Carbohydrate Attachment Research Articles

The amino acid sequence of the light chain of human high-molecular-mass kininogen.

The complete amino acid sequence of the light chain of human high-molecular-mass kininogen has been determined. The peptide chain contains 255 amino acid residues. The half-cystine, which forms the disulfide bridge to the heavy chain, was identified in position 225. Nine carbohydrate attachment sites were found. All carbohydrate side chains are O-glycosidically linked. Alignment of the present sequence with the bovine kininogen light chain sequence shows a high degree of homology, except for an extension of 22 amino acids within the histidine-rich part of the sequence. The histidine-rich region may have arisen by gene multiplication during evolution.

Complete amino acid sequence of human vitronectin deduced from cDNA. Similarity of cell attachment sites in vitronectin and fibronectin.

cDNA clones for vitronectin, a cell adhesion-promoting plasma and tissue protein, were isolated from a lambda gt11 library containing cDNA inserts made from human liver mRNA. The library was screened with anti-vitronectin antibodies and the positive clones were further identified with synthetic oligonucleotide probes deduced from the partial amino acid sequence of vitronectin. Nucleotide sequence analysis showed that the largest insert was 1545 bp long and contained the whole sequence corresponding to plasma vitronectin. It showed that vitronectin contains the entire 44-amino acid somatomedin B peptide at its NH2 terminus and, near its COOH terminus, a 34-amino acid glycosaminoglycan binding site in which half of the amino acids are basic residues. Three potential carbohydrate attachment sites are present in the sequence. An Arg-Gly-Asp sequence, which has previously been shown to be the cell attachment site in fibronectin, was found in vitronectin immediately after the NH2-terminal somatomedin B sequence. No other homologies with fibronectin were found. The Arg-Gly-Asp sequence appears to constitute the cell attachment site of vitronectin, since it is in the region where we have previously localized the cell attachment site, its presence correlate with cell attachment activity among the insert-coded polypeptides, and because previous results have shown that synthetic peptides containing the Arg-Gly-Asp sequence inhibit the cell attachment function of vitronectin. The discovery of an Arg-Gly-Asp cell attachment site in a protein with a known cell attachment function emphasizes the general importance of this sequence in cell recognition.

Complete nucleotide and derived amino acid sequences of the fourth component of mouse complement (C4). Evolutionary aspects.

The nucleotide sequence coding for the fourth component of mouse complement (C4) has been determined from a cloned genomic DNA fragment and a cloned cDNA fragment. The amino acid sequence of the protein was deduced. The single chain precursor protein (pro-C4) consists of 1719 amino acid residues. The mature beta, alpha, and gamma subunits contain 654, 766, and 291 amino acids, respectively. One potential carbohydrate attachment site is predicted for the beta chain, three for the alpha chain, and none for the gamma chain. From a comparison with human C4 cDNA sequence an extensive overall sequence homology, 79% in nucleotides and 76% in amino acids, is observed. There is conservation in both the position and number of cysteine residues in human and mouse C4. We compared the mouse C4 amino acid sequences with those of mouse C3 and human alpha 2-macroglobulin and the evolutionary relationship among these three proteins is discussed.

Complete primary structure of the human alpha 2 type V procollagen COOH-terminal propeptide.

Recently we presented the partial covalent structure of a type V collagen chain. Analysis of amino acids 796-1020 in the human alpha 2(V) Gly-X-Y region showed strong conservation of charged positions with the interstitial collagens but also revealed substitutions unique to type V. To gain more information about this procollagen and primarily to resolve the ambiguous nature of the 3' noncollagenous propeptide, we sequenced several cDNA clones coding for amino acids adjacent to the carboxyl end of the alpha chain. Here we report the complete primary structure of the alpha 2(V) COOH-terminal propeptide. In general, the latter sequence (270 residues) bears a greater degree of similarity to those of the interstitial rather than the basement membrane procollagens. Compared to the interstitial procollagens, however, more divergence has occurred in alpha 2(V) surrounding the conserved N-asparaginyl-linked carbohydrate attachment site at residues 171-173, and alpha 2(V) possesses an additional potential glycosylation site (Asn-Lys-Thr) located in a hypervariable region near the NH2 terminus. Although certainly premature to form any rigid hypothesis, a pattern emerges that may be characteristic of alpha 2 versus alpha 1 chains. Both the alpha 2(I) and alpha 2(V) telopeptides are devoid of a lysine, which in alpha 1 chains forms an interchain cross-link with residue 87 of the collagenous region. Also in contrast to the interstitial alpha 1 carboxyl propeptides is the absence in alpha 2(I) and alpha 2(V) of a cysteine that probably participates in an interchain disulfide bond. Therefore, one can speculate that those alpha 2 chains, represented only once in procollagen trimers, may not be under the same selective pressure as alpha 1 chains to maintain certain residues responsible for stabilizing the triple helical molecules.

The structure of prothrombin fragment 1 at 3.5-A resolution.

The structure of prothrombin fragment 1 has been determined at 3.5-A resolution by multiple isomorphous replacement methods with four heavy atom derivatives. The final average figure of merit is 0.72. There is a large cylindrical solvent region with an average diameter of 35-40 A along the entire length of the c axis (85 A) centered at about x = y = 1/2. The connected density forming the wall of this channel is not of sufficient extent to account for the 156 residues of fragment 1 and the two accompanying carbohydrate chains totaling 5000 in molecular weight. Deglycosylated fragment 1 crystallizes isomorphously with fragment 1, and a difference map between the two revealed that the sugar chains are severely disordered and reside in the solvent channel. Although the disordered carbohydrate and the complexity of five disulfides in a 126-residue sequence have hampered the complete tracing of the peptide chain, two-thirds of the molecule has been accounted for in the form of an unusually oblate ellipsoid of about 15 X 30 X 35 A. The folding of the molecule has little secondary structure (one alpha-helix (7%), 20% beta-structure) in agreement with dichroism measurements and one of the points of carbohydrate attachment is suggested from the deglycosylated difference map.

Evidence for a conformational change in deglycosylated glycoprotein hormones

Removal of long-chain asparagine-linked carbohydrates leads to loss of receptor-effector coupling in gonadotropin target tissue. This implies a direct interaction between carbohydrate and cell membrane components. To examine other mechanisms by which carbohydrate could activate post-receptor events, we have used sequence-specific and conformation-specific α-subunit radioimmunoassays as probes for conformational changes in deglycosylated choriogonadotropin and follitropin. Immunoreactivity of either hormone was enhanced 8–13-fold after removal of carbohydrate by anhydrous hydrogen fluoride. Removal of sialic acid alone had little effect on reactivity. Based on the specificity of the antisera, the effect could be localized to a region in the amino-terminus remote in the linear sequence from actual sites of carbohydrate attachment. The results suggest that a conformational change in the α-subunit could account, at least in part, for the observed effects of deglycosylation on glycoprotein hormone action. Carbohydrate Glycoprotein Follitropin Choriogonadotropin Conformation Membrane receptor

Human complement component C3: cDNA coding sequence and derived primary structure.

The complete cDNA coding sequence and derived amino acid sequence of human complement component C3 are presented. The encoded precursor molecule contains a signal peptide of 22 amino acid residues, the beta chain (645 residues), and the alpha chain (992 residues). The two chains are joined by four arginine residues not present in the mature protein. Several functionally important sites have been localized, such as the thiolester site, the cleavage site liberating the anaphylatoxin, and two sites of cleavage by the serine protease factor I, as well as a peptide fragment with leukocyte mobilizing activity. At least two carbohydrate attachment sites, one on each chain, have been identified. Human C3 has 79% identity to mouse C3 at the nucleotide level and 77% identity at the amino acid level. The protease alpha 2-macroglobulin and complement component C4 show considerable homology to C3, suggesting that the three proteins have evolved from a common ancestor.

Natural-Abundance, 13C-Nuclear Magnetic Resonance-Spectral Studies of Carbohydrates Linked to Amino Acids and Proteins

This chapter presents the 13 C-nuclear magnetic resonance-spectral studies of carbohydrates linked to amino acids and proteins. This technique is dealt with in terms of the information that may be gained about the glycoprotein without its modification, which would lead to the destruction of the native sample and possible microheterogeneity of the oligosaccharide chains. The covalent attachment of carbohydrates to amino acid residues of glycoproteins falls into three main categories including the glycosidic ( O -glycosyl) linkage, the N -glycosyl linkage, and the S -glycosyl linkage. The analysis of the anomeric-carbon region of two glycoproteins— human α 1 -acid glycoprotein (α 1 -AGP) and calf fetuin—provides additional information not available by more conventional methods. Any structural analysis of such a complex system must be based on the assignments of the various signals in the spectra. Their spectra shows that resonances corresponding to inner carbohydrate display residues a much larger line-width than resonances corresponding to outer residues. This reflects the relative mobility of such residues and aids in distinguishing between terminal residues and inner residues of the intact oligosaccharide chains of intact glycoproteins. The structural analysis of glycopeptides and glycoprotein containing O -glycosyl linkages is also elaborated in the chapter.

Structure of murine complement component C3. II. Nucleotide sequence of cloned complementary DNA coding for the alpha chain.

From the inserts of two recombinant plasmids isolated from a murine liver cDNA library, the nucleotide sequence coding for the C3 alpha chain was obtained and the corresponding amino acid sequence was derived. The alpha chain portion of the mRNA is 2979 nucleotides long, specifying a polypeptide of 993 amino acids. The molecular weight of the alpha chain, in the absence of carbohydrate, was calculated from the sequence data to be 112,933. Two possible carbohydrate attachment sites were predicted at residues 269 (Asn) and 947 (Asn). In addition, the positions for two putative factor I cleavage sites were predicted from a comparison with the cleavage sites in the human C3 alpha chain. The C3 alpha chain contains 24 cysteine residues, 10 of these clustered in the C-terminal 175 amino acids of the alpha chain. Together with the accompanying report (Lundwall, A, Wetsel, R.A., Domdey, H., Tack, B.F., and Fey, G.H. (1984) J. Biol. Chem. 259, 13851-13856), this study completes the nucleotide and amino acid structure of the murine precursor prepro-C3 molecule.

Structure of murine complement component C3. I. Nucleotide sequence of cloned complementary and genomic DNA coding for the beta chain.

The nucleotide sequence coding for the beta chain of murine C3 was determined from cloned cDNA and genomic DNA fragments. Sonicated subfragments were randomly inserted into the bacteriophage M13 and sequenced using the dideoxynucleotide technique. Each nucleotide was sequenced on average six times in these studies. The derived amino acid sequence includes a signal peptide and a tetra-arginine sequence between the beta and alpha subunits in the precursor polypeptide prepro-C3. Together with the accompanying report (Wetsel, R.A., Lundwall, A., Davidson, F., Gibson, T., Tack, B.F., and Fey, G.H. (1984) J. Biol. Chem. 259, 13857-13862), this paper completes the analysis of the coding sequences for the prepro-C3 polypeptide. The derived molecular weight of the unglycosylated beta chain (642 amino acids) is 70,641. The sequences of the first two introns in the murine C3 gene and of the 5'-flanking 106 nucleotides are also reported. The 5'-flanking region contains a TATA consensus sequence in agreement with an earlier report (Wiebauer, K., Domdey, H., Diggelmann, H., and Fey, G.H. (1982) Proc. Natl. Acad. Sci. U. S. A. 79, 7077-7081), presumed to be involving in regulating the expression of the C3 gene. A striking feature of the derived sequence was that only 3 cysteine residues were found, all located in the C-terminal part of the polypeptide chain. No carbohydrate attachment sites were predicted in the beta chain.

Renin cleavage of a human kidney renin substrate analogous to human angiotensinogen, H-Asp-Arg-Val-Tyr-Ile-His-Pro-Phe-His-Leu-Val-Ile-His-Ser-OH, that is human renin specific and is resistant to cathepsin D

A synthetic tetradecapeptide, H-Asp-Arg-Val-Tyr-Ile-His-Pro-Phe-His-Leu-Val-Ile-His-Ser-OH, which corresponds to the 13 amino terminal residues of human angiotensinogen plus a carboxy terminal serine to replace a suggested site of carbohydrate attachment, has been shown to be a good substrate for human kidney renin. At pH 7.2 and 37°C the K m or Michaelis constant was 8.4 ± 2.9 μ m, and the V m or velocity at infinite tetradecapeptide concentration was 11.3 ± 2.4 μmol angiotensin I made per hour per milligram renin. The tetradecapeptide was highly resistant to cleavage by mouse submaxillary renin. The tetradecapeptide was also slowly cleaved by human liver cathepsin D, by rabbit lung angiotensin-converting enzyme, and by reconstituted human serum, but did not yield angiotensin I. Thus, this synthetic renin substrate should permit more specific measurement of human kidney renin activity.

Tissue plasminogen activator: peptide analyses confirm an indirectly derived amino acid sequence, identify the active site serine residue, establish glycosylation sites, and localize variant differences.

Tissue plasminogen activator, separated into variants I and II (differing in Mr by 2000-3000), was reduced and [14C]carboxymethylated. Fragments from cleavages with enzymes and cyanogen bromide (CNBr) were separated by reverse-phase high-performance liquid chromatography and subjected to sequence degradations. All seven CNBr fragments were purified and found to be compatible with the cDNA-derived amino acid sequence [Pennica, D., Holmes, W. E., Kohr, W. J., Harkins, R. N., Vehar, G. A., Ward, C. A., Bennett, W. F., Ylverton, E., Seeburg, P. H., Heynecker, H. L., Goeddel, D. V., & Collen, D. (1983) Nature (London) 301, 214-221]. Chemical characterization of 93% of the 527 residues recovered in 50 peptides confirmed the indirectly deduced primary structure of the protein. The tryptic peptide patterns from the two variants were found to differ for one peptide (T15). Since carbohydrate was present in this peptide for variant I and since a marked difference in chromatographic behavior for T15 was observed in variant II, we conclude that carbohydrate differences in this peptide (i.e., Asn-184 in the numbering system of the cDNA-derived amino acid sequence) are the explanation for the size differences between variants I and II. Carbohydrate was also found at two other positions in the protein, corresponding to Asn-117 and Asn-448. However, a fourth potential glycosylation site, Asn-218, is apparently not utilized for carbohydrate attachment. The enzyme is inactivated by diisopropyl phosphorofluoridate, which covalently modifies the serine residue corresponding to position 478, identifying this as the active site serine residue.(ABSTRACT TRUNCATED AT 250 WORDS)

Structure and sequence of the chicken type II procollagen gene. Characterization of the region encoding the carboxyl-terminal telopeptide and propeptide.

The DNA sequence of two overlapping cDNA clones and a genomic lambda clone covering the region coding for 288 amino acids at the COOH terminus of the chicken type II procollagen gene is reported. This region consists of 4 exons coding for the last 15 amino acids of the triple helical domain and 273 amino acids which correspond to the COOH-terminal telopeptide and COOH-terminal propeptide. The sequence, base composition, and codon usage of this region of the type II procollagen gene show particularly high similarity to those of the chicken alpha 1(I) procollagen gene and differ from those of the alpha 2(I) and alpha 1(III) gene sequences. Two DNA tracts of low sequence similarity were observed. One of these regions spans the telopeptide and COOH-terminal propeptidase cleavage site, although 4-5 amino acids at the actual cleavage site are conserved compared with the alpha 1(I) and alpha 2(I) genes. A region of unusually high nucleotide sequence conservation is present in exon 2 (amino acids 171c - 186c ) consisting of approximately 45 nucleotides with only one or two base substitutions compared with the other procollagen genes. Within this conserved sequence is a site for carbohydrate attachment. The 3' nontranslated sequence of the type II procollagen mRNA is longer than that of either the alpha 1(I) or alpha 2(I) mRNA and contains several unusual long tracts consisting primarily of one or two bases. Although the canonical site for polyadenylation is not present, two related sequences, AACAAA and ATATAAA , are present 32 and 41 bases preceding the end of the major RNA species. The exon/intron structure of the type II procollagen gene is similar to that of other collagen genes which have been described. This DNA sequence provides the first extensive report of the amino acid sequence of chicken type II procollagen.

The amino acid sequence of the rabbit glycoprotein hormone alpha subunit

The amino acid sequence of the alpha subunit of rabbit (lagomorph) lutropin (lLH) has been determined. Overlapping peptides from trypsin and chymotrypsin digestions were isolated by reverse-phase high-pressure liquid chromatography (HPLC). Sequencing was by the dansyl-Edman procedure. Amide placements were established by HPLC analysis of the PTH amino acid derivatives. The proposed sequence of lLH alpha subunit is (asterisks denote carbohydrate attachment sites): This proposed sequence is highly homologous with the porcine, murine, ovine, and bovine glycoprotein hormone alpha subunit sequences. Two unusual proteolytic cleavages were observed: (1) a cleavage by trypsin between Asn-77 and Ala-78, and (2) a cleavage by chymotrypsin between Ala-45 and Arg-46. Similar enzymatic cleavages were previously reported for equine chorionic gonadotropin alpha subunit by Wardet al. and for these sites in the ovine LH alpha subunit by Liuet al. Chymotrypsin cleaved on the carboxyl side of methionine sulfone residues at positions 51 and 75.

Release of glucose-containing polymannose oligosaccharides during glycoprotein biosynthesis. Studies with thyroid microsomal enzymes and slices.

Incubations of thyroid microsomes with radiolabeled dolichyl pyrophosphoryl oligosaccharide (Glc3Man9-GlcNAc2) under conditions optimal for the N-glycosylation of protein resulted in the release, by apparently independent enzymatic reactions, of two types of neutral glucosylated polymannose oligosaccharides which differed from each other by terminating either in an N-acetylglucosamine residue (Glc3Man9GlcNAc1) or a di-N-acetylchitobiose moiety (Glc3Man9GlcNAc2). The first mentioned oligosaccharide, which was released in a steady and slow process unaffected by the addition of EDTA, appeared to be primarily the product of endo-beta-N-acetylglucosaminidase action on newly synthesized glycoprotein and such an enzyme with a neutral pH optimum capable of hydrolyzing exogenous glycopeptides and oligosaccharides (Km = 18 microM) was found in the thyroid microsomal fraction. The Glc3Man9GlcNAc2 oligosaccharide, in contrast, appeared to originate from the oligosaccharide-lipid by a rapid hydrolysis reaction which closely paralleled the N-glycosylation step, progressing as long as oligosaccharide transfer to protein occurred and terminating when carbohydrate attachment ceased either due to limitation of lipid-saccharide donor or addition of EDTA. There was a striking similarity between oligosaccharide release and transfer to protein with lipid-linked Glc3Man9GlcNAc2 serving as a 10-fold better substrate for both reactions than lipid-linked Man9-8GlcNAc2. The coincidence of transferase and hydrolase activities suggest the possibility of the existence of one enzyme with both functions. The physiological relevance of oligosaccharide release was indicated by the formation of such molecules in thyroid slices radiolabeled with [2-3H]mannose. Large oligosaccharides predominated (12 nmol/g) and consisted of two families of components; one group terminating in N-acetylglucosamine, ranged from Glc1Man9GlcNAc1 to Man5GlcNAc1 while the other contained the di-N-acetylchitobiose sequence and included Glc3Man9GlcNAc2, Glc1Man9GlcNAc2, and Man9GlcNAc2.

Coding assignment and nucleotide sequence of simian rotavirus SA11 gene segment 10: location of glycosylation sites suggests that the signal peptide is not cleaved.

A cloned DNA copy of simian rotavirus SA11 genomic segment 10 was used to confirm the assignment of the nonstructural glycoprotein NCVP5 to this gene. Determination of the nucleotide sequence for gene 10 indicated that NCVP5 is 175 amino acids in length and has an N-terminal hydrophobic region with the characteristics of a signal sequence for membrane translocation. Unexpectedly, this region was also the location for the only two potential glycosylation sites within the molecule, asparagine residues 8 and 18. The carbohydrates carried by NCVP5 were of the high-mannose type, Man9GlcNAc and Man8GlcNAc, with the mannose 9 species predominating; no complex oligosaccharides were present. If these asparagine residues are the sites for carbohydrate attachment, this implies that cleavage of the putative signal peptide does not occur during the maturation of this nonstructural glycoprotein.

Nucleotide sequences of complementary deoxyribonucleic acids for the pro alpha 1 chain of human type I procollagen. Statistical evaluation of structures that are conserved during evolution.

Nucleotide sequences were determined for two cloned cDNAs encoding for over three-fourths of the pro alpha 1 (I) chain of type I procollagen from man. Comparison with previously published data on amino acid sequences of the alpha 1 (I) chain of type I collagen made it possible to examine mutations in the transcribed products of the gene which have occurred during the evolution of man, calf, rat, mouse, and chick. Comparison of the nucleotide sequences with the corresponding sequences of cDNAs from chick [Fuller, F., & Boedtker, H. (1981) Biochemistry 20, 996] and with cDNAs for the pro alpha 2(I) chain from man [Bernard, M.P., Myers, J. C., Chu, M.-L., Ramirez, F., Eikenberry, E. F., & Prockop, D. J. (1983) Biochemistry 22, 1139] demonstrated that selective pressure during evolution for 250 million or more years acted more strongly on the structure of the pro alpha 1 (I) chain than on the pro alpha 2(I) chain. To improve the reliability of the comparison, the nucleotide sequences were examined with a modification of previous procedures for evaluating mutations in replacement sites and silent sites. The corrected divergence for replacement sites between the alpha 1 (I) chains was 6 +/- 0.8% whereas it was 15 +/- 1.9% for the alpha 2(I) chains. The C-propeptide domain of the pro alpha 1 (I) chain was also highly conserved with a corrected divergence at replacement sites of 5 +/- 0.9%, a value that was not distinguishable from the value previously found for the C-propeptide of the pro alpha 2(I) chain. Therefore, a large part of the structure of both C-propeptides appears to be under selective pressure. Inspection of changes in the C-propeptide of the pro alpha 1 (I) chain suggested that there was a highly conserved region around the carbohydrate attachment site similar to the highly conserved region of 37 amino acids previously found in the C-propeptide of the pro alpha 2(I) chain. Two statistical tests, however, were unable to confirm nonrandom distribution of changes in the C-propeptide of the pro alpha 1(I) chain. The same tests established the presence of a nonrandom distribution in nucleotide changes of the C-propeptide of the pro alpha 2(I) chain. The 3'-noncoding region of the cDNA for pro alpha 1(I) of human type I procollagen showed no homology with the same region in the chick.(ABSTRACT TRUNCATED AT 400 WORDS)

Sequence of the hemagglutinin gene from influenza virus A/Seal/Mass/1 /80

A double-strand DNA copy of the influenza virus A/Seal/Mass/1/80 (H7N7) [seal] hemagglutinin (HA) gene was cloned into the plasmid pAT153/P vuII/8 and sequenced to deduce the primary amino acid sequence. The gene is 1731 nucleotides long and codes for a protein of 560 amino acids with a nonglycosylated molecular weight of 62098 Da. The deduced amino acid sequence displays similarities to all other sequenced hemagglutinins by retaining six of seven potential glycosylation sites, showing conversation in the number and position of cysteine residues, conservation in the fusion and anchor peptides, and conservation in the putative receptor site of the molecule. However, three features of the primary amino acid sequence could be distinguished from the H7 amino acid sequence of A/fowl plague/Rostock/34 (FPV), another avian H7 influenza virus which does not produce disease in mammals. First, the seal HA sequence has three fewer amino acids in the connecting peptide region of the HA than FPV. This lack of multiple basic amino acids in the connecting peptide is similar to that found in avirulent H7 avian strains and to mammalian serotypes Hl, H2, and H3. Second, the seal HA has gained four additional proline residues, all in HA1, as compared to FPV. These residues may alter the tertiary structure of the HA and ultimately contribute to the biological features of this virus. Third, the seal HA has lost a potential carbohydrate attachment site at residue 149 which lies at the tip of the HA structure. The loss of this carbohydrate could alter the seal HAs interaction with host cell receptors.

Is adrenal proenkephalin glcosylated?

Adrenal proenkephalin contains the sequence AsnSerSer that is a typical site for the attachment of asparagine-linked carbohydrate. The 5300- and 18,200-Da bovine adrenal proteins derived from proenkephalin contain this recognition sequence and were therefore analyzed for the presence of both amino and neutral sugars. Less than 0.05 mol of amino sugar and less than 0.1 mol of neutral sugar were found per mole of each protein. No amino sugar was detected in other high-molecular-weight adrenal [Met]enicephalin-containing proteins. Together these findings indicate that bovine adrenal proenkephalin does not contain asparagine-linked carbohydrate.

Glycosylated insulin complexed to Concanavalin A. Biochemical basis for a closed-loop insulin delivery system.

The oligosaccharides maltose, maltotriose, mannotriose, and mannotetrose have been chemically attached to insulin molecules. Incubation of oligosaccharide and insulin at different molar ratios, with or without addition of cyanoborohydride, showed a nearly linear increase in carbohydrate attachment over time. The intravenous t1/2 of 125I-labeled sugar-insulin derivatives was identical to that of unmodified insulin (3.0 min). Biologic activity of these derivatives, assessed in rats by use of a blood glucose depression assay, did not differ significantly from control. These glycosylated insulin molecules are reversibly bound to the glucose-binding lectin Concanavalin A (Con A). Such sugar-insulin/lectin complexes serve as an insulin reservoir from which sugar-insulin molecules are displaced by glucose. Release of sugar-insulin molecules is a function of the particular sugar-insulin and of the ambient glucose concentration. Glucose displacement of glycosylated insulin complexed to Con A is in direct proportion to the amount of glucose present in the surrounding fluid. At each glucose concentration, the relative binding affinity of the maltotriose derivative is less than that of the mannotriose derivative, while the relative binding affinity of both maltotriose and mannotriose are less than that of the mannotetrose derivative. Prolonged incubation at 37 degrees C causes sugar-insulin, like unmodified insulin, to spontaneously aggregate into high-mol-wt, nondiffusable complexes. This aggregation phenomenon was found to be markedly inhibited when glycosylated insulins were synthesized utilizing partially sulfated insulin. Results from the studies described in this report provide the biochemical basis for a closed-loop, glucose-controlled insulin delivery system, utilizing glycosylated insulin complexed to Con A.