Macrophage Colony-stimulating Factor Mediates Invasion of Ovarian Cancer Cells through Urokinase

The macrophage colony-stimulating factor (CSF-1) is best known as a hematopoietic cytokine important to macrophage activation. Recently, the importance of CSF-1 and its receptor (encoded by the c-fms protoonco-gene) in epithelial ovarian cancer has also been recognized, with overexpression of CSF-1 denoting poor prognosis in ovarian cancer patients. During macrophage activation, CSF-1 promotes urokinase-type plasminogen activator (uPA) activity; in macrophages and in malignant cells of lung, breast, colon, and prostatic origin, uPA activity is strongly correlated with the ability to invade and, in the malignant cells, to metastasize. While there is clear evidence of CSF-1 and uPA expression in primary and metastatic ovarian cancer, the significance of their expression to invasion of these cells has not been explored. We find that all of our ovarian cancer cell lines which we have studied co-express CSF-1 and uPA transcripts and protein. Urokinase expression in these ovarian cancer cell lines correlates with the degree of tumorigenicity in nude mice, with the most virulent tumor resulting from Hey cells, a strong expressor of uPA. We studied the invasion of these primary and established ovarian cancer cells through a Matrigel (reconstituted basement membrane matrix) barrier. The ability of ovarian cancer cells to invade is strongly correlated with endogenous CSF-1 expression (Pearson's correlation, r = 0.91; P = 0.01). A total of 0.90 - 0.16% of Bix3 cells (very weak expressor of CSF-1) invaded through the barrier, in contrast to 6.95 - 0.75% of Hey cells (strong CSF-1 expressor) and 10.44 - 2.33% of Bixler cells (the strongest CSF-1 expressor). We studied the ability of two of the cell lines to invade human laminin and type IV collagen (Bix3, a weak invader of Matrigel, and Hey, a strong invader), to determine (a) whether our results on a Matrigel matrix may represent a relevant model for invasion in humans and (b) whether there is a potential confounding effect from the cytokines and proteases in Matrigel. On this human simple matrix, we confirm that Bix3 is a weakly invasive cell line (033 - 0.04% invasion) which contrasted to the strongly invasive Hey cell line (8.51 - 0.47%). Treatment of Bix3 cells with exogenous CSF-1 stimulates percentage of invasion by 2-fold and results in a similar increase in the level of uPA transcripts and cellular associated uPA antigen. Furthermore, cell surface-bound uPA increased from 74% in the absence of CSF-1 to 100% (fully saturated) in the presence of CSF-1. We show that CSF-1-stimulated invasion of Bix3 cells is completely blocked by anti-uPA antibodies but not by preimmune IgG. Collectively, these data suggest that stimulation of invasion by exogenous CSF-1 is mediated entirely through uPA activity. But uPA expression does not directly correlate with invasion. In fact, inhibition of uPA activity in Hey cells (the strongest uPA expressor) by a saturable amount of anti-uPA antibodies results in only a 50% inhibition of invasion. This suggests that, while uPA activity is associated with extent of intraperitoneal tumorigenicity in nude mice and mediates stimulation of invasion by exogenous CSF-1, it is an important, but not exclusive, mediator of ovarian cancer cell invasion.