Plant‐based production can result in covalent cross‐linking of proteins

Antibodies, antigens, enzymes for replacement therapies and virus-like particles (VLPs) have all been produced successfully in plants as part of the concept of "Molecular Farming" (Lomonossoff and D'Aoust, 2016). There have been several differences noted between plant-expressed proteins and their equivalents produced in other systems, such as CHO cells, particularly regarding their glycosylation.

Antibodies, antigens and enzymes for replacement therapies and virus-like particles (VLPs) have all been produced successfully in plants as part of the concept of 'Molecular Farming' (Lomonossoff and D'Aoust, 2016). There have been several differences noted between plant-expressed proteins, and their equivalents produced in other systems, such as CHO cells, particularly regarding their glycosylation. However, several publications have also indicated that preparations of plant-expressed proteins, including antibodies, VLPs and soluble molecules such as HIV gp120, have a higher proportion of multimers or aggregates than their CHO-or yeast-expressed equivalents (e.g. Mechtcheriakova et al., 2006;Ramessar et al., 2008;Rosenberg et al., 2013). The cause of this has not been investigated in detail because the molecules expressed in the different systems are often not identical and may have been purified to different extents.
As part of our studies on virus maturation, we have expressed the coat protein of the insect virus, Nudaurelia capensis omega virus (NxV) in both Nicotiana benthamiana (a heterologous expression system) and insect cells (the homologous system). NxV belongs to the Tetraviridae, a family of viruses with nonenveloped T = 4 capsids and single-stranded positive-sense RNA genomes that infect Lepidoptera. Maturation of the virus particles from procapsid to mature capsid involves a compaction of the particles from 48 to 41 nm diameter with a concomitant autocatalytic cleavage of the full-length coat protein (a-peptide) to the b and c peptides, with sizes of 70, 62 and 8 KDa, respectively. Experiments with NxV coat protein expressed in insect cells have shown that this maturation can be triggered by reducing the pH of a suspension of procapsids from 7.6 to 5.0 in vitro (Canady et al., 2000). We have recently shown that mature particles extracted from N. benthamiana leaves expressing the a-peptide are very similar in structure from the mature particles derived from insect cells and to contain the cleaved b and c peptides (Berardi et al., 2020;Castells-Graells, 2019). This indicates that the maturation process can be faithfully recapitulated in plants. By contrast, when particles were extracted from leaves as procapsids at pH 7.6 and then matured in vitro by reducing the pH to 5.0, the kinetics of cleavage of the a-peptide were consistently slower and the process less complete than found with the equivalent material derived from insect cells (Castells-Graells, 2019).
To investigate potential differences between plant and insect cell-derived procapsids, genes encoding the identical amino acid sequence of the a-peptide were expressed in N. benthamiana leaves and insect cells using plasmids pEAQ-HT-NxV and pFastBac-NxV, respectively (Agrawal and Johnson, 1995;Berardi et al., 2020;Castells-Graells, 2019); in both cases, VLPs were extracted in the procapsid form as previously described. For the plant-derived sample, procapsids were separated from mature capsids by sucrose gradient centrifugation, since some maturation occurs within the cells over time. While the procapsids isolated from insect cells contained almost exclusively the monomeric a-peptide, the plant-derived material contained additional higher molecular mass bands which appear to be multimers of the protein (Figure 1a). This is consistent with the previous identification of dimers by mass spectrometry and Western blotting in samples of plant-expressed NxV VLPs (Castells-Graells, 2019). Since identical conditions were used to extract and purify the procapsids and the denaturing conditions used for the SDS-PAGE analysis were the same in each case, the formation of the oligomers must be a specific consequence of using plants for expression. Given the fact that the oligomers resisted denaturation, it is probable that they result from covalent cross-linking of subunits.
To investigate whether cross-linking occurred within the plant cells or during extraction, we incubated insect cell-produced NxV procapsids with plant extracts prepared in the same buffer used to extract procapsids (50 mM Tris-HCl pH 7.6, 250 mM NaCl) and in which cleavage of the a-peptide does not occur. The conditions of the incubation (4 h at 4°C) were approximately the same as those used to prepare procapsids up to the sucrose gradient step. Western blot analysis (Figure 1b) showed that incubation of insect cell-produced procapsids in plant extracts results in the appearance of dimers and additional high molecular bands (lanes 2 to 4), not seen when the procapsids were incubated in buffer alone (lane 1). This pattern of higher bands is similar to that found in plant tissue expressing the a-peptide (lane 8), though in this sample maturation products are also seen since the procapsids were not purified. The formation of dimers occurred irrespective of whether the insect cell procapsids were incubated with extracts prepared from uninfiltrated leaves ( procapsids did not show any NxV-specific bands demonstrating the specificity of the antiserum. These results strongly imply that some component present in the plant extracts, such as peroxidases, induces the cross-linking of the NxV coat proteins and that this cross-linking occurs during extraction and purification. Varying the conditions of extraction using, for example, buffers at different pHs (5.0, 7.6, 10.0) or adding DTT to 1mM did not make a detectable difference to the result. The observation of cross-linking has implications for the production of proteins in plants. In the case of NxV, we propose that such cross-linking is detrimental to the protein rearrangements necessary for the efficient in vitro maturation of the procapsids, thus affecting the final assembly. We conclude that it is the time taken for the isolation and purification of the procapsids (approximately 4 h) that allows the cross-linking to occur, thereby interfering with the maturation process in vitro; however, cross-linking within cells prior to extraction might also occur. Cross-linking can still occur post-maturation, but this does not affect the structure of the particles.
Though cross-linking is clearly deleterious for the maturation of NxV capsids, in certain cases it may not be harmful or even beneficial. For example, in the case of plant-produced antibodies, the presence of aggregates increased the IC 50 of preparations several fold compared to CHO-produced material (Ramessar et al., 2008;Rosenberg et al., 2013). Likewise, partial cross-linking of subunits can also aid the stability and structural integrity of VLPs (Peyret et al., 2015); indeed, VLPs and virus particles intended for vaccine purposes are often deliberately cross-linked during formulation. Nonetheless, it is important to be aware of cross-linking in plant-produced material so that methods for its control or elimination can be developed. If the enzymes involved can be identified, cross-linking could be addressed by removing them, using genetic engineering techniques like CRISPR-Cas9 (Belhaj et al., 2015). Controlling cross-linking should increase interest using plants for the production of pharmaceuticals and other relevant biological products (Lomonossoff and D'Aoust, 2016). In lanes 1 to 4, NxV VLPs produced in insect cells were mixed with pH 7.6 buffer (1) or with different plant extracts (2, 3 and 4). Lanes 5 to 7 represent the corresponding plant extracts used in lanes 2 to 4 but without the added VLPs. Lane 8 contains an extract from plants infiltrated with pEAQ-HT-NxV-WT. The plant extract in lanes 2 and 5 was from uninfiltrated plant leaves, in lanes 3 and 6, from plants infiltrated with pEAQ-HT-EV and for lanes 4 and 7, from plants infiltrated with pEAQ-HT-GFP. In all cases, the plant material was blended with the pH 7.6 buffer and the immunodetection was with a polyclonal antibody for the NxV coat protein. M = SeeBlue Plus 2 pre-stained protein standards.