key: cord-0840583-l3xhx8r7
authors: Kasprzyk, Renata; Spiewla, Tomasz J.; Smietanski, Miroslaw; Golojuch, Sebastian; Vangeel, Laura; De Jonghe, Steven; Jochmans, Dirk; Neyts, Johan; Kowalska, Joanna; Jemielity, Jacek
title: Identification and evaluation of potential SARS-CoV-2 antiviral agents targeting mRNA cap guanine N7-Methyltransferase
date: 2021-07-23
journal: Antiviral Res
DOI: 10.1016/j.antiviral.2021.105142
sha: 6787f0711ba4847fb97db4421205b63016660a20
doc_id: 840583
cord_uid: l3xhx8r7

SARS-CoV-2, the cause of the currently ongoing COVID-19 pandemic, encodes its own mRNA capping machinery. Insights into this capping system may provide new ideas for therapeutic interventions and drug discovery. In this work, we employ a previously developed Py-FLINT screening approach to study the inhibitory effects of compounds against the cap guanine N7-methyltransferase enzyme, which is involved in SARS-CoV-2 mRNA capping. We screened five commercially available libraries (7039 compounds in total) to identify 83 inhibitors with IC(50) < 50 μM, which were further validated using RP HPLC and dot blot assays. Novel fluorescence anisotropy binding assays were developed to examine the targeted binding site. The inhibitor structures were analyzed for structure-activity relationships in order to define common structural patterns. Finally, the most potent inhibitors were tested for antiviral activity on SARS-CoV-2 in a cell based assay.

In 2019, a novel β-coronavirus (β-CoV) emerged, causing coronavirus disease 19 1 and leading to more than one hundred million infections around the world as a serious threat to global health. Due to its high similarity to the SARS and MERS coronaviruses, it was named severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). While multiple vaccines (i.e. Comirnaty® by BioNTech and Pfizer, mRNA-1273 by Moderna, as well as ChAdOx1 nCoV-19 by AstraZeneca) 2 have been developed to prevent the spread of COVID-19, effective medical treatment is currently lacking. The U.S. Food and Drug Administration (FDA) recently approved Veklury (remdesivir) for the treatment of hospitalized COVID-19 patients (ClinicalTrials.gov: NCT04323761). Other repurposed drugs, such as favipiravir, lopinavir, and ritonavir 3 , are still subject to clinical trials. Several monoclonal antibody treatments (e.g., tocilizumab, ClinicalTrials.gov: NCT02735707) 4 have also been approved for treatment of COVID-19 patients with mild or moderate symptoms at risk of progressing to severe disease and hospitalization.

Nevertheless, the ongoing pandemic necessitates the development of novel treatment for COVID-19 and possible future CoV infections. Compounds that target processes essential for virus multiplication and specifically inhibit crucial viral proteins may have potential as antiviral therapeutics. An example of a druggable target process is the 5' end capping of J o u r n a l P r e -p r o o f newly synthesized viral mRNAs, crucial for transcript stability and protein expression in infected human cells.

In eukaryotes, mRNA is modified at its 5' end via the addition of a 7-methylguanosine (m 7 G) cap, which protects transcripts from premature degradation, augments translation, and allows for discrimination between endogenous mRNAs and foreign RNA material (e.g. viral). The process of cap biosynthesis engages three enzymes, namely RNA 5'triphosphatase (TPase), guanylyltransferase (GTase), and N7-methyltransferase (N7-MTase), resulting in cap 0 structure formation, typical for yeasts and plants. 5 In the canonical (human) RNA capping pathway, TPase catalyzes RNA 5' triphosphate cleavage to release RNA 5' diphosphate and a free phosphate group. Thereafter, GTase transfers a GMP molecule from GTP onto the 5' end of RNA. Finally, N7-MTase specifically methylates the guanine N7 position, utilizing S-adenosyl-L-methionine (SAM) as a methyl group donor. In higher eukaryotes, the mRNA cap is subjected to additional methylation at the 2'-O position of the first transcribed nucleotide (cap 1) under the action of 2'-O methyltransferase. Some mRNAs undergo a second 2'-O methylation at the second transcribed nucleotide, forming the so-called cap 2. The functional significance of such 2'-O-methylation is still under investigation, but it presumably protects mRNA from translational shutdown triggered via the innate immune response as a result of type I interferon signaling pathway. Due to the binding specificities of human IFIT proteins, the innate immune system can effectively distinguish and target triphosphate RNAs and cap 0-carrying mRNAs (which come from exogenous sources), but not endogenous cap 1 and cap 2 mRNAs. 6 Viruses often harness their own mRNA capping pathways in order to augment the expression of viral proteins and escape the innate immune response of the host. 7, 8 These may include the canonical capping pathway or distinct non-canonical mechanisms, including m 7 GTP RNA capping way (alphaviruses), 9 GDP RNA capping (non-segmented negative-sense viruses e.g. Rabies virus) 10 or "cap snatching" (influenza virus). 11 Coronaviruses (CoVs) are positive-stranded RNA viruses with some of the most complex and largest viral genomes. Severe acute respiratory syndrome coronavirus (SARS-CoV) expresses 16 non-structural proteins (nsps), which are responsible for essential J o u r n a l P r e -p r o o f processes and are no part of the viral capsid. The SARS-CoV-2 genome shares 85% identity with that of SARS-CoV 12 , also encoding 16 non-structural proteins. In contrast to small RNA viruses, CoVs encode unique proteins to cap their mRNAs, presumably following the canonical or alphavirus-like capping pathway. 13 While the nsp13 enzyme is responsible for TPase and RNA helicase function for both SARS-CoV and SARS-CoV-2, 14, 15 there is still no experimental evidence of proteins with GTase activity engaged in RNA capping. 16 SARS-CoV mRNAs are N7-methylated by nsp14. 17 The nsp14 enzyme has also been reported to possess 3'-5' exonuclease (ExoN) activity, thus acting as an RNA proofreading enzyme. 18 The two catalytic domains of nsp14, ExoN and N7-MTase, are located at its N-and C-termini, respectively, and function independently. Nonetheless, the 62-527 amino acid sequence is required for both activities. The exonuclease activity of nsp14 is significantly enhanced upon binding with nsp10, while N7-MTase activity is not affected. 19, 20 The nascent cap 0 structure can be further methylated by SAM-dependent nsp16, which acts as a 2'-O-methyltransferase, requiring nsp10 as a cofactor for activation. The nsp16/nsp10 complex utilizes m 7 GpppA-RNA to synthesize the cap 1 structure. [21] [22] [23] Differences in capping machinery and mechanisms between coronaviruses and humans create new opportunities for antiviral drug development. Viral N7-methyltransferases have already been highlighted by various groups as a potential targets for antiviral therapies, [24] [25] [26] [27] including ones for SARS-CoV. 28 N7-MTase inhibition has been shown to suppress viral replication. 19, 29 Various small-molecule inhibitors of SARS-CoV nsp14 N7-MTase have already been discovered through different in vitro approaches, including large library screening 30 and rational drug design based on screening results or substrate structures. 31 Currently, the crystal structure of SARS-CoV-2 nsp14 N7-MTase domain remains unknown, impeding rational drug design (although the ExoN-nsp10 complex is already available -PDB code: 7DIY 32 ). However, the N7-MTase active site is highly conserved among Coronaviridae 33 , and the high amino acid identity with SARS-CoV nsp14 (> 95%) enables homology modelling. 34 This approach has been widely used for compound library virtual screening, including docking studies. [35] [36] [37] [38] J o u r n a l P r e -p r o o f Based on the structural comparison of SARS-CoV-2 nsp14 to its SARS-CoV homolog, Otava et al. carried out structure-guided design of substrate-based inhibitors, generating analogs of S-adenosyl-L-homocysteine (SAH) modified with aromatic groups at position 7 of the 7-deazaadenine derivative of SAH. 39 Inhibitors have also been synthesized and tested in vitro via radioactivity-based assays using biotinylated RNA and 3 H-SAM substrates 40 as well as via surface plasmon resonance.

We recently developed a fluorescence-based assay for the high-throughput screening (HTS) of potential N7-methyltransferase inhibitors. The assay is based on a Gp3A analog labeled with pyrene at the 3'-O-position of adenosine that undergoes fluorescence intensity change (quenching) upon methylation (Py-FLINT probe; Figure 1 ). We have applied the assay in studies on model N7-MTases, including Ecm1 from E. cuniculi, human RNMT-RAM, and Vaccinia capping enzyme (VCE) from Vaccinia virus. 41 In the current work, we adapted the assay to explore the N7-MTase activity of nsp14 from SARS-CoV-2 in order to identify potent inhibitors of viral mRNA N7-methylation with antiviral drug potential. The method allows for the direct monitoring of nsp14 activity in real time and is suitable for the exploration of inhibitors of both the nucleotide-and SAM-binding sites of N7-MTase. We employed the method for HTS experiments of five commercially available compound libraries, namely LOPAC® 1280 , Machine Learning SARS Targeted Library (OTAVA Chemicals), SARS-CoV-2 nsp16 Targeted Library (OTAVA Chemicals), FDAapproved Drug Library, and part of the Flavonoids Compound Library (ChemFaces), containing a total of 7039 compounds. The most potent hits selected from the screening were further evaluated to determine their IC50, validate their inhibitory properties using RNA substrates, determine selectivity against the human N7-MTase RNMT-RAM, and performed structure-activity relationship (SAR) analysis. To study the mechanism of action (MOA), we developed fluorescence anisotropy (FA) binding assays using either fluorescently labeled Gp3A or SAH analogs. Finally, the antiviral activity of identified nsp14 inhibitors was studied using human β-coronavirus SARS-CoV-2-infected Huh 7 (human hepatocellular carcinoma) cells. 42 We identified three compounds that inhibited viral replication in the cell culture model. 

Chemical synthesis of Gp3ApG and SAH-N6-6tFluo, nsp14 expression and purification, Py-FLINT assay optimization for nsp14, short RNA synthesis, and dot blot analysis were described in details in Supplementary Information. Py-FLINT and FA binding assays were performed using point fluorescence measurements in 96-well black, non-binding assay plates at 30 °C with microplate reader Biotek Synergy H1. For FA binding assays we used excitation (485 ± 20 nm) and emission (528 ± 20 nm) polarization filters.

For screening experiments each sample (well) contained 50 mM Tris-HCl, pH 7.5, a substrate (1 µM Py-FLINT probe), the SAM cosubstrate (20 µM), nsp14 enzyme (40 nM), an inhibitor (50 µM), and 5% of DMSO (if compound in library was dissolved in pure DMSO). The reaction components were preincubated for 15 min at 30 °C with mixing at 300 rpm. Before adding the enzyme, the plate was additionally incubated in a plate reader for 20 min at 30 °C with mixing at 300 rpm and point fluorescence registered every minute J o u r n a l P r e -p r o o f to obtain stable fluorescence signal. Immediately before fluorescence reading, 10 µL of an enzyme solution was added into each well to a total reaction volume of 150 µL. The reaction was monitored for 0.5−1 h. Initial rates were calculated by fitting a linear curve to the first 10 points (10 minutes).

IC50 determination experiments were performed analogously, but with different inhibitor concentrations (half-log dilutions logCinh <−2.5;2> for nucleotide-like inhibitors or logCinh. <−3;1.5>). If the inhibitor fluorescence interfered with the Py-FLINT probe emission, a control measurement of the inhibitor solution was performed under the same conditions and then the result was subtracted from the inhibition data. To determine IC50 parameters, a four-parameter dose-response equation was fitted as follows:

where A1 and A2 are the bottom and top asymptotes, respectively; Cinh. the inhibitor concentration; p is the Hill coefficient, and V/V0 is the ratio of the initial reaction rate with the inhibitor to that without the inhibitor.

To 

where ∥ is the parallel emission intensity, ⊥ the perpendicular emission intensity, and G is a grating factor equal to 0.994. For each sample the final FA values were calculated as the mean FA values from datapoints between 50 and 60 min. To calculate the total emission intensity we used the following equation:

J o u r n a l P r e -p r o o f

As the I values did not change more than 10% we did not corrected FA values on enhancement factor resulting from emission intensity changes. To determine KD the FA values were plotted as a function of protein concentration and the following binding curve was fitted:

where FAF is the fluorescence anisotropy of free probe, FAB is the fluorescence anisotropy of bound probe, Cp is the nsp14 concentration, and LT is the total probe concentration. 

The human hepatoma cell line Huh 7 (kindly provided by Ralf Bartenschlager, University 

We first assessed whether the nsp14 N7-MTase could methylate small dinucleotide substrate Gp3A. To that end, Gp3A methylation in the presence of nsp14 was monitored via RP HPLC and mass spectrometry, which confirmed the formation of N7-methylated product m 7 Gp3A and SAH as a byproduct ( Figure S1 ). We then determined the steady- (Table S1 ). Based on the determined KM value (3.5 ± 1.1 μM), the optimal probe concentration was set at 1 μM (approximate to KM). The catalytic efficiency (kcat/KM) 44 of nsp14-catalyzed N7-methylation was 0.025 s -1 μM -1 , which was almost two-fold higher than previously calculated for the RNMT-RAM enzyme (0.014 s -1 μM -1 ), 41 indicating that the Py-FLINT probe is a slightly better substrate for SARS-CoV-2 than human N7-MTase.

To determine the optimal concentration of the SAM cosubstrate, we incubated the Py- Figure S2B ).

The optimal SAM concentration was set to 20 μM based on the plateau position. Finally, the optimal conditions of nsp14 N7-MTase activity monitoring were established as follows:

1 μM Py-FLINT probe, 20 μM SAM, and 40 nM nsp14 (Table S2) .

To validate the method for HTS experiments, we determined z-factor value by measuring fluorescence intensity changes of the positive and negative control samples ( Figure S3 ).

The positive control samples (no inhibition) were prepared as a mixture of the Py-FLINT probe, SAM, and nsp14, while the negative control samples additionally contained 40 μM of sinefungin (universal MTase inhibitor). The calculated z-factor value of 0.79 implied that the assay met the requirement for HTS (z factor > 0.5) and could therefore be applied for compound library screening.

We employed the optimized Py-FLINT method for the HTS screening of commercially Figure 2 ). When comparing the number of hits to the total number of compounds included in each library, the largest contribution was observed for the Flavonoid Library (5.6%) and the lowest for the two combined libraries from OTAVA Chemicals (0.35%; Figure 2 ). The inhibitory properties of hit compound were subsequently explored.

All hits were further evaluated to determine their IC50 values. To this end, Py-FLINT probe Table 1 ).

Based on the determined IC50 values for LOPAC® 1280 hits, we divided the compounds into three sets. The first set (LI; Figure 3A , 4, Table 1 ) contained the ten most potent inhibitors (IC50 < 10 Figure 3C , 4, Table 1 ), which The first three compounds were even more potent than the sinefungin reference ( Table 1 ). The second compound set (FII)

included inhibitors with 10 μM < IC50 < 30 μM (10 compounds), while the third one (FIII) those with IC50 > 30 μM (16 compounds, Figure S6 , Table S5 ). Some of the hits, such as Table 1 .

In order to independently confirm whether the hits identified and characterized via Py-FLINT assay could inhibit nsp14 N7-MTase activity, we used RP HPLC to analyze the N7- Figure S8A ). We speculate that these compounds may interact with nucleotide substrates, influencing the IC50 values determined via Py-FLINT assays.

To additionally confirm nsp14 N7-MTase inhibition, we carried out the reaction on longer (35 nt) in vitro transcribed RNA substrates capped with the Gp3ApG trinucleotide, which ensures high RNA capping efficiency. 46 To analyze the reaction progress, we employed dot blot assays with product detection using the m 7 G-cap-specific antibody. To calculate the amount of N7-methylated product, a calibration curve was prepared using mixtures of Gp3ApG-RNA35 and m 7 Gp3ApG-RNA35 at different ratios. Unfortunately, many of the inhibitors, including SCH 202676, pyridostatin, tannic acid, or ebselen, interfered with RNA blotting and detection ( Figure S9 ). Hence, the IC50 values were determined only for selected non-interfering compounds sinefungin and p-benzoquinone ( Figure S10 ). We confirmed specific N7-methylation by nsp14 and observed that p-benzoquinone (IC50 1.93 ± 0.42 μM) was an even more potent inhibitor than sinefungin (21.0 ± 4.6 μM). 

We observed several frequently repeating structural patterns among the identified nsp14

inhibitors. In order to systematically review the structures and identify scaffolds potentially related to compound bioactivity, we used the SARvision|SM (Small Molecules) software (by Altoris, Inc., San Diego, CA). We selected nsp14 inhibitors with IC50 < 50 μM for the analysis, obtaining a total of 83 different structures. The most common pattern was chromone, found in 11 compounds, constituting 13% of the nsp14 inhibitors with an IC50 lower than 50 μM ( Figure 5 ). Usually the Chromone was present in flavonoid analogs, such as myricetin, morin, or multicaulisin. The other identified scaffolds are presented in Figure 5 and Figure S11 . Nsp14 inhibitors found in the OTAVA library were weaker compared to inhibitors from other libraries, however their fragment-based drug character is of interest for SAR analysis. Hence, we separately analyzed scaffolds identified in the structures of OTAVA inhibitors ( Figure S12 ). were marked in blue and red, respectively. Open circle data points indicate that the IC50 value was higher than 100 μM.

All 41 The second probe, designed for the SAM-binding site, was synthesized in four steps, starting from 6-chloropurine riboside ( Figure S14 ; SAH-N6-6tFluo).

We then confirmed that the fluorescence anisotropy (FA) of the probe-nsp14 complexes is higher than that of free probes ( Figure S15 ). Using Gp3A and sinefungin, ligands targeting nucleotide-and SAM-binding sites, respectively, we validated the assay's functionality (Figure 7, columns 1-3) . We used bovine serum albumin (BSA), a noninteracting protein, as a negative control. Both probes were also characterized for their binding affinity in FA-monitored saturation-binding experiments ( Figure S15B ) Table S7 . Extended data including control samples with BSA are shown in Figure S16 . 

The value similar to its EC50 (35.6 ± 1.0 μM), which is visible in its low SI value of 1.15 ±0.04.

Herein, we sought to apply our previously developed N7-MTase Py-FLINT assay 41 for the identification of inhibitors against the nsp14 N7-methyltansferase, which is involved in mRNA capping of SARS-CoV-2. We first confirmed that our assay could be used for monitoring nsp14 activity in real time. Thereafter, we optimized experimental conditions to 1.2 times lower than that of sinefungin, a well-known methyltransferase inhibitor. 48 We previously identified some of the inhibitors from the LOPAC® 1280 library, including myricetin and Reactive Blue 2, as compounds targeting Ecm1 N7-MTase from the E.

cuniculi parasite. 41 Overall, the hit compounds from the two OTAVA libraries were weaker nsp14 inhibitors. The two most potent ones with IC50 ~20 μM were selected for further studies together with inhibitors from the other libraries.

The hits found via Py-FLINT assays were validated through HPLC with Gp3A as a substrate. All of the identified nsp14 compounds were confirmed as inhibitors of nsp14 activity. However, the RP HPLC analysis revealed pyridostatin (LI3), mitoxantone (LI5) and ruthenium red (LI6) as weaker inhibitors of nsp14 than indicated by Py-FLINT assay results. We speculate that ruthenium red, a complex of ruthenium ion and ammonia molecules, pyridostatin, and mitoxantrone, which has positively charged amine groups located at alkyne chains, electrostatically interacted with the Py-FLINT probe, influencing fluorescence intensity. Ruthenium red targets various proteins, which makes it nonselective 49 and was thus excluded from further studies.

For further hit validation, we used 35 nt RNA capped with Gp3ApG trinucleotide as a substrate. m 7 Gp3ApG-RNA35, the product of N7-methylation by nsp14, was detected using an m 7 G-cap-specific antibody via dot blot assay. Unfortunately, most inhibitors showed interferences with the antibody, in which case the reaction product could not be reliably quantified. Therefore, the IC50 values were determined for two selected compounds, namely sinefungin (reference) and p-benzoquinone, confirming their inhibitory properties against nsp14.

To identify the structural patterns that could enhance the potency of inhibition against nsp14 N7-MTase, we performed SAR analysis of all 83 inhibitors with IC50 < 50 μM using SARvision|SM software by Altoris Inc. As a result, we identified chromone in the structures of 11 nsp14 inhibitors. Chromone is found in various compounds of plant origin, such as alkaloids and flavonoids, which often possess anti-inflammatory and antiviral properties. 50 Analogs of chromone have already been identified as inhibitors of the SARS-CoV-2 3Clike protease (3CLpro). 51 Herein, we demonstrated their ability to target another essential SARS-CoV-2 protein -the nsp14 N7-MTase. The flavonoid baicalein was previously identified as an nsp14 N7-MTase inhibitor via an in silico approach. 36 However, we did not observe such activity in our Py-FLINT assays during Flavonoids compound library

screening. An antraquinone motif was identified within 6 nsp14 inhibitors, including anthracyclines (i.e. doxorubicin), known for their anticancer activity. Three other scaffolds present in at least 5 compounds identified via SAR analysis were 2-aminothiophene-3carbaldehyde, naphthalene, and biphenyl. Although compounds selected from OTAVA libraries were relatively weak nsp14 inhibitors (IC50 > 20 μM), we found them of interest for SAR analysis. Scaffold identification for these compounds revealed frequently repeating heterocycle fragments containing sulfur atoms, such as thiophene, thiazole, or thiadiazole. All of the identified scaffolds could potentially favor SARS-CoV-2 ns14 N7-MTase inhibition.

As a next step of inhibitor evaluation, we tested the selectivity of identified compounds towards human RNA N7-MTase (RNMT) in complex with RNMT-activating miniprotein (RAM) using Py-FLINT assay. 41 and Evans Blue (dyes) exhibited EC50 values lower than 50 μM. The most selective compound with an EC50 of 3.58 ± 0.16 μM and an SI of 16.6 ± 6.1, was pyridostatin. This potency is in the same range as the EC50 of GS-441524 (the parent nucleoside of remdesivir) in this assay system. In binding assays, we observed that FA were higher for probe/nsp14/pyridostatin samples compared to probe/BSA/pyridostatin or probe/pyridostatin. This result suggested that pyridostatin may interact with both nucleotide probes and probe-nsp14 complexes. Hence, its antiviral mechanism may be more complex. Further studies are required to elucidate the mechanism of action.

Although most of the compounds did not exhibit potent antiviral activity, which may be due insufficient cellular permeation, the current results provide a basis for the design of novel biologically active compounds targeting SARS-CoV-2 capping machinery. We hope that the current results will contribute to future rational drug design for COVID-19 treatment.

Supplementary Information. Tables S1-S7, Figures S1-S17, Experimental procedures, and HRMS and NMR spectra for the synthesized compounds.

Corresponding Author 

Alternative Medicine Interventions for COVID-19

We thank Professor Victoria Cowling (University of Dundee) for the human N7-MTase RNMT-RAM, Professor Marcin Drag (Wroclaw University of Science and Technology) for sharing FDA-approved Drug Library, and Dr Karolina Drazkowska (University of Warsaw) for designing the cloning.

J o u r n a l P r e -p r o o f

☒ The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.☐The authors declare the following financial interests/personal relationships which may be considered as potential competing interests:Jacek Jemielity 25 th June 2021 J o u r n a l P r e -p r o o f