Rapamycin inhibits multiple stages of c-Neu/ErbB2–induced tumor progression in a transgenic mouse model of HER2-positive breast cancer

Amplification of the HER2 (also called ErbB2 or c-Neu) proto-oncogene occurs in 25% to 30% of breast cancers and confers a poor prognosis due to refractoriness to conventional chemotherapies and high relapse rates (1, 2). HER2/ErbB2 is a member of the epidermal growth factor family of receptor tyrosine kinases (3). In conjunction with ErbB3, HER2/ErbB2 activates a number of signaling pathways including the phosphatidylinositol 3-kinase (PI3K) pathway (4). When PI3K is activated, it signals through the Akt/protein kinase B kinase, which subsequently activates the growth regulator mammalian target of rapamycin (mTOR; ref. 5). The mTOR kinase regulates cell proliferation and growth through translational control of an array of proteins (6). mTOR phosphorylates p70S6 kinase (p70S6K), which, in turn, phosphorylates the ribosomal protein S6 (S6), promoting translation of proteins bearing a 5′-oligopyrimidine tract. mTOR also phosphorylates eIF-4E-BP1 (4EBP1), a negative regulator of the translational initiator eIF-4E. Phosphorylation of 4EBP1 induces its dissociation from eIF-4E, which enhances cap-dependent translation of numerous mRNAs including cell cycle regulators such as cyclin D1, an essential regulator of proliferation in HER2/ErbB2 cells (7, 8).

Elevated levels of PI3K, p70S6K, and eIF-4E as well as decreased levels of 4EBP1 are associated with tumor aggressiveness in vivo and in vitro (6, 9, 10). In human breast cancer, Zhou et al. (11) have linked mTOR activation to HER2-positivity; high levels of phosphorylated Akt, phosphorylated mTOR, and phosphorylated 4EBP1 were all associated with HER2/ErbB2 expression as well as decreased disease-free survival. Rapamycin, an inhibitor of mTOR, decreases proliferation of a number of mammary epithelial cell lines, particularly those that overexpress HER2/ErbB2, have activated Akt, or have up-regulated p70S6K (12–14). In addition, forced expression of HER2/ErbB2 in MCF-7 cells promotes invasiveness and colony formation that can be blocked by rapamycin (11, 14). In mice that develop mammary tumors in response to overexpression of the rat ErbB2 (c-Neu) gene, the rapamycin analogue RAD001 inhibits proliferation of tumor cells (15). In addition, growth of mouse mammary tumors induced by a constitutively active form of c-Neu (NeuYD) is also attenuated by rapamycin (16).

To date, the effects of rapamycin on multiple stages of ErbB2/Neu-induced tumor formation—ranging from early, hyperplastic changes through distal metastases—have not been assessed. In the present study, we used mouse mammary tumor virus (MMTV)-c-Neu mice (17) to determine if rapamycin has more global inhibitory effects on multiple stages of tumorigenic progression that extend beyond inhibition of proliferation of primary tumors. These mice express rat c-Neu/ErbB2 selectively in the mammary gland and have been used to study preneoplastic changes in the gland (18–20) and mechanisms of metastatic progression (21–23). Herein, we report that rapamycin induces regression of primary tumors that overexpress c-Neu/ErbB2 as well as hyperplastic epithelia. Rapamycin also attenuates proliferation of established lung metastases in these mice. Together, these data indicate that rapamycin can inhibit multiple stages of tumor progression, suggesting that inhibition of the mTOR pathway may be a useful chemopreventive strategy as well as an effective treatment for early-stage and late-stage HER2/ErbB2 positive breast cancers.

Primary antibodies used for immunoblotting were P-mTOR (Ser²⁴⁴⁸; Cell Signaling), mTOR (Cell Signaling), β-actin (Sigma), cyclin D1 (BD PharMingen), P-S6 ribosomal protein (Ser^235/236; Cell Signaling), Akt (Cell Signaling), P-Akt (Ser⁴⁷³; Cell Signaling), and total caspase-3 (Cell Signaling). Secondary antibodies used for immunoblotting were horseradish peroxidase–conjugated goat anti-mouse and anti-rabbit (Santa Cruz Biotechnology). Rapamycin was purchased from Eton Biosciences. Primary antibodies used for immunohistochemistry were P-S6 ribosomal protein (Cell Signaling), P-HER2/ErbB2 (Tyr⁸⁷⁷; Cell Signaling), P-histone-H3 (Upstate Biotechnology), cleaved caspase-3 (Asp¹⁷⁵; Cell Signaling), CD31 (DakoCytomation, clone JC70A), and Ki-67 (DakoCytomation, clone TEC-3). For terminal deoxyribonucleotidyl transferase–mediated dUTP nick end labeling (TUNEL) analysis, the ApopTag Plus Peroxidase In Situ Apoptosis Detection Kit (Chemicon International) was used.

FVB/N-TgN(MMTV-Neu)202Mul/J mice containing the rat proto-oncogene c-Neu targeted to mammary epithelium by the MMTV-long terminal repeat promoter (21) were purchased from The Jackson Laboratory and bred with FVB/N wild-type animals. Female progeny were housed in microisolator cages in pathogen-free conditions under a 12-h light/dark cycle and provided food and water ad libitum. Mice were genotyped as previously described (19). All animal studies were approved by the Case Western Reserve University Institutional Animal Care and Use Committee.

Gene expression microarray analysis of age-matched wild-type glands (n = 3 pools of 5 mice each) and primary tumors (n = 7) using the Affymetrix Murine U74Av2 GeneChip Array have previously been described (19). Briefly, computational analyses were done with Microarray Suite (v.5.0, Affymetrix), Data Mining Tool (DMT v.3.0, Affymetrix), MicroDB (v.3.0, Affymetrix), and GeneSpring (v.6.0, Silicon Genetics) software. Probes that received a call of “absent” by Affymetrix detection call on all GeneChips were not analyzed. Changed genes were identified by application of Welch's approximate t test using a Benjamini and Hochberg multiple testing correction to establish a false discovery rate of 5% (P < 0.05).

MMTV-c-Neu mice were randomly assigned to vehicle or rapamycin groups following palpable tumor detection. Mean tumor volumes at the start of the treatment paradigm were indistinguishable between the two groups (vehicle, 205 ± 58 mm³; rapamycin, 263 ± 77 mm³; P = 0.56). Rapamycin-treated mice received 150 μg of rapamycin, made by diluting 3 μL of a 50 mg/mL stock (in 100% ethanol) into 100 μL of vehicle (5.2% Tween 80, 5.2% polyethylene glycol-400 in 0.9% saline). Control mice received 3 μL of ethanol diluted in 100 μL of vehicle. Drug or vehicle was delivered by i.p. injection every other day for 32 days. Tumor volumes, body weights, and changes in grooming behavior were monitored during the treatment paradigm. Animals were euthanized if estimated tumor burden approached 10% of the total body weight. Tumor dimensions were measured by external calipers and volumes were estimated using the formula (L × W²)/2, where L is the largest diameter. Mice were killed by asphyxiation with CO₂, mammary tumors and tissues were removed, and lungs were examined for metastases. For short-term studies, mice were treated as described above, except that rapamycin or vehicle was given for either 5 days (two treatments) or 10 days (five treatments).

Tumors were collected and bisected. One half was frozen on dry ice and immediately stored at −80°C. The second half of the tumor as well as lungs were fixed overnight in 4% (w/v) paraformaldehyde in PBS. Fixed samples were paraffin embedded, sectioned at 5 μm, baked, and either stained with H&E or used for immunohistochemistry.

Viable tumor area was quantified by manually demarcating the tumor boundary and the boundary of the grossly necrotic portion of the tumor on an electronic image of an H&E-stained section. These areas were quantified by using AxioVision (v4.1) software (Zeiss), and viable tumor area was defined as the proportion of total tumor area that was not grossly necrotic.

Whole tissues were homogenized in 1.5 mL of nondenaturing protein lysis buffer [20 mmol/L Tris-HCl (pH 7.5), 1% Triton X-100, 100 mmol/L NaCl, 40 mmol/L NaF, 1 mmol/L EDTA, 1 mmol/L EGTA] with Protease Inhibitor Cocktail (Sigma) and 1 mmol/L Na₃VO₄. Following lysis on ice for 30 min, homogenates were clarified by centrifugation (12,000 × g) for 10 min at 4°C. Protein concentrations were quantified by Bradford Protein Assay (Bio-Rad). One hundred micrograms of protein were denatured in Laemmli sample buffer (Bio-Rad), resolved by SDS-PAGE, and transferred to polyvinylidene difluoride membranes (Bio-Rad). Membranes were blocked with 5% (w/v) skim milk in either PBS or PBS with 0.05% (v/v) Tween 20 (1 h, room temperature). Membranes were incubated with primary antibody in 5% (w/v) bovine serum albumin (overnight, 4 C) followed by incubation with the appropriate secondary antibody conjugated to horseradish peroxidase in 5% skim milk (1 h, room temperature). Bound antibodies were detected by chemiluminescence (Amersham). Protein bands were quantified by densitometric analyses of photographic films using the Bio-Rad GS-710 densitometer.

Boiling in 10 mmol/L citrate buffer (pH 6.0) for 15 to 20 min was used for antigen retrieval for the following antibodies: P-S6 (1:200), P-histone-H3 (1:100), cleaved caspase-3 (1:50), and Ki-67 (1:25). A 1-min boil and 15-min sub-boil in 1 mmol/L EDTA (pH 8.0) was used for antigen retrieval for the p877 HER2/ErbB2 antibody (1:100) and a 30-min boil in 1 mmol/L EDTA and 10 mmol/L Tris (pH 9.0) followed by a 20-min cool was used for antigen retrieval for the CD31 antibody (1:50). All primary antibodies were incubated overnight at 4°C. Primary antibodies were visualized by using either mouse or rabbit Envision Plus-HRP kits (DakoCytomation) or rabbit anti-rat biotinylated immunoglobulin G (1:100) in combination with streptavidin/HRP (1:200; both from DakoCytomation) and bound antibody was detected by 3,3′-diaminobenzidine reaction. All tissues were counterstained with Gill's hematoxylin #3 (Polysciences, Inc.). As a negative control, blocking buffer was used without addition of primary antibody for each tissue. In this mouse model, hyperplastic tissue is largely restricted to glands that also harbor a tumor. Thus, all studies examining epithelial hyperplasia involved analyses of tissue surrounding tumors.

Proliferation of tumor epithelia was quantified by counting the number of tumor cells staining positive for phosphorylated histone-H3 in a 20× field. At least five fields were counted per tumor—three from the outer regions and two from viable tissue in the more central regions. The number of cells per field was identical for rapamycin- and vehicle-treated tumors. When lung metastases were immunostained for P-histone-H3, a very small proportion of cells showed positive staining (ranging from 5/1,000 cells to 15/1,000 cells), possibly indicative of low basal rates of proliferation. In addition, the small size of the metastases resulted in a low number of total cells available for assessment. Thus, proliferation in lung metastases was quantified by determining the percentage of mammary epithelial cells staining positively for Ki-67, which stained a more quantifiable proportion of cells (up to ∼9% of cells), in all tumor metastases and emboli within a section of lung. Ki-67 expression occurs in multiple stages of the cell cycle; hence, antibodies to this protein inherently detect more proliferating cells than P-H3, a marker of mitosis. A minimum of 250 metastatic cells were counted per lung.

Tumor vascularity was quantified by measuring the average percentage of tumor area positively stained with an anti–CD31 antibody. Electronic images of tumor regions were quantified using Metamorph (v5.0) software. The number of tumor regions quantified varied, depending on tumor size, and ranged from 5 to 50 individual 20× fields.

Statistical tests were done using Microsoft EXCEL or SAS (SAS Institute). A general linear random-effects MIXED model incorporating repeated measures was used to test for differences in rates of tumor growth. For continuous data, a Student's t test assuming unequal variance was used to compare groups. A Fischer's exact test was used to compare proportions. P < 0.05 was considered statistically significant.

Previously, we characterized the progressive transcriptome changes associated with HER2/Neu-induced mammary cancer using microarray analyses of wild-type mammary glands and mammary tumors in MMTV-c-Neu transgenic mice (19). Further analysis of this data revealed that message levels for several components of the PI3K and mTOR pathways were altered. Specifically, mRNA levels for the p85 subunit of PI3K and p70S6K were elevated ∼3- to 4-fold in tumors compared with wild-type glands (Fig. 1A). In contrast, mRNA levels for 4EBP1 were decreased ∼3-fold in tumors compared with wild-type controls (Fig. 1A). PI3K is an upstream activator of mTOR whereas p70S6K is a downstream target of both mTOR and PI3K. Elevated levels of these factors would be expected to promote activation of mTOR and its downstream targets. Specifically, p85, the regulatory domain of PI3K, initiates PI3K signaling by binding to phosphorylated sites on ErbB3, the preferred dimerization partner of ErbB2 (24), which is also highly expressed in these tumors (25). The concomitant up-regulation of ErbB2, ErbB3, and p85 would be expected to promote PI3K activation, and this is supported by the activation of AKT observed in these tumors (26). Additional factors that can alter mTOR activity whose expression was changed include phosphatase and tensin homologue (PTEN) and p90 ribosomal S6 kinase 1 (RSK1; ref. 27; Supplementary Fig. S1A).³

Human HER2-positive breast cancers have active signaling through the mTOR pathway (11). Consistent with this observation, we found that levels of phosphorylated and total mTOR were elevated in MMTV-c-Neu tumors, as compared with wild-type glands (Supplementary Fig. S1B).³ To determine if pathways downstream of mTOR were also activated in MMTV-c-Neu–induced mammary gland tumors, wild-type mammary glands and tumors were examined by immunohistochemical staining for phosphorylated S6, a downstream target of p70S6K. Strong cytoplasmic staining was observed throughout tumors (Fig. 1D) and in hyperplastic tissue surrounding tumors (Fig. 1C). Epithelial cells in wild-type tissues also had marked phosphorylation of S6, suggesting that activation of the mTOR pathway is a common component of mammary epithelial homeostasis in normal cells (Fig. 1B). These data confirm that mTOR signaling is active in MMTV-c-Neu tumors.

Activation of the mTOR pathway in c-Neu–induced tumors suggested that rapamycin, an inhibitor of mTOR, may be an efficacious chemotherapeutic for these tumors. To test this hypothesis directly, five tumor-bearing MMTV-c-Neu mice were treated every other day with 150 μg rapamycin for 32 days and five additional tumor-bearing transgenic mice were treated with vehicle. Treated mice showed no significant changes in weight (Supplementary Fig. S2A) or grooming behavior. In contrast, when this dose was given daily, substantial weight loss was observed within 4 days (data not shown). In mice treated with the nontoxic dosing paradigm, we measured external tumor volumes with calipers every 4 days. Two of the rapamycin-treated mice had two tumors, which were individually monitored. Among vehicle-treated mice, the external volumes of all tumors increased during treatment, with final volumes ranging from 2- to 7-fold greater than baseline (Fig. 2A). In sharp contrast, five of the seven tumors in rapamycin-treated mice were reduced between 1.2- and 3.3-fold in size after 32 days. Two rapamycin-treated tumors increased in external volume by 1.1- and 1.4-fold. The average final fold changes in tumor volume within the vehicle- and rapamycin-treated groups were 4.0 and 0.8, respectively (P = 0.001, MIXED model). Together, these data show that rapamycin inhibits or regresses primary tumor growth at doses that do not promote overt toxicity.

H&E-stained histologic sections of vehicle-treated tumors collected at the end of the study revealed that the tumors were composed predominantly of densely packed viable tumor epithelia (Fig. 2B). In contrast, rapamycin-treated tumors typically contained large, centrally located regions of either acellular material or grossly necrotic epithelia (Fig. 2B). To quantify these histologic differences, the cross-sectional area of tumor that was not grossly necrotic was expressed as a ratio of the total tumor area (Supplementary Fig. S2B).³ Approximately 92% of the tumor area in vehicle-treated tumors was composed of viable tissue, whereas the average proportion of viable tissue in rapamycin-treated tumors was 50% (P < 0.05, Student's t test). Thus, changes in the external volumes in rapamycin-treated c-Neu/ErbB2 tumors were accompanied by significant necrotic cavitation. Of note, the rapamycin-treated tumor that had an increase in external volume by 1.4-fold also had a necrotic center.

The MMTV-c-Neu mice develop tumors in areas of hyperplasia. To determine if sustained activation of mTOR was essential for maintenance of hyperplastic lesions, the histopathology of mammary glands surrounding tumors was evaluated (Fig. 2C and D). For the vehicle-treated tumors, 4 of 5 (80%) tumors were accompanied by hyperplastic mammary epithelium. In contrast, extensive depletion of this hyperplastic epithelium was observed in 5 of 5 (100%) rapamycin-treated mice (P < 0.05, Fischer's exact test).

The c-Neu/ErbB2 receptor is activated by phosphorylation subsequent to dimerization with other ErbB receptors (4). To exclude the possibility that rapamycin-induced regression was due to altered transgene expression or activation, phosphorylated c-Neu/ErbB2 was examined by immunohistochemistry. Activation of c-Neu/ErbB2 was indistinguishable between vehicle- and rapamycin-treated tumor epithelia (Supplementary Fig. S3A and B).³ Previous studies have reported that rapamycin increases phosphorylated Akt (P-Akt) levels with short-term treatment and down-regulates P-Akt with long-term exposure of tumor cells (28, 29). In contrast to these reports, we found that P-Akt levels were unchanged in treated MMTV-c-Neu tumors (Fig. 3A). Thus, rapamycin does not block expression of the MMTV-c-Neu transgene or alter activation of c-Neu/ErbB2 or its proximal target, Akt.

To verify that rapamycin blocked mTOR activity, levels of phosphorylated mTOR and ribosomal subunit S6 were evaluated. Rapamycin did not alter the relative levels of phosphorylated mTOR (Fig. 3B), suggesting that upstream signaling cascades remained intact. Whereas the relative levels of phosphorylated mTOR remained unchanged, the efficacy of rapamycin in blocking downstream mTOR signaling was clearly shown by examining phosphorylated S6 in tumor sections. Phosphorylated S6 was present throughout tumor epithelia in vehicle-treated tumors but was greatly diminished or absent in rapamycin-treated tumors (Fig. 3C and D). Thus, rapamycin-induced regression of tumors is associated with a loss of mTOR signaling, despite sustained activation of the c-Neu/ErbB2 receptor.

To determine whether inhibition of tumor growth and regression was due to cytostatic drug effects, proliferation rates were measured by quantifying levels of phosphorylated histone-H3, a nuclear marker of mitosis, in primary tumors that had been treated for 32 days. Rapamycin-treated tumors had a 49% reduction in the number of proliferating cells within the viable regions of the tumor, as compared with vehicle-treated tumors (P < 0.05, Student's t test; Fig. 4A and Supplementary Fig. S4A and B).³ One reported mechanism of rapamycin-induced inhibition of proliferation in cultured mammary epithelial cells is down-regulation of cyclin D1 (12, 13). This is particularly salient to c-Neu/ErbB2 tumors that are dependent on cyclin D1 for growth (8). An analysis of whole-cell lysates collected from tumors revealed that cyclin D1 expression was decreased in four of five rapamycin-treated tumors, with a mean reduction of 73% compared with untreated control tumors (Fig. 4B and C; P < 0.05, Student's t test). Thus, the ability of rapamycin to block proliferation in primary tumors may be due, in part, to down-regulation of cyclin D1 expression.

To characterize the progression of histologic changes underlying tumor cavitation, tumors were collected from mice treated with rapamycin every other day for 5, 10, and 32 days (Supplementary Fig. S4D–F).³ At 5 days, some treated tumors had developed small foci of cells with pyknotic nuclei. By day 10, numerous foci, many of which had coalesced into larger bands of dying cells, were present throughout much of the central portion of the tumors. By 32 days, only small islands of viable tumor cells were found in the central regions of the tumors, with the majority of the central core being composed of diffusely staining material. A closer examination of the tumor epithelia within the central necrotic core showed that they had features of ischemic necrosis (30). Specifically, “mummified” cells were observed (i.e., cells that were structurally intact but had lost their ability to retain staining due to protein and nucleic acid degradation).

To characterize the mechanism of cell death in tumors, histologic sections were examined for evidence of DNA fragmentation by TUNEL analysis. Whereas mummified cells in the rapamycin-treated tumors were not TUNEL positive, foci of TUNEL positivity were found along the perimeters of the necrotic zone as well as in the viable tissue regions. However, similar foci were also observed in vehicle-treated tumors (data not shown). Isolated TUNEL-positive cells were rare in both vehicle- and rapamycin-treated tumors, indicating that cell death characterized by TUNEL staining only occurred within large groups of tumor cells. The percentage of isolated TUNEL-positive cells was also not different between the two treatment groups (data not shown). Thus, TUNEL positivity could not be directly linked to rapamycin treatment. To ascertain whether dying cells were undergoing caspase-3–dependent apoptosis, levels of activated (cleaved) caspase-3 were measured in vehicle- and rapamycin-treated tumors by immunoblotting. There were no significant differences in the relative amount of activated caspase-3 after either 5 days (Fig. 4D) or 32 days (data not shown) of treatment, suggesting that the extensive cell death in rapamycin-treated tumors was not associated with caspase-3 activation. These data show that rapamycin induced rapid, focal cell death and that the majority of the dead or dying cells in rapamycin-treated tumors were TUNEL negative, did not have activated caspase-3, and manifested the characteristic features of ischemic necrosis.

mTOR signaling is activated in tumor endothelia by growth factors such as vascular endothelial growth factor, and its inhibition by rapamycin decreases proliferation of endothelial cells (31–34). Thus, tumor cell death and cavitation in MMTV-c-Neu mice could be the consequence of disrupted blood flow caused by the direct effects of rapamycin on vascular endothelia. Vehicle-treated tumor endothelial cells stained positively for phosphorylated S6 (Fig. 5A), indicating that the mTOR pathway is active in these cells. In contrast, S6 phosphorylation was considerably diminished with rapamycin treatment (Fig. 5B), indicating that this drug inhibits mTOR activity in endothelial cells within tumors.

One manifestation of disrupted endothelial signaling is decreased microvascular density due, in part, to an inability to form new vessels. To assess whether rapamycin treatment altered intratumoral vasculature in MMTV-c-Neu tumors, tumor sections from mice treated for either 5 or 32 days were immunostained for CD31, a cell adhesion and signaling molecule restricted to vascular endothelial cells (Fig. 5C and D). When average areas of CD31-positive staining were compared in vehicle- and rapamycin-treated mice (1.9% versus 1.2%; Supplementary Fig. S5),³ there was no difference between treatment groups (P = 0.25, Student's t test). Thus, although rapamycin blocked mTOR signaling in tumor endothelia, it did not induce a measurable change in microvessel density within the remaining areas of viable tumor tissue.

Tumors in MMTV-c-Neu mice spontaneously metastasize to the lung (21). To assess whether rapamycin treatment affected metastases, lungs were collected from mice treated with vehicle or rapamycin for 32 days. Like the primary MMTV-c-Neu tumors, metastases had prominent phosphorylation of S6 that was blocked by rapamycin treatment (Supplementary Fig. S6A and B).³ To determine whether rapamycin decreased proliferation of metastatic cells, these cells were immunostained for Ki-67, a nuclear protein present in proliferating cells (Fig. 6A and B). When compared with the vehicle controls, rapamycin caused a significant 3-fold decrease in proliferation of metastatic cells (P < 0.05; Fig. 6C). Similar to the primary tumors, no change in apoptosis was detected between the two groups using TUNEL staining or immunohistochemistry for cleaved caspase-3 (data not shown). Unlike the primary tumors, rapamycin treatment did not promote gross necrotic cavitation of lung metastases. These data show that rapamycin blocks proliferation of metastatic tumor cells but does not cause apoptotic cell death.

Breast cancer is the most common malignancy in women, comprising approximately a third of all newly diagnosed cancers (35). Critical to the advancement of effective treatment modalities for this disease is the development and utilization of biologically relevant cancer models to assess the effect and mechanisms of action of molecularly targeted anticancer therapies. Studies presented herein extend previous research on the effects of rapamycin in the c-Neu/ErbB2 mammary tumor models by assessing the effect of rapamycin on hyperplasia of the mammary gland as well as metastatic tumor cells (15, 16).

Rapamycin induced regression of primary tumors as a result of extensive caspase-independent cell death accompanied by a decrease in proliferation. We have previously shown that tissues surrounding tumors have focal hyperplasia, express active c-Neu/ErbB2 and specific markers of human HER2/ErbB2 tumors, and have a unique mRNA expression signature that clusters between wild-type mammary glands and tumors (19). We found that rapamycin treatment causes regression of this peritumoral tissue, indicating that, like the primary tumors, hyperplastic cells may also require mTOR signaling for growth and survival. These results reveal that mTOR dependency may occur early in the c-Neu/ErbB2–induced tumorigenic process, suggesting that treatment of hyperplastic lesions with rapamycin could prevent or delay primary tumor formation. In support of this concept, rapamycin prevented primary tumor formation in MMTV-NeuYD mice (16) as well as secondary tumor growth in MMTV-c-Neu mice (15). The presence of phosphorylated S6 levels in normal epithelial cells was unexpected because rapamycin does not negatively affect mammary gland development in early pubertal mice (36). These observations suggest that mTOR signaling may play a more significant role in malignant c-Neu/ErbB2–driven epithelial growth than in the normal mammary gland.

Like human HER2/ErbB2-positive tumors, MMTV-c-Neu tumors are highly metastatic (21, 37). Rapamycin treatment impeded mTOR signaling in lung metastases, and this blockade was associated with a significant reduction in proliferation. Whereas previous studies have suggested that rapamycin inhibits metastatic progression, they focused on the use of experimental metastases (38–41). The results presented herein indicate that rapamycin is efficacious at inhibiting proliferation of spontaneous distant metastases arising in an autochthonous tumor model. Because this is a spontaneous lung metastasis model with considerable intrinsic variability in the onset and extent of metastatic progression, an assessment of the effect of rapamycin on the formation and growth of metastatic lesions following primary tumor formation will require both extended treatment paradigms on large numbers of mice and quantitative in vivo imaging of metastatic tumors.

The effects of rapamycin on proliferation of malignant epithelial cells from primary tumors and lung metastases were similar. In contrast, rapamycin-induced cell death was only seen in primary tumors and occurred in a focal pattern. The focal nature of the epithelial cell death suggests that it may be due to factors extrinsic to epithelial cells. This reasoning is supported by studies characterizing the intrinsic effects of rapamycin on HER2/ErbB2 expressing mammary epithelial cells in culture, which have shown that rapamycin blocks proliferation but generally does not promote apoptosis (33).

Our findings related to rapamycin-induced cell death differ somewhat from those reported in a study of the effects of rapamycin treatment of transplanted polyoma middle T–induced tumors (MMTV-PyV-mT; ref. 42). In these mice, an early transient increase in apoptosis was observed in tumor cells after 3 h of rapamycin treatment but not after 7 days. These observations prompted the authors to conclude that apoptotic cell death is a transient event that only occurs early in the treatment paradigm. Whereas this explanation is consistent with the mode of action of the drug in the PyV-mT transplant model, it would not explain our observation that the vast majority of cell death occurred after the 5th day of treatment when no changes in apoptotic levels were observed. This difference may be due to the fact that, whereas PyV-mT recapitulates some of the signaling of HER2/ErbB2, it is not identical (43) and their associated tumors have distinct histopathologies (44).

One explanation for the extensive, but focal, cell death associated with rapamycin treatment in the MMTV-c-Neu mice is vascular dysfunction. Rapamycin blocked mTOR signaling in vascular endothelia in vivo, an effect that would be expected to make these cells refractory to growth factor stimulation (33, 34). Despite endothelial mTOR inhibition, however, rapamycin did not induce significant changes in tumor vessel density. This finding contrasts with an earlier study which showed that rapamycin decreased microvessel density in a Neu-YD transgenic mouse model (16). The courses of tumor progression in Neu-YD and MMTV-c-Neu mice are distinct (45), and the differences in signaling properties between the two Neu transgenes may underlie the differences in rapamycin regulation of microvessel density.

Rapamycin-induced cell death was observed as early as 5 days after initiation of treatment in the MMTV-c-Neu tumors and was pervasive by day 10. The rapid and focal nature of tumor cell death may reflect acute changes in established blood flow rather than attenuation of vessel proliferation. Rapamycin can promote thrombotic occlusion of tumor vessels (46, 47). We observed histologic evidence of vessel occlusion, including clogged and dilated vessels and dying cells with features of ischemic necrosis, in tumors and peritumoral regions that were similar to those described for other types of tumors treated with rapamycin (data not shown). However, unlike these other studies, we could not definitively link this pathology to the effects of rapamycin because similar features were observed in vehicle-treated tumors. One explanation for this difference is that our study used a spontaneous tumor model that develops tumors with long latency and has intrinsic intertumor variability whereas previous studies examining rapamycin-induced vessel occlusion involved orthotopic transplant models that rapidly generate clonal, relatively homogenous tumors. Thus, whereas progressive vessel occlusion is a highly plausible mechanism accounting for tumor cell death and could explain why cell death was not observed within the relatively avascular lung metastases, we cannot rule out the possibility that death was caused by other mechanisms.

In summary, rapamycin inhibits multiple stages of c-Neu/ErbB2–induced tumorigenesis in vivo. In addition to inducing regression of primary tumors, we found that rapamycin inhibits proliferation in spontaneous lung metastases and causes regression of peritumoral hyperplasia. These findings underscore the exquisite dependence of c-Neu/ErbB2 tumorigenic progression on intact mTOR signaling. This study used a mouse model that develops spontaneous mammary gland tumors due to aberrant up-regulation of a nonmutated c-Neu/ErbB2 oncogene, comparable to that found in human tumors. Consequently, our findings provide strong support for further evaluating the efficacy of rapamycin in patients with HER2/ErbB2-positive breast cancers, including those with early-stage and late-stage disease. Of note, clinical trials using rapamycin for the treatment of breast cancer have yielded promising, but mixed, results (48). This may be due to the heterogeneity of this disease. Our studies suggest that HER2/ErbB2-positive cancers may be particularly sensitive to the effects of rapamycin.

We thank Dr. Nick Ziats for providing reagents and for assistance with blood vessel staining, David Danielpour for thoughtful discussions on mechanisms of rapamycin action, and Kristen Lozada for excellent technical assistance. These studies were facilitated by the Histology and Gene Expression Core Facilities of the Case Comprehensive Cancer Center (grant P30CA43703).