SirT1 modulates the estrogen–insulin-like growth factor-1 signaling for postnatal development of mammary gland in mice

A previously described SirT1 targeting construct, KOII [23], was used to generate mice harboring a conditional targeted mutation in the SirT1 gene (SirT1^co/co mice) (see Additional file 1). The breeding of SirT1^co/co mice and CMV-Cre transgenic mice results in mice harboring a germline-transmitted deletion of exon 4 of the SirT1 gene (SirT1^+/ko mice). Both SirT1^co/co mice and SirT1^+/co mice were used to establish breeding colonies for generating SirT1^co/co and SirT1^ko/ko mice, respectively. Both SirT1^co/co mice and SirT1^ko/ko mice were in a mixed 129SvJ/C57B6 background. Mice were housed in a special-pathogen-free facility and all procedures were approved by the University of Washington Animal Care and Use Committee. A PCR-based genotyping method was established to identify the wild-type, co, and ko loci of the SirT1 gene using three primers: 5' co primer, 5'-GGTTGACTTAGGTCTTGTCTG; 5' ko primer, 5'-AGGCGGATTTCTGAGTTCGA; 3' primer, 5'-CGTCCCTTGTAATGTTTCCC. Murine embryonic fibroblasts (MEFs) were isolated from the embryos between embryonic day 12.5 and embryonic day 14.5. The body weight was measured once a week.

A commercial radioimmunoassay (ALPCO Inc., Windham, NH, USA) was used. Samples (300 μl) were prepared via an alcohol extraction followed by a 2-day disequilibrium incubation at 4°C. Samples were measured in duplicate and data were calculated as the samples were counted. Six standards ranging from 10 to 500 ng/ml were used to generate the standard curve. The lower limit of detection was 10 ng.

To determine the sensitivity of embryonic stem cells after ionizing radiation or hydrogen peroxide treatment, 300 embryonic stem cells were seeded onto a 10-cm plate for 24 hours. The plates were either exposed to a ¹³⁷Cs source at indicated doses or were treated with embryonic stem media containing hydrogen peroxide at indicated concentrations for 30 minutes. Seven days after irradiation or hydrogen peroxide treatment, the embryonic stem colonies stained with crystal violet were counted.

Estrogen pellets in the form of 17β-estradiol, at 0.1 mg, and the 21-day releasing time were obtained from Innovative Research (Sarasota, FL, USA). One pelletwas subcutaneously implanted in the subscapular region ofindividual female mice using a sterilizedmetal trochar (Innovative Research).

The inguinoabdominal mammary fat pads were spread on microscope slides, fixed in Carnoy's fixative overnight, hydrated and stained with carmine alum stain (Sigma, St. Louis, MO, USA) overnight, and were then dehydrated, treated with xylene to remove fat, and mounted with Secure Mount (Fisher Scientific, Pittsburgh, PA, USA) and cover slips.

The mice were given a single dose of bromodeoxyurdine (BrdU) at 100 g/kg body weight via intraperitoneal injection 2 hours before euthanasia. The mammary fat pads, as well as other tissues, were collected and fixed in 10% formalin solution (Fisher Scientific). Paraffin-embedded sections were prepared at 4 μm thickness followed by standard H & E staining for histological analysis and by immunohistochemical staining for BrdU-labeled cells (Sigma) and apoptotic cells (Promega, Madison, WI, USA). To detect the presence of mouse milk proteins, a rabbit anti-mouse milk specific protein antibody (Nordic Immunology, Tillburg, The Netherlands) was used for immunofluorescent staining, followed by Texas Red anti-rabbit secondary antibody (Molecular Probes) and DAPI staining for nuclei (Molecular Probe, Eugene, OR, USA). Microscopic analyses of all histological findings were carried out on an AxioVert 200M microscope with AxioVision 4.5 software (Carl Zeiss, München-Hallbergmoos, Germany).

Protein extracts were prepared from mammary tissues as well as from MEFs. The following primary antibodies were used: anti-SirT1 (Upstate Biotechology, Lake Placid, NY, USA), anti-IGF-1 binding protein-1 (IGFBP-1) (R&D Systems, Minneapolis, MN, USA), and anti-actin and anti-IκBα (Santa Cruz Biotechnology, Santa Cruz, CA, USA). The corresponding secondary antibodies were horseradish peroxidase-conjugated anti-rabbit (Pierce, Rockford, IL, USA), anti-rat, and anti-goat (Santa Cruz Biotechnology).

The conditional targeted SirT1 mutant mice (SirT1^co/co mice) carry an insertion mutation of the neomycin-resistant gene and lox sequences in the SirT1 gene flanking exon 4 that encodes a conserved Sir2 motif (Figure 1a). The mutation does not affect the expression of SirT1 in SirT1^co/co mice (Figure 1b). As expected, SirT1^co/co mice are phenotypically indistinguishable from wild-type mice. To convert the SirT1 co allele into the SirT1 ko allele, SirT1^co/co mice were crossed with CMV-Cre transgenic mice to generate SirT1 heterozygotes carrying the SirT1^+/ko, CMV-Cre⁺ genotype. The expression of the CMV-Cre transgene catalyzes the deletion of exon 4 in most lineages of cells, including germ cells (Figure 1a). This SirT1 ko allele should be identical to the previously described Δex4 mutation [23]. SirT1^+/ko, Cre⁺ mice were backcrossed with wild-type mice to generate the mice harboring a germline-transmitted deletion mutation (that is, SirT1^+/ko mice). The breeding of SirT1^+/ko male mice and SirT1^+/ko female mice resulted in SirT1^ko/ko mice. The cells derived from either SirT1^+/ko mice or SirT1^ko/ko mice expressed a SirT1 mutant protein due to the inframe deletion of exon 4 (Figure 1b). This SirT1 mutant protein may not be functional, however, since SirT1^+/ko mice are phenotypically indistinguishable from wild-type mice and SirT1Δ^ex4/Δex4 mice were phenotypically identical to SirT1 null mice [23].

Similar to other strains of SirT1-deficient mice, nearly two-thirds of SirT1^ko/ko newborns die shortly after birth and the majority of surviving SirT1^ko/ko mice manifest growth retardation (Table 1 and Figure 1c) [22, 23]. Growth retardation may result from a systemic defect in hormonal regulation, DNA double-strand break repair, or other causal mechanisms. We found that SirT1^ko/ko mice have reduced levels of serum IGF-1 (Figure 1d). IGF-1 often acts as a local effector for pituitary GH. The serum level of GH in SirT1^ko/ko mice, however, appeared to be within the normal range (data not shown). On the other hand, yeast Sir2 and its associated Sir complex, as well as mouse SirT6, have also been implicated in DNA double-strand break repair and the maintenance of chromosome stability [32‐34]. We found that SirT1^ko/ko embryonic stem cells do not have increased sensitivity to either ionizing radiation or hydrogen peroxide treatment when compared with wild-type cells, or with positive control cells such as Ku70-deficient or Atm-deficient cells (Figure 1e,f). Both Ku70-deficient mice and Atm-deficient mice display growth retardation phenotypes [35, 36]. The growth retardation phenotype in SirT1^ko/ko mice is therefore probably not due to a defect in DNA damage repair, but rather results from a deficit in IGF-1 signaling.

Both male and female SirT1^ko/ko mice can be fertile, which is in contrast with the reported sterile phenotype in one strain of SirT1-deficient mice [22]. When male SirT1^ko/ko mice impregnated female SirT1^+/ko mice, the number of surviving SirT1^ko/ko offspring was reduced due to the partial perinatal lethal phenotype (Table 1). In a reciprocal approach, male SirT1^+/ko mice can also impregnate female SirT1^ko/ko mice (Table 1). After parturition, SirT1^ko/ko mothers exhibited normal nursing behavior. The pups did not survive for more than 3 days after birth, however, unless they were immediately removed and put under the care of a foster mother. All pups died of dehydration or starvation as a result of lack of milk in their stomachs.

To determine whether the SirT1^ko/ko mothers encountered a lactation defect, whole mount and histological analyses were used to characterize the morphological changes in the mammary glands. We found that virgin SirT1^ko/ko mice displayed impeded ductal morphogenesis up to 9 months of age, while age-matched virgin wild-type mice displayed extensive ductal elongation and branching (Figure 2a). The absence of ductal morphogenesis persisted in pregnant SirT1^ko/ko mice up to day 13 of pregnancy (Figure 2b). Despite the fact that pregnancy can induce ductal morphogenesis at the late stage of pregnancy, underdeveloped mammary glands in SirT1^ko/ko female mice cause a severe deficit in milk production, as shown by histological analysis as well as the immunofluorescence analysis using an anti-milk antibody (Figure 2b,c).

To characterize the developmental defect leading to lactation failure, we analyzed the mammary glands of SirT1^ko/ko female mice on lactation day 1. Terminal end buds (TEBs) are club-shaped transitional structures (see Figure 3a, upper panel). In wild-type female mice, TEBs form and precede ductal elongation and branching at the onset of puberty [17, 24, 25]. When TEBs/ducts reach the edge of mammary fat pads, TEBs regress as shown in the virgin mice in Figure 3a (upper panel). Interestingly, the transitional TEBs can be readily identified in SirT1^ko/ko female mice on lactation day 1, manifesting as either newly formed TEBs or as TEBs attached to developing ductal and alveolar structures (Figure 3a, lower panel). Moreover, SirT1^ko/ko mice displayed varying degrees of ductal development, ranging from a lack of any ductal structure to a fully developed ductal network, the latter of which is comparable with the morphology and cellularity of wild-type mammary tissues on day 13 of pregnancy (Figure 3a). These observations indicated that pregnancy could rescue impeded ductal morphogenesis in virgin SirT1^ko/ko mice.

Normal alveolar morphogenesis takes place under the control of additional hormones and growth factors during pregnancy and lactation [17, 24, 25]. These coordinated efforts are necessary to support a densely saturated lobuloalveolar system for secreting milk after parturition (see wild-type mice on lactation day 1 in Figures 2b and 3a). In some SirT1^ko/ko mice, the alveolar morphogenesis was initiated, as indicated by the presence of a few secretory alveolar epithelial cells within scattered alveolar structures sprouted from a well-established ductal branching network (Figures 2c and 3a, lower panel). The extent of alveolar morphogenesis in SirT1^ko/ko mice was inadequate, however, as shown by the severe deficit in milk production when compared with that in wild-type mice (Figure 2c).

To test whether the increased levels of estrogen in pregnant mice were sufficient to induce ductal morphogenesis in SirT1^ko/ko mice, a single estrogen pellet was implanted in the subscapular region of each virgin SirT1^ko/ko mouse and the mammary tissues were analyzed 14 and 21 days after the implantation. By day 14, wild-type mice displayed ductal side branching reminiscent of the morphological changes during early pregnancy (Figure 4a, left panel). In contrast, SirT1^ko/ko mice showed TEBs and ductal elongation, which are the characteristic features of ductal morphogenesis in pubertal wild-type mice. By day 21, ductal elongation and side branching were restored in SirT1^ko/ko mice and the extent of ductal morphogenesis was indistinguishable between wild-type and SirT1^ko/ko female mice (Figure 4a, right panel). These observations clearly indicated that exogenous estrogen alone is sufficient to rescue ductal morphogenesis in virgin SirT1^ko/ko mice. Furthermore, the presence of transitional TEBs on day 14 and of ductal side branching on day 21 suggested that the characteristic features of mammary gland development at puberty and during early pregnancy could be coupled in response to increasing levels of estrogen. Increased levels of estrogen, either during pregnancy or through implantation, were therefore sufficient to stimulate the differentiation of epithelial progenitor cells and ductal morphogenesis in virgin SirT1^ko/ko mice.

The finding of pregnancy-induced mammary gland development prompted us to investigate how mammary epithelial progenitor cells in pregnant SirT1^ko/ko mice can differentiate so rapidly. Both BrdU labeling and the TUNEL (Terminal deoxynucleotidyl transferase mediated dUTP Nick End Labeling) assay were used to measure the cell proliferation and apoptosis in the mammary tissues, respectively. We found in SirT1^ko/ko mammary tissues that about one-third of the cells in TEBs were BrdU-positive as compared with the number of BrdU-positive cells in the newly differentiated ductal or alveolar epithelial cells (Figure 5a(upper panel), 5b). This BrdU-positive finding differs from that in wild-type mammary tissues, in which less than 3% of ductal epithelial cells and about 25% of alveolar epithelial cells were BrdU-positive. The newly generated ductal epithelial cells in SirT1^ko/ko mice formed elongated ducts and side branches, while pushing TEBs towards the edge of mammary fat pad (Figure 3a, lower panel). The newly generated ductal epithelial cells can consequently further differentiate into a few observed alveolar cells in which milk proteins were detected (Figure 2c). These observations suggested that the proliferation and differentiation of epithelial progenitor cells, which is programmed to take place at different times of a female's reproductive life, could be concurrent at a late stage of pregnancy in SirT1^ko/ko mice.

Apoptosis is a part of the self-renewal and remodeling processes in normal mammary tissues [17, 24, 25]. Apoptotic epithelial cells can be readily identified in both ductal and alveolar epithelium of SirT1^ko/ko mice on lactation day 1 (Figure 5a, lower panel). The levels of apoptotic cells were significantly higher than those in wild-type mice on lactation day 1, but seemed to be closer to that in wild-type mice in mid-pregnancy (e.g. day 13 of pregnancy) (Figure 5c). The rush of proliferation and differentiation during pregnancy-rescued ductal development in SirT1^ko/ko mice manifested multiple features on lactation day 1 that could reassemble some characteristic features of wild-type mice in mid-pregnancy. Successive pregnancies, however, failed to improve the survival rate of either SirT1^+/ko pups or SirT1^ko/ko pups (three or four pregnancies in 3 months for each of three SirT1^ko/ko females tested), indicating that the lactation defect persisted. A SirT1 deficiency therefore affected ductal morphogenesis in virgin mice, and lobuloalveolar proliferation in parous mice.

The characterization of lactation failure has revealed two developmental defects in SirT1^ko/ko female mice. The impeded ductal morphogenesis in virgin SirT1^ko/ko mice bore some resemblance to the defective ductal morphogenesis seen in mice lacking GH, estrogen, or IGF-1, as well as in mice harboring mutations in the corresponding receptors, such as the GH receptor and ERα [17, 24‐31]. These previous studies have established the notion that GH and estrogen synergistically stimulate ductal morphogenesis, which is mediated by stromal cell-produced local IGF-1 in mammary tissues. Impeded ductal morphogenesis in virgin SirT1^ko/ko mice is therefore probably due to a deficit in IGF-1 signaling.

In addition to this early defect, pregnancy-induced mammary gland development in SirT1^ko/ko mice appeared to be arrested at the beginning of lobuloalveolar development, which is reminiscent of the phenotype in mice lacking IκB kinase alpha (IKKα) activity for NF-κB activation [37]. IKKα-deficient mice exhibit normal ductal morphogenesis at the onset of puberty and exhibit normal ductal side branching during early pregnancy. Owing to the absence of NF-κB-dependent cyclin D₁ expression, however, IKKα-deficient mice display defective alveolar epithelial cell proliferation and fail to lactate [37]. While SirT1 negatively regulates the transcription activity of FoxOs and NF-κB in mammalian cells [10, 11, 20], SirT1^ko/ko mice displayed the loss-of-function phenotypes. We hypothesized that SirT1 deficiency deregulates the expression of negative feedback signals and thereby desensitizes the cells to various types of stimulation. Indeed, among the many FoxOs and NF-κB-regulated genes, IGFBP-1 and IκBα encode negative feedback signals for IGF-1 signaling and NF-κB activation, respectively. We found that the basal levels of both IGFBP-1 and IκBα were increased in adipose tissues from virgin SirT1^ko/ko mice and in ductal epithelial cells from SirT1^ko/ko mice on lactation day 1 (Figure 3b). Meanwhile, the expression pattern of IGFBP-1 and IκBα in wild-type mammary tissues, shown in Figure 3b, was in agreement with the findings of others [16, 38, 39]. Namely, the expression of IGFBP-1 was downregulated during pregnancy, whereas NF-κB activity was increased at a late stage of pregnancy. The loss of SirT1 therefore increased the basal level of both IGFBP-1 and IκBα in multilineages of cells, and potentially increased thresholds for activating SirT1-modulated signaling pathways.

SirT1 deficiency did not affect the ability of mammary epithelial progenitor cells to differentiate into functional alveolar epithelial cells despite the lactation failure (Figure 2c). SirT1 deficiency may therefore alter the homeostasis of the signals for ductal morphogenesis and/or the cellular response to the signals. Increased expression of both IGFBP-1 and IκBα in adipose tissues and mammary epithelial cells could simply reflect a global deregulation of gene expression in SirT1^ko/ko cells in which the activity of both FoxOs and NF-κB is increased. IGFBP-1 is a potent inhibitor of IGF-1 in vivo [40], suggesting that the elevated basal level of IGFBP-1 in adipose tissues in SirT1^ko/ko mice could reduce the effect of IGF-1. Moreover, deregulated IGFBP-1 expression, but not deregulated IκBα expression, was reversed in response to increased levels of estrogen in the mammary fat pads after implantation (Figure 4b). The finding is relevant to the fact that the expression of IGFBP-1 is downregulated as the level of estrogen increases in pregnant wild-type mice (Figure 3b). Increased levels of estrogen therefore stimulate the production of stromal cell-derived local IGF-1, which overcomes the IGFBP-1 barrier in SirT1^ko/ko mice and, ultimately, reverses the increased levels of IGFBP-1 in mammary tissues.

It was noted that increased levels of estrogen and local IGF-1 did not interfere with the expression of IκBα (Figure 4b). To determine whether deregulated IκBα expression in SirT1^ko/ko cells can be reversed, we treated MEFs with TNFα and measured the kinetics of IκBα expression. Three independent lines of SirT1^ko/ko MEFs were used, which displayed varying basal levels of increased IκBα expression. Following IκBα degradation induced by TNFα, the levels of newly synthesized IκBα in all three SirT1^ko/ko MEFs were always higher than that of wild-type MEFs (Figure 6). This in vitro finding implies that SirT1 deficiency affects NF-κB signaling when the IκB kinase activation is normal.

Our current system might not be suitable for studying the effects of SirT1 deficiency on NF-κB activation in vivo. NF-κB signaling exhibits distinct biological responses because the expression of IκBα depends on NF-κB activation while the other isoforms of IκB (β and ε) are independent of NF-κB feedback [41]. For example, TNFα stimulation induces dampened oscillatory behavior after the first hour, whereas lipopolysacharide treatment leads to stable NF-κB activation [42, 43]. The NF-κB signaling in the developing mammary gland is under the temporal control of RANK signaling and IKKα activation [37]. The finding of increased levels of newly synthesized IκBα within the first hour indicated that SirT1 deficiency could disrupt the temporal control of NF-κB signaling to all types of stimulation (Figure 6). It remains unknown, however, whether the reduced IGF-1 signaling would affect the efficacy of RANK signaling and IKKα activation in developing mammary glands. The fact that individual SirT1^ko/ko mice manifested varying degrees of ductal morphogenesis (Figure 3a) makes it difficult to dissect the potential compound effect of SirT1 deficiency on both upstream IKKα activation and downstream NF-κB signaling. Other experimental systems, such as mammary epithelial cell-specific SirT1^ko/ko female mice, may be used to address this issue.

SirT1 is an integral part of the IIS system that checks and balances the efficacy of insulin and IGF-1 signals. The IGF-1 signaling of the IIS system is required for survival after birth. As illustrated in Figure 7a, the function of SirT1 parallels the linear IGF-1/IGF-1 receptor/phosphoinositide-3'-OH kinase/Akt signaling pathway, both of which can negatively regulate the transcription activity of FoxOs. Mice harboring targeted mutations in either IGF-1 or the IGF-1 receptor gene exhibit perinatal lethality and growth retardation [44, 45]. In the IGF-1 receptor-expressing cells, the binding of insulin and/or IGF-1 activates the phosphoinositide-3'-OH kinase and Akt kinase cascade for survival [8, 9, 13]. The phosphoinositide-3'-OH kinase/Akt signaling pathway is also highly conserved from worms to mammals, and functions to modulate energy metabolism and lifespan in lower organisms. The mammalian Akt kinase family has three members, Akt1–Akt3, which provide specificity and versatility. Mice lacking both the Akt1 and Akt2 genes display perinatal lethal and growth retardation phenotypes that are strikingly similar to that of either IGF-1 or IGF-1 receptor-deficient mice [46]. In this regard, SirT1^ko/ko mice, as well as other strains of SirT1-deficienct mice, also exhibit perinatal lethal and growth retardation phenotypes (Figure 1c and Table 1) [22, 23]. The consistent observation of perinatal lethal and growth retardation phenotypes from our study and studies with mice lacking IGF-1, the IGF-1 receptor, or Akt1/Akt2 suggests that upregulated FoxO activity unleashes the expression of downstream effector genes that lead to the birth-related stress. This birth-related stress may be related to the switch from a maternal level of IGF-1 to a neonatal level of IGF-1 because the surviving SirT1^ko/ko mice can live into adulthood and can reproduce despite the growth retardation phenotype (Figure 1c). In this context, at least one of the FoxO-deficient mice strains, FoxO3 (Foxo3a/FKHRL1)-deficient mice, grow normally and develop functional mammary glands [47].

IGFBP-1 is one of six IGF-1 binding proteins, but is the only one that can be negatively regulated by insulin [48]. IGF-1 binding proteins inhibit the action of local IGF-1 in a spatial and temporal manner [40, 49]. In mammalian cells, FoxO3 can directly bind to the promoter of the IGFBP-1 gene and activate the transcription [50‐52]. SirT1 acts to negatively regulate the activity of nuclear FoxO3 and to decrease the expression of IGFBP-1 when the IGF-1 signal is low (Figure 7a). Increased levels of IGFBP-1 can diminish the action of IGF-1 signals in vivo. In that way, SirT1 balances the homeostasis of IGF-1 and IGFBP-1 for varying levels of IGF-1 signals.

The homeostasis of IGF-1 and IGFBP-1 clearly affects fertility, as shown by the infertility phenotype in both IGF-1-deficient mice and IGFBP-1 transgenic mice [28, 40, 49]. The serum level of estrogen in IGF-1-deficient female mice appears significantly reduced, whereas in ERα-deficient mice, which are also infertile, there is a threefold increase [53]. In contrast, SirT1^ko/ko mice can be fertile despite the fact that the fertility was reduced as compared with that of wild-type females (Table 1). Although our preliminary assessment indicated that the serum level of estrogen in SirT1^ko/ko female mice was not overtly altered, it remains possible that a subtle irregularity or an individual variation of the estrogen level or ERα expression might be attributed to the reduced fertility. An alternative explanation for reduced fertility is that SirT1 deficiency deregulates the activity of FoxO3 and manifests an opposite effect of FoxO3 deficiency. FoxO3-deficient mice exhibit early depletion of ovarian follicles, which is reminiscent of premature ovarian failure in women [47, 54]. Premature ovarian failure manifests as early cessation of ovarian function and as premature menopause caused by genetic mutations, chemo/radiation therapy, or other unknown factors. It is possible that SirT1 deficiency counterbalances normal release of ovarian follicles and reduces the fertility in mice.

We hypothesize that, in virgin SirT1^ko/ko mice, increased levels of IGFBP-1 in adipose tissues effectively diminish IGF-1 signals, including the estrogen–IGF-1 signaling, and halt the differentiation of mammary epithelial progenitor cells (Figure 7b). Several results of this study provided support for this hypothesis. First, the expression of IGFBP-1 is increased in mammary fat pads from virgin SirT1^ko/ko mice, in which there is no evidence of any duct (Figures 2a and 3a). Second, both pregnancy and estrogen implantation can rescue impeded ductal morphogenesis in SirT1^ko/ko mice (Figures 3a and 4a), demonstrating that increased levels of estrogen can enhance the production of local IGF-1. Moreover, the varying degrees of rescue among individual mice, as well as individual mammary fat pads of the same mouse, also indicated that local IGF-1 must be halted in SirT1^ko/ko mice (Figure 3a, and data not shown). Finally, increased levels of estrogen specifically activate the IGF-1 signaling in mammary tissues, as shown by the reversal of IGFBP-1 expression in SirT1^ko/ko mice after estrogen implantation (Figure 4b). These results indicate that a gradient of IGF-1 signals, including both circulating and local IGF-1, may regulate mammary gland development at embryonic, postnatal, and reproductive stages, and that increased expression of IGFBP-1 decreases the efficacy of IGF-1 signals.

Mammary development may undergo estrogen/IGF-1-independent phases and estrogen/IGF-1-dependent phases, which encompasses a late stage of embryonic development to the onset of puberty and involves maternal IGF-1, circulating IGF-1, and estrogen-stimulated local IGF-1. By embryonic day 16.5, mammary mesenchymes are connected to primitive fat pads and begin to proliferate, which results in rudimentary ducts [17, 25, 55, 56]. We propose that this process marks the transition from estrogen/IGF-1-independent mammary stem cells to IGF-1-dependent epithelial progenitor cells, and that maternal IGF-1 is sufficient to stimulate the differentiation of epithelial progenitor cells to estrogen-independent ductal epithelial cells but not estrogen-dependent epithelial cells (Figure 7b). This explains why the development of rudimentary ducts is normal in either IGF-1 or ERα knockout mice and that TEBs form and regress immediately before and after birth, respectively [17, 26, 27]. During the prepubertal period, the rudimentary ducts grow isometrically under the influence of circulating IGF-1. Like maternal IGF-1, circulating IGF-1 is not sufficient to induce the differentiation of estrogen-independent ductal epithelial cells into estrogen-dependent epithelial cells, which will express ERα and become estrogen dependent (Figure 7b). At the onset of puberty, ovarian estrogen, in synergy with pituitary GH, causes a surge in the production of stromal cell-derived local IGF-1 and stimulates the differentiation of estrogen-independent ductal epithelial cells to estrogen-dependent epithelial cells, which begins postnatal development of mammary gland (Figure 7b). Impeded ductal morphogenesis results in virgin SirT1^ko/ko mice because local IGF-1 is apparently not sufficient to overcome increased IGFBP-1 in mammary adipose tissues. SirT1 therefore positively regulates the efficacy of the estrogen–IGF-1 signaling for ductal epithelial cell proliferation and differentiation.

Breast cancer results from the accumulation of inherited and/or somatic mutations. Lifetime exposure to estrogen is a major risk factor for breast cancer, and a number of lifestyle factors, such as diet, body fat, alcohol consumption, exercise, parity, and hormone replacement therapy, may influence a woman's exposure to estrogen. The compound personal risk to a woman is difficult to assess because the molecular link between these factors and the effect of estrogen is poorly understood. The result of this study demonstrated that SirT1 positively modulates the efficacy of the estrogen–IGF-1 signal. This suggests that if SirT1 serves as a lifetime sensor to dietary restriction and acute withdrawal of nutrients [14, 57], its activity could ultimately influence the risk of breast cancer. Interestingly, an epidemiological study of Dutch famine in 1945 indicated that an exposure to famine at prepubertal age increased the risk of breast cancer, most clearly seen in women who never gave birth [58]. Moreover, the serum levels of estrogen and IGF-1 were increased among those exposed to the famine, which suggests that an epigenetic adaptation phenomenon may take place. Assessing the potential correlation between the enzymatic activity of SirT1 and the risk of breast cancer is therefore of great interest in order to identify feasible targets for chemoprevention against breast cancer.

Several dietary polyphenols as well as small molecules have been identified as either agonists or antagonists of SirT1 [59, 60]. The pharmacological effects of these compounds in humans remain elusive given that SirT1 targets multiple transcription factors and exerts pleiotropic effects. Nonetheless, our study of SirT1^ko/ko mice demonstrated the potential utility of SirT1 antagonists. Specifically, reducing SirT1 activity alters the homeostasis of cellular responses, including deregulation of the expression of IGFBP-1 and IκBα, which can attenuate the stimulation from the estrogen–IGF-1 signaling and potentially desensitize mammary epithelial cells to NF-κB activation. SirT1 is therefore a candidate target for chemoprevention against breast cancer.