EP0378676A1 - Tumor necrosis enhancing factor and methods of preparation and use - Google Patents

Abstract

Factors which act in synergy with a tumor necrosis factor (TNF), to produce the expression of a tissue growth factor (of the thromboplastin or factor III type) in endothelial cells, have been isolated. Both protein factors are obtained from mouse meth A sarcoma cells. The first factor is characterized by a molecular weight of 40 to 50 KDA under non-reducing conditions and from 65 to 75 KDA under reducing conditions and by isoelectric points between 6.8 and 7.2. Said factor is further characterized: by its loss of activity when treated with protease K, by its capacity for elution from heparin-sepharose with 0.5 M NaCl, by its capacity for binding to a column of FPLC-ProRPC in reverse phase, by its capacity of elution from the column of FPLC-ProRPC in reverse phase by methanol gradient rising to about 50%, by its capacity of migration in the form of a band unit on an SDS-polyacrylamide gel and by its adsorption capacity on concanavalin A-Sepharose. The second factor is characterized by a molecular weight of 10 to 30 KDa by gel filtration, by its loss of activity when it is treated with trypsin, by its thermosensitivity, by its capacity of elution from Ultrogel. heparin with 0.5 M NaCl, by its binding capacity on a mono Q column and by its separation by elution using an ascending salt gradient with 0.4 M NaCl.

Description

TUMOR NECROSIS ENHANCING FACTOR AMD METHODS OF PREPARATION AND USE

This application is a continuation-in-part of U.S. Serial No. 219,650, filed July 15, 1988, the contents of which are hereby incorporated by reference into this application.

The invention described herein was Bade in the course of work under Grants Nos. CA 43902, HL 34625, HL 42833, and HL 42507 from the Public Health Service, U.S. Department of Health and Human Services. The U.S. Government has certain rights in this invention.

Background of the Invention

Throughout this application various publications are referenced by Arabic numerals. Full citations for these references may be found at the end of the specification immediately preceding the claims. The disclosures of^' these publications in their entireties are hereby incorporated by reference into this application to describe more fully the state of the art to which this invention pertains.

Tumor Necrosis Factor (TNF) is a cytokine (1) which has been shown to act on a variety of cellular targets including the endothelium (1) . Because endothelial cells form the luminal vascular surface, TNF-induced changes in endothelial cell properties could alter the basic blood-compatibility of the vessel wall. In this context, endothelium has been shown to provide specific receptors for TNF and a consequence of TNF interaction with these receptors includes modulation of coagulant properties (2-3). Thus, the coagulant phenotype of the endothelium is altered from the quiescent state, where mechanisms promoting the fluidity of blood predominate, to a perturbed state, where procoagulant pathways are enhanced (3) .

Although these changes could predispose to clot forma¬ tion, when mice are infused with doses of TNF in the range of 1-10 μg/animal, widespread clot formation is not observed. In contrast, at these same doses of TNF, necrosis of tumors does not occur in mice bearing meth A fibrosar-'_as and clot formation throughout the tumor bed is seen (4) .

An investigation was conducted then to examine whether a tumor-cell derived substance could act synergistically with TNF to necrotize tumors. Such a tumor cell-derived substance might explain why certain tumors respond to TNF while others do not, even though TNF binding sites are present. Such a tumor necrotizing substance could act on the vessel wall as well as on other cellular elements surrounding and forming the tumor. This substance could facilitate TNF-induced hemorrhagic necrosis via several mechanisms, one of which would be by further augmenting the increase in endothelial cell procoagulant activity which substances like TNF or other onokines bring about. Another aspect of this investigation concerned whether tumor cell-derived substance could increase the procoagulant activity which TNF induces in the endothelium.

As a result of this investigation it has been found that meth A fibrosarcoma cells in culture produce a factor, tumor necrosis enhancing factor (TNEF) , which increases production of a central procoagulant, tissue factor, whose synthesis in endothelium is induced by TNF.

Summary of the Invention

This invention concerns a purified tumor necrosis enhancing factor. In one preferred embodiment the polypeptide is characterized by an apparent molecular weight between about 40,000 and about 50,000 daltons under non-reducing conditions and between about 65,000 and about 75,000 daltons under reducing conditions, and by isoelectric focussing peaks at about 6.8 and about 7.2. The factor is further characterized by loss of activity upon treatment wit' protease K, by an affinity for Heparin-Sepharose from which the factor may be eluted with about 0.5 M NaCl, by the ability to bind to a reverse phase FPLC-ProRPC column, and by the ability to be eluted from a reverse phase FPLC-ProRPC column to which the factor is bound in an ascending methanol gradient at about 50%, by the ability to migrate as a single band on an SDS-polyacrylamide gel, and by adsorption to Concanavalin A-Sepharose.

This invention also concerns a purified, biologically active fragment of purified tumor necrosis enhancing factor which is characterized by an apparent molecular weight between about 10,000 and about 30,000 daltons, by the loss of activity upon treatment with trypsin or heat, by an affinity for Heparin ultrogel from which the fragment may be eluted with about 0.5 M NaCl, by the ability to bind to a Mono Q column, and by the ability to be eluted from a Mono Q column to which the fragment is bound in an ascending salt gradient at about 0.4 M NaCl.

This invention concerns a pharmaceutical composition comprising an amount of a tumor necrosis enhancing factor effective to enhance the tumor inhibitory activity of a cytotoxic agent and a pharmaceutically acceptable carrier.

This invention also concerns a pharmaceutical composition comprising an amount of a biologically active fragment of purified tumor necrosis enhancing factor effective to enhance the tumor inhibitory activity of a cytotoxic agent and a pharmaceutically acceptable carrier.

Brief Description of the Figures

Figure 1. Enhanced induction of endothelial cell tissue factor in response to TNF in the presence of tumor-conditioned medium.

A. Conditioned medium from meth A sarcoma cells at the indicated dilution was incubated with endothelial cells in serum-free medium (RPMI containing lOmM HEPES, pH 7.4, 20 μg/ml transferrin, 10 μg/ml insulin, 1 μg/ml polymyxin B and 5 mg/ml human serum albumin) . Cultures contained either serum-free medium alone (>.,, TNF (0.1 nM) alone (TNF) , undiluted conditioned medium alone (CM) , or TNF in the presence of conditioned medium at the indicated dilution (TNF + CM) . Each addition was in a volume of 10 μl and the volume of serum-free medium was 1 ml. After 7 hours of incubation at 37*C, monolayers were washed, fresh serum-free medium (0.5 ml) was added along with Factors Vila and/or X. After 8 minutes, the reaction was stopped and Factor Xa formation was assessed. The cross-hatched bar corresponds to cultures exposed to tumor-conditioned medium (1/2 dilution) in the presence of TNF followed by incubation for 1 hour at 37° with purified monoclonal antibody (10 μg/ml) to human tissue factor. Then, Factors Vila and X were added. Purified mouse IgG (10 μg/ml and 100 μg/ml) had no effect on effect on endothelial cell coagulant activity as measured in this assay. The darkened bar corresponds to cultures incubated with tumor-conditioned medium (1/2 dilution) in the presence of TNF in which the coagulant assay was carried out in the presence of only Factor X. Results shown are the mean and SEM of triplicate determinations. Factor Xa generation was the same as untreated controls when tumor-conditioned medium was added directly to the Factor Vlla-X incubation mixtur in the absence of endothelial cells.

B. Tumor-conditioned medium (obtained as described in A above) (1 ml) was applied to a Sephadex G150 column (0.9 x 55 cm), eluted with 10 mM HEPES (pH 7.4), 0.1 M NaCl and fractions (1.3 ml) were collected. Aliquots of the indicated column fractions, at a 1/4 dilution, were then incubated with endothelial cell onolayers in the presence of TNF (0.1 nM) for 7 hours at 37 'C. The tissue factor assay was carried out as described above (A) and Factor Xa formed over 7 minutes is shown (mean + SEM) . (TNF) denotes cells incubated with TNF a^-r_e (0.1 nM) and column buffer. Column fractions 1 to 9 were inactive in this assay. The gel filtration column was calibrated with standard proteins including ribonuclease (13,700), chymotrypsinogen A (25,000), ovalbumin (43,000) and albumin (67,000).

Figure 2. Results of Heparin Ultrogel (IBF

Biotechniques Co., Savage, Maryland) chromatography of a meth A fibrosarcoma culture supernatant collected in protein-free medium.

A. This panel (left hand) demonstrates that following adsorption of TNEF activity to the resin, it was eluted with 0.5 M NaCl, with the bulk of the protein, not at higher salt concentrations where FGF and tumor cell- derived mitogens are found.

B. The salt gradient elution in this panel (right hand) of Figure 2B shows elution of TNF activity towards the end of the salt gradient, well separated from much of the applied protein.

Figure 3. Chromatography of the pool eluted from the Heparin Ultrogel column on an FPLC (fast-pressure liquid chromatography) Mono Q^β column (Pharmacia

Inc. , Piscataway, New Jersey) . Elution occurred as a gradient of increasing salt was applied (elution ocurred at about 0.4 M NaCl). Again, TNEF activity was separated from the bulk of the applied protein.

Figure 4. Purification of tumor necrosis enhancing factor (TNEF) by chromatography of meth A-conditioned medium on Mono S and reverse phase FPLC columns.

A. Meth A-conditioned medium was harvested from cultures, subjected to ammonium sulfate precipitation, dialysis, a negative adsorption step to Q-Sepharose and then chromatographed on Mono S. The column was eluted with an ascending salt gradient (dashed line) and the maximal salt concentration at the end of the gradient was 0.5 M. Protein content of the fractions is plotted versus tissue factor activity induced in endothelial cultures following a 7 hr incubation of partially purified tumor necrosis enhancing factor (TNEF) with the cells (only results of fractions with peak activity are shown) .

B. The pool of activity eluted from the Mono S column was applied to a reverse phase ProRPC column and eluted with an ascending methanol gradient (maximal methanol concentration at the end of the gradient was 100%) . OD at 280 nm is plotted versus tissue factor activity induced in endothelial cultures as described in A above. The mean of duplicate determinations of tissue factor activity is shown in A and B above. Detailed Description of the Invention

This invention concerns a purified tumor necrosis enhancing factor. In one preferred embodiment the factor is a polypeptide characterized by an apparent molecular weight between about 40,000 and about 50,000 daltons under non-reducing conditions and by an apparent molecular weight between about 65,000 and about 75,000 daltons under reducing conditions, preferably about 44,000 daltons on a non-reducing SDS- polyacryla ide gel, and about 70,000 daltons on a reducing SDS-polyacrylamide gel, and by isoelectric focussing peaks about 6.8 and about 7.2. The factor is further characterized by loss of activity upon treatment with protease K, by an affinity for Heparin- Sepharose from which the factor may be eluted with about 0.5 M NaCl, by the ability to bind to a reverse phase FPLC-ProRPC column, and by the ability to be eluted from a reverse phase FPLC-ProRPC column to which the factor is bound in an ascending methanol gradient at about 50%, by the ability to migrate as a single band on an SDS-polyacrylamide gel, and by adsorption to Concanavalin A-Sepharose.

This invention also concerns a purified, biologically active fragment of purified tumor necrosis enhancing factor characterized by an apparent molecular weight between about 10,000 and about 30,000 daltons, preferably about 20,000 daltons, by the loss of activity upon treatment with trypsin or heat, by an affinity for Heparin Ultrogel from which the fragment may be eluted with about 0.5 M NaCl, by the ability to bind to a Mono Q column, and by the ability to be eluted from a Mono Q column to which the fragment is bound in an ascending salt gradient at about 0.4 M NaCl .

The purified tumor necrosis enhancing factor preferably is of animal, e.g., human, rat, or mouse origin. Heparin Ultrogel is obtained by coupling heparin to Ultrogel A4, i.e., 4% agarose, by epichlorohydrin using a six carbon spacer arm. Heparin Ultrogel may be obtained from IBF Biotechniques Co., Savage, Maryland. A Mono Q* column is an ion-exchange column prepacked with Mono Q^β, i.e., mono dispersed hydrophilic polymer beads with an extremely narrow particle size distribution for chromatographic resolution of proteins (pH range 2-12) . Mono Q* columns may be obtained from Pharmacia, Inc., Piscataway, New Jersey. Con A- Sepharose may be obtained by coupling Concanavalin A to Sepharose. Con A-Sepharose may be obtained from Pharmacia, Inc., Piscataway, New Jersey. Heparin- Sepharose may be obtained by coupling heparin to Sepharose. Heparin-Sepharose may be obtained from Pharmacia, Inc., Piscataway, New Jersey. ProRPC is a reverse-phase column and may be obtained from Pharmacia, Inc. , Piscataway, New Jersey.

This invention also concerns a purified DNA molecule, a cDNA molecule or an isolated genomic DNA molecule, encoding a tumor necrosis enhancing factor, and a purified DNA molecule, a cDNA molecule or an isolated genomic DNA molecule, encoding a biologically active fragment of tumor necrosis activating factor. Such DNA can be readily obtained by one skilled in the art utilizing well known methods, e.g. , the preparation of oligonucleotide probes and the use of such probes to obtain the DNA, such probes in turn being based upon the amino acid sequence of tumor necrosis enhancing factor which may be readily obtained by conventional methods such as automated DNA sequencing.

The invention provides a pharmaceutical composition comprising an amount of a tumor necrosis enhancing factor effective to enhance the tumor inhibitory activity of a cytotoxic agent, and a pharmaceutically acceptable carrier. Also provided is a pharmaceutical composition comprising an amount of a biologically active fragment of tumor necrosis enhancing factor effective to enhance the tumor inhibitory activity of a cytotoxic agent, and a pharmaceutically acceptable carrier. Pharmaceutically acceptable carriers include water, and other conventional carriers, e.g., sugars, such as mannitol, in water.

This invention also provides a pharmaceutical composition comprising an amount of tumor necrosis enhancing factor effective to enhance the tumor inhibitory activity of a cytotoxic agent, a pharmaceutically acceptable carrier, and further comprising an amount of tumor necrosis factor effective to inhibit tumor growth.

This invention also provides a pharmaceutical composition comprising an amount of a biologically active fragment of tumor necrosis enhancing factor effective to enhance the tumor inhibitory activity of a cytotoxic agent, a pharmaceutically acceptable carrier, and further comprising an amount of tumor necrosis factor effective to inhibit tumor growth.

This invention also provides a method of inhibiting the growth of tumor cells, particularly the localized inhibition of tumor cell growth within the vascular system, which comprises contacting the cells with an effective tumor inhibitory amount of a cytotoxic agent in the presence of an amount of a tumor necrosis enhancing factor effective to enhance the tumor inhibitory activity of the cytotoxic agent. In one embodiment, the cytotoxic agent is a polypeptide tissue factor having tumor inhibitory activity, preferably tumor necrosis factor.

This invention also provides a method of inhibiting the growth of tumor cells, particularly the localized inhibition of tumor cell growth within the vascular system, which comprises contacting the cells with an effective tumor inhibitory amount of a cytotoxic agent in the presence of an amount of the biologically active fragment of tumor necrosis enhancing factor effective to enhance the tumor inhibitory activity of the cytotoxic agent. In one embodiment, the cytotoxic agent is a polypeptide tissue factor having tumor inhibitory activity, e.g., tumor necrosis factor.

This invention also provides a method of inhibiting the growth of tumor cells which comprises contacting the cells with an effective amount of a pharmaceutical composition, the composition comprising an amount of a tumor necrosis enhancing factor effective to enhance the tumor inhibitory activity of a cytotoxic agent and a pharmaceutically acceptable carrier.

This invention also provides a method of inhibiting the growth of tumor cells which comprises contacting the cells with an effective amount of a pharmaceutical composition, the composition comprising an amount of the biologically active fragment of tumor necrosi enhancing factor effective to enhance the tumo inhibitory activity of a cytotoxic agent and pharmaceutically acceptable carrier.

This invention also provides a method of treating a subject having a tumor which comprises administering to the subject an amount of a pharmaceutical composition, the composition comprising an amount of tumor necrosis enhancing factor effective to enhance the tumor inhibitory activity of a cytotoxic agent and a pharmaceutically acceptable carrier, effective to necrotize the tumor.

This invention also provides a method of treating a subject' having a tumor which comprises administering to the subject an amount of a pharmaceutical composition, the composition comprising an amount of the biologically active fragment of tumor necrosis enhancing factor effective to enhance the tumor inhibitory activity of a cytotoxic agent and a pharmaceutically acceptable carrier, effective to necrotize the tumor.

This invention also provides a method of preparing a tumor necrosis enhancing factor which comprises recovering the factor from tumor cells and purifying the factor so recovered.

This invention also provides a method of preparing a biologically active fragment of tumor necrosis enhancing factor which comprises recovering the fragment from tumor cells and purifying the fragment so recovered.

This invention also provides a method of preparing a tumor necrosis enhancing factor which comprises expressing a purified DNA molecule encoding the tumor necrosis enhancing factor in a suitable host under conditions so that the factor is produced, recovering the factor so produced, and purifying the factor so recovered. Also provided is a method of preparing a biologically active fragment of tumor necrosis enhancing factor which comprises expressing a purified DNA molecule encoding the biologically active fragment of tumor necrosis enhancing factor in a suitable host under conditions so that the fragment is produced, recovering the fragment produced, and purifying the fragment s recovered. Suitable hosts include bacteria, e.g., Escherichia coli, and mammalian cells and yeast cells. Methods for expressing DNA in such hosts are well known in the art, as are methods for recovering polypeptides produced, such as tumor necrosis enhancing factor, in such hosts and purifying factors so recovered.

This invention is illustrated in the Experimental Details and Experimental Discussion sections whic follow. These sections are set forth to aid in a understanding of the invention but are not intended to, and should not be construed to, limit in any way th invention as set forth in the claims which follo thereafter.

Experimental Details

Materials and Methods

First Series of Experiments: In vivo Infusion and

Morphologic Studies.

For studies examining the effect of TNF on meth A sarcomas _in vivo, Balb/c mice were injected intradermally with meth A sarcoma cells (10 cells/animal; provided by Drs. Hoffman and Old,

Memorial Sloan-Kette..lug Cancer Center, New York, N.Y.

(7)). After 7-10 days, when tumors reached a size of about 1 cm in diameter, animals were injected intravenously via a tail vein with TNF alone (3 μg/animal) or TNF in the presence of either human fibrinogen (100 μg/ani.mal) or 125I fibrinogen (7.5 g μg/animal) . Recombinant human TNF (10 U/mg) was generously provided by BASF (F.R.G.), and this preparation was homogenous on SDS-PAGE and distinct from lymphotoxin, as described previously (8) . TNF was heat-inactivated as described previously (8) . Highly purified human fibrinogen, provided by Dr. J. eitz (Hamilton University, Ontario, Canada), was radiolabelled by the lactoperoxidase method (9) (150 m μCi/mg) and migrated as three bands of unequal intensity on reduced SDS-PAGE corresponding to the alpha, beta and gamma chains. At the indicated time after the TNF infusion, fibrin deposition/accumulation of radioactivity in the tumor was assessed as follows. For morphologic studies, mice were anestheized and subjected to whole-body beating heart perfusion fixation (90-110 mm Hg) . Incorporation of radioactivity into the mouse tissue was determined after infusion of 125I-fιbrιnogen by removing a piece of tissue, weighing it and counting the sample in a gamma-counter. The presence of fibrin in tumor tissue was studied by excising tumor tissue, cutting it up finely with a scalpel and then extracting it with buffer containing Triton X-100 (2%) and protease inhibitors (1.5 mM PMSF, 0.3 mM leupeptin, 20 μg/ml soybean trypsin inhibitor, 500 U/ml Trasylol>. The extract was reacted with an equal volume of rabbit anti-mouse IgG immunobeads (Biorad, Richmond, CA) for 2 hours at 37*C to remove mouse immunoglobulin. Control experiments in which a trace of radioiodinated mouse IgG was added to rissue extra-__ indicated that greater than 99% of the IgG was absorbed by the beads. The extract was then made 9M in urea and 10 mg/ml in dithiothreitol, boiled for 3 minutes and added to an equal volume of sample buffer for reduced Laemmli SDS- PAGE (10) . After boiling for 3 minutes, the mixture was centrifuged (10,000 RPM; -5 min) and the supernatant was subjected to SDS-PAGE (10%) and Western blotting. In each case, the same amount of tumor tissue was processed and the total protein loaded per lane of the gel was about the same. Following Western blotting, nitrocellulose membranes were reacted with monoclonal antibody specific for human fibrin (11) followed by

125 . .

I-affιnιty purified goat and anti-mouse IgG by the general method of Johnson et al. (12) as described previously (13) . The blot was then dried and exposed to Kodak X-Omat (XAR5) film (Eastman Kodak Co.,

Rochester, NY) in the presence of a Cronex intensifying screen (Dupont Co., Wilmington, DE) . Previous work has shown this antibody recognizes a polypeptide with M about 59,000 following SDS-PAGE and Western blotting of fibrin-containing samples (14) . Mice were infused with human fibrinogen in these studies to visualize fibrin in the tumor bed, because it appeared to show considerably greater immunoreactivity with the fibrin specific monoclonal antibody than murine fibrin based on control studies under these conditions. Lack of visualization of band(s) corresponding to the heavy and light chains of mouse immunoglobulins indicated that the removal of mouse immunoglobulin using the above procedure with anti-mouse IgG immunobeads was effective.

In certain experiments, mice were maintained on drinking water supplemented with the warfarin derivative 3(o--acetonylbenzyl)-4-hyc _^xycoumarin (0.7 mg/1) for three days prior to the TNF infusion (about 7 days after the meth A cells were injected) . Tumors grew to the same size in warfarin-treated and control animals. Prior to carrying out an experiment with the anticoagulated animals, a Factor X assay on the mouse plasma was performed (15) . Only animals with Factor X levels of less than 1% were used.

Second Series of Experiments: Isolation and

Characterization of the Biologically Active Fragment of Tumor Necrosis Enhancing Factor (TNEF)

Cell Culture and Tissue Factor Assay. Endothelial cells derived from human umbilical cord veins were prepared by the method of Jaffe (16) as modified by Thornton (17) . Experiments were carried out within 24 hrs of the cells achieving confluence in 9.6 cm wells (passages 1 to 5) . Cells were characterized as endothelial based on the presence of von Willebrand Factor antigen (18) , as described previously, and thrombomodulin activity (19) . Meth A sarcoma cells, provided by Drs. Hoffman and Old (7) , were grown in RMPI 1640 containing 10% fetal calf serum. Conditioned medium was obtained from meth A cells by placing them in serum-free medium (RPMI containing 10 mM HEPES, pH 7.4 20 μg/ml transferrin, 10 μg/ml insulin, 1 μg/ml poly yxin B and 5 mg/ml bovine serum albumin) for 48 hrs. Normal Balb/c mouse dermal fibroblasts were obtained from explant cultures of skin/subcutaneous tissue. They were maintained and used to prepare conditioned medium as described for meth A cells.

The tissue factor activity of endothelial cell monolayers was assayed using purified human Factors Vila (8nM) (provided by Dr. R. Bach, Mc-'-tt Sina Medical School, NY, NY) and X (1.5 μM) in serum-free medium at 23'C. At 8 minutes, a sample (0.2 ml) of th reaction mixture was removed and assayed for Factor X activity by monitoring hydrolysis of the chromogeni substrate benz-Ile-Glu-Gly-Arg-p-nitroanilide (20) . monoclonal antibody which blocks human tissue facto coagulant activity was provided by Dr. R. Bach.

The biologically active fragment of tumor necrosi enhancing factor (TNEF) may be prepared as follows:

A. In order to culture meth A sarcoma cells, a tissu explant was prepared. Adherent cells were grown t confluence in RPMI 1640 (Gibco) containing 10% feta calf serum and then split in 1:3 ratio using trypsin EDTA (Gibco) . Cells were then washed 3 times in calcium-magnesium-free Hanks' balanced salt solution and placed in serum-free medium (RPMI 1640 containin 10 mM HEPES (pH 7.4), 20 μg/ml polymyxin B and 5 mg/m bovine serum albumin) for 48 hours. Thereafter, th conditioned medium was harvested and incubated wit endothelial cell cultures as described below. B. Confluent cultures of human umbilical vein endo¬ thelial cells (10 cells/well grown by standard methods (5) ) were washed in calcium-magnesium-free Hanks' balanced salt solution and serum-free medium was added (1 ml/well). Then, either no further addition of Tumor Necrosis Factor (TNF) was made (o) , TNF alone (0.1 nM) was added (TNF), tumor-conditioned medium alone at 1/2 dilution was added (CM), or TNF (0.1 nM) in the presence of the indicated dilution (1/2, 1/10, 1/20) of conditioned medium was added for 7 hours at 37'C (TNF + CM). The final volume was 1 ml *■ ' each case. Monolayers were then washed and serum-free medium (0.5 ml) was added along with purified human Factors Vila (8 nM) and X (1.5 μM) at 23*C to assay for tissue factor activity (13) . A sample of the reaction mixture (0.2 ml) was removed at 8 minutes and assayed for Factor Xa activity by monitoring hydrolysis of the chromogenic substrate benz-Ile-Glu-Gly-Arg-p- nitroanilide (0.05 nM) (13). Results shown in Figure 1A are the mean and SEM of triplicate determinations. Factor Xa generation was the same as saline controls when tumor-conditioned medium was added directly to the Factor VIIA-X incubation mixture.

C. 1 ml of tumor-conditioned medium which is obtained as described in A above, was applied to a Sephadex G150 column (0.9 x 55 cm), eluted with a mixture of 10 mM HEPES (pH 7.4) and 0.1 M NaCl, and fractions (1.3 ml) were collected. Aliquots of the indicated column fractions, at a 1/4 dilution, were then incubated with endothelial cell monolayers in the presence of TNF (0.1 nM) for 7 hours at 37*C. The tissue factor assay was carried out as described in B above and Factor Xa formed over 7 minutes as shown in Figure B (mean +

SEM) . In Figure IB ^M, TNF" denotes cells incubated with TNF alone (0.1 nM) , "B" denotes cells incubated with column buffer alone, and "TNF + B^H denotes cells incubated with TNF (0.1 nM) and column buffer. Ear¬ lier column fractions were inactive in this assay. The gel filtration column was calibrated with standard proteins including ribonuclease (molecular weight (MW)=13,700 daltons), chymo-trypsinogen A (25,000), ovalbumin (43,000) and albumin (67,000).

The biologically active fragment of tumor necrosis enhancing factor (TNEF) was characterized in the following manner:

Exposure of cultured endothelium to meth A fibrosar- coma-conditioned medium, containing tumor necrosis enhancing factor secreted by meth A fibrosarcoma cells (Figure 1A) in the absence of TNF led to a small increase in procoagulant activity which was identified as tissue factor based on the Factor Vila-dependence of Factor Xa formation. TNF directly induced an increase in endothelial cell tissue factor activity. Addition of meth A fibrosarcoma conditioned medium, along with TNF, to endothelial cell cultures increased the procoagulant response in a dose-dependent manner. Similar enhancement of tissue factor activity was observed at several TNF concentrations. Because heat treatment of tumor-conditioned medium blocked its activity (Table 1) and all experiments were carried out in the presence of polymyxin (omission of polymyxin B had no effect) , endotoxin contamination is excluded to account for the observed effects. Similarly, no ₇- interferon, Interleukin 1, or TNF activity was detect¬ able in tumor-conditioned medium (4) . These data indi¬ cate that tumor cell product(s) , as distinct from Interleukin 1 or 7-interferon, can enhance the intro¬ duction of endothelial cell procoagulant activity in response to TNF. In other experiments it has been found that this activity is distinct from TGF (α and β) , FGF (acidic and basic) and Il-G.

Further characterization of the activity in meth A fibrosarcoma culture supernatants which enhanced endo¬ thelial cell tissue factor induction in response to TNF is shown by the experimental results in Table 1 and Figure IB. These results indicate that the activity Vas nondialyzable, heat and trypsin-sensitive, and on gel filtration chromatography eluted as a broad peak corresponding to a molecular weight range of 10,000 to 30,000 daltons (Figure IB).

Figure 2 shows the results of heparin Ultrogel (IBF Co.) chromatography of a meth A fibrosarcoma culture supernatant (the supernatant was collected in protein- free medium) . The left hand panel in Figure 2A demonstrates that following adsorption of TNEF activity to the resin, it was eluted with 0.5 M NaCl, with the bulk of the protein, not at higher salt concentrations where FGF and tumor cell-derived mitogens are found. The salt gradient elution in the right hand panel of Figure 2B shows elution of TNF activity towards the end of the salt gradient, well separated from much of the applied protein. Chro¬ matography of the pool eluted from the heparin Ultrogel column , on an FPLC (fast-pressure liquid chromatography) Mono Q column (Pharmacia, Inc., Piscataway, New Jersey) resulted in adsorption of TNEF activity to the resin with elution ocurring as a gradient of increasing salt was applied (elution occurred at about 0.4 M NaCl) (Figure 3). Again, TNEF activity was separated from the bulk of the applied protein. Thus, TNEF described herein is due to (a) Table 1

Characterization of activity in tumor-conditioned medi¬ um enhancing endothelial cell tissue factor in re- sponse to TNF*

Treatment Factor Xa formed (pmole/ml/well)

None 75 + 8

Heat 17 + 4

Acid 62 + 5

Base 77 + 7

Dialysis-retentate 71 + 8

Trypsin 21 + 3

No CM added 23 + 2

* Tumor-conditioned medium obtained from meth A sarco¬ ma cells was subjected to no treatment (none) or the indicated procedure: heat, 100*C for 10 minutes; acid, exposure to pH 2.0 for 5 minutes; base, exposure to pH 9.0 for 5 minutes; dialysis (molecular weight cut off 3,000-4,000); or trypsin, 50 μg for 1 hour at 37^*C (trypsin was inactivated at the end of the incubation period with^' diisopropylfluorophosphate; 1 mM diisopropylfluorophosphate alone had no effect on subsequent induction of tissue factor) . After each of these treatments, conditioned medium was added to endo¬ thelial cell cultures at a 1:4 dilution in the presence of TNF (0.1 nM) for 7 hours at 37*C. The tissue factor assay was carried out as described above and Factor Xa formed over 7 minutes as shown. "No CM added" indicates that no conditioned medium was added (TNF alone was present at 0.1 nM) . If neither conditioned medium nor TNF was added, Factor Xa formation was less than 5 pmole/ml/well. The mean and SEM of triplicat determinations are shown in Table 1. trypsin and heat-sensitive polypeptide(s) , molecular weight 10,000-30,000 daltons with a low affinity for Heparin Ultrogel. This activity is distinct from Interleukin 1, ₇-interferon, endotoxin, tumor necrosis factor, transforming growth factor (a and β) , Interleukin G, fibroblast growth factor, and multiple endothelial cell mitogens.

Thus, the results described above and taken together indicate that the biologically active fragment of tumor necrosis enhancing factor is (a) trypsin and heat- sensitive polypeptide(s) fragment with a molecular weight between about 10,000-30,000 daltons and further characterized by a low affinity for Heparin Ultrogel.

Third Series of Experiments: Purification of Tumor Necrosis Enhancing Factor

Cell Culture and Assays. Endothelial cells derived from human umbilical cord veins were prepared by the method of Jaffe (16) as modified by Thornton et al. (17) . Experiments were carried out within 24 hr of the cells achieving confluence and cultures were characterized as described previously. Meth A cells, provided by Drs. Hoffman and Old (Memorial Sloan- Kettering Cancer Center) (7) , were grown in RPMI 1640 containing 10% calf serum. Conditioned medium was obtained from meth A cells by placing them in serum- free medium (RPMI 1640 containing HEPES, lOmM, pH 7.4) for 12 hr at 37'C. Normal BALB/c dermal fibroblasts and bovine vascular fibroblasts were prepared from explant cultures of skin and aortic tissue, respectively. Induction of tissue factor in endothelial cultures was assessed by incubating tumor necrosis enhancing factor (TNEF) with cultures in serum-free medium (Medium 199 containing lOmM HEPES [pH=7.4], polymyxin B [1 μg/ml], bovine serum albumin-fatty acid free [5 mg/ml]) in the presence/absence of TNF for the indicated times. Tissue factor activity was determined by two-stage coagulant assay and, in a limited number of assays, vas also measured by studying formation of Factor Xa in the presence of purified Factors Vila (8 nM) and X (1.5 μM) by monitoring hydrolysis of the chromogenic substrate Benz-Ile-Glu-Gly-Arg-p-nitroanilide. Both of these assays have been previously described in this context (21) . Units of tissue factor activity were defined arbitrarily by assigning a value of 1U to an amount of meth A factor which induces an equivalent amount of tissue factor activity to that observed with 1 pg of purified human tissue factor reconstituted into phosphatidylserine/phosphatidylcholine vesicles (20:80) using the two-stage coagulant assay 80% phosphatidylcholine) . For these assays, tumor necrosis enhancing factor (TNEF) was incubated with endothelium alone or in the presence of TNF for 7 hr at 37'C. Assays were carried out with whole cells obtained in suspension following scraping from the dish with a rubber policeman (cell viability was >90% based on trypan blue exclusion) or, if indicated, with intact monolayers. Purified recombinant TNF (10 U/ml) was provided by Dr. Lomedico of Koffman-LaRoche (Nutley, NJ) . Tissue factor antigen content of endothelial cultures was determined (using an ELISA assay carried out by Dr. Jim Morrissey, Scripps Clinic and Research Foundation, La Jolla, CA) . Northern blots to determine levels of tissue factor mRNA were carried out b extracting total RNA from cells using the guanidiniu thiocyanate procedure (22) , electrophoretic fractionation of the RNA on a 1.2% agarose gel (23), and transfer to nitrocellulose. The cDNA probe for thrombomodulin (provided by Dr. E. Sadler, Wash U. , St. Louis, MO) was labelled using random hexamer labeling (Boehringer Mannheim random primed DNA labelling kit, Indianapolis, IN) . Hybridization of the cDNA probe to normal and meth A factor-treated RNA was performed at 42'C as described previously (23).

Tumor -.icrosis enhancing factor (TNEF) was studied fc_* its content of lipopolysaccharide (using the Li ulus amoebocyte assay carried out by Dr. Cerami, Rockefeller University, NY, NY) and assays for Interleukin-1 activity (using the D10 assay carried out by Dr. Killian, Hoffman LaRoche, Nutley, NJ) , Interleukin-6 activity (using the B cell proliferation assay carried out by Dr. May, Rockefeller University, NY, NY) , TNF activity (using the L929 assay carried out by Mr. DiPirro, SUNY at Buffalo, Buffalo, NY) and mitogenic activity (using the 3T3 mitogenesis assay carried out by Dr. Witte, Columbia University, NY, NY) . Antibody to murine Interleukin la was provided by Dr. Lomedico (Hoffmann-LaRoche) and antibody to murine TNFα was purchased from Genzyme (Boston, Mass) .

Tumor necrosis enhancing factor (TNEF) may be prepared as follows:

Meth A-conditioned medium, prepared as described above, was harvested from the cultures, sterile filtered (0.2 μm) , and after addition of octyl-ø-glucoside (final concentration 0.1%) and protease inhibitors (benzamidine, 5 mM, PMSF, 0.2 mM) , ammonium sulfate precipitation (80% saturation) was carried out. The ammonium sulfate pellet was then dissolved in tris (20 mM; pH 7.3)/octyl- -glucoside (0.1%) and dialyzed exhaustively against the same buffer. The retentate was applied to a Q-Sepharose column (2.5x20 cm; Pharmacia, Piscataway, NJ) equilibrated with tris (20 mM; pH 7.3) containing octyl-ø-glucoside and_. protease inhibitors (as above), and step-eluted with 0.1 M NaCl in the same buffer. The active fractions were pooled, dialyzed versus phosphate (25 mM; pH 6.8) containing octyl-j9-glucoside (0.1%) and applied to a FPLC Mono S column (HR 10'"-D; Pharmacia) equilibrated in the same buffer containing 4M urea. The column was eluted with an ascending salt gradient (0 to 0.5 M NaCl) and the active fractions were pooled, the pH was adjusted to 2.2 with trifluoroacetic acid and the sample applied to an FPLC ProRPC (HR 5/10; Pharmacia). The column was eluted with a linear gradient of increasing methanol concentration in the presence of trifluoroacetic acid (0.1%), the active fractions were pooled, dialyzed in the presence of SDS (0.1%) and concentrated b lyophilization. Samples were then dissolved in Laemmli sample buffer and preparative SDS-PAGE (12.5%) was carried out. Following electrophoresis, protein was visualized by silver staining (Biorad kit, Richmond, CA) or proteins were electroeluted using an ISCO Sampl 5 Concentrator (Lincoln, Nebraska) . For visualization o the isolated bands, aliquots of the samples wer lyophilized and re-run on 12.5% SDS-PAGE and staine (Biorad kit, Richmond, CA) . For determination o activity, SDS was removed after lyophilization b ^υ washing the samples in acetone-triethylamine-wate (9.5:0.5:0.5) three times. Prior to assay samples wer re-solubilized in tris (20 mM; pH 7.3)/octyl-ø glucoside (0.1%), chromatographed on Detoxi-gel column (Pierce, Rockford, 111.) and incubated with endothelia 5 cultures as described above. Protein concentrations were determined using a kit from Biorad, Richmond, CA. Tumor necrosis enhancing factor (TNEF) (2 ng) was treated with protease K (0.4 ng; Sigma) for 2 hr at 37*C, diluted and then added to endothelial cultures to test for induction of tissue factor. Control cultures were incubated with protease K alone and showed no induction of endothelial tissue factor.

Tumor necrosis enhancing factor (TNEF) may also be produced from other tumor lines. This tumor- necrotizing substance may also influence other properties of the vessel wall surrounding tumors in addition to TNF-induced expression of tissue factor. Tumor necrosis enhancing factor (TNEF) may also be involved in determining the immunogenicity of meth A sarcomas and may influence the fibrinolytic potential of the tumor vasculature.

Results

When normal and tumor-bearing mice were infused with high concentrations of TNF, 30 μg/animal or more, most of the animals died with thrombi in multiple organs, especially lung and liver, consistent with previous studies indicating the severe toxicity of TNF at these concentrations (24-25) . A TNF concentration of 10 μg/animal resulted in less marked systemic toxicity and thrombus formation. At 3 μg/animal, most animals survived without gross lesions in the normal vasculature, but hemorrhagic changes were observed in the tumors, indicating that this lower dose of TNF was triggering hemostatic abnormalities in the vascular bed of the meth A sarcoma, without widespread thrombohemorrhagic phenomena in other tissues. Fibrin deposition in the tumor vasculature after infusion of TNF (3 μg/animal) was assessed by measuring accumulation of radioactivity in the tumor in the presence of 125I fibrinogen. About ten times more radioactivity accumulated in the tumor bed of animals co-infused with TNF than in the meth A sarcomas of saline-infused controls. In contrast, other organs showed only a minimal increase in uptake of radioactivity after TNF infusion. Heat-treatment of

TNF, which prevents its b^ iing to cellular TNF receptors (8) , prevented the enhanced deposition of radioactivity in the tumor bed, indicating that TNF was the active agent. That activation of the coagulation mechanism with fibrin formation was responsible for the accumulation of radioactivity in the tumors is implied by the decreased incorporation of radioactivity in tumors from anticoagulated animals. Consistent with this hypothesis, after infusion of TNF Western blots of tumor extracts reacted strongly with a fibrin-specific monoclonal antibody. Treatment of animals with 3(α- acetonylbenzyl)-4-hydroxycoumarin considerably attenuated this band, and infusion of heat-treated TNF

(in place of native TNF) prevented appearance of the band corresponding to the fibrin-specific epitope.

These data indicate that activation of the coagulation mechanism leading to fibrin deposition in the tumor bed occurs in response to the TNF infusion, and fibrin could be visualized in the tumor vascular bed after TNF infusion. In animals infused with saline alone, there was no fibrin in the patent vessels of the tumors.

Thirty minutes after the infusion of TNF, however, fibrin was visible within the intravascular space closely associated with the endothelial surface. Fibrin was identified by the usual morphologic criteria (26) , and the characteristic ultrastructural periodicity of 21.09 nm. The scanning electron micrograph demonstrates fibrin strands apposed to the luminal endothelial cell surface, a situation never observed in control animals. Fibrin deposition was limited to the vessels in the tumor bed. At these early times, adherence of platelets and white cells on the vessel wall did not occur, and platelet thrombi were not seen in the spleen or other organs. This is consistent with the occurence of loc ized activation of coagulation within the tumor vasculature presumably initiated by endothelium. Two hours after the TNF infusion, occlusive thrombi with a prominent fibrin component were observed throughout the tumor. Concomitant with the appearance of these thrombi, unperfused areas were demonstrated in parallel studies carried out with Evan blue to visualize blood flow. At earlier times, 1 hour after TNF infusion, unperfused areas were focal, whereas 1-2 hours later large areas, up to 80% of the tumor, could not be reached by the dye. The presence of thrombi within the tumor vessels provides a potential link to Ϊ vitro studies demonstrating a procoagulant shift in endothelial cell hemostatic properties (1-6) .

Activation of coagulation with fibrin formation after the infusion of TNF was not unexpected, based on the results of previous studies showing that TNF could induce modulation of endothelial cell coagulant properties to favor clotting (1-6) . However, the localization of fibrin deposition to the tumor vascular bed was unexpected. One mechanism which could account for localized clot formation in response to TNF would be a vessel wall-dependent process accentuated in the tumor bed. The diffuse nature of fibrin deposition in the tumor vascular bed and its close association with the endothelial cell surface supported the hypothesis that a tumor-endothelial cell interaction might be involved. To examine this, supernatants of cultured meth A sarcoma cells (which had no intrinsic procoagulant activity) , generated under serum-free conditions, were incubated with endothelial cell monolayers and induction of tissue factor, a central initiator of coagulation (27) , was examined. Exposure of cultured endothelium to tumor-conditioneu medium alone led to at most a small, probably insignificant, increase in procoagulant activity. TNF, at a submaximal dose (0.1 nM) , induced an increase in endothelial cell procoagulant activity, as previously reported (4-5) . Addition of dilutions of tumor- conditioned medium, along with TNF, to endothelial cell cultures greatly increased the procoagulant response in a dose-dependent manner. The procoagulant activity was identified as -tissue factor based on the Factor Vila- dependence of Factor X activation and the inhibitory effect of anti-tissue factor antibody. Similar enhancement of tissue factor activity by tumor- conditioned medium was observed at several TNF concentrations. The experiments involve exposure of endothelium to TNF and tumor-conditioned medium for 7 hours, sufficient time for the maximal procoagulant response to develop. Enhanced tissue factor activity of endothelial cell monolayers was first noted at considerably earlier times. In contrast to the effect of medium derived from meth A cells, medium conditioned by non-transformed Balb/c (the same strain from which the meth A sarcoma was derived) fibroblasts had no effect on endothelial cell coagulant activity in the presence or absence of TNF. Preliminary characterization of the meth A-derived activity which augments the procoagulant response of endothelium to TNF indicates that it was nondialyzable (molecular weight cut off 3500) , heat-sensitive (100*c for 10 minutes) and trypsin-sensitive (50 μg for 4 hour at 37^*C). Gel filtration on Sephadex G150 demonstrated a broad peak of activity corresponding to a molecular weight range of about 10,000-30,000. Although the identity of the factor(s) responsible for the TNF- enhancing effect of tumor-conditioned medium is not clear, sensitivity to heat-treatment and the presence of polymyxin in all media (omission of polymyxin actually did not influence the results) make endotoxin contamination unlikely to account for the observed changes in endothelial cell TNF response. In this context, no Interleukin 1, fibroblast growth factor, TNF, gamma-interferon, or transforming growth factor- alpha activity was detectable in tumor conditioned medium. Because TGF. has been found in most mammalian ^β tissues and cell lines studied (34-36) , it was important to determine whether TGF^ or TGF ₂ had the capacity to enhance the endothelial cell procoagulant response to TNF. These studies employed human and porcine TGF_/3 because of the availability of purified material and the considerable sequence homology betweeen TGF from different species (murine, porcine and human) (34-36) . Neither TGF_/9₁ nor TGF0₂ (at concentrations up to 500 pM) enhanced endothelial cell procoagulant activity in the presence of TNF, suggesting that the activity in meth A culture supernatants was distinct from TGF .

Previous studies have indicated that culture supernatants from meth A cells can enhance endothelial tissue factor activity induced by TNF (37). This led us to design a purification procedure to identify the molecule(s) responsible, starting with serum-free, meth A-conditioned medium and using induction of endothelial tissue factor activity as an assay system. Culture supernatants were concentrated by ammonium sulfate precipitation (80% saturation) and then dialysis was carried out (Table 2) . The dialysate was adsorbed to Q-Sepharose (at pH 7.3), removing -90% of the total protein, and the supernatant was dialyzed and applied to a Mono S column (Figs. 4A and 4B) . Activity which enhanced endothelial tissue factor was identified, the fractions were pooled, dialyzed and applied to a reverse phase FPLC column (ProRPC) . Using an ascending methanol gradient, the activity was eluted from ProRPC at -50%. Silver stained SDS-gels of material eluted from ProRPC still showed a complex pattern , but electroelution of the gels demonstrated that activity which induced endothelial tissue factor was localized to a single region, the area at Mr -44,000 Da, which corresponded to a faint band on the gel. This led us to carry out preparative scale SDS-PAGE followed by electroelution with the pool from reverse phase chromatography. When material eluted from this gel was re-run on another SDS-gel, it migrated as a single band, with Mr -44,000 Da (non-reduced) and Mr-70,000 Da (reduced) , and biologic activity was detected in the corresponding areas of the gel (extensive dialysis, presumably allowing re-folding and disulfide bond formation, was required to obtain activity in sample from reduced gels) . Gel-eluted material was quite heat-resistant (98*C for 10 min), but could be inactivated by protease K treatment (the Mr -44,000 Da band on gels disappeared, in parallel, indicating that digestion had Table 2. Purification of tumor necrosis enhancing factor (TNEF)*.

Step Protein Activity Specific Yield Purific¬ Units+ Activity ation (for

5 (xlO^J) (U/mgxl0³) that step)

Meth A- condition- ed medium 79 mg 13 0.16 100%

Ammonium .sulfate precipita¬ tion 72 mg 92% 1

Fast Q 8 mg 65% 6

Mono S 2 mg 25% 15

^•jfroRPC 20 μg 11% 4

Preparative

SDS-Page and gel elution 10 ng 0.1 1000 0.8% 14

2^β The starting material was 1 liter of meth A-conditioned serum-free medium.

+ Units are defined arbitrarily: one unit is the amount of tissue factor activity induced in endothelial cells by tumor necrosis enhancing factor (TNEF) which is equivalent to 1 pg of purified tissue factor reconstituted into phospholipid 2gesicles (see Materials and Methods) .

30

35 probably taken place) . HPLC gel filtration demon¬ strated a major peak of activity corresponding to molecular weight -45-50,000 Da, indicating that tumor necrosis enhancing factor (TNEF) is present in culture supernatants probably as a monomer (data not shown) . Putting together the steps in this isolation procedure (Table 2) , a net purification of about 5000-fold was obtained and 10 nanograms was prepared from 1 liter of conditioned medium.

To further characterize the properties of tumor necrosis enhancing factor (TNEF) several approaches were used. Heparin-Sepharose chromatography demonstrated complete elution of adsorbed tumor necrosis enhancing factor (TNEF) activity by 0.5 M NaCl (i.e. at considerably lower ionic strength than fibroblast growth factor and a group of tumor-derived angiogenesis factors) and isoelectric focussing showed two major peaks of activity corresponding to pH-6.8 and 7.2. The activity also adsorbed to Concanavalin A- Sepharose, suggesting that it was a glycoprotein. Bioassays for TNF (L929) and studies with anti-murine TNFα antibody indicated the factor was distinct from TNF. Similar experiments assessing Interleukin 1 (using the D10 assay and anti-murine Interleukin 1-α) , Interleukin 6, and mitogenic activity for NIH 3T3 cells (which would reflect the presence of platelet-derived growth factors and/or certain other growth factors) , were also negative. Pilot studies have shown that neither transforming growth factor-^, a product of a variety of malignant cells, nor epidermal growth factor have similar activity when compared with tumor necrosis enhancing factor (TNEF) in terms of induction o endothelial procoagulant activity. Finally, samples of tumor necrosis enhancing factor (TNEF) were chromatographed over Detoxi-gel columns to remove possible contaminating lipopolysaccharide (there was no lipopolysaccharide detectable in the Limulus amoebocyte assay) and incubation of tumor necrosis factor (TNEF) with endothelium was carried out in the presence of polymyxin B. These findings, taken together with the other data cited above, indicate that it is highly unlikely that lipopolysaccharide contamination was responsible for the induction of endothelial procoagulant activity in the presence of tumor necrosis enhancing factor (TNEF) .

Although purified tumor necrosis enhancing factor (TNEF) had no procoagulant activity by itself, when it was incubated with endothelial cultures, induction of tissue factor synthesis was observed. Functional studies showed time and dose-dependent induction of tissue factor. Tissue factor activity was evident by 3 to 4 hours but did not decline until 24 hours, in contrast to the more rapid decline of endothelial tissue factor induced by TNF or Interleukin 1. Production of tissue factor procoagulant activity by tumor necrosis enhancing factor (TNEF) could be blocked by the addition of cycloheximide to endothelial cultures and the half-maximal effect occurred at a concentration of about 1 to 5 ng/ml. The rise in tissue factor activity was paralleled by an increase in tissue factor antigen. When low concentrations of TNF were co-incubated with cultures along with tumor necrosis enhancing factor (TNEF) , enhancement of tissue factor induction to levels greater than that observed with either agent alone were observed.

Discussion Although thrombosis is a major cause of morbidity and mortality, mechanisms involved in the pathogenesis of localized intravascular clot formation are largely uncharacterized. Models of intravascular clot formation derived from studies of the hemostatic plaque, rapid thrombus formation following contact of plasma factors and cellular elements of the blood with subendothelial cell components (29) , present a picture of thrombosis in which endothelial cell denudation is the critical initiating step. The results represented here indicate that TNF, an endogenously produced mediator of the host response, can selectively induce intravascular clot formation in the tumor vasculature of mice bearing meth A sarcomas in the presence of a viable endothelial monolayer and delineate another model of localized thrombosis. Following infusion of TNF, radioiodinated fibrinogen/fibrin accumulated in the tumor. The frank deposition of fibrin in the tumor vascular bed -indicates that activation of coagulation with clot formation was clearly involved. In support of this hypothesis, coumadin, an anticoagulant which prevents carboxylation of the vitamin K-dependent coagulation proteins and thereby down-regulates their effective interaction with membrane surfaces (30) , considerably decreased accumulation of radiolabel and appearance of the fibrin-specific epitope in the tumor bed. Furthermore, the intravascular localization of fibrin deposition with evidence of occlusive thrombi was noted in all animals studied up to 2 hours after the TNF infusion. In contrast, neither consistent focal uptake of 125I-fιbrιnogen/fιbrιn comparable in magnitude to that in the tumor bed nor thrombus formation was seen in the normal vasculature of tumor- bearing or normal mice. This indicates that th thrombogenic effects of TNF were being targeted to the tumor vascular bed.

A central question which evolves from these studies concerns possible mechanisms through which host and tumor-mediated process could locally enhance the effects of systematically infused TNF. Elaboration of cytokines/angiogenic agents and/or other factors, such as changes in blood flow in the tumor bed, could certainly be involved. The experiments presented suggest that tumor cells elaborate a distinct mediator(s) that potentiates the coagulant response of endothelium to TNF. Because TNF has been shown to induce a shift in endothelial cell coagulant properties favoring activation of coagulation (1-6) , enhancing this effect locally in a particular vascular bed could constitute a potent thrombogenic stimulus. The mechanism(s) through which tumor factor(s) modulate the responses of endothelium to TNF remains unclear. Pilot studies examining the affinity and number of TNF binding sites on endothelium after exposure to tumor- conditioned medium have not demonstrated a change in the binding parameters of 125I-TNF. This suggests that the effect of tumor cell products on endothelial cell reactivity to TNF is mediated distal to ligand-receptor interactions.

Locally-induced clot formation, occurring rapidly (initiated within 30 minutes of cytokine infusion) , is only one example of a focal TNF effect in tumors. By 4-8 hours after TNF infusion, tumor vessels are packed with leukocytic cells. Well-described toxic morphologic changes in the tumor cells and hemorrhagic necrosis also take place at later times (24,31). Some of these local changes, such as adhesion of leukocytes to endothelium (32) , may also be selectively enhanced in the tumor bed due to the effect of factor(s) in the tumor microenvironment such as that which enhances the TNF-induced procoagulant response. Although the relationship between early activation of coagulation in the tumor bed and tumor necrosis is unclear, pilot studies using mice anticoagulated with the same coumadin compound employed in these studies have shown attenuated tumor necrosis, based on reduction in tumor weight; by as much as 50% in anticoagulated compared with control mice. In this context, selective deposition of fibrin, a well-recognized stimulant of the inflammatory response (33) , in the tumor bed could be a factor promoting subsequent leukocytic infiltration. Further studies with a variety of anticoagulants and tumors will be required to gain further insight into this tissue.

The studies presented in this report indicate that intravascular ^• clot formation within the tumor vasculature is part of the early response of meth A sarcomas to TNF. The localization of clot formation to the tumor bed provides a potentially important model for examining mechanisms underlying intravascular thrombus formation.

The tumor microenvironment is the result of host and tumor-derived factors which are concentrated in the tumor bed. The studies described here demonstrate that meth A tumor cells elaborate a polypeptide, tumor necrosis enhancing factor (TNEF) , which induces endothelial tissue factor and enhances the procoagulant response to TNF. Tumor necrosis enhancing facto (TNEF) appears to be distinct from other cytokines which have been shown to induce endothelial tissu factor activity (such as TNF and Interleukin-1) , and is a potentially important component of the tumor microenvironment contributing to alterations in vascular function observed in the tumor bed. In this context, studies demonstrating that tumor necrosis activating factor (TNEF) enhances induction of endothelial tissue factor in response to low concentrations of TNF (8) emphasize the importance of considering the interaction of locally produced mediators as therapeutic agents. Although mechanisms responsible for the interaction of tumor necrosis enhancing factor (TNEF) and TNF are not yet clear, pilot radioligand binding studies have shown that meth A factor increases the affinity of endothelium for TNF, providing a possible basis for apparent enhancement of TNF-induced endothelial tissue factor production in the presence of tumor necrosis enhancing factor (TNEF) .

The studies presented here represent an initial characterization of murine tumor necrosis enhancing factor (TNEF) . Further work will be required to demonstrate the distribution and effects of tumor necrosis enhancing factor (TNEF) in neoplastic and normal cells/tissues. Initial studies suggest that tumor necrosis enhancing factor (TNEF) activity is not present in nontransformed murine and bovine fibroblasts, but is found and could be partially purified from several tumors which are sensitive to TNF (including the SA-1 murine tumor cell line and the FO-1 human melanoma cell line) . In contrast, tumor necrosis enhancing factor (TNEF) activity was not found in a murine tumor recently identified at the Gray Cancer Laboratories which is resistant to TNF. Although it is too soon to draw any conclusions concerning the effect of tumor necrosis enhancing factor (TNEF) on TNF- mediated perturbation of tumor vasculature, these findings indicate that further studies of this and other tumor-derived mediators will provide insights into mechanisms underlying the unique properties of vessels in the tumor bed.

Summary

Recent studies have indicated that TNF can promote activation of the coagulation mechanism by modulating coaguxant properties of endothelial cells. In this report, we demonstrate that infusion of low concentrations of TNF (3 μg/animal) into mice bearing meth A fibrosarcomas leads to localized fibrin deposition with formation of occlusive intravascular thrombi in close association with the endothelial cell surface. Studies with 125I-fibrinogen showed ten-fold enhanced accumulation of radioactivity in the tumor within 2 hours after TNF infusion. Western blots of tumor extracts subjected to SDS-PAGE and visualized with a fibrin-specific monoclonal antibody indicated that fibrin forms in the tumor after the TNF infusion.

Electron microscopic studies demonstrated fibrin strands, based on the characteristic 21 nm periodicity, which appeared to be adherent to the endothelial cell surface. Further ultrastructural studies indicated that fibrin formation, first evident within 30 minutes of the TNF infusion, led to occlusive thrombi limited to the tumor vascular bed, i.e., not in the normal mouse's vasculature within 2 hours, and was associated with an 80% reduction in tumor perfusion based on studies with Evans blue. In view of previous work concerning TNF induction of endothelial cell procoagulant activity, the hypothesis that tumor cell products prime the responses of endothelium to this cytokine was tested. Supernatants of cultured meth A fibrosarcomas obtained in serum-free conditions, which had no intrinsic procoagulant activity, considerably enhanced tissue factor induction in endothelium in response to submaximal concentrations to TNF. The factor(s) in the tumor conditioned medium appeard to be distinct from Interleukin 1, fibroblast growth factor, interferon, TNF, endotoxin, TGF-α and TGF-0. These studies delineate a novel model of localized clot formation in which thrombosis is initiated by a pathophysioiogic mediator, TNF, and provides an opportunity to examine mechanisms in the microenvironment directing clot formation to the tumor vascular bed.

Previous studies have shown that an early component of the vascular response to TNF includes intravascular clot formation localized to the neoplastic lesion which is closely associated with the endothelial surface and leads to reduced tumor blood flow. This has led us to identify tumor-derived mediators which enhance endothelial procoagulant activity and the cellular response to TNF, using murine meth A fibrosarcomas as a model system. A heat-stable, protease K-sensitive polypeptide with an apparent molecular weight of about 44,000 daltons on a non-reduced SDS-polyacrylamide gel and a molecular weight of about 70,000 daltons on a reduced SDS-polyacrylamide gel was purified about 5000- fold from serum-free culture supernatants of meth A cells by sequential Q-Sepharose, Mono S, reverse phase, and SDS-PAGE procedures. Based on immunologic criteria, biologic activity, and other molecular properties, tumor necrosis enhancing factor (TNEF) appears to be distinct from other cytokines and growth factors. This protein induced transcription of the tissue factor gene and expression of procoagulant activity by endothelium (half-maximal effect at about 1 ng/ml) . Furthermore, pre-incubation of tumor necrosis enhancing factor (TNEF) enhanced tissue factor induction in response to subsequent exposure to TNF. These data indicate that meth A tumors elaborate a potentially unique molecule which can alter hemostatic properties of the vessel wall, potentially modulating reactivity of the tumor vasculature to host response mediators and other agents.

References

1. Beutler, B. , and Cerami, A. 1986. Cachectin and tumor necrosis factor as two sides of the same biological coin. Nature 32:584

2. Old, L. 1986. Tumor necrosis factor. Science 230:630.

10 3. Tracey, K. , Beutley, B. , Lowry, S., Merryweather, J., Wolpe, S., Milsark, I., Hariri, R. , Fahey, T. , Zentella, A., Albert, J. , Shires, G. and Cerami, A. 1987. Shock and tissue injury induced by recombinant human cachectin. Science 234:470.

15

4. Nawroth, P. and Stern, D. 1986. Modulation of endothelial cell hemostatic properties by TNF. J. Exp. Med. 164:70.

_ 5. Bevilacqua, M. , Pober, J., Majeau, G., Fiers, W. , Cotran, R. , and Gimbrone, M. 1986. Recombinant TNF induces procoagulant activity in endothelium. PNAS (USA) 83:4533.

__ 6. Le, J., and Vilcek, J. 1987. Biology of Disease: 25

TNF and IL-1: Cytokines with multiple overlapping biological activities. Lab. Invest 56:234.

7. Old, L. , Benacerraf, B. , Clarke, D. , Carswell, E. 30 and Stockert, E. 1961. Role of the reticuloendothelial system in the host reaction to neoplasia. Cancer Res. 21:1281.

35 8. Nawroth, P., Bank, I., Handley, D., Cassimens, J. , Chess, L. and Stern, D. 1986. Tumor necrosis factor interacts with endothelial cell receptors to induce release of IL-1. J. Exp. Med. 163:1363.

9. David, G. and Reisfeld, R. , 1974. Protein iodination with solid state 1ac operoxidase. Biochem. 13:1014.

10. Laemroli, U. 1970. Cleavage of structural proteins during the assembly of the head of bacteriophage T. Nature 227:680.

11. Hui, K. , Haber, H. and Matsueda, G. 1983. Monoclonal antibodies to a synthetic fibrin-like peptide bind to human fibrin but not fibrinogen. Science 222:1129.

12. Johnson, D. , Gautsch, J. , Sportsman, J. and Elder, J. 1984. Improved technique utilizing non-fat dry milk for analysis of proteins and nucleic acids transferred to nitrocellulose. Gen. Anal. Tech. 1:3.

13. Stern, D. , Brett, J. Harris, K. and Nawroth, P. 1986. Participation of endothelial cells in the protein C-protein S pathway: the synthesis and release of protein S. J. Cell Biol. 102:1971.

14. Chen, F. , Haber, E. , and Matsueda, G. 1987. Characterization of fibrinogen and fibrin degradation that contains the -chain (1-7) epitope. Clin. Res. 35:422A. 15. Bajaj, P. and Mann. K. 1973. Simultaneous purification of bovine prothrombin and factor X. J. Biol. Chem 248:7729.

16. Jaffe, E. Culture and identification of large vessel endothelial cells. In E. Jaffe, ed. , Biology of Endothelial Cells., E. Jaffe, editor. Martinus Nijhoff, Boston. 1-13.

17. Thornton, S., Mueller, S. and Levine, E. 1983.

Human endothelial cells: use of neparin in cloning and long-term serial cultivation. Science 222:623.

18. Jaffe, E., Hoyer, L. and Nachman, R. 1973. Synthesis of antihemophilic factor antigen by cultured human endothelial cells. J. Clin. Invest. 52:2757.

19. Esmon, G. 1984. Protein C. Prog. Hemostasis Thromb. 7:25.

20. Van Dieijan, G. , Tars, G. , Rosing, J. and Hemker, H. 1981. The role of Factor VIII in the activation of Factor X. J. Biol. Chem. 256:3433.

21. Nawroth, P., Stern, D. , Kisiel, W. , and Bach, R. 1985. Cellular requirements for tissue factor generation by perturbed bovine aortic endothelial cells in culture. Thrombosis Res. 40:677-691.

22. Chirgwan, J., Przbyla, A., McDonald, R. , and

Rutter, W. 1979. Isolation of biologically active ribonucleic acid from sources enriched in ribonuclease. Biochemistry 18:5294-5299. 23. Yancopoulos, G. and Alt, F. 1985. Developmentally controlled and tissue-specific expression of unrearranged V gene segments. Cell 40:271-281.

24. Beutler, B. , Milsark, I., and Cerami, A. 1985. Cachectin/TNF: Production, distribution and metabolic fate in vivo. J. of Immunol. 135:3972.

25. Asher, A., Mule, J. , Reichert, C, Shiloni, E. and Rosenberg, S. 1987. Studies on the anti- tumor efficacy of systemically administered rTNF in vivo. J. of Immunol 138:963.

26. Bale, M. and Mosher, D. 1986. Effects of thrombospondin of fibrin polymerization and structure. J. Biol. Chem 261:862.

27. Nemerson, Y. and Bach, R. 1982. Tissue Factor revisited. Prog. Hemostasis Thromb. 6:237.

28. Ruff, M. and Gifford, G. 1980. Purification and physicochemical characterization of rabbit TNF. Immunol. 125:1671.

29. Robbins, S., Ramzi, C, and Kumar, V. 1984. Thrombosis. In Healing and Ripari, eds., Pathologic Basis of Disease, Saunders Co. , Philadelphia, pp. 91-92.

30. Suttie, J. ed. , Vitamin K Metabolism and Vitamin K-Dependent Proteins, University Park Press, Baltimore, MD. 31. Carswell, E. Old, L. , Kassel, R. , Green, S . , Fiore, N. and Williamson, B. 1975. An endotoxin- induced serum factor that causes necrosis of tumors. PNAS (USA) 72:3666.

32. Bevilacqua, M. , Wheeler, E. , Pober, J., Fiers, W. , Mendrick, R. , Contran, M. , Gimbrone, M. 1987. Endothelial cell-dependent mechanisms of leukocyte adhesion. In Movat, H. ed. Leukocyte Emigration and its Sequelae. S. Karger, Basel, p. 79.

33. Robbins, s., Ramzi, C. , and Kumar, V. 1984. In Healing and Repari, eds., Pathologic Basis of Disease, Saunders Co., Philadelphia, pp. 76-77.

34. Sporn, M., Roberts, A., Wakefield, L. and de Crombugghe. 1987. Some recent advances in the chemistry and biology of transforming growth factor-beta. J. Cell Biol. 105:1039-1045.

35. Cheifetz, S., Weatherbee, J. , Tsang, M. , Anderson, J., Mole, J., Lucas, R. and Massague, J. 1987. The transforming growth factor-beta system, a complex pattern of cross-reactive ligands and receptors. Cell 48:409-415.

36. Massague, J. 1987. The TGF-beta family of growth and differentiation factors. Cell 49:437-438.

37. Nawroth, P., Handley, D. , Matsueda, G., DeWaal, R. , Blohm, H., and Stern, D. 1988. J. Exp. Med. 168:637-647.

Claims

What is claimed is:

1. A purified tumor necrosis enhancing factor.

2. A purified tumor necrosis enhancing factor of claim 1 comprising a polypeptide characterized by an apparent molecular weight between about 40,000 and about 50,000 daltons under nonreducing conditions and by an apparent molecular weight between about 65,000 and about 75,000 daltons under reducing conditions, and by isoelectric focussing peaks at about 6.8 and about 7.2.

3. A purified tumor necrosis factor of claim 2 further characterized by loss of activity upon treatment with protease K, by an affinity for Heparin-Sepharose from which the factor may be eluted with about 0.5 M NaCl, by the ability to bind to a reverse phase FPLC-ProRPC column, and by the ability to be eluted from a reverse phase FPLC-ProRPC column to which the factor is bound in an ascending methanol gradient at about 50%, by the ability to migrate as a single band on an SDS-polyacrylamide gel, and by adsorption to Concanavalin A-Sepharose.

4. A purified tumor necrosis enhancing factor of claim 2 characterized by an apparent molecular weight of about 44,000 daltons on a non-reducing SDS-polyacrylamide gel and an apparent molecular weight of about 70,000 daltons on a reducing SDS- polyacrylamide gel.

5. A purified, biologically active fragment of the purified tumor necrosis enhancing factor of claim 2 characterized by an apparent molecular weight between about 10,000 and about 30,000 daltons, by the loss of activity upon treatment with trypsin or heat, by an affinity for Heparin Ultrogel from which the fragment may be eluted with about 0.5 M NaCl, by the ability to bind to a Mono Q column, and by the ability to be eluted from a Mono Q column to which the fragment is bound in an ascending salt gradient at about 0.4 M NaCl.

6. A purified, biologically active fragment of claim 5 characterized by an apparent molecular weight of about 20,000 daltons.

7. A purified DNA molecule encoding the tumor necrosis enhancing factor of any of claims 1 to 4.

8. A purified DNA molecule encoding the biologically active fragment of tumor necrosis enhancing factor of any of claim 5 or 6.

9. A cDNA molecule of claim 7.

10. A cDNA molecule of claim 8.

11. An isolated genomic DNA molecule of claim 7.

12. An isolated genomic DNA molecule of claim 8.

13. A pharmaceutical composition comprising an amount of the tumor necrosis enhancing factor of any of claims 1 to 4 effective to enhance the tumor inhibitory activity of a cytotoxic agent and a pharmaceutically acceptable carrier.

14. A pharmaceutical composition comprising an amount of the biologically active fragment of tumor necrosis enhancing factor of any of claim 5 or 6 effective to enhance the tumor inhibitory activity of a cytotoxic agent and a pharmaceutically acceptable carrier.

15. A pharmaceutical composition of claim 13 further comprising an amount of tumor necrosis factor effective to inhibit tumor growth.

16. A pharmaceutical composition of claim 14 further comprising an amount of tumor necrosis factor effective to inhibit tumor growth.

17. A method of inhibiting the growth of tumor cells which comprises contacting the cells with an effective tumor inhibitory amount of a cytotoxic agent in.the presence of an amount of the tumor necrosis enhancing factor of any of claims 1 to 4 effective to enhance the tumor inhibitory activity of the cytotoxic agent.

18. A method of inhibiting the growth of tumor cells which comprises contacting the cells with an effective tumor inhibitory amount of a cytotoxic agent in the presence of an amount of the biologically active fragment of tumor necrosis enhancing factor of any of claim 5 or 6 effective to enhance the tumor inhibitory activity of the cytotoxic agent.

19. A method of claim 17, wherein the cytotoxic agent is a polypeptide tissue factor having tumor inhibitory activity.

20. A method of claim 19, wherein the polypeptide tissue factor is tumor necrosis factor.

21. A method of claim 18, wherein the cytotoxic agent is a polypeptide tissue factor having tumor inhibitory activity.

22. A method of claim 21, wherein the polypeptide tissue factor is tumor necrosis factor.

23. A method of claim 17, wherein the inhibition of tumor cell growth is localized within the vascular system.

24. A method of claim 18, wherein the inhibition of tumor cell growth is localized within the vascular system.

25. A method of inhibiting the growth of tumor cells which comprises contacting the cells with an effective amount of the composition of claim 13.

26. A method of inhibiting the growth of tumor cells which comprises contacting the cells with an effective amount of the composition of claim 14.

27. A method of treating a subject having a tumor which comprises administering to the subject an amount of the pharmaceutical composition of any of claim 13 or 15 effective to necrotize the tumor.

28. A method of treating a subject having a tumor which comprises administering to the subject an amount of the pharmaceutical composition of any of claim 14 or 16 effective to necrotize the tumor.

29. A method of preparing the tumor necrosis enhancing factor of any of claims 1 to 4 which comprises recovering the factor from tumor cells and purifying the factor so recovered.

30. A method of preparing the biologically active fragment of the tumor necrosis enhancing factor of any of claim 5 or 6 which comprises recovering the fragment from tumor cells and purifying the fragment so recovered.

31. A method of preparing the tumor necrosis enhancing factor of any of claims 1 to 4 which comprises expressing the DNA of claim 7 in a suitable host so that the factor is produced, recovering the factor so produced and purifying the factor so recovered.

32. A method of preparing the biologically active fragment of the tumor necrosis enhancing factor of any of claim 5 or 6 which comprises expressing the DNA of claim 8 in a suitable host so that the fragment is produced, recovering the fragment so produced, and purifying the fragment so recovered.