Dual LC–MS Platform for High-Throughput Proteome Analysis

Oct 3, 2013 - A simple dual online ultra-high pressure liquid chromatography system (sDO-UHPLC) for high throughput proteome analysis. Hangyeore Lee ...
0 downloads 0 Views 2MB Size
Technical Note pubs.acs.org/jpr

Dual LC−MS Platform for High-Throughput Proteome Analysis Dennis J. Orton,† Mark J. Wall,‡ and Alan A. Doucette*,‡ †

Department of Pathology, Dalhousie University, 11th Floor Tupper Medical Building, Room 11B, Halifax, NS B3H 4R2, Canada Department of Chemistry, Dalhousie University, Room 212, Chemistry Building, Halifax, NS B3H 4R2, Canada



S Supporting Information *

ABSTRACT: We describe a dual-column interface for parallel chromatography to improve throughput during LC−MS experimentation. The system employs a high-voltage switch to operate two capillary column/nanospray emitters fixed at the MS orifice. Sequentially loading one column while operating the second nearly doubles the LC−MS duty cycle. Given the innate run-to-run variation of a nanospray LC−MS (12% RSD peak area; 2% retention time), the intercolumn variability of the platform showed no meaningful difference for proteome analysis, with equal numbers of proteins and peptides identified per column. Applied to GeLC analysis of an E. coli extract, throughput was increased using one of three methods: doubling the number of replicates, increasing the LC gradient length, or sectioning the gel into twice as many fractions. Each method increased the total number of identifications as well as detection throughput (number of peptides/proteins identified per hour). The greatest improvement was achieved by doubling the number of gel slices (10 vs 5). Analysis on the dual column platform provided a 26% increase in peptides identified per hour (24% proteins). This translates into ∼50% more total proteins and peptides identified in the experiment using the dual LC−MS platform. KEYWORDS: capillary liquid chromatography, high-throughput proteomics, nanospray ionization, dual-column mass spectrometry interface, duty cycle, parallel chromatography, multiplexed chromatography



INTRODUCTION Advances in instrumentation and associated methodology have expanded the capacity of liquid chromatography coupled to mass spectrometry (LC−MS) for in-depth proteome characterization. Together with improvements in MS resolution, scan speed, and sensitivity,1 the adoption of low flow capillary chromatography with online nanospray2,3 has provided researchers with a high-throughput tool for proteome analysis.4,5 Nevertheless, with reversed-phase chromatography, an extensive fraction of time must still be devoted to sample loading and column re-equilibration; this issue is further exaggerated under low-flow conditions. Given that extensive (multidimensional) fractionation of the proteome is often required for improved characterization, it follows that methods that improve proteomic throughput are an important objective for large-scale proteome analysis. The duty cycle of the LC−MS experiment can be defined as the fraction of time the MS instrument acquires useful data relative to the total analysis time. Duty cycle can therefore be improved by decreasing any “down time” experienced, for example, during column regeneration and sample loading. In the capillary LC−MS experiment, researchers have adopted various strategies to permit a higher rate of sample loading. The simplest perhaps is to inject the smallest possible volume of sample. However, this strategy necessitates reconstitution of the peptide mixture in a very low volume of solvent, which challenges the precision of solvent delivery as well as the © 2013 American Chemical Society

complete resolubilization of the sample. Increasing the injection volume implies that a greater proportion of the sample can be injected, although longer injection times will be required. At 250 nL/min, a 10 μL injection would take a minimum 40 min to load the sample. Trap column configurations can be employed, as they can accommodate higher flow rates, thereby dramatically decreasing the required loading time. If two pumps are employed (a high-flow pump for loading and a low-flow gradient pump to run samples), a trap column configuration offers the advantage of sequential sample loading and column regeneration while operating the alternate the analytical column. Unfortunately, elution of analytes through a trap column and onto the analytical column will cause peaks to broaden, thereby reducing separation efficiency and impacting proteome characterization.6 To preserve the advantage of the trap column in terms of sample cleanup (desalting), we may employ an offline column configuration.7,8 This also has the distinct advantage of being able to quantify the concentration of analyte in the sample to optimize the amount of sample injection for LC−MS. Trap column configurations are perhaps the most commonly adopted form of multiplexed chromatography for proteome analysis. Various instrument platforms, including commercial configurations, have been described to distribute MS Received: July 17, 2013 Published: October 3, 2013 5963

dx.doi.org/10.1021/pr400738a | J. Proteome Res. 2013, 12, 5963−5970

Journal of Proteome Research

Technical Note

acquisition among multiple column configurations.9−16 MUX interface couples multiple columns having independent electrospray nozzles to a single MS detector by rapid sampling from each column as the remaining emitters are blocked.15,17 Adaptation of this technology into a nanospray configuration has not been reported, making MUX technology of limited use for proteome analysis. In 2001, Smith et al. described the first true multiplexed configuration of capillary chromatography coupled to mass spectrometry,18 which was automated in 2004.19 This system demonstrated the potential for sequential regeneration of one column with sample loading and elution from another. Smith et al. later adapted the system into an automated LC−MS platform approaching 100% duty cycle.20 The later system employed four columns operating through two independent gradient pumps, along with four nanospray emitters, being coupled to MS on a moving stage.1,20 The actively moving stage enables the selection of a given spray emitter for MS data acquisition by positioning it in front of the MS orifice. Thibault et al. describe a multiplex platform for high-throughput proteomics that coupled two trap columns to two capillary columns, each within independent emitters also on a moving stage.21 Thibault’s strategy involved rapid cycling of data acquisition across the two simultaneous eluting columns with the moving stage. Characterization of complex proteome mixtures in this fashion may suffer as reduced acquisition time per column lowers the overall sensitivity. In another approach, Hanash et al. coupled four columns (two traps, two capillary columns) to a single nanospray emitter for sequential regeneration and loading on one column and analysis from the second.22 This was the first demonstration of a multiplexed nanoflow platform, wherein flow rates of 250 nL/min were employed. By eliminating the moving nanospray stage, this system provides a more robust approach for high-throughput analysis. However, coupling a single emitter to multiple columns through a switching valve introduced post-column dead volume, causing peak broadening and reducing sensitivity. In this work, we describe a fully automated, simple and robust dual column LC−MS system which nearly doubles the duty cycle over a conventional (single column) format for proteome analysis. The system eliminates the use of trap columns, packing two capillary columns that are integrated to two nanospray emitters which are fixed in position during operation. Voltage is applied to one of the two columns, while the second column is sequentially equilibrated and loaded for a subsequent run. The system is evaluated in terms of column-tocolumn reproducibility and validated for high throughput GeLC-MS analysis of an E. coli proteome extract. The system is a readily adaptable technology for improving throughput in proteome analyses.



Escherichia coli (strain K12) was obtained as a gift from Dr. Andrew Roger (Dalhousie University, Halifax, Canada) and maintained in Luria−Bertani media (L3022) from Sigma (Oakville, Canada). Agar (A1296), trypsin (T802), acrylamide (A3699), ammonium bicarbonate (ABC; A6141), along with formic acid (94318) and trifluoroacetic acid (TFA; T6508) were from Sigma (Oakville, Canada). Tris (161-0719), iodoacetamide (163-2109), dithiothreitol (DTT; 161−0611), and sodium dodecyl sulfate (SDS; 161-0302) were from BioRad (Hercules, CA). Water was purified to 18.2 MΩ·cm and solvents were of HPLC grade and from Fisher Scientific (Ottawa, Canada). Protein Extraction and SDS-PAGE

E. coli was harvested according to established protocols (Qiagen Manual for Good Microbiological Practices). E. coli proteins were extracted by two rapid freeze−thaw cycles, with homogenization using a PelletPestle (Fisher) and final suspension of the cell pellet in 1% SDS. Protein concentration was determined using the Pierce BCA assay (Fisher). A total of 50 μg of each E. coli protein extract was loaded into six lanes and resolved on a 12% T SDS-PAGE gel, followed by staining with Coomassie Brilliant Blue (Bio-Rad, 161-0400). Each lane was processed into five (sample A) or ten (sample B) gel slices (see Supplemental Figure S1 in the Supporting Information) and subject to in-gel trypsin digestion, as previously described.23 Peptides extracted from the gel were subject to automated off-line sample cleanup and quantitation on a reversed-phase column, as previously described.7 The cleaned fractions were dried in a SpeedVac and stored at −20 °C prior to analysis by LC−MS/MS. LC−MS Setup and Analysis

Extracted peptides were subject to LC−MS/MS analysis on an Agilent 1200 nanoflow reversed-phase liquid chromatography system coupled to a ThermoFisher LTQ linear ion trap mass spectrometer (San Jose, CA) through a modified dual column nanospray ionization source (spray voltage 2.5 kV), as described. The power supply from the LTQ nanospray source was routed through a custom designed high voltage switch, which directs voltage to one of the two capillary columns, connected via liquid junction through a 360 μm OD Upchurch MicroTee (Idex Health and Science). Capillary columns (30 cm × 75 μm i.d.) were packed in-house with reversed-phase Phenomenex (Torrance, CA) Jupiter beads (C12, 4 μm, 90 Å pore size) into silica PicoFrit Emitters (New Objectives, Woburn, MA), and connected to the MicroTee. The emitters were mounted to a piece of plastic affixed to the existing xyz stage of the LTQ nanospray interface. Peptides extracted from each gel slice were analyzed using one of two gradients between solvent A (0.1% formic acid/ water) and solvent B (0.1% formic acid/ACN). Samples (10 μL) were loaded onto each column using the Agilent 1050 pump with flow splitting T ahead of the autosampler to reduce the initial 100 μL/min flow from the pump to a rate of 300 nL/ min using a solvent comprising 5% B in solvent A. Peptides were eluted with the Agilent 1200 Nano pump at a flow rate of 250 nL/min. The 65 min gradient was as follows: 0 min, 5% B; 0.1 min, 7.5% B; 45 min, 20.0% B; 57.5 min, 25% B; 60 min, 35% B; 61 min, 80% B; 64.9 min, 80% B; 65 min, 5% B. The 120 min gradient was as follows: 0 min, 5% B; 0.1 min, 7.5% B; 90 min, 20% B; 115 min, 25% B; 120 min, 35% B; 121 min, 80% B; 125 min, 80% B; 125.1 min, 5% B. The LTQ was set to data-dependent mode (MS followed by MS/MS of top three

MATERIALS AND METHODS

Instrumentation and Materials

The two-position six-port valve was a Rheodyne MXP7980 high-pressure nano switch valve (Idex Health and Science, Oak Harbor, WA) equipped with external relay contact control. An Agilent 1200 G2226A Nano pump (Oakville, Canada) was used for gradient elution of peptides from the capillary columns, while an Agilent 1050 quaternary pump was used for sample loading through a G1313A Agilent autosampler, with 100 μL injection loop. A combination of PEEKsil (Upchurch) and silica capillary tubing (PolyMicro Technologies, Pheonix, AZ) was employed, all having inner diameters of 50 μm. 5964

dx.doi.org/10.1021/pr400738a | J. Proteome Res. 2013, 12, 5963−5970

Journal of Proteome Research

Technical Note

peaks). The timing of sample injection and column selection on the dual spray system is described in detail in Supplementary Table S1 in the Supporting Information. Data Analysis

Database searching used the SEQUEST algorithm within the Thermo Proteome Discoverer (v. 1.3) software package. MS spectra were searched against the Uniprot E. coli K12 database (4584 entries, downloaded February 14, 2013). Tolerances were set to 2.0 Da for precursor and 1.0 Da fragment ions. Allowable modifications included static carbamidomethylation (+57.0215 Da) of cysteine residues, dynamic phosphorylation (+79.9663 Da) of serine, tyrosine, and threonine residues, and dynamic oxidation (+15.9949 Da) of methionine, with up to two missed trypsin cleavages allowed. Peptide filters were adjusted to provide a false discovery rate of 1% or less by decoy database searching, and proteins required a minimum of two unique peptides for positive identification. (A spreadsheet of all identified proteins and peptides is provided in the Supporting Information.) Redundant protein identifications from the same peptides were also removed prior to analysis. Spectral counts (SpH) per protein is defined as the total number of peptide identifications per protein.



RESULTS AND DISCUSSION

System Design Figure 1. Schematic of the dual column LC−MS system. The system is composed of two independent pumps: the ‘load pump’ delivers solvent through the autosampler to the column, while the nanoflow ‘run pump’ delivers a solvent gradient to the opposing column. (A) With switch valve in position 1, and voltage to column 1, the run pump operates column 1 while column 2 is equilibrated then loaded with the load pump. (B) In position 2, both the solvent switching valve and the high voltage source are diverted, operating column 2 with the run pump while sample is loaded on column 1. (C) Photos of the system show: (i) the Agilent 1050 ‘load pump’, (ii) Agilent 1200 nanoflow ‘run pump’ atop the autosampler, (iii) dual nanoflow capillary columns, (iv) voltage application through the MicroTee, (v) the dual nanospray emitter tips in front of the MS source, wherein the upper column is in run mode while solvent builds on the end of the second tip during column equilibration and sample loading.

Here we describe a robust, low-cost, fully automated dual capillary LC-nanospray MS platform for high-throughput proteome analysis. The system maximizes the utility of the mass spectrometer through sequential loading of sample onto one of two analytical columns while acquiring data using the other column, thereby improving LC−MS duty cycle relative to a single column platform. Figure 1 provides a schematic of the dual column platform. An Agilent 1200 Nano flow binary pump (annotated ‘run pump’ in Figure 1) provides a stable flow of 250 nL/min for analyte elution from the capillary column (e.g., ‘Column 1’ in Figure 1A). Simultaneously, sample loading to the alternate column is provided through an Agilent 1050 isocratic pump (annotated as ‘load pump’ in Figure 1), connected to an Agilent 1100 autosampler. The load pump need not directly possess low-flow capabilities, as a simple splitting T will maintain a flow near 300 nL/min. At this flow rate, 65 min is sufficient to reequilibrate the column and load up to 10 μL of sample prior to column switching. A single two-position six-port switching valve (