The application of FAIMS gas analysis in medical diagnostics.

There is an ever increasing need to develop new tools to aid in the diagnosis and monitoring of human diseases. Such tools will ultimately reduce the cost of healthcare by identifying disease states more quickly and cheaply than current practices. One method showing promise is the analysis of gas-phase biomarkers from human breath, urine, sweat and stool that reflect bodily metabolism. Analysis of these volatiles by GC MS requires specialised infra-structure and staff, making it unsuitable for a clinical setting. Point of care sensor based technologies such as eNoses often suffer from stability and sensitivity issues. Field-Asymmetric Ion Mobility Spectrometry (FAIMS) has potential to fulfil this clinical need. In this paper we review the medical need, the technology, sampling methods and medical evidence thus far. We conclude with reflecting on future developmental steps necessary to bring the device into medical practice.

Cell therapy attenuates deleterious ventricular remodeling and improves cardiac performance after myocardial infarction.

BACKGROUND: Myocardial infarction (MI) promotes deleterious remodeling of the myocardium, resulting in ventricular dilation and pump dysfunction. We examined whether supplementing infarcted myocardium with skeletal myoblasts would (1) result in viable myoblast implants, (2) attenuate deleterious remodeling, and (3) enhance in vivo and ex vivo contractile performance. METHODS AND RESULTS: Experimental MI was induced by 1-hour coronary ligation followed by reperfusion in adult male Lewis rats. One week after MI, 10(6) myoblasts were injected directly into the infarct region. Three groups of animals were studied at 3 and 6 weeks after cell therapy: noninfarcted control (control), MI plus sham injection (MI), and MI plus cell injection (MI+cell). In vivo cardiac function was assessed by maximum exercise capacity testing and ex vivo function was determined by pressure-volume curves obtained from isolated, red cell-perfused, balloon-in-left ventricle (LV) hearts. MI and MI+cell hearts had indistinguishable infarct sizes of approximately 30% of the LV. At 3 and 6 weeks after cell therapy, 92% (13 of 14) of MI+cell hearts showed evidence of myoblast graft survival. MI+cell hearts exhibited attenuation of global ventricular dilation and reduced septum-to-free wall diameter compared with MI hearts not receiving cell therapy. Furthermore, cell therapy improved both post-MI in vivo exercise capacity and ex vivo LV systolic pressures. CONCLUSIONS: Implanted skeletal myoblasts form viable grafts in infarcted myocardium, resulting in enhanced post-MI exercise capacity and contractile function and attenuated ventricular dilation. These data illustrate that syngeneic myoblast implantation after MI improves both in vivo and ex vivo indexes of global ventricular dysfunction and deleterious remodeling and suggests that cellular implantation may be beneficial after MI.

A specific structural alteration in the heparan sulphate of human glomerular basement membrane in diabetes.

AIMS/HYPOTHESIS: Heparan sulphate proteoglycan is an important component of the glomerular anionic filtration barrier and its reduced amount in diabetes contributes to glomerular dysfunction. The objective of this study was to determine if there is also an alteration in the sulphation pattern of the diabetic heparan sulphate chains. METHODS: The heparan sulphate in the glomerular basement membrane/mesangial matrix from human diabetic and nondiabetic kidneys obtained at autopsy was fragmented by a hydrazine/nitrous acid procedure and after radiolabelling with NaB[3H]4, the disaccharide products were chromatographically resolved and quantified. RESULTS: Six sulphated disaccharides were identified in both the diabetic and nondiabetic samples and the molar distribution of these was similar, with the notable exception of the iduronic acid-2-O-sulphatectl--> 4glucosamine-3-O-sulphate species which occurred in the diabetic glomeruli in less than half the amount as in the nondiabetic samples (9.0% compared to 18.7% of total sulphated disaccharides, p < 0.005). CONCLUSION/INTERPRETATION: 3-O-sulphated glucosamine is a rare constituent of heparan sulphate occurring usually in a glucuronic acidbeta1--> 4glucosamine-3-O-sulphate(+/- 6-O-sulphate) sequence within the antithrombin-binding domain. In the glomerular basement membrane where the 3-O-sulphated glucosamine is present in substantial amounts, however, it occurs exclusively in an iduronic acid-containing sequence. It is likely that the recently discovered 3-O-sulphotransferase variant which specifically acts on the iduronic acidalpha1--> 4glucosamine sequence is decreased in human diabetes and moreover that this unusual disaccharide could be a component of a specific heparan sulphate domain which interacts with bioactive proteins.

Human anti-porcine T cell response: blocking with anti-class I antibody leads to hyporesponsiveness and a switch in cytokine production.

Intervention in the molecular interactions that lead to an immune response is possible at various stages of Ag recognition and T cell activation. Perturbation of the interaction of the TCR with the MHC/peptide ligand complex is one approach that has shown promise for autoimmunity and graft rejection in blocking T cell-activated responses. In this study, we investigated the effect of altering the target MHC class I molecule by blocking with Abs. We established a system that analyzed the human T cell response against MHC class I+/class II- porcine stimulatory cell targets. The primary human response against porcine smooth muscle cells was CD8+ T cell dependent. In the presence of F(ab')2 fragments of the MHC class I-reactive Ab, PT-85, the proliferative response was inhibited and production of IL-2 and IFN-gamma was blocked. Moreover, in a secondary response, proliferation was reduced and type 1 cytokine levels were inhibited. In contrast, levels of IL-10 and IL-4 were sustained or slightly increased. These findings indicate that Ab against MHC class I blocked the recognition of porcine cells by the human CD8+ T cells and altered the cytokine secretion profile. Thus, a single treatment with PT-85 F(ab')2 directed against the MHC class I molecule provides an attractive approach to the induction of T cell tolerance that may provide long-term graft survival in porcine-to-human cell transplantation.

Xenogeneic transplantation of porcine hepatocytes into the CCl4 cirrhotic rat model.

Liver support using extracorporeal devices and hepatocyte transplantation has received renewed interest for the management of acute and chronic liver failure. The aim of this study was to determine whether xenogeneic porcine hepatocytes could integrate into the liver parenchyma of cirrhotic Lewis rats when administered by an intrasplenic route. Cirrhosis was induced by carbon tetrachloride (CCl4) inhalation and confirmed histologically. Freshly isolated porcine hepatocytes were infused directly into the splenic pulp at laparotomy over a 5-15-min interval. Using (111)In-labeled hepatocytes, the degree of localization of porcine hepatocytes to the spleen and liver was found to be greater than 60% in both control and cirrhotic rats. Integration of porcine hepatocytes into the rat liver parenchyma was determined by immunohistochemical staining for porcine albumin in rat liver sections. Further confirmation was provided by in situ hybridization using a porcine-specific probe that binds to a distinct repetitive element (PRE) in porcine DNA. Evidence of integrated porcine hepatocytes was seen for over 50 days in animals under cyclosporine immunosuppression. These data demonstrate the integration of xenogeneic porcine hepatocytes into the liver of the cirrhotic rat and their ability to produce porcine albumin for up to 50 days.

Xenogeneic cell therapy: current progress and future developments in porcine cell transplantation.

The multitude of distinct cell types present in mature and developing tissues display unique physiologic characteristics. Cellular therapy is a novel technology with the promise of utilizing this diversity to treat a wide range of human degenerative diseases. Intractable diseases, disorders, and injuries are characterized by cell death or aberrant cellular function. Cell transplantation can replace diseased or lost tissue to provide restorative therapy for these conditions. The limited use of cell transplants as a basis for current therapy can, in part, be attributed to the lack of available human cells suitable for transplantation. This has prevented further realization of the promise of cell transplantation as a platform technology. Accordingly, cell-based therapies such as blood transfusions, for which the cells are readily available, are a standard part of current medical practice. Despite numerous attempts to expand primary human cells in tissue culture, current technological limitations of this approach in regard to proliferative capacity and maintenance of the differentiated phenotype has prevented their use for transplantation. Further, use of human stem cells for the derivation of specific cell types for transplantation is an area of future application with great potential, but hurdles remain in regard to deriving and sufficiently expanding these multipotential cells. Thus, it appears that primary cells are at present a superior source for transplantation. This review focuses on pigs as a source of a variety of primary cells to advance cell therapy to the clinic and implement achievement of its full potential. We outline the advantages and disadvantages of xenogeneic cell therapy while underscoring the utility of transplantable porcine cells for the treatment of human disease.

Analysis of polymorphism in porcine MHC class I genes: alterations in signals recognized by human cytotoxic lymphocytes.

Elucidation of the mechanism of the immune response against transplanted porcine tissue is critical for the success of xenografting in humans. Both human T cells and NK cells recognize MHC Ags, and human receptors may bind to MHC Ags across species barriers. Molecular characterization of porcine MHC class I clones from two MHC class I loci (P1 and P14) obtained from homozygous inbred miniature swine of three haplotypes (aa, cc, and dd), revealed extensive conservation between loci, suggesting that the genes were products of duplication from a common ancestral sequence. The level of homology between loci was similar to that between the haplotypes at each locus, suggesting that intergenic exchange had limited divergence of these genes. Comparison of the alleles indicated that the polymorphism occurred in the alpha-1 and alpha-2 domains of the class I heavy chain, while the alpha-3 domain was highly conserved among the six genes analyzed. Amino acids in the alpha-2 and alpha-3 domains responsible for the binding of human CD8 to MHC class I were largely conserved in the porcine genes, but several critical residues were altered. Comparison of sequences recognized by human NK cell inhibitory receptors revealed that the residues critical for recognition by these receptors were altered in the porcine genes; thus, the porcine class I molecules would be unable to inhibit lysis by human NK clones characterized to date. This finding provides a likely explanation for the susceptibility of porcine cells to cytolysis by human NK cells.

Structure of the O-linked oligosaccharides from a major thyroid cell surface glycoprotein.

A major glycoprotein at the surface of calf thyroid cells, GP-3 (M(r) 20,000), contains the I-antigenic activity of calf thyroid which has been attributed to its poly-N-acetyllactosamine N-linked saccharide chains (Edge, A. S. B., and Spiro, R. G., J. Biol. Chem. 260, 15332-15338, 1985). The present study demonstrated that alkaline borohydride treatment of GP-3 results in the release of five neutral and five acidic saccharides that were found to represent over 30% of the saccharides of this carbohydrate-rich glycoprotein. Three of the oligosaccharides contained terminal alpha1 --> 3-linked galactose residues which accounted for their affinity toward Bandeiraea simplicifolia I lectin. The saccharides could be grouped into several distinct categories on the basis of their internal sequence. A novel tetrasaccharide in GP-3 was shown to have the structure: Gal alpha1 --> 3Gal beta1 --> 6(Gal beta1 --> 3)GalNAcH2. An unsubstituted N-acetylgalactosamine unit and a Gal beta1 --> 3GalNAc disaccharide were prominent O-linked constituents, with the disaccharide serving as a core unit for the attachment of sialic acid residues to form tetra- and trisaccharides. A branched core structure, Gal beta1 --> 3(GlcNAcbeta1 --> 6)GalNAcH2 was shared by 4 of the 10 saccharides, the most complete of which was assigned the sequence NeuAc alpha2 --> 3Gal beta1 --> 3(Gal alpha1 --> 3Gal beta1 --> 4GlcNAc beta1 --> 6)GalNAcH2.

Human natural killer cells account for non-MHC class I-restricted cytolysis of porcine cells.

Despite similarities in the cellular response to allografts and xenografts, some aspects of the xenogeneic immune response are unique. We find that both freshly isolated and primed human peripheral blood lymphocytes manifest MHC unrestricted cytolysis of porcine cells. While natural antibody-mediated mechanisms account for variable levels of cytotoxicity, reproducible killing in the absence of human serum is attributable to natural killer (NK) cells. This was shown by cold target inhibition with K562 cells, increased antiporcine cytotoxicity after enrichment for CD56+ cells, and significantly reduced lytic activity after depletion of CD56+ cells. Increased anti-porcine cytotoxicity after mixed culture of human and porcine cells was due to differentiation of NK cells to lymphokine-activated killer (LAK) cells and was IL-2 dependent. After depletion of NK cells, T-cell-mediated anti-porcine cytotoxicity could also be demonstrated. We conclude that the human anti-porcine cellular cytotoxic response is due to multiple cell types that include T cells in addition to NK and LAK cells.

Reduction of serum cholesterol in Watanabe rabbits by xenogeneic hepatocellular transplantation.

Transplantation of xenogeneic hepatocytes would provide a novel therapy for liver disease and would help to solve the problem of an insufficient supply of donor organs. We have tested whether xenogeneic cells infused into the liver could correct the metabolic defect in the Watanable heritable hyperlipidemic (WHHL) rabbit, an animal model for homozygous familial hypercholesterolemia, and we have investigated whether the infused cells traverse the lining of the portal vasculature. We find that porcine hepatocytes are localized in the hepatic sinusoids after surgery and subsequently migrate out of the vessels and integrate into the hepatic parenchyma. The integrated porcine hepatocytes provide functional LDL receptors that lower serum cholesterol in the WHHL rabbit by 30-60% for at least 100 days.

Enzymatic removal of alpha-galactosyl epitopes from porcine endothelial cells diminishes the cytotoxic effect of natural antibodies.

Xenotransplantation of tissues between discordant species such as pig into human is not yet feasible due to the problem of hyperacute rejection. This rapid response to xenogeneic tissue is mediated by natural antibodies that react with antigens on the xenograft. A number of xenoantigens consist of carbohydrate residues, and a terminal galactose in alpha linkage has been shown to be involved in hyperacute rejection of pig-to-human xenografts. We show that alpha-linked galactose on porcine endothelial cells is a major epitope recognized by IgG and IgM antibodies present in monkey and human sera. Endothelial cells that had been treated with alpha-galactosidase did not react with fluorescein-labeled Griffonia simplicifolia I B4 (GS-IB4), a lectin that detects the alpha-galactosyl epitope on intact cells. The reactivity of both human and cynomolgus monkey serum with endothelial cells was decreased by 59% to 90% after treatment with coffee bean alpha-galactosidase. Using a colorimetric assay for cell viability, we show that natural antibodies present in the sera of cynomolgus monkey and humans are cytotoxic to porcine endothelial cells in the presence of exogenously added complement. When the terminal alpha-galactosy residues were removed by enzymatic digestion, the cytotoxic effect of natural antibodies on porcine endothelial cells was diminished by > 80%. Evaluation of the time course of reappearance of the alpha-galactosyl epitope at the cell surface revealed that 48 hr after alpha-galactosidase treatment, binding of GS-IB4 was diminished by 60%. These results suggest that glycosidase treatment of cells to be transplanted could prevent hyperacute rejection mediated by natural antibodies.

Porcine repeat element DNA: in situ detection of xenotransplanted cells.

Using a digoxygenin-labelled DNA probe derived from the porcine repeat element PRE-1, we have developed a protocol for the detection of transplanted porcine islets and hepatocytes against a background of murine host tissue. Analysis of this probe by Southern blotting indicated that PRE-1 hybridizes to pig genomic DNA but not to human or mouse DNA. On tissue sections, hybridizing probe was detected using alkaline phosphatase-conjugated antidigoxygenin antibody visualized with 5-bromo-4-chloro-3-indolyl-phosphate/4-nitro-blue tetrazolium chloride (BCIP/NBT) substrate. We have demonstrated sensitive and highly specific staining of porcine nuclei in fixed, paraffin embedded tissue sections, and have applied the technique to detect porcine pancreatic islets and hepatocytes transplanted into murine kidney and spleen. Application of this technique include detection of transplanted cells or organs across the variety of xenogeneic barriers.

Effect of dexamethasone on the carbohydrate chains of the insulin receptor.

Dexamethasone treatment of IM-9 lymphocytes and Fao hepatoma cells resulted in an increase in synthesis of the insulin receptor. The receptors synthesized after stimulation with the glucocorticoid had altered carbohydrate structure. The carbohydrate side chains of the insulin receptor were less branched on the dexamethasone-treated cells; i.e., the ratio of saccharides with three and four branches to those bearing only two branches was decreased. The predominant polymannose oligosaccharide after dexamethasone treatment was Man9GlcNAc (vs Man6GlcNAc in the control cell). Both of these changes are consistent with a less complete processing of the N-linked carbohydrate units and were not observed for the total cellular glycoproteins, whereas all glycoproteins manifested an increased sialylation in Fao cells after dexamethasone treatment. These data indicate that glucocorticoid treatment results in alterations in branching of carbohydrate side chains, in the size of polymannose chains and in sialylation of the insulin receptor.

Insulin receptor carbohydrate units contain poly-N-acetyllactosamine chains.

The insulin receptor was immunoprecipitated from cultured human lymphocytes (IM-9) and rat hepatocytes (Fao) after biosynthetic labeling with [3H]glucosamine or [3H]mannose, and the nature of the carbohydrate units was investigated. Digestion of the receptor from IM-9 lymphocytes with E. freundii endo-beta-galactosidase increased the migration of the insulin receptor alpha- and beta-subunits on sodium dodecyl sulfate-polyacrylamide gels and sharpened the electrophoretic bands; the alpha-subunit was converted from an apparent mol wt (Mr) of 123,000 to a Mr of 118,000, and the beta-subunit from a Mr of 92,000 to 89,000. The susceptibility of the insulin receptor to this enzyme indicates that its carbohydrate units contain poly-N-acetyllactosamine sequences. Affinity chromatography of receptor glycopeptides on Concanavalin-A-Sepharose revealed that the poly-N-acetyllactosamine units were attached to multiantennary glycopeptides that accounted for over 75% of the [3H]glucosamine incorporated into the IM-9 lymphocyte insulin receptor; the remaining radioactivity was present in polymannose units (primarily Man8GlcNAc2) and biantennary complex saccharides. Several differences in the carbohydrate chains of the insulin receptor from the Fao and IM-9 cells indicated that glycosylation was cell specific despite the occurrence of poly-N-acetyllactosamine chains in both cell types. The IM-9 lymphocyte receptor glycopeptides were larger (Mr, 3,200-9,500) and more susceptible to endo-beta-galactosidase than those from the Fao receptor (Mr, 3,000-5,000). Moreover, the released saccharides from the Fao receptor were found by exoglycosidase digestions and chromatographic comparison to standards to contain terminal sialic acid in both alpha 2----3 and alpha 2----6 linkage to galactose, whereas the IM-9 carbohydrate units contained only alpha 2----3-linked sialic acid.

Characterization of novel sequences containing 3-O-sulfated glucosamine in glomerular basement membrane heparan sulfate and localization of sulfated disaccharides to a peripheral domain.

Fragmentation of the heparan sulfate chains from bovine glomerular basement membrane (GBM) by hydrazine/nitrous acid treatment followed by NaB3H4-reduction yielded a mixture of six sulfated disaccharides containing D-glucuronic (GlcUA) or L-iduronic acid (IdUA) and terminating in 2,5-anhydro[3H]mannitol (AnManH2), in addition to the nonsulfated component GlcUA beta 1----4AnManH2. Among these products two novel disaccharide units were identified as IdUA alpha 1----4AnManH2(3-SO4) and IdUA(2-SO4)alpha 1----4AnManH2(3-SO4); these accounted for 22% of the total sulfated species indicating that there are 2-3 residues of 3-O-sulfated glucosamine/heparan sulfate chain. The disulfated disaccharide was shown through its release by direct nitrous acid treatment to be situated in a GlcNSO3-IdUA(2-SO4)-GlcNSO3(3-SO4) sequence which is distinct from that in which 3-O-sulfated glucosamine is located in the antithrombin-binding region of heparins. Analyses of heparan sulfate from lens capsule, a nonvascular basement membrane, indicated the absence of sequences containing 3-O-sulfated glucosamine, although otherwise the sulfated disaccharides produced by hydrazine/nitrous acid/Na-B3H4 treatment (GlcUA beta 1----4AnManH2(6-SO4), IdUA alpha 1----4AnManH2(6-SO4), IdUA(2-SO4)alpha 1----4AnManH2 and IdUA(2-SO4)alpha 1----4AnManH2(6-SO4] were the same as from GBM. Examination of the GBM heparan sulfate domains after nitrous acid treatment indicated that the O- as well as N-sulfate groups are clustered in an iduronic acid-rich 10-disaccharide peripheral segment, while the internal region (approximately 20 disaccharides) is composed primarily of repeating GlcUA beta 1----4GlcNAc units. The localization of chain diversity to the outer region may facilitate interactions of the heparan sulfate with other macromolecular components.

Presence of an O-glycosidically linked hexasaccharide in fetuin.

Examination by gel filtration, thin layer and anion exchange chromatography of the O-linked carbohydrate units released from fetuin by alkaline borohydride treatment indicated the presence in this glycoprotein of an acidic glucosamine-containing hexasaccharide in addition to the previously described tetra- and trisaccharides. The structure of the hexasaccharide was determined to be NeuAc alpha 2----3Gal beta 1----3[NeuAc alpha 2----3Gal beta 1----4GlNAc beta 1----6]GalNAc, on the basis of exoglycosidase digestion, periodate oxidation, and methylation analysis as well as hydrazine-nitrous acid fragmentation. The latter procedure when carried out on the reduced asialohexasaccharide yielded Gal----2-deoxygalactitol and Gal----anhydromannose which were shown to be derived, respectively, from Gal----N-acetylgalactosaminitol and Gal----GlcNAc sequences. Reductive amination of the Gal----anhydromannose disaccharide with [14C] methylamine permitted identification of its linkage as 1----4. While Diplococcus pneumoniae endo-alpha-DN-acetylgalactosaminidase acting on asialofetuin released the sialic acid-free tetra- and trisaccharides (Gal beta 1----3GalNAc), this enzyme did not cleave the peptide attachment of the asialohexasaccharide (Gal beta 1----3 [Gal beta 1----4GlcNAc beta 1----6] GalNAc). The number of O-linked hexa-, tetra-, and trisaccharides per fetuin molecule was determined to be 0.2, 0.7, and 2.1, respectively, on the basis of galactosaminitol analyses. The absence of O-linked N-acetylglucosamine-containing tetra- or pentasaccharides in fetuin suggest that the attachment of this sugar is a rate-limiting step; furthermore, the limited occurrence of the hexasaccharide may indicate that the addition of sialic acid to Gal beta 1----3GalNAc to form the NeuAc alpha 2----3Gal linkage precludes action of the GlcNAc transferase to form the branch point on the GalNAc residue.

Selective deglycosylation of the heparan sulfate proteoglycan of bovine glomerular basement membrane and identification of the core protein.

The heparan sulfate proteoglycan of the bovine glomerular basement membrane (Mr = 200,000, 30% carbohydrate by weight) has been deglycosylated by various chemical and enzymatic procedures to identify the core protein and provide information about the N- and O-linked saccharide units. Heparitinase digestion of the proteoglycan reduced its Mr to 143,000, consistent with the removal of its four glycosaminoglycan chains with the exception of short segments adjacent to the carbohydrate-protein linkage region, whereas nitrous acid treatment brought about a smaller reduction in size (to Mr = 168,000) which was shown to be due to the resistance of the internal portion of the heparan sulfate polymer to this reagent. Incubation of the heparitinase-digested proteoglycan with peptide N-glycosidase F decreased its Mr by about 8,000 and liberated oligosaccharides which were primarily acidic in nature; since endo-beta-N-acetylglucosaminidase H did not bring about any saccharide release, it appears that the N-linked carbohydrate units (three per molecule) occur exclusively as the complex type. Treatment of the proteoglycan with trifluoromethanesulfonic acid, a reagent which cleaves all saccharide units, yielded the core protein which migrated as a single discrete band (Mr = 128,000) on polyacrylamide gel electrophoresis. Although the native and heparitinase-treated proteoglycan reacted with concanavalin A and Bandeiraea simplicifolia I, the core protein had no affinity for these lectins, and this loss of reactivity can be attributed to the removal of the N- and small O-linked saccharides. However, the immunoreactivity of the deglycosylated protein with antiserum directed against the intact proteoglycan was to a large measure (80%) preserved, suggesting that the polyclonal response to this glomerular basement membrane glycoconjugate is primarily directed against determinants on the polypeptide portion.

Characterization of O-glycosidically linked oligosaccharides of rat erythrocyte membrane sialoglycoproteins.

The carbohydrate units of the rat erythrocyte membrane sialoglycoprotein rSGP-4 [Edge, A. S. B., & Weber, P. (1981) Arch. Biochem. Biophys. 209, 697-705] have been characterized. All of the carbohydrate of this Mr 19,000 glycoprotein occurs in O-glycosidic linkage to the peptide; following alkaline borohydride treatment and chromatography on Bio-Gel P-2, sialic acid containing oligosaccharides terminating in N-acetylgalactosaminitol were obtained. Their structures were determined by compositional analysis, exoglycosidase digestions, alkaline sulfite degradation, and periodate oxidation. The oligosaccharides were characterized for molecular weight and linkage by direct chemical ionization and gas-liquid chromatography/mass spectrometry, respectively. The structures are proposed to be NeuAc alpha 2----3Gal beta 1----3GalNAc-ol, Gal beta 1----3(NeuAc alpha 2----6)GalNAc-ol, NeuAc alpha 2----3Gal beta 1----3(NeuAc alpha 2----6)GalNAc-ol, and NeuAc alpha 2----3Gal beta 1----3(NeuAc alpha 2----3Gal beta 1----4GlcNAc beta 1----6)GalNAc-ol. Two of the N-acetylglucosamine-containing hexasaccharides were present per molecule of rSGP-4 along with two trisaccharides and seven tetrasaccharides.