WT CPMV and eVLP have the same antigenicity
CPMV is an icosahedral virus composed of 60 copies of both the Large (L) and the Small (S) coat protein subunit16. CPMV has a bipartite ssRNA genome (RNA-1, 6 kb and RNA-2, 3.5 kb). Each of the two genomic segments is encapsidated separately producing three fractions named for their sedimentation behavior in density gradients: CPMV-B(ottom) (containing the larger RNA-1), CPMV-M(iddle) (containing RNA-2) and a relatively small amount of naturally occurring empty particles (CPMV-T(op)). The protein components of CPMV-B, CPMV-M, CPMV-T and a recombinantly-expressed empty VLP (eVLP-CPMV) are almost identical structurally (Fig. 1a)13,14,15,16 with the only significant difference being in the degree of cleavage of a 24 amino acid extension to the C-terminus of the S subunit, which occurs during particle purification and storage13,14. This extension is located at the 5-fold vertices of the eVLP-CPMV structure13. A polyclonal antiserum raised against WT CPMV17 can detect both the L and S subunits in all three forms of WT CPMV and eVLP-CPMV (Fig. 1b). An Affimer phage display library3,4,5 was screened against purified eVLP-CPMV immobilized using a biotin-streptavidin linkage, to isolate specific binders. After three rounds of panning, 24 randomly picked clones were tested for binding eVLP-CPMV as well as WT CPMV using phage ELISA (Fig. 1c). To confirm effective binding these Affimer clones were tested against WT CPMV. The majority of the clones bound WT CPMV to a level comparable to that seen for eVLP-CPMV. However, two clones (numbers 14 and 22), did not show such binding for CPMV or eVLP-CPMV and represent false positives from the screening process and were not analysed further (Fig. 1c). Seven clones with the highest signal amplification, and therefore most likely the highest affinity binders for CPMV and CPMV-eVLP (labelled with black asterisks (*), Fig. 1c), were sequenced. Sequence analysis revealed that all seven Affimers were unique. A prerequisite for our analysis of Affimer proteins was the selection of variable loops that had a significantly different sequence, since a similar sequence in this region suggests the binding sites are most likely identical. Three Affimer proteins (2, 11 and 24) were not taken forward for further analysis as the sequence in their variable loops was close to that of at least one of the other sequenced Affimer proteins. The remaining four Affimer proteins (3, 9, 17 and 23) were chosen for protein production (labelled with red asterisks (*), Fig. 1c). These were sub-cloned with a C-terminal His-Tag into pET11 expression vectors and subsequently produced in E. coli and purified for further analysis (see Methods and Supplementary Fig. 1a).
Wild type CPMV and CPMV eVLP have the same antigenicity. (a) Overlay of published CPMV structures, 5a33, 5a32, 5FMO, 5MS1, 5MSH. The RMSD between these structures is 0.4. The Small (S) subunit is coloured blue and the Large (L) subunit is coloured green. (b) Coomassie blue-stained SDS-PAGE gel to show the total protein concentration of CPMV samples compared with a western blot which demonstrates that a polyclonal antiserum raised against WT CPMV detects both the L and S subunits of WT CPMV-T, M and B and eVLP-CPMV with the same efficiency. (c) Phage ELISA of Affimer proteins from 24 clones incubated in wells containing immobilised empty virus like particles (eVLPs) (pink), WT CPMV (blue) and a negative control (green), showing the 3,3′,5,5′-tetramethylbenzidine (TMB) product absorbance at 620 nm after 2 minutes. Clones labelled with a black asterisk (*) were selected as Affimer proteins suitable for detecting both eVLP and WT CPMV and were sequenced. Clones labelled with a red asterisk were identified as Affimer proteins with different sequences at the variable loops and used for further testing.
Structure determination of CPMV bound to Affimer 3
To analyse the binding mechanism of CPMV to an Affimer reagent we chose to determine the structure of an Affimer:CPMV complex using cryoEM. Initially all four Affimer proteins were analysed using negative stain EM in complex with CPMV to find the Affimer reagent most suitable for cryoEM. Anti-CPMV Affimer 3 was selected as it did not cause appreciable aggregation or clumping of CPMV particles in negative stain screens. A negative stain dataset was collected and a preliminary 3D reconstruction was produced in RELION. However, the resolution was not sufficient to visualise the bound anti-CPMV Affimer 3, probably due to the size of the Affimer (11 KDa).
In an attempt to obtain a high resolution structure of the Affimer:CPMV complex, cryoEM grids were produced. Mixing CPMV with anti-CPMV Affimer 3 prior to loading onto a cryoEM grid and plunge freezing caused catastrophic aggregation, and the resulting grids were not suitable for high resolution data collection18. To overcome this issue CPMV was applied to lacey carbon grids overlaid with a thin (<3 nm) layer of carbon. This causes the CPMV particles to be immobilized on the carbon support, anti-CPMV Affimer 3 was subsequently applied, excess solution was removed and the grid was finally washed prior to blotting and plunge freezing (see Methods and18 for more details). This method resulted in high quality cryoEM grids with particles that were evenly distributed and thus ideal for data collection (Fig. 2a). Data were collected and an EM density map produced (see Methods). The global resolution of the resulting structure was 3.4 Å (Fig. 2, Table 1). The viral capsid has the highest resolution (Fig. 2c), and previous atomic models for the L and S subunit fit into the density with clear resolution of the bulky side-chains in both capsid proteins (Fig. 2d). The L subunits (green) interact at the 2-fold symmetry axis of the icosahedral capsid and here we can see additional density that is attributed to anti-CPMV Affimer 3 (purple, Fig. 3a). Due to the icosahedral averaging applied during image processing and structure determination, these locations have two-fold symmetry applied, which is appropriate for the viral CP, but not for the bound Affimer. As a result, the non-two-fold symmetric Affimer structure is inappropriately averaged meaning there is not clear density for the backbone of the Affimer and we are unable to map the position of individual amino acids. This is consistent with the Affimer protein being flexible relative to the virus CP as the resolution appears to be worse at the “tip” of the Affimer protein (Fig. 2c). There also appears to be a gap in density at the “tip” of the Affimer, presumably due to a high degree of flexibility. It is therefore difficult to interpret the resulting structure. We visualise the Affimer reagent bound to each of the twenty 2-fold symmetry axis. Due to symmetry averaging it is difficult to determine the occupancy of the Affimer reagent at these binding sites. The EM density attributed to the Affimer reagent is not as strong as the CP EM density but it is similar, therefore we believe the sites are substantially occupied. EM density is visualized for the alpha helical domain of the Affimer with the base of the helix bound at the interface between two L subunits (Fig. 3b). The Affimer’s beta strands are not visible at high resolution (Fig. 3b), suggesting that there is some flexibility in the binding of the Affimer. If we analyse the unsharpened EM density map which retains low-resolution information, the density corresponding to the β sheet is visible, but the strands are not resolved (Fig. 3b). The variable loop of the Affimer is the location for molecular recognition and therefore this region should be specifically bound to the capsid. Previous structural analysis of Affimer proteins has shown the location of the variable loop (Supplementary Fig. 1b). Due to the location of the variable loops, we propose that the N-terminus of the Affimer reagent is facing down into the capsid as this would allow the Affimer variable loops to bind to a portion of the L subunit in the CPMV capsid (Fig. 3b).
Cryo-Electron Microscopy of CPMV bound to anti-CPMV Affimer 3. (a) A representative cryoEM motion corrected micrograph of CPMV bound to anti-CPMV Affimer 3. Scale bar is 100 nm. (b) 2D class averages calculated in Relion1.4. (c) 3D cryo-EM reconstruction of CPMV bound to anti-CPMV Affimer 3 coloured by local resolution. The highest resolution region of the map is ~ 3.3 Å and is located at the viral capsid (blue/green). The lowest resolution features of the map (>4 Å) are the bound anti-CPMV Affimer proteins (red). A key is shown for reference. (d) Representative cryoEM density to demonstrate side chain density in the CP. (e) Fourier shell correlation (FSC) curve of the masked map, unmasked map and corrected map. The resolution reported here was according to the 0.143 criterion.
Anti-CPMV Affimer 3 binds to CPMV at the 2-fold axis. (a) 3D cryo-EM reconstruction of CPMV bound to anti-CPMV Affimer 3. The CPMV S subunit is coloured blue and the L subunit is green. Extra EM density attributed to anti-CPMV Affimer 3 is shown in purple. Icosahedral (I1) symmetry was imposed during image processing. The 2-fold axis is indicated with a dashed red line. (b) Zoomed in view of bound anti-CPMV Affimer 3 (purple) side and top views. The crystal structure of an Affimer (PDB 4N6U) is fitted into the density. As anti-CPMV Affimer 3 is bound to the 2-fold axis, density is not visualized for the beta strands due to averaging required for image processing. The sharpened map contains high resolution information and only the alpha helical portion of the Affimer can be visualized. The unsharpened map contains low resolution information where density for the beta strands can be visualized. The 2-fold axis is indicated with a dashed red line.
Affimer 3 specifically binds to CPMV
For any binding reagent to be useful in diagnosis, it is essential that it does not produce false positive results. Therefore, we investigated whether the anti-CPMV Affimer 3 was specific for CPMV, or whether it would recognize the similar structural features of other icosahedral viruses. We therefore immobilized equal concentrations (10 µg/ml) of samples of the following VLPs; Murine Norovirus (MNV), Satellite Tobacco Necrosis Virus (STNV), Hepatitis B (Hep B) and Potato Leaf Roll Virus (PLRV) using a biotin-streptavidin linkage, and analyzed the ability of the anti-CPMV Affimer 3 to recognize them using an ELISA (Fig. 4a). Anti-CPMV Affimer 3 did not bind to any of the viruses tested but showed specificity for CPMV (both WT and eVLP).
Anti-CPMV Affimer 3 is specific to CPMV. (a) ELISA of immobilized VLPs – Murine Norovirus (MNV), Satellite Tobacco Necrosis Virus (STNV), Hepatitis B (Hep B) and Potato Leaf Roll Virus (PLRV) detection using anti-CPMV Affimer 3. Only CPMV and eVLP-CPMV were detected by anti-CPMV Affimer 3. (b) A schematic describing a sandwich ELISA. Anti-CPMV Affimer 3 is immobilized on a maxisorb plate, if CPMV is present it binds to anti-CPMV Affimer 3. Biotinylated anti-CPMV Affimer 3 binds to the CPMV capsid (if present) which then binds to streptavidin-HRP. HRP oxidises TMB to produce a colour change which is monitored by absorbance. (c) A sandwich ELISA of a blind test to detect the presence of CPMV in infected leaves. 6 leaves were a negative control and 6 leaves were infected with CPMV (diluted to either 1 in 5 or 1 in 20). Leaf extract numbers 2, 4, 5, 6, 9 and 11 were positive. These crude leaf extracts also showed increased absorbance demonstrating the detection method was successful. Error bars show standard deviation from the mean.
To further test the suitability of anti-CPMV Affimer 3 for future in-field diagnostics, we tested the ability of anti-CPMV Affimer 3 to detect CPMV in crude extracts from infected leaves using a sandwich ELISA (Fig. 4b). In the field if a plant was found to be infected with CPMV (or another plant virus) the plant would be destroyed to limit the spread of the infection to other plants. Purified CPMV was used as a positive control to optimize the ELISA conditions. A sandwich ELISA requires two binding reagents, in this case two Affimer reagents; a “capture Affimer” which is immobilized on a plate and a “detection Affimer” which will bind to CPMV and produce a colour change via a Biotin-Streptavadin-HRP complex on addition of TMB (if CPMV is present) (Fig. 4b). The “detection Affimer” is cysteine terminated which permits conjugation to a biotin linker. The biotin is used for detection of a positive result in a sandwich ELISA as the cysteine linked “detection Affimer” will bind to CPMV if present. The “detection Affimer” is then able to bind to Streptavidin bound to HRP which will produce a colour change in TMB that can be monitored using absorbance in a plate reader (Fig. 4b). The “capture Affimer” does not require biotin conjugation and therefore Affimer proteins are usually produced without cysteines. This means Affimer reagents are easier to produce and the lack of a cysteine is useful in many downstream experiments. To find the optimal parameters for the type of capture anti-CPMV Affimer 3 (i.e. cysteine-terminated or “normal” (non-cysteine) Affimer), the concentration and incubation time of anti-CPMV Affimer 3 were determined (Supplementary Fig. 2a). When used as “capture” reagent, the cysteine-terminated anti-CPMV Affimer 3 displayed increased background compared to the non-cysteine anti-CPMV Affimer 3 (*, Supplementary Fig. 2a), therefore, the non-cysteine anti-CPMV Affimer 3 was used as the “capture Affimer” in subsequent ELISAs. The greatest absorbance was achieved with overnight adsorption at 4 °C of 50 µg/ml anti-CPMV Affimer 3 and with 0.5 µg/ml biotinylated cysteine-terminated detection anti-CPMV Affimer 3 (green box, Supplementary Fig. 2). Therefore, for the “capture Affimer”, 50 µg/ml of cysteine free “normal” Affimer reagent was incubated overnight at 4 °C and 0.5 µg/ml biotinylated cysteine terminated Affimer reagent was used as “detection Affimer”. These conditions were used in all the following ELISAs. To extract the CPMV particles from infected leaves two methods were compared (Supplementary Fig. 2b and see methods). We intentionally kept the preparation of leaves as simple as possible without the requirement for scientific equipment. Upon analysis of the most dilute leaf extractions, the highest absorbance was achieved by manually grinding the leaves in PBS buffer and so this method was used in subsequent analysis.
Blind testing was performed in which twelve leaves were randomly numbered 1–12 (6 of which were inoculated with CPMV, and 6 of which were uninoculated negative controls in a blind experiment). The leaves were extracted and analysed using the optimised sandwich ELISA conditions described above (Supplementary Fig. 2a). This extraction and sandwich ELISA was repeated three times, using a fresh sample of the numbered leaf each time (Fig. 4c). The ELISA correctly identified the CPMV-infected samples, with significant absorbance being observed from inoculated leaves (leaves 2, 4, 5, 6, 9, 11) and little absorbance (i.e. below a threshold of 0.2 optical density) was observed for the negative controls (leaves 1, 3, 7, 8, 10, 12) demonstrating that anti-CPMV Affimer 3 can specifically detect CPMV infected leaves (Fig. 4c).