Evidence that specific high mannose structures directly regulate multiple cellular activities.

Previous studies have demonstrated that much of the immunomodulatory activity of the glycoprotein uromodulin can be attributed to attached oligosaccharides. Structural studies of isolated and purified saccharides derived from uromodulin suggest that the structure Man6GlcNAc2-asn can inhibit in vitro assays of antigen driven T cell proliferation. Based on these observations, we isolated a series of high mannose glycopeptides from a variety of natural sources and tested them for biological activity in a number of assays. We found that purified mannose rich glycopeptides are able to activate the hexose monophosphate (HMP) shunt, induce prostaglandin synthesis, and directly stimulate IL-1 synthesis. These in vitro effects appear to have in vivo counterparts. Thus in a species-restricted fashion, high mannose compounds are able to directly activate a delayed mononuclear cell infiltrate after intradermal injection. Our data suggest that specific mannose oligosaccharides may activate as well as inhibit cellular immune responses at several different levels. These findings support the hypothesis that specific saccharide structures could participate in the physiologic regulation of the immune response.

Sialomucin and lytic susceptibility of rat mammary tumor ascites cells.

The potential role of cell surface sialomucin in preventing natural killer (NK)-mediated lysis of tumor cell targets has been addressed by comparing the properties of 2 NK-resistant [ascites (ASC) and short-term cultured (STC)] and 2 NK-susceptible [tunicamycin-treated (TUN) and long-term cultured (LTC)] preparations of 13762 MAT-B1 rat mammary tumor cells. Both the ASC and STC cell preparations contain elevated levels of the sialomucin ASGP-1 relative to TUN and LTC preparations as determined by [3H]glucosamine labeling and by binding of peanut agglutinin. The major difference in the susceptibility to NK-mediated lysis appeared to be due to the differences in the susceptibility to lysis by lytic granules, rather than to differences in the ability to bind or trigger effector cells, since TUN and LTC cells were approximately 10-fold more sensitive to lysis by lytic granules than were ASC and STC cells. All preparations inhibited the lysis of the susceptible target YAC-1 by normal rat splenocytes, indicating an ability to bind these effector cells. Triggering of effectors, as monitored either by incorporation of 32P into phosphatidylinositol or by transmethylation of phosphatidylcholine, was similar for the positive control YAC-1, STC, TUN, and LTC, whereas ASC appeared to be defective in triggering effectors. These results suggest that tumor sialomucin blocks the final phase of lysis, but not the initial recognition of tumor cells by NK effectors.

Evidence that specific high-mannose oligosaccharides can directly inhibit antigen-driven T-cell responses.

Uromodulin is an 85 Kd immunosuppressive glycoprotein originally isolated from human pregnancy urine. It is unique in that most of its biologic activity can be attributed to attached oligosaccharides. Purified immunomodulatory oligosaccharides from uromodulin have been structurally characterized using 1H-NMR spectroscopy and shown to be Man6-7GlcNAc2(M6,M7). Based on these observations, we isolated high-mannose N-type oligosaccharides and glycopeptides from ovalbumin, soybean agglutinin, and yeast mannan and show that these high-mannose compounds directly inhibit in vitro antigen-driven T-cell proliferation from millimolar to nanomolar concentrations. The most active compound was a core mannose oligosaccharide derived from yeast mannan, M9(y), which acts to block early events required for normal antigen processing/presentation. These data emphasize the potential functional role of carbohydrate structure in regulating the human immune response.

Effect of tunicamycin on sialomucin and natural killer susceptibility of rat mammary tumor ascites cells.

The MAT-B1 and MAT-C1 ascites sublines of the 13762 rat mammary adenocarcinoma contain a dominant cell surface "complex" consisting of two glycoproteins: ascites sialoglycoprotein (ASGP)-1, a Mr 600,000-700,000 peanut agglutinin-binding sialomucin, and ASGP-2, a Mr 120,000 concancavalin A-binding glycoprotein (Sherblom et al., J. Biol. Chem., 255: 783-790, 1980; Sherblom and Carraway, J. Biol. Chem., 255: 12051-12059, 1980). Although both cell lines are resistant to lysis by natural killer cells, treatments which result in loss of cell surface ASGP-1 render the cells susceptible to natural killer cell lysis (Sherblom and Moody, Cancer Res., 46:4543-4546, 1986). Treatment of the ascites cells with 5 micrograms/ml tunicamycin for 24 h effectively inhibits glycosylation of ASGP-2 without affecting cell viability or total protein synthesis. Under these conditions, expression of ASGP-1 is depressed by at least 50% in both cell lines, as monitored by [3H]glucosamine incorporation and by binding of peanut agglutinin to intact cells. The size distribution of O-linked oligosaccharides in ASGP-1 from tunicamycin-treated versus control MAT-B1 cells is indistinguishable, as determined by Bio-Gel P-4 chromatography following alkaline-borohydride treatment. Complex isolated from either treated or control cells bands at the same density in a CsCl gradient containing Triton X-100 and contains a diffuse band corresponding to ASGP-2 by sodium dodecyl sulfate-polyacrylamide gel electrophoresis. Tunicamycin-treated cells, consistent with the reduced expression of ASGP-1, are significantly more susceptible to natural killer cell-mediated lysis, when compared to untreated controls. The results suggest that N-linked glycosylation is a prerequisite for sialomucin synthesis and/or complex formation.

Age-related changes in serum insulin-like growth factor-binding proteins in women.

Plasma insulin-like growth factor-I (IGF-I) concentrations are reported to decline with advancing age. Five IGF-binding proteins (IGF-BPs) have recently been characterized in human serum, although their biological role beyond circulatory transport of IGF-I is unknown. We studied plasma IGF-I (by RIA) and serum IGF-BPs (by Western ligand blotting) in healthy elderly (n = 21) and healthy young (n = 22) women to determine if aging alters IGF-I and its high affinity BPs. Plasma IGF-I was significantly lower in the elderly than in the young group (0.78 +/- 0.08 vs. 1.22 +/- 0.11 U/ml; P less than 0.005). The number and size of IGF-BPs did not differ between age groups, but the IGF-BP binding ratios (binding of one BP fraction/binding of all fractions) for the BP-53 acid-stable complex (41.5K and 38.5K BPs), the 30K IGF-BP, and the 24K IGF-BP were all lower in the elderly than in the young group (P less than 0.01 for each fraction, elderly vs. young). In contrast, the 34K IGF-BP binding ratio was significantly greater in the elderly than in the young (0.30 +/- 0.03 vs. 0.12 +/- 0.01; P less than 0.001) and correlated closely with advancing age (r = 0.64; P less than 0.01). The changes in IGF-BPs found in the elderly are quite similar to alterations in serum IGF-BPs previously reported in GH deficiency. Since several IGF-BPs in vitro have been shown to modulate the mitogenic activity of IGF-I, the serum IGF-BP changes noted above may be important for the growth and maintenance of connective tissue, muscle, and bone during the aging process.

Pregnancy-associated changes in oligomannose oligosaccharides of human and bovine uromodulin (Tamm-Horsfall glycoprotein).

The urinary glycoprotein uromodulin (Tamm-Horsfall glycoprotein) exhibits a pregnancy-associated ability to inhibit antigen-specific T cell proliferation, and the activity is associated with a carbohydrate moiety [Muchmore and Decker (1985) Science 229:479-81; Hession et al., (1987) Science 237:1479-84; Muchmore, Shifrin and Decker (1987) J Immunol 138:2547-53]. We report here that the Man6(7)GlcNAc2-R glycopeptides derived from uromodulin inhibit antigen-specific T cell proliferation by 50% at 0.2-2 microM, and further studies, reported elsewhere, confirm that oligomannose glycopeptides from other sources are also inhibitory, with Man9GlcNAc2-R the most inhibitory of those tested [Muchmore et al., J Leukocyte Biol (in press)]. In this work, we have extended the observation of pregnancy-associated inhibitory activity to a second species, and have compared the oligomannose profile of Tamm-Horsfall glycoprotein (nonpregnant) with that of uromodulin (pregnant) derived from both human and bovine sources. Surprisingly, there was a pregnancy-associated decrease in the total content of oligomannose chains due predominantly to a reduction in Man5GlcNAc2-R and Man6GlcNAc2-R. Man7GlcNAc2-R, which did not decrease with pregnancy, comprised a significantly greater proportion of the total oligomannose chains in pregnant vs. nonpregnant samples from both species (human; 34.6% vs. 25.9%: bovine; 14.4% vs. 7.2%).

IL-2, a lectin with specificity for high mannose glycopeptides.

Utilizing a solid phase binding assay, we have demonstrated that rIL-2 binds with high affinity to the human urinary glycoprotein uromodulin. This binding is specifically inhibited by the saccharides diacetylchitobiose and Man(alpha 1-3)(Man(alpha 1-6]Man-O-methyl and by the high mannose glycopeptides Man5GlcNAc2-R and Man6GlcNAc2-R, but not by Man9GlcNAc2-R. rIL-2 also binds OVA, a glycoprotein which contains approximately 50% high mannose chains at a single glycosylation site, and to yeast mannan. This binding is inhibited by the same battery of saccharides which inhibit the binding to uromodulin. The conclusion that rIL-2 is a lectin is further supported by the observation that the sequence of IL-2 shares 27% homology with a 33-residue sequence of the carbohydrate-binding domain of human mannose-binding protein. The potential physiologic relevance of the carbohydrate binding activity is further elucidated by studies which show that 1) binding of soluble rIL-2 to immobilized uromodulin is enhanced at a pH of 4 to5 in the presence of divalent cations, and 2) neither uromodulin nor the high mannose glycopeptide Man5GlcNAc2Asn blocks the binding of rIL-2 to the IL-2R. Thus the carbohydrate-binding site of rIL-2 is distinct from the cell surface receptor-binding site, and might function preferentially in acidic microenvironments.

The lectin-like interaction between recombinant tumor necrosis factor and uromodulin.

The polypeptide of uromodulin, an immunosuppressive glycoprotein isolated from human urine, has been shown to be identical to that of Tamm-Horsfall glycoprotein and is synthesized exclusively in the kidney (Hession, C., Decker, J. M., Sherblom, A. P., Kumar, S. (1987) Science 237, 1479-1484). Uromodulin binds recombinant murine interleukin 1 alpha with high affinity, and this binding can be inhibited by addition of specific saccharides (Muchmore, A. V., and Decker, J. M. (1987) J. Immunol. 138, 2541-2546). We now report that uromodulin binds recombinant human tumor necrosis factor (rTNF) with high affinity. Both diacetylchitobiose and Man(alpha 1-6)(Man(alpha 1-3]-Man-O-ethyl are effective inhibitors of the binding, whereas a wide variety of other saccharides are not inhibitory. Although Tamm-Horsfall glycoprotein contains predominantly tetraantennary N-linked chains, the binding to rTNF is unaffected by removal of terminal sialic acid, galactose, and N-acetylhexosamine residues. Fractionation of a Pronase digest of uromodulin by gel filtration yields material that inhibits the binding of uromodulin to rTNF but is of lower molecular weight than the major oligosaccharide. Uromodulin does not inhibit the cytotoxic activity of rTNF as monitored by lysis of tumor cell targets but effectively protects mice from lethal challenge with lipopolysaccharide, an event that may involve lymphokine toxicity. We have previously shown that rTNF binds to sections of human kidney and is localized in the same region as uromodulin. Thus, rTNF interacts with uromodulin via carbohydrate chains that are less processed than the major tetraantennary chain, and this interaction may be critical in promoting clearance and/or reducing toxicity of TNF and other lymphokines.

Bovine serum sialic acid: age-related changes in type and content.

1. The sialic acid content of newborn calf serum (4.8 mumol/ml) is approx. 3-fold higher than that of mature animals (1.4 mumol/ml) and decreases to 2.4 mumol/ml at 20 days of age. Colostrum-fed and colostrum-deprived calves have similar levels of sialic acid from birth to 14 days of age. 2. The high level of sialic acid in newborn calf serum is due predominantly to N-acetylneuraminic acid, since this sialic acid accounts for 93% of the total and since less than 5% of the sialic acid is O-acetylated. 3. Comparison of day 0 and day 20 serum by gel filtration and by SDS polyacrylamide gel electrophoresis demonstrates that the increase in sialic acid is associated with increased production and/or sialylation of components with MW of 45-60 kDa. 4. A high percentage (64%) of the sialic acid in newborn calf serum is detected with the lipid-linked sialic acid assay, relative to 20 day old (25%) or mature (18%) animals. 5. This indicates that the glycoproteins of newborn calf serum are more efficiently extracted under the conditions of this assay than glycoproteins of mature serum.

Uromodulin (Tamm-Horsfall glycoprotein): a renal ligand for lymphokines.

The protein portion of the immunosuppressive glycoprotein uromodulin is identical to the Tamm-Horsfall urinary glycoprotein and is synthesized in the kidney. Evidence that the glycoproteins are the same is based on amino acid sequence identity, immunologic cross-reactivity, and tissue localization to the thick ascending limb of Henle's loop. Nucleic acid sequencing of clones for uromodulin isolated from a complementary DNA bank from human kidney predicts a protein 639 amino acids in length, including a 24--amino acid leader sequence and a cysteine-rich mature protein with eight potential glycosylation sites. Uromodulin and preparations of Tamm-Horsfall glycoprotein bind to recombinant murine interleukin-1 (rIL-1) and human rIL-1 alpha, rIL-1 beta, and recombinant tumor necrosis factor (rTNF). Uromodulin isolated from urine of pregnant women by lectin adherence is more immunosuppressive than material isolated by the original salt-precipitation protocol of Tamm and Horsfall. Immunohistologic studies demonstrate that rIL-1 and rTNF bind to the same area of the human kidney that binds to antiserum specific for uromodulin. Thus, uromodulin (Tamm-Horsfall glycoprotein) may function as a unique renal regulatory glycoprotein that specifically binds to and regulates the circulating activity of a number of potent cytokines, including IL-1 and TNF.

Immunosuppression, sialic acid, and sialyltransferase of neonatal and maternal bovine serum.

Serum samples of 7 cows from -10 to +10 days following parturition and of 7 calves from 0 to 20 days following birth were tested for the ability to inhibit mitogen-stimulated proliferation of lymphocytes, for cortisol and progesterone concentrations, and for sialic acid and sialyltransferase activity. Calf serum inhibited phytohemagglutinin (PHA)-stimulated proliferation of lymphocytes, with maximal inhibition at 12-24 h following birth, whereas no consistent immunosuppressive activity was detected in the maternal serum. Sialic acid was greatly elevated in calf serum (4.8 +/- 0.2 mumol/ml) relative to adult control values (1.3 +/- 0.1) and decreased continuously from day 0 to day 20. Sialic acid of maternal serum was slightly elevated prior to parturition (1.7 +/- 0.1) and increased to peak at 2.5 +/- 0.1 on day 8 following parturition. Sialyltransferase of both maternal and calf serum increased dramatically following parturition to peak at day 2. For calf serum, a moderate correlation was observed between sialic acid and cortisol concentration (r = 0.71) and between sialic acid and suppression of PHA-stimulated proliferation (r = 0.60). The results demonstrate that serum of 12-24 h-old calves is immunosuppressive in vitro, and suggest that changes in sialic metabolism may accompany cortisol-related immunosuppression in these animals.

Cell surface sialomucin and resistance to natural cell-mediated cytotoxicity of rat mammary tumor ascites cells.

MAT-B1 and MAT-C1 ascites sublines of the 13762 rat mammary adenocarcinoma both contain sialomucin as a major cell surface component and are resistant to cytolysis by normal rat spleen lymphocytes [3 +/- 2% (SD) and 0 +/- 1%, respectively]. Susceptibility to lysis did not increase following treatment of cells with neuraminidase, fucosidase, or alpha- or beta-galactosidase. Treatment with trypsin significantly increased the susceptibility of MAT-B1 (14 +/- 3%) but not MAT-C1 (5 +/- 2%). Following 1 month in culture, the sialomucin content of MAT-B1 cells dropped from 30% to 8% (determined by glucosamine labeling) and natural cell-mediated cytolysis increased to 16 +/- 4%, whereas the sialomucin content and susceptibility of MAT-C1 cells did not change. The results indicate that the relatively minor changes associated with removal of cell surface sialic acid or fucose residues do not result in increased susceptibility of the ascites cells to cytolysis. However, susceptibility of MAT-B1 cells to lysis by normal rat spleen lymphocytes was inversely correlated with the amount of major glycoprotein (r = -0.96).

Sialyltransferase of bovine serum: age- and hormone-related changes.

Sialyltransferase activity of bovine serum with the acceptor asialofetuin exhibits a pH optimum at 6.0-6.5, no divalent cation dependence, and a Km for CMP-sialic acid of 0.05 mM. Although a 2-fold increase in sialyltransferase activity with the acceptor asialofetuin is observed in serum samples from 2-day-old vs 20-day-old calves, the relative activity towards other glycoprotein acceptors is not different between the groups. With the acceptor lactose, the major product (greater than 95%) for all samples is 3'-sialyllactose, suggesting that the elevated levels of sialyltransferase in 2-day-old calves are due to Gal-R (alpha 2-3) sialyltransferase.

Immunosuppression, sialic acid, and sialyltransferase of bovine serum as a function of progesterone concentration.

Serum samples from progesterone-oestrogen-treated ovariectomized Holstein cows (N = 4) were compared with samples from control ovariectomized Holstein cows (N = 4) to determine the effects of physiological levels (0-6 ng/ml) of circulating progesterone. The average progesterone level in treated animals rose from 1 ng/ml (Day 0) to plateau at 5 ng/ml (Days 12 to 36). Sera from progesterone-oestrogen-treated cows during Days 4 to 10 significantly suppressed stimulation of lymphocytes by phytohaemagglutinin as compared with sera from control cows (P = 0.02), whereas no differences were detected in serum samples from Days 12 to 36. Serum samples from progesterone-oestrogen-treated or control cows did not affect the stimulation of lymphocytes by pokeweed mitogen. Sialyltransferase activity (P = 0.0002) and sialic acid content (P = 0.006) were both significantly elevated in serum from progesterone-oestrogen-treated animals compared with controls during Days 8 to 16, whereas no significant differences were observed at later times. The results suggest that suppression of phytohaemagglutinin-induced stimulation, sialic acid content, and sialyltransferase activity are sensitive not to the circulating level of progesterone but rather to increases in progesterone concentration, with maximal effects observed at Days 8, 12 and 12, respectively, after the start of progesterone treatment. The work provides a preliminary basis for further studies on the mechanism of immunosuppression by steroids and during pregnancy.

N-Acetylneuraminic acid and N-glycolylneuraminic acid in the O-linked oligosaccharides of a tumor cell glycoprotein. Incorporation and distribution.

The MAT-B1 and MAT-C1 ascites sublines of the 13762 rat mammary adenocarcinoma, which differ in several cell surface properties, contain a major mucin-type glycoprotein, termed ASGP-1. The sialic acid content of MAT-C1 ASGP-1 is 2-3-fold greater than MAT-B1 ASGP-1 (Sherblom, A. P., Buck, R. L., and Carraway, K. L. (1980) J. Biol. Chem. 255, 783-790). Sialic acid analysis demonstrated that, whereas MAT-C1 ASGP-1 contained approximately equal amounts of N-acetylneuraminic acid (NeuAc) and N-glycolylneuraminic acid (NeuGl), MAT-B1 ASGP-1 was devoid of NeuGl. MAT-B1 microsomes also did not contain NeuGl. MAT-B1 cells incubated with [3H]N-acetylmannosamine did not synthesize either labeled CMP-NeuGl or free NeuGl, even though the CMP-sialic acid synthetase was active with the substrate NeuGl. Thus, MAT-B1 cells may be deficient in the enzyme N-acetylneuraminate monooxygenase. The O-linked oligosaccharides from both MAT-B1 and MAT-C1 ASGP-1 have been shown to contain a core tetrasaccharide Gal(beta 1-4)GlcNAc(beta 1-6)(Gal(beta 1-3]GalNAc in which both galactose residues may be linked to additional sugars (Hull, S. R., Laine, R. A., Kaizu, T., Rodriquez, I., and Carraway, K. L. (1984) J. Biol. Chem. 259, 4866-4877). The distribution of NeuAc and NeuGl between the two galactose termini of the core tetrasaccharide was examined for MAT-C1 ASGP-1. Oligosaccharides were released by alkaline-borohydride treatment of MAT-C1 ASGP-1 which had been labeled with [14C]glucosamine and galactose oxidase/B3H4. Following fractionation by Bio-Gel P-4, DEAE-Sephadex, and high-performance liquid chromatography, oligosaccharides were analyzed for NeuAc and NeuGl and for susceptibility to digestion with beta-galactosidase. Three disialylated oligosaccharides were identified containing 2 mol of NeuAc (5.5% recovery), 2 mol of NeuGl (4.5%), or 1 mol each of NeuAc and NeuGl (11.1%). For monosialylated oligosaccharides, NeuGl appeared preferentially associated with the Gal(beta 1-4)GlcNAc terminus (9.0%), whereas significant amounts of oligosaccharide containing NeuAc at both the Gal(beta 1-3)GalNAc (2.6%) and Gal(beta 1-4)GlcNAc (4.5%) termini were detected. Each of the major qualitative differences between MAT-B1 and MAT-C1 oligosaccharides, including the presence of NeuGl (MAT-C1), sulfate (MAT-B1), and alpha-linked galactose (MAT-B1), occurs at the Gal(beta 1-4)GlcNAc terminus.

Sialyltransferase of the 13762 rat mammary ascites tumor cells.

The MAT-B1 and MAT-C1 ascites sublines of the 13762 rat mammary adenocarcinoma differ in morphology, agglutinability with concanavalin A, and xenotransplantability. Both cell lines contain a major mucin-type glycoprotein, but the MAT-C1 (xenotransplantable) subline contains a 3-fold-greater content of sialic acid on the glycoprotein than does the MAT-B1 (nonxenotransplantable) subline. The present work indicates that whole cells of both lines incorporate radioactivity from labeled CMP-sialic acid into a component which comigrates with the major glycoprotein by sodium dodecyl sulfate polyacrylamide gel electrophoresis, and that label incorporated by MAT-B1 cells is released by alkaline-borohydride treatment. Sialyltransferase can be purified from 250- to 400-fold by chromatography of a Triton X-100 extract of microsomes on CDP-agarose. The purified fraction of both cell lines has a Km for CMP-sialic acid of 0.40 +/- 0.10 mM with asialofetuin as the acceptor, and gives 35 to 40% of the activity with the acceptor asialotransferrin as with asialofetuin. When assayed with a variety of acceptors, the MAT-C1 extract showed higher sialyltransferase activity at a pH below 6.5 than did the MAT-B1 extract. Analysis of the products following incubation with lactose yields only 3'-sialyllactose for both cell lines. The results indicate that the differences in MAT-B1 and MAT-C1 sialyltransferase when assayed with glycoprotein acceptors are not large enough to account for the differences in sialic acid content of the two cell lines.

Specificity of submaxillary gland sialyltransferases.

A method has been developed to determine the activities of specific sialyltransferases by analysis of the products of the reaction. This method, which utilizes high performance liquid chromatography, distinguishes addition of sialic acid to the N-acetylgalactosamine vs. galactose residues of the mucin disaccharide Gal beta(1 leads to 3)GalNac, and can be used to distinguish formation of the 3'- and 6'-isomers of sialyllactose. For the bovine, ovine, and porcine submaxillary extracts, more than 95% of the activity with asialo ovine submaxillary mucin is due to formation of NeuAc alpha(2 leads to 6)GalNAc. With lactose as the acceptor, more than 95% of the alpha(2 leads to 3) isomer is produced. Activity with asialofetuin is due solely to the O-linked chain, with relative activity toward the galactose vs. GalNAc residues of 0.32, 1.5, and 0.10 for bovine, ovine, and porcine, respectively. The rat submaxillary gland extract showed equal formation of 3'- and 6'-sialyllactose, and very low activity with asialo ovine submaxillary mucin. However, at least 40% of the activity toward the Gal beta(1 leads to 3)GalNAc disaccharide of asialofetuin was directed toward the GalNAc residue. The relative preference of the N-acetylgalactosaminide alpha(2 leads to 6) sialyltransferase for a monosaccharide vs. a substituted GalNAc may play a role in regulation of chain length during mucin synthesis.