The accessibility of blood, and hence plasma, from patients makes it a popular source of samples for clinical research. Routine quantitative analysis on these samples is primarily conducted using immunoassays, but as with any analysis, in addition to its benefits—such as simplicity, specificity, and rapid response—there are limitations and drawbacks inherent to the platform. For immunoassays, the desired antibodies are often unavailable or may lack sufficient specificity. Only a limited number of analytes can be measured simultaneously and the assays can only be used for validation, but not discovery, of biomarkers. Nanoflow liquid chromatography-mass spectrometry (nanoLC-MS)-based proteomics is emerging as a promising and powerful technique to overcome this latter issue, as it can be used in both biomarker discovery and biomarker validation. Despite this potential strength for use in plasma proteomics, nanoLC-MS-based methods remain arduous to implement in a reproducible, robust, and high-throughput manner in large patient cohorts.
Quantifying plasma proteins
Recent advances in nanoLC-MS-based proteomics have helped to overcome these significant challenges, making plasma proteomics more accessible to routine laboratories. However, the large dynamic range of protein abundance in plasma and nanoLC-MS’s limited implementation in large cohort studies prevents nanoLC-MS from becoming the go-to plasma proteomics method. A recent study hopes to overcome these barriers by developing a robust and streamlined shotgun plasma proteomics workflow for analysis of large sample cohorts.1
Very short nanoLC-MS/MS gradients were coupled with fast MS-based identification. The process was then optimized to allow high-throughput generation of plasma proteome profiles for clinical samples. This approach can provide valuable insights into the health and disease states of patients, in a meaningful timeline.
Rapid, high-throughput methods
In this study, plasma samples from 192 patients with acute inflammation were collected and depleted for the top 12 most abundant proteins. Tryptic peptides (100 ng per sample) were separated using a 100-runs-per-day LC method (Evosep One, Evosep Biosystems) for subsequent delivery to a trapped ion mobility spectrometry (TIMS) quadrupole time-of-flight (QTOF) mass spectrometer (timsTOF Pro), with parallel accumulation-serial fragmentation (PASEF, Bruker Daltonics).
TIMS, when combined with PASEF, offers uncompromised speed, sensitivity, and resolution for shotgun proteomics. TIMS is an ion mobility–based gas phase separation technique, which resolves sample complexity by taking advantage of the intrinsic property of collision cross-sectional values (CCS) in addition to high-performance liquid chromatography (HPLC) separation, which is based on hydrophobicity. This is followed by accurate mass-to-charge determination of the precursor ion and its collision-induced dissociation (CID)-produced fragment ions (MS/MS). PASEF principally adds a time-focusing aspect which boosts sensitivity as the ions are trapped (bunched according to their ion mobilities) and then scanned. The technique also improves the confidence in peptide identification for low-level signals at the same false-discovery rates (FDR).
By optimizing data acquisition for short gradients (11.5 min total sample injection to injection time) and applying PASEF, peptides were sequenced at speeds of >100 Hz with a high MS1 sampling rate for accurate quantification. The MS2 spectra quality of the low-abundance peptides can be increased by selecting them several times. Data was collected over an m/z range of 100 to 1700 for MS and MS/MS on the timsTOF Pro using an accumulation time and ramp time of 100 milliseconds.
After every 10th run, a quality control (QC) sample of pooled, peptide-digested plasma proteins was included to monitor LC-MS/MS performance over time, resulting in a total analysis time of 51 hours. 212 LC-MS/MS runs were submitted post acquisition for data analysis (PEAKS Studio, Bioinformatics Solutions), and close to 200 proteins were identified with quantification values across all of the QC runs, showing the robustness of the method. The patient sample data showed higher relative variability in the number of quantified protein groups (coefficient of variation [CV] = 20.2%) (Figure 1A) compared to the QC runs (CV = 3.5%). This was attributed to the process of blood collection (which is known to activate the coagulation cascade), processing of the blood, and biological variability. The low variability of the QC standards across the study displays the stability and reproducibility of the LC-MS/MS workflow.
Improving specificity of results
It has previously been demonstrated that peptide identifications from MS/MS spectra can be transferred between runs by matching retention time, m/z and intensity,2 a promising approach which boosts the number of identifications and lowers the number of missing values between runs. However, specificity can be low, as the 3D matching applied in traditional approaches without ion mobility leaves room for false-positive identifications in matched runs. Sophisticated, unique TIMS capabilities on modern mass spectrometers provide the possibility to acquire ion mobility information (1/K0) and calculate peptide CCS values. CCS values are physical parameters of peptides, and recent results show that they can be determined extremely reproducibly with TIMS (<1% relative deviation),3 which provides a new dimension of confidence in results.
The present study used 1/K0 to employ a 4D matching approach in PEAKS Studio, which includes an additional dimension (TIMS ion mobility) to match peptide features in retention time, 1/K0 and m/z for subsequent transfer of identifications from some runs to feature intensities matching in other runs that miss MS/MS identifications. Application of the 4D alignment resulted in a dramatic increase of quantified plasma proteins from 188 to 500 (median values) per sample in the QC standard dataset (Figure 1B). Additionally, the number of quantified proteins in the patient sample dataset was significantly higher when applying the 4D (median 478), and the relative variability is similar (CV = 16%) to the dataset without matching (Figure 1B).
By looking at the proteins quantified across the study with and without 4D matching, it is apparent that there is no systematic shift of quantification. The higher scatter of quantified proteins in the later runs is only visible in the patient samples, not the QC standards, demonstrating sample quality but not workflow variability (Figure 1C).
Achieving greater proteomic depth
Sample prefractionation is one method of achieving greater proteomic depth but results in long measurement times per sample, which is often compensated for with multiplexing and chemical labeling approaches.4–6 Although isobaric labeling methods are initially appealing, they suffer from radio distortion that can skew or compromise quantification of small or substantial changes in protein abundance between samples. Such an approach is limited for quantifying proteomic differences of potential new biomarkers, especially if they are of low abundance.
A label-free approach combining short gradients with the high selectivity of 4D matching provides a strong alternative that can be applied to thousands of samples, in a modest acquisition time. The data collected in this study covers a dynamic range of five orders of magnitude of protein intensity (abundance) (Figure 2).
Advancing plasma proteomics into the future
Together with the high throughput and sensitivity, this study has demonstrated TIMS-QTOF MS technology with PASEF as a powerful tool for biomarker discovery in large-sample cohorts. Blood plasma samples from 192 patients could be measured in less than two days of LC-MS/MS time, and robust identification and quantification was achieved with low sample volumes (100 ng). The unique characteristics of this technology allow the routine measurement of ion mobility information and enable high-quality 4D feature alignment in retention time, m/z, 1/K0 and MS1 intensity. The alignment boosts the number of quantified plasma proteins to 500 in a single 11.5 min LC-MS/MS run. Achieving this proteomic depth offers new possibilities to analyze large sample cohorts of hundreds to thousands of samples for biomarker discovery in blood plasma.
1. Kosinski T, et al. “Plasma proteomics goes high throughput – timsTOF Pro with PASEF and 4D feature alignment to quantify 500 plasma proteins in 11.5 mins”
2. Cox J, et al. (2014) Accurate proteome-wide label-free quantification by delayed normalization and maximal peptide ratio extraction, termed MaxLFQ. Mol. Cell. Proteomics; 13(9): 2513-26.
3. Meier F, et al. (2018) Online parallel accumulation – serial fragmentation (PASEF) with a novel trapped ion mobility mass spectrometer, Mol. Cell. Proteomics; 17(12): 2534-2545.
4. Keshishian H, et al. (2015) Multiplexed, Quantitative Workflow for Sensitive Biomarker Discovery in Plasma Yields Novel Candidates for Early Myocardial Injury. Mol. Cell. Proteomics; 14(9); 2375-93.
5. Lee SE, et al. (2017) The Plasma Proteome Is Associated with Anthropometric Status of Undernourished Nepalese School-Aged Children. J Nutr.; 147(3): 304-313.
6. Keshishian H, et al. (2017) Quantitative, multiplexed workflow for deep analysis of human blood plasma and biomarker discovery by mass spectrometry. Nat. Protoc., 12(8), 1683-1701.
Heiner Koch, PhD, is senior scientist, Gary Kruppa, PhD, is vice president, and Rohan Thakur, PhD, is managing director and executive vice president at Bruker Daltonics. For more information on the capabilities of the timsTOF Pro with PASEF, please visit www.bruker.com/products/mass-spectrometry-and-separations/lc-ms/o-tof/timstof-pro/overview.html.
We would like to thank Roman Fischer, PhD, from the Target Discovery Institute, Nuffield Department of Medicine, University of Oxford, Oxford, U.K., and investigators from Evosep Biosystems, Odense, Denmark, for their contributions.