Antibodies that bind to endoglin

Disclosed are antibodies that specifically bind to endoglin. Conjugates of the antibodies linked to diagnostic or therapeutic agents are also provided. Methods of using the antibodies and conjugates are also disclosed, including methods of targeting the vasculature of solid tumors through recognition of the tumor vasculature-associated antigen, endoglin.

 

What is claimed is:

1. A purified antibody that binds to the same epitope as the monoclonal antibody TEC-4 (ATCC xxxxx).

2. The purified antibody of claim 1, which is a monoclonal antibody.

3. The purified antibody of claim 2, which is monoclonal antibody TEC-4 (ATCC xxxxx).

4. The purified antibody of claim 1, in a pharmaceutically acceptable diluent or excipient.

5. The purified antibody of claim 1, wherein the antibody is linked to an anticellular agent capable of killing or suppressing the growth or cell division of endothelial cells.

6. The purified antibody of claim 5, which is a monoclonal antibody.

7. The purified antibody of claim 6, which is monoclonal antibody TEC-4 (ATCC xxxxx).

8. The purified antibody of claim 5, wherein the antibody is linked to a plant-, fungus- or bacteria-derived toxin.

9. The purified antibody of claim 8, wherein the antibody is linked to an A chain toxin, a ribosome inactivating protein, .alpha.-sarcin, aspergillin, restrictocin, a ribonuclease, diphtheria toxin or Pseudomonas exotoxin.

10. The purified antibody of claim 9, wherein the antibody is linked to an A chain toxin.

11. The purified antibody of claim 10, wherein the antibody is linked to ricin A chain.

12. The purified antibody of claim 11, wherein the antibody is linked to deglycosylated ricin A chain.

13. The purified antibody of claim 5, wherein the antibody is linked to a chemotherapeutic agent, a radioisotope or a cytotoxin.

14. The purified antibody of claim 13, wherein the antibody is linked to asteroid, an antimetabolite, an anthracycline, a vinca alkaloid, an antibiotic, an alkylating agent or an epipodophyllotoxin.

15. The purified antibody of claim 5, in a pharmaceutically acceptable diluent or excipient.

16. A purified antibody that binds to the same epitope as the monoclonal antibody TEC-11 (ATCC yyyyy).

17. The purified antibody of claim 16, which is a monoclonal antibody.

18. The purified antibody of claim 17, which is monoclonal antibody TEC-11 (ATCC yyyyy).

19. The purified antibody of claim 16, in a pharmaceutically acceptable diluent or excipient.

20. The purified antibody of claim 16, wherein the antibody is linked to an anticellular agent capable of killing or suppressing the growth or cell division of endothelial cells.

21. The purified antibody of claim 20, which is a monoclonal antibody.

22. The purified antibody of claim 21, which is monoclonal antibody TEC-11 (ATCC yyyyy).

23. The purified antibody of claim 20, wherein the antibody is linked to a plant-, fungus- or bacteria-derived toxin.

24. The purified antibody of claim 23, wherein the antibody is linked to an A chain toxin, a ribosome inactivating protein, .alpha.-sarcin, aspergillin, restrictocin, a ribonuclease, diphtheria toxin or Pseudomonas exotoxin.

25. The purified antibody of claim 24, wherein the antibody is linked to an A chain toxin.

26. The purified antibody of claim 25, wherein the antibody is linked to ricin A chain.

27. The purified antibody of claim 26, wherein the antibody is linked to deglycosylated ricin A chain.

28. The purified antibody of claim 20, wherein the antibody is linked to a chemotherapeutic agent, a radioisotope or a cytotoxin.

29. The purified antibody of claim 28, wherein the antibody is linked to asteroid, an antimetabolite, an anthracycline, a vinca alkaloid, an antibiotic, an alkylating agent or an epipodophyllotoxin.

30. The purified antibody of claim 20, in a pharmaceutically acceptable diluent or excipient.

BACKGROUND OF THE INVENTION

1. Field of the Invention

The present invention relates generally to methods and compositions for targeting the vasculature of solid tumors using immunological and growth factor-based reagents. In particular aspects, antibodies carrying diagnostic or therapeutic agents are targeted to the vasculature of solid tumor masses through recognition of tumor vasculature-associated antigens, such as endoglin, or through the specific induction of other antigens on vascular endothelial cells in solid tumors.

2. Description of Related Art

Over the past 30 years, fundamental advances in the chemotherapy of neoplastic disease have been realized. While some progress has been made in the development of new chemotherapeutic agents, the more startling achievements have been made in the development of effective regimens for concurrent administration of drugs, and our knowledge of the basic science, e.g., the underlying neoplastic processes at the cellular and tissue level, and the mechanism of action of basic antineoplastic agents. As a result of the fundamental achievement, we can point to significant advances in the chemotherapy of a number of neoplastic diseases, including choriocarcinoma, Wilm's tumor, acute leukemia, rhabdomyosarcoma, retinoblastoma, Hodgkin's disease and Burkitt's lymphoma, to name just a few. Despite the impressive advances that have been made in a few tumors, though, many of the most prevalent forms of human cancer still resist effective chemotherapeutic intervention.

The most significant underlying problem that must be addressed in any treatment regimen is the concept of "total cell kill." This concept holds that in order to have an effective treatment regimen, whether it be a surgical or chemotherapeutic approach or both, there must be a total cell kill of all so-called "clonogenic" malignant cells, that is, cells that have the ability to grow uncontrolled and replace any tumor mass that might be removed. Due to the ultimate need to develop therapeutic agents and regimens that will achieve a total cell kill, certain types of tumors have been more amenable than others to therapy. For example, the soft tissue tumors (e.g., lymphomas), and tumors of the blood and blood-forming organs (e.g., leukemias) have generally been more responsive to chemotherapeutic therapy than have solid tumors such as carcinomas. One reason for this is the greater physical accessibility of lymphoma and leukemic cells to chemotherapeutic intervention. Simply put, it is much more difficult for most chemotherapeutic agents to reach all of the cells of a solid tumor mass than it is the soft tumors and blood-based tumors, and therefore much more difficult to achieve a total cell kill. The toxicities associated with most conventional antitumor agents then become a limiting factor.

A key to the development of successful antitumor agents is the ability to design agents that will selectively kill tumor cells, while exerting relatively little, if any, untoward effects against normal tissues. This goal has been elusive to achieve, though, in that there are few qualitative differences between neoplastic and normal tissues. Because of this, much research over the years has focused on identifying tumor-specific "marker antigens" that can serve as immunological targets both for chemotherapy and diagnosis. Many tumor-specific, or quasi-tumor-specific ("tumor-associated"), markers have been identified as tumor cell antigens that can be recognized by specific antibodies. Unfortunately, it is generally the case that tumor specific antibodies will not in and of themselves exert sufficient antitumor effects to make them useful in cancer therapy.

Over the past fifteen years, immunotoxins have shown great promise as a means of selectively targeting cancer cells. Immunotoxins are conjugates of a specific targeting agent typically a tumor-directed antibody or fragment, with a cytotoxic agent, such as a toxin moiety. The targeting agent directs the toxin to, and thereby selectively kills, cells carrying the targeted antigen. Although early immunotoxins suffered from a variety of drawbacks, more recently, stable, long-lived immunotoxins have been developed for the treatment of a variety of malignant diseases. These "second generation" immunotoxins employ deglycosylated ricin A chain to prevent entrapment of the immunotoxin by the liver and hepatotoxicity (Blakey et al., 1987). They employ new crosslinkers which endow the immunotoxins with high in vivo stability (Thorpe et al., 1988) and they employ antibodies which have been selected using a rapid indirect screening assay for their ability to form highly potent immunotoxins (Till et al., 1988).

Immunotoxins have proven highly effective at treating lymphomas and leukemias in mice (Thorpe et al., 1988; Ghetie et al., 1991; Griffin et al., 1988) and in man (Vitetta et al., 1991). Lymphoid neoplasias are particularly amenable to immunotoxin therapy because the tumor cells are relatively accessible to blood-borne immunotoxins; also, it is possible to target normal lymphoid antigens because the normal lymphocytes which are killed along with the malignant cells during therapy are rapidly regenerated from progenitors lacking the target antigens. In Phase I trials where patients had large bulky tumor masses, greater than 50% tumor regressions were achieved in approximately 40% of the patients (Vitetta et al., 1991). It is predicted that the efficacy of these immunotoxins in patients with less bulky disease will be even better.

In contrast with their efficacy in lymphomas, immunotoxins have proved relatively ineffective in the treatment of solid tumors such as carcinomas (Weiner et al., 1989; Byers et al., 1989). The principal reason for this is that solid tumors are generally impermeable to antibody-sized molecules: specific uptake values of less than 0.001% of the injected dose/g of tumor are not uncommon in human studies (Sands et al., 1988; Epenetos et al., 1986). Furthermore, antibodies that enter the tumor mass do not distribute evenly for several reasons. Firstly, the dense packing of tumor cells and fibrous tumor stromas present a formidable physical barrier to macromolecular transport and, combined with the absence of lymphatic drainage, create an elevated interstitial pressure in the tumor core which reduces extravasation and fluid convection (Baxter et al., 1991; Jain, 1990). Secondly, the distribution of blood vessels in most tumors is disorganized and heterogeneous, so some tumor cells are separated from extravasating antibody by large diffusion distances (Jain, 1990). Thirdly, all of the antibody entering the tumor may become adsorbed in perivascular regions by the first tumor cells encountered, leaving none to reach tumor cells at more distant sites (Baxter et al., 1991; Kennel et al., 1991). Finally, antigen-deficient mutants can escape being killed by the immunotoxin and regrow (Thorpe et al., 1988).

Thus, it is quite clear that a significant need exists for the development of novel strategies for the treatment of solid tumors. One approach would be to target cytotoxic agents or coagulants to the vasculature of the tumor rather than to the tumor. Indeed, it has been observed that many existing therapies may already have, as part of their action, a vascular-mediated mechanism of action (Denekamp, 1990). The present inventors propose that this approach offers several advantages over direct targeting of tumor cells. Firstly, the target cells are directly accessible to intravenously administered therapeutic agents, permitting rapid localization of a high percentage of the injected dose (Kennel et al., 1991). Secondly, since each capillary provides oxygen and nutrients for thousands of cells in its surrounding `cord` of tumor, even limited damage to the tumor vasculature could produce an avalanche of tumor cell death (Denekamp, 1990; Denekamp, 1984). Finally, the outgrowth of mutant endothelial cells lacking the target antigen is unlikely because they are normal cells.

For tumor vascular targeting to succeed, antibodies are required that recognize tumor endothelial cells but not those in normal tissues. Although several antibodies have been raised (Duijvestijn et al., 1987; Hagemeier et al., 1986; Bruland et al., 1986; Murray et al., 1989; Schlingemann et al., 1985) none has shown a high degree of specificity.

The antibodies termed TP-1 and TP-3, which were raised against human osteosarcoma cells, have been reported to react with the same antigen present on proliferating osteoblasts in normal degenerating bone tissue. They also cross-react with capillary buds in a number of tumor types and in placenta, but apparently not with capillaries in any of the normal adult tissues examined (Bruland et al., 1986). It remains to be seen whether the TP-1/TP-3 antigen is present on the surface of endothelial cells or whether the antibodies cross-react with gut endothelial cells, as was found with another antibody against proliferating endothelium (Hagemeier et al., 1986). This antibody described by Hagemeier and colleagues (1986), termed EN7/44, reacts with a predominantly intracellular antigen whose expression appears to be linked to migration rather than proliferation (Hagemeier et al., 1986).

Immunotoxins in which the antibody portion is directed against the fibronectin receptor have also been proposed for use in killing proliferating vascular endothelial cells (Thorpe et aI., 1990). However, intravenous administration of an immunotoxin containing dgA linked to the anti-fibronectin receptor antibody termed PB1 did not result in reduced vascularization of tumors (Thorpe et al., 1990). Unfortunately, further studies also revealed that fibronectin receptors were too ubiquitous to enable good targeting of tumor vasculature.

Other molecular markers have been described that are specific for endothelial cells, although not for tumor endothelial cells. For example, an endothelial-leukocyte adhesion molecule, termed ELAM-1, has been identified that can be induced on the surface of endothelial cells through the action of cytokines such as IL-1, TNF, lymphotoxin or bacterial endotoxin (Bevilacqua et al., 1987). However, the art currently lacks methods by which such inducible molecules could be effectively employed in connection with an anti-cancer strategy. Thus, unfortunately, while vascular targeting presents promising theoretical advantages, no effective strategies incorporating these advantages have been developed.

SUMMARY OF THE INVENTION

The present invention addresses one or more of the foregoing or other disadvantages in the prior art, by providing a series of novel approaches for the treatment and/or diagnosis (imaging) of vascularized solid tumors. The invention rests in a general and overall sense on the use of reagents, particularly immunological reagents, to target therapeutic or diagnostic agents to tumor-associated vascular endothelial cells, alone or in combination with the direct targeting of tumor cells.

Such antibodies or growth factors will be referred to herein as "targeting agents". Thus, the targeting compounds of the invention may be either targeting agent/therapeutic agent compounds or targeting agent/diagnostic agent compounds. Further, a targeting agent/therapeutic agent compound comprises a targeting agent operatively attached to a therapeutic agent, wherein the targeting agent recognizes and binds to a tumor-associated endothelial cell marker. A targeting agent/diagnostic agent compound comprises a targeting agent operatively attached to a diagnostic agent, wherein the targeting agent recognizes and binds to a tumor-associated endothelial cell marker. The targeting agent/therapeutic agent compounds of the invention may be produced using either standard recombinant DNA techniques or standard synthetic chemistry techniques, both of which are well known to those of skill in the art.

In the case of diagnostic agents, the constructs will have the ability to provide an image of the tumor vasculature, for example, through magnetic resonance imaging, x-ray imaging, computerized emission tomography and the like.

In the case of therapeutic agents, constructs are designed to have a cytotoxic or otherwise anticellular effect against the tumor vasculature, by suppressing the growth or cell division of the vascular endothelial cells. This attack is intended to lead to a tumor-localized vascular collapse, depriving the tumor cells, particularly those tumor cells distal of the vasculature, of oxygen and nutrients, ultimately leading to cell death and tumor necrosis. In animal model systems, the inventors have achieved truly dramatic tumor regressions, with some cures being observed in combination therapy with anti-tumor directed therapy.

It is proposed that the various methods and compositions of the invention will be broadly applicable to the treatment or diagnosis of any tumor mass having a vascular endothelial component. Typical vascularized tumors are the solid tumors, particularly carcinomas, which require a vascular component for the provision of oxygen and nutrients. Exemplary solid tumors to which the present invention is directed include but are not limited to carcinomas of the lung, breast, ovary, stomach, pancreas, larynx, esophagus, testes, liver, parotid, biliary tract, colon, rectum, cervix, uterus, endometrium, kidney, bladder, prostate, thyroid, squamous cell carcinomas, adenocarcinomas, small cell carcinomas, melanomas, gliomas, neuroblastomas, and the like.

A preferred method of the invention includes preparing an antibody that recognizes an antigen or other ligand associated with the vascular endothelial cells of the vascularized tumor mass, linking, or operatively attaching the antibody to the selected agent to form an antibody-agent conjugate, and introducing the antibody-agent conjugate into the bloodstream of an animal, such as a human cancer patient or a test animal in an animal model system. As used however, the term "antibody" is intended to refer broadly to any immunologic binding agent such as IgG, IgM, IgA, IgE, F(ab').sub.2, a univalent fragment such as Fab', Fab, Dab, as well as engineered antibodies such as recombinant antibodies, humanized antibodies, bispecific antibodies, and the like.

Alternatively, growth factors, rather than antibodies, may be utilized as the reagents to target therapeutic or diagnostic agents to tumor-associated vascular endothelial cells, alone, or in combination with the direct targeting of tumor cells. Any growth factor may be used for such a targeting purpose, so long as it binds to a tumor-associated endothelial cell, generally by binding to a growth factor receptor present on the surface of such a tumor-associated endothelial cell. Suitable growth factors for targeting include, but are not limited to, VEGF/VPF (vascular endothelial growth factor/vascular permeability factor), FGF (which, as used herein, refers to the fibroblast growth factor family of proteins), TFG.beta. (transforming growth factor .beta.), and pleitotropin. Preferably, the growth factor receptor to which the targeting growth factor binds should be present at a higher concentration on the surface of tumor-associated endothelial cells than on non-tumor associated endothelial cells. Most preferably, the growth factor receptor to which the targeting growth factor binds should, further, be present at a higher concentration on the surface of tumor-associated endothelial cells than on any non-tumor associated cell type.

The agent that is linked to the antibody or growth factor targeting agent will, of course, depend on the ultimate application of the invention. Where the aim is to provide an image of the tumor, one will desire to use a diagnostic agent that is detectable upon imaging, such as a paramagnetic, radioactive or fluorogenic agent. Many diagnostic agents are known in the art to be useful for imaging purposes, as are methods for their attachment to antibodies (see, e.g., U.S. Pat. Nos. 5,021,236 and 4,472,509, both incorporated herein by reference). In the case of paramagnetic ions, one might mention by way of example ions such as chromium (III), manganese (II), iron (III), iron (II), cobalt (II), nickel (II), copper (II), neodymium (III), samarium (III), ytterbium (III), gadolinium (III), vanadium (II), terbium (III), dysprosium (III), holmium (III) and erbium (III), with gadolinium being particularly preferred. Ions useful in other contexts, such as X-ray imaging, include but are not limited to lanthanum (III), gold (III), lead (II), and especially bismuth (III). Moreover, in the case of radioactive isotopes for therapeutic and/or diagnostic application, one might mention iodine.sup.131, iodine.sup.123, technicium.sup.99m, indium.sup.111, rhenium.sup.188, rhenium.sup.186, gallium.sup.67, copper.sup.67, yttrium.sup.90, iodine.sup.125, or astatine.sup.211 (for a review of the use of monoclonal antibodies in the diagnosis and therapy of cancer, see Vaickus et al., 1991).

For certain applications, it is envisioned that the therapeutic agents will be pharmacologic agents will serve as useful agents for attachment to antibodies or growth factors, particularly cytotoxic or otherwise anticellular agents having the ability to kill or suppress the growth or cell division of endothelial cells. In general, the invention contemplates the use of any pharmacologic agent that can be conjugated to a targeting agent, preferably an antibody, and delivered in active form to the targeted endothelium. Exemplary anticellular agents include chemotherapeutic agents, radioisotopes as well as cytotoxins. In the case of chemotherapeutic agents, the inventors propose that agents such as a hormone such as a steroid; an antimetabolite such as cytosine arabinoside, fluorouracil, methotrexate or aminopterin; an anthracycline; mitomycin C; a vinca alkaloid; demecolcine; etoposide; mithramycin; or an antitumor alkylating agent such as chlorambucil or melphalan, will be particularly preferred. Other embodiments may include agents such as a coagulant, a cytokine, growth factor, bacterial endotoxin or the lipid A moiety of bacterial endotoxin. In any event, it is proposed that agents such as these may, if desired, be successfully conjugated to a targeting agent, preferably an antibody, in a manner that will allow their targeting, internalization, release or presentation to blood components at the site of the targeted endothelial cells as required using known conjugation technology (see, e.g., Ghose et al., 1983 and Ghose et al., 1987).

In certain preferred embodiments, therapeutic agents will include generally a plant-, fungus- or bacteria-derived toxin, such as an A chain toxins, a ribosome inactivating protein, .alpha.-sarcin, aspergillin, restrictocin, a ribonuclease, diphtheria toxin or pseudomonas exotoxin, to mention just a few examples. The use of toxin-antibody constructs is well known in the art of immunotoxins, as is their attachment to antibodies. Of these, a particularly preferred toxin for attachment to antibodies will be a deglycosylated ricin A chain. Deglycosylated ricin A chain is preferred because of its extreme potency, longer half-life, and because it is economically feasible to manufacture it a clinical grade and scale.

The present invention contemplates two separate and distinct approaches to the targeting of targeting agents, preferably antibodies, to the tumor vasculature. The first approach involves a targeting agent having a binding affinity for a marker found, expressed, accessible to binding, or otherwise localized on the cell surfaces of tumor-associated vascular endothelial cells as compared to normal, non-tumor associated vasculature. Further, certain markers for which a targeting agent has a binding affinity may be associated with the tumor-associated vasculature rather than on the tumor-associated endothelial cells, themselves. For example, such markers may be located on basement membranes or tumor-associated connective tissue. It is preferred that targeting agents specific for such non-endothelial cell markers be operatively attached to agents such as radioisotopes.

In the case of antibody targeting agents, such an approach involves the preparation of an antibody having a binding affinity for antigenic markers found, expressed, accessible to binding or otherwise localized on the cell surfaces of tumor-associated vascular endothelium as compared to normal vasculature. Such a targeting agent, preferably an antibody, is then employed to deliver the selected diagnostic or therapeutic agent to the tumor vasculature.

Naturally, where a therapeutic as opposed to diagnostic application is envisioned it will be desirable to prepare and employ an antibody having a relatively high degree of tumor vasculature selectivity, which might be expressed as having little or no reactivity with the cell surface of normal endothelial cells as assessed by immunostaining of tissue sections. Of course, with certain agents such as DNA synthesis inhibitors, and, more preferably, antimetabolites, the requirement for selectivity is not as necessary as it would be, for example, with a toxin, because a DNA synthesis inhibitor would have relatively little effect on the vascularation of normal tissues because the capillary endothelial cells are not dividing. Further, such a degree of selectivity is not a requirement for imaging purposes since cell death, and hence toxicity, is not the ultimate goal. In the case of diagnostic application, it is proposed that targeting agents, such as antibodies, having a reactivity for the tumor vasculature of at least two-fold higher than for normal endothelial cells, as assessed by immunostaining, will be useful.

This aspect of the invention rests on the proposition that because of their proximity to the tumor itself, tumor-associated vascular endothelial cells are constantly exposed to many tumor-derived products such as cytokines (including lymphokines, monokines, colony-stimulating factors and growth factors), angiogenic factors, and the like, that will bind to and serve to selectively elicit the expression of tumor endothelium-specific cell surface markers. For use in the present invention, the anti-tumor vasculature antibodies may be directed to any of the tumor-derived antigens which bind to the surface of vascular endothelial cells, and particularly to tumor-derived ligands, such as growth factors, which bind to specific cell surface receptors of the endothelial cells.

In connection with certain aspects of the invention, antibodies directed against tumor vasculature may be prepared by using endothelial cells isolated from a tumor of an animal, or by "mimicking" the tumor vasculature phenomenon in vitro. As such, endothelial cells may be subjected to tumor-derived products, such as might be obtained from tumor-conditioned media. Thus, this method involves generally stimulating endothelial cells with tumor-conditioned medium and employing the stimulated endothelial cells as immunogens to prepare a collection of antibodies, for example, by utilizing conventional hybridoma technology or other techniques, such as combinatorial immunoglobulin phagemid libraries prepared from RNA isolated from the spleen of the immunized animal. One will then select from the antibody collection an antibody that recognizes the tumor-stimulated vascular endothelium to a greater degree than it recognizes non-tumor-stimulated vascular endothelium, and reacts more strongly with tumor-associated endothelial cells in tissue sections than with those in normal adult human tissues, and producing the antibody, e.g., by culturing a hybridoma to provide the antibody.

Stimulated endothelial cells contemplated to be of use in this regard include, for example, human umbilical vein endothelial cells (RIIVE), human dermal microvascular endothelial cells (HDEMC), human saphenous vein endothelial cells, human omental fat endothelial cells, other human microvascular endothelial cells, human brain capillary endothelial cells, and the like. It is also contemplated that even endothelial cells from another species may stimulated by tumor-conditioned media and employed as immunogens to generate hybridomas to produce an antibodies in accordance herewith, i.e., to produce antibodies which cross react with tumor-stimulated human vascular endothelial cells, and/or antibodies for use in pre-clinical models.

As used herein, "tumor-conditioned medium" is defined as a composition or medium, such as a culture medium, which contains one or more tumor-derived cytokines, lymphokines or other effector molecules. Most typically, tumor-conditioned medium is prepared from a culture medium in which selected tumor cells have been grown, and will therefore be enriched in such tumor-derived products. The type of medium is not believed to be particularly important, so long as it at least initially contains appropriate nutrients and conditions to support tumor cell growth. It is also, of course, possible to extract and even separate materials from tumor-conditioned media and employ one or more of the extracted products for application to the endothelial cells.

As for the type of tumor used for the preparation of the media, one will, of course, prefer to employ tumors that mimic or resemble the tumor that will ultimately be subject to analysis or treatment using the present invention. Thus, for example, where one envisions the development of a protocol for the treatment of breast cancer, one will desire to employ breast cancer cells such as ZR-75-1, T47D, SKBR3, MDA-MB-231. In the case of colorectal tumors, one may mention by way of example the HT29 carcinoma, as well as DLD-1, MCT116 or even SW48 or SW122. In the case of lung tumors, one may mention byway of example NCI-H69, SW2, NCI H23, NCI M460, NCI M69, or NCI H82. In the case of melanoma, good examples are DX.3, A375, SKMEL-23, HMB-2, MJM, T8 or indeed VUP. In any of the above cases, it is further believed that one may even employ cells produced from the tumor that is to be treated, i.e., cells obtained from a biopsy.

Once prepared, the tumor-conditioned media is then employed to stimulate the appearance of tumor endothelium-specific marker(s) on the cell surfaces of endothelial cells, e.g., by culturing selected endothelial cells in the presence of the tumor-conditioned media (or products derived therefrom). Again, it is proposed that the type of endothelial cell that is employed is not of critical importance, so long as it is generally representative of the endothelium associated with the vasculature of the particular tumor that is ultimately to be treated or diagnosed. The inventors prefer to employ human umbilical vein endothelial cells (EUVE), or human dermal microvascular endothelial cells (HDMEC, Karasek, 1989), in that these cells are of human origin, respond to cytokine growth factors and angiogenic factors and are readily obtainable. However, it is proposed that any endothelial cell that is capable of being cultured in vitro may be employed in the practice of the invention and nevertheless achieve benefits in accordance with the invention. One may mention by way of example, cells such as EA.hy9.26, ECV304, human saphenous vein endothelial cells, and the like.

Once stimulated using the tumor-derived products, the endothelial cells are then employed as immunogens in the preparation of monoclonal antibodies. The technique for preparing monoclonal antibodies against antigenic cell surface markers is quite straightforward, and may be readily carried out using techniques well known to those of skill in the art, as exemplified by the technique of Kohler & Milstein (1975). Generally speaking, the preparation of monoclonal antibodies using stimulated endothelial cells involves the following procedures. Cells or cell lines derived from human tumors are grown in tissue culture for .gtoreq.4 days. The tissue culture supernatant (`tumor-conditioned medium`) is removed from the tumor cell cultures and added to cultures of RIIVEC at a final concentration of 50% (v/v). After 2 days culture the HUVEC are harvested non-enzymatically and 1-2.times.10.sup.6 cells injected intraperitoneally into mice. This process is repeated three times at two-weekly intervals, the final immunization being by the intravenous route. Three days later the spleen cells are harvested and fused with SP2/0 myeloma cells by standard protocols (Kohler & Milstein, 1975): Hybridomas producing antibodies with the appropriate reactivity are cloned by limiting dilution.

From the resultant collection of hybridomas, one will then desire to select one of more hybridomas that produce an antibody that recognizes the activated vascular endothelium to a greater extent than it recognizes non-activated vascular endothelium. Of course, the ultimate goal is the identification of antibodies having virtually no binding affinity for normal endothelium. However, for imaging purposes this property is not so critical. In any event, one will generally identify suitable antibody-producing hybridomas by screening using, e.g., an ELISA, RIA, IRMA, IIF, or similar immunoassay, against one or more types of tumor-activated endothelial cells. Once candidates have been identified, one will desire to test for the absence of reactivity for non-activated or "normal" endothelium or other normal tissue or cell type. In this manner, hybridomas producing antibodies having an undesirably high level of normal cross-reactivity for the particular application envisioned may be excluded.

The inventors have applied the foregoing technique successfully, in that antibodies having relative specificity for tumor vascular endothelium have been prepared and isolated. In one particular example, the inventors employed the HT29 carcinoma to prepare the conditioned medium, which was then employed to stimulate HUVE cells in culture. The resultant HT29-activated HUVE cells were then employed as immunogens in the preparation of a hybridoma bank, which was ELISA-screened using HT29-activated HUVE cells and by immunohistologic analysis of sections of human tumors and normal tissues. From this bank, the inventors have selected antibodies that recognized a tumor vascular endothelial cell antigen.

The two most preferred monoclonal antibodies prepared by the inventors using this technique are referred to as tumor endothelial cell antibody 4 and 11 (TEC4 and TEC11). The antigen recognized by TEC4 and TEC11 was initially believed to migrate as a doublet of about 43 kilodaltons (kD), as assessed by SDS/PAGE. However, as detailed herein, the present inventors subsequently determined this antigen to be the molecule endoglin, which migrates as a 95 kD species on SDS/PAGE under reducing conditions. The epitopes on endoglin recognized by TEC4 and TEC11 are present on the cell surface of stimulated HUVE cells, and only minimally present (or immunologically accessible) on the surface of non-stimulated cells.

Monoclonal antibodies have previously been raised against endoglin (Gougos and Letarte, 1988; Gougos et al., 1992; O'Connel et al., 1992; Buhring et al., 1991). However, analyzing the reactivity with HUVEC or TCM-activated HUVEC cell surface determinants by FACS or indirect immunofluorescence shows the epitopes recognized by TEC-4 and TEC-11 to be distinct from those of a previous antibody termed 44G4 (Gougos and Letarte, 1988).

The TEC-4 and TEC-11 mAbs are envisioned to be particularly suitable for targeting human tumor vasculature as they label capillary and venular endothelial cells moderately to strongly in a broad range of solid tumors (and in several chronic inflammatory conditions and fetal placenta), but display relatively weak staining of vessels in the majority of normal, healthy adult tissues. TEC-11 is particularly preferred as it shows virtually no reactivity with non-endothelial cells. Furthermore, both TEC-4 and TEC-11 are complement-fixing, which imparts to them the potential to also induce selective lysis of endothelial cells in the tumor vascular bed.

In addition to their use in therapeutic embodiments, TEC-4 and TEC-11 antibodies may also be used for diagnostic, prognostic and imaging purposes. For example, TEC-4 and TEC-11 may be employed to identify tumors with high vessel density, which is known to correlate with metastatic risk and poor prognosis. This is a marked advance over the laborious enumeration of capillaries labelled with pan-endothelial cell markers or the use of complex and subjective in vivo assays of angiogenesis. Indeed, studies are disclosed herein which indicate that TEC-4 and TEC-11 can distinguish between intraductal carcinoma in situ (CIS), an aggressive preneoplastic lesion and lobular CIS, which is associated with a more indolent clinical course.

TEC-4 or TEC-11 antibodies may be linked to a paramagnetic, radioactive or fluorogenic ion and employed in tumor imaging in cancer patients, where it is contemplated that they will result in rapid imaging due to the location of endoglin on the luminal face of endothelial cells. Furthermore, TEC-4 and TEC-11 are of the IgM isotype, which limits extravasation and enables more specific imaging of antigens in the intravascular compartment. This is in contrast to 44G4 which is an IgG1 antibody.

The present invention therefore encompasses anti-endoglin antibodies and antibody-based compositions, including antibody conjugates linked to paramagnetic, radioactive or fluorogenic ions and anti-cellular agents such as anti-metabolites, toxins and the like, wherein the antibodies bind to endoglin at the same epitope as either of the MAbs TEC-4 and TEC-11. Such antibodies may be of the polyclonal or monoclonal type, with monoclonals being generally preferred, especially for use in preparing endoglin-directed antibody conjugates, immunotoxins and compositions thereof.

The identification of an antibody or antibodies that bind to endoglin at the same epitopes as TEC-4 or TEC-11 is a fairly straightforward matter. This can be readily determined using any one of variety of immunological screening assays in which antibody competition can be assessed. For example, where the test antibodies to be examined are obtained from a different source to that of TEC-4 or TEC-11 , e.g., a rabbit, or are even of a different isotype, for example, IgG1 or IgG3, a competition ELISA may be employed. In one such embodiment of a competition ELISA one would pre-mix TEC-4 or TEC-11 with varying amounts of the test antibodies prior to applying to the antigen-coated wells in the ELISA plate. By using either anti-murine or anti-IgM secondary antibodies one will be able to detect only the bound TEC-4 or TEC-11 antibodies--the binding of which will be reduced by the presence of a test antibody which recognizes the same epitope as either TEC-4 or TEC-11.

To conduct an antibody competition study between TEC-4 or TEC-11 and any test antibody, one may first label TEC-4 or TEC-11 with a detectable label, such as, e.g., biotin or an enzymatic or radioactive label, to enable subsequent identification. In these cases, one would incubate the labelled antibodies with the test antibodies to be examined at various ratios (e.g., 1:1, 1:10 and 1:100) and, after a suitable period of time, one would then assay the reactivity of the labelled TEC-4 or TEC-11 antibodies and compare this with a control value in which no potentially competing antibody (test) was included in the incubation.

The assay may be any one of a range of immunological assays based upon antibody binding and the TEC-4 or TEC-11 antibodies would be detected by means of detecting their label, e.g., using streptavidin in the case of biotinylated antibodies or by using a chromogenic substrate in connection with an enzymatic label or by simply detecting the radiolabel. An antibody that binds to the same epitope as TEC-4 or TEC-11 will be able to effectively compete for binding and thus will significantly reduce TEC-4 or TEC-11 binding, as evidenced by a reduction in labelled antibody binding. In the present case, after mixing the labelled TEC-4 or TEC-11 antibodies with the test antibodies, suitable assays to determine the remaining reactivity include, e.g., ELISAs, RIAs or western blots using human endoglin; immunoprecipitation of endoglin; ELISAs, RIAs or immunofluorescent staining of recombinant cells expressing human endoglin; indirect immunofluorescent staining of tumor vasculature endothelial cells; reactivity with HUVEC or TCM-activated MUVEC cell surface determinants indirect immunofluorescence and FACS analysis. This latter method is most preferred and was employed to show that the epitopes recognized by TEC-4 and TEC-11 are distinct from that of 44G4 (Gougos and Letarte, 1988).

The reactivity of the labelled TEC-4 or TEC-11 antibodies in the absence of any test antibody is the control high value. The control low value is obtained by incubating the labelled antibodies with unlabelled antibodies of the same type, when competition would occur and reduce binding of the labelled antibodies. A significant reduction in labelled antibody reactivity in the presence of a test antibody is indicative of a test antibody that recognizes the same epitope, i.e., one that "cross-reacts" with the labelled antibody. A "significant reduction" in this aspect of the present application may be defined as a reproducible (i.e., consistently observed) reduction in binding of at least about 10%-50% at a ratio of about 1:1, or more preferably, of equal to or greater than about 90% at a ratio of about 1:100.

The present invention further encompasses antibodies which are specific for epitopes present only on growth factor/growth factor receptor complexes, while being absent from either the individual growth factor or growth factor receptor. Thus, such antibodies will recognize and bind a growth factor/growth factor receptor complex while not recognizing or binding either the growth factor molecule or growth factor receptor molecule while these molecules are not in growth factor/growth factor receptor complex form. A "growth factor/growth factor receptor complex" as used herein refers to a growth factor ligand bound specifically to its growth factor receptor, such as, by way of example only, a VEGF/VEGF receptor complex.

As it is envisioned that the growth factor/growth factor receptor complexes to which these antibodies specifically bind are present in significantly higher number on tumor-associated endothelial cells than on non-tumor associated endothelial cells, such antibodies, when used as targeting agents, serve to further increase the targeting specificity of the agents of the invention. Such antibodies may be of the polyclonal or monoclonal type, with monoclonals being generally preferred.

The second overall general approach presented by the present invention involves the selective elicitation of vascular endothelial antigen targets on the surface of tumor-associated vasculature. This approach targets known endothelial antigens that are present, or inducible, on the cell surface of endothelial cells. The key to this aspect of the invention is the successful manipulation of antigenic expression or surface presentation such that the target antigen is expressed or otherwise available on the surface of tumor associated vasculature and not expressed or otherwise available for binding, or at least to a lesser extent, on the surface of normal endothelium.

A variety of endothelial cell markers are known that can be employed as inducible targets for the practice of this aspect of the invention, including endothelial-leukocyte adhesion molecule (ELAM-1; Bevilacqua et al., 1987); vascular cell adhesion molecule-1 (VCAM-1; Dustin et al., 1986); intercellular adhesion molecule-1 (ICAM-1; Osborn et al., 1989); the agent for leukocyte adhesion molecule-1 (LAM-1 agent), or even a major histocompatibility complex (MHC) Class II antigen, such as HLA-DR, HLA-DP or HLA-DQ (dollins et al., 1984). Of these, the targeting of ELAM-1 or an MHC Class II antigen will likely be preferred for therapeutic application, with ELAM-1 being particularly preferred, since the expression of these antigens will likely be the most direct to promote selectively in tumor-associated endothelium.

The targeting of an antigen such as ELAM-1 is the most straightforward since ELAM-1 is not expressed on the surfaces of normal endothelium. ELAM-1 is an adhesion molecule that can be induced on the surface of endothelial cells through the action of cytokines such as IL-1, TNF, lymphotoxin or bacterial endotoxin (Bevilacqua et al., 1987). In the practice of the present invention, the expression of ELAM-1 is selectively induced in tumor endothelium through the use of a bispecific antibody having the ability to cause the selective release of one or more of the foregoing or other appropriate cytokines in the tumor environment, but not elsewhere in the body. This bispecific antibody is designed to cross-link cytokine effector cells, such as cells of monocyte/macrophage lineage, T cells and/or NK cells or mast cells, with tumor cells of the targeted solid tumor mass. This cross-linking is intended to effect a release of cytokine that is localized to the site of cross-linking, i.e., the tumor.

Bispecific antibodies useful in the practice of this aspect of the invention, therefore, will have a dual specificity, recognizing a selected tumor cell surface antigen on the one hand, and, on the other hand, recognizing a selected "cytokine activating" antigen on the surface of a selected leukocyte cell type. As used herein, the term "cytokine activating" antigen is intended to refer to any one of the various known molecules on the surfaces of leukocytes that, when bound by an effector molecule such as an antibody or a fragment thereof or a naturally-occurring agent or synthetic analog thereof, be it a soluble factor or membrane-bound counter-receptor on another cell, will promote the release of a cytokine by the leukocyte cell. Examples of cytokine activating molecules include CD14 and FcR for IgE, which will activate the release of IL-1 and TNF.alpha.; and CD16, CD2 or CD3 or CD28, which will activate the release of IFN.gamma. and TNF.beta., respectively.

Once introduced into the bloodstream of an animal bearing a tumor, such a bispecific construct will bind to tumor cells within the tumor, cross-link those tumor cells with, e.g., monocytes/macrophages that have infiltrated the tumor, and thereafter effect the selective release of cytokine within the tumor. Importantly, however, without cross-linking of the tumor and leukocyte, the bispecific antibody will not effect the release of cytokine. Thus, no cytokine release will occur in parts of the body removed from the tumor and, hence, expression of ELAM-1 will occur only within the tumor endothelium.

A number of useful "cytokine activating" antigens are known, which, when cross-linked with an appropriate bispecific antibody, will result in the release of cytokines by the cross-linked leukocyte. The most preferred target for this purpose is CD14, which is found on the surface of monocytes and macrophages. When CD14 is cross linked it will stimulate the monocyte/macrophage to release IL-1, and possibly other cytokines, which will, in turn stimulate the appearance of ELAM-1 on nearby vasculature. Other possible targets for cross-linking in connection with ELAM-1 targeting includes FcR for IgE, found on Mast cells; FcR for IgG (CD16), found on NK cells; as well as CD2, CD3 or CD28, found on the surfaces of T cells. Of these, CD14 targeting will be the most preferred due to the relative prevalence of monocyte/macrophage infiltration of solid tumors as opposed to the other leukocyte cell types.

In that MHC Class II antigens are expressed on "normal" endothelium, their targeting is not quite so straightforward as ELAM-1. However, the present invention takes advantage of the discovery that immunosuppressants such as Cyclosporin A (CsA) have the ability to effectively suppress the expression of Class II molecules in the normal tissues. There are various other cyclosporins related to CsA, including cyclosporins A, B, C, D, G, and the like, which have immunosuppressive action, and will likely also demonstrate an ability to suppress Class II expression. Other agents that might be similarly useful include FK506 and rapamycin.

Thus, the practice of the MHC Class II targeting embodiment requires a pretreatment of the tumor-bearing animal with a dose of CsA or other Class II immunosuppressive agent that is effective to suppress Class II expression. In the case of CsA, this will typically be on the order of about 10 to 30 mg/kg. Once suppressed in normal tissues, Class II antigens can be selectively induced in the tumor endothelium through the use of a bispecific antibody, this one having specificity for the tumor cell as well as an activating antigen found on the surface of helper T cells. Note that in this embodiment, it is necessary that T cells, or NK cells if CD16 is used, be present in the tumor to produce the cytokine intermediate in that Class II antigen expression is achieved using IFN-.gamma., but is not achieved with the other cytokines. Thus, for the practice of this aspect of the invention, one will desire to select CD2, CD3 or CD28 (most preferably CD28) as the cytokine activating antigen.

An alternative approach to using "cytokine-activating" bispecific antibodies might be to activate the patients peripheral blood leukocytes or tumor-infiltrating lymphocytes in vitro (using IL-2 or autologous tumor cells for instance), reinfuse them into the patient and then localize them in the tumor with a bispecific antibody against any reliable leukocyte-specific marker, including CD5, CD8, CD11/CD18, CD15, CD32, CD44, CD45 or CD64. In order to selectively localize those leukocytes that had become activated from within a mixed population, it is recommended that the anti-leukocyte arm of the bispecific antibody should recognize a marker restricted to activate cells, such as CD25, CD30, CD54 or CD71. Neither of these approaches is favored as much as the `cytokine-activating` antibody approach because cross-linking to tumor cells is not a prerequisite for cytokine secretion and thus the resultant induction of cytokine-induced endothelial cell antigens may not be confined to the tumor.

The targeting of the other adhesion molecules, ICAM-1, VCAM-1 and LAM-1 agent, will typically not be preferred for the practice of therapeutic embodiments, in that these targets are constitutively expressed in normal endothelium. Thus, these adhesion molecules will likely only be useful in the context of diagnostic embodiments. Furthermore, it is unlikely that ICAM-1 or VCAM-1 expression by normal endothelial cells would be inhibited in vivo by CsA because low levels of expression of both markers are constitutive properties of human endothelial cells (Burrows et al., 1991). However, it may still be possible to utilize one of these molecules in diagnostic or even therapeutic embodiments because their level of expression on the endothelial cell surface is increased 10-50 fold by cytokines. As a consequence, there may be a therapeutic or diagnostic `window` enabling use of anti-ICAM-1 or anti-VCAM-1 conjugates in an analogous way to the proven clinical utility of some antibodies against `tumor-associated` antigens whose expression differ quantitatively but not qualitatively from normal tissues.

The tumor antigen recognized by the bispecific antibodies employed in the practice of the present invention will be one that is located on the cell surfaces of the tumor being targeted. A large number of solid tumor-associated antigens have now been described in the scientific literature, and the preparation and use of antibodies are well within the skill of the art (see, e.g., Table II hereinbelow). Of course, the tumor antigen that is ultimately selected will depend on the particular tumor to be targeted. Most cell surface tumor targets will only be suitable for imaging purposes, while some will be suitable for therapeutic application. For therapeutic application, preferred tumor antigens will be TAG 72 or the HER-2 proto-oncogene protein, which are selectively found on the surfaces of many breast, lung and colorectal cancers (Thor et al., 1986; Colcher et al., 1987; Shepard et al., 1991). Other targets that will be particularly preferred include milk mucin core protein, human milk fat globule (Miotti et al., 1985; Burchell et al., 1983) and even the high Mr melanoma antigens recognized by the antibody 9.2.27 (Reisfeld et al., 1982).

In still further embodiments, the inventors contemplate an alternative approach for suppressing the expression of Class II molecules, and selectively eliciting Class II molecule expression in the locale of the tumor. This embodiment takes advantage of the fact that the expression of Class II molecules can be effectively inhibited by suppressing IFN-.gamma. production by T-cells, e.g., through use of an anti-CD4 antibody (Street et al., 1989). Thus, in this embodiment, one will desire to pretreat with a dose of anti-CD4 that is effective to suppress IFN-.gamma. production and thereby suppress the expression of Class II molecules (for example, on the order of 4 to 10 mg/kg). After Class II expression is suppressed, one will then prepare and introduce into the bloodstream an IFN-.gamma.-producing T-cell clone (e.g., T.sub.1 1 or CTL) specific for an antigen expressed on the surface of the tumor cells.

A preferred means of producing the IFN-.gamma.-producing T-cell clone is by a method that includes removing a portion of the tumor mass from the patient, extracting tumor infiltrating leukocytes from the tumor, and expanding the tumor infiltrating leukocytes in vitro to provide the IFN-.gamma. producing clone. This clone will necessarily be immunologically compatible with the patient, and therefore should be well tolerated by the patient. It is proposed that particular benefits will be achieved by further selecting a high IFN-.gamma. producing T-cell clone from the expanded leukocytes by determining the cytokine secretion pattern of each individual clone every 14 days. To this end, rested clones will be mitogenically or antigenically-stimulated for 24 hours and their culture supernatants assayed by a specific sandwich ELISA technique (Cherwinski et al., 1989) for the presence of IL-2, IFN-.gamma., IL-4, IL-5 and IL-10. Those clones secreting high levels of IL-2 and IFN-.gamma., the characteristic cytokine secretion pattern of TH1 clones, will be selected. Tumor specificity will be confirmed using proliferation assays. Furthermore, one will prefer to employ as the anti-CD4 antibody an anti-CD4 Fab, because it will be eliminated from the body within 24 hours after injection and so will not cause suppression of the tumor recognizing T cell clones that are subsequently administered. The preparation of T-cell clones having tumor specificity is generally known in the art, as exemplified by the production and characterization of T cell clones from lymphocytes infiltrating solid melanoma tumors (Maeda et al., 1991).

The invention contemplates that still further advantages will be realized through combination regimens wherein both the tumor endothelial vasculature and the tumor itself are targeted. Combination regimens may thus include targeting of the tumor directly with either conventional antitumor therapy, such as with radiotherapy or chemotherapy, or through the use of a second immunological reagent such as an antitumor immunotoxin. In fact, dramatic, synergistic antitumor effects were seen by the inventors when solid tumors were targeted with both an antitumor endothelial cell immunotoxin and an antitumor cell immunotoxin. Such combination therapy is founded theoretically on 1) the use of the endothelial-directed immunotoxin to kill those tumor cells that depend upon vascular oxygen and nutrients, and 2) the use of the tumor-directed immunotoxin to kill those tumor cells that may have an alternate source of oxygen and nutrients (i.e., those tumor cells lining the vasculature and those forming the outer boundary of the tumor mass). Thus, it is proposed that particular advantages will be realized through the targeting of agents both to tumor cell targets as well as to tumor endothelial cell targets.

The invention further contemplates the selected combinations of agents particularly adapted for use in connection with the methods of the present invention, defined as including a first pharmaceutical composition which includes a bispecific antibody recognizing an activating antigen on the cell surface of a leukocyte cell and a tumor antigen on the cell surface of tumor cells of a vascularized solid tumor, together with a second pharmaceutical composition comprising a second antibody or fragment thereof linked to a selected therapeutic or diagnostic agent that recognizes the induced endothelial antigen. In accordance with one aspect of the invention, these agents may be conveniently packaged together, being suitably aliquoted into separate containers, and the separate containers dispensed in a single package.

In particular embodiments, the activating antigen induced by the bispecific antibody will be CD2, CD3, CD14, CD16, FcR for IgE, CD28 or the T-cell receptor antigen, as may be the case. However, preferably, the bispecific antibody will recognize CD14, and induce the expression of IL-1 by monocyte/macrophage cells in the tumor, or recognize CD28 and induce the expression of IFN-.gamma. by T-cells in the tumor. Where IL-1 is the cytokine intermediate, the second antibody will preferably be one that recognizes ELAM-1, since this adhesion molecule will be induced on the surface of endothelial cells by IL-1. In contrast, where IFN-.gamma. is the intermediate, the second antibody will preferably be one that recognizes an MHC Class II antigen. In the later case, one might desire to include with the combination a third pharmaceutical composition comprising one of the cyclosporins, or another immunosuppressive agent useful for suppressing Class II expression.

Furthermore, in that the invention contemplates combination regimens as discussed above, particular embodiments of the invention will involve the inclusion of a third pharmaceutical composition comprising an antitumor antibody conjugated to a selected agent, such as an anti-tumor immunotoxin. In these embodiments, particularly preferred will be the targeting of tumor antigens such as p185.sup.HER2, milk mucin core protein, TAG-72, Lewis a, carcinoembryonic antigen (CEA), the high Mr melanoma antigens recognized by the 9.2.27 antibody, or the ovarian-associated antigens recognized by OV-TL3 or MOV18. These same antigens will also be preferred as the target for the bispecific antibody. Of course, where such a bispecific antibody is employed in combination with an antitumor antibody, it may be desirable to target different tumor antigens with the bispecific and antitumor antibody.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1a. Induction of I-E.sup.k on SVEC cells by IFN-.gamma. in regular medium, r.IFN-.gamma., or r.IFN-.gamma. plus excess neutralizing anti-IFN-.gamma. antibody. SVEC cells were cultured for 72 hours in regular medium ( - - - ), r.IFN-.gamma. ( . . . ) or r.IFN-.gamma. plus excess neutralizing anti-IFN-.gamma. antibody ( . . . ). Their expression of I-E.sup.k was then measured by M5/114 antibody binding by indirect immunofluorescence using the FACS (fluorescence-activated cell sorter). Other cultures were treated with r. IFN-.gamma. and stained with an isotype-matched control antibody ( . . . ).

FIG. 1b. Induction of I-E.sup.k on SVEC cells by IFN-.gamma. in C1300-conditioned media. SVEC cells were cultured for 72 hours in C1300-conditioned medium ( . . . ), C1300(Mu.gamma.)-conditioned medium ( . . . ) or C1300(Mu.gamma.)-conditioned medium plus excess neutralizing anti-IFN-.gamma. antibody ( . . . ). Their expression of I-E.sup.k was then measured as in FIG. 1a. Other cultures were treated with C1300(Mu.gamma.)-conditioned medium and stained with an isotype-matched control antibody (- - -).

FIG. 2a. Expression of I-E.sup.k and H-2K.sup.k by pure and mixed populations of C1300 and C1300(Mu.gamma.) cells stained with anti-I-E.sup.k antibody. C1300 cells ( . . . ), C1300(Mu.gamma.) cells (- - -), a mixture of C1300 and C1300(Mu.gamma.) cells in the ratio 7:3 cocultured in vitro ( . . . ) or cells recovered from a mixed subcutaneous tumor in a BALB/c nu/nu mouse ( . . . ) were stained with anti-I-E.sup.k antibody by indirect immunofluorescence using the FACS. No staining of any tumor cell population was seen with the isotype-matched control antibodies.

FIG. 2b. Expression of I-E.sup.k and H-2K.sup.k by pure and mixed populations of C1300 and C1300(Mu.gamma.) cells stained with anti-H-2K.sup.k antibody. C1300 cells ( . . . ), C1300(Mu.gamma.) cells (- - -), a mixture of C1300 and C1300(Mu.gamma.) cells in the ratio 7:3 cocultured in vitro ( . . . ) or cells recovered from a mixed subcutaneous tumor in a BALB/c nu/nu mouse ( - - - ) were stained with anti-H-2K.sup.k antibody by indirect immunofluorescence using the FACS. No staining of any tumor cell population was seen with the isotype-matched control antibodies.

FIG. 3. Tumorigenicity, growth, and tumor endothelial cell Ia.sup.d expression in pure and mixed subcutaneous C1300 and C1300(Mu.gamma.) tumors. BALB/c nu/nu mice were injected with a total of 2.times.10.sup.7 tumor cells in which the ratios of C1300: C1300(Mu.gamma.) cells were 10:0 (.DELTA.), 9:1 (.largecircle.), 8:2 (.circle-solid.), 7:3 (.diamond.), 5:5 (.box-solid.) 3:7 (.quadrature.) or 0:10 (.tangle-solidup.). The vertical axis shows the mean diameter of the tumors at various times after injection. Also shown are the percentage of animals in each group which developed tumors. The proportion of Ia.sup.d -positive vascular endothelial cells was categorized as follows: +, 75-100%; .+-., 25-75%; -, 0-5%; n.d., not determined because no intact blood vessels were visible. Standard deviations were <15% of mean diameters and are not shown.

FIG. 4a. Killing activity of anti-Class II immunotoxin (M5/114 dgA) against unstimulated SVEC mouse endothelial cells. The data shown are for treatment of cells with varying concentrations of ricin (.quadrature.); M5/114dgA (.circle-solid.); and the control immunotoxin CAMPATH-2dgA (.largecircle.).

FIG. 4b. Killing activity of anti-Class II immunotoxin (M5/114 dgA) against SVEC mouse endothelial cells stimulated with conditioned medium from the IFN-.gamma.-secreting tumor C1300(Mu-.gamma.). The data shown are for treatment of cells with varying concentrations of ricin (.quadrature.); M5/114dgA (.circle-solid.); and the control immunotoxin CAMPATH-2dgA (.largecircle.).

FIG. 5. This figure also shows the killing of SVEC cells under various conditions by the anti-Class II immunotoxin, M5/114dgA. The data shown are for treatment of cells with varying concentrations of the immunotoxin following treatment with IFN-.gamma. TCM (.largecircle.); C1300 TCM (.tangle-solidup.); C1300(Mu-.gamma.) TCM (.box-solid.); and C1300(Mu-.gamma.) treated with anti-IFN-.gamma. (.quadrature.).

FIG. 6a. Shows a comparison of killing activity of an anti-Class I (antitumor) immunotoxin (11-4.1-dgA, which recognized H-2K.sup.k) and an anti-Class II (anti-tumor endothelial cell) immunotoxin (M5/114-dgA) against a 70:30 mixed population of C1300 and C1300(Mu-.gamma.) cells. Data was obtained through treatment of the cells with ricin (.circle-solid.); the 11-4.1-dgA immunotoxin (.largecircle.); the M5/114-dgA immunotoxin (.box-solid.) and a control immunotoxin (.quadrature.).

FIG. 6b. Shows killing of cells freshly recovered from subcutaneous tumors in mice. Data was obtained through treatment of the cells with ricin (.circle-solid.); the 11-4.1-dgA immunotoxin (.smallcircle.); the M5/114-dgA immunotoxin (.box-solid.) and a control immunotoxin (.quadrature.).

FIG. 7a. Killing of pure populations of C1300 (.circle-solid.) and C1300(Mu-.gamma.) (.smallcircle.) by the antitumor cell immunotoxin, 11-4.1-dgA. Also shown are 70:30 mixed populations mixed in vitro or in vivo (i.e., recovered from S/C tumors). Also shown are controls, including ricin (.tangle-solidup.) and a control immunotoxin (.DELTA.).

FIG. 7b. Killing of pure populations of C1300 (.circle-solid.) and C1300(Mu-.gamma.) (.smallcircle.) by the antitumor cell immunotoxin, 11-4.1-dgA, as shown in FIG. 7a, with the controls, ricin (.tangle-solidup.) and a control immunotoxin (.DELTA.), for comparison.

FIG. 8. This figure shows the in vivo antitumor effects of the anti-endothelial cell immunotoxin, M5/114-dgA, at various doses, including 20 .mu.g (.smallcircle.) and 40 .mu.g (.box-solid.). These studies involved the administration of the immunotoxin intravenously 14 days after injection of tumor cells. Controls included the use of a control immunotoxin, CAMPATH-2-dgA (.DELTA.) and PBS +BSA (.tangle-solidup.).

FIG. 9. This figure is a histological analysis of 1.2 cm H&E-stained tumor sections 72 hours after treatment with 20 .mu.l of the anti-Class II immunotoxin, M5/114-d9A.

FIG. 10. This figure is a histological analysis of 1.2 cm H&E-stained tumor sections 72 hours after treatment with 100 .mu.g of the anti-Class II immunotoxin, M5/114-dgA.

FIG. 11. This figure is a representation of the appearance of a solid tumor 48-72 hours after intravenous immunotoxin treatment, and compares the effect achieved with anti-tumor immunotoxin, to that achieved with anti-endothelial cell immunotoxin therapy.

FIG. 12. This figure shows the antitumor effects of single and combined treatments with anti-Class I and anti-Class II immunotoxins in SCID mice bearing large solid C1300(Mu-.gamma.) tumors. SCID mice bearing 1.0-1.3 cm diameter tumors were injected intravenously 14 days after tumor inoculation with 20 .mu.g of Class II immunotoxin (.smallcircle.), 100 .mu.g Class I immunotoxin (.circle-solid.), or diluent alone (.tangle-solidup.). Other animals received the anti-Class II immunotoxin followed by two days later by the anti-Class I immunotoxin (.box-solid.), or vice versa (.quadrature.). Tumor size was measured at regular intervals and is expressed as mean tumor diameter .+-.SEM. Each treatment group consisted of 4-8 mice.

FIG. 13a. Gel electrophoretic analysis of proteins immunoprecipitated from 35S-labelled human umbilical vein endothelial cells (HUVEC), showing that TEC-4 and TEC-11 recognize endoglin. 12.5% SDS-PAGE gel of proteins immunoprecipitated under reducing (lanes 2-4) or non-reducing (lanes 5-7) conditions with TEC-4 (lanes 2,5), TEC-11 (lanes 3,6) or TEPC-183 (lanes 4,7). Lane 1: Position of the .sup.14 C-labelled standards of the molecular weights indicated. Lane 8: Positions of 95 kDa and 180 kDa species.

FIG. 13b. Reactivity of TEC-4 and TEC-11 with human endoglin transfectants, showing that TEC-4 and TEC-11 recognize endoglin. Parental murine L cells and L cell transfectants expressing human endoglin were incubated with purified MAb TEC-11, TEC-4 and 44G4 followed by FITC-conjugated F(ab').sub.2 goat anti-mouse IgG (H+L). The staining observed on the parental L cells with the MAb (white histograms) was indistinguishable from that observed with IgM and IgG1 controls. The L cell endoglin transfectants (black histograms) were specifically reactive with all 3 antibodies as revealed by the percentage of cells included within the gate and shown in parentheses.

FIG. 14. Crossblocking of TEC-4 and TEC-11 antibodies. Biotinylated antibodies (10 .mu.g/ml) were mixed with an equal volume of unlabelled TEC-4 antibody at 10 .mu.g/ml (.quadrature.) 100 .mu.g/ml () or 1000 .mu.g/ml (.box-solid.) or with unlabelled TEC-11 antibody at 10 .mu.g/ml (), 100 .mu.g/ml () and were added to HUVEC in PBS-BSA-N.sub.3. Indirect immunofluorescence staining was carried out as described in Example V with the exception that labelled antibody binding was detected with a streptavidin-phycoerythrin conjugate. Each group of histograms shows the percent blocking of the biotinylated antibody by the different concentrations of unlabelled antibodies. Bar: SD of triplicate determinations.

FIG. 15. Complement fixation by TEC-4 and TEC-11 antibodies. HUVEC were incubated with TEC-4 (.circle-solid.), TEC-11 (.box-solid.) or MTSA (.tangle-soliddn.) antibodies, washed and subsequently incubated with guinea-pig complement. Cell number and viability were determined by trypan blue dye exclusion.

FIG. 16a. Correlation between TEC-11 binding and cellular proliferation in HUVEC. HUVEC from sparse cultures (hatched histogram) or post-confluent cultures (open histogram) were stained with TEC-11 by indirect immunofluorescence. Also shown, confluent HUVEC stained with negative control antibody MTSA (stippled histogram). Endoglin.sup.lo and endoglin.sup.hi populations of post-confluent HUVEC were separated on a FACStar Plus cell sorter as indicated and subsequently analyzed for PaNA and DNA content.

FIG. 16b. Lack of correlation between LM142 binding and cellular proliferation in HUVEC. HUVEC from sparse (hatched histogram) or post-confluent (open histogram) were stained with the anti-vitronectin receptor antibody LM142. Also shown, sparse HUVEC stained with negative control antibody MTSA (stippled histogram)