Technical repeatability and efficiency in deep proteomics: a comparative study of various biological matrices

Understanding the technical repeatability of proteomic analyses is crucial for designing robust experimental studies. At the Proteomics Core Facility of the University of Athens, we emphasise the importance of conducting preliminary tests on different matrices to prove the feasibility of the chosen sample preparation method and ensure the consistency and reliability of results.

We analysed replicates of protein extracts from undepleted serum, CD138+ cell pellets, protein phenol phase samples after RNA extraction (PPP) of CD138+ cells, murine bronchoalveolar lavage (BAL) fluid, and murine heart tissue. Proteins from BAL and PPP samples were precipitated with solvents and pelleted by centrifugation. Following adaptations from the SPEED protocol (Doellinger et al., 2020), proteins were denatured with trifluoroacetic acid (TFA), reduced and alkylated, then digested overnight with trypsin. Purified peptides were separated onto a C18 reverse-phase column (nanoElute 2) during a 30-min elution gradient. For each matrix, we optimised the injected number of cells or nanograms of protein to maximise coverage and ensure scalability. Data was acquired on a timsTOF fleX (Bruker) using library-free Data-Independent Acquisition (DIA) Parallel Accumulation Serial Fragmentation. Data was analysed using DIA-NN.

Across the different sample types, we detected 400-5,000 protein groups, demonstrating high depth of analysis compared to other published studies. Serum, CD138+ cell pellet, and PPP samples showed robust repeatability with 81%, 81% and 81.5% of the 430, 4,800, and 3,600 protein groups detected below 20% coefficient of variations (CV), respectively. BAL samples were precipitated with ethanol, acetonitrile, or acetone, as well as with the SP3 method. Acetone precipitation yielded the best results regarding repeatability and protein identification, with an average of 1400 protein groups detected, 89% of which with CV <20%. Heart tissue analysis was much more challenging with no significant difference between TFA lysis and traditional RIPA buffer protein extraction. While more than 3,200 protein groups could be detected, only a quarter had a CV <20%.

These tests allow us to improve our protocols and bring knowledge as to the reliability of proteomics analyses. The use of TFA enhances biological safety and simplifies the preparation process. We have further streamlined our workflows with a 30-min gradient, significantly shorter than traditional proteomics protocols, achieving high throughput and scalable proteomics pipelines. Depth of analysis was achieved thanks to careful optimisation of DIA windows. Along with ensuring the feasibility of the analyses, understanding the unique properties of each matrix often has a significant impact on study design, sample selection or collection, and appropriateness of proteomics to answer research questions. We recommend performing such tests very early in any study planning to include proteomics.