Tissue factor is an angiogenic-specific receptor for factor VII-targeted immunotherapy and photodynamic therapy

Angiogenesis, the formation of new blood vessels from existing blood vessels, is a common and critical pathological process in many human diseases [1], including but not limited to cancer, age-related macular degeneration (AMD), endometriosis and rheumatoid arthritis. For example, in cancer, the newly formed blood vessels, usually termed as tumor angiogenic vessels or tumor neovasculature, can be differentiated from normal (resting, quiescent) vessels in normal tissues in morphology (vascular and endothelial shapes), and function (increased permeability) [2, 3]. As such, tumor neovasculature provides not only nutrients and oxygen for cancer cells to proliferate, but also serves as a conduit for cancer cells to metastasize. Thus, the tumor neovasculature is a key target for development of therapeutic agents. The search for biomarkers as target molecules specific for angiogenic vascular endothelial cells (VEC), the inner layer of pathological neovasculature, is promising and critical in developing novel neovascular-targeting therapies for these common human diseases [1].

One potential approach to eliminating pathological neovascularization involves attacking the subset of vascular endothelial cells that are angiogenic, via a cell-type-specific surface molecule coupled to a cell-killing agent. Based on the in vivo observation of TF expression on tumor vascular endothelial cells in breast cancer tissues from patients [4] and some mouse models of human tumor xenografts [5‐7], we have begun developing such an approach using tissue factor (TF), a 47-kDa membrane-bound receptor [4, 8‐11], as the target molecule on angiogenic VECs. To assemble this TF-directed treatment, we used, as the starting molecule, the principal ligand of TF, factor VII (fVII) [12]. To eliminate the coagulation activity of zymogenic fVII, a coagulation active site lysine 341 was replaced by an alanine (thereafter containing the mutation of K341A in fVII polypeptides unless specified) using site-directed mutagenesis [5, 13, 14]. Using this strategy, we have pursued two different therapeutic approaches; immunotherapy (by fusing fVII to an IgG1 Fc as an antibody-like immunoconjugate, i.e., fVII/IgG1 Fc, also called ICON) [5, 14] and photodynamic therapy (PDT) (by chemically conjugating fVII peptides with photosensitizers (PS), fVII/PS, for fVII-targeted PDT, abbreviated as fVII-tPDT) [15, 16] for the treatment of cancers including melanoma [5, 13], prostate [14], head and neck [17], breast [7, 15, 16] and lung [6], as well as age-related macular degeneration (AMD) [18‐20] and endometriosis [21] in preclinical animal models and in phase I and II clinical trials for AMD patients (NCT01485588) [22]. In vivo data confirm that the TF-targeted agents ICON and fVII-tPDT can selectively eradicate pathological neovasculature in cancer [13, 23], AMD [18‐20] and endometriosis [21]. However, it remains unclear which angiogenic growth factor induces TF expression commonly detected in these pathological angiogenesis-dependent malignancy and non-malignant diseases, whether TF is also selectively expressed by in vitro angiogenic VEC models and whether TF-targeted agents can selectively eradicate in vitro models of angiogenic VECs via targeting TF without harming quiescent VECs. This study is designed to address these questions by identifying a common angiogenic growth factor (for example, VEGF) in these angiogenic-dependent diseases, examining TF expression on in vitro models of angiogenic and quiescent VECs and then testing the selectivity and effectiveness of TF-targeted agents (ICON and fVII-PS) for binding and killing the in vitro angiogenic VECs.

Vascular endothelial growth factor (VEGF) plays a central role in angiogenesis-dependent cancer and non-malignant human diseases [24], such as macular degeneration [25], rheumatoid arthritis [26] and endometriosis [27]. Specifically, VEGF stimulates angiogenesis by binding to VEGR receptors on VECs in the pathological neovasculature (usually micro- or capillary vessels) in those angiogenesis-dependent diseases. In humans, there are three major types of blood vessels, namely micro- or capillary vessels, veins and arteries. Accordingly, VEC, the inner layer of blood vessels, can be isolated from microvessels, venous and arterial vessels. It is previously known that VEGF can induce TF expression on human umbilical vein endothelial cells (HUVEC), a commonly used VEC model in angiogenesis studies. To better mimic pathological angiogenesis, an ideal angiogenic VEC model should be derived from micro- or capillary vessels. This study is designed to investigate whether angiogenic VECs from micro- or capillary vessels express TF upon VEGF stimulation and whether fVII-dependent therapies target non-pathological (quiescent) endothelial cells using VEGF-stimulated and unstimulated VECs as in vitro models. Therefore, in the current study, we examined the expression patterns of TF on in vitro models of angiogenic and quiescent VECs from three types of blood vessels and sought to rigorously assess the validity of the claim that dimeric fVII-IgG1Fc (ICON) in immunotherapy and monomer fVII peptide in PDT are selective and effective in killing angiogenic VECs via TF targeting.

To test for expression of angiogenic surface receptors, we used VEGF to generate angiogenic VEC models in vitro, and tested for the presence of TF on primary human VECs derived from three major types of mammalian vessels: (1) HMVEC, human microvascular endothelial cells; (2) HUVEC, human umbilical venous endothelial cells; and (3) HAEC, human aortic endothelial cells.

We first examined the time course of TF expression on angiogenic and quiescent HMVEC (Fig. 1a), HUVEC (Fig. 1b) and HAEC (Fig. 1c). The Western blotting results in Fig. 1 showed that (1) TF protein, reported MW 47 kDa, shown in Fig. 1 as a band at ~50 kDa, was not detected in quiescent VEC (starved and unstimulated) (time point = 0 h, red arrows in TF expression fold chart) from all three VECs; (2) TF was detected in VEGF-treated angiogenic VECs (HMVEC, HUVEC and HAEC) within 2 h of VEGF stimulation; (3) and peaked at 4–6 h, then started decreasing at 6 h post-stimulation.

To confirm the time course of induced TF expression on endothelial cells, we stained VEC for TF and the endothelial marker CD31 and observed the stained VEC under confocal microscope. The results in Fig. 1d showed that TF was not detected on quiescent VEC (0 h). After VEGF stimulation, however, TF expression was detected at 2 h with a peak in fluorescent intensity at 4–6 h followed by a decrease that was still visible at 24 h.

Next, we examined the binding (specificity) of fVII therapeutic agents to angiogenic VECs. To generate angiogenic VEC with maximal TF expression, VECs were incubated with 1 nM VEGF for 4 h based on the results in Fig. 1. Confocal imaging confirmed that TF was expressed on angiogenic HMVEC (Fig. 2a), HUVEC (Fig. 2c) and HAEC (Fig. 2e) after 4 h stimulation with VEGF, but not on resting VECs (Fig. 2b, d, f). Taken the results in Figs. 1 and 2, we conclude that TF is selectively expressed on angiogenic VEC, which can be induced on VECs from all three types of blood vessels (vein, microvessels and artery).

To assess the binding specificity of murine fVII/human IgG1 Fc (mouse ICON) and murine fVII/Sp (simplified as mfVII in Fig. 2) to angiogenic VECs, 4 h VEGF-stimulated VECs (Fig. 2g, i, k) and unstimulated VECs (Fig. 2h, i, j) were analyzed by cell ELISA. To ensure that cell membrane-bound mfVII/hIgG1 Fc was intact, we separately used anti-mfVII and antihuman IgG Fc antibody horseradish peroxidase conjugates for detection of mfVII/hIgG1 Fc in the cell ELISA. The cell ELISA results in Fig. 2g, h, i are summarized as follows. (1) A mouse monoclonal anti-TF antibody (clone HTF1) demonstrated that angiogenic VECs expressed TF (HMVEC in Fig. 2g, HUVEC in Fig. 2i and HAEC in Fig. 2k), whereas quiescent VECs do not express TF (Fig. 2h, j, l). The TF results by cell ELISA are consistent with the Western blotting results (Fig. 1) and confocal imaging results (Fig. 2a, b, c, d, e, f). (2) Similar to observations made with the anti-TF antibody, fVII agents, either in dimeric (mfVII/hIgG1 Fc) or monomeric form (mfVII), selectively bound all three angiogenic VEC types (Fig. 2g, i, k, respectively). In contrast, fVII agents did not bind to any of the quiescent VECs (Fig. 2h, 2j, l). Further, results in Fig. 2f, g suggested that mfVII/hIgG1 Fc was intact when binding to angiogenic HMVEC and HUVEC since there was no difference between anti-mfVII and anti-hIgG antibodies for detection of mfVII/hIgG1 Fc in cell ELISA. These findings established that (1) TF is selectively expressed on angiogenic VECs; (2) fVII agents selectively bind angiogenic VECs, likely reflecting their binding to TF (See Figs. 1 and 2).

To confirm that selective binding of fVII agents to angiogenic VECs is mediated by TF, we carried out experiments that directly evaluated the ability of anti-TF antibodies to inhibit fVII binding to TF on angiogenic VECs. Two antihuman TF antibodies, mouse monoclonal antibody HTF1 and a goat polyclonal antibody (goat anti-HTF) (American Diagnostica), which were known to be able to block TF coagulation [29, 33], were compared for their relative inhibition of fVII binding to TF. Initial studies established that both anti-TF antibodies (p > 0.05) similarly bind to a breast cancer line (MDA-MB-231) that expresses high levels of TF [16] (Fig. 3a). Subsequently, goat anti-HTF was found to display a significantly stronger effect than HTF1 on inhibiting the binding of mfVII/hIgG1 Fc to MDA-MB-231 cancer cells (p < 0.001) (Fig. 3b). These observations suggested using goat anti-HTF as an inhibitory antibody for the experiments described in Fig. 3c, d, e.

To determine whether fVII is a ligand for TF on in vitro angiogenic VEC models, we developed an inhibitory cell ELISA assay in which angiogenic VEC were first incubated with various nanomolar concentrations of inhibitory goat anti-HTF for blocking TF or anti-EPCR (RCR-379) for blocking EPCR [30] as a control antibody, followed by incubation with 10 nM mfVII/hIgG1 Fc (to generate molar ratios 0.01:1–10:1, inhibitory antibody to mfVII/hIgG Fc). The binding of fVII/IgG1Fc was detected with anti-hIgG HRP conjugate. We found that goat anti-HTF completely (at ratios of 1:1 and 10:1) or partially (at ratios of 0.01:1 and 0.1:1) inhibit the binding of mfVII/hIgG1 Fc to all three angiogenic VECs (See Fig. 3c, d, e; HMVEC in Fig. 3c, HUVEC in Fig. 3d and HAEC in Fig. 3e). By contrast, control antibody did not inhibit fVII binding to angiogenic VECs, demonstrating that on angiogenic VECs, fVII acts via binding to TF. To further verify the specificity of fVII agents to TF versus the endothelial protein C receptor (EPCR) [34], we carried out immunoprecipitation–Western blotting (IP–WB) analysis using stable CHO-K1 cell lines expressing TF, EPCR, TF and EPCR and the control was untransfected CHO-K1 cells [35]. The results showed that mouse and human ICON could pull down TF efficiently in the presence and absence of EPCR (Fig. 3f), demonstrating that fVII agents (fVII/IgG1 Fc, ICON) specifically bind to TF.

To determine whether fVII agents are selective and effective in killing angiogenic VECs, we tested fVII-tPDT using photosensitizer SnCe6-conjugated monomeric mfVII peptides. We first confirmed that like goat anti-TF antibody (Fig. 4a), PS-conjugated mfVII/Sp (Fig. 4b) and mfVII/NLS (Fig. 4c) selectively binds to angiogenic HMVECs. Then, we tested the effectiveness and selectivity of fVII-tPDT for in vitro killing of angiogenic HMVEC using the following assays:

To better understand its mechanism of action of fVII-tPDT, we first treated angiogenic and resting HMVEC with fVII-tPDT (mfVII/NLS-SnCe6) or ntPDT (SnCe6) then immediately tested the VEC for apoptosis using Annexin V-FITC and for necrosis using propidium iodide (PI). We found that fVII-tPDT treatment induced immediate apoptosis and necrosis in angiogenic HMVECs (see Fig. 4g), similar to its mechanism in killing human lung cancer cells [6]. Importantly, fVII-tPDT did not induce detectable apoptosis and necrosis in quiescent HMVEC, whereas ntPDT had no effect in angiogenic and quiescent HMVEC (Fig. 4g).

Identification of specific anti-angiogenic target molecules in endothelial is crucial in developing anti-angiogenesis therapies. Toward this end, a major effort has focused on targeting VEGF and VEGFRs. Elevated levels of VEGFRs are detected on tumor VECs in many types of solid cancers [36]. Unfortunately, VEGFRs are not angiogenic-specific endothelial receptors since they are also expressed on normal endothelium in many organs [37, 38]. Nevertheless, this study defines the peak expression of TF on angiogenic VECs (4 h VEGF induction), which may be used as an optimized time point for generating in vitro VEC models for identification of more novel target molecules by revealing the differences in gene expression (for example, mRNAs and micro-RNAs) between angiogenic and quiescent VECs.

In this report, we show that, unlike VEGFRs, TF is a specific receptor for angiogenic VECs in vitro. It appears that TF is also a unique pathological angiogenic endothelial cell surface receptor in vivo because of its selective expression on angiogenic VECs in vivo in tumor vasculature [4‐7, 9, 14], ocular [18] and endometriotic [21] neovasculature from patients or animal models. In addition to TF on angiogenic VEC, TF is overexpressed on many cancer cells including solid cancer and leukemia [39]. Therefore, TF appears to be a common yet specific biomarker and target molecule for both the pathological neovasculature and cancer cells [6, 7, 16, 39].

Consequently, the use of fVII as targeting vehicle enabled our group to develop two TF-targeted therapeutics, namely fVII/IgG1 Fc (ICON) for immunotherapy and fVII/PS for tPDT, which display selectivity, effectiveness and safety in the treatment of lung and breast cancer (regardless of their drug resistance and other membrane targets) [5‐7, 13‐17, 23], macular degeneration [18‐20] and/or endometriosis [21] in preclinical mouse, rat and pig models and in phase I and II clinical trials for AMD patients (NCT01485588) [22]. In this study, we showed that both mfVII/IgG1 Fc and hfVII/IgG1 Fc bind to human TF, which is only expressed by angiogenic VECs upon VEGF stimulation. Thus, TF-targeted ICON and fVII-tPDT should not cause side effects to quiescent vessels, including microvessels, venous and arterial vessels, while they can selectively target and effectively eradicate angiogenic VECs via TF targeting. Nevertheless, caution should be taken as VEGF and other chemokines (IL-1β, TNF-α, etc.) may potentially induce TF expression on normal resting VECs. They should be monitored before and during the treatment of fVII agents, particularly when fVII agents are administered systemically. In the case of cancer, Drs. Yang and Rosenberg et al. [40] reported that 76 of 113 patients with metastatic renal cell carcinoma had a baseline VEGF level below the lower limit of detection (40 pg/ml). After anti-VEGF antibody bevacizumab (Avastin) therapy, the plasma VEGF levels increased to a range of 150–250 pg/ml. Similarly, a very recent study [41] reported that the plasma baseline levels of VEGF in 60 patients (Hongkong) with colorectal cancer metastases were about 10 pg/ml, and they were increased to 40 pg/ml after anti-VEGF antibody bevacizumab (Avastin) therapy [41]. Importantly, the group found that increased VEGF was in a complex with the neutralizing antibody that prevents VEGF from binding to VEGF receptors. In the cases of AMD and PCV (polypoidal choroidal vasculopathy), Tong et al. [42] reported that the aqueous humor level of VEGF was about 400 pg/ml in eyes with PCV and about 700 pg/ml in eyes with choroidal neovascularization of AMD. A Japanese group also reported aqueous humor level of VEGF was higher in AMD and PCV than control eyes. But they did not find the significant difference of VEGF levels between AMD and PCV. Nevertheless, the levels of VEGF reported in those published studies were at 100-fold lower than the concentration (1 nM VEGF = 38 ng/ml, MW 38 kDa) that we tested in this study. It remains to investigate the lowest VEGF concentration that is able to induce endothelial TF expression. If the blood VEGF level in patients is detected high (enough to induce TF expression on normal endothelial cells), anti-VEGF inhibitors may be administered to neutralize and clear VEGF from circulation, if any, prior to fVII-targeted therapies. Regardless of the VEGF levels in AMD and PCV, fVII-tPDT may be used for the treatment of these ocular diseases as long as TF expression is detected in the neovascularization lesions.

In conclusion, this study identifies that TF is an angiogenic-specific endothelial biomarker and verifies that TF is the therapeutic target specific for fVII agents. This study also elucidated for the first time that the specificity of fVII agents for angiogenic VEC is mediated via binding TF. Thus, this study not only identifies TF as an angiogenic endothelial receptor specific for fVII agents, but also helps predict therapeutic efficacy and potential side effects of these fVII-targeted agents. As such, supports use of TF-targeted therapeutics in clinical trials of several angiogenesis-dependent common human diseases, notably cancer, macular degeneration, endometriosis and rheumatoid arthritis, in which VEGF plays a central role in pathological angiogenesis.