Structural and In Vitro Functional Analyses of Novel Plant-Produced Anti-Human PD1 Antibody

Immunotherapy has emerged as a promising and effective treatment for cancer. The frequently used immunotherapy agents are immune checkpoint inhibitors, such as antibodies specific to PD1, PD-L1, or CTLA-4. However, these drugs are highly expensive, and most people in the world cannot access the treatment. The development of recombinant protein production platforms that are cost-effective, scalable, and safe is needed. Plant platforms are attractive because of their low production cost, speed, scalability, lack of human and animal pathogens, and post-translational modifications that enable them to produce effective monoclonal antibodies. In this study, an anti-PD1 IgG4 monoclonal antibody (mAb) was transiently produced in Nicotiana benthamiana leaves. The plant-produced anti-PD1 mAb was compared to the commercial nivolumab produced in CHO cells. Our results showed that both antibodies have similar protein structures, and the N-glycans on the plant-produced antibody lacks plant-specific structures. The PD1 binding affinity of the plant-produced and commercial nivolumab, determined by two different techniques, that is, enzyme-linked immunosorbent assay (ELISA) and surface plasmon resonance (SPR), are also comparable. Plant-produced nivolumab binds to human PD1 protein with high affinity and specificity, blocks the PD-1/PD-L1 interaction, and enhances T cell function, comparable to commercial nivolumab. These results confirmed that plant-produced anti-PD1 antibody has the potential to be effective agent for cancer immunotherapy.

complexed proteins, such as antibodies, and perform necessary posttranslational modifications 8 . Plant-produced recombinant proteins were shown to meet the same good manufacturing practice criteria as proteins produced in mammalian cells, such as appearance, stability, half-life, immunogenicity etc 9 . These advantages for plant platforms suggest that they represent an attractive alternative for producing biopharmaceuticals for developing countries 10,11 .
In this study, we further developed an anti-PD1 monoclonal antibody (mAb), nivolumab, as a possible cancer immunotherapeutic by applying plant technology that enables cost-effective and scalable antibody production. We produced nivolumab at 140 μg/g fresh Nicotiana benthamiana leaves within 6 days of infiltration. Plant-produced nivolumab retained similar structure and in vitro efficacy, compared to those of commercial mammalian cell-produced nivolumab. This study is the first to demonstrate the use of plant-derived mAb for cancer immunotherapy.

Expression and purification of anti-PD1 monoclonal antibodies in ΔXf N. benthamiana.
To increase the yield of monoclonal antibody (mAb) expression in plants, the coding sequences of the heavy chain (HC) and light chain (LC) of nivolumab 12 were optimized in silico with N. benthamiana-optimized codon, cloned into geminiviral vectors and transformed into Agrobacterium tumefaciens. A. tumefaciens harbouring pBY-Nivo-HC and pBYNivo-LC (Fig. 1A) were co-delivered into N. benthamiana leaves through vacuum infiltration. Co-expression of the HC and LC resulted in the assembly of the complete antibody consisting of 2HC and 2LC (Fig. 1B). The expressed antibody was purified using protein A affinity chromatography and subjected to SDS-PAGE. The nonreducing SDS-PAGE confirmed that the complete mAb assembled into its tetrameric form (Fig. 1C). The reducing gel electrophoresis confirmed that the HC and LC of anti-PD1 mAb were expressed in leaves with the expected molecular weights (Fig. 1D). Notably, differences in mobility are caused by the presence of the SEKDEL peptide on the HC and LC. Western blot analysis after reducing gel electrophoresis confirmed the expression of HC and LC with anti-human gamma and anti-human kappa antibodies (Fig. 1E,F, respectively). Moreover, the purified antibody was analyzed by size exclusion chromatography. The result confirmed that the purified plant-produced monoclonal antibody remains monomer as commercial nivolumab (Fig. 1G). The assembly of anti-PD1 mAb was confirmed by an ELISA that detects the assembled form of the antibody. ELISA results also showed that the anti-PD1 mAb was expressed at the highest levels on 6 dpi, producing approximately 140 µg/g leaf fresh weight (Fig. 2).
Plant-produced anti-PD1 exhibited mammalian-like N-linked glycosylation patterns. The N-glycan composition of the plant-produced nivolumab was compared to commercial CHO cell-produced antibody by liquid chromatography-electrospray ionization-mass spectrometry (LC-ESI-MS) analysis of tryptic glycopeptides (Fig. 3). As expected, the results confirmed that the N-glycan composition of nivolumab produced in ΔXF plants contained mainly GnGn (GlcNAc 2 Man 3 GlcNAc 2 ) structures that more closely resemble mammalian-type N-glycans. As expected, GnGnF (GlcNAc 2 Man 3 FucGlcNAc 2 ) carrying the mammalian-specific core fucosylation was the major N-glycosylated peptide in the commercial CHO cell-produced antibody (Fig. 3).
Secondary and tertiary structural characterization. The secondary and tertiary structures of plant-produced anti-PD1 were compared to commercial mammalian cell-produced antibody using circular dichroism (CD) and nuclear magnetic resonance (NMR) spectroscopy, respectively. The CD results (Fig. 4A) confirmed that the secondary structure of plant-produced anti-PD1 is similar to that of the commercial antibody. Both spectra indicate negative absorbance at 218 nm, which is a β-sheet-rich structure. For NMR spectroscopy, dispersed aromatic protons (Fig. 4B) and up-fielded methyl protons (Fig. 4C) were observed in both spectra, indicating that the tertiary structures are retained. Furthermore, both NMR spectra also confirmed that the tertiary structure of the plant-produced anti-PD1 antibody is similar to that of the commercial antibody.
Functional assays of plant-produced anti-PD1 mAb. The in vitro binding to human PD1 protein of the plant-produced anti-PD1 mAb was determined by ELISA, as shown in Fig. 5A. Both plant cell-produced and commercial mammalian cell-produced anti-PD1 mAbs showed similar binding to human PD1 protein.
To assess the binding of the antibody more quantitatively, a surface plasmon resonance (SPR) assay was performed with purified plant-produced or commercial mammalian cell-produced anti-PD1 mAb immobilized on  www.nature.com/scientificreports www.nature.com/scientificreports/ a Protein G sensor chip. The human PD1 protein was generated and flowed across the solid-phase anti-PD1 mAb at various concentrations. Plant-produced anti-PD1 mAb showed similar binding affinity and kinetics for human PD1 protein compared to commercial mammalian cell-produced mAb (Fig. 6). The association rate constant k a , dissociation rate constant k d , and equilibrium dissociation constant K D of the plant-produced and commercial  www.nature.com/scientificreports www.nature.com/scientificreports/ nivolumab are detailed in Table 1. The K D between human PD1 and the plant-produced and commercial antibodies are 14 nM and 12 nM, respectively (Table 1). This data confirmed that both mAbs can bind to human PD1 protein with high affinity similarly.
Using a luciferase reporter system, where luciferase expression is under the control of an nuclear factor of activated T-cell (NFAT) promoter, plant-produced anti-PD1 mAb induced concentration-dependent NFAT activation, similar to the commercial mammalian cell-produced antibody, with EC 50 values of 496 ng/ml and 544 ng/ ml, respectively, being observed (Fig. 7). Human IgG was used as a negative control in this study.
The ability of plant-produced anti-PD1 mAb to promote T cell responses was measured in vitro. The human PBMCs were stimulated by the Staphylococcal enterotoxin B in the presence of 1 μg/ml anti-PD1 mAb. Plant-produced nivolumab enhanced both IL-2 (Fig. 8A) and IFN-γ (Fig. 8B) secretion, comparing to human IgG4 negative control. In addition, the levels of IL-2 and IFN-γ enhancing from the activation of plant-produced nivolumab are not significantly different from commercial nivolumab. These experiments showed that  www.nature.com/scientificreports www.nature.com/scientificreports/ plant-produced anti-PD1 mAb can bind potently to human PD1 protein, block the PD-1/PD-L1 interaction, and activate T cell response.

Discussion
Immunotherapy is an attractive treatment for many cancer patients. In particular, treatment with monoclonal antibody (mAb)-based immune checkpoint blockade enhances the anti-tumour function of T cells, yielding promising results in clinical studies [8][9][10][11] . However, immunotherapy is highly expensive, which could reach up to 1 million US dollars per patient 12 . Therefore, most people in developing countries cannot access this treatment. Plant platform was developed recently for producing Our goal is to develop an alternative platform for producing effective immune checkpoint blockade mAb. In this study, we investigated the production of anti-human PD1 mAb in plants. Our results show that plant-derived anti-PD1 nivolumab mAb has similar structure and binding potency compared to mammalian cell-produced mAb. Moreover, the plant platform can address the cost and scalability issues associated with antibody production.
Transgenic plants are still prohibited in many countries 13 . To avoid concerns about genetically modified plants, transient expression technology was used in this study. Plants were recently developed to produce several monoclonal antibodies, which mostly are infectious therapeutic antibodies [15][16][17] and anti-cancer antibodies 18-20 . Our study firstly showed the development of plant-produced IgG4 monoclonal antibody for immunotherapy. The nivolumab heavy chain and light chain were modified by codon optimization based on the Nicotiana benthamiana codon bias. The sequence Ser-Glu-Lys-Asp-Glu-Leu (SEKDEL), the endoplasmic reticulum (ER) retention motif, was added to the C-terminus of the heavy chain. The SEKDEL motif has been shown to increase the recombinant protein levels for some antibodies 21 . Presumably because the ER is a favorable compartment for protein storage as it lacks proteases that may degrade the recombinant protein.
The nivolumab was rapidly expressed in N. benthamiana leaves within 6 days of infiltration and successfully assembled into a native IgG monomer form (Fig. 1C,G). With genetic optimization and ER retention, mAb accumulated at an average of 140 µg/g of fresh leaf weight (Fig. 2). Previous study was shown that the PD1 binding of commercial nivolumab is highly potent 14 . Compared to the commercial mAb, both ELISA and SPR techniques showed that plant-derived nivolumab has equivalent binding kinetics in vitro.
The N-linked glycosylation of proteins in plants is generally similar to that in mammalian cells, but plants contain specific xylose and fucose residues that are not found in mammals 22 . Therefore, there is concern about the immunogenicity of plant-specific N-glycans in humans. The clinical data on the immunogenicity of plant  www.nature.com/scientificreports www.nature.com/scientificreports/ N-glycans are still controversial [16][17][18] . However, several studies showed there is no cases with allergic or hypersensitivity symptoms 9,23-25 . For immune checkpoint blockade purposes, the IgG4 subtype was used because the induction of ADCC by IgG4 is notably weak or nonexistent 19 . This study aims to compare the anti-PD1 mAb produced in plant and mammalian cells.
Therefore, glyco-modified plants that almost completely lack β1,2-xylose and core α1,3-fucose residues on N-glycans were used in this study 20 . The N-glycan structure attached to the conserved N-glycosylation site in the heavy chain of nivolumab produced in these plants is not identical to mammalian cell-produced nivolumab (Fig. 3). In particular, the plant-produced antibody lacks core fucosylation and terminal galactosylation and displays oligomannosidic N-glycans (Man5-Man9). The oligomannosidic N-glycans are the result of SEKDEL-mediated retention in the endoplasmic reticulum (ER). The glycan variation in the Fc region of the IgG affects the binding of the Fc receptor and the antibody-dependent cell-mediated cytotoxicity (ADCC) 26,27 and the anti-PD1 IgG4 with oligomannosidic N-glycans may also be subjected to faster in vivo clearance due to binding to the mannose receptors 28 . The ADCC of different anti-PD1 IgG4 glycosylation variants as well as the in vivo half-life will be analysed in future research. To this end, the anti-PD1 IgG4 will be produced in wild-type plants containing specific xylose and fucose residues, in ΔXF plants and without the SEKDEL peptide to reduce the amount of oligomannosidic N-glycans. Without the SEKDEL peptide the recombinant protein will be trafficking through the Golgi apparatus where the N-glycans are processed from oligomannosidic to complex N-glycans 22 . In addition to the glycan structure, the secondary and tertiary structures of the antibody produced from plant and mammalian cells are compared. The data from CD and NMR spectroscopy confirmed that the secondary and tertiary of both antibodies are similar (Fig. 4).
In addition to the glycan and protein structures, the binding efficacy of the antibody is critical. In vitro binding of both antibodies to human PD1 was analysed by ELISA and SRP (Figs 5 and 6). Our results confirm that plant-produced anti-PD1 mAb can bind to human PD1 protein approximately as well as mammalian cell-produced anti-PD1 mAb. To compare the potency and stability of both antibodies, a PD-1/PD1-L1 blockade cell-based bioassay (Promega, USA) was used. Interestingly, our data showed that both antibodies had similar PD-1/PD1-L1 blockade patterns at the same concentration (Fig. 7).
In summary, we established a plant cell-based platform to rapidly express and assemble anti-PD1 IgG4 mAb. Our data also show the similarities in structure and PD1 binding efficacy between plant cell-based and mammalian cell-based antibodies, supporting the potential of plant-produced anti-PD1 mAb for use in cancer immunotherapy. However, human PD1 knock-in mice are needed to study the efficacy of plant-produced anti-PD1 mAb in vivo. From a manufacturing perspective, the production cost of plant-produced mAb is lower than that of the mammalian system. Therefore, a plant-based alternative platform might reduce the cost of the current treatment with mAb-based immune checkpoint blockade.

Methods
Expression vector construction. Nucleotide sequences encoding nivolumab heavy chain (HC) and light chain (LC) were optimized in silico with Nicotiana benthamiana-optimized codons. The signal peptide was added to the N terminus, and SEKDEL was added to the C-terminus of both HC and LC genes. Optimized HC and LC sequences were synthesized (Bioneer company, South Korea). The nivolumab HC and LC plasmids were cut with XbaI and SacI before ligation into a geminiviral vector 23 resulting in pBYNivo-HC and pBYNivo-LC, respectively (Fig. 1A). The plasmids were transformed into Agrobacterium tumefaciens strain GV3101. 24 were grown in a controlled plant room with a 16 h light/8 h dark cycle at 28 °C for 6 to 8 weeks. Plants were co-agroinfiltrated with GV3101 strains containing pBYNivo-HC or pBYNivo-LC at OD 600 0.2 by vacuum infiltration. Infiltrated leaves were harvested on 2, 4, 6, 8, and 10 days post-infiltration (dpi) to evaluate the antibody expression pattern. For other experiments, plant leaves were harvested at 6 dpi. A blender was used to homogenize infiltrated leaves with 1x PBS (phosphate-buffered saline: 137 mM NaCl, 2.7 mM KCl, 4.3 mM Na 2 HPO 4 , 1.47 mM KH 2 PO 4 ) at pH 7.4 and centrifuged at 26,000 g and 4 °C for 40 min. Crude extract was filtered with a 0.45-μm membrane filter and loaded into a protein A bead column. After washing the column with 1x PBS pH 7.4, the protein was eluted with 100 mM glycine pH 2.7 and neutralized with 1.5 M Tris base to final pH value of 7.4.
Size exclusion chromatography. Fifty microliters of 2 µM protein solution was loaded on a Superdex 200 10/300 GL column (GE Healthcare, Little Chalfont, UK). The protein was eluted with elution buffer (50 mM Tris-HCl pH 7.5 and 150 mM KCl) at a flow rate of 0.3 ml/min. The eluted proteins were detected by the absorbance value at 280 nm.