Mass spectrometry-based shotgun proteomics is a powerful tool to measure the abundance of many proteins in biological samples simultaneously. Proteomics experiments with bioinformatics analysis are routinely employed to identify biomarkers and discover associated biological complexes and pathways underpinning pathological mechanisms. With its high analyte specificity and potential quantitative accuracy, shotgun proteomics also has excellent potential to be adopted by research facilities and diagnostic laboratories for clinical sample analysis without the need to rely on antibodies1,2.
To prepare protein samples for shotgun proteomics analysis, proteins extracted from biological samples (e.g., cells and tissues) typically first need to be processed using lengthy protocols, including measuring the sample protein concentration, protein reduction, and alkylation, and enzymatic digestion into peptides. Moreover, proteins extracted in common lysis buffers containing detergents often require additional steps of buffer exchange or detergent removal before analysis because detergent can interfere with trypsin digestion and significantly degrade the performance of downstream liquid chromatography-tandem mass spectrometry (LC-MS/MS) analysis3. Peptides are typically further desalted, dried, and reconstituted in LC-MS/MS compatible solvents following enzymatic digestion. These protein biochemistry procedures can be labor-intensive and time-consuming. Thus, they continue to limit the throughput of proteomics workflows and contribute to the variability of acquired data4,5. Human errors and biases have been recognized as crucial factors affecting data variance and reproducibility6,7. To minimize human errors in mass spectrometry sample preparation workflows, automated pipetting robotic systems have been utilized to improve the throughput and reproducibility of protein identification and quantification from shotgun proteomics and targeted mass spectrometry analysis, where such advances have been hailed as instrumental for continuing the push for widespread adoption of proteomics technologies in critical research and clinical settings8,9,10,11,12,13. However, most existing protocols utilize robotic liquid handling platforms that require substantial investment and training, limiting their utility in many laboratories in the academic environment or otherwise with a limited budget.
This article describes a protocol that utilizes a low-cost, open-source robotic liquid handling system, the OT-2, to semi-automate a typical shotgun proteomics sample preparation workflow. The OT-2 has a lower cost than many other robotic liquid handling systems, and at the time of writing, costs approximately $5,000 US dollars. When factoring in the prices of different modules and labware, the total cost to set up experiments in this protocol at the time of writing is around $10,000, which renders it more affordable to a considerably broader set of laboratories over more expensive options. The OT-2 is compatible with open-source programming through Python scripts and offers great flexibilities in user-defined DIY protocol design. Using three in-house developed scripts, the protocols below cover executing a typical shotgun proteomics sample preparation workflow on the OT-2 station with an archetypical protein standard (bovine serum albumin; BSA) and a complex protein sample of a normal human heart lysate (Figure 1). The procedures for processing (1) a BSA sample and (2) a complex cardiac lysate sample are detailed in Protocol sections 1, 2, 5, 6 and 3, 4, 5, 6, respectively. Sera-Mag carboxylate-modified magnetic beads are utilized in single-pot solid-phase-enhanced sample preparation (SP3) to remove detergents and salts in the protein and peptide samples. Tryptic digests from bovine serum albumin and human heart proteins are further cleaned by SP3 beads and submitted for LC-MS/MS analysis. Mass spectra are then analyzed using the MaxQuant software for peptide and protein identification. Representative results performed by us show that the protocol achieves excellent technical coefficients of variation (CV) while saving bench time and is non-inferior to hand digest.