Differential Insulin-like Growth Factor I Receptor Signaling and Function in Estrogen Receptor (ER)-positive MCF-7 and ER-negative MDA-MB-231 Breast Cancer Cells¹

The IGF-IR³ is a ubiquitous, transmembrane tyrosine kinase that has been implicated in different growth-related and growth-unrelated processes critical for the development and progression of malignant tumors, such as proliferation, survival, and anchorage-independent growth, as well as cell adhesion, migration, and invasion (1, 2).

The IGF-IR is necessary for normal breast biology, but recent clinical and experimental data strongly suggest that the same receptor is involved in the development of breast cancer (1, 3). The IGF-IR is overexpressed (up to 14-fold) in ER-positive breast cancer cells compared with its levels in normal epithelial cells (1, 4, 5). The elevated expression and hyperactivation of the IGF-IR has been linked with increased radioresistance and cancer recurrence at the primary site (4). Similarly, high levels of IRS-1, a major signaling molecule of the IGF-IR, correlated with tumor size and shorter disease-free survival in ER-positive tumors (6, 7).

IGF-IR ligands, IGF-I and IGF-II, are strong mitogens for many hormone-dependent breast cancer cell lines and have been found in the epithelial and/or stromal component of breast tumors (1). Importantly, higher levels of circulating IGF-I predict increased breast cancer risk in premenopausal women (8). In vitro, activation of the IGF-IR, especially the IGF-IR/IRS-1/PI-3K pathway in ER-positive breast cancer cells, counteracts apoptosis induced by different anticancer treatments or low concentrations of hormones (1, 9, 10, 11). On the other hand, overexpression of either the IGF-IR or IRS-1 in ER-positive breast cancer cells improves responsiveness to IGF and, in consequence, results in estrogen-independent proliferation (1, 12, 13). In agreement with these observations, blockade of IGF-IR activity with various reagents targeting the IGF-IR or its signaling through IRS-1/PI-3K reduced the growth of breast cancer cells in vitro and/or in vivo (1, 12, 14, 15, 16, 17).

The requirement for the IGF-IR/IRS-1 pathway for growth and survival appears to be a characteristic of ER-positive, more differentiated, breast cancer cells. By contrast, ER-negative tumors and cell lines, often exhibiting less differentiated, mesenchymal phenotypes, express low levels of the IGF-IR and often decreased levels of IRS-1 (1, 17, 18). Notably, these cells do not respond to IGF-I with growth (1, 19, 20, 21, 22). Despite the lack of IGF-I mitogenic response, the metastatic potential of ER-negative breast cancer cells can be effectively inhibited by different compounds targeting the IGF-IR. For instance, blockade of the IGF-IR in MDA-MB-231 cells by an anti-IGF-IR antibody reduced migration in vitro and tumorigenesis in vivo, and expression of a soluble IGF-IR in MDA-MB-435 cells inhibited adhesion on the extracellular matrix and impaired metastasis in animals (14, 16, 23). These observations suggested that in ER-negative cells, some functions of the IGF-IR must be critical for metastatic cell spread. Here we addressed the possibility that in ER-negative cells, the IGF-IR selectively promotes growth-unrelated processes, such as migration and invasion, but is not engaged in the transmission of growth and survival signals. Using ER-negative MDA-MB-231 breast cancer cells, we studied IGF-I-dependent pathways involved in migration and the defects in IGF mitogenic signal. For comparison, relevant IGF-I responses were analyzed in ER-positive MCF-7 cells.

The pcDNA3-IGF-IR expression plasmid encoding the wild-type IGF-IR under the cytomegalovirus promoter was described before (13). The expression plasmids encoding constitutively active forms of Akt kinase, i.e., myristylated Akt and Akt with an activating point mutation (Akt/E40K), were obtained from Drs. P. Tsichlis and T. Chan (Kimmel Cancer Center) and were described before (24). The Akt plasmids contain the HA-tag, allowing for easy identification of Akt-transfected cells. The pCMS-EGFP expression vector encoding GFP was purchased from Clontech.

MDA-MB-231 cells were obtained from American Type Culture Collection. MDA-MB-231/IGF-IR clones were generated by stable transfection of MDA-MB-231 cells with the plasmid pcDNA3-IGF-IR using a standard calcium phosphate precipitate procedure (13). Transfectants resistant to 1 mg/ml G418 were screened for IGF-IR expression by fluorescence-assisted cell sorting analysis using an anti-IGF-IR mouse mAb α-IR3 (10 μg/ml; Calbiochem) and a fluorescein-conjugated goat antimouse IgG2 (1 μg/ml; Calbiochem). Cells stained with the secondary antibody alone were used as a control. Additionally, the parental MDA-MB-231 cells and MCF-7/IGF-IR clones 12 and 15 (13), all expressing known levels of the IGF-IR, were analyzed in parallel. IGF-IR expression in MDA-MB-213-derived clones was then confirmed by WB with specific antibodies (listed below). In growth and migration experiments, we used control MCF-7/pc2 and MDA-MB-231/5 M cell lines, which have been developed by transfection of MCF-7 and MDA-MB-231 cells with the pcDNA3 vector. MCF-7, MCF-7/pc2, and MCF-7/IGF-IR cells were described in detail previously (13).

Seventy % confluent cultures of MDA-MB-231 and MCF-7 cells were transiently cotransfected with an Akt expression plasmid and a plasmid pCMS encoding GFP (Akt:GFP ratio, 20:1) using Fugene 6 (Roche). Transfection was carried out for 6 h in phenol red-free DMEM containing 0.5 mg/ml BSA, 1 μm FeSO₄, and 2 mm l-glutamine (referred to as PRF-SFM; Ref. 13); the optimal DNA:Fugene 6 ratio was 1 μg:3 μl. Upon transfection, the cells were shifted to fresh PRF-SFM, and the expression of total and active Akt kinase at 0 (media shift), 2, and 4 days was assessed by WB with specific antibodies (see below). In parallel, the efficiency of transfection was evaluated by scoring GFP-positive cells. In all experiments, at least 40% of transfected cells expressed GFP, which indicated a high transfection efficiency. In addition, the expression of Akt plasmids was monitored by measuring the cellular levels of HA-tag and Akt proteins by WB.

MDA-MB-231 and MCF-7 cells were grown in DMEM:F12 (1:1) containing 5% CS. MDA-MB-231- and MCF-7-derived clones overexpressing the IGF-IR or expressing vector alone were maintained in DMEM:F12 plus 5% CS plus 200 μg/ml G418. In the experiments requiring 17β-estradiol- and serum-free conditions, the cells were cultured in PRF-SFM (13).

To analyze the growth in serum-containing medium, the cells were plated in six-well plates in DMEM:F12 (1:1) containing 5% CS at a concentration of 1.5–2.0 × 10⁵ cells/plate; the number of cells was then assessed by direct counting at 1, 2, and 4 days after plating. To study IGF-I-dependent proliferation, the cells were plated in six-well plates in the growth medium as above. The following day (day 0), the cells at ∼50% confluence were shifted to PRF-SFM containing 20 ng/ml IGF-I. Cell number was determined at days 1, 2, and 4.

The cells grown on coverslips in normal growth medium were shifted to PRF-SFM at 70% confluence and then cultured in the presence or absence of 20 ng/ml IGF-I for 0, 12, 24, 48, and 96 h. Apoptosis in the cultures was determined with the In Situ Cell Death Detection kit, Fluorescein (Roche), following the manufacturer’s instructions. The cells containing DNA strand breaks were stained with fluorescein-dUTP and detected by fluorescence microscopy. Cells that detached during the experiment were spun on glass slides using cytospin and processed as above. Apoptotic index (the percentage of apoptotic cells/total cell number in a sample field) was determined for adherent and floating cell populations, and the indices were combined.

Seventy % cultures were shifted to PRF-SFM for 24 h and then stimulated with 20 ng/ml IGF-I for 15 min, 1 h, 1 day, or 2 days. Proteins were obtained by lysing the cells in a buffer composed of 50 mm HEPES (pH 7.5), 150 mm 1% Triton X-100, 1.5 mm MgCl₂, 1 mm CaCl₂, 5 mm EGTA, 10% glycerin, 0.2 mm Na₃VO₄, 1% phenylmethylsulfonyl fluoride, and 1% aprotinin. The IGF-IR was immunoprecipitated from 500 μg of protein lysate with anti-IGF-IR mAb (Calbiochem) and subsequently detected by WB with anti-IGF-IR pAb (Santa Cruz Biotechnology). IRS-1 was precipitated from 500 μg of lysate with anti-IRS-1 pAb (UBI) and detected by WB using the same antibody. Tyrosine phosphorylation (PY) of immunoprecipitated IRS-1 or IGF-IR was assessed by WB with anti-phosphotyrosine mAb PY20 (Transduction Laboratories). Akt, ERK1/ERK2, and p38 MAPKs (active and total), and active GSK-3 were measured by WB in 50 μg of whole cell lysates with appropriate antibodies from New England Biolabs. The expression of HA-tag was probed by WB in 50 μg of protein lysate with anti-HA mAb (Babco). The intensity of bands representing relevant proteins was measured by laser densitometry scanning.

PI-3K activity was determined in vitro, as described by us before (25). Briefly, 70% cultures were synchronized in PRF-SFM for 24 h and then stimulated with 20 ng of IGF-I for 15 min or 2 days. Untreated cells were used as controls. IRS-1 was precipitated from 500 μg of cell lysates; IRS-1 IPs were then incubated in the presence of inositol and [³²P]ATP for 30 min at room temperature. The products of the kinase reaction were analyzed by TLC using TLC plates (Eastman Kodak). Radioactive spots representing phosphatidylinositol-3-phosphates were visualized by autoradiography, quantified by laser densitometry (ULTRO Scan XL, Pharmacia), and then excised from the plates and counted in a beta counter.

Chemotaxis and chemokinesis were tested in modified Boyden chambers containing porous (8-mm), polycarbonate membranes. The membranes were not coated with extracellular matrix. Briefly, 2 × 10⁴ cells (synchronized in PRF-SFM for 24 h) were suspended in 200 μl of PRF-SFM and plated into upper wells. Lower wells contained 500 μl of PRF-SFM. To study chemotaxis, IGF-I (20 ng/ml) was added to lower wells only; to assess chemokinesis, IGF-I was placed in either upper wells only, or in both wells. After 24 h, the cells in the upper wells were removed, whereas the cells that migrated to the lower wells were fixed and stained in Coomassie Blue solution (0.25 g of Coomassie blue:45 ml water:45 ml methanol:10 ml glacial acetic acid) for 5 min. After that, the chambers were washed three times with H₂O. The cells that migrated to the lower wells were counted under the microscope (10, 26).

LY294002 (Biomol Research Labs) was used to specifically inhibit PI-3K (27). UO126 (Calbiochem), a specific inhibitor of MEK1/2, was used to block ERK1 and ERK2 kinases (28), and SB203580 (Calbiochem) was used to down-regulate p38 MAPK (29). To determine optimal concentrations of the compounds, different doses (1–100 μm) of the inhibitors were tested in cells treated for 1, 8, 12, and 24 h in PRF-SFM. Additionally, the efficacy of all inhibitors in blocking the phosphorylation of relevant downstream targets (Akt, ERK1/ERK2, and p38 kinases) was determined by WB. In this experiment, the cells were stimulated with IGF-I (20 ng/ml) for 15 min. LY294002 and UO126 were supplemented simultaneously with IGF-I, whereas SB203580 was added 30 min before IGF-I treatment. Ultimately, for both growth and migration experiments, LY294002 was used at the concentration 50 μm, UO126 at 5 μm, and SB203580 at 10 μm. At these doses, the inhibitors did not affect cell proliferation and survival at 24 h, with the exemption of LY294002, which inhibited (by 20%) the proliferation of MCF-7/IGF-IR clone 12 in PRF-SFM. A shorter treatment (12 h) with LY294002 had no impact on the growth and survival of the cells (evaluated by cell proliferation and In Situ Cell Death Detection assays, as described above). Thus, the effects of LY294002 on migration were assessed at 12 h, whereas the actions of UO126 and SB203580 were assessed at 24 h of treatment.

To study growth-related and growth-unrelated effects of IGF-I in ER-negative cells breast cancer cells, we used the MDA-MB-231 cell line. These cells express low levels of the IGF-IR and do not respond to IGF-I with growth (19, 22). Because it has been established that mitogenic response to IGF-I requires a threshold level of the IGF-IR (e.g., in NIH 3T3-like fibroblasts, ∼1.5 × 10⁴ IGF-IRs; Refs. 30, 31), our first goal was to test whether increasing IGF-IR expression would induce IGF-I-dependent growth in MDA-MB-231 cells. To this end, several MDA-MB-231 clones overexpressing the IGF-IR (MDA-MB-231/IGF-IR cells) were generated by stable transfection, and the receptor content was analyzed by binding assay, fluorescence-assisted cell sorting analysis (data not shown), and WB (Fig. 1). We determined that MDA-MB-231 clones 2, 21, and 31 express approximately 3 × 10⁴, 1.5 × 10⁴, and 2.5 × 10⁵ IGF-IRs/cell, respectively, whereas the parental MDA-MB-231 cells express approximately 7 × 10³ IGF-IRs/cell (19). For comparison, ∼6 × 10⁴ IGF-IRs were found in ER-positive MCF-7 cells (Fig. 1; Ref. 13).

The analysis of growth profiles of different MDA-MB-231/IGF-IR clones indicated that overexpression of the IGF-IR never improved basal proliferation in normal growth medium, and in the case of clone 31, which expressed the highest IGF-IR content (∼2.5 × 10⁵ IGF-IRs/cell), an evident growth retardation at days 2 and 4 (P < 0.05) was observed (Fig. 2,A). In contrast, similar overexpression of the IGF-IR in ER-positive MCF-7 cells significantly augmented proliferation (Fig. 2,B). The growth of control clones MDA-MB-231/5 M and MCF-7/pc2 was comparable with that of the corresponding parental cell lines (Fig. 2).

Subsequent studies established that increasing the levels of the IGF-IR from 7 × 10³ up to 2.5 × 10⁵ was not sufficient to induce IGF-I-dependent growth response in MDA-MB-231 cells. In fact, similar to the parental and MDA-MB-231/5 M cells, all MDA-MB-231/IGF-IR clones were progressively dying in PRF-SFM with or without 20 ng/ml IGF-I (Fig. 3,A). In all ER-negative cell lines, the rate of cell death was significantly increased at days 2 and 4 of the experiment. Notably, at these later time points, MDA-MB-231/IGF-IR clone 31 was dying faster in the presence of IGF-I than in PRF-SFM and more rapidly than the parental cells (Fig. 3,A and data not shown). Conversely, in ER-positive cells, the stimulation of the IGF-IR always induced proliferation. In addition, at later time points, especially at day 4, the growth rate in IGF-I was significantly (P < 0.05) increased in MCF-7/IGF-IR cells relative to that in MCF-7 or MCF-7/pc2 cells (Fig. 3 B).

The analysis of the antiapoptotic effects of IGF-I in the above cell lines cultured for 48 h under PRF-SFM indicated that IGF-I reduced apoptosis, by ∼3-fold, in ER-positive cells, but it was totally ineffective in MDA-MB-231 and MDA-MB-231/IGF-IR cells (Table 1).

Next, we investigated molecular basis underlying the lack of IGF-I growth response in ER-negative cells. IGF-I signaling was studied in MDA-MB-231 cells, MDA-MB-231, clone 31, and in parallel, in ER-positive MCF-7 and MCF-7/IGF-IR cells. The experiments focused on IGF-IR tyrosine kinase activity and several postreceptor signaling pathways that are known to control the growth and survival of ER-positive breast cancer cells (and many other cell types), i.e., the IRS-1/PI-3K, Akt, and ERK1/ERK2 pathways (1, 17, 25, 32, 33, 34). We also analyzed other IGF-I effectors that have been shown to contribute to nonmitogenic responses in ER-positive breast cancer cells, such as p38 kinase and SHC (10, 26, 35).

Because both acute and chronic effects of growth factors are important for biological response (36), we studied IGF-IR signaling at different times after stimulation: 15 min, 1 h, 2 days, and 4 days. In both ER-positive and ER-negative cell types, IGF-I signaling seen at 15 min was identical to that at 1 h, whereas IGF-I response at 2 days was similar to that at 4 days. Thus, Fig. 4 demonstrates the representative results obtained with cells stimulated for 15 min and 2 days.

In MDA-MB-231 and MDA-MB-231/IGF-IR cells, IGF-IR and its major substrate, IRS-1, were tyrosine phosphorylated at both time points in a manner roughly reflecting the receptor levels. The activation of both molecules was stronger just after stimulation and weaker at 2 days of the treatment (Fig. 4,A). Analogous IGF-I effects were seen in MCF-7 cells and their IGF-IR-overexpressing derivatives (Fig. 4 B). A basal level of IGF-IR and IRS-1 tyrosine phosphorylation was observed in cells expressing high receptor levels. This effect most likely can be attributed to the autocrine stimulation of the IGF-IR by IGF-I-like factors (12).

One of the major growth/survival pathways initiated at IRS-1 is the PI-3K pathway (32, 37). The repeated measurements of IRS-1-associated PI-3K activity in vitro demonstrated that at 15 min after IGF-I addition, PI-3K activity was similar in both cell types, but at 2 days, in MDA-MB-231 and MDA-MB-231/IGF-IR cells, IGF-I did not stimulate PI-3K through IRS-1, or induced it very weakly, whereas in MCF-7 and MCF-7/IGF-IR cells, a significant level of PI-3K activation was observed (Fig. 5).

The in vitro activity of PI-3K was reflected by the stimulation of its downstream effector, Akt kinase. At 15 min, Akt was up-regulated in response to IGF-I an all cell lines, but at 2 days, no effects of IGF-I were seen in MDA-MB-231 and MDA-MB-231/IGF-IR cells, whereas up-regulation of Akt was still evident in MCF-7 and MCF-7/IGF-IR cells (Fig. 4, C and D). Akt is known to phosphorylate (on Ser-9) and down-regulate GSK-3β (23, 32, 34). We found that in both cell types, the phosphorylation of GSK-3β reflected the dynamics of Akt activity, with no induction of phosphorylation observed at 2 days in ER-negative cells (Fig. 4,C) and IGF-I-stimulated phosphorylation in MCF-7 and MCF-7/IGF-IR cells (by 40 and 120%, respectively; Fig. 4 D).

Another IGF-IR growth/survival pathway involves ERK1 and ERK2 kinases (1, 36, 38). This pathway was strongly up-regulated at 15 min and weakly induced at 2 days in MCF-7 and MCF-7/IGF-IR cells. In MDA-MB-231 and MDA-MB-231/IGF-IR cells, the basal activation of ERK1/2 kinases was always high, and the addition of IGF-I only minimally (10–20%) induced the enzymes at 15 min, with no effects seen at 2 days (Fig. 4, E and F).

p38, a stress-induced MAPK and a known mediator of nongrowth responses in breast cancer cells (35), was strongly stimulated by IGF-I in ER-negative cells at 15 min (Fig. 4,E). By contrast, in ER-positive cells, the enzyme was much stronger when induced at 2 days than at 15 min (Fig. 4 F). The stimulation of SHC, a substrate of the IGF-IR involved in migration and growth in ER-positive cells (10, 26), was weak in all cell types, and no differences in the activation patterns were observed (data not shown).

Previous results indicated that MDA-MB-231 and MDA-MB-231/IGF-IR cells are unable to sustain IGF-I-dependent activation of the PI-3K/Akt survival pathway when cultured in the absence of serum for 2–4 days. Consequently, we tested whether cell death under PRF-SFM conditions can be reversed by a forced overexpression of the Akt kinase. Two different expression plasmids encoding constitutively active forms of Akt, Myr-Akt, and Akt/E40K (24) were transiently transfected into MDA-MB-231 cells. The efficiency of transfection was at least 40% (by scoring GFP-positive cells); correspondingly, the transfected cells expressed elevated (by ∼40%) levels of the Akt protein and exhibited enhanced Akt phosphorylation (Fig. 6,A). The improved biological activity of Akt in the transfected cells was indicated by down-regulation of the prolonged ERK1/2 stimulation (39, 40), which was noticeable at day 2 (data not shown) and most pronounced at day 4 (∼50 and 40% for Myr-Akt and Akt/E40K, respectively; Fig. 6,B) The expression of constitutively active Akt mutants was reflected by a tendency of MDA-MB-231 cells to survive better at 2 days after transfection (at the time of the greatest Akt activity), but the differences did not reach statistical significance (P > 0.05; Fig. 6 C and data not shown).

To complement the above studies, we examined the importance of the PI-3K, ERK1/ERK2, and p38 kinase pathways in IGF-I-dependent growth and survival of ER-positive and ER-negative breast cancer cells using specific inhibitors (27, 28, 29). The efficacy of PI-3K and ERK1/ERK2 inhibitors was first tested by establishing their effects on the activity of target proteins (Fig. 7). Table 2 demonstrates the impact of the compounds on cell growth/survival at 2 days of treatment. The inhibition of PI-3K with LY294002 reduced the growth of MCF-7 and MCF-7/IGF-IR cells, but it did not influence or had only minimal effects on MDA-MB-231 and MDA-MB-231/IGF-IR cells. Furthermore, the action of LY294002 was counteracted by IGF-I in ER-positive, but not in ER-negative, cells. The inhibition of MEK1/2 and ERK1/ERK2 with UO126 reduced the growth and/or survival in both cell types, but only in MCF-7 and MCF-7/IGF-IR cells was IGF-I able to oppose this effect. Down-regulation of p38 kinase with SB203580 reduced the survival of MDA-MB-231 and MDA-MB-231/IGF-IR cells and to a lesser extent the growth and survival of MCF-7 and MCF-7/IGF-IR cells. IGF-I did not reverse the antimitogenic action of the p38 kinase inhibitor in either of the cell lines studied (Table 2). Cumulatively, these results suggested that in ER-positive cells, IGF-I transmits mitogenic signals through PI-3K and ERK1/ERK2 pathways. By contrast, IGF-I does not induce growth or survival signal through these pathways in ER-negative cells.

We investigated the nonmitogenic effects of IGF-I in ER-negative and ER-positive breast cancer cells. Unlike with the growth and survival responses, we found that the IGF-IR transmitted nonmitogenic signals in MDA-MB-231 and MDA-MB-231/IGF-IR cells. Specifically, in the chemotaxis experiments, IGF-I placed in lower wells stimulated migration of ER-negative cells in a manner reflecting IGF-IR content. Similarly, the same IGF-I doses induced migration in ER-positive cells (Table 3). The addition of IGF-I to the upper well or both upper and lower wells always suppressed chemotaxis of all cell lines (Table 3).

Using the inhibitors of PI-3K, ERK1/ERK2, and p38 kinases, we determined which pathways of the IGF-IR are involved in migration of ER-negative and ER-positive cells. The treatment was carried out for 24 h (UO126 and SB203580) or 12 h (LY294002) and did not affect cell growth and/or survival with or without IGF-I (see “Materials and Methods”). As demonstrated in Table 4, down-regulation of PI-3K with LY294002 inhibited basal migration of both cell types, with a more pronounced effect in ER-negative cells. Similarly, blockade of p38 kinase inhibited motility of all cell lines studied. The inhibition of MEK1/2 and ERK1/2 with UO126 never suppressed the migration of ER-positive and ER-negative cells; in fact, the compound stimulated cell motility. The addition of IGF-I as a chemoattractant significantly counteracted the effects of all three inhibitors; however, no clear association between the cellular levels of the IGF-IR and this competing action of IGF-I was noted (Table 4). These results suggested that IGF-I-dependent motility in both types of cells requires the PI-3K and p38 pathways but does not rely on the activity of ERK1/ERK2.

The experimental and clinical evidence supports the notion that hyperactivation of the IGF-IR may be critical in early steps of breast cancer development, promoting cell growth, survival, and resistance to therapeutic treatments. However, the function of the IGF-IR in the later stages of the disease, including metastasis, is still obscure (1). For instance, whereas the IGF-IR has been found overexpressed in primary breast tumors, its levels, similar to ER levels, appear to undergo reduction during the course of the disease (1, 18). According to Pezzino et al. (41), who studied the IGF-IR status in two patient subgroups representing either a low risk (ER- and progesterone receptor-positive, low mitotic index, diploid) or a high risk (ER- and progesterone receptor-negative, high mitotic index, aneuploid) population, there is a highly significant correlation between IGF-IR expression and better prognosis. Similar conclusions were reached by Peyrat and Bonneterre (42) and recently by Schnarr et al. (18). Therefore, it has been proposed that similar to the ER, the IGF-IR marks more differentiated tumors with better clinical outcome. However, it has also been argued that the IGF-IR may play a role in early steps of tumor spread because node-positive/IGF-IR-positive tumors appeared to have a worse prognosis than node-negative/IGF-IR-positive tumors (1, 42). In addition, quite rare cases of ER-negative but IGF-IR-positive tumors are associated with shorter disease-free survival (43).

In breast cancer cell lines, a hormone-dependent and less aggressive phenotype correlates with a good IGF-IR expression (1, 19, 42). By contrast, different ER-negative, breast cancer cell lines express low levels of the IGF-IR and generally do not respond to IGF-I with growth (1, 18, 19, 20, 21, 22). However, many ER-negative cell lines appear to depend on the IGF-IR for tumorigenesis and metastasis. For instance, blockade of the IGF-IR in MDA-MB-231 cells by anti-IGF-IR antibody reduced migration in vitro and tumorigenesis in vivo, and expression of a soluble IGF-IR in MDA-MB-435 cells impaired growth, tumorigenesis, and metastasis in animal models (1, 14, 16, 23). These observations suggest that some growth-unrelated pathways of the IGF-IR may be operative in the context of ER-negative cells.

Here we studied whether this particular IGF-I dependence of ER-negative breast cancer cells relates to the nonmitogenic function of the IGF-IR, such as cell migration. Our experiments indicated that the IGF-IR is an effective mediator of cell motility. Furthermore, IGF-I-induced migration was proportional to IGF-IR content. We demonstrated, for the first time, that in MDA-MB-231 ER-negative cells, IGF-IR signaling pathways responsible for cell movement include PI-3K and p38 kinases. Indeed, an acute IGF-I stimulation of MDA-MB-231 and MDA-MB-231/IGF-IR cells appears to induce both PI-3K and p38 kinases, suggesting that this short-time activation may be involved in migration. Both of these pathways have been shown previously to regulate cell motility in breast cancer cells and other cell types (35, 44). Interestingly, the migration of both ER-negative and ER-positive cells was enhanced by a specific MEK1/MEK2 inhibitor UO126. We observed this effect over a broad range of UO126 doses (1–10 μm) and in several MDA-MB-231- and MCF-7-derived clones; the same doses always suppressed cell proliferation in serum-containing medium and PRF-SFM (data not shown and Table 2). These peculiar effects suggest that MEK1/2 may represent a regulatory point balancing mitogenic and nonmitogenic cell responses.

In contrast with the positive effects of IGF-I on cell motility in ER-negative and ER-positive breast cancer cells, this growth factor never stimulated the proliferation of MDA-MB-231 cells, whereas it induced the growth of MCF-7 cells and MCF-7-derived clones overexpressing the IGF-IR. It is has been established by Rubini et al. (30) and Reiss et al. (31) that mitogenic response to IGF-I requires a threshold level of IGF-IR expression (in fibroblasts, ∼1.5 × 10⁴). Here, we demonstrated that increasing the levels of the IGF-IR from ∼7 × 10³ up to ∼2.5 × 10⁵ and subsequent up-regulation of IGF-IR tyrosine phosphorylation was not sufficient to induce IGF-I-dependent growth of MDA-MB-213 cells. Similar results were obtained by Jackson and Yee (21), who showed that overexpression of IRS-1 in ER-negative MDA-MB-435A and MDA-MB-468 breast cancer cells did not stimulate IGF-I-dependent mitogenicity. These authors suggested that the lack of IGF-I response, even in IRS-1-overexpressing ER-negative cells, was related to insufficient stimulation of ERK1/ERK2 and PI-3K pathways (21). Defective insulin response in ER-negative cell lines has also been described by Costantino et al. (45) and linked with an increased tyrosine phosphatase activity.

Our experiments suggested that the lack of IGF-I mitogenicity in MDA-MB-231 and MDA-MB-231/IGF-IR cells was not related to the impaired IGF-IR or IRS-1 tyrosine phosphorylation. The cells were also able to respond to an acute IGF-I stimulation with a marked activation of the PI-3K/Akt and ERK-1/ERK2 pathways. We hypothesize that this transient stimulation could be sufficient to induce some IGF-I response, such as cell motility. Mitogenic response, on the other hand, may rely on a more sustained activation of critical IGF-IR signals, as demonstrated before with mouse embryo fibroblasts (36). Indeed, the most significant difference in IGF-I signal between ER-negative and ER-positive cells rested in the impaired long-term stimulation of the PI-3K/Akt pathway; MDA-MB-231 and MDA-MB-231/IGF-IR cells were unable to sustain this IGF-I-induced signal for 1 or 2 days, whereas in MCF-7 and MCF-7/IGF-IR cells, the PI-3K/Akt pathway was still active at this time. The subsequent experiments with MDA-MB-231 cells transfected with constitutively active Akt mutants demonstrated that the increased biological activity of Akt was not sufficient to completely reverse cell death in PRF-SFM (Fig. 6,C and data not shown). This suggested that although a sustained Akt activity could be important in the survival of breast cancer cells, other pathways, or a proper equilibrium between Akt and other pathways (such as ERK1/2), are also critical. The latter possibility could be supported by our finding that hyperactivation of Akt down-regulates the ERK1/2 pathway. Normally, this pathway appears to play a role in the survival of ER-negative cells (Table 2).

In summary, our data suggest that IGF-IR signaling and function may be different in hormone-dependent and -independent breast cancer cells. In ER-positive MCF-7 cells, IGF-IR transmits various signals, such as growth, survival, migration, and adhesion. In ER-negative MDA-MB-231 cells, the growth-related functions of the IGF-IR become attenuated, but the receptor is still able to control nonmitogenic processes, such as migration. It is likely that this kind of evolution is also involved with the response to other growth factors. Epidermal growth factor, for instance, is an effective mitogen for ER-positive breast cancer cells but does not stimulate the proliferation or survival in MDA-MB-231 cells, despite high EGF-R expression (46). However, as demonstrated recently by Price et al. (46), EGF is a potent chemoattractant for MDA-MB-231 cells. EGF-induced migration in MDA-MB-231 cells requires PI-3K and phospholipase Cγ and is not inhibited by antagonists of ERK1/ERK2.

In conclusion, mitogenic and nonmitogenic pathways induced by growth factors in breast cancer cells may be dissociated, and attenuation of one is not necessarily linked with the cessation of the other. Delineating the nonmitogenic responses will be critical for the development of drugs specifically targeting metastatic cells.