Synthesis of 4” manipulated Lewis X trisaccharide analogues

Three analogues of the Lex trisaccharide antigen (β-D-Galp(1→4)[α-L-Fucp(1→3)]-D-GlcNAcp) in which the galactosyl residue is modified at O-4 as a methyloxy, deoxychloro or deoxyfluoro, were synthesized. We first report the preparation of the modified 4-OMe, 4-Cl and 4-F trichloroacetimidate galactosyl donors and then report their use in the glycosylation of an N-acetylglucosamine glycosyl acceptor. Thus, we observed that the reactivity of these donors towards the BF3·OEt2-promoted glycosylation at O-4 of the N-acetylglucosamine glycosyl acceptors followed the ranking 4-F > 4-OAc ≈ 4-OMe > 4-Cl. The resulting disaccharides were deprotected at O-3 of the glucosamine residue and fucosylated, giving access to the desired protected Lex analogues. One-step global deprotection (Na/NH3) of the protected 4”-methoxy analogue, and two-step deprotections (removal of a p-methoxybenzyl with DDQ, then Zemplén deacylation) of the 4”-deoxychloro and 4”-deoxyfluoro protected Lex analogues gave the desired compounds in good yields.


Introduction
A glycolipid displaying the dimeric Le x hexasaccharide (dimLe x ) has been identified as a cancer associated carbohydrate antigen, particularly prevalent in colonic and liver adenocarcinomas.In addition, an association between the fucosylation of internal GlcNAc residues in polylactosamine chains, and metastasis and tumor progression in colorectal cancers has been suggested [1][2][3][4][5][6].Unfortunately, dimLe x displays the Le x tri-saccharide (β-D-Galp(1→4)[α-L-Fucp(1→3)]-D-GlcNAcp), at the nonreducing end, and this antigenic determinant is also present at the surface of many normal cells and tissues, which include kidney tubules, gastrointestinal epithelial cells, and cells of the spleen and brain [7][8][9][10][11].Indeed, anti-Le x monoclonal antibodies (e.g., FH3, SH1) have been shown to recognize this Le x antigenic determinant as it exists in the hexasaccharide [1][2][3][4][5][6].Therefore, as our group embarks on the development of a therapeutic anticancer vaccine utilizing the Tumor Associated Carbohydrate Antigen (TACA) dimLe x as a target, an important factor to consider is an expected autoimmune response to the native Le x antigen.Fortunately an internal epitope displayed by dimLe x was discovered when monoclonal antibodies (mAbs) FH4 and SH2, raised against dimLe x , were isolated.Indeed, these mAbs were shown to bind to the dimLe x and trimLe x antigens but only weakly recognise Le x trisaccharide antigen [1][2][3][4][5][6].With this in mind, we focus our research on the discovery of analogues of dimLe x that can be used as safe vaccine candidates.Ideally, these analogues should display the internal epitopes that are recognized by anti-dimLe x SH2-like antibodies while being free of those that are recognized by anti-Le x SH1-like antibodies.
In order to abolish cross-reactivity with the Le x antigen, we have prepared [12][13][14] several analogues in an attempt to identify a suitable replacement for the nonreducing end Le x trisaccharide.In turn, we have compared the conformational behaviour of these analogues to that of the natural Le x -OMe 1 (Figure 1) through a mixture of stochastic searches and NMR analyses [15].The results pointed toward the preferential adoption, by all of analogues, of the stacked conformation that has been assigned for the Le x trisaccharide [16][17][18][19][20][21].The relative affinity of the anti-Le x monoclonal antibody SH1 [6] for the native Le x antigen 1 and our Le x analogues [12][13][14] was examined by competitive ELISA experiments using a Le x -BSA glycoconjugate as the immobilized ligand [15,[22][23][24].It was discovered that the Le x analogue 2, with a glucose unit in place of the galactose residue (Figure 1), did not bind to the SH1 mAb, even at high concentrations [15].This discovery suggests that, to conserve cross-reactivity with the natural Le x antigen, the nonreducing end galactose is essential, and that modifying this residue, particularly at O-4, may lead to the complete loss of this cross-reactivity [15].Currently, it is not known what the reason for this observed loss of binding is, since the binding affinities between proteins and carbohydrates are the result of a combination of factors [25][26][27][28][29].One of the main interactions is the formation of hydrogen bonds, either direct or water-mediated, between the amino acid residues of the protein and the key binding hydroxy groups of the ligand, which are arranged in clusters presented by different monosaccharide units.Other factors include favourable interactions of the nonpolar amino acid residues with the hydrophobic patches exhibited by the ligand, as well as high-energy water molecules being favourably displaced from the combining site.Binding affinity is therefore a result of combined enthalpic, entropic and solvation effects, frequently leading to a balance between favourable enthalpic and unfavourable entropic contributions [25][26][27][28][29]. Thus, only additional competitive ELISA studies with Le x -OMe analogue inhibitors containing strategic manipulations at the 4" site will provide further insight into specific binding interactions [15].The synthesis of a 4"-deoxy Le x trisaccharide analogue was reported recently by Dong et al. [30].Here, we report the synthesis of 4"-methyloxy, 4"-deoxychloro and 4"-deoxyfluoro Le x -OMe analogue inhibitors 3-5.

Synthesis of monosaccharide building blocks 8-11
Glucosamine acceptor 8 was prepared in two steps from the known [14] benzylidene 7: the benzylidene acetal was first hydrolyzed, and then the resulting diol (79%) was selectively benzoylated at O-6 (BzCl-collidine) giving acceptor 8 in 62% yield.The syntheses of trichloroacetimidate donors 9-11 are described on Scheme 1; they were all prepared from the known trichloroethyl galactoside 14 [57].Galactoside 14 was first prepared in three steps from galactose: (1) peracetylation (Ac 2 O-pyridine); (2) BF 3 •OEt 2 activation of the anomeric acetate and glycosidation with trichloroethanol; and (3) Zemplén deacetylation.This sequence of reactions gave the desired galactoside 14 in 78% yield and as a 9:1 α/β mixture, as assessed by 1 H NMR. It is important to point out that the second step in this sequence of reactions used conditions very similar to those used by Risbood et al. to prepare peracetylated trichloroethyl galactopyranoside from peracetylated galactose.Indeed, in agreement with their work [57], the ratio of α-anomer isolated here suggests a late anomerization of the β-glycoside during our extended reaction time (20 h) at reflux.The 4-methyloxy trichloroacetimidate donor 9 was then prepared in four steps from the anomeric mixture of galactoside 14.Tetraol 14 was stirred in a mixture of pyridine and dichloromethane at −10 °C and treated with 3.1 equivalents of pivaloyl chloride.Under these conditions the α-tripivaloate 15, selectively acylated at O-2, O-3 and O-6, was obtained pure and free of β-anomer (64%).The free hydroxy group in alcohol 15 was then deprotonated with sodium hydride and allowed to react with methyl iodide, yielding the 4-OMe galactoside 16 in very good yield.In turn, trichloroethyl galactoside 16 was converted to the trichloroacetimidate donor 9 in two steps: the anomeric trichloroethyl group was removed (Zn/AcOH), and then the resulting hemiacetal was treated with trichloroacetonitrile in the presence of DBU giving the desired α-trichloroacetimidate in good yield.
A Lattrell-Dax nitrite mediated inversion [58][59][60] of the 4-OH in galactoside 15 provided the glucoside 17, which was used as the common precursor to analogues 18 and 19.Treatment with sulfuryl chloride [61] gave the 4-chloro galactoside 18 in good yield, whereas triflation at O-4 followed by S N 2 displacement of the triflate by using tetrabutylammonium fluoride [62,63] gave the 4-fluoro galactoside analogue 19 in excellent yields.As described above for the preparation of donor 9 from glycoside 16, trichloroethyl galactosides 18 and 19 were, in turn, converted in two steps (Zn/AcOH then Cl 3 CCN/DBU) to the trichloroacetimidate donors 10 and 11, respectively, which were obtained in acceptable yields over the two steps.

Galactosylation at O-4 of N-acetylglucosamine acceptors
It has been well established that the hydroxy group at C-4 of N-acetylglucosamine is a poor nucleophile, with reduced reactivity toward glycosylation [64][65][66].However, we have reported the successful O-4 glycosylation of an N-acetylglucosamine monosaccharide acceptor using peracetylated gluco-and galactopyranose α-trichloroacetimidate donors under activation with 2 equivalents of BF 3 •OEt 2 at room temperature or 40 °C [14,67].We applied similar conditions to prepare disaccharides 20-22 (Table 1).Glycosylation of methyl glycoside 6 with the 4-methoxy donor 9 gave the best results when the reaction was carried out at 40 °C and left to proceed for 2 hours.Under these conditions, the desired disaccharide 20 was isolated in acceptable yields (Table 1, entries 1 and 2).Our attempts to reduce the number of equivalents of donor 9 used in the reaction always resulted in a lower yield of the desired disaccharide.Glycosylation of acceptor 8 with the 4-chloro galactosyl donor 10 appeared to be slower (Table 1, entries 3-5) than that of acceptor 6 with donor 9.The best results were obtained when the reaction was left to proceed for 3 rather than 2 hours (Table 1, entry 4), and the desired disaccharide 21 was then obtained in acceptable yield.Increasing the temperature to 60 °C did not increase the yield, presumably due to the degradation of the glycosyl donor at this higher temperature (Table 1, entry 5).Of the three glycosylations considered here, the coupling of acceptor 8 with the 4-fluoro donor 11 gave the highest yields (Table 1, entries 6 and 7).Indeed the desired disaccharide 22 was obtained in very good yields when the reaction was allowed to proceed for 2 hours at 40 °C.Once again, increasing the temperature to 60 °C offered no advantage and in fact led to a lower yield of the desired product.
From these three reactions, it is clear that the substituent at O-4 of a galactosyl donor impacts the outcome of glycosylation at O-4 of N-acetylglucosamine acceptors.Indeed, we have previously observed that galactosylations of such acceptors [68,69] usually result in lower yields (~70%) than those for analogous glucosylations, which usually provided around 90% of the desired disaccharides [14,67].The results described here indicate that 4-OAc galactosyl donors perform better than the 4-Cl donor 10, whereas the 4-OMe donor 9 performs as well as the 4-OAc analogues.In addition, of all of the galactosyl donors employed thus far in such reactions, the 4-fluorinated analogue seemed to perform the best.Thus the reactivity of galactosyl trichloroacetimidate donors towards the BF 3 •OEt 2 -promoted glycosylation at O-4 of N-acetylglucosamine glycosyl acceptors seems to follow the ranking 4-F > 4-OAc ≈ 4-OMe > 4-Cl.

Preparation of the protected Le x trisaccharide analogues
The synthesis of protected trisaccharide intermediates 26-28 is described in Scheme 2. First, acceptors 23-25 were prepared in good yields through the selective removal of the chloroacetate (thiourea) in disaccharides 20-22.Fucosylation of acceptor 23 with ethylthioglycoside 12 was first attempted under NIS and TMSOTf activation at low temperature (−30 °C).Under these conditions, the desired trisaccharide 26 was isolated in 78% yield but as an 8:2 mixture of the α and β-anomers as estimated by 1 H NMR.Although the desired anomer 26α could be obtained pure upon purification by HPLC, it was isolated in a less than desirable yield of 48%.We thus attempted the coupling of acceptor 23 and donor 12 under activation with excess MeOTf (5 equiv).Indeed, we have reported that such conditions allow glycosylation at O-4 of N-acetylglucosamine acceptors through the in situ formation of the corresponding N-methylimidate, temporarily masking the N-acetyl group in the acceptor [70,71].Thus, we expected a similar in situ formation of the methyl imidate in acceptor 23, which would further undergo fucosylation at O-3.However, since methylimidates are unstable when purified on silica gel, they were converted back to the acetamido before purification.Thus, once TLC had shown that all of the acceptor had been consumed, the reaction was worked up and the crude mixture was treated with Ac 2 O-AcOH prior to purification by chromatography [70,71].Under these conditions, the desired trisaccharide 26α was obtained pure and free of the β-anomer in 77% yield.
Similar reaction conditions were applied for the glycosylation of acceptors 24 and 25 with ethylthioglycoside 13.Interestingly, these fucosylation reactions proved to be slower and required additional equivalents of donor 13 to proceed.However, after treatment of the reaction mixtures with AcOH-Ac 2 O, the desired trisaccharides 27 and 28 were isolated in good yields (Scheme 2) and exclusively as the α-fucosylated trisaccharides.As previously observed for other similar analogues [66,72,73], careful analysis of the 1 H NMR spectra acquired for trisaccharides 26α, 27, 28 indicated that the glucosamine residue adopted a conformation distorted from the usual 4 C 1 chair conformation.
The distorted conformations of the N-acetylglucosamine ring in analogues 26α, 27, 28 were characterized by vicinal coupling constants of 6.2-6.6 Hz measured between the ring hydrogens H-2 to H-5 of this residue, rather than the expected values of 8.0-8.3Hz as observed, when measurable, for the same hydrogens in disaccharides 20-25.In addition, although signal overlap precluded its measurement in trisaccharides 26α and 28, the vicinal coupling constant measured between H-1 and H-2 in trisaccharide 27 (5.2Hz) also supported a distorted conformation for this residue (compare to the same coupling constant in compounds 23-25, ~7. 4 Hz).

Deprotection of trisaccharides 26-28
As described previously, the removal of various protecting groups, such as pivaloyl and benzyl groups here, can be accomplished efficiently in one step under Birch reduction conditions [15,68,69,74,75].Thus, treatment of trisaccharide 26α with sodium in liquid ammonia at −78 °C was followed by neutralization of the reaction mixture with AcOH and purification by gel permeation chromatography on a Biogel P2 column (water) to give the desired deprotected 4"-methoxy trisaccharide analogue 3 pure in 83% yield.Since such conditions were not compatible with the 4-chloro and 4-fluoro substituents in trisaccharides 27 and 28, these intermediates were converted in two steps to the desired deprotected analogues 4 and 5, respectively.Thus, removal of the p-methoxybenzyl group with DDQ in CH 2 Cl 2 /H 2 O (15:1 v/v) was followed by Zemplén deacylation, giving the target Le x analogues 4 and 5 in 78% and 75%, res-pectively, over the two steps.The structures of the final deprotected trisaccharides 3-5 were confirmed by HR-ESI mass spectrometry and NMR.

Conclusion
We describe here the efficient synthesis of three Le x methyl glycoside derivatives (3)(4)(5) in which the galactosyl 4-hydroxy group is either methylated (3) or replaced by a chlorine (4) or fluorine (5).Our results seem to indicate that galactosylation at O-4 of an N-acetylglucosamine acceptor under activation with excess BF 3 •OEt 2 can be significantly affected by the nature of the substituent present at C-4 of the galactosyl donor.Indeed, the best results were obtained with the 4-fluoro galactosyl donor, whereas the 4-chloro donor reacted less efficiently than the 4-O-methyl or 4-O-acetyl donors.Overall, this study also confirms our observation [68], that galactosylations at position 4 of N-acetylglucosamine acceptors are usually less successful than glucosylations [14,67].The final Le x -OMe analogues will be used as competitive inhibitors in future ELISA experiments to provide a better understanding of the binding process between the anti-Le x monoclonal antibody SH1 and the Le x antigen.