Ke Zhang, Xingjian Chen, Francesco Monticolo, Salvatore Sorrentino, Haochun Huang, Claire Callahan, Yueqing Qiao, Judy Zhou, Sonia Brodowska, Styliani Sapantzi, Jianhuan Qi, Yinghan Wu, Thai Nam Son Dang, Yanwan Cao, Sumin Kang, Francesca Viggiani, Chia-Kang Ho, Yanxin Xu, Koseki J Kobayashi-Kirschvink, Thang Mung, Hemali Phatnani, Zhixun Dou, Jeon Woong Kang, Peter T C So, Jian Shu
Aging and tissue repair involve heterogeneous remodeling across transcriptional, biochemical and cellular dimensions, yet prevailing definitions rely on isolated molecular markers that obscure how these states co-evolve. Here we present RamanOmics, a multimodal framework integrating label-free hyperspectral Raman imaging with single-nucleus RNA sequencing and spatial transcriptomics to link biochemical states with transcriptional programs at single-cell spatial resolution. Applied to young and old mouse lung and skin, RamanOmics reveals tissue-specific programs: lung senescent cells are enriched for extracellular matrix remodeling and transforming growth factor-β signaling, whereas skin senescence is dominated by epidermal differentiation genes (Krt10, Lor and Sbsn). Across tissues, we identified a conserved lipid-linked Raman signature (1,131-1,135 cm-1) marking p21+ senescent cells and developed a machine learning-derived, multimodal barcode enabling nondestructive senescence identification in situ. In a mouse wound-healing model, RamanOmics reveals reactivation of epidermal differentiation genes (Krt10, Lor and Sbsn) in senescent cells, alongside increased lipid-associated Raman signatures. Together, RamanOmics provides a tissue-agnostic framework for scalable, multimodal profiling of cellular states.