Strain-Promoted Alkyne-Azide Cycloadditions (SPAAC) Reveal New Features of Glycoconjugate Biosynthesis

Abstract

We have shown that 4-Dibenzocyclooctynol (DIBO), which can easily be obtained by a streamlined synthetic approach, reacts exceptionally fast in the absence of a Cu^I catalyst with azido-containing compounds to give stable triazoles. Chemical modifications of DIBO, such as oxidation of the alcohol to a ketone, increased the rate of strain promoted azide-alkyne cycloadditions (SPAAC). Installment of a ketone or oxime in the cyclooctyne ring resulted in fluorescent active compounds whereas this property was absent in the corresponding cycloaddition adducts, thereby providing the first example of a metal-free alkyne-azide fluoro-switch click reaction. The alcohol or ketone functions of the cyclooctynes offer a chemical handle to install a variety of different tags, thereby facilitating biological studies. It was found that DIBO modified with biotin combined with metabolic labeling with an azido-containing monosaccharide can determine relative quantities of sialic acid of living cells that have defects in glycosylation (Lec CHO cells). A combined use of metabolic labeling/SPAAC and lectin staining of cells that have defects in the Conserved Oligomeric Golgi (COG) complex revealed that such defects have a greater impact on O-glycan sialylation than galactosylation, whereas sialylation and galactosylation of N-glycans was similarly impacted. These results highlight that the fidelity of Golgi trafficking is a critical parameter for the types of oligosaccharides that are being biosynthesized by a cell. Furthermore, by modulating the quantity of biosynthesized sugar nucleotide, cells may have a means to selectively alter specific glycan structures of glycoproteins.

Introduction

Metal free cycloadditions between cyclooctynes and azides to give stable 1,2,3-triazoles have found wide utility in labeling glycans proteins and lipids of living cells, glycoprotein enrichment for proteomics, protein and oligonucleotide modification and tissue reengineering.^[1] These reactions, which have been coined “Strain-Promoted Alkyne-Azide Cycloadditions (SPAAC)” have also made entry in material sciences and have for example been employed for the assembly, crosslinking and surface modification of dendrimers,^[2] derivatization of polymeric nanostructures,^[3] and patterning of surfaces.^[4] The attraction of SPAAC is that it does not require a toxic metal, is highly efficient even in very complex milieu and proceeds efficiently at ambient temperature. Density functional theory (B3LYP) calculations of the transition states of cycloadditions of phenyl azide with acetylene and cyclooctyne indicate that the fast rate of the strain promoted cycloaddition is due to a lower energy required for distorting the 1,3-dipole and alkyne into the transition-state geometry.^[5]

The first generation of cyclooctynes (1) suffered from relatively slow reaction rates and as a consequence the scope of these reagents is rather limited (Figure 1). It has, however, been found that electron-withdrawing fluorine groups at the propargylic position of a cyclooctyne (DIFO, 2) dramatically increase the rate of strain-promoted cycloaddition with azides.^[6] The attractiveness of this methodology has, for example, been demonstrated by visualization of glycans in vivo at subcellular resolution during the development of zebrafish embryos.^[7]

Figure 1.

Reagents for labeling azido-containing biomolecules.

We have found that derivatives of 4-dibenzocyclooctynol (DIBO, 3) react fast with azido-containing saccharides and amino acids and can be employed for visualizing metabolically labeled glycans of living cells.^[8] While the fluorine atoms of DIFO (2) effect the rate enhancement by increasing interaction energies, the aromatic rings of 3 accomplish a similar increase in reaction rate through conformational effects that results in decreasing the distortion energy. Attractive features of DIBO (3) include easy access to the compounds by a simple synthetic approach, nontoxicity and straightforward attachment of a variety of probes. Furthermore, dibenzocyclooctynes can be generated photochemically by short irradiation by UV light of corresponding cyclopropenones thereby providing opportunities for the spatial and temporal controlled labeling of the target substrates.^[9] We have also shown that by employing nitrones and nitrile oxides as 1,3-dipoles, the rate of cycloaddition can be further enhanced and this technology has for example made it possible to selectively tag proteins at the N-terminus or perform sequential modifications of complex compounds.^[10] Furthermore, several analogs of DIBO have been reported that exhibit even higher rates of cycloaddition with azides.^[11]

We report here a streamlined approach for the preparation and modification of DIBO and it is shown that modification of the eight-membered ring, by for example installment of a ketone, affects the rate of cycloaddition. The presence of a ketone or oxime resulted in compounds that are fluorescent active. Interestingly, the corresponding cycloaddition products are non-fluorescent and hence compounds such as 11 and 12 provide a novel metal free alkyne-azide fluoro-switch click reaction, which for example can be exploited in monitoring reactions in real time. Metabolic labeling combined with SPAAC of wild type cells and cells that have known defects in their glycosylation machinery showed that relative quantities of sialylation of glycoconjugates can easily and reliably be established. Furthermore, a combined use of metabolic labeling/SPAAC and lectin staining revealed that a defect in the Conserved Oligomeric Golgi (COG) complex affects terminal processing of N-glycans to a greater extent than modification of O-glycans.

Results and Discussion

Chemical synthesis and physical properties of DIBO

4-Dibenzocyclooctynol (3) was easily prepared starting from commercially available dibenzosuberenone (4, Scheme 1), which was treated with TMSCHN₂ in the presence of BF₃.Et₂O resulting in ring expansion by carbene insertion to give trieneone 5. ^[12] The latter compound was reduced with NaBH₄ (→6), brominated with bromine in chloroform (→7), and then treated with LDA in THF^[13] to give target compound 3 in a yield of 57%. In an alternative approach, intermediate 6 was obtained by base-mediated ring opening of Kagan’s ether, which was prepared by a double Friedel-Crafts alkylation phenylacetaldehyde.^[14] The former synthetic route provides a higher overall yield (71% vs. 42%), requires fewer chemical steps, and is more scalable.

Scheme 1.

Reagents and conditions. (a) TMSCHN₂, BF₃.OEt₂, DCM, −10 °C, 3 h, 71%; (b) NaBH₄, EtOH/THF, 7 h, 100%; (c) Br₂, CHCl₃, 0.5 h , 58%; (d) LDA, THF, 0.5 h, 57%; (e) 4-nitrophenyl chloroformate, pyridine, DCM, 18 h, 92%; (f) propanol amine, Et₃N, DCM, 3 h, 89%; (g) tris(ethylene glycol)-1,8-diamine, Et₃N, DCM, 3 h, 80%; (h); Dess-Martin, DCM, 0.5 h, 92%; (i) (carboxymethyl)hydroxyamine, DCM/MeOH, 61%; (j) N-{2-[2-(2-amino-ethoxy)-ethoxy]-ethyl}-2-aminooxy-acetamide, AcOH, DCM/MeOH, 63%; (k) Biotin-OSu, Et₃N, DCM/MeOH, 89% for 14, 92% for 15.

The hydroxyl of 3 provides an opportunity for further functionalization and for example can be activated with 4-nitrophenyl chloroformate to give activated carbonate 8, which can be reacted with amines such as propanol amine and tris(ethylene glycol)-1,8-diamine to give carbamates 9 and 10, respectively. Alternatively, the alcohol of 3 can be oxidized to ketone 11 using Dess Martin periodate, which can then be modified by aminooxy derivatives. This procedure was employed for the preparation of compounds 12 and 13. It was observed that oxime formation was rather sluggish and the rate of reaction could not significantly be increased by the addition of aniline.^[15] Finally, acylation of the amino groups of compounds 10 and 13 with N-(Biotinyloxy)succinimide (biotin-OSu) and Et₃N gave biotin-labeled derivatives 14 and 15, respectively.

It was observed that compound 3 has an excellent shelf life and remained intact after treatment with nucleophiles such as thiols and amines. However, upon exposure to azides a fast reaction took place to give the corresponding triazoles in high yield. Rate measurements of cycloaddition of compounds 3, 9, 11 and 12 were conducted by UV spectroscopy at 25±0.1 °C. A calculated amount of 0.25 M solutions of benzyl azide, required to achieve a desired azide concentration (6 × 10⁻⁴ - 1.5 × 10⁻² M), was added to a thermally equilibrated 6 × 10⁻⁵ M solution of cyclooctyne in methanol. The progress of the reactions was monitored by the decay of the characteristic absorbance of acetylenes at circa 317 nm. Consumption of starting material followed a first order equation and pseudo-first order rate constants were obtained by least-square fitting of the data to a single exponential equation. The rate dependence as a function of the concentration of azide was linear and least-square fitting of the data to a linear equation produced bimolecular rate constants summarized in Table 1. The rate constant of cycloaddition of DIBO (3) with benzyl azide is two orders of magnitude higher than that of cyclooctyne 1 and similar to that of DIFO. Conversion of the alcohol of DIBO into a carbamate, as in compound 9, did not influence the rate of cycloaddition. However, ketone 11 reacted three times faster than the parent alcohol 3. Probably, oxidation of 3 to 11 induces a small change in the conformation of the cyclooctyne ring, which may be responsible for the observed enhancement of reaction rate. In this respect, density functional theory (B3LYP) calculations of the transition states of cycloadditions of phenyl azide with acetylene and cyclooctyne indicate that the fast rate of the strain promoted cycloaddition is due to a lower energy required for distorting the 1,3-dipole and alkyne into the transition-state geometry.^[5] This finding has been exploited in the design of more reactive cyclooctynes and for example it has been found that installment of an amide or cyclopropanation in the eight membered ring leads to higher rates of cycloaddition.^[16] Surprisingly, oxime 12 did not exhibit a faster rate of reaction than alcohol 3. Computational studies will be required to provide a rationale for these observations.

Table 1.

Bimolecular rate constants for the reactions of acetylenes with benzyl azide in methanol at 25 ± 0.1 °C.

Cyclooctyne	Product	Rate (M−1S−1)
3	16	0.0567 ± 0.0027
9	17	0.0696 ± 0.0019
11	18	0.2590 ± 0.0067
12	19	0.0611 ± 0.0035

The acid labile oxime linkage of compound 15 provides opportunities for catch and release strategies, which for example are required for glycoproteomic applications. In this respect, treatment of triazole 21, which was formed by reaction of 15 with benzyl azide, at pH=2 for 10 h resulted in hydrolysis of the oxime linkage. On the other hand, compound 20, which was formed by reaction of derivative 14 with benzyl azide, was stable under these conditions.

It was observed that compound 11 exhibits fluorescent properties and in methanol emits light with a maximum fluorescent intensity of 436 nm and a quantum yield of 36% when excited at 375 nm. Oxime 12 showed a maximum fluorescent intensity of 380 nm and a quantum yield of 34% when excited at 313 nm (Figure 2). Interestingly, triazoles 18 and 19, which were formed by a reaction of 11 and 12, respectively with benzyl azide, showed only very weak fluorescence. Previously, a number of Cu(I) mediated fluorogenic alkyne-azide click (CuAAC) reactions have been described which are based on the formation of a fluorescent 1,3-triazole from a non-fluorescent precursor such as azide or alkyne substituted coumarins, 1,8-naphthalimides or anthracenes.^[17] The reactions of 11 and 12 with benzyl azide are the first examples of a metal free alkyne-azide fluoro-switch click reaction. A unique feature of this reaction is that the triazole moiety is responsible for fluorescent quenching whereas this effect is not observed for the parent cyclooctyne. It is to be expected that fluorogenic SPAAC will provide opportunities to monitor progress of reactions in real time. Such an approach will in particular be advantageous for application in which a fluorescent cyclooctyne is attached to a surface or solid support.^{[4, 18]}

Figure 2.

Fluorescence spectra of compounds 11 and 18 (excited at 375 nm) and compounds 12 and 19 (excited at 313 nm) in methanol (10⁻⁵ M). AU indicates arbitrary fluorescence units.

Evaluation of DIBO for labeling glycoconjugates of living cells

The bioorthogonal chemical reporter strategy is emerging as a versatile method for labeling biomolecules such as nucleic acids, lipids, proteins and carbohydrates.^{[1a, 1c, 1d]} In this approach, an abiotic chemical functionality (reporter) is incorporated into a target biomolecule, which can then be reacted with a complementary bioorthogonal reagent linked to a probe. Azide is commonly employed as a reporter and can be installed into biomolecules using azido-containing biosynthetic precursors that can be accepted by the cell’s native or engineered biosynthetic machinery. For example, azido-containing glycoconjugates can be biosynthesized by metabolic labeling with peracetylated N-α-azidoacetylmannosamine (Ac₄ManNAz), which is an appropriate substrate for the cell’s glycosylation machinery.^[19] A subsequently bioorthogonal reaction can then covalently attach a probe to the azido function which in turn makes it possible to conduct a multitude of functional studies. A number of bioorthogonal reactions have been described for reactions with azides, however, SPAAC is emerging as a particularly attractive approach as it can be performed under physiological conditions and does not require a toxic metal catalyst.

To establish biotin-modified DIBO derivatives 14 and 15 as appropriate bio-orthogonal reagents, we employed these compounds to determine relative quantities of cell surface sialylation of wild type and mutant cells and compare the results with traditional lectin staining. It is well established that Ac₄ManNAz can be employed by the glycosylation machinery to install azido-containing sialic acid in various glycoconjugates and a subsequent reaction with 14 or 15 was expected to provide quantitative data on cell surface sialylation.^[19] Thus, Jurkat cells were cultured in the presence of 25 µM of Ac₄ManNAz for three days to metabolically introduce N-azidoacetyl-sialic acid (SiaNAz) moieties into glycoproteins and glycolipids. As a negative control, Jurkat cells were employed that were grown in the presence of peracetylated N-acetylmannosamine (Ac₄ManNAc). A time course experiment was conducted by exposing the cells to 30 µM of 14 and 15 for different time periods at room temperature, washed and then stained with avidin-FITC for 15 min at 4 °C. The efficiency of the two-step cell surface labeling was determined by measuring the fluorescence intensity of the cell lysates. Gratifyingly, the ManNAz-labeled cell exhibited strong fluorescent readings after staining with the two different DIBO derivatives and the cell labeling was almost complete after a reaction time of 60 min (Figure 3), whereas the control cells gave a very low fluorescence intensity showing that background labeling is negligible.

Figure 3.

Time course of cell surface labeling with compounds 14 and 15. Jurkat cells grown for 3 days in the presence of Ac₄ManNAc or Ac₄ManNAz (25 µM) were incubated with compounds 14 or 15 (30 µM) for 0 – 90 min at room temperature. Next, cells were incubated with avidin-FITC for 15 min at 4 °C, after which cell lysates were assessed for fluorescence intensity. AU indicates arbitrary fluorescence units. Data (n=3) are presented as mean ± SD.

Similar results were obtained when Chinese Hamster Ovary (CHO) K1 cells were subjected to the same procedure.

Metabolically labelled cells were also examined by confocal microscopy (Figures 4 and S1). Thus, adherent Chinese hamster ovary (CHO) cells were cultured in the presence of Ac₄ManNAz (100 µM) for two days.^[20] Next, cell surface azido moieties were reacted with 14 or 15 (30 µM) for 1 h at ambient temperature, and then visualized with avidin-AlexaFluor 488 for 15 min at 4 °C. Staining was mainly observed at the cell surface and as expected, blank cells exhibited very low fluorescence staining confirming that background labeling is negligible. We also found that the two-step labeling approach with 14 and 15 had no effect on cell viability as determined by morphology and exclusion of trypan blue (data not shown).

Figure 4.

Fluorescence images of cells labeled with compounds 14 and 15 and avidin-AlexaFluor 488. CHO cells grown for 2 days in the presence of a, b) Ac₄ManNAc or c, d) Ac₄ManNAz (100 µM) were incubated with compounds a, c) 14 or b, d) 15 (30 µM) for 1 h at room temperature. Next, cells were incubated with avidin-AlexaFluor 488 for 15 min at 4 °C and, after washing, fixing, and staining for the nucleus with the far-red-fluorescent dye TO-PRO-3 iodide, imaged. Merged indicates that the images of cells labeled with AlexaFluor (488 nm) and TO-PRO (633 nm) are merged and shown in green and red, respectively.

The concentration-dependency of the cell surface labeling was studied by incubating Jurkat and CHO-k1 cells with various concentrations of 14 or 15 followed by staining with avidin-FITC (Figure S2). As expected, cells displaying azido moieties showed a dose-dependent increase in fluorescence intensity and reliable labeling was achieved at a concentration of 3 µM of 14 or 15, however optimal results were obtained at concentrations ranging from 30 to 100 µM. No increase in labeling was observed at concentrations higher than 100 µM due to limited solubility.

Jurkat and CHO-K1 cells were also metabolically labeled with peracetylated N-α-azidoacetylgalactosamine (Ac₄GalNAz, 100 µM), which can be metabolized by a number of cells and installed on mucin type glycoproteins.^[21] Subsequent treatment of the CHO-K1 cells with 14 followed by avidin-FITC resulted in strong fluorescent labeling whereas weak labeling was observed for the Jurkat cells (Figure S3). These results are in agreement with the well-known fact that CHO cells produce significant quantities of mucins whereas this is not the case for Jurkat cells.^[22]

Having established optimal conditions for SPAAC of azido-modified glycoconjugates of living cells with DIBO reagents, labeling studies were performed with a panel of lectin-resistant (Lec) mutant CHO cells. These cell lines (Lec2, Lec13 and Lec32), which exhibit unique structural changes in surface carbohydrates reflecting specific defects in glycosylation reactions, were expected to be ideally suitable for validation of the SPAAC methodology. Lec2 cells have a mutation in the open reading frame of the CMP-sialic acid transporter and therefore are unable to translocate CMP-sialic acid into the lumen of the Golgi apparatus, resulting in a marked reduction in glycoprotein and ganglioside sialylation.^[23] Although very small amounts of sialic acid containing glycoconjugates are made by these cells,^[24] the major class of glycans are asialo, core fucosylated N-glycans having LacNAc moieties.^[25] Lec32 mutants also exhibit a defect in sialylation due to a reduced expression of CMP-sialic acid synthetase. As a result, these cells have an increase in terminal β-galactoside residues on cell surface glycoproteins.^[26] Lec13 mutants exhibit a reduced expression of GDP-Man-4,6-dehydratase activity, which results in a decrease in GDP-fucose biosynthesis and underfucosylation of glycoproteins and glycolipids.^[27] The N-glycan profiles of these cells show increased levels of core nonfucosylated N-glycans with the most abundant N-glycans being asialo-, mono- or di-sialylated structures.

Wild-type CHO-K1 and Lec2, Lec13 and Lec32 mutant cells were cultured in the presence of Ac₄ManNAz or Ac₄ManNAz (100 µM) for 2 days and then exposed to biotin-modified DIBO 14 for 1 h at room temperature. Next, the cells were washed and labeled with avidin-FITC at 4 °C and the fluorescence intensity measured. As expected, the wild-type and Lec13 cells gave similar and strong fluorescence intensity readings (Figure 5A). On the other hand, the Lec2 and the Lec32 mutants showed a significant reduction in staining intensity and, in the case of the Lec2 cells, the readings were barely above control indicating that these cells express very low levels of surface sialosides. Thus, the results of these studies are in agreement with the previously described defects in the mutant cell lines and hence support the notion that the chemical reporter strategy can be employed to determine relative quantities of sialylation of glycoconjugates of living cells. Furthermore, treatment of the wild-type cells with Vibrio cholerae neuraminidase led to a similar fluorescent reading as control cells, confirming selective azide incorporation into sialic acid.

Figure 5.

SiaNAz expression in CHO-K1 and CHO-glycosylation mutant cells and effect of neuraminidase treatment on cell surface labeling. A) CHO-K1 and CHO mutant cells grown for 2 days in the presence of Ac₄ManNAc or Ac₄ManNAz (100 µM) were incubated, either directly or after treatment with V. cholerae neuraminidase (50 mU mL⁻¹) in serum-free culture medium for 2 h at 37 °C, with compound 14 (30 µM) for 1 h at room temperature. Next, cells were incubated with avidin-FITC for 15 min at 4 °C, after which cell lysates were assessed for fluorescence intensity. To assess the effects of desialylation of CHO-K1 and CHO-glycosylation mutant cells on their recognition by PNA and RCA1, cells were incubated, either directly or after treatment with V. cholerae neuraminidase (50 mU mL⁻¹) in serum-free culture medium for 2 h at 37 °C with B) PNA-FITC (50 µg mL⁻¹) or C) RCA1-FITC (50 µg mL⁻¹) for 45 min on ice in the dark. Next cell lysates were assessed for fluorescence intensity. AU indicates arbitrary fluorescence units. Data (n=3) are presented as mean ± SD. Similarly, D) CHO-K1 (a, c) and Lec2 (b, d) cells were imaged, either directly (a, b) or after treatment with V. cholerae neuraminidase (c, d) with PNA-FITC.

The results of the metabolic labeling studies were compared with traditional lectin staining using FITC-labeled peanut (Arachis hypogaea) agglutinin (PNA) and FITC-labeled Ricinus communis (castor bean) agglutinin type 1 (RCA1), which mainly recognize terminal β-Gal-(1–3)-GalNAc residues of O-linked structures and β-Gal-(1–4)-GlcNAc (LacNAc) found on N-linked glycoproteins, respectively. Cells that have intact sialylation machinery modify β-galactosyl residues with sialic acid and hence display low reactivity against PNA and RCA1 lectins. Indeed, the wild type and Lec13 cells gave fluorescent intensities just above background whereas the Lec2 and Lec32 mutants, which have a defect in sialylation, showed strong staining (Figure 5B–D).

Treatment of the wild type and Lec13 cells with V. cholerae neuramidase resulted in fluorescent intensities similar to that of the Lec2 and Lec32 cells, indicating that the various cell types express similar quantities of galactosyl-containing glycoproteins. Furthermore, a similar neuraminidase treatment of Lec2 cells followed by staining with PNA-FITC or RCA1-FITC did not lead to a significant increase in fluorescent intensity demonstrating that these cells do not significantly modify their cell surface glycoconjugates with sialic acid. On the other hand, neuraminidase treatment of Lec32 cells resulted in an increase in fluorescence staining with RCA1-FITC, whereas it did not impact the reading of PNA-FITC. These results indicate that the N-linked glycans of Lec32 contain some sialosides whereas this modification is absent in O-linked residues.

The lectin staining (Figure 5B–D) and metabolic labeling followed by SPAAC (Figure 5A) gave similar results and in particular both approaches showed that Lec2 cell express very small quantities of sialosides whereas the Lec32 mutant attach some sialic acid to their glycoconjugates. Surprisingly, a shortage of CMP-Neu5Ac as in the Lec32 cell resulted in differential sialylation of N- and O-linked glycans and it appears that N- but not O-linked glycans are modified by some sialic acid. Finally, both approaches showed that a defect in fucosylation does not impact the level of glycoconjugate sialylation. Attempts were made to directly assess differences in sialylation within the Lec cell mutants by utilizing lectins that recognize α(2,3)- or α (2,6)-linked terminal sialic acid residues (MAA and LFA). Surprisingly, we found fluorescence intensity for both lectins to be comparable in wild-type and Lec2 mutants (data not shown), suggesting that these lectins may recognize additional sugar structures other than terminal sialic acid and are not suitable for this study. Thus, we believe that the chemical reporter strategy provides a more reliable approach to determine relative differences in glycoprotein sialylation.

Having established biotin-modified DIBO 14 as a reliable reagent for detection of cell surface sialosides, attention was focused on sialylation of the CHO mutants Cog1 (ldlB) and Cog2 (ldlC). These cell lines were identified in a genetic screen for mutations that block low-density lipoprotein receptor (LDLR) activity.^[28] Further examinations have shown that these cell lines have defects in the Conserved Oligomeric Golgi (COG) complex, which is a protein complex consisting of eight subunits (Cog1–8) that play a critical role in retrograde vesicle transport and intra-Golgi trafficking. Malfunctions in the COG complex impact Golgi integrity, resulting in defects in protein sorting and glycosylation.^[29] Mutations in COG subunits have also been observed in humans and result in severe congenital disorders of glycosylation (CDG).^[30]

Metabolic labeling of Cog1 and Cog2 cells with ManNAz followed by a SPAAC with 14 and staining with avidin-FITC showed that these cells produce sialylated glycoconjugates, however at a significantly reduced level compared to wild-type CHO-K1 cells (Figure 5A). Staining with PNA-FITC demonstrated that that the cells expose terminal galactosyl residues on their O-linked glycans (Figure 5B). Furthermore, treatment of the cells with V. cholerae neuraminidase led to a similar staining intensity as for wild-type cells indicating that both cells types express similar quantities of galactosyl moieties and thus it appears that in Cog1 and 2 cells, glycoprotein sialylation of O-glycans is more severely imparted than galactosylation. Interestingly, a different staining profile was obtained when RCA1-FITC was employed (Figure 5C) and in this case untreated and neuraminidase exposed cells gave similar but reduced fluorescent intensities highlighting that N-glycan sialylation and galactosylation are both affected in the Cog1 and Cog2 mutants. These results suggest that loss of COG complex function may affect the localization and/or stability of glycosyltransferases involved in terminal processing of N-glycans to a greater extent than those enzymes that modify O-glycans. In support of this hypothesis, recent studies have shown that the stability of β(1,4)-galactosyltransferase is altered in COG-depleted HeLa cells due to altered trafficking and proteasomal degradation.^[31] At this point, we can however not rule out the possibility that these differences are due to the type and amount of glycoprotein cargo that is modified in the Golgi of these cells.

Conclusion

The past several years has seen a rapid development of the bioorthogonal chemical reporter methodology for the labeling of glycoconjugates of living cells and whole organisms. In this paper, we present a streamlined chemical approach for the preparation and derivatization of DIBO, which is an ideal bioorthogonal reagent for the chemical reporter strategy. Attractive features of DIBO include easy access by a simple and scalable synthetic approach, nontoxicity and straightforward attachment of a variety of probes. The use of several cell lines with known defects in glycoconjugate glycosylation validated DIBO as a reagent for determining relative quantities of cell surface glycoconjugate sialylation. The chemical reporter strategy in combination with lectin staining revealed that O-glycan sialylation of Cog1 and 2 cells is more severely impacted than galactosylation. Surprisingly, sialylation and galactosylation of N-glycans were similarly affected in these mutant cell lines. These results suggest that loss of COG complex function may differently affect the localization and/or stability of glycosyltransferases involved in terminal processing of N- and O-glycans. Differential modulation of N- and O-linked sialylation was also observed in Lec32 cells, which exhibit a reduced expression of CMP-sialic acid synthetase, and in this case N-linked oligosaccharides acquire some sialic acid moieties whereas this is not the case for O-linked structures. It is well known that different cell types express different ensembles of glycans. The results of this study indicate that a limited availability of sugar nucleotides is one way of a cell to selectively modulate the structures of glycoprotein glycans. We anticipate multiple applications of the described chemical reporter methodology in glycobiology and glycomedicine, including the tagging and isolation of glycoproteins from cell and tissue extracts as well as the investigation of trafficking and turnover of glycoconjugates in healthy and diseased cells.

Experimental Section

General methods and materials

Chemicals were purchased from Aldrich or Fluka and used without further purification. Dichloromethane was distilled from CaH₂ and stored over molecular sieves 4 Å. Pyridine was distilled from P₂O₅ and stored over molecular sieves 4 Å. THF was distilled form sodium. All reactions were performed under anhydrous conditions under an atmosphere of Argon. Reactions were monitored by TLC on Kieselgel 60 F254 (Merck). Detection was by examination under UV light (254 nm). Flash chromatography was performed on silica gel (Merck, 70–230 mesh). Iatrobeads (60 µm) were purchased from Bioscan. ¹H NMR (1D, 2D) and ¹³C NMR were recorded on a Varian Merc 300 spectrometer equipped with Sun workstations. ¹H and ¹³C NMR spectra were recorded in CDCl₃ or CD₃OD, and chemical shifts (δ) are given in ppm relative to solvent peaks (CDCl₃: ¹H, δ 7.24; ¹³C, δ 77.0. CD₃OD: ¹H, δ 3.31; ¹³C, δ 49.0.) as internal standard for compounds. High-resolution mass spectra were obtained by an Applied Biosystems 4700 MALDI mass spectrometer in positive ion reflective mode using 2,5-dihydroxyl-benzoic acid in CH₃CN as matrix.

6H-Dibenzo[a,e]cyclooctatrien-5-one (5)

To a stirred solution of dibenzosuberenone 4 (2.88 g, 14.0 mmol) and BF₃·OEt₂ (2.59 mL, 21.0 mmol) in CH₂Cl₂ (30 mL) was added dropwise a solution of trimethylsilyl diazomethane (10.5 mL, 21.9 mmol) in CH₂Cl₂ (20 mL) at −10 °C over a period of 1 h. The reaction mixture was stirred at −10 °C for 2 h, and then poured into ice water. The aqueous layer was extracted with CH₂Cl₂ (3 × 100 mL) and the combined organic layers washed with brine, dried (MgSO₄), filtered and the filtrate concentrated under reduced pressure. The crude product was purified by flash chromatography over silica gel (2/1→1/2, v/v, hexanes/CH₂Cl₂) to give compound 5 as an amorphous solid (2.22 g, 72%). ¹H NMR (300 MHz, CDCl₃): δ 8.26 (1H, q, J = 1.4, 6.6 Hz), 7.13–7.43 (7H, m), 7.05 (2H, q, J = 3.8, 12.9 Hz), 4.06 (2H, s). ¹³C NMR (75 MHz, CDCl₃): δ 196.6, 136.9, 136.3, 135.4, 133.8, 133.1, 132.4, 131.4, 130.6, 129.3, 128.8, 128.0, 127.3, 126.9, 48.4. MALDI HRMS: m/z 243.0767 [M+Na⁺]. Calcd for C₁₆H₁₂NaO⁺: 243.0780.

5,6-Dihydro-dibenzo[a,e]cycloocten-5-ol (6)

To a stirred solution of 5 (2.20 g, 10 mmol) in a mixture of EtOH and THF (1/1, v/v, 120 mL) was slowly added sodium borohydride (0.757 g, 20 mmol). The reaction mixture was stirred for 7 h, after which TLC analysis indicated completion of the reaction. The reaction was quenched by slow addition of acetic acid (1 mL) and the solvents were evaporated. The residue was dissolved in CH₂Cl₂ (100 mL) and the resulting solution was washed with brine (100 mL), which was back extracted with CH₂Cl₂ (4 × 100 mL). The combined organic phases were dried (MgSO₄), filtered and concentrated under reduced pressure to give 6 as a white solid (2.22 g), which was directly used in the next step reaction without further purification. ¹H NMR (300 MHz, CDCl₃): δ 7.50 (1 H, m), 7.14–7.30 (7 H, m), 6.90 (2 H, q, J = 2.7, 12.0 Hz), 5.31 (1 H, q, J = 6.3, 10.0 Hz), 3.41 (2 H, m). ¹³C NMR (75 MHz, CDCl₃): δ 141.7, 136.7, 136.2, 134.5, 131.7, 131.5, 130.1, 129.9, 129.3, 128.7, 127.4, 127.2, 126.9, 125.9, 74.4, 42.7. MALDI HRMS: m/z 245.0949 [M+Na⁺]. Calcd for C₁₆H₁₄NaO⁺: 245.0937.

11,12-Dibromo-5,6,11,12-tetrahydro-dibenzo[a,e]cycloocten-5-ol (7)

To a stirred solution of 6 (2.22 g, 10 mmol) in CHCl₃ (50 mL) was added dropwise bromine (0.51 mL, 10 mmol). After stirring the reaction mixture for 0.5 h, TLC analysis indicated completion of the reaction. The solvent was evaporated under reduced pressure and the residue was purified by flash chromatography over silica gel (2/1→1/2, v/v, hexanes/CH₂Cl₂) to give 7 as light yellow oil (2.22 g, 58%). ¹H NMR (300 MHz, CDCl₃): δ 7.54-7.47 (2 H, aromatics), 7.31- 6.72 (6 H, aromatics), 5.77 (1 H, d, J = 5.4 Hz, CHBr), 5.22 (1 H, dd, J = 3.6, 15.9 Hz, CHOH), 5.19 (1 H, d, J = 5.4 Hz, CHBr), 3.50 (1 H, dd, J = 3.6, 15.9 Hz, CH₂), 2.75(1 H, dd, J = 3.6, 15.9 Hz, CH₂). ¹³C NMR (75 MHz, CDCl₃): δ 141.3, 140.0, 137.2, 134.0, 133.4, 131.5, 131.3, 130.9, 127.8, 126.2, 123.7, 121.3, 76.5, 70.0, 62.3, 32.2. MALDI HRMS: m/z 402.9313 [M+Na⁺]. Calcd for C₁₆H₁₄Br₂NaO⁺: 402.9304.

5,6-Dihydro-11,12-didehydro-dibenzo[a,e]cycloocten-5-ol (3)

To a stirred solution of 7 (1.53 g, 4.0 mmol) in tetrahydrofuran (40 mL) was added dropwise lithium diisopropylamide in tetrahydrofuran (2.0 M) (8 mL, 16 mmol) under an atmosphere of Argon. The reaction mixture was stirred for 0.5 h, after which it was quenched by the dropwise addition of water (0.5 mL). The solvents were evaporated under reduced pressure, and the residue was purified by flash chromatography on silica gel (2/1→ 0/1 v/v, hexanes/CH₂Cl₂) to give 3 as a white amorphous solid (0.50 g, 57%). ¹H NMR (300 MHz, CDCl₃): δ 7.67 (1 H, aromatics), 7.37-7.18 (7 H, aromatics), 4.57 (1 H, dd, J = 2.1, 14.7 Hz, CHOH), 3.04 (1 H, dd, J = 2.1, 14.7 Hz, CH₂), 2.86(1 H, dd, J = 2.1, 14.7 Hz, CH₂). ¹³C NMR (75 MHz, CDCl₃): δ 154.5, 150.6, 128.6, 127.1, 1127.0, 126.0, 125.8, 125.1, 124.7, 123.0, 122.7, 121.7, 111.9, 109.6, 74.2, 47.7.

Carbonic acid, 5,6-dihydro-11,12-didehydrodibenzo[ a,e]cycloocten-5-yl ester, 4-nitrophenyl ester (8)

To a solution of 3 (0.22 g, 1 mmol) in CH₂Cl₂ (30 mL) was added 4-nitrophenyl chloroformate (0.4 g, 2 mmol) and pyridine (0.4 mL, 5 mmol). After stirring for 4 h at room temperature, the mixture was washed with brine (2 × 10 mL), and the organic layer was dried (MgSO₄). The solvents were evaporated under reduced pressure, and the residue was purified by silica gel column chromatography (hexane/ethyl acetate, 10/1, v/v) to afford 8 (0.34 g, 89%). ¹H NMR (300 MHz, CDCl₃): δ 8.23-8.18 (2H, aromatics), 7.56-7.54 (2H, aromatics), 7.46-7.18 (8H, aromatics), 5.52 (1H, dd, J = 3.9, 15.3 Hz, CHOH), 3.26 (1H, dd, J = 3.9, 15.3 Hz, CH₂), 2.97 (1H, dd, J = 3.9, 15.3 Hz, CH₂); ^I3C NMR (75 MHz, CDCl₃): δ 154.5, 150.7, 149.1, 148.7, 129.0, 127.4, 127.3, 126.7, 126.5, 125.5, 125.2, 124.3, 124.0, 122.6, 122.4, 120.8, 120.6, 120.2, 112.2, 108.5, 80.6, 44.8; MALDI HRMS: m/z 408.0852 [M + Na⁺]. Calcd for C₂₃H₁₅NNaO₅ ⁺: 408.0842.

3-Hydroxypropyl-carbamic acid 5,6-dihydro-11,12-didehydrodibenzo[a,e]cycloocten-5-yl ester (9)

To a stirred solution of 8 (38 mg, 0.1 mmol) in CH₂Cl₂ (15 mL) was added 3-aminopropan-1-ol (15 mg, 0.2 mmol) and triethylamine (10 µL). The reaction mixture was stirred at room temperature for 12 h, after which the solvents were evaporated under reduced pressure and the residue purified by silica gel column chromatography (CH₂Cl₂/CH₃OH, 20/1, v/v) to afford 9 (25 mg, 77%). ¹H NMR (CDCl₃, 300 MHz) δ: 6.94–7.43 (m, 8H, aromatics), 5.42 (m, 1H, Ph-CH-O), 3.61(m, 2H, CH₂OH), 3.30 (m, 2H, CH₂NH), 3.08 (dd, 1H, J= 15.0, 1.8 Hz, PhHCH), 2.84 (dd, 1H, J= 15.0, 3.9 Hz, PhHCH), 1.53–1.68 (m, 2H, CH₂CH₂OH). ¹³C NMR (75 MHz, CDCl₃) δ: 150.8, 149.1, 128.9, 128.0, 127.0, 126.1, 126.0, 125.9, 125.8, 125.3, 125.1, 125.0, 122.8,122.6, 120.3, 111.9, 108.9, 58.6, 45,2, 36.8, 36.7, 31.6. MALDI HRMS: m/z 344.1246 [M+Na⁺]. Calcd for C₂₀H₁₉O₃NNa⁺: 344.1257.

{2-[2-(2-Amino-ethoxy)-ethoxy]-ethyl}-carbamic acid 5,6-dihydro-11,12-didehydro-dibenzo[a,e]cycloocten-5-yl ester (10)

To a stirred solution of 8 (77 mg, 0.2 mmol) and tris(ethylene glycol)-1,8-diamine (0.293 mL, 2 mmol) in CH₂Cl₂ (20 mL) was added Et₃N (0.139 mL, 1.0 mmol). The reaction mixture was stirred for 3 h, after which the solvent was removed under reduced pressure. The residue was purified by flash chromatography over Iatrobeads (8→30%, v/v, MeOH/CH₂Cl₂) to give 10 as a light yellow amorphous solid (0.063 g, 80%). ¹H NMR (300 MHz, CDCl₃): δ 7.51 (1H, d, J = 7.3 Hz), 7.24–7.37 (7H, m), 5.81 (1H, s, NH), 5.48 (1H, br), 3.50–3.68 (8H, m), 3.39 (2H, m), 3.16 (1 H, d, J = 14.8 Hz), 2.91 (2H, br), 2.88 (1H, d, J = 14.8 Hz), 2.57 (2H, br, NH₂). ¹³C NMR (75 MHz, CDCl₃): δ 155.7, 152.2, 151.1, 130.0, 128.1, 128.0, 127.2, 127.1, 126.3, 126.0, 123.9, 123.8, 121.3, 113.0, 110.0, 76.7, 72.8, 70.3, 70.2, 70.1, 70.0, 46.2, 41.5, 41.0. MALDI HRMS: m/z 417.1766 [M+Na⁺]. Calcd for C₂₃H₂₆N₂NaO₄ ⁺: 417.1785.

6H-11,12-Didehydro-dibenzo[a,e]cyclooctatrien-5-one (11)

To a stirred solution of 3 (0.172 g, 0.78 mmol) in CH₂Cl₂ (40 mL) was added Dess-Martin reagent (0.40 g, 0.94 mmol). The reaction mixture was stirred for 0.5 h after which TLC analysis indicated completion of the reaction. The reaction mixture was filter through a short pad of silica gel, which was washed with CH₂Cl₂. The filtrate was concentrated, and the residue was purified by flash chromatography over silica gel (1/1→0/1, v/v, hexanes/CH₂Cl₂) to give 11 as a white amorphous solid (0.158 g, 92%). ¹H NMR (300 MHz, CDCl₃): δ 7.29–7.57 (8H, m), 4.17 (1H, d, J = 10.6 Hz), 3.64 (1H, J = 10.6 Hz). ¹³C NMR (75 MHz, CDCl₃): δ 200.4, 154.7, 148.2, 131.21 (2 C), 131.18, 129.3, 128.2, 127.8, 126.3, 125.9, 122.2, 111.1, 109.4, 49.3. MALDI HRMS: m/z 241.0638 [M+Na⁺]. Calcd for C₁₆H₁₀NaO⁺: 241.0624.

(6H-11,12-didehydro-dibenzo[a,e]cycloocten-5-ylideneaminooxy)-acetic acid (12)

A solution of 6H-11,12-didehydro-dibenzo[a,e]cyclooctatrien-5-one 11 (21.8 mg, 0.1 mmol) and (carboxymethyl)hydroxylamine hemihydrochloride (21.8 mg, 0.2 mmol) in a mixture of MeOH, CH₂Cl₂ and HOAc (8 mL, 1/1/0.02, v/v/v) was stirred for 2 days. The solvents were removed under reduced pressure, and the residue was purified by flash chromatography on silica gel (EtOAc) to give 12 as an amorphous white solid (17.8 mg, 61%). ¹H NMR (300 MHz, CDCl₃): δ 7.54 (1H, d, J = 7.4 Hz), 7.46 (1H, d, J = 7.4 Hz), 7.18–7.39 (6 H, m), 4.53 (2 H, m), 4.23 (1 H, d, J = 12.8 Hz), 3.16 (1 H, d, J = 12.8 Hz). ¹³C NMR (75 MHz, CDCl₃): δ 175.2 & 173.6, 154.1 153.2, 130.7,129.5, 19.3, 129.2, 129.1, 128.1, 128.0, 127.1, 126.9, 125.5, 125.2, 122.7, 113.9, 111.2, 84.7, 68.3 & 67.1, 35.0 & 33.2. MALDI HRMS: m/z 314.0770 [M+Na⁺]. Calcd for C₁₈H₁₃NNaO₃ ⁺: 314.0788.

N-{2-[2-(2-Amino-ethoxy)-ethoxy]-ethyl}-2-(6H-11,12-didehydrodibenzo[a,e]cycloocten-5-ylideneaminooxy)-acetamide (13)

A solution of 11 (46 mg, 0.211 mmol), N-{2-[2-(2-amino-ethoxy)-ethoxy]-ethyl}-2-aminooxy-acetamide (84 mg, 0.251 mmol) and acetic acid (0.1 mL) in a mixture of MeOH and CH₂Cl₂ (4 mL, 1/1 v/v) was stirred for 2 days. The solvents were evaporated under reduced pressure, and the residue was purified by flash chromatography over Iatrobeads (4→15%, v/v, MeOH/CH₂Cl₂) to give 13 as a light yellow solid (56 mg, 63%). ¹H NMR (300 MHz, CD₃OD): δ 7.73 (1 H, d, J = 7.6 Hz), 7.30–7.56 (7 H, m), 4.48 (2 H, m), 4.35 (1 H, d, J = 14.7 Hz), 3.50–3.68 (8H, m), 3.24 (1 H, d, J = 13.6 Hz), 3.08 (2H, d, J = 5.3 Hz), 3.02 (2H, d, J = 5.3 Hz). ¹³C NMR (75 MHz, CD₃OD): δ 172.2, 162.0, 150.0, 149.0, 133.2, 131.0, 130.6, 129.8, 129.3, 128.5, 127.1, 126.8, 124.9, 124.3, 111.2, 110.5, 73.8, 71.4, 71.3, 70.4, 67.8, 40.6, 39.7, 39.1. MALDI HRMS: m/z 444.1875 [M+Na⁺]. Calcd for C₂₄H₂₇N₃NaO₄ ⁺: 444.1894.

{2-[2-(2-Biotinylamino-ethoxy)-ethoxy]-ethyl}-carbamic acid 5,6-dihydro-11,12-didehydro-dibenzo[a,e]cycloocten-5-yl ester (14)

To a solution of 10 (15.8 mg, 0.04 mmol) and biotin-OSu (20.5 mg, 0.06 mmol) in a mixture of MeOH and CH₂Cl₂ (7 mL, 1:1 v/v) was added 3 drops of Et₃N. The reaction mixture was stirred for 2 h, and then the solvents were removed under reduced pressure. The residue was purified by column chromatography over Iatrobeads (5–15% v/v MeOH/CH₂Cl₂) to give 14 as light yellow amorphic solid (22.1 mg, 89%). ¹H NMR (300 MHz, CD₃OD): δ 7.59 (1 H, aromatics), 7.42-7.33 (7H, aromatics), 5.44, (1H, dd, J = 5.0, 14.1 Hz, ArCHOH), 4.60, 4.46 (m, 2H, CHNH), 4.24 (s, 4H, OCH₂CH₂O), 3.72 (m, 4H, OCH₂), 3.64 (m, 2H, CH₂NH), 3.55 (m, 1H, CHS), 3.33 (dd, 1H, J = 4.8, 12.0 Hz), 3.23 (t, 2H, J = 6 Hz, CH₂NH₂), 3.22, (1H, dd, J = 5.0, 14.1 Hz, CH₂), 2.88, (1 H, dd, J = 5.0, 14.1 Hz, CH₂), 2.68 (d, 1H, J = 12.45 Hz), 2.20 (t, 2H, J = 7.5 Hz, CH₂CO), δ 1.4 (m, 6H, biotin-CH₂). ¹³C NMR (75 MHz, CD₃OD): δ 175.0, 164.9, 156.9, 152.5, 151.3, 129.9, 128.2, 128.1, 127.2, 127.1, 126.0, 125.7, 123.8, 121.2, 112.7, 109.8, 76.8, 70.2, 70.1, 69.8, 69.4, 62.1, 60.4, 55.8, 54.6, 46.0, 42.6, 40.6, 39.9, 39.1, 35.5, 28.6, 28.3, 25.6, 17.5, 16.1, 12.0; MALDI HRMS: m/z 643.2575 [M+Na⁺]. Calcd for C₃₃H₄₀N₄NaO₆S⁺ 643.2561.

N-{2-[2-(2-Biotinylamino-ethoxy)-ethoxy]-ethyl}-2-(6H-11,12-didehydro-dibenzo[a,e]cycloocten-5-ylideneaminooxy)-acetamide (15)

To a solution of 13 (8.4 mg, 0.02 mmol) and Biotin-OSu (10.2 mg, 0.03 mmol) in MeOH/CH₂Cl₂ (4 mL, 1/1, v/v) was added 2 drops of Et₃N. The reaction mixture was stirred for 2 h, after which the solvents were removed under reduced pressure. The residue was purified by column on Iatrobeads (5→15 v/v MeOH/CH₂Cl₂) to give 15 as a light yellow solid (11.9 mg, 92%). ¹H NMR (300 MHz, CD₃OD): δ 7.64 (1H, aromatics), 7.54-7.30 (7H, aromatics), 4.47 (m, 3H), 4.34 (m, 1H), 3.67 (s, 4H, OCH₂), 3.60-3.40 (m, 3H), 3.30 (m, 2H), 3.20 (m, 3H), 2.96-2.86 (m, 2H), 2.20 (t, 2H, J = 7 Hz, biotin-CH₂CO), 1.80-1.50 (m, 4H, biotin-CH₂) 1.48-1.38 (m, 2H, biotin-CH₂). MALDI HRMS: m/z 670.2665 [M + Na⁺]. Calcd for C₃₄H₄₁N₅NaO₆S⁺ 670.2670.

General procedure for the preparation of triazoles (16–19)

A solution of dibenzocyclooctyne derivative (3, 9, 11 or 12, 0.1 mmol) and benzyl azide (0.1 mmol) in MeOH (10 mL) was stirred for 3 h. The solvent was evaporated under reduced pressure, and the residue was purified by silica gel column chromatography to give the desired product (16, 17, 18 or 19, respectively) in a quantitative yield.

Compound 16: ¹H NMR (300 MHz, CD₃OD): δ 8.29-7.98 (1H, aromatic), 7.70-6.80 (12H, aromatics), 5.84-5.25 (2H, PhCH₂), 5.31-4.70 (1H, CHOH), 3.77-2.55 (2H, ArCH₂). MALDI HRMS: m/z 377.1507 [M + Na⁺]. Calcd for C₂₃H₂₀N₃NaO⁺ 377.1499.

Compound 17: ¹H NMR (300 MHz, CD₃OD): δ 8.00-6.90 (13H, aromatics), 6.03-5.26 (2H, PhCH₂), 5.10-4.77 (1H, CHOCO), 3.58 (2H, CH₂OH), 3.29 (2H, NHCH₂), 3.20-2.64 (2H, ArCH₂), 1.62 (2H, CH₂CH₂OH). MALDI HRMS: m/z 477.1889 [M + Na⁺]. Calcd for C₂₇H₂₆N₄NaO₃ ⁺ 477.1897.

Compound 18: ¹H NMR (300 MHz, CD₃OD): δ 8.09-6.92 (13H, aromatics), 5.63 (2H, PhCH₂), 3.70 (2H, COCH₂). MALDI HRMS: m/z 375.1328 [M + Na⁺]. Calcd for C₂₃H₁₈N₃NaO⁺ 375.1342.

Compound 19: ¹H NMR (300 MHz, CD₃OD): δ 7.59-7.06 (11H, aromatics), 7.04-6.97 (2H, aromatics), 5.89-5.32 (2H, PhCH₂), 4.68-4.43 (2H, CH₂CO₂H), 4.30-4.03 (1H), 3.07-2.96 (1H). MALDI HRMS: m/z 447.1439 [M + Na⁺]. Calcd for C₂₅H₂₀N₄NaO₃ ⁺ 447.1428.

Fluorescence measurement

The fluorescence spectra of compounds 11 and 18 (excited at 375 nm) and compounds 12 and 19 (excited at 313 nm) in methanol (10⁻⁵ M) were recorded on a spectrofluorometer FluoroMax-3 from Horiba Jobin Yvon.

Kinetics of the cycloaddition reaction

The rate measurements of cycloaddition of cyclooctynes with benzyl azide were conducted by UV spectroscopy at 25 ± 0.1 °C. A calculated amount of 0.25 M solutions of benzyl azide required to achieve desired azide concentration (6 × 10⁻⁴ - 1.5 × 10⁻² M) was added to a thermally equilibrated 6 × 10⁻⁵ M solution of acetylene in MeOH. Reactions were monitored by following the decay of the characteristic absorbance of acetylenes at ca. 317 nm. Consumption of starting material followed a first order equation and the pseudo first order rate constants were obtained by least-squares fitting of the data to a single exponential equation. The rate dependence as a function of the concentration of azide was linear. Least-squares fitting of the data to a linear equation produced bimolecular rate constants summarized in Table 1. In this respect, the UV spectroscopic method can be performed under pseudo first order conditions over a wide range of reagent concentrations making the analysis of second-order kinetic curves more reliable.

Reagents for biological experiments

Synthetic compounds 14 and 15 were reconstituted in DMF and stored at −80 °C. Final concentrations of DMF never exceeded 0.56% to avoid toxic effects. Ac₄ManNAc, Ac₄ManNAz, Ac₄GalNAc and Ac₄GalNAz were synthesized as reported^[32] and reconstituted in ethanol. Avidin-FITC and avidin-AlexaFluor 488 were obtained from Molecular Probes, V. cholerae neuraminidase was from Sigma-Aldrich, and PNA-FITC and RCA1-FITC were from EY Laboratories.

Cell culture conditions

Human Jurkat cells (Clone E6-1; ATCC) were cultured in RPMI 1640 medium (ATCC) with L-glutamine (2 mM), adjusted to contain sodium bicarbonate (1.5 g L⁻¹), glucose (4.5 g L⁻¹), HEPES (10 mM), and sodium pyruvate (1 mM). Chinese hamster ovary (CHO) cells (Clone K1; ATCC) were cultured in Kaighn’s modification of Ham’s F12 medium (ATCC) with L-glutamine (2 mM), adjusted to contain sodium bicarbonate (1.5 g L⁻¹). Mutant CHO cells (Lec2, Lec13, Lec32 mutants obtained from Dr. Pamela Stanley and Cog1 and Cog2 mutants (ldlB and ldlC) obtained from Dr. Monty Kreiger) were cultured in Minimum Essential Medium Alpha 1X (Cellgro) with Earle’s salts, ribonucleosides, deoxyribonucleosides and L-glutamine (2 mM). All media were supplemented with penicillin (100 u mL⁻¹) / streptomycin (100 µg mL⁻¹; Mediatech) and fetal bovine serum (FBS, 10%; Hyclone). Cells were maintained in a humid 5% CO₂ atmosphere at 37 °C.

Cell surface azide labeling

Jurkat cells were seeded at a density of 75,000 cells mL⁻¹ in a total volume of 40 mL culture medium in the presence of Ac₄ManNaz or Ac₄GalNaz (25 µM final concentration) and grown for 3 days, leading to the metabolic incorporation of the corresponding N-azidoacetyl sialic acid (SiaNAz) into their cell surface glycoproteins. Control cells were grown in the presence of Ac₄ManNac or Ac₄GalNac (25 µM final concentration) for 3 days. CHO cells were plated in 12-well plates (250,000 cells/well) and grown in medium that contained Ac₄ManNaz or Ac₄GalNaz (100 µM) and as control cells Ac₄ManNAc or Ac₄GalNac (100 µM) for 2 days. Expected cell number on day of click chemistry was ~1 × 10⁶ cells/well.

Sialidase pre-treatment

Cells were washed twice with serum-free culture medium and incubated with V. cholerae neuraminidase (50 mU mL⁻¹) in serum-free culture medium for 2 h at 37 °C, washed in PBS and subjected to the respective assay.

Click chemistry and detection by fluorescence intensity

Jurkat cells bearing azides and control cells were washed with labeling buffer (DPBS, pH 7.4 containing FBS (1%) and BSA (1%)) and transferred to round bottom tubes (1 × 10⁶ cells/sample). CHO cells (untreated or sialidase pre-treated) were left in the 12-well plates (~1 × 10⁶ cells/sample) and washed with labeling buffer. Next cells were incubated with the biotinylated compounds 14 or 15 (0–100 µM) in labeling buffer for 0–90 min at room temperature. The cells were washed three times with cold labeling buffer and then incubated with avidin-FITC (5 µg mL⁻¹) for 15 min at 4 °C in the dark. Following three washes and cell lysis in passive lysis buffer (Promega), cell lysates were analysed for fluorescence intensity (485 ex/520 em) using a microplate reader (BMG Labtech). Data points were collected in triplicate and are representative of three separate experiments. Fluorescence of Jurkat cell lysates was expressed as fluorescence (arbitrary units; AU) per 800,000 cells. CHO cell lysates were assayed for total protein using the bicinchoninic acid assay (BCA; Pierce Biotechnology) and fluorescence intensity was expressed as fluorescence (AU) per µg total protein.

Lectin binding assay

Untreated or sialidase pre-treated cells (~1 × 10⁶) were washed twice in cold PBS and subsequently incubated in 300 µL PBS containing PNA-FITC (50 µg mL⁻¹) or RCA1-FITC (50 µg mL⁻¹) for 45 min on ice in the dark. After washing with cold PBS, the cells were lysed in passive lysis buffer (Promega) and cell lysates were analyzed for fluorescence intensity (485 ex/520 em) using a microplate reader.

Detection of cell labeling and lectin staining by fluorescence microscopy

For cell surface labeling: CHO-K1 cells labeled with Ac₄ManNAc or Ac₄ManNAz (100 µM) for 2 days were seeded at a density of 50,000 cells/coverslip (22 mm) and allowed to adhere overnight in their original medium. After two washes with wash buffer (DPBS, supplemented with FBS (1%)), live cells were incubated with biotinylated compounds 14 or 15 (30 µM) in wash buffer for 1 h at room temperature, followed by three washes in wash buffer (10 min/wash). Next, the cells were incubated with avidin conjugated with AlexaFluor 488 (5 µg mL⁻¹) for 15 min at 4 °C. Cells were washed 3 times with wash buffer and fixed with formaldehyde (3.7% in PBS) at room temperature for 15 min. After the coverslips were washed four times in PBS (5 min/wash), the nucleus was labeled with the far red-fluorescent TO-PRO-3 iodide dye (Molecular Probes). The cells were mounted with PermaFluor (Thermo Electron Corporation) before imaging.

For lectin staining of cell surface glycans: CHO-K1 and Lec2 cells were seeded at a density of 50,000 cells/coverslip (22 mm) and allowed to adhere overnight. After two washes with serum-free culture medium, live cells were treated with V. cholerae neuraminidase (50 mU mL⁻¹) in serum-free culture medium for 2 h at 37 °C. Coverslips were washed with DPBS and incubated with PNA-FITC (50 µg mL⁻¹) in PBS supplemented with BSA (1%) for 45 min at 4 °C. Cells were washed with PBS and fixed and mounted as above.

Initial analysis was performed on a Zeiss Axioplan2 fluorescent microscope. Confocal images were acquired using a 60X (NA1.42) oil objective. Stacks of optical sections were collected in the z dimensions. The step size, based on the calculated optimum for each objective, was between 0.25 and 0.5 µm. Subsequently, each stack was collapsed into a single image (z-projection). Analysis was performed offline using ImageJ 1.39f software (National Institutes of Health, USA) and Adobe Photoshop CS3 Extended Version 10.0 (Adobe Systems Incorporated), whereby all images were treated equally.

Statistical Analysis

Statistical significance between groups was determined by two-tailed, unpaired Student’s t test. Differences were considered significant when P<0.05.