Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

The epidermal growth factor receptor (EGFR, or ErbB-1) subfamily is comprised of four transmembrane receptor tyrosine kinases, ErbB-1, -2, -3, and -4. These receptors have been implicated in the pathogenesis of various human cancers (1). The significance of ErbB-2 and/or ErbB-1 overexpression has been studied extensively in breast and lung cancer. For example, in lung adenocarcinomas, ErbB-2 overexpression relative to normal alveolar lung tissue has been found to correlate with a shortened survival (2). Overexpression of ErbB-1 in conjunction with autocrine ligand production has also been associated with decreased survival (3, 4).

Despite the association of ErbB-2 or -1 overexpression with poor prognosis in lung adenocarcinoma, few studies describing ErbB family protein interactions in bronchial epithelial progenitor cells are available. In any cell type, the tumorigenic contributions of these signaling molecules is not clear. ErbB-1 alone, in the absence of ligand, does not induce tumorigenicity of mouse fibroblasts (5). Similarly, the overexpression of ErbB2 alone is insufficient to induce malignant transformation in immortalized normal human mammary (6) or lung epithelial cells (7). It has been proposed that interactions of ErbB-2 and -1 are needed to induce a transformed phenotype. For example, overexpression of both ErbB-1 and ErbB-2 is necessary to induce a tumorigenic phenotype in NR6 mouse fibroblasts (8). In addition, kinase-deficient ErbB-2 proteins display a dominant negative mutant phenotype, inhibiting both normal ErbB-1 function and cell transformation induced by overexpressed ErbB-1 (9). Studies using transgenic mice suggest that interactions of ErbB-2 and transforming growth factor (TGF)- may also play a role in induction of tumorigenicity, inasmuch as transgenic strains expressing TGF- as well as wild-type ErbB-2 develop mammary tumors at an accelerated rate (10).

Although such evidence supports a role for ErbB-1/ ErbB-2 interactions in inducing tumorigenicity, the interactions of other ErbB family members and their ligands in induction of tumorigenicity of bronchial epithelial cells has not been studied. Ligand binding to a member of the ErbB family of receptors leads to the formation of not only receptor homodimers but also receptor heterodimers (11). The heterodimerization of ErbB-1 and -2 induced by ErbB-1 ligands was first demonstrated in mouse fibroblasts (12). More recently, all possible interactions among ErbB family members have been shown to occur in overexpression systems and tumor cell lines (11). Investigation of ligand stimulation in pairwise expression systems has defined three groups of EGF family ligands: those binding ErbB-1, EGF, amphiregulin (AR), and TGF-; those binding ErbB-1 and -4, betacellulin (BTC), epiregulin, and heparin-binding (HB) EGF-like growth factor; and those binding ErbB-3 and -4, the heregulins (HRGs) or Neu differentiation factors. Endogenous expression of these ligands by epithelial cells can occur and would be expected to influence the abundance of specific homo- or heterodimers. Because signaling properties of ErbB-1 and -2 depend upon their dimerization partner (13), this array of ligands and receptors has the capacity to respond sensitively to changing conditions through modulation of ligand-stimulated dimerization.

To examine contributions of the ErbB receptor family to malignant progression of human lung epithelial cells, an expression vector for c-erbB-2 was transfected into the immortalized human lung epithelial cell line BEAS-2B. The resulting cell clones represented a spectrum of human bronchial epithelial cells (HBECs) overexpressing ErbB-2 with an endogenous expression level of the other ErbB receptors and ligands. Clonal cell lines were screened for tumorigenicity. Only one of five tested clones was tumorigenic. This clone, B2BE6 (E6), expressed the ErbB-1 ligand TGF- as well as ErbB-2 and -1 (7). Clones expressing equivalent levels of ErbB-2 and -1, but not TGF-, failed to produce tumors. In addition, tumorigenic cells, producing high levels of ErbB-2 and TGF- (E6T), were transfected with an antisense TGF- expression vector. TGF- production was decreased 93% in the resulting cells (E6TA); ErbB-1/-2 heterodimer formation was blocked; and cells were not tumorigenic (14). We have demonstrated that both E6T and E6TA cells express, in addition to TGF-, comparable levels of AR, BTC, HB EGF-like growth factor, and HRGs 2a and 3 (15), indicating the possibility of autocrine stimulation of an array of heterodimers.

In the current study, we have examined ErbB family dimer formation in the paired cell lines E6T and E6TA. We demonstrate that all four ErbB receptors are expressed in both cell lines and that heterodimers with ErbB-1 appear to be the predominant species in this bronchial epithelial system. To examine contributions of ErbB-1 and -2 within this complex system, endoplasmic reticulum (ER) trapping was employed. Mitogen-activated protein kinase (MAPK) signaling was elevated 4- to 5-fold in E6T cells relative to E6TA, and this elevation was eliminated by blocking surface expression of either ErbB-1 or ErbB-2. Inoculation of nude mice with ErbB-1- or ErbB-2-trapped E6T cell lines demonstrated a contribution of each receptor to tumorigenicity. These results emphasize the importance of both ErbB-1 and ErbB-2 in maintenance of a tumorigenic phenotype in these human lung epithelial cells, and raise the possibility that TGF--modulated signaling through ErbB-3 and -4 heterodimers with ErbB-1 may also contribute significantly to tumorigenicity.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Cell Line Derivation and Culture

The BEAS-2B cell line is a nontumorigenic immortalized HBEC line derived from the infection of normal HBECs with SV40 Adeno 12 hybrid virus (16). It was grown in serum-free LHC-8 medium (Biofluids, Rockville, MD) according to established protocols (16). The BEAS-2B E6 cell line was derived by introducing the human c-erbB-2 expression vector (pLTRERBB-2neo) into BEAS-2B cells, as previously described (7). The BEAS-2B E6T cell line (referred to as E6T) was derived from BEAS-2B E6 cells that had been passaged once in nude mice and recultured in vitro. These cells were shown to be derived from BEAS-2B by karyotypic analysis (7). They were grown in serum-free LHC-8 medium (Biofluids) containing geneticin (200 µg/ml) (GIBCO BRL, Gaithersburg, MD). E6TA cells were prepared by introducing a TGF- antisense expression vector (pLTRTGF-HYG) into E6T cells, as previously described (14). These cells were grown in serum-free LHC-8 medium containing hygromycin B (200 µg/ml) (Boehringer Mannheim, Indianapolis, IN). MDA-MB-453 breast cancer cells (kindly provided by Dr. Ruth Lupu, University of California at Berkeley, Berkeley, CA) were grown in Iscove's modified Eagle's medium (Biofluids) supplemented with 10% fetal bovine serum (FBS) (Biofluids). MDA-MB-231 breast cancer cells (American Type Culture Collection, Rockville, MD) were grown in RPMI-1640 medium (Biofluids) supplemented with 10% FBS. A431 cells (kindly provided by Dr. Kathy Elliget, University of Maryland School of Medicine, Baltimore, MD) were grown in Dulbecco's modified Eagle's medium (DMEM) (Biofluids) supplemented with 10% FBS.

For basal conditions, E6T and E6TA cell lines were grown in LHC Basal (Biofluids), which was supplemented with insulin (5 µg/ml), transferrin (5 µg/ml), and selenium (5 ng/ml) (ITS) (Sigma, St. Louis, MO).

Protein Lysate Preparation, Immunoprecipitation, and Western Blotting

Cells were grown to 80% confluency in 100-mm tissue culture dishes in LHC-8 medium. The cells were starved for 16 h in LHC basal medium supplemented with ITS (Sigma). For stimulated samples, 15 min before lysis, cells were treated at 37°C with TGF- (7 ng/ml) (UBI, Lake Placid, NY). Cells were washed three times with cold N-2-hydroxyethylpiperazine-N'-ethane sulfonic acid (Hepes)-buffered saline (20 mM Hepes in calcium-free phosphate-buffered saline [PBS] with phenol red, pH 7.5) (Biofluids), lysed in RIPA buffer (50 mM Tris [pH 7.4], 150 mM NaCl, 1% Triton X-100, 1% deoxycholic acid, sodium salt, 0.1% sodium dodecyl sulfate [SDS], 100 µg/ml phenylmethylsulfonyl fluoride, 1 µg/ml aprotinin, 1 mM dithiothreitol, and 1 mM sodium orthovanadate) for 10 min, and scraped. The extracts were centrifuged at 40,000 × g for 30 min at 4°C. Protein concentrations were measured using the Bicinchoninic method (Pierce, Rockford, IL) according to the manufacturer's instructions. For Western analysis, lysates (100 µg per sample) were resolved on 8% SDS gels. For immunoprecipitation, 1 mg of lysate was incubated with 1µg of primary antibody overnight at 4°C. Protein A/G beads (Oncogene Science, Cambridge, MA) were added to the lysate and incubated for 1 h at 4°C. Immunoprecipitates were washed four times for 5 min each with cold PBS- 0.05% Tween-20. Beads were resuspended in 2× sample buffer, heated at 95°C for 5 min, and centrifuged. Supernatants were electrophoresed on 8% polyacrylamide gels. The gels were electrophoretically transferred overnight to Immobilon-P membranes. Membranes were blocked with 10% nonfat dry milk in 150 mM NaCl, 50 mM Tris (pH 7.5), and 0.1% Tween-20 (TBST) overnight followed by a 2-h incubation with primary antibody, then washed three times with TBST and incubated for 1 h with horseradish peroxidase-conjugated secondary antibody (1:10,000 dilution in 5% nonfat dry milk/TBST) (Amersham, Arlington Heights, IL). This was followed by three 10-min washes in TBST. Blots were developed using SuperSignal chemiluminescence reagent (Pierce) and luminescence detection film (Amersham) according to manufacturer's instructions. Anti- ErbB-2 (Ab-3 [IP/Western]; Oncogene Science); -ErbB-1 (1005 [IP/Western]; Santa Cruz Laboratories, Santa Cruz, CA); -ErbB-3 (c-17 [IP], Santa Cruz Laboratories; and Ab-6 [Western], Neomarkers, Union City, CA); -ErbB-4 (Ab-1 [IP] and Ab-2 [Western], Neomarkers); -actin (Boehringer Mannheim); antiphosphotyrosine (UBI); -active MAPK (Promega, Madison, WI) directed against the dually phosphorylated Thr/glu/tyr region in the catalytic core of mammalian extracellular regulated kinase (ERK)1 and ERK2; -p85-PI-3-kinase (Transduction Laboratories, San Diego, CA); and -MAPK, recognizing a c-terminal epitope of ERK1 and, to a lesser extent, ERK2 (c-16; Santa Cruz) were purchased from commercial sources. Densitometric analyses were performed using the Image Quant program of a Molecular Dynamics laser densitometer and performing comparisons of multiple exposures of blots to maintain a linear range.

Single-Chain Antibody Transfection and Viral Infection

The pBABE-5R and pBABE-RIR vectors were constructed to express a single-chain antibody specific for the extracellular domains of ErbB-1 and ErbB-2, respectively, and contain a puromycin resistance cassette (17, 18). The single-chain antibody contains an N-terminal signal peptide that directs it to the ER and an ER retention signal at its C-terminus directing the retention of these antibodies and their complex partners to the lumen of the ER (17). Single-chain antibody experiments utilized the Psi-2 and PA317 viral packing cell lines (20) for generating infectious virus. The ecotropic viral packaging mouse fibroblast cells, Psi-2, were grown to 40 to 60% confluence in 100-mm tissue culture dishes and transfected with 8 µg of either pBABE-5R, pBABE-RIR, or empty pBABE vectors using lipofectin reagent (GIBCO BRL) according to manufacturer's instructions. After a 5-h incubation at 37°C, the lipofectin-DNA mixture was replaced with DMEM supplemented with 10% FBS. At 2d after transfection, fibroblasts were grown in medium from a Serum Free Fibroblast Medium Kit (Sigma). At 24 h later, the serum-free media (viral supernatant) from Psi-2 cells were placed on 100-mm dishes (5 ml/dish) containing PA317 amphotropic mouse fibroblasts (1 × 10⁶ cells/dish) in the presence of 8 µg/ml hexadimethrine bromide (polybrene) (Sigma). After 24 h of infection, the Psi-2 viral supernatant was replaced with DMEM supplemented with 10% fetal calf serum. PA317 cells were expanded to 80% confluency before collection of conditioned viral supernatant. E6T lung epithelial cells grown to 50% confluency in 100-mm dishes were infected with PA317-conditioned viral supernatant for 24 h. The viral supernatant was then removed and E6T cells containing the retroviral vector were selected by growth in LHC-8 containing 0.5 µg/ml puromycin (Sigma).

Fluorescence-Activated Cell Sorter Analysis

Cells were starved for 16 h in LHC basal medium supplemented with ITS as described. Cells were harvested by trypsinization, pelleted, and resuspended in Hepes-buffered saline supplemented with 0.5% bovine serum albumin. Cells were incubated with antibody to ErbB-1 (528; Santa Cruz) or ErbB-2 (antibody 5; Oncogene Science), and a fluorescein isothiocyanate-conjugated antimouse secondary antibody. Cells were analyzed for surface expression of receptor on a Becton Dickinson FACS-Star.

Elk-1 Reporter Assay

Cells were grown to 40 to 60% confluence in six-well tissue culture plates in LHC-8 media. Cells were cotransfected using Lipofectin in duplicate with a Gal-Elk-1 plasmid (0.5 µg/well) and a 5× Gal-Luc reporter plasmid (0.25 µg/ml). A pRl-Tk plasmid (0.1 µg/well) was also cotransfected as an internal renilla luciferase control (Promega). Two days after transfection, the vector-containing cells were switched to LHC Basal medium for 18 h. Cells were lysed in 1× Passive Lysis Buffer (Promega). After lysis, 20 µl of the cell lysate was assayed for both firefly luciferase and renilla luciferase activity using the Dual Luciferase Reporter Assay Kit (Promega).

Tumorigenicity Assay

Athymic nude mice were inoculated subcutaneously in a single site with each of the cell lines tested (5 × 10⁶ cells per mouse, 10 to 20 mice per cell line) and were monitored weekly for tumor formation and growth.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

ErbB Family Heterodimer Formation

We have previously described a human lung adenocarcinoma model system in which ErbB-2 overexpression generated a tumorigenic cell line (E6T) endogenously producing high levels of TGF- (7). Inhibition of TGF- expression by antisense technology generated a nontumorigenic cell line (E6TA) (14). Further, we demonstrated that, although ErbB-1 and -2 are expressed at equal levels in both cell lines, the ErbB-1/-2 heterodimer is constitutively formed only in tumorigenic E6T. However, treatment of E6TA with exogenous TGF- can induce the ErbB-1/-2 heterodimer (14). We have recently shown (15) that both E6T and E6TA express, in addition to TGF-, the EGF family ligands AR, BTC, HB-EGF, and HRG 2a and 3, showing that these cell types can produce, in addition to TGF-, endogenous ligands for ErbB-1, -3, and -4. Thus, the E6T cell line models a human lung cancer cell developed from a progenitor producing endogenous levels of receptors and ligands but perturbed by overexpression of ErbB-2. The model system of E6T and E6TA provides an opportunity to evaluate ErbB family heterodimer formation and downstream signaling correlating with tumorigenicity by comparing cells expressing high (E6T) and low (E6TA) levels of TGF-.

Lysates of E6T and E6TA cells were immunoprecipitated with -ErbB-1 and sequentially blotted with antibodies to ErbB-1, -2, and -3 (Figure 1a). In an additional experiment (Figure 1b), the immunoprecipitation was repeated but membranes were sequentially probed for ErbB-1 and -4. As observed previously (14), the constitutive ErbB-1/-2 heterodimer was detectable only in tumorigenic E6T cells. Both ErbB-1/-3 and ErbB-1/-4 complexes were easily and comparably detectable in both cell lines, on the basis of densitometric comparison using ErbB-1 as an internal control. Interestingly, in these cells that produce AR, BTC, and HB EGF-like growth factor in addition to TGF-, ErbB-1/-2 heterodimer formation was eliminated by the decreased level of TGF- in E6TA cells, suggesting that association of this ligand in particular with ErbB-1 may, through physical interaction, create a conformation favoring ErbB-2 interaction.

View larger version (42K): [in this window] [in a new window]   Figure 1.   (a and b) ErbB-1 interactions with other ErbB receptors. Lysates (250 µg) of E6T and E6TA cells grown under basal conditions were immunoprecipitated with anti-ErbB-1 antibody. Immunoprecipitates (IP) were analyzed as described by immunoblot analysis (IB) using antibodies to the indicated receptors.

In the presence of endogenous HRG, a ligand for ErbB-3 and -4, these receptors would be expected to form dimeric species with each other and with ErbB-2 (21, 22). Immune precipitation with -ErbB-3 and probing for ErbB-2 or ErbB-4 were performed in E6T and E6TA under endogenous or exogenously HRG-stimulated conditions (Figures 2A and 2B). Heterodimer formation was compared with that detected in MDA 453 cells, which express ErbB-2, -3, and -4 but not HRG (23). As expected, ErbB-2/-3 complexes were present in both basal and HRG-treated E6T and E6TA cells but only in treated, HRG-negative MDA 453 cells (Figure 2A). Densitometric evaluation of ErbB-2 levels relative to ErbB-3 revealed that HRG treatment was required for heterodimer formation in MDA 453 cells, but that addition of HRG 1 to cells expressing HRG 2a and 3 did not increase the level of ErbB-2/-3 heterodimers seen in untreated cells (E6T, basal/HRG 1 = 0.9; E6TA, basal/HRG 1 = 1.2). Similarly, ErbB-3/-4 complexes were detected at equivalent levels in E6T cells under basal and stimulated conditions (Figure 2B), whereas stimulation of the ErbB-3/-4 heterodimer in MDA 453 cells required exogenous HRG. Immunoprecipitation with antibody to ErbB-4 under basal conditions followed by blotting for ErbB-2 detected ErbB-2/-4 heterodimers only in E6T cells (Figure 2C). Despite the equivalent levels of ErbB-1/-3 and -1/-4 heterodimers in E6T and E6TA (Figure 1), levels of ErbB-2/-3, -3/-4, and -2/-4 were lower or undetectable in E6TA (Figure 2). Exposures of E6TA lanes shown in Figures 2A and 2B are substantially longer than those for E6T. These observations suggest that autocrine HRG stimulation has achieved the maximal effect of HRG in these cells and demonstrate differential formation of heterodimer species other than ErbB-1/-2.

View larger version (33K): [in this window] [in a new window]   Figure 2.   (A) ErbB-2/ErbB-3 heterodimer formation. Lysates (1 mg) of MDA 453, E6T, and E6TA cells grown under basal conditions () or stimulated with HRG 1 (50 ng/ml, 15 min) (+) were immunoprecipitated (IP) with an anti-ErbB-3 antibody as described in MATERIALS AND METHODS. Immunoprecipitates were analyzed by immunoblotting (IB) with an ErbB-2 antibody. (B) ErbB-3/-4 heterodimer formation. Lysates (1 mg) of cells grown and treated as for A were immunoprecipitated with an anti- ErbB-3 antibody and immunoblotted with anti-ErbB-4. (C) ErbB-2/-4 heterodimer formation. Lysates (1 mg) of MDA 453, E6T, and E6TA cells grown under basal conditions were immunoprecipitated (IP) with anti-ErbB-4 antibody. Immunoprecipitates were analyzed by IB with an ErbB-2 antibody.

Comparison of Constitutive Levels of ErbB-3 and ErbB-4

These results, indicating low or undetectable ErbB-3 and -4 heterodimers in E6TA compared with E6T, contrast with the results shown in Figure 1, demonstrating equivalent levels of these receptors associated with ErbB-1 in both cell lines. Therefore, a direct comparison of expression levels of ErbB-3 and ErbB-4 was made. Figure 3 demonstrates the greater expression levels of both receptors in E6T compared with E6TA, and the increased detection of both receptors in cells pulsed with TGF- (Figure 3). The increase in ErbB-3 in E6T is not evident in the overexposure required for its detection in E6TA. These results are in agreement with those of Figure 2 but remain discordant with those showing equivalent ErbB-1 heterodimers. One explanation for these results is that ErbB-3 and -4 exist primarily as heterodimers with ErbB-1 and that the epitopes recognized by antibodies utilized for immunoprecipitation of ErbB-3 and -4 are unavailable when they are part of ErbB-1 heterodimer complexes. Reasoning that a short TGF- treatment might perturb and cause transient dissociation of some ErbB-1 complexes (24), levels of ErbB-3 and -4 in E6T and E6TA were studied with and without a 15-min TGF- treatment (Figure 3). These data indicate that both ErbB-3 and ErbB-4 can be detected more efficiently after the TGF- pulse and support the possibility of transient dissociation after ligand stimulation. The increased detection could represent homodimeric or monomeric species as well as heterodimers. However, the low levels of both ErbB-3 and ErbB-4 in E6TA relative to E6T contrast with the comparable and easily detectable quantities of these species complexed to ErbB-1 in both cell types (Figure 1) and suggest a role for TGF--induced signaling in modulating steady-state levels of these receptors. Together, these data demonstrate that ErbB-3 and -4 receptors are expressed at higher levels in tumorigenic E6T cells (Figure 3). However, only the ErbB-2/-3, -2/-4, and -3/-4 interactions reflect this differential. Thus, formation of these heterodimers as well as the ErbB-1/-2 heterodimer is inhibited in nontumorigenic E6TA cells with low TGF- expression.

View larger version (28K): [in this window] [in a new window]   Figure 3.   Constitutive expression levels of ErbB-3 and -4. Lysates (1 mg) of E6T and E6TA cells were grown under basal conditions and left unstimulated () or stimulated (+) with TGF- (7 ng/ml, 15 min) as described. The lysates were then immunoprecipitated with anti-ErbB-3 or anti-ErbB-4 antibodies and analyzed by Western blot analysis with the homologous antibody.

Exogenous TGF- but Not HRG Induces ErbB-3/PI-3-Kinase Interaction

Experimental observations by Gamett and colleagues (24) suggested that members of kinase-activated heterodimer complexes could transiently dissociate and form new associations with other receptors or with downstream signaling molecules. In E6T and E6TA cells, we have shown that a pulse of exogenous TGF- allows increased detection of ErbB-3 and ErbB-4 (Figure 3) whereas exogenous HRG 1 does not stimulate additional ErbB-2/-3 or ErbB-3/-4 heterodimer formation (Figure 2). These observations might be explained by the greater representation of ErbB-1- associated species that was detected after dissociation by TGF- stimulation. To test this possibility further, TGF- and HRG 1 were compared for their ability to stimulate an ErbB-3/PI-3-kinase (K) interaction. Basal or TGF-- stimulated lysates of E6T and E6TA were immunoprecipitated with antibody to the p85 subunit of PI-3-K. ErbB-3 complex formation, undetectable under basal conditions in either cell type, was stimulated by exogenous TGF- (Figure 4a). Interestingly, addition of the growth factor did not increase tyrosine phosphorylation of p85. This result might be predicted if the interaction were secondary to ErbB-1/-3 dissociation, inasmuch as p85 could already have been phosphorylated in association with this heterodimer. In contrast, exogenous HRG 1, which stimulated ErbB-3/p85 association in MDA 453 cells, did not bring about the interaction in E6T or E6TA (Figure 4b), in agreement with the earlier demonstration of lack of responsiveness of this cell pair to exogenous HRG and suggesting that the constitutive phosphorylation of p85 is stimulated by alternative pathways.

View larger version (34K): [in this window] [in a new window]   Figure 4.   Ligand-induced ErbB-3/PI-3-K interaction. (a) E6T and E6TA cells were grown under basal conditions and then left untreated () or treated with TGF- (7 ng/ml, 15 min) (+). Cell lysates (1 mg) were immunoprecipitated with anti-p85-PI-3-K antibody. Immunoprecipitates were analyzed by immunoblot analysis (IB) using the indicated antibodies. (b) MDA 453, E6T, or E6TA cells were grown under basal conditions and were left untreated () or treated with HRG 1 (50 ng/ml, 15 min). Cell lysates (1 mg) were immunoprecipitated with an anti-ErbB-3 antibody. Immunoprecipitates were analyzed by IB with the indicated antibodies.

Activation of Downstream Signaling Pathways

The relative increase in ErbB family heterodimers in E6T versus E6TA cells suggests that signaling induced by these species may be critical for tumorigenicity. Because ErbB signaling ultimately stimulates the ras-MAPK pathway, the degree of MAPK phosphorylation in E6T and E6TA cells was evaluated. The data in Figure 5 demonstrate that MAPK levels in E6T and E6TA are equivalent and comparable to those in the ErbB-1 positive MDA 231 cells. The broadening of the bands in E6T and E6TA relative to MDA 231 indicates greater MAPK activation in these cells. However, under basal conditions, E6T cells contain a higher level of the activated species than do E6TA cells. Densitometry of activated relative to total MAPK revealed a 4-fold elevation in E6T cells. These data support the conclusion that activation of signaling pathways initiated by the ErbB species in response to TGF- correlates with tumorigenicity in E6T cells.

View larger version (34K): [in this window] [in a new window]   Figure 5.   MAPK activation in E6T and E6TA cells. Lysates (1 mg) of MDA 231, E6TA, or E6T cells grown under basal conditions were immunoprecipitated with an anti-MAPK antibody. Immunoprecipitates were analyzed by Western analysis using an antibody against activated MAPK, then stripped and reprobed with MAPK antibody as described in MATERIALS AND METHODS.

Relationship of MAPK Signaling to Functional ErbB-1 and ErbB-2

The ErbB-1/-2 species is differentially formed in E6T in response to high levels of TGF- (14). To further evaluate the relative importance of ErbB-1/-2 heterodimer signaling, ErbB-1 or ErbB-2 functionality was attenuated in E6T cells by expressing receptor-specific single-chain antibodies (19, 25). E6T cells were infected with retroviral vectors encoding antibodies binding to epitopes in the external domain of ErbB-1 (pBR1R) or ErbB-2 (pB5R), and targeted to the lumen of the ER. Such molecules inhibit transit of the receptors through the ER (17). Fluorescence- activated cell sorter (FACS) scanning was used to evaluate the cell-surface expression of ErbB-1 in the selected E6TpB vector control and E6TpB-R1R cells expressing the anti-ErbB-1 vector. The data in Figure 6a document a 32% reduction in ErbB-1 surface expression. Cells stained only with secondary antibody were indistinguishable from unstained cells. Figure 6b shows FACS analysis of E6TpB stained with anti-ErbB-2 (right peak) or secondary antibody alone (left arrow). In comparison, E6TpB-5R cells expressing the ErbB-2 antibody showed a 97% reduction in surface expression of ErbB-2 with no effect on ErbB-1. All retrovirally infected cell lines were shown to produce the same high level of TGF- as did the parental E6T cells (data not shown).

View larger version (24K): [in this window] [in a new window]   Figure 6.   FACS analysis of ErbB cell surface expression. Retrovirally infected E6T pB (vector control), E6T pB-RIR (ErbB-1-trapped), or E6T pB-5R (ErbB-2-trapped) cells were analyzed for cell surface expression of ErbB-1 (a) or both ErbB-2 and ErbB-1 (b), as indicated.

The effect of internalization of ErbB-1 or ErbB-2 on MAPK signaling was evaluated functionally by measuring activation of the transcription factor, Elk-1, which responds to MAPK as well as other activation signals. Cells were cotransfected with a Gal-Elk-1 transactivator plasmid, a Gal-luciferase reporter plasmid, and a renilla luciferase construct as an internal control. The data in Table 1 reveal that E6T cells express a level of Elk activity 5-fold that of nontumorigenic E6TA cells, in agreement with the MAPK data in Figure 5. Compared with MAPK activation in E6T, E6TpB-R1R (ErbB-1-trapped) and E6TpB-5R (ErbB-2-trapped) cells showed reductions of 37 and 94%, respectively (Table 1). These values correlate with FACS data (Figure 6) indicating 32 and 97% reductions in surface expression, respectively, suggesting that the decrease in activation was a result of receptor trapping and indicating that ErbB-1 and -2 contribute the bulk of the excess MAPK activation seen in E6T cells.

Contribution of ErbB-1 and ErbB-2 to Tumorigenicity

The loss of tumorigenicity by E6TA cells suggested that the ErbB-1/-2 heterodimer whose formation and activation is stimulated by the high TGF- level (14) contributes significantly to tumorigenicity through activation of proliferative signaling pathways in these human lung epithelial cells (Table 1). Measurement of tumorigenicity of ErbB-1- or ErbB-2-trapped cell lines (Table 2) indicated clearly that functional ErbB-1 is required for tumorigenicity because the E6TpB-R1R cells were completely nontumorigenic and, of all ErbB-1 heterodimers, only the ErbB-2 combination was detected in tumorigenic E6T and absent in nontumorigenic E6TA (14). Although MAPK and FACS data (Table 1 and Figure 6) indicated efficient trapping of ErbB-2 in cells injected into mice, these cells were able to form tumors (Table 2). These data might be explained by a selection for the minority of cells with surface ErbB-2 because ER retention of ErbB-2, like ErbB-1 (26), would be expected to be antiproliferative. To examine the stability of ErbB-2 trapping in vitro and in vivo, E6TpB-5R cells were examined for ErbB-2 surface expression at early (p3) and late (p24) passage after cell sorting in comparison to a tumor explant from an E6TpB-5R-induced tumor. The data in Figure 7 show that, although some selection against trapping occurs during passage in vitro, selection for surface expression of ErbB-2 during in vivo tumor formation is more effective. In cells initially selected by FACS sorting and reanalyzed at passage 3 (Figure 7a), only 2.3% of cells showed ErbB-2 surface expression (higher energy fluorescence). In contrast, cells after 21 passages (Figure 7b) and a tumor explant culture (Figure 7c) show 16.8 and 83.4% surface expression of ErbB-2, respectively. These data indicate a strong in vivo selection for ErbB-2 surface expression correlating with tumorigenicity, and support a requirement for functional ErbB-2 as well as ErbB-1 for tumorigenicity.

View larger version (17K): [in this window] [in a new window]   Figure 7.   FACS analysis of ErbB-2 cell surface expression. Retrovirally infected E6T pB-5R (ErbB-2-trapped) cells from (a) early (p3) or (b) late (p24) in vitro passages, or from (c) a tumor that was explanted at Week 38 after inoculation, were analyzed for ErbB-2 cell surface expression as described in MATERIALS AND METHODS. The percentage of cells expressing ErbB-2 on their surface is indicated.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

It was the goal of the present study to understand how perturbations initiated by overexpression of c-erbB-2 in human lung epithelial cells contributed to their tumorigenic conversion. Experiments in model systems and breast cancer cell lines have demonstrated the importance of coexpression of ErbB-2 to heterodimer formation, ligand binding, and enhanced downstream signaling (11, 27). We reasoned that bronchial epithelial cells would model the spectrum of endogenous ligand and receptor expression found in lung-cancer progenitor cells. We therefore transfected the T antigen-immortalized nontumorigenic human lung epithelial cell line BEAS-2B with LTR-erbB2neo to model ErbB-2 overexpression in lung adenocarcinoma. Indeed, E6T, the tumorigenic cell line produced by this transfection, formed adenocarcinoma-like tumors in nude mice (7). The relevance of this model system to lung cancer is supported by the observations that nontumorigenic HBECs endogenously produce high levels of ErbB-1 and TGF- (28). Such cells would represent progenitors for tumorigenic cells such as E6T that, in addition to ErbB-1 and TGF-, express high levels of ErbB-2 (14). The downstream effects of autocrine TGF- secretion on formation of ErbB family heterodimer species were compared in E6T cells expressing high levels of TGF- and in the antisense derivative E6TA, in which TGF- secretion is reduced 93% (14). In addition, we examined differential activation of signaling in these cell lines and evaluated ErbB-1 and -2 contributions by receptor-trapping.

In recent studies, we have shown that both E6T and E6TA express AR, BTC, HB EGF-like growth factor, and HRGs 2a and 3 (15). Elimination of ErbB-1 surface expression (E6TpB-R1R) or a 93% reduction in TGF- (E6TA) eliminated the tumorigenicity (Table 2) (14) of E6T cells. The E6TpB-5R cell line with reduced surface expression of ErbB-2 proved to be unstable, with the percentage of surface ErbB-2 expression increasing with passage (Figure 7). The strong selection for ErbB-2 surface expression in the tumors induced by E6TpB-5R (Figure 7c and Table 2) supports the importance of this receptor for tumorigenicity. These data indicate that the ErbB-1/-2 heterodimer makes a significant contribution to the tumorigenicity of E6T cells and that its elimination by antisense downregulation of TGF- or spatial blocking of interaction between ErbB-1 and ErbB-2 inhibits tumorigenicity.

Although only TGF- downregulation was engineered in the derivation of E6TA from E6T (14), the endogenous expression levels of ErbB-3 and -4 protein also appear to be lower in these cells (Figure 3) whereas the levels of ErbB-1 and ErbB-2 are comparable in these two cell lines (14). The observation of readily detectable ErbB-1/-3 and -1/-4 heterodimers under basal conditions in E6TA as well as E6T cells (Figure 1) contrasts with the low levels of constitutive ErbB-3 and -4 in these cells (Figure 3) and the undetectable levels of ErbB-2/-4 and -3/-4 heterodimers (Figure 2). Immunoprecipitation with antibody to ErbB-1 allows clear detection of ErbB-4 in the same quantity of lysate in which it was undetectable when precipitated by antibody to ErbB-4 (Figures 1 and 3). In fact, ErbB-1/-3 and -1/-4 heterodimer levels do not vary significantly between E6T and E6TA. These data indicate that ErbB-3 and -4 are expressed at lower levels in E6TA cells, possibly reflecting a diminution in signaling initiated by TGF-. Further, in E6TA cells, all of ErbB-4 and the bulk of ErbB-3 are present in ErbB-1 heterodimers. One possible explanation is that the epitope recognized for immune precipitation by the ErbB-4 or -3 antibody is unavailable in heterodimers and that there is no appreciable ErbB-3 or -4 in homodimer or monomer species. However, the detection of ErbB-2/-3 heterodimers (Figure 2) and TGF- enhancement of ErbB-4 detection (Figure 3) makes this an unlikely possibility. Alternatively, if homodimer-heterodimer equilibria are in constant flux (24) in the presence of a pulse of exogenous TGF-, ErbB-4 or -3 might be temporarily displaced from ErbB-1, rendering these proteins more available to the immunoprecipitating antibody. These alternatives could be resolved by analysis of constitutive ErbB-3 and -4 by Western blotting. Unfortunately, attempts to evaluate ErbB-3 and -4 by Western blotting failed because these receptors were undetectable in the maximum protein that we could load by this technique (up to 500 µg). The suggestion that a TGF- pulse can stimulate dissociation of coreceptor partners is further supported by the stimulation of ErbB-3/p85-PI-3-K association by exogenous TGF- and the absence of its stimulation by exogenous HRG (Figure 4). In addition, the experiment of Figure 4a indicates that PI-3-K is constitutively phosphorylated in both E6T and E6TA, suggesting that PI-3-K phosphorylation in this system may be downstream of the ErbB-1/-3 and -1/-4 heterodimers. These observations after a TGF- pulse would not, therefore, represent a steady-state measurement. Under steady-state conditions, for example, it is known that EGF stimulates ErbB-1/-3 heterodimer formation (29). Further experiments will be necessary to unravel the role of these heterodimers in downstream signaling in this system.

The apparent dominance of ErbB-1 heterodimers in these bronchial epithelial cells contrasts with that of ErbB-2 heterodimers observed in other cell types (26). Further, given the basal expression of AR, BTC, HB EGF-like growth factor, and HRGs 2a and 3 (15), the elimination of the ErbB-1/-2 heterodimer (Figure 1), reduced downstream signaling (Figure 5 and Table 1), and loss of tumorigenicity (Table 2) consequent to TGF- reduction are surprising. It has been shown that the C-terminal regions of EGF and TGF- bind to different epitopes on ErbB-1 (30). Differences in binding patterns of the EGF-like ligands may result in different patterns of homo- and heterodimer formation. Further, TGF- may have quantitatively greater effects on induction of signaling because it selectively potentiates receptor recycling (31). In this bronchial epithelial cell system, reduction of TGF- eliminated the ErbB-1/-2 heterodimer. Recent reports have shown that formation of ErbB-1/-2 heterodimers increases recycling of ErbB-1 due to reduced endocytosis, endosomal sorting, and lysosomal targeting (32, 33). Further, it has been shown that the tyrosine phosphorylation pattern and consequent downstream signaling pathways activated by ErbB-1 and -2 depend on their dimerization partners (13). Thus, if ErbB-1/-2 heterodimers are eliminated by reduction in TGF-, the decrease in downstream signaling is justified.

In summary, this study reveals that human lung epithelial cells express all four ErbB family receptors. Overexpression of ErbB-2 in this context in cells expressing high levels of TGF- generated formation of the ErbB-1/-2 heterodimer, which was eliminated by antisense inhibition of TGF- production. Differential expression of ErbB-3 and -4 may also be influenced by TGF- expression and contribute to tumorigenic conversion, although proving this hypothesis will require more extensive investigation. Higher expression levels of these receptors found in E6T cells might also activate signaling pathways that contribute to tumorigenicity. However, the correlation of MAPK signaling with surface expression of both ErbB-1 and ErbB-2, as well as the findings that tumorigenic variants require high levels of TGF- (14), require surface expression of ErbB-1 (Table 2 and Figure 6), and select for surface expression of ErbB-2 (Table 2 and Figure 7), combine to strongly support the conclusion that formation of the ErbB-1/-2 heterodimer and activation of its downstream signaling pathways are critical for tumorigenicity.