Striking Confinement Effect: AuCl 4 - Binding to Amines in a Nanocage Cavity Juan D. Henao, Young-Woong Suh, † Jeong-Kyu Lee, Mayfair C. Kung,* and Harold H. Kung* Department of Chemical and Biological Engineering, Northwestern UniVersity, 2145 Sheridan Road, E136, EVanston, Illinois 60208-3120 Received August 6, 2008; E-mail: m-kung@northwestern.edu; hkung@northwestern.edu It is well-known that the protonation constant of an amine and dissociation constant of a carboxylic acid are strongly influenced by the ability of their environment to stabilize charges. Thus, they are affected by the dielectric constant of the medium and presence of other charged or polarizable groups nearby. For example, the protonation constant of the second lysine in acetoacetate decar- boxylase is shifted from a pK a of 10 to ∼6 due to electrostatic repulsion between two charged ammonium ions and the low dielectric constant of the hydrophobic environment of a protein, 1 where solvation stabilization and charge screening are much weaker than those in an aqueous medium. This phenomenon of pK a shift is also observed in nonbiological systems. For example, a pK a shift of 1-2 pH units is observed for the amines in the interior of a G4-PAMAM dendrimer, 2 ∼1 unit in a vesicle, 3 1-2 units in a micelle, 4,5 and as much as 4-5 units at the air-water interface. 3 Likewise, a difference of ∼2 pH units between the protonation constants of propylamine and octylamine in an aqueous solution has been attributed to the agglomeration of octylamine consequent to its hydrophobic octyl group, thus affecting solvation stabilization of the ammonium ion. 6 These shifts are all attributed to a combination of effects of electrostatic repulsion and a hydrophobic medium. The magnitude of pK a shift should be concentration dependent, since factors such as electrostatic repulsion involves interaction between molecules. One consequence of this can be illustrated with a simplistic example of amines in a small water droplet. Neglecting the effect of ionic strength, protonation of one amine increases the pH of a 2 nm diameter water droplet to ∼13.5 and a 10 nm droplet to ∼11.5 (Supporting Information, Figure S1). Since the protonation constant of primary amines is ∼10, these values suggest that no more than one amine group could be protonated in a nm-size water droplet. Recently, we have synthesized siloxane and carbosilane nano- cages with 2-5 nm diameter cavities. 7,8 One of these nanocages was designed to possess 7-9 amines as aminopropyl groups tethered to the interior surface of a porous siloxane shell that forms a 2 nm cavity. 7 Because of the considerations mentioned above, we expect that, in a near neutral pH solution of these nanocages, no more than one of the amine groups in each cavity would be protonated. The remaining amine groups, being neutral, should then be available for interactions that utilize the electron lone pair on the N atom. We have examined the binding of AuCl 4 - complexes to these interior amines. AuCl 4 - can bind to an amine either by ligand exchange with a neutral amine that results in changing the Au-Cl coordination number (eq 1) or by forming an ion pair with a protonated amine without any change in the Au-Cl coordination number (eq 2). Since these different modes of binding are sensitive to the state of the amine, they can be used to probe whether its protonation is affected by the nanocage environment. -RNH 2 + AuCl 4 - )-RN(H) 2 AuCl 3 + Cl - (1) -RNH 3 + + AuCl 4 - )-RNH 3 + [AuCl 4 ] - (2) The binding of the AuCl 4 - complex was monitored spectroscopi- cally using a combination of EXAFS and UV-vis spectroscopy, assuming that the Au-N coordination can be accurately ap- proximated with the Pt-N coordination. This assumption is reasonable because of the similar electronic configuration of Pt and Au atoms. The Au-Cl bond is longer than the Au-O or Au-N bond, and a Cl atom scatters X-ray more strongly than O or N atoms. Thus, these bonds can be distinguished in EXAFS (Figure 1 and S4). The ligand-to-metal charge transfer band at 227 nm in the UV region is also sensitive to a Cl versus O atom (Figure S2). From the known hydrolysis constants for AuCl 4 - (Table S1) and data from control experiments using different solutions of known pH, Au(III) and Cl - concentrations, a quantitative correlation between the Au-Cl coordination number (CN), and the extinction coefficient at 227 nm in UV spectroscopy can be established (Figure S3). This correlation was shown to be applicable for both aqueous and methanol solutions. The latter was used to dissolve the nanocages. From this correlation, changes in the Au-Cl CN could be quantified. When the formation of a Au-O bond by hydrolysis was either absent or could be accounted for by control experiments, the Au-N coordination could be determined. In addition to a methanol solution of HAuCl 4 , another set of control experiments used solutions of 3-aminopropylmethylbis(trimethylsiloxy)silane (AS), which mimics the aminopropyl group in the nanocage. Changes in the Au-Cl CN in methanol solutions of HAuCl 4 at different H + concentrations were followed as a function of time using † Current address: Korea Institute of Science and Technology, South Korea. Figure 1. Magnitude of the Fourier transformed EXAFS (without phase correction) for 430 μM Au(III) in anhydrous methanol, Cl/Au ) 4: (A) no added amine, log[H + ] )-3.7; (B) in the presence of AS, NH 2 /Au ) 1, log[H + ] )-4.7; (C) in the presence of nanocage, NH 2 /Au ) 1, log[H + ] ) -4.0. The main peak in curve C is slightly skewed to the left compared with the others because of contribution from a small, shorter bond length component (Au-N). Published on Web 11/12/2008 10.1021/ja806179j CCC: $40.75  2008 American Chemical Society 16142 9 J. AM. CHEM. SOC. 2008, 130, 16142–16143