CA2003361C - Human vascular permeability factor - Google Patents

Abstract

A process is disclosed for the production of a novel human vascular permeability factor in vitro comprising growing cells derived from the human histiocytic lymphoma cell line U-937 in nutrient culture medium at about 35° to 38°C for a sufficient time to elaborate vascular permeability factor and isolating the resulting vascular permeability factor from the spent cells or the cell culture conditioned medium.

Description

-1- 07-21(525).A
HUMAN VASCULAR PERMFsABILITY FACTOR
Background of the Invention This invention relates to a novel human vascular permeability factor and its in vitro production in a highly purified form from the human histiocytic lymphoma cell line; U-937.
Vascular permeabilii~y factors (VPFs) are proteins originally obtained from a variety of tumors which cause a rapid and reversible increase in blood vessel permeability when nanogram amounts are injected under the skin of a warm blooded mammal.
VPF activity has been found in tumor ascites fluid from guinea pigs, hamsters and mice and is secreted by these tumors and a variety of tumor cell lines in vitro according to Senger et al., Science 219, 983-985 (1983).
In U.S. Patent 4,45~~,550, a purified VPF is described which has the following characteristics:
(a) in an aqueous solution (0.01 M Na3P04, pH 7) whose concentration of lVaC1 is varied linearly, VPF is eluted from a heparin-~Sepharose chromatography column in a peak centered at ~0.4 NaCl;
(b) in an aqueous solution of Na3P04, pH
7.0, whose concentration is varied linearly, VPF is eluted from a hydroxylapatite column in a peak centered at 0.25 M Na3P04; and -- ~ 0 0 3 ;3 6 1 -2- 07-21(525)A
(c) when subjected to SDS gel electrophoresis in a polyacrylamide slab gel (0.375 M
tris-HC1, pH 8.8, 0.1% SDS) at 35 milliamps and 4°C., VPF is localized to a region .corresponding to a molecular weight between 34,000 and 45,000 daltons.
VPF of the foregoing characteristics was thus purified about 1800 fold from serum-free conditioned medium of guinea pig tumor cell culture or 10,000 fold from ascites fluid by a series of steps consisting of:
(a) affinity chromatography with a column of heparin-*Sepharose~
(b) chromatography with a column of hydroxylapaptite; and (c) sodium dodecylsulfate/
polyacrylamide gel electrophoresis.
According to said patent, as :little as 200 ng (5 x 10 12 moles) of this purified VPF increased the vascular permeability equivalE:nt to 1.25 Ng (4 x 10 9 moles) of histamine. Histamine is a standard permeability mediator describf:d by Miles and Miles, J. Physiol. 118, 228-257 (195:?). The VPF is said to have therapeutic value insofar as it enables blood nutrients to reach tissue with increased need for nutrients, as in wound healin<~.
*Trade-mark s' ~0 03361 -3- 07-21(525)A
According to Folkman and Klagsbrun, Science 235, 442-447 (1987), VPF causes leakage of proteins, including fibrinogen, from blood vessels, thereby initiating the formation of a fibrin gel which, in turn, may play a role in angiogenesis. See also Dvorak et al., J. Immunol. 1.?2(1), 166-174 (1979);
Dvorak, NEngl. J. Med. 315, 1650-1659 (1986);
Kadish et al., Tissue & Cell _11, 99 (1979); and Dvorak et al., J. Natl: Cancer Inst. 62, 1459-1472 (1979).
. A method of stimu~Lating endothelial cell growth is known which compr~.ses subjecting said cells to a growth stimulating amount of a highly Purified VPF. The highly purified VPF derived from guinea pig tumor cells has t:he following characteristics:
(a) it has a Mr about 34,000 - 40,000 as determined by sodium dodecyl-sulfate polyacrylamide gel electrophoresis (SDS/PAGE);
(b) it is a disulfide-linked protein dimer;
(c) it has a N-terminal amino acid sequence as.follows:
AlaProMetAlaGluGlyGluGlnLy;sProArgGluValValLys PheMetAspValTyrLysArgSerTy:rCysArgProIleGluMet LeuValAspIlePheGln; and rR X003361 -4- 07-21(525)A
(d) it exhibits substantial mitogenic activity to endothelial cells in culture.
The foregoing highly purified guinea pig VPF, also referred to as gVPF,, was isolated from , serum-free conditioned culturE~ medium of guinea pig tumor cells in a series of stf:ps comprising:
(a) affinity chromatography of said conditioned culture medium with a column of heparin-Sepharose CL-6B;
(b) cation exchange chromatography of the VPF active fractions from said affinity chromatography with a TSK SP-5-~PW column;
(c) high performance: liquid chromatography (HPLC) of the VPF
active fractions from said cation exchange chromatography with a Vydac C4 reversed phase HPLC column;
and (d) HPLC of the VPF active fractions from said C9 HPhC with a Vydac C18 reversed phase EIPLC column.
A method of producing antibodies against gVPF is known in which certain peptide fragments .25 of gVPF are used as immunogens .
Lobb et al., Int. J. Cancer 36, 473-478 (1985), describe a partially purified a VPF from a human adenocarcinoma cell line: HT-29 having a -5- 07-21(525)A
molecular weight of 45,000. 'this VPF, however, does not bind to immobilized heparin as does the VPF
derived from guinea pig tumor cell material by Senger and Dvorak.
Senger et al., Cancer Res. 46, 5629-5632 (1986), describe the production of VPF from a variety of human tumor cell lines, namely human osteogenic sarcoma, bladder sarcoma, cervical carcinoma and fibrosarcoma cell lines. However, none of these human cell lines were found to be as active as the guinea pig cell line 10 for the producton of VPF.
The finding of an improved human cell line and method for the production of human VPF therefrom in a highly purified form would have significant advantages.
Brief Description o:f the Invention The present inventors have investigated numerous human cell lines for the production of VPF
by in vitro cell culture methods but most of them have been eliminated as unsuitable candidates in view of their relatively poor VPF production or failure to produce VPF.
A cell line that ha;a unexpectedly been found by the inventors to have good growth characteristics in cell cultu:re and to be able to elaborate the desired human V7?F in suitable quantities is the human histiocytic lymphoma cell line U-937.
This cell line was originally established from cells from the pleural effusion of a patient with diffuse histiocytic lymphoma as reported by Sundstrom and Nilsson, Int. J. Cancer 17, 565-577 (1976). These cells are widely distributed as evidenced by publications and are also readily available to the public in an unrestricted culture deposit from the ,.

-6- 07-21(525)A
American Type Culture Collection, Rockville, Maryland, under accession number ATCC fRL 1593. Further background on these cells can. be had by reference to J. Exp. Med. 143, 1528-1533 (1976); Nature 279, 328-331 (1979); and J. Immunol. 125, 463-465 (1980).
A recent report on the use of U-937 cells to produce VPF-like activity was made by Beck and Habicht, J. Leukocyte Biol. 42, 568 Absts., Dec.
1987. However, the activity was not purified and chemical characterization or identity was not disclosed.
The human VPF produced by U-937 cells herein was identified by its inhibition and binding by rabbit polyclonal antibodies to guinea pig VPF. A
rabbit polyclonal antiserum to guinea pig VPF
inhibited the permeability activity produced by U-937 cells as determined in the assay of Miles and Miles, supra, (hereinafter also referred to as the Miles assay). This U-937 generated VPF activity was about 70% to 80% removed by binding to immunoabsorbents produced with protein A-Sepharose~ which had been reacted with the rabbit polyclonal antiserum to guinea pig VPF.
The process for the production of human VPF
in accordance with the invention comprises growing cells derived from the human histiocytic lymphoma cell line U-937 in serum-free nutrient culture medium at about 35° to 38°C for a sufficient time to elaborate VPF and isolating the resulting VPF from the spent cells or the cell culture conditioned medium.
A preferred method of isolating the human VPF from the cell culture conditioned medium of the U-937 cells comprises the following steps:

-7- 07-21(525)A
(a) cation exchange chromatography of said conditoned cell. culture medium, for example with a column of CM-cellulose, CM-Sephadex~, Amberlite~ IR-120H or S-Sepharose Fart Flow cation exchanger;
(b) metal affinity chromatography of the VPF active fractions from said cation exchange chromatography, for example with a Cu2+, Zn2+ or Ni2+/iminodi-acetic acid(IDA,)/Sepharose column; and (c) reverse phase H:PLC of the active VPF fractions from said method affinity chromatography, for example with a C4 or C1$ reverse phase HPLC column.
The thus purified human VPF is a protein of Mr 34,000-42,000. When subjected to N-terminal amino-acid sequence analysis, it was found to have a distinct and novel structure whereby it differed from gVPF in four of the first ten amino acid positions.
This novel human VPF or hVPF has the following N-terminal amino acid sequence:

Ala-Pro-Met-Ala-Glu-Gly-Gl:y-Gly-Gln- Asn As described hereinafter, the hVPF of this invention also contains several novel internal sequences.
Purified hVPF of the present invention was active in promoting vessel leakage at a dose of 22 ng (5.5 x 10 13 Moles) upon intr,adermal injection into guinea pigs. This highly purified hVPF thus is about 9 times more potent than the gVPF described in U.S.
G
F

-8- 07-21(525)A
Patent 4,456,550. Another advantage of hVPF of this invention is its human origin which is indicative of potential use as a human therapeutic compared to other agents of lesser purity or derived from guinea pig or other animals such as would cause immunological reactions.
Detailed Description of the Invention While the specification concludes with claims particularly pointing out and distinctly claiming the subject matter regarded as forming the present invention, it is believed that the invention will be better understood from the following detailed description of preferred embodiments of the invention taken in conjunction with the accompanying drawings in which:
FIG. 1 is a schematic representation which shows the stepwise isolation of the human VPF (hVPF) from the conditioned cell culture medium of U-937 cells in one embodiment of the invention.
FIG. 2 is a graphical representation which shows the elution patterns in the stepwise purification of the hVPF in four panels A, B, C and D
in the embodiment of FIG. 1 as follows:

20 03361.' -9- 07-21(525)A
A.) Cation Exchange Chromatography of hVPF.
Serum-free conditioned medium from U-937 cells (6 L of 6-fold concentrate) was adjusted to pH 7.0 and loaded at a flow rate of 60 ml/hr onto an S-Sepharose column (5 x 45 cm) equilibrated in 0.01 M sodium phosphate, pH 7Ø A linear gradient from 0.2 M to 0.8 M NaCl in the same buffer was used to elute hVPF.
B.) Metal Affinity Chromatography of hVPF.
The active eluate from the cation exchange column was concentrated to 20 ml by ultrafiltration and loaded onto a Sepharose (Fast Flow)/IDA/Cu+2 column equilibrated in 0.01 M sodium phosphate, pH 7.0, 2 M
sodium chloride, 0.5 M imidaz~ole. hVPF was eluted with a linear gradient of imidazole as shown.
C.) RP-HPLC of hVPF. The active eluate from the metal affinity column was loaded onto a C1$ RP-HPLC column equilibrated with 0.05%
trifluoroacetic acid (TFA) in water and eluted at 1 ml/min with a gradient of acetonitrile as shown.
D.) Sodium Dodecyls~ulfate-Polyacrylamide Gel Electrophoresis (SDS-PAGE) of RP-HPLC Fractions.
Aliquots were removed from fractions around the activity peak and analyzed by electrophoresis without reducing agent. Standards were first reduced with ~-mercaptoethanol.

~0 03361 -10- 07-21(525)A
FIG. 3 shows the ELISA of hVPF and gVPF. Rabbit anti-gVPF IgG was used to coat the polystyrene wells of microtiter plates. Various amounts of either hVPF (lower panel) or gVPF
(upper panel) were allowed to bind overnight, and the amount of bound antigen detected with biotin-anti-gVPF IgG followed by HRP-avidin. Wells were developed with HRP substrate, the absorbance was read at 490 nm. Wells containing readings that were off scale were diluted and re~-read in the linear range of the Bio-Tek microplate reader; the absorbances given are corrected for dilution.
Different x-axis scales were used for hVPF
and gVPF. Early and late bleed refers to both primary and secondary biotin-conjugated IgGs prepared from sera collected at the fourth bleed (after 3 immunizations) and at the ele~Tenth bleed (after 5 immunizations), respectively.
The U-937 cells can be cultured in well-known cell culture media such as basal medium Eagle's (BME), Dulbecco's modified Eagle medium (DMEM), medium 199, RPMI 1640 medium, and the like cell culture media such as described in detail by H.
J. Morton, In Vitro 6, 89-108 (1970). These conventional culture media contain known amino acids, mineral salts, vitamins, hormones and carbohydrates.
They are also frequently fortified with mammalian sera such as fetal bovine sertun (FBS). Other components which can be used p'.n the media are bovine serum albumin (BSA), growth factors such as trans-ferrin and insulin, protein h5rdrolysates such as lactalbumin hydrolysate, trypt~one, tryptose and peptone, as well as lipids, surfactants and the like materials. The U-937 cells preferably are cultured in serum-free media for the production of hVPF.

20 03361 ~' -11-~ 07-21(525)A
Methods for the large scale growth of mammalian cells are well-known and these methods can be used for the culture of the U-937 cells defined herein. Such methods are described, for example, by Tolbert et al., Biotech. Bioenq_ XXIV, 1671-1679 (1982); Tolbert and Feder, An.n. Rept. Ferm. Proc.
Vol. 6, Ch. 3, pp. 35-74 (1983); Harakas, Ibid. Vol.
7, Ch. 7, pp. 159-211 (1984); and references cited in said publications. U.S. Pat. Nos. 4,166,768;
4,289,854; and 4,537,860 disclose particularly useful methods and apparatus for the large scale growth and maintenance of cells for the production of protein-aceous materials. The methods and apparatus disclosed therein can be usE;d for the culture of the U-937 cells defined herein.
The cells also can be cultured on a large scale basis in nutrient medium at 37°C. in agitated suspension culture as described in U.S.
Pat. No. 4,289,854 and, after' a suitable growth period, can be maintained in the static maintenance reactor described in U.S. Pat.. No. 4,537,860 in which the medium is supplemented with 0.5% lactalbumin hydrolysate.
Although purification of the hVPF from the spent culture media can employ various known procedures for the separation. of proteins such as, for example, salt and solvent fractionation, adsorption with colloidal materials, gel filtration, ion exchange chromatography, affinity chromatography, immuno-affinity chromatography, electrophoresis and high performance liquid chromatography (HPLC), the above described three-step chromatographic method is preferred. Suitable metal affinity chromatography procedures are illustrated by Sulkowski, Trends Biotech. 3, 1-7 (1985).

-12- 07-21(525)A
In a preferred embodiment of the invention the U-937 cells are subcloned. or passaged through nude mice to improve the yield of hVPF. Such treatment of the cells has provided for the production of hVPF in quantities of up to 200 to 800 ng/ml of the conditioned medium. This is equivalent to or greater than the gVPF levels reported for the guinea pig cell Line 10.
The U-937 cells have been passaged through nude mice by injection of 1 x 10~ cells into the mouse peritoneum. Nude mice are an immunodeficient species in which human cells can grow and not be rejected. After about 3 to 4 weeks, the mice were sacrificed and soft tumors were removed from their abdomens. The tumors were mechanically dissociated into cells and these U-937 cells were again put into culture.
The following examples will further illustrate the invention although it will be appreciated that the invention is not limited to these specific examples or the details described therein.
Example 1 MATERIALS AND METHODS
Growth of U-937 Cells. U-937 cells originally obtained from the i~merican Type Culture Collection (ATCC) were subcloned in soft agar and selected for fast growth. One of these clones ~- ~0 03361 -13- 07-21(525)A
was selected for scale-up, but other clones, and even uncloned ATCC cells, also produced hVPF. The serum-free medium used contained the following components:
RPMI 1640, DME (high glucose), Ham's F12 in a 1:1:1 ratio; HEPES (25 mM, pH 7.10-'7.15); glutathione (1 mM); ethanolamine (20 NM); selenium (30 nM); NaHC03 (2 mM); CuS04 (5 nM); NH4VO3 (5 ~aM}; ZnS04 (0.5 NM);
MnS04 (0.5 nM); FeS04 (4 NM); bovine serum albumin, Miles "Pentex" (100 Ng/ml); iron rich transferrin, Miles (5 ~g/ml); bovine insulin (10 ~g/ml); ExCyte, Miles-lipid fraction (0.1% v/v); F-68 Pluracol~ (0.05%
w/v). The volume was adjusted to yield an osmolarity of 280 mOsm. The doubling tune was about 50-60 hours in this medium, whereas it was only 35-40 hours in medium containing 2% fetal bovine serum (FBS).
Cells were scaled-up in the serum-free medium from T-flasks into rol7.er bottles and then into small spinners. A 12 L :>pinner was then used to innoculate a 14 L perfusion chemostat which was perfused at a rate of approximately 3 ml medium/hour/ml of total packed cells. The culture was subsequently transferred t:o a 100 L perfusion chemostat, which was perfused under limiting nutrient conditions (1.5-2.0 ml medium~'hour/ml packed cells, or 0.1-0.15 ml/day/million cells). Cells were recycled to the reactor using an AG Tec:hnology hollow fiber-cartridge (model CFP-4-E;-6, 0.4 micron). Cell density ranged from 1.0 x 106 to 4.6 x 106 viable cells/ml, 3.0 to 23 ml/L of packed cells, and viability ranged from 64% to 84%. The production run in the 100 L reactor lasted 24 days, during which time a total of 1000 L of serum free conditioned medium was produced.
The permeate from the perfusion reactor was -14- 07-21(525)A
collected and stored at 4°C. Concentration was performed in 200-300 L lots on an Amicon DC-30 ultrafiltration apparatus with three low protein binding 10 kDa cutoff spiral cartridges (Amicon S10Y10) operated in parallel. A concentration of about 6 fold was achieved, including a phosphate buffered saline (PBS) wash of the cartridges and equipment that was pooled with the concentrate. The concentrate was stored at -20°C.
Vascular Permeability Assay. A Miles-type permeability assay (Miles and Miles, supra) was used to detect hVPF. Hairless guinea pigs (IAF/HA-H0, Charles River, Wilmington MA) were anesthetized by inhalation of methoxyflurane (Metofane~, Pitman-Moore, Inc.). A 1 ml volume of 0.5% (w/v) Evan's blue dye (Sigma Chemical Co.) prepared in sterile saline for injection (Abbott Laboratories) was injected intracardially into the circulation.
Samples for hVPF determination were prepared at appropriate dilutions in saline or phosphate buffered saline, and 200 ~1 volumes injected intradermally into sites on the back of the guinea pig. The presence of hVPF was indicated by an intense blue spot at the site of the :injection where dye (bound to serum protein) had :Leaked from the circulation into the tissues.
Cation Exchange Chromatography. Six liters of six-fold concentrated conditioned medium was adjusted to pH 7.0 with acetic acid and passed through a column (5 cm X 44 cm) of S Sepharose~ Fast Flow (Pharmacia) cation exchange gel equilibrated with 0.01 M sodium phosphate, pH 7Ø At 4°C a significant portion of the permeability enhancing ~0 03361 -15- 07-21(525)A
activity passed through the column, but at ambient temperature (25°C) 50 to 70% of the activity became bound to the column. This step was therefore regularly performed at room temperature. Sodium azide (0.01% w/v) was added to all buffers. Flow rates for loading and elution were 10 ml per minute.
After loading, the column was washed with 900 ml of 0.01 M sodium phosphate, pH 7.0, and then eluted with a 2.3 L linear gradient containing from 0.2 M to 0.8 M sodium chloride in the same buffer. Between runs, the column was washed with 0.1. M sodium hydroxide before re-equilibrating with 0.01 M sodium phosphate, pH 7Ø
Metal Affinity Chromatography. Metal affinity chromatography was performed using a copper-iminodiacetic acid complex covalently linked to agarose via a spacer arm. The gel was synthesized in two steps by first adding a,n epoxide-containing spacer arm to the agarose, and then reacting the activated gel with iminodiacet:ic acid.
Highly cross-linked agarose (Sepharose Fast Flow, Pharmacia) was repeatedly washed with distilled water to remove all buffers and preservatives and then dried by suction. About 100 g (100 ml) of this damp gel was suspended in 60 mil distilled water, and then 40 ml freshly prepared 2.5 M NaOH solution was added to the gently stirring a.garose suspension.
Then 100 ml diethyleneglycol d.iglycidyl ether, prepared as described by Gu et. al., Synthesis, 649-651 (1983), was added, and. the mixture was gently stirred at 30°C for 16 hours. The activated gel was repeatedly washed with distilled water to remove the excess epoxide and base. The washed, suction-dried gel contained 70 micromoles active epoxide groups per ~0 03361 -16- 07-21(525)A
mL gel. The activated gel was stored in distilled water at 4°C and generally used within 24 hours of preparation.
About 100 ml (100 g) of-the activated Sepharose Fast Flow was washed with distilled water, dried by suction, and suspended in 100 ml of 1.0 M
Na2NH(CHZCOZ)-H20 solution, which was adjusted to pH
- 11Ø This mixture was gently stirred at 65°C for 24 hours and then repeatedly washed with distilled water to remove excess ligand. The functionalized gel was stored in ethanol/water (25/75 v/v) at 4°C
until ready for use. Titration with thiosulfate showed the absence of epoxide groups, so capping with ethanolamine was deemed unnecessary. To determine the metal binding capacity of the gel, 10 ml of suction-dried gel was saturate=d with excess 50 mM
Cu(Clo4)2 and then carefully washed with distilled water. Finally, the bound copper was removed with an excess of 50 mM Na2H2EDTA. Using standardized copper-EDTA solutions for comparison, the copper content was photometrically determined to be 43 micro-moles Cu per milliliter of damp, suction-dried gel.
Chromatography was performed in a glass column (1.0 x 13 cm.) containing 10 ml of gel. The gel was charged with a 50 mM ~~olution of Cu(C104)2, pH = 4.5, and then saturated with a buffered imidazole solution (20 mM imidazole + 2 M NaCl + 50 mM NaH2P04, pH 7.0). After thoroughly washing with 1.0 M NaCl, the column was then equilibrated with the starting buffer (50 mM NaH2P04, pH 7.0, 2 M NaCl, 0.5 mM imidazole). The fractions from the cation exchange step containing permeability enhancing.
activity were pooled and concentrated from a volume of about 700 ml to 20 ml using an *Amicon YM5 *Trade-mark 20 03361 , -17- 07-21(525)A
membrane. Upon concentration, a protein precipitate usually formed that could be removed by filtration.
The sample was applied to the column at ambient temperature (25°C) and eluted at a flow rate of 0.5 ml/min over 500 minutes with .a linear gradient of imidazole (0.5 mM to 60 mM) in 50 mM NaH2P04, pH 7.0, 2 M NaCl.
Reverse-Phase HPLC. Reverse-phase HPLC
(RP-HPLC) was performed using a 4.6 mm x 25 cm Vydac column (Separations Group) containing 5 ~M packing with 330 angstrom pore size. The mobile phases were: "A", 0.05% trifluoroacE_tic acid (TFA) in water and "B", 0.05% TFA in acetonii~rile. After loading the sample, the column was washed with "A" until the absorbance again reached baseline, and then eluted with the following linear gradients: 0% to 20% "B"
over 20 minutes, 20% to 40% "B" over the next 80 minutes, and then 40% to 100% "B" over the next 20 minutes. All flow rates were 1 ml/min. Fractions were collected in siliconized glass tubes.
N-Terminal Amino Acid Sequence Analysis.
Automated Edman degradation chemistry was used to determine N-terminal amino acid sequence. An Applied Biosystems, Inc., model 470A c~as phase sequencer (Foster City, CA) was employed for the degradations [Hunkapiller, et al, Methods E,nzymol. 91, 399-413 (1983}]. The respective PTH-amino acid derivatives were identified by RP-HPLC analysis in an on-line fashion employing an Applied E~iosystems, Inc., Model 120A PTH Analyzer fitted with a Brownlee 2.1 mm I.D.
PTH-C,Q column. Yields of PTH: amino acids wPrP
determined by comparison with an external standard mixture. The average repetitive yield was calculated ~o033s~

-18- 07-21(525)A
by linear regression analysis of the log pmolar yield versus cycle number plot.
Similarities between the obtained sequence and known sequences were investigated using the computer program FAST A [Pearson and Lipman, Proc. Natl. Acad. Sci. USA 85, 2444-2448 (1988)].
Similarity searches were also performed against the National Biomedical Research ;Foundation (NBRF) protein sequence data base [S:idney et al., Nucleic Acids Res. 16, 1869-1:371 (1988), Release 17, June 1988] and the translated GENBANK DNA base [Bilofsky and Burks, Nucleic Acids Res. _16, 1861-1864 (1988), Release 56, June 1988].
Tryptic peptides were prepared from reduced and alkylated hVPF. hVPF (1 iunole) was dissolved in 100 ~1 of 0.5 M Tris HC1, pH 8.5, 6 M guanidine-HC1, 1 mM EDTA, and 5 mM dithiothreii:ol. The solution was incubated for 30 minutes at 37°C before the addition of sodium iodoacetate (to a final concentration of 5 mM) and incubation at 4°C overnight. After dialyzing against 2 M guanidine HC1, O.C)1 M Tris-HC1, pH 8.5, and then 0.1 M ammonium bicarbonate, 1 ~g of TPCK
treated trypsin (Sigma Chemical Co., St. Louis, MO) was added and the solution incubated overnight at 37°C. Peptides were separatecl by RP-HPLC using a °.
Nucleosil C18, 5 micron, 100 F,, 4.6 x 250 mm column (Macherey-Nagel, Inc.). The f.'low rate was 1 ml/min at room temperature. A linear gradient of 0 to 90%
acetonitrile in 0.1% trifluoro~acetic acid run over 270 minutes was used for elution.
Electrophoresis. Sodium dodecyl sultate-polyacrylamide gel electrophoresis (SDS-PAGE) was carried out on 10-15% gradient polyacrylamide gels using a Pharmacia PhastSystem. The buffer systems and the silver staining protocol were those that were recommended by the manufacturer.

~0 03361 -19- 07-21(525)A
Antibodies and Immunoassays. Rabbit polyclonal antiserum to gVPF (designated F001) was prepared by immunizing a New «ealand White (NZW) rabbit with repeated injections of gVPF. The first injection was in Complete Freund's Adjuvant followed by boosts in Incomplete Freund's Adjuvant. gVPF was purified by preparative SDS-PAGE [Senger et al, Science 219 983-985 (1983)] following purification by the method of Senger et al. _Fe~d. Proc. _46, 2102 (1987).
Sandwich enzyme linl~;ed immunosorbant assays (ELISA) for human and guinea pig VPF were performed as described below. The IgG fraction of F001 antiserum was purified using adsorption onto protein A-Sepharose. 500 ng/well of the IgG obtained from an 11th bleed serum was coated onto polystyrene microtiter plates for 3 hours. Then various known concentrations of human and guinea pig VPF were allowed to bind overnight. Th.e concentrations were estimated from the maximal dilution which still produced a detectable response in the Miles permeability assay. This concentration was 50 ng/ml of VPF. The amount of bound antigen was detected with 500 ng/well biotin-a-gVPF IgG (11th bleed F001 antiserum) for 2 hours followed by 1/2000 dilution of horseradish peroxidase (HRP)-avidin (Cappel Labs) for 90 minutes. Wells were developed with the HRP substrate o-phenylenediamine-2HC1 plus H202 and the absorbance was read at 490 nm in a BioTek reader.
It had been observed that the earlier bleeds of rabbit F001 contained a higher concentration and/or higher affinity of antibodies cross-reactive with hVPF compared to later -20- 07-21(525)A
bleeds. Therefore, an ELISA for hVPF was also performed with an early bleed (4th bleed) antiserum.
It was performed under the same conditions as described above except that 1500 ng/well of a biotin-F001 IgG (4th bleed) w<~s used to detect the amount of bound antigen.
RESULTS
U-937 Permeability Enhancing Activity. The serum-free conditioned medium from U-937 cells in culture produced a positive response when tested in the Miles permeability assay. These results were unexpected since many cells do not produce a VPF-like activity as seen from Example 2,~below. The Miles assay measures extravassation of Evan's blue dye-serum albumin complexes from the circulation after intradermal injection of a te~;t material. However, the assay is non-specific and could measure positive response from a variety of substances, including, for example, histamine. It was therefore not known initially if the permeability enhancing activity produced by U-937 cells was related to guinea pig tumor VPF. To test this, the medium was mixed with an immunoadsorbent composed of protein A-Sepharose~ and IgG obtained from polyclonal anti-sera against gVPF.
Most, but not all, of the permeability enhancing activity present in the U-937 medium was adsorbed using this procedure, but not when control IgG
obtained from rabbits not immunized with gVPF was used instead. Most of the permeability enhancing activity secreted into the medium of U-937 cells therefore appeared to be related to guinea pig tumor derived VPF. However, as discussed below, even though hVPF shares some immunocrossreactivity with gVPF, it is immunologically distinct from gVPF.

~- ~0 0336 1 -21- 07-21(525)A
hVPF Purification. Initial attempts at purification of U-937 cell derived permeability enhancing activity employed the purification method previously used for gVPF as described by Senger et al, _Fe~d. Proc. 46, 2102 (1987). The application of this method, or minor modifications thereof, did not produce homogeneous protein from U-937 cell conditioned medium, even though the chromatographic behaviour of the permeability enhancing activity was very similar to that of gVPF. A novel purification method was therefore developed that incorporated cation exchange chromatography, metal affinity chromatography, and RP-HPLC-(Fig. 1). In the first step, concentrated conditioned medium was passed over an S Sepharose cation exchange column (Fig. :?A). About 50-70% of the permeability enhancing aci~ivity was bound to the column at pH 7Ø The non-adsorbed activity was not characterized. The bound activity was eluted with a gradient of sodium chloride, and after concentration by ultrafiltration, loaded onto a metal affinity column (Fig. 2B). All detectable activity was tightly bound by the copper/IDA/Sepharose column, and was eluted after most of the other proteins in a gradient of imidazole. The final step utilized RP-HPLC (Fig.
2C) and resulted in elution of a group of Mr ~40 kDa proteins in the fractions associated with the peak of permeability enhancing activity (Fig.
2D). This method has been repeated numerous times and it reproducibly yielded Mr. ~40 kDa protein of about 90% or greater purity as analyzed by SDS-PAGE
with silver staining or by N-terminal sequence analysis. Approximately 1-2 Ng of pure protein can be obtained per liter of U-937 cell conditioned medium.

~0 03361 -22- 07-21(525)A
Dose Response of hVPF Induced Permeability Enhancement. Different amounts of hVPF
were tested in the Miles permeability assay. The lowest dilution producing a positive response was at a hVPF concentration of about 2.75 nM. This corresponds to an injection dose of 22 ng, or 0.55 picomoles of Mr 40 kDa hVPF. This is equivalent to only one-ninth the 200 ng required for a similar response by the gVPF
described in U.S. Patent 4,456,550.
Amino Acid Sequence of hVPF.
hVPF was subjected to N-terminal amino acid sequence analysis (Table 1). Complete identity was observed between hVPF and the guinea pig tumor derived gVPF for the first 6 positions, but the sequence diverged for the next 4 amino acids sequenced. The identity is tlZUS only 60% in this N-terminal region.
Table :L
2 0 Comparison of N-Terminal Sequences of hVPF and gVPF
Residue (pMole Yield) Cycle hVPF gVPF

1 Ala (627) Ala (838) 2 5 2 Pro (427) Pro (598) 3 Met (406) Met (358) 4 Ala (130) Ala (463) 5 Glu (85) Glu (456) 6 Gly (55) Gly (434) 3 0 7 Gly (76) Glu (537) 8 Gly (150) Gln (276) 9 Gln (34) Lys (179) 10 Asn (29) Pro (354) 3 5 Approximately 1 nmole (about 40 Ng as estimated by SDS-PAGE and silver staining) of hVPF was sequenced. The amino acids detected and the yield at each cycle (in parentheses) are shown. The average repetitive yield was 73%. These data are representative of several other runs on similar hVPF preparations.

~0033fi1 -23- 07-21(525)A
In order to obtain internal sequence information, hVPF was reduced, carboxymethylated with iodoacetic acid, and treated 'with trypsin. The resulting peptides were then separated using RP-HPLC.
Several of the isolated peptides were sequenced, as shown in Table 2. None of these sequences, nor the N-terminal sequences, showed significant homology to proteins present in published data bases. The novel hVPF of the present invention thus is substantially different from previously described proteins and from gVPF.
Table :>_ Sequence of hVPF Tryptic Peptides Peptide (Gln)GlnGlnLys Pro 27 (Arg)GlnGluGln Arg(Pro Lys) 53 Phe AspVal Tyr Gln Arg(Arg) Met 2 0 71 Ile LysPro Ser Cys Val F'ro Leu Met Phe Arg 93 Val IlePhe G5n Glu Tyr F'ro Asp Glu Asp Ile Glu Tyr io Reduced and carboxymethylated hVPF was treated with trypsin and the peptides separated by RP-HPLC. Peptides that were well isolated were sequenced. The number designation corresponds to the fraction number in which the peptide appeared.
Residues in parentheses are designated with a fairly low degree of confidence.

20 0338 1 ~~

-24- 07-21(525)A
Reactivity of hVPF ~W_ith Anti-gVPF
Antibodies. Polyclonal rabb it antibody against gVPF
was tested for cross-reactivity with hVPF
using a sandwith ELISA. Figure 3 shows that . hVPF is recognized by anti-gVl?F IgG, but that the concentration of hVPF required for detection is about 10 fold higher than that for gVPF. Furthermore, although the maximum absorbance attained for hVPF (1.5 absorbance units above background) was significant, it was about 3 fold lower than that attained usinct gVPF as antigen.
These results imply that the t:wo proteins are related, but that hVPF is immunologically distinct from gVPF.
Example 2 hVPF derived from U-937 cells was compared with VPF from several other cell lines by the Miles assay with the following results. The Mnng HOS human osteogenic sarcoma cell line was used as representative of the prior art human cell lines for production of VPF described by Senger et al., Cancer .
Res. 46, 5629-5632 (1986). The cells were extracted at a density of 1.5 to 2.0 x 106 cells/ml for 48 hours in serum-free basal medium, either RPMI 1640 or DME. The extract was collected, centrifuged to remove cells and debris, and then concentrated if necessary using *Centricon 10 membranes (Amicon Corp.). The extracts were then tested in the Miles assay. The results are set forth in Table 3.
*Trade-mark -25- 07- 21(525)A

Table 3 Cell Line Conditioned Miles Estimates Media of Concn.(a) Assay VPF Concn.

(units)(b) (ng/ml) (c) U-937(ATCC) Uncloned 1X 1 50 U-937(ATCC) 1X 2-8 100-400 Cloned (d) U-937(ATCC) Uncloned-Passaged through 1X 4-8 200-400 Nude mouse Mnng HOS

Human osteogenic lOX +++ (e) 30 (e) sarcoma JURKAT

Human acute lymphoblastic 12.3X 1-2 4-8 leukemia F2.11-2X

Human T cell-T cell lOX 0 <5 hybridoma Neonatal human 26.6X 0 <2 foreskin fibroblasts Human fetal 3 0 lung fibroblasts 26.6X 5 10 HEK

Human embryonic 16.9X 4 10 kidney 3 5 Human sarcoma 36X >16 >20 ~0 03361 -26- 07-21(525)A
(a) Conditioned media were concentrated X-fold as indicated prior to testing by the Miles assay.
(b) Activity of sample = greatest dilution of sample at which blue spot was detectable.
1 unit (u) = the amount of VPF in a sample producing the smallest detectable blue spot discernible from control injections without VPF.
(c) Estimate of VPF concentration in the unconcentrated spent media. in ng/ml based on 1 unit activity= 50 ng/ml, a relationship established with the guinea pig Line 10 VPF of U.S. Patent 4,456,550.
(d) Several clones, obtained by limiting dilution cloning procedures, produced activities in the 100-800 ng/ml range.
(e) This is an estimate based on the previously determined activity of Mnng HOS cells, since dilutions of the samples were not performed in this assay.
Various other examples will be apparent to the person skilled in the art .after reading the present disclosure without departing from the spirit and scope of the invention. It is intended that all such examples be included within the scope of the appended claims.

Claims (7)

1. Method for the production of human vascular permeability factor in vitro comprising growing cells derived from human histiocytic lymphoma cell line U-937 in nutrient culture medium at about 35° to 38°C for a sufficient time to elaborate vascular permeability factor and isolating the resulting vascular permeability factor from spent cells or the cell culture conditioned medium.

2. The method of Claim 1 in which the vascular permeability factor is isolated from the cell culture conditioned medium by (a) cation exchange chromatography of said conditoned cell culture medium;
(b) metal affinity chromatography of the VPF active fractions from said cation exchange chromatography; and (c) reverse phase HPLC of the active VPF fractions from said metal affinity chromatography.

3. The method of Claim 1 in which the nutrient culture medium is serum-free.

4. The method of claim 1 in which the cation exchange chromatography is carried out with a column of S-*Sepharose Fast Flow cation exchange, the metal affinity chromatography is carried out with a copper/iminodiacetic acid (IDA)/Sepharose column and the reverse phase HPLC is carried out with a C18 reverse phase HPLC column.

5. A human vascular permeability factor having a molecular weight of about 34 to 42 kDa as determined by non-reduced SDS-PAGE and an N-terminal amino acid sequence as follows:

Ala-Pro-Met-Ala-Glu-Gly-Gly-Gly-Gln-Asn.

6. The human vascular permeability factor of Claim 5 containing the following internal sequences:
(a)(Gln)Gln Gln Lys Pro, (b)(Arg)Gln Glu Gln Arg(Pro Lys), (c)Phe Met Asp Val Tyr Gln Arg(Arg), (d)Ile Phe Lys Pro Ser Cys Val Pro Leu Met Arg, and (e)Val Asp Ile Phe Gln Glu Tyr Pro Asp Glu Ile Glu Tyr.

7. The human vascular permeability factor produced by the method of Claim 1.