The mammary gland is a unique organ that primarily develops after birth. Newborn female mice only possess rudimentary mammary ducts near their nipples, which begin to invade the surrounding subcutaneous white adipose tissues in the prepubertal period and occupy the entire fat pad during puberty. The tips of the invading ducts contain a unique structure referred to as the terminal end buds (TEBs), which occasionally bifurcates into branch ducts. In addition, some side branches sprout laterally from the main ducts. Additional side branches of a different character begin to develop in the luteal phase, and these side branches will become the milk-secreting acini. After weaning, the acini involute with extensive apoptosis, whereas other ductal structures remain in their prepregnant states (Lund et al.,1996; Hinck and Silberstein,2005; Lanigan et al.,2007).
It is known that ovarian and pituitary hormones, including estrogen and growth hormone, tightly regulate mammary development by affecting not only mammary epithelial cells, but also stromal cells (Ruan and Kleinberg,1999; Mallepell et al.,2006). Under this hormonal control, mammary epithelial cells grow and differentiate to form the duct and acinus, whereas mammary stromal cells surrounding the epithelial cells produce a variety of growth factors and extracellular matrix (ECM) components to facilitate epithelial growth and differentiation (Hovey et al.,1999; Sternlicht et al.,2006).
The mammary stromal cell population mainly consists of fibroblasts, adipocytes, and vascular endothelium. The number and size of the stromal cells, and possibly their functions, depend on the stage of mammary gland development (Richert et al.,2000). In addition, the composition and functions of the stromal cell population may be affected by nutritional status, such as obesity, which suggests that epithelial growth and differentiation are modulated by these factors. Indeed, diet-induced obesity in mice results in impaired lobuloalveolar development during pregnancy and reduced lactation (Flint et al.,2005). Similar obesity-induced lactation failure may also occur in humans and rats, because obese women (body mass index [BMI] > 29.0) have shorter breastfeeding periods than lean women (BMI < 26.0) (Kugyelka et al.,2004) and neonatal pups from obese rat dams are afflicted with a high fatality rate due to insufficient milk intake (Rasmussen et al.,2001). The mechanisms underlying obesity-mediated lactation failure, however, remain to be elucidated.
To date, whether obesity affects pubertal ductal development is unknown. Although genetically obese ob/ob and db/db mice show impaired ductal development (Hu et al.,2002), both mouse strains are sterile due to a deficiency in leptin signaling that disturbs the hypothalamic-pituitary-gonadal axis (Chehab et al.,1996; Yu et al.,1997). Therefore, a lack of hypothalamic-pituitary-gonadal hormonal regulation in these genetically obese mice certainly influences ductal development. To clarify the effects of obesity on mammary ductal development, we have examined the mammary glands of diet-induced obese mice, and found that obesity reduces the branching frequency and width of the ducts, which is accompanied by enhanced collagen deposition and an abnormal myoepithelium.
As shown in Figure 1, mice given the HFD for 16 weeks gained approximately 10 g more weight than the mice fed the ND. The obese mice exhibited an increase in the weights of the abdominal and thoracic mammary glands (subcutaneous adipose tissue), mesenteric adipose tissue, and liver, whereas no change was observed in the skeletal muscle (Table 1). Concomitantly, the obese mice showed hyperglycemia, hyperinsulinemia, hyperlipidemia, and hyperleptinemia, whereas plasma adiponectin levels were not affected (Table 1).
Table 1. The Effects of the ND and HFD on Body Weight, Tissue Weights, and Plasma Levels of Nutrients, Insulin, and Adipocytokines
Diet-Induced Obesity Caused Abnormal Mammary Ductal Development in Nonpregnant Mice
To determine whether diet-induced obesity influenced mammary ductal development in nonpregnant mice, whole mammary glands were stained with carmine dye and examined. As shown in Figure 2A, mammary glands from 20-week-old mice fed ND contained a dense distribution of ducts, including side branching, and clear swollen tips. In contrast, mammary glands from HFD-fed mice were enlarged, reflecting an increase in adipose tissue mass. Moreover, the less densely distributed ducts were elongated with some side branching until they reached the border of the fat pad. Quantitative comparison of the ducts from obese mice with those from lean mice demonstrated that the total ductal length in the HFD-fed mice (686 ± 49 mm; n = 5) was significantly longer than that in the ND-fed mice (462 ± 37 mm, n = 5), whereas the numbers of branches were similar in the HFD-fed and ND-fed mice (498 ± 63 and 461 ± 8, respectively; n = 5 for each group). Accordingly, branching frequency calculated by dividing the branching number by the ductal length was reduced in the HFD-induced obese mice (Fig. 2B). Moreover, the width of the ducts near the swollen tips of HFD-fed mice was smaller than that of the ND-fed mice (Fig. 2C).
Because the ducts in the obese mice seemed poorly developed, we further examined them in a high-power field (Fig. 3A). Although the width of the ductal lumen in the HFD-fed mice was almost the same as that in the ND-fed mice (Fig. 3B), the thickness of the ductal epithelium layer in the HFD-fed mice was significantly smaller than that of the ND-fed mice (Fig. 3B). The mammary ducts were composed of two epithelial layers: the luminal epithelium and the basal myoepithelium. The latter was clearly labeled with the antibody specific for α-SMA (Fig. 3C). When we compared the ducts from mice fed the ND with those from the mice fed the HFD, almost 60% of the ducts in the HFD-fed mice were not lined completely with α-SMA-positive cells, whereas most ducts from the ND-fed mice were completely lined (Fig. 3C and D), suggesting that the decrease in the number of myoepithelial cells in the obese mice was the basis of the decreased ductal width observed in Figure 2C.
Diet-Induced Obesity Resulted in Abnormal Interstitial Cells and an Altered Extracellular Environment in the Mammary Glands of Nonpregnant Mice
We next examined the effects of diet-induced obesity on the interstitial cells and surrounding extracellular environment. As expected, adipocytes, which occupied the majority of the interstitial region, were enlarged in HFD-fed mice, compared with those of mice fed the ND (Fig. 3A). Moreover, the eosinophilic layer surrounding the ducts was thicker in the mice fed the HFD than in the mice fed the ND (Figs. 3A and 4C). This eosinophilic layer was also stained using Masson's trichrome method (Fig. 4A) and anti-type I collagen antibody (Fig. 4B), indicating that the layer was composed of fibrous collagen, the deposition of which was augmented in mice fed the HFD.
Carmine dye stained the mammary ductal epithelium intensely, whereas it faintly marked such stromal cells as fibroblasts and adipocytes. Unlike the mice fed the ND, however, mammary glands from the HFD-fed mice contained a number of cell aggregates in the stromal region that were densely stained with carmine (Fig. 5A; 72.8 ± 29.9 aggregates per gland). The aggregates, which resembled “crowns,” were composed of macrophages that had been recruited into the adipocyte tissue as previously reported (Weisberg et al.,2003). This was confirmed by the staining with antibody against F4/80, a marker of the macrophage (Fig. 5B). We then examined the mRNA expression of the proinflammatory cytokines IL-6 and TNFα. Although both genes were rarely expressed in the mammary glands of ND-fed mice, TNFα mRNA expression, but not IL-6 mRNA expression, tends to increase in all of the mice fed the HFD (Supp. Fig. S1, which is available online). Therefore, a small but significant number of macrophages recruited into the stromal region is possibly indicative of an inflammatory reaction surrounding tissue in the mammary gland from the HFD-fed mice.
We also examined the mRNA expression of some cytokines that affect mammary epithelial growth and functions (Bonnette and Hadsell,2001; Pollard,2001). Conventional and real-time quantitative PCR analyses showed that HGF, TGF-β1, and IGF-1 mRNA were detected in the mammary gland of ND-fed mice, but their expression was not significantly changed in the mice fed the HFD (Supp. Fig. S1). Gene expression analysis, however, revealed that leptin mRNA expression was significantly higher in the mammary glands of obese mice (Fig. 6A), which corresponded with a higher plasma leptin concentration (Table 1). In addition, leptin receptors were localized on the basal and luminal epithelial cells and periductal fibroblasts in the mammary glands of both ND- and HFD-fed mice (Fig. 6B), suggesting that leptin is the stromal-derived factor that affects mammary epithelial functions directly and causes the ductal abnormalities mentioned above. To test this hypothesis, BMEC proliferation was assessed in the absence or presence of leptin. As shown in Figure 6C, leptin reduced the cell number on day 4 (59.3 ± 8.6% of control samples; P < 0.05).
In the present study, we have demonstrated that diet-induced obesity in nonpregnant mice results in a typical nutritional disorder phenotype, including enlarged abdominal mammary glands (fat pads). Moreover, the obese mice display abnormal mammary ductal development, that is, reduced branching frequency and mammary duct width, a partial deficit in the myoepithelium, and increased collagen deposition surrounding the ductal epithelium (Figs. 2–4). Therefore, obesity-induced abnormalities in ductal development, especially incomplete lining of the luminal epithelium with myoepithelium, might lead to retardation of lobuloalveolar formation after pregnancy, and lactation failure because the myoepithelium plays an important role in the maintenance of ductal structures for milk ejection (Faraldo et al.,2005).
It has been suggested that stromal functions altered by obesity are one of causative mechanisms for impaired the ductal development, because obesity markedly alters the expression profile of adipocytokines and proinflammatory cytokines in adipose tissues (Ronti et al.,2006). Indeed, the mammary glands of obese mice fed the HFD contained larger adipocytes and more infiltrating macrophages (Figs. 3 and 5). However, among the factors tested, obesity only markedly increased leptin expression (Fig. 6A). Interestingly, leptin receptors were present on both the basal and luminal epithelial cells (Fig. 6B), and leptin directly inhibited the growth of BMEC (Fig. 6C) and HGF-induced BMEC ductal morphogenesis in three-dimensional collagen gels (Yamaji et al.,2007). Such leptin-dependent inhibition of cell proliferation is also reported in noncancerous mouse mammary epithelial cell lines as well as bovine cells (Silva et al.,2002; Baratta et al.,2003; Motta et al.,2007), although there is a report showing that leptin stimulates proliferation of one of the human breast cancer cell lines (Dieudonne et al.,2002). Thus, leptin increased in obesity likely plays a suppressive role in ductal growth, which resulted in the reduced ductal width and incomplete lining of the luminal epithelium with myoepithelium (Figs. 2 and 3). Furthermore, it is worth noting that ob/ob mice suffer several reproductive and endocrine abnormalities, which, with the exception of lactation, can be rescued by leptin administration (Malik et al.,2001). This result suggests that leptin replacement in ob/ob mice may inhibit lactation by directly affecting mammary epithelial cells.
In human hepatic stellate cells, leptin increases collagen production (Choudhury et al.,2006) and suppresses matrix metalloproteinase (MMP)-1 mRNA expression, an enzyme involved in the degradation of type I collagen fibril (Cao et al.,2007). Thus, increased leptin levels in obese mice likely act on the leptin receptor–expressing fibroblasts present in the periductal region (Fig. 6B), and promote collagen production and deposition as shown in Figure 4. In addition to leptin signaling, macrophages may be required for ductal outgrowth through a promotion of collagen fibrillogenesis around the TEB to facilitate the formation of the bud during pubertal ductal development (Ingman et al.,2006). Therefore, excess recruitment and activation of macrophages possibly contribute to the deposition of fibrillar collagen around the ducts observed in this study.
Obesity-induced abnormal deposition of collagen and/or increased levels of other ECM components, such as heparan sulfate proteoglycans, may be a reason for reduction of branching frequency (Fig. 2). This is supported by the fact that deletion of MMP-3, an ECM remodeling ectoenzyme expressed in periductal stroma and adipocytes, decreases the number of side branching ducts in the mammary gland (Wiseman et al.,2003). Such ECM remodeling in the obese mice is also a likely indirect mechanism to suppress myoepithelium proliferation, in addition to the direct suppressive effect of leptin mentioned above, because the proliferation of human mammary myoepithelial cells has been shown to depend on heparan sulfate proteoglycans (Sergeant et al.,2000).
In summary, we have shown that obesity suppresses mammary duct development in nulliparous nonpregnant mice, presumably through a remodeling of the mammary microenvironment and paracrine factors, such as leptin. Although additional studies are required to further elucidate the underlying mechanisms, our results have revealed some of the important roles of the local microenvironment during mammary ductal development.
Animals and Tissue Preparations
In this study, experimental procedures and animal care were carried out in accordance with the Guidelines of Animal Care and Use from Hokkaido University, and were approved by the University Committee for the Care and Use of Laboratory Animals. Female 3-week-old C57BL/6J mice (n = 10) were obtained from Nihon SLC (Shizuoka, Japan), housed under specific pathogen-free conditions at 24°C with a 12 h:12 h light: dark cycle, and given food and water ad libitum. One week after acclimation, mice were randomly divided into two groups (n = 5), which were fed either a high-fat diet (HFD: 507.6 kcal/100g [fat kcal 56.7%], Clea, Tokyo, Japan) containing 24.5 wt% unsaturated and 7.1 wt% saturated fatty acid, 25.5 wt% protein, and 2.9 wt% fiber or a normal diet (ND: 343.1 kcal/100g, Clea) containing 4.8 wt% fat, 25.1 wt% protein, and 4.2 wt% fiber. The body weights of the mice were measured every 6 days and the mice were sacrificed using cervical dislocation 16 weeks later. Blood was collected from the carotid artery into a heparinized tube. Plasma was immediately separated by centrifugation (12,000g, 3 min, 4°C) and stored at −80°C until it was used in an assay. Bilateral abdominal mammary glands (the fourth mammary gland) were excised and weighed. The right gland was spread on a glass slide for whole-mount staining, whereas half of the left gland was preserved at −80°C in RNAlater (Ambion, Austin, TX) and the other was in 10% formalin after removing the lymph node.
Measurement of Plasma Parameters
Plasma concentrations of glucose, nonesterified fatty acid (NEFA), insulin, leptin, and adiponectin were measured with the following commercial kits: glucose-B and NEFA-C tests (Wako Pure Chemical, Osaka, Japan), mouse insulin and leptin ELISA kits (Morinaga Bioscience Institute, Yokohama, Japan), and an adiponectin/Acrp30 immunoassay kit (R&D systems, Minneapolis, MN).
Whole-Mount Examination of Mammary Gland
Mammary glands were spread on glass slides, fixed in Carnoy's fixative (absolute ethanol, chloroform, and glacial acetic acid at a volume ratio of 6:3:1), washed with 70% ethanol, and rehydrated with decreasing concentrations of ethanol followed by distilled water. The gland was stained overnight at 4°C in carmine alum (0.2% carmine dye in 0.5% aluminum potassium sulfate), and dehydrated again with a series of 70, 80, 90, and 100% ethanol, and then with xylene. The gland was photographed with a digital camera under a dissecting microscope, and the image was analyzed using Photoshop (Adobe Systems, San Jose, CA) and NIH Image software. The length of the ducts and the number of branches were determined by tracing all of the ducts in the entire mammary gland. Branching frequency was calculated by dividing the number of branches by the length of the ducts. The duct width was measured at the “necks” of the swollen tips (see Fig. 2A, arrow) with approximately 200–370 branches per gland; the mean value was defined as the individual ductal width. Additionally, the crown-like structures found exclusively in the stromal region of the carmine-stained whole-mount glands from mice fed the HFD were counted.
The carmine-stained, whole-mount mammary gland was removed from the glass slide and embedded in paraffin. The gland was cut into 5-μm-thick sections to examine the duct along the long axis, and the sections were stained with hematoxylin and eosin (HE) or Masson's trichrome stain for collagen fibers. Pictures were obtained, and all ducts cut longitudinally in a section (5–13 ducts per a section) were selected. The thickness of the ductal epithelium layer and the layer surrounding the duct, and the width of the lumen, were measured for 46–50 ducts in each experimental group.
Localization of the myoepithelium was determined by immunohistochemical visualizing α-smooth muscle actin (α-SMA) using a Histfine SAB-PO kit (Nichirei, Tokyo, Japan). Briefly, the section was prepared from the carmine-stained, whole-mount mammary gland. Deparaffinized and rehydrated sections were sequentially incubated in methanol containing 3% H2O2 for 30 min, 10% rabbit normal serum for 1 hr, PBS with or without anti-α-SMA mouse monoclonal antibody (1:10,000 dilution in PBS; Sigma-Aldrich, St. Louis, MO) overnight at 4°C, biotin-conjugated rabbit anti-mouse IgG solution (1:30 dilution) for 1 hr, peroxidase-conjugated streptavidin solution (1:3 dilution) for 1 hr, and finally diaminobenzidine as a substrate for 40 sec. The sections were counterstained with hematoxylin. α-SMA expression in 20 ducts per mouse were examined in a blinded manner to determine whether the luminal epithelia of the ducts were completely or incompletely lined with α-SMA-positive cells.
Infiltration of macrophages was confirmed in the formalin-fixed paraffin-embedded gland by immunohistochemical staining with rat anti-F4/80 monoclonal antibody (1:200 dilution, AbD Serotec, Oxford, UK) and a Histfine MAX-PO kit (Nichirei).
Expression of leptin receptors (Ob-Rs) and collagen type I were examined in the section prepared from the carmine-stained, whole-mount mammary gland, by using rabbit anti-Ob-Re polyclonal antibody (1:50 dilution, Morinaga Bioscience Institute) and rabbit anti-collagen type I antibody (1:1,000 dilution, Rockland, Gilbertsville, PA), respectively.
Analysis of mRNA Expression
Total RNA was extracted from mammary glands using RNAiso (Takara Bio, Shiga, Japan) according to the manufacturer's protocol. RNA (2 μg) was treated at 70°C for 5 min and reverse transcribed using 100 U of Moloney murine leukemia virus reverse transcriptase (Invitrogen, Carlsbad, CA), 50 pmol of poly(dT) primer, and 20 nmol of dNTPs in a total volume of 20 μl at 37°C for 1 hr. After heating at 94°C for 5 min, PCR amplification was performed with 0.025 U of Taq polymerase (Ampliqon, Herlev, Denmark), 3 mM MgCl2, and 50 pmol of forward and reverse primers specific for the respective genes in a total volume of 50 μl. The primer pairs and PCR conditions for mouse leptin, tumor necrosis factor-α (TNFα), interleukin-6 (IL-6), insulin-like growth factor 1 (IGF-1), hepatocyte growth factor (HGF), transforming growth factor-β1 (TGF-β1) and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) are summarized in Table 2. After electrophoresis in a 2% agarose gel, the PCR products were stained with ethidium bromide.
PCR primers and conditions used in the present study are summarized. The first row of each column shows the gene name, GenBank accession number, and PCR product size, whereas the second row describes the PCR conditions. Annealing temperature and time, and cycles of the conventional PCR cycle are shown in parentheses, while annealing temperature and time, extension time, and temperature for fluorescence measurement of the real-time PCRs are given in square brackets. Denaturation and extension during the conventional PCRs were performed at 94°C for 30 sec and 72°C for 30 sec, respectively, whereas denaturation during real-time PCRs was performed at 94°C for 15 sec. The third and fourth rows show forward and reverse primer sequences for the respective genes.
To quantify the levels of leptin, HGF, and TGF-β1 mRNA expression, real-time PCR was performed with a fluorescence thermal cycler (Light Cycler System, Roche Diagnostics, Basel, Switzerland) using 0.5 μM of each primer (Table 2) and the respective cloned genes as standards. The fluorescence of SYBR Green (Qiagen, Hilden, Germany) at 530 nm was recorded at the end of the extension phase and analyzed using Light Cycler Software (Version 3). The level of β-actin mRNA was also determined as an internal control.
Cell Proliferation Assay With Bovine Mammary Epithelial Cells (BMECs)
BMECs were prepared as described previously (Yamaji et al.,2007); more than 95% of the purified cells were labeled with anti-cytokeratin antibodies, an epithelial marker. To further enrich the epithelial cell population, BMECs were embedded in a type I collagen gel (0.6 mg/ml Cellmatrix type I-A; Nitta Gelatin, Tokyo, Japan) and cultured in growth medium (Dulbecco's modified Eagle's medium [DMEM]/Ham's F-12 [1:1] [Sigma], supplemented with BSA [2 mg/ml; Sigma], penicillin [100 units/ml; Meiji Seika, Tokyo, Japan], streptomycin [100 μg/ml; Meiji Seika], gentamicin [10 μg/ml; Invitrogen], amphotericin B [250 ng/ml; Invitrogen], holotransferrin [10 μg/ml; Invitrogen], insulin [10 μg/ml; Sigma], cortisol [5 μg/ml; Sigma], epidermal growth factor [10 ng/ml, Wako], and cholera toxin [10 ng/ml; Wako]) at 37°C in a humidified atmosphere containing 5% CO2 in air for one week. To perform the cell proliferation assay, BMECs were freed from the collagen gel by treating them with M199 medium (Sigma) containing 1% collagenase and 0.1% soybean trypsin inhibitor, after which the cells were dispersed using trypsin-EDTA. The cells (1 × 103 cells) were then cultured in 96-well plates coated with type I collagen (Cellmatrix type I-C, Nitta Gelatin) and 200 μl of the growth medium overnight, and further cultured in a basal medium (DMEM/F-12 with antibiotics, BSA, holotransferrin, and cholera toxin) either in the absence or presence of 5 nM (80 ng/ml) leptin (PeproTech, London, UK; n = 4, each experiment was performed in triplicate) for a 4-day period. The number of cells was assessed every day using a Cell Titer-Glo luminescent viability assay kit (Promega, Madison, WI).
Data are expressed as means ± S.E.M. (standard error of the mean), and were analyzed using the unpaired two-tailed Student's t-tests. P values less than 0.05 were considered to be statistically significant.
The work was supported in part by JSPS Research Fellowships for Young Scientists to O. I. and Y. O.-O.