Figure 1.
Localization of MMP-2 in the human and murine bone microenvironment.
A, Fluorescent TRAcP staining (green) was used to localize osteoclasts (closed arrow heads) while immunofluorescence was used to localize osteoblast marker osteocalcin and MMP-2 (red) in human samples of breast to bone metastasis (n = 11). DAPI (blue) was used as a nuclear stain. H&E was used to identify tumor-bone morphology. Arrows indicate osteoblasts while open arrow heads indicate osteocytes. Scale bars represent 50 µm. The localization of MMP-2 in human GCT osteoclasts (TRAcP positive, brown) was also assessed. B, Immunohistochemistry for MMP-2 (brown) and osteocalcin (OC) in the wild type and MMP-2 null murine bone and tumor-bone microenvironment. Rabbit IgG was used as a negative control. Arrows indicate osteoblasts while open arrow-heads indicate osteocytes. Colorimetric TRAcP staining (red) was also used to identify osteoclasts (arrow heads).
Figure 2.
Host-derived MMP-2 impacts mammary tumor growth in the bone microenvironment.
A, Representative timeline of PyMT-Luc luciferase expression in FVB wild type (WT; n = 10) and MMP-2 null (MMP-2−/−; n = 10) mice. Hotter colors indicate areas of increased luciferase activity. B, Graph represents PyMT-Luc tumor growth over time as assessed by luciferase activity in WT and MMP-2−/− animals. C, PyMT-Luc fail to grow in the MMP-2−/− bone microenvironment as assessed over a 25 day period. PyMT-Luc were injected intratibially into syngeneic FVB wild type (WT; n = 10) or MMP-2 null (MMP-2−/−; n = 10) mice. WT mice were euthanized on day 15 due to tumor size. D, 17L3C-Luc, an unrelated cell line derived from the PyMT model of mammary gland tumorigenesis, was intratibially injected into wild type (WT; n = 5) or MMP-2 null (MMP-2−/−; n = 5) mice. Luciferase activity was assessed as a measure of tumor growth over a 7-day period. Data are mean ± SD; Asterisk denotes statistical significance (p<0.05).
Figure 3.
Host-derived MMP-2 impacts tumor survival in the bone microenvironment.
A, Representative photomicrograph of MCM2 positive proliferating (brown) cells in the wild type (WT) and MMP-2 null (MMP-2−/−) tumor-bone microenvironment. B, Proliferation in the tumor-bone microenvironment as a function of total cell number was assessed by staining for Mcm2 in tumor bearing tibias of WT and MMP-2 null mice at day 3 and day 6 post-surgery. C, Representative photomicrograph of caspase-3 positive apoptotic (brown) cells in the wild type (WT) and MMP-2 null (MMP-2−/−) tumor-bone microenvironment. D, Apoptosis in the tumor-bone microenvironment as a function of total cell number was assessed by staining for cleaved caspase-3 at day 3 and day 6 post surgery. Data are mean ± SD; n.s. implies a non-significant p value (p>0.05).
Figure 4.
Tumor induced osteolysis is controlled by host MMP-2.
A, Representative μCT scans of trabecular bone from tumor bearing and sham injected limbs of wild type (WT) and MMP-2 null (MMP-2−/−) mice at day 9. Graph illustrates ratio of bone volume to total volume (BV/TV). B, Representative H&E stained photomicrographs of tumor bearing tibias from WT and MMP-2−/− mice. Scale bars represent 1 mm. Graph represents the BV/TV ratio. C, Representative radiographic images from tumor injected WT and MMP-2−/− animals at day 9. Arrow indicates lytic tumor lesions in the wild type group. Graph represents the tumor volume (TuV) over total volume (TV) ratio. D, Representative TRAcP stained photomicrographs of tumor injected WT and MMP-2−/− animals at day 9. Scale bar represents 50 µm. Graph indicates the number of osteoclasts/mm of trabecular bone. Data are mean ± SD. Asterisk denotes that p<0.05.
Figure 5.
Absence of host MMP-2 does not impair osteoclast precursor function.
A, Quantitative analysis of isolated wild type (WT) and MMP-2 null (MMP-2−/−) CD11b+ve osteoclast precursor migration to 10% serum or control serum free media. B, Analysis of the osteoclastogenic capability of WT or MMP-2−/− osteoclast precursors. Arrows indicate TRAcP positive (red) multinucleated (blue) osteoclasts. Scale bars represent 50 µm. C, Analysis of the number of WT and MMP-2−/− mature osteoclasts formed/well in each group (n≥3/group). D, Pit formation assay on dentin discs to test the functionality of mature WT and MMP-2−/− osteoclasts. Asterisk denotes that p<0.05 while p = n.s. denotes non-significant p values.
Figure 6.
Osteoblast-derived MMP-2 controls TGF-β bioavailability.
A, Representative phase contrast photomicrographs of isolated wild type (WT) and MMP-2 null (MMP-2−/−) primary osteoblast morphology. Osteoblasts were characterized by differentiation in the presence or absence of OGM and measuring alkaline phosphatase activity. Asterisk denotes statistical significance (p<0.05) between the OGM and the respective control treated groups. Conditioned media derived from wild type (WT CM) and MMP-2 null (MMP-2−/− CM) primary osteoblasts was also assessed for the presence of MMP-2 using gelatin zymography. Arrow denotes latent MMP-2 while arrow head denotes active MMP-2. HT1080, a human firbosarcoma cell line was used as a positive control for MMP-2. B and C, PyMT-Luc cells were treated with conditioned media from wild type (WT CM) or MMP-2 null (MMP-2−/− CM) osteoblasts. The metabolic activity of PyMT-Luc cells treated with WT CM or MMP-2−/− CM was assessed by MTT assay (B) and survival was determined by soft agar colony formation (C). D, Levels of TGFβ in WT and MMP-2−/−CM were assessed by ELISA. Changes in TGFβ levels were also assessed after the addition of recombinant MMP-2 (rMMP-2; 100 ng/ml CM for 3 hours at 37°C). E, The ability of MMP-2 (100 ng recombinant MMP-2/ml CM 3 hours at 37°C) to process LTBP-3 derived from transfected COS-7 cells was assessed by immunoblot as described [32]. Plasmin (1 µg/ml of CM 1 hour at 37°C) was used as a positive control. F, LTBP-3 levels and the impact of recombinant MMP-2 on LTBP-3 processing in the CM derived from WT and MMP-2−/− osteoblasts was determined by immunoblot analysis. MW is indicated by 230 kDa. Arrow indicates full length LTBP-3.
Figure 7.
MMP-2 mediates tumor survival in a TGFβ dependent manner.
A, Tumor survival in the presence of 1 ng/ml recombinant TGFβ was assessed by soft agar colony formation assay. B, The number of PyMT-Luc colonies formed in response to 10 days of treatment with control media (5% serum containing αMEM) or conditioned media derived from wild type or MMP-2 null osteoblasts was analyzed in a 2D (plastic) assay (non-treated group). Under similar conditions the impact of TGFβ inhibition was also determined (matched isotype antibody was added to control wells). Differences are significant as assessed by ANOVA. C, The effects of blocking TGFβ in the CM derived from WT and MMP-2−/− osteoblasts was determined by soft agar colony formation assay. Data are mean ± SD. Asterisk indicates p<0.05 and n.s indicates non significance. Double asterisk denotes that the addition of recombinant MMP-2 to MMP-2−/− CM significantly enhances tumor survival compared to MMP-2−/− CM alone. D, Osteoblast-derived MMP-2 and TGF-β mediate tumor survival without impacting colony size. Using a soft agar colony formation assay, no difference in colony size was detected between the various treatment conditions. Data are mean ± SD.
Figure 8.
Osteoblast-derived MMP-2 impacts TGFβ activation and tumor survival in the in vivo tumor-bone microenvironment.
A, The levels of TGFβ in normalized tumor-bone lysates derived from WT and MMP-2−/− tumor or sham injected tibias was assessed by ELISA. B, Representative immunoblots for phospho-SMAD (pSMAD2), total smad2 (SMAD2), phospho-AKT (pAKT), total AKT (AKT) and actin (loading control) in the tumor-bone lysates derived from WT and MMP-2−/− mice. Densitometry on immunoblots generated from tumor bone lysates of at least 5 animals per group was used to generate graphs of the ratio of pSMAD2/SMAD2 and pAKT/AKT. Data are mean ± SD. *, p<0.05. C, Osteoblast derived MMP-2 controls TGFβ activation and mammary survival in the tumor-bone microenvironment. (1) Tumor derived signals induce osteoblast retraction from mineralized matrix and the resorption of the non-mineralized osteoid canopy. (2) We posit that MMP-2 processing of the factors that sequester TGFβ in a latent state such as LTBP-3 and LAP initiates TGFβ activation. (3) Our data show that the activation of TGFβ by osteoblast derived MMP-2 mediates tumor survival and we hypothesize this mini-vicious cycle is essential for the establishment of the traditional vicious cycle of tumor induced osteolysis.