20073
2006, 110, 20073-20076 Published on Web 09/26/2006
Comparison of Intra- vs Intermolecular Long-Range Electron Transfer in Crystals of Ruthenium-Modified Azurin Cristian Graˇ dinaru and Brian R. Crane* Department of Chemistry and Chemical Biology, Cornell UniVersity, Ithaca, New York 14853 ReceiVed: July 13, 2006; In Final Form: September 4, 2006
Selective metal-ion incorporation and ligand substitution are employed to control whether electrons tunnel over intra- or intermolecular separations in crystals of P. aeruginosa azurin modified with Ru-polypyridine complexes. Cu1+-to-Ru3+ electron transfer (ET) across a specific protein-protein interface in the crystal lattice has a time constant 5-10 times longer than ET between the same donor and acceptor within a single protein (τET ) 5 vs 0.5-1.0 µs). Slower intermolecular ET agrees well with a longer distance between redox centers across the inter-protein (18.9 Å) compared to the intra-protein separation (17.0 Å) and indicates that the closest donor/acceptor pair dominates crystal ET. Lowering the crystal pH accelerates inter-protein ET (τET ) 1.0 µs) but not intra-protein ET. Faster inter-protein ET likely results from a pH-induced peptide bond flip that perturbs hydrogen bonding in the path between Ru and Cu centers on adjacent molecules.
Introduction Regulation of biological electron transfer (ET) largely resides in the chemical structures of proteins and their ability to form specific complexes.1 The role of protein conformational change in modulating long-range ET is an open and important question in systems that range from energy transducing membrane complexes to enzymes that metabolize DNA.1 Moreover, the ability of structural changes to switch “on” and “off” reactivity not only concerns biological mechanisms but also has implications for the design of molecular electronics and other nanoscale devices.2 The blue-copper protein azurin is a long-standing model system for investigating the physical basis of long-range protein ET. Gray and co-workers have extensively investigated bluecopper redox properties, electron-tunneling pathways, and amino acid radical formation in azurin by attaching Ru (Os, Re) photochemical probes to specific histidine residues on the protein surface.3 In a different approach, Farver and Pecht have radiolytically generated disulfide radicals to transfer electrons to the azurin copper center.4 Azurin has also been the subject of pioneering electrochemical studies that have probed singlemolecule behavior and provided a basis for developing nanoscale bioelectronic devices.5 To better understand the impact of variable protein structure on both intra- and intermolecular ET, we have developed methods for studying such reactions in single crystals, which fix the association modes of molecules in an environment that can be subjected to structural analysis by X-ray diffraction.6 Unlike ET between soluble factors, the three-dimensional crystal lattice generates multiple fixed spatial relationships for the same donor (D) and acceptor (A). Thus, ET rate measurements across different separations in the same crystal lattice have the potential * To whom correspondence should be addressed. Phone: (607) 2548634. Fax: (607) 255-1248. Email:
[email protected].
10.1021/jp0644309 CCC: $33.50
to probe how structure and proximity tune reactivity. With this motivation, we have generated a crystal system of Ru-modified azurin in which photoinduced ET reactions can be driven specifically over two distinct D/A separations. Within these crystals, structural perturbations triggered by pH changes affect the composition and ET properties across one separation, but not the other. Sample Preparation. Pseudomona aeruginosa azurin was recombinantly expressed in E. coli and purified as previously described.7 The two azurin (Az) variants were prepared following established protocols.3b-d Briefly, the protein was incubated overnight with either 10 mM CuSO4 or 10 mM ZnSO4 in 100 mM acetate buffer, pH 4.5. The filtered supernatants were concentrated to 3-5 mM, dialyzed into 20 mM Tris-HCl pH 8.5, and then further purified by elution from a HiPrep QXL 16/10 anion exchange column against a gradient of 20 mM TrisHcl, 200 mM NaCl, pH 8.5. His83 of both CuAz and ZnAz was labeled by incubation with equimolar amounts of Ru2+(bpy)2(CO3)‚4H2O in 250 mM NaHCO3, pH 8.3. Labeled azurins were exchanged into 15 mM acetate buffer, pH 4.2, and then purified by elution from a Resource S cation-exchange column against a gradient of 300 mM acetate buffer, pH 4.5. Last, the ZnAzRu sample was dialyzed against 400 mM imidazole-HCl pH 7.0 for 4-5 days and luminescence followed at 630 nm as a marker of imidazole binding to Ru. Crystal Structure Determination. Cocrystals of [Ru2+(bpy)2(Im)(His83)]AzZn(II) and [Ru2+(bpy)2(OH2)(His83)]AzCu(II) (space group P21212, cell dimensions 70.14 × 58.77 × 53.17 Å3, two molecules per asymmetric unit) grew from 2 µL drops made from an equimolar mixture of 30 mg/mL Rumodified azurins in 25 mM HEPES pH 7.5 and reservoir. The drops were equilibrated against 500 mL of reservoir containing 20% PEG molecular weight 4000, 100 mM LiNO3, and 100 mM imidazole pH 7.0. Diffraction data (30.0-1.5 Å resolution, 84.2% complete, Rsym ) Σj|Ij - 〈I〉 |/ΣΣjIj ) 6.8%; overall © 2006 American Chemical Society
20074 J. Phys. Chem. B, Vol. 110, No. 41, 2006 signal-to-noise ) 〈I〉/sI ) 27.8) were collected on a Quantum210 CCD (Area Detector Systems Corporation) at the Cornell High Energy Synchrotron Source, beamline F2 (0.945 Å), and processed with DENZO.8 The structure was determined by molecular replacement with AMORE9 using a probe derived from the structure of Ru2+(2,2′-bipyridine)2(im)(His83)-AzCu(II) (pdb code 1BEX).3d Rigid-body, positional, and thermal factor refinement was carried out with CNS10 amidst rounds of manual rebuilding, incorporation of both ∆ and Λ enantiomers of Ru2+(bpy)2(im or OH2), and water placement with XFIT.11 In the final model (1.5 Å resolution, R-factor (working set) ) 21.4%; R-free ) 23.4%) all residues have favored backbone dihedral angles. Stereochemical restraints were removed from the copper ligand bonds in the later stages of refinement. Anomalous diffraction data were collected at the zinc and copper absorption edges and metal-ion occupancies refined against these data with MADPHSREF.12 Transient Crystal Spectroscopy. Crystals for transient absorption experiments were grown anaerobically and sealed inside 1 × 1 mm flat-faced quartz capillaries. Transient spectroscopy was carried out with a 75-W Xe-arc lamp probe light source (Thermo Oriel) pulsed with a laser diode driver to increase light intensity over the time window immediately following excitation. Probe light was focused to a 20-µmdiameter spot and collected 50 cm from the sample to reduce the intensity of sample luminescence. A Nd:YAG pumped optical parametric oscillator (Opolette) provided 8-ns 0.8-mJ pulses at 490 nm for excitation. Light intensity was monitored at 430 and 630 nm through collecting optics, which were coupled to an Oriel monochromator (1200 L/mm grating) attached to a Hamamatsu photomultiplier tube. A 500-MHzbandwidth Tektronix oscilloscope was used for digital conversion (9-bit resolution). Data were averaged from at least three different samples (∼128 excitation shots/sample). For the mixed-metal crystals normalized transmittance (r(t)) was fit to a function that accounts for absorbance changes associated with metal-ion redox state and luminescence from the *Ru2+(bpy)2(Im)(His), whose lifetime will depend on position in the crystal lattice. Total passed light intensity (Iout)
r(t) )
Iout(t) lim Iout(t)
) 1 + C1{exp[Ao ln 10(e-t/τET - e-t/τq)] -
tf∞
1} +
C2 -t/τq + R e-t/τnq) (1) (e 1+R
is due to photons from the two *Ru2+ emission processes (q, quenched; nq, nonquenched sites), absorbance from nonparticipating (np) Cu2+ sites, time-dependent loss and recovery of absorbance from participating Cu2+ sites: A0[1 - exp(-t/τET) + exp(-t/τq)], and some probe light leakage around crystals. C1 equals the portion of light passed or generated by the crystal; C2 equals the emission light at t ) 0 divided by Iout(t ) ∞); and R ) I(0)nq/I(0)q, or the fraction of *Ru2+ sites not quenched by neighboring Cu2+. C2 values reflect a number of variable parameters which include crystal size, crystal orientation, minor Cu1+ content, probe light intensity, and the ratio of quenched to nonquenched sites in the mixed-metal crystals. In instances where both sample luminescence and absorbance changes contribute to the collected light, eq 1 can be approximated by a multiexponential function only when τq is small compared to τET and the luminescence intensity is much less than the probe light passed through the sample. To reduce the number of parameters in the fits, the quenching rate constants from participating and nonparticipating sites were fixed at their
Letters
Figure 1. (A) Photochemistry of Ru-azurin mixed-metal crystals. (B,C) Strategy for controlling intra-protein (B) vs inter-protein (C) ET in Ru-azurin crystals.
values determined from luminescence measurements of homogeneous Cu2+ and homogeneous Zn2+ crystals. Luminescence lifetimes were recorded in the absence of probe light and deconvoluted from the instrument response function by Fourier methods. Consistent with all Cu2+ sites participating with neighboring *Ru2+ in the homogeneous crystals, terms involving τnq were unnecessary to fit the corresponding transmittance data. Emission terms were unnecessary to fit data taken at 430 nm (an isosbestic point for *Ru). Results and Discussion In the Ru-modified azurin system, excitation of the Ruphotosensitizer MLCT results in rapid reduction of the protein Cu2+ center (characterized by increased transmission at 630 nm) and concomitant formation of Ru3+ (characterized by increased transmission at 430 nm)3a (Figure 1a). The charge-separated state (Ru3+/Cu1+) then returns to the Ru2+/Cu2+ ground state by an activationless charge recombination reaction.3b,h,6a Azurins modified at His83 with Ru, Re, or Os photosensitizers give very similar ET kinetics in solution and in single crystals.6a Thus, despite the extensive intermolecular contacts of the crystal lattice, the reactivity is consistent with ET within one molecule over the 17 Å D/A separation (Figure 2a). However, the Ruazurin lattice provides an alternative D/A separation over an inter-protein contact that is also within range to support rapid electron tunneling (18.9 Å, Figure 2b). To selectively favor the intermolecular reaction, we employed a crystal doping strategy inspired by previous work on crystals composed from mixtures of cytochrome c containing either Zn- or Fe-porphyrins.6b Examination of the Ru-azurin lattice suggested that the unique lattice site of the asymmetric unit had the potential to accommodate slightly modified molecules, provided molecular contacts were not altered. We thus cocrystallized two different Ru-azurin variants: one containing an active donor, Ru2+(bpy)2(im)(His83) (where bpy is 2,2′-bipyridine and im is imidazole), but an inactive acceptor (Zn2+), and the other containing an inactive donor, Ru2+(bpy)2(OH2)(His83), but an active acceptor (Cu2+). The *Ru-aquo complex has too short a lifetime (
