Revealing global stoichiometry conservation architecture in cells from Raman spectral patterns

Changes in molecular profiles in cells are often correlated and suggested to be effectively low dimensional. However, what kinds of biological principles entail such constraints remains elusive. Here, we measure Raman scattering light fromEscherichia colicells under diverse conditions, whose spectral patterns convey their comprehensive molecular composition. We reveal that dimension-reduced Raman spectra can predict condition-dependent proteome profiles. Quantitative analysis of the Raman-proteome correspondence characterizes a low-dimensional hierarchical stoichiometry-conserving proteome structure. The network centrality of each gene in the stoichiometry conservation relations is biologically significant, correlating with its essentiality and evolutionary conservation. Furthermore, stoichiometry-conserving core components obey growth law and ensure homeostasis across conditions, whereas peripheral stoichiometry-conserving components enable adaptation to specific conditions. Mathematical analysis reveals that the stoichiometric architecture is reflected in major changes in Raman spectral patterns. These results uncover coordination of global stoichiometric balance in cells and demonstrate that vibrational spectroscopy can decipher such biological constraints beyond statistical or machine-learning inference of cellular states.