Aberrant ADAM10 expression correlates with osteosarcoma progression

Paraformaldehyde-fixed, paraffin-embedded human osteosarcoma tissue chip slides were purchased from US Biomax (Rockville, MD, USA). Each slide contained 40 duplicate samples of osteosarcoma tissues, and pathological stages included IA (n = 3), IB (n = 5), IIA (n = 10) and IIB (n = 22), according to the Musculoskeletal Tumor Society Staging System. This study was approved by the ethics committee of Daping Hospital of the Third Military Medical University.

Mouse anti-human CD31 monoclonal antibody (Ab), rabbit anti-human ADAM10 (cytoplasmic domain) polyclonal Ab and rabbit anti-human activated Notch1 NICD polyclonal Ab were all purchased from Abcam (Cambridge, UK). Isotype-specific control Abs were also obtained from Abcam and were used to control staining specificity. Fluorescein isothiocyanate (FITC)-conjugated goat anti-mouse IgG Ab (ZSGB-BIO, Beijing, China) and cyanine 3 (Cy3)-conjugated goat anti-rabbit IgG Ab (Beyotime Institute of Biotechnology, Jiangsu, China) were used as secondary Abs. Antifade mounting medium was also obtained from Beyotime Institute of Biotechnology.

Paraffin tissue sections were deparaffinized in xylene, rehydrated through graded ethanol series and washed in 10 mM phosphate-buffered saline (PBS), pH 7.4. Hematoxylin and eosin (H&E) staining was performed according to a standard protocol. The histological types were assessed by an experienced pathologist specializing in osteosarcoma diagnosis. For immunofluorescent staining, antigen retrieval was performed by incubating tissue sections for 20 minutes in 0.01 M sodium citrate buffer, pH 6.0, and heated in a microwave oven, then sections were permeabilized with 0.1% Triton X-100. After blocking with 5% bovine serum albumin/PBS for 30 minutes, three sequential slides were then incubated with anti-CD31 Ab, anti-ADAM10 (cytoplasmic domain) Ab and anti activated Notch1 (NICD) Ab, respectively. For double-immunofluorescence staining, two additional sequential slides were incubated with either anti-CD31 Ab (mouse anti-human) plus anti-ADAM10 Ab (rabbit anti-human) or anti-CD31 Ab plus anti-NICD Ab (rabbit anti-human), respectively. Isotype-specific control Abs were used to test staining specificity. Slides were incubated overnight at 4°C. After being washed with PBS, slides were stained with secondary FITC anti-mouse Ab and Cy3 anti-rabbit Ab for 30 minutes at room temperature (RT). Afterward, nuclei were stained with 4′,6-diamidino-2-phenylindole (DAPI) for 10 minutes at RT. Images were obtained using the MRC-600 digital confocal laser scanning system (Bio-Rad Laboratories, Hercules, CA, USA).

TRAP staining was performed on deparaffinized and rehydrated sections using the Acid Phosphatase, Leukocyte (TRAP) Kit (Sigma-Aldrich, St Louis, MO, USA) according to the manufacturer’s instructions. DAPI staining was performed afterward as mentioned above to visualize the nuclei.

Acquired images were exported and merged using Image-Pro Plus version 6.0 software (MediaCybernetics, Rockville, MD, USA). NICD and CD31 density percentages were measured as the percentage of positive FITC signal area against the whole field area under 100× magnification as described by Connor et al. [17]. ADAM10-positive tumor cells were counted in ADAM10/DAPI merged images under the same magnification, and analysis was performed with tools in Image-Pro Plus software. Statistical differences were determined using a two-tailed, unpaired Student’s t-test.

We first examined the expression of ADAM10 in osteosarcoma tissue chips. Under 400× magnification, a polarized and condensed expression pattern of cytoplasmic ADAM10 was observed in all cases and stages of osteosarcoma (Figure 1). In order to determine whether ADAM10 was differentially expressed among different stages of osteosarcoma, ADM10⁺ tumor cells were counted under 100× ADAM10/DAPI merged images. The number of ADM10⁺ tumor cells in stages IIA and IIB increased significantly compared with those in stage IA (Figure 1), suggesting that elevated ADAM10 levels in tumor cells might be involved in the progression of osteosarcoma.

After the immunofluorescent staining, osteosarcoma tissue chips were stained with H&E. Four types of osteosarcoma were recognized, including giant cell-rich osteosarcoma, osteoblastic osteosarcoma, fibroblastic osteosarcoma and chondroblastic osteosarcoma (Additional file 1: Figure S1, upper panels). We then reanalyzed the number of ADAM10⁺ tumor cells based on different histological types and found that osteoblastic osteosarcoma had significantly more ADAM10⁺ tumor cells compared with chondroblastic and fibroblastic osteosarcomas (Additional file 1: Figure S1, lower graph), suggesting that ADAM10 might be involved specifically in the pathogenesis of osteoblastic osteosarcoma.

As ADAM10 is a major regulator of Notch signaling via its shedding of Notch receptors [8, 9], we sought to determine whether increased expression of ADAM10 in osteosarcoma progression would also lead to increased activation of Notch1. We detected the expression of activated Notch1 (NICD) in another sequential slide. As shown in the upper panels of Figure 2, Notch1 was activated in osteosarcoma tissue at all stages; however, its activation was not significantly correlated with pathological staging, as shown by NICD density from stage IB to stage IIB. These results suggest that Notch1 was activated in osteosarcoma but that its activation might not contribute to disease progression.

Previous studies have shown that the expression of CD31 was significantly increased in metastatic osteosarcoma compared with primary osteosarcoma [16]. However, the differences between nonmetastatic groups are unclear. Hence we performed FITC-CD31 staining on all osteosarcoma tissues from stages IA to IIB. Although a trend of increased CD31 expression was observed as osteosarcoma progressed, no significant differences were found between different pathological stages in terms of CD31 density (Figure 3), suggesting that vascularity (represented by the expression of CD31) was not dramatically altered during the pathological progression of nonmetastatic osteosarcoma, a scenario different from metastatic groups.

We found that the multinucleated giant cells in giant cell-rich osteosarcoma stained positively with CD31 (Figure 4, left panels), suggesting their angiogenic property. Moreover, cytoplasmic ADAM10 and activated Notch1 were colocalized within these giant cells (Figure 4), suggesting that ADAM10/Notch1 signaling was activated in these angiogenic tumor cells. By TRAP staining, we observed that TRAP was absent in multinucleated cells in giant cell-rich osteosarcoma tissue, whereas TRAP⁺ cells were found in osteoblastic, fibroblastic and chondroblastic osteosarcoma tissue (Figure 5). Taken together, these results suggest that what were previously called osteoclast-like multinucleated giant cells [2] are not osteoclasts, but rather angiogenic tumor cells that involve ADAM10/Notch1 signaling activation.