Fig. 6: Representative molecular structures among the major heteroatomic formula classes in polluted and clean air.
Molecular markers in polluted and clean air (OPLS; Fig. 5) were structurally characterized by MS2 spectral library matching (red structures; Level 2) or by MS2 in silico prediction (blue structures; Level 3). Representative example structures are shown to illustrate the results (see also Supplementary data 4 – Identifications). a Polluted air was distinguished by a mixture of nitrogen- and sulfur-containing compounds, and highly oxygenated organic molecules. Several of these structures corresponded to aromatic N-heterocycles (e.g. nitrophenols, imidazoles, and diazine derivatives), which are structures that have been previously linked to light-absorbing BrC. The three structures in red from polluted air (N2, N2O, N2O5) were confirmed Level 1 by authentic reference standards. b Molecular hallmarks of clean air were mono-, di-, and tri-oxygenated organic molecules (O-O3 class), many of which were predicted in silico (Level 3) or annotated on the GNPS spectral library (Level 2) as biogenic compounds. These included terpenes, such as derivatives of alpha-pinene (CH) and other plant volatiles. An interesting exception of a compound in clean air with higher oxygen content (O6). was confirmed as acetyl portentol (Level 1; Fig. S21), a polyketide produced by the marine lichens of Roccella sp. native to Indian coastal habitats. Another biogenic compound was predicted in silico (Level 3) as viscosumic acid, a sesquiterpene produced by the annual herbs of Polygonum sp. native to Nepal, Bangladesh, and north-eastern India.
