Computational modeling-assisted discovery of the structural effects of phenylpropanoid derivatives on intracellular anti-Trypanosoma cruzi activity

Previous investigations have demonstrated that n-hexacosyl p-coumarate, a natural product isolated from the Brazilian plant Baccharis sphenophylla, exhibits activity against the extracellular (trypomastigote) forms of Trypanosoma cruzi. Structural modification of this compound led to the synthesis of the n-hexyl p-coumarate derivative, which showed significantly improved bioactivity, achieving an EC₅₀ value of 1.7 μM and a selectivity index of 71 against trypomastigote forms. However, when tested against intracellular amastigote forms, the compound displayed only moderate activity (EC₅₀ = 13 μM) and some cytotoxicity (CC₅₀ = 120 μM). To enhance efficacy and minimize toxicity, the double bond between C-7 and C-8 was reduced to a single bond, affording a series of phloretic acid derivatives ( 3a–3f ). These derivatives were synthesized via Fischer esterification and evaluated for their in silico ADMET properties, which indicated improved drug-likeness profiles. Notably, phloretic acid derivatives exhibited substantially lower cytotoxicity (CC₅₀ > 200 μM) compared with the p-coumaric acid analogues, while maintaining potent activity against amastigote forms (EC₅₀ ≈ 10 μM), comparable to the reference drug benznidazole. A structure–activity relationship (SAR) analysis based on machine learning was performed using a deep neural network (DNN) model combined with SHAP interpretation, allowing identification of key molecular fragments influencing predicted EC₅₀ values, as well as their correlation with molecular flexibility and lipophilicity. Collectively, these results underscore the potential of phloretic acid derivatives as lead candidates for the development of new anti- T. cruzi agents. • n -Hexyl p -coumarate showed potent bioactivity against trypomastigote forms of Trypanosoma cruzi but reduced potential against intracellular amastigotes. • To enhance efficacy and minimize toxicity, a series of floretic acid derivatives ( 3a–3f ) were synthesized and evaluated for their in silico ADMET properties, which indicated improved drug-likeness profiles. • Floretic acid derivatives ( 3a–3f ) exhibited substantially lower cytotoxicity (CC₅₀ > 200 μM) and displayed potent activity against amastigotes (EC₅₀ ≈ 10 μM), comparable to the reference drug benznidazole. • SAR analysis based on machine learning was performed using DNN and SHAP models, allowing identification of key molecular fragments.