Alkanediamide-Linked Bisbenzamidines Are Promising Antiparasitic Agents

A series of 15 alkanediamide-linked bisbenzamidines and related analogs was synthesized and tested in vitro against two Trypanosoma brucei (T.b.) subspecies: T.b. brucei and T.b. rhodesiense, Trypanosoma cruzi, Leishmania donovani and two Plasmodium falciparum subspecies: a chloroquine-sensitive strain (NF54) and a chloroquine-resistant strain (K1). The in vitro cytotoxicity was determined against rat myoblast cells (L6). Seven compounds (5, 6, 10, 11, 12, 14, 15) showed high potency against both strains of T. brucei and P. falciparum with the inhibitory concentrations for 50% (IC50) in the nanomolar range (IC50 = 1–96 nM). None of the tested derivatives was significantly active against T. cruzi or L. donovani. Three of the more potent compounds (5, 6, 11) were evaluated in vivo in mice infected with the drug-sensitive (Lab 110 EATRO and KETRI 2002) or drug-resistant (KETRI 2538 and KETRI 1992) clinical isolates of T. brucei. Compounds 5 and 6 were highly effective in curing mice infected with the drug-sensitive strains, including a drug-resistant strain KETRI 2538, but were ineffective against KETRI 1992. Thermal melting of DNA and molecular modeling studies indicate AT-rich DNA sequences as possible binding sites for these compounds. Several of the tested compounds are suitable leads for the development of improved antiparasitic agents.


Methodology
We used Surflex-Dock to study the DNA binding affinities of 16 bisbenzamidine compounds to the minor groove of AT-rich sites. The AAATTT sequence was the primary target of interest, but we also considered to target AAAGTTT and AAAGCTTT, because a single G or GC inserted short AT base pairs are common in eukaryotic parasite genomes. One DNA sequence rich in G-C base pairs (5′-d(GCGCGCGCGC)-3′) was also considered. For the AAATTT sequence we used 5′-d(CCAAATTTGC)-3′ as used in a recent study [2]. The AAAGTTT and AAAGCTTT sequences were constructed by inserting G and GC to the center of the 5′-d(CCAAATTTGC)-3′ sequence. The initial regular B-DNA duplexes were constructed using the Biopolymer/Build module in SYBYL-X1.3 and energy minimized using the Powell method and Tripos force filed. Initial 3D structures of all the l6 compounds were built using the sketch molecule function in SYBYL-X and energy minimized using the Powell method with Tripos force field and Gasteiger-Huckel charges.
To help position the bound bisbenzamidine compounds to the minor groove constituting AT rich base pairs, the model d(CCAAATTTGC) structure was aligned to the crystal structure of d(CGCGAATTCGCG)-pentamidine complex (ID64.pdb) [1] such that the central AATT sites were overlaid and the DNA sequence from 1D64 was removed. All the other model DNA duplexes were then aligned to this model such that crystal ligand is in the minor groove of the target sequences AAATTT, AAAGTTT, AAAGCTTT and GCGCGC. All the docking calculations were carried out using the standard protocols with Surflex-Dock, and 20 poses were retained for each compound based on the docking scores. Within the top 10 poses at least 4 to 9 poses were found to have identical or very similar poses. The standard deviations of the energies of these poses for most compounds were less than 0.5. A consensus score that combines multiple popular scoring functions, such as G_Score, PMF_Score, D_Score and ChemScore, was also calculated for each hit, and it is supposed to be more robust than any single function for evaluating ligand-receptor interaction and it has been validated in virtual high throughput screening [3]. Thus the total score with the best consensus score was used for ranking the binding affinities of benzamidine compounds to the double helix DNA with specific AT rich and GC sites.

Surflex-Dock Data
Docking scores of alkanediamide-linked bisbenzamidines and analogs to various DNA duplexes with specific central sequences are tabulated in Table S1.

Correlation Data
Correlation between the experiment pIC50 values in different cell lines and the docking score haven determined. The Pearson r values and the p values for statistical significance are tabulated in Table S2.

DNA-Ligand Binding Interactions
Two active compounds, 11 and 15, bound to the central -AAAGTTT-sites of 5′-d(CCAAAGTTTGC)-3′ duplex are depicted in Figure S2. In compound 11 one terminal amidine displayed a H-bond with O3′ of A5 and the other terminal amidine displayed a H-bond with O3′ of C17. In compound 15 one terminal amidine displayed two H-bonds with O4′ of A5 and N3 of A4, and the other terminal amidine displayed a H-bond with O3′ of A16. In addition, the alkanediamide nitrogens of the linkers in 11 formed two H-bonds with O2 of T7 and N3 of A5 and in 15 formed one H-bond with O2 of T8.