The amino acid and nucleotide sequence information were retrieved from the National Center for Biotechnology Information (NCBI), and Uniprot databases. The codons were first optimized to improve the expression in tobacco and lettuce. Since the codon usage by the specific amino acid in the human and within the different plant also vary, the codons were optimized based on the codon usage table of the N. benthamiana and L. sativa available at Kazusa DNA Research Institute (http://www.kazusa.or.jp). Moreover, the codon usage of highly expressed protein in plants, Rubisco and ubiquitin were analyzed in tobacco and lettuce using CAI calculator (http://genomes.urv.es/CAIcal/) and the codon of HA were adjusted. Importantly, the distribution of codon in the protein was maintained and the codon with zero usage frequency in the Rubisco and ubiquitin was avoided or used at a minimum. The optimized sequences of the full length of HA including the native signal peptide were synthesized (Macrogen, Seoul, Korea; http://www.macrogen.com). The synthesized HA DNA sequences were polymerase chain reaction (PCR) amplified by using the primer set HA3_SP_XhoI_F 5′-ATATCTCGAGATGAAGACTGTGATCGCCCTTTCTTATATTTTC-3′ and HA3_XbaI_R 5′-TTGTTCTAGAAATGCAAATGTTACATCTGATGTTGCCTC-3′ to generate HA with signal peptide and HA3_XhoI_F 5′-ATATCTCGAGATGCAGAACCTTCCAGGAAGTGAGAATAACGC-3′ and HA3_KDEL_XbaI_R 5′-TTGGTCTAGATCACAATTCATCCTTAATGCAAATGTTACATCTGATGTTGC-3′ to generated HA with KDEL at C terminus.

The amplified PCR product was digested with restriction enzymes XhoI and XbaI and ligated using the NEB T4 ligase (http://www.neb.com) into XhoI and XbaI site in geminiviral vector pBYR11ek2Md. For green fluorescent protein (GFP) control vector pBYR2ek2Md was used. The correct insertion of HA DNA was confirmed by colony PCR, restriction enzyme digestion, and Sanger sequencing.

Agrobacterium tumefaciens harboring the respective vector was cultured in yeast extract peptone medium containing kanamycin (50 mg/L), rifampicin (50 mg/L), and gentamicin (50 mg/mL) at 28℃ for overnight with 200 rpm to get the cell concentration around 1.6–1.8 optical density (O.D.). Bacterial cell density was determined by measuring O.D. at 600 nm. The agrobacterium cells were then collected by centrifugation at 5,000 rpm for 7 minutes at 4℃. Then, the pellet was re-suspended in the infiltration buffer (10 mM MES-KOH [pH 5.7], 10 mM MgSO₄·7H₂O) to give the O.D.₆₀₀ of 0.2. N. benthamiana of 6- to 7-week-old, grocery-purchased red lettuce, and hydroponically-grown green lettuce were used for agroinfiltration. Agroinfiltration was done either by 1 mL syringe without needle for an individual leaf, or vacuum for the whole plant completely submerging them within a desiccator and applying vacuum for 2 minutes for tobacco and 5 minutes for lettuce at 70 mbar to allow the infiltration of agrobacteria into the apoplast spaces of submerged plant tissue. The leaf surface was washed after infiltration, then grown those infiltrated plants in 25℃, 50% humidity environment with a 16:00/8:00-hour day/night cycle.

The leaves were harvested 4-day-after infiltration for protein detection and quickly frozen into liquid nitrogen and stored in −80℃ until further analysis. Hand-held ultraviolet lamp at 365 nm was used for visual detection of GFP in the plant (Fig. 1A, B). To obtain the expression levels of rHA, sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) and western blot were carried out. For total protein extraction, agroinfiltrated leaves of 50 mg (fresh weight) were ground to the fine powder in liquid nitrogen and then mixed with 150–200 µL Laemmli loading buffer (50 mM Tris-HCl [pH 6.8], 2% sodium dodecyl sulfate, 5% beta-mercaptoethanol, 10% glycerol, 0.2 mg/mL bromophenol blue, 1× protease inhibitor cocktail [GenDEPOT, Katy, TX, USA]) followed by 10-minute incubation in boiling water and centrifuged for 10 minutes. Alternatively, 200–300 µL of extraction buffer (50 mM Tris pH 8.0, 150 mM NaCl, 2 mM ethylenediaminetetraacetic acid, 0.1% Triton X-100, 1× protease inhibitor cocktail [GenDEPOT]) was added to 100 mg and centrifuged at 13,000 rpm at 4℃ for 30 minutes. The supernatant was transferred to a new tube and protein concentration was determined by the Bradford assay. Total protein of 20 µg was resolved in the 10% Tris-glycine SDS-PAGE gel. The gel was then either stained with Coomassie brilliant blue G-250 or electrotransferred to polyvinylidene difluoride membrane (Merck, Darmstadt, Germany) using FLASBlot transfer buffer (https://advansta.com). The membrane was probed with a rabbit polyclonal influenza A H3N2-HA antibody (PA5-34930, Thermo Fisher, Waltham, MA, USA) in a dilution factor of 1:3,000 followed by the secondary antibody-alkaline phosphatase goat anti-rabbit IgG (AP-100-1, Vector Laboratories, Burlingame, CA, USA) in a dilution of 1:4,000 in phosphate buffered saline. Protein in the membrane was detected by using the ECL detection reagent. For the yield quantification, 100 ng and 125 ng commercially available H3N2-HA (recombinant partial length, Ab124562, Abcam, Cambridge, MA, USA) were used as a positive control. The plant made rHA yield was calculated by analyzing the protein band intensity in the membrane using the Image J software (National Institutes of Health, Bethesda, MD, USA) and compared with that of commercially available recombinant.

The efficient production of a particular recombinant protein in a plant is a challenging because there are many combinatorial parameters governing the successful expression of the proteins. In order to find a right expression vector and the appropriate host plant that express the HA (H3N2) efficiently, we rationally chose DNA viral vector system and tested transient expression in the two host plants. In addition, to prevent the species-specific bias in the codon usage, synonymous mutations were introduced for optimal expression. So prior to cloning, we optimized the viral specific codon of HA from influenza A/canine/Korea/S3001/2015 (H3N2) towards the usage of preferred codon in tobacco and lettuce avoiding the use of rare codons that could negatively impact on HA expression.

To test whether the codon-optimized sequences of HA (H3N2) can be translated into recombinant protein in plant system, we expressed the HA gene in N. benthamiana, a well-established host plant for the production of the plant-based pharmaceutics. To this end, we cloned the full length of HA into the geminiviral vector expression system. The geminiviral vector expression system, pBYR11ek2Md is based on bean yellow dwarf virus (BeYDV), a DNA virus replicon system, to replicate huge amount of DNA template which produces the higher level of heterologous protein after transcription and translation in plant cell [192021]. The Agrobacterium tumefaciens harboring HA-carrying geminiviral vector was infiltrated into the tobacco leaves with the needless syringe or vacuum pump and the expression pattern was evaluated at the protein level. GFP-containing geminiviral vector, pBYR2eK2Md was also infiltrated as an expression control for the agroinfiltration procedure and monitor plant growth after infiltration (Fig. 1A). The extracted protein from the infiltrated leaves was resolved in SDS-PAGE gel and HA protein was detected by western blot analysis using rabbit polyclonal anti-influenza H3N2 HA antibody. The rHA was successfully expressed in tobacco leaves as revealed by the presence of strong protein band in the western blot (Fig. 1B).

We then expressed rHA in an edible host plant lettuce which is amenable for the development of plant-produced oral vaccines. Lettuce offers an economical, quick and cold-chain free oral vaccine production platform. Although, transgenic expression system has been used to express the vaccine candidate protein against the human infectious viral disease in the lettuce, there is only a few report on transient expression of influenza vaccine candidate in lettuce [22232425]. Geminiviral vector system has also been applied to express the antibodies and antigens in transgenic system [202126].

We intended to express the rHA in lettuce because it can be further developed into a quick and inexpensive vaccine production system for pandemic and seasonal CIV. The production of vaccine antigen in lettuce could be especially more accountable at the time of emergency if we also could use the grocery lettuce for the transient production of a vaccine protein. Hence, we first tested whether geminiviral expression systems do express a recombinant protein in the lettuce bought from grocery stores. We vacuum-infiltrated the agrobacterium harboring the GFP in geminiviral vector (pBYR2eK2Md) into red lettuce head purchased from grocery store and green lettuce grown in the lab hydroponics system. The geminiviral vector system successfully worked for the GFP in the red lettuce head and green lettuce (Fig. 1C).

We further cloned the full length of the HA with the native signal peptide in pBYR11ek2Md vector and transiently expressed them in grocery-purchased lettuce (Fig. 1B). The expression level of rHA was determined 4-day-after infiltration by the western blot and probed with H3N2 HA-specific antibody. The rHA was successfully expressed in lettuce which was shown by the protein band in lettuce as similar to tobacco protein in the western blot; however, the higher level of rHA protein was produced in tobacco (Fig. 1B). We quantified the yield of rHA protein to be 9.3-13 mg/kg fresh weight of tobacco leaves (Fig. 1D).

Several studies suggest the importance of considering the glycosylation aspect for influenza vaccine design. During 2016–2017 influenza season, human H3N2 vaccine used at that time was not as effective as expected because of the loss of the HA glycosylation site at antigenic site B in the egg-produced vaccine, which emphasizes the need of considering protein glycosylation during vaccine design [2728].

In view of the possible impact of different glycosylation patterns of the rHA in its surface affinity, antigenicity, and efficacy of the vaccine, we were interested in producing CIV rHA with different glycosylation patterns. We fused the full length of rHA with the ER retention signal, KDEL to prevent the entry of rHA into the Golgi apparatus where the complex glycosylation occurs. To confirm and to compare the production of rHA without complex glycan by KDEL, we expressed the full length of rHA with native signal peptide as well (N-SP) (Fig. 2A). Geminiviral vector harboring rHA with native signal peptide or KDEL signal (KDEL) were infiltrated to both tobacco and lettuce leaves and the expressed proteins were analyzed in western blot. rHA protein was produced in the presence of either native signal peptide or KDEL signal in both tobacco and lettuce (Fig. 2B, C). However, the size of the rHA protein expressed with KDEL is smaller than that produced with the native signal peptide (Fig. 2B, C). KDEL signal retains rHA in the ER avoiding the further addition of the carbohydrate on rHA. In this case, the rHA protein size is proportional to the basal glycans being added in the ER. On the other hand, the native signal peptide with HA produced rHA protein of size around 75 kDa which is approximately 11 kDa bigger than the calculated size (64 kDa) of HA most possibly because of extra molecular weight of complex glycans added to HA by glycosylation in Golgi (Fig. 2B, C). We checked for the glycosylation sites in HA by using online glycosylation prediction site www.dtu.dk and found 6 N-glycosylation sites in asparagine residues (Fig. 2D) [29]. The similar nature of HA was reported for H5N1 where the observed protein size was bigger than the calculated when expressed full sequence in cytosol and apoplast [30]. It is interesting to observe the higher level of basal-glycosylated rHA production in lettuce than in tobacco, whereas higher level of complex-glycosylated rHA in tobacco (Figs. 1B, 2B, C).