V V V L V ML L L L S LW GAHY Y R T Q…

TCT

…Y R DL L A V V AAS QK K Q A I T A L V V V S I V A L AV L I I T C V L I HC C QV R K HC…

Fig. 1. EGF ligand chimera constructs. (A) EGFR ligands are synthesized as integral membrane proteins that have variable membrane anchoring domains. Chimeras containing the receptorbinding domain of EGF (core) attached to the stem, transmembrane domain (TM) and cytoplasmic domain of either EGF or TGF-␣ were constructed as shown. (B) Amino acid sequence of each chimera showing the published cleavage site (arrowheads).

Autocrine EGFR-mediated cell migration

3691


Fig. 2. Sparse-labeled monolayer cell migration assay. Fluorescently labeled cells were plated 1:10 with unlabeled cells to form a confluent monolayer in a 96well plate. Images were acquired at 15-minute intervals for 6-8 hours using a Cellomics KineticScan. Labeled cells within the monolayer were tracked using Imaris. Cell paths from a sample field for a single condition are shown. Individual cell paths are displayed as a function of time.

surface of HMECs in culture (Burke et al., 2001). To determine the effects of autocrine EGF release on surface receptor levels, we measured steady-state binding of 125I-labeled 13A9, a noncompetitive EGFR monoclonal antibody (mAb), after 5 hours in the presence or absence of 2 nM exogenous EGF in the different cell lines (Hendriks et al., 2003). We confirmed receptor downregulation of the parental cells incubated with exogenous EGF (Fig. 5A). The ECT and TCT cells displayed lower levels of surface receptors than control cells, even in the absence of exogenous ligand. The addition of soluble EGF, however, reduced the receptor level of all cells to the same final degree (Fig. 5A). The TCT cells displayed the lowest steadystate EGFR levels, which were only marginally influenced by exogenous ligand addition. The effect of increasing the ligand release rate on the level of EGFR activation arises from a convolution of two countervailing processes: a greater fractional occupancy of surface receptors due to increased capture of ligand, combined with a lesser number of surface receptors due to increased endocytic downregulation. To assess the net outcome, we measured EGFR tyrosine phosphorylation using the Bio-Plex quantitative immunoprecipitation bead-based assay. As shown in Fig. 5B, despite the decrease in surface EGFR, the levels of phosphorylated EGFR were indeed enhanced at greater ligand release rates. This is consistent with a long-standing finding for exogenous EGF treatment that the degree of EGFR

downregulation is inversely proportional to receptor activity (Knauer et al., 1984). Thus, overall receptor activation increases concomitantly with increasing autocrine ligand release rates, despite the increasing associated degree of receptor downregulation that is induced. Autocrine signaling leads to steady ERK activation We next sought to determine how a key downstream signal varied with autocrine ligand release rate in concert with the increasing levels of EGFR activation. We found that autocrine presentation led to elevated phosphorylation levels of the EGFR-induced MAP kinase, ERK 1/2, measured by western blot over an 8 hour time-course (see Fig. 6A). Quantitative ERK1/2 phosphorylation dynamics were determined using the BioPlex assay. The cell medium was spiked with EGF to achieve a final concentration of 2 nM exogenous EGF or switched to serum-free media containing 10 ␮g/ml 225 EGFRblocking antibody (mAb225) for 2 hours prior to washing and replacing with fresh serum-free media. Although the parental cells exhibited a dramatic increase in phospho-ERK levels after EGF stimulation, the ECT and TCT cells showed phosphorylated ERK that increased to a significantly lesser degree following exogenous EGF stimulation before returning to basal levels (less than twofold increase of ECT and TCT cells compared with the greater than sixfold increase of parental cells; see Fig. 6B). Incubation with the receptor-

Fig. 3. EGF ligand production leads to proportional receptor binding and uptake. HMECs were treated with increasing amounts of adenovirus containing TCT and ECT expression constructs. (A) At approximately 40 hours after infection, the rate of EGF appearance in the extracellular medium was measured by ELISA over a 3 hour period in either the presence (white squares/circles) or absence (black squares/circles) of 10 ␮g/ml EGFR-blocking antibody 225. The amount of ligand observed in the presence of 225 is considered total EGF production, whereas the difference between the presence and absence of 225 reflects the amount of EGF consumed at that production level. (B) The relationship between the amount of EGF produced and consumed at each adenovirus MOI is shown both for TCT constructs (squares) and ECT constructs (circles).

3692


A

Molecules EGF Released/Cell

4

x10

blocking mAb225 for 2 hours abolished ERK phosphorylation in the autocrine cells (see Fig. 6B). After removing mAb225, several washes and adding fresh serum-free media, ERK phosphorylation steadily increased because of autocrine presentation over a 2 hour period before reaching initial levels. After adding varying batimastat concentrations, we found phospho-ERK1/2 to be an asymptotically increasing function of autocrine ligand release rate (Fig. 6C).

6

ECT TCT 3 ~6000 2 Slope = Ligand Release Rate (molecules/cell/min)

1

~20

0

EGF in Media at 8 hrs (ng/ml)

C

EGF in media (ng/ml)


B

0

2

0 0.001

0.01

4

6 Time (hr)

8

10

10

100

8

10

3

2

1

3

0.1 1 Batimastat Conc (μM)

Serum-Free Media 0.12 μM Batimastat 0.33 μM Batimastat 0.86 μM Batimastat

2

1

0 0

2

4 6 Time (hr)

Fig. 4. EGF ligand release can be quantitatively tuned across a wide dynamic range. (A) Ligand release rate is measured from the slope of EGF accumulation over time. After 16 hours in serum-free media, cells were switched to fresh serum-free media or to media containing an inhibitor and samples were collected at 2-hour time intervals from triplicate wells. EGF concentration was measured using an EGF ELISA. The approximate ligand release rates (molecules of EGF released/cell/minute), verified in independent experiments, are shown next to each linear fit. (B) TCT release into the media over an 8 hour time period is inhibited using increasing concentrations of batimastat. Fitting the conditioned EGF concentration to a doseresponse curve reveals a median inhibition concentration of 0.33 ␮M batimastat. (C) Adding increasing concentrations of batimastat down-modulates the linear ligand release of TCT cells. Error bars represent one standard deviation from the mean of triplicate samples.

Autocrine presentation induces normal proliferation but enhanced migration To determine the degree to which cell responses might depend on the rate of autocrine ligand release, and whether they might differ for chronic (autocrine) versus acute (exogenous) stimulation, we measured cell proliferation and cell migration. These two responses are also crucial elements driven by EGFR activity in tumor progression. Cells were counted under the low and high ligand release rates in serum-free media and compared to parental cells in regular DFCI-1 culture media containing 2 nM EGF. The relative growth rates, calculated from an exponential fit to the growth kinetics, were found to be essentially invariant: 3.7, 3.6 and 3.2% increase/hour for the parental, ECT and TCT cells, respectively (Fig. 7). Thus, while these cells are dependent on EGF for cell growth (Stampfer et al., 1993), the low and high levels of autocrine signaling induced by the ligand chimeras did not alter the levels beyond control behavior. Cell migration was measured using high-throughput imaging of multiple individual cells within a monolayer. This assay monitors the movement of a population of confluent cells while capturing individual cell variation and avoiding artifactual single-cell migration behavior arising from expression/cloning variability. Movement within the context of a cell population is a more physiologically relevant situation than dispersed, low-density single-cell tracking (Condeelis et al., 2005). After cells were incubated in serum-free media for 15 hours, cells were switched to fresh serum-free media with or without exogenous EGF, batimastat or mAb225. Cells were monitored via the Cellomics KineticScan instrument for 6-8 hours with images acquired at 15-minute intervals. Exogenous EGF stimulation increased parental cell migration in a dosedependent manner, but the ECT and TCT cells exhibited even more vigorous motility (Fig. 8A,B). Population-averaged values of cell speed were determined from the individual cell paths under the various experimental conditions (>200 paths were analyzed in each case), and Fig. 8B shows that the graphical trends observed in Fig. 8A are quantitatively valid. In addition, Fig. 8B shows that the high migration speeds induced by autocrine ligand stimulation are decreased by treatment with 10 ␮g/ml mAb225 and 10 ␮M batimastat, as expected. Interestingly, exogenous EGF did not significantly enhance the autocrine ligand-induced cell motility, possibly because of the relatively unaltered ERK phosphorylation under these conditions (Fig. 6B). Monitoring migration speed as a function of time showed that the trends in Fig. 8B are maintained throughout the experiment; moderate reduction occurs over time for all conditions, most probably caused by mild cytotoxic effects of the cell tracker dye or phototoxicity (see supplementary material Fig. S3). We were interested in whether the chronic exposure to EGF in the autocrine cells increased cell motility because of gene


A

B 10000

Serum-Free Media 2nM EGF

8000 pTyrEGFR (AU)

Surface Receptors (#/cell)

400,000 300,000 200,000

Parental ECT TCT TCT + 0.12μM Batimastat TCT + 0.33μM Batimastat TCT + 0.86μM Batimastat

6000 4000 2000

100,000 0 0

0 Parental


3693

ECT

TCT

2000

4000

6000

8000

Ligand Release Rate (molecules/cell/min)

Fig. 5. Autocrine signaling leads to surface EGFR downregulation and tyrosine phosphorylation in a ligand-release-rate-dependant manner. (A) Surface receptor numbers were quantified using an equilibrium-binding assay with 125I-labeled 13A9 mAb. Cells were serum-starved overnight before incubating the cells in fresh serum-free media containing labeled 13A9 alone (black bars) or with 2 nM exogenous EGF (gray bars) and allowed to reach steady state (5 hours), as described in the Materials and Methods. Surface-bound antibody was determined by counting acid strip solutions on a gamma counter. Receptor numbers were normalized to the average cell number counted from parallel plates. Error bars represent one standard deviation from the mean of triplicate wells. (B) Phospho-EGFR levels were measured using a quantitative immunoprecipitation bead-based tyrosine EGFR BioRad Bio-Plex phosphoprotein detection assay. Parental, ECT or TCT cells were switched to fresh serum-free media for 2 hours and then lysed. In addition, TCT cells were incubated for 2 hours with increasing concentrations of batimastat (0.12, 0.33 and 0.86 ␮M) to achieve additional release rates prior to lysis. Error bars represent standard deviation from triplicate lysates.

expression effects independent of ligand presentation. Thus, after 14 hours under autocrine conditions in serum-free media, we incubated the ECT and TCT cells for 2 hours with 10 ␮M of batimastat to block ligand shedding followed by stimulation with 2 nM of exogenous EGF. As shown in Fig. 8C, exogenous stimulation led to similar levels of stimulated cell motility of the parental, ECT and TCT cells treated with batimastat. The slight increases in cell speed of the ECT and TCT cells is probably due to the incomplete inhibition of ligand shedding by batimastat (the ligand release rate of TCT treated with 10 ␮M batimastat is ~160 molecules/cell/minute compared with ~6000 in serum-free media alone, data not shown). Therefore, the chronic exposure to autocrine EGF does not appear to ‘prime’ the autocrine cells for elevated cell speed in response to EGF. In addition, migration of parental cells after 16 hours of pre-treatment with exogenous EGF demonstrated that chronic exposure to exogenous ligand did Fig. 6. Autocrine stimulation leads to sustained ERK phosphorylation. (A) Phosphorylation of ERK1/2 measured over an 8-hour time course. Cells were incubated in serum-free media (SFM) and lysed at the indicated time points (16, 20, 22 or 24 hours). Blots were stripped and re-probed for total ERK1/2. (B) Phospho-ERK (pERK1/2) dynamics measured using a quantitative immunoprecipitation bead-based ERK1/2 BioRad Bio-Plex phosphoprotein detection assay. Cells were incubated in serum-free media for 16 hours and then spiked with 2 nM exogenous EGF (black circles, squares and diamonds) or blocked with 10 ␮g/ml of 225 for 2 hours prior to three PBS washes and serum-free media replacement (white circles and squares). Error bars represent one standard deviation from the mean of triplicate lysates. (C) ERK phosphorylation was measured from triplicate lysates after 2 hours in serum-free media. In addition, TCT cells were incubated for 2 hours in batimastat (0.12, 0.33, 0.86 and 10.00 ␮M) to achieve additional release rates before lysis. Error bars represent standard deviation of triplicate lysates.

3694


Cell Number

1,000,000

100,000 Parental DFCI1 Parental SFM ECT SFM TCT SFM 10,000 0

24

48

72

96

Time (hr)


Fig. 7. Autocrine presentation stimulates normal cell proliferation. Cell growth rates were measured by cell counting with a Vi Cell XR. Cell number was measured over 3 days in regular DFCI-1 media or serum-free media (SFM). Error bars represent standard deviation of triplicate measurements.

not in itself lead to the high cell speed of the TCT cells (see supplementary material Fig. S4). Chronic exposure to exogenous EGF did appear to increase the sensitivity of the parental cells to the lowest concentration of EGF (0.2 nM). However, autocrine presentation resulted in higher maximal cell speeds than could be achieved under exogenous stimulation alone. Autocrine ligand presentation also appeared to yield increased cell persistence (data not shown). In previous studies of EGF ligand-stimulation-mode effects on HMECs under sparsely distributed single-cell migration conditions we had similarly observed greater directional persistence of migration in response to autocrine presentation compared with exogenous presentation (Maheshwari et al., 2001). However, persistence behavior is very difficult to interpret rigorously in context of the confluent monolayer assay used in our current study, because of the inherent continuous cell-cell interactions and the geometric restriction of migration. Cell migration speed is proportional to ERK1/2 phosphorylation as autocrine ligand release rate is varied Given the concomitant increases in ERK1/2 phosphorylation levels and cell migration speed as autocrine EGF release rate is increased, we sought to ascertain the quantitative nature of this relationship. Fig. 9A shows the average cell migration speed against phospho-ERK1/2 levels measured at 2 hours after treatment for the parental, ECT and TCT cells (see supplementary material Fig. S5 for phospho-ERK measurements alone). Migration speed was directly proportional to phospho-ERK1/2 levels for the autocrine cells (ECT, TCT) as well as for the parental cells, but the slope of this relationship was much steeper for the autocrine-induced cells relative to exogenous-stimulated parental cells. Even though exogenous-stimulated parental cells can generate phospho-ERK1/2 levels as high as or higher than the autocrineinduced cells, the former cannot attain nearly as great a migration speed. To test whether these differential ERK-migration relationships are merely correlative or are causative, the cells

were treated with a spectrum of concentrations of PD98059 – a MEK inhibitor – under both exogenous and high autocrine stimulation. (Quantitative dose-response relationships for both phospho-ERK and migration speed as functions of inhibitor concentration are shown in supplementary material Fig. S6.) When the migration speeds in the presence of PD98059 were plotted against the corresponding phospho-ERK values, the same relationships were obtained as in the case of varying ligand release rates (Fig. 9B). Therefore, we conclude that cell migration speed depends crucially on the levels of phosphoERK in the case of both autocrine and exogenous ligand presentation. However, autocrine signaling induces a higher migration speed for any given level of ERK activity. Discussion EGFR family signaling plays a key role in governing multiple cell phenotypic behaviors, including proliferation, migration and differentiation (Wells, 1999). Moreover, its dysregulation is associated with various pathologies, such as tumor progression (Condeelis et al., 2005; Wells, 2000; Normanno et al., 2006), and it is considered to be a broadly useful therapeutic target system for cancer and other diseases (Bublil and Yarden, 2007; Coffey et al., 2007; Johnston et al., 2006). The vast preponderance of basic science and clinical application studies has focused on the receptors and downstream signaling pathways as key regulators of pathophysiological cell behavior. However, the crucial and highly complex nature of ligand presentation in the EGFR system is beginning to be appreciated (Carpenter, 2000; Harris et al., 2003; Singh and Harris, 2005). Autocrine stimulation by EGF family ligands does not function only in pathologies but has been found vital for many developmental and remodeling processes that invoke cell proliferation and migration (Conti et al., 2006; El Hader et al., 2005; Chu et al., 2005; Gu et al., 2007; Semino et al., 2006; Xu et al., 2007; Ross et al., 2001; Rosen, 2003). Because both mammary gland development and breast cancer invasion have been shown to use increased autocrine EGF ligand activity (Rosen, 2003; Kenny and Bissell, 2007; Seton-Rogers et al., 2004) we explored this cellular process in terms of migration and proliferation in mammary epithelial cells. Although information has been gathered concerning the regulation of autocrine ligand release by ligand structural features and metalloprotease identities (Sahin et al., 2004; Dong et al., 2005; Ohtsu et al., 2006), little is understood about how the protease-mediated ligand release rates govern receptor-mediated signaling and consequent cell behavior. Even such a basic foundational point as the simple ‘doseresponse’ dependencies of receptor downregulation, downstream signaling and phenotypic cell functions on autocrine ligand release rate have not been established for endogenous ligand stimulation, despite the centrality of this kind of information for understanding the underlying regulatory mechanisms. Moreover, there has been little exploration of potential differences between cell responses induced by the common but non-physiological ‘acute’ stimulation by exogenous ligand treatment versus those arising from the more physiological (but much more difficult to generate experimentally) ‘chronic’ stimulation by endogenous ligand. To begin addressing the question of how different autocrine


A

Parental

ECT

3695

TCT

Serum-Free Media

+2nM EGF

-100

B

40 Average Cell Speed (μm/hr)

Fig. 8. Autocrine stimulation leads to elevated cell migration speed. Cell migration was measured in a sparse-labeled monolayer high-throughput assay. Parental, ECT and TCT cells were starved for 15 hours and then switched to fresh serumfree media, 0.2 nM EGF, 2 nM EGF, 10 ␮g/ml mAb225 or batimastat (0.12, 0.33, 0.86 or 10.00 ␮M) ~1 hour before imaging. Fluorescence images were acquired at 15minute intervals for 8 hours and imported in Imaris for single-cell tracking of the labeled cells. Cell paths were analyzed in Matlab. (A) Cell paths are shown starting at each origin. (B) Average cell migration speed under each condition. (C) The average cell speed was measured for all cell types when stimulated with exogenous EGF while ligand cleavage at the cell surface was inhibited with batimastat. After 14 hours of starvation in serum-free media, the cells were treated for 2 hours with 10 ␮M batimastat and then switched to fresh batimastat alone or with 2 nM EGF. Cell speed is compared to that measured under serum-free media alone for each cell type. Error bars represent s.e.m., and n >200 cells per condition.

35 30

35 30

25

25

20

20

15

15

10

10

5

5

Serum-Free Media 0.12μM Batimastat 0.33μM Batimastat 0.86μM Batimastat 10μM Batimastat

0 Parental

C

0 100 μm

40

Serum-Free Media 0.2nM EGF 2nM EGF 10μg/ml 225 mAb

0 ECT

TCT

Parental

TCT

40 Average Cell Speed (μm/hr)


+10μg/ml 225 mAb

35 30 25

10μM Batimastat 10μM Batimastat + 2nM EGF Serum-Free Media

20 15 10 5 0 Parental

ligand production rates influence cell responses, we developed a set of EGFR ligand chimeras that vary in their transmembrane and cytosolic anchor regions (ECT and TCT; see Fig. 1) to ‘coarse-grain tune’ EGF autocrine ligand release rate with the accompanying ‘fine-tuning’ by the metalloprotease inhibitor batimastat. We have shown

ECT

TCT

previously that batimastat is a very effective inhibitor of ligand shedding in HMECs, is relatively non-toxic and shows few obvious secondary effects (Dong et al., 1999). In line with these expectations, we saw little difference in the migration of cells in the presence and absence of batimastat when treated with exogenous EGF. Although using metalloprotease


Cell Migration Speed (μm/hr)

A

B 40 Parental SFM Parental + 0.2nM EGF Parental + 2nM EGF ECT SFM TCT SFM TCT + 0.12μM Bat TCT + 0.33μM Bat TCT + 0.86μM Bat TCT + 10μM Bat

Increase Ligand Release Rate 30 20 10

Parental +EGF

0 0

2000

4000

6000

8000

pERK 1/2 (AU)

Cell Migration Speed (μm/hr)

3696

40 30 20 10

Increase MEK Inhibition

0 0

2000 4000 pERK 1/2 (AU)

Parental SFM Parental + 2nM EGF Parental + EGF + 0.04μM PD Parental + EGF + 0.2μM PD Parental + EGF + 1μM PD Parental + EGF + 5μM PD Parental + EGF + 25μM PD TCT SFM TCT + 0.04μM PD TCT + 0.2μM PD TCT + 1μM PD TCT + 5μM PD TCT + 25μM PD

6000


Fig. 9. Cell migration speed is a function of ligand release rate and ERK phosphorylation. ERK phosphorylation was measured from triplicate lysates using a BioPlex assay (see supplementary material Fig. S5 for ERK data alone). The parental cells were lysed after a 2-hour incubation with serum-free media (SFM) or exogenous EGF (0.2 and 2.0 nM). ECT cells were lysed after a 2-hour incubation in fresh serum-free media. In addition, TCT cells were lysed after a 2-hour incubation in serum-free media or increasing concentrations of batimastat (0.12, 0.33, 0.86 and 10.00 ␮M Bat). (A) Cell migration speed (shown in Fig. 7, error represents s.e.m., and n >200 cells) is plotted as a function of ERK phosphorylation (error represents standard deviation from triplicate lysates). (B) Parental and TCT cells were pre-incubated with multiple concentrations of the MEK inhibitor PD98059 (0.04, 0.20, 1.00, 5.00 and 25.00 ␮M PD) for 30 minutes. Parental cells were then stimulated with 2 nM of EGF under the same PD conditions. Cell migration speed is shown as a function of ERK1/2 phosphorylation measured at 2 hours after exogenous stimulation (see supplementary material Fig. S6 for ERK and migration data separately).

inhibitors is not an ideal way to control ligand-shedding rates, it is currently the most practical approach for cells in which retroviral-mediated gene transduction must be used. We used EGF as the mature receptor-binding core ligand in our chimeras because our aim focused on basic questions concerning the effects of ligand presentation rather than a focus on a specific physiology or pathology application for which it is not necessarily the ligand family member involved. For future studies aimed at particular tissue development or homeostasis applications, in which other EGF ligand family members are involved (including TGF␣, AR and HB-EGF), the behavior of the ligand(s) most relevant to that application should be compared in the correspondingly relevant cell type. To demonstrate the usefulness of our approach, we measured proximal receptor downregulation and activation as a function of ligand release rate. Although receptor downregulation has been previously demonstrated in HMECs using exogenous EGF (Burke et al., 2001), we found that we could progressively decrease surface receptor levels by increasing ligand release rates (Fig. 5A). This was anticipated based on our earlier biophysical studies using the mouse B82 cell line (DeWitt et al., 2001). The ECT cells have sub-saturating receptor downregulation that can be further downregulated with exogenous EGF, whereas the TCT cells have sustained low levels of surface receptors that are not further reduced by additional exogenous EGF treatment. These quantitative surface receptor measurements suggest that, in ECT cells, VLVR, where VL is the rate of ligand production and VR is the rate of EGFR production (DeWitt et al., 2001). As anticipated (Knauer et al., 1984), decreasing total numbers of cell surface receptor levels were correlated with enhanced EGFR activation (Fig. 5B). This is consistent with the idea that the occupancy-induced internalization of the EGFR, which leads to downregulation, is a mechanism to enhance ligand capture rather than attenuate signaling (Shankaran et al., 2007); that is, endocytosis permits clearing of ligand/receptor complexes from the cell surface,

permitting ensuing new rounds of ligand binding by recycled or newly synthesized receptors. Overall, the relationship between ligand production, receptor levels and receptor activation corresponds to our previous findings with nonresponsive, transformed model cell lines, indicating that the ligand-receptor dynamics of autocrine systems are consistent and reproducible. A powerful aspect of the current study is that, by using our HMEC experimental system, we can link ligand-receptor dynamics to biologically relevant cell responses. Downstream of the EGFR, both the ECT- and TCT-expressing cells displayed sustained ERK phosphorylation and responded only transiently to exogenous EGF stimulation (Fig. 6A,B). Thus, after 1 hour of treatment with exogenous EGF, the ECT and TCT cells returned to their steady levels of ERK phosphorylation, whereas the parental cells achieved much higher levels of activation that remained above autocrine levels at 2 hours of stimulation. This initial peak difference can be largely accounted for by the lower receptor levels present on the ECT and TCT cells relative to the parental cells, due to the autocrine ligand-induced receptor downregulation (Fig. 5A). The dynamic overshoot and subsequent relaxation to a signaling steady-state that we observed following initial stimulus treatment has been considered to arise from a quantitative mismatch of receptor-ligand binding rates versus signal activation and deactivation rates (Kholodenko et al., 1999; Resat et al., 2003). It is conceivable that these dynamics might be significantly influenced by negative-feedback mechanisms known to be induced by acute ligand stimulation (Amit et al., 2007; Hackel et al., 2001). Conversely, after blocking autocrine activation for 2 hours with mAb225, EGFR tyrosine phosphorylation and ERK phosphorylation steadily increased in serum-free media (Fig. 6B; also see supplementary material Fig. S2). Autocrine stimulation never led to the high peaks or transient dynamics characteristic of acute exogenous stimulation, even at the highest ligand release rates.


Autocrine EGFR-mediated cell migration We found that cell proliferation stimulated by low and high levels of autocrine ligand release was similar to that observed in response to exogenous EGF (Fig. 7). In contrast, we found a significant difference with respect to cell migration behavior, with migration speed exhibiting a strong quantitative dependence on ligand release rate (Fig. 8). This relationship was verified in both a positive and negative manner; higher speed resulted from the higher release rate TCT construct compared with the lower release rate ECT construct, and lower speeds resulted from diminution of TCT release by increasing concentrations of the protease inhibitor batimastat. Furthermore, we found that migration speed appears to be proportional to the levels of ERK phosphorylation – whether modulated by increasing ligand shedding (Fig. 9A) or by increasing the concentration of the MEK inhibitor PD98059 (Fig. 9B). ERK is known to be crucial for the regulation of multiple key biophysical processes underlying cell migration (Brahmbhatt and Klemke, 2003; Glading et al., 2001; Huang et al., 2004). This separation of cell responses by presentation of ligand has significant implications for epithelial-stromal communications and might explain in part how these separate compartments can communicate even when using ‘cues’ that are also self-produced. Interestingly, the migration behavior of cells driven by autocrine stimulation and those driven by exogenous EGF stimulation separate onto disparate curves in the speed-versus-pERK ‘dose-response’ plot (Fig. 9A,B). The cells driven by autocrine ligand shedding attained much higher migration speeds than those driven by exogenous ligand for any given level of ERK phosphorylation. This finding might reflect different pools of ERK being activated (Glading et al., 2001); alternatively, or concomitantly, it could mean that the EGFR signaling network can act in both a ‘local’, spatially restricted mode directing cell migration, and a global mode inducing cell proliferation. A recent mass spectrometry study of EGFR signaling in these same 184A1 HMECs demonstrated that numerous different substrates are phosphorylated and multiple signaling pathways are activated by the addition of a bolus of EGF, but only a subset of these pathways are correlated with enhanced cell migration (Wolf-Yadlin et al., 2006). Autocrine signaling, even in our modified cell system, involves the spatially regulated release of ligand prior to its binding and activation of the EGFR. Thus, it is likely that the pattern of substrate phosphorylation is distinct between cells treated with exogenous EGF versus endogenously produced EGF. However, because significant differences between autocrine and exogenous ligand effects were not found for cell proliferation, some important EGFR effectors might not be spatially restricted. Indeed, it has been shown that EGFR signaling at the apical versus basolateral surfaces of cells results in the phosphorylation of common as well as distinct substrates (Kuwada et al., 1998). Further studies are being pursued to critically test this hypothesis. In any event, the kind of rigorously quantitative dose-response capabilities that we have established here should play a vital role in examining molecular pathway regulatory mechanisms in future studies. Materials and Methods Reagents and antibodies Batimastat {BB-94; [4-(N-hydroxyamino)-2R-isobutyl-3S-(thienylthiomethyl)succinyl]-L-phenylalanine-N-methylamide} was custom synthesized by Kimia Corporation (Santa Clara, CA). ERK phosphorylation was inhibited with the small molecule MEK inhibitor PD98059 (Calbiochem). Anti-EGFR 13A9 mAb,

3697

which binds non-competitively to both occupied and empty EGFR (Winkler et al., 1989), was a gift from Genentech (South San Francisco, CA). The receptor-blocking anti-EGFR mAb225 was isolated from a hybridoma cell line obtained from the American Type Culture Collection (Gill et al., 1984). Anti-EGF antibodies for EGF ELISA include 236 (R&D Systems) and a rabbit polyclonal described previously (Wiley et al., 1998). The tertiary ELISA detection antibody was an alkaline phosphatase-conjugated goat anti-rabbit antibody (Sigma). Recombinant human EGF was obtained from Peprotech. ERK antibodies used in western blots were purchased from Cell Signaling. Luminex phospho-tyrosine EGFR and phosphoERK1/2 BioPlex kits and lysis buffer were purchased from Bio-Rad.

Construction of ligand chimeras and cell lines The TCT ligand (EGF with the membrane-anchoring region and cytoplasmic tail of TGF-␣) was assembled initially in pBluescript using a construct that contained the mature coding sequence of EGF (amino acids 971-1023) preceded by the signal sequence of the EGFR. A BglII site was engineered at the 3⬘ end of the EGF sequence followed by an XbaI site from the original pBluescript multicloning region. Ligand-specific sequences that included the juxtamembrane, transmembrane and cytoplasmic tails were amplified using PCR primers that contained a BglII site in the 5⬘ primer and an XbaI and BamHI site in the 3⬘ primer. The following primer sequences were used: TGF-␣ front primer: 5⬘-ACTTAAGATCTCCTGGCCGTGGTGGCTGCCAGCCAGA-3⬘; TGF-␣ back primer: 5⬘-CCGCTCTAGAACTAGTGGATCCCCTCTTCAGACCACTGTTTCTGAGTG-3⬘. Following assembly in pBluescript, the chimeric ligand was subcloned into the retroviral vector pBMIRESpuro (Garton et al., 2002). Retroviral supernatants were collected from transfected HEK293 cells and used to transduce the parental HMEC cell line. The ECT ligand (EGF with the membrane-anchoring region of EGF) was constructed as described previously (Wiley et al., 1998; Dong et al., 2005). These constructs are shown schematically in Fig. 1. The HMEC line 184A1 (Stampfer et al., 1993) was obtained from Martha Stampfer (Lawrence Berkeley National Laboratory, University of California, Berkeley, Berkeley, CA). The HMECs were retrovirally transduced with the ECT and TCT constructs and maintained in DFCI-1 medium (Band and Sager, 1989). Serum-free media consisted of DFCI-1 complete media lacking fetal bovine serum, bovine pituitary extract or EGF supplements. Unless otherwise noted, cells were incubated for 16 hours in serum-free media before starting each experiment. All experiments were performed using cells within 15 passages of thawing. Adenovirus vectors encoding the TCT and ECT ligands were produced by insertion of TCT or ECT DNA into the pADTet7 shuttle vector (Hsia et al., 2003) using the BstXI (blunted) and NotI sites. The recombinant pADtet-TCT vector was co-transfected into HEK 293 cells expressing Cre recombinase (kind gifts from Dan Streblow and Jay Nelson, OHSU, TX) together with ␺5 viral DNA (Hardy et al., 1997). The pAdTet7 vector contains the Tet-responsive cassette that includes the tet-responsive enhancer sequences, CMV minimal promoter and SV40 late polyA signal as well as a single loxP site. The ␺5 adenoviral backbone DNA contains an E1A/E3-deleted adenoviral genome. The recombinant viral construct can be propagated to high titer because no protein is expressed unless cells are co-infected with a second adenovirus encoding the Tet transactivator (Streblow et al., 1999). Recombinant adenovirus was expanded in HEK293-Cre cells, and stocks were titered by limiting dilution. To express ligands using adenovirus, HMECs were plated using a minimum of 1.2⫻106 cells into 60 mm dishes 8 hours prior to infection. For infection, equal amounts of recombinant and transvirus were added in a total of 2.5 ml together with polybrene at 6 ␮g/ml. The cells were incubated overnight in virus, washed and split into two wells of a six-well plate for assays.

EGF release rates Cells expressing ligands from retrovirus vectors were plated in DFCI-1 media in 12-well plates on day 1 at a density of 60,000-80,000 cells/well, and switched to serum-free media on day 2. After 16 hours cells were switched to 1 ml of fresh serum-free media alone or containing 225 or batimastat at varying concentrations. At 2-hour time intervals, media was collected from parallel plates, centrifuged at 16,873 g for 10 minutes at 4°C, and the supernatant was frozen. The EGF concentration of the conditioned media, measured by ELISA, was converted to molecules of EGF and normalized to the average cell number per well measured with a Vi Cell XR. The rate of EGF release was determined from the slope of a linear fit. Cells expressing ligands from adenovirus were changed to DFCI-1 lacking EGF following plating and incubated overnight. The next morning (about 40 hours after the addition of virus) the EGFR-blocking mAb225 (10 ␮g/ml) was added to half of the cells for 1 hour. Cells were washed in PBS and then serum-free medium +/– 225 mAb225 was added to the plates. After a 3-hour incubation, the medium was collected and the EGF concentration of the conditioned media was measured by ELISA, was converted to molecules of EGF and normalized to the average cell number per well.

Surface receptor levels Anti-EGFR mAb 13A9 was labeled with 125I (Perkin Elmer) using iodobeads (Pierce) as previously described (Burke and Wiley, 1999). Labeled 13A9 was used

3698


to measure surface EGFR numbers as previously described (Hendriks et al., 2003). Briefly, 150,000 cells were plated in six-well plates on day 1 and switched to serumfree media on day 2. After 19 hours, cells were switched to fresh serum-free media containing 600 ng/ml of radiolabeled 13A9 alone or in addition to 2 nM EGF and incubated at 37°C for 5 hours. Surface-bound antibody was removed with an acid strip and quantified using a Gamma counter from triplicate wells. Parallel plates were trypsinized and counted using a Vi Cell XR to determine the average number of cells per well. The experiment was repeated on a separate day to confirm the results, and surface receptor numbers are reported here as an average per cell measured in triplicate wells on a single day.


EGFR and ERK phosphorylation Cells were plated in six-well plates at 150,000 cells/well. Cells were starved in serum-free media for 16 hours, treated as described in the figure legends, rinsed once with cold PBS and subsequently scraped in lysis buffer, and prepared as previously described (Janes et al., 2003) for western blotting or in Phosphoprotein Lysis Buffer (Bio-Rad) for the Luminex assays. Lysates were incubated on ice for 10 minutes and then centrifuged at 16,873 g for 10 minutes at 4°C. The supernatants were frozen at –80°C until use. Lysates were analyzed using a bicinchonicic assay (Pierce) to determine the total protein concentration. For western blotting, equal amounts of total lysate protein were diluted in lysis buffer and 4⫻ sample buffer, run on 7.5% gels, and transferred to PDVF membranes (Bio-Rad) before blocking with 5% BSA in TBS-T and incubating membranes overnight at 4°C with a primary p44/42 ERK antibody (Cell Signaling). The membranes were stripped and reprobed for total ERK (Cell Signaling). For the Luminex assays, which are quantitative beadbased assays, lysates were diluted in Bio-Rad Lysis Buffer and Bio-Rad Assay Buffer to a final protein concentration of 150 ␮g/ml (7.5 ␮g of protein per well). Linearity of the pTyrEGFR and pERK 1/2 Bio-Rad assays was checked using varying ratios of mixed lysate from EGF stimulated and unstimulated parental cells (see supplementary material Fig. S1) and results were used to determine the optimal loading per well. Phosphotyrosine EGFR and phosphorylated ERK 1/2 (Thr202/Tyr204, Thr185/Tyr187) were both measured using Bio-Rad phosphoprotein single-plex kits on a Luminex as per protocol (Bio-Rad).

Proliferation Cells were plated in six-well plates at a low initial density of 50,000 cells/well in regular DFCI-1 media. On day 2, cells were counted using a Vi Cell XR and parallel plates were switched to either regular or serum-free media. Cells from different plates were counted in triplicate at 24-hour intervals for a total of 3 days in each media. ECT and TCT cell growth was similar under both media conditions, and data are shown comparing relative growth rates of autocrine-stimulated proliferation in serum-free media to parental cells. Specific cell growth rate constant was determined from an exponential fit.

Migration Single-cell migration was measured from a cell monolayer using a modified highthroughput assay (Kumar et al., 2006); see Fig. 2. A confluent plate of cells was incubated with 8 ␮M of a fluorescent cell tracker dye, CMFDA (Invitrogen), for 25 minutes at 37°C, rinsed with PBS and then trypsinized along with a parallel plate of unlabeled cells. Cells were counted with a Vi Cell XR, mixed at a ratio of 1:10 labeled to unlabeled cells and plated at 40,000-50,000 cells per well in a 96-well plate (Packard) in DFCI-1 complete media. The optimal cell density to obtain a nonovercrowded confluent monolayer was determined by plating increasing numbers of cells under the same conditions and visualizing cell density with a bright-field microscope (data not shown). After 4 hours, cells were switched to 100 ␮l of serumfree media and after 15 hours the medium was switched to 100 ␮l of serum-free media with or without stimulation. MEK inhibition involved a 30-minute preincubation with PD98059 (0.04, 0.2, 1, 5 and 25 ␮M) prior to stimulation (see Fig. 8 legend). Migration of the parental cells was also measured after 16 hours of EGF treatment, which is potentially similar to the constant exposure of EGF under autocrine conditions. Parental cells were pre-treated with exogenous EGF (0.2, 2 and 25 nM) for 15 hours and then stimulated with fresh EGF-containing media (same concentration) at around 1 hour prior to imaging. Fluorescent images were acquired using a Cellomics KineticScan at 15-minute intervals for 6-8 hours. Image files were exported to Imaris for cell tracking and dynamic cell coordinates were analyzed in Matlab. All cell movement paths lasting the duration of the experiment were included in the analysis. We report the average cell speed and standard error of the mean (s.e.m.) for a minimum of 200 cells measured from multiple wells per condition.

The authors wish to thank Christina Lewis for assistance with the receptor downregulation assay, and Neil Kumar and Hyung-Do Kim for migration assay development; a portion of the latter studies were undertaken in the Whitehead-MIT BioImaging Center. Financial support is gratefully acknowledged from National Cancer Institute grants R01-CA096504 and U54-CA112967 to D.A.L., R01GM069668 to A.W., a Whitaker Foundation Graduate Fellowship to

E.J.J., and the Laboratory Directed Research and Development Program at the Pacific Northwest National Laboratory to L.K.O. and H.S.W.

References Amit, I., Citri, A., Shay, T., Lu, Y., Katz, M., Zhang, F., Tarcic, G., Siwak, D., Lahad, J., Jacob-Hirsch, J. et al. (2007). A module of negative feedback regulators defines growth factor signaling. Nat. Genet. 39, 503-512. Arribas, J., Lopez-Casillas, F. and Massague, J. (1997). Role of the juxtamembrane domains of the transforming growth factor-alpha precursor and the beta-amyloid precursor protein in regulated ectodomain shedding. J. Biol. Chem. 272, 17160-17165. Band, V. and Sager, R. (1989). Distinctive traits of normal and tumor-derived human mammary epithelial cells expressed in a medium that supports long-term growth of both cell types. Proc. Natl. Acad. Sci. USA 86, 1249-1253. Baselga, J., Mendelsohn, J., Kim, Y. M. and Pandiella, A. (1996). Autocrine regulation of membrane transforming growth factor-alpha cleavage. J. Biol. Chem. 271, 32793284. Bates, S. E., Valverius, E. M., Ennis, B. W., Bronzert, D. A., Sheridan, J. P., Stampfer, M. R., Mendelsohn, J., Lippman, M. E. and Dickson, R. B. (1990). Expression of the transforming growth factor-alpha/epidermal growth factor receptor pathway in normal human breast epithelial cells. Endocrinology 126, 596-607. Borrell-Pages, M., Rojo, F., Albanell, J., Baselga, J. and Arribas, J. (2003). TACE is required for the activation of the EGFR by TGF-alpha in tumors. EMBO J. 22, 11141124. Brahmbhatt, A. A. and Klemke, R. L. (2003). ERK and RhoA differentially regulate pseudopodia growth and retraction during chemotaxis. J. Biol. Chem. 278, 1301613025. Bublil, E. M. and Yarden, Y. (2007). The EGF receptor family: spearheading a merger of signaling and therapeutics. Curr. Opin. Cell Biol. 19, 124-134. Burke, P. M. and Wiley, H. S. (1999). Human mammary epithelial cells rapidly exchange empty EGFR between surface and intracellular pools. J. Cell. Physiol. 180, 448-460. Burke, P., Schooler, K. and Wiley, H. S. (2001). Regulation of epidermal growth factor receptor signaling by endocytosis and intracellular trafficking. Mol. Biol. Cell 12, 18971910. Carpenter, G. (2000). EGF receptor transactivation mediated by the proteolytic production of EGF-like agonists. Sci. STKE 2000, PE1. Chen, W. N., Woodbury, R. L., Kathmann, L. E., Opresko, L. K., Zangar, R. C., Wiley, H. S. and Thrall, B. D. (2004). Induced autocrine signaling through the epidermal growth factor receptor contributes to the response of mammary epithelial cells to tumor necrosis factor alpha. J. Biol. Chem. 279, 18488-18496. Chu, E. K., Foley, J. S., Cheng, J., Patel, A. S., Drazen, J. M. and Tschumperlin, D. J. (2005). Bronchial epithelial compression regulates epidermal growth factor receptor family ligand expression in an autocrine manner. Am. J. Respir. Cell Mol. Biol. 32, 373380. Coffey, R. J., Washington, M. K., Corless, C. L. and Heinrich, M. C. (2007). Menetrier disease and gastrointestinal stromal tumors: hyperproliferative disorders of the stomach. J. Clin. Invest. 117, 70-80. Condeelis, J., Singer, R. H. and Segall, J. E. (2005). The great escape: when cancer cells hijack the genes for chemotaxis and motility. Annu. Rev. Cell Dev. Biol. 21, 695718. Conti, M., Hsieh, M., Park, J.-Y. and Su, Y.-Q. (2006). Role of the epidermal growth factor network in ovarian follicles. Mol. Endocrinol. 20, 715-723. DeWitt, A. E., Dong, J. Y., Wiley, H. S. and Lauffenburger, D. A. (2001). Quantitative analysis of the EGF receptor autocrine system reveals cryptic regulation of cell response by ligand capture. J. Cell Sci. 114, 2301-2313. DeWitt, A., Iida, T., Lam, H. Y., Hill, V., Wiley, H. S. and Lauffenburger, D. A. (2002). Affinity regulates spatial range of EGF receptor autocrine ligand binding. Dev. Biol. 250, 305-316. Dong, J., Opresko, L. K., Dempsey, P. J., Lauffenburger, D. A., Coffey, R. J. and Wiley, H. S. (1999). Metalloprotease-mediated ligand release regulates autocrine signaling through the epidermal growth factor receptor. Proc. Natl. Acad. Sci. USA 96, 6235-6240. Dong, J., Opresko, L. K., Chrisler, W., Orr, G., Quesenberry, R. D., Lauffenburger, D. A. and Wiley, H. S. (2005). The membrane-anchoring domain of epidermal growth factor receptor ligands dictates their ability to operate in juxtacrine mode. Mol. Biol. Cell 16, 2984-2998. Eguchi, S., Frank, G. D., Mifune, M. and Inagami, T. (2003). Metalloproteasedependent ErbB ligand shedding in mediating EGFR transactivation and vascular remodeling. Biochem. Soc. Trans. 31, 1198-1202. El Hader, C., Tremblay, S., Solban, N., Gingras, D., Beliveau, R., Orlov, S. N., Hamet, P. and Tremblay, J. (2005). HCaRG increases renal cell migration by a TGF␣ autocrine loop mechanism. Am. J. Physiol. Renal Physiol. 289, F1273-F1280. Fan, H. and Derynck, R. (1999). Ectodomain shedding of TGF-alpha and other transmembrane proteins is induced by receptor tyrosine kinase activation and MAP kinase signaling cascades. EMBO J. 18, 6962-6972. Fischer, O. M., Hart, S., Gschwind, A. and Ullrich, A. (2003). EGFR signal transactivation in cancer cells. Biochem. Soc. Trans. 31, 1203-1208. Forsten, K. E. and Lauffenburger, D. A. (1992). Interrupting autocrine ligand-receptor binding: comparison between receptor blockers and ligand decoys. Biophys. J. 63, 857861. Garton, K. J., Ferri, N. and Raines, E. W. (2002). Efficient expression of exogenous genes in primary vascular cells using IRES-based retroviral vectors. Biotechniques 32, 830, 832, 834 passim.


Autocrine EGFR-mediated cell migration Gill, G. N., Kawamoto, T., Cochet, C., Le, A., Sato, J. D., Masui, H., McLeod, C. and Mendelsohn, J. (1984). Monoclonal anti-epidermal growth factor receptor antibodies which are inhibitors of epidermal growth factor binding and antagonists of epidermal growth factor binding and antagonists of epidermal growth factor-stimulated tyrosine protein kinase activity. J. Biol. Chem. 259, 7755-7760. Glading, A., Uberall, F., Keyse, S. M., Lauffenburger, D. A. and Wells, A. (2001). Membrane proximal ERK signaling is required for M-calpain activation downstream of epidermal growth factor receptor signaling. J. Biol. Chem. 276, 23341-23348. Gu, L., Li, B., Yang, X., Hu, X., Huang, X., Hertz, L. and Peng, L. (2007). Depolarization-induced, glutamate receptor-mediated, and transactivation-dependent extracellular-signal regulated kinase phosphorylation in cultured cerebellar granule neurons. Neuroscience 147, 342-353. Hackel, P. O., Gishizky, M. and Ullrich, A. (2001). Mig-6 is a negative regulator of the epidermal growth factor receptor signal. Biol. Chem. 382, 1649-1662. Hardy, S., Kitamura, M., Harris-Stansil, T., Dai, Y. and Phipps, M. L. (1997). Construction of adenovirus vectors through cre-lox recombination. J. Virol. 71, 18421849. Harris, R. C., Chung, E. and Coffey, R. J. (2003). EGF receptor ligands. Exp. Cell Res. 284, 2-13. Hendriks, B. S., Opresko, L. K., Wiley, H. S. and Lauffenburger, D. (2003). Coregulation of epidermal growth factor receptor/human epidermal growth factor receptor 2 (HER2) levels and locations: quantitative analysis of HER2 overexpression effects. Cancer Res. 63, 1130-1137. Hsia, D. A., Mitra, S. K., Hauck, C. R., Streblow, D. N., Nelson, J. A., Ilic, D., Huang, S., Li, E., Nemerow, G. R., Leng, J. et al. (2003). Differential regulation of cell motility and invasion by FAK. J. Cell Biol. 160, 753-767. Huang, C., Jacobson, K. and Schaller, M. D. (2004). MAP kinases and cell migration. J. Cell Sci. 117, 4619-4628. Janes, K. A., Albeck, J. G., Peng, L. X., Sorger, P. K., Lauffenburger, D. A. and Yaffe, M. B. (2003). A high-throughput quantitative multiplex kinase assay for monitoring information flow in signaling networks: application to sepsis-apoptosis. Mol. Cell. Proteomics 2, 463-473. Janes, K. A., Gaudet, S., Albeck, J. G., Nielsen, U. B., Lauffenburger, D. A. and Sorger, P. K. (2006). The response of human epithelial cells to TNF involves an inducible autocrine cascade. Cell 124, 1225-1239. Johnston, J. B., Navaratnam, S., Pitz, M. W., Maniate, J. M., Wiechec, E., Baust, H., Gingerich, J., Skliris, G. P., Murphy, L. C. and Los, M. (2006). Targeting the EGFR pathway for cancer therapy. Curr. Med. Chem. 13, 3483-3492. Kenny, P. A. and Bissell, M. J. (2007). Targeting TACE-dependent EGFR ligand shedding in breast cancer. J. Clin. Invest. 117, 337-345. Kholodenko, B. N., Demin, O. V., Moehren, G. and Hoek, J. B. (1999). Quantification of short term signaling by the epidermal growth factor receptor. J. Biol. Chem. 274, 30169-30181. Knauer, D. J., Wiley, H. S. and Cunningham, D. D. (1984). Relationship between epidermal growth factor receptor occupancy and mitogenic response. Quantitative analysis using a steady state model system. J. Biol. Chem. 259, 5623-5631. Kumar, N., Zaman, M. H., Kim, H. D. and Lauffenburger, D. A. (2006). A highthroughput migration assay reveals HER2-mediated cell migration arising from increased directional persistence. Biophys. J. 91, L32-L34. Kuwada, S. K., Lund, K. A., Li, X. F., Cliften, P., Amsler, K., Opresko, L. K. and Wiley, H. S. (1998). Differential signaling and regulation of apical vs. basolateral EGFR in polarized epithelial cells. Am. J. Physiol. 275, C1419-C1428. Lauffenburger, D. A., Oehrtman, G. T., Walker, L. and Wiley, H. S. (1998). Real-time quantitative measurement of autocrine ligand binding indicates that autocrine loops are spatially localized. Proc. Natl. Acad. Sci. USA 95, 15368-15373. Maheshwari, G., Wiley, H. S. and Lauffenburger, D. A. (2001). Autocrine epidermal growth factor signaling stimulates directionally persistent mammary epithelial cell migration. J. Cell Biol. 155, 1123-1128. Maly, I. V., Wiley, H. S. and Lauffenburger, D. A. (2004). Self-organization of polarized cell signaling via autocrine circuits: computational model analysis. Biophys. J. 86, 1022. Markowitz, S. D., Molkentin, K., Gerbic, C., Jackson, J., Stellato, T. and Willson, J. K. (1990). Growth stimulation by coexpression of transforming growth factor-alpha and epidermal growth factor-receptor in normal and adenomatous human colon epithelium. J. Clin. Invest. 86, 356-362. Normanno, N., De Luca, A., Bianco, C., Strizzi, L., Mancino, M., Maiello, M. R., Carotenuto, A., De Feo, G., Caponigro, F. and Salomon, D. S. (2006). Epidermal growth factor receptor (EGFR) signaling in cancer. Gene 366, 2-16. Ohtsu, H., Dempsey, P. J. and Eguchi, S. (2006). ADAMs as mediators of EGF receptor transactivation by G protein-coupled receptors. Am. J. Physiol. Cell Physiol. 291, C1C10.

3699

Pandiella, A. and Massague, J. (1991). Multiple signals activate cleavage of the membrane transforming growth factor-alpha precursor. J. Biol. Chem. 266, 57695773. Resat, H., Ewald, J. A., Dixon, D. A. and Wiley, H. S. (2003). An integrated model of epidermal growth factor receptor trafficking and signal transduction. Biophys. J. 85, 730-743. Rosen, J. M. (2003). Hormone receptor patterning plays a critical role in normal lobuloalveolar development and breast cancer progression. Breast Dis. 18, 3-9. Ross, M. A., Sander, C. M., Kleeb, T. B., Watkins, S. C. and Stolz, D. B. (2001). Spatiotemporal expression of angiogenesis growth factor receptors during the revascularization of regenerating rat liver. Hepatology 34, 1135-1148. Sahin, U., Weskamp, G., Kelly, K., Zhou, H. M., Higashiyama, S., Peschon, J., Hartmann, D., Saftig, P. and Blobel, C. P. (2004). Distinct roles for ADAM10 and ADAM17 in ectodomain shedding of six EGFR ligands. J. Cell Biol. 164, 769779. Semino, C. E., Kamm, R. D. and Lauffenburger, D. A. (2006). Autocrine EGF receptor activation mediates endothelial cell migration and vascular morphogenesis induced by VEGF under interstitial flow. Exp. Cell Res. 312, 289-298. Seton-Rogers, S. E., Lu, Y., Hines, L. M., Koundinya, M., LaBaer, J., Muthuswamy, S. K. and Brugge, J. S. (2004). Cooperation of the ErbB2 receptor and transforming growth factor beta in induction of migration and invasion in mammary epithelial cells. Proc. Natl. Acad. Sci. USA 101, 1257-1262. Shankaran, H., Resat, H. and Wiley, H. S. (2007). Cell surface receptors for signal transduction and ligand transport: a design principles study. PLoS Comput. Biol. 3, e101. Shvartsman, S. Y., Wiley, H. S., Deen, W. M. and Lauffenburger, D. A. (2001). Spatial range of autocrine signaling: modeling and computational analysis. Biophys. J. 81, 1854-1867. Singh, A. B. and Harris, R. C. (2005). Autocrine, paracrine and juxtacrine signaling by EGFR ligands. Cell. Signal. 17, 1183-1193. Sporn, M. B. and Todaro, G. J. (1980). Autocrine secretion and malignant transformation of cells. N. Engl. J. Med. 303, 878-880. Stampfer, M. R., Pan, C. H., Hosoda, J., Bartholomew, J., Mendelsohn, J. and Yaswen, P. (1993). Blockage of EGF receptor signal transduction causes reversible arrest of normal and immortal human mammary epithelial cells with synchronous reentry into the cell cycle. Exp. Cell Res. 208, 175-188. Streblow, D. N., Soderberg-Naucler, C., Vieira, J., Smith, P., Wakabayashi, E., Ruchti, F., Mattison, K., Altschuler, Y. and Nelson, J. A. (1999). The human cytomegalovirus chemokine receptor US28 mediates vascular smooth muscle cell migration. Cell 99, 511-520. Tsao, M. S., Zhu, H. and Viallet, J. (1996). Autocrine growth loop of the epidermal growth factor receptor in normal and immortalized human bronchial epithelial cells. Exp. Cell Res. 223, 268-273. Wells, A. (1999). EGF receptor. Int. J. Biochem. Cell Biol. 31, 637-643. Wells, A. (2000). Tumor invasion: role of growth factor-induced cell motility. Adv. Cancer Res. 78, 31-101. Wells, A., Welsh, J. B., Lazar, C. S., Wiley, H. S., Gill, G. N. and Rosenfeld, M. G. (1990). Ligand-induced transformation by a noninternalizing epidermal growth factor receptor. Science 247, 962-964. Wiley, H. S., Woolf, M. F., Opresko, L. K., Burke, P. M., Will, B., Morgan, J. R. and Lauffenburger, D. A. (1998). Removal of the membrane-anchoring domain of epidermal growth factor leads to intracrine signaling and disruption of mammary epithelial cell organization. J. Cell Biol. 143, 1317-1328. Wiley, H. S., Shvartsman, S. Y. and Lauffenburger, D. A. (2003). Computational modeling of the EGF-receptor system: a paradigm for systems biology. Trends Cell Biol. 13, 43-50. Wilsbacher, J. L., Moores, S. L. and Brugge, J. S. (2006). An active form of Vav1 induces migration of mammary epithelial cells by stimulating secretion of an epidermal growth factor receptor ligand. Cell Commun. Signal. 4, 5. Winkler, M. E., O’Connor, L., Winget, M. and Fendly, B. (1989). Epidermal growth factor and transforming growth factor alpha bind differently to the epidermal growth factor receptor. Biochemistry 28, 6373-6378. Wolf-Yadlin, A., Kumar, N., Zhang, Y., Hautaniemi, S., Zaman, M., Kim, H. D., Grantcharova, V., Lauffenburger, D. A. and White, F. M. (2006). Effects of HER2 overexpression on cell signaling networks governing proliferation and migration. Mol. Syst. Biol. 2, 54. Xu, K. P., Yin, J. and Yu, F. S. (2007). Lysophosphatidic acid promoting corneal epithelial wound healing by transactivation of epidermal growth factor receptor. Invest. Ophthalmol. Vis. Sci. 48, 636-643. Yarden, Y. and Sliwkowski, M. X. (2001). Untangling the ErbB signalling network. Nat. Rev. Mol. Cell Biol. 2, 127-137.