Chuan
Wan‡
a,
Zhanfeng
Hou‡
a,
Dongyan
Yang‡
c,
Ziyuan
Zhou
d,
Hongkun
Xu
a,
Yuena
Wang
a,
Chuan
Dai
a,
Mingchan
Liang
b,
Jun
Meng
d,
Jiean
Chen
b,
Feng
Yin
*b,
Rui
Wang
*b and
Zigang
Li
*ab
aState Key Laboratory of Chemical Oncogenomics, School of Chemical Biology and Biotechnology, Peking University Shenzhen Graduate School, Shenzhen, 518055, P. R. China. E-mail: lizg.sz@pku.edu.cn; lizg@szbl.ac.cn
bPingshan Translational Medicine Center, Shenzhen Bay Laboratory, Shenzhen, 518118, P. R. China. E-mail: wangrui@szbl.ac.cn; yinfeng@szbl.ac.cn
cCollege of Chemistry and Chemical Engineering, Zhongkai University of Agriculture and Engineering, Guangzhou, 510225, P. R. China
dNational Cancer Center/National Clinical Research Center for Cancer/Cancer Hospital & Shenzhen Hospital, Chinese Academy of Medical Sciences and Peking Union Medical College, Shenzhen, 518116, P. R. China
First published on 3rd December 2022
Visible-light-mediated methods were heavily studied as a useful tool for cysteine-selective bio-conjugation; however, many current methods suffer from bio-incompatible reaction conditions and slow kinetics. To address these challenges, herein, we report a transition metal-free thiol-sulfoxonium ylide photo-click reaction that enables bioconjugation under bio-compatible conditions. The reaction is highly cysteine-selective and generally finished within minutes with naturally occurring riboflavin derivatives as organic photocatalysts. The catalysts and substrates are readily accessible and bench stable and have satisfactory water solubility. As a proof-of-concept study, the reaction was smoothly applied in chemo-proteomic analysis, which provides efficient tools to explore the druggable content of the human proteome.
Fig. 1 Cysteine-selective bioconjugations. (a) Reported methods for cysteine-selective bioconjugation. (b) This work: visible-light-induced thiol-sulfoxonium ylide click reaction. |
Alternatively, several visible-light-induced methods, including thiol–ene49–51 and thiol-yne reactions,52S-arylation53,54 and desulfurative functionalization,55,56 have been well established as efficient Cys-based bioconjugation methods (Fig. 1a). The multidimensional controls and novel reaction pathway of these methods provide versatile tools for bioorthogonal applications.57,58 Despite the advances, the thiyl-radical based reactions suffered from relatively slow kinetics and non-biocompatible conditions, such as the use of transition metal catalysts and/or stoichiometric amounts of the oxidant. Another challenge is the side reactions caused by the highly oxidizing conditions, which may severely disrupt the structure of the protein or cause unwanted cross-linking. Consequently, the requirement of a visible-light-induced bioconjugation that possess the advantages of both the photocatalytic and classical methods, i.e. spatiotemporal control, fast reaction kinetics and high chemo-selectivity under mild and biocompatible conditions, is still unmet.
Sulfoxonium ylide is one of the most stable and industrially safe ylides that contains a nucleophilic carbon attached to sulfoxide.59–61 Significant efforts were devoted to the study of X–H insertion of sulfoxonium ylides (Fig. S1†).62–70 Brønsted acid or transition metal catalysts are generally needed for these methods, and they mainly go through either a nucleophilic addition or an electrophilic metal carbenoid pathway.59,60 Previously, sulfoxonium ylide was utilized to design cathepsin X-selective activity-based probes (ABPs) by Edgington-Mitchell et al.;71 the thiol-sulfoxonium ylide reaction is relatively slow and requires specific conditions and substrates.64 Motivated by the hydrogen atom transfer (HAT) pathway of photocatalysis,72 we envisioned that the sulfoxonium ylide species may act as a sacrificial hydrogen acceptor to provide the highly reactive sulfoxonium species, and then undergo a fast nucleophilic substitution with thiol groups.64
Herein, we report a novel visible-light-induced thiol-sulfoxonium ylide click reaction that enables Cys-selective bioconjugation under physiological conditions (Fig. 1b). Readily accessible, bench stable and water-soluble sulfoxonium ylides were prepared and utilized in this study. By exploring the conditions in aqueous media, derivatives of riboflavin (vitamin B2) behaved as the most efficient photosensitizers. The practicality of the reaction was further investigated with a series of Cys-containing peptides and proteins. In addition, a chemo-proteomic application was performed to further validate its biocompatibility and possibility as an efficient tool for exploring the druggable content of the human proteome.
To further optimize the reaction conditions, we defined the thiol-sulfoxonium ylide reaction with a riboflavin tetraacetate (RFTA) photocatalyst under 450 nm light in water as the standard conditions (86% yield, entry 1, Table 1), and the observed second-order reaction constant k2 was estimated to be 0.172 M−1 s−1 (Fig. S4†). The light irradiation and photosensitizer (RFTA) are both essential for this reaction (entry 2–7). To examine the influences of different solvents, phosphate buffer (PB, pH 7.4 in D2O), protic solvent methanol-d4 as well as aprotic solvent DMSO-d6 and acetonitrile-d3 were used as the reaction solvent (entry 8–11). Interestingly, the reactions in aqueous and protic solvents gave significantly higher yields (87% and 74%) than that in aprotic solvents (18% and 26%), and the disulfide by-product was detected as the main product in aprotic solvents (see detailed data in the ESI†), hinting at the importance of a hydrogen source. Notably, the α-carbonyl methylene of product 2a was fully deuterated in D2O (Fig. S4†). Furthermore, the addition of a radical trapper (2,2,6,6-tetramethylpiperidine-1-oxyl, TEMPO) fully quenched the reaction (entry 12), indicating a radical involving pathway in this reaction. Additional reaction conditions, including oxygen, temperature and scale variants, were investigated to assess the sensitivity of the current protocol (Table 1).74,75 The average yields for typical protic and aprotic solvents were assessed, respectively. As a result, except for aprotic solvents, the transformation was shown to be insensitive, suggesting the robustness of this thiol-sulfoxonium ylide reaction.
Entry | Deviations from the standard conditionsa | Yieldb (%) |
---|---|---|
a Standard conditions: 1a (50 mM), 2 (25 mM) and photocatalysts (2% mol) under light irradiation (450 nm) for 20 min in D2O at rt. b Yields were determined by 1H NMR with dimethyl sulfone (MSM) as the internal standard. c Methanol-d4 was used as the solvent. | ||
1 | None | 86 |
2 | Dark (12 h) | 0 |
3 | Dark (50 °C, 12 h) | 0 |
4 | White light | 58 |
5 | UV light (365 nm) | 0 |
6 | Absence of RFTA | 0 |
7 | Flavin mononucleotide (FMN) instead of RFTA | 81 |
8 | PB buffer (pH 7.4) instead of D2O | 87 |
9 | Methanol-d4 instead of D2O | 74 |
10 | DMSO-d6 instead of D2O | 18 |
11 | Acetonitrile-d4 instead of D2O | 26 |
12c | Addition of TEMPO | 0 |
Then different sulfoxonium ylides and thiol substrates were tested to gain more insights into this photoreaction. For the scope of sulfoxonium ylides, 1b, 1d, 1e, 1f and 1i with electron-donating groups, 1c with an electron-withdrawing group and 1g and 1h with a hetero-aromatic ring were tested (Fig. 2). The isolated yields of products (2c) from electron-withdrawing sulfoxonium ylides were higher than those of electron-donating compounds (2b, 2d, 2e, 2f and 2i), and the hetero-aromatic ring containing substrates gave moderated yields, such as 2g (79%) and 2h (74%). For the scope of thiol substrates, glutathione (GSH) 3, propanethiol 4, mercaptoethanol 5 and 4-mercaptopyridine 6 were subjected to the reaction with 1a, and satisfactory yields (76 to 91%) were achieved (see more substrates in Fig. S5†). As a result, the thiol-sulfoxonium ylide reaction was carried out smoothly with satisfactory yield and substrate tolerance.
The UV-vis absorption spectra showed that the visible light (400–500 nm) was exclusively absorbed by RFTA (Fig. 3a and S7†). Stern–Volmer luminescence quenching studies were then performed between RFTA and substrate 1a and 2, respectively (Fig. 3b and S8†). Apparently, favorable luminescence quenching between RFTA and 1a was observed, but no significant interactions between RFTA and 2 were observed. Thus, the energy transfer or electron transfer event between RFTA and 1a was proved. In addition, a control reaction was carried out in the absence of a thiol substrate, and the sulfoxonium ylide was decomposed to DMSO, indicating the highly reactive nature of the potential intermediate (Fig. 3c and S9a†).
To examine the proton transfer event, the hydrogen sources were investigated in deuterated protic and aprotic solvents. As summarized in Fig. 3d (see detailed data in Fig. S9†), the reactions were performed smoothly in protic solvents and most of the carbonyl-α-carbon was deuterated. In contrast, the desired reactions were inefficient in aprotic solvents and the α-carbon was exclusively hydrogenated. Consequently, we concluded that the hydrogen source is mainly from the solvent in the protic solvents, and thiol may be the hydrogen source in the aprotic solvents.
Based on the mechanistic experiments, a photocatalytic activation of sulfoxonium ylide was proposed (Fig. 3e). The flavin (Fl) photocatalyst is light-promoted to the singlet-excited state followed by intersystem crossing to result in the triplet-excited state [Fl]*.76 The triplet-excited flavins were reported as a strong single electron oxidant (Ered1/2 = 2.2 V versus Fc/Fc+ electrode for RFTA) and they should undergo facile single electron transfer (SET) with sulfoxonium ylide (Ered1/2 = 1.3 V versus Fc/Fc+ electrode for 1a) to furnish radical cation I along with the radical anion [Fl]˙− (see the ESI† for experimental details). The flavin species can act as a base (pKa [RFTA-H]˙ = 8.3),77,78 favoring proton transfer (PT) from the protic solvent or thiol group. This unique property of flavin derivatives can partially explain their outstanding catalytic efficiency. The radical cation I could abstract a hydrogen atom from the redox state [Fl-H]˙ to turn over the ground-state photocatalyst Fl79 and generate the highly reactive sulfoxonium species II. The nucleophilic substitution between the sulfoxonium II and ionized thiol is a fast step64 which furnishes the target product and DMSO.
With the thiourea additive, the conditions for peptide modification were further optimized. Photoredox conditions are essential in the presence of thiourea (Fig. S10†). RFTA was still the most efficient photocatalyst in the reaction of peptide, and the kinetic investigation demonstrated that the starting peptide was fully converted within 1 min. In addition, degassing and nitrogen protection of the model peptide reaction could further decrease the formation of by-products.
Then various sulfoxonium ylides and peptides were subjected to the visible-light-induced thiol-sulfoxonium ylide reaction under optimized conditions (Fig. 4). First, the sulfoxonium ylide 1a–1i and four additional substrates 1j–1m were reacted with model peptide 7. Except for the highly electron-donating substrates, most of the reactions gave satisfactory yields (57–89%). Moreover, four short peptides 8–11, containing various nucleophilic residues, were designed and prepared for the investigation of AA tolerance. Under the standard conditions with sulfoxonium ylide 1a, moderate to high yields were obtained for all of the four peptides (49% to >95%). Furthermore, we tested a model peptide 12, which contains all of the 14 reactive AA residues, for the chemo-selective study of the reaction. 68% of Cys-adducted product 12a was obtained, highlighting the excellent chemo-selectivity of the thiol-sulfoxonium ylide reaction. And the biologically relevant substrates (bioorthogonal handle 1l and biotin 1m) reacted with peptide 12 in a different manner. Only a trace amount of 12l was detected, and moderate yield (55%) of 12m was observed. To further examine the practicality of the reaction, two protein fragments 13 and 14 were derived from two important cancer targets epidermal growth factor receptor (EGFR) and estrogen receptor (ER), respectively. Similarly, the reactions of 1a and 1m with 13 and 14 gave higher yields (52–92%) than that of 1l (23–29%). All of the peptide products were characterized by MS/MS analysis on the cysteine site, and the adducted positions were further confirmed by the MS/MS searching for all the potential functionalization on nucleophilic residues in product 12a, 13a and 14a, which demonstrated that the cysteine site is the only possible reaction position (see detailed data in the ESI†). To further evaluate the chemo-selectivity, we have performed reactions on peptides that contain nucleophilic residues but without free cysteine (Fig. S11†), and no product was observed under the standard conditions for the thiol-sulfoxonium ylide reaction. All in all, the current thiol-sulfoxonium ylide protocol presents a versatile platform for Cys-specific bioconjugation under physiological conditions.
Having optimized the reaction conditions with peptides, the visible-light-induce thiol-sulfoxonium ylide reaction for chemical modification of proteins was then investigated (Fig. 5). In order to eliminate the interference of possible side reactions, nitrogen protection was performed for protein modification (Fig. 5a). Bovine serum albumin (BSA) was used as a model protein due to its unique free Cys34 residue. ESI-TOF MS analysis of reactions between BSA and 1a/1m afforded >90% modification (Fig. 5b and c). In addition, the reaction between 1a and ubiquitin-conjugating enzyme 2C (UBE2C) gave >70% modification (Fig. 5d). Significantly, an exceedingly fast (10 s reaction time) reaction was found. Furthermore, the Cys34 selectivity of the thiol-sulfoxonium ylide reaction was confirmed by LC-MS/MS analysis on the BSA-1a adduct (Fig. 5e), and circular dichroism (CD) analysis was also performed to check the potential conformational change (Fig. 5f). Similar CD spectra were observed for the control BSA and two adducts, highlighting that there was no significant change in their secondary structural content under the photoreaction conditions. Thus, the thiol-sulfoxonium ylide photo-click reaction was capable of covalently modifying proteins with controllable and ultrafast kinetics and outstanding cysteine selectivity under mild and biocompatible conditions.
With the efficient labeling of proteins, we then applied the thiol-sulfoxonium ylide reaction in chemo-proteomic applications. As a relatively low-reactive electrophile, the biological application of a sulfoxonium ylide warhead was highly reliant on the design of ligand-based probes. For example, the practicability of sulfoxonium ylide electrophiles as one ABP to detect the cathepsin X activity was evidenced.71 But, it's difficult to directly apply the sulfoxonium ylide warhead to globally profile the reactive and ligandable cysteinome. In fact, due to the impact of the off-target effect, researchers usually use different warheads in the study of chemoproteomics and covalent ligand/inhibitors, respectively.46–48 Thus, we proposed that the visible-light-activated condition may enable the direct application of sulfoxonium ylide as a probe to profile the functional proteome, and thus provide more visions to explore the druggable contents by using this hydrophilic, stable and cysteine-selective probe.
Along these lines, we conducted protein profiling by both gel-based and MS-based protocols (Fig. 6a). Initially, we checked the western-blot (WB) analysis of the biotinylated BSA-1m adduct, and obvious bands emerged (Fig. 6b), demonstrating that the biological function of biotin was preserved after the reaction. Next, we switched the protein to human cells (HeLa and MCF7 cell lysates) and found that 1m exhibited relatively higher labeling effects in HeLa cells (Fig. 6c). We then proceeded to investigate the labeling efficiency and cysteine selectivity of 1m in a competition assay with the known highly reactive and cysteine-selective reagent, iodoacetamide (IAM). 1m exhibited strong immunofluorescence intensity at a concentration of 2, 5, and 10 mM. Pretreatment of excess IAM successfully decreased the intensity of the bands, indicating that the labeling of 1m predominantly occurs at cysteine residues with a high cysteine selectivity (Fig. 6d). Furthermore, we employed the MS-based proteomics technique, and 3246 modified cysteine sites from HeLa cells were identified by the light-induced labelling of 1m (see more details in ESI Table S3†). The percentage of unique peptide modification for each nucleophilic amino acid was plotted for 1m. As shown in Fig. 6e, 1m primarily labeled cysteine residues with a significantly high cysteine reactivity (>90%). Interestingly, the alignment of local sequences flanking the modified cysteines in HeLa showed that 1m prefers to label cysteine residues that surrounded by serine residues (Fig. 6f). As an example, by analyzing the b, y ion mode, annotated MS/MS of the 1m-labeled RPL37 peptide was sufficient to confirm the Cys-site selectivity of 1m labelling in the chemoproteomic studies (Fig. 6g). Taken together, as a proof-of-concept study, the visible-light-induced conditions have enabled the direct application of the thiol-sulfoxonium ylide reaction for protein profiling in the proteome. This result not only validated the biocompatibility of the photo-click reaction, but also provided a possibility to develop a covalent ligand/inhibitor for cysteine by using the same reactive chemotypes of chemo-proteomics analysis under appropriate conditions.
Footnotes |
† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d2sc05650j |
‡ These authors contributed equally to this work. |
This journal is © The Royal Society of Chemistry 2023 |