Preparation of 1-Amino-1-Deoxyoligosaccharides and Derivatives Thereof

Certain proteins synthesized by animals, insects, plants and unicellular eukaryotes such as yeast are glycosylated; that is, oligosaccharide chains are covalently attached to the polypeptide backbones. These glycoproteins are primarily localized on the external surface of the plasma membranes of cells and on the lumenal face of intracellular organelles such as lysosomes or the Golgi apparatus. They are also secreted into extracellular fluids.

The carbohydrates moieties of the glycoproteins are believed to be involved in a variety of biological functions, influencing both the biological activity of individual glycoproteins as well as inter-and intracellular recognition events. The carbohydrate units of a particular glycoprotein can modulate the solubility of the protein, protect it from heat denaturation or proteolytic degradation, control its circulatory lifetime, direct its site of uptake in an animal, and influence its antigenicity.

The carbohydrate moieties attached to proteins exposed on the external face of the plasma membrane of cells are believed to be important determinants for biological recognition. They serve as cell surface receptors for certain growth factors, hormones, toxins, viruses, bacteria, and lectins. Cell surface carbohydrates are also involved in the interactions of cells with the extracellular matrix and with other cells.

In view of the known and postulated biological functions of the oligosaccharides of glycoproteins, these carbohydrate units have become the subject of intense interest and experimental investigation. Studies of the carbohydrates include efforst to analyze structure, to determine routes of subject of intense interest and experimental investigation. Studies of the carbohydrates include efforts to analyze structure, to determine routes of biosynthesis and catabolism, to examine the specificity of individual enzymes in the metabolic pathways, to modulate the biological activity of glycoproteins by alteration of their carbohydrates, to attach carbohydrates to proteins, and, of course, to determine their biological functions.

Free reducing oligosaccharides obtained from natural sources, and derivatives of the oligosaccharides, have been used extensively for such investigations. For most structural studies, it is desirable to introduce a reporter group such as a radioactive, fluorescent, or UV-active chromophore at the reducing end of the oligosaccharide chain. The most common procedures to accomplish this are reduction with NaB[.sup.3 H].sub.4 (S. Takasaki and A. Kobata, Meth. Enzymol., 50, 50-54 (1978)) or reductive amination with 2-aminopyridine and sodium borohydride. (S. Hase et al., J. Biochem (Tokyo), 95, 197-203 (1984)).

The resulting products and similar derivatives are also useful as substrates for studies of glycosyltransferases and glycosidases, and for studies of the interaction of carbohydrate chains with receptors.

Free reducing oligosaccharides or their derivatives have also been coupled through the reducing sugar to other materials for a variety of purposes. Examples include coupling of the oligosaccharide to a resin for use as an affinity ligand; coupling to a carrier protein in order to raise antibodies against the carbohydrate; and coupling to a protein to alter the protein's biological activity. Potential improvements in the biological activity of a protein include an increase in solubility, an increase in stability to proteolytic degradation or heat denaturation, a reduction in antigenicity or immunogenicity, an increase or decrease in serum lifetime, or a targeting to a specific cell type in the body such as hepatocytes or macrophages.

Many strategies have been employed for coupling carbohydrates to proteins or other substrates (Y. C. Lee and R. T. Lee, The Glycoconjugates, Vol. IV, M. I. Horowitz (Ed.), Academic Press, N.Y., 1982, pp 57-83; J. D. Aplin and J. C. Wriston, Jr. CRC Crit. Rev., 10, 259-306 (1981)). These include, but are not limited to, diazo coupling, thiocarbamylation, amidation, amidination, reductive alkylation, transglutamination and the use of bifunctional reagents such as 2,4,6-trichloro-s-triazine and 2, 4-diisocyanotoluene.

The reactions used to prepare the various oligosaccharide derivatives generally result in reduction of the pyranose ring to the open chain (alditol) form or in formation of an O- or S-glycosidic linkage at the reducing end of the oligosaccharide. These derivatization reactions can be difficult to carry out when working with oligosaccharides derived from Asn-linked glycan chains.

The present invention is concerned with enzymatic methods for obtaining 1-amino-1-deoxyoligosaccharides from peptides and proteins containing Asn-linked oligosaccharides and converting them to N-glycoside derivatives that preserve the beta-glycosylamine linkage and the pyranosyl ring of Asn-linked glycoproteins.

The present invention provides an enzymatic procedure for obtaining 1-amino-1-deoxyoligosaccharides from any glycopeptide or glycoprotein containing Asn-linked oligosaccharides, describes derivatives of the aminodeoxyoligosaccharides that are useful for studies of the structure, biosynthesis, and function of the carbohydrate moieties, and provides procedures for coupling the aminodeoxy sugars to proteins to enhance their biological properties.

FIG. 1 illustrates the mechanism for hydrolysis of Asn-linked oligosaccharides by amidase and N-glycosidase activities.

FIG. 2 illustrates the coupling of acyl derivatives of 1-amino-1-deoxyoligosaccharides to the .epsilon.-amino group of a lysine residue.

FIG. 3 illustrates the coupling of acyl derivatives of 1-amino-1-deoxyoligosaccharides to the thiol group of cysteine.

FIG. 4 illustrates the coupling of acyl derivatives of 1-amino-1-deoxyoligosaccharides to the carboxylate groups of glutamic acid or aspartic acid.

FIG. 5 illustrates the coupling of acyl derivatives of 1-amino-1-deoxyoligosaccharides to the amide group of glutamine.

FIG. 6 illustrates the hydrolysis of turkey ovomucoid glycopeptide by aspartyl N-acetyl glycosylamine amidohydrolase from almond emulsin. Portions of the .sup.1 H NMR spectrum recorded at six time points in the reaction are plotted (not to scale). The NMR signals for the anomeric proton (5.020 ppm) and N-acetyl group (2.000 ppm) of the N-acetylglucosamine-oligosaccharide attached to the asparagine residue and for the two beta-protons of asparagine (2.893 and 2.766 ppm) decrease in intensity as the hydrolysis reaction proceeds. At the same time, NMR signals appear for the N-acetyl group (2.037 ppm) in the reducing end residue of the free carbohydrate moiety and for the two beta-protons of aspartic acid (2.730 and 2.620 ppm) formed during the hydrolysis reaction. The NMR signal at 5.188 ppm is the anomeric proton of the alpha-anomer of the final equilibrated product; the appearance of the signal is delayed, consistent with the enzyme-catalyzed formation of a 1-amino-1-deoxyoligosaccharide which undergoes a subsequent nonenzymatic hydrolysis with release of ammonia.

FIG. 7 illustrates trapping of the 1-amino-1-deoxyoligosaccharide that is formed in the reaction of beta-aspartyl N-acetylglycosylamine amidohydrolase with turkey ovomucoid glycopeptide. FIG. 7A shows, a portion of the .sup.1 H-NMR spectrum of the TOG prior to hydrolysis by the enzyme. FIG. 7B shows the NMR signal, after a 1 h incubation of TOG with the enzyme, the reaction was quenched with acetic anhydride. The NMR signal at 5.045 ppm is the anomeric proton for the 1,2-diacetamido intermediate trapped in the beta conformation. The signal partially overlaps the anomeric proton of a mannose residue in the oligosaccharide (5.058 ppm). FIG. 7C shows the NMR signal, after a 4 h incubation the quantity of intermediate trapped in much less because of hydrolysis of the aminodeoxyoligosaccharide. The signal at 5.188 ppm is the anomeric proton of the alpha-anomer of the reducing end in the final equilibrated product. FIG. 7D shows the NMR signal for, the fully equilibrated carbohydrate moiety from the hydrolysis reaction.

FIG. 8 illustrates the N-acetyl groups of the carbohydrate moiety of turkey ovomucoid glycopeptide (A) and the 1-acetamido derivative formed when the hydrolysis reaction is quenched with acetic anhydride after a 1 h incubation (B). As shown in FIG. 8A, the N-acetyl group at 2,000 ppm is that of the N-acetylglucosamine attached to asparagine. FIG. 8B shows that, upon acetylation, a beta-1-acetamido derivative of the 1-amino-N-acetylglucosamine-oligosaccharide intermediate is formed. The NMR signals at 2.002 and 2.000 ppm are those of the 1-acetamido and 2-acetamido groups of the reducing terminal N-acetylglucosamine residue.

FIG. 9A is an .sup.1 H-NMR spectrum recorded at times 5.4-5.0 minutes for hydrolysis of turkey ovomucoid glycopeptide by aspartyl N-acetylglycosylamine amidohydrolase from Flavobacterium meningosepticum.

FIG. 9B is an .sup.1 H-NMR spectrum recorded at times 2.2-1.6 minutes for hydrolysis of turkey ovomucoid glycopeptide by aspartyl N-acetylglycosylamine amidohydrolase from Flavobacterium meningosepticum.

FIG. 9C is an .sup.1 H-NMR spectrum recorded at times 5.4-5.0 minutes for hydrolysis of turkey ovomucoid glycopeptide by aspartyl N-acetylglycosylamine amidohydrolase from Flavobacterium meningosepticum.

FIG. 9D is an .sup.1 H-NMR spectrum recorded at times 2.2-1.6 minutes for hydrolysis of turkey ovomucoid glycopeptide by aspartyl N-acetylglycosylamine amidohydrolase from Flavobacterium meningosepticum.

Procedures exist for the preparation of various glycosylamines by the reaction of ammonia or certain amines with free reducing monosaccharides (for example, H. S. Isbell and H. L. French, Meth. Carb. Chem., Vol. VIII, 255 (1980)). The resulting glycosylamines are stable at alkaline pH, but undergo rapid hydrolysis at pH values between about 4 and 6. In principle it should also be possible to convert free reducing oligosaccharides to glycosylamines by treatment with ammonia.

The 1-amino-1-deoxyoligosaccharides of the present invention are obtained by treatment of glycopeptides or glycoproteins that contain one or more Asn-linked oligosaccharides with a beta-aspartylglycosylamine amidohydrolase at neutral or alkaline pH. This class of enzyme cleaves the N-acetylglucosaminylaspartate bond to generate an aspartic acid residue and a 1-amino-1-deoxyoligosaccharide. The aminodeoxyoligosaccharide subsequently can undergo a nonenzymatic reaction to release ammonia (FIG. 1). As in the case of the simple glycosylamines, the rate of release of ammonia is pH dependent, being rapid at acidic pH and slow at alkaline pH.

Enzymes with aspartylglycosylamine amidohydrolase activity were first described in the chemical literature in 1965 (E. D. Kavergneva, Izv. Akad Nauk S.S.S.R., Ser. Khim., 10, 1911 (1965); M. Murakami and E. H. Eylar, J. Biol. Chem., 240, 556 (1965); C. P. J. Roston, J. C. Caygill, and F. R. Jeoms, Biochem. J., 97, 43 (1965).) These and similar enzymes from a variety of sources have been shown to be beta-aspartyl N-acetylglucosylamine amidohydrolases (M. Makino et al., Biochem. Biophys. Res. Comm., 24, 961 (1966); A. L. Tarentino and F. Maley, Arch. Biochem. Biophys., 130, 295-303 (1969); J. Conchie and I. Strachan, Biochem. J., 115, 709-715 (1969); J. Conchie et al., Biochem. J., 115, 717-723 (1969)). Their preferred substrate is beta-aspartyl N-acetylglucosamine (Asn-GlcNAc), although certain Asn-oligosaccharides are also hydrolyzed. Substitution of the amino or carboxyl groups of the asparagine gives rise to a substrate that is not hydrolyzed by the enzyme. Thus glycopeptides and glycoproteins are not substrates. The beta-aspartyl N-acetylglucosylamine amidohydrolases are believed to play a role in the catabolic breakdown of Asn-linked oligosaccharides. Deficiency of the human aspartyl N-acetylglucosylamine amidohydrolase, which has been purified to homogeneity from liver (M. M. McGovern et al., J. Biol. Chem., 258, 10743-10747 (1983), leads to a genetic deficiency disease called aspartylglucosaminuria. Clinical symptoms include mental retardation, facial and skeletal abnormalities, and excretion of high levels of Asn-GlcNAc in the urine.

Aspartyl N-acetylglucosylamine amidohydrolases have been purified from several sources including pig sera, hen oviduct, rat liver, and human liver. The enzymes generally have a broad pH optimum between 4.0 and 6.0. The reported molecular weight of the proteins varies between 80,000 and 110,000 kD).

Although it was originally believed that the aspartyl N-acetylglucosylamine amidohydrolase cleaved the N-glycosidic bond of the Asn-GlcNAc linkage, it was subsequently proposed that the enzyme is an amidase (FIG. 1). (M. Makino et al, Biochem. Biophys. Res. Comm., 24, 961 (1966)). The experimental evidence supporting this conclusion is:

1. The observation that at low pH stoichiometric amounts of aspartic acid, ammonia and N-acetylglucosamine are formed, but at alkaline pH, only a small percentage of the expected ammonia is released until the reaction mixture is acidified.

2. The detection of a compound with a chromatographic mobility identical to authentic 1-amino-1-deoxy-N-acetylglucosamine.

Aspartyl N-acetylglucosylamine amidohydrolases that appear to produce an aminodeoxy sugar intermediate include the enzyme from pig sera (M. Makino et al., Biochem. Biophys. Res. Comm., 24, 961 (1966)), hen oviduct (A. L. Tarentino and F. Maley, Arch. Biochem. Biophys., 130, 295-303 (1969), and rat liver (J. Conchie and I. Stracham, Biochem. J., 115, 709-715 (1969). In addition, it was observed that the enzyme from pig sera hydrolyzed an ovalbumin Asn-oligosaccharide to yield an oligosaccharide that had the characteristics of an aminodeoxy derivative.

In 1977, Takahashi (Biochem. Biophys. Res. Comm., 76, 1194-1201 (1977)) described an enzyme isolated from almond emulsin that was capable of cleaving the beta-aspartylglycosylamine linkage present in glycopeptides and certain glycoproteins but not in unsubstituted asparaginyl derivatives. In a subsequent paper, evidence was presented that the hydrolysis of the aspartyl N-acetylglycosylamine linkage preceded the release of ammonia. In that paper the authors concluded that the reaction catalyzed by the almond enzyme was analogous to the two step reaction carried out by the aspartyl N-acetylglucosylamine amidohydrolase and that "although the formation of an intermediate 1-amino-N-acetylglucosamine-oligosaccharide, has not been confirmed, . . . these experimental data support the view that the enzyme is not an N-glycosidase but an amidase." (N. Takahashi and H. Nishibe, J. Biochem., 84, 1467-1473 (1978)). This statement not withstanding, the same enzyme from almond emulsin and a second enzyme from Flavobacterium meningosepticum have been termed peptide: N-glycosidases following isolation and characterization (T. H. Plummer, Jr. and A. L. Tarentino, J. Biol. Chem., 256, 10243-10246 (1981); A. L. Tarentino and T. H. Plummer, Jr., J. Biol. Chem., 257, 10776-10780 (1982); E. M. Taga et al, Biochemistry, 23, 815-822 (1984); T. H. Plummer, Jr. et al, J. Biol. Chem., 259, 10700-10704 (1984); A. L. Tarentino et al, Biochemistry, 24, 4665-4671 (1985)). The almond enzyme has a broad specificity, being able to hydrolyze the bond between high mannose-, hybrid- and complex-type oligosaccharides and asparaginyl residues in glycopeptides and certain glycoproteins (N. Takahashi and H. Nishibe, Biochem. Biophys. Acta, 657, 457-467 (1981); H. Ishihara et al., ibid, 661, 274-279 (1981)). However, the enzyme acts slowly, if at all, on many glycoproteins, preferring glycopeptide substrates (A. L. Tarentino and T. H. Plummer, Jr., J. Biol. Chem., 257, 10776-10780 (1982)).

Although the aspartyl N-acetylglycosylamine amidohydrolase from almond emulsin has nearly the same pH rate profile as the aspartyl N-acetylglucosylamine amidohydrolases described earlier, it differs in several significant ways:

1. Its preferred substrates are glycopeptides;

2. The enzyme is unable to cleave Asn-oligosaccharides;

3. The enzyme has a lower molecular weight (67,000 kD); and

4. Glycopeptides bearing a single monosaccharide (GlcNAc) are not substrates. An enzyme that appears to be similar to the one from almond emulsin has been detected in jack bean meal (K. Sugujama et al, Biochem. Biophys. Res. Comm., 112 155-160 (1983)).

Finally, an enzyme with "peptide: N-glycosidase" activity has been detected in fermentation broths of Flavobacterium meningosepticum (T. H. Plummer, Jr., et al, J. Biol. Chem., 259, 10700-10704 (1984)) and purified to homogeneity (A. L. Tarentino et al Biochemistry, 24, 4665-4671 (1984)).

The Flavobacterium enzyme resembles the amidohydrolase from almond emulsin in that it:

1. Prefers glycoproteins and glycopeptides as substrates;

2. Is unable to hydrolyze Asn-oligosaccharides or glycopeptides bearing a single GlcNAc.

3. Has a similar broad specificity toward different oligosaccharides.

However several significant differences also are apparent. The Flavobacterium enzyme:

1. Is more active in hydrolyzing glycoprotein substrates;

2. Has a much lower molecular weight (35,000 kD); and

3. Has a much higher pH optimum (8.6).

In view of these significant differences, it is possible that the Flavobacterium amidohydrolase has an entirely different mechanism of action. The enzyme does appear to have amidase activity because during cleavage of Asn-oligosaccharides from glycopeptides, the asparagine residue is converted to aspartate (T. H. Plummer, Jr. et al., J. Biol. Chem.259, 10700-10704 (1984)). We are aware of no studies of the mechanism of the enzyme.

There is tremendous structural variety in the oligosaccharides linked by asparagine residues to proteins. The present invention can be applied to any that are substrates for an amidohydrolase. It is typical for a glycoprotein to exhibit microheterogeneity with respect to the oligosaccharides present on the molecule. Thus it will usually be necessary to purify either the aminodeoxyoligosaccharide or a derivative if a single homogeneous material is desired. In most cases it will probably prove easier to purify the derivative. Purification procedures will likely be some form of chromatography such as lectin, gel filtration, ion exchange, silica gel, or reverse phase chromatography.

Examples of relatively homogeneous oligosaccharides that may be obtained directly from glycoproteins without extensive purification include: H.sub.2 N--GlcNAc--GlcNAc--Man.sub.9 from soybean agglutinin, H.sub.2 N--GlcNAc--GlcNAc--Man.sub.5-8 from ribonuclease B, biantennary complex chains from fibrinogen or human transferrin, biantennary complex chains with a proximal fucose residue from bovine IgG, triantennary complex chains from fibrinogen or human complex chains from alpha.sub.1 -acid glycoprotein.

The enzymes from almond emulsin and Flavobacterium meningosepticum are particularly useful for releasing the aminodeoxyoligosaccharides because they have been purified to homogeneity and are commercially available (Seikagku, Japan; Genzyme Corporation, Boston, Mass.), they have a broad substrate specificity, they can utilize glycopeptides and glycoproteins as substrates, and they are active at neutral and alkaline pH. In the present invention the 1-amino-1-deoxyoligosaccharides produced in the reaction by the amidohydrolases are observed directly for the first time. Furthermore, it is demonstrated that the released amino deoxyoligosaccharides exist largely as the beta-anomer. Finally a 1-amino-1-deoxyoligosaccharide is trapped for the first time as a stable derivative.

It is known that acyl derivatives of glycosylamines can be prepared and that the acylated glycosylamines are generally stable compounds that do not undergo hydrolysis in aqueous solution (H. S. Isbell and H. L. French, Meth. Carb. Chem., Vol VIII, 255 (1980)). The Asn-oligosaccharide linkage in glycoproteins is itself an example of a stable acyl derivative of a glycosylamine. To date no aminodeoxyoligosaccharide prepared by enzymatic hydrolysis of a glycopeptide has been converted to an acyl derivative: The starting glycosylamines have previously been prepared through reaction of a free reducing sugar with ammonia.

In the present invention it is demonstrated that it is possible to form acyl derivatives of aminodeoxyoligosaccharides released from glycopeptides and glycoproteins by an amidohydrolase. In general a reactive acyl derivative such as an acid chloride, acid anhydride, or other active acyl compound, a large number of which are well known in the chemical literature, is mixed with the released 1-amino-1-deoxyoligosaccharide at alkaline or neutral pH. To minimize hydrolysis of the glycosylamine the enzymatic reaction and trapping is best carried out between pH 8 and 9. The trapping reaction can be performed in the presence of the hydrolyzed peptide or glycoprotein, but it is preferable to first remove the peptide or polypeptide. This can be accomplished by precipitation of the protein by addition of a solvent such as ethanol, dioxane or tetrahydrofuran. The trapping reaction can be performed in aqueous solution or in mixed solvent systems such as alcohol/water, tetrahydrofuran/water, dioxane/water, dimethylformamide or dimethyl sulfoxide. A base such as triethylamine, pyridine or sodium bicarbonate may be added to neutralize acid released during the reaction.

In certain instances the trapping reaction can be performed in a manner that produces the desired product directly. In other cases it will be desirable to trap the aminodeoxyoligosaccharide with a bifunctional linker that can itself then be coupled to another material in a subsequent reaction.

1. Structural Analysis of Oligosaccharides

Underivatized oligosaccharides are difficult to analyze by conventional means because of the lack of a "reporter" group. Indeed, it is routine procedure to introduce a label such as a fluorescent group or radioactive group to increase sensitivity. Two of the most common labeling procedures involve reductive amination with 2-aminopyridine and sodium borohydride (S. Hase et al., J. Biochem (Tokyo), 95, 197-203 (1984)) and reduction with sodium borotritide (S. Takasaki et al., Meth. Enzymol, 83, 263-268 (1982)). Each of these techniques has been employed to analyze oligosaccharides released by aspartaryl N-acetylglycosylamine amidohydrases, presumably following release of ammonia to generate the free reducing oligosaccharide (I. Ishii et al., J. Chromat., 345, 134-139 (1985)). The labeled oligosaccharides are then analyzed by techniques such as gel filtration chromatography on BioGel P4 columns (K. Yamashita et al., Meth. Enzymol., 83, 105-126 (1982)), high pressure liquid chromatography (I. Ishii et al., J. Chromat., 345, 134-139 (1985); J. Baenziger and M. Natowicz, Anal. Biochem., 112, 357-361 (1981); S. J. Mellis and J. Baenziger, Anal. Biochem., 114, 276-280 (1981) and exoglycosidase digestion (K. Yamashita et al., Meth. Enzymol., 83, 105-126 (1982)).

In one embodiment of the present invention, the aminodeoxyoligosaccharides released from glycopeptides or glycoproteins by an amidohydrolase can be trapped directly by a reactive derivative of the reporter group. Examples include, but are not limited to:

1. Formation of dimethylaminonaphthylsulfonamide (Dansyl) derivatives by treatment with dansyl chloride.

2. Formation of fluorescein derivatives by treatment with fluorescein isothiocyanate.

3. Formation of 4-dimethylaminoazobenzene 4'-sulfonamide (Dabsyl) derivatives by treatment of an aminodeoxyoligosaccharide with Dabsyl chloride.

4. Formation of a [.sup.3 H]- or [.sup.14 C] 1-acetamido-1-deoxyoligosaccharide by treatment of an aminodeoxyoligosaccharide with [.sup.3 H]- or [.sup.14 C] acetic anhydride.

2. Preparation of Substrates for Glycosyltransferases and Endo-and Exoglycosidases

The derivatives of the 1-amino-1-deoxyoligosaccharides described in the previous section also are useful as substrates for glycosyltransferases and exo-and endoglycosidases. Simple mono-and disaccharides are often used to assay these enzymes (for example, p-nitrophenylglycosides are used to assay exoglycosidases), but the substrates often bind poorly to the enzyme. It would be a significant advantage to perform the assays with oligosaccharides more closely resembling the natural substrates. In addition to accomplishing this objective, the 1-amino-1-deoxyoligosaccharide derivatives of the present invention maintain the beta-glycosylamine linkage and the pyranose form of the N-acetylglucosamine residue previously bonded to the asparagine.

3. Probes for Use in Cell Biology

The various derivatives described above have additional applications in the field of cell biology. These applications include studying the interaction of oligosaccharides with receptors on or isolated from cells, bacteria, or viruses. The derivatives may also be used to localize oligosaccharide binding proteins in tissue slices or whole animals.

4. Formation of Neoglycoproteins

For reasons described earlier, it is of interest to develop methods for attaching oligosaccharides to proteins (thereby forming "neoglycoproteins"). There are several reports of coupling synthetic glycosylamines to a protein by formation of an amide linkage with carboxyl groups of the protein. Wriston (FEBS Lett., 33, 93-95 (1973) treated chymotrypsinogen with 1-ethyl-3(3-dimethyl-aminopropyl)carbodiimide (EDC) and synthetic 1-amino-1-deoxyglucose. He reports coupling about 3 equivalents of aminoglucose to each chymotrysinogen molecule.

In a series of papers, Moczar and coworkers describe several glycosylamine derivatives of lysozyme prepared by EDC coupling at acidic pH (E. Moczar, Experientia, 29, 1576-1577 (1973); E. Moczar and J. Leboue, FEBS Lett., 50, 300-302 (1975); E. Moczar and G. Vass, Carbohydr. Res., 50, 133-141 (1976)). The glycosylamines that were employed included lactosylamine, glucosylamine, galactosylamine and mannosylamine, each prepared by reaction of the corresponding sugar with ammonia. The preferred coupling method of the present invention involves trapping the aminodeoxyoligosaccharide with a bifunctional reagent. This derivative is subsequently coupled to the protein.

A variety of chemistries, well known in the chemical literature, can be employed to couple the oligosaccharide derivative to the protein. (For example, see Y. C. Lee and R. T. Lee, in The Glycoconjugates, Vol. IV., M. I. Horowitz, Ed., Academic Press, 1982, pp 57-83; J. D. Aplin and J. C. Wriston, Jr., CRC Crit. Rev., 10, 259-306 (1981)).

With this method there is a "linker" or "spacer" interposed between the protein and the oligosaccharide. This strategy allows one to employ a wider range of coupling chemistries, minimizes side reaction with the protein and makes the oligosaccharide more accessible for binding by other molecules.

Among the amino acids with side chains that it is possible to derivatize are the .epsilon.-amino group of lysine, the hydroxyl group of serine, theorine or tyrosine, the thiol group of cysteine, the carboxylate group of aspartic acid and glutamic acid, the imidazole ring of histidine, the aromatic ring of phenylalamine, tryptophan or tyrosine and the amide group of glutamine. The C- and N-terminal groups are also subject to derivatization. Examples of preferred coupling procedures follow below:

i. .epsilon.-amino group of lysine and/or the N-terminal amino group of the protein (FIG. 2)

The 1-amino-1-deoxyoligosaccharide released by the amidohydrolase is first trapped by treatment with an excess of a cyclic anhydride such as succinic anhydride or glutaric anhydride. The free carboxylic acid group of the linker is then activated by formation of a derivative such as the N-hydroxysuccinimide ester. This activated derivative can then be reacted with the protein of choice.

ii. Thiol group of cysteine (FIG. 3)

The 1-amino-1-deoxyoligosaccharide is trapped by treatment with a linker possessing an activated acyl derivative and a maleimide moiety. The resulting product can be used to derivatize the protein of choice.

iii. Carboxylate groups of aspartic acid, glutamic acid or the C-terminal amino acid (FIG. 4)

The 1-amino-1-deoxyoligosaccharide is trapped by treatment with a linker possessing an activated acyl derivative and a protected amino group. After deprotection, the resulting product can be coupled to the protein with the aid of a coupling reagent such as EDC or carbonyldiimidazole. Although it is also possible to couple the aminodeoxyoligosaccharide directly to the protein as mentioned earlier, the procedure in FIG. 4 has the advantage that one does not have to be concerned with decomposition of the aminodeoxyoligosaccharide to ammonia and the free oligosaccharide at the preferred pH of the reaction with EDC (pH 4 to 6).

iv. The amide group of glutamine (FIG. 5)

The ##STR1## derivative described in Figure can be substituted for the NH.sub.2 group of glutamine through a reaction catalyzed by the enzyme transglutaminase (S. Yan and F. Wold, Biochemistry, 23, 3759-3765 (1983)).

One application of the derivatization reaction is to extend the serum lifetime of proteins. For example, it is known that small (30,000 kD) proteins rapidly cleared from the circulation primarily by the kidney, while larger proteins are removed by other mechanisms. Thus therapeutic proteins such as interleukin-2 have a short serum half-life because of renal clearance. By attaching appropriate oligosaccharides clearance by the kidney can be prevented. The optimum carbohydrate chains to attach in this case would be complex oligosaccharides, for example, those derived from fibrinogen, transferrin, fetuin or alpha-1-acid glycoprotein. Any of the coupling procedures described earlier could be employed.

A second application of the derivatization reaction is to target proteins to specific cell types. Cells with sugar-specific cell surface receptors can bind proteins in the circulation bearing the complementary carbohydrates and internalize the protein into the cell. There are three well-known sugar binding receptors: the asialoglycoprotein (galactose specific) receptor on hepatocytes, the mannose/N-acetylglucosamine receptor on reticuloendothelial cells, and the fucose receptor on hepatocytes and lymphocytes. (For a review, see E. Neufeld and G. Ashwell, in The Biochemistry Glycoproteins and Proteoglycans, W. Lennarz, Ed., Plenum Press, New York, 1980, 241-266). By coupling appropriate 1-amino-1-deoxyoligosaccharides, a protein can be targeted to each of these cell types. To target a protein to macrophages, the oligosaccharides from a glycoprotein such as soybean agglutinin or ribonuclease B could be employed. To target a protein to cells bearing the asialoglycoprotein receptor, one could employ complex oligosaccharides isolated from fibrinogen, transferrin, fetuin or alpha-1 acid glycoprotein after treatment (neuraminidase or dilute acid) to remove the terminal sialic acid residues.

5. Ligands for resins, beads, and surfaces. Using chemistries similar to those described for preparation of neoglycoproteins it is possible to prepare derivatives of various solid materials such as resins beads and films. Applications for these derivatized materials include:

i. Use as an affinity resin for the purification of receptors, lectins, glycosyltransferases, and glycosidases.

ii. Use in diagnostic kits for detection of oligosaccharides, receptors, lectins, glycosyltransferases and glycosidases.

Examples 1 and 2 that follow are taken from J. M. Risley and R. L. van Etten, J. Biol. Chem., 260, 15488-15494 (1985).

To a solution of 1 mg of turkey ovomucoid glycopeptide (a heterogeneous mixture of three structurally related complex type oligosaccharides purified as described by E. M. Taga et al., Biochemistry, 23, 815-822 (1984)) in 50 mM phosphate, p.sup.2 H 7.0 buffer in 99.96% .sup.2 H.sub.2 O equilibrated at the desired temperature was added the almond amidohydrolase enzyme to give a total volume of 0.5 ml. The reaction was followed by .sup.1 H NMR. The hydrolysis of 1-amino-1-deoxy-N-acetylglucosamine in 50 mM phosphate, p.sup.2 H 7.0, buffer in 99.96% .sup.2 H.sub.2 O to 2-acetamido-2-deoxy-D-glucose was followed by .sup.1 H NMR as a reference compound. The proton peaks of interest were integrated and the pseudo first-order rate constant was calculated from the slope of a plot of the natural logarithm of the peak area as a function of time.

The two-step hydrolysis reaction of turkey ovomucoid glycopeptide (TOG) to intermediate (INT) to the final carbohydrate product (CHO), which is assumed to be irreversible, is described by two independent reactions. ##STR2## where k.sub.1, and k.sub.2 are the rate constants for each step. The set of differential equations which describes the reaction in Equation 1 is ##EQU1##

The differential equations are solved by the variation of parameters method. When the initial concentrations of INT and CHO are zero, the solution reduces to ##EQU2## where [TOG] is the initial concentration of the enzyme substrate.

The .sup.1 H NMR spectrum of 2-acetamido-2-deoxy-D-glucose was recorded in order to model the (temperature-dependent) chemical shifts that were likely for the sugar at the reducing end of the new oligosaccharide that was produced during the enzyme-catalyzed hydrolysis reaction. The cheical shifts observed were 2.043 ppm for the N-acetyl group and 5.195 ppm (J.sub.1,2 =3.58 Hz) and 4.709 ppm (J.sub.1,2 =8.40 Hz) for the anomeric proton, the former the alpha-anomer and the latter the beta-anomer. The carbohydrate moiety of the turkey ovomucoid glycopeptide that is isolated from the first domain is a complex-type oligosaccharide. Spectra from a typical hydrolysis reaction of turkey ovomucoid glycopeptide incubated with the almond amidohydrolase and followed by .sup.1 H NMR are shown in FIG. 6. The anomeric proton and the N-acetyl group for the N-acetylglucosamine-oligosaccharide attached to asparagine in the turkey ovomucoid glycopeptide have .sup.1 H NMR chemical shifts of 5.020 ppm (J.sub.1,2 =9.56 Hz) and 2.000 ppm, respectively, and the two beta-protons of asparagine have .sup.1 H NMR chemical shifts of 2.766 ppm and 2.893 ppm. Other .sup.1 H NMR signals observed in FIG. 6 are for the anomeric protons of mannose-4 (5.058 ppm) and mannose-4' (5.018, 5.0097, 4.908 ppm), for the .epsilon.-methylene protons of lysine (3.002 ppm, J=7.32 Hz), for the .gamma.-methylene protons of glutamic acid (2.204 ppm, J=7.98 Hz), and for the N-acetyl groups (2.081-2.0229 ppm) of the turkey ovomucoid glycopeptide. During the hydrolysis reaction, the N-acetyl signal at 2.000 ppm, the beta-proton signals of asparagine at 2.766 and 2.893 ppm, and the anomeric proton signal at 5.020 ppm decrease simultaneously as the glycopeptide is hydrolyzed by the enzyme, while at the same time signals at 2.037 ppm as well as at 2.620 and 2.730 ppm appear. A new .sup.1 H NMR signal also appears at 5.188 ppm, but its appearance is delayed. Based on the model sugar compound, the NMR signals at 2.037 ppm and at 5.188 ppm have been assigned to the N-acetyl group and to the anomeric proton (alpha-anomer), respectively, of the N-acetylglucosamine-oligosaccharide in the free carbohydrate moiety. The NMR signals at 2.620 and 2.730 ppm are assigned to the two beta-protons of aspartic acid in the polypeptide chain. The eight .sup.1 H NMR signals at 2.000 ppm, 2.766 ppm, 2.893 ppm, 5.020 ppm, 2.037 ppm, 2.620 ppm, 2.730 ppm, and 5.188 ppm were all used to follow the course of the reaction.

The delayed appearance of the .sup.1 H NMR signals at 5.188 ppm was consistent with the formation of a 1-aminooligosaccharide intermediate in the hydrolysis reaction. A 1-aminooligosaccharide intermediate would be expected to have distinct .sup.1 H NMR signals for the anomeric proton and for the N-acetyl group of the sugar bearing the 1-amino group. However, no .sup.1 H NMR signal for the anomeric proton of a 1-aminooligosaccharide intermediate could be detected and the .sup.1 H NMR signals for the N-acetyl groups did not change following the appearance of the signal at 2.037 ppm. The latter observation was consistent with the formation of a 1-aminooligosaccharide intermediate, providing that the chemical shifts for the N-acetyl group in the intermediate and in the N-acetylglucosamine-oligosaccharide were nearly identical. To test these possibilities, we utilized 1-amino-1-dideoxy-N-acetylglucosamine as a model compound for the hydrolysis of the probable 1-aminooligosaccharide intermediate formed in the apparent two-step enzyme-catalyzed hydrolysis reaction.

The rate of hydrolysis for the model compound was readily measured using .sup.1 H NMR. Upon hydrolysis, the .sup.1 H NMR signal for the anomeric proton of the model compound shifts downfield from 4.189 ppm (J.sub.1,2 =9.35 Hz) to 5.195 ppm (J.sub.1,2 =3.38 Hz) and 4.709 ppm (J.sub.1,2 =3.38 Hz), the former the alpha-anomer and the latter the beta-anomer of the product, N-acetylglucosamine, while the chemical shift of the N-acetyl group is unchanged. The pseudo-first order rate constants for the enzyme-catalyzed reaction and for the hydrolysis of the model compound are given in Table I; based on least squares standard error analysis, the precision in the pseudo-first order rate constant is .+-.5%. These limited data give energies of activation for the enzyme-catalyzed reaction of 9.5 kcal mol.sup.-3 and for the model compound of 17.7 kcal mol.sup.1.

A solution of 1 mg of turkey ovomucoid glypeptide in 0.4 ml of 50 mM phosphate, p.sup.2 H 7.0 buffer in 99.96% .sup.2 H.sub.2 O was equilibrated at 37.degree. C. in a water bath. Enzyme (1.2 mg in 0.1 ml of buffer) was added and the solution was incubated at 37.degree. C. The hydrolysis reaction was checked by .sup.1 H NMR. At the end of the incubation, acetic anhydride (5-10 .mu.l) was added to trap the intermediate as the 1-acetamido-N-acetylglucosamine-oligosaccharide and to quench the enzyme activity. The solution was applied to a Bio-Gel P-4 column (104.times.3 cm) equilibrated with double-distilled, deionized water at room temperature, and 3 ml fractions were collected and monitored for the presence of polypeptide by measuring A.sub.280 nm and for carbohydrate by the phenol-sulfuric acid method. The carbohydrate-containing fractions were lyophilized and equilibrated with deuterium oxide: 3 times in 99.75% .sup.2 H, 2 times in 99.96% .sup. 2 H, and 2 times in 99.996% .sup.2 H, for 24 h each followed by lyophilization. The .sup.1 H NMR of the carbohydrate was recorded in 99.996% .sup.2 H.sub.2 O. A standard reference compound was prepared by reacting acetic anhydride with 1-amino-1-dideoxy-N-acetylglucosamine in 99.996% .sup.2 H.sub.2 O to give 1,2-diacetamido-1,2-dideoxy-beta-D-glucopyranose. The .sup.1 H NMR signal for the anomeric proton moved downfield from 4.189 ppm (J.sub.1,2 =9.35 Hz) to 5.053 ppm (J.sub.1,2 =9.33 Hz) upon acetylation of the model compound and the 2-acetamido group moved upfield from 2.047 ppm to a position where the two acetamido groups had chemical shifts of 2.011 and 2.009 ppm; the coupling constant for the anomeric proton indicates that, as expected, the 1-amino model compound retained the beta-conformation upon acetylation.

The anomeric protons for mannose-4 (5.058 ppm; J.sub.1,2 =9.56 Hz), and mannose-4' (5.018, 5.007, 4.908 ppm) in turkey ovomucoid glycopeptide are shown in FIG. 7A. When the glycopeptide was incubated at p.sup.2 H 7.0 and 37.degree. C. with almond amidohydrolase and the reaction was quenched with acetic anhydride after 1-h (FIG. 7B), 2-h, and 4-h (FIG. 7C) incubation times, the .sup.1 H NMR signal of the anomeric proton for the reducing end N-acetylglucosamine-oligosaccharide of the isolated carbohydrate moiety was shifted downfield to 5.045 ppm (J.sub.1,2 =9.81 Hz); the signal partially overlaps the .sup.1 H NMR signal of the anomeric proton for mannose-4. The .sup.1 H NMR signal of the alpha-anomer of the N-acetylglucosamine reducing end in the final carbohydrate product appears at 5.188 ppm (J.sub.1,2 =2.27 Hz). Based on the model compound, the NMR signal at 5.045 ppm is the anomeric proton of acetylated 1-aminooligosaccharide in the beta-conformation-the postulated intermediate. The .sup.1 H NMR spectra were integrated and the amounts of the acetylated 1-aminooligosaccharide and the final carbohydrate product were calculated as a percentage (.+-.5%) of the total isolated product. The theoretical percentages of the products expected were calculated from the kinetic data using the differential equations (Equation 3) for the two-step reaction. After a 1-h reaction time, 82% of the product was trapped as the intermediate and 18% was present as the final carbohydrate product. Consistent with this, the predicted quantities were 3% glycopeptide (not hydrolyzed by enzyme), 82% intermediate, and 15% final product. After a 2-h incubation before trapping, the product was a mixture of 72% intermediate and 28% completely hydrolyzed carbohydrate (67% of the intermediate and 32% of the final product were expected on the basis of our kinetic model). Finally, after 4-h rection time before trapping, the intermediate represented 40% and the free carbohydrate was 60% of the total product (the predicted quantities were 43% and 57% respectively). The .sup.1 H NMR spectrum of the same region in the completely hydrolyzed and equilibrated carbohydrate moiety shows resonance signals of the anomeric protons for the alpha-anomer of the reducing and N-acetylglucosamine, the mannose-4, and the mannose 4' (FIG. 7D). The .sup.1 H NMR signal of the N-acetyl group of the N-acetylglucosamine-oligosaccharide attached to asparagine in the glycopeptide appears at 2.000 ppm (FIG. 8A). The .sup.1 H NMR spectrum of the N-acetyl groups of the acetylated carbohydrate moiety after 1-h incubation is shown in FIG. 8B). The two .sup.1 H NMR signals that appear at 2.000 and 2.002 ppm are those of the two N-acetyl groups at the reducing end of the sugar. The .sup.1 H NMR spectra of the N-acetyl groups recorded after 2-h and 4-h reaction times showed that the two signals at 2.000 and 2.002 ppm had decreased in magnitude, while a .sup.1 H NMR signal appeared at 2.037 ppm. Thus the .sup.1 H NMR data for the trapping experiment give a clear, direct indication that 1-aminooligosaccharide is formed as an intermediate in the almond amidohydrolase-catalyzed hydrolysis of turkey ovomucoid glycopeptide.

To a solution of 0.7 mg of TOG in 50 mM phosphate, p.sup.2 H 8.0 buffer in 99.96% .sup.2 H.sub.2 O equilibrated at 37.degree. was added 2 units of N-Glycanase.TM. (Genzyme, Boston, Mass.) to give a final reaction volume of 58 ul. The reaction was followed by H-NMR as described for the almond emulsion enzyme reaction.

The hydrolysis of the TOG was followed by disappearance of the H-NMR signal for the N-acetyl group (2.00 ppm) and the anomeric proton (5.058 ppm) of the N-acetylglucosamine group linked to the asparagine residue. The production of the released 1-amino-1-deoxyoligosaccharides was followed by monitoring the appearance of the acetyl signal at 2.037 ppm. Under the conditions of the reaction, the hydrolysis of TOG was approximately 80% complete after 36 h (FIG. 9). The reaction can be accelerated by the addition of greater amounts of enzyme.

Unlike the reaction performed at pH 7.0 in Example 1, the released beta-1-amino-1-deoxyoligosaccharide is stable at pH 8.0. Thus even after 36 h there is no appreciable H-NMR signal for the alpha-anomer (5.19 ppm) of hydrolyzed aminodeoxysugar.

As in example 1, the 1-amino-1-deoxyoligosaccharide was trapped by the addition of acetic anhydride.

The practicality of trapping the glycosylamines with dansyl chloride was investigated with the model compound 2-acetamido-1-amino-1,2-dideoxyglucose. The aminodeoxysugar (44 mg) in 2 ml of 0.2M NaHCO.sub.3 was mixed with 1.1 ml of acetone containing dansyl chloride (60 mg). The reaction mixture was heated with stirring at 70.degree. for 2 h. At the completion of the reaction a homogeneous solution was obtained, with only a trace of the original yellow color remaining. Analysis of the reaction mixture by silica gel thin-layer chromatography (methylene chloride/methanol, 5/1 by vol) indicated the presence of the following reaction components: dansyl chloride (rf 0.99), dimethylaminonaphthalene sulfonamide (rf 0.95), dimethylaminonaphthalene sulfonic acid (rf 0.34) a trace of 2-acetamido-1-amino-1,2-dideoxy-glucose (rf 0.06), and a new product (rf 0.55). The reaction product corresponding to the spot at rf 0.55 was purified by silica gel chromatography (2.times.15 cm column; eluted with methylene chloride/methanol, 6/1 by vol.). A portion of the desired product was obtained free of contaminants (the elution volume of this product was 100-110 ml). After concentration under reduced pressure, 12 mg of a yellow glass was obtained. H-NMR analysis confirmed that the material was the expected 2-acetamido-1,2-dideoxy-1-dimethylaminonaphthalenesulfonamide-B-D-glucoses : 1.50 (s, 3H, CH.sub.3 --C--NH), 2.85 (s, 6H, (CH.sub.3).sub.2 --N), 3.10-3.90 (m, 6H, pyranose ring protons), 4.55, (d,1H, anomeric, J=9.4H), 7.27-7.59 (m, 3H, aromatic) 8.25-8.54 (m, 3H, aromatic). There appeared to be a small amount (<5%) of the alpha-anomer (.delta. 5.10) present in the isolated product. The UV spectrum of the product gave two absorbance peaks (EtOH) with .lambda.max at 251 nm and 336 nm.