Thiocholine-Mediated One-Pot Peptide Ligation and Desulfurization

Thiol catalysts are essential in native chemical ligation (NCL) to increase the reaction efficiency. In this paper, we report the use of thiocholine in chemical protein synthesis, including NCL-based peptide ligation and metal-free desulfurization. Evaluation of thiocholine peptide thioester in terms of NCL and hydrolysis kinetics revealed its practical utility, which was comparable to that of other alkyl thioesters. Importantly, thiocholine showed better reactivity as a thiol additive in desulfurization, which is often used in chemical protein synthesis to convert Cys residues to more abundant Ala residues. Finally, we achieved chemical synthesis of two differently methylated histone H3 proteins via one-pot NCL and desulfurization with thiocholine.

In this study, we report NCL-based peptide ligation and desulfurization using thiocholine as a thiol additive. Thiocholine is an odorless and highly hydrophilic compound bearing an alkyl thiol group with a pKa of~7.8 at room temperature [17]; therefore, it was expected to be useful as a peptide thioester or thiol additive in NCL and desulfurization bearing an alkyl thiol group with a pKa of ~7.8 at room temperature [17]; therefore, it was expected to be useful as a peptide thioester or thiol additive in NCL and desulfurization ( Figure 1). We evaluated the thiocholine peptide thioester in terms of hydrolysis resistance and NCL reactivity by comparing it with other alkyl thioesters. Then, the use of thiocholine as an additive in desulfurization was also examined. Finally, we semisynthesized histone H3 proteins tethering lysine dimethylation or trimethylation by thiocholine-mediated one-pot NCL/desulfurization reaction.

Synthesis of Thiocholine Peptide Thioester
Commercially available acetylthiocholine iodide was acid-hydrolyzed to quantitatively obtain thiocholine chloride (Scheme S1) according to the previous literature [18]. Then, a model peptide 1 bearing C-terminal thiocholine thioester was synthesized to investigate its hydrolysis resistance and NCL reactivity. First, a heptapeptide bearing Cterminal hydrazide was synthesized via standard 9-fluorenylmethoxycarbonyl (Fmoc) solid-phase peptide synthesis (SPPS) with 59% isolated yield and >98% purity ( Figure  S1A). Then, the hydrazide moiety was converted to thioester by NaNO2 activation [19,20] followed by thiocholine treatment to afford thiocholine peptide thioester 1 (41% isolated yield and >95% purity) ( Figure S1B). In the same procedure, four other different peptide thioesters composed of TFET, MTG, MESNa, and thioglycolic acid (TGA) were also synthesized to determine the relative properties of thiocholine thioester (Figure S1C-F). Notably, the peptide thioesters used here had a C-terminal Gly residue, which is known to have higher NCL reactivity and lower hydrolytic resistance than other C-terminal amino acid residues [15,21].

Hydrolytic Stability and NCL Reactivity of Thiocholine Thioester
To evaluate the hydrolytic stability of thiocholine thioester, peptide 1 (1 mM) was dissolved in NCL buffer containing 200 mM NaH2PO4 (pH 7.0), 6 M Gn·HCl, and 40 mM TCEP, incubated at 25 °C, and analyzed by reversed-phase HPLC. The degree of hydrolysis gradually increased and reached 35% at 1 h ( Figure 2A). The pseudo first-order rate constant k1 of the hydrolysis was estimated as 7.5 × 10 −7 s −1 by determining the concentration of hydrolyzed peptide 2 at each time point using a calibration curve ( Figures 2B, S2A,D and S3). In the same analysis of the four other peptide thioesters, TFET and MTG thioesters showed similar k1 values of 7.3 × 10 −7 and 6.9 × 10 −7 s −1 , respectively, whereas MESNa and TGA thioesters showed relatively slower hydrolytic rates (k1 = 1.0 × 10 −7 and 1.6 × 10 −7 s −1 , respectively), suggesting that the hydrolysis rates were consistent with their pKa values.

Synthesis of Thiocholine Peptide Thioester
Commercially available acetylthiocholine iodide was acid-hydrolyzed to quantitatively obtain thiocholine chloride (Scheme S1) according to the previous literature [18]. Then, a model peptide 1 bearing C-terminal thiocholine thioester was synthesized to investigate its hydrolysis resistance and NCL reactivity. First, a heptapeptide bearing C-terminal hydrazide was synthesized via standard 9-fluorenylmethoxycarbonyl (Fmoc) solid-phase peptide synthesis (SPPS) with 59% isolated yield and >98% purity ( Figure S1A). Then, the hydrazide moiety was converted to thioester by NaNO 2 activation [19,20] followed by thiocholine treatment to afford thiocholine peptide thioester 1 (41% isolated yield and >95% purity) ( Figure S1B). In the same procedure, four other different peptide thioesters composed of TFET, MTG, MESNa, and thioglycolic acid (TGA) were also synthesized to determine the relative properties of thiocholine thioester (Figure S1C-F). Notably, the peptide thioesters used here had a C-terminal Gly residue, which is known to have higher NCL reactivity and lower hydrolytic resistance than other C-terminal amino acid residues [15,21].

Hydrolytic Stability and NCL Reactivity of Thiocholine Thioester
To evaluate the hydrolytic stability of thiocholine thioester, peptide 1 (1 mM) was dissolved in NCL buffer containing 200 mM NaH 2 PO 4 (pH 7.0), 6 M Gn·HCl, and 40 mM TCEP, incubated at 25 • C, and analyzed by reversed-phase HPLC. The degree of hydrolysis gradually increased and reached 35% at 1 h ( Figure 2A). The pseudo first-order rate constant k 1 of the hydrolysis was estimated as 7.5 × 10 −7 s −1 by determining the concentration of hydrolyzed peptide 2 at each time point using a calibration curve ( Figures 2B, S2A,D and S3). In the same analysis of the four other peptide thioesters, TFET and MTG thioesters showed similar k 1 values of 7.3 × 10 −7 and 6.9 × 10 −7 s −1 , respectively, whereas MESNa and TGA thioesters showed relatively slower hydrolytic rates (k 1 = 1.0 × 10 −7 and 1.6 × 10 −7 s −1 , respectively), suggesting that the hydrolysis rates were consistent with their pKa values.
We then investigated NCL reactivity of the peptide thioesters with N-terminal Cys peptide 3 prepared by standard Fmoc SPPS ( Figure S2B). Thiocholine thioester peptides 1 and 3 were mixed in NCL buffer at 25 • C, and analyzed by reversed-phase HPLC. Notably, the peptide concentration was set at 0.1 mM, which was lower than a conventional condition around 1 mM, in order to make the initial reaction rate easily detectable. Time-course HPLC analysis showed that the ligated peptide 4 gradually appeared and reached 7.6% after 10 min ( Figure 2C). The apparent second-order rate constant k 2 was estimated as 1.1 M −1 s −1 by determining the concentration of ligated peptide 4 at each time point using a calibration curve ( Figures 2D, S2C,E and S4). TFET and MTG thioesters showed relatively higher k 2 values of 3.4 and 1.6 M −1 s −1 , respectively, whereas MESNa and TGA thioesters showed significantly slower NCL rates (k 2 = 6.9 × 10 −2 and 4.1 × 10 −2 M −1 s −1 , respectively). We then investigated NCL reactivity of the peptide thioesters with N-terminal Cys peptide 3 prepared by standard Fmoc SPPS ( Figure S2B). Thiocholine thioester peptides 1 and 3 were mixed in NCL buffer at 25 °C, and analyzed by reversed-phase HPLC. Notably, the peptide concentration was set at 0.1 mM, which was lower than a conventional condition around 1 mM, in order to make the initial reaction rate easily detectable. Time-course HPLC analysis showed that the ligated peptide 4 gradually appeared and reached 7.6% after 10 min ( Figure 2C). The apparent second-order rate constant k2 was estimated as 1.1 M −1 s −1 by determining the concentration of ligated peptide 4 at each time point using a calibration curve ( Figures 2D, S2C,E and S4). TFET and MTG thioesters showed relatively higher k2 values of 3.4 and 1.6 M −1 s −1 , respectively, whereas MESNa and TGA thioesters showed significantly slower NCL rates (k2 = 6.9 × 10 −2 and 4.1 × 10 −2 M −1 s −1 , respectively).
These two experiments revealed that the thiocholine thioester could be classified as a reactive alkyl thioester, such as TFET and MTG thioesters, and its reactivity corresponded to the pKa value of thiocholine. Therefore, we reasoned that thiocholine would be useful as a thiol catalyst in NCL-based peptide ligation. These two experiments revealed that the thiocholine thioester could be classified as a reactive alkyl thioester, such as TFET and MTG thioesters, and its reactivity corresponded to the pK a value of thiocholine. Therefore, we reasoned that thiocholine would be useful as a thiol catalyst in NCL-based peptide ligation.

Desulfurization with Thiocholine
Thiol additives are usually used in free-radical desulfurization reactions for efficient propagation of radical species [8,[22][23][24]. We investigated the reactivity of thiocholine as a thiol additive in the desulfurization-mediated Cys-to-Ala conversion. First, an internal Cys-containing model peptide 5 was synthesized via standard Fmoc SPPS ( Figure S5). Then, peptide 5 was desulfurized in the presence of 80 mM thiocholine, 20 mM VA-044, and 300 mM TCEP ( Figure 3A). After reaction for 1 h, starting peptide 5 was almost quanti-tatively converted to desulfurized peptide 6 ( Figures 3B and S6). When the desulfurization rate was compared with that of other thiol additives, thiocholine, MESNa, and glutathione (GSH) exhibited similar fast kinetics, whereas MTG and MPAA showed significantly slower rates ( Figures 3C and S7). Notably, the reaction mixture containing MTG reached completion within several hours, while desulfurization with MTG is usually performed with additional alkyl thiol additives presumably to accelerate the reaction rate [14,[25][26][27]. We assume that the slow reaction rate of MTG could be caused by the stable alkyl radical generated after the TCEP-mediated sulfur atom abstraction. This alkyl radical should be stabilized by resonance effects with its carbonyl group and the energy barrier for the subsequent hydrogen atom abstraction might be high [28]. On the other hand, MPAA did not reach completion after a few days as previously reported [11]. Next, the concentration dependence of the thiol additive was examined with/without 5, 20, 80, 150, and 300 mM thiocholine ( Figures 3D and S8). Surprisingly, the desulfurization reaction reached completion without any thiol additive after 1 h, suggesting that external thiol-mediated radical propagation would not be an essential step in desulfurization as recently reported [29][30][31]. The presence of 20-150 mM thiocholine moderately accelerated the reaction rates and achieved full conversion. On the other hand, 300 mM thiocholine significantly reduced the reaction rate. This was probably due to the deficiency of TCEP, which is necessary to abstract sulfur atoms from thiol moieties of Cys or thiol additives in the course of the desulfurization mechanism [30].
Molecules 2023, 28, x FOR PEER REVIEW 4 of 13 Then, peptide 5 was desulfurized in the presence of 80 mM thiocholine, 20 mM VA-044, and 300 mM TCEP ( Figure 3A). After reaction for 1 h, starting peptide 5 was almost quantitatively converted to desulfurized peptide 6 ( Figures 3B and S6). When the desulfurization rate was compared with that of other thiol additives, thiocholine, MESNa, and glutathione (GSH) exhibited similar fast kinetics, whereas MTG and MPAA showed significantly slower rates ( Figures 3C and S7). Notably, the reaction mixture containing MTG reached completion within several hours, while desulfurization with MTG is usually performed with additional alkyl thiol additives presumably to accelerate the reaction rate [14,[25][26][27]. We assume that the slow reaction rate of MTG could be caused by the stable alkyl radical generated after the TCEP-mediated sulfur atom abstraction. This alkyl radical should be stabilized by resonance effects with its carbonyl group and the energy barrier for the subsequent hydrogen atom abstraction might be high [28]. On the other hand, MPAA did not reach completion after a few days as previously reported [11]. Next, the concentration dependence of the thiol additive was examined with/without 5, 20, 80, 150, and 300 mM thiocholine ( Figures 3D and S8). Surprisingly, the desulfurization reaction reached completion without any thiol additive after 1 h, suggesting that external thiol-mediated radical propagation would not be an essential step in desulfurization as recently reported [29][30][31]. The presence of 20-150 mM thiocholine moderately accelerated the reaction rates and achieved full conversion. On the other hand, 300 mM thiocholine significantly reduced the reaction rate. This was probably due to the deficiency of TCEP, which is necessary to abstract sulfur atoms from thiol moieties of Cys or thiol additives in the course of the desulfurization mechanism [30].

One-Pot NCL/Desulfurization for Histone H3K4me3 Semisynthesis
To demonstrate the utility of thiocholine-mediated NCL and desulfurization in chemical protein synthesis, we conducted one-pot semisynthesis of Arabidopsis trimethylated histone H3 (H3K4me3), which is related to transcriptional activation [32][33][34][35][36]. The full-length H3 sequence (135 aa) was divided into two peptide segments at the Ser28-Ala29 junction ( Figure 4A). The N-terminal segment 8 was synthesized as MESNa thioester, which was more stable than thiocholine thioester and suitable for isolating peptide thioester, via NaNO2 activation of C-terminal hydrazide peptide 7 ( Figure S9). The C-terminal 107 aa segment was prepared as an N-terminal Cys peptide through E. coli expression of N-terminally His-tagged protein and following tag removal by enterokinase cleavage. The recovered crude peptide 9 was purified by ion exchange chromatography, treated

One-Pot NCL/Desulfurization for Histone H3K4me3 Semisynthesis
To demonstrate the utility of thiocholine-mediated NCL and desulfurization in chemical protein synthesis, we conducted one-pot semisynthesis of Arabidopsis trimethylated histone H3 (H3K4me3), which is related to transcriptional activation [32][33][34][35][36]. The fulllength H3 sequence (135 aa) was divided into two peptide segments at the Ser28-Ala29 junction ( Figure 4A). The N-terminal segment 8 was synthesized as MESNa thioester, which was more stable than thiocholine thioester and suitable for isolating peptide thioester, via NaNO 2 activation of C-terminal hydrazide peptide 7 ( Figure S9). The C-terminal 107 aa segment was prepared as an N-terminal Cys peptide through E. coli expression of N-terminally His-tagged protein and following tag removal by enterokinase cleavage. The recovered crude peptide 9 was purified by ion exchange chromatography, treated with methoxyamine to open the thiazolidine ring [37], and finally purified by HPLC ( Figure S10).
Peptide ligation was conducted with peptide thioester 8 and N-terminal Cys peptide 9 in NCL buffer containing 80 mM thiocholine and monitored by HPLC ( Figure 4B). Within one minute, MESNa thioester 8 was converted to thiocholine thioester 8′ and ligation product 10 was observed as a main peak at 4 h ( Figure S11). Then, radical initiator VA-044 and sulfur scavenger TCEP were added to start desulfurization in a one-pot manner. After 16 h, the observed main peak in the HPLC chart was fractionated and purified. As a result, desulfurized product 11, full-length H3K4me3, was successfully obtained with high purity and identified by MALDI-TOF mass spectrometry ( Figure 4C). In the same thiocholine-mediated one-pot NCL/desulfurization strategy, we also completed the synthesis of Arabidopsis H3K9me2 (Figures S12 and S13), which is known to be correlated with gene silencing [34,[38][39][40][41][42][43]. SDS-PAGE of full-length H3K4me3 and H3K9me2 confirmed that these products were obtained with high purity ( Figure 4D).

General
All Fmoc or Boc protected amino acids were purchased from Watanabe Chemical   Figure S13).
Peptide ligation was conducted with peptide thioester 8 and N-terminal Cys peptide 9 in NCL buffer containing 80 mM thiocholine and monitored by HPLC ( Figure 4B). Within one minute, MESNa thioester 8 was converted to thiocholine thioester 8 and ligation product 10 was observed as a main peak at 4 h ( Figure S11). Then, radical initiator VA-044 and sulfur scavenger TCEP were added to start desulfurization in a one-pot manner. After 16 h, the observed main peak in the HPLC chart was fractionated and purified. As a result, desulfurized product 11, full-length H3K4me3, was successfully obtained with high purity and identified by MALDI-TOF mass spectrometry ( Figure 4C). In the same thiocholine-mediated one-pot NCL/desulfurization strategy, we also completed the synthesis of Arabidopsis H3K9me2 (Figures S12 and S13), which is known to be correlated with gene silencing [34,[38][39][40][41][42][43]. SDS-PAGE of full-length H3K4me3 and H3K9me2 confirmed that these products were obtained with high purity ( Figure 4D).

General
All Fmoc or Boc protected amino acids were purchased from Watanabe Chemical (flow rate = 7.5 mL/min) in a linear gradient. For the analytical HPLC measurements of the peptides, the 5C 18 AR-II column (4.6 ID × 250; Nacalai), Protein-R column (4.6 ID × 250; Nacalai), or Jupiter 5C 4 300 Å column (4.6 ID × 250; Phenomenex) was used with a binary mixture of A and B as the mobile phase (flow rate = 1.0 mL/min) in a linear gradient, as described in each figure or protocol. All HPLC charts indicated in this paper were monitored at 220 nm. All heating protocols were conducted in the COOL-INCUBATOR HCRCS2V75W-A0602 (IKUTA Sangyo, Kawasaki City, Japan).

Synthesis of Model Peptide Thioesters
C-terminal peptide hydrazide, Ac-TRLYRVG-NHNH 2 was first synthesized with Cl-Trt(2-Cl)-Resin (Watanabe Chemical, 1.34 mmol/g, 200 µmol scale). After swelling the resin in DMF for 20 min, Fmoc-hydrazine (50.8 mg, 200 µmol) and DIEA (96.8 µL, 0.558 mmol) in DMF (12 mL) were added to the resin, and the mixture was agitated at room temperature for 16 h. After Fmoc quantification, Fmoc-SPPS was conducted using Initiator+ Alstra. After elongation of the peptide chain, the N-terminal Fmoc group was removed manually by adding 20% piperidine in DMF twice for 5 min each, and the N-terminal free amino group was acetylated in 25% Ac 2 O/CH 2 Cl 2 for 5 min. After washing the resin with DMF and CH 2 Cl 2 , a cleavage cocktail (95.0% TFA, 2.5% triisopropylsilane [TIPS], and 2.5% H 2 O) was added to the reaction column and shaken at room temperature for 2 h. Then, the solution was filtered to remove the resin, and 10 times the volume of cold diethyl ether was added to the filtered solution to precipitate the peptide. This suspension was vortexed and centrifuged at 10,000× g at 3 • C for 5 min. Ether was removed by decantation. The precipitate was washed with diethyl ether twice. The crude peptide was dissolved in a water/acetonitrile mixture containing 0.1% TFA, purified by preparative HPLC, and identified by MALDI-TOF mass spectrometry (MS). After lyophilization, the peptide hydrazide was obtained as a white solid (96.7 mg, 84.3 µmol, 59%). The HPLC profile and MS data after purification are shown in Figure S1.

Hydrolysis Kinetics of Peptide Thioesters
Each peptide thioester (1.0 mM) and TCEP (40 mM) were dissolved in NCL buffer (0.2 M NaH 2 PO 4 [pH 7.0], 6 M Gn·HCl) and incubated at 25 • C. A 5 µL aliquot of the reaction solution was taken and quenched with 40 µL of acidic NCL buffer (pH 3.0) and 5 µL of 500 mM TCEP aq. (pH 7.0). Then, 40 µL of each sample was injected into an analytical HPLC equipped with a 4.6 ID × 250 mm COSMOSIL 5C 18 -AR-II column (Nacalai tesque). HPLC peaks were monitored at 220 nm in a linear gradient with water/acetonitrile containing 0.1% TFA (gradient: 15-50% for 30 min). The concentration of hydrolyzed product peptide 2 at each time point was calculated using the standard curve. The pseudo first-order rate constants of hydrolysis (k 1 ) were determined using a least squares line fitting the following equation: ln ([A]/[A] 0 ) = −k 1 t.

NCL Kinetics of Peptide Thioesters
To a solution of peptide 3 (final conc., 0.11 mM) and TCEP (final conc., 40 mM) in NCL buffer (0.2 M NaH 2 PO 4 [pH 7.0], 6 M Gn·HCl; 400 µL) was added a solution of peptide 1 in NCL buffer (final conc., 0.10 mM). The NCL reaction was performed at 25 • C (pH 7.0), and 50 µL of reaction solution was taken and quenched with 400 µL of acidic NCL buffer (pH 3.0) and 50 µL of 500 mM TCEP aq. (pH 7.0). Then, 40 µL of each sample was injected into an analytical HPLC equipped with a 4.6 ID × 250 mm COSMOSIL 5C 18 -AR-II column (Nacalai tesque). HPLC peaks were monitored at 220 nm in a linear gradient with water/acetonitrile containing 0.1% TFA (gradient: 15-50% for 30 min). The concentration of ligated peptide 4 at each time point was calculated using the standard curve. The apparent second-order rate constants of NCL (k 2 ) were determined using a least squares line fitting the following equation: Considering that some of the peptide thioesters were partially hydrolyzed before starting the experiment, initial concentrations of peptide thioesters (=[A] 0 ) were calculated by subtracting the amount of hydrolyzed peptide 2 at 0 min. Since the hydrolysis rate of peptide thioesters was significantly slower than the NCL rate, the concentration of the hydrolyzed peptide 2 was assumed to be constant. To the reaction solution was added 50 µL of VA-044 (100 mM), and the reaction mixture was incubated at 37 • C under an argon atmosphere (pH 7.0). For analysis of each reaction, 20 µL of reaction solution was taken and quenched with 5 µL of ascorbic acid (500 mM) in acidic NCL buffer (pH 3.0), followed by injection into an analytical HPLC equipped with a 4.6 ID × 250 mm COSMOSIL 5C18-AR-II column (Nacalai tesque). HPLC peaks were monitored at 220 nm in a linear gradient with water/acetonitrile containing 0.1% TFA (gradient: 5-30% for 30 min).
After elongation of the peptide chain, the N-terminal Fmoc group was removed by adding 20% piperidine in DMF twice for 5 min each. After N-terminal deprotection, a cleavage cocktail (95.0% TFA, 2.5% TIPS, 2.5% H 2 O) was added to the reaction column and shaken at room temperature for 2.5 h. Then, the solution was filtered to remove the resin, and 10 times the volume of cold diethyl ether was added to the filtered solution to precipitate the peptide. This suspension was vortexed and centrifuged at 10,000× g at 3 • C for 5 min. Ether was removed by decantation. The precipitate was washed twice with diethyl ether. Then, the crude peptide was dissolved in a water/acetonitrile mixture containing 0.1% TFA, purified by preparative HPLC, and identified by MALDI-TOF MS. After lyophilization, H3K4me3 peptide hydrazide 7 and H3K9me2 peptide hydrazide 12 were obtained as white solids (31.7 mg, 7.35 µmol, 17% for peptide 7; 12.8 mg, 2.98 µmol, 6.0% for peptide 12).

Preparation of Histone H3 C-Terminal Peptide Segment by E. coli Expression
The DNA fragment encoding Arabidopsis thaliana H3.1 mutant (Ala29Cys) with the recognition sequence of enterokinase (Asp-Asp-Asp-Asp-Lys) was inserted into pET-15b (Novagen, Houston, TX, USA). Notably, the DNA fragment for the recognition sequence of enterokinase was integrated between residues Ser28 and Cys29 of H3.1. The recombinant histone H3.1 mutant was produced as N-terminally hexa-histidine (His 6 )-tagged proteins in Escherichia coli BL21 (DE3) harboring the minor tRNA expression vector (Codon (+) RIL, Stratagene, San Diego, CA, USA). Bacterial cultures were grown on LB plates containing ampicillin (100 µg/mL) at 37 • C for 16 h, then inoculated into TB medium containing ampicillin (100 µg/mL) and cultured at 37 • C. Expression of H3.1 mutant was induced by the addition of 0.5 mM isopropyl-β-D-thiogalactopyranoside when the cell density reached A600 = 0.4-0.6, followed by culture at 37 • C for 12 h. Purification of the H3.1 mutant was performed using a modified version of the previously described method [45,46]. Briefly, His 6 -tagged H3.1 mutant protein was isolated and purified under denaturing conditions using nickel-nitrilotriacetic acid agarose (Ni-NTA) resin (QIAGEN, Hilden, Germany). The enterokinase light chain (New England BioLabs, Ipswich, MA, USA) was added to the purified H3.1 mutant protein (0.57 ng/mg of H3.1 mutant) to remove the His 6 -tagged N-terminal region (aa 1-28), and the mixture was incubated at 20 • C for 16 h. The reaction product was then mixed with 1.5 times the volume of cation exchange buffer (20 mM NaOAc pH5.2, 6 M urea, 10 mM 2-mercaptoethanol, 1 mM EDTA) and subjected to a HiPrep SP HP 16/10 cation exchange column (Cytiva, Muskegon, MI, USA). The truncated H3.1 (aa 29-135) was then eluted by a linear gradient of 0-900 mM NaCl with 20 column volumes of cation exchange buffer. The purified truncated H3.1 was dialyzed four times against water, frozen in liquid nitrogen, and then lyophilized. Then, the truncated H3.1 was purified by preparative HPLC. To convert the N-terminal thiazolidine impurity to the desired N-terminal Cys peptide, to peptide 9 (final conc., 1 mM) dissolved in denaturing buffer (0.2 M NaH 2 PO 4 [pH 5], 6 M Gn·HCl) was added MeONH 2 ·HCl (final conc., 200 mM) and TCEP (final conc., 20 mM). The reaction solution was incubated at 37 • C (pH 5). After 15 h, the reaction solution was purified by HPLC and identified by MALDI-TOF mass spectrometry. After lyophilization, peptide 9 was obtained as a white solid (19.3 mg).

Conclusions
In this study, we proposed and demonstrated thiocholine as a new thiol catalyst/additive in NCL and desulfurization, which are two key reactions in chemical protein synthesis. Our results revealed that thiocholine peptide thioester exhibited practical NCL reactivity and moderate hydrolytic stability among several alkyl thioesters, indicating that these properties reflected the pKa of thiocholine (~7.8). To our surprise, this thiol compound showed better activity in desulfurization than MTG, which has a similar pKa value as thiocholine. Finally, we demonstrated semisynthesis of Arabidopsis histone H3K4me3 and H3K9me2 through thiocholine-mediated one-pot NCL/desulfurization. Thiocholine is expected to be a useful option in the field of chemical protein synthesis because of its high hydrophilicity and odorless nature.