Quantitative analysis of global protein lysine methylation by mass spectrometry

Abstract

Since protein activity is often regulated by post-translational modifications, the qualitative and quantitative analysis of modification sites is critical for understanding the regulation of biological pathways that control cell function and phenotype. Methylation constitutes one of the many types of post-translational modifications that target lysine residues. Although lysine methylation is perhaps most commonly associated with histone proteins and the epigenetic regulation of processes involving chromatin, methylation has also been observed as an important regulatory modification on other proteins, which has spurred the development of methods to profile lysine methylation sites more globally. As with many post-translational modifications, tandem mass spectrometry represents an ideal platform for the high-throughput analysis of lysine methylation due to its high sensitivity and resolving power. The following protocol outlines a general method to assay lysine methylation across the proteome using SILAC and quantitative proteomics. First, cells are labeled by SILAC to allow for relative quantitation across different experimental conditions, such as cells with or without ectopic expression of a methyltransferase. Next, cells are lysed and proteins are digested into peptides. Methylated peptides are then enriched by immunoprecipitation with pan-specific antibodies against methylated lysine. Finally, the enriched peptides are analyzed by LC-MS/MS to identify methylated peptides and their modification sites and to compare the relative abundance of methylation events between different conditions. This approach should yield detection of a couple hundred lysine methylation sites, and those showing differential abundance may then be prioritized for further study.

1. Introduction

Methylation is a conserved post-translational modification of proteins, characterized by the enzymatic transfer of a methyl group from S-adenosyl-methionine (SAM) to a lysine or arginine side chain. The side chain nitrogen of lysine is capable of accepting up to three methyl groups, and therefore, may exist in a monomethylated (Kme1), dimethylated (Kme2), or trimethylated state (Kme3) (DesJarlais & Tummino, 2016). Although they do not affect the charge state of lysine like an acetyl group, methyl groups increase the size and hydrophobicity of lysine and limit its potential to participate in hydrogen bonding networks (Lanouette, Mongeon, Figeys, & Couture, 2014; Moore & Gozani, 2014). Methylation of lysine may also create a new binding surface to promote interaction with proteins that possess sequences, such as chromodomains, that recognize methylated lysine (Greer & Shi, 2012). Methylation will also block other types of modifications, such as acetylation and ubiquitination, at a given lysine.

Lysine methylation has been studied extensively in the context of histones and chromatin biology. Histone methylation contributes to the regulation of gene expression, particularly through the recruitment of “reader” proteins (Allis & Jenuwein, 2016). Depending on the site of methylation, histone methylation can represent a repressive mark or an activating mark. For instance, trimethylation of lysine 9 on histone H3 (H3K9me3) is associated with silenced chromatin, whereas trimethylation of lysine 4 on histone H3 (H3K4me3) is associated with active chromatin (Greer & Shi, 2012; Kouzarides, 2007). However, the occurrence of lysine methylation is not restricted to histone proteins. For example, p53 undergoes methylation at several lysine residues with consequent effects on its stability and transcriptional activity (Biggar, 2014; Han et al., 2019; Moore et al., 2013). Methylation has also been characterized on other transcription factors, kinases, splicing components, and translation factors (Biggar, 2014; Carlson & Gozani, 2016; Han et al., 2019).

Lysine methylation is catalyzed by protein lysine methyltransferases (KMTs) and is removed by lysine demethylases (KDM), thus making it a dynamic and reversible modification. The seven-beta-strand family and the SET (Su(var), Enhancer of Zeste, and Trithorax) domain family form the two major groups of KMTs in humans with about 125 members in the former and 50 members in the latter (Petrossian & Clarke, 2010). Except for DOT1L, which methylates histone H3 lysine 79, all KMTs targeting histones contain the catalytic SET domain and thus belong to the SET family (Carlson & Gozani, 2016; Greer & Shi, 2012; Lanouette et al., 2014). In general, histone methyltransferases (HMTs) show selective activity towards particular histone residues, such as the case for EZH2 and H3K27 (Greer & Shi, 2012; Herz, Garruss, & Shilatifard, 2013). However, G9a and SETD7 are notable for their ability to act on a broader range of substrates beyond histones, and some KMTs appear to solely target non-histone substrates (Biggar, 2014; Carlson & Gozani, 2016). Currently, a major area of interest is matching specific KMTs to specific substrates by profiling lysine methylation patterns on arrays or in cells as discussed in more detail below. KDMs balance the activity of KMTs by removing methyl groups from lysine residues through an oxidation reaction that involves either FAD or α-ketoglutarate. LSD1 and LSD2 represent the two FAD-dependent monoamine oxidases that demethylate H3K4me1 and H3K4me2. In contrast, the JmjC dioxygenases are to demethylate all states of lysine methylation (Kme1, Kme2, and Kme3) on histones and non-histone substrates through a mechanism that involves Fe²⁺ and α-ketoglutarate (DesJarlais & Tummino, 2016; Fan, Krautkramer, Feldman, & Denu, 2015; Greer & Shi, 2012).

Given its biological importance and relevance to diseases such as cancer (Biggar, 2014; Greer & Shi, 2012; Han et al., 2019; Mazur et al., 2014; Moore & Gozani, 2014), numerous methods have been developed to identify methylated proteins and their specific modification sites. Generally, they can be classified into profiling lysine methylation either from a selected methyltransferase or across the proteome without a specific enzyme in mind (Carlson & Gozani, 2014; Lanouette et al., 2014). The former approach involves assaying the activity of a purified methyltransferase on peptide or protein arrays in vitro (Levy et al., 2011; Liu et al., 2013; Rathert et al., 2008). Methylation events can be detected through the use of radiolabeled SAM or methylation-specific antibodies. If sequence motifs are identified from substrates on the array, the proteome can then be searched for additional candidate substrates. Instead of using substrate arrays, it is also possible to engineer a methyltransferase to accept a chemically tagged SAM analogue, allowing substrates to be labeled in cells and then purified for identification by mass spectrometry (Islam et al., 2012). Simple pulldowns of methyltransferases with subsequent analysis by mass spectrometry have also been used to identify potential substrates (Sbirkov et al., 2017). However, purification of methylated proteins or detection of methyltransferase activity on a protein array does not guarantee that the specific methylation site will be identified.

The second type of approach involves profiling lysine methylation across the proteome more broadly without consideration of a specific methyltransferase, generally using affinity reagents that recognize methylated lysine independent of the surrounding amino acid sequence followed by mass spectrometry analysis (Carlson & Gozani, 2014; Lanouette et al., 2014). For instance, one group utilized the MBT domain of L3MBTL1, which effectively enriched for roughly 500 proteins modified by Kme1 or Kme2 (Moore et al., 2013). However, since the purification occurred at the protein level rather than peptide level, less than 30 methylation sites were identified. Ideally, mass spectrometry analysis is performed after enrichment of methylated peptides to increase the probability of identifying specific methylation sites. To this end, degenerate peptide libraries carrying a central methylated lysine have been used as immunogens to generate pan-specific antibodies that recognize Kme1, Kme2, or Kme3, independent of surrounding amino acids. Using these antibodies as affinity reagents, up to hundreds of lysine methylation sites were identified from tryptic digests of cell lines (Bremang et al., 2013; Cao & Garcia, 2016; Cao, Arnaudo, & Garcia, 2013; Guo et al., 2013; Olsen et al., 2016; Ong, Mittler, & Mann, 2004). In a slight variation to this approach, another group propionylated unmodified and mono-methylated lysine residues and then used an antibody against propionylated, mono-methylated lysine for enrichment (Wu et al., 2014). Notably, the results of these studies indicate that Kme1 is more abundant than Kme2 or Kme3 and that lysine methylation is less prevalent than arginine methylation, at least as assayed by these antibodies and cell types (Bremang et al., 2013; Cao et al., 2013; Guo et al., 2013). Additional manipulations or isotope labeling schemes can be introduced upstream of harvesting cells for enrichment of methylated peptides. For instance, SILAC (stable isotope labeling by amino acids in cell culture) may be used to compare the abundance of methylated peptides across different experimental conditions, such as after knockdown or inhibition of a methyltransferase, which could reveal candidate substrates of that enzyme on a global scale (Olsen et al., 2016). Heavy methyl SILAC (hmSILAC), in which cells are incubated with labeled methionine to generate labeled SAM, and therefore labeled methyl groups, has also been used to increase confidence in the assignment of methylation sites since the mass shift introduced by a methyl group could also be explained by an amino acid substitution (Ong et al., 2004). However, labeling by hmSILAC is dependent on efficient production of labeled SAM and active turnover of methylation, which may vary for different methylation sites and experimental systems.

Global profiling of lysine methylation sites by mass spectrometry without affinity reagents has also been explored. Because trypsin typically fails to cleave at methylated lysine and arginine residues, modified peptides are expected to have higher charge states than unmodified peptides without missed cleavages. Thus, chromatography methods that fractionate peptides based on charge, such as strong cation exchange (SCX) and HILIC, permit some degree of enrichment of methylated peptides, though this approach generally has been less successful than affinity-based approaches (Wang et al., 2016). Chemical derivatization to neutralize the charge of unmodified lysine and arginine residues prior to SCX offered some improvement (Ning et al., 2016).

The following protocol details a strategy to enrich and identify peptides with methylated lysines using pan-specific antibodies and LC-MS/MS analysis. After labeling cells by SILAC, proteins are extracted and digested with trypsin. Peptides are desalted and then incubated with magnetic beads bearing antibodies against Kme1 and Kme2 to enrich for methylated peptides. Following elution and a final desalting step, peptides are detected by reverse-phase LC-MS/MS and the presence of methylated lysine can be verified by inspection of MS/MS fragmentation spectra.

2. Materials

2.1. Equipment

Incubator for cell culture

Cell culture supplies (e.g. flasks, trypsin, PBS)

Sonicator

Centrifuge

Tubes

Cuvettes

Spectrophotometer

Heating block

Nanodrop

Magnetic rack

Desalting cartridges

Speed-vac

C18 stage tips

LC-MS/MS system (e.g. Dionex UltiMate 3000 in line with Thermo QE-HF)

Software for data analysis (e.g. MaxQuant or ProteomeDiscoverer (Thermo))

2.2. Reagents

Cells

Growth medium (e.g. DMEM with 1x Pen/Strep and 10% FBS)

Proline, dissolve at 110 mg/ml (955 mM) in sterile PBS for 1000x stock

Light L-arginine (Arg0, Sigma A8094), dissolve at 66.3 mg/ml in sterile PBS for 1000x stock

Heavy ¹³C₆,¹⁵N₄ L-arginine HCl (Arg10, Silantes 201604102) dissolve at 84 mg/ml in sterile PBS for 1000x stock

Light L-lysine HCl (Lys0, Sigma L8662), dissolve at 139.9 mg/ml in sterile PBS for 1000x stock

Heavy ¹³C₆,¹⁵N₂ L-lysine HCl (Lys8, Silantes 211604102), dissolve at 146 mg/ml in sterile PBS for 1000x stock

Dialyzed FBS

SILAC base DMEM (DMEM − arginine − lysine + 10% dialyzed FBS + 1x Penicillin/Streptomycin + 110 mg/L proline)

Light SILAC medium (SILAC base DMEM + 139.9 mg/L Lys0 + 66.3 mg/L Arg0)

Heavy SILAC medium (SILAC base DMEM + 146 mg/L Lys8 + 84 mg/L Arg10)

Cell lysis buffer (8M urea, 50 mM NH₄HCO₃ or 50 mM Tris HCl, pH 8), supplement with protease inhibitors immediately prior to use

Protein concentration assay (e.g. Bradford reagents)

Dithiothreitol

Iodoacetamide

50 mM NH₄HCO₃ or 50 mM Tris HCl, pH 8

Trypsin, sequencing-grade

Trifluoroacetic acid (TFA)

Desalting buffers (acetonitrile, 50% acetonitrile / 0.1% TFA, 0.1% TFA)

Phosphate-buffered saline (PBS)

Sodium phosphate, dibasic (100 mg/ml, 373 mM)

Protein A beads

Anti-Kme1 antibodies

Anti-Kme2 antibodies

High salt PBS (PBS with additional 0.8 M NaCl)

C18 stage tip buffers (methanol, 0.1% TFA, 50% acetonitrile / 0.1% formic acid)

LC Solvent A (0.1% formic acid in water)

LC Solvent B (80% acetonitrile in water with 0.1% formic acid)

Analytical C18 columns for nano-LC

3. Procedure

3.1. Growing and labeling cells with SILAC

1. Grow cells of interest under appropriate conditions in preparation for labeling proteins by SILAC. For instance, culture 293T cells in 10 ml of complete DMEM in T-75 flasks at 37°C in a humidified incubator with 5% CO₂. When starting a culture from a frozen stock, we passage cells at least twice before beginning SILAC.

2. Prepare heavy and light SILAC media. Although the mass concentrations of the heavy and light versions of arginine and lysine are different, the final molar concentrations in the media are equivalent (766 μM lysine and 380 μM arginine). Other basal media besides DMEM may be used according to the growth requirements of the cells of interest. However, it is critical that exogenous sources of lysine and arginine are eliminated or minimized as much as possible such that they will not compete for labeling when cells are grown in heavy and light SILAC media. Hence, the use of dialyzed FBS is recommended. Incomplete labeling of cells with Lys8 and Arg10 will complicate the interpretation of results.

3. When ready to start SILAC, harvest cells in conditioned medium in a 50 ml tube. If using adherent cells, use trypsin or an appropriate dissociation method to remove cells from the culture vessel.

4. Resuspend cells in 20 ml of PBS to wash. If desirable to equalize cell numbers across different conditions, determine cell concentration with hemacytometer or cell counter. Equalize cell numbers by discarding appropriate volume of cells.

5. Pellet cells by centrifugation (300 x g, 5 mins) and discard the supernatant.

6. Resuspend cells in 20 ml of PBS and remove half to another 50 ml tube so that there are two equivalent aliquots of cells. Generally, it is advisable to perform reciprocal SILAC labeling in parallel to increase confidence in the proteomics data. For instance, if two experimental conditions (control versus treated, or vector containing gene of interest versus empty vector) will be compared, prepare heavy and light cultures of cells for each condition. In this way, a differentially abundant peptide should have a large heavy versus light ratio when comparing cells with one labeling scheme and a small heavy versus light ratio when comparing cells with the reciprocal labeling scheme.

7. Pellet cells by centrifugation as above.

8. Discard the supernatant. Resuspend one cell pellet in 10 ml of light SILAC medium and the other pellet in 10 ml of heavy SILAC medium.

9. Transfer cells to T-75 flasks and continue growing cells in 37°C humidified incubator with 5% CO₂ until they reach 70-80% confluence. To conserve media, we generally label cells on a smaller scale in T-75 flasks for at least two passages to generate a population of largely labeled cells, and then we expand the labeled cells as necessary for the final passages.

10. Quantitative incorporation of the SILAC labels is expected to require several cell doublings and should be verified before terminating the culture and harvesting the final cell pellet. After expanding 293T cells beyond the initial T-75 cultures, we have achieved satisfactory labeling after 2 additional passages (~7 days). To test if SILAC labeling is complete, remove a small aliquot (2-4 million cells) when passaging cells. While the passaged cells continue to grow in culture over the next several days, perform cell lysis, protein digestion, and peptide desalting as below, expect for using smaller capacity C18 stage tips. Analyze the peptide digest by LC-MS/MS. Cells cultured in the heavy SILAC condition should show complete conversion of peptides to the heavy form (+8 Da or +10 Da depending on their lysine or arginine content). If unlabeled peptides are still detected, harvest another aliquot of cells at the next passage and repeat the analysis while continuing to grow the cells in SILAC medium.

3.2. Cell lysis and protein digestion

1. Lyse cells in cell lysis buffer for 10 mins on ice. Use a volume of 5-10x the packed cell pellet volume. Alternatively, cells may be fractionated into subcellular compartments.

2. Sonicate the cell lysate to reduce viscosity from released DNA. For our probe-tip sonicator, we use 5 cycles of 15 sec with 15 sec of rest in between at power setting 2, which yields a power output of 6-7 W. Keep the lysates on ice during sonication and sonicate as needed until the lysate is no longer viscous.

3. Pellet insoluble debris by centrifugation in a microcentrifuge for 10 mins at full speed (e.g. 26591 x g) at 4°C.

4. Remove the supernatant, which represents the protein extract, and measure the protein concentration by Bradford assay.

5. Prepare a standard amount of lysate in a standard amount of cell lysis buffer (e.g. 1 mg in 1 ml). Mix lysates from heavy and light SILAC conditions at 1:1 ratio by mass. We recommend starting with at least 1 mg of input material in total.

6. Reduce disulfide bonds by adding dithiothreitol to the lysate at a final concentration of 10 mM and incubating for 30 mins at 56°C.

7. After cooling the lysate to room temperature, add iodoacetamide to the lysate to a final concentration of 50 mM. Incubate at room temperature in the dark for 40 mins to alkylate reduced cysteines.

8. Dilute 1 volume of lysate with 4 volumes of 50 mM NH₄HCO₃ or 50 mM Tris HCl, pH 8 to reduce the urea concentration (e.g. 4 ml of diluent to 1 ml of lysate).

9. Digest the lysate by adding trypsin and incubating overnight at 37°C with rotation. We typically use a trypsin/substrate ratio of 1/200 – 1/500 by mass. Alternative proteases may also be used to obtain complementary sequence coverage. Check the manufacturer’s recommendations and adjust the protocol accordingly if additional components are necessary for the digestion (e.g. CaCl₂) or if interfering substances require more dilution.

10. Acidify the digest by adding trifluoroacetic acid to 1%. The pH should be 2-3.

11. Remove insoluble debris by centrifuging the digest for 5 mins at maximum speed (e.g. 3400 x g for a swinging bucket centrifuge).

12. Desalt the digest. For larger scale preparations, we use Oasis HLB SepPak cartridges. The binding capacity is expected to be roughly 5-10% of the resin mass. The following volumes are for a 30 mg cartridge with a reservoir of 1 cc. A centrifuge, vacuum manifold, or syringe (without a needle) can be used to drive liquid through the column. Place the column in a tube or a beaker to collect the waste.

    1. Activate the resin with 1 ml of acetonitrile.
    2. Wash the resin with 1 ml of 50% acetonitrile/0.1% TFA.
    3. Equilibrate the resin with three washes of 1 ml of 0.1% TFA.
    4. Load the acidified digest.
    5. Wash the resin three times with 1 ml of 0.1% TFA.
    6. Place the column over a 1.5 ml tube to collect the desalted peptides.
    7. Elute the peptides with 1 ml of 50% acetonitrile / 0.1% TFA.
    8. Dry the peptides in a speed-vac.

3.3. Immunoprecipitation of methyl-lysine peptides

1. Resuspend the dried peptide digest in phosphate-buffered saline (PBS) at 0.5-1 mg/ml. Expect 50% recovery after trypsin digestion and desalting. Add concentrated sodium phosphate dibasic (e.g. 100 mg/ml Na₂HPO₄) to neutralize pH to 7-8 if necessary. We typically add Na₂HPO₄ to 0.1M final concentration. Alternatively, peptides may be resuspended in an appropriate buffer for offline fractionation by strong cation exchange prior to immunoprecipitation.

2. Pellet insoluble debris by centrifugation for 5 mins at 17,000 x g.

3. Remove supernatant and measure absorbance of 1 μl on a NanoDrop instrument at 280 nm to approximate peptide concentration (1 abs unit ≈ 1 mg/ml).

4. Prepare peptide solution for immunoprecipitation that contains a standard amount of peptide in a standard volume of PBS. Typically, we aim for a minimum of 500-1000 μg of peptides per immunoprecipitation at a concentration in the range of 0.5-1 mg/ml.

5. Remove 10-20 μg of peptides for an input sample. Store at −20°C overnight.

6. Prepare antibody mixture containing both anti-mono-methyl-lysine (Kme1) and anti-di-methyl-lysine (Kme2). For each mg of peptide digest, we add 50 μg of each anti-Kme1 and anti-Kme2 in a final volume of 1 ml of PBS. This ratio should ensure that the antibody is in far excess of antigen, but it can be further optimized if desired. We use customized anti-Kme1 and anti-Kme2 produced by ProteinTech using a synthetic library of mono-methylated or di-methylated peptides. The library consists of peptides following the sequence pattern CX₆KX₆, where X represents any amino acid except cysteine; and the central lysine is unmodified, mono-methylated, or di-methylated. We further purified polyclonal antibodies from the anti-serum by affinity chromatography with the peptide libraries covalently coupled to cysteine-reactive resin.

7. Prepare protein A beads. For each immunoprecipitation, we use 70 μl of slurry per 112 μg of antibody. We generally prepare a master batch of antibody-charged beads that is later divided into aliquots for each immunoprecipitation sample. It is advisable to prepare slightly more beads than necessary to account for pipetting errors (e.g. if there are 2 samples, prepare enough beads for 2.5-3 samples). Prior to removing the beads, ensure that they are thoroughly resuspended. First, wash the protein A slurry three times with 1 ml PBS. To do so, add 1 ml of PBS to the beads, resuspend well by pipetting, place the tube on a magnetic rack for 30 seconds (or until the beads separate from solution), and then discard supernatant. Remove the tube from the magnetic rack and repeat twice more. Proceed immediately to the next step to prevent the beads from drying out.

8. Resuspend the washed protein A beads in 1 ml of antibody mixture from the prior step.

9. Incubate the antibodies and protein A beads on a rotator for 1 hr at room temperature or for several hours (e.g. 4 h) at 4°C.

10. Separate protein A beads from solution by placing the tube on a magnetic rack for 30 seconds.

11. Wash the beads with 3 x 1 ml PBS as in the previous step to remove unbound antibody.

12. Resuspend the beads in a small volume of PBS (e.g. 100 μl per each immunoprecipitation).

13. Aliquot an equal volume of resuspended beads to each peptide digest for immunoprecipitation.

14. Incubate the charged beads with the peptide digests overnight at 4°C on a rotator.

15. Place the tubes on a magnet and separate the beads from the solution. Remove the supernatant containing unbound peptides and save for analysis if desired.

16. Wash the beads sequentially with 1 x 1 ml with PBS, 1 x 1 ml with high salt PBS, 1 x 1 ml with PBS, and 1 x 1 ml with water.

17. Elute the captured peptides by resuspending the beads in 50 μl of 1% trifluoroacetic acid. Incubate for 5 mins at room temperature. The pH should be less than 3.

18. Remove the beads by placing the tubes on a magnet. Collect the supernatant containing the eluted peptides.

19. Desalt the input and eluted peptides with C18 stage tips.

    1. Trim off roughly 5 mm from the tapered end of a P1000 pipette tip with a clean scalpel
    2. Punch out a piece of C18 material with the trimmed pipette tip
    3. Use capillary tubing to push the C18 material into a P200 pipette tip to create a stage tip
    4. Place the stage tip into an adaptor fitted onto a 1.5 ml or 2 ml collection tube
    5. Activate the stage tip with 30 μl of methanol and spin in a microfuge at 1000 x g for 1 min. For this and all subsequent steps, ensure that all the liquid passes through the resin before proceeding to the next step. If necessary, increase the centrifugation time or speed.
    6. Equilibrate the tip with 30 μl of 0.1% TFA and centrifuge as above. Repeat for a total of two equilibration steps.
    7. Acidify the input sample by adding TFA to 1% final concentration. The enriched sample already contains 1% TFA. Pipette a small amount of sample (< 1 μl) onto pH paper to check that the pH is 3 or lower. If necessary, add more TFA.
    8. Load the sample into the stage tip and centrifuge as above.
    9. Wash the stage tip with 30 μl of 0.1% TFA. Centrifuge as above. Repeat for a total of two washes.
    10. Transfer the stage tip with the adaptor to a clean 1.5 ml tube. Add 30 μl of 50% acetonitrile / 0.1% formic acid to elute the desalted peptides.

20. Dry down the samples in a speed-vac prior to analysis by LC-MS/MS.

3.4. Analysis by LC-MS/MS

1. Desalted peptides should be resuspended in 0.1% formic acid for analysis by LC-MS/MS.

2. Peptides can be separated by reverse-phase separation using a C18 column and gradient with Solvent A as 0.1% formic acid and Solvent B as 0.1% formic acid in 80% acetonitrile. We make analytical columns of 10-20 cm by packing fused silica (Polymicro Tech, 75 μm i.d.) with C18 material (ReproSil-Pur 120 C18-AQ, 3 μm, from Dr. Maisch GmbH). Separation from 2% solvent A to 50% solvent B should occur over a minimum of 60 minutes. It is recommended that the reserved peptide input should be analyzed over a 120 minute gradient because of its greater complexity.

3. Inject approximately 1 μg of peptide material from the input sample and roughly a quarter or half of the material recovered from the enrichment.

4. The following represents an example mass spectrometric method developed for a Thermo Q-Exactive-HF mass spectrometer. All MS instrument methods should be optimized for specific columns, gradients and samples.

   1. Full MS1 scan from 300-1500 m/z with a minimum of 120,000 resolution at 200 m/z. The AGC target should be 1x10⁶.
   2. High-resolution MS/MS spectra should be acquired with a 7,500 or 15,000 resolution at 200 m/z and an AGC target of 1x10⁵ charges with a maximum AGC fill time of 200 ms. HCD fragmentation energy can be set to NCE 28. Dynamic exclusion is optimized for typical peak widths depending on the column and the chromatography, with an n = 1 selection and exclusion for the duration of the peak. Charge state +2 to +6 as a minimum should be selected for fragmentation with a cycle of n = 15.

3.5. Data analysis

1. RAW files should be searched against an appropriate protein database. Precursor mass tolerance should be set at 10 ppm or less. Product ion mass tolerance should be set at 20 ppm for high resolution MS/MS. Peptides should be searched with a minimum of 2 missed cleavages to account for impaired trypsin activity at methylated lysines. Usage of alternative proteases can potentially increase sequence coverage to identify additional modification sites or verify sites identified from tryptic digests. If charge states of z = +6 were selected for fragmentation, then up to 4 missed cleavages could be allowed in the search parameters. Fixed modifications should include carboamidomethylation at cysteine. Variable modifications should include mono-, di-, and tri-methylation at lysine. Additional options for variable modifications include acetylation at lysine, oxidation at methionine, and mono- and di-methylation at arginine. High resolution MS/MS is able to distinguish between acetylated lysine and tri-methylated lysine as the difference in mass is greater than 10 ppm. Furthermore, as peptides are enriched, it is appropriate to consider proteins identified by a single methylated peptide. If using SILAC, heavy isotope labels at arginine and lysine should also be included as variable modifications, and SILAC should be indicated as the method of quantitation.

2. Peptide spectral matches (PSM) identified to contain a post-translational modification are at risk of a higher false discovery rate (FDR) than the applied FDR cutoff (Hart-Smith, Yagoub, Tay, Pickford, & Wilkins, 2015). PSMs are prone to incorrect placement of the modification, and these instances should be perceived as false positives. N-terminal methylation of peptides can be included as a variable modification as a way to highlight the false discovery rate of modified peptides. A peptide (rather than protein n-terminus) identified with an n-terminal methylation will quickly highlight poor spectral quality because of insufficient ion coverage to appropriately localize the methylation site.

3. For more stringent evaluations, several criteria may be evaluated in addition to the basic PSM statistical analysis:

   1. Greater confidence PSM assignment should be given to PSM with clearer isolation windows from the MS¹.
   2. Several search algorithms report a delta mod score, referring to the score difference between the identified modification within the sequence and the next scoring modification. PSMs with low delta mod scores have lower confidence without further evaluation than PSMs with larger delta mod scores.
   3. Peptides can be evaluated by sequence for confidence on the placement of the modification. Endopeptidase trypsin preferentially cleaves at unmodified lysines and arginines. Theoretically, the modification would be expected to occur on the internal lysine residue, giving rise to a “missed” cleavage, rather than the C-terminal lysine or arginine. PSMs with C-terminal methylated lysines should be verified by inspection of the MS/MS data to support the site assignment.

4. PSMs can be manually verified to ensure a confident sequence assignment and PTM site localization. Most data processing methods will provide both theoretical fragment masses as well as the annotated spectrum. The most intense peaks in the spectrum should be assigned as fragment masses, and the PSM should have sufficient sequence coverage to unambiguously place the modification. Several example spectra are discussed below, highlighting both the strengths and risks of each PSM.

5. Labeling cells by SILAC facilitates the quantitative analysis of lysine methylation across two conditions. If SILAC is specified as the quantitation method, MaxQuant and ProteomeDiscoverer will report the heavy to light ratios (H/L) for peptides and proteins. A histogram of the log2-transformed H/L ratios should reveal a symmetric distribution with a center near zero, indicating equivalent mixing of the heavy and light isotopically-labeled lysates prior to trypsin digestion. Identification of methylated peptides showing differential abundances across conditions may be based on a specified threshold value (e.g. log2(H/L) greater than 1 or less than −1) (Olsen et al., 2016). Alternatively, a more objective approach is to consider the spread of the distribution and define peptides as differentially abundant when the log2(H/L) falls outside a specified number of standard deviations (e.g. two) from the mean (Pagala et al., 2015). Generating a complementary SILAC data set using a reciprocal labeling scheme, in which the heavy and light labels are switched between conditions, can be especially useful in selecting true positives. In this case, the log2(H/L) ratios of the two data sets can be plotted against each other, and differentially abundant peptides should be easily visible as points with a high log2(H/L) along one axis and a low log2(H/L) on the other. Inclusion of replicates to allow for the calculation of inferential statistics (e.g. Student’s t test) can also be useful for detecting peptides with robust and reproducible changes in abundance across two conditions.

6. In some cases, a perceived change in PTM abundance may manifest indirectly from a change in the abundance of the protein on which the PTM occurs rather than a direct change in PTM occupancy. Thus, analyzing the input material prior to enrichment is critical for detecting changes in protein abundances. To normalize for any changes in protein abundance, the change in PTM abundance across two conditions should be divided by the corresponding change in the protein abundance.

4. Expected results

Using this enrichment protocol, we can detect roughly 150 mono-methyl peptides and 50 di-methyl peptides, which is in line with previous studies (Bremang et al., 2013; Guo et al., 2013). The advantage of utilizing an enrichment method is in the simplification of the mass spectral data. The variety of methylated proteins are highlighted in Figure 1, which shows a western blot of a cell lysate probed with antibodies against Kme1 and Kme2. The complexity of the tryptic digest prior to enrichment (input) versus the enriched digest can be observed in Figure 2. In contrast to the input TIC in Fig. 2F, which shows a multitude of overlapping peaks, the enriched TIC in Fig. 2A shows visibly distinct peaks. Further evaluation of extracted precursor peaks representing specific peptides highlights the degree of the enrichment. For instance, the peak displayed in Fig. 2G is clearly amplified after enrichment (Fig. 2B) and is even visible in the enriched TIC (Fig. 2A). A second example is depicted in Fig. 2C and 2H. While the signal is not increased after the enrichment, the isolation of the precursor has fewer contaminating ions in the enriched sample versus the input sample (compare Fig. 2D to Fig. 2I).

Figure 1.

Cytosolic (C) and nuclear (N) extracts were prepared from 293T cells, resolved by PAGE, and transferred to a nitrocellulose membrane. The membrane was divided into 3 sections and then blotted separately with antibodies against Kme1 or Kme2. Normal rabbit IgG served as a negative control. GAPDH and histone H4 served as loading controls for the cytosolic and nuclear fractions, respectively.

Figure 2.

LC-MS/MS data from a typical enrichment are displayed in A-E with data from the corresponding input displayed in F-J. Total ion chromatograms for the MS1 results are plotted for the enrichment (A) and the input (F). Extracted ion chromatograms are plotted for two peptides after enrichment in B and C or before enrichment in G and H (input). The extracted ion chromatogram and the related isolation window for 671.3729 m/z, z=2+ (GSFK(me1)YAWVLDK from Eukaryotic Translation Elongation Factor 1 Alpha) are presented in B and E for the enrichment, respectively, as well as G and J for the input, respectively. The extracted ion chromatogram and the related isolation window for 550.2890 m/z, z=3+ (SK(me1)QFYNQTYGSRK(me2) from OTU Domain-containing Protein 4) are presented in C and D for the enrichment, respectively, as well as H and I for the input, respectively.

Data can be searched with software that is designed to handle SILAC workflows, such as Proteome Discoverer ver. 2.2 or higher (Thermo), or MaxQuant (Cox & Mann, 2008; Tyanova, Mann, & Cox, 2014). If Proteome Discoverer is used, the peptide spectral matches will be identified by a search algorithm such as Mascot, Sequest, or Byonic, and the quantitation will be performed by Proteome Discoverer. If MaxQuant is utilized, the search algorithm is Andromeda and the quantitation is performed by MaxQuant.

Performing a database search to focus on post-translational modifications often requires a more thorough analysis of results than a standard proteomics workflow (Hart-Smith et al., 2015). In addition to matching the correct peptide sequence to the correct peptide spectrum, the modification needs to be assigned to the correct position in the peptide sequence. For the former, false discovery rates and peptide statistics have been characterized for large proteomic data sets (Hart-Smith et al., 2015). However, the responsibility of correctly annotating PTM sites largely falls on the analyst. Lysine methylation adds a layer of complexity to the analysis as the +14.01565 Da mass shift can be accounted for by other sequence variations or modifications. For instance, the mass shifts between glycine and alanine, serine and threonine, and threonine and aspartic acid are all 14 Da. Furthermore, the nominal masses of arginine and di-methylated lysine as well as tri-methylated lysine and acetylated lysine are similar at 156 Da and 170 Da, respectively. The use of high resolution MS¹ and MS/MS can largely distinguish between these similar but distinct nominal masses.

Modified peptides identified from database search results require an additional layer of review by the analyst. Three example PSMs are presented in Figs. 3-5, representing high, medium, and low confidence assignments. The precursor isolation window and the fragment ion coverage are easily reviewed to determine if the sequence assignment is consistent with the fragmentation data. The precursor isolation window should be viewed from the MS¹ scan immediately prior to the MS/MS scan. The charge state of the precursor ion should be verified as well as quality of the precursor isolation. A ‘clean’ isolation contains only the precursor ions of interest with no isolation interference. Compared to the ideal situation shown in Fig. 3B, Fig. 5C contains significantly more isolation interference within the isolation window. Both the selected precursor and a second unrelated precursor fall within the isolation window. The second precursor contributes its own fragment ions to the MS/MS spectrum (Fig. 5A), resulting in a mixed spectrum. In addition to the isolation window, the fragment ion coverage should also be examined. First, the most abundant fragment ions in the MS/MS should be assigned and/or justified, such as b/y ions for HCD spectra or unfragmented precursor, respectively. The ¹²C ions should be assigned, as in Fig. 3A. Additionally, there should be a sufficient number of fragments to confidently localize the modification to the correct residue, ideally with several fragment ion pairs, as in Fig. 3A. The assignment of fragment ions should be manually reviewed across spectra of high, medium, and low quality to establish a rough threshold at which PSMs and methylation assignments become more reliable.

Figure 3.

Fragment coverage map for a high confidence, unambiguous assignment of a methylated lysine peptide (GSFk(me1)YAWVLDK from Eukaryotic Translation Elongation Factor 1 Alpha). The lowercase letter indicates the methylation site assignment. MS/MS data is shown in A with y ions depicted in red and b ions depicted in blue. Residual, unreacted precursor is shown in pink. The most abundant fragment ions can be identified in the spectrum. The isolation window is shown in B with the window itself highlighted in blue. The isolation window is clear except for the precursor of interest.

Figure 5.

Fragment coverage map for a low confidence assignment of a methylated lysine peptide (MDSTEPPYSQk(me2)R from Eukaryotic Translation Elongation Factor 1 Alpha). MS/MS data is shown in A with y ions depicted in red and b ions depicted in blue. The lowercase letter indicates the methylation site assignment. The fragment ion coverage is displayed with weak coverage for lower mass y-ions. Closer examination of the expanded spectrum in B clearly shows the y₁ ion peak is not the ¹²C peak. Additionally, a potential y₁ ion at 189.12 indicates the possibility of a monomethylated arginine residue instead of a methylated lysine. The isolation window in C is highlighted with blue, showing a z=2+ species co-isolated with the z=3+ precursor of interest.

Figs. 4 and 5 highlight questionable or incorrect assignment of fragment ions. Fig. 4 highlights an example where the modification is placed on the C-terminal lysine, rather than the internal lysine representing a missed cleavage. In Fig. 4B, fragments with questionable assignments are highlighted with arrows. Several high intensity fragment ions without assignments are detected in Fig. 4A and B, even though the precursor ion of interest is the major species in the isolation window (Fig. 4C). Despite the presence of many unassigned fragment ions, there is still a strong y-ion series to support the assignment of the peptide sequence and PTM site. Fig. 5 shows an example of a methylation event localized to the lysine of a peptide terminating in KR, which is questionable since methylation may occur on both lysine and arginine. The placement of the methylation on the lysine instead of the arginine residue is likely due to inclusion of lysine methylation but not arginine methylation as a variable modification in the search parameters. Additionally, a fragment ion is identified at 189.18 in Fig. 5B, indicating a potential methylated arginine. Since the isolation window in Fig. 5C contains significant interference, several of the most abundant fragment ions in Fig. 5A are unassigned, and the assignment of the modification site relies on a single, questionable fragment ion. Thus, there is little confidence in the assignment of this PSM. Further examination of additional MS/MS data from the same precursor or from a different charge state of the precursor is needed to support the assignment.

Figure 4.

Fragment coverage map for a medium confidence assignment of a methylated lysine peptide (QIAAKQYk(me1) from ADP/ATP Translocase). MS/MS data is shown in A with y ions depicted in red and b ions depicted in blue. The lowercase letter indicates the methylation site assignment. The fragment ion coverage is displayed with all fragment ions as identified. However, closer inspection of the individual fragment ion assignments (magnified region shown in B) results in multiple retracted assignments, specifically all y-ions in B. Not all of the most abundant ions are identified as a component of the peptide. Remaining unreacted precursor is shown in pink. The isolation window is shown in C with the window itself highlighted in blue. The isolation window is not a ‘clean’ isolation.

5. Summary

Like other PTMs, lysine methylation is important for the regulation of biological pathways, particularly in the context of histones and epigenetics. However, other proteins besides histones are subject to methylation, and thus, there is growing interest in mapping lysine methylation sites more broadly in cells as well as comparing methylation patterns across different experimental conditions. To this end, one approach consists of enriching methylated peptides by immunoprecipitation for analysis by LC-MS/MS. SILAC performed upstream of the enrichment provides for easy and precise relative quantitation of methylated peptides across two experimental conditions, facilitating the identification of methylation sites undergoing dynamic regulation and which may be of further interest.