Source-linked AI summary
MassChroQ: A versatile tool for mass spectrometry quantification
Benoît Valot, Olivier Langella, Edlira Nano, Michel Zivy
TL;DR
LC-MS quantification tools often target particular labeling strategies or instrument resolutions. MassChroQ aligns data and quantifies peptides from extracted ion chromatograms, with evaluation showing 1.4% technical reproducibility CV and 0.98 correlation to protein quantity.
Problem
Peptide-specific slopes in protein-quantity relationships require quantitative analyses to account for peptide-specific effects.
Method
MassChroQ uses normalization and two-way ANOVA to estimate and remove peptide-specific effects before comparing peptide quantities.
Results
Removing peptide-specific effects put all peptides on the same scale without changing their curve shapes.
Takeaways & Limitations
Peptide-specific effects can be adjusted to support comparable representation of peptide relationships across quantities.
Abstract
from arXiv · showhide
Recently, many software tools have been developed to perform quantification in LC-MS analyses. However, most of them are specific to either a quantification strategy (e.g. label-free or isotopic labelling) or a mass-spectrometry system (e.g. high or low resolution). In this context, we have developed MassChroQ, a versatile software that performs LC-MS data alignment and peptide quantification by peak area integration on extracted ion chromatograms. MassChroQ is suitable for quantification with or without labelling and is not limited to high resolution systems. Peptides of interest (for example all the identified peptides) can be determined automatically or manually by providing targeted m/z and retention time values. It can handle large experiments that include protein or peptide fractionation (as SDS-PAGE, 2D-LC). It is fully configurable. Every processing step is traceable, the produced data are in open standard format and its modularity allows easy integration into proteomic pipelines. The output results are ready for use in statistical analyses. Evaluation of MassChroQ on complex label-free data obtained from low and high resolution mass spectrometers showed low CVs for technical reproducibility (1.4%) and high coefficients of correlation to protein quantity (0.98). MassChroQ is freely available under the GNU General Public Licence v3.0 at http://pappso.inra.fr/bioinfo/masschroq/.
Supplementary materials and methods · 1 Protein extraction and digestion
Yeast proteins were extracted by TCA/acetone precipitation, solubilized in urea, thiourea, and CHAPS, then quantified, reduced, alkylated, and trypsin-digested before TFA acidification.
- 1 Protein extraction and digestion: Yeast proteins were extracted from cell pellets using a TCA/acetone precipitation method.Proteins were precipitated in 10% TCA with 0.07% 2-Mercaptoethanol in acetone.
- 1 Protein extraction and digestion: The protein pellet was rinsed in 0.07% 2-Mercaptoethanol in acetone before resuspension.Resuspension used 8 M urea, 2 M thiourea, and 2% CHAPS.
- 1 Protein extraction and digestion: Protein concentration was determined with the PlusOne 2-D Quant Kit from GE Healthcare.The kit was used after resuspension in the urea, thiourea, and CHAPS solution.
- 1 Protein extraction and digestion: Proteins were diluted 10 times with 50 mM ammonium bicarbonate before enzymatic digestion.The dilution followed reduction and alkylation.
- 1 Protein extraction and digestion: Digestion used trypsin at a 1/50 (w/w) ratio and 37 °C overnight.Trypsin was from Promega, and digestion was stopped by acidification with TFA.
2 LC-MS/MS analysis
The study used two 90-minute nanoLC-MS/MS workflows with comparable peptide loading and chromatographic dimensions but different LC systems, gradients, mass spectrometers, and acquisition settings. One workflow used an LTQ XL ion trap, whereas the other used an LTQ-Orbitrap Discovery with 15,000-resolution FTMS scans.
- LC separation: The first workflow used a NanoLC-Ultra system with a 5–30% B gradient for 60 min at 300 nL/min, yielding 90-minute runs.Buffers were 0.1% HCOOH in water and 0.1% HCOOH in ACN; regeneration and equilibration used 95% B and 95% A.
- Mass spectrometry: The ion-trap workflow acquired enhanced-profile full MS scans from m/z 300 to 1300, followed by centroid MS/MS of the three major ions and 45-second dynamic exclusion.MS/MS used qz = 0.25, 30 ms activation time, and 35% collision energy.
- LC separation: The second workflow used an Ultimate 3000 system with a 4–36% B gradient for 60 min at 300 nL/min, also producing 90-minute runs.Its buffers contained 0.1% HCOOH and 3% ACN in A or 80% ACN in B, with regeneration and equilibration at 100% B and 100% A.
- Mass spectrometry: The Orbitrap workflow acquired profile-mode FTMS scans from m/z 300 to 1300 at 15,000 resolution, followed by centroid LTQ MS/MS of the two major ions and 90-second dynamic exclusion.MS/MS used qz = 0.25, 30 ms activation time, and 35% collision energy.
3 Protein identification
Protein identification used X!Tandem searches against the Saccharomyces Genome Database with contaminants and reversed-sequence decoys, followed by filtering for stringent peptide and protein evidence. The criteria yielded a 0.3% false discovery rate, and redundant proteins were grouped by shared peptides and specific-peptide subgroups.
- Database search: X!Tandem searched the Saccharomyces Genome Database, a contaminant database, and a reversed-protein-sequence decoy database.The Saccharomyces database contained 5885 entries.
- Protein grouping: Proteins sharing peptide sequences were grouped, and members with at least one peptide specific relative to other group members were reported as subgroups.Grouping addressed redundancy when the same peptide sequence occurred in several proteins, and grouped proteins had similar functions.
4 Statistical analysis
Normalized peptide data accounted for global quantitative differences between LC-MS runs and were analyzed after log10 transformation. BSA peptide responses were linear but had peptide-specific slopes, which were estimated and removed without changing curve shapes.
- Normalization: Normalization divided each peptide value by its median ratio to a chosen reference LC-MS run.This accounted for possible global quantitative variations between runs.
- Statistical analysis: Statistical analyses used log10-transformed normalized peptide data.
- Peptide-specific effects: BSA peptides showed linear relationships with protein quantity, but their slopes were peptide-specific.
- Peptide-specific effects: A two-way ANOVA estimated peptide-specific effects using peptide and BSA quantity as factors.The estimated effect of each BSA peptide was removed from normalized data.
- Peptide-specific effects: Removing peptide-specific effects put all peptides on the same scale without altering curve shapes.