Ping Zhangab, Kenneth Ngb and Chang-Chun Ling*a
aDepartment of Chemistry, Alberta Ingenuity Center for Carbohydrate Science, University of Calgary, Calgary Alberta, T2N 1N4, Canada. E-mail: ccling@ucalgary.ca; Fax: +1 (403) 289-9488; Tel: +1 (403) 220-2768
bDepartment of Biological Science, Alberta Ingenuity Center for Carbohydrate Science, University of Calgary, Calgary Alberta, T2N 1N4, Canada
First published on 13th November 2009
The toxins TcdA and TcdB produced by the human pathogen Clostridium difficile gain entrance to host epithelial cells by recognizing cell-surface carbohydrate ligands. Inhibiting the attachment of these toxins to host cells has been proposed to be a viable therapy to treat C. difficile infections. Glycan array screening previously revealed that the LeA-LacNAc pentasaccharide binds strongly to TcdA. Here we report the efficient syntheses of the pentasaccharide and a structurally related tetrasaccharide motif. These compounds will be used to better define the carbohydrate-binding specificity of toxins from C. difficile, which will hopefully lead to the development of improved therapeutics.
![]() | ||
Fig. 1 Structures of ligands bounds to toxin A from CFG screen experiments and synthetic targets. |
![]() | ||
Fig. 2 Retrosynthetic analysis for both tetrasaccharide 2 and pentasaccharide 4. |
Scheme 1 showed the synthesis of N-phthalimido protected glucosamine thioglycosides 13 and 14. The known compound 1623 was converted to 17 in large scale and high yield by the treatment of 16 with boron trifluoride etherate and p-chlorophenylthiol which is a preferred thiol because it is less malodorous compared to the more coventional ethanethiol or thiophenol. We found that if an in situ acetylation step was carried out after the reaction, we could convert the remaining thiol to the corresponding thioacetate thus essentially eliminating the unpleasent smell. In agreement with the literature,24 the β-OAc of 16 was found to react much faster (4 h) than the corresponding α-OAc (up to 72 h). No chromatography was needed to isolate the thioglycoside 17 as a simple recrystallization from a mixture of ethyl acetate–hexane afforded the pure thioglycoside 17 in 76% yield, illustrating the advantage of using p-chlorophenylthiol as the aglycone. The subsequent removal of acetates was carried out using a milder Zémplen transesterification condition22 with a guanidine/guanidinium buffer as the deacetylation reagent in anhydrous methanol; compound 18 crashed out from the reaction media in pure form and in almost quantitative yield (98%). The more conventional condition (sodium methoxide in methanol) was found to be less suitable as it resulted in a partial opening of the phthalimido group. To synthesize compound 14, a transacetalation was carried out on triol 18 with benzaldehyde dimethyl acetal under the catalysis of camphorsulfonic acid to give the desired alcohol 14 in excellent yield (93%). To regioselectively protect the primary alcohol of 18, the stanylidene acetal-mediated benzylation was found to be efficient; thus a solution of 18 with 1 equivalent of dibutyltin oxide in toluene was heated to reflux to form intermediate stanylidene acetal which allowed the installation of a benzyl group at O-6 in a regioselective manner to give compound 19 in 67% yield. To synthesize alcohol 13, another dibutyltin oxide-mediated alkylation was carried out with p-methoxybenzyl (MBn) chloride; compound 13 was obtained in moderate yield (33%) together with the regioisomer 20 which has the MBn group installed at the O-3 position (30% yield).
![]() | ||
Scheme 1 Syntheses of glucosaminyl donors/acceptors. |
The synthesis of the common disaccharide 5 (Scheme 2) began with the preparation of acceptor 22 which was obtained by coupling the 6-azidohexan-1-ol 21 with thioglycoside 19 under the activation of N-iodosuccinimide (NIS)/trifluoromethanesulfonic acid (TfOH).
![]() | ||
Scheme 2 Synthetic route to the common disaccharide acceptor 5. |
Despite the presence of two free hydroxyl groups in donor 19, glycosylation proceeded in a remarkably smooth way to afford 22 in 85% yield. Clearly, the primary hydroxyl group of 21 exhibited sufficiently high reactivity to compete with all inter/intramolecular self-condensation side reactions. Coupling of 12 with 22 under the activation of NIS/TfOH condition gave the desired disaccharide 23 in moderate yields (47%). To improve the glycosylation efficiency, we switched the donor to imidate 11 to afford the desired disaccharide 23 in slightly improved yield (60%) under the catalysis of trimethylsilyl trifluoromethanesulfonate (TMSOTf) at −78° C. Finally, a selective deprotection of acetates was carried out using guanidine/guanidium buffer conditions to afford the triol 5 in high yield (84%).
We next investigated the preparation of βGal(1→3)βGlcNPhth disaccharide donor 6 using the designed acceptor 13. We first attempted the glycosylation with the peracetylated galactopyranosyl imidate 10 in anhydrous dichloromethane (Scheme 3) under the catalysis of TMSOTf. By starting the reaction at 0° C and slowly warming the reaction up to room temperature, surprisingly, no desired glycoside 6 was formed even if we used a large excess amount of 10 (up to 3.0 equivalents). The major compound isolated from the reaction mixture was the unreacted acceptor 13 and a small amount of rearranged product 25 from donor. Attempts to change the reaction solvent to others such as anhydrous acetonitrile also failed. When we employed the more reactive imidate donor 11, we were still unable to observe the formation of the desired disaccharide 24. The imidate donor quickly rearranged to form the glycosyl amide 26 and the acceptor 13 was recovered.
![]() | ||
Scheme 3 Attempts to prepare disaccharide 6 and 24 from 13. |
Previously, it was well-established25 that if a tribenzylated L-fucopyranosyl residue is attached to the O-4 position of a N-phthalimido protected glucosamine acceptor, the OH-3 became inaccessible for glycosylations due to steric hindrance. Here the fact that we used a much smaller and electron-donating MBn group at O-4, and still could not glycosylate the OH-3 position, is surprising. In another experiment (Scheme 4), we reacted the analogous acceptor 19 which had no protecting group at O-4 with a reactive fucosyl imidate donor 28 (1.3 equivalents), we found that both OH-3 and OH-4 were reactive even at low temperature (−78° C); the OH-3 exhibited even higher reactivity than OH-4 as the corresponding disaccharide 29 and 30 were formed and isolated in 52% and 27% yields respectively. We therefore concluded that the low reactivity of OH-3 in 13 was still due to the steric shielding from both the MBn group at O-4 and the 2-phthalimido group; probably both groups form a pocket which traps OH-3 and renders it unreactive towards the oxocarbenium intermediates formed from the glycosyl donors.
![]() | ||
Scheme 4 Probing the sterically shielding effect on OH-3 of 2-phthalimido protected acceptor. |
Based on the above probing results, the advantage of using 4,6-benzylidenated 14 to synthesize the required disaccharide 9 became evident, because the cyclic benzylidene group provides simultaneous protection to both the O-4 and O-6 positions while making OH-3 available for reactions; most importantly the smaller cyclic acetal relieves steric congestion around OH-3, making it more accessible than OH-3 in 13. Indeed, as shown in Scheme 5, when 14 was subjected to reaction with imidate 10 under the activation of TMSOTf, the desired disaccharide 7 was formed in a smooth manner and isolated in good yield (75%).
![]() | ||
Scheme 5 Preparation of neolactosamine and Lewis A donor. |
Using the triethylsilane26 as the reducing agent under the catalysis of boron trifluoride etherate, the 4,6-benzylidene group in 7 was regioselectively opened to afford the expected disaccharide alcohol 9 in almost quantitative yield (98%). The free OH-4 in 9 should be much less steric hindered due to the higher rotational freedom of benzyl group attached to the O-6 position thus is readily available for α-fucosylation. We applied the armed-disarmed glycosylation principle by reacting the less reactive thioglycoside 9 (disarmed) with the more reactive thiofucoside 15 (armed) under the activation of NIS/TfOH to afford the desired Lewis A trisaccharide donor 8 in excellent yield (84%)
The success in preparing the requires donors 8 and 9 allowed us to proceed further to synthesize the target tetrasaccharide 2 and pentasaccharide 4. As shown in Scheme 6, under the NIS/TfOH-promoted glycosylation conditions using triol 5 as the acceptor and thioglycoside 9 as the donor, the desired tetrasaccharide 31 was obtained in excellent yield (87%). As expected, despite the presence of three hydroxyl groups, the glycosylation proceeded in a remarkably regioselective manner to give only OH-3′ glycosylated product. The regioselectivity of glycosylation was established by a series of NMR experiments: 1D 1H and 2D 1H-1H COSY NMR revealed a broad doublet (3J 1.9 Hz) at 2.83 ppm, which appeared to correlate with H-2′, suggesting that it's the hydroxyl group (OH-2′) attached to the C-2′ position (the reducing end galactosyl unit). This was further confirmed by a selective 1D TOCSY experiment (see ESI†): we selectively excited the hydroxyl signal at 2.83 ppm, and observed that the magnetization was indeed transferred to the H-1′, H-2′, H-3′ and H-4′ protons (the reducing end galactosyl unit). These results confirmed that the glycosylation occurred at OH-3′. Next, the acetyl and N-phthalimido groups of 31 were concurrently removed in one step by the treatment with hydrazine hydrate in methanol under refluxing conditions. TLC revealed that the deprotection went very clean, only one spot was observed; the intermediate diamine was not isolated but subjected to a regioselective acetylation with acetic anhydride in methanol to provide the diamide which was subsequently debenzylated in 10% acetic acid–methanol under the catalysis of 20% Pd(OH)2 on charcoal to give the targeted tetrasaccharide 2 in 48% overall yield after three steps. Compound 2 was isolated by passing through a C18 Sep-Pak cartridge using a gradient of MeOH–H2O (0% → 10%) as the eluent.
![]() | ||
Scheme 6 Synthesis of tetrasaccharide 2. |
Similarly, when the Lewis A donor 8 was reacted with 5 under similar conditions as for 9, the pentasaccharide 32 was smoothly formed and isolated in very good yield (70%) and excellent regioselectivity (Scheme 7). Following the same deprotecting sequences as for 31, the desired pentasaccharide 4 was obtained in 55% overall yield after three steps.
![]() | ||
Scheme 7 Synthesis of pentasaccharide 4. |
The structures of both compounds 2 and 4 were confirmed by 1D and 2D NMR experiments, and also by high resolution mass spectrometry (see ESI†).
In conclusion, we have presented a concise route to synthesize a pentasaccharide containing the LeA-LacNAc sequence which has previously been shown to bind to TcdA from C. difficile. By applying the same strategy, a related tetrasaccharide was also prepared. Carefully choosing orthogonally protected donors and acceptors was found to be the key to the success of the syntheses and by effectively applying the armed-disarmed principle, we have significantly reduced the number of required steps thus improved the overall efficiency of the syntheses. These compounds are currently being used for binding and crystallographic studies, which will hopefully better define the molecular basis for carbohydrate binding and ultimately lead to the development of novel therapeutics.
Data for 13 (Found: C, 65.47; H, 5.30; N, 2.02%. C35H32NO7SCl requires C, 65.06; H, 4.99; N, 2.17%); [α]D +13.5 (c 0.7, CHCl3); δH (400 MHz, CDCl3) 7.92–7.78 (m, 2H, Phth), 7.78–7.68 (m, 2H, Phth), 7.45–7.29 (m, 7H, Ar), 7.21–7.12 (m, 4H, Ar), 6.84 (d, 2H, J 8.6 Hz, Ar), 5.54 (d, 1H, J 10.2 Hz, H-1), 4.70–4.56 (m, 4H, Bn), 4.42 (ddd, 1H, J 10.3, 8.7, 4.2 Hz, H-3), 4.20 (dd, 1H, J 10.3, 10.3 Hz, H-2), 3.88–3.79 (m, 2H, H-6a + H-6b), 3.78 (s, 3H, OMe), 3.67 (ddd, 1H, J 9.8, 3.9, 2.1 Hz, H-5), 3.57 (dd, 1H, J 9.5, 8.7 Hz, H-4), 2.25 (d, 1H, J 4.4 Hz, OH-3); δC (100 MHz, CDCl3) 159.50, 138.13, 134.22, 134.20, 131.63, 130.41, 130.11, 129.64, 128.97, 128.45, 127.77, 127.75, 123.81, 114.08 (Ar), 83.06 (C-1), 79.28 (C-5), 78.72 (C-4), 74.37 (Bn or MBn), 73.51 (MBn or Bn), 72.73 (C-3), 68.88 (C-6), 55.57 (C-2), 55.26 (OMe–MBn); m/z (ESI-MS) calcd for [C35H32NO7SCl + Na]+ 668.2, found 668.2.
Data for 20 (Found: C, 65.46; H, 5.25; N, 1.96%. C35H32NO7SCl requires C, 65.06; H, 4.99; N, 2.17%); [α]D +52.0 (c 0.6, CHCl3); δH (400 MHz, CDCl3) 7.86–7.80 (m, 1H, Phth), 7.77–7.64 (m, 3H, Phth), 7.42–7.28 (m, 7H, Ar), 7.15 (d, 1H, J 8.5 Hz, Ar), 6.95 (d, 2H, J 8.6 Hz, Ar), 6.45 (d, 2H, J 8.6 Hz, Ar), 5.51 (d, 1H, J 10.1 Hz, H-1), 4.65 (d, 1H, J 12.1 Hz, Bn or MBn), 4.63 (d, 1H, J 11.8 Hz, MBn or Bn) 4.53 (d, 1H, J 11.8 Hz, MBn or Bn), 4.47 (d, 1H, J 12.1 Hz, Bn or MBn), 4.23 (dd, 1H, J 9.9, 8.1 Hz, H-3), 4.18 (dd, 1H, J 10.1, 10.1 Hz, H-2), 3.87–3.76 (m, 3H, H-4 + H-6a + H-6b), 3.69 (ddd, 1H, J 9.5, 4.8, 4.8 Hz, H-5), 3.63 (s, 3H, OMe–MBn), 2.90 (d, 1H, J 2.7 Hz, OH-4); δC (100 MHz, CDCl3) 158.93, 137.68, 134.24, 134.11, 133.94, 133.91, 131.55, 130.38, 130.11, 129.64, 128.97, 128.54, 127.93, 127.77, 123.47, 123.21, 113.53 (Ar), 83.29 (C-1), 79.31 (C-3), 78.00 (C-5), 74.20 (Bn or MBn), 73.89 (C-4), 73.76 (MBn or Bn), 70.50 (C-6), 54.90 (OMe–MBn), 54.43 (C-2); m/z (ESI-MS) calcd for [C35H32NO7SCl + Na]+ 668.2, found 668.2.
Footnote |
† Electronic supplementary information (ESI) available: Additional experimental procedures for compounds 12, 14–15, 17–18, 25–26, 28–30 and 1H and 13C NMR spectra of all synthesized compounds. See DOI: 10.1039/b914193f |
This journal is © The Royal Society of Chemistry 2010 |