Establishment and validation of an in vitro co-culture model for oral cell lines using human PBMC-derived osteoclasts, osteoblasts, fibroblasts and keratinocytes

Indirect co-culture models with osteoclasts including oral cell lines may be influenced by M-CSF and RANKL in the common cell medium. Therefore, we investigated the viability and proliferation of osteoblasts (OB), fibroblasts (FB) and oral keratinocytes (OK) under stratified medium modification and assessed the differentiation of osteoclasts in each co-culture. The impact of M-CSF and RANKL in the common OC co-culture was assessed for OB, FB and OK via MTT assay via DAPI control. The multinuclearity and function of OC were evaluated by light microscopy, DAPI staining, resorption assay and FACS analysis. The PBMC showed the highest differentiation into OC after an incubation period of 7 days. Furthermore, co-culture with OB enhanced the number of differentiated multinucleated OC in comparison with monoculture, whereas co-culture with OK decreased PBMC multinuclearity and OC differentiation. FB did not influence the number of differentiated OC in a co-culture. RANKL and M-CSF reduction had no impact on OC differentiation in co-culture with FB or OB, whereas this medium modification for OK attenuated PBMC multinuclearity and OC differentiation in all approaches. Supplementation of RANKL and M-CSF can be modified for a co-culture of PBMC with FB or OB without disturbing OC differentiation. Thus, pathogenic processes of bone remodelling involving OB, OC, FB and OK in the oral cavity can be investigated thoroughly.

isolation and cryopreservation of pBMc. Whole blood collected from volunteers was provided from the Department of Transfusion Medicine, University hospital of Lübeck. The blood was first centrifuged for 20 min at 22 °C and 3500 RPM to obtain the buffy coat. Buffy coats were transferred into anticoagulant EDTA tube and peripheral blood mononuclear cells (PBMC) were isolated by gradient density centrifugation with Histopaque-1077 (Sigma-Aldrich, St Louis, USA). For this procedure 10 ml of "buffy coat" was mixed with 10 ml of warm (37 °C), sterile phosphate-buffered saline (PBS, Life Technologies, Darmstadt, Germany), layered over 10 ml of Histopaque and centrifuged (800g, 30 min, 22 °C, with brake off). The cell layer direct on top of the Histopaque contains PBMC. This layer was transferred into a new 50 ml tube, resuspended in 10 ml of warm PBS, diluted with 10 ml PBS (= 20 ml volume) and centrifuged (300 g, 3 min, 22 °C, with brake on). Subsequently, cells were counted in a haematocytometer, aliquoted, frozen and stored in liquid nitrogen (Fig. 1).
Osteoclast differentiation of PBMC. Thawed PBMCs were plated in cell-culture dishes at 8.5 × 10 7 cells per dish in DMEM, 10% FBS, 1% antibiotics, 5% l-Glutamine and 25 ng/ml human M-CSF. After an expansion time of 72 h, the PBMCs were incubated with PBS/EDTA (Life Technologies, Darmstadt, Germany) for 10 min, detached for counting and settled in experimental set up. For osteoclast differentiation in the experimental set-up, medium was supplemented with 25 ng/ml human M-CSF (Peprotech, Rocky Hill, USA) and 50 ng/ml human RANKL (Peprotech, Rocky Hill, USA).

Validation of osteoclasts via resorption assay and tRAp and DApi staining.
First, DAPI (4′, 6-diamidino-2′-phenylindole, dihydrochloride; Roche, Basel, Switzerland) staining was performed to count the number of nuclei per cell. After rinsing once in PBS, cells were fixed with 100 µl 4% paraformaldehyde (Merck Millipore, Burlington, USA) for 15 min at 22 °C and then incubated with 1 g/ml DAPI in PBS for 5 min. Cell nuclei glow blue when exposed to 345 nm lights (Immunofluorescence microscopy, Axioobserver Z.1, Zeiss, Jena, Germany). Three images were captured per well. These were analysed using the ImageJ software 17 . Only cells containing more than three nuclei were considered multinuclear cells. Second, TRAP activity staining was performed to prove the osteoclast character of the giant cells. The cells were fixed and stained using the acid phosphatase, leukocyte (TRAP) kit (Sigma-Aldrich; Steinheim, Germany) according to the manufacturer's instructions. Third, the osteoclast bone resorption assay was performed using a commercially available bone resorption assay kit (CosMo Bio, Tokyo, Japan) (Fig. 2). The resorption assay in the mono-culture arm was performed for 7 days only to ensure that the osteoclasts going in co-culture are functionally active cells. The cells were incubated on fluoresceinated calcium phosphate-coated microplate with RANKL (50 ng/ml) in the presence or absence of M-CSF (25 ng/ml). Bone resorption activity was evaluated by detecting the fluorescence intensity of conditioned medium at an excitation wavelength of 485 nm and an emission wavelength of 535 nm.

Mtt assay for cell proliferation.
Recent studies demonstrated that MTT assay provides high sensitivity, reproducibility, stimulation index and a wide linear response range for cell growth in comparison to other methods 21,22 . For MTT assay, viable cells are detected and quantified by mitochondrial conversion of 3-(4,5dime thylthiazol-2-yl)-2,5-diphenyltetrazoliumbromid (MTT, Sigma Aldrich, St Louis, USA). According to the instructions of the manufacturer, the different cells of the co-culture inserts were set in new wells and rinsed with PBS. Afterwards, 200 μl of MTT dye (1:10 in DMEM/ 10% FBS) was added, incubated for 1 h and rinsed again with PBS. Absorption in the control and experimental groups was measured using a multiwell ELISA reader at a wavelength of 570 nm (Clariostar, BMG-Labtech, Ortenberg, Germany).
Study design. PBMC were seeded at a density of 8.5 × 10 7 cells. After a 3-day expansion phase of PBMC, osteoclast differentiation was induced by 50 ng/ml RANKL and 25 ng/ml M-CSF. Differentiation was evaluated for 12 days and controlled by the resorption assay, TRAP and DAPI staining and FACS analysis. After setting the optimal time for the harvest of osteoclasts and the proof of osteoclast function, an indirect co-culture was initiated. A trans-well indirect co-culture system with inserts containing a membrane of 0.4 μm pore size was used to test cell lines separately (ThinCerts cell culture inserts; 0.4 μm pore size, Greiner-bioOne, Kremsmünster, Austria). This model has the advantage to allow exchange of paracrine signaling and response to soluble signalling factors between the upper (insert) and lower chamber of the well. Interaction among cell lines environment remains possible while cells are kept separated for further investigations.
Osteoclasts were seeded in 24-well-plates (Techno Plastic Products, Trasadingen, Switzerland), and the different oral cell lines (fibroblasts, osteoblasts and keratinocytes) were placed in the semipermeable inserts onto the osteoclasts for 5 days (Fig. 1). Co-culture of keratinocytes and PBMC was performed using standard keratinocyte medium or DMEM/F12 (1% or 5% KO-SR) instead of osteoclast medium. Different concentrations of RANKL and M-CSF were tested (RANKL: 50 ng/ml and 25 ng/ml; M-CSF: 25 ng/ml and 12.5 ng/ml) in OB-PBMC and FB-PBMC co-culture but not in OK-PBMC. After 5 days, an MTT assay was performed to investigate the proliferation of osteoblasts, keratinocytes and fibroblasts in co-culture. Further, DAPI staining was performed on the remaining osteoclasts in each co-culture (Fig. 1).

Results
OC differentiation of PBMC in a monoculture. OC differentiation was investigated by obtaining multinucleated cells from PBMC, FACS analysis of RANKL and calcitonin receptor expression, as well as by the functional resorption assay.
Resorption assay. Considering day 7 as a peak for the multinuclearity and RANK / calcitonin receptor expression in PBMC, we investigated the resorption function of OC at this time. The calcium phosphate (CaP)-coated plates showed distributed pit formation, indicating the resorption capacity of differentiated OC. As a shamcontrol, only PBMC medium was administered on the (CaP)-coated plates. The use of 25 ng/ml M-CSF and 50 ng/ml RANKL for OC differentiation led to a significant increase in the resorption activity of OC in contrast to RANKL-free M-CSF (mean of fluorescence intensity: M-CSF + RANKL: 2419 ± 64 versus M-CSF: 2227 ± 36; p = 0.012).

Discussion
Osteoclasts are a crucial cell type in bone remodelling. They are the main target of bisphosphonates and RANKL inhibitors in patients treated for osteoporosis, bone metastases and bone resorption-induced hypercalcemia. Administration of these drugs leads, therefore, to disturbed bone turnover and subsequent necrosis of the jaw 1,2 .
In the current study we established an indirect co-culture model to investigate contactless cell-cell interaction between PBMC-derived osteoclasts and the main oral cell lines in vitro. Thus, the effect of soluble substances and media on viability and proliferation of osteoblasts, fibroblasts and keratinocytes can be better studied and understood.

OC differentiation from PBMC in co-culture with OB, FB and OK. Several protocols have been
reported for osteoclast differentiation from PBMC in vitro 16,23,24 . These protocols are often time-consuming and incapable of generating sufficient numbers of osteoclasts 10 . Bernhardt et al. compared different PBMC isolation methods for obtaining a homogenous cell population and for improving osteoclast generation. They concluded that simple density gradient centrifugation, as performed in the current study, leads to optimal osteoclast differentiation between 9 and 16 days 25 . In this study, PBMCs were differentiated to osteoclasts by adding 25 ng/ml M-CSF and 50 ng/ml RANKL to the standard medium. Using this method, we could show that PBMC had the highest number of multinucleated cells after 7 days. A longer cultivation period did not improve OC differentiation. We assessed the advantage of RANKL supplementation to induce active osteoclasts in contrast to M-CSF alone. In accordance with our results, Cody et al. showed high OC differentiation after 7-8 days by adding a lower concentration of M-CSF to the standard medium 10 . Similarly, they observed more efficient OC formation in medium supplemented with 50 ng/ml RANKL compared with 10 ng/ml, whereas increasing the concentration to 100 ng/ml did not increase the osteoclast yield further 10 . In contrast to these findings, Costa-Rodriguez et al. showed that long-term supplementation-up to 21 days-with conditioned media derived from supernatants from either fibroblast or osteoblast cell cultures stimulated osteoclastogenesis of PBMC without addition of M-CSF and RANKL, since both cultures used to prepare the conditioned media express RANKL and M-CSF  proliferation of oB, fB and oK in co-culture with oc. The proliferation of the osteoblasts was compromised using medium with 25 ng/ml M-CSF and 50 ng/ml RANKL, while this effect was not evident at lower M-CSF and RANKL concentrations (12.5 ng/ml and 25 ng/ml, respectively). The intervention in the RANKL/ OPG ratio might be a reason for this observed impact on OB proliferation. It is known that RANKL up-regulation and OPG down-regulation lead to bone loss via several endogenous factors that regulate the RANKL/ RANK/OPG pathway, including cytokines and mesenchymal transcription factors 28,29 . The establishment of an indirect keratinocyte-PBMC co-culture seems to represent a challenging issue, as we could not keep the OK and  www.nature.com/scientificreports/ OC in co-culture viable or differentiated for the initial 5 days. Both cell lines underwent significant decreases in proliferation and differentiation in both media. The proliferation loss of OK in OC medium supports the assumption by Shipley et al., who reported on the advantage of serum-free medium for in vitro culturing of keratinocytes 30 . Compared with the standard medium of each cell line, co-culturing using DMEM/F12 with 5% of knock-out serum replacement solution, however, results in a high proliferation rate of OK along with a relatively reasonable OC differentiation.

conclusion
We show that OC differentiation of PBMC could be reached after 7 days in monoculture. Both OB and FB enhance PBMC differentiation onto OC in an indirect co-culture model and thus allow reduction of M-CSF and RANKL concentration in the medium without impact on OC differentiation. The results presented above contribute to the establishment of an indirect co-culture model for the investigation of environmental and soluble factors, which influence bone remodelling in pathogenic processes of the oral cavity, where OB, OC, FB and OK interact. The mechanism underlying altered osteoclasts differentiation in co-culture is an interesting and relevant aspect from the physiologic point of view, that has not been investigated in the current study but may be addressed in future studies to better develop therapeutic approaches. funding Open Access funding enabled and organized by Projekt DEAL.