EA041325B1 - PROTEIN PURIFICATION METHOD - Google Patents

Description

The field of technology to which the invention belongs

The present invention relates to the field of protein purification, more specifically to the field of antibody purification.

State of the art prior to the invention

In the field of therapeutics, the use of proteins and antibodies, and antibody-derived molecules in particular, has continuously become necessary and important, and thus a need for controlled manufacturing processes has arisen in parallel. The commercialization of therapeutic proteins requires their production in large quantities. For this purpose, proteins are often expressed in host cells and then must be isolated and purified before they can be obtained in administrationable forms.

The most common class of antibody molecules is immunoglobulin G (IgG), a heterotetramer consisting of two heavy chains and two light chains. The IgG molecule can be subdivided into two functional subunits: (1) a crystallizing fragment (Fc), which forms the terminal portion of an antibody and interacts with cell surface receptors to activate the immune response; and (2) an antigen binding fragment (Fab) that mediates antigen recognition. The Fc region contains two pairs of constant domains (CH2 and CH3) of two paired heavy chains, while the Fab region consists of a variable domain followed by a heavy chain constant domain (VH and CH1, respectively) that are paired with a light chain variable and constant domain ( VL and CL, respectively). The Fc and Fab regions are separated by a hinge region that contains disulfide bonds holding the two chains together. Full length IgG antibodies have traditionally been purified by methods that include an affinity chromatography capture step using protein A derived from Staphylococcus aureus. The high binding specificity of protein A and the Fc region of antibodies makes it possible to remove more than 98% of impurities in one step by this chromatography method, starting directly from complex solutions, such as media collected from cell cultures. The high factor achieved at this stage of the process helps to simplify the entire subsequent purification process. Typically, only trace impurities remain to be removed after this purification step (high molecular weight aggregates, residual host cell proteins, leachable Protein A) and this can usually be achieved with one or two subsequent chromatography steps.

In addition, in antibodies that contain a specific framework subgroup 3 of the human heavy chain variable region, an alternative protein binding site A has been described (Sasso et al. J. Immunol. 1991, 147:1877-1883, Human IgA and IgG F(ab') ₂ that bind to staphylococcal protein A belong to the VHIII subgroup), also referred to as the heavy chain variable region VH3 domain and often classified as a secondary interaction. Although all five protein A domains (A, B, C, D, E) bind to the Fc region of IgG, only the D and E domains show significant binding to VH3. In the context of IgG purification using protein A affinity chromatography, which is based on interaction between the Fc domain of IgG and protein A, these interactions of the VH3 domain with protein A were considered undesirable given that they affect the elution profile of the antibody being purified, and alternative resins have been developed and are commercially available that contain only the B domain of protein A, such as SuRe ® from GE Healthcare.

However, many antibodies and antibody-derived molecules currently available and/or under development do not contain Fc regions, and further adaptation of their purification methods is needed. The binding of protein A to the VH3 regions described above could provide the use of protein A in the production of such antibodies.

A particular requirement for antibody purification is the isolation of the desired antibody, or a molecule derived from the antibody in monomeric form, or substantially free of higher molecular weight molecules such as dimers and trimers.

Molecules of certain antibodies have an increased tendency to form multimers, which is the result of random pairing of the variable domain with the variable domains of adjacent molecules. This is especially true for more complex antibody-derived molecules in which linker regions are used between different domains of interest.

Thus, there remains a need in the art for additional methods for the production and purification of antibodies and antibody-derived molecules in monomeric form, in particular when these molecules do not contain an Fc region.

Brief description of the drawings

Fig. 1 is a chromatogram showing the elution profile of A26Fab-645dsFv from protein A resin eluting with a pH gradient.

In FIG. 2 shows the total amount of protein, monomeric and multimeric A26Fab-645dsFv, present in each of the fractions obtained after the chromatography shown in FIG. 1 as analyzed by size exclusion HPLC.

Fig. 3 is a chromatogram showing the elution profile of the Fc construct from resin with protein A when eluting with a pH gradient.

In FIG. 4 shows the total amount of protein, monomeric and multimeric forms present in each of the fractions obtained after the chromatography shown in FIG. 3 as analyzed by size exclusion HPLC.

Fig. 5 is a chromatogram showing the elution profile of A26Fab-645dsFv from protein A resin in pH stepping elution.

In FIG. 6 shows the analysis by size exclusion HPLC of the fractions obtained after the Protein A chromatography shown in FIG. 5. The dotted line represents the analysis of the fraction eluting at pH 3.8, while the continuous line represents the analysis of the fraction eluting at pH 3.0.

In FIG. 7 shows non-reducing SDS-PAGE analysis of the fractions obtained after protein A chromatography of A26Fab-645dsFv in the pH stepping elution shown in FIG. 5. Lane 1 shows molecular weight markers, lane 2 shows a sample of decontaminated cell culture supernatant as applied for protein A chromatography, lane 3 shows the elapsed fraction from protein A chromatography, lane 4 shows the fraction obtained after elution at pH 3.8, and lane 5 represents the fraction obtained after elution at pH 3.0.

In FIG. 8 shows a reduction SDS-PAGE analysis of the fractions obtained after protein A chromatography of A26Fab-645dsFv in the pH stepping elution shown in FIG. 5. Lane 1 shows molecular weight markers, lane 2 shows a sample of decontaminated cell culture supernatant as applied for protein A chromatography, lane 3 shows the elapsed fraction from protein A chromatography, lane 4 shows the fraction obtained after elution at pH 3.8, and lane 5 represents the fraction obtained after elution at pH 3.0.

In FIG. 9 shows size exclusion HPLC analysis of fractions obtained after A26Fab-645dsFv protein A gradient elution chromatography as described in Example 4 showing total protein, monomeric and multimeric A26Fab-645dsFv.

In FIG. 10 shows size exclusion HPLC analysis of fractions obtained after A26Fab-645dsFv protein A gradient elution chromatography as described in Example 5 showing total protein, monomeric and multimeric A26Fab-645dsFv.

In FIG. 11 shows size exclusion HPLC analysis of fractions obtained after A26Fab-645dsFv protein A gradient elution chromatography as described in Example 6 showing total protein, monomeric and multimeric A26Fab-645dsFv.

In FIG. 12 is a schematic representation of the monomeric Fab-dsFv and multimeric versions of Fab-dsFv, as well as their main components: Fab and dsFv. This diagram illustrates the possible monomers, dimers and trimers. However, in fact, when dimers and trimers form cyclic structures, all linkers would be the same length.

In FIG. 13-20 show various sequences of antibody molecules and their components.

In FIG. 21 is a chromatogram showing the elution profile of TrYbe from protein A resin eluting with a pH gradient.

In FIG. 22 shows the analysis by size exclusion HPLC of the fractions obtained after the Protein A chromatography shown in FIG. 21.

In FIG. 23 shows non-reducing (A) and reducing (B) SDS PAGE analysis of the fractions obtained after TrYbe protein A chromatography eluting with the pH gradient shown in FIG. 21.

In FIG. 24 shows the total amount of protein, monomeric and multimeric TrYbe present in each of the fractions obtained after the chromatography shown in FIG. 21 as analyzed by size exclusion HPLC.

In FIG. 25 is a chromatogram showing the elution profile of BYbe from protein A resin eluting with a pH gradient.

In FIG. 26 shows HPLC analysis of the fractions obtained after the Protein A chromatography shown in FIG. 25.

In FIG. 27 shows non-reducing (A) and reducing (B) SDS PAGE analysis of fractions obtained after BYbe protein A chromatography eluting with the pH gradient shown in FIG. 25.

In FIG. 28 shows the total amount of protein, monomeric and multimeric BYbe present in each of the fractions obtained after the chromatography shown in FIG. 25 as analyzed by size exclusion HPLC.

In FIG. 29 is a schematic representation of alternative Fab-scFv monomer formats amenable to purification by the method of the invention.

In FIG. 30 is a schematic representation of alternative Fab-2x dsscFv (TrYbe®) and Fab-dsscFv-dsFv monomer formats amenable to purification by the method of the invention.

Detailed description of the invention

The present invention satisfies the above need by providing a new method for producing an antibody containing a human VH3 domain in monomeric form. Currently at

- 2 041325 binding of human VH3 domains and protein A revealed the effect of avidity. This discovery is unexpected given that it is not described for the interaction of Fc regions and protein A and has led to the development of a new method that allows the isolation of monomeric antibodies containing the human VH3 domain from a mixture containing monomeric and multimeric forms of the antibody.

In a first embodiment, the present invention relates to a method for producing an antibody containing a human VH3 domain in monomeric form, comprising:

a) applying a mixture containing an antibody containing a human VH3 domain, in monomeric and multimeric form, to a chromatographic material containing protein A, where said protein A contains a domain D and/or E, under conditions that ensure the binding of said antibody to protein A ; And

b) isolating an antibody containing a human VH3 domain in monomeric form, wherein the antibody containing a human VH3 domain does not contain an Fc region.

In a second alternative embodiment, the present invention relates to a method for producing an antibody containing a human VH3 domain, comprising:

a) expression of the antibody in the host cell;

b) obtaining a mixture containing the antibody, host cells and other impurities;

c) purifying the antibody using at least a protein A chromatography step, wherein said protein A contains a D and/or E domain; And

d) obtaining an antibody containing a human VH3 domain, wherein the antibody containing a human VH3 domain does not contain an Fc region.

In a third alternative embodiment, the invention relates to a method for separating an antibody comprising a human VH3 domain in monomeric form from an antibody in multimeric form, comprising:

a) applying a mixture containing an antibody containing a human VH3 domain, in monomeric and multimeric form, to a protein A-containing chromatographic material, where said protein A contains a D and/or E domain;

b) allowing said antibody to bind to protein A;

c) applying an elution buffer that selectively breaks the binding of the antibody in monomeric form;

d) obtaining the resulting eluate; and optional

e) applying a second elution buffer that breaks the binding of the antibody in multimeric form, and obtaining this second eluate, where the antibody containing the human VH3 domain does not contain an Fc region.

In a fourth alternative embodiment, the invention relates to a method for separating an antibody containing a human VH3 domain in monomeric form from an antibody in multimeric form, comprising:

a) applying a mixture containing an antibody containing a human VH3 domain, in monomeric and multimeric form, to a protein A-containing chromatographic material, where said protein A contains a D and/or E domain;

b) providing binding of the antibody in multimeric form;

c) obtaining the antibody in monomeric form in the filtrate; and optional

d) application of an elution buffer that selectively destroys the binding of the antibody in multimeric form; And

e) obtaining an eluate resulting from d), where the antibody containing the human VH3 domain does not contain an Fc region.

In a further embodiment, the method of the invention further comprises one or more chromatography steps to remove remaining impurities. Typically, such steps use a non-affinity chromatography step using a solid phase with appropriate functionality for use in size exclusion chromatography, cation exchange chromatography, anion exchange chromatography, mixed mode chromatography, hydrophobic chromatography and hydrophobic charge induction chromatography. They can be carried out in binding mode and elution mode or in flow mode. In flow mode, impurities bind or have reduced mobility in the solid phase, while the desired protein is isolated in the eluate or in the flow fraction. Suitable solid phases for use in chromatography, such as granular resins or membranes with suitable functionality, are readily available to those skilled in the art. In a specific embodiment of the method of the invention, the method further comprises an anion exchange chromatography step carried out in flow mode.

In a further specific embodiment, the method of the invention comprises a protein A chromatography step followed by a first chromatography step that is an anion exchange chromatography to obtain a protein containing filtrate and a second chromatography step that is a cation exchange chromatography followed by a protein containing eluate. .

Alternatively, the method of the invention comprises protein A chromatography followed by a first chromatography step which is a cation exchange chromatography followed by a protein containing eluate and a second chromatography step which is an anion exchange chromatography to obtain a protein containing filtrate .

Typically, Protein A chromatography is carried out in a binding or elution mode where the binding of the desired protein to the solid phase allows impurities, such as contaminating proteins, to pass through the chromatographic medium while the desired protein remains bound to the solid phase. The bound desired protein is then obtained from the solid phase using an elution buffer that disrupts the mechanism by which the desired protein is bound to said solid phase.

In a further embodiment, the method of the invention includes adding a first solution to the protein A-containing chromatographic material after application of the mixture containing the human VH3 domain-containing antibody in monomeric and multimeric form, such that the unbound material goes into solution.

In a further embodiment of the method of the invention, an elution buffer is applied to the protein A-containing chromatographic material such that the bound antibody is released.

In a further embodiment of the method of the invention, the eluate obtained after protein A chromatography is enriched in monomeric antibody compared to multimeric antibody relative to the applied mixture.

As understood by one skilled in the art, in the context of the present invention, the protein in the eluate after Protein A chromatography contains an increased percentage of antibodies in monomeric form relative to the mixture prior to the Protein A chromatography step.

In a specific embodiment, the eluate resulting from protein A chromatography contains at least 50%, at least 60%, at least 70%, 75%, 80%, 85%, or at least 90% of an antibody containing a human VH3 domain, in monomeric form.

In a further alternative embodiment where the antibody is to be isolated in multimeric form, the protein contained in the eluate after Protein A chromatography contains an increased percentage of the antibody in multimeric form relative to the mixture before the Protein A chromatography step. In a specific embodiment, the eluate obtained after chromatography with protein A, contains at least 50%, at least 60%, at least 70%, 75%, 80%, 85% or at least 90% of an antibody containing a human VH3 domain in multimeric form.

In a further embodiment of the method of the invention, said protein A is a native recombinant protein A.

There are many chromatographic materials available to the person skilled in the art containing said native recombinant protein A, such as, for example, MabSelect® (GE Healthcare), Absolute® (Novasep), Captiv A® (Repligen) or Amsphere® (JSR).

In a specific embodiment of the method of the invention, the bound antibody is released from the Protein A-containing chromatographic material by applying an elution buffer at a pH suitable to disrupt antibody binding. This pH is molecule dependent and is typically empirically determined by one skilled in the art and adjusted to achieve the desired endpoint, i.e. it may be desirable to obtain the greatest possible amount of monomer from the supported mixture, or it may be desirable to obtain the highest possible purity of the monomer. In a particular embodiment of the method according to the invention, the pH of the elution buffer is pH 3.0 to 4.5, preferably pH 3.2 to 4.3, pH 3.5 to 4, preferably pH 3.6 to 3.9 or pH 3.8.

Buffers suitable for use as wash and elution buffers in protein A chromatography are readily available in the art and can be selected from phosphate buffered saline (PBS), Tris, histidine, acetate, citrate or MES buffers as non-limiting examples. (imidazole 2-(N-morpholino)ethanesulfonic acid), BES (N,N-(6uc-2hydroxyethyl)-2-aminoethanesulfonic acid), MOPS (3-(N-morpholino)propanesulfonic acid) or HEPES (N-2- hydroxyethylpiperazine-N'-2-ethanesulfonic acid).

In a specific embodiment, the method of the invention comprises applying a mixture containing an antibody containing a human VH3 domain, in monomeric and multimeric form, to a protein A-containing chromatographic material, wherein said protein A contains a D and/or E domain, under conditions that allow binding said protein A antibody; applying the first solution or wash buffer such that unbound material goes into solution; applying an elution buffer to the protein A-containing chromatographic material so that the bound antibody is released; and obtaining an antibody containing a human VH3 domain in monomeric form, wherein the resulting solution is enriched with an antibody containing a VH3 domain in monomeric form relative to the applied mixture, and wherein the antibody containing a human VH3 domain does not contain an Fc region.

In a further specific embodiment, the method of the invention comprises applying a mixture containing an antibody containing a human VH3 domain, in monomeric and multimeric form, to a protein A-containing chromatographic material, wherein said protein A contains the D domain.

- 4 041325 and/or E, under conditions that ensure the binding of the specified antibody to protein A; applying the first solution or wash buffer so that unbound material goes into solution; applying an elution buffer to a protein A-containing chromatographic material such that the bound antibody is released in monomeric form; and obtaining an antibody containing a human VH3 domain in monomeric form, wherein the resulting solution is enriched with an antibody containing a VH3 domain in monomeric form relative to the applied mixture, and wherein the antibody containing a human VH3 domain does not contain an Fc region.

In a further embodiment, the VH3 domain-containing antibody is selected from Fab', F(ab') ₂ , scFv, Fab-Fv, Fab-scFv, Fab-(scFv)2, Fab-(Fv)2, diabody, triobody and tetratel.

In a further embodiment of the method of the invention, an antibody comprising a VH3 domain contains at least two human VH3 domains.

In a further embodiment of the method of the invention, an antibody comprising a human VH3 domain specifically binds to OX40.

In one embodiment of the method of the invention, the antibody is a FabFv or a disulfide-stabilized form thereof as described in PCT/EP2014/074409, incorporated herein by reference.

In one embodiment, the antibody comprises a binding domain specific for human serum albumin, specifically with CDRs or variable regions, as described in WO 2013/068563, incorporated herein by reference.

In a further embodiment of the method of the invention, said human VH3 domain containing antibody comprises:

Heavy chain CDR1, CDR2 and CDR3 as defined in SEQ ID NO: 1, SEQ ID NO: 2 and SEQ ID NO: 3, respectively; And

Light chain CDR1, CDR2 and CDR3 as defined in SEQ ID NO: 4, SEQ ID NO: 5 and SEQ ID NO: 6, respectively.

In a further specific embodiment of the method of the invention, the antibody is A26Fab-645dsFv, which contains a Fab fragment that specifically binds to OX40 and an Fv fragment that specifically binds to serum albumin, where both fragments are stabilized by a disulfide bond as defined in WO 2013/ 068563, incorporated herein by reference.

In a further specific embodiment, the method of the invention comprises applying a mixture containing A26Fab-645dsFv in monomeric and multimeric form to a protein A-containing chromatographic material, wherein said protein A contains a D and/or E domain, under conditions that allow said antibody to bind to protein A; applying the first solution or wash buffer so that unbound material goes into solution; applying an elution buffer to a protein A-containing chromatographic material such that the bound antibody is released in monomeric form; and obtaining A26Fab-645dsFv in monomeric form, where the resulting solution is enriched with A26Fab-645dsFv in monomeric form relative to the applied mixture and where the pH of the elution buffer is from 3.5 to 4.2, preferably from 3.6 to 4.1, from 3, 7 to 4.0, preferably 3.8 to 3.9 or 3.8.

In a further specific embodiment, the method of the invention further comprises applying a second elution buffer to a protein A-containing chromatographic material to obtain bound A26Fab-645dsFv in multimeric form, wherein the pH of said second elution buffer is pH below 3.5, preferably pH below 3.4 , preferably pH below 2.8-3.2, preferably pH 2.9-3.1, preferably pH 3.0.

Thus, the present invention provides a bispecific antibody fusion protein that binds to human OX40 and human serum albumin, comprising:

a heavy chain comprising sequentially from the N-terminus a first heavy chain variable domain (VH#1), a CH1 domain, and a second heavy chain variable domain (VH#2);

a light chain comprising, in series from the N-terminus, a first light chain variable domain (VL#1), a CL domain, and a second light chain variable domain (VL#2), wherein said heavy and light chains are aligned such that VH#1 and VL# 1 form the first antigen-binding site, and VH#2 and VL#2 form the second antigen-binding site;

where, specifically, the antigen that binds to the first antigen-binding site is human OX40, and the antigen that binds to the second antigen-binding site is human serum albumin;

where the first heavy chain variable domain (VH#1) contains the sequence shown in SEQ ID NO: 1 for CDR-H1, the sequence shown in SEQ ID NO: 2 for CDR-H2 and the sequence shown in SEQ ID NO: 3 for CDR-H3, and the first light chain variable domain (VL#1) contains the sequence shown in SEQ ID NO: 4 for CDR-L1, the sequence shown in SEQ ID NO: 5 for CDR-L2, and the sequence shown in SEQ ID NO: 6 for CDR-L3;

- 5 041325 where the second variable domain of the heavy chain (VH#2) contains the sequence shown in SEQ ID NO: 11, and the second variable domain of the light chain (VL#2) contains the sequence shown in SEQ ID NO: 12, and the second variable the heavy chain domain (VH#2) and the second light chain variable domain (VL#2) are linked by a disulfide bond.

In one embodiment, a peptide linker is present between the CH1 domain and the second heavy chain variable domain (VH#2). In one embodiment, a peptide linker is present between the CL domain and the second light chain variable domain (VL#1). In one embodiment, the first heavy chain variable domain (VH#1) contains the sequence shown in SEQ ID NO: 8. In one embodiment, the first light chain variable domain (VL#1) contains the sequence shown in SEQ ID NO: 7. In one embodiment, the heavy chain contains or consists of the sequence shown in SEQ ID NO: 15. In one embodiment, the light chain contains or consists of the sequence shown in SEQ ID NO: 16.

Thus, in one embodiment, a bispecific antibody fusion protein is provided that binds to human OX40 and human serum albumin, with a heavy chain containing the sequence shown in SEQ ID NO: 15 and a light chain containing the sequence shown in SEQ ID NO: 16.

In one embodiment, the antibody molecule, eg Fab-dsFv format, is the antibody molecule described in PCT/EP2014/074409 or WO 2014/019727, incorporated herein by reference.

In another embodiment, the antibody molecule is a Fab-scFv format fusion protein as described in WO 2013/068571, incorporated herein by reference.

In another embodiment, the antibody molecule is a multispecific antibody molecule comprising or consisting of:

a) a polypeptide chain of formula (I)

VH-CH1-X-V1; and

b) a polypeptide chain of formula (II)

VL-CL-Y-V2;

where e VH is a heavy chain variable domain;

CH 1 is a heavy chain constant region domain, eg domain 1;

X is a bond or linker;

Y is a bond or linker;

V 1 is dsFv, sdAb, scFv or dsscFv;

V L is a light chain variable domain;

CL is a domain from a light chain constant region such as Ckappa;

V 2 is dsFv, sdAb, scFv or dsscFv, where at least one of V1 or V2 is dsFv or dsscFv as described in WO 2015/197772, incorporated herein by reference.

In one particular embodiment, the antibody molecule is a Fab-2x dsscFv format polyspecific antibody molecule described in WO 2015/197772, incorporated herein by reference.

In a further specific embodiment, the Fab-2x dsscFv format polyspecific antibody molecule is a trivalent antibody, i. each of the Fvs binds to a different epitope.

In a further specific embodiment, the format of the multispecific antibody molecule is the Fab-dsscFv-dsFv format as described in WO 2015/197772, incorporated herein by reference.

An antibody that can be produced by the method of the present invention can be obtained by culturing eukaryotic host cells transfected with one or more expression vectors encoding the recombinant antibody. Preferably the eukaryotic host cells are mammalian cells, more preferably Chinese hamster (CHO) cells.

Mammalian cells can be cultured in any medium that supports their growth and the expression of antibodies in them, preferably the medium is a chemically defined medium that does not contain animal-derived products such as animal serum and peptone. There are various cell culture media available to the person skilled in the art containing various combinations of vitamins, amino acids, hormones, growth factors, ions, buffers, nucleosides, glucose or an equivalent energy source at appropriate concentrations to promote cell growth and protein production. Additional components of the cell culture media, which are known to those skilled in the art, may be included in the cell culture medium at appropriate concentrations at various points during the cell culture cycle.

- 6 041325

Mammalian cell culture can be carried out in any suitable container, such as a stirred flask or bioreactor, which may or may not be operated in a fed-batch mode, depending on the scale of production required. These bioreactors may be of the mixing tank type or bottom air fed type. Various large scale bioreactors are available with capacities greater than 1000-50000L, preferably 5000-20000L or up to 10000L. Alternatively, smaller scale bioreactors, such as 2 to 100 L, can also be used to produce an antibody by the method of the invention.

Typically, the antibody or antigen-binding fragment thereof that can be produced by the methods of the present invention is in the supernatant of a mammalian host cell culture, typically a CHO cell culture. When treating CHO cultures where the desired protein, such as an antibody or antigen-binding fragment thereof, is secreted into the supernatant, said supernatant is collected by methods known in the art, typically by centrifugation.

Thus, in a particular embodiment of the invention, the method prior to purification of the protein includes the step of centrifugation and obtaining the supernatant. In a further specific embodiment, said centrifugation is a continuous centrifugation. For the avoidance of doubt, supernatant means the liquid above the pelleted cells obtained after cell culture centrifugation.

Alternatively, said supernatant can be obtained using decontamination methods known to those skilled in the art, such as depth filtration, for example. Thus, in a particular embodiment of the invention, the method prior to protein purification includes the steps of depth filtration and supernatant preparation.

Alternatively, the host cells are prokaryotic cells, preferably Gram-negative bacteria. More preferably, the host cells are E. coli cells. Prokaryotic host cells for protein expression are well known in the art (Terpe, K. (2006). Overview of bacterial expression systems for heterologous protein production: from molecular and biochemical fundamentals to commercial systems. Appl. Microbiol. Biotechnol. 72, 211- 222.). Host cells are recombinant cells that are genetically engineered to produce the desired protein, such as an antibody fragment. Recombinant E. coli host cells can be obtained from any suitable E. coli strain, including MC4100, TG1, TG2, DHB4, DH5a, DH1, BL21, K12, XL1Blue, and JM109. One example is E. coli strain W3110 (ATCC 27,325), a widely used host strain for recombinant protein fermentation. Antibody fragments can also be obtained by culturing modified E. coli strains, for example E. coli strains that are metabolically mutated or deficient in protease, such as those described in WO 2011/086136, WO 2011/086138 or WO 2011/086139 included in this document as a reference.

Typically, the antibody that can be purified by the methods of the present invention, depending on the nature of the protein, the scale of production and the E. coli strain used, is located in the periplasm of the E. coli host cell or in the culture supernatant of the host cells. Methods for targeting proteins to these compartments are well known in the art (Makrides, S.C. (1996). Strategies for achieving high-level expression of genes in Escherichia coli. Microbiol. Rev. 60, 512-538.). Examples of suitable signal sequences for directing proteins to the E. coli periplasm include the PhoA, OmpA, OmpT, LamB and OmpF E. coli signal sequences. Proteins can be directed into the supernatant using natural secretory pathways or by inducing limited outer membrane permeability causing protein secretion, examples of which are the use of the pelB leader, protein A leader, bacteriocin release protein co-expression, mitomycin-induced bacteriocin release protein, together with addition to the medium for cultivation of glycine and co-expression of the kil gene to increase membrane permeability. Most preferably, in the methods of the invention, expression of the recombinant protein occurs in the periplasm of E. coli host cells.

Expression of the recombinant protein in E. coli host cells can also be controlled by an inducible system when expression of the recombinant antibody in E. coli is under the control of an inducible promoter. Many inducible promoters suitable for use in E. coli are well known in the art and, depending on the promoter, expression of the recombinant protein can be induced by various factors such as temperature or the concentration of a particular substance in the growth medium. Examples of inducible promoters include the E. coli lac, tac, and trc promoters, which induce lactose or the non-hydrolysable analogue of lactose, isopropyl-β-D-1-thiogalactopyranoside (IPTG), and the phoA, trp, and araBAD promoters, which induce phosphate, tryptophan, and L-arabinose. respectively. Expression can be induced, for example, by adding an inducer or by changing the temperature, where the induction is temperature dependent. When induction of recombinant protein expression is achieved by adding an inducer to the culture, the inducer can be added in any suitable manner depending on the fermentation system and the inducer, for example, by one or more batch additions, or by gradual addition of the inducer during feeding. It should be understood that there may be a delay between the addition of the inducer and the actual induction of protein expression, for example, when the inductor is lactose, as long as any previously added carbon source is used before lactose, there may be a delay before induction of protein expression begins.

E. coli host cell cultures (fermentable cultures) can be grown in any medium that supports E. coli growth and recombinant protein expression. The medium may be any chemically defined medium, such as those described in Durany O., et al. (2004). Studies on the expression of recombinant fuculose-1-phosphate aldolase in Escherichia coli. Process Biochem. 39, 1677-1684.

The cultivation of E. coli host cells can be carried out in any suitable container, such as a stirred flask or a fermenter, depending on the scale of production required. Various large scale fermenters are available with capacities ranging from over 1,000L to approximately 100,000L. Preferably, fermenters from 1000 to 50000 liters are used, more preferably from 1000 to 25000, 20000, 15000, 12000 or 10000 liters. Smaller scale fermenters with capacities ranging from 0.5 to 1000 liters can also be used.

The fermentation of E. coli can be carried out in any suitable system, eg, continuous, batch or fed-batch mode, depending on the protein and desired yields. If necessary, intermittent mode can be used with portioned additions of nutrients or inducers. Alternatively, fed-batch and pre-induction batch cultures can be used at the maximum specific growth rate that can be maintained using the nutrients originally present in the fermenter and one or more nutrient feeds used to control growth rate until fermentation is complete. In a fed-batch mode, pre-induction can also be used to control the metabolism of E. coli host cells and to achieve higher achievable cell densities.

If desired, host cells can be collected from the fermentation medium, for example, host cells can be collected from a sample by centrifugation, filtration, or concentration.

In one embodiment, the method of the present invention, prior to protein isolation, includes the step of centrifuging and harvesting the cells.

In E. coli fermentation methods where the protein of interest, such as an antibody fragment, is located in the periplasmic space of the host cell, the protein must be released from the host cell. The release can be carried out by any suitable method, such as cell lysis by mechanical treatment or application of pressure, freeze-thaw treatment, osmotic shock, extractants or heat treatment. Such isolation techniques for protein release are well known in the art. Thus, in a particular embodiment, the method of the invention prior to protein purification includes an additional step of isolating the protein.

In a further embodiment, the method of the invention further comprises isolating host cells from the cell culture medium, harvesting a protein in a protein extraction step, obtaining a protein-containing mixture resulting from the protein extraction step, and purifying said protein from the mixture, wherein said purification comprises at least at least one protein A chromatography step.

In a particular embodiment, the extraction step includes adding extraction buffer to the sample and obtaining the resulting protein. Preferably, the extraction step is carried out for a suitable period of time and at a suitable temperature to recover the protein in its natural conformation and is empirically optimized for each particular protein. In a particular embodiment of the present invention, said extraction step is carried out at temperatures from 25 to 35°C, from 27 to 33°C, preferably from 29 to 31°C. The protein extraction step is carried out over a period of time, which is also empirically optimized depending on the particular protein and the temperature used. In a specific embodiment, said extraction step is carried out for a period of 4 to 20 hours, for a period of 6 to 18 hours, preferably 8 to 12 hours. In a specific embodiment, the protein extraction step is carried out for a period of 8 to 12 hours at temperatures from 29 to 31°C, preferably within 11 hours at 30°C.

In an alternative embodiment, an extraction buffer is added to the sample and then the sample is subjected to a heat treatment step. The heat treatment step is preferably as detailed in US Pat. No. 5,665,866, incorporated herein by reference.

After the extraction step, the mixture containing the desired protein, such as an antibody, can be subjected to a centrifugation and/or filtration step.

In a further specific embodiment, the method of the invention, after the extraction step and before purifying the protein from said mixture, may include the step of adjusting the pH of the mixture containing the desired protein.

As used herein, the term antibody or antibodies refers to monoclonal or polyclonal antibodies. As used herein, the term antibody or antibodies as non-limiting examples includes recombinant antibodies that are obtained through recombinant technologies, as is known in this field. The antibody or antibodies include antibodies of any species, in particular mammalian species, including antibodies with two essentially whole heavy and two essentially whole light chains, human antibodies of any isotype, including IgD, IgG1, IgG _2a , IgG2b, IgG ₃ , IgG ₄ , IgE, and antibodies that are produced as dimers of this basic structure, including IgA1, IgA ₂ , or pentamers such as IgM and modified variants thereof, antibodies from non-human primates such as chimpanzees, baboons, cynomolgus or cynomolgus monkeys, rodent antibodies, for example mice or rats; rabbit, goat, or horse antibodies; and camelid (eg, camel or llama, such as Nanotela™) antibodies and derivatives thereof, or avian species, such as chicken antibodies, or fish antibodies, such as shark antibodies. Also, the term antibody or antibodies refers to chimeric antibodies in which the first portion of at least one antibody heavy and/or light chain sequence is derived from a first species and the second portion of the antibody heavy and/or light chain sequence is derived from a second species. Chimeric antibodies of interest herein include primatized antibodies containing antigen-binding variable domain sequences derived from a non-human primate (e.g., marmosets such as baboon, rhesus monkey, or cynomolgus monkey) and human constant region sequences. Humanized antibodies are chimeric antibodies that contain a sequence derived from non-human antibodies. Advantageously, humanized antibodies are human antibodies (recipient antibody) in which the recipient's hypervariable region residues are replaced by hypervariable region [or complementarity determining region (CDR)] residues of a non-human species (donor antibody) such as mouse, rat, rabbit, chicken, or non-human primate, with the required specificity, affinity and activity. In most cases, human (recipient) antibody residues outside the CDR, i.e. in the framework region (FR) are further substituted with corresponding non-human residues. In addition, humanized antibodies may contain residues that do not exist in either the recipient antibody or the donor antibody. These modifications are made to further improve the performance of the antibody. Humanization reduces the immunogenicity of non-human antibodies in humans, thus facilitating the use of antibodies to treat human diseases. Humanized antibodies and several different technologies for their production are well known in this field. The term antibody or antibodies also refers to human antibodies that can be obtained as an alternative to humanization. For example, transgenic animals (eg, mice) can be generated that, when immunized, can produce the full repertoire of human antibodies with no production of endogenous mouse antibodies. For example, homozygous deletion of the antibody heavy chain junction (JH) region gene in chimeric and germline mutant mice has been described as resulting in complete inhibition of endogenous antibody production. Transfer of a set of human germline immunoglobulin genes to such germline mutant mice after immunization of a transgenic animal carrying human germline immunoglobulin genes with said antigen results in the production of human antibodies with specificity for that particular antigen. Technologies for obtaining such transgenic animals and technologies for isolating and producing human antibodies from such transgenic animals are known in the art. Alternatively, in a transgenic animal, such as a mouse, only the immunoglobulin genes encoding mouse antibody variable regions are replaced with the corresponding human immunoglobulin variable region gene sequences. Mouse germline immunoglobulin genes encoding antibody constant regions remain unchanged. Thus, antibody effector functions in the immune system of the transgenic mouse, and thus B cell development, are substantially unchanged, which may lead to an improved antibody response to antigen challenge in vivo. Once the genes encoding the particular antibody of interest have been isolated from such transgenic animals, the genes encoding the constant regions can be replaced with human constant region genes to produce a wholly human antibody. Other in vitro methods for producing human antibodies are based on display technologies, such as phage display or ribosome display technologies, using recombinant DNA libraries that are obtained at least in part artificially or from donor immunoglobulin variable (V) domain gene repertoires. Phage and ribosome display technologies for producing human antibodies are well known in the art. Human antibodies can also be obtained from isolated human B cells that are immunized ex vivo with an antigen of interest and then fused to form hybridomas that can then be screened for the optimal human antibody. As used herein, the term antibody or antibodies also refers to aglycosylated antibodies.

Antibody molecules for use in any of the embodiments of the invention include

- 9 041325 antibody fragments such as Fab, Fab', F(ab') ₂ and Fv and scFv fragments, as well as diabodies, including formats such as BiTE® (T-cell-attracting bispecific activators) and DART™ (redirector technology). dual affinity antibodies), tribodies, tetrabodies, minibodies, domain antibodies (dAbs) such as sdAb, VHH and VNAR fragments, single chain antibodies, bispecific, trispecific, tetraspecific or polyspecific antibodies derived from antibody fragments, or antibodies comprising as non-limiting examples of the construct Fab-Fv, Fab-scFv, Fab(Fv)2 or Fab-(scFv)2. Antibody fragments as defined above are known in the art. For clarity, it should be understood that Fab-Fv refers to a construct containing one Fv region and one Fab region linked in any order, i.e. Fab-Fv or Fv-Fab, where the last amino acids in one region are followed by the first amino acids in the next region, or vice versa. Similarly, a Fab-scFv should be understood to refer to a construct containing one scFv region and one Fab region linked in any order, and in the case of Fab, to any of the remaining polypeptide chains, i.e. Fab-scFv or scFv-Fab, where the last amino acid in one region is followed by the first amino acid in the next region, or vice versa. Similarly, Fab-(Fv)2 should be understood to refer to a construct containing two Fv regions and one Fab region linked in any order, i.e. Fab-Fv-Fv, Fv-Fab-Fv, or Fv-Fv-Fab, where the last amino acids in one region are followed by the first amino acids in the next region, or vice versa. Similarly, Fab-(scFv) ₂ should be understood to refer to a construct containing two scFv regions and one Fab region linked in any order, and in the case of Fab, to any of the remaining polypeptide chains, resulting in 20 possible combinations.

Typically, these constructs between the first region (eg Fab) and the second region (eg Fv) contain a peptide linker. Such linkers are well known in the art and may be one or more amino acids, typically optimized by one skilled in the art for length and composition. Alternatively, said regions can be linked directly, ie. without peptide linker.

Examples of suitable linker regions for linking a variable domain to a Fab or Fab' are described in WO 2013/068571 and WO 2014/096390, incorporated herein by reference, and include movable linker sequences and rigid linker sequences as non-limiting examples. Flexible linker sequences include those described in Huston et al., 1988, 10 PNAS, 85:5879-5883; Wright & Deonarain, Mol. Immunol., 2007, 44(11):2860-2869; Alfthan et al., Prot. Eng., 1995, 8(7):725-731; Luo et al., J. Biochem., 1995, 118(4):825-831; Tang et al., 1996, J. Biol. Chem. 271(26):15682-15686 and Turner et al., 1997, JIMM 205, 42-54.

As used herein, the term VH3 domain refers to framework 3 subset of the human immunoglobulin heavy chain variable region. Antibody heavy chain variable domains are classified into distinct subfamilies (VH1 to VH6) based on DNA sequences and protein homologies (Walter et al. Am. J. Hum. Genet. 42:446-451, 1988, Analysis for genetic variation reveals human immunoglobulin VH-region gene organization Schroeder et al Int Immunol 1990 2(1):41-50, Structure and evolution of mammalian VH families.

As used herein, the term Fc region refers to the Fc region of a natural antibody, which is a constant region dimer lacking the heavy chain 1 (CH1) constant domain. As is known in the art, the Fc region of an antibody is a crystallizable fragment (Fc) obtained after cleavage of a native antibody with pepsin or papain.

As used herein, an antibody that does not contain an Fc region refers to an antibody that does not contain natural heavy chain constant domain 2 (CH2), natural heavy chain constant domain 3 (CH3), and natural heavy chain constant domain 4 ( CH4).

Residues of the Fc region responsible for binding to protein A have previously been described in this area (Nagaoka et al. Single amino acid substitution in the mouse IgG1 Fc region induces drastic enhancement of the affinity to protein A, PEDS. Vol. 16, Issue 4, pages 243-245). Thus, a person skilled in the art can design an antibody with an Fc region that has lost the ability to bind to protein A, and such an antibody will be suitable for use in the method of the present invention.

As used herein, the term multimer or multimeric form refers to forms of antibodies consisting of the domains of two or more monomers in which all domains are correctly folded and paired. Examples of multimers are provided in FIG. 12, where various antibody molecules are correctly folded and each VH domain is paired with a complementary VL domain. For clarity, it should be understood that VH-VL complementary pairs cooperatively bind the same antigen.

As used herein, the term Protein A, or Staphylococcal Protein A, is a type I membrane protein covalently bound to the cell wall of most strains of the gram-positive bacterium Staphylococcus aureus. It has high affinity for IgG of various species, such as human, rabbit and guinea pig, but only weak interaction with ruminant and mouse IgG. Protein A interacts with antibodies in two specific binding events: the classical binding site on the Fc portion of human IgG1, IgG2 and IgG4 and the alternative site

- 10 041325 binding, located on the Fab part of human IgG, IgM, IgA and IgE, which contain the heavy chains of the VH3 subfamily. The most commonly reported molecular weight of Staphylococcus aureus protein A is approximately 42,000 Da. Recombinant streptococcal protein A consists of 299 amino acids and has a theoretical molecular weight of 33.8 kDa as estimated by SDS-PAGE.

Protein A consists of three regions: S, which is a signal sequence that is processed during secretion; five homologous IgG-binding domains E, D, A, B and C and an XM anchoring region in the cell wall. The truncated protein lacking the X region has a molecular weight of approximately 31 kDa. The domains are able to bind independently to the Fc portion of IgG1, IgG ₂ and IgG ₄ but show only weak interaction with IgG ₃ . In addition, all natural protein A domains show comparable binding to Fab, which is described to be mediated by the D and E regions.

As used herein, the term OX40 refers to a molecule also known as CD134, TNFRSF4, ACT35, or TXGP1L, which is a member of the TNF receptor superfamily, acting as a co-stimulatory receptor, with subsequent recruitment of CD28 and OX40 required for optimal T-cell proliferation and survival. cells.

As used herein, the terms specific binds to, specific binding to a given molecule, and their antibody equivalents mean that the antibody binds to that specified molecule with sufficient affinity and specificity to achieve a biologically significant effect. The antibody of choice will typically have a binding affinity for that molecule, for example an antibody can bind to a given molecule with a Kd value between 100 nM and 1 pM. Antibody affinities can be determined, for example, by a surface plasmon resonance assay such as a BIAcore assay; enzyme-linked immunosorbent assay (ELISA); and competitive assays (eg, RIA). Within the meaning of the present invention, an antibody that specifically binds to this specified molecule can also bind to another molecule, for example, as a non-limiting example, as in the case of a bispecific antibody.

Examples

Example 1 Purification of A26Fab-645dsFv with Protein A by elution with a pH gradient.

Expression, isolation and purification of A26Fab-645dsFv using E. coli.

A26Fab-645dsFv (an antibody fragment that binds to human OX40 and serum albumin) was expressed as a heterologous protein in E. coli W3110 host cells after induction with IPTG (isopropyl-β-D-1-thiogalactopyranoside) and the heterologous protein was isolated from the periplasmic space host cells by adding 100 mM Tris/10 mM EDTA buffer adjusted to pH 7.4 and protein extraction step at 30°C. Cellular material was removed by centrifugation, and then the cell extract containing the heterologous protein was purified from impurities using a combination of centrifugation and filtration through a 0.22 μm filter.

Purification of A26Fab-645dsFv with Protein A by pH gradient elution.

The decontaminated E. coli extract was applied to a 5 ml HiTrap MabSelect (GE Healthcare) protein A chromatography column equilibrated in Dulbecco's phosphate-buffered saline (PBS), pH 7.4. First, the column was washed with PBS to remove unbound material, and then the bound material was eluted with a pH gradient from 7.4 to 2.1. The eluted material was fractionated and analyzed by size exclusion HPLC (size exclusion chromatography-high performance liquid chromatography) using size exclusion HPLC with TSK gel G3000SWXL (Tosoh Corporation). SEC HPLC analysis was used to determine the % monomers and multimers present in each fraction. The monomer retention time of A26Fab-645dsFv was approximately 9 minutes. Dimer, trimer, tetramer and higher order structures exhibited retention times of less than 9 minutes and were collectively referred to as multimeric or higher molecular weight molecules (HMWS).

Elution from the protein A chromatography column showed two peaks along the pH gradient, see FIG. 1. Higher levels of monomer were observed in the first peak compared to higher levels of multimeric molecules in the second peak, see table. 1 and FIG. 2.

A26Fab-645dsFv lacks an Fc, thus protein A binding occurs via the framework of the VH3 variable region subclass of human V regions. It is believed that the increased binding of multimeric molecules was due to the increased avidity of these molecules to protein A. Multimeric molecules contain more VH3 regions and thus bind more strongly to the protein A resin, requiring a lower pH to elute.

- 11 041325

Table 1

SEC analysis of G3000 fractions after purification of A26Fab-645dsFv with protein A

when eluting with g pH Radiant Fraction Volume (ml) Conc, (mg/ml) Protein (mg) HMWS (%) Monomer (%) A1 3.3 0.3 0.9 4.0 96.1 A2 3.1 0.6 1.9 6, 7 93.3 AZ 3.2 1.6 5.0 15.0 85.0 A4 3.2 1.6 5.1 22.7 77.3 A5 3.2 1.8 5.8 44.2 55.8 Ab 3.3 1.4 4.6 53.8 46.2 A7 3.3 0.5 1.5 54.7 45.3 A8 1, oh 0.2 0.2 50.6 49.4

Example 2 Purification of an Fc Construct with Protein A by pH Gradient Elution.

Expression and decontamination of multimeric Fc using CHO.

An Fc construct containing a human IgG1 Fc domain fused to a human IgM terminal that induces assembly of Fc into multimers was expressed in a Chinese hamster cell line (CHO DG44) with stable dihydrofolate reductase (DHFR) deficiency. Cells were transfected using Nuclefector (Lonza) according to the manufacturer's instructions with a plasmid vector containing the DHFR gene as a selectable marker and genes encoding the product. Transfected cells were selected in a medium without hypoxanthine and thymidine and in the presence of the DHFR inhibitor methotrexate. Cultivation was carried out in stirred flasks in a periodic mode and the collection was carried out after 14 days.

Purification of the cell culture supernatant from impurities was carried out by centrifugation (4000xg for 60 min at room temperature) followed by deep and sterile filtration.

The cell culture supernatant with impurities removed was concentrated and all constructs containing Fc were purified using a 5 ml HiTrap MabSelect SuRe protein A chromatography column (GE Healthcare) equilibrated in PBS pH 7.4. The column was washed with PBS and the bound material was eluted with 0.1 M citrate pH 3.4. The eluted material was buffered with PBS pH 7.4.

Purified Fc constructs were applied to a 5 ml HiTrap MabSelect protein A chromatography column (GE Healthcare) equilibrated in PBS pH 7.4. First, the column was washed with PBS to remove unbound material, and then the bound material was eluted with a pH gradient, pH 7.4 to 2.1, see FIG. 3. The eluted material was fractionated and analyzed by size exclusion HPLC (G3000 size exclusion HPLC, Tosoh Corporation). SEC HPLC analysis was used to determine the % monomeric and multimeric Fc present in each fraction. The Fc monomer retention time was approximately 9.3 minutes. Trimer, hexamer, and higher order retention times were less than 9.4 minutes and were collectively referred to as multimeric or higher molecular weight molecules (HMWS), see Table 1. 2 and FIG. 4.

table 2

Analysis of SEC G3000 fractions after purification of multimeric Fc with protein A by elution with a pH gradient

Fraction Total protein (mg) % monomeric Fc % multimeric Fc A1 0.89 2.3 97.6 A2 3.13 4.7 95.3 AZ 3.14 9.5 90.4 A4 0.94 7.8 92.3

The elution of the Fc construct proceeded as a single peak with small shoulders at the upper and lower kinks. However, analysis of the fractions by size exclusion HPLC revealed that the monomeric and multimeric forms of the molecule elute in parallel, see table. 2 and FIG. 4. Separation of different molecules during gradient elution was not observed.

The Fc construct multimers contain several Fc regions but no variable regions and therefore lack the ability to bind protein A through the human VH3 domain. The monomeric and multimeric Fc constructs co-elute from the Protein A column in a gradient elution, demonstrating that the avidity of the Fc region to the resin is not indicative of elution. Thus, molecules containing a monomeric and multimeric Fc region cannot be efficiently separated by this method. This is in contrast to Examples 1 and 2, where eluting VH3 domain-containing antibodies lacking an Fc region from Protein A material could separate monomers from multimers using a pH gradient.

- 12 041325

Example 3 Purification of A26Fab-645dsFv with Protein A by pH stepping elution.

Expression and purification of A26Fab-645dsFv impurities using CHO.

A construct that binds to human OX40 and serum albumin was expressed in a Chinese hamster cell line (CHO DG44) with stable dihydrofolate reductase (DHFR) deficiency. Cells were transfected by electroporation using Nuclefector (Lonza) according to the manufacturer's instructions with a plasmid vector containing the DHFR gene as a selectable marker and genes encoding the product. Transfected cells were selected in a medium without hypoxanthine and thymidine and in the presence of the DHFR inhibitor methotrexate. Cultivation was carried out in stirred flasks in a periodic mode and the collection was carried out after 14 days.

Purification of the cell culture supernatant from impurities was carried out by centrifugation (4000xg for 60 min at room temperature) followed by deep and sterile filtration.

Purification of A26Fab-645dsFv with Protein A by pH stepping elution.

The decontaminated cell culture supernatant was applied to a 5 ml HiTrap MabSelect (GE Healthcare) protein A chromatographic column equilibrated in Dulbecco's phosphate-buffered saline (PBS) pH 7.4. The column was first washed with PBS to remove unbound material and then the bound material was eluted first at pH 3.8 followed by a second elution step at pH 3.0, see FIG. 5. The eluted material was fractionated and analyzed by size exclusion HPLC (size exclusion HPLC with G3000, Tosoh Corporation) and SDS-PAGE was performed with 4-20% Tris/glycine under reducing and non-reducing conditions.

SEC HPLC analysis was used to determine the % monomers and multimers present in each fraction. The A26Fab-645dsFv monomer retention time was approximately 9 minutes. Dimer, trimer, tetramer and higher order retention times were less than 9 minutes and were collectively referred to as multimeric or higher molecular weight molecules (HMWS), see FIG. 6 and tab. 3.

Table 3

SEC analysis of G3000 fractions after purification of A26Fab-645dsFv with protein A by pH stepping elution

Volume (ml) Conc, (mg/ml) Total protein (mg) HMWS (%) Monomer (%) pHZ, 8 13.8 0.67 9.2 21.3 78.7 pHZ, 0 7.5 1.04 7.8 96.8 3.2

Non-reducing SDS-PAGE confirmed the levels of monomer and multimers obtained above in the elution peaks. A26Fab-645dsFv monomer migrated in the range of 97-66 kDa and was the main band in lane 4, corresponding to elution at pH 3.8, A26Fab-645dsFv multimers migrated as several bands over 120 kDa and were the main components of the bands in lane 5, corresponding to the fraction obtained after elution at pH 3.0, see FIG. 7.

Reducing SDS-PAGE confirmed that all bands in non-reducing SDS-PAGE were for A26Fab-645dsFv, see FIG. 8.

After elution at pH 3.8, a single peak was obtained in 2.8 column volumes with a slight tail at the bottom inflection. Analysis by size exclusion HPLC showed that this peak contained 79% A26Fab-645dsFv in monomeric form, a high amount of monomer was also confirmed by non-reducing SDS-PAGE analysis. After elution at pH 3.0 one peak was obtained. Size-exclusion HPLC analysis showed that this peak was 97% A26Fab-645dsFv in multimeric form, a large amount of multimer was also confirmed by non-reducing SDS-PAGE analysis.

The above results demonstrate that efficient separation of monomer and multimer molecules by VH3 binding to protein A is possible. This is different from Fc binding to protein A as demonstrated in the previous example.

Example 4 Purification of A26Fab-645dsFv with Protein A (Amsphere) by gradient elution.

Expression and purification of A26Fab-645dsFv impurities using CHO.

A construct that binds to human OX40 and serum albumin was expressed in a Chinese hamster cell line (CHO DG44) with stable dihydrofolate reductase (DHFR) deficiency. Cells were transfected by electroporation using Nuclefector (Lonza) according to the manufacturer's instructions with a plasmid vector containing the DHFR gene as a selectable marker and genes encoding the product. Transfected cells were selected in a medium without hypoxanthine and thymidine and in the presence of the DHFR inhibitor methotrexate. Cultivation was carried out in stirred flasks in a periodic mode and the collection was carried out after 14 days.

Purification of the cell culture supernatant from impurities was carried out by centrifugation (4000xg for 60 min at room temperature) followed by deep and sterile filtration.

- 13 041325

Purification of A26Fab-645dsFv with Protein A (Amsphere) by pH gradient elution.

The supernatant purified from impurities was applied to an Amsphere protein A chromatographic column (column volume 5 ml with a layer height of 10 cm) and equilibrated in Dulbecco's phosphate-buffered saline (PBS) pH 7.4. The column was then washed with PBS to remove unbound material, and then the bound material was eluted with a pH gradient, pH 6.0 to 2.1. The eluted material was fractionated and analyzed by SEC (size-exclusion chromatography, ultra high performance liquid chromatography) using a BEH450 2.5 μm Acquity SEC-SELC column. SEC SALC analysis was used to determine the % monomers and multimers present in each fraction. The monomer retention time of A26Fab-645dsFv was approximately 2.2 minutes. The dimer, trimer, tetramer, and higher order structures exhibited retention times of less than 2.2 minutes and were collectively referred to as multimeric or higher molecular weight molecules (HMWS).

Elution from the Amsphere Protein A chromatography column showed two peaks along the pH gradient, see FIG. 9. Higher levels of monomer were observed in the first peak compared to higher levels of multimeric molecules in the second peak, see table. 4 and FIG. 9.

A26Fab-645dsFv lacks an Fc, thus binding to Amsphere protein A occurred via the framework of the VH3 variable region subclass of human V regions. It is believed that the increased binding of multimeric molecules was due to the increased avidity of these molecules to protein A. Multimeric molecules contain more VH3 regions and thus bind more strongly to the protein A resin, requiring a lower pH to elute.

Table 4

Analysis of fractions after purification of A26Fab-645dsFv with Amsphere Protein A by elution with a pH gradient by size exclusion chromatography

Fraction Volume (ml) Conc. (mg/ml) Protein (mg) HMWS (%) Monomer (%) B15 5 0.54 2.68 0.0 85.2 B14 5 1.01 5.05 1.8 95.7 B13 5 1.14 5.7 48.2 49.7 AT 12 5 3.32 16, 61 96.1 3.9 AT 11 5 1.85 9.235 96.1 2.8 AT 10 5 0.51 2.535 79.9 7.0

Example 5 Purification of A26Fab-645dsFv with Protein A (NovaSep Absolute) by gradient elution.

Expression and purification of A26Fab-645dsFv impurities using CHO.

A construct that binds to human OX40 and serum albumin was expressed in a Chinese hamster cell line (CHO DG44) with stable dihydrofolate reductase (DHFR) deficiency. Cells were transfected by electroporation using Nuclefector (Lonza) according to the manufacturer's instructions with a plasmid vector containing the DHFR gene as a selectable marker and genes encoding the product. Transfected cells were selected in a medium without hypoxanthine and thymidine and in the presence of the DHFR inhibitor methotrexate. Cultivation was carried out in stirred flasks in a periodic mode and the collection was carried out after 14 days.

Purification of the cell culture supernatant from impurities was carried out by centrifugation (4000xg for 60 min at room temperature) followed by deep and sterile filtration.

Purification of A26Fab-645dsFv with Protein A (NovaSep Absolute) by pH gradient elution.

The supernatant purified from impurities was applied to a NovaSep Absolute protein A chromatographic column (column volume 5 ml, layer height 10 cm) and equilibrated in Dulbecco's phosphate-buffered saline (PBS) pH 7.4. First, the column was washed with PBS to remove unbound material, and then the bound material was eluted with a pH gradient, pH 6.0 to 3.0. The eluted material was fractionated and analyzed by size exclusion HPLC using a BEH450 2.5 μm Acquity SEC HPLC column. SEC SALC analysis was used to determine the % monomers and multimers present in each fraction. The monomer retention time of A26Fab-645dsFv was approximately 2.2 minutes. The dimer, trimer, tetramer, and higher order structures exhibited retention times of less than 2.2 minutes and were collectively referred to as multimeric or higher molecular weight molecules (HMWS).

Elution from the NovaSep Absolute protein A chromatography column showed two peaks along the pH gradient, see FIG. 10. Higher levels of monomer were observed in the first peak compared to higher levels of multimeric molecules in the second peak, see table. 5 and FIG. 10.

A26Fab-645dsFv lacks an Fc, thus binding to NovaSep Absolute protein A occurred via the VH3 variable region subclass scaffold of human V regions. It is believed that the increased binding of multimeric molecules was due to the increased avidity of these

- 14,041,325 molecules to Protein A. Multimeric molecules contain more VH3 regions and thus bind more strongly to the Protein A resin, requiring a lower pH to elute.

Table 5

Fraction exclusion analysis after purification of A26Fab-645dsFv with protein A NovaSep Absolute by pH gradient elution

Fraction Volume (ml) Conc, (mg/ml) Protein (mg) HMWS (%) Monomer (%) B15 2 0.36 0.72 0 99.7 B14 2 1.21 2.42 0 99.7 B13 2 0.46 0.92 4.6 94.8 AT 12 2 0.64 1.28 87.3 12.2 AT 11 2 2.05 4.1 98.4 1.4 AT 10 2 2.68 5.36 99 0.8 AT 9 2 0.67 1.34 97.2 2.3 AT 8 2 0.37 0.74 94.8 4.3 wb 2 0.03 0.06 94.7 4

Example 6 Purification of A26Fab-645dsFv with Protein A (AcroSep) by gradient elution.

Expression and purification of A26Fab-645dsFv impurities using CHO.

A construct that binds to human OX40 and serum albumin was expressed in a Chinese hamster cell line (CHO DG44) with stable dihydrofolate reductase (DHFR) deficiency. Cells were transfected by electroporation using Nuclefector (Lonza) according to the manufacturer's instructions with a plasmid vector containing the DHFR gene as a selectable marker and genes encoding the product. Transfected cells were selected in a medium without hypoxanthine and thymidine and in the presence of the DHFR inhibitor methotrexate. Cultivation was carried out in stirred flasks in a periodic mode and the collection was carried out after 14 days.

Purification of the cell culture supernatant from impurities was carried out by centrifugation (4000xg for 60 min at room temperature) followed by deep and sterile filtration.

Purification (AcroSep) of A26Fab-645dsFv with Protein A by pH gradient elution.

The decontaminated supernatant was applied to an AcroSep protein A chromatographic column (1 ml volume column with a layer height of 1.5 cm) and equilibrated in Dulbecco's phosphate-buffered saline (PBS) pH 7.4. First, the column was washed with PBS to remove unbound material, and then the bound material was eluted with a pH gradient, pH 6.0 to 3.0. The eluted material was fractionated and analyzed by size exclusion HPLC using Acquity BEH450 2.5 µm SEC HPLC column. SEC HPLC analysis was used to determine the % monomers and multimers present in each fraction. The monomer retention time of A26Fab-645dsFv was approximately 2.2 minutes. The dimer, trimer, tetramer and higher order structures exhibited retention times greater than 2.2 minutes and were collectively referred to as multimeric or higher molecular weight molecules (HMWS).

Elution from the AcroSep Protein A chromatography column showed one broad peak eluting with a pH gradient, see FIG. 11. The reduced resolution may have been due to the reduced layer height, however, increased levels of monomer were still observed at the upper inflection of the peak compared to the increased levels of multimeric molecules at the lower inflection of the elution peak, see table. 6 and FIG. eleven.

A26Fab-645dsFv lacks an Fc, thus binding to AcroSep Absolute protein A occurred via the VH3 variable region subclass scaffold of human V regions. It is believed that the increased binding of multimeric molecules was due to the increased avidity of these molecules to protein A. Multimeric molecules contain more VH3 regions and thus bind more strongly to the protein A resin, requiring a lower pH to elute.

- 15 041325

Table 6

Fraction exclusion analysis after purification of A26Fab-645dsFv with AcroSep Absolute Protein A by pH gradient elution

Fraction Volume (ml) Conc, (mg/ml) Protein (mg) HMWS (%) Monomer (%) B13 1 0.17 0.17 28.9 71.1 AT 12 1 0.22 0.22 44.6 55 AT 11 1 0.35 0.35 67.1 32.8 AT 10 1 0.52 0.52 87.4 12.4 AT 9 1 0.73 0.73 96.2 3.6 AT 8 1 1.00 1.00 98.6 1.2 AT 7 1 1.19 1.19 99.1 0.7 wb 1 1.05 1.05 99.2 0.6 AT 5 1 0.54 0.54 98.6 0.9 AT 4 1 0.22 0.22 97.5 1.4

Example 7 Purification of Trybe® with Protein A by elution with a pH gradient.

Purification of Trybe® with Protein A by elution with a pH gradient.

Expression and decontamination of Trybe® using CHO.

A multispecific trivalent antibody molecule of the Fab-2x dsscFv format as described in WO 2015/197772 (TrYbe®) was expressed in a Chinese hamster (CHO DG44) cell line with stable dihydrofolate reductase (DHFR) deficiency. Cells were transfected using Nuclefector (Lonza) according to the manufacturer's instructions with a plasmid vector containing the DHFR gene as a selectable marker and genes encoding the product. Transfected cells were selected in a medium without hypoxanthine and thymidine and in the presence of the DHFR inhibitor methotrexate. The expression was carried out in its own way with periodic feeding to obtain a large number of cells.

Purification of the cell culture supernatant from impurities was carried out by centrifugation (4000xg for 60 min at room temperature) followed by deep and sterile filtration. The decontaminated cell culture supernatant was applied to 5 ml HiTrap MabSelect (GE Healthcare) equilibrated in Dulbecco's phosphate buffered saline (PBS) pH 7.4. The column was washed with PBS and the bound material was eluted with a pH gradient, pH 7.4 to 2.1, see FIG. 21. The eluted material was fractionated and analyzed by size exclusion HPLC G3000 and SDS-PAGE with 4-20% Tris/glycine (reducing and non-reducing). SEC HPLC analysis was used to determine the % monomers and multimers. The TrYbe® monomer retention time was approximately 9.4 minutes. Dimer, trimer, tetramer and higher order retention times were less than 9.4 minutes and were collectively referred to as multimer molecules or HMWS. For analysis of HPLC size exclusion chromatograms, see FIG. 22.

Elution with a pH gradient showed two peaks with a shoulder at the lower inflection, see FIG. 21. These three fractions were analyzed by SDS-PAGE, see FIG. 23. Several bands were observed in the elution profile on the SDS-PAGE recovery. The monomer migrated in the range of molecular weight bands 116-200 kDa. All bands that migrate above the monomer are collectively referred to as multimer molecules or HMWS. Light chain impurities with no interchain disulfide bonds and no heavy and light chain disulfide bonds migrate in the molecular weight marker range of 37-55 kDa. Fraction 1 (lane 2) predominantly consisted of monomer and light chain molecules with little or no visible HMWS. In fraction 2 (lane 3), the monomer makes up the main band. Although the HMWS bands are visible, the levels are significantly reduced relative to the levels observed in fraction 3 (lane 4). In the reducing gel, all product related molecules are reduced to heavy and light chains with a small band of non-reducible material visible in lanes 3 and 4. Lane 2 confirms that fraction 1 is highly enriched in light chain related molecules.

The first peak (1) was identified by SDS-PAGE as light chain impurities and contained 2% HMWS when analyzed by size exclusion HPLC. The second peak (2) was identified as TrYbe® and contained 72% monomer and 22% HMWS, see table. 7 and FIG. 24. The bottom fold at peak 2, fraction 3, was identified as TrYbe® and contained 6% monomer and 92% HMWS.

Fc is absent in TrYbe®, thus protein A binding occurred via the framework of the VH3 variable region subclass of human V regions. It is believed that the increased binding of multimeric molecules was due to the increased avidity of these molecules to protein A. Multimeric molecules contain more VH3 regions and thus bind more strongly to the protein A resin, requiring a lower pH to elute.

- 16 041325

Table 7

SEC G3000 Analysis of Fractions After Purification of Trybe® with Protein A by pH Gradient Elution

Fraction Protein (mg) HMWS (%) Monomer (%) 1 3.9 2.0 97.3 2 33, 9 22.1 72.0 3 0.4 91.5 6, 4

Example 8 Purification of BYbe with Protein A by elution with a pH gradient.

Expression and purification of BYbe impurities using CHO.

The Fab-scFv (Bybe) fusion protein was constructed essentially as described in Example 4 of WO 2013/068571 using different variable region sequences. The construct was expressed in a Chinese hamster (CHO DG44) cell line with stable dihydrofolate reductase (DHFR) deficiency. Cells were transfected using Nuclefector (Lonza) according to the manufacturer's instructions with a plasmid vector containing the DHFR gene as a selectable marker and genes encoding the product. Transfected cells were selected in a medium without hypoxanthine and thymidine and in the presence of the DHFR inhibitor methotrexate. After culture in the shake flask step, growth and production were evaluated and the 24 most expressing clones were selected for evaluation by the shake flask method with fed-batch. A 3 L stirred flask was inoculated with 1 L of culture at an initial density of 0.3×10 ⁶ viable cells/mL and maintained at 36.8° C. under 5% CO ₂ . Nutrient supplementation was performed from day 3 to day 12 and glucose was added as a bolus supplement when the concentration fell below 5.8 g/l. The culture was harvested on day 14 by centrifugation at 4000xg for 60 min followed by 0.2 µm filtration.

Purification of BYbe with protein A by elution with a pH gradient.

The decontaminated cell culture supernatant was applied to a 4.7 ml HiScreen MabSelect column (GE Healthcare) equilibrated in phosphate buffered saline (PBS) Sigma pH 7.4. The column was washed with PBS followed by 90% 0.2 M sodium phosphate/10% citric acid, pH 7.4, and the bound material was eluted with a pH gradient of 7.4 to 2.1, see FIG. 25. The eluted material was fractionated and analyzed by size exclusion HPLC G3000 and SDS-PAGE with 4-20% Tris/glycine (reducing and non-reducing). Analysis by size exclusion HPLC was used to determine the % of monomers and multimers. The BYbe monomer retention time was approximately 9.6 minutes. Dimer, trimer, tetramer and higher order retention times were <9.6 min and were collectively referred to as multimer molecules or HMWS. For analysis of fractions by size exclusion HPLC chromatograms, see FIG. 26.

A single peak was observed eluting over the pH gradient, see FIG. 25. The peak elution fractions were analyzed by SDS-PAGE, see FIG. 27. On non-reducing SDS-PAGE, several bands were observed in the elution profile. The monomer migrated close to the level of the 98 kD molecular weight marker. All bands migrating above the monomer (in the range of molecular weight markers from 250 to 98 kDa) are collectively designated multimeric molecules or HMWS. Heavy and light chains not bound by disulfide bonds migrate at the level of markers of molecular weights of 50-64 kDa and 30 kDa, respectively. Fractions B9-B6 (lanes 6-9) were predominantly monomer with little or no visible HMWS. In fractions B5-B3 (lanes 10-12) there was an increase/increase in the intensity of the HMWS bands as the acidity of the pH gradient increased. In the reducing gel, all molecules related to the product were reduced to heavy and light chains.

The elution peak was identified as BYbe and contained a total of 84% monomer and 16% HMWS, see table. 8 and FIG. 28.

BYbe lacks Fc, thus binding to the A protein occurred via the framework of the VH3 variable region subclass of human V regions. The increased binding of the multimeric molecules is thought to be due to the increased avidity of these molecules to protein A. The multimeric molecules contain more VH3 regions and thus bind more strongly to the protein A resin, requiring a lower pH to elute.

-

Claims (10)

Table 8 SEC G3000 Analysis of Fractions After Purification of BYbe with Protein A by pH Gradient Elution Fraction Volume (ml) Concentration (mg/ml) Protein (mg) HMWS (%) Monomer (%) B9 2.12 0.026 0.055 0.0 100.0 B8 2.14 0.103 0.220 0.0 100.0 B7 2.16 0.527 1.138 0.0 100.0 B6 2.18 2.395 5.221 0.7 99.3 B5 2.20 4,290 9.438 5.4 94.6 B4 2.22 1.886 4.187 28.8 71.2 B3 2.24 0.308 0.690 52.2 47.8 B2 2.26 0.047 0.106 40.8 59.2 Average: 16.0 84.0 CLAIM 1. A method for separating an antibody containing a human VH3 domain in monomeric form from an antibody in multimeric form, comprising: a) applying a mixture containing an antibody containing a human VH3 domain, in monomeric and multimeric form, to a protein A-containing chromatographic material, where said protein A contains a D and/or E domain; b) allowing said antibody to bind to protein A; c) applying an elution buffer that selectively destroys the binding of the antibody in monomeric form; d) obtaining the resulting eluate, where the antibody containing the human VH3 domain does not contain the Fc region; where an antibody containing a human VH3 domain contains at least two human VH3 domains. 2. The method according to claim 1, wherein the elution buffer in step c) has a pH of 3.0 to 4.5. 3. The method according to claim 1 or 2, further comprising applying an elution buffer that selectively destroys the binding of the antibody in multimeric form after protein A chromatography and obtaining the resulting eluate. 4. The method according to any one of the preceding claims, wherein a first solution is added to the protein A-containing chromatographic material after application of a mixture containing an antibody containing a human VH3 domain in monomeric and multimeric form, such that the unbound material goes into solution. 5. The method of claim 1 or 2, wherein the elution buffer is applied to the protein A-containing chromatographic material such that the bound antibody is released. 6. The method of any one of the preceding claims, wherein said protein A is a native recombinant protein A. 7. The method according to any of the preceding claims, wherein the VH3 domain containing antibody is selected from F(ab') ₂ , Fab-Fv, Fab-scFv, Fab-(scFv) ₂ , Fab-(Fv) ₂ , Fab-dsFv , diatel, triotel and tetratel. 8. The method of any one of the preceding claims, wherein the VH3 domain-containing antibody comprises a heavy chain variable domain as set forth in SEQ ID NO: 11 and a light chain variable domain as set forth in SEQ ID NO: 12. 9. The method of any one of the preceding claims, wherein the VH3 domain-containing antibody specifically binds to OX40. 10. The method according to claim 9, where the specified antibody containing the VH3 domain contains: Heavy chain CDR1, CDR2 and CDR3 as defined in SEQ ID NO: 1, SEQ ID NO: 2 and SEQ ID NO: 3, respectively; And Light chain CDR1, CDR2 and CDR3 as defined in SEQ ID NO: 4, SEQ ID NO: 5 and SEQ ID NO: 6, respectively. -