A combined bioprocess and artificial intelligence framework for comparative evolutionary genomics of Rhodotorula mucilaginosa

Abstract

Oleaginous red yeasts have attracted considerable interest as sustainable microbial platforms for the simultaneous production of lipids and high-value carotenoids. In addition, understanding the evolutionary conservation of key biosynthetic enzymes can provide valuable insights for strain improvement and metabolic engineering. In the present study, Rhodotorula mucilaginosa, isolated from petroleum refinery wastewater, was evaluated for lipid and carotenoid production under batch and fed-batch cultivation in a 5-L bioreactor. Furthermore, the evolutionary conservation of acetyl-CoA carboxylase (ACC) and phytoene synthase (PSY), two key enzymes involved in lipid and carotenoid biosynthesis, was investigated through comparative genomic and phylogenetic analyses.

Fed-batch cultivation with glucose supplementation significantly enhanced biomass accumulation and metabolite production compared with batch operation. The maximum biomass concentration reached 4.37 ± 0.35 g L⁻¹, with lipids accounting for 57.22 ± 0.74% (w/w) of the dry biomass. Total carotenoid production attained 48.45 ± 2.44 mg L⁻¹ during the stationary phase, including 21.40 ± 1.55 mg L⁻¹ of β-carotene. Intracellular lipid and carotenoid accumulation was confirmed using Nile Red fluorescence microscopy, field emission scanning electron microscopy (FESEM), and Raman spectroscopy.

Comparative genomic and phylogenetic analyses revealed a high degree of conservation of ACC and PSY across Rhodotorula species. ACC exhibited strong conservation of functionally important catalytic regions, particularly the biotin carboxylase and carboxyltransferase domains containing the conserved GGHGT motif. Similarly, PSY retained the characteristic aspartate-rich DXXXD catalytic motifs essential for carotenoid biosynthesis, although moderate sequence variation was observed in terminal non-catalytic regions. Machine learning-assisted frameworks, including BetaAlign and Phyloformer, were further explored as emerging tools for enhancing sequence alignment, phylogenetic analysis, and evolutionary interpretation of lipid- and carotenoid-biosynthetic enzymes.

These results demonstrate the potential of R. mucilaginosa as an efficient microbial cell factory for the co-production of lipids and carotenoids. The observed conservation of key catalytic domains, together with sequence diversification in non-catalytic regions, provides new insights into the evolution of lipid and carotenoid biosynthesis in Rhodotorula and establishes a foundation for future metabolic engineering and industrial biotechnology applications.