Catalytic function and substrate specificity of the papain-like protease domain of nsp3 from the Middle East respiratory syndrome coronavirus

Abstract

The papain-like protease (PLpro) domain from the deadly Middle East respiratory syndrome coronavirus (MERS-CoV) was overexpressed and purified. MERS-CoV PLpro constructs with and without the putative ubiquitin-like (UBL) domain at the N terminus were found to possess protease, deubiquitinating, deISGylating, and interferon antagonism activities in transfected HEK293T cells. The quaternary structure and substrate preferences of MERS-CoV PLpro were determined and compared to those of severe acute respiratory syndrome coronavirus (SARS-CoV) PLpro, revealing prominent differences between these closely related enzymes. Steady-state kinetic analyses of purified MERS-CoV and SARS-CoV PLpros uncovered significant differences in their rates of hydrolysis of 5-aminomethyl coumarin (AMC) from C-terminally labeled peptide, ubiquitin, and ISG15 substrates, as well as in their rates of isopeptide bond cleavage of K48- and K63-linked polyubiquitin chains. MERS-CoV PLpro was found to have 8-fold and 3,500-fold higher catalytic efficiencies for hydrolysis of ISG15-AMC than for hydrolysis of the Ub-AMC and Z-RLRGG-AMC substrates, respectively. A similar trend was observed for SARS-CoV PLpro, although it was much more efficient than MERS-CoV PLpro toward ISG15-AMC and peptide-AMC substrates. MERS-CoV PLpro was found to process K48- and K63-linked polyubiquitin chains at similar rates and with similar debranching patterns, producing monoubiquitin species. However, SARS-CoV PLpro much preferred K48-linked polyubiquitin chains to K63-linked chains, and it rapidly produced di-ubiquitin molecules from K48-linked chains. Finally, potent inhibitors of SARS-CoV PLpro were found to have no effect on MERS-CoV PLpro. A homology model of the MERS-CoV PLpro structure was generated and compared to the X-ray structure of SARS-CoV PLpro to provide plausible explanations for differences in substrate and inhibitor recognition.

Importance: Unlocking the secrets of how coronavirus (CoV) papain-like proteases (PLpros) perform their multifunctional roles during viral replication entails a complete mechanistic understanding of their substrate recognition and enzymatic activities. We show that the PLpro domains from the MERS and SARS coronaviruses can recognize and process the same substrates, but with different catalytic efficiencies. The differences in substrate recognition between these closely related PLpros suggest that neither enzyme can be used as a generalized model to explain the kinetic behavior of all CoV PLpros. As a consequence, decoding the mechanisms of PLpro-mediated antagonism of the host innate immune response and the development of anti-CoV PLpro enzyme inhibitors will be a challenging undertaking. The results from this study provide valuable information for understanding how MERS-CoV PLpro-mediated antagonism of the host innate immune response is orchestrated, as well as insight into the design of inhibitors against MERS-CoV PLpro.

FIG 1

MERS-CoV PLpro constructs, expression, and enzymatic activities in HEK293T cells. (A) MERS-CoV PLpro constructs. Wild-type (aa 1485 to 1802) (WT), catalytic cysteine mutant (Cys1594/A [aa 1485 to 1802]; CA), and three UBL-deleted mutant (N20 [aa 1505 to 1802], N40 [aa 1524 to 1802], and N60 [aa 1545 to 1802]) proteins were fused to a V5 epitope tag at the C terminus for V5 antibody detection. (B) Trans-cleavage activity of MERS-CoV PLpro in HEK293T cells expressing SARS-CoV nsp2/3-GFP. Lysates were harvested at 24 h posttransfection, and protein expression was analyzed by Western blotting. DeISGylating (C) and deubiquitinating (D) activities of MERS-CoV PLpro constructs were also examined. HEK293T cells were transfected with MERS-CoV PLpro expression plasmids for WT, CA, and UBL-deleted mutant (N20, N40 or N60) proteins, along with myc-ISG15, E1, E2, and E3 ISGylating machinery plasmids to test the deISGylating (C) activity, or with a Flag-Ub expression plasmid to test the deubiquitinating (D) activity of each PLpro construct. Cells were lysed at 18 h posttransfection and analyzed by Western blotting. The strong bands indicate ISGylated (C) and ubiquitinated (D) proteins. The figure shows representative data from at least two independent experiments.

FIG 2

Interferon antagonism activity of MERS-CoV PLpro. HEK293T cells were transfected with plasmids expressing either wild-type (WT) PLpro, catalytic mutant PLpro (CA), or a UBL-deleted PLpro mutant (N20, N40, or N60). Cells were also transfected with plasmids expressing IFN-luciferase, Renilla luciferase, and the stimulator MDA5 (indicated at the top of the figure). At 16 h posttransfection, cells were lysed and luciferase activity was measured. Experiments were performed in triplicate. Error bars represent standard deviations of the means.

FIG 3

Purification of MERS-CoV PLpro_1484–1802. (A) SDS-PAGE analysis of whole-cell lysate and purified MERS-CoV PLpro, which ran at the expected molecular mass of 37 kDa. Lane M, molecular size marker. (B) SEC-MALS traces of MERS-CoV PLpro at different protein concentrations. MERS-CoV PLpro at 4.2 mg/ml, 2.1 mg/ml, and 1.0 mg/ml eluted at the same retention time from a SEC column. The M_w, determined from the molar mass from the MALS analysis, corresponded to a monomer for the peak of each concentration. All analyzed peak areas were monodispersed (M̄_w/M̄_n value of <1.01), as shown by the horizontal traces.

FIG 4

MERS-CoV and SARS-CoV PLpro activities with three ubiquitin-based substrates. The activities of MERS-CoV PLpro (gray circles) and SARS-CoV PLpro (black circles) with each substrate are shown in panels A to C. Dose-response curves of the inhibition by free Ub and ISG15 are shown in panels D and E. Data were fit to the Michaelis-Menten equation unless the catalytic activity exhibited a linear response to the substrate concentration. In such a case, data were fit to the equation v/[E] = k_cat/K_M[S], where [E] and [S] are the concentrations of enzyme and substrate, respectively. The error bars represent the standard deviations for a minimum of triplicate samples.

FIG 5

Ubiquitin chain specificity of MERS-CoV and SARS-CoV PLpros. The in vitro cleavage of K48-linked Ub₍₅₎ (A) and Ub₍₄₎ (D) by MERS-CoV PLpro and SARS-CoV PLpro, respectively, and of K63-linked Ub₍₆₎ by MERS-CoV PLpro (B) and SARS-CoV PLpro (E) is shown. (C) Cleavage of linear Ub₍₄₎. (F) Analysis of Ub₍₂₎ accumulation during SARS-CoV PLpro-mediated processing of K48-linked substrates. Processing of the substrates is shown by the production of lower-molecular-weight bands at progressive time points, in minutes (′) or hours (h). The locations of the different Ub species are shown. Lane M, molecular size marker.

FIG 6

MERS-CoV PLpro and HCoV-NL63 PLP2 inhibition by a series of SARS-CoV PLpro inhibitors. The percent inhibition of SARS-CoV PLpro, HCoV-NL63 PLP2, and MERS-CoV PLpro activity in the presence of SARS-CoV PLpro inhibitors is shown by a graph. Percent inhibition was calculated from two independent assays with a fixed concentration of 100 μM compound, and data are shown as mean % inhibition. Error bars representing the positive and negative deviations from the average values were removed for clarity. The difference between each independent measurement was less than 10% for the entire set of data. Highlighted in bold are the best SARS-CoV PLpro inhibitor candidates, including compound 3k, also shown in Fig. 7C. (Inset) Chemical structure of compound 3k.

FIG 7

Analysis of MERS-CoV PLpro subsites, active site, and ridge region of the thumb domain. (A) Homology model of MERS-CoV PLpro (gray surface, yellow cartoon), displaying the canonical right-hand architecture, with thumb, palm, and zinc finger domains, and with an additional UBL domain at the N terminus. Modeled Ub (pink) is positioned onto the Ub-binding domain in the zinc finger with its C terminus extending into the active site. The areas highlighted with boxes are the regions of the thumb domain and palm domain predicted to be responsible for MERS-CoV PLpro divergence from SARS-CoV PLpro substrate and inhibitor specificities. (B) Enzyme subsites, displaying the predicted intermolecular interactions with the Ub C terminus. Green dashed lines indicate the H bonds between SARS-CoV PLpro (blue cartoon) and the Ub C terminus that are predicted to be conserved in MERS-CoV PLpro. The black dashed lines indicate H bonds or salt bridges that are predicted to be lost in the MERS-CoV PLpro–Ub C terminus interaction. Amino acids involved in SARS-CoV PLpro–Ub C terminus interactions are shown in blue font, and the predicted corresponding amino acids in MERS-CoV PLpro are shown in black font. Residues highlighted in bold are the nonconserved amino acid substitutions in MERS-CoV PLpro. (C) SARS-CoV PLpro in complex with compound 3k (orange ball-and-stick model) (PDB entry 4OW0), overlaid on MERS-CoV PLpro and a homology model of HCoV-NL63 PLP2 (green). The amino acid residues important for SARS-CoV PLpro–inhibitor interactions are shown (blue font), along with the predicted corresponding amino acids in HCoV-NL63 PLP2 (green font) and MERS-CoV PLpro (black font). The residues highlighted in bold are the nonconserved substitutions in MERS-CoV PLpro. At the bottom of panel C is a comparison between SARS-CoV, HCoV-NL63, and MERS-CoV PLpro amino acid compositions of the β-turn/loop (highlighted with an arrow), known to be important for the inhibitor-induced-fit mechanism of association of compound 3k and SARS-CoV PLpro. (D) Comparison of the active site and oxyanion hole, showing the corresponding amino acids in SARS-CoV, HCoV-NL63, and MERS-CoV PLpros. (E) Overlay of SARS-CoV PLpro and MERS-CoV PLpro ridge regions of the thumb domain. Amino acid numbering is defined as follows: for SARS-CoV PLpro, aa 1 corresponds to aa 1540 in the polyprotein; for HCoV-NL63 PLP2, aa 1 corresponds to aa 1578 in the polyprotein; and for MERS-CoV PLpro, amino acid 1 corresponds to amino acid 1480 in the polyprotein.

FIG 8

Comparison between MERS-CoV PLpro β-turn region and enzyme subsites identified via molecular modeling and the recently reported X-ray crystal structure. A structural superposition between the refined homology model of MERS-CoV PLpro (yellow cartoon) and the recently reported X-ray crystal structure (PDB entry 4P16; green cartoon), which was reported during review of the manuscript, yields a C-α root mean square deviation (RMSD) value of 2.1 Å for 268 atoms aligned. The 2F_o − F_c electron density map from PDB entry 4P16 is contoured at 1σ (shown as gray mesh) and confirms the presence and locations of the amino acids predicted for the enzyme subsites by the structural model (labeled amino acids; shown as sticks). The residues comprising the β-turn in the 4P16 structure are missing in the X-ray structure due to the lack of associated electron density. The refined homology model contains this loop region and therefore serves as a useful structural model for understanding the interactions between the loop and substrates or inhibitors. The striking similarity between the X-ray crystal structure and our energy-minimized structural model demonstrates the high quality of our computational analyses and makes it a good model for predicting a potential conformation for the β-turn of MERS-CoV PLpro.

FIG 9

Multiple-sequence alignment, generated with ESpript, presenting the secondary structure elements on top, as follows: squiggles, α-helices; black arrows, β-strands; and TT, turns. Highlighted are the highly conserved areas (blue outlined boxes) containing the conserved residues (red filled boxes), homologous residues (red font), and divergent residues (black font). The structural elements were generated using the X-ray crystal structure of SARS-CoV apo-PLpro (PDB entry 2FE8). MERS-CoV PLpro UBL truncation sites N20, N40, and N60 are marked in purple, and the catalytic triad residues are highlighted with asterisks. α-Helix 2 (highlighted with a green box), containing the amino acid residues important for SARS-CoV PLpro interaction with K48-Ub₂ and ISG15, is highly divergent among CoV PLpros. The amino acid residues important for interactions with SARS-CoV PLpro inhibitors are highlighted with a blue filled box. The β-turn/loop at the inhibitor-binding site (highlighted with a black outlined box) is highly divergent among CoV PLpros. Sequence accession numbers are as follows: SARS-CoV PLpro_21541–1854, AAP13442.1; HCoV-NL63 PLP_21578–1876, YP_003766.2; MERS-CoV PLpro_1484–1802, AFS88944.1; HCoV-HKU1 PLP2_1648–1955, YP_173236; HCoV-OC43 PLP2_1562–1870, CAA49377.1; HCoV-229E PLP2_1599–1905, CAA49377.1; PHEV-CoV PLP2_1561–1871, YP_459949.1; PRCV-CoV PLP2_1484–1780, DQ811787; TGEV-CoV PLP2_1487–1783, CAA83979.1; FCoV PLP2_1441–1920, AAY32595; CCoV PLP2_1441–1920, AFX81090; BCoV PLP2_21562–1870, NP_150073; and MHV-A59 PLP2_1606–1915, NP_068668.2.

FIG 10

Model for the processing of K48-linked Ub₍₄₎ by SARS-CoV PLpro and MERS-CoV PLpro. Schematic diagrams show two mechanisms for the recognition of the distal Ub (B and C) from K48-linked Ub₍₄₎ (A). (B) The distal Ub-interacting subsites SUb1 and SUb2 are shown for a bivalent mode of recognition, with one Ub subsite at the zinc finger and the second Ub subsite at the ridge region of the thumb domain. (C) The monovalent mechanism of distal Ub recognition uses only the SUb1 site at the zinc finger. The position of the substrate's scissile bond in the active site is indicated with a red arrow, and the reaction progress is shown as product accumulations 1, 2, and 3. (D) SARS-CoV PLpro has a bivalent mode of recognition toward K48-linked polyubiquitin chains (mechanism 1) and has a high affinity for K48-linked di-Ub molecules. In the case of K48-linked Ub₍₄₎, the first cleavage event occurs through the bivalent interaction of the SARS-CoV PLpro zinc finger and ridge region of the thumb domain with di-Ub, producing two di-Ub cleavage products. Subsequent cleavage events occur much more slowly due to the less favorable binding of mono-Ub than di-Ub molecules. (E) MERS-CoV PLpro interacts with K48-linked polyubiquitin chains via a monovalent mode of recognition (mechanism 2) and has a moderate affinity for mono-Ub molecules. Cleavage of K48-linked Ub₍₄₎ occurs through the monovalent interaction of the MERS-CoV PLpro zinc finger with mono-Ub, with no significant differences in the rates of processing tetra-, tri-, and di-Ub species. Other possible cleavage routes are shown with blue arrows.