Peripheral blood mononuclear cells (PBMCs) were obtained from 12 healthy donors’ buffy coats, through a density gradient by the Ficoll-Paque TM PLUS (Cytiva, Marlborough, USA). PBMS were in part frozen for subsequent NK separation and in part utilized for monocyte separation through CD14 microbeads and Midi MACS cells separation system (Miltenyi Biotec, Bergisch Gladbach, Germany). The purity of CD14⁺ cells (≥ 95%) was checked by flow cytometry. Cells were plated in 24- or 6- multi-well plates (Corning, NY, USA), in Alpha-Minimal Essential Medium (α-MEM, supplied by Life Technologies, NY, USA), supplemented with 10% fetal bovine serum (FBS) and 1% Penicillin-Streptomycin (Euroclone, Pero, Milan). To induce OC differentiation, recombinant human M-CSF (25 ng/ml) and RANK-L (10 ng/ml) (PeproTech, East Windsor, NJ, USA) were added every 3 days and the cultures were maintained at 37°C and 5% di CO₂ for 8–10 days.

OC photos were obtained through an optical microscope, and tartrate-resistant acid phosphatase (TRAP) positive multinucleated cells with three or more nuclei were stained by TRAP staining kit (Cosmo Bio Co. LTD. Tokyo, Japan). The optical density of the TRAP enzymatic activity was quantified in the supernatants of OCs on the different cell culture conditions through Multiskan FC Microplate Photometer. The interaction between OCs and NK cells were monitored and registered for 72h by JulyBr Live cell movie analyzer (Nanoentek, Waltham, MA, USA).

Annexin V staining was performed on OCs following the manufacturer’s protocol (Miltenyi Biotec, Bergisch Gladbach, Germany). Briefly, Annexin V- FITC Kit allows to detect apoptotic dead cells as Annexin V positive and propidium iodide (PI) negative. All samples were acquired through MACsQuant 10 and analyzed with MACsQuantify software (Miltenyi Biotech, Bergisch Gladbach, Germany).

OCs were analyzed by flow cytometry, according to a previously published protocol (23). Briefly, OCs were detached by ACCUTASE solution (Sigma-Aldrich, St. Louis, MO, USA), and incubated with Hoechst 33342 Solution (BD Bioscence, Franklin Lakes, New Jersey, USA) for 30 minutes at 37°C. Successively, cells were washed with PBS and incubated with primary anti-human Calcitonin R antibody (CalcR IgG2a) (R&D,Systems, Minneapolis, USA) for 30 minutes at 4°C, then with the appropriate secondary IgG2a A647 and the PE-conjugated antibody mouse anti-human CD51-61 (BD Bioscence, Franklin Lakes, New Jersey, USA) for 30 minutes at 4°C. The expression of NK cell receptors’ ligands on OCs were performed using the following primary monoclonal antibodies (mAb) produced in laboratory (UOC Patologia e Immunologia, IRCCS Ospedale Policlinico San Martino, Genoa, Italy): anti-PVR (L95-IgG1), anti-Nectin2 (L14-IgG2A), while anti-ULBP1 (MAB1380, R&D System Minneapolis, USA), anti-ULBP2 (MAB1298), anti-B7H6 (MAB7144), anti-NID1 (MAB2570) were purchased from R&D System, Minneapolis, USA. The secondary antibodies used were mouse anti-IgG1 PE (1070-09, Southern Biotech, Birmingham, USA) and Alexa Fluor 647(A21240, Thermofisher Scientific, Waltham, Massachusetts, USA). All samples were analyzed using MACsQuant 10 and computed with MACsQuantify software (Miltenyi Biotech, Bergisch Gladbach, Germany).

After OCs were formed on plates (about 8–9 days), they were detached, plated on bone slices (ids, Boldon, United Kingdom) and cultured for 14 days. Then, OC resorption was evaluated through Scanning Electron Microscopy (SEM) using a ZEISS EVO 50 XVP (Oberkochen, Germany), equipped with a LaB6 source. Briefly, bone slices were undergone to alcohol dehydration (NaOH 50%, NaOH 70%, NaOH 90%, NaOH 95%, NaOH 99%), next to minimize the charge effects, the surface of the samples was previously coated with a thin gold layer (∼10 nm), then analyzed by SEM. Images of the total surface of the bone slices were analyzed to quantify the bone resorption area, by ImageJ software 1.53t.

Frozen PBMCs, previously isolated and frozen as described in 2.1, were thawed to isolate NK cells through a negative selection, by NK Cell Isolation Kit (130-092-657, Miltenyi Biotec, Bergisch Gladbach, Germany), according to the manufacturer’s instructions. The overall purity of the utilized NK cells was ≥ 95%, Supplementary Figure S1. Cells were cultured at 200.000 cell/well in RPMI 1640 (Lonza, Basel, Switzerland) + 10% FBS + 1% Penicillin-Streptomycin, 1% Glutamine (Euroclone, Pero, Milano) supplemented with IL-15 (5 ng/ml, IL-15 ImmunoTools, Friesoythe, Germany) in U-bottom tissue culture plates (Corning, NY, USA). Medium and IL-15 were refreshed after 3 days, thus NK cells received 2 stimulations with IL-15 for up to 7 days, then they were utilized for the different co-culture conditions, without adding further IL-15. We checked the viability of NK cells by 7-AAD staining solution by Miltenyi Biotec, Supplementary Figure S2.

The immunophenotype of NK cells were analyzed through staining by the following primary monoclonal antibodies (mAbs), produced in laboratory (UOC Patologia e Immunologia, IRCCS Ospedale Policlinico San Martino, Genoa, Italy): AZ20 (IgG1, anti-NKp30), Z231 (IgG1, anti-NKp44), GN18 (IgG3, anti-DNAM1). We also used commercially available mAbs as CD69 PE, CD3 PercP, CD16 APC-Vio770, TIGIT (Miltenyi Biotec, Bergisch Gladbach, Germany), CD56 PE-Cy7 (anti-human Beckaman Coulter, Fullerton, CA, USA), and LEAF anti-TIM3 (345004, Biolegend, San Diego, California). The secondary antibodies used were mouse anti-IgG1 PE (1070-09, Souther Biotech, Birmingham, USA) and Alexa Fluor-647(A21240, Thermofisher Scientific, Waltham, Massachusetts, USA).

NSCLC cell line, A549 (ATCC, Manassas, Virginia) was plated in adherent condition in F-12K Nut Mix 1x (Gibco, Billings, MT, USA) with 10% FBS, 1% Penicillin-Streptomycin.

To obtain CSCs (A549s), A549 cells were detached by Tripsin (Euroclone, Pero, Milano), counted and plated at 10.000 cells/well in Ultra-Low Attachment 6-multi-well (Thermo Fisher Scientific, Carlsbad, CA, USA), in a serum-free stem cell medium (SCM), constituted by DMEM/F12K (Gibco, Billings, MT, USA), supplemented with B27 (Gibco, Billings, MT, USA), EGF 20 ng/mL (Sigma Aldrich, Burlington, MA, USA), bFGF10 ng/mL (PeproTech, East Windsor, NJ, USA), Heparin 2 g/mL (Stemcell Technologies, Vancouver, Canada), and 1% Penicillin-Streptomycin. After 10 days, spheroid-like structures were originated, then dissociated by ACCUTASE and analyzed by flow cytometry through the staining with the following anti-human monoclonal antibodies CD133 PE-Vio770, Nectin-2 PE, Nectin-4 Vio Bright V423 (Miltenyi Biotec, Bergisch Gladbach, Germany), CXCR4 APC (BD Bioscence, Franklin Lakes, New Jersey, USA), PVR BV510 (Biolegend, San Diego, USA).

Co-cultures with autologous OCs and NK cells (ratio 1:1) and with OCs+A549s+NK cells (ratio 1:1:1) were plated on flat bottom 96-well plates for 3 days. The same co-cultures were performed on bone slices in SCM with M-CSF (25 ng/mL) and RANKL (10 ng/ml) for 14 days, to allow the OC resorption activity. NK cells and A549s were co-cultured (ratio 1:1) for 3 days in SCM.

The NK degranulation activity was evaluated by CD107a staining on NK cells on day 7 (Supplementary Figure S3) and after 3 days of culture in medium without IL-15 or in co-cultures with A549s and OCs. Briefly, NK cells were co-cultured with A549s, OCs and both OCs and A549s (ratio of 1:1:1) in the presence of CD107a antibody (Biolegend, San Diego, USA). After 1hour, GolgiStop (BD Biosciences, Franklin Lakes, New Jersey, USA) was added to the co-cultures, and cells were incubated for 5 additional hours. Next, cells were recovered, washed once with PBS and stained for CD56 and CD16. In addition, anti-TIGIT blocking antibody (50ug/ul) (Biolegend, San Diego, USA) was added in the different co-cultures.

Statistical analyses were performed using GraphPad Prism version 8.0. Two-sided and paired Student’s t-test was utilized to determinate statistically significant differences between two groups. Statistical analysis among more than two groups was performed by one-way Anova with Tukey’s post-hoc test. Data are expressed as means ± standard deviation (SD), unless otherwise indicated. Statistical significance was defined as a p-value less than 0.05.

CSCs have been widely reported to be implicated in the metastatic process. In this study, we explored the interactions between CSCs, OCs, and NK cells as fundamental players in bone metastasis process. The crosstalk among OCs, NK cells and CSCs was investigated by setting up in vitro co-cultures of OCs, NK cells and/or A549 spheres (A549s), enriched for CD133⁺ CSC subpopulation. After 72 hours of co-culture, we reported that multinucleated, TRAP⁺ OCs (Figures 1A, E) decreased when cultured with NK cells, (Figures 1B, E, *p < 0.05) and with both NK cells+A549s (Figures 1D, E, **p < 0.01).

We observed a significant decrease in OC numbers (p < 0.05), when co-cultured with NK cells and A549s (Figures 1D, E) compared to A549s alone (Figures 1C, E) suggesting a modulatory role of NK cells in OC viability or differentiation. Consistently, TRAP enzymatic activity was markedly decreased in OCs co-cultured with NK cells (Figure 1F, *p < 0.05). Notably, TRAP activity in OCs was not significantly impacted by A549s alone, but it differed significantly when cultured with both NK cells and A549s (Figure 1F, **p < 0.01), indicating a suppressive effect of NK cells on OC function, which may be modulated by the presence of CSC-enriched A549 spheres.

Since we observed a reduction in OC number after co-culture with NK cells, we real-time monitored OCs and NK cell co-culture for 72h. We showed that the death of OCs was not immediate, but mainly concentrated at 48-72h, with apoptosis representing the main mechanism of cell death (Suppl. Video). Analysis of Annexin V expression in OCs cultured alone (Figure 2A) or with NK cells (Figure 2B) revealed a significant increase of Annexin V in presence of NK cells p < 0.01(Figure 2C), while PI was negative for both OCs alone (Figure 2D) and OCs+NK (Figure 2E).

After 14 days of co-culture, OC resorption activity on bone slices (Figure 3A), that was a measurement of OC functionality, was significantly reduced with NK cells, p < 0.05 (Figures 3B, E), whereas was not altered in co-culture with A549s compared to OCs alone (Figures 3C-E). Notably, in the presence of NK cells, A549s adhered to the bone slice surface and exhibited proliferative behavior (Figures 3D, F), thereby precluding a quantification of the resorption area under these conditions. Nonetheless, the evaluation of TRAP activity revealed a significant reduction of OC resorption activity both with NK cells alone or with A549s, p < 0.05 (Figure 3G). These findings indicate that NK cells do not exert an inhibitory effect on cancer cell proliferation but conversely, they could facilitate tumor cell growth.

NK cells are heterogeneous: the most characterized subsets are the cytokine producer (CD56^brightCD16⁻) and the cytotoxic (CD56^dimCD16⁺) NK cells. We assessed whether these 2 subsets were modified after co-culture with OCs, A549s and OCs+A549s, for 3 days. We observed a significant increase of CD56^dim NK cell subset following co-cultured with A549s (* p < 0.05) OCs (**p < 0.01) and with both OCs and A549s (*p < 0.05) (Figure 4A) compared to NK cells alone. Conversely, the CD56^bright NK cell subset showed a decreasing trend in the presence of A549s and OCs+A549s (Figure 4B).

We further evaluated the modulation of activating (DNAM-1, NKp44, NKp30) and inhibitory receptors (TIGIT and TIM-3) on NK cells in response to co-culture with OCs and A549s. DNAM-1, a key activating receptor involved in NK cell-mediated cytotoxicity, was significantly downregulated on NK cells co-cultured with OCs or OCs/A549s (**p < 0.01) compared to control NK cells (Figure 4C), suggesting a potential impairment of NK cell activation. Similarly, NKp44 expression was significantly altered in NK cells co-cultured with both OCs and A549s (*p < 0.05) respect to A549s alone (Figure 4D), suggesting a dynamic regulation of this receptor under complex microenvironmental stimuli. Finally, we showed that NKp30 levels did not change (Figure 4E), whereas the activation marker CD69 was upregulated in NK cells co-cultured with A549s (Figure 4F), reflecting an activated phenotype in the presence of tumor cells.

Among the inhibitory receptors, we focused on TIGIT and TIM3, showing a significantly higher expression of TIM-3 by NK co-cultured with A549s (*p < 0.05), and with OCs+A549s (**p < 0.01) (Figure 4G). TIGIT expression was elevated in NK cells cultured with OCs and OCs+A549s compared to NK cells alone (*p < 0.05), but not with A549s alone (*p < 0.05) (Figure 4H). These findings suggest that OCs exert an inhibitory effect on NK cells, as evidenced by downregulation of the activating receptor DNAM-1 and upregulation of the inhibitory receptors TIM-3 and TIGIT, thereby dampening the NK cell-mediated response against A549s. Consequently, NK cells may facilitate the adhesion of A549 cells on the surface of bone slices and their proliferation (Figure 3F).

Since we showed the modulation of NK cells receptors by co-cultures with OCs, we next investigated the expression on OCs of ligands for NK cell receptors on OCs. Specifically, we analyzed PVR and Nectin-2, ligands for both the activating receptor DNAM-1 and the inhibitory receptor TIGIT; NID-1, a ligand for the activating receptor NKp44; and B7H6, a ligand for NKp30, respectively controlling cytotoxic response and NK cell-mediated killing. PVR and Nectin-2 expression showed a slight but not statistically significant increase, whereas NID-1 and B7H6 expression were significantly decreased when OCs were co-cultured with NK cells (Figure 5A, *p < 0.05). The observed downregulation of NID-1 (Figure 5B, *p < 0.05) and B7H6 (Figure 5B, **p < 0.01) suggests a mechanism by which OCs may reduce NK cell activation, contributing to immune evasion. However, OCs co-cultured with A549 cells alone or with A549 cells and NK cells did not display any significant changes in NID-1 or B7H6, whereas the decrease of NID-1 (Figure 5B, *p < 0.05) and B7H6 (Figure 5B, **p < 0.01) was consistently confirmed in presence of NK cells alone, highlighting the modulatory impact of NK-OC interactions on ligand expression and subsequent NK cell functionality.

To assess whether OCs exert an inhibitory effect on the degranulation activity of NK cells, we co-cultured them with OCs for 72 hours, then we tested their capability to respond to A549s. NK cells degranulated against A549s as highlighted by CD107a expression (Figure 6A, *p < 0.05). Notably, NK cells previously cultured with OCs showed a significant reduction CD107a expression, when upon exposure to A549s compared (Figure 6A, *p < 0.05). Next, we evaluated the CD107a expression without pre-conditioning NK cells with OCs (Figure 6B). NK cells increased their degranulation in presence of A549s (****p < 0.0001) and with A549s+OCs (**p < 0.01). The level of CD107a expression was higher in NK+A549s than in NK+OCs (***p < 0.005), and likewise elevated in NK cells co-cultured with both OCs and A549 cells relative to OCs alone, *p < 0.05. Given the observed increase in TIGIT expression on NK cells in co-cultures with OCs and A549s, we investigated CD107a expression in the presence or absence of a TIGIT-blocking antibody (Figure 6C). The treatment with anti-TIGIT antibody did not affect degranulation in presence of A549s alone, but significantly enhanced CD107a expression when NK cells were co-cultured with OCs, *p < 0.05. NK cell activity was significantly reduced in co-culture with OCs compared to A549s (**p < 0.01), without any correlation with anti-TIGIT treatment. Interestingly, NK cell degranulation did not significantly increase in co-culture with OCs+A549s compared to OCs alone. However, the addition of anti-TIGIT antibody induced a significant increase in CD107a expression (**p < 0.01), suggesting that the presence of A549s alone is not sufficient to overcome the inhibitory effect exerted by OCs, and that the TIGIT pathway plays a critical role in the crosstalk between NK cells and OCs.

To evaluate whether NK cell degranulation correlates with a reduction in multinucleated TRAP⁺ OCs, we performed TRAP staining under different culture conditions: OCs alone (Figure 6D), OCs co-cultured with NK cells with or without anti-TIGIT (Figures 6 E, F), and OCs co-cultured with NK cells+A549s with or without anti-TIGIT (Figures 6 G, H). The treatment with anti-TIGIT antibody induced a marked reduction of multinucleated TRAP+ OCs (Figures 6 F, H), in line with the observed increase in NK cell degranulation activity.

We analyzed the CD133⁺/CXCR4⁺ cell population, previously identified as the subset of metastasis-initiating cells (4), within the A549s cultured alone, or in co-culture with NK cells, OCs or both. A significant reduction of the CD133⁺/CXCR4⁺ subset was observed when A549s were co-cultured with NK cells, p < 0.01 (Figure 7A), confirming NK cell-mediated cytotoxicity against A549s. However, in the presence of NK cells+OCs, this cell population was not significantly reduced, suggesting that the immunosuppressive effect exerted by OCs may enable A549s to evade NK cell killing.

We further analyzed on CD133⁺ A549s the expressions of Nectin-2, Nectin-4 and PVR, which are the ligands of the DNAM-1 and TIGIT, the NK cell receptors significantly modulated in the different co-cultures.

We further analyzed on CD133⁺ A549s the expression of Nectin-2, PVR and Nectin-4, which is a ligand expressed by tumor cells. Nectin-2 expression showed a slight increase in A549s co-cultured with NK cells+OCs, but it was not statistically significant (Figure 7B). In contrast, Nectin-4 expression displayed a downregulation in CD133⁺ A549s when cultured with NK cells or OCs. PVR expression was very low in the CD133⁺ subset without differences in the different culture conditions (data not shown).

These findings may partially explain the lack of enhanced NK cell degranulation observed following anti-TIGIT treatment in co-cultures with NK cells+A549s or with OCs+A549s. In particular, the absence of significant modulation in TIGIT ligands on A549s suggests a limited contribution of the TIGIT axis in these settings, reinforcing the idea that OCs, rather than A549s, are the primary drivers of NK cell inhibition via TIGIT engagement.