Limits for resolving isobaric tandem mass tag reporter ions using

2 days ago - A popular method for peptide quantification relies on isobaric labeling such as tandem mass tags (TMT) which enables multiplexed proteome...
0 downloads 0 Views 1MB Size
Subscriber access provided by UNIV OF WESTERN ONTARIO

Technical Note

Limits for resolving isobaric tandem mass tag reporter ions using phase constrained spectrum deconvolution Christian D Kelstrup, Konstantin Aizikov, Tanveer S Batth, Arne Kreutzman, Dmitry Grinfeld, Oliver Lange, Daniel Mourad, Alexander A. Makarov, and Jesper V. Olsen J. Proteome Res., Just Accepted Manuscript • DOI: 10.1021/acs.jproteome.8b00381 • Publication Date (Web): 16 Sep 2018 Downloaded from http://pubs.acs.org on September 16, 2018

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 29 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of Proteome Research

Limits for resolving isobaric tandem mass tag reporter ions using phase constrained spectrum deconvolution Christian D. Kelstrup1, Konstantin Aizikov2, Tanveer S. Batth1, Arne Kreutzman2, Dmitry Grinfeld2, Oliver Lange2, Daniel Mourad2, Alexander A. Makarov2, Jesper V. Olsen*,1 1)

The Novo Nordisk Foundation Center for Protein Research, University of Copenhagen,

DENMARK 2)

Thermo Fisher Scientific, Bremen, GERMANY

Corresponding Author *Phone: +45-353-25022. E-mail: [email protected].

ABSTRACT A popular method for peptide quantification relies on isobaric labeling such as tandem mass tags (TMT) which enables multiplexed proteome analyses. Quantification is achieved by reporter ions generated by fragmentation in a tandem mass spectrometer. However, with higher degrees of multiplexing, the smaller mass differences between the reporter ions increase the mass resolving power requirements. This contrasts with faster peptide sequencing capabilities enabled by lowered mass resolution on Orbitrap instruments. It is therefore important to determine the mass resolution limits for highly multiplexed quantification when maximizing

ACS Paragon Plus Environment

1

Journal of Proteome Research 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 2 of 29

proteome depth. Here we defined the lower boundaries for resolving TMT reporter ions with 0.0063 Da mass differences using an ultra-high-field Orbitrap mass spectrometer. We found the optimal method depends on the relative ratio between closely spaced reporter ions and that 64 ms transient acquisition time provided sufficient resolving power for separating TMT reporter ions with absolute ratio changes up to 16-fold. Furthermore, a 32 ms transient processed with phase-constrained spectrum deconvolution provides >50% more identifications with >99% quantified, but with a slight loss in quantification precision and accuracy. These findings should guide decisions on what Orbitrap resolution settings to use in future proteomics experiments relying on isobaric TMT reporter ion quantification.

KEYWORDS TMT multiplexing, TMT10-plex, resolution requirement, phase-constrained spectrum deconvolution, ΦSDM, signal processing, super-FT resolution, quantitative proteomics, Orbitrap

INTRODUCTION Quantitative shotgun proteomics deals with large-scale measurements of peptide abundances using liquid chromatography (LC) coupled to mass spectrometry (MS), where quantification is achieved through either stable isotope dilution or label free quantification (LFQ) methods. A prominent choice for isotopic labeling is the isobaric labeling of peptides using TMT reagent sets.1,2 This chemical labeling approach enables multiplexing via a stable heavy isotopically coded tag that consists of an amine-reactive group, a balancer group and a reporter ion group. The differentially TMT labeled peptides are indistinguishable in mass but fragmentation enables relative quantification using the differential reporter ions that are released upon collisionallyinduced dissociation (CID). While the balancer group can also be used for quantification3,4, the

ACS Paragon Plus Environment

2

Page 3 of 29 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of Proteome Research

most commonly used approach is to use the characteristic reporter ions. These were initially limited in multiplexing to four- or six-plex, but with the higher resolution tandem mass spectrometric (MS/MS) instrumentation becoming readily available, the mass differences below integer masses can be resolved and therefore a higher sample multiplexing can be achieved.5,6 However, the required mass resolution is demanding even for modern MS instruments associated with the overhead costs related to the speed of acquisition as resolving power of Fourier transform (FT)-MS scales with acquisition time. An MS instrument is a central part of the general shotgun proteomics workflow and a key element for fast analysis of highly multiplexed TMT labeled samples. One of the most popular mass analyzers for proteomics is the Orbitrap™ mass analyzer, which belongs to the FT-MS family of instruments.7,8 Enhanced FT (eFT™) calculation9,10 has increased the obtained resolution for the same transient signal and although it is superior to conventional (magnitude) FT, the enhanced method has a fundamental limit in achievable resolution imposed by Fourier uncertainty. In recent years novel super-FT resolving methods has demonstrated the ability to resolve isobaric TMT reporter ions on FT-MS transients substantially shorter than otherwise would be required by FT-based signal processing techniques.11–13 Particularly, a new computational approach termed the phase-constrained spectrum deconvolution method (ΦSDM) promises further improvement in spectral quality and MS acquisition rate albeit at an additional computational cost.14 Here, we experimentally investigate the lower limit of the mass resolution required to resolve closely spaced reporter ions with 6.32 mDa mass differences. We further describe an implementation of ΦSDM FT-MS that enables higher resolution for TMT reporter ions with short acquisition transients in real time. The implemented solution operates using the existing hardware in contrast to alternative solutions for higher resolution that require additional hardware such as the FTMS Booster (Spectroswiss, Lausanne, Switzerland). This is

ACS Paragon Plus Environment

3

Journal of Proteome Research 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 4 of 29

accomplished by applying ΦSDM only in a very narrow frequency/mass band containing the reporter ions. This should allow faster acquisition by shortening the Orbitrap transient and thereby lowering the resolution in the MS/MS spectra except for the reporter ion regions which due to the ΦSDM processing will keep a high resolution. Another advantage of the ΦSDM method is that it does not require reporter m/z to be defined a priori unlike alternatives such as the Least Squares Fitting method11 and therefore should be more tolerable to unexpected isobaric interferences. We further investigate when it can be useful to use these short transients processed with ΦSDM for TMT quantitation. We have previously shown that faster acquisition methods require short ion injection times for optimal parallel acquisition mode of operation and therefore require abundant, complex samples on short gradients in order to have a high ion flux that produces MS/MS spectra that contains enough ions to be identified.15,16 Any compromise in ion flux such as extension of the gradient or reduction of sample amount hampers fast acquisition methods as MS/MS spectra may have too low ion counts and thereby poor chances of identification. Data quality under such circumstances can be kept high by allowing longer ion injection times; however, the resolution can then also be increased due to the previously mentioned parallel acquisition. The most relevant area to evaluate resolution limits for resolving TMT reporter ions are therefore with abundant samples, analyzed on short gradients. In any other circumstance, ion flux and not the resolution will most likely be the limiting factor for acquisition speed. The focus is kept on the resolution requirements for TMT10-plex, which are here seen as independent to a broader discussion of general advantages and shortcomings of the TMT labeling strategy. Other limitations have been described elsewhere in great detail and include labeling challenges, accuracy challenges due to co-fragmentation of impure precursors as well as altered gas-phase fragmentation and charge-state distributions.17–20 The specific aim of the

ACS Paragon Plus Environment

4

Page 5 of 29 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of Proteome Research

present work has been to describe the trade-offs with fast acquisition methods for quantitative proteomics with TMT10-plex labeling.

MATERIALS AND METHODS Instrument modification Transients were processed directly on the instrument computer of the commercial Q Exactive™ HF-X mass spectrometer (Thermo Fisher Scientific, Bremen, Germany) with a research-grade data analysis software with in-house implementation of the ΦSDM algorithm. eFT processing was conducted with the default settings at MS/MS transient lengths of 16, 32, 64, and 128 ms. The ΦSDM algorithm recovers a frequency distribution of phased harmonic components approximated on a refined frequency grid from Fourier Transform (FT) spectrum. The method solves a constrained quadratic optimization problem iteratively with respect to unknown mass distribution to fit the observed FT MS signal. The computational procedure is implemented in C++ and employs Alternating Directions Method of Multipliers (ADMM) with the Moreau-Yosida regularization.21 For the complete description of the algorithm the interested reader is referred to the supplementary material section of the original ΦSDM publication.14 To ensure real time computation, ΦSDM was calculated in 0.22 Da wide spectral windows centered at the reporter ion regions of a TMT 10-plex at 126, 127, 128, 129, 130, and 131 Da. The computations in all spectral windows were conducted concurrently in a dedicated thread. The phase was constrained to within a ±5° cone. The number of iterations was set to 50. Frequency grid refinement was set to the factor of 16. To the best knowledge of Thermo Fisher Scientific authors, ΦSDM algorithm is planned to be made available commercially once it passes through all stages of the internal development

ACS Paragon Plus Environment

5

Journal of Proteome Research 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 6 of 29

process. Until then, Thermo Fisher Scientific authors are ready to make it available to a limited number of interested parties under a customer-privileged license agreement.

Cell culture Immortalized human epithelial cervix carcinoma adherent cells (HeLa), human U-2-OS osteosarcoma cells (U2OS), and human retinal pigment epithelial cells (RPE1) were grown in 15 centimeter dishes in DMEM (Gibco, Waltham, USA) media containing 2 mM L-glutamine supplemented with 10% fetal bovine serum (Gibco, Waltham, USA), 100 U/ml penicillin (Life Technologies, Carlsbad, USA), 100 µg/ml streptomycin in 37°C incubator supplemented with 5% CO2. Human neuroblastoma SH-SY5Y and human Jurkat T-lymphocytes cell lines were cultured in RPMI (Gibco, Invitrogen) with the same supplements as listed above. Growing cells were allowed to obtain between 80-90% confluency prior to harvesting.

Sample preparation and trypsin digestion Media was removed from the plates and the cells washed twice with cold phosphate containing buffer (1x PBS). Cells were rapidly lysed and cysteines were reduced and alkylated in a single step as previously described.22 Briefly, 4ml of boiling 6M guanidine hydrochloride (Gnd-HCl) containing 10 mM chloroacetamide and 10 mM tris(2-carboxyethyl)phosphine was added directly to the plates and the cells were manually collected by scraping. Lysis buffer containing cells were boiled for an additional 10 minutes at 99°C followed by sonication (Vibra-Cell VCX130, Sonics, Newtown, CT, USA) for 2 minutes with pulses of 1 s on and 1 s off at 50% amplitude. Protein concentration was determined by Bradford assay (Bio-Rad, Hercules, USA). Lys-C protease (Wako Chemicals, Richmond, USA) was added at a ratio of 1:100 (w/w) and digested for 2 hours at 37 °C. The concentration of Gnd-HCl was reduced to less than 1M with 25mM ammonium bicarbonate prior to the addition of trypsin (Life Technologies, Carlsbad,

ACS Paragon Plus Environment

6

Page 7 of 29 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of Proteome Research

USA) at a ratio of 1:50 and allowed to digest overnight at 37 °C. After digestion, the sample was acidified using trifluoroacetic acid (TFA) to a 1% final concentration in order to quench protease activity followed by centrifugation for 5 minutes at 4000 g to remove residual cell debris. Supernatant was cleared of all salts/buffers and tryptic peptides were purified using solid phase extraction on C18 Sep-Pak (Waters Corporation, Milford, USA) by gravity. Peptides were eluted with 50% acetonitrile (ACN), 0.1% formic acid (FA). Peptides were dried and ACN was removed from the peptide containing sample by SpeedVac (Eppendorf, Germany) at 45°C. Peptide concentration was measured using a NanoDrop™ 2000 spectrometer (Thermo Fisher Scientific, Waltham, USA).

TMT labeling Peptides were labeled using TMT-10plex isobaric tags (Thermo Fisher Scientific, San Jose, USA) according to manufacturer's instructions. Briefly, peptides were dissolved in 40% ACN containing 40 mM HEPES buffer at pH 8.5. TMT reagent dissolved in neat ACN was added to peptide samples and the reaction was carried out at room temperature for 1 hour. Reaction was quenched with the addition of 1% hydroxylamine for 15 minutes. TMT-plex 1-10 labeled samples were mixed into one and acidified to 1% TFA. ACN of labeled mixed samples was evaporated by speedvac and additional buffers and excess TMT reagent was removed using C18 solid phase extraction as described above. Peptide concentration of labeled samples was again measured using the NanoDrop 2000 spectrometer. Peptides from HeLa cells were mixed to form the five test samples with known mixing ratios. Another four cell line sample was created where peptides from four cell lines was mixed in equal amounts (w/w). In detail, U2OS was labeled with TMT127N and TMT129N, Jurkat with TMT127C and TMT129C, RPE1 with TMT128N and TMT130N, SH-SY5Y with TMT128C and 130C.

ACS Paragon Plus Environment

7

Journal of Proteome Research 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 8 of 29

High-pH Fractionation Labeled peptides were fractionated as previously described15,23 using a reversed-phase Acquity CSH C18 1.7 µm 1 × 150 mm column (Waters, Milford, MA) on an UltiMate 3000 high-pressure liquid chromatography (HPLC) system (Dionex, Sunnyvale, CA) operating at 30 µL/min. Buffer A (5 mM ammonium bicarbonate) and buffer B (100% ACN) were used. Peptides were separated by a linear gradient from 5% B to 35% B in 55 min, followed by a linear increase to 70% B in 8 min. 46 fractions were collected without concatenation directly into a 96-microwell plate. The loading amount was kept constant at 250 ng per injection, estimated by measuring absorbance at 280 nm on a NanoDrop 2000 spectrometer.

Nanoflow LC/MS/MS The nano LC/MS/MS analysis was performed as previously explained with few alterations.15 Briefly, the peptide solution was adjusted in volume to 0.1 µg/µl and kept in loading buffer (5% ACN and 0.1% TFA) prior to autosampling. The EASY-nLC™ 1200 system (Thermo Fisher Scientific, San Jose, USA) was equipped with an integrated column oven (PRSO-V1, Sonation GmbH, Biberach, Germany) maintaining temperature at 40 °C for the in-house packed 15 cm, 75 µm ID analytical capillary column with 1.9 µm Reprosil-Pur C18 beads (Dr. Maisch, Ammerbuch, Germany) which was interfaced online with the mass spectrometer. FA 0.1% was used to buffer the pH in the two running buffers. The 15 minute gradient went from 8% to 24% ACN in 12.5 minutes followed by 24% to 36% in 2.5 minutes. Longer gradients were multiple of this, i.e. the 30 minutes gradient went from 8% to 24% ACN in 25 minutes followed by 24% to 36% in 5 minutes. All gradients were followed by a washout by a 0.5 minute increase to 64% ACN which was kept for 4.5 minutes. Flow rate was kept at 350 nL/minute. Re-equilibration was done in parallel with 2 µl sample pickup and prior to loading with a minimum requirement of 0.5 µL 0.1% FA buffer at a pressure of 800 bar.

ACS Paragon Plus Environment

8

Page 9 of 29 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of Proteome Research

The Q Exactive HF-X instrument was configured as described above with custom Tune (version 2.9) instrument control software that enabled a Tune toggle for ΦSDM for TMT10-plex. Spray voltage was set to 2 kV in positive polarity, funnel RF level at 40, and heated ion transfer tube temperature at 275 °C. The methods employed were Full MS/ DD-MS/MS. Full MS resolutions were set to 60,000 at m/z 200 and full MS target was 3E6 with an IT of 45 ms. Mass range was set to 350-1400. Target value for fragment spectra was set at 1E5, and intensity threshold was kept at 1E5. Isolation width was set at 0.8 m/z. Dynamic exclusion was enabled and set at 30 s. A fixed first mass of 100 m/z was used. Normalized collision energy was set at 33%. Peptide match was set to off, and isotope exclusion was on. Charge state exclusion rejected ions having unassigned charge states or that were singly charged or had a charge state above 5. Full MS data were acquired in the profile mode with fragment spectra recorded in the centroid mode. For each of the MS/MS transient lengths, specific settings were set to keep parallel acquisition optimal and MS cycle time approximately constant. The MS/MS acquisition at 7500 resolution or 16 ms transient was combined with a loop count of 40 and an injection time of 11 ms. The 15,000 resolution or 32 ms transient MS/MS acquisition was combined with a loop count of 28 and an injection time of 22 ms. The 30,000 resolution or 64 ms transient had a loop count at 14 and an injection time of 54 ms. The 60,000 or 128 ms transient had a loop count of 7 and an injection time of 118 ms. A separate control experiment was performed on the five main samples just after the primary experiment where all instrument methods were also measured with a constant 86 ms injection time and a loop count of 10.

Data analysis All raw LC-MS/MS files were analyzed in one combined analysis using the MaxQuant software suite (version 1.6.1.2) with the integrated Andromeda search engine, where each raw file was configured as a separate experiment.24 For the database search, fixed modifications were set to

ACS Paragon Plus Environment

9

Journal of Proteome Research 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 10 of 29

carbamylation of cysteines and TMT-10 were specified as label on N-terminal and lysines residues with mass tolerance set to 0.003 Da. Variable modifications were set to oxidation of methionines, protein N-terminal acetylation, deamidation of asparagine or glutamine, and pyroglutamate formation from N-terminal glutamine. The database was the UniProt human reference proteome release 2018_02 without isoforms containing 21,010 protein sequences. In addition, the included contaminant database from MaxQuant contains enzymes and common contaminants. FDR was set to 1% on the PSM, site, and protein level. Otherwise default values were used. Reading of intensity-to-noise values was performed with the raxport command line tool (freely available at http://code.google.com/p/raxport/) together with custom scripts in Perl to filter the large lists. All data were filtered to only focus on identified spectra. Subsequent data analysis was performed in R.25 The TMT labeled HeLa sample was analyzed without normalization and a N:C ratio was calculated for each closely spaced reporter ion pair. The results from the four cell line sample was log2-transformed and normalized first by channel wise median subtraction to correct for possible mixing errors followed by row wise normalization the log-transformed reporter intensities. Significance was evaluated using Wilcoxon rank sum and signed rank tests. Principal component analysis was performed without further scaling. Heatmap including clustering was performed using the heatmap3 package without further scaling.26 Design of figures was achieved in Adobe Illustrator CS6. An overview of the acquired data can be found in the supplementary information (Table S-1). The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium27 via the PRIDE partner repository with the data set identifier PXD011009. Reviewer account details: Username: [email protected] Password: oDtLXqXY

ACS Paragon Plus Environment

10

Page 11 of 29 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of Proteome Research

RESULTS AND DISCUSSION TMT10-plex quantification relies on mass resolution and accurate measurements of the reporter ions with mass differences as small as 3.6 mDa, which places a high demand on the mass resolving power of the MS/MS measurements. It is known, however, that resolution and speed are a trade-off for FT-MS instruments and increases in acquisition speed has been shown to improve throughput and increase proteome coverage depth.15,23 It is therefore important to uncover the lower limits of the resolution requirements for resolving isobaric TMT reporter ions with identical integer masses, and thereby establish resolution settings that are sufficient to achieve accurate quantification, as well as settings at which no practical benefits with higher resolution is observed. Further, TMT quantification is expected to benefit from new signal processing methods targeted to improve the obtained resolution.

Traditionally, separating closely spaced MS peaks required them to be baseline resolved or have a low degree of overlap known as the ‘10% valley resolution definition’.28 This criterion

ACS Paragon Plus Environment

11

Journal of Proteome Research 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 12 of 29

mainly deals with peaks of similar intensity and largely ignores that it is more difficult to separate peaks that have intensity-ratios of 1:4 or higher. This is easily demonstrated through modeling where resolution needs to be higher in order to separate peaks of different intensities (Figure 1a). Due to this effect, an insufficient resolving power results in closely spaced peaks appearing either merged or having their intensities corrupted by the FT interferences.29 For high abundance ratios, the smaller peak might simply disappear in the noise band and be reported as a missing data-point. Missing data-points are usually not a problem in the TMT-based quantification when sufficient MS/MS resolution settings are used. However, for the low resolution settings, missing data-points start to appear, which can be used as a simple measure for how well the quantitative method separates closely spaced peaks. Other possible indicators can be the mass accuracy, which might be affected if the peaks merge, or quantitative descriptive parameters such as accuracy and precision. These considerations form the basis for the current report. Using the ΦSDM as a super FT resolving approach has recently been shown to increase the mass resolving power of the Orbitrap mass analyzer relative to the eFT method currently implemented on the commercial instrument (Figure 1b).14 The ΦSDM algorithm hence supplements the eFT in the data processing of the transient signal to the mass spectrum. Implementation of the ΦSDM algorithm has here been optimized to run on existing instrument hardware with no downsides in acquisition speed (Supplementary Figure S-1a). This was accomplished by limiting the ΦSDM calculations to 6 small m/z ranges, in which the TMT10-plex reporter ions are present (Figure 1c). For the rest of the spectrum, the eFT algorithm was used to enable real-time processing with the current instrument hardware. In greater detail, each 0.22 m/z window was chosen such that the corresponding number of associated frequency bins does not fall below a tested minimal threshold width which ensures reliability of the results. This minimal width is a function of the chosen mass range itself (as the ion oscillation frequency is a

ACS Paragon Plus Environment

12

Page 13 of 29 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of Proteome Research

function of mass) and the resolution settings (as longer transients equal finer frequency bins for a fixed sampling rate). A 5 degree phase angle was chosen such that any (unavoidable) electronic timing jitter, which will be reflected into a small phase distribution broadening around the calibrated phase is safely comprised by this phase slack. The number of iterations and grid refinement settings were set to 50 and 16, respectively, as these settings were found to deliver satisfactory convergence and resolution for experimental as well as artificial test input. Each window had the same processing settings as it is desirable to ensure mutual comparability of the results from the chosen windows (as long as the windows and settings are set reasonably as described above). More windows could theoretically be added if more TMT channels become available, where the settings described above will also be appropriate. The benefit of these additional ΦSDM calculations provides a resolution improvement in this narrow mass range that can be seen as substantial (Figure 1d).

ACS Paragon Plus Environment

13

Journal of Proteome Research 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 14 of 29

To design an empirical test to define the lowest resolution needed to resolve TMT reporter ions with 6.32 mDa mass differences, peptides from a HeLa digest were labeled with the eight central channels from a TMT10-plex kit (Figure 2a). These contain four pairs of closely spaced reporter ions, which were mixed in different known ratios. In total, five sample mixtures with known TMT ratios were designed (sample I-V), where four (sample I-IV) contained the four closely spaced peak pairs in ratios from 8:1 to 1:8 (Figure 2a). Sample V contained the more extreme ratios 1:16 and 1:64. All five samples were subsequently measured using data dependent acquisition on short 15 minute gradients with seven different main MS methods.The MS methods used the three shortest transient lengths available i.e. 16, 32, and 64 ms with either eFT or ΦSDM processing set for the TMT reporter region. A further control MS method

ACS Paragon Plus Environment

14

Page 15 of 29 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of Proteome Research

was included in the form of a 128 ms transient with eFT processing, which is known to have high enough resolution to baseline resolve isobaric TMT reporter ions (Figure 2a). As all peptides in the sample were labeled in the same way, each spectrum within a raw file should show the same reporter ion distribution and could therefore be seen as a technical replicate measurement with variable precursors and ion counts. Further, the five different samples were all expected to generate similar conclusions and can therefore be seen as replica on a different layer. Finally, a replica control experiment was run using constant instead of variable ion injection times. However, the main comparative analysis was applied to the seven different MS methods, which together with five samples and two replicates resulted in seventy raw files in this part of the experiment. Samples I to IV were found to behave very similarly with regard to the overall number of identifications and quantifications and they were therefore grouped for this analysis (Figure 2b). The number of peptide spectrum matches (PSMs) increased with shorter transients; this is a known potential benefit from faster Orbitrap transients when the ion flux is high enough.15 A comparison of eFT to ΦSDM showed no significant differences in total number of identifications which confirmed that the computational costs of ΦSDM are kept within the limitations of the current MS hardware. The challenge with fast acquisition methods is readily apparent when quantified PSMs are investigated (Figure 2b). For eFT-based quantification, the 16 ms and 32 ms transients did not produce any PSMs with all TMT reporter ions present (fully quantified) indicating problems in resolving the closely spaced peaks. In contrast, the 64 ms eFT method showed great performance for samples I-IV, with >99% of ~4800 PSMs being fully quantified. The 128 ms eFT method, however, lead to 45% less PSMs (~2600) with similar quantification coverage (>99%) of all PSMs. For sample V, the challenge of quantifying high ratios came into effect and only 35% of PSMs could be fully quantified using the 64 ms eFT method while >99% quantified

ACS Paragon Plus Environment

15

Journal of Proteome Research 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 16 of 29

PSMs was maintained for the 128 ms eFT method that also gave the total highest number of fully quantified PSMs. We found a small software issue that labeled reporter ions as exception peaks in the raw data for the 8:1 ratio with the 128 ms eFT transient. This should be fixed in a software update in the future, and we therefore decided to skip this data-point from this part of the analysis. Comparing eFT to ΦSDM showed no significant difference in the total number of identifications, which confirms that the relative computational costs of ΦSDM are practically non-existent as it operates within the limitations of the current MS hardware. Overall, quantitative performance was found to be improved. The 16 ms ΦSDM methods could fully quantify ~5000 PSMs from sample I-IV with the 32 ms increasing this to ~7460 (Figure 2b). With eFT, no reporter ion sets were fully quantified with these very short transient lengths. However, the faster 32 ms ΦSDM method reported here quantified 55% more than the 64 ms method with eFT. For the extreme ratio sample V, ΦSDM was able to show an improvement as the 64 ms ΦSDM method now kept >99% PSMs fully quantified but with 80% higher total PSMs than the classical 128 ms eFT method. Improvements for ΦSDM were also seen for mass errors for the 16 and 32 ms transients while no significant difference was seen for the 64 ms transient (Supplemental Figure S-1b and S-1c). These results showed that the 16 ms eFT method failed to resolve closely spaced reporter ions showing only one peak instead of two. While the mass precision was good as the closely spaced reporter ion pairs were consistently not resolved it showed the relative worst mass accuracy for the lowest ratios which fits with the interpretation that peaks of similar height collapsed to a single peak with an average mass to charge value. In contrast, the 16 ms ΦSDM was found to resolve the closely spaced peaks but still suffers from relatively poor precision with relatively high m/z spread. The 32 ms eFT method was more complicated to interpret with variable precision and accuracy likely due to only partially resolved closely spaced reporter ions

ACS Paragon Plus Environment

16

Page 17 of 29 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of Proteome Research

of especially the low ratios. The 32 ms ΦSDM method and methods with longer transients with either ΦSDM or eFT processing showed good and similar mass accuracy and precision indicating that these methods could all resolve the majority of the closely spaced reporter ions.

In further detail, the relative number of quantified ratios was verified to depend on the ratio between the heavy N and C version of the closely spaced TMT reporter ions (Figure 3a). Here it was also evident that 16 ms eFT method was not enough to resolve any reporter ion pairs, whereas the 32 ms eFT method had an approximately 70% chance of producing a ratio readout for ratios with changes up to two-fold. This was evaluated as too poor performance for practical use and therefore these short transients with eFT quantification were excluded from further analysis. In contrast, the longer 64 ms and 128 ms eFT methods generally performed

ACS Paragon Plus Environment

17

Journal of Proteome Research 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 18 of 29

comparably in terms of relative number of quantified PSMs with the only difference that the 64 ms transient displaying difficulties with the most extreme 1:64 ratio, where only 50% of PSMs were quantified. With ΦSDM replacing eFT for the TMT mass region, the fastest 16 ms transient was found to separate ratios up to 4 fairly consistently while higher ratios have >5% chance of producing missing values. Increasing the transient to 32 ms was only found to cause problems for the most extreme 1:64 ratio. It can be speculated that a currently unavailable 48 ms transient processed with ΦSDM would provide a nice compromise if such large ratios would need to be resolved. Next, quantitative differences in precision and accuracy were evaluated (example for sample III in Figure 3b with all results in Supplementary Figure S-2a). The most accurate and precise methods were unsurprisingly the longer transients while higher ratios showed worse precision. Between the 64 ms and 128 ms eFT methods there was no significant difference in accuracy measured by median (p = 0.29; n = 20). Precision in the form of the geometric coefficient of variation (CV) was significantly different when ratios were considered (paired analysis, p = 0.002; n = 20) but only slightly so as both methods were within 1% absolute difference with the average CV of the 128 ms eFT method at 15.4% and the 64ms eFT method at 14.5% CV (Figure 3c and Supplementary Figure S-2b). This suggests the 64 ms eFT as an attractive method for fast acquisition as it seemed to give good quantitative results at the lower limit of the resolution requirement. Comparing ΦSDM and eFT for the 64 ms transient showed a small significant difference with ΦSDM processing giving on average a 2% more extreme median (p < 0.001; n = 20) and a worse CV averaging at 18.2% (p < 0.001; n = 20). This indicates a quantitative disadvantage by enabling ΦSDM when resolution is sufficient, which made ΦSDM less interesting for this and longer transient lengths. For the shorter 32 ms and 16 ms, where ΦSDM is required for resolving the reporter ions, accuracy and precision was worse than longer transients. This was

ACS Paragon Plus Environment

18

Page 19 of 29 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of Proteome Research

especially evident for the higher ratios, e.g. for the 32 ms ΦSDM method 1:1 ratio had a modest absolute 1.4% larger CV than the 128 ms eFT method while the absolute CV for the 1:2, 1:4, and 1:8 ratios were 1.8%, 5.4%, and 17.8% larger, respectively. As faster acquisition methods have shorter ion injection time and hence lower ion counts in each MS/MS spectrum this could affect the quantitative readouts. Analyzing the ratio readout as a function of intensity-to-noise was therefore found relevant. Across all samples a lower intensity-to-noise was found to worsen precision and an overall lower intensity-to-noise was found for the faster acquisition methods (Figure 3d and Supplementary Figure S-3). The fast 32 and 16 ms ΦSDM methods provided lower precision and accuracy than slower acquisition methods and this was found to be partly due to processing method and partly due to overall lower intensity-to-noise. Yet, although the 16 ms ΦSDM method did work and could resolve most peaks, actual use-cases seem almost nonexistent as the 32 ms ΦSDM method gave better quantitative data with a similar number of identifications. Hypothetically, a higher average ion flux than the ones tested here could potentially saturate the 32 ms ΦSDM method and tip the balance in favor of the 16 ms ΦSDM method. Although, this is an unlikely scenario short term, it can be speculated that future instrument advancements probably will increase ion flux or provide more advanced ion accumulation that could allow for this method to perform better.

ACS Paragon Plus Environment

19

Journal of Proteome Research 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 20 of 29

The 32 ms ΦSDM method or the 64 ms eFT method can arguably be seen as the two most interesting methods, both close to the limit for quantifying TMT reporter ions. Based on the results presented so far, it can be expected that the faster 32 ΦSDM method would trade a benefit in increased qualitative performance compared to the 64 ms eFT which would have an improved quantitative performance. It can also be hypothesized that under changed experimental conditions where the average ion flux is lowered either through dilution of sample amounts, reduction of sample complexity, or with the use of longer gradients, any benefit in increased proteome depth from the faster acquisition will likely be lost as the number of ions in each spectrum becomes too low. In order to test this, a new sample was created where peptides from four cell lines were mixed in duplicates in equal amounts as described previously and a series of experiments were conducted with this sample. The first was a dilution series with the two most interesting acquisition methods used as variables and measured in technical duplicates. The faster 32 ms

ACS Paragon Plus Environment

20

Page 21 of 29 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of Proteome Research

ΦSDM method gave consistently 180% MS/MS spectra across all dilutions compared to the 64 ms eFT method (Figure 4a). However, the qualitative benefit decreased from 162% identifications at the highest 500 ng peptide load on column to 121% at 10 ng peptide load. When the gradient was extended, the faster 32 ms ΦSDM method consistently gave above 180% MS/MS spectra compared to the 64 ms eFT method on the same gradient length or 95% of what could be achieved on a twice as long gradient (Figure 4b). Again, translating this to qualitative performance showed a decreasing benefit with longer gradients i.e. at short 15 minute gradients the identifications per minute for the 32 ms ΦSDM method was at 164% compared to the 64 ms eFT method while this benefit on longer gradients dropped to 113% (Figure 4c). Finally, we evaluated the impact of offline fractionation of the four cell line sample using classical two dimensional high/low pH reversed-phase fractionation with short LC gradients.15,23 In the experimental setup, each of 46 fractions was analyzed by repeated injection and measurement with either the 64 ms eFT or the 32 ms ΦSDM method acquiring technical duplicates of each in alternating run order. The number of identified spectra, unique peptides, and proteins were found highest for the 32 ms ΦSDM method as shown (Figure 4d). Further the relative number of PSMs that had all used reporter ion channels quantified were 94.8% for both 64 ms eFT replica and 96.1% in both replica for the 32 ms ΦSDM method. Comparing the the total numbers between the and the 64 ms eFT method, the number of quantified PSMs was 30.2% higher, the number of unique peptides were 22.2% and the number of protein groups were 7.6% higher (Figure 4d). Looking at quantitative differences for all unique peptides that were found across all replicates using a principal component analysis, the acquisition method seemed not to influence overall grouping indicating very similar quantitative performance. A hierarchical cluster analysis using correlations showed a small systematic difference for each acquisition method but with overall similar performance between the acquisition methods. In

ACS Paragon Plus Environment

21

Journal of Proteome Research 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 22 of 29

summary, the fractionation experiments showed that it was possible to get deeper proteome coverage than using 64 ms eFT with very similar quantitative performance using the 32 ms ΦSDM method.

CONCLUSION This systematic comparative study shows that two methods in particular seems relevant for future TMT10-plex or even higher multiplexing experiments. The 32 ms transient with ΦSDM provided the overall best performance by achieving the highest number of quantified PSMs and lowest costs in precision and accuracy for absolute intensity ratio changes up to 16-fold. The slower, 64 ms transient with eFT had slightly better precision and accuracy for larger ratios with the same absolute ratio limitation of 16 but with overall improved ion statistics due to longer fill times. This limitation is practically small as absolute ratios above 16 between reporter ions in the same integer mass channel are rarely reported in the literature. However, if larger ratios are expected, longer transients can be used. These findings should be helpful for making decisions on what transient to use for proteomics investigations relying on reporter ion quantification with high resolution requirements. The 32 ms transient with ΦSDM accelerates experiments utilizing isobaric reporter ions of TMT under high ion flux conditions with only small quantitative tradeoffs.

FIGURES Figure 1: Resolving closely spaced peaks. a) Simulations of two closely spaced Gaussian peaks with different volumes and widths representing different resolutions and intensities. Centroids after basic peak detection is shown in red and the ability of calling peaks resolved is subjectively evaluated with a change in background color where red indicate unlikely to be

ACS Paragon Plus Environment

22

Page 23 of 29 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of Proteome Research

resolved, yellow is uncertain and green is likely resolved. b) Schematic depiction of the signal processing in Orbitrap mass spectrometry starting from a transient, calculation of the frequency spectrum, followed by conversion to a mass spectrum. c) The employed strategy for limiting ΦSDM to only work on a very small part of the mass spectrum. d) The relationship between resolution settings in the software, the corresponding length of the recorded transient for the ultra-high-field Orbitrap and the resolution in the reporter ion region with either eFT or ΦSDM. Figure 2: Experimental design and overview results. a) HeLa samples labeled with the eight central TMT reporter ions, representing four integer mass pairs. These are mixed in five different set of ratios forming five different samples that are measured with seven different MS methods. b) Bar chart showing number of identified and quantified spectra for the measured samples. Error bars indicate the standard deviation across different samples. Figure 3: More detailed results. a) Line chart summary where the ratio of closely spaced reporter ions (N:C) ratio is divided by of the relative number of quantified ratios vs the total number of identifications. b) Box plot for sample III of five working acquisition methods and four ratios. Horizontal width indicates how many spectra were quantified in each condition and horizontal lines indicate expected ratios. c) Bar chart showing the geometric coefficient of variation for each ratio grouped by acquisition method and TMT channel. d) Scatter plot showing the measured ratios versus the minimum intensity to noise for sample III for all working acquisition methods. A 300 point moving average line is shown for each group together with a lighter area covering the moving average +/- the moving standard deviation to make trends more visible. Figure 4: Comparison of the 64 ms eFT and the 32 ms ΦSDM acquisition method. a) Bar chart showing the average result of a dilution series with 15 minute gradient analysis. b) Bar chart showing the average result of longer gradients. c) Bar chart showing the speed of

ACS Paragon Plus Environment

23

Journal of Proteome Research 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 24 of 29

identified spectra vs the gradient length. d) Bar chart showing the average number of identified and fully quantified spectra, unique peptide sequences, and protein groups found with the two methods. e) Scatter plot showing the scores for the first two components from a principal component analysis of peptides found quantified in all conditions. The largest grouping effect is annotation with the cell type and the second largest with the difference between replica in different TMT channels (“Ch. replica”). f) Heat map displaying protein intensities after median normalization for each peptide across all 32 conditions with white indicating a missing value. The clustering is depicted as a dendrogram with annotated grouping based on channel replica, injection technical replica, acquisition method, and cell type.

ASSOCIATED CONTENT Supporting Information The following files are available free of charge at ACS website http://pubs.acs.org: Figure S-1: Characteristics of transient lengths and FT processing method. Figure S-2: Additional data on precision and accuracy. Figure S-3: Additional data on intensity dependency. Table S-1: Overview of acquired raw data

AUTHOR INFORMATION Notes The authors declare the following competing financial interest(s): K.A., A.K., D.G., O.L, D.M., and A.M. are employees of Thermo Fisher Scientific, the manufacturer of the Q Exactive HF-X instrument used in this research.

ACS Paragon Plus Environment

24

Page 25 of 29 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of Proteome Research

ACKNOWLEDGMENT Work at The Novo Nordisk Foundation Center for Protein Research (CPR) is funded in part by a generous donation from the Novo Nordisk Foundation (grant no. NNF14CC0001). Part of this work has been funded as part of the MSmed project that has received funding from the European Union’s Horizon 2020 Research and innovation program under grant agreement no. 686547. We would like to thank the PRO-MS Danish National Mass Spectrometry Platform for Functional Proteomics and the CPR Mass Spectrometry Platform for instrument support and assistance. We thank Dorte Breinholdt Bekker-Jensen for assistance in cell culture and peptide digestion. J.V.O. was supported by the Danish Cancer Society (R90-A5844 KBVU project grant).

ABBREVIATIONS ACN, acetonitrile; CID, collisionally-induced dissociation; CV, coefficient of variation; eFT, enhanced Fourier transform; FA, formic acid; FT, Fourier transform; Gnd-HCl, guanidine hydrochloride; LC, liquid chromatography; LFQ, label free quantification; m/z, mass-to-charge; MS, mass spectrometry; MS/MS, tandem mass spectrometry; PSMs, peptide spectrum matches; TFA, trifluoroacetic acid; TMT, tandem mass tag; ΦSDM, phase-constrained spectrum deconvolution method

REFERENCES (1) Thompson, A.; Schäfer, J.; Kuhn, K.; Kienle, S.; Schwarz, J.; Schmidt, G.; Neumann, T.; Johnstone, R.; Mohammed, A. K. A.; Hamon, C. Tandem mass tags: a novel quantification strategy for comparative analysis of complex protein mixtures by MS/MS. Anal. Chem. 2003, 75 (8), 1895–1904.

ACS Paragon Plus Environment

25

Journal of Proteome Research 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 26 of 29

(2) Ross, P. L.; Huang, Y. N.; Marchese, J. N.; Williamson, B.; Parker, K.; Hattan, S.; Khainovski, N.; Pillai, S.; Dey, S.; Daniels, S.; et al. Multiplexed protein quantitation in Saccharomyces cerevisiae using amine-reactive isobaric tagging reagents. Mol. Cell. Proteomics 2004, 3 (12), 1154–1169. (3) Wühr, M.; Haas, W.; McAlister, G. C.; Peshkin, L.; Rad, R.; Kirschner, M. W.; Gygi, S. P. Accurate multiplexed proteomics at the MS2 level using the complement reporter ion cluster. Anal. Chem. 2012, 84 (21), 9214–9221. (4) Stadlmeier, M.; Bogena, J.; Wallner, M.; Wühr, M.; Carell, T. A Sulfoxide-Based Isobaric Labelling Reagent for Accurate Quantitative Mass Spectrometry. Angew. Chem. Int. Ed Engl. 2018, 57 (11), 2958–2962. (5) McAlister, G. C.; Huttlin, E. L.; Haas, W.; Ting, L.; Jedrychowski, M. P.; Rogers, J. C.; Kuhn, K.; Pike, I.; Grothe, R. A.; Blethrow, J. D.; et al. Increasing the multiplexing capacity of TMTs using reporter ion isotopologues with isobaric masses. Anal. Chem. 2012, 84 (17), 7469–7478. (6) Werner, T.; Becher, I.; Sweetman, G.; Doce, C.; Savitski, M. M.; Bantscheff, M. Highresolution enabled TMT 8-plexing. Anal. Chem. 2012, 84 (16), 7188–7194. (7) Hu, Q.; Noll, R. J.; Li, H.; Makarov, A.; Hardman, M.; Graham Cooks, R. The Orbitrap: a new mass spectrometer. J. Mass Spectrom. 2005, 40 (4), 430–443. (8) Zubarev, R. A.; Makarov, A. Orbitrap Mass Spectrometry. Anal. Chem. 2013, 85 (11), 5288–5296. (9) Michalski, A.; Damoc, E.; Hauschild, J.-P.; Lange, O.; Wieghaus, A.; Makarov, A.; Nagaraj, N.; Cox, J.; Mann, M.; Horning, S. Mass spectrometry-based proteomics using Q Exactive, a high-performance benchtop quadrupole Orbitrap mass spectrometer. Mol. Cell. Proteomics 2011, 10 (9), M111.011015. (10)

Lange, O.; Damoc, E.; Wieghaus, A.; Makarov, A. Enhanced Fourier transform for Orbitrap mass spectrometry. Int. J. Mass Spectrom. 2014, 369, 16–22.

ACS Paragon Plus Environment

26

Page 27 of 29 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of Proteome Research

(11)

Aushev, T.; Kozhinov, A. N.; Tsybin, Y. O. Least-squares fitting of time-domain signals for Fourier transform mass spectrometry. J. Am. Soc. Mass Spectrom. 2014, 25 (7), 1263– 1273.

(12)

Kozhinov, A. Fourier transform mass spectrometry at the uncertainty principle limit for improved qualitative and quantitative molecular analyses, Thesis, EPFL, Lausanne, 2015.

(13)

Kozhinov, A. N.; Tsybin, Y. O. Filter diagonalization method-based mass spectrometry for molecular and macromolecular structure analysis. Anal. Chem. 2012, 84 (6), 2850– 2856.

(14)

Grinfeld, D.; Aizikov, K.; Kreutzmann, A.; Damoc, E.; Makarov, A. Phase-Constrained Spectrum Deconvolution for Fourier Transform Mass Spectrometry. Anal. Chem. 2017, 89 (2), 1202–1211.

(15)

Kelstrup, C. D.; Bekker-Jensen, D. B.; Arrey, T. N.; Hogrebe, A.; Harder, A.; Olsen, J. V. Performance Evaluation of the Q Exactive HF-X for Shotgun Proteomics. J. Proteome Res. 2018, 17 (1), 727–738.

(16)

Kelstrup, C. D.; Jersie-Christensen, R. R.; Batth, T. S.; Arrey, T. N.; Kuehn, A.; Kellmann, M.; Olsen, J. V. Rapid and Deep Proteomes by Faster Sequencing on a Benchtop Quadrupole Ultra-High-Field Orbitrap Mass Spectrometer. J. Proteome Res. 2014, 13 (12), 6187–6195.

(17)

Everley, R. A.; Huttlin, E. L.; Erickson, A. R.; Beausoleil, S. A.; Gygi, S. P. Neutral Loss Is a Very Common Occurrence in Phosphotyrosine-Containing Peptides Labeled with Isobaric Tags. J. Proteome Res. 2017, 16 (2), 1069–1076.

(18)

Sandberg, A.; Branca, R. M. M.; Lehtiö, J.; Forshed, J. Quantitative accuracy in mass spectrometry based proteomics of complex samples: the impact of labeling and precursor interference. J. Proteomics 2014, 96, 133–144.

(19)

Thingholm, T. E.; Palmisano, G.; Kjeldsen, F.; Larsen, M. R. Undesirable ChargeEnhancement of Isobaric Tagged Phosphopeptides Leads to Reduced Identification

ACS Paragon Plus Environment

27

Journal of Proteome Research 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 28 of 29

Efficiency. J. Proteome Res. 2010, 9 (8), 4045–4052. (20)

Ow, S. Y.; Salim, M.; Noirel, J.; Evans, C.; Rehman, I.; Wright, P. C. iTRAQ underestimation in simple and complex mixtures: “the good, the bad and the ugly.” J. Proteome Res. 2009, 8 (11), 5347–5355.

(21)

Boyd, S.; Parikh, N.; Chu, E.; Peleato, B.; Eckstein, J. Distributed Optimization and Statistical Learning via the Alternating Direction Method of Multipliers. Foundations and Trends® in Machine Learning 2011, 3 (1), 1–122.

(22)

Jersie-Christensen, R. R.; Sultan, A.; Olsen, J. V. Simple and Reproducible Sample Preparation for Single-Shot Phosphoproteomics with High Sensitivity. Methods Mol. Biol. 2016, 1355, 251–260.

(23)

Bekker-Jensen, D. B.; Kelstrup, C. D.; Batth, T. S.; Larsen, S. C.; Haldrup, C.; Bramsen, J. B.; Sørensen, K. D.; Høyer, S.; Ørntoft, T. F.; Andersen, C. L.; et al. An Optimized Shotgun Strategy for the Rapid Generation of Comprehensive Human Proteomes. Cell Syst 2017, 4 (6), 587–599.e4.

(24)

Cox, J.; Mann, M. MaxQuant enables high peptide identification rates, individualized ppb-range mass accuracies and proteome-wide protein quantification. Nat. Biotechnol. 2008, 26 (12), 1367–1372.

(25)

Team, R. C. The R Project for Statistical Computing: R Foundation for Statistical Computing, Vienna, Austria; ISBN 3-900051-07-0. Available: http://www. R-project. org/. Accessed 2013 Sept 29, 2012.

(26)

Zhao, S.; Guo, Y.; Sheng, Q.; Shyr, Y. Heatmap3: an improved heatmap package with more powerful and convenient features. BMC Bioinformatics 2014, 15 (Suppl 10), P16.

(27)

Vizcaíno, J. A.; Deutsch, E. W.; Wang, R.; Csordas, A.; Reisinger, F.; Ríos, D.; Dianes, J. A.; Sun, Z.; Farrah, T.; Bandeira, N.; et al. ProteomeXchange provides globally coordinated proteomics data submission and dissemination. Nat. Biotechnol. 2014, 32 (3), 223–226.

ACS Paragon Plus Environment

28

Page 29 of 29 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of Proteome Research

(28)

Murray, K. K.; Boyd, R. K.; Eberlin, M. N.; Langley, G. J.; Li, L.; Naito, Y. Definitions of terms relating to mass spectrometry (IUPAC Recommendations 2013). J. Macromol. Sci. Part A Pure Appl. Chem. 2013, 85 (7), 1515–1609.

(29)

Rockwood, A. L.; Erve, J. C. L. Mass spectral peak distortion due to Fourier transform signal processing. J. Am. Soc. Mass Spectrom. 2014, 25 (12), 2163–2176.

For TOC Only:

ACS Paragon Plus Environment

29