Structural and functional differences between glycosylated and non-glycosylated forms of human interferon-beta (IFN-beta).

PURPOSE: Two recombinant IFN-beta products have been approved for the treatment of multiple sclerosis, a glycosylated form with the predicted natural amino acid sequence (IFN-beta-1a) and a non-glycosylated form that has a Met-1 deletion and a Cys-17 to Ser mutation (IFN-beta-1b). The structural basis for activity differences between IFN-beta-1a and IFN-beta-1b, is determined. METHODS: In vitro antiviral, antiproliferative and immunomodulatory assays were used to directly compare the two IFN-beta products. Size exclusion chromatography (SEC), SDS-PAGE, thermal denaturation, and X-ray crystallography were used to examine structural differences. RESULTS: IFN-beta-1a was 10 times more active than IFN-beta-1b with specific activities in a standard antiviral assay of 20 x 10(7) IU/mg for IFN-beta-1a and 2 x 10(7) IU/mg for IFN-beta-1b. Of the known structural differences between IFN-beta-1a and IFN-beta-1b, only glycosylation affected in vitro activity. Deglycosylation of IFN-beta-1a produced a decrease in total activity that was primarily caused by the formation of an insoluble disulfide-linked IFN precipitate. Deglycosylation also resulted in an increased sensitivity to thermal denaturation. SEC data for IFN-beta-1b revealed large, soluble aggregates that had reduced antiviral activity (approximated at 0.7 x 10(7) IU/mg). Crystallographic data for IFN-beta-1a revealed that the glycan formed H-bonds with the peptide backbone and shielded an uncharged surface from solvent exposure. CONCLUSIONS: Together these results suggest that the greater biological activity of IFN-beta-1a is due to a stabilizing effect of the carbohydrate on structure.

Identification of endogenous protein-associated carbohydrate ligands for E-selectin.

A comparative analysis of carbohydrate libraries derived from cell lines binding E-selectin was used to identify endogenous protein-associated carbohydrate ligands for E-selectin. Three structures, which together constitute less than 1% of the total cell surface protein-associated carbohydrate, were unique to cell lines capable of binding E-selectin, including neutrophils and the monocytic cell line U937. All are tetra-antennary N-linked structures, with a sialic acid alpha 2 --> 3 galactose beta 1 --> 4 (fucose alpha 1 --> 3) N-acetyl glucosamine beta 1 --> 3 galactose beta 1 --> 4 (fucose alpha 1 --> 3) N-acetyl glucosamine lactosaminoglycan extension (sialyl-di-Lewis X [S-diLe(x)]) on the arm linked through the C4 residue on the mannose. While all contained the expected 3-SLe(x) sub-structure, these native structures have an additional fucosylated lactosamine unit. Direct evidence that these S-di-Lex-containing structures are high-affinity ligands for E-selectin came from the use of recombinant soluble E-selectin-agarose affinity chromatography. These three carbohydrate structures bound specifically to the E-selectin column, while 3-SLe(x) itself does not bind under identical conditions.

Human type I interferon receptor, IFNAR, is a heavily glycosylated 120-130 kD membrane protein.

Type I interferons (IFNs) bind and signal through cell surface receptors that share at least one common component. One candidate for such a component is the interferon-alpha receptor (IFNAR). Genetic studies have shown that the IFNAR gene product is required for response to many type I interferons. However, these studies also suggest that the IFNAR protein interacts with an additional receptor component(s) to form functionally complete type I IFN receptors. Although these genetic studies have contributed significantly to understanding the type I IFN receptors. Although these genetic studies have contributed significantly to understanding the type I IFN receptors, little biochemical characterization of IFNAR and its function has been reported. To facilitate biochemical studies of the IFNAR gene product, a monoclonal antibody, GB8, recognizing the extracellular domain of IFNAR was prepared. The epitope for GB8 maps to the second extracellular domain of IFNAR between amino acids 278 and 293. GB8 identifies IFNAR in western blots of cell membranes as a broad band with molecular mass ranging from 100 to 150 kD in membranes from CHO cells overexpressing the human IFNAR gene to 136-150 kD in Daudi cell membranes. Such variations in the mean value and the range of molecular mass between IFNAR in different cell lines suggest differences in glycosylation. The majority of glycosylation is N-linked, although there may also be a small amount O-linked oligosaccharide. Deglycosylation of IFNAR in Daudi cell membranes results in a 70 kD IFNAR species, indicating that nearly half of the apparent molecular mass of Daudi cell IFNAR is contributed by carbohydrate moieties.

Isolation and characterization of natural protein-associated carbohydrate ligands for E-selectin.

A comparative analysis of carbohydrate 'libraries' derived from cell lines binding E-selectin with differing avidity identified endogenous protein-associated carbohydrate ligand candidates for E-selectin. Three unusual structures, which constitute less than 3% of cell surface protein-associated carbohydrate, were unique to the E-selectin-binding cells, including neutrophils and the monocytic cell line U937. All are tetraantennary N-linked structures with a NeuAc alpha 2-->3Gal beta 1-->4(Fuc alpha 1-->3)GlcNAc beta 1-->3Gal beta 1-->4- (Fuc alpha 1-->3)GlcNAc lactosaminoglycan extension (diSLex) on the arm linked through the C4 residue on the mannose. While all contained the expected SLex [NeuAc alpha 2-->3Gal beta 1-->4(Fuc alpha 1-->3)GlcNAc] moiety, these structures have an additional fucosylated lactosamine unit. Direct evidence that these diSLex-containing structures are, indeed, high-affinity ligands for E-selectin came from the use of recombinant soluble E-selectin-agarose affinity chromatography. We found that these three carbohydrate structures bound specifically to the E-selectin column. SLex itself does not bind under identical conditions. In summary, these related structures: (1) all possess an unusual 3-sialyl di-Lewis x extension on one arm of an N-linked tetraantennary glycan; (2) of the cells tested, are present only on E-selectin-binding leukocytes and leukocytic cell lines; (3) bind to E-selectin with a relatively high affinity (Kd < microM) and one greater than that of 3-sialyl Lewis x or 3-sialyl Lewis a; and (4) represent a very small percentage of the protein-associated carbohydrate. These carbohydrate structures appear to be present on only a very small number of cell surface proteins and may alone be responsible for the specificity of E-selectin-dependent adhesion.

The ELAM ligand fucosyltransferase, ELFT, directs E-selectin binding to a secreted scaffold protein: a method to produce and purify large quantities of specific carbohydrate structures.

Many biological processes depend on cell surface recognition of receptor-ligand pairs. Some receptors, such as the selectins, recognize specific carbohydrate structures as part of their ligands. The ability to synthesize such ligands for use in the study of cell adhesion mechanisms or as inhibitors of a variety of pathological conditions would be extremely useful. However, the chemical or enzymic in vitro synthesis of carbohydrate-based ligands has thus far been difficult and costly. We have used E-selectin and its carbohydrate ligand as a model system to test if it is possible to express specific carbohydrate structures on a secreted, glycosylated and easily purified scaffold protein and to use this newly modified protein as a functional adhesion molecule. We co-expressed a fucosyltransferase (ELFT) and a secreted immunoglobulin-LFA3 fusion protein (LFA31G) in the same cell to modify the carbohydrate structures on the secreted LFA3IG scaffold protein (we refer to this novel protein as X-LFA3IG). Using glycosidase digestion, lectin binding, carbohydrate composition analysis and antibody-binding assays, we show that approximately 50% of the potential N-linked carbohydrate sites on X-LFA3IG are, indeed, modified and that the modification is the addition of fucose. Furthermore, we show that X-LFA3IG contains epitopes recognized by anti-Slex antibodies, and, using an E-selectin-specific adhesion assay, we demonstrate that X-LFA3IG is a functional ligand for E-selectin. This in vivo approach for generating specific carbohydrate structures could be generalized to produce and purify large quantities of other biologically important carbohydrate structures.

Human eosinophil adherence to vascular endothelium mediated by binding to vascular cell adhesion molecule 1 and endothelial leukocyte adhesion molecule 1.

Adherence of human eosinophils to cytokine-stimulated endothelial cells, which was only partially due to CD18-dependent pathways, was also mediated by binding to endothelial leukocyte adhesion molecule 1 (ELAM-1) and vascular cell adhesion molecule 1 (VCAM-1). Eosinophils bound specifically to both recombinant soluble ELAM-1 and recombinant soluble VCAM-1. Eosinophil binding to recombinant soluble VCAM-1 and to transfected CHO cells expressing VCAM-1 was inhibited with anti-VCAM-1 (4B9) and anti-very late activation antigen 4 (anti-VLA-4; HP1/2 or HP2/1) monoclonal antibodies. Eosinophils, but not neutrophils, expressed VLA-4 detected by cytofluorography. Eosinophil adherence to tumor necrosis factor alpha-stimulated human umbilical vein endothelial cells was partially blocked by monoclonal antibodies against ELAM-1 (BB11) and VCAM-1 (4B9) and against VLA-4 (HP2/1). Thus, while both eosinophils and neutrophils can bind to activated endothelial cells by adherence to ICAM-1 and ELAM-1, only eosinophils expressed VLA-4 and adhered to VCAM-1 on activated endothelial cells. Eosinophil adherence to VCAM-1 might provide a mechanism contributing to the selective recruitment of eosinophils into tissue sites of inflammation.

Expression and functional characterization of a soluble form of endothelial-leukocyte adhesion molecule 1.

Endothelial leukocyte adhesion molecule 1 (ELAM1) is a leukocyte adhesion molecule induced on human venular endothelium in vitro and in vivo by inflammatory stimuli. A truncated cDNA for ELAM1 has been constructed, stably expressed in Chinese hamster ovary cells, and the secreted recombinant soluble form of ELAM1 (rsELAM1) purified to homogeneity by immunoaffinity chromatography. rsELAM1, when immobilized on plastic, is fully functional as an adhesion protein, and selectively binds only cells known to bind cell-surface ELAM1 expressed on human endothelial cells, including the myelomonocytic cell line HL60 and the colon carcinoma cell line HT29. Immobilized rsELAM1 also binds human PMN, monocytes, NK cells, and T cells. T cell subset analyses indicate preferential binding of CD4+ T memory cells. However, rsELAM1 is only a weak inhibitor of ELAM1-mediated adhesion. rsELAM1 should prove valuable for the further study of the role of ELAM1 expressed on the vascular wall during the inflammatory response.

ELFT: a gene that directs the expression of an ELAM-1 ligand.

The LECCAMs are a family of cell adhesion molecules implicated in certain inflammatory processes. ELAM-1, a LECCAM found on the surface of activated endothelial cells, can mediate adhesion of neutrophils, monocytes, and certain cell lines to endothelial cells in vitro. No ligand for any LECCAM has yet been fully characterized. Here we report the cloning of a cDNA, ELFT (ELAM-1 ligand fucosyltransferase), that can confer ELAM-1 binding activity when transfected into nonbinding cell lines. ELFT encodes a 46 kd protein that has alpha(1,3)fucosyltransferase activity, suggesting that a fucosylated carbohydrate structure is an essential component of the ELAM-1 ligand. Furthermore, ELFT is expressed specifically in cell types that bind to ELAM-1, suggesting that this enzyme is an important regulator of inflammatory events in vivo.

Affinity chromatography with biotinylated RNAs.

Small, reversibly biotinylated RNAs as described here are versatile ligands for affinity chromatography of RNA-binding components. These RNAs can be attached to a solid support by binding to avidin and used as ligands, or they may be hybridized to another RNA which acts as the ligand. The incorporation of a disulfide bond in the linker arm connecting biotin to the RNA makes it possible to dissociate the RNA from avidin under mild conditions. Our results regarding the binding and elution of the biotinylated RNA may be applied to other, reversibly biotinylated molecules.

Phylogenetic survey of proteins related to synapsin I and biochemical analysis of four such proteins from fish brain.

A phylogenetic survey of proteins immunologically related to Synapsin I, a major synaptic vesicle-associated phosphoprotein in mammals was carried out. Proteins antigenically related to Synapsin I were found by use of radioimmunoassay and other radioimmunochemical techniques in the nervous systems of several vertebrate and invertebrate species, which included birds, reptiles, amphibians, fish, echinoderms, arthropods, and mollusks. Four proteins present in fish brain, antigenically related to Synapsin I, were further studied and found to resemble mammalian Synapsin I in several respects. Like Synapsin I, the fish proteins were present in high amounts in nervous tissue, were enriched in synaptosomal fractions of brain where they were substrates for endogenous protein kinases, were acid extractable, and were sensitive to digestion by collagenase. In addition, two-dimensional peptide-mapping analysis revealed some homology between major phosphopeptide fragments of Synapsin I and the fish proteins. The results indicate that proteins related to Synapsin I are wide-spread in the animal kingdom.

Purification of DNA from formaldehyde fixed and paraffin embedded human tissue.

The ability to isolate DNA from preserved human tissues would provide numerous experimental opportunities. In this report it is shown that DNA can be extracted from tissues prepared for routine histopathological examination (i.e., fixed with formaldehyde and embedded in paraffin). Although the extracted DNA is not intact, it is double stranded, cleavable with restriction endonucleases, and suitable for a variety of standard techniques used in molecular biology.

Hypomethylation of DNA from benign and malignant human colon neoplasms.

The methylation state of DNA from human colon tissue displaying neoplastic growth was determined by means of restriction endonuclease analysis. When compared to DNA from adjacent normal tissue, DNA from both benign colon polyps and malignant carcinomas was substantially hypomethylated. With the use of probes for growth hormone, gamma-globin, alpha-chorionic gonadotropin, and gamma-crystallin, methylation changes were detected in all 23 neoplastic growths examined. Benign polyps were hypomethylated to a degree similar to that in malignant tissue. These results indicate that hypomethylation is a consistent biochemical characteristic of human colonic tumors and is an alteration in the DNA that precedes malignancy.

Adenylate kinase of Escherichia coli: evidence for a functional interaction in phospholipid synthesis.

Previous genetic and biochemical experiments have suggested that the adenylate kinase of Escherichia coli may be directly involved in phospholipid synthesis through formation of a complex with sn-glycerol-3-phosphate acyltransferase, the membrane-bound enzyme that catalyzes the first step in phospholipid synthesis. In this paper we report direct experiments to test this hypothesis. A mutation within the adenylate kinase structural gene is described that results in a temperature-sensitive phospholipid synthesis (assayed in vivo) and a temperature-sensitive acyltransferase. The adenylate kinase activity of this strain is only minimally altered either in vitro or [as assayed by adenosine 5'-triphosphate (ATP) levels] in vivo. This result demonstrates that the inhibition of phospholipid synthesis is not the result of reduced ATP levels. We report the purification of E. coli adenylate kinase to homogeneity; and find that the addition of homogeneous wild-type adenylate kinase to membranes containing a mutationally altered temperature-sensitive acyltransferase results in thermal stabilization of the acyltransferase activity. Ovalbumin has no such protective effect. Purified E. coli inner membranes contain several proteins that are precipitated by addition of anti adenylate kinase antibody to detergent-solubilized membranes.

Distribution of protein I in mammalian brain as determined by a detergent-based radioimmunoassay.

A radioimmunoassay has been developed for measuring protein I, a basic, neuron-specific protein associated with nerve terminals. The procedure utilizes the detergents NaDodSO4 and Nonidet P-40 to prevent nonspecific adsorption of this highly charged protein to various surfaces. By use of this procedure, it has been possible to show that protein I comprises approximately 0.4% of the total protein in cerebral cortex of several mammalian species. In addition, the amount of protein I was determined in about 40 regions of cat brain. The results suggest that measurement of protein I may provide a quantitative method for estimating the density of nerve terminals in various regions of the mammalian nervous system.

Phospholipid synthesis in Escherichia coli. Characteristics of fatty acid transfer from acyl-acyl carrier protein to sn-glycerol 3-phosphate.

Two kinetically distinguishable sn-glycerol 3-phosphate (glycerol-P) acryltransferase activities were detected in Escherichia coli inner membranes using acyl-acyl carrier protein (ACP) substrates. The first system was characterized as having a Michaelis constant (Km) for glycerol-P of 90 microM and utilized palmitoyl-ACP to form primarily 1-acylglycerol-P. Palmitoyl-CoA and cis-vaccenoyl-ACP were also utilized by this system but, with these substrates, significantly more phosphatidic acid was formed as compared to palmitoyl-ACP. Although palmitoyl-ACP and palmitoyl-CoA had kinetically indistinguishable glycerol-P sites, distinct acyl donor binding sites were inferred from kinetic experiments using acyl carrier protein as an acyltransferase inhibitor. A second enzyme system, characterized as having a Km for glycerol-P of 700 microM, was found using palmitoyl-ACP as a substrate. This acyltransferase had a slightly higher pH optimum than the low Km acyltransferase activity, and phosphatidic acid was the major product. Two degradative reactions were identified in this system. One reaction yielded diacylglycerol when palmitoyl-ACP was the substrate. The other degradative reaction produced glycerol. Glycerol was formed in all incubations but was most pronounced when palmitoyl-ACP was the substrate.

The positional distribution of fatty acids in Escherichia coli phospholipids is not regulated by sn-glycerol 3-phosphate levels.

Large changes in the intracellular concentration of sn-glycerol 3-phosphate had no effect on the acyl chain distribution of the phospholipids of Escherichia coli. This result directly contradicts the prediction by other workers based on in vitro experiments.

Neuronal protein phosphorylation: recent studies concerning protein I, a synapse-specific phosphoprotein.

Recent studies are described concerning the physiological role and tissue distribution of Protein I, a synapse-specific phosphoprotein. This protein is present in both the central and the peripheral nervous system and appears to be enriched in synaptic vesicles. It is a major substrate for both Ca2(-)- and cyclic AMP-dependent phosphorylation in the brain. In slices of the facial motor nucleus from rat brainstem, serotonin (5-HT) added to the incubation medium stimulates the phosphorylation of Protein I. This effect is potentiated by low concentrations of isobutylmethylxanthine and is prevented by the serotonin antagonist mianserin.