Method for coupling enzymatically activated glycoconjugates to a modifying compound

The present invention relates to a method for coupling of an enzymatically activated glycoconjugate to a modifying compound as well as a glycoconjugate coupled to a modifying compound obtainable by said method. Furthermore, the present invention is directed to the use of an oxidase enzyme in said method and also to pharmaceutical compositions comprising a coupled glycoconjugate obtained by said method.

Background of the invention

There is a large number of therapeutically interesting proteins with pharmacokinetic drawbacks such as instability in solution, poor bioavailability, short half-life and rapid degradation due to in vivo immunoreactivity. Quite often conjugation of proteins for in vivo applications with water-soluble polymers such polyethylene glycol (PEG) or starch derivatives may improve the pharmacokinetic profile of a protein drug.

The conjugation of proteins typically involves reactive groups derived from the side chains of sterically available amino acids. Some of the most common amino acid targets for this purpose are lysine, cysteine and arginine. Because a protein usually comprises more than one of the target amino acids, the probability is high that conjugation will take at different position(s) on the protein surface. This can lead to considerable heterogeneity of the coupling products.

At least one of the reaction partners, i.e. protein or water-soluble polymer, requires prior activation. However, currently available activation methods are not suited to distinguish among several different amino acids of the same species (for example lysines) on the protein's surface. As a result, a complex mixture of positional isomers of the conjugation product is generally obtained. The separation of these isomers is a

very complex and work-intensive task. In some instances the necessary isolation cannot be accomplished with standard chromatography techniques.

Often the coupling of protein and polymer is accompanied by a reduction or even a loss of the protein's activity. This phenomenon is mostly due to the random conjugation reactions discussed before, which with a certain probability will result in a modification in or near the active site of the protein and/or the binding site of antibodies or receptors. Furthermore, even a covalent coupling with a polymer distant to the site of activity or binding site may result in a modified protein ^' s tertiary structure which can be expected to reduce or abrogate bioactivity.

For glycoconjugates, particularly proteins with carbohydrate chains on their surface, i.e. glycoproteins, a different approach for coupling polymers is available. Because most glycoproteins only have a limited number of glycosylation sites, in some cases only one site, coupling via carbohydrate chains may provide a higher level of coupling specificity than coupling to the more abundant amino acids. A further advantage of this strategy is that coupling takes place at some distance from the protein because the carbohydrate chain of the glycoprotein acts as a natural spacer. As a consequence, the coupling to carbohydrate moieties typically results in no or minor modifications for the tertiary protein structure and yields a more homogeneous coupling product than amino acid targeted coupled polymers.

But even this approach has some inherent drawbacks. The carbohydrate residues generally require a previous activation step. The activation step is typically an oxidation of vicinal diols to dialdehydes with periodate and interferes with a number of amino acids residues such as, for instance, methionine. Most often, this unselective oxidation method will create more than one activation site and, therefore, more than one coupling site. Consequently, a heterogeneous coupling product is often encountered. Furthermore, these strong oxidizing conditions may partially modify the protein's chemistry (e.g., by oxidation of Met). Therefore, chemical oxidation can lead to loss of activity.

The enzyme oxygen 6-oxidoreductase (E.G. 1.1.3.9) isolated from Dactylium dendroides, also known as galactose oxidase (GAO), is a copper enzyme catalyzing the following reaction:

GAO has been used in a number of diagnostic and biosensor applications (e.g., Vega FA et al., Anal. Chim. Acta, 373, p.57-62 (1998); Tkac J. et al., Biotechnol. Tech., 13, p. 931-936 (1999); Szabo EE et al., Biosens. Bioelectron., H, p. 1051- 1058 (1996)). It was also used to introduce a label into polysaccharides on cell surfaces (Calderhead DM et al., J. Biol. Chem., 263, p. 12171-12174 (1988)) as well as for the detection of saccharide tumor markers in colon cancer (Said IT et al., Histol. Histopathol., 14, p. 351-357 (1999)). Furthermore, GAO has also been used to modify galactose-containing polysaccharides in order to reduce molecular weight and/or viscosity (e.g., US2002/0076769A1 or WO01/62938).

In view of the above the object of the present invention is to provide an improved method for coupling glycoproteins to modifying compounds such as polymers, preferably water soluble polymers such as PEG or starch derivatives. It is another object of the invention to achieve an improved selectivity for carbohydrate-based coupling of glycoconjugates and a further object to provide a mild and selective activation method so that the conjugation reaction does not interfere with the glycoprotein's original protein structure and the biological activity of the original glycoprotein is maintained in the coupling product.

Summary of the invention

The object of the invention is solved by a method for coupling an enzymatically activated glycoconjugate to a modifying compound, comprising the steps of:

a) activating by enzymatic oxidation of at least one primary and/or secondary hydroxyl group of at least one oligo- or polysaccharide moiety in a glycoconjugate to an aldehyde or ketone group, and

b) reacting the modifying compound with said aldehyde and/or ketone group.

The activation by enzymatic oxidation of the primary and/or secondary hydroxyl groups of the oligo- or polysaccharide moiety of a glycoprotein in step a) according to the invention is accompanied by mild reaction conditions and high selectivity. The non-saccharide part of the glycoconjugate, its structure and its function typically remains unchanged due to the mild reaction conditions and the selectivity of the respective enzyme.

Herein, the term "glycoconjugate" is defined as any substance having an oligo- or polysaccharide moiety with primary and/or secondary hydroxyl groups. Said substance can be a protein, an oligo- or polypeptide or a lipid.

Preferably, said glycoconjugate is a glycoprotein, more preferably a biologically active glycoprotein, most preferred a therapeutically active glycoprotein. It is also preferred that the glycoprotein is a human or human-derived glycoprotein.

In a preferred embodiment the oligo- or polysaccharide moiety of the glycoconjugate comprises at least one saccharide selected from the group comprising of N-acetyl glucosamine, N-acetyl galactosamine, mannose, galactose, sialic acids and fucose.

In further preferred embodiment at least one primary hydroxyl group of at least one oligo- or polysaccharide moiety in a glycoconjugate is enzymatically oxidized to an aldehyde group.

Preferably, the enzymatic oxidation (activation) of the glycoconjugate is performed by a saccharide-specific oxidase. Saccharide-specific oxidases require mild, often physiological conditions and demonstrate high selectivity for the saccharide. For saccharide-specific oxidases the saccharide specificity as well as the saccharide-

specific content of the oligo- or polysaccharide of the glycoconjugate will determine the number of aldehyde groups in the activated glycoconjugate.

More preferably the saccharide-specific oxidase is a galactose-specific oxidase. Among the galactose-specific oxidases, 6-oxidoreductases are particularly preferred. A galactose oxidase (GAO) was originally isolated from Dactylium dendroides by Avigad et al. (J. Biol. Chem, 237, p. 2736-2743 (1962)). In a preferred embodiment, the 6-oxidoreductase for use in the invention is derived from Dactylum dendroides.

GAO is also available as a recombinant protein from E. coli (Sun et al., Protein Eng., 14. P- 699-704 (2001)), Pichia pastoris (Whittaker et al., Protein Expr. Purif., 20, p. 105-111 (2000)), mammalian cells (Maffia et al., WO01/62938) and others such as Fusarium species.

Because of their higher purity the recombinant GAO enzymes are preferred for use in the method according to the invention. More preferably, the 6-oxidoreductase employed is a recombinant protein, most preferably one produced by E. coli, Pichia pastoris or Fusarium species. Alternative preferred GAOs may be produced by other yeasts, such as Saccharomyces cerevisiae or Hansenula polymorpha, by mammalian or by insect cells. In a most preferred embodiment, the galactose- specific oxidase is galactose oxidase EC 1.1.3.9 or another oxygen 6- oxidoreductase.

For practical reasons, it is preferred to use the enzyme in an immobilized form. Most convenient is the use of enzyme attached to a insoluble matrix, such as a bead, a sheet or a membrane. The insoluble matrix material may be glass, a polymer or any other inert insoluble matrix material available to the skilled person. The enzyme should be firmly attached to the insoluble matrix, e.g. by covalent bonds. Methods for attaching protein to commercially available matrices are known to the skilled person.

GAO has quite a broad specificity for primary hydroxyl groups, but oxidation in the presence of C6-OH groups in galactose is much preferred over other primary OH- groups. The wild type enzyme is a glycoprotein of about 68 kD consisting of 639 amino acids with a carbohydrate content of about 1.7% by weight. It has a single

polypeptide chain with a type 2 copper centre as prosthetic group. The catalysed reaction is the following:

(Scheme 1)

A tyrosin is located in the active site of GAO, which is involved in the redox mechanism as tyrosyl radical / tyrosyl anion. In combination with a copper ion a two- electron transfer is accomplished:

Cu(l)/Tyr ^" ===^ Cu(ll)/Tyr ^" + 2 e ^" red

(Scheme 2)

The electrons released during the oxidation reaction are transferred to oxygen. The Cu(ll)/Tyr ^" system is regenerated to the Cu(l)/Tyr system by the oxidation of the primary OH-group at the C6 of the galactose to the aldehyde.

An important feature for substrate recognition of GAO is the configuration at the C4 of the pyranose. Besides D-galactose, D-gulose and D-talose may also act as substrates, whereas D-glucose or D-mannose are not recognized by this enzyme. This is also true for derivatives of D-galactose with substituents in the C4 position, for which a loss of activity results. Substituents in the C2 position are much better tolerated, although with a reduced reaction rate. Some galactosides are more rapidly oxidized than galactose itself, especially oligo- and polysaccharides of the guaran type. The intraglycosidic bond between a terminal galactose and a subterminal glycosyl unit may be β(1->6), α(1->6) or α(1->4). ^β (1->4) linkages have a much lower affinity for the enzyme.

Galactose oxidase can be completely inhibited by hydroxylamine, hydrazine and cyanide. The enzyme is also strongly inhibited by its product hydrogen peroxide above a concentration of 2 mM. At 10 mM the inhibition is complete. GAO is quite stable at room temperature but rapidly looses activity at higher temperatures or pH values above 8.0.

In a preferred embodiment the method of the invention further comprises the step of degrading any produced hydrogen peroxide.

More preferably, the hydrogen peroxide is degraded by catalase and/or peroxidase (POD).

The preferred addition of catalase to the oxidation reaction in step a) of the method of the present invention degrades the hydrogen peroxide formed by the GAO reaction into oxygen and water. A considerable portion of the GAO molecules will remain in a semi-active state after the catalysis reaction, a phenomenon which is not well understood despite the availability of 3D structures of the enzyme. It has been suggested, that the use of "one-electron-oxidizers" would be able to convert the enzyme into an active state again. Examples of such oxidizing agents are Fe- cyanide, [Co(Phen) ₃ ] ^3" and others.

For reactivating the semiactive GAOs after the catalysis reaction it is preferred to employ the enzyme peroxidase (POD), more preferably peroxidase from horse radish, as a biological oxidizer.

In a more preferred embodiment the method of the invention is one wherein the hydrogen peroxide is degraded by catalase and peroxidase (POD) so that the hydrogen peroxide is degraded by catalase and at the same time GOA is reactivated by POD.

The complete scheme for the above three enzyme reaction for the oxidation of galactose is illustrated below.

POD

Catalase H ₂ O ₂ *. 2 H ₂ O + O ₂

(Scheme 3)

In another preferred embodiment of the method of the invention step a) involves at least the three enzymes oxygen 6-oxidoreductase, preferably GAO, catalase and peroxidase (POD).

The reaction conditions of the above complex system can be optimized by employing raffinose as a substrate. A yield of 80% aldehyde groups was achieved at room temperature with a reaction time of 45 hrs in a reaction mix with 50 mM potassium phosphate buffer, pH 6, 0.1 mM CuCI ₂ , 10 mM raffinose and 0.1 U GAO, 2 U catalase and 0.1 U peroxidase per 1 μmol substrate, respectively. It is important to provide sufficient oxygen by bubbling O ₂ through the reaction mix during the incubation.

Hence, it is preferred that step a) of the method of the invention is performed at a temperature in the range of 0 to 50 ⁰ C for a time period of 30 min to 120 hours in the presence of sufficient oxygen in the reaction mixture.

More preferred, step a) is performed at a temperature in the range of 15 to 40 ⁰ C for a time period of 2 to 30 hours.

In an exemplary optimized three-enzyme model system for oxidizing the glycoprotein fetuin the yield calculated as mol aldehyde per mol protein was below 10 %, confirming that galactose residues subterminal to sialic acids are poor substrates for

the GAO. Because fetuin contains up to three sialic acid residues per glycosylation site, it should be desialylated to become an improved substrate for the galactose oxidase. Desialylation with sialidase according to standard methods is described in the literature (e.g., Schauer: Sialic Acids: Chemistry, Metabolism and Function: Cell Biology Monographs (1983).

Asialofetuin isolated after sialidase treatment and oxidized with GAO / catalase / POD under optimized conditions will yield about 50 % oxidation of the available galactose units.

In a preferred embodiment, non-sialylated or desialylated glycoproteins are employed for the method of the present invention.

In order to facilitate the purification of oxidized glycoproteins the oxidizing enzymes can be employed in an immobilized form. This is highly advantageous in that it enables one to reuse the enzymes for several biotransformations due to the higher stability of the enzymes in the immobilized form. Each enzyme can be immobilized separately, so that different ratios of the individual enzymes can be used if desired by mixing the appropriate amounts of immobilized enzymes. As a suitable immobilization technique many of the methods known in the art are suitable (e.g.: Chibata, I: Immobilized Enzymes (Kodansha Scientific Books, 1979). N-hydroxy succinimide activation or azlacton coupling to agarose, glass or polymer beads are preferred techniques of immobilization of enzymes for oxidation. The immobilized enzymes can then easily be removed from reaction mixture by filtration.

In a preferred embodiment of the present invention, the oxidizing enzyme is immobilized on a solid insoluble matrix in order to simplify the purification of the substrate, e.g by filtration. Preferred ratios of the three preferred enzymes 6- oxidoreductase, preferably GAO, catalase and peroxidase are in the range of 1-10 : 10-50 : 1-10 for 6-oxidoreductase : catalase : POD, more preferred in the range of 1- 2 : 5-20 : 2-4. Further preferred ratios are in the range of 1 : 0.1 to 100 for GAO : catalase and 1 : 0.5 to 50 for GAO : peroxidase, preferably horseradish peroxidase. Alternatively, lactoperoxidase, myeloperoxidase, glutathion peroxidase or manganese peroxidase may be employed.

As used herein, the term "modifying compound" means any compound that will modify the biological properties of the original glycoconjugate when coupled according to the invention. For example, said modification may influence the glycoconjugate 's own biological activity, its stability in solution and/or pharmacokinetic behaviour in vivo, such as, e.g. bioavailability, half-life and rapid degradation due to immune reactivity. It is preferred that the modifying compound does not merely modify physical properties of the glycoconjugate, e.g. change the molecular weight or provide a label for detection, but actually modifies the biological activity of the glycoconjugate in at least one way.

In a preferred embodiment, the nucleophilic group of the modifying compound is selected from amine group, a hydroxyl group, a thiol group, hydrazide and a guanidino group.

In a more preferred embodiment, there is a group X covalently attached to the modifying compound, wherein X is a bifunctional linker comprising a nucleophilic group or a combination of such bifunctional linkers. Again, it is preferred to select the nucleophilic group from an amine group, a hydroxyl group, a thiol group, a guanidine group and hydrazide.

Suitable bifunctional linkers are well known in the art and can be found, for example, in the catalog of the Pierce company, Rockford, IL, USA (Pierce 2005-2006 Applications Handbook & Catalog at www.piercenet.com).

Also preferred is that the linker is a sulfhydryl-reactive linker, preferably N-[β- maleimidopropionic acid]hydrazide ^» TFA (MPH), 4-(4-N-maleimidophenyl)butyric acid hydrazide (MPBH), N-[e-maleimidocaproic acid] hydrazide (EMCH), 4-(N-maleimido- methyl)cyclohexan-1-carboxylhydrazid HCI (M ₂ C ₂ H), 3-(2-pyridyldithio) propionyl- hydrazid (PDPH) or N-[k-maleimidoundecanoic acid]-hydrazide (KMUH).

In one embodiment of the invention the linker connects an SH- and an NH-group. Linkers connecting an SH- and an NH-group are e.g. linkers selected from AMAS (N- α(maleimidoacetoxy)succinimidester), BMPS (N- ^β (maleimidopropyloxy)succinimid-

ester), GMBS (N-γ(maleimidobutyτyιoxy) succinimidester), EMCS (N-ε(maleimido- caproyloxy)succinimidester), MBS (m-(maleimidobenzoyl)-N-hydroxysuccinimid- ester), SMCC (succinimidyl-^N-maleimidomethyl) cyclohexan-1-carboxylat), SMPB (succinimidyl-4-(p-maleimidophenyl) butyrat), SPDP (succinimidyl-S-^-pyridyldithio) proprionat), Sulfo-EMCS (N-ε(maleimidocaproyloxy)sulfosuccinimidester) and Sulfo- GMBS (N-γ(maleimidobutyryloxy)sulfosuccinimidester).

In another embodiment of this invention, linkers are used which connect two SH- groups. Examples for linkers having that ability are selected from BMB (1.4-bis- maleimidobutan), BMDB (1.4-bis-maleimido-2.3-dihydroxybutan), BMH (bis-maleimi- dohexan), BMOE (bis-maleimidoethan), DTME (dithio-bis-maleimidoethan), HBVS (1.6-hexan-bis-vinylsulfon), BM(PEO) ₃ (1.8-bis-maleimidotriethylenglycol) and BM (PEO) ₄ (1.11-bis-maleimidotetraethylenglycol).

Furthermore, according to the present invention linkers may be used, which connect an NH- and an NH-group. Examples for such linkers are selected from BSOCOES (bis-(2-(succinimidyloxycarbonyloxy) ethyl) sulfon, BS ³ (bis-(sulfosuccinimidyl) suberat), DFDNB (1.5-difluor-2.4-dinitrobenzol), DMA (dimethyladipimidat HCI)), DSG (disuccinimidylglutarat), DSS (disuccinimidylsuberat) and EGS (ethylenglycol-bis- (succinimidylsuccinat).

In still another embodiment of the present invention, the linker transforms an amino group into a sulfhydryl group and is preferably 2-iminothiolane-HCI (Traut ^' s reagent).

Finally, it is another preferred embodiment that the modifying compound is bound to the linker by a carbonyl group or a carboxyl group.

Alternatively, a group of the formula -Y-NH ₂ is covalently attached to the modifying compound, wherein Y is a bifunctional linker selected from the bifunctional linkers recited in any of claims 16 to 25, In yet another embodiment, a group of the formula -Y-NH-NH ₂ is covalently attached to the modifying compound, wherein Y is a bifunctional linker selected from the bifunctional linkers recited in any of claims 16 to 25, and in still another embodiment a group of the formula -Y-CO-NH-NH ₂ is

covalently attached to the modifying compound, wherein Y is a bifunctional linker selected from the bifunctional linkers recited in any of claims 16 to 25.

Preferably, the modifying compound is a polymer, more preferably a biocompatible polymer, most preferably a water-soluble biocompatible polymer. The modifying compound preferably carries a nucleophilic group.

It is preferred that the polymer is selected from the group consisting of starch, gelatine, dextran and albumin.

More preferably, the polymer is hydroxyalkyl starch (HAS), an amino-functionalized hydroxyalkyl starch or a sulfhydryl functionalized hydroxyalkyl starch being most preferred.

A very preferred polymer for practicing this invention is a semi-synthetic polysaccharide hydroxyalkyl starch. Particularly preferred among hydroxyalkyl starches is hydroxyethyl starch. Due to the natural raw starting material, amylopectin, and the production process during which a certain extent of cleavage of the polymer chains is required, the hydroxyalkyl starch is produced as a molecular homogeneous substance with defined molecular weight but comprises a mixture of molecules of different sizes which may also be differently substituted by hydroxyalkyl groups. The characterization of such mixtures requires statistical measures. For the weight average molecular weight the mean molecular weight (Mw) is called upon. This value is generally determined by means of a light-scattering detector that provides the average molecular weight of all components of a polydisperse mixture.

The average value (Mn) indicates the number average molecular weight currently used for calculating the real number of moles in a specimen. This value is measured by its effect on osmotic pressure by means of a membrane osmometer.

There are two different substitution degrees available for defining the substitution of HAS by hydroxyalkyl groups. The substitution degree (MS, molar substitution) is defined as the average number of hydroxyethyl groups per anhydroglucose unit. It is

determined from the total number of hydroxyethyl groups in a specimen by ether cleavage and subsequent quantitative determination of ethyl iodide and ethylene.

The substitution degree DS (degree of substitution) is defined as the proportion of the substituted anhydroglucose units of all anhydroglucose units. It can be determined from the measured amount of the unsubstituted glucose after hydrolysis of a specimen. It follows from these definitions that the MS is normally greater than the DS. In the case where only a monosubstitution is present, each substituted anhydroglucose unit carries only one hydroxyethyl group and MS equals DS.

It is well known in the field of hydroxyalkyl research that besides molecular weight and the degree of substitution the C2/C6 ratio is also of great importance. Hydroxyalkyl groups can be attached to the -OH in C2, C3 and C6 of the glucose moieties. The action of α-amylase is the major mechanism of in vivo breakdown of hydroxyalkyl starches. This enzyme preferably cleaves unsubstituted anhydroglucose units, whereas substituted anhydroglucose units are split much slower. Additionally, hydrolysis by α-amylase is more retarded by substitution on C2-OH than on C6-OH, because the C2 position is much closer to the actual cleavage site at C1. This means that with an increase in the degree of substitution and an increase in the C2/C6 ratio a greater delay in breakdown and elimination of hydroxyalkyl starch occurs. In this respect, only hydroxyalkyl starches having a high C2/C6 ratio or being highly substituted are useful for pharmaceutical purposes.

The above mentioned factors for hydroxyalkyl starch play an important role for the determination of its vascular persistence and its fate in the body after administration.

In principle, distribution, degradation and elimination are determined by the above physicochemical properties (MW, MS, DS and C2/C6 ratio). The main mechanisms governing HAS-pharmacokinetic behaviour are the renal clearance (related to the polymer size) and the degradation rate by α-amylase (in relation with the extent of substitution).

After intravascular degradation, HAS is almost exclusively eliminated by the kidney. As soon as the size-limit is reached by enzymatic degradation (60 to 7OkD) the glomerular filtration removes the starch polymer from blood stream. Also, to a certain

extent free glucose may result which will be metabolised in normal biological pathways. Gastrointestinal elimination does not play any role in the elimination process for hydroxyalkyl starches.

Preferably, the polymer for practicing the method of the invention is hydroxyalkyl starch (HAS) with a molecular weight in the range of 1 kD to 500 kD.

More preferably, the hydroxyalkyl starch (HAS) has a molecular weight in the range of 5 kD to 200 kD.

It is further preferred that the molar substitution of hydroxyalkyl starch for use according to the invention is in the range of 0.1 to 0.9, more preferably in the range of 0.2 to 0.6.

It is also preferred that the C2/C6 ratio of hydroxyalkyl starches for use in the present invention is in the range of 1 to 10.

It should be noted that the most preferred hydroxyalkyl starch (HAS) is hydroxyethyl starch (HES). In the following paragraphs and in the examples the invention is further explained with reference to HES without being meant to be limited to this most preferred embodiment.

HES has been used for 15 years in volume substitution therapy as so called plasma expander. The broad clinical experience with this polymer has shown an extra- ordinary high biocompatibility with respect to toxicity, antigenicity, immunogenicity and rate of clinical side effects. Moreover, in contrast to PEG hydroxyalkyl starch is a biodegradable polymer with a very well studied pharmacokinetic profile, which lead to predictable degradation rates according to the structural parameters MW, MS, DS and C2/C6 ratio as discussed above.

HES is commercially produced by hydroxyethylation of starch using ethylene oxide or 2-chloroethanol as described, for example, by Sommermeyer et al. (EP0402724). HES has been commonly employed for the production of colloidal plasma volume expanders. The field of volume substitution (e.g. in case of hemorrhagic shock) or

hemodilution (e.g. for arterial occlusive disease, Fontaine Il B, III) is today inconceivable without the use of colloidal plasma substitutes. HES is highly accepted in clinical applications because of its very low antigenicity, immunogenicity and toxicity and its well known predictable pharmacokinetic behaviour.

These favourable properties make hydroxyethyl starch the polymer of choice for protein conjugation. Consequently, different methods for coupling HES to proteins or peptides were already described in the art and develop "HESylation" technology as an alternative to "PEGylation" (see Hemberger & Orlando: WO03/074087). Exploiting the high water solubility of this polymer conjugates with low-soluble drugs will increase the solubility by a factor of several 1000 in most cases (Orlando & Hemberger. WO03/074088).

For HES coupled to glycocoηjugates or glycoproteins via the enzymatically activated saccharide moiety according to the invention there are a number of advantages over previous approaches. Due to the often limited number of specific glycosylation sites in a glycoconjugate (preferably only one glycosylation site) the selectivity of the coupling site on the protein is greatly enhanced and predictable. The method of the invention also provides a distance between protein and polymer (i.e. the oligosaccharid acting as a "natural spacer"). This will minimize the impact of the polymer on the tertiary structure of any glycoconjugate and therefore result in a high retention of the original biological activity of the glycoprotein in the HES-coupled product.

Biologically active glycoprotein-HES conjugates are very valuable as therapeutic protein drugs, because of their increased stability, longer life span in vivo and reduced antigenic or immunogenic potential. Conjugates produced according to the methods of the invention may also be used as diagnostic tools, for example, when antibodies or antibody fragments are used as the glycoprotein part.

In a preferred method for functionalizing HAS, preferably HES, the starch derivative may be selectively oxidized at the reducing end aldehyde to the corresponding acid by methods already described by Hemberger & Orlando in WO03/074087 and WO03/074088. This acid may be readily converted into the corresponding lacton by

dehydration. To react with the aldehyde function of the activated glycoprotein, the HES lacton has to be further functionalized.

This can be done by reacting the HES lacton with hydrazine or diaminobutane in a water-free organic solvent, preferably DMSO or DMF. The resulting HES hydrazide may then be coupled to the enzymatically activated glycoprotein to form a stable hydrazone bond. Alternatively, monosulfhydryl-HAS may be used, which has advantages in purification in comparison to the hydrazine derivative. In this case the HES-SH may be reacted with 4-maleimidopropionic acid hydrazide (MPH) or 4-(4-N- maleimidophenyl)butyric acid hydrazide (MPBH) to yield hydrazide derivatives which are coupled via the above mentioned linkers to the SH group of the HES-SH. By reacting these hydrazides with an activated glycoprotein a stable hydrazone bond in the same manner as described above may be obtained between the selectively functionalized HES and the enzymatically oxidized carbohydrate part of the glycoprotein.

To obtain a hydrazide-functionalized HAS molecule one may start from the selectively oxidized HAS as described in WO03/074087 or the corresponding lacton. After deprotection the reaction with N-BOC-hydrazin results in the derivative HAS- CO-NHNH ₂ without a linker molecule. Alternatively, a bifunctional linker can be introduced very easily by standard methods if desired. Preferably, these linkers have biocompatible, non-toxic, non-antigenic and non-immunogenic properties.

In another preferred embodiment an SH-functionalized HAS derivative will provide for higher reactivity and higher purity. Its synthesis may be accomplished via cysteamine coupling as described in the example section below. The HAS-SH is then reacted with a suitable SH-reactive linker to yield a hydrazide of the general formula HAS-X- CO-NHNH ₂ , wherein X is the linker.

For the final coupling reaction in step b) the pre-treated conjugation partners (activated glycoconjugate and nucleophilic functionalized modifying compound) may be dissolved together in a suitable buffer system, e.g. 0.1 M sodium acetate solution (pH ~ 5.5) and be maintained under moderate stirring conditions for 20 to 24 hrs at 20°-25°C. Of course, the buffer conditions have to be adjusted according with the

protein target ^' s specific tolerance. The reaction product can be isolated and analyzed, for example, by gel permeation chromatography as well as by SDS-PAGE and subsequent protein- and glyco-specific staining.

In a further preferred embodiment, the glycoprotein for use in the method of the present invention is selected from the group consisting of growth factors, preferably epidermal growth factor (EGF), nerve growth factor (NGF), platelet-derived growth factor (PDGF), differentiation factors, preferably erythropoetin (EPO), granulocyte- colony stimulating factor (G-CSF), granulocyte/macrophage-colony stimulating factor (GM-CSF), chemokines and cytokines, preferably interferons, more preferably α- interferon, β-interferon, γ-interferon and their interferone subtypes, interleukins, more preferably interleukines 1 to 12, asialofetuin, and monoclonal antibodies as well as fragments thereof.

The above incomplete list of therapeutically active glycoproteins demonstrates that numerous therapeutic glycoproteins may benefit from enzymatic coupling method according to the present invention. Coupled with modifying compounds such as water-soluble polymers, e.g. PEG or starch, a higher stability, longer in vivo half life and/or lower toxicity, antigenicity and immunogenicity and consequently less side effects during the clinical application may be accomplished.

In step b) the enzymatically and selectively oxidized (activated) glycoconjugates are coupled to at least one modifying compound by reacting at least one nucleophilic moiety of the modifying compound with at least one of the aldehyde and/or ketone moieties of the activated glycoconjugate.

In a preferred embodiment the method of the invention further comprises the step of reducing the coupling bond resulting from step b), if further stabilization of the bond between the glycoconjugate and the modifying compound, e.g. HAS, is desired.

In yet another preferred embodiment of the invention the coupling between the aldehyde and/or ketone function of the enzymatically activated glycoconjugate and the nucleophilic modifying compound will form a Schiff ^' s base structure which may

be reduced to a stable amine by suitable reduction agents such as boron hydride or more preferred cyanoboron hydride.

As demonstrated more generally above and in the specific embodiments of the example section below, the method of the invention allows for the highly selective coupling of enzymatically activated aldehydes and/or ketones on glycoconjugates to nucleophilic modifying compounds in very high to moderate yields and retention of the biological activity of the glycoprotein target. The major advantage of the method of the invention is that unlike to the methods of the prior art a predictable, selective, stable covalent coupling with good recovery of the biological activity is achieved, which yields a homogenous coupling product with respect to a 1 :1 stochiometry and the specific coupling site.

Because of the new predictable, selective and stable covalent coupling method of the present invention resulting in excellent recovery of the original biological activity of the glycoconjugate, which also yields a homogenous coupling product with respect to a 1 :1 stochiometry and the specific coupling site, the method of the present invention will result in novel products which have not been available with the previous methods in the art. Therefore, in a further aspect, the present invention also provides new glycoconjugates coupled to a modifying compound, obtainable by a method according to the invention.

Preferably, the polymer of said glycoconjugate is hydroxyalkyl starch (HAS).

More preferably, the polymer is hydroxyalkyl starch (HAS) with a molecular weight in the range of 1 kD to 500 kD.

Further preferred is that the polymer is hydroxyalkyl starch (HAS) with a molecular weight in the range of 5 kD to 200 kD.

Also preferred is that the polymer is hydroxyalkyl starch (HAS) with a molar substitution in the range of 0.1 to 0.9, more preferably in the range of 0.2 to 0.6.

Furthermore, it is preferred that the C2/C6 ratio of HAS is in the range of 1 to 10.

Preferably, the new glycoconjugate according to the invention is coupled to a modifying compound by means of a bifunctional linker.

Also preferred is that the hydroxyalkyl starch is hydroxyethyl starch (HES).

More preferably, the hydroxyethyl starch (HES) is conjugated to a glycoprotein by means of a hydrazone bond.

In a most preferred embodiment, the glycoconjugate produced according to the invention comprises a glycoprotein selected from the group consisting of growth factors, preferably epidermal growth factor (EGF), nerve growth factor (NGF), platelet-derived growth factor (PDGF), differentiation factors, preferably erythropoetin (EPO), granulocyte-colony stimulating factor (G-CSF), granulocyte/macrophage- colony stimulating factor (GM-CSF), chemokines and cytokines, preferably interferons, more preferably α-interferon, β-interferon, γ-interferon and their interferone subtypes, interleukins, more preferably interleukines 1 to 12, asialofetuin, and monoclonal antibodies as well as fragments thereof.

A third aspect of the invention is directed to the use of an oxidase enzyme in a method according to the invention.

Last but not least, the present invention relates to a pharmaceutical composition comprising a coupled glycoconjugate produced according to the invention and optionally a pharmaceutically acceptable excipient.

The glycoconjugate in the pharmaceutical composition is preferably selected from the group consisting of growth factors, preferably epidermal growth factor (EGF), nerve growth factor (NGF), platelet-derived growth factor (PDGF), differentiation factors, preferably erythropoetin (EPO), granulocyte-colony stimulating factor (G-CSF), granulocyte/macrophage-colony stimulating factor (GM-CSF), chemokines and cytokines, preferably interferons, more preferably α-interferon, β-interferon, γ-

interferon and their interferone subtypes, interleukins, more preferably interleukines 1 to 12, asialofetuin and monoclonal antibodies as well as fragments thereof.

More preferably, the pharmaceutical composition of the invention comprises a glycoconjugate coupled to a hydroxyalkyl starch (HAS), preferably a hydroxyethyl starch (HES).

Figures

Fig. 1 illustrates the substrate specificity of GAO in the presence of catalase and POD. The yield is calculated from the estimation of the newly formed aldehyde groups by the BCA methods (for details see example 7).

Fig. 2 illustrates the absence of sialic acids in the desialylated fetuin by isoelectric focussing (for details see example 12)

Fig. 3 illustrates the oxidation of asialofetuin by GAO in the presence of catalase and POD. The yield is calculated from the estimation of the newly formed aldehyde groups by the BCA methods (for details see example 13)

Fig. 4 demonstrates the formation of new aldehyde groups by glycan detection (for details see example 13)

Fig. 5 shows the analysis of the coupling product of oxidized asialofetuin with HES- MPH in example 15 by SDS-PAGE. From the results of the SDS-PAGE a coupling yield of about 50% was estimated, i.e. about 50% of the oxidized asialofetuin remained unchanged after the reaction (see example 15 for details).

Fig. 6 shows the GPC analysis of the coupling reaction mixture from example 14.

The following examples are presented for illustrating preferred embodiments of the present invention only. They are not intended to be construed as limiting to the scope of the present invention.

Examples

Example 1: Synthesis of HES-hydrazide

Hydroxyethyl starch with a molecular weight of 70 kD and a degree of substitution of 0.5 (HES ₇₀ ) was used for this experiment. HES _7O was selectively oxidized as described in WO03/074087 to the corresponding carboxylic acid (c(- COOH) = 32 nmol/mg). OxHES ₇₀ in the lacton form was treated with a 2-fold molar excess of N-BOC-hydrazine in water-free DMSO under argon atmosphere for 24 hrs at 50 °C. The reaction product was precipitated by adding an ice-cold mixture of acetone/methanol (4:1) and washed until no N-BOC-hydrazine was detected on TLC. The precipitate was dissolved in water, treated with a strong cation exchanger to remove traces of the starting material and lyophilized. For deprotection of the amino function the lyophilisate was dissolved in water/MeOH (3:1), cooled on ice and treated with gaseous HCI under moderate stirring. The reaction was monitored with ninhydrin on TLC plates and stopped upon completion of the reaction by adding 1 M potassium phosphate until a pH of about 7. The reaction mixture was extensively dialyzed against water under argon atmosphere at 4 °C. The amount of hydrazide in the product was estimated using the TNBS method (Inman JK, Biochem., 8, p. 4074- 4082(1969)). Briefly, the substance was dissolved in 950 μl 0.5 M sodium carbonate buffer, pH 10.5 and 50 μl 0.25% 2,4,6-trinitrobenzosulfonic acid (TNBS) were added and incubated for 2 hrs at 37 °C. After cooling the absorbance was read λ=425 nm. The yield was about 72%.

Example 2: Synthesis of HES-SH

2.5 g (0.1 mmol) HES ₇₀ were dissolved in 50 ml buffer (0.1 M NaHCO ₃ , 2 mM EDTA, pH 8.4). 450 mg (2 mmol) cysteamine dihydrochloride and 500 mg (8 mmol) sodium cyanoboron hydride were added and the pH was adjusted to 7.5 with diluted sodium

hydroxide solution. The reaction mixture was kept at 37 ⁰ C for one week. Thereafter, the solution was dialyzed against dest. water and lyophilized. The amount of disulfide bonds was determined by treatment with 2-nitro-5-thiosulfobenzoate (NTBS) as described by Thannhauser, Konishi et Scherga, Anal. Biochem., 138, 181-188, 1984 and was found to be 32 nmol/mg. 1.0 g of this product was dissolved in 50 ml buffer (0.1 M NaHPO ₄ , 1mM EDTA, pH 8) under argon atmosphere. An 2-fold excess of 1 ,4-Dithioerythritol (DTE) was added and the reaction mixture was stirred for 2 hrs at 20 ⁰ C. The solution was dialyzed against 0.01 M HCI, 10 mM ascorbic acid for 2 days. The amount of thiol-groups determined by treatment with 5,5'-dithio-bis-(2- nitrobenzoic acid) (DTNB) as described by Pierce (Instructions Ellman ^' s reagent, No. 22582, Pierce, Rockland, IL) was 30 nmol/mg.

Example 3: Synthesis of HES-MPH

a) HES ₇₀

Sulfhydryl-functionalized HES ₇₀ from example 2 was treated with a 10-fold excess of the crosslinker 4-maleimidopropionic acid hydrazide trifluoracetate (MPH) in dry DMSO under argon atmosphere for 4.5 hrs at room temperature. At the end of the reaction the HES derivative was precipitated by adding cold acetone, centrifuged and redissolved in water. The product was purified by a Hi-prep SEC 26/10 column (Amersham Biotech). The high molecular weight fraction was pooled and lyophilized. The yield of the modification reaction was determined by quantifying the hydrazide group with TNBS at 500 nm. The yield was about 75%.

b) HES ₅₀

In a variation to scheme a) a HES derivative was used, wherein the SH-group was protected by a disulfide bond as a cystamin derivative. In this case an HES species with MW=50 kd and DS=O.5 was employed. With this HES ₅₀ an even better yield of close to 100% was obtained.

Example 4: Synthesis of HES-MPBH

a) HES ₇₀

Sulfhydryl-functionalized HES70 from example 2 was treated with a 10-fold excess of the crosslinker 4-(4-N-maleimidophenyl)butyric acid hydrazid x HCI (MPBH) in dry

DMSO under argon atmosphere for 4.5 hrs at room temperature. At the end of the reaction the HES derivative was precipitated by adding cold acetone, centrifuged and redissolved in water. The product was purified by a Hi-prep SEC 26/10 column (Amersham Biotech). The high molecular weight fraction was pooled and lyophilized. The yield of the modification reaction was determined by quantification of the hydrazide group with TNBS at 500 nm. The yield was about 80%.

b) HES ₅₀

This reaction was done as described in example 3 b) but with MPBH as cross linker. The yield of the coupling reaction determined as described before was about 83%.

Example 5: Quantitative estimation of the degree of oxidation

To determine the degree of oxidation in a given carbohydrate a modified BCA assay was used. Aldehyde groups formed by enzymatic oxidation were oxidized by Cu ²⁺ ions in alkaline solution. The resulting Cu+ ions were bound by 2,2 ^' bis-cinchoninate (BCA) to form a stable blue complex with an absorption maximum at 560 nm. D- Glucose was used a reference to calibrate the assay.

Example 6: Oxidation of raffinose by GAO

Raffinose, a non-reducing trisaccharide, was dissolved in 50 mM potassium phosphate buffer, pH 6.0, with 0.1 mM CuCI ₂ at a concentration of 10 mM. 0.1 u galactose oxidase, 2 u catalase from bovine liver and 0.1 u peroxidase from horse radish were added per 1 μmol raffinose, respectively. The reaction mix was stirred under oxygen pressure for 20 hrs. The yield determined by the BCA assay as described in example 5 was 40% under optimized conditions.

Example 7: Substrate specificity of GAO

In glycoproteins the carbohydrate chains may be either attached to Ser/Thr as O- glycosides or to Asn as N-glycosides. A typical structure of O-glycosides is the Gal- containing disaccharide D-Gal-β[1-3]-D-GalNAc, whereas D-Gal-β[1-4]-D-GlcNAc represents a typical N-glycosidic structure. Both structures as the corresponding O- methyl glycosides were oxidized by the three-enzyme system described above to

determine a possible preference of the enzyme for one of these structures. The results (Fig. 1) clearly show a strong preference for the D-Gal-β[1-3]-D-GalNAc structure over D-Gal-β[1 -4]-D-GlcNAc as substrate.

Example 8: Immobilization of GAO

10 mg of recombinant GAO (Whittaker) purified from Pichia pastoris was dissolved in 0.1 M sodium carbonate buffer pH 8 and incubated with 10 ml activated CH- sepharose, comprising a 6-aminohexanoic spacer and N-hydroxysuccinimide as reactive group. After incubation for 4 hrs at 4 °C the matrix was washed with coupling buffer and incubated with 1 M ethanolamine for about 1 hr. The chemical coupling yield was close to 100 %, i.e. no GAO enzyme was detected in the wash fractions. The activity of the immobilized enzyme was about 60% compared to the free one.

Enzymatic activity of the galactose oxidase was determined according to Avigad et al. (J. Biol. Chem, 237, p. 2736-2743 (1962)). Briefly, 1.7 ml 0.1 M potassium phosphate buffer, pH 6.0, containing 0.5% o-tolidine were added to 1.5 ml 10% galactose in H ₂ O. After adjustment of the temperature to 25 °C 0.1 ml peroxidase (60 u/ml) was added and the reaction was started with 0.1 ml GAO preparation. The change of absorbance with time was read at 425 nm and the activity was calculated from the linear portion of the curve. For the immobilized enzyme an aliquot of the coupling product was used in the assay.

Example 9: Immobilization of catalase

50 mg catalase from bovine liver (Sigma) were incubated with 10 ml activated CH- sepharose similar to the method described in example 8. The chemical yield in this case was about 75%, with only 40% of the total activity in the immobilized state. Therefore, immobilization on azlactone-fractogel (Merck) was explored as an alternative. Again, 50 mg catalase in 50 mM potassium phosphate buffer pH 7 were incubated with 10 ml activated gel for 2 hrs at 4 °C . The gel was then washed with coupling buffer and treated with 0.2 M glycine, pH 8, for 1 hr and washed again with coupling buffer. Under these conditions a chemical yield of 80% was obtained with about 70% of the enzymatic activity retained.

The activity of catalase was measured as described by Beers & Sizer (J. Biol. Chem, 195, p. 133ff (1952)). Briefly, to 1.9 ml water, 1.0 ml of 0.059 M hydrogen peroxide was added and equilibrated to 25 ⁰ C. The reaction was started by the addition of 0.1 ml catalase in 50 mM potassium phosphate buffer, pH 7.0. The change of absorbance at 240 nm with time was read and the enzymatic activity was calculated from the linear portion of the curve.

Example 10: Immobilization of POD

The azlactone method was also applied to the immobilization of peroxidase (POD). 10 mg POD were used per 10 ml activated gel under the same conditions described in example 8. This amount of POD was quantitatively coupled to the matrix. The yield in terms of enzymatic activity was about 80%.

To determine the enzymatic activity 1.0 ml 100 mM potassium phosphate buffer, pH 7.2, was equilibrated to a temperature of 25 °C. 16.7 μl 20.1 mM guajacol solution and 10 μl 12.3 mM hydrogen peroxide were added. To this solution 33 μl of peroxidase from horse radish (Sigma) in phosphate buffer was added and the reaction was followed for 15 min at 436 nm in a spectrophotometer.

Example 11 : Oxidation of fetuin by GAO

Oxidation of fetuin (Sigma) was performed at concentrations up to 1 mg/ml in 50 mM potassium phosphate buffer, pH 6.0, with variable ratios of the enzymes GAO, catalase and POD at 25 °C for up to 120 hrs under sterile conditions. As expected from literature data very low yields were obtained (< 10% of the fetuin molecules had an aldehyde function according to the BCA assay described in example 5).

Example 12: Preparation of asialofetuin

In order to get a better protein substrate for the GAO the sialic acid residues of fetuin were removed. This was achieved by treatment with an appropriate sialidase enzyme. 1 ml immobilized sialidase from vibrio cholerae (Galab, Geesthacht, Germany) with 1 U/ml gel was used to treat 10 mg fetuin for 20 hrs at 37 ⁰ C in 50 mM

sodium acetate buffer, pH 5.5, containing 1 mM CaCI ₂ . The absence of sialic acids in the treated fetuin was demonstrated by isoelectric focussing (Fig. 2)

Example 13: Oxidation of asialofetuin by GAO

The optimized incubation mix for the oxidation of asialofetuin in the three-enzyme approach consisted of 50 mM potassium phosphate buffer, pH 6.0, 1 U/ml GAO, 20 U/ml catalase, 1.5 u/ml POD and 0.5 mg/ml asialofetuin in a total volume of 1 ml. The reaction was run at 25 ⁰ C for up to 120 hrs under sterile conditions in an oxygen atmosphere. The degree of oxidation was determined by the BCA method described in example 5. Because the presence of proteins will also cause an increase in the absorbance at 560 nm, although much lower than the effect of the aldehyde oxidation, this was corrected by measuring the calibration curve in the presence of protein. The yield was calculated as % aldehyde groups per available galactose units and was in the range of 55%. This yield was reached after about 30 hrs incubation time (see. fig. 3). Fig. 4 demonstrates the presence of newly formed aldehyde groups in the asialofetuin by glycan detection. The reaction product was separated by SDS- PAGE and blotted onto a nitrocellulose membrane. The membrane was treated with digoxigenin-3O-succinyl-ε-aminocaproic acid hydrazide to form a stable hydrazone between aldehydes of asialofetuin and the hydrazid. The presence of the digoxigenin derivative was detected by an digoxigenin-specific antibody coupled to alkaline phosphatase with NBT/X-phosphate as a substrate.

Example 14: Coupling of oxidized asialofetuin with HES-hydrazide

0.5 mg oxidized asialofetuin from example 13 was dissolved in 1 ml 0.1 M K- phosphate buffer, pH 5.5. To this solution HES _7O hydrazide was added to obtain a 10- fold molar excess of the hydrazide over the aldehyde groups of the glycoprotein. Incubation was typically done at room temperature for 24 hrs. The coupling product was analyzed by SDS-PAGE and gel permeation chromatography. From the results of the SDS-PAGE a coupling yield of about 70% could be estimated, i.e. about 30% of the oxidized asialofetuin remained unchanged after the reaction. Fig. 6 shows the distribution between the coupling product and the remaining oxidized asialofetuin in a GPC analysis.

Example 15: Coupling of oxidized asialofetuin with HES- MPH

Similar to example 14 0.5 mg oxidized asialofetuin from example 13 were dissolved in 1 ml 0.1 M potassium phosphate buffer, pH 5.5. To this solution HES ₅₀ -MPH was added to obtain a 10-fold molar excess of the hydrazide over the aldehyde groups of the glycoprotein. Incubation was usually done at room temperature for 24 hrs. The coupling product was analyzed by SDS-PAGE and gel permeation chromatography.

From the results of the SDS-PAGE a coupling yield of about 50% was estimated, i.e. about 50% of the oxidized asialofetuin remained unchanged after the reaction (fig. 5).

Example 16: Coupling of oxidized asialofetuin with HES- MPBH

Similar to example 15 0.5 mg oxidized asialofetuin from example 13 were dissolved in 1 ml 0.1 M potassium phosphate buffer, pH 5.5. To this solution HES ₅₀ -MPBH was added to obtain a 10-fold molar excess of the hydrazide over the aldehyde groups of the glycoprotein. Incubation was typically done at room temperature for 24 hrs. The coupling product was analyzed by SDS-PAGE and gel permeation chromatography.

From the results of the SDS-PAGE a coupling yield of about 70% could be estimated, i.e. about 30% of the oxidized asialofetuin remained unchanged after the reaction.

Example 17: Oxidation of erythropoetin by GAO

1 mg EPO was desialylated by the same procedure as described in example 12 for fetuin. 0.5 mg of the asialo-EPO in 50 mM potassium phosphate buffer, pH 6.0, was treated with 1 U immobilized GAO, 10 U immob. catalase and 2.5 U immob. POD at

25 °C for 72 hrs under sterile conditions and with sterile oxygen bubbling through the suspension. The degree of oxidation was determined by the BCA method described in example 12. About 30% of all available galactose units were oxidized to the corresponding aldehyd.

Example 18: Coupling of oxidized erythropoetin with HES-MPBH

0.5 mg of the oxidized asialo-EPO from example 17 can be dissolved in 1 ml 0.1 M potassium phosphate buffer, pH 5.5. HES ₅₀ -MPBH in a 5-fold molar excess was

added and the reaction was allowed to proceed for 48 hrs at room temperature. SDS- PAGE analysis and gel permeation chromatography show that the majority the EPO can be found in the high molecular range, i.e. is conjugated to the HES polymer.

The biological activity of the HES-EPO can be measured in vitro using the cell line NFS-60 according to Hara K et al. (Experimental Hematology, 16, 256-61 (1988)). Briefly, growth proliferation of this cell line was measured in response to EPO by a colorimetric method. Alamar blue (Biosource, Camarillo, CA) was used to stain the cells and measured at 570 nm.

Example 19: Oxidation of granulocyte-colony stimulation factor by GAO

1 mg G-CSF may be desialylated by the same procedure described in example 12 for fetuin. 0.75 mg of the asialo-G-CSF in 50 mM potassium phosphate buffer, pH 5.5, is treated with 1 U immobilized GAO, 20 U immob. catalase and 2 U immob. POD at 25 ⁰ C for 72 hrs under sterile conditions and with sterile oxygen bubbling through the suspension. As expected for O-glycosidic linked oligosaccharides the degree of oxidation reached in this case is much higher compared to example 17.

Example 20: Coupling of oxidized granulocyte-colony stimulating factor with HES-hydrazide

HESγo-hydrazide prepared according to example 1 may be used as coupling partner for the oxidized asialo-G-CSF. The reaction conditions applied in this case are similar to example 18. 0.5 mg of the oxidized glycoprotein were dissolved in 1 ml 0.1 M potassium phosphate buffer, pH 6.5. A 10-fold excess of the HES ₇₀ -hydrazide is added and the reaction is allowed to proceed over night at room temperature. SDS- PAGE and GPC were again used to analyze the coupling product. The coupling efficiency was slightly lower than observed in example 18 with EPO, but still in an acceptable range.

The method of Yamaguchi et al. (Biol. Pharm. Bull. 20, p. 943-947 (1997)) was used to determine the in vitro biological activity of HES-G-CSF. Briefly, the promyelocytic human cell line HL-60 was pretreated with DMSO and retinoic acid to induce differentiation. In response to G-CSF under these conditions the cell line responded

with an increase in proliferation. This proliferation response was measured photometrically with Alamar blue as described in example 18.

Example 21 : Oxidation of interleukin-2 by GAO

Recombinant human IL-2 produced in CHO cells was used as additional target for the enzymatic oxidation. In this case no desialylation reaction was performed, because analytical data had shown that this preparation was incompletely sialylated. 100 μg of the glycoprotein were treated with the three enzymes in a ratio of 0.5 U/10 U/2 U for GAO/catalase/POD in 50 mM potassium phosphate buffer, pH 5.5, at 25 °C for 40 hrs under sterile conditions and with sterile oxygen bubbling through the suspension. The reaction was stopped and the product was analyzed by the BCA method. The oxidation yield was found to be about 45%.

Example 22: Coupling of oxidized interleukin-2 with HES-hydrazide

The coupling of the oxidized IL-2 was done with HES ₇ o-hydrazide prepared according to example 1. The reaction mix from example 21 was directly incubated with a 20-fold excess of HES ₇ o-hydrazide and the reaction was run for 24 hrs at room temperature. The coupling product was analyzed by SDS-PAGE and GPC as already shown for example 13 and 14. About half of the IL-2 present was found in the coupling product.

The in vitro bioactivity of the HES-IL-2 was estimated using a cell proliferation assay with the cell line CTLL-2, which specifically responds to active IL-2 (Weston L et al. Immunology and Cell Biology, 76, 190-192 (1998)). The proliferation response to IL-2 is detected by measuring the absorbance of Alamar blue as described in example 18. About 28 % of the original CTLL-2 proliferating activity was detected in the HES-IL-2 coupling product.