Functional integration of protein A binding ability to antibody fragments for convenient and tag-free purification

ABSTRACT

Although the development of small therapeutic antibodies is important, the affinity tags used for their purification often result in heterogeneous production and immunogenicity. In this study, we integrated Staphylococcus aureus protein A (SpA) binding ability into antibody fragments for convenient and tag-free purification. SpA affinity chromatography is used as a global standard purification method for conventional antibodies owing to its high binding affinity to the Fc region. SpA also has a binding affinity for some variable heavy domains (VH) classified in the VH3 subfamily. Through mutagenesis based on alignment and structural modeling results using the SpA-VH3 cocrystal structure, we integrated the SpA-binding ability into the anti-CD3 single-chain Fv. Furthermore, we applied this mutagenesis approach to more complicated small bispecific antibodies and successfully purified the antibodies using SpA affinity chromatography. The antibodies retained their biological function after purification. Integration of SpA-binding ability into conventional antibody fragments simplifies the purification and monitoring of the production processes and, thus, is an ideal strategy for accelerating the development of small therapeutic antibodies. Furthermore, because of its immunoactivity, the anti-CD3 variable region with SpA-binding ability is an effective building block for developing engineered cancer therapeutic antibodies without the Fc region.

Keywords:

• single-chain variable fragment
• tag-free purification
• small therapeutic antibody
• small bispecific antibody
• cancer immune therapy
• Protein A purification

Introduction

Monoclonal antibodies are widely used as therapeutic agents for various diseases that are difficult to cure using conventional drugs [1–3] Although monoclonal antibodies exhibit considerable therapeutic benefits owing to their high specificity and binding affinity to target antigens, the high production costs associated with antibodies produced in mammalian expression systems have restricted their further applications. Small antibodies, mainly composed of antibody variable domains without the Fc region, may be useful for overcoming the limitations of full-length antibodies [4,5], as they can be produced using cost-effective microbial expression systems [6]. Furthermore, small antibody fragments exhibit better penetration ability into cancer tissues than conventional antibodies do and are easy to engineer [7]. Notably, the construction of small bispecific antibodies (bsAbs) that can simultaneously bind two target molecules is an attractive strategy because of the multiple functions of these molecules [8–12]. Various small bsAbs have been developed; for example, the bispecific T-cell engager blinatumomab (Blincyto®), which was approved by the United States Food and Drug Administration in 2014, has been successfully used to treat acute lymphoblastic leukemia [13]. Moreover, three small bsAbs (ozoralizumab, flotetuzumab, and AFM13) are currently being reviewed in pivotal phase II studies [1].

In most cases, small antibodies are purified by immobilized metal affinity chromatography (IMAC) [14] based on the affinity of the histidine tag (His tag) for the metal ion coupled to the resin. However, in the therapeutic application of small antibodies, the presence of a tag may induce immunogenicity and heterogeneous preparation of the protein of interest and require additional operations to remove the leaked metal ions derived from the chromatography resin [15,16]. Although other tag-assisted affinity chromatography systems, such as those using a hemagglutinin tag, FLAG tag, and Myc tag, can be used to purify small antibodies, these systems are costly because IgG antibodies are required as purification ligands [17–19]. Various affinity tags that do not require IgG ligands have also been reported; however, their immunogenicity and difficulty in removing tags before their pharmaceutical application remain a concern.

Staphylococcus aureus protein A (SpA) is a 42 kDa protein consisting of five immunoglobulin-binding domains (domains E, D, A, B, and C), with each domain built from three helices [20–22]. Helix I and II of SpA can bind to the CH2 and CH3 domains of the Fc region with high affinity [23], explaining why SpA is universally used as a ligand for IgG antibody purification. SpA has also been used to monitor the production rate during the cultivation phase and the purification efficiency of each chromatographic step during the mass production of IgG [24,25]. Furthermore, SpA can bind to some variable heavy domains (VHs) classified in the VH3 subfamily [26]. The cocrystal structure of domain D of SpA and Fab with the VH3 subfamily has revealed important amino acid residues in VH3 for SpA-binding. Using this information, some research groups have successfully converted the non-SpA-binding variable domain of the heavy chain of heavy-chain antibodies (VHH) to SpA-binding VHH through amino acid mutagenesis [27,28]. As therapeutic VHHs have gained attention over the past few years [5], these studies have contributed to advancements in the therapeutic VHH field, particularly in the purification and purification tracing of VHH. However, conventional antibody-based small antibodies such as Fv and single-chain Fv (scFv) are readily available and can be developed further. These antibody fragments can be used as components of more complicated small antibodies with higher-order structures, such as bispecific diabodies (Dbs) [8] and bispecific tandem scFvs (taFvs) [10]. However, except for VHH, the integration of SpA-binding ability into antibody fragments has not been reported.

On the basis of the hypothesis that SpA-binding ability can be integrated into small antibodies by mutagenesis through sequence alignment and structural modeling, we aimed to develop a versatile method to integrate SpA-binding ability. First, we engineered anti-CD3 scFv (Figure 1a), derived from Orthoclone OKT3 (muromonab-CD3). This clone is the most widely utilized anti-CD3 antibody clone, specifically for developing bsAbs because of its immunoactivity [29–32]. Second, we applied the resultant OKT3 mutant with SpA-binding ability to develop two types of bsAbs: single-chain Db (scDb) and taFv (Figure 1a). Both bsAbs were purified using SpA affinity chromatography without affinity tags and retained their biological function. Integration of SpA-binding ability with conventional antibody fragments is an ideal strategy for accelerating the development of small therapeutic antibodies. Additionally, the OKT3 mutant with its SpA-binding ability is a powerful and versatile tool for integrating the ability of easy purification (through SpA) into small bsAbs.

Figure 1. (a) schematic depiction of small antibodies we used in this study. (b, c) structural prediction and analysis of interaction interface at position 65. (b) mutagenesis of aspartic acid at position 65 on the VH domain of 2A2; collision sites are depicted with a yellow dotted line. (c) glycine in the VH domain of 2A2 classified as VH3 in the cocrystal structure.

Materials and methods

Preparation of OKT3-scFv mutants

We used a humanized version of the anti-CD3 scFv (OKT3-scFv, Supplementary Figure S1), to construct several engineered antibodies [33,34]. Each OKT3-scFv expression vector was constructed based on a humanized sequence and T7 promoter-based pRA vector, and each mutation was introduced by the standard protocol of QuikChange site-directed mutagenesis using each primer set (Supplementary Table S1) [33]. OKT3-scFv and its mutants were prepared using an E. coli expression system as previously reported [35]. Briefly, the E. coli BL21 Star (DE3) strain was transformed with each pRA-OKT3-scFv vector individually, and the culture supernatant was recovered. The resulting OKT3-scFv concentrated using ammonium sulfate from the culture supernatant was purified by IMAC. Here, each OKT3-scFv sample was loaded onto poly-prep® chromatography columns (Bio-Rad, #7311550) packed with Ni Sepharose 6 Fast Flow (Cytiva, #17531802), and wash step was conducted using 0.5 M NaCl/20 mM PPB (pH7.4) buffer. The elution steps were subsequently conducted by increasing the imidazole concentration in buffer to 20, 50, 150, 300, 1000 mM. The fraction eluted with buffer containing 300 mM imidazole was collected and applied for size-exclusion chromatography using a Superdex^TM 200 Increase 10/300 GL column (Cytiva, #28990944) to fractionate the monomer of each OKT3-scFv. The purity of the product was confirmed using sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) under reducing conditions.

SpA-binding ability using SpA-immobilized column

Two hundred μL of each purified sample concentrated to 1.0 nmol/mL was loaded onto poly-prep® chromatography columns packed with 100 µL of rProtein A Sepharose Fast Flow (Cytiva, #17127903) and the flow-through fraction was collected. In particular, the column was first washed with two column volumes (CVs) of phosphate-buffered saline (PBS), followed by three elution steps conducted with 2CVs of 0.1 M glycine-HCl (pH 3.0) buffer. Each fraction was analyzed by SDS-PAGE under reducing conditions. The protein on gel after SDS-PAGE was transferred onto the nitrocellulose membrane. The membrane was treated with 5% skimmed milk/PBS-T buffer for 50 min, chemiluminescence was detected using a horseradish peroxidase-conjugated anti-His tag antibody (QIAGEN, #34460) as a detection antibody.

Site-specific mutagenesis of hEx3

As described in our previous report, we designated the VH and VL regions of the humanized anti-epidermal growth factor receptor (EGFR) antibody 528 as 5H and 5L and the VH and VL regions of the humanized anti-CD3 antibody OKT3 as OH and OL [29]. To construct hEx3-scDb-LH (Supplementary Figure S2) or hEx3-ta6 (Supplementary Figure S3) with eight mutations (hEx3-scDb-LH-8m or hEx3-ta6-8m, respectively), we introduced the mutations (D65G and D82aN on OH and K19R, N65G, T70S, E81Q, S82aN, and R82bS on 5H) by overlap extension polymerase chain reaction or QuikChange site-directed mutagenesis using pRA-hEx3-scDb-LH or pRA-hEx3-ta6 as template vectors. In the overlap extension polymerase chain reaction, the gene of hEx3-scDb-LH-8m was constructed by using two DNA fragments, which were amplified by each primer set (T7 promoter Fw with hEx3 E81Q/S82aN/R82bS Rev and hEx3 E81Q/S82aN/R82bS Fw with T7 terminator Rev). To construct the gene of hEx3-ta6-8m, two DNA fragments amplified by two different primer sets (T7 promoter Fw with ta6 5H Rev and ta6 5H Fw with T7 terminator Rev) and one DNA fragment amplified by the primer set (scDb-8m 5H Fw with scDb-8m 5H Rev) were used for an overlap extension polymerase chain reaction. Each amplified gene was inserted into the pROXb3 expression vector to construct pROXb3-hEx3-scDb-LH-8m or -hEx3-ta6-8m using restriction enzymes for expression in B. choshinensis, as described previously [34]. The gene of tag-free hEx3-scDb-LH-8m (hEx3-scDb-LH-8m(-)) was amplified by polymerase chain reaction using a primer set (scDb tag remove Fw with scDb tag remove Rev) to introduce a stop codon at the C-terminus of hEx3-scDb-LH-8m. The amplified hEx3-scDb-LH-8m(-) gene was inserted into the pROXb3 expression vector to construct pROXb3-hEx3-scDb-LH-8m(-).

Preparation of each hEx3 from B. choshinensis

hEx3 was prepared using a B. choshinensis expression system, as previously reported [34]. Briefly, the B. choshinensis HPD31 strain was transformed with each vector individually, and culture supernatant was recovered. The resulting hEx3 concentrated using ammonium sulfate from the culture supernatant was purified by IMAC. Here, each hEx3 sample was loaded onto poly-prep® chromatography columns packed with Ni Sepharose 6 Fast Flow, and a washing step was conducted using a PBS buffer with 50CVs. The elution steps were subsequently conducted by increasing the imidazole concentration in buffer to 10, 50, 150, 200, 300, 500 mM. The fractions eluted with buffer containing 150 mM or 200 mM imidazole was collected and applied for size-exclusion chromatography using a Superdex^TM 200 Increase 10/300 GL column to fractionate the monomer of each hEx3. Rather than IMAC, tag-free hEx3-scDb-LH-8m(-) was purified by SpA affinity chromatography. hEx3-scDb-LH-8m(-) sample concentrated using ammonium sulfate from the culture supernatant was loaded onto poly-prep® chromatography columns packed with rProtein A Sepharose Fast Flow. The column was first washed with 20CVs of PBS buffer and 5CVs of 0.1 M Glycine-HCl buffer (pH4.0), followed by five elution steps conducted with 3CVs of 0.1 M Glycine-HCl buffer (pH3.0). The fractions containing hEx3-scDb-LH-8m(-) was collected and applied for size-exclusion chromatography using a Superdex^TM 200 Increase 10/300 GL column to fractionate the monomer. The purity of the product was confirmed using SDS-PAGE under reducing conditions.

SpA-binding ability on plate assay

0.5 mM of recombinant Bipo Resin Protein A, kindly provided by ProteinExpress Co., Ltd., was immobilized on black 96-well Immune Maxisorp plates (Thermo Fisher Scientific, #437111) in carbonate-bicarbonate buffer (Sigma-Aldrich, #C3041) and the plate was washed three times. After blocking it with PBS containing 1% (w/v) bovine serum albumin, the plate was washed three times. hEx3-scDb-LH or hEx3-scDb-LH-8m labeled with fluorescein using the Fluorescein Labeling Kit – NH2 (Dojindo Laboratories, #LK01) was added to the plates and incubated for 1 h, followed by three washing steps. The fluorescence intensity of the plates was measured.

Cell-binding ability evaluated by flow cytometry

Approximately 5 × 10⁵ target cells were incubated for 30 min on ice with 20 pmol of each hEx3. After two washes with PBS, 100 µL of rabbit anti-hEx3 serum, kindly provided by Immuno-Biological Laboratories Co., Ltd., was added to the cells and incubated for 30 min on ice. After two washes with PBS, the cells were exposed to 1 µg of Alexa Fluor® 594 conjugated goat anti-rabbit IgG (Abcam, #ab150080) on ice for 30 min. The stained cells were analyzed by flow cytometry using a BD Accuri C6 flow cytometer (BD Biosciences).

In vitro growth inhibition assay

The inhibition of cancer cell growth was evaluated as previously reported [36]. Briefly, a human bile duct carcinoma (TFK-1)-expressing EGFR cell line [37,38] was used as target, and lymphokine-activated killer cells with the T-cell phenotype (T-LAK) were used as effector cells. T-LAK cells were induced as previously described [36]. Briefly, peripheral blood mononuclear cells were cultured for 48 h at a density of 1 × 10⁶ cells/mL in medium supplemented with 100 IU/mL recombinant human interleukin-2 (Shionogi Pharmaceutical Co., Ltd., #6399411D1022) in a culture flask that was pre-coated with 10 µg/mL of anti-CD3 monoclonal antibody. The in vitro inhibition of cancer cell growth was evaluated using an MTS assay kit (CellTiter 96 AQueous Non-Radioactive Cell Proliferation; Promega, #G5421), as previously reported [33].

Results

We successfully integrated the SpA-binding ability into anti-CD3 OKT3-scFv by mutagenesis based on sequence alignment and structural modeling. This mutagenesis approach has been applied to small bispecific antibody formats, such as scDb and taFv, without any negative effects on antibody biological functions.

Design of anti-CD3 scFv with binding ability to SpA

To integrate SpA-binding ability into anti-CD3 scFv (OKT3-scFv), we designed mutants based on the structure of the VH-SpA complex. CD3 forms protein complexes with T-cell receptors on T-cells and, therefore, is an attractive target for cancer immunotherapy. The VH domain of OKT3 (OH) cannot bind to SpA, even though OH belongs to the VH3 subfamily (unpublished data). These data indicate that not all VHs in the VH3 subfamily have SpA-binding ability. The following key residues in VH3 for SpA binding were identified through crystal structure analysis: G15, S17, R19, K57, Y59, K64, G65, R66, T68, S70, Q81, N82a, and S82b (residues were numbered according to the Kabat numbering method) [27,39]. Among the key residues, sequence alignment showed that OH had three unfavorable residues for SpA binding: T57, D65, and D82a (Table 1). Position 57 could tolerate lysine (K), threonine (T), or arginine (R) for SpA-binding, whereas asparagine (N) at position 82a was important for the interaction with S33 in SpA through a hydrogen bond [26]. Since there is no information about the contribution level of position 65 to interact with SpA, we used a molecular visualization tool, PyMol (https://pymol.org/2/), which combines data from mutagenesis studies. The predicted structure showed a collision with the side chain of N43 on SpA when position 65 on the VH domain of 2A2 was mutated to aspartic acid (D). Glycine (G) at position 65 on the VH domain of the original 2A2 may avoid this collision (Figure 1b, c). Thus, we selected the D65G and D82aN mutations to confer OKT3-scFv with the SpA-binding ability.

Table 1. Sequence alignment of SpA-binding positions of VH used in this study with SpA binding VH classified in VH3 subfamily (residues were numbered based on the Kabat numbering method).

RESIDUE	15	17	19	57	59	64	65	66	68	70	81	82A	82B
2A2	G	S	R	K	Y	K	G	R	T	S	Q	N	S
OH*	G	S	R	T	Y	K	D	R	T	S	Q	D	S
5H**	G	S	K	T	Y	K	N	R	T	T	E	S	R
REGION	FR1			CDR2				FR3

*OH is the VH domain of OKT3.

**5H is the VH domain of 528.

Preparation and SpA-binding evaluation of OKT3-scFv mutants

To evaluate the effects of mutagenesis of D65 and/or D82a, we prepared the following OKT3-scFv mutants: OKT3-scFv-D65G, OKT3-D82aN, OKT3-D65G/D82aN, OKT3-D65A/D82aN, and OKT3-D65N/D82aN. We designed OKT3-scFv-D65A/D82aN to verify whether the smallest side chain, -CH3, could tolerate VH-SpA binding and OKT3-scFv-D65N/D82aN to evaluate the effect of charge conversion in the side chain. Each OKT3-scFv single or double mutant was prepared in an Escherichia coli expression system and purified using IMAC, followed by gel filtration analysis to fractionate the monomers. Peaks corresponding to the monomers were observed in the chromatograms; successful purification of each OKT3-scFv mutant was confirmed by sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) under reducing conditions (Supplementary Figure S4). The SpA-binding ability of each purified OKT3-scFv was evaluated by performing SpA affinity chromatography using rProtein A Sepharose Fast Flow (Cytiva). Only OKT3-scFv-D65G/D82aN showed improved binding to the SpA-packed column (Figures 2 and S5). Interestingly, OKT3-D65A/D82aN and OKT3-D65N/D82aN did not exhibit this effect. These results indicate that mutations at position 65, except for glycine, prevented SpA-binding, at least in OKT3-scFv. Therefore, we developed an SpA-binding OKT3-scFv mutant through amino acid mutagenesis at positions 65 and 82a.

Figure 2. SpA-binding evaluation of OKT3-scFv mutants using SpA-packed column. Each fraction was eluted with two column volumes (CVs) of buffer.

Integration of SpA-binding ability to bsAb format

To develop a therapeutic antibody that can be purified by SpA affinity chromatography, we designed a mutant of humanized bsAb, hEx3, consisting of an anti-epidermal growth factor receptor (EGFR) antibody 528 and anti-CD3 antibody OKT3, as reported previously [33]. On the basis of hEx3-scDb-LH, which was in a bispecific scDb format with a VL-VH domain order and promising variant [29], we applied the same mutagenesis approach as used for OKT3 to 528 to increase its SpA-binding ability. In addition to D65G/D82aN mutations in OH, we introduced the following mutations in the VH domain of 528 (5H): K19R, N65G, T70S, E81Q, S82aN, and R82bS (Table 1). The resulting hEx3-scDb-LH mutant (hEx3-scDb-LH-8m) and hEx3-scDb-LH were expressed in a Brevibacillus choshinensis expression system, as described previously [34], and purified by IMAC and gel filtration to fractionate the monomers. Taking an approach similar to what we did with OKT3-scFvs, we confirmed the presence of hEx3-scDb-LH monomers with high purities in the main peaks by SDS-PAGE under reducing conditions (Figures 3 and S6). Unlike hEx3-scDb-LH, hEx3-scDb-LH-8m was captured on an SpA-packed column and eluted using an appropriate buffer (Figure 4a). Next, we evaluated the SpA-binding affinity by performing an SpA-immobilized plate assay using fluorescein isothiocyanate-labeled hEx3-scDb-LHs. The fluorescence intensity of hEx3-scDb-LH-8m increased in a concentration-dependent manner (Figure 4b), and the dissociation constant (K_D) was calculated as 1.4 × 10⁻⁶ (M). This K_D was comparable to previously reported values for the interaction between VH and each SpA domain [40] and is considered adequate for purifying hEx3-scDb-LH-8m using SpA affinity chromatography. Thus, we constructed scDb with SpA-binding ability and adequate K_D by introducing mutations.

Figure 3. Preparation of hEx3-scDb-LH and hEx-scDb-LH-8m. (a) gel filtration chromatographs of each hEx3-scDb-LH and (b) sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) analysis of monomer fractions, as indicated by arrows.

Figure 4. SpA-binding evaluation of hEx3-scDb-LH and hEx3-scDb-LH-8m. (a) SpA-binding evaluation using an SpA-packed column. Each fraction was eluted with two column volumes (CVs) of buffer. (b) evaluation of SpA binding in an SpA-immobilized plate assay.

Evaluation of functions of hEx3-scDb-LH mutant

To confirm the mutagenic effects on the biological functions of hEx3-scDb-LH, we evaluated the target cell-binding ability using flow cytometric analysis and growth inhibition effects on cancer cells in an MTS assay using 3-(4,5-dimethylthiazole-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2 H-tetrazolium inner salt as a detection reagent. Both hEx3-scDb-LHs showed similar cell-binding abilities in EGFR-positive TFK-1 cells and CD3-positive T-LAK cells (Figures 5a and S7). In addition, the growth inhibition of cancer cells mediated by the cross-linking of TFK-1 cells and T-LAK cells by bsAbs was confirmed in a concentration-dependent manner to similar levels between hEx3-scDb-LH and hEx3-scDb-LH-8m (Figure 5b). These results suggest that the introduced mutations did not severely influence antigen recognition and growth inhibition by hEx3-scDb-LH, as each contact site with SpA and the antigens were distant from each other.

Figure 5. Evaluation of antibody functions before and after multiple mutagenesis. (a) cell-binding ability to EGFR-positive TFK-1 cells (left panel) and CD3-positive T-LAK cells (right panel) evaluated by flow cytometric analysis. (b) growth inhibition of cancer cells evaluated in an MTS assay using 5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2 H-tetrazolium inner salt as a detection reagent. The ratio of TFK-1:T-LAK was 1:4.

Preparation and evaluation of hEx3-scDb-LH-8m without affinity tags

Finally, we examined whether tag-free purification could be achieved using SpA affinity chromatography with high purity. We constructed and expressed hEx3-scDb-LH-8m without affinity tags (hEx3-scDb-LH-8m(-)) in the B. choshinensis expression system and purified hEx3-scDb-LH-8m(-) from the culture supernatant using SpA affinity chromatography. The capture of hEx3-scDb-LH-8m and hEx3-scDb-LH-8m(-) on the SpA column was confirmed by SDS-PAGE under reducing conditions (Figure 6a), and single peaks corresponding to monomers with high purity were observed in gel filtration analysis and SDS-PAGE under reducing conditions (Figure 6b, c). Notably, a peak shift to a smaller molecular weight was observed in hEx3-scDb-LH-8m(-), caused by the removal of c-Myc and His tag. Flow cytometric analysis showed that the binding ability of hEx3-scDb-LH-8m(-) was similar to those of other hEx3-scDb-LHs for each target cell (Figures 7a and S7). Interestingly, slightly stronger growth inhibition effects were observed at 1 nM of hEx3-scDb-LH-8m(-) (Figure 7b), which was consistent with the results of our previous study [41]. While the detailed mechanism is unknown, local structural differences may affect cross-linking between target cells. These results indicate that the functional small bsAbs were purified through SpA affinity chromatography, similar to conventional intact antibodies and Fc-fusion proteins, without affinity tags.

Figure 6. Purification of hEx3-scDb-LH-8m and hEx3-scDb-LH-8m(-) from culture supernatant in SpA affinity chromatography. (a) sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) analysis after single-step SpA affinity chromatography. (b) gel filtration chromatographs of each hEx3-scDb-LH-8m and (C) SDS-PAGE analysis of monomer fractions indicated by arrows.

Figure 7. Evaluation of tag-free hEx3-scDb-LH-8m functions. (a) cell-binding ability to EGFR-positive TFK-1 cells (left panel) and CD3-positive T-LAK cells (right panel) was evaluated by flow cytometric analysis. (b) growth inhibition of cancer cells was evaluated in an MTS assay using 5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2 H-tetrazolium inner salt as a detection reagent. The ratio of TFK-1:T-LAK was 1:4.

Application to another small bsAb format

To verify the versatility of our mutagenesis approach, we applied this strategy to taFv, which is another small bsAb format, and evaluated its SpA-binding ability. We previously identified an attractive hEx3 taFv variant, hEx3-ta6, with high expression levels and high cancer growth inhibition effects conferred by its comprehensive domain rearrangements [34]. hEx3-ta6 and hEx3-ta6-8m with the same mutations as hEx3-scDb-LH-8m were prepared using a method similar to that used for hEx3-scDb-LHs. The results of SpA affinity chromatography showed that the amount of hEx3-ta6-8m, in contrast with that of hEx3-ta6, decreased in the flow-through and wash fractions and increased in the eluted fractions (Figure 8a). hEx3-ta6 and hEx3-ta6-8m showed similar inhibitory effects on cancer growth (Figure 8b). Thus, our mutagenesis approach for integrating SpA-binding ability appears to be versatile for small antibody formats.

Figure 8. Evaluation of hEx3-ta6 and hEx3-ta6-8m. (a) SpA-binding evaluation using SpA-packed column. Each fraction was eluted with two column volumes (CVs) of buffer. (b) growth inhibition of cancer cells was evaluated in an MTS assay using 5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2 H-tetrazolium inner salt as a detection reagent. The ratio of TFK-1:T-LAK was 1:4.

Discussion

Small antibodies have been applied in various life science fields, including immunotherapy and biosensors [4,5,42]. Artificial tags are widely used to purify small antibodies [14]; however, use of these tags may cause adverse events, such as the generation of heterogeneous proteins of interest during the preparation process or immunogenicity during clinical use [15,16]. SpA affinity chromatography is often applied to purify intact antibodies or Fc-fusion proteins because of its high binding affinity for the Fc region [43]. SpA can bind to some VHs classified in the VH3 subfamily [26]. In some mutagenesis studies, the SpA-binding ability was integrated into the non-SpA-binding VHH [27,28]. However, small therapeutic antibodies that are already approved or those that will be approved may include various VH subfamilies and formats [1]. Some therapeutic small bsAbs without SpA-binding ability, such as blinatumomab, must be purified by using a tag or other artificial tag systems [44].

We first identified the amino acids crucial for SpA-binding and verified whether OKT3-scFv, which belongs to the VH3 subfamily but does not bind to SpA, could acquire SpA-binding ability through mutagenesis. Importantly, the VH on the cocrystal structure was derived from a human antibody; therefore, we predicted that limited additional immunogenicity would be induced. Based on the amino acid alignment results, different amino acids between SpA-binding VH3 and VH of OKT3-scFv (OH) were observed at positions 57, 65, and 82a (Table 1). We did not replace Thr57 in CDR2 for two reasons: Thr57 is involved in CD3 binding via a hydrogen bond and van der Waals interaction, and threonine and arginine at this site were tolerated for SpA-binding in a previous study [26,45]. Although position 65 is also found in CDR2, similar to position 57, aspartic acid is not related to CD3 binding, and side chain collision with the SpA domain prevents SpA-binding. In addition, according to previous cocrystal structural studies, asparagine at position 82a is critical for forming a hydrogen bond with serine 33 on SpA [26]. Thus, we prepared and evaluated mutants at positions 65 and 82a to convert non-SpA-binding OKT3-scFv to SpA-binding OKT3-scFv (Figure 2, Supplementary Figure S5). At position 65 on OKT3-scFv, similar to the previous in silico prediction of VHH [27], mutagenesis of asparagine did not confer SpA-binding ability. Notably, mutagenesis of alanine also abolished SpA-binding ability, most likely because of steric hindrance caused by the presence of a small methyl group on the side chain. Our results show that only glycine, with no side chain, was tolerated at position 65. Thus, glycine is important for SpA-binding, as is asparagine at position 82a, at least in OKT3-scFv. These results may guide the design of future small functional antibodies that enable SpA affinity purification.

Next, we integrated SpA-binding ability into a small bsAb with a more complicated structure than that of scFv. We have developed several bsAbs, including those with small molecular weights [33,46,47]. We applied the same mutagenesis approach as OKT3-scFv to hEx3-scDb-LH [29,46], a promising bsAb that strongly inhibits cancer growth. The binding ability of hEx3-scDb-LH-8m to SpA was confirmed using SpA affinity chromatography and an SpA-immobilized plate assay (Figure 4). In hEx3-scDb-LH-8m, lysine at position 19 was replaced with arginine, which decreased the number of reactive groups for fluorescent labeling; the numbers of fluorescein per protein were 6.0 for hEx3-scDb-LH and 5.3 for hEx3-scDb-LH-8m. Nevertheless, hEx3-scDb-LH-8m showed stronger fluorescence intensity than hEx3-scDb-LH, indicating that SpA binding was improved by mutagenesis. The calculated K_D was comparable to previously reported values for VH-SpA binding [40], and we purified small bsAbs without using affinity tags (Figure 6). Because excessively strong SpA-binding requires harsh elution conditions, such as lower pH or salinity, we integrate a moderate but adequate SpA-binding ability to purify hEx3-scDb-LH. Although the recovery rate was approximately 70% compared with that of IMAC purification, the preparation of tag-free cancer therapeutic antibodies has several advantages, and the recovery rate can be improved by optimizing the purification conditions. Finally, we verified whether the mutagenesis approach could enhance structural versatility using another small bispecific format, hEx3-ta6, the taFv variant of hEx3 (Figure 8). In addition, we confirmed that mutagenesis did not severely affect the antigen-binding ability and cytotoxicity of all the evaluated small bsAbs (Figures 5a, b, 7a, b and 8b). This is because the binding sites for antigens and SpA are mainly separate from each other, demonstrating the versatility of our mutagenesis approach.

Conclusion

We integrate the SpA-binding ability into anti-CD3 OKT3-scFv by performing mutagenesis based on alignment and structural modeling. This strategy was also applied to small bsAbs and those engineered bsAbs purified using SpA affinity chromatography; bsAbs were found to retain their biological functions. Conferring conventional antibody fragments with SpA-binding ability simplifies the purification and monitoring of the production process and thus can accelerate the development of small therapeutic antibodies. Furthermore, because of its immunoactivity, the OKT3 mutant with SpA-binding ability can serve as a building block for the development of small, engineered antibodies such as bsAbs.

Table

SpA	Staphylococcus aureus protein A
bsAbs	bispecific antibodies
IMAC	immobilized metal affinity chromatography
His tag	histidine tag
VH	variable heavy domain
VHH	variable domain of heavy chain of heavy chain antibody
scFv	single-chain Fv
Dbs	diabodies
taFvs	tandem scFvs
scDb	single-chain Db
OH	VH domain of OKT3
SDS-PAGE	sodium dodecyl sulfate polyacrylamide gel electrophoresis
EGFR	epidermal growth factor receptor
5H	VH domain of 528
KD	dissociation constant
CVs	column volumes
PBS	phosphate-buffered saline
MTS	3-(4,5-dimethylthiazole-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2 H-tetrazolium inner salt

Authors’ contributions

Atsushi Kuwahara: Investigation, Writing – Original Draft; Misae Nazuka: Investigation; Yuri Kuroki: Investigation; Kohei Ito: Methodology; Shunsuke Watanabe: Methodology; Izumi Kumagai: Conceptualization; Ryutaro Asano: Conceptualization, Writing – Review & Editing. All authors read and approved the final manuscript.

Ethics statement

This study was approved by the biosafety subcommittee for safe handling of living modified organisms at Tokyo University of Agriculture and Technology (Permission number: R2–60) and carried out according to the guidelines of the committee.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

The authors confirm that the data supporting the findings of this study are available within the article and its supplementary materials.

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/21655979.2023.2259093

Additional information

Funding

This work was supported by Grants-in-Aid for Scientific Research from the Japan Society for the Promotion of Science (JSPS) under Grant [number 20H02534, 21K18321, and 22H02915] and the Japan Agency for Medical Research and Development (AMED) under Grant [numbers 22cm0106278 and 22ak0101143].