WO2012158932A2 - REMOVAL OF α-GAL EPITOPES FROM THERAPEUTIC GLYCOPROTEINS FOR AVOIDING INTERACTION WITH THE ANTI-GAL ANTIBODY - Google Patents

Abstract

The invention provides methods comprising applying a therapeutic glycoprotein preparation having one or more α-gal epitopes having a terminal Gala α1-3Gal or terminal α-galactosyls to a column having covalently or non-covalently coupled α-galactosidase in porous beads or on particulate material in order to destroy the α-gal epitope by enzymatic removal of the terminal galactose by the α-galactosidase. The invention further provides methods comprising destruction of one or more α-gal epitopes having a terminal Gala α1-3Gal or terminal a- galactosyls on therapeutic glycoproteins by mixing the glycoproteins with tagged α- galactosidase in order to destroy the α-gal epitope by enzymatic removal of the terminal galactose and to subsequently remove the α-galactosidase.

Description

REMOVAL OF a-GAL EPITOPES FROM THERAPEUTIC GLYCOPROTEINS FOR AVOIDING INTERACTION WITH THE ANTI-GAL ANTIBODY

This application claims priority under 35 U.S.C. § 119(e) to co-pending U.S. Provisional Application Serial No. 61/519,263, filed on May 18, 2011, herein incorporated by reference in its entirety.

FIELD OF INVENTION

The present invention is related to the field of therapeutic glycoproteins. In particular, the present invention provides compositions and methods for avoiding the binding of the natural anti-Gal antibody to monoclonal antibodies, and to natural or recombinant glycoproteins or to other molecules that have cc-gal epitopes (carbohydrate chains with terminal Galal-3Gal) and which are injected or infused for therapeutic purpose into humans or other vertebrates producing the anti-Gal antibody. The prevention of this binding eliminates the risk of complement activation, allergic reactions and accelerated removal of the injected glycoproteins, which otherwise may occur if the anti-Gal antibody binds to a-gal epitopes on said glycoproteins.

BACKGROUND OF THE INVENTION

Therapeutic glycoproteins which are introduced into the blood stream or into various tissues by injection or infusion, range from a variety of monoclonal antibodies, to recombinant enzymes, hormones and various glycoproteins with biological activities. When produced in vertebrate cells, the asparagines (Asn or N)-linked carbohydrate chains of these glycoproteins are synthesized on asparagines (N) which are part of an amino acids sequence N-X-S/T (glycosylation sites with the sequence of asparagine- any amino acid- serine or threonine). If these carbohydrate chains carry a-gal epitopes (Gakxl-3Gaipi-4GlcNAc-R or Galal-3Gaipi- 3GlcNAc-R), such therapeutic glycoproteins may be greatly affected by the natural anti-Gal antibody, and thus, may affect the recipient of the glycoprotein. Anti-Gal is the only natural antibody known to be produced in large amounts in all humans who are not severely immunocompromized. Anti-Gal is produced in humans as -1% of IgG (Galili et al, J Exp Med 160: 1519, 1984), >1% of IgM, and is readily detected in various body secretions as an IgA antibody (Hamadeh et al., Clin Diagnos Lab Immunol 2: 125, 1995). Recently, anti-Gal was shown to be produced in a proportion of the population as IgE antibody that mediates allergic reactions (Chung et al., New Engl J Med 358: 1109, 2008). In was further shown that one of the causes which may lead to production of anti-Gal IgE antibody is a bite of the tick Ambyomma americanum (Commins et al J Allergy Clin Immunol 127: 1286, 2011). Anti-Gal interacts specifically with a-gal epitopes (Galili et al., J Exp Med. 162:573, 1985; Galili et al., J Exp Med 165:693, 1987). Whereas anti-Gal is naturally produced in humans, apes and Old World monkeys (monkeys of Asia and Africa) (Galili et al, Proc Natl Acad Sci (USA), 84:1369, 1987), the a-gal epitopes is synthesized by cells in nonprimate mammals, prosimians and New World monkeys (monkeys of Central and South America), since the latter group of mammals has active al,3galactosyltranferase ( l,3GT), the enzyme that synthesizes within the Golgi apparatus a-gal epitopes (Galili et al., Proc Natl Acad Sci (USA), 84: 1369, 1987; Galili et al., J Biol Chem, 263: 17755, 1988). Anti-Gal readily interacts in vivo with a-gal epitopes as indicated in xenotransplantation studies. Transplantation of pig organs or cells (i.e. cells expressing a-gal epitopes) into Old World monkeys or humans results in anti-Gal binding to these epitopes and activation of complement which lyses the xenograft cells, resulting in rapid (hyperacute) rejection (Galili, Immunol Today, 14:480, 1993; and Collins et al., J Immunol, 154:5500, 1995). A similar interaction may occur between anti-Gal and a-gal epitopes on therapeutic glycoproteins injected or infused into humans. This interaction may lead to at least three apparent detrimental outcomes: 1. Activation of the complement system in the serum resulting in the production of complement cleavage factors such as C5a and C3a which, if produced in high enough concentration, may cause systemic inflammatory response syndrome (SIRS) that has symptoms similar to sepsis. 2. If the treated patient has anti-Gal IgE antibodies bound to mast cells, the binding of the a-gal epitopes on the therapeutic glycoprotein to these IgE antibodies will result in cross-linking of the IgE molecules on the mast cells, thus inducing the activation of the mast cells and induction of an allergic reaction, which may ultimately cause an anaphylactic shock. 3. The formation of immune complexes between anti-Gal and the injected glycoprotein may induce early removal of the protein from the circulation by the reticuloendothelial system, as a result of the binding of Fc portion of the complexed anti-Gal with Fey receptors (FcyR) on macrophages of the reticuloendothelial system. Therefore, prevention of anti-Gal binding to therapeutic glycoproteins is desirable in patients treated with such glycoproteins. SUMMARY OF THE INVENTION

The present invention is related to the field of therapeutic glycoproteins. In particular, the present invention provides compositions and methods for processing glycoprotein molecules with linked a-gal epitopes for elimination of the a-gal epitopes from these molecules thereby preventing the in vivo binding of the natural anti-Gal antibody to the a-gal epitopes. The present invention provides methods for prevention of allergic reactions, prevention of complement activation and prevention of accelerated removal of the injected therapeutic glycoprotein, by avoiding in vivo anti-Gal binding to these therapeutic glycoproteins. This is achieved by the use of an enzyme that cleaves (hydrolyses) the terminal galactose from the a-gal epitope thereby it destroys a-gal epitopes. In one embodiment, the enzyme, referred to as a-galactosidase, is used as a solid phase in a column of porous beads through which the solution of the therapeutic glycoprotein is passed for the destruction of the a-gal epitopes. The enzyme a-galactosidase may be isolated from a natural source or in a recombinant form. In another embodiment the enzyme is used in a soluble form which is mixed with the therapeutic glycoprotein. After the destruction of the a-gal epitope, the enzyme may be removed by its selective binding to a matrix such as, but not limited to, an agarose column with a ligand that specifically removes the enzyme from the solution mixture.

Thus, in one embodiment, the invention provides a method, comprising applying a therapeutic glycoprotein preparation having one or more a-gal epitopes having a terminal Galal-3Gal or terminal a-galactosyls to a column having covalently or non-covalently coupled α-galactosidase in porous beads or on particulate material in order to destroy the a-gal epitope by enzymatic removal of the terminal galactose by the α-galactosidase. In one embodiment, the therapeutic glycoproteins are monoclonal antibodies. In another embodiment, the therapeutic glycoproteins are recombinant glycoproteins. In a further embodiment, the therapeutic glycoproteins are natural glycoproteins. In another embodiment, the α-galactosidase is obtain from a natural source including, but not limited to microbial source, fungal source, or plant source. In yet another embodiment, the α-galactosidase is a recombinant a-galactosidase produced in a prokaryotic gene expression system or in a eukaryotic gene expression system or in a transgenic animal. The invention also provides a method, comprising destruction of one or more a-gal epitopes having a terminal Galal-3Gal or terminal a-galactosyls on therapeutic glycoproteins by mixing the glycoproteins with tagged α-galactosidase in order to destroy the a-gal epitope by enzymatic removal of the terminal galactose and to subsequently remove the α-galactosidase. In one embodiment, the tagged α-galactosidase is a recombinant α-galactosidase produced in a prokaryotic gene expression system or an eukaryotic gene expression system or in a transgenic animal and the enzyme is tagged with an oligopeptide, polypeptide, protein or other tag that enables the removal of the α-galactosidase from the reaction mixture by passing the mixture through a column of porous beads or with particulate material carrying linked ligand to the tag. In another embodiment, the α-galactosidase is tagged with an oligopeptide of 6 histidines (Hise), or tagged with fusion proteins as protein A, Fc portion of an immunoglobulin molecule, glutathione S-transferase, or tagged with biotin, and the corresponding columns with matrix for removal of the tagged α-galactosidase including nickel-sepharose column, IgG coupled column, protein A coupled column, glutathione coupled column, or avidin coupled column, respectively. In a further embodiment, the α-galactosidase is not tagged.

BRIEF DESCRIPTION OF THE DRAWINGS

The following are illustrations of the present invention and are not intended as limiting in any manner.

Figure 1 illustrates schematically a-gal epitopes on N-linked carbohydrate chains of therapeutic glycoproteins and the destruction of these epitopes by α-galactosidase. The a-gal epitopes are marked with rectangles of broken lines. When injected or infused into humans, anti- Gal binds to a-gal epitopes (left glycoprotein). The a-gal epitopes can be destroyed by the enzyme α-galactosidase. This enzyme cleaves (i.e. hydrolyzes) the linkage between the terminal galactose and the penultimate galactose in the a-gal epitope (Galal-3Gaipi-4GlcNAc-R), resulting in the removal of the terminal galactose. The remaining terminal carbohydrate structure of Gai i-4GlcNAc-R on the carbohydrate chains of the therapeutic glycoprotein (right glycoprotein) does not bind the anti-Gal antibody and is referred to as N-acetyllactosamine.

Figure 2 shows a schematic illustration of a porous bead within the column of a- galactosidase attached to a matrix of agarose beads (called also Sepharose beads). The column is also referred to as a-galactosidase column, a-galactosidase in linked to the matrix either covalently as in agarose beads activated with cyanogens bromide, or linked non-covalently. One non-limiting example of a non-covalently linked enzyme is recombinant α-galactosidase with a tag of a number of histidines (His₆ tag) that links the enzyme to nickel-agarose. The therapeutic glycoprotein passing through the column is subjected to the enzymatic activity of a- galactosidase linked to the column, resulting in the removal of terminal galactose from the carbohydrate chains of the glycoproteins and thus, destruction of the a-gal epitopes.

Figure 3 demonstrates the in vitro binding of the monoclonal anti-Gal antibody M86 (Galili et al., Transplantation 65: 1129, 1998) to a-gal epitopes of various glycoproteins prior (Figure 3A), and post treatment with recombinant a-galactosidase (10 Units/ml) (Figure 3B). The glycoproteins served as solid-phase antigens in an enzyme-linked immunosorbent assay (ELISA). The ELISA wells were coated with the ^g/ml glycoproteins in carbonate buffer pH 9.5 by standard methods known to those skilled in the art. Binding of the monoclonal anti-Gal M86 in Figures 3A and 3B was determined by the subsequent binding of goat anti-mouse IgM coupled to horseradish peroxidase (HRP) and color development with O-phenylene diamine (OPD). Binding of anti-Gal in human serum to untreated glycoproteins in ELISA is presented in Figure 3C, whereas binding of anti-Gal in human serum to glycoproteins treated with a- galactosidase is presented in Figure 3D. Binding of anti-Gal in human serum to a-gal epitopes was determined by the subsequent binding of rabbit anti-human IgG coupled to HRP and color development with OPD. Figure 3A provides a graph of the binding of monoclonal anti-Gal M86 to synthetic a-gal epitopes on bovine serum albumin (a-gal BSA) (·), bovine thyroglobulin (■), a-gal fetuin (A) or mouse laminin (♦), human thyroglobulin (□), human laminin (0), or original fetuin lacking a-gal epitopes (Δ). In Figure 3B binding to the various glycoproteins with a-gal epitopes was determined after the ELISA wells were incubated for 2h at room temperature with 10 Units/ml of recombinant α-galactosidase. Symbols as in Figure 3A. Figure 3C provides a graph of the binding of anti-Gal in human serum to the various glycoproteins with symbols as in Figure 3A. In Figure 3D the binding anti-Gal in human serum was evaluated after the assayed glycoproteins were treated with α-galactosidase. Note that monoclonal anti-Gal and human serum anti-Gal readily bind to the a-gal epitopes on a-gal BSA, bovine thyroglobulin, a-gal fetuin or mouse laminin. However, following incubation of these glycoproteins with a- galactosidase, the binding of the monoclonal or human anti-Gal greatly diminishes because of the enzymatic destruction of a-gal epitopes. The residual binding at high antibody concentrations (i.e. at low dilutions) in wills treated with a-galactosidase is the result of nonspecific binding of M86 monoclonal antibody or of polyclonal IgG antibody to the ELISA wells.

Figure 4 illustrates the activation of complement in human serum, by human anti-Gal binding to a-gal epitopes on glycoproteins, and prevention of this activation by a-galactosidase mediated destruction of a-gal epitopes on the glycoproteins. Figure 4A is a schematic of the readout system for complement activity involving the cytolysis of the anti-Gal producing hybridoma cells M86 (Galili et al, Transplantation 65: 1129, 1995). These cells have anti-Gal bound to their cell membranes because, in addition to production of a monoclonal anti-Gal, these cells express a-gal epitopes. Upon addition of complement, as in the addition of human serum to M86 cells, the complement is activated by the monoclonal anti-Gal bound to the a-gal epitopes on the M86 hybridoma cell membrane, resulting in complement mediated lysis (cytolysis) of the M86 cells. In contrast, if anti-Gal in the serum binds to a-gal epitopes on glycoproteins and activates serum complement as a result of this binding, the complement in the serum is consumed. If consumed, the complement is not available for subsequent cytolysis of M86 cells upon addition of the cells to the serum containing the glycoproteins. Figure 4B provides a graph showing the cytolysis of M86 cells by complement after incubation at 37°C for lh with human serum at various dilutions of the serum (·). In contrast, heat inactivation of serum (by pre-incubation for 30min at 6°C) results in inactivation of the complement within the human serum and hence no cytolysis of the M86 cells is observed at any serum dilution (O). The cytolysis of M86 cells by human serum is dependent only on complement activity. This is indicated by a similar cytolysis observed with normal rabbit serum which completely lacks anti- Gal antibodies (since rabbits, like all nonprimate mammals have a-gal epitopes on their cells) (■). Figure 4C provides a graph showing that the interaction of anti-Gal in human serum with a- gal epitopes on glycoproteins results in complement activation and consumption, as measured by a loss of serum cytolytic activity. Human serum was incubated for 2h at 37°C at serial two-fold dilutions with lOmg/ml a-gal BSA (O) that has 21 a-gal epitopes per molecule or with 5mg/ml a-gal fetuin (□) that has 9 a-gal epitopes per molecule. Subsequent addition of M86 cells to these suspensions for lh at 37°C indicated that the extent of cytolysis greatly decreased in comparison to cytolysis by complement in the corresponding human serum dilutions in Figure 4B. This decrease in cytolysis was the result of complement consumption following human serum anti-Gal binding to a-gal epitopes on a-gal BSA, or on a-gal fetuin. Pre-incubation of a- gal BSA or of a-gal fetuin with 10 Units/ml recombinant a-galactosidase (Zhu et al., Arch Biochem Biophys 324:65, 1995) resulted in destruction of a-gal epitopes on the glycoproteins and thus, prevented the subsequent interaction of the glycoprotein with human serum anti-Gal antibody and the activation and consumption complement. Therefore, incubation of human serum with α-galactosidase treated a-gal BSA (·) or of a-gal fetuin treated with a- galactosidase (■), did not significantly affect the cytolytic activity of the serum with M86 cells as presented in Figure 4B, since no significant complement activation occurred.

Figure 5 illustrates the destruction of a-gal epitopes on a glycoprotein with a-gal epitopes passed through an α-galactosidase column. 0.5 ml of the glycoproteins a-gal BSA (•)in saline, at a concentration of 100μg/ml, was passed through 1ml columns of a- galactosidase-agarose for a period of 30min at room temperature. The glycoprotein was collected in 2ml saline and plated in ELISA wells at serial two-fold dilutions in carbonate buffer pH 9.5 starting at 10μg/ml. In parallel, the original glycoprotein solution of a-gal BSA (O) was plated at similar two-fold dilutions, starting at 10μg/ml. After 2h incubation at 37°C and overnight incubation at 4°C, the ELISA plate was blocked with 1% BSA in PBS. The presence of a-gal epitopes on the glycoprotein attached to the wells was determined by incubation for 2h of the monoclonal anti-Gal antibody M86, followed by washes, addition of FfRP coupled goat anti- mouse IgM and color development with OPD. The passage through the α-galactosidase column resulted in complete elimination of a-gal epitopes from a-gal BSA, as indicated by the absence of anti-Gal M86 binding even at the highest concentration of the treated a-gal BSA attached to the wells.

Figure 6 illustrates human complement activation by a-gal BSA passed through an a- galactosidase column. Human serum was incubated for 2h at 37°C at serial two-fold dilutions with 100μ§Λ 1 α-gal BSA (O) or with 100μg ml a-gal BSA that was passed through an a- galactosidase column (·) and subsequently cytolysis of M86 cells was measured. Cytolysis of M86 cells by human serum in the absence of a-gal BSA is described as (■). Similar to Figure 4C, human serum incubated with a-gal BSA displays decreased cytolysis of M86 cells because of complement activation and consumption as a result of anti-Gal binding to oc-gal epitopes of oc- gal BSA. However, if the a-gal BSA was passed through an a-galactosidase column, consumption of complement was much lower, because a large proportion of the a-gal epitopes on a-gal BSA were destroyed by α-galactosidase within the column. These findings imply that the removal of the a-gal epitopes on therapeutic glycoproteins by passing these glycoproteins through an α-galactosidase column will prevent subsequent in vivo binding of anti-Gal to the glycoproteins and thus, avoid the adverse effects caused by this antibody binding to the therapeutic glycoproteins. DEFINITIONS

To facilitate an understanding of the present invention, a number of terms and phrases are defined below:

The term "therapeutic glycoproteins" as used herein, refers to any protein molecule that carries carbohydrate chains and is used for clinical treatment by infusion, injection intravenously or to any tissue such as intramuscular or intradermal. Therapeutic glycoproteins may originate from blood, tissue extracts or secretion of various animals, or may be produced in natural or recombinant form from various cells grown in cultures. Some non-limiting examples of therapeutic glycoproteins are: monoclonal antibodies, recombinant or natural hormones, recombinant growth factors and cytokines.

The term "a-gal epitope" as used herein, refers to any molecule, or part of a molecule, with a terminal structure comprising Galal-3Gaipi -4GlcNAc-R, Galal -3Galpl-3GlcNAc-R, or any carbohydrate chain with terminal Galal-3Gal, or terminal a-galactosyl at the non-reducing end, and which are capable of binding natural or elicited anti-Gal antibodies.

The term "glycoproteins with a-gal epitopes" as used herein, refers to any protein with at least one carbohydrate chain which has one or more a-gal epitopes, or any oligopeptide or polypeptide and which has one or more a-gal epitopes.

The term "N-acetyllactosamine" as used herein refers to any molecule, or part of a molecule, with a terminal structure comprising Gai i-4GlcNAc-R, or Gal l-3GlcNAc-R. The term anti-Gal antibodies, also called anti- -gal-antibodies, or anti-a-galactosyl- antibodies or anti-galactose al-3galactose antibodies or anti-Galotl-3Gal antibodies and refers to naturally occurring or elicited antibodies that bind to a-gal epitopes.

The term a-galactosidase refer to a glycosidase enzyme that cleaves (i.e. hydrolyzes) terminal galactose units on carbohydrate chains that are linked in an a-anomeric linkage to the penultimate carbohydrate unit that may be galactose or any other carbohydrate. A natural source of a-galactosidase may include, but not limited to, green coffee beans, soy beans, or fungi. Recombinant α-galactosidase is produced by cloning an α-galactosidase gene from any of the natural sources, insertion of the cloned gene into an expression system of microbial or eukaryotic cell or transgenic animal and production of the enzyme in the expression system cells. The recombinant α-galactosidase may be produced in its native form (i.e. without any tag), or linked to a tag such as, but not limited to, (His)6, protein A, Fc portion of an immunoglobulin molecule, or glutathione S-transferase.

The term "α-galactosidase column" refers to a column of agarose beads, Sepharose beads Sephacryl beads, or any other matrix to which any α-galactosidase enzyme in linked covalently or non-covalently.

The term "al,3-galactosyltransferase," refers to an enzyme synthesizing a-gal epitopes. The enzyme is naturally expressed in most mammals with the exception of humans, apes and Old World monkeys. This enzyme is also referred to as "al,3GT". The carbohydrate structure produced by the enzyme is immunogenic in man and most healthy people have high titer natural anti a-gal antibodies, also referred to as anti-galactosea 1-3 galactose antibodies or anti-Gal antibodies. In some embodiments, the term "al,3GT" refers to a common marmoset gene (e.g., Callithrix jacchus - GENBANK Accession No. S71333) and its gene product, as well as its functional mammalian counterparts (e.g., other New World monkeys, prosimians and non- primate mammals, but not Old World monkeys, apes and humans). In other embodiments, the term "al,3GT" refers to mouse al,3GT (e.g., Mus musculus - nucleotides 445 to 1560 of GENBANK Accession No. NM_010283), bovine al,3GT (e.g., Bos taurus - GENBANK Accession No. NM_177511), feline al,3GT (e.g., Felis catus - GENBANK Accession No. NM_001009308), ovine al,3GT (e.g., Ovis aries - GENBANK Accession No. NM_001009764), rat al,3GT (e.g., Rattus norvegicus - GENBANK Accession No. NM_145674) and porcine al,3GT (e.g., S s scrofa - GENBANK Accession No. NM_213810). The a,l,3GT gene is also referred to as Ggtal gene.

The term "a-gal fetuin" refers to the bovine fetal serum glycoprotein fetuin on which the terminal carbohydrate epitope SA(sialic acid)-Gaipi-4GlcNAc-R was enzymatically replaced with a-gal epitopes (Gal l-3Galpl-4GlcNAc-R) (Chen et al., Glycobiology 11:577, 2001).

The term "anti-Gal binding epitope", as used herein, refers to any molecule or part of molecule that is capable of binding the anti-Gal antibody.

The terms "tag" refers here to an oligopeptide, polypeptide, protein, or any molecule that can be attached to a natural or recombinant a-galactosidase and which can bind to a corresponding molecule (ligand) on a column, thereby enabling the specific removal of a- galactosidase from solution.

The terms "patient" and "subject" refer to a mammal or an animal that is a candidate for receiving medical treatment.

The terms "saline" refers to physiologic salt solution of 0.9% NaCl in distilled water.

GENERAL DESCRIPTION OF THE INVENTION

The present invention relates to the field of therapeutic glycoproteins and describes methods and compositions for preventing adverse effects of therapeutic glycoproteins that have a-gal epitopes on their carbohydrate chains and thus, can interact in vivo with the natural anti- Gal antibody (left carbohydrate chain in Figure 1).

I. Anti-Gal Antibodies and a-Gai Epitopes

Anti-Gal is an abundant natural antibody in humans constituting -1% of serum immunoglobulins (Galili et al., J Exp Med, 160:1519, 1984). This antibody interacts specifically with the a-gal epitope (Galal-3Gai i-4GlcNAc-R or Galal-3Gaip i-3GlcNAc-R) on glycolipids and glycoproteins (Galili, Springer Semin Immunopathol, 15:155, 1993). Anti-Gal is produced throughout life as a result of antigenic stimulation by bacteria of the gastrointestinal tract (Galili et al., Infect Immun, 56:1730, 1988). The a-gal epitope is synthesized by the glycosylation enzyme al,3galactosyltransferase (al,3GT) and expressed in very large amounts on cells of non-primate mammals, prosimians and in New World monkeys (Galili et al., J Biol Chem, 263: 17755, 1988). The al,3GT gene was inactivated in ancestral Old World primates. Thus humans, apes, and Old World monkeys lack a-gal epitopes and produce high titer anti-Gal antibodies (Galili, Springer Semin Immunopathol, 15: 155, 1993). Anti-Gal antibodies bind avidly in vivo to -gal epitopes when administered to humans or Old World monkeys. This is particularly evident in the context of xenotransplantation, where the in vivo binding of anti-Gal to a-gal epitopes on transplanted pig heart or kidney is the main cause for the rapid rejection of such grafts in humans and Old World monkeys (Galili, Immunol Today, 14:480, 1993; Collins et al., J Immunol, 154:5500, 1995).

One of the main mechanisms mediating xenograft rejection is the activation of the complement cascade due to anti-Gal binding to a-gal epitopes on the endothelial cells of the xenograft. This results in the destruction of these endothelial cells by the activated complement molecules, causing collapse of the vascular bed and xenograft ischemia followed by rapid rejection of the xenograft (Collins et al., J Immunol, 154:5500, 1995). This in situ interaction of anti-Gal with a-gal epitopes may also occur when therapeutic glycoproteins carrying these epitopes are injected into patients. The in situ interaction of anti-Gal with a-gal epitopes on therapeutic glycoproteins may result in a number of adverse effects, some of which are detailed below.

II. Synthesis of cc-Gal Epitopes on Therapeutic Glycoproteins by the Glycosylation Enzyme l,3galactosyltransferase (al,3GT)

a-gal epitopes are synthesized on carbohydrate chains of glycoproteins and glycolipids in the Golgi apparatus of cells containing al,3GT by the following reaction:

Gal l-4GlcNAc-R + UDP-Gal ^l'^3GT _► Gala 1-3 Gal l -4GlcNAc-R + UDP

N-Acetyllactosamine uridine diphosphate- a-gal epitope

galactose

The number of a-gal epitopes differs between various glycoproteins and depends on the number of carbohydrate chains per protein that can be subjected to the enzymatic activity of l,3GT in the trans-Golgi apparatus in cells, and on the activity of al,3GT in comparison to other competing glycosyltransferases within the trans-Golgi compartment where al,3GT is active, such as sialyltransferase (Smith et al, J Biol chem. 265:6225, 1990). The various glycosyltransferases in the trans-Golgi apparatus compete with each other for "capping" the N- acetyllactosamine of the carbohydrate chains with the carbohydrates they transfer to the chain. Thus, if a given hybridoma is generated by fusion between a mouse myeloma fusion partner cells and lymphocytes, the monoclonal antibody produced by such hybridoma cells will have a-gal epitopes (Borrebaeck et al., Immunol Today 14:477, 1993). This is because the al,3GT encoded by the al,3GT gene of the myeloma fusion partner will synthesize a-gal epitopes on a proportion of the carbohydrate chains present on all monoclonal antibodies. Similarly, recombinant glycoproteins produced in cells that contain active al,3GT may have a-gal epitopes because the recombinant glycoprotein passing through the trans-Golgi compartment will be subjected to the activity of al,3GT. Non-limiting examples are recombinant human Factor VIII produced in baby hamster kidney cells (Hironaka et al., J Biol Chem 267:8012, 1992) and recombinant human Interferon β produced in mouse epithelial cells (Kagawa et al., J Biol Chem 263 : 17508, 1988). Analysis of the carbohydrate chains on these recombinant glycoprotein demonstrated presence of a-gal epitopes on their carbohydrate chains. Thus, any therapeutic glycoprotein produced in mammalian cells that contain active l,3GT may have a-gal epitopes. Similarly, recombinant glycoproteins produced in vivo, such as in the cells of mammary glands having active al,3GT and secreted in the milk of mammalian species including, but not limited to, cow, sheep or goat, may have a-gal epitopes synthesized by al,3GT in cells of the mammary glands. Indeed, a-gal epitopes were demonstrated as free oligosaccharides in ovine milk (Urashima et al., Biochim Biophys Acta 992:375, 1989), or bovine milk (Urashima et al., Biochim Biophys Acta 1073:225, 1991).

III. Adverse In Vivo effects of the Interaction between Anti-Gal Antibodies And a-Gal Epitopes on Therapeutic Glycoproteins

When injected into treated individuals, therapeutic glycoproteins that have a-gal epitopes, can bind via their a-gal epitopes to anti-Gal IgE antibodies on mast cells, induce degranulation of these cells resulting in an allergic response. This was demonstrated in patients treated with the monoclonal antibody to epidermal growth factor receptor named "cetuximab" (Chung et al. New Engl J Med 358: 1109, 2008). Glycoproteins with a-gal epitopes binding in vivo anti-Gal may also induce complement activation in the blood. If present in high enough concentration, the complement cleavage peptides C3a and C5a can induce a systemic response called "systemic inflammatory response syndrome" ("SIRS") which has adverse effects similar to septic shock, including disseminated intravascular coagulopathy "DIC" (Guo and Ward Recent Patents Anti-Infect Drug Disc 1:57, 2006; Ward, Focus on Complement 5:2, 2007). Anti- Gal bound to cc-gal epitopes on glycoproteins can further decrease the half life of the glycoprotein in the circulation because of accelerated removal of the immune complexes with anti-Gal by the reticuloendothelial system. This was demonstrated by the inverse correlation between the half life of monoclonal antibodies injected into the blood of patients and the number of oc-gal epitopes on these monoclonal antibodies (Borrebaeck et al., Immunol Today 14:477, 1993). The higher the number of these epitopes the shorter the half life of the antibody in the circulation.

IV. Elimination of a-Gal Epitopes on Glycoproteins by a-galactosidase

This invention teaches compositions and methods for elimination of oc-gal epitopes on therapeutic glycoproteins by the use of the enzyme α-galactosidase. As shown in Figure 1, this enzyme cleaves the terminal galactose by hydrolysis of the a 1,3 linkage between the terminal galactose of the a-gal epitope and the penultimate galactose in the following reaction.

a-galactosidase

Galal-3Galpl-4GlcNAc-R — ► Galpl-4GlcNAc-R + Gal

As a result of this enzymatic reaction, the carbohydrate chains of therapeutic glycoproteins will have the terminal structure Gaipi-4GlcNAc-R, or Gaipi-3GlcNAc-R, which is referred to as N- acetyllactosamine (Right structure in Figure 1). α-galactosidase may be isolated from natural sources such as, but not limited to, green coffee beans, soy beans, or fungi, α-galactosidase has been obtained also in a recombinant form. One non-limiting example of α-galactosidase that of the coffee bean α-galactosidase gene that was cloned and inserted into the genome of the yeast Pichia pastoris (Zhu et al., Arch Biochem Biophys 324:65, 1995). This yeast produces the recombinant α-galactosidase and secretes it into the yeast growth medium. The enzyme can subsequently purified from the growth medium by using on an ion exchange chromatography column as known to those skilled in the art.

The cloned a-galactosidase gene may be inserted into genomes of eukaryotic or prokaryotic cells comprising various gene expression systems and the recombinant α-galactosidase may be produced in its native form (i.e. without any tag), or linked to a tag such as, but not limited to,

(His)6, protein A, Fc portion of an immunoglobulin molecule, or glutathione S-transferase. The recombinant a-galactosidase may be subsequently produced in vitro and isolated from the growth medium, or in vivo in transgenic animals containing the α-galactosidase transgene and producing it in various cells. A non-limiting example is production of this recombinant enzyme by the cells of the mammary glands of animals transgenic for the α-galactosidase gene and isolation of the recombinant enzyme from the milk of such transgenic animals.

The present invention provides methods for elimination of a-gal epitopes from therapeutic glycoproteins by incubating them with the enzyme α-galactosidase as illustrated in Figure 1.

Incubation of glycoproteins carrying a-gal epitopes with α-galactosidase will result in the destruction of a-gal epitopes on the therapeutic glycoproteins and thus, will prevent subsequent in vivo binding of the natural anti-Gal antibody to the glycoproteins.

V. Prevention of Contamination of Enzymatically Treated Therapeutic Glycoproteins Preparations by a-galactosidase

Whereas a-gal epitopes on therapeutic glycoproteins may be destroyed by a-galactosidase, these glycoproteins have to be injected into treated individuals in a pure form, i.e. without contaminating α-galactosidase. The present invention provides methods and compositions for the prevention of contamination of therapeutic glycoprotein solution by α-galactosidase used for the destruction of a-gal epitopes. Two non-limiting methods are as follows:

1. Use of Solid-Phase α-galactosidase - This invention teaches a method for the destruction of a-gal epitopes on therapeutic glycoproteins by α-galactosidase that is linked to a matrix (i.e. solid-phase α-galactosidase) such as, but not limited to, porous agarose beads, Sepharose beads or Sephacryl beads. As illustrated in Figure 2, natural or recombinant α-galactosidase is covalently or non-covalently linked to the porous beads or any other matrix by a variety of methods known to those skilled in the art. Two non-limiting examples are the use of cyanogens bromide activated agarose beads for covalent linking of α-galactosidase to the beads, and the use of α-galactosidase tagged with (His)6 for the non-covalent linking of this enzyme to nickel- agarose beads (also referred to as nickel-Sepharose beads). A column is constructed from the beads containing solid-phase α-galactosidase. In order to remove a-gal epitopes from therapeutic glycoproteins, preparations of such glycoproteins in solutions are passed through the solid-phase α-galactosidase column. The flow rate through the column may vary according to the type of glycoprotein and its concentration. The α-galactosidase molecules bound to the matrix of the column cleave the terminal galactose units from the a-gal epitopes, resulting in glycoproteins with N-acetyllactosamines as terminal structures on their carbohydrate chains, instead of a-gal epitopes as illustrated in Figure 2.

2. Use of soluble α-galactosidase with a tag for removal of the enzyme from the therapeutic glycoprotein solution- This invention teaches an alternative method for preventing contamination by α-galactosidase of enzymatically treated therapeutic glycoprotein preparation, by the use of soluble α-galactosidase that can be removed from the mixture of glycoproteins and enzyme by passing the solution through a column capturing the enzyme. This can be achieved by several methods, including but not limited to, tagging the recombinant a-galacosidase with an oligopeptide, polypeptide, protein, or any molecule which has a corresponding ligand on the column. Several non-limiting examples are: 1. Tagging recombinant α-galactosidase with a His- His-His-His-His-His tag (i.e. an oligopeptide of 6 histidines (His)e) by genetic engineering methods known to anyone skilled in the art. The a-galactosidase-(His)6 recombinant enzyme is added to the therapeutic glycoprotein solution for the destruction of a-gal epitopes. After completion of the enzymatic reaction (described in Figure 1), the solution is passed through a nickel-sepharose which specifically binds the enzyme through the -(His)₆ tag. This results in elimination of a-galactosidase. 2. Tagging recombinant α-galactosidase with glutathione S- transferase enables subsequent removal of the enzyme with columns containing glutathione. 3. Tagging recombinant α-galactosidase with the protein A enables the subsequent removal of the enzyme in columns of IgG bound to beads.

EXPERIMENTAL The following examples are provided in order to demonstrate and further illustrate certain preferred embodiments and aspects of the present invention and are not to be construed as limiting the scope thereof.

In the experimental disclosure which follows, the following abbreviations apply: kDa (kilodalton); rec. (recombinant); N (normal); M (molar); mM (miliimolar); μΜ (micromolar); mol (moles); mmol (millimoles); mol (micromoles); nmol (nanomoles); pmol (picomoles); g (grams); mg (milligrams); μg (micrograms); ng (nanograms); 1 or L (liters); ml (milliliters); μΐ (microliters); cm (centimeters); mm (millimeters); μπι (micrometers); nm (nanometers); C (degrees Centigrade); al,3GT (al,3galactosyltransferase); APC (antigen presenting cell); BSA (bovine serum albumin); ELISA (enzyme linked immunosorbent assay); HRP (horseradish peroxidase) IFNy (interferon-γ); knockout (KO); OD (optical density); OPD (ortho phenylene diamine); PBS (phosphate buffered saline); rabbit RBC (rabbit red blood cells);

EXAMPLE 1: DESTRUCTION OF a-GAL EPITOPES ON GLYCOPROTEINS BY a- GALACTOSIDASE

The recombinant a-galactosidase originally cloned from the coffee beans cc-galactosidase gene (Zhu et al., Arch Biochem Biophys 324:65, 1995) is very effective in destruction of the a- gal epitopes on glycoproteins. This can be demonstrated by ELISA with various glycoproteins carrying a-gal epitopes, used as solid-phase antigens in ELISA wells. These glycoproteins included mouse laminin with 50 a-gal epitopes/molecule (Arumugham et al., Biochim Biophys Acta 883: 112, 1986), bovine thyroglobulin with 12 a-gal epitopes/molecule (Spiro and Bhoyroo J Biol Chem 259:9858, 1984), a-gal fetuin originating from the bovine serum fetuin, on which 9 a-gal epitopes were synthesized by the use of recombinant al,3GT (Chen et al., Glycobiology 11 :577, 2001), and a-gal BSA which is a bovine serum albumin molecule carrying 21 synthetic a-gal epitopes, each linked via a 14 carbon spacer to the protein molecule (Dextra, Reading, UK). Human thyroglobulin and human laminin were used as control glycoproteins lacking a-gal epitopes because humans have no al,3GT activity (Galili et al., J Biol Chem, 263: 17755, 1988). In addition, bovine fetuin was used as control lacking a-gal epitopes since it carries 3 N-linked carbohydrate chains, each with 3 branches that are capped with sialic acid (SA) as the terminal epitope SA-Gaipi-4GlcNAc-R (Green et al., J Biol Chem 263: 18253, 1988). Expression of a-gal epitopes on the various glycoproteins could be demonstrated by the binding of the monoclonal anti-Gal antibody M86 (Galili et al, Transplantation 65: 1129, 1998) to the glycoproteins in ELISA wells. The monoclonal anti-Gal antibody M86 readily bound to mouse laminin, bovine thyroglobulin, a-gal fetuin and to a-gal BSA (Figure 3A). However, this monoclonal anti-Gal antibody did not bind to human thyroglobulin, human laminin, or fetuin, in accord with the complete lack of a-gal epitopes on these glycoproteins (Figure 3A). Incubation of the glycoproteins having a-gal epitopes and bound to ELISA wells, with 10 Units/ml of a- galactosidase for 2h at room temperature, resulted in destruction of a-gal epitopes. This was indicated by the subsequent lack of monoclonal anti-Gal antibody binding to wells coated with bovine thyroglobulin, mouse laminin, a-gal fetuin or a-gal BSA after these wells were incubated with recombinant a-galactosidase (Figure 3B).

Anti-Gal IgG within human serum also readily bound to bovine thyroglobulin, mouse laminin, and to a-gal fetuin, but not to the glycoproteins lacking a-gal epitopes, including human thyroglobulin, human laminin, or fetuin (Figure 3C). Incubation with 10 Units/ml a- galactosidase resulted in the subsequent inability of anti-Gal in human serum to bind to these glycoproteins because of the destruction of a-gal epitopes on these glycoproteins by this enzyme (Figure 3D). These findings demonstrate the efficacy of recombinant a-galactosidase in eliminating a-gal epitopes from carbohydrate chains of various glycoproteins. The observed residual binding represents nonspecific binding of human IgG to the ELISA wells

EXAMPLE 2: PREVENTION OF COMPLEMENT ACTIVATION BY a- GALACTOSIDASE MEDIATED ELIMINATION OF a-GAL EPITOPES ON GLYCOPROTEINS

One of the potentially harmful effects of the in vivo interaction between anti-Gal and therapeutic glycoproteins expressing a-gal epitopes is the activation of the complement cascade as a result of this antigen/antibody interaction. Generation of the complement cleavage products C3a and C5a in the circulation may lead to the induction of a septic shock like syndrome, referred to as systemic inflammatory response syndrome (SIRS) by a mechanism that is similar to serum sickness, however the effects are much faster and can occur within minutes after administration of the glycoprotein. If potent enough, this systemic response may cause kidney, lung and heart failure (Guo and Ward Recent Patents Anti-Infect Drug Disc 1 :57, 2006; Ward, Focus on Complement 5:2, 2007).

The activation of complement by anti-Gal/a-gal epitope interaction also could be demonstrated in vitro by measuring complement activity in human serum prior and post addition of glycoproteins with a-gal epitopes. Viability of the anti-Gal hybridoma M86 cells incubated with the tested serum served as a readout system for activation of complement. M86 cells were originally generated by fusion of mouse myeloma cells (containing active ccl,3GT) with lymphocytes from anti-Gal producing l,3GT knockout mouse (Galili et al., Transplantation 65: 1129, 1998). The cloned anti-Gal hybridoma cells produce monoclonal anti-Gal IgM antibody, as well as maintain al,3GT activity synthesizing a-gal epitopes (contributed by the myeloma fusion partner). A portion of anti-Gal produced by the M86 cells, binds to the a-gal epitopes expressed on these anti-Gal secreting cells (Figure 4A). Thus, when M86 cells are added to human serum, they are readily lysed by the complement cascade that is activated by the monoclonal anti-Gal IgM molecules bound to its antigen, (i.e. ligand) a-gal epitopes on these cells. The complement mediated cytolysis is detected even at a serum of 1 :64 (Figure 4B). Inactivation of complement in human serum by 30 min incubation at 56°C results in complete elimination of complement mediated lysis (Figure 4B). The cytolysis of the M86 cells is dependent only on presence of complement and is not dependent on anti-Gal in the serum. This is indicated by the observation that normal rabbit serum, which has complement, but lacks anti- Gal (like all nonprimate mammals), induces complement mediated cytolysis of M86 cells at a level similar to that of human serum (Figure 4B).

Evaluation of complement consumption serves as a sensitive measure for complement activation as a result of serum anti-Gal binding to a-gal epitopes on glycoproteins. The assayed glycoprotein is added to serum and incubated for 2h at 37°C. Subsequently, M86 cells are added for incubation of lh at 37°C. Interaction of anti-Gal with a-gal epitopes on glycoproteins will result in complement activation and consumption in the period of 2h incubation prior to addition of M86 cells. If complement is consumed, there will be less complement available for M86 cytolysis. Thus, decreased cytolysis will reflect complement consumption as a result of complement activation due to binding of anti-Gal to glycoproteins with a-gal epitopes, prior to addition of the M86 cells. Incubation of human serum at various dilutions with lOmg/ml of a-gal BSA resulted in almost complete consumption of complement, as indicated by the subsequent lack of complement mediated cytolysis of M86 cells at serum dilutions of 1 :4-1:64 (Figure 4C). However, incubation of a-gal BSA with 10 Units/mi of a-galactosidase for 4h at room temperature, prior to incubation with human serum, resulted in complete elimination of complement activation (i.e. effective cytolysis of M86 cells that are added to the serum after the 2h incubation with enzymatically treated a-gal BSA), because of the destruction of a-gal epitopes by the enzyme. This finding implies that a similar enzymatic treatment will prevent the subsequent in vivo activation of complement because the administered glycoproteins lacks a-gal epitopes and thus, it does not bind anti-Gal antibodies.

High complement activation was also observed with a-gal fetuin because of the 9 a-gal epitopes on its carbohydrate chains (Figure AC). As with a-gal BSA, enzymatic destruction of a- gal epitopes on a-gal fetuin by pre-incubation with 10 Units/ml α-galactosidase resulted in elimination of anti-Gal/ a-gal epitopes interaction (as in Figure 3B and 3D) and thus, only marginal complement consumption was observed, i.e. the extent of M86 cell cytolysis in suspensions containing enzymatically treated a-gal fetuin did not differ significantly from that in suspensions lacking glycoproteins (Figure 4C).

EXAMPLE 3: a-GALACTOSIDASE COLUMNS DESTROY a-GAL EPITOPES ON a- GAL BSA

The present invention teaches how to destroy a-gal epitopes on glycoproteins with a-gal epitopes by passing them through a column of porous beads, or of any other matrix to which the enzyme α-galactosidase is coupled covalently or bound non-covalently. The present example describes a column of α-galactosidase coupled to cyanogen bromide activated agarose beads, as schematically illustrated in Figure 2, and the efficacy of such a column in destroying a-gal epitopes on glycoproteins is demonstrated.

Cyanogen bromide activated agarose beads (also called Sepharose beads) (Sigma Co. St Louis, MO) were placed in a small 1 ml column attached to a 10ml reservoir above it. The column was equilibrated with carbonate buffer, pH 8.3. Subsequently, 50ml of recombinant a- galactosidase (Zhu et al, Arch Biochim Biophys 324:65, 1995) at a concentration of 100 Units/ml in carbonate buffer, pH 8.3, were passed through the 1ml column of the cyanogens bromide activated beads. The enzyme solution was passed twice through the column to enable binding of recombinant a-galactosidase to the matrix of the beads. The beads in the column were then incubated overnight in a solution of 0.2M glycine in carbonate buffer pH 8.3 in order to block the remaining reactive groups on the beads which did not bind the enzyme. Subsequently, the column was washed with 20ml carbonate buffer followed by 100ml solution of 0.5M NaCl, then with 30ml of saline.

A solution of 100μg/ml a-gal BSA (bovine serum albumin carrying 21 cc-gal epitopes per molecule) in saline was applied on the α-galactosidase coupled column at a total volume of 0.5ml (i.e. 50μg a-gal BSA) in three 166.6μ1 aliquots loaded on the column in 10 min intervals with flow stopping in between the loadings. Subsequently, 1.5ml saline was added to the column. The effluent was collected at a total volume of 2ml. The final concentration of a-gal BSA in the effluent is 25μ§/ηι1. It is assumed that the a-gal epitopes on a-gal BSA passing through the beads of the column will be destroyed by the a-galactosidase attached to the beads as illustrated schematically in Figure 2. To study this assumption, the a-gal BSA in the effluent was placed in serial two-fold dilutions in carbonate buffer pH 9.5, starting at a concentration of 10μg/ml of the glycoprotein. In parallel, the original a-gal BSA solution was brought to a concentration of 25μg ml and placed in serial two-fold dilutions in carbonate buffer pH 9.5, starting at a concentration of 10μg/ml. Both glycoproteins were allowed to attach to the ELISA wells for 2h at 37°C followed by 20h at 4°C. After blocking of the wells with 1% BSA in PBS, the wells were washed and the monoclonal anti-Gal antibody M86 was placed for 2h in the wells, at a dilution of 1 :5. This monoclonal antibody binds to a-gal epitopes and thus enables the detection of these epitopes on the enzymatically treated or untreated a-gal BSA in the ELISA wells. The ELISA plates were then washed and horseradish peroxidase (HRP) coupled goat anti-mouse IgM antibodies added to the wells for lh (Galili et al., Transplantation 65: 1129, 1998). Subsequently, the plates were washed and color development performed with ortho-phenylene-diamine (OPD), as in Figure 3 above. The untreated a-gal BSA attached to ELISA wells at different concentrations carries multiple synthetic a-gal epitopes linked through a 14 carbon spacer to BSA and thus can be readily detected by monoclonal anti-Gal antibody binding yielding 1 O.D. even at a coating concentration of 0.15μg/ml (Figure 5). In contrast, no monoclonal anti-Gal antibody bound to the a-gal BSA passed through the a-galactosidase column even at a coating concentration of lC^g/ml (Figure 5). This implies that a-galactosidase attached to the beads destroyed all the a-gal epitopes of the a-gal BSA passing through the column.

EXAMPLE 4: GLYCOPROTEIN PASSED THROUGH oc-GALACTOSIDASE COLUMNS LOSES ITS ABILITY TO ACTIVATE COMPLEMENT

The prevention of complement activation following anti-Gal/a-gal epitope interaction could be demonstrated in human serum with a-gal BSA that was passed through the a-galactosidase column, i.e. a-gal BSA on which a-gal epitopes were destroyed by agarose bound a- galactosidase. As shown in Figure 4 above and in Figure 6, incubation of M86 cells 37°C for lh with human serum, at various dilutions of the serum, results in the extensive cytolysis of the cells as a result of activation of complement in the human serum by the monoclonal anti-Gal antibody bound to these cells. Pre-incubation of the human serum with 100Lig/ml of a-gal BSA for 2h at 37°C resulted in activation of a large proportion of the complement and its consumption prior to the addition of the M86 cells. Therefore, as many as 80% of M86 cells are not killed upon their incubation with the serum containing a-gal BSA at a 1 :2 dilution of the serum (Figure 6). At higher serum dilutions, no complement mediated M86 cytolysis is observed in presence of a-gal BSA because of complement consumption by the anti-Gal/a-gal epitope interaction (Figure 6). However, when a-gal BSA was passed through the α-galactosidase column, then added to the human serum dilutions at a concentration of 100μg/ml, the complement consumption was greatly reduced as indicated by the extensive complement mediated cytolysis of M86 cells (Figure 6). This decrease in complement consumption was the result of destruction of a-gal epitopes on a-gal BSA by the α-galactosidase column, thus a much lower binding of anti-Gal to a-gal BSA occurred prior to the addition of M86 cells to the serum at various dilutions.

These findings demonstrate in vitro the prevention of one of the possible adverse outcomes of a-gal glycoproteins administered into humans- the systemic activation of complement in the blood, which under severe conditions may lead to systemic inflammatory syndrome response (SIRS). Passing of a-gal proteins through the α-galactosidase column results in the destruction of a-gal epitopes. Therefore, subsequent administration of such glycoproteins into patients will not be followed by binding of anti-Gal to the glycoproteins, thus, complement activation will be avoided. Similarly, the marked decrease or complete elimination of anti-Gal binding is likely to prevent interaction with anti-Gal IgE molecules on mast cells and thus, it will diminish the risk of allergic reactions mediated by this antibody.

Claims

We claim 1. A method, comprising applying a therapeutic glycoprotein preparation having one or more a-gal epitopes having a terminal Galal-3Gal or terminal a-galactosyls to a column having covalently or non-covalently coupled α-galactosidase in porous beads or on particulate material in order to destroy said a-gal epitope by enzymatic removal of the terminal galactose by said a- galactosidase.

2. The method of Claim 1, wherein said therapeutic glycoproteins are monoclonal antibodies.

3. The method of Claim 1, wherein said therapeutic glycoproteins are recombinant glycoproteins.

4. The method of Claim 1, wherein said therapeutic glycoproteins are natural glycoproteins.

5. The method of Claim 1, wherein said α-galactosidase is obtain from a natural source including, but not limited to microbial source, fungal source, or plant source.

6. The method of Claim 1, wherein said α-galactosidase is a recombinant a-galactosidase produced in a prokaryotic gene expression system or in a eukaryotic gene expression system or in a transgenic animal.

7. A method, comprising destruction of one or more a-gal epitopes having a terminal Gala 1-3 Gal or terminal a-galactosyls on therapeutic glycoproteins by mixing said glycoproteins with tagged α-galactosidase in order to destroy said a-gal epitope by enzymatic removal of the terminal galactose and to subsequently remove said a-galactosidase.

8. The method of Claim 7, wherein said tagged a-galactosidase is a recombinant ot- galactosidase produced in a prokaryotic gene expression system or an eukaryotic gene expression system or in a transgenic animal and said enzyme is tagged with an oligopeptide, polypeptide, protein or other tag that enables the removal of said α-galactosidase from the reaction mixture by passing said mixture through a column of porous beads or with particulate material carrying linked ligand to said tag.

9. The method of Claim 7, wherein said α-galactosidase is tagged with an oligopeptide of 6 histidines (Hise), or tagged with fusion proteins as protein A, Fc portion of an immunoglobulin molecule, glutathione S-transferase, or tagged with biotin, and the corresponding columns with matrix for removal of the tagged α-galactosidase including nickel-sepharose column, IgG coupled column, protein A coupled column, glutathione coupled column, or avidin coupled column, respectively.

10. The method of Claim 7, wherein said α-galactosidase is not tagged.