Human OS cell lines SAOS-2 (HTB-85), 143B (CRL-8303), MG63 (CRL-1427), human fetal osteoblastic cells (hFOB; CRL-11372) and mouse macrophage cell line RAW264.7 (TIB-71) cells were obtained from American Type Culture Collection (ATCC) (35). Human metastatic LM7 and MG63.2 were derived from SAOS-2 and MG63 respectively (35). CCH-OS-O and CCH-OS-D cells were derived from the primary tumor of patients (35). The OS17 cell line (36) was used to generate OS17-GFP cells and provided by Dr. Amer Najjar (MD Anderson). LM7-GFP was generated in our laboratory.

The mouse metastatic OS cell line K7M3 was derived from K7 (37) mouse OS cells (38). Dunn and DLM8 cell lines were obtained from Dr. Asai Tatsuya (39). Luciferase expressing K7M3 cells (K7M3-luc) were kindly provided by Dr. Timothy M. Fan (University of Illinois) originally generated by S.Y. Kim at NCI (40, 41). All cells except hFOB, were cultured and maintained in Dulbecco’s modified Eagle’s medium supplemented with 10% heat-inactivated fetal bovine serum, 1% Penicillin/Streptomycin, 1% L-Glutamine, 1X nonessential amino acids, 1X minimal essential medium vitamin solution. The hFOB cells were maintained in DMEM/F12 (1:1) containing 10% FBS and 300 µg/ml neomycin (G418). All cells were regularly tested for mycoplasma contamination using MycoAlert mycoplasma detection kit (Lonza).

Animal experiments performed in the study were in accordance with approved IACUC protocols. BALB/c female mice (6–8 weeks old) were used in the studies and were purchased from Charles River at NCI Frederick. Each group contained at least 5 mice. For the experimental OS-lung metastasis model, BALB/c mice were injected i.v with 0.25x10⁶ K7M3-luc cells. Animals were randomized after tumor cells injection. Blinding was not done for the in vivo experiments. Mice were injected with D-luciferin (GoldBio, 150 mg/kg) for bioluminescence In vivo Imaging system (IVIS) imaging to assess tumor burden.

The orthotopic OS model was also employed to assess the therapeutic effect on established lung metastases that arose from the primary tumor (38). Briefly, the right leg of 7–8 weeks old BALB/c mice was dehaired 24 hrs before tumor cell injection. K7M3-luc cells (0.2x10⁶ cells in 10 µl of PBS) were injected into the right tibial medullary cavity using a 27-gauze needle and 100 µl Hamilton syringe to establish an orthotopic primary OS tumor. The right leg bearing primary tumor was amputated at day 20 post intra-tibial tumor cell injection, a time when lung micrometastases are present (38).

WP1066 (29, 42) was administered via oral gavage (50 mg/kg) in vehicle of DMSO/PEG300 (1:4) every day for 5 days a week for three weeks (29, 32, 43). Control cohorts received vehicle only. For anti-CD47 treatment, mice were treated with 400 µg of anti-CD47 antibody (Clone MIAP410; Bioxcell) in 200 µl PBS intra-peritoneally every 48 hrs three doses a week for total of eight doses as shown in experimental schema (44). In combination treatment, WP1066 treatment was initiated earlier than anti-CD47 antibody treatment to achieve stable plasma concentration that requires multiple administrations (29).

To generate tumor spheroids, OS17-GFP and LM7-GFP cells (1x10⁴) were seeded in clear, round-bottom 96 well ultra-low attachment plates (Corning) in complete DMEM media. The plates were then centrifuged for 5 mins at 1000 g and incubated at 37°C. After 24 hrs, the spheroids were treated with the indicated concentration of WP1066 or DMSO only as control. Plates were incubated in the IncuCyte S3 live-imaging system (Essen Biosciences) and scanned every 6 hrs for 5 days.

MDSCs were generated from bone marrow isolated from naïve female BALB/c mice as described before (45). Single cells isolated from femur and tibia were plated (5x10⁶ cells/plate) in non-tissue culture treated 10 cm plates in complete RPMI media (basal RPMI with 10% FBS and penicillin and streptomycin) supplemented with recombinant mouse GM-CSF (40 ng/ml; Biolegend) and IL6 (40 ng/ml; Biolegend). After 4 days in culture, Bone marrow derived MDSCs (BM-MDSCs) were harvested and washed with complete RPMI for assays. To determine the effect of WP1066 on BM-MDSCs, BM-MDSCs were replated in 10 cm non-TC plates (5x10⁶ cell) and treated with the indicated concentration of WP1066. After 24 hrs cells were harvested and analyzed for surface markers including CD11b, Ly6G, Ly6C, and Ki67 (proliferation marker) expression by flow cytometer.

For T cell suppression assays (46), splenocytes harvested from naïve BALB/c mice were incubated in tissue culture treated plates for 45 mins to isolate T cells. T cells were labelled with 2.5 µM Cell Trace Carboxyfluorescein succinimidyl ester (CFSE, Thermofisher) and cocultured with BM-MDSCs and CD3/28 dynabeads (Thermofisher) at 1:1:1 ratio in 200 µl complete RPMI media for 72 hrs. Cells were harvested, stained for CD4 and CD8, and analyzed for proliferation in flow cytometer (BD Fortessa).

In vitro macrophages were derived by culturing bone marrow cells in complete RPMI supplemented with recombinant mouse M-CSF (20 ng/ml; Biolegend) for 6 days with media replaced at day 4. At day 6 adherent cells were harvested with TrpLE (Thermofisher) and replated in 6 well plate (1x10⁶ cells per well) in 2 ml complete RPMI media without M-CSF. After overnight culture, the cells were either treated with WP1066 (4 µM) or DMSO for 24 hrs. Cells were then harvested and stained for macrophage markers including CD11b, F4/80, MHC-II, and CD80 and analyzed by flow cytometer.

Lungs draining cervical lymph nodes (LDLN) and lungs were harvested from mice at day 16 post tumor cell injection for analysis. Single cells suspensions from the LDLN were prepared by mashing the tissue through 70 µm cell strainer. Single cells from lungs were prepared by enzymatic digestion of finely chopped lung tissues in complete RPMI supplemented with collagenase-IV (Worthington) and DNAse (Sigma) at 37°C for 30mins. Digested tissues were then mashed through 70 µm cell strainer to obtain single cell suspensions. Red blood cells were lysed using RBC lysis buffer (Biolegend). Single cells from lymph nodes and lungs were washed with PBS and stained with Fixable Zombie violet dye for 10 mins at 4 C in the dark. Cells were then stained with Fc block (CD16/32 antibody) and stained for surface markers with a combination of antibodies ( Supplementary Table 1 ) diluted in FACS buffer (PBS supplemented with 1% BSA and 0.05% NaN3) with Brilliant Violet stain buffer for 30 mins at 4 C in the dark. After incubation, cells were washed in FACS buffer and fixed using fixation buffer (Thermofisher). For intracellular/intranuclear staining, fixed cells were washed with permeabilization buffer followed by staining with antibody in the buffer for 30 mins at 4 C. Cells were acquired using BD Fortessa and analyzed using FlowJo v. 10 software.

Single cells from lungs and LDLNs (2x10⁶) were activated with phorbol 12-myristate 13-acetate (PMA, 50 ng/ml) and Ionomycin (500 ng/ml) for 4 hrs in presence of Brefeldin (Biolegend) and Monensin (Biolegend). Cells were harvested after 4 hrs and stained as described before. Cells were stained for surface markers followed by intracellular for IFN-γ and TNF-α.

All statistical analysis was performed in Graph Pad Prism v. 9. Data are represented as mean ± standard error of the mean (SEM). For comparison between two groups, an unpaired two-tailed student’s t-test or Mann-Whitney test was used as indicated in figure legends. One way ANOVA followed by Tukey’s multiple comparison test was performed for comparing more than two groups. Survival analysis was performed using the log-rank (Mantel-Cox) test. p-value of <0.05 was considered statistically significant.

Data generated in the study are included in the manuscript and the supplementary information file. The raw data are available upon request from the corresponding author.

First, we evaluated the status of activated STAT3 in OS cell lines. We performed western blot analysis on cellular lysates of multiple metastatic and non-metastatic mouse and human OS cell lines. In line with previous reported observations, phosphoSTAT3 (pSTAT3^Y705) was detected at different intensities in all tested cell lines ( Supplementary Figures 1A, B ). We next investigated if WP1066 treatment inhibits STAT3 activation in OS cell lines. Mouse (K7M3 and DLM8) and human OS cells (OS17 and LM7) were treated with different concentrations of WP1066. WP1066 treatment for 24 hrs reduced the level of pSTAT3^Y705 in a dose dependent manner as compared to DMSO-treated cells ( Supplementary Figures 1C, D ). This suggests that pSTAT3^Y705 is constitutively expressed in OS cells and that WP1066 inhibits activated STAT3.

Given the critical role of STAT3 in cell proliferation in multiple tumor models, we investigated the effect of inhibiting STAT3 activation by WP1066 in OS cells. To determine if WP1066 had a cytotoxic effect, mouse (K7, K7M3, Dunn, DLM8) and human (OS17, LM7, MG63, MG63.2, 143B, HOS) OS cells were incubated with different concentrations of WP1066 for 72 hrs and analyzed for cell viability by resazurin assay ( Figure 1A ; Supplementary Figure S1E ). WP1066 inhibited proliferation of both mouse and human OS cells in a concentration dependent manner ( Figures 1A, B ). The IC₅₀ against mouse OS cells ranged from 2.79 µM to 6.01 µM with K7M3 being more sensitive than K7 ( Figure 1B ; Supplementary Figure S1E ). The IC50 for human OS cells ranged from 2.3 µM to 3.5 µM with OS17 being more sensitive than LM7 ( Figure 1B ; Supplementary Figure S1E ). Moreover, we assessed if WP1066 induced tumor cell apoptosis using a fluorescence-based assay to assess the kinetics of activation of caspase-3/7. WP1066 treatment activated caspase-3/7 across all tested cell lines. The higher dose of WP1066 (5 µM) induced swift activation of caspase-3/7, whereas a gradual increase over time was observed when cells were treated with the lower dose of WP1066 (2.5 µM) compared to DMSO ( Figure 1C ). As another measure for apoptosis, we also assessed the level of PS by Annexin V. Similar kinetics for the expression of PS (Annexin V⁺) was observed following treatment with WP1066, supporting the conclusion that WP1066 induced apoptosis ( Figure 1D ).

We also investigated WP1066-induced toxicity using tumor spheroids. Tumor spheroids were generated from OS17-GFP and LM7-GFP cells and were treated with different concentrations of WP1066. Without WP1066 treatment, OS17 and LM7 derived tumor spheroids remained stable and maintained strong GFP-mean fluorescence intensity (MFI) throughout the duration of the experiment, indicating viability. By contrast, the cytotoxicity of WP1066 against tumor spheroids was demonstrated by a significant drop in GFP-MFI in a concentration dependent manner ( Figure 1E ).

WP1066-induced apoptosis may upregulate the expression of “eat me signals” which have been shown to enhance phagocytosis (47). To determine if WP1066 enhanced tumor cell phagocytosis, K7M3 cells were treated with WP1066 (2.5 µM) for 24 hrs. CFSE-labelled WP1066- or DMSO-treated K7M3 cells were co-cultured with RAW264.7 mouse macrophages for 4 hrs. Flow cytometric analysis revealed that there was a significant increase in the phagocytosis of WP1066-treated K7M3 cells compared to those treated with DMSO ( Figure 1F ). This data supports that WP1066 treatment enhanced the ability of macrophages to phagocytose tumor cells. Taken together, these data suggest that WP1066 suppresses cell proliferation, induces apoptosis, and enhances phagocytosis.

We next sought to investigate the effect of WP1066 on MDSCs and macrophages generated from bone marrow cells. First, we assessed the effect of WP1066 on viability and proliferation of MDSCs. In vitro generated BM-MDSCs were treated for 24 hrs with WP1066 and then analyzed by flow cytometry for the markers CD11b, Ly6C, and Ly6G to differentiate between monocytic MDSCs (M-MDSCs, CD11b⁺Ly6C^hiLy6G^-) and polymorphonuclear MDSCs (PMN-MDSCs, CD11b+Ly6C+Ly6G+) (45). WP1066 treatment significantly reduced the frequency of live M-MDSCs as compared with the DMSO treated control ( Figure 2A ). However, there was no statistical difference in percentage of live PMN-MDSCs between the treated and control groups. When analyzed for the expression of Ki67, a proliferation marker, WP1066 significantly reduced the mean fluorescence intensity (MFI) of Ki67 in both M-MDSCs and PMN-MDSCs ( Figure 2B ). These data indicate that M-MDSCs are more susceptible to the cytotoxic effect of WP1066 as compared to PMN-MDSCs. However, WP1066 suppressed the proliferation of both M-MDSCs and PMN-MDSCs.

Next, we evaluated whether the immunosuppressive capacity of MDSCs was affected by WP1066. BM-MDSCs were treated with WP1066 for 24 hrs and then evaluated for T cell suppression capacity by co-culture assay. DMSO treated BM-MDSCs suppressed the proliferation of CFSE-labelled T cells. CFSE^low daughter generation was not observed as compared with T cells activated with dynabeads alone. In contrast, CD4 and CD8 T cell proliferation was increased when co-cultured with BM-MDSCs treated with WP1066 in a dose dependent manner ( Figure 2C ). These observations are in line with other studies where inhibition of STAT3 has been shown to induce apoptosis and attenuate the suppressive capacity of MDSCs (48, 49). Taken together, these data suggest that WP1066 significantly affects the viability, proliferation, and suppressive capacity of BM-MDSCs.

Given the significant role of STAT proteins in plasticity and differentiation of macrophages (50), we next determined the in vitro effect of WP1066 on BMDM. WP1066 treatment for 24 hrs significantly increased the frequency of macrophages expressing MHC-II (CD11b⁺F4/80⁺MHC-II⁺) and co-expressing MHC-II and CD80, markers associated with M1-like or proinflammatory phenotype ( Figure 2D ). In addition, WP1066 increased the MFI of CD80 and MHC-II in macrophages as compared with DMSO control ( Figure 2D ). The increase in MHC-II and co-stimulatory molecule CD80 on macrophages following WP1066 treatment suggests that STAT3 inhibition by WP1066 activates macrophages and enhances antigen presentation. Taken together, the described effects on MDSCs and macrophages indicates a potential positive immunomodulatory effect by WP1066 on myeloid cells through suppression of MDSCs and activation of anti-tumor macrophage function.

The syngeneic K7M3 experimental metastasis model was used to determine the therapeutic effect of WP1066 on established lung metastases. K7M3-luc cells were injected intravenously (i.v) and lung metastases were validated by IVIS imaging before treatment on day 3 post tumor cell injection. Mice were treated with WP1066 (o.g) or vehicle control 24 hrs after imaging ( Figure 3A ). WP1066 treatment significantly reduced the tumor growth ( Figures 3B, C ) and improved the survival. The median survival time (MST) for WP1066-treated cohort was 48 days as compared to 42 days for the control group (p=0.037). Interestingly, 1 out of 17 mice treated with WP1066 survived more than 70 days, whereas all control mice died by day 50. This suggests the potential therapeutic activity of WP1066 against OS-lung metastases.

WP1066 treatment resulted in only a modest 6-day increase in MST, highlighting limited activity. We reasoned that combining WP1066 with immunotherapy could enhance therapeutic efficacy. The CD47-SIRPα axis, also referred to as an innate immune checkpoint, regulates the activation of innate immune cells (24). Neutralizing CD47-SIRPα axis with anti-CD47 antibody enhances tumor phagocytosis by macrophages, and activation of Natural Killer cells (NK), Dendritic cells (DCs), neutrophils and macrophages (24–26). In line with previous reports, flow cytometric and immunoblot analysis revealed that both human and mouse OS cells constitutively express CD47 ( Supplementary Figure S2 ), suggesting a potential mechanism of immune evasion (51). These findings prompted us to investigate whether combining WP1066 with anti-CD47 antibody would improve the anti-tumor responses. Tumor bearing mice were treated with WP1066 alone, anti-CD47 alone or WP1066 + anti-CD47 ( Figure 4A ). No significant change in body weight was observed in any of the groups ( Figure 4B ). Combination therapy significantly inhibited tumor growth (p=0.028) and prolonged survival (p<0.0001) compared to the vehicle treated control cohort ( Figures 4C–E ). The MST for mice that received combination therapy was 63 days compared to 47 days for the control group. Monotherapy with anti-CD47 also resulted in a significant decrease in tumor growth (p=0.048, Figure 4D ) and an increase in MST (p<0.001, Figure 4E ) compared to control. While there was a significant increase in MST in mice treated with combination therapy compared to those treated with WP1066 alone (p=0.027), the improvement in MST between mice treated with combination therapy versus anti-CD47 alone did not reach statistical significance (MST: 63 days vs 50 days, p =0.083). However, using the hazard ratio (log-rank test) to compare the anti-CD47-treated group with the combinatorial therapy group indicated that the mice treated with the monotherapy had more than twice the risk of death compared to those receiving combination therapy (Hazard ratio: 2.237, 95% CI: 0.7257 to 6.894). This represents a ~55% reduction in risk with the combination therapy, which is biologically meaningful. These data suggest that combination therapy can improve the anti-tumor response of WP1066 against established lung metastases.

As the therapeutic efficacy of immunotherapy has been shown to be linked to immune cell infiltration into the tumor and tumor draining lymph nodes, we performed flow cytometric analysis of lungs and LDLNs at day 16 post tumor cell injection ( Figure 5A ; Supplementary Figure S3 ). Analysis of lungs revealed that the combination treatment increased central memory CD8⁺ T cells (CD8Tcm; CD8⁺CD44⁺CD62L⁺) ( Figure 5A ). In addition, combination treatment significantly reduced PD1 expressing CD8 T cells (CD8⁺PD1⁺), a marker of dysfunction when compared with vehicle treated control cohort ( Figure 5A ). While combination treatment also increased FoxP3⁺ expressing regulatory T cells (CD4⁺FoxP3⁺; Tregs) the CD4 Tem (CD4⁺CD44⁺CD62L^-): Treg ratio, was increased suggesting increased number of effector memory CD4⁺T cells to Tregs. Furthermore, analysis of cytokine secreting cytotoxic CD8⁺ T cells revealed that combination treatment increased IFN-γ secreting CD8⁺ T cells, a marker of activated T-cells ( Figure 5A ). Analysis of cytotoxic NK cells in lungs suggested that combination therapy increased the level of differentiated mature NK cells (CD49b⁺CD11b^hiCD27^-CD62L⁺). Furthermore, IFN- γ expression in NK cells was significantly increased in mice that received both the treatments ( Figure 5B ). Interestingly, we observed a similar trend in animals treated only with anti-CD47 antibody ( Figures 5A, B ).

Analysis of antigen presenting cells (APCs) infiltrating the lungs revealed that combination treatment significantly increased the frequency of M1-like or proinflammatory activated macrophages (CD11b⁺F4/80⁺MHC-II⁺CD80⁺) versus control cohort ( Figure 5C ). Combination therapy also increased activated myeloid derived monocytic DCs co-expressing MHC-II and CD80 (CD11b⁺CD11c⁺MHC-II⁺CD80⁺). Furthermore, APCs infiltrating the lungs of animals in the combination cohort had significantly reduced expression of PDL1 (associated with immune suppression) and increased expression of MHC-II ( Figure 5C ).

These observations were consistent with the immune cell profile in the LDLNs. LDLN from animals treated with both agents had an increased CD8: Treg ratio with significantly increased CD8 Tcm and reduced CD4⁺ Tregs and reduced dysfunctional CD8⁺PD1⁺ ( Supplementary Figure S3 ). Mice treated with combination therapy also had a significant increase in cytotoxic CD8 T cells co-secreting IFN-γ and TNF-α ( Supplementary Figure S3A ).

Taken together, as summarized in Figure 5D , these data suggest that the therapeutic efficacy of combination therapy was associated with an augmented pro-inflammatory anti-tumor immune response mediated by increased infiltration of activated immune cells into the lungs and LDLNs, including IFN-γ secreting cytotoxic CD8 T cells, NK cells and activated macrophages and monocytic DCs. These pro-inflammatory effects were paralleled with decreased immune suppression as observed by a decrease in Tregs and dysfunctional PD1 expressing CD8 T cells in combination treatment cohort.

Next, we evaluated the therapeutic effect of combination therapy using our orthotopic intra-tibial OS model in which lung metastasis form spontaneously from a primary tumor in the leg (38). This model better reflects how lung metastases develop in patients. For this purpose, K7M3-luc cells were injected into the right tibia of BALB/c mice. Mice were treated with the first cycle of therapy using WP1066, CD47 or combination therapy as shown in Figure 6A . Amputation of the leg-bearing primary tumor was performed 3 weeks after tumor cell injection to prevent further metastatic spread. This was followed by the treatment strategy shown in the schema ( Figure 6A ). Although the lung tumor burden, as represented by tumor flux, at day 45 post tumor cell injection was not statistically significant among the groups ( Figure 6B ), there was a significant difference in the survival time. In control group, 60% of mice died by day 45 and 100% had to be euthanized by day 70. By contrast the death rates were 10% and 30% in mice treated with WP1066 and anti-CD47 respectively. There were no deaths in the combination therapy group at 45 days. Survival at day 70 was 0%, 20%, 20% and 70% in control, WP1066, anti-CD47, and combination therapy treated mice respectively ( Figure 6C ). The MST of the mice treated with combination therapy was 78 days compared to 51 and 50 days for mice treated with WP1066 and anti-CD47 respectively. These data suggest that combined treatment with WP1066 and anti-CD47 is also more effective in prolonging survival in an orthotopic OS model.