ML-AMPs designed through machine learning show antifungal activity against C. albicans and therapeutic potential on mice model with candidiasis

Abstract

Aims

C. albicans resistant strains have led to increasingly severe treatment challenges. Antimicrobial peptides with low resistance-inducing propensity for pathogens have been developed. A series of antimicrobial peptides de novo designed through machine learning by our research team were named ML-AMPs. In the present research, the antifungal activity of ML-AMPs against C. albicans and its therapeutic potential on Candidiasis mice model were studied.

Main methods

MTT methodology was performed to measure the minimum inhibitory concentrations. Absorbance photometry was utilized to evaluate the erythrocyte toxicity. Optical microscopy was operated to observe C. albicans hyphae. Crystal violet staining was employed to assess biofilm inhibition and reduction. Colony counting was performed to determine the time-kill kinetics. Scanning electron microscopy and fluorescent staining were used to investigate the underlying mechanism of antifungal action. Candidiasis mice model was established to evaluate the in vivo efficacy of ML-AMP2.

Key findings

ML-AMPs exhibited strong anti-Candida activity, with minimum inhibitory concentrations against C. albicans ranging from 3.85 to 12.37 μg/mL. Notably, they exhibited robust fungicidal effects on fluconazole-resistant C. albicans. Moreover, they exhibited fast-killing kinetics, as well as low resistance potential. Additionally, ML-AMPs could effectively inhibit the formation of mycelium and biofilm, and more prominently, their ability to reduce biofilm was higher than that of fluconazole. ML-AMPS increased the permeability of C. albicans cell membrane and induced ROS accumulation. Among ML-AMPs, ML-AMP2 performed the best, which promoted the recovery of Candidiasis mice model.

Significance

ML-AMP2 holds great promise as a candidate molecule for effectively treating drug-resistant C. albicans infections.