Fluorescence-Based Activity Screening Assay Reveals Small Molecule Inhibitors of Vaccinia Virus mRNA Decapping Enzyme D9.

Vaccinia virus (VACV) represents a family of poxviruses, which possess their own decapping machinery as a part of their strategy to eliminate host mRNAs and evade the innate immune response. D9 is one of the two encoded VACV decapping enzymes that is responsible for cap removal from the 5' end of both host mRNA transcripts and viral double-stranded RNAs. Little is known about the structural requirements for D9 inhibition by small molecules. Here, we identified a minimal D9 substrate and used it to develop a real-time fluorescence assay for inhibitor discovery and characterization. We screened a panel of nucleotide-derived substrate analogues and pharmacologically active candidates to identify several compounds with nano- and low micromolar IC50 values. m7GpppCH2p was the most potent nucleotide inhibitor (IC50 ∼ 0.08 μM), and seliciclib and CP-100356 were the most potent drug-like compounds (IC50 0.57 and 2.7 μM, respectively). The hits identified through screening inhibited D9-catalyzed decapping of 26 nt RNA substrates but were not active toward VACV D10 or human decapping enzyme, Dcp1/2. The inhibition mode for one of the compounds (CP-100356) was elucidated based on the X-ray cocrystal structure, opening the possibility for structure-based design of novel D9 inhibitors and binding probes.

Introduction

Eukaryotic mRNAs are co-transcriptionally modified by the addition of a 7-methylguanosine triphosphate moiety at the 5′ end. This modification, referred to as the 5′ cap, protects mRNA from premature degradation by 5′-exonucleases and is necessary for canonical translation initiation. Only specialized decapping enzymes can remove the 7-methylguanosine cap from an mRNA body, thereby exposing it to 5′-to-3′ degradation. Hence, mRNA decapping is an important process, aiding the control of gene expression. Eukaryotic cells produce several decapping enzymes that participate in bulk mRNA decay, mRNA quality control, and specialized degradation pathways.[1−4] Two major decapping enzymes involved in bulk mRNA degradation are Nudix hydrolase, Dcp2, which is involved in the 5′–3′ degradation pathway and scavenger decapping enzyme, DcpS, which cleaves m7G-capped structures released as the products of 3′–5′ mRNA decay.[5,6] Moreover, other proteins possessing intrinsic decapping activity have been identified and include Nudix family hydrolases (e.g., Nudt3 and Nudt16) and DXO decapping exoribonuclease protein, which is responsible for degrading aberrantly capped RNAs.[7] Precise control of the activity of those enzymes is crucial to maintain the cell homeostasis.

Some viruses express their own decapping machinery as a part of their strategy to inhibit the cellular translation and promote the synthesis of viral proteins. Till date, four virus-encoded decapping enzymes have been described: D9 and D10 from vaccinia virus (VACV), g5R from African swine fever virus, and L375 Nudix enzyme from mimivirus.[8−11] All three viruses belong to the family of nucleocytoplasmic large DNA viruses.

The viral decapping enzymes differ greatly in the substrate specificity but share the catalytic motif and the ability to cleave m7GpppN-RNAs to release m7GDP as a product (Figure A). They represent a class of phosphohydrolases that cleave nucleoside diphosphates linked to another moiety X (Nudix). This Nudix motif consists of a 23 amino acid characteristic sequence GX5EX5[UA]XREX2EEXGU, wherein U represents an aliphatic, hydrophobic residue, and X represents any amino acid.[12] Albeit VACV is not pathogenic to humans, it represents a family of large DNA viruses, including zoonotic viruses, which can jump to humans and/or are pathogenic to livestock and humans. D9 and D10 from VACV are therefore prototype viral decapping enzymes, the studies of which may aid in elucidating how viruses use their decapping machinery to avoid the innate immune response and influence the host translation.

Figure 1

D9 as a viral RNA decapping enzyme. (A) Biochemistry of RNA cleavage and (B) summary of the key D9 functions during viral infection.

To ensure mRNA stability and translational activity, VACV adds 5′ caps onto viral mRNAs using the virus-encoded capping machinery.[13−15] Early VACV mRNAs are modified to form either m7GpppAm or m7GpppGm caps, whereas intermediate and late transcripts carry only m7GpppAm caps.[16,17] D9 is expressed in the early stage of infection and[8] together with D10, it prevents accumulation of dsRNAs, thereby minimizing cellular antiviral response mediated by PKR and OAS/RNase L.[13,18] D9 exhibits specificity for N7-methylated cap structures and a preference for longer RNA substrates (>30 nt).[8] Unmethylated (Gppp-capped) RNAs are not cleaved. The enzyme is strongly inhibited by uncapped RNA, whereas N7-methylated guanine nucleotides such as m7GpppG, m7GTP, and m7GDP are less potent D9 inhibitors. Both D9 and D10 enzymes trigger the degradation of multiple cellular and viral mRNAs to promote the selective translation of viral mRNAs[19] and reduce the levels of dsRNAs, which are the main trigger of innate immune response (Figure B). Interestingly, VACV deficient in either D9 or D10 is more immunogenic due to dsRNA accumulation, replicates more efficiently in cancerous than in healthy cells, and thus acts as effective oncolytic virus.[20] Significant effort has been made to understand the role of D9 and D10 in VACV replication; nonetheless, the differential functions of these enzymes, their regulation, and interdependence remain elusive.

Structural and functional studies on cellular decapping enzymes, both in vitro and in vivo, can be greatly facilitated by the design of specific small-molecule ligands and inhibitors.[21−23] As such, in this work, we aimed to design molecular tools that could find utility in studying the activity of recombinant D9 and aid discovery of small-molecule inhibitors of this enzyme. Little has been determined about cap-related substrate specificity of D9 or design of specific inhibitors. The so-far-reported inhibition studies relied on isotopically labeled RNA substrates, the use of which entails relatively high cost and low throughput. Here, we sought to design a new simple approach that can be used to study the activity of VACV D9 and discover inhibitors via high-throughput screening (HTS). To this end, we developed fluorescence intensity-based (FLINT) assay with a small-molecule fluorescent probe (Figure A). The designed method was adopted to the HTS format to screen an in-house library of nucleotide-derived compounds and a commercially available library of pharmacologically active compounds (LOPAC1280, Sigma Aldrich). The identified hits were verified on short RNA substrates and cross-examined against VACV D10 and human decapping enzyme, Dcp1-Dcp2. Finally, one of the most potent inhibitors was crystalized in complex with D9, providing a deeper insight into the inhibition mode.

Figure 2

Development of an activity-based assay for D9. (A) Anticipated working mode of fluorescence intensity probes for D9; (B) structures of the evaluated fluorescent probes; and (C) time-dependent fluorescence intensity changes induced by D9 cleavage. Reaction conditions: probe 1a–1e (100 nM), D9 (50 nM) in 10 mM Tris·HCl containing 100 mM KOAc, 2 mM MgCl2, and 0.5 mM MnCl2, pH 7.5. For each probe, fluorescence changes at emission maximum were monitored (1a—exc. 490/em. 512 nm; 1b, 1d, and 1e—345/378 nm, and 1c—420/489 nm). (D) Time-dependent emission changes for probe 1b recorded at 1 min interval under conditions given in (C).

Results

Development of Activity-Based Assay for D9 Enzyme

We first synthesized a set of fluorescently labeled nucleotides as potential fluorogenic substrates for D9. Knowing that D9 specifically cleaves 7-methylguanosine-capped transcripts, all probes were equipped with a 7-methylguanine nucleotide moiety as a recognition element and a fluorescent tag sensitive to m7G presence as a cleavage-responsive element (Figure A). 7-Methylguanosine has been shown to act as a quencher of electron-rich fluorescent dyes, giving the basis for the development of fluorescent turn-on probes.[24] To select the most suitable probe in terms of substrate and fluorescent properties, we tested several fluorescent dyes [pyrene (Py), perylene (Pe), and boron dipyrromethene (BODIPY-FL)] and different oligophosphate chain lengths (from tri- to pentaphosphate; Figure B). The compounds were subjected to enzymatic cleavage by D9, and emission spectra were recorded over time (Figure C,D). In all cases, the gradual increase in fluorescence intensity was observed, consistent with cleavage by D9 enzyme. The greatest changes were observed for probes labeled with pyrene, but the efficiency varied depending on the length of the oligophosphate chain. The system response decreased with the length of the oligophosphate (over 14-, 4-, and 3-fold enhancement of the fluorescence signal for tri-, tetra-, and pentaphosphate, respectively); triphosphates were also the most rapidly cleaved substrates (Figure S1). Therefore, the pyrene-labeled m7GTP analogue (1b) was selected for further studies as a turn-on probe with the optimal substrate and emission properties. As m7GTP-Py is structurally different from natural D9 RNA substrates, we tested whether the hydrolysis mechanism remains similar. The probe was subjected to D9 cleavage, and the reaction progress was monitored using RP-HPLC (Figure S2A,B). Appropriate fractions containing hydrolysis products were collected and identified by mass spectrometry (Figure S2C–F), proving conclusively that D9 enzyme cleaves the probe liberating m7GDP and monophosphate of the rest of the molecule, that is, with the same regioselectivity as observed for capped RNA.

Probe 1b was then characterized in more detail and employed to develop high-throughput screening assay in 96-well plates (Figure A). Buffer compositions, concentrations of divalent metal ions, and protein-stabilizing agents were adjusted to optimize the reaction conditions. Manganese (Mn2+) ions were essential for efficient enzyme activity (Figure S3A), whereas addition of BSA as a stabilizing protein significantly improved reproducibility (Figure S3B). Initial hydrolysis rates for varying 1b concentrations were next determined by measuring fluorescence intensity changes as a function of time. Plotting the initial rates as a function of substrate concentrations enabled the determination of Michaelis–Menten kinetic parameters for the probe (KM, Vmax, and kcat; Figure B). Based on those results, we established optimal conditions (Table ) that ensured sufficient fluorescence signal response, proper reaction kinetics, high protein stability, and good reproducibility. Z′ factor[25] determined under the optimized conditions was 0.75 (Figure C), indicating that the quality of the assay is sufficient for the application in high-throughput experiments.

Figure 3

Validation of fluorescence-based assay for D9 activity and evaluation of an in-house library of nucleotide-based compounds. The assay was performed at 30 °C in a 96-well format using 3 μM 1b, in the presence of the putative inhibitor (half-logarithmic dilution from 100 μM) and 5 nM D9. (A) Application of the fluorescence-based method to monitor D9 activity in the HTS mode; (B) determination of the kinetic parameters for probe 1b (0–15 μM) in the presence of 5 nM D9. Data shown are means from three independent experiments ±SEM; (C) determination of the Z-factor for D9 inhibition assay. Reaction conditions: 3 μM 1b, 5 nM D9 in the absence (positive control) or presence (negative control) of the 100 μM nucleotide inhibitor. The experiment was carried out in a 96-well plate reader in 10 mM MOPS·HCl, pH 7.0 buffer containing 100 mM KOAc and 2 mM MgCl2, 0.3 mM MnCl2, 2 mM DTT, and 0.1% BSA at 30 °C; and (D) determined IC50 values. Only compounds with IC50 < 60 μM are shown. All IC50 values, along with compound structures, are shown in Table S1; (E) representative inhibition curves and (F) structures of the most potent compounds from mono-, di-, and trinucleotide groups. Data are shown as mean values from triplicate experiment ±SEM.

Table 1

Optimized Conditions for D9 Hydrolytic Activity Assay Based on Probe 1b

conditions used in D9 hydrolytic activity experiments
probe concentration	3 μM
enzyme concentration	5 nM
buffer	10 mM MOPS·HCl, pH 7.0, 100 mM KOAc, containing 2 mM MgCl2, 0.3 mM MnCl2, 2 mM DTT, and 0.1% BSA
excitation/emission wavelength	345/378 nm
temperature	30 °C
preincubation	15 min, 300 rpm

Inhibition Studies

To identify the structural preferences for D9 inhibition by substrate analogues, the optimized D9 assay was employed to evaluate an “in-house” library consisting of unmodified mono-, di-, and trinucleotides and their analogues. To assess whether such approach is feasible, we first tested susceptibility of selected nucleotides to D9 cleavage. A set of 15 compounds with different modifications were incubated with D9 for 2 h under assay conditions. Most of the compounds were unsusceptible or did not undergo significant hydrolysis during initial 30 min of the analysis (Figure S4).

Next, the half-maximal inhibitory concentrations (IC50) were determined for all nucleotides from the library by observing the degradation of 1b in the presence of different concentrations of the putative inhibitor (Figure D, Table S1). The determined IC50 values varied between 0.077 and over 100 μM. The strongest D9 inhibitors (IC50 below 1 μM, Figure E,F) were identified among mononucleotides containing m7G moiety and three or more phosphate groups, indicating that both positively charged 7-methylguanosine and negatively charged phosphate bridge are required for tight interactions with the protein. Dinucleotide cap analogues generally showed weaker inhibitory properties (IC50 from 2.6 up to 46 μM) but relatively strong inhibition was observed for several dinucleoside 5′,5′-tetraphosphates, including m7GmpppCH2pG (IC50 = 2.6 ± 0.4 μM) and m7GpSpppSm7G D3 (IC50 = 3.6 ± 0.4 μM); the latter has been previously identified as an inhibitor of yeast decapping enzyme SpDcp1/Dcp2.[26] The most potent compound in the group of trinucleotides was m7GmpppGmpG with IC50 value of ∼4 μM. Interestingly, both in the group of dinucleotides and trinucleotides, 2′-O methylation of 7-methylguanosine notably increased inhibitory properties, suggesting that this modification might increase binding affinity for D9. At the same time, compounds that were 2′-O methylated either at 7-methylguanosine or within the second nucleotide showed the highest susceptibility to hydrolysis by D9 (Figure S4), which might suggest that this is a binding affinity-related effect.

In addition, we screened a commercially available compound library consisting mostly drug-like compounds (LOPAC1280, Sigma-Aldrich). Compounds were screened at 30 μM under conditions given in Table (Figure A). The relative reaction progress for each compound was calculated by normalizing the initial rate of the reaction in the presence of the tested compound to the initial rate in the absence of any inhibitor. The criterion of the cutoff for inhibitor selection for further evaluation was set to 30% of the maximal initial rate (dotted line), which yielded six potential hits. The hits were then further evaluated, including determination of their IC50 values (Figure B,C). Three most potent compounds had IC50 value below 10 μM, and they included an adenine derivative seliciclib (IC50 = 0.57 ± 0.05 μM), quinazolinamine derivative CP-100356 (IC50 = 2.7 ± 0.7 μM), and a purine derivative SCH 58261 (4.5 ± 0.8 μM) (Figure D).

Figure 4

Identification of D9 inhibitors by LOPAC1280 library screening. (A) Relative reaction progress of D9-catalyzed (5 nM) hydrolysis of probe 1b (3 μM) with 1280 compounds from the LOPAC library (30 μM). Data are means from the duplicate experiment; (B) IC50 values for six selected hits from the screening experiment (30% cutoff for inhibitor selection). Numerical values are shown in Table ; (C) inhibition curves for identified inhibitors—one compound had to be tested separately due to its spectroscopic properties. Data are mean values ± SEM from three independent experiments; (D) structures of six most potent hits, and their IC50 values are shown in Table .

Table 2

IC50 Values for Inhibitors Identified by HTS Screening

LOPAC ID	compound	IC50 (μM)
P2_H5	amsacrine hydrochloride (A 9809)	12.6 ± 1.7
P3_B7	CP-100356 monohydrochloride (PZ0171)	2.7 ± 0.7
P4_A6	seliciclib	0.57 ± 0.05
P5_A5	CGS-15943 (C-199)	20.1 ± 5.2
P10_F4	BIO (B 1686)	95 ± 18
P16_F10	SCH 58261 (S4568)	4.5 ± 0.8
Reference Compounds
	m7GpppCH2p	0.18 ± 0.01
	m7GpppG	15 ± 2

Verification of Inhibitory Properties toward D9 for the Identified Hits

To further validate the hits, we checked whether the three most potent inhibitors from the LOPAC library and the best hit from nucleotide-derived library inhibit D9-catalyzed decapping of short capped RNA transcripts. A 26-nt long RNA substrate co-transcriptionally capped with m7GpppApG (cap 0 structure)[27] was subjected to D9-catalyzed hydrolysis in the presence of the evaluated compound at three different concentrations (1, 10, and 100 μM). The reaction progress was analyzed electrophoretically in reference to the reaction performed in the absence of the inhibitor (Figure A). The band intensities corresponding to capped and uncapped RNAs were quantified densitometrically, and the decay of the substrate was plotted as a function of time (Figures B,C, S5). All compounds showed inhibitory activity toward RNA decapping, and their relative potencies were in good agreement with IC50 values determined by the fluorescence assay using probe 1b. Finally, we tested the specificity of the compounds in the context of VACV D10 and human decapping enzyme Dcp1/2, which are also m7GpppRNA hydrolases belonging to the Nudix family. We did not observe any inhibitory activity toward Dcp1/2 for the tested compounds at concentrations up to 100 μM, thereby confirming their selectivity toward D9 (Figures D,E, S6). m7GpppCH2p and CP-100356 showed weak inhibitory activity toward D10 (Figure S7). The most potent drug-like D9 inhibitor identified in this study, seliciclib, showed very good selectivity for D9.

Figure 5

Verification of inhibitory properties and selectivity of the identified hits in a decapping assay on short RNA substrates. (A) Idea of RNA decapping assay based on short m7Gppp-RNA substrates (26 nt) and electrophoretic analysis and (B) representative gel (right) from single experiment with compound CP-100356 and densitometric analysis (left) of gel band intensities. m7Gppp-capped RNA (20 ng) was incubated for 1 h at 37 °C with D9 (3 nM) without or with the presence of the tested inhibitor at 1, 10, or 100 μM. Samples from different time points were analyzed by polyacrylamide gel electrophoresis (PAGE) and stained with SYBR Gold. The data for all tested compounds are shown in Figure S5; (C) results from triplicate experiments ±SEM for four identified hits confirming their inhibitory properties toward D9 decapping enzyme; and (D,E) analogous experiments with human Dcp1/2 complex (11 nM)—data show no inhibitory properties of tested compounds toward this enzyme. More details are given in the Supporting Information.

Crystallographic Structure of Vaccinia Virus D9 with CP-100356

To better understand the mechanism of inhibition, we determined the crystal structure of wild-type D9 bound to inhibitor CP-100356 (P3_B7) to 1.78 Å (Figure A, Table ). This crystal form has two copies of D9 in the asymmetric unit (all-atom RMSD 1.811 Å), each bound to 1 molecule of CP-100356. The conformation of CP-100356-bound D9 is nearly identical to a recently published structure of D9 bound to m7GDP product in the post-catalytic, inactive conformation (all-atom RMSD 2.55 Å). Our crystal structure reveals that the CP-100356 is positioned in the cap-binding pocket by continuous stacking of the 6,7-dimethoxyquinazoline moiety between aromatic residues F54 and Y158 in a manner similar to m7G cap recognition in the product-bound structure, PDB 7SEZ (Figure B,C). These conserved residues are essential for efficient cap hydrolysis.[28] However, in the complex with CP-100356, Y158 is significantly reoriented compared to the position occupied in the complex with m7GDP (Figure B,C), presumably to allow more efficient π–π stacking interaction with the aromatic quinazoline system of the ligand. Unlike the m7G cap in PDB 7SEZ, which is oriented by hydrogen bond base pair mimicry between the guanine base and residues D151 and E16, the CP-100356 inhibitor is not oriented by hydrogen bonding. Rather, the 6,7-dimethoxy-1,2,3,4-tetrahydroisoquinoline moiety extends into a small cavity past D151 and E16 (Figure C). It is clear from our crystal structure that CP-100356 inhibits D9 cap hydrolysis by stabilizing the enzyme in an inactive conformation, preventing conserved aromatic residues F54 and Y158 from binding capped mRNA substrate. Attempts to crystallize D9 with the other most potent inhibitors identified from our screen, seliciclib (P4_A6) and SCH 58261 (P16_F10), were unsuccessful. Nonetheless, based on the similarity of these compounds to natural nucleobase moieties, we hypothesize that they likely bind in a similar manner, that is, by stacking between F54 and Y158.

Figure 6

CP-100356 occupies the cap-binding pocket of D9. (A) Alignment of D9 crystal structures obtained by co-crystallizing with m7GDP (teal, PDB 7SEZ) or CP-100356 (yellow, PDB 7T7H). The m7GDP and CP-100356 molecules are green and orange, respectively. The all-atom RMSD is 2.55 Å. (B) Close-up view showing positioning of the m7GDP product in PDB 7SEZ. The methylated guanine base is stacked between conserved aromatic residues F54 and Y158 and hydrogen bonds with E16 and D151. (C) Close-up view showing positioning of the CP-100356 inhibitor in the cap-binding site, sandwiched between F54 and Y158.

Table 3

Data Collection and Refinement Statistics

	D9 + CP-100356 (PDB: 7T7H)
Data Collection
space group	P212121
Cell Dimensions
a, b, c (Å)	56.77, 82.77, 96.31
α, β, γ (deg)	90, 90, 90
resolution (Å)	62.77–1.78 (1.844–1.78)a
Rmerge	0.04373 (0.9509)
I/σ(I)	30.35 (2.29)
completeness (%)	99.79 (97.95)
redundancy	12.9 (12.0)
Refinement
resolution (Å)	62.77–1.78
no. unique reflections	44,130
Rwork/Rfree	0.1864/0.2182
No. Atoms
protein	3294
ligand/ion	84
water	194
B-Factors
protein	48.3
ligand/ion	61.3
water	50.8
RMS Deviations
bond lengths (Å)	0.019
bond angles (deg)	1.97
Ramachandran (%)
favored	98
allowed	1.74
outliers	0.26

Values in parentheses are for the highest resolution shell, and the data set was collected from a single crystal.

Conclusions

Literature reports that VACV D9 requires RNA body for decapping activity, and that the decapping efficiency increases with the length of the RNA.[8] Surprisingly, in this work, we discovered a set of minimal substrates (activity probes) for D9 that consist of m7GTP molecule connected to a fluorescent tag via the terminal phosphate moiety. The activity probes were evaluated as D9 substrates to select the most efficient and responsive internally quenched probes suitable for high-throughput assay development. Pyrene-labeled m7GTP (probe 1b) had the most favorable properties, both in term of hydrolysis kinetics and fluorescent response upon cleavage. Probably both the m7G nucleotide and the pyrene moiety contribute to recognition by D9 because the corresponding unlabeled compounds such m7GTP and m7GpppG are very poor D9 substrates (as determined by RP HPLC). The probe was used to develop fluorescent HTS assay to screen for D9 inhibitors. The addition of Mn2+, beside Mg2+ ions, and the presence of BSA were necessary to maintain high protein activity and stability. The optimized assay was characterized by a Z′ factor of 0.75, that is, suitable for HTS. The screening experiments were performed on two distinct compound libraries. The first library consisted of nucleotide derivatives, most of which were substrate analogues of varying sizes (mono-, di-, and trinucleotides), and the second was commercially available LOPAC). The nucleotide screening revealed that highly negatively charged 7-methylguanosine mononucleotides are the most potent inhibitors of D9. The LOPAC screening revealed several heteroaromatic compounds as potential hits, which were then further evaluated to show good IC50 values and verified inhibitory properties on a longer set of substrates. The most interesting hits identified in the screening were seliciclib and CP-100356. Seliciclib, a 2,6-diaminopurine derivative, was the most potent D9 inhibitor, which combined with its very low inhibitory activity against D10 and hDcp1/2 making it a good candidate for future development. Albeit we were not able to determine the crystal structure of D9 in complex with this inhibitor, its structure suggests that the recognition may be mediated by amino acid-forming nucleic acid-binding sites. CP-100356 was another potent and selective inhibitor of D9 with high potential of future development. The crystal structure of recombinant D9 in complex with CP-100356 was determined, which revealed that the compound targets the m7GDP-binding site and forms stacking interactions similar to those observed for 7-methylguanine, suggesting a competitive inhibition mode. This, combined with the observed selectivity of the compound upon cross-examination with VACV D10 and human PNRC2-Dcp1/Dcp2 complex, indicates that the cap-binding site in D9 is distinctive enough from other decapping Nudix family hydrolases to enable the development of small molecules selectively targeting this chemical space.

Methods

Expression and Purification of Recombinant Viral D9 and D10 Proteins

Recombinant viral D9 decapping enzyme with additional C-terminal His-tag sequence was expressed, as previously described.[28] The concentration of the protein was determined spectrophotometrically by assuming ε280 = 14,900 M–1 cm–1. The enzyme was stored at −80 °C in a storage buffer (10 mM MES, pH 7.0, 300 mM NaCl, 1 mM DTT, and 10% glycerol).

Viral D10 decapping enzyme was expressed following a similar protocol, and the detailed procedure will be published elsewhere.

Expression and Purification of Recombinant Human PNRC2-Dcp1/2 Complex

Human PNRC2-Dcp1/2 decapping complex genes were obtained in the plasmid vector (pETDuet_PNRC2-Dcp1/2). Protein complex (for sequence details, see Figure S8) with His-tag was overexpressed in the E. coli BL21(DE3) RIL strain in LB medium with ampicillin (100 μg/mL). Cells were grown to OD600 ∼0.5 at 37 °C, then temperature was adjusted to 20 °C. When the culture reached OD600 of 0.6–0.7, protein expression was induced by adding isopropyl β-D-1-thiogalactopyranoside (IPTG) to the final concentration of 1 mM, followed by overnight incubation at room temperature with shaking. Cells were then harvested by centrifugation (7000g, 4 °C) and lysed by sonication in 50 mM sodium phosphate (pH 7.5), 300 mM NaCl, 10 mM imidazole, 5% glycerol, 1 mM DTT with lysozyme (1 mg/mL), and protease inhibitors (1 mM PMSF, 10.5 μM leupeptin, and 1 μM pepstatin A). The lysate was centrifuged at 40,000g for 30 min at 4 °C, and the protein was purified by immobilized metal affinity chromatography on the GE Healthcare HisTrap HP column. The column was washed with 50 mM phosphate (pH 7.5), 1 M NaCl, 30 mM imidazole, and 1 mM DTT, and the PNRC2-Dcp1/2 complex was eluted using 50 mM phosphate buffer (pH 7.5), 300 mM NaCl, 500 mM imidazole, and 1 mM DTT. The supernatant was loaded onto the GE Healthcare HisTrap Heparin HP column and eluted with 50 mM sodium phosphate (pH 7.5), 1 M NaCl, and 1 mM DTT. The last step of purification was size exclusion chromatography on the GE Healthcare HiLoad 16/600 200 pg column. The purified human PNRC2-Dcp1/2 complex was concentrated using 30 kDa Amicon filters and flash frozen in liquid nitrogen. The concentration of the protein was determined spectrophotometrically by assuming ε280 = 82,445 M–1 cm–1. The enzyme was stored at −80 °C in a storage buffer (50 mM Hepes, pH 7.5, 150 mM NaCl, 1 mM DTT, and 20% glycerol).

Fluorescent Probe Synthesis

All solvents and chemical reagents for non-fluorescent nucleotide synthesis were purchased from Sigma-Aldrich and used without any pre-treatment, unless otherwise indicated. BODIPY FL iodoacetamide was acquired from ThermoFisher Scientific. Pyrene azide and perylene azide were purchased from Lumiprobe. Guanosine and guanosine 5′-monophosphate disodium salts were purchased from Carbosynth. m7GTPγS triethylammonium salt was synthesized, as described previously.[29] Probes 1b, 1d, and 1e were synthesized, as described previously.[24]

Fluorescent nucleotides were purified using analytical or semi-preparative HPLC. Analytical HPLC was performed on Agilent Tech. Series 1200 using the (RP)Supelcosil LC-18-T HPLC column (4.6 × 250 mm, flow rate 1.3 mL/min) with a linear gradient of acetonitrile (0–100% from 0 to 15 min) in 0.05 M ammonium acetate buffer (pH 5.9). UV detection was performed at 260 nm, as well as fluorescent label absorption and fluorescence maximum. Semi-preparative RP HPLC was performed on the same apparatus equipped with the Grace Vision HT C18 HL column (250 cm × 22 mm, 10 μm, flow rate 5.0 mL/min) with a linear gradient of acetonitrile (0–100% from 0 to 120 min) in 0.05 M ammonium acetate buffer (pH 5.9) and UV detection at 260 nm and at the fluorescent tag absorption and emission maximum. The structure and homogeneity of each final product were confirmed by RP HPLC and high-resolution mass spectrometry HRMS (ESI–) with a Thermo Scientific LTQ Orbitrap Velos mass spectrometer.

m7GTPγS-BODIPY-FL (1a)

m7GTPγS triethylammonium salt (7.5 mg, 8.8 μmol) was mixed with BODIPY-FL iodoacetamide (1.0 mg, 2.4 μmol) in 113 μL of DMSO. The reaction was carried out at room temperature for 1 h and stopped by addition of 0.85 mL of H2O. The precipitated fluorescein dye remains were centrifuged. The reaction product was purified using analytical RP HPLC (Method A3). The collected eluate was lyophilized repeatedly to afford 0.5 mg (0.6 μmol) of 1a as an orange solid. HPLC yield: 58%. Yield after purification: 25%.

HRMS (−)ESI m/z; found, 841.0976, calcd for C25H31BF2N8O14P3S–: 841.3535.

m7GTPC4Pe (1c)

Aqueous solution of m7Gp3C4H5[30] triethylammonium salt (1.5 mg, 1.9 μmol) was mixed with a solution of perylene azide (2.0 mg, 4.4 μmol) in DMSO (80 μL), followed by addition of 1 M aqueous solutions of CuSO4·5H2O (0.8 mg, 3 μmol, 3 μL) and 1 M sodium ascorbate (1.2 mg, 6 μmol, 6 μL). The reaction was carried out at 37 °C for 24 h with shaking (350 rpm) and quenched by addition of Na2EDTA (1.8 mg, 4.8 μmol) and water (100 μL). The precipitated perylene dye was separated from the mixture by centrifugation. The reaction product was purified by semi-preparative RP HPLC to give 1c (0.15 mg, 0.14 μmol) as an ammonium salt. HPLC yield: 81%. Yield after purification: 7.3%.

HRMS (−)ESI m/z; found, 1024.2221, calcd for C42H45N9O16P3–: 1024.2203.

Spectroscopic Measurements

Fluorescence emission spectra were recorded using the Agilent Cary Eclipse fluorescence spectrophotometer equipped with the xenon lamp under temperature-controlled conditions. All experiments were performed at 37 °C in the Hellma quartz cuvette (10 × 4 mm) with a sample volume 1000 μL in 10 mM Tris·HCl (pH 7.5), 100 mM KOAc, 2 mM MgCl2, 0.5 mM MnCl2, and 2 mM DTT. Before each measurement, the buffer was degassed. Emission spectra were registered for all excitation maxima (detailed wavelengths for each probe: 1a—exc. 490/em. 512 nm; 1b, 1d, and 1e—345/378 nm; and 1c—420/489 nm).

Susceptibility to D9 Hydrolysis/Probe Hydrolysis Experiments

The enzymatic activity of D9 was assayed at 30 °C with mixing at 300 rpm in 10 mM Tris·HCl (pH 7.5) containing 100 mM KOAc, 2 mM DTT, 2 mM MgCl2, and 0.3 mM MnCl2. Reaction mixtures contained 30 μM (nucleotide analogue) or 100 μM (probe 1b) of the tested compound and 25 nM of the recombinant protein. Enzymatic reaction progress was examined after 5, 15, 30, 45, and 60 min. Aliquots were collected, and the reaction progress was terminated by heat inactivation for 3 min at 95 °C, followed by centrifugation. The samples were analyzed by RP HPLC (Agilent Technologies Series 1200). Appropriate fractions from probe 1b hydrolysis were collected and identified based on the mass measurements. Mass spectra were recorded on the AB Sciex API 3200 spectrometer.

Point Fluorescence Measurements—General Information

Point fluorescence measurements were performed using the BioTek Synergy H1 microplate reader. Before each measurement, the reaction buffer was degassed. Experiments were performed in the Greiner 96-well, black, non-binding plates at 30 °C. Before the measurement, the plate was preincubated for 15 min at 30 °C with mixing (300 rpm). The point fluorescence was registered with 1 min interval and detection at 345 nm excitation wavelength and 378 nm emission wavelength.

Kinetic Parameter Determination

To determine the kinetic parameters of enzymatic hydrolysis of probe 1b by D9 enzyme, point fluorescence measurements were performed. Each well of the plate contained an appropriate buffer (10 mM MOPS·HCl, pH 7.0, 100 mM KOAc, containing 2 mM MgCl2, 0.3 mM MnCl2, and 2 mM DTT) with addition of 0.1% BSA, a substrate (probe 1b at a concentration range 0–15 μM) and an enzyme (5 nM of D9) to a final volume of 150 μL. The measurement was carried out on until full saturation, which allowed us to convert the initial rates from au/min to mol/min/mg. The initial rates were calculated by fitting a linear curve to the first 10 points (10 min) of the hydrolysis reaction. To calculate the kinetic parameters, the Michaelis–Menten model was applied with the following formula fitted to the experimental data using GraphPad Prism softwarewhere V is the initial rate of the reaction, Et is the concentration of enzyme sites, kcat is the turnover number indicating the catalytic efficiency, Cprobe is the concentration of the probe 1b, Km is the Michaelis–Menten constant, and Vmax is the maximum enzyme velocity.

Z′ Factor Determination

To determine the Z′ factor of the method, point fluorescence measurements were performed. The positive control contained a mixture of probe 1b (3 μM) and enzyme D9 (5 nM) in 10 mM MOPS·HCl, pH 7.0, 100 mM KOAc, containing 2 mM MgCl2, 0.3 mM MnCl2, and 2 mM DTT. The negative control additionally included a nucleotide inhibitor (30 μM of m7GpppCH2p[31]) to fully hinder the D9 activity. The value of Z′ parameter was calculated as followswhere σpos and σneg are the standard deviations and μpos and μneg are the mean values of the positive and negative controls, respectively.

High-Throughput Screening of the LOPAC1280 Library

To screen the LOPAC1280 Library, point fluorescence measurements were performed. Each plate well contained an appropriate buffer (10 mM MOPS·HCl, pH 7.0, 100 mM KOAc, containing 2 mM MgCl2, 0.3 mM MnCl2, and 2 mM DTT) with addition of 0.1% BSA, a fluorescent substrate (3 μM of probe 1b), an inhibitor (30 μM of each LOPAC compound), and an enzyme (5 nM of D9) to a final volume of 150 μL. To determine the relative reaction progress, the ratio of the initial reaction rate with inhibitor to the initial rate without inhibitor was calculated. Nine out of the 1280 compounds showed interference with pyrene fluorescence and were tested separately with different settings of fluorescence read-out (different gains).

IC50 Parameter Evaluation

Experiments to determine IC50 parameters were performed as point fluorescence measurements. Each well contained an appropriate buffer (10 mM MOPS·HCl, pH 7.0, 100 mM KOAc, containing 2 mM MgCl2, 0.3 mM MnCl2, and 2 mM DTT) with addition of 0.1% BSA, a fluorescent substrate (3 μM of probe 1b), an inhibitor (in half-logarithmic dilution, cinh ∈ 0, 100 μM), and an enzyme (5 nM of D9) to a final volume of 150 μL. To determine the IC50 values, a three parameter dose–response formula with standard hill slope from GraphPad Prism software was fittedwhere V/V0 is the relative initial rate of the reaction, Imin and Imax are bottom and top plateaus, and Cinh is the concentration of the inhibitor.

All tested compounds come from LOPAC1280 (Table ) or our in-house library of cap analogues (determined IC50 values with compound’s structures and synthesis references are available in Table S1).

RNA Synthesis

Short RNA sequences were synthesized by in vitro transcription (IVT) using T7 class II promoter Φ2.5 initiated by ATP (TAATACGACTCACTATTA).[32] The typical transcription reaction (200 μL) was carried out for 4 h at 37 °C and contained 5 mM each of UTP/GTP/CTP (Thermo Fisher Scientific), 1.25 mM ATP (Thermo Fisher Scientific), and 6.5 mM of trinucleotide cap analogue (m7GpppApG), RNA Polymerase Buffer (Thermo Fisher Scientific), 1 μM of annealed oligonucleotides as DNA template (CAGTAATACGACTCACTATTAGGGAAGCGGGCATGCGGCCAGCCATAGCCGATCA and TGATCGGCTATGGCTGGCCGCATGCCCGCTTCCCTAATAGTGAGTCGTATTACTG), 1 U/μL RiboLock RNase inhibitor (Thermo Fisher Scientific), 20 mM MgCl2, and T7 RNA polymerase (20 U/μL Thermo Fisher Scientific). Uncapped transcripts (IVT) were obtained as mentioned above but in the presence of 5 mM ATP and without addition of cap analogue. After 4 h incubation, 7 U of DNase I (Thermo Fisher Scientific) was added, and the mixture was incubated for another 30 min at 37 °C. The reaction mixture was stopped by the addition of equimolar amounts of Na2EDTA water solution (EDTA to Mg2+) and RNA phenol–chloroform extraction, followed by RNA precipitation (3 M NaOAc pH 5.2, ethanol). The crude RNA sample was next purified using the HPLC and Phenomenex Clarity 3 μM Oligo-RP column (150 × 4.6 mm) (the method is given below) and precipitated as before. To obtain a homogeneous RNAs at 3′ end, appropriate sample underwent trimming by DNAzyme 10–23 (TGATCGGCTAGGCTAGCTACAACGAGGCTGGCCGC). The reaction mixture contains equal molar concentrations of RNA substrates and DNAzyme in 50 mM Tris (pH 8.0) and 50 mM MgCl2 for 1 h at 37 °C. The reaction was then purified using HPLC, and the desired product was precipitated with ethanol as mentioned above. This procedure allows us to obtain a 3′ homogeneous 25-nt (uncapped) or 26-nt (capped) transcripts. The quality of the obtained short RNAs was checked on 15% acrylamide/7 M urea/TBE gels with SYBR Gold (Invitrogen) staining, and the concentration was determined spectrophotometrically.

HPLC method: A: 100 mM TEAAc, B: 200 mM TEAAc/ACN; 10–30% of eluent B in 30 min.

RNA Decapping Assay

20 ng of 26-nt capped RNA transcript was incubated with an appropriate decapping enzyme (3 nM D9 or 11 nM Dcp1/2 or 200 nM D10) in an appropriate buffer with addition of 0.1% BSA at 37 °C. After being indicated, times reactions were terminated by adding equal volume of the loading dye (4.5 M urea, 50% formamide, 20 mM EDTA, 0.03% bromophenol blue, and 0.03% xylene cyanol) and flash freezing. Samples were then resolved by PAGE on denaturing 15% acrylamide/7 M urea/TBE gel and were stained with SYBR Gold (Invitrogen) and visualized using a Typhoon FLA 9500 (GE Healthcare). The band intensities corresponding to capped and uncapped RNAs were quantified densitometrically using 1D Gel Image Analysis Software TotalLab CLIQS.

Buffer used in D9 experiments: 10 mM MOPS·KOH (pH = 7.0), 100 mM KOAc, 2 mM MgCl2, 0.3 mM MnCl2, and 2 mM DTT.

Buffer used in Dcp2 experiments: 50 mM Tris·HCl pH = 8.0, 50 mM NH4Cl, 0.01% Igepal, 5 mM MgCl2, 2 mM MnCl2, and 1 mM DTT.

Buffer used in D10 experiments: 10 mM MOPS·KOH pH = 7.0, 100 mM KOAc, 2 mM MgCl2, 0.3 mM MnCl2, and 2 mM DTT.

X-ray Crystallography

The CP-100356 inhibitor was dissolved in DMSO at a concentration of 5 mg/mL. The protein–inhibitor complex was prepared by mixing D9 and CP-100356 in crystallization buffer containing 20 mM MgCl2 to a final concentration of 10 mg/mL D9 and 3 mM cap analogue substrates and incubated at room temperature for 30 min. The protein–substrate complex was mixed with well solution at a 1:1 ratio, and crystals were grown by hanging drop vapor diffusion at room temperature. D9:CP-100356 well solution contained 20% PEG 3350 and 100 mM NaF. Crystals were flash frozen in liquid nitrogen using a cryoprotectant consisting of well solution with 25% glycerol. All data sets were collected on beamline 8.3.1 at the Advanced Light Source at 100 K and wavelength 1.11583 Å using the Pilatus3 S 6 M detector and indexed, integrated, and scaled using XDS,[33] Pointless and Aimless[34] via automated beamline software ELVES.[35] Phasing was carried out by molecular replacement with the previously solved D9 structure in the post-catalytic conformation, 7SEZ.[28] The structure was then iteratively refined in PHENIX[36] and manually adjusted in COOT.[37]