Recombinant hepatitis B surface antigen (HBsAg) constitutes currently used vaccines against hepatitis B virus, and has been successfully employed as a carrier for foreign epitopes. With the aim of developing an inexpensive, easily administered vaccine source for global immunization, several groups have expressed HBsAg in plant systems. Transgenic plant-derived HBsAg assembles into virus-like particles (VLPs) and is immunogenic in both mice and humans. However, HBsAg expression is relatively low in transgenic plant systems. The time-consuming and labour-intensive process of generating transgenic plants also significantly limits high-throughput analyses of various HBsAg fusion antigens. In this paper, the high-yield rapid production of HBsAg in plant leaf using a novel viral transient expression system is described. Nicotiana benthamiana leaves infiltrated with the MagnICON viral vectors produced HBsAg at high levels, averaging 295 µg/g leaf fresh weight at 10 days post-infection, as measured by a polyclonal enzyme-linked immunosorbent assay. Transiently expressed HBsAg accumulated as the full-length product, formed disulphide-linked dimers, displayed the conformational ‘a’ antigenic determinant and assembled into VLPs. Immunization of mice with partially purified HBsAg elicited HBsAg-specific antibodies. Furthermore, it was found that transient production of HBsAg using vacuum infiltration of whole plants, rather than syringe infiltration of leaves, was readily scalable, and greatly improved the accumulation of correctly folded HBsAg that displays the protective ‘a’ determinant.
Infection with hepatitis B virus (HBV) is highly prevalent worldwide, despite the development of safe and effective vaccines. This is partly because the relatively high cost of traditional vaccination is still prohibitive for most developing countries. Injectable HBV vaccines that are currently used contain recombinant hepatitis B surface antigen (HBsAg) produced in yeast (McAleer et al., 1984) or mammalian (Diminsky et al., 1997) cells. These vaccine products require strict downstream purification and processing prior to use. HBsAg has also been successfully used as a carrier for foreign epitopes (Delpeyroux et al., 1986; von Brunn et al., 1991; Mancini et al., 1994).
With the aim of developing an inexpensive, easily administered vaccine source for global immunization, several groups have expressed HBsAg in a number of transgenic plants or plant cell systems (reviewed in Streatfield, 2005; Thanavala et al., 2006). Transgenic plant-derived HBsAg assembles into virus-like particles (VLPs) (Mason et al., 1992; Kong et al., 2001; Smith et al., 2003), similar to those in yeast recombinant HBsAg vaccines, and stimulates anti-HBsAg antibody and T-cell responses when injected into mice (Thanavala et al., 1995). Importantly, transgenic plant tissues expressing HBsAg are orally immunogenic in mice (Kong et al., 2001) and human volunteers (Kapusta et al., 1999; Thanavala et al., 2005). Some investigators have reported transgenic expression of HBsAg fusions with homologous and heterologous peptides or epitopes (Sojikul et al., 2003; Huang et al., 2005; Shchelkunov et al., 2006; Youm et al., 2007) as improved or multivalent vaccine candidates. However, in general, HBsAg expression is relatively low in transgenic plant systems (reviewed in Streatfield, 2005). The time-consuming and labour-intensive process of generating transgenic plants also significantly limits high-throughput analyses of various HBsAg fusion antigens and testing of their immunogenicity. Thus, a technology platform that allows high-level rapid HBsAg production in plants is needed.
As an alternative to stable transgenic plants, transient expression systems using plant virus-based vectors have the advantage of rapid high-level expression of the recombinant protein (reviewed in Twyman et al., 2003; Canizares et al., 2005; Yusibov et al., 2006). Plant viruses efficiently infect plant cells and replicate autonomously to very high copy numbers, resulting in high-level expression of recombinant vaccines in a relatively short time. For example, with a tobacco mosaic virus (TMV) vector, the VP1 protein of the foot and mouth disease virus was produced as an intact protein at levels of 50–150 µg/g leaf fresh weight (FW) within 10 days post-infection (dpi) (Wigdorovitz et al., 1999). Recently, a ‘deconstructed’ TMV-based vector system (MagnICON), based on the in planta assembly of functional viral vectors by the recombination of separate provector modules, has been developed for the high-level, rapid production of foreign proteins (Marillonnet et al., 2004, 2005; Gleba et al., 2005). It has been proven to be highly effective in the production of vaccine antigens, such as protective antigens against Yersinia pestis (Santi et al. 2006) and hepatitis B core VLPs (Huang et al., 2006).
In this study, the high-yield rapid production of HBsAg in Nicotiana benthamiana leaf is described using the MagnICON system. It was found that transiently expressed HBsAg was correctly assembled and was immunogenic in mice. Moreover, HBsAg production was readily scaled up by vacuum infiltration of vectors into leaves, which surprisingly resulted in greatly improved accumulation (compared with syringe infiltration) of correctly folded antigen that displays the protective ‘a’ determinant.
High-level transient expression of HBsAg in N. benthamiana leaves
The MagnICON vector system (Marillonnet et al., 2004) consists of three modules: 5′ and 3′ viral provectors, and an expression cassette for integrase, which mediates recombinational joining of the provectors. These three modules are co-delivered into N. benthamiacna leaves by agroinfiltration to express the protein of interest. A 3′ provector, pICH-HBsAg (Figure 1), was constructed for the transient expression of HBsAg. Provector pICH-GFP (Figure 1), expressing green fluorescent protein (GFP), was used as a reporter to confirm the efficiency of infiltration.
Infiltrated leaves were collected over a 14-day time course and analysed for HBsAg expression by polyclonal enzyme-linked immunosorbent assay (ELISA), which determines total HBsAg. Table 1 shows that total HBsAg levels increased to 10 dpi, reaching 294.5 µg/g FW, and then decreased slightly at 12 and 14 dpi. pICH-HBsAg-infiltrated leaves developed slight necrosis at 10 dpi, which became more severe at 12 and 14 dpi. The time course study was repeated twice and similar results were obtained. The effects of subcellular targeting were investigated using HBsAg fused to the C-terminal endoplasmic reticulum (ER) retention signal ‘SEKDEL’, and no significant difference was observed in either total HBsAg or ‘a’ determinant Auszyme monoclonal reactive antigen. Expression was also examined using the 5′ provector pICH13840 which contains an ER signal peptide, and no detectable expression was obtained (data not shown). Our efforts were therefore focused on the characterization of unmodified HBsAg expressed with the MagnICON system.
Table 1. Transient expression of hepatitis B surface antigen (HBsAg) in Nicotiana benthamiana leaves over a 14-day time course
Nicotiana benthamiana leaves infiltrated with pICH-HBsAg were collected at different time points and tested for total HBsAg by polyclonal ELISA and for ‘a’ determinant by Auszyme monoclonal assay.
dpi, days post-infection; FW, fresh weight; ND, not determined.
Data reported are means ± standard deviation from four independently infiltrated samples.
Data were obtained from pooled extracts from four independently infiltrated samples.
12.5 ± 2.5
233.7 ± 13.3
241.0 ± 30.2
294.5 ± 30.5
220.4 ± 45.5
224.8 ± 25.6
The integrity of transiently expressed HBsAg was examined by sodium dodecylsulphate-polyacrylamide gel electrophoresis (SDS-PAGE) and Western blotting using a polyclonal anti-HBsAg serum. Under reducing conditions, the yeast-derived recombinant HBsAg standard migrated at ~24 kDa (Figure 2, lane 1). pICH-HBsAg-infiltrated leaf samples produced a strong 24-kDa band (Figure 2, lane 4); by contrast, non-infiltrated or pICH-GFP-infiltrated leaf extracts yielded no signals (Figure 2, lanes 2 and 3, respectively). The Western blotting results indicated that transiently expressed HBsAg accumulated as the full-length product.
Characterization of transiently expressed HBsAg
Dimerization of HBsAg via intermolecular disulphide bonds is critical to its immunogenicity (Mishiro et al., 1980). To test whether transiently expressed HBsAg forms dimers, pICH-HBsAg leaf extract was examined by SDS-PAGE under non-reducing conditions. Western blotting showed plant-derived HBsAg predominantly as an ~48-kDa band representing the dimer form. Bands representing monomer, trimer and tetramer forms were also evident at much lower levels (Figure 2, lane 5). Thus, plant transiently expressed HBsAg forms dimers and higher order oligomers via intermolecular disulphide bonds.
Extracts from pICH-HBsAg-infiltrated leaves were assayed for the conformational ‘a’ determinant using the Auszyme monoclonal antibody ELISA. The ‘a’ determinant levels were at a maximum at 10 dpi, representing 7.5% of total HBsAg (Table 1). Although this proportion is relatively low, it nonetheless indicates the potential to display the conformational ‘a’ determinant in transiently expressed HBsAg.
It has been shown that transgenic plant-derived HBsAg assembles into VLPs (Mason et al., 1992; Kong et al., 2001; Smith et al., 2003). To determine whether the MagnICON-expressed HBsAg also forms VLPs, sucrose gradient sedimentation of pICH-HBsAg leaf extracts was performed to obtain evidence of particle behaviour. Figure 3a shows that the Auszyme reactivity profiles of sucrose gradient fractions are similar for MagnICON-expressed HBsAg and yeast-derived HBsAg VLPs (Valenzuela et al., 1982), strongly suggesting that MagnICON-expressed HBsAg assembles into VLPs. Plant-derived HBsAg was prepared for negative stain electron microscopy by an immunoaffinity procedure that depends on the binding of antigen to Auszyme monoclonal antibody specific for the ‘a’ determinant. Eluted material showed particles with a size consistent with that expected for VLPs (Figure 3b), whereas similarly treated material from a negative control leaf (vector without HBsAg) showed no such particles (data not shown).
HBsAg production can be readily scaled up with improved ‘a’ determinant accumulation by a vacuum infiltration method
Vacuum infiltration of whole N. benthamiana plants, rather than syringe infiltration of individual leaves, was used to demonstrate convenient scale-up. Similar results were obtained from three independent experiments; the results from a representative experiment with four replicate plants are shown in Table 2. The levels of total HBsAg ranged between 205 and 243 µg/g FW, similar to those in leaves infiltrated with a syringe (Table 1). Surprisingly, the accumulation of the ‘a’ determinant was greatly improved (approximately threefold) relative to that in leaves infiltrated using a syringe, with the percentage of the ‘a’ determinant relative to total HBsAg averaging ~26% (Table 2).
Table 2. Transient expression of hepatitis B surface antigen (HBsAg) in Nicotiana benthamiana leaves using vacuum infiltration
Total HBsAg (µg/g FW)
‘a’ determinant (µg/g FW)
‘a’ determinant/total HBsAg (%)
Nicotiana benthamiana leaves were infiltrated with pICH-HBsAg by a vacuum method, collected at 10 days post-infection, and assayed for total HBsAg by polyclonal enzyme-linked immunosorbent assay (ELISA) and for the ‘a’ determinant using the Auszyme monoclonal kit.
FW, fresh weight; SD, standard deviation.
Mean ± SD
230 ± 17.2
61 ± 26.8
26.1 ± 10.77
Transiently expressed HBsAg is immunogenic in mice
To determine the immunogenicity of transiently expressed HBsAg, the material was partially purified over a sucrose gradient. This material was then adsorbed to alum and injected (1 µg per dose) intraperitoneally at weeks 0, 1 and 2. The serum anti-HBsAg antibody titres increased, peaked at week 17, and then gradually declined (Figure 4). The anti-HBsAg titres were 403 mIU/mL at week 12, and peaked at 638 mIU/mL. This result is consistent with a previous report (Huang et al., 2005), which revealed that mice immunized with transgenic plant (HB117)-derived HBsAg showed an average titre of 346 mIU/mL at week 12. In order to validate the plant-derived antigen as equivalent to the licensed recombinant yeast-derived vaccine antigen, the mice were boosted at week 25 with yeast recombinant HBsAg; the antibody titres increased immediately, with a peak response of 830 mIU/mL at week 32. In another study, mice immunized with extracts from empty vector pICH11599-transfected leaves did not produce any HBsAg-specific antibody (data not shown). These data indicate that transiently expressed MagnICON-vectored HBsAg is equivalent in immunogenicity to HBsAg derived from stable transgenic plants.
Transient expression offers the unique advantage of rapidity for the analysis of expression vectors, protein accumulation and conformation. Previously, it has been shown that Agrobacterium-mediated leaf infiltration with conventional plant expression vectors produces sufficient material for conformational analysis of vaccine antigens (Huang and Mason, 2004). In this study, it was demonstrated that the yield of HBsAg from leaves using MagnICON viral vectors was remarkably high (Table 1). This technology is highly scalable, and the resulting antigen levels are sufficient for conducting preclinical and clinical trials. By contrast, the HBsAg expression in stable nuclear transgenic plants is relatively low. By vacuum infiltration, expression of the ‘a’ determinant averaged 60 µg/g leaf FW (0.26% total soluble protein), whereas transgenic plants yielded only 0.01% total soluble protein in tobacco leaf (Mason et al., 1992), 0.008% total soluble protein in potato leaf (Ehsani et al., 1997), 5.5 ng/g FW in lettuce leaf (Kapusta et al., 1999), and 300 ng/g leaf FW and 10 ng/g fruit FW in cherry tomatillo (Gao et al., 2003). Thus, it is clear that the MagnICON transient expression system is superior to transgenic plants in terms of both the amount of antigen produced and the time frame required. Rapid transient expression is particularly valuable to vaccine production efforts in emergency situations, such as pandemic outbreaks or terrorism events, which require a quick response.
The primary structure of HBsAg contains 14 cysteines per 226 amino acid monomers (Valenzuela et al., 1979), providing the opportunity for the formation of numerous disulphide bonds. In mammalian cells, HBsAg forms dimers via intermolecular disulphide bonds in the ER lumen (Huovila et al., 1992), and further oligomerizes to assemble into VLPs in a post-ER, pre-Golgi compartment (Huovila et al., 1992). Previous studies have shown that the disulphide-bonded HBsAg dimers retain all antigenic determinants and full immunogenicity of the 22-nm VLPs (Mishiro et al., 1980), whereas HBsAg monomers are poorly immunogenic (Mishiro et al., 1980; Wampler et al., 1985). Several reports have shown that plant-derived HBsAg from either transgenic plants or plant cells (Smith et al., 2002a,b, 2003), or from transient expression (Huang and Mason, 2004), accumulates primarily as the dimeric form. In this study, it has been demonstrated that the majority of MagnICON-expressed HBsAg is present as dimers under non-reducing conditions and as monomers under reducing conditions (Figure 2), indicating that disulphide bonds hold the dimers together. Thus, plant-derived HBsAg from different expression systems forms disulphide-linked immunogenic dimers, in agreement with the immunogenicity of plant-derived HBsAg observed in this study (Figure 4) or reported previously (Kong et al., 2001; Smith et al., 2003; Sojikul et al., 2003). Mice immunized with very low doses of partially purified plant-derived antigen (1 µg HBsAg adsorbed to alum and delivered intraperitoneally) produced anti-HBsAg responses. Moreover, the antibody responses of plant HBsAg-immunized mice were boosted efficiently by inoculation with yeast recombinant HBsAg, which is the licensed vaccine. Thus, immunization with the plant-derived HBsAg produces B cells that can be recalled by stimulation with yeast-derived HBsAg, showing consistency of antigenicity and strongly suggesting the equivalence of the plant- and yeast-derived HBsAg.
In recombinant expression systems, only a portion of HBsAg can be folded correctly to form the ‘a’ determinant. The monoclonal reactivity of purified yeast-derived HBsAg varies from preparation to preparation in the range of 20%–50% of the reference human antigen (McAleer et al., 1984). The percentage of the ‘a’ determinant relative to total polyclonal reactive HBsAg is 21% for transgenic potato tubers (Smith et al., 2003) and 6%–37% for transgenic plant cell suspension cultures, depending on the culture stage (Smith et al., 2002b). Kumar et al. (2005) have reported that 67.87% of total HBsAg expressed in transgenic banana reacts with monoclonal antibody; however, the overall HBsAg level in this system was only 38 ng/g leaf FW. In the present study, it was found that, for leaves infiltrated using a syringe, only a small portion (≤ 7.5%) of total polyclonal antibody-reactive HBsAg displayed the conformational ‘a’ determinant (Table 1). The relatively low percentage of the ‘a’ determinant is probably a result of the saturation of the cell's capacity to process HBsAg as it is rapidly produced at very high levels. However, when the leaves were infiltrated using a vacuum, greatly improved ‘a’ determinant accumulation was obtained (up to 37.6% of total HBsAg). The vacuum infiltration study was conducted three times with similar results. Currently, the mechanism of this effect is not clear. However, one possibility is that cells in young growing leaves are transfected by vacuum, whereas syringe infiltration targets mature non-growing leaves because of their much greater ease of manipulation. Growing leaves are likely to possess a greater content of ER, which is utilized in the production of cell wall components, and thus may process HBsAg more efficiently.
It may be possible to further enhance the ‘a’ determinant formation of MagnICON-vectored HBsAg. First, improvement of the in vivo post-translational processing capacity by the upregulation of chaperone proteins, such as protein disulphide isomerase (PDI), may improve ‘a’ determinant formation. Second, prolonging the processing time of HBsAg in the ER by fusion with an ER retention peptide may lead to improved folding and disulphide bond formation. Richter et al. (2000) and Kumar et al. (2005) showed that the ‘SEKDEL’ retention signal enhanced ‘a’ determinant accumulation somewhat in transgenic potato and banana, respectively. However, another study in transgenic tobacco NT-1 cells showed that the addition of the C-terminal ‘SEKDEL’ produced no significant difference in the accumulation of HBsAg with the ‘a’ determinant (Sojikul et al., 2003), consistent with the observations in the current study. The failure of the ER retention signal to produce enhanced expression is also consistent with the observation that unmodified HBsAg accumulates in ER-derived vesicles in potato leaves and tobacco cells (Smith et al., 2003), indicating ER retention of antigen without ‘SEKDEL’. Third, in vitro manipulation could convert the poorly immunogenic monomer form of yeast-derived HBsAg to strongly immunogenic, extensively disulphide-bonded VLPs (McAleer et al., 1984; Wampler et al., 1985). For HBsAg derived from transgenic plant cell cultures, Smith et al. (2002a) have demonstrated that the Auszyme reactive fraction relative to total HBsAg can be increased to 75%–80% by adjusting the pH of the extraction buffer and the storage time. Whether such improvement can be reproduced for MagnICON-vectored HBsAg is being investigated.
Although our original research goal was to provide inexpensive, easily administered vaccines in the form of edible transgenic plant material, it is realized that plant-derived vaccines, like any other drug, will be subject to the strict regulations of the US Food and Drug Administration or other similar national agencies. Requirements, including dose standardization, must be met before a plant-derived vaccine is approved. To ensure dosing consistency, vaccine antigen-expressing plant materials must be subjected to downstream processing to some extent, including steps ranging from simple freeze–drying to chromatography purification. As a successful example, the first plant-made veterinary vaccine recently approved by the United States Department of Agriculture consists of the protective antigen of the Newcastle disease purified from transgenic plant cell cultures in an injectable form (http://www.dowagro.com/newsroom/corporatenews/2006/2006131b.htm). A major goal in our group remains the development of plant-based oral subunit vaccines, which will probably contain purified antigen and oral adjuvant in a formulation that is resistant to the harsh conditions in the gut. In the light of this, the high-level, rapid HBsAg expression (up to 0.64 mg total HBsAg/g leaf FW) obtained here establishes a solid economic mass base, greatly facilitating the downstream processing, purification and formulation of a vaccine product.
Viral vectors, including pICH11599, pICH14011, pICH15879 and pICH-GFP, have been described previously (Marillonnet et al., 2004; Huang et al., 2006) (Figure 1). pICH13840 is identical to pICH15879, but also contains the signal peptide sequence from an apple pectinase gene (GenBank accession L27743; Marillonnet et al., 2004). The codon-optimized HBsAg gene and the same with the hexapeptide ‘SEKDEL’ extension at the C-terminus were released from pHB217 or pHB221, respectively (Sojikul et al., 2003), by NcoI/SacI digestion and ligated into pICH11599 from the same sites, resulting in pICH-HBsAg (Figure 1) and pICH-HBsKDEL. The HBsAg vectors were used in concert with either pICH15879, which contains no targeting peptide, or pICH13840, which contains the ER signal peptide, in order to evaluate the effect of fusing an N-terminal ER-targeting domain to HBsAg.
Viral provectors were introduced into Agrobacterium tumefaciens strain GV3101 by electroporation. The resulting strains were grown overnight and used to infiltrate leaves of 6–8-week-old glasshouse-grown N. benthamiana plants using either a syringe, as described previously (Huang et al., 2006), or a vacuum. For vacuum-based infiltration, Agrobacterium cultures were spun down and resuspended in infiltration buffer [10 mm 2-(N-morpholino)ethanesulphonic acid (MES), 10 mm MgSO4, pH 5.5] to a final optical density at 600 nm (OD600) of 0.1. Three litres of Agrobacterium solution were used in a vacuum infiltration apparatus consisting of a vacuum desiccator (Belart, Pequannock, NJ, USA) and pump (Welch model 25468-01, Niles, IL, USA). N. benthamiana plants were inverted into the Agrobacterium solution and the desiccator was sealed. A vacuum was then applied to the system at 50–60 cmHg for 1 min. The vacuum was released and reapplied once more to ensure complete infiltration of the leaves.
Total soluble protein from N. benthamiana leaf was extracted and measured as described previously (Huang et al., 2005). Total HBsAg was quantified by a sandwich ELISA as described by Sojikul et al. (2003). The ‘a’ determinant of HBsAg was quantified using the Auszyme monoclonal kit (Abbott Diagnostics, Abbott Park, IL, USA). For Western blot analysis, plant extract or yeast-derived HBsAg standard (Advanced ImmunoChemical, Long Beach, CA, USA) was subjected to SDS-PAGE and blotting as described previously (Huang and Mason, 2004). Sucrose gradient sedimentation of plant extracts or yeast-derived HBsAg standard was performed as described previously (Huang et al., 2005). Immunoaffinity purification of HBsAg from crude leaf extract, using ‘a’ determinant-reactive Auszyme monoclonal antibody and negative stain electron microscopy, was performed as described previously (Huang and Mason, 2004). Briefly, leaf extract was incubated with one Auszyme monoclonal antibody-coated bead (Abbott Diagnostics) for 16 h at room temperature. After washing with water, bound HBsAg was eluted with 0.2 m glycine (pH 2.5) and immediately neutralized with tris(hydroxymethyl)aminomethane (Tris) base. The resulting material was stained with 0.5% aqueous uranyl acetate, and transmission electron microscopy was performed with a Philips CM-12S microscope (Philips, Eindhoven, The Netherlands).
HBsAg from pICH-HBsAg-infiltrated leaves was partially purified as described previously (Huang et al., 2005), and mixed (1 : 1) with Imject® alum (Pierce, Rockford, IL, USA) according to the manufacturer's instructions. Female BALB/c mice (6–8 weeks old), purchased from the National Cancer Institute, Frederick, MD, USA (Clarence Reeder), were immunized intraperitoneally with 1 µg alum-adsorbed antigen at weeks 0, 1 and 2. At week 25, mice were immunized intraperitoneally with 0.5 µg of alum-adsorbed yeast-derived recombinant HBsAg. Mice were bled via the retro-orbital plexus prior to and weekly following immunizations. The sera were separated and stored at –70 °C until evaluation for anti-HBsAg-specific antibodies using the AUSAB enzyme immunoassay (EIA) kit (Abbott Diagnostics), as described previously (Huang et al., 2005).
The authors thank Dr Yuri Gleba (Icon Genetics, Halle, Germany) for the Icon viral expression system, and Barbara Gonzales and Michael Lopker for technical assistance. This work was supported by National Institutes of Health grant AI 42836 to YT and HSM.