Induction and detection of antibodies to squalene.

An enzyme-linked immunosorbent assay (ELISA) utilizing antigen coated on hydrophobic polyvinyldiene fluoride (PVDF) membranes is described for detecting antibodies that bind to squalene (SQE). Because of the prior lack of availability of validated antibodies to SQE, positive controls for the assay were made by immunization with formulations containing SQE to create monoclonal antibodies (mAbs) that reacted with SQE. Among eight immunogens tested, only two induced detectable murine antibodies to SQE: liposomes containing dimyristoyl phosphatidylcholine, dimyristoyl phosphatidylglycerol, 71% SQE, and lipid A [L(71% SQE+LA)], and, to a much lesser extent, an oil-in-water emulsion containing SQE, Tween 80, Span 85, and lipid A. In each case, lipid A served as an adjuvant, but neither SQE alone, SQE mixed with lipid A, liposomes containing 43% SQE and lipid A, nor several other emulsions containing both SQE and lipid A, induced antibodies that reacted with SQE. Monoclonal antibodies produced after immunizing mice with [L(71% SQE+LA)] served as positive controls for developing the ELISA. Monoclonal antibodies were produced that either recognized SQE alone but did not recognize squalane (SQA, the hydrogenated form of SQE), or that recognized both SQE and SQA. As found previously with other liposomal lipid antigens, liposomes containing lipid A also induced antibodies that reacted with the liposomal phospholipids. However, mAbs were also identified that reacted with SQE on PVDF membranes, but did not recognize either SQA or liposomal phospholipid. The polyclonal antiserum produced by immunizing mice with [L(71% SQE+LA)] therefore contained a mixed population of antibody specificities and, as expected, the ELISA of polyclonal antiserum with PVDF membranes detected antibodies both to SQE and SQA. We conclude that SQE is a weak antigen, but that antibodies that specifically bind to SQE can be readily induced by immunization with [L(71% SQE+LA)] and detected by ELISA with PVDF membranes coated with SQE.

Oil-in-water liposomal emulsions: characterization and potential use in vaccine delivery.

Emulsification of mineral oil by phospholipids donated by liposomes composed of dimyristoyl phosphatidylcholine, dimyristoyl phosphatidylglycerol, cholesterol, and lipid A by extrusion resulted in the formation of oil-in-water liposomal emulsions containing a substantial number of intact liposomes. Increasing the proportion of liposomes from 25 mM to 150 mM phospholipid and increasing the oil content from 2.5% (v/v) to 42.5% (v/v) changed the flow characteristics of the emulsions from fluid liquid-like to viscous. Likewise, the degree of stability of the emulsions was liposomal phospholipid concentration-dependent, ranging from partial emulsification in the range 25-100 mM to complete stabilization in the range 125-150 mM. Despite some loss of liposome integrity, as evidenced by the release of liposomal trapped glucose, emulsification of liposomes containing encapsulated prostate-specific antigen (PSA) exhibited antigen-specific immunostimulation in mice. These results suggest that liposomes containing encapsulated antigen can serve as constituents for the formulation of oil-in-water vaccines.

Immunologic approaches to the treatment of prostate cancer.

The presence of several organ-specific molecules that could serve as immunogens or targets of an immune attack, the nonessential nature of the prostate gland, the substantial failure rate after treatment of the primary tumor, and the lack of effective chemotherapy for metastatic disease make prostate cancer an ideal candidate for immunotherapy. This report reviews the current status of two novel approaches to the treatment of prostate cancer. The first is an effort to induce antitumor immunity by enriching the cytokine environment within the primary cancer by intraprostatic injection of Leukocyte Interleukin (Cel-Sci Corp, Vienna, VA), a mixture of natural cytokines that includes interleukin-1 beta (IL-1beta), IL-2, granulocyte-macrophage colony-stimulating factor (GM-CSF), interferon gamma (IFN-gamma), and tumor necrosis factor alpha (TNF-alpha). The second approach uses OncoVax-P (Jenner Biotherapies, Inc, San Ramon, CA), a vaccine consisting of liposome-encapsulated recombinant prostate-specific antigen (PSA) and lipid A. When administered as an emulsion or in association with bacillus Calmette-Guérin (BCG)/cyclophosphamide or GM-CSF with or without IL-2/cyclophosphamide, immunologic tolerance is broken as evidenced by the generation of humoral and cellular immunity. Both of these approaches have been shown to be feasible and safe, and are now being tested in patients with less advanced disease to determine if manipulation of the immune system can favorably influence clinical outcome.

Cytotoxic T lymphocytes to Ebola Zaire virus are induced in mice by immunization with liposomes containing lipid A.

An eight amino acid sequence (TELRTFSI) present in the carboxy terminal end (aa 577-584) of membrane-anchored GP, the major structural protein of Ebola virus, was identified as an H-2k-specific murine cytotoxic T cell epitope. Cytotoxic T lymphocytes (CTLs) to this epitope were induced by immunizing B10.BR mice intravenously with either irradiated Ebola virus or with irradiated Ebola virus encapsulated in liposomes containing lipid A. The CTL response induced by irradiated Ebola virus could not be sustained after the second round of in vitro stimulation of immune splenocytes with the peptide, unless the irradiated virus was encapsulated in liposomes containing lipid A. The identification of an Ebola GP-specific CTL epitope and the requirement of liposomal lipid A for CTL memory recall responses could prove to be a promising approach for developing a vaccine against Ebola virus infection.

Transcutaneous immunization with bacterial ADP-ribosylating exotoxins as antigens and adjuvants.

Transcutaneous immunization (TCI) is a new technique that uses the application of vaccine antigens in a solution on the skin to induce potent antibody responses without systemic or local toxicity. We have previously shown that cholera toxin (CT), a potent adjuvant for oral and nasal immunization, can induce both serum and mucosal immunoglobulin G (IgG) and IgA and protect against toxin-mediated mucosal disease when administered by the transcutaneous route. Additionally, CT acts as an adjuvant for coadministered antigens such as tetanus and diphtheria toxoids when applied to the skin. CT, a member of the bacterial ADP-ribosylating exotoxin (bARE) family, is most potent as an adjuvant when the A-B subunits are present and functional. We now show that TCI induces secondary antibody responses to coadministered antigens as well as to CT in response to boosting immunizations. IgG antibodies to coadministered antigens were also found in the stools and lung washes of immunized mice, suggesting that TCI may target mucosal pathogens. Mice immunized by the transcutaneous route with tetanus fragment C and CT developed anti-tetanus toxoid antibodies and were protected against systemic tetanus toxin challenge. We also show that bAREs, similarly organized as A-B subunits, as well as the B subunit of CT alone, induced antibody responses to themselves when given via TCI. Thus, TCI appears to induce potent, protective immune responses to both systemic and mucosal challenge and offers significant potential practical advantages for vaccine delivery.

Immunization with cholesterol-rich liposomes induces anti-cholesterol antibodies and reduces diet-induced hypercholesterolemia and plaque formation.

Immunization of rabbits with a protein-free formulation consisting of liposomes containing 71% cholesterol and lipid A as an adjuvant induced anticholesterol antibodies that caused complement-dependent lysis of liposomes lacking lipid A. The antibodies, immunoglobulin G (IgG) and immunoglobulin M (IgM), also recognized nonoxidized crystalline cholesterol as an antigen by enzyme-linked immunosorbent assay (ELISA). The effects of immunization against cholesterol on elevations in serum cholesterol and development of atherosclerosis were examined in rabbits fed a diet containing 0.5% to 1.0% cholesterol. Although the mean serum cholesterol level, mainly in the form of very-low-density lipoprotein cholesterol, rose as much as 60-fold in the nonimmunized rabbits, the elevation was significantly less--as much as 35% lower--in the immunized rabbits. Elevation of serum cholesterol was accompanied by an apparent drop in the level of antibodies on initiating the diet, followed by a rebound on stopping the diet, thus suggesting that the antibodies were adsorbed to cholesterol that was present in circulating lipoproteins. When lipoprotein fractions--composed of either very-low-density and intermediate-density lipoproteins derived from cholesterol-fed nonimmunized rabbits or human low-density lipoproteins--were tested as capture antigens by solid-phase ELISA, reactivity was observed with IgG and IgM antibodies present in the serum of immunized rabbits. Immunization also resulted in a marked decrease in the risk of developing atherosclerosis. Analysis of aortic atherosclerosis by quantitative histologic examination and fatty streaks by automated morphometric probability-of-occurrence mapping showed diminished atherosclerosis in most areas of the aorta in vaccine recipients. It is proposed that immunization with liposomes containing 71% cholesterol and lipid A can reduce diet-induced hypercholesterolemia and atherosclerosis.

Murine IgG subclass antibodies to antigens incorporated in liposomes containing lipid A.

The IgG subclass responses to antigens incorporated in liposomes containing lipid A were investigated using a synthetic malarial antigen (SPf66) and cholera toxin (CT). The antigen-specific IgG subclass response was determined in BALB/c mice immunized with either: (a) SPf66 encapsulated in liposomes containing lipid A, (b) CT bound to the surface of liposomes containing lipid A, or (c) both encapsulated SPf66 and surface-bound CT in the same liposomes. In each case the antibodies to SPf66, CT and lipid A demonstrated an IgG2a predominance. Liposomes containing lipid A not only increased the magnitude of the antibody response to liposomal antigens but elicited predominantly IgG2a subclass antibodies as well.

Complement-dependent phagocytosis of liposomes by macrophages: suppressive effects of "stealth" lipids.

We have previously reported that complement-opsonized liposomes composed of dimyristoyl phosphatidylcholine and cholesterol are actively phagocytozed by murine peritoneal macrophages and that such complement-induced phagocytosis can be suppressed by the presence of liposomal phosphatidylinositol (Proc. Natl. Acad. Sci. USA 81, 1984). We now report suppressive effects of other liposomal lipids, including monosialoganglioside (GM1) and sulfogalactosylceramide. Complement-dependent phagocytosis was almost completely suppressed by liposomes containing GM1 or phosphatidylinositol and partially suppressed when liposomes contained sulfogalactosylceramide. Although the mechanism of suppression of complement-induced phagocytosis by these liposomal lipids is not yet completely understood, it does not seem to involve the early stages of complement activation resulting in opsonization of liposomes with complement. We conclude that suppression of complement-induced phagocytosis by phosphatidylinositol, GM1, or sulfogalactosylceramide occurs at a step after liposome opsonization.

The bimodal growth response of Swiss 3T3 cells to the B subunit of cholera toxin is independent of the density of its receptor, ganglioside GM1.

The B subunit of cholera toxin, a protein which binds specifically to cell surface ganglioside GM1, has been shown to have a bimodal effect on DNA synthesis in Swiss 3T3 fibroblasts. The B subunit induced cellular proliferation of confluent and quiescent cells while it inhibited the growth of the same cells when they were sparse and rapidly dividing. The amount of cell surface GM1 increased when the cells reached confluency. To examine the hypothesis that the variation in levels of GM1 was responsible for the bimodal effect, we increased GM1 levels in rapidly dividing cells by insertion of exogenous GM1 or by treatment of the cells with neuraminidase to convert polysialogangliosides to GM1. Even after the level of GM1 was increased to levels similar to those found in confluent cells, the B subunit still inhibited, rather than stimulated, their growth. Therefore, this result indicates that the bimodal response to the B subunit is not solely a function of the concentration of cell surface GM1; rather it is the growth stage that determines the fate of the signal transduced by the interaction of the B subunit and ganglioside GM1.

Lipid signalling pathways in normal and ras-transfected NIH/3T3 cells.

The role of ras oncogenes in cellular signalling pathways involving phospholipid breakdown was studied in untransfected and proto-H-ras and mutated H-, K- and N-ras transfected NIH/3T3 cells. When the cells were grown at low cell densities, all of the ras transfected cells had 2-4 fold higher diacylglycerol (DAG) levels compared to growing NIH/3T3 cells. At high cell densities, DAG levels decreased in the former and increased in contact inhibited NIH/3T3 cells. In this regard, only cells transformed by mutated cellular and viral H-ras oncogenes (but not by the H-ras proto-oncogene) had elevated DAG levels compared to contact inhibited NIH/3T3 cells. The basal levels of inositol phosphates in ras transfected cells were not significantly different from NIH/3T3 cells and did not vary with cell density. Thus, the elevated DAG levels are not a consequence of increased phosphoinositide hydrolysis. The latter was stimulated by serum and bombesin only in normal and proto-H-ras transfected cells. In contrast, stimulation by bradykinin was observed only in cells transformed by mutated cellular ras oncogenes. Furthermore, aluminum fluoride stimulated phosphoinositide breakdown in the latter cells indicating that there was no uncoupling of the G protein from phospholipase C. Treatment of ras transfected cells with dibutyryl cyclic AMP (DB-cAMP), which causes an inhibition of growth and a reversal of the transformed morphology, did not alter the basal levels of inositol phosphates, DB-cAMP, however, did lower DAG levels in some of the transformed cell lines, but elevated DAG levels in low density NIH/3T3 cells. These findings indicate that the ras gene product p21 is not involved in phosphoinositide hydrolysis and that DAG levels do not correlate with cell growth in either normal or ras transfected NIH/3T3 cells. Thus, p21 appears to alter cell growth through mechanism(s) independent of lipid signalling pathways.

Asymmetric distribution of gangliosides in rat renal brush-border and basolateral membranes.

Highly enriched brush-border and basolateral membranes isolated from rat renal cortex were used to study the distribution of endogenous gangliosides in the two distinct plasma membrane domains of epithelial cells. These two membrane domains differed in their glycolipid composition. The basolateral membranes contained more of both neutral and acidic glycolipids, expressed on a protein basis. In both membranes, the neutral glycolipids corresponding to mono-, di-, tri- and tetraglycosylceramides were present. The basolateral membranes contained more diglycosylceramide than the brush-border membranes. The major gangliosides found were GM4, GM3, and GD3 with minor amounts of GM1 and GD1a. The latter were identified and quantified by sensitive iodinated cholera toxin binding assays. When the distribution of individual gangliosides was calculated as a percent of total gangliosides, the brush-border membranes were enriched with GM3, GM1 and GD1a compared to the basolateral membranes, which were enriched with GD3 and GM4. The observation of a distinct distribution of glycolipids between brush-border and basolateral membranes of the same epithelial cell suggests that there may be a specific sorting and insertion process for epithelial plasma membrane glycolipids. In turn, asymmetric glycolipid biogenesis may reflect differences in glycolipid function between the two domains of the epithelial plasma membrane.

Subcellular distribution and biosynthesis of rat liver gangliosides.

Gangliosides have generally been assumed to be localized primarily in the plasma membrane. Analysis of gangliosides from isolated subcellular membrane fractions of rat liver indicated that 76% of the total ganglioside sialic acid was present in the plasma membrane. Mitochondria and endoplasmic reticulum fractions, while containing only low levels of gangliosides on a protein basis, each contained approx. 10% of total ganglioside sialic acid. Gangliosides also were present in the Golgi apparatus and nuclear membrane fractions, and soluble gangliosides were in the supernatant. Individual gangliosides were non-homogeneously distributed and each membrane fraction was characterized by a unique ganglioside composition. Plasma membrane contained only 14 and 28% of the total GD1a and GD3, respectively, but 80-90% of the GM1, GD1b, GT1b and GQ1b. Endoplasmic reticulum, when corrected for plasma membrane contamination, contained only trace amounts of GM1, GD1b, GT1b and GQ1b, but 11 and 5% of the total GD1a and GD3, respectively. The ganglioside composition of highly purified endoplasmic reticulum was similar. Ganglioside biosynthetic enzymes were concentrated in the Golgi apparatus. However, low levels of these enzymes were present in the highly purified endoplasmic reticulum fractions. Pulse-chase experiments with [3H]galactose revealed that total gangliosides were labeled first in the Golgi apparatus, mitochondria and supernatant within 10 min. Labeled gangliosides were next observed at 30 min in the endoplasmic reticulum, plasma membrane and nuclear membrane fractions. Analysis of the individual gangliosides also revealed that GM3, GM1, GD1a and GD1b were labeled first in the Golgi apparatus at 10 min. These studies indicate that gangliosides synthesized in the Golgi apparatus may be transported not only to the plasma membrane, but to the endoplasmic reticulum and to other internal endomembranes as well.

Alteration of glycolipids in ras-transfected NIH 3T3 cells.

Glycosphingolipid alterations upon viral transformation are well documented. Transformation of mouse 3T3 cells with murine sarcoma viruses results in marked decreases in the levels of gangliosides GM1 and GD1a and an increase in gangliotriaosylceramide. The transforming oncogenes of these viruses have been identified as members of the ras gene family. We analyzed NIH 3T3 cells transfected with human H-, K- and N-ras oncogenes for their glycolipid composition and expression of cell surface gangliosides. Using conventional thin-layer chromatographic analysis, we found that the level of GM3 was increased and that of GD1a was slightly decreased or unchanged, and GM1 was present but not in quantifiable levels. Cell surface levels of GM1 were determined by 125I-labeled cholera toxin binding to intact cells. GD1a was determined by cholera toxin binding to cells treated with sialidase prior to toxin binding. All ras-transfected cells had decreased levels of surface GM1 and GD1a as compared to logarithmically growing normal NIH 3T3 cells. Levels of GM1 and, to a lesser extent, GD1a increased as the latter cells became confluent. Using a monoclonal antibody assay, we found that gangliotriaosylceramide was present in all ras-transfected cells studied but not in logarithmically growing untransfected cells. Interestingly, gangliotriaosylceramide appeared when the latter cells became confluent. These results indicated that ras oncogenes derived from human tumors are capable of inducing alterations in glycolipid composition.

Blood coagulation abnormalities in Fischer strain rats bearing tumors.

This study reports an unusual subpopulation of rats bearing transplanted tumors approximately 50% of the total and apparently unrelated to the presence of metastases, that exhibited shortened bleeding times despite reduced platelet numbers and/or fibrinogen levels. The remaining rats exhibited the expected inverse relationships between bleeding time and platelet numbers and/or fibrinogen level. Tumors were hepatomas and squamous cell carcinomas initially induced in the Fischer strain of rats and carried by passage through tissue culture or syngenic recipient animals.

Fibronectin binding to gangliosides and rat liver plasma membranes.

Binding of fibronectins to gangliosides was tested directly using several different in vitro models. Using an enzyme-linked immunoabsorbent assay (ELISA), gangliosides were immobilized on polystyrene tubes and relative binding of fibronectin was estimated by alkaline phosphatase activity of conjugated second antibody. Above a critical ganglioside concentration, the gangliosides bound the fibronectin (GT1b congruent to GD1b congruent to GD1a greater than GM1 much greater than GM2 congruent to GD3 congruent to GM3) in approximately the same order of efficiency as they competed for the cellular sites of fibronectin binding in cell attachment assays (Kleinman et al., Proc natl acad sci US 76 (1979) 3367). Alternatively, these same gangliosides bound to immobilized fibronectin. Rat erythrocytes coated with gangliosides GM1, GD1a or GT1b bound more fibronectin than erythrocytes not supplemented with gangliosides. Using fibronectin in which lysine residues were radioiodinated, an apparent Kd for binding to mixed rat liver gangliosides of 7.8 X 10(-9) M was determined. This value compared favorably with the apparent Kd for attachment of fibronectin to isolated plasma membranes from rat liver of 3.7 X 10(-9) M for fibronectin modified on the tyrosine residue, or 6.4 X 10(-9) M for fibronectin modified on lysine residues. As shown previously by Grinnell & Minter (Biochem biophys acta 550 (1979) 92), fibronectin modified on tyrosine residues did not promote spreading and attachment of CHO cells. It did, however, bind to cells. In contrast, lysine-modified fibronectin both bound to cells and promoted cell attachment. Plasma membranes isolated from hepatic tumors in which the higher gangliosides that bind fibronectin were depleted bound 43-75% less [125I]fibronectin than did plasma membranes from control livers. The findings were consistent with binding of fibronectins to gangliosides, including the same gangliosides depleted from cell surfaces during tumorigenesis in the rat.

Dictyosome-like structures from guinea-pig testes lack galactosyltransferase, a Golgi apparatus marker.

More than twenty different enzyme activities of fractions containing dictyosome-like structures (DLS) as a dominant cell component were monitored. Plasma membrane vesicles were a major contaminant of the DLS fractions, which, presumably as a consequence, were enriched somewhat in plasma membrane markers. The lysosomal enzymes arylsulfatase and latent acid phosphatase were present in the DLS fractions as were the Golgi apparatus activities thiamine pyrophosphatase and nucleoside diphosphatase. The presence of the latter two enzymes in DLS, plus NADH-ferricyanide reductase, has been verified from cytochemistry. On the other hand, the Golgi apparatus marker, galactosyltransferase, was not enriched in DLS fractions and appeared to be absent. This latter finding, verified from cytochemistry with isolated DLS fractions and, in situ, from [3H]galactose incorporation by testis tubules with analysis by autoradiography, provides the first clear biochemical characteristic that serves unequivocally to distinguish DLS from conventional Golgi apparatus.

Coupling of uridine-5'-diphosphate (UDP) formation and nicotinamide adenine dinucleotide (NAD+) reduction for cytochemical localization of glycosyltransferases.

A technique applicable to the cytochemical localization of glycosyltransferases through a series of coupled enzyme reactions is described. Uridine-5'-diphosphate (UDP) formed by glycosyltransferases is first phosphorylated to uridine-5'-triphosphate (UTP) by nucleoside 5'-diphosphate kinase. The UTP plus exogenous glucose-1-phosphate is converted into UDP-glucose by uridine-5'-diphosphoglucose pyrophosphorylase. UDP-glucose is then oxidized by uridine-5'-diphosphoglucose dehydrogenase to form UDP-glucuronic acid and reduced nicotinamide adenine dinucleotide (NADH). The NADH is utilized by membrane-located NADH-ferricyanide oxidoreductases in the presence of a copper salt to form electron-dense deposits of cupric ferrocyanide (Hatchett's brown). Using this technique, galactosyltransferase has been localized in cisternae (including the central midregions of the cisternae) of Golgi apparatus isolated from rat liver. Reactivity is absent from the cis-most cisternae and membrane elements. The reaction is dependent on UDP-galactose and inhibited by ethylene diaminetetraacetic acid and puromycin. the latter is a known inhibitor of galactosyltransferase of rat liver Golgi apparatus. The reaction is adaptable by varying the sugar nucleotide donor and acceptor to any glycosyltransferase utilizing UDP-sugars (except UDP-glucose). Presently it is restricted to isolated membrane fractions and permeabilized cells due to the need for accessibility of reagents and coupling enzymes.