High-coverage immunopeptidomics using timsTOF mass spectrometers with Thunder-DDA-PASEF boosted by MS2Rescore

doi:10.21203/rs.3.rs-4849156/v1

Download PDF

Method Article

High-coverage immunopeptidomics using timsTOF mass spectrometers with Thunder-DDA-PASEF boosted by MS²Rescore

https://doi.org/10.21203/rs.3.rs-4849156/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

Major histocompatibility complex (MHC, or Human leukocyte antigen, HLA) peptide ligands can be exploited to develop immunotherapies targeting immunogenic disease-specific immunopeptides, such as virus- or cancer mutation-derived peptides. Liquid chromatography-coupled with mass spectrometry (LC-MS)-based immunopeptidomics is the gold standard for identifying MHC ligands. We previously optimized a workflow enabling the identification of more than 10,000 MHC class I ligands per cell line. This process comprises three major steps: (I) a high-recovery immunopeptidome enrichment, (II) an optimized MS acquisition in the timsTOF Pro called Thunder-Data-Dependent Acquisition with Parallel Accumulation-SErial Fragmentation (Thunder-DDA-PASEF), (III) and peptide identification using PEAKS XPro boosted by MS²Rescore data-driven rescoring. Here, we describe our workflow for deep-coverage immunopeptidomics step-by-step, from sample preparation to data analysis and validation.

Analytical Biochemistry

Mass Spectrometry

Immunology

Bioinformatics

Immunopeptidomics

human leucocyte antigen

major histocompatibility complex

sample preparation

immunoprecipitation

mass spectrometry

rescoring.

Major histocompatibility complex (MHC, or Human leukocyte antigen, HLA) peptide ligands, known as immunopeptides, play a crucial role in immune surveillance by triggering immune responses when T-cells recognize presented epitopes. This can be exploited to develop immunotherapies targeting immunogenic disease-specific immunopeptides, such as virus- or cancer mutation-derived peptides (1). Liquid chromatography-coupled with mass spectrometry (LC-MS)-based immunopeptidomics is the gold standard for identifying MHC ligands. However, it is challenging due to the high diversity, extensive dynamic range, and individual-specificity of the immunopeptidome (2, 3). To address these challenges, we previously optimized a process enabling the identification of more than 10,000 MHC class I ligands per cell line by exploiting the additional resolution, speed, and sensitivity provided by ion mobility separation (IMS) in timsTOF mass spectrometers (4).

The process of immunopeptide identification comprises three main steps. (I) Immunopeptide – MHC complexes are enriched by antibody-based immunoprecipitation (IP) and immunopeptides are subsequently isolated. We use a previously optimized method (5) with slight modifications (4, 6). (II) The peptides are analyzed by LC-MS. We use a nanoElute 2 coupled with timsTOF Pro2 with an LC-IMS-MS method called Thunder-Data-Dependent Acquisition with Parallel Accumulation-SErial Fragmentation (Thunder-DDA-PASEF) (4, 7). To maximize immunopeptide coverage, this method exploits an extended ion mobility resolution, the “high sensitivity mode” in timsTOF instruments, and a semi-selective fragmentation of precursors corresponding to possible immunopeptides. (III) Peptides are then identified from the raw data. We use PEAKS XPro, followed by rescoring with MS²Rescore to boost peptide identification and confidence (8, 9). Briefly, peptide retention time (RT), IMS, and fragmentation are predicted using DeepLC, ionmob, and the timsTOF MS²PIP model, respectively (10, 11, 12, 13, 14). Then, peptide identifications are rescored based on these features using a semi-supervised learning algorithm through Mokapot (15).

Here, we describe our workflow for high-coverage immunopeptidomics step-by-step, from sample preparation to data analysis and validation. First, samples, such as cultured cells, are obtained (section 3.1), and the antibody for the isolation of peptide-MHC complexes is immobilized on CNBr-activated agarose beads (section 3.2). The cells or tissues are lysed using a non-denaturant buffer (section 3.3) to maintain the integrity of the peptide-MHC complexes. Then, the peptide-MHC complexes are enriched by IP (section 3.4), the peptides are eluted under acidic conditions, separated from longer polypeptides by ultra-filtration (section 3.5), and the flow-through is further purified by solid-phase extraction (SPE, section 3.6). The peptides are then analyzed by LC-IMS-MS on a nanoElute 2 LC coupled with a timsTOF Pro 2 MS, using Thunder-DDA-PASEF (4) (section 3.7). Raw data is processed using PEAKS XPro as a first step for peptide identification (section 3.8), reviewed and exported (section 3.9), and then rescored using MS²Rescore (section 3.10). The quality of the peptide identification results can be assessed using diverse metrics (section 3.11), such as the total coverage, the peptide length distribution, binding prediction to the sample-specific MHCs, and their clustering into the corresponding sequence motifs.

We evaluate the quality of the procedure using the lymphoblastoid cell line JY as a benchmarking and quality control sample. We enrich the MHC class I (MHC1) ligands starting from 1x10⁸ cells per sample preparation replicate. Then, we analyze each sample in triplicate, injecting the equivalent of 1x10⁷ cells per injection replicate. This typically results in more than 15,000 total peptides with 80–90% of 8-13mers, out of which 85 to 95% are predicted to bind the JY MHC alleles (HLA-A*02:01, B*07:02, C*07:02). Thus, our workflow yields more than 10,000 predicted MHC1 ligands per sample in average.

3.1. Critical consumables and substances

Low-binding tubes: To minimize sample loss, we prepare samples and antibodies using Eppendorf™ polypropylene protein LoBind tubes.
Use only LC-MS-grade solvents and formic acid (FA). For instance: Acetonitrile ROTISOLV® ≥99,98%, Ultra LC-MS, Carl Roth, cat. No. HN40.1. Water ROTISOLV®, Ultra LC-MS, Carl Roth, cat. No. HN43.1. Formic acid ROTIPURAN® ≥99%, LC-MS-grade, Carl Roth (cat. no. 1EHK.1).
Use analytical- or LC-MS-level purity grades for all other substances.
Centrifugal concentrators, Vivacon® 500, 10,000 molecular weight cut-off (MWCO ultrafilters, with flat bottom) Hyrosart®, Sartorius (cat. no. VN01H02 from VWR). See Note 1.
Affinity coupling media, CNBr-Activated Sepharose™ 4B (agarose beads, 90 µm), VWR (cat. no. 17-0430-01).
Antibody W6/32. Anti-Human MHC-A, B, C (MHC Class I) - Purified In vivo GOLD Functional Grade, 50 mg, Leinco (cat. no. H263-50mg) at the USA, or imported to Europe by Hölzel-biotech (cat. no. LEIN-H263-50mg).
Solid phase extraction (SPE) plate for peptide desalting and cleaning (Note 2): Oasis Hydrophilic-Lipophilic-Balanced (HLB) 96-well µElution Plate, 2 mg Sorbent per Well, 30 µm.

3.2. Equipment for sample preparation

Tube rotator: We use the Intelli-Mixer Rotator with Vortexer, which has pre-set programs (neoLabLine, cat. No. 7–0045) and adequate adaptors.
Ultrasound bath with cooling system or sonicator lance: System for cell lysis (Note 3). We use the Bioruptor system (Diagenode, cat. No. B01020002).
Rocker: rocking platform shaker with variable speed and adjustable tilt.
Extraction plate manifold for 96-well plates with vacuum pump.
Vacuum centrifuge for sample drying.
Benchtop centrifuge for 15 mL and 50 mL tubes, refrigerated.
Refrigerated benchtop centrifuge for 500, 1500, and 2000 µL tubes, capable of reaching 17,000 x g.

3.3. Cell culture, harvesting and storage

The note “in SC” denotes steps that require aseptic or sterile conditions (SC) and should be performed in a laminar flow hood (LFH).

Penicillin/streptomycin (Pen/Strep): aliquot 5 mL of 10,000 units/mL.
Sodium pyruvate: weigh sodium pyruvate and add MilliQ water for a 1 M solution. Sterile filter with 0.2 µm filter.
Fetal bovine serum (FBS, in LFH): Heat-inactivate FBS for 30 min at 56°C in a water bath. Aliquot FBS in 50 mL tubes and centrifuge for 30 min, 4°C, 3,600 x g. Decant FBS in new 50 mL tubes and store either in the fridge for short-term storage or in the freezer for long-term storage.
Complete medium: Add 5 mL Pen/Strep, 5 mL of sodium pyruvate, and 50 mL of FBS to 500 mL of RPMI medium.
Phosphate-buffered saline (PBS) stock: For 5 L of 10X PBS, weigh 402 g NaCl and 78 g of NaH₂PO₄. Add MilliQ water and adjust the pH to 6.6 with NaOH, and sterilize the solution by filtering.
PBS working solution (Note 4): 1X PBS. Dilute 10X PBS (sterile) to 1/10th with MilliQ water, and sterilize the solution by autoclaving.

3.4. Coupling the antibody to CNBr-Sepharose

HCl 1 mM: In a glass bottle, prepare a stock of 1 M HCl by adding 830 µL concentrated HCl to 9.97 mL of LC-MS-grade water. Dilute 50 µL of 1 M HCL with 49,95 mL of LC-MS-grade water.
Blocking buffer: 0.2 M glycine. Weigh 750 mg glycine and fill up with LC-MS-grade water to 50 mL.
Washing buffer: 1X PBS. Weigh 478 mg PBS and fill up with LC-MS-grade water to 50 mL.
Coupling buffer: 0.5 M NaCl and 0.1 M NaHCO₃. Weigh 880 mg NaCl and 250 mg NaHCO₃. Transfer them to one tube and dissolve them in approximately 25 mL of LC-MS-grade water. Adjust the pH with NaOH to 8.3. Fill up with LC-MS-grade water up to 30 mL.
Antibody (Note 5): An antibody recognizing the MHC of interest should be used. Typically, we use the anti-pan-HLA (HLA-A, -B, -C) class 1 antibody W6/32 for class 1 ligands. We purchase the antibody from LEINCO (cat. No. H263) via the European distributor Hölzel (cat. No. LEIN-H263). The antibody is delivered in 1X PBS at concentrations above 5 mg/mL.

3.5. Preparation of lysate (cell pellet).

Solubilization buffer: PBS with protease inhibitor cocktail (PIC). Dissolve 400 mg of chaps and one tablet of cOmplete protease inhibitor tablet (EDTA-free) with 33 mL of 1X PBS for the 2X solubilization buffer. For the 1X solubilization buffer, dilute 2X solubilization buffer 1/2 with 1X PBS, e.g., 10 mL PBS and 10 mL 2X solubilization buffer.

3.6. Immuno-affinity purification of MHC-peptide complexes and elution of ligands

Use LC-MS-grade water for all solutions unless indicated otherwise.

Trifluoroacetic acid (TFA) 10%: add 9 mL of LC-MS-grade water to a glass vial. Add 1 mL of 100% TFA on top.
TFA 0.2%: add 9.98 mL of LC-MS-grade water to a glass vial. Add 20 µL of 100% TFA on top.

3.7. Ultrafiltration

TFA 1%: add 9.9 mL of LC-MS-grade water to a glass vial. Add 100 µL of 100% TFA on top.

3.8. Desalting and drying

LC-MS-grade Acetonitrile (ACN) 100%
TFA 0.1%: add 9.99 mL of LC-MS-grade water to a glass vial and 10 µL of 100% TFA.
Elution buffer (Note 6): 35% ACN, 0.1% TFA. Add 350 µL of ACN to 640 µL of LC-MS-grade water and 10 µL of 10% TFA.
Formic acid (FA) 0.1%:

3.9. LC-MS instrumentation and consumables

NanoLC-MS instrumentation: NanoLC-MS analysis is performed using a nanoElute 2 coupled to a timsTOF Pro 2 mass spectrometer equipped with a Captive Spray source (Bruker Daltonics) (Note 7).
Analytical column: C18 Reversed-phase (RP) Aurora 25 cm x 75 µm ID, 120Å pore size, 1.7 µm particle size, acquired from IonOpticks (Australia).
Mobile phases: Phase A. 99.9% water with 0.1% formic acid. Phase B. 99.9% acetonitrile with 0.1% formic acid. Solvents and formic acid must be LC-MS-grade.

3.10. Software for data analysis

PEAKS XPro: PEAKS (18) is a proprietary software commercialized by Bioinformatics Solutions Inc. (BSI, Waterloo, Ontario, Canada). For comments about the PEAKS version and alternative software, see Note 8 and Note 9, respectively.
MS²Rescore v3 (Fig. 1): Open any internet browser and access the MS²Rescore website (https://ms2rescore.readthedocs.io/en). On the left panel, click on Installation (1), then Windows Installer (2). Click “install for windows” (3) to download the exe file. Once downloaded, install it following the wizard. This will install the software and graphical user interface (Note 10).

4.1. Cell culture, harvesting and storage

JY cells are cultured in complete medium as suspension cells. All these steps are performed under aseptic conditions, following biosafety measures according to the samples being processed.

Harvest cells from flasks: dispatch the cell suspension into as many 50 mL tubes as needed for the cell suspension intended to be harvested. Centrifuge at 220 x g at room temperature (RT) for 10 min., and discard the supernatant.
Wash two times with PBS: add approximately 20 mL of 1X PBS per 50 mL tube, gently resuspend with a serological pipette, and pool the cell pellets if necessary. Centrifuge at 220 x g and RT for 10 min, and carefully discard the supernatant by decanting.
Aliquot cells into pellets of 1 x10⁸ cells per replicate: Add approximately 20 mL of 1X PBS 50 mL tube and gently resuspend by pipetting up and down. Then, take an aliquot and estimate the total cell amount (e.g., with Trypan blue in Neubauer chamber). Distribute the volume required for 1x10⁸ cells in aliquots in 5 mL or 15 mL low-binding tubes, centrifuge at 220 x g and RT for 10 min and carefully discard the supernatant.
Freeze and store the tubes with the cell pellets at -80°C until usage.

4.2. Coupling the antibody to CNBr-activated Sepharose

The preparation of antibody-beads is scaled according to the total number of cells to be processed, as described in Table 1. To prepare beads for up to 9 x10⁸ cells, use a 15 mL tube. To prepare beads for more cells, use a 50 mL falcon. Here, we will use the scaled values indicated for six samples of 1x10⁸ cells. Use a different tube for each antibody to couple.

Table 1

Preparation of Sepharose beads for antibody binding.
No. of Samples	1	6	9
Total cells	1x10⁸	6x10⁸	9x10⁸
Sepharose beads	24 mg	144 mg	216 mg
HCL 1 mM	1.67 mL	10 mL	15 mL
Glycine 0.2 M	1.67 mL	10 mL	15 mL
Coupling Buffer	0.67 mL	4 mL	6 mL
PBS	11 mL	16 mL	19 mL

Weigh and wash the CNBr Activated Sepharose™ beads: Weigh 24 mg of beads per sample in a 15 mL low-binding tube, add 10 mL of 1 mM HCl (Note 11), and gently agitate the tubes by continuous rotation at 15 rpm for 30 min at 15 rpm (Rotator in program C2, without shaking) (Note 12).
Prepare the beads for coupling: Centrifuge for 4 min at 20 x g at RT (brakes = 2), carefully remove the supernatant using a pipette and discard it (Note 13). Add 2 mL of coupling buffer to pre-equilibrate the beads at pH 8.3 (to avoid antibody denaturation) and mix gently.
Couple the antibody to the beads: To ensure a homogenous distribution of antibody in the stock solution, mix it by gently pipetting up and down, then transfer the equivalent volume for 0.6 mg of antibody per 1x10⁸ cells sample (e.g. 3.6 mg for six samples) to the tube containing the beads (Note 14), top up to 4 mL with coupling buffer, and incubate at RT for 2 h with rotation at 20 rpm (Rotator, program F5).
Wash the antibody-beads: Centrifuge them for 4 min at 20 x g at RT (brakes = 2), carefully remove the supernatant using a pipette and discard it.
Block the remaining active sites on the CNBr Activated Sepharose™ beads: Add 10 mL of blocking solution (0.2 M Glycine) to the washed antibody-beads, and incubate them for 60 minutes at RT with rotation at 20 rpm (Rotator program F5).
Wash the blocked antibody beads two times (Note 15): Centrifuge the antibody beads for 4 min at 20 x g, at RT (brakes = 2), carefully remove the supernatant using a pipette, discard it, add 5 mL of 1X PBS, resuspend beads by gentle agitation and mild shaking by hand, and repeat this step.
Split the washed antibody-beads into individual tubes for each sample: Add 1X PBS to the stock of antibody-beads to top up the level to 6 mL (Note 16), resuspend the beads by gentle agitation and mild shaking by hand, and split the antibody-beads suspension according to the number of samples into low-binding tubes (5 mL, 2 mL, or 1.5 mL, depending on the lysate volume).

4.3. Preparation of lysate for cell pellets (day 1)

Thaw the cell pellet on ice. Add approximately 2X volumes of solubilization buffer per volume of cell pellet. For example, if the pellet volume is approximately 2.5 mL, add 2.5 mL of 2X solubilization buffer. We typically use 2.5 mL of buffer for 5x10⁸ JY cells. Incubate on ice on the rocker with gentle rocking for 1 hour (speed = 15, tilt = 15).
Dilute the lysate to ½: Add the same volume of 1X solubilization buffer as the previous step, and incubate on ice on the rocker for 1 hour (speed = 15, tilt = 15).
Sonicate the lysate to ensure complete dissociation: Sonicate in Bioruptor once for 15 cycles of 30 s on/off, high mode, 4°C (Note 17).
Incubate the samples on a rocker for 1 hour (speed = 15, tilt = 15). Meanwhile, cool down the centrifuge to 4°C.
Remove cell debris by centrifugation: Transfer lysate into 1.5 mL low-binding tubes, using multiple tubes per sample if necessary. Centrifuge lysate for 20 min at 17,000 x g at 4°C. Carefully transfer the supernatant into new 1.5 mL low-binding tubes. Centrifuge the supernatant again for 30 min at 17,000 x g at 4°C (Note 18).

4.4. Immuno-affinity purification of MHC-peptide complex (day 1) and elution of ligands (day 2)

Prepare antibody-beads for adding the lysate: Centrifuge the antibody-Sepharose beads (prepared in Section 3.3) for 4 min at 20 x g, at RT, carefully remove the supernatant using a pipette and discard it.
Incubate the lysate with the antibody-beads: Transfer all supernatants into the low-binding tube with the antibody-Sepharose-bead pellet as prepared in section 3.2. Incubate with slow rotation overnight, on the rotator at 10 rpm, 4–8°C, program C2.
Recover the antibody-beads with bound peptide-MHCs, and the unbound fraction (supernatant): Centrifuge the samples for 4 min at 20 x g at 4°C (Note 19). Carefully transfer the supernatant to a fresh tube. Store the supernatant (unbound fraction) at 4°C if used within two days otherwise at -80°C (Note 20).
Wash the peptide-MHC-antibody-beads two times with PBS: Add 5 mL PBS (1 mL /1.5 mL for 1.5 mL/2 mL tubes). Rotate samples for 10 min at 10 rpm at 4°C with program C2. Centrifuge MHC-IP beads for 4 min at 20 x g at 4°C. Carefully remove and discard the supernatant. Add 5 mL PBS (or 1 mL for 2 mL tubes, or 1.5 mL for 2 mL tubes) and rotate the sample on a rotator for 30 min at 10 rpm at 4°C with program C2. Centrifuge the beads for 4 min at 20 x g at 4°C (Note 19). Carefully discard the supernatant.
Wash the peptide-MHC-antibody-beads with water: Add 5 mL water to the beads pellet (or 1 mL for 1.5 mL tubes, or 1.5 mL for 2 mL tubes). Incubate on the rocker for 1 hour (speed = 15, tilt = 15). Centrifuge MHC-IP beads for 4 min at 20 x g at 4°C (Note 19). Carefully discard the supernatant. Remove the remnant liquid from the tube using a 20 µL pipette (Note 21).
Elute the immunopeptides (Note 22): elute samples seven times in total. For the first elution only, add 15 µL 10% TFA per mg AB to the MHC-IP beads, e.g., 3 mg AB = 45 µL, 0.6 mg AB = 9 µL 10% TFA. Shake gently by hand. Incubate on ice on the rocker for 15 min (speed = 25 rpm, tilt = 15). While waiting, prepare the collection tubes. Centrifuge the MHC-IP beads for 4 min at 20 x g and 4°C. Carefully transfer the supernatant (eluate) into the prepared 1.5 mL low-binding Tube (on ice). Repeat six times more using 200 µL 0.2% TFA each time and pool all eluate in one tube (total volume = 1.4 mL).
Incubate overnight at 4°C (Note 23).

4.5. Ultrafiltration (day 3)

Prewash the ultrafilter: centrifuge the ultrafiltration unit twice with 300 µL 1% TFA for 20 min at 16,300 x g and 4°C.
Prepare two collection tubes: prepare two fresh 1.5 mL low-binding tubes to recover the flow-through of each sample by cutting off the caps (Note 24).
Centrifuge the eluate for 10 min at 16,300 x g, at 4°C.
Ultrafiltrate sample: set the filter unit on top of one of the collection tubes, add 350 µL of the eluate on the filter (Note 25), and centrifuge for 15 min at 16,300 x g/ 4°C. Repeat this step once more (Note 26).
Transfer the filter to the second collection tube and repeat the previous step until all the eluate has been ultrafiltered.

4.6. Desalting and drying (day 3)

For desalting, a 96 well HLB and vacuum manifold is used (Note 2). For each step of solvent elution through the cartridge, after adding the solvent to the wells in use (Note 27), activate the vacuum system by closing the chamber and starting the pump. Keep on operating until the liquid elutes, but avoid letting the solid phase dry completely since this affects recovery (Note 28).

Mount the Oasis HLB plate in the vacuum manifold over a waste tray or a 96-well plate.
Condition the wells: add 200 µL 100% ACN to each well in use of the Oasis HLB plate and activate the vacuum system.
Equilibrate the wells: add 400 µL of 0.1% TFA and activate the vacuum system. Repeat this step once more.
Load samples (Note 29): At this stage, the same well can be used for both collection tubes of the same sample, but do not mix samples. Add up to 800 µL of sample at once in a well and activate the vacuum system (Note 30). Repeat until all the sample is run through.
Wash the HLB wells: add 800 µL 0.1% TFA per well. Afterward, wash with 200 µL of water per well.
Mount the plate over a fresh low-binding 96-well plate (Note 31).
Elute the samples: make sure a collection plate is set under the plate and the lid is removed. Add 50 µL of 35% ACN/0.1% TFA and activate the vacuum system to elute the sample. Then, repeat this step once more.
Prepare samples for drying: Take the collection plate from the vacuum manifold and centrifuge it for 1 min, 800 x g at 4°C. Prepare 0.5 mL low-binding tubes and punch a hole into the lid from inside to outside. Transfer the samples into the tubes and freeze at -80°C.
Dry and store the samples (Note 32): Evaporate the eluates in the vacuum centrifuge until the samples are nearly dry. They can then be stored at -80°C until dissolving them for LC-MS measurement.
Prepare the samples for LC-MS measurement: Dissolve samples in 15 µL of 0.1% FA each by pipetting up and down the walls of the tube. Then, transfer the sample into an LC-MS vial. Place the LC-MS vials in the autosampler of the nanoElute 2 for immediate analysis (recommended), or store them at -20°C.

4.7. LC-MS configuration and analysis

Mount the CaptiveSpray source on the timsTOF Pro 2 mass spectrometer and set up the nanoElute 2 LC following recommendations of the provider (Bruker). Install the analytical column on the nanoElute 2 system as recommend by the providers (IonOpticks and Bruker). No trap column is required as the desalted peptides are directly injected onto the C18 analytical column in the present protocol.
Program the LC gradient using the method editor in Hystar (Note 33): To setup your chromatographic method, choose “one-column separation” mode in the chromatographic method editor, set the column temperature to 50°C and choose default settings for separation column equilibration and sample loading. Using default settings, column is equilibrated by a flush with four column volumes of solvent A. Equilibration and sample loading are both conducted at 800 Bar (Note 34). Program the following gradient for peptide separation in the method editor: Increase solvent B from 2% (v/v) to 17% (v/v) in 23 min and 25% (v/v) in 34.5 min, then to 37% (v/v) over 3.8 min, followed by ramping up B% to 95% (v/v) over another 3.8 min. Wash column for 5 min with 95% (v/v) B. Set the flow rate to 400 nL/min (see also Table 2).
Create a new DDA-PASEF method for the timsTOF Pro 2: Open timsControl on the MS acquisition computer. Open the default DDA-PASEF method, which should have Scan Mode = PASEF. Save the method with a new name by clicking Method > Save as…, or press keys Ctrl + Shift + S. Enter. Select the directory to save the method, and enter a representative method name. For instance: “Thunder_DDA_MHC1p_300ms_065-175invK0_03MSMS_100-1700mz_LSAon_1600V”.
Program the MS method (Note 35): Configure the MS settings in timsControl as shown in Table 3. At the end, save the method.

Optimal capillary voltage (CV) may vary between columns and over time due to column ageing, we typically use 1,400 to 1,600 V. We recommend verifying it as suggested by Bruker. Place the filter with spiked calibrant masses on the source, set the LC to a flow similar to the gradient start (400 nL/min, 2% B). Set the CV to 1500 V and then reduce it by 50 V every one minute until the signal drops by 50%, then increase it by 300 V. This corresponds to the optimal CV. We typically use capillary voltages between 1500 and 1600 V.
Pre-pulse storage can be lower to prioritize ions with lower m/z (16).
When the high-sensitivity mode is activated, do not inject more than 50 ng of tryptic digest or 200 ng immunopeptide sample to avoid overloading the detector.
m/z range may be increased to 100 − 20,00 m/z for MHC class II ligands (17).
Most MHC class I ligands elute between 0.65 and 1.75 Vs/cm² but this may be adapted in function of the sample-specific profile, or for analyzing MHC class II ligands.
TIMS accumulation time should be the same as ramp time to ensure maximum duty cycle usage.
A TIMS ramp time of 300 ms is optimal for analyzing MHC class I samples of 1x10⁶ to 2x10⁸ JY cell equivalents per injection in a timsTOF Pro 2. However, this may be adapted, for instance with a shorter ramp time for higher sample loads. Importantly, it should be decreased to a maximum of 200 ms for timsTOF SCP instruments.
The number of PASEF frames or MS/MS ramps used for fragmentation should is modulated to keep cycle time between 1 and 1.5 s. It may be modified if the TIMS ramp time is modulated: 10 frames for 100 ms, 5 frames for 200 ms.
Most MHC1 peptides are detected at a charge state between 1 and 3. However, the range should be modified for MHC class II ligands (2 to 6).
The cycle overlap should be about 50% the number of PASEF frames, or one less. For example, 4 for 10 frames, 2 for 5 frames, 1 for 3 frames.
The collision energy could be adapted, but the optimal ramp may vary in function of the sample-specific peptide profile (17).

Design the PASEF polygon (Fig. 2) (Note 36): Find the TIMS View in timsControl, which is typically situated at the middle-top of the window. If there is already a polygon (Compound Region) on the 1/K0 vs. m/z heatmap, right-click on it and select “Delete Compound Region” (Fig. 2a). Create two new PASEF polygons using the coordinates indicated in Table 3 as coordinates for the edges. Right-click the empty heatmap and select “Define Polygon Compound Region” (Fig. 2b). Create the first polygon with the coordinates in the inset table in Fig. 2c. Place the cursor at the first coordinate (e.g., 420 ± 2 m/z vs. 0.65 ± 0.02 1/K0) and left-click, then do the same for the remaining three coordinates of the polygon. When the polygon is complete, right-click on the heatmap and select “Finish Compound Region” (Fig. 2c). Repeat this step with the coordinates for the second polygon to obtain the Thunder polygon Fig. 2d). At the end, save the method.

Table 2

LC setup and method settings.
Setting	Value
LC	nanoElute 2
Trap column	None
Analytical column:	Aurora 25 cm x 75 µm 1.6 µm, IonOpticks
Column temperature	50°C
Phase A	Water + 0.1% formic acid
Phase B	Acetonitrile + 0.1% formic acid
Separation column equilibration	4 column volumes
Separation column equilibration pressure	800 Bar
Loading method	µL pick-up (with 2 µL pick-up volume)
Sample loading pressure	800 Bar
Gradient
Flow (constant)	400 nL/min
Time	%B
0	2
23	17
34.5	25
38.3	37
42.1	95
47	95

Table 3

MS setup and method settings for timsTOF Pro 2. The notes indicate critical settings. Optimal settings vary between the different timsTOF models (Note 35).
Category	Setting	Value	Unit	Notes
Source	Source Type	Captive Spray
	Capillary	1,600	V	a
	Dry Gas	3	L/min
	Dry Temperature	180	°C
Tune > General > Transfer	Deflection 1 Delta	70	V
	Funnel 1 RF	500	Vpp
	isCID Energy	0	eV
	Funnel 2 RF	200	Vpp
	Multipole RF	200	Vpp
Tune > General > Collision Cell	Collision energy	10	eV
Tune > General > Collision Cell	Collision RF	1,500	Vpp
Tune > General > Quadrupole	Ion Energy	5	eV
Tune > General > Quadrupole	Low Mass	200	m/z
Tune > General > Focus Pre TOF	Transfer time	60	µs
Tune > General > Focus Pre TOF	Pre-pulse storage	12	µs	b
Tune > General > Detection	High Sensitivity Detection	TRUE		c
Tune > Processing > MS Peak detection	Sum of Intensities (Area)	FALSE
	Absolute Threshold	10
	Absolute Threshold (per 100 ms Accu Time)	3
Tune > Processing > PASEF Data Reduction	Denoising mode	No reduction
Tune > Processing > Mobility peak detection	Absolute Threshold	5,000
MS settings	m/z range	100–1,700	Th	d
	Polarity	Positive
	Scan mode	PASEF
TIMS settings	1/k0 Start	0.65	Vs/cm²	e
	1/k0 End	1.75	Vs/cm²
	Accumulation time	300	ms	f
	Ramp time	300	ms	g
	Duty cycle (locked)	100	%
	Ramp rate	3.27	Hz
	MS Averaging	1	x
MS/MS > Precursor Ions	Number of PASEF ramps	3	frames	h
	Total cycle time	1,220	s
	Charge Minimum	1		i
	Charge Maximum	3
	MS Repetitions	1	x
	Cycle Overlap	1		j
	Summation Width	25	pts
	Max. No. of Peaks	3
MS/MS > Scheduling	Precursor repetitions	Linear
	Target intensity	20,000
	Intensity threshold	1,000
MS/MS > Active Exclusion	Active exclusion	TRUE
MS/MS > Active Exclusion	Release after	0.4	min
MS/MS > Collision Energy (ramp)	1/k0	0.65–1.60	Vs/cm²	k
MS/MS > Collision Energy (ramp)	Collision Energy	20.0–59.0	eV
MS/MS > Isolation Width	Mass	700–800	Th
MS/MS > Isolation Width	Width	2.00–3.00	Th

Perform first mass then mobility calibration according to the protocol of the manufacturer: Load the acquisition method created in the previous step in timsControl by clicking on “Method”, then “Load”, then navigate to the corresponding folder, select the file and click on “Open”. If the mass spectrometer was in standby mode, set the instrument to operation by clicking on the on/off icon, then “Operate.” We suggest calibrating the TIMS every three days or before starting a batch. If the instrument was previously on standby or the TIMS settings changed, wait 30 minutes before calibrating to ensure the source has stabilized. Calibrate the TOF and the TIMS according to the recommendations of the provider (Bruker). For instance, to calibrate the TIMS, use the built-in “inject calibrant” function on the nanoElute to inject 2 µL of 2 mM sodium formate and calibrate the TOF using the corresponding list. To calibrate the TIMS, replace the 2 µm filter on the source with a similar filter spiked with the calibrant masses, let the calibrant diffuse into the source for 5 minutes, then place back the filter that will be used during measurements, adjust the TIMS pressure and calibrate using the “filter mass calibrants” list.
Create a new sample list (Fig. 3a, b). In Hystar, create a new list by clicking on New [1] and save it by clicking Save as… [2]. Enter a representative name [3], and click OK [4].
Configure the sample list (Fig. 3c): In the sample list table, fill the rows with the corresponding information, including the Vial position [5], Sample ID [6] which is included in the file name, Sample Comment with additional information [7], Injection Method as “Standard” [8], and the Separation and MS Methods. To select the LC method, click on the arrow pointing down next to the cells on the “Separation Method column” [9], select “Open,” navigate to the folder containing the method, select it, and click on “Open.” To choose the Thunder-DDA-PASEF acquisition method, click on the arrow pointing down next to the cells on the “MS Method column” [10], select “Open,” navigate to the folder containing the method, select it, and click on “Open.” The methods can be copied and pasted on the remaining cells or dragged down by clicking a cell, then clicking on the little black square that appears on the right-bottom corner of the cell [11], and dragging it up or down. In the column “Line,” the boxes for all the samples to be measured [12].
Upload sample conditions and verify the setup (Fig. 3d): Save the sample list [13], then right-click on the first row [14] and click on “Upload Sample Conditions” [15]. This will upload the LC method to the nanoElute 2 and the acquisition method to timsControl and the MS.
Start the measurements (Fig. 3f): in Hystar, click the “Start” button on top of the Sample List [16], and click “Start Sequence” [17].

4.8. Peptide identification in PEAKS XPro

In PEAKS XPro, it is possible to process data in three different manners. (I) Following the Project Wizard to load the data and configure the search for the whole project, which results in a single report for all the samples. (II) Using the Workflow to process individual samples one after the other using the same settings, resulting in individual reports for each sample. (III) Loading the data first, then right-clicking on individual samples or the whole project and starting the search by selecting DB Search. We will explain the procedure using the Project Wizard (I) for simplicity.

Example data can be downloaded from the Jpost repository (https://repository.jpostdb.org/entry/JPST002044.3). Here, we will use the files corresponding to three injections of 10x10⁶ cell equivalents of JY MHC1 peptides corresponding to Fig. 6 in the original publication (4) (files G230411_011_Slot2-28_1_9449.d, G230411_020_Slot2-28_1_9458.d, and G230411_029_Slot2-28_1_9467.d)

Prepare the protein database FASTA file: It should contain the proteome corresponding to the samples processed, such as the human proteome from UniProt (https://www.uniprot.org/proteomes/UP000005640), and possible contaminants, including the mouse IG2-2a-kappa used for the enrichment. See Note 37 for detailed instructions.
Configure the FASTA protein database in PEAKS (Fig. 4a-d): Open PEAKS XPro, then click on Window/Configuration/Database [1, 2]. Click on Browse… [3], navigate to the location of your local FASTA database, and select it [4]. Select the proper FASTA format database using the drop-down menu [5]. Give your database a name [6] and click on Validate Database to verify if the FASTA file is compatible [7]. Finally, save to the FASTA database list by clicking Add/Update [8].
Enable singly-charged features (Fig. 5): It is essential to perform this step before opening PEAKS if the fragmentation of singly charged ions was enabled during acquisition, as it is done in Thunder-DDA-PASEF for MHC1 ligands (Note 38). If PEAKS was already open, close it. Open the Windows Explorer and go to the following path: C:\PEAKSStudioXpro\algorithmpara [1]. Open “feature_detection_para.properties” [2] with any text editor (e.g., notepad). Go to the end of the document and change “timstof_feature_min_charge = 1” instead of the default 2 [3], save the file, and close it.
Create a new project in PEAKS and add data (Fig. 6a-e): Click on New Project [1] to open the Project Wizard. In the Project Wizard window, enter a project name [2]. If needed, select a different Project Location by clicking on “Browse…” [3] and selecting the desired directory (Note 39). Then, click on “Add Data…” [4], navigate to the directory where your raw files are located, select the files [5] and click on “Open” [6]. The files will appear on the left side as a list [7]. Click on “Select All” [8] and click on the button with an icon showing a plus symbol and three Erlenmeyer flasks to add each file to a new sample [9] (Note 40). If desired, edit the sample names by clicking them and writing the replacement text [10]. Using the drop-down menus on the first sample [11], select Enzyme = None; Instrument = TIMS-TOF; Fragment = CID; Acquisition = DDA. Click on “Copy to Whole Project” [12] to use the same settings for all the samples. Verify that all the files to be processed were included and that the settings are correct (Note 41). Click on “Workflow Selection >” [13].
Select workflow (Fig. 7a): In the Workflow Selection, select “Sequence Database Search” [1], then click on Identification > [2].
Configure Data Refinement Settings (Fig. 7b): In the Data Refinement window, tick the boxes of the following options [3]. To enable recalibration of m/z values, select Correct Precursors [DDA], Mass only. Enable the Associate feature with Chimera Scan [DDA]. To restrain the peak picking to the expected precursors, select Filter Features, keep features satisfying, m/z between 200 and 1500, and charge between 1 and 4 (Note 42). Then, click on the Database Search > button [4].
Configure the de novo-based database search on the PEAKS Search window (Fig. 7c): The following settings are advised for MHC1 ligands acquired using Thunder-DDA-PASEF in timsTOF Pro instruments. Setup Error Tolerance [5]: Precursor Mass Tolerance = 15 ppm using monoisotopic mass and Fragment Ion Tolerance, 0.03 Da (timsTOF). Setup Enzyme [6]: Enzyme = None, Digest mode = Unspecific. Set maximum allowed variable PTM per peptide = 2 [7]. Setup the database [8]: select the database configured in step 2. General Options [9]: enable Estimate FDR with decoy-fusion and turn off PEAKS PTM and SPIDER searches (Note 43).
Configure peptide posttranslational modifications (PTM) (Fig. 7c-e): On the PEAKS Search window, click on Set PTM [10]. In the PTM options window, select tab Common [11], then select “Acetylation (Protein N-term)” [12] and click on the “=>” button next to “Select Variable PTM” [13] to add it as a variable modification. Repeat the last step for “Oxidation (M)” in tab Common, and Cysteinylation in tab Uncommon [14] (Note 44). When all the modifications are added [15], click “OK” [16].
Start the search (Fig. 7c): Verify that the settings are correct and click Finish [17]. PEAKS will start loading and then processing the data (Fig. 7f). The time required depends on the computer configuration and the number and size of the files.

4.9. Reviewing and exporting results

When PEAKS finishes the process, the results nodes are added on the Results View at the left side of the window, below the files or project names.

Configure the report for initial inspection at 1% Peptide FDR (Fig. 8a-c). To open the report, double-click on the database node (PEAKS #) [1]. This may take some time. The following settings are recommended for a first inspection of the results without further rescoring. In the Summary tab, configure the filtering setting [2]: click on the Peptides FDR button and select 1.0% in the pop-up window; for Proteins, enable the case − 10lgP, then use the drop-menus to set − 10lgP ≥ 0 and ≥ 0 unique peptides, and do not allow with significant peptides. Click on Apply [3] to implement the changes; this may take some time. When PEAKS finishes, Fig. 1 in the summary report should show an FDR of around 1.0% (it may be slightly lower or higher).
Explore the results Summary (Fig. 8c). Explore the report generated at 1% FDR to evaluate the data quality (Note 45). Further data quality metrics are described in Section 3.11.
Export the results in CSV format (Fig. 9a, b). In the Summary tab, click on Export [1] and select the tab Text Formats [2]. Tick (at least) the options [3]: DB search peptides (peptide.csv), DB search peptide-spectrum matches (DB search psm.csv) (Note 46), and Filters for documentation purposes. If required, click on Browse [4] and select the desired destination directory, then click OK. Finally, click Export [5] to generate the files (Note 47).
Configure results for 100% FDR (all peptides) and enable showing decoys in PEAKS XPro (Fig. 9c-e): This step is essential to rescoring with MS²Rescore. In the Summary tab, select Peptides − 10lgP ≥ 0 using the drop menu [6], then click on Apply [7]. This will result in PEAKS showing the results without FDR filtering (i.e., 100% FDR. Click on Window and select Preferences [8]. In the Preferences windows, select General / Display Options [9] and enable Show Decoys Hits [10], then click OK [11], and click again on Apply [7]. Verify that decoys are included in the reports (Note 48).
Export the results for rescoring (Fig. 9f). To export the identification spectra files, click on Export [12] and select the tab For Third Party [13], tick the option peptides - mzIdentML (peptides_1_1_0.mzid) [14]. Use a different folder name than for the 1% exports, such as adding “_FDR100” at the end [15]. Finally, click on Export to start generating the files [16]. Depending on the computer performance, number of files and peptides identified, this may take some time. Export the peptide, PSM, and filter CSV reports as described in step 3 (Fig. 9a, b, points 1 to 5), entering the “…_FDR100” folder name or another destination to avoid overwriting the 1% FDR files (Note 49).
Verify that the results were correctly exported. The resulting folder should contain one mgf file for each raw data file and a single peptides_1_1_0.mzid file. These are the files that MS²Rescore will use.

4.10. Rescoring using MS²Rescore v3

The main settings to be modified in the MS²Rescore GUI are mentioned in the steps below. As shown in Fig. 10, the default values can be used for the remaining settings.

Configure MS²Rescore main settings (Fig. 10a-d): Open MS²Rescore. Then, select the files to be rescored. In the Select identification file section, click on Browse files [1], navigate to the files exported at 100% FDR including decoys [2], select the mzid file [3], and click Open [4]. Then, in the section Select MGF/mzML file or directory, click on Browse directories [5], navigate to the corresponding folder [6), and click on Select Folder [7]. Using the drop menu, select the number processors you want to use dependant on the maximum number of processors and on the available memory on the machine [8] (Note 50, Note 51).
Configure the Advanced settings for PEAKS XPro results (Fig. 10e): Click on the Advanced tab [9], then enter this text "index=(\\d+)", without quotation marks, in the cases PSM ID regex pattern and Spectrum ID regex pattern [10] (Note 52).
Configure the Feature generators settings for timsTOF data (Fig. 10f): Click the Feature generators tab [11]. Then, select the MS²PIP model timsTOF from the drop-menu [12] and set the MS2 error tolerance in Da [13] to the same value used for PEAKS XPro (0.03 Da). DeepLC is by default enabled without transfer learning which can stay this way [14] (Note 53). To enable ion mobility prediction as a rescoring feature, click “Enable ionmob” and select the GRUPredictor [15].
Configure the Rescoring engine (Fig. 10g): Click on the Rescoring engine tab [16] and verify that the default settings are correct. Mokapot should be selected as Rescoring Engine [17], and Write TXT output files should be enabled [18].
Start MS²Rescore: Click on Start (Fig. 10a; [19]). The Progress will be shown in the Progress window section. At the end, a message indicating that “Program finished successfully!” will pop up. Click on Close (Fig. 10h; [20]).
Explore the overview MS²Rescore report section (Fig. 11): Open the html report generated in the output directory. Evaluate, in the Overview tab, the proportion of PSMs and Peptides gained by rescoring, the score distribution of the decoys vs. target peptides, the false discovery rate, and the identification overlap.
Explore the rescoring features of the MS²Rescore report section: in the html report go to the rescoring features tab where on the top you can examine the feature weights on the rescoring model (Fig. 12), and the performance of the models (Fig. 13). The former shows the individual feature weight and the sum for each feature generator highlighting the contribution of each to the rescoring process, while the latter show the individual performance of the main 3 feature generators: MS²PIP, DeepLC, and Ionmob. Check the performance of each to be sure that the feature generation process went smoothly. This entails that the correlation of MS²PIP is sufficiently high (Note 54) and that the DeepLC (Note 55) and Ionmob predictions are in line with the observations (Fig. 13). Otherwise something might have gone wrong during processing of PSMs or with the predictors itself. This can then be resolved by tweaking the settings in the advanced rescoring tab (Note 56).

4.11. Data evaluation

The performance of the whole procedure can be assessed, for instance, by counting the number of unique sequences, evaluating their peptide length distribution, and their predicted binding to the HLA alleles of the sample (1, 2). Here, we provide instructions using Excel for simplicity. However, other strategies can provide a deeper insight, as detailed in Note 57.

Open the peptides results file (Fig. 14a, b, c): In Excel (or another editor), click on File, then browse, and select “All files” using the drop menu in the pop-up window [1]. Then, navigate to the repository containing the results [2], select peptides_1_1_0.ms2rescore.mokapot.peptides.tsv [3], and click Open [4]. In the Text Import Wizard window: in Step 1 of 3 [5], select the file type as delimited, start import at row 1, and my data has a header, then click Next [6]; in Step 2 of 3, select Tab [7] as only delimiter and click Finish [8].
Obtain the stripped sequences and peptide length (Fig. 14d): To obtain the stripped sequences, select cell N1 (first empty column after the table), name it Sequence, write the following function in cell N2 [9], then copy it in all the cells of the column: =SUBSTITUTE(SUBSTITUTE(SUBSTITUTE(C2, "[Acetyl]-",""), "[Oxidation]", ""), "[Cysteinyl]", "") (Note 58). To obtain the peptide length, select cell O1 (next column), name it Length, write the following function in cell O2, then copy it in all the cells of the column [10]: =LEN(N2).
Filter the table to the required level of FDR (Fig. 14d-f): Select a cell on the results table, then click Data and Filter to add the filter options [11]. Click on the arrow pointing down next to mokapot q-value [12], select number Filters and then Less than…. [13] In the Custom AutoFilter window select “is less than” using the drop menu, enter 0.01 (for 1% FDR) [14] and OK [15].
Create a table with unique sequences and length (Fig. 15a-e): Select the whole table, right-click on it, and select copy [1], then open a new tab, right-click on the first cell, and select paste as values [2]. To keep unique sequences only, select the whole table [3], navigate to the Data toolbar, click Remove Duplicates [4], and click OK in the pop-up window [5]. A pop-up window will appear, indicating the number of unique values conserved, click OK. To format the resulting cells as a table, select the cells [6], navigate to the Home toolbar [7], click Format as table [8], select any format you like. In the pop-up window Format as Table, make sure the correct cells are shown and the option My table has headers is selected in the pop-up window and click OK [9].
Create a summary table counting peptides by length (Fig. 15f-g): To create a summary PivotTable, click on the table created in the previous step [10], navigate to the Design toolbar [11], click Summarize Pivot Table [12], then click OK in the pop-up window to create the table in a new tab. To summarize the number of peptides in function of their length, in the PivotTable Fields window, drag Length to Rows and Sequence to Values [13]. The table should show the peptide length in the first column (Row Labels) and the Count of Sequence in the second [14].
Create a peptide length histogram (Fig. 15g, h): Select any cell of the PivotTable [14], navigate to the Analyze toolbar, select PivotChart [15], then choose Column Clustered and click OK on the Insert Chart pop-up window. The plot is created [16].
Evaluate the peptide length distribution on the resulting histogram (Fig. 15h): Since MHC class I ligands are typically 8 to 13 amino acids long [16], these should comprise more than 70% of the identifications. Since most of the ligands are 9-mers, there should be a peak at this point. A deviation in this distribution could indicate a low MHC enrichment.

Note 1 In our experience, the Vivacon® 500, 10,000 MWCO concentrators/ultra-filters (Sartorius) made of stabilized cellulose-based membrane provide the best peptide recovery for MHC class I and II ligands. Other membrane materials, as well as the 5,000 MWCO concentrators, result in poor peptide recovery.

Note 2 Other SPE materials can be used, such as C18. However, we typically obtain higher peptide recovery yields using HLB.

Note 3 Ultrasound cleaning baths commonly used for cleaning metal parts or solvent degassing are unsuitable for cell lysis. If a proper ultrasound system is unavailable, it may be replaced by other homogenization or lysis procedures, but the immunopeptide recovery may be affected.

Note 4 This PBS is only used for cell culture and not in further downstream processes.

Note 5 The antibody should not contain any further protein (e.g., BSA) or conserving agents (e.g., sodium azide). Alternatively, the antibody can be produced and purified from hybridomas. Other anti-MHC antibodies can be used. For instance, we use L243 for MHC-DR. For other MHCs, please see reference (2) where diverse options are further described.

Note 6 Prepare the elution buffer right before using it.

Note 7 Similar methods could be applied to other LC instruments. To transfer the MS method to other timsTOF instruments and further insight into the effect of the settings, we suggest reading previous publications and optimizing similar methods for timsTOF Pro 2 (4, 16) or SCP (17). Importantly, the TIMS ramp time should be limited to not more than 200 ms in SCP instruments to minimize unintentional fragmentation.

Note 8 PEAKS XPro is the predecessor of PEAKS 11. PEAKS 11 includes a built-in rescoring system. However, the “boost” is based only on retention time. In addition, it is currently impossible to export results with the decoys for rescoring by external software. Thus, we use the PEAKS XPro version, which supports the export of results at 100% FDR, including decoys, for downstream processing with MS²Rescore.

Note 9 Sage and Fragpipe are both fast and free alternatives to PEAKS. Sage (19) can be easily combined with MS²Rescore. Sage can be easily run with SearchGUI (20) and all files can be directly input to MS²Rescore, when setting the file type to Sage. Another alternative to PEAKS and MS²Rescore is Fragpipe. this is a free software package continuously developed by the Nesvizhskii lab (https://fragpipe.nesvilab.org/). The built-in search engine MSFragger (21) combined with the rescoring algorithm MSBooster (22) can provide similar immunopeptidome coverage compared to our workflow using PEAKS XPro + MS²Rescore, also starting from Bruker timsTOF raw files (.d).

Note 10 MS²Rescore can also be used as a python package or a standalone command line tool. Further information can be found in the documentation website, including instructions for installation (https://ms2rescore.readthedocs.io/en/latest/installation/), and dedicated help sections for both the python package usage (https://ms2rescore.readthedocs.io/en/latest/tutorials/in-depth-python-api/) and the command line implementation (https://ms2rescore.readthedocs.io/en/latest/cli/).

Note 11 For this step, a low pH is required to avoid an activation and, therefore, unspecific binding to the Sepharose beads.

Note 12 To ensure that beads are homogeneously mixed with the buffers and substances, the agitation steps are performed by continuously rotating the tubes containing the beads. This is performed at low speeds to avoid damaging the beads.

Note 13 If bead pellets get “disturbed” by any point of taking off the supernatant, re-centrifuge. Also, use a small pipette tip to remove the remaining liquid. The coupling of antibodies to the beads works best at a high pH. The coupling buffer will support the correct pH. Remove as much HCl as possible to avoid reducing the pH due to the prior added HCl.

Note 14 Take an aliquot (e.g., 50 µL) of the antibody-bead mixture immediately after the addition of the antibody (t₀) and 120 min (t_end) after the centrifugation step. To evaluate antibody-bead binding efficiency, it is possible to run an SDS-PAGE or perform a protein concentration assay.

Note 15 Whenever washing beads, try to wash them down from the wall of the tube to ensure the best immersion of beads in cleaning solutions.

Note 16 : The volume should be easily divisible by the number of samples. We usually use either 1,000 µl or 500 µL per sample. At this step, beads can be stored in the fridge for up to a month. For prolonged storage, it is advised to validate the performance.

Note 17 It is also possible to use a sonication lance for three times 30 s on and a break of 30 s on ice in between. Make sure to clean the lance between samples with LC-MS-grade isopropanol and water.

Note 18 Take a 50 µL aliquot of the whole lysate for Western Blot or proteomics analysis: Store on ice and freeze at -80°C. If possible, quickly freeze in liquid nitrogen before storing at -80°C.

Note 19 It is possible to take a 20 µL aliquot of the intermediary supernatants during the washing steps to evaluate MHC depletion and loss by Western Blot. If this is done, take the aliquot carefully (without disrupting the pellet) after centrifugation and before discarding the supernatant.

Note 20 The unbound fraction can be used for sequential enrichment of peptide-MHC complexes, such as MHC class I followed by MHC class II. The sequential preparation should be prepared on the same or the next day without freezing the sample for better recovery. In addition, to maximize immunopeptide recovery, it is possible to re-incubate the unbound fraction with antibody-beads and process this as an additional sample. We recommend independently analyzing a small sample (0.5 µL or 1/30th of the final sample) resulting from each incubation during method establishment for new types of samples. In subsequent preparations, the samples can be pooled and analyzed together if the second incubation results in a considerable number of peptides identified (e.g., more than 30% of incubation 1).

Note 21 The narrower tip helps to remove the low volume without disrupting the beads.

Note 22 : When adding 0.2% TFA for elution, be sure to wash down beads on the wall of the tube. Place the tube on ice to ensure washing (90°C angle).

Note 23 The overnight incubation at 4°C results in the precipitation of some larger polypeptides. If multiple incubations are performed, the eluate can be stored until all the incubations are ready and then continued with ultrafiltration and desalting of all the samples at once. The eluates can be stored at 4°C for one day or frozen (-20°C) for up to a week.

Note 24 Do not use 2 mL tubes for recovering the ultrafiltrate as they don’t fit well with the filter units.

Note 25 If the ultrafiltration is performed using a tilt rotor, do not add more than 350 µL of sample at once. Otherwise, the sample may spill from the unit when tilted. If the ultrafiltration is performed using a swing rotor, it may be possible to use larger volumes after testing it properly.

Note 26 As long as you continue to add eluate, it does not matter if there is still some remaining liquid on the filter. However, for the last step, ensure that all the liquid passes through.

Note 27 It is not necessary to fill all the wells of the plate. Unused wells can be used for other preparations.

Note 28 If solvent passes through some wells at markedly lower speed than through others, keep continuing the steps described below for those wells, where liquid passes through fast. In case of wells where liquid passes through at lower speed continue the protocol once liquid has run through or when the rest of the samples is done.

Note 29 It is possible to collect the flow-through (FT) by mounting a 0.5 mL low-binding Tube under the HLB well within the vacuum chamber. Remember to demount after collecting the FT.

Note 30 It is possible to keep the vacuum system working while adding the sample. However, do not let the wells get dry. To avoid this, restrict the air flow by compressing the vacuum tubing slightly during sample application.

Note 31 Samples can also be eluted into 0.5 mL tubes. Ensure that the 0.5 mL Tube is “locked” underneath the plate properly. If a plastic “bridge” is in the way, use the scissors to cut a vertical line for tube insertion.

Note 32 If drying samples in 0.5 mL tubes, place them in 1.5 mL tubes as adaptors to prevent tube loss. Lyophilization takes 30 min to 1h 30 min.

Note 33 The gradient described in this chapter works well for JY MHC class I peptides. However, immunopeptides may present different elution patterns in function of their MHC binding motifs. Thus, the gradients could be further optimized for the samples to be analyzed. In addition, longer gradients can be used for higher sample loading amounts.

Note 34 To extend column life time, the sample loading pressure and the equilibration volume can be lowered, e.g., to 600 Bar and 5.0 column volumes, respectively.

Note 35 Optimal settings vary between the different timsTOF models. The following publications can be consulted for more information: timsTOF Pro 2 DDA-PASEF (4, 16), timsTOF SCP DDA-PASEF (17).

Note 36 Since MHC class I peptides are typically 8 to 13mers originating from diverse enzymatic cleavages, some may only be detected as singly charged ions. Thus, for MHC class I ligands, the Thunder isolation polygon, including singly charged peptides, is used (Fig. 2d). MHC class II peptides are typically 12 to 25mers and less prone to be detected exclusively as singly charged peptides. Thus, the TIMS range can be set to 0.65 to 1.4 1/K0, and the standard polygon can be used to fragment only the multiply charged ions (Fig. 2a).

Note 37 Download or generate a FASTA protein database. For instance, the canonical human proteome can be downloaded from Uniprot (https://www.uniprot.org/). At the left of the search bar, select “Proteomes,” then enter the human proteome ID “UP000005640”. In the section “Components,” click on “Download,” select “Download only reviewed (Swiss-Prot) canonical proteins,” format “FASTA,” Compressed “No,” click on “Download,” and save it in the desired directory. Download the sequences corresponding to the mouse Igg from Uniprot (accession codes: P01837, P01864, P20760). Open the file(s), add a tag to the headers indicating they are contaminants (for example, >sp|Cont_P01864|GCAB_MOUSE), and copy the sequences with the headers at the end of the proteome FASTA file. Download a suitable set of contaminant proteins (FBS proteins from cell culture, keratins, etc.). We use the lists published by Hao Group (23), available at github.com/HaoGroup-ProtContLib. Make sure they have a tag in the name indicating that they are contaminants, and copy the sequences with the headers at the end of the proteome FASTA file.

Note 38 PEAKS is configured by default to avoid picking singly charged features from timsTOF data. If this setting is not changed, the singly charged peptides may still be identified from the MS/MS spectra, but the peptide area may be assigned to zero since features won’t be obtained during peak picking. However, if singly charged ions were not selected for fragmentation, it is better to keep the default configuration.

Note 39 PEAKS converts the raw data into its format and stores these files in the project folder. Thus, it is crucial to ensure that the project directory has enough memory.

Note 40 It is also possible to manually select files and add them to samples. Files corresponding to different biological, preparation, or injection replicates should each be added as individual samples. Multiple raw files can be added to one sample when they result from fractions (e.g., by high pH C18) of the sample.

Note 41 Verify that the data was loaded with the correct settings and the data selected was acquired using DDA-PASEF. Otherwise, PEAKS may still try to process the data. If the wrong instrument or fragmentation mode is selected, it can result in suboptimal peptide identification. If the wrong Acquisition method was selected, PEAKS may stop the processing and indicate no spectrum was found. If DIA-PASEF data is selected, PEAKS will still try to process it but fail to identify peptides.

Note 42 For MHC class II, If singly charged peptides were not fragmented, this setting can be changed to 2 to 6.

Note 43 Contaminant database is not selected because contaminants are included already in the protein database. This is to avoid that PEAKS removes peptides from proteins considered as contaminants, which could provide an indication of sample quality, such as proteins from the cell culture medium or segments of the antibody used for IP.

Note 44 These modifications were selected based on results from JY MHC1 immunopeptides obtained following the procedure shown in this protocol. After obtaining the Database results, the PEAKS PTM search, and these three PTMs showed the highest numbers of PSMs. The user may want to adapt the PTMs in function of their sample or objectives. Cysteine carbamidomethylation is NOT included since it is not performed during sample preparation.

Note 45 The Summary report in PEAKS provides several summary plots that help evaluate the data quality and the performance of the identification process, such as number of peptides identified, feature m/z distribution and RT distribution, and the precursor mass error. In addition, the Quality Control report can be accessed by right-clicking the report node in the Results View and selecting Quality Control.

Note 46 The peptide.csv file contains all the peptides identified by the database search but does not include those detected only by de novo.

Note 47 WARNING: The results files are created with the settings and filters activated at the moment of the export. For instance, filtered at the FDR or score threshold setup, showing decoys or not, and applying any search filter used in the results tables within PEAKS (e.g., on the Peptide results tab). Remember to apply the FDR threshold, activate or deactivate the decoys, and deactivate any undesired filters before exporting the results.

Note 48 To verify that the decoys are enabled, select the Proteins tab and search for Accession contains = #DECOY. The Accession column should include identifications with this tag.

Note 49 After applying the 100% FDR setting (-10lgP ≥ 0), PEAKS may show an FDR below 100% but higher than 1% in Fig. 1. This usually is not a problem and the results will still contain all the PSMs.

Note 50 Selecting “-1” (default value) will select all of the available processors but sometimes this can result in a bug where MS²Rescore keeps trying to process the files but never finishes. Therefore, if one wishes to select as much cores as possible you should select the highest number of cores in the dropdown menu.

Note 51 WARNING: Selecting the highest number of cores does not always result in the fastest processing, due to more overhead that is created by using more cores which also results in more memory usage. This could even lead to termination of the rescoring due to memory issues. Therefore, it is advised to not always set this to the highest number of cores, especially on bigger machines. Typically, 16 to 32 cores should work reasonably fast without much memory overhead, for very big servers this could be higher 64 or even 128, as memory is typically less of an issue. For smaller computers or laptops it could thus be preferable to lower the number of cores instead of using all of them.

Note 52 The regex patterns differ depending on the search engine used or is sometimes not necessary to provide them, however, for PEAKS this is necessary. In previous versions, we used “(index=[0–9]*)” Spectrum ID regex pattern and “” (no value) as PSM ID regex pattern for PEAKS results. However, the "index=(\\d+)" pattern is more stable. More information can be found in the MS²Rescore documentation website: https://ms2rescore.readthedocs.io/en/latest/userguide/configuration/#mapping-psms-to-spectra.

Note 53 Transfer learning might be advisable for datasets with a very different gradient, i.e. much shorter, used than for training the original models or when the calibration does not seem to align predictions with observations well (which can be seen in the rescore report, see 4.10 step 7). The number of epochs can then be set to 20 to 50 for larger datasets or even higher to 100 to 200 for small datasets with low initial confident PSMs (above 1% FDR) for transfer learning.

Note 54 Lower prediction accuracy could be due to selecting the wrong model, having to high or low MS2 error tolerance, or due to faulty parsing of the PSMs with the spectrum id pattern which results in mismatching spectrum with the identified peptides and therefore a lot of 0 pearson correlations.

Note 55 DeepLC errors can be due to faulty calibration of the predicted retention times, or too few positive identifications, i.e. < 1%FDR. One way to resolve this is by changing the number of calibration set size. Enabling transfer learning can also resolve this this when the number of epochs is set sufficiently high, especially for datasets with lower number of PSMs below the 1% FDR threshold.

Note 56 If MS²Rescore problems persist, it could be due to a very big discrepancy between the training data of the feature generator and the provided data. It could be therefore be wise to disable the according feature generator for rescoring and open an issue on https://github.com/compomics/ms2rescore/.

Note 57 When the HLA alleles of the sample are known, we recommend using MhcVizPipe (24) to evaluate the data. Once installed, MhcVizPipe is easy to use and generates an interactive HTML report including the peptide length distribution, HLA binding prediction using NetMHCpan (25) and clustering using GibsCluster (26). The software and its utilization are well documented on github: https://github.com/CaronLab/MhcVizPipe?tab=readme-ov-file.

Note 58 C2 must be the Peptide column. If different modifications were used, or PEAKS export is being evaluated, adapt the function accordingly.

6. Acknowledgments and funding

We would like to acknowledge Lucas Kleinort (HI-TRON, Mainz) for his technical assistance and contributions to sample preparation. This work was funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) SFB1292 TP13 (H.S.) and TPQ01 (S.T.), the Federal Ministry of Education and Research (BMBF) (MSCoreSys, DIASyM, Fkz: 161L0218, 16LW0241K to S.T.) the Helmholtz-Institute for Translational Oncology Mainz (HI-TRON Mainz) – a Helmholtz institute by DKFZ, Mainz, Germany. S.T., acknowledge funding from Bundesministerium für Bildung und Forschung, (BMBF) as part of the National Research Node “Mass spectrometry in Systems Medicine” (MSCoreSys) [031L0217A/B]. A.D, L.M. and R.G. acknowledge funding from the Research Foundation Flanders (FWO) [1SE3722, G010023N, G028821N, 12B7123N]. L.M. acknowledges funding from the European Union's Horizon 2020 Programme (H2020-INFRAIA-2018-1) [823839], support from the Horizon Europe Project BAXERNA 2.0 [101080544], and funding from the Ghent University Concerted Research Action [BOF21/GOA/033].

Shapiro IE, Bassani-Sternberg M (2023) The impact of immunopeptidomics: From basic research to clinical implementation. Semin Immunol 66:101727. https://doi.org/10.1016/j.smim.2023.101727
Purcell AW, Ramarathinam SH, Ternette N (2019) Mass spectrometry–based identification of MHC-bound peptides for immunopeptidomics. Nat Protoc 14(6):1687–1707. https://doi.org/10.1038/s41596-019-0133-y
Becker JP, Riemer AB (2022) The Importance of Being Presented: Target Validation by Immunopeptidomics for Epitope-Specific Immunotherapies. Front Immunol 13(April):1–10. https://doi.org/10.3389/fimmu.2022.883989
Gomez-Zepeda D, Arnold-Schild D, Beyrle J, Declercq A, Gabriels R, Kumm E, Preikschat A, Łącki MK, Hirschler A, Rijal JB, Carapito C, Martens L, Distler U, Schild H, Tenzer S (2024) Thunder-DDA-PASEF enables high-coverage immunopeptidomics and is boosted by MS²Rescore with MS²PIP timsTOF fragmentation prediction model. Nat Commun 15(1):2288. https://doi.org/10.1038/s41467-024-46380-y
Nelde A, Kowalewski DJ, Stevanović S (2019) Purification and Identification of Naturally Presented MHC Class I and II Ligands. In P. van Endert (Ed.), Antigen Processing: Methods and Protocols (pp. 123–136). Springer New York. https://doi.org/10.1007/978-1-4939-9450-2_10
Hahlbrock J (2017) MHC-Klasse-I vermittelte Antigenpräsentation: systembiologische Analyse in humanen Krebszelllinien und Charakterisierung der ER-residenten Aminopeptidase ERMP1. https://doi.org/10.25358/openscience-1391
Meier F, Brunner AD, Koch S, Koch H, Lubeck M, Krause M, Goedecke N, Decker J, Kosinski T, Park MA, Bache N, Hoerning O, Cox J, Räther O, Mann M (2018) Online parallel accumulation–serial fragmentation (PASEF) with a novel trapped ion mobility mass spectrometer. Mol Cell Proteomics 17(12):2534–2545. https://doi.org/10.1074/mcp.TIR118.000900
Declercq A, Bouwmeester R, Hirschler A, Carapito C, Degroeve S, Martens L, Gabriels R (2022) MS²Rescore: Data-Driven Rescoring Dramatically Boosts Immunopeptide Identification Rates. Mol Cell Proteom 21(8):100266. https://doi.org/10.1016/j.mcpro.2022.100266
Buur LM, Declercq A, Strobl M, Bouwmeester R, Degroeve S, Martens L, Dorfer V, Gabriels R (2024) MS 2 Rescore 3.0 Is a Modular, Flexible, and User-Friendly Platform to Boost Peptide Identifications, as Showcased with MS Amanda 3.0. J Proteome Res 1–23. https://doi.org/10.1021/acs.jproteome.3c00785
Bouwmeester R, Gabriels R, Hulstaert N, Martens L, Degroeve S (2021) DeepLC can predict retention times for peptides that carry as-yet unseen modifications. Nat Methods 18(11):1363–1369. https://doi.org/10.1038/s41592-021-01301-5
Teschner D, Gomez-Zepeda D, Declercq A, Łącki MK, Avci S, Bob K, Distler U, Michna T, Martens L, Tenzer S, Hildebrandt A (2023) Ionmob: a Python package for prediction of peptide collisional cross-section values. Bioinformatics 39(9). https://doi.org/10.1093/bioinformatics/btad486
Degroeve S, Martens L (2013) MS²PIP: a tool for MS/MS peak intensity prediction. Bioinformatics 29(24):3199–3203. https://doi.org/10.1093/bioinformatics/btt544
Degroeve S, Maddelein D, Martens L (2015) MS 2 PIP prediction server: compute and visualize MS 2 peak intensity predictions for CID and HCD fragmentation. Nucleic Acids Res 43(W1):W326–W330. https://doi.org/10.1093/nar/gkv542
Gabriels R, Martens L, Degroeve S (2019) Updated MS²PIP web server delivers fast and accurate MS² peak intensity prediction for multiple fragmentation methods, instruments and labeling techniques. Nucleic Acids Res 47(W1):W295–W299. https://doi.org/10.1093/nar/gkz299
Fondrie WE, Noble WS (2021) mokapot: Fast and Flexible Semisupervised Learning for Peptide Detection. J Proteome Res 20(4):1966–1971. https://doi.org/10.1021/acs.jproteome.0c01010
Hoenisch Gravel N, Nelde A, Bauer J, Mühlenbruch L, Schroeder SM, Neidert MC, Lemke S, Dubbelaar ML, Wacker M, Dengler A, Klein R, Mauz P-SS, Löwenheim H, Hauri-Hohl M, Martin R, Hennenlotter J, Stenzl A, Heitmann JS, Salih HR, Walz JS (2023) TOFIMS mass spectrometry-based immunopeptidomics refines tumor antigen identification. Nat Commun 14(1):1–12. https://doi.org/10.1038/s41467-023-42692-7
Phulphagar KM, Ctortecka C, Jacome ASV, Klaeger S, Verzani EK, Hernandez GM, Udeshi ND, Clauser KR, Abelin JG, Carr SA (2023) Sensitive, High-Throughput HLA-I and HLA-II Immunopeptidomics Using Parallel Accumulation-Serial Fragmentation Mass Spectrometry. Mol Cell Proteom 22(6):100563. https://doi.org/10.1016/j.mcpro.2023.100563
Zhang J, Xin L, Shan B, Chen W, Xie M, Yuen D, Zhang W, Zhang Z, Lajoie GA, Ma B (2012) PEAKS DB: De Novo Sequencing Assisted Database Search for Sensitive and Accurate Peptide Identification. Mol Cell Proteom 11(4). https://doi.org/10.1074/mcp.M111.010587. M111.010587
Lazear MR (2023) Sage: An Open-Source Tool for Fast Proteomics Searching and Quantification at Scale. J Proteome Res 22(11):3652–3659. https://doi.org/10.1021/acs.jproteome.3c00486
Barsnes H, Vaudel M (2018) SearchGUI: A Highly Adaptable Common Interface for Proteomics Search and de Novo Engines. J Proteome Res 17(7):2552–2555. https://doi.org/10.1021/acs.jproteome.8b00175
Kong AT, Leprevost Fv, Avtonomov DM, Mellacheruvu D, Nesvizhskii AI (2017) MSFragger: ultrafast and comprehensive peptide identification in mass spectrometry–based proteomics. Nat Methods 14(5):513–520. https://doi.org/10.1038/nmeth.4256
Yang KL, Yu F, Teo GC, Li K, Demichev V, Ralser M, Nesvizhskii AI (2023) MSBooster: improving peptide identification rates using deep learning-based features. Nat Commun 14(1):4539. https://doi.org/10.1038/s41467-023-40129-9
Frankenfield AM, Ni J, Ahmed M, Hao L (2022) Protein Contaminants Matter: Building Universal Protein Contaminant Libraries for DDA and DIA Proteomics. Journal of Proteome Research, 2022.
04.27.489766
. https://doi.org/10.1021/acs.jproteome.2c00145
Kovalchik KA, Ma Q, Wessling L, Saab F, Duquette JD, Kubiniok P, Hamelin DJ, Faridi P, Li C, Purcell AW, Jang A, Paramithiotis E, Tognetti M, Reiter L, Bruderer R, Lanoix J, Bonneil É, Courcelles M, Thibault P, Sirois I (2022) MhcVizPipe: A Quality Control Software for Rapid Assessment of Small- To Large-Scale Immunopeptidome Datasets. Mol Cell Proteomics 21(1):0–14. https://doi.org/10.1016/j.mcpro.2021.100178
Reynisson B, Alvarez B, Paul S, Peters B, Nielsen M (2021) NetMHCpan-4.1 and NetMHCIIpan-4.0: Improved predictions of MHC antigen presentation by concurrent motif deconvolution and integration of MS MHC eluted ligand data. Nucleic Acids Res 48(W1):W449–W454. https://doi.org/10.1093/NAR/GKAA379
Andreatta M, Alvarez B, Nielsen M (2017) GibbsCluster: Unsupervised clustering and alignment of peptide sequences. Nucleic Acids Res 45(W1):W458–W463. https://doi.org/10.1093/nar/gkx248

The authors declare no competing interests.

Download PDF

Version 1

posted

You are reading this latest preprint version

High-coverage immunopeptidomics using timsTOF mass spectrometers with Thunder-DDA-PASEF boosted by MS²Rescore

Status:

Version 1

Abstract

Figures

Introduction

Materials

Methods

Notes

Declarations

References

Additional Declarations

Status:

Version 1