Redox-Active Biomolecular Architectures and Self-Assembled

Redox-Active Biomolecular Architectures and Self-Assembled Monolayers Based on a Cyclodecapeptide Regioselectively Addressable Functional Template...
1 downloads 0 Views 284KB Size
8134

Langmuir 2006, 22, 8134-8143

Redox-Active Biomolecular Architectures and Self-Assembled Monolayers Based on a Cyclodecapeptide Regioselectively Addressable Functional Template Charles H. Devillers,†,‡ Didier Boturyn,‡ Christophe Bucher,*,† Pascal Dumy,*,‡ Pierre Labbe´,† Jean-Claude Moutet,† Guy Royal,† and Eric Saint-Aman† Laboratoire d’Electrochimie Organique et de Photochimie Re´ dox, UMR CNRS 5630, Institut de Chimie Mole´ culaire de Grenoble, FR CNRS 2607, UniVersite´ Joseph Fourier, BP 53, 38041 Grenoble Ce´ dex 9, France, and Laboratoire d’Etudes Dynamiques et Structurales de la Se´ lectiVite´ , UMR CNRS 5616, Institut de Chimie Mole´ culaire de Grenoble, FR CNRS 2607, UniVersite´ Joseph Fourier Grenoble 1, BP 53, 38041 Grenoble Ce´ dex 9, France ReceiVed February 20, 2006. In Final Form: June 7, 2006 A nanometer scale redox active biomolecular architecture has been successfully synthesized through an efficient chemoselective oxime based coupling between ferrocenyl groups and a regioselectively addressable cyclodecapeptide. This molecular tool exhibits electronic, structural, and chemical properties driven by the biomimetic recognition activity of the polypeptide skeleton associated to the well-defined electrochemical activity of metallocenyl probes. Biomolecular materials obtained by confinement of the redox cyclopeptide in self-assembled monolayers on gold surfaces shows efficient through-bond electron transfer from the ferrocenes to the electrode surface via the peptidic backbone, as well as markedly improved sensing properties toward anionic species in organic electrolyte, as compared to those observed in homogeneous solution.

Most of the biological processes rely on the recognition properties of peptidic structures toward species with different sizes, geometries, charges, and functions. Numerous natural cyclopeptides are especially involved in the transport and regulation of ionic substrates in cells, owing to their ability to selectively bind and release positively or negatively charged molecular materials.1 The tremendous diversity of amino acids, building blocks of functional proteins with specific physicochemical properties (charge, polarity, hydrophobicity/hydrophilicity), makes supramolecular polypeptidic architectures very attractive to chemists working in the field of molecular recognition and sensing.2 The duality of amide groups, able to interact with substrates through various combinations of donating/accepting hydrogen or dative bonds, as observed in the so-called phosphate or sulfate binding proteins,3,4 naturally leads chemists to consider natural or artificial peptide based macromolecules as potential receptors for the recognition of a wide range of anionic substrates. Incorporating metal complexes in a biological assembly is thus of the greatest interest since it produces redox-active biomaterials which may act as sensing and/or switching devices.5-13 In this rapidly growing research area, amino acids

or peptide-ferrocene conjugates have recently been the subject of intense investigations.14,15 As a matter of fact, due to its unique structural and well behaved redox properties combined with an extensive chemical reactivity, the ferrocenyl group is undoubtly the most convenient building block for the construction of redoxresponsive bioconjugates15 and generally speaking to build selective receptors that can effectively convert a molecular recognition event into a physically measurable quantity, e.g., an electrochemical signal.16 Another crucial step toward electrochemical devices is the immobilization of redox chemosensors onto electrodes, which requires the introduction of suited functionality allowing electrode surface derivatization. Lysine-containing cyclodecapeptides called “regioselectively addressable functional templates” (RAFTs)17 were previously described as interesting molecular building blocks for the oriented presentation of recognition18 or complexing sites.19 Introducing redox active probes on a regioselectively addressable cyclodecapeptide RAFT scaffold especially opens up very interesting perspectives toward the molecular electrochemical recognition of negatively charged species. Moreover, RAFT molecules exhibit two spacially, well separated faces that can be selectively

* To whom correspondence should be addressed. (P.D.) Tel: (33) 04 76 51 46 85. Fax: (33) 04 76 51 43 82. E-mail: [email protected]. (C.B.) Tel: (33) 04 76 51 46 82. Fax: (33) 04 76 51 42 67. E-mail: [email protected]. † UMR CNRS 5630. ‡ UMR CNRS 5616.

(9) Geisser, B.; Alsfasser, R. Inorg. Chim. Acta 2003, 344, 102. (10) Plumb, K.; Kraatz, H.-B. Bioconjugate Chem. 2003, 14 (3), 601. (11) Huang, H.; Mu, L.; He, J.; Cheng, J.-P. J. Org. Chem. 2003, 68 (20), 7605. (12) Sheehy, M. J.; Gallagher, J. F.; Yamashita, M.; Ida, Y.; White-Colangelo, J.; Johnson, J.; Orlando, R.; Kenny, P. T. M. J. Organomet. Chem. 2004, 689 (9), 1511. (13) Prins, L. J.; Scrimin, P. Artificial (Pseudo)peptides for Molecular Recognition and Catalysis. In Functional Synthetic Receptors; Schrader, T., Hamilton, A. D., Eds.; Wiley-VCH Verlag GmbH & Co. KGaA: Weinheim, Germany, 2005; p 1. (14) Zatsepin, T. S.; Andreev, S. Y.; Hianik, T.; Oretskaya, T. S. Russ. Chem. ReV. 2003, 72 (6), 537. (15) Van Staveren, D. R.; Metzler-Nolte, N. Chem. ReV. 2004, 104 (12), 5931. (16) Beer, P. D.; Gale, P. A.; Chen, G. Z. Coord. Chem. ReV. 1999, 185-186, 3. (17) Dumy, P.; Eggleston, I.; Servigni, S.; Sila, U.; Sun, X.; Mutter, M. Tetrahedron Lett. 1995, 36 (8), 1255. (18) Renaudet, O.; Dumy, P. Org. Lett. 2003, 5 (3), 243. (19) Scheibler, L.; Dumy, P.; Stamou, D.; Duschl, C.; Vogel, H.; Mutter, M. Tetrahedron 1998, 54 (15), 3725.

(1) Ovchinnikov, Y. A.; Ivanov, V. T.; Shkrob, A. M. Membrane actiVe complexes; Elsevier: Amsterdam, 1974. (2) Vandenheuvel, D. J.; Kooyman, R. P. H.; Drijfhout, J. W.; Welling, G. W. Anal. Biochem. 1993, 215, 223. (3) Luecke, H.; Quiocho, F. A. Nature 1990, 347 (6291), 402. (4) Copley, R. R.; Barton, G. J. J. Mol. Biol. 1994, 242, 321. (5) Severin, K.; Bergs, R.; Beck, W. Angew. Chem., Int. Ed. Engl. 1998, 37 (12), 1635. (6) Schmitt, J. D.; Sansom, M. S. P.; Kerr, I. D.; Lunt, G. G.; Eisenthal, R. Biochemistry 1997, 36 (5), 1115. (7) Gallagher, J. F.; Kenny, P. T. M.; Sheehy, M. J. Inorg. Chem. Commun. 1999, 2 (8), 327. (8) Saweczko, P.; Kraatz, H.-B. Coord. Chem. ReV. 1999, 190-192, 185.

10.1021/la060491m CCC: $33.50 © 2006 American Chemical Society Published on Web 08/16/2006

Redox ActiVe Biomolecular Architecture

Langmuir, Vol. 22, No. 19, 2006 8135

addressed. The attachment of grafting sites, e.g., thioalkanes, to one face of the template thus offers strong opportunities to produce micro- and nanoscale devices through gold surface micropatterning with self-assembled monolayers (SAMs) of functional cyclopeptides.19-21 With the ultimate aim of producing redox-active peptide-based sensors, we report here the synthesis and the self-assembly on gold surfaces of a ferrocene-cyclopeptide conjugate. The redox properties of the ferrocenoyl cyclopeptide and the rates of electron transfer through the self-assembled monolayers have been investigated by voltamperometry. The anion chemosensing properties of these biomolecular architectures and biomaterials have also been investigated. Self-assembled monolayers have shown markedly improved sensing properties toward the dihydrogen phosphate anion in organic electrolyte, as compared to those observed in homogeneous solution. Experimental Section Electrochemical Measurements. Cyclic voltammetry experiments were performed using a conventional three electrodes system (660B model potentiostat, CH-Instruments). An automatic iR compensation was performed before each cyclic voltammetry experiment conducted in homogeneous media (solutions of 1). Electrode potentials were referred to an Ag|AgCl|NaCl (sat.) reference in aqueous media and toward an Ag|Ag+ (10-2 mol L-1) reference in hydro-organic media. The working glassy carbon (3 mm diameter) and gold (2 mm and 25 µm diameter) disk electrodes were purchased from CH-Instruments. Carbon electrodes were polished with 2 µm diamond paste before use. The cleaning procedure of the gold microelectrodes included successive polishing steps with alumina (3, 1, 0.3, and 0.05 µm), each step being followed by sonication in ultrapure (10 MΩ) water for 5 min. The gold surfaces were then activated by cycling at 10 V s-1 in aqueous 0.5 M sulfuric acid between 0.2 and 1.6 V until a reproducible voltammogram corresponding to a clean surface was obtained.22 This procedure was repeated before each experiment. Prior to the assembling procedure, two voltammograms were recorded at 0.1 V s-1 in 0.5 mol L-1 sulfuric acid between 0.2 and 1.5 V in order to check the surface state and to obtain roughness information. Assuming a charge of 420 µC cm-2 for the reduction of the gold oxide monolayer,22 a roughness factor between 2.2 and 2.7 was obtained for each gold electrode. Immediately after activation, the gold electrodes were modified by immersion in a 10-3 mol L-1 CH3CN/H2O (50:50 v/v) solution of 2 for 30 min followed by rinsing with the same solvent and with deionized water. Quartz Crystal Microbalance. Energy dissipation quartz crystal microbalance (QCM-D) experiments were performed on Q-Sense D 300 equipment (Q-Sense AB, Go¨teborg, Sweden) by monitoring simultaneously changes in the resonance frequencies (∆f) and energy dissipation factors (∆D) due to the assembling process. The QCM chip was excited to oscillate in the thickness-shear mode at its fundamental resonance frequency (f1 ) 5 MHz) and odd overtones (n ) 3, 5, and 7) by applying a radio frequency voltage across the electrodes. Measurements were performed by periodically disconnecting the oscillating crystal from the circuit in a computer controlled way and measuring the decay time τ0 of the exponentially damped sinusoidal voltage signal over the crystal caused by switching the voltage applied to the piezoelectric oscillator. The Q-Sense software then allows to calculate the dissipation factor, Dn, via relation (1) Dn )

1 2 ) πfnτ0 ωnτ0

(1)

(20) Scheibler, L.; Dumy, P.; Boncheva, M.; Leufgen, K.; Mathieu, H.-J.; Mutter, M.; Vogel, H. Angew. Chem., Int. Ed. Engl. 1999, 38 (5), 696. (21) Leufgen, K.; Mutter, M.; Vogel, H.; Szymczak, W. J. Am. Chem. Soc. 2003, 125 (29), 8911. (22) Vela, M. E.; Salvarezza, R. C.; Arvia, A. J. Electrochim. Acta 1990, 117, 35.

where fn is the resonance frequency at fundamental (n ) 1) and odd overtones (n ) 3, 5, 7), and τ0 is the relaxation time constant. These data give information on the adsorption process as well as on certain viscoelastic properties of the adsorbed film. In the case of homogeneous quasi-rigid films with reasonable thicknesses, the frequencies shifts are proportional to the ∆m mass uptake per unit area that can then be deduced from the Sauerbrey23 relationship -∆fSauerbrey )

1 m nC f

(2)

where the mass sensitivity C equals 17.7 ng cm-2 Hz-1 at f1 ) 5 MHz. To go beyond the Sauerbrey approximation, the ∆fn and Dn experimental data can be analyzed by using the framework developed by Voinova et al.24 on the assumption that the film is an homogeneous and isotropic viscoelastic layer. Before starting a QCM-D experiment, the gold surface was cleaned with UV-ozone during 10 min and then dipped for 10 min in ethanol. AFM imaging has shown that such a chemical treatment does not modify the morphology of the transducer surface and that surface roughness remains close to one.25 All solutions were previously degassed in order to avoid bubble formation in the QCM-D measuring chamber and experiments were conducted in exchange mode. Ellipsometry. Ellipsometric measurements were performed using an imaging ellipsometer EP3-SE from Nanofilm Technology GmbH, Germany. Experiments were conducted ex situ under air conditions at a wavelength of 630.2 nm and at variable angles of incidence ranging from 50° to 80°. Optical modeling was performed using the EP3View software from Nanofilm Technology GmbH. A threelayer model, substrate/layer/ambient air was used to fit the data. The optical properties of the bare gold substrate (a QCM-D gold coated quartz crystal) were previously measured. Materials. Protected amino acids, Sasrin, and ChloroTrityl resins were obtained from Advanced ChemTech Europe, Bachem Biochimie, and France Biochem. PyBOP was purchased from France Biochem, and other reagents were obtained from Aldrich or Acros. RP-HPLC was performed on Waters equipment equipped with a 600 controller and a Waters 2487 Dual Absorbance Detector. The purity of peptide derivatives was analyzed on an analytical column (Macherey-Nagel Nucleosil 120 Å 3 µm C18 particles, 30 × 4.6 mm) using the following solvents: solvent A, water containing 0.09% (v/v) TFA; solvent B, acetonitrile containing 0.09% (v/v) TFA and 9.91% (v/v) H2O; flow rate of 1.3 mL min-1 was employed and UV absorbance was monitored at 214 and 250 nm simultaneously. Preparative column (Delta-Pak 100 Å 15 µm C18 particles, 200 × 2.5 mm) was used to purify the crude peptides (when necessary) using the same solvent system at a flow rate of 22 mL min-1. ESI mass spectra were recorded on an Esquire 3000 (Bruker) spectrometer. The analyses were performed in the positive mode for peptide derivatives using 50% aqueous acetonitrile as solvent. (See the Supporting Information for HPLC profiles and mass spectra.) General Procedure for Solid-Phase Peptide Synthesis. Assembly of all protected peptides was carried out manually using the Fmoc/t-Bu strategy in a glass reaction vessel fitted with a sintered glass frit. Coupling reactions were performed using 1.5-2 equiv of N-R-Fmoc protected amino acid (relative to the resin loading) activated in situ with 1.5-2 equiv of PyBOP and 3-4 equiv of diisopropylethylamine (DIPEA) in DMF (10 mL‚g-1 resin) for 30 min. The coupling efficiency in manual synthesis was assessed by Kaiser and/or TNBS tests. N-R-Fmoc protecting groups were removed by treatment with a piperidine (20%)/DMF (80%) solution (v/v) for 10 min (10 mL g-1 resin). The process was repeated three times and the completeness of deprotection verified by UV absorption of the piperidine washings at 299 nm. Synthetic linear peptides were recovered directly upon acid cleavage (1% (v/v) TFA in CH2Cl2) from Sasrin and ChloroTrityl resins. The resins were treated for 3 (23) Sauerbrey, G. Z. Z. Phys. 1959, 155, 206. (24) Voinova, M. V.; Rodahl, M.; Jonson, M.; Kasemo, B. Phys. Scr. 1999, 59 (5), 391. (25) Mokrani, C.; Fatisson, J.; Guerente, L.; Labbe, P. Langmuir 2005, 21 (10), 4400.

8136 Langmuir, Vol. 22, No. 19, 2006 min repeatedly until the resin beads became dark purple. The combined washings were concentrated under reduced pressure, and white solid peptides were obtained by precipitation from ether. They were analyzed by RP-HPLC and if necessary further purified on a preparative column. General Procedure for Cyclization Reactions. All linear peptides (0.5 × 10-3 mol L-1, 1 molar equiv) were dissolved in DMF, and the pH values were adjusted to 8-9 by addition of DIPEA. PyBOP (1.2 molar equiv) was then added and the solution stirred at room temperature for 1 h.26 DMF was removed under reduced pressure and the residue dissolved in the minimum of CH2Cl2. Diethyl ether was added to precipitate the peptide which was triturated and washed with ether (×3) to obtain a crude material used in the next step without further purification. A detailed representation of the multistep synthesis of 2 is depicted in the Supporting Information. The following compounds were synthesized according to the procedures reported in ref 17: c[Lys(Boc)-Ala-Lys(Boc)-Pro-GlyLys(Boc)-Ala-Lys(Boc)-Pro-Gly] 3, c[Lys-Ala-Lys-Pro-Gly-LysAla-Lys-Pro-Gly] 7, c[Lys(-COCH2ONHBoc)-Ala-Lys(-COCH2ONHBoc)-Pro-Gly-Lys(-COCH 2 ONHBoc)-Ala-Lys(COCH2ONHBoc)-Pro-Gly] 8, c[Lys(-COCH2ONH2)-Ala-Lys(COCH 2 ONH 2 )-Pro-Gly-Lys(-COCH 2 ONH 2 )-Ala-Lys(COCH2ONH2)-Pro-Gly] 4. c[Lys(Boc)-Lys(Alloc)-Lys(Boc)-ProGly-Lys(Boc)-Ala-Lys(Boc)-Pro-Gly] 5 and c[Lys(Boc)-LysLys(Boc)-Pro-Gly-Lys(Boc)-Ala-Lys(Boc)-Pro-Gly] 9 were synthesized according to the procedures reported in ref 26. c[Lys(-COCH2ONHdCH-Fc)-Ala-Lys(-COCH2ONHdCHFc)-Pro-Gly-Lys(-COCH 2 ONHdCH-Fc)-Ala-Lys(COCH2ONHdCH-Fc)-Pro-Gly] 1. To a solution of 4 (131.9 mg, 0.105 mmol) in 20 mL of a solution containing 50% (v/v) acetate buffer (0.1 mol L-1, pH ) 4.6) in CH3CN was added ferrocene carboxaldehyde (137.7 mg, 0.643 mmol). The reaction mixture, protected from light, was stirred for 7 h at room temperature. The crude solid, obtained after evaporation of the solvents under reduced pressure, was purified by RP-HPLC and lyophilized to afford compound 1 as a yellow powder (100 mg, 49.0 µmol, 46%). Mass spectrum (ES-MS, positive mode) calcd. 2038.66, found 2040.5. c[Lys(Boc)-Lys(Fmoc-Cys(t-Buthio))-Lys(Boc)-Pro-Gly-Lys(Boc)-Ala-Lys(Boc)-Pro-Gly] 10. To a solution of 9 (154 mg, 0.108 mmol) dissolved in 15 mL of DMF, whose pH was adjusted to 8 by addition of DIPEA, was added Fmoc-Cysteine-tButhio (78 mg, 0.18 mmol) and PyBOP (94 mg, 0.18 mmol). After stirring the resulting solution at room temperature for 1 h, DMF was removed under reduced pressure. The crude solid material was dissolved in the minimum of CH2Cl2, precipitated with diethyl ether, washed three times with ether, and dried to afford 10 as a white powder (157 mg, 85 µmol, 79%). Mass spectrum (ES-MS, positive mode) calcd. 1833.0, found 1833.7. c[Lys-Lys(Fmoc-Cys(t-Buthio)-Lys-Pro-Gly-Lys-Ala-Lys-ProGly] 11. Compound 10 (157 mg, 85 µmol) was dissolved in 20 mL of a TFA/DCM mixture (50/50 v/v) and stirred for 1 h at room temperature. The crude product was concentrated under reduced pressure, washed with diethyl ether, and dried to yield compound 11 as a white powder (114 mg, 79 µmol, 93%). Mass spectrum (ES-MS, positive mode) calcd. 1432.8, found 1433.6. c[Lys(-COCH2ONHBoc)-Lys(Fmoc-Cys(t-Buthio)-Lys(COCH2ONHBoc)-Pro-Gly- Lys(-COCH2ONHBoc)-Ala-Lys(COCH2ONHBoc)-Pro-Gly] 12. To a solution of 11 (114 mg, 79 µmol) dissolved in 20 mL of DMF was added DIPEA to adjust the pH to 8. BocNHOCH2COSu27 (186.8 mg, 0.65 mmol) was then added to the mixture which was stirred for 1 h at room temperature. After removing the solvent under reduced pressure, the crude product was washed three times with diethyl ether and dried to yield 12 as a white powder (130 mg, 61 µmol, 77%). Mass spectrum (ES-MS, positive mode) calcd. 2125.0, found 2125.8. c[Lys(-COCH 2 ONHBoc)-Lys(Cys(t-Buthio))-Lys(COCH2ONHBoc)-Pro-Gly- Lys(-COCH2ONHBoc)-Ala-Lys((26) Boturyn, D.; Coll, J.-L.; Garanger, E.; Favrot, M.-C.; Dumy, P. J. Am. Chem. Soc. 2004, 126, 5730. (27) Ide, H.; Akamatsu, K.; Kimura, Y.; Michiue, K.; Makino, K.; Asaeda, A.; Takamori, Y.; Kubo, K. Biochemistry 1993, 32, 8276.

DeVillers et al. COCH2ONHBoc)-Pro-Gly] 13. Compound 12 (130 mg, 61 µmol) was dissolved in 20 mL of a 20% piperidine/80% DMF (v/v) solution and stirred for 30 min at room temperature. After removing the solvent under reduced pressure, the crude product was dissolved in the minimum of CH2Cl2 and precipitated with Et2O. The resulting precipitate was washed three times with diethyl ether and dried to afford 13 as a white powder (92 mg, 48 µmol, 79%). Mass spectrum (ES-MS, positive mode) calcd. 1903.0, found 1904.8. c[Lys(-COCH2ONH2)-Lys(Cys(t-Buthio))-Lys(COCH 2 ONH 2 )-Pro-Gly-Lys(-COCH 2 ONH 2 )-Ala-Lys(COCH2ONH2)-Pro-Gly] 6. Compound 13 (92 mg, 48 µmol) was dissolved in 20 mL of a solution containing 50% TFA/5% TIS/5% H2O (v/v) in CH2Cl2 and stirred for 1 h at room temperature. The crude product was concentrated under reduced pressure and washed with diethyl ether. The product was then purified by RP-HPLC to afford 3 as a white powder (51 mg, 34 µmol, 70%). Mass spectrum (ES-MS, positive mode) calcd. 1502.8, found 1503.6. c[Lys(-COCH2ONHdCH-Fc)-Lys(Cys(t-Buthio)-Lys(COCH2ONHdCH-Fc)-Pro-Gly-Lys(-COCH2ONHdCH-Fc)Ala-Lys(-COCH2ONHdCH-Fc)-Pro-Gly] 2. To a solution of 6 (39.9 mg, 26.5 µmol) in 5 mL of a solution containing 2.5 mL of an aqueous acetate buffer (0.1 mol L-1, pH ) 4.6) in CH3CN was added ferrocene carboxaldehyde (27.5 mg, 128.5 µmol). The reaction mixture, protected from light, was stirred for 7 h at room temperature. The crude solid, obtained after evaporation of the solvents under reduced pressure, was purified by RP-HPLC and lyophilized to afford 2 as a yellow powder (30 mg, 13.1 µmol, 49%). Mass spectrum (ES-MS, positive mode) calcd. 2286.8, found 2288.4.

Results and Discussion Synthesis. Cyclic polypeptide architectures known as “RAFT” (regioselectively addressable functionalized template) contain orthogonally protected attachment sites pointing toward both sides of a macromolecular scaffold.17,28,29 Such regioselectively addressable systems are thus ideally designed to readily and selectively introduce complementary functionalities on a peptidebased backbone through an oxime based ligation technique (A, Scheme 1).30 This highly efficient grafting procedure is compatible with a wide variety of chemical functions, without any coupling reagent and with minimal chemical manipulations. Owing to its great anion binding potential combined with its unique chemical reactivity, a RAFT cyclopeptide was selected as a suitable scaffold to regioselectively introduce redox active metallocene probes and a disulfur grafting group on the opposite faces of a peptide based template backbone (B and C, Scheme 1). The convenient choice of “cornerstone” protected amino acids and their relative positioning within the peptide primary sequence afforded the desired cyclodecapeptides exhibiting one (B) and two (C) attachments faces, respectively. Their linear precursor, containing side-chain protected lysines, H-Lys(Boc)-Lys(Alloc)Lys(Boc)-Pro-Gly-Lys(Boc)-Ala-Lys(Boc)-Pro-Gly-OH and HLys(Boc)-Ala-Lys(Boc)-Pro-Gly-Lys(Boc)-Ala-Lys(Boc)-ProGly-OH were obtained using a standard Fmoc/tBu solid-phase chemistry on an acid-labile Sasrin resin.17 The head to tail cyclizations of both linear peptides were performed in DMF under high dilution with PyBOP upon reacting the C-terminal glycine with the N-terminal lysine R-amino group to afford quantitatively the already known RAFT molecules 3 and 5. According to a known procedure, the aminoxy addressable cyclopeptide 4 could be obtained from the cyclic decapeptide 3 in three steps (Figure 1; overall yield 64%). Conversely, the attachment of a sulfur-based grafting group on the opposite face of the macrocyclic backbone required a (28) Dumy, P.; Eggleston, I. M.; Esposito, G.; Nicula, S.; Mutter, M. Biopolymer 1996, 39 (3), 297. (29) Peluso, S.; Ru¨ckle, T.; Lehmann, C.; Mutter, M.; Peggion, C.; Crisma, M. ChemBioChem 2001, 2, 432. (30) Shao, J.; Tam, J. P. J. Am. Chem. Soc. 1995, 117, 3893.

Redox ActiVe Biomolecular Architecture

Langmuir, Vol. 22, No. 19, 2006 8137

Figure 1. Preparation of Fc-containing peptides 1 and 2. Reagents and conditions: (a) 50% TFA/CH2Cl2 (v/v), 1 h; BocNHOCH2CO-NHS, DIPEA, DMF, 1 h; TFA/CH2Cl2/TIS/H2O (50/40/5/5) (v/v), 1 h; (b) Fc-CHO, acetate buffer (0.1 mol L-1, pH 4.6)/CH3CN (50/50) (v/v), 7 h; (c) Pd(PPh3)4, PhSiH3, CH2Cl2, 1 h; FmocCys(S-t-Bu)OH, PyBOP, DIPEA, DMF, 1 h; TFA/CH2Cl2 (50/50) (v/v), 1 h; BocNHOCH2CO-NHS, DIPEA, DMF, 1 h; DMF/piperidine (80/20) (v/v), 30 min; TFA/CH2Cl2/TIS/H2O (50/40/5/5) (v/v), 1 h; (d) Fc-CHO, acetate buffer (0.1 mol L-1, pH 4.6)/CH3CN (50/50) (v/v), 7 h. Scheme 1. Schematic Representation of Redox Active Regioselectively Addressable Functionalized Templates (RAFT) Exhibiting One (B) and Two (C) Attachments Faces

modification of the original strategy through the use of an Alloc protected lysine group which could be selectively deprotected to introduce a Cys(t-Buthio) residue. Removal of the Alloc group from peptide 5, carried out using the well-known Pd0/PhSiH3 procedure,31 afforded the desired derivatives containing a free amino group in excellent yields. Acylation of this amine with Fmoc-Cys(t-Buthio) was readily achieved using a PyBOP coupling strategy. The corresponding (Fmoc-Cys(t-Buthio)-RAFT(Boc)4 could be obtained, in sufficient purity to carry out the subsequent step, by thorough (31) Dessolin, M.; Guillerez, M.-G.; Thieriet, N.; Guibe, F.; Loffet, A. Tetrahedron Lett. 1995, 36 (32), 5741.

washing of the solid formed upon adding an excess of diethyl ether to the crude reaction mixture. This intermediate was readily deprotected with 50% (v/v) trifluoroacetic acid at room temperature in dichloromethane for 1 h, and acylated with succinimide ester of N-Boc-O-(carboxymethyl)-hydroxylamine to yield the functionalized templates (Fmoc-Cys(t-Buthio))-RAFT(-ONHBoc)4. The ultimate Fmoc cysteine protecting group was then smoothly removed using a 20% piperidine/80% DMF solution (v/v). Treatment of this ultimate intermediate with a mixture of 50% TFA, 5% triisopropylsilane (TIS), 5% water, and 40% CH2Cl2 (v/v) to remove the Boc groups followed by RP-HPLC purification provided the aimed (Cys(t-Buthio))RAFT(COCH2ONH2)4 6 in six steps (overall yield 28%).

8138 Langmuir, Vol. 22, No. 19, 2006

Figure 2. (A) Cyclic voltammetric response (ν ) 100 mV s-1) on a glassy carbon electrode (diameter 3 mm) of a 2.5 × 10-4 mol L-1 solution of 1 in 0.1 M TBAP/70% CH3CN/30% H2O (v/v) vs Ag/ Ag+ reference electrode. Dotted trace corresponds to simulated voltammogram obtained with Digisim by assuming four successive nernstian electron transfers with Em° ) 0.171 mV and E1°, E2°, E3°, and E4° respectively equal to Em° - (RT/F) ln 4, Em° - (RT/F) ln(3/2), Em° + (RT/F) ln 4, Em° + (RT/F) ln(3/2). (B) Plot of Ipa as a function of ν1/2.

Peptides 4 and 6, exhibiting four highly reactive aminooxy functions, provided the key intermediates to assemble, by chemoselective oxime formation, an array of ferrocene probes on a nanometer scale molecular peptide-based architectures. The aimed redox informative receptors 1 and 2 were thus easily obtained through oxime formations between an excess of ferrocene carboxaldehyde and the free aminooxy-containing RAFTs 4 and 6 in acetate buffer/acetonitrile (Figure 1). As inferred by HPLC, the ligation step proceeded cleanly and the reactions were generally complete after 7 h by using a slight 1.2-fold excess of aldehyde compounds relative to the aminooxycontaining template. The yields of the reactions were ranging from 46 to 49% after purification steps. Electrochemical Properties of RAFT-Ferrocene. The electrochemical properties of the RAFT-ferrocene 1 in solution was studied by cyclic voltammetry using a glassy carbon electrode (diameter 3 mm) in a 70% CH3CN/30% H2O (v/v) hydro organic mixture containing 0.1 mol L-1 TBAP as supporting electrolyte. These conditions turned out to be best suited to avoid adsorption phenomena onto electrode surfaces evidenced through ill-behaved voltametric curves. The electrochemical behavior of molecules with multiple redox centers has been the subject of numerous studies.32,33 It has been especially demonstrated34 that electron transfers to or from molecules containing identical, noninteracting, electroactive centers should yield a single current-potential curve similar to that observed with single electroactive center (∆Ep ) 58 mV at 25 °C) but with a magnitude determined by the total number of redox centers. When each center is characterized by the same standard potential Em° and adheres to the Nernst equation independently of the oxidation state of any of the other centers in the molecule, it is possible to calculate the formal potentials corresponding to each pair of successive oxidation states of the multi-centers molecules. The RAFT ferrocene conjugate 1 exhibits nanometer scale dimensions (Figure 4) which strongly suggests the absence of electrostatic-based interactions between each redox active centers, at least in highly solvating aqueous media disfavoring hydrogen-bond-based self-association processes. Theoretically, the successive ferrocene based formal oxidation potentials (E1°, E2°, E3°, and E4°) should respectively equal Em° - (RT/F) ln 4, Em° - (RT/F) ln(3/2), Em° + (RT/F) ln(3/2), and (32) Bard, A. J. Pure Appl. Chem. 1971, 25 (2), 379. (33) Ammar, F.; Saveant, J. M. J. Electroanal. Chem. 1973, 47 (1), 115. (34) Flanagan, J. B.; Margel, S.; Bard, A. J.; Anson, F. C. J. Am. Chem. Soc. 1978, 100 (13), 4248.

DeVillers et al.

Figure 3. QCM-D results recorded upon grafting 2 onto gold covered quartz crystal transducers. The frequency shifts are normalized toward the overtone number as ∆fn/n. n corresponds to the 5th and 7th overtones (that is 25 and 35 MHz) of the fundamental quartz crystal resonance frequency (5 MHz). The thiol-RAFT-ferrocene concentration is 10-3 mol L-1 in 50% CH3CN/50%H2O (v/v). The experiment was performed in exchange mode.

Em° + (RT/F) ln4.34 In agreement with previous work,34 we verified using digital simulation that four chemically equivalent, noninteracting, redox centers lead to a theoretical difference ∆Ep between anodic and cathodic peak potentials equal to 58 mV at 25 °C and that the peak current intensity equals four times that recorded for a single monoelectronic Nernstian system. The voltammetric behavior of RAFT-ferrocene at a concentration of 2.5 × 10-4 mol L-1 revealed one single wave (Figure 2) characterized by Epa ) 0.209 ( 0.005 V, Epc ) 0.134 ( 0.005 V (vs Ag/Ag+ reference electrode), and ∆Ep ) 0.075 ( 0.005 V. These potential characteristics remained unchanged for potential sweeping rates lower than 400 mV s-1, which indicates that under these conditions the system exhibits a Nernstian character. Moreover, the linear evolution of the anodic peak current intensity as a function of the sweeping rate (ν) square root is in full agreement with a diffusion controlled electrochemical response (Figure 2). The standard potential of ferrocene, Em°, could thus be estimated from the half-wave potential E1/2 ) (Epa + Epc)/2 ) 0.171 ( 0.005 V. A diffusion coefficient D1 ) 2.3 × 10-6 cm2 s-1 was assessed, assuming it is independent of the RAFT-ferrocene oxidation states and that the peak current intensity (Figure 2) corresponds to the successive transfer of four electrons. The experimental voltammogram was also compared to digitally simulated voltammograms (Figure 2) using the parameters previously determined. A satisfying agreement could be obtained, although a somewhat higher ∆Ep of 75 mV was observed in the experiment as compared to the theoretical value of 58 mV. Such a slight deviation from simple theory, already observed for molecules with multiple noninteracting sites,35 can be attributed to non fully compensated ohmic drop, slight interaction between adjacent redox centers, structural changes, or changes in diffusion coefficients upon adding or removing electrons from the molecules. Self-assembly of the disulfur-containing redox active RAFT 2 onto gold surfaces was characterized by QCM-D and cyclic voltammetry experiments. Figure 3 shows a typical grafting QCM-D experiment performed in 50% CH3CN/50% H2O (v/v). Frequency shifts have been normalized toward the overtone number as ∆fn/n. Perfect stabilization of both frequency and (35) Tarraga, A.; Molina, P.; Curiel, D.; Desamparados Velasco, M. Organometallics 2001, 20 (11), 2145.

Redox ActiVe Biomolecular Architecture

Langmuir, Vol. 22, No. 19, 2006 8139

Figure 4. Molecular modeling of 2 using InsightII-Discover 2000 (A) and its corresponding schematic representation (B).

dissipation signals could not be obtained in such media, and a continuous linear drift was observed. Acetonitrile is indeed known to slowly modify the mechanical properties of O-rings used to maintain the quartz transducer in the measuring chamber. Taking into account this linear drift, we could however analyze QCM-D experiments conducted with the overtones 5 and 7, known to be less sensitive to mechanical holding of the quartz transducer. Injection at t ) 1 min of 2 (2 mL, 10-3 mol L-1) led to a fast decrease of the resonance frequency during the first minute followed by a stabilization. Concomitantly, the energy dissipation remained well below 10-6 (not shown). Upon rinsing the quartz with pure solvent (at t ) 8 min), the resonance frequency reached a new value from which frequency shifts ∆fn/n of 5.3 and 5.9 Hz were respectively obtained for overtones 5 and 7. These results are thus in good agreement with a rapid formation of a grafted thiol-RAFT-ferrocene layer onto the gold surface. The low energy dissipation monitored upon adding 2 onto the gold surface is related to a weak increase in the crystal oscillation dampening usually observed in the case of rigid self-assembled monolayers. These experimental results indicate that the self-assembled biomolecular layer of 2 behaves as a thin rigid film, which in turn allowed us to apply the Sauerbrey equation to assess the ∆m mass uptake per unit area. An average surface mass of 99 ng cm-2 could be deduced for this biomolecular monolayer. A surface coverage of 4.33 × 10-11 mol cm-2 and molecular surface of about 384 Å2 could then be assessed assuming that the mass uptake was only due to the grafting of 2 (molecular weight 2288 g mol-1). This molecular coverage has to be compared with the size of the peptide-based architecture 2 assessed through molecular modelization. The cyclodecapeptide dimensions could be estimated around 15 Å × 6 Å when the distance between the ferrocene and disulfur groups reached up to 25 Å in a stretched conformation (Figure 4). On the other hand, the distance between two adjacent ferrocene groups could reach up to 30 Å. These estimations clearly suggest that the ferrocenyl arms conformation, more than the cyclodecapeptide scaffold dimension, control the surface occupied by a RAFT-ferrocene molecule grafted on the surface. In the case of a chemical adsorption leading to a parallel arrangement between the rigid cyclopeptide backbone and the gold surface, the coverage of one molecule should thus at least equal 225 Å2 (15 Å × 15

Å) which is fully compatible with the 384 Å2 value extracted from QCM-D. Ellipsometric measurements allowed us to definitively confirm the film thickness preliminarily assessed from QCM-D and molecular modeling experiments. Since the ellipsometry equations did not yield independent estimates of the refractive index and thickness, a refractive index of 1.45 was assumed for the SAM layer of 2 (the value of 1.45 is typically used for organic layer). Using a classical three-phase optical model, the fitting procedure led to a thickness of d ) 2.31 ( 0.16 nm and k ) 0.080 ( 0.047 (mse ) 1.15) (see ESI). It is noteworthy that the film thickness calculated from ellipsometric measurements well agreed with both molecular modeling predictions and QCM-D analysis. A nonzero extinction coefficient (k) had to be considered to account for the negative shift of ψ observed upon grafting 2 on the gold substrate. This negative shift can be explained by a non-negligible film absorption at λ ) 630.2 nm, the wavelength at which the ellipsometric measurements were recorded. Such weak absorbance was attributed to traces of ferricinium (that absorbs light at this wavelength) formed upon air-drying the ferrocenecontaining gold material. To further investigate the molecular organization of these nanostructured materials, cyclic voltammetry experiments were conducted with self-assembled monolayers (SAM) of 2 on gold microelectrodes (diameter 25 µm). All experiments were performed on carefully polished and cleaned microelectrodes as described in the Experimental Section. The electrodes roughness factor was systematically determined by cyclic voltammetry in aqueous sulfuric acid (0.5 mol L-1) as electrolyte in order to normalize the SAM coverage toward the effective gold surface. The exact geometric areas of these microelectrodes were furthermore confirmed using ferrocene-methanol as a diffusing nernstian redox probe in solution. Figure 5 shows a series of CV (cyclic voltammetry) curves recorded in 0.1 mol L-1 aqueous NaClO4 at increasing scanning rates ν for a gold microelectrode modified with 2. For values of ν < 1V s-1, the CV response exhibits symmetric cathodic and anodic waves typically observed for immobilized redox species. From the background-subtracted curve, the surface coverage and specific area occupied by one immobilized molecule could be estimated through signal integration (Coulombic charge)

8140 Langmuir, Vol. 22, No. 19, 2006

DeVillers et al.

Figure 5. (A) Cyclic voltammograms (vs Ag|AgCl) recorded in 0.1 mol L-1 aqueous NaClO4 on a gold microelectrode (diameter 25 µm) modified with a SAM of 2 at increasing scan rates from 0.1 to 5000 V s-1. (B) Plot of (Epa - E1/2) and (Epc - E1/2) as a function of log(ν) with E1/2 ) Epa - Epc calculated for ν ) 0.1 V s-1.

followed by roughness correction. Even after roughness correction, a somewhat higher surface coverage could be measured on microelectrodes as compared to the QCM-D results obtained on gold coated quartz crystals. This small difference could be attributed to the difference in gold surface morphology as a consequence of a different roughness that is about 1 for quartz crystals and around 2.5 for polycrystalline gold microelectrodes.

Εven at slow scan rates, ∆Ep is about 20 mV and not zero as expected for an immobilized redox species (Figure 5). Although we used microelectrodes and high ionic strength, such deviation from ideal behavior is often observed and tentatively attributed to non fully compensated ohmic drop. However, the shape and peak potentials (Figure 5) remained unchanged up to 1 V s-1. These electrochemical features are compatible with a Nernstian

Redox ActiVe Biomolecular Architecture

behavior (at low scan rates) with E1/2 ) 0.453 ( 0.005 V vs Ag|AgCl which logically compares with the half wave potential measured for 1 in homogeneous media. ∆Efwhm (peak width at half heigh) is about 110 mV, somewhat higher than the ideal of 90 mV. This can be a consequence of some disorder in the SAM structure and subsequent dispersity in the ferrocene local environment and standard potential E°. At higher scan rates (ν >1 V s-1), the anodic and cathodic peaks are progressively shifted respectively toward more positive and negative potentials (Figure 5), indicating that under these conditions the electrode response is governed by electron transfer kinetics. In parallel, it can be observed that, as voltammetric peak potentials move to larger overpotentials, the wave shapes become progressively broader and the peak current (normalized with ν) smaller. This behavior is particularly evident for the anodic wave. These effects have already been studied in detail in the case of ferrocene alkane thiol monolayer electrode kinetics.35 In particular, it has been shown that linear sweep voltammograms in which overpotential η approaches the Marcus reorganization energy λ, differ substantially from those following classical Butler-Volmer kinetics. In the case of Butler-Volmer kinetics36 faster sweep rates and smaller electron transfer kinetics simply cause the peak potentials to shift to higher overpotentials without affecting wave shapes and peak intensities.37 In contrast, when Marcus heterogeneous electron transfer kinetics apply, the peak currents, and consequently the wave shapes, do not scale linearly with sweep rate and are quite sensitive to the value of λ. In addition, calculation based on Marcus theory leads to slower electron-transfer kinetics than the Butler-Volmer theory.36 Voltammograms of Figure 5 also indicate an evident asymmetry since broadening of the anodic wave is more pronounced than the cathodic one, making impossible to measure anodic peak potentials at high ν. This observation suggests than the anodic and cathodic transfer coefficients Ra and Rc are different from 0.5. Since the Marcus theory has only been thoroughly developped with Ra ) Rc ) 0.5,36 we choose, in a first approximation, to apply the Butler-Volmer model to analyze the cathodic waves depicted in Figure 5 and estimate both k° (standard heterogeneous electron transfert rate constant) and Rc (cathodic transfer coefficient). Applying the Laviron model36 to the cathodic part of Figure 5 (only for overpotentials higher than 0.1 V) allowed us to extract k° ) 6350 ( 2000 s-1 and Rc ) 0.2 ( 0.05. Although k° may be overestimated by using Butler-Volmer kinetics, it is striking to observe that SAMs of 2 exhibit electron-transfer rates in the same order of magnitude as the ones recorded by Kraatz et al.38 for ferrocenyl-oligopeptides SAMs. Using oligopeptidic spacers of various length, it was shown38 that the heterogeneous electron transfer (ET) rate does not depend on the donor-acceptor distance. As measured by ellipsometry, ET rate constants remain comprised between 4300 and 7000 s-1, whereas the thickness of the ferrocenyl-oligopeptide SAMs increases from 5 to 11 Å.38 This lack of distance dependence could be explained by efficient through-bond electronic coupling of the ferrocene through the peptide spacer to the Au surface. In the case of the RAFTferrocene SAMs, the donor-acceptor distance between ferrocene and Au surface is expected from molecular modeling (Figure 4) to be about twice that recorded by Kraatz and co-workers.38 The fact that we observed an ET rate in the same magnitude strongly agrees with an efficient through-bond electronic coupling of ferrocene via the peptidic backbone. (36) Laviron, E. J. Electroanal. Chem. 1979, 101 (1), 19. (37) Tender, L.; Carter, M. T.; Murray, R. W. Anal. Chem. 1994, 66 (19), 3173. (38) Bediako-Amoa, I.; Sutherland, T. C.; Li, C.-Z.; Silerova, R.; Kraatz, H.-B. J. Phys. Chem. B 2004, 108 (2), 704.

Langmuir, Vol. 22, No. 19, 2006 8141

Figure 6. Redox dependent binding of an anion A by the RAFTferrocene receptor.

Electrochemical Anion Sensing Properties of the RAFTFerrocene Receptor 1 in Homogeneous Solution or SelfAssembled on a Gold Electrode. The ability of 1 to transduce a binding event into a measurable electrochemical signal was first investigated in homogeneous media. From a general point of view, redox and complexation equilibria for a ferrocene-based host in the presence of a given anion can be summarized in a simple square scheme (Figure 6), leading to the theoretical relationship (3)

∆E0 ) E0c - E0f ) (RT/F) ln(K/K+)

(3)

where E0c and E0f and K and K+ are the formal potential of the complexed and free receptor and the association constants of the complexes of the reduced and oxidized redox receptor, respectively.39,40 This relationship indicates that change in the electroactivity of the receptor upon complexation depends on the interactions balance between the targeted species and the reduced or oxidized state of the receptor. In simple terms, since an anion binds more strongly to the cationic ferrocenium form of a ferrocenyl receptor, it becomes easier to oxidize, and the potential of the ferrocene/ferrocenium system shifts negatively following its interaction with an anion. The more the interactions get reinforced through the receptor oxidation, the more the potential gets shifted negatively upon complexation. According to the respective values of K and K+, one can observe two different perturbations in the electrochemical response of the receptor: (i) a gradual potential shift of the voltammetric curve (one wave behavior) or (ii) the growth of a new, distinct voltammetric curve which develops at the expense of the original one following progressive addition of the guest species (two wave behavior).39 The electrochemical response of 1 in CH3CN turned out to be strongly affected in the presence of dihydrogen phosphate. When studied using cyclic voltammetry (CV), the addition of increasing amounts of these anionic species (n-tetrabutylammonium salt) to an acetonitrile solution of 1 led to the progressive decrease of the initial ferrocene-based wave at E1/2 ) 0.245 V, along with the growth at a less negative potential (Epa ∼ 0.1 V) of a new peak corresponding to the complexed receptor (Figure 7). These electrochemical features are similar to those already observed with other ferrocene containing redox-responsive receptors.41 This remarkable two-wave behavior turned out however to be complicated by adsorption phenomena, as evidenced by the growth on the reverse cathodic scan of sharp, intense reduction peaks attributed to the dissolution of oxidized ion pairs [1+A-] adsorbed on the electrode surface during the anodic scan. A clear and simple two-wave type behavior could be observed using differential pulse voltammetry (DPV) which allowed a quantitative analysis of the sensing properties of 1 toward H2PO4(39) Miller, S. R.; Gustowski, D. A.; Chen, Z. H.; Gokel, G. W.; Echegoyen, L.; Kaifer, A. E. Anal. Chem. 1988, 60 (19), 2021. (40) Beer, P. D.; Gale, P. A.; Chen, G. Z. J. Chem. Soc., Dalton Trans. 1999, 12, 1897. (41) Beer, P. D.; Hayes, E. J. Coord. Chem. ReV. 2003, 240 (1-2), 167.

8142 Langmuir, Vol. 22, No. 19, 2006

DeVillers et al.

Figure 7. Evolution of the CV waves (A) and of the DPV peak (B, scan rate 10 mV s-1) of 1 (5 × 10-4 mol L-1, CH3CN, 0.1 mol L-1 TBAP) as a function of added equivalents of TBAH2PO4: 0 (b), 2 (O), 2.4 (9), 2.7 (0), 3 (2), 3.4 (4), 3.7 (f). and 4 (g).

(Figure 7). The initial DPV peak at Ep ) 205 mV corresponding to the free ligand completely disappeared and the new corresponding to the complexed receptor peak (Ep ≈ 95 mV) reached a maximum intensity after adding 4 molar equiv of H2PO4-. This stoechiometry strongly suggested the ability of the electrogenerated tetra-ferrocenium receptor to host 4 equiv of anionic species to form an overall neutral complex. Adding an excess of the target anion induced a progressive shift of this new wave toward less positive potential, as well as a significant decrease in intensity (80% drop of the initial current was thus observed after adding 30 molar equivalents of anions). The observed ∆Ep value corresponds to the apparent association constant K+ between H2PO4- and the RAFT-ferrocene receptor 1 in its electrochemically oxidized ferrocenium form (Figure 6), which is 65 times larger than K related to the anion binding property of the neutral ferrocenyl receptor. The recognition behavior can thus be explained in terms of H-bonding interactions between the anion and the polypeptidic architecture, and the magnitude of the electrochemical sensing is essentially dependent on the additional ion-pairing interactions that develop between the anion and the ferrocenium moieties of the oxidized receptor. The electrochemical sensing properties of 1 toward the dihydrogen phosphate anion well compares with that already observed in acetonitrile with other neutral ferrocenyl receptors containing amide groups able to interact with anions through a combination of donating/accepting hydrogen or dative bonds.42,43 The use of pure DMSO as solvent not only suppressed the adsorption phenomena observed upon adding dihydrogen phosphate but also significantly lower the host-guest interactions and its corresponding redox signature. DMSO is known to form strong H bonds with H donors and thus interacts with 1 through the numerous amide functions of the molecule. As a consequence, the perturbation of the redox activity upon anion binding turned out to be reduced, with a shift that reached only -45 mV after the addition of 4 molar equiv of dihydrogen phosphate. Moreover, the host-guest interactions weakening were also evidenced by a “one-wave” progressive shift monitored throughout the titration, in contrast with the “two-wave” amperometric type of sensing observed in the less polar and competitive acetonitrile medium. It needs to be mentioned that the redox activity of the ferrocene groups was also significantly altered in DMSO as proven by a poor reversibility of the oxidation process. This effect, which proved to be even emphasized upon adding the targetted phosphate anions, can be reasonably attributed to poor stability of the ferrocenium species in such dipolar aprotic medium known to (42) Reynes, O.; Gulon, T.; Moutet, J.-C.; Royal, G.; Saint-Aman, E. J. Organomet. Chem. 2002, 656 (1-2), 116. (43) Reynes, O.; Maillard, F.; Moutet, J.-C.; Royal, G.; Saint-Aman, E.; Stanciu, G.; Dutasta, J.-P.; Gosse, I.; Mulatier, J.-C. J. Organomet. Chem. 2001, 637-639, 356.

enhance the anion nucleophilic character through an important solvation of the accompanying cation.44 Such effect was conversely not observed in pure water which strongly solvates neutral, anionic and cationic species through the existence of important H-bond networks established with both H-donor and H-acceptor molecules. The electrochemical signature of the ferrocene groups did however not significantly evolve upon adding increasing amounts of dihydrogen phosphate anions in accordance with a non favorable large solvation energy of both hosts and guests partners involved in the recognition process. The study and thorough exploitation of the electrochemical recognition properties of such macromolecular redox active polypeptide architecture were thus significantly hampered by solvent dependent adsorption or solvation processes. To overcome these important limitations inherent to homogeneous media, we turned our attention toward functional redox active materials. The result of adsorption of species onto surfaces followed by their spontaneous organization is a major research field with particularly promising applications in analytical chemistry. As detailed in the first part of the present article, a successful grafting of the thiol-appended RAFT ferrocene 2 on a gold electrode was readily achieved, and the self-organized redox-active layer exhibited a well-defined electrochemical activity characterized by symmetric cathodic and anodic waves shapes. Investigation by cyclic voltammetry of the electrochemical recognition properties of these modified electrodes toward dihydrogen phosphate anions was first conducted in acetonitrile (0.1 mol L-1 TBAP). A concentration in TBAH2PO4 above 5 × 10-5 mol L-1 was necessary to induce a noticeable perturbation of the original electroactivity evidenced by a well behaved “two-wave” electrochemical recognition process. Two waves of equal intensities corresponding to the free peptide and phosphate complexes were clearly observed as [TBAH2PO4] reached 1 × 10-4 mol L-1 (Figure 8B). However, the SAM electrodes submitted to high concentrations of phosphate presented a poor stability, as shown by the progressive loss of redox activity upon repetitive cycling. It should be emphasized that in the presence of H2PO4- the ferrocene-localized wave of the immobilized receptor is shifted to a more negative potential than that already observed in homogeneous solution. This point will be discussed below. The recognition properties in acetonitrile (0.1 mol L-1 TBAP) of the SAM-modified electrodes toward the dihydrogen phosphate anions were further assessed using differential pulsed voltammetry. The ferrocene-based electrochemical activity of the free peptide was observed at 0.25 V (vs Ag|Ag+) as a well defined peak. Changes in the redox signature upon adding increasing (44) Miller, J.; Parker, A. J. J. Am. Chem. Soc. 1961, 83, 117.

Redox ActiVe Biomolecular Architecture

Langmuir, Vol. 22, No. 19, 2006 8143

Figure 8. (A) Evolution of the DPV peak (scan rate 10 mV s-1) of SAM of 2 on gold electrodes as a function of [TBAH2PO4] in CH3CN + 0.1 mol L-1 TBAP. The overall exchanged charge for each electrode was corrected to obtain the same integral value calculated for the initial peak. (B) Cyclic voltammogram recorded in acetonitrile (0.1 mol L-1 TBAP) wherein [TBAH2PO4] ) 10-4 mol L-1.

amounts of the targetted anion were then monitored. To get reproducible and accurate results, sensing of phosphate was thus performed using a new, freshly prepared modified electrode for each concentration investigated. The electrochemical data recorded in phosphate-containing media were then corrected through the normalization of the overall exchanged charge to take into account the specific surface coverage Γ (mol cm-2) of each modified electrode. This procedure led to a clear two-wave like evolution with a progressive decrease of the initial ferrocenebased DPV peak at Ep ) 0.25 V, along with the growth at a less positive potential of a new peak corresponding to the complexed receptor (Figure 8). As proven by the observation of two equally intense DPV peaks (Figure 8A), only half of the adsorbed peptides were involved in phosphate recognition as [H2PO4-] reached 10-4 mol L-1. For a phosphate concentration of 5.4 × 10-4 mol L-1, a negative potential shift of around -200 mV could be monitored. This ∆Ep value corresponds to an apparent association constant K+ between H2PO4- and the ferrocenium form of the immobilized RAFT-ferrocene receptor that is 2000 times larger than in its neutral ferrocenyl form. Taking into account that the K+/K ratio was estimated around 65 for 1 in the homogeneous phase, the increase in the anion association constant observed upon confining the receptor to the electrode surface is thus remarkable. As already suggested for other devices synthesized from self-assembled layers of ferrocene-based anion receptors on gold surfaces45 or colloids46,47 steric strains and host surface preorganization effects could be responsible for this significant amplification of the redox sensing properties of the immobilized RAFT-ferrocene receptor.

Conclusion We successfully designed and synthesized a nanometer scale redox active biomolecular architecture through an efficient chemoselective oxime-based coupling between ferrocenyl groups and a regioselectively addressable cyclodecapeptide. This molecular tool exhibits electronic, structural, and chemical properties driven by the biomimetic recognition activity of the polypeptide skeleton associated with the well defined electrochemical activity of metallocenyl probes. Biomolecular materials on gold electrodes were obtained by confinement of redox cyclopeptides in a selfassembled monolayer. Marcus heterogeneous kinetics, implying significant reorganization energy consecutive to the electron (45) Beer, P. D.; Davis, J. J.; Drillsma-Milgrom, D. A.; Szemes, F. Chem. Commun. 2002, 16, 1716. (46) Labande, A.; Astruc, D. Chem. Commun. 2000, 12, 1007. (47) Labande, A.; Ruiz, J.; Astruc, D. J. Am. Chem. Soc. 2002, 124, 8, 1782.

Table 1. Electrochemical Properties of SAM of 2 Recorded on Au Microelectrodes (0.1 mol L-1 Aqueous NaClO4, ν ) 1 V s-1)a microelectrodes (diameter ∼25 µm)b Epa (V) Epc (V) E1/2 (V) ∆Ep ) Epa-Epc (V) ∆Efwhm,a (V) ∆Efwhm,c (V) Qa/τ (C) Qc/τ (C) Γa(RAFT) (mol cm-2) Γc(RAFT) (mol cm-2) σa(RAFT t) (Å2) σc(RAFT) (Å2)

0.462 ( 0.009 0.442 ( 0.003 0.453 (0.005 0.020 ( 0.006 0.112( 0.006 0.110 ( 0.006 1.18 × 10-10 ( 0.31 × 10-10 -1.27 × 10-10 ( 0.49 × 10-10 6.2 × 10-11 ( 1.6 × 10-11 6.7 × 10-11 ( 2.6 × 10-11 267 ( 67 248 ( 96

a E : anodic pic potential. E : cathodic pic potential. E Pa pc 1/2 half wave potential. ∆Efwhm,a and ∆Efwhm,c: peak width at half heigh. Qa/τ and Qc/τ: charge integrated under the anodic or cathodic wave and corrected by roughness factor. Γa(RAFT) and Γc(RAFT): surface concentrations of compound 2 in the SAM. σa(RAFT) and σa(RAFT): molecular surface area. b Standard deviations are calculated from four experiments.

transfer processes, had to be considered to account for the unusual CVs shapes observed at high sweeping rates. Kinetic studies strongly suggest an efficient through-bond electronic coupling of immobilized ferrocene groups via the peptidic backbone. Concentration of anion binding sites in nanostructured materials led to enhanced electrochemical recognition properties as proved by the amperometric type of sensing monitored in acetonitrile upon adding increasing amounts of dihydrogen phosphate. We believe that confinement of peptidic structures bearing electrophore moieties is a promising pathway towards in situ redox sensing of biological events. These results thus open up new exciting perspectives in the field of biomimetic electrochemical sensing and/or activation relying on the intimate association between protein like receptors and redox active reporters/initiators. Acknowledgment. We thank the “Institut des Nanosciences de Grenoble” (Idnano), the “Institut Universitaire de France” (IUF), and the innovation center for nanobiotechnologies (Nanobio) for support and the CECIC for giving us access to molecular modeling facilities. Supporting Information Available: Detailed representation of the multistep synthesis of 2, and HPLC profiles and mass spectra of compounds 1, 2, 6, and 10-13. This material is available free of charge via the Internet at http://pubs.acs.org. LA060491M