In-silico Investigation of Antitrypanosomal Phytochemicals from Nigerian Medicinal Plants

Background Human African trypanosomiasis (HAT), a parasitic protozoal disease, is caused primarily by two subspecies of Trypanosoma brucei. HAT is a re-emerging disease and currently threatens millions of people in sub-Saharan Africa. Many affected people live in remote areas with limited access to health services and, therefore, rely on traditional herbal medicines for treatment. Methods A molecular docking study has been carried out on phytochemical agents that have been previously isolated and characterized from Nigerian medicinal plants, either known to be used ethnopharmacologically to treat parasitic infections or known to have in-vitro antitrypanosomal activity. A total of 386 compounds from 19 species of medicinal plants were investigated using in-silico molecular docking with validated Trypanosoma brucei protein targets that were available from the Protein Data Bank (PDB): Adenosine kinase (TbAK), pteridine reductase 1 (TbPTR1), dihydrofolate reductase (TbDHFR), trypanothione reductase (TbTR), cathepsin B (TbCatB), heat shock protein 90 (TbHSP90), sterol 14α-demethylase (TbCYP51), nucleoside hydrolase (TbNH), triose phosphate isomerase (TbTIM), nucleoside 2-deoxyribosyltransferase (TbNDRT), UDP-galactose 4′ epimerase (TbUDPGE), and ornithine decarboxylase (TbODC). Results This study revealed that triterpenoid and steroid ligands were largely selective for sterol 14α-demethylase; anthraquinones, xanthones, and berberine alkaloids docked strongly to pteridine reductase 1 (TbPTR1); chromenes, pyrazole and pyridine alkaloids preferred docking to triose phosphate isomerase (TbTIM); and numerous indole alkaloids showed notable docking energies with UDP-galactose 4′ epimerase (TbUDPGE). Polyphenolic compounds such as flavonoid gallates or flavonoid glycosides tended to be promiscuous docking agents, giving strong docking energies with most proteins. Conclusions This in-silico molecular docking study has identified potential biomolecular targets of phytochemical components of antitrypanosomal plants and has determined which phytochemical classes and structural manifolds likely target trypanosomal enzymes. The results could provide the framework for synthetic modification of bioactive phytochemicals, de novo synthesis of structural motifs, and lead to further phytochemical investigations.


Introduction
Human African trypanosomiasis (HAT), also known as sleeping sickness, is caused by the single-celled kinetoplastid parasites, Trypanosoma brucei, which are transmitted to humans by infected tsetse flies (Glossina spp.). Two sub-species of T. brucei (rhodesiense and gambiense) cause the two different forms of the disease. T. b. rhodesiense is found in southern and eastern Africa while T. b. gambiense is found in the western, central and some parts of eastern Africa. T. b. gambiense now accounts for about 90% of all reported cases of sleeping sickness. A third subspecies, T. b. brucei, does not cause HAT because of its susceptibility to lysis by human apolipoprotein L1 [1].
Current chemotherapies of HAT are directed either to the early or late stages of the disease. All the clinically available HAT chemotherapeutic drugs have been noted to be ineffective, and they also have severe side-effects. The only drug candidate in clinical trials for the treatment of HAT is the nitroimidazole fexinidazole. Fexinidazole is currently in clinical study for the treatment of the late stage form of HAT [2,3]. It is worth noting that the number of reported cases of HAT fell in the past decade, and it has also been suggested that a possible elimination of the disease might be in sight [4]. This is a very delightful development for this ''neglected'' tropical disease, and it is our hope that continued research into new and effective chemotherapy against HAT remains an integral part of public health initiatives in endemic communities.
Medicinal plants from Nigeria's lush rainforest, as well as her very diverse montane and savanna vegetation, continue to play a vital role in her healthcare system. For tens of millions of Nigerians, indigenous traditional medicine is the major -and sometimes the only -access to pharmacological agents [5]. There have been several published reports on the biological activity of Nigerian plants, but most of the bioactive components of those plants have not been characterized. However, the country's big and loosely-regulated traditional medicine industry continues to promote the efficacy of extracts and concoctions made from most of the plants. A number of Nigerian plants have been used traditionally in West Africa to treat protozoal infections and many of these have shown in-vitro antiprotozoal activity (Table S1).
Several T. brucei protein targets have been identified and experimentally validated [6]. In addition to validated targets, several potential targets have been predicted in silico [7]. For a recent review of phytochemical agents that show activities against parasitic protozoans and protozoan biochemical targets, see [8,9]. Some of the potential T. brucei drug targets that we considered in this work include adenosine kinase [10], pteridine reductase 1 [11], dihydrofolate reductase [12], trypanothione reductase [13], cathepsin B [14], heat shock protein 90 [15], as well as sterol 14ademethylase (CYP51) [16], nucleoside hydrolase [17], triose phosphate isomerase [18], nucleoside 2-deoxyribosyltransferase [19], UDP-galactose 49 epimerase [20] and ornithine decarboxylase [21]. In this computational study, we have evaluated the interaction of compounds that were isolated from some antitrypanosomal Nigerian medicinal plants (Table S1) against potential protein drug targets in Trypanosoma brucei for which X-ray crystal structures were available from the Protein Data Bank (PDB). We strove to address the questions of which phytochemical agents might be responsible for the observed antitrypanosomal activity and what are the likely targets of those phytochemicals. In doing so, we hope to identify particular classes of phytochemical agents that can be exploited for antiparasitic chemotherapy.
brucei sterol 14a-demethylase, TbCYP51 (PDB 3gw9 [16]), T. brucei nucleoside hydrolase, TbNH (PDB 3fz0 [31]), T. brucei triosephosphate isomerase, TbTIM (PDB 1iih, PDB 6tim [32], and PDB 4tim [33]), T. brucei nucleoside 2-deoxyribosyltransferase, TbNDRT (PDB 2a0k, PDB 2f64, and PDB 2f67 [19]), T. brucei UDP-galactose 49-epimerase, TbUDPGE (PDB 1gy8 [20]), and T. brucei ornithine decarboxylase, TbODC (PDB 1f3t [34], PDB 1njj [35], and PDB 1qu4 [21]). All solvent molecules and the cocrystallized ligands were removed from the structures. Molecular docking calculations for all compounds with each of the proteins were undertaken using Molegro Virtual Docker v. 4.3 [36,37], with a sphere large enough to accommodate the cavity centered on the binding sites of each protein structure in order to allow each ligand to search. If a co-crystallized inhibitor or substrate was present in the structure, then that site was chosen as the binding site. If no co-crystallized ligand was present, then suitably sized cavities were used as potential binding sites. Standard protonation states of the proteins based on neutral pH were used in the docking studies. The protein was used as a rigid model structure; no relaxation of the protein was performed. Assignments of charges on each protein were based on standard templates as part of the Molegro Virtual Docker program; no other charges were set. Each ligand structure was built using Spartan '08 for Windows [38]. The structures were geometry optimized using the MMFF force field [39]. Flexible ligand models were used in the docking and subsequent optimization scheme. As a test of docking accuracy and for docking energy comparison, co-crystallized ligands were re-docked into the protein structures. Different orientations of the ligands were searched and ranked based on their energy scores. The RMSD threshold for multiple cluster poses was set at ,1.00 Å . The docking algorithm was set at maximum iterations of 1500 with a simplex evolution population size of 50 and a minimum of 30 runs for each ligand. Each binding site of oligomeric structures was searched with each ligand. The lowestenergy (strongest-docking) poses for each ligand in each protein target are summarized in Tables S2-S20.

Author Summary
Traditional herbal medicine continues to play a key role in health, particularly in remote areas with limited access to ''modern medicines''. Many plants are used in traditional Nigerian medicine to treat parasitic diseases. While many of these plants have shown notable activity against parasitic protozoa, in most cases the mode of activity is not known. That is, it is not known what biochemical entities are being targeted by the plant chemical constituents. In this work, we have carried out molecular docking studies of known phytochemicals from Nigerian medicinal plants used to treat human African trypanosomiasis (sleeping sickness) with known biochemical targets in the Trypanosoma brucei parasite. The goals of this study were to identify the protein targets that the medicinal plants are affecting and to discern general trends in protein target selectivity for phytochemical classes. In doing so, we have theoretically identified strongly interacting plant chemicals and their biomolecular targets. These results should lead to further research to verify the efficacy of the phytochemical agents as well as delineate possible modifications of the active compounds to increase potency or selectivity.
3-chromene, 6-acetyl-2,2-dimethylchroman, and O-methylenececalinol exhibited selectivity for TbTIM with docking energies comparable to the co-crystallized ligand, 3-phosphoglyceric acid (221.6 kcal/mol). The flavonoid 39,49,5,59,6,8-hexamethoxyflavone, on the other hand, showed selective docking to TbPTR1 and TbUDPGE. Nour and co-workers [49] have examined the antitrypanosomal activities of several methylated flavonoids and a chromene from A. conyzoides. The flavonoids all have similar antitrypanosomal activities with IC 50 values ranging from 3.0 to 6.7 mg/mL. The chromene, O-methylencedalinol, on the other hand, was much less active (IC 50 = 78.4 mg/mL). The docking energies for many of the protein targets was much more negative (stronger docking) for the flavonoids than for the chromene. Thus, for example, there is good correlation between log(IC 50 ) and docking energies of the ligands with TbPTR1 or with TbUDPGE (R 2 = 0.712 and 0.751, respectively).

Garcinia kola
Polyphenolic compounds, flavonoids, biflavonoids, etc., have been isolated and identified from Garcinia kola (Table S8) [40,66]. G. kola biflavonoids docked favorably with TbAK and TbODC. The biflavonoids do not dock at the adenosine binding sites of TbAK, but rather in a pocket between the two sites bounded by residues Asn222, Gly298, Ala297, Thr264, Asp266, Glu225, Arg132, and Asn195 (see Fig. 3). Likewise, biflavonoid docking with TbODC does not occur at the ornithine/putrescine binding site or the geneticin binding site, but rather in a pocket bounded by Asp243, Asp385, Val335, Asp332, Ala334, Ala244, and Arg277 (Fig. 4). This would suggest that if G. kola biflavonoids inhibit either TbAK or TbODC, they act as allosteric inhibitors of these proteins. The two tocotrienols garcinal and garcinoic acid, on the other hand, docked more favorably with TbUDPGE. Key interactions of the tocotrienols with the protein are hydrogen-bonding of the phenolic -OH of the ligands with Pro253 and Phe255, hydrogen-bonding of the carbonyl group of the ligand side chains with Arg268, hydrogen-bonding of the pyran ring oxygen atom with Arg235, face-to-face pp interactions of the ligand aromatic rings with Phe255, and hydrophobic interactions of the tocotrienol ligands with Leu222, His221 and the NAD cofactor (Fig. 5 top). The prenylated benzophenone kolanone docked very strongly with TbNH (docking energy = 237.1 kcal/mol) in the nucleoside binding site (Fig. 6), a hydrophobic pocket bounded by Trp80, Phe178, Asn171, Trp205, and Val277, with additional hydrogen-bonding with Asn171.

Lawsonia inermis
The sterols and triterpenoids [40] from Lawsonia inermis showed preferential docking to TbCYP51 (T. brucei sterol 14a-demethylase) (Table S10). This is perhaps not surprising since the normal substrates for this enzyme are sterols. The laxanthones from L. inermis showed preferential docking to TbPTR1 with docking energies comparable to the co-crystallized ligand. In addition, they docked in the same positions and orientations as pseudocolumbamine and pseudopalmatine from Enantia chlorantha (see above and Fig. 2).

Morinda morindoides
The phytochemistry of Morinda morindoides [40] is dominated by flavonoid glycosides [77] and phenylpropanoid-conjugated iridoid glycosides [78] (Table S12). Of these, epoxygaertneroside and morindaoside were selectively strongly binding ligands for TbAK, and morindaoside also docked strongly to TbPTR1. A number of gaertneroside derivatives showed docking selectivity for TbDHFR (see Table S12), while dehydroepoxymethoxygaertneroside docked very strongly with TbNH, occupying the nucleoside binding site (Fig. 6) with the same hydrophobic interactions as kolanone and oruwacin (above). It is unlikely that these glycosides will remain intact in vivo, and hydrolysis may be necessary for absorption and general bioavailability [79]. Of the flavonoid aglycones from M. morindoides, apigenin, chrysoeriol, kaempferol, quercetin, and ombuin selectively docked with TbPTR1.

Newbouldia laevis
T. brucei triosephosphate isomerase, TbTIM, is the likely protein target for the phytochemical agents of Newbouldia laevis. Both furanonaphthoquinones [82,83] and pyrazole alkaloids [84,85] from this plant showed remarkable selective affinity for this protein (Table S14). The monomeric furanonaphthoquinone ligands all occupy the same site with hydrogen bonding of the furan oxygen and C(9) carbonyl oxygen to Lys313; C(4) carbonyl oxygen with Ser513 and Val514; a van der Waals surface provided by Val533, Gly534, Gly535; and a hydrophobic pocket to accommodate the isopropenyl moiety provided by Ile472, Gly512, and Leu532 (see Fig. 9). Similarly, the pyrazole alkaloid 49-hydroxywithasomnine has key hydrogen-bonding interactions between the pyrazole ring nitrogens and Ser513 and Val514. The aromatic ring lies in the hydrophobic pocket made up of Ile472 and Leu532, and there is an additional hydrogen-bonding interaction between the phenolic -OH group and His395 (Fig. 9).

Picralima nitida
Picralima nitida glycosylated coumestans [89] showed strong binding to most of the protein targets, except for TbNDRT (Table  S16). They are, for example, along with Acacia nilotica flavonoid gallates, the only ligands that dock to rhodesain with docking energies comparable to the co-crystallized ligand. Of the ligands examined in this work, coumestan 2 is the strongest-binding ligand for TbAK (244.1 kcal/mol) and TbUDPGE (243.7 kcal/mol). The corresponding aglycones, 4, 5, and 6, however, were selective for TbPTR1 as well as TbUDPGE.

Prosopis africana
The phytochemistry of Prosopis africana is characterized by piperidine alkaloids (Table S17) [40,90]. These ligands exhibited similar docking energies with all protein targets, owing presumably to the small, flexible nature of the compounds. They did, however, show slightly better affinity for TbAK.

Securidaca longipedunculata
Cinnamate esters from Securidaca longipedunculata [40] showed selective docking to TbTIM while S. longipedunculata xanthones [40,93] had a docking preference for TbPTR1. Both of these protein targets have relatively small binding sites, which are more suitable for the small ligands (Fig. 10). The xanthones also docked relatively strongly with TbUDPGE. The S. longipedunculata indole alkaloids dehydroelymoclavine and alkaloid A [94] docked strongly to TbPTR1 and TbAK, respectively, as well as with TbUDPGE (see Table S19).

Strychnos spinosa
Phytochemicals isolated from Strychnos spinosa include secoiridoids [95], indole, pyridine, and naphthyridine alkaloids [40], sterols and triterpenoids [96] (Table S20). Relatively small pyridine and naphthyridine alkaloids from Strychnos spinosa showed preferential docking to TbPTR1 and/or TbTIM. The indole alkaloids akagerine, 10-hydroxyakagerine, and kribine also docked preferentially to TbPTR1, while iridoid glucosides preferred TbUDPGE. Both sterols and triterpenoids from S. spinosa selectively docked with TbCYP51, similar to what was observed with L. inermis sterols and triterpenoids (see above), but these compounds also showed notably strong docking with TbODC. Interestingly, a comparison of docking energies of triterpenoid and steroid ligands with their antitrypanosomal activities [96] shows no correlation, even comparing TbCYP51 docking or TbODC  docking. Plots of log(IC 50 ) vs. docking energies gives R 2 values of 0.043 and 0.007 for TbCYP51 and TbODC, respectively. It may be that inhibition of some other protein target [8,9] is the biochemical mechanism of activity for these compounds.
In terms of natural products drug discovery, it is useful to examine whether different phytochemical classes show selectivity for particular protein targets. Simple flavonoid ligands showed docking preferences for TbPTR1 and TbUDPGE. Flavonoid gallates, on the other hand, were shown to be promiscuous docking ligands to all protein targets, but were particularly strongly docking with TbAK, TbPRT1, TbCYP51, and TbNH. Likewise, flavonoid glycosides tended to be promiscuous docking agents, but with preference for TbAK, TbPRT1, and TbNH. Oligomeric flavonoids (tannin-like polyphenolics) showed strong docking to TbAK. The diversity of flavonoid structures has led to diverse biological activities, including antiprotozoal activity, but the modes of antiprotozoal activity have not been well elucidated [97].
As previously noted (see above), triterpenoid ligands were largely selective for TbCYP51. Withanolide triterpenoids also showed a docking preference for TbCYP51, while limonoids preferentially docked with TbAK as well as TbCYP51. Not surprisingly, sterols showed a propensity to dock with TbCYP51, but also docked strongly with TbUDPGE.
All of the anthraquinone ligands examined in this docking study, docked with strong binding energies to TbPRT1. Likewise, xanthone ligands exhibited docking selectivity for TbPTR1. Naphthoquinones, on the other hand, docked preferentially with TbTIM. Most chromene ligands also showed notable docking energies to TbTIM. The phenylpropanoids examined showed preferences for TbTIM as well as TbUDPGE, while glycoside derivatives of phenylpropanoids showed selectivity for TbDHFR.
Berberine alkaloids docked preferentially to TbPTR1 while aporphine alkaloids showed some selectivity for TbPTR1 and TbUDPGE. Piperidine alkaloids were also selective for  TbUDPGE. Pyrazole and pyridine alkaloids, on the other hand, preferred docking to TbTIM. A total of 93 indole alkaloids were examined in this docking study and many of them showed notable docking energies with TbUDPGE and some with TbAK and TbPTR1. Glycoside derivatives of alkaloids also preferentially docked with TbUDPGE.
Overall, the protein objects most targeted by the phytochemical ligands in this study were TbUDPGE, targeted by many alkaloids; TbPTR1, preferred by planar-like ligands; TbCYP51, which docked terpenoid ligands well; and TbAK, which docked many different classes of phytochemicals. Those proteins least preferred in terms of docking energies were rhodesain, TbCatB, and TbNDRT.
Rhodesain and TbCatB are both cysteine proteases with relatively small binding sites. It may be that the docking energies reflect the fact that only relatively small ligands, with inherently small docking energies, can fit well into the binding sites of these two proteins. The docking energies do not, however, reflect the potential for covalent bonding to the active sites of these proteins. It is useful, therefore, to examine small electrophilic ligands for   [41]). Right: With TbCatB (PDB 3hhi [46]). Note the proximity and orientation of the quinone moiety with the cysteine sulfur atoms in the active sites. doi:10.1371/journal.pntd.0001727.g014 Docking of Antitrypanosomal Phytochemicals www.plosntds.org energetically favorable docking orientations that would allow for reaction of nucleophilic amino acid side chains to the electrophilic sites of the ligands.
Although umbelliferone does not dock with particularly strong energies to rhodesain or TbCatB, it does dock in poses such that the nucleophilic Cys25 of rhodesain or Cys122 of TbCatB are poised to undergo conjugate addition to the pyrone ring (Fig. 11). The S atom of Cys25 is 3.16 Å from C(4) of docked umbelliferone in rhodesain, while in TbCatB, Cys122 is 3.65 Å from C(4) of umbelliferone. Coumarins have been shown to be trypanocidal agents [98] and it has been suggested that umbelliferone undergoes conjugate addition with available cysteine thiol groups [99].
Many naphthoquinones have been shown to be antitrypanosomal [100], and are suspected to interfere with redox thiol metabolism by inhibition of TbTR [101,102]. There are docking poses, albeit not the lowest-energy poses, of isoplumbagin (docking pose energy = 29.9 kcal/mol) and lawsone (docking pose energy = 28.6 kcal/mol) with TbTR such that these quinone ligands are in the proximity of reduced trypanothione (Fig. 12). Similarly, both isoplumbagin and lawsone dock with the cysteine proteases rhodesain and TbCatB with the electrophilic carbons near the active-site cysteine residues (Fig. 13). N. laevis furanonaphthoquinones (a-lapachone derivatives) also dock with rhodesain in poses such that the nucleophilic Cys25 can undergo Michael addition to the quinone ring (Fig. 14). None of the furanonaphthoquinones docked near the trypanothione thiol groups in TbTR, however.
This in-silico investigation suggests that trypanosomal phytochemicals may target different protein targets. There are several caveats to these docking results: (a) many of the phytochemical agents may not be bioavailable due to limited solubility, membrane permeability, hydrolysis, or other metabolic decomposition; (b) tannins and other polyphenolics are promiscuous protein binding agents and are likely, therefore, not selective antitrypanosomal ligands; (c) the docking studies do not account for synergism in bioactivity of phytochemicals; (d) this current study does not address the binding of ligand to human homologous isozymes, which may also be targeted; (e) there are likely additional phytochemicals in each of the medicinal plants that have not been isolated or identified; and (f) there are likely additional trypanosomal proteins or other biochemical targets that have not yet been identified. Nevertheless, this in-silico molecular docking study has provided evidence for what phytochemical classes and structural manifolds are targeting particular trypanosomal protein targets and could provide the framework for synthetic modification of bioactive phytochemicals, de novo synthesis of structural motifs, and further phytochemical investigations.