1930

Bioconjugate Chem. 2009, 20, 19301939

Protein Binding and the Electronic Properties of Iron(II) Complexes: An Electrochemical and Optical Investigation of Outer Sphere Effects
Kylie D. Barker, Amanda L. Eckermann, Matthew H. Sazinsky,, Matthew R. Hartings, Carnie Abajian, Dimitra Georganopoulou, Mark A. Ratner, Amy C. Rosenzweig,, and Thomas J. Meade*,,,|
Departments of Chemistry, Biochemistry and Molecular and Cell Biology, Neurobiology and Physiology, and Radiology, Northwestern University, 2145 Sheridan Road, Evanston, Illinois 60208. Received June 17, 2009; Revised Manuscript Received
September 1, 2009

Metalloenzymes and electron transfer proteins inuence the electrochemical properties of metal cofactors by controlling the second-sphere environment of the protein active site. Properties that tune this environment include the dielectric constant, templated charge structure, van der Waals interactions, and hydrogen bonds. By systematically varying the binding of a redox-active ligand with a protein, we can evaluate how these noncovalent interactions alter the electronic structure of the bound metal complex. For this study, we employ the well-characterized avidin-biotin conjugate as the protein-ligand system, and have synthesized solvatochromic biotinylated and desthiobiotinylated iron(II) bipyridine tetracyano complexes ([Fe(BMB)(CN)4]2- (1) and [Fe(DMB)(CN)4]2- (2)). The binding afnities of 1 and 2 with avidin are 3.5 107 M-1 and 1.5 106 M-1, respectively. The redox potentials of 1 and 2 (333 mV and 330 mV) shift to 193 mV and 203 mV vs Ag/AgCl when the complex is bound to avidin and adsorbed to a monolayer-coated gold electrode. Upon binding to avidin, the MLCT1 band red-shifts 20 nm for 1 and 10 nm for 2. Similarly, the MLCT2 band for 1 red-shifts 7 nm and the band for 2 red-shifts 6 nm. For comparison, the electronic properties of 1 and 2 were investigated in organic solvents, and similar shifts in the MLCT bands and redox potentials were observed. An X-ray crystal structure of 1 bound to avidin was obtained, and molecular dynamics simulations were performed to analyze the protein environment of the protein-bound transition metal complexes. Our studies demonstrate that changes in the binding afnity of a ligand-receptor pair inuence the outer-sphere coordination of the ligand, which in turn affects the electronic properties of the bound complex.

INTRODUCTION
The properties of coordinatively saturated transition metal complexescanbeprofoundlyinuencedbytheirenvironment(1-7). The ligands of a transition metal complex can interact with surrounding solvent molecules, macromolecular species, or proteins to affect the electronic state of the metal complex. These noncovalent interactions (known as outer-sphere or secondsphere coordination) include hydrogen bonding, - stacking, electrostatic, hydrophobic, and van der Waas interactions (8-11). The concept of outer-sphere coordination was introduced by Werner to explain phenomena such as solvents of crystallization and the formation of adducts between amines and coordinatively saturated complexes (12, 13). In 1981, Stoddart and co-workers presented the rst crystallographic evidence of this type of interaction, in which a crown ether is associated with a Pt-ammine complex through noncovalent interactions (14). In recent years, it has been shown that interactions with the second sphere of coordination complexes affect the physical properties. For example, interaction of 18-crown-6-ether with the second sphere of a series of ruthenium-ammine complexes results in changes in the spectroscopic and electrochemical properties of the complexes (6, 7). Second-sphere coordination is important in biological processes such as iron siderophore recognition, and is currently
* E-mail: tmeade@northwestern.edu. Department of Chemistry. Current Address: Department of Chemistry, Pomona College. Ohmx Corporation, 1801 Maple Avenue, Suite 6143, Evanston, Illinois 60201. Departments of Biochemistry and Molecular and Cell Biology. | Departments of Neurobiology and Physiology and Radiology.

being investigated as a means to mimic the diverse functionality of metalloenzymes (9, 15, 16). The ability to modulate the properties of transition metal complexes through second-sphere ligand design is of interest for the development of articial metalloenzymes that combine the features of both heterogeneous and enzyme catalysis (17, 18). Ward and co-workers use chemogenetic protein engineering to optimize articial metalloenzymes comprising a catalyst anchored to a biological macromolecule (19-21). Second-sphere coordination is known to play an important role in activity and enantioselectivity in these catalytic reactions (17, 19-22). Previously, we reported a series of redox active transition metal-modied biotin complexes designed to probe the effects of protein binding on the electrochemical response (Figure 1) (23). The biotin-avidin system was chosen as a model due to its strong binding interaction (Ka ) 1015 M-1) and extensive structural characterization (24-26). Electrochemical studies demonstrated that the avidin-bound complexes are shielded and unable to effectively interact with the electrode. However, the bound complexes can be accessed via electrochemical mediators, suggesting that the complex is only partially shielded in the binding pocket of the protein (23). Quantitative results could not be obtained for the electrochemical potential of the proteinbound complex. Therefore, to investigate the effects of the protein binding on the electronic properties of the bound complex, we chose to use [Fe(bpy)(CN)4]2- complexes (Figure 1). [Fe(bpy)(CN)4]2- complexes possess electrochemical and optical properties that are highly sensitive to the local environment (4, 27-29). To investigate how differences in ligand-receptor binding energetics inuence the second-sphere coordination environment

10.1021/bc900270a CCC: $40.75 2009 American Chemical Society Published on Web 09/29/2009

Outer Sphere Effects of Iron(II) Complexes

Bioconjugate Chem., Vol. 20, No. 10, 2009 1931

Figure 1. Structures of [Fe(BMB)(CN)4]2- and [Fe(DMB)(CN)4]2-. Solvatochromic [Fe(bpy)(CN)4]2- complexes are targeted to bind avidin by modication with biotin (1) or desthiobiotin (2) through amide derivatization of 4-aminomethyl-methyl-2,2-bipyridine.

of a small molecule when bound to a protein, we designed two redox-active, solvatochromic transition metal complexes that bind to avidin with different afnities. Herein, we directly correlate systematic modications of the complex environment to observed changes in electrochemical and spectroscopic signals. The inuence of avidin binding on the electrochemical and spectroscopic properties of [Fe(BMB)(CN)4]2- (1) and [Fe(DMB)(CN)4]2- (2) is discussed in light of X-ray crystallographic, electrochemical, optical, computational, and thermodynamic studies.

EXPERIMENTAL SECTION
Materials and Methods. Deglycosylated avidin was used for crystallization studies and obtained from Belovo, Inc. Egg white avidin, used for thermodynamic studies, was obtained from Molecular Probes. Poly(ethylene glycol) was obtained from Fisher Scientic. All other chemicals were obtained from Aldrich. K2[Fe(BMB)(CN)4] (1) and K2[Fe(DMB)(CN)4] (2) were prepared according to previously reported methods (23). Electrochemistry. Cyclic voltammetry experiments were carried out in a conventional three-electrode cell using a CH Instruments 660A workstation (Austin, TX), glassy carbon working electrode (3 mm dia, CH Instruments), Pt wire counter electrode, and Ag/AgCl reference electrode. The reference electrode was calibrated to Fc/Fc+. The electrode was polished with alumina on a polishing cloth and the sample purged with N2 before each experiment. All electrochemical potentials (E1/2) are reported vs Ag/AgCl [where E1/2 is dened as (Epa + Epc)/2]. Cyclic voltammograms of 1 were collected in solvents of varying acceptor number (AN). A 200 L aliquot of a 50 mM aqueous solution of 1 was added to 3 mL of a 50/50 solvent/ 100 mM PBS buffer (pH 7.2) mixture where the solvents were water (AN ) 55), methanol (AN ) 41), DMSO (AN ) 19), or DMF (AN ) 16) (30). Cyclic voltammograms of 1 and 2 in 0%, 10%, 20%, 30%, 40%, and 50% DMF in PBS buffer were collected. Samples were prepared by adding appropriate amounts of solvent and water to 1.5 mL 100 mM PBS buffer (pH 7.2) to make a 3 mL sample with a constant electrolyte concentration of 50 mM. A 200 L aliquot of a 50 mM aqueous solution of 1 or 2 was added to the solution. For example, the 10% DMF sample was prepared by adding 0.3 mL DMF and 1.2 mL water to 1.5 mL 100 mM PBS buffer. In order to obtain cyclic voltammograms of 1 and 2 bound to avidin, a gold ball electrode was prepared by melting the end of a 0.127-mm-diameter gold wire into a ball approximately 0.75 mm in diameter. The electrode was cleaned electrochemi-

cally by immersing it in 2 N H2SO4 and cycling the potential between 0 and 1.6 V at a scan rate of 500 mV/s. The electrode was rinsed with water followed by ethanol and immediately immersed in a 5 mM ethanolic solution of 11-mercaptoundecanoic acid for 16 h. The electrode was washed with ethanol followed by water and 10% NH4OH(aq). It was then immersed in an aqueous solution consisting of 0.1 mM 1 or 2 and 4 mg/ mL avidin for 2 h. The electrode was washed with water, and cyclic voltammograms were obtained in 100 mM PBS buffer (pH 7.2). Absorption Spectroscopy. UV-vis absorption studies were performed using an Agilent 8453 spectrophotometer (Foster City, CA). Samples were prepared as described for electrochemical experiments. UV-vis spectra of 1 and 2 bound to avidin were obtained by preparing a solution of each in 100 mM PBS buffer and adding a 1.5-fold excess of avidin. Isothermal Titration Calorimetry. Thermodynamic studies of 1 and 2 were carried out using a VP-ITC microcalorimeter from MicroCal, Inc. (Northampton, MA). In a typical experiment, the reference cell was lled with pure distilled water. The reaction cell was lled with 1.8 mL of a 6 M solution of avidin (24 M avidin protomer) in pH 7.0 phosphate buffer. The injection syringe was lled with a 300 M solution of 1. After thermal equilibration to 25 C, 1 was injected into the reaction cell in 5 L aliquots every 200 s until saturation was achieved (18 injections). A reference offset of 25% was used. The experiment was repeated injecting 1 into pH 7.0 phosphate buffer. The data were subtracted from the binding data to correct for heats of dilution. The data were t using the Origin 7.0 OneSites model to determine binding constants (Ka) and heats of binding (H). The above procedure was repeated for compound 2. Crystallization and Data Collection. The K2[Fe(BMB)(CN)4] complex was incubated with avidin for 1 h at room temperature. The avidin-[Fe(BMB)(CN)4]2- complex was crystallized at 18 C by using the sitting drop vapor diffusion method. A protein solution containing 3.0 mg/mL avidin bound by [Fe(BMB)(CN)4]2- in phosphate buffer pH 7.0 was combined with an equal volume of precipitant solution comprising 100 mM sodium acetate (pH 4.6) and 25% PEG 3000. The resulting crystals were harvested after six weeks. The crystals were hexagonal, space group P3221, with unit cell dimensions a ) b ) 74.20, c ) 85.82. X-ray data were collected at 100 K at the Advanced Photon Source on beamline 5-ID from crystals ash-frozen in liquid nitrogen after soaking in a cryo-protectant composed of the well solution plus 20% ethylene glycol. The diffraction data were integrated with MOSFLM (31) and scaled by using SCALA (32). The statistics are presented in Table 5. Structure Determination and Renement. Phases were determined by molecular replacement using the program PHASER (33) and avidin from the pdb le 1LDO as the search model. The ligand-bound form of avidin was modeled with XtalView (34), rened with CNS (35), and validated with PROCHECK (36). Molecular Modeling. Force eld parameters were derived for [Fe(BMB)(CN)4]2- and [Fe(BMB)(CN)4]2-. Atomic partial charges were taken from the Mulliken population analysis of ab initio geometry minimizations. These calculations were carried out at the Hartree-Fock (HF) level in the program Gamess (37). A Hay-Watt effective core potential was used to describe the iron atoms (38). The carbon and nitrogen atoms in the bipyridyl and cyano ligands were described by 6-31+G* basis set (39). All other atoms were described by the 6-31 G* basis set (40). Geometries and charges were calculated for both 2+ and 3+ metal ionization states.

1932 Bioconjugate Chem., Vol. 20, No. 10, 2009

Table 1. Variation of MLCT Wavelengths and Electrochemical Potential of K2[Fe(BMB)(CN)4] with Acceptor Number (AN) solvent water MeOH DMSO DMF avidin AN 55 41 19 16 MLCT1 (nm) 480 496 512 519 500 MLCT2 (nm) 341 348 356 357 348 E1/2 (mV) 333 257 195 181 193

Barker et al.

Iron-ligand bonds were modeled using the points-on-a-sphere methodology (41). In this method, iron-nitrogen and iron-carbon bond lengths are held constant (acquired from crystal structures of similar compounds) (42) and the atom-iron-atom bond angles are determined by the electrostatic interaction of the atoms around the metal center. The crystal structure of the biotin-avidin complex (obtained from the Protein Data Bank) was used as the starting geometry for the molecular dynamics simulation (26). Hydrogen atoms and other atoms missing from the crystal structure were added referencing standard amino acid residues from AMBER. The ab initio optimized geometry coordinates for the [Fe(bpy)(CN)4]2- complex were appended to the coordinates of the crystal structure of avidin bound biotin for each of the four binding avidin protomers. The energy of the avidin/{[Fe(BMB)(CN)4]2-}4 tetrameric complex was minimized using a combined steepest descent and conjugate gradient minimization technique. The minimized structure was solvated with a 75 75 75 box of water of approximately 13 800 TIP3P water molecules. The solvated protein structure was minimized using the method described above. Molecular dynamics (MD) simulations were carried out for the avidin/{[Fe(BMB)(CN)4]2-}4 tetrameric complex using the Amber 94 force eld (43). Constant volume periodic boundary conditions were employed; the simulation boundary box constructed was slightly larger than the extent of the water buffer used to solvate the protein. A dielectric constant of 4 was assigned to the boundary box to dampen the effects of electrostatics on the system. MD simulations were run for 1 ns recording energy and trajectory information every picosecond. The isopotential surfaces were rendered with HARLEM from a calculation of the electrostatic potential by solving the Poisson-Boltzmann equation using the method of nite differences (44). Electrostatic calculations were carried out using the partial charges on the avidin/{[Fe(BMB)(CN)4]2-}4 complex to represent the charge distribution. The static dielectric constant as a function of position, also used in the Poisson equation, was mapped by dening the low dielectric volume of the protein by a solvent-accessible surface (SAS). A solvent probe radius of 1.4 was used to dene the SAS. Atoms contained inside the boundary of the surface are assigned a low dielectric constant of 4, outside the surface the dielectric is assigned value of 80 for water. These calculations yield the isopotential surface surrounding the modied ligands. Calculations were made for atoms within a 10 shell around the binding ligand. This calculation provides an estimate of the electric eld immediately surrounding the ligand and indicates portions of the protein with net positive or net negative charge.

Figure 2. (A) Correlation of the K2[Fe(BMB)(CN)4] MLCT bands with solvent acceptor number. Closed () symbols represent MLCT2; open symbols (O) represent the MLCT1. (B) Correlation of the K2[Fe(BMB)(CN)4] E1/2 with solvent acceptor number. Cyclic voltammograms were collected at 100 mV/s using a glassy carbon electrode and Ag/AgCl reference. Values are reported vs Ag/AgCl.
Table 2. Variation of MLCT Wavelengths and Electrochemical Potential of K2[Fe(BMB)(CN)4] with Increasing DMF % DMF 0 10 20 30 40 50 avidin MLCT1 (nm) 480 490 496 510 514 519 500 MLCT2 (nm) 341 345 347 349 352 357 348 E1/2 (mV) 333 301 270 246 221 211 193

Table 3. Variation of MLCT Wavelengths and Electrochemical Potential of K2[Fe(DMB)(CN)4] with Increasing DMF % DMF 0 10 20 30 40 50 avidin MLCT1 (nm) 486 494 501 510 519 529 496 MLCT2 (nm) 345 351 357 360 363 368 351 E1/2 (mV) 330 302 269 238 212 184 203

RESULTS
Electrochemistry. The E1/2 of 1 and 2 in PBS pH 7.2 are 333 mV and 330 mV, respectively. These values are in agreement with previously reported values and comparable to thatof[Fe(4,4-dimethyl-2,2-bipyridine)(CN)4]2- (312mV)(23,45). The E1/2 of 1 increases with increasing solvent acceptor number (AN) where AN describes the Lewis acidity of the solvent (Table 1, Figure 2) (30, 46). When DMF is introduced incrementally to an aqueous solution of 1 or 2, the E1/2 decreases linearly (Tables 2 and 3, Figure 3). The plot of current (i) vs

square root of scan rate (V) is linear for 1 and 2 in all solvent systems. The average Ep (difference of the peak potentials) for 1 and 2 is 64 mV and is consistent for all solvent systems. The ia/ic ratio is 0.8 for 1 and 1.0 for 2 in all solvent systems. An electrochemical signal is not observed for 1 or 2 bound to avidin in solution (23). When adsorbed to a mercaptoundecanoic acid (MUA) SAM-coated gold electrode, 1 has an E1/2 of 193 mV and 2 has an E1/2 of 203 mV. At 100 mV/s, the Ep for 1 in this experiment is 85 mV with an ia/ic ratio of 1.1 (Supporting Information Figure S1). The Ep for 2 is 75 mV with an ia/ic ratio of 1.1 (Figure 4). Absorption Spectroscopy. For an aqueous solution of 1, a charge transfer (CT) band is observed at 293 nm and two MLCT bands are observed at 341 nm [metal d f bipyridine * (2) (MLCT2)] and 480 nm (metal d f bipyridine * (1) (MLCT1)) (47). The max of the MLCT peaks are red-shifted

Outer Sphere Effects of Iron(II) Complexes

Bioconjugate Chem., Vol. 20, No. 10, 2009 1933

Figure 3. Electrochemistry of K2[Fe(BMB)(CN)4] and K2[Fe(DMB)(CN)4] in increasing concentrations of DMF in PBS buffer at pH 7.2. (A) Representative cyclic voltammograms of K2[Fe(DMB)(CN)4] (B) Correlation of the E1/2 of K2[Fe(BMB)(CN)4] with increasing amounts of DMF (C) Correlation of the E1/2 of K2[Fe(DMB)(CN)4] with increasing amounts of DMF. Cyclic voltammograms were collected at 100 mV/s using a glassy carbon electrode and Ag/AgCl reference. Values are reported vs Ag/AgCl.

Figure 4. Cyclic voltammogram of (A) K2[Fe(DMB)(CN)4] and (B) K2[Fe(BMB)(CN)4] adsorbed to a mercaptoundecanoic acid SAM on a gold ball electrode. The cyclic voltammograms were collected at 100 mV/s in 100 mM PBS at pH 7.2 with Ag/AgCl as a reference.

linearly with decreasing solvent acceptor number (Table 1, Figure 2). When DMF is introduced incrementally to an aqueous solution of 1, the max values of the MLCT peaks are red-shifted proportionally (Table 2, Figure 5). Upon addition of avidin to 1 in PBS buffer, the MLCT1 band shifts 20 nm (833 cm-1) and the MLCT2 band shifts 7 nm (590 cm-1) (Table 2). When an excess of biotin is added to the solution of 1-avidin, the max values of the MLCT peaks are shifted back to the original values. For an aqueous solution of 2, a CT band is observed at 293 nm and two MLCT bands are observed at 345 nm (MLCT2) and 486 nm (MLCT1). When DMF is introduced incrementally to an aqueous solution of 2, the max values of the MLCT peaks are red-shifted proportionally (Table 3, Figure 5). Upon addition of avidin to 2 in buffer, the MLCT1 band shifts 10 nm (415

cm-1) and the MLCT2 band shifts 6 nm (495 cm-1) (Table 3). When an excess of biotin is added to the solution of 2-avidin, the max values of the MLCT peaks are shifted back to the original values. Isothermal Titration Calorimetry. For 1 and 2, the amount of energy released per mole of complex titrated was plotted versus the molar ratio of complex to protein. The curve was best t using the Origin OneSites model to give the binding constant (Supporting Information Figure S1). The reduced 2 value was 1.3 and 1.1 for 1 and 2, respectively. Binding constants and heats of binding for 1 and 2 binding to avidin are summarized in Table 4. Structure of the [Fe(BMB)(CN)4]2--Avidin Complex. Complex 1 bound to both protomers in the 3.1 resolution crystal structure of the avidin complex is shown in Figure 6. As expected, the biotin group binds to a site identical to one previously characterized (26). In one protomer, electron density for the [Fe(BMB)(CN)4]2- group is observed on the surface of the protein, sandwiched among three loops: L3,4; L5,6; and L7,8 (Figure 6B). This conguration appears stabilized by several hydrogen bonding pairs including interactions of N12, S16, Y33, T35, T77, and N118 with the ureido portion of the complex, S73 and S75 with the carbonyl of the amide bond, and S31 and S102 with the transition metal portion of the complex. In the second protomer, electron density for the [Fe(BMB)(CN)4]2group and residues 41-44 on the L3,4 loop were not observed. Molecular Modeling. In the course of the molecular dynamics simulation, several conformations were found to be accessible to 1 when bound to avidin (Supporting Information Figure S2). The conformation in closest agreement with the crystal structure was chosen to compare the second-sphere environment of 1 and 2 when bound to avidin. In this conformation, the distances from the R carbon of Ser 41 and the metal atom of 1 and 2 are 5.76 and 5.92 , respectively. The distance between Ser 101 and the metal atom was found to be approximately 10 for both 1 and 2. The distance between Ser 102 and the metal atom was found to be approximately 5 for both 1 and 2. Loss of architecture in strands 3 and 4 is observed and loops 7,8 and 5,6 are disordered for 2 bound to avidin. The models of 1 and 2 bound to avidin are compared in Figure 7. The isopotential surface of the portion of the protein that surrounds avidin (Figure

1934 Bioconjugate Chem., Vol. 20, No. 10, 2009

Barker et al.

Figure 5. Absoprtion spectroscopy of K2[Fe(BMB)(CN)4] and K2[Fe(DMB)(CN)4] in increasing concentrations of DMF in PBS buffer at pH 7.2. (A) Representative UV-vis spectra of K2[Fe(DMB)(CN)4]. (B) Correlation of the K2[Fe(BMB)(CN)4] MLCT bands. (C) Correlation of the K2[Fe(DMB)(CN)4] MLCT bands with increasing amounts of DMF. The wavelengths of the MLCT bands are linearly correlated to the solvent fraction indicating that no preferential solvation of the complexes occurs.
Table 4. Binding Constants and Enthalpy of Binding for Iron(II) Complexes ligand K2[Fe(BMB)(CN)4] K2[Fe(DMB)(CN)4] K (M-1) 3.5 107 1.5 106 H (kcal/mol) -104 -80 Table 5. Data Collection, Phasing, and Renement Statistics data collection beamline wavelength () resolution range ()a total observations unique observations completeness (%) redundancy I/ Rsymb (%) Renement no. of reections, working/test c Rwork (%) Rfree (%)d Protein atoms ligand atoms water molecules r.m.s.d. bond length () r.m.s.d. bond angle () average B-value (2) DND-CAT (Sector 5) 1.279 30-3.1 (3.27-3.10) 64 647 (9347) 9587 (742) 99.9 (100) 12.3 (12.6) 28.0 (5.7) 9.9 (40.6) 8292/975 25.5 33.0 1856 61 22 0.009 1.4 65.5

8) is mainly positive in nature. This portion of the protein should have favorable interactions with the net negative charge of 1 and 2.

DISCUSSION
The redox properties of metal complexes in proteins are inuenced by changes in the surrounding protein environment (e.g., cytochrome c and blue copper proteins) (48-50). Metal cofactors bound to proteins experience a solvation environment with a lower local dielectric constant than that of aqueous buffer. The dielectric constant () of water is 80, whereas is estimated to be between 4 and 20 inside a protein (51-53). Changes in the dielectric constant affect the energetics for electronic transitions of metal redox centers. Further, the protein imparts a structured charge environment around the metal complex due to the partial charges of each atom. This charge structure, with its partial positive or negative properties, stabilizes more strongly either the reduced or oxidized state of the metal. Finally, the protein can be viewed as a scaffold containing electrophilic or nucleophilic residues capable of multiple interactions with the metal cofactor. In order to probe changes in outer-sphere coordination induced by changes in ligand-receptor binding, we have synthesized a biotinylated iron(II) complex (1) and a desthiobiotinylated iron(II) complex (2). These redox-active, solvato-

a Values in parentheses are for the highest-resolution shell. b Rsym ) i hkl |Ii(hkl) - I(hkl)|/hkl I(hkl), where Ii(hkl) is the ith measured diffraction intensity and I(hkl) is the mean of the intensity for the Miller index (hkl). c Rwork ) hkl |Fo(hkl)| - |Fc(hkl)|/hkl|Fo(hkl)|. d Rfree ) Rwork for a test set of reections (5%).

chromic transition metal complexes are designed to bind to the protein avidin. Biotin and desthiobiotin bind to avidin with Kas of 1015 M-1 and 1013 M-1, respectively (54). Therefore, weaker ligand-receptor binding is expected to give the [Fe(DMB)(CN)4]2- complex slightly more exibility within the binding pocket as compared to [Fe(BMB)(CN)4]2-. Therefore, the dielectric constant, charge structure, van der Waals, and

Outer Sphere Effects of Iron(II) Complexes

Bioconjugate Chem., Vol. 20, No. 10, 2009 1935

Figure 6. Structure of [Fe(BMB)(CN)4]2- bound to avidin. (A) The asymmetric unit shows [Fe(BMB)(CN)4]2- bound to two avidin protomers. The other two protomers are related by crystallographic symmetry. 2|Fo| - |Fc| omit maps are depicted around the modeled ligand at 1 (green) and 4 (red). (B) View of ligand-protein interactions. Amino acids in green interact with biotin as observed in crystal structures of native D-biotin bound to avidin (26). Residues in gray hydrogen bond to the [Fe(bpy)(CN)4] portion of the complex. The 2|Fo| - |Fc| omit density is superposed at 4.

Figure 7. A view of the nal structure of the molecular dynamics simulation of avidin bound by (A) [Fe(BMB)(CN)4]2- and (B) [Fe(DMB)(CN)4]2-. One monomer of the tetramer avidin is shown.

Figure 8. Isopotential surfaces of the avidin binding pocket (shown in space lling mode) as calculated by the program HARLEM. Blue corresponds to positive charges, and red corresponds to negative charges. This representation indicates that there are favorable interactions between the binding pocket and the negatively charged 1 (shown) and 2.

hydrogen bonding contacts near the metal will differ for 1 and 2. We have used a variety of techniques to observe differences in the electronic structure of these complexes as imparted by the protein. Electronic Structure, Solvatochromism, and Acceptor Number. Iron(II) tetracyano complexes are known to be highly sensitive to changes in environment (4, 27, 29, 55, 56). [Fe(bpy)CN)4]2- is a paramagnetic complex with C2V symmetry that has completely lled d orbitals (dxy, dxz, dyz), and these orbitals act as the HOMO set for the molecule. Oxidation of

the iron(II) to iron(III) species requires removal of an electron from one of these metal-centered orbitals. The environmental sensitivity of these complexes is largely due to noncovalent interactions with the lone pair electrons on the cyano ligands. Solvent molecules interact noncovalently with these lone pairs. Strong electron acceptors such as water decrease the electron density at the CN and therefore increase backbonding with the metal ion. This interaction stabilizes the iron(II) state with respect to the iron(III) state. Depending on solvent, the LUMO in [Fe(bpy)(CN)4]2- might be either the unlled metal d-orbitals (dz2 and dx2-y2) or the * levels of the bipyridine ligands (28). The rst optically allowed transitions from the d orbitals, however, are to the * orbitals of the bipyridine ligand (47). Weak acceptors such as DMF decrease the backbonding as compared to water. Therefore, as DMF is added to the system, water molecules are displaced and the iron(II) state becomes destabilized. A solvent with a high acceptor number such as water will result in a higher-energy metal to ligand charge transfer (MLCT) as compared to weak acceptors such as DMF. These changes in MLCT energy are signicant enough to be observed spectroscopically. Likewise, stabilization of the iron(II) complex will result in higher oxidation potentials that can be observed by electrochemical methods such as cyclic voltammetry. The absorbance of 1 in various solvents was measured to investigate the sensitivity of the physical properties of 1 to second-sphere effects. The solvents were chosen to represent a range of ANs. We observed a larger shift of the low-energy MLCT1 band compared to the MLCT2 band. This result indicates that the solvent is interacting with both the cyano and bipyridine ligands to stabilize the * orbitals of the bipyridine

1936 Bioconjugate Chem., Vol. 20, No. 10, 2009

Barker et al.

ligand to varying degrees. This phenomenon has been described for cyanoiron(II) complexes by Toma and co-workers and demonstrates that changes in the second-sphere environment around any part of the molecule, whether it is the bipyridine or the cyano ligands, result in spectroscopic changes (56). As increasing amounts of DMF are added to an aqueous solution of 1 or 2, the energy of MLCT bands decreases linearly (Table 2, Figures 3 and 5). This linear correlation in a binary solvent system demonstrates the effect of an increasingly nonpolar environment and indicates that no preferential solvation of the complexes occurs (56). DMF displaces water molecules from the second sphere of the metal complex, and the complexes do not have a tendency to maintain a solvation shell of water when a nonpolar solvent is introduced. To investigate the sensitivity of the redox potentials of 1 and 2 to second-sphere coordination, the electrochemical potential E1/2 of 1 and 2 was determined in a variety of solvents. As the AN is decreased, the E1/2 decreases (Table 1, Figure 2). Further, as increasing amounts of DMF are added to an aqueous solution of 1 or 2, the E1/2 decreases linearly (Table 2, Figures 3 and 5). These solvent experiments provide a basis for rough comparison with avidin-binding results. Upon addition to an aqueous solution of 1 or 2, DMF replaces water in the vicinity of the metal complex; when 1 or 2 binds to avidin, water will be displaced from the metal center in a similar way. X-ray Crystal Structure and Binding Properties. Isothermal titration calorimetry was used to measure the binding constants of 1 and 2 to avidin. The binding afnities of 1 and 2 for avidin are 3.5 107 M-1 and 1.5 106 M-1, respectively. These values demonstrate a signicant decrease in binding afnity compared to that of native D-biotin (K ) 1 1015 M-1) and D-desthiobiotin (K ) 1 1013 M-1). The data t best to the case where all four binding sites are identical, and binding is independent. Previous studies by Livnah and co-workers demonstrate that when bound to avidin, modication of biotin at the carboxyl terminus by a caproic acid results in disruption of the hydrogen bonding of Thr 38, Ala 39, and Thr 40 with the carboxyl terminus of biotin. This disruption results in a decrease in binding afnity (7 to 8 orders of magnitude) for biotin analogues that have an amide at the carboxylic acid (57). The crystal structure shows two unique protomers (Figure 6). Avidin is a tetramer; the other two protomers are related by crystallographic symmetry. It is apparent that 1 is in a relatively hydrophobic (low dielectric) environment and interacts directly with the protein via electrostatic and hydrogen bonding interactions. While several biotinylated-transition metal complexes have been synthesized, only one other has been structurally characterized bound to avidin (20, 23, 58-61). Similar to the observations of Ward et al., there are short contacts between the metal center of 1 and amino acid residues. Importantly, there is no major reorganization of the host protein. The three-dimensional structure of 1 bound to avidin shows that the cyano ligands of the iron complex interact with specic residues in the avidin binding pocket. In one protomer, the Ser 41 residue of the L3,4 loop interacts directly with a CN group on the iron(II) complex, and Ser 102 of the L5,6 loop interacts with a CN group through a water molecule. We expect these interactions with the cyano groups to have an observable effect on the electrochemical properties of 1 and 2 bound to avidin. The stabilizing, attractive interaction of the protein with the cyano groups prevents the loops from opening as they do in the cases where the carboxyl terminus is derivatized by caproic acid (40, 57, 62). In the second protomer, electron densities for the [Fe(bpy)CN)4]2- moiety of 1 and residues 41-44 on the L3,4 loop were not observed. These differences are most likely due to disorder imposed by crystal packing. This conclusion is sup-

ported by ITC data that are best t for a system with four identical binding events. We have used the program HARLEM to calculate the isopotential surface of avidin and show that the binding pocket of avidin is positive in nature (Figure 8). 1 and 2 are negatively charged and have small polar ligands that should promote interactions with the protein. This portion of the protein should have favorable interactions with the net negative charges of 1 and 2. Molecular Dynamics Simulations. Molecular dynamics simulations were carried out with 1 and 2 bound to avidin to compare the local environments of the transition metal complexes when bound to avidin. Several similar conformations were found to be accessible to 1 and 2 when bound to avidin in the course of the molecular dynamics simulation (conformations of 1 shown in Supporting Information Figure S2). The conformation in closest agreement with the crystal structure of 1 bound to avidin was chosen to compare the second-sphere environment of 1 and 2 when bound to avidin. For comparison, the distance between the apex of loop 3,4 (Ser 41) and the metal center was measured. This distance is specically of interest because the crystal structure shows that Ser 41 has a direct interaction with 1. The distance from the R-carbon of Ser 41 to the iron center in 1 is 5.76 , whereas in 2, the distance is 5.92 . We did not observe a water molecule between Ser 102 and the cyano group in the molecular dynamics simulation, as was observed in the crystal structure of avidin with 1. The MD simulations show that Ser 101 has a stronger interaction with the binding ligand than does Ser 102. The distance between Ser 101 and the iron center is 5 , while the distance between Ser 102 and the iron atom is 10 . These observations are slightly different than values determined from the crystal structure; however, they are reasonable due to the uxional nature of the protein in the molecular dynamics simulations. While the location of the ligand within the protein in the crystal structure is very informative, it is only a snapshot of the ensemble of congurations that the metal complex samples during a spectral or electrochemical measurement. The MD simulations further show that, when 2 is bound, the protein structure is more disordered. Figure 7 shows the disorder in strands 3 and 4 and loops 7,8 and 5,6 for 2 as compared to 1 bound to avidin. The environment of 2 is slightly more open and water-accessible than the environment that 1 experiences. Electronic Affects of Avidin Binding. UV-vis spectroscopy was used to measure roughly the HOMO-LUMO gaps of 1 and 2 while bound to the protein. The optical spectra in aqueous solution show two MLCT bands, one to each of the rst two * levels of the bipyridine ligands (MLCT1 at lower energies, 480 nm for 1 and 486 nm for 2; and MLCT2 at higher energies, 341 nm for 1 and 345 nm for 2). Upon binding of avidin to 1, the MLCT1 band red-shifts 20 nm and the MLCT2 band redshifts 7 nm. Upon binding of avidin to 2, the MLCT1 band red-shifts 10 nm and the MLCT2 band red-shifts 6 nm. When an excess of biotin is added to displace 1 or 2 from avidin, the max of the MLCT bands shifts back to the original values reported for the compounds in aqueous solution. These results demonstrate that the outer-sphere coordination of the molecule changes with the ligand receptor binding. The UV-vis studies show that the molecule with the lower binding constant (2) has smaller changes in MLCT bands compared to that of 1, consistent with a higher acceptor number environment. The spectroscopic shifts upon avidin binding to 1 and 2 are similar to the shifts observed when increasing amounts of DMF are added to aqueous solutions of 1 or 2. The protein binding event displaces water molecules from the metal center just as DMF replaces water molecules that are interacting with the metal center. This solvent system provides a convenient model

Outer Sphere Effects of Iron(II) Complexes

Bioconjugate Chem., Vol. 20, No. 10, 2009 1937

for systematically varying the environment to crudely mimic protein binding. Electrochemistry. The redox potentials of 1 and 2 while bound to avidin were found to shift by -140 and -127 mV, respectively. These potentials were measured by electrostatically adsorbing the protein to a monolayer of mercaptoundecanoic acid (MUA). These dramatic shifts are most likely due to the interaction of the complex with the monolayer as well as the protein. Being sandwiched between the protein and the monolayer minimizes any interaction with bulk water and places the complexes in an even lower dielectric environment compared to avidin in solution. The protein/monolayer combination destabilizes the iron(II) state with respect to the iron(III) state, compared with the complex in aqueous solution, raising the energy level of the orbitals that comprise the HOMO. The HOMO is destabilized by 1130 cm-1 and 1050 cm-1 for 1 and 2, respectively. Immobilization techniques are commonly used to study electroactive proteins (61, 63-70). In our case, a monolayer of MUA was formed and then washed with NH4OH(aq) in order to leave a number of negatively charged acids on the surface (the pI of avidin is 10, and therefore the protein should be positively charged at pH 7). Figure 8 shows the charge distribution at the binding pocket of avidin at pH 7. There is a signicant area with a positive charge near the binding site. This positive charge ensures that a portion of the avidin on the monolayer will be oriented so that the bound iron complexes will make effective contact. The observed electrochemical signal is likely from these bound complexes (61). Since avidin is a tetramer, it is likely that a maximum of two of the binding sites are in contact with the monolayer at any given time.

CONCLUSION
We have established that the electronic structures of [Fe(BMB)(CN)4]2- and [Fe(DMB)(CN)4]2- are highly sensitive to noncovalent interactions. The optical spectrum shifts up to 20 nm in organic solvents with lower acceptor numbers than water. The shift is linearly dependent on the solvent acceptor number indicating there are specic interactions between the solvent and the CN and bpy ligands. Similarly, the electrochemical response of 1 and 2 in various organic solvents shows a shift that is linearly dependent on solvent AN; shifts of up to 150 mV are observed. These data suggest that the electronic structure of 1 and 2 is sensitive to second-sphere coordination changes and solvation. The results from solvent studies conrm that the complexes can be used as sensitive probes to investigate the electronic properties of protein binding sites. Binding of 1 and 2 to avidin results in visible changes in the electronic spectra. The shifts are similar to those observed in the solvent studies. We expect the binding event to exclude water from the metal complex, resulting in a lower AN environment. For example, binding of 1 to avidin can be compared to a solution of 1 in 23% DMF in water. Our data show that the environments of 1 and 2 in avidin are observably different. ITC measurements show that the binding of 2 is weaker than that of 1 by a factor of 20. Therefore, we expect 1 and 2 to sample slightly different environments in the binding pocket. Indeed, small differences are observed in both the optical and electrochemical data. Upon binding to avidin, the optical spectrum shifts 20 and 7 nm for 1, while the same transitions shift 10 and 6 nm for 2. The binding of 2 can be compared to 12% DMF in water. The difference in redox potential shifts is small (10 mV), yet well above the error for such a measurement. The three-dimensional X-ray structure of 1 bound to avidin shows 1 partially buried in a hydrophobic pocket with a direct

interaction between the protein and the transition metal complex. An open-loop conformation, expected from the presence of the amide, would prevent the avidin from shielding the complex from bulk water and minimize the observable change in electrochemical potential and absorbance properties. Isopotential calculations, however, show a patch of positive charge near the binding site that should favor binding of the negatively charged 1 and 2. These interactions may partially counteract the effects of having the amide in place of the carboxylic acid. Finally, molecular dynamics simulations show that, when 2 is bound to avidin, the protein structure is slightly more disordered. However, for the MD snapshot most resembling the crystal structure, the distances between the metal center and select residues are nearly the same as when 1 is bound. Together, these results demonstrate that even small changes in the binding of a ligand (1 or 2) result in small yet experimentally observable changes in second-sphere coordination imposed by the protein. Specically, these changes include the partial charge structure, van der Waals interactions, and hydrogen bonds. Small changes in these weak interactions are powerful enough to induce changes in the electrochemical and optical properties of the metal cofactors. The synthetic versatility and the sensitivity of [Fe(bpy)(CN)4]2- complexes to local environmental changes makes these complexes candidates for use in electronic and optical protein detection devices for any number of ligand-receptor pairs. For example, by systematic change of the protein binding ligand (e.g., adding or removing hydrogen bond donors) and observing the effect of protein binding on the spectroscopic properties, the nature of an unknown binding pocket can be evaluated. Further, the shift in the electrochemical potential indicates whether solvent is being excluded from the metal center and therefore can be used to evaluate the hydrophobicity of the binding site. Sensitivity to small changes in binding afnities indicates that these species can be used as probes for remotely evaluating the noncovalent factors that affect ligand-receptor binding. Our future studies will translate this approach to self-assembled monolayer systems for investigating the effects of protein binding on the electrochemical properties of the metal complex.

ACKNOWLEDGMENT
We thank Belovo, Inc. for providing deglycosylated avidin. We thank Harry Gray, Igor Kurnikov, Baudilio Tejerina, and Alec Durrell for helpful discussions. This work was funded by the Nanoscale Science and Engineering Initiative of the National Science Foundation under NSF Award Number EEC-0647560 and Ohmx Corporation. MR thanks the MURI program of the DoD for support. Supporting Information Available: Molecular modeling overlay with crystal structure, isothermal titration calorimetry data, cyclic voltammograms of 1 in 0-50% DMF/water. This material is available free of charge via the Internet at http:// pubs.acs.org.

LITERATURE CITED
(1) Lever, A. B. P. (1984) Inorganic Electronic Spectroscopy, 2nd ed., Vol 33, Elsevier, Amsterdam. (2) Nigam, S., and Rutan, S. (2001) Principles and applications of solvatochromism. Appl. Spectrosc. 55, 362a370a. (3) Dodsworth, E. S., Hasegawa, M., Bridge, M., and Linert, W. (2004) Solvatochromism. Compr. Coord. Chem. II 2, 351365. (4) Alousy, A., Blundell, N. J., Burgess, J., Hubbard, C. D., and Van Eldik, R. (2002) Solvatochromism and piezochromism of dicyano-, tricyano-, and tetracyano-diimine-iron(II) complexes. Transition Met. Chem. (Dordrecht, Neth.) 27, 244252.

1938 Bioconjugate Chem., Vol. 20, No. 10, 2009 (5) Gritzner, G., Danksagmuller, K., and Gutmann, V. (1976) Outer-sphere coordination effects on redox behavior of Fe(CN)63-Fe(CN)64- couple in non-aqueous solvents. J. Electroanal. Chem. 72, 177185. (6) Ando, I. (2004) Hydrogen bonding of 18-crown-6 ether to ruthenium-ammine complexes at second sphere. Coord. Chem. ReV. 248, 185203. (7) Todd, M. D., Dong, Y., and Hupp, J. T. (1991) Crown ether encapsulation effects upon optical electron-transfer energetics in a symmetrical mixed-valence system. Inorg. Chem. 30, 4685 4687. (8) Beck, M. T. (1968) Chemistry of outer-sphere complexes. Coord. Chem. ReV. 3, 91115. (9) Harwani, S., and Telford, J. R. (2005) Outer-sphere coordination of metal complexes in biological and cyclodextrin-based systems. Chemtracts: Inorg. Chem. 18, 437448. (10) Raymo, F. M., and Stoddart, J. F. (1996) Second-sphere coordination. Chem. Ber. 129, 981990. (11) Colquhoun, H. M., Stoddart, J. F., and Williams, D. J. (1983) Second-sphere coordination -- a novel role for molecular receptors. Angew. Chem., Int. Ed. 25, 487582. (12) Werner, A. (1913) Neuere Anschauungen auf dem Gebiete der anorganischen Chemie, 3rd ed, Vieweg, Braunschweig. (13) Werner, A. (1912) Ber. Dtsch. Chem. Ges. 64, 121. (14) Colquhoun, H. M., Stoddart, J. F., and Williams, D. J. (1981) Formation and X-ray crystal structure of [Pt(H2NCH2CH2NH2)2.18crown-6]n2+[PF6]2n-. A hydrogen bonded stepped-chain copolymer. Chem. Commun. 847849. (15) Shook, R. L., and Borovik, A. S. (2008) The effects of hydrogen bonds on metal-mediated O2 activation and related processes. Chem. Commun. 60956107. (16) Bjerre, J., Rousseau, C., Marinescu, L., and Bols, M. (2008) Articial enzymes, chemzymes: current state and perspectives. Appl. Microbiol. Biotechnol. 81, 111. (17) Steinreiber, J., and Ward, T. R. (2008) Articial metalloenzymes as selective catalysts in aqueous media. Coord. Chem. ReV. 252, 751766. (18) Ward, T. R. (2008) Articial enzymes made to order: combination of computational design and directed evolution. Angew. Chem., Int. Ed. 47, 78027803. (19) Pordea, A., and Ward, T. R. (2008) Chemogenetic protein engineering: an efcient tool for the optimization of articial metalloenzymes. Chem. Commun. 42394249. (20) Creus, M., Pordea, A., Rossel, T., Sardo, A., Letondor, C., Ivanova, A., Le Trong, I., Stenkamp, R. E., and Ward, T. R. (2008) X-ray structure and designed evolution of an articial transfer hydrogenase. Angew. Chem., Int. Ed. 47, 14001404. (21) Loosli, A., Rusbandi, U. E., Gradinaru, J., Bernauer, K., Schlaepfer, C. W., Meyer, M., Mazurek, S., Novic, M., and Ward, T. R. (2006) (Strept)avidin as host for biotinylated coordination complexes: Stability, chiral discrimination, and cooperativity. Inorg. Chem. 45, 660668. (22) Pordea, A., Creus, M., Panek, J., Duboc, C., Mathis, D., Novic, M., and Ward, T. R. (2008) Articial metalloenzyme for enantioselective sulfoxidation based on vanadyl-loaded streptavidin. J. Am. Chem. Soc. 130, 80858088. (23) Eckermann, A. L., Barker, K. D., Hartings, M. R., Ratner, M. A., and Meade, T. J. (2005) Synthesis and electrochemical characterization of a transition-metal-modied ligand-receptor pair. J. Am. Chem. Soc. 127, 1188011881. (24) Green, N. M. (1990) Avidin and streptavidin. Methods Enzymol. 184, 5167. (25) Livnah, O., Bayer, E. A., Wilchek, M., and Sussman, J. L. (1993) Three-dimensional structures of avidin and the avidinbiotin complex. Proc. Natl. Acad. Sci. U.S.A. 90, 507680. (26) Livnah, O., and Sussman, J. L. (1990) Crystal forms of avidin. Methods Enzymol. 184, 9093. (27) Gritzner, G., Danksagmuller, K., and Gutmann, V. (1976) Outer-sphere coordination effects on the redox behaviour of the Fe(CN)63-/Fe(CN)64- couple in non-aqueous solvents. J. Electroanal. Chem. 72, 177185.

Barker et al. (28) Winkler, J. R., Creutz, C., and Sutin, N. (1987) Solvent tuning of the excited-state properties of (2,2-bipyridine)tetracyanoferrate(II): direct observation of a metal-to-ligand charge-transfer excited state of iron(II). J. Am. Chem. Soc. 109, 34703471. (29) Burgess, J., Chambers, J. G., and Haines, R. I. (1981) Solvatochromic behavior of intramolecular charge-transfer spectra of inorganic diimine complexes. Transition Met. Chem. (Dordrecht, Neth.) 6, 145151. (30) Mayer, U., Gutmann, V., and Gerger, W. (1975) Acceptor number - quantitative empirical parameter for electrophilic properties of solvents. Monatsh. Chem. 106, 12351257. (31) Leslie, A. G. W. (1992) Recent changes to the MOSFLM package for processing lm and image plate data. Joint CCP4+ESF-EAMCB Newsletter on Protein Crystallography 26. (32) (1995) Collaborative Computational Project, The CCP4 suite: Programs for protein crystallography. Acta Crystallogr., Sect. D D50, 760-763. (33) McCoy, A. J., Grosse-Kunstleve, R. W., Storoni, L. C., and Read, R. J. (2005) Likelihood-enhanced fast translation functions. Acta Crystallogr., Sect. D D61, 458464. (34) McRee, D. E. (1999) XtalView/Xt - A versatile program for manipulating atomic coordinates and electron density. J. Struct. Biol. 125, 156165. (35) Brunger, A. T., Adams, P. D., Clore, G. M., Delano, W. L., Gros, P., Grosse-Kunstleve, R. W., Jiang, J.-S., Kuszewski, J., Nilges, N., Pannu, N. S., Read, R. J., Rice, L. M., Simonson, T., and Warren, G. L. (1998) Crystallography and NMR system (CNS): A new sofware system for macromolecular structure determination. Acta Crystallogr., Sect. D 54, 905921. (36) Laskowski, R. A., MacArthur, M. W., Moss, D. S., and Thorton, J. M. (1993) PROCHECK: A program to check the stereochemical quality of protein structures. J. Appl. Crystallogr. 26, 283291. (37) Schmidt, M. W., Baldridge, K. K., Boatz, J. A., Elbert, S. T., Gordon, M. S., Jensen, J. H., Koseki, S., Matsunaga, N., Nguyen, K. A., et al. (1993) General atomic and molecular electronic structure system. J. Comput. Chem. 14, 13471363. (38) Hay, P. J., and Wadt, W. R. (1985) Ab initio effective core potentials for molecular calculations. Potentials for the transition metal atoms scandium to mercury. J. Chem. Phys. 82, 270283. (39) Hehre, W. J., Radom, L., and Pople, J. A. (1972) Molecular orbital theory of the electronic structure of organic compounds. XII. Conformations, stabilities, and charge distributions in monosubstituted benzenes. J. Am. Chem. Soc. 94, 14961504. (40) Huberman, T., Eisenberg-Domovich, Y., Gitlin, G., Kulik, T., Bayer, E. A., Wilchek, M., and Livnah, O. (2001) Chicken avidin exhibits pseudo-catalytic properties: biochemical, structural, and electrostatic consequences. J. Biol. Chem. 276, 3203132039. (41) Hay, B. P. (1993) Methods for molecular mechanics modeling of coordination compounds. Coord. Chem. ReV. 126, 177236. (42) Lescoueezec, R., Lloret, F., Julve, M., Vaissermann, J., and Verdaguer, M. (2002) [Fe(bipy)(CN)4]- as a versatile building block for the design of heterometallic systems: synthesis, crystal structure, and magnetic properties of PPh4[FeIII(bipy)(CN)4] H2O, [{FeIII(bipy)(CN)4}2MII(H2O)4] 4H2O, and [{FeIII(bipy)(CN)4}2ZnII] 2H2O [bipy ) 2,2-Bipyridine; M ) Mn and Zn]. Inorg. Chem. 41, 818826. (43) Cornell, W. D., Cieplak, P., Bayly, C. I., Gould, I. R., Merz, K. M., Jr., Ferguson, D. M., Spellmeyer, D. C., Fox, T., Caldwell, J. W., and Kollman, P. A. (1995) A second generation force eld for the simulation of proteins, nucleic acids, and organic molecules. J. Am. Chem. Soc. 117, 51795197. (44) Kurnikov, I. personal communication. (45) Terrettaz, S., Becka, A. M., Traub, M. J., Fettinger, J. C., and Miller, C. J. (1995) -Hydroxythiol monolayers at Au electrodes. 5. Insulated electrode voltammetric studies of cyano/bipyridyl iron complexes. J. Phys. Chem. 99, 1121611224.

Outer Sphere Effects of Iron(II) Complexes (46) Gutmann, V. (1976) Empirical parameters for donor and acceptor properties of solvents. Electrochim. Acta 21, 661670. (47) Bryant, G. M., Fergusson, J. E., and Powell, H. K. J. (1971) Charge-transfer and intraligand electronic spectra of bipyridyl complexes of iron, ruthenium, and osmium. I. Bivalent complexes. Aust. J. Chem. 24, 25773. (48) Machczynski, M. C., Gray, H. B., and Richards, J. H. (2002) An outer-sphere hydrogen-bond network constrains copper coordination in blue proteins. J. Inorg. Biochem. 88, 375380. (49) Worrall, J. A. R., Schlarb-Ridley, B. G., Reda, T., Marcaida, M. J., Moorlen, R. J., Wastl, J., Hirst, J., Bendall, D. S., Luisi, B. F., and Howe, C. J. (2007) Modulation of heme redox potential in the cytochrome c6 family. J. Am. Chem. Soc. 129, 9468 9475. (50) Shifman, J. M., Gibney, B. R., Sharp, R. E., and Dutton, P. L. (2000) Heme redox potential control in de novo designed fourR-helix bundle proteins. Biochemistry 39, 1481314821. (51) Schutz, C. N., and Warshel, A. (2001) What are the dielectric \constants\ of proteins and how to validate electrostatic models? Proteins: Struct., Funct., Genet. 44, 400417. (52) Simonson, T., and Perahia, D. (1995) Internal and interfacial dielectric properties of cytochrome c from molecular dynamics in aqueous solution. Proc. Natl. Acad. Sci. U.S.A. 92, 1082 1086. (53) Rodgers, K. K., and Sligar, S. G. (1991) Surface electrostatics, reduction potentials, and the internal dielectric constant of proteins. J. Am. Chem. Soc. 113, 94199421. (54) Green, N. M. (1975) Avidin. AdV. Protein Chem. 29, 85 133. (55) Shraydeh, B. F., and Zatar, N. (1994) Donor and acceptor number effects on the solvatochromic behavior of bis(2,2bipyridyl)biscyanoiron(II) in binary aqueous mixtures. Monatsh. Chem. 125, 655659. (56) Toma, H. E., and Takasugi, M. S. (1983) Spectroscopic studies of preferential and asymmetric solvation in substituted cyanoiron(II) complexes. J. Solution Chem. 12, 547561. (57) (a) Eisenberg-Domovich, Y., Pazy, Y., Nir, O., Raboy, B., Bayer, E. A., Wilchek, M., and Livnah, O. (2004) Structural elements responsible for conversion of streptavidin to a pseudoenzyme. Proc. Natl. Acad. Sci. U.S.A. 101, 59165921. (b) Pazy, Y., Kulik, T., Bayer, E. A., Wilcheck, M., and Livnah, O. (2002) Ligand exchange between proteins. J. Biol. Chem. 277 (34), 2089230900. (58) Lo, K. K.-W., and Hui, W.-K. (2005) Design of rhenium(I) polypyridine biotin complexes as a new class of luminescent probes for avidin. Inorg. Chem. 44, 19922002.

Bioconjugate Chem., Vol. 20, No. 10, 2009 1939 (59) Lo, K. K.-W., Hui, W.-K., and Ng, D. C.-M. (2002) Novel rhenium(I) polypyridine biotin complexes that show luminescence enhancement and lifetime elongation upon binding to avidin. J. Am. Chem. Soc. 124, 93449345. (60) Collot, J., Gradinaru, J., Humbert, N., Skander, M., Zocchi, A., and Ward, T. R. (2003) Articial metalloenzymes for enantioselective catalysis based on biotin-avidin. J. Am. Chem. Soc. 125, 90309031. (61) Jhaveri, S. D., Trammell, S. A., Lowy, D. A., and Tender, L. M. (2004) Electron conduction across electrode-immobilized neutravidin bound with biotin-labeled ruthenium pentaamine. J. Am. Chem. Soc. 126, 65406541. (62) Pazy, Y., Raboy, B., Matto, M., Bayer, E. A., Wilchek, M., and Livnah, O. (2003) Structure-based rational design of streptavidin mutants with pseudo-catalytic activity. J. Biol. Chem. 278, 71317134. (63) Yokoyama, K., Leigh, B. S., Sheng, Y., Niki, K., Nakamura, N., Ohno, H., Winkler, J. R., Gray, H. B., and Richards, J. H. (2008) Electron tunneling through Pseudomonas aeruginosa azurins on SAM gold electrodes. Inorg. Chim. Acta 361, 1095 1099. (64) Yue, H., and Waldeck, D. H. (2006) Understanding interfacial electron transfer to monolayer protein assemblies. Curr. Opin. Solid State Mater. Sci. 9, 2836. (65) Tarlov, M. J., and Bowden, E. F. (1991) Electron-transfer reaction of cytochrome c adsorbed on carboxylic acid terminated alkanethiol monolayer electrodes. J. Am. Chem. Soc. 113, 1847 1849. (66) Feng, Z. Q., Imabayashi, S., Kakiuchi, T., and Niki, K. (1997) Long-range electron-transfer reaction rates to cytochrome c across long- and short-chain alkanethiol self-assembled monolayers: Electroreectance studies. J. Chem. Soc., Faraday Trans. 93, 13671370. (67) Clark, R. A., and Bowden, E. F. (1997) Voltammetric peak broadening for cytochrome c/alkanethiolate monolayer structures: dispersion of formal potentials. Langmuir 13, 559565. (68) Chobot, S. E., Hernandez, H. H., Drennan, C. L., and Elliott, S. J. (2007) Direct electrochemical characterization of archaeal thioredoxins. Angew. Chem., Int. Ed. 46, 41454147. (69) Ye, T., Kaur, R., Wen, X., Bren, K. L., and Elliott, S. J. (2005) Redox properties of wild-type and heme-binding loop mutants of bacterial cytochromes c measured by direct electrochemistry. Inorg. Chem. 44, 89999006. (70) Armstrong, F. A., Hill, H. A. O., and Walton, N. J. (1988) Direct electrochemistry of redox proteins. Acc. Chem. Res. 21, 407413.
BC900270A