Systematic evaluation of the performance metrics for spatial proteomics workflow and its application on colorectal cancer samples

Recent advances in mass spectrometry instrumentation have led to a great increase in sensitivity, thereby facilitating the analysis of single cells and other low input material sample types. Laser capture microdissection (LCM), where spatially-defined regions of interests (ROIs), such as individual or populations of cells, are extracted from a biospecimen under the microscope, generates highly-defined samples of low input material. Visual proteomics, which combines high sensitivity proteomics technology and high precision LCM sampling has been shown by pioneers in the field to yield differences at a spatial resolution within a single tissue slice; however, it remains a niche technique and is yet to be widely adapted and applied in an industry setting.

Here, we investigate the basic performance metrics of visual proteomics, including sensitivity and reproducibility of the pipeline. First, we optimized the gradient and acquisition settings of the LC-MS/MS setup using a serial dilution of commercially available HeLa digest. Samples were measured in data independent acquisition (DIA) mode using the Bruker timsTOF Ultra mass spectrometer. We identified on average more than 5,000 protein groups from 250 pg of standard HeLa digest using a library free approach in Spectronaut. Next, we assessed the robustness, scalability, and minimal sample limit of a typical LCM workflow. Homogenous areas of murine liver FFPE tissue ranging from 150,000 µm2 to 1,250 µm2 were laser capture dissected using a Leica LMD7 microscope. We identified on average approximately 1,500 protein groups with a CV of 20%. As proof of concept, we applied our optimized workflow to analyze proteomic differences in colorectal cancer tumor tissue and near adjacent to tumor areas from clinical samples.

Finally, we discuss the common practical issues typically associated with such cutting-edge technology. We highlight issues valuable to the community, from sample preparation to data acquisition and analysis. For example, classical bottom-up proteomics commonly involves system performance checks using HeLa or K562 tryptic peptide digests, whereas an equivalent control standard for visual proteomics currently does not exist. Due to the differences in scale and composition between classical MS and LCM-derived samples, a customized performance test for proper quality management in an industrial environment is a must. Scalability requirement provides another challenge for industry. The translation of systematic image analysis through laser capture to LC-MS/MS is a major bottle neck that needs to be overcome to reach the throughput required for a commercial setting. Ultimately, it is important to set these standards correctly as early as possible to ensure that the data produced matches the rapid technological advancements currently seen in this emerging field.