3D printed hardware for automation of proteomics sample preparation at the Meso-Scale

• Proteomics research can benefit from automated sample preparation • The Andrew+ was adapted for digestion and desalting using 3D printed hardware • The automated method’s lower recovery was rectifiable by loading more material • The Andrew+ automated methods are recommended for standard sample applications Mass spectrometry-based proteomics is the dominant method for measuring peptides and proteins from complex mixtures. In bottom-up approaches, proteins are digested or proteolyzed prior to LC-MS/MS analysis. Peptides are fragmented, and proteins are inferred via peptide spectral matching (PSM). The throughput of this process is surprisingly low; a proteomics core facility might analyse <20 samples/day per instrument using UHPLC-MS/MS. Because of this, automation in proteomics is rare, and virtually all preparation is performed by hand. We developed 3D printed hardware and automated sample preparation modules for a lower-cost Andrew Alliance pipetting robot. The robot operates on simple principles, using traditional pipettes and follows protocols closely resembling manual preparation. Here, we present modular protocols for the major techniques in proteomics preparation: in-solution and S-Trap digestion; Tip and solid-phase extraction (SPE) based desalting. Both approaches yield dense protein identifications from complex proteomes. Automated samples had high reproducibility: ∼60% of proteins identified from in-solution and S-Trap digested samples had a measured CV of ≤20%. In contrast, 52% of in-solution digested and 63% of S-Trap digested of proteins identified from manually prepared samples had CVs ≤20%. Automated sample digestion and tip-based desalting had reduced ≅ 70% and 40% quantitative yield respectively compared to manual preparation according to the protein label-free quantification (LFQ). Increasing injection amount to normalize the yield restored protein and peptide identifications which demonstrates the differences between manual and automated methods were predominantly due to reduced recovery. Overall, automation of bottom-up proteomics sample preparation at the meso-scale offers increased reproducibility in non sample-limited applications.