Pilot-Scale Screening of Clinically Approved Drugs to Identify Uridine Insertion/Deletion RNA Editing Inhibitors in Trypanosoma brucei

Abstract

RNA editing pathway is a validated target in kinetoplastid parasites (Trypanosoma brucei, Trypanosoma cruzi, and Leishmania spp.) that cause severe diseases in humans and livestock. An essential large protein complex, the editosome, mediates uridine insertion and deletion in RNA editing through a stepwise process. This study details the discovery of editosome inhibitors by screening a library of widely used human drugs using our previously developed in vitro biochemical Ribozyme Insertion Deletion Editing (RIDE) assay. Subsequent studies on the mode of action of the identified hits and hit expansion efforts unveiled compounds that interfere with RNA-editosome interactions and novel ligase inhibitors with IC₅₀ values in the low micromolar range. Docking studies on the ligase demonstrated similar binding characteristics for ATP and our novel epigallocatechin gallate inhibitor. The inhibitors demonstrated potent trypanocidal activity and are promising candidates for drug repurposing due to their lack of cytotoxic effects. Further studies are necessary to validate these targets using more definitive gene-editing techniques and to enhance the safety profile.

Graphical Abstract

African and American trypanosomiasis and various forms of leishmaniasis are caused by Trypanosoma brucei, Trypanosoma cruzi, and Leishmania species. They belong to the order Kinetoplastea due to their distinctive kinetoplast, a DNA-containing structure within their single, large mitochondrion.^1,2 Currently, there are no vaccines and few highly effective drugs without side effects available for these diseases, underscoring the pressing need for new drug development efforts.^3–6

We employed a drug-centric drug repositioning strategy,⁷ to identify novel treatments, linking a known drug to a new target with its associated indication, specifically antiparasitic activity. Among several targets validated as unique and conserved among these kinetoplastids,⁸ we focus on mitochondrial uridine insertion/deletion editing. This target is particularly compelling for drug discovery due to its essential role and extensive drug-binding landscape.⁹ These microorganisms possess a cryptic mitochondrial gene expression system. The genes transcribed from the mitochondrial genome necessitate decryption and editing to acquire functionality. This process, termed mitochondrial uridine insertion/deletion editing, is executed by a large protein complex known as the editosome. Comprising over 70 distinct proteins, the holoenzyme editosome has been described as an RNA editing catalytic complex (RECC), RNA-editing substrate-binding complex (RESC), and RNA-editing helicase 2 complex.¹⁰ The RECC catalyzes the enzymatic activities and has been targeted in various screens^11–15 as most of the catalytic core components are essential for the survival of the parasites.⁹ Editing is directed by untranslatable short RNAs (30–60 nt) known as the guide RNAs (gRNAs), mostly transcribed from mitochondria minicircles. Initially, a hybrid of gRNA and pre-edited mRNA forms through the complementarity between the 5′ anchor region of the gRNA and the pre-edited mRNA This results in the protrusion of single-stranded uridines within the mRNA (deletion sites) or purine nucleotides within the gRNA (insertion sites). An endonuclease in the complex then recognizes and cleaves at the first unpaired nucleotide adjacent to the 5′ anchor duplex. The specific isoform of RECC containing KREN1 + KREPB8 + KREX1 cleaves U-deletion sites, whereas the isoforms with KREN2 + KREPB7 or KREN3 + KREPB6 cleaves U-insertion sites.^16–20 Depending on the type of editing, either the exonuclease KREX2 removes uridine residues or the terminal uridylyl transferase KRET2 adds them, guided by sequence complementarity with the gRNA. Finally, the ligase (KREL1 for deletion or KREL2 for insertion) joins the two RNA fragments to produce the final edited mRNA. Although the two ligases are spatially separated in two different subcomplexes and perform distinct functions, only KREL1 is shown to be essential for the parasite’s viability, and it can compensate for the lack of KREL2.^21–24

Different methods have been employed to discover new inhibitors of this pathway,^{11–15,25–28} including both virtual and in vitro screens. Our laboratory has developed a FRET-based full-round deletion RNA editing assay,²⁹ which has been utilized in both pilot-scale¹³ and high-throughput screening (HTS)¹² to discover new inhibitors of this pathway. Further studies are underway to evaluate the hits from these screenings. We enhanced our FRET-based assay, now termed the Ribozyme Insertion Deletion Editing (RIDE) assay, by modifying it to bypass the rate-limiting step of endonuclease activity. This assay simultaneously monitors both insertion and deletion editing within the same reaction, as shown in Figure 1.³⁰ The RIDE assay uses ribozymes that bind and cleave a specific FRET substrate. Postediting, these ribozymes cleave their specific substrates to release a detectable signal. This assay is adaptable for high-throughput applications and requires significantly lower ATP concentrations—10,000 times less than the full-round assay—thereby facilitating the discovery of competitive inhibitors.

Figure 1.

Schematic representation of the RIDE assay. This assay integrates both uridine insertion and deletion RNA editing processes (postendonucleolytic activity) into a single multiplex reaction. The pre-edited RNA sequence, guided by the gRNA, undergoes editing by the editosome, activating two hammerhead ribozymes (HHR1 for insertion and HHR2 for deletion) from a catalytically inactive state. These activated ribozymes are then capable of cleaving their respective FRET substrates. Fluorescent signals emitted upon cleavage of these FRET substrates indicate the quantity of fully edited RNA molecules present.

In our search for potential drug candidates that can suppress RNA editing, we conducted an in vitro biochemical screen of the National Institutes of Health Clinical Collections (NCC) library. This library comprises 707 small molecules, including FDA-approved drugs and late-stage development candidates, which have been tested against various targets and disease agents previously.^31–40 Using our refined assay, we identified five compounds inhibiting in vitro RNA editing and four exhibiting antiparasitic activity. Further investigations and hit expansion efforts revealed that these compounds interfere with multiple catalytic steps in our in vitro mode of action assays and include novel ligase inhibitors with IC₅₀ values in the low micromolar range, with two compounds disrupting RNA-protein interactions. Significantly, these compounds have shown potent antiparasitic effects at micromolar concentrations, with minimal cytotoxicity to host cells. Our findings suggest new potential targets for these compounds. By leveraging existing preclinical and/or clinical data on these molecules, we can optimize these hits into lead compounds, thereby advancing the development of effective antikinetoplastid drugs. Future studies must confirm their target specificity and rigorously enhance their safety profiles.

RESULTS

Screening for RNA Editing Inhibitors Using the RIDE Assay.

To discover new RNA editing inhibitors, we utilized the RIDE assay to screen the NIH Clinical Collection³⁰ (Figure 1). This HTS method includes both insertion and deletion editing in a single reaction, omitting endonuclease activity. We employed a KREL1-tagged tandem affinity purified editosome from T. brucei as the protein source and conducted the screening at 20 μM concentration. Compounds reducing RNA editing activity by more than 50% compared to the control were considered hits, resulting in an acceptable screen performance with a signal-to-background ratio greater than 3 and a mean Z′ factor of 0.8 ± 0.1.

We identified 11 primary hits from the library. These hits represent a 1.5% hit rate, with ten exhibiting reproducible inhibitory effects (Figure 2)—a 0.14% false-positive discovery rate after confirmation using the multiplex assay and individual precleaved assays. The hits included four drugs from the anthracycline family, three from the flavonoids, and three singleton compounds.

Figure 2.

Screening cascade and chemical structures of initial hits. The pilot-scale screening identified ten primary hits two compound groups (flavonoids and anthracyclines) and three distinct compounds (mitoxantrone, cefixime, and natamycin). After excluding nonspecific hits, five final hits remained, including selected flavonoids, cefixime, and natamycin. These final hits are marked with an asterisk.

Counter Screening for Specificity (Interference Assay).

A counter screen was conducted to identify compounds that hindered ribozyme activity independent of the editosome to ascertain the specificity of the inhibition observed in the initial screening. This validation process involved assessing the impact of potential inhibitors on edited hammerhead ribozymes in the absence of the editosome. Among the ten hits, five inhibitors were found to nonspecifically impede ribozyme activity at 40 μM concentration (Figure 3). Notably, members of the anthracyclines family (daunorubicin, idarubicin, doxorubicin, epirubicin) and Mitoxantrone exhibited this nonspecific inhibition. Additionally, the activity of these compounds was assessed by monitoring ribozyme cleavage activity on a gel, which revealed distinct effects—some impeded FRET cleavage (doxorubicin, epirubicin, and mitoxantrone), while others interfered with fluorescence detection (daunorubicin and idarubicin). This approach enabled the characterization of each compound’s nonspecific mechanism of action in the RIDE assay and led to the elimination of promiscuous hits from the initial hits.

Figure 3.

Counter-screening initial hits for elimination of false positives. The ten primary hits were subjected to counter-screening against the active ribozyme to determine potential interference with (A) fluorescence in the FRET-based assay and (B) cleavage of the FRET substrate in a gel-based format. Controls without the ribozyme and without drugs were included to establish baseline activity. Suramin served as an additional negative control within the assay. Mean values and standard deviations were calculated from the FRET-based assay data and are shown for two independent replicates.

RNA Editing Inhibitors’ Efficacy in RIDE Assay and Their Effect on Different Parasite’s Viability.

Dose–Response Analysis and Comparative Efficacy.

Each confirmed hit was tested in a dose–response manner in the multiplex assay, and their IC_50^s were determined using a three-parameter nonlinear regression model in GraphPad (Figure 4). Among flavonoids, epigallocatechin gallate (EGCG) showed its inhibitory effects even at 100 nM, whereas isoquercitrin and hyperoside were less potent, with IC₅₀ values in the low to moderate micromolar range. Cefixime and natamycin impaired the FRET assay in the moderate micromolar range, with their IC_50^s ranging from 6 to 55 μM. Cefixime, isoquercitrin, and EGCG demonstrated effects on both insertion editing and deletion editing within a similar concentration range. This suggests that they may target proteins involved in regulating both types of editing. Conversely, natamycin and hyperoside exhibited stronger activity against insertion and deletion, respectively. This indicates that the proteins they target may influence both types of editing, but with a greater impact on one over the other.

Figure 4.

Dose–response curves for confirmed RNA editing inhibitors. Each confirmed hit was tested in the RIDE assay under a range of concentrations to establish dose–response profiles. The IC₅₀ for each inhibitor was determined for both insertion and deletion editing. The IC₅₀ values are presented in μM with a 95% confidence interval in parentheses. Mean values and standard deviations are depicted for each data set based on a minimum of two independent experiments.

Subsequently, we tested the hits against T. brucei using the Alamar blue assay, and EC₅₀ values against other kinetoplastids, such as T. cruzi and leishmania species, were added to Table 1 (data sourced from the literature). The results from other sources were consistent with our finding for T. brucei. Cefixime showed no effects on T. brucei viability or growth in our experiments, and no data was available for its impact on other kinetoplastids. Flavonoids showed varying antitrypanosomatid effects in the moderate micromolar range, with EGCG being the most potent among them, displaying EC₅₀ of 20, 41 and 200 μM against T. brucei, leishmania, and T. cruzi, respectively. Hyperoside showed mediocre efficacy against all three parasite species, with EC₅₀ values ranging from 50 to 65 μM.

Table 1.

Antitrypanosomal Activity of Confirmed Hits and Their Effect on the Editing In Vitro^f

drug name	EC50				IC50
	T. brucei	T. cruzi	leishmania	human cells	insertion editing	deletion editing
cefixime	NI	ND	ND	>200 μM (MT-4)33a	55.45 μM [35.81–88.49]	40.37 μM [22.49–76.3]
EGCG	20 μMb	200 μM41b	41 μM41b	246 μM (HGF-1)42a	0.074 μM [0.03761–0.1197]	0.422 μM [0.1821–0.9522]
isoquercitrin (quercitrin)	>20 μMb (62 μM)41b	ND (67 μM)41b	ND (39.5 μM)41b	500 μM (CHO)43c [3.3 mM (Vero E6)]44d	21.81 μM [12.35–41.17]	16.54 μM [6.821–46.85]
natamycin	12 μMb	ND	15 μM45e	100 μM (macrophage)45e	6.42 μM [4.41–9.47]	26.57 μM [15.27–48.75]
hyperoside	50 μMb	64.7 μM41b	64.7 μM41b	194 μM (L6)41b	61.37 μM [30.45–161.4]	20.93 μM [13.87–32.27]

^aValues are averages of at least three separate experiments.

^bMean values from at least two replicates (variation is a maximum of 20%).

^cIndependent experiment with a minimum of three replicates for each end point.

^dThe number of replicates was not indicated.

^eConducted at least three times in triplicates.

^fThis table presents the EC₅₀ values for each compound against T. brucei parasites, as measured using the Alamar blue assay. Suramin (one of the current drugs for human African trypanosomiasis) and no-drug served as the positive and negative controls in our viability assay, respectively. Additional data from the literature are included where available (see references). Information on the effects of these compounds on human cells is also provided. ND and NI abbreviations indicate “not determined” and “no inhibition” within the tested range, respectively. IC₅₀ values against RNA editing assays were added from Figure 4 to compare with their effect on cell viability. The number of replicates for each data is mentioned below.

Isoquercitrin had no observable effects on T. brucei viability at the tested concentrations. However, data for its analog, quercitrin, showed it impacting parasite viability within the 39.5–67 μM range. We predict, if present, the isoquercitrin EC₅₀ for T. brucei will be more than 20 μM. No data was found for isoquercitrin against T. cruzi and leishmania.

Apart from EGCG, which had a very low IC₅₀ in the RIDE assay, and cefixime, which showed no effect on the viability of T. brucei, the other compounds (natamycin, hyperoside, and isoquercitrin) affected both editing and the viability of the parasites in the same concentration ranges, suggesting the editing as one of their probable targets which warrants further investigation.

We also compiled data on the effects of hits on different human cells and included this in Table 1 to provide insight into their selectivity toward the parasites. Given that these compounds are selected from the NIH clinical collection, they were expected to show low or no effects on mammalian cells. Their selectivity and safety for human use make these hits promising candidates for repurposing in treating kinetoplastid diseases. Further studies will be needed to enhance the selectivity index for these compounds.

Mode of Action Analysis for Confirmed Hits.

Each confirmed hit (EGCG, isoquercitrin, natamycin, cefixime, and hyperoside) was tested in the in vitro mode of action assays to evaluate its effect on individual catalytic steps of uridine insertion/deletion RNA editing and its impact on RNA-protein interaction (Figure 5). We previously developed fluorescent gel-based assays to monitor each catalytic step—endonuclease, exoUase, TUTase, and ligase.¹² Additionally, electrophoretic mobility shift assay (EMSA) was utilized to determine if a hit impairs the editosome’s ability to bind the RNA substrate. Suramin served as the positive control, known to inhibit all enzymatic steps and interfere with RNA protein interactions, which is presumed to be its mode of action against the editosome.^11,12,46

Figure 5.

Characterization of the hits’ mode of action. Each compound was assessed for its effect on the editosome’s individual catalytic steps of RNA editing. These include (A) RNA ligation by ligase, (B) uridine removal by exoUase, and (C) uridine addition by TUTase. (D) Compounds were also evaluated for their ability to disrupt RNA-protein interactions using EMSA. (E) The endonuclease activity assay served as a control to confirm specificity, as this assay is not encompassed by the RIDE assay. RNA substrates and corresponding fluorophores for each assay are depicted, and diagrams representing the RNA hybrids highlight the targeted enzymatic activities. In these assays, suramin served as a positive control (indicative of 100% inhibition), while DMSO were the negative control (indicative of no inhibition).

All flavonoids (EGCG, isoquercitrin, and hyperoside) inhibited the ligase, exoUase, and TUTase activities but did not affect the endonuclease or RNA-protein interactions (Figure 5). This lack of effect might be due to the higher protein concentration present in these two assays.

Cefixime inhibited ligase, endonuclease, and TUTase but did not affect exoUase activity (Figure 5). It also hindered RNA-protein interaction in the gel shift assay, suggesting it may interact with an RNA-binding protein involved in all the assays except for exoUase activity.

Natamycin inhibited all the enzymatic activities (ligase, TUTase, exoUase, and endonuclease). It moderately affected the gel shift assay, indicating its mode of inhibition likely involves preventing the protein from binding to the substrate.

Exploring Target Specificity in RNA Ligase Inhibition.

Each activity assay is designed to be independent of the other steps, meaning that only the activity of a specific enzyme (e.g., ligase) is required for the assay. If a compound inhibits multiple enzymatic activities, it likely targets a common element rather than inhibiting different enzymes separately. To ascertain whether the hits that hindered the ligation step of RNA editing, facilitated by the editosome, could also impact the recombinant enzyme governing the same process, we examined these compounds against recombinant RNA editing ligase (Figure 6). This approach helped determine if the compounds could bind directly to the ligase while disrupting the activity of the editosome through interference with RNA-protein interactions, particularly binding to mRNA-binding proteins such as Mitochondrial RNA binding Proteins. Mordant Black 25 (MrB), a previously identified ligase inhibitor, served as the positive control.¹³

Figure 6.

Evaluating the efficacy of hits on ligation activity. (A) The inhibitory effects of all identified hits were assessed in our FRET-based ligation assay using two protein sources recombinant ligase (right) and the editosome complex (left), with assays conducted at two different ATP concentrations (100 nM and 100 μM). This analysis showed differential inhibitory patterns among the compounds with varying ATP concentrations. (B) The IC₅₀s were calculated for the compounds against recombinant ligase (rKREL1) and rKREL1/KREPA2 complex. Data represent means and standard deviations, based on duplicate experimental replicates.

The results show that all hits except hyperoside inhibited the recombinant RNA editing ligase 1 at a low ATP concentration (100 nM) (Figure 6A), indicating that cefixime, EGCG, natamycin, and isoquercitrin bind to the ligase and hinder the ligation process in vitro. This discovery opens up new possibilities for creating targeted inhibitors.

In a preliminary analysis, we tested the drugs at 40 μM with two ATP concentrations (100 nM and 100 μM) in the ligation assay to test the effect of high ATP concentrations on the compounds’ ability to inhibit editing (Figure 6A). Remarkably, the inhibition by hyperoside decreased from approximately 77 to 35% when tested against the editosome. The increase in ATP concentration also impacted EGCG and isoquercitrin to some degree, reducing their inhibitory effects on the editosome from 88 and 70 to 50 and 37%, respectively. However, it only slightly affected isoquercitrin’s inhibition of the recombinant protein, reducing it from 81 to 68%, and had no discernible effect on EGCG.

Natamycin’s efficacy against rKREL1 decreased slightly from 85 to 66%, and its mediocre inhibitory effect remained largely unaffected against the editosome (47 to 36%). Cefixime fully inhibited the ligation assay with both the editosome and the rKREL1 at both ATP concentrations. To evaluate the efficacy of the hits against the ligase, we determined the IC₅₀ of cefixime, EGCG, isoquercitrin, and natamycin in the ligation assay using either rKREL1 or the rKREL1-KREPA2 complex (Figure 6B). Utilizing KREPA2 would provide insight into whether the presence of an interacting partner affects the inhibitors’ efficacy. KREPA2 enhances ligation activity compared to recombinant REL1 alone,⁴⁷ potentially impacting the inhibitors’ efficacy. Except for EGCG, we observed less inhibition when using rKREL1/KREPA2 as the protein source. Hyperoside did not affect either rKREL1 or rKREL1/KREPA2, indicating that its target lies elsewhere in the editosome.

Cefixime, isoquercitrin, and natamycin showed significantly reduced inhibition against rKREL1/KREPA2 compared to rKREL1, possibly due to the enhanced ligation activity of rKREL1/KREPA2 or conformational changes in KREL1 induced by KREPA2. The IC_50^s of Natamycin and isoquercitrin were even higher than their IC_50^s in the RIDE assay, suggesting an alternative mode of action within the editosome context. Cefixime’s IC₅₀ was ten times higher against rKREL1/KREPA2 than rKREL1, yet it exerts its effect in the same range of concentrations observed in the RIDE assay mediated by the editosome.

This study showed evidence that the identified compounds inhibit RNA ligase activity and potentially disrupt RNA-editosome interactions in vitro. However, it is crucial to acknowledge the limitations of our assay methods, which are primarily biochemical and do not provide direct proof of mechanism within a cellular context. The observed effects might also be mediated through mechanisms not explored in this study, including potential off-target interactions. Future studies employing knock-in/knockout techniques and in vivo models will be essential to confirm these findings and elucidate the precise mechanisms by which these compounds exert their antiparasitic effects.

Molecular Docking Studies.

Molecular docking studies identified the binding location and interaction mode of EGCG with KREL1 through blind docking using AutoDock Vina (Figure 7). EGCG was found to bind within the ATP binding pocket of the full-length KREL1, which includes both the catalytic N-terminal and C-terminal domains involved in RNA recognition and ligase autoadenylylation (Figure 7A–C).⁴⁷ EGCG exhibited a binding affinity of −9 kcal/mol, compared to −8.6 kcal/mol for docked ATP. Figure 7D illustrates the superimposition of EGCG and ATP, showing their similar binding characteristics. Both molecules bind in analogous poses; EGCG’s benzopyran group and ATP’s adenine base are located deepest within the ATP-binding cleft, while the remainder of their structures extend outward from the binding pocket. According to predictions, EGCG’s benzopyran is expected to occupy the positions analogous to ATP’s adenine and ribose sugar (Figure 7D), illustrating mimicry in their binding configuration. Interactions include Phe209, Val286, and Lys87 engaging with the benzopyran of EGCG and the adenine base of ATP, respectively. Additionally, the hydroxyl group attached to carbon 7 (C7) of EGCG’s benzopyran forms a hydrogen bond with Glu86, and ATP’s adenine base hydrogen bonds with Tyr58. ATP forms a pi–σ bond with Ile61, while EGCG’s galloyl group hydrogen bonds with the same residue (Figure 7C,E). EGCG also forms a pi-anion interaction with the highly conserved residue Lys307, crucial for the first step of adenylylation reaction,⁴⁸ whereas ATP is hydrogen bonded to Lys307 through its α-phosphate group. EGCG also forms two hydrogen bonds with catalytic Lys87, involving the oxygen atoms from EGCG’s ring C and the galloyl group attachment. However, ATP engages Lys87 only via a pi-alkyl bond. EGCG Rings B and D (the galloyl group) efficiently fill the space occupied by ATP’s phosphate chain (Figure 7D), binding tightly within the ligase. A hydroxyl group on EGCG’s Ring B forms a hydrogen bond with the conserved His89 (crucial for strand joining^48–50). Additionally, EGCG’s Ring B forms a pi-anion interaction with Glu159, a motif III conserved residue, whereas Glu159 binds the ATP ribose sugar and phosphate groups of ATP via hydrogen and charge interactions. Amino acids that interact with both EGCG and ATP in the full-length protein are outlined in black (Figure 7C).

Figure 7.

Molecular docking studies of EGCG against the full-length AlphaFold model of KREL1. (A) Structure of KREL1 and its two domains (B) 3D view of the EGCG docked in the ATP binding pocket (C) 2D ligand interaction diagram of EGCG in the binding site. Amino acids that were shown to interact with the docked ATP are outlined in black (D) EGCG superimposed on the ATP (both docked against the full-length KREL1) (E) 2D ligand interaction diagram of ATP docked against the full-length KREL1.

We also observed an unfavorable donor–donor interaction between Arg309 and an oxygen atom on EGCG’s galloyl group, suggesting a potential for induced fit effect due to the protein’s static nature in our simulation. This observation supports the hypothesis that EGCG’s inhibitory effect on KREL1 at low micromolar concentration may be enhanced by protein conformational adjustments. While ATP forms more interactions within the pocket, EGCG’s similar docking pose and critical residue engagement highlight its potential as a basis for developing potent KREL1 inhibitors.

Expanding the Ligase Inhibitor Pool.

To expand the repertoire of ligase inhibitors, we acquired an additional set of 36 compounds, which includes flavonoids, several suramin analogs (previously known as KREL1 inhibitors), and nucleotide analogs. We screened these compounds at a concentration of 50 μM against rKREL1, the rKREL1/KREPA2 complex, and the REL1-TAP tagged purified editosome for ligation activity (Figure 8A). We selected this concentration because the IC₅₀ values of flavonoids (excluding EGCG) fell within the moderate micromolar range. Screening compounds with closely related structures at 50 μM ensures we capture all potential inhibitors. If a compound fails to exhibit any inhibitory effect at this concentration, it confirms its lack of activity and is deemed unworthy of further pursuit. ABMA, an ATP analog, also demonstrated a 50% inhibitory effect within a similar range, justifying the same rationale for nucleotide analogs.³⁰

Figure 8.

Expansion of the hit compounds. (A) A selected set of 36 available analogs, including flavonoids, suramin (a previously identified editosome inhibitor^11,12,14), and nucleotide analogs were tested at 50 μM in a ligation assay. Three protein sources were used recombinant ligase rKREL1, rKREL1/KREPA2 complex, and the editosome. Compounds achieving greater than 80 percent inhibition were considered significant, and their chemical structures are depicted. (B) To confirm the specificity of these compounds for ligase activity within the editosome, their effect on the editosome’s endonuclease activity were also evaluated. The color coding in the illustrations—red for inhibition and green for no inhibition—indicates the outcome of these assessments.

Among this set, we found nine compounds that inhibit ligation activity, targeting either all three protein sources [rKREL1, rKREL1/KREPA2 and the editosome], two protein sources [rKREL and editosome], or exclusively the editosome. The high hit rate can be attributed to the selection of these compounds as analogs of the hits identified in this study (flavonoids) and from previous studies (suramin).^12,14 We identified two compounds (nucleotide analogs), MRS2289 and MRS2295, that solely inhibited ligation within the editosome, warranting further investigation to elucidate their mode of action. The synthesis of MRS2285 (compound 7 from⁵¹), MRS2289 (compound 17 from⁵²), and MRS2295 (compound 15 from⁵²) was previously described. Furthermore, we observed that suramin analogs impacted the activity of all protein sources like suramin, as shown previously.^11,12,14

We used an endonuclease assay as a counter-screen to determine whether a compound specifically affects the “ligation” step among the enzymatic activities of the editosome. As shown in Figure 8B, five compounds also inhibited the endonuclease activity, indicating that their target protein within the editosome influences other editing activities. MRS2884, which is a polyamidoamine (PAMAM) dendrimer conjugate,⁵³ exhibited aggregation within the well, representing its mode of action. However, two flavonoids (MRS8266 and MRS8267) did not affect endonuclease activity, suggesting their target within the editosome could be ligase or its interacting partners. Additionally, MRS8267 was inactive against the rKREL1/KREPA2, possibly due to the conformational change or the enhanced activity of the complex compared to the recombinant protein alone.

DISCUSSION

Repurposing drugs is an effective strategy for developing treatments for trypanosomatid diseases. To advance this research, we used a novel multiplex assay to screen the NCC library for inhibitors targeting the uridine insertion/deletion RNA editing, which is unique to kinetoplastids and absent in humans. We initially identified ten effective drug candidates against the editosome, the protein complex crucial for this editing process. Our promising findings include two compound series—three flavonoids, four anthracyclines, and three individual compounds.

Among the primary hits, the anthracyclines family—including daunorubicin, idarubicin, doxorubicin, epirubicin—and mitoxantrone were eliminated from further studies due to their interference with the ribozyme activity and the detection system. This aligns with previous research indicating that anthracycline derivatives are characterized by distinctive red fluorescence emission associated with their conjugate systems,^54,55 and their ability to intercalate with DNA^56–59 or form adducts with it.⁶⁰ Also, a recent study proposed a model illustrating doxorubicin docked within the major groove of the 32 bp U-helix, establishing polar contacts with G-U base pairs.⁶¹ This aligns with our findings indicating that doxorubicin inhibits the active ribozyme from cleaving the substrate, as demonstrated in the polyacrylamide gel data (Figure 3B). Mitoxantrone, a classic DNA intercalator that also forms DNA adducts, similarly inhibited the FRET cleavage of the active ribozyme.^62,63 Although these inhibitors were not pursued in this study, some could serve as promising starting points for targeting the RNA structure to inhibit editing, as proposed in another study.⁶¹

After confirming the initial hits (cefixime, natamycin, EGCG, isoquercitrin, and hyperoside), the IC₅₀ values of these compounds in the multiplex assay were determined. However, except for EGCG, which showed submicromolar activity, all other hits showed low to moderate micromolar efficacy. Moreover, their antiparasitic activity also fell within the moderate micromolar range. Considering that the in vitro data shows that the IC₅₀ of enzyme inhibition concentration is in the same range as the cell-based activity (except for EGCG), we speculate that RNA editing could potentially be the target for the compounds exhibiting antiparasitic effects, which warrants further investigation. In the case of EGCG, the enzyme inhibition occurs at a concentration lower than that required for efficacy against the parasite, suggesting either poor compound uptake by the parasite or by the mitochondrion where the editosome resides.

Additionally, we analyzed the mode of action of the identified hits against the editosome, revealing their inhibition of multiple enzymatic activities and RNA-protein interaction, as demonstrated in EMSA. This suggests that the compounds may target either the integrity of the complex or its RNA binding capability. Many previously discovered compounds showed this mode of action, including MrB, aurintricarboxylic acid (ATA), PPNDS, NF449,¹³ and several compounds from our recent HTS.¹² Even though the inhibitors target noncatalytic regions of the editosome, having these inhibitors allows us the opportunity to refine them further, potentially leading to the development of more effective lead compounds. Flavonoids, in particular, have been proposed as promising candidates for further optimization in treating kinetoplastid diseases.^41,64–67 However, it is worth noting that flavonoids are known for their promiscuous activities, yet they continue to be pursued as potential drug leads.^68,69 For instance, EGCG was identified in three other studies screening the NIH clinical collection, reporting various activities, including the inhibition of HIV-integrase, modulation of nonviral gene delivery to adipose-derived hMSCs, and inhibition of amyloid-β.^31,33,37

Given that ligation activity was targeted by the identified hits, assessing these compounds against the recombinant protein could confirm their ability to inhibit the isolated enzyme. Among the five confirmed hits, we found that hyperoside did not inhibit the recombinant enzyme to the same extent as the editosome, unlike the other hits. This underscores a critical point that inhibiting a specific enzymatic activity within a complex does not necessarily mean that the enzyme responsible for that activity is being targeted. Also, a compound that inhibits a recombinant enzyme may not necessarily target the same protein in a complex containing that enzyme. Moreover, increasing the ATP concentration by a thousand-fold (up to 2X drug concentration in the reaction) resulted in over a 30 percent reduction in the inhibition of flavonoids (EGCG, isoquercitrin, and hyperoside) (Figure 5) when assessed against the editosome, but not when evaluated against the recombinant ligase. This suggests that these flavonoids competed with ATP in the presence of the editosome. Flavonoids have been shown to inhibit kinases by competitively mimicking kinase substrates, such as triphosphate (ATP and GTP).^70–72

Cefixime, another confirmed hit, maintained its inhibitory effect against both the editosome and the recombinant ligase even with increased ATP concentrations. As a cephalosporin antibiotic, cefixime can establish covalent bonds through its beta-lactam group with its intended targets.^73,74 However, since it did not affect the viability of T. brucei, it likely could not reach its target inside the mitochondria. Further studies focusing on formulation or modifications to functional groups are required to enhance its cell permeability.

Natamycin has been suggested to induce mitochondrial membrane depolarization, elevate intracellular Ca²⁺ levels, and cause significant changes to the plasma membrane in Leishmania cells. Consequently, these disruptions lead to depletion of cellular ATP levels and the generation of reactive oxygen species, ultimately culminating in apoptosis-like and necrotic cell death.⁴⁵ We propose that natamycin may also influence RNA editing within the mitochondria, where the edited products (such as those involved in the electron transport chain) are crucial for ATP generation. However, further investigation is needed to validate the editosome as the in vivo target of natamycin.

We also aimed to increase the pool of ligase inhibitors available for future studies. Therefore, we screened a small group of compounds, including flavonoids, suramin analogs, and nucleotide analogs. Our rationale was based on three initial hits that were flavonoids that inhibited ligation. Additionally, another study previously identified myricetin as an inhibitor of rKREL1/KREPA2.¹⁴ These flavonoids share a common structural scaffold, comprising a 15-carbon skeleton organized into three rings (designated as A, B, and C) with different substituents attached.^75,76 We identified two additional flavonoid compounds (dihydrorobinetin and (+)-dihydroquercetin) that inhibited ligation. The potential mechanisms underlying the bioactivity of flavonoids in cells may involve the formation of protein-flavonoid adducts or flavonoid-triggered protein oxidation.⁷⁷ Nucleophilic residues in proteins might form covalent bonds with flavonoid quinones, or flavonoids could oxidize specific amino acids such as cysteine, methionine, or tyrosine.⁷⁷ Notably, even minor structural modifications can significantly affect the inhibitory activity among flavonoids.^78,79 For instance, we found that dihydrorobinetin and (+)-dihydroquercetin inhibited ligation, whereas apigenin and naringenin did not. Dihydrorobinetin and (+)-dihydroquercetin are structurally similar dihydroflavonols, differing from the flavonoids apigenin and naringenin, which have a 3-hydroxy-4-keto function in the B-ring absent in the dihydroflavonols.

Suramin, known to inhibit editosome activities as well as rKREL1/KREPA2,^11,12,14 led us to explore its analogs, leading to the discovery of NF449 and NF023, which are known to inhibit editing activities.^11,13 We also identified PSP0739, which shares functional groups with suramin and inhibited ligation. PSP0739 and suramin were found to inhibit the P2Y12 receptor, which uses ADP as its primary physiological agonist.^80–83 These three compounds (PSP0739, NF449, and NF023) act similarly to suramin, inhibiting the recombinant ligase, rKREL1/KREPA2, and ligation by the editosome. They also inhibited the endonuclease activity by inhibiting RNA-protein interactions (EMSA data not shown).

As the ligase hydrolyzes ATP, we tested a group of nucleotide analogs and identified three additional hits that can be further utilized in SAR studies for lead compound development.

CONCLUSION

Employing a drug repurposing approach to identify novel uridine insertion/deletion RNA editing inhibitors, we screened the NCC library using our recently developed RNA editing insertion/deletion RIDE assay. This effort yielded five promising compounds (cefixime, natamycin, EGCG, isoquercitrin, and hyperoside), most of which exhibited significant antiparasitic activity. These compounds demonstrated inhibition across multiple enzymatic steps of editing performed by the editosome. Notably, cefixime and natamycin were found to inhibit RNA-protein interactions, as evidenced by EMSA analysis. Both cefixime and EGCG exhibited low micromolar efficacy against recombinant RNA editing ligase 1. Furthermore, we broadened the repertoire of ligase inhibitors by screening a selection of flavonoids, nucleotide analogs, and suramin analogs. Our findings highlight the discovery of novel RNA editing inhibitors that hold promise for further optimization toward lead drug development, especially considering their established safety profiles in humans. Such developments could involve structural modifications to reduce toxicity or strategies to improve target specificity, ensuring that any repurposed drug maintains efficacy while minimizing adverse effects in clinical settings.

METHODS

RNA Preparation.

RNA substrates used in the multiplex assay, 5′Ins, 3′Ins, 5′Del, 3′Del2, gHHRc, and FRET substrates, were synthesized by Integrated DNA Technologies (IDT, Coralville, IA) as previously described.³⁰ The gHHRc competitor, a DNA sequence, was also prepared by IDT. Fluorescent-labeled RNA sequences used in the mode of action assays (5′Del-/C3/, 5′Ins-/Cy5/, PreA6-/FAM/) were chemically synthesized and labeled by IDT. All sequences have been detailed previously.^12,30

Compound Libraries.

The NIH clinical collection library (NCC), consisting of 707 compounds, was obtained from the National Center for Advancing Translational Sciences (NCATS), NIH. The compounds were dissolved in DMSO at a concentration of 10 mM, distributed across nine 96-well plates. Daughter plates were prepared with drugs at 200 μM for screening purposes. Fresh powders of the final hits were also ordered from Sigma, and the stocks were prepared.

Ribozyme Insertion/Deletion Editing Assay.

This assay, which encompasses both uridine insertion and deletion editing, was performed as described before with a minor modification.³⁰ First, appropriate RNA hybrids for deletion {[5′ Del (2 pmol), 3′ Del2 (2 pmol), gHHRc (4 pmol)] and insertion editing [5′ Ins (2 pmol), 3′ Ins (2 pmol), gHHRc (4 pmol)] for insertion editing} were annealed in separate microtubes. This was achieved by denaturing at 70 °C in a water bath for 5 min, then gradually cooling them to 27 °C at 1 °C per minute. The RNA hybrids were then mixed and added to the master mix. The final reaction contained 25 mM HEPES (pH 7.9), 10 mM Mg (OAc)2, 50 mM KCl, 1 mM EDTA, 0.1 μM ATP, 5 mM CaCl2, 0.1% Triton X-100, 1 μL of the purified editosome (calmodulin eluate from KREL1-TAP tag purification^12,30,84 from T. brucei, approximately at 40 ng/mL), 10 μM UTP and a compound of interest (DMSO, a small molecule, or positive control like suramin), in a 20 μL final volume. The assay was set up in 96-well plates, securely covered securely, and incubated for 4 h at 28 °C. To separate the edited ribozyme from the gRNA, 40 pmol of DNA guide competitor, fully complementary to the gRNA, was added. This mixture was then incubated at 85 °C for 10 min and afterward at 25 °C for 5 min. Then 20 pmol of each of FRET1 and FRET2 substrates were added, and FAM and Cy5 signals were kinetically monitored at 37 °C every minute for 30 min using an RT-qPCR machine.

Electrophoretic Mobility Shift Assay.

To study the effect of compounds on the RNA-protein interaction, we used a fluorescent-based gel shift assay as described before.¹² Briely, gA6[14] gRNA labeled with 6-carboxyfluorescein by Integrated DNA Technologies (IDT, Coralville, IA) served as the probe. For each reaction, 10 pmol of RNA was preheated at 95 °C for 5 min in a dry block incubator and then cooled to room temperature (RT). The probe was then incubated with 5 μL of 10-fold concentrated KREL1 TAP-tagged editosome complex (2.5 nmol per reaction) before addition to the mixture. Each EMSA reaction contained 20 mM Tris–HCl (pH 7.9), 150 mM KCl, 5 mM MgCl2, 100 μg/mL BSA, 10% glycerol, 1 mM DTT, 20 units of RNasin ribonuclease inhibitor (Promega, Madison, WI), 10 pmol of the RNA probe and the protein–drug mixture, totaling 20 μL in volume. The mixture was incubated at RT for 30 min to allow RNA binding by the editosome. It was then mixed 11 with 30% glycerol and loaded on a 4% (w/v) native polyacrylamide gel. The gel was run for 15–20 min at 150 V in 0.5X TBE buffer (Tris, boric acid, EDTA) at 4 °C and analyzed using ChemiDoc mp imaging system (Bio-Rad, Hercules, CA) to visualize the fluorescent-labeled RNA bands.

Counter-Screen Assay.

In this assay, genuine editosome inhibitors were differentiated from compounds that interfered with the FRET assay, such as those inhibiting the ribozyme activity or hampering fluorescent signals. This assay follows the same condition as the multiplex assay, with one minor modification each reaction contained only one pmol of active HHR (A6Rbz) and no editosome. The reaction was incubated at 27 °C for 30 min. Subsequently, 20 pmol of FRET substrate was added, and the fluorescent signal was kinetically read to assess the ribozyme’s activity in the presence of compounds. In parallel, after adding FRET substrates and a further incubation at 37 °C for 30 min, the reaction was mixed with the 2X TBE-urea loading buffer, heated to 95 °C, and then loaded on a 20% polyacrylamide, 7 M urea denaturing gel. The gel was run for 1 h at 18 W and directly visualized using a ChemiDoc mp imaging system (Bio-Rad, Hercules, CA).

Viability Assay.

We used the Alamar blue assay to test the effects of compounds on the viability of T. brucei PRA-382 parasites (Lister 427 VSG 221 bloodstream-form⁸⁵) in a dose–response manner using 96-well plates, as previously described.^12,86,87 It is important to note that this assay does not distinguish between trypanostatic and trypanocidal effects. Briefly, parasites were seeded at a density of 2000 cells/ml in each well and incubated at 37 °C for 24 h. The next day, compounds were added to each well at their desired concentrations, with the final assay volume set at 200 μL. The plate was then incubated at 37 °C for an additional 48 h to allow the compounds to exert their effects. Subsequently, we added 20 μL of Alamar blue reagent (from Life Technologies, Carlsbad, CA) to each well, followed by further incubation at 37 °C for 4 h. Fluorescence was measured at 590 nm using a BioTek Synergy H4 Hybrid Microplate Reader (Agilent Technologies, CA, USA).

Mode of Action Assays.

Each enzymatic step of uridine insertion and deletion editing (endonuclease, TUTas, exoUase, and ligase) was replicated and monitored by a unique fluorescent gel-based in vitro assay, as previously described.¹² The assays were adaptations of former radioactivity-based editing assays, performed similarly to the multiplex assay but with variations in the RNA substrates used, gel analysis, and some assay components.^{13,30,88–90} Trimolecular RNA hybrids were prepared by heating the RNA substrates at 70 °C for 5 min and allowing them to cool to RT. The sequences used in assays were as follows [5′ Ins-/Cy5/, 3′ Ins, gHHR] for TUTase, [5′ Del-/Cy3/, 3′ Del, gHHRc] for ExoUase, [5′ Ins-/Cy5/, 3′ Del, gHHRc] for ligase and [PreA6-/Cy5/, gA6] for endonuclease activity assays. These RNAs were then mixed with the same master mix as used in the multiplex assay, with the following modifications that 100 nM ATP and 100 μM UTP were added only to the ligase and TUTase activity assays. ATP was not added to the endonuclease, TUTase, and exoUase assays to prevent inadvertent ligation by ligase. Each assay included 50 fmol of the KREL1 TAP-tagged purified editosome protein, except for the endonuclease assay, which used 250 fmol. After the reactions, a loading dye (7 M Urea in TBE buffer) was mixed in 11 ratio with the reaction mixture, then loaded onto a 20% polyacrylamide with 7 M urea and run for 2 h at 18 W. Gel analysis was performed using the ChemiDoc mp system (Biorad) to detect fluorophore-labeled RNA bands.

In precleaved assays like the FRET-based ligation assay, RNA substrates [5′ Ins, 3′ Del, gHHRc] were used under the multiplex assay conditions along with the desired ATP concentrations and protein sources, including recombinant KREL1⁴⁷ (rKREL1), or recombinant KREL1_51–459/KREPA2_56–176 (rKREL1/KREPA2) commercially expressed in BL21(DE3) Escherichia coli and purified by GenScript (Piscataway, NJ, USA), or the KREL1-TAP-tagged purified editosome from T. brucei.³⁰ Other precleaved assays, including precleaved insertion and precleaved deletion assays, followed the same format as the multiplex assay but with specific precursor RNA substrates added for insertion or deletion accordingly.

Molecular Docking Studies.

To investigate the potential interactions of EGCG with KREL1, we performed blind docking studies using the full-length AlphaFold model of KREL1. Docking was carried with the AutoDock Vina package of PyRx 0.8, as described before^91–93 (https://pyrx.sourceforge.io/). The protein model, obtained from the AlphaFold2 database (AF-P86927-F1), was prepared using BIOVIA Discovery Studio, uploaded in PyRx 0.8 and converted to the PDBQT format. The 3D SDF format file of the ligand, EGCG, was downloaded from PubChem, subjected to energy minimization, and converted to PDBQT format using the OpenBabel plugin of PyRx.⁹⁴ Within AutoDock Vina, we selected both the ligand and target protein and defined a grid box to cover the entire protein with dimensions of X 104.9 Å, Y 126.1551 Å, and Z 123.3374 Å, and center coordinates X 0.4318, Y 12.7782, and Z 29.7122. The optimal pose, which showed the lowest binding affinity and zero RMSD, was chosen for further analysis using BIOVIA Discovery Studio. We used the docking pose of ATP within the full-length KREL1 as a control to compare against the blind docking results for EGCG.