Live Eco-AI, Electrophoresis-Correlative Data-Dependent Acquisition with Artificial Intelligence-Based Data Processing Democratizes Single-Cell Mass Spectrometry Proteomics

Abstract

Single-cell mass spectrometry (MS) recently emerged with unprecedented sensitivity to characterize cellular proteomes. Still, limited access to sensitive and fast mass spectrometers impedes its broad adoption, due chiefly to budget constraints. Here, we democratize single-cell MS to a budget-conscious alternative with contemporary performance. A custom-built inexpensive capillary electrophoresis (CE) platform electrophoresis-correlatively (Eco) sorted peptide m/z signals. Detection re-employed a decade-old, slower yet affordable quadrupole-quadrupole-orbitrap (Q-Exactive Plus, QE+, Thermo). Data-dependent acquisition (DDA) with narrow, 1.6-to-4 Th isolation tracked the m/z trends in [~]100% efficiency, outperforming the data-independent reference. The ensuing chimeric spectra were deciphered using artificial intelligence (AI) via CHIMERYS (Thermo). Live Eco-AI augmented our QE+ to [~]15 peptides/spectrum, on par with more advanced analyzers (Lumos and Exploris compared). 1 ng HeLa proteome digest gave 2,142 proteins, topping 969 on a contemporary LC Qq-OT-ion trap reference (Fusion Lumos, Thermo). From [~]250 pg, approximately a cell, 1,799 HeLa-proteins were detected in <15-min-effective electrophoresis. To establish proof of principle, Live Eco-AI was employed to profile 1,524 proteins among differentially fated single stem cells (50-to-75 micrometer diameter) in the Xenopus laevis (frog) blastula. The quantitative data revealed previously unknown proteome reorganization during differentiation into the dorsal and ventral lineages.