Metabolic perturbations in the edible plant Houttuynia cordata Thunb. induced by microplastic pollution and development of a rapid detection method.

The widespread use of agricultural plastic products and plastic packaging has led to microplastics (MPs) becoming a new class of environmental pollutants, posing threats to both human health and plant growth. However, current research on MPs' impact on plants predominantly focuses on yield and morphological traits, while the mechanisms by which MPs affect volatile flavor compounds in plants remain poorly understood. Additionally, existing detection methods are often hampered by cumbersome sample preparation procedures and high costs. This study employed Houttuynia cordata Thunb., a traditional culinary herb and edible vegetable widely used in Asian cuisine, prized for its distinctive "fishy" aroma-as a model plant. Using headspace solid-phase microextraction coupled with gas chromatography-mass spectrometry (HS-SPME-GC-MS), we systematically investigated how microplastics (MPs) stress of varying polymer types (LDPE, HIPS), concentrations (0.5-50 mg·L-1), and particle sizes influences the accumulation of volatile compounds in rhizomes and leaves. Concurrently, we developed a novel rapid detection strategy integrating offline pyrolysis, portable mass spectrometry, and machine learning algorithms. A total of 46 volatile compounds were identified and relatively quantified. Multivariate statistical analysis revealed that MPs concentration is the primary driver of changes in the volatile metabolic profile of H. cordata, with the metabolic response in rhizomes being significantly stronger than in leaves. As MPs concentration increased, key terpenoid compounds such as bornyl acetate in rhizomes decreased significantly, while aldehydes/ketones like 2-undecanone showed marked enrichment. Furthermore, the Gradient Boosting discriminative model achieved an accuracy rate of 97.86% in distinguishing between uncontaminated and MPs-contaminated samples. The Logistic Regression model accurately identified PE vs. HIPS contamination types. Moreover, the XGBoost model successfully provided preliminary warnings for low (0.1 mg·g-1), medium (0.5 mg·g-1), and high (1 mg·g-1) contamination levels. SHAP interpretability analysis confirmed that characteristic ions such as m/z 104 (styrene monomer) and m/z 140 (decene) were key contributors to model predictions. This study not only elucidates the perturbation patterns of key volatile components in H. cordata under MPs stress, highlighting the heightened sensitivity of rhizomes as direct contact sites but also provides a cost-effective, high-throughput approach for rapid screening and tiered early warning of MPs contamination without requiring complex sample pretreatment. This work offers critical scientific and technical support for ensuring the safety of edible plants and establishing relevant risk assessment standards.