AU713531B2 - Method of diagnosis of atherosclerosis using anti-cholesterol antibodies - Google Patents

Description

WO 97/21099 PCT/US9 6 193 81 METHOD OF DIAGNOSIS OF ATHEROSCLEROSIS USING ANTI-CHOLESTEROL ANTIBODIES The present application is in the field of antibodies particularly anti-cholesterol antibodies and more specifically relates to the diagnosis of atherosclerosis and diseases related to cholesterol.

BACKGROUND OF THE INVENTION Arteriosclerosis, a generic term for thickening and hardening of the arterial wall, is responsible for the S 15 majority of deaths in the United States and most westernized societies. One type of arteriosclerosis is atherosclerosis, the disorder of the larger arteries that underlies most coronary artery disease, aortic aneurysm, and arterial disease of the lower extremities, and also plays a major role in cerebrovascular disease. Atherosclerosis is by far the leading cause of death in the United States, both ~above and below age 65 in both sexes. E.L. Bierman, "Atherosclerosis and Other Forms of Arteriosclerosis," Ch.

208, p. 1106 in Harison's Principles of Internal Medicine, WO 97/21099 PCT/US96/19381 13th edition, eds. K.J. Isselbacher, et al. (McGraw-Hill, Inc.

NY 1994).

Atherosclerosis and Cholesterol Atherosclerosis is characterized by infiltration of cholesterol and appearance of foam cells in lesions of the arterial wall. This is followed by a complex sequence of changes involving platelets, macrophages, smooth muscle cells, and growth factors that produces proliferative lesions.

These distort the vessels and make them rigid. In individuals with elevated plasma cholesterol levels, there is an increased incidence of atherosclerosis and its complications. W.F. Ganong, Review of Medical Physiology, 17th edition, p. 281 (Appleton Lange Norwalk, CT 1995).

Cholesterol Metabolism Cholesterol is absorbed from the intestine and incorporated into the chylomicrons formed in the mucosa.

After the chylomicrons discharge their triglyceride in adipose tissue, the chylomicron remnants bring cholesterol to the liver. The liver and other tissues also synthesize cholesterol. Some of the cholesterol in the liver is excreted in the bile, both in the free form and as bile acids. Some of the biliary cholesterol is reabsorbed from the intestine.

Most of the cholesterol in the liver is incorporated into very low density lipoproteins, and all of it circulates in lipoprotein complexes.

WO 97/21099 PCT/US96/19381 The transport system for endogenously produced cholesterol is made up of very low density lipoproteins (VLDL), intermediate-density lipoproteins (IDL), lowdensity lipoproteins (LDL), and high-density lipoproteins (HDL). The protein constituents of these proteins are called apoproteins. The major apoproteins are called apo E, apo C, and apo B. Apo B has two forms, apo B-48, characteristic of the exogenous lipid transport system, and apo B-100, characteristic of the endogenous system. VLDL are formed in the liver and transport triglycerides formed from fatty acids and carbohydrates in the liver to extrahepatic tissues. After their triglyceride is largely removed by lipoprotein lipase, they become IDL. The IDL give up phospholipids and, through the action of lecithincholesterol acyltransferase, pick up cholesteryl esters formed from cholesterol in the HDL. Some IDL are taken up by the liver. The remaining IDL then lose more triglyceride and protein, and become LDL. During this conversion, they lose apo E, but apo B-100 remains.

LDL provide cholesterol to the tissues. In the liver and most extrahepatic tissues, LDL are taken up by receptor-mediated endocytosis in coated pits. The receptors recognize the apo B-100 component of the LDL. They also bind apo E but do not bind apo B-48. LDL are also taken up by a lower affinity system in the macrophages and some other cells. When overloaded by a high plasma level of HDL, the macrophages become full of cholesteryl esters and make up the foam cells that appear early in atherosclerotic lesions. The LDL receptor on macrophages WO 97/21099 PCT/US96/19381 and related cells is called the scavenger receptor. W.F.

Ganong, pp. 277-279; M. J. Malloy, et al., "Agents Used in Hyperlipidemia," ch. 34, p. 522, in Basic and Clinical Pharmacology, ed. B.G. Katzung (Appleton Lange Norwalk, CT 1995).

Cholesterol leaving cells is absorbed in HDL, lipoproteins that are synthesized in the liver and the intestine. Some of the HDL contain apo E and bind to LDL receptors on other cells, thus transporting cholesterol from one cell to another. W.F. Ganong, pp. 277-279.

Cholesterol and the Immune Response For over 70 years, sporadic reports have appeared in the literature documenting the induction of antibodies that react with cholesterol. Early studies used indirect methods, such as complement lysis of cholesterol-containing liposomes (Swartz, G. C. R. Alving, et al., 1988, Proc.

Natl. Acad. Sci. USA 85: 1902.; Watanabe, et al., 1991, Infect. Immun. 59: 2200.; Aniagolu, et al., 1995, J.

Immunol Methods,182: 85; Wassef, N. et al., 1989, J.

Immunol. 143; 2990; Alving, et al., 1989, Bichem. Soc. Trans. 17: 637) to quantify such antibodies; more recently several ELISA were developed to demonstrate direct binding of antibodies to cholesterol as an antigen. Swartz, G. C. R. Alving, et al., 1988, Proc. Natl. Acad. Sci. USA 85:1902; Watanabe, T.,et al., 1991, Infect. Immun. 59: 2200; Aniagolu, G.

M. Swartz, J. Dijkstra, J. W. Madsen, J. J. Raney, and S. J.

Green, 1995, J. Immunol Methods 182: 85; Wassef, N. M., WO 97/21099 PCT/US96/19381 C. R. Alving, et al., 1989, J. Immunol. 143; 2990; Alving, C.R.,et al., 1989, Bichem. Soc. Trans. 17: 637. See also U.S. Patent No. 4,885,256 to Alving. A wide variety of cholesterol conjugates (cholesterol or cholesterol-ester protein conjugates, cholesterol ester complexed to polylysine, cholesterol covalently linked to phospholipids and complexed to polylysine) or cholesterol-containing substances (LDL, HDL, Mycoplasma, cholesterol-rich liposomes) can induce such antibodies with specificity for cholesterol (Wadsworth, et al., 1935, J. Immunol. 29: 135; Klopstock, et al., 1964, J. Immunol. 92: 515; Sato, et al., 1972, Immunochem. 9: 585; Sato, et al., 1976, Biomedicine 24: 385; Sato, et al., 1976, Jap. J. Exp.

Med. 46: 213; Hara, et al., 1979, Chem. Phys. Lipids 23: 7; Swartz, G. et al., 1988, Proc. Natl. Acad. Sci.

USA 85: 1902.; Watanabe, et al., 1991, Infect. Immun.

59: 2200), and virtually no reactivity with other sterols.

Berger, et al., 1932, Z. Imunitdt. 76: 16. The demonstration that most experimental animals and humans have readily detectable naturally occurring antibodies to cholesterol (Wassef, N. C. R. Alving, et al., 1989, J.

Immunol. 143; 2990; Alving, 1989, Biochem. Soc.

Trans. 17: 637) opens the issue of what role these antibodies play in vivo in cholesterol metabolism and atherosclerosis.

Autoantibodies to cholesterol have been thought to influence the development of atherosclerosis. Cholesteroldependent complement activation has been implicated as a WO 97/21099 PCT/US96/19381 possible mechanism for the pathogenesis of atherosclerosis (reviewed in Alving, et al., 1991, Crit. Rev.

441. Libby, et al., 1991, Lab. Invest. 64: Other studies have indicated that vaccination of rabbits with b-lipoproteins, or with cholesterol esters conjugated to albumin or b-lipoproteins, reduced elevated serum cholesterol levels and aortic atherosclerosis induced by a high cholesterol diet (Ger6, et al., 1959, Lancet, ii: 6; Ger6, et al., 1961, Lancet, ii: 1119; Bailey, J. et al., 1964, Nature: 201, 407; Bailey, J. 1967, Rev.

Atheroscl.9: 204; Bailey, J. et al., 1967, in The Reticuloendothelial System and Atherosclerosis. N. R. Di Luzio, et al., eds, Plenum Press, New York, p. 433). It has recently been reported that immunization of LDL receptor deficient rabbits with malondialdehyde-modified

LDL

reduced atherosclerotic lesions by 20% (Palinski, et al., 1995, Proc. Natl. Acad. Sci. USA 92: 821 However, it is not clear from the majority of these studies whether an immune response to cholesterol or cholesterol ester was elicited. One researcher has reported the induction of antibodies against the cholesterol-ester portion of the albumin conjugate (Bailey, J. 1967, Rev. Atheroscl.9: 204.). However, the biological effects of anticholesterol antibodies are still unknown.

It is an object of the invention to provide a means of removal of excess cholesterol from the circulation.

WO 97/21099 PCTIUS96/19381 It is a further object of the invention to provide a non-invasive means of imaging atherosclerotic lesions early in the disease process.

It is a further object of the invention to provide a means to assay biological samples, such as blood, serum or samples drawn from atherosclerotic lesions, for the amount and relative ratio of VLDL, IDL, LDL, and HDL molecules.

SUMMARY OF THE INVENTION Antibodies that selectively and differentially bind to lipoproteins are described. Methods of use of anticholesterol antibodies are described for diagnosis and treatment of atherosclerosis. In one embodiment of the invention, anti-cholesterol antibodies are used to assay samples drawn from atherosclerotic lesions for the amount and relative ratio of lipoprotein molecules. In another embodiment of the invention, the anti-cholesterol antibodies are used as fusiogens to promote fusion of VLDL, IDL and LDL molecules and liposomes, to promote clearance of cholesterol from the body.

BRIEF DESCRIPTION OF THE FIGURES Figure 1 is a graph of the induction of antibodies to cholesterol with 71 mol cholesterol liposomes. BALB/c mice were inoculated twice (2 weeks apart) with liposomes WO 97/21099 PCT/US96/19381 containing 71 mol cholesterol. One week after the second inoculation, mice were bled and serum assessed for anticholesterol-binding activity using the PVDF-ELISA.

Absorbance values on the ordinate are the are the mean of triplicate determinations, with the standard deviation indicated. The ordinate gives the reciprocal of the serum dilution. Background absorbance amounted to 0.200.

(squares): immune serum; (diamonds): pre-immune serum.

Figure 2 is a bar graph of the effect of epitope density of liposomal cholesterol on induction of antibodies to cholesterol. BALB/c mice were inoculated twice (2 weeks apart) with liposomes containing various amounts of cholesterol. Serum was collected one week after the boost and anticholesterol activity assessed with the PVDF-ELISA as described below. Antibody titers (ordinate) are the mean of triplicate determinations. Pre-immune titers were less than 800. The abscissa presents the mole percentage of liposomal cholesterol.

Figure 3 is a bar graph of the determination of anti-phospholipid and anti-lipid A antibodies in immune serum. PVDF-ELISA wells were coated with cholesterol micrograms), DMPC (7.5 micrograms), or DMPG micrograms) (abscissa). Polystyrene-ELISA wells were coated with lipid A (0.5 micrograms). Immune serum (obtained as for Figure 1) was assessed for binding activity to the individual liposomal constituents as described below. Titers were calculated from triplicate r ^i?/19 3 81 0 3 JUL 199r, determinations (abscissa). Titers were calculated from triplicate determinations. Pre-immune titers were less than 800.

Figure 4 is a graph of the binding of the anticholesterol monoclonal antibody 2C5-6 to cholesterol.

The binding of the affinity chromatography-purified, murine IgM anticholesterol monoclonal antibody 2C5-6 was assessed using the PVDF-ELISA. Absorbance values SD are the mean of triplicate determinations. (squares): anticholesterol IgM; (circle): nonspecific murine IgM. The absorbance at 405 nm is presented on the ordinate, the nanograms per milliliter of monoclonal antibody is presented on the abscissa.

Figure 5A, 5B and 5C are graphs of anti-cholesterol antibody binding to intact human VLDL/IDL, LDL and HDL, in ELISA wells coated with various amounts of lipoprotein protein. As indicated, varying amounts of HDL, LDL and VLDL/IDL were used to coat the wells of polystyrene-ELISA plates. After removal of unbound lipoprotein, dilutions of immune serum were assessed for binding activity. Absorbance values are the mean of triplicate determinations.

Absorbance values of pre-immune serum did not exceed 0.250. On the abscissa is plotted the reciprocal of serum dilution. On the ordinate is measured the absorbance at 405 nm.

AMENDED SHEET AMENDED SHEET FUl/US 9 6 19 3 8 1 IPEA/US o0 JUL 1991r Figure 6 is a graph of anti-cholesterol antibody binding to intact human VLDL/IDL, LDL and HDL, in ELISA wells coated with corresponding amounts of unesterified cholesterol. Polystyrene-ELISA plates were coated with 50 microliters of a solution containing either 0.05 micrograms VLDL/IDL, 0.15 micrograms LDL, or 220 micrograms HDL (based on protein). After removal of unbound lipoprotein, binding of the monoclonal antibody 2C5-6 was assessed. After performing the ELISA, the amounts of total and unesterified cholesterol adsorbed to the wells was determined for each individual lipoprotein.

The obtained values are indicated in the Figure. On the abscissa is plotted the reciprocal of the serum dilution, on the ordinate is plotted the absorbance at 405 nm.

Figures 7A and 7B are graphs of the binding of anti-cholesterol antibodies to total lipid extracts from human lipoprotein. Fig. 7A is a graph of the total lipid extracted from either VLDL/IDL, LDL, or HDL was bound to the PVDF membrane of ELISA plate wells micrograms of total cholesterol per well). Fig. 7B is a graph of the matched amounts of the intact lipoproteins micrograms total cholesterol per well) were used to coat the wells of PVDF-ELISA plates. Both the lipid extracts and the intact lipoproteins were incubated with the indicated dilutions of immune and pre-immune serum and further processed as described in the Methods section.

Absorbance values are the mean plus or minus the standard cT R4 4 deviation of triplicate determinations. On the abscissa is -oAMEND AMENDED SHEET 0 IN/US 0 3 JUL 1997 plotted the absorbance values at 405 nm, on the ordinate is plotted the reciprocal of the dilution.

Figures 8A and 8B are photographs of an anti-cholesterol immunoblot of TLC-separated lipid extracts from human lipoproteins. Total lipid extracts of VLDL/IDL, LDL, and HDL (10 micrograms total cholesterol) were separated by TLC and lipids were then transferred to PVDF membranes. Membranes were probed with anticholesterol immunoreactive serum and stained with peroxidase-conjugated anti-IgM antibody as described below. Fig. 8A: TLC plate stained after the transfer with ferric chloride spray. Fig. 8B: Immunoblot (mirror image). C: cholesterol; CE: cholesterol esters; TG: triglycerides; PL: phospholipids.

Figures 9A, 9B, and 9C are graphs of the effect of temperature on anticholesterol antibody binding to cholesterol, cholesterol-rich liposomes, and VLDL/IDL.

Cholesterol (Fig. 9A), 71 mol cholesterol liposomes (Fig. 9B), and VLDL/IDL (Fig. 9C) were absorbed onto PVDF-ELISA plates (10 micrograms free cholesterol/ well). Binding of immune serum was assessed at 4°C (squares) or 37 0 C (diamonds). Preimmune serum was also examined at 4 0 C (circles) and 37 0 C (triangles). Absorbance values are the mean standard deviation of triplicate determinations. On the abscissa is plotted the absorbance values at 405 nm, on the ordinate is plotted the reciprocal of the dilution.

AMENDED

SHEET

S" PCT/US 9 6 19 3 8 1 IPEA/US 0 3 JUL 1997 Figures 10A, 10B and 10C are graphs of the effects of anticholesterol reactive serum on lipoproteins in suspension. VLDL/IDL (Fig. 10A), LDL (Fig. and HDL (Fig. 10C) (10 micrograms total cholesterol), were admixed with the anticholesterol IgM monoclonal antibody 2C5-6 or irrelevant control IgM antibodies micrograms protein) for 30 minutes at 4°C. Aliquots were then analyzed for aggregation by flow cytometry. A shift of the curve to the right is indicative of an increase in size.

Figures 11A through 11E are photomicrographs of lipoproteins and liposomes after incubation with anticholesterol antibodies. Human VLDL/IDL (Fig. 11A), LDL (Fig. 11B), and HDL (Fig. 11C) or cholesterol-rich (71 mol SUV (Figs. 11D and 11E) were admixed with monoclonal anticholesterol IgM antibodies (2C5-6; micrograms protein), incubated at 4°C, and checked under the microscope after 30 minutes and at later time points.

Although observable after 30 to 60 minutes, the obtained structures were photographed after overnight incubation to optimize the images. In contrast to HDL (Fig. 11C), incubation of the antibodies with VLDL/IDL (Fig. 11A) and LDL (Fig. 11B) resulted in the appearance of spherical, droplet-like structure with diameters up to micrometer. Similarly, incubation with small unilamellar vesicles (diameter 0.1 micrometer) resulted in the formation of large vesicular structures (diameter up to micrometers) (Fig. 11D). In the absence of antibodies, the -12- AMENDED SHEET PCT/US 9 6 19381 IPEA/US 0 3 JUL 1997 liposomes were not visible at the light microscopic level (Figure liE). Magnification: 200x.

Figure 12 is a graph of showing the detection of antibodies to cholesterol in the serum of BALB/c mice inoculated with liposome containing 71 cholesterol and 2 micrograms monophosphoryl lipid A, using polyvinylnitrocellulose composite on which cholesterol was adsorbed. On the abscissa is plotted cholesterol (micrograms/well), and on the ordinate is plotted absorbance at 405 nm. (square immune; circles preimmune). Each point is the mean of three determinants plus or minus the standard deviation.

Figure 13 is a graph of cholesterol detection using various dilutions of immune sera added to wells containing cholesterol immobilized on polyvinyl nitrocellulose. On the abscissa is plotted cholesterol (micrograms/well), and on the ordinate is plotted absorbance at 405 nm.

DETAILED DESCRIPTION OF THE INVENTION Novel antibodies that selectively and differentially bind to cholesterol and to the different classes of lipoproteins are provided by the present invention.

Methods of use of anticholesterol antibodies are described for diagnosis and treatment of atherosclerosis.

A method of determining the amount VLDL, IDL PA and LDL, in a biological sample comprising the steps of: -13- AMENDEO SHEET WO 97/21099 PCT/US96/19381 contacting a biological sample with an anticholesterol antibody; measuring the formation of antigen-antibody complexes by; and determining the amount of VLDL, IDL and LDL in the sample.

While the biological sample, may be derived from any suitable biological source, such as blood, serum, tissue and the like, in one embodiment of the method it is desirable that the biological sample is derived from an atherosclerotic lesion.

It is also desirable that the antibody is produced by the hybridoma cell line deposited at the ATCC under Designation Number ATCC HB-8995.

15 In one embodiment of the invention, the ATCC HB-8995 derived anti-cholesterol antibodies are used to assay samples drawn from atherosclerotic lesions for the amount and relative ratio of VLDL, IDL, LDL, and HDL molecules.

In another embodiment of the invention, the anticholesterol antibodies are used as fusiogens to promote fusion of VLDL, LDL molecules and liposomes, to promote clearance of cholesterol from the body.

The antibodies of the present invention can be labeled with paramagnetic ions, such as Eu or Gd, or a stable free radical for in vivo immunodetection. Additionally, the antibodies of the present invention can be labeled with 1 11 n for use in computer assisted tomography. Still further, labels useful for in vitro assays, such as radioactive isotopes, color generating reagents or light emitting 14- WO 97/21099 PCT/US96/19381 reagents, may be conjugated or otherwise associated with the antibodies of the present invention.

Labeled antibodies of the present invention may be administered to a patient to localize cholesterol and lipoprotein areas by non-invasive imaging techniques.

Labeled antibodies of the invention may be contacted with biological samples in vitro assays systems to determine the presence and amount of cholesterol and the relative ration of various lipoproteins.

It shall be understood that the term "cholesterol", as used herein regarding the binding specificity of the antibodies of the invention, includes structural analogs of cholesterol that are bound by the antibodies. The invention may be further understood with reference to the following non-limiting examples. All publications cited herein are incorporated by reference.

Example 1: Induction of anti-cholesterol antibodies in mice after inoculation with cholesterol-rich liposomes.

Materials and Methods Reagents. 1,2-Dimyristoyl-sn-glycero-3-phosphocholine (DMPC), 1, 2 -dimyristoyl-snglycero-3-phosphoglycerol (DMPC), and cholesterol (5-cholesten-3-beta-ol) were obtained from Avanti Polar Lipids, Inc. (Alabaster, AL).

Cholesteryl beta-epoxide (cholestan-5-beta,6-beta-epoxy-3beta-ol); 7-dehydrocholesterol (5,7-cholestadien-3-beta-ol); epicholesterol (5-cholesten-3-alpha-ol); 7-ketocholesterol WO 97/21099 PCT/US96/19381 (5-cholesten-3-beta-ol-7-one); 4-beta-hydroxycholesterol (5-cholesten-3-beta-4-beta-diol); 7-beta-hydroxycholesterol (5-cholesten-3-beta-7-beta-diol); and lanosterol (8,24,(5alpha)cholestadien-4,4,14-alpha-trimethyl-3-beta-ol), were obtained from Steraloids, Inc. (Wilton, NH). Cholestene cholestenone (4-cholesten-3-one); cholesteryl alpha-epoxide (cholestan-5-alpha-6alpha-epoxy-3-beta-ol); cholesteryl hemisuccinate (5-cholesten-3-beta-ol hemisuccinate); cholesteryl myristate (5-cholesten-3-beta-ol myristate); cholesteryl oleate cholesten-3-beta-ol oleate); cholesteryl sulfate (5-cholesten-3-beta-ol sulfate, sodium salt); coprostane beta-cholestane); dihydrocholesterol (5-alpha-cholestan-3beta-ol); ergosterol (5,7, 2 2-cholestatrien-24-beta-methyl-3beta-ol); 7-alpha-hydroxycholesterol (5-cholesten-3-beta,7alpha-diol); and 25-hydroxycholesterol (5-cholesten-3were purchased from Sigma Chemical Company (St. Louis, MO). Primarily monophosphoryl lipid A from Salmonella minnesota, R595 lipopolysaccharide was prepared by List Biological Laboratories, Inc. (Campbell, CA). Most lipid stock solutions were made in chloroform, whereas some of the steroid stocks were kept in ethanol. All solutions were stored at 20 C.

Cholesterol liposomes. Cholesterol-rich multilamellar vesicles (MLV; 71 mol cholesterol) used for induction of antibodies to cholesterol were composed of DMPC, DMPG, and cholesterol in a molar ratio of 9:1:25 and contained the adjuvant monophosphoryl lipid A at 25 micrograms per 16- WO 97/21099 PCT/US96/19381 mole phospholipid. For antibody binding studies, liposomes having lower amounts of cholesterol or completely lacking lipid A were occasionally also prepared.

The lipids were aliquoted from stock solutions in chloroform into a pyrogen-free, round-bottom flask, and the solvent was removed by rotary evaporation. The lipids were further dried under high vacuum for 2 hours. The dry film was hydrated by adding sterile deionized water to achieve a 50 mM total phospholipid and vortexing until all the lipid was resuspended. The material was subsequently incubated for 2 hours at room temperature and occasionally vortexed. Aliquots of the liposomal preparations were lyophilized in vaccine vials and prior to use reconstituted in PBS to 10 mM with respect to total phospholipids.

To study anticholesterol antibody-induced aggregation of liposomes by flow cytometry and light microscopy, MLV prepared without lipid A were sonicated to obtain small unilamellar vesicles (SUV). One to two ml of an MLV suspension (diluted in PBS to 5 mM phospholipid) was transferred to a capped, glass culture-tube. Next, the lipid was sonicated under a nitrogen atmosphere to opalescence (four 15 minute periods with intermittent vortexing) in a bath-type sonicator (Laboratory Supplies, Inc., Hicksville, NY) at 30 The SW were used on the day of preparation.

Immunization to cholesterol. For induction of antibodies to cholesterol, groups of 5 male inbred mice (C3H/HeJ, C3H/HeOuJ, C57BL/6J, C57BL/LDLr-/-, C3H.MRLFaslpr, 17- PCT/US96/19381 WO 97/21099 DBA/2J, CBA/CaJ, CBA/CaHN-xid/J, A/J, BXSB/MpJYaa, NZB/B1NJ, HRS/J, BALB/cByJ, and BALB/cByJ-nu strain; Jackson Lab, Bar Harbor, ME) were immunized intraperitoneally with 100 microliters of the cholesterol-rich MLV (2.5 micromoles of cholesterol).

After 2 weeks, the mice were boosted and bled 1 to 2 weeks later from the tail vein to collect immune serum.

Pre-immune sera were collected just before the primary inoculation.

Monoclonal antibody to cholesterol and other antibodies.

The hybridoma cell line 2C5-6 (ATCC HB-8995), secreting a monoclonal IgM antibody that reacts with cholesterol (Swartz, G. et al., 1988, Proc. Natl. Acad. Sci. USA 15 85: 1902.), was a gift from Dr. C. R. Alving, Walter Reed Army Institute of Research. As described previously, collected ascites fluid from Balb/c mice was centrifuged and the supernatant after heat inactivation stored at -70 0 C. IgM was purified by affinity chromatography on a 20 mannan-binding protein column (Pierce, Rockford, IL).

o' Ascites fluid containing a monoclonal IgM antibody specific for phosphatidylinositolphosphate was also obtained from Dr. C. A. Alving (Alving, C.R. Alving, et al., 1987, Clin. Exp. Immunol.: 69, 403.). Isotype control, non-specific murine IgM was purchased from Southern Biotechnology Associates, Inc., Birmingham,

AL.

Lipid extraction from lipoproteins, thin layer A chromatography and transfer to PVDF membranes for 18- WO 97/21099 PCT/US96/19381 immunodetection. Total lipid was extracted from the human lipoproteins VLDL/IDL (1.006-1.019 g/cc), LDL (1.019-1.063 g/cc), and HDL (1.063-1.21 g/cc) (Organon Teknika Corp.-CAPPEL Res. Product, Durham, NC) according to the method of Folch et 1957, J. Biol.

Chem. 226: 497.). Upon extraction, lipids recovered in the chloroform phase were dried, resuspended in pure chloroform, and stored at Lipoprotein extracts and pure, lipid standards were separated on TLC silica gel plates and detected according to the procedure of Aniagolu, S. J. Green, et al., 1995, J.

Immunol Methods 182: For immunodetection of cholesterol on TLC plates, the separated lipids were transferred to hydrophobic polyvinylidenefluoride membranes (PVDF), incubated with antibodies and stained as described (Aniagolu, et al., 1995, J. Immunol Methods 182: Cholesterol, lipoprotein, liposome, and lipid A ELISA. To quantitate the amount of anti-cholesterol activity in serum or ascites fluid, the PVDF-ELISA was employed as described previously (Aniagolu, et al., 1995, J. Immunol Methods 182: Briefly, 5-10 micrograms cholesterol in 50 microliters ethanol was applied directly to microtitration plates containing a PVDF membranes (Immobilon-P; Millipore Corp., Bedford, MA). After drying, the wells were treated for 1 hour with FCS blocking buffer consisting of 10 heat-inactivated FCS (Biofluids, Inc., -19- WO 97/21099 PCT/US96/19381 Rockville, MD) in PBS. Serum samples were serially diluted in blocking buffer, and 100 microliters was added to duplicate wells. After 1 hour incubation with slow agitation, wells were washed four times with the blocking buffer, followed by a 1 hour incubation with a 1/1500 dilution of peroxidase conjugated anti-mouse IgG (H+L chain; BioRad, Richmond, CA) or IgM (Kirkegaard Perry, Gaithersburg, MD). Membranes were then washed four times with PBS, and incubated with 100 microliters of 2,2'-azino-di[3-ethylbenz-thiazoline-6-sulfonic acid] solution (ABTS peroxidase substrate system, Kirkegaard Perry, Gaithersburg, MD) and monitored for 10-15 minutes until adequate color change. The reactant microliters) was then transferred to a clear-plastic microtiter plate and the absorbance at 405 nm determined using a Bio-Rad microplate reader. As indicated, the PVDF-ELISA was occasionally also used for the determination of anti-phospholipid and anti-steroid activity.

For the lipoprotein and liposome ELISA, the wells of polystyrene immulon-2 plates (Dynatech Labs, Inc., Chantilly, VA) were coated for 1 hour with 50-100 microliters of a VLDL/IDL, LDL, HDL, or liposome dilution in PBS, containing the indicated amounts of protein or lipid. For reasons of comparison, PVDF plates were occasionally also used to assess binding to either lipoproteins or liposomes. After coating, the wells were washed and further processed as described for the cholesterol ELISA. In some lipoprotein ELISA experiments, the wells washed extensively after WO 97/21099 PCT/US96/19381 determination of the absorption, to remove residual chromophore. Next, the cholesterol was extracted from individual wells by three washes with 100 ml isopropanol.

The combined extracts of 12 wells were pooled and dried in a Speed Vac concentrator and total cholesterol was assayed after resuspension in 50 microliters isopropanol. Free cholesterol was calculated from the free over total cholesterol ratios measured for each of the lipoprotein stock solutions used in the ELISA.

The lipid A ELISA was performed as described by Freudenburg, et al., 1988, Infection 7: 322., using polystyrene immulon-2 plates. The wells were coated with micrograms lipid A in 50 microliters ethanol and after drying, the plates were further processed as described for the cholesterol ELISA.

In this example, antibody titers are defined as the reciprocal of the highest serum dilution having an absorbance of at least two standard deviations above the background level. The background was determined by replacing the antiserum or purified primary (monoclonal) antibody with blocking buffer.

Flow Cytometry and photomicroscopy. Lipoprotein or liposome dilution (10 microliters) containing the indicated amounts of protein or lipid were admixed with microliters of a purified 1 mg/ml solution of anti-cholesterol monoclonal IgM, diluted in 300 ul saline (sterile, tissue culture grade, Sigma Chem. Co., St. Louis, -21- WO 97/21099 PCT/US96/19381 MO), and incubated for 30 minutes in polypropylene tubes at 4°C. As a control, the lipoproteins were also incubated with 10 ug of control, non-specific IgM. Subsequently, the samples were analyzed by FACSort (Becton-Dickinson, San Jose, CA) flow cytometery. In order to diminish background noise, double distilled water was used as sheath fluid. Beads of 1, 2, and 6.8 micrometers were employed to check signal linearity and as particle size standards. In addition, plain saline, lipoprotein, liposome, and monoclonal antibody signals were collected to establish reagent signal background controls. The logarithmic, amplified signals of FSC and SSC from 10,000 events were collected, and the list mode acquired data were later analyzed with LYSYS II and CELLQuest programs (Becton-Dickinson).

Aliquots of the incubations were also transferred to a 96-well polystyrene plate, lipoprotein aggregates allowed to settle, and visible structures photographed using Kodak film (ASA 400) in an Olympus CK-2 inverted-photomicroscope with Hoffman contrast optics.

Miscellaneous methods. Phospholipid concentrations in stock solutions and in liposomal suspensions in water were determined by assessing phosphorous according to the method of Bartlett (Bartlett, G. 1959, J. Biol. Chem.

234, 466.). Total cholesterol and free, unesterified cholesterol in lipoproteins and lipid extracts (resuspended in 50 ul isopropanol) were determined using enzymatic -22- WO 97/21099 PCT/US96/19381 colorimetric methods purchased as kits from Wako Chemicals USA, Inc. (Richmond, VA).

Results Induction of an antibody response to cholesterol. As shown previously (Swartz, G. et al., 1988, Proc. Natl. Acad.

Sci. USA 85: 1902.; Freudenberg, M.A.,et al., 1988, Infection 7: 322.), intraperitoneal inoculation of Balb/cByJ mice with lipid A-containing, 71 mol cholesterol liposomes induces a humoral response to pure cholesterol.

Compared to pre-immune serum, a 100-fold increase in anti-cholesterol activity was obtained in the serum of vaccinated mice (Figure The immune anti-cholesterol antibody titers were usually in excess of 50,000, whereas pre-immune titers amounted to less than 800. The pre-immune titers do not represent anti-cholesterol activity and are due to non-specific antibody adsorption. The use of detergents to reduce non-specific binding in sterol ELISA is not -possible because of the high solubility of such antigens in detergent solutions.

The density of cholesterol in the liposomal membrane considerably influenced the magnitude of the immune response. Liposomes containing amounts of 56 mol cholesterol or more, stimulated optimal activity (Figure The presence of lipid A in the liposomes was found to be essential for the induction of an anti-cholesterol response. Without lipid A, the titers obtained with 71 mol cholesterol liposomes were not different from the pre-immune values (not shown). Only IgM anti-cholesterol -23- WO 97/21099 PCT/US96/19381 antibodies were detected in the sera of mice vaccinated with the cholesterol-rich liposomes. IgG, IgA, and IgE were not detectable.

A variety of other inbred mouse strains (C3H/HeJ, C3H/HeOuJ, C57BLt6J, C57BL/LDLr-/-, C3H.MRL-FaslPr(4), DBA/2J, CBA/CaJ, CBA/CaHN-xid/J, A/J, BXSB/MpJYaa, NZB/B1NJ, HRS/J), including nude mice (BALB/cByJ-nu), were also found to respond to 71 mol cholesterol liposomes. Both males and females responded similarly (data not shown). No serum-associated anti-cholesterol activity was detectable after vaccination of C3H/HeJ mice, which is consistent with this strain's low responsiveness to lipid A. However, substituting the liposomal lipid A for other adjuvants, such as liposome-encapsulated E. coli heat-labile enterotoxin (Clements, J. et al., 1988, Vaccine 6, 269.), resulted in comparable anti-cholesterol levels in the sera of these animals (data not shown).

Besides an anti-cholesterol response, we also observed weak anti-DMPC, but no anti-DMPG activity in the sera of mice vaccinated with the 71 mol liposomal cholesterol formulation (Figure As for cholesterol, the anti-DMPC antibodies were of the IgM isotype. In addition, the immune sera contained significant amounts of antibodies to the adjuvant lipid A, which were found to be of both the IgG and IgM isotypes (Figure The observation that lipid A incorporated into phospholipid vesicles is able to induce antibodies to both liposomal phospholipids and itself is consistent with previous studies (Dancey, G. et al., -24- WO 97/21099 PCT/US96/19381 1977, Immunol. 119, 1868; Schuster, B. et al., 1979, J.

Immunol, 122: 900.

Specificity of the anti-cholesterol antibodies. First, the activity of an anti-cholesterol monoclonal antibody to cholesterol was tested using the PVDF-ELISA. This IgM antibody (2C5-6) was previously generated against 71 mol cholesterol liposomes (Swartz, G. et al., 1988, Proc.

Natl. Acad. Sci. USA 85: 1902.). From the results presented in Figure 4, we calculated an anti-cholesterol titer of approximately 100,000 for the purified, 1 mg/ml IgM monoclonal stock solution. It is noteworthy that the anti-cholesterol monoclonal IgM did not cross react with either the liposomal phospholipids DMPC and DMPG or lipid A (not shown).

In a subsequent set of experiments, the binding of both polyclonal antiserum and the monoclonal antibody to various steroids and derivatives thereof was assessed by substituting cholesterol for the indicated steroids in the

PVDF-ELISA.

0 M M WO 97/21099 WO 9721099PCT/US96119381 Table 1: Immunoreactivity of steroids and derivatives with anti-cholesterol antibodies Immunoreactivea Cholesterol (5-cholesten-3beta-ol) Dihydrocholesterol (5alpha-cholestan-3-beta-ol) 7-Dehydrocholesterol (5,7-cholestandien-3beta-ol) Ergosterol (5 ,22-cholestatrien-24beta-methyl-3-beta-ol) Cholesteryl aipha-epaxide (cholestan-5-alpha,6alpha-epoxy-3-beta-ol) Chalesteryl beta-epaxide (cholestan-5-beta,6beta-epoxy-3-beta-ol) 7-Ketocholesterol (5-cholesten-3beta-ol-7-one) 4-beta-Hydroxycholesterol (5-cholesten-3-beta, 4-betadial) 7-alpha-Hydroxycholesterol (5-cholesten-3-beta, 7alpha-diol) Non.-immunoreactiveb Epicholesterol (5-cholesten-3-alpha-ol) Cholestene Chalestenone (4-cholesten-3-one) Coprostane cholestane) Cholesteryl sulfate (5-cholesten-3-beta-ol sulfate) Chalesteryl myristate (5-cholesten-3-beta-ol myristate) Cholesteryl oleate (5-cholesten-3-beta-ol aleate) Chalesteryl hemisuccinate (5-cholesten-3-beta-ol hemnisuccinate) Lanosterol (8,24,5aipha-cholestadien 4,4,14alpha-trimethyl-3-beta-ol) 7-beta-Hydroxycholesterol (5-cholesten-3-beta, 7-betadial) -Hydraxycholesterol (5-cholesten-3-beta- a Equivalent amounts (10 micrograms/well) of the indicated lipids diluted in ethanol t< 2 chloroform) were dried onto PVDF-ELISA plates. The binding of both poly- and monoclonal anti-cholesterol antibodies was then assessed as described in the Experimental Procedures. b Immunoreactive: titers range from 26- WO 97/21099 PCT/US96/19381 3,000-100,000; non-immunoreactive: titers 1000. With cholesterol as the antigen, the titers of the polyclonal antiserum and the monoclonal antibody solution employed in this experiment amounted to approx. 50,000 and 100,000, respectively.

As shown in Table 1, the specificity of the antibodies was to unesterified cholesterol and structurally similar sterols containing a 3-beta-hydroxyl group ergosterol). In contrast, anti-cholesterol binding activity was significantly diminished if the 3-beta-hydroxyl was altered by epimerization epicholesterol), by substitution with hydrogen cholestene), by oxidation to a keto-group cholestenone), or by esterification cholesteryl sulfate or cholesteryl oleate). Changes in the structure of the P-ring (degree of saturation, dihydrocholesterol or 7-dehydrocholesterol; or degree of oxidation, i.e., cholesteryl epoxide or 7-ketocholesterol) did not seem to affect the affinity of the anti-cholesterol antibodies. The introduction of an additional beta-hydroxyl group at the 4-position of the A ring (4-beta-hydroxycholesterol) also did not change the efficiency of antibody binding to the sterol. On the other hand, lanosterol, which possesses two methyl groups at the 4-position of the A-ring, did not significantly interact with the antibodies. Taken together, these results suggest that a 3-beta-hydroxyl group is required for significant binding of both the poly- and monoclonal antibodies to such steroids Depending on the nature and number of functional groups, additions to the A-ring may interfere with antibody binding, whereas the -27i_*ll _X__ii~ WO 97/21099 PCT/US96/19381 structure of the P-ring appears to be of lesser importance for the interaction.

Anti-cholesterol antibody interaction with human lipoproteins as detected by ELISA. Inasmuch as previous studies suggested a possible role for antibodies to cholesterol in dietary induced atherosclerosis (Bailey, J. M., et al., 1964, Nature: 201, 407; Bailey, J. et al., 1967, in The Reticuloendothelial System and Atherosclerosis. N. R.

Di Luzio, et al., eds, Plenum Press, New York, pp. 433.), we next examined whether the anti-cholesterol antibodies would interact with the human lipoproteins VLDL/IDL, LDL, and HDL. As a first approach, the binding of anti-cholesterol serum to lipoproteins adsorbed onto ELISA plates was determined. The wells of polystyrene plates were coated with various dilutions (3 to 100 micrograms protein per well) of each type of lipoprotein. After removal of unbound lipoprotein, the binding of cholesterol antiserum to each of the lipoproteins was assessed. Figure 5 shows that at all the protein concentrations used to coat the wells, the anti-cholesterol antibodies preferentially and unexpectedly bound to VLDL/IDL and LDL, and not to HDL. Similar results were obtained for binding of the monoclonal anti-cholesterol antibody to these lipoproteins (not shown).

Since neither the polyclonal antiserum or mAb bound to HDL, we sought to determine if HDL was present in the ELISA plate wells in the same concentration as the other lipoproteins. We measured the amounts of cholesterol -28- WO 97/21099 PCT/US96/19381 adsorbed to the wells after performing the ELISA. Wells were coated with dilutions of each lipoprotein so that similar amounts of free cholesterol became stably adsorbed to the wells.

To rule out this possibility, we measured the amounts of cholesterol adsorbed to the wells after performing the ELISA. In the next experiment the wells were coated with such dilutions of each lipoprotein, that similar amounts of free cholesterol became stably adsorbed to the wells.

Figure 6 shows that, even with a slight excess of free-HDL-cholesterol bound to the wells, antibody titers for binding of the anti-cholesterol monoclonal antibody to both VLDL/IDL and LDL were in excess of 20,000, whereas negligible interaction was observed with HDL (titer <800).

Similar results were obtained when anti-cholesterol immune serum was tested under these conditions (not shown).

The failure of anti-cholesterol antibodies to bind to intact HDL was not due to the inability of the HDL cholesterol per se to interact with the antibody, because total lipid extracts from VLDL/IDL, LDL, and HDL (adjusted to equivalent amounts of total cholesterol) reacted equally well with immune serum in the PVDF-ELISA (Figure 7A). On the other hand, when PVDF-ELISA plates were incubated with matched amounts (based on total cholesterol) of the intact lipoprotein preparations, the anticholesterol serum interacted again exclusively with VLDL/IDL and LDL (Figure 7B). To verify that anticholesterol immunoreactive serum recognized unesterified cholesterol in the lipoprotein extracts, the lipids were 29- WO 97/21099 PCT/US96/19381 separated by thin-layer chromatography, transferred to a PVDF membrane, and then probed with anti-cholesterol serum. As shown in Figure 8, the unesterified cholesterol band of each of the lipoprotein extracts reacted strongly immunopositive, whereas the phospholipid bands showed weak staining, and cholesterol ester and triglyceride were not reactive at all. The latter observations are in agreement with the results of PVDF-ELISA studies which demonstrated that the anti-cholesterol antiserum contains some anti-phospholipid DMPC) activity (Figure 3), but does not react with the major constituents of the hydrophobic core of lipoproteins, cholesterol esters (Table I) and triacylglycerols (not shown). In conclusion, both human VLDL/IDL and LDL interacted with the murine anti-cholesterol antibodies, whereas no significant binding to HDL could be demonstrated using the ELISA.

This was a surprising and unexpected result.

Effect of temperature on the interaction of anti-cholesterol antibodies with liposomal, lipoprotein-associated, and pure cholesterol. The physical properties of membranes are dependent on environmental factors such as temperature.

For instance, both the rotational and lateral mobility of individual lipid molecules increases with increasing temperature. Inasmuch as the anti-cholesterol antibody binding studies presented in the previous sections were all performed at room temperature, we next examined the effects of low (4 and physiological temperature (37 °C) on antibody binding to pure cholesterol and cholesterol in WO 97/21099 PCT/US96/19381 the membrane of liposomes and lipoproteins (Figure 9).

PVDF plates were used for all the antigen formulations, and the ELISA was performed at room temperature, except for the primary antibody binding step and the subsequent washes, which were done at either 4 'C or 37 The binding of the anti-cholesterol antibodies to cholesterol itself was found to be similar at 4 'C and 37 'C (Figure 9A). However, with either 71 mol cholesterol liposomes (not containing lipid A) or VLDL/IDL coated onto the wells, binding decreased about 50 and 80 respectively, when the temperature was increased from 4 to 37 'C (Figures 9B and 9C). A similar, but even more drastic decrease in antibody binding was observed with LDL at 37 'C (not shown). The binding of pre-immune serum was not found to be significantly affected by the temperature in these experiments (Figure These observations may suggest that, in contrast to pure cholesterol, the increased mobility of membrane-associated cholesterol at 37 'C diminishes stable attachment during the interaction between antibody and antigen.

Anti-cholesterol antibody interaction with lipoproteins and liposomes in suspension. The interaction between lipoproteins and anti-cholesterol antibodies was also examined in suspension, employing both flow-cytometric and light-microscopic techniques. When analyzed by flow cytometry, the interaction of the anti-cholesterol monoclonal antibody with human VLDL at 4 'C resulted in an increase in medium size of the particles (Figure -31i_ I IL_=~ILml~ WO 97/21099 PCT/US96/19381 Similar results were obtained with LDL (Figure On the other hand, the antibodies did not change the size distribution of HDL (Figure 10C). Control, non-specific IgM had no effect on the distribution profile of the tested lipoproteins. Binding of the anti-cholesterol antibodies to VLDL and LDL presumably resulted in aggregation of the lipoproteins, as indicated by the change in size distribution.

When examined microscopically, large (up to micrometers diameter) spherical, droplet-like structures were found in suspensions of VLDL/IDL and LDL incubated with the anti-cholesterol monoclonal IgM antibody (Figure 11A). Control IgM, or an irrelevant antiphosphatidylinositolphosphate monoclonal antibody had no effect (data not shown). No visible structures could be detected in the suspensions containing HDL and anticholesterol antibodies (Figure 11A). The smooth appearance of the obtained structures suggests that these particles are not simple aggregates of individual VLDL or LDL particles, but are the result of a fusion or a fusion-like process between a significant number of lipoprotein particles (the individual lipoproteins have diameters between 0.01 and 0.075 micrometers (Bradley, W. et al., 1988, in Biology of Cholesterol, P. L. Yeagle, ed., CRC Press, Boca Raton, FL, pp. It is of interest that the structures obtained during incubation at 4 °C were not stable when transferred to room temperature or to 37 oC.

At room temperature, the smooth structures slowly decreased in size and finally dissolved completely. This -32- WO 97/21099 PCT/US96/19381 reaction was found not to be reversible upon subsequent cooling to 4 oC.

To investigate the interaction of anti-cholesterol antibodies with cholesterol containing membranes in a less complex system, 71 mol cholesterol liposomes were sonicated to decrease their size (diameter less than 0.1 micrometer). The obtained small unilamellar vesicles (SUV) bind anti-cholesterol antibodies (not shown) similar to unsonicated, multilamellar vesicles (Figure 9B). When the SUV were incubated with the anti-cholesterol IgM at 4 °C and analyzed by flow cytometry, the vesicles were found to have increased in size, the majority having dimensions larger than 1 micrometer (not shown). Under the microscope, large vesicular structures of various sizes (2 micrometers to 30 micrometers in diameter), most probably the product of antibody-induced fusion of the SUV, were easily distinguishable (Figure 11B). In contrast to the structures obtained with VLDL and LDL, the SUV products did not disintegrate at higher temperatures, but even fused into larger, more complex structures upon warming (not shown). The large vesicular structures could also be formed from the SUV at room temperature or at 37 Incubation of the SUV in the presence of control IgM did not result in a detectable increase in size when assessed by flow cytometry or in the appearance of visible structures under the microscope.

Similar phenomena as described for the interaction of the anti-cholesterol monoclonal antibody with the lipoproteins and liposomes were observed using anti-cholesterol -33- WO 97/21099 PCT/US96/19381 antiserum (not shown). The results obtained in this section indicate that the interaction of anti-cholesterol antibodies with cholesterol-containing, lipid mono- or bilayers may, depending on the conditions, result in destabilization and subsequent fusion of these membranes.

DISCUSSION

We observed that anti-cholesterol antibodies: (1) recognize the 3-beta-hydroxyl group of sterols, bind to the lower density lipoproteins, VLDL/IDL and LDL, but not HDL, and induce a fusion-like reaction upon binding to these lipoproteins. Only anti-cholesterol antibodies of the IgM subtype could be detected in the sera of mice boosted with lipid A containing, cholesterol-rich liposomes. In contrast, naturally occurring antibodies to cholesterol observed in humans were reported to be of both the IgM and IgG subtype. Swartz, G. et al., 1988, Proc.

Natl. Acad. Sci. USA 85: 1902; Alving, et al., 1989, Bichem. Soc. Trans. 17: 637. The observation that the liposomal cholesterol formulation also induced anticholesterol antibodies in nude, athymic mice characterizes the response as thymus-independent (Pantelouris, E. M., 1968, Nature 217, 370.

Moreover, the fact that CBA/CaHN-xid mice also responded, suggests that CD5+ B lymphocytes are not essential for the induction of anti-cholesterol antibodies (Hayakawa, et al., 1986, Eur. J. Immunol. 16, 450).

These results indicate that the specificity of the anticholesterol antibodies was for steroids containing a free -34- WO 97/21099 PCT/US96/19381 beta-hydroxyl group on the 3-position of the steroid nucleus. This specificity is similar to the requirements for the binding of both Semliki Forest virus and bacterial cytolysins to cholesterol-containing membranes (Kielian, M. et al., 1984, J. Virol. 52: 281.; Alouf, J. 1980, Pharmacol. Ther. 11, 611.).

In the presence of lipid A, more than 50 mol of cholesterol in the liposomes was necessary to obtain optimal anti-cholesterol antibody levels in mice. Conversely, immune serum and also a monoclonal antibody to cholesterol were found to bind only to cholesterol-rich liposomes or to pure cholesterol. For example, when binding of the monoclonal antibody to liposomes was assessed by complement-induced release of glucose, significant release occurred at 50 mol of cholesterol, whereas 43 mol was insufficient. Optimal release was induced with liposomes containing well over 50 mol of cholesterol (Swartz, G. M.,et al., 1988, Proc. Natl. Acad.

Sci. USA 85: 1902.). Similar results were obtained when monoclonal antibody binding was measured by ELISA, using liposomes with varying amounts of cholesterol as the capture antigen (results not shown). These observations suggest that for optimal binding of the antibodies to cholesterol-containing phospholipid bilayers consisting of DMPC and DMPG, the molar ratio of cholesterol over phospholipid has to amount to at least 1. At such ratios, clusters of free cholesterol exist in membranes (Collins, J.

et al., 1982, J. Lipid Res. 23, 291).

WO 97/21099 PCT/US96/19381 Binding of the anti-cholesterol antibodies to pure cholesterol attached to an ELISA plate appeared not to be affected when the temperature was raised from 4 to 37 °C.

On the other hand, binding at 37 °C was diminished when the antigen was presented in a liposomal bilayer or in the monolayer of lipoproteins. The decrease observed with the 71 mol cholesterol liposomes, which by definition contain clusters of cholesterol in the membrane (Collins, J.

et al., 1982, J. Lipid Res. 23, 291), may result from the increased mobility of the lipid molecules at 37 °C.

Finally, antibody-lipoprotein binding at 37 °C may result in the release of antibody-lipoprotein fragments from the ELISA plate. Therefore, it is possible that the ELISA at 37 °C underestimates the actual interaction of anticholesterol antibodies with the fixed lipoprotein.

The interaction of both liposomes and the lower density lipoproteins with anti-cholesterol antibodies in suspension resulted in membrane destabilization. The liposomes appeared to have fused into vesicles with considerably increased diameters, whereas the lipoproteins were transformed into large, spherical structures, presumably by a fusion-like process.

Immunizing rabbits with cholesterol-ester conjugated to serum albumin or with heterologous LDL was reported to reduce the development of dietary-induced aortic atherosclerosis (Sachs, 1925, Biochem. Z. 159: 491; Bailey, J. M.,et al.,1967, in The Reticuloendothelial System and Atherosclerosis. N. R. Di Luzio, et al., eds., Plenum Press, New York, p. 433). Antibodies to the conjugated -36- WO 97/21099 PCT/US96/19381 cholesterol ester were detectable after vaccination with the cholesterol-ester-albumin complex (Bailey, J. 1967, Rev. Atheroscl. 9: 204.

EXAMPLE 2: Binding of murine monoclonal and polyclonal IgM antibodies at 37 degrees Celsius to lipid particles isolated from advanced human athromas, but not to LDL derived from serum.

METHODS

Tissue and Serum Samples. Human aortic tissue was obtained from either autopsy of individuals within 24 hours of death or from surgery samples during aortic aneurysm repair and frozen (at -70 0 C) immediately. In order to obtain grossly normal tissue, abdominal aortic samples from pediatric cases and thoracic or aortic arch material from patients with no to minimal disease was used.

Samples were grouped into three categories: as grossly normal or no lesion, minimal presence of plaques or fatty streaks, or severe disease or ulcerating lesions. Table 2 summarizes the source of tissue assessed for anti-cholesterol IgM antibody.

-37- WO 97/21099 PCT/US96/19381 Table 2.

Clinical summary on the human aortic tissue evaluated for antibody to unesterified cholesterol Extent of disease Source of tissue age (yrs) gender cause of death Normal or no lesions 3 wks female birth defect 9 female lymphoma (AML)*& male motor vehicle acident

(MVA)*

31 female sepsis*# 49 male general surgery 17 female MVA Fatty streak male MVA male MVA 31 female sepsis# 49 male general surgery 44 male cerebral aneurism Ulcerating lesion@ 47-74 male aortic aneurism repair *tissue from thoraic aorta; all others tissue from aortic arch.

tissue from same individual, except from two different regions of the aorta.

tissue from 7 males consisting of ages: 47, 61, 64, 68, 70, 74, and &MVA- motor vehicle accident, ARF- acute renal failure, GSW- gun shot wound, AML-acute myelocytic leukemia.

Extraction of Immunoglobulin from Atheroma. Aortas were homogenized in approximately 2-5 ml of PBS per gram of wet tissue. Immunoassay on these materials was normalized for the extracts protein concentration.

Isolation and Purification of Aortic Liposomes. The method followed was that of Chao, et al., 1990, A. J. Path., -38- WO 97/21099 PCT/US96/19381 136: 169. Briefly, aortic tissue dissected free of adventia was rinsed in PBS, blotted dry, and weighed. It was then finely minced in approximately 6 ml per gram wet tissue in chilled normal saline supplemented with 0.1% EDTA, 0.05% glutathione, and 0.02% NaN3. After two 1500 x g centrifugations, the supernatant was passed through a 0.45 micron filter and was concentrated on a stirred cell concentrator with a 10,000 MW cut-off. After dialyzing overnight against 1.5 M NaCl (pH 7.4) containing 0.1% EDTA, 0.05% glutathione and 0.02% NaN3, the extract was subjected to gel filtration chromatography (Bio-gel A- Richmond CA) using a 1.6 x 50 cm column as described. Next, The void volume fractions containing cholesterol were pooled and subjected to density gradient centrifugation. The gradient was constructed from bottom to top using 3 ml of 1.100 g/ml, 3 ml of void volume fraction (density adjusted to 1.061 g/ml), 3 ml of 1.019 g/ml, and 2.5 ml of 1.006 g/ml NaCl solution and centrifuged at 170,000 g for 22 hours at 40C. Fractions containing liposomes of various densities were located visually and collected by aspiration from the top. Fractions were then dialyzed against PBS and 0.02% NaN3.

ELISA for anti-cholesterol IgM binding activity to cholesterol and liposomes. To quantify antibodies specific for these antigens, PDVF microtiter plates (Immobilon P Millipore) were either spotted with 10 microliters of a 1 mg/ml ethanol solution of cholesterol or cholesterol oleate and allowed to air dry. If coated with liposomes the plates were washed 4 times with 250 ul of PBS per well. Next, -39- WO 97/21099 PCT/US96/19381 the plates were blocked with 10% heat inactivated fetal calf serum and then incubated with primary and secondary antibody as described (Aniagola et al., 1995, J.

Immunol. Methods 182: 85). In all cases specific absorbance was calculated by subtracting a no-antigen or cholesterol oleate negative control from that of the test material.

Measurement of IgM levels. PDVF plates were coated with anti-total human immunoglobulin. After blocking for nonspecific binding, titration curves of serial dilutions of the aortic homogenates were then constructed using either affinity purified IgM specific anti-sera as the secondary antibody (Kirkegard and Perry). Results of the curve were then compared with known commercial standards to quantify the amount of IgM. Anti-cholesterol IgM concentration in the aortic homogenates was calculated by subtracting the quantity of IgM bound to the negative control well from that bound to the cholesterol coated well.

The same standard curve noted above was used for determining the quantity of both the cholesterol specific and non-specific IgG in this calculation. When possible these assays were performed in the same microtiter trays to control for plate to plate variation.

Murine antibody. The production of the murine anticholesterol anti-sera and the murine anti-cholesterol monoclonal antibody (2C5-6) are described in Swartz, G.

et al., 1988. PNAS, 85:1902.

_Ciill___lLL_/I~----1II~XI~IIX ~1II~II WO 97/21099 PCT/US96/19381 Isolation ofLDL. Plasma was obtained from healthy fasted subjects. LDL was isolated by sequential ultracentifugation as described in Patsch, J. et al., 1986, in Methods in Enzymology, Vol. 129, J. Albers, et al., eds., Academic Press, New York, NY, p. 37.

Measurement of Protein and Cholesterol. For most immunoassays total cholesterol measurements and total protein were run on a Hitachi 717 clinical laboratory using Bohreinger Manheim reagents. This clinical system has a sample requirement of approximately 20 microliters for both assays, and allowed for more immunoassays to be run per extraction.

RESULTS

To determine whether autoantibodies to cholesterol were sequestrated in the blood vessel walls from athrogenic aortas, we extracted antibodies from the tissue and measured for anti-cholesterol IgM binding activity. A total of 18 human aortas were examined for the presence of anticholesterol antibodies (Table The severity of disease was graded as either no lesion, fatty streaks, or ulcerating lesions. Both the recovery of anti-cholesterol IgM and total IgM increased with the severity of disease. However, as the severity of disease increased, the amount of specific anti-cholesterol IgM relative to total IgM increased: 4-times as much specific anti-cholesterol IgM was found in the ulcerating lesions compared to the no lesion group; 0.22 vs.

0.06% specific anti-cholesterol antibody IgM per total Ig, -41s__l__llC__l_~jlllI_ ~~-L-tll*l~tl WO 97/21099 PCT/US96/19381 respectively Although there was no significant difference between the fatty streak and the no lesion group a marked distinction was found when the fatty streak and ulcerating lesion groups were compared to the no lesion group (p<0.001).

To assess whether lesion-derived lipid particles were immunoreactive, particles were isolated and assessed for specific anti-cholesterol binding activity with both polyclonal and monoclonal antibodies. To assess binding activity, both the low density (D<1.01) and high density (D>1.02) extracellularly located lipid particles found in advanced human lesions were isolated and absorbed to ELISA plates. The density for both particles have a reported unesterified cholesterol to phospholipid molar ratio of 2.5:1 (Chao et al., 1990, A. J. Path., 136: 169). This ratio is in the range which has previously been described to support optimal binding of murine anticholesterol antibodies to reconstituted 71 mol cholesterol liposomes (Swartz, G. 1988, PNAS. 85:1902). Serum from mice previously immunized with 71% mol cholesterol liposomes containing the adjuvant, MPLA, reacts with esterified and unestified cholesterol lipid particles isolated from human lesions at 37oC, whereas significantly less activity was found with LDL from the same subject. Preimmune serum did not react under the same conditions.

The murine monoclonal anti-cholesterol antibody selectively binds to both of the lesion-derived lipid particles at 370C, whereas no reactive was observed with LDL under the same conditions. Collectively, these data suggest that -42- WO 97/21099 PCT/US96/19381 lipid particles derived from human athroma are uniquely immunoreactive with a specific Ig that recognizes unesterified cholesterol.

DISCUSSION

The results reported here indicate that polyclonal and monoclonal antibodies specific for cholesterol and related sterols can bind to cholesterol-rich liposomes extracted from advanced athromas, but not to LDL derived from serum. This gives immunological evidence for the presence of a unique interaction between Ig and high concentrations of unesterified cholesterol found in the aortic liposome preparations (Hui, S. 1988, in The Biology of Cholesterol, P. L. Yeager, Ed. CRC Press. Boca Raton, Fl. p. 213.), and may indicate that the biological response to this structure contributes to atherogenesis. Previous studies have detected both the presence of an anti-cholesterol Iglike binding activity in normal human sera, and the ability of human sera to initiate the complement dependent lysis of high cholesterol liposomes (Alving et al., 1977, J.

Immunol. 118:342.). The data presented here suggest that anti-cholesterol IgM may correlate with atherosclerotic disease, but not with plasma cholesterol levels.

Example 3: An ELISA for detection of anti-sterol antibodies using polyvinylnitrocellulose and antisterol antibodies -43- C~___ilYX_ _IIYC~_ II~ -I WO 97/21099 PCT/US96/19381 The ELISA plate uses a polyvinylnitrocellulose composite material that can resist organic solvent chloroform/ethanol) which are necessary for solubilizing sterols. Aniagolu, et al., 1995, J. Immunol. Methods, 182:85. The initial coating of the sterol-antigen is done in the presence of a chloroform/ethanol solution. Due to the hydrophobic nature of the nitrocellulose composite, cholesterol and ergosterol tightly binds allowing the replacement of the organic solvent with aqueous buffer without causing the sterol to precipitate. Once in a physiological buffer, the procedure is similar to that of a conventional protein-based ELISA. In using this method we have detected antibodies to cholesterol in serum of BALB/c mice inoculated with liposomes containing 71% cholesterol and 2 ug monophosphoryl lipid A as shown in Figure 12 below. Each point is the mean of three determinants S.D.

Compared to other assays, this approach is the most sensitive to date. As shown in Figure 13, 600 ng of immobilized cholesterol is needed on the polyvinyl nitrocellulose for detection of antisera to cholesterol. In this experiment, dilution of immune serum were added to wells containing varying concentrations of cholesterol, as indicated on the horizontal axis (see Figure 13).

Experimental results also show that ergosterol can complex tightly with the polyvinyl nitrocellulose and is reactive to anti-sera from mice previously inoculated with liposomal- -44- WO 97/21099 PCT/US96/19381 ergosterol containing monophosphoryl lipid A, whereas sera from naive mice is not reactive to ergosterol.

The proposed ELISA, using immobilized sterol on polyvinyl nitrocellulose, is superior to the traditional radioimmunoassays and bioassays because it does not require a radioligand; the ligand does not have to be chemically modified or cross-linked for binding; insolubility of the ligand is of no concern; it is equally as sensitive and reproducible as a radioimmunoassay; the assay is both quantitative and qualitative; it is a quicker and cheaper assay than the RIA and other traditional methods; it is direct and can be converted for use for any hydrophobic sterol or steroid; and it is not biologically hazardous.

WO 97/21099 PCT/US96/19381 Modifications of the above-described methods and device are contemplated by the invention, and meant to fall within the scope of the appended claims.

-46-

Claims (3)

1. A method of selectively determining the sum total amount of cholesterol in Very Low-Density Lipoprotein (VLDL), Intermediate Density Lipoprotein (IDL) and Low Density Lipoprotein (LDL), but excluding cholesterol in High Density Lipoprotein (HDL), in a biological sample, comprising the steps of: contacting a biological sample with an anti-cholesterol antibody capable of selectively detecting cholesterol in VLDL, IDL and LDL, but excluding cholesterol in HDL; measuring the formation of cholesterol-antibody complexes, and thereby selectively determining the sum total amount of cholesterol in VLDL, IDL and LDL, but excluding cholesterol in HDL, in the biological sample. 15

2. A method of claim 1 wherein the biological sample is derived from an atherosclerosis lesion. a

3. A method of claim 1 or claim 2, in which the antibody is produced by the hybridoma deposited at the American Type Culture Collection under Designation Number ATCC HB-8995. S4. A method of selectively determining the sum total amount of cholesterol in Very Low-Density Lipoprotein (VLDL), Intermediate Density Lipoprotein (IDL) and Low Density Lipoprotein (LDL), but excluding 25 cholesterol in High Density Lipoprotein (HDL), in a biological sample, comprising the steps of: contacting a biological sample with an anti-cholesterol antibody capable of selectively detecting cholesterol in VLDL, IDL and LDL, but excluding cholesterol in HDL; measuring the formation of cholesterol-antibody complexes, and thereby selectively determining the sum total amount of cholesterol in VLDL, IDL and LDL, but excluding cholesterol in HDL, in the biological sample; measuring the formation of total cholesterol in the biological sample; and determining the ratio of cholesterol in VLDL, IDL and LDL, but excluding HDL, to the total cholesterol in the biological sample, and thereby determining the ratio of cholesterol in HDL to the total cholesterol in the biological sample. Dated this first day of October 1999 ENTREMED, INC. Patent Attorneys for the Applicant: F B RICE CO *o