• Open Access

Expression of multiple proteins using full-length and deleted versions of cowpea mosaic virus RNA-2


* Correspondence (fax +44 1603450045; e-mail george.lomonossoff@bbsrc.ac.uk)


The use of multiple copies of vectors based on either full-length or deleted versions of cowpea mosaic virus RNA-2 for the production of heteromeric proteins in plants was investigated. Co-infiltration of two full-length RNA-2 constructs containing different marker genes into Nicotiana benthamiana in the presence of RNA-1 showed that the two foreign proteins were efficiently expressed within the same cell in inoculated tissue. Furthermore, the proteins were co-localized to the same subcellular compartments, an essential prerequisite for heteromer formation. However, segregation of two separate RNA-2 molecules, and therefore expression of the two proteins, was observed on systemic spread of the recombinant viruses. Thus, efficient assembly of heteromeric proteins is likely to occur only in inoculated tissue. To determine the optimum approach for expression in inoculated tissue, the heavy and light chains of the blood group-typing immunoglobulin G (IgG) C5-1 were inserted into full-length and deleted versions of RNA-2, and the constructs were agroinfiltrated in the presence of RNA-1. The results obtained showed that full-size IgG molecules accumulated using both approaches, but that the levels were significantly higher when deleted RNA-2 vectors were used. The levels were also greatly enhanced by the inclusion of an endoplasmic reticulum retention signal at the C-terminus of the heavy chain. As the potential benefit of using full-length RNA-2 constructs, the ability to spread systemically, appears to be irrelevant to the production of heteromeric proteins, the use of deleted versions of RNA-2 is clearly advantageous, particularly as they offer the benefit of biocontainment.


Cowpea mosaic virus (CPMV) is a bipartite RNA plant virus which has been developed into two types of vector system suitable for the production of heterologous polypeptides in plants (Liu et al., 2005; Sainsbury et al., 2007). Both types of system are based on the modification of RNA-2, but differ in whether a full-length or deleted version is used. In both cases, however, replication of the modified RNA-2 is achieved by co-inoculation with RNA-1.

The expression system based on a full-length version of RNA-2 involves the fusion of the foreign protein to the C-terminus of the RNA-2-derived polyprotein, release of the free foreign polypeptide being mediated by the action of the 2A catalytic peptide sequence from foot-and-mouth disease virus (FMDV; Gopinath et al., 2000). The resulting RNA-2 molecules are capable of spreading both within and between plants. This strategy has been used to express a number of recombinant proteins, such as the hepatitis B core antigen (HBcAg) and small immune proteins (SIPs), in cowpea plants (Alamillo et al., 2006; Mechtcheriakova et al., 2006; Monger et al., 2006). The plant-expressed HBcAg has been shown to self-assemble into core particles, and the SIP molecules, consisting of a single-chain antibody (scFv) fused to a domain from the constant region of a heavy chain, are able to dimerize in planta and neutralize infectivity in vivo.

Although successful, the use of a full-length viral vector has disadvantages in terms of the size constraints of inserted sequences and concerns about biocontainment. To address these, a system based on a deleted version of CPMV RNA-2 has recently been developed (Cañizares et al., 2006). In this system, the region of RNA-2 encoding the movement protein and both coat proteins has been removed. However, the deleted molecules still possess the cis-acting sequences necessary for replication by the RNA-1-encoded replicase, and thus high levels of gene amplification are maintained without the concomitant possibility of the modified virus contaminating the environment. With the inclusion of a suppressor of gene silencing, such as a helper-component proteinase (HcPro) from potato virus Y (PVY) (Brigneti et al., 1998), in the inoculum in addition to RNA-1, the deleted CPMV vector may be used as a high-yielding transient expression system (Sainsbury et al., in press). However, by contrast with the situation with a vector based on full-length RNA-2, replication is restricted to inoculated leaves.

Although CPMV vectors have been used successfully to express multichain complexes consisting of a single type of polypeptide, many therapeutic proteins, such as antibodies, are heteromeric, consisting of complexes of more than one polypeptide. The simplest way of achieving the expression of heteromeric proteins would be to inoculate plants with multiple RNA-2 molecules, each expressing an individual component, in the presence of RNA-1. However, for this approach to work and produce correctly assembled complexes, it is essential that the component chains all be expressed within the same cell. This, in turn, means that the multiple RNA-2 molecules must all be delivered to, and replicated within, the same cells. It was reasoned that it should be possible to achieve this without encountering problems of virus exclusion found with vectors based on monopartite viruses (Dietrich and Maiss, 2003; Giritch et al., 2006), as both RNA-2 molecules would be dependent on the RNA-1-encoded replicase for their amplification. The purpose of this study was therefore to investigate whether this could be achieved using CPMV vectors based on both full-length and deleted versions of RNA-2. This was initially investigated using RNA-2 molecules expressing different marker proteins, and was subsequently used to investigate the expression of a target molecule, the antihuman murine monoclonal immunoglobulin G (IgG) antibody C5-1, which is used in blood group typing. This was chosen as it has been successfully expressed in transgenic plants, where it retains its essential characteristics. The specificity and affinity of plant-produced C5-1 are identical to those of hybridoma-produced C5-1, and both have similar half-lives in mice (Khoudi et al., 1999). Furthermore, when produced in alfalfa, C5-1 is extremely stable during extraction and for at least 12 weeks in dry hay. The results obtained showed that it was possible to use two different RNA-2 molecules containing the heavy and light chains to express assembled IgG in plants. However, co-localization of the expressed marker proteins occurred efficiently only in inoculated leaves, with segregation occurring when the full-length RNA-2 constructs moved systemically. It was also shown that the deleted versions of RNA-2 were a particularly efficient means of achieving high levels of expression in inoculated leaves, and had the advantage that they provided a high degree of biocontainment.


Protein co-expression with full-length vector

To create two versions of RNA-2 whose expression could be monitored individually, the reporter genes green fluorescent protein (GFP) and Discosoma red fluorescent protein (DsRed) were inserted downstream of the FMDV 2A sequence in the vector pBinP-NS-1 (Liu et al., 2005). In both cases, the coding sequence was preceded by the signal peptide from the Arabidopsis basic chitinase gene (Samac et al., 1990), and followed by the HDEL endoplasmic reticulum (ER) retention signal (Denecke et al., 1990; Napier et al., 1992), to give constructs pBinP-NS-ER-GFP and pBinP-NS-ER-DsRed, respectively (Figure 1). This was performed so that the marker proteins would be localized and retained within the same subcellular compartment as required for the assembly and accumulation of antibodies. Infiltration of either of these modified RNA-2 molecules in the presence of RNA-1 and HcPro from PVY in Nicotiana benthamiana leaves resulted in high levels of the marker proteins being produced, as determined by Western blot analysis (Figure 2a). HcPro is a well-known suppressor of silencing that blocks the maintenance of post-transcriptional gene silencing (Brigneti et al., 1998). It is known to enhance the efficiency of infectivity by agroinoculated viral cDNAs (Chiba et al., 2006), and its presence significantly boosts expression levels from CPMV-based vectors (Cañizares et al., 2006). Co-infiltration of both RNA-2 molecules in the presence of RNA-1 and HcPro resulted in the expression of both proteins at levels similar to those seen when the individual constructs were infiltrated.

Figure 1.

Structure of the constructs used in this study and schematic diagram of the immunoglobulin G (IgG) heteromeric complex. HDEL and SEKDEL 3′ retention signals indicate endoplasmic reticulum (ER) retention (orange boxes). 2A, foot-and-mouth disease virus (FMDV) 2A self-cleaving peptide (black boxes); HC, C5-1 heavy chain (dark grey boxes); HcPro, helper-component protein; L, large coat protein; LC, C5-1 light chain (light grey boxes); MP, movement protein; Pro, proteinase; ProC, proteinase cofactor; S, small coat protein; SP, leader peptide from Arabidopsis basic chitinase (blue boxes).

Figure 2.

Expression of multiple proteins using the full-length version of cowpea mosaic virus (CPMV) RNA-2. (a) Western blot analysis of green fluorescent protein (GFP) and Discosoma red fluorescent protein (DsRed) co-expression (top panel) in samples taken from tissue either not infiltrated (lane 1) or infiltrated with Agrobacterium cultures containing pBinP-S-1-NT and pBIN61-HcPro plus one of the following: lane 2, pBinP-NS-ER-GFP; lane 3, pBinP-NS-ER-DsRed; lane 4, both pBinP-NS-ER-GFP and pBinP-NS-ER-DsRed. Bottom panel shows Coomassie-stained samples as a loading control. (b) Confocal microscopy of tissue infiltrated with pBinP-S-1-NT, pBIN61-HcPro, pBinP-NS-ER-GFP and pBinP-NS-ER-DsRed. Scale bar: 200 µm for 10× panels and 50 µm for 40× panels.

Confocal laser scanning microscopy was used to determine whether the expression of the two marker genes occurred in the same cells when leaves were co-infiltrated with pBinP-NS-ER-GFP and pBinP-NS-ER-DsRed in the presence of RNA-1 and HcPro. Examination of the expression of GFP and DsRed individually showed that most (> 95%) cells in the inoculated area expressed both proteins, and that these appeared to be correctly localized in the ER (Figure 2b). Merging the images obtained with GFP and DsRed showed a high degree of co-localization within cells, an essential precondition for the formation of heteromeric protein complexes (Figure 2b, panels showing merged images).

Multiple modified RNA-2 molecules segregate on systemic spread of CPMV

One of the potential advantages of full-length viral vectors is the possibility of systemic spread of the modified virus throughout the plant, with the concomitant expression of the foreign protein throughout the plant. To investigate whether the co-localization of the two marker proteins observed in inoculated tissue is maintained during systemic movement, upper, systemically infected leaves of plants co-inoculated with pBinP-NS-ER-GFP and pBinP-NS-ER-DsRed, either singly or together, in the presence of RNA-1 and HcPro were examined. Visible symptoms of systemic infection generally developed 2–3 weeks after inoculation in all cases (Figure 3, panels marked ‘visible light’). The systemically infected tissue of plants inoculated with pBinP-NS-ER-GFP alone showed clear GFP expression, as detected by both ultraviolet (UV) illumination and phosphor imaging (Figure 3, panels marked ‘UV’ and ‘FITC settings’). This was confirmed by confocal microscopy (data not shown). This result is consistent with that previously reported for RNA-2 expressing a cytoplasmic version of GFP in N. benthamiana (Liu and Lomonossoff, 2002). Likewise, systemically infected leaves of plants inoculated with pBinP-NS-ER-DsRed alone showed both visible symptoms and DsRed expression, as determined by phosphor imaging (Figure 3, panels marked ‘visible light’ and ‘Cy3 settings’). However, when plants were inoculated with both pBinP-NS-ER-GFP and pBinP-NS-ER-DsRed, only DsRed expression could be detected in the systemically infected leaves (Figure 3). This contrasts strongly with the situation found in inoculated tissue, and implies that virus derived from pBinP-NS-ER-DsRed somehow outcompetes that from pBinP-NS-ER-GFP when systemic movement takes place. This difference in expression may reflect the fact that the GFP and DsRed sequences differ in both size and composition, with the DsRed sequence being 48 bases shorter than GFP. This difference in size may confer a replicative advantage on the DsRed-containing RNA-2 molecules, as increasing insert size is known to have a negative effect on the replication rate (Liu et al., 2005). Alternatively, there could be a difference in the efficiency of encapsidation of the two RNA-2 molecules, which could affect the rate of systemic spread. It also indicates that an exclusion mechanism may be operating in systemically infected tissue, in which the previous presence of one type of RNA-2 prevents a second RNA-2 from being expressed. Thus, the more rapid replication and movement of the DsRed construct would effectively prevent expression of the GFP construct.

Figure 3.

Segregation of foreign protein expression from recombinant cowpea mosaic virus (CPMV) on systemic movement. Systemically infected leaves from plants 22 days after infiltration with pBinP-S-1-NT, pBIN61-HcPro and recombinant RNA-2 clones, as indicated, were visualized under visible and ultraviolet (UV) light. The fluorescein isothiocyanate (FITC) and cyanine 3 (Cy3) settings on a phosphorimager detect green fluorescent protein (GFP) and Discosoma red fluorescent protein (DsRed), respectively. Scale bar, 20 mm.

To investigate whether such an exclusion mechanism is operating, two additional RNA-2-based constructs, pBinP-NS-ER-CFP and pBinP-NS-ER-YFP, were created. These contained marker genes [cyan fluorescent protein (CFP) and yellow fluorescent protein (YFP)] which share 97% nucleic acid sequence identity (compared with < 50% between GFP and DsRed) and are identical in length, and thus RNA-2 molecules containing them should have identical replication and encapsidation efficiencies. The new clones were constructed in the same way as pBinP-NS-ER-GFP, so that the marker genes were targeted to, and retained within, the ER (Figure 4a). When the two constructs were co-infiltrated into leaves in the presence of RNA-1 and HcPro, both proteins were co-localized in infiltrated leaf patches as before (data not shown). The upper leaves developed symptoms of systemic infection after 2–3 weeks. Examination of the symptomatic patches by confocal laser scanning microscopy revealed delimited areas expressing one or other, but not both, of the marker genes (Figure 4b). The fact that expression from both RNA-2 molecules can be detected in systemically infected leaves indicates that, as expected, they have similar capacities for systemic movement. However, the observation that the expression of CFP and YFP did not overlap but, rather, was confined to separate patches of cells (Figure 4b, panels marked ‘merge’) indicates that, even if the foreign sequences within different RNA-2 molecules are almost identical, segregation of expression occurs as the recombinant viruses spread through the plant. Although the cause of this segregation or mutual exclusion is not clear, it indicates that it will not be possible to simultaneously express two different polypeptides within the same cell in systemic tissue using two different RNA-2 molecules. Consequently, the ability of full-length CPMV-based vectors to spread systemically will not be an advantage for the co-expression of different polypeptide chains.

Figure 4.

Spatial separation of highly similar recombinant cowpea mosaic virus (CPMV) in systemically infected leaves. (a) Structure of the full-length RNA-2 constructs used to monitor the endoplasmic reticulum (ER)-retained co-expression of cyan fluorescent protein (CFP) and yellow fluorescent protein (YFP). HDEL retention signal is shown in orange. (b) Confocal microscopy of tissue systemically infected by CFP- and YFP-expressing recombinant CPMV. 2A, foot-and-mouth disease virus (FMDV) 2A self-cleaving peptide (black boxes); L, large coat protein; MP, movement protein; S, small coat protein; SP, leader peptide from Arabidopsis basic chitinase (dark blue boxes). Scale bar: 200 µm for 10× panels and 50 µm for 40× panels.

IgG expression with CPMV vectors

Given that the ability of CPMV vectors based on full-length RNA-2 to move systemically does not confer any advantage with regard to heteromeric protein formation, the efficacy of CPMV-based vectors in full-length and deleted forms (Cañizares et al., 2006) for the expression of a murine monoclonal antibody (C5-1) in agroinfiltrated leaves was compared.

To this end, the light and heavy chain coding sequences of the antibody were assembled in separate full-length or deleted RNA-2 molecules in native (secreted) and ER-retained forms. In each case, the native murine signal peptide was incorporated at the N-terminus of the sequence and ER retention was achieved, where required, by the addition of a SEKDEL sequence at the C-terminus. Schematic representations of the full-length and deleted RNA-2 constructs obtained are presented in Figure 1. The expression strategy from the full-length RNA-2, including the release of the antibody chains through the action of the 2A catalytic peptide, was identical to that used for the marker proteins. In the deleted version, the region encoding the Ig chain was flanked by nucleotides 1–512 from the RNA-2 5′ untransformed region (UTR) and 3330–3481 from the 3′ UTR. These terminal sequences have been shown to be sufficient to permit replication of the RNA-2-derived construct by RNA-1 provided that a suppressor of silencing is present (Cañizares et al., 2006).

For each expression strategy assayed, antibodies were expressed by co-infiltrating a light chain-containing RNA-2 and a heavy chain-containing RNA-2 in the presence of RNA-1 and HcPro. For both the full-length and deleted versions of the vector, light and heavy chain constructs were used in different combinations, allowing ER retention of the light and/or heavy chain of the antibody. All leaves of a 6-week-old plant were completely filled with appropriate mixtures of Agrobacterium suspension.

Following a 7-day incubation period, infiltrated plants were harvested and analysed by immunological methods. Western blot analysis, using peroxidase-conjugated anti-mouse IgG antibodies, of protein extracts electrophoresed on non-reducing gels showed a major band of approximately 220 kDa (Figure 5), corresponding in size to a control murine IgG1. This indicates that fully assembled C5-1 antibodies accumulate in infiltrated plant tissue, irrespective of the expression strategy employed. The unexpectedly high apparent molecular weight of the fully assembled antibody (220 kDa instead of 150 kDa) is commonly observed for Igs when analysed by electrophoresis using Tris-Glycine running buffers (Giritch et al., 2006; Peterson et al., 2006). In addition to the fully assembled antibody, the Western blot revealed immunoreactive fragments at approximately 130, 95, 60, 50 and 35 kDa, with band intensities that varied with global accumulation level. Similar fragments have been described for other antibodies expressed in plant cell cultures (Sharp and Doran, 2001), transgenic plants (Khoudi et al., 1999) and plants agroinfiltrated with a viral vector (Giritch et al., 2006).

Figure 5.

Western blot analysis of C5-1 expression. Tissue was infiltrated with pBinP-S-1-NT, pBIN61-HcPro and combinations of C5-1-containing the recombinant RNA-2 clones as indicated. Native, both heavy chain (HC) and light chain (LC) secreted; HC/KDEL, HC endoplasmic reticulum (ER)-retained; LC/KDEL, LC ER-retained; HC + LC/KDEL, both HC and LC ER-retained.

A dramatic improvement in the level of antibody accumulation was obtained when using the deleted RNA-2 vector rather than the full-length RNA-2 vector. Furthermore, ER retention increased the accumulation of antibodies in the leaves, but to a lesser extent with the full-length RNA-2 vector than the deleted version (Figure 5). The antibody accumulation level was quantified by enzyme-linked immunosorbent assay (ELISA) using a polyclonal goat anti-mouse IgG (heavy chain specific) for capture and a peroxidase-conjugated goat anti-mouse IgG (heavy and light chain) for detection. In the native form, antibody accumulation reached 0.4% of total soluble protein (TSP) when produced using the full-length RNA-2 (Figure 6), corresponding to a yield of 19 mg/kg of fresh weight (FW) of leaves. However, by contrast with the Western blot analysis, which indicated a slight increase in antibody accumulation on addition of an ER retention peptide to the heavy chain or both the light and heavy chain, ELISA showed a decrease in the accumulation of ER-retained forms when produced with the full-length RNA-2 vector. The cause of this discrepancy, if significant, is not yet understood. As expected from Western blot analysis, use of the deleted RNA-2 vector for C5-1 expression increased antibody accumulation, in native form to 0.75% TSP, and maximal antibody accumulation was obtained when combining ER retention of the heavy chain or both the light and heavy chain with the use of the deleted RNA-2 vector, with up to 1.9% TSP (corresponding to 74 mg/kg FW).

Figure 6.

Enzyme-linked immunosorbent assay of C5-1 expression. Tissue was infiltrated with pBinP-S-1-NT, pBIN61-HcPro and combinations of C5-1-containing the recombinant RNA-2 clones as indicated. Native, both heavy chain (HC) and light chain (LC) secreted; HC/KDEL, HC endoplasmic reticulum (ER)-retained; HC + LC/KDEL, both HC and LC ER-retained. Values represent the percentage of extracted total soluble protein (TSP).


It has been shown that multiple copies of RNA-2 containing different polypeptides can be used to produce heteromeric proteins in plants. Using constructs expressing both reporter proteins and antibody chains, it has been demonstrated that at least two foreign proteins can be produced at high levels and co-directed to the same location in each cell of inoculated tissue. This is probably a result of the bipartite nature of the CPMV genome, allowing two related viral vectors to replicate efficiently in the same cell. This represents the first occasion on which a bipartite virus has been used to express assembled antibodies in plants.

Previous reports on the expression of assembled antibodies have all used monopartite viruses. Verch et al. (1998) and Alamillo et al. (2006) used vectors based on tobacco mosaic virus (TMV) and potato virus X (PVX), respectively, and, in each case, infected plants with separate constructs expressing the heavy and light chains. Although the antibodies produced were shown to be able to assemble, and were capable of binding their cognate antigen, in neither case was the yield of protein quantified. One problem with the use of multiple copies of a monopartite virus for the expression of multiple proteins is that segregation occurs even in infiltrated tissue (Dietrich and Maiss, 2003; Giritch et al., 2006). To avoid this problem, Giritch et al. (2006) used two different viral vectors, based on TMV and PVX which are known to exist synergistically in nature, to express the heavy and light chains of a human monoclonal IgG. This approach allowed large amounts of antibody to be produced, but required the co-infiltration of six different Agrobacterium cultures (Giritch et al., 2006). In addition, the PVX-based constructs were based on a full-length version of the virus which produces infectious virus particles. Thus, the system is not fully biocontained.

A notable finding from the current study is that substantially higher levels of assembled antibody can be obtained using deleted rather than full-length versions of RNA-2. This may reflect differences in the efficiency of translation and/or replication between the two types of RNA-2 molecule. Full-length versions of RNA-2 direct the accumulation of only low levels of marker genes when expressed in the absence of RNA-1 (Cañizares et al., 2006; F. Sainsbury and G. P. Lomonossoff, unpubl. data), indicating that replication of these RNA molecules is essential for high levels of expression. By contrast, deleted versions of RNA-2 can express substantial levels of protein even in the absence of replication, although these levels can be enhanced by expression of RNA-1 (Cañizares et al., 2006). These differences in behaviour most probably reflect differences in the stability of the two types of RNA-2 in the absence of replication. Whatever the reason, it is clear that the use of deleted versions of RNA-2 has distinct advantages for the expression of heteromeric proteins in inoculated tissue, not only in terms of biocontainment but also in terms of the expression levels obtainable.

A potential advantage of using full-length rather than deleted versions of RNA-2 is that expression is not limited to inoculated tissue. However, the experiments using full-length constructs containing marker genes showed that segregation of expression occurred on systemic movement of the recombinant viruses. When CFP- and YFP-containing RNA-2 constructs were inoculated together, both molecules appeared to have similar abilities to spread systemically; however, protein expression from the systemic infections did not overlap. In the case of GFP and DsRed expression, the separation was more severe, with only DsRed detectable in systemic leaves. Spatial separation, or the exclusion of a virus from tissue infected by another virus, is well documented. Although some viruses coexist synergistically (Pruss et al., 1997), some combinations of different viruses (Dietrich and Maiss, 2003), or even different strains of the same virus (Hull and Plaskitt, 1970; Dietrich and Maiss, 2003; Giritch et al., 2006), are unable to replicate in the same tissue. Often discussed in terms of cross-protection, it has been shown that, for mixed infections of the unrelated potex- and tobraviruses, this phenomenon is RNA-mediated and based on gene silencing (Ratcliff et al., 1999). However, in our case with mixed populations of differently labelled CPMV RNA-2 molecules, the mechanism behind the observed spatial separation is not known. We believe that it is unlikely to be caused by gene silencing, as similar segregation was found when RDR6i plants (Schwach et al., 2006), in which such silencing is suppressed, were co-inoculated with two different RNA-2 constructs (F. Sainsbury and G. P. Lomonossoff, unpubl. data). It may be relevant that RNA derived from pBinP-NS-ER-GFP can be detected in the systemic leaves of plants co-inoculated with pBinP-NS-ER-DsRed, despite the fact that no GFP expression can be detected (data not shown). This suggests that some form of translational competition may be involved.

A possible solution to the problem of the spatial separation found in systemic tissue when full-length CPMV RNA-2 molecules are used might be to adapt the approach used by Verver et al. (1998). This employed two RNA-2 molecules with complementing deletions, and effectively created a tripartite virus as all three RNA molecules were essential for virus propagation. However, it was found that the two deletion mutants tended to recombine to restore a wild-type, and careful design of the constructs would be required to minimize this.

Despite the high level of expression obtained using the deleted version of RNA-2, this approach suffers from the limitation that expression is restricted to inoculated tissue. If necessary, this disadvantage could be addressed in two ways. The first would be to increase the amount of tissue that is infiltrated. This could readily be achieved using methods such as the vacuum infiltration of whole plants. The second would be to use the combined transgene/virus vector approach described by Cañizares et al. (2006). In this, a deleted version of RNA-2 harbouring GFP (1-GFP) was used to transform N. benthamiana, and the resulting plants were crossed with plants transgenic for RNA-1 and HcPro. Plants harbouring all three transgenes showed very high levels of GFP expression. It is highly probable that a similar result would be obtained if two deleted RNA-2 molecules carrying different inserts were incorporated, and would result in stably transformed lines expressing high levels of the heteromeric protein.

All the work reported in this article concerned the expression of foreign proteins in the model plant N. benthamiana. This host was selected as it is particularly amenable to agroinfiltration, with a minimum amount of damage being caused to the inoculated tissue during the process. This is particularly important for studies using fluorescent marker genes, such as GFP, as wound-derived autofluorescence can interfere with the assessment of gene expression. This is a significant drawback when using the natural host of CPMV, cowpea (Vigna unguiculata), which possesses a thick cuticle, which must be extensively damaged to allow constructs to be agroinfiltrated into leaves. However, cowpea has the advantage as an expression system that it is an edible plant and has already been used for the expression of HBcAg and SIPs (Alamillo et al., 2006; Mechtcheriakova et al., 2006; Monger et al., 2006). We have recently conducted preliminary experiments on the expression of a human anti-human immunodeficiency virus (anti-HIV) IgG in cowpea, which have shown that deleted RNA-2 molecules can be used to produce assembled antibodies in this host (F. Sainsbury and G. P. Lomonossoff, unpubl. data). This, coupled with the recent report of the successful regeneration of transgenic cowpea (Popelka et al., 2006), indicates that the expression of high levels of heteromeric therapeutic proteins in an edible plant will be possible. Furthermore, given the apparent efficiency of co-localization of proteins within the same cell, it also seems likely that heteromeric proteins consisting of more than two polypeptides could be successfully produced using the multiple RNA-2 approach described in this article.

Experimental procedures


Plasmids containing CPMV sequences based on the binary transformation vector pBINPLUS (van Engelen et al., 1995) have been described previously. Full-length CPMV RNA-1 was provided by pBinP-S-1-NT (Liu and Lomonossoff, 2002). Full-length RNA-2 constructs were based on pBinP-NS1 (Liu et al., 2005). Deleted versions of RNA-2 were based on 35S-1-GFP (Cañizares et al., 2006). C5-1 light and heavy chain coding sequences were amplified from pGA643-kappa and pGA643-gamma, respectively (Khoudi et al., 1999).

Construction of full-length CPMV RNA-2 clones

The construction of pBinP-NS-ER-GFP was based on pCP2/S-2A-GFP (Gopinath et al., 2000). This contains the complete sequence of CPMV RNA-2, the 3′ end of which has been modified to express GFP downstream of an in-frame FMDV 2A sequence. The leader peptide from Arabidopsis basic chitinase (KTNLFLFLIFSLLLSLSSAEF; Samac et al., 1990) was inserted between the 2A sequence and GFP, and an HDEL ER retention signal was added to the 3′ end of the GFP sequence to give pCP2/S-2A-AraSP-GFP-HDEL (F. Sainsbury et al., in preparation). The plasmid was digested with EcoRI and BamHI, which cut the RNA-2 sequence in the 5′ UTR and 3′ end of the movement protein, respectively. The fragment containing the marker gene was inserted into similarly digested pBinP-NS1 to give pBinP-NS-ER-GFP. DsRed, CFP and YFP sequences were amplified by polymerase chain reaction (PCR) from pDsRed2, pECFP and pEYFP, respectively (Clontech-Takara Bio Europe, Saint-Germain-en-Laye, France). The forward primer incorporated an EagI site followed by the 3′ portion of the basic chitinase leader peptide, and the reverse primer introduced an HDEL retention sequence followed by the stop codon and StuI site. EagI/StuI digestion of pCP2/S-2A-AraSP-GFP-HDEL released GFP together with the 3′ portion of the leader peptide, and the remaining fragment became the recipient vector for EagI/StuI-treated DsRed, CFP and YFP PCR products. The plasmids generated were digested with EcoRI and BamHI and inserted into similarly digested pBinP-NS1 to give pBinP-NS-ER-DsRed, pBinP-NS-ER-CFP and pBinP-NS-ER-YFP, respectively.

Plasmids pBinP-C5-1HC, pBinP-C5-1LC, pBinP-C5-1HC-ER and pBinP-C5-1LC-ER were assembled as follows. Coding sequences of C5-1 light and heavy chain were amplified by PCR with oligonucleotides containing ApaI (forward primer) and StuI (reverse primer) restriction sites (Table 1). ApaI/StuI-digested PCR products were inserted into similarly digested pBinPS2NT (Liu et al., 2005).

Table 1.  Oligonucleotide primers used for the assembly of cowpea mosaic virus (CPMV)-based vectors for the expression of C5-1. Bold text denotes restriction sites, italicised text indicates SEKDEL sequence, underlined text designates C5-1 sequences, and bold underlined text designates homology to 35S-1-GFP.

Construction of deleted CPMV RNA-2 clones

A deleted CPMV RNA-2 containing the C5-1 light chain between the ApaI and StuI sites was assembled using a PCR-based ligation procedure described by Darveau et al. (1995), employing the oligonucleotides shown in Table 1. A first PCR amplification was performed using primers CPMV-3207C and LC C51-MP(CPMV).R with 35S-1-GFP (Cañizares et al., 2006) as template. A second fragment was generated by PCR amplification using primers ApaI-LC (C5-1).1c and LC (C5-1)-StuI.r with pGA643-kappa as template. In parallel, a third PCR amplification was performed with primers LC C51-3′UTR(CPMV).C and CPMV-4926.R with 35S-1-GFP as template. The three fragments obtained were mixed together, and the mixture was used as template for a fourth PCR reaction with primers CPMV-3207C and CPMV-4926.R. The resulting amplification fragment was digested with DraIII and AscI, and ligated into the same sites in 35S-1-GFP, creating pBD-C5-1LC. Plasmids pBD-C5-1HC, pBD-C5-1LC-ER and pBD-C5-1HC-ER were created by inserting ApaI/StuI fragments from pBinP-C5-1HC, pBinP-C5-1LC-ER and pBinP-C5-1HC-ER, respectively, into similarly digested pBD-C5-1LC.


Binary plasmid constructs were maintained in Agrobacterium tumefaciens strain LBA4404 or AGL-1. Agroinfiltration was carried out as described previously (Monger et al., 2006).

Confocal microscopy

Confocal laser scanning microscopy was carried out using a Leica SP2 (Leica Microsystems, Wetzlar, Germany) with a × 10 or × 40 oil immersion objective. Excitation of GFP utilized the 488-nm channel of an argon ion laser and emission from 505 to 555 nm was detected. DsRed was excited by the 543-nm channel of a helium/neon laser and emission wavelengths between 570 and 650 nm were detected. CFP and YFP were excited by the 458-nm and 514-nm channels of the argon ion laser, respectively, with CFP detected between 475 and 500 nm and YFP detected between 530 and 570 nm. For the detection of multiple marker proteins in the same tissue, scanning was carried out in sequential mode.

Immunological detection of GFP and DsRed

Infiltrated leaf tissue was homogenized in the presence of protein extraction buffer [50 mm Tris-HCl pH 7.25, 150 mm NaCl, 2 mm ethylenediaminetetraacetic acid (EDTA), 0.1% (v/v) Triton X-100]. Lysate was clarified by centrifugation and protein concentrations were determined by the Bradford assay. Approximately 30 µg of protein extract was separated by sodium dodecylsulphate-polyacrylamide gel electrophoresis (SDS-PAGE) under reducing conditions (Laemmli, 1970), and electroblotted on to nitrocellulose. The membrane was simultaneously probed with the polyclonal Living Colors A.v. peptide antibody for GFP detection and Living Colors DsRed monoclonal antibody (Clontech-Takara Bio Europe). Both anti-mouse and anti-rabbit horseradish peroxidase-conjugated secondary antibodies were applied (Amersham Biosciences, Little Chalfont, Surrey, UK), and bands were visualized by electrochemiluminescence (ECL) captured on Hyperfilm (Amersham Biosciences).

Whole leaf imaging

Systemically infected leaves were photographed with a Nikon D1x digital camera (Nikon, Kingston, Surrey, UK) under polarizing light for visible light images or under UV illumination from a Blak-Ray B-100AP UV lamp (Blak-Ray, Upland, CA, USA) for the detection of GFP. The fluorescence detection mode of a Fujifilm phosphorimager FLA-5000 (Fujifilm, Tokyo, Japan), fitted with an FLA filter tray, was used, with fluorescein isothiocyanate (FITC) settings (excitation at 473 nm; emission at 530 nm; bandpass, 20 nm) for the detection of GFP, and cyanine 3 (Cy3) settings (excitation at 532 nm; emission at 570 nm; bandpass, 20 nm) for the detection of DsRed.

Analysis of C5-1 expression

For Western blot analysis, proteins from total crude extracts or purified antibody were separated by SDS-PAGE and electrotransferred on to polyvinylidene difluoride (PVDF) membranes (Roche Diagnostics Corporation, Indianapolis, IN, USA) for immunodetection. Membranes were blocked with 5% skimmed milk and 0.1% Tween-20 in Tris-buffered saline (TBS-T), and probed with a peroxidase-conjugated goat anti-mouse IgG (heavy and light chain) antibody (Jackson ImmunoResearch, West Grove, PA, USA). Immunoreactive complexes were detected by chemiluminescence using luminol as the substrate (Roche Diagnostics Corporation). Horseradish peroxidase enzyme conjugation of human IgG antibody was carried out using the EZ-Link Plus® Activated Peroxidase Conjugation Kit (Pierce Biotechnology, Rockford, IL, USA).

Quantification of C5-1 accumulation was performed by ELISA as follows. Multiwell plates (Immulon 2HB, ThermoLab System, Franklin, MA, USA) were coated with 2.5 µg/mL of goat anti-mouse antibody specific to IgG1 heavy chain (Sigma, St. Louis, MO, USA) in 50 mm carbonate buffer (pH 9.0) at 4 °C for 16–18 h. Plates were then blocked with 1% casein in phosphate-buffered saline (PBS) (Pierce Biotechnology) for 1 h at 37 °C. A standard curve was generated with dilutions of a purified mouse IgG1 as control (Sigma). When performing the immunoassays, all dilutions (control and samples) were carried out in a plant extract obtained from plant tissue infiltrated with pBinP-S-1-NT alone to eliminate any matrix effect from plant proteins in the extract. Plates were incubated with protein samples and standard curve dilutions for 1 h at 37 °C. After three washes with 0.1% Tween-20 in PBS (PBS-T), the plates were incubated with a peroxidase-conjugated goat anti-mouse IgG (heavy and light chain) antibody (Jackson ImmunoResearch) at 0.04 µg/mL in blocking solution for 1 h at 37 °C. The washes with PBS-T were repeated and the plates were incubated with a 3,3′,5,5′-tetramethylbenzidine (TMB) Sure Blue peroxidase substrate (KPL, Gaithersburg, MD, USA). The reaction was stopped with 1 m HCl and the absorbance was read at 450 nm. Each sample was assayed in triplicate, and the concentrations were interpolated in the linear portion of the standard curve.


The authors would like to thank Grant Calder, Andrew Davis and Hannes Vogler for technical assistance, and David Baulcombe for providing pBIN61-HcPro. Part of the work described in this article was funded under the EU FP6 ‘PharmaPlanta’ project. F.S. acknowledges funding from a Marie Curie Early Stage Training Fellowship MEST-CT-2004-504273 and the Trustees of the John Innes Foundation.