Oral Presentation 24th Annual Lorne Proteomics Symposium 2019

Universal Solid-Phase Protein Preparation (USP3) for high-throughput clinical proteomics (#57)

Laura F Dagley 1 2 , Giuseppe Infusini 1 2 , Andrew I Webb 1 2
  1. The Walter and Eliza Hall Institute of Medical Research, Parkville, VIC, Australia
  2. Department of Medical Biology, University of Melbourne, Parkville, VIC, Australia

Selecting a sample preparation strategy for mass spectrometry-based proteomics is critical to the success of quantitative workflows, especially in the context of large clinical cohorts of patient samples where accuracy and reproducibility are of paramount importance. The factors that dictate the overall utility of sample preparation methodologies are balanced between efficiency, sensitivity and robustness versus speed, scalability and flexibility. Here we present a universal, solid-phase protein preparation (USP3) method which is rapid, robust and scalable, facilitating high-throughput protein sample preparation for bottom-up mass spectrometry (MS) analysis.

We have demonstrated the utility of the USP3 workflow in a patient cohort of children diagnosed with acute rheumatic fever and rheumatic heart disease, conditions which are highly prevalent in indigenous populations of Oceanian countries, including Australia. We obtained neat plasma and serum samples from children diagnosed with these conditions from ethnically-diverse populations including plasma samples from the Northern Territory of Australia (n=48; 26 ARF, 3 RHD, 7 healthy controls, 12 alternative diagnoses), serum samples from Fiji (n= 38; 9 ARF, 12 RHD, 17 healthy controls) and serum samples from New Caledonia (n=38; 28 ARF, 10 healthy controls). Neat plasma/serum samples were subjected to on-bead digestion (USP3 method) followed by high-resolution MS analysis on a Bruker Impact II UHR-QTOF instrument. An in-house developed XIC-based feature detection method was then used for MS1-based peptide quantification. To determine the most relevant peptides that can be developed into liquid-biopsy signatures, we employed a Random-forest based machine-learning algorithm. Utilising this machine learning approach, we identified a set of statistically-relevant peptide signatures that distinguish between patients with ARF from all other patient groups. A number of these peptides are unique to the Australian cohort whilst there are others that consistently stratify the patient groups across the diverse genetic ancestries. Efforts are now underway to translate these findings into a rapid clinical diagnostic test using the Noviplex™ Plasma Prep cards, with initial feasibility studies resulting in the identification of ~360 unique proteins from single-shot MS analyses.