Aptamer Displacement Screen for Flaviviral RNA Methyltransferase Inhibitors

Reagents

ATA, SAM, S-adenosyl-L-homocysteine (SAH), and GTP were purchased from Sigma-Aldrich (St. Louis, MO). FL-dT10 (5′ FAM-TTT-TTT-TTT-T-OH 3′) was synthesized by IDT, Inc. (Coralville, IA).

MTase Purification

The DNA sequence that specifies the first 296 amino acids of YFV NS5 (17D) was cloned into pet28B with an N-terminal poly His tag, and protein was expressed in the BL21 Rosetta strain from Invitrogen (Madison WI). Cultures were induced with 0.5 mM isopropyl β-D-1-thiogalactopyranoside for 6 h at room temperature (RT), pelleted, and frozen. Frozen pellets were resuspended in lysis buffer and sonicated. Following centrifugation to remove insoluble material, the enzyme was purified by affinity chromatography on Ni⁺⁺ resin from Thermo Scientific (Madison, WI). Protein was dialyzed against methyltransferase buffer (50 mM HEPES [pH 7.2], 1 mM MgCl₂, and 6 mM KCl) and stored at −80 °C until use. WNV and DENV type 2 MTases were expressed in the M15 expression system from Qiagen (Valencia, CA) and purified as described above.

Fluorescence Polarization Assay

Fluorescence polarization (FP) assays were performed in 20 mM HEPES (pH 7.2) and 5 mM MgCl₂, using purified YFV MTase protein, and 10 nM FL-dT10 in a total volume of 30 µL dispensed into 384-well black microtiter plates. Binding saturation experiments were performed by dilution of purified YFV MTase serially into MTase buffer containing FL-dT10. Plates were incubated at 37 °C for 15 min and read on a Tecan Safire2 fluorescence microtiter plate reader (Tecan, Männedorf, Switzerland). For experiments containing small molecules, 1 µL of a 30× stock of compound was added to the wells prior to addition of the reaction mixture. Determination of EC₅₀ for ATA was performed as described above, using 500 nM YFV MTase. For FP experiments containing vRNA, 80 ng of YFV RNA transcript representing the first 250 nucleotides of the 5′-terminal YFV genome was added to each test well. WNV and DENV MTase were used at 500 nM for binding experiments.

Capped RNA Substrates

Capped RNA in vitro transcripts for methyltransferase assays were generated by a procedure similar to that described by Ray et al. ⁹ Briefly, capped RNA substrates representing the first 250 nucleotides of the 5′-terminal YFV genome were transcribed in vitro from polymerase chain reaction (PCR) products containing a bacteriophage T7 class II φ2.5 promoter that allows an adenosine triphosphate (ATP)–initiated transcript using T7 RNA polymerase. ²¹ In vitro YFV RNA transcripts were capped in the presence or absence of SAM with α-[³²P]-GTP and Vaccinia virus capping enzyme from CellScript (Madison, WI) to generate GpppA-RNA and m7GpppA-RNA. The resulting capped RNA was purified through a G-50 column from GE Healthcare (Madison, WI), followed by an RNeasy column obtained from Qiagen prior to use in methylation assays.

Cap Methylation Assays

The methylation reaction was performed at 37 °C for 1 h in MTase buffer, containing 5000 cpm GpppA-YFV RNA, 1.0 µM YFV MTase, and 100 µM SAM. For testing of inhibitors, MTase was preincubated with ATA for 10 min at 37 °C, and reactions were initiated by the addition of RNA and SAM. Methylation reactions were digested for 1 h at 68 °C with Nuclease P1 (US Biological, Swampscott, MA), and the resultant cap modification products were analyzed by polyethyleneimine cellulose thin-layer chromatography in 0.2 M ammonium sulfate.

Electrophoretic Mobility Shift Assay

Electrophoretic mobility shift assay (EMSA) analysis was performed by incubating 1 µM YFV MTase and 5000 cpm cap RNA representing the first 250 bp of the YFV genome with ATA and SAH for 30 min at 25 °C in 10 mM Tris (pH 9.0) and 1 mM MgCl₂. After incubation, samples were fractionated on a 7% polyacrylamide gel in 100 mM Tris (pH 9.5) and 10 mM NaCl as buffer. After electrophoresis was complete, the gel was analyzed by autoradiography for RNA binding to MTase.

Bioactives Screen

Screening of the National Institutes of Health (NIH) clinical collection was performed as described for FP assays above, using 500 nM YFV MTase, 10 nM FL-dT10, and library test compound, 33 µM. After dispensing 30-µL reaction mixtures to microtiter plate wells, library compounds were added to the wells by pin transfer of 100 nL. Plates were incubated at 37 °C for 30 min and read on a Tecan Safire2 for fluorescence measurements. ATA was used at a concentration of 10 µM as positive control for displacement of FL-dT10.

Z′ Analysis

The Z′ of the assay was calculated using the equation 1 – [3(SD1 + SD2)]/|µ1 – µ2| to compare untreated and ATA-treated wells. ²² In this formula, |µ1 – µ2| represents the absolute value of the difference between the average measurement of FP in the test and control reactions, and (SD1 + SD2) represents the sum of the standard deviations of the same two measurements. Under the conditions of the assay, FP was measured with and without test compound in a reaction containing 500 nM YFV MTase and 10 nM FL-dT10, incubated alone or with 4 µM ATA for 30 min at 37 °C; 30 wells were used for each condition tested.

Decathymidylate (dT10), a YFV RNA Surrogate Probe to Test the RNA Binding Site of YFV MTase

In undertaking the present studies, we have validated displacement of a surrogate ligand in the form of a short DNA sequence that occupies the vRNA binding site on the MTase. The selection of dT10 as a probe was based on the observed action of phosphocellulose as a pseudo-affinity medium for the absorption of DNA- or RNA-modifying enzymes.^23–26 We asked whether dT10, based on its phosphate content, might specifically bind to the YFV capping MTase via the RNA binding site. Oligo dT10 adopts a random coil configuration, with minimal ability to assume a secondary structure; it represents approximately one helical turn. Decathymidylate was derivatized at its 5′ end with fluorescein to form the probe, FL-dT10, that was then tested for its ability to bind to YFV MTase. Complex formation between FL-dT10 and YFV MTase as a function of added protein was monitored by FP, with results shown in Figure 1A . The K_d for FL-dT10 was 231 ± 12 nM, calculated directly as a function of FL-dT10 binding using nonlinear regression analysis. The K_d of binding of the FL-dT10 is similar to that described for the binding of five-base RNA by DENV, WNV, and YFV MTases. ¹⁹ Comparison of DENV, WNV, and YFV MTases for FL-dT10 binding in Figure 1B shows that DENV and WNV MTases are also capable of binding FL-dT10. The assay described above establishes that a negatively charged DNA oligonucleotide can act as a surrogate for the vRNA substrate.

OPEN IN VIEWER

Figure 1.

Binding of FL-dT10 to yellow fever virus (YFV) MTase measured by fluorescence polarization (FP). (A) Determination of K_d for the interaction between FL-dT10 and YFV MTase as a function of MTase concentration. (B) Binding of FL-dT10 by West Nile virus (WNV) and dengue virus (DENV) MTases and displacement by aurintricarboxylic acid (ATA). The reaction mixture containing FL-dT10 was supplemented with 500 nM YFV, WNV, or DENV MTase, followed by addition of 4 µM ATA as indicated, and the resultant binding was determined by FP.

YFV MTase uses both the YFV RNA and GTP together, with the guanylyltransferase activity inherent in NS5A, as substrates to form a cap (GpppA) at the 5′-end of the vRNA. This reaction is followed by methylation of the cap, with SAM as the methyl donor, first on guanine to form cap 0 (N-7MeGpppA), followed by methylation on the 2′-O of adenosine to form cap 1 (N-7GpppA2′-O). To learn whether the MTase ligands GTP and SAM displace FL-dT10, we incubated YFV MTase with FL-dT10 to form complexes in the presence of known ligands as competitors. Results, shown in Figure 2A , indicate that the complex is disrupted by ATA (a known nucleic acid binding protein inhibitor)^27–31 but not by GTP or SAH, suggesting that FL-dT10 selectively interacts with the binding site for vRNA and does not affect the binding sites for GTP and SAM/SAH. The ability of ATA to disrupt the complex is mimicked by YFV RNA, as shown in Figure 2B . Interference from GTP on fluorescein 3′ end–labeled five-base RNA binding and a weak positive effect on binding in the presence of SAM were noted previously. ¹⁹ Owing to the need for longer RNAs for N-7 methylation, and in the absence of defined RNA binding sites on the MTase, we propose that FL-dT10 may only partially overlap with the binding site of these short RNAs. Fixation of the FL-dT10 to YFV MTase confirmed that the site of FL-dT10 binding was critical for MTase activity (Supplemental Data), supporting the concept that FL-dT10 selectively interacts at an RNA binding site.

OPEN IN VIEWER

Figure 2.

Specificity of FL-dT10 displacement by MTase ligands. (A) The stability of yellow fever virus (YFV) MTase-FLdT10 complexes (lane 1) was tested in the presence of 10 µM and 100 µM ligand for guanosine-5′-triphosphate (GTP; lanes 2 and 3), S-adenosyl-L-homocysteine (SAH; lanes 4 and 5), and aurintricarboxylic acid (ATA; lanes 6 and 7). (B) The stability of YFV MTase-FL-dT10 complexes was measured in the presence of 250-bp YFV RNA transcript and ATA.

ATA, a Small-Molecule Inhibitor of Protein Nucleic Acid Interactions

ATA has been described as generally inhibiting nucleic acid–protein interactions. ATA was also shown to inhibit the DENV and WNV MTases, with a proposed binding site linked to the KDKE RNA-binding motif. ²⁰ We therefore tested the ability of ATA to disrupt FL-dT10–YFV MTase complexes. Results, shown in Figure 3 , indicate that ATA disrupted the binding of FL-dT10 to YFV MTase with an EC₅₀ = 1.54 ± 0.16 µM, which is close to the IC₅₀ for ATA inhibition of human DNA MTases DNMT1, 0.68 µM, and DNMT3a, 1.4 µM. ³² In addition, the IC₅₀ for ATA inhibition of short capped RNA 2′-O methylation was 2.3 ± 0.3 µM for DENV MTase and 4.2 ± 0.2 µM for WNV MTase. ²⁰

OPEN IN VIEWER

Figure 3.

Displacement of FL-dT10 from yellow fever virus (YFV) MTase by aurintricarboxylic acid (ATA). The EC₅₀ for destabilization of FL-dT10-YFVMTase complexes by ATA was determined by fluorescence polarization and calculated using nonlinear regression analysis.

We next determined whether YFV MTase–vRNA bound complexes are also sensitive to the disruptive effects of ATA. [³²P]-GTP-labeled cap YFV RNA was incubated with YFV MTase, resulting in the formation of MTase-RNA complexes. Reactions containing the formed complexes were supplemented with ATA, and the resultant products of the reaction were fractionated by polyacrylamide gel electrophoresis ( Fig. 4 ). As shown in Figure 4 , lane 1 contains [³²P]-GTP–labeled cap YFV RNA alone, while lane 2 displays the retardation of the RNA by added YFV MTase in the absence of inhibitor. Lanes 3 to 5 show the breakdown of the complex by addition of 10, 50, and 100 µM ATA. In contrast, 10, 50, and 100 µM SAH (lanes 6–8) failed to destabilize the complex. Experiments performed with 0.1-µM and 1-µM concentrations of ATA did not show displacement of the YFV RNA from the MTase, with displacement of RNA occurring at 10 µM ATA (Supplemental Data). Thus, we infer that ATA competes with the vRNA for binding to the RNA binding site of the MTase.

OPEN IN VIEWER

Figure 4.

Displacement of yellow fever virus (YFV) RNA from YFV MTase by aurintricarboxylic acid (ATA) or S-adenosyl-L-homocysteine (SAH). Destabilization of YFV RNA-YFV MTase complexes by ATA. The YFV RNA was incubated with YFV MTase and supplemented with increasing concentrations of ATA or SAH as shown. Reaction mixtures were analyzed by polyacrylamide gel electrophoresis gel electrophoresis followed by autoradiography. Lane 1, YFV cap RNA; Lane 2, cap RNA + MTase (complete); lanes 3–5, complete plus 10, 50, and 100 µM ATA, respectively; lanes 6–8, complete plus 10, 50, and 100 µM SAH, respectively.

To determine the functional significance of the FL-dT10 binding studies described, we tested inhibition of the enzymatic activity of YF MTase by ATA. Test samples containing MTase, [³²P]-GTP–labeled cap RNA, and SAM were assembled. The capped vRNA comprised the 250-nt 5′UTR region of the YFV genome. Results are shown in Figure 5 . The same inhibitory reagents that displace FL-dT10 from YFV MTase were shown to inhibit its enzymatic activity, further validating the use of FL-dT10 as a surrogate ligand for the vRNA. As shown in Figure 5 , ATA inhibited both N-7 and 2′-O methylation at approximately the same concentration of 12.5 µM. The IC₅₀s for ATA inhibition were previously reported for WNV to be 4.2 µM for 2′-O and 38 µM for N-7, as well as 2.3 µM and 127 µM for DENV. ²⁰ However, these values were collected with two different lengths of RNA substrates. It would be interesting to know whether WNV and DENV show a similar IC₅₀ profile for both N-7 and 2′-O when using only long RNA substrates, as shown here with YFV MTase. The need for higher concentrations of ATA for inhibition of longer RNA substrates is consistent with the polymeric binding characteristics of ATA. Taken together, these data support a mechanism of ATA fully displacing or competing for binding with the vRNA on the MTase. The studies reported above support the use of FL-dT10 as a convenient surrogate probe to replace vRNA in a high-throughput screen to discover YFV MTase inhibitors.

OPEN IN VIEWER

Figure 5.

Inhibition of yellow fever virus (YFV) MTase activity by aurintricarboxylic acid (ATA). N-7 and 2′-O activity of YFV MTase was measured in the presence of ATA. The methylation reaction was allowed to proceed for a sufficient time to obtain significant cap 1 (N-7 and 2′-O methylation). ATA was added to the reaction at the concentrations indicated. The IC₅₀ was obtained by densitometry and nonlinear regression analysis. The IC₅₀ for ATA was determined to be 12.5 ± 2 µM. Data shown represent three independent experiments. Lane 1, [³²P]-GTP labeled cap; lane 2 complete, cap 0 and cap 1 formed by the YFV MTase from labeled cap; lanes 3–9 complete, supplemented with 1, 2, 5, 10, 20, 50, and 100 µM ATA, respectively.

High-Throughput Screen of the NIH Clinical Collection

To validate the FP assay for use as a high-throughput discovery screen, we calculated a Z′ using 4 µM ATA as our control inhibitor. This analysis resulted in a Z′ of 0.83 (Supplemental Data). Since the assay was readily adaptable to high-throughput screening (HTS), we deployed the assay against the NIH Clinical Collection, a library of 742 known bioactives. This library was screened for the ability to disrupt FL-dT10-YFV MTase complex formation. The distribution of hits for a screen of this library is shown in Figure 6 . Notable hits include benzbromarone (a uricosuric agent) and multiple cephalosporins such as ceftatrizine. A list of primary hits is included in the supplemental information that accompanies this article.

OPEN IN VIEWER

Figure 6.

High-throughput screen of the National Institutes of Health Clinical Collection. The test reaction contained 500 nM MTase and 10 nM FL-dT10. In total, 30 µL of test reaction per well was dispensed into 384-well plates and supplemented with 100 nL library test compound added by pin transfer. Fluorescence polarization was determined and plotted as a function of sample number. ATA, aurintricarboxylic acid.

Multiple screening hits involving cephalosporins and benzofurans were obtained, suggesting that upon suitable modification, they might be repurposed for the treatment of flaviviral infections. Taken together, these data establish the use of a fluorescein-labeled dT10 as a suitable surrogate ligand for the YFV vRNA. Although this assay does not directly measure MTase activity, it is well suited to HTS and also represents a new way to screen for inhibitors of nucleic acid binding proteins. DNA methyltransferases have been considered targets for cancer chemotherapy.^33–35 It will be interesting to learn the extent to which the primary assay described here will be adaptable to the discovery of other nucleic acid–modifying enzymes.