High throughput proteomics enabled by a fully automated dual-trap and dual-column LC-MS

Abstract

This technical note describes a dual-column liquid chromatography system coupled with mass spectrometry (LC-MS) for high throughput bottom-up proteomic analysis. This system made full use of two 2-position 10-port valves and a binary pump with an integrated loading pump of a commercial LC to provide successive operation of two parallel subsystems. Each subsystem consisted of a set of trap column and analytical column. A T-junction union was used to split the mobile phase from the loading pump into two parts. This allowed one set of columns to be washed and equilibrated, followed by the injection of the next sample while the previous sample was eluting and being analyzed on the other set of columns, thereby greatly increasing the analysis throughput. This approach showed high reproducibility for the analysis of HeLa tryptic digests with the average relative standard deviation (RSD) of 1.75%, 6.90%, and 5.19% for the identification number of proteins, peptides, and peptide-spectrum matches (PSMs), respectively, across 10 consecutive runs. The capacity for peptides and proteins identification, as well as proteome depth, of the dual-column LC system was comparable to conventional single-column system. Due to its simple equipment requirements and fabrication process, this method has seen to be highly accessible for other laboratories.

INTRODUCTION

Proteins play a crucial role in living organisms, participating in various cellular functions such as catalyzing reactions, transporting molecules, and transmitting messages.^1–3 Analyzing proteins in cells, tissues, organs, and biofluids has provided valuable information to help understand protein-protein interactions,⁴ disease mechanisms,⁵ biomarker discovery,⁶ drug development,⁷ and therapeutic intervention assessment.⁸ Liquid chromatography-tandem mass spectrometry (LC-MS)-based proteomics has been widely recognized as a highly effective tool for in-depth proteomic profiling of biological and clinical samples due to its robustness, flexibility, high sensitivity, high separation resolution, and capacity to provide molecular weight and structural information.^{9, 10}

In recent years, mass spectrometric technologies have significantly improved in data acquisition speed, sensitivity, mass resolving power, and mass accuracy.^10–14 However, throughput has been a longstanding challenge in MS-based proteomics due to lengthy sample processing and analysis times. This issue is even more critical in single-cell proteomics, where a large number of samples must be analyzed for statistically relevant results.^{10, 15, 16} Therefore, the development of high-throughput MS-based proteomic methods is crucial. Efforts have been made to increase analysis throughput, primarily through the use sample multiplexing,^17–20 multi-column LC-MS systems^21–27 and improved separation efficiency.^{28, 29} Multi-column LC-MS systems show promise as they can be combined with other strategies to improve analysis throughput.

Among these approaches, multi-column LC-MS systems can be combined with other methods to enhance analysis throughput further and considered to be one promising strategy. Recently, Webber et al. developed a dual-trap and dual-column nanoflow LC platform that enabled profiling of up to 200 single-cell proteomes daily.²¹ This platform achieved a duty cycle (the fraction of time acquiring useful data) ranging from 58% to 83% when the overall cycle time increased from 7 to 30 minutes. Kreimer et al. reported a dual trap single-column method that achieved a 90% duty cycle with a throughput of 15 min per sample.²² In another method developed by Livesay et al., a nearly 100% duty cycle was achieved using a four-column LC system.²⁶ However, this approach requires 11 two-position valves to control the system, making the configuration and operation highly complex. This issue has also been encountered in other studies where more valves were needed than the LC instrument could accommodate.^{23, 30–32} In some previously mentioned methods, multiple emitters and custom-pulled columns are directed at the MS inlet to maximize the analysis throughput. During operation, one emitter’s voltage is typically kept on to ionize the sample, while the voltage for the other emitter is turned off to prevent mutual interference. However, this can lead to a problem during washing of the column in which the voltage is off, as the absence of ionization can cause contaminants to accumulate on the spray tip, leading to unstable ionization and affecting the quality of the subsequent analysis.

In this study, we present a versatile dual-column LC system that utilizes two 2-position 10-port valves and a binary pump with an integrated loading pump from a commercial LC system. This system is controlled using commercial LC software without additional software or devices. It consists of two sets of trapping and analytical columns and shares a single ESI emitter, eliminating the need for aligning multiple emitters with the MS inlet simultaneously. The system can automatically perform consecutive operations with minimal dead time between runs, resulting in high throughput due to the elimination of sample injection, column washing, and equilibration time. Additionally, it reduces the risk of contamination to the mass spectrometer by connecting the channel to a waste bottle during column washing. This dual-column LC system can be easily implemented in most commercial LC systems, making it highly accessible for other laboratories.

EXPERIMENTAL SECTION

Reagents and Materials.

Tris(2-carboxyethyl)phosphine hydrochloride (TCEP), Igepal CA-630, urea, Trizma^® base, Sodium chloride (NaCl), Ethylenediaminetetraacetic acid (EDTA), HPLC grade acetone and chloroform, hydrochloric acid (HCl), trifluoroacetic acid (TFA), and formic acid (FA) were purchased from Sigma-Aldrich (St. Louis, MO, USA). HeLa Protein Digest Standard (Cat. 88328) was purchased from Thermo Fisher Scientific (Waltham, MA, USA). LC-MS grade acetonitrile (ACN) and methanol (MeOH) were purchased from Honeywell Burdick & Jackson (Muskegon, MI, USA). Ultrapure water was prepared using a Milli-Q system (Millipore, Bedford, MA, USA).

Dual-column LC System Configuration and Operation.

The schematic of our proposed dual-column two-LC system is depicted in Figure 1. The system was configurated using a DIONEX UltiMate 3000 RSLCnano system (ThermoFisher Scientific, San Jose, CA, USA). This system consisted of two subsystems connected using two 2-position 10-port valves. Each subsystem contained a parallel set of trapping column and analytical column. The two subsystems shared a sampling system, a loading pump, a gradient pump and an ESI emitter. The loading pump was operated at microflow rate (30 μL/min) for sampling loading, column washing and column re-equilibration. The gradient pump was operated at nanoflow rate (300 nL/min) for the gradient elution of samples. A flow-splitter (T junction union) was applied to the loading pump to divide the flow into two channels with one channel connected to the position 3 of the sample injection valve for sample injection and trap column washing and the other channel connected to position 5 of valve 2 for analytical column washing. This system allowed one subsystem to elute and analyze a sample while the other subsystem was washing and regenerating the columns and injecting another sample simultaneously

Figure 1.

Schematic of the dual-column LC system. The system consists of two subsystems with each subsystem containing a set of trap column and analytical column. The two subsystems share the sample injection component, loading pump, gradient pump, ESI emitter and mass spectrometer. (A) Sample is eluting from Trap 2 and Column 2 and analyzing with mass spectrometer. Meanwhile, Trap 1 and Colum 1 is under washing and equilibration, followed by injection of the next sample and transmission of the sample to Trap 1. (B) The roles of two sets of columns are reversed.

PepMap 100 C18 column (5 μm, 0.3 × 5 mm, Thermo Fisher Scientific, Waltham, MA, USA) was used as the trap column. Acclaim PepMap RSLC C18 column (2 μm, 75 μm × 15 cm, 100 Å, Thermo Fisher Scientific, Waltham, MA, USA) was used as the analytical column. Solvent A (water/FA, 100/0.1) and solvent B (ACN/water/FA, 80/20/0.1) were used as the mobile phases for loading pump. Solvent C (water/FA, 100/0.1) and solvent D (ACN/FA, 100/0.1) were used as the mobile phases for gradient pump.

Mass Spectrometry analysis.

The MS analyses were performed using a Thermo Q Exactive HF quadrupole-Orbitrap mass spectrometer equipped with a nanospray ESI source (ThermoFisher Scientific, San Jose, CA, USA) in positive mode with the spray voltage of 1.8 kV and capillary temperature at 250 °C. The MS1 scan range was 200–2000 m/z with MS1 resolution of 60, 000 and MS1 automatic gain control (AGC) target of 3E6. The 15 most intense ions were selected for MS2 analyses with MS2 resolution of 30,000 and MS2 AGC target of 1E5. The maximum injection time for both MS1 and MS2 was set at 100 ms. The normalized collisional energy (NCE) of the higher-energy collisional dissociation (HCD) was set at 30% for ion fragmentation. Mass spectrometry data were acquired using data-dependent acquisition (DDA) mode.

Data Analysis.

RAW files were searched with MSFragger and performed against UniProt human reference database (released February 2022). The precursor and fragment mass tolerances were set at 20 ppm. The maximum missed cleavage was 2. The peptide length was 7–50 with the peptide mass ranged from 500 to 5000. Carbamidomethylation of cysteine residues was set as fixed modification and N-terminal acetylation and methionine oxidation were set as variable modifications. False discovery rate (FDR) was set at 1% at protein level.

RESULTS AND DISCUSSION

In this study, we introduce a dual-column LC system combined with MS to enhance proteomic analysis throughput. This innovative approach fully utilizes a commonly available commercial UltiMate 3000 RSLCnano system, including two 2-position 10-port valves, a loading pump, and a gradient pump. Compared to existing high-throughput proteomic analysis methods,^{21, 23–27, 30, 32} the construction of our proposed system is relatively simple. As outlined in the experimental section, this system only requires two trap columns, two analytical columns, LC, and MS, without the need for additional power supplies, sample loops, valves, ESI emitters, or specialized software. These advantages make it highly accessible for other laboratories.

In this system, a T-junction union splits the mobile phase from the loading pump into two parts. One end is used for washing and equilibrating the trap column, as well as for sample injection, while the other end is used for washing and equilibrating the analytical column. Simultaneously, another set of trap and analytical columns are connected, and samples are eluted and analyzed. To minimize dead time between samples, a 2-step approach is implemented for switching between the sets of trap and analytical columns. For example, during one run, trap 2 and column 2 are connected for sample analysis (Figure S1A). The subsequent sample is injected four minutes before the current run's completion to avoid waiting time for sample injection (Figure S1B). 0.5 minutes before the end of the run, valve 1 is switched while keeping valve 2 unchanged. This allows the transition of the next sample from trap 1 to position 3 of valve 2 while the previous sample is still being analyzed in column 2 (Figure S1C). After the previous run is completed, valve 2 is switched to connect trap 1 with column 1, and the sample between positions 1 of valve 1 and position 3 of valve 2 is transferred to column 2 for analysis (Figure S1D). Further operational details, including valve positions and gradient profiles, are presented in Figures S1-S2. Notably, during column washing, the channel is connected to a waste bottle, ensuring that contaminants such as surfactants are eluted into the waste rather than being transferred to the mass spectrometer. This helps to reduce the risk of contamination to the instrument.

The reproducibility of this dual-column system, both between runs and columns, was assessed by analyzing 10 consecutive runs of 500 ng HeLa tryptic digested peptides. As illustrated in Figure 2A, B, the system displayed excellent robustness and reproducibility, evidenced by consistent heatmap, peak shapes and retention times with minimal variation across runs and columns. Moreover, the average relative standard deviation (RSD) values for the identification number of proteins, peptides, and peptide-spectrum matches (PSMs) across 10 replicates in columns 1 and 2 were 1.75%, 6.90%, and 5.19%, respectively, further substantiating the system's reliability (Table S1). It is worth noting that the consistency of the two sets of columns also affects the RSD, with highly consistent columns resulting in lower RSD values.

Figure 2.

(A) Heatmap and (B) Chromatogram of 10 consecutive analysis of 500 ng HeLa tryptic digested peptides with dual-column system. (C) Schematic description of the operation steps of dual-column system. (D) Heatmap and (E) Chromatogram of a single run of 500 ng HeLa tryptic digested peptides using single-column system.

Prior studies have indicated that sample injection, column washing, and column equilibration can take up to 50% of the total analysis time in a single run using a conventional single-column platform.^{25, 33} In contrast, this dual-column LC method reduces the time needed for these steps, allowing for more samples to be analyzed within a given time frame. As demonstrated in Figure 2C, only the first sample required waiting time for sample injection. Starting from the second sample, sample injection, column washing, and column equilibration occurred concurrently with the elution and analysis of the previous sample. Specifically, the dual-column LC system required an overall analysis time of 60 minutes for a 60-minute elution gradient, while the single-column system needed approximately 93 minutes (including 3 minutes for sample injection and 30 minutes for column washing and equilibration) (Figure 2D, E). By adopting the dual-column LC approach, the analysis throughput increased by over 35% for a 60-minute gradient, and a duty cycle of nearly 90% was attained when analyzing large-scale samples consecutively. For a shorter gradient such as 30-min, the analysis throughput can be increased by more than 50%. Importantly, using narrower columns and capillaries can further boost analysis throughput by minimizing the dead time between runs.

The dual-column LC system's effectiveness in identifying peptides and proteins was compared with the single-column LC system. In the dual-column system, an average of 1865 proteins, 8148 peptides, and 13979 PSMs were identified from 500 ng of HeLa tryptic digested peptides in 10 replicates. The single-column system identified 1792 proteins, 8084 peptides, and 14346 PSMs in column 1 and 1899 proteins, 8809 peptides, and 14460 PSMs in column 2 (Figure 3A). Pairwise analysis of the dual- and single-column systems revealed high correlation coefficients, with Pearson's correlation coefficients ranging from 0.86 to 0.92 across 10 consecutive runs using the dual-column LC system and from 0.84 to 0.92 between 10 consecutive runs obtained from dual-column systems and single runs from single-column systems (Figure 3B). Furthermore, the upset plot analysis showed that the identified proteins had a high degree of overlap, with a maximum of 1.15% of the proteins being unique to a single run when using the dual-column LC system and up to 1.36% when using single-column systems (Figure 4). These results suggest that, despite a 35% shorter overall analysis time, the dual-column LC system can identify peptides and proteins with accuracy and proteome depth comparable to that of the conventional single-column system.

Figure 3.

Evaluation of capability for the dual-column LC system. (A) Numbers of proteins, peptides and PSMs identified from 500 ng HeLa tryptic digested peptides with dual- and single-column system. (B) Pairwise correlation of protein identifications across 10 consecutive runs with dual-column system and single runs in column 1 and column 2 with single-column systems.

Figure 4.

Upset plot analysis of the proteins identified from 500 ng HeLa tryptic digested peptides using dual- and single-column systems.

In summary, the dual-column method has proven to be an effective way for high throughput proteomic analysis. This method has shown an increase in analysis throughput by over 35% for a 60-minute gradient, while maintaining accuracy and depth of the proteome analysis comparable to the conventional single-column system. As a proof-of-concept study, the robustness, reproducibility, and effectiveness of the method have been evaluated through analysis of widely used HeLa tryptic digested peptides. Due to its ease of configuration and the absence of any additional system components, this method is highly promising to be used in other laboratories.

CONCLUSIONS

In this technical note, we present a dual-column LC system combined with MS to facilitate high-throughput bottom-up proteomic analysis. This system was developed by fully utilizing two 2-position 10-port valves and the binary pump with an integrated loading pump from a commercial LC system. The dual-column LC system substantially improved analysis throughput by enabling one set of trap and analytical columns to analyze a sample while the other set underwent washing and equilibration. As a proof-of-concept study, the system was applied to the analysis of HeLa tryptic digested peptides. The system demonstrated good robustness and reproducibility. Furthermore, the dual-column LC system's ability to identify proteins and peptides was comparable to that of a conventional single-column LC system. Since no additional equipment or software is needed and the fabrication process is straightforward, this approach holds great potential for adoption in other laboratories.