Expression and Purification of a Bispecific Antibody against CD16 and Hemagglutinin Neuraminidase (HN) in E. Coli for Cancer Immunotherapy

Background: Immunotherapy of cancer by bispecific antibodies (bsAb) is an attractive approach for retargeting immune effector cells including natural killer (NK) cells to the tumor if the proper expression and purification of the bsAb for such applications could be addressed. Herein, we describe E. coli expression of a recombinant bsAb (bsHN-CD16) recognizing NK-CD16 and hemagglutinin neuraminidase (HN) of Newcastle Disease Virus (NDV). This bsAb might be efficient for ex vivo stimulation of NK cells via coupling to HN on the surface of the NDV-infected tumor cells. Methods: A bsAb-encoding pcDNA3.1 vector (anti-HN scFv-Fc-anti-CD16 scFv) was used as a template, and the scFv segments (after enzymatic digestion and cutting of the Fc part) were rejoined to construct the Fc-deprived bsAb (anti-HN scFv-anti-CD16 scFv; bsHN-CD16). The constructed bsHN-CD16 was inserted into the HindIII and BamHI site of the T7 promoter-based pET28a plasmid. Following restriction analyses and DNA sequencing to confirm the cloning steps, bsHN-CD16 encoding pET28a was transformed into the E. coli (Rosetta DE3 strain), induced for protein expression by IPTG, and the protein was purified under native condition by Ni/NTA column using imidazole. Results: Analyses by SDS-PAGE and Western Blotting using Rabbit anti-human whole IgG-HRP conjugate, confirmed the expression and purification of the bsAb with the expected full size of 55 kDa and yields around 8% of the total protein. Conclusions: Results showed efficient production of the bsAb in E. coli for future large-scale purification.


Introduction
Cancer with approximately 9.6 million (1 out 6) annual deaths, was one of the main causes of death worldwide in 2018 (1). Due to insufficiency of conventional cancer therapies, novel and promising treatment approaches such as immunotherapy are strongly sought.
Natural killer (NK) cells are non-B, non-T lymphocytes, and as a part of the innate immune system, they can recognize and kill virus-infected or transformed cells without prior sensitization (2). The activity of NK cells is regulated by a balance between inhibitory and activating receptors on the surface of the NK cells. There are different activating receptors on the surface of NK cells comprising: Natural killer receptors (NCRs), NKG2D, and CD16 (FcγRIII) (2,3). It is shown that Rep. Biochem. Mol. Biol, Vol.9, No.1, Apr 2020 53 decreased activity of peripheral blood lymphocyte specially NK cells is associated with higher cancer risk (4)(5)(6). Of note, the inhibitory effect of the tumor microenvironment might also decline the effector functioning of the NK cells by downregulation of activating receptors such as NCRs (7)(8)(9). Therefore, the development of therapeutic approaches to revive NK cells functioning is a major concern in the immunotherapy of cancer (10). Application of bispecific antibodies (bsAb) is one of the promising approaches to preserve the effector function of NK cells against tumor cells (11). Bispecific antibodies are protein engagers that bind to two distinct ligands at the same time and thereby activate the immune effector cell against the target cell (12). In this context, bsAbs which can effectively activate NK cells by binding to the EpCAM antigen (a tumor associated antigen) (13) or CD33 (14) on the surface of the tumor cell from one side and to the CD16 activating receptor on the surface of NK cells from the other side were reported. However, due to the immune suppressive condition of the tumor region, it might be needed to enhance the danger signals that are required to elicit a strong immune response similar to bacteria/virus infection (15). In this context, infection of the tumor cells by a safe, non-lytic virus with potent immuno-stimulatory effects, such as Newcastle Disease Virus (NDV) Ulster strain was reported (15). This strategy provided the required strong danger signals via the tumor cell surface-expressed, viral hemagglutininneuraminidase (HN) protein. Accordingly, bsAbs designed to attach the surface-displayed HN from one end and the CD3 or CD28 receptors on the T cells from the other end could efficiently activate these immune (T) cells against the tumor (16). This final report suggested the possibility of applying the same strategy to activate NK cells against tumor cells, too.
An important point for the success in the application of recombinant bsAbs is their efficient and appropriate expression in a proper host. To this end, the expression of several bsAbs in different expression systems, including E. coli, have been reported (13,(17)(18)(19). Indeed, E. coli expression system, due to the ease of manipulation, availability of a high number of vectors with various promoters and high yields, is the first choice in most of the recombinant protein expression applications (20)(21)(22). Also, there are plenty of reports on the application of this host for expression of bsAbs in proper structure and high yields (13,17,18,20). In the present study, we successfully expressed a recombinant bsAb (bsHN-CD16) recognizing CD16 and NDV-HN in E. coli for the final application in cancer immunotherapy by activation and retargeting of NK cells towards NDV-infected tumor cells.

Bacterial strains and Plasmids
E. coli DH5-alpha strain was used for cloning purposes. pET28a (Novagen, USA) was used as a bacterial expression vector, and E. coli strain Rosetta (DE3) was used as an expression host. Propagation of E. coli for protein expression was in 2x LB medium (Merck, Germany).

Cloning and construction of the recombinant bsAb-encoding plasmid
The primary bsAb-encoding construct (anti-HN scFv-Fc-anti-CD16 scFv) that had been subcloned in the pcDNA3.1(-) vector was engineered by Dr. Frank Momburg (German Cancer Research Center, DKFZ). The construct contained an scFv antibody derived from the hybridoma HN.B (23) (kindly provided by Dr. V. Schirrmacher, German Cancer Research Centre) and an scFv antibody derived from the hybridoma A9 (24) (kindly provided by Dr. Gerhard Moldenhauer, German Cancer Research Centre) linked by the Fc portion of human IgG1. This plasmid was used as a template, and the scFv segments (after enzymatic digestion and cutting of the Fc part) were rejoined to construct the Fc-deprived bsAb (antiHN scFv-anti-CD16 scFv; bsHN-CD16). The constructed bsHN-CD16 was inserted into the HindIII and BamHI site of the T7 promoter-based pET28a plasmid. Colony-PCR, digestion analyses, and sequencing were performed to verify that the cassette is cloned correctly in the frame. All cloning and molecular procedures were based on conventional protocols (25).

Expression of the bsAb in E. coli
The recombinant plasmid was transformed into E. coli Rosetta DE3 competent cells. Transformed bacteria were grown in 800 ml 2x Luria Broth Rep. Biochem. Mol. Biol, Vol.9, No.1, Apr 2020 52 supplemented with 50 ug/ml kanamycin and 25ug/ml chloramphenicol. Induction of expression was done when OD600 reached 0.6 by the addition of 1 mM isopropyl-b-Dthiogalactopyranoside (IPTG) (Merck, Germany) at 37 °C in a shaker incubator. Three hours after induction, bacteria were harvested by centrifugation at 4000 x g, 15 minutes at 4 °C and the pellet stored at -70 °C after washing the pellet with PBS with the same centrifugation conditions.

Protein Purification
Protein purification was done under native condition by affinity chromatography using Ni/NTA column according to the manufacturer's protocol (The QIA expressionist TM . 2003). In brief, the bacterial pellet was thawed, and about 3 ml of lysis buffer (50 mM NaH2PO4, 300 mM NaCl, 10 mM imidazole, pH 8) was added and gently shaken in the refrigerator for about 1 hour. Next, sonication was done to lyse the bacterial cell completely (240 W, 20 s sonication, 20 s rest on ice for 15 cycles). To remove the pellet, centrifugation was done with 10000g, 30 min at 4 °C. The supernatant was collected to be applied on the Ni/NTA column. About 1 ml of resin was added to the column, and after removing the alcohol, the resin was washed with 3 ml of the lysis buffer. Next, the lysate supernatant solution was mixed very gently (about 50 rpm) with the resin on a rotary shaker at 4 °C in a 15 ml falcon tube. The mixture was loaded into a column, and the flow-through was collected. The column was washed twice with 4 ml of wash buffer (50 mM NaH2PO4, 300 mM NaCl, 20 mM imidazole, pH 8) and then the protein was eluted 5 times with 0.5 ml elution buffer (50 mM NaH2PO4, 300 mM NaCl, 250 mM imidazole, pH 8) which were collected in tubes to be analyzed by SDS-PAGE later. The purified protein under native condition was further passed through 30 kDa Amicon-filters to ensure removal of any protein bands below this size.

Protein characterization
The pellet from 1 ml of the medium before and after induction was characterized by SDS-PAGE and then by Western blotting using rabbit anti-human whole IgG HRP conjugate (Sigma, UK) using standard protocols (25). In brief, after separation by electrophoresis, protein bands were transferred to polyvinylidene difluoride membrane (PVDF) (Roche, Germany) in wet condition and blocked by 5% skim milk in PBS plus 0.1% Tween-20 (PBS-T) overnight at 4 °C. Blots were then incubated with anti-human whole IgG HRP conjugate with a concentration of 1:2000 in PBS-T for two hours. After washing steps with PBS-T, detection step was done by addition of 3,3'-diaminobenzidine (DAB) (Sigma-Aldrich). Seven mg DAB in 10 ml PBS in the presence of 14 µl hydrogen peroxide was added to the blot, and the expected bands were observable after a few minutes.
Protein expression yield was measured by image analysis using the open source software, FiJi (26). Briefly, the image of the acrylamide electrophoresis gel was acquired by a commercial camera. The resulted image was converted to a 32-bit image, and the background was subtracted to reduce the noise level. Then the ratio of the area under the curve related to the protein of interest divided to the total area under the curve related to the bacterial crude lysate to obtain protein expression yield.

Construction of the bsAb (Anti HN-Anti CD16)encoding vector for expression in E. coli and restriction analyses
The primary bsAb (anti-HN scFv-Fc-anti-CD16 scFv) in pcDNA3.1(-) vector coding for a tetravalent bispecific antibody for secretory expression in mammalian cells was used as the template for developing bsAb coding sequence in bacterial vector. Since the Fc part in mammalians is post-translationally N-glycosylated and is not required for efficient production in bacteria, the Fc portion was deleted. To this end, first, the anti-HN part was separated by digesting the pcDNA3.1(-) plasmid containing the primary bispecific antibody by NheI at 37 °C. The intended band (about 940 bp) was extracted from the agarose gel and cloned into a new intact pcDNA3.1(-), which was digested by the same enzyme followed by 1 µg alkaline phosphatase, using T4 ligase. The positive clones with correct orientation were selected by colony-PCR using CMV-F and BGH-R universal primers, and digestion with EcoRI, where expected 960 and 5418 bp bands relating to the positive clones with correct orientation could be observed (Fig. 1). Since for developing a construct without Fc, it was necessary to separate the segment containing-CD16 scFv, the main bsAb-coding pcDNA3.1 plasmid was digested by HindIII and XbaI, and the 833 bp segment was extracted from the gel and cloned in a new pcDNA3.1 plasmid backbone. This construct was confirmed by digestion with PstI restriction enzyme where the 4061 bp and 1840 bp segments could be seen on the gel (Fig. 2). Subsequently, to generate the pre-final construct (bsHN-CD16 in pcDNA3.1), the anti-HN coding plasmid was digested with NheI, while the anti-CD16 coding sequence was digested by XbaI restriction enzyme. Since NheI and XbaI create compatible cut sites, the anti-HN scFv coding sequence was inserted beside the anti-CD16 coding sequence inside the pcDNA3.1(-) by T4 ligase, after the alkaline phosphatase treatment for ten minutes. Colony-PCR and digestion with EcoRI confirmed the accuracy of cloning and the correct orientation (Fig. 3). Finally, the bacterial expression vector, pET28a, and the pre-final construct were digested by NcoI and HindIII. The cut with NcoI removed the ER leader sequence of the original construct. The insert (antiHN-antiCD16) was extracted from the gel and ligated to the digested vector according to the standard protocol by T4 ligase (25). The final construct was confirmed by colony-PCR, digestion by BstEII restriction enzyme to generate expected 4676, 1357, and 809 bp segments, and finally sequencing (Fig. 4).

Expression, purification, and characterization of the E. coli-derived bsHN-CD16 Ab
As shown in Fig. 5A, induction of the E. coli expression hosts harboring the pET28a-encoding bsAb by IPTG indicated the expected 55 kDa protein band corresponding to the bsHN-CD16. Western blotting by rabbit anti-human whole IgG HRP conjugate confirmed the nature of the expressed antibody (Fig. 5B). Image analysis of the SDS-PAGE protein bands (Fig. 5A) by Fiji software (26) indicated a yield of 8% of total bacterial protein, that showed more than 90% purity after purification with Ni-NTA method under native condition. This purity was further enhanced by passage through 30 kDa Amicon filters (Fig. 6).

Discussion
In this study for the first time, to our best of knowledge, we have successfully cloned, expressed and purified a recombinant bispecific antibody (bsHN-CD16) in E. coli strain Rosetta (DE3) with acceptable yields for potential large-scale plans. This immune engager is supposed to bind to the viral HN protein displayed on the surface of the NDV infected tumor cells from one side and to the CD16 activating receptor on the surface of the NK cells from the other side. Since E. coli expression system, due to fast, cost-effective and efficient yield of protein production, is the first choice for protein production (20,21), we decided to produce the bsAb in this bacterial expression system.
As shown in figures 1 to 3, the challenging step in the construction of the E. coli-expression vector for this bsAb was deletion of the Fc region. In fact, Fc fragment is a highly glycosylated structure in Abs and considering the unavailability of the glycosylation system in E. coli cells, presence of the Fc in the target constructs might result in the inhibition or reduction of the protein-expression yields in E. coli (22). Also, the CD16 on the NK cell might act as the receptor for the Fc part of the antibodies (FcγRIII) (27). Thus, deletion of the Fc part from the construct might ensure that binding of the bsAb to the CD16, occurs only via the CD16 scFv part. Consistent with our approach, there are other studies addressing the construction of an NK engager without Fc part for retargeting NK cells toward the tumor cells (13,14,17,28).
Several other studies reported the expression of bispecific antibodies with the ability of engaging CD16 on the surface of the NK cells in E. coli system, including: bsCD16-CD133 (against colorectal cancer cell) (18), bsCD16-CD33 (against HL-60 cell line, a human leukaemia cell line) (14) or bsCD16-EpCAM (against HT-29 cell line, colorectal cell line) (29). Of note, most of these prior mentioned studies reported the use of BL2 strain of E. coli as an expression host. However, since the CD16 scFv in our engager (as well as in the prior reports) is a human-derived sequence, using Rosetta (DE3) strain in our study, as a BL21 derivative designed to enhance the expression of eukaryotic proteins, might be a better choice. The Rosetta (DE3) strain supplies tRNA genes for AGG, AGA, AUA, CUA, CCC, GGA (codons rarely used in E. coli) on a Col-E1 compatible chloramphenicolresistant plasmid (30) and might decrease the need for codon optimization.
Image analysis by Fiji software (27) determined the yield of our bsAb expression, which was about 8% of the total bacterial proteins. Although this yield was acceptable, there are reported strategies to further improve the expression yields of bispecific antibodies in E. coli expression systems (28). Accordingly, in one study, incorporation of a segment of human muscle aldolase (hma) in the bsAb was reported to improve the expression yields by 20-fold which resulted in enhanced purity from 10% to 80% (31). In another study, the inclusion of the 218s linker (GSTSGSGKPGSGEGSTKG) improved proteolytic stability and reduced the degree of protein aggregation (28). Recently, by optimizing the cultivating media and induction condition (M9 minimal medium and 24 h shaking at 16 °C after induction by 1 mM IPTG), expression levels for a bsAb in E. coli BL-21 strain were improved to an acceptable level of 0.6 mg/200 ml (19). Therefore, it might be possible to further improve the expression level of our bsAb with similar strategies.
Due to the presence of 6xHis-tag (C-terminal poly-histidine tag with a free carboxyl group) in our bsAb, the purification of bsAb was performed by Ni/NTA column and under native condition. Although the yield of purification under native condition is less than denaturing condition, to avoid solubilizing and refolding steps, and to retain the natural structure of the bsAb, we decided to choose native purification protocol. As shown in Fig. 6A, several unwanted small sized bacterial proteins were present in our natively purified bsAb preparations which were removed by passing through a 30 kDa Amicon filter (Fig.  6B). Our approach is in contrast to several prior studies that had purified the E. coli-derived bsAbs under denaturing condition (13,17,31). In fact, despite an acceptable purity of 80 to 90% after denaturing condition of protein purification, returning the natural structure and biological activity to the produced antibodies would probably require a complex refolding processes (13,17,31).
Altogether, in the present study, to our best of knowledge, we provided the first report on successful cloning and expression of a bispecific antibody (bsHN-CD16) for recognition of the CD16 and NDV-HN in the Rosetta (DE3) strain of E. coli expression system. This bsAb is designed to bind to CD16 activating receptor on NK cells from one side and to the viral protein HN on the surface of the NDV-infected tumor cell from the other side. Further studies on this bsAb, including binding and cytotoxicity assays, may demonstrate the potential of this bsAb for application in cancer immunotherapy. There is no conflict of interest.