We designed an HTS assay to identify compounds capable of coordinating to the heme iron of CYP51. Following target-based screening and the further validation steps summarized in , we identified 185 high affinity hits and 57 partially overlapping hits with anti-T. cruzi
activity, including a dozen sub-micromolar inhibitors of T. cruzi
infection in mammalian cells. Hits identified in the primary two-wavelength screen were evaluated by cheminformatic analysis, spectral screening and cross-validation against T. cruzi
amastigotes. Informatics feedback served to maximize scaffold diversity and reduce the number of compounds subject to re-evaluation in the more laborious spectral mode. As a result, the two pools, of 185 and 57 hits, were structurally diverse with respect to the number, size and arrangement of rings, as well as the nature of the linkers and substituting groups constituting the compound frameworks (Tables S3 and S4
). In summary, although two-wavelength mode is more efficient in terms of throughput, it generated >60% false-positive hits. Although more time-consuming to read, analyze and tabulate, the spectral mode is a more comprehensive methodology than the two-wavelength mode in terms of efficiently filtering and ranking true positive hits. Overall, based on validation tests for representative compounds, we expect the set of hits identified in this work to contain a high percentage of true positive hits and to be a valuable resource for further development of Chagas Disease applications.
Flowchart diagram of screening and validation steps.
At the high hit rate we observed, we were unlikely to miss important chemotypes by excluding 102 compounds violating Lipinski rule of five from validation in the full spectral mode and cross-validation in the T. cruzi
assay. The process of optimization generally yields larger and more hydrophobic compounds, and thus it is industry practice to focus on smaller “lead-like” compounds to allow MW and logP to increase during lead development 
. On the other hand, a rationale for looking at hydrophobic hits in more detail is justified by the fact that the CYP51 inhibitors itraconazole and posaconazole do not comply with this rule. The potency of posaconazole in the treatment of invasive fungal infections is attributed to accumulation of drugs in the tissues of internal solid organs rather than to free levels of drug in plasma 
. In isolated cells and solid tissues, both posaconazole and itraconazole are mostly concentrated in cellular membranes, which are also the site of their molecular target, CYP51 
. Given that the potent in vivo
activity of posaconazole and other novel CYP51 inhibitors is attributed to their special pharmacokinetic properties, such as large volumes of distribution and long terminal half-life 
, it may be worth revisiting the 102 hits initially excluded by the Lipinski criteria.
The T. cruzi
–active pool contained both high affinity hits (score 4 or 5) and those with lower binding scores (), suggesting that low score compounds might have been acting via alternative mechanisms. To test this possibility, the binding and activity of the highest rated hit, Compound 1
, was confirmed in manual assays. Compound 1
, had activity an order of magnitude higher than any other hit identified in these studies (). At 100 nM concentration, Compound 1
suppressed sterol biosynthesis in T. cruzi
amastigotes more efficiently than posaconazole (), thus becoming a candidate for further evaluation in an animal model of infection. Cheminformatic searching in the SEA databases 
revealed a significant resemblance of the top hit to inhibitors of P450 drug targets involved in eicosanoid synthesis and fatty acid metabolism. For example, thromboxane-A synthase was highly ranked with regard to the top hit and the structurally related sub-micromolar hits 4
, as were the fatty acid ω-hydroxylases of the CYP4 family. Surprisingly, the lowest E-values were obtained for glutaminyl-peptide cyclotransferase, which is unrelated to CYP family by sequence or structure. While the ability of drugs to bind to unintended targets in the human host can lead to undesirable side effects, the same binding to a target in a pathogen infecting a human can cure infection. In the context of a deficit of drugs for neglected diseases, the unexpected promiscuity of CYP51 encountered in this work could be utilized in a “piggy-back” strategy 
for identifying a chemical starting point for anti-parasitic therapy development, while structure-activity relationships derived from parasite assays could lead to disease-specific clinical candidates 
The structural motif present in the cohort of antifungal azole drugs was apparent in both pools. For instance, hit 1, 2-(4-chlorophenoxy)-1-(4-chlorophenyl)-2-(1H-1,2,4-triazol-1-yl)ethanone (binding score 5) (Compound 13
), which ranked thirteenth in the anti-T. cruzi
activity assay, resembled miconazole (), one of the first broad-spectrum antimycotic agents and the conazole progenitor of orally active antifungal agents such as ketoconazole, itraconazole, posaconazole, fluconazole and voriconazole 
. Manual UV-vis titration confirmed that Compound 13
binds and saturates CYP51 at equimolar 0.5 µM concentrations, suggesting at least low nanomolar binding affinity (). According to the SciFinder online resource, compounds with close structural similarities to Compound 13
were patented in the 1970's as potent fungicides. To confirm a relationship between this structural motif and biological activity, we extracted membrane sterols from T. cruzi
amastigotes isolated from infected mouse myoblasts treated with Compound 13
. Consistent with the inhibition of CYP51, GC-MS analysis showed an increase in lanosterol and eburicol precursors and a concomitant decline in episterol, fecosterol and other 14-demethylated intermediates (). Compound 13
was effective against T. cruzi
amastigotes with an EC50
of 1.3 µM and was trypanocidal at 2.5 µM.
The shape of the posaconazole molecule defines its binding mode to CYP51. Its long “tail” group extends into the substrate-binding tunnel 
. In the mouth of the tunnel, posaconazole adopts alternative conformations and makes multiple points of contact with amino acid residues on the protein surface 
. Extensive interactions of posaconazole with catalytically non-essential residues in CYP51 are strikingly consistent with the pattern of drug resistance in fungi. The points of posaconazole contact are mutation hot spots in azole-resistant isolates of the pathogenic fungi Aspergillus fumigatus
and Candida albicans
. Some of these mutations confer cross-resistance to all azole drugs 
. The spread of drug resistance due to mutation of the target in response to therapy might be avoided in T. cruzi
if the binding properties of new inhibitors were improved using target structure as a guide. This concept has been successfully used for a panel of 9 FDA-approved HIV protease inhibitors developed with extensive use of structure-based drug design 
. In order to stay within a substrate envelope and focus hit-to-lead optimization chemistry on main-chain interactions, we are pursuing co-crystallization of low molecular weight hits with CYP51 for x-ray structure analysis. Compared to the larger posaconazole (MW 700 g/mol), the molecular weight of hits active against T. cruzi
ranged from 215 to 480, with the majority (39 of 57) below 360; Compounds 1
are 323 and 348 g/mol, respectively. Furthermore, with its “fragment-like” characteristics 
of the lowest molecular weight and logP among the sub-micromolar T. cruzi
hits, Compound 5
has good prospects in hit-to-lead optimization, as does one of the smallest high affinity spectral hits, 5,8-dibromoisoquinoline (Compound 45
in Table S3
; MW 287 g/mol), consisting only of an iron-coordinating heterocyclic module.
In conclusion, diversification of leads for CYP51 inhibitors should offer the medicinal chemist new choices in terms of chemical accessibility and prospects for lead optimization. Multiple leads lower the risk of drug attrition in the case of undesirable ADMET properties and also cover a patent-free space, one of the keys to the successful anti-Chagasic therapy of the future. In the course of this work we have (i) further developed a UV-based high throughput screening methodology for P450 enzymes; (ii) diversified the chemical scaffold space for CYP51 inhibitors; (iii) identified a potent T. cruzi inhibitor and a diverse array of low molecular weight hits with high affinity to CYP51 for hit-to-lead optimization; (iv) related CYP51 to other pharmacologic targets by computational ligand chemistry. This effort has allowed us to identify molecules already produced by pharmaceutical companies for future experimental testing against T. cruzi.