Multiparameter ranking of carbazoles for anti-trypanosome lead discovery

The criteria for the progression of hits in the discovery of leads for human African trypanosomiasis (HAT), a neglected disease caused by the microbial eukaryote Trypanosoma brucei , are not standardized. Hits are advanced upon meeting thresholds for drug-like molecules. Following those principles, pharmacokinetics (C max and AUC 0 – 6h ) and anti-trypanosome characteristics predicted the arrest of T. brucei proliferation in mice by three curaxins. Unexpectedly, while CBL0137 cured HAT in a mouse model, CBL0174 and CBL0187 — structural analogs of CBL0137 with similar drug-like properties — failed to control T. brucei division. We here propose an alternative strategy that integrates physicochemical, metabolic, pharmacokinetic, pharmacodynamic, tissue distribution, and trypanocidality parameters into calculating a score for ranking compounds in hit-to-lead campaigns. Data from our studies of curaxins support the feasibility of this goal. Serum dropped the anti-trypanosome potency of CBL0174 and CBL0187 considerably. Delayed trypanocidal concentrations (DTC 25 and DTC 90 ) were used to study modes of curaxin actions in trypanosomes. Ef ﬁ cacy of CBL0137 in mice correlated with (i) a high AUC 0 – 6h : DTC 90 ratio, (ii) blocking of transferrin endocytosis, and (iii) the inhibition of protein synthesis. Hydroxylation of the carbazole prevented CBL0137 from inhibiting endocytosis of transferrin. The multiparametric score “ Curaxin HAT lead ef ﬁ cacy (CHLE) ” score was calculated using pharmacokinetic, physicochemical, metabolic, brain exposure, and pharmacodynamic data; CBL0137 was the highest scoring hit. Complementing these observations and predictive of performance of curaxins in mice, CBL0137, but not CBL0174 or CBL0187, was trypanocidal after the exposure of trypanosomes to AUC 0 – 6h amounts of the hits for 6 hours in vitro . We discuss a role for CHLE scores in ranking curaxins for anti-HAT lead discovery. The principles used to develop CHLE scores may be used to calculate new ones for other scaffolds during the discovery of leads for HAT or other infectious diseases.


Introduction
Human African trypanosomiasis (HAT), a threat to ~60 million people in the countries of Sub-Saharan Africa, is caused by the protist Trypanosoma brucei sp.(Hasker et al., 2022).The trypanosome life cycle involves transit through a tsetse fly vector in which bloodstream trypanosomes differentiate into an insectstage form (procyclic) by expressing a different surface coat protein (Bangs, 2018;Faria et al., 2022) and altering several metabolic pathways (Michels et al., 2021).Under natural conditions, T. brucei is found in many host tissues, including the brain (Crilly and Mugnier, 2021).Switching of the major coat protein (variant surface glycoprotein) during "antigenic variation" (Bangs, 2018;Faria et al., 2022) poses a significant impediment to vaccine development.Consequently, HAT is managed with chemotherapy.
Of the limited number of drugs for treating chronic HAT, melarsoprol is toxic and others require frequent intravenous administration (Dickie et al., 2020).Fexinidazole is the first oral drug developed for chronic HAT in 40 years (Deeks, 2019;Pelfrene et al., 2019).Reports of trypanosome resistance to fexinidazole or nifurtimox (Sokolova et al., 2010;Wyllie et al., 2016;Kasozi et al., 2022) draw attention to a persistent threat posed by parasite "accommodation" of drug pressure, highlighting a need to sustain development of new chemical entities to combat HAT (Baker and Welburn, 2018;De Rycker et al., 2018).
To construct new guidelines to identify carbazole leads for HAT, we made a significant effort to understand differences in the performance of curaxin CBL0137 and its analogs in vivo.We found that (i) serum reduces the anti-trypanosome activity of some curaxins, (ii) exposure of curaxins varies in mice, and (iii) the compounds' biological actions (i.e., molecular pharmacodynamics (PD)) in T. brucei are not identical despite their extensive structural similarity (Table 1).We deduced that curaxin's potential to cure HAT in mice is associated with meeting multiple criteria.The candidate inhibits endocytosis of transferrin and protein synthesis in the trypanosome.The candidate's exposure (area under the time-concentration curve over 6 h (AUC 0-6h )) is ten times higher than trypanocidality (DTC 90 ), the concentration that is 90% trypanocidal 24-h after a 6-h exposure of trypanosomes to the compound.
Five factors, including pharmacokinetics, trypanocidality, tissue distribution, and modes of action in T. brucei, were used to develop a multi parameter score for ranking candidate curaxins in lead optimization studies for HAT drug development.A "curaxin HAT lead efficacy (CHLE)" score was calculated using pharmacokinetic, physicochemical, metabolic, brain exposure, and pharmacodynamic data.Principles used to EC 50 is the concentration of curaxin that inhibits proliferation by 50% when trypanosomes (4 × 10 3 /mL) are exposed to the compound continuously for 48 h.DTC is the delayed trypanocidal concentration of test compounds.Here, curaxin treatment is performed for 6 h with trypanosomes (5 × 10 5 /mL).After drug rinse-off, a trypanosome culture is initiated at a density of 1 × 10 5 cells/mL and survival is tracked for 24 h.DTC 50 inhibits trypanosome proliferation by 50% after drug wash off.
develop CHLE scores may be deployed to calculate new ones for other chemotypes during the discovery of leads for either HAT or different infectious diseases.
Intestinal absorption of the three test compounds was predicted using monolayers of human Caco-2 cells (Hubatsch et al., 2007).Apparent permeabilities of CBL0137, CBL0174, and CBL0187 from the apical to the basal side (P app A-to-B) were 19.0 ± 0.42 × 10 −6 cm/ s, 17.8 ± 0.35 × 10 −6 cm/s, and 14.0 ± 1.91 × 10 −6 cm/s, respectively.Permeabilities from the basal to the apical side (P app B-to-A) were 22.5 ± 0.14, 21.9 ± 3.39, and 21.7 ± 3.39 × 10 −6 cm/s, respectively (Table 2).This assay predicts permeability in the intestine and the fraction of the drug absorbed after oral administration, correlating with in vivo absorption (Panse and Gerk, 2022).Drugs with P app of less than 1 × 10 −6 cm/s are categorized as poorly absorbed, those with P app value between 1 and 10 × 10 −6 cm/s as absorbed moderately, and drugs with P app greater than 10 × 10 −6 cm/s as well absorbed (Yee, 1997).Thus, all three carbazoles are predicted to possess good in vivo absorption characteristics.
In pharmacokinetic studies, plasma levels of all three compounds were determined.Mice received single oral 40, 60, or 50 mg/kg doses of CBL0137, CBL0174, or CBL0187, respectively.These dose levels corresponded to the repeated maximum tolerated dose (rMTD) (see Methods section) of each compound.Plasma samples were obtained at 1, 2, 4, and 6 h post-dosing and analyzed.CBL0137 had the highest plasma concentration of the three compounds: C max of CBL0137, CBL0174, and CBL0187 were 1.86, 1.04, and 0.72 µM in female mice, respectively (Table 3), which were 33-, 28-, and 31-fold higher than EC 50 of T. brucei AnTat1.1 (Table 1) for corresponding compounds.Male and female mice had similar C max (Supplementary Figure S1) and similar pharmacokinetic profiles (Table 3).
CBL0137, CBL0174, and CBL0187 reached maximum plasma concentrations (C max ) 3, 1, and 1.5-h post dosing, respectively.T 1/2 of CBL0137 and CBL0187 could not be determined over 6 h since their concentrations did not drop below 50% of C max over that period.Thus, for CBL0137 and CBL0187, t 1/2 is greater than 6 h.In the case of CBL0174, t 1/2 was 3.61 h (Table 3).The mean residence time (MRT last ) was longest for CBL0137 (3.02 h) and shortest for CBL0174 (2.58 h) (Table 3).
Brain-to-plasma concentration ratios of the carbazoles ranged from 5.3 (CBL0137) to 27.9 (CBL0174) (Supplementary Figure S2).Thus, all curaxins had higher concentrations in the brain than plasma.These data predicted that CBL0137, CBL0174, and CBL0187 would control parasite load in trypanosome infections of mice.

Trypanocidality of curaxins
Trypanocidality of the compounds revealed after a short-term cell exposure supports their potential for selection of lead candidates.The premise is that drugs administered to mice are briefly in contact with trypanosomes before elimination.Therefore, we sought hits that, after a 6-h treatment and wash-off, prevented the proliferation of trypanosomes.From dose-response data, compound concentrations that caused 50% cidality (DTC 50 ) were: CBL0137, 0.29 µM; CBL0174, 0.11 µM; CBL0187, 0.11 µM (Table 1; Figure 1).Thus, all three compounds appeared to be irreversibly trypanocidal in vitro.Justification for using delayed trypanocidal concentrations (DTCs) of curaxins to study their biological effects Identifying modes of action can be difficult when hits are identified through phenotypic screening (Buckner et al., 2020), and there are no guidelines for systematically addressing this problem, so we devised protocols to assess modes of action for such hits (Sharma et al., 2022).Experiments of this nature require internal controls, so it is important to simplify the comparison of the biological effects of two or more drugs when determining how much of each compound to use per experiment.Most appropriate for such studies are concentrations that were normalized for solubility in culture medium, physicochemical properties, metabolism, and trypanocidality (Bachovchin et al., 2019;Sharma et al., 2022).We therefore determined DTC 25 concentrations of curaxins (Figure 1; Table 1).
DTC 25 is the concentration of hit that after 6-h treatment of trypanosomes (and drug wash-off) has no immediate effect on parasite viability but, following a 48-h culture, causes a 25% reduction in number of trypanosomes (Bachovchin et al., 2019;Sharma et al., 2022).To determine DTC 25 , trypanosomes were incubated with various amounts of curaxins for 6 h, rinsed, and then transferred to a drug-free medium for 48 h.The concentration of drug causing a 25% reduction in cell Delayed trypanocidal concentration (DTC) of CBL0137 analogs: T.b. brucei (5 × 10 5 /mL HMI-9) was treated with serial dilutions of compounds in 24well plates for 6 h.Trypanosomes were washed, enumerated with a Coulter Counter, and transferred into a drug-free HMI-9 medium for 24 h in a 24-well plate (1 × 10 5 cells/mL).After 24 h, cells were enumerated with a Coulter counter (see flow-chart).Data obtained were analyzed using GraphPad Prism to determine DTC 25 , DTC 50, and DTC 90 values.
numbers under these conditions (DTC 25 ) was determined as 180 nM (CBL0137), 54 nM (CBL0187), and 48 nM (CBL0174) (Figure 1; Table 1).Because DTC 25 is normalized for the growth of trypanosomes, the amount of hit causing the effect is expected to differ as reported here.
Once the biological effects of hits are determined using DTC amounts, it is very tempting to find the amount of each hit blocking a physiologic effect (i.e., EC 50 ).However, for the purpose of determining the importance of a physiologic process in proliferation inhibition, those EC 50 s are not directly relevant to cell division.That is, if at DTC 25 the event (e.g., endocytosis) is not blocked by the hit, it is difficult to rationalize the event's relevance for proliferation inhibition by the hit.Thus, those EC 50 data are best saved for studies focused strictly on the named process (e.g., translation or endocytosis or DNA synthesis).

CBL0174 and CBL0187 cannot limit proliferation of trypanosomes in mice
Since CBL0137 is a lead for anti-HAT drug discovery (Thomas et al., 2016), we tested whether CBL0174 and CBL0187 could serve as backup leads using a chronic HAT model established in mice with a pleomorphic AnTat1.1 strain of T. brucei (Claes et al., 2009).At least three waves of parasitemia were observed in mice after infection with T. brucei AnTat1.1 (i.e., vehicle-treated), as expected from the phenomenon of antigenic variation (Bangs, 2018).Parasitemia typically peaked between 10 7 and 10 8 trypanosomes/mL before receding (Figure 5A) and increasing again (Matthews, 2015).Untreated mice died 45-50 days after infection.In CBL0137-treated mice, no parasites were detected in the blood after two doses of drug (Figure 5A).Parasitemia remained undetectable for at least 93 days (Figure 5A), indicating that the mice were cured of chronic HAT.
C max or AUC 0-6h amounts of CBL0187 and CBL0174 are not trypanocidal in vitro The area under the time curve (AUC) accounts for cumulative drug exposure over 6 hours (AUC 0-6h ) (Table 3).In our studies, AUC is quoted for 6 h for all curaxins, thereby normalizing the time component in the units of measurement and allowing us to focus on the concentration parameter in AUC data.A 6-h time point was selected for our work because it is the doubling time of bloodstream T. brucei.AUC 0-6h s were 8.27 ± 0.58 µM (CBL0137), 3.14 ± 0.25 µM (CBL0174), and 2.95 ± 0.18 µM (CBL0187) (Table 3).Because hits that killed T. brucei after only a short exposure seemed the best candidates to advance in lead optimization, T. brucei was adapted to grow in medium containing 50% serum, after which recovery of cell division after curaxin treatment was monitored.This experimental setup using medium containing 50% serum mimics trypanosome blood better than regular culture medium containing 10% serum (Hirumi and Hirumi, 1994;Creek et al., 2013).We treated trypanosomes for 6 h with curaxin amounts equivalent to either C max or AUC 0-6h , rinsed the compounds, and determined the lag time for proliferation recovery (set at two cell divisions) (Figure 6).
After treatment with C max equivalents of curaxins (Figure 6A), CBL0137 was trypanocidal because no proliferation of cells was detected after leaving the culture without drug for 84 h (long enough for four cell divisions in 50% serum, as the recovery time of control cells is 18 h).CBL0187-treated trypanosomes resumed proliferation in 24 h (Figure 6A).As the recovery of the proliferation time of control T. brucei in 50% serum was 18 h, we subtracted 18 h from the measured doubling time of 24 h and recorded 6 h as the latency period (Table 4).CBL0187 was not trypanocidal, suggesting that it might not perform well in a mouse model of HAT.Likewise, CBL0174 was cytostatic in 50% serum with a delay in the resumption of proliferation of 66 h (Figure 6A; Table 4).These data predict that CBL0174 will not be efficacious in a mouse model of HAT.
The exposure of T. brucei to AUC 0-6h amounts of curaxins in medium containing 50% serum produced similar results to using C max quantities of curaxins were used.CBL0137-treated trypanosomes failed to proliferate and began dying after 60 h (Figure 6B).The lag in proliferation recovery in culture medium containing 50% serum was 30 h for CBL0187 and 66 h in the case of CBL0174 (Table 4; Figure 6B).
Taken together, these observations (Figure 6) predict data obtained from trypanosome-infected mice treated with curaxins (Figure 5), in that CBL0137 cured the infection, whereas CBL0174 and CBL0187 could not arrest trypanosome proliferation.

Discussion
An empirical formula for ranking curaxins as candidates for anti-HAT efficacy studies in a mouse model of disease African trypanosomes persist in vertebrates because of "antigenic variation" that allows the evasion of immune responses mounted by a host (Bangs, 2018;Pays and Nolan, 2021).Monomorphic T. brucei strains (e.g., T. brucei Lister427) are used routinely in laboratory models of HAT.Infected with these strains, untreated mice die within 10 days, in part because of high parasite loads.In contrast, pleomorphic trypanosomes (e.g., T. brucei AnTat1.1)sustain infection for up to 50 days by limiting trypanosome density and by antigenic variation (Faria et al., 2022), with parasitemia occurring in waves (Figure 5A).
Curaxin CBL0137 (40 mg/kg daily oral dose) cured mice infected with pleomorphic T. brucei AnTat1.1 (Figure 5A), confirming CBL0137 as a lead for HAT drug development.It is prudent in drug development to have a few backup compounds in case of unforeseen liabilities with the primary lead.Two promising backups for CBL0137 were CBL0174 and CBL0187 (Thomas et al., 2016), both of which have acceptable anti-trypanosome, metabolic, and physicochemical properties (Tables 1, 2; Figure 1).However, neither curaxin controlled parasitemia in mice (Figure 5B).
Using three structure-matched curaxins, we found a combination of factors that were associated with the efficacy of Plasma C max and AUC 0-6h of CBL0174 and CBL0187 in mice are not trypanocidal in vitro: Trypanosoma brucei Lister427 (10 5 cells/mL) cultured in HMI-9 containing 50% FBS was treated with plasma C max (A) or plasma AUC 0-6h (B) (see Table 5) for 6 h.Trypanosomes were rinsed and resuspended (10 5 cells/mL) in a drug-free medium, and their density was determined every 12 h for 84 h.TABLE 4 Latency in proliferation recovery (h) of trypanosomes after 6-h treatment with C max or AUC 0-6h curaxins in medium containing 50% serum.

Concentration of curaxin
Latency of proliferation (h) CBL0137 in a mouse model of HAT.Multiple parameters were considered: trypanocidality (Table 1), exposure of trypanosomes to curaxins in mice (Table 3), tissue distribution of curaxins in mice (Supplementary Figure S2), molecular effects of the carbazoles in trypanosomes (Figures 2-4; Supplementary Table S3), and serum shift of EC 50 (Supplementary Figure S3).From this information, we constructed an empirical formula, the "curaxin HAT lead efficacy" (CHLE) score (defined in the legend of Table 5 and discussed below).CBL0137 had the highest CHLE score.
The rationale for the CHLE formula (Table 5) is as follows.Factors whose values are desirable in large numbers are numerators (e.g., DTC 90 :AUC 0-6h , and brain-blood distribution ratio).Denominators were parameters preferred in small numbers (e.g., metabolism and residual translation inhibition).Two variables received additional weighting in acknowledgement of their high value in these studies.First, DTC 90 :AUC 0-6h was multiplied by 10 to emphasize its importance in pharmacokinetics (PK)/ pharmacodynamics (PD)/ modes of action models.AUC 0-6h is obtained from PK work in mice.That value must significantly exceed the trypanocidality (DTC 90 ) of the curaxin; otherwise, chances of the hit controlling the proliferation of T. brucei in mice is low.Second, we added the brain-blood distribution of curaxins as a denominator because anti-HAT drugs are expected to cross the blood-brain barrier to eliminate parasites in the central nervous system.Tropism of T. brucei in vertebrates is undergoing revision (Crilly and Mugnier, 2021), so other tissues harboring trypanosomes may deserve consideration in future iterations of CHLE scores.
We developed a new proliferation recovery assay to complement predictions from CHLE scores.Here, hit concentrations from pharmacokinetics (in mice) are used in vitro against cultured trypanosomes.With AUC 0-6h amounts of curaxins, CBL0137 was trypanocidal, whereas CBL0187 and CBL0174 delayed cell division by 30 or 66 h, respectively (Figure 6; Table 4).Curaxins with CHLE scores equivalent to or exceeding that of CBL0137 are expected to be lead compounds for HAT drug development.CHLE scores will be used to rank new curaxins (Singh et al., 2023) before efficacy tests in mice for anti-HAT activity.
Regarding general applications of our findings, we anticipate that formulae like CHLE scores can be developed by others working with matched hits in other pathogen drug development programs.General guidelines are as follows.CHLE score was calculated with five parameters: (i) ratio of AUC 0-6h to DTC 90 , (ii) brain-plasma distribution, (iii) serum shift of EC 50 , (iv) transferrin endocytosis stimulation, and (v) translation inhibition.Parameters (i)-(iii) specify pharmacokinetic and tissue distribution thresholds and may be regarded as constants that all potential leads for the disease must exceed (see traditional screening funnel in Figure 7).In contrast, parameters (iv) and (v) are pharmacodynamic, obtained from pathogen-hit interactions.For that reason, those data are likely to be specific for each chemical class as well as the pathogen under study.The choice of molecular pathways assessed for a chemical scaffold may be determined empirically or can be guided by known functions of the hit's targets in the pathogen-see Guyett et al. (2016), Huang et al. (2016), Guyett et al. (2017), Sullenberger et al. (2017), Barbosa da Silva et al. (2021), andMeyer andShapiro (2021) for examples.
The CHLE score has predictive potential: its application calls for data from pharmacokinetics and tissue distribution in addition to pharmacodynamics and metabolism properties.Pharmacokinetic and tissue distribution data are not available for a new set of curaxins (Singh et al., 2023), so the score function cannot yet be applied to that set of compounds.
Tf uptake: effect of curaxins (DTC 25 ) on endocytosis on transferrin is determined in comparison to a DMSO control (see Figure 4).Stimulation of endocytosis is recorded; inhibition receives a value less than 1.When transferrin uptake is stimulated by the curaxin, fractional increase has a value higher than 1.
New studies (e.g., discovery of physiologic targets of curaxins-Mensa-Wilmot (2021); Sanz-Rodriguez et al. ( 2022)) are needed to provide molecular explanations for the different effects of CBL0137 and CBL0187 on Tf endocytosis.

Limitations of the study
No agreement exists on the precise features of small molecules to guide their selection to progress in drug development programs.Different chemotypes or pharmacophores may have unique effects on the pharmacodynamic and pharmacokinetic properties of hits.Consequently, conclusions from our studies with curaxins apply to that set of carbazoles used in anti-trypanosome drug development.Whether principles from our work may be applied to other pharmacophores in infectious disease drug discovery will require a new set of studies beyond the scope of this work.
We acknowledge that we were very optimistic in exposing trypanosomes to AUC 0-6h or C max in our in vitro model of exposure of trypanosomes to curaxins in mice because neither concentration is sustained for 6 h in a mouse infection.Nevertheless, the results from that set of studies is consistent with data from treating infected mice with the three curaxins, giving credence to our newly developed proliferation recovery assay using AUC 0-6h or C max amounts of hits to rank candidates for lead compound selection.

Formulation of compounds
The test compounds (CBL0137: 1 1) in the form of hydrochlorides were provided by Incuron, Proposed sequence for determining criteria used for ranking analogs of CBL0137 before testing in mouse model of HAT.
Frontiers in Drug Discovery frontiersin.org11 Sharma et al. 10.3389/fddsv.2024.1430927Inc.(Buffalo, NY).Synthesis of the curaxins is described in US patent number PCT/US 2009/059558.For in vitro studies, stock solutions of compounds were prepared in dimethyl sulfoxide (DMSO).The final concentration of DMSO in the culture medium was ≤0.1%.For mouse studies, the compounds were dissolved in a vehicle of N-methyl-2-pyrrolidone (10% v/v) and 0.2% hydroxypropyl methylcellulose (90% v/v).

Physicochemical properties of curaxin analogs
Thermodynamic aqueous solubility (pH 7.4).Various amounts of compounds in powdered form were added to phosphate buffer (0.1 M, pH 7.4).Suspensions were kept on a shaker at 25 °C for 20-24 h.After centrifugation, compounds in the supernatant were quantified using LC-MS/MS.
Intrinsic metabolic clearance.Human liver microsome intrinsic clearance (HLM Cl int ) was assessed as described (Konsoula and Jung, 2008) at 37 °C for 15 min.The test compound was added to a final concentration of 1 μM, and the residual concentrations were tracked with LC-MS/MS at defined time points.
Caco-2 bidirectional permeability assay.Caco-2 bidirectional permeability assay was performed at Absorption Systems (Exton, Pennsylvania).Briefly, monolayers of Caco-2 cells (clone C2BBe1) were grown to confluence on collagen-coated, microporous membranes in 12-well plates.Permeability assay buffer (Hanks' balanced salt solution containing 10 mM HEPES and 15 mM glucose, pH 7.4) in the receiver chamber contained 1% bovine serum albumin.Compounds dissolved in assay buffer (5 µM) were added either on the apical (A) or the basolateral side (B) and incubated at 37 °C with 5% CO 2 in a humidified incubator.Samples were collected from donor and receiver chambers at 120 min.Each determination was performed in duplicate.The flux of lucifer yellow dye was measured post-experimentally for each monolayer to ensure that no monolayers were damaged during the flux period with compounds.Drug amounts were determined by LC-MS/MS.Apparent permeability (P app ) and percent recovery were determined as per Hubatsch et al. (2007).The efflux ratio was calculated as P app (B-to-A)/P app (A-to-B).
Maximum tolerated dose of compounds in mice.To determine a single maximum tolerated dose (sMTD), male and female Swiss Webster mice were first given a single 50 mg/kg dose of compound formulated in N-methyl-2-pyrrolidone (10% v/v) and 0.2% hydroxypropyl methylcellulose (90% v/v).Mice were monitored daily for up to 7 days post-administration for overall condition using a Total Condition Scoring (TCS) system (Supplementary Table S1) and body weight changes.If greater than 20% body weight loss was recorded, the dose was lowered in two-fold amounts in subsequent tests.If no body weight changes occurred, the single dose was increased to 100 mg/kg, and the mice were observed for up to 7 days.When no toxicity was observed, sMTD values were utilized to determine the repetitive maximum tolerated dose (rMTD).Male and female Swiss-Webster mice were treated with sMTD for 10 consecutive days.If they showed no overt signs of toxicity or greater than 20% body weight loss, the dose was considered the rMTD.If adverse effects were recorded, the daily doses were reduced two-fold in follow-up work.At the end of the study, mice were euthanized by CO 2 asphyxiation followed by cervical dislocation.
Pharmacokinetics.The plasma concentration and tissue distribution of drugs were determined by Absorption Systems.Briefly, male and female Swiss Webster mice were administered a single 40 mg/kg, 60 mg/kg, or 50 mg/kg dose of CBL0137, CBL0174, or CBL0187, respectively.These dose values corresponded to rMTD for each of these compounds.Blood and brain samples were obtained at 1, 2, 4, and 6 h post-dosing, and plasma and brain concentrations of the drug were determined using LC-MS/MS.Stock standards for calibration were prepared in a matrix comprising equal volumes of water and mouse blood containing K 2 -EDTA as an anticoagulant.Working solutions were prepared in 50:50 acetonitrile-water and added to the matrix to final concentrations of 1,000, 500, 100, 50.0, 10.0, 5.00, 1.00, and 0.500 ng/mL for all compounds.Pharmacokinetic parameters were determined using Phoenix WinNonlin (v8.0) software with sparse sampling.Maximum plasma concentration (C max ) and the time to reach maximum plasma concentration (t max ) were determined from the four data points.The area under the time concentration curve (AUC last ) was calculated using the linear trapezoidal rule with calculation to a 6-h data point.Plasma half-life (t 1/2 ) was calculated from 0.693/slope of the terminal elimination phase.Samples below the limit of detection (0.500 ng/mL) were treated as 0.

Delayed trypanocidal assay
Mid-logarithmic growth-phase T. b.Lister427 were inoculated into HMI-9 medium (5 × 10 5 cell/mL).Cells were incubated with either 0.1% DMSO (vehicle) or serial dilutions of drug in 24-well plates for 6 h at 37 °C/5% CO 2 , following which trypanosomes were pelleted (3,000 g, 5 min), rinsed with drugfree HMI-9 medium, and cells enumerated with a Coulter counter.Trypanosomes were then inoculated at 1 × 10 5 cells/ mL in 24-well-plates or 384-well plates and cultured for 24 h, after which trypanosomes were checked for viability with a hemocytometer and enumerated with a Coulter counter.Each experiment was performed thrice, and non-linear regression graphs plotted with GraphPad Prism were used to obtain DTC 50 values (Bachovchin et al., 2019).

Endocytosis of proteins
Curaxin or DMSO-treated trypanosomes (5 × 10 5 /mL) were washed with serum-free HMI-9 medium (500 uL).Trypanosomes were resuspended in serum-free HMI-9 medium (100 µL) supplemented with either transferrin-AlexaFluor647 conjugate (50 μg, Invitrogen, Eugene, OR) or BSA-AlexaFluor647 conjugate (50 μg, Invitrogen, Eugene, OR) for 30 min at 37 °C, 5% CO 2 .Cells were transferred to an ice-water bath, washed with cold PBS/G at 4 °C (3,000 g for 5 min), resuspended in PBS/G (500 μL) containing propidium iodide (3 µM), and analyzed on a flow cytometer (Beckman Coulter CytoFLEX S).FlowJo software (FlowJo, LLC) was used to gate trypanosomes based on size and shape (forward and side scatter features), and fluorescence intensity of endocytosed protein was measured in viable cells (negative for propidium iodide uptake).FlowJo software was used to determine the mean fluorescence intensity of each cargo in trypanosomes (at least 10,000 events).Data from three independent biological experiments were analyzed with a two-tailed Student's t-test with unequal variance to determine the possible statistical significance of differences in means of fluorescence intensities of endocytosed cargo.

Treatment of chronic HAT in a mouse model of disease
Drug efficacy studies in a mouse model of HAT were conducted following recommendations in the "Guide for the Care and Use of Laboratory Animals" of the National Institutes of Health.The protocol was approved by the Institutional Animal Care and Use Committee (IACUC) at the University of Georgia.After the study, the mice were euthanized by CO 2 asphyxiation followed by incision to form a bilateral pneumothorax, as approved by the American Veterinary Association.
Female Swiss-Webster mice (20-25 g; 8-10 weeks old) were acclimatized for 1 week before the commencement of the study.For the CBL0137 efficacy study, mice (n = 12) were infected intraperitoneally with 1 × 10 4 T. brucei AnTat1.1 and randomly distributed into two groups on day 0. After microscopic confirmation of infection on day 2, the treatment group received a single oral dose of CBL0137 (40 mg/kg) on days 2, 3,4,5,8,9,10,11,14,and 15. For work with CBL0174 and CBL0187,mice (n = 12) were infected intraperitoneally with 1×10 4 pleomorphic T. brucei AnTat1.1 and randomly distributed into three groups on day 0. After confirming parasitemia by hemocytometry on day 3, mice in treatment groups B and C were given either one oral dose of CBL0174 (60 mg/kg) or CBL0187 (50 mg/kg), respectively, on days 3, 4, 5, 6, 7, 8, and 9.In both studies, untreated mice were infected with 1 × 10 4 trypanosomes and then received vehicle alone.Parasites in blood collected from a tail venipuncture and diluted 8fold in RBC lysis buffer (Qiagen) were monitored with a hemocytometer (Neubauer Bright-line).During the study, mice with parasitemia above 10 8 trypanosomes/mL were euthanized for humane reasons.Mice without parasitemia for 90 days were considered cured.

Effect of 50% serum on curaxin inhibition of trypanosome proliferation
HMI-9 medium (Hirumi and Hirumi, 1989) was prepared with different final concentrations of fetal bovine serum (FBS) (20, 35, and 50%).T. brucei Lister 427 was cultured in each medium for at least 7 days.Thereafter, EC 50 of CBL0137, CBL0174, or CBL0187 was determined in the modified media as described earlier (Supplementary Figure S3).EC 50 measurements were obtained from three independent experiments in each medium.Possible statistical significance of differences in EC 50 values for control (20% FBS containing) and modified (35 or 50% FBS containing) media was determined using Student's t-test.

Recovery of trypanosome proliferation after a 6-h exposure to curaxins
The in vivo plasma exposure of trypanosomes to drug is expressed as area under the curve (AUC) in pharmacokinetic analysis.Theoretically, the same drug exposures in vitro and in vivo will be achieved within 6 h if the in vitro compound concentration × 6 h value was equal to in vivo AUC 0-6h .We hypothesized that for a compound to be effective in vivo, a 6-h in vitro exposure of T. brucei to a concentration equaling AUC 0-6h corrected for plasma protein binding (AUC 0-6h unbound ) ought to halt proliferation of the cells.AUC 0-6h unbound is equal to the AUC 0-6h × fraction of drug unbound to plasma (PF U ) (Table 4).
Trypanosomes (1 × 10 5 /mL) were treated with a concentration equivalent of the AUC 0-6h unbound (Table 4) of each compound.After 6 h, cells were washed and resuspended (1 × 10 5 cells/mL) in drug-free medium.Trypanosome densities were determined every 12 h thereafter using a hemocytometer.Since the density of control trypanosomes (treated with 0.1% DMSO) increases four-fold in 12 h, we considered any inhibition of proliferation "recovered" when trypanosome density reached 4 × 10 5 /mL.In a complementary study, trypanosomes cultured in media containing 10%, 20%, 35%, or 50% FSB and then treated with C max or AUC 0-6h (Table 3) for 6 h.Proliferation recovery time was recorded post drug wash-off as described above.

Quantification and statistical analysis
Data for all biological assays were obtained from three independent experiments with technical duplicates.Statistical tools and p-value thresholds for experiments are described in respective sections; data were analyzed using GraphPad Prism version 9.0 (GraphPad, San Diego, CA).

FIGURE 2
FIGURE 2 Curaxins reduce DNA synthesis in Trypanosoma brucei: T. b. brucei (5 × 10 5 /mL) was treated with DMSO (D: 0.1%), DTC 25 of CBL0137 (180 nM), CBL0174 (48 nM), or CBL0187 (54 nM) for 15 min and labeled with EdU (300 µM) for 1 h.Incorporated EdU was tracked by clicking with azide-Alexa Fluor-488 and observed with a fluorescence microscope.For quantitation, at least 100 cells were analyzed from each sample.(A) Representative images of the EdU-positive signal (green) with DAPI (blue) in 0.1% DMSO (top), CBL0174-treated (middle), or CBL0187-treated cells (bottom).(B) Representative images for quantitation of nuclear EdU brightness (nEdU).CellProfiler 3.1.9was used to extract the EdU fluorescence signal over the DAPI area.Lines represent the median with error bars denoting standard deviation (SD) from three biological replicates.The statistical significance of the distribution of EdU brightness in different treatment groups was analyzed with a Mann-Whitney test.

FIGURE 4 CBL0137
FIGURE 4 CBL0137, but not CBL0174 and CBL0187, reduces transferrin endocytosis: Trypanosoma brucei (5×10 5 cells/mL) in HMI-9 medium was treated with DMSO (0.1%), or DTC 25 of each curaxin at for 6 h.Cells were washed and resuspended in serum-free medium.Trypanosomes were incubated with fluorescent endocytosis ligands transferrin (Tf) or bovine serum albumin (BSA) for 30 min (37 °C).A flow cytometer was used to detect fluorescence intensity per cell.Propidium iodide (PI) was used to stain and gate out non-viable cells.Histograms present fluorescence intensity of Tf-AF488 (A) or BSA-AF647 (B) for every cell (n > 10,000 for each cargo) in DMSO or drug-treated samples.Points on the scatter plot represent mean fluorescence intensities of Tf (C) or BSA (D) from three biological replicates (calculated with FlowJo).Bars indicate mean ± SD.Student's t-test was used to evaluate statistical significance of differences in mean intensity of endocytosed ligand in DMSO-treated sample compared to curaxin-treated samples.

TABLE 2
Cell permeability properties of CBL0137 analogs.

TABLE 3
Pharmacokinetic properties of CBL0137, CBL0174, and CBL0187 in Swiss-Webster mice after a single oral dose of 40 mg/kg, 60 mg/kg, and 50 mg/kg, respectively.Area under the time curve, calculated to the 6 h time point.ND: not determined due to a lack of quantifiable data points trailing the c max .