Attribution of PM_2.5-Induced Transcriptomic Perturbation to Toxic Components

Area of Science:

• Environmental Health Sciences
• Toxicology
• Molecular Biology

Background:

• Ambient fine particulate matter (PM2.5) poses health risks, but mass alone doesn't capture its full impact.
• Understanding the specific chemical components driving PM2.5 toxicity is crucial for effective risk assessment.

Purpose of the Study:

• To develop an interpretable chemotranscriptomic framework linking PM2.5 chemical components to molecular perturbations.
• To identify toxicity-relevant PM2.5 constituents beyond mass-based evaluations.

Main Methods:

• Collected and fractionated urban roadside and coastal PM2.5.
• Utilized advanced chemical characterization (LC/GC × GC-HRMS, ICP-MS) and transcriptomic profiling.
• Applied machine learning (Random Forest, SHAP) for feature attribution and mechanistic corroboration.

Main Results:

• Urban PM2.5 showed higher cytotoxicity, primarily driven by extractable fractions.
• Distinct cellular responses were observed: urban PM2.5 induced oxidative stress and cell cycle suppression, while coastal PM2.5 triggered immune signaling and apoptosis.
• Identified key chemical drivers including plasticizers, combustion products, copper (urban), and aged organics, nickel (coastal).

Conclusions:

• The chemotranscriptomic framework effectively attributes PM2.5 toxicity to specific chemical components.
• Prioritizing hazardous PM2.5 constituents based on mechanistic insights is more informative than mass-based assessments.
• This approach aids in identifying critical pollutants for targeted public health interventions.