Polysialic acids (PSAs) are naturally occurring unbranched polymers of sialic acid produced by certain bacterial strains and in mammals in certain cells [Roth et. al., 1993]. They can be produced in various degrees of polymerisation from n = about 80 or more sialic acid residues down to n = 2 by limited acid hydrolysis or by digestion with neuraminidases, or by fractionation of the natural, bacterially derived forms of the polymer. The composition of different polysialic acids also varies such that there are homopolymeric forms i. e. the alpha-2, 8-linked polysialic acid comprising the capsular polysaccharide of E. coli strain K1 and the group-B meningococci, which is also found on the embryonic form of the neuronal cell adhesion molecule (N-CAM). Heteropolymeric forms also exist-such as the alternating alpha-2, 8 alpha-2, 9 polysialic acid of E. coli strain K92 and group C polysaccharides of N. meningitidis. Sialic acid may also be found in alternating copolymers with monomers other than sialic acid such as group W135 or group Y of N. meningitidis. Polysialic acids have important biological functions including the evasion of the immune and complement systems by pathogenic bacteria and the regulation of glial adhesiveness of immature neurons during foetal development (wherein the polymer has an anti-adhesive function) [Muhlenhoff et. al., 1998; Rutishauser, 1989; Troy, 1990,1992 ; Cho and Troy, 1994], although there are no known receptors for polysialic acids in mammals. Polysialic acid is also present in human milk.

CD36 is a new member of mammalian polysialic acid-containing glycoprotein [Yabe et. al., 2003]. The alpha-2, 8-linked polysialic acid of E. coli strain K1 is

also known as'colominic acid'and is used (in various lengths) to exemplify the present invention.

The alpha-2, 8 linked form of polysialic acid, among bacterial polysaccharides, is uniquely non-immunogenic (eliciting neither T-cell or antibody responses in mammalian subjects, even when conjugated to immunogenic carrier proteins) which may reflect its status as a mammalian (as well as a bacterial) polymer. Shorter forms of the polymer (up to n=4) are found on cell-surface gangliosides, which are widely distributed in the body, and are believed to effectively impose and maintain immunological tolerance to polysialic acid. In recent years, the biological properties of polysialic acids, particularly those of the alpha-2, 8 linked homopolymeric polysialic acid, have been exploited to modify the pharmacokinetic properties of protein and low molecular weight drug molecules [Gregoriadis, 2001; Jain et. al., 2003 and 2004; US-A-5846, 951: WO-A-0187922; Jain et. al., 2004]. Polysialic acid derivatisation gives rise to dramatic improvements in circulating half-life for a number of therapeutic proteins including catalase and asparaginase [Fernandes and Gregoriadis, 1996 and 1997], and also allows such proteins to be used in the face of pre-existing antibodies raised as an undesirable (and sometimes inevitable) consequence of prior exposure to the therapeutic protein [Fernandes and Gregoriadis, 2001]. There are questions about the use of PEG with therapeutic agents that have to be administered chronically, since PEG is only very slowly biodegradable [Beranova et. al., 2000] and high molecular weight forms tend to accumulate in the tissues [Bendele et. al., 1998; Conover et. al., 1997]. Likewise there are concerns about the utility of PEG in therapeutic agents that may require high dosages, since accumulation of PEG may lead to toxicity. The alpha-2, 8 linked polysialic acid (PSA) therefore offers an attractive alternative to PEG,. being an immunologically invisible biodegradable polymer which is naturally part of the human body, and which degrades, via tissue neuraminidases, to sialic acid, a non-toxic saccharide.

Our group has described, in previous scientific papers and in granted patents, the utility of natural polysialic acids in improving the pharmacokinetic properties of protein therapeutics [Gregoriadis, 2001; Fernandes and Gregoriadis; 1996,1997, 2001; Gregoriadis et. al., 1993,1998, 2000; Hreczuk-Hirst et. al., 2002; Mital, 2003; Jain et. al., 2003; US-A-5846951; WO-A-0187922]. We also describe new mono-functional derivatives of PSAs, which allow new compositions and methods of production of PSA- derivatised proteins (and other forms of therapeutic agent) in our co-pending application filed the same date as the present application. These new materials and methods are particularly suitable for the production of PSA- derivatised therapeutic agents intended for use in humans and animals, where the chemical and molecular definition of drug entities is of major importance because of the safety requirements of medical ethics and of the regulator authorities (e. g. FDA, EMEA).

Jennings and Lugowski, in US 4,356, 170, describe derivatisation of bacterial polysaccharides to proteins via an activated non-reducing terminal unit. In one example the polysaccharide is subjected to a preliminary reduction step but the product is then subjected to an oxidation step. They suggest this approach where the terminal unit is N-acetyl mannosamine, glucose, glucosamine, rhamnose and ribose.

Of the various methods, which have been described to attach polysialic acids to therapeutic agents [US-A-5846951; WO-A-0187922], none of these involve chemical modification of the reducing end, because of its weak reactivity towards therapeutic proteins. Nevertheless, despite the inefficiency of this reaction, we have observed that it gives rise to unintentional by-products during conjugation reactions intended to produce conjugates with protein via an introduced aldehyde at the non-reducing (opposite) end of the polymer by the schemic shown in Figure 1. The potential for such products is evident in published studies of catalase, insulin and asparaginase [Fernandes and Gregoriadis; 1996,1997, 2001; Jain et. al., 2003 and 2004], where the hemiketal of the reducing terminal of the

natural (chemically unmodified) form of the polymer gives rise to protein conjugates at a low level of efficiency (less than 5% of protein becoming derivatised, see further below in the reference examples, and table1) during reductive amination.

The reducing end sialic acid residue of compounds such as polysialic acid (PSA) spontaneously forms an open ring ketone by tautomerisation (Fig. 2). In the dynamic equilibrium between ring and linear structures of the reducing end sialic acid residue, the ketone moiety is present on only a subpopulation of PSA molecules at any one instant. Thus as illustrated in figure 2, in solution, the terminal sialic acid residue at the reducing end of polysialic acid exists in a tautomeric equilibrium. The ring-open form, although in low abundance in the equilibrium is weakly reactive with protein amine groups.

The reactivity of the reducing end of colominic acid (CA), though weak towards protein and other targets, is sufficient to be troublesome in the use as pharmaceutical excipients etc. of the kind likely to be preferred by regulatory authorities for use in man and animals. Polysialic acids such as colominic acids have been used in cosmetics in JP-A-04089417, JP-A- 05194173 and JP-A-04-036213 for their properties as moisture-retaining viscosity-increasing, and similar functions.

Sialic acids have been used to coat filters for use to remove viral particles from the air (US 5851395).

Colominic acids have been used as therapeutically effective anti- inflammatory compounds, for instance to treat renal diseases, hepatitis, and other indications where its use as an immunomodulator or as an inhibitor of chemotaxis of neutrophils is of therapeutic value. (US 5516764, US 5763420 and US 5668115 and JP-A-05-025050). Complexes of colominic acids with metals such as aluminium or ruthenium may be useful, especially in vaccines, in the prophylaxis and treatment of cerebiospinal meningitis (US 4740589). JP-A-05-043468 suggests colominic acids are effective in the

treatment of arthrorheumatism. In JP-A-04-279526 colominic acid has nephritis inhibitory properties and inhibits platelet coagulation.

The conventional CA used in the above disclosures may react weakly with body components (e. g. glycation) forming a Schiffs base via the ketal form of the reducing terminal unit. The colominic acid may be irritating or toxic/allergenic or may block protein surface; or hard to remove etc.

Now therefore we have solved the problems by developing a new method for use of sialic acid compounds having deactivated sialic acid groups at what would be the reducing terminal. These deactivated compounds may be of use as a pharmaceutical excipient.

In the invention there is provided a new process for producing an unreactive derivative of sialic acid in which a sialic acid starting material having a reducing sialic acid unit is subjected to a passivation step in which the ketal ring of the reducing sialic acid unit is opened and reacted to prevent reformation of the ketal ring (Figure 3).

The starting material is preferably a di-, oligo-or poly-saccharide compound. More preferably the starting material is an oligo-or poly-sialic acid. The source of the polysialic acid may be bacterial or mammalian, such as E coli k92, E. coli k1, Group B polysaccharides of meningococci, cow's milk or from the embryonic form of N-CAM. Hekro-polymers such as group W135 or group Y N. meningitidis.

The sialic acid starting material used in the process of the invention is preferably a compound having a sialic acid unit at the reducing terminal end joined to an adjacent, usually a saccharide, unit through its eight carbon atom. In an alternative embodiment, the reducing sialic acid unit is joined to an adjacent saccharide unit through the 9 carbon atom.

In the process of the invention it is preferred that the end product of the process has a terminal saccharide unit at the non-reducing end which is a inert group. A passivated non-reducing end of a saccharide compound may be created by subjecting a starting material to a preliminary step, prior to the passivation step of selective oxidation to oxidise any vicinal diol group

to an aldehyde, whereby in the passivation step the aldehyde is also passivated. The invention is of particular utility where the terminal unit of non-reducing end of polysaccharide starting material is a sialic acid unit.

The invention is of most utility where the starting material is a di-, oligo-or poly-sialic acid, that is a compound in which sialic acid units are joined to each other in a linear chain, with the units joined 2-8,2-9 or alternating 2-8,2-9 to one another.

In the invention a polysaccharide may comprise units other than sialic acid in the molecule. For instance sialic acid units may alternate with other saccharide units. Preferably, however, the polysaccharide consists substantially only of units of sialic acid. Preferably these are joined 2-8 and/or 2-9.

Preferably the starting material has at least 2, more preferably at least 25, more preferably at least 50, for instance at least 200 or more, saccharide units. For instance the polysaccharide may comprise at least 5 sialic acid units. Preferably the polysaccharide has a low polydispersity of molecular weight. We have found it is possible to fractionate polysialic acid for instance using ion-exchange chromatography. One process is described below in the worked examples.

In the process the passivation step is preferably a ring opening reduction, in which the ketal is reduced to an alcohol. This is shown in Figure 3 for the preferred starting material polysialic acid wherein the units are 2-8 linked.

Suitable reduction conditions (for steps) may utilise hydrogen with catalysts or, preferably hydrides, such as borohydrides. These may be immobilised such as Amberlite (trade mark) -supported borohydride.

Preferably alkali metal hydride such as sodium borohydride is used as the reducing agent, at a concentration in the range 1uM to 0. 1 M, a pH in the range 6.5 to 10, a temperature in the range 0 to 60°C and a period in the range 1 min to 48 hours. The reaction conditions are selected such that pendant carboxyl groups on the polysaccharide (e. g. such groups in

polysialic acid) are not reduced. Where a preliminary oxidation step has been carried out on the non-reducing terminal unit, the aldehyde group generated is reduced to an alcohol group not part of a vicinal diol group.

This reaction is shown for polysialic acid in Figure 4. Other suitable reducing agents are cyanoborohydride under acidic conditions, e. g. polymer supported cyanoborohydride or alkali metal cyanoborohydride, L-ascorbic acid, sodium metabisulphite, L-selectride, triacetoxyborohydride etc.

According to an alternative embodiment of the passivation step of the invention the sialic acid reducing unit is reacted with a primary amine or hydrazide reagent, preferably under reducing conditions to end-cap the unit via a secondary amine or hydrazone linkage. Suitable reagents are ammonia, ethylamine, hydroxylamine, ethanolamine, hydrazine phenylhydrazide etc. The reaction scheme for a polysialic acid is shown in Figure 5.

Generally this aspect of the invention is part of a process in which the sialic acid starting material also has a non-reducing terminal unit which is activable and is required to be inert. In such a process the non-reducing terminal unit is generally passivated by activating then inactivating the sialic acid non-reducing terminal units. Such a reaction is, for instance selective oxidation of a vicinal diol moiety e. g. of a non-reducing sialic acid terminal unit, followed by inactivation/passivation simultaneously with the passivation of the reducing terminal.

In a preliminary oxidation step of a non-reducing saccharide unit, preferably a sialic acid unit, the selective oxidation should preferably be carried out under conditions such that there is substantially no mid-chain cleavage of the polysaccharide chain, that is substantially no molecular weight reduction. Enzymes which are capable of carrying out this step may be used. Most conveniently the oxidation is a chemical oxidation. The reaction may be carried out with immobilised reagents such as polymer- based perrhuthenate. The most straight forward method is carried out with dissolved reagents. The oxidant is suitably perrhuthenate, or, preferably,

periodate. Oxidation may be carried out with periodate at a concentration in the range 1 mM to 1 M, at a pH in the range 3 to 10, a temperature in the range 0 to 60°C for a time in the range 1 min to 48 hours.

Between any oxidation step and reduction step the respective intermediate must be isolated from oxidising and reducing agents, respectively, prior to being subjected to the subsequent step. Where the steps are carried out in solution phase, isolation may be by conventional techniques such as expending excess oxidising agent using ethylene glycol, dialysis of the intermediate, ethanol precipitation and ultrafiltration to concentrate the aqueous solution. The product mixture from the reduction step again may be separated by dialysis and ultrafiltration. It may be possible to devise reactions carried out on immobilised oxidising and reducing reagents rendering isolation of product straightforward.

The process of the invention produces a sialic acid derivative product which is inert and is suitable as a pharmaceutical excipient, or for other uses for which colominic acid has been used in the prior art, e. g. as described above. The product may be suitable for uses as pharmaceutical excipient, a wetting agent, moisturising agent, conditioning agent (less viscous, compared to PEG), as a water-soluble base (e. g. easily removable by water etc. ), as a flocculating agent e. g. , for suspensions, as an osmotic agent (e. g. in an osmolic pump), a diluent or vehicle, a suspending agent (prevention of aggregation etc.), complexing agent (e. g. complexes with metal ions, Ca2+), a cation exchange resin, use in filters (e. g. for microorganism such as viruses), as coatings on membranes or other separation media, a solubilising agent (e. g. cationic drugs), a precipitating agent (e. g. proteins, concentration dependent), a cryoprotecting agent (eg. for proteins), a disintegrating agent (e. g. for tablets tec. ), for surface coating (e. g. masking bitter taste of drugs etc. ), in a separation system such as a biphasic system (with PEG, for instance), a stabiliser (e. g. based on charge and/or pH etc. ), a viscosity enhancer, a lubricant, glidant and anti-adhesive for powders and granules etc.

The products have particular utility in drug delivery systems. In such systems the product may be used for the preparation of hydrogels e. g. by cross linking e. g. carboxyl groups using EDC in dry environment, ester bonds are formed between carboxyl and hydroxyl groups. Ester bonds are known to hydrolyse in aqueous environment, and these cross-links may therefore be used to control dissolution of a hydrogel matrix in which a pharmaceutical active is dispersed. The products may also be used in other drug delivery systems such as based on microspheres, nanospheres, colon targeting, bioadhesion, mucoadhesion, osmotic pump systems, other matrix systems including depot systems. The product may also have utility in conventional drug delivery, generally not as a water-insoluble material but rather a water-soluble material, used as a matrix, in a suppository, in ointments (e. g. water-based ointments) and eye or ear drops.

The products may also be useful as diluents and carriers in various compositions such as of use in cosmetics, for instance in hair-styling products, creams, e. g. skin creams, nail polish etc.

Preferably compositions comprising the compounds are aqueous, but the products may also be used in organic solvent compositions such as based on or produced in DMSO.

It is believed the product compounds may themselves have some biological activities and may be useful as pharmaceutical actives. Examples of drug uses are as a laxative, as an anti-inflammatory, as an antiviral, as neuraminidase inhibitor, as an affinity based binding agent as an immunomodulator or as an antibacterial.

Pharmaceutical compositions comprising the product and a diluent may be aqueous or non-aqueous.

According to the invention there is also provided a novel compound which is a sialic acid derivative in which the terminal unit at the reducing end includes an inert moiety and at a non-reducing end is also inert.

The novel compound is preferably a di-, oligo-or poly-saccharide, most preferably a polysialic acid. Preferably the compound comprises

saccharide units consisting only of sialic acid units or, alternatively, may comprise other saccharide units in addition to sialic acid units. The compound may generally be as described above in relation to the first aspect of the invention.

The novel compound is substantially inert under normal conditions in the presence of drug or other biologically active compounds. Thus it is substantially free of reactivity via the non-reducing end of the an oligo or polysaccharide chain to biological molecules such as biologically active compound, for instance a pharmaceutically active compound.

The novel compound may have the general formula I in which R'is an organic group; Gly is a glycosyl group; RI and R3 are selected from one of the following combinations: i) R2 is H and R3is OH; ii) Ruz ils H and R3 is NR4R5 where R4 is selected from H and Cl-6 alkyl and R5 is selected from H, C1-24 alkyl, aryl and NR62where each R6 is independently selected from H, C1-6 alkyl and aryl ; or iii) R2 and R3 together are =N-R7 where R7is H, Cl-24 alkyl, aryl or -NR32 where each R8 is independently selected from H, C1-24 alkyl and aryl ; Ac is acetyl ; and

n is 0 or more.

In the general formula I the broken bonds from two carbon atoms to the oxygen atom which is part of the glycosyl group and to the hydroxyl group denote that there is a bond from either one of the carbon atoms to the oxygen atom of the glycosyl group and a bond from the other carbon atom to the hydroxyl group. Thus the terminal group may be joined through the carbon atom which was the 8-carbon of the reducing sialic acid unit from which the novel compound is derived or the 9-carbon.

In the compound of the formula 1, Gly is a glycosyl group. Where n is 0, the compound may be a monosialic acid derivative. Where n is 1 or more the compound is a di-, oligo-or polysaccharide. The glycosyl groups of such a compound may be the same or different from one another. In one embodiment the groups are the same and are preferably sialic acid units, joined to one another 2-8,2-9 or alternatively 2-8/2-9. Preferably n is at least 4, more preferably at least 9, for instance at least 20, or even 50 or more, up to 1000 or sometimes even more than this. Preferably n is up to 500.

The group R'is an organic group which may be a glycosyl group, e. g. a naturally occurring non-reducing terminal unit. For the preferred embodiment of the compound which is a derivative of an oligo-or poly-sialic acid, the group RI may be a sialyl group or a derivatised version thereof, namely the product of a cycle of oxidation and reduction of a non-reducing unit. In such compounds R'is where RI and R3 have the same meaning as at the other end of the molecule.

Where R is H and R3 is NR4R5, the compound is an unreactive hydrazine or secondary amine. The sialic acid unit is, in this instance,

end-capped to passivate it. As described above these compounds are produced by reaction with amines or hydrazide under reducing conditions.

Where R'and R'together are =NR'the compound is an intermediate and is potentially reactive. Before it is used in an application where its lack of reactivity is important it should generally be reduced, to form the corresponding compound in which R is H and R3 is NR4R5 and at least one of R4 and R5 is H.

The reaction scheme for end-capping the reducing terminal sialic acid unit of polysialic acid is illustrated in Figure 5.

The invention further provides compositions comprising the novel compounds and a diluent or carrier, as well as pharmaceutical compositions comprising the novel compounds, and a pharmaceutical acceptable carrier or diluent. The compositions may be liquid or solid, or semi-solid, such as gel-form. The compositions may be suitable for food or beverage use, in natraceuticals and in cosmetics. Pharmaceutical compositions may be administered orally, intravenously, intraperitoneally, intramuscularly, subcutaneously, intranasal, intradermal or intratracheally, topical, etc.

The new deactivated (passivated) form of the polysialic acid or other polysaccharide is more conducive to the use as an additive to compositions comprising biologically active compounds, for instance in pharmaceutical acceptable products, since it avoids the risk of by-product formation which is otherwise inadvertently created by use of compounds with unmodified reducing ends. The novel compounds are biodegradable, non- immunogenic, non-antigenic and non-allergenic as well as being cost- effective. The compounds are stable at elevated temperatures. The compounds will have utility in any of the applications for which colominic acids have been used in the prior art, as well as a preferable replacement for PEG.

The protein reactivity (by reductive amination) of the various intermediates described in figure 4 is described in Table 1. Notably, these data demonstrate that the intermediate'CAOR', created by borohydride

reduction of the periodate oxidised polymer, is inert towards protein targets, proving that both its aldehyde and hemiketal moieties have been destroyed by borohydride reduction. This new scheme of reaction (Fig. 4 or 5), method elegantly avoids the need to purify away the intended product from the various unintended products (described in Fig. 2), which are completely avoided in this new reaction scheme.

Brief Description of the Drawings: Figure 1 is a reaction scheme of the prior art polysialic acid activation process; Figure 2 is a reaction scheme showing the tautomerisation of the reducing terminal unit of a polysialic acid; Figure 3 is a reaction scheme of the simples embodiment of the present invention using polysialic acid as the starting material ; Figure 4 is a reaction scheme of an alternative embodiment of the present invention using polysialic acid as the starting material ; Figure 5 is a reaction scheme of an alternative embodiment of the present invention.

Figure 6 is the results of the native PAGE of polysialic acids and of the products of the reaction scheme shown in Figure 3 and of intermediates of the reaction scheme shown in Figure 4 as described in Example 1; Figure 7 shows further SDS-PAGE results of Example 1; Figure 8 shows the GPC results of Example 2; Figure 9 shows the SDS-PAGE results of Example 3; and Figure 10 shows the DSC results of Example 3.

The invention is illustrated further in the accompanying examples.

Examples Materials Ammonium carbonate, ethylene glycol, polyethylene glycol (8KDa), sodium cyanoborohydride (> 98% pure), sodium meta-periodate and molecular weight markers were obtained from Sigma Chemical Laboratory, UK. The colominic acid (CA) used, linear a-(2<8)-linked E. coli K1 polysialic

acids (22.7 kDa) was from Camida, Ireland, radioactive iodide (Na'251) was purchased from Amersham, UK. Other materials included 2,4 dinitrophenyl hydrazine (Aldrich Chemical Company, UK), dialysis tubing (3.5KDa and 10KDa cut off limits ; Medicell International Limited, UK), Sepharose SP HiTrap, PD-10 columns (Pharmacia, UK), Tris-glycine polyacrylamide gels (4-20% and 16%), Tris-glycine sodium dodecylsulphate running buffer and loading buffer (Novex, UK). Deionised water was obtained from an Elgastat Option 4 water purification unit (Elga Limited, UK). All reagents used were of analytical grade. A plate reader (Dynex Technologies, UK) was used for spectrophotometric determinations in protein or CA assays.

Methods Protein and colominic acid determination Quantitative estimation of polysialic acids (as sialic acid) with the resorcinol reagent was carried out by the resorcinol method [Svennerholm, 1957] as described elsewhere [Gregoriadis et al., 1993; Fernandes and Gregoriadis, 1996,1997]. Fab (protein) was measured by the BCA colorimetric method [Brown et. al., 1989].

Reference Example Table 1. Yields of covalent PSA-protein conjugates generated by reductive amination with sodium cyanoborohydride using the natural form of polysialic acid (colominic acid, CA) from E. coli, via its weakly reactive reducing end. CA = colominic acid; CAO = oxidised colominic acid as in Fernandes and Gregoriadis, 1996; Jain et al 2003. Sodium cyanoborohydride (NaCNBH3) was used at a concentration of 4mg ml-1.

The molar ratios in column 1 are the ratio of starting CA (0) to protein. n=3, standard deviation.

Degree of modification with CA molar Preparation ratio (CA: protein) Catalase+CAO+NaCNBH3 (10: 1) 0. 770. 16 Catalase+CAO+NaCNBH3 50 : 1) 2. 590. 08 Catalase+CA+NaCNBH3 (50: 1) 0. 550. 05 Catalase+CA (50: 1) 0. 65+0. 04 Insulin+CAO+NaCNBH3 (25: 1) 1. 60 + 0. 16 Insulin +CAO+NaCNBH3 (50: 1) 1. 65 _ 0. 14 Insulin +CAO+NaCNBH3 (100: 1) 1. 74 + 0. 12 Insulin +CA+NaCNBH3 (25: 1) 0. 2 + 0. 02 Insulin +CA+NaCNBH3 (50: 1) 0. 21 _ 0. 04 Insulin +CA+NaCNBH3 (100: 1) 0. 24 + 0. 06 Example 1 Preparation of inert polysialic acid : 1.1 Preparation of reduced colominic acid (R-CA) (Fig. 3) Freshly prepared sodium borohydride (0.15 M in 0.1 M NaOH) was added to CA (22.7 kDa; 100mg/ml, 1 ml) and the pH adjusted to 8-8. 5. The mixture was stirred at 20°C for to 2h in the dark and the pH reduced to 7.0.

The reduced colominic acid (R-CA) was dialysed (3.5 kDa molecular weight cut dialysis tubing) against 0. 01 % ammonium carbonate buffer pH (7) at 4°C and concentrated by ultrafiltration over a 3.5 kDa membrane to-500 ul. The sample was the lyophilized and stored at 4°C until further required.

1.2 Determination of the oxidation state of CA and derivatives Qualitative estimation of the degree of colominic acid oxidation was carried out with 2,4 dinitrophenylhydrazine (2,4-DNPH), which yields sparingly soluble 2, 4 dinitrophenyl-hydrazones on interaction with carbonyl compounds. Non-oxidised (CA), oxidised (CAO), and reduced (R-CA) (5mg each), were added to the 2,4-DNPH reagent (1. 0ml), the solutions were shaken and then allowed to stand at room temperature. Oxidation was quantified according to a known method [Park and Johnson, 1949]. This is based on the reduction of ferricyanide ions in alkaline solution to ferric

ferrocyanide (Persian blue), which is then measured at 630nm. In this instance, glucose was used as a standard.

1.3 R-CA (non-) conjugates with growth hormone (R-CA-GH) An XK50 column (Amersham Biosciences, UK) was packed with 900 ml Sepharose Q FF (Amersham Biosciences) and equilibrated with 3 column volumes of wash buffer (20mM triethanolamine ; pH 7.4) at a flow rate of 50ml/min. CA (25 grams in 200 ml wash buffer) was loaded on column at 50 ml per minute via a syringe port. This was followed by washing the column with 1.5 column volumes (1350ml) of washing buffer.

The bound CA was eluted with 1.5 column volumes of different elution buffers (Triethanolamine buffer, 20 mM pH 7.4, with OmM to 475mM NaCI in 25 mM NaCI steps) and finally with 1000mM NaCI in the same buffer to remove all residual CA and other residues (if any).

The samples were concentrated to 20 ml by high pressure ultra filtration over a 5kDa membrane (Vivascience, UK). These samples were buffer exchanged into deionised water by repeated ultra filtration at 4°C. The samples were analysed for average molecular weight and other parameters by GPC and native PAGE (stained with alcian blue).

Growth hormone was dissolved in 0. 15 M PBS (pH 7.4) and mixed with reduced CA; as a positive control oxidised CA fractions were conjugated to Fab. The CAO was the oxidised version of the CA as supplied. The N- CAO was an oxidised product of the narrow molecular weight fraction of CA separated as described having an average molecular weight of 29.0. The CA compounds were individually added to protein (2mg) in a CA: protein molar ratios (25: 1), sodium cyanoborohydride was added to a final concentration of 4 mg/ml. The reaction mixtures were sealed and stirred magnetically for 48h at 352°C. The mixtures were then subjected to ammonium sulphate ((NH4) 2SO4) precipitation by adding the salt slowly whilst continuously stirring, to achieve 70% w/v saturation, stirred for 1 h at 4°C, then spun (5000xg) for 15 min and the pellets resuspended in a saturated solution of (NH4) 2SO4 and spun again for 15 min (5000xg). The precipitates recovered

were redissolved in 1 ml PBS pH 7.4 and dialysed extensively (24 h) at 4°C against the same buffer.

Shaking was kept to a minimum to avoid concomitant denaturation of the protein. Protein product was characterised by SDS-PAGE.

Results The integrity of the internal alpha-2, 8 linked Neu5Ac residues after borohydride treatment was analysed by and native-PAGE (Fig 6) and the gel obtained for the oxidised (CAO) produced as in the Reference Example, reduced (R-CA) materials were compared with that of native CA. It was found that oxidized (CAO), reduced (R-CA) and native CA exhibit almost identical elution profiles, with no evidence that the reduction steps give rise to significant fragmentation of the polymer chain. No aggregation was seen in process control with R-CA as compared to native protein.

Quantitative measurement of the oxidation state of CA was performed by ferricyanide ion reduction in alkaline solution to ferrocyanide (Prussian Blue) [Park and Johnson, 1949] using glucose as a standard [results are shown in table 2]. Table 2 shows that the oxidized colominic acid was found to have a greater than stoichiometric (>100%) amount of reducing agent, i. e.

111.9 mol % apparent aldehyde content comprising the combined reducing power of the reducing end hemiketal and the introduced aldehyde (at the other end). No reactivity was seen in R-CA demonstrating that the neutralisation of hemiketal of CA had been successfully accomplished by borohydride reduction.

The results of quantitative assay of colominic acid intermediates (Table 2) were consistent with the results of qualitative tests performed with 2,4 dinitrophenylhydrazine which gave a faint yellow precipitate with the native CA, and intense orange colour with the aldehyde containing forms of the polymer, resulting in an intense orange precipitate after ten minutes of reaction at room temperature, no colour change or precipitate was seen for the reduced CA.

The SDS-page of protein-polymer conjugation shows that there is no conjugation and reduction in the aggregation of protein in presence of R-CA as compared to other process control, i. e. protein in PBS under similar reaction conditions shows some aggregation in channel 3 (Fig 7). Channels 4 and 5 of Figure 7 are positive controls with 22.7 (CAO) 29.0 kDa (N-CAO) conjugated to Fab.

Table 2: Degree of activity (oxidation) of various colominic acid intermediates in the reduced reaction scheme using glucose as a standard (100%, 1 mole of aldehyde per mole of glucose). CA species Degree of oxidation (%) colominic acid (CA) 15. 8 colominic acid-oxidised 111. 9 (CAO) colominic acid-reduced 0 ; Not detectable (R-CA) Example 2 Preparation of colominic oxidised reduced (CAOR) 2.1 Activation of colominic acid Freshly prepared 0.1 M sodium metaperiodate (NalO4) solution was mixed with CA (1 00mg CA/ml Na'04) at 20°C and the reaction mixture was stirred magnetically for 15 min in the dark. A two-fold volume of ethylene glycol was then added to the reaction mixture to expend excess Nal04and the mixture left to stir at 20°C for a further 30 min. The oxidised colominic acid was dialysed (3.5KDa molecular weight cut off dialysis tubing) extensively (24 h) against a 0. 01 % ammonium carbonate buffer (pH 7.4) at 4°C. Ultrafiltration (over molecular weight cut off 3.5kDa) was used to concentrate the CAO solution from the dialysis tubing. Following concentration to required volume, the filtrate was lyophilized and stored at - 40°C until further use.

2.2 Reduction of colominic acid This was done as reported in Example 1.1.

2.3 Determination of the oxidation state of CA and derivatives This was done as reported in Example 1.2.

2.4 Gel Permeation Chromatography Colominic acid samples (CA, CAO and CAOR) were dissolved in NaN03 (0.2M), CH3CN (10%; 5mg/ml) and were chromatographed on over 2x GMPWXL columns with detection by refractive index (GPC system: VE1121 GPC solvent pump, VE3580 RI detector and colation with Trisec 3 software, Viscotek Europe Ltd. Samples (5mg/ml) were filtered over 0. 45, um nylon membrane and run at 0.7cm/min with 0.2M NaNO3 and CH3CN (10%) as the mobile phase.

Results Colominic acid, a polysialic acid, is a linear alpha-2, 8-linked homopolymer of N-acetylneuraminic acid (Neu5Ac) residues. Periodate, however, is a powerful oxidizing agent and although selective [Fleury and Lange, 1932] for carbohydrates containing hydroxyl groups on adjacent carbon atoms, it can cause time-dependent cleavage to the internal Neu5Ac residues. Therefore, in the present work exposure of colominic acids to oxidation was limited to 15 min using 100 mM periodate at room temperature [Lifely et. al., 1981]. Moreover, as periodate decomposes on exposure to light to produce more reactive species, reaction mixtures were kept in the dark. The integrity of the internal alpha-2, 8 linked Neu5Ac residues post periodate and borohydride treatment was analysed by gel permeation chromatography and the chromatographs obtained for the oxidised (CAO) and oxidised reduced (CAOR) materials were compared with that of native CA. It was found (Fig. 8) that oxidized (15 minutes) (CAO), and reduced (CAOR), and native CA exhibit almost identical elution profiles, with no evidence that the successive oxidation and reduction steps give rise to significant fragmentation of the polymer chain. The small peaks are indicative of buffer salts.

Quantitative measurement of the oxidation state of CA was performed by ferricyanide ion reduction in alkaline solution to ferrocyanide (Prussian Blue) [Park and Johnson, 1949] using glucose as a standard [results are shown in table 3]. Table 3 shows that the oxidized colominic acid was found to have a greater than stoichiometric (>100%) amount of reducing agent, i. e.

112 mol % of apparent aldehyde content comprising the combined reducing power of the reducing end hemiketal and the introduced aldehyde (at the other end). No reactivity was seen in CAOR and R-CA demonstrating that the neutralisation of both the aldehyde and the hemi ketal of CAO had been successfully accomplished by borohydride reduction.

The results of quantitative assay of colominic acid intermediates in the double oxidation process using ferricyanide (Table 3) were consistent with the results of qualitative tests performed with 2,4 dinitrophenylhydrazine which gave a faint yellow precipitate with the native CA, and intense orange colour with the aldehyde containing forms of the polymer, resulting in an intense orange precipitate after ten minutes of reaction at room temperature.

Table 3: Degree of activation of various colominic acid intermediates in using glucose as a standard (100%, 1 mole of aldehyde per mole of glucose). n=3 + s. d. CA species Degree of oxidation colominic acid (CA) 16. 1 + 0. 63 colominic acid-oxidised (CAO) 112.03 4.97 colominic acid-reduced (CAOR) 0; Not detectable Example 3 Comparison of non-conjugated mixed Fab/CA with Fab-colominic acid conjugates Fab was dissolved in 0.15 M PBS (pH 7.4) and covalently linked to different colominic acids (CA, CAO, and CAOR) via reductive amination in the presence of sodium cyanoborohydride (NaCNBH3). Colominic acid from each step of the synthesis (starting material and products of each of

Examples 2.1 and 2.2) together with Fab in a CA: Fab molar ratios (100: 1) were reacted in 0.15 M PBS (pH 7.4 ; 2ml) containing sodium cyanoborohydride (4mg/ml) in sealed vessels with magnetic stirring at 352°C in an oven. The mixtures was then subjected to ammonium sulphate ((NH4) 2SO4) precipitation by adding the salt slowly whilst continuously stirring, to achieve 70% w/v saturation. The samples, stirred for 1 h at 4°C, were centrifuged for 15 min (5000xg) and the pellets containing polysialylated Fab suspended in a saturated solution of (NH4) 2SO4 and centrifuged again for 15 min (5000xg). The precipitates recovered were redissolved in 1 ml 0. 15M Na phosphate buffer supplemented with 0.9% NaCI (pH 7.4 ; PBS) and dialysed extensively (24 h) at 4°C against the same PBS. The dialysates were then assayed for sialic acid and Fab content and the conjugation yield was expressed in terms of CA: Fab molar ratio. Controls included subjecting the native protein to the conjugation procedure in the presence of non-oxidised CA or in the absence of CA, under the conditions described. Stirring was kept to a minimum to avoid concomitant denaturation of the protein. Polysialylated Fab was further characterised by size exclusion chromatography, ion exchange chromatography and SDS-PAGE (Fig 9).

In previous experiments [Jain et. al., 2003; Gregoriadis, 2001] with other proteins it was found that optimal CA: Fab (derived from sheep IgG) molar conjugation yields required a temperature of 352°C in 0.15 M PBS buffer at pH 6-9 for 48h. The imine (Schiff base) species formed under these conditions between the polymer aldehyde and protein was successfully reduced with NaCNBH3 to form a stable secondary amine [Fernandes and Gregoriadis, 1996; 1997]. Exposure of protein to periodate-oxidised natural CA generates a metastable Schiff's base CA-protein adduct (as reported for the polysialylation of catalase [Fernandes and Gregoriadis, 1996]. Likewise, in the reaction of oxidised forms of CA with Fab, we first created a metastable Schiff's base adduct, by incubation of the oxidised polymer with Fab for 48 h at 37°C which was then consolidated by selective reduction (reductive amination) with NaCNBH3 (which reduces the Schiff's base imine structure,

but not the aldehyde moiety of the polymer). In order to characterise the protein reactivity of the various CA intermediates. Fab was subjected to reductive amination in the presence of natural CA (CA), oxidized CA (CAO) and oxidised-reduced CA (CAOR). For these studies 22.7kDa PSA was used, at CA: Fab molar ratio of (100: 1). After 48h of incubation in the presence of NaCNBH3, CA-Fab conjugates were isolated from reaction mixtures by precipitation with ammonium sulphate (as described in the "Examples") and the results expressed in terms of CA: Fab molar ratios in the resulting conjugates (Table 4).

Table 4: Synthesis Fab (protein) colominic acid compounds. CA species tested Molar conjugation ratio (CA: Fab) attained colominic acid (CA) 0.21 : 1 (weakly reactive) colominic acid-oxidised 2. 81 : 1 (CAO) (highly reactive) colominic acid-oxidised-not detectable reduced (CAOR) (reactivity destroyed) It is evident from Table 4 that when natural, non-oxidized CA (in the presence of cyanoborohydride) was used, a significant but low level of conjugation was observed (resulting in a 0.21 : 1, CA: Fab molar ratio) via reaction with the hemiacetal group of CA at its reducing end.

Formation of the CA-Fab conjugates was further confirmed by the co- precipitation of the two moieties on addition of (NH4) 2SO4 (CA as such does not precipitate in the presence of the salt). Evidence of conjugation was also confirmed by ion exchange chromatography (IEC, not shown) and polyacrylamide gel electrophoresis (SDS-PAGE ; Fig. 9).

For ion-exchange chromatography, polysialylated Fab obtained by (NH4) 2SO4 precipitation was redissolved in sodium phosphate buffer (50mM, pH 4.4) and subjected to IEC using a Sepharose SP HiTrap column (cation exchange). In contrast with results indicating complete resolution of CA (in the wash) and Fab (in eluted fractions), both CA and Fab from the 48h reaction samples co-eluted in the wash fractions, demonstrating the presence of CA-Fab conjugate.

Fig. 9 shows the results of the SDS-PAGE of the antibody Fab reaction products produced as described above. These data confirm that the molecular weight distributions of the products of the incubation of CAOR with Fab and the Fab starting material are very similar. It is also evident from Fig.

9 that when Fab conjugates were prepared from CAO (i. e. periodate oxidised natural CA), showed a wide molecular weight distribution, elevated from the molecular weight of underivatised Fab control. This is consistent with the known polydispersity of the natural polymer reported in our previous published works. It is also evident from Fig 9 that only trace amounts of underivatised Fab remained in each CAO-Fab conjugate sample.

Differential Scanning Calorimetry (DSC) The DSC of CAOR was performed by setting the cycle for a 10°C/min heating rate from 0° to 300°C. The results are shown in Figure 10.

Results The thermal degradation of CAOR was found to occur at more than 200°C. This high degradation temperature indicates good thermal stability of CA under normal processing temperature used for pharmaceutical, food and cosmetic industries.

References Bendele, A. , Seely, J. , Richey, C., Sennello, G. , Shopp, G., Renal tubular vacuolation in animals treated with polyethylene-glycol conjugated proteins, Toxicological sciences, 42 (1998) 152-157.

Beranova, M. , Wasserbauer, R. , Vancurova, D. , Stifter, M. , Ocenaskova, J. , Mora, M., Biomaterials, 11 (2000) 521-524.

Cho, J. W. and Troy, F. A., Polysialic acid engineering: Synthesis of polysialylated neoglycosphingolipid by using the polytransferase from neuroinvasive E. colí K1, Proceedings of National Academic Sciences, USA, 91 (1994) 11427-11431.

Convers, C. D. , Lejeune, L. , Shum, K. , Gilbert, C. , Shorr, R. G. L, Physiological effect of polyethylene glycol conjugation on stroma-free bovine hemoglobin in the conscious dog after partial exchange transfusion, Artificial organ, 21 (1997) 369-378.

Dyer, J. R. , Use of periodate oxidation in biochemical analysis, Methods of Biochemical Analysis, 3 (1956) 111-152.

Fernandes, A. I., Gregoriadis, G., Polysialylated asparaginase: preparation, activity and pharmacokinetics, Biochimica et Biophysica Acta, 1341 (1997) 26-34.

Fernandes, A. I., Gregoriadis, G. , Synthesis, characterization and properties of polysialylated catalase, Biochimica et Biophysica Acta, 1293 (1996) 92- 96.

Fernandes, A. I., Gregoriadis, G. , The effect of polysialylation on the immunogenicity and antigenicity of asparaginase: implications in its pharmacokinetics, International Journal of Pharmaceutics, 217 (2001) 215- 224.

Fleury, P. , Lange, J. , Sur l'oxydation des acides alcools et des sucres par I'acid periodique, Comptes Rendus Academic Sciences, 195 (1932) 1395- 1397.

Gregoriadis, G. , Drug and vaccine delivery systems, in: PharmaTech, World Markets Research Centre Limited, London (2001) 172-176.

Gregoriadis, G. , Fernandes, A. , McCormack, B., Mital, M. , Zhang, X, Polysialic acids: Potential for long circulating drug, protein, liposome and other microparticle constructs, in Gregoriadis, G and McCormack, B (Eds), Targeting of Drugs, Stealth Therapeutic Systems, Plenum Press, New York (1998) 193-205.

Gregoriadis, G. , Fernandes, A., Mital, M. , McCormack, B., Polysialic acids:

potential in improving the stability and pharmacokinetics of proteins and other therapeutics, Cellular and Molecular Life Sciences, 57 (2000) 1964- 1969.

Gregoriadis, G., McCormack, B. , Wang, Z. , Lifely, R., Polysialic acids: potential in drug delivery, FEBS Letters, 315 (1993) 271-276.

Hreczuk-Hirst, D. , Jain, S. , Genkin, D. , Laing, P. , Gregoriadis, G., Preparation and properties of polysialylated interferon-a-2b, AAPS Annual Meeting, 2002, Toronto, Canada, M1056 Jain, S. , Hirst, D. H. , McCormack, B., Mital, M. , Epenetos, A. , Laing, P., Gregoriadis, G., Polysialylated insulin : synthesis, characterization and biological activity in vivo, Biochemica et. Biophysica Acta, 1622 (2003) 42- 49.

Jain, S. , Hirst, D. H. , Laing, P. , Gregoriadis, G., Polysialylation : The natural way to improve the stability and pharmacokinetics of protein and peptide drugs, Drug Delivery Systems and Sciences, 4 (2) (2004) 3-9.

Jennings, H. J. , Lugowski, C., Immunogenicity of groups A, B, and C meningococal polysaccharide tetanus toxoid conjugates, Journal of Immunology, 127 (1981) 1011-1018.

Lifely, R. , Gilhert, A. S. , Moreno, C. C., Sialic acid polysaccharide. antigen of Neisseria meningitidis and Escherichia coli : esterification between adjacent residues, Carbohydrate Research, 94 (1981) 193-203.

Mital, M. , Polysialic acids: a role for optimization of peptide and protein therapeutics, Ph. D. Thesis, University of London, 2003 (submitted).

Muflenhoff, M. , Ectehardt, M. , Gerardy-Schohn, R., Polysialic acid : three- dimensional structure, biosynthesis and function, Current opinions in Structural Biology, 8 (1998) 558-564.

Park, J. T. , Johnson, M. J. , A submicrodetermination of glucose, Journal of Biological Chemistry, 181 (1949) 149-151.

Roth, J. , Rutishauser, U. , Troy, F. A. (Eds.), Polysialic acid : from microbes to man, Birkhauser Verlag, Basel, Advances in Life Sciences, 1993.

Rutishauser, U., Polysialic acid as regulator of cell interactions in: R. U.

Morgoles andR. K. Margalis (eds. ), Neurobiology of Glycoconjugates, pp 367- 382, Plenum Press, New York, 1989.

Shriner, R. L., Fuson, R. D. C. , Curtin, D. Y., Morill, T. C. , The Systematic Identification of Organic Compounds, 6th ed., Wiley, New York, 1980.

Svennerholm, L. , Quantitative estimation of sialic acid ll : A colorimetric resorcinol-hydrochloric acid method, Biochimca et Biophysica Acta, 24 (1957) 604-611.

Troy, F. A. Polysialylation of neural cell adhesion molecules, Trends in Glycoscience and Glycotechnology, 2 (1990) 430-449.

Troy, F. A., Polysialylation : From bacteria to brain, Glycobiology, 2 (1992) 1- 23.