A universal strategy for direct immobilization of intact bioactivity-conserved carbohydrates on gold nanoparticles

Abstract

Carbohydrate-functioned gold nanoparticles have been highlighted for sensing complicated carbohydrates but their performance is far from ideal due to the lack of bioaffinity-conserved ways to function the particles with intact carbohydrates. Herein, we present an innovative method to directly immobilize intact carbohydrates on gold nanoparticles with cyanuric chloride as a linker. Both small and large carbohydrates were successfully immobilized on GNPs at a surface coverage of 43.5% (tested by mannose). Pre-treatment of the carbohydrates was avoided and no loss of their recognition ability towards lectins was observed after immobilization, even in the case of monosaccharides. The mechanism was attributed to the three σ-linking bonds which are rotatable and spaced at 120°, enabling the immobilized carbohydrates to find their best spatial position to favour their recognition. The method can minimize immobilizing chemical reactions and ease the manipulation by simply controlling the reaction temperature. It is cost-effective and extendable to the immobilization of other hydroxyl compounds and amines as well.

---

Introduction

Carbohydrates are of great significance to reveal and understand physiological processes and disease mechanisms, to help develop and screen new drugs, and to guide clinical diagnosis and treatment.^1–3 They have become a major target and been used as biomimetic functional molecules for surface modification because of their specific molecular characteristics and actions in living systems.⁴ However, they are the most complicated substances and their detection and analysis are full of challenges. The state-of-the-art of their research includes the exploration of chemo-and bio-sensors based on nanoparticles (NPs). One of the most common studied sensing strategies is based on gold nanoparticles (GNPs), for example, GNPs-based local surface plasmon resonance (LSPR) which is a promising analytical platform for assaying different glycomic samples and for the diagnosis or treatment of carbohydrates-dependent diseases such as glioma and stroke.^5–14

One of the prerequisite steps in the exploitation of GNPs-based sensors is to functionalize the GNPs with a target carbohydrate. Physical adsorption and chemical reaction are two easily thought of strategies, but the physical adsorption is not very suitable for carbohydrates which have much less absorptibility than other molecules such as proteins. Chemical reaction is thus the top choice which forms covalent bond between carbohydrates and GNPs, being more stable in applications.^{4,5,11,15–18} Nevertheless carbohydrates are less reactive than other biochemicals (e.g., proteins), the easiest adaptable reaction lies in the aldehyde group which allows to directly immobilize carbohydrates on hydrazide- or hydroxylamine-terminated GNPs. After reductive amination reaction, the immobilized carbohydrates are very stable. The problem is that it is only applicable to reducing carbohydrates, and easily generates cyclic products which lose or change the recognition affinities of carbohydrates, harmful to monosaccharides.^19,20

Another widely studied technique is chemisorption of thiolated carbohydrates on GNPs. Thiols can easily deposit on gold surface by formation of Au–S covalent bond. But the bottleneck is the difficulty to introduce thiol groups onto carbohydrates by common synthetic approaches.^{4,5,11,15–18}

The third way is to immobilize carbohydrates by photochemical reactions. For example, with perfluorophenyl azide (PFPA) as a linker, all intact carbohydrates can be immobilized on gold surface. However, PFPA has not yet been commercialized, need one to synthesize in laboratory. Furthermore, PFPA-based photochemical immobilization has to be performed in solid state to reach high immobilization ratio. This is not very convenient for carbohydrate solutions, and the recognition affinity will be lost for the immobilization of small carbohydrates.²¹

A common problem in use of chemical adsorption strategy is the loss of recognition affinity of the immobilized carbohydrates. Although some of the loss may be recovered by introduction of a “soft arm” between the gold surface and the linker or between the linker and carbohydrates, it complicates the immobilization. A better immobilization chemistry is thus highly desirable.

We have thus focused on the exploitation of a method able to directly anchor the intact carbohydrates on GNPs, with chemistry as easy as possible. Considering that a carbohydrate molecule contains many hydroxyl groups which have not yet been well explored, the hydroxyl-oriented reactions were researched and tried, of which cyanuric chloride (CC)-based reactions^22–25 were systematically checked and tried. In this paper, an innovated CC-based method is established for the direct immobilization of intact or underivatized carbohydrates on GNPs by one-pot synthesis (Scheme 1).

Scheme 1 Schematic illustrations of directly immobilizing the intact carbohydrates on gold nanoparticles (GNPs) via cyanuric chloride (CC) chemistry and of the interaction of the obtained carbohydrate-modified GNPs with lectins. The chemistry includes three key steps: formation of hydroxyl-terminated GNP surface; modification of the hydroxyl terminal with CC; and immobilization of carbohydrate on the CC-terminated GNP surface.

CC is a heterocyclic tri-functional compound, its most prominent feature is that the three chlorines atoms on the triazine ring can be stepwise replaced by hydroxyl or amino groups through the regulation of reaction temperature. The formed σ-linking bond can turn around to adjust their spatial position, thus endows the molecules immobilized on a solid surface to rotate for better recognition position. This immobilization strategy has been applied to the preparation of carbohydrates-functioned chips in our laboratory.²⁶ Here, we demonstrate that the CC-based chemistry is ideally applicable to the immobilization of intact and/or natural carbohydrates on GNPs, with bioaffinity towards lectins well conserved. Inherently, the method has cut the chemical steps to the minimum and much eased the immobilizing manipulation. The method is extendable to the immobilization of other hydroxy and amino compounds on GNPs.

Experimental section

Materials and reagents

Concanavalin A (Con A), peanut agglutinin (PNA), and 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide (EDC) hydrochloride were purchased from Sigma (St. Louis, Mo, USA). Cyanuric chloride, 11-mercaptoundecanoic acid (MUA), 11-mercapto-1-undecanol (MUOH) and HAuCl₄ were obtained from Aldrich (Milwaukee, WI, USA). N-hydroxysuccinimide (NHS) was purchased from Acros (New Jersey, USA). NaOH, NaCl, CaCl₂, MnCl₂, Tween 20, absolute ethanol, ethanolamine (EOA) and tris-(hydroxylmethyl)-aminomethane (Tris) were of analytical reagent grade from Beijing Chemical Works (Beijing, China). All carbohydrates were of biochemical reagent grade from Beijing Reagent Work (Beijing, China). All reagent and solvents were used as received. Aqueous solutions were all prepared with pure water (>18.2 MΩ cm) from a Milli-Q academic system (Billerica, MA, USA).

Synthesis of GNP

GNPs were prepared following a modified procedure of two-phase system.²¹ HAuCl₄ aqueous solution (0.25 mM, 100 mL) was heated to boiling and sodium citrate solution (1 wt%, 1.8 mL) was added quickly under vigorous stirring. The solution was allowed to boil for an additional 5 min until the color of the solution became dark purple and finally light red. Tween 20 (0.1%, v/v) was then added to protect the formed GNPs by standing for overnight at room temperature. The obtained GNP protected by Tween 20(GNP–Tween 20)solution was used directly or stored at 4 °C before use.

Preparation of carbohydrate-functioned GNPs

A mixture of MUOH and MUA in ethanol solution (50 μL at 20 mM total thiol concentration at a molar ratio of 3 : 1 MUOH/MUA) was added into the GNPs–Tween 20 solution (4 mL) and reacted for 4 h at room temperature under vigorous stirring. The reacted product was centrifuged at 12 000 rpm for 20 min on an Eppendorf 5430R Centrifuge (Eppendorf Co. Ltd, German), and the precipitate of MUOH/MUA-modified GNPs was harvested and washed three times with 4 mL water at pH 7.0–8.0. After resuspension of the prepared GNPs in 1 mL water at pH 9.0, it was gradually added with 20 μL of 100 mM CC (in acetone) under vigorous stirring in an ice bath for 1.5 h. During reaction, the suspension was kept at pH 8.0–9.0 by replenishing 1 M NaOH solution when necessary. The reaction was further conducted by addition of 55 mM mannose or other carbohydrates and stirred at 25 °C for 8 h while keeping pH at ca. 9.0. After centrifugation at 12 000 rpm for 20 min, the precipitate was harvested, washed three times with water, and re-suspended in water under sonication. All the pH value of the solutions and suspensions was adjusted with 1 M NaOH.

For biological tests, the carbohydrate-functioned GNPs were blocked by mixing with 50 mM EDC aqueous solution and 25 mM NHS. After reaction for 20 minutes at room temperature, the precipitate was collected and re-suspended in 100 mM EOA solution at pH 8.6. After standing for 1 h, the blocked carbohydrate-functioned GNPs were washed three times with water and re-suspended in water.

Determination of immobilization efficiency of carbohydrates on GNPs

The immobilization efficiency of carbohydrates on GNP surfaces or the surface covering rate was measured with mannose as an example, based on specific reaction of carbohydrates with anthrone reagent, i.e. 0.1% anthrone in 98% concentrated sulfuric acid.²⁷ Briefly, an anthrone solution freshly prepared in concentrated H₂SO₄ (0.5 wt%, 1 mL) was added into 0.5 mL aqueous carbohydrate at 0.05, 0.1, 0.2, 0.3, 0.4, 0.5, 0.6 mM, respectively, under stirring in an ice bath. The solution was then heated to 100 °C and stirred for further 10 min. After cooled to room temperature, the resulting suspensions were subjected to UV-vis spectrophotometric determination at 620 nm. The quantity of immobilized carbohydrates on GNPs was calculated from the change of mannose concentrations before and after immobilization, with a calibration curve plotted by peak intensity vs. mannose concentration (Fig. S2 in ESI†).

Characterization of GNPs, carbohydrate-functioned GNPs and their lectin-recognition complex

The optical absorption spectra of GNPs, carbohydrate-functioned GNPs and their lectin-recognition complexes were measured using a TU-1900 UV-visible spectrometer (Beijing Purkinje General Instrument Co., Ltd, Beijing, China). The scanning electron microscopy (SEM) was measured using a model JSM-6701F SEM (JEOL, Tokyo, Japan).

For the recognition of the related GNPs with lectins, Con A or PNA solutions at 100 nM were prepared in Tris–HCl buffer (pH 7.6, 25 mM Tris, 1 mM CaCl₂, 1 mM MnCl₂, 0.1% Tween 20 (v/v)). After the Con A or PNA solution was introduced into the carbohydrate-functioned GNP solution and incubated for 30 min at room temperature, the color change was commonly observed by the naked eye and photographed with a Canon digital camera, model EOS 450D (Japan). In addition, optical absorption and SEM characterizations were also performed as above described.

Results and discussion

Chemistry to immobilize intact carbohydrates on GNPs

The proposed strategy aimed at immobilizing any intact or untreated carbohydrates on GNPs via the amazing CC linker as illustrated in Scheme 1. The immobilization chemistry includes only three key steps: creating hydroxyl-terminal on GNPs, anchoring CC on the hydroxyl terminal; and immobilizing the carbohydrates on top of CC layer. The successful introduction of each chemical was systematically followed by the change of UV-vis spectra (Fig. 1) and partially confirmed by SEM and our previous data. Fig. 1 shows that the first hydroxyl-termination on GNPs causes the resonance absorption shifting from 524 nm (GNP–Tween 20) to 526 nm (GNP–MUOH/MUA). Surprisingly, the anchoring of CC largely broadened the adsorption band and shifted the peak wavelength to even 536 nm. The possible reason lies in that the CC is in nature hydrophobic. This increases the aggregation tendency of GNPs–CC. This tendency was much prohibited by further immobilization of mannose on CC. This made the spectral band shrink and shift back the peak wavelength to 533 nm. The immobilization process of intact carbohydrates on GNPs was also characterized by SEM (Fig. S1 in ESI†), with (A), (B) and (C) corresponding to GNP–Tween 20, GNP–MUOH/MUA, and GNP–mannose respectively, which showed the dispersed state of GNPs and were in good agreement with the absorption spectra. In combination with the data of ¹H NMR used in the characterization of carbohydrates-functioned chips,²⁶ we can conclude the successful immobilization of carbohydrates on GNPs. This is ultimately confirmed by the recognition tests of the functioned-GNPs with related lectins (see below).

Fig. 1 LSPR absorption-based UV-vis spectra measured at the different stages of preparing mannose-functioned GNPs. a, GNPs; b, GNPs–Tween 20; c, GNPs–MUOH/MUA; d, GNsP–CC; e, GNPs–mannose.

Clearly, the immobilization starts from the first step to terminate the GNP surface with hydroxy ready for anchoring the CC. There is a matured and easy chemistry to directly hydroxylate the gold surface, that is, to use MUOH. The –SH group at the end of MUOH can self-assemble on gold, forming Au–S bond while leaving the hydroxyl at its other end pointing up for CC reaction.

The problem, however, is that the MUOH assembled GNPs aggregated seriously. To solve this problem, we added MUA together with MUOH to create static repelling force among GNPs by use of the charging carboxylic group on MUA. This was shown to work ideally in neutral or alkaline environments. The GNPs–MUOH/MUA did not have obvious broadened absorbance (Fig. 1), indicating that the GNPs–MUOH/MUA dispersed very well. Further inspection revealed that the stability of the GNPs–MUOH/MUA is closely related to the ratio of MUOH to MUA. Fig. 2 shows that the absorption spectra and the color depend very much on the molar ratio of MUOH to MUA from 1 : 1, 3 : 1, 5 : 1 to 8 : 1. The color varies from wine red to dark purple. When the ratio of MUOH/MUA equals to 5 : 1, the absorption wavelength did not appear obvious redshift which confirmed that the particle dispersed well; at the MUOH/MUA ratio of 5 : 1, a weak aggregation of GNP appeared; as the MUOH/MUA ratio reached 8 : 1, the aggregation became very apparent. In order to guarantee the stability of GNPs in later modifications, the ratio of MUOH to MUA is better kept below 5 : 1 and the ratio of 3 : 1 was recommended. The GNPs–MUOH/MUA (3 : 1) showed wine red color and were stable under alkaline conditions but they may still aggregate in acidic conditions.

Fig. 2 Absorption spectra and color changes of GNPs–MUOH/MUA at the MUOH-to-MUA molar ratio of 1 : 1, 3 : 1, 5 : 1 and 8 : 1.

On the hydroxyl-terminated GNPs, a layer of CC was easily formed at a low temperature (ca. 0–4 °C), followed by a layer of carbohydrates. Here, one-pot synthesis procedure was used to realize the immobilization of underivatized carbohydrates on GNPs. The three chlorines on CC could be stepwise replaced by hydroxyl or amino groups through the regulation of reaction temperature. The mild reaction conditions and simple experimental procedure help to maintain structural integrity of the sugar ring and keep the GNP stable in the reaction process which provides essential conditions for the conservation of carbohydrates bioactivity.

There were only limited examples focused on the reaction between CC and carbohydrates and to the best of our knowledge, no attempt has been made to use CC to directly immobilize intact carbohydrates on GNP surface. It should be noted that CC should first react with hydroxyl group rather than amino groups when immobilize hydroxyl compounds, otherwise the immobilization of carbohydrates will be failed.²²

Immobilization efficiency

Immobilizing efficiency was defined as a rate of the immobilized carbohydrate amount over the theory-calculated amount, which is equal to the surface-covering ratio on GNPs. The ratio was determinated by use of mannose as a testing sample. The real amount of carbohydrate immobilized on GNPs was measured by the anthrone-sulfuric acid colorimetric method,²⁷ which is a well-established assay for the quantitative analysis of carbohydrates. A calibration curve was obtained by plotting the absorption peak intensity (at 620 nm) against the mannose concentration (Fig. S2 in ESI†). The measured/theoretical ratio was 43.54% which is fairly high.

Recognition ability of the immobilized carbohydrates on GNPs

To investigate whether or not the immobilized carbohydrates can retain their recognition affinity, mannose-functioned GNPs were incubated with Con A. As known Con A is tetrameric, each monomer has one carbohydrate binding site specific to mannose. In order to avoid the interferences of carboxyl groups (on GNPs), metal ions (Ca²⁺/Mn²⁺, in the lectins buffer) and nonspecific adsorption, EOA (100 mM) was utilized as a blocking reagent which did not influence the stability of GNP suspensions. Fig. 3 shows that GNPs–mannose suspension changes its color to purple after addition of 100 nM (final concentration) Con A in Tris–HCl buffer (incubated for 30 min at room temperature) but keeps unchange of the color if only Con A-free Tris–HCl buffer is added. This indicates the existence of mannose's recognition ability on mannose-function GNPs. It is the recognition reaction that causes the aggregation of mannose-functioned GNPs and in turn the color variation through binding with Con A. After standing for 24 h, the mass aggregation or precipitation became very obvious. The recognition of GNPs–mannose with Con A also caused the maximum absorption of UV-vis spectra a significant redshift, for more than 30 nm. The results were also confirmed by SEM (Fig. S1(D)†). These results demonstrate not only the successful immobilization of carbohydrates on GNPs but more importantly the conservation of recognition ability of the on GNP-immobilized carbohydrates. This reveals a probability to establish a visual method to study the recognition of carbohydrates with the lectins by simply use of carbohydrate-functioned GNPs.

Fig. 3 Interaction of carbohydrate-functioned GNPs–mannose with 100 nM Con A illustrated by the UV-vis spectra. a, EOA–GNPs–mannose; b, EOA–GNPs–mannose with Tris–HCl buffer; c, EOA–GNPs–mannose with 100 nM Con A.

The competitive effect of the interaction was further checked by trehalose-functioned GNPs. Fig. 4A shows that the GNPs–trehalose, after incubated with 100 nM Con A for 30 min, have a significant redshift of absorption wavelength for 25 nm; and Fig. 4B illustrates that the competitive effect can easily be visualized by color variation and especially by precipitation. Compared with GNPs–trehalose (Fig. 4B-a), and GNPs–trehalose plus Tris–HCl buffer (Fig. 4B-b). GNPs–trehalose with 100 nM Con A (Fig. 4B-c) leads to significant precipitation and fading of the color due to the formation of GNPs–trehalose–Con A complex. The precipitation was effectively prohibited after addition of 200 mM glucose (Fig. 4B-d) because glucose can strongly compete with the GNPs–trehalose to deprive the Con A from the binding GNPs. In addition, the glucose was added excessively to enhance the competition. This observation further confirms the conservation of the recognition affinity of the immobilized carbohydrates, and we can concluded that the GNPs-carbohydrates are equivalent to the free states of carbohydrate molecules in solutions.

Fig. 4 (A) Interaction of carbohydrate-functioned GNPs–trehalose with 100 nM Con A followed by UV-vis spectra; (B) competitive effect of GNPs–trehalose and glucose. a, GNP–trehalose; b, GNP–trehalose with Tris–HCl buffer; c, GNP–trehalose with 100 nM Con A; d, GNP–trehalose with 100 nM Con A and 200 mM glucose. All recorded after 5 h.

Method applicability

The method was demonstrated to be capable of immobilizing all the intact carbohydrates (what we have) on GNPs, including monosaccharides of mannose, glucose, fructose and galactose (Fig. 5A), disaccharides of maltose, trehalose, sucrose, cellobiose and lactose (Fig. 5B), and larger molecules such dextran and cyclodextranes. The original recognition affinity of carbohydrates was well conserved on the carbohydrates-functioned GNPs as demonstrated by reaction with Con A or PNA. For instance, mannose-, glucose-, fructose-, maltose-, trehalose- and sucrose-functioned GNPs selectively combined with Con A but not PNA, leading to a significant redshift obviously depending on the type of saccharides; while the galactose- and lactose-functioned GNPs specifically formed complex with PNA but not Con A, also with significant redshift. Their recognition behavior showed no difference with carbohydrates in free solutions. They are hence applicable sensing method and have the advantages of color and precipitation variations for visualization.

Fig. 5 UV-vis spectra recorded from the interaction of 100 nM lectin (Con A or PNA) with (A) monosaccharide-, and (B) disaccharide-functioned GNPs. The experiments were repeated three times and the results were consistent and reproducible.

It is exciting that fructose showed an obvious recognition effect with Con A after its immobilization on GNPs (Fig. 5A), much the same as on a chip demonstrated in our previous work.²⁶ It is known that the fructose has so weak recognition ability with Con A that it is hardly observed even in free solution unless at high concentration. This suggests that the immobilization of it on GNPs could possibly has concentrating effect. This further reveals the advantage of this chemistry over others which are commonly unable to realize the recognition between fructose and Con A.

Conclusions

An innovative method was developed for the direct immobilization of various intact or untreated carbohydrates on gold nanoparticles. It minimizes the immobilization reaction steps and eases the reaction manipulation by taking the advantage of the temperature-regulable reaction of CC. By use of CC as a key linker, the immobilization efficiency tested by mannose could reach a fairly high level of 43.54%. The most exciting feature is that it can conserve the original recognition ability of carbohydrates on their functioned GNPs. The reason lies in that the CC offers three rotary σ-bonds or axes, able to make the immobilized carbohydrate molecules find their best spatial position to favor the recognition. In this case, even the very weak fructose can show its obvious recognition effect on its functioned GNPs towards Con A. By this method, the inherent aggregation of GNPs was successfully avoided by introduction of surficial charges with a recipe of 3 : 1 MUOH/MUA. The method is also applicable to the functionalization of GNPs with other hydroxyl substances and amines, other than carbohydrates. In short, it is more or less universal, easy in use, and cost-effective.

Acknowledgements

This work was supported by grants from the National Natural Science Foundation of China (Nos. 21235007 and 21135006) and Chinese Academy of Sciences.

References

1. A. Varki, Glycobiology, 1993, 3, 97–130.
2. G. E. Ritchie, B. E. Moffatt, R. B. Sim, B. P. Morgan, R. A. Dwek and P. M. Rudd, Chem. Rev., 2002, 102, 305–319.
3. C. Ding, S. Qian, Z. Wang and B. Qu, Anal. Biochem., 2011, 414, 84–87.
4. C. Earhart, N. R. Jana, N. Erathodiyil and J. Y. Ying, Langmuir, 2008, 24, 6215–6219.
5. C. C. Lin, Y. C. Yeh, C. Y. Yang, G. F. Chen, Y. C. Chen, Y. C. Wu and C. C. Chen, Chem. Commun., 2003, 2920–2921,  10.1039/b308995a.
6. M. J. Hernaiz, J. M. de la Fuente, A. G. Barrientos and S. Penades, Angew. Chem., Int. Ed., 2002, 41, 1554–1557.
7. E. Mahon, T. Aastrup and M. Barboiu, Chem. Commun., 2010, 46, 5491–5493.
8. Y. J. Chuang, X. Zhou, Z. Pan and C. Turchi, Biochem. Biophys. Res. Commun., 2009, 389, 22–27.
9. K. El-Boubbou, D. C. Zhu, C. Vasileiou, B. Borhan, D. Prosperi, W. Li and X. Huang, J. Am. Chem. Soc., 2010, 132, 4490–4499.
10. B. Nolting, J. J. Yu, G. Y. Liu, S. J. Cho, S. Kauzlarich and J. Gervay-Hague, Langmuir, 2003, 19, 6465–6473.
11. C. L. Schofield, B. Mukhopadhyay, S. M. Hardy, M. B. McDonnell, R. A. Field and D. A. Russell, Analyst, 2008, 133, 626–634.
12. M. C. Martos-Maldonado, M. B. Thygesen, K. J. Jensen and A. Vargas-Berenguel, Eur. J. Org. Chem., 2013, 2793–2801.
13. Y. K. Lyu, K. R. Lim, B. Y. Lee, K. S. Kim and W.-Y. Lee, Chem. Commun., 2008, 4771–4773,  10.1039/b807438k.
14. D. C. Kennedy, G. Orts-Gil, C.-H. Lai, L. Mller, A. Haase, A. Luch and P. H. Seeberger, J. Nanobiotechnol., 2014, 12.
15. A. L. Martin, B. Li and E. R. Gillies, J. Am. Chem. Soc., 2009, 131, 734–741.
16. K. Niikura, K. Nagakawa, N. Ohtake, T. Suzuki, Y. Matsuo, H. Sawa and K. Ijiro, Bioconjugate Chem., 2009, 20, 1848–1852.
17. Y. Sato, T. Murakami, K. Yoshioka and O. Niwa, Anal. Bioanal. Chem., 2008, 391, 2527–2532.
18. M. Alvarez-Paino, G. Marcelo, A. Munoz-Bonilla, J. Rodriguez-Hernandez and M. Fernandez-Garcia, Polym. Chem., 2013, 4, 986–995.
19. Z. l. Zhi, A. K. Powell and J. E. Turnbull, Anal. Chem., 2006, 78, 4786–4793.
20. M. R. Lee, D. W. Jung, D. Williams and I. J. Shin, Org. Lett., 2005, 7, 5477–5480.
21. X. Wang, O. Ramstrom and M. Yan, J. Mater. Chem., 2009, 19, 8944–8949.
22. J. R. Dudley, J. T. Thurston, F. C. Schaefer, D. Holmhansen, C. J. Hull and P. Adams, J. Am. Chem. Soc., 1951, 73, 2986–2990.
23. D. W. Kaiser, J. T. Thurston, J. R. Dudley, F. C. Schaefer, I. Hechenbleikner and D. Holmhansen, J. Am. Chem. Soc., 1951, 73, 2984–2986.
24. J. T. Thurston, J. R. Dudley, D. W. Kaiser, I. Hechenbleikner, F. C. Schaefer and D. Holmhansen, J. Am. Chem. Soc., 1951, 73, 2981–2983.
25. G. Blotny, Tetrahedron, 2006, 62, 9507–9522.
26. K. Liang and Y. Chen, Bioconjugate Chem., 2012, 23, 1300–1308.
27. A. Leyva, A. Quintana, M. Sanchez, E. N. Rodriguez, J. Cremata and J. C. Sanchez, Biologicals, 2008, 36, 134–141.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c6ra16511g

---