(A) Nude mice were orthotopically implanted with 10⁷ wild-type (Wt) or LOX shRNA-expressing MDA-MB-231 human breast tumor cells. Lungs were excised 6 weeks later, and frozen serial sections were stained with either H&E or with pan-cytokeratin (green) to identify tumor cells and CD11b (red) to identify myeloid cells. Arrows indicate pulmonary cell clusters (foci). Scale bar = 75µm.
(B) Lungs from mice bearing Wt or LOX shRNA-expressing tumors were homogenized and analyzed by flow cytometry for numbers of CD11b+ myeloid cells, c-Kit+ (CD117) myeloid progenitor cells, or F4/80+ mature macrophages. Data are mean ± SEM. * = p<0.05.
(C) Tumor-free mice were injected with the indicated CM daily for 3 weeks. Homogenized lungs were analyzed for CD11b, c-Kit, and F4/80 positive cells by flow cytometry. LOX Ab = LOX-specific antibody; BAPN = small molecule inhibitor of LOX. Data are mean ± SEM. * = p<0.05 relative to control; ** = p<0.05 relative to mice injected with Wt CM.
(D) Merged immunofluorescent staining of LOX (green) and CD11b+ cells (red) in representative lung sections from tumor-free mice injected daily for 3 weeks with the indicated CM. Co-localization is indicated in yellow. Low/high LOX = relative concentration of purified LOX; MMP-I = matrix metalloproteinase inhibitor. Scale bar = 150µm.
(E) Image analysis of lung sections from mice in Figure 1D. Data indicate the relative area of LOX (green) or CD11b (red) staining relative to control mice. Data are mean ± SEM. * = p<0.05 relative to control; ** = p<0.05 relative to mice injected with Wt CM.

(A) Matrigel-coated wells were incubated with the indicated additives for 24hr; Matrigel cross-linked with glucose was included for comparison. Solutions were removed and CD11b+ cells isolated from murine whole bone marrow were added. Numbers of CD11b+ cells remaining in solution were quantified after 2.5hr. Data are mean ± SEM. * = p<0.05 relative to control; ** = p<0.05 relative to matrix pre-incubated with LOX.
(B) Naïve mouse lung tissue was excised and a 2cm³ piece was incubated in serum-free media containing either LOX or glucose for 6hr. Media was changed, CD11b+ cells were added, and the numbers of cells remaining in solution after 12hr were quantified. Data are mean ± SEM.
(C) Lung tissue from B was frozen, sectioned, and stained for LOX (green) and CD11b+ cells (red). Scale bar = 300µm.
(D) Gelatin zymography showing MMP-2 activity of monocytes in contact with Matrigel or collagen IV pre-incubated with the indicated CM.
(E) Gelatin zymography showing MMP-2 activity of freshly harvested BMDCs in contact with collagen IV pre-incubated with the indicated CM.
(F) Matrigel filters were incubated with the indicated CM or purified protein for 24hr. The CM was then removed and freshly harvested whole murine bone marrow cells were allowed to invade through the “modified” Matrigel. BMDCs that invaded through the modified Matrigel were also stained for CD11b and c-Kit. Data are mean ± SEM. * = p<0.05 relative to control; ** = p<0.05 relative to matrices pre-incubated with Wt CM.
(G) Matrigel filters were pre-incubated as in 2F, and invasion of isolated CD11b+ cells or c-Kit+ cells through the modified matrix was quantified. Data are mean ± SEM. * = p<0.05 relative to control; ** = p<0.05 relative to matrices pre-incubated with LOX.
(H) Mice with wild-type bone marrow or MMP-2 knockout bone marrow were injected daily with the indicated CM for 3 weeks prior to flow cytometry analysis of lungs for CD11b, c-Kit, or F4/80 positive cells. Data are mean ± SEM. * = p<0.05 relative to no CM mice; ** = p<0.05 relative to mice with Wt bone marrow injected with Wt CM.
(I) Immunofluorescent staining for LOX, CD11b+ cells, and MMP-2 in representative frozen serial sections of lungs from mice with wild-type or MMP-2 knockout bone marrow injected with Wt CM. Scale bar = 75µm.

(A) ELISA to detect collagen IV remodeling via peptide formation. Collagen IV peptides exogenously added to media (white bars) are provided for comparison. Matrigel pre-incubated for 24hr with the indicated additives (grey bars) were subsequently contacted with CD11b+ cells for 24hr. Peptides released into the surrounding media were quantified with the ELISA. Plasma samples from the indicated tumor-bearing mice (black bars) were also analyzed. Data are mean ± SEM. * = p<0.05 relative to control; ** = p<0.05 relative to matrices pre-incubated with LOX (grey) or relative to Wt tumor-bearing mice (black).
(B) Sections of lungs from Wt tumor-bearing mouse illustrating loss of collagen IV antibody epitope in some areas through LOX-mediated collagen IV remodeling. Collagen I staining was not affected. Scale bar = 150µm.
(C) Numbers of isolated CD11b+ cells or c-Kit+ cells invading through naïve Matrigel toward collagen IV peptides in the bottom of the transwell. Data are mean ± SEM. * = p<0.05 relative to control (no peptides).
(D) Flow cytometric quantification of CD11b+ cells and tumor cells (human pan-cytokeratin-positive) in lungs of mice bearing LOX shRNA-expressing tumors. Mice were “pre-conditioned” by injection of Wt CM or purified LOX protein for 2 weeks after tumor implant. Lungs were harvested 6 weeks after tumor implant. Data are mean ± SEM. * = p<0.05 relative to control mice; ** = p<0.05 relative to mice pre-conditioned with LOX protein.
(E) Flow cytometric quantification of CD11b+ cells and tumor cells in lungs of mice with LOX shRNA-expressing tumors treated with the indicated solutions. Data are mean ± SEM. Clod = clodronate; * = p<0.05 relative to control shRNA tumor-bearing mice; ** = p<0.05 relative to shRNA tumor-bearing mice injected with LOX protein.
(F) Same experiment as E, using 4T1 murine mammary tumor cells in Balb/c mice. Pulmonary cell foci (clusters) were quantified from H&E-stained lung tissue. Data are mean ± SEM. * = p<0.05 relative to control 4T1 shRNA tumor-bearing mice; ** = p<0.05 relative to 4T1 shRNA tumor-bearing mice injected with LOX protein.
(G) Flow cytometry analysis for CD11b+ cells in lungs of tumor-free Balb/c mice or Balb/c mice bearing LOX shRNA-expressing 4T1 tumors injected with the indicated CM daily for 3 weeks. Pulmonary cell foci were quantified from H&E-stained lung tissue. Data are mean ± SEM. * = p<0.05 relative to control mice; ** = p<0.05 relative to mice injected with Wt CM.
(H) Lungs of Balb/c mice implanted with LOX shRNA-expressing 4T1 tumors and injected with the indicated CM daily for 3 weeks. Arrows indicate macroscopic lung metastases.

(A) Serial sections from liver tissue of Wt tumor-bearing mouse stained for fibronectin, LOX, and BMDCs. Scale bar = 50µm.
(B) Flow cytometry analysis for CD11b+ cells in livers and brains of Nude mice bearing MDA-MB-231 human breast tumors or Balb/c mice with 4T1 murine mammary tumors. Data are mean ± SEM. ** = p<0.05 relative to Wt tumor-bearing mice.
(C) Tissue microarrays (TMAs) of clinical metastases stained for CD11b+ cells, LOX, or c-Kit+ cells. Samples from normal cerebral cortex and liver are provided as negative controls. Metastatic and normal TMAs were stained simultaneously, and were photographed with identical microscope and camera settings. Arrows indicate regions of CD11b+ cells or c-Kit+ cells. Scale bar = 150µm.
(D) (1) Hypoxic primary tumor cells secrete LOX into the bloodstream. (2) LOX accumulates in the lungs of tumor-bearing mice and cross-links collagen IV. (3) Adhesion of CD11b+ cells to cross-linked matrix increases BMDC MMP-2 activity. Collagen IV remodeling by LOX and MMP-2 leads to peptide formation, invasion of CD11b+ cells, and increased recruitment of BMDCs. (4) LOX-dependent formation of the pre-metastatic niche enhances metastatic growth.