Effect of foot-and-mouth disease virus 3 C protease B 2 b-strand proline mutagenesis on expression and processing of the P 1 polypeptide using a plasmid expression vector

The production of experimental molecular vaccines against foot-and-mouth disease virus utilizes the viral encoded 3C protease for processing of the P1 polyprotein. Expression of wild type 3C protease is detrimental to host cells. The molecular vaccine constructs containing the 3C protease L127P mutant significantly reduce adverse effects associated with protease expression while retaining the ability to process and assemble virus-like particles. In published 3C protease crystal structures, the L127 residue is contained within the B2 b-strand as part of the A2–B2 b-sheet. To provide insight into the mechanism by which the L127P mutant alters the properties of the 3C protease, we performed scanning proline mutagenesis of residues 123–128 of the B2 b-strand and monitored expression and P1 processing. Simultaneously, we utilized random mutagenesis of the full 3C sequence to identify additional mutations presenting a phenotype similar to the L127P mutation. Six of the tested mutants enhanced expression over wild type, and the I22P, T100P and V124P mutations surpassed the L127P mutation in certain cell lines. These data areinterpreted in conjunction with published 3C protease crystal structures to provide insight into the mechanism by which these mutations enhance expression.


INTRODUCTION
Foot-and-mouth disease virus (FMDV) is an Aphthovirus member of the Picornaviridae family and the cause of footand-mouth disease, characterized by lameness and loss of productivity among most cloven-hooved livestock species.Approximately 2.35 billion doses of FMDV vaccines are administered yearly [1].The most commonly used vaccine is derived from chemically inactivated live virus typically produced in the BHK-21 cell line [2][3][4][5].While highly effective, manufacturing of inactivated FMDV vaccines carries the risk of an accidental outbreak.Next-generation molecular vaccines utilize only the FMDV proteins necessary for assembly of virus-like particles (VLPs), eliminating the risk of accidental outbreaks.
Molecular vaccines require expression of the FMDV 3C protease (3C pro ) for processing of the FMDV P1 polyprotein into structural proteins, VP0, VP1 and VP3, critical for the formation of VLPs [6][7][8][9][10][11][12].The separation of VP0 into VP4 and VP2 is not processed by 3C pro .In polio it is dependent on the packaging of RNA [13,14], but can occur independently of RNA in FMDV [15].The highly conserved 3C pro utilizes a catalytic triad of H46, D84 and C163 to induce processing of the FMDV polyprotein and multiple host proteins critical for cell functions [16][17][18][19][20]. Expression of wild type 3C pro results in the processing of host proteins critical for cell functions, restricting the viability of vaccine platform options.
The recently reported 3C pro mutant, L127P, reduces processing of host cell proteins while maintaining FMDV P1 processing [12].The L127P mutation was identified using a plasmid expressing the FMDV 2A protein, Gaussia luciferase (GLuc), and the 3C pro .The FMDV 2A protein is able to produce separate proteins from a single open reading frame [21,22] while GLuc is a naturally secreted luciferase [23].This system allows for monitoring the effects of 3C pro expression on the yield of plasmid expressed antigen in transfected cell cultures [8,12].We sought to investigate the role played by proline disruption of the B 2 b-strand in the phenotype observed with the L127P mutation, by utilizing scanning proline mutagenesis on residues 123-128 and the same GLuc reporter assay previously used to identify and evaluate the L127P mutation.Additionally, following random mutagenesis, analysis of novel 3C pro mutants identified three mutations, I22P, L23P and T100P, which enhanced plasmid protein expression and retained P1 processing.The availability of published crystal structures for 3C pro allows interpretation of presented results in relation to residue location and overall 3C pro structure.

MATERIALS Isolation of pJJP mutant constructs
Constructs were synthesized into the pJJP plasmid by Gen-Script and contained nucleotide sequences encoding the P1-2A polypeptide from FMDV O1 Manisa iso87 (GenBank accession no.AY593823), the FMDV Asia 1 Shamir 3C (GenBank accession no.AY593800) and the D1D2A-SGLuc D1M-secreted biomarker [8].Select cut sites were removed using silent mutations to facilitate easier cloning.Constructs were transformed into NEB 5-Alpha competent Escherichia coli (New England Biolabs) according to the manufacturer's instructions for plasmid propagation.Cultures were grown at 37 C overnight at 250 r.p.m. in Terrific broth with 100 µg carbenicillin ml À1 .Plasmids for transfection were isolated using the QIAGEN EndoFree Plasmid Maxi kit (Qiagen).

Random mutagenesis of 3C pro
Random mutagenesis of FMDV 3C pro was performed utilizing a Diversify PCR Random Mutagenesis Kit (Takara), as recommended by the manufacturer, for two mutations per 1000 bp.The template was a synthesized WT FMDV Asia 1 Shamir 3C (GenBank accession no.AY593800) in pUC57kan (GenScript), while primers utilized for the amplification were Nde-3C-F2 (CATATGAGTGGTGCCCCACCGAC) and Rand-XhoXba-R (CCGATTCTAGACTCGAGTTA).PCR clean-up was performed using a QIAquick PCR Purification kit (Qiagen), and PCR product digested by restriction enzymes NdeI and XhoI (New England Biolabs) as suggested by the manufacturer.Digested PCR product was cloned into a pSNAP-tag (T7)À2 vector (New England Biolabs), which had been similarly digested, using T7 ligase (Roche) as per the manufacturer's suggestions.

Bacterial plating and expression
Transformation of T7 Express Competent E. coli (New England Biolabs) was performed as suggested by the manufacturer.Prior to plating, transformation was put in 10 ml of 100 µg carbenicillin ml À1 Terrific broth for 3 h at 37 C.After incubation, 200 µl was plated on fresh +IPTG/+X-GAL/+CARB LB plates (Teknova) for colony growth and incubated at 37 C overnight.Isolated colonies were grown in 4 ml of +CARB LB media overnight for plasmid isolation using a QIAprep Spin Miniprep kit (Qiagen).Purified plasmids were sequenced using primers T7 (TAATACGAC TCACTATAGGG) and pSNAP-SR (CGGATATAGTTCC TCCTTTC) for determination of nucleotide sequence.Plasmids encoding mutants of interest were then re-transformed into T7 Express Competent E. coli, cultured and plated as previously described [12], to validate whether or not expression had an effect on colony growth.
Evaluation of plasmid protein expression and processing HEK293-T cells (ATCC CRL-3216) were grown in six-well plates (VWR) using 293 growth media (1Â DMEM, 10 % fetal bovine serum, 1Â antibiotic-antimycotic, 1Â nonessential amino acids).For transfections, cells were rinsed with 2 ml of DPBS, and 1 ml of fresh media was added to each well prior to transfection with 4 µg of plasmid DNA and 10 µl of lipofectamine 2000 (Invitrogen).Cells were incubated with 5 % CO 2 at 37 C overnight prior to harvesting of cells.
Luciferase activity was detected in media separated from transfected cells using a 96-well BioSystems Veritas luminometer (Turner Biosystems) with 100 µl of 1 : 4 transfection media-diluted sample in each well.Readings were taken immediately following injection of 100 µl of 50 µg water-soluble coelenterazine l À1 (NanoLight Technologies, Pinetop AZ) using an integration time of 0.5 s both before and after injection of substrate.
Transfected HEK293T cell lysates were removed from plates using 500 µl of M-PER (Invitrogen).After cryofracturing, samples were mixed with 10 µl DNAse and incubated for 1 h at 37 C. Following incubation, 60 µl of sample was mixed with 30 µl of 4Â NuPage LDS Sample Buffer (Invitrogen), heated at 97 C for 10 min and loaded into wells of 10well NuPage 4-12 % Bis-Tris gel (Invitrogen).Gels were electrophoresed in 1Â MES buffer (Invitrogen) at 200 V for 35 min prior to transfer onto membranes using the iBlot 2 system (Invitrogen).
Membranes were incubated in 5 % milk blocking buffer for 1 h at room temperature, then washed three times with 1Â PBS-T (EMD Millipore) for 5 min each.Primary antibodies were added at 1 : 50 dilution for mouse monoclonal antibodies F1412SA [27] and 12FE9.2.1 (28), at 1 : 250 dilution for an anti-VP3 rabbit polyclonal antibody and at 1 : 1000 dilution for eIF4AI (ab31217, ABCAM), and incubated for 1 h at room temperature.Membranes were washed three times with 1Â PBS-T for 5 min after the primary antibody incubation, and 1 : 500 dilutions of the secondary antibodies, goat anti-mouse-HRP or goat anti-rabbit-HRP (KPL) were applied to appropriate membranes for 1 h at room

CHO-K1 cell culture work
Chinese hamster ovary (CHO-K1) cells (ATCC CCL-61) were cultured on six-well plates (Costar) in the recommended F-21K Media (ATCC) supplemented with 5 % FBS.Transfection, cell lysate harvesting and luciferase assays were performed as described above.Western blots were performed as described above, with the exception of using 1-Step Ultra TMB Blotting solution (ThermoFisher) in place of SIGMAFAST 3,3'-diaminobenzidine tablets.

Transmission electron microscopy
Transmission electron microscopy utilized a Hitachi 7600 with a 2kÂ2 k AMT camera at 80 kV.Sample preparation was performed as previously described [12].

Mutations of B 2 b-strand residues
In FMDV 3C pro crystal structures [24][25][26], the L127 residue is a component of the B 2 b-strand, forming a b-sheet with the A 2 b-strand.Neither strand contains any member of the catalytic triad, nor do they comprise the substrate-binding pocket (Fig. 1).To investigate whether the effect of L127P mutation is residue specific, we performed scanning proline mutagenesis for each residue comprising the B 2 b-strand in published crystal structures, residues 123-128 (Fig. 1).We constructed expression plasmids encoding the FMDV O1 Manisa (O1M) P1-2A polypeptide, 3C pro and the D1D2A- GLuc D1M biomarker.The D1D2A-GLuc D1M biomarker utilizes the FMDV 2A protein to produce a secreted luciferase allowing for easy and rapid quantification of protein expression from plasmids transfected into cell culture [8].Not including L127P or the C163A negative control, four of the five tested residues resulted in an enhancement of expression over wild type (WT) (Fig. 2a).The L127P mutation maintained the highest enhancement in expression over WT among the B 2 b-strand proline mutants, while mutant V124P also showed strong enhancement (P<0.0001 by Student's t-test).Mutants D123P, G125P and I128P were equivalent and enhanced expression over WT but not to the same level as V124P and L127P.Interestingly mutant R126P had no difference in expression from WT (Fig. 2a).
Lysates from transfected HEK293-T cells were examined for FMDV O1M P1-2A polypeptide processing (Fig. 2b).All mutations, with the exception of the C163A activity knockout, retained the ability to process the O1M P1 polypeptide and displayed fully processed VP2, VP3 and VP1 (Fig. 2b).

Residues identified by random mutagenesis
Identification of the L127P mutation was the result of polymerase infidelity in the process of cloning [12].To identify novel mutations in different structural regions of the 3C pro with similar phenotypes, we utilized a commercially available random mutagenesis kit to enhance polymerase infidelity and repeated the cloning of 3C pro into a bacterial expression vector.Previous reports observed that bacteria expressing WT 3C pro grew poorly on plates, [12], and subsequently we selected colonies which grew under protein expression inducing conditions to identify potential mutations of interest.Five mutations were identified, two of which contained premature stop codons and were not evaluated further.The remaining mutants, containing T100P, L23P or L42H, were re-plated on both +CARB/+IPTG and +CARB plates to confirm reduced adverse effects.E. coli with L23P or T100P had enhanced colony growth confirmed (Fig. 3), while L42H was determined to be a false positive (data not shown).
Since both L23P and T100P represent mutations to proline, we also included adjacent proline mutants I22P and I99P to understand the impact of proline in these regions.HEK293-T cells were transfected with pJJP plasmids expressing either WT, I22P, L23P, I99P, T100P, V124P, G125P, R126P, L127P or C163A for comparison of enhancement of plasmid-derived protein expression.Mutations identified outside of the A 2 -B 2 b-sheet, I22P, L23P, I99P and T100P, all showed an enhancement of expression when compared to WT, with the I22P mutation showing expression levels greater than L127P (Figs 4a and S1, available in the online version of this article).
Lysates from transfected cells were evaluated for O1M P1 processing by western blotting.Mutants showed different degrees of processing efficacy (Fig. 4b).Mutant T100P showed processing and the presence of VP2.However, it also showed unprocessed P1 and a weak fusion peptide of VP3 and VP1 (VP3-VP1).Both I22P and L23P showed similar processing profiles with VP3-VP1 present, suggesting difficulties in processing the VP3-VP1 junction (Fig. 4b).In spite of this, arrays of VLPs were observed in HEK293-T cells transfected with pJJP plasmids containing the I22P mutant (Fig. 4c).Mutant I99P enhanced plasmid expressed protein output over WT but failed to show any processing of the O1M P1-2A polypeptide (Fig. S1).

Combination of identified mutations
Identified mutations represent alterations to different structures of the 3C pro .Residues V124 and L127 are part of the B 2 b-strand, I22 and L23 are on the A 1 b-strand and T100 is on a loop near the F 1 b-strand.We evaluated whether utilizing mutants I22P, T100P or V124P, in combination with L127P, would further enhance protein expression.Three double mutation constructs, I22P/L127P, T100P/L127P and V124P/L127P, were expressed and evaluated for both luciferase output and O1M P1-2A processing.Plasmid-derived protein expression for all combinations was below the levels observed with L127P alone (Fig. 4a), with only V124P/ L127P retaining a high degree of processing and the presence of VP2 (Fig. 4b).The I22P/L127P construct retained weak processing but failed to show any VP2 and only Host eIF4AI processing 3C pro processes a critical component of the eukaryotic cell's translational machinery, eIF4AI.Transfected HEK293-T cell lysates were examined for host cell eIF4AI processing by single mutants I22P, L23P, T100P, V124P, G125P and L127P, as well as double mutants I22P/L127P and V124P/ L127P, for comparison to WT and C163A (Fig. 5).Strong bands corresponding to processed eIF4AI were observed in WT and R126P; a weak band was observed in G125P, and a very faint band was detectable in V124P (Fig. 5).

Production cell line expression
Human eIF4GI is reported as resistant to processing by FMDV 3C pro due to a proline at residue 713 [28].Previous work utilizing the L127P mutation relied heavily on the human-derived HEK293-T cell line [12].To test the broader applicability of these mutations, transfections of cell lines used for either pharmaceutical production (CHO-K1) or FMDV research (BHK-21) were performed utilizing a subset of tested mutations.
In CHO-K1 cells mutant L127P produced the highest luciferase levels among single mutations.All mutations that enhanced expression over WT in HEK293-T cells also enhanced expression over WT in CHO-K1 cells (blue bars in Fig. 6).
Transfection of BHK-21 cells (red bars) with mutant I22P yielded the highest luciferase levels, and similar levels were associated with V124P (Fig. 6).Mutant L127P showed reduced enhancement relative to I22P, T100P and V124P, but all were greater than WT (Fig. 6).
Processing efficiency of tested 3C pro mutants in both CHO-K1 and BHK-21 cell lines was similar to that observed in HEK293-T cells (Fig. 7a).Electron microscopy confirmed VLP formation with the L127P mutant in transfected CHO-K1 cells (Fig. 7b), and with the V124P mutant in transfected BHK-21 cells (Fig. 7c).

DISCUSSION
The identification of the L127P mutation to FMDV 3C pro [12] led to the question of how a single mutation could cause such a dramatic change in proteolytic activity.We have shown that multiple mutations to the B 2 b-strand can result in enhancement of plasmid-derived protein expression and reduction of host eIF4AI processing.Furthermore, we identified mutations to other structures of the 3C pro that result in a similar phenotype.

Relation of B 2 b-ribbon mutants to published structures
In published 3C pro crystal structures [24][25][26], the A 2 -B 2 bsheet is not a component of the substrate-binding pocket but is in close proximity to the loop containing the C163 residue, giving rise to speculation that it plays a role in substrate specificity [24].Our results appear to validate this hypothesis, as processing of the P1-2A is retained while host cell eIF4AI processing is lost or impaired for almost all B 2 b-strand mutants.The dramatic differences in expression seen among proline mutants of the B 2 b strand suggest that location of the mutated residue is critically important.
Despite being adjacent to residue L127, mutant R126P showed no enhancement to expression and it retained host eIF4AI processing.The published crystal structure 2WV5 shows the amino group of R126 as distal to any carbonyl on the A 2 b-strand, while the amino group of L127 is in position to form a hydrogen bond with the carbonyl of residue V119 at a distance of 2.1 Å (Fig. 8a).This provides an explanation for why R126 showed little difference from WT despite being adjacent to L127, as R126 does not contribute to the hydrogen-bonding stabilization of the A 2 -B 2 b-sheet.Conversely, the loss of hydrogen-bonding potential of the L127 amino group results in the loss of a key stabilization interaction.Mutations D123P, G125P and I128P moderately enhance expression compared to WT.One explanation for this moderate enhancement is that they represent the result of restrictions in torsional angles conferred by the presence of proline on the polypeptide backbone but not a loss in hydrogen bonding.Residue D123 is part of the turn between the A 2 and B 2 b-strands and the loss of any hydrogen bonding to the amino group, if present, is unlikely to be a factor in its enhancement.Glycine, like proline, has unique structural properties because the lack of a side chain confers on it greater structural flexibility, making the mutation of glycine to proline a mutation from the most flexible amino acid to the least.Like R126, the amino group for I128 is too distal from the A 2 b strand to form a hydrogen bond (Fig. 8a).The mutation of I128 to proline may influence the amino group of residue of F129, 2.0 Å from the carbonyl group of residue G117 on the A 2 b-strand, through restrictions on the torsional angles of the polypeptide backbone Mutant V124P showed a significant enhancement over WT (P<0.0001 by Student's t-test).One explanation for this is the influence of glycine at residue 125 in conjunction with the mutation of V124 to proline.Residue V124 may offset the flexibility of glycine at residue 125, possibly through the formation of a hydrogen bond with residue N121 at a distance of 2.0 Å, minimizing disruption of the b-strand structure.In mutant V124P this stabilization may be removed, and the combination of the proline-glycine pair at residues 124 and 125 may cause greater disruption of secondary structure than either alone.
It is also possible to examine this dataset in relation to the side chains of the mutated residues.In the B 2 b-strand, the guanidinium side-chain for R126 is distal to the side chain of residue Y162, as opposed to V124 and L127 whose sidechains are directly adjacent (Fig. 8b).Residue Y162 is adjacent to C163, a component of the catalytic triad, providing a means by which A 2 -B 2 b-sheet structural alterations, and/or changes in side-chain structure, could influence substrate binding and processing.The subset of moderately enhancing mutations, D123P, G125P and I128P, may all cause displacement of the side-chains of V124 and L127 through changes in the torsional angle of the polypeptide backbone.

Relation of I22P, L23P and T100P mutants to published structures
Random mutagenesis led to the identification of three mutants outside of the A 2 -B 2 b-sheet that enhance output while retaining P1 processing, I22P, L23P and T100P (Fig. 9a).Residues I22P and L23P are contained in the A 1 bstrand (Fig. 9b), while T100P is situated on a loop structure adjacent to the F 1 b-strand and localized between the N-and C-terminal helices (Fig. 9c).The A 1 -B 1 b-strand pair, residues 18-31, is adjacent to the peptide-binding pocket (Fig. 9b) and undergoes conformational changes upon binding to substrate [25].This suggests that mutations I22P and L23P may result in alterations of the structure of the substrate-binding pocket and consequently substrate specificity.
Mutants I99P and T100P represent dramatically different P1 processing efficiencies despite being adjacent.Both are localized in the loop adjacent to the F 1 b-strand (Fig. 9d), in close proximity to the N-and C-terminal a-helices but from www.microbiologyresearch.orgby IP: 54.70.40.11On: Fri, 02 Aug 2019 21:10:29 A mutation to proline is notable because proline is a known disrupter of b-sheet structures, due to its cyclic structure preventing formation of the hydrogen-bonding network between b-strands and placing restrictions on the torsional angles of the polypeptide backbone.Published crystal structures of 3C pro , 2WV5, 2BHG and 5HM2, show residue L127 as part of the B 2 b-strand which, along with the B 2 ' bstrand, comprises a b-sheet with the A 2 b-strand [24-26].

Fig. 1 .
Fig. 1.The locations of the A 2 -B 2 b-sheet (red and blue), catalytic residues (orange) and substrate (green) are shown using the 2WV5 structure [25].A focus on the A 2 -B 2 b-sheet shows side-chain location in relation to the A 2 -B 2 b-sheet structure.

Fig. 2 .
Fig. 2. Effect of scanning proline mutagenesis on residues 123-128.(a) Luciferase readings for B 2 b-ribbon proline mutants in comparison to WT and C163A (a 3C pro activity knockout used as a negative control).The R126P mutant is closest to WT, 8.04Â10 8 and 9.23Â10 8 RFU 0.5 s -1 , respectively.Each bar represents six readings each for three separately transfected wells averaged together.(b) Western blot of transfected HEK293-T cell lysates for B 2 b-ribbon proline mutants.

Fig. 3 .
Fig. 3. Validation of selected random mutants in E. coli.E. coli transformed with WT, L23P or T100P 3C pro grown on +CARB or +CARB/ +IPTG plates for evaluation of the effect of 3C pro expression on bacterial colony growth.Plating under inducing conditions (+IPTG) resulted in a dramatic reduction of colony-forming units for bacteria expressing WT, but little observable reduction for bacteria expressing L23P or T100P.

Fig. 4 .
Fig. 4. Comparison of random and site-directed proline mutants.(a) Luciferase readings for HEK293-T cells transfected with pJJP plasmids containing 3C pro proline mutations at residues 22, 23 and 100, along with previously tested proline mutations 124-127 and double-mutant constructs.Each bar represents six readings each for two separately transfected wells averaged together.(b) Western blots of corresponding cell lysates that showed processing.(c) Arrays of O1 Manisa VLPs in HEK293-T cells transfected with the I22P mutant pJJP plasmid.

Fig. 6 .
Fig. 6.Evaluation of expression for selected mutants in additional cell lines.Luciferase readings for CHO-K1 (blue) and BHK-21 (red) cells transfected with pJJP plasmids containing 3C pro proline mutations at residues 22, 23, 100, 124, 125, 126 and 127 and double-mutant combinations.Each bar represents six readings each for two separately transfected wells averaged together for CHO-K1 cells, and seven readings from a single transfected well for BHK-21 cells averaged together.

Fig. 7 .
Fig. 7. Evaluation of P1 processing for selected mutants in additional cell lines.(a) Western blots of transfected CHO-K1 and BHK-21 cell lysates utilizing F14 and 12FE9 monoclonal antibodies.(b) Arrays of O1 Manisa VLPs in CHO-K1 cells transfected with pJJP plasmid containing the L127P mutation.(c) VLPs formed in BHK-21 cells transfected with pJJP plasmid containing the V124P mutation.

Fig. 8 .
Fig. 8. Proline mutagenesis implications on A 2 -B 2 b-sheet structure.(a) Structure of the amino acid backbone comprising the A 2 -B 2 b sheet.Amino groups of critical residues are identified with black arrows.Residues 128, 126 and 124 are highlighted in yellow, and residues 117, 119 and 121 are highlighted in green.Distances between atoms is in angstroms and denoted by dotted lines.(b) Structure of the A 2 -B 2 b-sheet relative to the loop containing the Y162 and C163A residues.