Glycosylation is one of the most abundant post-translational modifications in eukaryotic cells, leading to structurally diverse biomolecules that mediate cellular interactions in processes ranging from cancer progression to immune response.^[1] With no corresponding genetic code and a plethora of possible structures,^[2] understanding the functions of complex oligosaccharides, as well as isolating them from natural resources, is challenging. Obtaining sufficient amounts of pure and homogenous oligosaccharides^[3] and further modifying isolated oligosaccharides typically requires multiple synthetic steps and purifications. These difficulties have directed research efforts toward the chemical synthesis of oligosaccharides, with a focus on developing methodologies that reduce the number of protecting-group manipulations, control stereochemistry of the glycosidic bonds, and eliminate labor-intensive purifications.

Contemporary advances in synthesis emphasize the efficiency of preparing complex oligosaccharides by using automated synthesis,^[4] one-pot procedures,^[5] and chemoenzymatic processes.^[6] One of the most exciting developments is exemplified in the multi-gram production of iGb3, which was obtained as the free trisaccharide by using E. coli.^[7] This approach holds enormous potential and provides impetus for the concomitant development of generalized methods for conjugating free sugars. Indeed, Castillon et al. recently showed that commercially available iGb3 could be per-O-acetylated (per-O-Ac) and converted into a glycosyl iodide, which underwent glycosylation with the tin acetal of ceramide (Scheme 1).^[8] This finding encouraged us to explore a similar strategy to introduce functionalized linkers that could be useful tools for multivalent display of carbohydrate head groups.^[9] We imagined a conjugation methodology that would begin with readily available per-O-acetylated oligosaccharides derived from either synthetic or natural sources. Activation of the anomeric acetate, and subsequent trapping with a suitable acceptor, would afford the corresponding glycoconjugate after removal of the acetate protecting groups. In this manner, various sugars could be functionalized for the preparation of sugar arrays for high-throughput screening^[10] and applications in chemotherapeutics development.^[11] We were especially interested in controlling the stereochemistry at the anomeric position because the chiral center may affect bioassay results.^[12]

Based upon earlier results from our lab, our efforts focused on the idea of using trimethylene oxide (TMO) as an acceptor. We had reported the unique reactivity of armed ether-protect­ed glycosyl iodides with oxa- and thiocycloalkane acceptors (Scheme 2)^[13] to yield iodoalkyl conjugates with high stereo-control. The S_N2-like displacement provided a convenient way to functionalize fully protected sugars in two consecutive steps.

Preparing benzyl-protected monosaccharides with an anomeric acetate is relatively straightforward, but requires several steps and is not well suited for oligosaccharides. Per-O-silylation of oligosaccharides can be achieved in a single step, but conversion of the anomeric silyl ether into the corresponding glycosyl iodide is often complicated by competing cleavage of internal glycosidic linkages, resulting in a complex mixture of degradation products. In contrast, acetylation is a commonly used protecting group for carbohydrate isolation and purification.^[14] The electron-withdrawing nature of acetyl protecting groups stabilizes glycosidic linkages and provides well-resolved proton NMR spectra, facilitating quality control and compound characterization. However, a major challenge of utilizing ester-protected glycosyl iodides is their lack of reactivity. Per-O-acetylated glycosyl iodides have been shown to survive aqueous extractions, column chromatography, and are even so stable that X-ray crystal structures have been obtained.^[15] The “disarmed”^[16] nature of ester-protected glycosyl iodides necessitates using a promoter (Lewis acids or metals) or heat to afford reasonable yields of the glycosylation products. In addition, C-2 ester-protected glycosyl donors often form orthoesters as a side reaction.^[17] For example, in the case of iGb3 noted above, stannyl alkoxides were needed for efficient addition and a Lewis acid was required to avoid orthoester formation.^{[8, 18]} Other groups have successfully prepared β-linked glycosides from ester-protected glycosyl iodides by using various activating reagents, including NIS/I₂/TMSOTf (NIS=N-iodosuccinimide, Tf=triflate),^[17] I₂/DDQ (DDQ=2,3-dichloro-5,6-dicyano-1,4-benzoquinone),^[19] and NBS/Zn (NBS=N-bromosuccinimide) salts.^[20] These methods all have their merits, but the substrates were limited to monosaccharides and, in most cases, long reaction times were required. These findings suggested to us that a rapid and generalized conjugation methodology for ester-protected glycosyl iodides is still needed.

Herein, we report an anomeric-functionalization methodology that combines per-O-Ac glycosyl iodide formation and I₂-promoted TMO addition under microwave irradiation (Figure 1). The formed β-glycosides were functionalized with an iodopropyl linker at the anomeric position, serving as a useful chemical handle for further manipulations. With the functionalized glycosides in hand, multivalent glycoconjugates and glycopeptides were constructed to demonstrate the utility of this methodology.

Previous reports from our lab showed per-O-TMS glycosyl iodides to be highly reactive donors, which when coupled with various acceptors form glycoconjugates with exquisite stereo-control.^[21] Although the methodology worked well with monosaccharide-derived iodides, TMSI-promoted glycosylation of oligosaccharides was complicated and consistently led to a complex reaction mixture in a short period of time (<30 min). Crude MS analysis revealed bond cleavage, trans-silylation, and desilylation products.^{[21a, 22]} Mindful of the fact that glycosyl iodide reactivity is highly attenuated by acetate protecting groups, we prepared C-6,6′ acetate analogues 1 and 2 by using regioselective silyl exchange technology (ReSET) of per-O-silyl lactose.^[22,23] We hoped that incorporation of the acetates would protect the inter-glycosidic linkage from cleavage and that the presence of the remaining silyl ethers would sufficiently activate the iodide. Remarkably, when compound 1 or 2 was treated with TMSI, the corresponding iodide (3) was obtained without internal-bond cleavage or silyl migration.^[22] Trimethylene oxide (TMO) was then introduced to the iodide and the reaction mixture was kept in the dark for 36 h. After acidic methanolysis to remove the silyl protecting groups and reacetylation for isolation and characterization purposes, the β-TMO adduct (4) was obtained in 56% yield over four steps (Scheme 3). The glycosylation required no promoter or heat, indicating that iodide 3 is a highly reactive donor.

Encouraged by these results, we turned our attention to per-O-acetylated oligosaccharides. Investigations began with the study of β-lactose octaacetate (5β), which was treated with TMSI and monitored by NMR spectroscopy to observe iodide formation. Unlike benzylated or silylated analogues, which formed the corresponding α-iodides within minutes, 5β first formed 6β and then gradually anomerized to the α-iodide (6α) over six to eight hours.^[24] When treating the β-iodide (6β) with 1.5 equivalents of TMO at room temperature, the corresponding orthoester (7) was obtained in 88% yield after eight hours. In contrast, the α-iodide (6α) was unreactive at room temperature; even after five days at 40°C only a low yield of the desired glycoconjugate (4) was produced, along with unreacted 6α (Scheme 4). These results led us to conclude that a suitable promoter that avoids orthoester formation is necessary to optimize the production of 4.

Several groups have observed orthoester formation during glycosylation studies.^{[15b, 19b, 20a, 25]} Common methods to circumvent this side reaction include introducing a Lewis acid or applying heat to the reaction.^{[20b, 26]} Initially, we followed the method reported by Stachulsky et al.,^[15b] utilizing NIS and triflic acid, but only 20–30% yield of 4 was obtained. We then explored I₂-promoted glycosidation^[19] in combination with microwave irradiation to heat the reaction more efficiently.^[27] Having observed significant reactivity differences between the lactosyl iodide anomers, our study evaluated I₂ activation of 6α and 6β independently (Scheme 5). In each case, commercially available per-O-acetylated lactose was treated with TMSI to generate the lactosyl iodide, which was monitored by real-time NMR spectroscopy^[24] or TLC analysis; the two lactosyl iodide anomers are distinguishable by TLC analysis (see the Supporting Information, Figure S1). Once each iodide was in hand, TMO and I₂ were added to the flask and the reaction was subjected to microwave irradiation (Scheme 5). We hypothesized that the anomers might have different reactivity and glycosylation profiles, but the yields were similar, being only slightly higher for the β-anomer. Reactions of 6α and 6β with TMO gave target molecule 4 in 51–57% yield, di-TMO adduct 8 in 12–15% yield, and per-O-Ac lactose 5 in 9–13% yield, suggesting that both anomers proceed through a common reaction intermediate (Scheme 5).

Based upon studies by Field et al.,^[19b] Murakami et al.,^[20a] and Demchenko et al.,^[28] a proposed mechanism of I₂-promoted TMO addition is shown in Figure 2. Both α-iodide and β-iodide can be activated by I₂, forming a better leaving group at the anomeric position. The C-2 acetate is positioned to displace the activated β-triiodide intramolecularly, forming the α-face-blocking oxonium ion. Direct attack by TMO at the anomeric position affords the β-glycoside after iodide opening of the activated oxetane. The small amount of α-TMO-adduct formation may result from TMO attack on the α-face, but this occurs to a lesser extent because neighboring-group participation predominates.

The di-TMO adduct (8) had a similar R_f value as the target mono-TMO adduct (4), making isolation difficult and lowering the yield. Different reaction temperatures (0–70°C) were screened, but compound 8 was observed in all instances. We reasoned that the side product could be forming by means of a radical side reaction and decided to introduce butylated hydroxytoluene (BHT), a common radical scavenger,^[29] to suppress possible radical formation (Table 1). When using CH₂Cl₂ as solvent, compound 4 was obtained in 61% yield, but trace amounts of 8 were still observed (Table 1, entry 1). When changing the solvent to CHCl₃, compound 4 was isolated in 51% yield and compound 8 did not form, as evidenced by TLC and crude MS analysis. However, the reaction in CHCl₃ also gave the α-TMO adduct in trace amounts. This adduct usually co-eluted with per-O-Ac lactose (Table 1, entry 2). Using benzene as the solvent led to target molecule 4 in 54% yield, but it took longer to generate the iodide and the amount of α-TMO-adduct side product slightly increased (Table 1, entry 3).

Taking purification and reaction time into consideration, we decided to use 1.2 equivalents of TMSI in CHCl₃ for iodide formation, 1.5 equivalents of TMO, 1.0 equivalent of I₂, and 50 mol% of BHT, under microwave irradiation at 70°C for 20 min (Table 1 entry 2), as the standard conditions to explore the methodology with different per-O-Ac sugars (Table 2). Commercially available per-O-Ac monosaccharides (9β, 10β, 10α, and 11α) were first subjected to the glycosyl iodide/TMO addition reaction. Iodide formation time was established at the time the per-O-acetylated starting material had been completely consumed. In general, the β-acetates (9β and 10β) reacted faster than the α-acetates (10α and 11α) in TMSI-promoted iodide formation,^[30] and led to slightly better yields of the corresponding TMO adduct (Table 2, entries 1 and 2 versus Table 2 entries 3 and 4). β-Glucose pentaacetate (9β) and β-galactose pentaacetate (10β) reacted to form the corresponding iodides in less than one hour, and afforded the corresponding TMO adducts (13 and 14) in 84 and 87% yields, respectively (Table 1, entries 1 and 2). Galactose was slightly more reactive than glucose in both iodide formation and TMO glycosidation. In contrast, 10α required a longer reaction time (2.5 h) to be completely consumed in the iodide formation step. After I₂-promoted TMO glycosylation, the reaction afforded 14 in a slightly lower yield of 79% (Table 1, entry 3). We attributed the increased reactivity of the β-anomers to neighboring-group participation. Nevertheless, using either α- or β-acetates gave rise to the same β-TMO adduct (14), providing further evidence of common reaction intermediates. The methodology was extended to include α-mannose pentaacetate (11α). Iodide formation of 11α required 4.5 h to complete, and the corresponding α-TMO adduct (15) was obtained in 71% yield (Table 2, entry 4).

Not only could we apply the methodology to the monosaccharides and per-O-Ac lactose (5β), but also the per-O-Ac globotriaose (12), which was obtained from ReSET modules.^[22] Compound 12 was a mixture of anomeric acetates in a ratio of α/β=1:5. The iodide formation was completed within one hour and led to the corresponding β-TMO adduct (16) in 72% yield. It is worth noting that the per-O-Ac globotriaose (α/β=1:1) starting material could be isolated in the same reaction. Based on the recovery of starting material, the yield of the desired β-TMO adduct (16) was 81% (Table 2, entry 6). These encouraging results illustrate the broad potential for modification of other per-O-acetylated oligosaccharides, whether obtained commercially or isolated from a natural source.

With the iodopropyl-containing oligosaccharides in hand, we started exploring further applications of the functionalized oligosaccharides. Both compounds 4 and 16 were transformed to azide compounds 17 and 19 in over 90% yields. After global deacetylation by using NaOMe/MeOH, compounds 18 and 20 were obtained in quantitative yields and contained useful “chemical handles” for further modifications (Scheme 6).

Cu-catalyzed azide–alkyne cycloaddition (CuAAC)^[31] allowed us to construct multivalent glycoconjugates. Tetraalkyne 21^[32] was chosen as the core structure for the display. When azido lactose 17 was stirred with 21 in DMF/H₂O with CuSO₄·5H₂O, sodium ascorbate and tris-(benzyltriazolylmethyl)amine (TBTA),^[33] the protected lactose tetramer was obtained after 20 h. The protected lactose tetramer was treated with NaOMe/MeOH for one hour to afford the fully deprotected lactose tetramer (22) in 78% yield over two steps. When using azido triglobotriaose 19 as the reagent, globotriaose tetramer 23 was obtained in 60% yield after the two-step CuAAC–deprotection procedure (Scheme 7). Both compounds 22 and 23 were water soluble, and served as important substrates for ongoing biological studies in our laboratory.

Previously, our lab demonstrated using copper-catalyzed azide–alkyne cycloaddition to construct a synthetic trivalent mimotope of HIV gp120 conjugated to pan allelic HLA DR binding epitope (PADRE), which displayed enhanced binding affinity to HIV-1-neutralizing monoclonal antibody, MAb b12.^{[32a, 34]} PADRE,^[35] a known artificial T-helper (T_H) epitope containing thirteen amino acids, has not only been used in HIV research, but also in cancer-vaccine development as an immunogenic carrier to stimulate the immune response. The combination of tumor-associated carbohydrate antigens (TACAs) and PADRE has been demonstrated by Dumy et al. as a new family of immunostimulants.^[36] This new class of multivalent glycoconjugates could also provide promising antitumor-vaccine candidates in the future.^{[11c, 37]} Applying the TMO-addition concept and click chemistry, trimeric globotriaose–PADRE conjugate 25 was constructed as a potential cancer-vaccine candidate (Scheme 8). Following the published protocol,^{[32, 38]} compounds 20 and 24^[32] were reacted with sodium ascorbate, aminoguanidine, CuSO₄·5H₂O, and tris(3-hydroxypropyltriazolylmethyl)amine ligand (THPTA) in PBS buffer pH 7.4 containing 5% DMSO, under O₂-free conditions overnight. After desalting, purification, and lyophilization procedures, the trimeric globotriaose–PADRE conjugate (25) was obtained as a fluffy white solid in 75% yield, and was analyzed by MALDI-TOF MS and analytical HPLC (see the Supporting Information, Figure S2 and S3). The biological activity of this construct is currently under investigation.

A rapid, step-economical, one-pot methodology for the functionalization of per-O-Ac oligosaccharides has been developed. The methodology integrates in situ iodide formation by using TMSI and I₂-promoted TMO addition to form iodopropyl-linked per-O-Ac β-glycosides. A proposed mechanism for the I₂-promoted TMO addition by using per-O-Ac iodide highlights neighboring-group participation at the C-2 position. The mechanism explains the observed stereochemistry, and microwave irradiation hastens the process allowing efficient conjugation within 20 min, compared to hours or days in previously reported methods.

The iodopropyl linker serves as a useful chemical handle for further manipulation. One application is to transform the terminal iodide into the corresponding azide for click chemistry. With the proper alkyne cores, multivalent glycoconjugates were readily prepared in respectable yields. These multivalent glycoconjugates (23 and 25) contain tumor-associated carbohydrate antigens (TACAs) and immuno-stimulating peptides (PADRE), presenting potential applications as cancer-vaccine or adjuvant candidates.