Multiple and Single Reaction Monitoring Mass Spectrometry for Absolute Quantitation of Proteins

Abstract

The use of mass spectrometry (MS) to measure proteins has grown exponentially over the past 25 years. This growth has been primarily driven by the advent of proteomics in which scientists have developed methods to identify and quantitate as many proteins in a complex mixture as possible. Early studies trended towards the development of techniques that enabled greater quantitative coverage of the proteome. Many of these developments focused on relative quantitation in which the change in the abundances of proteins in comparative samples was measured. However, relative quantitation only allows a limited number of samples to be compared. This deficiency led to the development of technologies that allowed the relative quantitation of an unlimited number of samples to be measured, but what was still lacking was an emphasis on the ability of MS to measure the absolute abundance of proteins. A more recent technology trend has taken full advantage of the analytical attributes afforded in the use of MS for protein measurements. This trend utilizes the accuracy, sensitivity, specificity, and multiplexed capabilities of MS to quantity specific proteins within complex mixtures. Combined with the use of stable isotope-labeled internal standards, MS assays are now being developed to quantitate key diagnostic and prognostic proteins within clinical samples such as serum, plasma, urine, and cerebrospinal fluid. This chapter describes the technology behind the development of MS-based clinical protein assays and provides examples of where these assays are being used in diagnostic and prognostic settings.

1. Introduction

Mass spectrometry (MS) has evolved over the past three decades to take its place amongst the premier analytical techniques in use today. The development of MS instrumentation can be traced back to the field of physics, where it was developed as a technique to study the electron [1]. Until the 1940’s, MS was still primarily used within the domain of physicists until novel instrument designs by Alfred Nier enabled MS to be used to separate isotopes such as carbon-13 (¹³C) as well as uranium-235 and uranium-238 [2]. It was during the 1940’s that mass spectrometers became commercially available and were widely used by industrial chemists to quantitatively measure molecules whose identities and structures were already known within mixtures [3]. As a result of the efforts of Fred MacLafferty, Klaus Biemann, and Carl Djerassi, the fragmentation mechanisms that occurred during the process of measuring molecules using MS were determined, enabling unknown components within mixtures to be identified [3].

While the analysis of small organic molecules became increasingly routine, it was not until the 1980’s that methods for analyzing macromolecules were developed, due to the challenge of finding ways to get large molecules into the gas phase without causing them to fragment. The pioneering work of John Fenn and Koichi Tanaka in the areas of electrospray ionization (ESI) [4] and matrix-assisted laser desorption and ionization (MALDI) [5], respectively, paved the way for macromolecules to be analyzed using MS. It was these developments, especially ESI, from which the field of MS-based proteomics was born.

2. Mass spectrometry and the development of proteomics

The ability to analyze large macromolecules using MS was the beginning of the proteomics revolution. The term proteomics was first coined in 1995 and referred to the identification of individual proteins that had been separated using two-dimensional gel electrophoresis [6]. Initially the proteomics revolution started small, only using MS to identify proteins in simple mixtures [7]. However, as MS technology advanced, the number of proteins identified in increasingly complex biological samples skyrocketed [8, 9, 10]. For example, the number of proteins identified in human serum increased from fewer than 500 to over 4000 in less than 20 years [11, 12]. A recent study even identified an astonishing 340,000 proteins across 100 taxonomically diverse organisms, doubling the number of proteins that had previously been identified with direct experiment evidence [13].

As the ability to identify large numbers of proteins became commonplace, advancements in the field turned to focus on quantitation. The initial methods were built to compare the relative abundances of proteins in two samples. The most used method involved stable isotope labeling of proteins that were being translated in cultured prokaryotic or eukaryotic cells [14, 15, 16]. In this method, two identical cell lines are cultured in vitro. One of the cell lines is cultured in normal media (e.g., DMEM, RPMI, etc.) while the other is cultured in the same medium containing a heavy stable isotope labeled component (e.g., ¹³C-labeled amino acid, ¹⁵N-enriched medium, etc.) [17, 18]. One sample is treated with a specific perturbation, while the other acts as a control. At some time-point the two cultures are combined, the proteins extracted and digested into tryptic peptides, are then analyzed using liquid chromatography (LC) coupled directly online to MS (Figure 1). Peptides originating from the separate cultures are distinguished by their heavy isotopic content and are quantitated based on their peak area. This measurement allows the relative abundance of the protein (through its peptide surrogate) in the two samples to be compared and the effect of the perturbation on the protein to be determined. While most of the proteins will not show any change in abundance, studies often identify tens to hundreds of proteins whose abundance changes several-fold due to the perturbation [19, 20].

Figure 1.

Use of stable isotope labeling for comparing the relative abundances of proteins in proteome samples. In this method, cells are grown in normal media or media to which a heavy version of a stable isotope has been added. The heavy stable isotope is incorporated into the proteins as they are translated. The proteome samples are combined in equal ratios and the combined sample is proteolytically digested into peptides. The mass spectrum of the isotope labeled and unlabeled peptides are acquired to compare their relative abundance in the two samples. The final step is the automated identification of the peptide using tandem mass spectrometry (MS/MS).

Once scientists understood that entire proteomes could be analyzed, a variety of methods for conducting comparison studies were quickly developed. Some of these methods included chemical labeling with isotope tags [21, 22, 23], isotope labeling in vivo [24], and methods to compare proteomes based on MS peak intensity [25] and number of identified peptides [26]. While label-free methods such as MS peak intensity and peptide number increased the number of samples that could be compared, the samples needed to be prepared and analyzed in a consistent manner for the comparative measurements to be analytically valid. It was not until the development of the Sequential Window Acquisition of all Theoretical Spectra (SWATH) method that the relative abundances of proteins within an unlimited number of samples could be compared [27].

2.1 The biomarker movement

The capability of comparing the relative differences in thousands of proteins in complex samples opened an entire new use for quantitative MS. While much of the focus was on determining how various perturbations changed cellular proteomes, scientists also began to explore how this technology could be used to identify biomarkers of various diseases [28, 29, 30]. The premise was simple - acquire a cohort of clinical samples (e.g., plasma, serum, urine, etc.) from a group of patients with a specific disease and a cohort from healthy, matched controls. Each sample would be separately analyzed and the relative abundance of proteins in samples obtained from disease-affected and healthy individuals would be compared [31, 32]. The combined increase in mass spectrometer technology along with this simple sample preparation method resulted in this procedure being utilized extensively in biomarker research. For example, in over the past 50 years, over 30,000 manuscripts containing the keywords “mass spectrometry” and “biomarker” have been listed on PubMed, with almost 97% of these published in the past two decades.

While proteomics has brought about the ability to measure thousands of proteins in hundreds of samples, it also presents several analytical challenges. Since large-scale proteomic biomarker discovery studies can require several weeks, deterioration of overall analytical performance increases in likelihood. For example, chromatographic performance and MS instrument performance can decline as impurities within the biological samples accumulate. A decline in overall performance makes it difficult to determine if differences observed between different cohorts of samples are due to biological or analytical variance. Therefore, robust quality control (QC) strategies are required to optimize the chances of identifying legitimate biomarkers.

Over the past couple of decades there have been thousands of potential biomarkers for a variety of diseases reported in literature. Unfortunately, between the years 1993 and 2008, only 22 novel protein-based tests were approved by the United States Food and Drug Administration (FDA) [33]. There are many issues that have created this discrepancy. Some of them can be understood in the context of how, and how many, samples must be analyzed at each stage of a biomarker discovery project. As shown in Figure 2, proteomic biomarker studies generally follow four stages: i) discovery, ii) qualification, iii) verification, and iv) validation. In many ways, the stages are like investigational new drug (IND) clinical trials in scope and number of samples required. The discovery stage typically involves the analysis of tens of samples, however, data on the relative abundance of thousands of analytes is recorded. After this initial data set is scrutinized, only proteins that show a change in abundance between the cohorts are focused on. Any of these proteins that still show a difference between the two cohorts are specifically “targeted” in a larger number of samples. In targeting, the mass spectrometer is instructed to ignore all peaks except those that arise from the qualified proteins that show an abundance difference. Target strategies are aimed at increasing the analytical stringency of measurements, thereby increasing the chance that the abundance difference observed is a result of biological variance. In the final validation stage, specific assays that measure the absolute abundance of a very small number (i.e., less than 10) of proteins are designed and used to analyze thousands of clinical samples. Any protein that passes this final stage may then become approved for clinical use.

Figure 2.

The steps in biomarker discovery. Biomarker discovery proceeds through the four stages of discovery, qualification, verification, and validation prior to a molecule being reliable for clinical use. Moving through these four stages requires an increasing number of samples to be analyzed, but fewer analytes within each sample to focus on. Not only does the number of samples and analytes vary during these four stages, but so do the analytical and biological variance observed within the samples. The analytical techniques used during the initial discovery stage possess high analytical variance since many analytes must be surveyed. As the study progresses to validation, the analytical variance decreases substantially as the analytical technique chosen is tuned to measure a smaller number of analytes.

One of the biggest hurdles in gaining FDA approval occurs early within the process. The discovery phases of biomarker studies generally result in a significant percentage of the measured proteins showing a difference in their relative abundances between the two cohorts of samples. This feature makes selection of which proteins to focus on in the following stages difficult. Another challenge is one that is common to IND trials: access to proper samples. Progressing through all the stages of a biomarker discovery project ultimately requires access to well over a thousand samples that must be carefully stratified, acquired, stored, and processed. The difficulty in maintaining such a rigorous environment is not a simple task especially when multiple laboratories are involved in the biomarker discovery and validation process. This necessary rigor has been a major contributor to the lack of biomarkers that have been validated using this approach.

2.2 The movement from relative to absolute quantitation

As the number of proteomes and samples that could be compared increased, there was still something lacking in the results: the ability to measure a protein’s absolute abundance. While most experiments performed in basic research measure the relative abundance of proteins in samples (e.g., western blotting, immunofluorescence, etc.), knowing the absolute abundance of a molecule allows an unlimited number of samples to be compared. More importantly, measuring the absolute abundance of molecules is the foundation of many medical tests such as comprehensive and basic metabolic panels [34]. If MS were to become a major clinical technique, it also needed to be able to measure the absolute abundance of proteins. It was not that the technology was not available: in fact, MS had been used for decades for measuring the absolute quantity of metabolites in clinical samples. Gas chromatography coupled with MS had become mature enough that the U.S. Environmental Protection Agency adopted this technique as its standard method for quantitating several key pollutants [35]. Probably the most well-known quantitative MS assay is the “in-born errors of metabolism” test, which analyze newborn blood samples for defects in fatty acid, organic acid, and amino acid metabolism [36]. While this technique measured metabolites and not proteins, its value is inarguable as millions of infants worldwide and more than 500 confirmed disorders have been screened using this method [36, 37].

3. Absolute protein quantitation

Methods for measuring the quantity of proteins using MS determine either their relative or absolute abundances. While most researchers employ the former approach, there is an increasing movement towards the latter approach. This trend is primarily due to the desire to apply MS for clinical applications. The biggest advantage of measuring a molecule’s absolute abundance is the ability to compare results from samples acquired and analyzed anywhere in the world using a standard operating procedure. Absolute quantitation also allows the abundance of different molecules within a single sample to be directly compared. For example, if a cell surface receptor is found to be increased in abundance, a correlation to other proteins involved in its signaling pathway can also be determined if absolute abundances are being measured. Measuring the relative abundance only allows the researcher to determine if a protein’s abundance differs between samples taken from different individuals. It does not provide any information related to how a protein’s abundance has changed in relation to other proteins within the same sample.

Although antibody-based methods (i.e., immunoassays) currently dominate the protein assay field in clinical laboratories, these methods are not without their disadvantages [38]. Immunoassays can suffer from lot-to-lot antibody variation, high cost, and their need for relatively high sample volumes. In addition, a vast majority of immunoassays only measure a single analyte per experiment. On the positive side, immunoassays are easily automated and there is a large workforce of scientists that are currently trained in conducting these types of experiments.

Using MS to conduct clinical assays has several distinct advantages over immunoassays [39]. Mass spectrometry methods are highly sensitive and require very little sample volume. Since samples are usually fractionated using liquid chromatography prior to MS analysis, hundreds of molecules can be quantitated per experiment. The stable isotope labeled standards, which are added to samples to enable absolute quantitation, are easily synthesized, and do not suffer from lot-to-lot variation.

3.1 Multiple reaction monitoring-mass spectrometry

Determining protein absolute abundance using MS is conducted using a technique called multiple reaction monitoring- or single reaction monitoring-MS (MRM-MS or SRM-MS) [40, 41]. These techniques are sometimes referred to as targeted MS because instead of measuring as many proteins as possible, they are used to quantify specific (or targeted) proteins in a mixture. While enzyme-linked immunosorbent assays (ELISAs) have been the dominant technique for measuring the absolute abundance of protein in complex mixtures, MRM- and SRM-MS methods offer several key advantages. MRM- and SRM-MS assays are not limited by the availability of suitable antibodies. This advantage allows these assays to be readily customizable to suit any new targets that are discovered. MRM- and SRM-MS assays also provide a direct measurement of the analyte of interest, whereas ELISAs can suffer from cross-reactivity associated with antibodies [42].

MRM-MS is generally performed using triple quadrupole or triple quadrupole ion trap mass spectrometers because of their linear quantitation range, however, other types of instruments (e.g., Orbitraps, quadruple-TOFs, etc.) can also be used [43, 44]. As shown in Figure 3, the first step of an MRM-MS (and SRM-MS) experiment is to digest the proteome into peptides [43, 44, 45]. The peptides are then separated using LC coupled directly on-line with the mass spectrometer. As they elute into the first quadrupole (Q1), the specific peptide of interest is isolated by applying a radiofrequency/direct current (RF/DC) potential that prevents all other peptides from passing through this region. The isolated peptide then moves onto the second quadrupole (Q2) and is fragmented, generally through collisions with an inert gas. All the fragments enter the third quadrupole (Q3), however, only around 3–5 are isolated and detected. A SRM-MS experiment is done in a very similar manner, except only a single fragment is isolated in Q3 and detected.

Figure 3.

Multiple reaction monitoring-mass spectrometry (MRM-MS) for the absolute quantitation of proteins.

So why are 3–5 fragments selected in an MRM-MS experiment but only one in a SRM study? The answer partly comes down to accuracy. Ensuring that the correct peptide is being measured is foundational to quantitative proteomics. It is inadequate to identify a peptide only based on molecular weight (MW) because the complexity of a proteome sample means many peptides will have the same (or very similar) MWs. It is unlikely, however, that two peptides will have identical fragment ions. In MRM-MS, multiple fragment ions are monitored to ensure the correct peptide is being measured, while in SRM-MS, only a single fragment ion is being monitored. In general MRM-MS is used regularly for peptide measurements while SRM-MS is used for measuring small metabolites. Therefore, the rest of this chapter will focus only on MRM-MS quantitative studies.

To measure the absolute quantity of the peptide, a known amount of stable isotope labeled internal standard with the same sequence as the peptide of interest is added to the sample. The amount of endogenous peptide can then be determined by comparing its peak area to that of the internal standard.

As with any assay, an MRM-MS assay is only useful if it is specifically measuring the protein of interest. Therefore, rules have been established to ensure the specificity of the MRM-MS assay [46]. The peptide selected to act as the surrogate must be unique to the targeted protein. To ensure its uniqueness, the sequence of the peptide must be compared to all sequences within the corresponding species-specific protein database. The peptide should be between 7 and 20 amino acids long since this size range is ideal for efficient generation of fragments. It should also be within all the different isoforms of the protein (if any). If the experiment is designed to measure the abundances of different isoforms of a protein, the peptides selected need to be unique to each isoform. To prevent any change to the peptide during sample processing, it should be void of any easily oxidizable amino acids (i.e., cysteine, methionine, etc.). In addition, it is best to measure multiple (i.e., 3–5) peptides from the same protein whenever possible so that the results from each can be used to confirm the result.

3.2 Validating MRM-MS

Prior to its widespread use, MRM-MS needed to prove itself as an accurate technique for measuring protein abundance. As the goal of using MRM-MS was to measure biomarkers in clinical samples, the robustness of the technique had to be extensively validated in serum, plasma, and urine samples [47, 48, 49]. In one of these validation studies [47], eight participating laboratories were asked to measure the absolute abundance of 11 peptides originating from 7 different proteins. The aim of this study was to determine the intra- and inter-laboratory variation of MRM-MS assays at different sites. Five of the proteins were non-human in origin, to eliminate any unpredictable interference from endogenous proteins already within the plasma sample. The other two proteins were the prostate cancer biomarker, human prostate-specific antigen (PSA) [50] and the acute phase response protein, C-reactive protein (CRP) [51].

This study was conducted in three phases [47], with the measured outcome of each phase being the absolute recovery of the peptides based on comparison to the signals generated by the heavy isotope labeled internal standards. The three phases differed in the amount of sample preparation required at each site. In phase I, the participating labs received plasma samples pre-digested using trypsin that already contained the 11 peptides of interest. Each lab had to simply run the MRM-MS assay. The inter- and intra-laboratory reproducibility and precision in this phase was excellent for all peptides. In phase II, each laboratory had to perform the tryptic digestion step and clean up the sample prior to MRM-MS analysis. The additional sample preparation steps had minimal impact on the precision as the coefficient of variation (CV) between the labs remained less than 15%. In phase III, all the sample preparation step were performed on-site (i.e., additional of heavy isotope labeled internal standards, tryptic digestion, sample clean-up, etc.) prior to MRM-MS analysis. While the overall peptide recovery was significantly lower (i.e., 119.8%, 79.6%, and 48.9% for phases I, II, and III, respectively), the interlaboratory CVs remained below 25% for 8 of the 11 peptides of interest. This study highlighted areas of special concern, primarily in sample preparation steps, that need to be standardized to ensure accuracy in absolute protein abundance measurements.

3.3 Multiplexing the technology

The combination of LC separation with MS analysis provides an analytical platform capable of detecting thousands of individual components in a clinical sample such as plasma, serum, or urine. Imagine what could be done if many of these components were valid biomarkers for various conditions such as cancer and even a fraction of these detectable components could be accurately quantitated. Patients could provide blood and urine samples at yearly physicals that could be used to diagnose specific cancer types at very early stages. As early detection is a key to surviving cancer, the ability to accurately quantify specific biomarkers in routinely acquired clinical sample would substantially decrease the death rate due to many different cancers.

While this multiplexing capability is currently employed for measuring many in-born errors such as metabolism [36], steroid hormones [52], and lipids [53], proteins are not included. However, there is a current effort to expand its use to proteins, and in the largest study to date, an assay to quantitatively measure 267 proteins using MRM-MS was designed [54]. To test its veracity, the 267 proteins were measured across 21 different commercially available human plasma samples. Within these proteins, 61 had been FDA-approved for use in laboratory developed tests (LDTs) [55] and a further 67 were putative cardiovascular disease biomarkers [56]. The major goal of this study, beyond showing the utility of MRM-MS as an analytical tool, was to develop an easy-to-use standardized kit that would provide reproducible and transferable clinical results in a variety of laboratory settings.

In this study, a known amount of stable isotope labeled peptides corresponding to specific endogenous target peptides was added to each commercial plasma sample. The absolute abundances were then calculated by determining the peak area ratios between the endogenous target peptides and their corresponding stable isotope standard and comparing this ratio to previously constructed standard curves. The standard curves were prepared on all three days using 5–8 standards. These curves allowed both the lower and upper limits of quantitation (LLOQ and ULOQ) to be calculated for each protein. Each sample was analyzed each day for three separate days to evaluate robustness. Of the 267 targeted proteins, 144 were quantified in at least 5 of the 21 samples. Just over one-third of these were FDA approved as biomarkers or for use in LDTs. In addition, a total of 111 proteins were quantified in all 21 samples over the course of the three-day analysis stage. No protein in any of the samples had a concentration above the ULOQ, however, 110 proteins were below the LLOQ in all 21 samples in all 3 separate analyses. Over half of the proteins whose concentrations were below the LLOQ have associations with various conditions including cardiomyopathy, epilepsy, neuropathy, rheumatoid arthritis, and lung cancer. Although a measurement below the LLOQ prevents a protein from being quantified, this observation is still important for evaluating various diseases in which a specific protein may be upregulated, and its concentration falls within the limit of detection (LOD) of this multiplexed assay.

As the goal of this study was to develop a kit that could be used across various laboratories, it was important that the assay possessed consistency. Comparison of the 21 samples showed high consistency as a median of 145 proteins (standard deviation 15.8) were quantified over the three days of analysis, as shown in Figure 4A., with an 84% overlap of quantifiable proteins per sample over the three analyses. Overall, the data showed that the MRM-MS assay is highly reproducible for quantifying multiple protein targets in a very complex biological sample. The CVs were calculated for all the proteins that could be quantified (Figure 4B), and approximately 70% of the CVs were below 10%, with only about 8% greater than 20%. Taken together, the data points towards the reliability of multiplexed MRM-MS assays quantitating proteins within clinical laboratories.

Figure 4.

(A) Number of proteins that were quantified within 21 plasma samples analyzed over a three-day period. To be counted, each protein had to meet acceptable analytical criteria. (B) Distribution of percentage of quantitated proteins based on their coefficients of variation (CVs) of their abundances measured over three days.

Another important characteristic of any analytical technique, specifically one aiming to be used to measure biomarkers, is its dynamic range. Dynamic range refers to the range of concentrations that a technique can measure [57]. Since the concentration range of proteins in human plasma spans 12 orders of magnitude, it is important that MRM-MS provide a comparable dynamic range [58]. The developed MRM-MS assay displayed a dynamic range of six orders of magnitude ranging from a high of serum albumin (747 pmol) to a low of p-selectin (1 fmol). P-selectin is up-regulated in endothelial cells and platelets in sickle-cell anemia patients and contributes to their symptoms of pain and vaso-occlusion [59]. Recently, Novartis was granted approval for the use of crizanlizumab as a treatment for reducing these symptoms through its binding to p-selectin [60, 61]. The coefficients of variation for 86% of the measured proteins were less than 15%. The differences in protein concentrations in the 21 samples ranged from a highly consistent 1.1-fold for metalloproteinase inhibitor 2 to a widely variable 69-fold for serum amyloid A1/A2. The highest variability observed across the samples for an FDA-approved biomarker was the 60-fold variation seen for C-reactive protein, which along with serum amyloid A is known to increase several hundred-fold in concentration due to an inflammatory response [62].

Working with MRM Proteomics, Inc. and Cambridge Isotopes Laboratories, Inc., investigators coupled the necessary reagents and instructions into the commercially available PeptiQuant™ Assay Kits for perform the analysis described above [63]. The kits are available as either PeptiQuant Plus Quality Control Kits for evaluating the performance of a lab-developed MRM-MS assay or the PeptiQuant Plus Biomarkers Assessment Kit that analyzes human or mouse plasma samples for more than 125 disease-related biomarkers. The development of these kits not only shows that MRM-MS is an important tool for diagnosing and monitoring diseases such as cancer, heart disease, etc. but the commercialization of the PeptiQuant Assay Kits demonstrates that the actual clinical use is not far off in the future.

4. Applications of MRM-MS

An immediate use of MRM-MS is the improvement on existing techniques for measuring known disease biomarkers. Hepatocyte growth factor receptor (HGFR), also known as Met, is a tyrosine kinase membrane receptor that is measured as a biomarker for various cancers [64]. When over-activated, Met activates several biological activities that can result in an invasive oncogenic phenotype. The overexpression of Met is directly correlated with tumor aggression and poor patient outcome. Such correlation has been observed in gastroesophageal and esophageal adenocarcinomas [65]. As a result, Met is a prime therapeutic target and several monoclonal antibodies (mAb) that bind to and inhibit either HGF or Met are currently being tested in clinical trials. In fact, a patient with stage IV gastroesophageal cancer that had a high MET gene copy number and concordant Met expression showed complete response to the Met-specific mAb onartuzumab [66]. In a phase II trial, rilotumumab treatment increased the survival of patients with tumors exhibiting a high Met expression level compared to those with low Met levels [67].

Selection of the correct treatment of Met-driven tumors depends on the ability to measure Met expression in patient samples. These measurements are routinely performed using immunohistochemistry (IHC) [68]. Unfortunately, IHC is only semi-quantitative; simply providing a “score” of 0, +1, +2, or + 3 that signifies the levels of protein expression. The score is generated by considering the staining intensity as well as the percentage of stained cells. The score is also prone to subjective bias since the colorimetric signal produced during IHC is interpreted by a pathologist [69]. While imaging software has attempted to standardize the technique, IHC is still not uniformly and systematically applied for quantitating specific proteins in tumor biopsies.

To overcome the subjectivity and semi-quantitative nature of IHC, Liquid-Tissue-SRM was developed specifically to measure the absolute abundance of specific proteins in formalin-fixed paraffin-embedded (FFPE) tissue [70, 71, 72]. In this method, tumor cells are extracted from tissue sections using laser microdissection (Figure 5). A lysate is prepared from these cells and digested into peptides using trypsin, and a stable-isotope labeled version of a peptide within the target protein (in this case Met) is added to the peptide mixture and the sample is analyzed using LC-SRM-MS.

Figure 5.

Liquid tissue-selected reaction monitoring-mass spectrometry (SRM-MS) workflow for quantitation of proteins extracted from formalin-fixed paraffin-embedded (FFPE) tissue sections. In this method, specific areas procured from deparaffinized tissue sections are placed in a tube. The proteins within the cells are extracted and digested into peptides. A known amount of heavy isotope labeled peptide corresponding to a peptide of interest is added and the samples are analyzed using SRM-MS to measure the absolute abundance of the targeted endogenous peptide.

To test the method, 130 FFPE gastroesophageal cancer tissues were analyzed using Liquid Tissue, IHC, mean MET gene copy number/nucleus, and MET/CEP7 gene copy number ratio using fluorescence in situ hybridization (FISH) [72]. The correlation between Liquid Tissue and IHC results was low (R² = 0.537). The correlation between Liquid Tissue results and the MET/nucleus and MET/CEP7 gene copy number results, however, was high (R² = 0.898).

In a similar study, an MRM-MS method was developed to quantitate human epidermal growth factor receptor 2 (HER2) in breast cancer tumors [73]. Overexpression of HER2 is associated with breast cancer and higher levels of the protein correlate with poorer patient outcomes [74]. While the study ultimately sought to develop a method with higher quantitative accuracy than IHC, it also was attempting to address a limitation with FISH. Specifically, FISH is routinely used to evaluate samples assigned an IHC score of +2. Unfortunately, the process is time-consuming and expensive. Therefore, establishing an MRM-MS method that obviated the need for both IHC and FISH would improve on existing methods for differentiating HER2 status as well as reducing costs.

In this study, 210 FFPE tissue sections were cut from breast cancer samples [73]. After processing the samples using a similar Liquid Tissue process as described above, the absolute levels of HER2 and the epithelial cell-specific protein, adhesion molecule A, were measured using MRM-MS. To normalize the results, the absolute amount of HER2 was divided by that of adhesion molecule A, which is not expected to change based on tumor status. In contrast to IHC alone, the MRM-MS analysis was able to distinguish HER2 2+/FISH positive and HER2 2+/FISH negative samples, with an area under the receiver operator characteristic curve (AUROC) of 0.908. The results show that MRM-MS assays provide more accurate HER2 expression levels than IHC and can eliminate uncertainty in the decision making of oncologists seeking the proper treatment of HER2 positive breast tumors.

4.1 Taking MRM in vivo

While measuring the absolute abundance of a protein is a hallmark of diagnosing and/or monitoring many disease states, there is value in measuring protein turnover as well, since it can detect abnormalities in protein clearance rates and anticipate potential build up of proteins in the future. While protein turnover could be measured by adding stable isotope labeled standards to a series of samples acquired at several time points, this method suffers from analytical variabilities and places a significant burden on the patient. To overcome these challenges, investigators at Washington University in St. Louis developed an in vivo stable isotope labeling method to study protein turnover. While stable isotope labeling in vivo had been done previously on animal species including rats and mice [75], this was the first case of the procedure being performed on humans [76]. This method, termed stable isotope labeling kinetics (SILK), used in vivo incorporation of a stable isotope labeled amino acid to measure the fractional synthesis (FSR) and fractional clearance rates (FCR) of specific proteins. The focus of SILK was to measure the FSR and FCR of beta-amyloid (Aβ) to evaluate the role that this protein’s production and clearance plays in plaque formation [77].

In the SILK procedure, human subjects are intravenously infused over a 9-hour period with ¹³C₆-labeled leucine (¹³C₆-Leu) dissolved in normal saline (Figure 6). As ¹³C₆-Leu enters the blood stream, it is incorporated into proteins that are being actively translated within cells. Blood samples are then taken from the patient every hour for up to 48 hours. This time provides the data necessary to measure the uptake of ¹³C₆-Leu into newly translated Aβ during the earlier time points and the degradation of this protein population over the later time points. To simplify the mixture, an antibody is used to immunoprecipitate Aβ from each sample. The purified Aβ is then digested into peptides and analyzed using MRM-MS.

Figure 6.

Methodology for conducting stable isotope labeling kinetics (SILK) study to determine fractional synthesis (FSR) and fractional clearance rates (FCR) of beta-amyloid (Aβ) in vivo. In a SILK study, a solution of ¹³C₆-labeled leucine (¹³C₆-Leu) solution is given intravenously to the subject. Aliquots of cerebrospinal fluid (CSF) are acquired at various timepoints after initiation of the ¹³C₆-Leu infusion. Amyloid β is immunoprecipitated from the CSF samples and digested into smaller peptides prior to multiple reaction monitoring mass spectrometry (MRM-MS) analysis. The ratio of the ¹³C₆-labeled peptide to its unlabeled counterpart is plotted to calculate the FSR and FCR of Aβ in the subject.

The MRM-MS data yields the quantitative ratio between the ¹³C₆-Leu labeled and unlabeled Aβ peptides. During the early time points (i.e., between about hours 8–16), this ratio increases as the pool of ¹³C₆-Leu in the cell increases [77]. Measuring the slope of this increase allows the FSR to be calculated. Later time points see a drop in the ¹³C₆-labeled/unlabeled ratio as the pool of ¹³C₆-Leu is diminished (i.e., between about hours 20–28), and proteins that had incorporated the heavy amino acid start to be degraded. The FCR is calculated by measuring the slope of this decreasing ratio. Any changes in the FSR or FCR could potentially lead to a diagnosis of Alzheimer’s disease, a notoriously difficult condition to diagnose in early stages - when intervention would have the greatest impact in preventing its progression.

4.2 Microbial identification

Bacteremia, a bacterial infection in blood, is a serious condition as it allows the bacteria to spread to any part of the body [78]. The infection can lead to sepsis, a dysregulated response to an infection, and remains a major cause of hospital morbidity and mortality [79]. The conventional method of identifying bloodstream infections require overnight subculturing followed by identification of the relevant species and antimicrobial susceptibility testing. Although automated, this process still requires about 18–24 hours to complete [80]. The MALDI time-of-flight (MALDI-TOF) MS procedure can provide identification results in less than an hour using bacterial pellets obtained from blood culture vials. This reduction in time can have a tremendous impact on survival of patients with sepsis. Since there is an inverse relationship between the time to sepsis diagnosis and patient mortality, decreasing the time required to identify bacteria-specific infections would be expected to have a major beneficial impact on public health [81, 82, 83].

Arguably the most successful application of MS in the clinic is the use of MALDI-TOF MS for the identification of microorganisms. While the identification of bacteria using MS dates to the early 1970’s, the techniques then relied mainly on the detection of lipids of microorganisms grown in agar [84]. It was in the 1990’s that the ability to obtain MALDI-TOF MS profiles of the cell contents of bacteria after sonication was demonstrated [85]. This ability led to the generation of “fingerprints” that could be assigned to specific bacterial types, which enabled rapid identification of bacteria using MS [86]. MALDI-TOF MS methods are becoming increasingly commonplace in clinical microbiology laboratories, because of its capability to reduce identification times by over 75% [87].

Sample preparation in MALDI-TOF MS procedures is simple [88]. Briefly, 1 mL of blood culture broth is mixed and incubated for 2 minutes with an extraction strip that contains a lysis buffer (Figure 7). This treatment lysis red blood cells but allows microorganisms to remain intact. The microorganisms are then captured onto a membrane surface by immersing a filter wand into the lysate and applying a vacuum. After washing using a series of buffers, the microorganisms attached to the filter are transferred to a MALDI plate target spot by by tapping the filter onto the surface of the plate. Cyano-4-hydroxycinnamic acid and formic acid are then applied to the spots containing the microbial sample, which is allowed to dry prior to MALDI-TOF MS data acquisition.

Figure 7.

Determination of microbes in blood using MALDI-MS. A buffer is added to blood to lyse red blood cells followed by the insertion of a filter wand. Application of a vacuum to the wand draws the microbes to its surface, where they are washed prior to being spotting on a target of a MALDI plate. After addition of the MALDI matrix, the mass spectrum of the microbes is recorded. To identify the unknown microbe, the recorded spectrum is compared to a reference library of spectra acquired of known microbes.

After acquisition of the MALDI-TOF MS spectra, microorganisms contained within the sample are identified by matching the acquired protein and peptide peak profiles to reference profiles contained within annotated databases. A previous study showed that the proteins detected using this method are primarily composed of ribosomal, DNA-binding, and cold shock proteins [89].

5. Conclusions

The past 30 years has seen a major shift in protein science. Before 1990, protein science was almost exclusively a “one-at-a-time” process in which a feature (e.g., abundance, size, structure, binding partner, etc.) of a single protein was studied per experiment. With the advent of ESI and the coupling of LC with MS, the relative abundance of thousands of proteins could now be compared in complex biological samples. This capability accelerated the growth of proteomics and resulted in several investigators using this technology to search for biomarkers of various diseases. Unfortunately, as the data accumulated it became evident that potential biomarkers could only be validated if their absolute abundance could be measured.

The obvious competitor to MRM-MS assays are antibody-based methods such as ELISAs. ELISAs are routinely performed and with the use of standards, can provide results that measure the absolute abundance of a specific protein. However, many scientists argue that MRM-MS methods are superior owing to their specificity. Not only can antibody measurements suffer from cross-reactivity, but there are not antibodies available for every protein isoform currently known. While there are some protein isoforms that MRM-MS may not be able to measure, such as specific phosphorylated forms or those with very high homology, it is generally conceded that MRM-MS assays can be developed for a broader range of proteins and other types of biomolecules.

While MRM-MS does have several advantages over antibody-based techniques, it is not without its disadvantages. MRM-MS assays may not always have the sensitivity to match the endogenous level of the target protein within its biological setting. In addition, the signal produced by MS is often negatively impacted by the presence of other components in the mixture. Therefore, extensive prefractionation or molecule-specific enrichment of serum, plasma, and tissue samples may be necessary to optimize the LLOQ of an MRM-MS assay. If the goal is to develop an MRM-MS assay that measures a panel of different proteins, the fractionation and/or enrichment steps need to be applicable to all the targeted species. Incorporation of any additional steps not only decreases the throughput but introduces additional analytical variability. Since antibody-based methods generally do not require additional prefractionation or enrichment steps, their throughput is more constant.

Another disadvantage of MRM-MS methods is cost. As MRM-MS capabilities increased with the development of more sensitive, higher resolution mass spectrometers, the cost of these instruments likewise increased. It is not uncommon for a typical LC–MS system necessary for conducting MRM-MS assays to cost more than 400,000 U.S. dollars. Along with the costs of reagents (especially stable isotope standards) and software, MRM-MS represents a significant cost per analysis. While MRM-MS has multiplexing capabilities, only one sample can be run per instrument, whereas hundreds (or thousands) of samples can be analyzed on a single ELISA plate. Increasing sample throughput for an MRM-MS assay requires purchasing additional LC/MS systems, which is an additional major capital equipment expense.

MRM-MS methods for measure the absolute quantity of proteins will continue their rapid progress and likely become routinely used in diagnostic and prognostic medicine. Their ability to measure multiple physiological biomarkers in easily acquired biological samples represents a tremendous advancement in diagnostic medicine. Imagine giving a blood sample at an annual physical and having it analyzed for biomarkers that are diagnostic for various cancers, myopathies, cardiomyopathies, neurological disorders, inflammation, etc. Routine detection of these conditions at an early stage would have an enormous impact on life expectancy. This scenario is the ultimate advantage of MRM-MS assays over antibody-based tested in future medicine; no hypothesis would be required to identify a current disease condition. Much like how genome-wide sequencing is becoming highly accessible to the public, quantitative protein screening via MRM-MS assays will soon follow.

Acknowledgments

The authors would like to thank Cedarville University for support during the preparation of this chapter.

Conflict of interest

The authors declare no conflict of interest.