Tumor cells: Human breast carcinoma (MCF7, and MDA-MB-231) and renal carcinoma (Caki-1, and ACHN) cell lines were purchased from the American Type Cell Culture Collection (ATCC). CRBM-1990 cells were previously isolated and characterized from a BM of renal cell carcinoma (Avnet et al., 2004). Breast and renal carcinoma cell lines were maintained in complete Iscove’s modified Dulbecco’s medium (IMDM). Complete medium was supplemented with 10% heat-inactivated fetal bovine serum (FBS, Euroclone), plus 100 U/ml of penicillin, and 0.1 mg/ml of streptomycin (Life Technologies). Cells were cultured at 37°C, in a humidified atmosphere of 5% CO₂. Unless otherwise indicated, cells were maintained at pH 7.4.

Human osteoblasts: hOB from healthy donors were purchased from VWR International PBI. Human MSCs from bone marrow (BM-MSCs) were purchased from Lonza (Euroclone). hOBs were maintained in complete Dulbecco’s modified Eagle’s medium (DMEM), and BM-MSC in complete minimum essential medium Eagle’s alpha modified (α-MEM). Complete media were supplemented with 10% heat-inactivated fetal bovine serum (FBS, Euroclone), plus 100 U/ml of penicillin, and 0.1 mg/ml of streptomycin (Life Technologies). Cells were cultured at 37°C, in a humidified atmosphere of 5% CO₂. Unless otherwise indicated, cells were maintained at pH 7.4.

Human osteoclasts: Human OC (hOC) were derived from buffy-coats of at least two healthy donors (AVIS, Bologna), as previously described (Avnet et al., 2007). Peripheral blood mononuclear cells (PBMC) were isolated on Ficoll-Hystopaque gradient (GE Healthcare) and washed with PBS. To obtain differentiated hOC, isolated cells were resuspended in complete high-glucose DMEM (Euroclone) supplemented with 10% heat-inactivated characterized FBS (Celbio), plus 100 U/ml of penicillin, and 0.1 mg/ml of streptomycin (Life Technologies) (OC complete medium), and plated on different supports, depending on the type of assay, at a cell density of 3 × 10⁶ cells/cm². PBMCs were incubated at 37°C in a humidified atmosphere of 5% CO₂. After 2 h, non-adherent cells were removed, and the medium was replaced with different media depending on the type of assay.

Human CD14⁺ monocytes: To obtain an enriched population of OC precursors, CD14⁺ cells were derived from PBMCs isolated from buffy-coats of two healthy donors by immunomagnetic separation using an anti-CD14 monoclonal antibody (MiniMACS; Myltenyi Biotec). Briefly, PBMCs were isolated from buffy-coats, as described above, washed with MACS buffer (PBS at pH 7.2, supplemented with 0.5% bovine serum albumin and 2 mmol/L ethylene diamine tetraacetic acid), and clumps were removed by passing cells through a 30-μm prefilter. Cells were then centrifuged at 400 × g for 15 min. The cell pellet was suspended in MACS buffer (10⁷ cells in 80 μl), mixed with 20( μl of anti-CD14 MACS antibody-coated microbeads (Miltenyi Biotec), and incubated for 15 min at R°T. The cell suspension was applied to an LS-positive selection column that was previously washed with 1 ml of MACS buffer, and placed in a magnetic separation unit. The column was rinsed with 3.5 ml of MACS buffer, then removed from the magnetic separation unit, and positive bound cells were flushed with 2.5 ml of buffer.

Culture medium at specific pH was obtained by adjusting the concentration of sodium bicarbonate according to the Henderson–Hasselbalch equation, as previously described (Avnet et al., 2013). To model acidic conditions, the pH of the medium was adjusted to 6.8. At different time-points and at the end-point of each experiment, the pH of the culture supernatants was measured to confirm the maintenance of the prefixed pH values during the incubation period, by using a micro-electrode (pH 301, HANNA Instruments). For the measurement of the pH of the culture supernatant, the medium was collected by using plastic tips and tubes that were pre-incubated at the same temperature and CO₂ concentration of the incubator for at least an hour. Then, collected medium was centrifuged and, the pH value of the supernatant was measured at room temperature and CO₂ conditions, as quick as possible, to avoid that the measurement could be affected by the changes in atmospheric temperature and CO₂. pH value was obtained as an average of three sequential measurements for each of the three different samples.

Conditioned medium of carcinoma cells: Conditioned medium (CM) of the carcinoma cell lines was used for the OC differentiation and for the Type I collagen degradation assays, as described in the next paragraphs. It was obtained from cultures seeded in T75 cell culture flasks and cultured at standard conditions. In particular, when 70% confluence was reached, cells were washed with PBS, and incubated with complete high-glucose DMEM. After 48 h, the supernatant was collected, centrifuged, and stored at −80°C until use.

Conditioned medium of hOB: CM of hOB was used for the functional assays with cells of the OC lineage (OC differentiation, Type I collagen degradation, and CD14⁺ cell migration assays), as described in the next paragraphs. It was obtained from acid-stressed or not-stressed hOB (CM hOB^{pH 6⋅8} and CM hOB^pH7⋅4, respectively) by seeding cells in T25 cell culture flasks (3 × 10⁵ cells/flask) in standard conditions. After adhesion, hOBs were washed with PBS and cultured for 20 h in α-MEM plus 0.1% FBS (low-serum medium). Cells were then washed with PBS, and incubated again with low-serum medium at pH 6.8 or 7.4 for 10 h. The medium was then replaced with low-serum medium at pH 7.4, and cells were incubated for an additional 48 h for all the conditions. At the end of the incubation, the supernatants were collected, centrifuged, and stored at −80°C until use.

To quantify the levels of IL-6, IL-8, or RANKL secreted from hOB by ELISA assay, as described in the next paragraph, we used two different hOB primary cell cultures: (a) for IL-6 and IL-8, we used the supernatant of the already differentiated cells: commercial hOB cells were seeded at standard conditions (3 × 10⁴ cells/well) in 24-well plates. After adhesion, in order to ensure low-serum adaptation, cells were washed with PBS and cultured in α-MEM plus 0.1% FBS (low-serum medium) for 20 h. Cells were then washed with PBS, and incubated with low-serum medium at pH 6.8 or at pH 7.4. After 24 h, the supernatants were collected, centrifuged, and stored at -80°C until use; (b) for RANKL, since commercial hOBs do not secrete detectable levels of RANKL, we used hOBs that were previously induced to differentiate from mesenchymal precursors: BM-MSCs were seeded (3 × 10⁴ cells/well) in 24-well plates at standard condition, and then cultured in osteogenic medium (complete α-MEM supplemented with 50 μg/ml of L-ascorbic acid 2-phosphate, 10^–8 M dexamethasone, and 10 mM β-glycerophosphate, Sigma). After 10 days of culture, to ensure low-serum adaptation, cells were washed with PBS and cultured for 20 h in low-serum osteogenic medium. Cells were then washed with PBS, and incubated again with low-serum osteogenic medium at two different pH (6.8 or 7.4). After an additional 24 h, the supernatants were collected, centrifuged, and stored at −80°C until use.

Human OC precursors (PBMC) were isolated from buffy-coats of at least two healthy volunteers as previously described, and seeded in eight-well chamber slides. After 2 h of incubation at 37°C in a humidified 5% CO₂ atmosphere, the medium was replaced, and the resulting mononuclear adherent precursors were cultured for 7 days with collected CM at a defined ratio depending on the specific experiment (see Table 1A), and compared with the respective negative control.

For all the cell cultures, medium was changed every 3 or 4 days. When specifically mentioned, anti-IL-6 monoclonal antibody (TCZ, 100 μg/ml, Roche) or anti-IL-8 antibody (anti-IL-8 Ab, 5 μg/ml, R&D Systems) were added every 24 h. After 7 days of culture, cells were analyzed for TRACP expression (Acid Phosphatase, Leukocyte kit, Sigma-Aldrich), according to the manufacturer’s protocols, dark incubated with 2.25 μg/ml of Hoechst 33258 (Sigma-Aldrich) at RT for 10 min, and visualized with a Nikon Eclipse E800M fluorescence microscope (Nikon). TRACP-positive cells that showed more than three nuclei were considered as OC and counted. For assays, at least four replicates were performed.

Testing the direct effect of acidic pH: hOC precursors (PBMC) isolated from buffy-coats of at least two healthy volunteers were seeded in 96-well plates coated with europium-conjugated type I collagen (Osteolyse Assay Kit, Human Collagen, Lonza). After 2 h from seeding, non-adherent cells were removed, and cells were cultured for 7 days in fresh complete high-glucose DMEM, added with 50 ng/ml of RANKL and 10 ng/ml of M-CSF (Peprotech) (differentiation medium) in order to obtain osteoclastogenesis. Then, medium was replaced with acidic or neutral differentiation medium, and cells were incubated for an additional 4 days. At the end-point, the cell supernatants were transferred into wells containing the fluorophore-releasing reagent. Fluorescence emission was immediately measured with a microplate reader (Infinite F200pro, Tecan) using an excitation wavelength of 340 nm and an emission wavelength of 615 nm. Data were expressed as relative fluorescence units (RFU), and hOC-mediated degradation of human bone collagen was considered to be directly proportional to the RFU recorded. In order to exclude the type I collagen degradation that was not the result of cell activity, we subtracted the signal obtained with cell-free acidic differentiation medium. All experiments were performed in quadruplicates.

Testing the effect of the cell secretome: Differentiated hOC were obtained after 7 days of incubation with differentiation medium, as described above. Then, the medium was replaced, and cells were incubated for an additional 4 days with the collected CM, at a defined ratio depending on the specific experiment (see Table 1B), and compared with the respective control.

For the experiments with TCZ or anti-IL-8 Ab treatment, the CM of hOB were added or not with 100 μg/ml TCZ or with 5 μg/ml anti-IL-8 Ab every 24 h. At the end-point, the cell supernatants were transferred into wells containing the fluorophore releasing reagent. Fluorescence emission was immediately measured as above described. Data were expressed as relative fluorescence units (RFU) after blank subtraction, and hOC-mediated degradation of human bone collagen was considered directly proportional to the RFU recorded. For all assays, at least eight replicates were performed.

RNA isolation from carcinoma cells: Total RNA was extracted from cell cultures using TRIzol reagent (Invitrogen, Thermo Fisher Scientific) and reverse transcribed with MuLV reverse transcriptase (Applied Biosystems, Thermo Fisher Scientific). For carcinoma cell lines, RNA was isolated from semi-confluent cells. Cells were first plated in complete IMDM and incubated for 24 h in a standard incubator. Then, for an additional 24 h for normoxic condition (21% O₂), cells were maintained in the same incubator, whereas for hypoxic conditions (1% O₂), cells were transferred to a hypoxic Invivo2-400 incubator (Ruskin Technologies).

RNA isolation from human osteoblasts: For RNA isolation from hOB under acidic conditions, cells were first plated in complete D-MEM. After adhesion, to ensure low-serum adaptation, cells were washed with PBS, and pre-treated for 20 h with alpha-MEM plus 0.1% FBS (low-serum medium). Then, at time 0 (T0), after washing with PBS, cells were incubated with low-serum Alpha-MEM at pH 6.8. RNA was collected after 3 and 24 h of incubation.

Quantitative reverse transcription polymerization chain reaction (qRT-PCR) protocols and primers: qRT-PCR was performed by amplifying 1 μg of cDNA using the light cycler instrument and the universal probe library system (Roche Applied Science). Probes and primers were selected using a web-based assay design software (ProbeFinder, ¹). Primer sequences and probes are listed in Table 2.

Results were normalized to hydroxymethylbilane synthase (HMBS) or beta-actin (β-actin), according to the 2-ΔΔCT method (Avnet et al., 2017). For the experiments performed at pH 6.8, we used HMBS since we have previously demonstrated its stability under acidic conditions in human osteoblast-like cells (Lemma et al., 2018). Assay was repeated with four replicates.

Seahorse technology: Tumor cells were seeded on an XF microplate in complete IMDM, starved of glucose for 2 h, and then supplied with 2 g/L of D-glucose (Sigma-Aldrich). The complete ECAR analysis consisted of four stages: basal (without drugs), glycolysis induction (addition of 10 mM glucose), maximal glycolysis induction (addition of 2 μM oligomycin in order to impair oxidative phosphorylation), and glycolysis inhibition (addition of 100 mM 2-DG). The measurements of the proton production rates (PPRs) were obtained by using a Seahorse Extracellular Flux (XF-96) analyzer (Seahorse Bioscience). We considered PPR values obtained before the addition of D-glucose to be representative of non-glycolytic extracellular acidification (−glycEA), and those obtained after the addition of D-glucose to be representative of extracellular acidification derived from glycolysis (+glycEA). Data were normalized to the total protein content using the BCA protein assay (Pierce). This type of assay has a great variability since metabolic activity can change very quickly and is very sensitive to small variations in the microenvironmental conditions. Thus, for this test, the experiment was performed three times, and each condition had 15 intra-assay replicates (n = 45).

PHMed assay: As already previously described (Perut et al., 2014), 16 × 10⁶ cells were washed twice in pHMed solution [80% normal saline, 10% unbuffered RPMI-1640 (Sigma-Aldrich) and 10% FBS], and incubated in pHMed in suspension for 3 h at 37°C. Cells were then centrifuged (10 min at 500 × g), and supernatant was collected for extracellular pH (pHe) measurement. pHe was immediately quantified by a digital pH-meter (pH 301, HANNA Instruments). The experiment was repeated with six replicates for both breast and renal carcinoma cell lines.

CD14⁺ cells were isolated from PBMC after immunomagnetic separation, as previously described. The isolated cells were then suspended in 200 μl of alpha-MEM containing 0.1% BSA and seeded into Transwells with 8-μm pores (9 × 10⁴ cells/Transwell), in the upper compartment. In the lower compartment, 800 μl of alpha-MEM added with 5% of FBS at pH 7.4, or 800 μl of alpha-MEM added with 5% of FBS at pH 6.8, or 800 μl of CM hOB^{pH 7⋅4}, or 800 μl of CM hOB^{pH 6⋅8} were placed. Cells were then incubated in 5% CO₂ at 37°C, and allowed to migrate for 36 h. At the endpoint, migrated cells on the lower side were fixed in methanol, stained with crystal violet solution, and counted from nine random fields (20 × lens) in each well. The assay was repeated with six replicates.

Analysis of cell culture supernatants: hOB supernatants were analyzed for IL-6, IL-8, and RANKL content using Human IL-6 DuoSet ELISA kit, Human CXCL8/IL-8 Quantikine ELISA kit (RD System), and Human sRANK Ligand Standard ABTS ELISA Development kit (Peprotech), respectively. Normalization was performed with respect to the total protein content as measured by BCA protein assay. IL-6, IL-8, TRACP5b, and CTX were quantified in serum samples using the Human IL-6 DuoSet, Human CXCL8/IL-8 Quantikine (RD System), BoneTRAP^® (TRACP5b), and Serum Crosslaps^® (CTX-I) (Immunodiagnostic Systems) ELISA kits, respectively. The assay was repeated with four replicates.

Analysis of serum from patients: We collected blood samples from 29 patients at the time of BM diagnosis (20 from breast carcinomas, six from renal carcinomas, three from thyroid carcinomas) and that were enrolled since February 2014 to June 2017 by the Rizzoli Orthopedic Institute and the ANT foundation in Bologna, and by the Istituto Scientifico Romagnolo Per Lo Studio E La Cura Dei Tumori (IRST) in Meldola. The samples were collected at clinical presentation, before surgical treatment of the bone lesion, with signed informed consent and with institutional ethical committee approval (No. 0037602 of November 14, 2013). Blood samples were centrifuged (2,500 rpm, 8 min), and sera were aliquoted and stored at −80°C until the ELISA assays were performed.

Animal experiment was conducted as a service by the Pharmatest Company (Turku, Finland), with the approval of the National Committee for Animal Experiments (license number ESAVI-2077-04 10 07-2014). Two 5- to 6-week-old female athymic nude mice (Hsd: Athymic nude, Foxn1nu; Envigo, Netherlands) were exposed to analgesic (Temgesic; buprenorphine, 0.1 mg/kg body weight) at least 30 min before the inoculation, and then injected with MDA-MB-231 (1 × 10⁵ cells in 20 μl of PBS) into the bone marrow of right proximal tibia. At 4 weeks after intratibial inoculation, the mice were sacrificed with CO₂, and the death was confirmed by cervical dislocation. The tumor-bearing tibias were collected to 10% neutral-buffered formalin (NFB) for histological evaluation. Paraffin-embedded tumor-bearing tibias were prepared for histology. The tumor-bearing tibias were decalcified in ethylenediaminetetraacetic acid (EDTA) for 2 weeks. After de-calcification, the samples were dehydrated in an increasing series of ethanol concentrations, defatted in xylene, and embedded in paraffin (TEK III Paraffin Wax, Sakura, Netherlands). Then midsagittal 4-μm sections were obtained from each animal. For the histology, the sections were deparaffinized in xylene and rehydrated in a series of descending ethanol concentrations.

Sections were stained for hematoxylin and eosin (H&E-Orange G; Leica Biosystem and Sigma-Aldrich) for quantification of tumor area, and TRACP (Pararosaline; Sigma-Aldrich) for the identification of OC.

Five-micrometer-thick tissue sections were mounted on a glass slide covered with 2% silane solution in acetone. After dewaxing in Citro Histoclear (Histo-Line Laboratories, Milano, Italy) and rehydration in ethanol, tissue sections were incubated first in a 3% hydrogen peroxide solution and then in a 2% bovine serum albumin solution in order to block the endogenous peroxidases and non-specific binding, respectively. Incubation with the rabbit anti-ATP6V1B2 or rabbit anti-LAMP2 (Sigma, Saint Louis, Missouri, United States) primary antibody followed. Tissue sections were then incubated with a biotinylated secondary antibody, covered with DAB, and counterstained with Mayer’s hematoxylin (EnVision FLEX, High pH Link visualization system, Agilent Technologies, Santa Clara, United States).

Because of the small number of observations, data were not considered to be normally distributed, and therefore, non-parametric tests were used. Statistical analysis was performed using GraphPad Prism 7 software (San Diego, CA, United States). For the difference between two groups, the Mann–Whitney U-test was used. For the correlations between serum levels of the different proteins analyzed, the Spearman’s rank test was used. Data were expressed as the mean ± standard error (SEM), and only p-values < 0.05 were considered for statistical significance.

We considered different carcinomas that metastasize to bone. Most of them, such as the breast carcinoma cell lines MDA-MB-231 and MCF7, can directly promote differentiation of hOC (Figure 1A), whereas only MCF7 can induce a slight increase in OC activity (type I collagen degradation, Figure 1B) at in vitro standard culture condition and under physiological pH 7.4. Conversely, for renal carcinoma, we already reported that ACHN and CRBM-1990 cells induce hOC formation only indirectly, via the stimulation of stromal cells, such as endothelial cells (Cenni et al., 2006). This appeared not to be the case for Caki-1 cells that, in this study, directly induced OC differentiation and activity (Figures 1A,B).

However, carcinoma cells may also modulate OC differentiation and activity through the acidification of the extracellular space that may derive both from hypoxia-dependent or -independent mechanisms. Thus, to confirm the ability of mammary and renal carcinoma cells to lower the extracellular pH, both under normoxia and hypoxia, we analyzed the mRNA expression of key regulators of intracellular and extracellular pH at different oxygen tension (Figure 2A).

Specifically, we analyzed the mRNA levels of the proton extruders carbonic anhydrase 9 (CA9) and V-ATPase. We found that renal carcinoma cell lines expressed CA9, V₁B₂, V₀C, and V₁G₁ subunits of V-ATPase that were confirmed also in breast carcinoma cell lines. Furthermore, CA9 expression was significantly increased under reduced oxygen conditions with respect to normoxia (p = 0.0209 for MDA-MB-231 and ACHN, p = 0.0495 for MCF7, p = 0.0339 for Caki-1, Figure 2A). In contrast, among the different V-ATPase subunits, only for the V₀C, we observed a trend of hypoxia-induced increased expression that was significant for the MDA-MB-231 cell line (p = 0.0433, Figure 2A). By using the Seahorse analyzer, we then directly quantified the acidification ability of both mammary and renal carcinoma cells (namely, the measurement of proton production rate, PPR) resulting from the glycolytic metabolism (+glycEA) or, independently, to glycolysis (−glycEA) (Figures 2B,C). Notably, all the examined carcinoma cell lines strongly acidified the extracellular space, and we still detected a meaningful secretion of protons in the -glycEA condition, except for MCF7 breast carcinoma cells (Figure 2C). We further confirmed this finding also by measuring the supernatant of the unbuffered cell cultures through the use of a micro-electrode. Although less precise, this approach enlightened that, in vitro, carcinoma cells acidified the extracellular fluid to an average pH value of 6.72 ± 0.06 (Figure 2D). Then, we investigated the effects of acidosis on cells of the BM microenvironment by using a medium buffered at pH 6.8 to mimic the acid supernatant of carcinoma cells in all the following experimental protocols. We selected the pH value that corresponded to the average pH of the supernatant of the carcinoma cell lines included in this study (Figure 2D, pHMed assay) that was also the same pH value of the tumor microenvironment, as assessed in a model of breast carcinoma, assuming that the intratumoral pH of BM deriving from breast cancer may well correspond to that of primary lesions (Longo et al., 2016). To dissect the role of tumor-derived extracellular acidification, we used fresh acid medium instead of the acidified supernatant of carcinoma cells, thereby avoiding other confounding factors like the secretion of paracrine molecules, including tumor-derived growth factors. Interestingly, we found that low pH directly enhances the recruitment of human CD14⁺ monocytes (Figure 3A, pH 6.8 vs. 7.4, p = 0.025). The incubation with acidic medium also significantly enhanced type I collagen degradation in cultured OC (Figure 3B).

To confirm the close relationship between OC and acidifying carcinoma cells, we developed a xenograft model of BM by intra-tibial injection of MDA-MB-231 cells (Salerno et al., 2010; Di Pompo et al., 2017). We used MDA-MB-231 cells since they are derived from a breast carcinoma that, in human patients, causes osteolytic lesions more frequently than other carcinomas (Weilbaecher et al., 2011) and because, adversely to MCF7 cells, they can develop osteolytic bone metastases in vivo (Walenta et al., 2001; Robey et al., 2008). Moreover, we previously demonstrated that MDA-MB-231 cells have a Warburg phenotype characterized by an increased NADH/NAD+ ratio, GLUT1 expression, glycolytic rate, and, as a result, increased acidification in the extracellular microenvironment. On the contrary, the metabolism of the less-aggressive breast carcinoma cells MCF7 is mainly based on oxidative phosphorylation (Lemma et al., 2017). We then considered V-ATPase and LAMP2 expression as indirect markers of acidifying tumor cells and acidified tumor area (Avnet et al., 2013, 2019a; Damaghi et al., 2015). In the tumor sections obtained from this model, we observed the presence of actively resorbing OC that were identified by the staining for TRACP activity and V₁B₂ V-ATPase strong signals, with particular regard to the resorbing cell side that attached to the bone (Figure 4). As expected, the V-ATPase subunit was also expressed by carcinoma cells. In the tumor area in which carcinoma cells are coupled with OC, we also observed LAMP2 staining in cancer cells (Figure 4), indicating that those areas that are characterized by the presence of actively resorbing OC were associated with acidifying carcinoma cells.

In addition to a direct effect, we speculated that extracellular acidosis indirectly recruits hOC precursors by inducing a chemotactic secretome in human OB (hOB). As a confirmation, when we used the conditioned medium of acid-stressed hOB (CM hOB^pH6⋅8), the migratory activity of CD14⁺ monocytes was significantly higher compared with conditioned media from hOB^pH7⋅4 or to acidic medium itself (Figure 3A, p = 0.0062). Notably, acid-stimulated hOB also indirectly promoted OC differentiation as CM hOB^pH6⋅8 significantly increased the number of mature hOC in respect with CM of not acid-stressed human hOB (CM hOB^pH7⋅4) (p = 0.0209, Figure 5A).

To go deeper inside the identification of pro-osteoclastogenic growth factors that are released by acid-stimulated hOB, we found that exposure to acidic medium significantly increased the level of secreted free RANKL protein, with respect to neutral conditions (Figure 5B, p = 0.0339), and, after 24 h, the level of the mRNA expression of the macrophage colony-stimulating factor (M-CSF) and TNF-α (p = 0.0274 and p = 0.0205 vs. time 0, respectively, Figure 5C). Similarly, also the mRNA transcription of the pro-inflammatory cytokines IL-6 and IL-8 was increased (at 3 and 24 h for IL-6 and IL-8, respectively, p = 0.0008 for both IL-6 and IL-8, Figure 5C). Protein analysis confirmed the augmented secretion of the two pro-inflammatory cytokines at low pH conditions (p = 0.0209 for both IL-6 and IL-8, Figure 5D). Notably, the acid-stimulated levels of expression of the inflammatory mediators and pro-osteoclastogenic factors IL-6 and IL-8 were significantly higher in hOB than in the tumor cell line MDA-MB-231 (p = 0.0339, Figure 5E).

Our preliminary in vitro data suggested that an acidic microenvironment, in addition to directly promote the migration of OC precursors and the bone resorption activity of mature OC, activates an inflammatory and pro-osteoclastogenic secretome in OB. In more detail, we speculated that the acidic microenvironment of BM further exacerbates osteolysis, both directly and by regulating the paracrine interaction between OB and OC, in favor of OC differentiation and OC-mediated bone resorption (Figure 6).

By analyzing circulating markers in human patients, we then indirectly verified whether the inflammatory cytokines that are possibly released by locally acid-stimulated osteoblasts in the BM microenvironment are also released at a systemic level. If this hypothesis is correct, the circulating levels of IL-6 and IL-8 may reflect a local increase in the respective cytokines in the bone tissue and may parallel the levels of osteolysis in human patients. As a first step, we validated the tartrate-resistant acid phosphatase 5b (TRACP5b) as a specific and sensitive marker of OC activity (Halleen et al., 2000, 2001; Avnet et al., 2008; Savarino et al., 2010) in a continuous series (n = 29) of patients with osteolytic BM from breast, renal, and thyroid carcinoma (Table 3).

In this series of patients, we confirmed a significant correlation (p = 0.0322) between the serum levels of the two bone resorption-related markers, the TRACP5b and the cross-linked telopeptide of type I collagen (CTX) (Figure 7A).

Subsequently, by ELISA assays, we found a highly significant correlation between TRACP5b and IL-8 (p < 0.0001) and a lack of correlation with IL-6 (Figure 7A). On the contrary, IL-6 correlated with C-reactive protein (CRP) (Figure 7A).

Notably, in our in vitro model, IL-8 was more highly expressed in acid-stressed hOB with respect to IL-6 (2,332.53 ± 176.41 versus 469.78 ± 7.08 ng/mg of total protein, respectively, Figure 5D).

Based on the evidence obtained, IL-6 and IL-8 are possible regulators of the proposed mechanistic model (see Figure 6). We then sought to investigate the effectiveness of using the neutralizing anti-IL-6R antibody TCZ and anti-IL-8 antibody (Figure 6) that is already clinically used to treat different diseases (Kang et al., 2015). When OC precursors were cultured in the presence of CM hOB^pH6⋅8 that was added with TCZ or with an anti-IL-8 antibody, hOC differentiation significantly decreased (p = 0.0008 and p = 0.0066, respectively; Figure 7B). Conversely, no significant inhibitory effects were observed after exposure to CM hOB^pH7⋅4, added with TCZ. These data suggest that the pro-osteoclastogenic effect exerted by the acid-induced secretome of OB specifically relays on IL-6 and IL-8 secretion, while the stimulatory effect of CM hOB^pH7⋅4 is possibly mediated by factors other than IL-6. Based on our data, IL-8 may be one of these mediators since neutralizing IL-8 strongly affected osteoclastogenesis after exposure to both neutral and acid-induced secretome of OB (p = 0.0065 and p = 0.0066, respectively). Regarding OC resorption activity, only TCZ significantly inhibited type I collagen degradation induced by CM hOB at both pH (p = 0.0184 for CM hOB^pH7⋅4 and p = 0.0007 for CM hOB^pH6⋅8). However, the percentage of inhibition obtained with TCZ treatment was comparable between CM hOB^pH6⋅8 and CM hOB^pH7⋅4 (23.62 versus 21.30%, respectively, Figure 7C), suggesting that the acid-induced secretome of hOB is more effective in promoting OC differentiation rather than OC activity.