Protecting-group-free S-glycosylation towards thioglycosides and thioglycopeptides in water

Abstract

A facile and green S-glycosylation method has been developed featuring protecting-group-free and proceeding-in-water like enzymatic synthesis. Glycosylation of fluoride donors with thiol sugar acceptors using Ca(OH)₂ as a promoter afforded various thioglycosides in good yields with exclusive stereoselectivity. This method also enabled the successful production of S-linked oligosaccharides and S-linked glycopeptides.

Accessing large numbers of well-defined glycans and glycoconjugates is fundamental for structural and functional studies of glycans, which are crucial in understanding many physiological and pathological processes. Although numerous glycosylation methods have been developed, the assembly of complex glycans remains a time-consuming and labor-intensive process. In addition, the majority of chemical glycosylation methods are unsustainable with the use of organic solvents and environmentally harmful catalysts. The development of more straightforward, affordable, sustainable, and more efficient glycosylation methods is one of the primary goals in carbohydrate chemistry.¹ The idealized strategy would be a glycosylation process that occurs in the complete absence of protecting groups under mild and neutral conditions.² Recently, several facile ways to avoid or decrease protection group manipulations have been developed;³ however, protecting-group-free synthetic glycosylation reactions that produce oligosaccharides under aqueous conditions are still scarce.⁴ In a sustainable approach using water as a solvent, glycosylation should be highly efficient as water is a potentially competitive nucleophile for most activated glycosyl donors.⁵ In addition, glycosyl donors that do not undergo hydrolysis or hydrolyze slowly are required.

Enzymatic synthesis is an attractive alternative to chemical methods. Glycosyltransferases or glycosidases are used to catalyze the formation of glycosides employing pre-fashioned sugar nucleotides or non-reducing sugars as glycosyl donors in buffered water.⁶ Although enzymatic methods are protecting-group-free, highly efficient, selective, and synthetic-applicable enzyme catalysts are limited. Besides, their strict substrate specificities further limit their application. The development of new chemical glycosylation methods that could be conducted in water by avoiding unsustainable solvents and catalysts has been a promising direction.

S-Glycosylation, in which thiol serves as the nucleophile instead of hydroxyl, has drawn significant attention in recent years.⁷ Owing to the similar conformation to that of the natural O-linked counterparts, sulfur analogues could be tolerated by most biological systems and are less susceptible to acid-catalyzed or enzymatic hydrolysis.⁸ Besides, the replacement of ether units by thioethers was observed to enhance biological activities such as antimicrobial and antitumor activities.⁹ Therefore, thioglycosides have great potential to be developed into therapeutics and are gaining increasing interest as targets in the pharmaceutical industry.¹⁰ Importantly, a number of thioether-containing drugs, e.g., clindamycin, retapamulin, singulair, and sugammadex, have been widely used in the clinic.

Generally, the strategy for synthesizing thioglycosides includes the coupling of an appropriately activated glycosyl donor to a thiol acceptor. Various thioligation reactions have been reported to date.^7a,11 However, tedious protection and deprotection steps are usually required to obtain the S-linked glycosides, which again involve the employment of various unsustainable solvents and catalysts. On the other hand, unlike advancements in the enzymatic synthesis of O-linked oligosaccharides, natural glycosyltransferases and glycosidases, or engineered glycosynthases are rarely found to be capable of forming S-linked oligosaccharides. The Withers lab developed mutants of glycosidases that catalyzed the coupling between a thiol acceptor and either a dinitrophenyl donor (Fig. 1a)¹² or a fluoride donor (Fig. 1b)¹³ to generate thioglycosides. Although these reactions are efficient, they pose another challenge of obtaining properly engineered enzymes, and their substrate specificity limits the generation of a wide variety of thioglycosides. Therefore, a simple, efficient, green, and protecting-group-free chemical strategy for S-glycosylation is still highly demanded, ideally in an aqueous solution.

Fig. 1.

Protecting-group-free synthesis of thioglycosides. (a) Enzymatic synthesis of β-linked thioglycosides; (b) enzymatic synthesis of α-linked thioglycosides; and (c) chemical synthesis of thioglycosides reported in this work. DNP = 2,4-dinitrophenyl; pNP = para-nitrophenyl.

In recent years, fluorine and sulfur groups have become crucial components of pharmaceuticals in the medicinal industry.^10a Glycosyl fluorides have displayed relative stability in water and have been widely applied in significant glycosylation reactions.¹⁴ A number of synthetic methods for obtaining glycosyl fluorides and thiol-modified glycans¹⁵ have been developed. Herein, we report a facile and protecting-group-free method towards thioglycosides using fluoride donors and thiol acceptors under basic aqueous conditions (Fig. 1c). This chemical thioligation method provides an easier, more practical, cleaner, and efficient alternative for the synthesis of thioglycosides using renewable water and thereby eliminating organic pollution.

The fluoride-thiol coupling method was evaluated and optimized using coupling glucopyranosyl fluoride 1 and 6-SH modified methyl glucopyranoside 2 (Table 1). Firstly, we tested NaH as a base to promote the coupling reaction, given its standard implementation in the glycosylation of glycosyl halides.¹⁶ Unfortunately, this reaction did not yield any product (entry 1). It was observed that as the reaction proceeded, thiol acceptor 2 was oxidized to a disulfide. In this context, tris(2-carboxyethyl) phosphine (TCEP) was added to act as a reducing reagent in the reaction, but still no product was formed (entry 2). Stronger bases like NaOH or KOH could not promote this coupling reaction as well (entries 3 and 4). We then adopted the condition of Ca(OTf)₂/45% aqueous NMe₃, which was used to achieve the regioselective O-glycosylation of unprotected sucrose or analogues with fluoride donors in an aqueous medium.¹⁷ However, the coupling between 1 and 2 under this condition was unsuccessful (entry 5). Recently, Wadzinski and coworkers reported that fluoride donors could be efficiently coupled to phenols in the presence of Ca(OH)₂ in water.⁵ So, we tested Ca(OH)₂ as a base to promote the coupling reaction between fluoride donor 1 and thiol acceptor 2 in water (entry 6). Surprisingly, the S-linked disaccharide 3 was successfully afforded with nearly complete conversion within 0.5 h. The high yield and fast rate may be attributed to the strong nucleophilicity of thiol and the formation of CaF₂ precipitation, which drive the reaction in the direction of product formation. In addition to Ca(OH)₂, other alkaline earth metal hydroxides, including Sr(OH)₂ (entry 7) and Ba(OH)₂ (entry 8), were tested to promote this coupling reaction as well. Only Sr(OH)₂ was able to promote this coupling reaction in a moderate yield. Although the lanthanide hydroxides, including La(OH)₃, Nd(OH)₃, Sm(OH)₃, Gd(OH)₃, could also drive fluoride precipitation, no corresponding thioglycoside was generated (entries 9–12). Therefore, we reasoned that the basicity of these hydroxides might not be strong enough to promote this coupling reaction. We then evaluated the impact of donor equivalents where the use of 1.5 equiv. resulted in a satisfactory yield of 79% (entry 13) and 2 equiv. of the donor gave a high yield of 92% (entry 14). When 1.5 equiv. of donor 1 was used, the thiol acceptor 2 could not be completely consumed even when the reaction time was extended to up to 3 h. On the other hand, decreasing the concentration of the thiol acceptor in the system from 0.5 M (entry 6) to 0.1 M (entry 15) had a minor influence on the overall coupling yield (98% to 82%). Because the nature of the solvent is known to influence the stereoselectivity and the yield of glycosylation reactions, other solvents were evaluated for the fluoride–thiol coupling, including DMF, DMSO, DMF/H₂O (1/1), and CH₃CN/H₂O (1/1) (entries 17–20). It turned out that the reaction only proceeded well in the CH₃CN/H₂O solvent system besides water. Moreover, a reaction of 5 mmol of 2 (1.05 g) with 15 mmol of 1 (2.73 g) resulted in a 92% yield of 3 (1.71 g), demonstrating the scalability of this reaction.

Table 1.

Optimization of S-glycosylation^a

Entry	1 (equiv.)	Base	Base (equiv.)	Solvent	2 (Conc.)b	Yieldc
1	3	NaH	3	DMF	0.5 M	0
2	3	NaH/TCEP	3	DMF	0.5 M	0
3	3	NaOH	3	H2O	0.5 M	0
4	3	KOH	3	H2O	0.5 M	0
5	3	Ca(OTf)2	3	45% NMe3	0.5 M	0
6	3	Ca(OH)2	3	H2O	0.5 M	98%
7	3	Sr(OH)2	3	H2O	0.5 M	54%
8	3	Ba(OH)2	3	H2O	0.5 M	0
9	3	La(OH)3	3	H2O	0.5 M	0
10	3	Nd(OH)3	3	H2O	0.5 M	0
11	3	Sm(OH)3	3	H2O	0.5 M	0
12	3	Gd(OH)3	3	H2O	0.5 M	0
13	1.5	Ca(OH)2	1.5	H2O	0.5 M	79%
14	2	Ca(OH)2	2	H2O	0.5 M	92%
15	3	Ca(OH)2	3	H2O	0.1 M	82%
16	3	Ca(OH)2	3	H2O	0.2 M	85%
17	3	Ca(OH)2	3	DMF	0.5 M	0
18	3	Ca(OH)2	3	DMSO	0.5 M	0
19	3	Ca(OH)2	3	DMF/H2O	0.5 M	0
20	3	Ca(OH)2	3	CH3CN/H2O	0.5 M	93%

^aThe reaction time is 1 h.

^bThe amount of compound 2 is 0.2 mmol.

^cYield of isolated product 3.

With the optimized glycosylation reaction conditions, we investigated the substrate compatibility of this fluoride–thiol coupling reaction. An array of glycopyranosyl fluorides and thiol acceptors were synthesized (ESI†), thus enabling the synthesis of a wide variety of thioglycosides. As shown in Table 2, most coupling reactions proceeded in high yields giving a series of β-thioglycosides. Notably, S-glycosylation of the hindered secondary glycosyl acceptor 11 with glucopyranosyl fluoride 1 was also efficient, affording compound 12 in a high yield of 91% (entry 5). The yield of the coupling between galactopyranosyl fluoride 13 and methyl 6-SH-glucopyranoside 2 or 4-SH-glucopyranoside 11 was moderate (entries 6 and 7), which may be due to the observed quick decomposition of the galactopyranosyl fluoride compared to other glycopyranosyl fluorides. Interestingly, S-glycosylation of mannopyranosyl fluoride 16 with thiol acceptor 2 gave the α-thioglycoside 17 (entry 8). Glycosylation of lactosyl fluoride 18 with thiol acceptor 2, 4, or 21 gave thiotrisaccharide 19, 20 or thiotetrasaccharide 22 in a high yield demonstrating the applicability of this method in synthesizing thiooligosaccharides (entries 9–11). It is worth mentioning that these thioglycosides could be mimics of their native O-linked sugars, which confer valuable biological potential. For instance, the coupling between fluorogalactose 13 and methyl 4-SH-glucopyranoside 11 yielded the S-lactose analogue 15, which can be used in the enzymatic synthesis of thio-Globo-series glycans (a class of tumor-associated antigens) and human milk oligosaccharides.

Table 2.

S-Glycosylation of various glycopyranosyl fluorides and thiol acceptors^a

Entry	Donor	Acceptor	Product	Yieldb
1				94%
2				92%
3				94%
4				90%
5				91%
6				62%
7				55%
8				83%
9				87%
10				91%
11				80%

^aReaction conditions: acceptor (0.3–0.6 M, 1 equiv.), donor (3 equiv.), Ca(OH)₂ (3 equiv.), H₂O (0.3–0.5 mL), r.t., 1 h.

^bIsolated yield.

Next, we tested an iterative protocol for the synthesis of S-linked oligosaccharides. As shown in Scheme 1, disaccharide 23 was obtained from compound 3 by the sequential 6′-tosylation, peracetylation, 6′-thioacetylation, and deacetylation in a total yield of 80% over four steps. Acetylation and deacetylation were carried out only to improve the quality of compound 23, and these steps could be avoided in a large-scale synthesis. Glycosylation of compound 1 with 23 afforded S-linked trisaccharide 24 in 92% yield, demonstrating the applicability of generating S-linked oligosaccharides using this fluoride–thiol coupling reaction. The reactions employing the protected 6-thiol donors in the iterative glycosylation were sluggish due to their low solubility in water, and therefore, the thio-modifications were done subsequently after the glycosylation.

Scheme 1.

Iterative S-glycosylation.

S-Linked glycopeptides have been pursued owing to their enhanced stability towards enzymatic degradation compared to their O-linked counterparts.¹⁸ The synthesis of S-linked glycopeptides is generally achieved by utilizing glycosylated cysteine (Cys) as a building block with suitable protecting groups on both carbohydrates and amino acids.¹⁹ Here, we demonstrated that this aqueous phase fluoride–thiol coupling strategy is practical and efficient for the direct S-glycosylation of Cys-containing peptides to generate thioglycopeptides in an economical and environmentally friendly way. For example, the tripeptide l-glutathione (l-γ-glutamyl-l-cysteinyl-glycine; GSH) is a common constituent in most animal cells and many bacteria. It participates in various critical physiologic functions, such as signal transduction, immune response, maintenance of cellular osmolality, defense against reactive oxygen species (ROS) and free radicals, and protein folding.²⁰ Our method enabled the direct S-glycosylation of a purified tripeptide 25 with glucopyranosyl fluoride 1, affording the thioglycopeptide 26 in 90% yield (Scheme 2a). Additionally, α-synuclein is an intrinsically disordered soluble protein, and its aggregate is associated with synucleinopathies, including Parkinson’s disease.²¹ O-Glycosylation of α-synuclein at Thr72 has an essential function in preventing protein aggregation.²² Herein, a pentapeptide mimic replacing threonine 72 with Cys was synthesized via SPPS (solid-phase peptide synthesis) using an automated peptide synthesizer (ESI†). The peptide was cleaved from the resin by a cleavage cocktail and successively protected the N-terminus as a BOC amide. The purification of the peptide favored the formation of a disulfide bond. Thus, the crude peptide was directly used in performing S-glycosylation to obtain an α-synuclein based thioglycopeptide 27 in an overall yield of 44% (Scheme 2b). We also synthesized a natural thioglycopeptide fragment of sublancin, which has antimicrobial activities against a couple of Gram-positive bacteria, including methicillin-resistant Staphylococcus aureus (MRSA).²³ It has been shown that the S-linked glycosyl unit on sublancin is essential for antimicrobial activity.²⁴ A crude heptapeptide sequence surrounding the β-linked d-glucopyranose unit in sublancin was synthesized by following the same strategy as used for the pentapeptide synthesis, and then coupling of glucopyranosyl fluoride 1 in the presence of Ca(OH)₂ successfully afforded thioglycopeptide 28 in an overall yield of 39% (Scheme 2c). To the best of our knowledge, this is the first direct chemical S-glycosylation of peptides in water without the use of protecting groups and thus eliminating the excessive use of toxic solvents.

Scheme 2.

Synthesis of thioglycopeptides. TFA = trifluoroacetic acid; TIS = triisopropylsilane; and DODT = 2,2′-(ethylenedioxy)diethanethiol.

Finally, a plausible glycosylation mechanism was proposed as shown in Scheme 3 and was supported by the in situ NMR experiments. For the 1,2-cis-α-fluoride glycoside donors, firstly, fluoride coordinates with calcium as indicated by the chemical shift of lactosyl fluoride 18 in the presence of Ca(OH)₂ in D₂O after 5 min in in situ ¹⁹F NMR experiments (ESI†). Then a thiol attacks the anomeric carbon of the donor from the β side, giving rise to β-thioglycoside via the S_N2 mechanism²⁵ (Scheme 3a). It was also observed that the fluoride donor was hydrolyzed over time. In comparison, for the 1,2-trans-α-fluoride glycoside (for instance, mannosyl fluoride 16), α-thioglycoside was observed rationalizing the involvement of double inversion with an intramolecular attack forming an epoxide-based intermediate prior to glycosylation^4b,5,26 (Scheme 3b). This observation was supported by the instantaneous hydrolysis of mannosyl fluoride 16 with Ca(OH)₂ in D₂O in in situ NMR experiments (ESI†), thus indicating the neighboring 2-axial OH participation. Therefore, the configuration of 2-OH with respect to the anomeric position plays a crucial role in the stereochemical outcome.

Scheme 3.

Proposed mechanisms. (a) Glycosylation of 1,2-cis-α-fluoride donors with thiol acceptor; and (b) glycosylation of 1,2-trans-α-fluoride donors with thiol acceptor.

Conclusions

In summary, a direct and environmentally friendly S-glycosylation method with glycosyl fluoride as a donor and thiol modified methyl glycoside as an acceptor was developed in the presence of Ca(OH)₂ in water at room temperature. Similar to the enzymatic synthesis, the chemical fluoride–thiol coupling reaction proceeds rapidly under mild conditions, producing thioglycosides in high yields with an exclusive stereoselectivity. This method also enables the iterative synthesis of S-linked oligosaccharides and facile synthesis of various thioglycopeptides in water without using protecting groups. Although using more than 1 equivalent of coupling reagent is a drawback in the current approach, Ca(OH)₂ is a more affordable and greener reagent compared to other commonly used glycosylation reagents, and significantly, the coupling reaction eliminates the use of toxic reagents and solvents which makes the whole process environmentally friendly. Moreover, the side product CaF₂ could be easily removed by filtration. Collectively, the fluoride-thiol coupling reaction provides an efficient, protecting-group-free, and sustainable approach to obtain biopharmaceutically useful thioglycosides and thioglycopeptides.