4,6-O-[1-Cyano-2-(2-iodophenyl)ethylidene] Acetals. Improved Second Generation Acetals for the Stereoselective Formation of β-d-Mannopyranosides and Regioselective Reductive Radical Fragmentation to β-d-Rhamnopyranosides. Scope and Limitations

Abstract

The [1-cyano-2-(2-iodophenyl)]ethylidene group is introduced as an acetal protecting group for carbohydrate thioglycoside donors. The group is easily introduced under mild conditions, over short reaction times, and in presence of a wide variety of other protecting groups by the reaction of the 4,6-diol with triethyl (2-iodophenyl)orthoacetate and trimethylsilyl triflate, followed by trimethylsilyl cyanide and boron trifluoride etherate. The new protecting group conveys strong β-selectivity with thiomannoside donors and undergoes a tin mediated radical fragmentation to provide high yields of the synthetically challenging β-rhamnopyranosides. The method is also applicable to the glucopyranosides when high α-selectivity is observed in the coupling reaction and α-quinovosides are formed selectively in the radical fragmentation step. In the galactopyranoside series, α-glycosides are formed selectively on coupling to donors protected by the new system, but the radical fragmentation is unselective and gives mixtures of the 4- and 6-deoxy products. Variable temperature NMR studies for the glycosylation step, which helped define an optimal protocol, are described.

INTRODUCTION

Rhamnopyranosides are significant components of bacterial capsular polysaccharides, which are involved in the propagation of disease states. Their synthesis presents a means of understanding these biointeractions and routes to possible vaccines against them. While l-rhamnose and its glycosides are widespread, the d-series is being found with greater frequency and the preparation of oligosaccharides bearing this novel subunit has become the object of increasing interest from standpoints of characterization and biophysical investigation.¹ The scarcity of d-rhamnose itself distinguishes the problem of d-rhamnoside synthesis from that of the l-series for which the obvious starting point is the readily available l-rhamnose. The issue of glycosidic bond formation in the two enantiomeric series may be reconciled to a single problem by the chemical synthesis of d-rhamnose or of a suitably protected derivative. Indeed, this has been the method of choice in other laboratories, with approaches to d-rhamnopyranosyl donors beginning from d-mannose and including an iodination/reduction sequence at the C6 hydroxyl or an Hanessian-type NBS-mediated cleavage of the 4,6-O-benzylidene protected mannosides.^2,3 We reason, however, given the present state of the art in the synthesis of the β-l-rhamnopyranosides,⁴ that this is a less than ideal approach to the β-d-rhamnopyranosides. This is amply documented by a recent synthesis of a β-d-rhamnopyranoside at the core of a trisaccharide by means of a 2-O-sulfonyl protected rhamnosyl donor, when the anomeric selectivity was limited to 1.1:1 in favor of the β-anomer.⁵ Rather, we have preferred an approach in which the β-d-rhamnopyranosidic linkage is introduced in the form of a β-d-mannopyranoside which, taking advantage of the β-directing effect of the 4,6-O-benzylidene acetal protecting group, is currently one of the easier types of glycosidic bond to prepare with reproducibly high stereoselectivity.⁶ Once the glycosidic bond has been formed the 4,6-O-benzylidene acetal is then cleaved reductively to afford the β-d-rhamnopyranoside. The most obvious approach to the reductive cleavage of the 4,6-O-benzylidene acetal is the Hanessian-Hullar N-bromosuccinimide mediated cleavage to the 6-bromo-6-deoxy mannoside followed by hydrogenolytic cleavage of the bromine atom, or the more recent Roberts’ radical cleavage with a catalytic thiol, which leads directly to the 6-deoxy system.⁷ However, neither system is really compatible with the presence of benzyl ethers and like, owing to essential hydrogen atom abstraction step from the acetal.⁸ One rare exception to this rule is the high yield cleavage of a benzylidene acetal of a 1,2-diol in the presence of benzyl ethers with NBS on a significant scale and in excellent yield reported in the course of a synthesis of a cyclosporine component.⁹ It should be noted, however, that the rate of hydrogen atom abstraction from the acetal position of 1,3-dioxolanes is approximately an order of magnitude faster than that from the comparable position in 1,3-dioxanes,¹⁰ which accounts for the selectivity over benzylic hydrogen atom abstraction in this example.

The central role of benzyl ether type protecting groups in modern oligosaccharide synthesis and their incompatibility with the Hanessian-Hullar and Roberts’ 4,6-O-benzylidene acetal fragmentations spurred the search in our laboratory for alternative means of generation of 2-substituted-1,3-dioxan-2-yl type radicals, not dependent on hydrogen atom abstraction, suitable for use in stereoselective glycosylation reactions. This search led to our development of a first generation solution in the form of the 4,6-O-[α-(2-(2-iodophenyl)ethylthiocarbonyl)benzylidene] group as a surrogate for the 4,6-O-benzylidene acetal (Scheme 1) and its subsequent employment in the total synthesis of the lipopolysaccharide E. hermanii ATCC 33650/33652 (Scheme 2).¹¹ In this synthesis both α- and β-rhamnosyl linkages were formed from donors equipped with the novel acetal protecting group, and a single radical reaction step was used to uncover the two latent rhamnopyranosides simultaneously. This protecting group provided excellent β-selectivities in coupling reactions and the high yields in the fragmentations are decreased only to a minor extent by formation of a by-product from reduction of the intermediate benzylidene radical to the benzylidene radical itself. Notwithstanding this success, the method suffers from two limitations: the lengthy sequence required to prepare the thiol, and the transesterification required to introduce the thiol ester, which limits functional group compatibility.

Scheme 1.

4,6-O-[α-(2-(2-Iodophenyl)ethylthiocarbonyl)benzylidene] Radical Fragmentation.

Scheme 2.

Synthesis of lipopolysaccharide E. hermanii ATCC 33650/33652

It was hoped that broader versatility could be achieved by another radical methodology, perhaps also capable of decreasing the reduction by-product observed with the first generation system. On the basis of seminal work by Beckwith on the radical migration of cyano groups, and subsequent applications to synthesis by Rychnovsky and co-workers,¹² we hypothesized that a 2-cyano-1,3-dioxane would serve as a suitable precursor to the 1,3-dioxan-2-yl radical by intramolecular transfer of the cyano group to an appropriately placed radical. The successful implementation of this second generation method is described herein.

RESULTS AND DISCUSSION

Acetal Design and Development

Before embarking on the development of the second generation system it was necessary to ascertain the effect of the 2-substituent on the fragmentation, more especially the regioselectivity, of substituted 1,3-dioxan-2-yl radicals. Additionally, it was reasoned that substitution of a methyl group for the phenyl group of the traditional benzylidene fragmentation could destabilize the incipient radical at the acetal carbon leading to a faster rate of fragmentation and diminishing amounts of reduced product.

To this end we prepared a modified first generation system beginning with the diol 5. Collins et al. observed that acetal exchange with the methyl pyruvate dimethyl acetal provides high yields of an undesired isomerization product; in our hands similar results were observed and the by-product proved all but intractable in most solvent systems.¹³ Of the literature methods surveyed, Ziegler’s proved the most workable, yielding 31% of a 2:1 (equatorial methyl: axial methyl) mixture of isomers, ^{14, 15} with the remainder of the starting material undergoing decomposition during the course of reaction, a result observed by other groups in use of the method. ¹⁶ The separated isomer with an equatorial methyl group was then smoothly transesterified at room temperature to give the new donor 8 in 76% yield.

Benzenesulfinyl piperidine (BSP) mediated coupling¹⁷ of donor 8 in presence of the hindered base, tri-tert-butyl pyrimidine (TTBP)¹⁸ proceeded to give disaccharide 10 in 76% yield with 5% recovered starting material, whereas coupling with the stronger thiophile generated from the Ph₂SO/Tf₂O combination¹⁹ saw complete activation with an 80% yield. In both instances only the β-isomer could be isolated by silica gel chromatography, in keeping with observations made with the first generation system itself. Dropwise addition of AIBN and tributyltin hydride to substrate 10 in refluxing toluene, according to the procedure used with the first generation system, provided the β-d-rhamnoside 11 in 74% yield with 15% of the ethylidene acetal by-product 12 isolated after deacylation and column chromatography (Scheme 3). This result parallels very closely those obtained earlier with the first generation series and indicates that the 1,3-dioxan-2-yl radical need not carry a 2-phenyl substituent for a highly regioselective cleavage favoring fragmentation of the primary C-O bond. The analogous observation was also made by the Roberts’ group with their thiyl radical-based hydrogen atom abstraction/fragmentation system. The fact that results with methyl prove similar to those with phenyl suggests that the transition state for fragmentation is late and the contribution from destabilization of the radical is equally countered by loss of the delocalization obtained in the conjugated benzoate ester, again in accord with theoretical calculations performed by Roberts.⁷

Scheme 3.

Preparation and Fragmentation of a Pyruvate-based Mannosyl Donor.

Turning to the use of nitriles as radical precursors we first investigated briefly the intermolecular abstraction of this group by a stannyl radical, based on the demonstration of Curran and co-workers of radical mono decyanation of gem-dicyano acetals using AIBN/Bu₃SnH.²⁰ Accordingly, donor 14 was synthesized via Utimoto’s Lewis acid assisted TMSCN protocol.²¹ From a mixture of diastereomers of the preformed orthoester, the latter reaction provided a single isomer cleanly and in good yield. The diphenyl sulfoxide/Tf₂O coupling protocol yielded 67% of the pure β-anomer 16 with 20% of the unactivated donor recovered. The incomplete activation was unexpected given use of the strong thiophile; the cause was not investigated at this stage, however, due to uncertainty of how the product would behave under conditions of radical fragmentation. Unfortunately, repeated attempts at heating in toluene with AIBN/Bu₃SnH induced no fragmentation and the disaccharide was recovered quantitatively each time (Scheme 4). Evidently, the generation of a somewhat stabilized 2-methyl-1,3-dioxan-2-yl radical is insufficient to promote the desired intermolecular nitrile abstraction reaction.

Scheme 4.

Radical Decyanation Based Ethylidene Fragmentation.

We turned, therefore, to intramolecular nitrile abstraction following the precedent of the Beckwith and Rychnovsky groups.¹² Thus the requisite orthoester 19 was prepared according to standard Pinner synthesis via the imidate salt 18 of 2-iodo-phenyl-acetonitrile in 57% over two steps with the known acid ester as the major by-product.²² Acid-catalyzed orthoester exchange in presence of the acid ester, followed directly by BF₃.OEt₂-promoted cyanation, without prior purification provided the desired donor in 80% yield after silica gel chromatography. The presumed stereochemical outcome of cyanation of 20, based on stereoelectronic effects,²³ was rigorously proven by x-ray crystallographic analysis (Fig. 1) of the β-mannoside 21 after coupling to methyl 2,3-O-isopropylidene-α-l-rhamnopyranoside by the Ph₂SO/Tf₂O protocol (Scheme 5). Dropwise addition of tributyltin hydride and AIBN to a solution of mannoside 21 in toluene at reflux finally gave the β-rhamnopyranoside 22 in 76% yield, with no indication of the formation of the regioisomeric 4-deoxy product (Scheme 5).

Figure 1.

X-ray structure of disaccharide 21.

Scheme 5.

Nitrile Transfer Based Fragmentation.

Glycosylation and Radical Fragmentation

With proof of principle for the complete second generation sequence in hand, a series of VT-NMR experiments, described in detail below, helped definite an optimal protocol for preparative scale couplings in which the mixture of thioglycoside, and diphenyl sulfoxide was warmed to −20 °C after addition of triflic anhydride then cooled back to −78 °C before the acceptor was introduced. In this manner a number of couplings were conducted in high yield, with excellent β-selectivity as reported in Table 1.

Table 1. Glycosylation reactions with [1-cyano-2-(2-iodophenyl)]ethylidene protected donors.

Entry	Donor	Acceptor	Product	1JCH (Hz)	Procedurea (% yield, β: α selectivity)
1				β: 158.6	A: 63 % (β-only) B: 92 % (β-only)
2	20			β: 156.1	B: 77 % (β-only)
3	20			β: 157.1 α: 173.7	B: 72% (8.4:1)
4	20			β: 154.9	B: 71% (β-only)
5	20			β: 153.5	B: 82% (β-only)
6		9		β: 164.9 α: 175.0	A: 41% (1:5)
7		39		β: 159.9 α: 171.2	A: 33% (1:4) B: 25% (1:5)
8		39	48	-	A: 31% (1:4)
9		39		β: 164.9 α: 169.2	B: 65% (1:10)
10		9		β: 164.9	A: 96% (β-only)
11	37	39		β: 159.9	A: 75%(β-only)
12	37	40		β: 159.9	A: 86% (β-only)
13	37	42		β: 160.9	A: 89% (β-only)

^aIsolated yields of analytically pure material. Procedure A: Tf₂O added at −78 °C and reaction maintained at −78 °C until ~1.5 h after addition of acceptor. Procedure B: Tf₂O added at −78 °C and reaction mixture warmed to −20 °C for ~20 min. Then cooled to −78 °C and acceptor added and stirred at −78 °C for ~1.5 h.

With the couplings in hand, other measures were then considered to favor the desired fragmentation pathway. To this end, tris(trimethylsilyl)silane was employed as a weaker hydrogen donor and addition times and temperatures were adjusted until optimized parameters were found,²⁴ as summarized in Table 2. Of note, the silane proved an inefficient propagator and an amount of starting material was recovered in all such trials. Also, it was found that the pathway leading to the 6-deoxy sugars was favored at higher temperature, though, no conditions were obtained which eliminated formation of the reduced acetal altogether. No 4-deoxy sugars were observed in the ¹H-NMR spectra of the crude reaction mixtures indicating complete regioselectivity of radical fragmentation in the mannose series. Overall, therefore, the 4,6-O-[1-cyano-2-(2-iodophenyl)]ethylidene group provides the β-d-rhamnopyranosides rapidly and in high yields from thiomannosides.

Table 2.

Optimization of the Radical Fragmentation with Disaccharide 21.

Entry	Solvent	Initiator	Propagator	Addition	6-Deoxy 22 Yield (%)	Acetal 23 Yield (%)	Recovered 21 (%)
1	Toluene	AIBN	Bu3SnH	3 h	76	15	0
2	Toluene	AIBN	((CH3)Si)3SiH	3 h	68	6	15
3	Xylenes	AIBN	Bu3SnH	3 h	56	2	30
4	Xylenes	AIBN	Bu3SnH	2 h	81	5	0
5	Xylenes	Bz2O2	Bu3SnH	2 h	61	10	0
6	Xylenes	AIBN	((CH3)Si)3SiH	2 h	70	5	20

Subsequently, the optimized fragmentation conditions were applied to the complete series of β-d-mannosides, resulting in each case in high isolated yields of the corresponding β-Drhamnopyranosides, as set out in Table 3, entries 1-5.

Table 3.

Radical fragmentations of coupled β-d-mannopyranosides and α-d-glucopyranosides.

Entry	Substrate	Major Product (a)	Reduction Product (b)	% Yield (a:b)a
1	21			86% (16:1)
2	43			89% (14:1)
3	44			66% (6:1)
4	45			83% (trace b)
5	46			94% (8:1)
7	48			72% (trace b)
8	49			69% (trace b)b

^aAll yields refer to isolated yields from 2 h addition of 1.5 eq. Bu₃SnH and 0.2 eq. AIBN in refluxing xylenes.

^b17% unreacted starting material recovered.

To probe the scope and limitations of this second generation system the glucopyranose and galactopyranose systems were also investigated. In the glucose series four donors were prepared, equipped with 2-naphthylmethyl, p-methoxybenzyl, and standard benzyl protecting groups as outlined in Scheme 6. Contrary to our expectations, couplings to glucosyl donors 29, 30, and 32 resulted in relatively complex reaction mixtures and, ultimately, low yields of the coupled products (Table 1, entries 7 and 8). Neither the use of higher equivalents of activator, nor variation of temperature was able to increase these coupling yields. Only the 2,3-di-O-benzyl protected donor 31 produced satisfactory yields and only upon employing protocol B, as developed with the mannose donor 20. In line with the precedent for coupling to 2,3-di-O-benzyl-4,6-O-benzylidene protected glucosyl donors, the reactions were α-selective. ²⁵ As discussed below, VT-NMR experiments were again conducted in an attempt to understand the problematic couplings.

Scheme 6.

Preparation of Glucosyl Donors.

Only two examples of the radical fragmentation were studied in the glucose series, nevertheless, excellent selectivities for fragmentation of the primary bond leading to the 6-deoxy-d-glucoside, or d-quinovoside, were obtained (Table 3, entries 7 and 8). This selectivity parallels exactly that seen with the first generation system in the glucose series, as well as that seen by Roberts’ and Hanessian in their fragmentation of standard 4,6-O-benzylidene acetals in the glucopyranose series. In other words, the radical fragmentation in the glucose series is no different from that in the mannose case with very high selectivity for formation of the 6-deoxy isomer.

Working in the galactopyranose series we began with the known benzylidene acetal proctected thioglycoside 33,²⁶ which was converted to the second generation acetal by a three step protocol including removal of the benzylidene group with neopentyl glycol and catalytic camphorsulfonic acid, introduction of the orthoester and, finally, the nitrile (Scheme 7). As expected for the more reactive galactopyranoside series, activation of this donor was efficient. However, the addition of 1-adamantanol, usually an excellent glycosyl acceptor, resulted not in the isolation of the anticipated β-galactoside 36 but in the formation of a relatively complex mixture. To elucidate this problem we returned to the more plentiful benzylidene acetal 33 which on activation and attempted coupling delivered a 74% yield of an unexpected disaccharide 34.²⁷ We were able to circumvent this problem, with its obvious roots in acetoxy migration and neighboring group participation, simply by switching to the corresponding benzoate ester 37 when the adamantanyl β-galactopyranoside 38 was formed in 89% yield. A series of further couplings were then conducted with this donor as reported in Table 1 (entries 12-15), all of which proceeded in high yield and with excellent β-selectivity. The synthesis of donor 37 (Scheme 7) serves to highlight two advantages of the second generation radical precursor over and above the first generation system (Scheme 1), both of which revolve around compatibility with ester protecting groups. First, the second generation system may be introduced in the presence of esters, whereas the transesterification step required for the first generation system precluded their use. Second, saponification may be readily carried out in the presence of the second generation system under standard conditions which was obviously not possible for the first generation, thiol ester-based radical precursor.²⁸

Scheme 7.

Preparation and Coupling of Galactoside Donors.

Radical fragmentation of the galactosides prepared in this manner led to mixtures of the 6-deoxy (fucopyranosides) products and the 4-isomers, typically with a slight excess of the 4-deoxy regioisomer (Table 4). To facilitate separation of the mixture of regioisomers chemoselective saponification of the primary ester in the 4-deoxy product was achieved with guanidine in ethanol.²⁹ The lack of regioselectivity in these galactoside fragmentations matches well that found by Roberts’ in his thiyl radical mediated cleavage of galactose-based benzylidene acetals.³⁰ The observed regioselectivity and the high degree of agreement with the Roberts’ work prompted a re-inspection of our earlier work on the application of the first generation radical trigger to the galactose series. Thus, we reported earlier that acetals 65 afforded an 89/9 mixture of 67 and 69 and that acetal 66 gave exclusively the 4-deoxy product 68.^11a Scrutiny of the spectral data for 65-69 revealed these structures to have been misassigned in the original publication, and we now revise these structures, along with their precursors described in the original supporting information, to the galactofuranoside substrates 70 and 71 and the 5- and 6-deoxygalactofuranoside products 72, 73, and 74. The error in assignment of these structures arose because of a methyl galactopyranoside to methyl galactofuranoside rearrangement^31,32 that had gone undetected during the Lewis acid (TMSOTf) mediated acetalization of 75 which it is now clear gave the furanosides 76 and not the pyranosides 77. The sequence employed originally in the intended synthesis of 65 and 66 was predicated on the need to introduce the acetal and complete the subsequent transesterification step to the thiol ester before introduction of any ester protecting groups. It serves to highlight, therefore, one of the significant advantages of the second generation series presented here, namely the full compatibility with esters as apparent in Scheme 7 and as discussed above.

Table 4.

Radical fragmentations of β-d-galactopyranosides.

Entry	Substrate	Major Product (a)	Major Product (b)	Reduction Product (c)	% Yield (a:b:c)b
1	50				79% (7.5:7.5:1)
2	51				73% (2.8:1.8:1)b
3	52				70% (3.2:2.8:1)b
4	38				83% (5:4.5:1)

^aAll yields refer to isolated yields from 2 h addition of 1.5 eq. Bu₃SnH and 0.2 eq. AIBN in refluxing xylenes.

^bYields after radical reaction and selective saponification of the ester on the primary hydroxyl with 1 eq. guanidine in CH₂Cl₂:EtOH (1:9).

VT NMR Studies on Glycosylation Intermediates

The unexpectedly incomplete activation of the glycosyl donor 20 at −78 °C, with 20% unchanged thioglycoside recovered from the example reported in Scheme 5, prompted an investigation by variable temperature NMR spectroscopy, our method of choice for probing such questions.³³ Accordingly, a CD₂Cl₂ solution of donor 20 and diphenyl sulfoxide was treated at −78 °C and the spectrum recorded. The anomeric peak of donor 20 at δ 5.26 (Figure 2a) was immediately consumed after addition of Tf₂O, giving rise to two new peaks, the first at δ 6.09 and the second δ 5.90 (Figure 2b). Upon warming in ten degree intervals to −30 °C, it was observed that the first peak decreased and then disappeared altogether as the second peak increased relative to the methylene chloride peak present at δ 5.32 (Figure 2c). Further warming witnessed the disappearance of the peak at δ 5.90 and emergence of a new peak at δ 6.28 at 0 °C. In a subsequent experiment it was observed that, after enriching the peak at δ 5.90 by warming to −20 °C the reaction mixture could be recooled to −78 °C with no detriment, and then quenched with methanol to give the anomeric mixture of methyl glycosides. The last result suggests that the peak at δ 5.90 is the active species, the glycosyl triflate, in such couplings, which decomposes at 0 °C to be replaced by a signal at δ 6.28. We tentatively assign this signal to glycal 90 (H1) on the basis of the previous isolation of such 2-alkoxyglycals from this type of experiment.²⁵

Figure 2.

Low temperature NMR spectra for a) thiomannoside 20 at −78 °C, before activation, b) glycosyl triflate (δ 5.90) and by-product (δ 6.09) at −78 °C, and c) the glycosyl triflate alone at −20 °C, after conversion of the other intermediate.

Exactly analogous results were observed with the 2,3-di-O-benzyl protected glucosyl donor 31 (See supporting information for VT-spectra). Thus, on addition of Tf₂O to a CD₂Cl₂ mixture of 31, and Ph₂SO at −78 °C the substrate was consumed and two new anomeric signals were formed at δ 6.02 (d, J = 3.3 Hz) and δ 5.63 (d, J = 9.6 Hz), representing an α- and a β-derivative, respectively. On warming, the signal at δ 5.63 was converted to that at δ 6.02 by −20 °C, which we assign to the α-triflate. This triflate was stable to −10 °C at which temperature the onset of decomposition was observed. If the reaction mixture was warmed to −20 °C and then recooled to −78 °C (Table 1, protocol B) a good yield of the coupled product could be obtained. Addition of triflic anhydride to donor 32 likewise led to the formation of two new peaks at δ 5.95 (d, J = 2.7 Hz) and δ 5.70 (d, J = 9.0 Hz), indicative of an α- and a β-derivative, respectively (See supporting information for VT-spectra). Quenching of the mixture at −78 °C with methanol resulted in the loss of the signal at δ 5.95 in favor of an anomeric mixture of methyl glucosides, which suggests this resonance to be that of the α-glucosyl triflate. However, in the contrast to the di-O-benzyl system 31, and to the mannosyl donor 20, on warming of the mixture from the activation of 32 to −50 °C the upfield peak (δ 5.70) was observed to have decreased, relative to the methylene chloride peak at δ 5.32, in favor of a new, poorly resolved peak at δ 5.61. By −30 °C the δ 5.70 peak had disappeared completely and the anomeric triflate peak, at δ 5.95 was diminished in size, while the resonance at δ 5.61 had both increased in intensity and was better resolved into a doublet with coupling constant of 4.0 Hz, consistent with an α-glucosyl derivative. Finally, at −20 °C, the downfield peak had disappeared completely, leaving only the δ 5.61 resonance. The product corresponding to this last anomeric signal was isolated and characterized as decomposition product 91, arising from electrophilic aromatic substitution by the anomeric oxacarbenium ion on the naphthylmethyl protecting group. The difference between the performance of the 2,3-di-O-benzyl donor 31 and the 2,3-di-O-naphthylmethyl donors 30 and 32, apparent from the preparative scale coupling reactions (Table 1) is therefore seen to be the result of the more facile aromatic substitution onto the naphthalene ring, which competes with the conversion of the unknown intermediate at δ 5.70 to the anomeric triflate on warming. Although no VT-NMR experiments were conducted, a parallel problem presumably underlies the poor yield obtained with the 2,3-di-O-(4-methoxybenzyl) donor 29 (Table 1, entry 6).

The one remaining unknown is the identity of the substances characterized by anomeric resonances at δ 6.09 in the mannose series and δ 5.63 (J = 9.6 Hz, 2,3-di-O-benzyl system) and δ 5.70 (J = 9.7 Hz, 2,3-di-O-naphthylmethyl system) in the glucose series, and which is slowly converted to the corresponding α-anomeric triflates on warming. We considered O-glycosyl sulfoxonium ions, as observed by Gin, ³⁴ but excluded this possibility with the aid of an experiment in which the diphenyl sulfoxide was preactivated at −78 °C with triflic anhydride, thereby removing all nucleophilic sulfoxide from the reaction mixture, before addition of the thioglycoside, when the same resonances were displayed as in the normal mode of activation. We also considered the possibility of glycosyl thiosulfonium salts, such as might arise from reaction of a glycosyl triflate or an oxacarbenium ion with disulfide formed in situ from reaction between the thioglycoside and the activating mixture. ³⁵ However, the addition of diethyl disulfide to the initial reaction mixture arising from activation of 32 resulted in loss of both the anomeric triflate signal at δ5.95 and the signal in question at δ5.70, giving a spectrum devoid of resonances in the region δ 5.4 - 7.0. That this occurrence was not simply due to decomposition, but to the formation of a third, likely thiosulfonium intermediate, was confirmed by subsequently increasing the probe temperature to −20 °C, where the familiar decomposition product due to electrophilic aromatic substitution was observed to appear. The identities of the products, with anomeric peaks at δ 6.09 in mannose and δ 5.70 in glucose, therefore remain unknown at present, but it appears likely that they are initial intermediates formed on reaction between the thioglycoside and the activated diphenyl sulfoxide, and persist due to the somewhat disarmed nature of the donors. These results call to mind the work of Lowary and co-workers who observed a number of intermediates by VT-NMR spectroscopy during the activation of a series of 2,3-anhydrothiofuranoside sulfoxides with Tf₂O which only coalesced to the apparent glycosyl triflate on warming to −40 °C.³⁶ If, as Bols has demonstrated in recent work, the effect of 4,6-O-acetals in disarming sugar donors is partially torsional and partially electronic, it must be assumed that an apical nitrile or other, more severely electron withdrawing group could be capable of enhancing the electron-withdrawing, disarming effect,³⁷ leading to enhanced stability of any intermediate sulfonium salts along the pathway to the glycosyl triflate. By a similar token, once formed the glycosyl triflates should be more stable than those observed with simple benzylidene acetal protecting groups as is reflected in the decomposition temperature of 0 °C. This in turn leads to tighter transient oxacarbenium triflate ion pairs and enhanced β-selectivity³⁸ over that seen with the simple benzylidene acetals.

CONCLUSION

The [1-cyano-2-(2-iodophenyl)]ethylidene group, an acetal protecting group for carbohydrate thioglycoside donors has been developed. The group is easily introduced under mild conditions, over short reaction times, and in presence of a wide variety of other protecting groups. It conveys strong β-selectivity with thiomannoside donors and undergoes a tin mediated radical fragmentation to provide high yields of the synthetically challenging β-rhamnopyranosides.

EXPERIMENTAL SECTION

Ethyl (2-iodophenyl)iminoacetate hydrochloride (18)

To 5.00g dry 2-iodophenylacetonitrile was added 1.400 mL (1.2 equiv.) absolute ethanol under inert atmosphere. The solution was cooled to 0 °C, saturated with anhydrous HCl gas, and then allowed to stand at 0 °C overnight. The resultant white solid was triturated with ethyl ether, filtered and dried under vacuum, providing the product in 92% yield. White solid. Mp: 125-128 °C (sublimes). ¹H NMR (DMSO d₆): δ 7.86 (d J = 8.0 Hz, 1H), 7.43 (d J = 6.5 Hz, 1H), 7.38 (t J = 7.0 Hz, 1H), 7.06 (dt J = 1.0, 7.5 Hz, 1H), 4.44 (q J = 7.0 Hz, 2H), 4.20 (s, 2H), 1.20 (t J = 7.0 Hz, 3H). ¹³C NMR (DMSO d₆): δ 176.7 (C), 139.8, 135.9 (C), 131.7, 130.4, 129.2, 101.9 (C), 70.2 (CH₂), 43.7 (CH₂), 13.8. Anal. Calcd. for C₁₀H₁₃ClINO %C: 36.89, %H: 4.02; found C: 36.79, H: 3.91%.

Triethyl (2-iodophenyl)orthoacetate (19)

6.00g of imidate 18 were dissolved in 40 mL absolute ethanol and stirred under inert atmosphere for 2 days at room temperature. A volume of ethyl ether equal to that of ethanol used (40 mL) was added to the reaction mixture. The reaction was then filtered through Celite to remove the precipitated ammonium chloride and solvents evaporated to yield the crude orthoester together with the by-product acid-ester in 62% yield by NMR. The orthoester was best used as the crude mixture itself.

General procedure for preparation of 4,6-O-[1-Cyano-2-(2-iodophenyl)ethylidene] protected donors

1 equiv. of relevant diol and 0.05 equiv. CSA were dissolved in dry CH₂Cl₂ (to ~0.1M) and stirred under inert atmosphere at 0 °C. 1.2 equiv. (based on NMR) crude orthoester 19 in a sparing amount of CH₂Cl₂ was then added dropwise to the reaction mixture over ~10 min. The reaction was allowed to warm to room temperature and left to stir for ~3 hr., after which time TLC showed that no starting material remained. The mixture was diluted with CH₂Cl₂ and washed with aqueous NaHCO₃, brine, and dried over Na₂SO₄. Evaporation of solvents and azeotropic removal of water with benzene provided the crude orthoester as a mixture of isomers, which was reacted without further purification. The crude orthoester was taken up in CH₂Cl₂ (to ~0.1M) and 4 equiv. TMSCN was added in one portion. The mixture was stirred at 0 °C while 0.4 equiv. BF₃(OEt₂) in a sparing amount of CH₂Cl₂ was added dropwise. The reaction mixture was stirred at room temperature ~1.5 hr. further, after which time TLC showed that all starting material had been converted to a slightly more polar compound. Solid K₂CO₃ was added and the mixture stirred ~10 min. before being diluted with CH₂Cl₂ and washed with aqueous NaHCO₃, brine, and dried over Na₂SO₄. Column chromatography provided the donors in 74-89% yield from the diols.

Ethyl 2,3-di-O-benzyl-4,6-O-[1-cyano-2-(2-iodophenyl)]ethylidene-α-d-thiomannopyranoside (20)

0.72g (1.8 mmol) dry diol 5 and 0.77g (2.1 mmol, 1.2 equ.) crude orthoester 19 were combined according to the general procedure to give 0.93g 20 (80% over two steps) after column chromatography (eluent 10:1 Hexanes:EtOAc). [α]²⁴_d: +90.7 (c 1, CHCl₃). IR (thin film): 2251 (CN) cm⁻¹. ¹H NMR (CDCl₃): δ 7.90 (dd J = 1.0, 8.5 Hz, 1H), 7.52 (dd J = 1.0, 7.5 Hz, 1H), 7.39-7.29 (m, 11H), 6.99 (dd J = 1.5, 8.0 Hz, 1H), 5.26 (d J = 1.5 Hz, 1H), 4.77 (d, J = 12.5 Hz, 1H), 4.69 (d J = 12.5 Hz, 1H), 4.66 (d J = 11.5 Hz, 1H), 4.55 (d J = 11.5 Hz, 1H), 4.52-4.49 (m, 1H), 4.12-4.09 (m, 3H), 3.89 (dd J = 1.0, 3.0 Hz, 1H), 3.82 (dd J = 3.5, 9.5 Hz, 1H), 3.56 (dd J = 6.5, 14.5 Hz, 2H), 2.63-2.53 (m, 2H), 1.25 (t J = 7.5 Hz, 3H). ¹³C NMR (CDCl₃): δ 139.9, 138.3, 137.8, 135.8, 131.6, 129.4, 128.5, 128.4, 128.3, 128.2, 127.9, 127.70, 127.65, 114.6, 103.1, 96.9 (C), 83.9 (¹J_CH = 166.2 Hz), 78.0, 76.7, 76.3, 73.24, 73.18, 65.8, 63.6, 49.0, 25.5, 14.9. HRMS (ESI): m/z calcd. for C₃₁H₃₂INO₅S (M + Na)⁺ 680.0944, found 680.0936.

General procedure for coupling of thiogalactopyranoside 37 (Protocol A)

1 equiv. of dry donor, together with 1.3 equiv. diphenyl sulfoxide, 1 equiv. TTBP, and freshly activated molecular sieves was taken up in dry CH₂Cl₂ (0.05M in substrate) and brought to −70°C under inert atmosphere. 0.080 mL (0.477 mmol, 1.4 equ.) Triflic anhydride was then added. After stirring at −70°C for ~30 min 2 equivalents of the acceptor alcohol dissolved in dry CH₂Cl₂ was added all at once. The mixture was allowed to stir at −70 °C for 2 hr., before being filtered, quenched with NaHCO₃, washed with brine, and dried over Na₂SO₄. Evaporation of solvent and column chromatography provided the coupled products.

General procedure for coupling of thiomannopyranoside 20 (Protocol B)

1.0 equiv. of dry donor, together with 1.5 equiv. diphenyl sulfoxide and 3 equiv. TTBP was dissolved in dry CH₂Cl₂ (0.05M to substrate with the thiomannosides and 0.01M to substrate with the thioglucosides) and brought to −70 °C under inert atmosphere. 1.7 equiv. Triflic anhydride was then added and the mixture allowed to rise to −20 °C over 30 min. After stirring at −20 °C for ~15 min, the mixture was then brought back to −70 °C and 2 equivalents of the acceptor alcohol dissolved in 1 mL dry CH₂Cl₂ was added all at once. The mixture was allowed to stir at −70 °C for 2 h, before being quenched with NaHCO₃, washed with brine, and dried over Na₂SO₄. Evaporation of solvent and column chromatography provided the coupled products.

Methyl 4-O-(2,3-di-O-benzyl-4,6-O-[1-cyano-2-(2-iodophenyl)]ethylidene-β-d-mannopyranosyl)-2,3-O-isopropylidene-α-l-rhamnopyranoside (21)

Coupling of 0.100 g (0.15 mmol) 20 with 0.066 g (0.30 mmol) 9 according to protocol B afforded 0.114 g (0.14 mmol, 92%) 21 as a white solid. [α]²⁴_d: −42.4 (c 1, CHCl₃). Mp: 112-115°C. IR (KBr pellet): 2230 (CN) cm⁻¹. ¹H NMR (CDCl₃): δ 7.87 (dd J = 1.0, 8.0 Hz, 1H), 7.49 (dd J = 1.5, 7.5 Hz, 1H), 7.39-7.22 (m, 11H), 6.97 (dt J = 2.0, 8.0, 1H), 4.94 (s, 1H), 4.86 (s, 1H), 4.84 (d J = 12.0 Hz, 1H), 4.79 (d J = 12.0 Hz, 1H), 4.54 (d J = 12.5 Hz, 1H), 4.48 (d J = 12.5 Hz, 1H), 4.38 (t J = 10.0 Hz, 1H), 4.14 (s, 1H), 4.12 (d J = 2.5 Hz, 1H), 4.08 (d J = 1.5 Hz, 1H), 3.91 (d J = 2.5 Hz, 1H), 3.63-3.62 (m, 2H), 3.56-3.49 (m, 3H), 3.40 (s, 3H), 3.20 (m, 1H), 1.51 (s, 3H), 1.34 (s, 3H), 1.31 (d J = 6.0 Hz, 3H). ¹³C NMR (CDCl₃): δ 139.8, 138.4, 138.3, 135.8, 131.4, 129.4, 128.4, 128.3, 128.2, 128.1, 127.6, 127.53, 127.46, 114.7, 109.3, 103.0, 100.1 (¹J_CH = 158.6 Hz), 97.8 (¹J_CH = 168.7 Hz), 96.7, 78.3, 78.1, 76.3, 76.2, 76.1, 74.8, 72.5, 66.6, 65.9, 64.1, 55.0, 49.0, 27.9, 26.4, 17.7. Anal. Calcd. for C₃₉H₄₄INO₁₀ C: 57.57, H: 5.45; found C: 57.34, H: 5.57%.

General procedure for radical fragmentation of glycopyranosides

1 Equiv. of substrate was dissolved in degassed xylenes (to ~0.006M) and brought to reflux under argon. Over 2 h, 1.5 equiv. Bu₃SnH and 0.20 equiv. AIBN in degassed xylenes (to ~0.025 M in Bu₃SnH) was added via syringe pump to the refluxing reaction mixture. Upon completion of the addition, the mixture was cooled to room temperature, solvent evaporated and taken up in 5 mL ethanol. 2.0 Equiv. NaBH₄ was added and the reaction stirred for ~15 min. The ethanol was removed under vacuum and the mixture was diluted with CH₂Cl₂ and washed with water and brine. Evaporation of solvents followed by column chromatography provided the deoxygenated sugars.

Methyl 4-O-(2,3-di-O-benzyl-4-O-(2-cyanophenyl)acetyl-β-d-rhamnopyranosyl)-2,3-O-isopropylidene-α-l-rhamnopyranoside (22) and by-product Methyl 4-O-(2,3-di-O-benzyl-4,6-O-[2-(2-cyanophenyl)]ethylidene-β-d-mannopyranosyl)-2,3-O-isopropylidene-α-l-rhamnopyranoside (23)

According to the general procedure 0.093 g (0.114 mmol) 21 yielded 0.062 g (0.092 mmol, 81%) 22 and 0.004 g (0.006 mmol, 5%) 23. 22: Clear oil. [α]²⁴_d: −56.2 (c 0.5, CHCl₃). IR (thin film): 2229 (CN) cm⁻¹. ¹H NMR (CDCl₃): δ 7.60 (dd J = 1.5, 7.5 Hz, 1H), 7.42 (dd J = 1.0, 7.0 Hz, 1H), 7.38-7.16 (m, 12H), 5.23 (t J = 10.0 Hz, 1H), 4.89 (s, 1H), 4.88 (d J = 13.0 Hz, 1H), 4.87 (s, 1H), 4.73 (d J = 12.5 Hz, 1H), 4.37 (d J = 12.5 Hz, 1H), 4.20 (d J = 12.5 Hz, 1H), 4.14-4.08 (m, 2H), 3.90 (d J = 2.5 Hz, 1H), 3.81 (q J = 15.5 Hz, 2H), 3.70-3.64 (m, 2H), 3.44 (dd J = 3.5, 10.0 Hz, 1H), 3.39 (s, 3H), 3.39-3.35 (m, 1H), 1.51 (s, 3H), 1.34 (d J = 6.0 Hz, 3H), 1.23 (d J = 6.5 Hz, 3H). ¹³C NMR (CDCl₃): δ 168.8, 138.7 (C), 138.1, 137.5, 132.9, 132.8, 130.6, 128.3, 128.1, 128.0, 127.8, 127.5, 127.4, 127.1, 117.7, 113.3, 109.4, 99.3 (¹J_CH = 157.4 Hz), 97.9 (¹J_CH = 171.2 Hz), 79.4, 78.5, 77.5, 77.2, 76.1, 74.1, 74.0, 71.0, 70.6, 64.3, 54.9, 39.7, 27.9, 26.5, 17.7, 17.5. HRMS (ESI): m/z calcd. for C₃₉H₄₅NO₁₀ (M + Na)⁺ 710.2941, found 710.2937. 23: Clear oil. [α]²⁴_d: −41.2 (c 0.5, CHCl₃). ¹H NMR (CDCl₃): δ 7.60 (dd J = 1.0, 8.0 Hz, 1H), 7.49 (dt J = 1.5, 7.5 Hz, 1H), 7.42-7.20 (m, 12H), 4.93 (s, 1H), 4.88-4.84 (m, 3H), 4.76 (d J = 12.5 Hz, 1H), 4.54 (d J = 12.5 Hz, 1H), 4.48 (d J = 12.5 Hz, 1H), 4.13-4.06 (m, 3H), 3.90 (t J = 10.0 Hz, 1H), 3.89 (d J = 3.0 Hz, 1H), 3.71 (t J = 5.0 Hz, 1H), 3.61 (m, 2H), 3.53 (dd J = 3.5, 10.0 Hz, 1H), 3.38 (s, 3H), 3.23-3.19 (m, 3H), 1.50 (s, 3H), 1.33 (s, 3H), 1.30 (d J = 6.0 Hz, 3H). ¹³C NMR (CDCl₃): δ 139.9, 138.5, 132.8, 132.5, 131.0, 128.4, 128.3, 128.1, 127.5, 127.4, 127.2, 118.2, 113.7, 109.4, 101.3 (¹J_CH = 156.1 Hz), 100.1 (¹J_CH = 158.6 Hz), 97.9 (¹J_CH = 167.4 Hz), 78.6, 78.4, 78.0, 77.8, 76.3, 76.1, 74.7, 72.4, 68.3, 54.9, 39.4, 27.9, 26.4, 17.7. HRMS (ESI): m/z calcd. for C₃₉H₄₅NO₁₀ (M + Na)⁺ 710.2941, found 710.2933.

Methyl 2,3,4-tri-O-benzyl-6-O-[2,3-di-O-benzoyl-4-deoxy-6-O-(2-cyanophenyl)acetyl-β-d-galactopyranosyl]-α-d-glucopyranoside (81), Methyl 2,3,4-tri-O-benzyl-6-O-[2,3-di-O-benzoyl-4-O-(2-cyanophenyl)acetyl-β-d-fucopyranosyl]-α-d-glucopyranoside (82), and by-product Methyl 2,3,4-tri-O-benzyl-6-O-[2,3-di-O-benzoyl-4,6-O-(2-cyanophenyl)ethylidene-β-d-galactopyranosyl]-α-d-glucopyranoside (83)

According to the general procedure 0.086 g (0.079 mmol) 51 yielded after column chromatography 0.045 g (0.047 mmol, 61%) of a 1.5:1 mixture of 6-deoxy and 4-deoxy products as determined by NMR, together with 0.010 g (0.010 mmol, 13%) 83. The mixture of deoxy sugars was dissolved in 2 mL EtOH:CH₂Cl₂ (9:1) and 0.003 g (1 equiv.) guanidine (obtained by neutralization of the HCl salt with NaOEt and filtration under argon) was added. The mixture was stirred for 15 min., after which time TLC showed complete disappearance of one of the two isomers and emergence of a more polar compound. Aqueous workup with extraction into CH₂Cl₂ and column chromatography afforded 0.027 g 82 (36% from 51) and 0.015 g 81 (24% from 51). 81: Clear oil. [α]²⁴_d: +21.6 (c 0.25, CHCl₃). ¹H NMR (CDCl₃): δ 7.93-7.88 (m, 4H), 7.51-7.48 (m, 1H), 7.40-7.17 (m, 18H), 7.03 (m, 2H), 5.45 (dd J = 7.5, 10.0 Hz, 1H), 5.37 (dt J = 5.5, 11.0 Hz, 1H), 4.88 (d J = 11.0 Hz, 1H), 4.75 (d J = 12.0 Hz, 1H), 4.69 (d J = 11.0 Hz, 1H), 4.66 (d J = 7.5 Hz, 1H), 4.60 (d J = 12.5 Hz, 1H), 4.49 (d J =4.0 Hz, 1H), 4.45 (d J = 11.0 Hz,1H), 4.28 (d J = 11.0 Hz, 1H), 4.10 (dd J = 1.5, 10.5 Hz, 1H), 3.88 (t J = 9.5 Hz, 1H), 3.80-3.66 (m, 2H), 3.46-3.35 (m, 2H), 3.24 (s, 3H), 2.27 (ddd J = 1.0, 5.0, 12.0 Hz, 1H), 2.06 (t J = 6.5 Hz, 1H), 1.82 (q J = 11.5 Hz, 1H). ¹³C NMR (CDCl₃): δ 166.0, 165.3, 138.8, 138.2, 133.3, 133.0, 129.73, 129.70, 129.4, 129.3, 128.5, 128.4, 128.3, 128.2, 127.9, 127.6, 1275, 101.4 (¹J_CH = 155.4 Hz), 98.1 (¹J_CH = 170.2 Hz), 81.9, 79.7, 79.1, 77.3, 77.2, 75.6, 74.7, 73.5, 72.6, 72.4, 71.7, 69.5, 68.3, 64.8, 55.1, 32.0, 25.7. HRMS (ESI): m/z calcd. for C₄₈H₅₀NO₁₂ (M + Na)⁺ 841.3200, found 841.3206. 82: Clear oil. [α]²⁴_D: +10.0 (c 0.15, CHCl₃). ¹H NMR (CDCl₃): δ 7.88 (dd J = 1.0, 7.5 Hz, 2H), 7.78 (d J = 7.5 Hz, 2H), 7.61 (d J = 7.5 Hz, 1H), 7.47 (q J = 8.0 Hz, 2H), 7.40 (t J = 6.5 Hz, 1H), 7.36-7.24 (m, 17H), 7.21 (t J = 7.5 Hz, 2H), 7.11 (d J = 6.0 Hz, 2H), 5.69 (dd J = 8.0, 10.5 Hz, 1H), 5.50 (d J = 3.0 Hz, 1H), 5.40 (dd J = 3.0, 10.0 Hz, 1H), 4.89 (d J = 11.0 Hz, 1H), 4.73 (d J = 12.0 Hz, 1H), 4.69 (d J = 10.5 Hz, 1H), 4.68 (d J = 8.0 Hz, 1H), 4.58 (d J = 12.0 Hz, 1H), 4.55 (d J = 11.0 Hz, 1H), 4.46 (d J = 3.0 Hz, 1H), 4.37 (d J = 11.0 Hz, 1H), 4.15 (d J = 9.0 Hz, 1H), 4.03 (d J = 17.0 Hz, 1H), 3.95 (d J = 17.0 Hz, 1H), 3.92-3.89 (m, 2H), 3.74-3.69 (m, 2H), 3.41 (dd J = 3.5, 9.5 Hz, 1H), 3.37 (t J = 9.5 Hz, 1H), 3.20 (s, 3H), 1.30 (d J = 6.5 Hz, 3H). ¹³C NMR (CDCl₃): δ 169.4, 165.6, 165.2, 138.8, 138.2, 137.1, 133.3, 133.1, 132.9, 130.5, 129.7, 128.5, 128.41, 128.37, 128.34, 128.1, 127.9, 127.8, 127.7, 127.6, 117.4, 113.5, 101.4 (¹J_CH = 157.4 Hz), 97.9 (¹J_CH = 173.2 Hz), 82.0, 79.8, 77.6, 75.6, 74.7, 73.4, 72.1, 71.6, 69.6, 69.4, 68.3, 55.0, 39.1, 16.2. HRMS (ESI): m/z calcd. for C₅₇H₅₅NO₁₃ (M + Na)⁺ 984.3571, found 984.3568. 83: Clear oil. [α]²⁴_D: +8.6 (c 0.5, CHCl₃ ). IR (thin film): 2224 (CN), 1728 (CO) cm⁻¹ . ¹H NMR (CDCl₃): δ 7.91 (dd J = 1.0, 8.5 Hz, 2H), 7.88 (dd J = 1.0, 8.5 Hz, 2H), 7.54-7.40 (m, 4H), 7.36 (t J = 7.0 Hz, 2H), 7.33-7.23 (m, 15H), 7.18-7.10 (m, 4H), 5.84 (dd J = 8.0, 10.0 Hz, 1H), 5.23 (dd J = 3.5, 10.0 Hz, 1H), 4.90 (d J = 11.0 Hz, 1H), 4.79 (dd J = 4.0, 6.5 Hz, 1H), 4.71 (d J = 12.0 Hz, 1H), 4.70 (d J = 10.5 Hz, 1H), 4.68 (d J = 8.5 Hz, 1H), 4.61 (d J = 12.0 Hz, 1H), 4.58 (d J = 12.0 Hz, 1H), 4.43 (d J = 4.0 Hz, 1H), 4.42 (d J = 11.0 Hz, 1H), 4.28 (d J = 4.0 Hz, 1H), 4.22 (dd J = 1.0, 12.5 Hz, 1H), 4.16 (dd J = 1.5, 11.0 Hz, 1H), 3.90 (t J = 8.5 Hz, 1H), 3.88 (dd J =2.0, 13.0 Hz, 1H), 3.74 (ddd J = 1.5, 5.0, 10.0 Hz, 1H), 3.67 (dd J = 5.0, 11.0 Hz, 1H), 3.49 (d J = 1.5 Hz, 1H), 3.38 (dd J = 3.5, 9.5 Hz, 1H), 3.34 (dd J = 8.5, 9.5 Hz, 1H), 3.30-3.20 (m, 2H), 3.19 (s, 3H). ¹³C NMR (CDCl₃): δ 165.8, 165.2, 139.6, 138.8, 138.3, 138.2, 133.4, 133.0, 132.4, 132.2, 132.0, 129.9, 129.7, 128.5, 128.39, 128.37, 128.1, 127.92, 127.90, 127.6, 127.5, 126.9, 118.1, 113.0, 101.4 (¹J_CH = 163.7 Hz), 100.1 (¹J_CH = 170.0 Hz), 97.8 (¹J_CH = 171.2 Hz), 82.0, 79.8, 77.8, 75.6, 74.7, 73.3, 72.9, 72.5, 69.7, 69.0, 68.4, 68.1, 66.5, 55.0, 39.4. HRMS (ESI): m/z calcd. for C₅₇H₅₅NO₁₃ (M + Na)⁺ 984.3571, found 984.3535.