Introduction

Breast cancer is the most common cancer in women worldwide, with more than one million new cases diagnosed each year [1]. Breast cancer metastasizes primarily to the bone, lung, lymph nodes, liver and brain, and distant metastasis of this kind accounts for nearly 90% of deaths due to breast cancer. Therefore, strategies for its treatment need to be developed. Breast cancer represents a heterogeneous disease comprising several intrinsic subtypes with distinct clinical outcomes, such as basal-like, HER2-enriched, luminal A, luminal B, and normal-like [2,3,4]. Of these, the luminal A subtype carries the best prognosis with the lowest metastatic potential, whereas the basal subtype carries the poorest prognosis with the highest metastatic potential [1, 5, 6].

Several studies have thus far examined the ionizing radiation sensitivity of primary and metastatic cancer. Rantanen et al. found that the clonogenic survival after X‑irradiation of UT-EC-2A (established from a patient with endometrial adenocarcinoma) is comparable to that of UT-EC-2B (established from the same patient but 17 months later from metastasis in the left supraclavicular fossa) [7]. Huerta et al. compared SW480 (established from a patient with primary colon adenocarcinoma) with SW620 (established from the same patient but 6 months later from a lymph node metastasis), and found that SW620 is more resistant to γ‑induced apoptosis than SW480 [8, 9]. van den Aardweg et al. reported wide variations in clonogenic survival after X‑irradiation among eight uveal melanoma cell lines (four primary and four metastatic from different patients) independently of whether or not a cell line is metastatic [10]. However, there have been no such reports for breast cancer.

Radiotherapy is used for metastasis from breast cancer [11,12,13,14], for which the dose is determined by considering the characteristics of the primary cancer based on clinical experience, but not that of metastatic cancer. Thus, we set out to establish the metastatic cell lines from the basal subtype cell line, and characterize the radiosensitivity of such metastatic cell lines.

Materials and methods

Cells and cell cultures

Human breast adenocarcinoma cell lines MDA-MB-231 (ATCC® HTB-26TM) and MCF7 (ATCC® HTB-22TM) were purchased from the American Type Culture Collection (ATCC, Manassas, VA). MDA-MB-231 is a triple-negative, basal subtype that was established from a pleural effusion in a 51-year-old Caucasian woman with a metastatic mammary carcinoma [15, 16]. MCF7 (Michigan Cancer Foundation-7) is an estrogen receptor (ER)-positive, progesterone receptor (PgR)-positive, luminal A-subtype that was established from a pleural effusion in a 69-year-old Caucasian woman with a metastatic mammary carcinoma [17, 18]. Unless otherwise specified, all cell cultures were maintained in RPMI1640 (R8758, Sigma-Aldrich, St Louis, MO) containing 10% fetal bovine serum (FBS) at 37˚C in a humidified atmosphere of 5% CO2 in air. For all the experiments, control cells were sham-manipulated and handled in parallel with the test cells.

Establishment of pGL4.5 cell line and its metastatic cell lines by in vivo selections

The pGL4.50[luc2/CMV/Hygro] vector (E1310, Promega, Madison, WI) encodes the luciferase reporter gene luc2, cytomegalovirus promoter and Hygromycin B resistance gene. MDA-MB-231 cells were lipofected with the pGL4.50[luc2/CMV/Hygro] vector using Lipofectamin2000 (Invitrogen, Carlsbad, CA), and were maintained in RPMI1640 containing 10% FBS and 0.1 mg/ml Hygromycin B (Life Technologies, Carlsbad, CA). The resulting cell line was named “pGL4.5” cells, which can be visualized with luciferase based, noninvasive bioluminescence imaging [19].

To obtain metastatic cell lines derived from pGL4.5 cells, 4‑ to 5‑week old female BALB/cAJcl-nu/nu athymic nude mice purchased from CLEA Japan (Tokyo, Japan) were used. All animal experiments were approved by the Animal Experiments Committee of Fukushima Medical University (approval No. 2506), and conducted according to the guidelines of the committee. Mice were anesthetized via intraperitoneal injection of a mixture of medetomidine (0.2 μg/g body weight: BW), midazolam (2.2 μg/g BW) and butorphanol tartrate (2.8 μg/g BW).

To establish a cell line metastatic to the lymph nodes, pGL4.5 cells were orthotropically implanted into the mammary fat pad (Fig. S1). Briefly, with a 27-gauge needle, 1 × 106 pGL4.5 cells in 0.1 ml were inoculated in about 1 min into the mammary fat pad of anesthetized mice [20]. At 20–30 days after inoculation of pGL4.5 cells, D‑Luciferin (in vivo grade VivoGloTM Luciferin, Promega) was intraperitoneally injected into anesthetized mice at 150 µg/g BW. At 5 min after injection, metastatic lymph node lesions were visualized with bioluminescence imaging using the IVIS Lumina Imaging System (Xenogen Corporation, Alameda, CA), and isolated. Cells obtained from isolated lymph nodes were expanded in RPMI1640 containing 10% FBS, 0.1 mg/ml Hygromycin B, 200 U/ml penicillin and 200 µg/ml streptomycin, and the resulting cell line metastatic to the lymph nodes was named “NLN0001” cells (Fig. S2). Likewise, NLN0001 cells underwent the same orthotropic implantation procedures one more time, and the resulting cell line was named “NLN0104” cells (Fig. S2).

MDA-MB-231 cells form metastases in the bone, lung and elsewhere from the bloodstream when inoculated into the left cardiac ventricle of immunodeficient mice [21]. Therefore, to establish cell lines metastatic to the bone or lung, pGL4.5 cells were subjected to left ventricular cardiac injection (Fig. S1). In brief, after observing a back flow into the syringe, 1 × 105 pGL4.5 cells in 0.1 ml were inoculated with a 27-gauge needle in about 1 min into the left ventricle of anesthetized mice [19, 21]. With the same method as described for NLN0001 cells, metastatic bone or lung lesions were isolated, followed by expansion and reinoculation of cells. After repeating the left ventricular cardiac injections three times, the cell lines metastatic to the bone and lung were obtained and named “NBOS0042” and “NLU0041”, respectively (Fig. S2). For more detailed information, see the legend to Fig. S2.

Table S1 shows the frequency of metastasis to the bone, lung and lymph nodes in pGL4.5 cells and its metastatic cell lines. Metastatic frequency was 0–50% in pGL4.5 cells, which was increased to 83% for the lymph nodes in NLN0001 cells established by the in vivo selection once. Moreover, metastatic frequency was 100% for the lymph nodes in NLN0104 cells established by the in vivo selection twice, as well as for the bone and lung in NBOS0042 and NLU0041, respectively, both of which were established by the in vivo selections three times.

Cell proliferation assay

For in vivo experiments, 1 × 106 cells were inoculated into each of six mammary glands (three on the left side and three on the right side) in 4‑ to 5‑week old female BALB/cAJcl-nu/nu athymic nude mice. Tumor size was measured at 4–37 days after inoculation, and the tumor volume (mm3) was calculated as (π / 6) × L × W 2, where L and W are the major and minor axial lengths in millimeters, respectively. For in vitro experiments, 3 × 104 cells were plated into each well of a 12-well plate. At every 24–196 h after plating, cells were counted using a Coulter Z‑1 counter (Beckman Coulter, Fullerton, CA). Growth curves were fitted against the data for the tumor volume (in vivo) at 22–37 days after inoculation and the cell numbers (in vitro) at 24–120 h after plating to the exponential equation y = a * exp (b * x), where y is the tumor volume (mm3) or the cell numbers, x is time in days for the tumor volume and in h for the cell numbers, a is intercept, and b is slope. Doubling time (T D) in days for the tumor volume and h for the cell numbers was calculated as ln (2 / b).

Alamar Blue assay was also performed in vitro as described [22]. In brief, 3 × 103 cells were plated into each well of a 96-well plate. At every 24–120 h after plating, 10% Alamar Blue in PBS (Molecular Probes, Eugene, OR) was added, and cells were incubated at 37˚C/5% CO2 for 4 h, followed by measurements of fluorescence with an excitation wavelength at 560 nm and an emission wavelength at 590 nm using a microplate reader (Varioskan® Flash, Thermo Fisher Scientific, Vantaa, Finland). For each cell line, the data for 48–120 h time points are shown as fold changes relative to the data for the 24 h time point.

Wound healing assay

A wound healing assay was conducted as described [23]. Briefly, 5 × 105 cells were plated into each well of a 12-well plate. At 24 h after plating, confluent cell monolayers were scrape-wounded with a 200 µl micropipette tip (H-110-96RS, QSP, Petaluma, CA). At 0, 4, 8 and 22 h after wounding, images were captured with an Olympus IX71 microscope equipped with an UPIan FLN10X/0.3 objective and DP2-BSW software. The distance between wound edges was measured in three randomly chosen regions for each sample. The percentage of wound closure was calculated as 100 × (d 0 − d t) / d 0, where d 0 is the distance between wound edges at 0 h and d t is that at 4, 8 or 22 h after wounding.

Matrigel invasion assay

A Matrigel invasion assay was carried out using BD BioCoatTM MatrigelTM Invasion Chamber (BD Biosciences, Bedford, MA) according to the manufacturer’s instruction. In brief, for each well of a 24-well transwell chamber plate, 0.5 ml of RPMI1640 containing 1 × 104 cells but without FBS was added to the upper compartment, and 0.5 ml of RPMI1640 containing 5% FBS but without cells was added to the lower compartment. Chambers were incubated at 37˚C/5% CO2 for 22 h. Invasive cells having passed through to the lower surface of the membrane filter were fixed with methanol, stained with 0.05% crystal violet (Wako, Osaka, Japan), and counted.

Clonogenic survival assay

Cell survival was determined by the clonogenic survival assay as described [24, 25]. Briefly, 5 × 104 cells were seeded onto a 3.5-cm dish at 2 days prior to irradiation. Cells were irradiated with single, graded doses of X‑rays at room temperature from an X‑ray generator (MBR-1520R-3, Hitachi Medico, Tokyo, Japan) operated at 150 kV and 20 mA with a 1 mm aluminum equivalent filter, at a source to surface distance of 40 cm and a dose rate of 4 Gy/min. Cultures were reseeded into 6‑cm dishes in triplicate at 24 h after irradiation, and were incubated for 2 weeks, upon which cultures were fixed in methanol and stained with 0.05% crystal violet. Only colonies consisting of 50 or more cells were scored as survivors. Survival curves were fitted against the means of five or six independent experiments to the linear quadratic equation, predicated upon which the dose required to reduce the surviving fraction to 0.1 (10% survival dose, D 10) was determined. The dose modifying factor (DMF) was calculated as D 10 for NBOS0042, NLU0041 or NLN0104 cells divided by that for pGL4.5 cells.

TUNEL assay

Apoptosis was detected by the terminal deoxynucleotidyl transferase (TdT)-mediated dUTP-biotin nick-end labeling (TUNEL) method, using the ApopTag® Peroxidase In Situ Apoptosis Detection kit (S7100, Chemicon International), as described [26]. In brief, 2 × 104 cells were plated onto a 1.2-cm-diameter round coverslip (Matsunami Glass, Osaka, Japan) at 2 days prior to irradiation. At every 24–120 h after irradiation with 6 Gy of X‑rays at 4 Gy/min as aforementioned, cells were fixed on ice in 1% paraformaldehyde and permeabilized in 0.5% Triton X‑100. After endogenous peroxidase was quenched with 3% hydrogen peroxide, the coverslips were reacted with TdT enzyme and peroxidase-conjugated anti-digoxigenin antibody. Positive TUNEL staining was developed with diaminobenzidine and indicated by a brown precipitate. The cells were counterstained with 0.5% methyl green, and viewed under an Olympus IX71 microscope after mounting. For each sample, 1000 or more cells were analyzed.

Statistical analysis

All data were calculated as the means and standard deviations (SD) of three or more independent experiments. Statistical comparisons between groups were made by two-sample Student’s t‑test assuming unequal variances, and a p value of 0.05 or less was considered to be significant.

Results

Proliferation of metastatic cell lines is increased in vivo but not in vitro

The in vivo tumor growth of MCF7, pGL4.5 and NLN0001 cells was analyzed (Fig. 1). As expected, MCF7 cells that served as a negative control here did not grow significantly (e. g., p = 0.35 between 22 and 37 days after inoculation). In contrast, at ≥12 days after inoculation, pGL4.5 cells exhibited significant growth compared with MCF7 cells (p < 0.001), and growth of NLN0001 cells was significantly enhanced compared with pGL4.5 cells (p < 0.02). Next, the in vitro proliferation of pGL4.5, NBOS0042, NLU0041 and NLN0104 cells was evaluated with the two different methods, but there was no significant difference among cell lines (Fig. S3). These results indicate that metastatic cell lines possess an increased proliferative potential in vivo, but not in vitro.

Fig. 1
figure 1

Cell proliferation in vivo. Tumor volume (≥two mammary glands/mouse) was evaluated at 4–37 days after 1 × 106 cells (MCF7 [filled squares], pGL4.5 [filled circles] and NLN0001 [filled triangles] cells) were orthotropically inoculated into each of six mammary glands of athymic nude mice (three mice/cell line) (*p < 0.05, **p < 0.005 between NLN0001 and pGL4.5. # p < 0.001 between MCF7 and pGL4.5. Growth curves were fitted against the data for the tumor volume (mm3) at 22–37 days after inoculation to the exponential equation y = a * exp (b * x), where y, x, a and b are the tumor volume (mm3), time (days), intercept and slope, respectively. Doubling time (T D) in days was calculated as ln (2 / b). N.A. not available)

Full size image

Chemotaxis and invasion of metastatic cell lines are increased in vitro

Many studies have reported that metastatic cancer cells are more chemotactic and invasive than primary cancer cells [27,28,29]. Therefore, chemotaxis and invasion were assessed. The wound healing assay (Fig. 2) revealed that wound closure occurs in three metastatic cell lines (NBOS0042, NLU0041 and NLN0104 cells) significantly faster than in pGL4.5 cells at 4–22 h after wounding (p < 0.03), with no significant difference amongst three metastatic cell lines at 22 h after wounding (p > 0.2). A Matrigel invasion assay (Fig. 3) showed that the fraction of invasive cells was significantly higher in three metastatic cell lines compared with pGL4.5 cells (p < 0.044), without significant difference among three metastatic cell lines (p > 0.67). These results indicate that metastatic cell lines have an increased chemotactic (migratory) and invasive potential in vitro.

Fig. 2
figure 2

Chemotaxis in vitro tested with a wound healing assay. Wound closure was assessed at 0, 4, 8 and 22 h after confluent monolayers of pGL4.5, NBOS0042, NLU0041 and NLN0104 cells were scrape-wounded. The distance between wound edges was measured in three randomly chosen regions for each sample, and the data represent the means and SD of three independent experiments with triplicate samples (*p < 0.05, **p < 0.005 vs pGL4.5 cells. Scale bars, 200 µm)

Full size image
Fig. 3
figure 3

Invasion in vitro tested with a Matrigel invasion assay. Invasive cells that migrated across a Matrigel-coated porous membrane for 22 h were counted in pGL4.5, NBOS0042, NLU0041 and NLN0104 cells. The data are presented as the means and SD of three to five independent experiments with triplicate or quadruplicate measurements (*p < 0.05 vs pGL4.5 cells)

Full size image

Metastatic cell lines are more radioresistant in vitro

The clonogenic survival assay (Fig. 4) revealed that three metastatic cell lines (NBOS0042, NLU0041 and NLN0104 cells) are more resistant to X‑induced clonogenic inactivation than pGL4.5 cells (p < 0.007), with no significant difference amongst three metastatic cell lines (p > 0.11). DMF was 1.37, 1.39 and 1.32 for NBOS0042, NLU0041 and NLN0104 cells, respectively. TUNEL assay (Fig. 5) showed that three metastatic cell lines are more resistant to X‑induced apoptosis compared with pGL4.5 cells (p < 0.04), without significant difference amongst three metastatic cell lines (p > 0.1). These results indicate that metastatic cell lines are more radioresistant in vitro.

Fig. 4
figure 4

Sensitivity to X‑induced clonogenic inactivation in vitro determined with a clonogenic survival assay. pGL4.5 (left upper panel), NBOS0042 (right upper panel), NLU0041 (left lower panel) and NLN0104 (right lower panel) cells were replated for colony formation at 24 h after X‑irradiation at indicated doses. The data represent the means and SD of five or six independent experiments with triplicate measurements (*p < 0.01 vs pGL4.5 cells)

Full size image
Fig. 5
figure 5

Sensitivity to X‑induced apoptosis in vitro determined with TUNEL assay. pGL4.5 (circles), NBOS0042 (squares), NLU0041 (diamonds) and NLN0104 (triangles) cells were fixed for TUNEL staining at every 24 h up to 120 h after irradiation with 6 Gy of X‑rays. More than 1000 cells were analyzed for each sample, and the data are presented as the means and SD of three or four independent experiments with duplicate samples (*p < 0.05, **p < 0.005 vs NBOS0042, NLU0041 and NLN0104 cells)

Full size image

Discussion

Radiotherapy has been employed to treat metastasis. Whereas it is critical to understand the characteristics of metastatic cancer, the experimental animal model was unavailable that can compare the radiosensitivity of primary cancer and its metastatic counterpart. Here we have established a basal-like breast cancer cell line that can be visualized with bioluminescence imaging, and its derived cell lines that are highly metastatic to the bone, lung or lymph nodes by means of repetitive in vivo selections via orthotropic injections into the mammary fat pad or left ventricular cardiac injections (Table S1, Figs. S1 and S2). With this model, primary cancer and its metastatic counterpart can be compared directly both in vivo and in vitro. In addition, this model reflects more actual in vivo situations than other models, such as a bone metastasis model via a direct tumor implantation in bone or a lung metastasis model by tumor injection into a tail vein.

Our present data show that compared with pGL4.5 cells, NLN0001 cells have an increased proliferative potential in vivo (Fig. 1), and that the other three metastatic cell lines (NBOS0042, NLU0041 and NLN0104 cells) are more chemotactic (migratory), invasive, and resistant to X‑induced clonogenic inactivation and apoptosis in vitro (Figs. 2, 3, 4 and 5).

A significantly increased proliferation observed in vivo in NLN0001 cells was not observed in vitro in other three metastatic cell lines (NBOS0042, NLU0041 and NLN0104 cells) (Fig. 1 and Fig. S3). In this regard, stromal cells are reported to promote angiogenesis and growth of human prostate tumors in a xenograft model [30], and human cancer-derived fibroblasts are known to stimulate proliferation and migration of human breast cancer cells via leptin signaling leading to increased tumor progression [31]. Such a supply of leptin, cytokines or other secretory factors from the tumor stroma that occurs in vivo does not occur in vitro, and this may account for our observed difference in vivo and in vitro. To test this possibility and further understand a different proliferative response observed in vitro and in vivo, proliferation in vitro (in the absence or presence of cytokines or other related secretory factors) and in vivo of NLN0001, NBOS0042, NLU0041 and NLN0104 cells needs to be tested.

Our in vitro analyses showed that three metastatic cell lines (NBOS0042, NLU0041 and NLN0104 cells) have increased radioresistance, which may be accounted for, at least in part, by resistance to radiogenic apoptosis (Figs. 4 and 5). Such apoptotic radioresistance was observed despite increased proliferation in vivo. In this respect, human Namalwa Burkitt lymphoma cells that received continuous proteasome inhibition with bortezomib are known to exhibit an apoptosis resistant and hyperproliferative phenotype, where an increase in Hsp27 and a decrease in p53 accompany such apoptosis resistance [32]. These warrant the in vivo validation experiments to irradiate tumors arising from our metastatic cell lines, analyze apoptosis in situ, and examine its underlying mechanisms in vitro.

Here we used only one basal subtype cell line (i. e., MDA-MB-231 cells) to establish pGL4.5 cells, and established more than ten metastatic cell lines each for the bone, lung and lymph nodes from pGL4.5 cells, from which we used only three metastatic cell lines (i. e., NBOS0042, NLU0041 and NLN0104 cells, respectively) to minimize the number of mice subjected to experiments and due to limited manpower. However, further characterization of several metastatic cell lines other than NBOS0042, NLU0041 and NLN0104 cells, as well as the use of several basal subtype cell lines other than MDA-MB-231 cells, to establish independent metastatic cell lines will be important to generalize our present findings, warranting further studies.