Simple Peptide Quantification Approach for MS-Based Proteomics Quality Control

Abstract

Despite its growing popularity and use, bottom-up proteomics remains a complex analytical methodology. Its general workflow consists of three main steps: sample preparation, liquid chromatography coupled to tandem mass spectrometry (LC-MS/MS), and computational data analysis. Quality assessment of the different steps and components of this workflow is instrumental to identify technical flaws and avoid loss of precious measurement time and sample material. However, assessment of the extent of sample losses along with the sample preparation protocol, in particular, after proteolytic digestion, is not yet routinely implemented because of the lack of an accurate and straightforward method to quantify peptides. Here, we report on the use of a microfluidic UV/visible spectrophotometer to quantify MS-ready peptides directly in the MS-loading solvent, consuming only 2 μL of sample. We compared the performance of the microfluidic spectrophotometer with a standard device and determined the optimal sample amount for LC-MS/MS analysis on a Q Exactive HF mass spectrometer using a dilution series of a commercial K562 cell digest. A careful evaluation of selected LC and MS parameters allowed us to define 3 μg as an optimal peptide amount to be injected into this particular LC-MS/MS system. Finally, using tryptic digests from human HEK293T cells and showing that injecting equal peptide amounts, rather than approximate ones, result in less variable LC-MS/MS and protein quantification data. The obtained quality improvement together with easy implementation of the approach makes it possible to routinely quantify MS-ready peptides as a next step in daily proteomics quality control.

Figure 1

Schematic outline of the sample preparation and quantification steps in a generic bottom-up proteomics experiment. Peptide concentration is usually estimated based on protein quantification before proteolytic digestion (dark gray), but standard methods for performing this quantification at the peptide level, right before LC–MS/MS (red), would be highly desirable. Optional steps of the protocol are written in gray.

Figure 2

Dynamic quantification range of the microfluidic spectrophotometer for complex peptide mixtures. Yeast peptide digest solutions with concentrations ranging from 10 ng/μL to 1.0 μg/μL (black dots) can be accurately measured, as shown by a strong correlation between the theoretical and measured peptide concentration values. Below 50 ng/μL (red dots), peptide measurements are no longer reliable, as absorbance values reach the lower limit of detection of the spectrophotometer. Dots represent measurements of four independent replicate dilutions.

Figure 3

Quality control metrics from LC–MS/MS analysis of 10 different amounts of a trypsin-digested K562 cell extract. (a) Theoretical and measured peptide amounts for two independent experiments. In both experiments, all measured values were lower than predicted, pointing to peptide losses during sample solution preparation. (b) Bar plots indicating the numbers of identified peptide and proteins per analytical run. Colors represent the results of two replicate series analyzed on two different LC columns. (c) Representative heatmap visualizing the intensities of quantified proteins in each of the 10 analytical runs measured using LC column 1. Missing values are shown in gray. Based on numbers of identified and quantified proteins a rather broad window ranging from 1.49 to 5.20 μg of peptides was found as an optimal injection amount. (d) Scatter plots of average MS2 ion injection time, median peak area, and median signal-to-noise ratio (S/N) of identified peptides in each analytical run. (e) Scatter plots of the median peak area of hydrophilic, intermediate hydrophobic, and hydrophobic identified peptides. For both replicate series, from around 2–3 μg, peak areas decline steeply for hydrophilic peptides, while they go up progressively for hydrophobic peptides, two types of peptide classes having a lower and higher affinity for the used precolumn, respectively. Colors represent the results of two replicate series analyzed on two different LC columns. Loss curve fitting was used to generate the connecting lines. Based on the loss of hydrophilic peptides, 3 μg was defined as an optimal peptide injection amount for this particular LC–MS/MS setup (dashed lines).

Figure 4

Comparison of LC–MS/MS and protein quantification reproducibility for two sets of analytical runs on trypsin-digested Hek293T cell extracts. (a) Plots of the average MS2 injection time, median peak area, and median signal-to-noise ratio values are shown for the identified peptides in each of five replicates from the corrected and “noncorrected” groups of samples. For each data point, the exact injected peptide amount based on quantification on the microfluidic spectrophotometer is shown on the right. Vertical bars represent the mean ± standard deviation. (b) Density plots and boxplots showing the distribution of the coefficients of variation (CV) of raw intensities (left) and (normalized) maxLFQ intensities (right) for all proteins quantified in the ten analytical runs (n = 2677). For both raw and normalized protein intensities, a shift to higher CV values was found for the noncorrected samples (Mann–Whitney U test p-value < 2.2 × 10^–16 for the shift in both raw and normalized distributions).