Gold Glyconanoparticles for Mimics and Measurement of Metal Ion

The essential involvement of the lactose moiety in Ca2+ complexation was shown by control measurements on related d-glucose-derivatized nanoparticles,...
1 downloads 8 Views 373KB Size
Download PDF
1156

Langmuir 2006, 22, 1156-1163

Gold Glyconanoparticles for Mimics and Measurement of Metal Ion-Mediated Carbohydrate-Carbohydrate Interactions Angela J. Reynolds, Alan H. Haines, and David A. Russell* School of Chemical Sciences and Pharmacy, UniVersity of East Anglia, Norwich, Norfolk, NR4 7TJ U.K. ReceiVed August 19, 2005. In Final Form: NoVember 14, 2005

To mimic and measure calcium ion-mediated carbohydrate-carbohydrate interactions, four lactose derivatives have been synthesized for assembly on gold nanoparticles. The series of lactose derivatives varied by the length of the thiolated ethylene glycol anchor chain [O(CH2CH2O)mCH2CH2SH; where m ) 0, 1, 2, and 3] used to self-assemble the carbohydrates to the preformed gold nanoparticles of ca. 16 nm diameter. Upon addition of calcium ions to the lactose-stabilized nanoparticles, rapid carbohydrate-carbohydrate interactions were visualized and subsequently measured using UV-visible spectrometry and transmission electron microscopy (TEM). The nanoparticle aggregates formed via metal-mediated carbohydrate-carbohydrate interactions could be readily redispersed through the addition of EDTA. Multiple reaggregation and redispersion cycles were achieved, confirming that the aggregation process was due to metal ion-mediated carbohydrate interactions rather than calcium chelation by residual citrate ions on the particle surface. The essential involvement of the lactose moiety in Ca2+ complexation was shown by control measurements on related D-glucose-derivatized nanoparticles, where a significantly reduced aggregation response was obtained only at high ion concentrations. Other group 2 metal ions with radii larger than that of calcium, viz., barium and strontium, were also shown to mediate the aggregation of the lactose-stabilized nanoparticles. The induced aggregation of the lactose nanoparticles was determined to be quantitatively dependent upon the calcium ion concentration. Furthermore, the analytical sensitivity of the calcium-induced aggregation and the linear dynamic range were dependent on the length of the ethylene glycol anchor chain. The shortest ethylene glycol chain (m ) 0) gave the most sensitive response with the optimum limit of detection (0.8 mM Ca2+), whereas the longest ethylene glycol chain (m ) 3) provides a measurement of calcium ion concentration over the largest linear dynamic range (10-35 mM Ca2+). This work has shown that the self-assembled deposition of lactose derivatives on gold nanoparticles provides multivalent carbohydrate surfaces that can be used as mimics for the measurement of biologically relevant carbohydrate-carbohydrate interactions. Additionally, this study has highlighted the importance of the structure and length of the ligand that anchors the carbohydrate sugar to the gold particle surface to facilitate such carbohydrate interactions and for “tuning” the analytical characteristics of bioassays developed using metal nanoparticle technology.

Introduction Calcium is ubiquitous within human cells and consequently of great physiological importance. Interactions between calcium ions and carbohydrates have been implicated in many physiological processes such as calcium transport, calcification, cellcell adhesion, and binding of glycoproteins to cell surfaces1 as well as cell recognition.2 Numerous studies have shown that carbohydrates bind to calcium ions, and other cations, when the sugars provide contiguous axial-equatorial-axial hydroxyl groups on six-membered rings or vicinal cis-cis-triol groups on five-membered rings.3,4 The size of the cation is also important in complex formation. Carbohydrate-metal complexes form only with cations of ionic radii larger than 0.8 Å, the optimum size being 1.0 Å.3,5 Thus calcium ions, with ionic radii of 0.99 Å, are ideal for complex formation. An obvious example of a carbohydrate-calcium complex is provided by lactose. Lactose in milk is well known to increase the rate at which calcium is absorbed from the human gastrointestinal tract.6 It is the formation of this metal-carbohydrate complex that is the focus of the present study. * To whom correspondence should be [email protected]. Tel/Fax: +44-1603-593012.

addressed.

E-mail:

(1) Bugg, C. E. J. Am. Chem. Soc. 1973, 95, 908-913. (2) Bucior, I.; Burger, M. M. Curr. Opin. Struct. Biol. 2004, 14, 631-637. (3) Angyal, S. J. Tetrahedron 1974, 30, 1695-1702. (4) Lu, Y.; Deng, G. C.; Miao, F. M.; Li, Z. M. J. Inorg. Biochem. 2003, 96, 487-492. (5) Angyal, S. J. Pure Appl. Chem. 1973, 35, 131-146. (6) Cochet, B.; Jung, A.; Friessen, M. Gastroenterology 1983, 84, 935.

Previously, we have studied carbohydrate-lectin interactions by depositing a thiolated derivative of a sugar on a planar gold surface or preformed gold nanoparticles and measuring the respective interaction via surface plasmon resonance7,8 or UVvisible spectroscopy using the intense surface plasmon absorption band characteristic of gold nanoparticles of ca. 16 nm diameter.8,9 The latter technique was first demonstrated by Mirkin et al. when they developed a colorimetric assay for the measurement of DNA hybridization.10,11 Gold nanoparticles exhibit a large surface plasmon absorption band centered at 520 nm. Thus, a solution of such particles appears red. A thiolated ligand is selfassembled to the gold nanoparticle surface and upon “recognition” of a target species aggregates the particles. The aggregation process brings the metal particles into closer proximity, and coupling interactions result in a red shift and broadening of the surface plasmon absorption band. A solution of the aggregated nanoparticles therefore appears blue. This approach has been used subsequently by numerous groups for the colorimetric detection of lectins,12-14 cations (potassium, lithium, and lead),15-18 and anti-protein A.19 (7) Revell, D. J.; Knight, J. R.; Blyth, D. J.; Haines, A. H.; Russell, D. A. Langmuir 1998, 14, 4517-4524. (8) Karamanska, R.; Mukhopadhyay, B.; Russell, D. A.; Field, R. A. Chem. Commun. 2005, 3334-3336. (9) Hone, D. C.; Haines, A. H.; Russell, D. A. Langmuir 2003, 19, 71417144. (10) Mirkin, C. A.; Letsinger, R. L.; Mucic, R. C.; Storhoff, J. J. Nature 1996, 382, 607-609. (11) Elghanian, R.; Storhoff, J. J.; Mucic, R. C.; Letsinger, R. L.; Mirkin, C. A. Science 1997, 277, 1078-1081.

10.1021/la052261y CCC: $33.50 © 2006 American Chemical Society Published on Web 01/06/2006

Gold Glyconanoparticles for Carbohydrate-Carbohydrate Interactions

Figure 1. Structures of lactose derivatives 12, 13, 14, and 17 (m ) 1, 2, 3, and 0, respectively) and control glucose derivative 11 used to stabilize gold nanoparticles.

On the basis of crystallographic data, two molecules of lactoses a disaccharide consisting of D-glucose and D-galactosesare required to chelate calcium ions.1 The chelation occurs via oxygen atoms of the HO(2)-HO(3) pair of hydroxyl groups of the glucose moiety of one lactose molecule and those of the HO(3)-HO(4) pair of hydroxyl groups of the galactose moiety of a second lactose molecule. The calcium ion is thus surrounded by a shell composed of eight oxygen atomssfour from lactose hydroxyl groups and four from water molecules.1 Our aim was to mimic in an aqueous medium this calcium-mediated interaction between the two separate lactose molecules and to measure such complexation using the colorimetric changes associated with the formation of aggregated gold nanoparticles. To achieve this aim, four lactose derivatives (Figure 1), each with a thiolated ethylene glycol anchor chain, were synthesized and deposited onto preformed gold nanoparticles. To confirm the importance of the disaccharide component in effecting efficient Ca2+ complexation, a related monosaccharide containing D-glucose was also investigated (Figure 1). In addition to the previously cited work studying carbohydratelectin interactions using gold nanoparticles, a number of other groups have deposited carbohydrates onto gold nanoparticles, albeit particles of ca. 2 nm diameter. An example of such work includes the investigation of carbohydrate-mediated self-recognition of marine sponge cells,20 and of direct relevance to this present study is that of Penade´s et al. who have studied calciummediated carbohydrate-carbohydrate recognition.21-23 The Penade´s group has self-assembled disulfide-modified carbohydrates on gold nanoparticles to study carbohydrate interactions, although because the particles were only 2 nm in diameter colorimetric changes upon aggregation would not have been (12) Otsuka, H.; Akiyama, Y.; Nagasaki, Y.; Kataoka, K. J. Am. Chem. Soc. 2001, 123, 8226-8230. (13) Lin, C. C.; Yeh, Y. C.; Yang, C. Y.; Chen, G. F.; Chen, Y. C.; Wu, Y. C.; Chen, C. C. Chem. Commun. 2003, 2920-2921. (14) Ishii, T.; Otsuka, H.; Kataoka, K.; Nagasaki, Y. Langmuir 2004, 20, 561-564. (15) Lin, S. Y.; Liu, S. W.; Lin, C. M.; Chen, C. H. Anal. Chem. 2002, 74, 330-335. (16) Obare, S. O.; Hollowell, R. E.; Murphy, C. J. Langmuir 2002, 18, 1040710410. (17) Kim, Y.; Johnson, R. C.; Hupp, J. T. Nano Lett. 2001, 1, 165-167. (18) Liu, J.; Lu, Y. J. Am. Chem. Soc. 2003, 125, 6642-6643. (19) Thanh, N. T. K.; Rosenzweig, Z. Anal. Chem. 2002, 74, 1624-1628. (20) Carvalho de Souza, A.; Halkes, K. M.; Meeldijk, J. D.; Verkleij, A. J.; Vliegenthart, J. F. G.; Kamerling, J. P. ChemBioChem 2005, 6, 828-831. (21) de la Fuente, J. M.; Barrientos, A. G.; Rojas, T. C.; Rojo, J.; Can˜ada, J.; Ferna´ndez, A.; Penade´s, S. Angew. Chem., Int. Ed. 2001, 40, 2258-2261. (22) Herna´iz, M. J.; de la Fuente, J. M.; Barrientos, A. G.; Penade´s, S. Angew. Chem., Int. Ed. 2002, 41, 1554-1557. (23) Barrientos, A. G.; de la Fuente, J. M.; Rojas, T. C.; Ferna´ndez, A.; Penade´s, S. Chem.sEur. J. 2003, 9, 1909-1921.

Langmuir, Vol. 22, No. 3, 2006 1157

visible. This group has reported the self-assembly of both the trisaccharide LewisX and indeed lactose onto gold nanoparticles. In the presence of calcium ions, the LewisX-functionalized particles exhibited calcium-mediated carbohydrate recognition as evidenced by the aggregation of particles visualized using TEM. However, no such particle aggregation was observed when 10 mM calcium was added to lactose-functionalized nanoparticles.21,23 With consideration of the known coordination of calcium between two lactose molecules and its considerable biological importance, the distinct absence of calcium-mediated lactose recognition is surprising. However, a recent report by this group has suggested that some lactose nanoparticle aggregation does indeed occur when calcium is added to the particles, albeit considerably less aggregation than that observed for the LewisX-functionalized particles.24 Here we report that lactose-functionalized gold nanoparticles can indeed exhibit calcium-mediated carbohydrate-carbohydrate recognition. This has been achieved through two important differences with that of previous reports: (i) The gold nanoparticles have been synthesized using the Turkevich method,25 which uses citrate as both a reductant and capping agent. Consequently, these particles are water-soluble, readily functionalized with carbohydrate ligands, and 16 nm in diameter, enabling the aggregation of particles to be visualized, both visually and spectrophotometrically. (ii) The thiolated ligand anchoring the carbohydrate to the gold surface is ethylene glycol-based rather than a methylene hydrocarbon chain. It is anticipated that the ethylene glycol-derived chain would allow water molecules to penetrate between the individual components of the selfassembled carbohydrate monolayer on the gold particle surface, facilitating calcium coordination between two lactose molecules (which on crystallographic evidence includes water molecules). As a consequence of our approach, we have been able to show that the aggregation of the lactose nanoparticles was quantitatively dependent on the calcium ion concentration. In a further element of this study, we have varied the length of the ethylene glycol anchor chain of the lactose monolayer to establish the direct relationship between the interparticle distance and the analytical sensitivity for the detection of the calcium ion concentration based on the surface plasmon absorption signal. Materials and Methods Reagents. All reagents were of analytical research grade and used without further purification. Aqueous solutions were prepared using a 10 mM tris stock solution at pH 7.6. Milli-Q water was used throughout. Synthesis of Carbohydrate-Derived Thiols. The carbohydrate glycosides containing a thiol functionality at the terminal position of the aglycone moiety were prepared in a multistep sequence from D-glucose and lactose, employing as the initial glycosidation step a boron trifluoride etherate-catalyzed reaction between an appropriate alcohol and the peracetate of either of the two carbohydrates. Thus, treatment of β-D-glucose pentaacetate with tetraethylene glycol (Scheme 1) afforded glycoside 3, containing a terminal hydroxy group in the aglycone portion that could be mesylated by treatment with methanesulfonyl chloride in pyridine. Displacement of the methanesulfonyloxy group was readily achieved with the thioacetate anion to yield thioacetate 7, the deacetylation of which on treatment with methoxide in methanol gave ω-mercapto glycoside 11. In a related fashion, starting with lactose octaacetate, diethylene glycol, triethylene glycol, and tetraethylene glycol, ω-mercapto glycosides 12, 13, and 14 were prepared. (24) de la Fuente, J. M.; Eaton, P.; Barrientos, A. G.; Mene´ndez, M.; Penade´s, S. J. Am. Chem. Soc. 2005, 127, 6192-6197. (25) Enu¨stu¨n, B. V.; Turkevich, J. J. Am. Chem. Soc. 1963, 85, 3317-3328.

1158 Langmuir, Vol. 22, No. 3, 2006

Reynolds et al. Scheme 1

Scheme 2

Although a similar sequence could be used to prepare ω-mercaptoethyl glycoside 17, separation of the initial glycosidation product 2-hydroxyethyl β-lactoside from the reaction of β-lactose octaacetate and ethylene glycol was not straightforward because of the similarity in its mobility on chromatography with the di-glycoside byproduct. A more satisfactory procedure was to prepare bromoethyl β-lactoside 15 (Scheme 2) on which the thioacetate displacement was then performed to give 16, which could then be deacetylated to afford 17. Aqueous solutions for use in the derivatization of the gold nanoparticles were made directly from the de-acetylated materials produced in the catalytic trans-esterification reactions. Glycoside Derivatives of Oligoethyleneglycols. (11-Hydroxy3,6,9-trioxaundecyl)2,3,4,6-tetra-O-acetyl-β-D-glucopyranoside 3. The following procedure gives the general method used for the synthesis of this series of compounds. To a stirred and cooled (0 °C) solution of 1,2,3,4,6-penta-O-acetyl-β-D-glucopyranoside,26 1 (1.3 g, 3.3 mmol), and tetraethyleneglycol (1.94 g, 10 mmol) in CH2Cl2 (7 mL) in a flask fitted with a drying tube and rubber septum was added boron trifluoride diethyl etherate (4.73 g, 4.2 mL, 33.3 mmol) from a syringe. The ice bath was removed, the reaction mixture was allowed to rise to ambient temperature, and after 3 h it was diluted with CH2Cl2 (25 mL) and washed successively with water (2 × 25 mL), NaHCO3(aq) (2 × 25 mL), water (2 × 25 mL) and then dried. TLC (ethyl acetate) indicated starting pentaacetate as the fastestrunning component and the required glycoside 3 as the slowestrunning component; with an intermediate Rf, this material was shown by 1H NMR spectroscopy to be the corresponding acetylated diglycoside of tetaethyleneglycol. Column chromatography initially with ethyl acetate as the eluant gave the di-glycoside derivative, and then elution with ethyl acetate-ethanol (9:1) gave, as a thick syrup, the title compound 3 (0.81 g, 46%). m/z (ES): 542.2 [M + NH4]+. Found (ES): [M + NH4]+ 542.2447. C22H40NO14 requires m/z 542.2443. (5-Hydroxy-3-oxapentyl)2,2′,3,3′,4′,6,6′-hepta-O-acetyl-β-lactoside 4. 1,2,2′,3,3′,4′,6,6′-octa-O-acetyl-β-lactoside27 2 (3.39 g, 5 mmol) and diethyleneglycol (1.59 g, 15 mmol) were dissolved in CH2Cl2 (10 mL), and to the stirred, cooled (0 °C) solution was added boron trifluoride diethyl etherate (4.4 mL, 4.93 g, 35 mmol). The cooling bath was removed, and the reaction was allowed to reach ambient temperature and then stored for another 5 h. Workup in a manner similar to that described for 3 gave a crude product (2.86 g) shown to contain by TLC (EtOAc-hexane, 2:1) some lactose octaacetate, a slightly slower-running component identified as the di-glycoside of the diol, and as the least mobile component the required mono-glycoside. Column chromatography in the same solvent system afforded the octaacetate and then the di-glycoside (0.17 g,), after which elution with ethyl acetate gave mono-glycoside (26) Wolfrom, M. L.; Thompson, A. In Methods in Carbohydrate Chemistry; Whistler, R. L., Wolfrom, M. L., BeMiller, J. N., Eds.; Academic Press: New York, 1963; Vol II, p 212. (27) Hudson, C. S.; Johnson, J. M. J. Am. Chem. Soc. 1915, 37, 1270-1275.

4 (1.39 g, 38%). m/z (ES): 742.2 [M + NH4]+. Found (ES): [M + NH4]+ 742.2762. C30H48NO20 requires m/z 742.2764. (8-Hydroxy-3,6-dioxaoctyl)2,2′,3,3′,4′,6,6′-hepta-O-acetyl-β-lactoside 5. The reaction of 1,2,2′,3,3′,4′,6,6′-octa-O-acetyl-β-lactoside27 2 (3.39 g, 5 mmol) and triethyleneglycol (2.25 g, 15 mmol) in CH2Cl2 with boron trifluoride diethyl etherate (4.4 mL, 4.93 g, 35 mmol) in the same manner as for the preparation of 3 and column chromatography with ethyl acetate as the eluent gave starting octaacetate, followed by the di-glycoside (0.51 g). Further elution with ethyl acetate-ethanol (9:1) gave the title compound 5 (1.47 g, 38%). m/z (ES): 786.4 [M + NH4]+. Found (ES): [M + NH4]+ 786.3024. C32H52NO21 requires m/z 786.3026. (11-Hydroxy-3,6,9-trioxaundecyl)2,2′,3,3′,4′,6,6′-hepta-O-acetylβ-lactoside 6. The reaction of 1,2,2′,3,3′,4′,6,6′-octa-O-acetyl-βlactoside27 2 (2.26 g, 3.3 mmol) and tetraethyleneglycol (1.94 g, 10 mmol) in CH2Cl2 with boron trifluoride diethyl etherate (3.31 g, 2.96 mL, 23.4 mmol) in the same manner as for the preparation of 3 and workup in the usual manner gave a crude product (2.41 g,) seen by TLC (ethyl acetate) to contain a trace of starting octaacetate, a slower-moving di-glycoside, and a slowly moving major component. On column chromatography and elution with ethyl acetate, the octaacetate and the di-glycoside were eluted first, and final elution with ethyl acetate-ethanol (9:1) gave title compound 6 (1.58 g, 58%). m/z (ES): 830.3 [M + NH4]+. Found (ES): [M + NH4]+ 830.3293. C34H56NO22 requires m/z 830.3288. Acetylthioglycoside Derivatives of D-Glucose and Lactose. (11Acetylthio-3,6,9-trioxaundecyl)2,3,4,6-tetra-O-acetyl-β-D-glucopyranoside 7. With the exception of compound 16 (see later), the following procedure represents a general method used for the synthesis of this series of compounds. To a solution of alcohol 3 (0.4 g, 0.76 mmol) in pyridine (5 mL) was added methanesulfonyl chloride (0.1 mL. 0.148 g, 1.3 mmol), and after 1 h, CH2Cl2 (25 mL) was added and the organic solution was washed sequentially with 1 M hydrochloric acid (100 mL), NaHCO3(aq) (50 mL), and water (50 mL) and then dried. TLC (EtOAc, two developments) indicated complete conversion of the starting alcohol into the faster-running methanesulfonate (0.42 g, 85%), which was used directly in the next step. A mixture of the methanesulfonate (0.38 g, 0.63 mmol) and potassium thioacetate (0.144 g, 1.26 mmol) in butanone (25 mL) was heated under reflux for 2 h, when TLC (EtOAc) showed that the starting material had been replaced with the slightly fasterrunning thioacetate. The solvent was removed under reduced pressure, and the residue was distributed betwen CH2Cl2 (25 mL) and water. The separated organic layer was dried then concentrated, and the product was subjected to column chromatography (EtOAc-hexane, 1:1) to give thioacetate 7 (0.26 g, 71%). m/z (CI): 600.3 [M + NH4]+. Found (ES): [M + NH4]+ 600.2320. C24H48NO14S requires m/z 600.2321. (5-Acetylthio-3-oxapentyl)2,2′,3,3′,4′,6,6′-hepta-O-acetyl-β-lactoside 8. Methanesulfonylation of (5-hydroxy-3-oxapentyl)2,2′,3,3′,4′,6,6′-hepta-O-acetyl-β-lactoside 4 (0.6 g, 0.83 mmol) in pyridine (5

Gold Glyconanoparticles for Carbohydrate-Carbohydrate Interactions mL) with methanesulfonyl chloride (0.1 mL, 0.148 g, 1.3 mmol), as described for the preparation of compound 7, afforded the chromatographically homogeneous methanesulfonate (0.64 g, 90%). A mixture of a solution of this ester (0.59 g) in butanone (25 mL) with potassium thioacetate (0.162 g, 1.4 mmol) was heated under reflux for 1 h, when TLC (EtOAc) indicated complete conversion into a faster-running component. Column chromatography elution initially with EtOAc-hexane (1:1) to remove the color and then with with EtOAc-hexane (2:1) gave title compound 8 (0.42 g, 76%). m/z (ES): 800.3 [M + NH4]+. Found (ES): [M + NH4]+ 800.2650. C32H50NO20S requires m/z 800.2641. (8-Acetylthio-3,6-dioxaoctyl)2,2′,3,3′,4′,6,6′-hepta-O-acetyl-βlactoside 9. (8-Hydroxy-3,6-dioxaoctyl)2,2′,3,3′,4′,6,6′-hepta-Oacetyl-β-lactoside 5 (0.5 g, 0.65 mmol) was reacted with methanesulfonyl chloride (0.1 mL, 0.148 g, 1.3 mmol) in pyridine (5 mL) in the manner described for the preparation of 7 to afford the methanesulfonate (0.51 g, 93%). A solution of the latter compound (0.44 g) in butanone (25 mL) was heated under reflux in the presence of potassium thioacetate (0.118 g, 1 mmol) for 1.5 h. Column chromatography of the crude product with EtOAc-hexane (1:1) to remove color and then with EtOAc-hexane (2:1) gave acetylthio compound 9 (0.33 g, 77%). m/z (CI): 844.4 [M + NH4]+. Found (ES): [M + NH4]+ 844.2908. C34H54NO21S requires m/z 844.2904. (11-Acetylthio-3,6,9-trioxaundecyl)2,2′,3,3′,4′,6,6′-hepta-O-acetylβ-lactoside 10. The reaction of (11-hydroxy-3,6,9-trioxaundecyl)2,2′,3,3′,4′,6,6′-hepta-O-acetyl-β-lactoside 6 (0.59 g, 0.73 mmol) with methanesulfonyl chloride (0.1 mL, 0.148 g, 1.3 mmol) in pyridine (5 mL) and isolation of the product, as described for the preparation of 7, gave the corresponding methanesulfonate (0.65 g), a portion (0.57 g) of which was then dissolved in butanone (25 mL) and the solution heated under reflux in the presence of potassium thioacetate (0.146 g, 1.3 mmol). Column chromatography of the crude product with EtOAc-hexane (1:1) and then EtOAc-hexane (2:1) gave title compound 10 (0.41 g, 72%). m/z (ES): 888.2 [M + NH4]+. Found (ES): [M + NH4]+ 888.3156. C36H58NO22S requires m/z 888.3166. (2-Acetylthioethyl)2,2′,3,3′,4′,6,6′-hepta-O-acetyl-β-lactoside 16. A solution of (2-bromoethyl)2,2′,3,3′,4,4′,6,6′-octa-O-acetyl-βlactoside28 15 (0.73 g, 0.98 mmol) in butanone (25 mL) containing potassium thioacetate (0.228 g, 2 mmol) was heated under reflux for 2 h when TLC indicated complete conversion to a new product. Workup and column chromatography in the manner described for 7 using EtOAc-hexane (1:1) as the eluent afforded, as a foam, thioacetate 16 (0.61 g, 84%). m/z (CI): 756.4 [M + NH4]+. Found (ES): [M + NH4]+ 756.2392. C30H46NO19S requires m/z 756.2379. De-acetylation of Acetylthioglycosides to Give Compounds 11-14 and 17. Dry methanol (20 mL) was added to an accurately weighed sample of the acetylthioglycoside (approximately 0.5 mmol), and a small amount of sodium (approximately 5 mg) was then added to form sodium methoxide in situ and bring about de-acetylation through trans-esterification. After 12 h, TLC indicated complete reaction, so the solvent was removed under reduced pressure and the residue was stored in a vacuum desiccator over P2O5 for 12 h. An aqueous solution of known strength of the de-acetylated product was prepared by adding to it a sufficient quantity of water accurately from a pipet in order to achieve a concentration of 0.85 M. Aliquots from this solution were used for the preparation of the functionalized gold nanoparticles for the calcium-binding experiments. Preparation of Citrate-Reduced Gold Nanoparticles. Watersoluble gold nanoparticles were prepared according to the method reported by Turkevich25 using sodium citrate as both a reducing and capping agent. Hydrogen tetrachloroaurate trihydrate (12.5 mg, HAuCl4) was dissolved in 100 mL of water to give a pale-yellow solution. Sodium citrate (50 mg) was dissolved in 50 mL of water. Both solutions were heated to 60 °C when the sodium citrate solution was rapidly added to the gold solution. The temperature was then increased to 85 °C with continuous stirring for 2 h. A deep-red solution was formed because of the intense surface plasmon absorption band of the gold nanoparticles. (28) Davis, B. G.; Maughan, M. A. T.; Green, M. P.; Ullman, A.; Jones, J. B. Tetrahedron: Asymmetry 2000, 11, 245-262.

Langmuir, Vol. 22, No. 3, 2006 1159

Figure 2. TEM images of the lactose-stabilized gold nanoparticles (a) as prepared using lactose derivative 14, (b) following addition of 10 mM Ca2+ ions, and (c) following subsequent addition of 10 mM EDTA. The scale bar represents 50 nm in each instance. Self-Assembly of Thiolated Lactose Derivatives on Gold Nanoparticles. Five milligrams of each thiol-derivatized carbohydrate was added to 10 mL of freshly prepared citrate-reduced gold nanoparticles. Self-assembly was facilitated by leaving the solution for 48 h. To remove any unbound carbohydrate, the solution was diluted to 100 mL with water, transferred into centrifuge tubes, and then centrifuged (Beckman Avanti J-25 centrifuge) at 23 750g for 25 min. The supernatant was removed, and the modified particles were resuspended in tris buffer. The centrifugation step was repeated another three times to ensure complete removal of any unbound carbohydrate. The solution was then resuspended back to the original

1160 Langmuir, Vol. 22, No. 3, 2006

Figure 3. UV-visible spectral changes observed following addition of increasing concentrations of calcium ions to the lactose-stabilized nanoparticles (derived from 14). 0 mM, s; 2 mM, - -; 4 mM, - ‚ -; 5 mM, - ‚ ‚ -; 7 mM, - -; 10 mM, - ‚ -; 20 mM, ‚‚‚.

Reynolds et al.

Figure 5. UV-visible spectra of citrate-capped gold nanoparticles (s); Ca2+ (10 mM)-induced aggregation of citrate-capped nanoparticles (- -); and citrate-capped nanoparticles following addition of Ca2+ and subsequent addition of 10 mM EDTA (‚ ‚).

Figure 6. Measurement of metal ion-mediated aggregation of lactose nanoparticles (derived from 14). The kinetics of aggregation were measured at 650 nm over 10 min for each metal ion. 10 mM BaCl2, - ‚ -; 10 mM SrCl2, ‚ ‚ ; 10 mM CaCl2, - -; 10 mM MgCl2, - ‚ ‚ -; 10 mM LaCl3, - ‚ -; 10 mM NaCl, ‚‚; 10 mM KCl, - -; 0 mM CaCl2, s.

Figure 4. Schematic representation of the formation of large aggregates following calcium-mediated carbohydrate-carbohydrate interaction of the lactose nanoparticles. It should be noted that one calcium ion is coordinated to a galactose moiety of a lactose nanoparticle and a glucose moiety on a second lactose nanoparticle. Water molecules that presumably are associated with each Ca2+ are omitted for clarity. volume of 10 mL and evaporated under vacuum until an absorbance maximum of ca. 0.8 was obtained. Instrumentation. 1H NMR spectra were recorded at 300 MHz on a Varian Gemini FT spectrometer or at 400 MHz on a Varian Unity Plus spectrometer in CDCl3 unless stated otherwise, with Me4Si as an internal standard. 13C NMR spectra were similarly recorded at 75 MHz on a Varian Gemini FT spectrometer. Coupling constants, J values, are given in Hz.29 Where appropriate, signal assignments were deduced by DEPT, COSY, and HSQC NMR experiments.29 Low- and high-resolution mass spectra were recorded by the EPSRC Mass Spectrometry Service Centre at the University College of Swansea. TLC was performed on silica gel (MacheryNagel) SIL G-25UV254, and compounds on developed plates were detected either by viewing with a UV lamp (254 nm) or by dipping into a 5% solution of sulfuric acid in ethanol followed by heating to 150 °C. Column chromatography was performed with Kieselgel 60 (70-230 mm mesh, Merck). Where mixed solvents were used, (29) See Supporting Information.

the ratios given are v/v. Organic solutions were dried over anhydrous Na2SO4. Reactions were maintained at 0 °C by means of an ice bath. All UV-visible spectra were recorded using a Hitachi U-3000 spectrophotometer. Transmission electron microscopy (TEM) measurements were performed using a JEOL 2000EX TEM using an accelerating voltage of 200 kV. Samples were prepared by drop casting 5 µL of the sample onto holey copper grids.

Results and Discussion Lactose-Stabilized Gold Nanoparticles and CalciumInduced Aggregation. In the first instance, the lactose derivative (14) with the longest ethylene glycol anchor chain was used to formulate lactose-stabilized nanoparticles. Following the 48 h period of self-assembly, the lactose displaced the citrate cap surrounding the surface of the gold nanoparticles. The TEM image (Figure 2a) shows the presence of dispersed nanoparticles. The average diameter of the nanoparticles was determined to be ca. 16 nm (standard deviation ) 1.5) on the basis of the measurement of a sample of 150 particles. The absorption spectrum of the citrate-capped gold nanoparticles exhibits an intense maximum at 520 nm, which is typical of the surface plasmon absorption band of gold particles of this diameter. Once the lactose-stabilized nanoparticles had been formulated, the surface plasmon absorption band maximum red-shifted to 523 nm (Figure 3). Aggregation of the lactose glyconanoparticles was mediated by the addition of calcium ions in the 2-20 mM concentration range. TEM and UV-visible spectroscopy both provide evidence of the aggregation process. Figure 2b shows

Gold Glyconanoparticles for Carbohydrate-Carbohydrate Interactions

Langmuir, Vol. 22, No. 3, 2006 1161

Figure 7. Effect of ethylene oxide anchor chain length on the sensitivity of calcium-mediated particle aggregation. (A-D) Detected response for lactose derivatives 12-14 and 17, respectively. (E) The four calibration curves for direct comparison: lactose derivative 17 (9), 12 (b), 13 (2), and 14 (().

the TEM image of the aggregates formed upon addition of 10 mM calcium ions to the lactose nanoparticles. Visibly, the nanoparticle solution changed color from red to purple following the addition of calcium ions. The changing absorption spectrum of the lactose nanoparticles, upon addition of increasing concentrations of Ca2+, is shown in Figure 3. It is apparent (from Figure 3) that the surface plasmon absorption band shifts substantially to the red with increasing Ca2+ concentration, a change characteristic of the formation of large aggregates of nanoparticles, 9 to provide a broad feature with a maximum at ca. 650 nm. Recently, Penade´s et al. reported that their lactose-coated nanoparticles produced only small aggregates, approximately 100 nm in diameter, upon addition of 10 mM calcium.24 Their lactose derivative was assembled onto ca. 2 nm gold particles

via a hydrophobic thiolated methylene anchor chain. The aggregated nanoparticles seen in Figure 2b are clearly much larger than 100 nm. Crystallographic evidence1 has determined that calcium binds in an octa-coordinate fashion surrounded by two lactose molecules and four water molecules. One lactose molecule is coordinated to the calcium ion through O(3) and O(4) of the galactose moiety whereas the second is coordinated through O(2) and O(3) of the glucose moiety. However, hydration of the binding site is equally important. It is thought that the ethylene glycol anchor chain of the lactose derivative used in this study provides a site that facilitates calcium hydration and consequent metal ion-mediated carbohydrate-carbohydrate interaction of the lactose nanoparticles. A schematic representation of the calcium-lactose nanoparticle aggregate is shown in Figure 4.

1162 Langmuir, Vol. 22, No. 3, 2006

Reversible and Selective Nature of Particle Aggregation. To establish whether the calcium-mediated lactose-lactose interaction is reversible, a 10 mM solution of EDTA was added to the aggregate. Immediately, the aggregated solution returned to the red color of the dispersed particles with the associated absorption spectrum also returning to the original spectrum of the lactose-derivatized particles. The TEM image in Figure 2c shows the resultant monodisperse lactose nanoparticles following the addition of EDTA. It was subsequently possible to cycle the aggregation/redispersion of the lactose nanoparticles multiple times following removal of the EDTA-Ca2+ complex via centrifugation on each occasion. Sodium citrate is also known to be a calcium ion chelating agent. Because citrate is used in the formation of the gold nanoparticles, a control experiment was performed to ensure that the aggregation observed was not due to any residual citrate on the surface of the lactose nanoparticles. Calcium ions were added to a freshly prepared sample of citrate-capped nanoparticles. Aggregation of the particles occurred, although this aggregation proved to be nonreversible upon addition of EDTA (Figure 5). It is likely that the calcium-induced aggregation of the citratecapped particles was due to charge destabilization of the negative citrate ions on the gold surface. Another control experiment was performed to establish whether the ethylene glycol linker chain was significantly involved in the calcium ion-mediated complexation. A D-glucose derivative, 11, (Figure 1) was synthetically produced for this control. Upon addition of 10 mM calcium ions to a solution of the glucosestabilized nanoparticles, no detectable absorbance change at 650 nm was observed (data not shown). Only at significantly higher concentrations, viz. 30-100 mM, did any aggregation of the glucose particles occur, suggesting that the linker had minimal influence on the metal ion complexation between lactose molecules. As previously noted, the size of the cation used to form a complex with lactose is important. It should be anticipated that ions with radii larger than that of calcium (0.99 Å)3,5 will also mediate complex formation between lactose nanoparticles. Other metal ions were added to the lactose-stabilized nanoparticles. The kinetics of absorbance intensity changes of lactose nanoparticles were measured at 650 nm (the surface plasmon absorption band maximum following aggregation)9 for 10 min following the addition of each metal ion. It can be seen in Figure 6 that other group 2 metal ions also induce aggregation of the lactose nanoparticles. Although magnesium has a minimal effect, enhanced aggregation occurs with strontium and barium. Indeed, an increased response in aggregation occurred as the ionic radii increased (i.e., Mg < Ca < Sr < Ba). It should be noted, however, that ions such as barium and strontium have limited biological relevance. No response was observed upon addition of sodium, potassium, or lanthanum ions. Effect of Ethylene Glycol Anchor Chain Length. Mirkin et al. have previously shown that changes in the optical properties of DNA-modified nanoparticles are dependent upon the spacer length of the oligonucleotide, which affects both the interparticle distance and the number of particles within an aggregate structure.30 Additionally, it was established that different spacer lengths provide kinetic control over aggregate size and hence the optical properties.30 We were interested to establish whether the aggregation of lactose nanoparticles was quantitatively dependent on the metal ion concentration and whether such a concentration dependence was variable on the basis of the chain length of the (30) Storhoff, J. J.; Lazarides, A. A.; Mucic, R. C.; Mirkin, C. A.; Letsinger, R. L.; Schatz, G. C. J. Am. Chem. Soc. 2000, 122, 4640-4650.

Reynolds et al.

Figure 8. TEM image of aggregated lactose nanoparticles formulated using lactose derivative 17. Aggregation was induced through addition of 1.4 mM calcium ions. The scale bar represents 50 nm.

ethylene glycol-derived spacer. For this study, three other lactose derivatives were synthesizeds12, 13, and 17 (Figure 1). The aggregation of each lactose-nanoparticle system was mediated through the addition of varying concentrations of calcium ions. For each lactose derivative, aggregation of the nanoparticles was induced, although the concentration of calcium used to achieve full aggregation was significantly different in each instance. Calibration data are shown in Figure 7 for each of the lactose derivatives used to stabilize the gold nanoparticles. The curves were obtained through the measurement of the change in absorbance intensity 30 s after addition of the cation as a function of Ca2+ concentration.9 There was a quantitative relationship between the concentration of calcium ions and the aggregation of the particles for each lactose derivative (Figure 7A-D). However, as is evident from Figure 7, the length of the ethylene glycol anchor chain has a profound effect on the intensity of the changing absorbance signal and the concentration range of calcium ions that initiate the aggregation process (Figure 7E). It is clear that the rate of change of the surface plasmon absorption band is inversely dependent upon the length of the linker and that the lactose derivative with the shortest chain length exhibits the largest absorbance change at the fastest rate. In terms of bioassay development, the largest change in signal, with the greatest sensitivity and lowest limit of detection (0.8 mM Ca2+), was obtained from the lactose derivative with the shortest ethylene glycol anchor chain (17). However, the derivative with the longest ethylene glycol anchor chain (14) provided a measurement of calcium ion concentration with the largest linear dynamic range (10-35 mM Ca2+). The intensity of the absorbance signal is directly related to both the proximity of the particles following aggregation and the density of the nanoparticle aggregate. The TEM image of nanoparticles stabilized with the short-chain lactose (17) derivative clearly shows a dense, highly extended aggregate following the addition of 1.4 mM calcium ions (Figure 8). This image is in contrast to the image of the “nondense” aggregate obtained for the long-chain lactose (14) derivative (using 10 mM Ca2+) shown in Figure 2b and accounts for the differences in the absorbance intensity seen in Figure 7. The TEM images and absorbance intensity calibration data obtained for derivatives 12 and 13 are intermediate between those of 17 and 14. The data shown in Figure 7 confirm that the aggregation of glyconano-

Gold Glyconanoparticles for Carbohydrate-Carbohydrate Interactions

particles can provide a quantitative bioassay for the colorimetric determination of biologically relevant metal ions such as calcium, the sensitivity and dynamic working range of which can be tuned through variation of the ethylene glycol anchor chain length.

Conclusions We have shown that biologically relevant, metal ion-mediated carbohydrate-carbohydrate interactions can be mimicked using gold glyconanoparticles. To achieve such calcium-induced carbohydrate interactions between lactose molecules, the presence of water molecules is essential. It has been shown that the use of an ethylene glycol chain to anchor lactose derivatives to gold nanoparticles provides a sufficiently hydrophilic site to enable calcium ions not only to chelate with a glucose moiety from one lactose molecule and a galactose moiety of a second lactose molecule but also to allow access of the requisite water molecules. The aggregation of lactose-derivatized gold nanoparticles has been measured via the red shift of the surface plasmon absorption band. The aggregation of lactose nanoparticles was found to be quantitatively dependent on calcium ion concentration. In a study of the anchoring ligand, the analytical sensitivity and working

Langmuir, Vol. 22, No. 3, 2006 1163

dynamic range of the calcium-induced aggregation was shown to be dependent upon the length of the ethylene glycol chain. This work has clearly highlighted the fact that for the creation of glyco-mimics, or for the development of metal nanoparticlebased bioassays in general, the construction of the ligand used to anchor the molecular recognition component to the gold particle surface is of fundamental importance. Acknowledgment. This work was partially supported by the EPSRC (grant no. GR/S64134/01). We thank Rick Evans-Gowing for technical assistance with the TEM measurements and Dr. Balaram Mukhopadhyay for his expert recording of the NMR spectra. The determination of the low- and high-resolution mass spectra was achieved by the EPSRC Mass Spectrometry Service Centre at Swansea. The assistance of Claire Schofield with the production of the figures is gratefully acknowledged. Supporting Information Available: Chemical shift data and coupling constants. This material is available free of charge via the Internet at http://pubs.acs.org. LA052261Y