Protein Mass Spectrometry Made Simple

Recent progress in protein mass spectrometry has opened the door for the discovery of autoantibody targets in several kidney diseases. We discuss what readers should know about protein mass spectrometry to understand the methodology and new findings as they materialize.

In recent years, mass spectrometry has seen growing use in the identification of endogenous antigens that are the targets of autoantibodies. For example, a major breakthrough was achieved when two podocyte cell surface proteins (the PLA2 receptor and Thrombospondin type-1 domain-containing 7A) were identified that bind autoantibodies present in membranous nephropathy.^1,2 Recently, as reported in the Journal of the American Society of Nephrology, a member of the protein DnaJ homolog subfamily was identified as a likely target antigen in fibrillary GN.^3,4 Similarly, the proximal tubule protein megalin, which functions as an albumin clearance receptor, has been identified as a target antigen in anti-brush border antibody disease, a newly described form of AKI.⁵

Progress has been catalyzed by breathtaking improvements in the speed and mass resolution of mass spectrometers, coupled with better quantification strategies and better bioinformatic methods. On the basis of current successes and anticipated progress, we can predict additional discoveries that speak to the molecular basis of kidney disease. Anticipating future publications about the use of protein mass spectrometry to identify disease biomarkers, what should readers know to evaluate the validity of the conclusions? Here, we provide a simple “how it works” overview of protein mass spectrometry as it is usually performed and outline some basic principles of data analysis relevant to quantification and elimination of false-positive results.

How Bottom-Up Protein Mass Spectrometry Works

The most common approach to identify proteins by mass spectrometry is frequently referred to as “bottom-up” analysis (Figure 1). With this, proteins in a sample are digested using various recombinant proteases, most often trypsin. The digestion can be done in solution or alternatively samples can be run on SDS-polyacrylamide gels and slices of the gel can be subjected to in-gel trypsin digestion. The tryptic peptides are acidified to give them a positive charge. The resulting peptide ions are injected into a HPLC column (the liquid chromatography of liquid chromatography–tandem mass spectrometry [LC-MS/MS]) which sends them to the mass spectrometer over a long period of time (typically 60–120 minutes). The mass spectrometer consists of two analytical stages (the tandem mass spectrometry of LC-MS/MS). At any given point during an LC-MS/MS run, peptides are seen by the first stage of the tandem mass spectrometer (MS1) as a series of peaks, each with a different mass-to-charge ratio. These mass-to-charge ratios provide part of the information needed to identify the sequences of the tryptic peptides. To identify them unambiguously, the dominant peptides in the MS1 are fragmented, most often by colliding the tryptic peptides with inert gases (e.g., nitrogen or argon) with just the right amount of kinetic energy to cleave one and only one peptide bond, producing two fragments (collision-induced dissociation). The cleavage site, however, varies stochastically so that when the cleavage products are displayed in the second stage of the tandem mass spectrometer (MS2), one sees a series of peaks corresponding to all N-terminal fragments superimposed over all C-terminal fragments. The difference in mass-to-charge ratio between neighboring peaks identifies the amino acid between cleaved peptide bonds because each peptide has its own unique residue mass (except for leucine and isoleucine, which have the same mass). In principle, the entire amino acid sequence can be manually read off from the MS2 spectrum. In practice, however, the amino acid sequence is identified by pattern matching between the MS2 spectrum and spectra predicted from a database of all tryptic peptides coded by the appropriate genome. Overall, this process of liquid chromatography–mediated stratification of tryptic peptides, identification of mass-to-charge ratios at the MS1 level, and fragmentation to get MS2 spectra can identify tens of thousands of tryptic peptides from a single sample. Each identification is scored and low-quality identifications are eliminated, typically to obtain a prespecified false-positive rate (e.g., 1%). This procedure can, in principle, produce many false-positive identifications when projected to tens of thousands of tryptic peptides. However, multiple tryptic peptides from a single protein can be identified giving a high degree of redundancy in LC-MS/MS–mediated protein identification. This greatly increases confidence in identifications at a protein level. For example, if the false-positive probability (P) is 0.01 for a particular tryptic peptide identification and six different peptides from a given protein are identified, the overall probability of false-positive identification of the protein would be (0.01)⁶ or 10⁻¹². In the anti-brush border antibody study cited above, for example, 17 peptides mapping to megalin were observed, thereby giving a very high confidence identification.⁵

Figure 1.

Identification of Proteins by Mass Spectrometry. Proteins can be isolated either by laser capture microscopy (LCM) of tissue or by resolution of extracted proteins by gel electrophoresis. Proteins are then further resolved by HPLC and quantified and identified by tandem mass spectrometry.

How Proteins Are Quantified in LC-MS/MS

Biomarker proteins can be recognized in a biologic specimen by an increase in their abundance relative to appropriate control samples. Quantification can be achieved by a variety of methods that can be divided into label-free and labeling methods. A detailed treatment of these methods is beyond the scope of this short tutorial. The simplest label-free method is spectral counting. To compare the abundance of a particular protein between two samples using spectral counting, one simply adds up the total number of peptide spectra that were mapped to the specified protein. This can include the same peptide under different guises (e.g., different charges) or different peptides. Typically, to compare samples, the total number of spectra can be normalized by the total number of spectra obtained for all proteins in the sample. To assure that the comparison is done correctly, normalized spectral counts for several housekeeping proteins can be used. More precise protein quantification can be carried out using isobaric tag labeling methods (Isobaric tags for relative and absolute quantitation or tandem mass tag) in which chemical tags linked to the tryptic peptides allow several samples to be mixed and the relative abundance of each peptide to be compared across all samples. The reader can find information on more sophisticated quantification methods in a recent review article.⁶

Special Aspects of Sample Processing

An important step in protein mass spectrometry analysis is sample enrichment. The simple idea here is that if a disease process is specific to a particular cell type, then biomarkers are more likely to be found when the sample is processed to enrich the cell type of interest. A convenient modality for cell type enrichment is laser capture dissection.⁷ Laser capture has been applied to proteomic analysis of kidney tissue as outlined in Figure 1.⁸ The value of laser capture microscopy is that allows us to enrich for structures in the kidney that are only affected by specific processes. Alternatively, proteins can be enriched by simple resolution on acrylamide gels. Because of dramatic increases in the overall sensitivity of the mass spectrometers used for protein mass spectrometry, it has been possible to analyze smaller and smaller samples. A few thousand cells may be adequate to identify a few thousand proteins, allowing deep analysis of kidney biopsy samples.^9,10

Assessing the Validity of LC-MS/MS Protein Identification and Quantification

When a particular protein is identified as a potential disease biomarker, what should the reader look for in the article to assess the validity of the claim? First, they should identify the false discovery rate per peptide; this value is commonly set to 1% in the peptide search. Second, they should ascertain how many unique peptides were found that map to the identified protein; two or more peptides are usually needed to have confidence in the identification. Third, the reader should look at the spectra; the MS2/fragmentation spectra should show strong peaks that coincide with the theoretical spectra that they matched to. With regard to quantification, the reader should ask whether all of the identified peptides corresponding to the identified biomarker show greater abundance in the test sample than in the control sample. The most critical element, however, is the reproducibility across multiple patients.

Disclosures

None.