Abstract
Monkeypox is an emerging, rapidly spreading disease with pandemic potential. It is caused by the monkeypox virus (MPXV), a dsDNA virus from the Poxviridae family, that replicates in the cytoplasm and must encode for its own RNA processing machinery including the capping machinery. Here, we present the crystal structure of its 2’-O-RNA methyltransferase (MTase) VP39 in complex with the pan-MTase inhibitor sinefungin. A comparison of this 2’-O RNA MTase with enzymes from unrelated ssRNA viruses (SARS-CoV-2 and Zika) reveals a surprisingly conserved sinefungin binding mode implicating that a single inhibitor could be used against unrelated viral families.
Introduction
Monkeypox is a disease caused by the monkeypox virus (MPXV), an Orthopox virus belonging to the Poxviridae family, clinically resembling the eradicated smallpox1. Until recently, its occurrence was limited to central and western Africa where its natural reservoir consists of rodents and primates2; however, it can also infect humans, and the mortality rate is estimated to be 3 - 6%. This rate is lower than the deadly smallpox, but much higher than COVID-19 that is caused by SARS-CoV-2. Monkeypox has become increasingly spread across the globe with number of cases exponentially growing reaching somewhat alarming 54900 cases, and it has been reported in some 100 countries3. It is no wonder that the prospective of another viral pandemic alarmed the public health authorities as well as the general public.
MPXV is a dsDNA virus; however, it replicates in cytoplasm, implying that it must encode for DNA and RNA replication machinery4 because the human enzymes are located in the nucleus. Besides the DNA-dependent DNA polymerase and the RNA-dependent RNA polymerase, it also encodes for the RNA capping machinery. In an early stage of cellular infection, the RNA cap is important for the initiation of translation of the viral RNA (vRNA)5,6. This cap plays an important role in the innate immunity as uncapped RNA is recognized by the innate immunity IFIT and RIG-I sensors 7,8. The cap is also important for stability of the mRNA, thus the poxviruses encode for decapping enzymes to prevent accumulation of dsRNA later during infection which would induce the innate antiviral response9. Indeed, protection against innate immunity is of the uttermost importance for the MPXV, and it also encodes poxin, an enzyme that blocks the cGAS-STING pathway triggered by the presence of dsDNA in cytoplasm10.
Chemically, the mRNA cap is a structure where an N7-methylated guanosine is linked via a triphosphate to the 5’ end of the RNA yielding m7GpppRNA, which is also referred to as cap-0. In the fully matured cap (cap-1), the first nucleotide is also methylated at the 2’-O position of the ribose moiety. In the Poxviridae family, cap-0 is synthesized in cytoplasm by the heterodimeric capping enzyme that has RNA 5’ triphosphatase (RTPase), guanylyltransferase (GTase) and (guanine-N7)-methyltransferase (MTase) activity11. The addition of another methyl group to the 2’-O position of the adjacent ribose converts cap-0 to cap-1. This step is also important in preventing an innate immunity response12 and is catalyzed by the MPXV 2’-O MTase VP3913. Here, we present the structure of the MPXV VP39 in complex with the pan-MTase inhibitor sinefungin. The structure reveals the mechanism of VP39 inhibition and comparison of this structure to the 2’-O MTases from unrelated viruses SARS-CoV-2 and Zika has important implications in the design of pan-antivirals based on MTase inhibitors.
Results
Overall structure and enzymatic characterization of the MPXV VP39 MTase
To obtain the structure, we expressed and purified recombinant MPXV VP39 in E. coli. and performed screening of the crystallization conditions. Initial crystals were of poor quality but diffraction quality crystals were obtained upon a few rounds of seeding. The crystals diffracted to 2Å resolution and belonged to the orthorhombic P212121 space group. The structure was solved by molecular replacement and refined to Rwork = 20.81% and Rfree = 23.24% and good geometry (detailed in Table 1). The structure revealed a mixed α/β fold that resembles the Rossmann fold. A central β-sheet is composed of β-sheets β2-β10 that adopt the shape of the letter J. Interestingly, this arrangement was also observed for the SARS-CoV-2 2’-O MTase nsp1614. The central sheet is stabilized from one side by helices α1, α2, α6 and α7, and by α3 and α7 from the other. These opposite sides of the J-like β-sheet are connected by α 5, β1 and β11 (Figure 1A, B).
We tested two substrates that differed in the penultimate base (m7GpppA-RNA vs. m7GpppG-RNA) to verify the enzymatic activity of our recombinant VP39 enzyme. Both substrates were accepted; however, the guanine base was clearly preferred in the penultimate position (Figure 1C). That is in contrast to coronaviruses where adenine is preferred at this position15. We also verified that sinefungin was able to inhibit the VP39 MTase; under the conditions used (detailed in M&M), it exhibited IC50 = 40.7 ± 6.2 μM (Figure 1D).
Analysis of the VP39-sinefungin interaction
The electron density for sinefungin was immediately visible upon molecular replacement in the central pocket of the VP39 MTase (Figure 1). Sinefungin occupied the SAM binding pocket (Figure 2A). Its adenine moiety lies within a deep canyon lined with hydrophobic sidechains from residues Phe115, Val116, Val139 and Leu159 and makes two hydrogen bonds with the backbone of Val116 (Figure 2B). The ribose and amino acid moieties are also involved in hydrogen bonding: the ribose ring directly binds the Asp95 and Arg97 via its hydroxyl groups while the amino acid moiety forms hydrogen bonds with the sidechains of Gln39 and Asp138 (Figure 2B).
Based on the previously solved crystal structure of vaccinia VP39 in complex with RNA16, we constructucted a model of a ternary complex sinefungin:RNA:VP39 to illustrate the molecular mechanism of sinefungin’s action (Figure 3). It efficiently protects the 2’-O-ribose position; its amino group is in the close vicinity of the 2’ ribose hydroxyl group where the sulphur atom of SAM would be located otherwise.
Implications for inhibitor design
The SAM binding canyon has two ends: one borders with the RNA binding pocket and serves to position SAM for the methyltransferase reaction; the other end adjacent to the adenine base of sinefungin is apparently unoccupied (Figure 3). However, a close inspection of this location reveals a network of water molecules that are interconnected by hydrogen bonds and also bond the adenine moiety and residues Val116, Glu118 and Asn156 (Figure 4). Compounds based on the sinefungin scaffold bearing a moiety that would displace these waters and interact with Val116, Glu118 and Asn156 directly could be an exceptionally good binder because displacement of water molecules can have a favorable entropic effect17,18.
Next, we compared the catalytic sites of VP39 with the catalytic sites of 2’-O-ribose MTases from unrelated medically important viruses, namely SARS-CoV-2 and Zika. The resemblance of the SAM binding sites is remarkable in both cases. Sinefungin is virtually in the same conformation in MPXV VP39, SARS-CoV-2 nsp16 and Zika NS5 (Figure 5). The catalytic tetrad (Lys41, Asp138, Lys175, Glu218 for MPXV) is absolutely conserved among these unrelated viruses including the conformation of these residues. However, differences do exist in the binding mode of the nucleobase as well as the sugar ring. The adenine base is bound to the backbone of Val116 (Figure 2) and held in place by a complex water network in the case of VP39 (Figure 4), while in SARS-CoV-2 it is held by the side chain of Asp114 (Figure 5A) and in Zika by the side chain of Asp131 (Figure 5B). The binding mode of the ribose ring is virtually identical between VP39 and SARS-CoV-2 nsp16, as both of these viruses utilize an aspartate residue (Figure 5A). However, in Zika it is held in place by His110 and Glu111. Interestingly, all these viruses use an aspartate residue to interact with the amino group of sinefungin’s amino acid moiety, further highlighting the similarity of the SAM binding site between a dsDNA virus and two distinct +RNA viruses.
Discussion
Viruses are a global threat to human health as illustrated by the recent COVID-19 pandemic. Even viral diseases with mortality rates ~ 1% can kill millions of people if they spread globally. Monkeypox is already present on every continent, its spread was very likely facilitated by cessation of the vaccination against smallpox in the 1980s since the smallpox vaccine is effective against the monkeypox (~80% efficiency)19. With an estimated mortality rate of 3 - 6%, the MPXV has the potential to cause a devastating pandemic20. On the other hand, the effective FDA-approved vaccine JYNNEOS21 is available, as is the FDA-approved drug - tecovirimat22. Encouragingly, MPXV is a dsDNA virus, and the mutation rate of DNA viruses is much lower than that of RNA viruses, such as SARS-CoV-2 or Zika. Nonetheless, it would be wise to develop more drugs to ensure their availability and efficiency.
RNA MTases are present in multiple unrelated viral families, including both RNA and DNA viruses23; yet, none of the FDA approved drugs target a viral MTase. Most current antivirals are inhibitors of polymerases, integrases or proteases. In fact, only a handful of inhibitors of viral MTases, mostly targeting flaviviruses24,25, were known prior to the COVID-19 pandemic. However, recently many inhibitors of SARS-CoV-2 MTases were reported by us and others26–30.
It remains to be established if inhibitors of MPXV MTases are viable antiviral targets; however, the high conservation of VP39 across orthopoxviruses suggests an important role in the viral lifecycle. In addition, the similarity of the dsDNA virus VP39 SAM binding site with the 2’-O MTases from two +RNA viral families (flaviviruses and coronaviruses, Figure 5) suggests that the development of a panantiviral MTase inhibitor should be feasible, especially since the catalytic tetrad is absolutely conserved. However, to obtain specific antiviral inhibitors that will not interfere with the human RNA 2’-O MTases, it might be necessary to take advantage of specific features of the viral enzymes, such as a large hydrophobic pocket above the adenine base that is present in coronavirus nsp1628,31 or the complex water network adjacent to the adenine base in the case of MPXV VP39. Nonetheless, inhibitors of viral MTases are urgently needed to further elucidate the role of these enzymes during viral infection.
Materials and Methods
Protein expression and purification
The gene for the monkeypox (USA-May22 strain) viral MTase VP39 was codon-optimized for expression in Escherichia coli and commercially synthesized (Azenta). Subsequently it was cloned into a pSUMO (with 8xHis-SUMO tag) using restriction cloning (BamHI and NotI sites). The recombinant protein was purified according to our established protocols for viral methyltransferases32. Briefly, E. coli BL21(DE3) cells were transformed with the VP39 expression plasmid and incubated in LB medium supplemented with ZnSO4 (10 μM) and ampicillin (100 μg/ml) and cultured at 37°C. When the cells reached optical density [OD] = 0.6 the production of VP39 was induced by the addition of IPTG (isopropyl-b-D-thiogalac-topyranoside) (250 μM) and the temperature was lowered to 20°C for 15 h. The cells were pelleted by centrifugation and resuspended in lysis buffer (25 mM Tris, pH 8, 300 mM NaCl, 30 mM imidazole pH = 8.0, 3 mM b-mercaptoethanol) and lysed by sonication. The clarified lysate was subjected to affinity chromatography on Ni2+ agarose resin (Macherey-Nagel) and washed extensively with lysis buffer. The recombinant protein was eluted with lysis buffer supplemented with 300 mM imidazole pH = 8.0. The 8xHis-SUMO tag was cleaved by Ulp1 protease during dialysis against the lysis buffer. Subsequently the Ulp1 and His8x-SUMO tag were removed by a second round of affinity chromatography. Unbound fraction was concentrated and loaded on HiLoad 16/600 Superdex 75 gel filtration column (Cytivia) in size exclusion buffer: 10mM Tris, pH 7.4, 150 mM NaCl, 2mM b-mercaptoethanol and 5% glycerol. The purified VP39 was concentrated to a set of different concentrations from 7 to 45 mg/ml, mixed with sinefungin (1 mM) and directly used for crystallization trials.
Crystallization and data collection
VP39 protein supplemented with 1 mM sinefungin and it was plated using the sitting drops method. 200 nl protein solution and 200 nl reservoir solution were mixed in several commercial crystallization screens using a mosquito robot (SPT Labtech). Only low-quality crystals were generated in the initial experiments, which were then used as crystallization seeds in further experiments. The initial crystals were crushed with a glass rod and diluted 10 000x to produce seeds. The seeding screen was prepared using Dragonfly (SPT Labtech) by gradually decreasing the PEG concentration of the original hit (200 mM lithium citrate, 20% (w/v) PEG 3350). The use of Angstrom Additive Screen TM (Molecular Dimensions) improved the quality of the seed. Finally, thin plate shaped crystals grew within 24 hours at 18°C in 200 mM lithium citrate and 14.5% (w/v) PEG 3350 and were cryoprotected in a 20% (v/v) glycerol-enriched well solution and snap-frozen in liquid nitrogen.
The crystallographic dataset was collected from a single crystal on the BL14.1 beamline at the BESSY II electron storage ring operated by the Helmholtz-Zentrum Berlin33. The dataset was collected at the temperature of 100 K using the wavelength of 0.9184 Å. The crystals diffracted to 2Å resolution and belonged to the P212121 space group. The data was integrated and scaled using XDS34. The structure of the MPXV VP39/sinefungin complex was solved by molecular replacement using the structure of the vaccinia virus VP39/SAH complex as a search model (pdb entry 1VP3)35. The initial model was obtained with Phaser36 from the Phenix package37. The model was further improved using automatic model refinement with Phenix.refine38 from the Phenix package37 and manual model building with Coot39. Statistics for data collection and processing, structure solution and refinement are summarized in Table 1. Final Ramachandran statistics were as follows: favored 99.64%, allowed 0.36%, outliers 0.00%. Structural figures were generated with the PyMOL Molecular Graphics System v2.0 (Schrödinger, LLC). The atomic coordinates and structural factors were deposited in the Protein Data Bank (https://www.rcsb.org) under the PDB accession code 8B07.
RNA substrate preparation
For the 2’-O-MTase, an m7GpppG-capped RNA or an m7GpppA-capped RNA were used. These RNAs were 35 nucleotides long and they were prepared by in vitro transcription using a DNA template (5’-CAGTAATACGACTCACTATAGGGGAAGCGGGCATGCGGCCAGCCATAGCCGATCA-3’) and the TranscriptAid T7 high-yield transcription kit (Thermo Scientific). The reaction was performed in a 50-μl mixture containing 1× TranscriptAid reaction buffer, 7.5 mM nucleoside triphosphates [NTPs], 1 μg template DNA, and 1× TranscriptAid enzyme mix and 6 mM cap analog m7GpppG (Jena Bioscience) or m7GpppA (prepared chemically according to a published protocol40). The reaction was incubated at 37°C for 8 h. Next, the RNA was purified using RNA Clean & Concentrator-5 from Zymo Research. DNA was removed by DNase I treatment performed directly on the column.
2’-O-MTase assay
The reaction was performed in a total volume of 15 μl and all the compounds were diluted in DEPC water to prevent contamination with RNases. The reaction mixture contained 4 μM SAM (S-Adenosyl-L-methionine), 4 μM m7GpppG-capped RNA or m7GpppA-capped RNA in the MTase reaction buffer (5mM Tris pH 8.0, 1mM TCEP, 0.1 mg/ml BSA, 0.005% Triton X-100, 1mM MgCl2). The reaction was initiated by adding 500 nM VP39, mixed, spun, and incubated at 30°C for 120 min. The reactions were stopped by adding 30 μl of 7.5% formic acid and analyzed on an Echo mass spectrometry system coupled with a Sciex 6500 triple-quadrupole mass spectrometer operating with an electrospray ionization source. The rate of MTase activity was measured as the amount of the product of the reaction S-adenosylhomocysteine (SAH). The spectrometer was run in the multiple-reaction-monitoring (MRM) mode with the interface heated to 350°C. The declustering potential was 20 V, the entrance potential 10 V, and the collision energy 28 eV. Ten nanoliters was injected in the mobile phase (flow rate of 0.46ml/min; 70% acetonitrile with 0.1% formic acid). The characteristic product ion of SAH, m/z 385.1 >134.1, was used for quantification.
For measurements of inhibition of the 2’-O-MTase reaction by sinefungin, the above described protocol was used with only m7GpppG-capped RNA as a substrate. Sinefungin was added to the reaction mixture in the concentration range from 0 to 100 μM and all measurements were performed in triplicates, i.e. the measurements were taken from distinct sample reactions that were prepared using the same stocks of the enzyme, inhibitor, and other chemicals. The half maximal inhibitory concentration (IC50) was calculated by non-linear regression and curve fitting using the equation: y(x) = max/(1+x/IC 50), where y is the substrate (SAM) conversion rate, x is the concentration of inhibitor (sinefungin), and max is the maximal substrate (SAM) conversion rate in the absence of inhibitor.
Acknowledgment
We thank the Helmholtz-Zentrum Berlin für Materialien und Energie for the allocation of synchrotron radiation beamtime. This research was funded by the project the National Institute Virology and Bacteriology (Programme EXCELES, Project No. LX22NPO5103) - Funded by the European Union - Next Generation EU. The Academy of Sciences of the Czech Republic (RVO: 61388963) is also acknowledged.