Refinement of paramagnetic bead-based digestion protocol for automatic sample preparation using an artificial neural network

Sergio Ciordia; Fátima Milhano Santos; João M L Dias; José Ramón Lamas; Alberto Paradela; Gloria Alvarez-Sola; Matías A Ávila; Fernando Corrales

doi:10.1016/j.talanta.2024.125988

Refinement of paramagnetic bead-based digestion protocol for automatic sample preparation using an artificial neural network

Talanta. 2024 Mar 23:274:125988. doi: 10.1016/j.talanta.2024.125988. Online ahead of print.

Authors

Sergio Ciordia¹, Fátima Milhano Santos¹, João M L Dias², José Ramón Lamas¹, Alberto Paradela¹, Gloria Alvarez-Sola³, Matías A Ávila³, Fernando Corrales⁴

Affiliations

¹ Functional Proteomics Laboratory, Centro Nacional de Biotecnología, CSIC, Calle Darwin 3, Campus de Cantoblanco, 28049, Madrid, Spain.
² Department of Medical Genetics, University of Cambridge, Cambridge, United Kingdom; Early Cancer Institute, University of Cambridge, Cambridge, United Kingdom.
³ Hepatology Laboratory, Solid Tumors Program, Center for Applied Medical Research (CIMA), University of Navarra, 31008, Pamplona, Spain; National Institute for the Study of Liver and Gastrointestinal Diseases (CIBERehd, Carlos III Health Institute), 28029, Madrid, Spain; IdiSNA, Navarra Institute for Health Research, 31008, Pamplona, Spain.
⁴ Functional Proteomics Laboratory, Centro Nacional de Biotecnología, CSIC, Calle Darwin 3, Campus de Cantoblanco, 28049, Madrid, Spain. Electronic address: fcorrales@cnb.csic.es.

PMID: 38569368
DOI: 10.1016/j.talanta.2024.125988

Abstract

Despite technological advances in the proteomics field, sample preparation still represents the main bottleneck in mass spectrometry (MS) analysis. Bead-based protein aggregation techniques have recently emerged as an efficient, reproducible, and high-throughput alternative for protein extraction and digestion. Here, a refined paramagnetic bead-based digestion protocol is described for Opentrons® OT-2 platform (OT-2) as a versatile, reproducible, and affordable alternative for the automatic sample preparation for MS analysis. For this purpose, an artificial neural network (ANN) was applied to maximize the number of peptides without missed cleavages identified in HeLa extract by combining factors such as the quantity (μg) of trypsin/Lys-C and beads (MagReSyn® Amine), % (w/v) SDS, % (v/v) acetonitrile, and time of digestion (h). ANN model predicted the optimal conditions for the digestion of 50 μg of HeLa extract, pointing to the use of 2.5% (w/v) SDS and 300 μg of beads for sample preparation and long-term digestion (16h) with 0.15 μg Lys-C and 2.5 μg trypsin (≈1:17 ratio). Based on the results of the ANN model, the manual protocol was automated in OT-2. The performance of the automatic protocol was evaluated with different sample types, including human plasma, Arabidopsis thaliana leaves, Escherichia coli cells, and mouse tissue cortex, showing great reproducibility and low sample-to-sample variability in all cases. In addition, we tested the performance of this method in the preparation of a challenging biological fluid such as rat bile, a proximal fluid that is rich in bile salts, bilirubin, cholesterol, and fatty acids, among other MS interferents. Compared to other protocols described in the literature for the extraction and digestion of bile proteins, the method described here allowed identify 385 unique proteins, thus contributing to improving the coverage of the bile proteome.

Keywords: Artificial neural network; Automation; Bile; Opentrons; Proteomics; SP3; Sample preparation.