Zoledronic acid alters hematopoiesis and generates breast tumor-suppressive bone marrow cells

Experiments were performed in 6 to 7-week-old female NCr-Nu (nude), CrTac:NCr-Foxn1 ^nu mice (Taconic Laboratories, Hudson, NY, USA) and 6 to 7-week-old female C57BL/6 mice (The Jackson Laboratory, Bar Harbor, ME, USA). All procedures were performed in accordance with the regulations of Boston Children’s Hospital Institutional Animal Care and Use Committee (protocol 12-11-2308R).

ZOL [1-hydroxy-2-(1H-imidazole-1-yl)ethylidene-bisphosphonic acid] (Zometa, Novartis Pharmaceuticals, Cambridge, MA, USA) was diluted to a working concentration of 4 mg/mL in 1X Hank’s Balanced Buffer Solution (HBBS, Gibco, 14065-056) and stored at 4 °C until use. A single dose of 100 μL of 1X HBBS or 100 μg/kg ZOL was administered to mice via intraperitoneal (i.p.) injection.

Femora and tibiae were dissected free into 2% FBS in PBS. Bone marrow cells (BMCs) were collected by centrifugation of mouse femora and tibiae at 6000–7000 g for 4 minutes at 4 °C. Cells were then incubated with red blood cell (RBC) lysis solution (BioLegend 420301) for 5 minutes on ice, washed once with BMC buffer, re-suspended in 0.5 mL of sterile BMC buffer, and passed through a 5-mL polystyrene round-bottom tube with a cell-strainer cap (Corning, 352235). One fourth of the spleen was mechanically dissociated with a razor blade, incubated with RBC lysis solution (BioLegend 420301) for 5 minutes on ice, washed once with PBS, and filtered through a 70-μM nylon mesh filter. Next, 200 μL of peripheral blood was incubated with RBC lysis solution (BioLegend 420301) for 10 minutes at room temperature, washed once with PBS passed through a 5-mL polystyrene round-bottom tube with a cell-strainer cap (Corning, 352235). At the experimental end points, blood was collected by intracardiac puncture with a 27-gauge needle into ethylenediaminetetraacetic acid (EDTA) Microtainer tubes (BD Pharmingen). Complete blood counts were obtained using a HEMAVET® hematology analyzer (Drew Scientific). Plasma was prepared by centrifugation of whole blood at 1500 g for 8 minutes at 4 °C.

BMCs (preparation as previously described) were incubated with appropriate antibodies for 30 minutes at 4 °C. Dead cells were eliminated from analysis using 7-aminoactinomycin D (7-AAD; BioLegend), and flow cytometry gating was used to exclude debris and cell clumps. Countbright Absolute counting beads (Life Technologies, C36950) were added to the samples to quantify the numbers of cells of any given lineage and report absolute differences between cohorts. Data were acquired on a Canto II or a FACSAria IIu/FACSDiva (BD Biosciences). Data analyses were performed using FlowJo software (TreeStar). Antibody information, dilutions, and cell gating strategies are included in Additional file 1: Table S1, Additional file 2: Table S2 and Additional file 3: Figure S1.

BrdU incorporation was performed per manufacturer’s instructions using the BD Pharmigen APC BrdU Flow Kit (Catalog number 552598). Mice were injected i.p. with 100 μL of 1 mg BrdU solution 2 hours prior to tissue harvest. BMCs (preparation as previously described) and cell-surface antigens were stained using the antibodies indicated in Additional file 1: Table S1. Cells were fixed and permeabilized with BD Cytofix/Cytoperm Buffer and treated with 100 μL of 300 μg/mL DNase to expose incorporated BrdU prior to staining with APC BrdU antibody (BD Biosciences, 557892).

BMCs, spleen cells, and peripheral blood cells (preparation as previously described) were passed through a 70-μM nylon mesh filter. BMCs (3 × 10⁴) were prepared and plated into 35-mm Petri dishes in Mouse Colony-Forming Unit (CFU) Methocult™ Assays (Stemcell Technologies Inc., M3434) according to the manufacturer’s instructions. Colonies were counted after 10 days of culture.

Femora were dissected free and fixed in 4% (wt/vol) paraformaldehyde (PFA) for 24 hours, stored in 70% ethanol for 24 hours, embedded in paraffin, and sectioned onto ProbeOn Plus slides (Fisher Scientific). Tibiae were dissected free and fixed in ice cold 4% PFA (pH 7.4 in PBS) for 48–72 hours and decalcified in 0.5 M EDTA/0.5% PFA (pH8, in PBS) for 2 weeks at 4 °C followed by paraffin embedding. Proliferating BMCs were identified by staining with an antibody to Ki67 (Abcam, catalog number ab15580, dilution 1:200) and a secondary Vectastain kit (ABC, Rabbit IGG, PK-6101) with AEC Chromogen Substrate (Dako, K3461). Tissues were counter stained with hematoxylin (Vector laboratories, VWR101098-062). Images were captured with identical exposure and gain for any given experiment, using a Nikon Eclipse Ni microscope. Staining was quantified using ImageJ (NIH) and CellProfiler (The Broad Institute) image analysis software.

Osteoblast and osteoclast number/mm trabecular bone was quantified on H&E or tartrate-resistant acid phosphatase (TRAP)-stained sections of tibiae covering all trabecular bone surfaces 200 μm away from the epiphysis, as described previously [21]. Osteoblasts were identified by their large Golgi complex, single, distinct nucleus, cuboidal shape and proximity to endosteal surfaces. Osteoclasts were identified by their bright pink appearance after TRAP staining. Two non-serial 3-μm-thick sections were assessed per sample using an Olympus BX53 (×20 objective) and OsteoMeasure software (Osteometrics). Detailed scoring methodologies are illustrated in Additional file 4: Figure S2A.

Proteoglycan-rich extracellular matrix in the epiphysis and metaphysis were visualized by toluidine blue staining on sections of tibiae. Two non-serial 3-μm-thick sections were assessed per sample using an Olympus BX53 a Leica DMRB microscope (×2.5 objective) and the area of interest (AOI) within a total tissue area of 1 mm² was calculated using OsteoMeasure software (Osteometrics; see Additional file 4: Figure S2B). Cortical and trabecular bone surfaces not directly connected to the dense network of trabeculi at the epiphysis were excluded from analysis.

The trabecular bone volume of femora was analyzed using a SkyScan 1272 device (SkyScan). Femora were scanned using 200 mA, 51 kV, a 0.5-mm aluminum filter, medium camera resolution of 2016 × 1344 and pixel size set to 4.3 μm. Images were reconstructed using NRecon software and a bone volume of interest (VOI) was determined by interactively drawing on the two-dimensional images. Analysis was started from a fixed offset 0.7 mm away from the proximal end of the growth plate covering a length of 1.5 mm. Grayscale images of the VOI were converted into binary images (threshold 95 − 225) and bone parameters were calculated using the CTAn software.

Tibiae were fixed in ice cold 4% PFA (pH 7.4, in PBS) for 4 hours at 4 °C, decalcified in 0.5 M EDTA (pH 8, in PBS) for at least 24 hours, transferred to 20% sucrose/2% polyvinylpyrrolidone (PVP) solution (in PBS) overnight at 4 °C followed by embedding in 8% gelatine/20% sucrose/2% PVP and 30-μm-thick sections were prepared using a cryostat at -20 °C as described by Kusumbe et al. 2014 [23].

Bone marrow vasculature was visualized by immunofluorescent staining using antibodies against the vascular endothelial cell marker Endomucin (1:100, Endomucin V.7C7, rat monoclonal, Santa Cruz, sc-65495) and CD31 (1:100, DIA-310: Anti-CD31 (Ms) from Rat (Clone: SZ31)). Alexa Fluor 555, goat anti-rat IgG, (LifeTechnologies, A21434 (1:200)) was used as the secondary antibody.

Images were acquired using a Nikon A1 confocal microscope, NIS-Elements-software Version 4.30, CFI Plan Fluor 20x MI (NA 0.75) or a Nikon inverted Ti eclipse, NIS-Elements software Version 4.30, Plan Apo 20x (NA 0.75). For a subset of samples, Z-stacks of 20-μm depth (0.9-μm intervals, 23 steps) were acquired and 3D projections reconstructed using Fiji (ImageJ, Version 2.0.0-rc-24/1.49 m, Image > Stacks > 3D projections). Brightness and contrast for the entire images were adjusted using Fiji or NIS-Elements-software. The Nikon Eclipse Ti microscope was used to produce tile scans (8 × 8). Aperio ImageScope software was used to track vessels positively stained for endomucin in an average area of bone marrow of 615 x 1095 μm from the top of the growth plate.

Human breast cancer MDA-MB-231 BO2F11 luciferase-transfected bone-tropic cells were a kind gift from the van der Pluijm laboratory (Leiden University Medical Center, The Netherlands) [24]. Cells tested negative for mycoplasma and were validated by short tandem repeat (STR) profiling by the Molecular Diagnostics Laboratory at the Dana-Farber Cancer Institute.

Mice were anesthetized with 2.5% isoflurane and injected i.p. with 150 mg/kg D-luciferin (Perkin-Elmer). Luciferase-positive cells were detected using a Xenogen IVIS imaging system (Caliper Life Sciences) 10 minutes after luciferin injection. Luminescence signal was detected for the regions of interest as radiance (p/sec/cm²/sr) and analyzed using the Living Image Software Version 4.1 (Caliper Life Sciences).

BMCs were freshly harvested from cohorts of donor mice (preparation as described previously) and immediately admixed with MDA-MB-231 BO2F11 tumor cells in DMEM with 10% Matrigel (BD, Growth Factor Reduced, 356230). For each injection, 750,000 BMCs from a given donor mouse were admixed with 250,000 tumor cells in 100 μL DMEM/10% Matrigel. BMCs from each donor mouse were distributed bilaterally into three recipient nude mice, with six possible tumors per donor. Tumor growth was monitored by in vivo imaging system (IVIS) imaging of all mice and caliper measurements were taken for all palpable tumors. Volume was estimated as 0.5 × length × width².

All data were analyzed with the use of GraphPad Prism Software (Version 6). Data are expressed as mean ± SEM. Results were analyzed using Student’s t test, unless otherwise indicated, and were considered statistically significant if the p value was ≤0.05.

To determine whether ZOL impacts hematopoiesis, we used two different strains of mice - nude and C57BL/6 - that are commonly used in breast cancer research. While patients with osteoporosis or metastatic bone disease are often treated for chronic disease [25], our goal was to evaluate the effects of ZOL on hematopoiesis in the absence of overt bone disease. We also reasoned that effects on hematopoiesis should be analyzed over a time period when ZOL is known to be bioavailable in the bone. ZOL is known to concentrate in the bone within 24 hours of administration and is cleared during bone turnover, which occurs at a rate of around 0.7% per day in the mouse and thus, takes 2 weeks to complete [26]. Hence, we administered a single, clinically relevant dose of 100 μg/kg ZOL (comparable to the 4-mg clinical dose that has been well-established to inhibit osteoclast activity in vivo [21]) to cohorts of immunocompromised (nude) and immunocompetent (C57BL/6) mice and analyzed hematopoietic cells at various time points over a course of 2 weeks (Fig. 1a).

In the nude mice, the numbers of HSCs (LSK/CD150+/CD48-/CD34-/Flt3-) in the marrow of ZOL-treated mice were elevated 1.8-fold (p = 0.003) relative to the control cohort 3 days after administration, after which the numbers returned to baseline (Fig. 1b, Table 1). In agreement with these results, numbers of both short-term HSCs (ST-HSCs, LSK Flt3+ CD34-) and long-term HSCs (LT-HSCs, LSK Flt3- CD34-) were elevated 4.1-fold (p = 0.01) and 4.8-fold (p = 0.06), respectively (Table 1).

Similar results were observed in C57BL/6 mice; however, the maximal effect on HSC populations occurred 2 days later than in the nude mice. Specifically, no differences were observed between the control and the ZOL-treated cohorts 3 days after administration, while at the 5-day time point, there was a 1.8-fold increase in HSCs (p = 0.04), a 2.0-fold increase in LT-HSCs (p = 0.03), and a 1.7-fold increase in ST-HSCs (p = 0.05), after which these numbers returned to baseline and remained stable until the experimental end point at day 20 (Fig. 1g, Table 1).

To determine whether increased cell proliferation could account for the observed changes in HSCs numbers, we injected ZOL-treated and vehicle-treated mice with BrdU and analyzed HSCs by flow cytometry. The BrdU+ HSCs as a percentage of total HSCs were not significantly different between the ZOL-treated and control-treated cohorts at any time point analyzed, even at time points as early as 24 and 48 hours after ZOL treatment (Fig. 1c, Additional file 5: Figure S3d). These data suggest that while ZOL treatment results in a significant expansion of HSCs in the marrow, the percentage of proliferating cells is constant and therefore, the increase in numbers of HSCs is likely due to differences in their proliferation rates rather than to HSCs amassing in a quiescence phase.

Given that the HSC numbers were significantly elevated at 3 days (nude mice) and 5 days (C57BL/6 mice) after ZOL treatment, we focused on these time points to determine if ZOL was increasing the numbers of functional HSCs. To do so, we performed CFU assays using BMCs isolated from both vehicle-treated and ZOL-treated mice at these time points. We found that ZOL significantly increased the number of CFUs in both strains of mice (nude mice, day 3: control 5.2 ± 1.4, ZOL 10.4 ± 1.6, p = 0.04; C57BL/6 mice, day 5: control 11.6 ± 4.0, ZOL 21.6 ± 5.7, p = 0.05) (Fig. 1d). Additionally, no increases in the number of CFUs were observed in the spleen or peripheral blood 3 days after ZOL administration as compared to control treatment, indicating that the effect of ZOL was specific to the bone marrow (Additional file 5: Figure S3c).

To determine whether the increased HSC function could be ascribed to direct effects of ZOL on HSCs, we isolated BMCs from naïve mice and subjected them to the CFU assay in the presence of either vehicle or 10 μM ZOL in vitro (a dose comparable to the 100 μg/kg dose used in vivo). Ten days after treatment, no differences in the number of CFUs between the groups were observed (Fig. 1e), suggesting that ZOL did not act directly on HSCs to mediate their expansion in vivo and that this expansion was not attributed to direct drug storage and release from the bone matrix.

We next determined whether ZOL affected the proliferation of other bone marrow hematopoietic cells. To start, we stained sections of tibiae with an antibody specific for the proliferation marker Ki67. The numbers of Ki67+ cells per field in the marrow of ZOL-treated nude mice were increased by around 82% (p < 0.0001) and in C57BL/6 mice by around 74% (p = 0.0007), compared to the respective vehicle controls (Fig. 1f). These results prompted us to carry out a more detailed analysis of the specific hematopoietic progenitor cell (HPC) populations at various time points after administration of control treatment or ZOL, using flow cytometric analysis (Additional file 1: Table S1, Additional file 2: Table S2 and Additional file 3: Figure S1).

Compared to control treatment, ZOL had significant effects on a number of BMC populations in the nude mice (Table 1, Fig. 1g). Specifically, there was a 1.3-fold increase in the Lin-Sca1+cKit+ (LSK) populations (p = 0.02) accompanied by a non-significant 1.8-fold increase in multipotent progenitors (MPPs, LSK/CD150-/CD48+) 3 days after ZOL-treatment (p = 0.1). Common myeloid progenitor cell (CMP, Lin-Sca1-cKit+IL7Ra-CD34 + FcγRII/III-) numbers were elevated 2.0-fold (p = 0.03) and 3.4-fold (p = 0.01) at the 3-day and 5-day time points, respectively. Megakaryocyte-erythroid progenitors (MEPs, Lin-Sca1-cKit + IL7Ra-FCγRII/II-CD34-) were elevated 1.8-fold (p = 0.03) on day 5. Numbers of granulocyte-macrophage progenitors (GMPs, Lin-Sca1-cKit + IL7R-CD34 + FcγRII/III+) appeared to be elevated 5 days after ZOL treatment compared to the control cohort, although this change was not statistically significant (1.4-fold increase, p = 0.2). Common lymphoid progenitors (CLPs, Lin-/sca^int/ckit^intILRa+), which are present in nude mice, were decreased 1.5-fold (p = 0.04) at the 5-day time point in the marrow of ZOL-treated mice. In all cases, the numbers of these various progenitor cells returned to baseline by day 10.

In the C57BL/6 mice, the changes in HPC numbers followed some of the same trends as those observed in the nude mice, although with different kinetics (Fig. 1g, Table 1). The numbers of MPPs were increased 1.4-fold by day 5 (p = 0.004) and 1.3-fold by day 10 (p < 0.0001). CMPs and MEPs were elevated 1.5-fold (p = 0.1) and 1.6-fold (p = 0.07), respectively, after 15 days, although these numbers were not statistically significant. GMPs were significantly elevated 1.3-fold (p = 0.02) at the 15-day time point in the marrow of ZOL-treated mice. In all cases, the numbers of these various progenitor cells returned to baseline by day 20. Additionally, we observed a significant 2.2-fold (p = 0.0003) decrease in the number lymphoid biased progenitor (LBP) populations in the C57BL/6 mice 5 days after ZOL treatment, after which these numbers were not statistically different from those in the control cohort. In these C57BL/6 mice, we also observed significant 3.1-fold (p = 0.0001) and 2.5-fold (p = 0.003) increases in the number of common lymphoid progenitor (CLP) populations 5 and 10 days after ZOL treatment, respectively. The decrease in LBPs and increase in CLPs indicate that it is possible that the effect of ZOL on the bone marrow drives LBPs toward differentiation into the downstream CLP populations.

Collectively, these results established that ZOL has profound and transient effects on hematopoiesis in both nude and C57BL/6 mice. In particular, the effects of ZOL appeared to favor differentiation toward common lymphoid progenitors and expansion of myeloid lineage progenitors, which may be accounted for by the significant expansion in HSCs. Additionally, our results suggest that the effects of ZOL on HSC proliferation and expansion are indirect.

A number of reports have demonstrated that bisphosphonate treatment reduces circulating monocytes and neutrophils [27, 28]. It is also well-established that hematopoiesis in the marrow compensates for significant alterations in circulating blood cell numbers [29]. Hence, we next investigated how the changes to hematopoietic cells in the marrow related to peripheral blood counts during the approximate 2-week time period after ZOL administration.

We analyzed circulating blood cell populations at various time points after ZOL or control treatment in both strains of mice. In the C57BL/6 mice, there was around 49% reduction in neutrophils (p = 0.05) and around 83% reduction in monocytes (p = 0.003) in the blood of the ZOL-treated cohort relative to the control cohort on day 5 (Fig. 1g, Table 2). Platelets were increased compared to control at 3, 5 and 10 days after ZOL administration, with approximately 49% elevation observed at the 5-day time point (p = 0.01) (Fig. 1g, Table 2). Moreover, on day 5 there were significant reductions in the numbers of eosinophils (p = 0.03) and basophils (p = 0.01) in the ZOL-treated mice (Additional file 5: Figure S3b).

In nude mice, we observed a trend toward reduced numbers of circulating neutrophils and monocytes with ZOL treatment relative to vehicle control, although these numbers were not statistically significant (Fig. 1g, Table 2). There were also no statistically significant differences in the numbers of other blood cell populations between cohorts at these time points in the nude mice (Table 2, Additional file 5: Figure S3a).

Taken together with the bone marrow analysis, these results indicate that reductions in peripheral neutrophils and monocytes were accompanied by an increase in their progenitor cells in the bone marrow after administration of a single, clinically relevant dose of ZOL. In the C57BL/6 mice, the increase in peripheral blood platelets observed 5 days after ZOL treatment preceded the increase in MEPs observed in the bone marrow, which was observed 15 days after ZOL treatment. Additionally, although we observed decreases in the LBP and increases in the CLP populations in the bone marrow of C57BL/6 mice 5 days after ZOL treatment, there was not a corresponding change in the number of lymphocytes in the peripheral blood observed at this same time point.

As HSC maintenance and hematopoiesis are tightly regulated by cells in the endosteal niche, in particular osteoblasts and osteoclasts, we endeavored to determine whether the ZOL-induced changes in hematopoietic cell populations occurred in conjunction with the well-established effects on these particular bone cells. We therefore analyzed osteoclast and osteoblast parameters in the cohorts of nude mice at the time points at which we had analyzed hematopoietic stem cell populations.

As expected, ZOL caused a significant reduction in osteoclast activity, as measured by N-telopeptide of type 1 collagen (NTX) plasma concentrations, at both 3 and 5 days after treatment, with levels returning to baseline by day 10 (day 3: p < 0.0001, day 5: p = 0.0005, day 10: p > 0.999; Fig. 2a). The decrease in osteoclast activity observed at day 3 was accompanied by an 11.7-fold reduction in the number of osteoclasts lining trabecular bone surfaces of the ZOL-treated cohort relative to the control-treated cohort (p < 0.0001; Fig. 2b, c).

Although the long-term therapeutic effects of bisphosphonates are thought to be a consequence of direct inhibition of osteoclast-mediated bone resorption, there is increasing evidence that these drugs also affect osteoblasts, due to the tight coupling of osteoclast and osteoblast activity [30, 31]. We observed a transient but significant reduction in plasma procollagen I N-terminal propeptide (PINP) (a measure of osteoblast activity) 3 and 5 days after ZOL administration, after which it was not significantly different relative to the control cohort (day 3: p < 0.0001, day 5: p < 0.0001, day 10: p = 0.4, Fig. 2d). Concomitant with these findings, the number of osteoblasts lining trabecular bone surfaces was reduced 3.4-fold relative to control (p = 0.007, Fig. 2e, f).

To establish whether these changes in bone cell number and activity manifested in alterations to bone structure, we analyzed trabecular bone volume 3 days after ZOL administration. As expected, no significant alterations in trabecular bone volume or number were observed at this early time point (Fig. 2g-i). However, as measured by proteoglycan staining, we observed increased extracellular matrix deposition in the metaphysis 3 days after treatment (1.3-fold increase relative to control, p = 0.02) with a further increase 8 days after ZOL treatment (1.5-fold increase relative to control, p = 0.001, Fig. 2j, k). Although gross total bone density had not yet been measurably affected at these relatively early time points, increases in extracellular matrix deposition in the metaphysis in this study was consistent with our previous finding that ZOL increases trabecular bone extracellular matrix deposition and that increases in bone volume are not observed until 10 days following treatment [21].

These results established that a single dose of ZOL decreases the number and activity of osteoblasts and osteoclasts within a few days of administration, after which activity returns to baseline. Moreover, although gross changes in bone density were not apparent, significant changes to the bone extracellular matrix had begun to manifest. The kinetics of these changes to bone mirrored those with which alterations to hematopoietic progenitors (described above) had occurred in the wake of ZOL treatment.

Proliferative HSCs are reported to reside in peri-vascular niches within the bone [32, 33]. Moreover, it has been suggested that ZOL has anti-angiogenic activity in vitro and reduces circulating pro-angiogenic vascular endothelial growth factor (VEGF) in patients with cancer [34‐37]. Based on these findings, and on a recent report that HSC maintenance is regulated by specific bone marrow blood vessel types [23], we investigated the effects of ZOL on the bone microvasculature. To do so, we stained bone tissue sections from ZOL and control treated mice for the endothelial cell marker endomucin, which is expressed by the vessels in both the metaphysis and diaphysis [23].

In the control cohort, we observed a distinct organization of endomucin-positive microvasculature, with elongated, branched tubular vessels occupying the metaphysis and a more sinusoidal patterning in the diaphysis, as expected. Although no quantitative difference was observed in the number of vessels and average length of vessels (Additional file 6: Figure S4b), the endomucin-positive vasculature of ZOL-treated animals appeared to comprise shorter, discontinuous vessels with a deficit of sinusoidal patterning in the diaphysis and an apparent elongation of the branched zone within the metaphysis (Fig. 3a-f, Additional file 6: Figure S4a).

Using a second endothelial cell-surface marker, CD31, which is expressed by the vessels in the metaphysis but is absent in the diaphysis [23], we observed no overt differences in organization and structure of the CD31⁺ bone marrow vasculature (Additional file 6: Figure S4a) in the ZOL-treated cohort when compared to control; however, the zone occupied by columnar vessels appeared to be extended, in agreement with the endomucin staining (Fig. 3e). VEGF plasma levels were not affected 3 days after a single dose of ZOL or control (p = 0.3, control: 266.2 ± 8.80 pg/mL vs ZOL: 251.7 ± 7.92 pg/mL, Fig. 3g), indicating that ZOL did not change VEGF plasma levels at this time point.

Although not easily quantifiable, our observations support the notion that a single dose of ZOL modifies the organization of the bone microvasculature in areas known to comprise HSC niches and thus could, at least in part, account for the expansion in HSCs that we observed 3 days after ZOL administration. Whether these ZOL-induced modulations to the microvasculature have significant effects on hematopoiesis will be the subject of future studies.

In recent clinical trials, adjuvant ZOL has been correlated with reduced breast cancer recurrence, although the precise mechanism(s) of action for this protective effect remains unknown [25]. Given that hematopoietic cells play an important role in breast cancer progression [38, 39] and that our findings demonstrated that ZOL has profound effects on hematopoietic stem and progenitor cells, we wished to determine whether the ZOL-induced alterations in hematopoiesis have an effect on breast tumor outgrowth. We based our modeling on the notion that the outgrowth of disseminated cancer cells encountering a bone marrow environment that had been modified by ZOL could be profoundly affected by those ZOL-induced changes to hematopoietic cells that we had observed.

To exclusively test the effects of ZOL-modified bone marrow in the absence of overt bone disease, we compared the ability of BMCs isolated from control-treated and ZOL-treated animals to modulate tumor cell growth using our previously reported BMC functional assay [40] (Fig. 4a). To do so, BMCs were isolated from mice at selected time points after control or ZOL treatment and admixed with MDA-MB-231 bone-tropic (BO2F11) human breast tumor cells in a ratio of 3:1. The admixtures were immediately injected subcutaneously into recipient cohorts of nude mice and tumor incidence and progression were assessed over a period of 14 days (Fig. 4a). In all cases, we confirmed that the vast majority of cells in the BMC preparations were CD45+ hematopoietic cells, with <0.005–0.02% of the cells being CD45-negative (data not shown).

We first started with BMCs harvested at those time points at which maximal effects on hematopoiesis were observed, namely, 3 days in nude mice and 5 days in C57BL/6 mice. BMCs prepared from the control cohort of nude donor mice 3 days after vehicle treatment had no effect on tumor growth; the incidence of tumor formation was approximately 83% in the cohort of mice that had received these admixtures and in a cohort of mice that had been injected with tumor cells only (i.e., without any admixed BMCs) (Fig. 4b, c). In striking contrast, BMCs prepared from nude mice 3 days after ZOL treatment significantly inhibited tumor outgrowth; only 50% of the injected admixtures formed tumors (Fig. 4b). Similarly, tumor incidence was only approximately 17% in the mice that had received donor BMCs harvested from C57BL/6 mice 5 days after ZOL treatment while tumors formed in approximately 83% of the mice that had received control BMCs (Fig. 4b). These results established that ZOL generates tumor-suppressive bone marrow but did not reveal whether the changes we had observed in hematopoiesis were correlated with the inhibitory function of the bone marrow.

We reasoned that if ZOL-induced changes to hematopoiesis could define the tumor-inhibitory bone marrow, then BMCs isolated from control-treated and ZOL-treated mice at time points when HSC and HPC counts had returned to baseline (by 10 days in nude mice and 15 days in C57BL/6 mice) should no longer have tumor suppressive capacity. Interestingly, however, BMCs isolated from ZOL-treated donor mice at these later time points after treatment also had tumor-inhibitory function. Specifically, BMCs harvested from nude donor mice 10 days after ZOL treatment inhibited the outgrowth of MDA-MB-231 bone-tropic tumor cells (approximately 12% incidence) as compared to control-treated donor bone marrow (approximately 88% incidence) (Fig. 4b, c). Likewise, donor BMCs harvested from C57BL/6 mice 15 days after ZOL treatment resulted in decreased incidence of tumor formation (approximately 50%) relative to the control-treated donor cohort (100%) (Fig. 4b). Furthermore, in those tumors that did form, 5-day ZOL-treated BMCs in the nude mice reduced tumor growth kinetics compared to control-treated bone marrow cells (Additional file 7: Figure S5a) and 10 day ZOL-treated BMCs in the nude mice and 5 day ZOL-treated BMCs in the C57BL/6 mice reduced tumor size compared to control-treated bone marrow cells (Fig. 4b and Additional file 7: Figure S5b).

To rule out the possibility that tumor growth inhibition was the consequence of some residual drug that may have been retained in the donor bone marrow and introduced during the admixing process, we tested whether the tumor cells were susceptible to ZOL treatment. First, MDA-MB-231 BO2F11 cells were treated in vitro with a dose comparable to the dose used in vivo and no statistically significant differences were observed in the ZOL-treated cells compared to the control-treated cells in the cell proliferation growth assays (Additional file 7: Figure S5c). Second, to determine the likelihood that ZOL affected tumor cells in vivo, mice were administered one dose of ZOL 3 days prior to subcutaneous injection with MDA-MB-231 bone-tropic cells, and then re-dosed with ZOL every 5 days. Tumor growth was monitored over 14 days. There were no statistically significant differences in total tumor burden at the experimental end point in the ZOL-treated cohorts compared to the control cohorts (Fig. 4d), indicating that ZOL does not directly inhibit growth of these tumor cells, consistent with earlier reports [41].

As mammary tumors may also impact hematopoiesis, we wished to determine if the impact of ZOL on hematopoiesis occurred in the presence of a primary mammary tumor. We injected MDA-MB-231 bone-tropic cells orthotopically and after 30 days of tumor growth, we treated the mice with control or ZOL for 3 days before assessing the bone marrow by flow cytometry. Consistent with our earlier observations (Fig. 1) we observed expansion of the LSK compartment with ZOL, even in the presence of a primary mammary tumor (Fig. 4e).

Taken together, our findings established that ZOL treatment generates tumor-suppressive bone marrow cells. Moreover, these findings indicated that the tumor-suppressive action of ZOL is longer-lasting than the transient alterations to bone and hematopoietic cell numbers.