Evaluation of carboxyfluorescein-labeled 7-methylguanine nucleotides as probes for studying cap-binding proteins by fluorescence anisotropy

Fluorescence anisotropy (FA) is a powerful technique for the discovery of protein inhibitors in a high-throughput manner. In this study, we sought to develop new universal FA-based assays for the evaluation of compounds targeting mRNA 5′ cap-binding proteins of therapeutic interest, including eukaryotic translation initiation factor 4E and scavenger decapping enzyme. For this purpose, a library of 19 carboxyfluorescein probes based on 7-methylguanine nucleotides was evaluated as FA probes for these proteins. Optimal probe:protein systems were further investigated in competitive binding experiments and adapted for high-throughput screening. Using a small in-house library of compounds, we verified and confirmed the accuracy of the developed FA assay to study cap-binding protein binders. The applications of the most promising probes were then extended to include evaluation of allosteric inhibitors as well as RNA ligands. From this analysis, we confirmed the utility of the method to study small molecule ligands and evaluate differently 5′ capped RNAs.


Results and discussion
Optimization of the probe and binding studies. As the initial step in the development of an FA method, different structures of fluorescent probes were explored. As a starting point for the design of the probes, we used several known cap-derived eIF4E and DcpS binders differing in structural complexity (Fig. 1). As a label, we chose carboxyfluorescein (FAM) because of its many advantages in the context of FA assays, including high quantum yield and short half-life of the excited state (~ 4 ns) 36 , which is beneficial for small molecular probes 37 . As a result, we synthesized and tested a set of carboxyfluorescein-labeled cap analogs differing in size (from mono-to trinucleotides) and the site of fluorophore attachment (Fig. 1, Fig. S1). The fluorophore was attached using different chemical strategies to either the terminal phosphate, the 2′ or 3′ hydroxyls of m 7 G or G ribose moiety, or the N6-position of adenine. Fluorescent probes for studies with DcpS were additionally modified within the triphosphate bridge to make them resistant to enzymatic hydrolysis. To this end, different phosphate modifications were explored, including a bridging modification (β-γ-O to CH 2 ) 38 , nonbridging modification (γ-O-to-S) 39 , and a recently reported phosphorothiolate modification (5′-PSL) 30 . As a reference, we included a 30-nt long capped-RNA probe that was previously used for binding studies with Drosophila melanogaster eIF4E 40 . This probe could be considered a mimic of the natural ligand of eIF4E (mRNA), wherein the probe was placed 16 nt away from the 5′ end, thereby minimizing its impact on protein binding.
All the fluorescently labeled compounds were evaluated as FA probes for eIF4E and/or DcpS proteins. The optimal probe fulfilling the requirement for unbiased K D estimation and development of competition binding assays should have high affinity for the target protein and a stable intrinsic fluorescence intensity that remains unchanged over time and upon binding to the target protein 41,42 . To select optimal probes, we performed direct binding experiments, in which each probe (at a constant concentration) was mixed with increasing concentrations of eIF4E or DcpS. We also performed negative control experiments for select probes using Bovine Serum Albumin (BSA) to confirm lack of unspecific interactions at concentrations up to 2.5 µM (Fig. S2). To check whether the emission of the ligand changed upon binding to the specific protein, values of total intensities were calculated as the sum of the parallel and double perpendicular intensities for each binding experiment 43,44 . Probes were compared based on the brightness enhancement factor g, demonstrating enhancement of total intensity between the free and bound forms of the probe. We observed that the changes in emission intensity during protein-probe complex formation strongly depended on the site of cap analog labeling and the linker length (Table 1). For both tested proteins, the greatest changes in fluorescence intensity were observed when the label was located at the ribose moiety. Compounds with the label at the 2′ position were more sensitive to environmental changes than those labeled at the 3′ position (a 1.2-fold difference for probes 3b and 3c). In contrast, cap analogs labeled at the N6 position of adenine had the most stable fluorescence signal. These dependencies changed with modifications within the phosphate bridge (e.g., the additional phosphate group in probe 1c decreased the fluorescence intensity stability in comparison with probes 1a and 1b). The unfavorable  www.nature.com/scientificreports/ effects of binding-sensitive fluorescence intensity could be successfully eliminated by changing the length of the linker (Table 1). For DcpS protein, longer linkers were associated with more stable fluorescence intensity (probe 2i was 1.4 times less sensitive than probe 2h). In the case of eIF4E, the smallest changes in fluorescence intensity were observed for the medium-length linker (the intensity change for probe 2b was 4%, whereas those for probes 2a and 2c were 10% and 9%, respectively). The stability of fluorescence intensity was also affected by FAM regioisomerism; in binding studies with DcpS, isomer 5 of FAM led to significantly greater changes in fluorescence intensity than isomer 6. To determine the dissociation constant (K D ) values, a 1:1 binding model was fitted to the obtained binding curves (Table 1, Fig. 2A). For probes characterized by Δg values greater than 0.1, the FA values were appropriately corrected before K D determination 44 . Among fluorescent probes tested against eIF4E, mononucleotide cap analogs (1a, 1b, 1c, 1e, 1f) bound the protein with significantly higher affinity than other probes. The affinity for eIF4E was the highest for mononucleotide analogs carrying a tetraphosphate chain (compound 1b with a K D that was 5.7-fold lower than that of the triphosphate probe 1a). Despite the high affinity for eIF4E, probe 1b showed the lowest FA response upon transition from the free to bound state, which affected method quality. Interestingly, further elongation of the tetraphosphate bridge to pentaphosphate did not improve the binding. The affinity of the probes containing phosphorothioate modification (1e, 1f) to eIF4E was dependent on the absolute configuration of the stereogenic P center (probe 1e bound to eIF4E 1.5 times stronger than probe 1f), consistent with previous data reported for unlabeled compounds 39 . Phosphorothioate substitution improved the binding compared with unmodified probe (1e showed a K D 1.4 times lower than 1a); however, the impact was less favorable than phosphate bridge elongation. Dinucleotide probes had generally weaker binding affinities than mononucleotide probes; the most potent dinucleotide probe 2d had a K D that was 1.3-fold higher than that of probe 1a. In contrast, trinucleotide probes had a binding affinity in the range corresponding to mononucleotide probes. This result suggested that the third nucleotide eliminated the unfavorable influence of the second nucleotide by forming new contacts between eIF4E and the additional nucleotide or by rearrangement of the cap structure inside the eIF4E binding pocket. However, the K D of oligonucleotide probe 4a was sevenfold higher than that of 3a, suggesting that further addition of nucleotides emulating the mRNA body may counteract this effect, resulting in negligible contribution of the mRNA body to the eIF4E:cap interaction. For DcpS, we evaluated the hydrolysis-resistant probes 1d, 1e, 1f, 2e, 2f, 2g, 2h, and 2i (Table 2). Owing to the presence of a stereogenic P-center, the phosphorothioate probes existed in the form of two P-diastereoisomers, designated as D1 and D2 according to their order of elution during reverse-phase high-performance liquid chromatography. The diastereoisomers varied in binding affinity towards DcpS enzyme (e.g., probe 1f had a K D that was 1.8-fold lower than that of probe 1e). Moreover, both phosphorothioate probes bound to DcpS with affinity higher than the corresponding probe with methylenobisphosphonate modification (1.3-and 2.4-times higher binding affinity compared with 1d). Unexpectedly, probe 2d carrying the 5′-phosphorothiolate moiety was found to be susceptible to DcpS-catalyzed hydrolysis under assay conditions and thus was not suitable for this assay, despite the fact that other compounds carrying this moiety have been shown to be resistant and potent inhibitors of DcpS 30 . The most promising probes for DcpS were found among dinucleotide cap analogs. The lowest K D value was obtained for cap analog 2i carrying a methylenebisphosphonate moiety and labeled at the 3′ position of ribose with a long linker. Development and validation of an FA competitive binding assay for eIF4E. After preliminary evaluation of the probes, we aimed to develop an FA-based binding assay for eIF4E. To this end, we selected three mononucleotide probes characterized by medium to high binding affinity (1a, 1b, and 1e; Fig. 2A). Using these probes, we performed probe-displacement experiments (competitive FA assays), in which an unlabeled Table 1. Binding affinities of mono-, di-, tri-, and oligonucleotide probes for eIF4E together with fluorescence enhancement factors (g). FA experiments were performed in black 96-well plates using a Biotek Synergy H1 plate reader. Each well (200 µL) contained a fluorophore-tagged probe (0.5, 1, 2, or 10 nM) and increasing concentrations of the desired protein (from 0 to 2.5 µM).  45 . The determined EC 50 and Hill slope values are shown in Table 3.
The results indicated that for low-and moderate-affinity ligands (half maximal inhibitory concentration [IC 50 ] ≥ 200 nM), the Hill slope was close to 1, which was expected for a 1:1 binding model. However, the Hill slope was higher than 1 for high-affinity ligands, indicating that the probe affinity was too low to properly evaluate these ligands. As expected, the steepness of the curves was lowest for probe 1b, which had the highest affinity for eIF4E. Hence, the results indicated that the high-affinity probe 1b could be used to accurately measure the binding affinity of highly potent compounds, as also confirmed by the best correlation with the experimental data obtained from direct binding experiments using tsFQT (Fig. 2B). Table 2. Binding affinities of mono-and dinucleotide probes for DcpS together with fluorescence enhancement factors (g). FA experiments were performed in black 96-well plates using a Biotek Synergy H1 plate reader. Each well (200 µL) contained a carboxyfluorescein-tagged probe (0.5, 1, 2, or 10 nM) and increasing concentrations of the desired protein (from 0 to 2.5 µM).  www.nature.com/scientificreports/ Next, we evaluated whether FA assays based on probes 1a, 1b, or 1e could be adopted for high-throughput screening. We first determined the assay quality based on Z' factor estimation for all three systems, i.e., 1a:eIF4E, 1b:eIF4E, and 1e:eIF4E (Fig. 3A). Probe-protein complex was used as a negative control sample (high FA), and a mixture of probe, eIF4E, and m 7 GTP (excess) was used as a positive control (low FA). The determined Z' factors were 0.74 for the 1a:eIF4E system and 0.78 for the 1e:eIF4E system. After 60 min, the Z' factors were    www.nature.com/scientificreports/ still higher than 0.5. Thus, systems 1a:eIF4E and 1e:eIF4E could be successfully applied in a high-throughput screening format. Unfortunately, in a similar test for 1b:eIF4E, we obtained a Z' factor less than 0.5, with poor signal separation between positive and negative controls. Therefore, this system was considered inappropriate for high-throughput screening owing to the low signal-to-noise ratio. The reduced response window in comparison to other probes could be a result of increased rotational mobility caused by the additional phosphate group. Using the 1e:eIF4E system, a small in-house library of ligands was screened against eIF4E (Fig. 3B). The library contained mainly dinucleotide cap analogs modified within a triphosphate bridge, some standard mononucleotides, and non-nucleotide ligands. The binding affinities of these ligands were evaluated in previous studies. The screening was performed under optimized conditions in the presence of each tested inhibitor (750 nM). All tested dinucleotide cap analogs effectively competed for eIF4E, regardless of modification. However, the combination of an imidophosphate group with phosphate chain elongation appeared to have the most stabilizing effect on the protein-ligand complex. This observation was consistent with the literature data, showing that m 7 GpNHpppG had the highest association constant (K AS = 112.3 ± 1.8 μM −1 ) among the tested ligands 46 . The screening also revealed the unfavorable impact of reducing the number phosphate groups on the binding (m 7 GSpppG to m 7 GSppG or m 7 GDP to m 7 GMP), consistent with literature data 30,47 . As expected, compounds without the m 7 G moiety did not bind to eIF4E. The allosteric inhibitor 4EGI-1, which binds to eIF4E at a different site than the cap, did not influence the fluorescence of the probe-protein complex under these conditions. All of the above results validated the FA method developed with probe 1e.
Testing of allosteric binding with eIF4E. eIF4E protein is a component of eukaryotic initiation translation complex 4F (eIF4F) and together with eIF4A and eIF4G proteins is required for initiation of the translation process 48 . 4EGI-1 is an inhibitor of eIF4E and eIF4G association and consequently leads to inhibition of capdependent translation 49 . Therefore, disruption of the eIF4E-eIF4G interaction is another important target for cancer therapy. For identification of small-molecule inhibitors of the eIF4E-eIF4G interaction, an FA assay has been developed previously 49 . The binding event was monitored by evaluating changes in FA resulting from the interaction of fluorescein-labeled 4G peptide with eIF4E with a K D of 25 μM. Because only the 4G-binding site was observed, the potential connection between 4G-and cap-binding was not elucidated.
Next, we tested whether probe 1a could be used to study the binding of inhibitors outside the cap-binding site, such as 4EGI-1. Although 4EGI-1 targets eIF4E at a binding site different from that of the fluorescent probes, we hypothesized that if 4EGI-1 binding evoked conformational changes in the proteins, FA readouts may be affected. Therefore, we conducted an experiment similar to the competitive test, but using increasing concentrations of 4EGI-1. Interestingly, we observed changes in the fluorescence anisotropy signal at 4EGI-1 concentrations exceeding 10 μM; the magnitude of these changes suggested that the fluorescent probe was released from the cap-binding site. The EC 50 value for this interaction was 35.3 ± 4.4 μM (Fig. 4A). One possible explanation for this observation was that 4EGI-1 binding to eIF4E may trigger structural rearrangements, leading to allosteric inhibition of both interactions, i.e., cap-eIF4E and eIF4G-eIF4E 50 . To verify this, we performed direct binding assays for probe 1a in the presence or absence of a high concentration of 4EGI-1 (100 μM; Fig. 4B). The results showed that the binding of probe 1a to eIF4E was at least sevenfold weaker in the presence of 4EGI-1. This suggested the interdependence of the 4G-and cap-binding sites and revealed that our method could also be used for the identification and analysis of allosteric inhibitors of cap-dependent translation. For the first time, we showed that 4EGI-1 destabilized the cap-eIF4E complex.
Capped oligonucleotide evaluation using FA. The biophysical aspects of cap-protein interactions are most often investigated using synthetically modified mono-and dinucleotide cap analogs. Despite many attempts to use fluorescently labeled and capped oligonucleotide probes to evaluate eIF4E binding 40 , their use is limited by synthetic complexity and consequently low availability. Therefore, we tested whether an FA-based competitive approach was suitable for evaluation of label-free capped oligonucleotides.  Fig. 5). This observation was consistent with data obtained for 3′-ARCA dinucleotide cap analogs, in which the α/β-bisphosphonate modification weakens the affinity to eIF4E by approximately 2.3-times 38 . Thus, we showed that the FA method could be successfully used to study the affinity of oligonucleotides to eIF4E protein.

Establishment and validation of an FA competitive assay for DcpS. Using similar assumptions as
the for eIF4E competition assay, we established conditions to study ligands of DcpS. For initial evaluation, we chose four high-affinity fluorescent probes, i.e., three dinucleotide probes (2g, 2 h, and 2i) and one mononucleotide probe (1f). Probes were tested with four DcpS inhibitors, i.e., m 7 GMP, m 7 GDP, m 7 GpNHppG, and RG3039, which differed in affinity to DcpS 33,46 . For each tested compound, we performed competition experiments to determine the EC 50 s of the selected probe:DcpS system (Fig. 6, Table 4). The affinities of the selected probes for DcpS increased in the following manner: 1f < 2g < 2h < 2i. For the two lower affinity systems, i.e., 1f:DcpS and 2g:DcpS, we did not observe any separation of dose-response curves for three of the four inhibitors. In those systems, only the weak m 7 GMP inhibitor was accurately characterized. Characterization of the potent inhibitors was limited by insufficient probe affinity (Hill slope: 1.5-3.5). Using the high-affinity systems 2 h:DcpS and 2i:DcpS, all curves were sufficiently separated, even for the two most potent DcpS inhibitors (RG3039 and m 7 GpNHppG). The obtained dose-response curves for the highest affinity compound, i.e., RG3039, were characterized by high Hill slope values (> 3.6 for both systems). The binding curves obtained for RG3039 did not permit determination of affinity owing to the poor representation of binding curves because the total protein concentration significantly exceeded the K D of the inhibitor. This result indicated that probes 2 h and 2i were still not optimal for quantitative studies of such potent inhibitors. Overall, we observed strong dependence of the ability to characterize potent inhibitors on the affinity of the probe (Fig. 6). Besides the limitations mentioned above, systems 2 h:DcpS and 2i:DcpS correctly assessed the inhibitory potencies of the selected inhibitors. The order of the tested compounds in terms of their binding affinities towards DcpS was consistent with data obtained using the fluoride-release (FR) fluorescent method 28 . FR assays use an artificial DcpS substrate, 7-methylguanosine 5′-fluoromonophosphate (m 7 GMPF), which is hydrolyzed by the enzyme to release fluoride. Fluoride activates the fluorogenic probe bis-(tert-butyldimethylsilylfluorescein) in a concentration-dependent manner; hence, the fluorescence signal is proportional to the enzymatic reaction progress. Using this activity-based assay, over 70 cap analogs were characterized as DcpS inhibitors, including compounds selected for FA method validation, i.e., m 7 GMP (IC 50 = 97 ± 21 μM), m 7 GDP (IC 50 = 5.2 ± 1.2 μM), m 7 GpNHppG (IC 50 = 3.2 ± 0.9 μM), and RG3039 (IC 50 = 0.048 ± 0.010 μM) 28 .
Because probe 2i showed the lowest K D value toward DcpS and was the most effective for characterization of potent inhibitors, such as RG3039, the 2i:DcpS system was chosen for high-throughput method optimization. The Z' factor was determined under conditions optimized for the competition assay. The Z' value exceeded 0.8 for incubation times up to 1 h, making the assay suitable for screening experiments (Table 5). A screening experiment was then performed using the same compound library as that used for eIF4E screening. The results highlighted the impact of the triphosphate bridge modification on the affinity for the protein. Cap analogs modified with phosphorothioate and phosphorothiolate moieties (m 7 GSpp S pG D1, m 7 GSpp S pG D2, m 7 GSpp S pSG D1, and m 7 GSpp S pSG D2) were the most potent inhibitors. The combination of these two modifications afforded compounds with properties similar to RG3039, which was previously identified as a potent DcpS inhibitor using FR assays 30 . Cap analogs containing imidophosphate and methylenebisphosphonate moieties (e.g., m 7 GpCH 2 ppG, m 7 GpCH 2 pppG, m 7 GpNHppG, and m 7 GpNHpppG) were also strong DcpS inhibitors (showing an inhibitory   www.nature.com/scientificreports/ potency similar to that of m 7 GDP) but were significantly weaker than RG3039. The FA method also enabled the identification of unstable compounds, e.g., hydrozylable ligands for which determination of affinity is problematic (as observed by FA signal changes during the experiment; Fig. 7). Among the tested ligands, m 7 Gp S ppG D1, m 7 Gpp BH3 pG D1, and m 7 Gpp BH3 pG D2 were recognized as slowly hydrolyzed DcpS substrates, which are difficult to identify using other screening methods. Despite the limitations of the FA method to characterize strong DcpS inhibitors, the screening assay was found to be suitable for the discovery and preliminary evaluation of DcpS inhibitors.

Conclusions
FA is a powerful technique that is widely used to study protein-ligand interactions. In this study, we used FA to develop new methods for searching small-molecule inhibitors of cap-binding proteins. In the first step, we characterized a set of fluorescent probes. As probes, we used fluorescently labeled m 7 G nucleotide analogs resembling natural substrates or ligands interacting with test proteins. We verified the influence of the bridging modification and cap-fluorophore linker length on affinity towards eIF4E and DcpS and the tested fluorescence sensitivity to binding. Based on these studies, we selected the most promising probe candidates for competitive studies and ligand characterization. Selected probe:protein systems were used to determine EC 50 and Hill slope parameters for known ligands of eIF4E and DcpS. The obtained values correlated well with literature data. Probes characterized by high affinity to the target and good FA responses were adapted to high-throughput screening assays. As a result of this analysis, we developed FA methods for both eIF4E and DcpS. The methods could be successfully used for ligand screening purposes and EC 50 parameter determination. eIF4E ligands have been extensively studied owing to the involvement of eIF4E in tumorigenesis and its role as a therapeutic target in many cancers. New ligands of eIF4E could facilitate the identification of novel anticancer agents. Notably, we found that the FA method could be used to study allosteric eIF4E ligands, such as 4EGI-1. Furthermore, high affinity of the 5′ cap for eIF4E is also crucial for the design of efficiently translated therapeutic mRNAs 51 . We showed that the FA method developed in this study was suitable for evaluation of small molecules as well as capped RNAs. This approach is a novel method that could be applied for the implementation of high-throughput approaches in therapeutic mRNA optimization and quality control. Screening of potential DcpS inhibitors is a new field of research, and few inhibitor families have been identified. DcpS plays general roles in the control of gene expression and has been independently linked to SMA, intellectual disability, and AML. Thus, identification of novel DcpS inhibitors could facilitate further studies of the connections between inhibitory and therapeutic effects because the mechanisms of action are still unknown. Fluorescent probes 1a-1f, 2a-2i and 3a-3c were synthesized chemically using methods based on phosphorimidazolide chemistry. The fluorescent labelling with fluorescein was carried out either by copper-catalyzed azide-alkyne cycloaddition or amide bond formation by N-hydroxysuccinimide chemistry. Further details on the chemical synthesis are included in the Supporting Information S1. Probe 4a was purchased from TriLink Biotechnology. eIF4E and DcpS expression and purification. Murine eIF4E (residues 28-217) was expressed in E. coli and purified as described previously 26 . Briefly, high level expression of eIF4E obtained at conditions of 0.5 mM isopropyl-β-D-thiogalactoside (IPTG) at 37 °C induced the formation of inclusion bodies. Inclusion bodies containing eIF4E were solubilized in 50 mM HEPES/KOH (pH 7.2) buffer containing 10% glycerol, 6 M guanidine hydrochloride and 2 mM DTT. Protein was then refolded during a two-step dialysis against buffer with decreas-  Expression of recombinant His-tagged human DcpS was performed in BL21(DE3) RIL strain and induced overnight at 18 °C using 0.5 mM IPTG as described previously 26 . Cells were harvested, resuspended in buffer (50 mM Tris pH 7.6, 500 mM NaCl, 20 mM imidazole) with lysozyme (0.1 mg/ml) and protease inhibitors (1 mM PMSF, 1 µM pepstatin A, 0.3 µM aprotinin) and then lysed using sonication. Lysate was clarified by centrifugation at 35 000 × g for 40 min at 4 °C. The cell supernatant was passed over a 5 mL HisTrap HP (GE Healthcare) affinity column and Ni-NTA-bound proteins were eluted using 50 mM Tris buffer pH 7.6 containing 500 mM NaCl and 400 mM imidazole. The enzyme hDcpS was purified to homogeneity on HiLoad 16/600 Superdex 200 pg (GE Healthcare) gel filtration column using 50 mM Tris·HCl pH 7.6, 200 mM NaCl, 2 mM DTT buffer. Protein was aliquoted and stored in the presence of 10% glycerol at -80 °C.

Synthesis of fluorescent probes.
Preparation of differently capped oligonucleotide ligands. Short RNAs were prepared as described previously 30  To generate uncapped RNA in reaction mixture cap analog was omitted whereas concentration of GTP was increased to 0.5 mM. The crude RNAs were purified using RNA Clean & Concentrator-25 (Zymo Research). Quality of transcripts was checked on 15% acrylamide/7 M urea gels, whereas concentration was determined spectrophotometrically. To remove in vitro transcription by-products of unintended size RNA samples were gelpurified using PAA elution buffer (0.3 M sodium acetate, 1 mM EDTA, 0.05% Triton X-100), precipitated with isopropanol and dissolved in water. In the direct binding experiments aimed at determining K d values for protein-probe complexes, the fluorescent probe at a constant concentration was mixed with an increasing concentration of the protein (0-2.5 μM). Concentration of fluorescent probes in binding experiment are provided in Table 6. Before FA measurements, the plates containing analyzed samples were incubated for 10 min at 25 °C with simultaneous shaking at 300 rpm, then the protein was added to each well, samples were incubated for additional 3 min. The FA readouts were performed in a microplate reader at 25 °C. FA signals were recorded for 20 min with 60 s interval.
The FA values for each timepoint were calculated according to the following equation: where I is the parallel emission intensity, P is the perpendicular emission intensity, and G is the grating factor. The value of the G factor was equal 0.994. For each sample, the final FA value taken to K D determination was the mean FA value from all datapoints determined for timepoints between 10 and 20 min.
To determine the dissociation constants, FA values were plotted as a function of protein concentration and the binding curves were fitted using the following equation: where FA is the determined fluorescence anisotropy, A F is the fluorescence anisotropy of free probe, A B is the fluorescence anisotropy of probe-protein complex, L T is the total ligand concentration.
The following equation was used to calculate total fluorescence intensity of the probe: where I T -total fluorescence intensity, I -the parallel emission intensity, I ⊥ -the perpendicular emission intensity.
(1) www.nature.com/scientificreports/ The calculated values of the total fluorescence intensity were plotted against the protein concentration and the curve described by the Eq. 2 was fitted. Enhancement factor g was determined from the following equation: If the change in total fluorescence intensity due to binding was greater than 10%, the correction for the calculation of the bound fraction of probe was applied 44 : where A-measured fluorescence anisotropy, A free -the fluorescence anisotropy of free probe, A bound -the fluorescence anisotropy of probe-protein complex, g-enhancement factor. FA competition assay. For competitive binding assay, a mixture containing the probe and the protein was incubated with the tested ligand. The exact concentrations of eIF4E or DcpS used for evaluation of particular probes are summarized in Table 7.
In competitive measurements, a constant concentration of the protein and fluorescent probe and increasing concentration of the test ligand were used. At least 12 point dilutions of the tested compound were used. The experiments were carried out in 96-well plates. Each sample contained a mixture consisting of the probe, tested ligand and buffer (same as for the direct binding assay). The samples were incubated for 10 min at 25 °C with simultaneous shaking, then the protein was added to each well, incubated for additional 3 min, followed by the measurement of fluorescence anisotropy. FA signals were recorded for 20 min with 2 min intervals. For each sample, the final FA value taken to EC 50 determination was the mean FA value from all datapoints determined for timepoints between 10 and 20 min.
The EC 50 value, i.e., the ligand concentration required for 50% displacement of the probe from the complex with protein were calculated according to the equation: where FA is the measured fluorescence anisotropy, Top and Bottom are asymptotes, L is the ligand concentration, HillSlope is the steepness of the curve.   where SD n and SD p are the standard deviations, and μ n and μ p represent the means of the FA values obtained from the negative and positive controls, respectively. For the screening experiments a small in-house library containing 21 ligands was used. Experiments were conducted in the same manner as Z' factor determination, with exception that the ligand concentration was modified (eIF4E screening conditions: 1 nM 1e, 100 nM eIF4E, 750 nM ligand, DcpS screening conditions: