Isolation Window Optimization of Data-Independent Acquisition Using Predicted Libraries for Deep and Accurate Proteome Profiling

Abstract

In silico spectral library prediction of all possible peptides from whole organisms has a great potential for improving proteome profiling by data-independent acquisition (DIA) and extending its scope of application. In combination with other recent improvements in the field of mass spectrometry (MS)-based proteomics, including sample preparation, peptide separation, and data analysis, we aimed to uncover the full potential of such an advanced DIA strategy by optimization of the data acquisition. The results demonstrate that the combination of high-quality in silico libraries, reproducible and high-resolution peptide separation using micropillar array columns, as well as neural network supported data analysis enables the use of long MS scan cycles without impairing the quantification performance. The study demonstrates that mean coefficient of variations of 4% were obtained even at only 1.5 data points per peak (full width at half-maximum) across different gradient lengths, which in turn improved proteome coverage up to more than 8000 proteins from HeLa cells using empirically corrected libraries and more than 7000 proteins using a whole human in silico predicted library. These data were obtained using a Q Exactive orbitrap mass spectrometer with moderate scanning speed (12 Hz) and perform very well in comparison to recent studies using more advanced MS instruments, which underline the high potential of this optimization strategy for various applications in clinical proteomics, microbiology, and molecular biology.

Deep Time Course Proteomics of SARS-CoV- and SARS-CoV-2-Infected Human Lung Epithelial Cells (Calu-3) Reveals Strong Induction of Interferon-Stimulated Gene Expression by SARS-CoV-2 in Contrast to SARS-CoV

“…Samples were prepared as biological triplicates and analyzed using single-shot DIA-based proteomics, with an optimized workflow for deep and accurate protein profiling. 20 In total, 8391 proteins were identified in a 3 h gradient, of which 7478 proteins were consistently quantified and used for further analysis (Pearson correlation >0.98, median coefficient of variation between 0.048 and 0.062 within each triplicate, data completeness 98.3%). Viral replication was verified by qPCR of the cell culture supernatants.…”

High‐throughput, in‐depth and estimated absolute quantification of plasma proteome using data‐independent acquisition/mass spectrometry (“HIAP‐DIA”)

“…In other intensity bins, median CV was about 10% for 40 and 50 DIA segments. As quantification reproducibility of peptides in lowest peptide intensity bin depended on both number of data points and co‐isolation degree, more or less DIA segments would increase CV [32]. 40 or 50 DIA segments was optimal for good quantification reproducibility with shown CVs.…”

Unbiased antimicrobial resistance detection from clinical bacterial isolates using proteomics

“…The LC-MS/MS measurement relies on data-independent acquisition (DIA), which circumvents the protein detection bias caused by the stochastic nature of data-dependent acquisition (DDA). This strategy allows collection of proteomes with deep coverage (Doellinger et al , 2020a). However, it introduces big challenges for data analysis, which need to be covered by tailor-made concepts.…”

“…The workflow further involves very rapid (~30 min) and universal sample preparation using acid extraction, without the need for protocol adaptation to certain bacterial species. Potentially, this method, termed SPEED, should enable to fully automate sample preparation including bacterial cell lysis in the future, as it uniquely omits the need for physical disruption (Doellinger et al , 2020a).…”