Leptin Mediates Tumor–Stromal Interactions That Promote the Invasive Growth of Breast Cancer Cells

For the past 3 decades, cancer research focused predominantly on the characteristics of breast cancer cells. Recently, clinical and experimental studies revealed that both tumor initiation and progression are related to the complex interactions that transpire within the tumor microenvironment. The stromal compartment is composed of mesenchymal cells (fibroblasts, adipocytes, blood cells) and extracellular matrix (ECM; lamin, fibronectin, collagen, proteoglycans, and so on), and signals from these cells come as soluble secreted factors, ECM components, or direct cell–cell contacts. Growth factors, cytokines, adipokines, proteases, and vascular-stimulating factors are involved in stroma-mediated procancerous activities (1–4). The chemokines CXCL12, CXCL14, and CCL7 stimulated tumor cell proliferation and invasion in vitro and in vivo and increased tumor angiogenesis and macrophage presence at tumor sites (5–7). The interleukins (IL)-1 and -8 induced cancer progression by enhancing metastasis and cachexia (8, 9). As important adipocyte-derived endocrine and paracrine mediator, the adipokine leptin has been correlated with breast cancer occurrence. Indeed, leptin synthesis and plasma levels increase with obesity, a pandemic condition that influences both risk and prognosis of breast cancers (10).

The processes of heterotypic signaling involve a constant bidirectional cross-talk between stromal cells and malignant cells. Stromal cells influence tumor invasiveness and malignancy, whereas at the onset and during breast cancer progression, the microenvironment is reorganized by cancer cells (11). Tumors recruit stromal fibroblasts in a process referred to as the desmoplasmic reaction, and these carcinoma-associated fibroblasts (CAF) are reprogrammed to produce growth factors, cytokines, and ECM-remodeling proteins, that acting in autocrine and paracrine fashion support tumor proliferation and invasion into surrounding tissues (4). Moreover, a variety of these factors may activate estrogen receptor-α (ERα; ref. 12).

Estrogens and its receptor play a crucial role in regulating breast cancer growth and differentiation. Variant forms of ERα due to alternative splicing or gene mutation have been reported, but their clinical significance is still unresolved (13, 14). A naturally occurring mutation at nucleotide 908, introducing a lysine to arginine transition at residue 303 within the hinge domain of the receptor (K303R-ERα), was identified in one third of premalignant breast hyperplasias and one half of invasive breast tumors. This mutation correlated with poor outcomes, older age, larger tumor size, and lymph node–positive disease (15, 16). Other studies did not detect the mutation in invasive cancers (17–20), but our studies suggest that the detection method used might be insensitive. However, K303R expression was found at low frequency in invasive breast tumors by Conway and colleagues (21). K303R mutation allows ERα to be more highly phosphorylated by different kinases, and it alters the dynamic recruitment of coactivators and corepressors (22–24). Mutant overexpression in MCF-7 breast cancer cells increased sensitivity to subphysiologic levels of estrogen and decreased tamoxifen responsiveness when elevated growth factor signaling was present (15, 25). K303R-ERα mutation also conferred resistance to the aromatase inhibitor anastrozole (23, 26), suggesting a pivotal role for this mutation in more aggressive breast cancers.

The aim of this study was to elucidate the mechanisms underlying tumor–stroma interaction in ERα-positive breast cancer cells. First, we investigated how tumoral microenvironment pressure, exerted by CAFs, impacts breast cancer cell proliferation, migration, and invasiveness in relation to the expression of wild-type (WT) or the K303R-ERα. We then defined the effect that a single factor leptin has on stroma-mediated breast cancer progression. Finally, we examined the bidirectional interactions between CAFs and breast cancer cells, leading to increased malignancy.

The following reagents and antibodies were used: leptin, 17β-estradiol, and epidermal growth factor (EGF) from Sigma; ICI182780 from Tocris Bioscience; AG490, AG1478, PD98059, and LY294002 from Calbiochem; ERα, ERβ, glyceraldehyde-3-phosphate dehydrogenase (GAPDH), ObRl, ObRs, Ob, Akt, and pAkt^Ser437 antibodies from Santa Cruz Biotechnology; and mitogen-activated protein kinase (MAPK), Janus-activated kinase (JAK)2, STAT3, pMAPK^{Thr202/Tyr204}, pJAK2^Tyr1007/1008, pSTAT3^Tyr705/pERα^Ser118, and pERα^Ser167 from Cell Signaling Technology.

Generation of yellow fluorescent protein (YFP)-tagged expression constructs, YFP-WT and YFP-K303R-ERα, was done as described in the work of Cui and colleagues (22). XETL plasmid, containing an estrogen-responsive element, was provided by Dr. Picard, University of Geneva, Geneva, Switzerland.

MCF-7 and SKBR3 cells were acquired in 2010 from American Type Culture Collection where they were authenticated, stored according to supplier's instructions, and used within 4 months after frozen aliquots resuscitations. YFP-WT and YFP-K303R-ERα stably expressing MCF-7 cells, MCF-7 and SKBR3 pools stably transfected with YFP-WT and YFP-K303R-ERα were generated as described earlier (23, 26). Immortalized normal human foreskin fibroblasts BJ1-hTERT were provided by Dr Lisanti, Jefferson University, Philadelphia, PA. Every 4 months, cells were authenticated by single tandem repeat analysis at our Sequencing Core; morphology, doubling times, estrogen sensitivity, and mycoplasma negativity were tested (MycoAlert, Lonza).

Human breast cancer specimens were collected in 2011 from primary tumors of patients who signed informed consent. Following tumor excision, small pieces were digested (500 IU collagenase in Hank's balanced salt solution; Sigma; 37°C for 2 hours). After differential centrifugation (90 × g for 2 minutes), the supernatant containing CAFs was centrifuged (500 × g for 8 minutes), resuspended, and cultured in RPMI-1640 medium supplemented with 15% FBS and antibiotics. CAFs between 4 and 10 passages were used, tested by mycoplasma presence, and authenticated by morphology and fibroblast activation protein (FAP) expression.

CAFs were incubated with regular full media (48–72 hours). Conditioned media were collected, centrifuged to remove cellular debris, and used in respective experiments. Alternatively, conditioned media were collected from WT- and K303R-ERα–expressing MCF-7 cells incubated in 5% charcoal-stripped FBS (72 hours).

Expression profiles were determined with Affymetrix GeneChip Human Genome U133 plus 2.0 arrays. Data quality and statistical analyses were conducted as described in the work of Barone and colleagues (23). Microarray study followed MIAME (Minimum Information About a Microarray Experiment) guidelines. All data are available in our previous publication (23).

Protein extracts were subjected to SDS-PAGE as described (27). Immunoblots show a single representative of 3 separate experiments.

Cells were fixed with 4% paraformaldehyde, permeabilized with PBS + 0.2% Triton X-100 followed by blocking with 5% bovine serum albumin (1 hour at room temperature), and incubated with anti-ObR antibody (4°C, overnight) and with fluorescein isothiocyanate–conjugated secondary antibody (30 minutes at room temperature). IgG primary antibody was used as negative control. 4′,6-Diamidino-2-phenylindole (DAPI; Sigma) staining was used for nuclei detection. Fluorescence was photographed with OLYMPUS BX51 microscope, 100× objective.

The gene expression of FAP, Ob, cyclin D1, pS2, cathepsin D, and 36B4 was evaluated by reverse transcription PCR (RT-PCR) method as described in the work of Catalano and colleagues (27). The gene expression of ObRl, ObRs, CXCR4, insulin receptor (IR), IL-2RB, IL-6R, EGF receptor (EGFR), insulin-like growth factor-1 receptor (IGF1R), fibroblast growth factor receptor 3 (FGFR3), ERα, EGF, IL-6, and insulin was assessed by real-time RT-PCR, using SYBR Green Universal PCR Master Mix (Bio-Rad). Each sample was normalized on 18S mRNA content. Relative gene expression levels were calculated as described (28). Primers are listed in Supplementary Table S1.

ERα transactivation assays were conducted as described in the work of Catalano and colleagues (29).

After 4 days of treatments, cell proliferation was assessed by MTT (Sigma) and expressed as fold change relative to vehicle-treated cells.

After 4 days of treatment, cell numbers were evaluated by trypsin suspension of samples followed by microscopic evaluation using a hemocytometer.

Anchorage-independent growth assays were conducted as described in the work of Barone and colleagues (26).

Data represent 3 independent experiments carried out in triplicate.

Cell monolayers were scraped and treated as indicated. Wound closure was monitored over 24 hours; cells were fixed and stained with Coomassie brilliant blue. Images represent 1 of 3 independent experiments (10× magnification, phase contrast microscopy).

Cells treated with or without leptin were placed in the upper compartments of Boyden chamber (8-μm membranes; Corning Costar). Bottom well contained regular full media. After 24 hours, migrated cells were fixed and stained with Coomassie brilliant blue. Migration was quantified by viewing 5 separate fields per membrane at 20× magnification and expressed as the mean number of migrated cells. Data represent 3 independent experiments, assayed in triplicate.

Matrigel-based invasion assay was conducted in invasion chambers (8-μm membranes) coated with Matrigel (BD Biosciences; 0.4 μg/mL). Cells treated with or without leptin were seeded into top Transwell chambers, whereas regular full medium was used as chemoattractant in lower chambers. After 24 hours, invaded cells were evaluated as described for transmigration assays. Data represent 3 independent experiments, assayed in triplicate.

In vivo studies were conducted as described in the work of Mauro and colleagues (30).

Leptin was measured by a competitive in-house immunoassay (Chematil) following manufacturer's protocol. Results are presented as nanograms per cell.

Protein G-agarose beads were incubated with anti-leptin or IgG antibodies. Antibody–beads complexes were incubated with CAF conditioned media and centrifuged. Leptin immunodepletion was verified by radioimmunoassay (RIA).

Data were analyzed for statistical significance with 2-tailed Student t test using GraphPad Prism 4. SDs are shown. Survival curves were computed by Kaplan–Meier method and compared using 2-sided log-rank tests.

Epithelial–stromal interactions support tumor cell proliferation and invasion. Thus, we first investigated the role of tumoral microenvironment in influencing breast cancer phenotype in relation to the expression of WT- or K303R-ERα mutant receptor.

We used ERα-positive MCF-7 cells stably transfected with YFP-WT or YFP-K303R-ERα expression vectors as experimental models for breast cancer. We chose this approach because wild-type receptor was present along with K303R-ERα in invasive breast tumors (16). Stable clones were screened for ERα expression using immunoblot analysis (Fig. 1A). Two clones stably expressing YFP-WT (WT1-2) or YFP-K303R-ERα (K303R1-2) are shown along with wild-type or mutant receptor stable pools (WT P and K303R P). As stromal cells, we used CAFs isolated from biopsies of primary breast tumors. CAFs possessed the basic fibroblast characteristics with long and spindle-shaped morphology and highly expressed the FAP (Fig. 1B). To create in vitro conditions that can mimic the complex in vivo microenvironment, we used coculture experiments. Breast cancer cells were incubated with regular full media, CAF-derived conditioned media, or normal fibroblasts conditioned media, and growth was evaluated by soft agar assays (Fig. 1C). As previously shown (23, 26), control basal growth of mutant-expressing cells was elevated compared with wild-type–expressing cells. CAF-derived conditioned media significantly increased colony numbers in both WT- and K303R-ERα–expressing cells; however, CAF-derived conditioned media enhanced K303R-expressing cell growth at a higher extent than wild-type–expressing cells. We then examined the ability of CAF-derived conditioned media to promote WT- and K303R-expressing cell movement in wound-healing scratch assays (Fig. 1C). The mutant cells moved the farthest in either direction to close the gap compared with wild-type–expressing cells. CAF-derived conditioned media promoted net movement of wild-type–expressing cells compared with full media, but K303R-expressing cells exposed to CAF-derived conditioned media moved at higher rate to close the gap in the cell bed. As expected, CAFs possessed a higher ability to enhance both proliferation and motility of breast cancer cells than normal fibroblasts (Fig. 1C). CAF-derived conditioned media–induced cell growth and migratory potential was blocked by inhibition of the classic cytokine JAK2/STAT3 signaling cascade (AG490) and the ERα signaling inhibitor (ICI182780), although to a higher extent in K303R clones (Fig. 1D). All functional effects described so far are the results of exposure to the total complement of CAF-secreted proteins. However, it is desirable, albeit experimentally difficult, to define the contribution of a single factor. Thus, we addressed which CAF-secreted factor may promote breast cancer cell growth and motility.

Diffusible growth factors, interleukins, chemokines, and adipokines implicated as mediators of stromal–epithelial interactions are involved in breast carcinoma initiation and progression. To determine changes in gene expression for the different receptors of CAF-secreted factors that may be responsible for the different sensitivity of wild-type and mutant clones to CAF-derived conditioned media exposure, we conducted microarray analysis. Gene expression profile comparing RNA isolated from K303R-expressing with WT-expressing cells are shown in Table 1 and Supplementary Table S2. K303R-ERα expression induced several genes potentially involved into tumor–stroma interactions; however, the leptin receptor (ObR) gene was the most highly induced (2.4-fold; Table 1). We also observed increased expression of different leptin signaling downstream effectors such as JAK2, the transcription factors fos, and STAT, as well as the suppressor of cytokine signaling 3 (Supplementary Table S3). To validate the microarray study, YFP-WT and YFP-K303R-ERα–expressing cells were evaluated for a panel of genes using real-time PCR (Fig. 2A). K303R-associated induction could be confirmed for all of them, and again, the gene encoding the long and short leptin receptor isoforms (ObRl/ObRs) was the most highly upregulated in mutant-expressing cells. However, we did not detect any differences in IGF1R mRNA expression levels between the 2 cells, although microarray analysis showed a significant decrease of IGF1R. ERα RNA levels were similar between WT- and K303R-ERα–expressing cells.

The increase in both ObRl and ObRs was then confirmed by evaluating protein levels using immunoblotting (Fig. 2B) and immunofluorescent staining of WT- and K303R-ERα–expressing cells (Fig. 2C, red).

Given the gene expression profile identified in the microarray study, we defined the impact that a single factor leptin may have on K303R-ERα breast cancer cell progression. First, time course–response studies were conducted to analyze phosphorylation of leptin downstream effectors using immunoblot analysis (Fig. 2D). WT-expressing cells exhibited low basal levels of phosphorylated JAK2, STAT3, Akt, and MAPK that were increased in a time-dependent manner after leptin treatment. In contrast, K303R-expressing cells showed elevated constitutive phosphorylation of these signaling molecules in control vehicle conditions, that was slightly increased after leptin treatment. Thus, the mutant ERα expression was associated with increased leptin signaling activation.

Leptin directly activates ERα in the absence of its own ligand in MCF-7 breast cancer cells (29). As a consequence of the enhanced leptin signaling, we found increased ERα transcriptional activity and upregulated mRNA levels of the classical ERα target genes cyclin D1, pS2, and cathepsin D in both control and leptin-treated conditions in K303R-ERα–expressing cells. In addition, the mutant exhibited elevated pS118 and pS167 YFP-K303R-ERα levels (Supplementary Fig. S1).

We next used these stably transfected clones as model systems to study leptin sensitivity, in relation to mutant receptor expression. First, we evaluated leptin effects on growth using anchorage-dependent growth assays (Supplementary Fig. S2). As expected, in both WT- and K303R-expressing cells, treatment with leptin (100 ng/mL) increased cell proliferation. However, low leptin treatment (10 ng/mL) significantly enhanced cell viability only in K303R-expressing clones. We also evaluated leptin-mediated proliferative effects in anchorage-independent growth assays (Fig. 3A). Leptin treatments at 100 and 1,000 ng/mL concentrations enhanced colony numbers in all 4 clones tested, although to a higher extent in mutant-expressing cells. Again, leptin at 10 ng/mL increased anchorage-independent growth only in K303R cells. The increase in colony numbers induced by leptin was reversed by the JAK2/STAT3 inhibitor AG490 (Fig. 3B). We also used the antiestrogen ICI182780 and found that this treatment suppressed anchorage-independent growth of both cell lines, indicating that ER expression remains important in growth regulation of these cells (Fig. 3B).

We next evaluated the ability of increasing doses of leptin to influence cell migration in wound-healing scratch assays (Fig. 3C). Again, the mutant cells moved the farthest in either direction to close the gap compared with WT-expressing cells. Leptin treatments at 100 and 1,000 ng/mL promoted cell motility in both WT- and K303R-expressing cells, although to a higher extent in mutant cells. Interestingly, leptin at 10 ng/mL stimulated migration only in K303R-expressing cells. Then, the capacity of cells to migrate across uncoated membrane in transmigration assays or to invade an artificial basement membrane Matrigel in invasion assays was tested in the presence of leptin (Fig. 3D). Although wild-type cells exhibited little motile and no invasive behavior in vitro, our data clearly showed that mutant receptor expression increased both motility and invasion of cells. High doses of leptin increased the number of migrated and invaded cells in both clones and again low doses of leptin stimulated motility and invasion only of cells expressing the K303R receptor. As expected, treatment with AG490 and ICI182780 resulted in a clear reduction of both control-untreated and leptin-induced cell motility in wound-healing scratch assays, especially in K303R-expressing cells (Fig. 3E). We recently published that K303R-ERα MCF-7 xenograft tumors grew faster than WT-ERα tumors (26). In addition, MCF-7 xenograft tumors doubled control value after 13 weeks of leptin exposure (30). Thus, we determined whether the mutant receptor–expressing breast cancer cells might exhibit an increased sensitivity to leptin stimulation also in vivo. We found that in mice treated with leptin, all xenografts derived from cells with K303R-ERα expression doubled in size within 6 weeks of treatment, whereas none of xenografts from WT-ERα–expressing cells doubled in size during this experiment (Fig. 3F). Thus, expression of the mutant generated a leptin hypersensitive phenotype in vitro and in vivo.

We next assessed the role of leptin in the context of heterotypic signaling working in tumor–stroma interactions. First, RIA measurement in CAF-derived conditioned media showed that leptin secretion varied from 10 ± 4.5 ng per 200,000 cells. RT-PCR evidenced Ob mRNA expression in CAFs; CAFs also expressed ObR long isoform, but they did not express ObR short isoform, ERα, or ERβ (Supplementary Fig. S3). Leptin was then immunodepleted from CAF-derived conditioned media by leptin-specific antibodies, and resulting media were tested for the ability to induce anchorage-independent growth and migration of breast cancer cells. Leptin depletion (conditioned media + LepAb) significantly decreased growth- and migration-promoting activities of CAF-derived conditioned media, particularly in K303R-expressing cells (Fig. 4A). Conditioned media treated with a nonspecific mouse IgG had no effects, suggesting that the neutralizing effects of leptin antibodies were specific. Our results identify leptin as a main molecular player that mediates CAF effects on tumor cell growth and migration.

Leptin activates via JAK2 the MAPK and phosphoinositide 3-kinase (PI3K)/Akt pathways (31). Thus, we investigated the specific signaling involved in the CAF–breast cancer cells interaction and found that the PI3K/Akt inhibitor LY294002 was more effective in inhibiting CAF-induced proliferation and migration than the MEK1 inhibitor PD98059 (Fig. 4B).

To extend the results obtained, we generated pools of YFP-WT and YFP-K303R-ERα stable transfectants in ERα-negative SKBR3 breast cancer cells (Supplementary Fig. S4). As previously shown for MCF-7 cells, we found a significant increase in both long and short leptin receptor isoforms mRNA in mutant-expressing cells. Again, treatment with leptin at 100 and 1,000 ng/mL significantly increased colony numbers of wild-type clones and to a higher extent of K303R-expressing cells; however, 10 ng/mL of leptin enhanced anchorage-independent growth only in K303R-expressing clones. Similarly, low leptin promoted migration only in mutant cells. Finally, we tested CAF-derived conditioned media for its effects on cell growth and migration. We found a great induction of anchorage-independent growth and motility after treatment with conditioned media, especially in K303R-expressing SKBR3 pools. Leptin-immunodepleted conditioned media strongly reduced conditioned media proliferative and migratory promoting activities on K303R cells, confirming that leptin hypersensitive phenotype was associated with K303R-ERα expression in different cellular backgrounds.

CAFs and tumor cells cross-talk via different soluble factors, where the effects on both subpopulations determine the final outcome of the tumorigenic process. Thus, as a final step of this study, we defined the effects of conditioned media from WT- and K303R-ERα–expressing breast cancer cells on CAF phenotype. Treatment with K303R-derived conditioned media elicited a dramatic alteration in the shape of CAFs in vitro, accompanied by an increased FAP mRNA expression (Fig. 5A). K303R-derived conditioned media also stimulated CAF viability and motility compared with WT-derived conditioned media effects (Fig. 5B), suggesting that how soluble K303R-ERα cell–secreted factors may generate a more activated CAF phenotype. Because leptin synthesis is influenced by different humoral factors (32–34), we evaluated the effects of breast cancer–derived conditioned media in modulating leptin secretion from CAFs. Incubation of CAFs with K303R-derived conditioned media increased leptin mRNA expression and leptin release compared with WT-derived conditioned media, whereas no differences were detected in leptin levels among WT- and K303R-derived conditioned media (Fig. 5C). Finally, to investigate the paracrine factor by which breast cancer cells may affect CAF phenotype, we used microarray analyses to measure the expression of different genes known to be associated with CAFs and/or leptin secretion (11, 32–35). Our results showed that the genes encoding for EGF (2.8-fold), IL-6 (1.2-fold), and insulin (1.2-fold) were induced in mutant cells, and real-time PCR confirmed that the EGF gene was most highly upregulated (Fig. 5D). Thus, we evaluated the role of EGF. First, addition of EGF in WT-derived conditioned media mimicked the induction of K303R-derived conditioned media on CAF motility, and the EGFR signaling inhibitor (AG1478) reduced K303R-derived conditioned media effects (Fig. 5E). Second, treatment with AG1478 reversed K303R-derived conditioned media stimulated FAP mRNA expression (Fig. 5F). Third, Ob mRNA expression and leptin secretion from CAFs cocultured with K303R-derived conditioned media were significantly decreased in the presence of AG1478 (Fig. 5F).

Our data show that K303R-ERα-expressing breast cancer cells through their soluble secreted factors may take advantage of the plastic nature of reactive surrounding cell populations, as CAFs, to generate a tumor-enhancing microenvironment.

ERα expression has important implications for breast cancer biology and therapy. Fuqua and colleagues identified a lysine to arginine transition at residue 303 of ERα (K303R-ERα) in 30% of breast hyperplasias and in 50% of invasive breast cancers (15, 16), although using another detection method, the mutation was identified in only 6% of tumors (21); thus the frequency is still unresolved. This mutation was associated with older age, larger tumor size, lymph node positivity, and shorter time to recurrence—all features related to a more aggressive breast cancer phenotype. Because of the recently recognized importance of tumor–stroma cross-talk in promoting breast cancer progression and metastasis, it is imperative to elucidate the molecular events occurring between cancer cells and adjacent stroma at the site of primary tumors to provide new treatment options for breast cancer.

Here, we elucidated the complex interactions between peritumoral tissue, locally derived factors, and neoplastic cells in dependency of ERα status, with a special focus on leptin effects in influencing the behavior of breast cancer cells bearing the naturally occurring K303R-ERα mutation. We proposed a model in which leptin, secreted from CAFs, binds to its receptor, activates K303R-ERα, and promotes proliferation, migration, and invasiveness of K303R-ERα–expressing breast cancer cells. In turn, K303R cells release factors as EGF that “educate” CAFs to enhance secretion of leptin, which, acting back on malignant cells, may establish a positive feedback loop between cancer and stromal cells to further support breast tumor progression (Fig. 6).

The phenotype of malignant cells appears regulated not only by cell autonomous signals but also is dependent on heterotypic signals coming from surrounding stromal cells, able to create a specific local microenvironment to tightly control breast cancer proliferation and differentiation (36–38).

We defined the molecular interactions between stromal fibroblasts isolated from biopsies of primary breast tumors (CAFs), WT-, and K303R-ERα–expressing MCF-7 breast cancer cells. The initial conditioned media experiments showed that the entire complement of secretory proteins released by CAFs have more profound effects on K303R-ERα–expressing cell proliferation and migration than on WT-ERα cells. We evidenced an important role for JAK2/STAT3 and ERα signaling pathways in conditioned media–mediated effects. Our microarray study pointed to the regulation of several important transcriptional programs of growth factors and cytokines receptors that, acting as mediators of stromal–epithelial interactions, are potentially involved in carcinoma progression. Among them, the gene encoding for leptin receptor was the most highly induced in K303R-expressing breast cancer cells.

Leptin is primarily synthesized from adipocytes but is also produced by other cells, including fibroblasts (39–41). We showed, for the first time, Ob mRNA expression and leptin secretion in CAFs. CAFs expressed ObR long isoforms, implying that an autocrine feedback loop may exist. Leptin immunodepletion from CAF-derived conditioned media substantially reduced the growth- and migration-promoting activities of CAFs. As one of the leptin downstream effectors (31), we found that the PI3K/Akt inhibitor LY294002 was effective in inhibiting CAF-induced effects.

Because fibroblasts are the principal cellular component of the stroma, our results suggest that in the breast microenvironment CAFs through leptin signaling may become the main actor in influencing tumor cell behavior, especially in K303R-ERα–expressing breast cancer cells.

Leptin, a pleiotropic molecule that regulates food intake, hematopoiesis, inflammation, cell differentiation, and proliferation, is also required for mammary gland development and tumorigenesis. Indeed, leptin and its receptor isoforms (ObRs) have been detected in mammary epithelium and breast cancer cell lines and are overexpressed in cancer tissue compared with healthy epithelium, with a positive correlation between ObR and ERα expression (42, 43). Real-time PCR, immunoblotting, and immunofluorescent experiments revealed an increase in mRNA and protein expressions of ObR long and short isoforms in K303R-ERα–expressing cells. We also showed that the mutant expression was associated with enhanced leptin signaling activation and increased sensitivity to leptin stimulation on growth, motility, and invasiveness. Moreover, a significant increase in the growth of leptin-treated mutant tumors was observed in vivo.

Leptin is a potent modulator of the estrogen signaling pathway (29, 44). On the contrary, estradiol modulates ObR expression in rat brain, through a putative estrogen-responsive element in its promoter (45, 46), and others showed that estradiol induces leptin and ObR expression in MCF-7 breast cancer cells (43). Thus, leptin and estrogen might cooperate in sustaining estrogen-dependent breast carcinoma growth. We showed an increased S167 and S118 phosphorylation of the K303R receptor, an enhanced K303R-ERα transactivation, and a more pronounced upregulation of classical estrogen-regulated genes in K303R-expressing cells. Indeed, the pure antiestrogen ICI182760 drastically suppressed leptin-stimulated anchorage-independent growth and motility of mutant cells.

These results suggest that the mutation may potentiate the role of ERα as an effector of leptin intracellular signal transduction, which may enhance cell proliferation, migration, and invasiveness, contributing to the more aggressive phenotype of K303R-associated breast cancers.

In the same way as tumor microenvironment plays active roles in shaping the fate of a tumor, cancer cells actively recruit fibroblasts into the tumor mass, in particular, the subpopulation named CAFs. This cell type is defined on the basis of the morphologic characteristics or expression of markers as the FAP (1–4).

Studies addressing these issues are heterogeneous in terms of cell systems used, tumor cell types, and fibroblast sources. Experimental systems have used different tumor-derived conditioned media to stimulate CAFs, and others have cocultured tumor cells with normal fibroblasts or mesenchymal stem cells and measured chemokines levels in the resulting conditioned media. For instance, fibroblasts growth with tumor cells resulted in increased production of chemokines whose source is in CAFs themselves. Chemokines produced under these “mixed” conditions promoted tumor promalignancy activities (6, 9, 47). We showed increased leptin mRNA expression and secretion by CAFs in response to soluble K303R-ERα cell–secreted factors compared with WT-derived conditioned media, suggesting that K303R cells have the ability to instruct their surrounding fibroblasts to augment leptin production, thereby enhancing tumor growth. This further indicates that interactions between the 2 subpopulations are actually bidirectional.

These interactions become more productive when tumor cells have a higher aggressiveness phenotype (47–49). CAFs exposed to K303R cell–derived conditioned media acquired a more activated phenotypic characteristic, as revealed by an altered morphology, an increased FAP mRNA expression, and enhanced proliferative and migratory capabilities. We identified the epidermal growth factor, known to affect CAFs phenotype and leptin secretion (11, 32–35), as the factor responsible of the paracrine activation of the surrounding stroma.

Thus, a positive feedback loop is established which leads to the development and growth of the tumor.

Our study highlights the functional importance of tumor–host cross-talk in impacting malignant cell behavior and implies several clinical implications. First, because K303R mutation was identified in breast premalignant hyperplasia, it is tempting to speculate that this specific mutation hypersensitive to leptin signaling may promote or accelerate the development of cancers from premalignant breast lesions, further increasing risk in obese women. Second, understanding the key genes involved differently in relation to ERα status in tumor–stroma interactions may help to identify novel biomarkers for breast cancer. Finally, our findings support the development of new therapeutics targeting stroma signaling components (e.g., leptin) to be implemented in the adjuvant therapy for improving clinical care and reducing mortality from breast cancer.

No potential conflicts of interest were disclosed.

The authors thank Dr. F. Romeo for providing the breast cancer specimens.

This work was supported by Reintegration AIRC/Marie Curie International Fellowship in Cancer Research to I. Barone; Lilli Funaro Foundation to I. Barone and L. Gelsomino; European Commission/FSE/Regione Calabria to L. Gelsomino; AIRC grant (IG11595) to S. Catalano, D. Bonofiglio, C. Giordano, and S. Andò; AIRC grant (IG8804) to G. Bossi; and NCI RO1 CA72038 to S.A.W. Fuqua.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.