Existing thermal shift-based mass spectrometry approaches are able to identify target proteins without chemical modification of the ligand, but they are suffering from complicated workflows with limited throughput. Herein, we present a new thermal shift-based method, termed matrix thermal shift assay (mTSA), for fast deconvolution of ligand-binding targets and binding affinities at the proteome level. In mTSA, a sample matrix, treated horizontally with five different compound concentrations and vertically with five technical replicates of each condition, was denatured at a single temperature to induce protein precipitation, and then, data-independent acquisition was employed for quick protein quantification. Compared with previous thermal shift assays, the analysis throughput of mTSA was significantly improved, but the costs as well as efforts were reduced. More importantly, the matrix experiment design allowed simultaneous computation of the statistical significance and fitting of the dose–response profiles, which can be combined to enable a more accurate identification of target proteins, as well as reporting binding affinities between the ligand and individual targets. Using a pan-specific kinase inhibitor, staurosporine, we demonstrated a 36% improvement in screening sensitivity over the traditional thermal proteome profiling (TPP) and a comparable sensitivity with a latest two-dimensional TPP. Finally, mTSA was successfully applied to delineate the target landscape of perfluorooctanesulfonic acid (PFOS), a persistent organic pollutant that is hard to perform modification on, and revealed several potential targets that might account for the toxicities of PFOS.

Cell surface is the primary site for sensing extracellular stimuli. The knowledge of the transient changes on the surfaceome upon a perturbation is very important as the initial changed proteins could be driving molecules for some phenotype. In this study, we report a fast cell surface labeling strategy based on peroxidasemediated oxidative tyrosine coupling strategy, enabling efficient and selective cell surface labeling within seconds. With a labeling time of 1 min, 2684 proteins, including 1370 (51%) cell surface-annotated proteins (cell surface/plasma membrane/extracellular), 732 transmembrane proteins, and 81 cluster of differentiation antigens, were identified from HeLa cells. By comparison with the negative control experiment using quantitative proteomics, 500 (68%) out of the 731 significantly enriched proteins (p-value < 0.05, ≥2-fold) in positive experimental samples were cell surface-annotated proteins. Finally, this technology was applied to track the dynamic changes of the surfaceome upon insulin stimulation at two time points (5 min and 2 h) in HepG2 cells. Thirty-two proteins, including INSR, CTNNB1, TFRC, IGF2R, and SORT1, were found to be significantly regulated (p-value < 0.01, ≥1.5-fold) after insulin exposure by different mechanisms. We envision that this technique could be a powerful tool to analyze the transient changes of the surfaceome with a good time resolution and to delineate the temporal and spatial regulation of cellular signaling.