Several different viruses have been used for protein production in plants, and each has advantages as well as disadvantages [28–32]. Many of the “first generation” plant virus vectors [28, 33] utilized whole plant systemic virus infections to give desired proteins. While many of these have proven to be very useful there are some significant drawbacks. Systemic infections can take several days to fully develop. Protein production is then asynchronous and yields can vary in different tissues . Recombinant viruses also show size constraints for the inserted sequence, often coding sequences of only 1 kb or less (encoding a protein of only ~35 kDa) can be inserted . Then as the infection develops the viruses partially or completely excise the inserted recombinant sequence, leading to loss of the desired intact protein product [28, 35]. Furthermore, some viruses (e.g. those with icosahedral capsids such as CMV) may have even more severe size constraints if RNA encapsidation is a requirement for development of the systemic infection. Then if the coding sequence for the desired protein is large, insertion into the viral RNA may preclude encapsidation, thereby preventing efficient spread.
Recent progress in developing Agrobacterium tumefaciens delivered plant virus-based protein production systems has been made by several research groups using different plant viruses [31, 34, 36]. These plant virus-based amplicon systems offer many advantages including the fact that non-transgenic plants can be used, the desired protein production is rapid, the product can accumulate to high levels, and virus-based expression can be temporally regulated to be almost synchronous in all infiltrated areas. Because a majority of the infiltrated cells become simultaneously infected, virus movement to new cells is not necessary, encapsidation of recombinant RNAs is not an issue. These “second generation” virus-based systems also can retain larger foreign coding sequences and thus produce larger proteins in plants , here we produced enzymatically active 56,000 MW E1.
In our previous work, we used the estradiol-inducible, CMV-based CMViva to produce α anti-trypsin (AAT) in non-transgenic N. benthamiana plants . CMViva has all three CMV genome components in one large 28 kbp plasmid, which, due to its large size is difficult to manipulate. Thus, here we took approaches to develop CMV-based inducible (CMVin) as well as autonomously replicating (CMVar) systems, both of which are more easily manipulated and might be able to give high accumulation of heterologous proteins in plants. First, we separated the CMV genomic RNA cDNAs onto two plasmids, one containing the RNA 1 and 2 replication-associated genome components and the other containing the CMV RNA 3 genome segment. The CMV RNA3 component is rather small in size, 2.2 kb, and is easy to manipulate and to engineer to contain restriction enzyme sites to allow for easy removal of the CMV CP gene and replace it with any gene of interest. The desired restriction enzyme sites were introduced into the intergenic region of RNA 3. As expected, these altered the 5′ untranslated leader sequence of the resulting mRNA (RNA 4). The sgRNA promoter (for RNA 4 transcription) is within the minus strand of RNA 3 and is recognized by the RNA-dependent RNA polymerase and mRNA transcription is initiated. For CMV-Q RNA3, the transcription initiation starts at nt position 1167 in the intergenic region, which is upstream of the modified leader sequences. Our analyses demonstrated that the RNA 3 modifications affected RNA 3 and RNA 4 accumulation, but showed even more unpredictable effects on resulting protein accumulation. It does not appear that these can be attributed only to start codon context [18, 19] as the same construct (RNA 3) showed different protein yields whether the RNA was delivered using CMVin vs.
In contrast to CMViva, both CMVinII and CMVarI and II variants require mixing A. tumefaciens cells containing different plasmids which are then co-infiltrated into plants and T-DNA from the different A. tumefaciens cells containing the CMV plasmids must be transferred to the same plant cell for the complete CMV amplicon. For CMVinII variants this is then followed by induction using estradiol, which resulted in high level accumulation of the proteins tested here (CP, GFP, E1). However, like for CMViva, the CMVinII RNA 1 deletion does not allow for its replication, only translation of the newly transcribed mRNA. Therefore we also developed the non-inducible autonomously replicating CMV-based system, CMVarI and II. Wildtype CMVar (expressing the CMV CP) replicated to very high levels and even caused systemic infections in plants. However, when genes for GFP or E1 were substituted for the CP gene, both proteins were produced in plants within the infiltrated areas, and quantitative analyses showed that high levels of proteins accumulated for both CMVin vs.
CMVar, particularly at 6 days post-infiltration.
Although CMVinII E 6 and 8 variants gave slightly more active E1 in most experiments, CMVarI and II variants offers advantages in ease of use (e.g. no need to add the inducer) and thus, two additional approaches to improve accumulation of the desired protein product were investigated. Like most viruses having genomes composed of multiple segments, CMV genomic RNAs can be mixed (reassortment) to achieve genetic diversity [37, 38], and this offers opportunities for using CMV to produce desirable proteins in different plant species, as has been demonstrated also by others . Therefore, we generated CMV reassortant genotypes by substituting CMV subgroup I genomic RNAs 1 and 2 derived from a more virulent CMV, with the original CMV Q subgroup II RNA 3, giving CMVarI. Comparison of CMVarI and CMVarII G, E variants showed higher GFP and E1 for CMVarI G, E variants. However, the CMVarI Ewt showed higher E1 accumulation than did the corresponding CMVinII Ewt, but CMVinII E variants 6 and 8 gave the higher E1 accumulation than CMVar I 6 and 8 variants thus showing that reassortment alone was not sufficient.
As another alternative, we generated a MP C-terminal 33 amino acid deletion mutant. Cell-to-cell movement in CMV-infected plants requires interactions between the CP and MP . Our CMV-based systems including CMViva, CMVarI and II and CMVinII are cell-to-cell movement deficient since they lack the CP and thus, desired recombinant proteins are produced only within infiltrated cells. However, previous workers demonstrated that the CMV MP C-terminal 33 amino acids are essential to recognize and interact with the CP . When this region is deleted, the CMV infections can spread cell-to-cell even in the absence of CP . In support of this the CMVarI and CMVarII 33 G variants showed high CMV-based GFP production (Figure 6; and see ). When we created MP 33 amino acid deletion constructs and tested them, they showed increased production of not only GFP in CMVarI and II 33 G variants, but also of E1 in CMVarI and II 33E variants (Table 3, Figures 5 and 6), and the highest yields of active E1 were obtained using the CMVarI 33E variants.
Other workers have produced versions of E1 in various transgenic plants with gene expression driven by different promoters. For example, full-length E1 containing the catalytic domain, linker and carbohydrate binding domain has been previously produced in transgenic tobacco plants. Based on the resulting E1 activity, yields of up to 0.25% on average of total leaf soluble proteins were shown with Mac promoter, a chimeric promoter of the CaMV 35S and mannopine synthase gene . Similar yields were shown also with CaMV 35S promoter . In transgenic Z. mays seeds, the full-length E1 was produced using Glob-1 (Maize embryo-preferred globulin-1 promoter) and yields up to 6% TSP were obtained . In transgenic rice (Oryza sativa) plants, 35S driven E1 lacking the carbohydrate binding domain but only containing the catalytic domain gave yields up to 4.9% TSP . Thus, our yields of up to 0.4% TSP of intact E1 in nontransgenic N. benthamiana plants are similar to those achieved for intact E1 in transgenic tobacco, but less than those in more specialized systems. Furthermore, CMVin and CMVar-based production of the desired protein can be temporally regulated to give almost synchronous protein accumulation over a very short time period, even a few days.