Improved SILAC Quantification with Data-Independent Acquisition to Investigate Bortezomib-Induced Protein Degradation

Abstract

Stable isotope labeling by amino acids in cell culture (SILAC) coupled to data-dependent acquisition (DDA) is a common approach to quantitative proteomics with the desirable benefit of reducing batch effects during sample processing and data acquisition. More recently, using data-independent acquisition (DIA/SWATH) to systematically measure peptides has gained popularity for its comprehensiveness, reproducibility, and accuracy of quantification. The complementary advantages of these two techniques logically suggests combining them. Here we develop a SILAC-DIA-MS workflow using free, open-source software. We empirically determine that using DIA achieves similar peptide detection numbers as DDA and that DIA improves the quantitative accuracy and precision of SILAC by an order of magnitude. Finally, we apply SILAC-DIA-MS to determine protein turnover rates of cells treated with bortezomib, an FDA-approved 26S proteasome inhibitor for multiple myeloma and mantle cell lymphoma. We observe that SILAC-DIA produces more sensitive protein turnover models. Of the proteins determined to be differentially degraded by both acquisition methods, we find known proteins that are degraded by the ubiquitin-proteasome pathway, such as HNRNPK, EIF3A, and IF4A1/EIF4A-1, and a slower turnover for CATD, a protein implicated in invasive breast cancer. With improved quantification from DIA, we anticipate that this workflow will make SILAC-based experiments like protein turnover more sensitive.

Keywords: data independent acquisition; protein degradation; protein turnover; pulse SILAC; quantitative proteomics.

Figure 1.. An approach for quantifying pulse SILAC peptides using free, open-source software.

A pulse SILAC experiment is performed (A) and the data is acquired following a chromatogram library approach (B) in which the time₀ sample(s) are injected multiple times with gas phase fractionation (GPF). The GPF time₀ spectra files are searched against a predicted spectral library using EncyclopeDIA, creating a time₀ chromatogram library (i) which is then used to align the light peptide retention times across the entire pulse SILAC experiment (ii). Using the database of retention time-aligned light peptide detections, the heavy peptides are paired and chromatograms extracted using Skyline (iii). Quantitative values for light and heavy SILAC peptides are used to fit protein turnover models and assess statistical significance (C).

Figure 2.. Comparison of PSM and peptide detections in SILAC-DDA vs SILAC-DIA.

(A) Detections in three SILAC proteome samples (100% light SILAC, 100% heavy SILAC, and 50/50% light/heavy SILAC “mix”) are compared from three replicates each of DDA and DIA (windowing scheme of 75×8 m/z, staggered). For DDA analysis, the unique PSMs at 1% FDR are used, without using match-between-runs across samples; for DIA analysis, the unique peptides at 1% FDR are used. (B) The number of unique peptides detected at 1% FDR in each technical triplicate of two DIA windowing schemes is compared for the same 50/50% light/heavy SILAC sample. Schemes are described by the number of isolation windows (e.g. “75”) followed by the width of the isolation windows (e.g. “x8mz”), and whether the isolation windows are staggered (“stagger”) or fixed.

Fig 3.. SILAC-DIA improves dynamic range and quantitative accuracy in benchmark experiments.

The measured log10(heavy/light) ratios in four dilutions of heavy/light HeLa proteome samples are compared using DDA MS1 quantification with match-between-runs (A, B) enabled versus DIA MS2 quantification (C, D). The samples represent a 70%/30% heavy/light proteome (pink, “0.7”), 50%/50% (orange, “0.5”), 30%/70% (blue, “0.5”), 10/90% (green, “0.1”), 1%/99% (cyan, “0.01”), and a 0.1%/99.9% (yellow, “0.001”). Boxplots depict the first and third quartiles (25th and 75th percentiles) of the Log10(Heavy/Light) values for each sample, with whiskers representing 1.5x the interquartile range.

Figure 4.. Protein turnover for bortezomib treated cell cultures.

(A, B) Time is shown horizontally with the fraction of light/total protein vertically. DMSO treatment is shown in blue and bortezomib treatment in green. The shape of these models indicates degradation rate, where proteins with sharply decreasing light fractions are being rapidly degraded, while proteins with more gradual decreases in light fraction are more slowly degraded. (C, D) Volcano plots of significantly differential protein half-lives are shown for DDA and DIA analyses of the same samples for DDA (left) and DIA (right).