Effective Treatment of Established Bone Metastases Can Be Achieved by Combinatorial Osteoclast Blockade and Depletion of Granulocytic Subsets

Bone is the third most common site of metastasis for a variety of solid tumors including lung, breast, prostate, colorectal, melanoma, and others, with about 70% of patients with metastatic prostate and breast cancers harboring bone metastasis (1, 2). In addition to leading to significant morbidity and mortality, bone metastases lack effective therapeutic approaches, and the most successful treatments for these patients are aimed at reducing skeletal-related events (SRE), such as fractures, bone pain, hypercalcemia, and spinal cord compression (3–6).

Significant effort has been devoted towards the understanding of the mechanisms of cancer cell dissemination into the bone. The “vicious cycle,” originally proposed by Mundy and Guise, proposes an intimate interaction between tumor cells and bone-resorbing cells [osteoclasts (OC); refs. 7–9]. Factors released by tumor cells reaching the bone microenvironment drive osteoclastogenesis, and OC-driven bone destruction releases bone-stored factors that in turn stimulate proliferation and recruitment of more cancer cells to the bone microenvironment (7, 8). Breaking the tumor/bone vicious cycle by targeting OC activation with antiresorptive therapies at disease onset has been effective in reducing tumor burden and SREs in preclinical studies (10). However, in the clinical setting, zoledronic acid (ZA), a bisphosphonate compound suppressing bone resorption, or neutralization of receptor activator of nuclear factor kappa-Β ligand (RANKL), the key osteoclastogenic cytokine, fails to show antitumor effects in patients with breast cancer, with only a subset of postmenopausal subjects showing improved overall survival following ZA administration (11, 12). These clinical findings suggest that in addition to the OCs, other cells are likely to be involved in supporting tumor growth in bone.

The immune system plays a pivotal role in the surveillance against tumor progression by recognizing transformed cells expressing abnormal proteins (tumor antigens) and killing them. Tumor cells can, however, develop diverse strategies to escape immune surveillance, ranging from secreting immune suppressive cytokines and recruiting immune-suppressor populations, helping cancer cells to escape killing by natural killer (NK) and T cells, to expressing molecules that activate immune checkpoints, and thus impair antitumor immunity (13). Myeloid-derived suppressor cells (MDSC) are immature myeloid populations originating in the bone marrow, specialized in suppressing antitumor immune responses. MDSC accumulation correlates with poor prognosis in a variety of cancers (14, 15). MDSCs are generally identified by the coexpression of α_M integrin (CD11b) and Gr1 in mice or CD33 in humans (14). This myeloid population is further subdivided into granulocytic (Ly6G⁺Ly6C^low in mice, CD15⁺ in human) or monocytic (Ly6G⁻Ly6C^high in mice, CD14⁺ in human) subsets with distinct abilities to suppress antitumor T-cell responses (16, 17). Due to the expression of markers shared by other myeloid populations, in this article, we use the terminology “MDSC” only for CD11b⁺Gr1⁺ cells with proven ability to suppress antitumor immune responses; otherwise, we refer to them based on their surface markers.

MDSC depletion and/or their functional inactivation have been proven to be effective in reducing tumor burden in a variety of tumor and metastasis models (18, 19). However, the role of MDSCs in establishing and maintaining bone metastases has only become apparent in recent years. Bone marrow CD11b⁺Gr1⁺ cells isolated from tumor-bearing mice have increased osteoclastogenic potential, support tumor growth, and promote osteolytic lesions in a process limited by administration of ZA (20, 21). We previously reported that increased MDSC numbers can enhance tumor growth in the bone by reducing antitumor T-cell responses, even in mice with impaired OC differentiation (22). These studies indicate that CD11b⁺Gr1⁺ or MDSC populations can support tumor growth in bone via multiple mechanisms, either by becoming OCs and contributing to tumor-induced bone loss, thus perpetuating the tumor/bone vicious cycle, or by creating an immune suppressive environment where tumor cells grow unabated. Nevertheless, the efficacy of MDSC targeting in the context of bone metastases has not been fully evaluated.

In this study, we demonstrate that anti-Gr1 delivery to deplete CD11b⁺Gr1⁺ populations, although effective in reducing established primary soft tissue tumors, fails to reduce tumor burden in bone due to accumulation of OC precursors. Amelioration of established skeletal metastases can be achieved through the combined actions of Gr1⁺ cell depletion and OC inhibition. We also find that patients with breast cancer with progressive bone metastases, despite antiresorptive treatment, have increased granulocytic populations compared with patients with stable skeletal metastatic disease. This study suggests that granulocytic and OC populations together contribute to progression of skeletal metastases.

Firefly luciferase (FI)–conjugated murine Lewis lung carcinoma cells (LLC; C57BL/6, LLC-Fl), B16-F10 murine melanoma cells (C57BL/6, B16-Fl), and 4T1 murine mammary tumor cells (BALB/c, 4T1-Fl) were cultured at 37°C with 5% CO₂ in complete media: DMEM, high glucose (Gibco, catalog no. 11965) supplemented with 2 mmol/L L-glutamine (Gibo, catalog no. 25030), streptomycin (100 μg/mL), penicillin (100 IU/mL; Gibco, catalog no. 15140), 1 mmol/L sodium pyruvate (Corning, catalog no. 25–000-CI), and 10% FBS (Amizona Scientific LLC, catalog no. AM-SM-002). All the cell lines above are generous gift from Dr. Katherine N. Weilbaecher (Washington University, St. Louis, MO), originally purchased from ATCC. The species of origin and identity of each cell line was validated by karyotyping and full exome sequence analysis. All cell lines were tested for Mycoplasma every 2 months. Aliquots for each cell line were used for maximum 1 month after initial thaw.

C57BL/6 or BALB/c wild-type (WT) mice were purchased from The Jackson Laboratory. Animals were housed in a pathogen-free animal facility at Washington University (St. Louis, MO). All animal studies (6- to 8-week-old mice were used) were conducted in accordance with, and with the approval of an Institutional Animal Care and Use Committee (IACUC). Because no differences in tumor growth were noted between males and females based on initial experiments, animals of both sexes were used for subsequent experiments.

To establish tumors, B16-Fl, LLC-Fl, and 4T1-Fl tumor cells were suspended in PBS and either inoculated subcutaneously (s.c.; 10⁵ cells) in the left flank, injected into the right tibia (10⁴ cells; intratibial, i.t.), or injected in the left cardiac ventricle (10⁴ cells; intravenous, I.v.) of sex- and age-matched mice (C57BL/6 for B16-Fl and LLC-Fl tumor lines, BALB/c for 4T1-Fl tumor line). All animals were sacrificed and tissues collected 2 weeks after tumor inoculation. For s.c. tumors, tumor measurements were performed every 2 or 3 days with a caliper, and volumes were calculated using the following formula: V = 0.5 [length (mm) × width (mm)²]. For i.t. and I.v. tumors, growth curves were determined by bioluminescence, as described below. The mAb anti-mouse anti-Gr1 (clone RB6–8C5, BioXCell) was used to deplete Gr1⁺ (Ly6G/Ly6C) myeloid cells, and rat IgG2b isotype control (clone LTF-2, BioXCell) was used as antibody control. Treatment consisted of intraperitoneal (i.p.) injections of anti-Gr1 and control antibody at a concentration of 12.5 mg/kg three times a week for 2 weeks starting either at the time of tumor inoculation (prophylactic setting) or 7 days later (therapeutic setting). ZA was used to block OC activity in vivo. ZA (Novartis Pharma AG) was administered subcutaneously 7 days after tumor inoculation at the dose of 0.75 μg per mouse. This dose of ZA was designed to produce drug levels similar to those achieved with the clinical dosing regimen of 4 mg Zometa for the treatment of bone metastases (23).

Tumor growth in bone (from i.t. and I.v. tumor models) was monitored by bioluminescence imaging (BLI) every 2 to 3 days starting 7 days after tumor inoculation using an IVIS 50 (PerkinElmer; Living Image 4.3.1, with exposures of 1 second to 1 minute, binning 2 to 8, FOV 12.5 cm, f/stop1, & open filter). D-luciferin (150 mg/kg in PBS; Gold Biotechnology) was injected into the mice i.p. and imaged using isoflurane anesthesia (Pivetal, NDC 46066–755–04; 2% vaporized in O₂) with isoflurane vaporizer (Highland Medical Equipment). The total photon flux (photons/second) was measured from regions of interest (ROI) using the Living Image 3.2 program (Caliper Life Sciences). For I.v. injections, mice with extrapleural intrathoracic tumors were excluded from analysis. Because tumor cells can migrate to both legs, sum of the bioluminescence total flux for both legs was calculated and used to graph tumor growth curves.

All human samples were obtained in accordance with guidelines set by the Institutional Review Board of Washington University (IRB ID#: 201102244) and followed federal and state guidelines. All participants gave written informed consent under the IRB-approved protocol prior to inclusion in the study, including access of archival tumor tissue for research. Samples were deidentified prior to sharing with collaborators. All studies were conducted in compliance with the Declaration of Helsinki.

Patients had diagnosis of estrogen receptor (ER)–positive, HER2-negative breast cancer, stage IV, with bone metastases, prior to first-line systemic therapy for metastatic breast cancer or had prior therapy for metastatic breast cancer but met the following criteria: (i) no prior chemotherapy or immune therapy in the past 2 months, (ii) patients currently stable or progressing on hormonal therapy or hormonal therapy combination or starting hormonal therapy or hormonal therapy combination, and (iii) no limitation on the number of prior hormonal therapy or chemotherapy treatments. Radiologic tumor assessment was required within 1 month prior to or after the collection of the baseline blood sample to serve as the baseline tumor assessment. Exclusion criteria included: uncorrected coagulopathy, bleeding tendency, or other conditions that might increase the risk of a biopsy, blood draw, or other procedure; any reason that would make the patient unlikely to comply with study requirements or be incapable of providing appropriate consent (e.g., confusion, infirmity, alcoholism, etc.). Prior history of other invasive malignancies was not an exclusion criterion, unless the disease was active and progressing at the time of protocol screening.

Peripheral blood was collected at enrollment, every 3 months (to coincide with regular imaging), and time of disease progression from 9 patients with stage IV HER2-negative, ER-positive ductal carcinoma with bone metastases being treated with ZA (7 patients) or denosumab (anti-RANKL, 2 patients) and established skeletal metastases (metastases at non–skeletal sites were either stable or absent at the time of sample collection) on standard-of-care treatment under an IRB-approved protocol (IRB ID#: 201102244).

To isolate peripheral blood mononuclear cells (PBMC), EDTA-treated whole blood was diluted to a volume of 20 mL with PBS, transferred to a 50 mL conical tube, and underlaid with 15 mL of Ficoll (Atlantal Biologicals). Tubes were centrifuged at 400 × g for 30 minutes. The PBMC fraction was collected at the interface layer and washed three times with 40 mL of PBS. After counting, a minimum of 5 × 10⁶ PBMCs were frozen in 10% [volume for volume (v/v)] DMSO (Sigma-Aldrich, catalog no. D5879) in FBS and stored at −80°C for subsequent analysis as described below. Patient and disease information, including metastatic sites and treatment regimen, are included in Table 1. Patients were divided into 2 groups based on routine skeletal CT or bone scintigraphy scans. As part of standard protocol, restaging CT scans of the chest, abdomen, and pelvis were performed approximately every 3 months, and nuclear medicine (NM) bone scintigraphy scans were performed every 6 months. Bone imaging studies corresponded closely in timing (within 2 weeks) to the blood samples used for flow cytometry analysis.

Immediately upon sacrifice, single-cell suspensions were prepared from bone marrow, spleen, and tumor. In brief, bone marrow cells were harvested from tibias and femurs by centrifugation, whereas spleens were mechanically disassociated with 1 cc syringe plunger, and individual cell suspensions obtained through 70-μm cell strainer. Tumor tissues were minced and digested with collagenase A (3.0 mg/mL; Roche) and DNase I (50 U/mL; Sigma-Aldrich) in serum-free media for 30 minutes at 37°C vortexed for 10 seconds every 15 minutes. Cells were filtered through 40-μm nylon strainers (Thermo Fisher Scientific) and washed twice in PBS with 2% FBS. Red blood cells were then removed with red blood cell lysing buffer Hybri-Max (Sigma-Aldrich, catalog no. R7757). Cells were washed once and stained in PBS with 0.5% BSA, 2 mmol/L EDTA (Sigma-Aldrich, catalog no. 03690) and 0.01% NaN₃ (Fisher Scientific, catalog no. S2271) with the following anti-mouse antibodies: allophycocyanin (APC)-conjugated anti-Gr1 (RB6–8C5) or -Ly6C (HK1.4) from eBioscience; phycoerythrin (PE)-conjugated antibodies to CD11b (M1/70) and FITC-conjugated anti-Ly6G (1A8) from BD Biosciences. The respective isotype-matched conjugated controls were purchased from eBioscience (catalog no. 17–4732–81, 17–4321–81) and BD Biosciences (catalog no. 555743, 555573), respectively. Corresponding isotype controls yielded no significant staining. Acquisition was performed on a BD FACS Calibur and the dedicated software BD CellQuest Pro. Data were analyzed with FlowJo software.

For analysis of patient PBMCs, the following cocktail of anti-human antibodies was used: Brilliant Violet 605–conjugated anti-CD45 (2D1), peridinin-chlorophyll-protein cyanine (Cy) 5.5 (PerCP Cy5.5)–conjugated anti-CD33 (WM53), Brilliant Violet 421 (BV421)–conjugated anti-CD11b (ICRF44), APC-conjugated anti-CD14 (63D3), FITC-conjugated anti-CD15 (HI98), and PE-conjugated anti–HLA-DR (L243) from Biolegend. Once stained following the protocol above, cells were fixed in BD Cytofix fixation buffer (BD Biosciences). Samples were acquired using the BD LSR-Fortessa cytometer or BD X-20 cytometer with the dedicated software BD FACSDiva. Data were analyzed with FlowJo software (BD Biosciences).

Freshly isolated splenocytes (10⁶ cells/mL) from naïve C57BL/6 mice were depleted of red cells, labeled with CFSE (1 μmol/L; Invitrogen) for 10 minutes at 37°C, and washed with fresh culture media according to the manufacturer's instructions. CD11b⁺Ly6G^high, CD11b⁺Ly6C^intLy6G^int, and CD11b⁺Ly6C^high cells from the bone marrow of B16 s.c. injected mice receiving IgG or anti-Gr1 in the prophylactic setting were sorted with BD FACSAria III Cell Sorter using the following anti-mouse antibodies: APC-conjugated anti-CD11b (M1/70), PE-conjugated anti-Ly6C (HK1.4), and PerCP Cy5.5–conjugated anti-Ly6G (1A8) from Biolegend. Sorted subsets (>95% pure populations) were cocultured for 72 hours in RPMI 1640 (10% FBS) in U-bottom 96-well plates with 2×10⁵ carboxyfluorescein diacetate succinimidyl ester (CFSE)–labeled naïve splenocytes at different ratios (1:10 and 1:5). During coculture, CFSE-labeled splenocytes were stimulated with anti-CD3 (Thermo Fisher Scientific, 17A2, 10 μg/mL) and anti-CD28 (Thermo Fisher Scientific, 37.51, 5 μg/mL). In a parallel experiment, whole bone marrow from IgG- or anti-Gr1–treated B16 tumor–bearing mice were cocultured with CFSE-labeled splenocytes at different ratios (2:1 and 1:1). To further analyze the CD3⁺, CD4⁺, and CD8⁺ T-cell proliferation, cells were stained with BV421-conjugated anti-CD3e (145–2C11, BD Horizon), Brilliant Violet 785 (BV785)–conjugated anti-CD4 (RM4–5, BioLegend), and Brilliant Ultraviolet 395 (BUV395)–conjugated anti-CD8a (53–6.7, BD Horizon) and analyzed for CFSE dilution using a BD X-20 cytometer and FlowJo software. Data were normalized to the percentage of CD3⁺, CD4⁺, and CD8⁺ T-cell proliferation from stimulated CFSE⁺ splenocytes, calculated as 100%. Experiments were performed in triplicate.

Fifty thousand whole bone marrow cells were isolated from long bones of 4- to 8-week-old C57BL/6 mice and cultured in osteoclastogenic media containing α minimum essential medium (α-MEM) and 10% heat-inactivated FBS, glutathione-S-transferase-RANKL (GST-RANKL, 100 ng/mL), and 1% CMG 14-12 culture supernatant containing the equivalent of CSF-1 (10 ng/mL; ref. 24). GST-RANKL was a generous gift from Steven L. Teitelbaum (Washington University). When multinucleated OCs appeared, all wells were fixed in 4% (v/v) paraformaldehyde in PBS (Polysciences), stained for tartrate-resistant acidic phosphatase (TRAP) using leukocyte acid phosphatase kit (Sigma-Aldrich, catalog no. 387A), and counted. To determine the effect of the anti-Gr1 on OC differentiation, the antibody or the isotype control was supplemented to the culture media for 10 days at 2.5 μg/mL. In some experiments, whole bone marrow cells were also cultured in the presence of 5% LLC tumor–conditioned media in addition to osteoclastogenic cytokines.

To assess the osteoclastogenic potential of CD11b⁺Ly6G^high, CD11b⁺Ly6C^intLy6G^int, and CD11b⁺Ly6C^high populations, bone marrow from C57BL/6 naïve mice was stained with the following anti-mouse antibodies: BV605-conjugated anti-CD45 (30-F11) from Biolegend; Alexa Fluor 700–conjugated anti-CD11b (M1/70) and PE-Cy7–conjugated anti-Ly6C (HK1.4) from Invitrogen; and FITC-conjugated anti-Ly6G (1A8) from BD Biosciences. Each subset was sorted with BD FACSAria III Cell Sorter (>95% purity). Cells were seeded into 96-well tissue culture treated plates at 1, 2, or 5 × 10⁴ cells per well in osteoclastogenic media and cultured at 37°C in 5% CO₂, changing media every 2 days, and cells were fixed on day 6 and 10.

Freshly removed LLC tumor–bearing tibias were fixed in 10% neutral-buffered formalin (DiRuscio & Associates, Inc., catalog no. 415–25) for 24 hours and decalcified in 14% EDTA for 10 days. Tissue was paraffin-embedded and sectioned 5-μm thick at the histology core of the Washington University Musculoskeletal Research Center and stained for TRAP [sodium acetate (Sigma-Aldrich, catalog no. S8750), L(+) tartaric acid (Sigma-Aldrich, catalog no. T1807), Napthol AS-MX phosphate (Sigma-Aldrich, catalog no. N4875), and Fast Red TR salt (Sigma-Aldrich, catalog no. F8764)] or hematoxylin (Leica Microsystem, catalog no. 3801571) and eosin (H&E; Leica Microsystem, catalog no. 3801619). ImageJ software (NIH) was used to quantitate the number of OCs, defined as TRAP-positive per bone marrow area. An Olympus BX41 Phase Contrast & Darfield Microscope (Olympus Optical Co.) was used [2× objective lens (Olympus PLN2X), 10× objective lens (Olympus UPLFLN 10×2), 20× objective lens (Olympus UPLFLN20×), QImaging camera (Model Number 01-RET-OEM-F-CLR-12; QImaging), BIOQUANT OSTEO 2021 software v21.5.60 (BIOQUANT Image Analysis Corporation)].

In vitro experiments were done in triplicate and analyzed using Student t test. T-cell suppression assays with sorted cells comparing various groups were analyzed using two-way ANOVA. In calculating two-tailed significance for equality of means, equal variances were assumed for the two populations. In vivo experiments were done with at least 4 to 8 mice per group (the number of mice used for each experiment is specified in the figure legends), and multiple groups or timepoints were analyzed by two-way ANOVA, followed by the Bonferroni multiple-comparison test. Histologic analysis of TRAP-stained bone sections was done with 3 to 4 tumor-bearing bones per group and analyzed using Student t test. Results were considered significant at P < 0.05. All statistical analyses were performed with GraphPad Prism 9.2.0 software for Windows (GraphPad Software).

Based on the importance of CD11b⁺Gr1⁺ populations in supporting tumor growth at various soft tissue sites, we aimed to determine whether depletion of CD11b⁺Gr1⁺ cells could prevent or reduce the growth of skeletal metastases. We adopted various tumor models known to metastasize to bone, namely the LLC cell line, the triple-negative breast cancer cell line (4T1), and the B16-F10 melanoma cell line. FI-conjugated tumor cells were injected directly into the mouse tibia (i.t.) to study the effects of CD11b⁺Gr1⁺ cell depletion on the growth of tumor cells in the bone microenvironment. Because CD11b⁺Gr1⁺ cells are increased in circulation in tumor-bearing mice (25), we also examined the antitumor effects of CD11b⁺Gr1⁺ cell depletion in mice inoculated with tumor cells in the left ventricle (l.v.), as a model of tumor dissemination to bone. For these experiments, anti-Gr1 was administered in a therapeutic fashion in mice with established tumors in bone (day 7 after tumor inoculation; Fig. 1A). Differently from soft tissue tumors (Supplementary Fig. S1A and B; refs. 26–28), anti-Gr1 administration did not reduce the growth of established LLC or triple-negative 4T1 breast cancer cells inoculated into the tibias of C57BL/6 or BALB/c mice, respectively (Fig. 1A). To determine whether CD11b⁺Gr1⁺ cells played a crucial role in the early tumor development in bone, anti-Gr1 was delivered on the same day of tumor cell inoculation (day 0). Similarly, to established tumors, delivery of anti-Gr1 in the prophylactic setting was not sufficient to prevent the growth of 4T1 tumor cells and B16 melanoma cells injected either i.t. or l.v. (Fig. 1B). These results demonstrate that anti-Gr1 treatment fails to reduce skeletal metastases in multiple tumor models, and mouse backgrounds, in the therapeutic and prophylactic settings.

To understand the difference in antitumor efficacy of Gr1⁺ cell depletion between soft tissue tumors and skeletal metastases, we analyzed the percentage of CD11b⁺Gr1⁺ cells in our models. As expected, the frequency of CD11b⁺Gr1⁺ cells in primary B16 and LLC s.c. tumors and spleens of mice receiving anti-Gr1 in the therapeutic setting was reduced compared with IgG controls (Supplementary Fig. S1C and S1D). CD11b⁺Gr1⁺ cells were not depleted, but rather increased in the bone marrow of mice bearing skeletal LLC and 4T1 tumors treated with anti-Gr1, independent of the time of administration (Supplementary Fig. S2A and S2B). Higher numbers of CD11b⁺Gr1⁺ cells were also found in bone marrow of mice injected with B16 cells i.t. and l.v. and receiving anti-Gr1 at time of tumor inoculation (Supplementary Fig. S2B). Similar results were observed in bone marrow of mice bearing s.c. tumors (Supplementary Fig. S1C and S1D). These findings suggest that the failure of anti-Gr1 to exert antitumor effects in skeletal metastases might be due to poor depletion efficiency of bone-marrow resident CD11b⁺Gr1⁺ cells.

CD11b⁺Gr1⁺ cells comprise monocytic and granulocytic populations that can be identified by the differential expression of Ly6C (monocytic) or Ly6G (granulocytic) markers. We thus determined whether anti-Gr1 induced the expansion of a specific subset of CD11b⁺Gr1⁺ cells in the bone marrow. Consistent with previous reports (27), anti-Gr1 significantly depleted the Ly6G^+/high subset both in the bone marrow and primary site of mice bearing s.c. LLC tumors, whereas the monocytic Ly6C^+/high and Ly6C^intLy6G^int populations were only depleted in soft tissue tumors (Fig. 2A and B). We also noted that the Ly6C^intLy6G^int population was 2- to 3-fold higher in the bone marrow of mice treated with anti-Gr1 compared with IgG controls (Fig. 2A). The bone marrow of mice bearing skeletal LLC and B16 tumors showed efficient depletion of Ly6G^+/high cells and accumulation of Ly6C^intLy6G^int cells following anti-Gr1 administration (Fig. 2C,–E). Similar findings were observed in spleens (Fig. 2F,–H). These results suggest that the failure of anti-Gr1 to protect from bone metastases might be due to the accumulation of a CD11b⁺Ly6C^intLy6G^int population.

To determine whether bone marrow CD11b⁺Ly6C^intLy6G^int cells had similar immune-suppressive functions to the Ly6G^+/high and Ly6C^+/high MDSC subsets, we performed T-cell proliferation assays. CD11b⁺Ly6C^intLy6G^int, Ly6G^+/high, and Ly6C^+/high subsets were sorted from the bone marrow of B16 s.c. tumor–bearing mice treated with IgG or anti-Gr1 and cocultured with naïve CSFE-labeled splenocytes (ratios 1:10 and 1:5) and stimulated with anti-CD3 and anti-CD28. CD3⁺, CD4⁺, and CD8⁺ T-cell proliferation was assessed 72 hours later via flow cytometry. Although Ly6G^+/high cells underwent significant cell death, thus impacting our ability to confirm their immune-suppressive effects, we observed that Ly6C^+/high cells from both IgG- and anti-Gr1–treated mice reduced T-cell proliferation compared with splenocytes alone (Fig. 3A and B, top; Supplementary Fig. S3A). The Ly6C^int/Ly6G^int subset, however, did not show any immune-suppressive effects, independent of IgG or anti-Gr1 (Fig. 3A and B, bottom). To evaluate whether the elevated numbers of Ly6C^intLy6G^int cells in anti-Gr1–treated mice could still impact T-cell proliferation, whole bone marrow cells from mice bearing B16 s.c. tumors treated with anti-Gr1 or IgG were cultured with splenocytes at 2:1 and 1:1 ratios. Whole bone marrow cells from anti-Gr1–treated mice failed to reduce T-cell proliferation compared with IgG controls (Fig. 3C and D; Supplementary Fig. S3B). These results indicate that bone marrow CD11b⁺Ly6C^intLy6G^int cells, which accumulate following anti-Gr1 administration, lack immune-suppressive capacities.

Because CD11b⁺Gr1⁺ cells have been shown to differentiate into bone-resorbing OCs (21), cells known to support development and progression of skeletal metastasis (7), we sought to determine whether CD11b⁺Ly6C^intLy6G^int cells possessed OC differentiation potential. We sorted CD11b⁺Ly6G^high, CD11b⁺Ly6C^intLy6G^int, and CD11b⁺Ly6C^high bone marrow populations from naïve mice and cultured them in vitro using RANKL and CSF-1 cytokines. Although CD11b⁺Ly6G^high did not adhere and died in osteoclastogenic medium, CD11b⁺Ly6C^intLy6G^int and CD11b⁺Ly6C^high cells generated OCs, although the latter formed OCs a day earlier and in higher numbers (Fig. 4A).

Next, we cultured whole bone marrow cells from IgG- or anti-Gr1–treated non–tumor-bearing mice in osteoclastogenic medium and observed higher number of OCs when cells were isolated from mice receiving anti-Gr1 (Fig. 4B, top). Similarly, bone marrow cells from naïve animals cultured in osteoclastogenic media supplemented with tumor-conditioned media, formed more OCs in the presence of anti-Gr1 (Fig. 4B, bottom). To further confirm that anti-Gr1 treatment promoted OC differentiation in the tumor-bearing mice, bone marrow cells were isolated from anti-Gr1– or isotype control–treated mice bearing LLC skeletal tumors. To avoid the expansion of tumor cells in vitro, LLC tumor cells were injected into the right tibia and bone marrow cells were isolated from the left tibia, as no differences were observed in the percentage of CD11b⁺Gr1⁺ cells between the two legs. Similar to naïve mice, higher numbers of OCs formed from bone marrow cells isolated from tumor-bearing mice receiving anti-Gr1 (Fig. 4C). Confirming these in vitro observations, histologic analysis of the LLC tumor–bearing leg from mice receiving anti-Gr1 showed increased number of TRAP⁺ OCs infiltrating tumors compared with IgG controls (Fig. 4D and E). These results suggest that anti-Gr1 induces the expansion of CD11b⁺Ly6C^intLy6G^int cells, which represent an OC precursor subset.

Because OCs support tumor growth in the skeletal microenvironment, we hypothesized that anti-Gr1 failed to decrease tumor growth in bone due to increased number of OC precursors. To test this hypothesis, we evaluated if using ZA to block OC activity in combination with anti-Gr1 (to deplete Ly6G^high granulocytes) would reduce the growth of established skeletal tumors. For this purpose, we established a model of bone metastasis insensitive to the antitumor effects of OC blockade, consisting of delayed ZA administration starting 7 days after intratibial or intracardiac LLC tumor cell inoculation. Seven days after the tumors were established, mice were randomized into 4 groups: anti-Gr1, isotype control, ZA, and anti-Gr1 + ZA. Consistent with our previous results, the administration of anti-Gr1 alone did not affect tumor growth compared with control (Fig. 5A). Similarly, no difference in tumor growth was observed in mice receiving ZA alone compared with anti-Gr1 or isotype controls (Fig. 5A). The combined therapy (anti-Gr1 Ab + ZA) significantly reduced LLC tumor growth (Fig. 5A). Anti-Gr1 increased the numbers of CD11b⁺Ly6C^intLy6G^int cells in the bone morrow and spleen, both in the presence or absence of ZA (Fig. 5B), indicating that ZA does not alter immature myeloid populations, but rather OC activity. We further analyzed the antitumor effects of the combined therapy in the 4T1 bone metastasis model. We observed significant reduction of established breast cancer metastasis in bone with anti-Gr1 + ZA but not by each agent alone (Fig. 5C). This data suggests that treatment of established tumors in bone can be achieved by blocking OC differentiation/activity and concomitant depletion of granulocytic populations in the bone marrow.

ZA treatment is an effective strategy to limit SREs in patients with breast cancer with skeletal metastasis (29). However, a subset of patients develops new skeletal lesions or have expansion of existing metastases, despite ZA treatment. Therefore, we asked whether increased number of granulocytic populations possess protumorigenic functions and are responsible for the lack of ZA antitumor effects in patients with breast cancer with progressive skeletal metastases. We collected PBMCs from 9 patients with stage IV HER2-negative, ER-positive ductal carcinoma with established skeletal metastases being treated with ZA or denosumab (anti-RANKL) at time of study inclusion and sample acquisition (Table 1). Patients were divided into 2 groups based on routine skeletal CT or bone scintigraphy scans: 5 with progressive bone metastatic disease and 4 with stable metastatic bone disease. In both groups, metastases at non–skeletal sites were either stable or absent at the time of sample collection. CD11b⁺CD15⁺ granulocyte populations (Fig. 6A, gating strategy) remained unchanged or were decreased in the stable group (Fig. 6B and C), whereas the populations were increased from baseline in the progressive group (Fig. 6D and E). In particular, patient #536 showed progression of a lesion involving the right proximal femur, indicated by higher bone scintigraphy signal intensity (Fig. 6E, top). This lesion eventually progressed to a pathologic fracture requiring surgical hip replacement. Similarly, patient #575 showed a new lesion in a thoracic vertebral body. This lesion progressed to a pathologic fracture spinal cord compression requiring surgical decompression and stabilization (Fig. 6E, bottom). Taken together, these data suggest that progression of skeletal metastases might be regulated by multiple myeloid-cell populations, rather than solely by mature OCs, and that increased granulocytic cells could be responsible for metastatic bone disease in face of OC blockade.

Currently, there are no effective curative treatments for metastatic bone disease. In the clinic, denosumab (anti-RANKL) aimed at blocking OC differentiation, and ZA, to block bone resorption, efficiently reduce SREs but are not sufficient to improve survival in patients with advanced breast cancer, and ZA antitumor effects are only seen in a small subset of postmenopausal patients in the early-stage setting (12). Here, we provide evidence that established tumors in bone, refractory to ZA treatment, can be effectively treated by a combination of OC blockade and depletion of Gr1⁺ cell subsets. Although inhibition of OC activity is aimed at blocking the tumor/bone vicious cycle, depleting granulocytic populations can potentially improve antitumor immune responses. Our finding suggests that attacking tumor cells residing in the bone where they are protected from chemotherapeutic agents can be achieved by antiresorptive and immunomodulatory approaches.

A role for the immune system in regulation of bone metastases has been previously established. Depletion or inhibition of T cells and NK cells enhances tumor growth in bone in mice (30, 31). Defective antitumor immune responses can also increase skeletal metastases in mouse models with reduced numbers of OCs, which were expected to be protected from tumor growth in bone by interfering with the classical view of the tumor/bone vicious cycle (30). Such findings emphasize the importance of the immune system in modulating tumors in bone, thereby adding additional players to the tumor/bone vicious cycle model. Indeed, although OC inhibition can decrease tumor growth in bone in immune-deficient mice, the residual tumor burden in mice depleted of CD4⁺ or CD8⁺ T cells is higher compared with immune-competent animals receiving the same antiresorptive therapy (30). Conversely, prophylactic administration of anti-CTLA4 to activate antitumor T-cell responses has been effective at reducing skeletal metastases in a melanoma mouse model (30).

In the clinic, immune checkpoint blockade (ICB) is commonly used to treat various solid tumors; however, ICB alone has limited efficacy in breast cancer (32, 33), especially when the tumor metastasizes to bone (34, 35). A preclinical study has shown that ICB fails to prevent skeletal metastases in a prostate tumor model, while suppressing tumor growth at extraskeletal sites. ICB increases the numbers of CD4⁺ and CD8⁺ T cells both in soft tissues and in the bone, although subset-specific analysis shows absence of Th1 cells in the bone microenvironment due to a local increase in bone-derived TGFβ (36). Combinatory treatment of ICB and anti-TGFβ is necessary to suppress the growth of skeletal metastases. Similarly, we found that sole depletion of CD11b⁺Gr1⁺ immune-suppressive myeloid populations, while limiting extraskeletal tumor growth, is not sufficient to reduce skeletal metastases both in the prophylactic and therapeutic settings and concomitant OC blockade is required. ZA has been shown to target macrophage populations in addition to the OCs (37–39). Although we did not observe any changes in F4/80⁺ cells in bone marrow of mice with skeletal metastases treated with ZA and/or anti-Gr1 versus IgG, we also were not able to separate the tumor-infiltrating macrophages from the rest of the bone marrow populations. Nevertheless, the fact that ZA as single agent could not ameliorate established tumors in bone suggests that in addition to targeting OCs and possibly tumor-associated macrophages, concomitant depletion of Gr1⁺ populations is required for effective suppression of tumor growth in bone.

Similar to CD11b⁺Gr1⁺ cells in mice, human CD15⁺ cells have immunosuppressive/tumor-promoting effects in patients with cancer (40). We found that the number of CD15⁺ granulocytic cells increased in patients with breast cancer with progressive skeletal metastases, despite receiving antiresorptive therapies. This result suggests that an increased number of CD15⁺ cells in circulation may be used as a predictive biomarker for resistance to OC inhibitory therapy in patients with metastatic breast cancer, and that a combinatorial strategy using antiresorptive therapies with immunomodulatory agents targeting CD15⁺ cells might prove effective in efforts to treat skeletal metastases. At the moment, there is no anti-Gr1 equivalent for humans, and such treatment could cause severe neutropenia in patients. However, certain chemotherapeutic agents (gemcitabine, cisplatin, 5-fluorouracil, 5-azacytidin) and tyrosine kinase inhibitors (sunitinib, ibrutinib) have been shown to reduce the numbers of circulating and tumor-infiltrating MDSCs in mouse models and patients (41). Several studies report the antitumor effects of preventing MDSC recruitment to tumor sites by blocking chemokine receptors or the inhibition of MDSC immunosuppressive functions by using COX2 inhibitors and others (42). Although these approaches have multiple cellular targets, including the tumor cells, our findings support exploring their use, in combination with OC blockade in patients with established bone metastatic disease.

Depletion of Gr1⁺ cells achieved by anti-Ly6G or anti-Gr1 has been used to reduce primary tumor growth and extraskeletal metastases in numerous mouse models (27, 43, 44). However, these approaches are transient, leading to a rebound effect, in part, due to compensation in bone marrow and spleen in response to the peripheral loss of Gr1⁺ populations (45). We found that anti-Gr1 efficiently depleted Ly6G^high cells but induced accumulation of a CD11b⁺Ly6G^int population. This Ly6G^int population was increased in the bone marrow and spleen of anti-Gr1–treated mice but not at the tumor site. A similar finding has been reported following anti-Ly6G administration, where the depletion of Ly6G^high subsets is followed by a second wave of granulopoiesis and increased numbers of Ly6G^int cells (45). Because Ly6G is upregulated during granulocyte maturation, this Ly6G^int population could represent immature granulocytes that express lower Ly6G and have impaired functionality. Mice infected with various pathogens show that delivery of anti-Ly6G/Gr1 delays pathogen clearance (46, 47). We also found that the immature Ly6G^int cells, which were increased in anti-Gr1–treated mice, lacked immune-suppressive functions compared with MDSCs, but they were still responsible for limiting the antitumor effects of anti-Gr1 treatment by supporting to the tumor/bone vicious cycle.

In line with previous reports demonstrating the osteoclastogenic potentials of CD11b⁺Gr1⁺ cells (21, 48), we found that accumulation of CD11b⁺Ly6G^int cells following anti-Gr1 led to higher number of OCs in vitro and in vivo compared with IgG-treated mice. Ly6G^high cells do not have osteoclastogenic potentials, and rather represent granulocytes that are depleted with anti-Gr1. In contrast, the Ly6C^high subset has higher osteoclastogenic potential than Ly6G^int cells, but it is also present in much lower numbers. Our finding suggests that treatment of bone metastases with anti-Gr1, while reducing protumorigenic CD11b⁺Ly6G^high and Ly6C^high cells, leads to accumulation of Ly6G^int OC precursors that may induce activation of the tumor/bone vicious cycle.

In conclusion, herein we demonstrate that combination of anti-Gr1 and OC inhibition is necessary to treat established metastases in bone, possibly by unleashing antitumor immune responses and blocking the tumor/bone vicious cycle.

A.-H. Capietto reports grants from NIH, Shriners Hospital, and Siteman Cancer Center during the conduct of the study, as well as personal fees from Genentech outside the submitted work. S. Lee reports grants from NIH, Shriners Children's Hospital, and Siteman Cancer Center and personal fees from Kwanjeong Educational Foundation during the conduct of the study. E. Eul reports grants from NIH during the conduct of the study. C.X. Ma reports grants and personal fees from Puma and Pfizer, as well as personal fees from Eli Lilly and Company, Novartis, Seattle Genetics, AstraZeneca, Bayer, Sanofi, Athenex, and Eisai outside the submitted work. R. Faccio reports grants from NIH, Shriners Children's Hospital, and Siteman Cancer Center during the conduct of the study. No disclosures were reported by the other authors.

A.-H. Capietto: Conceptualization, data curation, formal analysis, validation, investigation, visualization, methodology, writing–original draft, writing–review and editing. S. Lee: Data curation, formal analysis, validation, investigation, visualization, methodology, writing–original draft, writing–review and editing. D. Clever: Resources, investigation, visualization, writing–original draft. E. Eul: Investigation, visualization. H. Ellis: Resources, validation. C.X. Ma: Resources, validation, writing–review and editing. R. Faccio: Conceptualization, resources, supervision, funding acquisition, validation, visualization, methodology, writing–original draft, project administration, writing–review and editing.

The authors thank the Musculoskeletal Histology Core of Musculoskeletal Research Center supported by the NIH P30 Grants AR057235 and P30 AR074992, the Washington University Bright Institute and Molecular Imaging Center (NIH P50 CA94056ADD), and the Siteman Flow Cytometry Core at Washington University in St. Louis. This research was supported by grants from NIH grants R01 AR066551 (to R. Faccio) and CA235096, as well as grants from Shriners Hospital (P19–07408 CR, to R. Faccio) and the Siteman Investment Program, Siteman Cancer Center (Pre-R01 Program, to R. Faccio), Saint Louis Men's Group Against Cancer (to C.X. Ma), and Kwanjeong Lee Chong Hwan Educational Foundation of Korea (KEF-2018, to S. Lee).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.