Advancing sustainable agriculture in acidic soils through artificial intelligence-driven functional microbiome mining and microbial-mediated crop resilience

Acidic soils, covering ~40% of the world’s arable land, pose severe constraints on crop productivity due to aluminum (Al) toxicity. Traditional approaches to studying microbial contributions to plant Al tolerance have been limited by the inability to efficiently isolate and characterize functional microorganisms from complex environmental samples. To address this, we developed an ​Artificial Intelligence-Assisted Raman-Activated Cell Sorting (AI-RACS)​​ system, which integrates single-cell Raman spectroscopy, optical tweezers, and AI-driven automation to enable high-throughput, label-free sorting of microbial cells based on their metabolic activity under stress conditions. Applied to acidic red soils, AI-RACS successfully isolated Al-tolerant strains by quantifying metabolic activity via deuterium oxide (D₂O) probing, outperforming conventional cultivation methods. These isolates were used to construct ​synthetic microbial communities (SynComs)​​ that enhanced rice resilience in acidic soils. In field trials, SynCom inoculation increased rice yield by ​26.36%​, reduced root Al accumulation by ​26.5%​, and improved phosphorus availability by solubilizing legacy soil phosphorus. Mechanistic studies revealed that microbial cooperation underpins SynCom efficacy: for instance, Pseudomonas sp. and Rhodococcus sp. exhibited enhanced Al tolerance via ​quinolone-mediated cross-feeding, where degradation of the signaling molecule HHQ reinforced cell walls and optimized metabolic activity under Al stress. Further research demonstrated that SynComs activate host plant adaptations by ​remodeling root cell walls. Specifically, microbes upregulated xyloglucan endotransglucosylase (XET) activity and brassinosteroid biosynthesis, reducing Al binding sites in roots and decreasing Al accumulation by ​47.5%​​. This synergy between microbial metabolic support and host cell wall modification highlights a novel pathway for mitigating Al toxicity. Our work establishes a scalable framework from AI-RACS-driven functional strain identification to SynCom application, that bridges microbiome ecology and crop resilience. These advances offer practical strategies for sustainable agriculture in acidic soils, leveraging microbial tools to enhance food security without relying on chemical amendments.