Multigram synthesis of an orthogonally protected pentasaccharide for use as glycan precursor in a Shigella flexneri 3a conjugate vaccine: application to a ready-for-conjugation decasaccharide

The rapidly growing interest for carbohydrate-based bioactive molecules calls for strategies enabling appropriate design and large scale delivery of the glycan moiety. Here, we described a robust and high-yielding chemical...


Introduction
Understanding the importance of carbohydrates as mediators of biological processes has substantiated major advances in oligosaccharide synthesis to overcome limitations associated to isolates from natural sources. 1 Various strategies are being explored, among which enzymatic, chemo-enzymatic and chemical routes. The latter feature the largest versatility in providing an access to natural as well as non-natural oligosaccharides. Programmable, one-pot solution phase and methods for solid phase and HPLC-assisted automated strategies exemplify some of the major ongoing investigations to accelerate the chemical synthesis of complex glycans. [2][3][4][5] Significant developments have facilitated the expedited synthesis of diverse well-defined oligosaccharides, including homopolymers of increasing chain length, [6][7][8] and the total synthesis of the largest chemically assembled polysaccharide to date, [9][10][11] paving the way to useful probes for investigating further yet poorly understood carbohydrate-mediated vital biological events. When considering large heteropolymers and highly branched targets, solution phase iterative block synthesis has remained an attractive strategy. [12][13][14][15][16][17][18][19][20] In particular, the successful delivery of glycans featuring several repeats strongly relies on the identification of building blocks empowering iterative homologation with high and reproducible glycosylation yields, while also obeying regio-and stereoselectivity criteria in addition to qualifying for efficient full deprotection. Another significant challenge for relevant building block design stems from the need for a synthesis enabling the large-scale production of these essential intermediates to subsequently deliver usable amounts of the extended glycan targets. 21,22 Herein, we tackle this relevant issue in the context of vaccine development.
Shigellae are Gram negative bacteria and the cause of shigellosis, a major diarrheal disease responsible for a high burden notably among children aged 1-5 years living in low-and middle-income settings. 23,24 Shigella is on the WHO pathogen priority list. Epidemiological data, among which the increasing antimicrobial resistance observed among field isolates, call for the development of a multivalent Shigella vaccine. 23,25 Toward this goal, conjugate vaccines based on the bacterial polysaccharide antigens, or surrogates thereof, have been the subject of major interest. 26 As part of the ongoing developments, we have proposed the first synthetic glycanprotein conjugate vaccine candidate against endemic shigellosis. 27 While many antibacterial glycovaccine candidates use haptens corresponding to one repeating unit of the homologous natural polysaccharide antigens, 28 the selected glycoconjugate prototype comprises a chemically synthesized pentadecasaccharide corresponding to a three core repeating unit portion from the Shigella flexneri 2a O-antigen (O-Ag). 29 It was produced according to good manufacturing practice and  30 demonstrated to be safe and immunogenic in adult volunteers in the frame of first-in-human clinical trial. 31 These achievements have provided strong support to serotype broadening. In this context, our current efforts aim for a vaccine candidate against S. flexneri 3a (SF3a), another prevalent Shigella serotype for which a vaccine is in high demand. 24 The SF3a O-Ag is made up of a branched pentasaccharide repeat (E)AB Ac C Ac D ( Figure 1), featuring (12)-trans-linked Lrhamnoses (A, B, C) and a N-acetyl-D-glucosamine residue (D). Rhamnose A is 3-O--D-glucosylated (E). Acetylation at position 2 C is stoichiometric. In contrast, position 6 D is O-acetylated to a 40% extent only. 30 Epitope mapping has revealed the immunodominant 2 C -O-acetyl (Ac) moiety and the importance of chain length for protective antibody recognition. 32 Molecular modeling simulations supported by NMR analysis of O-Ag segments from 12 S. flexneri serotypes featuring the same backbone, among which those relevant to S. flexneri 2a and SF3a, suggested similar backbone conformational behavior. 33 This study also revealed a dynamic behavior of the end-chain -D-glucopyranosyl residue (13)-linked to rhamnose A differing from that predicted for glucose side-chains located on internal repeats. 33 Overall, convincing evidences support the assumption that oligosaccharides achieving SF3a O-Ag functional mimicry encompass at least two repeating units. Otherwise, the role of the non-stoichiometric 6 D -O-acetylation remains undisclosed. Aiming at establishing a lead hapten candidate for SF3a vaccination, we report a straightforward multi-step chemical synthesis of pentasaccharide 1, 17 as the lead common precursor to the (E)AB Ac CD and (E)AB Ac C Ac D modules, their combinations and oligomers thereof, as found in the native SF3a O-Ag (Scheme 1). Going beyond our previous disclosures while aiming at scalability and robustness, the orthogonally protected pentasaccharide building block was produced in several 10gram amounts. Emphasis was put on (i) restraining the handling of toxic and poor user-friendly reagents, in particular by circumventing the notoriously questionable tin chemistry and by avoiding concerns related to hydrazine and its derivatives especially when involved at an advanced stage of a multi-step synthesis, (ii) limiting the repeated use of low abundant catalysts despite their remarkable potential as exemplified with iridium-based compounds, and (iii) reducing the number of demanding purification steps involving column chromatography by promoting crystalline intermediates and fine-tuning of reaction parameters, while (iv) achieving high yielding conversions fulfilling regio-and stereoselectivity criteria. It is well-appreciated that concern for the latter increases when addressing glycosylation steps involved in large oligosaccharide blockwise synthesis. Herein, significant inputs feature handy metal-catalyzed protecting group manipulation, advanced crystalline intermediates, fine-tuned 1,2-cis and block glycosylation steps, and a meaningful reduction of the number of column chromatography, the latter being known to qualify as a bottleneck when aiming at large-scale synthesis. 22 Furthermore, the proof-of-concept is established as the potential of the selected pentasaccharide building block is next demonstrated in the synthesis of a ready-for-conjugation linker-equipped decasaccharide corresponding to a dimer of the repeating unit of the SF3a O-Ag.

Results and discussion
Building block 1 was designed as an allyl glycoside, allowing easy conversion into a donor or an acceptor. 17 It is 2 C -O-acetylated as in the SF3a O-Ag. In contrast, the second site of natural Oacetylation is masked as a 4 D ,6 D -O-benzylidene (Bzl) acetal, allowing for the chemoselective late stage modification at OH-6 D . Non-interfering hydroxyl groups are benzylated, and the site of elongation (OH-2 A ) features a levulinoyl ester, which fulfills criteria for stability, anchimeric assistance and orthogonality, in particular to the 2 C -acetate. 34  pentasaccharide 1 is readily accessible from the known B Ac CD and EA allyl glycosides, 2 17 and 3, 35 respectively. These key intermediates are obtained from are commercially available in bulk amounts. Substantiating our previous report, 36 diol 5 is routinely obtained in at least 90 g amount in four steps and over 80% yield (Scheme 1). It performs as an exquisite common precursor to the known acceptor A/C (14) and donor B (13). 35

Synthesis of the BCD trisaccharide 2.
Going beyond the original tin-mediated regioselective benzylation of 1,2-cis diols 37 and the inherent toxicity of tin reagents used in stoichiometric amount, elegant procedures enabling the site-selective modification of carbohydrates have been developed. 38,39 The recently reported iron(III)-based catalysts, Fe(dibm) 3 , offering high regioselectivity, broad scope and high reactivity, 40 and its cheaper although equally efficient analog Fe(dipm) 3 , 41 called our attention (Scheme 2). Readily obtained from the inexpensive FeCl 3 ‧6H 2 O, these reagents are considered non-air sensitive, non-toxic and environmentally benign. 42 Gratifyingly, Fe(dibm) 3 -promoted benzylation of diol 5 in the absence of additive proceeded at 80 °C in acetonitrile to give the desired alcohol 7 (92%) together with its regioisomer 7a (5%). Satisfactorily, Fe(dipm) 3 performed as well. The 19:1 regioselectivity compares nicely with the 87% yield achieved using tin chemistry. 35 Advantageously, purification is simpler. Next, instead of using a large excess of levulinic anhydride prepared up front, Steglich esterification of alcohol 7 gave levulinate 8, 35 which was in turn deallylated into hemiacetal 9. 35 As an attempt to avoid the previously adopted efficient, albeit expensive, [Ir(COD){PCH 3 (C 6 H 5 ) 2 } 2 ] + PF 6catalyst and its necessary hydrogen-mediated activation ( Table 1, Entry 1), 17 we have favoured the use of more Earth-abundant metal catalysts, focusing primarily on well-explored palladium derivatives (Table  1) amid numerous possible reagents, [43][44][45] to complete the anomeric deallylation step. 46 Unexpectedly, Pd(PPh 3 ) 4 used in combination with mild acids 47 led at best in partial conversion to propen-1-yl 10 (Entries 2 and 3). Therefore, established protocols involving Pd(II) catalysts, which are generally more stable and less expensive than Pd(0) derivatives, were considered instead. Diverging from previous observations, 48 PdCl 2 in buffered AcOH/AcONa was low-yielding (Entry 4). Although the phenomenon was barely reported, methyl glycoside 11 was repeatedly isolated when using PdCl 2 in methanol (Entries 5 and 6), while the Wacker-type products 43,45 12a/12b were formed in DMF (Entries 7 and 8). We reasoned that changing DMF to a non-polar solvent used in combination with water as the proton source would prevent side-oxidation. Indeed, conversion to propen-1-yl 10 was slow, but oxidized 12a/12b were not observed in DCM/H 2 O (Entry 9). Otherwise, changing DMF for THF led to low conversion (Entry 10). Gratifyingly, heating rhamnoside 8 to 50 °C for 2 h in DCM/H 2 O (3:1) containing PdCl 2 (4 mol%) allowed faster completion and provided hemiacetal 9 in quantitative yield post iodine addition (Entry 11). These yet unreported easy-to-handle conditions were adopted on the large scale (Entries 12 and 13).  Remarkably, trichloroacetimidate 13 easily obtained by reacting hemiacetal 9 and trichloroacetonitrile in the presence of a base, 49 is now routinely prepared on the 40 g scale (92%) from alcohol 7 in three steps and no intermediate purification (Scheme 3). Donor 13 is stable for at least a month at -20 °C despite being isolated as a syrup.
The stepwise conversion of diol 5 into the BC donor 17 is a robust process (Scheme 3), reaching 69% over four steps on a 5-10 g scale. 17 Herein, this conversion was achieved without intermediate purification reaching an overall yield of 86%, which was proven reproducible upon scaling up. Indeed, donor 17 was isolated in 30 g amount (84%) starting from 13 g of diol 5. Noteworthy features in doing so include the reaction of acceptor 14, readily obtained from diol 5 as a 95:5 mixture of regioisomers, with a reduced excess of donor 13 (1.1 instead of 1.2 equiv.) and the use of the newly established Pd(II)-mediated anomeric deallylation protocol without any yield loss, as demonstrated for the independent conversion of rhamnobioside 15 into trichloroacetimidate 17 (91%).
Otherwise, acceptor 18 (88%) was achieved from tetraacetate 4 in four steps as described. 50 In line with expectation, 17 the TMSOTf-promoted [18+17] glycosylation proved to be highly efficient (Scheme 4). Crystalline B Ac CD 2 of acceptable purity for the next step was isolated in 97% yield from 11 g of crystalline 18 and a slight excess (1.15

Synthesis of the EA donors 34 and 35.
The synthesis of disaccharide 3 was another opportunity for improvement (Scheme 5, Table 2). Originally, the essential 1,2cis EA linkage was achieved from diol 5 as established in the late nineties 51 to give alcohol 24 in 61% yield over three steps (Entry 1). 36 Since then, a better understanding of factors affecting stereoselectivity has guided several reports on strategies addressing the challenge of anomeric control during 1,2-cis glycosylation. 52,53 Therefore, going beyond original achievements, 36 while considering scaling up, easy-toimplement alternatives were explored. Relying on the originally favored glucosyl donor 21 54  realizations, parameters such as the promotor, its amount, and the solvent (Entries 2-7) were varied without any observed meaningful improvement. While avoiding formation of the Chapman rearrangement product 26, 55 use of the (Nphenyl)trifluoroacetimidate (PTFA) donor 22 56 resulted in loss of / selectivity (Entry 11). In agreement with former investigations, changing donor 21 for the corresponding known thiophenyl glycoside 19 57 and fluoride 20 58 met no success (Entries 14 and 15). Next, the possible remote anchimeric assistance of the protecting groups masking on the [E+A] glycosylation outcome 52 was examined. In view of their orthogonal properties and easy access by means of the selective 6-O-debenzylation of hemiacetal 6 (Scheme 6), 59 glucosyl donors bearing a temporary 6-O-acetyl ester or 6-O-tert-butyldiphenylsilyl ether (TBDPS), respectively, were considered. However, the enhanced / ratio expected from long-range 6-O-acyl-assistance using donors 28 60 and 29 or from a foreseeable steric hindrancecontrolled -glucosylation by means of the silylated analogs 31 61 and 32 was not observed in our hands providing the condensation products in at best a 7:3 / ratio (Schemes S2, not described). Leaving aside promising albeit more demanding strategies involving specific protecting group manipulation, 52 we turned to investigate the potential of exogenous nucleophiles to control stereoselectivity 62 when considering solely the more readily available tetrabenzyl donors, and in particular imidates 21 and 22, as the simplest possible E precursors. We prioritized the DMF-modulated glycosylation    more reactive EDC (Scheme 7). This successful in situ activation of levulinic acid advantageously replaced the formerly adopted conditions. 35 Deallylation, whether conventional 35 or using thenewly established aforementioned PdCl 2 protocol, delivered the known hemiacetal 33 35 quantitatively for direct conversion into imidates 34 35 and 35. Alternatively, the fully protected 3 was evolved into those same donors without intermediate purification.
Scaling up this efficient three-step conversion provided trichloroacetimidate 34 in 67% yield in combination with hemiacetal 33 (30%), post chromatography. Obviously, the recovery of a meaningful amount of 33 was attributable to donor hydrolysis on the column, suggesting that careful consideration be given to the purification step for large scale development. Satisfactorily, the more stable PTFA donor 35 was isolated in 40 g amount in an excellent 90% yield over three steps.

Synthesis of the EAB Ac CD pentasaccharide building block 1.
Restraining the number of column chromatographies, the twostep conversion of the fully protected 2 (25 g) into pentasaccharide 1 employed the crude acceptor 36 17 (Scheme  8). Satisfactorily, the independent use of donor 34 or 35 (1.15 equiv.) ensued a good 80% yield from the fully protected 2, or rather a 88-94% corrected yield based on recovered 36. Adding to the overall improved strategy of the (E)AB Ac CD building block (1), this compares favourably with original stepwise achievements. 17 In particular, the proof of concept having been established, we are confident that additional fine-tuning on the two-step conversion on a large scale will contribute to increase further the isolated yield pentasaccharide 1. On the way forward toward this aim, we also envisioned alternatives to hydrazine acetate involving less toxic reagents for the selective delevulination at OH-2 B of the 2 C -O-acetyl B Ac CD precursor (2). While the former remains from far the method most frequently encountered, it is not without drawback. In particular, we have previously observed the partial reduction of the olefinic bond of the allyl aglycon in pentasaccharide 1 concomitant to hydrazinolysis of the 2 A -O-levulinoyl ester. 17 Inspiration from earlier findings, 66 encouraged the investigation of sulfite as a handy reagent. However, resulting at best in incomplete conversion despite prolonged reaction time (not described), neither the original conditions nor their modified version implemented in the context of oligonucleotide synthesis 67 fulfilled our expectations. Optimization was not attempted. Instead, implementation of user-friendly conditions enabling the high-yielding delevulination of trisaccharide 2 took advantage of a previous report from R. Adamo's group. 68 Indeed, replacing hydrazine acetate by the more acceptable ethylenediamine provided alcohol 36 in a selective manner suggesting that these conditions could be adopted in the future (Scheme 8).

Synthesis of the spacer-equipped pentasaccharide 45 and decasaccharide 46.
Having achieved an improved synthesis of the fully protected 1, the next step consisted in ensuring that this key building block fulfilled expectations when evolved into a donor and a linkerequipped acceptor, respectively. Toward this aim, pentasaccharide 1 was submitted to conventional deallylation into hemiacetal 38, which was in turn converted to the corresponding PFTA donor 39 in high yield (Scheme 9). Pleasingly, running the two steps without any intermediate purification also resulted in an efficient conversion, reaching repeatedly over 85% yield on a multigram scale.
Alcohol 37 was used as a model acceptor for the [5+5] glycosylation envisioned next (Table 3). Comforting the promising outcome of the ethylenediamine-mediated delevulination of the BCD trisaccharide 2, it was isolated in an excellent 89% yield upon heating the fully protected 1 in the presence of excess ethylenediamine (Scheme 9). The [39 + 37] coupling proved to be high-yielding in all the conditions that were tested (Table 3). However, tendencies were revealed. In particular, some unconsumed acceptor was always observed when the 39:37 ratio was below 1.3 (Entries 4- 6). Glycosylation proceeded in a large range of temperatures to give decasaccharide 40 in over 80% yield, but formation of an unidentified side-product was observed repeatedly at temperatures higher than -40 °C (Entries 1-3). Changing DCM for toluene while keeping the temperature at -40 °C and using 1.3 equivalents of donor had no obvious influence (Entries 1 and 7). Having identified high yielding glycosylation conditions, we turned to synthesis of the linker-equipped decasaccharide 46.
Glycosylation of donor 39 with 2-azidoethanol was achieved in toluene at -40 °C to give the -linked azidoethyl glycoside 41, which was isolated in a good 78% yield (Scheme 10). Hydrazinolysis of the 2 A -O-levulinoyl ester provided acceptor 42 in a yield equivalent to that obtained for the corresponding allyl glycoside 37. 17 In support to the selection of azidoethyl glycoside 42 as precursor to larger linker-equipped SF3a oligosaccharides, its full deprotection promoted by Pd(OH) 2 in an hydrogen atmosphere was uneventful, permitting the smooth concomitant hydrogenolysis and reduction of all protecting groups in place, to give the expected aminoethyl ARTICLE Please do not adjust margins  pentasaccharide 45 in a satisfactory 70% yield post RP-HPLC chromatography.
Interestingly, transferring the most promising [5+5] glycosylation conditions to the 2-azidoethyl-equipped acceptor revealed that the [39+37] glycosylation was somewhat sensitive to both solvent and temperature (not described). In agreement with original findings, 17 running the condensation in non-polar toluene at -40 °C were identified as the best conditions providing decasaccharide 43 in a reproducible 90% average yield (Scheme 10). Subsequent delevulination gave alcohol 44, which was next submitted to a one-step full deprotection. While enabling the concomitant cleavage of the two 4 D ,6 D -Obenzylidene acetals and 16 benzyl ethers in addition to the simultaneous reduction of the two 2 D -trichloroacetamides and azide moiety, the Pd(OH) 2 -catalyzed hydrogenation/hydrogenolysis of the azidoethyl glycoside 44 in tBuOH/DCM/H 2 O into the aminoethyl decasaccharide 46 was more demanding than that of its counterpart 42 into pentasaccharide 45. The use of a higher Pd(OH) 2 amount combined to a longer reaction time at ambient temperature and pressure furnished the conjugation-ready 46 in a good 52% yield post RP-HPLC (Scheme 10). Nevertheless, the observed drop in the yield of the two O-Ag repeating unit segment 46 versus the one repeating unit oligosaccharide 45 suggested that improvement might be needed for the full deprotection of larger oligomers featuring an aminoalkyl aglycon and a higher number of trichloroacetamide groups.

Conclusions
This study aimed at achieving a robust process enabling the largescale synthesis of pentasaccharide 1, and at demonstrating that this orthogonally protected building block could serve as a suitable precursor to a donor and an acceptor, whose combination would provide ready-for-conjugation oligosaccharides for use in the development of a synthetic carbohydrate-based conjugate vaccine candidate against SF3a. A robust and convenient 26-step synthesis, featuring four crystallizations and only nine column chromatographies -mother liquors included -of pentasaccharide 1 from crystalline 1,3,4-6-tetra-O-acetyl-D-glucosamine, L-rhamnose and tetrabenzyl-D-glucose was described. The upgraded synthesis combines several independent step-specific improvements involving greener, less demanding, more stereoselective and user-friendly protocols, also promoting crystalline intermediates and multigram scale validation. Notably, relevant improvements of interest in a broader context include the implementation of the easy-to-handle PdCl 2 in DCM/H 2 O for high-yielding anomeric deallylation and catalytic Fe(dipm) 3 for the 3-O-etherification step of diol 5. Paving the way to further scale up and vaccine development against SF3a, the (E)AB Ac CD pentasaccharide 1 was readily delivered in over 30 g amounts and subsequently converted into the linker-equipped hapten 45 by means of acceptor 42. Alternatively, the fully protected 1 was efficiently transformed into donor 39. Lastly, the proof-ofconcept for building block selection enabling a robust [5+5] chain elongation strategy from pentasaccharide 1 was successfully demonstrated by delivering the ready-for-conjugation decasaccharide 46, which corresponds to a two repeating unit segment of the SF3a O-Ag. Aminoethyl glycosides 45 and 46 and larger SF3a O-Ag segments, are ideal substrates for use as components of well-defined glycoconjugates for in vivo study, which represents the next aim.

Allyl (2,3,4,6-tetra-O-benzyl-α-D-glucopyranosyl)-(1→3)-4-Obenzyl-α-L-rhamnopyranoside (24). 51 Route 1 (1 g scale).
Trimethyl orthoacetate (820 µL, 6.45 mmol, 1.9 equiv.) and monohydrated PTSA (10 mg, 0.05 mmol, 0.015 equiv.) were added to diol 5 (1.0 g, 3.4 mmol, 1.0 equiv.) in anhyd. MeCN (2.3 mL) at rt. After stirring at rt for 1 h, 80% aq. AcOH (2.3 mL) was added at 0 °C and the mixture was stirred at this temperature for 30 min. TLC (Tol/EtOAc, 70: 30) showed total consumption of the intermediate orthoester. DCM was added along with water and the two layers were separated. The aq. phase was extracted with DCM and the combined organic phases were washed successively with sat. aq. NaHCO 3 and brine, dried over Na 2 SO 4 , filtered and concentrated to dryness to give crude acceptor 14. TfOH (300 µL, 3.4 mmol, 1.0 equiv.) was slowly added to a solution of the latter and trichloroacetimidate 71  Please do not adjust margins extracted with DCM and the combined organic phases were washed successively with sat. aq. NaHCO 3 and brine, dried over Na 2 SO 4 , filtered and concentrated to dryness to give the crude acceptor 14. DMF (5.28 mL, 67.9 mmol, 20 equiv.) and activated 4 Å MS (0.5 g) were added to a mix of the latter and the PTFA donor 22 (3.14 g, 4.42 mmol, 1.3 equiv.) in anhyd. DCM (44 mL) and the suspension was stirred at rt under Ar for 30 min, then at -78 °C for 15 min. TfOH (0.3 mL, 3.40 mmol, 1.0 equiv.) was added very slowly at -78 °C. The reaction mixture was then stirred for 1 h while slowly warming up to rt. TLC (cHex/EtOAc, 70:30) showed the complete disappearance of rhamnoside 14 and the presence of less polar products. The reaction mixture was neutralized with Et 3 N. EtOAc was added along with water and the two layers were separated. The aq. phase was extracted with EtOAc and the combined organic layers were washed successively with sat. aq. NaHCO 3 and brine, dried over Na 2 SO 4 , filtered and evaporated. MeONa (25% in MeOH, 1.17 mL, 5.1 mmol, 1.5 equiv.) was added to the obtained crude in DCM/MeOH (11:8, 50 mL). After stirring at rt overnight, TLC (Tol/EtOAc, 80: 20) revealed that the glycosylation products had reacted and more polar products were present. DOWEX H + resin was added and the mixture was stirred 30 minutes before filtering and washing thoroughly with MeOH. Et 3 N (few drops) were added and volatiles were evaporated. Purification by flash column chromatography (Tol/EtOAc, 100:0 to 90: 10) gave the desired α anomer 24 as a pale yellow oil (2.2 g, 79% over 3 steps) and the commercially available hemiacetal 6 (345 mg).  (27). 60 Ac 2 O/TFA (4:1, 30 mL) was added to hemiacetal 6 (2.5 g, 5.0 mmol, 1.0 equiv.) at 0 °C and the suspension under Ar was stirred at rt for 3 h. TLC (cHex/EtOAc, 60: 40) showed conversion of the starting 6 into less polar products. Cold water (50 mL) was added at 0 °C and the mixture was stirred for 15 min at rt, then neutralized with sat. aq. NaHCO 3 . EtOAc was added and the two layers were separated. The aq. phase was extracted repeatedly with EtOAc and the combined organic layers were washed with sat. aq. NaHCO 3 , brine and dried over Na 2 SO 4 . Volatiles were evaporated and the crude was solubilized in DMF (20 mL). Hydrazine (60% in water, 0.22 mL, 7.0 mmol, 1.5 equiv.) and AcOH (0.40 mL, 7.0 mmol, 1.5 equiv.) were added at rt and the mixture was stirred at this temperature for 20 h. TLC (cHex/EtOAc, 60: 40) showed conversion of the 1,6-di-O-acetyl intermediate into more polar products. EtOAc and water were added and the two layers were separated. The aq. layer was extracted repeatedly with EtOAc and the combined organic phases were washed with sat. aq. NaHCO 3 , brine and dried over Na 2 SO 4 . Flash column chromatography on silica gel (cHex/EtOAc, 100:0 to 50: 50) gave the known hemiacetal 27 as a white solid (/ 65:35, 1.99 g, 87%). An analytical sample was obtained by means of a second purification. 1 (30). 61 A mixture of Ac 2 O/TFA (4:1, 30 mL) was added at 0 °C to hemiacetal 6 (2.5 g, 5.0 mmol, 1.0 equiv.) under Ar and the suspension was stirred at rt for 3 h, at which time TLC (cHex/EtOAc, 60:40) indicated conversion of the starting 6 into less polar products. Cold water (50 mL) was added at 0 °C and the mixture was stirred for 15 min at this temperature, then neutralized with 4 M aq. NaOH. EtOAc was added and the two layers were separated. The aq. layer was extracted with EtOAc and the combined organic phases were washed with brine and dried over Na 2 SO 4 . Volatiles were evaporated and MeONa (25% in MeOH, 5 mL) was added to the crude intermediate stirred in MeOH (20 mL) at rt. After stirring overnight at this temperature, TLC (cHex/EtOAc, 60: 40) showed the complete disappearance of the intermediate and the presence of more polar products. Dowex H + resin was added portionwise while the suspension was gently stirred until neutralisation. The suspension was filtered and volatiles were evaporated. DMAP (0.11 g, 0.93 mmol, 0.2 equiv.), imidazole (0.76 g, 11.1 mmol, 2.4 equiv.) and tertbutyldiphenylsilyl chloride (TBDPSCl, 1.44 mL, 5.55 mmol, 1.2 equiv.) were added to a solution of the crude in DMF (40 mL) at 0 °C. The mixture was stirred overnight at rt, at which time more imidazole (2.0 equiv.) and TBDPSCl (1.1 equiv.) were added. After 3 h, water and Et 2 O were added and the two layers were separated. The aq. phase was extracted with Et 2 O and the combined organic phases were washed with sat. aq. NaHCO 3 , brine and dried over Na 2 SO 4 . Volatiles were evaporated under reduced pressure, two successive purifications by flash column chromatography on silica gel (cHex/EtOAc, 100:0 to 95:5) gave hemiacetal 30 as a colorless oil (mostly α/β mixture, 1.7 g) contaminated with tertbutyldiphenylsilanol (10 mol%). Only the major isomer is described. 1 (2,3,4,6-Tetra-O-benzyl-α-D-glucopyranosyl)-(1→3)-4-O-benzyl-2-O-levulinoyl-α-L-rhamnopyranosyl (N-phenyl)trifluoroacetimidate (35). Levulinic acid (9.7 g, 83 mmol, 2.0 equiv.), EDC (15.5 g, 75 mmol, 1.8 equiv.) and DMAP (3.4 g, 17 mmol, 0.4 equiv.) were added to alcohol 24 (34 g, 42 mmol, 1.0 equiv.) in anhyd. DCM (210 mL). The mixture was stirred at rt for 60 h, at which time TLC (Tol/EtOAc, 80: 20) showed the full consumption of the starting 24 and the presence of a more polar product. The reaction mixture was diluted with water and DCM. The two layers were separated and the aq. phase was extracted with DCM repeatedly. The combined organic layers were washed successively with sat. aq. NaHCO 3 , water and finally brine. The organic layer was dried on Na 2 SO 4 , filtered, and volatiles were evaporated. PdCl 2 (621 mg, 2.1 mmol, 0.05 equiv., 60% purity) was added to the crude 3 in DCM/H 2 O (3:1, 420 mL). The biphasic mixture was stirred at 50 °C for 3 h. TLC (Tol/EtOAc, 80: 20) showed conversion of the starting 3 into a less polar product. Iodine (10.7 g, 84 mmol, 2.0 equiv.) in THF (50 mL) was added slowly to the solution at rt. After stirring at this temperature for 2.5 h, TLC (Tol/EtOAc, 8:2) showed conversion of the intermediate into a more polar product. Sat. aq. Na 2 S 2 O 3 was added and the biphasic mixture was filtered over a pad of Celite  . The organic phase was washed with sat. aq. NaHCO 3 , water and brine. The organic phase was dried on Na 2 SO 4 , filtered, and concentrated to dryness. PTFACl (10 mL, 63 mmol, 1.5 equiv.) and K 2 CO 3 (11.6 g, 84 mmol, 2.0 equiv.) were added slowly to the crude hemiacetal in acetone (420 mL) under Ar, at rt. The suspension was stirred at this temperature for 16 h. TLC (cHex/EtOAc, 70:30) showed the consumption of the intermediate 33 and the presence of less polar products. After filtration over a pad of Celite  , thorough washing of the solids with DCM and concentration of the filtrate to dryness, the residue was purified by column chromatography (cHex/EtOAc, 100:0 to 80: 20) to give donor 35 (39. 20) afforded the desired pentasaccharide 1 as a beige foam (31.5 g, 80% over 2 steps).