A protein carrier-assisted one-pot sample preparation coupled with liquid chromatography (LC) – selected reaction monitoring (SRM) termed cLC-SRM is a convenient method for multiplexed targeted proteomics analysis of small numbers of cells, including single cells. It capitalizes on using excessive exogenous protein as a carrier and high-specificity LC-SRM for targeted quantification.
Protein analysis of small numbers of human cells is primarily achieved by targeted proteomics with antibody-based immunoassays, which have inherent limitations (e.g., low multiplex and unavailability of antibodies for new proteins). Mass spectrometry (MS)-based targeted proteomics has emerged as an alternative because it is antibody-free, high multiplex, and has high specificity and quantitation accuracy. Recent advances in MS instrumentation make MS-based targeted proteomics possible for multiplexed quantification of highly abundant proteins in single cells. However, there is a technical challenge for effective processing of single cells with minimal sample loss for MS analysis. To address this issue, we have recently developed a convenient protein carrier-assisted one-pot sample preparation coupled with liquid chromatography (LC) – selected reaction monitoring (SRM) termed cLC-SRM for targeted proteomics analysis of small numbers of human cells. This method capitalizes on using the combined excessive exogenous protein as a carrier and low-volume one-pot processing to greatly reduce surface adsorption losses as well as high-specificity LC-SRM to effectively address the increased dynamic concentration range due to the addition of exogeneous carrier protein. Its utility has been demonstrated by accurate quantification of most moderately abundant proteins in small numbers of cells (e.g., 10-100 cells) and highly abundant proteins in single cells. The easy-to-implement features and no need for specific devices make this method readily accessible to most proteomics laboratories. Herein we have provided a detailed protocol for cLC-SRM analysis of small numbers of human cells including cell sorting, cell lysis and digestion, LC-SRM analysis, and data analysis. Further improvements in detection sensitivity and sample throughput are needed towards targeted single-cell proteomics analysis. We anticipate that cLC-SRM will be broadly applied to biomedical research and systems biology with the potential of facilitating precision medicine.
Recent technological advances in genomics (transcriptomics) allow for comprehensive and precise analysis of the genome (transcriptome) in single cells1,2,3. However, single-cell proteomics technologies are lagging far behind but are just as important as genomics (transcriptomics) technologies4,5,6,7,8. Furthermore, protein abundance cannot necessarily be inferred from mRNA abundance9, and the proteome is more complex and dynamic than the transcriptome10. Given these challenges, a large number of mixed populations of cells (i.e., bulk cells) are generally used to generate comprehensive proteome data11,12,13. However, such bulk measurements average out stochastic variations of individual cells, thus obscure important cell-to-cell variability (i.e., cell heterogeneity)4,14. Limitations of such bulk measurements become even more severe when the cells of interest only account for a small portion of the total populations of cells (e.g., cancer stem cells within tumors at an early-stage cancer). Therefore, there is a huge knowledge gap between single-cell proteomics and genomics (transcriptomics).
Antibody-based immunoassays (e.g., flow or mass cytometry) are predominantly used for targeted proteomic analysis of single cells6,7,15,16,17,18. However, they suffer from low multiplex, limited specificity, and unavailability of antibodies for new proteins of interest. Mass spectrometry (MS)-based targeted proteomics has emerged as an alternative for accurate protein quantification because of its being antibody-free, high multiplex (≥150 proteins in a single analysis19), high quantitation accuracy (absolute amounts or concentrations), and high specificity and reproducibility (≤10% CV)20,21,22,23. Recent significant progress in sample preparation has made MS-based single-cell proteomics possible for quantitative analysis of highly abundant proteins from single human cells. However, MS-based single-cell proteomics is still at the early infancy stage. For example, the most advanced MS platform coupled with ultralow-flow RPLC flow rates can only allow label-free MS detection and quantification of ~670-870 proteins out of the total ≥12,000 proteins in single HeLa cells24,25.
Currently, there are six MS-based single-cell proteomics approaches available for analysis of single mammalian cells, in which four are for global proteomics (nanoPOTS: nanowell-based Preparation in One pot for Trace Samples26; iPAD-1: integrated proteome analysis device for single-cell analysis27; OAD (oil-air-droplet) chip-based single cell proteomic analysis28; SCoPE-MS: single cell proteomics by mass spectrometry29) and the other two are for targeted proteomics (cLC-SRM: carrier-assisted liquid chromatography (LC) – selected reaction monitoring (SRM)30; cPRISM-SRM: carrier-assisted high pressure, high-resolution separations with intelligent selection and multiplexing coupled to SRM31). However, all these approaches have technical drawbacks. nanoPOTS, iPAD-1, and OAD downscale sample processing volume to 2-200 nL and are not ready for broad benchtop applications26,27,28. For SCoPE-MS, a TMT (tandem mass tag) carrier is added after single-cell processing, so it cannot effectively prevent surface adsorption losses during sample processing when a single tube is used for single-cell processing29, resulting in low reproducibility with a correlation coefficient of only ~0.2-0.4 between replicates32. For cLC-SRM and cPRISM-SRM, using exogenous proteins as a carrier is more suitable for targeted proteomics because peptides from excessive exogenous proteins are frequently sequenced by MS/MS, which greatly reduces the chance for sequencing low abundant endogenous peptides30,31. Unlike global proteomics for relative quantification, the two targeted proteomics approaches can provide accurate or absolute protein analysis of small numbers of cells with high reproducibility using heavy isotope-labeled internal standards at known concentrations. When compared to cPRISM-SRM that requires prior high-resolution PRISM fractionation, resulting in many fraction samples that need to be analyzed, cLC-SRM has a significant advantage in sample throughput without fractionation and can simultaneously quantify hundreds of proteins in a single analysis but with relatively lower detection sensitivity30. Therefore, cLC-SRM is more accessible and should have broader utilities for accurate multiplexed protein analysis of small numbers of cells as well as mass-limited samples.
Herein we describe a detailed protocol to perform cLC-SRM for convenient targeted proteomics analysis of small numbers of human cells, including single cells. The protocol consists of the following major steps: cell sorting by FACS (fluorescence activated cell sorting), cell lysis and digestion processed in low-volume single polymerase chain reaction (PCR) tubes, LC-SRM data collection, and SRM data analysis using publicly available Skyline software (Figure 1). Its broad utility was demonstrated along with our previously well-established SRM assays by absolute targeted quantification of EGFR/MAPK pathway proteins in 1-100 MCF7 or MCF10A cells and determination of pathway protein copies per cell at a wide dynamic range of concentrations30. We anticipate that with the detailed protocol most proteomics researchers can readily implement cLC-SRM in their laboratories for accurate protein analysis of ultrasmall samples (e.g., rare tumor cells) to meet their project needs.
Figure 1: Overview of all steps in cLC-SRM (carried-assisted one-pot sample preparation coupled with liquid chromatography –selected reaction monitoring). Nonhuman cell lysate digests (e.g., Shewanella oneidensis) are used to pretreat PCR tubes for coating tube surface. Small numbers of human cells or single cells sorted by FACS are collected into pretreated PCR tubes. BSA protein (or nonhuman cell lysate proteome) carrier, heavy internal standard (IS), and TFE (or DDM) are added into sample tubes sequentially for facilitating cell lysis and reducing surface adsorption losses. Conceptually, the combined DDM and nonhuman cell lysate proteome carrier will work well for cLC-SRM. Cell lysis is conducted by sonication, and protein denaturation is achieved by heating at high temperature. DTT and IAA reagents are used for reduction and alkylation, respectively (this step is optional). Trypsin is added for digestion with much higher ratios of trypsin over protein amount than that for standard trypsin digestion. The cap of sample tube is removed and then PCR tube is inserted into LC vial for direct LC-SRM analysis. Collected SRM data are analyzed by using publicly available Skyline software. Please click here to view a larger version of this figure.
NOTE: The step-by-step cLC-SRM analysis is shown in Figure 1.
1. Pretreatment of PCR tubes
2. FACS sorting
3. Addition of protein carrier, heavy internal standards, and TFE
4. Cell lysis and protein denaturation
5. Reduction and alkylation (optional)
6.Trypsin digestion
7. Preparation for direct LC-SRM analysis
8. LC-SRM analysis
9. Data Analysis
Small amounts of MCF7 cell lysates (0.5-20 ng equivalent to 5-200 cells) were first used to evaluate the performance of cLC-SRM by targeted quantification of EGFR/MAPK pathway proteins because they are more uniform with less variations when compared to small numbers of cells sorted by FACS. As shown in Figure 2A, XICs clearly shows detection of SRM transitions for ATADDELSFK derived from GRB2 present at ~220,000 copies per MCF7 cell34. cLC-SRM enabled reproducible quantification of endogenous ATADDELSFK down to 5 MCF7 cell equivalents with a S/N ratio of 14 and ~1,800 zmol of quantification sensitivity. The resultant calibration curves displayed excellent linearity with LOQs of 5 cells for high-abundance GRB2 protein and 20 cells for moderate-abundance PTPN11 protein (Figure 2B). The median SRM technical CV for all target peptides across all the data points was ~9%, consistent with the technical reproducibility of standard LC-SRM with CV below 10%20,23,35,36,37,38.
Figure 2: Sensitivity and accuracy of cLC-SRM for multiplexed quantification of EGFR/MAPK pathway proteins. (A) Extracted ion chromatograms (XICs) of transitions monitored for ATADDELSFK derived from GRB2 at different numbers of MCF7 cell equivalents: 548.8/924.4 (blue), 548.8/853.4 (purple), 548.8/738.4 (chestnut). (B) Calibration curves for quantifying high abundance GRB2 and moderate-abundance PTPN11 with the use of the best responsive interference-free transitions, ATADDELSFK (548.8/924.4) for GRB2 and SNPGDFTLSVR (596.9/496.3) for PTPN11. Three and two SRM replicates were performed for 0-10 and 20-200 MCF7 cell equivalents, respectively. (C) Comparison of SRM signal between 10 and 100 MCF7 cells sorted by FACS. Each sample consists of two biological replicates with the addition of ~30 fmol of heavy peptide standards per replicate. XICs of transitions monitored for ATADDELSFK derived from GRB2: 548.8/924.4 (blue), 548.8/853.4 (purple), 548.8/738.4 (chestnut); XICs of transitions monitored for FYGAEIVSALDYLHSEK derived from AKT1: 648.0/897.9 (blue), 648.0/816.4 (purple), 648.0/283.1 (chestnut); XICs of transitions monitored for LPSADVYR derived from SOS1: 460.7/807.4 (blue), 460.7/710.3 (purple), 460.7/404.2 (chestnut). This figure has been modified from Zhang et al30 with the explicit permission from ACS publisher. Please click here to view a larger version of this figure.
cLC-SRM was next applied to measure EGFR pathway proteins in 10 and 100 intact MCF7 cells sorted by FACS. cLC-SRM enabled detection of high- and moderate-abundance proteins in 10 intact MCF7 cells (Figure 2C). Interestingly, low-abundance AKT1 protein (~3700 copies per MCF7 cell) was also detected with the average S/N ratio of 7, suggesting ~60 zmol of absolute sensitivity of cLC-SRM. With the cell number increased to 100, most previously identified important EGFR pathway proteins (22 out of the total 32 proteins which have a wide dynamic concentration range)34 were reliably detected and quantified by cLC-SRM. Furthermore, we have recently tested whether a short gradient time (e.g., 5 min vs standard 45 min) is sufficient for rapid sensitive cLC-SRM analysis. As shown in Figure 3, XICs clearly shows endogenous detection of VLTPTQVK peptide derived from PEBP1 at ~744,000 copies per cell34 in single MCF10A cells sorted by FACS with a S/N ratio of 5 and ~1,240 zmol of quantification sensitivity. As expected, with the cell number increased to 50 and 75, stronger SRM signal was observed with detection of all three transitions, which have the same pattern as its corresponding heavy internal standard (Figure 3A). The calibration curve has linearity with R2 of 0.89 (Figure 3B). Thus, a short gradient is feasible for cLC-SRM presumably because sample complexity for ≤20 ng of tryptic peptides (10 ng BSA and ≤100 human cells ≈ 10 ng proteins) from carrier-assisted small numbers of cells can be effectively addressed by high-resolution capillary RPLC separation with high loading capacity of ≥200 ng. Taken together, these results have shown that cLC-SRM can be used for multiplexed, sensitive, absolute quantification of target proteins in small numbers of human cells including single cells.
Figure 3: An example of targeted proteomics analysis of small numbers of MCF10A cells including single cells with cLC-SRM at short LC gradient. (A) Comparison of SRM signal among 1, 50 and 75 MCF10A cells sorted by FACS. ~30 fmol of internal standard was added to each sample. XICs of transitions monitored for VLTPTQVK derived from PEBP1: 443.3/673.4 (blue), 443.3/572.3 (purple), 443.3/213.2 (chestnut). For single cells, the transition 443.3/213.2 was removed due to severe matrix interference. (B) Calibration curve for quantifying PEBP1 with the use of the best responsive interference-free transition, VLTPTQVK (443.3/673.4). Please click here to view a larger version of this figure.
cLC-SRM is a convenient targeted proteomics method that enables accurate multiplexed protein analysis of small numbers of cells including single cells. This method capitalizes on protein carrier-assisted one-pot sample preparation, in which all steps including cell collection, multistep cell lysis and digestion, and transfer of peptide digests to capillary LC column for MS analysis are performed in one pot (e.g., single tube or single well) (Figure 1). This 'all-in-one' low-volume one-pot processing effectively maximizes the recovery of small numbers of cells for targeted proteomics analysis by greatly reducing possible surface absorption losses.
There are two critical steps for cLC-SRM analysis: 1) after FACS sorting, immediate centrifugation is required to ensure that the collected cells are at the bottom of tubes or wells because a low volume (~15 µL) is used to process small numbers of cells; 2) prior to the addition of trypsin for digestion, the TFE levels need to be reduced down to ≤10% for effective digestion. To avoid drying out samples resulting in unrecoverable loss, frequently checking the sample volume (~4 µL of the remaining volume) is suggested during SpeedVac concentrating. One-pot sample preparation can be further simplified by removal of reduction and alkylation steps with negligible effect on digestion efficiency. One drawback is that cysteine containing peptides cannot be selected as surrogate peptides for target proteins. To avoid the SpeedVac concentrating step for TFE removal we considered replacing TFE with MS-compatible surfactants (e.g., DDM: n-Dodecyl β-D-maltoside) for cell lysis because frequently checking the sample volume severely prevents automation of sample processing with low throughput. This was evidenced by our recent data for DDM-assisted global single-cell proteomics analysis (unpublished). In addition, a single carrier protein (e.g., BSA) may not be sufficient to reduce surface adsorption losses for every protein without bias. This was evidenced by a few undetected proteins at >30,000 copies per cell (e.g., MAP2K1) in 10 MCF7 cells that should have been detected30. Non-human cell lysates presumably can be used as a more effective proteome carrier in contrast to our current BSA carrier because there are many different types of nonhuman proteins (e.g., >3000 proteins for Shewanella oneidensis) with less interference for SRM detection of human proteins. We are working on several different ways to further improve cLC-SRM performance with the potential of moving towards sensitive targeted single-cell proteomics.
When compared to other MS-based single-cell proteomics methods that require specific devices, cLC-SRM can be readily implemented in any MS proteomics laboratory without additional investment on specific devices for sample processing. But it can only be used for quantitative analysis of target proteins of interest. Unlike global single-cell proteomics methods for relative untargeted quantification, cLC-SRM can provide accurate or absolute quantification of target proteins in small numbers of cells including single cells. Furthermore, cLC-SRM has significant advantages over conventional targeted single-cell proteomics using antibody-based immunoassays in terms of multiplexing and quantitation specificity and accuracy. Future developments will focus on significant improvements in detection sensitivity and sample throughput for enabling rapid absolute quantification of most proteins (e.g., important negative feedback regulators) in single human cells. Automation of cLC-SRM is another direction to improve its robustness, reproducibility, and the overall sample throughput with commercially available liquid handlers. We anticipate that cLC-SRM will be broadly used for multiplexed accurate quantification of target proteins in small subpopulations of cells and rare tumor cells as well as mass-limited samples. In turn, it will greatly benefit current biomedical research and systems biology.
The authors have nothing to disclose.
This work was supported by NIH Grant R21CA223715 (TS) and UG3CA256967 (TS and HL). The experimental work described herein was performed in the Environmental Molecular Sciences Laboratory, Pacific Northwest National Laboratory, a national scientific user facility sponsored by the United States of America Department of Energy under Contract DE-AC05-76RL0 1830.
2 mL glass LC vial | Microsolv | 9502S-WCV | Vessel to hold PCR tube for autosample injection |
BSA | Sigma-Aldrich | P0834-10×1mL | Carrier protein for greatly reducing surface adsorption losses |
DTT | Thermo Scientific | A39255 | Reagent for reduction |
Formic acid | Thermo Scientific | 28905 | Reagent for stopping enzyme reaction |
IAA | Thermo Scientific | A39271 | Reagent for alkylation |
Peptide internal standards | New England peptide | Targeted quantification of EGFR/MAPK pathway proteins | |
RT-PCR tube | GeneMate Bioexpress | T-3035-1 | 0.2 mL PCR tube for one-pot sample preparation |
Skyline software | University of Washington | Publicly available for SRM data analysis | |
Sonicator | Hielscher Ultrasound Technology | UTR200 | Sonication on ice for cell lysis |
Speed Vac concentrator | Thermo Scientific | Reduction of the percentage of TFE for effective trypsin digestion | |
TFE | Sigma-Aldrich | 18370-10×1mL | 60% TFE for cell lysis |
Thermocycler w/ heated lid | Peltier Thermal Cycler | PTC-200 | Heating for protein denaturation |
Trypsin Gold | Promega | V528A | Enzyme for protein digestion |
Waters BEH C18 column | Waters | C18 column for peptide separation |
.