Technical Note pubs.acs.org/ac
Novel Parallelized Quadrupole/Linear Ion Trap/Orbitrap Tribrid Mass Spectrometer Improving Proteome Coverage and Peptide Identification Rates Michael W. Senko,† Philip M. Remes,† Jesse D. Canterbury,† Raman Mathur,† Qingyu Song,† Shannon M. Eliuk,† Chris Mullen,† Lee Earley,† Mark Hardman,† Justin D. Blethrow,† Huy Bui,† August Specht,† Oliver Lange,‡ Eduard Denisov,‡ Alexander Makarov,‡ Stevan Horning,‡ and Vlad Zabrouskov*,† †
Thermo Fisher Scientific, 355 River Oaks Parkway, San Jose, California 95134, United States Thermo Fisher Scientific (Bremen) GmbH, Hanna-Kunath Strasse 11, Bremen 28199, Germany
‡
S Supporting Information *
ABSTRACT: Proteome coverage and peptide identification rates have historically advanced in line with improvements to the detection limits and acquisition rate of the mass spectrometer. For a linear ion trap/Orbitrap hybrid, the acquisition rate has been limited primarily by the duration of the ion accumulation and analysis steps. It is shown here that the spectral acquisition rate can be significantly improved through extensive parallelization of the acquisition process using a novel mass spectrometer incorporating quadrupole, Orbitrap, and linear trap analyzers. Further, these improvements to the acquisition rate continue to enhance proteome coverage and general experimental throughput.
C
advances, there is still a mismatch between the number of species observed in survey spectra and those selected for MS/ MS. One of the largest limitations has been the effective MS/ MS acquisition rate, i.e., the production rate of MS/MS spectra which can be matched to a peptide through a database search. Current commercial instruments achieve 6−8 Hz, while it has been suggested that >25 Hz is necessary to address all observable species.12 In this report, we describe a novel mass spectrometer based on the combination of quadrupole, linear ion trap (LT), and Orbitrap technologies. The specific configuration of components allows for the spatial and temporal separation of ion isolation, fragmentation, and detection. Consequently, significant increases are obtained for both survey and MS/MS acquisition rates, producing a notable increase in peptide identification rates for traditional shotgun proteomics experi-
onsiderable progress has been made in the comprehensive analysis of complex protein digests using LC/MS/MS over the last dozen years. Initial attempts at comprehensive proteome analysis of a single organism identified only 1500 out of an expected ∼4000 proteins in baker’s yeast.1 Due to the limited throughput and dynamic range of the 3D trap, this required 1 mg of sample, extensive fractionation, and 80 h of analysis. Subsequent use of a linear trap/Orbitrap hybrid provided nearly complete proteome coverage with 4000 proteins identified from 100 μg of sample in 100 h.2 Most recently, a single dimension LC separation identified 3923 proteins from 4 μg of yeast proteome digested with LysC in 4 h using the latest quadrupole/Orbitrap hybrid.3 To a large extent, these achievements were due to innovations with instrumentation, specifically improvements in nano-LC separations4−6 and Orbitrap-based mass spectrometers.7−9 Recent improvements in sensitivity and spectral acquisition rates allow generation of more data in less time.10,11 The high resolving power and mass accuracy of the survey spectrum combined with rapid, sensitive MS/MS provide a powerful means for investigating complex protein digests. Despite these © XXXX American Chemical Society
Received: October 1, 2013 Accepted: November 19, 2013
A
dx.doi.org/10.1021/ac403115c | Anal. Chem. XXXX, XXX, XXX−XXX
Analytical Chemistry
Technical Note
ments, with nearly complete yeast proteome coverage obtained in two hours.
■
EXPERIMENTAL SECTION LC-MS/MS Analysis. Four micrograms of Saccharomyces cerevisiae digest13 (courtesy of Gygi lab) were separated using a nLC1000 pump (2−25% acetonitrile over 110 min, 300 nL/ minute) with a 500 mm × 75 μm Easy Spray emitter packed with 2 μm particles. All MS data was generated using an Orbitrap Fusion Tribrid mass spectrometer (Thermo Scientific, San Jose, CA). The electrospray ions are introduced into the mass spectrometer through a heated ion transfer tube (275 °C), followed by a stacked ring ion guide (RF-lens)14,15 evacuated by a rotary vane pump to ∼2 Torr. The rear of the guide includes a discharge ion source,16 which can be used for ion− ion reactions17 and/or internal calibration.18 To prevent neutral species and charged solvent clusters from entering the quadrupole mass filter (4 mm r0, 1.1 MHz), the line of sight is interrupted by a curved quadrupole with an axial DC field. After passing through the mass filter, ions travel through the Ctrap8 and are accumulated in a nitrogen filled (5−10 mTorr) multipole (2.75 mm r0, 2.8 MHz), referred to here as the ion routing multipole (IRM). This multipurpose ion storage device supports optional fragmentation during accumulation and subsequently routes ions to one of the mass analyzers through the use of an axial DC field, which is beneficial for rapidly and efficiently moving stationary ions to the desired destination. For high resolution analysis, ions are transported to a C-trap and then a compact high field Orbitrap analyzer19 operated with −5 kV (positive ions) applied to the central electrode with enhanced Fourier transform20,21 used for processing. For high sensitivity MS/MS analysis, ions are transported to a dual pressure LT,10 scanning at a rate of 66.6 Th/ms. The detector for the LT incorporates two high-energy dynodes that direct secondaries to a single electron multiplier.22 During data dependent acquisition, Orbitrap survey spectra were scheduled for execution at least every 3 s, with the embedded control system determining the number of MS/MS acquisitions executed during this period. AGC targets for Orbitrap survey and LT MS/MS spectra were 200 000 and 10 000 respectively. The Orbitrap target was lower than previously used to account for improved ion transmission with the new instrument configuration. Higher energy collisional dissociation (HCD) was performed in the IRM, using a normalized collision energy of 35%. Monoisotopic precursor selection and dynamic exclusion (60 s duration, 10 ppm mass tolerance, 50 000 list size) were enabled. Bioinformatics. All raw files were searched with Proteome Discoverer 1.4, using Sequest in combination with Percolator to score and rank spectral matches using a 1% false discovery rate.23,24 To evaluate the depth of penetration into the yeast proteome, the results were processed and annotated as previously described.10
Figure 1. Schematic representation of the instrument.
and each analyzer, minimizing ion losses and leaving the rear of the LT accessible for implementation of additional dissociation techniques.25−28 The ion optical configuration and associated control electronics allow for HCD in the IRM and collisionally induced dissociation (CID) or electron transfer dissociation (ETD) in the LT. Each technique can be performed at any stage of an MSn experiment with spectra recorded with either analyzer. This flexibility of ion movement requires that ion transfer efficiency through the instrument, and in particular back and forth between the IRM and high pressure cell of the LT, be near unity or overall instrument sensitivity will suffer proportionally. Due to the numerous optical elements and the interdependency of applied potentials, the traditional sequential optimization process was ineffective. Transfers are instead optimized using a genetic algorithm.29,30 This process is more robust than voltage scans of each element, with less susceptibility to source instability and an ability to optimize multiple dependent elements. The results of the genetic optimization can be characterized by measuring the signal loss after a round trip transfer between any two storage devices. For this instrument, typical efficiencies are on average >90% (Figure S-1, Supporting Information). Orbitrap Analyzer. This work used a compact high-field Orbitrap analyzer19 with an increase of the applied potential from 3.5 to 5.0 kV. This provides an additional 18% in resolving power for any fixed detection period.9 The resulting resolving power at m/z 200 is ∼500 per millisecond of detection or 120 000 FWHM for a typical 256 ms transient and an ultimate resolving power of ∼500 000. Acquisition Rate Improvements. The acquisition rate of a mass spectrometer can be improved by reducing the time associated with any step required to produce a spectrum, such as ion accumulation, isolation, activation, or analysis. Here, the time spent performing waveform based isolation in the LT (>4 ms, m/z and width dependent) is essentially eliminated through the addition of the quadrupole mass filter, since the desired ions are isolated as they transit this device. The only time penalty is the ∼1 ms required to traverse the distance necessitated by the inclusion of the mass filter in the ion path. A secondary benefit of the mass filter is the possibility for space charge free isolation. Waveform based isolation in an ion trap is inherently susceptible to space charge effects due to coaccumulated background ions,31 and the mass filter improves the ability to reliably analyze low abundance species.
■
RESULTS AND DISCUSSION Tribrid Architecture. The mass spectrometer used here comprises three mass separation devices: a quadrupole mass filter, an ultrahigh field Orbitrap analyzer, and dual-pressure LT analyzer (Figure 1). In contrast with prior LT hybrids, where the high resolution analyzer is located to the rear,8 this instrument uses a Tconfiguration. This reduces the distance between the ion source B
dx.doi.org/10.1021/ac403115c | Anal. Chem. XXXX, XXX, XXX−XXX
Analytical Chemistry
Technical Note
Figure 2. Schematic of parallel execution and event pipelining. Top row: Orbitrap scans. Middle and bottom rows demonstrate event pipelining: precursors identified in the previous Orbitrap scan are fragmented and accumulated in IRM (middle row) concurrent with LT acquisition of the previous MS/MS precursor (bottom row). Black rectangles show the injection time for the next Orbitrap scan.
selection of precursors using the improved signal-to-noise ratio, resolution, and mass accuracy of the full time domain signal but also saves ∼200 ms per cycle, since MS/MS acquisition can commence as soon as ions are injected into the FT analyzer instead of waiting to produce a preview spectrum. Instrument Performance with Complex Peptide Mixtures. Using a maximum survey spectral period of three seconds, 14533 MS and 107413 HCD MS/MS spectra were acquired in a single 140 min run (110 min gradient), identifying 29 539 unique peptides (FDR < 1%) and 3880 proteins. The average MS/MS acquisition rate is 14.5 Hz (Figure 3),
The time spent performing resonance excitation based activation (CID) in an ion trap (10−30 ms) is essentially eliminated through the injection of ions into the IRM at elevated energies (HCD), producing spectra comparable to collision cell dissociation even with the ions being trapped instead of passed directly through to the next optical device. The use of the IRM for dissociation brings additional benefits beyond speed, including the reduction of the low mass cutoff for CID32 and reported spectral quality improvements, particularly for phosphopeptides.33,34 A second method to improve the acquisition rate is simultaneous execution of ion processing steps from separate spectra, a technique known as “pipelining” in computing. Namely, ions for spectrum “N” can be scanned from the LT while accumulation of ions occurs in the IRM for spectrum “N +1”.35 In this mode, the instrument does not fit into the traditional classification of “tandem-in-time” or “tandem-inspace”36 but in a more exclusive space of “tandem-in-time-andspace”. For the case where the accumulation time matches the analysis time, pipelined operation can provide a 2-fold improvement in the acquisition rate. In practice, ion trap MS/MS acquisition rates exceed 22 Hz, nearly double that of prior Orbitrap-based hybrids.19 Out of Order Execution. The prior LT based hybrid instruments provided some limited parallelization of the Fourier transform (FT) and LT analyzers.37 Acquisition of a complete high resolution survey spectrum (60k−100k@ 400 m/z) required 750 ms of time domain data. However, using a “preview mode”, after
