The B cell death function of obinutuzumab-HDEL produced in plant (Nicotiana benthamiana L.) is equivalent to obinutuzumab produced in CHO cells

Plants have attracted attention as bio-drug production platforms because of their economical and safety benefits. The preliminary efficacy of ZMapp, a cocktail of antibodies produced in N. benthamiana (Nicotiana benthamiana L.), suggested plants may serve as a platform for antibody production. However, because the amino acid sequences of the Fab fragment are diverse and differences in post-transcriptional processes between animals and plants remain to be elucidated, it is necessary to confirm functional equivalence of plant-produced antibodies to the original antibody. In this study, Obinutuzumab, a third generation anti-CD20 antibody, was produced in N. benthamiana leaves (plant-obinutuzumab) and compared to the original antibody produced in glyco-engineered Chinese hamster ovary (CHO) cells (CHO-obinutuzumab). Two forms (with or without an HDEL tag) were generated and antibody yields were compared. The HDEL-tagged form was more highly expressed than the non-HDEL-tagged form which was cleaved in the N-terminus. To determine the equivalence in functions of the Fab region between the two forms, we compared the CD20 binding affinities and direct binding induced cell death of a CD20-positive B cells. Both forms showed similar CD20 binding affinities and direct cell death of B cell. The results suggested that plant-obinutuzumab was equivalent to CHO-obinutuzumab in CD20 binding, cell aggregation, and direct cell death via binding. Therefore, our findings suggest that Obinutuzumab is a promising biosimilar candidate that can be produced efficiently in plants.


Introduction
A recent breakthrough in cancer treatment employs immunotherapy with monoclonal antibodies that bind to a highly expressed target on cancer cells [1,2]. In contrast to radiation PLOS ONE | https://doi.org/10.1371/journal.pone.0191075 January 11, 2018 1 / 16 a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 immunogenicity was not achieved, the antibody removing plant-specific glycosylation by tagging the HDEL, which relieved the possibility of an allergenic response (antibody with HDEL tag has the high mannose type of oligosaccharide which found in mammals), can be capable of causing B cell death. Conversely, if the Fab function of Obinutuzumab produced in N. benthamiana leaves can be demonstrated as equivalent to that produced in CHO cells, this result would indicate that the protein sequence of the Fab portion of Obinutuzumab is not altered by plant-specific post-transcriptional processes. If this indeed is the case, it would be very easy to produce Obinutuzumab in a variety of transgenic plants or via highly productive expression systems that have solved the issue of plant-specific glycosylation.

Construction of obinutuzumab expression vectors in N. benthamiana and CHO cells
Light and heavy chain sequences of Obinutuzumab were obtained from GenBank (EA770939) and long chain cDNAs were synthesised by Bioneer (Daejeon, Korea) and plant codon-optimised according to the manufacturer's protocol (Bioneer, http://eng.bioneer.com, Daejeon, Korea). Original and plant codon-optimised light and heavy chains were inserted into the pCAMBIA 1300 binary vector (p35S:BiP ss-plant-obinutuzumab light chain/heavy chain-HDEL). For ER localisation, the signal sequence of BiP (BiP ss) and the ER retention signal sequence, HDEL, were ligated into the N-terminal and C-terminal regions, respectively. Codon and amino acid sequence are described in supplementary figures (S1 to S3 Figs). The resulting constructs (35S:BiP ss-plant codon-optimized light chain Obinutuzumab, 35S:BiP ssplant codon-optimized heavy chain Obinutuzumab, 35S:BiP ss-plant codon-optimized heavy chain Obinutuzumab:HDEL, 35S:BiP ss-original light chain Obinutuzumab, 35S:BiP ss-original heavy chain Obinutuzumab, and 35S:BiP ss-original heavy chain Obinutuzumab:HDEL) were transformed into Agrobacterium GV3101 competent cells using the freeze-thaw method into N. benthamiana (N. benthamiana L.) plants [27,28]. PCR followed by sequencing of products was performed to verify the constructs. To produce Obinutuzumab and rituximab in CHO cells (ATCC 1 CCL-61™, ATCC) the light and heavy chains of original Obinutuzumab and rituximab were inserted into the pLenti6.1 vector (ThermoFisher Scientific), then each clone was transfected into HEK cells to generate lentiviruses. HEK cells (ATCC 1 CRL-3216™, ATCC) grown in media containing lentiviruses with light and heavy chain inserts were mixed and used to infect CHO cells. In this way, CHO cells produced light and heavy chain-assembled Obinutuzumab and rituximab.

Agrobacterium-mediated infiltration into N. benthamiana leaves
N. benthamiana plants, aged 4-5 weeks, were used for infiltration. Each light and heavy chain construct was individually transformed into the GV3101 strain of Agrobacterium via the freeze-thaw method. Transformed Agrobacteria were incubated in YEP medium containing 50 mg/ml kanamycin and 50 mg/ml rifampicin at 28˚C for 2 days. Agrobacteria were resuspended in infiltration solution (10 mM MES, pH 5.7, 10 mM MgCl 2 , and 500 μM acetone) and infiltrated into the abaxial side of leaves using a syringe as described previously [29]. Infiltrated leaves were harvested after 3-4 days for all experiments.
buffer containing 50 mM Tris-HCl (pH 7.2), 150 mM NaCl, and protease inhibitor cocktail (Sigma-Aldrich) followed by incubation for 15 min at 4˚C [30]. The protein suspensions were centrifuged twice at 13,200 g for 20 min at 4˚C and the supernatants were collected. Total protein was quantified using Bradford reagent.

plant-obinutuzumab purification from crude N. benthamiana leaf protein extracts
Agro-infiltrated N. benthamiana leaves were harvested and Obinutuzumab was purified as follows. Total soluble proteins were extracted with two volumes of plant protein extraction buffer (50 mM Tris-HCl pH 7.2, 150 mM NaCl), and a protease inhibitor cocktail (Sigma-Aldrich); the extract was centrifuged at 16,000 g for 20 min. The supernatant was filtered through two layers of Miracloth to remove insoluble material then re-centrifuged at 16,000 g for an additional 30 min. The clarified extract was filtered through 0.2 μm pore filters (Sartorius Stedim Biotech GMBH, Gottingen, Germany) then loaded onto a protein A affinity chromatography column (Pierce, GE Healthcare Life Sciences, Baie d'Urfe, Quebec, Canada). The column was washed with extraction buffer, and antibodies were eluted using 100 mM glycine pH 3.0. Elution fractions were immediately neutralised with 2 M Tris-HCl (pH 7.4) and analysed using Ponceau S staining and western blot to determine antibody quantity and purity. Antibodies were stored at -80˚C until used.

Confocal fluorescence microscopy analysis
Wild-type and Obinutuzumab-infiltrated N. benthamiana leaves were fixed for 3 days with 3.7% paraformaldehyde in PBS (pH 7.2). Some samples were co-infiltrated with a p35S:BiP ss-RFP (red fluorescence protein)-HDEL construct to indicate ER localisation. The fixed leaves were paraffin-embedded, sectioned at 10 μm thickness, and adhered to gelatine-coated slides. The anti-human Ig Fc-specific FITC-conjugated antibody (109-095-008, Jackson Laboratories, 1:2000 dilution in blocking solution: 5% horse serum, 5% goat serum, 1% gelatine in PBS) was applied to the samples at room temperature for 30 minutes, and the slides were washed three times with PBS. Nuclei were stained with DraQ (ab108410, Abcam). FITC, RFP, and DraQ fluorescence was evaluated using confocal microscopy (Zeiss 780, Jena, Germany).

Binding affinity test using flow cytometry analysis
To evaluate binding specificity, mCherry-CD20-expressing HEK cells were stained with the various purified antibodies. Anti-human Ig Fc-specific FITC-conjugated antibody was used as a secondary antibody and DAPI was used for nuclear staining. FITC, mCherry, and DAPI fluorescence were imaged by confocal microscopy (Zeiss 780). To compare the binding affinities of the antibodies, flow cytometry (FACS, BD Biosciences, Franklin Lakes, NJ, USA) was used. Ramos cells (5 x 10 5 ) were stained with plant-obinutuzumab-HDEL or CHO-obinutuzumab at 1, 10, 100, 1000, and 10,000 nM for 15 minutes followed by anti-human Ig Fc-specific FITC-conjugated secondary antibody (109-095-008, 1:200 dilution; Jackson Laboratories) for 30 minutes at 4˚C. Cells were twice washed with culture medium then analysed. An antihuman Ig Fc-specific FITC-conjugated antibody only-treated sample was used for the negative control. The binding was measured as the mean fluorescence intensity of each sample.

Ramos cell depletion via direct antibody binding
Ramos cells were resuspended in RPMI media and plated on a round-bottom 96-well plate (at a density of 5 x 10 5 cells/well). The respective antibody dilution was added and incubated for 24 hours at 37˚C and 5% CO 2 . To stain live cells, calcein AM (Invitrogen, C34852) was freshly dissolved in DMSO (final concentration 10 nM) and added to each well. The retention of calcein in live cells was used as a readout by flow cytometry (FACS).

Statistical analyses
Data are presented as the means ± standard error of the mean. Statistical analysis was performed by Student's t-test followed by Tukey's multiple comparisons using the GraphPad Prism software package (version 5.0), as appropriate. P<0.05 was considered statistically significant.

Obinutuzumab binary vector construction for plant-based monoclonal antibody expression
To express Obinutuzumab in N. benthamiana leaves, we obtained protein sequences from a drug bank (https://www.drugbank.ca) and patent information (US20050123546). We synthesised the nucleotide sequences of the full-length light and heavy chains as original mammalian codon and plant codon optimised sequences (S2 and S3 Figs, S1 Fig for amino acid sequence with signal sequences). In order for the heavy and light chains to assemble in the ER, the localisation signal sequence from the ER protein BiP was attached to the N-terminal region of the light and heavy chains. Meanwhile, the ER retention signal sequence HDEL was attached to each chain to permit accumulation in the ER (Fig 1A). To confirm the incorporation of each construct into the N. benthamiana genome via agrobacterium-mediated infiltration, the vector inserts were verified by PCR. The genomic DNA of each construct infiltrated into N. benthamiana leaves was used as template with the primer pairs indicated in Fig 1A (arrows, S4  Fig for sequences). The proper sizes of PCR products confirmed that the heavy and light chain constructs were stably incorporated into the N. benthamiana genome via agrobacterium-mediated infiltration (Fig 1B).

Comparison of the integrity, expression level, and cellular localisation of plant-obinutuzumab-HDEL and plant-obinutuzumab in N. benthamiana leaves
Total protein extracts were separated by polyacrylamide gel electrophoresis (PAGE) under reducing and non-reducing conditions followed by immunoblot using an HRP-conjugated anti-human Ig Fc specific antibody, and an HRP-conjugated mouse Ig-specific antibody.
To verify the expression of each construct, the amount of total protein in each extract (8 μg) was compared to 50 ng and 100 ng purified CHO cell-produced rituximab (CHO-RTX), as control (Fig 2A). After comparing the protein expression level of original and plant codonoptimized constructs (S5 Fig), we used the original sequence heavy chain and light chain constructs for further experiments, because those showed slightly increased expression level than plant codon optimized constructs. In case of the no-HDEL heavy chain, a~35 KD truncated protein was detected, which implied that the heavy chain secreted into apoplasts was cleaved by an unknown proteinase, as previously reported [25,31]. On the other hand, the heavy chains with HDEL-tags did not show any truncated forms. These data suggested that the light and heavy chain-HDEL-infiltrated N. benthamiana leaves successfully expressed plant-obinutuzumab-HDEL with similar biochemical protein properties to Rituximab. To confirm the localization of each antibody in N. benthamiana leaves, plant-obinutuzumab-HDEL and plant-obinutuzumab were evaluated using immunohistochemistry with FITCconjugated human Ig-specific antibodies. To indicate the location of the ER, RFP (red fluorescent protein)-tagged BIP was co-infiltrated with plant-obinutuzumab-HDEL or plantobinutuzumab. Fig 2C shows that most plant-obinutuzumab-HDEL and plant-obinutuzumab were found in the ER, but plant-obinutuzumab-HDEL showed more protein expression in the ER than plant-obinutuzumab.

The yield of plant-obinutuzumab-HDEL produced in N. benthamiana leaves
To measure yields, various amounts of CHO-RTX were loaded with 20 μg total soluble protein from N. benthamiana leaves expressing HDEL-tagged and untagged antibodies, and immunoblot analysis using HRP-conjugated human Fc-specific antibodies was performed. According to the standard curve derived from rituximab, the production yields of plant-obinutuzumab-HDEL comprised approximately 140 ng out of 20 μg of total soluble proteins, which corresponded to~0.7% of total proteins (Fig 3). In the case of plant-obinutuzumab, the heavy chain appeared to have been cleaved in the Fab region because the truncated heavy chain was copurified, which implied the Fc portion remained intact to bind protein A. We did not calculate the production yield of plant-obinutuzumab and we used plant-obinutuzumab-HDEL for further experiments.

Specific epitope recognition of plant-obinutuzumab-HDEL
To confirm the specific epitope recognition of plant-obinutuzumab-HDEL, immunocytochemistry analysis was performed with CHO-obinutuzumab as a control. The mCherry-CD20 plasmid was transfected into HEK cells and immunocytochemistry was performed with plantobinutuzumab-HDEL and FITC-conjugated secondary human Fc-specific antibodies. plantobinutuzumab-HDEL only bound to mCherry CD20-expressing HEK cells, similar to CHOobinutuzumab. In addition, flow cytometry analysis was performed to compare their binding affinities (Fig 4C shows representative images). Incubation of Ramos cells (CD20-positive B cell line) with 10 μg/ml plant-obinutuzumab-HDEL or CHO-obinutuzumab resulted in similar binding affinities. A dose-dependent increase in affinity is shown and summarised in Fig  4D. CHO-obinutuzumab and plant-obinutuzumab-HDEL showed similar binding affinities, which were about half of the affinity of rituximab-CHO; IgG showed no binding at any dose. These data clearly indicated that plant-obinutuzumab specifically bound to CD20 and the binding affinity was equivalent with to CHO-obinutuzumab. Obinutuzumab produced in plant leaf

plant-obinutuzumab and CHO-obinutuzumab binding induced similar CD20 and B cell changes
The binding of rituximab to CD20 induces polarisation of lipid rafts and tight clustering of caveolins; however, Obinutuzumab binding to CD20 does not induce polarisation of lipid rafts [3,14]. To determine whether plant-obinutuzumab-HDEL had a similar effect, co-localisation of CD20 and caveolin was evaluated (Fig 5). Ramos cells were incubated with 10 μg/ml plantobinutuzumab-HDEL or CHO-obinutuzumab for 15 minutes, fixed with a 4% paraformaldehyde PBS solution, and immunostained with a FITC-conjugated human Fc-specific secondary antibody. As shown in Fig 5, after treatment with plant-obinutuzumab-HDEL or CHO-obinutuzumab, CD20 was distributed in a punctate manner on the cell surface, while CD20 was distributed in clusters on the surface of Ramos cells incubated with rituximab-CHO (Fig 5A). In addition, caveolin aggregation also was found with CD20 on rituximab-CHO treated cells, but not CHO-obinutuzumab and plant-obinutuzumab-treated cells (Fig 5B). Obinutuzumab causes homotypic adhesion (HA) of cells and DBCC-mediated B cell death via binding to CD20 [3,13,14]. As shown in Fig 5C, plant-obinutuzumab and CHO-obinutuzumab resulted in HA that was not observed with rituximab-CHO. Most interestingly, the rate of cell death induced by direct binding of plant-obinutuzumab and CHO-obinutuzumab was very similar; cell death was not observed in rituximab-CHO-treated or IgG control-treated cells (Fig 5D). These data clearly demonstrated that plant-obinutuzumab-HDEL and CHOobinutuzumab had similar effects on B cells via binding to CD20. The similarities in CD20 binding activities (Fig 4) and the effect on Ramos cells (Fig 5) demonstrate that plant-

Discussion
Obinutuzumab is a monoclonal antibody used to deplete CD20-expressing lymphoma cells and B cells via ADCC and direct binding [3,[13][14][15]. Of the 16 different anti-CD20 antibodies clinically available at present [32], Obinutuzumab is recognized as superior based on low minimal residual disease (MRD) and increased progression-free survival [2]. Therefore, the demand for Obinutuzumab is expected to be high. The anticipated demand for Obinutuzumab will be a significant opportunity to demonstrate the superiority of plant antibody production platforms. Our binding affinity test showed that plant-obinutuzumab-HDEL and CHO-obinutuzumab showed similar binding to CD20 on Ramos cells, and furthermore, both antibodies showed similar extend of the B cell depleting activity via physical binding to CD20 that are typical characteristics of type II anti-CD20 antibodies [32]. Accordingly, we suggest that plant-obinutuzumab-HDEL can be used to kill CD20-expressing lymphoma cells just by direct binding without complements or effector cells, as similar to CHO-obinutuzumab.
In our study, plant-obinutuzumab-HDEL was expressed at a ratio of 0.7% of the total soluble protein, and was purified at a ratio of~55%. Previous studies have shown that the yield of IgG in Nicotiana benthamiana ranges from 105.1 μg/g (IgG for HIV virus [33]) to 500 μg/g (IgG for Ebola virus GP1) [34] depending on the expressed protein. Compared to previous reports, the average yield of plant-obinutuzumab-HDEL, 14.8 μg/g, was relatively low. To increase yields, a special high yield multi-protein expression system is required, such as the BeYDV (bean yellow dwarf virus)-derived DNA replicon system [34].
Our data agreed with previous reports that the HDEL tag in IgG promotes accumulation of the monoclonal antibodies in the ER (Fig 3), resulting in high expression [21,23,35], but also protects the antibodies from possible cleavage within apoplasts. The HDEL tag resulted in high mannose-type glycosylation, which accounted for more than 80% of total N-glycans; Man 7-GlcNAc 2 , the same type of glycosylated residue found in humans [23], was most abundant. A small amount of xylose residue (~5%), acquired during transport from non-ER compartments, was also present in HDEL-tagged IgG [23]. However, small amounts of xylose have low probability to cause any immunogenic responses in humans. In support, Elelyso, a plant-produced recombinant glucocerebrosidase, has not been reported to cause any dangerous events [19]. On the other hand, plant-obinutuzumab not tagged with HDEL showed truncation of the protein in its N-terminal region, which resulted in unusable Fab-deleted antibodies (Fig 3A). There are reports that the IgG secreted within apoplasts undergoes cleavage through specific aspartic, cysteine, and serine peptidases, and the acidic pH in the extracellular space facilitates these proteolytic activities in N. benthamiana [25,31]. If the pH of the total protein extract during purification is maintained at a high enough level, it is possible to solve the antibody cleavage issue.
The immunogenicity and hypersensitivity caused by plant-specific glycosylation are still highly debated [19], but no evaluation of bio-drugs can be too strict for human safety. In addition, plant-specific glycosylation is considered the main obstacle for rapid industrialisation of plant-produced bio-drugs. Until now, there have been no reports of human studies that show plant-derived proteins are more immunogenic than mammalian-derived bio-therapeutics [19]. caused by CHO-obinutuzumab and plant-obinutuzumab-HDEL were compared to IgG and CHO-rituximab. Each antibody (at 1 μg/ml, 10 μg/ml, and 30 μg/ml) was incubated with Ramos cells for 14 hours and cell death was measured by lose of calcein-AM dye via FACS analysis. Three independent experiments are shown as means ± s.e.m. ÃÃ P < 0.01; ÃÃÃ P < 0.001. https://doi.org/10.1371/journal.pone.0191075.g005 Nevertheless, careful observation of the progress toward the use of plant-derived proteins is needed for more efficient and rational development of plant-produced bio-drugs. Li et al. reported that TALEN-mediated knock-out of xylosyl-transferase and fucosyl-transferase was unsuccessful in eliminating all xylose and fucose residues in Obinutuzumab due to compensatory activity likely from other isotypes of xylose and fucose transferases [20]. While Cas-9-mediated genome editing shows the most promise for eliminating these enzymes, it requires a significant amount of time to perform. At present, the use of RNAi in transgenic plants to inhibit expression [11] of xylosyl-transferase and fucosyl-transferase is the best way to achieve prevent plant-specific glycosylation. Considering the growth of plants, transgenic plants with all genes deleted by Cas-9 may not be a better production platform than transgenic plants using RNAi. However, a more economical and mammalian protein risk-free production platform is in high demand for the entire bio-drug industry. To this end, a plant-based protein production platform is the most promising approach that ultimately will contribute to worldwide welfare.
There have not been many successful examples of producing antibodies in plants so far. Considering its value, E. coli is the cheapest protein production host. However, plants are considered to be a better production platform for bio-drugs because of the presence of clean and fresh images of plants as well as post-translational modifications such as glycosylation. Although there are many similarities between the post-transcriptional processes of animals and plants studied so far, differences between the post-transcriptional processes as well as plant-specific O-glycosylation are not completely understood, which creates other possible bottlenecks to production and medical use of plant proteins. Furthermore, there are few reports of equivalence between plant-produced and mammalian proteins, so additional studies are necessary to demonstrate equivalence in order to truly industrialize the use of plants as a medical protein production platform. Our results clearly demonstrated that the Fab portion of plant-obinutuzumab-HDEL had very similar functional characteristics to CHO-obinutuzumab ( Fig 5); furthermore, plant-obinutuzumab-HDEL was equivalent to CHO-obinutuzumab with respect to cell death of B cells induced by Fab binding to CD20. Plant-specific glycosylation-free plants can be used to produce Obinutuzumab, which may be an improved bio-better production platform for monoclonal antibodies with strong ADCC capability that are similar to Obinutuzumab produced in glyco-engineered CHO cells.
In conclusion, we demonstrated that Obinutuzumab is a promising candidate as a plantproduced monoclonal antibody by showing that the Fab region of plant-obinutuzumab-HDEL has equivalent ability to bind CD20 and causes direct binding mediated B cell death compared to CHO-obinutuzumab.