High Level Production of Monoclonal Antibodies Using an Optimized Plant Expression System

Abstract

Biopharmaceuticals are a large and fast-growing sector of the total pharmaceutical market with antibody-based therapeutics accounting for over 100 billion USD in sales yearly. Mammalian cells are traditionally used for monoclonal antibody production, however plant-based expression systems have significant advantages. In this work, we showcase recent advances made in plant transient expression systems using optimized geminiviral vectors that can efficiently produce heteromultimeric proteins. Two, three, or four fluorescent proteins were coexpressed simultaneously, reaching high yields of 3-5 g/kg leaf fresh weight or ~50% total soluble protein. As a proof-of-concept for this system, various antibodies were produced using the optimized vectors with special focus given to the creation and production of a chimeric broadly neutralizing anti-flavivirus antibody. The variable regions of this murine antibody, 2A10G6, were codon optimized and fused to a human IgG1. Analysis of the chimeric antibody showed that it was efficiently expressed in plants at 1.5 g of antibody/kilogram of leaf tissue, can be purified to near homogeneity by a simple one-step purification process, retains its ability to recognize the Zika virus envelope protein, and potently neutralizes Zika virus. Two other monoclonal antibodies were produced at similar levels (1.2-1.4 g/kg). This technology will be a versatile tool for the production of a wide spectrum of pharmaceutical multi-protein complexes in a fast, powerful, and cost-effective way.

Keywords: Zika virus; glycosylation; heteromultimeric proteins; monoclonal antibodies; pharming; plant-based biopharmaceuticals; transient expression.

Figure 1

A generalized schematic of the plant expression vectors used in this study. pBYKEAM is a replicating plant expression vector based on BeYDV with optimized 5′ and 3′ UTRs and is the vector of choice for high expression of a single target gene (Diamos et al., ; Diamos and Mason, ; Rosenthal et al., 2018). pBYKEAM2 allows simultaneous co-expression of two genes from a single replicon containing optimized 5′ and 3′ UTRs. pBYKEMd2 is a previous iteration of pBYKEAM2 that does not contain double terminators. The remaining vectors contain multiple expression cassettes arranged as either a single large replicon or multiple replicons (when the gene cassettes are separated by SIR/LIR) with unoptimized 5′ and 3′ UTRs. For expression of antibodies, the barley alpha amylase signal sequence for ER-targeting is also present at the start of the gene. LIR, the long intergenic region from BeYDV; 35S, the 35S promoter with duplicated enhancer region from cauliflower mosaic virus; PsaK2T 5′ , the truncated 5′ UTR from the N. benthamiana psaK gene; Ext 3′ , the tobacco extensin terminator with intron removed; NbACT 3′ , the 3′ UTR from the N. benthamiana ACT3 gene; Rb7 MAR, the tobacco Rb7 matrix attachment region; Rb7 MARd, a truncation of the Rb7 MAR to remove unwanted restriction enzyme sites; SIR, the short intergenic region from BeYDV; Rep/RepA, the replication proteins from BeYDV; TMV 5′ , the 5′ UTR from tobacco mosaic virus; RbcS 3′ , the 3′ UTR from the pea rbcS gene; VspB 3′ , the 3′ UTR from the soybean vspB gene; AMV 5′ , the 5′ UTR from alfalfa mosaic virus; TEV 5′ , the 5′ UTR from tobacco etch virus; YFG, insertion site for the gene of interest; GFP, green fluorescent protein; DsR, DsRed fluorescent protein; YFP, yellow fluorescent protein; CFP, cyan fluorescent protein; 6D8H, the 6D8 heavy chain; 6D8L, the 6D8 light chain.

Figure 2

Simultaneous co-expression of up to four proteins with comparisons of optimized BeYDV vectors to unoptimized BeYDV vectors. Leaves of N. benthamiana were agroinfiltrated with either optimized (pBYKEAM or pBYKEAM2) vectors or unoptimized vectors (pBY-GR, pBY-GCR, referred to as “old”) for GFP, DsRed, CFP, and YFP in various combinations and (A) imaged at 5 dpi under UV illumination; or (B) separated by non-reducing or reducing SDS-PAGE and either viewed under UV transillumination or stained with Coomassie gel. Rubisco large subunit (RbcL) and the monomeric-sized band of all fluorescent proteins are indicated. (C) The relative total expression of each combination of constructs was analyzed using ImageJ software to quantify the band intensity of SDS-PAGE reducing gels. The mean band intensity is given ± standard error from 3 independently infiltrated leaf samples, where the expression of pBYKEAM-GFP was arbitrarily defined as 1. The bar colors are estimates of the relative production of each construct by non-reducing SDS-PAGE followed by ImageJ analysis using 3 independently infiltrated leaf samples. Statistical significance was calculated using student's t-test.

Figure 3

The role of replicon size and configuration in expression and replication of BeYDV vectors. (A) Leaf DNA was extracted at 3 DPI from uninfiltrated (Wt), or samples infiltrated with the indicated vectors, then digested with indicated restriction enzymes (K, KpnI; S, SacI) and run on a 1% agarose gel. G(SL)R refers to pBY-G(SL)R; GR refers to pBY-GR. Replicon positions are indicated with arrow head. (B) Fluorimetric analysis of GFP and DsRed accumulation showing efficient co-expression of two fluorescent proteins from either single [pBY-GR] or dual [pBY-G(SL)R] replicon vectors. Dilutions of total soluble protein extracts were subjected to spectrofluorimetry using excitation and emission wavelengths of 485 and 538 nm for GFP, and 544 and 590 nm for DsRed. G(SL)R, pBY-G(SL)R; GR, pBY-G(SL)R. Data are means ± S.D. from three independently infiltrated samples. (C) DNA from leaves of uninfiltrated (Wt), or infiltrated with indicated vectors were separated on 0.8% agarose gels before and after restriction enzyme digestion. GFP+CFP+DsRed indicates co-infiltrated sample with Agrobacterium mixture of pBY-GFP, pBY-CFP, and pBY-DsRed. Restriction enzyme XhoI was used for pBY-GR and GFP+CFP+DsRed. For pBY-GCR, restriction enzyme SalI was used. Expected replicon positions are indicated with arrow heads.

Figure 4

Simultaneous co-expression of three fluorescent proteins. N. benthamiana leaves were infiltrated with Agrobacterium strains harboring expression vectors as indicated (on the left). At 2 DPI the infiltrated leaf samples were examined with confocal laser scanning microscope. For co-infiltration, mixture of Agrobacterium strains harboring pBY-GFP, pBY-CFP, and pBY-DsRed were used. Excitation lasers of 488, 458, and 543 nm and detection windows of 550–560, 470–500, and 614–646 nm were employed to detect GFP, CFP, and DsRed signals, respectively. For plant autofluorescence chlorophyll detection, the excitation laser of 633 nm with detection window of 630–700 nm was used.

Figure 5

IgG production of three mAbs using optimized plant-expression vectors. Leaves of N. benthamiana agroinfiltrated with unoptimized (6D8 old) or optimized (6D8, HSV8, c2A10G6) BeYDV vectors were harvested at 5 DPI and protein extracts were analyzed for IgG production by ELISA using human IgG as a reference standard. Columns represent results from three independently infiltrated leaf samples ± standard error.

Figure 6

Western blot analysis of plant-derived 6D8. Protein samples were separated on a 4-20% SDS-PAGE gradient gel under denaturing and reducing condition (A,B) or under non-reducing condition (C) and blotted onto a PVDF membrane. The membrane was incubated with a goat anti-human gamma chain antibody or goat anti-human kappa chain antibody to detect heavy chain (A) or light chain (B,C). Wt: Protein samples extracted from uninfiltrated leaves; lanes marked pBYR-H(SL)L, protein samples extracted from the leaves infiltrated with dual replicon construct pBYR-H(SL)L; lanes marked pBYR-HL, protein extracted from the single replicon construct pBYR-HL. (D) Commassie blue stained gel is shown for normalized total protein loading.

Figure 7

Characterization of c2A10G6. (A) A N. benthamina leaf at 4 DPI was examined for signs of chlorosis or necrosis. Faint chlorosis was visible, but there was no visible necrosis. (B) To test whether the clarified leaf extracts of the antibody constructs were stable upon acid-precipitation, 1N phosphoric acid was added to a final acid volume of 4% of the total soluble extract. Following a 6-min incubation, the samples were neutralized with 1M Tris base. For comparison, a sample of the leaf extract pre-acid precipitation was also included along with a control uninfiltrated leaf extract that was not treated with acid. All three samples were mixed with non-reducing sample buffer and loaded on a 4–15% polyacrylamide gel for analysis by Coomassie-staining. (C) The same samples described in part A were run on a 4–15% polyacrylamide gel for analysis by Western blot. The Western blot was detected with HRP-conjugated goat anti-human IgG (kappa only). (D) After protein G affinity purification, samples of the purified antibody were run on SDS-PAGE gels under non-reducing or reducing conditions as noted. The left panel shows the results following a silver stain. Only the c2A10G6 band is visible. The two panels on the right show the results of the purified antibody run under non-reducing and reducing conditions on a stain-free gel. S, soluble fraction pre-acid precipitation; AP, samples subjected to acid-precipitation; U, uninfiltrated clarified leaf extract; NR, non-reducing conditions; R, reducing conditions.

Figure 8

Binding of c2A10G6 to ZIKV envelope glycoprotein. (A) Varying dilutions of clarified protein extract containing ZsE was directly bound to a polystyrene plate and probed with purified c2A10G6 to assess whether the antibody recognized the fusion loop epitope. Bound antibody was detected with goat anti-human IgG (kappa only) HRP conjugate. An uninfiltrated negative control was included to assess the level of any non-specific binding to native plant proteins. (B) The ability of c2A10G6 to recognize the fusion loop epitope was analyzed via a western blot. Clarified protein extracts containing ZprME or an uninfiltrated control were probed with purified c2A10G6 and detected with HRP-conjugated goat anti-human IgG antibody.

Figure 9

The neutralizing activity of c2A10G6 against ZIKV. Dilutions of c2A10G6 or 6D8 (negative control) were co-incubated with 100 PFU of ZIKV and Vero cells for 3 days. Plaques were counted and percent neutralization and EC₅₀ were calculated. Experiments were performed twice with technical triplicates for each sample. Bars represent the standard deviation (SD) of the mean.