Applied Machine Learning Toward Drug Discovery Enhancement: Leishmaniases as a Case Study.

Drug discovery (DD) research is a complex field with a high attrition rate. Machine learning (ML) approaches combined to chemoinformatics are of valuable input to this field. We, herein, focused on implementing multiple ML algorithms that shall learn from different molecular fingerprints (FPs) of 65 057 molecules that have been identified as active or inactive against Leishmania major promastigotes. We sought to build a classifier able to predict whether a given molecule has the potential of being anti-leishmanial or not. Using the RDkit library, we calculated 5 molecular FPs of the molecules. Then, we implemented 4 ML algorithms that we trained and tested for their ability to classify the molecules into active/inactive classes based on their chemical structure, encoded by the molecular FPs. Best performers were random forest (RF) and support vector machine (SVM), while atom-pair and topology torsion FPs were the best embedding functions. Both models were further assessed on different stratification levels of the dataset and showed stable performances. At last, we used them to predict the potential of molecules within the Food and Drug Administration (FDA)-approved drugs collection to present anti-Leishmania effects. We ranked these drugs according to their anti-Leishmanial probability and obtained in total seven anti-Leishmania agents, previously described in the literature, within the top 10 of each model. This validates the robustness of the approach, the algorithms, and FPs choices as well as the importance of the dataset size and content. We further engaged these molecules into reverse docking experiments on 3D crystal structures of seven well-studied Leishmania drug targets and could predict the molecular targets for 4 drugs. The results bring novel insights into anti-Leishmania compounds.

The In Silico Identification of Potential Members of the Ded1/DDX3 Subfamily of DEAD-Box RNA Helicases from the Protozoan Parasite Leishmania infantum and Their Analyses in Yeast.

DEAD-box RNA helicases are ubiquitous proteins found in all kingdoms of life and that are associated with all processes involving RNA. Their central roles in biology make these proteins potential targets for therapeutic or prophylactic drugs. The Ded1/DDX3 subfamily of DEAD-box proteins is of particular interest because of their important role(s) in translation. In this paper, we identified and aligned the protein sequences of 28 different DEAD-box proteins from the kinetoplast-protozoan parasite Leishmania infantum, which is the cause of the visceral form of leishmaniasis that is often lethal if left untreated, and compared them with the consensus sequence derived from DEAD-box proteins in general, and from the Ded1/DDX3 subfamily in particular, from a wide variety of other organisms. We identified three potential homologs of the Ded1/DDX3 subfamily and the equivalent proteins from the related protozoan parasite Trypanosoma brucei, which is the causative agent of sleeping sickness. We subsequently tested these proteins for their ability to complement a yeast strain deleted for the essential DED1 gene. We found that the DEAD-box proteins from Trypanosomatids are highly divergent from other eukaryotes, and consequently they are suitable targets for protein-specific drugs.

Infection of Human Neutrophils With Leishmania infantum or Leishmania major Strains Triggers Activation and Differential Cytokines Release.

Leishmaniases are neglected diseases, caused by intracellular protozoan parasites of the Leishmania (L.) genus. Although the principal host cells of the parasites are macrophages, neutrophils are the first cells rapidly recruited to the site of parasites inoculation, where they play an important role in the early recognition and elimination of the parasites. The nature of early interactions between neutrophils and Leishmania could influence the outcome of infection. Herein we aimed to evaluate whether different Leishmania strains, responsible for distinct clinical manifestations, could influence ex vivo functional activity of neutrophils. Human polymorphonuclear leukocytes were isolated from 14 healthy volunteers and the ex vivo infection of these cells was done with two L. infantum and one L. major strains. Infection parameters were determined and neutrophils activation was assessed by oxidative burst, degranulation, DNA release and apoptosis; cytokine production was measured by a multiplex flow cytometry analysis. Intracellular amastigotes were rescued to determine Leishmania strains survival. The results showed that L. infantum and L. major promastigotes similarly infected the neutrophils. Oxidative burst, neutrophil elastase, myeloperoxidase activity and apoptosis were significantly increased in infected neutrophils but with no differences between strains. The L. infantum-infected neutrophils induced more DNA release than those infected by L. major. Furthermore, Leishmania strains induced high amounts of IL-8 and stimulated the production of IL-1ß, TNF-α, and TGF-ß by human neutrophils. We observed that only one strain promoted IL-6 release by these neutrophils. The production of TNF-α was also differently induced by the parasites strains. All these results demonstrate that L. infantum and L. major strains were able to induce globally a similar ex vivo activation and apoptosis of neutrophils; however, they differentially triggered cytokines release from these cells. In addition, rescue of intracellular parasites indicated different survival rates further emphasizing on the influence of parasite strains within a species on the fate of infection.

Development of metal hydroxide nanoparticles from eggshell waste and seawater and their application as flame retardants for ethylene-vinyl acetate copolymer (EVA).

Recently, calcium hydroxide and magnesium hydroxide nanoparticles have attracted a lot of research interest in different sectors: food, packaging, health, automotive construction and food application. In the present study, we report development of bio-material calcium hydroxide nanoparticles (Ceg-Ca(OH)2), obtained from chicken eggshell collected from the food industries as well as magnesium hydroxide nanoparticles obtained from seawater (Seaw-Mg(OH)2). The flame-retardant behavior of Ethylene-Vinyl Acetate copolymer (EVA) containing different blends of Ceg-Ca(OH)2 and Seaw-Mg(OH)2 nanoparticles has been evaluated using cone calorimeter. Our results showed the interest of combining both nanoparticles. In fact, the partial substitution of small Seaw-Mg(OH)2 content (10 wt%) by Ceg-Ca(OH)2 enables further reduction of pHRR from 251 to 206 kW/m2 without any reduction of the composite time to ignition (52 s). Furthermore, the partial substitution of 40 wt% Seaw-Mg(OH)2 nanoparticles by Ceg-Ca(OH)2 enables high flame retardant effect as well as the generation of cohesive residue.