Figure 1.
Determination of SARS-CoV 3CLpro proteolytic rate using the protein substrate by FRET assay.
(A) Schematic diagram illustrating the principle of the FRET assay. The autocleavage sequence of 3CLpro is inserted between CFP and YFP in the protein substrate. When CFP is excited at 430 nm, YFP emits fluorescence at 530 nm through FRET. Cleavage of the peptide bond at SAVLQ↓SGF by 3CLpro separates CFP and YFP, leading to a decrease in the emitted fluorescence at 530nm. (B) After digestion by 4 µM of 3CLpro for one hour, the protein substrate of 58 kDa (lane 1) was separated into two products of 28 kDa and 30 kDa (lane 2). (C) The protein substrate cleaved by 1 to 4 µM of 3CLpro led to a time-dependent decrease in fluorescence at 530 nm. Observed rate constant, kobs, was obtained by fitting the data to a single exponential decay. (D) The plot of kobs against [3CLpro] yielded a straight line. The specific activity, kobs/[3CLpro], was determined by the slope of the plot.
Figure 2.
Profiling the substrate specificity at P5 to P3' positions.
198 single substitution variants were created by saturation mutagenesis of the autocleavage sequence at P5 to P3' positions. Specific activity on each of variants was determined, and normalized by that on WT substrate to obtain the relative activity.
Table 1.
Correlation between SARS-CoV 3CLpro activity and structural properties of substituting residues.
Figure 3.
Substrate specificity for SARS-CoV 3CLpro.
The relative activity significantly correlated with various structural properties of substituting residues. (A) At P5 position, the relative activity correlated well with the β-sheet propensity (r = 0.711, p<0.001). (B) At P4 position, significant correlation was observed for hydrophobicity (r = 0.587, p = 0.008). The correlation was improved (r = 0.942, p<0.001) when only residues with side chain volumes of <80 Å3 (Ala, Asn, Asp, Cys, Glu, Gly, Pro, Ser, Thr and Val) were included. (C) The relative activity on P3 variants were correlated with β-sheet propensity (r = 0.510, p = 0.022). Increase in the correlation (r = 0.729, p = 0.001) was found after neglecting charged residues (Arg, Asp, Glu and Lys). (D) Only variants with hydrophobic residues (Ala, Cys, Ile, Leu, Met, Phe, and Val) at P2 position were cleavable. (E) The relative activity on P1' variants with side chain volume of <50 Å3 (Ala, Cys, Gly and Ser) were higher than that on others.
Figure 4.
Modeling how 3CLpro recognizes P1-His.
In the 3CLpro-substrate complex (PDB: 2Q6G), amide group of P1-Gln of the WT substrate sequence (thin stick) forms hydrogen-bonds with the Nε2 atom of His163 and the backbone carbonyl group of Phe140. P1-Gln was substituted to His (thick stick) in silico using the program SWISS-PDBViewer [24]. The rotamer of P1-His was selected to avoid steric hindrance and to optimize for hydrogen bond formation. The modeled structure was then energy minimized using a GROMOS force-field implemented in SWISS-PDBViewer. It was found that P1-His can fit into the substrate binding pocket and form hydrogen bond to the amide group of Asn142.
Figure 5.
Super-active substrates were created by combining the best residues at P5 to P1 positions.
Three variants with double-substitution (grey bar) and three variants with triple-substitution (solid bar) were created, and their relative activities were measured. The relative activities of FVVLQ↓SGF, TVVLQ↓SGF, VVVLQ↓SGF, FVRLQ↓SGF, TVRLQ↓SGF and VVRLQ↓SGF were 2.11±0.26, 1.87±0.19, 1.80±0.17, 2.10±0.34, 2.84±0.25 and 2.71±0.29, respectively.
Table 2.
Summary of SARS-CoV 3CLpro substrate specificity at P5 to P3' positions.