WO2021208839A1

WO2021208839A1 - Purification of bispeciifc antibodies

Info

Publication number: WO2021208839A1
Application number: PCT/CN2021/086488
Authority: WO
Inventors: Yifeng Li; Ying Wang; Gaili GUO; Jinyu Han
Original assignee: Wuxi Biologics (Shanghai) Co., Ltd.; WuXi Biologics Ireland Limited
Priority date: 2020-04-13
Filing date: 2021-04-12
Publication date: 2021-10-21
Also published as: CN115734969A

Abstract

Provided is a process of purifying a bispecific antibody comprising a TCR constant domain from a fluid comprising said bispecific antibody and potential impurities, wherein said process comprises subjecting said fluid to a Protein A chromatography followed by an anion exchange chromatography and a mixed-mode chromatography in the specified order, thereby to obtain a harvest of purified bispecific antibody; wherein the anion exchange chromatography is conducted in bind-elute mode.

Description

PURIFICATION OF BISPECIIFC ANTIBODIES

FIELD OF THE INVENTION

The present invention is directed to methods of purifying a bispecific antibody (bsAb) containing a TCR constant domain, particularly a bsAbs according to the WuXiBody ^TM technology.

BACKGROUND

Bispecific antibodies (bsAbs) are artificial antibodies that can bind two epitopes on the same or a distinct target. Their capability to simultaneously engage two targets enables novel mechanisms of action (e.g., dual inhibition of signaling circuits and recruitment of effector cells) . Thus, bsAbs have emerged as promising tools for the treatment of cancers and other diseases. Based on their formats, bsAbs can be broadly divided into two main categories: IgG-like molecules comprising an Fc region and smaller non-IgG-like molecules lacking the Fc region. Each has its own advantages and limits. For instance, while the Fc-free bsAbs exhibit enhanced penetration into solid tumours, IgG-like bsAbs are advantageously capable of providing a longer circulation half-life and supporting secondary immune functions. In general, construction and production of IgG-like bsAbs, which represent greater complexity, is technically more challenging than that of the Fc-free counterparts.

IgG-like bsAbs can be further divided into two subgroups in accordance with their structural symmetry. Asymmetric IgG-like bsAbs are normally derived from two parental monoclonal antibodies (mAbs) with different binding specificity and consequently contain four distinct chains: two different heavy chains (HCs) and two different light chains (LCs) . When these chains are coexpressed in the producing cell, misassembled species, such as homodimers and molecules containing mispaired LCs, could account for about 90%of the total mass if they are allowed to pair randomly [1] . To address this problem, a variety of protein engineering approaches have been developed to enforce desired chain pairing [2] . Among them, knobs-into-holes (KiH) and CrossMab technologies are the most widely used ones for overcoming the HC homodimerization and LC mispairing problems, respectively [3, 4] . For IgG-like bsAbs adopting a symmetric format, bispecificity is usually achieved by fusion of a second antigen-binding unit to either terminus of the HC or LC [1, 2] . As symmetric bsAbs achieve their functionality through chain extension rather than introducing distinct chains, they normally have fewer number of associated byproducts than asymmetric bsAbs. Nevertheless, symmetric bsAbs have shown significantly increased tendency to form aggregates [5, 6] , which was once reported in a case to be present at a level as high as 50% [6] . Aggregates are likely formed through intermolecular domain swapping as a result of increased chain length and flexibility.

WuXiBody ^TM (also referred to as "WuXiBody" herein below) is an innovative bsAb platform developed by WuXi Biologics. Its key feature is the replacement of one parental mAb’s CH1/CL constant domain with the T cell receptor (TCR) constant domain [7] . WuXiBody ^TM design ensures cognate HC-LC pairing, the same goal as that being aimed by the CrossMab technology. BsAbs based on WuXiBody can adopt either asymmetric or symmetric format (Fig. 1) . For asymmetric WuXiBody-based bsAbs, heterodimerization is promoted by the KiH technology.

Still, impurities exist and need be addressed. Accordingly, there is a need for new and/or improved schemes for purifying a target bsAb from impurities, which allows the final purified product to meet the biotechnology industry requirements for the production of diagnostic and therapeutic products.

SUMMARY OF THE INVENTION

In general, the present disclosure provides a process of purifying a bispecific antibody comprising a TCR constant domain from a fluid comprising said bispecific antibody and potential impurities, wherein said process comprises subjecting said fluid to a Protein A chromatography followed by an anion exchange chromatography and a mixed-mode chromatography in the specified order, thereby to obtain a harvest of purified bispecific antibody; wherein the anion exchange chromatography is conducted in bind-elute mode.

The process of the invention is particularly suitable for purifying a bispecific antibody according to the WuXiBody ^TM technology, and is highly effective to remove the process-related impurities and the product-related impurities, including host cell components, homodimers, aggregates and the byproducts lacking the TCR constant domain of design.

DESCRIPTION OF THE DRAWINGS

Figure 1: Schematic representation of the four WuXiBody-based bsAbs (referred to as "Molecules A to D" ) in different formats, each with pI given beneath. As seen, Therein, molecules A and B are in asymmetric format while C and D are in symmetric format.

Figure 2: Protein A chromatograms for the four WuXiBody-based bsAbs. Insets: non-reducing SDS-PAGE analysis of relevant fractions, including M: protein marker; L: load; E: eluate; S: strip.

Figure 3: AEX chromatograms for the four WuXiBody-based bsAbs. Insets: non-reducing SDS-PAGE analysis of relevant fractions, including M: protein marker; L: load; W: wash; E: eluate; S: strip.

Figure 4: Mixed-mode chromatograms for the four WuXiBody-based bsAbs. The particular mixed-mode resin used was Capto MMC ImpRes for molecules A-C and Capto adhere ImpRes for D. Insets: non-reducing SDS-PAGE analysis of relevant fractions, including M: protein marker; L: load; W: wash; E: eluate; S: strip.

Figure 5: SEC-HPLC chromatograms for the four WuXiBody-based bsAbs, i.e., molecules A to D in their final purified form.

DETAILED DESCRIPTION OF THE INVENTION

As used herein, “a” and “the” are understood to include both singular and plural unless otherwise specified expressively.

As use herein, “or” is understood to be inclusive and is interchangeable with “and/or” unless otherwise specified expressively.

For a feature defined by a value or a value range, embodiments of the feature include not only the values and ranges listed herein but also each of the individual integers and fractions within the specified ranges and ranges formed from any two of the values listed herein, allowing for predictable deviation from a specific value.

Each of the scientific terms, as used herein, has the standard definition and meaning as commonly understood and recognized by the skilled in the art, unless otherwise specified.

In the present disclosure, the following Abbreviations are used: bsAb, bispecific antibody; mAb, monoclonal antibody; HC, heavy chain; LC, light chain; KiH, knobs-into-holes; TCR, T cell receptor; CH1, constant region 1 of the heavy chain; CL, constant region of the light chain; pI, isoelectric point; IEX, ion exchange; AEX, anion exchange; MMC, mixed-mode chromatography; SEC-HPLC, size-exclusion chromatography-high performance liquid chromatography; CV, column volume; HCP, host cell protein, SDS-PAGE, sodium dodecyl sulfate polyacrylamide gel electrophoresis

WuXiBody ^TM is an innovative bsAb platform developed by WuXi Biologics. Its key feature is the replacement of one parental mAb’s CH1/CL constant domain (formed from the CH1 and the CL constant regions as in a canonical Fab fragment) with a TCR constant domain (e.g., one formed from the Cβ region and the Cα region of TCR) . The purpose of such sequence engineering is to promote cognate HC-LC pairing. A new and universal characteristic of WuXiBody-based bsAbs gained thereby is their greatly reduced pIs. This is because that the TCR constant domain has a relatively low pI and hence the bsAbs containing same have pIs lower than that of regular mAbs. The inventors find that this bonus feature facilitates separation of the target bsAb from non-TCR-containing byproducts by IEX chromatography, especially in the case of bsAbs of an asymmetric format. In particular, the non-TCR-containing half antibody and its corresponding homodimer can be easily removed by AEX as they possess charges different from that of the target bsAb under a selected condition.

In the present disclosure, we demonstrated that a potential platform approach, which contains Protein A chromatography, AEX chromatography and mixed-mode chromatography, is desirable for purification of WuXiBody-based bsAbs in various formats. In particular, AEX and mixed-mode chromatography are highly effective at removing non-TCR-containing byproducts and aggregates, respectively. Advantageously, this platform will greatly save time and effort on downstream process development for future WuXiBody projects.

Accordingly, the present disclosure provides a process of purifying a bispecific antibody comprising a TCR constant domain (also referred to as "TCR-containing bsAbs" herein below) from a fluid comprising said bispecific antibody and potential impurities, wherein said process comprises subjecting said fluid to a Protein A chromatography followed by an anion exchange chromatography and a mixed-mode chromatography in the specified order, thereby to obtain a harvest of purified bispecific antibody; wherein the anion exchange chromatography is conducted in bind-elute mode.

The Protein A chromatography for product capture may be conducted according to the standard protocol. The elution can be either a gradient pH elution or a stepwise pH elution. In some embodiments, the Protein A chromatography is eluted by stepwise pH elution. In some instances, the stepwise pH elution comprises a stage at a pH ranging from about 3.0 to about 4.0, such as about pH 3.6.

AEX chromatography for intermediate purification is typically run in flow-through mode. Most mAbs have a relatively high pI (6.5-9.0) . Consequently, at a suitable pH (i.e., below the target protein’s pI) and conductivity, the product passes through the column whereas negatively charged impurities (e.g., HCPs) bind to the resin. However, in the present invention, for TCR-containing bsAbs having lowed pIs, the AEX chromatography is conducted in bind-elute mode. In this mode, the AEX can not only remove the non-TCR-containing high pI byproducts, including the non-TCR-containing homodimers as in the case of asymmetric format as represented by molecule A and molecule B, but also remove those byproducts having an even lower pI, including the TCR-containing homodimers as in the case of asymmetric format as represented by molecule A and molecule B. In some embodiments, the AEX chromatography column is eluted at about pH 8.0. In some embodiments, the AEX chromatography column is eluted at an ionic strength ranging from about 130 to about 190 mM, preferably about 130 to about 165 mM. In some embodiments, the AEX chromatography column is eluted with a buffer at about pH 8.0 (e.g., a 50 mM Tris-HAc buffer, pH 8.0) comprising about 130 to about 190 mM, preferably about 130 to about 165 mM NaCl.

In some embodiments, the process of the invention may optionally include a step of intermediate depth filtration after the Protein A chromatography and before the AEX chromatography. In general, post-Protein A neutralization followed by depth filtration is a robust and effective step for HCP clearance. The target pH for neutralization is usually between about 5.0 and about 7.5. The rationale is that most CHO HCPs have pIs ranging from about 4.5 to about 7.0 and become less soluble when titrated to this pH range whereas mAbs, normally having higher pIs, are not affected. However, in some embodiments of the present invention, a relatively low pH value may be used for the TCR-containing bsAbs having lowed pIs compared normal mAbs. In some embodiments, in this step, the eluate from Protein A chromatography is titrated to a pH ranging from about pH 4 to about pH 5 and is then filtered to obtain the filtrate for the subsequent treatment. In some instances, the filtrate is then subjected to the AEX chromatography.

In some embodiments, the mixed-mode chromatography may be Capto MMC ImpRes or Capto adhere ImpRes. And, the mixed-mode chromatography may be conducted in bind-elute mode. Capto MMC ImpRes and Capto adhere ImpRes are both multimodal ion exchangers designed for optimal aggregate removal [8-10] . Their ligands contain groups that are capable of ion exchange (cation and anion exchange for MMC and adhere, respectively) , hydrophobic interaction and hydrogen bonding. We previously demonstrated that both resins are highly effective at removing antibody aggregates [11] .

In some instances, the fluid comprising bsAbs and potential impurities may be a cell culture harvest from a recombinant production of the bsAb. In one embodiment, the culture harvest is pre-treated, which may include clarification. For instances, the culture harvest may be clarified (e.g., by centrifugation or filtration) to provide the fluid or "load" for chromatography.

In the present application, the TCR-containing bsAb may be an IgG-like bsAb according to the WuXiBody ^TM technology, as described in WO 2019/057122 A1 [7] . For instance, the bsAb may comprise a TCR constant domain in one or more Fab units or arms in said bsAb. As understood and as exemplified in Fig. 1, bsAbs may comprise more than two Fab arms or extended Fab arms comprising multiple Fab units. As used herein, the term "Fab arm" and the term "Fab unit" both refer to a moiety structurally and/or functionally equivalent to the stereotype Fab fragment as commonly understood in the field of engineered antibodies.

The TCR constant domain is typically formed from the constant regions from cognate chains of TCR. In some embodiments, the TCR constant domain is formed from a Cβ region and a Cα region of TCR linked by an inter-chain disulfide linkage.

The TCR-containing bsAbs may be of either an asymmetric format or a symmetric format. Typically, an asymmetric format will adopt the KiH design to facilitate heterodimerization.

In some embodiments, the TCR-containing bsAb is asymmetric, which comprises a TCR constant domain in place of the CH1/CL constant domain in one of the Fab arms, wherein the CH1-replaced chain has a "knob" mutation and the CH1-comprising chain has a "hole" mutation. Specifically, for instance, the bsAb may have the asymmetric format of molecule A in Fig. 1, wherein a Cβ region of TCR is comprised in place of the CH1 domain in the heavy chain having the "knob" mutation and a Cα region in place of the CL domain in the cognate light chain.

The term "CH1/CL constant domain" , as used herein, refer to the domain formed by paring between the corresponding CH1 and CL regions.

In some other embodiments, the TCR-containing bsAb is asymmetric, which comprises a TCR constant domain in place of the CH1/CL constant domain in one of the Fab arms, wherein the CH1-replaced chain has a "hole" mutation and the CH1-comprising chain has a "knob" mutation. Specifically, for instance, the bsAb may have the asymmetric format of molecule B in Fig. 1, wherein a Cβ region of TCR is comprised in place of the CH1 domain in the heavy chain having the "hole" mutation and a Cα region in place of the CL domain in the cognate light chain.

In some other embodiments, the TCR-containing bsAb is symmetric, which comprises two identical extended Fab arms linked to the N-terminus of the Fc region, wherein either extended Fab arm comprises a first Fab unit and a second Fab unit, wherein the first and the second Fab units have different specificities and at least one of them comprises a TCR constant domain in place of the CH1/CL constant domain. Specifically, for instance, the bsAb may have the symmetric format of molecule C in Fig. 1, which comprises extended Fab arms wherein the first Fab unit comprises a Cβ region of TCR in place of the CH1 domain and a Cα region in place of the CL domain in the cognate light chain, and wherein the first Fab unit is fused at the C-terminus of the Cβ region to the N-terminus of the second Fab unit in the heavy chain.

In some other embodiments, the TCR-containing bsAb is symmetric, which comprises a first pair of Fab arms (i.e., two identical first Fab arms) linked to the N-terminus of the Fc region and a second pair of Fab arms (i.e., two identical second Fab arms) linked to the C-terminus of the Fc region, wherein the first and the second Fab arms have different specificities and at least one of them comprises a TCR constant domain in place of the CH1/CL constant domain. Specifically, for instance, the bsAb may have the symmetric format of molecule D in Fig. 1, wherein the first Fab arm comprises a Cβ region of TCR in place of the CH1 domain and a Cα region in place of the CL domain in the cognate light chain, and wherein the first Fab arm is fused at the C-terminus of the Cβ region to the N-terminus of Fc region in the heavy chain.

It should be understood that by "in place of" and "replaced" in the above description of the TCR constant domain with reference to the CH1 and CL regions, it does not mean that the bsABs are produced by any particularly process of construction including deletion, insertion or substitution of said regions, but merely refers to the location with reference to the stereotype IgG antibody and stereotype Fab fragment.

In some other embodiments, the TCR-containing bsAb has a pI ranging from about 5.5 to about 6.5, more specifically about 6.0 to about 6.5 and even more specifically about 6.1 to about 6.5.

As can be understood, impurities in bsAb production may include process-related impurities (e.g., HCPs and host DNA) and product-related impurities (e.g., homodimers, low-molecular-species such as half antibodies, 3/4 antibodies, HC dimer, free HC and free LC, and high-molecular-species such as aggregates) . For TCR-containing bsAbs, like those according to WuXiBody ^TM, quite a significant part will be impurities lacking the TCR constant domain of design, which is referred to as "non-TCR-containing byproducts" in this disclosure, like non-TCR-containing half antibodies and homodimers. At the same time, for symmetric bsAbs, the major product-related impurities are aggregates. As demonstrated in the Example below, in a process according to the present invention, Protein A chromatography effectively removes process-related impurities (e.g., HCPs and DNA) , then AEX effectively removes non-TCR-containing byproducts and mixed-mode chromatography effective removes aggregates.

Thereby, the process of the invention can give a harvest of purified bispecific antibody substantially free of impurities. By "substantially free of impurities" , in context of the present disclosure, it refers to a purity of at least 95%, at least 96%, at least 97%, at least 98%, at least 99%or even higher, including a purity of 100%.

EXAMPLE

The inventions will be described in further details by referring to the following specific examples. It should be understood that these examples are provided for illustration only and by no means to limit the scope of the invention. Experiments were conducted as specified or otherwise according to the conventional practices, such as the teachings in technique manuals, e.g., Molecular Cloning: A Laboratory Manual, or according to manufacturers' instructions.

Materials and methods

Materials

Sodium dihydrogen phosphate monohydrate, disodium hydrogen phosphate dihydrate, ammonium sulfate, ethanol, sodium acetate trihydrate, sodium chloride, sodium hydroxide, glycine and Tris (hydroxymethyl) aminomethane were purchased from Merck (Darmstadt, Germany) . Acetic acid, L-arginine, L-arginine hydrochloric acid, histidine and histidine monohydrochloride were purchased from J. T. Baker (Phillipsburg, NJ, USA) . MabSelect SuRe LX, Capto MMC ImpRes, Capto adhere ImpRes and HisCale 26/40 column (inner diameter: 26 mm, length: 40 cm) and XK16/40 column (inner diameter: 16 mm, length: 40 cm) were purchased from GE Healthcare (Uppsala, Sweden) . Poros 50HQ and MOPS SDS Running Buffer (20X) were purchased from Thermo Fisher Scientific (Waltham, MA, USA) . A1HC depth filter (MA1HC23CL3, 23 cm ²) and VL11/25 column (inner diameter: 11 mm, length: 25 cm) were purchased from Millipore (Billerica, MA, USA) . TSKgel G3000SWxl stainless steel column (7.8 x 300 mm) was purchased from Tosoh (Tokyo, Japan) . 30%acrylamide/bis-acrylamide solution (37.5: 1) , TEMED and Precision Plus Protein Unstained Standards were purchased from Bio-Rad Laboratories (Hercules, CA, USA) . Ammonium persulfate, Coomassie blue R-250, glycerol, sodium dodecyl sulfate, iodoacetamide, Bis-Tris and Bromophenol blue were purchased from Sigma-Aldrich (St. Louis, MO, USA) . The four WuXiBody-based bsAbs, referred to as "molecules A to D" respectively herein below, were expressed in CHO-K1 cells grown in HyClone ActiPro culture medium supplemented with Cell Boost 7a and 7b (the medium and feeding supplements are from GE Healthcare) . The culture harvest was clarified by centrifugation to give a clarified harvest.

Equipment An AKTA pure 150 system installed with Unicorn software version 7.3 (GE Healthcare, Uppsala, Sweden) was used for column chromatography. pH and conductivity was measured using SevenExcellence S470 pH/Conductivity meter (Mettler-Toledo, Columbus, OH, USA) . Protein concentration was measured using a NanoDrop 2000 spectrophotometer (Thermo Fisher Scientific, Waltham, MA, USA) . An Agilent 1260 liquid chromatography instrument (Agilent Technologies, Santa Clara, CA, USA) was used for size-exclusion chromatography-high performance liquid chromatography (SEC-HPLC) . The bioreactor system from Applikon Biotechnology (Delft, Netherlands) was used for cell cultivation. The Sorvall LYNX 6000 superspeed centrifuge from Thermo Fisher Scientific was used for clarification of cell culture harvest.

Methods

Protein A chromatography

MabSelect SuRe LX was packed in a 2.6 cm diameter column with 20.2 cm bed height. The column volume (CV) is approximately 107.2 ml. Column load is the culture harvest clarified by centrifugation. For all runs, the column was loaded at 30 mg of protein per ml of resin. The system was run at a flow rate of 242 cm/h (residence time: 5 min) . All runs were conducted in bind-elute mode and the bound protein was eluted by stepwise pH elution. Detailed information for Protein A procedure is summarized in Table 1. All chromatograms were recorded by monitoring UV absorbance at 280 nm. Eluate from each run was subjected to host cell protein (HCP) , purity and concentration measurement.

Table 1. Details of Protein A chromatography.

^aNot applicable.

Intermediate depth filtration

Millipore A1HC (MA1HC23CL3, 23 cm ²) was used for post-Protein A intermediate depth filtration. Protein A eluate was first neutralized to pH 5.0 (for molecules A, C and D) or pH 4.8 (for molecule B) and was then loaded to the filter at a density of 1000 g/m ². The filtrate from each run was analyzed for HCP level, monomer purity and measured for product recovery.

AEX chromatography

AEX chromatography was conducted with POROS 50 HQ resin in bind-elute mode using a 1.6 cm diameter column with 24.0 cm bed height. The CV is approximately 48.2 ml. Details for each step is listed in Table 2. For all runs, the column was loaded at 40 mg of protein per mL of resin. For all chromatographic runs, the system was set at a flow rate of 288 cm/h (residence time: 5 min) . All chromatograms were recorded by monitoring UV absorbance at 280 nm. Eluate from each run was analyzed for HCP level and monomer purity. Step recovery was calculated based on concentration measurement.

Table 2. Details of AEX chromatography.

^aFor molecule A the column was equilibrated at pH 7.0, and for molecules B-D the column was equilibrated at pH 8.0.

^bNot applicable.

^cFor molecule A the filtrate was pH adjusted to 7.0, and for molecules B-D the filtrate was pH adjusted to 8.0.

^dFor molecule A only.

^eFor molecules B-D, respectively.

^fFor molecules A-D, respectively.

Capto MMC/adhere ImpRes chromatography

Capto MMC ImpRes and Capto adhere ImpRes chromatography were performed in bind-elute mode using a 1.1 cm diameter column with 20.0 cm bed height (CV: ～19.0 mL) and a 1.1 cm diameter column with 18.0 cm bed height (CV: ～17.1 ml) , respectively. The column was loaded at 40 mg and 30 mg protein per ml of resin for MMC and adhere, respectively. Related information for each step is listed in Table 3. The system was run at a flow rate of 240 cm/h and 216 cm/h for Capto MMC ImpRes and Capto adhere ImpRes, respectively (residence time: 5 min) . All chromatograms were recorded by monitoring UV absorbance at 280 nm. Eluate from each run was analyzed for HCP level and monomer purity. Step recovery was calculated based on concentration measurement.

Table 3. Details of MMC ImpRes/adhere ImpRes ^a chromatography.

^aFor molecules A-C MMC ImpRes was used as the second polishing, whereas for molecule D adhere ImpRes was used.

^bFor MMC ImpRes and adhere ImpRes, respectively.

^cNot applicable.

^dFor MMC ImpRes and adhere ImpRes, respectively.

^eFor molecules A-D, respectively.

^fFor molecules A and B only (wash 3 was not applied to molecules C and D) .

^gFor molecules A and B, respectively.

^hFor molecules A-D, respectively.

ⁱFor molecules A-D, respectively.

Non-reducing sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE)

SDS-PAGE was performed using 8%non-gradient Bis-Tris gels, which were cast in-house following standard protocol. 3.5X gel buffer (1.25 M bis-Tris, pH 6.5-6.8) was also prepared in-house. This buffer was used for casting both separating gel (8%) and stacking gel (4%) . Electrophoresis was carried out at a constant voltage of 80 V for 2 h. Gels were stained using Coomassie blue and de-stained with destaining solution containing 10%acetic acid, 20%ethanol and 70%water.

SEC-HPLC

SEC-HPLC analysis was performed on an Agilent 1260 liquid chromatography instrument using a Tosoh TSKgel G3000SWxl stainless steel column (7.8 x 300 mm) . 100 μg of sample was injected per run. The mobile phase consisted of 50 mM sodium phosphate, 300 mM sodium chloride at pH 6.8. Each sample was eluted isocratically for 20 min at a flow rate of 1.0 ml/min. Protein elution was monitored by UV absorbance at 280 nm.

HCP measurement

HCP level of samples at different stages was measured using the 3rd generation generic ELISA kit from Cygnus Technologies (Southport, NC, USA) following manufacturer's instructions. The detection range is 3-100 ng/ml. Serial dilutions of samples were made to keep the measurement within the calibration range. Absorbance was measured at 450 nm (absorbance) and 650 nm (reference) using Infinite 200 PRO plate reader (Tecan,

Switzerland) .

Process

Protein A chromatography for product capture

The clarified harvest fluid was loaded onto a MabSelect SuRe LX column. The Protein A chromatography for product capture was conducted as described above and the column was eluted by stepwise pH elution (Fig. 2) .

In general, Protein A chromatography effectively removes process-related impurities (e.g., HCPs and DNA) but has a limited capacity at removing product-related impurities (e.g., half antibody and aggregates) . Relevant quality data for Protein A eluate and Protein A step yield are listed in Table 4.

Table 4. Protein A eluate quality data and Protein A step yield.

Molecule	HCP (ppm)	SEC-HPLC (%)	Yield (%)
A	946	85.7/7.6/6.7 ^a	100.0
B	1945	86.3/11.1/2.6	100.0
C	3655	89.6/9.4/0.9	94.2
D	3145	90.4/9.6/ND ^b	100.0

^aPercentage of monomer, high-molecular-weight species and low-molecular-species, respectively.

^bNot detected.

As seen, for all four bsAbs, HCPs were reduced to a level ranging from approximately 1000 ppm to 4000 ppm, which was at the average to low-end of the level normally detected at this stage. SEC-HPLC based purity of Protein A eluate was ranging from approximately 86%to 90%. According to the SDS-PAGE results (Fig. 2, insets) and the SEC chromatogram (data not shown) , the post-Protein A byproducts for asymmetric (A and B) and symmetric (C and D) molecules are low-molecular-weight species and aggregates, respectively. These distinct and characteristic impurity profiles of these two groups of bsAbs are as expected and consistent with previous studies [4-6] .

Neutralization and intermediate depth filtration

The Protein A eluate was neutralized and filtered as described above. We chose a relatively low pH value (5.0 for molecules A, C and D, and 4.8 for molecule B) for neutralization as a WuXiBody-based bsAb has a significantly reduced pI in comparison to that of a normal mAb. Molecule B has the lowest pI among the four bsAbs studied and relatively large amount of precipitate was observed when the corresponding Protein A eluate was adjusted to pH 5.0. Consequently, Protein A eluate was titrated to pH 4.8 for this molecule. Even at this pH, the step yield for molecule B is lower than that of the other three molecules. The data shown in Table 5 suggest that on average HCP level for these bsAbs was moderately reduced post this step.

Table 5. Intermediate depth filtration filtrate quality data and filtration recovery.

Molecule	HCP (ppm)	SEC-HPLC (%)	Yield (%)
A	272	86.2/6.8/6.9 ^a	98.0
B	1458	86.6/10.7/2.5	88.9
C	2121	91.2/7.8/0.9	94.9
D	1537	90.4/9.6/0.1	97.6

AEX chromatography

The filtrates from intermediate depth filtration were then subjected to the AEX chromatography as described above. The AEX chromatography was conducted in bind-elute mode for each of the four bsAbs which have relatively low pIs (6.1-6.5) (Fig. 3) .Quality data for AEX eluate are summarized in Table 6.

Table 6. AEX eluate quality data and AEX step yield.

Molecule	HCP (ppm)	SEC-HPLC (%)	Yield (%)
A	116	93.8/5.5/0.7 ^a	89.2
B	306	92.9/5.5/1.6	74.7
C	147	93.0/6.0/1.0	88.1
D	126	90.4/9.5/0.1	88.7

In all cases, this step further reduced HCPs to a lower level. In addition, for bsAbs in asymmetric format (i.e., Molecules A and B) , their SEC purity was greatly improved by this step as those non-TCR-containing byproducts with high pIs shall not bind under the selected conditions. It is worth noting that those removed non-TCR-containing byproducts include potential homodimer. In the case of asymmetric format (i.e., Molecules A and B) , this step also effectively separated the target bsAbs from the undesired TCR-containing homodimer which has an even lower pI. Whereas the homodimer cannot be monitored by SEC-HPLC as it has a size similar to that of the target, we previously demonstrated that its presence and removal can be disclosed by analytical hydrophobic interaction chromatography [12-14] . For bsAbs in symmetric format (i.e., Molecules C and Format D) , there is no or very limited improvement on SEC purity by this step. It is not totally unexpected. For symmetric bsAbs, the major product-related impurities are aggregates and previous studies suggested that IEX chromatography has a limited capacity on removing aggregates [11] .

Capto MMC/adhere ImpRes chromatography

For molecules A to C, MMC ImpRes was used as the final polishing step whereas for molecule D adhere ImpRes was used. The MMC chromatographic runs were conducted as described above. The corresponding chromatograms are shown in Fig. 4, and relevant quality data for the final purified products are summarized in Table 7. In all cases, this final polishing step further reduced HCP to a level below the generally accepted criteria (i.e., 100 ppm) . In addition, SEC purity was improved to 99.6%, 97.1%, 96.5%and 95.3%for molecules A to D, respectively (Fig. 5) .

Table 7. MMC/adhere ^a eluate quality data and MMC/adhere step yield.

Molecule	HCP (ppm)	SEC-HPLC (%)	Yield (%)
A	6	99.6/0.4/ND ^b, c	70.5
B	29	97.1/2.9/ND	88.5
C	19	96.5/2.8/0.8	93.7
D	40	95.3/4.7/ND	68.2

^aFor molecules A-C MMC, ImpRes was used as the second polishing, whereas for molecule D adhere ImpRes was used.

^bPercentage of monomer, high-molecular-weight species and low-molecular-species, respectively.

^cNot detected.

In view the foregoing, it will be obvious that certain changes and modifications as equivalents can be practiced within the scope of the appended claims.

REFERENCES

[1] C. Klein, C. Sustmann, M. Thomas, K. Stubenrauch, R. Croasdale, J. Schanzer, U. Brinkmann, H. Kettenberger, J.T. Regula, W. Schaefer, Progress in overcoming the chain association issue in bispecific heterodimeric IgG antibodies, MAbs 4 (2012) 653-663.

[2] S. Krah, C. Sellmann, L. Rhiel, C.

S. Dickgiesser, J. Beck, S. Zielonka, L. Toleikis, B. Hock, H. Kolmar, S. Becker, Engineering bispecific antibodies with defined chain pairing, N. Biotechnol. 39 (2017) 167-173.

[3] J.B. Ridgway, L.G. Presta, P. Carter, 'Knobs-into-holes' engineering of antibody CH3 domains for heavy chain heterodimerization. Protein Eng. 9 (1996) 617-621.

[4] W. Schaefer, J.T. Regula, M.

J. Schanzer, R. Croasdale, H. Dürr, C. Gassner, G. Georges, H. Kettenberger, S. Imhof-Jung, M. Schwaiger, K.G. Stubenrauch, C. Sustmann, M. Thomas, W. Scheuer, C. Klein, Immunoglobulin domain crossover as a generic approach for the production of bispecific IgG antibodies. Proc. Natl. Acad. Sci. USA 108 (2011) 11187-11192.

[5] J.S. Michaelson, S.J. Demarest, B. Miller, A. Amatucci, W.B. Snyder, X. Wu, F. Huang, S. Phan, S. Gao, A. Doern, G.K. Farrington, A. Lugovskoy, I. Joseph, V. Bailly, X. Wang, E. Garber, J. Browning, S.M. Glaser, Anti-tumor activity of stability-engineered IgG-like bispecific antibodies targeting TRAIL-R2 and LTbetaR, MAbs 1 (2009) 128-141.

[6] J. Schanzer, A. Jekle, J. Nezu, A. Lochner, R. Croasdale, M. Dioszegi, J. Zhang, E. Hoffmann, W. Dormeyer, J. Stracke, W.

C. Ji, G. Heilek, N. Cammack, M. Brandt, P. Umana, U. Brinkmann, Development of tetravalent, bispecific CCR5 antibodies with antiviral activity against CCR5 monoclonal antibody-resistant HIV-1 strains, Antimicrob. Agents. Chemother. 55 (2011) 2369-2378.

[7] WO 2019/057122 A1.

[8] GE Healthcare Life Sciences, Polishing of monoclonal antibodies using Capto ^TM MMC ImpRes in bind and elute mode, Application note, 29-0373-49, AA, 2013.

[9] GE Healthcare Life Sciences, Capto ^TM MMC ImpRes, Data file, 29-0356-74, AB, 2015.

[10] GE Healthcare Life Sciences, Selective removal of aggregates with Capto adhere, Application Note, 28-9078-93, AA, 2007.

[11] X. Zhang, T. Chen, Y. Li, A parallel demonstration of different resins' antibody aggregate removing capability by a case study, Protein Expr. Purif. 153 (2019) 59-69.

[12] T. Chen, J. Han, G. Guo, Q. Wang, Y. Wang, Y. Li, Monitoring removal of hole-hole homodimer by analytical hydrophobic interaction chromatography in purifying a bispecific antibody, Protein Expr. Purif. 164 (2019) 105457.

[13] J. Tang, X. Zhang, T. Chen, Y. Wang, Y. Li, Removal of half antibody, hole-hole homodimer and aggregates during bispecific antibody purification using MMC ImpRes mixed-mode chromatography, Protein Expr. Purif. 167 (2020) 105529.

[14] T. Qin, Y. Wang, Y. Li, Separating antibody species containing one and two kappa light chain constant region by KappaSelect affinity chromatography, Protein Expr. Purif. 171 (2020) 105618.

Claims

A process of purifying a bispecific antibody comprising a TCR constant domain from a fluid comprising said bispecific antibody and potential impurities, wherein said process comprises subjecting said fluid to a Protein A chromatography followed by an anion exchange chromatography and a mixed-mode chromatography in the specified order, thereby to obtain a harvest of purified bispecific antibody; wherein the anion exchange chromatography is conducted in bind-elute mode.
The process of claim 1, wherein the Protein A chromatography is eluted by gradient pH elution or stepwise pH elution.
The process of claim 1, wherein the Protein A chromatography is eluted by stepwise pH elution.
The process of claim 3, wherein the stepwise pH elution comprises a stage at a pH ranging from about 3.0 to about 4.0
The process of claim 4, wherein the stepwise pH elution comprises a stage at about pH 3.6.
The process of claim 1, wherein the anion exchange chromatography column is eluted at about pH 8.0.
The process of claim 1, wherein the anion exchange chromatography column is eluted at an ionic strength ranging from about 130 to about 190 mM
The process of claim 7, wherein the anion exchange chromatography column is eluted at an ionic strength ranging from about 130 to about 165 mM.
The process of claim 1, further comprising a step of intermediate depth filtration after the Protein A chromatography and before the anion exchange chromatography, wherein the eluate from Protein A chromatography is titrated to a pH ranging from about pH 4 to about pH 5 and is then filtered to obtain the filtrate.
The process of claim 1, wherein the mixed-mode chromatography is Capto MMC ImpRes or Capto adhere ImpRes and is conducted in bind-elute mode.
The process of claim 1, wherein the bispecific antibody has a pI ranging from about 5.5 to about 6.5.
The process of claim 11, wherein the bispecific antibody has a pI ranging from about 6.0 to about 6.5.
The process of claim 1, wherein the bispecific antibody is a bispecific antibody according to the WuXiBody ^TM technology, which comprises a TCR constant domain in one or more Fab units in the bispecific antibody.
The process of claim 1, wherein the bispecific antibody is asymmetric.
The process of claim 14, wherein the bispecific antibody is of the KiH format.
The process of claim 1, wherein the bispecific antibody is symmetric.
The process of claim 1, wherein said impurities comprise one or more selected from the group consisting of host cell proteins, DNAs, byproduct species lacking TCR constant domain, homodimers comprising TCR constant domains and aggregates.
The process of claim 1, wherein said harvest of purified bispecific antibody is substantially free of impurities.