Mining stress-tolerant microorganisms from complex microbiomes is pivotal for the development of robust microbial chassis. However, conventional culture-first methods were laborious, low throughput, and inefficient for high-performance cells. Here, we developed and applied a high-throughput microfluidic optical tweezers-based Raman-activated cell sorting (RACS) system coupled with deuterium oxide (D₂O)-labelled single-cell Raman spectroscopy (SCRS). Leveraging a high screening throughput of ∼ 2,400 cells/h and a sorting accuracy of 91.3%, we successfully and efficiently enriched highly ethanol-tolerant cells from pit mud microbiomes. In a single sorting run, the system enriched 177 highly metabolic-active cells under ethanol stress before cultivation. Targeted cultivation on MRS medium yielded 6 isolates, all showing strong tolerance to 8% (v/v) ethanol in a 7 h SCRS-based assessment, whereas conventional culture-first screening achieved only 2 out 9 (22.2%) success. Genome sequencing and strain-specific transcriptomic profiling further provided molecular support for the ethanol-tolerant phenotypes of Lactiplantibacillus plantarum F4 (Raman Tolerance Index = 85.1 ± 3.41%) and Staphylococcus epidermidis F5 (RTI = 62.2 ± 1.09%). These molecular responses support the physiological relevance of the Raman screening signal. Overall, this integrated workflow achieved a 4.5-fold improvement in enrichment, and a 6.86-fold increase in assessment efficiency compared with conventional methods. Therefore, by sorting target cells based on metabolic activity in a screen-before-culture manner, D₂O-RACS is a powerful and versatile platform for efficient mining of stress-tolerant cells.

The filamentous fungus Aspergillus cristatus is the principal fermentative microorganism that defines the quality of Fuzhuan brick tea (FBT). However, its specific enzymatic repertoire has not been systematically characterized at the genome level, limiting the broader biotechnological deployment of this fungus. Here we combined whole-genome sequencing, comparative genomic profiling against four well-characterized Aspergillus species (Aspergillus niger, Aspergillus oryzae, Aspergillus nidulans, and Aspergillus fumigatus), CAZyme annotation, and an empirical 19-enzyme activity panel benchmarked against a 50-strain Aspergillus reference set to define the enzymatic specificity of A. cristatus strain CCNH008. The 28.1 Mb hybrid assembly was the most compact genome among the five compared species and encoded the smallest CAZyme repertoire of the compared genomes, with a strongly restrained pectinolytic complement. Enzymatic profiling revealed a distinctive phenotype characterized by moderate cellulolytic capacity, marked polyphenol oxidase and protease activities, and a genotype–phenotype discordance in which pectinase activity ranks among the highest in the 50-strain panel despite a limited predicted pectinolytic gene complement. Comparative genomic profiling positioned CCNH008 between the lignocellulose-degrading specialist A. niger and the starch-adapted food fungus A. oryzae, consistent with adaptation to the unique chemical environment of tea leaves. The safety profile was assessed by in silico PHI/DFVF virulence factor screening and in vitro antifungal susceptibility testing. These findings establish a genomic and enzymatic foundation for the rational exploitation of A. cristatus in tea fermentation and related food biotechnology applications.