SEARCH

SEARCH BY CITATION

Keywords:

  • Transforming growth factor β;
  • Ionizing radiation;
  • Mammary stem cell;
  • Epithelial-mesenchymal transition;
  • Breast cancer;
  • Notch;
  • In silico modeling;
  • Multiscale

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. Author Contributions
  9. Disclosure of Potential Conflicts of Interest
  10. References
  11. Supporting Information

Children exposed to ionizing radiation have a substantially greater breast cancer risk than adults; the mechanism for this strong age dependence is not known. Here we show that pubertal murine mammary glands exposed to sparsely or densely ionizing radiation exhibit enrichment of mammary stem cell and Notch pathways, increased mammary repopulating activity indicative of more stem cells, and propensity to develop estrogen receptor (ER) negative tumors thought to arise from stem cells. We developed a mammary lineage agent-based model (ABM) to evaluate cell inactivation, self-renewal, or dedifferentiation via epithelial-mesenchymal transition (EMT) as mechanisms by which radiation could increase stem cells. ABM rejected cell inactivation and predicted increased self-renewal would only affect juveniles while dedifferentiation could act in both juveniles and adults. To further test self-renewal versus dedifferentiation, we used the MCF10A human mammary epithelial cell line, which recapitulates ductal morphogenesis in humanized fat pads, undergoes EMT in response to radiation and transforming growth factor β (TGFβ) and contains rare stem-like cells that are Let-7c negative or express both basal and luminal cytokeratins. ABM simulation of population dynamics of double cytokeratin cells supported increased self-renewal in irradiated MCF10A treated with TGFβ. Radiation-induced Notch concomitant with TGFβ was necessary for increased self-renewal of Let-7c negative MCF10A cells but not for EMT, indicating that these are independent processes. Consistent with these data, irradiating adult mice did not increase mammary repopulating activity or ER-negative tumors. These studies suggest that irradiation during puberty transiently increases stem cell self-renewal, which increases susceptibility to developing ER-negative breast cancer. Stem Cells 2014;32:649–661


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. Author Contributions
  9. Disclosure of Potential Conflicts of Interest
  10. References
  11. Supporting Information

Ionizing radiation is one of very few environmental exposures known to increase breast cancer risk [1], with the greatest risk conferred by exposure before the age of 20 [2]. More than 50,000 women in the United States have been treated with high-dose (≥20 Gy) chest irradiation for a pediatric or young adult cancer. Children treated for cancer with radiotherapy have a 2.9 relative risk of a second malignancy compared to those who did not receive radiation therapy [3]. According to the Childhood Cancer Survivor Study, breast cancer risk is greatest among young women treated for Hodgkin's lymphoma, but it is also elevated among women who received moderate dose chest radiation for other pediatric and young adult cancers [3]. A review of 11 retrospective studies and 3 case-control studies of women treated with radiation at a young age showed that the cumulative incidence of breast cancer by 40–45 years of age ranged from 13% to 20% [4]. This is substantially higher than in the general population in whom the cumulative incidence of invasive breast cancer by age 45 is only 1% but is similar to that in women with a BRCA gene mutation where the cumulative incidence by age 40 ranges from 10% to 19% [5].

The relatively restricted window of carcinogen susceptibility during puberty has been attributed to either a greater number of target cells or to a critical period of regulation in developing human breast [1]. Notably, breast cancers of women exposed to therapeutic radiation during childhood are significantly more likely to be estrogen receptor (ER)-, progesterone receptor-, and HER-2-negative, so-called triple-negative, compared to age-matched controls [6], and the tumors exhibit an aggressive expression profile [7]. Triple-negative breast cancers are postulated to arise from transformed stem/progenitor cells [8] based on the “cell of origin” hypothesis, which predicts that malignant transformation of stem cells give rise to tumors that are less differentiated and more aggressive. This fundamental programming remains evident in the biology, behavior, and signature of the cancer subtype (reviewed in [9]).

The mammary gland undergoes extensive postnatal ductal morphogenesis, proliferation, and symmetric stem cell division under the influence of the ovarian hormones during puberty. We have previously shown that irradiating mice during puberty elicits a mammary stem cell (MaSC) signature in the expression profiles, increases the mammary repopulating efficiency, and expands a population marked by cell surface proteins associated with stem cells [10]. Notably, tumors arising from Trp53 null epithelium transplanted to irradiated mice are twice more likely to be ER negative compared to tumors arising from transplants in sham-irradiated mice. We identified a metaprofile from expression profiles of tumors arising in irradiated mice compared to those arising in unirradiated mice. Remarkably, this metaprofile clusters radiation preceded cancers and is strongly associated with ER negative human breast cancer, which suggests that common processes could contribute to radiation-preceded and sporadic ER negative cancers [11].

Tissue-specific stem cells or early progenitor cells are considered by many to be the critical cellular target in carcinogenesis, based in part on the idea that stem cell transformation can lead to unlimited progeny [2, 12-17]. Mechanisms that affect stem cell numbers include compensatory regeneration, increased self-renewal, and dedifferentiation to a stem-like state. Compensatory regeneration is observed in irradiated bone marrow and intestine during regeneration after stem cells death [13, 18]. Radiation can also induce cell inactivation via senescence [19]. Alternatively, some stem cells are thought to be resistant to radiation, and signals might increase symmetric division or self-renewal [20]. An additional mechanism is that in which non-stem cells dedifferentiate to replace stem cells [21]. A special case of this is epithelial to mesenchymal transition (EMT), which can recapitulate stem cell programs [22]. This possibility is particularly relevant in irradiated tissues since transforming growth factor β (TGFβ) is a key signal for EMT [23, 24]. Radiation induces TGFβ activation both in vitro and in vivo [25, 26] and primes cultured human mammary epithelial cells to undergo TGFβ-mediated EMT [26, 27].

Here, we used computational modeling to evaluate putative mechanisms underlying increased mammary stem cells following exposure to different radiation qualities and doses. Mammary repopulation and tumorigenesis were evaluated in mice, and modeling predictions were tested in vitro using surrogate markers of lineage commitment in the MCF10A human breast epithelial cell line.

Materials and Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. Author Contributions
  9. Disclosure of Potential Conflicts of Interest
  10. References
  11. Supporting Information

Mice

Mammary transplantation, repopulating activity, mammary stem cell marker fluorescence-activated cell sorting, and tumorigenesis assays were conducted as described previously [10]. In brief, BALB/c mice were irradiated at 10 weeks of age and tissue was collected at the designated times. Mice were irradiated with institutional approval with γ-radiation using a 137Cs source, or 350 MeV/amu Si26, or 600 MeV/amu Fe56 ions delivered at the NASA Space Radiation Laboratory of Brookhaven National Laboratory. In some experiments, mammary glands were cleared of endogenous epithelium at 3 weeks of age. Some mice were transplanted at that time with syngeneic Trp53 null mammary fragments and irradiated at 10 weeks. Another group of 10-week old mice were irradiated and transplanted 3 days later with Trp53 null mammary fragments. Mice were monitored for 600 days for tumor development, which were resected at a volume of approximately 1 cm3.

Cell Culture

MCF10A (ATCC, Manassas, VA) were cultured in serum-free medium with or without recombinant human TGFβ (400 pg/mL; R&D Systems, Minneapolis, MN, Minneapolis, http://www.rndsystems.com) added at the time of plating and added every other day with medium change as previously described [26]. MCF10A cells were stably transfected with control and Let7c- miRNA reporter viruses, a kind gift of Greg Hannon, as previously described [28]. Cell cultures were irradiated (3–200 cGy) with 6 MeV X-rays using a Varian 2300 linear accelerator 4 hours postplating in serum-free medium, and phenotype was assessed at the indicated time postplating. In confluent cell experiments, cells were irradiated (2 Gy) 7 days postplating in serum-free medium, and phenotype was assessed at 5 days later. MCF10A cells stably expressing a Let7c-miRNA reporter plated in six-well plates were cultured for 7 days; flow cytometry was used to determine the percentage of Ds-Red cells. To inhibit Notch signaling, a γ-secretase inhibitor (GSI), Compound E (500 nM, Enzo Life sciences), was added prior to irradiating cells. Histograms were generated using Diva software on LSRII (Becton Dickinson, Franklin Lakes, NJ, http://www.bd.com) and analyzed by Flow Jo software (v7.6.4).

Immunofluorescence and Image Acquisition

Cells grown on eight-well Lab-Tek cover glass slides (Nalge Nunc International, NY) were fixed in methanol for 20 minutes at −20°C, blocked with 0.5% casein, and incubated with the cytokeratin 14 (CK14) rabbit antibodies (Covance Biotechnologies, Princeton, NJ, http://www.covance.com) and cytokeratin 18 (CK18) (Chemicon, Temecula, CA, http://www.chemicon.com) mouse antibodies overnight at 4°C. Slides were washed three times with phosphate buffered saline (PBS) and incubated for 1 hour at room temperature with donkey anti-rabbit Alexa 594 and donkey anti-mouse Alexa 488 from Invitrogen (Carlsbad, CA, http://www.invitrogen.com). Notch was detected in parafomaldehyde-fixed cells with antibodies to activated Notch (Abcam, Cambridge, U.K., http://www.abcam.com; AB8925). 4′,6-diamidino-2-phenylindole (DAPI) was used for nuclear counterstaining, and samples where mounted with Vectashield (Vector Laboratories, Burlingame, CA, http://www.vectorlabs.com). Cells were imaged using a ×40 objective with 0.95 numerical aperture Zeiss Plan-Apochromat objective on a Zeiss Axiovert equipped with epifluorescence. Images were analyzed using ImageJ (NIH, Bethesda, MA) journals developed in-house.

In Silico Models

Both agent-based models (ABMs) were created as extensions of our previous work [29] and were written in Java using the Repast Simphony libraries (Supporting Information). All simulations were performed on the Lawrencium supercomputer (212-node, 1696 processor Linux cluster, Lawrence Berkeley National Laboratory) or on Amazon High-CPU Extra Large EC2 instances. Both models are detailed in Supporting Information Methods.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. Author Contributions
  9. Disclosure of Potential Conflicts of Interest
  10. References
  11. Supporting Information

Radiation Dose and Quality Induced Stem Cell Signature

We previously showed that the MaSC signature defined by Visvader and colleagues [30] is enriched in mammary mRNA expression profiles up to 1 month after exposure to 10 cGy γ-radiation [10]. To assess the generality and durability of this response, we analyzed mammary mRNA expression profiles from mice exposed to different doses and radiation qualities. Radiation quality is important since densely ionizing is more carcinogenic than sparsely ionizing radiation [31]. Tissues analyzed at 1 and 4 weeks, but not 12 weeks, following exposure to 100 cGy sparsely ionizing γ-rays or 10 or 30 cGy densely ionizing radiation were enriched for both a Notch-dependent signature [32] and the MaSC signature [30] compared to control profiles (Supporting Information Fig. 1). Ingenuity Pathway Analysis identified NOTCH1, CTNBB1, TGFB1, TGFB2, and WNT16 as upstream regulators (p < .05) for all radiation treatments at Week 1. At Week 4, only CTNNB1 and TGFB1 persisted. E-cadherin (CNTBB1) is highly regulated during EMT, while Notch and TGFβ are both implicated as signals in stem cell self-renewal. These data suggest that a durable but transient stem cell signature is present in irradiated mammary gland.

image

Figure 1. In silico mammary model and MASCa frequency predictions as a consequence of different radiation mechanisms. (A): Schematic of the mammary lineage components in the model. MASCa and PROGa can both self-renew. Differentiated EPITHa have two possible hormone receptor states (HR +/−). (B): An in silico gland consists of DUCTa, each of which is a lattice to contain MASCa (red), PROGa (yellow), and EPITHa (blue). Shown in the top inset is a growing DUCTa with MASCa and PROGa actively migrating to the tip and dividing to elongate the lattice. During symmetric division, MASCa create a nonmigratory daughter MASCa. After cell agents fill the DUCTa lattices, the DUCTa bifurcate to create two new elongating DUCTa. (C): At week 9, morphogenesis ceases with all migratory MASCa and PROGa differentiating to EPITHa. (D): The rate of MASCa symmetric division, freqSymDiv_MASCa was estimated to be 0.12 in order to match the 1/2,000 MASCa frequency we previously reported [10]. The heat map shows MASCa frequency of a simulated sham-irradiated gland measured at week 18. (E–K): The effects on MASCa frequency after simulating different radiation mechanisms in a simulated “juvenile” mammary gland (during morphogenesis at week 3) were tested (n = 25). Simulations indicate that a transient increase in MASCa symmetric division or transient dedifferentiation, but not cell inactivation through death or cell senescence, could induce the radiation-induced twofold increase in MaSC frequency in irradiated “juvenile” mice previously reported. (L–R): Simulations of irradiated mature glands (after completion of morphogenesis at week 12) were performed to further discriminate between self-renewal and dedifferentiation. (Q): The model predicts MASCa frequency is not affected by transiently increasing self-renewal after morphogenesis is complete. (R): In contrast, dedifferentiation initiated at week 12 increased MASCa frequency 4.5-fold. ***, p < .001. Abbreviations: DUCTa, duct agents; ER, estrogen receptor; EPITHa, epithelial cell agent; MASCa, mammary stem cell agent; PROGa, progenitor agent.

Download figure to PowerPoint

In Silico Mammary Epithelial Population Dynamics

Given the experimental challenges of testing whether radiation increases (a) cell inactivation (i.e., death) that then triggers proliferation; (b) signals for stem cell self-renewal, or (c) dedifferentiation, we turned to modeling to evaluate these potential mechanisms. ABMs previously developed for mammary acinar morphogenesis [29] were modified to simulate mammary lineage population dynamics during puberty. Mammary ABM consisted of interacting agents (computer programs) simulating mammary stem cells, progenitor cells, and differentiated cells (Fig. 1A). In this model, mammary stem cell agents (MASCa) can self-renew by symmetric division to gives rise to two MASCa or can asymmetrically divide to give rise to one MASCa and one progenitor agent, PROGa. In contrast to MASCa, PROGa can only self-renew for a limited number of times before becoming a fully differentiated epithelial cell agent (EPITHa) (Supporting Information Table S1). Fully differentiated EPITHa have limited proliferative potential. Although stem cells are thought to reside in the basal cell layer, our simplified model lacks this positional information. All cell agents reside within duct agents (DUCTa), each of which is a monolayer grid corresponding to a duct (Fig. 1B) with pre-ordained dimensions derived from measurements in mammary gland whole mounts.

During simulated puberty, MASCa and PROGa agents actively divide and migrate to fill the DUCTa lattice (Fig. 1B). MASCa agents undergoing symmetric division produce nonmigratory MASCa, resulting in the random distribution of MASCa, consistent with analysis of mammary epithelial composition using multiscale in situ sorting [33]. Proliferation is initiated during puberty (i.e., week 3 in the mouse), and cell agents divide, filling DUCTa lattices. When a grid is full, bifurcation leads to two DUCTa, and the process is repeated, until simulated week 9, representative of an adult gland when morphogenesis is complete (Fig. 1C). We swept the range for each unknown parameter in the ABM to identify values that resulted in branch growth rates that matched analysis of in vivo mammary outgrowth. Identified and fixed parameters are shown in Supporting Information Table S1. A key parameter was the rate of MASCa symmetric division, freqSymDiv_MASCa, which was found to be 0.12 (Fig. 1D). This value was obtained by selecting simulations leading to a MASCa frequency of approximately 1/2,000 to match the frequency of mammary repopulating cells we previously reported using limiting dilution assay [10].

Irradiating mice during puberty (4 weeks of age) with either 10 or 100 cGy doubled the stem cell frequency measured in adult mammary gland [10]. We simulated radiation-induced cell inactivation, self-renewal, or dedifferentiation in a “juvenile” mammary gland (i.e., has not completed morphogenesis) to determine their impact on MASCa frequency at simulated week 18. Radiation-induced cell senescence was modeled by randomly assigning agents to undergo permanent growth arrest; even extremely high values for random cell senescence (60% of cells) did not affect MASCa frequency (Fig. 1E). Radiation-induced cell death was simulated by randomly deleting agents. We also modeled selective cell death, where MASCa, PROGa, or EPITHa agents were preferentially deleted. High values (60%) of random and selective MASCa killing resulted in modest but significant decreases in MASCa frequency, 0.84-fold change and 0.76-fold change, respectively (Fig. 1F, 1G). In contrast, neither differential cell killing of PROGa or EPITHa resulted in a significant change in MASCa frequency (Fig. 1H, 1I). Thus, no mode of cell inactivation increased MASCa frequency.

As expression profiling showed that Notch signaling was present at 1 and 4 weeks after irradiation but resolved by 12 weeks, we simulated a transient increase in self-renewal. Increasing the rate of MASCa symmetric division in simulated juvenile glands from 0.12 to 0.16 for a duration equivalent to 4 weeks was sufficient to double MASCa at 18 weeks (Fig. 1J). Dedifferentiation was modeled by enabling a random fraction of EPITHa to become nonmigratory MASCa starting at the time of simulated radiation. TGFβ and E-cadherin, which are involved in EMT, were transiently invoked in expression profiles. Therefore, dedifferentiation was also transiently initiated for a duration equivalent to 4 weeks. A dedifferentiation rate of 0.0001% in simulated juvenile tissue led to an approximately two-fold increase in MASCa frequency (Fig. 1K). Thus, ABM simulations suggest that increased self-renewal or dedifferentiation, but not compensatory regeneration, are mechanisms that could increase stem cell frequency in irradiated juvenile mice.

Radiation therapy results in a much lower risk of breast cancer in adult women and the smaller risk is not associated preferentially with ER negative cancer [34]. Therefore, we used the ABM to predict the effect of radiation on MASCa after morphogenesis was complete, comparable to an adult gland. Radiation time was set at simulated week 12, and MASCa frequencies were predicted for week 18. As in the juvenile setting, neither random cell senescence nor selective PROGa cell death had any effect on MASCa frequency (Fig. 2L, 2O). Random cell death and selective MASCa death led to greater decreases in MASCa frequency in simulated adult versus simulated juvenile gland (Fig. 2M, 2N). Differential EPITHa death in simulated irradiated adult glands led to a small (1.2-fold) but significant increase in MASCa frequency (Fig. 2P). In contrast to simulations of irradiated juveniles, transiently increasing self-renewal in adults did not affect MASCa frequency (Fig. 1Q), because significant proliferation is a necessary condition. On the other hand, dedifferentiation initiated at week 12 increased MASCa frequency 4.5-fold (Fig. 1R). Thus, modeling predicted that dedifferentiation could increase MaSC frequency in irradiated adult mice, but that self-renewal would not.

image

Figure 2. Lineage marker status of MCF10A treated with radiation and/or TGFβ. MCF10A outgrowths in humanized mouse fat pad consist of morphologically bilayered epithelium. (A): Whole mounts (left) and histology (middle, ×10; right, ×40). (B): CK14 (red) and CK18 (green) dual stain. (C): Human specific DNA, COT1, in situ hybridization of humanized fat pad. Note that both fibroblasts and epithelium are human. (D): Negative control for COT1 in situ hybridization in mouse mammary gland. (E): Dual immunostaining of basal CK14 (red, left panel) and luminal CK18 (green, middle panel) in MCF10A cells. Nuclei are counterstained with DAPI (blue). Rightmost panel is merged to show the presence of double CK14/18 positive cells. Scale bar = 50 µm. Cells in which both markers simultaneously colocalize appear yellow in the merged image. (F): Quantitation of basal CK14, luminal CK18, and double positive populations under four treatment conditions (sham, treatment with TGFβ, and/or 200 cGy γ-radiation). Frequencies are calculated as the total number of cells expressing each or both of the CK markers over the total number of cells in the population. Three biological replicates were randomly imaged for a total analysis of 6,935 cells. The frequency of cells expressing both CKs were significantly increased (p < .001, χ2) after treatment with TGFβ or radiation and TGFβ compared to the sham irradiated controls. ***, p < .001. (G): Quantitation (mean ± S.D.) of cells positive for both basal CK14 and luminal CK18 in MCF10A cultured with TGFβ as a function of graded doses of γ-radiation (blue) or Si particle radiation (red) compared to γ-irradiated cells cultured without TGFβ. Very low doses of either radiation type significantly increased the frequency of double positive cells. Abbreviations: CK, cytokeratin; DAPI, 4′,6-diamidino-2-phenylindole; IR, irradiation; TGFβ, transforming growth factor β.

Download figure to PowerPoint

Based on the prediction of different outcomes depending on the mechanism of response, we irradiated 12-week-old adult mice with 10 or 100 cGy. Six weeks later, mammary epithelial cells were isolated and analyzed. Similar cell numbers were recovered from irradiated mouse mammary glands, as is expected given these low doses. The mammary stem cell enriched lin−/Cd24med/Cd49hi fraction was analyzed by flow cytometry [35]. This population increases in cells isolated from mice irradiated as juveniles [10]. However, the lin−/Cd24/Cd49 profiles of sham and irradiated adult mammary cells were comparable (Supporting Information Fig. 2A, 2B). Functional analysis of repopulating potential using limiting dilution into cleared mammary fat pads is the gold standard to assess mammary stem cells [36] and is increased in cells isolated from mice irradiated as juveniles [10]. Consistent with the lineage analysis, mammary repopulating activity was similar between irradiated adult mice and contemporaneous sham-irradiated mice (Supporting Information Fig. 2C, 2D). Therefore, the prediction that dedifferentiation would increase mammary stem cells in adult mice is wrong, which suggests that increased self-renewal signaling is the most likely mechanism.

Mammary Epithelial Lineage Commitment

To test the prediction that self-renewal was more likely than dedifferentiation, we turned to the MCF10A nonmalignant human mammary epithelial cell line, which can be primed by radiation to undergo TGFβ-mediated EMT [26, 27]. MCF10A cell cultures consist of distinct CK18 and CK14 expressing cells and reportedly contain a small population of cells expressing stem cell markers [37]. We found that MCF10A exhibit a remarkable capacity to generate histiotypic ductal outgrowths in a humanized mammary gland (Fig. 2A–2D). Ducts formed by MCF10A consist of luminal CK18 and basal CK14 bilayer (Fig. 2B), which were confirmed as human by in situ hybridization for COT1 DNA (Fig. 2C, 2D). The human breast epithelium consists of luminal, CK18 epithelial cells and basal, CK14 epithelial cells and a rare population in which basal and luminal CKs colocalize is thought to represent a pluripotent progenitor cell [38]. Both single and double positive CK18 and CK14 cells are evident in MCF10A cultures (Fig. 2E). Thus, we used CK18 and CK14 CK expression in MCF10A as a model to further study radiation effects on epithelial lineage commitment.

EMT, elicited by overexpression of transcription factors or exposure to TGFβ, is strongly associated with acquisition of stem cell markers and function [39]. Here we tested whether EMT primed by radiation affected the CK14/18 distribution of MCF10A cells. The distribution of CK14, CK18, and CK14/18 positive cells in cultures arising from irradiated cells was not significantly different from sham-irradiated populations (Fig. 2F). Significantly more CK14/18 cells and less CK18 cells were present in cultures exposed to TGFβ compared to control cultures or irradiated cultures. In contrast, nearly a third of the cells in irradiated cultures treated with TGFβ were CK14/18, which was significantly higher than in single treatment or control cultures (p < .001, Fig. 2F). The dose-response for EMT following either densely or sparsely ionizing radiation is switch-like, that is, doses as low as 3 cGy are sufficient and higher doses are quantitatively similar [27]. Thus we examined the frequency of CK14/18 cells following graded doses of either sparsely or densely (350 MeV/amu Si) ionizing radiation. The frequency of CK14/18 cells exhibited a very similar dose-response (Fig. 2G), which was substantied by nonsignificant Pearson and linear regression coefficients. The correlation between EMT and CK14/18 cells seemed to support a dedifferentiation mechanism.

We reasoned that a dedifferentiation mechanism would mean that CK14/18 cells are the progeny of lineage restricted, single keratin positive cells, while increased self-renewal would increase symmetric division of double positive cells to produce similar double-positive daughters. To test this, we analyzed the lineage characteristics as a function of time in culture. The percentage of CK14/18 dual-labeled cells was unexpectedly greatest at 24 hours after plating in control and irradiated cultures and gradually decreased over time while CK14 cells increased (Fig. 3). TGFβ-treated cultures began with and maintained more double-positive cells than those cultured without TGFβ. As above, irradiated, TGFβ-treated cultures contained the highest percentage of CK14/18 positive cells. To test whether symmetric division of pre-existing progenitors could account for the population distribution over time, we formulated an ABM consisting of bipotent cell agents (BPa), representing CK14/18 positive cells, defined by their ability to undergo symmetric self-renewal or asymmetric lineage commitment to either a basal (BCa) or luminal (LCa) cell agent type (Fig. 4A). Monolayer culture lag and log phase were simulated by assigning each agent with one cell cycle time for the first three simulated days a second cell cycle time for the remainder of the week-long experiments. Contact inhibition was included such that a cell agent could divide only if there was neighboring space. Dedifferentiation of BCa or LCa to BPa was also included (Fig. 4A). A parameter sweep determined that turning off dedifferentiation led to the best fit of the simulated treatment conditions (Fig. 4B, 4C), suggesting that this mechanism is the least likely to explain the in vitro results. Moreover, the best-fit simulation results using dedifferentiation are noticeably worse than the fit using increased symmetric division (Supporting Information Fig. S3A–S3C). The fitted parameters indicated that the probability of symmetric self-renewal increased in TGFβ-treated cells and the greatest difference was observed between controls and irradiated, TGFβ-treated cells (Fig. 4B). As the in vivo ABM predicted that self-renewal would only be effective during morphogenesis when there is considerable proliferation, MCF10A were grown to confluence (7 days) to reduce proliferation, then irradiated and cultured for additional 5 days with and without TGFβ (Supporting Information Fig. S4A). Under these conditions of low proliferation, neither radiation nor treatment with TGFβ affected the proportion of CK14/18 cells compared to control cultures (p = .323, Supporting Information Fig. S4B).

image

Figure 3. The distribution of MCF10A CK subpopulations changes after radiation and/or TGFβ exposure. To evaluate lineage commitment, control or 200 cGy γ-irradiated MCF10A cells were cultured with TGFβ and fixed at the indicated time prior to immunostaining for CK18 and CK14 lineage markers. Ten random images were acquired from each biological triplicate at 24, 48, 72, 120 and 168 hours after radiation and/or TGFβ treatment. Assignment of CK status is represented as mean ± SE at each time point. Abbreviations: CK, cytokeratin; IR, irradiation; TGFβ, transforming growth factor β.

Download figure to PowerPoint

image

Figure 4. Comparison of in vitro lineage distribution with in silico modeling after radiation and/or TGFβ exposure. (A): The in silico model consists of BPa, LCa, and BCa cell agents. (B): The best fit parameters for the model. The parameter α is the probability for BPa symmetric division. The parameter γ dictates the probability for asymmetric luminal division. The parameters δL and δB determine the probability for dedifferentiation of LCa or BCa to BPa, respectively. Tc1 and Tc2 are cell agent cycle times for the first 3 days and the remaining days in culture, respectively. The best fit parameters indicate that both TGFβ treatment alone and radiation with TGFβ treatment increases self-renewal, but the greatest difference in the probability of symmetric self-renewal is observed between sham and cells that were irradiated and TGFβ treated. (C): Experimental data points indicating the number of cells (mean ± SE) counted for a set surface area randomly scanned (n = 3). Red stars, CK14+ cells; green squares, CK18+ cells; yellow diamonds, double positive cells. Shaded areas indicate simulation results (red, BCa; green, LCa; yellow, BPa; mean ± SD, n = 50) using the model with the best fit parameters yielding the lowest squared residual error for all three cell type growth kinetics and for all four experimental conditions. Insets show example simulations at 24-hour, 96-hour, and 168-hour time points (red, BCa; green, LCa; yellow, BPa) along with an in vitro image at 168-hour for each treatment (CK14, red; CK18, green; nuclear counter stain, DAPI blue). Abbreviations: BPa, bipotent cell agent; BCa, basal cell agent; IR, irradiation; LCa, luminal cell agent; TGFβ, transforming growth factor β.

Download figure to PowerPoint

To further test whether TGFβ is a necessary signal for self-renewal in vivo, Tgfb1 heterozygote and wild-type littermates were irradiated with 10 cGy at 4 weeks of age. Although null genotypes are embryonic or perinatal lethal, Tgfb1 heterozygote mammary gland has 70%–90% less TGFβ than wild-type tissue [40]. As previously reported [10], mammary repopulation frequency doubled in the mammary gland of irradiated wild-type mice. In contrast, the mammary repopulating activity of irradiated Tgfb1 heterozygote mice was similar to unirradiated mice (Supporting Information Fig. S5). These data provide functional validation of the requirement for radiation-induced TGFβ in vivo. Together, ABMs of in vivo and in vitro mammary cell fate decisions and experimental data support the hypothesis that radiation stimulates self-renewal, which requires both TGFβ and active proliferation.

Radiation-Induced Notch and TGFβ Promote Self-Renewal

As a further test of this hypothesis, MCF10A were stably transduced with the retrovirus Let7c-reporter. Stem cells downregulate Let-7 miRNA [41]. Ibarra et al. constructed a reporter in which Let-7c miRNA silences Ds-Red expression in differentiated cells. Therefore, the lack of Let-7c in stem/progenitor cells leads to Ds-Red cells [28] (Fig. 5A). As evident in dual phase-fluorescence micrographs, monolayer cultures contain few Ds-Red cells (Fig. 5B) and mammospheres grown in non-adherent, growth restricted conditions contained only one or two Ds-Red cells (Fig. 5C). To test whether Ds-Red positive progenitor cells underwent self-renewal, we viably sorted the population using flow cytometry into Ds-Red positive and negative populations and cultured them independently. The Ds-Red positive sorted population gave rise to cultures containing both Ds-Red positive cells and negative cells but the Ds-Red negative sorted population did not yield Ds-Red cells. The percentage of Ds-Red positive cells in cultures initiated from sorted Ds-Red-cell increased from 1.7% ± 0.08% to 3.7% ± 0.18%, indicative of self-renewal.

image

Figure 5. Radiation and TGFβ increase the frequency of cells expressing the Let-7 reporter in a Notch-dependent fashion. (A): Diagram of the assay in which Let-7c expression represses Ds-Red reporter in differentiated cells. Cells lacking Let-7c are red. (B): Phase/fluorescence merged images of MCF10A expressing the Let-7c reporter after radiation and/or TGFβ treatment. Scale bar = 100 µm. (C): Representative images of MCF10A mammospheres expressing the Let-7c reporter after radiation and/or TGFβ treatment. Mammospheres treated with GSI contained DS-Red cells. (D): Fluorescence-activated cell sorting analysis of Ds-Red-cell frequency of cells from sham, after radiation and/or TGFβ treatment (gray) compared to those treated with GSI (black). Ds-Red cells increased in cultures treated with TGFβ or radiation and TGFβ compared to the sham controls (mean ± SE from three biological replicates), which was blocked by Notch inhibition. Ds-Red positive sorted population continued to give rise to both Ds-Red positive cells and negative cells, whereas the Ds-Red negative cells did not yield Ds-Red progenitor cells under any condition. (E): Representative images of Notch and β-catenin immunostaining illustrating the nuclear localization of Notch induced by 200 cGy γ-radiation and TGFβ. Scale bars = 50 µm. (F): The frequency of cells that exhibit nuclear Notch staining is significantly increased by either single or double treatments. Incubation (4-hour) with GSI exhibit Notch immunoreactivity, evidence of antibody specificity. *, p < .05; **, p < .01; ***, p < .001; χ2. Abbreviations: DMSO, dimethylsulfoxide, vehicle control; DAPI, 4′,6-diamidino-2-phenylindole; GSI, gamma secretase inhibitor; IR, irradiation; TGFβ, transforming growth factor β.

Download figure to PowerPoint

Radiation alone did not affect the frequency of Ds-Red cells as measured by flow cytometry (Fig. 5D). In contrast, TGFβ treatment significantly increased the Ds-Red population (p < .01), which was further enhanced by irradiation (p < .001). When Ds-Red cells were sorted from irradiated cultures treated with TGFβ, 39.6% ± 0.48% of the subsequent population was Ds-Red, consistent with increased self-renewal. In contrast, sorting Ds-Red negative cells from sham, TGFβ, radiation or irradiated, TGFβ treated cultures generated cultures without Ds-Red positive cells, which support the conclusion that dedifferentiation does not contribute to the increase in Ds-Red cells that we observed after exposure to radiation and TGFβ. We then modeled the self-renewal capacity of Ds-Red positive progenitor cells and found that, in accord with the results for the CK14/18 double positive population, the probability of symmetric divisions of these cells increased after radiation and TGFβ exposure (0.68 in sham cultures; 0.99 in cultures treated with radiation and TGFβ).

Stem cells, radiation, and EMT can be linked through their common association with the Notch pathway [42]. Notch activation precedes increased mammary repopulating activity in irradiated tissue based on expression profiling, expression of Notch target genes, and protein localization [10]. We localized nuclear Notch in MCF10A cells in sham and double treated cultures (Fig. 5E). As shown by quantitation (Fig. 5F), all treatments significantly increased cells exhibiting nuclear Notch, even though only double treatments induce EMT or CK14/18 dual labeled cells. We next tested whether this was functionally required for the increase in Let-7c reporter cells by using a GSI, which prevented the nuclear localization of Notch, indicative of activation (Fig. 5F). GSI blocked the increase in the frequency of Ds-Red cells in TGFβ treated cultures (Fig. 5D), but did not affect the presence of Ds-Red cells in mammospheres (Fig. 5C), which indicates that concurrent Notch and TGFβ induced by radiation co-operate to stimulate self-renewal reported by Let-7c.

Radiation Does Not Affect Tumor ER Status After Ductal Morphogenesis Is Complete

Our prior studies showed that the expansion of stem cells in mice irradiated during puberty correlates with development of aggressive, hormone receptor negative tumors arising in the radiation chimera model [10]. The current modeling and experiments shows that proliferation is necessary for stem cell expansion, thus irradiation of adult mice does not increase stem cells. If so, irradiation after cessation of ductal morphogenesis should not shift Trp53 tumor type in the mammary chimera model. To test this, we compared tumors arising in mice that were transplanted with Trp53 null mammary tissue and irradiated after morphogenesis was complete to those occurring in mice that were irradiated and subsequently transplanted with Trp53 null mammary tissue (i.e., ductal morphogenesis comparable to puberty ensues after irradiation). In both cases, the mice were 10 weeks old when exposed to sparsely or densely ionizing radiation and Trp53 null tumors arose 9–15 months later. Significantly more ER negative tumors arose in mice irradiated prior to Trp53 null transplantation compared to sham-irradiated mice (Fig. 6), as shown previously [10]. In contrast, the proportion of ER negative tumors was unaffected in mice that were irradiated after Trp53 ductal morphogenesis was complete. Thus, the increased frequency of ER-negative cancer correlates with increased mammary stem cells as a function of radiation exposure in young animals, while neither occurs when adult mice are irradiated.

image

Figure 6. Radiation exposure before completion of morphogenesis increases ER-negative tumors while irradiation after morphogenesis does not affect ER status. Tumors that arose from Trp53 null fragments undergoing morphogenesis before or after irradiation were scored for ER status using the Allred scoring system. (A): As previously reported, ER-negative tumors were significantly increased (χ2, p = .03, 95% CI) when morphogenesis occurred after exposure to either sparsely ionizing (36%, n = 25) or densely ionizing (42%, n = 64) radiation compared to sham (63%, n = 51). (B): In contrast, exposure to either sparsely or densely ionizing radiation after morphogenesis was complete did not affect the ER status of subsequent tumors (χ2 p [mt] .05, 95% CI). *, p < .05. Abbreviations: ER, estrogen receptor; IR, irradiation.

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. Author Contributions
  9. Disclosure of Potential Conflicts of Interest
  10. References
  11. Supporting Information

Exposure to radiation and the risk of developing breast cancer is unusually age-dependent [2]. Women exposed to radiation during puberty have the greatest risk of developing breast cancer [2]. Likewise, irradiated young Sprague-Dawley rats develop more mammary cancer while irradiated adults do not [43]. Notably, radiation-preceded breast cancer in women is much more likely to be aggressive and ER negative [6]. Radiation-preceded breast cancer also exhibit distinct mRNA expression profiles associated with aggressive breast cancers [7]. Our data indicate that increased stem cell self-renewal during ductal morphogenesis is the biological basis for puberty as window of susceptibility to radiation. Gene expression data show that radiation transiently induces well-established self-renewal signals, Notch and TGFβ, which in turn increase mammary stem cell self-renewal when coupled with proliferation during puberty. In contrast, radiation does not affect mammary repopulating activity after morphogenesis is complete. Our experimental data provide new biological insight into basis for the greater risk of radiation exposure during puberty that is well documented by radiation epidemiology.

The hypothesis that mammary stem cells or early progenitors are the critical cellular targets for carcinogenesis broadly motivates studies of intrinsic and extrinsic control of mammary stem cell self-renewal [2, 12-17]. The murine mammary hierarchy has been studied using marker analysis, lineage tracing, and a gold standard of function based on the ability to repopulate the mammary gland [36], similar to bone marrow repopulation used to define the hematopoietic lineages. Recent studies that attempt to define stem cell hierarchy indicate that each assay has limitations but that the repopulation assay is a very stringent test for capacity that may not be evident during homeostasis [44]. We tested the prevailing hypothesis that cell kill per se increases stem cell self-renewal by using radiation dose and densely ionizing radiation, both of which increase cell death. If stem cell activity is a function of cell loss, we would expect a proportional effect of dose and radiation quality on either stem cell qualitative (i.e., signature) or quantitative (i.e., repopulating) assays. However, enrichment of mammary stem cell signatures and increased repopulating activity were dose and radiation quality independent. This puzzling observation led us to weigh possible mechanisms using an in silico approach. An ABM of a simplified mammary hierarchy was developed to compare the consequences of assuming compensatory regeneration, increased self-renewal, or dedifferentiation to a stem-like state. These simulations rejected compensatory regeneration in that no mode of cell inactivation increased MASCa frequency.

In contrast, the ABM suggested that either an increase in self-renewal or dedifferentiation could increase MASCa frequency. Based on the observation that radiation results in a much lower risk of breast cancer in adult women [34], we used the ABM to predict the effect of self-renewal or dedifferentiation on MASCa in an irradiated adult gland, after ductal morphogenesis is complete. Interestingly, the ABM predicted that dedifferentiation could increase MaSC frequency in irradiated adult mice, but that self-renewal would not. This prediction was tested in adult mice, which did not show an increase in mammary repopulating cells, supporting self-renewal as the likely mechanism. The ABM predicted that active morphogenesis is a necessary condition to enhance stem cell activity after exposure to ionizing radiation. In essence, the combination of symmetric division with a transient signal for self-renewal in the ABM led to more stem cells for the same total number of epithelial cells.

Weinberg and colleagues have uncovered a strong link between EMT and stem cell reprogramming [39, 45, 46]. Since the in vivo ABM predicted that either self-renewal or dedifferentiation could increase MASCa, but the comparison between juvenile and adult mice supported self-renewal, we used the nonmalignant human mammary epithelial cell line, MCF10A, to further test this conclusion. The MCF10A cell line has remarkable morphogenic capacity in a humanized stroma, consists of distinct luminal and basal cell types, and contains a dual luminal and basal CK expressing population, which Petersen and colleagues identified as a progenitor in human breast [38]. These properties provide a useful model to study signaling in epithelial lineage commitment. The increase in MCF10A cells expressing both basal and luminal keratins exhibits switch-like radiation dose dependence similar to that shown for EMT [26, 27]. However, the ABM describing MCF10A population dynamics indicated that EMT and selfrenewal occur in parallel rather than in sequence and that, as found in vivo, significant proliferation is necessary for putative progenitor cells to increase. A very small population of MCF10A cells is also Let7c-negative, which is postulated to be a surrogate marker of progenitor cells [28]. Irradiation did not affect the frequency of Let7c-negative cells unless cultures were treated with TGFβ. This increase of Let7c negative cells was blocked by γ-secretase inhibition of Notch processing. These data indicate that even though irradiated MCF10A undergo TGFb mediated EMT, the primary mechanism by which radiation increased stem/progenitor cells is self-renewal mediated by TGFβ and Notch.

It is somewhat surprising that TGFβ is necessary for self-renewal since TGFβ overexpression suppresses self-renewal in normal murine mammary gland [47]. Yet Tgfb1 heterozygote mice supported the requirement for TGFβ in radiation-induced mammary repopulating activity, and Fuchs and colleagues report that damage-induced transient TGFβ activation overcomes hair follicle stem cell quiescence [48]. We speculate that radiation-induced TGFβ acts differently than homeostatic signaling because radiation also induces Notch effectors in vivo [10]. Radiation induced Notch increases self-renewal in cancer stem cells [49-51]. Although nuclear Notch was robustly induced by radiation, CK14/18 dual labeled and Let7c-negative, Ds-Red positive MCF10A cells only increased in irradiated cultures treated with TGFβ, indicating that Notch and TGFβ work together to stimulate self-renewal. Since TGFβ also mediates the acceleration of tumorigenesis in irradiated mice [10], this effect raises concerns that, while broadly characterized as a tumor suppressor, TGFβ can act to promote processes that prime tissue to develop cancer.

CONCLUSIONS

Together these data suggests that increased stromal TGFβ and epithelial Notch in mice irradiated during puberty mediate a transient increase in symmetric division that increases mammary repopulating cells. Similarly, recent studies using irradiated mammosphere cultures showed that p21 mediates stem cell resistance to p53-dependent apoptosis [20] and, interestingly, p21 is a direct target of TGFβ signaling [52]. Increased mammary stem cells and the subsequent development of aggressive, ER-negative breast cancer, which are functionally linked by the cell-of-origin hypothesis [9], are highly correlated in our murine models. The so-called window of susceptibility, evident from the epidemiology of radiation exposure in humans, is not observed when adults are exposed. Our experimental data support the hypothesis that the breast is susceptible to a transient increase in stem cell self-renewal when exposed to radiation during puberty, which primes the adult tissue to develop aggressive cancer decades later.

Acknowledgments

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. Author Contributions
  9. Disclosure of Potential Conflicts of Interest
  10. References
  11. Supporting Information

We thank Michael Gonzalez, William Chou and Jessica Chang for technical assistance. This research was supported by NASA Specialized Center for Research in Radiation Health Effects, NNX09AM52G and by DOE Low-Dose Radiation program (M.H.B.H.).

Author Contributions

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. Author Contributions
  9. Disclosure of Potential Conflicts of Interest
  10. References
  11. Supporting Information

J.T., I. F.-G., S.V., and J.-H.M.: collection and/or assembly of data, data analysis and interpretation, and manuscript writing; H. M.-R. and I.I.-B.: collection and/or assembly of data; D.H.N.: data analysis and interpretation; S.V.C. and M.H.B.-H.: conception and design, financial support, collection and/or assembly of data, data analysis and interpretation, manuscript writing, and final approval of manuscript. J.T. and I.F.G. contributed equally to this article.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. Author Contributions
  9. Disclosure of Potential Conflicts of Interest
  10. References
  11. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. Author Contributions
  9. Disclosure of Potential Conflicts of Interest
  10. References
  11. Supporting Information

Additional Supporting Information may be found in the online version of this article.

FilenameFormatSizeDescription
stem1533-sup-0001-suppinfo.doc2488KSupporting Information
stem1533-sup-0002-supplegends.doc36KSupporting Information Legends

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.