Direct target site identification of a sulfonyl-triazole covalent kinase probe by LC-MS chemical proteomics

Abstract

Chemical proteomics is widely used for the global investigation of protein activity and binding of small molecule ligands. Covalent probe binding and inhibition are assessed using liquid chromatography-tandem mass spectrometry (LC-MS/MS) to gain molecular information on targeted proteins and probe-modified sites. The identification of amino acid sites modified by large complex probes, however, is particularly challenging because of the increased size, hydrophobicity, and charge state of peptides derived from modified proteins. These studies are important for direct evaluation of proteome-wide selectivity of inhibitor scaffolds used to develop targeted covalent inhibitors. Here, we disclose reverse-phase chromatography and MS dissociation conditions tailored for binding site identification using a clickable covalent kinase inhibitor containing a sulfonyl-triazole reactive group (KY-26). We applied this LC-MS/MS strategy to identify tyrosine and lysine sites modified by KY-26 in functional sites of kinases and other ATP-/NAD-binding proteins (>65 in total) in live cells. Our studies revealed key bioanalytical conditions to guide future chemical proteomic workflows for direct target site identification of complex irreversible probes and inhibitors.

Graphical Abstract

INTRODUCTION

Kinases constitute a large and diverse class of proteins with greater than 500 members in the human proteome¹. Kinases catalyze the adenosine triphosphate (ATP)-dependent transfer of a phosphate group to protein or small molecule substrates². These enzymes are key mediators of signal transduction to regulate cell metabolism, growth, and survival in response to external stimuli³. The reversible phosphorylation of substrate proteins on serine, threonine, and tyrosine residues can alter protein conformation and activation, subcellular localization, and protein-protein interactions^4–5. Thus, kinases act as molecular switches to regulate cell biology through post-translational modification of signaling proteins. Given their role in cancer, inflammatory, and neurodegenerative diseases, kinases are prominent drug targets⁶.

Methods capable of studying endogenous kinase activity and inhibition by small molecules are needed to advance development of potent and selective kinase inhibitors. Several chemical proteomic methods including ATP acyl phosphate activity-based probes^7–9 and bead-immobilized kinase inhibitors (kinobeads^10–11) can be used for functional profiling of kinases by liquid chromatography-tandem mass spectrometry (LC-MS/MS). Although widely adopted for parallel analysis of hundreds of kinases, the reagents used for the aforementioned methods are not cell permeable, which precludes live cell studies. Cell permeable affinity-based kinase probes containing a photoreactive diazirine group have been developed but show a reduced target scope (~20 intracellular kinases) in proteomic analyses^12–13.

Recently, a cell-permeable pan-kinase probe (XO44) was shown to be effective for chemical proteomic evaluation of the kinome¹⁴. The XO44 kinase probe contains a pyrimidine 3-aminopyrazole group for binding recognition and a sulfonyl-fluoride reactive group^15–20 for facilitating covalent modification with lysine residues in kinase active sites. XO44 was capable of profiling dasatinib binding against ~130 endogenous kinases in intact cells. Target deconvolution using XO44 was accomplished by LC-MS/MS detection of tryptic peptides generated from probe-modified proteins enriched by affinity chromatography. While effective for protein-level identification, the exact amino acid site(s) modified by XO44 on these kinase targets could not be ascertained from tryptic peptide digest analyses. Site of binding analyses using XO44 were pursued but yielded LC-MS/MS data that enabled identification of only a single binding site on SRC kinase (K295)¹⁴. These results are not surprising given that probe reaction, especially larger probes such as XO44, at an amino acid site increases the molecular mass, hydrophobicity, and charge state of resulting probe-modified peptides that complicate LC-MS/MS identification. To address these structurally complex probe adducts, custom proteomic workflows have been developed to understand LC-MS/MS fragmentation mechanisms and increase the ability for binding site identifications of covalent kinase inhibitors that target cysteines^21–22.

Here, we identified chromatography and LC-MS/MS fragmentation conditions tailored for chemical proteomic evaluation of covalent kinase probes that produce large complex adducts with a target site. We synthesized a sulfonyl-triazole²³ analog of XO44 (labeled as KY-26) that contains a more reactive triazole leaving group in order to modify tyrosine and lysine residues, and thereby increase our capability to capture peptides that contain binding site residues. Our ability to identify KY26-modified sites on kinases and other target proteins was greatly improved by (1) replacing conventional C18 stationary phase in our analytical HPLC columns with PLRP-S (polystyrene/divinylbenzene) media, (2) including electron transfer dissociation (ETD) in the MS data acquisition methods and (3) utilizing hydrophilic interaction liquid chromatography (HILIC) to reduce contaminant ions for LC-MS/MS analyses. These modifications were important for identifying nearly 70 protein targets and corresponding binding sites of KY-26 as proof of concept for improved LC-MS/MS methodology for chemical proteomics.

EXPERIMENTAL METHODS

HPLC analysis of KY-26 reactions with amino acid mimetics and synthetic peptides

Reaction progress of KY-26 or XO44 with p-cresol or n-butylamine (16.5 μmol, 3.3 eq.) in the presence of TMG base (1,1,3,3-tetramethylguanidine, 1.1 eq) was evaluated by monitoring probe consumption and quantified based on the signal from the caffeine standard using HPLC. Synthetic peptide reactions were conducted by mixing the peptide (Ac-RLNERHYGGLTGLNK-NH₂, 50 nmol, 1.0 eq.) with 1.1 eq of TMG. KY-26 (550 nmol, 11.0 eq) was added to the mixture and the reaction was kept at 37 °C until the reaction achieved at least 50% conversion. Full details of this assay can be found in Supporting Methods.

Gel-based chemical proteomics

Jurkat cells were grown to 80% confluency and treated with either DMSO or probe at the designated final concentration (KY-26 or XO44, 1,000X stock in serum-free media) and incubated at 37 °C with 5% CO₂ for 30 min. Cells were harvested and lysed in PBS buffer containing EDTA-free protease inhibitors. Addition of the rhodamine fluorescent tag was accomplished by CuAAC and fluorescently-labeled proteins visualized by SDS-PAGE and in-gel fluorescence scanning. For KY-26 and XO44 live cell treated samples, 50 μL aliquots of proteome were used for gel experiments. For KY-26 lysate labeling, 49 μL aliquots of cell lysate were used for each dose and time point before adding 1 μL of 50X KY-26 stock (5 μM final) and incubated for 30 min at 37 °C. For competition experiments, ATP or KY-424 (1 μL of a 50X stock) was added to a proteome sample (48 μL) and incubated for 30 min at 37 °C before adding KY-26 (1 μL of a 50X stock, 5 μM final). Full details of gel-based assays used for in vitro and live cell chemical proteomic studies can be found in Supporting Methods.

LC-MS/MS analysis of KY-26-modified peptides

Probe-modified synthetic peptide was reconstituted in 5% AcOH and diluted to 5 pmol/μL concentration and analyzed using C18 (3 μm) or PLRP-S (3 μm) in a fused silica capillary (360 μm O.D. × 75 μm I.D.) on an Agilent 1100 Series Binary HPLC interfaced with a Thermo Scientific LTQ-XL. Probe-modified peptides from live cell studies (subjected to offline HILIC cleanup) were pressure loaded into a nanocapillary analytical column (10 cm, 3 μm 1000 Å PLRP-S packing material in 360 μm o.d. × 75 μm i.d. fused silica), with an integrated electrospray tip. Samples were initially electrosprayed into an in-house modified LTQ Velos Orbitrap mass spectrometer. Samples confirmed to contain KY-26 modified peptides were then analyzed on an Orbitrap Fusion Tribrid mass spectrometer.

Data analysis

Identification of KY-26 modified peptides was accomplished by searching data files using Byonic version 3.3.4. Data were searched against the SwissProt human protein database (January 6, 2020). All searches included a 1% false discovery rate. All modified peptides were manually inspected to verify peptide sequence and probe modification. If a peptide sequence identified by Byonic was incorrect, the peptide was identified by de novo sequencing and matched to a protein by searching the UniProt database.

RESULTS

KY-26 shows enhanced solution and proteome reactivity compared with XO44

We synthesized a sulfonyl-triazole analog of XO44 named KY-26 for our chemical proteomic studies (Figure 1). Our rationale for selecting the triazolide in place of the fluoride as a leaving group is based on our past studies that demonstrated enhanced reactivity at protein sites for sulfur–triazole exchange (SuTEx) compared with sulfur–fluoride exchange (SuFEx) chemistry^24–25. The sulfonyl-triazole reactive group was connected to the 2-aminopyrazole kinase-recognition unit through an amide linkage to increase the electron-withdrawing character of the adduct group²⁶ for enhanced reactivity of KY-26. Full details of the synthesis and characterization of KY-26 and analogs can be found in Figure S1 and Supporting Information.

Figure 1. Chemical structures of XO44 and KY-26 probes.

The SuFEx molecule, XO44, has been used to enrich for kinases in live cells by modifying catalytic lysine residues. The intact modification is difficult to detect with a purified protein, which may be due to the stability of the molecule. Modifications were made to the compound by synthesizing the SuTEx probe analog KY-26, predicted to modify both lysine and tyrosine residues found within kinase active sites. The triazole replaced the fluorine as the leaving group, and an amide bond para to the sulfonyl group was added based on previous studies for SuTEx structure activity relationships²⁶.

Initially, we compared solution reactivity of KY-26 and XO44 against nucleophiles that mimic tyrosine (p-cresol) and lysine (n-butylamine) side-chains by high-performance liquid chromatography (HPLC) as previously reported²⁵ and described in Supporting Methods. In these assays, the reactivity of electrophilic compound is evaluated by its depletion, resulting in reduced UV signals, upon reaction with nucleophile as a function of time. Covalent reaction in solution was facilitated by addition of 1,1,3,3-tetramethylguanidine (TMG) base. Although both probes showed time-dependent reaction with p-cresol, KY-26 was more reactive as evidenced by near-complete depletion by 20 min while XO44 reaction was incomplete at the longest time point tested (90 min, Figure 2A and B). Interestingly, although XO44 was reported as a lysine-targeted kinase probe, we observed minimal reaction with n-butylamine in our HPLC studies. In agreement with the tyrosine chemoselectivity reported for SuTEx probes^24–25, we observed reduced activity of KY-26 against n-butylamine compared with p-cresol (Figure 2B).

Figure 2. Comparing solution reactivity of XO44 and KY-26 against nucleophiles.

Reaction of XO44 (A) and KY-26 (B) with n-butylamine and p-cresol as a function of time. The addition of 1,1,3,3-tetramethylguanidine (TMG) base catalyzed the covalent reaction. The UV signals from each compound at the measured time point compared with t = 0 is used to quantify the percentage of substrate consumed (percent conversion). Overall, the results indicate KY-26 reacts more rapidly than XO44 with nucleophiles in solution. (C) In-gel fluorescence results for XO44 (SuFEx) and KY-26 (SuTEx) labeled proteins from in situ treatments of Jurkat cells (5 μM probe, 30 min). Rhodamine azide tags were appended to probe-modified proteomes by CuAAC to detect modified proteins from treated cells. HPLC data shown are representative of n = 3 independent experiments.

In addition to solution reactivity, we compared activity of KY-26 and XO44 in proteomes using gel-based chemical proteomics. Considering that both SuFEx¹⁴ and SuTEx²⁶ probes are cell permeable, we performed chemical proteomic studies under live cell treatment conditions. In brief, Jurkat cells were treated with KY-26 or XO44 (5 μM, 30 min, 37 °C) followed by cell lysis, separation of soluble and membrane lysates, and conjugation of a rhodamine tag to the alkyne handle of probe-modified proteins by copper-catalyzed azide-alkyne cycloaddition (CuAAC²⁷). Probe-modified proteins were visualized by SDS-PAGE and in-gel fluorescence scanning. In agreement with our HPLC findings, we observed substantially increased reactivity of KY-26 compared with XO44 in proteomes from probe-treated cells as evidenced by increased fluorescence intensity across the entire molecular weight range in our gel-based studies (Figure 2C). Although KY-26 was generally more reactive, our gel-based analyses revealed shared and distinct proteome-wide targets for XO44 and KY-26. Specifically, several protein bands were labeled more prominently in proteomes from XO44- compared with KY-26-treated cells, which supports differences in selectivity between these probes (Figure 2C). To test specificity of KY-26 labeling activity, we performed in vitro competition studies with both free ATP (10 – 0.5 mM) and a non-clickable analog of KY-26 (1 and 0.5 mM KY-424; Figure 1). We observed concentration-dependent blockade of KY-26 labeling of proteomes with pretreatments from both competitors. These findings support specific detection of probe labeling events within the ATP-binding site of target proteins that is dependent on the KY-26 inhibitor scaffold (Figure 3).

Figure 3. KY-26 labeling activity is dependent on molecular recognition.

Competition of KY-26 labeling of proteins in Jurkat proteomes as assessed by gel-based chemical proteomics. Pretreatment with free ATP (10 – 0.5 mM, 30 min, 37 °C; A) or a non-clickable version of KY-26 (KY-424, at 1 and 0.5 mM, 30 min, 37 °C; B) resulted in concentration-dependent blockade of KY-26 probe labeling (5 μM, 30 min). Rhodamine-azide tags were appended to probe-modified proteomes by CuAAC to detect modified proteins from KY-26 labeled lysates.

Next, we performed concentration- and time-dependent labeling studies with KY-26 in Jurkat cells to identify optimal probe labeling conditions. We treated cells at varying concentrations (2.5 – 25 μM) and harvested cells for gel-based chemical proteomic analyses at different time points (15 – 120 min). We observed both concentration- and time-dependent probe labeling (Figure S2). The latter finding further supports a covalent mode of action for KY-26 activity. Based on these findings, we concluded that a live cell-treatment condition of 12.5 μM of KY-26 for 2 hr was optimal for probe labeling in situ (Figure S2).

Improving reverse-phase chromatography of KY-26 probe-modified peptides using PLRP-S media

The ability to accurately identify binding sites from covalent probe modification is dependent on chromatographic separation of tryptic peptide digests of the proteome for MS identification. Probe-modified peptides generated from target proteins by protease digestion are conjugated to (desthio)biotin by CuAAC and enriched by avidin chromatography. Reverse-phase chromatography using C18 media separate these probe-modified peptides for site of binding identification using LC-MS/MS. While suitable for small covalent probes, larger and more structurally complex versions such as KY-26 are not likely to be efficiently eluted using standard reverse-phase LC conditions. We tested this hypothesis by using modifying synthetic peptides with KY-26 and comparing retention, elution, and MS detection of resulting probe-modified peptides under different LC conditions.

A synthetic peptide with the sequence Ac-RLNERHYGGLTGLNK-NH₂ was reacted with KY-26 in solution. The progress of reaction was tracked by HPLC (UV detection) to confirm at least 50% conversion before subjecting to LC-MS/MS analyses (see Supporting Methods for details). The N and C-termini of the peptide were acetylated and amidated, respectively, to prevent reactions at the peptide termini. The substrate peptide contained a tyrosine and a lysine to provide multiple sites for KY-26 modification that was facilitated by the addition of TMG base. Prior to LC-MS/MS analyses, a desthiobiotin tag was conjugated by CuAAC in order to model a probe-modified peptide detected by chemical proteomics (Figure S3). Under typical C18 reverse-phase conditions, we could not detect the KY-26-modified peptide in our reaction mixture (Figure S4) while the unmodified peptide was detected. Even with increased concentration of organic solvents in the mobile phase, the KY-26-modified peptide was not detected using the C18 analytical column (data not shown; see Supporting Methods for details of LC conditions tested). KY-26 is a large probe molecule that adds 946.4232 Da to a peptide after covalent reaction. Consequently, our findings show that C18 media, while appropriate for standard tryptic peptide analysis, was not suitable for reverse-phase LC of peptides modified with bulky probe adducts.

PLRP-S media has been used to elute hydrophobic molecules such as vancomycin, and in the chromatographic separation of proteins, including monoclonal antibodies^28–30. Thus, we surmised that PLRP-S could be a suitable alternative for chromatographic separation of KY-26-modified peptides. Analysis of reaction mixtures using a PLRP-S analytical column yielded detection of the KY26-modified peptide (Figure 4). In support of our hypothesis, we observed a ~8 min retention time difference between the modified and unmodified peptides, which indicates a substantial increase in hydrophobicity following KY-26 modification (Figure 4). In summary, our findings support the PLRP-S stationary phase as an effective alternative to C18 for reverse-phase LC-MS/MS detection of KY-26 and potentially other covalent probes that produce bulky adducts on amino acid sites to increase hydrophobicity of resulting peptides.

Figure 4. PLRP-S chromatographic separation of a KY-26-modified peptide product.

(A) Chromatograms that contain the base-peak chromatogram (BPC), extracted ion chromatograms (EIC) for the vasoactive standard peptide, angiotensin standard peptide, unmodified synthetic peptide and KY-26 modified synthetic peptide (modification on tyrosine). By using a PLRP-S column, typically used for whole protein separation, KY-26-modified peptide could be separated from the unmodified peptide. An added benefit of PLRP-S columns was that gradients could be shortened from 2–3 hr to 1 hr, improving the throughput for chemical proteomic experiments. (B) Structures and observed m/z of probe + TMG and hydrolyzed probe side products ([M+2H]⁺²) of the peptide modification reactions.

Electron-transfer dissociation improved sequence coverage of a KY-26 modified synthetic peptide

After identifying suitable LC conditions, we sequenced the KY-26 modified synthetic peptide by MS analysis to identify the site of KY-26 modification. Initially, we performed fragmentation by collisionally-activated dissociation (CAD), which yielded reasonable sequence coverage including the identity of the tyrosine residue modified by KY-26 (Figure 5A). In addition to the standard b- and y- ion series, we observed additional fragment ions that are derived from the desthiobiotin affinity tag (240 and 197 m/z; Figure 5B). These diagnostic fragment ions from KY-26 modification are consistent with findings from our previous SuTEx probe studies using similar fragmentation (higher-energy C-trap dissociation or HCD)³¹. Our CAD studies also revealed that KY-26 modification results in increased peptide charge state (+4 ion; 679.35 m/z versus +3 because of an additional proton on the probe moiety). This result was expected as kinase inhibitors such as the binding element of KY-26 contain heterocycles that increase gas-phase basicity and impact the charge state of resulting peptides subjected to LC-MS/MS analysis²².

Figure 5. CAD MS2 spectrum of KY-26 tyrosine modification of a synthetic peptide.

(A) The +4 ion (679.35 m/z) of the peptide containing a KY-26 modification was selected for an MS2 scan in the ion trap. With CAD, predominant fragment ions from the fragmentation of the desthiobiotin tag at 240 and 197 m/z are present in the MS/MS spectra. The peptide and site localization of the KY-26 modification on the tyrosine residue (highlighted in red) is identified but the b and y-ion series are incomplete. (B) Zoomed image of the CAD spectrum with peaks normalized to the largest peak in each subsection. Ions originating from fragmentation of the probe are highlighted in orange, b-ions are highlighted in blue, and y-ions are highlighted in green.

Higher charge state peptides yield complicated product ion spectra that contain multiply charged fragment ions that reduce the accuracy of search algorithms used for peptide identifications. The incomplete sequencing of the KY-26-modified peptide by CAD is likely a result of its higher charge state. We tested this hypothesis by sequencing the KY-26-modified peptide using electron-transfer dissociation (ETD), which is well-suited for analysis of higher charge state peptides (z ≥ +3)³². Another beneficial feature of ETD is the ability to preserve labile modifications (e.g. phosphorylation³³) that would be predicted to reduce fragments generated from the desthiobiotin tag. Using ETD, we achieved near-complete sequencing (c- and z-ion series) of the KY-26-modified peptide including the site of probe modification (Figure 6A). The 240 or 197 m/z fragment ions were not detected and indicative of preservation of the desthiobiotin tag (Figure 6B). We did observe a 949 m/z fragment ion that corresponded to loss of the KY-26 modification, which could be used to confirm the probe modification on peptides (Figure S5).

Figure 6. ETD MS2 spectrum of KY-26-tyrosine modification of a synthetic peptide.

(A) The +4 ion (679.35 m/z) of the peptide containing a KY-26 modification was selected for an MS2 scan in the ion trap. Using ETD, instead of desthiobiotin fragments, the loss of the entire KY-26 side chain is observed but not in high abundance. The benefit of ETD is that labile bonds are preserved, and the intact modification can be easily localized, as seen in this spectrum containing a near complete c and z-ion series. (B) Zoomed image of the same ETD MS2 spectrum with peaks normalized to the largest peak in each subsection; c-ions are highlighted in blue, and z-ions are highlighted in green.

Increasing the number of KY-26 target site identifications in live cell chemical proteomic studies

Next, we treated Jurkat cells with KY-26 using our optimized treatment conditions (12.5 μM, 2 hr) followed by cell lysis and chemical proteomics analysis using our tailored experimental workflow (Figure S6). We decided to employ a combination of HCD and ETD to take advantage of the benefits of both MS dissociation methods. Poly(ethylene glycol) (PEG)-related species are common polymer contaminants in MS samples that can be introduced into samples from plastics, pharmaceuticals, and personal care products³⁴. Desalting of samples by reverse-phased C18 resin (e.g. StageTips³⁵), while effective for removing salts, are not able to remove polymers that bind to these resins³⁶. Thus, we used hydrophilic interaction liquid chromatography (HILIC), which has been previously shown to be effective for removing PEG polymers³⁷, to reduce contaminant ions in our LC-MS/MS analyses (Figure S7). Probe-modified peptides derived from KY-26-targeted proteins from live cell treatments were analyzed on a Thermo Scientific Orbitrap Fusion Tribrid instrument that is capable of high resolution data acquisition employing both HCD and ETD fragmentation. Additional details of the HILIC cleanup, chemical proteomics, and LC-MS/MS analysis can be found in the Supporting Methods.

We identified KY-26-modified tyrosine and lysine sites on probe-modified peptides from kinases and other target proteins (Table S1). Importantly, KY-26-modified lysines were catalytic residues that resided in kinase active sites. These findings support the initial rationale for choosing the pyrimidine 3-aminopyrazole for mediating binding recognition of XO44¹⁴. We also confirmed KY-26 would modify tyrosine residues in kinase hydrophobic binding pockets and specifically, within the nucleotide binding domain (Table S1). These findings combined with our recent report using SuTEx probes²⁶ support tyrosines as ligandable sites for future development of covalent kinase inhibitors. We found that the use of HCD/ETD compared with HCD fragmentation alone increased the number of detected probe-modified proteins and peptides containing the KY-26 modified site (Figure 7).

Figure 7. KY-26 modified proteins and binding sites using HCD/ETD compared with HCD alone.

Proteins (A) and peptides (B) were identified using HCD and ETD compared with HCD alone. We identified 5 proteins unique to HCD analysis (stress-induced phosphoprotein 1, GTP-binding nuclear protein, septin-7, heat shock protein 90-beta, and MAP/microtubule affinity-regulating kinase 4). Akin to the total number of proteins identified, additional probe-modified peptides were identified when ETD was included in analyses. These results indicate that ETD analysis substantially improves protein and peptide identification from KY-26-modified peptides. All results presented in the above diagrams are peptides analyzed from tryptic digests. A full list of probe-modified peptide sequences including the site of modification can be found in Table S1.

Next, we explored alternative proteases in order to improve LC-MS/MS identification. Trypsin is a widely used protease for LC-MS/MS analysis and generates peptides in the range of 700 – 1500 Da³⁸. While predictable and ideal for CAD and HCD MS/MS analysis, tryptic peptides are not always well-suited for ETD analysis, which requires peptides with higher charge density³⁸. Furthermore, high sequence homology within kinase active sites may confound the ability to differentiate peptides from related members including, for example, different kinase isoforms. We chose chymotrypsin as a second protease for our LC-MS/MS studies with the goal of producing larger peptides for ETD analysis and improving kinase identification from enriched probe-modified peptides. The addition of chymotrypsin aided in the detection of 4 new kinase targets and 6 non-kinase targets of KY-26 (Table S1).

In summary, we found that the combination of HCD and ETD along with both trypsin and chymotrypsin protease allowed assignment of KY-26 modification across >65 probe-modified peptides in proteomes from live cell treatments. The target proteins were enriched for kinases with modifications occurring at the expected catalytic lysine and novel tyrosine sites. We also identified several non-kinase protein targets including probe modifications in the nucleotide binding domains of ATP- and NAD-binding proteins. Collectively, these data support the ability of KY-26 to target nucleotide binding domains in live cells.

DISCUSSION

Targeted covalent inhibitors are emerging as enabling probe molecules^39–41 and effective drug compounds^42–44. Methods capable of direct identification of site of binding (i.e. covalent adduct of probe with a target protein amino acid site) are needed to understand mode of action of larger probe scaffolds and guide development of targeted covalent inhibitors with improved selectivity. In contrast with smaller covalent probes and ligands, targeted covalent inhibitors are generally larger in molecular mass, which complicates binding site identifications by increasing the hydrophobicity and charge state of resulting probe-modified peptides analyzed by chemical proteomics. Consequently, a common alternative approach is the LC-MS/MS detection of tryptic peptides generated from probe-modified proteins enriched by affinity chromatography for protein-level identification.

To facilitate site of binding analyses for targeted covalent inhibitors, we present LC-MS/MS conditions tailored for chemical proteomic evaluation of a SuTEx probe based on a kinase inhibitor scaffold (KY-26). Akin to other targeted covalent inhibitors, KY-26 modification increases the molecular weight (+946 Da), hydrophobicity, and charge imparted onto peptides from modified proteins. We tested and identified chromatography conditions and dissociation strategies to guide LC-MS/MS analysis of bulky probe adducts introduced by KY-26. We applied these optimized LC-MS/MS conditions to identify tyrosine and lysine sites modified by KY-26 in functional sites of kinases and other ATP-/NAD-binding proteins (>65 in total) in live cell chemical proteomic studies. Competition of KY-26 labeling with free ATP and a non-clickable analog supports molecular recognition as an important feature of KY-26 labeling activity (Figure 3).

We found that C18 media was ineffective for reverse-phase separation of KY-26-modified peptides due to the increased hydrophobicity of peptides. PLRP-S was identified as an alternative medium for our analytical columns used for nanoflow LC, which enabled the retention and elution of KY-26-modified peptides (Figure 4). PLRP-S is advantageous due to its chemical and mechanical stability, and unlike C18, does not contain surface silanols which result in analyte tailing²⁸. We found that KY-26 modification changed the chromatography of probe-modified peptides substantially when compared to the unmodified peptide; the difference in elution times (~8 min) is indicative of increased hydrophobicity from KY-26 modification (Figure 4).

We identified MS dissociation strategies to increase coverage of identified KY-26-modified proteins and corresponding sites. Specifically, we demonstrated the benefits of including ETD fragmentation in chemical proteomic workflows including increased sequence coverage on high charge state peptides that result from KY-26 modification. ETD was first described for sequencing phosphopeptides and has since been deployed for LC-MS/MS analysis of various post-translational modifications (glycosylation, palmitoylation, etc.)^{32–33, 45–46}. The ability to preserve labile bonds with ETD was also important for reducing fragment ions from the desthiobiotin tag to reduce complexity of MS/MS spectra and increase sequence coverage (Figure 6). The combination of HCD and ETD in our LC-MS/MS studies facilitated increased identification of KY-26 modified sites in proteomes from probe treated cells. Analyses of probe-modified peptides with HCD/ETD yielded 51 target site assignments compared with 35 sites from HCD alone (Figure 7 and Tables S1). We also found peptides that were too small for differentiating between kinase members in our analyses of tryptic samples (< 6 amino acids, e.g., K[+KY-26]K, Y[+KY-26]IEK). The use of chymotrypsin as an alternative protease to generate larger peptides for LC-MS/MS protein identification yielded 4 new KY-26 kinase targets (Table S1).

We are cognizant that the number of probe-enriched kinases identified using KY-26 is lower than reported for XO44 despite higher reactivity for KY-26. Our approach enriches for and detects probe-modified peptides derived from KY-26 labeled proteins and thus measures probe-bound proteins exclusively. In contrast, protein-level identification strategies for assigning targets to XO44 and additional targeted covalent inhibitors measure tryptic peptides derived from proteins bound to affinity resin. While some proteins are enriched by affinity chromatography through direct probe binding, indirect mechanisms (e.g. protein-protein interactions with probe-bound proteins) may artificially inflate reported protein targets. Follow-up studies are needed to further improve the detection capabilities of our approach to complement and verify protein-level methodologies for identifying targets of covalent inhibitors with increased structural complexity. We envision testing additional reversed phase or ion exchange resins as well as incorporating reactive groups with specificity for other amino acids⁴⁷. Our LC-MS/MS workflow may also prove useful for detecting peptides modified by photoreactive probes⁴⁸ through improved chromatography and sequence coverage.

In summary, KY-26 along with additional covalent kinase probes/inhibitors^{14, 21, 41} are among a collection of activity-based probes used for chemical proteomic evaluation of kinase function and inhibitor binding. The development of appropriate LC-MS/MS methodology, including those described in this report, are needed to support these chemical proteomic efforts to advance basic and translational investigations of the human kinome.