Automation of peptide desalting for proteomic liquid chromatography - tandem mass spectrometry by centrifugal microfluidics

For large-scale analysis of complex protein mixtures, liquid chromatography - tandem mass spectrometry (LC-MS/MS) has been proven to be one of the most versatile tools due to its high sensitivity and ability to both identify and quantify thousands of proteins in a single measurement. Sample preparation typically comprises site-specific cleavage of proteins into peptides, followed by desalting and concomitant peptide enrichment, which is commonly performed by solid phase extraction. Desalting workflows may include multiple liquid handling steps and are thus error prone and labour intensive. To improve the reproducibility of sample preparation for low amounts of protein, we present a centrifugal microfluidic disk that automates all liquid handling steps required for peptide desalting by solid phase extraction (DesaltingDisk). Microfluidic implementation was enabled by a novel centrifugal microfluidic dosing on demand structure that enabled mapping multiple washing steps onto a microfluidic disk. Evaluation of the microfluidic disk was performed by LC-MS/MS analysis of tryptic HEK-293 eukaryotic cell peptide mixtures desalted either using the microfluidic disk or a manual workflow. A comparable number of peptides were identified in the disk and manual set with 19 775 and 20 212 identifications, respectively. For a core set of 10 444 peptides that could be quantified in all injections, intensity coefficients of variation were calculated based on label-free quantitation intensities. The disk set featured smaller variability with a median CV of 9.3% compared to the median CV of 12.6% for the manual approach. Intensity CVs on protein level were lowered from 5.8% to 4.2% when using the LabDisk. Interday reproducibility for both workflows was assessed by LC-SRM/MS analysis of samples that were spiked with 11 synthetic peptides of varying hydrophobicity. Except for the most hydrophilic and hydrophobic peptides, the average CV was lowered to 3.6% for the samples processed with the disk compared to 7.2% for the manual workflow. The presented centrifugal microfluidic DesaltingDisk demonstrates the potential to improve reproducibility in the sample preparation workflow for proteomic mass spectrometry, especially for application with limited amount of sample material.

Tryptic digestion of human serum for proteomic mass spectrometry automated by centrifugal microfluidics

Mass spectrometry has become an important analytical tool for protein research studies to identify, characterise and quantify proteins with unmatched sensitivity in a highly parallel manner. When transferred into clinical routine, the cumbersome and error-prone sample preparation workflows present a major bottleneck. In this work, we demonstrate tryptic digestion of human serum that is fully automated by centrifugal microfluidics. The automated workflow comprises denaturation, digestion and acidification. The input sample volume is 1.3 μl only. A triplicate of human serum was digested with the developed microfluidic chip as well as with a manual reference workflow on three consecutive days to assess the performance of our system. After desalting and liquid chromatography tandem mass spectrometry, a total of 604 proteins were identified in the samples digested with the microfluidic chip and 602 proteins with the reference workflow. Protein quantitation was performed using the Hi3 method, yielding a 7.6% lower median intensity CV for automatically digested samples compared to samples digested with the reference workflow. Additionally, 17% more proteins were quantitated with less than 30% CV in the samples from the microfluidic chip, compared to the manual control samples. This improvement can be attributed to the accurate liquid metering with all volume CVs below 1.5% on the microfluidic chip. The presented automation solution is attractive for laboratories in need of robust automation of sample preparation from small volumes as well as for labs with a low or medium throughput that does not allow for large investments in robotic systems.