Synthetic phosphorylation of p38α recapitulates protein kinase activity.

Through a "tag-and-modify" protein chemical modification strategy, we site-selectively phosphorylated the activation loop of protein kinase p38α. Phosphorylation at natural (180) and unnatural (172) sites created two pure phospho-forms. p38α bearing only a single phosphocysteine (pCys) as a mimic of pThr at 180 was sufficient to switch the kinase to an active state, capable of processing natural protein substrate ATF2; 172 site phosphorylation did not. In this way, we chemically recapitulated triggering of a relevant segment of the MAPK-signaling pathway in vitro. This allowed detailed kinetic analysis of global and stoichiometric phosphorylation events catalyzed by p38α and revealed that site 180 is a sufficient activator alone and engenders dominant mono-phosphorylation activity. Moreover, a survey of kinase inhibition using inhibitors with different (Type I/II) modes (including therapeutically relevant) revealed unambiguously that Type II inhibitors inhibit phosphorylated p38α and allowed discovery of a predictive kinetic analysis based on cooperativity to distinguish Type I vs II.

(b) The region around Thr180 has net negative charge (red) while the region around Ala172 is positive (blue).

Synthetic Chemistry
Melting points were determined under a light microscope at low power and are uncorrected. Infrared spectra were recorded using a Bruker TENSOR 27 spectrometer. Nuclear magnetic resonance (NMR) spectra were recorded using a Bruker AV-400 spectrometer, proton NMR ( 1 H NMR) at 400 MHz and carbon NMR ( 13 C NMR) at 101 MHz Chemical shifts (δ) are reported in parts per million (ppm) and are referenced relative to the residual proton-containing solvent ( 1 H NMR: 2.52 ppm for DMSO; 13 C NMR: 41.2 ppm for DMSO). Accurate mass spectra (ESI) were recorded using a Micromass Bruker MicroTOF instrument. TLC was performed using aluminium plates pre-coated with Merck Kieselgel 60 F254. The plates were visualised using either ultraviolet (UV) light or potassium permanganate staining as appropriate.

Media and Bacterial Strains
All bacterial handling was done in a sterile environment, either within close proximity to a Bunsen burner flame or inside a biological safety cabinet. Samples of p38α mutant plasmid construct p38α-C119S/C162S/A172C was kindly sent by Dr. Richard Bazin and co-workers, Pfizer, Sandwich, Kent. LB was bought as pre-formulated dry granules (Melford) and diluted according to the manufacturer's specifications. SOC and NZY + media were prepared from biological laboratory reagents. 2YT medium was bought as pre-formulated dry granules (Sigma Aldrich) and diluted according to the manufacturer's specifications. All other biological laboratory buffer and media reagents were bought from common S5 suppliers and used as purchased. BL21(DE3) and NovaBlue E. coli chemically competent cells (Novagen) were handled according to the manufacturer's instructions, thawing on ice before incubation with DNA. NovaBlue was used for cloning, while BL21(DE3) was used for protein expression. XL1-Blue E. coli chemically competent cells (Agilent) likewise were handled according to the manufacturer's instructions. XL1-Blue was used for cloning.

Mutagenesis and Bioinformatics
Site-directed mutagenesis was done using PfuUltra DNA polymerase AD (Agilent), dNTPs (Sigma Aldrich) and DpnI (NEB). Primers for mutagenesis were designed using the MutaPrimer design function in SimVector 4 from Premier Biosoft, which follows the guidelines given by Stratagene. The sequences were then sent to Life Technologies (Invitrogen) for custom oligonucleotide synthesis. Plasmid purification was done by using the QIAprep R Miniprep Kit (Qiagen) according to the manufacturer's recommendations. LyseBlue reagent had been added to buffer P1 from the Miniprep kit to aid identification of cell lysis. DNA and protein concentration determination was done using a NanoDrop 1000 spectrophotometer (Thermo Scientific), which allowed the calculation of yield. The wavelengths used were a ratio of 260/230 nm and 280-260 nm respectively. The extinction coefficients for proteins were calculated using ProtParam tool (ExPASy). 3,4 Other bioinformatics tools from the ExPASy website for various applications were also used.

Protein Expression and Purification of p38α Variants
Fast protein liquid chromatography (FPLC) was done using aÄKTA Purifier TM system (Amersham Biosciences, now part of GE Healthcare). Pre-packed columns for FPLC and G-25 desalting columns (SpinTrap TM , MiniTrap TM and MidiTrap TM ) were purchased from GE Healthcare. Unless otherwise stated, the analytes were filtered before loading onto the column for each chromatographic step. Dialysis was performed using Slide-A-lyser R MWCO 10 000 dialysis cassettes (Thermo Scientific). Concentration of protein was performed using Vivaspin 500, 15 or 6, MWCO 10 000 spin columns (Sartorius Stedim Biotech) as appropriate. Autoclaving was performed at 121 • C for 10-15 min. Solutions which were used as media or buffers and which could not be autoclaved were filtered through either 0.2 µm cellulose acetate filters, 0.2 µm polyamide filters, or 0.2 µm Minisart R filter cartridges (Sartorius Stedim Biotech), as appropriate. In particular, all filtrations of buffers, solutions and lysates involved in protein purification or chemical reactions involving proteins were done through 0.2 µm filters. Filtration of volumes of <2 mL (for HPLC) was done using 0.2 µm spin filters (Corning). HPLC samples were otherwise clarified by centrifugation (13 200  Solid LB agar medium was prepared by suspending LB granules (6.25 g) and agar (3.75 g) in MilliQ water (250 mL). The suspension was autoclaved as previously described to give a solution which once sufficiently cooled, was supplemented with the appropriate antibiotic (ampicillin: final concentration: 100 µg/mL) and poured into petri dishes (10-20 mL per plate).
Glycerol Stocks: For cloning strains (XL1-Blue), glycerol stocks were made by adding bacterial broth (750 µL) to glycerol solution (250 µL of a 60% stock solution in water) and stored at −80 • C. For expression strains (BL21(DE3)), glucose (28 µL of a 2 M stock solution in water) was added as a supplement before adding the bacterial broth (722 µL) and stored as above.

Gene Sequence Analysis
DNA sequencing was done by Geneservice, Oxford using the primers pGEX forward (5 -GGGCTGGCA AGCCACGTTTGGTG-3 ) and pGEX reverse (5 -CCGGGAGCTGCATGTGTCAGAGG-3 ) for genes contained in pGEX vectors . The forward and reverse complimentary sequences returned from the gene sequencing service were analysed by sequence alignment of the two returned sequences, followed by string comparison to the desired reference sequence. Sequence data manipulation was done using scripts written in MATLAB. In the case of serious mismatching between observed and desired sequences, the raw data were visualised using FinchTV version 1.4.0.

Protein Modification Reactions
Sodium thiophosphate was bought from Sigma Aldrich and used as purchased. Shaking incubation of the reactions was performed using PHMT-PSC-18 from Grant Bio. Incubation without shaking was done in a waterbath at the appropriate temperature. LC-MS analysis was done on a Micromass LCT Premier instrument using either a ProSwift R RP-4H monolith, 1×50 mm column (Dionex) or a Chromolith R FastGradient RP-18 endcapped 50×2 mm monolithic HPLC column (Merck). A linear gradient was run from 5-95% of solvent mixture B (1% formic acid in acetonitrile) into solvent mixture A (1% formic acid in water). The gradient was run with a flow rate of 0.4 mL/min over 4 min for the ProSwift column and with a flow rate of 0.3 mL/min over 6 min for the Chromolith column. Modified proteins used in subsequent reactions had their yield determined by volume measurement using a Gilson-type pipette, and concentration measurement by UV absorbance as described previously, making the assumption that the functional group incorporated in the chemical reaction did not have a large effect on the extinction coefficient as predicted for the protein's unmodified sequence. In determining the protein concentration in reaction mixtures during reaction monitoring, the samples of the reaction mixtures were first filtered and compared against a blank "reaction" (the reaction buffer with the other reagents, but no protein), which was also filtered. Analysis of the MS data collected was done using MassLynx version 4.1. Deconvolution of the protein ion series was done using the in-built maximum entropy algorithm with calibration of the MS done with equine myoglobin from heart.

Enzymatic Reactions and Activity Assay
Sequencing grade modified trypsin (Promega), ATF2 (Sigma Aldrich), ATP (Sigma Aldrich), active and unactive p38α (Merck Millipore) and [γ-32 P]-ATP (Perkin Elmer) were all used as per manufacturer's instructions. More specifically, active p38α purchased had been obtained by initial NiNTA purification of the non-phosphorylated p38α. The kinase was activated using MKK6-EE (constitutively active mutant) before subsequent re-purification by NiNTA agarose. 5  were subsequently added and the mixture returned to reflux at 80 • C for 2 h. As the reaction progressed, the reaction mixture first turned red and red vapour was observed and after 1 h, then transiently lost its colour before turning black. The mixture was then cooled to 0 • C and allowed to stand for 30 min before being filtered and washed with Et 2 O (40 mL). The filtrate concentrated in vacuo to give 2,5dibromoadipoyl dichloride (1) as a dark red oil. The product was not further purified or characterised and was directly subjected to the next reaction. DNA Purification and Storage: After glycerol stocks were made, the remaining cells were then pelleted by centrifugation (3000 rpm, 20 min, 4 • C). The resulting cell pellet was resuspended in P1 buffer and the plasmid DNA purified out using the QIAprep R Miniprep kit. The plasmid DNA was eluted in EB buffer and stored at −20 • C. DNA sequencing confirmed that the sequences of the plasmid samples received were as expected.

Preparation
Sequence of pGEX-2T-p38α-Cys119S/C162S/A172C: Mutagenesis Reaction: The mutagenesis reactions were made up by adding the following components in the given order: After this time, DpnI (0.25 µL of a 10 U/µL solution) was added and the mixture incubated for 1-2 h at 37 • C. Extent of DNA synthesis was determined by running samples (2.5 µL) of the reaction mixtures on a 0.8 % agarose gel supplemented with ethidium bromide (5.3 µL of a 1 % stock solution) in TAE buffer (50 mL) with the gel run at 150 V in TAE buffer. The samples were loaded with 6× DNA loading dye (0.5 µL). The gel was visualised under a UV lamp and only those reactions were plasmid DNA was observed were transformed into E. coli. Where there were 2 reactions containing the same template/primer combination, only the reaction with lower template concentration was subsequently used.
Transformation: The selected reaction mixtures were transformed into XL1-Blue E. coli supercompetent cells by heat shock. Reaction mixture (1 µL) was added to aliquots of freshly thawed cells (25 µL) and incubated on ice for 5 min. Heat shock was then performed at 42 • C for 40 s before returning the cells onto ice and incubating for a further 2 min. The cells were fed with NZY + broth (250 µL) and incubated in a shaker incubator at 37 • C, 250 rpm for 1 h.

Bacterial Culturing:
The cells were plated onto LB agar supplemented with ampicillin (100 µg/mL). The plates were incubated at 37 • C for 21-24 h. 3 colonies of each construct were then selected from the plates and cultured further in LB medium (3×10 mL), again supplemented with ampicillin (100 µg/mL) and the cultures incubated for a further 13-22 h in a shaker incubator at 37 • C, 250 rpm.
DNA Purification and Storage: Glycerol stocks of the cultures were made before cells were harvested by centrifugation (3750 rpm, 10 min, 4 • C) of the cultures and the supernatant decanted off. The cells were resuspended in P1 buffer for plasmid purification by Miniprep, where the manufacturer's protocol was used. DNA sequencing confirmed the colonies containing the desired constructs, which were selected for either expression or further rounds of mutagenesis.

General Procedure for Protein Expression of p38α Mutants -Cys172 and -Cys180
Transformation: Plasmid DNA of pGEX-2T-p38α-Cys172 and -p38α-Cys180 (1 µL each) were transformed into BL21(DE3) E. coli chemical competent cells (20-25 µL) by heat shock. The heat shock conditions used were the same as that described for cloning. After feeding with SOC medium (100-125 µL), the cells were incubated in a shaker incubator at 37 • C, 250 rpm for 1 h.
Protein Expression: After glycerol stocks of the cell starter cultures were made, one of the cultures (2×1-4 mL) was used a inoculate a larger culture in 2YT medium (2×800 mL) supplemented with ampicillin (100 µg/mL). These larger cultures were grown to OD 600 = 0.4 at 37 • C, 180 rpm. The temperature was reduced to 20 • C and the cultures grown further to OD 600 = 0.65-0.85 before protein expression was induced with IPTG (800 µL of a 1 M solution in water, final concentration: 1.0 mM). Expression was allowed to continue at 20 • C, 180 rpm for 14.5-16 h. The cells were harvested by pelleting with centrifugation (8000 rpm, JA10 rotor, 3×10 min, 4 • C). The pellets were flash frozen in liquid nitrogen and stored at −80 • C.

Protein Purification of p38α-Cys172
Cell Lysis: The frozen pellets (10 g) were combined and thawed on ice. Ice cold GST lysis buffer (PBS pH 7.3, 0.5 mM TCEP) (40 mL) was added and the mixture stirred until the pellets were completely resuspended. Lysosyme (50 mg) was added and the mixture stirred on ice for a further 1.5 h, after which the cells were sonicated using a sonicator equipped with a microtip (15×2 s blasts). DNase (1 mg) was then added and the mixture stirred on ice for 20 min before centrifugation (22 000 rpm, JA25.50 rotor, 20 min, 4 • C) and the supernatant taken.
GST Affinity Chromatography: The supernatant was further clarified by sequential filtration through 0.8 µm, 0.45 µm and 0.2 µm filters before loading onto a pre-packed 5 mL GSTrap TM HP GST affinity column using anÄTKA Purifier R system, chasing through (5 mL, 1 CV) with GST lysis buffer. The column was washed (40 mL, 8 CV) with GST wash buffer (PBS pH 7.3, 1 % Triton X-100) and further washed (50 mL, 10 CV) with GST lysis buffer. Elution was with a linear gradient (75 mL, 15 CV, 5 mL fractions) from lysis buffer to GST elution buffer (50 mM Tris pH 8.0, 20 mM L-glutathione). The fractions containing the p38α mutant fusion protein, as determined from the FPLC report file and by SDS-PAGE analysis, were combined.  Anion Exchange Chromatography: The protein was diluted by 2× with anion exchange start buffer (25 mM HEPES pH 7.5, 5 % glycerol, 0.5 mM TCEP) before being re-filtered and loaded onto a 5 mL HiTrap TM Q HP anion exchange column. The column was washed (50 mL, 10 CV) with anion exchange start buffer before the protein was eluted using a linear gradient (100 mL, 20 CV, 5 mL fractions) from start buffer to 50 % anion exchange elution buffer (same as start buffer but with 1.0 M NaCl). The protein corresponding to the major peak in the UV trace from the FPLC report file and with the correct mass as determined from SDS-PAGE analysis was taken and these fractions were combined.

Protein Purification of p38α-Cys180
p38α-Cys180 was purified using a similar procedure as used for p38α-Cys172: Cell Lysis: A pellet (14.0 g) containing p38α-Cys180 was thawed on ice and re-suspended (60 mL) in ice-cold GST lysis buffer (PBS pH 7.3, 0.5 mM TCEP). Lysozyme (70 mg) was then added and the mixture stirred at 4 • C for 2 h before being sonicated with a sonicator equipped with a microtip (15×2 s blasts, 60 % amplitude, 1291 J total energy). The lysate was further treated with DNase (1 mg) at 4 • C for 1 h before pelleting by centrifugation (20 000 rpm, JA25.50 rotor, 40 min, 4 • C and the supernatant taken. GST Affinity Chromatography: The cleared lysate was further clarified by filtration (0.2 µm filter) and loaded in 2 batches onto a pre-packed GSTrap TM HP GST affinity column (GE Healthcare) on an ATKA Purifier system. On each run, the lysate was chased through (5 mL, 1 CV) with more lysis buffer before washing, first (40 mL, 8 CV) with GST wash buffer (PBS pH 7.3, 1 % Triton X-100), then again (50 mL, 10 CV) with lysis buffer. The bound proteins were eluted (75 mL, 15 CV, 5 mL fractions) by a linear gradient from lysis buffer to GST elution buffer (50 mM Tris pH 8.0, 20 mM L-glutathione). The fractions containing the desired protein from both separations were determined from the UV trace of the FPLC report file and from SDS-PAGE analysis, and combined.
GST Tag Cleavage: The combined fractions were partially concentrated (23 mL final volume) by Vivaspin (10 000 MWCO) before dialysis against PBS at 4 • C for 16 h. Thrombin (400 U of a 1 U/µL solution) in PBS was added and the protein incubated at room temperature with gentle rocking for 4 h before the thrombin was inhibited with PMSF (170 µL of a 10 mg/mL solution) in isopropanol and the protein returned on ice. Small amounts of precipitate had formed on this reaction.
GST Tag Rebinding: After filtering, the protein was re-loaded (2 mL fractions collected) onto the GSTrap TM HP GST affinity column, again in 2 batches and in each run, the column was washed (10 mL, 2 CV) with lysis buffer before elution (75 mL, 15 CV) with 100 % elution buffer. The fractions from the flow-through containing the target protein of both batches were determined from the UV trace of the FPLC report file and from SDS-PAGE analysis, and combined.  Anion Exchange Chromatography: The combined fractions were diluted 2× with anion exchange start buffer (25 mM HEPES pH 7.5, 5 % glycerol, 0.5 mM TCEP), re-filtered, and loaded as a single batch onto a 5 mL HiTrap TM Q HP anion exchange column (GE Healthcare). The column was further washed (50 mL, 10 CV) with start buffer before elution (100 mL, 20 CV, 5 mL fractions) with a linear gradient from start buffer to 50 % anion exchange elution buffer (same as start buffer but with additional 1000 mM NaCl). SDS-PAGE analysis of the fractions determined that the desired protein corresponded to 2 distinct peaks in the UV trace of the FPLC report file, with the first peak being the major one. Protein corresponding to this major peak was taken as the desired protein and further treated. p38α-Cys172 (500 µL) of a 1.8 mg/mL solution, 22 nmol, 1 equiv.) in p38α storage buffer (50 mM HEPES, pH 7.5, 50 mM NaCl, 5% glycerol, 0.5 mM TCEP) was thawed on ice and the buffer exchanged using a G-25 MiniTrap TM desalting column (GE Healthcare) to p38α reaction buffer (the same as storage buffer but without TCEP). A suspension of 2 (252 µL of a 3.9 mg/mL suspension, 3.3 µmol, 150 equiv.) in p38α reaction buffer was added and the mixture shaken at 37 • C, 550 rpm for 4 h. LC-MS after this time showed >95% conversion to the Dha product. The reaction mixture was desalted using a G-25 MidiTrap column to give p38α-Dha172, which was either kept on ice for immediate use, or flash frozen in liquid nitrogen and stored at −80 • C for short-term use: Yield: 755 µL of a 0.68 mg/mL solution in p38α reaction buffer, 57%, (calculated mass: 41 403, observed mass: 41 404).

Synthesis of p38α-Dha180
p38α-Cys180 was thawed on ice and the buffer exchanged using a G-25 MiniTrap TM desalting column (GE Healthcare) to p38α reaction buffer (50 mM HEPES pH 8.0, 50 mM NaCl, 5 % glycerol). To the resulting protein (580 µL of a 2.51 mg/mL solution, 35 nmol, 1 equiv.) was added a suspension of dibromide 2 (410 µL of a 3.9 mg/mL suspension, 5.3 µmol, 150 equiv.) in the same buffer. The mixture was then shaken at 37 • C, 550 rpm for 2 h. LC-MS showed >95 % consumption of the unmodified protein p38α-Cys180 with some residual bromide adduct intermediate. The reaction mixture was desalted again using a G-25 MidiTrap TM desalting column (GE Healthcare) and the protein incubated on ice for 18 h. LC-MS showed >95% conversion to the Dha product p38α-Dha172. The reaction mixture was divided into aliquots and kept on ice for immediate further use: Yield: 1250 µL of a 0.87 mg/mL solution in p38α reaction buffer, 75%, (calculated mass: 41 373, observed mass: 41 375).

Comparison of Chemical Reaction Kinetics for Phosphocysteine Formation
Mass spectra from the timecourses of the phosphocysteine forming reactions with both p38α-Dha172 and p38α-Dha180 were quantitatively analysed (see Section 8.2.2 for more details) which revealed that the reaction with p38α-Dha172 was faster than with p38α-Dha180.

Electrostatic Calculations
The crystal structure file (1R3C.pdb) was prepared for electrostatic surface calculation using PDB2PQR 7,8 (CHARMM forcefield) before use of the ABPS PyMOL plugin 9 for calculation of the surface itself. The resultant images were manipulated in PyMOL.

Surface Accessibility Calculations
Surface accessibility was assessed using Naccess program 10,11 , calculated using 1R3C.pdb. The default probe size (1.4Å) was used. Conversion of the percentage data from Naccess into colour information for the figures was done using scripts written in Python. The relative percentages of total side-chain accessibility were used for this conversion. The percentage data used is plotted below. Due to the methods used in the calculation, percentages of over 100% are possible.

ATF2 Homology Modelling
For the structure of ATF2, a homology model was made using sequence (amino acids 1-287 from Bos Taurus, sequence ID: AA133291.1) obtained by BLAST search 12 of sequence from 1BHI.pdb. The homology model was then calculated using Phyre2 13 to give the images as displayed in the figures.

General Procedure for Measurement of Circular Dichroism Spectra
Prior to measuring the spectra, samples not already in p38α reaction buffer (50 mM HEPES pH 7.5, 50 mM NaCl, 5% glycerol) were desalted into this using a combination of G-25 SpinTrap TM desalting column (GE Healthcare) and repeated concentration/dilution by Vivaspin (MWCO 10 000). Samples were then diluted (0.36-0.72 mg/mL) to an appropriate final volume (190-220 µL), loaded into a cuvette with thin pathlength (1.0 mm) before the spectra were collected using a Chirascan (Applied Photophysics). The spectra were collected as a temperature melts (Wavelength range: 180-260 nm, Temperature range: 10 • C→90 • C→10 • C, Temperature step: 10 • C, Temperature equilibration time: 450 s, Repeat scans: 5). Data collected were exported as raw data and reprocessed in MATLAB to give the final 3D plots (mean average of repeat scans taken, spectra smoothed using Savitzky-Golay filtering. Smoothing parameters: MATLAB function: sgolayfilt from Signal Processing toolbox, Polynomial order: 1, Window size: 5).

General Procedure for LC-MS/MS Analysis including Sample Preparation
The protocol is adapted from the one associated with the "in-solution tryptic digestion and guanidination kit" (Thermo Scientific), with modifications made to the reduction step.
Sample Preparation: A sample of the protein of interest (10.5 µL) was diluted with ammonium bicarbonate (15 µL of a 50 mM solution) in water. The volume was made up with water (1.5 µL) and the mixture heated at 60 • C for 10 min. Iodoacetamide (3 µL of a 100 mM solution in water) was then added, the reaction mixture protected from light and incubated at room temperature for 30 min. The sample was digested by the addition of trypsin (1.0 µL of a 0.1 µg/µL solution), incubating at 37 • C for 2 h. A second batch of trypsin (1.0 µL) was added and incubation continued at either 30 • C for 18 h, or 37 • C for 2 h. The digested sample was flash frozen in liquid nitrogen and stored at −80 • C for further purification of the peptides.
Sample Purification: Peptides obtained from digestion were desalted using a C18 Sep-Pak cartridge (Waters). The cartridge was equilibrated by washing 5 mL with solution B (65% acetonitrile, 35% water, 0.1% formic acid), followed by washing (10 mL) with solution A (98% water, 2% acetonitrile, 0.1% formic acid). The sample was loaded onto the cartridge and the cartridge further washed (10 mL) with solution A. The peptides were then eluted (2×1 mL) with solution B, collecting the two fractions separately. The majority of the acetonitrile was removed in vacuo before the peptides were dried by lyophilisation. Finally, the dried peptides redissolved in solution A (20 µL).

Data Collection and Analysis:
The sample of redissolved, purified peptides was given to the Mass

LC-MS/MS Analysis
The samples were analysed on an Orbitrap Elite (Thermo Fisher Scientific, DE) connected to a UHPLC Proxeon EASY-nLC 1000 and an EASY-Spray nano-electrospray ion source with EASY-Spray column (Thermo Fischer Scientific, DK). Peptides were trapped on an Acclaim PepMap R trapping column (100 µm i.d. × 20 mm, 5 µm C18) and separated on an EASY-spray Acclaim PepMap R analytical column (75 µm i.d. × 500 mm, RSLC C18, 2 µm, 100Å). Samples were loaded at a pressure of 500 bar with 100% solvent A (0.1% formic acid in water) and the peptides separated by a linear gradient (length: 15 min, 7% to 30% solvent B (0.1% formic acid in acetonitrile), flow rate: 200 nL/min). Full scan MS spectra were acquired in the Orbitrap (350-1500 m/z, resolution 120 000, AGC target 1 × 10 6 , maximum injection time 250 ms). CID and ETD spectra were acquired in Ion Trap (resolution 7500, AGC cation target 3 × 10 4 , AGC Anion target 2 × 10 5 , maximum injection time 100 ms). After the MS scans, the 5 most intense peaks were selected for fragmentation based on data-dependent decision tree (DDDT). The signal threshold for fragmentation of parent ion was set to 500 ion counts. For CID fragmentation, the normalized collision energy and default charge state was set to 35% and 2 respectively. For ETD fragmentation, reaction time was set to 50 ms and supplemental activation was enabled. Charge state screening was enabled; parent ion with unassigned charge state and charge state 1 was rejected. Dynamic exclusion was enabled (exclusion size list 75, exclusion duration 5 s). The setting for DDDT is as such: ETD fragmentation was followed for charge states 3, 4 and 5 with m/z less than 750. For charge states greater than 5, ETD fragmentation was always performed. In all other cases, CID activation was performed. In a separate experiment, HCD was used. All the settings for the instrument were kept the same, except HCD spectra were acquired in Orbitrap and the 3 most intense peaks were selected for fragmentation and normalized collision energy was set to 32%.

Data Analysis
The raw data files generated were processed using MaxQuant software (Version 1.4.1.2), integrated with Andromeda search engine as described elsewhere. 14,15 For identification of phosphocysteine peptides, Andromeda searched peak lists against the sequence of p38α as well as against a list of common contaminants. Trypsin was selected as specific digestion mode with maximum number of missed cleavages set to be 4. In case of p38α-pCys180 protein; acetylation (N-term), carbamidomethylation (C), cysteine to serine, oxidation (M), phosphorylation (STY), threonine to carbamidomethylation, threonine to cysteine, threonine to dehydroalanine, and threonine to phosphocysteine were used as variable modifications. For p38α-pCys172, similar rational variable modifications were used. All spectra were manually validated.
8 Activity Assay with Substrate Detection by Mass Spectrometry (MS) 8

Data Analysis and Quantification
LC-MS datasets were processed using MassLynx as previously described (Section 2.5). Data from the deconvoluted spectra were output from MassLynx as a spectrum list for quantitative analysis using MATLAB R2012b.
In MATLAB, quantities of the three substrate species (unphosphorylated, monophosphorylated and diphosphorylated) were estimated from relative peak intensity, setting the baseline at the median of all the values in the spectrum list. The peaks were picked by taking the maximum intensities from the three ranges where the peaks of these species were expected.
To convert the MS data so that it could be compared to the data from electrophoretic radioassay (ERA), the MS signals from the two phosphorylated species were added according to the equation

Data Analysis and Quantification
The MS data was analysed and quantified using the procedure described in Section 8.

Data Reanalysis for Intermediate ATF2 Species Adducts
Close observation of the mass spectra revealed an adduct mass from the parent peak of +37-40 Da, which could be explained by cyanate adduct formation from the urea quench. 16,17 Reanalysis of the data were done to take this adduct into account, adding the mass of the adduct onto the corresponding parent peak and repeating the subsequent timecourse and kinetic analysis as described above. The two analyses revealed the same trends as observed in the previous analysis.

Control to Determine Phosphocysteine Stability against Assay Buffer Conditions
Due to the potential for phosphocysteine to be unstable towards reducing buffer conditions 18 , the stablity of the chemically modified p38α-pCys180 was verified. The same buffer conditions as described in Section 8.3 were used. 10× kinase assay buffer (0.5 mL), ATF2 storage buffer (3.7 mL), water (40 µL), ATP (0.5 mL of a 5 mM solution, pH 7.0) and p38α reaction buffer (217 µL) were combined. An aliquot (993 µL) was taken and p38α-pCys180 added (6.64 µL of a 1.0 mg/mL solution in p38α reaction buffer, final concentration: 80 nM). The protein was left to incubate at room temperature for 18 h. After this time, the protein was concentrated using Vivaspin 500 (MWCO 10 000) and the concentrated sample analysed by LC-MS. Analysis showed no detectable amounts of dephosphorylation product (p38α-pCys180: calculated mass: 41 487, observed mass: 41 487, p38α-Cys180: calculated mass: 41 407).

Control to Determine Thiophosphate Stability against DTT Treatment
To p38α-pCys180 (10 µL of a 0.5 mg/mL solution) in p38α reaction buffer was added DTT ( Proteolytic digestion of phosphorylated ATF2 was done using the FASP protocol 19 with 10 kDa Microcon filtration devices (Millipore). Protein sample (1-2 µg) was mixed with urea solution (200 µL of an 8 M solution in 50 mM ammonium bicarbonate buffer), loaded into the filtration devices and centrifuged (14 000 ×g, 15 min). The concentrate was reduced in the filtration device by adding DTT (100 µL of a 50 mM solution in 50 mM ammonium bicarbonate buffer) at 50 • C for 20 min, followed by centrifugation (14 000 ×g, 15 min). The sample was alkylated with iodoacetamide (100 µL of a 50 mM solution in 50 mM ammonium bicarbonate buffer) for 25 min in the dark. Excess alkylating agent was removed by centrifugation (14 000 ×g, 15 min), followed by washes of the centrifugal unit with ammonium bicarbonate buffer (3×100 µL). The resulting concentrate was diluted with ammonium bicarbonate buffer (50 µL of a 50 mM solution) containing trypsin (final enzyme:protein ratio: 1:50 (w/w)). The sample was incubated overnight at 37 • C. Following overnight digestion, peptides were eluted to a fresh Eppendorf tube by centrifugation. Ammonium bicarbonate buffer (50 µL) was added to the centrifugal unit and centrifuged again to elute the remaining peptides. The digested protein was diluted (final concentration: 100 fmol/µL) with 1% formic acid for analysis. LC-MS/MS analysis of the samples was then performed according to the conditions in Section 7.3.

Data Analysis
The data were processed similarly as for phosphopeptide data from the p38α variants (Section 7.3). The raw data files generated were processed using MaxQuant software (Version 1.4.1.2), integrated with Andromeda search engine as described elsewhere. 14,15 For identification of phosphorylated peptides, Andromeda searched peak lists against the human database (UniProt) as well as against a list of common contaminants. Trypsin was selected as specific digestion mode with maximum number of missed cleavages set at 2. Acetylation (N-term), oxidation (M) and phosphorylation (STY) were used as variable modifications and carbamidomethylation (C) as fixed modification. All spectra were manually validated.
The analysis showed that only two of the three possible phosphopeptides corresponding to Thr69 and Thr71 phosphorylation were detected: 1. mono-phosphorylation at Thr69 and 2. bis-phosphorylation at both sites. As was consistent with observations from whole protein ESI-MS, MS intensity for the mono-phosphorylated peptide was stronger than for the bis-phosphorylated one.
Ni 2+ Affinity Chromatography: Using anÄKTA Purifier FPLC system, the clarified lysate was loaded (110 mL) onto a pre-packed 5 mL HisTrap TM HP Ni 2+ affinity column, chasing through (17.5 mL, 3.5 CV) with more lysis buffer. The column was washed (50 mL, 10 CV) with Ni 2+ affinity wash buffer (same as lysis buffer but with 50 mM imidazole) and the bound protein eluted using a linear gradient (100 mL, 20 CV, 5 mL fractions) from wash buffer to Ni 2+ affinity elution buffer (same as lysis buffer but with 500 mM imidazole). The fractions containing the desired protein were determined from the UV trace of the FPLC report file and from SDS-PAGE analysis, and collected. Size Exclusion Chromatography and Storage: The collected fractions were concentrated by Vivaspin (MWCO 10 000) to a final volume ≈2 mL before filtering. The concentrated proteins were loaded onto a HiLoad TM 16/60 Superdex 200 gel filtration column and filtered (180 mL, 1.5 CV, 1.2 mL fractions) into MEK1 storage buffer (50 mM HEPES pH 7.5, 150 mM NaCl, 5 % glycerol, 0.5 mM TCEP). The fractions containing the desired protein were determined from the UV trace of the FPLC report file and their purity was determined from SDS-PAGE analysis. The pure and impure fractions of the desired protein were collected separately, re-concentrated by Vivaspin (MWCO 10 000, 3.1 mg/mL final concentration), divided into aliquots, flash frozen in liquid nitrogen and stored at −80 • C: Yield: 12

Synthesis of MEK1-Ethanolylcysteine222
To demonstrate that the installed Dha was reactive, MEK1-Dha222 was treated with β-mercaptoethanol.
To an aliquot of MEK1-Dha222 (50 µL of a 1.0 mg/mL solution, 1.1 nmol, 1 equiv.) in MEK1 reaction buffer was added β-meracaptoethanol (1 µL of a 10% v/v solution in MEK1 reaction buffer, 1.4 µmol, 1260 equiv.). The mixture was incubated at room temperature for 1 h. LC-MS after this time showed complete conversion to MEK1-ethanolylcysteine222. LC-MS/MS analysis of a sample prepared using the method as described in Section 7.3 showed that the adduct was installed in the desired position: