Water-soluble allyl sulfones for dual site-specific labelling of proteins and cyclic peptides† †Electronic supplementary information (ESI) available. See DOI: 10.1039/c6sc00005c

Allyl sulfones as efficient disulfide rebridging agents for site-specific protein modifications with up to two additional functionalities in water.


Introduction
The spatially dened chemical modication of proteins represents a vibrant eld of research, with great impact on various research areas such as the elucidation of protein functions, 1 monitoring cellular processes, 2 the development of new biocatalysts, 3 the construction of biomaterials 4 and the generation of novel therapeutics. 5 For most of these applications, reproducible and well-dened protein conjugates need to be achieved with retained structural integrity and biological function. Despite great progress in this area, the modication of native proteins at a distinct location still represents a major challenge. Methods for site-selective protein modication can be roughly divided into two categories, with the rst targeting a specic amino acid at the protein surface based on its abundance and accessibility. The relatively rare amino acid cysteine is a popular target for single-site modication. 6 However, only very few proteins offer accessible, unpaired cysteine residues and consequently cysteine point mutations need to be introduced. 7 N-terminal modication represents a powerful approach, which has certain limitations if the terminus is critical for function. 8 The second method involves the incorporation of unnatural amino acid, allowing subsequent chemical modication with a high level of site-selectivity. 9 However, tedious synthesis of aminoacylated tRNA can limit its general applicability. 10 As a complementary strategy, disulde rebridging represents a versatile technique facilitating the selective modication of accessible disuldes on proteins and peptides. 11,12 This strategy involves a Michael acceptor system and two cysteines in close vicinity and it is particularly attractive for medicinal applications, 13 since most therapeutically relevant proteins offer at least one disulde bond close to the surface. 14 Disulde rebridging reagents such as bis-sulfones (Scheme 1B) and dibromomaleimides have been applied for the functionalization of peptides 13,15,16 and Fab fragments 17,18 that provide just one accessible disulde bridge. Additionally, multiple disul-des of antibodies 18,19 and therapeutic proteins 20 have been functionalized previously. Bioconjugation reagents combining both the maleimide and bis-sulfone function offer efficient cross-conjugation of two different thiol-containing proteins, peptides or oligonucleotides. 21 The bis-sulde bioconjugates produced by disulde rebridging reagents can be disintegrated under certain stimuli, which offers great potential in biomedical sensing, medical diagnosis and controlled drug release. 22,23 However, until now, both low water-solubility and reactivity have limited the application of bis-sulfones and organicaqueous co-solvents need to be added, which could lead to denaturation of the protein structure. 15 In addition, disulde functionalization with bis-sulfones usually yields the monofunctionalized protein and the attachment of multiple functionalities to a single disulde bridge has not been achieved yet. [24][25][26] Herein, we rst introduce allyl sulfones as efficient disulde rebridging agents providing improved reactivity without in situ activation, stability, high water-solubility and site-specicity for precise protein modications with up to two additional functionalities. We demonstrate the broad applicability of our approach on the basis of the peptide hormone somatostatin (SST), bovine insulin as well as lysozyme with one, three and four disulde bridges, respectively. Excellent siteselectivity and predictability allow the functionalization of the most solvent accessible disulde at the protein surface by retaining functional activity. Noteworthy, site-specic disulde modication of insulin and lysozyme is demonstrated for the rst time. Allyl sulfones offer the combination of up to three different functionalities at a single site in a modular fashion. In this way, access to customized bioconjugates is granted by e.g. attaching a chromophore and a purication tag in a single step by simply adjusting the pH.

Results and discussion
Design of allyl sulfones for protein modication Water-soluble allyl sulfones 1 were designed based on the essential features of disulde rebridging agents, which include an effective leaving group (e.g. p-toluene sulfonyl group and halogen) and an electron-withdrawing keto-group conjugated to a double bond. Allyl sulfones 1 rebridge the most accessible disuldes of peptides and proteins aer disulde reduction via sequential addition-elimination reactions (Scheme 1A). First, the mild reduction of a solvent accessible disulde releases two free thiols in close vicinity. Next, allyl sulfones 1 react by thiol addition of the double bond with subsequent elimination of the p-toluene sulnic acid group to form a second Michael system. The second thiol group then undergoes the next Michael addition reaction under concomitant formation of the three-carbon bridge. Compared to the reported and widely used bissulfones 2a, the allyl sulfones 1a-e lack two hydrophobic benzene groups, which provide low n-octanol-water partition coefficients (log P o/w ) indicating improved water-solubility (Scheme 1). They also offer more effective disulde rebridging, since bis-sulfones 2 require in situ activation to form the reactive mono-sulfones 3 and this equilibrium step is usually not quantitative (Scheme 1B).
Allyl sulfones 1a-e were readily obtained following a convenient three step reaction sequence (Scheme 1C). A short triethylene glycol or hexaethylene glycol chain was monosubstituted with the desired functionality and then condensed with methacrylchloride to form the corresponding methacrylate 5. Allyl sulfones 1a-e were synthesized via a tandem iodosulfonylation-dehydroiodination reaction. 27 This one-pot reaction rst generated the intermediate vinyl sulfones 6, which isomerized to the thermodynamically more stable allyl sulfones 1a-e under reux and basic conditions. 27 Ethynyl-and aminoallyl sulfones 1a and 1c are attractive building blocks for generating further rebridging reagents with the desired functionalities by Cu(I) catalyzed cycloaddition and condensation reactions, respectively.

Site-directed functionalization of somatostatin, bovine insulin and lysozyme
Somatostatin, bovine insulin and lysozyme were selected for site-directed modications with allyl sulfones, which contain one, three and four disulde bridges, respectively. Disulde rebridging of the peptide hormone somatostatin (SST) was accomplished by allyl sulfone 1a and bis-sulfone 2a in aqueous and organic/aqueous solutions, respectively (Fig. 1). Disulde rebridging using bis-sulfone 2a required at least 40% acetonitrile (ACN), whereas the reaction with allyl sulfone 1a proceeded in aqueous media (Fig. 1B). The corresponding products 7a and 7b were puried by HPLC in 49% and 41% yield, respectively, and characterized via HR-MALDI-TOF MS (Fig. S23 †). The ester linkage of 7a showed high stability to hydrolysis at physiologically pH (pH 6-8, Insulin is a polypeptide hormone excreted by the pancreas, which plays a crucial role in carbohydrate and fat metabolism and it offers three disuldes. 28 Very recently, Loh et al. have demonstrated selective cysteine modication on reduced bovine insulin using allenamides in ammonium carbonate buffer containing 33% THF, resulting in two separate, fully modied chains A and B. 29 Herein, allyl sulfone 1d selectively "reannealed" the interchain disulde bond of bovine insulin in aqueous buffer as shown in Fig. 2. Bovine insulin was treated with 1.2 equiv. of the mild reduction reagent tris(2-carboxyethyl) phosphine (TCEP) and 2 equiv. of biotin-allyl sulfone 1d sequentially at pH 7.8 and the resulting reaction mixture was incubated at RT for 24 h (Fig. 2A) (Fig. 2C, le). The average mass (Dm/z 399 relative to bovine insulin) matched the calculated M.W. of 6132.7. Aer in-solution digestion of BTinsulin 8 by chymotrypsin, peptide fragments were obtained and analyzed by LC-MS 2 (Fig. 3A). The modication site was determined as the C20 (chain A)-C19 (chain B), which is consistent with the higher surface accessibility calculation implemented in the soware package Molecular Operating Environment (MOE, Fig. S4 †). 30 Due to the improved solubility and reactivity of allyl sulfones, more than one disulde modi-cation could be accomplished if desired. By applying the reducing agent and 1a in excess to bovine insulin, the attachment of up to three bioconjugation reagents to different disul-des has been detected suggesting that even a step-by-step rebridging could be achieved, which paves the way to dual or even higher modied insulin derivatives (Fig. S25 †).
We further tested allyl sulfone 1b for the site-specic modi-cation of the enzyme lysozyme (from hen egg white). Lysozyme (1,4-b-N-acetylmuramidase) hydrolyzes the b (l-4) glycosidic bonds, and induces bacterial cell death through lysis of the cell walls. It has received great interest for various applications in medicine, cosmetics and food industry due to its anti-bactericidal activity. Moreover, it is a disulde rich protein, contains four disulde bonds and the disulde C6-C127 is predicted to be most solvent accessible based on the calculation of MOE (Fig. S5 †). Lysozyme was treated with 1.05 equiv. of TCEP followed by 2 equiv. of allyl sulfone 1b and the resulting reaction mixture was incubated at RT for 24 h (Fig. 2B). The product coumarin-lysozyme 9 (C-Lyso) was isolated in 19% yield by FPLC (ÄKTA from GE) using a Hi Trap phenyl HP column (1 mL, from GE) based on the hydrophobicity. Noteworthy, 30% of native lysozyme was recovered during the FPLC purication, which was again recycled for further functionalization. Protein modication oen proceeds in low yields and without further purication. Herein, the yield of modied lysozyme was assessed aer purication. In comparison, the only unpaired and accessible cysteine residues of HSA and BSA, 31 (Fig. 2C, right) (Fig. S6 †). HR-ESI-MS also revealed a decarbonylation fragment with mass signals at 1731.76513 (calcd mass: 1731.77504), due to decarbonylation from the pyrone ring of the coumarin to form a benzofuran ring, as reported before. 32 The coumarin modied peptide fragment was then sequenced by LC-MS 2 , which again demonstrated modication of disulde C6-C127 (Fig. 3B). C-Lyso 9 has similar circular dichroism (CD) spectra as native lysozyme, indicating that the secondary and the tertiary structures of lysozyme remained unchanged aer modication (Fig. S7 †). Finally, the catalytic activities of C-Lyso 9 and native lysozyme were recorded applying a lysozyme activity kit (Sigma Aldrich, Cat. no. LY0100) according to the manufacturer's instructions.
The catalytic activities of C-Lyso and lysozyme have been determined as 367.7 AE 16.5 units per nmol and 412.7 AE 21.0 units per nmol, respectively, demonstrating that lysozyme's functional activity was retained aer modication (Fig. S8,  Table 1 †). These results underline that disulde rebridging of proteins via allyl sulfones 1 proceeds with high selectivity, efficiency, under mild aqueous conditions and without perturbation of the tertiary structure and function of lysozyme.

Allyl sulfones as reactive sites facilitating protein multifunctionalization
In the next step, we have explored allyl sulfones 1 as versatile building blocks for convenient preparation of more complex bioconjugates containing up to three different functionalities (Fig. 4A). Bis-sulfones react with thiols at pH 8 by in situ-elimination of p-toluene sulnic acid yielding a monosulfone. 21 In contrast, allyl sulfones undergo a Michael reaction with thiols already at pH 6 and at pH 8 in a sequential fashion (Fig. 4B), which was monitored by LC-MS. First, allyl sulfone 1a reacted with 2 equiv. cysteine at pH 6 forming the mono-cysteine adduct 10. This reaction was already completed aer only 15 min reaction time (Fig. S9 †). Only minor traces of the p-toluene elimination product 11 and no bis-cysteine adduct 17 were detected within 1 h. However, at pH 8 and aer 1 h incubation, the elimination product 11 and the bis-cysteine adduct 17 were formed as major products (Fig. S9 †). In the second step, the mono-cysteine adduct 10 was further reacted with 10 equiv. of glutathione (GSH) at pH 6 and pH 8. At pH 8, the reaction with GSH yielded product 12 almost quantitatively, while at pH 6, no reaction occurred even aer 24 h of incubation (Fig. S10 †). Therefore, optimal conjugation conditions require reaction with the rst thiol containing molecule at pH 6 for about 1 h reaction time followed by the addition of the second thiol containing molecule at pH 8 and incubation for 24 h.
Allyl sulfone 1e has been used to derivatize the single cysteine at the recombinant green uorescent protein (SH-GFP) with a single biotin tag and a uorescent probe (Fig. 4C). Such modications are attractive since they combine purication via affinity chromatography (e.g. avidin beads) and a uorescence tag for detection. Compound 1e was treated with 1.5 equiv. of SH-biotin (28) at pH 6 for 1 h and the product was puried by column chromatography. The resultant products were characterized by LC-MS. Elimination of p-toluene sulnic acid already occurred during column chromatography, resulting in the detection of the major product 13 (m/z ¼ 617 [M + 2H] 2+ ) and very minor quantities of product 14 (m/z ¼ 695 [M + 2H] 2+ ) (Fig. S11 †). Both 13 and 14 (20 equiv.) were further reacted with SH-GFP at pH 8 overnight at RT. The modied GFP 15 was puried by size exclusion chromatography using a Sepharose G-25 matrix and Milli-Q water as eluting solvent in 90% yield. Successful conjugation was demonstrated by gel electrophoresis, optical spectra, and the detection of biotin via western blotting (Fig. S12 †). Gel electrophoresis analysis proved the formation of conjugate 15 due to the intensive uorescent band at 30-31 kDa corresponding well to the M.W. of 30 511 g mol À1 (Fig. S12A †). Excitation of the conjugate 15 at 492 nm revealed a decrease of GFP uorescence at 512 nm and an increase of the lissamine rhodamine B (Rho B) emission at 594 nm, indicating energy transfer from GFP to Rho B (Fig. S12C †) due to the presence of both chromophores within one molecule. Based on the absorption spectra and extinction coefficients, the labeling efficiency of 55% of the Rho B on SH-GFP was calculated (ESI †). The labeling of biotin on GFP was veried by western blotting (Fig. S12B †). To conrm that the biotin group also has similar labeling efficiency as Rho B, the conjugate 15 was incubated with streptavidin agarose and 91% of the conjugate was immobilized on the streptavidin agarose (ESI †). In contrast, no conjugation occurred at pH 6 according to SDS-PAGE, optical spectra and western blotting (Fig. S12 †).

Conclusions
The convenient three-step synthesis of water-soluble allyl sulfones yields versatile and efficient reagents for the spatially dened modication of bioactive cyclic peptides and proteins under mild, aqueous conditions preserving tertiary structure and function. The reaction could even be conducted at low temperature, which is particularly important for many temperaturesensitive proteins such as antibodies. Compared to known bissulfones, disulde rebridging using allyl sulfones proceeds more efficiently since in situ activation is not required anymore. The improved solubility and reactivity of allyl sulfones facilitated the site-specic disulde functionalization of native proteins insulin and lysozyme for the rst time. Moreover, allyl sulfones offer the unique opportunity to conjugate two additional thiol containing molecules to the thus modied protein in a step-wise fashion by simply increasing the pH from 6 to 8, which is attractive for imparting additional functions e.g. uorescence and purication tags. As proof of concept, GFP was equipped with a single Rho B chromophore and a biotin affinity tag at a single site. We believe that our results are of great signicance for the construction of precise multifunctional peptide and protein conjugates, which is of emerging importance to achieve detectable protein therapeutics with e.g. improved pharmacokinetic parameters.