Synthesis of Azide-Modified Chondroitin Sulfate Precursors: Substrates for “Click”- Conjugation with Fluorescent Labels and Oligonucleotides

Azidopropyl-modified precursors of chondroitin sulfate (CS) tetrasaccharides have been synthesized, which, after facile conversion to final CS structures, may be conjugated with alkyne-modified target compounds by a one-pot “click”-ligation. RP HPLC was used for the monitoring of the key reaction steps (protecting group manipulation and sulfation) and purification of the CS precursors (as partially protected form, bearing the O-Lev, O-benzoyl, and N-trichloroacetyl groups and methyl esters). Subsequent treatments with aqueous NaOH, concentrated ammonia, and acetic anhydride (i.e., global deprotection and acetylation of the galactosamine units) converted the precursors to final CS structures. The azidopropyl group was exposed to a strain-promoted azide–alkyne cycloaddition (SPAAC) with a dibenzylcyclooctyne-modified carboxyrhodamine dye to give labeled CSs. Conjugation with a 5′-cyclooctyne-modified oligonucleotide was additionally carried out to show the applicability of the precursors for the synthesis of biomolecular hybrids.


■ INTRODUCTION
Chondroitin sulfate (CS) is a linear sulfated polysaccaharide that plays pivotal roles in many biological processes such as cell division, neuronal development, viral invasion, cancer metastasis, and spinal cord injury. 1−5 CS is composed of repeating β-D-glucuronic acid (GlcA) and N-acetyl-β-D-galactosamine (GalNAc) units arranged in the sequence by GlcA-β(1 → 3)-GalNAc-β(1 → 4) glycosidic bonds with variable high and low sulfation patterns. The highly sulfated regions with a specific arrangement of the sulfate groups define the binding motifs for certain proteins (for example neurotrophins, selectins, chemokines, and midkine). 6,7 Because of the biological activities of CS, it has a great potential in the drug development related to malignant tumors, multiple sclerosis, and neuronal degeneration processes, 8 for example. Several research groups have reported synthetic 9−17 or semisynthetic 18,19 procedures for the preparation of different CS derivatives. Even a solid-phase approach has been utilized. 20 Biological activity of synthetic CS di-, tetrasaccharide derivatives, 21−27 including CS-based glycomimetic polymers, 28,29 and glycopeptides, 30,31 has been extensively studied. In addition, CS-modified cell-surface glycal engineering 30 and CS-based microarrays 32,33 have been reported. For the drug delivery purposes, nanoparticles and liposomes 34 carrying genes 35 and anticancer drugs 36 have been coated with polymeric CS to provide targeted delivery via CD44 (a cell surface receptor)-mediated endocytosis.
Alternative procedures for the preparation of structurally well-defined, appropriately sulfated, homogeneous CS polysaccharide derivatives are still demanding. Only a micromolescale synthesis of conjugated or labeled CSs is usually required to give sufficient amounts of compounds for the biological activity studies. Multistep miniature manipulation of an oligosaccharide may be complex, however, especially if the product is strongly hydrophilic and lacks a chromophore. Therefore, It may be beneficial to design appropriately protected and UV-detectable precursors, which may be readily purified by RP HPLC and converted then via quantititative global deprotection step to the desired products that may be finally "fished out" by an appropriate label or a chromophoric conjugate group. 37,38 In the present study, azidopropylmodified CS tetrasaccharide precursors (18, 21, 23, and 27) have been synthesized on a 6 μmol scale. The precursors could be homogenized by RP HPLC and quantitatively converted to final CS structures prior to a one-pot "click"-ligation with alkyne-modified target compounds. Tetrasaccharide 16 acted as a general starting material for the synthesis of 18, 21, and 23. Different sulfation patterns at C4 and C6 sites of the GalNAc residues (18 vs 21) were obtained by utilizing an orthogonality of the benzylidene and chloroacetyl/acetyl protections. The key reaction steps, that is, the protecting group manipulation and sulfation, could be monitored by RP HPLC. The homogenized CS precursors (18,21,23, and 27, bearing the O-Lev, Obenzoyl, and N-trichloroacetyl groups and methyl esters) were converted to final CS structures by subsequent treatments with aqueous NaOH, concentrated ammonia, and acetic anhydride (i.e., the global deprotection and acetylation of the galactosamine units) and exposed then to a strain-promoted azide− alkyne cycloaddition (SPAAC) with a dibenzylcyclooctynmodified carboxyrhodamine dye to give labeled CSs 24−26 and 29. Conjugation with a 5′-cyclooctyne-modified oligonucleotide was additionally carried out to show the applicability of the precursors (18,21,23) for the synthesis of biomolecular hybrids.
Synthesis of Fully Protected CS-Tetrasaccharide Precursors 13 and 15. Unexpected stereochemical outcome of glycosylations with 4,6-O-benzylidene protected 2-deoxy-2trichloroacetamido-D-galactosyl trichloroacetimidate acceptors has been reported. 40 Glycosylation of simple alcohols with monomeric N-acyl galactosamines may give the expected βanomer as a major product (e.g., synthesis of 6 from 5, Scheme 2), but the extent of the α-anomer may become significant in glycosylation of D-glucuronic acid-derived acceptors and with disaccharidic N-acyl galactosamine donors. This has been attributed to unfavorable interactions in the transition state between the 4,6-benzylidene group and the approaching Dglucuronic derived acceptors. 41 As outlined in Scheme 3, glycosylation of 10 with disaccharide donor 12 gave tetrasaccharide 13 with an α-glycosidic bond in 32% isolated yield and only a substantial amount of β-glycosylation was observed. The preference for the β-anomer may be increased by replacing the benzylidene group to two acyl protections. 41 Glycosylation of 10 with previously reported disaccharide donor 14 14  expected β-glycosidic bond in a 33% isolated yield (the major byproduct was hydrolyzed 14).
Protecting Group Manipulation and Sulfation To Obtain CS-Tetrasaccharide Precursors 18, 21, and 23 with Different Sulfation Pattern. The chloroacetyl groups of 15 were selectively removed with a mixture of thiourea, pyridine, and ethanol, and the obtained tetrasaccharide 16 was used as a shared starting material for 18, 21, and 23 (Scheme 4, note the definitions of GalNAc-1 and GalNAc-2 to clarify the readability of the text). The required protecting group manipulation and sulfation were carried out on a 6 μmol scale and the key reaction steps were monitored by RP HPLC (see RP HPLC profiles of the crude product mixtures: a−h/ Scheme 4). Orthogonality of the benzylidene and chloroacetyl/ acetyl protections was utilized to obtain different sulfation patterns at C4 and C6 sites of the N-acetyl galactosamine residues (GalNAc-1 and GalNAc-2, 18 vs 21). To obtain 18, the exposed 4-and 6-OH groups (GalNAc-2) were first quantitatively sulfated using SO 3 ·TMA (17, cf. ii/Scheme 4 and RP HPLC profile b/Scheme 4). The acid-catalyzed benzylidene removal (GalNAc-1) was then carefully optimized, as premature elimination of the sulfonate groups may occur in acidic conditions. An acceptable result was obtained with 0.1% aqueous TFA (5 h at room temperature, iii/Scheme 4) that removed the benzylidene group (GalNAc-1) and kept the sulfonate groups (GalNAc-2) mainly intact (18, traces of elimination side products observed, c/Scheme 4). To obtain 21, the exposed 4-and 6-OH groups (GalNAc-2) were first acetylated (19, cf. iv and d/Scheme 4), and then the benzylidene group (GalNAc1) was removed by 80% aqueous acetic acid to give 20 (1 h at 100°C, cf. v and e/Scheme 4). It may be worth mentioning that the chloroacetyl groups are too prone to acid-catalyzed hydrolysis, and hence, the acyl replacement (at GalNAc2, 19 vs 16) was required prior to the benzylidene removal (GalNAc1). The exposed 4-and 6-OH groups (GalNAc1) were finally sulfated as above (ii and f/ Scheme 4) to give 21. To obtain 23, the benzylidene group of 16 (GalNAc-1) was removed (22,
Global Deprotection, N-Acetylation, and Labeling of the CSs. Quantitative global deprotection of 18, 21, and 23 (protected by O-Bz, O-Lev, and N-trichloroacetyl groups and methyl esters) was required prior to the labeling or conjugation steps (Schemes 5 and 6). Aqueous sodium hydroxide (3 h at 55°C ) was used for the hydrolysis of methyl esters of the GlcA residues and of the ester protections (O-Bz and O-Lev) (i/ Scheme 5). The remained N-trichloroacetyl groups were then removed by concentrated ammonia (4 days at 55°C, ii/Scheme 5). An aliquot of the crude product mixtures was labeled as below, and the completion of the global deprotection was verified by LC-MS analysis (data not shown). The exposed amino groups were then acetylated with a mixture of acetic anhydride and triethylamine in aqueous acetonitrile (iii/ Scheme 5), the plausible O-acetyl groups were removed by concentrated ammonia (iv/Scheme 5), and then all CS structures were exposed to SPAAC with dibenzylcyclooctyne-PEG4−5/6-carboxyrhodamine to give labeled CSs 24−26 (v/ Scheme 5). The global deprotection, N-acetylation and labeling were carried out using 0.3 μmol of the CS precursors (18, 21, or 23) and 0.6 μmol (2 equiv) of the label. As seen in the example of RP profiles of the crude product mixtures (a/ Scheme 5), the transformation from 18, 21, and 23 to 24−26 could be successfully carried out. Overall isolated yield in each case was ca. 60%. The labeled CS tetrasaccharide 29 and CS disaccharide 30 were synthesized from 9 and 13 using exactly the same procedures: The benzylidene protections were removed and the exposed hydroxyl groups were sulfated as described to 23 from 16 above. The obtained CS precursors (27,28) were purified by RP HPLC and then the transformation, including global deprotection, N-acetylation and labeling with the carboxyrhodamine dye, was carried out. Overall isolated yields of 29 and 30 (calculated from 9 and 13) were 11 and 12%, respectively. The authenticity of 24−26, 29, and 30 was verified MS (ESI-TOF) spectroscopy (cf. Table 1 and the Supporting Information).
SPAAC Conjugation with a Cyclooctyne-Modified Oligonucleotide. DNA-directed immobilization (DDI) of carbohydrates may be used to provide high throughput tools to study protein carbohydrate interactions on a DNA-based micro array. 42 The carbohydrates are usually conjugated to the 5′ end of appropriate DNA sequences and immobilized then via hybridization with the complementary strands at a specific location of the array. 42 For the preparation of the carbohydrateoligonucleotide conjugates, a straightforward and high yielding procedure is needed. As the global deprotection/N-acylation step above proved virtually quantitative (cf. RP HPLC analysis of crude product mixture of 24, a/Scheme 5), the applicability of the azidopropyl modified CS precursors (18,21, and 23, Scheme 4) was evaluated to gain 5′-CS-oligonucleotide conjugates (32−34, Scheme 6). 18, 21, and 23 (20 nmol) were exposed to the global deprotection and N-acetylated as above and treated then with a 5′-cyclooctyne modified 2′-deoxy oligoribonucleotide (31, the same sequence previously used for DDI on a carbohydrate microarray 42 ). The reactions were

Bioconjugate Chemistry
Article performed using an excess of the oligonucleotide (1.5 equiv, 1 mmol L −1 solution in water, overnight at 55°C), and the completion of the reactions was verified by an RP-HPLC analysis (Scheme 6). In each case, the crude product mixture contained the remained excess of the oligonucleotide (31), but the desired CS-glycoconjugate (32, 33, and 34, conversion yield according to peak areas: 70, 75, and 90%, respectively) could be readily homogenized. The authenticity of the products was verified by MS(ESI-TOF) spectroscopy (cf. Table 1 and the Supporting Information).
HP Neuron Cell Imaging with the Labeled CS Tetrasaccharide 29 and Disaccharide 30. In the nervous system, CS bound to core protein forms CS proteoglycan (CSPG), a major component of the brain extracellular matrix. It functions as a regulator of plasticity and axon guidance during brain development and inhibits regeneration in the adult central nervous system. 43 Interestingly, CSPG has been found to bind to receptors raising the possibility that CS conjugates may be taken up into neuronal cells. To test this and to demonstrate that even CS structures with a nonregular α-glycosidic bond may show activity, preliminary cellular uptake studies with 29 were carried out. Hippocampal neurons were incubated with labeled disaccharide 30 (modest affinity control) and tetrasaccharide 29 at 50 nM for 10 days, after which confocal imaging through nuclei was carried out to determine whether uptake had occurred. At this concentration, 29 was visible as bright puncta in the cytosol that resembled endosomes surrounding the nucleus and in the neurites (Figure 1). This distribution pattern is consistent with receptor-mediated uptake. A distinctly different pattern was obtained when the carboxyrhodamine dye alone was used. This was likely internalized via phagocytosis. When used at a higher concentration (200 to 500 nM), 29 uptake was observed in more than 80% of neurons and the signal was detectable even in the distal neurites. Uptake of 29 was not observed in neurons that were older than 15 days in vitro. The labeled disaccharide 30 was not taken up at all by hippocampal neurons at any of the tested concentrations (50−500 nM). These results suggest that 29 may be specifically taken up via receptor-mediated endocytosis in neurons, whereas 30 is not.

■ CONCLUSIONS
In this primarily synthetic description, azidopropyl-modified precursors of CS tetrasaccharides have been prepared, which, after facile conversion to final CS structures (i.e., global deprotection and N-acetylation: (1) 0.1 mol L −1 aq NaOH, 3 h  aq MeCN) may be conjugated with alkyne-modified target compounds by a one-pot "click"-ligation. RP HPLC was used for the monitoring of the key reaction steps (protecting group manipulation and sulfation) and purification of the CS precursors. The global deprotection/N-acetylation of the precursors proved to be virtually quantitative converting the precursors to final CS structures, which were then exposed to SPAAC conjugation with dibenzylcyclooctyn-modified carboxyrhodamine dye and a cyclooctyne modified oligonucleotide to give labeled CSs (24−26 and 29) and CS-oligonucleotide conjugates (32−34), respectively. While most of the synthetic procedures of CSs requires specialized expertise in carbohydrate chemistry, the present miniature procedure of the precursors has been designed to be more readily accessible by multidisciplinary bioorganic research groups offering new tools for biologists. Particularly our approach provides robust analytical method for protecting group manipulations of different sulfation patterns using HPLC and MS (ESI-TOF). Moreover, an azide group (not an alkyne) was introduced to the precursors, which may be more straightforward choice for the synthesis biomolecular hybrids via SPAAC (as demonstrated by 32−34). To demonstrate the potential neuronal activity of the labeled CSs, preliminary cellular uptake studies with hippocampal neurons and 29 vs 30 were demonstrated. More detailed affinity studies and applications of the CSprecursors are currently underway in our laboratory. The preparation of 5′-CS-oligonucleotides proved efficient, which may, via DDI, be expanded to protein−CS interaction studies on a DNA-based micro array. Moreover, conjugated CSs may find applications as a novel targeted delivery strategy for therapeutic oligonucleotides, which should be studied in more detail.

■ EXPERIMENTAL PROCEDURES
General Remarks. CH 2 Cl 2 , DMF, toluene, pyridine, and methanol were dried over molecular sieves. Solid reagents were dried over P 2 O 5 in a vacuum desiccator. The NMR spectra were recorded at 400 or 500 MHz. Chemical shifts are given in ppm using internal TMS or solvent residual signals as reference. Appropriate 1D and 2D NMR methods (e.g., TOCSY, COSY, DEPT, and HSQC) were used for peak assignment. The mass spectra were recorded using a MS (ESI-TOF) spectrometer. RP HPLC analysis and purification of the oligosaccharides were performed using a Thermo ODS Hypersil C18 (150 × 4.6 mm, 5 μm) analytical column. The correct yields of 18, 21, and 23 (i.e., amounts of the RP HPLC isolated products synthesized using 5.9 μmol 16 as a starting material) were determined by comparing the intensity of the 1 H NMR signals to a known amount of acetonitrile.

Bioconjugate Chemistry
Article exactly the same procedure as described for the transformation of 16 to 23 above. The acyl protected precursors (27 and 28) were purified by RP HPLC and exposed then to the treatments with aqueous NaOH, concentrated ammonia, acetic anhydride and carboxyrhodamine dye as described for the transformation of 18, 21, and 23 to 24−26 above. Isolated yields of 29 and 30 were 11 and 12%, respectively (overall yields calculated from 9 and 12). The authenticity of the products was verified by MS (ESI-TOF) spectroscopy ( Table 1).
HC Neuron Cell Imaging. Hippocampal neurons were prepared from rat as previously. 44

■ ACKNOWLEDGMENTS
The financial support from the Academy of Finland (No: 308931) and the Finnish Cultural Foundation are acknowledged.

Bioconjugate Chemistry
Article