Structure-Activity Relationship of Dibenzylideneacetone Analogs Against the Neglected Disease Pathogen, Trypanosoma brucei

Abstract

Trypanosoma brucei is a protozoan parasite that causes Human African Trypanosomiasis (HAT), a neglected tropical disease (NTD) that is endemic in 36 countries in sub-Saharan Africa. Only a handful drugs are available for treatment, and these have limitations, including toxicity and drug resistance. Using the natural product, curcumin, as a starting point, several curcuminoids and related analogs were evaluated against bloodstream forms of T. b. brucei. A particular subset of dibenzylideneacetone (DBA) compounds exhibited potent in vitro antitrypanosomal activity with sub-micromolar EC₅₀ values. A structure-activity relationship study including 26 DBA analogs was initiated, and several compounds exhibited EC₅₀ values as low as 200 nM. Cytotoxicity counter screens in HEK293 cells identified several compounds having selectivity indices above 10. These data suggest that DBAs offer starting points for a new small molecule therapy of HAT.

Graphical Abstract

Human African Trypanosomiasis (HAT), also known as sleeping sickness, is a neglected tropical disease (NTD) caused by the protozoan parasite Trypanosoma brucei.¹ There are three subspecies of T. brucei, which are all transmitted by the bite of an infected tsetse fly: T. b. brucei which primarily infects cattle and other animals, and T. b. gambiense and T. b. rhodesiense, which cause HAT.^{1, 2} HAT is endemic in sub-Saharan Africa with approximately 55 million people at risk of infection. ^{2, 3} The disease occurs in two stages: the initial, and hemolymphatic, stage 1, and the stage 2 infection that affects the central nervous system (CNS).² Current drug treatments for HAT include pentamidine and suramin, which are used as treatments for stage 1 infection by T. b. gambiense and T. b. rhodesiense, respectively.⁴ For stage 2 infections, the drugs traditionally used, e.g., melarsoprol, and eflornithine combined with nifurtimox, can induce severe adverse events, must be administered intravenously under medical supervision, and are associated with drug resistance.^4–11 Recently, fexinidazole, which treats both stage 1 and 2 infection by T. b. gambiense, was approved as the first oral drug against HAT.^{12, 13} Nevertheless, with the parasite’s ability to evolve resistance,^{11, 14} new therapies should be identified.

Curcumin (1) is a natural product isolated from the perennial herb, Curcuma longa L. (turmeric), and is a traditional ethnomedicine.^{15, 16} First discovered in the early 19^th century, curcumin and structurally related curcuminoids made their way to modern medicine in the mid- to late-20^th century,¹⁵ and are biologically active in a variety of biomedical applications,^{17, 18} including as anticancer,^19–22 anti-inflammatory,^{23, 24} and antiparasitic agents.^25–28 In spite of its utility, curcumin has been described as a pan-assay interference (PAIN)²⁹ and a poor lead compound^{30, 31} due to its instability,^{32, 33} as well as its poor adsorption, distribution, metabolism, excretion and toxicological/pharmacokinetic (ADMET/PK) properties.^33–36

Structurally, curcumin is a symmetrical molecule characterized by aryl rings separated by an α,β-unsaturated β-diketone linker. In vivo, curcuminoids are reduced to hexahydrocurcuminoids and form conjugates with sulfate and glucuronic acid.^{32, 37, 38} In vitro, one of the key degradation pathways of curcuminoids occurs via autoxidation of the α,β-unsaturated β-diketone linker to a bicyclopentadione 2 (Figure 1A).^{32, 39} To improve the in vitro and in vivo stability of curcumin and related curcuminoids, studies have explored derivatives in which the site of curcuminoid degradation has been modified.⁴⁰ Structurally derived from the curcuminoid backbone, dibenzylideneacetone (DBA, 3, Figure 1B) compounds bear α,β-unsaturated ketone aryl linkers that provide for improved in vitro stability and in vivo PK relative to curcumin.⁴⁰ Interestingly, DBA compounds are structurally similar to chalcones (e.g., 4, Figure 1B), which are considered privileged structures in medicinal chemistry and are a component of some marketed drugs.^{41, 42} DBA analogs have antiparasitic activity, including against Leishmania donovani,^{43, 44} Leishmania amazonensis,⁴⁵ Plasmodium falciparum,^46–48 Trichomonas vaginalis⁴⁹ and Trypanosoma cruzi.^{45, 50–52} Despite the continued interest in DBAs, their antiparasitic activity against Trypanosoma brucei, including any associated structure-activity relationship (SAR), has not been reported.

Figure 1.

(A) Chemical structure of curcumin and its major degradation product. (B) Chemical structures of DBA and chalcone.

Here, several natural and synthetic curcuminoids^53–58 were evaluated in vitro against bloodstream forms of T. b. brucei Lister 427. Also, curcuminoid analogs, including 26 DBAs, which bear alternative aryl linkers, were explored and evaluated. Finally, all compounds were evaluated for cytotoxicity against human embryonic kidney (HEK)293 cells. Test compounds (3, 5–47) were synthesized as described previously.^53–58 Curcumin (1) and the positive control, pentamidine, were purchased from Sigma-Aldrich. The purity of all compounds was ≥95% as determined by LC/MS.

All compounds (1, 3, 5–47) were first tested at a concentration of 4 μM against bloodstream forms of T. b. brucei Lister 427, and at a concentration of 20 μM for cytotoxicity against HEK293 cells, using the SYBR green^{59, 60} and resazurin⁶¹ cell viability assays, respectively (see Supporting Information for assay details and conditions). Compounds that were sufficiently active at these concentrations, i.e., > 50% growth inhibition, were then tested in an eight-point dose-response assay. The results are summarized in Tables 1–3.

Table 1.

Activity of curcumin and curcuminoids against T. b. brucei and HEK293 cells.

Cpd ID	Structure R	T. b. brucei EC50 (μM)a	HEK293 CC50 (μM)b	Cpd ID	Structure R	T. b. brucei EC50 (μM) a	HEK293 CC50 (μM)b
1		1.78 ± 0.03	5.56 ± 0.05	9		>4	>20
5		>4	>20	10		3.99 ± 1.56	5.56 ± 0.05
6		3.29 ± 0.12	6.73 ± 0.77	11		>4	>20
7		>4	>20	12		>4	>20
8		>4	>20	Pentamidine c		0.012 ± 0.002	>4

^aAntitrypanosomal activity against T. b. brucei Lister 427 parasites was assessed using the SYBR green cell viability assay after 72 h.

^bCytotoxicity of test compounds against HEK293 cells was assessed using the resazurin cell viability assay after 48 h. At least three assay replicates were performed each in duplicate. Means ± SD values are shown.

^cPentamidine was used as the positive drug control.

Table 3.

Activity of symmetrical and asymmetrical DBA analogs against T. b. brucei and HEK293 cells.

Cpd ID	Structure		T. b. brucei EC50 (μM)a	HEK293 CC50 (μM)b
	R	R’
3			0.80 ± 0.20	8.33 ± 0.48
28			>4	>20
29			>4	>20
30			0.24 ± 0.07	1.45 ± 0.11
31			0.84 ± 0.20	9.74 ± 1.20
32			1.42 ± 0.35	7.64 ± 1.90
33			0.20 ± 0.06	3.58 ± 0.60
34			>4	3.70 ± 0.08
35			>4	5.49 ± 0.22
36			0.29 ± 0.03	5.79 ± 0.35
37			1.25 ± 0.09	5.85 ± 0.60
38			0.37 ± 0.06	5.54 ± 0.56
39			0.66 ± 0.15	5.63 ± 0.36
40			>4	9.79 ± 0.22
41			3.74 ± 0.30	3.70 ± 0.10
42			1.93 ± 0.48	>20
43			2.15 ± 0.12	11.35 ± 0.48
44			2.33 ± 0.38	7.05 ± 0.92
45			2.51 ± 0.39	>20
46			0.20 ± 0.03	2.88 ± 0.86
47			0.41 ± 0.02	6.61 ± 0.90
Pentamidine c			0.012 ± 0.002	>4

^a-cSee Table 1 footnote.

As previously reported,^{62, 63} curcumin exhibited low micromolar activity against bloodstream forms of T. b. brucei with a 50% effective concentration (EC₅₀) of 1.78 μM (Table 1; published EC₅₀ values for curcumin are 2.5 μM⁶² and 2.7 μM⁶³). Modifications around the aryl rings led to curcuminoids 5–12, that were active (EC₅₀ values <4 μM) or inactive (EC₅₀ >4 μM) against the parasite. Removal of 3-methoxy and 4-hydroxy from 1 to 5 resulted in no measurable activity in first pass screens. Likewise, aryl substitutions with 3-methoxy (7), 3,4-dimethoxy (8) and 2,4-dimethoxy (9) were inactive. Interestingly, derivatives with more electron rich aryl rings, i.e., 6 and 10¸ exhibited activity against T. b. brucei with EC₅₀ values of 3.29 and 3.99 μM, respectively. Curcuminoids bearing one or more electron withdrawing groups on the aryl rings (11 and 12) were inactive. Overall, the SAR suggests a preference for curcuminoids with highly electron-rich aryl rings; however, as these compounds exhibited only moderate or no activity against T. b. brucei compared to the positive control drug, pentamidine (EC₅₀ = 0.012 μM), they were not pursued further.

Next, a focused set of 15 analogs with alternative aryl linkers were explored, including those derived from dibenzylideneacetone (DBA), and its reduced derivatives, 1,5-diphenylpentan-3-one (DPN) and 1,5-diphenylpentan-3-ol (DPL). These included five sets of matched molecular pairs that bear the same aryl substitutions and vary only by the oxidation state of the linker (Table 2). From this set, three compounds, 13, 16 and 19, displayed sub-micromolar activity against T. b. brucei. All three compounds belonged to the DBA class, whereas their corresponding DPN and DPL analogs were inactive at 4 μM. Compounds were also tested for cytotoxicity against mammalian HEK293 cells. Although 13, 16 and 19 showed some cytotoxicity with CC₅₀ values between 0.30 and 4.55 μM, 16 and 19 were more active against T. b. brucei than HEK293 cells with selectivity indices (SI = HEK293 CC₅₀/T. b. brucei EC₅₀) above 10.

Table 2.

Activity of dibenzylideneacetone, 1,5-diphenylpentan-3-one and 1,5-diphenylpentan-3-ol analogs against T. b. brucei and HEK293 cells.

Cpd ID	Structure		T. b. brucei EC50 (μM)a	HEK293 CC50 (μM)b
	Linker	R
13			0.36 ± 0.06	0.30 ± 0.03
14			>4	>20
15			>4	>20
16			0.42 ± 0.02	4.55 ± 0.17
17			>4	>20
18			>4	>20
19			0.31 ± 0.04	4.27 ± 0.18
20			>4	>20
21			>4	>20
22			>4	>20
23			>4	>20
24			>4	>20
25			>4	>20
26			>4	>20
27			>4	>20
Pentamidine c			0.012 ± 0.002	>4

^a-cSee Table 1 footnote.

Additional DBA analogs were explored and tested against T. b. brucei, including compounds that are either structurally symmetrical (3, 28–34) or asymmetrical (35–47) (Table 3). DBA 3, which bears symmetrical, unsubstituted aryl rings, was active with an EC₅₀ value of 0.80 μM. In contrast, the symmetrical derivatives 3-methoxy (28) and 4-methoxy (29) compounds were inactive at 4 μM. Interestingly, whereas 30 (3,3-dimethoxy) exhibited good activity with an EC₅₀ value of 0.24 μM, its analog 34, which has similar substitutions but bears a cyclohexane ring at the DBA linker, was inactive.

Regarding the asymmetrical DBA analogs (35–47), 11 of the 13 compounds exhibited moderate (EC₅₀ values between 1 and 4 μM) to potent (EC₅₀ <1 μM) antitrypanosomal activity. In general, asymmetrical compounds with electron rich rings (36–39) exhibited good activity against T. b. brucei with EC₅₀ values ranging from 0.29 to 1.25 μM. Exceptions to this pattern included compounds 35 and 40, which, despite having electron-rich aryl rings, showed no activity. Conversely, asymmetrical analogs with electron deficient rings (42–45) showed comparatively less activity with EC₅₀ values ranging from 1.93 to 2.51 μM. Interestingly, the asymmetrical DBA, 46, which bears an electron-withdrawing nitro group on one aryl ring, was one of the most potent analogs with an EC₅₀ value of 0.20 μM. Overall, the SAR of the DBA analogs suggests that the electronic properties of the aryl rings may influence potency against T. b. brucei, and that there may also be some preference for certain functional groups, particularly a nitro (46), that is not necessarily dictated by the electronic properties of the aryl rings. All DBA analogs were tested for cytotoxicity against HEK293 cells. In general, the 19 antitrypanosomal DBA analogs exhibited some cytotoxicity, though the most active compounds were more active against T. b. brucei with 17 analogs having SI values above 3, and 10 analogs with SI values above 10 (Figure 2).

Figure 2.

Plot of the log of T. b. brucei EC₅₀ values vs. HEK293 CC₅₀ values. Indicated are those DBA analogs that exhibit SI values above 1 (red line), 3 (blue line) and 10 (black line).

To measure the speed of the antitrypanosomal activity of the DBA analogs, a time-kill experiment was conducted using 46. T. b. brucei parasites were incubated with 46 (EC₉₉ = 0.90 ± 0.16 μM) at 1, 2 and 4 μM. As a comparison, parasites were also incubated with the positive control drug, pentamidine, at 0.5, 1 and 2 μM (Figure 3). Compound 46 required less than 24 h (less than 6 h at 2 and 4 μM) to kill the parasite at all concentrations tested, whereas pentamidine required >30 h to kill all of the parasites (Figure 3).

Figure 3.

Compound 46 is a faster-acting antitrypanosomal compound than pentamidine. Parasites were quantified by hemocytometer cell counting and the counts were normalized to a negative (DMSO) control. Means of three replicates ± SD bars are shown.

To distinguish trypanocidal activity from trypanostatic activity, we performed a washout experiment using compound 46. We first incubated 1, 2 and 4 μM of the compound with trypanosomes for 6, 4 and 2 h, respectively, and then washed the remaining parasites three times. The parasites did not regain the ability to divide after monitoring for a further 48 h, suggesting that 46 is trypanocidal rather than trypanostatic (Figure S1, Supporting Information).

Curcumin and curcuminoid analogs have established biological activity, including as anticancer,^19–22 anti-inflammatory,^{23, 24} and antiparasitic agents.^25–28 Nonetheless, these natural products have been considered to be poor lead compounds^29–31 due to their instability^{32, 33} and weak ADMET/PK properties.^33–36 In the present study, curcumin (1) and curcuminoids (5–12; Table 1) were evaluated for their antiparasitic activity against bloodstream forms of T. b. brucei. Whereas curcumin 1, and curcuminoids 6 and 10, resulted in moderate antitrypanosomal activity (EC₅₀ values of 1–4 μM), the remaining eight curcuminoids were inactive (EC₅₀ values >4 μM). To explore structurally related derivatives with improved metabolic stability,⁴⁰ compounds with alternative aryl linkers were explored, including DBAs and their reduced analogs, DPN and DPL (Table 2). Through the matched molecular pair analysis of the DBA, DPN and DPL analogs, it was established that the presence of the dienone in DBA, which may serve as a Michael acceptor, is essential for antitrypanosomal activity. This is demonstrated by the loss of activity when the Michael acceptor is removed, e.g., 13 with an EC₅₀ value of 0.36 μM vs 14 and 15 with EC₅₀ values >4 μM (Table 2).

Compounds with Michael acceptors are known antiparasitics.^{50, 62–64} Due to the intrinsic reactivity of Michael acceptors, a possible MOA may the involve reaction with nucleophilic residues, including with the catalytic site of cysteine proteases. A study by Ettari et al. has shown that curcumin, in combination with related Curcuma longa L. natural product, genistein, inhibits the major cathepsin L-like cysteine protease (TbrCATL)⁶⁵ of T. b. rhodesiense.⁶⁶ In the context of mono-enones, studies by Changtam et al. explored curcuminoid analogs bearing Michael acceptor mono-enones, several of which were active against bloodstream forms of T. b. brucei, Leishmania major promastigotes and Leishmania mexicana amastigotes.⁶² In a follow up study by Alkhaldi et al., the mechanism of action (MOA) of these mono-enones in T. b. brucei was explored, and their antitrypanosomal activity was attributed to the formation of trypanothione adducts as a result of Michael addition between the mono-enone and trypanothione. These mono-enone Michael acceptors may, therefore, deplete the parasite of essential thiols, leading to parasite death.⁶³ For the DBA analogs evaluated here, a similar MOA may be possible (Figure 4).

Figure 4.

Proposed MOA of Michael acceptors as antitrypanosomal agents,⁶³ the influence of DBA aryl substitutions and reactivity towards Michael addition, and the stability of the Michael adduct that may form.⁶⁷ TSH = trypanothione; EWG = electron withdrawing group; EDG = electron donating group.

From the present evaluation of 26 Michael acceptor DBA analogs against T. b. brucei, several aspects of the SAR were observed. In particular, the 26 DBA analogs differed by 1) the electronic properties of the aryl rings, 2) their symmetry, and 3), for a small number of compounds, the cyclic nature of the aryl linker (e.g., compounds 33 and 34). Using the unsubstituted DBA 3 as the reference compound, 13, 16, 22, 28–31, 34–41 and 47 have electron-donating groups on one or both aryl rings, whereas 19, 25, 32, 33 and 42–46 have electron-withdrawing groups on one or both aryl rings. The electronic nature of the aryl rings is expected to influence the reactivity towards Michael addition, as well as the stability of the Michael adduct that may form (Figure 4). In general, electron rich analogs were moderately more active (8 out of 16 compounds exhibited EC₅₀ values <1 μM) than electron deficient analogs (3 out of 9 exhibited EC₅₀ values <1 μM). Studies have shown that adducts formed by Michael acceptors bearing electron-rich aryl rings are expected to be more stable and less reversible than those bearing electron-deficient aryl rings,⁶⁷ which may explain why electron rich DBA analogs were generally more active. Interestingly, despite this overall trend in the SAR, 46, an asymmetrical nitro DBA analog bearing one electron deficient aryl ring, was one of the most active analogs with an EC₅₀ value of 0.20 μM. It is known that nitro-DBAs increase the production of reactive oxygen species (ROS) and damage the parasite antioxidant system in Trypanosoma cruzi.⁵⁰ This cidal activity against T. cruzi was attributed to the presence of the nitro group within the DBA analog that may be involved in the production of ROS.⁶⁸ Due to the potent activity of 46 against T. b. brucei, a similar MOA in T. b. brucei is possible.

In addition, the SAR of both symmetrical (3, 13, 16, 19, 22, 25, 28–34) and asymmetrical DBAs (35–47) was evaluated. As DBA analogs have two alkenes and therefore two sites for Michael addition, the key difference between the symmetrical and asymmetrical analogs is the possible formation of either one or two Michael adducts, respectively (see Figure 4 for an example of two generic Michael adducts). Generally, the symmetrical DBAs were more active (7 out of 13 exhibited EC₅₀ values <1 μM) than the asymmetrical DBAs (5 out of 13 with EC₅₀ values <1 μM). However, upon examining the symmetrical compound, 31, and its structurally closest asymmetrical analog, 39, the asymmetrical analog was slightly more potent against T. b. brucei (respective EC₅₀ values of 0.84 and 0.66 μM). This suggests that further study of the SAR is required to assess whether symmetry contributes to the antitrypanosomal activity of DBA analogs.

Finally, whereas most of the DBA analogs studied have linear dienone aryl linkers, two analogs with cyclic dienones, the piperidinone, 33, and the cyclohexanone, 34, were also included in the study. Compound 33 exhibited one of the most potent activities against T. b. brucei (EC₅₀ value of 0.20 μM) whereas 34 exhibited no activity (EC₅₀ value > 4 μM). Additional exploration of these cyclic dienones is required to evaluate the SAR of linear vs cyclic DBAs.

All compounds were also tested for toxicity against HEK293 cells and those that were active against T. b. brucei tended to have measurable activity against the mammalian cell line, although with a broad range of SI values (Figure 2). These data might suggest a general MOA irrespective of cell type. As per the MOA suggested by Alkhaldi et al. that involves the interaction with trypanothione in trypanosomes, mammalian cells also have essential thiols, not least, glutathione, that could form glutathione adducts with the DBAs.⁶³

To conclude, the present study identifies a series of DBA analogs as potent antitrypanosomal agents with EC₅₀ values as low as 0.20 μM and SI values up to >10. Additionally, whereas the DBA analogs were comparatively less active than pentamidine, the time-to-kill experiments show that analog 46 is a relatively faster trypanocide. The data encourage the continued investigation of DBAs as the possible basis for a novel chemotherapy of HAT.