Introduction

Single cell proteomics (SCP) is an emerging technique advancing our understanding of cellular protein heterogeneity. Hitherto, the community is pushed to report on the highest detected and quantified proteins, but it is important to note that the protein content of the desired sample is in the few-hundred picogram range and even a small amount of contamination can lead to large effects in the measured data. In this project, we aim to explore and identify the cause of potential contamination, by investigating a range of experimental parameters during cell isolation and sample preparation. Next, our aim is to assess technical and biological variability of SCP measurements as biological variability is key to assess single cell heterogeneity and is essential to understand molecular heterogeneity in cells.

Material and Methods

All experiments were performed on the CellenONE cell isolation platform and the investigated parameters include the different cell types, the target plate type (96-well plate and Cellenion EVO96 proteoCHIP), the use of laminar flow cabinet, as well as extensive cleaning of the CellenONE prior to usage. Next to dummy samples (zero cells dispensed in the well; N>=16), each plate also contained single cells (N>=36) and a few samples with higher cell counts up to 50 cells. All samples were processed and analyzed using a nanoElute2 LC system coupled to a timsTOF SCP-MS. Additionally, various data analysis software packages (Bruker ProteoScape, Spectronaut V18, and DIANN 1.8.1) have been evaluated as the best practice for analyzing a cohort that contains a varying number of cells per cohort.

Results

Preliminary data analysis has been performed on "0 cells" samples (treated identically to the single cell samples that were generated in parallel) from isolations of N2102Ep cells into either a standard 96-well plate or the CellenONE EVO96 proteoCHIP using both Spectronaut and DIANN. Overall, both tools showed that the usage of the EVO96 proteoCHIP resulted in lower contamination (57% and 97% decrease in Spectronaut and DIANN, respectively). While DIANN overall provided a lower identification of proteins, 7 (636 precursors) proteins compared to 284 (989 precursors) in Spectronaut. For more results see Figure. Further tests will include the bulk samples in the analysis runs to evaluate the effect of boosting (allow peptides that are identified in one sample, to be used in the search in other samples from the same cohort) and to filter the detected proteins based on the number of peptides associated with them to evaluate the reliability of the protein inference process as implemented in both Spectronaut and DIANN.

Conclusions

In conclusion, as SCP is gaining momentum, we hope with our obtained knowledge to raise awareness and eventually provide a general recommendation to the SCP community on how contamination can be avoided or minimized, and how best to account for any contamination in the data analysis process.