Divalent Triazole-Linked Carbohydrate Mimetics: Synthesis by Click Chemistry and Evaluation as Selectin Ligands

Starting from an enantiopure 3-amino-substituted pyran derivative, the synthesis of a series of divalent 1,2,3-triazole-linked carbohydrate mimetics is described. The preparation of the required 3-azido-substituted pyran proceeds smoothly by copper-catalyzed diazo transfer. Using different conditions for the Huisgen-Meldal-Sharpless cycloaddition, this azide reacts with several diynes to furnish the desired divalent carbohydrate mimetics bearing rigid or flexible linker units. The in situ generation of the 3-azidopyran in the presence of Cu/C as catalyst followed by the reaction with the alkyne allows a direct one-pot transformation from the 3-aminopyran to the


Introduction
Synthetic C-glycosides and other carbohydrate mimetics are compounds of great interest due to their similarity to carbohydrate structures. [1,2] Their modified properties may influence bioavailability, selectivity, and affinity in important biological processes involving carbohydrates. Our group discovered an alkoxyallene-based [3] straightforward access to enantiopure 3aminopyrans of general structure A and the closely related ringexpanded 4-aminooxepanes (Scheme 1). [4] Among compounds A, the 6,6-dimethyl-substituted 3-aminopyran 1 is the most easily available derivative [4a,4b] and hence its properties were investigated in detail. [4e,4f ] Most remarkably, the connection of 1 to gold nanoparticles via amide bonds and subsequent O-sulfation of the pyran hydroxyl groups provided a multivalent macromolecular species that turned out to be a highly potent inhibitor of L-and P-selectin with IC 50 values in the picomolar range. [5] L-and P-selectins are crucial components in the inflammatory process [6] and therefore compounds inhibiting their activity are of interest as potential therapeutics. [7] The mentioned desired click products. We also examined the Sakai-Westermann method that transfers primary amines with the aid of α,α-dichlorotosylhydrazones into 1,2,3-triazoles. These copper-free click conditions were applied for the first time to the preparation of a divalent compound. The O-sulfation of the carbohydrate mimetics was achieved using the SO 3 -DMF complex under careful 1 H-NMR control. Five polysulfated compounds could be obtained in pure form and these were tested by surface plasmon resonance spectroscopy as inhibitors of L-selectin giving IC 50 values between 45 nM and 50 μM.
aminopyran-decorated gold nanoparticles are densely packed with 1000-1200 ligands. For an understanding of the activity of these multivalent species, we studied analogous compounds that contain only a few of the aminopyrans ligands. Therefore we synthesized divalent and trivalent compounds of general structure B employing metal-catalyzed reactions (Scheme 1). [8] We also connected two or three aminopyrans units 1 via amide Scheme 1. Routes to di-and trivalent carbohydrate mimetics B, C, and D.
The cycloaddition of 2 and 3 proceeded smoothly with methods A and C providing 1,4-disubstituted 1,2,3-triazole derivative 5 in excellent yields, but the purification of the product obtained by method A is not as straightforward due to the presence of the fairly large amounts of the TBTA ligand. Employing the acid/base-catalyzed method B the desired product 5 was obtained with a lower yield of 67 %. Due to the relatively high polarity of 1,2,3-triazole 5 an extraction was not possible and separation of the product from the acid and base was less efficient. For the synthesis of 1,2,3-triazole derivative 6 from alkyne 4 only method C was examined that furnished the desired product quantitatively.
Since we planned to test the prepared carbohydrate mimetics in biological assays, it should be assured that no copper is present in the obtained compounds. The three samples of 1,2,3triazole derivative 5 obtained by the alternative CuAAC methods A, B, and C depicted in Scheme 3 were submitted to quantitative copper ICP-MS analyses. Values in the range between 4-8 ppb for 63 Cu and 65 Cu were obtained; these are lower than 15 ppb which is regarded as the sensitivity limit of this method.
For the synthesis of divalent structures diynes 7 and 8 with rigid arene centers and diynes 9 and 10 with flexible aliphatic linker units were chosen (Scheme 4). Alkynes 7, 8, and 9 and 3azidopyran 2 were subjected to method A using CuI and TBTA. Although all desired divalent bis-1,2,3-triazoles 11-14 were obtained in good yields, the reactions of alkynes 8 and 9 also provided small amounts of the corresponding mono-1,2,3-triazoles.
The acid/base-catalyzed CuAAC (method B) was examined with diyne 9. Using toluene as solvent and a mixture of Hünig base/acetic acid led to a fast reaction and a slight improvement of the yield. Apparently, the use of TBTA as additive is not necessary, whose removal is sometimes tedious. [13] Due to the excellent results observed with Cu/C-catalyzed CuAAC (method C) for the synthesis of the model 1,2,3-triazoles and the easy purification of the obtained compounds, these conditions were also applied to the synthesis of divalent compound 14. The commercially available diyne 10 and 3-azidopyran 2 furnished the corresponding product 14 in a very high yield.
In our study on 4-aminooxepanes we also prepared a bis-(azido)oxepane derivative that underwent a twofold CuAAC with an excess of 1,6-heptadiyne affording an unsymmetrical diyne 15 (Scheme 5). [13] In order to enhance the degree of complexity, compound 15 was treated with 3-azido-substituted pyran 2 employing the approved method C. The reaction proceeded smoothly, but product 16 with three carbohydrate mimicking subunits and four connecting 1,2,3-triazole moieties was extremely polar and hence hard to purify, resulting in a disappointingly low yield of 24 %.
The Cu/C matrix contains copper(I) as well as copper(II) species and therefore the Lipshutz group explored the direct onepot transformation of amines into the desired 1,2,3-triazoles by in situ generation of the corresponding organic azides employing TfN 3 . [22] We tested these reported one-pot conditions for the synthesis of the desired carbohydrate mimetics. It should be possible to start from 3-aminopyran 1 and after in situ generation of the 3-azidopyran 2, an alkyne can be added and the desired cycloaddition product should be obtained. Lipshutz et al. reported that the method proceeds in dichloromethane as solvent, [22] but we found for our compounds that under these conditions only unchanged 3-aminopyran 1 and 3-azidopyran 2 and no 1,2,3-triazoles were isolated. An optimization revealed that acetonitrile was a suitable solvent and we could successfully apply the one-pot procedure to the syntheses of 1,2,3triazole 5 and bis-1,2,3-triazole 14 (Scheme 6). For the preparation of the simple 1,2,3-triazole 5 we used the convenient and safe NfN 3 as reagent (method D) and the product was isolated in satisfying 75 % yield. The divalent product 14 was prepared with TfN 3 as a reagent for the diazo transfer (method E) and the yield was only 38 %; no control experiment with NfN 3 was performed to identify the reason for the lower efficiency in this example. In any case, the two experiments depicted in Scheme 6 indicate that the one-pot procedure for the diazo transfer and CuAAC may be a good alternative to the step-wise route. [23] Scheme 6. One-pot syntheses of 1,2,3-triazole 5 and of divalent bis-1,2,3triazole 14 starting from amine 1 using a Cu/C-catalyzed one-pot procedure. The methods presented above proved to be very suitable for the (3+2) cycloaddition leading to the desired 1,2,3-triazoles, however, they all require copper as metal catalyst. Orthogonal ligation protocols have been widely used as an alternative for the common CuAAC. [24] A so far less explored possibility for the regioselective preparation of 1,4-disubstituted triazoles employs the Sakai reaction of primary amines with α,α-dichlorotosylhydrazones. [14] In 2012, Westermann et al. reinvestigated and optimized this method for the synthesis of 1,2,3-triazoles and the modification of biologically significant target compounds. [15] The impressive results prompted us to briefly study the potential of this copper-free alternative for the construction of carbohydrate mimetics starting directly from 3-aminopyran 1. For the first trial the simple α,α-dichlorotosylhydrazones 17 and 19 were chosen (Scheme 7). [15] Following Westermann's protocol for the Sakai reaction the simple 1,2,3-triazoles 18 and 20 were obtained in excellent or good yield. To further evaluate the efficacy of this procedure, the synthesis of a divalent carbohydrate mimetic was also attempted. For this purpose, the bis(α,α-dichlorotosylhydrazone) 21 was prepared in two steps starting from dimethyl adipate and dichloromethane. [25] The metal-free connection of this compound with 3-aminopyran 1 furnished the bis-1,2,3-triazole 22 in good yield. To the best of our knowledge, this transformation is the first application of the Sakai-Westermann method for the synthesis of a divalent compound. With easily obtained and bench stable starting materials this method proved to be a good alternative for the metal-free direct conversion of primary amines into compounds with 1,2,3-triazole units without the need to generate potentially hazardous azide intermediates. For the evaluation as selectin ligands the prepared oligohydroxylated compounds had to be converted into the corresponding O-sulfated samples which is not a trivial challenge. [26] We recently reported optimizations of the O-sulfation step of multivalent carbohydrate mimetics connected with amide bonds and the commercially available sulfur trioxide-DMF complex was found to be the reagent of choice. [9] It was employed in (deuterated) DMF, thus allowing to follow the reaction progress with high-resolution 1 H-NMR spectroscopy. Here we present the successful application of this method for the O-sulfation of 3-azidopyran 2 and multivalent carbohydrate mimetics bearing 1,2,3-triazole units (Scheme 8). After O-sulfation all sulfuric acid monoesters were converted into the corresponding sodium salts using either a 1 M solution of sodium hydroxide (method G/a) or 0.5 M solution of this base (method G/b). The resulting products were finally purified by dialysis. 3-Azidopyran 2 and 1,2,3-triazole 18 were successfully converted into the desired products 23 and 24 (Scheme 8). Only low yields were observed for the two simple compounds due to their low molecular mass that leads to considerable losses of mass during the dialysis purification. Higher yields were observed for the O-sulfated divalent compounds 25-27. In agreement with our previous results, method G/b is favorable for the compounds with 1,2,3-triazole substructures. The two divalent compounds 11 and 12 (Scheme 4) with rigid arene linker units were also treated under the approved conditions with the sulfur trioxide-DMF complex and they were completely consumed, but after dialysis high-resolution 1 H-NMR spectroscopy revealed that complex mixtures of compounds were formed. To our regret, these interesting compounds could therefore not be investigated further by SPR experiments.
The five O-sulfated samples obtained were investigated by surface plasmon resonance (SPR) spectroscopy and as earlier found for this type of compounds a competitive binding assay was applied. [27] The resulting IC 50 values for binding to L-Selectin are compiled in Figure 1 and should be regarded as preliminary results that need further validation. For the monovalent O-sulfated 3-azidopyran 23 the lowest IC 50 value of 45 nM was found, however, this result has to be regarded with caution since the dose-response curve ascended slightly at higher concentrations. For the four 1,2,3-triazole derivatives 24-27 IC 50 values in the range of 0.6-50 μM were determined. The most potent compound 25 is characterized by the smallest distance between the two O-sulfated pyran moieties. The low number of comparable compounds does not allow a sound interpretation of their observed inhibitor properties. At the moment it is pure speculation that fairly close O-sulfated pyran moieties as in the divalent compound 25 (and in the multivalent polysulfated pyran-decorated gold nanoparticles mentioned in the introduction) are a prerequisite for high activity of these compounds.

22.
After O-sulfation of the compounds with the sulfur trioxide-DMF complex five products were isolated in sufficient purity. They were investigated by SPR as inhibitors of L-selectin and IC 50 values between 45 nM and 50 μM were found. These preliminary results need further confirmation and inclusion of more compounds in order to prove or disprove a multivalency effect. [29] In summary, the presented synthetic results show that the azide-alkyne (3+2) cycloadditions and the related methods furnish carbohydrate mimetics with interesting structural features.

Experimental Section
Reactions were generally performed under an inert atmosphere (argon) in flame-dried flasks. Solvents and reagents were added by syringe. Solvents were dried using standard procedures and were purified with a MB SPS-800-dry solvent system. Triethylamine was distilled from CaH 2 and stored with KOH under an argon atmosphere. Commercially available reagents were used as received without further purification unless otherwise stated. Products were purified by flash chromatography on silica gel (230-400 mesh, Merck or MACHEREY-NAGEL) or by ion exchange resin (DOWEX® 50WX8-200 Sigma-Aldrich). DOWEX® Na + was freshly prepared by washing DOWEX® with a saturated solution of NaCl. Unless stated otherwise, yields refer to analytical pure samples. TLC analyses were performed on silica gel coated aluminum plates (Merck). Products were detected by UV and by using staining reagents (cerium/molybdenum reagent, KMnO 4 , and ninhydrin).
NMR spectra were recorded with BRUKER (AV 500, AV 700) and JEOL (ECP 500) instruments. Chemical shifts (δ) are listed in parts per million (ppm) and are reported relative to solvent residual signals: CuI (0.23 equiv.) were added, followed by the addition of the corresponding alkyne (1.2 equiv.) and NEt 3 (0.23 equiv.). The resulting mixture was stirred at r.t. for the indicated time. The solution was quenched with 7 N NH 3 in MeOH and the resulting mixture was filtered through a short silica gel column and washed with a mixture of CH 2 Cl 2 /7 N NH 3 in MeOH (10:1). After removing the solvents in vacuo, the crude product was purified by flash column chromatography.
General Procedure (GP2), CuAAC using DIPEA/AcOH (Method B): To a suspension of 3-azidopyran 2 (1.0 equiv.) in toluene (1 mL/ mmol) were added CuI (0.4 equiv.), the corresponding alkyne (1.2 equiv.) and finally diisopropylethylamine (DIPEA) (4.0 equiv.) and HOAc (4.0 equiv.). The reaction mixture was stirred at r.t. during the indicated time. The solution was quenched with 7 N NH 3 in MeOH and the resulting mixture was filtered through a short silica gel column and washed with a mixture of CH 2 Cl 2 /7 N NH 3 in MeOH (10:1). After removing the solvents in vacuo, the crude product was purified by flash column chromatography.  9:1), Cu/C (0.1 equiv.), NEt 3 (5.0 equiv.) and the corresponding alkyne were added and the reaction mixture was stirred at 60°C. After the total consumption of TfN 3 or NfN 3 (controlled by TLC, CH 2 Cl 2 / MeOH, 9:1) the mixture was filtered through a pad of Celite® and washed with MeOH. The product was purified by flash column chromatography.

General
General Procedure (GP5), Metal free triazole synthesis using tosylhydrazones (Method F): To a cooled solution of 3-aminopyran 1 (1.0 equiv.) in EtOH (12 mL/mmol), KOAc (3.0 equiv.) was added and the resulting solution was stirred for 10 min at 0°C. The corresponding α,α-dichlorotosylhydrazone (1.3 equiv.) was dissolved in MeCN (6 mL/mmol) and added dropwise to the reaction mixture. Stirring was continued at r.t. until the reaction was complete (monitored by TLC). All volatiles were removed in vacuo and the crude product was purified by flash column chromatography.

General Procedure (GP6), O-sulfation (Method G):
The corresponding polyol (1.0 equiv.) was dissolved in [D 7 ]DMF (0.6-1.0 mL). The solution was cooled to 0°C and SO 3 -DMF (3.0 equiv. per OH) was added. The reaction mixture was stirred at r.t. during the indicated time. The reaction progress was followed by 1 H-NMR spectroscopy (700 MHz). When indicated, additional SO 3 -DMF (1.0-3.0 equiv. for each OH group) was added and the reaction mixture was stirred at r.t. for the additional time given until full conversion was observed. The obtained sulfated intermediates were directly converted into the corresponding sodium salts either by method G/a or method G/b.
Method G/a: The reaction mixture was cooled to 0°C and an aq. 1 M solution of NaOH was added dropwise until pH 10-12 was reached. The solvents were removed in vacuo and the crude product was purified by dialysis in H 2 O.
Method G/b: The reaction mixture was cooled to 0°C and an aq. 0.5 M solution of NaOH was added dropwise until pH 7-9 was reached. The reaction mixture was filtered through an ion exchange DOWEX® Na + column. The solvents were removed in vacuo and the crude product was purified by dialysis in H 2 O.
The final products were filtered through a syringe filter (diameter 25 mm; pore size 0.2 μm; PTFE membrane) when indicated. followed by slow addition of freshly prepared NfN 3 [17] (475 mg, 1.46 mmol). The mixture was stirred for 24 h, then glycine hydrochloride (554 mg, 5.00 mmol) was added and the suspension was stirred for another 24 h. The solution was quenched with 7 N NH 3 in MeOH and the copper salt was filtered off through a short silica gel column. The crude product was purified by flash column chromatography (silica gel, CH 2 Cl 2 100 %, CH 2 C 51.82, H 7.69, N 13.94; found C 51.96, H 7.72, N 13.36.