Skip to Main Content

Targeted Proteomics

Quantitative mass spectrometry-based proteomics approaches fall into two categories: discovery and targeted. Discovery methods, such as Data-Dependent Acquisition (DDA) and Data-Independent Acquisition (DIA) [e.g., label-free quantitation (LFQ, Asara et al., 2008)], maximize peptide identifications and quantifications per run. In contrast, targeted approaches focus on predefined peptides of interest, optimizing chromatographic separation and data acquisition for higher sensitivity and throughput. While untargeted MS is a valuable tool for the discovery of novel biomarker candidates, targeted MS is better suited for biomarker assay development and validation due to its more focused “target(s)” that allows for increased sensitivity, accuracy, and precision (Wenk et al., 2024). Targeted approaches include Multiple Reaction Monitoring (MRM) (Kulyyassov et al., 2021), also known as Selected Reaction Monitoring (SRM) (Shi et al., 2012), and Parallel Reaction Monitoring (PRM) (Rauniyar et al., 2015, Kulyyassov et al., 2021). Targeted approaches offer two key advantages: they ensure consistent peptide quantification across runs, eliminating “missing values” that hinder statistical analyses in discovery datasets, and they mitigate the inherent DDA bias toward highly abundant peptides.

Depiction of typical MRM, PRM, and DIA analyses

Figure 1: Depiction of typical MRM, PRM, and DIA analyses. In MRM a specific peptide precursor ion is selected in Q1 and fragmented in q2. Selected product ions (typically three) from the target peptide are detected in Q3. In PRM the first two steps are similar to MRM; however, instead of selecting specific product ions, all fragment ions from the target peptide are detected using a HR/AM mass analyzer. In DIA, an “untargeted” approach, all peptide ions within a predefined m/z window are isolated in Q1, fragmented together in q2, and the complex mixture of product ions is detected by an HR/AM mass analyzer. This process is repeated sequentially across the full m/z range. (From Shi et al. 2016, with permission).

As shown in Figure 1, MRM is performed on a triple quadrupole mass spectrometer (QqQ MS; Shi et al., 2012) that employs two stages of mass selection and operates with a relatively long dwell time (e.g., 10 milliseconds per precursor/product ion pair) within a narrow m/z window (e.g., ±0.02 m/z). This results in exceptional selectivity and sensitivity, surpassing full-scan global proteomics by at least one to two orders of magnitude (Shi et al., 2012). MRM offers high-throughput analysis, a linear dynamic range of up to 5 orders of magnitude, and the ability to quantify ng/ml concentrations of preselected peptides from tryptic digests of complex samples, including plasma (Stahl-Zeng et al., 2007). In collaboration with the Keck Laboratory’s Proteomics Resource, the Targeted Proteomics Core (TPC) of the Yale/NIDA Neuroproteomics Center developed a robust 90-minute LC-MRM assay for mouse/rat postsynaptic density fractions, quantifying 337 peptides from 112 proteins based on 15 observations per protein (Colangelo et al., 2015). Given the advantages of PRM for biomarker protein quantification (see below), we now primarily recommend using MRM for small-molecule analysis. For instance, the TPC utilized MRM to measure cotinine, the primary nicotine metabolite, in human cortex samples (McClure-Begley et al., 2016).

More recently, hybrid mass spectrometers have been developed that are suitable for PRM. In these instruments the third quadrupole is replaced with a high-resolution and accurate-mass (HR/AM) analyzer. Examples include quadrupole-Orbitrap (e.g., Q Exactive, Orbitrap Fusion Tribrid) and quadrupoletime-of-flight (e.g., Triple-TOF) instruments. As shown in Figure 1, PRM resembles MRM in peptide precursor isolation and fragmentation, but unlike MRM—where only predefined product ions (transitions) are monitored—PRM acquires full MS/MS spectra for each precursor, enabling the parallel detection of all product ions in the HR/AM mass analyzer. This distinction gives PRM higher selectivity than MRM in complex biological samples, allowing better discrimination of target peptide signals from co-eluting interferences (Domon et al. 2015). While PRM and MRM exhibit similar linearity, dynamic range, precision, reproducibility, and sensitivity (Schiffman et al., 2014; Rauniyar, 2015; Ronsein et al., 2015), PRM is preferred due to the superior resolution and accuracy of the mass analyzers employed for such analyses (e.g., Orbitraps and TOFs) compared to the QQQs typically used for SRM/MRM. The enhanced resolving power of PRM improves quantification by better distinguishing analytes with similar m/z ratios, while greater mass accuracy increases confidence in ion assignment (Gallien et al., 2013; Gallien and Domon, 2015). Key advantages of PRM are its >200 fold increased resolution and 50-fold greater mass accuracy, enhancing selectivity and confidence in protein identification (Table 1).

Table 1: Comparison of MRM vs PRM
Description MRM1 PRM2
Q1 used to isolate precursor peptide Yes Yes
Max #Peptides quantified ~500
Instrument Used in Keck Lab Triple Quad/Linear Ion Trap AB Sciex 4000 QTRAP QqQorbi Fusion Tribrid
Quantitation approach for product ions Sequentially monitor 2–5 transitions/peptide Monitor all transitions in parallel then select 4–10 transitions/peptide for post-run integration
Resolution 0.7 Da (285 @ m/z = 200) 0.0033 Da (60,000 @ m/z = 200)
Mass Accuracy 250 ppm 5 ppm
¹MRM data from AB Sciex instrument specs.
²PRM data from Gallien & Domon (2015) and Shi et al. (2016)

Bezstarosti et al. (2024) demonstrated the high sensitivity of PRM-based MS, surpassing immunoblotting in both sensitivity and quantitative accuracy. The detection limit for proteolytic peptides of a purified target protein reached the mid- to low-attomole range, increasing by an order of magnitude in complex biological matrices. In addition, PRM is a two-step process that changes the paradigm of targeted experiments by decoupling acquisition and data processing. PRM simplifies method development by obviating the need to select transitions prior to data acquisition. Since full MS2 scans are collected for each peptide of interest, each peptide can be interrogated by 4-10 interference-free transitions that can be chosen post-run, as opposed to the 2-5 pre-selected ions typically used for MRM. As a result, the best interference-free PRM transitions that will be used for quantitation can be chosen iteratively after data acquisition. Method development and data acquisition are thus simplified, with only the precursor ion m/z and elution time of each targeted peptide being required to develop a PRM assay. Balancing cycle time and resolution for large-scale peptide targeting is achievable through intelligent acquisition strategies that enhance sensitivity, specificity, and quantitative accuracy (van Bentun & Selbach, 2021). These methods optimize MS/MS acquisition by restricting scans to anticipated elution windows, enabling quantification of more peptides. Key strategies include: (1) retention time-based scheduling (e.g., Picky, iRT, MaxQuant.Live) to optimize peptide acquisition timing and (2) spike-in triggered acquisition (e.g., SureQuant, Pseudo-PRM, TOMAHAQ, Scout-MRM), where labeled synthetic peptides initiate targeted scans. Both approaches improve mass spectrometer efficiency, either increasing sensitivity or expanding the number of detectable targets per run (van Bentun & Selbach, 2021). The use of internal, stable isotope-labeled standards enables MRM and PRM to provide reproducible and accurate quantification of target small molecules and proteins, respectively, across many samples. In PRM, at least two "heavy" stable isotope-labeled surrogate tryptic peptides per protein of interest are usually synthesized. The heavy peptides are spiked into the sample at known concentrations prior to LC-PRM analysis. Quantification is achieved by comparing the signal intensities of heavy internal standard peptides to their corresponding "light" analyte counterparts. The protein quantity is determined from the mean concentration of its surrogate peptides. PRM's quantification scale is limited to ~500 peptides or ~125 proteins per analysis (Gallien et al., 2015), assuming two surrogate peptides per protein, each in light and heavy forms. This level of concurrent quantification requires precise LC retention time control during the LC-MRM run. The TPC developed a scheduled PRM assay to quantify 50 mouse postsynaptic density (PSD) proteins using 138 stable isotope-labeled synthetic peptides along with their 138 naturally occurring “light” counterparts (Wilson et al., 2019). This assay analyzed 1,705 fragment ions, averaging 6.2 fragment ions per peptide and 2.8 peptides per protein, yielding ~34 data points/protein.

To address the limitations in multiplexing, DIA was introduced for proteome-wide quantification of target proteins. This approach uses high-specificity DIA to generate comprehensive product ion maps that are then subjected to targeted data extraction based on the MRM concept. In DIA, wide precursor acquisition windows are predefined to span the entire m/z range of proteolytic peptides. All peptides within each window are fragmented, generating high-accuracy product ion spectra for all detectable peptides. This results in highly multiplexed spectra over the LC elution time, requiring advanced data processing and interpretation. DIA relies on either a targeted (e.g. spectral library based) or a non-targeted (e.g. library free) data extraction strategy. In the targeted strategy, utilizing a priori information—such as peptide retention time and product ion intensity—from spectral libraries for confident peptide identification. The library free approach typically utilizes a predictive algorithm/model to search the DIA data; these often includes machine learning tools. DIA has lower sensitivity, specificity, and reproducibility than MRM and PRM due to shorter dwell times, broader precursor isolation windows, and the lack of internal standards (Gillett et al., 2012). MRM, for example, offers at least 10-fold higher sensitivity. Both PRM and DIA can validate post-translational modifications. Partnering with the Keck Proteomics Resource, the TPC used DIA to quantify 2,134 mouse PSD proteins (Wilson et al, 2019) and phosphorylation at 29 Kalirin 7 (Kal7) sites in rat nucleus accumbens and prefrontal cortex following cocaine exposure (Miller et al., 2017). The TPC developed a DIA assay to examine G-CSF’s effect on attenuated cocaine-seeking, quantifying 4,410 proteins in the rat nucleus accumbens (NAc) and 5,519 in the medial prefrontal cortex (mPFC), with G-CSF altering 39 (0.9%) proteins in the NAc and 409 (7.4%) in the mPFC. To validate these results, the TPC devised a PRM assay that independently quantified 15 key mPFC proteins identified as downregulated by DIA during abstinence. The resulting PRM data confirmed the DIA findings with a median percent difference of just ±0.7% (Hofford et al., 2021).

Literature Cited