Plant virus‐derived nanoparticles decorated with genetically encoded SARS‐CoV‐2 nanobodies display enhanced neutralizing activity

Summary

Viral nanoparticles (VNPs) are a new class of virus‐based formulations that can be used as building blocks to implement a variety of functions of potential interest in biotechnology and nanomedicine. Viral coat proteins (CP) that exhibit self‐assembly properties are particularly appropriate for displaying antigens and antibodies, by generating multivalent VNPs with therapeutic and diagnostic potential. Here, we developed genetically encoded multivalent VNPs derived from two filamentous plant viruses, potato virus X (PVX) and tobacco etch virus (TEV), which were efficiently and inexpensively produced in the biofactory Nicotiana benthamiana plant. PVX and TEV‐derived VNPs were decorated with two different nanobodies recognizing two different regions of the receptor‐binding domain (RBD) of the SARS‐CoV‐2 Spike protein. The addition of different picornavirus 2A ribosomal skipping peptides between the nanobody and the CP allowed for modulating the degree of VNP decoration. Nanobody‐decorated VNPs purified from N. benthamiana tissues successfully recognized the RBD antigen in enzyme‐linked immunosorbent assays and showed efficient neutralization activity against pseudoviruses carrying the Spike protein. Interestingly, multivalent PVX and TEV‐derived VNPs exhibited a neutralizing activity approximately one order of magnitude higher than the corresponding nanobody in a dimeric format. These properties, combined with the ability to produce VNP cocktails in the same N. benthamiana plant based on synergistic infection of the parent PVX and TEV, make these green nanomaterials an attractive alternative to standard antibodies for multiple applications in diagnosis and therapeutics.

Introduction

Nanotechnology is a rapidly expanding research area focused on the utilization of nanoscale materials for a broad range of applications. Numerous platforms have been developed to produce nanoparticles, ranging from chemical synthesis to repurposing bionanomaterials such as those derived from viruses, known as viral nanoparticles (VNPs). VNPs can be used as building blocks for novel materials to support a range of functions of potential interest in biotechnology and nanomedicine, including vaccine and antibody platforms, targeted bioimaging or drug delivery. VNPs exhibit outstanding structural characteristics and easy functionalization: they self‐assemble with precise symmetry and polyvalency, are stable under a wide range of conditions, are biocompatible and biodegradable and can be genetically functionalized (Chung et al., 2020; Rybicki, 2020; Sainsbury and Steinmetz, 2023). Viral capsids are ideal scaffolds for the presentation of epitopes by fusing them at an appropriate site on the selected viral coat protein (CP). Attempts to display foreign proteins fused to all copies of the CP usually result in defective particles or in significantly reduced yields. However, several works have shown that this effect can be alleviated by incorporating wild‐type subunits in order to produce mosaic particles (Castells‐Graells et al., 2018; Lee et al., 2014). This can be achieved by including the well‐known, short picornavirus 2A peptide sequence between the fused polypeptides, which results in a mixed population of wild‐type and modified proteins via a cotranslational ribosomal skip mechanism (Kim et al., 2011).

Molecular farming refers to the recombinant production in plants of compounds of interest, including pharmaceuticals (Chung et al., 2022; Hefferon, 2019). Plants, which only require sunlight, water, atmospheric carbon dioxide and some inorganic salts to grow, can be considered low‐cost bioreactors. Production in plants is easily scaled up and down, since each single plant can be considered an individual bioreactor, and their products are free of human and livestock pathogens. In addition, plant cells contain endo‐membrane systems that mediate post‐translational modifications similar to those of mammalian cells (Margolin et al., 2020). A growing number of plant‐made pharmaceuticals are in clinical trials, and some of them have reached the market (Eidenberger et al., 2023; Fischer and Buyel, 2020; Schillberg and Finnern, 2021).

Viral nanoparticles derived from several plant viruses, both spherical and rod‐shaped, have already been successfully developed as scaffolds to support the display of peptides genetically encoded or chemically conjugated to structural viral proteins. With the aim of developing novel recombinant vaccines or diagnostic reagents, research in this area has mainly focused on the development of plant viruses carrying antigenic epitopes from human or animal pathogens (Chung et al., 2022; Peyret et al., 2021). This way, the activity of poorly immunogenic antigens can be boosted by the intrinsic capacity of VNPs to induce humoral and cellular immunity. Some of the most widely used plant viruses for nanotechnology developments have been cowpea mosaic virus (CPMV) (Beatty and Lewis, 2019; Ortega‐Rivera et al., 2021) and tobacco mosaic virus (TMV) (Lomonossoff and Wege, 2018; Wu et al., 2022), which have icosahedral and rod virion morphologies, respectively. However, the identification of alternative viral species that may allow for larger recombinant loads and an expanded host range of plant bioreactors is of interest. In particular, the larger surface area of filamentous viruses could provide more potential binding and acceptor sites for functionalization compared to spherical particles (Le et al., 2019).

Potato virus X (PVX), the type member of the genus Potexvirus, is a promising tool for epitope presentation (Le et al., 2019; Röder et al., 2019; Shukla et al., 2020). PVX particles are flexible rods of 500 nm consisting of 1270 identical 25‐kDa CP subunits. This virus consists of a plus‐sense, single‐stranded RNA genome of 6.4 kb in length that encodes five proteins (Verchot‐Lubicz et al., 2007). The last open reading frame (ORF) at the 3′ end of the genome encodes the CP. The N‐terminal portion of the CP is exposed on the virion surface, making amino‐terminal fusions to the CP the ideal strategy for epitope presentation. PVX particles displaying different peptides of immunological interest on the surface have been produced in plants, improving its capacity as subunit vaccine candidates (Röder et al., 2019). On the other hand, tobacco etch virus (TEV), which belongs to the genus Potyvirus, is a flexuous elongated virus, with a length of about 750 nm and a plus‐sense, single‐stranded RNA genome of 9.5 kb (Wylie et al., 2017). Its genome encodes a single polyprotein that is processed into a series of mature proteins. Some potyviruses have already been used as nanoscaffolds for peptide presentation, not only by generating recombinant viruses encoding the foreign peptide in fusion to the viral CP, but also by producing virus like particles (VLPs) from chimeric CPs expressed in heterologous systems (Martínez‐Turiño and García, 2020).

Severe acute respiratory syndrome coronavirus 2 (SARS‐CoV‐2) is responsible for a devastating recent human pandemic (COVID‐19). The rapid spread of SARS‐CoV‐2 has generated a large demand for diagnostic and therapeutic reagents and plants have been proposed as potential manufacturing platforms (Capell et al., 2020; Lico et al., 2020; Lobato Gómez et al., 2021; Tusé et al., 2020). The mature SARS‐CoV‐2 viral particle is formed by four structural proteins, with the receptor‐binding domain (RBD) of the Spike (S) protein being one of the principal immunodominant antigens in infected patients (Harvey et al., 2021). With a growing worldwide demand, monoclonal antibodies represent a major class of biopharmaceutical products for therapeutic and diagnostic applications. Among alternative expression systems, plants are promising bioreactors for the large‐scale production of recombinant antibodies for therapeutic, prophylactic and diagnostic applications (Chen, 2022; Donini and Marusic, 2019). In addition to full‐size monoclonal antibodies, smaller antibody fragments capable of antigen binding are also actively studied and employed in medicine and research (Julve Parreño et al., 2018; Malaquias et al., 2021; Wang et al., 2021). An interesting alternative to monoclonal antibodies are nanobodies or variable domain of heavy‐chain antibodies (VHHs). VHHs are derived from animals belonging to the family Camelidae and have a number of properties that make them attractive therapeutic molecules, such as their small size and simple structure, which do not compromise their high affinity and specificity (Mitchell and Colwell, 2018; Muyldermans, 2013). Interestingly, engineering VHHs as multimeric constructs has been shown to enhance their target affinity by means of an increased avidity (Xiang et al., 2020; Zupancic et al., 2021).

In this study, we report the production of genetically encoded VNPs derived from PVX and TEV that are highly decorated with neutralizing nanobodies targeting SARS‐CoV‐2 RBD. This was achieved by using different picornavirus 2A peptides to modulate the degree of nanobody decoration on the VNP surface. Correct assembly of recombinant VNPs was confirmed by immuno‐electron microscopy (IEM) and the ability to efficiently bind the corresponding antigen was demonstrated by enzyme‐linked immunosorbent assay (ELISA). The plant‐made recombinant VNPs completely neutralized SARS‐CoV‐2 pseudoviruses in cell culture infection assays, demonstrating their functionality. Notably, pseudovirus neutralization activity of the multivalent VNPs was substantially higher than that of the constituent nanobodies in a standard divalent formulation, highlighting the utility of VNPs as therapeutic and diagnostic scaffolds.

Results

Plant‐based production of PVX‐derived nanoparticles decorated with nanobodies against SARS‐CoV‐2

Based on the outstanding attributes of VNPs that make them ideal scaffolds for peptide presentation, we first aimed to combine their multimeric nature with the structural simplicity of nanobodies to develop multivalent nanoparticles able to bind the RBD domain of SARS‐CoV‐2. For this purpose, we selected the previously described nanobody VHH‐72 (Wrapp et al., 2020), hereafter named VHH1. Starting from a PVX infectious clone, the cDNA coding for VHH1 was inserted at the amino terminus end of PVX CP (Figure 1a); VHH1 was tagged with a Flag epitope at its carboxyl terminus to facilitate detection. Anticipating possible limitations on the infectivity of the resulting recombinant PVX clone (PVX‐VHH1), we also built derivative clones in which several picornavirus ribosome skipping domains, namely P2A, E2A and F2A (Kim et al., 2011), were inserted between the nanobody and the CP to produce partially decorated VNPs (PVX‐VHH1‐P2A, PVX‐VHH1‐E2A and PVX‐VHH1‐F2A; Figure 1a). The sequences of all recombinant viruses in this work is in Figures S1 and S2. N. benthamiana plants were agroinoculated with the series of PVX recombinant clones and the wild‐type PVX (PVX‐wt) as a control. The upper leaves of plants inoculated with PVX‐wt showed typical symptoms of infection at 7 days post‐inoculation (dpi). Similar symptoms were observed in plants inoculated with PVX‐VHH1‐P2A, PVX‐VHH1‐E2A and PVX‐VHH1‐F2A at 10 dpi, while plants inoculated with PVX‐VHH1 showed no apparent symptoms of infection (Figure 1b). RT‐PCR amplification followed by gel electrophoresis separation was used to analyse PVX progeny in systemic symptomatic leaves at 10 dpi. An RT‐PCR product likely corresponding to wild‐type CP PVX (CP_PVX; 208 bp) was amplified from plants inoculated with PVX‐wt (Figure 1c, lane 1), while plants inoculated with PVX‐VHH1‐P2A, PVX‐VHH1‐E2A and PVX‐VHH1‐F2A produced PCR products whose positions matched those expected for recombinant clones maintaining the VHH1 insert (745 bp; Figure 1c, lanes 3–5, respectively). Despite the absence of symptoms, a slight band corresponding to an amplification product of approximately 700 bp was also observed for the PVX‐VHH1 sample (Figure 1c, lane 2). These results suggest a defect in the accumulation of PVX‐VHH1, but also indicate efficient infectivity, stability and accumulation of the PVX recombinant clones that contain the different 2A peptides.

Figure 1.

Production of PVX‐derived VNPs decorated with an anti‐SARS‐CoV‐2 RBD nanobody. (a) Schematic representation of the PVX genome indicating the position where a heterologous sequence coding for an anti‐RBD nanobody (VHH1) was inserted, along with different picornavirus 2A peptides (P2A, E2A and F2A). (b) Pictures of upper leaves from representative plants mock‐inoculated or agroinoculated with different recombinant viruses, as indicated, taken at 14 dpi. (c) Plant tissues were collected at 14 dpi and the presence of the recombinant PVX was analysed by RT‐PCR with oligonucleotide primers indicated with arrows in (a). Amplification products were separated via electrophoresis in an agarose gel. Lanes 1–5, plants agroinoculated with PVX‐wt, PVX‐VHH1, PVX‐VHH1‐P2A, PVX‐VHH1‐E2A and PVX‐VHH1‐F2A, respectively; lane 6, mock‐inoculated plant. (d and e) Western blot analyses of protein extracts from infected tissues using an antibody against (d) CP_PVX or (e) the Flag epitope. Lanes 1–5, plants agroinoculated with PVX‐wt, PVX‐VHH1, PVX‐VHH1‐P2A, PVX‐VHH1‐E2A and PVX‐VHH1‐F2A, respectively. The positions and sizes of protein standards are indicated on the left. Black arrows indicate the positions of the VHH1‐F2A‐CP_PVX fusion, free CP_PVX and free VHH1. (f) The ratio between VHH1‐2A‐CP_PVX and free CP_PVX was quantified using the ImageJ software and plotted.

Next, the accumulation of free and VHH‐fused CP_PVX in infected tissues was investigated by western blot analysis with anti‐CP_PVX and anti‐Flag polyclonal and monoclonal antibodies, respectively. Reaction with the anti‐CP_PVX antibody produced a single band with the expected position (25 kDa) in the lane corresponding to a plant inoculated with PVX‐wt (Figure 1d, lane 1). No signal was detected for PVX‐VHH1 (Figure 1d, lane 2), while lanes corresponding to plants inoculated with PVX‐VHH1‐P2A, PVX‐VHH1‐E2A and PVX‐VHH1‐F2A exhibited bands at the expected sizes of free CP_PVX (25 kDa) and the VHH1‐2A‐CP_PVX fusion (44 kDa) (Figure 1d, lanes 3–5, respectively). Other bands were observed, most notably a series above the 44 kDa CP_PVX fusions, which we attribute to CP dimers. In these samples, the amount of CP_PVX decreased inversely to the amounts of the VHH1‐2A‐CP_PVX fusions, following the ribosome skipping activity P2A > E2A > F2A (Kim et al., 2011). Reaction with the anti‐Flag antibody exclusively produced positive signals in the lanes corresponding to the plants inoculated with PVX‐VHH1‐P2A, PVX‐VHH1‐E2A and PVX‐VHH1‐F2A (Figure 1e, lanes 3–5, respectively). The amounts of the VHH1‐2A‐CP_PVX fusions again inversely correlated with the 2A ribosome skipping activity P2A > E2A > F2A. Together, these results support that PVX recombinant clones with a CP_PVX‐fused VHH1 only accumulate efficiently in N. benthamiana when a 2A skipping peptide is inserted. Remarkably, the use of 2A peptides with different activities in vivo allows to modulate the rate of VNP decoration (Figure 1f). For further work, we selected the F2A peptide because of the optimal balance between VNP yield and degree of decoration.

Subsequently, we sought to extend the results obtained above to a second RBD‐binding nanobody and to purify and analyse the correct assembly of the recombinant VNPs. For this, we chose the Ty1 nanobody (Hanke et al., 2020), hereafter named VHH2 (Figure S3), which targets a different RBD epitope than VHH1. A cDNA encoding VHH2 was inserted in place of the VHH1 sequence in the PVX infectious clone to produce PVX‐VHH2‐F2A (Figure 2a). N. benthamiana plants were agroinoculated with PVX‐VHH1‐F2A or PVX‐VHH2‐F2A, as well as PVX‐wt, as a control. Systemic infection symptoms were observed at 10 dpi (Figure 2b, top). Accumulation of both PVX recombinant clones in symptomatic upper leaves was confirmed by RT‐PCR with primers flanking the insertion site and by western blot with an anti‐Flag antibody (Figure 2b, bottom). VNPs were extracted from infected tissues and fractionated by centrifugation in a sucrose gradient. Western blot analysis of the fractions showed efficient separation of the free nanobody molecules on the light gradient fractions and the assembled nanobody‐containing VNPs in the heavy fractions (Figure 2c). VNP‐containing fractions were pooled, dialyzed and concentrated using centrifugal filter units. VNPs were finally analysed by SDS‐polyacrylamide gel electrophoresis (PAGE) followed by Coomassie brilliant blue staining of the gel. The results confirmed the presence of free CP_PVX along with VHH‐fused CP_PVX in the purified samples (Figure 2d). Based on the intensity of bands arising from protein standards, yields of approximately 100 μg of VHH‐2A‐CP per gram of plant tissue (fresh weight) was estimated for both VNPs. IEM was next used to confirm the correct assembly of the purified VNPs. For this analysis we used a primary anti‐Flag antibody followed by a secondary anti‐rabbit antibody conjugated to 20‐nm gold spheres. Three representative images of the two VNP preparations (PVX‐VHH1‐F2A and PVX‐VHH2‐F2A) as well as PVX‐wt negative controls are shown in Figure 2e. The images revealed the typical PVX filamentous particles homogeneously decorated with gold spheres, which were absent in the PVX‐wt controls. All together, these results support the successful production of PVX‐derived VNPs partially decorated with VHH1 or VHH2 in N. benthamiana plants when the picornavirus F2A peptide is used.

Figure 2.

Purification of PVX‐derived VNPs presenting nanobodies against two different epitopes of SARS‐CoV‐2 RBD. (a) Two different VHH targeting distinct RBD epitopes were cloned in the PVX vector fused to the CP using the F2A skipping peptide and were produced by agroinoculation of N. benthamiana plants. (b) Top, pictures of upper leaves from representative plants mock‐inoculated or agroinoculated with different recombinant viruses as indicated, taken at 14 dpi. Bottom, RT‐PCR and western blot analyses of infected plants, as indicated. (c) Schematic representation of the VNP purification process. Symptomatic tissue was collected and homogenized in extraction buffer. The extract was subjected to ultracentrifugation on a sucrose gradient. Positive fractions, according to a western assay, were pooled, dialyzed against PBS, concentrated and analysed by SDS‐PAGE and IEM. (d) Purified VNPs were analysed by SDS‐PAGE and the gel stained with Coomassie brilliant blue. The ratios between VHH‐F2A‐CP_PVX and free CP_PVX were quantified using a BSA standard curve and the ImageJ software. Lanes 1–3, VNPs purified from tissues inoculated with PVX‐F2A‐VHH1, PVX‐F2A‐VHH2 and PVX‐wt, respectively. (e) IEM analysis of purified VNPs. The presence of the VHHs presented on the surface was revealed with a primary anti‐Flag antibody and a secondary anti‐rabbit antibody conjugated to 20‐nm gold particles. Three different images are shown for each VNP preparation. A PVX‐wt control is also included.

Plant‐based production of TEV nanoparticles decorated with nanobodies

We next explored whether the same concept could be extended to TEV, another filamentous virus, which is a member of the largest family (Potyviridae) of RNA viruses infecting plants. A cDNA coding for VHH1 was inserted in a TEV infectious clone to produce an amino‐terminal fusion with TEV CP (CP_TEV). Again, VHH1 was tagged at the carboxyl terminus, but in this case the E epitope was used to facilitate selective detection (TEV‐VHH1; Figure 3a; Figure S2). As before, we also built derivative clones with the different picornavirus 2A peptides (TEV‐VHH1‐P2A, TEV‐VHH1‐E2A and TEV‐VHH1‐F2A; Figure 3a). N. benthamiana plants were agroinoculated with wild‐type TEV (TEV‐wt) and the different TEV recombinant clones. Upper leaves of plants inoculated with TEV‐wt showed typical symptoms of infection at 7 dpi, an observation extended to the recombinant clones at 10 dpi. Notably, TEV‐VHH1 carrying the nanobody directly fused to the CP_TEV with no 2A skipping peptide also produced symptoms of infection (Figure 3b). Tissues from symptomatic upper leaves were collected at 10 dpi and the presence of the heterologous sequences in the viral progenies was evaluated by RT‐PCR amplification followed by electrophoretic analysis. As expected, a 501‐bp product was amplified from plants inoculated with TEV‐wt, which corresponded to the full‐length CP_TEV, while no product was obtained from mock‐inoculated plants (Figure 3c, lanes 1 and 2). Plants inoculated with the four TEV recombinant clones harbouring the VHH1 moiety produced PCR products whose positions matched those expected for recombinant clones containing the heterologous sequences, with or without 2A peptides (1054 and 984 bp, respectively; Figure 3c, lanes 3–6). We next analysed the accumulation of CP_TEV and the VHH1‐fused derivatives by western blot, using anti‐CP_TEV and anti‐E‐tag polyclonal and monoclonal antibodies, respectively. For plants inoculated with TEV‐wt, anti‐CP_TEV antibody produced a band corresponding to the expected size (30 kDa) (Figure 3d, lane 1, lower black arrow), while the plant inoculated with TEV‐VHH1 showed a band shifting to the upper part of the gel (49 kDa) (Figure 3d, lane 5, upper black arrow). In the case of the three viral clones carrying the 2A peptides, bands corresponding to the fused VHH1‐2A‐CP_TEV and the free CP_TEV were detected (Figure 3d, lanes 2–4). As previously observed for PVX clones, the amount of free CP_TEV decreased inversely to the amount of fused VHH1‐2A‐CP_TEV, following the known ribosome skipping activity of the peptides (P2A > E2A > F2A). Reaction with the anti‐E antibody detected bands arising from proteins of the expected size for the VHH1‐2A‐CP_TEV fusion in the lanes corresponding to plants inoculated with TEV‐VHH1‐E2A and TEV‐VHH1‐F2A (Figure 3e, lanes 2–4, upper black arrow) and VHH1‐CP_TEV for the plant inoculated with TEV‐VHH1 (Figure 3e, lane 5, upper black arrow). Again, the ratio of decorated over free CP_TEV resulted in the expected skipping efficiency of the 2A sequences (F2A > E2A > P2A; Figure 3f). Together, these findings indicate that TEV represents an additional platform to produce nanobody‐decorated VNPs in plants. Notably, TEV allowed direct fusion of VHH1 to CP_TEV and consequent production of fully decorated VNPs.

Figure 3.

Production of TEV‐derived VNPs decorated with an anti‐RBD nanobody. (a) Schematic representation of the TEV genome indicating the position where a heterologous sequence coding for an anti‐RBD nanobody (VHH1) was inserted, along with different picornavirus 2A peptides. Recombinant viruses were agroinoculated in N. benthamiana plants. (b) Pictures of upper leaves from representative plants mock‐inoculated or agroinoculated with different viral vectors as indicated, taken at 14 dpi. (c) Plant infected tissues were collected at 14 dpi and the presence of the TEV progeny analysed by RT‐PCR with oligonucleotide primers indicated with arrows in (a). Amplification products were separated via electrophoresis in an agarose gel. Lane 1, mock‐inoculated plant; lanes 2–6, plants agroinoculated with TEV‐wt, TEV‐VHH1‐P2A, TEV‐VHH1‐E2A, TEV‐VHH1‐F2A and TEV‐VHH1, respectively. (d and e) Western blot analyses of protein extracts using an antibody against (d) CP_TEV or (e) an antibody against E‐tag epitope. Lanes 1–5, plants agroinoculated with TEV‐wt, TEV‐VHH1‐P2A, TEV‐VHH1‐E2A, TEV‐VHH1‐F2A and TEV‐VHH1, respectively. The positions and sizes of protein standards are indicated on the left. Black arrows indicate the positions of the VHH1‐F2A‐CP_TEV fusion, free CP_TEV and free VHH1. (f) The ratio between VHH1‐2A‐CP_TEV and free CP_TEV was quantified using the ImageJ software and plotted.

We then assayed the production of the second nanobody (VHH2) using TEV as a scaffold. A cDNA encoding VHH2 was inserted in the same position as VHH1 obtaining TEV‐VHH2. Strikingly, when we agroinoculated plants with this recombinant clone, they did not develop symptoms of infection nor accumulated of CP_TEV according to a western blot analysis. However, the insertion of the F2A peptide in between VHH2 and the amino‐terminal end of CP_TEV restored the infection capacity of the viral clone (TEV‐VHH2‐F2A; Figure 4a). This result suggests that the viability of fully nanobody‐decorated TEV particles is not general and depends on the specific nanobody. Based on these results, in order to purify and analyse the TEV‐derived VNPs, we agroinoculated plants with TEV‐VHH1 or TEV‐VHH2‐F2A, as well as with TEV‐wt as a control. Symptoms of infection were observed at 10 dpi (Figure 4b, top). Symptomatic upper leaves were analysed by RT‐PCR with primers flanking the insertion site and by western blot with anti‐CP_TEV or with anti‐E‐tag antibodies. Results confirmed the presence of the heterologous sequences in the viral progeny and the accumulation of the expected VHH fusion products (Figure 4b, bottom). Next, VNPs were purified from infected tissues and fractionated in a sucrose gradient (Figure 4c). Purified VNPs were analysed by SDS‐PAGE followed by Coomassie brilliant blue staining of the gel in order to quantify the yield and the ratio of VHH‐fused CP_TEV over free CP_TEV. This analysis confirmed the presence of the single VHH1‐CP_TEV fusion in plants inoculated with TEV‐VHH1 and the coexistence of free CP_TEV along with VHH2‐F2A‐CP_TEV in plants inoculated with TEV‐VHH2‐F2A (Figure 4d, lanes 2 and 3, respectively). VHH1‐CP_TEV and VHH2‐F2A‐CP_TEV accumulation was estimated in approximately 25 and 50 μg per gram of fresh tissue, respectively. Next, in order to confirm the correct assembly of the corresponding VNPs, viral particles were analysed by IEM using a primary anti‐E antibody followed by a secondary anti‐mouse antibody conjugated to 10‐nm gold spheres. Micrographs showed the typical filamentous and flexuous TEV virions homogeneously surrounded of gold particles in the case of TEV‐VHH1 and TEV‐VHH2‐F2A VNPs. Gold staining was absent in the TEV‐wt control subjected to the same treatment (Figure 4e). All together, these results support the successful production of TEV‐derived VNPs completely decorated with VHH1 or partially decorated with VHH2 in N. benthamiana plants.

Figure 4.

Purification of TEV derived VNPs presenting VHHs against two different epitopes of RBD. (a) Two different VHH targeting distinct RBD epitopes were cloned in the TEV vector: VHH1 fused directly to the CP_TEV and VHH2 using the F2A skipping peptide. There were then produced by agroinoculation of N. benthamiana plants. (b) Top, pictures of upper leaves from representative plants mock‐inoculated or agroinoculated with different viral vectors as indicated, taken at 14 dpi. Bottom, RT‐PCR and Western blot analyses of infected tissues, as indicated. (c) Western blot analysis of TEV‐derived VNPs using an anti‐E antibody. The VNP preparation from plants inoculated with TEV‐VHH1‐F2A was separated by centrifugation on a sucrose gradient. (d) Purified VNP preparations were analysed by SDS‐PAGE. The ratio between VHH‐CP_TEV and free CP_TEV were quantified by Coomassie brilliant blue staining using a BSA standard curve and the ImageJ software. Lanes 1–3, VNPs purified from plants inoculated with TEV‐wt, TEV‐VHH1 and TEV‐F2A‐VHH2, respectively. (e) Purified VNPs were analysed by TEM and the presence of VHH presented on their surface was detected by IEM with a primary anti‐E antibody and a secondary anti‐mouse antibody conjugated to 10‐nm gold particles. Three different images are shown for each viral purification. Images of a TEV‐wt negative control are also included.

Functional analysis of nanobody‐decorated VNPs

We next aimed to analyse the functionality of the different nanobody‐decorated VNPs. For this purpose, we performed an ELISA against a recombinant version of SARS‐CoV‐2 RBD polypeptide (Figure 5a). Plates coated with RBD antigen were incubated with serial dilutions of purified PVX or TEV‐derived VNPs presenting the VHH1 or VHH2 nanobodies, as well as with serial dilutions of the corresponding wild‐type virions as negative controls. Specific VNP binding was detected with anti‐CP_PVX or anti‐CP_TEV polyclonal antibodies conjugated to alkaline phosphatase. The recombinant RBD polypeptide was successfully detected with both PVX‐ and TEV‐derived VNPs decorated with either VHH1 or VHH2 with similar efficiency, according to the dilution assay, thereby demonstrating that all VHHs retained their antigen‐binding activities (Figure 5b,c).

Figure 5.

Functionality of the multivalent VNPs decorated with VHHs on the surface. (a) Schematic representation of the ELISA to detect RBD. (b and c) Plates (96‐well) were coated with 100 ng/well of recombinant RBD, incubated with serial dilutions of PVX or TEV‐derived VNPs, as indicated, extensively washed and finally revealed using a (b) HRP‐conjugated anti‐CP_PVX or a (c) HRP‐conjugated anti‐CP_TEV. Bars represent the means and standard deviations for triplicate samples. Data were analysed by two‐way ANOVA with Tukey's post‐test (***P < 0.001; **P < 0.01). (d) The neutralization capacity of the VNPs against SARS‐CoV‐2 RBD domain was assessed using a GFP‐expressing VSV pseudotyped with the Wuhan‐Hu‐1 S protein. Pseudoviruses were pre‐incubated with serial dilutions of VNPs and used to infect a monolayer of Vero‐E6‐TMPRSS2 cells seeded in 96 well plates. At 16 h post‐inoculation, GFP expression in each well was quantified using a live‐cell microscope as a proxy of virus infectivity. (e) Neutralization assay with the PVX‐derived VNPs expressing VHH1 or VHH2. (f) Neutralization assay with the TEV‐derived VNPs expressing VHH1 or VHH2. Bars represent the means and standard deviations for triplicate samples. Data were analysed by two‐way ANOVA with Tukey's post‐test (***P < 0.001; **P < 0.01).

Going a step further, we assayed the ability of the VNPs decorated with the VHHs to neutralize viral entry into cells. For this, we performed a cell culture neutralization assay using a green fluorescent protein (GFP)‐expressing vesicular stomatitis virus (VSV) pseudotyped with the Wuhan‐Hu‐1 S protein, which have been extensively used to evaluate neutralizing antibodies to different SARS‐CoV‐2 S variants and show good correlation with neutralization of the real virus (Schmidt et al., 2020). Pseudoviruses were pre‐incubated with serial dilutions of the individual VNPs and then added to the cell monolayer in 96 well plates. Following a 16‐h incubation, the GFP signal in each well was quantified as a proxy of infection using a live‐cell microscope (Figure 5d). Wild‐type virions as well as an extract from mock‐inoculated plants subjected to the same purification process that VNPs were used as negative controls. For all four VHH‐decorated VNPs, results showed a concentration‐dependent decrease in GFP fluorescence. Almost complete pseudovirus neutralization was observed at the highest VNPs concentrations (Figure 5e,f). The estimated IC₅₀ for the PVX‐derived VNPs was approximately 0.3 μg/mL for both VHHs, while for TEV‐derived VNPs the estimated IC₅₀ was 0.4 μg/mL for the fully decorated VHH1 and 1.5 μg/mL for the VHH2 partially decorated particles. Taken together, these results support that both multivalent TEV and PVX‐derived VNPs, completely or partially decorated with VHHs, are able to bind the RBD antigen and to inhibit the interaction with its cellular receptor.

Multivalent VNPs show increased neutralizing activity versus free VHHs

Because artificially linked nanobodies in the format of dimers or trimers increase their avidity for the antigens (Zupancic et al., 2021), we wondered whether the multivalent VNPs, carrying hundreds of VHH units, display improved neutralization activity when compared with the classic benchmark of a dimeric VHH. To evaluate this, we performed pseudovirus neutralization assays comparing the partially decorated PVX‐VHH1‐F2A (estimated 650 binding sites per VNP) and the fully decorated TEV‐VHH1 (estimated 2000 binding sites per VNP) with a plant‐produced VHH1 fused to the human IgG1 Fc dimerization domain (Diego‐Martin et al., 2020). We performed the neutralization assays with the same molarity of VHH1 binding sites, that is the same number of total binding domains regardless of their distribution. For this purpose, the protein composition and concentration of each sample was analysed by SDS‐PAGE and Coomassie brilliant blue staining (Figure 6a). Pseudovirus neutralization activities of PVX‐VHH1‐F2A and TEV‐VHH1 were substantially superior to that of dimeric VHH1. The partially decorated PVX VNPs exhibited slightly better activity than the fully decorated TEV VNPs. Based on IC₈₀ estimation, both types of VNPs were at least 10‐fold more effective neutralizing pseudoviruses than the dimeric VHH counterpart (Figure 6b).

Figure 6.

Pseudovirus neutralization activity of multivalent VNPs. (a) SDS‐PAGE followed by Coomassie brilliant blue staining analysis of purified VHH‐FcIgG dimers, PVX VNPs and TEV VNPs. Concentration of each VHH sample was measured comparing with a BSA standard curve using the Image J software. (b) Schematic representation of the different VHH formats; from left to right, VVH‐FcIgG dimers, PVX‐derived VNPs of ~500 nm length composed of ~1270 CP subunits partially decorated with VHH1 and TEV‐derived VNPs of ~730 nm composed of ~2000 CP subunits fully decorated with VHH1. (c) Neutralization assay with equal molarity of VHHs. Bars represent the means and standard deviations for triplicate samples. Data were analysed by two‐way ANOVA with Tukey's post‐test (***P < 0.001; **P < 0.01).

VNPs cocktail production by synergistic viral infection

To avoid immune evasion due to punctual mutations in the pathogen genome, therapeutic strategies usually involve the use of antibody cocktails (Koenig et al., 2021; Xiang et al., 2020). These cocktails combine at least two different antibodies that, despite targeting the same pathogen, do not compete with each other to bind the antigen, leading to a stronger neutralizing activity. Our final aim was to directly produce such a cocktail in a single biofactory plant, taking advantage that PVX and TEV can synergistically co‐infect the same plant (Mascia and Gallitelli, 2016; Pruss et al., 1997). For that, N. benthamiana plants were co‐agroinoculated with the TEV‐VHH1‐F2A and PVX‐VHH2‐F2A (Figure 7a). Symptoms of systemic infection were observed at 10 dpi, and the presence of both recombinant VHH‐fused CPs was evaluated in protein extracts from infected tissues by western blot analyses with antibodies against CP_TEV (Figure 7b, top) and against CP_PVX (Figure 7b, bottom). Plants individually inoculated with each recombinant virus and plants inoculated with the wild‐type viruses were used as controls. Both VHH1‐F2A‐CP_TEV and VHH2‐F2A‐CP_PVX were detected in the upper leaves of co‐inoculated plants, confirming the successful systemic coinfection. We then co‐purified the TEV and PVX VNPs and collected sucrose gradient fractions that were positive for at least one of the viruses. Purified VNPs were analysed by SDS‐PAGE and Coomassie brilliant blue staining. Bands corresponding to free CPs and VHH‐fused CPs were observed (Figure 7c). Purified VNPs were analysed by IEM and the presence of both TEV and PVX VNPs with VHH1 or VHH2, respectively, presented on their surfaces was detected by co‐immunostaining with an anti‐E‐tag (VHH1) and an anti‐Flag (VHH2) conjugated to 10‐ and 20‐nm gold particles, respectively (Figure 7d).

Figure 7.

Production and functional analysis of a VNP cocktail with two different VHH against RBD. (a) A synergistic infection system was developed by co‐agroinoculation of N. benthamiana plants with the recombinant TEV‐VHH1‐F2A and the recombinant PVX‐VHH2‐F2A. A picture of upper leaves from a representative plant at 14 dpi is shown. (b) Western blot analyses of protein extracts from infected tissues using an antibody against CP_TEV (top) or an antibody against CP_PVX (bottom). Lanes 1–4, plants agroinoculated with TEV‐wt, TEV‐VHH1‐F2A, PVX‐wt or PVX‐VHH2‐F2A, respectively; lane 5, plants co‐agroinoculated with TEV‐VHH1‐F2A and PVX‐VHH2‐F2A. The positions and sizes of protein standards are indicated on the left. Black arrows indicate the positions of the VHH‐CP fusions and free CPs. (c) TEV and PVX‐derived VNPs mix were co‐purified and analysed by SDS‐PAGE and Coomassie brilliant blue staining. (d) Correct assembly of purified VNPs was observed by TEM negative‐staining and the presence of VHH1 or VHH2 presented on their surface was detected by co‐immunostaining with an anti‐VHH1 conjugated to 10‐nm gold particles and an anti‐VHH2 conjugated to 20‐nm gold particles. Four different representative images are shown. (e) Functionality of both recombinant VNPs on the cocktail was analysed by ELISA detected using an HRP‐conjugated anti‐CP_TEV (top) or an HRP‐conjugated anti‐CP_PVX (bottom). In both cases, the purified single TEV or PVX‐derived VNPs were used as positive controls. (f) Pseudovirus neutralization capacity of the VNP cocktail compared with the single TEV or PVX‐derived VNPs. Bars represent the means and standard deviations for triplicate samples. Data were analysed by two‐way ANOVA with Tukey's post‐test (***P < 0.001; **P < 0.01).

Functionality of the VNP cocktail was analysed by ELISA using a horseradish peroxidase (HRP)‐conjugated anti‐CP_TEV (Figure 7e, top) or an HRP‐conjugated anti‐CP_PVX (Figure 7e, bottom). Separately purified TEV‐VHH1‐F2A or PVX‐VHH2‐F2A VNPs were used as positive controls. The VNP cocktail efficiently bound SARS‐CoV‐2 RBD, and no significant difference was observed with the single TEV‐VHH1‐F2A or PVX‐VHH2‐F2A (Figure 7e). It is worth to note that the goal of the ELISA was to discard interference between both VNPs. Therefore, no sensitivity improvement can be expected, because each secondary antibody only detects one kind of VNP. Finally, pseudovirus neutralization capacity of the VNP cocktail was compared with those of the single TEV‐VHH1‐F2A or PVX‐VHH2‐F2A. The cocktail exhibited a significantly higher neutralization activity than the two components separately at the most diluted concentration assayed (Figure 7f).

Discussion

Plant virus‐derived nanoparticles have arisen interest in biotechnology and nanomedicine because, in addition to being biocompatible and biodegradable, they self‐assemble with a precise stoichiometry, can be genetically functionalized and can be produced inexpensively on a large scale in plants used as green biofactories (Chung et al., 2022; Eidenberger et al., 2023). Moreover, they are less likely to have unexpected adverse interactions with animal cells because they are derived from plant viruses (Shukla et al., 2020). With the aim of redirecting viral cell targeting toward therapeutic applications, substantial progress has been made in incorporating targeting peptides to external viral surfaces (Beatty and Lewis, 2019; Chung et al., 2020; Le et al., 2019). Several strategies for antibody display have been developed, most of which rely on chemical conjugation (Park et al., 2020; Uhde‐Holzem et al., 2016). These approaches usually require multiple steps, leading to decreased yields and increased costs. Alternative strategies for stable and scalable manufacturing of antibody‐nanoparticle conjugates directly from the plant bioreactor are, therefore, required.

In this work, we combined the multimeric self‐assembly capacity of plant viruses with the simplicity of the single chain nanobodies in order to develop genetically encoded multivalent VNPs that can be produced in large amounts in biofactory plants. The recent COVID‐19 pandemic has shown how a global pathogen outbreak entails a sudden demand for therapeutic, diagnostic and research reagents that can strain production capacities under non‐pandemic conditions. Production of some of these reagents, such as antigens and antibodies, is hampered by the high cost and limited scalability of traditional manufacturing platforms. Plant‐made recombinant proteins may contribute to a global response in such an emergency scenario, based on speed, scalability, safety and cost‐effectiveness (Capell et al., 2020; Lico et al., 2020; Lobato Gómez et al., 2021; Tusé et al., 2020). The SARS‐CoV‐2 envelope S glycoprotein contains the RBD, which binds to the extracellular domain of receptor angiotensin converting enzyme 2 (ACE2) and mediates the virus entry into host cells. RBD is, therefore, a preferred target when developing neutralizing antibodies. As a proof of concept, we selected two different previously reported neutralizing nanobodies that target the RBD domain and prevent ACE2 engagement (VHH‐72 and Ty1, renamed VHH1 and VHH2 in this work for simplicity) (Hanke et al., 2020; Wrapp et al., 2020). We aimed to produce these two nanobodies attached to VNPs in plants using two filamentous flexuous viruses, PVX and TEV, which belong to distinct families and are able to synergistically co‐infect plants. In this regard, spherical VLPs for nanobody display were previously produced in plants by means of assembling recombinant agrobacterium‐expressed fused CPs, called tandibodies (Peyret et al., 2015). In contrast, in this study, we functionalized the CP subunits to obtain flexuous rod‐shaped biomaterials for nanobody presentation via genetic fusion in the viral genome. Of note, the larger surface area of filamentous viruses provides more potential binding and acceptor sites for functionalization, compared to isometric particles (Le et al., 2019). It is worth mentioning that the fusion of nanobodies to protein scaffolds to increase their size (Megabodies and Legobodies), as well as to fluorescent proteins or enzymes to be used as reporters (RANbodies), have been described for different applications and have been shown to retain the full antigen‐binding specificity and affinity (Uchański et al., 2021; Wu and Rapoport, 2021; Yamagata and Sanes, 2018).

We first analysed the possibility of producing fully decorated genetically encoded PVX‐derived nanoparticles by the direct fusion of the nanobodies to the amino terminus of the viral CP. Nonetheless, we also assayed a series of picornavirus 2A peptides to split the nanobody and CP moieties to produce VNPs with different degrees of decoration. Previous studies showed that small peptides shorter than 15 amino acids can be directly linked genetically to the amino terminus of PVX CP, while the 2A‐based overcoating strategy enables the presentation of larger immunogenic peptides or fluorescent proteins that would otherwise interfere with assembly and function by reducing the number of CP‐insert fusions on the virion (Röder et al., 2019; Shukla et al., 2018). An early work, in which PVX particles were decorated with a functional scFv antibody fragment against the herbicide diuron, used the foot and mouth disease virus (FMDV) 2A peptide to introduce a ribosomal skip in the CP gene (Smolenska et al., 1998). A similar strategy was used here, although extended to an array of different 2A peptides to generate VNPs partially decorated with nanobodies presenting different ratios of fused CPs. Subsequent studies engineered PVX nanoparticles to present a domain from protein A, a bacterial immunoglobulin‐binding protein, as an antibody capture strategy, which enabled non‐covalent antibody display with orientational control (Uhde‐Holzem et al., 2016). More recently, a site‐specific conjugation strategy was also developed to functionalize icosahedral CPMV derived nanoparticles presenting a homogeneous display of full‐length antibodies (Park et al., 2020). On the one hand, our results indicated that direct fusion interfered with efficient recombinant virus accumulation, most probably due to steric hindrance on the correct particle assemble. On the other hand, our results also indicated successful accumulation in plant tissues of PVX‐derived VNPs decorated with the nanobodies when 2A peptides were used (Figures 1 and 2). We observed a ribosomal skipping activity gradient of P2A > E2A > F2A for the 2A peptides in our VNP context, in agreement with those previously reported in other animal and plant cells (El Amrani et al., 2004; Kim et al., 2011; Liu et al., 2017). Although in this work we selected the most decorated PVX VNPs harbouring the F2A peptide, this result opens the possibility to modulate the degree of decoration of VNPs by selecting the appropriate 2A peptide, for example when the particle assembly capacity is compromised or when the excess of exogenous peptide could have detrimental effects. A combination of fractionation in sucrose gradient, western blot and, particularly, IEM analyses demonstrated the correct assembly and the multivalent binding capacity of the PVX‐derived VNPs (Figure 2). These analyses also indicated that, for the selected PVX‐VHH1‐F2A and PVX‐VHH2‐F2A, approximately half of viral CPs contain fused nanobodies.

Second, we extended our analysis to TEV, another filamentous flexuous virus. TEV represents the largest family of plant RNA viruses with more than two hundred species currently characterized (Wylie et al., 2017), which could extend this technology to a broad range of host plants. Notably, fully decorated TEV‐derived VNPs, in which no 2A peptide was used, accumulated efficiently in infected tissues in the case of VHH1, but not for VHH2, indicating that fully decorated TEV nanoparticles can be directly produced in plants, at least for some nanobodies. Both inserted moieties were approximately the same size (19 and 18 kDa for VHH1 and VHH2, respectively). As previously observed with PVX‐derived VNPs, the use of different 2A peptides allowed to obtain different degrees of decoration (Figure 3). Insertion of F2A peptide facilitated production of nanobody‐decorated TEV VNPs in the case of VHH2 with, again, approximately half of the CPs decorated with the nanobody (Figure 4). Finally, a combination of sucrose gradient fractionation, western blot and IEM analyses demonstrated the proper assembly of the TEV‐derived VNPs and their multivalent binding capacity (Figure 4). Different potyviruses were previously used as nanoscaffolds for short antigenic peptide presentation, yielding increased immunomodulatory properties (Martínez‐Turiño and García, 2020). Most of these strategies relied on the production of VLPs by assembling chimeric CPs, generally presenting short antigenic peptides (Yuste‐Calvo et al., 2019). In a recent work, the correct expression of turnip mosaic virus (TuMV; genus Potyvirus) VLPs presenting a food allergen peptide of 90 residues was achieved by the addition of a linker between the fused proteins (Pazos‐Castro et al., 2022). Regarding genetically encoding VNPs, zucchini yellow mosaic virus (ZYMV; Potyvirus), plum pox virus (PPV; Potyvirus) and TuMV were developed to express short peptides at the N‐terminal end of CP (Martínez‐Turiño and García, 2020). TEV VNPs, with a genetically encoded nanobody against GFP, have also been recently described (Martí et al., 2022).

As an alternative to more specialized and long‐lasting PCR diagnostic tests, there is a critical need for protein‐based diagnostic reagents, particularly antibodies, for less complex tests, such as ELISA or lateral flow assays, which are faster and prone to high‐throughput and point‐of‐care formats. Hence, we assessed the suitability of all our produced VNPs as ELISA reagents for SARS‐CoV‐2 RBD detection, showing that all of them successfully detected the antigen (Figure 5). Although we did not perform a direct comparison, considering that each of our VNPs comprises hundreds of CP subunits, we hypothesize that this should significantly amplify the signal, thereby increasing the assay sensitivity compared with classical antibody assays. This idea has been recently proven using the filamentous M13 phage as a bifunctional probe with antigen recognition and signal amplification functions. A recombinant phage displaying a SARS‐CoV‐2‐specific binding peptide fused to the pIII structural protein was used for target binding, while the pVIII structural protein provided multiple binding sites (∼2700 copies) to achieve signal amplification using an anti‐M13‐HRP, establishing the so‐called sandwiched phage‐based enzyme‐linked chemiluminescence immunoassay (ELCLIA) (Liu et al., 2022).

Fourth, we analysed the capacity of our VNPs for virus neutralization and compared the multivalent nanoparticles with the benchmark dimeric format. Due to obvious restrictions inherent to working with infectious biosafety level 3 live SARS‐CoV‐2, we opted for a pseudovirus assay based on VSV pseudotyped with the SARS‐CoV‐2 glycoprotein S (Lu et al., 2021, 2023). Purified VHH‐expressing VNPs exhibited efficient pseudovirus neutralizing activity, with an IC₅₀ of approximately 0.3 μg/mL for both PVX‐derived VNPs, and IC₅₀ of 0.4 and 1.5 μg/mL for the TEV‐derived VNPs displaying VHH1 or VHH2, respectively. This difference probably arises from the F2A peptide presence that leads to less decorated particles. It is worth noting that these figures represent the amounts of VNPs and not the concentration of VHHs.

Based on their small size and lack of a light chain, nanobodies can be flexibly reshaped into multimeric versions, such as dimers or trimers, by tandem linking. In several works, engineered multivalent nanobodies were shown to exhibit enhanced binding affinity compared with the original monomeric versions (Liu et al., 2023; Wrapp et al., 2020). In this sense, the exterior surface of VNPs could work as a platform for multivalent presentation. Previous studies of the impact of multivalency on SARS‐CoV‐2 nanobody affinity and neutralization activity have only focused on relatively low valences, like dimers and trimers. It is expected that much higher valencies (hundreds) could lead to significant improvements in neutralization activity. To compare the neutralization activity of our multivalent VNPs, we used a previously reported version of VHH1 dimerized through fusion to the human IgG1 Fc domain (Diego‐Martin et al., 2020). Neutralization activity of PVX‐VHH1‐F2A and TEV‐VHH1 was approximately one order of magnitude higher than a comparable amount of the dimeric VHH1 (Figure 6). Importantly, in this assay, neutralization activity was evaluated as a function of the concentration of binding domains and not from the concentration of the multivalent complexes. Therefore, the observed increase in neutralization activity results from a synergistic improvement in the avidity of the nanoparticles and not from an increase in the number of binding domains, as has already been reported (Zupancic et al., 2021). Results of this experiment also showed that, although being smaller and not fully decorated, PVX‐VHH1‐F2A was significantly more efficient neutralizing the pseudovirus than TEV‐VHH1. We hypothesize that this is a consequence of normalizing the amount of VHH domains in the assay. Since PVX‐VHH1‐F2A was approximately half decorated, this means double amount of VNPs in the assay. Therefore, a balance between VNP concentration and degree of decoration must be found to obtain the best performance. It is worth noting here that protein quantification in our work was based on densitometry analysis of polyacrylamide gels stained with Coomassie brilliant blue and that a more precise quantification will be required to figure out in what extend nanobody‐decorated VNPs exhibit higher neutralization activity than the corresponding nanobodies in a dimeric format. Furthermore, the variability in terms of VHH/VNP density obtained in each purified batch must be taken into account.

Since the beginning of the current pandemic, SARS‐CoV‐2 circulating strains have acquired mutations that enhanced virus transmission not only by increasing affinity for its human ACE2 receptor but also by facilitating escape from immune neutralization (Chen et al., 2021; Harvey et al., 2021). In this context, the use of nanobody cocktails that recognize multiple epitopes is expected to be a more resistant strategy against eventual escape variants which may appear in response to selective pressure from single antibody therapy (Jin et al., 2021). In vitro evolution experiments have demonstrated that simultaneous targeting of two neutralizing epitopes severely hampers the emergence of viral escape mutants (Koenig et al., 2021; Xiang et al., 2020). Those experiments showed that combining two nanobodies imposed the requirement for a minimum of two amino acid substitutions to confer resistance to the nanobody cocktail, elevating the genetic barrier for escape. Given that VHH1 and VHH2 bind to different regions in RBD, we envisioned them as ideal partners for the generation of a multivalent VNP cocktail (Jin et al., 2021). Taking advantage of the well characterized synergism in PVX and TEV coinfections, we successfully produced a VNP cocktail combining TEV‐VHH1‐F2A and PVX‐VHH2‐F2A from the same host plant. This cocktail exhibited superior pseudovirus neutralizing activity versus the single VNPs (Figure 7).

The results presented in this study extend the use of plant viruses to produce VNPs for nanobody presentation. The large size of VNPs may contribute to half‐life improvement of circulating nanobodies (Lu et al., 2021). Highly stable multivalent nanobodies could be nebulized and exploited for the development of inhalable prophylactic formulations, allowing administration through respiratory routes (Liu et al., 2023). Therefore, the combination of potent neutralization activities and expected higher stability of VNP multivalent nanobodies makes them interesting candidates for therapeutic applications. Furthermore, when combined with medical payloads such as contrast agents or therapeutics, VNPs could serve as a targeted nanoparticle technology in molecular imaging diagnostic and drug delivery. In order to avoid severe off‐target effects of systemic drugs in medicine, several efforts are being made to develop targeted therapy using VNPs, in which drugs are either delivered specifically to tumour cells or activated specifically within them (Chung et al., 2020). As with all new approaches, there are challenges to overcome. For biomedical applications, the interactions of the CPs with the immune system need to be fully addressed. Although UV irradiation and chemical inactivation could be used to ensure VNPs are innocuous, they need to be carefully evaluated for in vivo toxicity and stability (Narayanan and Han, 2017). Moreover, while certain bottlenecks for expression of recombinant proteins in plants still remain, mainly related to purification and downstream processing as well as to regulatory issues, a wide range of recombinant proteins have been produced and their efficacy and safety have been demonstrated in both preclinical and clinical studies (Donini and Marusic, 2019). Finally, a good manufacturing practice (GMP) platform based on N. benthamiana could provide a reproducible, flexible system for rapid production of therapeutic antibodies (Swope et al., 2022).

In summary, in this work we have produced genetically encoded plant virus‐derived VNPs decorated with SARS‐CoV‐2 nanobodies in N. benthamiana, which have demonstrated superior neutralizing activity against pseudoviruses in infection cell assays. Further work must involve the presentation of nanobodies with different specificities, and even the combination of different functional molecules, beyond antibodies, into a single VNP. This study paves the way for the development of a new variety of tools and materials derived from plant viruses useful in biotechnology and nanomedicine.

Experimental procedures

Plasmid construction

Plasmid pLBPVX contains a full‐length PVX cDNA (GenBank accession number MT799816.1) flanked by the cauliflower mosaic virus (CaMV) 35S promoter and terminator. Plasmid pGTEVa (Bedoya et al., 2012) contains the cDNA of an infectious TEV variant with the GenBank accession number DQ986288 (G273A, A1119G), flanked by the CaMV 35S promoter and terminator in a binary vector derived from pCLEAN‐G181 (Thole et al., 2007). Derivatives from pLBPVX and pGTEV were constructed using standard molecular biology techniques, including polymerase chain reaction (PCR) with the high‐fidelity Phusion DNA polymerase (Thermo Scientific), DNA digestion with restriction enzymes followed by Gibson assembly of DNA fragments (Gibson et al., 2009) using the NEBuilder HiFi DNA assembly master mix (New England Biolabs). In pLBPVX‐VHH, the nanobody cDNA, flanked by Flag and c‐Myc epitopes, was inserted at the 5′ end of PVX CP ORF. In pLBPVX‐VHH‐P2A, PVX‐VHH‐E2A and PVX‐VHH‐F2A, the cDNAs corresponding to the respective picornavirus 2A peptide was inserted between those corresponding to VHH and the viral CP (Kim et al., 2011). In pGTEV‐VHH, the nanobody cDNA, flanked by E and c‐Myc epitopes, was inserted at the 5′ end of TEV CP cistron. The three initial codons of TEV CP, including silent mutations, were duplicated to mediate NIaPro proteolytic processing. In pGTEV‐VHH‐P2A, pGTEV‐VHH‐E2A and pGTEV‐VHH‐F2A, the cDNAs corresponding to the respective picornavirus 2A peptides were inserted between those corresponding to VHH and the viral CP (Kim et al., 2011). The full sequence of all plasmids in this work is in Figure S1.

Plant inoculation

Nicotiana benthamiana plants were grown at 25 °C under a 16/8 h day/night cycle in growth chambers. The strain C58C1 of A. tumefaciens, carrying the pCLEAN‐S48 helper plasmid (Thole et al., 2007), was transformed with the plasmids containing the different TEV clones mentioned above. Transformed bacteria were selected in plates with 50 μg/mL rifampicin, 50 μg/mL kanamycin and 7.5 μg/mL tetracycline. For the PVX clones, the C58C1 A. tumefaciens strain without the helper plasmid was used and transformed bacteria were selected in plates with 50 μg/mL rifampicin and 50 μg/mL kanamycin. Individual colonies of the different clones were further grown for 24 h at 28 °C in liquid media up to an optical density at 600 nm (OD₆₀₀) of 0.5–1. Cells were harvested by centrifugation, resuspended in agroinoculation solution (10 mM MES‐NaOH, pH 5.6, 10 mM MgCl₂ and 150 μM acetosyringone) at OD₆₀₀ of 0.5; the culture was further incubated for 2 h at 28 °C. With a needleless syringe, cultures corresponding to PVX or TEV clones were used to infiltrate fully expanded upper leaves from plants 4–6 weeks old. Immediately following infiltration, plants were watered and transferred to a growth chamber under a 12‐h day/night and 25 °C cycle. Aliquots of symptomatic tissues from upper leaves and the equivalent tissues from mock‐inoculated controls were harvested at 14 dpi, frozen in liquid nitrogen and stored at −80 °C until use.

RT‐PCR analysis of the viral progeny

Total RNA was purified from leaf tissue aliquots using silica‐gel columns (Zymo Research). For PVX analysis, aliquots of the RNA preparations were subjected to reverse transcription (RT) using the RevertAid reverse transcriptase (Thermo Scientific) and primer 5′‐CTCTTTAATTGCTGCTGC‐3′. RT products were subjected to PCR amplification using the high‐fidelity Phusion DNA polymerase and primers 5′‐GCCATTGCCGATCTCAAGCCAC‐3′ and 5′‐GCTACTATGGCACGGGCTGTAC‐3′, flanking the PVX CP amino‐terminal sequence. For TEV analysis, reverse transcription was carried out using primer 5′‐TCATAACCCAAGTTCCGTTC‐3′, while PCR amplification was performed with primers 5′‐CATCTGTGCATCAATGATCGAA‐3′ and 5′‐GTGTGGCTCGAGCATTTGACAA‐3′, flanking the TEV CP amino‐terminal region. PCR products were separated via electrophoresis in 1% (w/v) agarose gels that were subsequently stained with 1% (w/v) ethidium bromide.

Western blot analysis

Aliquots of frozen tissue (approximately 50 mg) were ground with a mill (Star‐Beater, VWR) using a 4‐mm diameter steel ball for 1 min at 30 s⁻¹. Three volumes of protein extraction buffer (60 mM Tris–HCl, pH 6.8, 2% [w/v] sodium dodecyl sulphate [SDS], 100 mM dithiothreitol [DTT], 10% [v/v] glycerol, 0.01% [w/v] bromophenol blue) were added. Samples were thoroughly vortexed, incubated for 5 min at 100 °C and clarified with centrifugation for 5 min. Aliquots of the supernatants were separated via SDS‐PAGE in 12% (w/v) polyacrylamide gels. Proteins were electro‐blotted to polyvinylidene difluoride (PVDF) membranes for 1 h. Membranes were blocked in 5% (w/v) non‐fat milk in washing buffer (10 mM Tris–HCl, pH 7.5, 154 mM NaCl, 0.1% [w/v] Nonidet P‐40) for 1 h and were then incubated overnight at 4 °C with conjugated antibodies in blocking solution at 1:10 000 dilution. Membranes were washed three times with washing buffer prior to detection. PVX CP and TEV CP were detected using polyclonal antibodies (Agdia) conjugated to alkaline phosphatase (AP). Flag and E epitopes were detected using monoclonal antibodies (Thermo Scientific) conjugated to HRP. AP and HRP were finally revealed using CSPD (Roche) and SuperSignal West Pico PLUS chemiluminescent (Thermo Scientific) substrates, respectively. Images were recorded using an Amersham ImageQuant 800 (Cytiva).

Viral particle purification

Symptomatic leaf tissue was weight and homogenized using a Polytron (Kinematica) in the presence of 2 volumes of cold extraction buffer (50 mM sodium phosphate, 10 mM EDTA, 10% ethanol, pH 7.2). The mix was filtered with Miracloth (Millipore) and clarified via centrifugation for 20 min at 7800 rpm. Supernatant was brought to 10 mM 2‐mercaptoethanol and 1% Triton X‐100 and incubated for 1 h. The preparation was centrifuged again for 20 min at 7800 rpm. Supernatant was collected and VNPs were sedimented by ultracentrifugation (2 h, 130 000 g, 4 °C). VNP‐containing pellet was resuspended in 1 mL of 50 mM sodium phosphate, 1% Triton X‐100 and loaded on a 10 mL sucrose gradient ranging from 10% to 50% or 70% for PVX or TEV, respectively. After ultracentrifugation (3 h, 130 000 g, 4 °C), 1‐ml fractions were collected and the presence of VNPs was evaluated by western blot analysis. Positive fractions were pooled, dialyzed against phosphate buffer saline (PBS) and concentrated using Centricon 100 kDa filter columns (Millipore). Preparations were stored at 4 °C until use.

Viral particle quantification

Aliquots of VNP preparations were separated by SDS‐PAGE and visualized by Coomassie brilliant blue staining. A bovine serum albumin (BSA) standard curve was used to quantify the concentration of VNPs using the ImageJ software. As a complementary approach, viral genome copies in the purified samples were quantified by RT‐quantitative PCR (qPCR). VNPs were denatured incubating them 5 min at 98 °C. The extracted viral RNA was used as template for RT‐qPCR following the protocol described above. qPCR was performed using the iTaq Universal SYBR Green Supermix (Bio‐Rad) and primers 5′‐AGTGGCACTGTGGGTGCTGGTGTTG‐3′ and 5′‐GTGTGGCTCGAGCATTTGACAA‐3′ at the 5′ end of TEV CP ORF.

Immunogold electron microscopy staining

Aliquots of purified VNPs (15 μL) were adsorbed to carbon film‐coated 200 mesh copper grids (EMS) for 10–15 min at room temperature and then blocked with 15 μL of 1% BSA in PBS. After blocking, the grids were incubated with 15 μL of one or more primary antibody described above for at least 2 h (diluted 1:50 in blocking buffer), washed twice with PBS and incubated with 10 μL of gold‐conjugated secondary antibody for 1 h. After incubation, the grids were washed twice with PBS and 4 times with H₂O (1 min each). Grids were finally stained with 10 μL of 2% (w/v) phosphotungstic acid (PTA; pH 7.0) for 3 min and dried at room temperature until transmission electron microscopy was carried out. TEM preparations were analysed using a JEM‐1400Flash (120 kV) electron microscope (JEOL, Japan).

Enzyme‐linked immunosorbent assays

Plates (96 wells) were coated with 100 ng/well of recombinant RBD antigen (Thermo Scientific, #RP‐87678) in Coating Buffer (15 mM Na₂CO₃, 35 mM NaHCO₃, pH 9.8). After 2 washes with 300 μL of PBS supplemented with 0.1% (v/v) Tween‐20 (PBS‐T), plates were blocked with 200 μL of blocking solution (5% milk in PBS‐T) for 2 h at room temperature. Next, plates were incubated for 2 h at room temperature with serial dilutions (1:5) of purified nanobody‐decorated VNPs in blocking solution starting at 1 μg of VNPs per well (100 μL). After 4 washing steps, 1:2000 alkaline phosphatase‐conjugated anti‐TEV CP (Agdia SRA 49501) or anti‐PVX CP (Agdia SRA 10002) in blocking solution was added. After 1 h, plates were washed 4 times with PBS‐T and incubated with 100 μL of 1 mg/mL disodium 4‐nitrophenyl phosphate (PNP; Sigma‐Aldrich). Absorbance was finally measured at 415 nm using a multi‐well plate reader (TECAN). Statistical analysis was performed using the GraphPad Prism 8 software. Bars represent the means and standard deviations for triplicate samples. Data were analysed by two‐way ANOVA with Tukey's post‐test (***P < 0.001; **P < 0.01).

Virus neutralization assay

The neutralization capacity of the VNPs against the SARS‐CoV‐2 was assessed using a GFP‐expressing vesicular stomatitis virus pseudotyped with the Wuhan‐Hu‐1 Spike protein as previously described (Sánchez‐Sendra et al., 2022), but using VERO‐E6‐TMPRSS2 cells. All tests were done in triplicate using fivefold VNP dilutions ranging from 1:20 to 1:62 500, with ~1000 foci forming units per well. Following 16 h of infection, the GFP signal in each well was quantified using a live‐cell microscope (Incucyte SX5, Sartorius). Background fluorescence from uninfected wells was subtracted from all well in which infection was detected, and the GFP fluorescence in each VNP‐treated dilution was standardized to the average fluorescence observed in mock‐treated samples. Any value resulting in a relative GFP signal of <0.001 was assigned a value of 0.001 to eliminate negative values.

Conflict of interest

The authors declare no conflict of interest.

Author contributions

F.M. and J.A.D. conceived the work with input from the rest of the authors. F.M., E.L.S., J.Z. and L.R. performed the experiments. All authors analysed the data. F.M. and J.A.D. wrote the manuscript with input from the rest of the authors. All authors discussed and revised the manuscript.

Supporting information

Figure S1 Nucleotide sequences of PVX‐wt (GenBank accession number MT799816.1) and the derived recombinant viruses PVX‐VHH1, PVX‐VHH1‐P2A, PVX‐VHH1‐E2A and PVX‐VHH1‐F2A.

Figure S2 Nucleotide sequences of TEV‐wt (GenBank accession number DQ986288, including silent mutations G273A and A1119G in red) and the derived recombinant viruses TEV‐VHH1, TEV‐VHH1‐P2A, TEV‐VHH1‐E2A and TEV‐VHH1‐F2A.

Figure S3 VHH2 was cloned replacing VHH1 in all constructs.

Data availability statement

The data that supports the findings of this study are available in the supplementary material of this article.