TEBs during puberty, thereby demonstrating the importance of Bim for ductal morphogenesis. By contrast, no major involution defects are observed in Bim null mammary glands after forced weaning, suggesting that the apoptotic involution mechanism is distinct from luminal clearing during development. Over-expression of BCL- 2 was found to partially suppress body cell apoptosis and to disrupt TEB structure. However, lumen formation was not inhibited, and mature virgin mammary glands developed normally in this transgenic model. In mature females, antiapoptotic BCL-2 and BCL-W mRNA and protein are downregulated during involution, whereas proapoptotic BAX, BAK, and BAD proteins are upregulated during lactation and early involution. The relative levels of Bcl-xS/Bcl-xL mRNAs also increase at the onset of involution. Studies in transgenic mice demonstrate that several BCL-2 family members contribute to the molecular control of apoptosis in the involuting breast. Over-expression of BCL-2 in wap-Bcl-2 transgenic mice inhibits alveolar cell apoptosis during involution. Conditional deletion of Bcl- x in the mouse mammary epithelium results in accelerated apoptosis during involution but does not compromise mammary function during lactation. Last but not least, disruption of Bax in the mammary epithelium reduces apoptosis levels during the first stage of involution but does not affect the second phase of involution.

p53, a well-known transcription factor that regulates the expression of several BCL-2 family members (Bax, Bak, Noxa, Puma, Bcl-2), is upregulated at the onset of involution. Its disruption delays involution, highlighting a physiologic role in this process. Investigators wondered whether p53 could be involved in Bax upregulation in the involuting breast. However, if p53 does regulate the transcription of the cell cycle inhibitor p21, it does not seem to induce Bax. The role of upregulated p53 in inducing proapoptotic Noxa and Puma or in downregulating Bcl-2 has not been investigated.

Altogether, the current literature shows that BIM is required for lumen clearance during ductal morphogenesis, whereas BCL-2, BCL-X, and BAX are important for the molecular control of apoptosis during involution.

In summary, apoptosis in the normal breast is under the molecular control of several extracellular and intracellular factors. The BCL-2 family member BIM is required for lumen clearance during epithelial morphogenesis. Involution is the result of the coordinated regulation of a complex network of proteins (Figure 23-2). Milk accumulation causes an alveolar stretch, leading to the disruption of prosurvival cell-matrix and cell–cell

adhesions. Truncation of the β-catenin binding domain of E-cadherin was shown to precede epithelial apoptosis in early mammary involution. Local milk stasis also induces the expression of several proapoptotic cytokines

– LIF, TGFβ3, and death ligands – that trigger apoptosis through the death receptor pathway and the STAT3 pathway. Survival pathways such as IGF and the PI3K/AKT signaling pathway are inhibited, whereas downstream targets are upregulated to ensure the transition to the second phase. For instance, dimeric transcription factors AP-1 and macrophage markers are upregulated at the end of the first phase to initiate the shift to the second phase of involution. AP-1 is known to regulate the transcription of the matrix metalloproteinase stromelysin-1 (MMP-3), whereas macrophage markers help recruit macrophages for the phagocytosis of apoptotic bodies. In the second phase of involution, activated matrix metalloproteinases degrade the ECM and finally trigger the massive anoikis of the remaining secretory alveoli.

3. APOPTOSIS IN BREAST CANCER

Disruption of balance between cell death and proliferation is considered a major factor in the growth of tumors or their regression during therapy. This balance can be disrupted in two ways in tumors: by increasing proliferation and/or decreasing apoptosis. There is evidence that tumor growth results from both uncontrolled proliferation and reduced apoptosis. In premalignant stages, major alterations in apoptosis, cell proliferation, and cell cycle regulators would arise, allowing the later progression of the disease.

The susceptibility of the mammary gland to tumorigenesis is influenced by its development particularly during puberty and pregnancy, when marked changes in cell proliferation, invasion, differentiation, and apoptosis occur. Indeed, terminal ducts that are highly proliferative in early adulthood are all the more susceptible to carcinogen exposure at that period. Moreover, the process of involution co-opts some of the programs of wound healing, creating a proinflammatory stroma that can promote tumor progression and that explains the high rate of metastases reported in pregnancyassociated breast cancer. In fact, the developing breast shares many properties (proliferation, invasion, angiogenesis, proinflammatory stroma) with breast cancer, and many signaling pathways that regulate processes such as invasion, proliferation, or apoptosis in the normal breast can be corrupted by tumor cells to their own advantage.

The extraordinary developing capacity of the normal breast underlies its great susceptibility to tumorigenesis and may explain why breast cancer is the most common type of nonskin cancer and the second leading cause of cancer death in American women. Most breast cancers arise from the epithelium in the undifferentiated terminal duct lobular unit, leading to cancer of the ducts (ductal carcinoma, approximately 90% of breast carcinomas) or cancer in the milk-producing glands (lobular carcinoma, approximately 10% of breast carcinomas). The development of breast cancer has been described as a multistep process with progressive phenotypic changes from hyperplasia with or without atypia through in situ carcinoma to invasive carcinoma capable of invading surrounding tissues and eventually metastasizing. The role of apoptosis in breast carcinogenesis and progression has been the focus of many investigations, and the main results are discussed next.

3.1. Apoptosis in breast tumorigenesis and cancer progression

Apoptosis status (occurrence, apoptotic rates, molecular regulation) has been analyzed at different stages of breast cancer development, in 3D culture systems, murine models, and patient samples. The number of apoptotic cells as a percentage of cells present, or the number of apoptotic cells per square millimeter of neoplastic tissue, is usually described as the apoptotic index (AI), as opposed to the mitotic index (MI), the percentage of proliferating cells.

In contrast with normal breast, premalignant breast cancer lesions fail to respond to normal apoptotic stimuli for lumen clearance and mammary involution. Indeed, hyperplasia with atypia and carcinoma in situ are characterized by a complete or partially filled lumen. Debnath et al. used a 3D culture of the mammary epithelial cell line MCF-10A to investigate the importance of enhanced proliferation versus apoptosis inhibition for lumen filling. Neither enhancing proliferation (by over-expressing mitogenic oncoproteins) or inhibiting apoptosis (by over-expressing BCL-2 or BCLXL) was sufficient to induce lumen filling. By contrast, oncoproteins such as ERBB2 and IGF-1R that simultaneously promote proliferation and prevent apoptosis induced lumen filling. ERBB2 was shown to prevent normal luminal apoptosis by downregulating BIM. Therefore, in 3D culture models, enhanced proliferation requires a concomitantly blocked apoptosis to cause neoplasia. Consistently, hyperplasia implants in mice are also unresponsive to normal apoptotic signals during mammary gland involution and fail to regress upon

forced weaning. Phosphorylated AKT1 and BCL-2 protein levels are higher in those hyperplasias than in the normal regressed mammary gland, suggesting that inhibition of cell death creates a permissive cellular environment for neoplastic transformation. This inhibition of apoptosis is consistent with reduced AI in a carcinogenesis model in rats. In this animal model, mammary tumors are preceded by hyperplastic and premalignant lesions arising mostly in TEBs, as well as in ducts and alveoli. Quantification of MI and AI showed that the percentage of proliferating cells is similar in TEBs to those in terminal end bud hyperplasia (TEBH), carcinomas in situ (CIS), and carcinomas, whereas the percentage of apoptotic cells (AI) is relatively high in TEBs and decreased in TEBH, CIS, and carcinomas. This indicates that, in this model, neoplastic transformation of mammary epithelial cells in TEBs is not associated with an increase in cell proliferation, but rather with a decrease in apoptotic cell death. In patients, reduced apoptosis is also detected in noninvolved tissue from cancer-containing breasts when compared with agematched benign tumors and normal breast tissue from women without cancer after menopause. Hyperplasias are thus associated with reduced apoptosis when compared with normal tissue both in mouse models and in the normal surrounding tissue of breast tumors. The reduction in apoptosis may lead to the preservation of genetically aberrant cells, hence favoring neoplastic development.

Whereas hyperplasia formation requires reduced apoptosis, malignant progression from hyperplasia to invasive carcinoma is usually associated with an increase in both cell proliferation and apoptosis. In breast cancer, high AIs have been correlated with several pathologic parameters, such as high MI, high tumor grade, lack of tubule formation, tumor necrosis, absence of BCL-2 and estrogen receptor (ER) expression, expression of p53, and poor overall survival. Rates of apoptosis are thus related to tumor grade and are higher in more aggressive tumors that exhibit higher rates of proliferation. The current hypothesis is that apoptosis may help selecting clonal subpopulations with high growth potential during breast cancer progression.

Discordant results are found for the transition from in situ carcinoma to invasive carcinoma. Some investigators reported a reduction in AIs from ductal carcinomas in situ to invasive carcinomas. These discrepancies could come from the different methods used for apoptosis detection (microscope counting or terminal deoxynucleotidyl transferase dUTP nick end labeling [TUNEL]) or from heterogeneous sample cohorts (age and number of patients, tissue differentiation, etc.). Mommers et al.