Tumor detection by imaging proteolytic activity

Abstract

The cell surface protease membrane-type serine protease-1 (MT-SP1), also known as matriptase, is often upregulated in epithelial cancers. We hypothesized that dysregulation of MT-SP1 with regard to its cognate inhibitor hepatocyte growth factor activator inhibitor-1 (HAI-1), a situation that increases proteolytic activity, might be exploited for imaging purposes to differentiate malignant from normal tissue. In this study, we show that MT-SP1 is active on cancer cells and that its activity may be targeted in vivo for tumor detection. A proteolytic activity assay with several MT-SP1-positive human cancer cell lines showed that MT-SP1 antibodies that inhibit recombinant enzyme activity in vitro also bind and inhibit the full-length enzyme expressed on cells. In contrast, in the same assay, MT-SP1-negative cancer cell lines were inactive. Fluorescence microscopy confirmed the cell surface localization of labeled antibodies bound to MT-SP1-positive cells. To evaluate in vivo targeting capability, 0.7 to 2 nmoles of fluorescently labeled antibodies were administered to mice bearing tumors that were positive or negative for MT-SP1. Antibodies localized to MT-SP1-positive tumors (n = 3), permitting visualization of MT-SP1 activity, whereas MT-SP1-negative tumors (n = 2) were not visualized. Our findings define MT-SP1 activity as a useful biomarker to visualize epithelial cancers using a noninvasive antibody-based method.

Figure 1

Antibodies are specific for active MT-SP1. SPR binding curves of MT-SP1 (black) and the catalytically inactive mutant R15A (gray) to the A11 Fab show robust binding to active enzyme and a lack of binding to the zymogen, indicating that the inhibitor is specific for the active form of the enzyme. Binding assays with enzyme concentrations up to 1 μM showed similar results (not shown).

Figure 2

MT-SP1-specific inhibitors reduce cell-associated proteolysis of a fluorescent P1-arginine protease substrate. Spectrofluor-tPA was added to seven human cancer cell lines with and without α-MT-SP1 Fab. Because the inhibitors are specific, any reduction in proteolysis can be attributed to an inhibition of active MT-SP1. Black bars represent total P₁-Arg proteolytic activity of uninhibited cells, while the grey bars show the activity of cells incubated with α-MT-SP1 Fab. Those cell lines which express MT-SP1 mRNA – MDA-MB-468, MCF-7, HT29 and LNCaP - do show a reduction in proteolysis upon the addition of antibody inhibitor, while the MT-SP1-negative cell lines MDA-MB-231, HT1080, and COLO 320 DM do not.

Figure 3

Fluorescently-labeled α-MT-SP1 scFv localizes to MT-SP1-positive cells in cell culture. (a) HT-29, LNCaP and MCF-7 cells express MT-SP1 mRNA while MDA-MD-231 and COLO 320DM cells (b) do not. Antibodies do not significantly displace HAI-1. HT-29 cells which have been incubated with recombinant HAI-1 (c) show much less surface labeling upon exposure to fluorescently labeled scFv than those which have not been blocked (d). Scale bar = 50 μm. Images were corrected as described in supplemental materials.

Figure 4

α-MT-SP1 antibodies are able to selectively target MT-SP1-positive tumors in vivo. AlexaFluor 680-labeled A11 IgG was injected and imaged using a 2-D fluorescent imager. Tumors are indicated by arrows. In MCF-7/Luc+ xenograft mice, the tumors become selectively labeled with minimal background after 50 hours (a). In the MDA-MB-231/Luc+ (MT-SP1-negative) models, however, no tumor labeling was seen over the same time period (b). As an alternate method for detecting the tumors, mice were injected with approximately 2 mg of luciferin and imaged after 10-15 minutes. Because the Luc+ cells have been engineered to express luciferase, a bioluminescent signal is seen at the tumor, confirming location and viability of the xenograft.