Herstatin, an Autoinhibitor of the Human Epidermal Growth Factor Receptor 2 Tyrosine Kinase, Modulates Epidermal Growth Factor Signaling Pathways Resulting in Growth Arrest*

Herstatin is an autoinhibitor of the ErbB family consisting of subdomains I and II of the human epidermal growth factor receptor 2 (ErbB-2) extracellular domain and a novel C-terminal domain encoded by an intron. Herstatin binds to human epidermal growth factor receptor 2 and to the epidermal growth factor receptor (EGFR), blocking receptor oligomerization and tyrosine phosphorylation. In this study, we characterized several early steps in EGFR activation and investigated downstream signaling events induced by epidermal growth factor (EGF) and by transforming growth factor α (TGF-α) in NIH3T3 cell lines expressing EGFR with and without herstatin. Herstatin expression decreased EGF-induced EGFR tyrosine phosphorylation and delayed receptor down-regulation despite receptor occupancy by ligand with normal binding affinity. Akt stimulation by EGF and TGF-α, but not by fibroblast growth factor 2, was almost completely blocked in the presence of herstatin. Surprisingly, EGF and TGF-α induced full activation of MAPK in duration and intensity and stimulated association of the EGFR with Shc and Grb2. Although MAPK was fully stimulated, herstatin expression prevented TGF-α-induced DNA synthesis and EGF-induced proliferation. The herstatin-mediated uncoupling of MAPK from Akt activation was also observed in Chinese hamster ovary cells co-transfected with EGFR and herstatin. These findings show that herstatin expression alters EGF and TGF-α signaling profiles, culminating in inhibition of proliferation.

The ErbB family of receptor tyrosine kinases includes the prototypical EGFR, 1 human epidermal growth factor receptor (HER)-2, HER-3, and HER-4. The ErbB receptors consist of an extracellular ligand binding domain (ECD), a single transmembrane domain, and a cytoplasmic tyrosine kinase domain (1)(2)(3). Several growth factors containing EGF-like domains bind with high affinity to each ErbB receptor except HER-2, which appears to function solely as a transactivating co-receptor (4 -6). Growth factor binding induces homomer and heteromer interactions between ErbB family members, which are required for kinase activation and receptor autophosphorylation in trans (2,7,8). The tyrosine phosphorylation sites on ErbB receptors provide docking sites for signaling proteins that execute such diverse cellular responses as survival, proliferation, migration, differentiation, and apoptosis (9,10). Given this wide range of action, regulation of ErbB signaling is critical, and misregulation has been implicated in the pathogenesis of many cancers (4 -10).
The ErbB receptors transduce signals through the mitogenactivated protein kinase (MAPK) cascade and the phosphatidylinositol 3-kinase (PI3K)/Akt signaling pathway. Generally, EGF-like growth factors concomitantly activate these two pathways (11)(12)(13), whereas several EGFR inhibitors suppress both signaling cascades (12,14,15). Although the MAPK and the PI3K/Akt pathways are commonly coupled, recent evidence indicates that Akt is more important in initiating proliferation and survival signals (11,16). Targeted inactivation of Akt inhibits cell growth (17,18), inducing arrest in G 1 independent of MAPK activation by EGF (12). In addition, EGF induction of Akt activity protects against Fas-induced apoptosis by a MAPK-independent mechanism (19). Because activation of Akt protects against drug-induced death of human breast cancer cells (12,18,19), inhibitors that target the Akt pathway should be effective in enhancing tumor cell death.
ErbB splice variants that encode truncated ECDs have been suggested to modulate ErbB signaling (30) either by sequestering growth factors (31) or by altering receptor interactions (29,32). One of these, herstatin, is a secreted alternative product of the HER-2 gene containing ECD subdomains I and II followed by an intron-encoded 79-amino acid sequence (32). Herstatin has been shown to bind to EGFR and HER-2 and to block homomeric and heteromeric receptor interactions (29,32). In contrast to dominant negative mutants, herstatin does not require a membrane anchor to achieve complex formation and trans inhibition, suggesting that its novel C-terminal domain may confer high affinity binding to the receptors. Indeed, the intron-encoded domain, expressed as a recombinant peptide, binds to HER-2 and the EGFR (29,32). Although herstatin inhibits the initial steps of receptor activation, its impact on ligand binding and intracellular signaling events has not been examined. In light of the novel structure and receptor binding properties of herstatin, determination of its effects on signaling is required to understand its mechanism of action and impact on the biology of the ErbB receptors.
In this study, we demonstrate that herstatin selectively modulates signaling cascades triggered by EGF and TGF-␣. These results suggest that this naturally occurring, alternative HER-2 product provides a novel mechanism for generating signaling diversity by EGFs.

EXPERIMENTAL PROCEDURES
Cell Culture and Generation of Stable Herstatin EGFR3T3 Clones-EGFR3T3 cells were derived from NIH3T3 cells by transfection with human EGFR in mammalian expression vector pCDNA3.1 (Invitrogen). Stably transfected clones were selected in DMEM ϩ 10% FBS supplemented with 0.4 mg/ml G418. A clonal line expressing high levels of EGFR was transfected with human herstatin in pCDNA3.1/Hygro (Invitrogen) using Superfect reagent (Qiagen) as per the manufacturer's instructions. Control 3T3 cell lines were generated by transfection with herstatin alone or with the corresponding empty vector. Stable cell lines were selected with 0.1 mg/ml hygromycin B and maintained in DMEM ϩ 10% FBS containing 0.4 mg/ml G418 and 0.1 mg/ml hygromycin B. Chinese hamster ovary (CHO) cells were grown in DMEM supplemented with 10% fetal bovine serum.
Antibodies-Herstatin polyclonal antibody was generated as described previously (32) and used at a dilution of 1:10,000. All antibodies were diluted into TBST (Tris-buffered saline plus 0.005% (v/v) Tween 20). Herstatin monoclonal antibody was a generous gift from Upstate Biotechnology (Lake Placid, NY) and was used at a 1:1000 dilution. Antibodies to MAPK and Akt were obtained from Cell Signaling and used at a 1:1000 dilution. Phospho-specific polyclonal antibodies to MAPK (phosphorylated at T 202 and Y 204 ) and Akt (phosphorylated at S 473 ) were also obtained from Cell Signaling and used at a 1:1000 dilution. Rabbit polyclonal anti-EGFR antibody was obtained from Santa Cruz Biotechnology, Inc. and used at a 1:1000 dilution. Monoclonal anti-Shc antibody was obtained from Santa Cruz Biotechnology, Inc. and used at a 1:1000 dilution. Rabbit polyclonal anti-Grb2 antibody was obtained from Santa Cruz Biotechnology, Inc. and used at a 1:1000 dilution. Rabbit polyclonal anti-p185HER-2/neu antibody was characterized previously (33) and used at a 1:10,000 dilution. Anti-phosphotyrosine monoclonal antibody was obtained from Sigma and used at a 1:10,000 dilution.
Transient Transfections-Cells were grown to ϳ80% confluence in 6-well plates and then the plasmid DNAs indicated in the figure legends were introduced using LipofectAMINE reagent (Invitrogen) as per the manufacturer's instructions. Transfection efficiencies between samples were compared by co-transfection with fluorescent green protein expression plasmid (Invitrogen) and inspection by fluorescence microscopy. Proteins were analyzed at 40 h after DNA introduction.
Receptor Internalization Assays-Cells were grown to 70% confluence, serum-starved for 20 h in 0.5% FBS, washed twice in ice-cold PBS, and incubated with EGF (Upstate Biotechnology) at 100 ng/ml in cold DMEM for 2 h at 4°C. Cells were then rinsed twice with PBS, placed in pre-warmed DMEM, and returned to 37°C to allow receptor internalization. At various time points, cells were placed on ice and washed twice with ice-cold PBS, and then cell surface proteins were labeled with freshly dissolved EZ-link Sulfo-NHS-LC-Biotin (Pierce) at 0.5 mg/ml in PBS for 30 min at room temperature. To quench the biotinylation reaction, cells were placed on ice and washed twice with cold PBS containing 0.2 mg/ml bovine serum albumin and twice with PBS. EGFR from lysed cells was immunoprecipitated (see below). Samples were resolved by SDS-PAGE in a 6% polyacrylamide gel, electrotransferred to nitrocellulose membrane, and overlaid with streptavidin-horseradish peroxidase at 1 g/ml in TBST (Pierce). Biotinylated proteins were visualized by exposing blots to x-ray film (X-Omat; Eastman Kodak Co.) after treatment with Supersignal West Pico reagent (Pierce).
Immunoprecipitations-Cells were washed in PBS and then lysed on ice in MTG (50 mM Tris, pH 8.0, 100 mM NaCl, 10% (v/v) glycerol, 1 (v/v) Nonidet P-40, and 2 mM sodium orthovanadate) containing protease and phosphatase inhibitor mixtures I and II (Sigma; used as per the manufacturer's recommendations). Cell lysate was cleared by centrifugation, and protein concentrations were quantified by Bradford assay (Bio-Rad). EGFR from 150 g of cell lysate was precipitated by overnight incubation with 1 g of anti-EGFR at 4°C. Signaling complexes from 500 g of cell extract were precipitated by overnight incubation with 2 g of anti-Grb2 at 4°C. Immune complexes were bound to 25 l of protein G-Sepharose (Amersham Biosciences) by co-incubation for 40 min at 4°C, centrifuged, and washed three times with 1 ml of ice-cold MTG (EGFR) or PBS (Grb2). Immune complexes were boiled in SDS-PAGE sample buffer for 5 min and analyzed as a Western blot. Western Blot Analysis-Western blotting was conducted as described previously (29). Briefly, cells were lysed, and protein concentrations were quantified as described for immunoprecipitations. Lysates were boiled in SDS-PAGE sample buffer and loaded onto polyacrylamide gels at 30 g/lane. After electrophoresis, proteins were transferred onto nitrocellulose, stained with Ponceau S, incubated with antibody as described above, and visualized by exposure to x-ray film (X-Omat; Kodak) after treatment with SuperSignal West Pico reagent or Super-Signal West Dura reagent (Pierce). Blots were stripped with the Re-Blot Western blot recycling kit (Chemicon International, Inc.) as per the manufacturer's instructions.
[ 3 H]Thymidine Incorporation Assay-Cells were grown to 70% confluence, starved for 24 h in DMEM with 0.5% FBS, and then treated with various concentrations of TGF-␣ for 18 h at 37°C. Ligand was removed, and cells were incubated in the presence of [ 3 H]thymidine (1 Ci/ml in DMEM) for 4 h at 37°C. Cells were washed with cold PBS, incubated in 10% trichloroacetic acid at room temperature for 10 min, and washed twice with 5% trichloroacetic acid. DNA was precipitated with 100% ethanol and then solubilized by incubation in 0.2 N NaOH for 10 min at room temperature. Samples were neutralized with 0.4 N HCl and counted in a scintillation counter.

125
I-EGF Binding-The 125 I-EGF binding analysis was conducted as described previously (35). Briefly, cells were grown to 70% confluence, serum-starved for 24 h, and incubated at 4°C for 2 h with 175 pM 125 I-EGF (NEN) and different amounts of unlabeled EGF (total EGF concentrations ranged from 175 pM to 10 nM) in binding buffer (DMEM plus 50 mM Hepes, pH 7.4, and 5% (w/v) bovine serum albumin). Cells were washed and then extracted in 0.1 N NaOH plus 0.1% (w/v) SDS, and bound 125 I-EGF was quantified by gamma counting.

Herstatin Reduces EGF-stimulated Tyrosine Phosphorylation and Down-regulation of the EGFR-Previous studies have
shown that transient coexpression of herstatin with the EGFR in CHO cells results in diminished dimerization and tyrosine phosphorylation of the receptor in response to EGF (29). To further characterize the effects of herstatin on EGFR and to investigate its impact on EGF-induced intracellular signaling, stable cell lines that express different levels of herstatin were derived from NIH3T3 cells expressing human EGFR (termed EGFR3T3 cells). Efforts to stably express herstatin in several tumor cell lines that overexpress EGFR as well as HER-2 were thwarted, presumably because of an inhibitory effect on cell survival. EGFR3T3 cells were transfected with control or herstatin expression plasmid, and clonal populations were isolated by selection with hygromycin B. Varied levels of herstatin were expressed in herstatin-transfected cell lines but not in the control-transfected cells (Fig. 1A).
To investigate the effects of herstatin on ligand-induced activation of the receptor, herstatin-and control-transfected EGFR3T3 cells were serum-starved for 20 h and treated with saturating concentrations (10 nM) of EGF for 20 min, and phosphotyrosine levels of the EGFR were assessed. Although clone 1 produced the greatest amount of herstatin, each of the three herstatin-expressing clones exhibited a similar reduction in tyrosine phosphorylation of the EGFR ( Fig. 1B; see also Figs. 4B and 6A), suggesting that maximal inhibition was achieved. The EGFR expression levels were comparable between parental cells and the three different clones, showing that expression of herstatin did not down-regulate the EGFR (Fig. 1B).
Because receptor tyrosine phosphorylation is required for endocytosis (36), we hypothesized that reduced EGFR autophosphorylation levels in the presence of herstatin alter EGFmediated receptor down-regulation from the cell surface. Parental and herstatin-expressing EGFR3T3 cells were saturated with EGF at 4°C and then returned to 37°C, a temperature permissive for internalization. After various incubation times, cell surface proteins were biotinylated, and EGFR was immunoprecipitated and visualized by streptavidin-horseradish peroxidase overlay. In the parental cell line, cell surface EGFR was reduced by 80% at 10 min and was undetectable by 20 min after the removal of the temperature block (Fig. 1C), in agreement with other studies (37,38). In herstatin-expressing cells, however, cell surface EGFR was reduced only about 35% at 10 min, and about one-third of the EGFR remained at the cell surface after 20 min of incubation at 37°C. Delayed downregulation is consistent with previous findings showing that EGFR degradation, a process contingent upon endosomal location of the receptor, is blocked in CHO cells that transiently express herstatin (29).
Herstatin Inhibits EGF-stimulated Akt Phosphorylation, but MAPK Is Fully Activated-Signal transduction through the ErbB family includes both the MAPK and the PI3K/Akt signaling pathways, which are generally stimulated simultaneously by growth factors (11)(12)(13). To examine the role of herstatin in EGF signaling downstream from the receptor, we treated EGFR3T3 parental and herstatin-expressing cells with saturating amounts of EGF (16 nM) and observed the kinetics of MAPK and Akt phosphorylation over a 1-h time course. In the parental cell line, the highest level of phospho-Akt was detected at 20 min after EGF addition ( Fig. 2A). In the herstatinexpressing cells, however, little phospho-Akt was observed in EGF-treated cells, with maximal levels reaching only 2% of the parental controls. Phosphorylation of Akt was almost completely abolished in the presence of herstatin; interestingly, there was no reduction of phospho-MAPK levels. In parental and herstatin-expressing cells, the time course and extent of MAPK activation were identical; maximal activation was achieved by 5 min and declined at 20 min after the addition of EGF ( Fig. 2A). To determine whether induction of MAPK in herstatin-expressing cells was due to ectopic overexpression of EGFR (39 -41), we investigated EGF signaling through endogenous receptors in 3T3 cells that express herstatin. As in the EGFR3T3 cells, we observed preferential EGF activation of MAPK and abrogation of Akt phosphorylation in the presence of herstatin (Fig. 2B). These data demonstrate that the signaling profile caused by herstatin expression is not affected by ectopic overexpression of EGFR.
EGF Induces the Formation of a Signaling Complex Containing EGFR, phospho-Shc, and Grb2 in Herstatin-expressing Cells-After EGF treatment, the adaptor protein Shc binds the autophosphorylation domain of EGFR, is tyrosine-phosphorylated, and recruits the Grb2-Sos complex from the cytoplasm to the plasma membrane (42). To investigate whether EGF stimulation of MAPK in the presence of herstatin was induced by EGFR through Shc and Grb2, we immunoprecipitated Grb2 and examined the immune complex by Western blotting. In both parental and herstatin-expressing cells, EGF-dependent association of EGFR and Shc with Grb2 was detected (Fig. 3). Furthermore, Shc was tyrosine-phosphorylated to a similar extent (Fig. 3), providing evidence for an active signaling complex. Because HER-2 is the preferred heterodimer partner of EGFR (4 -6), endogenous HER-2 may be present in the EGFR-Shc-Grb2 signaling complex, contributing to EGF stimulation of the MAPK signaling cascade. However, HER-2 could not be detected in the signaling complex immunoprecipitated from 500 g of either parental or herstatin-expressing cells (data not shown), even though the high titer antibody used (33) detects p185HER-2 in 20 g of 3T3 cell extract. These data suggest that Shc and Grb2 transduce the herstatin-mediated EGF signal from EGFR to components of the MAPK cascade, with no evidence of HER-2 involvement.
Characterization of the Effects of EGF Concentration on Intracellular Signaling in the Presence and Absence of Herstatin-Previous studies have shown that at very low EGF concentrations, MAPK activation occurs in the absence of Akt activation and EGFR tyrosine phosphorylation (43). We therefore examined whether signaling in EGFR3T3 cells exhibits a similar sensitivity to EGF concentration. Maximal stimulation of MAPK occurred independently of Akt activation in EGFR3T3 cells treated with 0.01 nM EGF (Fig. 4A), suggesting that ectopic overexpression of EGFR did not eliminate sensitivity to very low concentrations of EGF. At 0.1 nM EGF, a subsaturating concentration, maximal stimulation of Akt was observed in parental cells, but stimulation of Akt was inhibited in the herstatin-expressing cells. Therefore, preferential inhibition of the Akt pathway by herstatin could reflect a quantitative reduction in effective EGF concentration by competitive nM EGF for 20 min at 37°C. Cell extracts were resolved by 6% SDS-PAGE and analyzed as a Western blot using anti-phosphotyrosine antibody. The blot was stripped as described under "Experimental Procedures" and reprobed with anti-EGFR. Results are representative of three independent experiments. C, herstatin (clone 1)-and mocktransfected EGFR3T3 cells were serum-starved as described above, incubated with 16 nM EGF at 4°C for 2 h, and incubated at 37°C for the durations indicated. Cell surface proteins were biotinylated as described under "Experimental Procedures." EGFR was immunoprecipitated from 150 g of lysate, and immune complexes were separated by 6% SDS-PAGE, transferred to nitrocellulose, and then overlaid with streptavidin-horseradish peroxidase. Films were scanned by imaging densitometry (BioRad, model GS700) to quantitate streptavidin-horseradish peroxidase signal. Similar results were observed in two independent experiments.
inhibition. However, after treatment with either 1 or 100 nM EGF (Ͼ100 times K D ), EGFR phosphotyrosine levels were diminished ϳ10-fold, and Akt activation was reduced to the same extent (Fig. 4B). These data suggest that herstatin has a qualitative impact on intracellular signaling that is independent of EGF concentration.
Herstatin Expression Does Not Alter the Binding Affinity of EGF for EGR3T3 Cells-Previous studies demonstrated that herstatin binds to the extracellular domain of the HER-2 receptor (32) and forms a stable complex with EGFR (29). Inhibition of EGFR by herstatin could be caused by interference with EGF binding. To examine this possibility, we characterized the binding affinity of EGF to parental and herstatinexpressing clones of EGFR3T3 cells. The cells were incubated with subsaturating amounts (175 pM) of 125 I-EGF, and its dis-placement by unlabeled EGF was measured (Fig. 5). The displacement curve exhibited by the parental cells was indistinguishable from that of cell lines that expressed either high (clone 1) or low (clones 2 and 3) levels of herstatin (see Fig. 1A). Moreover, Scatchard analysis of the binding data, as described in the legend to Fig. 6, revealed the same apparent dissociation constant of ϳ500 pM in the absence and presence of different levels of herstatin. These studies show that herstatin expression does not prevent EGF binding or alter the EGF binding affinity, suggesting that herstatin is not a competitive inhibitor of EGF-mediated EGFR activation. Herstatin therefore modulates signaling of receptors that are complexed with EGF.
Herstatin Inhibits TGF-␣-induced Receptor Phosphorylation and Akt Phosphorylation, Whereas MAPK Activation Is Unaffected-TGF-␣ is an EGFR ligand that has increased mitogenic and transforming potency compared with EGF (44). Although these ligands compete for receptor binding, they exhibit subtly different binding properties to the EGFR (45) and show distinct co-receptor-dependent signal potentiation (37). Despite these differences, parental and herstatin-expressing EGFR3T3 cells treated with TGF-␣ displayed signaling profiles similar to those observed in response to EGF stimulation. Depression of EGFR tyrosine phosphorylation occurred in the herstatin-expressing cells, particularly at high (100 nM) concentrations of TGF-␣ (Fig. 6A). Moreover, phospho-Akt levels were markedly decreased, whereas MAPK activation was unaffected at both low (1 nM) and high (100 nM) concentrations of TGF-␣ (Fig. 6A).

Herstatin Expression Does Not Alter FGF-2 Stimulation of Akt-
The strong suppression of EGF-and TGF-␣-induced Akt phosphorylation in the EGFR3T3 cells that stably express herstatin could be an indirect effect of herstatin expression or chronic ErbB receptor inhibition. To examine the integrity of the PI3K/Akt pathway, we monitored phospho-Akt levels after treatment with FGF-2, a growth factor that activates a heterologous receptor tyrosine kinase. Parental and herstatin-expressing EGFR3T3 cells were serum-starved, exactly as done before EGF treatment, and then cells were incubated with saturating amounts of FGF-2 (10 nM) for 20 min. FGF-2 treatment stimulated the tyrosine phosphorylation of an 119-kDa protein, the approximate size of the FGF receptor (Fig.  6B). Furthermore, FGF-2 increased phospho-Akt levels to an equivalent extent in the presence and absence of herstatin (Fig. 6B). These data demonstrate the functional integrity of the PI3K/Akt pathway and suggest that the reduction of phospho-Akt levels caused by herstatin expression is specific to EGF-and TGF-␣induced signaling.

FIG. 2. EGF-induced signaling in parental and herstatin-expressing EGFR3T3 cells.
A, herstatin (clone 1)-and mock-transfected EGFR3T3 cells were serum-starved and incubated with 16 nM EGF at 37°C for the durations indicated. Cell lysates were separated by 7.5% and 10% SDS-PAGE and then subjected to Western blot analysis using antibodies specific for phospho-Akt (phospho-Ser 473 ) and phospho-p42/p44 MAPK (phospho-T 202 and Y 204 ). Exposed films were scanned by imaging densitometry to quantitate phospho-Akt signal. Blots were stripped and probed with Akt antibody. Results are representative of six independent experiments. B, herstatin-and mock-transfected 3T3 cells were serum-starved and incubated with 10 nM EGF at 37°C for the durations indicated. Cell lysates were separated by 7.5% and 10% SDS-PAGE and then subjected to Western blot analysis as described above.

FIG. 3. EGF induces EGFR-Shc-Grb2 complex formation in herstatin-expressing EGFR3T3 cells.
Herstatin (clone 1)-and mock-transfected EGFR3T3 cells were serum-starved and incubated with 10 nM EGF at 37°C for 20 min. Proteins complexed to anti-Grb2 were immunoprecipitated from 500 g of cell lysate, separated by 8% and 12% SDS-PAGE, and then subjected to Western blot analysis using the antibodies indicated. The anti-phosphotyrosine blot was stripped and reprobed with anti-Shc antibody. Results are representative of two independent experiments.

Herstatin Uncouples MAPK from Akt Activation in Tran-
siently Transfected CHO Cells-To evaluate whether the uncoupling of phospho-Akt from MAPK activation by herstatin was a feature confined to the 3T3 cell background, we examined the EGF signaling profile in CHO cells, which do not express endogenous EGFR (25). The cells were transiently transfected with EGFR and different levels of herstatin expression plasmids and then treated with EGF. Fig. 6C illustrates that EGF induction of MAPK phosphorylation were unaffected by expression of different amounts of herstatin. In contrast, Akt activation was inhibited in proportion to herstatin expression levels.
Herstatin Depresses Mitogenic Stimulation by EGF and TGF-␣-To test whether herstatin affected the proliferative functions of the EGFR growth factors, TGF-␣-induced mitogenesis was assessed by measuring DNA synthesis. Cells were first forced into quiescence by serum starvation for 40 h and then treated with different concentrations of TGF-␣. The ligand was removed, and cells were incubated with [ 3 H]thymidine to quantify DNA synthesis. In the presence of herstatin, we observed a striking decrease in TGF-␣-induced [ 3 H]thymidine incorporation that was not overcome at high ligand concentrations (Fig. 7A). The impact of herstatin on EGF-mediated cell proliferation was examined by quantitation of live cells by the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide assay (34). Equal numbers of EGFR3T3 control and herstatin-transfected cells were plated, serum-starved, and then treated for 72 h with vehicle or EGF. Herstatin expression resulted in a significant reduction in viable cells in the absence of EGF (p ϭ 0.007; Fig. 7B), which may reflect diminished survival under conditions of serum deprivation. Whereas EGF treatment significantly increased the control EGFR3T3 cells (p ϭ 0.006; Fig. 7B), cells expressing herstatin displayed no significant proliferation in response to EGF treatment. These results demonstrate that herstatin interrupts TGF-␣-and EGF-mediated mitogenic signal transduction, resulting in inhibition of proliferation.

DISCUSSION
Investigating the mechanisms employed by ErbB inhibitors and their impact on signaling is critical to understanding the biology of these receptors and to developing anti-receptor tyrosine kinase therapeutics. Whereas several inhibitors of the EGFR have been investigated, herstatin is distinguished by its novel structure, consisting of subdomains I and II of the HER-2 ECD and an intron encoded-C-terminal domain (32). Furthermore, herstatin is the only naturally occurring inhibitor of the EGFR in mammalian cells that exerts its action on the initial events in receptor activation: dimerization and autophosphorylation (29,32). In this study, we show that herstatin selectively modulates the intracellular signaling pathways stimulated by EGFR ligands. EGF binds to its receptor with normal affinity in the presence of herstatin, yet receptor tyrosine phosphorylation and down-regulation are suppressed. Whereas herstatin allows full ligand stimulation of the MAPK pathway and its upstream effector, Shc, Akt phosphorylation is selectively blocked, resulting in suppression of cell growth.
Herstatin is a secreted protein that binds to and inhibits the EGFR (29). Although the binding site has not been mapped, previous observations suggest that herstatin associates with the ECD of ErbB receptors (32). We therefore examined whether herstatin may interfere with binding of EGF. Our results demonstrate that neither the binding affinity nor the overall number of EGF binding sites was significantly altered in cells that expressed either low or very high levels of herstatin. Therefore, herstatin appears to inhibit EGFR that is occupied by growth factor. Two other classes of EGFR inhibitors compete with ligand binding, including the Drosophila ligand FIG. 4. Effects of EGF concentration on signaling in parental and herstatin-expressing EGFR3T3 cells. A, herstatin (clone 1)-and mock-transfected EGFR3T3 cells were serum-starved, treated with 0, 0.01, and 10 nM EGF at 37°C for 20 min, and analyzed by Western blot as described in the Fig. 2 legend. Staining with Ponceau S confirms equal loading. B, herstatin (clone 1)-and mocktransfected EGFR3T3 cells were treated with l or 100 nM EGF as described in A. Exposed films were scanned by imaging densitometry to quantitate phosphotyrosine signal. Results are representative of two independent experiments. Argos (46,47) and the EGFR monoclonal antibody C225 (48). Because herstatin inhibits ligand-occupied receptor, it may be effective even when growth factors are overproduced by tumors (49).
In this study, we demonstrate that herstatin uncouples intracellular signaling pathways triggered by EGF and TGF-␣. Whereas EGFR tyrosine phosphorylation and Akt activation are inhibited, Shc and MAPK are fully activated. This is in contrast to the intracellular signaling effects of several other inhibitors of EGFR including the quinazoline inhibitors (12), the p185neu dominant negative mutant (15), and the C225 monoclonal antibody inhibitor (14), which suppress EGF-induced phosphorylation of both MAPK and Akt. Similar to herstatin, these inhibitors reduce EGFR tyrosine phosphorylation and cause prolonged retention of the EGFR at the cell surface.
In agreement with previous studies (43), we also observed uncoupling of MAPK and Akt signaling cascades at very low concentrations of EGF (0.01 nM), suggesting that either limiting amounts of ligand or the presence of herstatin preferentially stimulated the MAPK cascade. The signaling effects of herstatin can not be explained by a reduction in the effective concentration of growth factor; the EGF binding affinity was unaffected by herstatin, and selective suppression of Akt signaling was observed at both subsaturating (0.1 nM) and very high ligand concentrations (100 nM).
Because phosphorylation of specific tyrosine residues on activated receptors is responsible for the recruitment and activation of distinct intracellular signaling molecules (51), inhibition of some, but not all, EGFR phosphorylation sites by herstatin may cause preferential recruitment of effector molecules (52). Evaluation of this possibility will require phosphopeptide mapping of the EGFR from the herstatin-expressing cells. However, even in the absence of EGFR tyrosine phosphorylation, either by kinase-impaired EGFR (39 -41) or EGFR missing its tyrosine phosphorylation sites (53), MAPK is activated, whereas Akt is not stimulated by EGF (16,54). In these cases (53,55), as well as with herstatin, EGFR appears to be involved in activation of the MAPK cascade, as shown by its presence in a signaling complex containing tyrosine phosphorylated Shc and Grb2. An endogenous receptor tyrosine kinase or a cytoplasmic kinase such as src (56) may also participate in MAPK activation by EGF. Endogenous HER-2, the preferred EGFR heteromeric partner, is unlikely to be responsible for MAPK activation because herstatin effectively disrupts HER-2 homomeric FIG. 6. Herstatin effects on signaling induced by EGF, TGF-␣, and FGF-2. A, herstatin (clone 1)-and mock-transfected EGFR3T3 cells were serum-starved and treated with l or 100 nM TGF-␣ at 37°C for 20 min and analyzed as described in Fig. 4B. Results are representative of two independent experiments. B, herstatin (clone 1)-and mock-transfected EGFR3T3 cells were serum-starved and treated with 100 nM FGF-2 at 37°C for 20 min, and cell lysate was subjected to Western blot analysis using antibodies against phosphotyrosine and phospho-Akt. The blot was stripped and incubated with Akt antibody. C, CHO cells were grown to 80% confluence and transfected with 1.5 g of EGFR and 0.5 or 1.5 g of herstatin expression plasmids. 20 h after transfection, cells were serum-starved, treated with 10 nM EGF for 20 min at 37°C, and analyzed for phospho-Akt and phospho-p42/p44 MAPK by Western blot. Equal loading was confirmed by Ponceau S staining. and heteromeric interactions (29). Although we were unable to detect HER-2 in the EGF-induced signaling complex, other endogenous tyrosine kinases may be involved in MAPK activation in herstatin-expressing cells.
A major question in signaling through receptor tyrosine kinases concerns the mechanism by which growth factors can stimulate diverse cellular responses (52,57,58). For the ErbB receptors, one level of diversity is achieved through generation of different ErbB dimer pairs that have been found to differentially stimulate intracellular signaling pathways (10,51,52,57). Alternatively, signaling by the same dimer pair may be altered in response to activation by different growth factors (8,22,52). Additionally, EGF ligands have been shown to promote diverse cellular responses, depending on the cell type (59). The results presented here point to a novel mechanism of generating diversity by which a single ligand, EGF, can achieve altered signal output within a given cell context. Our studies suggest that herstatin, a HER-2 receptor variant expressed in fetal kidney and liver cells, can selectively alter EGF signal output, resulting in growth arrest.
The Ras/MAPK pathway was previously suggested to be the major mitogenic signaling pathway initiated by the EGFR (60).
In the presence of herstatin, full stimulation of MAPK by TGF-␣ was insufficient to drive entry into S phase, as suggested by the suppression of DNA synthesis. Moreover, EGF stimulation of MAPK was insufficient to stimulate cell proliferation in the presence of herstatin. Because Akt activation was strongly inhibited, components of the PI3K signaling pathway that are required for growth factor-induced proliferation may not be activated in the presence of herstatin. This finding is in agreement with recent observations that interruption of the MAPK pathway by EGFR blockade with quinazoline inhibitors is not the cause of G 1 arrest, but rather interruption of PI3K function is required (12).
Signaling through the EGF receptor is often enhanced in human cancers through overexpression of the receptor and autocrine stimulation by ligands produced by the tumor (49). Enhanced EGFR signaling in several carcinomas is directly coupled to inappropriate phospho-Akt survival signals, rendering many cancers resistant to apoptotic signals, including those activated by radiation and chemotherapies (12,19). EGF stimulation of Akt kinase activity has been proposed as a major mechanism behind enhanced survival conferred by inappropriate EGF signaling. The activation of Akt appears to be both required and sufficient for the antiapoptotic function of EGF (19). Results presented here point to the effectiveness of herstatin in blockage of Akt activation and inhibition of proliferation stimulated by EGF. Previous studies also demonstrate the effectiveness of herstatin in blocking proliferation stimulated by HER-2 overexpression, which often occurs in the same tumor cells that have enhanced EGFR signaling. The results presented here further support the potential utility of herstatin in the development of therapeutics against cancers with ErbB receptor involvement.