A peptide sequence found in human TRMT1 fits the cleavage consensus for the SARS-CoV-2 Main Protease (Mpro).

A) Overview of the structure of the SARS-CoV-2 Mpro homodimer (PDB 7BB2) with substrate peptide residues (P4-P3-P2-P1-P1′-P2′-P3′-P4′) illustrated in the Mpro active site (inset); proteolytic cleavage takes place between substrate residues P1 and P1′ (dotted line). B) The TRMT1(527-534) sequence found in a linker region between the TRMT1 SAM methyltransferase (MTase) and Zinc Finger (ZF) domains is consistent with the SARS-CoV-2 Mpro cleavage consensus sequence.

SARS-CoV-2 Mpro cleaves full length human TRMT1 which impacts methyltransferase activity and tRNA binding.

A) Western blots of recombinantly purified full-length TRMT1 incubated with 10 µM catalytically inactive (Cys145Ala) or active (Wild-type) SARS-CoV-2 Mpro at 37 °C. Incubation with WT Mpro results in proteolysis of FL TRMT1 and the appearance of cleavage products corresponding the ZF domain (observed with both anti-TRMT1(609-659) and anti-TRMT1(460-659) antibodies) and the MTase domain (observed with only anti-TRMT1(460-659) antibody). B) Western blots of endogenous human TRMT1 in HEK293T cell lysate incubated with 10 µM of either catalytically inactive (Cys145Ala) or active (Wild-type) Mpro at 37 °C. Endogenous FL TRMT1 is stable in human cell lysate over the course of a 2-hour incubation with C145A Mpro (left) and is rapidly proteolyzed upon incubation with WT Mpro (right). GAPDH was stained in conjunction with TRMT1 antibodies and used as a loading control. C) Recombinant, FL TRMT1 cleaved for 18 hours with WT Mpro (complete cleavage confirmed by Western blot using anti-TRMT1(460-659), top panel), has no observable methyltransferase activity on a human full-length tRNAphe substrate (bottom panel, pink squares); in contrast, robust tRNA methylation activity is still seen with TRMT1 incubated for 18 hours with Mpro Cys145Ala or no protease (bottom panel, gray circles and blue triangles, respectively). TRMT1 tRNA methyltransferase activity was measured by monitoring radiolabel incorporation into tRNA substrate in reactions with cofactor 14C-SAM. D) Recombinant, FL TRMT1 cleaved for 18 hours with WT Mpro (complete cleavage confirmed by Western blot using anti-TRMT1(460-659), top panel), has reduced binding affinity (∼6-fold change) for human tRNAPhe (bottom panel, pink squares) compared to uncleaved TRMT1 that had been incubated for 18 hours with either Mpro Cys145Ala or no protease (bottom panel, gray circles and blue triangles, respectively). TRMT1-tRNA binding was measured by electrophoretic mobility shift assay (EMSA), where bound and unbound tRNA species at different TRMT1 concentrations were separated, visualized by SYBR Gold staining, and quantified using ImageJ to obtain fraction bound values. Methyltransferase and binding assays in C and D were carried out in triplicate and errors are shown as SEM; fitted kinetic and binding parameters are shown in Table S2.

Structure of human TRMT1(526-536) peptide bound to SARS-CoV-2 Mpro. Structure of human TRMT1(526-536) peptide bound to SARS-CoV-2 Mpro.

A) TRMT1 peptide (blue) bound in Mpro active site (gray) showing substrate binding pockets S1, S2, S4, and S3′. Fo-Fc omit electron density map of TRMT1 peptide bound to Mpro is shown contoured at 1σ. TRMT1 Gln P1, an ultra- conserved residue in the Mpro cleavage consensus which is critical for Mpro-mediated proteolysis, is nestled in the S1 pocket of the Mpro active site. The scissile P1 – P1′ amide linkage of TRMT1 is colored orange. B) The TRMT1 P1 Gln amide is positioned for cleavage near Mpro catalytic dyad residues His41 and Cys145Ala in the protease active site. C & D) Direct hydrogen bond contacts formed between Mpro residues (white) and the bound TRMT1 peptide (light blue) are illustrated as yellow dashed lines; C and D are views rotated by 90 °, highlighting different TRMT1-Mpro hydrogen bonding interactions. Mpro Phe140 and His163 recognize the TRMT1 P1 Gln530 sidechain; TRMT1 Asn532 and Mpro Asn142 engage in sidechain-sidechain interaction; additional backbone hydrogen bond contacts include Mpro Thr24-TRMT1 Thr534, Mpro Thr26-TRMT1 Asn532, Mpro Asn142- TRMT1 Asn532, Mpro Glu166-TRMT1 Arg528, and Mpro Gln189-TRMT1 Leu529; many of these interactions are consistent with canonical Mpro-peptide substrate contacts in the active site.

Analysis of Mpro-peptide structures illustrate two distinct substrate binding modes, P3′-in and P3′-out.

A) Comparison of known Mpro substrate cleavage sequences and the P2′ Ψ backbone dihedral angles measured in the corresponding C145A Mpro-peptide structures for each substrate. We included all known C145A Mpro-viral peptide structures in this analysis, except those that were missing the P3′ residue or had poorly-defined electron density for the C-terminal portion of the peptide; structures used in this analysis are PDB IDs: 7MGS, 7T8M, 7DVW, 7T9Y, 7TA4, 7TA7, 7TC4, and 9DW6. Additionally, since a C145A Mpro-nsp6/7 structure was not available, we used an H41A Mpro-nsp6/7 structure (PDB 7VDX) for this analysis. B) Section of an Mpro-bound peptide substrate showing residues P1′, P2′, and P3′, with the key P2′ Ψ dihedral angle illustrated with a curved arrow; the four backbone atoms that define the P2′ Ψ dihedral angle are labeled and highlighted with blue circles (P2′N–P2′Cα–P2′C–P3′N). C) Alignment of peptide substrate backbones in the Mpro active site reveals two distinct binding modes at the C-terminal end of the bound peptides characterized by P2′ Ψ dihedral angles ≥ 157° (nsp4/5, nsp5/6, nsp8/9, nsp9/10, nsp10/11, nsp15/16) or ≤ 116° (TRMT1, nsp6/7). Peptide overlays were generated by aligning SARS-CoV-2 Mpro-peptide substrate structures in PyMOL. The location of the P2′ Ψ dihedral angle in the substrate peptide backbone is denoted with a star. D) Alignment of nsp4/5- and TRMT1- bound Mpro structures showing divergent C-terminal peptide substrate binding modes in the Mpro active site. The backbone geometry of nsp4/5 (P2′ Ψ = 168°) positions the P3′ Phe sidechain away from the Mpro surface (‘P3′-out’ conformation), while the TRMT1 backbone geometry (P2′ Ψ = 115°) positions the P3′ Phe sidechain toward the Mpro active (‘P3′-in’ conformation) site where it displaces Mpro Met49 to open and occupy the S3′ pocket.

Human TRMT1 peptides are cleaved with similar catalytic efficiencies to known Mpro substrates.

A) Kinetics of nsp4/5, nsp8/9, and TRMT1 peptide cleavage by Mpro. To initiate the reaction, 50nM enzyme was added to 100-0.097 µM peptide. Each fluorogenic peptide was conjugated with a quenching moiety, and upon peptide cleavage, the fluorescence of the cleavage product was measured to determine initial rates of the reaction. Nsp4/5 cleavage rates were faster than those observed for the nsp8/9 or TRMT1 peptides, but nsp8/9 and TRMT1 sequences exhibit similar Mpro-mediated cleavage rates. B) The catalytic efficiency (kcat/KM) of TRMT1 peptide cleavage by Mpro is similar to that for nsp8/9 peptide cleavage; both of these substrates are cleaved significantly slower than the nsp4/5 sequence. This suggests that TRMT1 is a feasible substrate for Mpro. C) Illustration of changes in Mpro Met49, Asn142, and Gln189 residue positioning in TRMT1-bound (white) versus nsp4/5-bound (orange) structures. The TRMT1 peptide is shown in blue; nsp4/5 peptide is not shown. D) No major changes in catalytic efficiency are observed for nsp4/5 and TRMT1 peptide cleavage upon mutagenesis of key Mpro residues involved in TRMT1 binding and recognition. Primary fluorogenic kinetic data used to construct the plots and determine the kinetic parameters shown in A – D are listed in Dataset S1; numerical kcat, KM, and kcat/KM values are listed in Table S3.

The Mpro-targeted cleavage sequence is conserved in most mammalian TRMT1 proteins, except rodents where TRMT1 is resistant to cleavage.

A – C) The Mpro-targeted TRMT1 cleavage site sequence (human TRMT1 residues 526-536) is highly conserved in primates (A) and most mammals (B), with the notable exception of rodents (C), where the glutamine Q530 residue most critical for Mpro-directed cleavage is substituted to a lysine in Muroidea. Sequence logo plots of the cleavage site in TRMT1(526-536) were produced with WebLogo3. The human reference sequence is in black and orange residues show the differences. D) WT human TRMT1 is cleaved over the course of a 2 hour incubation with Mpro (left Western blot panel), whereas human TRMT1(Q530K), which has the Q to K mutation found in Muroidea, is entirely resistant to cleavage during a 2 hour incubation with Mpro (right Western blot panel).

Molecular dynamics (MD) simulations confirm dominant peptide binding conformations and suggest discrimination in cleavage kinetics result from catalytic steps that follow initial binding and nucleophilic attack.

A) Distribution of the sum of the minimum distance for P3′ Phe residue in nsp4/5 or TRMT1 from three residues (Thr25, Met49, Cys44) which form the S3′ subsite; P3′-in and P3′-out conformations are illustrated above the distribution plot. The much larger proportion of TRMT1 at smaller distances reflects the peptide’s preference for binding in the P3′-in conformation where TRMT1 P3′ Phe occupies the S3′ pocket during the majority of the MD simulation. B) Distribution of the attack angle of the nucleophilic Mpro Cys145 sulfur atom and the substrate carbonyl carbon atom in the to-be-cleaved amide bond (S–C=O angle θ, top illustration) during the course of the MD simulation. Although nsp4/5 has a higher proportion of attack angles observed closer to the optimal 90 degrees compared to TRMT1, consistent with faster nsp4/5 cleavage kinetics, this small preference is insufficient to explain the 200-fold faster cleavage kinetics of nsp4/5 observed in experimental proteolysis assays.