Mechanistic Investigations into the Application of Sulfoxides in Carbohydrate Synthesis

Abstract The utility of sulfoxides in a diverse range of transformations in the field of carbohydrate chemistry has seen rapid growth since the first introduction of a sulfoxide as a glycosyl donor in 1989. Sulfoxides have since developed into more than just anomeric leaving groups, and today have multiple roles in glycosylation reactions. These include as activators for thioglycosides, hemiacetals, and glycals, and as precursors to glycosyl triflates, which are essential for stereoselective β‐mannoside synthesis, and bicyclic sulfonium ions that facilitate the stereoselective synthesis of α‐glycosides. In this review we highlight the mechanistic investigations undertaken in this area, often outlining strategies employed to differentiate between multiple proposed reaction pathways, and how the conclusions of these investigations have and continue to inform upon the development of more efficient transformations in sulfoxide‐based carbohydrate synthesis.


Introduction
The widespread use of sulfoxides in organic chemistry is ar esult of their rich andv aried reactivity [1] showcased by an enviable plethora of reactions. Well-studied examples include the use of dimethyl sulfoxide in the oxidation of alcohols, [2] the activation of sulfoxides in Pummerer-type reactions, [3] and pericyclic reactions of sulfoxides, such as the Mislow-Evans rearrangement. [4] However,f ew fields have benefited more from the diversec hemical capabilities of sulfoxides than modern synthetic carbohydrate chemistry, [5] for which they often play integral roles as leaving groups,o ra sa ctivating agents in high yieldingg lycosylationr eactions. An all-encompassing review of the use of sulfoxidesi nc arbohydrate chemistry has been forsaken here in favour of an in-depth analysis of the elegant mechanistic investigations performed in this area, which have begun to underpin many of the contemporary theories regarding stereoselectivity and efficiency in challenging sulfoxidebased carbohydrate synthesis. Included will be ad iscussion on the use of glycosyl sulfoxides as glycosyld onors, as well as the application of sulfoxide reagents in dehydrative glycosylations, glycal activation and thioglycoside donor activation.

Glycosyl sulfoxides
The use of thioglycoside donors has been widespreads ince their introduction by Ferrier. [6] Then ext substantial step forward in the use of thioglycoside derivativesc ame from Kahne and co-workers [7] who originally developed the concept of using as ulfoxide glycosyl donor after unsuccessful attempts to glycosylate deoxycholic ester derivative 1 (Scheme 1), in which the target axial alcoholi sv ery unreactived ue to 1,3-diaxial steric hindrance. Sulfoxide glycosylation reactions with benzylated donor 2 and deoxycholic ester 1 afforded glycoside 3 in excellent yield, in an umber of differentsolvents (Scheme 1).
Activation of the sulfoxide was achieved with triflic anhydride at À78 8C, and proceeded through putative sulfonium triflate species 4.F urther examples with benzyl and pivaloylprotected donors were also high-yielding, and included the first example of glycosylation of an amide nitrogen atom, using trimethylsilyl acetamide-an early demonstration of the potentialu tility of glycosyl sulfoxides as novel glycosyld onors. Kahne and co-workersn oted the glycosylation of less reactive trimethylsilyl acetamide stalled at À78 8C, but re-initiated between 0 8Cand ambient temperature over 12 h. [7] Having previously demonstrated the reactivity of glycosyl sulfoxides at low temperatures, the authors postulated anyr eactive intermediates present at À78 8Cw ould decompose at highert emperatures. This implied that glycosylation at the highert emperatures occurredv ia an unidentified more stable intermediate. After furtheri nvestigation, this unknown intermediate was subsequently assigned as ag lycosyl sulfenate as the sulfenate 5 and disaccharide 6 were isolated in a2 :1 ratio (Scheme 2) following activation of fucose donor 7 at À60 8C. [8] Application of glycosyl sulfenates as donors had previously been performed at 0 8C; [9] therefore, the isolated glycosyls ulfenate 5 seemed al ikely candidate as ar eactive intermediate in the sulfoxide reactions at higher temperatures. Scheme1.The challenging glycosylation of ad eoxycholic ester is feasible using sulfoxide-based glycosyl donors. DTBMP = 2,6-di-tert-butyl-4-methylpyridine.
Subsequently,f ormation of glycosyl sulfenates from glycosyl sulfoxides was achieved using catalytic triflic anhydride. [8] Based upon this observation am echanism to accountf or formation of both glycosides and glycosyl sulfenates in sulfoxide glycosylations was proposed (Scheme 3). Following these mechanistic insights, Kahne and co-workers developed as trategy to scavenge byproducts in the sulfoxide glycosylationr eaction using 4-allyl-1,2-dimethoxybenzene, [10] an improvement that aided their program of challenging synthetic endeavours including the synthesis of the blood-group antigens, [11] the calicheamicino ligosaccharide [12] and the ciclamycin trisaccharide. [12] Stereoselective synthesis of b-mannopyranosides and a-glucopyranosides While pursuing ar adical-based solution [13] to the ubiquitous problem of stereoselective b-mannopyranoside synthesis, [14] Crich and co-workerss erendipitously uncovered an unappreciated level of complexity in Kahne's sulfoxide glycosylation method. [15] When using benzylidene acetal protected donor 8, Crich observed that the stereoselectivity of the reaction was dependento nt he order of addition of the acceptor and activating agent (Scheme 4). If donor 8 and acceptor 9 were premixed in diethyl ether and then activated with triflic anhydride, a-mannopyranoside 10 a was formed stereoselectively (in situ activation protocol, Scheme 4a). However,w hen the donor 8 was activated with triflic anhydridei nd iethyl ether prior to the addition of the acceptor 9,acomplete reversal in selectivity was observed and b-mannopyranoside 10 b wasf ormed stereoselectively (pre-activation protocol, Scheme 4b).
The utility of this new methodology for direct b-mannopyranoside formationw as demonstrated with an umber of acceptor alcohols. However,i tw as noted that the benzylidene acetal was essential for selectivity.W hen the fully benzylated equivalent donor was used the selectivity of the reaction was reduced significantly (a/b 2:1). The mechanistic rationale deployed to explain theseo bservations involved inferring the presence of ag lycosylt riflate intermediate 11 (Scheme 5). [16] In the proposed mechanism,t he fate of the oxacarbenium ion 12 depends on the order of addition of the reagents. In the absence of the acceptor( pre-activation), ap utative a-glycosyl triflate 11 is formed which reacts with an acceptor alcohol Martin  with inversion of configuration to afford b-mannopyranoside 13.A lternatively,w hen activation occurs in the presence of the acceptora lcohol (in situ activation)t he oxacarbenium ion 12 affords a-mannopyranoside 14.
In this hypothesis, the observed b-selectivity arises from S N 2type attack of the alcohol on the a-triflate species 11 (glycosyl tosylates with similar reactivity had previouslyb een disclosed). [17] This observation was initially substantiated by increased bselectivities (a/b 1:13!1:32) when less bulky O-2-benzyl donor 15 was used in al ess-ionizing dichloromethane solvent. It should also be noted that other groups have established that pre-activation of Crich's benzylidene acetal donors is not necessarily ap rerequisite for b-mannoside selectivity when glycosylations are performed in dichloromethane as opposed to diethyl ether. [18] Subsequente vidence for the existence of a-triflate species came from low-temperature NMR studies of the glycosylation reaction. [19] Using simplified donor 16 the mechanism was probedb ya ctivation at À78 8Cw itht riflic anhydride (Scheme 6). Within acquisition of the 1 HNMR spectrumanew intermediate had formed with ac haracteristic H1 shift of d = 6.20 ppm, and a 13 CNMR C1 shift of d = 104.6 ppm. [17] The intermediate was assigned as glycosyl triflate 17,a nd subsequently afforded b-mannopyranoside 18 on addition of methanol.
Ak ey point established by Crich is the necessity of the benzylidene acetal-protecting group for b-selective mannosylations. [16,19] This is attributed to the increased conformational constraint imposed on the sugar ring by the benzylidene acetal, which disfavours the formation of the half-chair oxacarbenium ion, [20] thus promoting the formation of a transdecalin-like glycosyl triflate intermediate.
An unexpected reversal of stereoselectivity was observed when glycosylation of glucosyl sulfoxide donors was performed. Thea uthors isolated only a-glycosides selectively (Scheme 7b), comparedt om annosyl sulfoxide donors, which afforded b-glycosides selectively (Scheme 7a). [21] The benzylidene acetal protecting group wasa gain ap rerequisite for stereoselectivity (although glycosylations with glucosyls ulfoxide 19 and triflic anhydride afford a-glucosides,b etter yields and selectivities were achieved by activation of thioglucosides with PhSOTf). [22] The authors postulated selectivity arises from reactiono ft he acceptorw ith transientg lycosyl triflates 20 (Scheme8). The www.chemeurj.org mechanistic rationale used for the gluco series differs from that of the manno series, in that the reactive intermediate is b-glucosyl triflate 20 b rathert han a-glucosyl triflate 20 a.A Curtin-Hammett kinetic scheme [23] was invokedt oe xplain selectivity,i nw hich the reaction proceeds through the less stable, and thus more reactive b-glucosyl triflate 20 b.
These initial explorations weref ollowed up with an umber of mechanistic studies on the chemistry of glycosyl sulfoxides and glycosyl triflates. [24] However, until recently there remained ad egree of ambivalence over whether the stereoselective attack on glycosylt riflates trulyp roceeded through an S N 2-like or S N 1-like mechanism. To jettison any ambiguity,C rich retooled two classical approaches fore lucidating chemical reaction kinetics-employing ac ation-clock experiment, [25] and an atural abundancek inetic isotope study [26] to unequivocally prove the reaction proceeds through an S N 2-like mechanism. Crich's cation-clock was developedt od istinguish between different mechanisms by measuring the relative kinetics between a-a nd b-O and b-C-mannopyranosylations and ac ompeting intramolecular cyclisation (Scheme 9). Following triflic anhydride activation of the mannopyranosyl sulfoxide 22,w hich bears ap rospective internal Sakurai nucleophile, am ajor 23 (bface attack affords the 4 C 1 chair conformer) and minor product 24 (a-face attack affords a 1 S 5 twist-boat conformer) were formed. The formation of both products was rationalised by intramolecular attack from either the a-o rb-face of the B 2,5 twist-boat mannosylo xacarbenium ion 25, [27] which exists in equilibriumwith aglycosyl triflate 26.The authors then repeated triflic anhydride activation experiments, but rapidly followed with the addition of increasing quantities of isopropanol as ag lycosyl acceptor. This reaction manifold allowed the quantification of individual mannopyranosyl anomers 27 b and 27 a formation with respect to the intramolecular cyclisation products 23 and 24,asafunction of isopropanol acceptorconcentration. This methodology was also repeated with trimethyl methallylsilane as an external competing C-nucleophile,t o report on the kinetics of C-glycoside formation.
The cation-clock experiment demonstrated firstly that the ratio of formation of b-isopropyl mannoside 27 b to cyclised products increases as isopropanol concentration increases; therefore, the formation of b-O-mannosidesi sf irst order with respectt on ucleophile concentration. Conversely,t he ratios of formation of a-isopropyl mannoside 27 a and b-C-mannoside 28 to cyclised products did not change with increasing nucleophile concentration, and was thus deemedz eroth order overall with respect to nucleophile concentration.
These resultsa re consistentw ith S N 2-like isopropanol attack on an a-mannosylt riflate,o ra na-contact ion pair,i na ccordance with Crich's earlier postulate;t he formationso ft he a-isopropylmannoside 27 a,and b-C-mannoside 28 were consistent with an S N 1-like isopropanol attack on an oxacarbenium ion or as olvent-separated ionp air. [25a] This study was closely followed by ac omplementary measurement of primary kinetic isotope effects (KIEs) using natural abundance of 13 Ca nd very highfield NMR spectroscopy (200 MHz for 13 CNMR) to measuret he formation of a-a nd b-mannopyranosidesa nd a-a nd b-glucopyranosidesv ia transientg lycosyl triflates. [26] Ab iased system facilitated erosion of the natural selectivity of the glycosylation reaction, allowing 13 C-1 signals of both anomeric products to be measured, using the benzylidene acetal carbon as an internal standard (Scheme 10). The ratios calculated were then compared to the same ratio in the glycosyl sulfoxide starting material. The calculated KIEs for the formation of the b-mannopyranosides 29 b, a-a nd b-glucosides 30 b and 30 a were all in the lower range expected for ab imolecular reaction( 1.03-1.08), while the KIE measured for the formationo fa-mannopyranoside 29 a (1.005 AE 0.002) was in the range for au nimolecular reaction( 1.00-1.01). These resultsa gainp rovidedf urther confirmationf or the formation of b-mannopyranosides through an exploded S N 2-like transition state, and a-manno- Our own mechanistic studies in this field of stereoselective glycosylation of glycosyls ulfoxides have been focussedu pon the activation and reactivity of oxathiane-S-oxided onors 33 and 34 (Scheme 11). [28] The trans-decalin motif present in these oxathianes conferred unanticipated stability on aryl sulfonium ions 35 and 36,t ot he extent that their formation could be monitored with NMR at ambient temperature, following triflic anhydride activation in the presence of electron-rich arenes. [28b] All protectedd erivatives of the oxathiane ketal-S-oxide displayedc omplete a-anomeric stereoselectivity,e ven at 50 8C, suggestive of an S N 2-like attack on the aryl sulfonium ion from the a-face. While still highly a-stereoselective, the oxathianeether-S-oxide also afforded b-glycosides, indicative of at least partial S N 1-like attack on an oxacarbeniumi on, and raised the question of whether the exchange of an axial methoxy group for ah ydrogen atom could effect ac hange in the mechanism from stereospecific S N 2-like attack to ah ighly stereoselective S N 1-like attack. However,D FT calculations using model structures indicated that both the oxathiane ketal and ether were equally likelyt or eact by an S N 2-like mechanism, discounting this tantalising proposition. Instead calculations of the relative stabilityo ft he relevant oxacarbenium ion conformers: 4 H 3 38 (S N 1-like attack upon which affords a-glycosides) and 3 H 4 37 (attack upon which affords b-glycosides) indicatei ti sm ore likely the erosion in a-stereoselectivity resultsf rom an increase in the population of 3 H 4 conformers upon removal of the axial methoxyg roup (Scheme 12).

Dehydrative glycosylation
Sulfoxides have also been used as activating agents in glycosylation reactions to facilitate in situ formation of reactive glycosylatings pecies. Gin and co-workersi dentified sulfoxides as the ideal reagents for dehydrative glycosylation of hemiacetal donors. [29] In ar epresentative example, ac ombination of Ph 2 SO and triflic anhydridew as used to pre-activate hemiacetal The first step of the mechanismi sa ssumedt ob ea ctivation of Ph 2 SO by triflic anhydride to give trifloxysulfonium ion 40. This speciesc ould then react with hemiacetal 41 through its S IV centre to afford an oxosulfonium intermediate 42 (Scheme 14 a), or through its S VI centret oa fford glycosyl triflate 43 (Scheme 14 b). The nearq uantitative incorporation of the label into recovered Ph 2 SO (47 AE 5 18 O-incorporation, as 2equiv of Ph 2 SO were used) ruled out the pathway involving glycosylt riflate 43 (Scheme 14 b). 1 HNMR spectroscopy was used to identify the presence of an oxosulfonium triflates pecies and ag lycosyl pyridinium species as reactioni ntermediates. The analogous glycosyl triflate previously synthesised by Crich and co-workers [19] was not observed in the reaction mixture. The authors noted the observed formation of glycosyl pyridinium species does not necessarily imply it is ar eactive intermediate involved in glycoside formation.
Following the initial studies by Gin andc o-workers [29,30] into the use of sulfoxides in dehydrative glycosylations, the method was utilised in variouso ther examples [31] including in the efficient synthesis of sialosides. [32] Sulfoxide covalent catalysis Mechanistic studies into the dehydrative glycosylation (vide supra)s uggested the possibility of using catalytic amounts of Ph 2 SO in the reaction;h owever, attempts to reduce the amount of Ph 2 SO were plagued by self-condensation of the sugar. [30a] To circumvent this problem Gin and co-workers developed ac atalytic protocol using an ucleophilic sulfonate counteranion 44 that reactedt of orm an anomeric sulfonate 45 as a" resting state" for the activated hemiacetal (catalytic cycle, Scheme 15). [33] For the protocol to work catalytically the sulfonate counteranion neededt ob en ucleophilic enough to displace/regenerate the sulfoxide 46,w hile the anomeric sulfonate 45 had to be reactive enough to afford glycosides 47,b ut also stable enough to prevents elf-condensation with the hemiacetal 48. Screening identified dibutyl sulfoxide and diphenyl sulfonic anhydride as the ideal combinationf or glycosyl sulfoxide-based covalentcatalysis (Scheme 16). [33] An elegant and exhaustive labelling study [34] was undertaken to confirm the postulated mechanism, using dynamic 18 O-label monitoring by low-temperature 13 CNMR spectroscopy. [35] Scheme13. Dehydrative glycosylation using Ph 2 SO and triflic anhydride.

Sulfoxide-based activation of glycal donors
Glycal donors 49 had previously been activated in at wo-step procedure using oxidising agent dimethyldioxirane (DMDO) [36] to afford C(2)-hydroxy pyranosides 50.G in and co-workers extended their use of sulfoxides as activating agents to achieve the same goal in ao ne-pot process. [37] The combination of Ph 2 SO and triflic anhydride (2:1 ratio) facilitatedt he formation of 2-hydroxy pyranosides 50 from glycal donors 49, by a complex oxidative mechanism that was thought to proceed via an 1,2-anhydropyranose intermediate 51 (Scheme 17).
The mechanism of the glycosylation reaction was again elegantlyd issected using labelling studies. [38] Transfer of the 18 Ol abel from Ph 2 SO to C(2)-OHw as observed (Scheme 18).
In mechanism a( Scheme 19 a) the glucal donor 54 is activated by diphenylsulfonium ditriflate 55,b efore excess Ph 2 SO reacts with sulfonium species 56 to afford disulfonium species 57.O na ddition of methanol, the s-sulfurane intermediate 58 [39] forms ands ubsequently fragments with expulsion of diphenyl sulfide to afford 1,2-anhydropyranoside 53.T he approacho fd iphenylsulfonium ditriflate 55 to the b-face of the glycal is ultimately responsible for the stereocontrol in the glycosylation reaction. Alternatively,i nm echanism b (Scheme 19 b), the excess Ph 2 SO gives rise to an oxygenbridged disulfonium salt 59.A ttack by the glucal donor 54 at the bridging oxygen atom would afford C-2-oxosulfonium dication 60 (or the analogous pyranosyl triflate 61). On addition of methanol, s-sulfurane intermediate 62 forms and affords 1,2-anhydropyranose 53 by fragmentation. The stereocontrol of the reactioni sn ow governed by approach to the least sterically hindered a-face by oxygen-bridgedd isulfonium salt 59.
The key difference between mechanismsa and bi st hat the oxosulfonium speciesi se ither connected to C-1 (Scheme19a) or C-2 (b). Thisd ifference in connectivity was exploited in order to determine which mechanistic pathway was traversed. [38] When using 13 C-1-labelled glucal donor 63 in a 13 CNMR trackinge xperiment, small perturbationsi ns ignals were measured when the 13 Cl abel was directly connected to an 18 O-label (Scheme 20). [35] Ac omparison of the C-1 signals using unlabelled Ph 2 SO and labelledP h 2 SO (60 % 18 O-incorporation) made it possible to distinguish whether the disulfonium species Scheme19. a) Proposed mechanism for glycal activation, incorporating disulfonium species 57.b)Proposed mechanism for glycal activation, incorporating C-2-oxosulfonium dication 60.
The data from this labelling experiment, therefore, inferred that the reaction proceeded by mechanism a( Scheme19a). Identical experiments using the analogous 13 C-2-labelled glucal also confirmed al ack of connectivity between 13 C-2 and 18 O, therefore discounting mechanism b( Scheme 19 b) as a possibility.

Sulfoxide-based activation of thioglycosides
The combination of sulfoxide reagents and triflic anhydride has also been applied to the activation of thioglycoside donors. In the pursuit of an expedient route to the aforementioned reactiveg lycosyl triflate intermediate 17 (Scheme 6), Crich and co-workersi dentified electrophilicb enzene sulfenyl triflate (PhSOTf) as an effective reagent for the activation of armed and disarmed thioglycosides. [21] In situ generation of PhSOTf (from benzene sulfenyl chloride (PhSCl)a nd silver triflate) and subsequent thioglycoside 66 activation provided access to glycosyl triflates 67 quantitativelya tl ow temperatures. The advantage of this methodo ver the glycosyl sulfoxide approacht og lycosyl triflates 67 is the exclusion of the sulfide oxidation step prior to the final glycosylation reaction(Scheme 21).
The necessary in situ synthesis of PhSOTf, ar esult of its marked reactivity and inherent instability,m ade the process arduous however.T on avigate this problem shelf -stable S-(4-methoxyphenyl) benzenethiosulfinate (MPBT) 68 (Scheme 22) was developed and showedr eactivity in the activation of armed thioglycosides, [40] but lacked potencyi nc ombination with disarmed donors. An alternative shelf-stable sulfinamide (BSP) 69 showedm uch more promise with ar ange of thioglycoside donors and acceptors, examples included glycosylations with primary, secondary and tertiary alcohols, affording glycosides in excellent yields. [41] At estament to the efficacy of the BSP/triflic anhydride activation of thioglycosides is the wealth of examples in the literature[24c ,42].T hese notably include use in ao ne-pot "reactivity-based" synthesis of aF uc-GM 1 oligosaccharide, [43] used with 2,3-oxazolidinone N-acetyl glucosamine donors [44] and the activation of 2-dialkyl phosphate thioglycosided onors. [45] Despite the obvious utility of the activation strategy, attempts to glycosylate unreactive 2,3-carbonate-protected rhamnopyranoside donors were unsuccessful using either MPBT or BSP/triflic anhydride. To solve this problem van der Marel and co-workersi ntuitively [29,37] opted to useacombination of Ph 2 SO/triflic anhydride as ap romoter,a nd discovered an even more potent reagent system for the activation of thioglycoside donors. [46] The replacemento ft he electron-donating piperidine ring in BSP with ac onventional phenyl group presumablyd estabilises the adjacent charge on sulfur,a nd thus increases ther eactivity of the sulfonium species.G lycosylation of disarmed donors proceeded in excellent yields (Scheme23), and selectivities were in line with the proposed formation of glycosyl triflates as intermediate speciesi nt he glycosylation reaction.
Attempts to activate thioglycoside 70 with Ph 2 SO/triflic anhydride or BSP/triflic anhydride in the presence of glycosyl acceptors were unsuccessful as the reactive alcohols equestered the activating sulfonium species to afford proposed byproduct 71 (Scheme 24), [47] reiterating the necessity of preactivation of the donor.Similarly,chemoselective glycosylations were initially plagued by putative transients pecies 72,f ormed on activation of at hiophenyl donor. [46a] Yields werel ow as the www.chemeurj.org disaccharide products formed were activated by sulfonium triflate species 72 and subsequently hydrolysed on workup. Yields could be increased, however,b yt he addition of triethyl phosphite (TEP) as ar eagent to quencht he sulfonium triflate species 72 at low temperature before decomposition could take place. Ar ange of other glycosidict ransformationsh ave also been effected using thioglycosides in combination with Ph 2 SO/triflic anhydride. [48] An impressive example illustrated the advantage of Ph 2 SO over the less reactive BSP in conjunction with triflic anhydride. The former wast he only reagent successful in the glycosylations of 5-N-7-O-oxazinanone-protected sialoside donors, [49] and more conventional peracetylated thiosialoside donors were also efficiently activatedw ith Ph 2 SO/triflic anhydridet oa fford sialosides in excellent yields and a-selectivities, [50] with excessP h 2 SO essential to suppress problematic glycal formation. [51] In this example, the authors observef ormationo fo xosulfoniums alts at low temperature and proposeg lycal formation by elimination of the C-2-oxosulfonium leaving group is reduced in these intermediates.

Stereochemical preferences of glycosyl sulfoxides
Although al ack of detailed studies have been reported on the activation of thioglycosides by sulfonium triflate species, the observations discussed vide supra implied that glycosyl sulfides attack the S IV centre of sulfonium triflate species, or similar reactive intermediates. We providedf urther strong evidence that this is the case and also gained insight into the stereochemicalp references governing glycosyl sulfoxide formation in an ovel transfer sulfoxidation reaction, by once again using the glycosyl oxathiane as as caffold for serendipitous mechanistic explorations. [52] When Ph 2 SO/Tf 2 Oa ctivation of the ring sulfur in the oxathiane 73/74 was attempted, hopefulo fs tereoselective glycosylation, we were insteads urprised to observe stereoselectiveo xidation to the oxathiane-S-oxide 75/76 (Scheme 25). DFT calculations indicated that the most-stable stereoisomer was formed preferentiallyw hen startingf rom both oxathiane ketal 73 and oxathiane ether 74,w hile lowtemperature 1 HNMR also demonstratedt hat the product was formed within minutes at À60 8Ci nt he absence of adventitious water or alcohol. We hypothesised that the reaction must proceedt hrough an ovel sulfoxide transfer mechanism after isotopic labelling studies using Ph 2 S 18 O( 87 %l abelled) unequivocally proved the oxygen in the sulfoxide product originated from Ph 2 SO (Scheme 25).
Further detailed 18 O-isotopic labellings tudies providede vidence for anumber of steps that must occur during the sulfoxidationr eaction, including that the first committed step in the mechanism must be the reaction of the oxathiane sulfur atom with an activated Ph 2 SO species and aP h 2 SO oxygen atom must become covalently bound to the oxathiane sulfur atom. Althoughw ew ere never ablet oo bserve or isolate diphenyl sulfidef rom the sulfoxidation reaction, the quantitative formation of triaryl sulfonium salt 82 (Scheme 26) was confirmed by HPLC mass spectrometric comparison of the crude product mixture with authentic samples of sulfonium salt 82 of known concentration, thus proving diphenyl sulfide must also be produced during the reaction and then react with some activated Ph 2 SO speciest op roduce the triarylsulfonium salt byproduct. Severalm echanistic pathways could be proposed and were www.chemeurj.org consistentw ith these observations( Scheme 26). [52] In the first (Scheme26, a), oxathiane 77 initially attacks an electrophilic oxygen atom in triflyloxy sulfoniumi on 55 to produce activated oxathiane 78 and diphenyl sulfide. Activated oxathiane 78 could then react with the excess Ph 2 SO to provide oxodisulfonium ion 79.S imilarly, 79 could also be formed by an alternative pathway (b) which also involves reactiona ta ne lectrophilic oxygen atom, but on this occasion dication 59.H owever, based on literature precedent, vide supra,w ed eemed routes (a) and (b) to be lessl ikely than attack at the softer electrophilic sulfur atoms in intermediates 55 and 59 (Scheme 26 c,d).
If oxathiane 77 were to reacta tt he sulfonium centreso f cation 55 (route c) or dication 59 (route d), ad ithiadication intermediate 80 would be produced (although seemingly unlikely,i ntermediate dithiadications have been synthesised previously by reaction between as ulfide and an activated sulfoxide). [33] SubsequentP h 2 SO attack at the oxathiane sulfur atom of the dithiadication would then afford oxodisulfoniumi on 79. Thus, regardless of the early steps in the reaction, all pathways converge on oxodisulfonium ion 79.T he final step in the reaction is then aq uench of the oxodisulfoniumi on by diphenyl sulfide to afford the oxathiane-S-oxide 81 and triaryl sulfonium ion 82.Wefavoured route (d) as the pathway for the formation of the dithiadication, which involves attack on the dication 59-first, postulated by Gin and co-workers(Scheme 19) as the reactive intermediate in a2 :1 Ph 2 SO/Tf 2 Oa ctivationm ix, and then confirmed by our own experiments in this study using 19 FNMR and 18 O-labelling studies. Extension of the labelling studies to as imple non-glycosyl oxathiane, demonstrated that the stereoselective sulfoxidation was not limited to substrates containing as ugar ring that have the ability to interconvert between axial and equatorial-orientated intermediates through anomericb ond breaking and generation of an oxacarbenium ion, followed by bond rotation and then intramolecular ring closing. It must therefore also be possible for the axial and equatorial activateds ulfoxide intermediates to also interconvert througha ni ntermolecular attack of Ph 2 SO on the activated oxodisulfonium ion 79,w here the lowest-energy stereoisomer is quenched to afford the lowest-energy sulfoxide (Scheme 26). An umber of other detailed mechanistic studies have also been used to dissects ome of the more nuanced stereochemical preferences observed in glycosyls ulfoxide formation. [53] Including Crich and co-workers [54] who established inherent stereochemical trends in the oxidation of thioglycosides. The authors concluded that (R) s sulfoxides are strongly favoured when axial-(a)-thioglycosides are oxidised, as the exo-anomeric effect leads to shielding of the of pro-S sulfur lone pair under the ring and exposes the pro-R lone pair to the solvent, whereas equatorial-(b)-thioglycosides afford sulfoxide diastereomers with reduced inherents ubstrate stereocontrol, only weakly favouringt he (S) s sulfoxide. An example of the dominance of this stereochemical preference observed for axial-(a)-thioglycoside oxidationw as noted in the preferential formation of an axylopyranosyl sulfoxide in as eemingly unlikely inverted 1 C 4 chair conformation.T oi nvestigate this preference Crich deployedaglycosyl allyl sulfoxide-sulfenater earrangement to probe the kinetic and thermodynamic preferences of sulfoxide Scheme26. a-d)Possibler eaction pathways for the oxidation of generic oxathiane 77.M echanisms are depicted as S N 2p rocesses for simplicity,a lthoughi tis likely that some mechanisms may proceed via sulfurane intermediates. Reproduced from ref. [52]. formationf rom thioxylosides. As expected oxidation of b-thioxyloside 83 b preferentially afforded the (S) s sulfoxide 84 b (S) s as the major (kinetic) product (Scheme 27 a), while the a-thioxyloside 83 a afforded the inverted 1 C 4 conformer of (R) s sulfoxide 84 a (R) s as the major (kinetic) product (Scheme 27 b). In the former b-series,f ollowing thermala llyl sulfoxide 84-sulfenate 85 rearrangement in deuteriobenzene, the thermodynamic product provedt ob et he same as the kinetic product. However,f ollowing thermal equilibrationo ft he latter 1 C 4 conformer of the sulfoxide 84 a (R) s ,c onversely thermodynamic reversion to the minor kinetic product 84 a (S) s occurred.
The observation that the kinetic sulfoxide 84 a (R) s exists in the triaxial inverted 1 C 4 conformer is explained by the authors as ap reference for minimising repulsions between the sulfoxide (S)-O and C2-O2d ipoles, which are unfavourably alignedi n the minor 4 C 1 conformer of the (R) s diastereomer,b ut following thermodynamic equilibration to the 84 a (S) s diastereomer, the preference to ring flip is obviated by al acko fd ipole repulsion, meaning 84 a (S) s exists in the expected 4 C 1 conformer.
a-Thioglycosides and analogous a-sulfoxides of S-phenyl mannoazideu ronate donors were also shown to exist primarily in the 1 C 4 confirmation, [55] as opposed to the corresponding b-thioglycoside/sulfoxide anomers that adopt a 4 C 1 chair in line with the observations made for xylopyranosylsulfoxides.

Conclusions
Since their first deployment as an anomeric leaving group over 25 years ago, sulfoxides have become increasingly attractive to synthetic carbohydrate chemists because of their penchant for facilitating interesting and unexpected transformations. As exampleso fs uch transformationsi nt he literature have multiplied, so has the ability of chemists to harnessa nd direct this complex reactivity.T his has led to the emergence of significant roles for sulfoxides as mediators in arange of innovative mechanistic strategies for probing glycosylationa nd other cognate reactions, includingt he development of cation clocks, mass spectrometry and 13 CNMR isotopic-labelling studies, and DFT molecular-modelling studies. Feedback from thesem echanistic studies has in-turn led to improvements in the reactivity,a nd anomeric stereoselectivity of sulfoxide glycosyl donors for the synthesis of challenging and complex oligosaccharides, as well as ap anel of increasingly potent thioglycoside activators for the synthesis of biologically important deoxy sugars, among others. These pioneering studies have also begun to influence the manneri nw hichc arbohydrate chemists approach and rationalise glycosylations using other classes of glycosyld onor.