Selective function-blocking monoclonal human antibody highlights the important role of membrane type-1 matrix metalloproteinase (MT1-MMP) in metastasis

Abstract

The invasion-promoting MT1-MMP is a cell surface-associated collagenase with a plethora of critical cellular functions. There is a consensus that MT1-MMP is a key protease in aberrant pericellular proteolysis in migrating cancer cells and, accordingly, a promising drug target. Because of high homology in the MMP family and a limited success in the design of selective small-molecule inhibitors, it became evident that the inhibitor specificity is required for selective and successful MT1-MMP therapies. Using the human Fab antibody library (over 1.25×109 individual variants) that exhibited the extended, 23-27 residue long, VH CDR-H3 segments, we isolated a panel of the inhibitory antibody fragments, from which the 3A2 Fab outperformed others as a specific and potent, low nanomolar range, inhibitor of MT1-MMP. Here, we report the in-depth characterization of the 3A2 antibody. Our multiple in vitro and cell-based tests and assays, and extensive structural modeling of the antibody/protease interactions suggest that the antibody epitope involves the residues proximal to the protease catalytic site and that, in contrast with tissue inhibitor-2 of MMPs (TIMP-2), the 3A2 Fab inactivates the protease functionality by binding to the catalytic domain outside the active site cavity. In agreement with the studies in metastasis by others, our animal studies in acute pulmonary melanoma metastasis support a key role of MT1-MMP in metastatic process. Conversely, the selective anti-MT1-MMP monotherapy significantly alleviated melanoma metastatic burden. It is likely that further affinity maturation of the 3A2 Fab will result in the lead inhibitor and a proof-of-concept for MT1-MMP targeting in metastatic cancers.

Keywords: MT1-MMP; antibody; cancer; metastasis; proteinase.

Figure 1. The 3A2 Fab is a selective, low nanomolar inhibitor of MT1-MMP

A. The clone, the sequence and the length of the CDR-H3 region in the selected Fab binders of MT1-MMP. B. Fab ELISA with the selected Fab binders of MT1-MMP. Left, the biotin-labeled catalytic domain of MT1-MMP (MT1-CAT) was captured onto streptavidin-coated wells of a 96-well plate. The Fab antibodies (0-8,000 nM) were allowed to bind to MT1-CAT. The bound antibodies were detected using HRP-conjugated anti-human Fab and a TMB/E substrate. Data are means ± SE from three individual experiments performed in triplicate. Right, the K_d values were calculated from the reactions in which a half of MT1-CAT was complexed with the added Fab. C. Inhibition of the MT1-MMP cleavage activity by the selected Fab antibodies. Left, the dose-response inhibition by the Fab fragments. The cleavage activity of MT1-CAT (5 nM) was measured in the presence of the increasing concentrations of the Fab antibodies (0-5,000 nM) using a Mca-PLGL-Dpa-AR-NH₂ substrate (1 μM). The residual cleavage activity was expressed in percent relative to a “no Fab” control. Data are means ± SE from 3 individual experiments conducted in triplicate. Right, the IC₅₀ values for the selected Fab antibodies. *^a and *^b, the weak inhibitory and non-inhibitory Fabs, respectively. D. The 3A2 Fab antibody is a selective inhibitor of MT1-MMP. The individual CAT of MT-MMPs (5 nM, each) were co-incubated with the increasing concentrations of the 3A2 Fab antibody (0-5,000 nM). The residual cleavage activity was measured using a Mca-PLGL-Dpa-AR-NH₂ substrate (1 μM). Left, the representative dose-response curves of the 3A2 Fab antibody against MT1-CAT. Right, the IC₅₀ values of the 3A2 Fab antibody in the individual MT-MMPs. RFU, relative fluorescence unit; *^c, no inhibition at the highest Fab concentration used. E. The 3A2 Fab antibody inhibits MT1-MMP proteolysis of AAT. AAT (2 μM) was co-incubated with MT1-CAT alone (40 nM, no inhibitor) or jointly with the 3A2 or DX2400 Fab fragments (20 and 200 nM, each), TIMP-1 (1,000 nM), TIMP-2 (20 nM) or GM6001 (1,000 nM). The reactions were analyzed by SDS-PAGE followed by Coomassie staining. DX, DX2400.

Figure 2. The 3A2 Fab antibody inhibits the functional activity of cellular MT1-MMP

A. The 3A2 Fab and the DX2400 Fab and IgG antibodies inhibited activation of the proMMP-2 zymogen by cellular MT1-MMP in HT1080 cells. Top, to induce proMMP-2 activation, cells were stimulated using phorbol 12-myristate 13-acetate (PMA; 50 ng/ml). Cells were then co-incubated with the 3A2 and DX2400 antibodies (20-200 nM, each) and also with the non-inhibitory MT1-MMP 3G4 IgG antibody (20-200 nM), TIMP-1 (1,000 nM), TIMP-2 (100 nM) and GM6001 (1,000 nM) controls. Medium aliquots were next analyzed by gelatin zymography to identify the status of MMP-2. Cells alone, no inhibitors were added to the cells. Bottom, the digitized zymogen:activation intermediate ratio in the MMP-2 samples. White and grey rectangles, zymogen and activation intermediate, respectively. The numbers indicate the percentage of the activation intermediate relative to the total combined amount of the zymogen and the intermediate. B. The 3A2 Fab and the DX2400 Fab and IgG do not affect cell viability. Normal mammary epithelial 184B5 cells were incubated alone (cells alone) or co-incubated with the antibodies (1,000 nM, each). Cell viability was measured using a luminescent ATP-Lite assay. Data are means ± SE from three individual experiments performed in triplicate. C. The 3A2 Fab antibody inhibits COL-I degradation by cellular MT1-MMP. MT1-MMP-deficient 184B5-mock and MT1-MMP-overexpressing 184B5-MT1 cells were plated onto COL-I layers and then incubated alone or co-incubated for 5 days with the 3A2 Fab (200 nM), DX2400 Fab or IgG (200 nM and 100 nM, respectively), TIMP-1 (1,000 nM), TIMP-2 (100 nM) or GM6001 (1,000 nM). After the removal of cells, COL-I was stained with Coomassie. The representative images from three independent experiments performed in triplicate are shown. D. Cell invasion through COL-I. 184B5-mock (mock) and 184B5-MT1 (MT1) cells (1×10⁵, each) were allowed to migrate alone (no inhibitor) or in the presence of the 3A2 or DX2400 Fab fragments (500 nM, each) or the indicated concentrations DX2400 IgG. GM6001 (1,000 nM) and 10% FBS were used as a control and a chemoattractant, respectively. Migration efficiency was calculated relative to MT1 cells, no inhibitor and 10% FBS (=100%). Data are means ± SE from three individual experiments conducted in triplicate. **, P < 0.05. DX, DX2400.

Figure 3. The 3A2 Fab antibody inhibits the functional activity of murine MT1-MMP

A. Murine melanoma B16F1-mMT1 cells stably transfected with murine MT1-MMP were co-incubated with the purified proMMP-2 zymogen alone (cells alone; 5-10 nM) or jointly with the 3A2 or DX2400 Fab antibodies (25-200 nM each; top and bottom panels, respectively). Where indicated, GM6001 (1,000 nM) was added to the cells. Medium aliquots were next analyzed by gelatin zymography to identify the status of MMP-2. B. The 3A2 Fab antibody inhibits COL-I degradation by murine cellular MT1-MMP. B16F1-mMT1 cells were plated onto COL-I layers and then incubated alone (no inhibitor) or co-incubated for 5 days with the 3A2 Fab (200 nM), DX2400 Fab and IgG (200 nM and 100 nM, respectively), and GM6001 (1,000 nM). After the removal of cells, COL-I was stained with Coomassie. The representative images from three independent experiments performed in triplicate are shown. DX, DX2400.

Figure 4. The 3A2 Fab reduces both the frequency and the size of melanoma metastatic nodules in mice

A. The catalytically active MT1-MMP is expressed in B16F1-mMT1 cells. Left, the status of MMP-2 (gelatin zymography; top panel) and MT1-MMP (Western blotting with the AB8345 antibody; bottom panel) in B16F1-mock and B16F1-mMT1 cells. Right, the fluorescent MP-3653 reporter (25 nM) reports the presence of the catalytically active MT1-MMP (green) in B16F1-mMT1 cells but not in B16F1-mock cells. DAPI (blue). Scale bar, 10 μm. B. Schematic representation of our injection protocol. Athymic mice received a single tail vein injection of B16F1-mock or B16F1-mMT1 on day 1 followed by the intraperitoneal injection of the 3A2 Fab (10-15 mg/kg) on days 1-12. Mice were euthanized and the lungs harvested on day 23. C, Top, representative images of the lungs obtained from the intact control (normal), B16F1-mock (mock), B16F1-mMT1 (mMT1) and B16F1-mMT1+3A2 animal groups (mMT1+3A2). Scale bar, 5 mm. Bottom, Western blotting (WB) of the lung extracts (20 μg total protein each) using the MT1-MMP AB8345 antibody. D. The weight and the number of the pulmonary metastatic lesions in the B16F1-mock, B16F1-mMT1 and B16F1-mMT1+3A2 mice. Normal, the lungs from the intact mice control. **, P < 0.05; NS, not significant.

Figure 5. The 3A2 Fab antibody competes with TIMP-2, but not with hydroxamate inhibitor, for its binding to MT1-MMP

A. The 3A2 and DX2400 Fab antibodies compete between themselves and also with TIMP-2 for their binding to MT1-MMP. 3A2/TIMP-2 and DX/TIMP-2, ELISA results in which the immobilized 3A2 and DX2400 Fab antibodies were each co-incubated with MT1-CAT (25 nM) and the indicated TIMP-2 – MT1-CAT molar ratios. 3A2/DX and 3A2/GM6001, ELISA results in which the immobilized 3A2 was co-incubated with MT1-CAT (25 nM) and the indicated DX2400 Fab or GM6001 – MT1-CAT molar ratio, respectively. In each ELISA, the bound MT1-MMP was then quantified using the rabbit polyclonal MT1-MMP antibody followed by the HRP-conjugated donkey anti-rabbit IgG and a TMB/E substrate. No MT1, MT1-CAT was not added. MT1, only MT1-CAT (25 nM) was added (=100%). Data are means ± SE from three individual experiments conducted in triplicate. **, P < 0.05. B. The 3A2 and DX2400 antibodies do not directly interact with the catalytic zinc vicinity. Left, the fluorescent MP-3653 reporter (25 nM) with a hydroxamate warhead did not detect the catalytically active MT1-MMP in MT1-MMP-deficient MCF7-mock cells. Right panels, MCF7-MT1 cells were left alone (no inhibitor) or co-incubated with the fluorescent MP-3653 reporter (25 nM) alone or jointly with the 3A2 Fab, the DX2400 Fab or IgG, the 3G4 IgG control, TIMP-1 (1,000 nM, each), TIMP-2 (50 nM) and GM6001 (100 nM). Scale bar, 10 μm. DX, DX2400.

Figure 6. The 3A2 Fab competes with TIMP-2 binding to MT1-CAT

A. The predicted structure of the hypothetical MT1-CAT·TIMP-2·3A2 Fab·GM6001 quadri-molecular complex. MT1-CAT is shown as cartoon (green), TIMP-2 and the 3A2 Fab are shown as yellow and cyan surfaces. GM6001, red sticks; the Phe260 residue of the MT1-CAT sequence, black sticks; the catalytic and structural zinc ions in MT1-CAT, black and grey spheres, respectively; the structural calcium ion, green sphere. A putative region where TIMP-2 clashes with the 3A2 moiety is shown in purple. The figure summarizes a detailed superimposition analysis of the available crystal structures of the tudor domain of human TDRD3 in complex with an anti-TDRD3 Fab (PDB 3PNW), MT1-CAT complexed with TIMP-2 (PDB 1BQQ) and the anthrax toxin lethal factor bound to GM6001 (PDB 4PKW). B, Unlike TIMP-2, the 3A2 Fab does not bind to the catalytic zinc vicinity in MT1-MMP. Left, close-up of the hypothetical MT1-CAT·TIMP-2·GM6001 complex shows that the bound GM6001 penetrates into the space occupied by TIMP-2 [46, 48, 49]. As a result, TIMP-2 and GM6001 compete for their binding to MT1-MMP. Right, two rotated close-ups of the MT1-CAT·3A2 Fab ·GM6001 complex clearly indicate that the 3A2 Fab cannot interact with the catalytic zinc vicinity (black sphere) in the MT1-MMP active site. As a result, the 3A2 Fab did not compete with GM6001 for the binding to MT1-CAT.