Laminin activates the p185HER2 oncoprotein and mediates growth inhibition of breast carcinoma cells.

The interaction between laminin and the oncoprotein encoded by the c-erbB-2 oncogene was studied in vitro and in vivo in human breast carcinomas. In vitro analysis of breast carcinoma cell lines overexpressing p185HER2 revealed that laminin, but not fibronectin, induced tyrosine phosphorylation and down-modulation of oncoprotein membrane expression. Laminin also specifically inhibited growth of p185HER2-positive cell lines. No direct binding between the recombinant extracellular domain of p185HER2 and laminin was found. Induction of oncoprotein down-modulation by anti-integrin antibodies and coprecipitation of the oncoprotein with the beta 4 integrin subunit indicate that the interaction between p185HER2 and laminin occurs through integrin molecules. The relevance of this in vitro observation was verified in vivo by analysing the prognostic value of p185HER2 overexpression as a function of laminin production on archival paraffin-embedded sections of 887 primary breast tumours. The results revealed an association between p185HER2 overexpression and unfavourable prognosis in tumours negative for laminin production, whereas in laminin-producing tumours, the oncoprotein overexpression was not associated with tumour aggressiveness.

We have recently shown (Campiglio et al., 1994) that a6f4 integrin capping, induced by anti-integrin MAbs, also gives rise to p185HER2 clustering and tyrosine phosphorylation of this receptor. Laminin, the ligand of the a6,B4 receptor (Lee et al., 1992), increases the basal level of pl85HER2 phosphorylation (Campiglio et al., 1994), suggesting a role for adhesion molecules in the activation of the oncoprotein.
In the present study, we investigated the effects of soluble laminin on pl85HER2 activation and expression in breast carcinoma cell lines overexpressing the oncoprotein. To verify the relevance of the in vitro observations on in vivo tumour aggressiveness, we also evaluated the prognostic value of c-erbB-2 overexpression relative to laminin production in a series of 887 primary breast carcinomas.
For the proliferation assay, cells were seeded at a density of 250 x 103 per well in triplicate 6-well plates in the presence or absence of exogenously added adhesion molecules (50 pg ml -'). After 4 days, cells were trypsinised and counted under the light microscope. Murine laminin, purified from mouse Engelbreth Holm Swarm tumour cells, and human fibronectin, purified from plasma, were obtained from Sigma and Boehringer Mannheim (Germany) respectively.
Flow cytometric analysis Antigens expressed by laminin-treated cells were quantitated by indirect immunofluorescence with fluorescein isothiocyanate-conjugated goat anti-mouse Ig (Kirkegaard and Perry Laboratories Inc., Gaithersburg, MD, USA) and the following purified MAbs: MGR2, directed against the extracellular domain of p185HER2 ; MGR1, directed against the extracellular domain of the EGF receptor ; and W6/32, directed against a monomorphic determinant of human HLA-A,B,C molecules (Parham et al., 1979). Fluorescence intensity was measured using a FACScan flow cytometer with LISYS II software (Becton Dickinson, Mountain View, CA, USA).

Extracellular domain (ECD) treatment
The human c-erbB-2 receptor extracellular domain (ECD) was produced using the baculovirus expression system (Luckow et al., 1989). A 2 kb cDNA encoding amino acids 1 to 622 of the erbB-2 protein (Akiyama et al., 1986) was cloned into the expression transfer vector pVL1393 using SmaI and XbaI cloning sites. Stop codons were inserted at the 3' end of the ECD coding sequence. The protein was expressed in Sf9 cells and purified from the culture medium 4 days after infection with the recombinant virus. Recombinant protein (20 pg) purified by immunoaffinity chromatography was labelled with 1251I by the Bolton-Hunter reagent (Amersham, Little Chalfont, UK). The functionality and specificity of the labelled molecule were assessed in a binding assay using polystyrene beads coated with MAbs MGR2 and MGR4 against the oncoprotein ECD (Centis et al., 1992;Tagliabue et al., 1991), or with an unrelated MAb (Martignone et al., 1992). Binding of labelled ECD was assayed in wells of 96-well plates adsorbed with laminin, fibronectin, bovine serum albumin (BSA) or MAb MGR2 at a concentration of 10 pg per well.
Co-immunoprecipitation MDA MB453 breast carcinoma cells were treated with laminin (50 pg ml -') for 5 min and lysed in the absence of ionic detergent. Lysate (2 mg) was immunoprecipitated with the following MAbs: MAR4 directed against the P1 integrin subunit ; and 3E1 directed against the #4 integrin subunit (Telios, San Diego, CA, USA).
Immunoprecipitates were subjected to immunoblot analysis with MAb c-neu Ab3 (Oncogene Science), and the proteins were visualised with the ECL detection system (Amersham). Filters were stripped in a buffer containing 62.5 mM Tris-HCl (pH 6.8), 2% sodium dodecyl sulphate (SDS) and 100 mM pmercaptoethanol for 30 min at 65°C and reprobed with the indicated antibodies.
Surgical specimens Paraffin-embedded tissues obtained surgically from 887 patients with breast cancer and collected in our Institute from January 1968 to December 1971 were examined. In patients with histologically positive axillary lymph nodes, surgery was combined with subsequent radiotherapy on supraclavicular and internal mammary lymph nodes. No systemic treatment was administered until the time of relapse.

Statistical analysis
Overall survival of patients from the date of surgery was evaluated, considering only deaths from breast cancer as events. Survival rates were evaluated using the actuarial lifetable approach. The log-rank method was used to analyse the differences in survival curves.

Results
In vitro analysis of pJ85HE" and laminin interaction Three cell lines, two overexpressing c-erbB-2 (SKBr3 and MDA MB453) and one not expressing the oncogene (MCF-7), were cultured in the presence of laminin or fibronectin for 4 days and tested for expression and activation of the c-erbB-2 oncoprotein. Laminin induced a down-modulation of the membrane expression of the oncoprotein as compared with untreated or fibronectin-treated cells on the two cell lines overexpressing p185HER2, whereas no change was found for MCF-7 cells, which do not express this oncoprotein (  M -L nI 1). This down-modulation induced by laminin was reproducible and significant, whereas no effect on EGFR and class I HLA expression was observed (Table I). Time course analysis of pl85HER2 down-modulation revealed decreased oncoprotein expression on the cell surface of p185HER2_ overexpressing cell lines after 1 h of laminin treatment, with further down-modulation until 4 days, when decreased p185HER2 expression plateaued (Figure 2).
To determine whether the rapid membrane downmodulation of oncoprotein was associated with activation of p185HER2, the state of receptor tyrosine phosphorylation was determined in the same cell lines treated or not with soluble laminin. As detected by immunoblot analysis of anti-p185HER2 immunoprecipitates with anti-P-Tyr antibodies, tyrosine phosphorylation of oncoprotein was increased in SKBr3 cells after treatment with soluble laminin (Figure 3). A smaller, but still significant, increase was observed in MDA MB453 cells after the same treatment (data not shown).
Proliferation assay of different breast carcinoma cells grown in the presence of 50 ug ml-' laminin or fibronectin for 4 days indicated that laminin, but not fibronectin, induces a 40% inhibition of cell growth in both cell lines overexpressing p185HER2 (SKBr3 and MDA MB453), whereas no variation was found for MCF-7 cells (Figure 4).
Analysis to determine whether the two molecules interact directly indicated no binding of radiolabelled p185HER2 ECD Laminin and p185HER2 E Tagliabue et al 1429 to laminin or to fibronectin and BSA used as negative controls, whereas the same labelled molecule was recognised by a MAb directed against the p185HER2 ECD ( Figure 5).
Based on the observed colocalisation of pl85HER2 and the a6,B4 integrin receptor in a lung carcinoma cell line (Campiglio et al., 1994), the possibility of an indirect interaction mediated by this integrin receptor was investigated. The down-modulation of oncoprotein membrane expression was analysed in SKBr3 and MDA MB453 cells after treatment with MAb GoH3 (Immunotech, Inc.,   Westbrook, ME, USA), which is directed against the laminin binding site of the a6 integrin subunit. Laminin and the anti-o6 MAb each induced a significant (P <0.05) decrease in pl85HER2 membrane expression, and no further downmodulation was observed when cells previously treated with GoH3 were seeded in the presence of laminin (Table II). To investigate the mechanism of interaction between laminin and p185HER2 further, fresh lysates, obtained from MDA MB453 cells treated with laminin or untreated, were subjected to immunoprecipitation with MAbs to #1 and /4 integrin subunits and immunoblotted with an anti-pl85HER2 MAb. As shown in Figure 6, the oncoprotein was recovered from ,B4 immunoprecipitates and the amount of coprecipitated p1 85HER2 was slightly increased in cells treated with laminin. No   the cases respectively. All of the 197 cases that overexpressed p185HER2 showed a characteristic staining at the cell membrane level (Figure 7a). Tumours were considered laminin positive (244 cases) when they displayed staining at the membrane and/or cytoplasmic level (Figure 7b). Laminin production by itself was not found to be associated with any other prognostic factors, such as nodal status, tumour size and grading (Pellegrini et al., 1995). Survival curves showed a strong correlation between poor prognosis and p185HER2 overexpression (P<0.01), whereas laminin production per se had no impact on survival. However, when these two parameters were analysed together, the prognostic value of the oncoprotein was significantly influenced by laminin production (Figure 8). For patients with laminin-negative tumours (643 cases), p185HER2 overexpression was significantly associated with poor prognosis (P <0.01). By contrast, no statistically significant differences in survival rate as a function of p185HER2 overexpression were found in patients with laminin-producing tumours (244 cases).

Discussion
The present study shows that laminin, a molecule of the extracellular matrix, can functionally interact with the c-erbB-2 oncoprotein. In vitro laminin treatment of breast carcinoma cells overexpressing p185HER2 resulted in activation of the oncoprotein as measured by tyrosine phosphorylation, down-modulation of membrane expression and, ultimately, inhibition of cell proliferation. The survival data for breast cancer patients clearly indicate the relevance of this interaction in the proliferation of in vivo human tumours, since p185HER2 overexpression, which is strongly associated with poor prognosis in laminin-negative tumours, almost completely lost its prognostic significance in lamininproducing tumours. No direct binding between laminin and the c-erbB-2 oncoprotein was detected, suggesting that their interaction is mediated by other molecules. Indeed, the decrease in p185HER2 membrane expression upon treatment with an anti-a6 MAb strongly suggests that the interaction is mediated through laminin-specific integrins. Consistent with this suggestion, coprecipitation of c-erbB-2 with the ,B4 subunit demonstrates that these molecules are structurally associated on the cell membrane, and their association is increased and stabilised by laminin binding. The interaction appears to be restricted to the /4 integrin, since no oncoprotein was found associated with the /1 integrin. The interaction of laminin integrins induces receptor clustering at the plasma membrane and consequent activation of p185HER2. Up-regulation of membrane protein tyrosine phosphorylation by integrin clustering has been reported (Kornberg et al., 1991), and our recent study in which clustering of the a6#4 integrin receptor in a lung carcinoma cell line was shown to enhance tyrosine phosphorylation of overexpressed pl85HER2 (Campiglio et al., 1994) indicates an activation of this oncoprotein in the presence of laminin. Overexpression of p185HER2 at the tumour cell surface is generally thought to lead to a growth-promoting signal (Di Fiore et al., 1987), and p185HER2 uses a signal transduction pathway that is localised at the plasma membrane level (Aronheim et al., 1994;Ben-Levyet al., 1994). The laminin-induced removal of pl85HER2 from the cell surface may result in decreased growth of those tumour cells that overexpress this receptor. Consistent with the hypothesis that the oncogenic potential of p185HER2 is restricted to the membrane form, antibody-mediated internalisation of this receptor was associated with inhibition of tumour growth (Hurwitz et al., 1995). Moreover, a cytoplasmic localisation of c-erbB-2 in breast carcinomas has been linked to better prognosis compared with tumours that showed cell membrane localisation of this receptor (Zschiesche et al., 1994).
However, the possibility remains that laminin, upon interacting with integrins, becomes available for binding to other molecules of the HER family that are normally not involved in such a phenomenon owing to insufficient binding affinity. Indeed, heregulin, the ligand for c-erbB-3 and c-erbB-4 (Tzahar et al., 1994;Sliwkowski et al., 1994;Plowman et al., 1993b), represents a family of molecules sharing an EGF-like domain, which appears to be responsible for receptor recognition (Panayotou et al., 1989;Holmes et al., 1992;Wen et al., 1992). Similar EGF-like domains are also present in the short arms of laminin . Assuming that EGF-like regions in different proteins are an essential motif for proteinprotein interaction, the corresponding domain of laminin might be responsible for the interaction of this adhesion molecule with members of the HER receptor family. In fact, the interaction only with integrins does not explain the differences between SKBr3 and MDA MB453 cells in level of activation, since the latter, which are less responsive to laminin, actually express a higher amount of oc6,B4 integrin than do SKBr3 cells, but show a different pattern of HER family molecules (Plowman et al., 1993a;Kraus et al., 1993;King et al., 1988).
No morphological evidence of cell differentiation, which has been reported to occur when p185HER2 activation leads to cell growth inhibition (Peles et al., 1992;Bacus et al., 1992), was observed after laminin treatment of cells. However, SKBr3 and MDA MB453 cells have a high level of aneuploidy, which might prevent a clear response to differentiation stimuli.
Although the interaction between laminin and pl85HER2 occurs by an indirect mechanism, the in vitro and in vivo data demonstrate the importance of this interaction for tumour progression. These data may open new approaches to therapeutic intervention of breast carcinoma that exploit the ability of laminin to reduce p185HER2-related tumour aggressiveness. For this, agents that up-regulate laminin production by the tumour or laminin-derived peptides might be suitable.