Introduction

The natural feature of some biological molecules to form well defined super molecular structures has been increasingly exploited as a valuable technology. In particular, the peculiarity of viruses to self assemble during replication and infection into protein based nanoparticles has received much attention since the early days of nanotechnology. Many examples include non-enveloped virus particles composed of the multimerization of a single coat protein (CP) that precisely assembles into structural architectures. Bacteriophages (Bar et al. 2008; Newton-Northup et al. 2009), animal (Murata et al. 2009; Santi et al. 2008) and plant viruses (Gonzalez et al. 2009; Lico et al. 2009) have been all investigated for their use as tools in a broad range of applied fields, extending from biomedicine to materials science (Lewis et al. 2006; Nam et al. 2006). Viral nanoparticles (VNPs) have a size particularly suitable for nanoscale applications and can offer several advantages over non biological nanoparticles: they are homogeneous in size (monodispersion), they bear an intrinsic robustness being at the same time naturally biodegradable and they have a large surface area to mass ratio and a defined, repetitive and symmetric macromolecular organization.

A common requirement for advanced nanoscale applications is to fully take advantage of multivalency and payload containment and transport. Also in this respect VNPs present several benefits. Viral capsids have been designed by nature to carry nucleic acids and in some cases non-structural proteins in their interior cavity, where they are confined and protected during transfer throughout extremely different environments. Moreover, many viruses are able to achieve tropism to certain organs and cell types, often as a result of characteristics related to the specific affinity of exposed portions of their CP to target cell receptors (Ashley et al. 2011; Singh 2009). Different strategies for the modification of VNPs’ outer and inner surfaces as well as for entrapment of molecules in their inside have been addressed in order to thoroughly take advantage of the described virus properties (Grasso and Santi 2010). The growing knowledge of the structure of viruses at high resolution level, together with the ease of manipulation of viral genomes, has allowed the development of a vast plethora of chimeric virus nanoparticles (CVNPs). Via genetic fusion of heterologous coding sequences (CDSs) to defined positions of the CP gene (cp), peptides with the most diverse biological functions have been displayed on VNPs outer surface (Gleiter and Lilie 2001; Hajitou et al. 2006; McCormick et al. 2006; van Houten et al. 2006) and, in a reduced number of cases, on their inner surface (Beterams et al. 2000). As an alternative or in combination with genetic modification, bioconjugation has been broadly addressed to expose not only peptides but also entire complex proteins, organic molecules and polymers such as nucleic acids and polyethylene glycol (Raja et al. 2003; Smith et al. 2006; Strable et al. 2004). The amine functional group of lysine is a common target for conjugation through N-hydroxysuccinimide (NHS) ester based chemistry (Steinmetz et al. 2010; Wang et al. 2002). Entrapment into VNPs has mainly been achieved in vitro, in presence of high concentrations of foreign molecules, by disassembly/reassembly of capsids (Chen et al. 2006; Comellas-Aragones et al. 2007; Ren et al. 2006) or by the reversible opening of viral gated pores (Douglas and Young 1998; Loo et al. 2008). Both methods require controlled tuning of physicochemical conditions such as temperature, pH and concentration of metal ions.

In particular, also plant viruses have received a great deal of interest as potential source of VNPs. Spherical as well as rod shape particles have been manipulated to develop novel nanotechnological tools for drug delivery, bioimaging, biomaterials and nanostructured electronic devices (Fowler et al. 2001; Lewis et al. 2006; Nam et al. 2006; Prasuhn et al. 2007; Ren et al. 2007). The vast majority of plant viruses has a small, single stranded, positive-sense RNA genome, that can be simply retro-transcribed into a cDNA infectious clone. Their limited genome size, together with the precise knowledge of their sequence, offers the possibility of de novo synthesis of their entire genome, making natural bioblocks even easier to obtain and use for the construction of new systems. The vast majority of plant viruses is non-infectious toward other organisms and therefore their use in biomedicine presents no biological hazard. Moreover, plants have been extensively reported as excellent bioreactor systems. They are safe, cost-effective and highly scalable. VNPs production in plants and their subsequent purification is simple, efficient, high yielding and quick (Chen and Lai 2013; Lai and Chen 2012; Santi et al. 2006; Scholthof et al. 1996).

Tomato bushy stunt virus (TBSV), the prototype member of the Tombusviridae family, has a monopartite, single stranded, positive-sense RNA genome of nearly 4,8 kb that contains five open reading frames (ORF) between a 5′-uncapped untranslated region (UTR) and a non-polyadenilated 3′ UTR. The third 5′-proximal ORF encodes the unique viral CP, p41, which is translated from a subgenomic mRNA into a product of ~41 kDa (White and Nagy 2004). The ternary structure of the CP subunit is defined by three separate domains: the RNA binding domain (R) formed by the N-terminal residues, the shell domain (S) and the C-terminal protruding (P) domain. The viral capsid is made of 180 copies of the CP subunit that self assembles into particles of ~30 nm in diameter with a spherical shape originated from a T = 3 icosahedral symmetry. In the context of this quaternary arrangement the CP R domain is placed at the interior of the virus particle, while the C-terminal P domain is exposed on the external surface (Harrison et al. 1978; Olson et al. 1983).

In this study we present the development of an efficient and versatile system for VNPs production in Nicotiana benthamiana based on the TBSV pepper isolate (Szittya et al. 2000). In particular, we report the construction of a vector which allows the generation of genetic fusion at the C-terminal end of the TBSV CP, resulting in the display of peptides outside the viral protein capsid. We tested the system by progressively fusing 4 CDSs of commonly used protein tags, one after the other in a modular fashion. We proved at different levels that all the obtained CVNPs were able to retain infectivity and proper structural assembly, reaching with the longest chimera the correct exposure of 56 additional amino acids. We used one of the chimera to address bioconjugation of biotin on genetically introduced lysine residues and we tuned an experimental method for virus reversal permeability, setting the ground for foreign molecules entrapment. Finally, we have optimized a purification protocol, determined yield and developed production standards.

Materials and methods

Genetic engineering

The construction of TBSV chimeras displaying specific tag peptides in fusion with the viral CP was achieved by standard recombinant DNA technologies (Sambrook and Russell 2001). All the first steps of genetic engineering were performed on a subclone obtained by NotI/NcoI digestion of the TBSV–P plasmid (Szittya et al. 2000) (Fig. 1a) followed by the introduction of the generated fragment, encoding for the C-terminal region of the CP, into the pBluescript SK (+) vector. A multi-cloning site (MCS) was inserted first by transforming the cp stop codon (Fig. 1b) into a ClaI restriction recognition site, which has then been used for vector digestion and ligation with the compatible in vitro annealed oligonucleotides (MCS-For, MCS-Rev; Online Resource 1) harboring the ApaI and PacI sites. The substitution of the amber stop codon with the ClaI sequence was obtained using the QuikChange® II-E Site-Directed Mutagenesis Kit (Stratagene) by means of two primers presenting the desired mutation (TBSV Mut-For, TBSV Mut-Rev; Online Resource 1). At this stage, the modified 3′-terminal cp sequence was moved back into the TBSV genome context upon NotI/NcoI digestion, generating the TBSV–vector (the modified region is detailed in Fig. 1c). The Flag CDS (corresponding to the DYKDDDDK peptide, Chiang and Roeder 1993), designed using preferential codon bias of N. benthamiana and spaced from the last cp nucleotide by a GGPGGGG linker CDS, was the first tag introduced in the polylinker of the previously generated vector. The vector was digested ApaI/PacI and ligated with the in vitro annealed Flag encoding oligonucleotides (Flag-For, Flag-Rev; Online Resource 1), owning compatible protruding ends, thus generating the TBSV–Flag (Fig. 1d). Subsequently, the TBSV–Flag construct was digested ApaI/PacI and the compatible in vitro annealed cMyc–Flag fusion CDS (cMyc-For, cMyc-Rev; Online Resource 1) (Kawarasaki et al. 2003) was introduced, producing the TBSV–cMyc (bearing the EQKLISEEDL cMyc peptide, derived from the c-myc proto-oncogene product, in addition to the Flag sequence; Fig. 1e). Next, the TBSV–cMyc construct was opened at ApaI site and the compatible in vitro annealed HA oligonucleotides (HA-For, HA-Rev; Online Resource 1) (coding for the YPYDVPDYA peptide, derived from the influenza Hemagglutinin, Laize et al. 1997) were ligated generating, upon correct orientation colony screening, the TBSV–HA (Fig. 1f). The same strategy was then used to clone the V5 fragment (corresponding to the GKPIPNPLLGLDST epitope, derived from the P/V proteins of the paramyxovirus simian virus 5, SV5; Kolodziej et al. 2009) using a synthetic oligonucleotides pair (V5-For, V5-Rev; Online Resource 1) creating the TBSV–V5 (Fig. 1g). In all cases, to separate each tag module from the others, GPG linker CDSs have been introduced by design into the encoding tag oligonucleotides. As for the Flag CDS, all tags CDSs were in silico optimized for N. benthamiana codon usage. To produce an alternative approach to viral in vitro transcription for infection, the T7 promoter of the original TBSV–P construct was replaced with the Cauliflower mosaic virus (CaMV) 35S promoter (Odell et al. 1985). The CaMV 35S sequence was amplified by PCR using the pPVX204 expression vector (Baulcombe et al. 1995) as a template and a pair of specific primers: the TBSV–35S–For primer (Online Resource 1), owning a 5′ BamHI extension, and the TBSV–35S–Rev primer (Online Resource 1). The 5′ TBSV genome context was amplified by PCR using a forward primer (TBSV-5′ context-For; Online Resource 1) covering the authentic 5′ end of TBSV genome (Hearne et al. 1990) and a reverse primer (TBSV-5′ context-Rev; Online Resource 1) with 5′ AgeI extension. Next, both PCR products properly restricted were included in a three fragments ligation with the TBSV–cMyc construct, previously digested BamHI/AgeI, generating the 35S–TBSV–cMyc construct (Fig. 1h). The Agrobacterium tumefaciens nopaline synthase (NOS) termination sequence (Depicker et al. 1982) was introduced downstream of the 3′ viral end. Also in this case, the NOS sequence was amplified by PCR using the pPVX204 expression vector as a template and two specific primers: the forward primer (PCR–NOS–For; Online Resource 1) with a 5′ XmaI extension and the reverse primer (PCR–NOS–Rev; Online Resource 1) with a 5′ SacI extension. Also the Hepatitis delta virus (HDV) antigenomic ribozyme (HRz) sequence was introduced (Chowrira et al. 1994) (the fragment, owning XmaI compatible ends, was generated by in vitro annealing of two complementary oligonucleotides; HRz–For, HRz–Rev; Online Resource 1). Subsequently, the 35S–TBSV–cMyc construct and the NOS fragment were XmaI/SacI digested and included in a three fragments ligation with the HRz sequence. The resulting constructs were screened for the presence of NOS (for the 35S–TBSV–cMyc.NOS vector; Fig. 1i) or of both fragments in the correct orientation (for the 35S–TBSV–cMyc.riboNOS vector; Fig. 1j). All clones were verified by sequencing (BMR Genomics) with specific primers (TBSV–1For, TBSV–2Back and ScreenNOS-Rev; Online Resource 1). All primers and oligonucleotides used in this work were synthesized at MWG-Biotech.

Fig. 1
figure 1

Schematic representation of TBSV cDNA constructs. a Genome organization of TBSV–P plasmid. The 5 ORFs encoding the respective proteins with their molecular masses (in kDa) are represented as boxes: p33 and p92 (RNA-dependent RNA polymerase), p41 (CP), p22 (movement protein), p19 (silencing inhibitor). b Sequence detail of the original and c of the genetically engineered CP stop codon (asterisk) region (TBSV–vector). The first amino acids of the GGPGGGG linker are reported. The underlined sequences represent the restriction enzyme sites used to clone the peptides of interest. As simplification, it is illustrated only the 3′-cp engineered region of the d TBSV–Flag, e TBSV–cMyc, f TBSV–HA, g TBSV–V5 constructs. The dotted boxes represent the GGPGGGG spacer and the striped boxes the GPG amino acid linker. h–j Genome organization of 3 TBSV–cMyc expression vectors based on the CaMV 35S promoter and the A. tumefaciens NOS terminator h 35S–TBSV–cMyc, i 35S–TBSV–cMyc.NOS, j 35S–TBSV–cMyc.riboNOS. The dotted line schematically represents the whole genomic sequence between the first 5′ TBSV nucleotide and the p41 ORF. The HDV antigenomic ribozyme sequence is indicated as “HRz” named box

Full size image

Plant infection

Six to eight weeks old N. benthamiana plants, grown in controlled conditions (24 °C, 16 h light/8 h dark), were inoculated with different TBSV constructs as described below.

cDNA templates of TBSV–P (that in vivo produces TBSV wild-type particles, TBSV–wt), TBSV–Flag, –cMyc, –HA and –V5 constructs were linearized by XmaI digestion. One μg of completely linearized cDNA was used for in vitro transcription following the manufacturer’s instructions (MEGAscript® T7 High Yield Transcription kit, Ambion Applied Biosystems). Integrity of the infectious transcribed RNA was verified by TBE agarose gel electrophoresis.

Plants were mechanically inoculated abrading the adaxial side of 2 leaves per plant using the TBSV infectious RNAs or the TBSV in planta expression vectors mixed with carborundum (silicon carbide) powder (VWR International). For each leaf inoculation, approximately 2–3 μg of the in vitro transcribed RNA were diluted in ultrapure sterile water up to a volume of 50 μl.

Plasmidic DNA of 35S–TBSV–cMyc, –cMyc.NOS and –cMyc.riboNOS constructs was extracted using the Plasmid Maxi kit (QIAGEN). Primary-infected plants were obtained by direct inoculation with cDNA vector on 2 healthy leaves per plant (30 μg for each leaf), as previously described; only the 35S–TBSV–cMyc cDNA was linearized by XmaI digestion before infection.

Reinfections were performed using saps from infected leaves. Briefly, a raw leaf sap was prepared by homogenizing the symptomatic systemic tissue of an infected plant in Phosphate Buffered Saline (1× PBS: 151 mM NaCl, 8,4 mM Na2HPO4 × 12 H2O, 1,86 mM NaH2PO4 × H2O).

VNPs purification

Virus particles were purified following the protocol of Burgyan and Russo (1998) with some modifications. Infected leaves were collected 7–9 days post-infection (dpi) and ground to a fine powder. Plant material was homogenized with 3 ml/g of 50 mM sodium acetate, pH 5.3, supplemented with 1 % ascorbic acid and a cocktail of protease inhibitors (P9599, Sigma) and the obtained solution was clarified by low-speed centrifugation (8,000×g for 15 min at 4 °C). During homogenization plant material lowered the pH of the extract and therefore NaOH was added to reach pH 5.0. The supernatant was then ultra-centrifuged for 1 h at 90,000×g at 4 °C. The obtained pellet was resuspended in 50 mM sodium acetate, pH 5.3. Quality of the preparations was verified by a silver stained 13.5 % SDS–PAGE. Virus concentration was evaluated using the Bradford micro-assay procedure (Bio-Rad) and confirmed through an SDS–PAGE followed by a Coomassie Brilliant Blue staining.

Electron microscopy

Purified virus particles were collected on carbon/formvar film coated 400 mesh copper grids (Electron Microscopy Sciences) and stained with 2 % (w/v) uranyl acetate aqueous solution. Grids were then analyzed by using a transmission electron microscope JEM 1200 EXII (Jeol) and images were acquired through a CCD Camera SIS Veleta (Olympus) at the Interdipartimental Center of Electronic Microscopy (University of Tuscia).

Sucrose gradient

A discontinuous sucrose density gradient was prepared by layering successive decreasing concentrations of sucrose solutions. Briefly, 2 ml each of 60, 50, 40, 30, 20 and 10 % (w/v) sucrose dissolved in PBS were poured into ultracentrifuge tubes and a volume of 200 μl of the crude extract was loaded on top of the gradient. The analytical separation occurred by ultra-centrifugation at 90,000×g for 2 h at 4 °C. Eleven fractions were collected from the top to the bottom of the tube. One hundred μl of each sample were finally analyzed by ELISA using a specific anti-Flag antibody or the DSMZ kit (AS-0098), which employs specific antibodies for the TBSV–wt detection.

SDS–PAGE analysis and immunoassays

Plant purified virus particles were mixed with standard Laemmli Loading Buffer supplemented with EDTA to a final concentration of 5 mM, denatured by heating for 1 min at 60 °C and resolved on a 13.5 % SDS–PAGE (Laemmli 1970). Proteins were electrophoretically transferred to a polyvinylidene difluoride (PVDF) membrane (Millipore) using a semidry blotting apparatus (TE70X semi-dry transfer unit, Hoefer). Subsequently, after blocking and washings, the membrane was probed with the specific primary antibody followed by incubation with the proper secondary horseradish peroxidase (HRP)-conjugated antibody. Signal detection was obtained by enhanced chemiluminescence (Immobilon Western Chemiluminescent HRP Substrate, Millipore).

Equal amounts of total soluble proteins (about 20 μg for each well) or plant purified virus particles were also analyzed by ELISA. Briefly, ELISA Maxisorp multiwell plates (NUNC) were coated with 100 μl of sample per well. Subsequently, 100 μl of the specific primary antibodies were added to each well. After incubation and washing the plates were probed with the proper secondary HRP-conjugated antibodies. Colorimetric reaction was developed using 2,2′-azino-bis(3-ethylbenzthiazoline-6 sulphonic acid) substrate (ABTS, KPL) and read at 405 nm on an automated ELISA reader (Sunrise, Tecan).

The primary antibodies and the respective dilutions used in this work were: mouse monoclonal M2 anti-Flag (1:4,000; F3165, Sigma), rabbit polyclonal anti-cMyc, -HA, or anti-V5 (1:2,000; PA1-22826, -86676, -32392 respectively, Thermo Scientific). For detection, anti-mouse IgG (NXA931, GE Healthcare) or sheep anti-rabbit IgG (PA1-84416, Thermo Scientific) HRP-conjugated secondary antibodies diluted 1:2,500 were used.

RNA extraction and RT-PCR analysis

Seven to nine dpi, total RNA was extracted from symptomatic inoculated and systemic leaves by using an RNeasy plant mini kit (QIAGEN) and RT-PCRs were performed by using 1 μg of extracted RNA for each sample and following the manufacturer’s instructions (ProtoScript M-MuLVTaq RT-PCR kit, New England Biolabs). cDNA was synthesized by using random primers and PCR was carried out with TBSV specific primers (RT-PCR1For and TBSV–2back; Online Resource 1) to amplify the 3′ cp region (obtaining a 648 bp fragment in the wild-type context). The resulting fragments were finally purified and sequenced (BMR Genomics).

VNPs reversable swelling

Approximately 4 μg of purified TBSV–wt or –Flag particles were incubated in swelling buffer (0.1 M Tris, 50 mM EDTA, pH 8.5) or control buffer (0.1 M Tris, pH 5.5) at 4 °C for 2 h. To verify VNPs pore gating, ethidium bromide (EtBr) wad added during the incubation at a final concentration of 0.5 μg/ml. Samples were resolved by electrophoresis in 1 % agarose gel using an electrophoresis buffer containing 38 mM glycine and 5 mM Tris at pH 8.5. The gel was prepared in the same buffer and, after samples run, it was stained in swelling buffer with the addition of EtBr at a final concentration of 0.5 μg/ml. Detection of virus particles was achieved by either visualization under UV light of the encapsidated RNA or Coomassie staining of the viral capsid protein.

RNAse protection assay

Four to six micrograms of TBSV–wt or –Flag particles were treated in swelling buffer or control buffer as described above. After 2 h 50 ng of ribonuclease A (New England Biolabs) were added to the reaction and the incubation was left at 4 °C for additional 30 min. Encapsidated RNA was extracted by the addition of an equal volume of a 1:1 phenol–chloroform solution (v/v) and SDS at a final concentration of 1 % (w/v). The mixture was agitated vigorously and then centrifuged to separate the two phases. The aqueous phase was recovered and the RNA was precipitated by the addition of 1/10 volume of 3 M sodium acetate and 3 volumes of ice cold ethanol. Finally, the RNA was sedimented by centrifugation and the pellet resuspended in RNAse free water. Integrity of the viral extracted RNA was analyzed by TBE 1 % agarose gel electrophoresis and visualized under UV light after EtBr staining.

Bioconjugation of biotin

In vitro biotinylation experiments were performed on TBSV–wt or –Flag particles following the EZ-Link® Micro NHS–polyethylene-oxide (PEO4)-Biotinylation Kit (Thermo Scientific) manufacturer’s instructions. Approximately 200 μg of VNPs were mixed with NHS–PEO4 biotin reagent at tenfold molar excess and incubated 2 h in ice and then one additional hour at 4 °C. The excess non-reacted and hydrolyzed biotin reagent was removed by chromatography (Zeba™ Desalt Spin Column, 7000 MWCO; Thermo Scientific) and the purified samples (approximately 2 μg each) were analyzed by western blotting and ELISA using a streptavidin HRP-conjugated as probe (N100, Pierce).

Results

Genetic engineering

Several studies have demonstrated that the C-terminal portion of the TBSV CP is exposed on the exterior surface of the assembled virion (Harrison et al. 1978; Olson et al. 1983). Hence a TBSV based vector, for the generation of a peptide display system, was designed to contain the heterologous peptides CDSs genetically fused to the 3′ end of the viral cp gene. To this purpose, a MCS has been inserted in this region, through several subcloning steps, as described in the materials and methods section, involving as original source the TBSV–P plasmid (Szittya et al. 2000), which contains the whole viral genome sequence as cDNA downstream the T7 promoter (Fig. 1a). In essence, the thymine (T) and the guanine (G) contained in the native cp amber stop codon (TAG) (Fig. 1b) were converted into a specific ClaI recognition sequence (ATCGAT) by site directed mutagenesis. Subsequently, a MCS sequence, carrying the ApaI and PacI restriction sites, was inserted by means of in vitro annealed pair of oligonucleotides displaying 5′ and 3′ ClaI compatible ends, generating the TBSV–vector (Fig. 1c). Several chimeras were created by the insertion of different peptides CDSs into the TBSV–vector using the previously introduced unique restriction recognition sites. Four common tag peptides, named Flag (DYKDDDDK), cMyc (EQKLISEEDL), HA (YPYDVPDYA) and V5 (GKPIPNPLLGLDST), were chosen as candidates to test the potential of the TBSV to sustain the display of increasingly longer peptide sequences. As a result, four different constructs were generated by the progressive addition of the 4 tags CDSs assembled together one after the other in a modular fashion. Tag peptides encoding oligonucleotides were designed using N. benthamiana preferential codon bias and were subcloned, upon in vitro annealing, to generate the 4 different engineered viral constructs encoding chimeric CPs progressively increasing in length: TBSV–Flag (Fig. 1d), TBSV–cMyc (including also Flag peptide CDS, Fig. 1e), TBSV–HA (including also Flag and cMyc peptides CDSs, Fig. 1f) and TBSV–V5 (including also Flag, cMyc and HA peptides CDSs, Fig. 1g), respectively resulting in extensions of 15, 28, 40 and 56 amino acids compared to the wild-type CP. The TBSV–vector is devoid of the original stop codon thus, during the in silico design process, a stop codon was introduced at the 3′ end of the Flag CDS, which represents the most distal module in every construct.

Moreover, the 5′ end of each fusion peptide CDS was in silico designed in order to generate, upon genetic fusion, a GGPGGGG linker peptide between the CP and the first tag module, and a GPG linker peptide between each of the subsequent adjacent modules. In fact, GPG based spacers were adopted to provide flexibility between the CP and the exogenous fragment, to avoid undesirable secondary structures that might interfere with chimeric CP folding and chimeric virus assembly, and to foster spatial and functional independence of multiple tag peptides.

To produce an alternative approach for the plant infection, which is normally obtained through the inoculation of infectious in vitro transcribed RNA, an in vivo transcription model strategy was designed. Using the TBSV–cMyc construct as a template, the CaMV 35S promoter sequence was introduced in place of the original T7 promoter, thus generating the 35S–TBSV–cMyc (Fig. 1h). Two additional constructs, both harbouring the cDNA of the TBSV–cMyc under the control of the CaMV 35S promoter, have been then assembled modifying the 3′ end context downstream the virus sequence: the 35S–TBSV–cMyc.NOS (Fig. 1i) additionally contains the termination sequence of the nos gene of A. tumefaciens and the 35S–TBSV–cMyc.riboNOS (Fig. 1j) has the HRz sequence introduced between the 3′ end of the genome and the NOS sequence. This particular ribozyme element, already employed by Scholthof (1999), when properly inserted is capable, through auto catalytic activity, to generate 3′ RNA terminal sequences identical to those of the TBSV genome, eliminating regions derived from nuclear transcription and normally not present on the infectious viral RNAs, such as the poly-adenylation signals.

Infectivity and in vivo assembly of VNPs

TBSV–P (that in vivo produces TBSV–wt particles), –Flag, –cMyc, –HA and –V5 vectors were linearized by digestion with XmaI restriction enzyme, purified and used for the in vitro transcription of infectious RNAs. TBSV RNAs were used to inoculate 6–8 weeks old N. benthamiana plants by gently rubbing the surfaces of 2 leaves per plant with silicon carbide and distributing on each leaf approximately 2–3 μg of RNAs. The symptomatology was constantly monitored and documented by photo capturing (data not shown). Plants inoculated with infectious TBSV RNAs showed the characteristic chlorotic local lesions onto inoculated leaves at 2–3 dpi. At 5–7 dpi symptoms appeared systemically showing mottling, severe distortion and stunting of the leaves. Necrosis of the apical shoots and finally the death of the plant occurred between 11 and 14 dpi. Differences between plants inoculated with infectious TBSV–wt or –chimeras RNAs, in terms of symptoms progression and severity, were not observed.

Leaves showing systemic infection were harvested at 7–9 dpi and virus particles were purified following the protocol of Burgyan and Russo (1998) with some modifications. The recovery was about 300 μg of VNPs per gram of fresh leaf tissue. The quality of each purification was verified by a silver stained SDS–PAGE (Fig. 2a).

Fig. 2
figure 2

Evaluation of particles purification and assembly. a Silver stained SDS–PAGE of purified TBSV–wt (1 μg, lane 1) and TBSV–Flag (1 μg, lane 2). Standards molecular weights are shown on the right of the figure. b Sucrose gradient sedimentation analysis. Crude extracts from TBSV–wt and –Flag N. benthamiana infected plants were analyzed on a 10–60 % sucrose gradient; 11 fractions were collected from the top to the bottom. The presence of TBSV particles in the collected fractions was determined by ELISA using anti-TBSV–wt CP (filled circle) and anti-Flag peptide (empty triangle) antibodies

Full size image

To verify the proper assembly of TBSV–chimeras, crude extracts of symptomatic TBSV–wt or –Flag infected leaves were subjected to sucrose sedimentation analysis. Eleven fractions were collected from the sucrose gradient and further analyzed by ELISA with specific antibodies against the viral CP or the Flag tag. The sedimentation profiles of TBSV–wt and –Flag crude extracts were similar (Fig. 2b).

To obtain direct evidence of TBSV nanoparticles formation, purified TBSV–wt and –chimeras particles were subjected to transmission electron microscopy analysis. For all samples, VNPs with a diameter of about 30 nm were observed (Fig. 3).

Fig. 3
figure 3

Visualization by negative staining and high resolution electron microscopy of purified TBSV–wt and –chimeras particles. a–c TBSV–wt displayed at different magnifications, d TBSV–Flag, e TBSV–cMyc, f TBSV–HA, g TBSV–V5

Full size image

CVNPs characterization

The leaves of primary infected plants showing systemic symptoms were harvested at 7–9 dpi and the genetic stability of the chimeric cp genes was verified by RT-PCR and sequencing. The mobility shift of PCR fragments on agarose gel (Fig. 4a) represents the different length of the constructs at genomic level, and the sequencing of the bands confirmed the presence of the correct heterologous sequences. To further characterize the TBSV constructs, purified TBSV–wt and all TBSV–chimeras were subjected to denaturation. Different protocols were tested to avoid protein aggregates (data not shown), finding as the best parameters heating at 60 °C for 1 min in the presence of EDTA at a final concentration of 5 mM. A quantity of 1 μg of purified denatured proteins for each construct was resolved by 13.5 % silver stained SDS–PAGE (Fig. 4b, upper panel). For all chimeric CPs, differences observed in the mobility through the acrylamide gel confirmed their expected sizes. Since all CVNPs displayed the Flag tag as part of the CP fusion peptide, the same samples were analyzed by western blotting using the anti-Flag specific antibody (Fig. 4b, lower panel). Results showed the presence of the Flag tag in all the fusion assemblies, confirming the expected molecular weight of the different chimeric CPs (Fig. 4b, lower panel). Finally, the absence of degradation products indicated that proteins were stable under purification and storage conditions. ELISA analysis was performed with antibodies specific to each single tag to assess the correct exposure of the fusion peptides on the surface of assembled CVNPs (Fig. 4c). A total protein raw extract was prepared from the infected tissue of plants inoculated with all the different constructs. Normalized extracts were evaluated with ELISA and a crude extract of a not-infected N. benthamiana plant was used as a negative control. Specific binding of the anti-tag antibodies directed against the chimeras was observed. Anti-Flag, -cMyc, -HA and -V5 antibodies were able to recognize the corresponding tag peptide (Fig. 4c), expressed as C-terminal fusion of the CP, demonstrating the correct presence of the peptides at protein level, their exposure on virion surface and the accessibility for antibodies binding.

Fig. 4
figure 4

Molecular characterization of TBSV vectors. a Evaluation of the presence of the heterologous sequences at genomic level. Agarose gel electrophoresis of RT-PCR covering the 3′ cp end of the TBSV–wt (fragment of 648 bp) and of all TBSV–chimeras. b SDS–PAGE analysis of denatured particles. TBSV–wt and all the TBSV–chimeras CPs were analyzed by silver staining (upper panel) and western blot (lower panel) using an anti-Flag antibody. c ELISA analysis of properly assembled TBSV particles. TBSV–wt and all the TBSV–chimeras VNPs were probed separately with specific anti-tag antibodies. A total soluble protein extract from a not-infected (NI) N. benthamiana plant was used as a negative control

Full size image

Access to the internal cavity of VNPs

In general two main strategies have been adopted to load a cargo inside VNPs: (1) in vitro assembly of the VNPs (Chen et al. 2006; Comellas-Aragones et al. 2007; Ren et al. 2006) and (2) reversal viral pore gating (Douglas and Young 1998; Loo et al. 2006), both achieved in presence of high concentration of the exogenous molecules.

In the case of the TBSV, the inner core can be reached by switchable pH dependent gating (Aramayo et al. 2005; Kruse et al. 1982). Many aspects of virus assembly were addressed and physicochemical conditions for the reversible opening of gated pores were defined. We applied slightly alkaline conditions and divalent ion chelators to promote conformational changes necessary for particle loading. As proof of concept, the loading of the small EtBr molecule into swollen TBSV nanoparticles was tested.

Approximately 4 μg of purified TBSV–wt or –Flag particles were incubated at 4 °C for 2 h in swelling buffer (0.1 M Tris, 50 mM EDTA, pH 8.5; Fig. 5a, lanes 2, 4) or control buffer (0.1 M Tris, pH 5.5; Fig. 5a, lanes 1, 3), both containing EtBr (0.5 μg/ml). The samples were then resolved by agarose gel electrophoresis and visualized under UV light. According to the nature of the EtBr molecule, the signal was visible only when this compound penetrated into cavity of swollen particles intercalating the viral RNA (Fig. 5a, upper panel, lanes 2, 4). Therefore, fluorescence was not detectable in lanes containing the virions being in compact state (Fig. 5a, upper panel, lanes 1, 3). When the whole gel, after the run, was incubated in swelling buffer and fresh EtBr was added, the particles previously in compact state underwent swelling and became visible once exposed to UV light (Fig. 5a, middle panel, lanes 1, 3). The assembly of the TBSV based nanoparticles after the treatments has been proved by staining the same gel with Coomassie Brilliant Blue, highlighting the co-migration of genomic RNA with capsid proteins (Fig. 5a, lower panel). To further test cations depletion and alkaline conditions as prerequisites for the accessibility to the viral RNA contained in the VNPs’ cavity, TBSV–wt or –Flag particles were challenged for susceptibility to RNAse activity in both control and swelling buffer (Fig. 5b). RNA nuclease processing was investigated to evaluate digestion: viral genomic RNA was extracted from VNPs used in each treatment and analyzed by agarose gel electrophoresis. The RNA enclosed in the TBSV nanoparticles at pH 8.5 supplemented with EDTA 50 mM was completely degraded (Fig. 5b, lanes 5, 6) whereas the genome within virions at pH 5.5 resulted fully protected (Fig. 5b, lanes 3, 4), as the untreated particles’ controls (Fig. 5b, lanes 1, 2).

Fig. 5
figure 5

Evaluation of TBSV internal cavity accessibility. a EtBr loading by controlled TBSV capsid expansion. Upper panel agarose gel of TBSV–wt (lanes 1, 2) and TBSV–Flag particles (lanes 3, 4) incubated with EtBr in swollen (lanes 2, 4) or compact state (lanes 1, 3). Further, the same gel was incubated in swelling buffer supplemented with EtBr (middle panel) and then stained with Coomassie Blue (lower panel). b Ribonuclease A protection assay. Agarose gel electrophoresis of TBSV extracted RNAs. Untreated TBSV particles were used as control of genome extraction (lane 1 TBSV–wt, and 2 TBSV–Flag). Purified particles were treated with ribonuclease A in compact (lane 3 TBSV–wt, and lane 4 TBSV–Flag) or swollen state (lane 5, TBSV–wt, and lane 6, TBSV–Flag) before RNA extraction

Full size image

Chemical derivatization of VNPs

Bioconjugation of biotin by NHS ester based chemistry was tested to investigate the capability of TBSV nanoparticles to be addressed by chemical derivatization. NHS–PEO4 biotin (Pierce) reacts with primary amines creating a stable amide bond through NHS ester. Although the TBSV CP contains 13 lysine residues in its primary structure, it has been shown that, in the arrangement of the quaternary structure of the properly assembled virion, primary amines contained in their side chain are not accessible (Harrison et al. 1978; Olson et al. 1983). As previously described, the required lysine residues were indeed introduced, by genetic modification, on the outer surface of each of the chimeric TBSV nanoparticles through the Flag tag fusion (DYKDDDDK) which is ubiquitously present on each CVNPs as last heterologous portion at the C-terminus of the chimeric CPs (see Fig. 1 for the related c-DNA constructs).

TBSV–wt or –Flag nanoparticles were incubated with a tenfold molar excess of NHS–PEO4–Biotin for 24 h. The non-reacted biotin reagent and the byproducts of the reaction (the NHS leaving group) were then removed by chromatography during the desalting step on a spin column. A sample from each purified reaction was analyzed by ELISA and western blot using a HRP-conjugated streptavidin as a probe (Online Resource 2a, b, lower panel). The results confirmed that TBSV–wt is unsuitable for conjugation to biotin by NHS ester based chemistry (Online Resource 2b, lower panel, lanes 1, 2). In contrast, the presence of biotin-nanoparticles conjugates was successfully confirmed for TBSV–Flag particles (Online Resource 2b, lower panel, lanes 3, 4). The presence of TBSV CPs was additionally detected through a silver stained SDS–PAGE (Online Resource 2b, upper panel).

Infections by in planta transcription of cDNA based constructs

An alternative in vivo transcription model based on plant expression strategy was designed. Three constructs have been assembled: 35S–TBSV–cMyc, –cMyc.NOS and –cMyc.riboNOS (Fig. 1h–j). To test the infectivity of the 3 different expression vectors, 5 plants of N. benthamiana were inoculated on 2 leaves per plant with each of the different purified cDNA expression constructs at 3 different concentrations: 10, 20 or 30 μg. The symptomatology was constantly monitored and documented by photo capturing. Using 10 μg of plasmid none of the vectors was able to establish the infection. With 20 μg, at 10 dpi only upper leaves of plants inoculated with 35S–TBSV–cMyc.riboNOS vector showed evident systemic viral symptoms. With 30 μg the 35S–TBSV–cMyc.riboNOS vector produced systemic symptoms at already 6–8 dpi, whereas the 35S–TBSV–cMyc.NOS showed some mild signs of infection only at 19 dpi. All the attempts performed with linearized 35S–TBSV–cMyc vector, used at 3 different concentrations for inoculation, failed to induce both local symptoms development and systemic virus spread.

As final biological prove of infection a cycle of re-inoculations was performed using the sap obtained homogenizing the symptomatic systemic tissue of all infected plants. All primary infections were confirmed.

The use of 30 μg of the 35S–TBSV–cMyc.riboNOS type of vector showed to be so efficient that it could be employed routinely to establish primary infections for production purposes, thus avoiding the in vitro transcription step.

Discussion

Nanotechnology’s need for defined, repetitive and nanoscale structures can find in VNPs an almost inextinguishable source of buildingblocks, scaffolds and particulate arrangements. It is estimated that viruses are the second most prevalent biological entity on the planet (Suttle 2005). Many different viruses of the most diverse origin have been studied for their potential use as structured bioinspired nanomaterials and nanodevices (Grasso and Santi 2010). Plant virus based systems are among the most advanced and exploited. More than 800 plant virus genomes are completely sequenced (data retrieved from viruses complete genomes at NCBI database: http://www.ncbi.nlm.nih.gov), the majority having a small single stranded positive RNA nature. Increasing knowledge on virus structure, genetics, replication, host range and fitness has already allowed the development of well established platforms (Gopinath et al. 2000; Kumagai et al. 1993; Lico et al. 2006; Liu et al. 2005; Yildiz et al. 2012; Ziegler et al. 2000).

In this light, the TBSV pepper isolate infectious cDNA clone was used for genetic engineering. Construct design strategy was developed to address the sequential insertion of 4 peptide modules to test the ability of the system to accommodate increasingly longer sequences. The choice of employing the well known and commonly used FLAG, cMyc, HA and V5 tags was made for the availability of commercial reagents and protocols. A sequence coding for a GGPGGGG linker peptide, spacing the CP and the heterologous portion, was designed in all constructs. This sequence was employed primarily for the well documented ability of GPG to interrupt protein secondary structures, providing flexibility to the residing region and, consequently, promoting each peptide to assume its original conformation independently of other modules (Wriggers et al. 2005). For these reasons, sequences were designed to have also GPG linkers between each module. FLAG was the first peptide introduced and it is therefore present at the C-terminal end of each CVNP (Fig. 1d–g). Its octapeptide sequence (DYKDDDDK) contains lysine and aspartic acid residues that, together with cysteines, are the most commonly addressed amino acids for chemical derivatization of VNPs (Steinmetz et al. 2010).

Nicotiana benthamiana plants are excellent bioreactors for supporting VNPs production in both laboratory setting and large scale production facilities. These plants are easy to grow and handle, being required only 8 weeks from seeds to the proper developmental stage for infection. Moreover, N. benthamiana plants are neither food nor feed and their use for production of plant made recombinant pharmaceuticals is so diffused that can almost be considered a standard (Goodin et al. 2008; Lico et al. 2008; Twyman 2005; Twyman 2008; Webster et al. 2009). Mechanical leaf rubbing inoculation was performed to test the different constructs. Simptomatology and kinetics of infection were very homogeneous among different chimeras. Production time can be estimated in just 10 days from tissue inoculation to purified VNPs. Chimeras were able to self assemble in vivo as demonstrated by comparing, on a sucrose gradient, each sedimentation profile to TBSV–wt (Fig. 2b). Direct prove of assembly was also verified by electron microscopy, that allowed to unequivocally demonstrate CVNPs homogeneity regarding both dimensions and shape (Fig. 3). The purification procedure is reliable, efficient and clean. The system consistently allows to produce approximately 300 mg of purified VNPs per kg of fresh infected leaves. Data collected from several independent experiments suggest that VNPs yield after purification was very similar among different CVNPs.

Plant material deriving from inoculated and systemically infected leaves was used to characterize the different CVNPs at protein level and to evaluate vector stability at nucleic acid level. ELISA was employed to test the reactivity of plant crude extracts, using the different anti-tags antibodies available. Data not only confirmed the expected display of each exogenous peptide module, but also revealed that the progressive fusion of different peptides did not affect the structure required for their molecular recognition. In fact, in the multi-modules assemblies (TBSV–cMyc, –HA and –V5) each tag was efficiently recognized when individually probed with the respective antibody (Fig. 4c). In general terms, independent structural folding would be particularly precious when using multiple display assembly of antigens for vaccine development, which is the oldest and one of most promising applications of plant based VNPs in biomedicine (Chen and Lai 2013; Lico et al. 2009; Lico et al. 2012; Montague et al. 2011; Steinmetz et al. 2009; Tyulkina et al. 2011). In addition, this feature could be effectively employed in targeted delivery of VNPs for displaying more than one moiety on the particle surface (Work et al. 2004). Correct fusion protein expression was also confirmed on purified CVNPs by western blot analysis, using the anti-Flag antibody. The mobility shift among the chimeric and wild-type CPs was clearly related to the respective sizes (Fig. 4b).

We were successfully able to display up to 56 additional amino acids as C-terminal fusion (TBSV–V5). Regardless from the attempt of displaying the entire GFP (238 aa), which resulted in a non-infectious construct (data not shown), TBSV–V5 was the longest assembly we tested, leaving the door open to even longer fusions. In any case it has to be stressed that length is not the only parameter to consider: charge, shape, dimension and affinity are all fundamental features, which mostly rely on the chemical nature of the fused heterologous sequence. Hence, each new chimera must be evaluated empirically on a case to case basis.

Construct instability has been reported in different systems. In fact, sometimes the added sequence can impair viral fitness, thus viral genome can rearrange with single or multiple mutations to acquire more stability. It is also possible a complete recombinational event with the total extrusion of the heterologous portion to reset viral genome to the initial wild-type configuration (Avesani et al. 2007; Haviv et al. 2006; Lico et al. 2006). Integrity of the sequence at nucleic acid level was here evaluated by RT-PCR and fragments sequencing (Fig. 4a). Recombination of chimeric replicons, isolated from inoculated and systemic leaves of first and second generation infections, has never been detected. Beyond second generation we have sporadically experienced some construct instability, especially when the largest fusions were concerned (data not shown). In this regard, we have found very convenient to use in vitro transcribed RNA directly for production purposes and we believe that it is indeed feasible to use just first generation material for small scale production. Under this view and with the prospect of a scaled up process, we designed and assembled an in planta expression cDNA based construct, that could be delivered to plants simply in the form of a circular DNA plasmid under the control of sequences active in plants, therefore avoiding the enzymatic treatment necessary for the in vitro transcription. Consequently, a high copy plasmid was used as backbone to support the assembly of a plant expression construct containing: the CaMV 35S promoter, the infectious viral cDNA, the HRz sequence and the NOS terminator (35S–TBSV–cMyc.riboNOS, Fig. 1j). Upon mechanical leaf rubbing inoculation we demonstrated that 30 μg of this vector are able to reproducibly establish infections comparable, in regards of symptomatology and progression, to the one generated by in vitro transcribed RNAs. The employment of HRz sequence, previously reported for TBSV (Scholthof 1999), promotes an autocatalytic process at RNA level that results, upon cleavage of the acquired poly (A) tail, in the generation of the correct viral 3′ end, essential for an efficient replication (Chowrira et al. 1994).

In vitro chemical modification of VNPs relies on the covalent binding of an activated conjugate to functional groups of amino acids residues of the virus structural proteins. The amine functional group of lysine is a common target for conjugation through NHS ester based chemistry (Smith et al. 2006; Wang et al. 2002). Thirteen lysine residues are part of the 388 amino acids long TBSV CP (pepper isolate) (Szittya et al. 2000), 8 of which reside in the internal R domain, 5 in the S domain and none in the protruding P domain. Therefore no primary amines are exposed and addressable on the outer viral shell (Hsu et al. 2006). Nevertheless, Flag (DYKDDDDK), which is the most distal peptide on each chimera, carries 2 lysine residues, one as the very last carboxi-terminus. Biotin NHS esters were used to probe for reactive amine groups exposed on Flag CVNPs. Although the number of addressed lysine residues has not been evaluated and it will be the object of future studies, effective biotinylation of Flag CVNPs was proved using streptavidin probes in both western blot and ELISA (Online Resource 2a and 2b, lower panel). Conversely, as predicted, it was not possible to functionalize unmodified wild-type virions.

Chemical derivatization of lysine functional groups has been exploited using other plant derived VNPs systems such as the case of successfully displaying fluorescent dyes for in vivo imaging (Lewis et al. 2006) and folic acid for cancer cells specific targeting (Ren et al. 2007). Moreover, bioconjugation of biotin on plant VNPs has been used to employ the strong non-covalent bond between biotin and streptavidin. In essence this interaction can be used to redecorate VNPs surface with proteins of interest fused to streptavidin. By uncoupling the expression of the VNP scaffold and of the protein to expose, very large assemblies, otherwise very difficult to generate, have been achieved (Smith et al. 2006).

The employment of other native or genetically introduced amino acid functional groups, such as carboxylates from aspartic and glutamic acids, thiol from cysteine, and tyrosine together with a more quantitative characterization and optimization of conjugation’s conditions will greatly expand the range of different chemistries, and their combinations, for addressing VNPs.

Viral capsids play different roles according to the influence of the surrounding molecular environment. Protection, delivery and discharge of their genetic cargo are fundamental functions for virus propagation. Many aspects of structural transition sustaining these viral functions have been addressed using a variety of biochemical and biophysical techniques setting the conceptual ground for using these objects as nanocontainers (Johnson and Chiu 2000; Ma et al. 2012; Yildiz et al. 2011). Although the interior plant virus cavity has not been addressed as much the external surface, several studies exploited specific viral structural properties to gain access to VNPs’ inner core. Some plant viruses efficiently assemble into mature capsids in vitro. The process of disassembly/reassembly can be promoted by providing the proper physicochemical conditions. Consequently, cargo entrapment can be established by capsid reassembly in presence of high concentration of the exogenous molecules (Loo et al. 2006; Ren et al. 2006). On the other hand, the inner side of icosahedral viruses, such as TBSV, can be reached by reversible opening of virion gated pores. In fact, it has been demonstrated that many of these structures, at the interface of their capsomeric subunits, carry divalent cations, typically Ca2+ (Speir et al. 1995) but also Mg2+ (Lucas et al. 2002), which appear to be essential for virion structural stability. In particular, the TBSV swelling process has been widely studied by different approaches which revealed that its inner core can be reached by switchable pH and divalent ion-dependent gating (Aramayo et al. 2005). In fact, by depletion of divalent cations (using EDTA 50 mM) in slightly alkaline conditions (pH 8.5) at 4 °C for 2 h we were able to address the interior cavity of TBSV particles setting the ground for cargo loading. We first demonstrated that genomic RNAs, contained inside VNPs in relaxed (swollen) state, were accessible to RNAase, while genomes contained in VNPs in compact state were inaccessible and therefore protected from degradation (Fig. 5b). Moreover, switching between the two conformational states, we were able to load EtBr inside the VNPs as clearly revealed by the fluorescence emission derived from the intercalation of this fluorophore into the nucleic acids contained inside the viral cavity (Fig. 5a). We decided to use EtBr mainly for analytical reasons in order to provide a proof of concept in the prospect of loading TBSV core with chemical molecules of pharmacological activity. In fact, EtBr has a size and chemical nature in the range of many drugs and it is currently used in veterinary medicine to treat trypanosomosis in cattle (Stevenson et al. 1995).

In conclusion, in this study we have presented the development and characterization of a TBSV based nanovector that can be used for the encapsulation of small molecules and on which a predictable and programmable genetic and/or chemical engineering can be performed. Safety, production efficiency, industrial scalability and versatility related to possible modifications are peculiar strengths of this system. Modification by design of novel VNPs properties, together with the possibility of generating organized arrays using VNPs as nanoblock components, have the potential to create multifunctional structures able to diagnose diseases, deliver therapeutic agents, monitor treatment progression, support catalytic reactions and sustain in vitro cell proliferation and differentiation. However, any new functionality must be carefully evaluated in regard of the VNPs properties and fate. More detailed studies concerning the in vivo behavior of VNPs in terms of biocompatibility, bioavailability, biodistribution, pharmacokinetics, immunogenicity and toxicity are of paramount importance so that this technology will be able to live up to its potential.