Defining the Scope of the Acid-Catalyzed Glycosidation of Glycosyl Bromides

Abstract

Following the recent discovery that traditional silver(I) oxide-promoted glycosidations of glycosyl bromides (Koenigs–Knorr reaction) can be greatly accelerated in the presence of catalytic TMSOTf, reported herein is a dedicated study of all major aspects of this reaction. A thorough investigation of numerous silver salts and careful refinement of the reaction conditions led to an improved mechanistic understanding. This, in turn, led to a significant reduction in the amount of silver salt required for these glycosylations. The progress of this reaction can be monitored by naked eye, and the completion of the reaction can be judged by the disappearance of characteristic dark color of Ag₂O. Further evidence on higher reactivity of benzoylated a-bromides in comparison to that of their benzylated counterparts has been acquired.

Introduction

Since first chemical O-glycosylations performed by Michael,^[1] carbohydrate synthesis has evolved into a heated area of interest to the organic community due to the biological importance of glycans and their limited availability from natural sources. In 1901, Michael’s glycosidation of glycosyl halides with phenolate was extended by Koenigs and Knorr to regular alcohols.^[2] Silver carbonate (Ag₂CO₃) was added to these reactions with the primary intention to scavenge hydrogen halide, produced during the reaction. Over the course of extended studies over the following three decades, Helferich and others began to realize that rather than having a sole purpose of acid scavengers, silver salts are able to promote the reaction by helping cleave the anomeric carbon-halogen bond.^[3] Traditional Koenigs-Knorr glycosylations are slow, and in subsequent decades, enhanced understanding of the reaction helped to evolve it into one of the most popular methods for glycosylation.^[4] We now know that glycosidation of halides can be significantly accelerated by using other metal salt-based promoters including more reactive silver salts,^[5] mercury,^[3b–d,6] cadmium,^[7] zinc,^[8] indium,^[9] tin,^[10] iron,^[11] etc.^[12]

The field has also seen advancement in the synthesis of glycosyl halides. For instance, per-acetylated^[13] and per-benzoylated bromides^[13e,14] are commonly obtained from the corresponding glycosyl esters by reaction with HBr. Thioglycoside precursors can also be readily converted in the corresponding bromides by reaction with bromine.^[15] This protocol is commonly used to obtain perbenzylated or other unstable bromides. More recently, other types of promoters that do not rely on the use of metal salts have been introduced. Selected examples include halogen or halonium ion with or without additives,^[16] organocatalysts,^[17] diarylborinic acid,^[18] glycosylation modulators such as pyridine,^[19] DMF,^[20] 3,3-difluoroxindole (HOFox),^[15d,21] phenanthroline,^[22] and super critical CO₂.^[23] However, many of these reactions still suffer from low rates and some have a limited substrate scope. Some approaches require multistep synthesis of specialized activators or demand excess of toxic reagents. Recently, we reported that traditional silver(I) oxide-promoted glycosylations can be greatly accelerated in the presence of catalytic TMSOTf.^[24] These reaction conditions are very mild and allow for maintaining neutral pH while providing high rates and excellent glycosylation yields (Scheme 1). Presented herein is our in-depth study of this acid-catalyzed Koenigs–Knorr glycosylation reaction.

Scheme 1.

Cooperative activation of glycosyl bromides with Ag₂O and TMSOTf.

Results and Discussion

Previously we demonstrated that glycosidation of per-benzoylated mannosyl bromide donor 1^[25] with glycosyl acceptor 2^[26] produces disaccharide 3^[27] within 10 min in 99% (Table 1, entry 1). This reaction was performed in the presence of 2.0 equiv Ag₂O and 0.20 equiv of TMSOTf at O °C in DCM. We chose these conditions as a benchmark for further refinement of this promising reaction in application to a broad range of substrates. We began our experimentation by screening various silver salts keeping all other parameters constant. Glycosylation reaction between mannosyl donor 1 and primary 6-OH glycosyl acceptor 2 in presence of silver toluenesulfonate (AgOTs) and TMSOTf produced disaccharide 3 in only 3% yield (entry 2). Despite all our attempts to improve the outcome of this reaction, we could not improve the yield beyond modest 50% yield (see the Supporting Information for further details). Silver methanesulfonate (AgOMs) afforded disaccharide 3 in 58% yield, but required 90 min to get to this point (entry 3). Glycosylation reaction with silver sulfate (Ag₂SO₄) provided disaccharide 3 in 99% yield, but required 90 min to complete (entry 4).

Table 1.

Activation of mannosyl bromide 1 in presence of different silver salts and catalytic TMSOTf.

Entry	Silver salt	Time	Yield of 3
1	Ag2O	10 min	99%
2	AgOTs	30 min	3%
3	AgOMs	90 min	58%
4	Ag2SO4	90 min	99%
5	AgCO2CF3	10 min	5%
6	AgBF4	20 min	78%
7	AgOCN	10 min	96%
8	AgOAc	10 min	80%
9	AgOBz	10 min	97%
10	AgNO3	10 min	< 2%
11	Ag2CO3	10 min	99%
12	AgOTf (no additive)	10 min	83%

Silver trifluoroacetate (AgCO₂CF₃) produced only 5% of disaccharide 3, along with substantial amounts of the hydrolyzed donor (entry 5). This rapid hydrolysis has possibly occurred via the intermediacy of glycosyl trifluoroacetate that is unstable under these reaction conditions. Glycosidations of mannosyl donor 1 in the presence of silver tetrafluoroborate (AgBF₄) or silver cyanate (AgOCN) afforded disaccharide 3 in good to excellent yields of 78 or 96%, respectively (entries 6 and 7). An efficient activation was also achieved in the presence of silver acetate (AgOAc) or silver benzoate (AgOBz). These reactions afforded disaccharide 3 in 80% or 97% yield, respectively (entries 8 and 9). Apart from the disaccharide, the corresponding anomeric acetate (30%, entry 8) and benzoate (8%, entry 9, yields based on donor 1) side products were also isolated from these reactions. Glycosylations in the presence of silver nitrate (AgNO₃), performed with and without TMSOTf, produced mannosyl nitrate,^[28a] along with the trace amount of disaccharide 3 (entry 10). Glycosylation reaction in the presence of silver carbonate (Ag₂CO₃) produced disaccharide 3 in 99% yield in 10 min (entry 11). Reaction in the presence of silver trifluoromethanesulfonate (AgOTf), a known effective activator of glycosyl halides, produced disaccharide 3 in the absence of TMSOTf in 83% yield within 10 min (entry 12). It should be noted that AgOTf-promoted glycosylation was accompanied by the formation of several unidentifiable by-products. Overall, silver salts investigation revealed that many silver salts are capable of co-promoting TMSOTf- catalyzed reaction. Among these, Ag₂O, Ag₂SO₄ and Ag₂CO₃ are most suitable silver salts that activate mannosyl bromide 1 rapidly and efficiently. Other important traits of these salts are their high stability, relative insensitivity to moisture, and relatively low cost.

Encouraged by excellent outcome of a number of reactions surveyed in Table 1, entries 1, 4 and 11 in particular, investigation of the role of different Lewis and Bronsted acids as co-promoters for this reaction was undertaken. Ag₂O was chosen to remain as the benchmark silver salt for this study. As previously mentioned, standard TMSOTf-co-catalyzed glycosidation of donor 1 with acceptor 2 afforded disaccharide 3 in 99% yield in 10 min (Table 2, entry 1). Replacing TMSOTf with 0.20 equiv of MeOTf did not give satisfactory results: the glycosylation proceeded slowly and afforded disaccharide 3 in only 21% yield (entry 2). Increasing the amount of MeOTf to 1.0 equiv was sufficient to obtain disaccharide 3 in 95% yield, but this reaction required 24 h to complete (entry 2). In this context, 1.0 equiv MeOTf-promoted glycosylation in the absence of Ag₂O produced only trace amount of disaccharide 3 after 24 h (see the Supporting Information for complete experimental data). Glycosylation conducted in the presence of TESOTf had a comparable rate to that of TMSOTf during the first 10 min. However, the reaction showed no further progress, and when it was stopped after 1 h disaccharide 3 was obtained in only 64% yield (entry 3). This was explained by rapid consumption of TESOTf to form 6-O-TES protected side product from acceptor 2. An efficient activation of donor 1 was observed in the presence of catalytic AgOTf (0.20 equiv) additive. As a result, disaccharide 3 was obtained in 99% yield in 5 min (entry 4). For comparison, AgOTf (0.20 equiv) in the absence of Ag₂O produced disaccharide 3 in only 8% yield (see the SI). An efficient albeit slower activation of mannosyl bromide 1 was achieved in the presence of bismuth(III) triflate additive. This reaction afforded disaccharide 3 in 83% yield in 5 h (entry 5). When BF₃-Et₂O, a non-triflate based Lewis acid, was added to the Ag₂O-promoted glycosylation of bromide donor 1, disaccharide 3 was produced in 75% yield in 15 min (entry 6). For comparison, activation with BF₃-Et₂O (20 to 100 mol%) in the absence of Ag₂O produced no disaccharide (see the SI).

Table 2.

Investigation of different Lewis/Bronsted acids as co-promoters for the Ag₂O-mediated activation of mannosyl bromide 1.

Entry	Additive [equiv]	Temp, Time	Yield of 3
1	TMSOTf (0.20)	0°C, 10 min	99%
2	MeOTf (0.20 to 1.0)	0°C-r.t., 24 h	21 to 95%
3	TESOTf (0.20)	0°C-r.t., 15 min	64%
4	AgOTf (0.20)	0°C, 5 min	99%
5	Bi(OTf)3 (0.20)	0°C-r.t., 5 h	83%
6	BF3-Et2O (0.20 to 0.40)	0°C, 15 min	75%
7	MsOH (0.20 to 1.0)	0°C-r.t., 30 h	9%
8	H2SO4 (1.0)	0°C, 1.5 h	48%
9	HClO4 (1.0)	0°C, 5 min	89%
10	TfOH (0.20)	0°C 5 min	99%
11	Tf2O (0.20)	0°C-r.t., 60 h	63 %

Having screened a variety of Lewis acid additives, we turned to investigating Ag₂O-promoted glycosylations in the presence of Bronsted acid additives. Activation of mannosyl bromide 1 with Ag₂O in the presence of 0.20 equiv methanesulfonic acid (MsOH) produced disaccharide 3 in 9% yield, even after 30 h (entry 7). Increasing MsOH to 1.0 equiv did not show any improvement. Glycosylations in presence of 0.20 equiv of stronger acids such as sulfuric acid (H₂SO₄) and perchloric acid (HClO₄) were also sluggish. However, in the presence of stoichiometric amounts of these acids disaccharide 3 was produced in 48 and 89% yield, respectively (entries 8 and 9). A modest yield of disaccharide 3 obtained in presence of H₂SO₄ was due to excessive hydrolysis of donor 1. In contrast, trifluoromethanesulfonic acid (TfOH) turned out to be an excellent co-promoter, ever in catalytic amounts. The TfOH-catalyzed reactions were much cleaner than those in the presence of all other protic acids, on a par with TMSOTf. As a result, disaccharide 3 was obtained in 99% yield in 5 min (entry 10). In contrast, glycosylation in the presence of triflic anhydride (Tf₂O) was very slow. Disaccharide 3 was obtained in a moderate yield of 63% after 60 h (entry 11). Overall, the investigation of acidic co-promoters revealed that catalytic TMSOTf, TfOH or AgOTf rapidly (5–10 min) produce disaccharide 3 in nearly quantitative yields. Complete details of several other Lewis/Bronsted acid-catalyzed glycosylations are available as a part of the SI.

Followed the promoter system optimization, we turned to investigating the effect of the reaction solvent. The most promising silver salt promoters Ag₂O, Ag₂SO₄ and Ag₂CO₃ along with TfOH co-catalyst were selected for this study. All glycosylation reactions in DCM or 1,2-DCE afforded disaccharide 3 in excellent yields of 94–99% (Table 3, entries 1, 2). Glycosylation reactions in diethyl ether provided disaccharide 3 in moderate to good yields of 63–90%, but the reactions required 4–24 h to complete (entry 3). A similar reactivity trend was observed for reactions in 1,4-dioxane although the yields of disaccharide 3 were even lower (entry 4). Glycosylations in THF produced only trace amounts of disaccharide 3 (entry 5) due to incompatibility of THF that presumably underwent ring opening polymerization under these reaction conditions.

Table 3.

Investigation of the reaction solvents for the silver salt/TfOH co-promoted glycosylations.

Entry	Solvent	Silver salt	Temp, Time	Yield of 3
1	DCM	Ag2O	0°C, 5 min	99%
		Ag2CO3	0°C, 10 min	99%
		Ag2SO4	0°C, 1.5 h	94%
2	DCE	Ag2O	0°C, 10 min	99%
		Ag2CO3	0°C, 10 min	95%
		Ag2SO4	0°C, 30 min	99%
3	Et2O	Ag2O	0°C-r.t., 5 h	63%
		Ag2CO3	0°C-r.t.,4h	79%
		Ag2SO4	0°C-r.t., 24 h	90%
4	dioxane	all	r.t., 4–20 h	19–63%
5	THF	all	0°C-r.t., 20 h	< 5%
6	MeCN	all	0°C-r.t., 5–30 h	24–44%
7[a]	acetone	all	0°C-r.t., 2–24 h	51–64%
8	toluene	Ag2O	0°C, 10 min	99%
		Ag2CO3	0°C, 10 min	99%
		Ag2SO4	0°C, 5 min	81 %
9	benzene	Ag2O	0°C, 10 min	99%
		Ag2CO3	0°C, 10 min	98%
		Ag2SO4	0°C, 5 min	91 %
10	hexanes	all	0°C-r.t., 20 h	< 5%

^[a]A solvent insertion product was obtained instead (see the Supporting Information for details).

As evident from results surveyed in Table 3 (entries 1–3), it became apparent that the rate of glycosylations in presence of Ag₂SO₄ are somewhat slower for all investigated solvents. This was attributed to the lower solubility of Ag₂SO₄. With the anticipation of the beneficial effect of polar solvents on the rates of Ag₂SO₄-mediated reactions we decided to investigate reactions in acetonitrile. Somewhat unexpectedly, very slow reaction rates have been encountered for all silver salt promoted glycosylations in acetonitrile. As a result, disaccharide 3 was produced in poor yields of 24–44%, even in prolonged experiments, up to 30 h (entry 6). Similar reaction rates have been obtained from glycosylations in acetone (entry 7). In this case, however, a solvent insertion product^[28b] was obtained instead of disaccharide 3 (see the Supporting Information for details).

To rationalize the reduction in reaction rates in ethereal and other polar solvents, acetone and acetonitrile, we decided to look at possible differences in the mechanistic pathways for this reaction in different classes of solvents. Previously, we suggested the reaction mechanism by which this cooperative catalysis proceeds (Scheme 2).^[24] Silver is halophilic, therefore, the process originates by bromide complexation with Ag₂O resulting in intermediate A that also forms in the classic Koenigs–Knorr reaction. Intermediate A will ultimately dissociate, but Ag₂O is too weak a promoter to effectively pull the leaving group and rapidly pass the energy barrier required for the dissociation rate determining step (RDS) to take place. When catalytic additive is available (TMSOTf in our previous study), a strongly ionized species B are formed. In non-polar solvents, DCM in our previous study, intermediate B will readily dissociate leading to the production of glycosyl cation C that subsequently reacts with the acceptor (ROH). The leaving group departure is the first irreversible step in this pathway due to the precipitation of AgBr. We anticipate that Ag₂O and other identified silver salts Ag₂SO₄ and Ag₂CO₃ act similarly. Also, the effect of Lewis or protic acid, such as newly identified TfOH, is similar to that described for TMSOTf. However, this mechanistic pathway may differ dramatically when the reaction is performed in a polar ethereal or nitrilic solvents.

Scheme 2.

Previously suggested mechanistic pathway for reactions in non-polar solvents (DCM).

The reaction in polar solvents still begins by bromide complexation with a silver salt (AgOX) resulting in intermediate A’ (Scheme 3). In the presence of a catalytic acidic additive (EY) strongly ionized species B’ are formed. In polar solvents, intermediate B’ becomes solvated (D), which can pull the promoter away from the leaving group (E) hence slowing the association of the silver ion with the anomeric bromide, necessary for the irreversible departure of the leaving group. More ionized silver salts and reaction intermediates will be solvated stronger, which could be the reason for a greater reduction in glycosylation rates observed for Ag₂SO₄ in all polar solvents. Glycosylcation C can also interact with nucleophilic ethereal and nitrilic solvents via participation leaving to the anomeric adducts. Such adducts may play critical roles in affecting the stereoselectivity of glycosylations.^[4,29] However, we believe that the formation of these reaction intermediates have no effect on the reduction in the rate of reaction, because their formation would not have a detrimental effect on the rate-determining step.

Scheme 3.

Plausible reaction pathway in polar solvents.

To support this general hypothesis, non-polar solvents such as toluene, benzene, and hexanes were investigated. A very rapid activation of mannosyl donor 1 in toluene was observed in presence of all silver salts, and disaccharide 3 was obtained in 81–99% yield in 5–10 min (Table 3, entry 8). Interestingly, activation with Ag₂SO₄ was even faster than with other silver salts. A similar reactivity trend was observed in benzene, and disaccharide 3 was obtained in 91–99% yield in 5–10 min (entry 9). Again, reaction in the presence of Ag₂SO₄ was faster than with other silver salts, and slightly reduced yields were due to the formation of glycosyl sulfate by-product. Practically no reaction took place in hexanes most likely due to poor solubility of glycosyl donor and acceptor (entry 10). Overall, the solvent study demonstrated that DCM, 1,2-DCE or toluene are the most suitable reaction solvents for this glycosylation reaction. It has also been observed that the reactivity of silver salts can be modulated in different solvents. For instance, glycosylations in the presence of Ag₂SO₄ in polar solvents was the slowest, which was reversed in toluene and benzene.

Turning back to the general reaction mechanism depicted in Scheme 2, glycosylation of intermediate C and subsequent deprotonation produce TfOH. The latter reacts with AgOTMS, released during the RDS along with AgBr, to produce TMSOTf that becomes available for the next catalytic cycle for the activation of complex A. Also generated is unstable AgOH that undergoes the loss of water, scavenged by the molecular sieves (MS), and contributes to the regeneration of Ag₂O and helps to maintain the overall neutral pH of the reaction medium. The proposed regeneration cycle of Ag₂O implies that excess silver salt, 2.0 or even 3.0 equiv used in our previous study, may not be necessary.^[24] To investigate whether a successful glycosylation can be performed with as little as 0.50 equiv of Ag₂O that is needed to produce 1 equiv AgBr, we reacted mannosyl bromide 1 with acceptor 2 in presence of Ag₂CO₃ (0.50 equiv) and TMSOTf (0.20 equiv) in DCM. Disappointingly, this reaction produced only trace amounts of disaccharide 3 (<5%).

After careful refinement of conditions, replacement of reagents and optimizing their ratios (see the Supporting Information for complete details) we discovered that glycosylations with donor 1 indeed can be conducted in the presence of only 0.50 equiv of silver salts. Thus, Ag₂CO₃ was replaced with Ag₂O, TMSOTf was replaced with TfOH to streamline the Ag₂O regeneration cycle because protic acid-catalyzed reactions produce AgOH and TfOH directly. With these adjustments we determined that 0.50 equiv of Ag₂O is indeed sufficient for glycosylations with bromide 1. Thus, when acceptor 2 was glycosylated with donor 1 under the refined conditions in the presence of 0.50 equiv Ag₂O and 0.35 equiv of TfOH in toluene at 0°C disaccharide 3 was obtained in 99% yield in 5 min (Table 4, entry 1). Inspired by this result, we moved forward to investigate the scope of these new reaction conditions. Glycosylation of secondary 2-OH acceptor 4^[26] in the presence of 0.50 equiv Ag₂O in combination with a slightly increased amount of TfOH (0.40 equiv) was also very rapid and clean. The corresponding disaccharide 5^[15d] was obtained in 97% yield in 5 min (entry 2). Glycosylations of 3-OH acceptor 6^[26] and even very unreactive 4-OH acceptor 8^[26] were rapid, efficient, and clean. The respective disaccharides 7^[24] and 9^[30] were afforded in excellent yields of 96–97% in 5 min (entries 3 and 4). Benzylidene protected 3-OH acceptor 10^[31] and diacetone galactose 6-OH acceptor 12 were then investigated. We feared that cyclic acetal and ketal protecting groups could be unstable in the presence of TfOH. However, the reactions in the presence of 0.30–0.35 equiv TfOH were very successful, and the respective disaccharides 11^[24] and 13^[32] were obtained in excellent yields of 92–94% (entries 5 and 6). Glycosidation of donor 1 with electronically deactivated primary 6-OH acceptor 14^[33] and cholesterol 16 afforded disaccharide 15^[34] and glycoside 17^[9b] in 99% and 79% yield, respectively (entries 7 and 8). To further demonstrate the applicability of the new reaction to the synthesis oligosaccharides, we investigated a possibility of double glycosylation using diol acceptors 18^[35] and 20.^[36] These reactions rapidly produced the desired trisaccharide products 19 and 21 in 10 min in 91 and 70% yield, respectively (entries 9 and 10). A lower efficiency in the latter case was attributed to the intermediacy of the 1–6-linked disaccharide, in which the 4-OH is deactivated both sterically by the bulky sugar unit attached to the adjacent primary position and electronically by neighboring benzoyl groups.

Table 4.

Glycosidation of mannosyl bromide 1 with various glycosyl acceptors in the presence of 0.5 equiv of Ag₂O.

It should be noted that Ag₂CO₃ and Ag₂SO₄ generally provided excellent results with all reactive primary glycosyl acceptors. In glycosylations of secondary acceptors, however, complete consumption of glycosyl acceptors was rarely observed that resulted in lower yields, typically in the range of 75–90%. In case of Ag₂CO₃, we noticed competitive hydrolysis due to the generation of water by-product. In case of Ag₂SO₄, we detected the formation of glycosyl sulfate as the side product in reactions in toluene. Further tweaking of the reaction conditions helped us to enhance the utility of these silver salts. For instance, water by-product could be addressed by using additional amounts of molecular sieves. Since Ag₂SO₄ is a very reactive promoter in toluene, mixing it with Ag₂O (0.25 equiv each) was beneficial for suppressing the formation of glycosyl sulfate. This also allowed for the additional reduction of TfOH to 0.30 equiv (see the Supporting Information for complete details of this study).

To further broaden the scope of this reaction, we investigated other series of glycosyl donors using the newly established benchmark: 0.50 equiv of Ag₂O and 0.40 equiv of TfOH as the promoter system in toluene as the reaction solvent. This study is summarized in Table 5. First, per-benzoylated galactosyl bromide 22^[37] was glycosidated with primary acceptors 2, 12 and 14. The respective disaccharides 24–26^[38] were smoothly produced in excellent yields of 95–97% in 5–10 min (entries 1–3). Glycosylation of secondary acceptors 4, 6, and 8 with galactosyl bromide 22 was also very efficient and swift, and the respective disaccharides 27–29^[27,39] were obtained in excellent yields of 93–97% in 5–10 min (entries 4–6). We then switched to investigating glycosidation of per-benzoylated glucosyl donor 23.^[14c,40] The reactivity of glucosyl donor 23 was lower than that of its mannosyl 1 and galactosyl 22 counterparts under these reaction conditions, and this observation was consistent with our previous findings made for reactions in the presence of excess Ag₂O with TMSOTf co-promoter.^[24] Nevertheless, glycosylation of acceptors 2, 4 and 8 with donor 23 proceeded very successfully and afforded the respective disaccharides 30–32^[27,30] in commendable yields of 86–98% after 15–60 min (entries 7–9). The lower yield of 86% in this series was obtained for the glycosylation of the highly hindered 4-OH acceptor 8.

Table 5.

Glycosidation of per-benzoylated galactosyl and glucosyl bromide donors 22 and 23.

Interestingly, it was noticed that glycosylation of glycosyl bromides in the presence of 0.50 equiv Ag₂O proceeded to complete decolorization of the reaction mixture within 2–35 min depending on the reaction. The reaction is very dark at the beginning due to the presence of Ag₂O (Figure 1A). The color gradually changes as Ag₂O is getting consumed. A significant color change is indicative of the reaction is nearing the completion. Very little residual Ag₂O is present at this time point (Figure 1B). Complete decolorization is indicative of the reaction completion, at this time point no Ag₂O (and glycosyl bromide donor) is remaining in the flask because it have been completely converted to AgBr, a pale-yellow solid (Figure 1C). Also refer to a representative 2 min video of a typical experiment available as a part of the Supporting Information. This indicative decolorizations was observed in all reactions performed in the presence of 0.50 equiv Ag₂O, regardless of the amount of TfOH, solvent, and the nature glycosyl acceptors. These variables could affect the time at which the decolorization occurs, but not the phenomenon itself.

Figure 1.

Color change is indicative of the reaction progress and completion (see the Supporting Information for video).

Encouraged by excellent reactivity profile of all per-benzoylated (disarmed) mannosyl, galactosyl and glucosyl bromides with several secondary and primary glycosyl acceptors, an effort was undertaken to glycosylate perbenzylated (armed) glycosyl bromides. While per-benzoylated bromides can be synthesized, crystallized, and stored for prolonged time, perbenzylated bromides are typically generated from thioglycosides directly prior to their application. Following our previously developed procedure, per-benzyalted glycosyl bromides 36–38^[41] were generated from the corresponding ethylthio glycosides 33–35^{[15b,16a,41a,42]} by reaction with Br₂. First test glycosylations of armed glycosyl bromide donors 36–38 were conducted with 2-OH glycosyl acceptor 4 using the newly refined reaction conditions, 0.50 equiv Ag₂O and 0.40 equiv TfOH in toluene. Quite unexpectedly, very sluggish reactions and poor yields of disaccharides have been encountered. Increasing Ag₂O to 1.0 equiv showed only a marginal increase in yields. To eliminate any variables, we repeated this glycosylation with per-benzoylated glycosyl bromides 1, 22 and 23 that were generated from the corresponding thioglycosides directly prior to application. Glycosylation of these bromides with glycosyl acceptor 4 in presence of Ag₂O (0.50 equiv) and TfOH (0.40 equiv) produced only traces of disaccharides (<10%). This gave us an idea that the way bromides are prepared has a dramatic effect on the outcome of glycosylations. Ag₂O is thiophilic and may get partially consumed (poisoned) by the remaining sulfur-containing species that could be remaining after the bromination step. Excess Ag₂O (2–3 equiv) used in our previous study^[24] was sufficient to overcome this issue, but new reactions performed with only 0.50 equiv of Ag₂O resulted in fast consumption of silver that becomes unavailable for mediating the glycosylation.

To troubleshoot this issue, we envisaged two possibilities. First, conduct subsequent search for suitable activators that would not be poisoned by sulfur. Second, perform a work-up (or conduct other suitable procedure) to remove the sulfur species remaining after the bromination step. Unpredictably, and to our delight, we observed that all glycosyl bromides generated in situ could be efficiently activated with Ag₂SO₄ (0.70 equiv) and TfOH (0.40 equiv). Thus, when perbenzylated glycosyl bromides 36–38 generated from the corresponding ethylthio glycosides 33–35 were glycosidated with 2-OH glycosyl acceptor 4 the respective disaccharides 39–41^[11] were obtained in 87–92% yield (Table 6, entries 1–3). We noted a somewhat low reaction rates because all glycosylations with perbenzylated bromides required 8 h to complete under these reaction conditions. For comparison, when glycosylations were repeated with per-benzoylated bromides 1, 22, and 23 generated in situ, the respective disaccharides 5, 27 and 31 were readily obtained in 30–35 min in 85–93% yield (entries 4–6). This finding of faster reaction times with supposedly less reactive, disarmed per-benzoylated bromides contradicts the armed-disarmed theory.^[43] On the other hand, these results reinforce our previous observations that per-benzoylated bromides are more reactive in the acid-catalyzed Koenigs–Knorr glycosylations than their benzylated counterparts.^[24] The glycosylation reactions with perbenzylated bromides gave anomeric mixtures of products because they lack a participating group at C-2.

Table 6.

Glycosidation of glycosyl bromide donors generated from the corresponding thioglycosides.

^[a]reaction was conducted at 0°C for 7 h, and then allowed to warm to r.t. over a period of 1 h.

To broaden the scope of the Ag₂SO₄/TfOH co-promoted glycosidation of glycosyl bromides generated in situ, we also investigated reactions of per-benzoylated glucosyl bromide 23 with several other glycosyl acceptors. Glycosylations of acceptors 4 and 6 in presence of Ag₂SO₄ (0.70 equiv) and TfOH (0.40 equiv) afforded disaccharides 31 and 42^[26] in excellent yields of 94 and 93%, respectively (entry 7). Even sterically hindered 4-OH glycosyl acceptor 8 produced disaccharide 32 in 72% yield after 1.5 h (entry 8). Glycosylation of primary glycosyl acceptors 2, 12 and 14 smoothly produced the respective disaccharides 30,^[30] 43^[44] and 44^[27] in excellent yields of 91–97% (entries 9, 11).

We have also explored a possibility of removing sulfur-containing species after the bromination step. To eliminate a possibility of sulfur containing species being immobilized on molecular sieves, bromination of ethylthio glucosides was conducted in the absence of molecular sieves. Upon completion, the reaction mixture was concentrated and dried in vacuo with the anticipation that volatile thiol and/or (di)sulfide species produced in this reaction would be completely removed. Following this protocol, benzoylated bromides 1, 22 and 23 could be quickly glycosidated with acceptor 4 affording the corresponding disaccharides in 92–94% yield in 10 min. Interestingly, this approach was practically ineffective with perbenzylated bromides 36–38, proving much lower reactivity of these donors under these reaction conditions (see the Supporting Information for complete details).

Conclusions

In conclusion, the present study unveils that apart from silver oxide, other silver salts are suitable promoters for the efficient activation of glycosyl bromides in cooperation with several compatible Lewis/Bronsted acids. The nature of solvents plays a crucial role for the newly standardized cooperative promoter system. While the polar solvents may assist in solubilization of these inorganic salts, solvation of loosely bound silver cations or free silver cations may also occur. This hinders the activation of glycosyl bromides and results in dramatic reduction in reaction rates. In this report, a successful reduction in the amount of silver salt to 0.50 equiv of Ag₂O has been achieved. This led a naked eye visual detection of the progress and completion of the glycosylation reaction. Efficient glycosidations of manno-, galacto- and glucosyl bromides with a series of differently protected primary and secondary acceptors have been performed. An unusual reactivity trend has been observed for bromides of the manno, galacto and gluco series. Thus, the classical disarmed sugars were deemed much more reactive than their armed counter parts. Our lab is currently engaged in studying the activation of other glycosyl halides and also ways of stereocontrolling glycosylation reactions under these novel reaction conditions.

Experimental Section

General glycosylation procedures

Method A.

Glycosidation of per-benzoylated glycosyl bromides in presence of silver salts and acid co-promoters.

A mixture of glycosyl donor (0.045 mmol), glycosyl acceptor (0.036 mmol), and freshly activated molecular sieves (3 Å, 90 mg unless mentioned) in an appropriate solvent (1.0 mL) was stirred under argon for 1 hour at room temperature. A silver salt (0.50–2.0 equiv) was added and the resulting mixture was stirred for 5 minutes at room temperature. The mixture was cooled to 08C, a Lewis or a Brønsted acid (0.20–1.0 equiv) was added and the resulting mixture was stirred vigorously for the time and temperature indicated in Tables. After that, the reaction was quenched with one drop of triethylamine. The solids were filtered off through a pad of Celite and rinsed successively with CH₂Cl₂. The combined filtrate (~40 mL) was washed with water (2×10 mL). The organic phase was separated, dried with Na₂SO₄, and concentrated in vacuo. The residue was purified by column chromatography on silica gel (ethyl acetate—hexane or toluene gradient elution) to afford disaccharides in yields listed in Tables.

Method B.

Glycosidation of glycosyl bromides obtained from the thioglycoside precursor in presence of silver salts and acid co-promoters.

A thioglycoside precursor (0.045–0.051 mmol) and freshly activated molecular sieves (3 Å, 90 mg) in CH₂Cl₂ (1.0 mL) were stirred under argon for 1 h at room temperature. The mixture was cooled to 0°C, Br₂ (1.3 equiv) was added, and the resulting mixture was stirred for 2–5 min. After that, the volatiles were removed under reduced pressure, the residue was co-evaporated with dry toluene (1.0 mL), and dried in vacuo for 1 h. A silver salt (0.50 or 0.70 equiv) and glycosyl acceptor (0.036–0.041 mmol) were added, and the resulting solid was additionally dried in vacuo for 1 h. The required reaction solvent (1.0 mL) was added and the resulting mixture was stirred for 5 min at room temperature. The mixture was cooled to 0°C, triflic acid (0.20–1.0 equiv) was added, and the resulting mixture was stirred vigorously for the time indicated in Tables. The reaction was then quenched with one drop of triethylamine. The solids were filtered off through a pad of Celite and rinsed successively with CH₂Cl₂. The combined filtrate (~40 mL) was washed with water (2×10 mL). The organic phase was separated, dried with Na₂SO₄, and concentrated in vacuo. The residue was purified by column chromatography on silica gel (ethyl acetate—hexane or toluene gradient elution) to afford a glycoside derivative in yields listed in Tables. Anomeric ratios (if applicable) were determined by comparison of the integral intensities of relevant signals in ¹H NMR spectra.