A comparison of machine learning algorithms for chemical toxicity classification using a simulated multi-scale data model

Background  

Bioactivity profiling using high-throughput in vitro assays can reduce the cost and time required for toxicological screening of environmental chemicals and can also reduce the need for animal testing. Several public efforts are aimed at discovering patterns or classifiers in high-dimensional bioactivity space that predict tissue, organ or whole animal toxicological endpoints. Supervised machine learning is a powerful approach to discover combinatorial relationships in complex in vitro/in vivo datasets. We present a novel model to simulate complex chemical-toxicology data sets and use this model to evaluate the relative performance of different machine learning (ML) methods.     

Glucose-6-phosphatase as a cytochemical marker of endoplasmic reticulum in human leukocytes and platelets

Leukocytes and platelets, freshly isolated from normal human blood, were tested cytochemically for glucose-6-phosphatase (G-6-Pase) by a modified Wachstein-Meisel method. The enzyme was present in the endoplasmic reticulum (ER) and perinuclear cisternae of all five types of leukocytes and in the ER of platelets. The reaction product from the cytochemical test distinguished the ER from other intracellular membrane-limited cisternae (i.e., the smooth pinocytic tubules of monocytes and the surface-connected canalicular system of platelets) and thus is a valuable marker of the ER. The cytochemical test also showed that the ER of polymorphonuclear leukocytes (PMN), usually obscured by abundant granules in cells prepared for morphological examination, is more extensive than formerly appreciated. This is the first demonstration of G-6-Pase in human leukocytes. Its precise role in leukocyte metabolism can now be investigated.     

Beta-AMYLASE4, a noncatalytic protein required for starch breakdown, acts upstream of three active beta-amylases in Arabidopsis chloroplasts

This work investigated the roles of beta-amylases in the breakdown of leaf starch. Of the nine beta-amylase (BAM)-like proteins encoded in the Arabidopsis thaliana genome, at least four (BAM1, -2, -3, and -4) are chloroplastic. When expressed as recombinant proteins in Escherichia coli, BAM1, BAM2, and BAM3 had measurable beta-amylase activity but BAM4 did not. BAM4 has multiple amino acid substitutions relative to characterized beta-amylases, including one of the two catalytic residues. Modeling predicts major differences between the glucan binding site of BAM4 and those of active beta-amylases. Thus, BAM4 probably lost its catalytic capacity during evolution. Total beta-amylase activity was reduced in leaves of bam1 and bam3 mutants but not in bam2 and bam4 mutants. The bam3 mutant had elevated starch levels and lower nighttime maltose levels than the wild type, whereas bam1 did not. However, the bam1 bam3 double mutant had a more severe phenotype than bam3, suggesting functional overlap between the two proteins. Surprisingly, bam4 mutants had elevated starch levels. Introduction of the bam4 mutation into the bam3 and bam1 bam3 backgrounds further elevated the starch levels in both cases. These data suggest that BAM4 facilitates or regulates starch breakdown and operates independently of BAM1 and BAM3. Together, our findings are consistent with the proposal that beta-amylase is a major enzyme of starch breakdown in leaves, but they reveal unexpected complexity in terms of the specialization of protein function.     

Spatio-temporal patterns of mortality in Pemphigus populicaulis and P. populitransversus on cottonwoods

Summary Galls of the aphid species Pemphigus populitransversus and P. populicaulis from a single cottonwood in Clinton, New York were censused for mortality approximately every ten days during the summer of 1978. P. populitransversus had significantly higher survival than P. populicaulis. Both species of aphid showed considerably elevated mortality later in the season. Mortality was not spatially random, but neighboring galls had a tendency to die at close to the same time. This tendency to spatially autocorrelated mortality varied from time to time, being more pronounced earlier in the season than later. Later in the season, galls on the outer portions of branches tended to have a higher mortality rate than galls closer to the main trunk of the tree. This effect was highly significant in P. populicaulis, but only suggestive in P. populitransversus.The proportion of dead galls in samples of the two species taken at other times and places was consistent with the data from the censuses. In addition, the potential of two common predators, Anthocoris sp. and Leucopis sp., for producing autocorrelated mortality in Pemphigus was assessed in these collections. There was some association of Leucopis with regions of higher mortality, but this was not the case in Anthocoris. Leucopis tended to be found in galls of P. populicaulis more frequently than in P. populitransversus unlike Anthocoris, suggesting that some of the differential mortality between P. populitransversus and P. populicaulis might be due to this dipteran predator.     

Quinoline alkaloids as intercalative topoisomerase inhibitors

Quinoline alkaloids are abundant in the Rutaceae, and many have exhibited cytotoxic activity. Because structurally related antitumor alkaloids such as camptothecin and fagaronine are known to function as intercalative topoisomerase poisons, it is hypothesized that cytotoxic Stauranthus alkaloids may also serve as intercalative topoisomerase inhibitors. To test this hypothesis theoretically, ten Stauranthus quinoline alkaloids were examined for potential intercalation into DNA using a molecular docking approach. Four of the alkaloids (stauranthine, skimmianine, 3′,6′-dihydroxy-3′,6′-dihydrostauranthine, and trans-3′,4′-dihydroxy-3′,4′-dihydrostauranthine) were able to intercalatively dock consistently into DNA. In order to probe the intermolecular interactions that may be responsible for intercalation of these quinoline alkaloids, density functional calculations have been carried out using both the B3LYP and M06 functionals. M06 calculations indicated favorable π–π interactions between either skimmianine or stauranthine and the guanine–cytosine base pair. Furthermore, the lowest-energy face-to-face orientation of stauranthine with guanine is consistent with favorable dipole–dipole orientations, favorable electrostatic interactions, and favorable frontier molecular orbital interactions. Likewise, the lowest-energy face-to-face orientation of stauranthine with the guanine–cytosine base pair reveals favorable electrostatic interactions as well as frontier molecular orbital interactions. Thus, not only can quinoline alkaloids dock intercalatively into DNA, but the docked orientations are also electronically favorable.      

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.