SPECTROSCOPIC AND THEORETICAL STUDIES OF T1 CU, CUA AND

CYTOCHROME C: GEOMETRIC AND ELECTRONIC STRUCTURE CONTRIBUTIONS

TO ELECTRON TRANSFER

A DISSERTATION SUBMITTED TO THE DEPARTMENT OF CHEMISTRY AND THE COMMITTEE ON GRADUATE STUDIES OF STANFORD UNIVERSITY IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY

Xiangjin Xie January 2010

© 2010 by Xie Xiangjin. All Rights Reserved. Re-distributed by Stanford University under license with the author.

This work is licensed under a Creative Commons Attribution- Noncommercial 3.0 United States License. http://creativecommons.org/licenses/by-nc/3.0/us/

This dissertation is online at: http://purl.stanford.edu/cq701yv2053

ii I certify that I have read this dissertation and that, in my opinion, it is fully adequate in scope and quality as a dissertation for the degree of Doctor of Philosophy.

Edward Solomon, Primary Adviser

I certify that I have read this dissertation and that, in my opinion, it is fully adequate in scope and quality as a dissertation for the degree of Doctor of Philosophy.

T Stack

I certify that I have read this dissertation and that, in my opinion, it is fully adequate in scope and quality as a dissertation for the degree of Doctor of Philosophy.

Robert Waymouth

Approved for the Stanford University Committee on Graduate Studies. Patricia J. Gumport, Vice Provost Graduate Education

This signature page was generated electronically upon submission of this dissertation in electronic format. An original signed hard copy of the signature page is on file in University Archives.

iii

ABSTRACT

Metal sites that are known to be involved in biological electron transfer (ET) include

Type 1 Copper (T1 Cu), CuA, cytochromes, and the 1-, 2-, 3-, and 4-iron sulfur centers (rubredoxin, ferredoxins, and high potential iron-sulfur proteins (HiPIPs)). These ET sites generally exhibit unusual spectroscopic features reflecting novel geometric and

electronic structures that contribute to function. My focuses are on T1 Cu, CuA, cytochrome c proteins utilizing a wide-range of spectroscopies combined with density functional calculations to understand active site electronic structures, the origin of their geometric structures, and possible contributions to function. Five major achievements are: 1) defined the temperature dependent absorption feature of T1 Cu site in nitrite reductase (NIR) and provided insight into the entatic/rack nature of the blue Cu site in plastocyanin; 2) addressed the interesting absorption features of the T1 Cu site in P. pantotrophus pseudoazurin and demonstrated the spectral probes of the weak axial ligation in metalloprotein; 3) resolved a two-state issue in the mixed-valence binuclear

CuA centers in cytochrome c oxidases (CcO) and nitrous oxide reductases (N2O) by a combination of density functional calculations and spectroscopy analyses, and evaluated

proteins role in CuA sites and their contributions to ET function; 4) determined that the

Cu-Cu interaction in CuA keeps the site delocalized even upon loss of a Histidine (NHis) ligand due to protonation, and defined the contribution of σ delocalization to efficient

ET; 5) investigated the nature of the Fe-SMet bond in ferricytochrome c.

(1) Thermodynamic Equilibrium between Blue and Green Copper Sites and the Role of the Protein in Controlling Function Spectroscopies and density functional theory calculations indicate that there are large temperature-dependent absorption spectral changes present in green nitrite reductases (NiRs) due to a thermodynamic equilibrium between a green and a blue type 1 (T1) copper site. The axial methionine (Met) ligand is unconstrained in the oxidized NiRs, which results in an enthalpically favored (ΔH ≈ 4.6 kcal/mol) Met-bound green copper site at low temperatures, and an entropically favored (TΔS ≈4.5 kcal/mol, at room

iv temperature) Met-elongated blue copper site at elevated temperatures. In contrast to the NiRs, the classic blue copper sites in plastocyanin and azurin show no temperature- dependent behavior, indicating that a single species is present at all temperatures. For these blue copper proteins, the polypeptide matrix opposes the gain in entropy that would be associated with the loss of the weak axial Met ligand at physiological temperatures by constraining its coordination to copper. The potential energy surfaces of Met binding indicate that it stabilizes the oxidized state more than the reduced state. This provides a mechanism to tune down the reduction potential of blue copper sites by > 200 mV.

(2) Variable Temperature Spectroscopic Study on Pseudoazurin: Effects of Protein Constraints on the Blue Cu Site. The T1 copper site of Paracoccus pantotrophus pseudoazurin exhibits significant absorption intensity in both the 450 and 600 nm regions. These are σ and π SCys to Cu2+ charge transfer (CT) transitions. The temperature dependent absorption, EPR, and resonance Raman (rR) vibrations enhanced by these bands indicate that a single species is present at all temperatures. This contrasts the temperature dependent behavior of the T1 center in nitrite reductase, which has a thioether ligand that is unconstrained by the protein. The lack of temperature dependence in the T1 site in pseudoazurin indicates the presence of a protein constraint similar to the blue Cu site in plastocyanin where the thioether ligand is constrained at 2.8 Å. However, plastocyanin exhibits only π CT. This spectral difference between pseudoazurin and plastocyanin reflects a coupled distortion of the site where the axial thioether in pseudoazurin is also constrained, but at a shorter 2+ Cu–SMet bond length. This leads to an increase in the Cu –SCys bond length, and the site undergoes a partial tetragonal distortion in pseudoazurin. Thus, its ground state 2+ wavefunction has both σ and π character in the Cu –SCys bond.

(3) The Two State Issue in the Mixed-Valence Binuclear CuA Center in Cytochrome

c Oxidase and N2O Reductase

For the CuA site in the protein, the ground and lowest energy excited-states are σu* and

πu, respectively, denoting the types of Cu-Cu interactions. EPR data on CuA proteins show a low g|| value of 2.19 deriving from spin-orbital coupling between σu* and πu, -1 which requires an energy gap between σu* and πu of 3000−4500 cm . On the other hand,

v from paramagnetic NMR studies, it has been observed that the first excited-state is thermally accessible and the energy gap between the ground state and the thermally accessible state is 350 cm-1. This study addressed this apparent discrepancy and evaluated the roles of the two electronic states, σu* and πu, in electron transfer (ET) of

CuA. The potential energy surface calculations show that both NMR and EPR results are

consistent within the electronic/geometric structure of CuA. The anti-Curie behavior

observed in paramagnetic NMR studies of CuA results from the thermal equilibrium

between the σu* and πu states, which are at very close energies in their respective

equilibrium geometries. Alternatively, the EPR g-value analysis involves the σu* ground state in the geometry with a short dCu-Cu where the πu state is a Frank−Condon excited- state with the energy of 3200 cm-1. The protein environment plays a role in maintaining

CuA in the σu* state as a lowest-energy state with the lowest reorganization energy and high-covalent coupling to the Cys and His ligands for efficient intra- and intermolecular ET with a low-driving force.

(4) Perturbations to the Geometric and Electronic Structure of the CuA Site: Factors that Influence Delocalization and their Contributions to Electron Transfer Using a combination of electronic spectroscopies and DFT calculations, the effect of pH perturbation on the geometric and electronic structure of the CuA site has been defined. Descriptions are developed for high pH (pH = 7) and low pH (pH = 4) forms of

CuA azurin and its H120A mutant which address the discrepancies concerning the extent of delocalization indicated by multifrequency EPR and ENDOR data. Our resonance Raman and MCD spectra demonstrate that the low pH and H120A mutant forms are essentially identical and are the perturbed forms of the completely delocalized high pH

CuA site. However, in going from high pH to low pH, a seven-line hyperfine coupling pattern associated with complete delocalization of the electron (S = 1/2) over two Cu

coppers (ICu = 3/2) changes into a four-line pattern reflecting apparent localization. DFT calculations show that the unpaired electron is delocalized in the low pH form and reveal that its four-line hyperfine pattern results from the large EPR spectral effects of 1% 4s orbital contribution of one Cu to the ground-state spin wave function upon protonative loss of its His ligand. The contribution of the Cu−Cu interaction to electron

vi delocalization in this low symmetry protein site is evaluated, and the possible functional significance of the pH-dependent transition in regulating proton-coupled electron transfer in cytochrome c oxidase is discussed.

(5) The Fe-Smet Bond in Ferricytochrome c DFT calculations calibrated with experiment data were used to define the nature of the

Fe-SMet bond in ferricytochrome c. This is inspired by the studies of NiR.

vii

ACKNOWLEDGMENTS

To my research advisor Prof. Edward I. Solomon, whose deep insight in science, great passion in research, and thoughtful care in heart have been driving and elevating my work into a doctoral thesis. I am deeply impressed by his knowledge on numerous areas of research, and benefit tremendously from him. I feel highly grateful for his constant support and mentoring over the years. I am thankful to my dissertation committee members for their help and support taking time out of their busy schedules to read my thesis and be on my defense committee. I thank my collaborators Prof. Keith Hodgson, Prof. Britt Hedman, Dr. Ritimukta

Sarangi and Munzarin Fatema Qayyum on CuA projects. Thanks are also due to Prof. Isabel Moura, Dr. Sofia R. Pauleta, Prof. Sun Un, Prof. Charles P. Scholes, Prof. Yi Lu, Prof. Jame A. Fee and the students in their groups who provided the samples and helpful discussions on various aspects of the projects. Thanks are also due to the past and present members of Solomon and Hodgson group. Thank Peng for introducing me into Solomon group. Thank Lipika for training me to run MCD and PES experiments. Thank Jungjoo and Barry from whom I learned a lot about EPR. Thank Mike Vance for introducing me into Raman experiments. Thank Takehiro and Ashley for Ti-sapphire laser setup. Thank S. I. Gorelsky and Marcus for their advices on DFT calculations. Thank Chrissy, Lei and Jordi for maintaining the group computer cluster and their prompt help in getting my troubled computers running. Thank Dey and Som for collaborating on thermodynamic studies of T1 Cu site. Thank Ryan for taking over my research projects, running some experiments for me in pseudoazurin and cyt c studies, and proof-reading my thesis. Thank Prof. Bozhen Chen for her help in cyt c studies. Thank Marina, Pin-pin, and Monita for the help in running MCD. Thank TJ, Julia, and Adam for being fellow classmates and the help over six years. Thank Mike Clay, Mike Neidig, Bryan, Adrienne, Pieter, Ken, Lei, Mat, Dave, Christian, Mads, Caleb, Ray Ana, Ning, Sam, Shaun, Esther and other group members socializing with me. Special thanks to all my friends in Prof. Dai’s group for enjoying Stanford life with me.

viii Thanks are also due to Prof. Piero Pianetta, Curtis Troxel, Dan Brehmer, Dr. Yun Sun, Dr. Zhi Liu, Dr. Shiyu Sun, Dr. Serena Debeer George, Dr. Rosalie Hocking, and staff members of SSRL for their technical advices and assistances at beam lines, where I spent a lot of time doing UHV experiments using synchrotron radiation at SSRL. I am also thankful to Prof. T. Daniel P. Stack and his group members for a pleasant stay in his group during my first year at Stanford. Thanks are also due to the department student service officer Roger Kuhn, who makes my everyday life as an international graduate student a lot easier. Thanks are also due to Carmen, Clara and Elayna for their excellent administrative assistance in the Solomon group. I must thank my parents, parents-in-law, brothers, brothers-in-law, sister and sister-in- law for their encouragement and support, and my lovely son, William, for his cooperation in my thesis writing. Special thanks to my beloved wife, Li, for her understanding, support and trust over the years.

ix TABLE OF CONTENTS

Abstract ------iv

Acknowledgements ------viii

Table of Contents ------x

List of Tables ------xv

List of Figures ------xvii

1 Background

1.1 Introduction ------2

1.2 T1 Cu Sites ------4

1.3 CuA Sites ------7

1.4 Cytochorme c ------16

1.5 Overview of Thesis ------18

1.6 References ------20

2 Thermodynamic Equilibrium between Blue and Green Copper Sites and the Role of the Protein in Controlling Function

2.1 Introduction ------29

2.2 Experimental Methods ------30

2.2.1 Materials ------30

2.2.2 Spectroscopic Studies ------30

2.2.3 Computational Details ------31

2.3 Results ------29

2.3.1 Low Temperature Spectroscopy ------32

2.3.2 Room Temperature Spectroscopy ------35

2.3.3 Thermodynamics ------37

2.4 Analysis ------39

2.4.1 Oxidized Site ------39

2.4.2 Reduced Site ------43

2.5 Discussions ------45

x 2.6 Acknowledgements ------47

2.7 References ------47

3 Variable Temperature Spectroscopic Study on Pseudoazurin: Effects of Protein Constraints on the Blue Cu Site

3.1 Introduction ------54

3.2 Experimental Methods ------55

3.2.1 Protein Purification ------55

3.2.2 Absorption and Magnetic Circular Dichroism Spectroscopy ------56

3.2.3 Electron Paramagnetic Resonance ------56

3.2.4 Resonance Raman ------57

3.2.5 DFT Calculations ------57

3.3 Results and Analysis ------58

3.3.1 Low Temperature Electronic Spectroscopy ------58

3.3.2 Low Temperature Resonance Raman Spectroscopy ------61

3.3.3 Electron Paramagnetic Resonance ------62

3.3.4 Variable Temperature Absorption Spectra ------65

3.3.5 Variable Temperature Resonance Raman Spectroscopy ------66

3.4 Discussion ------67

3.5 Acknowledgments ------69

3.6 References ------69

4 The Two State Issue in the Mixed-Valence Binuclear CuA Center in

Cytochrome c Oxidase and N2O Reductase

4.1 Introduction ------75

4.2 Experimental Details ------76

4.2.1 Sample Preparation ------76

4.2.2 Absorption and MCD Measurements ------76

4.2.3 Computational Details ------77

4.3 Results and Analysis ------79

xi 4.4 Acknowledgements ------80

4.5 Supporting Information ------80

4.6 References ------89

5 Perturbations to the Geometric and Electronic Structure of the CuA Site: Factors that Influence Delocalization and their Contributions to Electron Transfer

5.1 Introduction ------96

5.2 Experiments ------98

5.2.1 Sample Preparation ------98

5.2.2 Electron Paramagnetic Resonance ------98

5.2.3 Electronic Absorption and MCD Spectroscopies ------98

5.2.4 Resonance Raman Spectroscopy ------99

5.2.5 XAS Data Acquisition ------99

5.2.6 DFT Calculations ------100

5.3 Results and Analysis ------101

5.3.1 Spectroscopy ------101

5.3.1.1 Electron Paramagnetic Resonance ------101

5.3.1.2 Electronic Absorption and MCD Spectroscopy ------103

5.3.1.3 Resonance Raman Spectroscopy ------106

5.3.1.4 EXAFS Studies ------107

5.3.2 DFT Calculations ------110

5.3.2.1 Optimized Structures ------110

5.3.2.2 Ground State Electronic Structure ------112

(i) Ground State Wavefunction ------112

(ii) Cu Hyperfine Coupling ------113

5.3.2.3 Excited States ------115

(i)TD-DFT ------116

(ii) The ψ → ψ* MCD C-term ------118

5.4 Discussion ------120

xii 5.4.1 Spectral Probes of Delocalization ------120

5.4.2 Factors Effecting Delocalization ------121 5.4.3 Correlation of Electronic Structure to ET and its Regulation by

+ 122 [H ] ------

(i) Reorganization Energy ------122

(ii) Superexchange Pathways ------123

(iii) Possible Contribution to Function ------125

5.5 Summary ------125

5.6 Acknowledgements ------126

5.7 Supporting Information ------127

5.8 References ------133

6 The Fe-Smet Bond in Ferricytochrome c

6.1 Introduction ------142

6.2 Calculation Details ------143

6.3 Results and Analysis ------144

6.3.1 DFT Calculations ------144

6.3.1.1 Geometric Structure ------144

6.3.1.2 Energies ------145

6.3.2 TD-DFT ------147

6.4 Discussion ------153

6.5 References ------154

APPENDIX Oxygen Binding of Water-Soluble Cobalt Porphyrins in Aqueous Solution

A.1 Introduction ------158

A.2 Experiments ------159

A.3 Results and Analysis ------159

A.4 Acknowledgements ------163

A.5 Supporting Information ------164

xiii A.5.1 Preparation of compound 2 ------164

A.5.2 Preparation of compound 1Co ------165

A.5.3 Preparation of compound 3H2 ------165

A.5.4 Preparation of compound 3H2 ------166

A.5.5 Preparation of compound 1Co ------166

A.5.6 Preparation of compound 3Co (method B) ------166

A.5.7 Preparation of compound 1Co from 3Co (method B) ------167 A.5.8 X-band (9.37GHz) EPR Spectra of 1Co Buffer Solutions (C >

1mM) Under Different Conditions (recorded at 4mW and 77K) -- 167

A.6 References ------170

xiv LIST OF TABLES CHAPTER 2 Table 2.1 Calculated free energies for loss of an axial Met ligand (constrained

at different Cu-SMet distances) from an oxidized copper site at room

temperature. ------43 Table 2.2 Calculated free energies for loss of an axial Met ligand (constrained

at different Cu-SMet distances) from a reduced copper site at room

temperature. ------44

CHAPTER 3 Table 3.1 Low temperature (5 K) electronic spectroscopic parameters for

pseudoazurin and plastocyanin. ------61 Table 3.2 EPR parameters of P. pantotrophus pseudoazurin and a comparison

of those parameters to other blue copper sites. ------64 Table 3.3 Calculated EPR parameters for Cu sites in plastocyanin and

pseudoazurin by ORCA 2.6 ------64 Table 3.4 Cu characters in β-LUMO by Löwdin reduced orbital population

analysis (%). ------65

CHAPTER 4

Table 4.S1 The g-values and energy gap between the πu and σu* states from EPR, the energy of the thermally accessible state probed by NMR for

CuA proteins from various resources. ------83 Table 4.S2 Bond distances, NPA-derived atomic charges and atomic spin

densities, and the total Cu, Sthiolate and Nimidazole contributions to β-

spin LUMO of the CuA models. ------85 Table 4.S3 Bond energy decomposition analysis (kcal mol-1) for the 30-atom

CuA model. ------86

Table 4.S4 Reorganization energies (λ) of the ET pathway with the CuA center in

the σu* and πu ground states. ------86

xv CHAPTER 5

Table 5.1 EPR parameters of the high and low pH forms of CuA azurin, and its

H120A mutant. ------103

Table 5.2 EXAFS Least Squares Fitting Results for WT CuA at high pH and

low pH and for the H120A mutant. ------109

Table 5.3 The bond distances in the X-ray crystal structure of CuA azurin, DFT

geometry optimized 96-atom and 97-atom models. ------111 Table 5.4 Mulliken atomic spin densities of 96-atom and 97-atom models from

Gaussian 03 calculations. ------113 Table 5.5 Löwdin population analyses of Cu Characters in β-LUMO of 96-atom

and 97-atom models from Orca 2.5 calculations. ------113 Table 5.6 Calculated Cu hyperfine Coupling Constants of the 96-atom and 97- -4 -1 atom models with Program ORCA 2.5 (10 cm ). ------114

Table 5.7 CuA to Heme a ET rate comparison between the high and low pH

forms of CuA. ------125

Table 5.S1 Complete EXAFS Least Squares Fitting Results for WT CuA at pH6. 130

Table 5.S2 Complete EXAFS Least Squares Fitting Results for WT CuA at pH4. 131 Table 5.S3 Complete EXAFS Least Squares Fitting Results for H120A mutant. 132

CHAPTER 6 Table 6.1 Experimental and calculated bond distances in heme sites without or with a Met ligand. ------145 Table 6.2 Calculated electronic energies for the loss of the axial Met ligand. --- 146 Table 6.3 Calculated free energies for the loss of the axial Met ligand 147 (T=298K). Table 6.4 Calculated d-d transitions for in Fe-P-Im-DMS and Fe-P-Im-DMS. - 148 Table 6.5 The transitions near near the Q-band in Fe-P-Im-DMS. ------149

Table 6.6 The transitions near near the Q-band in Fe-P-(Im)2. ------149

xvi LIST OF FIGURES

CHAPTER 1

Figure 1.1 Electron transfer metal active sites in bioinorganic chemistry. ------3 Figure 1.2 (A) Absorption spectra of plastocyanin (left ε scale) and normal Cu2+ 2- complex D4h [CuCl4] (right ε scale). (B) X-band EPR spectrum of 2- plastocyanin (blue) and D4h [CuCl4] (Black). ------4 Figure 1.3 S K-edge and Cu L-edge XAS as probes of ligand-to-metal covalency. (A) Comparison of S K-edge spectra of blue copper and

CuA (bottom), energy level diagram depicting S 1s to RAMO transition (top), β2 is the amount of S p character in the RAMO. (B)

Comparison of Cu L-edge spectra of blue copper and CuA (bottom),

energy level diagram depicting Cu 2p to RAMO transition (top). ------5 Figure 1.4 Low temperature absorption and magnetic circular dichrisom spectra of a series of T1 Cu proteins. Low-temperature absorption spectra

show the redistribution of spectral intensity from SCys π to σ CT bands. Redox active orbitals calculated with SCF-Xα-SW showing rotation from a π type interaction in plastocyanin (left top) to σ (+π ) mixture in nitrite reductase (left bottom). MCD spectra show a shift of the LF transitions to higher energy for nitrite reductase relative to 6 plastocyanin indicating a tetragonal distortion. ------Figure 1.5 Continuum of coupled tetragonal distortions in a series of perturbed blue copper proteins. Crystal structures show contraction of the Cu-

SMet bond associated with elongation of the Cu-SCys bond and

tetragonal εu -like mode Jahn-Teller distortion. ------7

Figure 1.6 Scheme of electron and proton flow in cytochrome c oxidase. ------8

Figure 1.7 Q-band (red) and X-band (Black) EPR spectra of CuA protein. ------9 Figure 1.8 Low temperature absorption and resonance Raman profiles of wild

type CuA site from B. subtilis. Raman modes: ν1, “accordion” mode;

ν2, mixed Cu-S/Cu-N stretching mode; ν3, out-of-phase “twisting”

xvii Cu-S stretching modes; ν4, symmetric breathing mode. ------12

Figure 1.9 (A) Mixed-Valent (MV) Model; (B) Homovalent (II,II) Model. ------13 Figure 1.10 Assignment of ψ–ψ* transition in the MV model complex. Comparison of low-temperature absorption (a) and MCD (b) spectra for the mixed valence model complex (red lines) and the homovalent model complex (blue lines) reveals the ψ–ψ* band at 5560 cm-1 that

is present only in the MV complex. ------14

Figure 1.11 D2h-idealized Molecular orbital splittings in MV model and CuA,

showing the separate contributions of Cu-S (hCu-S) and Cu-Cu (hCu-Cu)

bonding interactions to 2HAB. σ and π labels refer to the symmetries

of Cu-Cu interactions only. ------14

Figure 1.12 The spectroscopic definition of the electronic structure of CuA. (A) Comparison of absorption spectra shows an increase in ψ–ψ* transition energy relative to the mixed valence model. (B) Sulfur K-

edge XAS reveals similar bridging thiolate covalency in CuA center

and the MV model. ------15

Figure 1.13 Potential energy surfaces in the Q- mode for CuA and the MV model

showing the strong stabilization for valence delocalization in CuA

due to metal-metal bonding. ------15

Figure 1.14 Schematic of cytochrome c pathways. ------17

CHAPTER 2 Figure 2.1 Absorption spectra of resting WT NiR at 7 K (green) and room

temperature (blue). ------33 Figure 2.2 Resonance Raman spectra of resting WT NiR excited at A) 458 nm and B) 568 nm. 77 K data are in bold lines and 25 0C in dotted lines. Inset: resonance Raman spectrum of the pure Blue component of NiR (blue spectrum, obtained by subtracting the spectrum of the green species at 77 K, 568 nm from the 25 0C, 568 nm spectrum and

renormalization) and of the M182T variant of NiR (red spectrum). - 34

xviii Figure 2.3 Resonance Raman profiles overlaid with the absorption spectra. The 364 cm-1 peak associated with the green copper species, while the -1 420 cm peak reflects the blue copper component of NiR. ------35 Figure 2.4 Temperature dependence of the absorption spectra of resting WT NiR

A) in buffer with ethylene glycol (40:60) and B) in buffer solution. - 38

Figure 2.5 Plot of lnKeq vs. 1/T, where Keq ~ 1 at 298 K. ------39 Figure 2.6 The geometry of A) oxidized and B) reduced type 1 Cu sites (only Cu and coordinated N, S atoms are shown for clarity, and the arrows indicate the major angle changes). Red – crystal structure of NiR at

100 K, short Cu-SMet bond length; Green – fully geometry optimized

structure, short Cu-SMet bond length; Blue – partial geometry

optimized structure with Cu-Sthioether distance constrained at 4 Å of

the long bond length. ------42

Figure 2.7 Potential energy surface as a function of Cu-Sthioether distance (solvation included using a PCM with ε = 4.0). Solid line-oxidized

Cu site, dashed line-reduced Cu site. ------42

CHAPTER 3 Figure 3.1 Structures of T1 Cu sites (plastocyanin in black, pseudoazurin in

grey). ------58 Figure 3.2 Electronic absorption (5 K) and MCD (5 K, 7 T) spectra of P. pantotrophus pseudoazurin. Both spectra were taken on a glass sample of 1.5 mM pseudoazurin in a 10 mM phosphate buffer (pH

6.9)/glycerol (50:50 v/v). ------60 Figure 3.3 Resonance Raman spectra of P. pantotrophus pseudoazurin with laser excitation of 593 and 450 nm at 77 K. The sample consisted of

~1.5 mM pseudoazurin in 10 mM phosphate buffer (pH 6.9). ------62 Figure 3.4 X-Band (A) and Q-band (B) EPR data on pseudoazurin (solid line: experiment; dashed line: simulation; simulated results are listed in Table 2); high frequency EPR data 95 GHz and 285 GHz (C).

xix Experimental condition: X-band data, 9.39 GHz, were taken at 77K; Q-band, 33.83 GHz, were taken at 50 K, EPR spectra at 95 GHz and

285 GHz were recorded at 4.2 K. ------63 Figure 3.5 Absorption spectra of pseudoazurin at 5 K (solid line) and 295 K (dotted line), Spectra were taken on a sample of 1.5 mM pseudoazurin in 10 mM phosphate buffer (pH 6.9)/glycerol (50:50

v/v). ------65 Figure 3.6 Resonance Raman spectra of pseudoazurin with laser excitation at 607 nm at 77 and 295 K. Spectra were taken on a sample of 1.5 mM

pseudoazurin in 10 mM phosphate buffer (pH 6.9). ------66 Figure 3.7 Continuum of coupled tetragonal distortions in a series of perturbed blue copper proteins (Pc: plastocyanin, CBP: cucumber basic protein,

NiR: nitrite reductase). ------68

CHAPTER 4

Figure 4.1 (top) DFT-Optimized structures of CuA without the protein * environment: (A) σu state and (B) πu state (only Cu2S2 cluster is

shown for simplicity, internuclear distances are given in Å, CuM denotes the Cu atom with the axial Met ligand); (bottom) β-spin + LUMO (contour value = 0.03) of the [Cu2(SCH3)2(imz)2] complex * in the (A) σu and (B) πu states. ------75 Figure 4.2 (A) Room-temperature absorption, (B) low-temperature (5K) MCD spectra of Tt CuA; (C) TD-DFT calculated absorption spectrum of the CuA model (total absorption –black and individual components –

red). ------80 Figure 4.3 (A) The ground state and the first excited state potential energy

surfaces (black lines refer to the CuA cluster in the vacuum and green – the cluster in the protein environment) and (B) Mayer bond order

between the two Cu atoms in the GS of CuA as a function of the Cu-

Cu distance. ------81

xx Figure 4.4 Calculated ET rate ratios kσ / kπ in the heme c→CuA and CuA→heme a

pathways as a function of the ET driving force (ΔG). ------82

Figure 4.S1 Structures of the CuA models for DFT calculations: (A) 30-atom

model (no Cu axial ligands), (B) 51-atom model (with CH3-S-CH3

and CH3-CO-NH-CH3 as axial ligands), (C) 217-atom model based 23 on the high-resolution structure of an engineered CuA azurin (PDB ID: 1CC3). The structures shown correspond to the geometries of the

σu* ground state. ------86

Figure 4.S2 Structure (top and side view) of the 217-atom CuA model. Gray- colored atoms indicate the frozen atoms in the geometry optimization; the coordinates of the atoms shown in blue were

optimized. ------87 Figure 4.S3 Mayer bond orders for the four Cu-S bonds (black lines and circles)

of the CuA cluster (51-atom model), the sum of the Cu-S bond orders (blue open circles) and the sum of the Cu-S and Cu-Cu bond orders

(blue squares). CuM is the Cu atom of the Cu2S2 cluster with the

thioether axial ligand. ------88

Figure 4.S4 NPA-derived atomic spin densities for the Cu and Sthiolate atoms of

the CuA cluster (A: the 51-atom model , B: the 217-atom model ).

CuM designate the Cu atom with the thioether axial ligand. ------89

CHAPTER 5 Figure 5.1 (A) X-band EPR spectra. Microwave frequency, 9.46 GHz; temperature 77K; (B) Q-band EPR spectra. Microwave frequency, 33.86GHz; temperature 77K; high pH form (⎯), low pH form (⎯),

H120A mutant (⎯), and XSophe simulations (⎯). ------102 Figure 5.2 Low-temperature absorption and MCD spectra of the High pH form

(⎯), the low pH form (⎯) and the H120A mutant (⎯). ------104 Figure 5.3 Resonance Raman spectra of (A) S→Cu CT band exitation and (B) ψ→ψ* band excitation (High pH form ⎯, low pH form ⎯ and

xxi H120A mutant ⎯; solvent peaks are marked with stars). ------105 Figure 5.4 (A) EXAFS data and (B) the corresponding non-phase shift

corrected Fourier transforms of w.t. CuA azurin at high pH (⎯) and

Low pH (⎯) and H120A mutant (⎯). ------108 Figure 5.5 DFT geometry optimized structures (A) 96-atom model of the high-

pH form; (B) 97-atom model of the low-pH form/H120 mutant. ------111 Figure 5.6 β-LUMOs with the contour values of 0.03 a.u. (A) 96-atom model of

the high pH form; (B) 97-atom model of the low pH form. ------112

Figure 5.7 Simplified molecular orbital diagram of the in-plane Cu2S2 core in

CuA site demonstrates the coupling between two C2v NCuS2

monomers each with its 3d 22 orbital highest in energy, the − yx 115 3d orbital second, followed by the 3d 2 , 3d and 3d orbitals. xy z xz yz ------Figure 5.8 TD-DFT calculated absorption spectra of the high pH form model

(solid line) and the low pH form model (dashed line). ------117 Figure 5.9 Graphic prediction of the C-term sign for the ψ→ψ* transition (i.e. b σg →σu*). ------119

Figure 5.10 Potential energy surfaces in Q- mode for the low-pH form of CuA

-1 2 site. The specific parameters used are: 2HAB =12000 cm , Λ / k− = -1 122 2450 cm , ΔE = 0.120 eV. ------Figure 5.11 Proposed ET pathways in bovine heart CcO based on Pathways

analysis (reference 34). The Cys200 and His204 CuA-to-heme a

pathways are comparable in rate. ------124 Figure 5.S1 (A) 5K 1T MCD, LT Abs and RT CD of the high pH form. Abs and MCD peaks are fitted simultaneously, and compared to RT CD peak

fitting. ------127 Figure 5.S1 (B) 5K 1T MCD, LT Abs and RT CD of the low pH form. Abs and MCD peaks are fitted simultaneously, and compared to RT CD peak

fitting. ------128

Figure 5.S2 ~77K resonance Raman excitation profiles of CuA Azurin: (a) High-

xxii pH form; (b) Low-pH form. ------129 Figure 5.S3 Fourier transforms (non phase shift corrected) and EXAFS data

(Inset) of WT P.a purple CuA construct at high pH. Data (―), Fit

(―). ------130 Figure 5.S4 Fourier transforms (non phase shift corrected) and EXAFS data

(Inset) of WT P.a purple CuA construct at low pH. Data (―), Fit (―). 131 Figure 5.S5 Fourier transforms (non phase shift corrected) and EXAFS data

(Inset) of the H120A mutant of P.a purple CuA. Data (―), Fit (―). -- 132

Figure 5.S6 (A) X-band EPR spectrum of the high pH form of CuA; (B) XSophe simulations of A. (C) XSophe simulations of A with only the A- tensor rotation considered. (D) XSophe simulations of A with Fermi

contact change only on CuO site. (E) X-band EPR spectrum of the

low pH form of CuA. ------133

CHAPTER 6 Figure 6.1 The structures of cytochrome c models: A) 55-atom model; B) 149 143

atom model. ------Figure 6.2 Schematic of molecular orbital diagram of Fe-P-Im-DMS. The table- inset give the symmetry correlations between porphyrin and Fe d

orbital ------150

Figure 6.3 Schematic of molecular orbital diagram of Fe-P-(Im)2. ------151 Figure 6.4 Figure 6.5 TD-DFT calculated absorption spectrum of the Cyt c model of A) FeIII with His and Met axial ligand; B) FeIII with bisHis axial ligand; C) experimental absorption spectrum of ferricytochrome

c. ------152

APPENDIX

Figure A.1 Design of model compounds. ------160

Figure A.2 Dioxygen binding of 1Co in the presence of ligand 2. ------160 Figure A.3 UV-vis spectra of 1Co in PH 7.0 phosphate buffer solution (C = ~

xxiii 6 5x10 M): (a) 1Co under N2; (b) 1Co + ligand 2 (~100 equivalents)

under N2; (c) 1Co + ligand 2 (~100 equivalents) under O2. ------161

- Figure A.4 UV-vis absorption spectra of the oxygen titration course (changing

from 2- 1Co to 2-1Co(O2)). ------162 Figure A.5 EPR Spectra of Complex 2-1Co under PH 7.0 buffer solution at 77K

(1Co: C > 1mM, ligand 2: 10 equivalents): (a) under N2; (b) after

oxygenation. ------163

Figure A.S1 In the absence of ligand 2, at pH 7.0, under N2, g|| = 2.02, A|| = 110G,

g⊥ is not resolved. ------167

Figure A.S2 (a) In the presence of ligand 2, at pH 7.0, under N2, g|| = 2.02, A|| =

86G, AN = 16G, g⊥ = 2.31; (b) In the presence of ligand 2, at pH 7.0,

under O2, g|| = 2.08, A|| = 17G, g⊥ = 2.01, A⊥ = 11G. Spin quantification shows a:b ~100:66 (intensity lost over time). ------168 -

Figure A.S3 (a) In the presence of ligand 2, at pH 5.0, under N2, g|| = 2.02, A|| =

85G, AN = 15G, g⊥ = 2.31; (b) In the presence of ligand 2, at pH 5.0,

under O2 (oxygenation adduct decomposed). Spin quantification

shows a:b ~100:0.7. ------169

Figure A.S4 (a) In the presence of ligand 2, at pH 9.0, under N2, g|| = 2.02, A|| =

83G, AN = 15G, g⊥ = 2.31; (b) In the presence of ligand 2, at pH 9.0,

under O2, g|| = 2.08, A|| = 17G, g⊥ = 2.01, A⊥ = 11G. Spin

quantification shows a:b ~100:4. ------170

xxiv

Chapter 1

Background*

* Reproduced in part with permission from Chem. Soc. Rev, 2008, 37, 613-872. Copyright © 2004 Royal Society of Chemistry.

1 1.1 Introduction Living systems require the integrated function of electron transfer (ET) in numerous important activities including photosynthesis, DNA damage and repair, respiration, and nitrogen fixation. ET in biological systems can involve transferring electrons between redox centers within a protein (intraprotein ET) or between proteins (interprotein ET), and many redox active centers are metalloproteins with unique spectral features associated with the unique geometric and electronic structure of their metal site(s). The metal active sites involved in biological ET are illustrated in Figure 1.1-5 In Cu

bioinorganic chemistry, these are Type 1 copper (T1 Cu), and CuA sites. Both have Cu centers in approximately trigonal coordination environment. The T1 Cu site has a highly

covalent thiolate (SCys) and two histidine (NHis) ligands in equatorial position, and a 6 weak/or no ligand in the axial position. In the CuA site, there are two coppers. Each

copper has a NHis and two bridging SCys equatorial ligands with additional weak trans axial ligands, one on each copper ( a sulfur atom from methionine (SMet) or an oxygen 7, 8 atom from backbone carbonyl (OGlu)). In iron-sulfur sites, there are either one Fe atom

with four SCys ligands in a distorted tetrahedral ligand field (rubredoxin) or 2 to 4 Fe clusters (ferredoxins and high potential iron-sulfur proteins (HiPIPs)) with each Fe in a distorted tetrahedral coordination environment formed by bridging sulfides and terminal thiolate ligands. The Fe centers in all iron-sulfur ET sites are high-spin in all redox states. In cytochomes, the most common ET sites are cytochromes b and c.9 The cytochrome b protein has a non-covalently bonded protoporphyrin IX, while cytochrome c protein has the heme prosthetic group covalently bonded to the protein through thioether linkages. The cytochromes (cyt) have varying axial ligands. There are of their two histidines in cyt b, and a histidine and a methionine in cyt c (Figure 1.1) creating strong ligand fields and low-spin Fe centers. In all cases, the ligand field and protein environment tune the reduction potential of the sites into their physiological range,3 and ET is rapid with a low reorganization energy (λ, little change in geometry with redox) and large electronic 10, 11 coupling through the protein (HDA) as described by the ET rate equation 1.1.

o 2 π 2 ⎡ (ΔG + λ) ⎤ kET = 2 ⋅(H DA ) ⋅exp⎢− ⎥ (1.1) h λkBT ⎣ 4λkBT ⎦

2 This dissertation focuses on spectroscopic and theoretical studies of T1 copper, CuA and cytochrome c sites to understand their geometric and electronic structure contributions to electron transfer. Thus, an overview of these sites is given in this chapter.

Figure 1.1 Electron transfer metal active sites in bioinorganic chemistry.

3 1.2 T1 Cu Sites T1 Cu sites in the oxidized (resting) state have unique spectroscopic features. The well

documented T1 Cu site or blue copper, plastocyanin (ligand set: 2Nhis and 1 Scys 2+ equatorial ligands, 1SMet axial ligand), exhibits an intense SCys → Cu charge transfer absorption band in the 600 nm region (ε ∼ 5000 M-1 cm-1, Figure 1.2A), resulting in its pronounced blue color.12 The hyperfine coupling of the electron spin (S = 1/2) to the copper nuclear spin (ICu = 3/2) in Electron Paramagnetic Resonance (EPR) spectrum is 63 x 10-4 cm-1 in contrast to the value of >150 x 10-4 cm-1 observed for normal Cu2+ complexes (Figure 1.2B)13. These unique spectral features are associated with the unusual geometric and electronic structure of the blue copper site, in particular the highly 14, 15 covalent Cu-SCys interaction.

A B

Figure 1.2 (A) Absorption spectra of plastocyanin (left ε scale) and normal Cu2+ complex 2- D4h [CuCl4] (right ε scale). (B) X-band EPR spectrum of plastocyanin (blue) and D4h 2- [CuCl4] (Black).

The covalency of the redox active molecular orbital (RAMO) of plastocyanin has been quantified through ligand K- and metal L-edge X-ray absorption spectroscopies (XAS).15, 16 S K-edge and Cu L-edge provide direct probes of the ligand and metal character, respectively, in their ½ occupied RAMO. The S pre-edge at 2470 eV reflects the transition from the S 1s orbital to the RAMO (Figure 1.3A top). Since the 1s orbital is

4 localized on the sulfur and s → p is electron dipole allowed, the intensity of the S pre- edge is directly proportional to the S 3p character mixed into the RAMO due to covalent interaction with the metal, i.e., the covalency of the sulfur-metal bond. The Cu L pre- edge at 930 eV reflects the Cu 2p → RAMO transition (Figure 3B top). The 2p orbital is localized on the Cu nucleus and p → d is electronic-dipole-allowed; therefore, the intensity of this pre-edge also reflects covalency, in this case the Cu d character in the RAMO. As shown in Figure 1.3, the RAMO of blue copper site in plastocyanin has 38%

SCys p and 41% Cu d character. This highly covalent bonding activates efficient long- range ET through the cysteine protein residue.

A B

Blue Copper Blue Copper Blue Copper Cu 38% A 38% 41% 44%

CuA 2x23% =46%

Figure 1.3 S K-edge and Cu L-edge XAS as probes of ligand-to-metal covalency. (A)

Comparison of S K-edge spectra of blue copper and CuA (bottom), energy level diagram depicting S 1s to RAMO transition (top), β2 is the amount of S p character in the RAMO.

(B) Comparison of Cu L-edge spectra of blue copper and CuA (bottom), energy level diagram depicting Cu 2p to RAMO transition (top).

5

LT Abs LT MCD

Figure 1.4 Low temperature absorption and magnetic circular dichrisom spectra of a series of T1 Cu proteins. Low-temperature absorption spectra show the redistribution of

spectral intensity from SCys π to σ CT bands. Redox active orbitals calculated with SCF- Xα-SW showing rotation from a π type interaction in plastocyanin (left top) to σ (+π ) mixture in nitrite reductase (left bottom). MCD spectra show a shift of the LF transitions to higher energy for nitrite reductase relative to plastocyanin indicating a tetragonal distortion.

Cucumber basic protein (CBP) and green nitrite reductase (NiR) have the same ligand set as plastocyanin yet exhibit perturbed spectral features (Figure 1.4). In the past, we have defined the coupled distortion model shown in Figure 1.5 that describes geometric changes in T1 Cu proteins in going from plastocyanin to CBP to green NiR along with

6 the protein color change from blue to green. Two extremes of the coupled distortion coordinate are represented by the T1 Cu site in fungal laccase which has no axial ligand,

and is blue in color, and green NiR, which has a strong axial SMet ligand. In the coupled distortion model, three dominant geometric changes occur in going from a blue to a green 2+ site: 1) a decrease in the Cu -SMet bond length which increases its donor interaction with 2+ 2+ the Cu , 2) this leads to an increase in the length of the Cu -SCys bond, and 3) a distortion from a more tetrahedral to a more tetragonal structure, which reflects a Jahn- 2+ 2+ Teller (εu ) rotation of the SCys-Cu -SMet plane into the NHis1-Cu -NHis2 plane. These geometric changes result in spectroscopic changes for the different proteins along the coupled distortion coordinate from blue to green. These spectroscopic changes are: 1) an increase in the ε450/ε600 ratio in the absorption spectra; 2) a decreased effective resonance

Raman vibrational frequency, <νCu-Scys>; and 3) an increase in the ligand field transition (d → d) energies in MCD spectra (Figure 1.4).

Figure 1.5 Continuum of coupled tetragonal distortions in a series of perturbed blue copper proteins. Crystal structures show contraction of the Cu-SMet bond associated with elongation of the Cu-SCys bond and tetragonal εu -like mode Jahn-Teller distortion.

1.3 CuA Sites 17-22 The CuA site, found in cytochrome c oxidases (CcOs) , nitrous oxide reductases 23, 24 25, 26 (N2ORs), and a nitric oxide reductase (NOR), is responsible for rapid intra- and intermolecular electron transfer (ET) in biological systems. CcOs catalyze the terminal

7 step of enzymatic aerobic respiration by coupling the four electron reduction of O2 to

H2O to the generation of a proton electrochemical gradient, which is the energetic driving force for the synthesis of adenosine triphosphate (ATP). As shown in Figure 1.6, the

catalytic active site in CcO is a heme-Cu binuclear center while a second heme and CuA serve as ET intermediaries between cytochrome c and the catalytic center. NOR reduces two NO to N2O + H2O, and N2OR is the terminal denitrification enzyme reducing N2O to

N2 + H2O. In all enzymes, the CuA site serves as an ET conduit between an electron source and the catalytic active site.

Figure 1.6 Scheme of electron and proton flow in cytochrome c oxidase.

27, 28 The two coppers in an oxidized CuA site are separated by ~2.4 Å and bridged by 29-32 two SCys forming a planar ‘diamond core’ . In its ET process, CuA shuttles between the reduced state [2Cu]2+ and the oxidized state [2Cu]3+. The latter is mixed valent (MV).

In the MV state of the binuclear Cu, the extra electron can be either localized on Cua

(with a wavefunction given by ψ[Cua(I)Cub(II)]), localized on Cub (wavefunction =

8 ψ[Cua(II)Cub(I)]), or delocalized between the two Cu centers as described by equation 1.2.33 2 1/2 ψground = (1-α ) ψ[Cua(I)Cub(II)] + αψ[Cua(II)Cub(I)] (1.2)

In the Robin and Day classification scheme,34 MV complexes are characterized as class I, II or III based on the value of α2, the extent of delocalization. A site with the extra electron completely localized, α2 = 0, is called a class I MV complex. In class II MV sites, partial delocalization of the extra electron occurs (i.e. 0 < α2 < 0.5). The completely delocalized case (α2 = 0.5) corresponds to the class III MV limit, which exhibits strikingly different spectroscopic features from those of the individual localized cases.

For CuA, its EPR signal in the g|| region (Figure 1.7) exhibits a seven-line hyperfine coupling of the unpaired electron (S = 1/2) to both Cu (I = 3/2) centers.35 This indicates

that CuA is at the delocalized Class III limit. The extent of delocalization is determined by the electronic coupling, HAB, associated with bonding interactions between the valence

orbitals on Cua and Cub. Note that this electronic coupling can involve both direct overlap of the valence orbitals of the two metals and superexchange through the bridging thiolate ligands.

g||

2.30 2.25 2.20 2.15 2.10 2.05 2.00 1.95

g

Figure 1.7 Q-band (red) and X-band (Black) EPR spectra of CuA protein.

9 An oxidized metal center will generally have shorter metal-ligand bond lengths than a reduced metal center. This distortion (symmetric contraction or elongation of all ligand-

metal bonds of metal centers A and B in their breathing modes, QA and QB) in an MV site gives an energy stabilization term that can trap the oxidation on the ligand contracted metal center. This is shown in Chart 1. Following the Piepho, Krausz and Schatz (PKS)

model in the Q- mode, which is the antisymmetric combination of the breathing modes -1/2 36 (Q_ = 2 (QA - QB), bottom of Chart 1), the combined effect of electronic coupling, 1 HAB, and vibronic trapping, λQ , on the delocalization of the ground state is given by 2 −

equation 1.3 in terms of the dimensionless coordinate x_ (x− = (k− / λ)Q− , k_ is force 1/2 constant and λ is the vibronic coupling parameter where λ ≈ k_(n Δrredox), in which n is

the number of metal-ligand bonds and Δrredox is the difference in the metal-ligand bond

length between the oxidized and reduced centers). The fact that the CuA center is a class III MV system indicates that the electronic coupling has overcome the vibronic trapping for this site. This lack of vibronic trapping plays a significant role in lowering the Frank-

Condon barrier to ET by the CuA center in biology.

2 1/ 2 1 ⎛ λ2 ⎞ ⎡1 ⎛ λ2 ⎞ ⎤ E ± = ⎜ ⎟x2 ± ⎢ ⎜ ⎟ x2 + H 2 ⎥ (1.3) 2 ⎜ k ⎟ − 2 ⎜ k ⎟ − AB ⎝ − ⎠ ⎣⎢ ⎝ − ⎠ ⎦⎥

Chart 1

10 How does the covalency of CuA correlate to that of the very well documented Blue copper site4, 37? From Figure 1.3A, the S K-edge XAS intensity of the blue copper pre- edge is about twice that of CuA. However, for CuA, the pre-edge reflects the covalency/thiolate and must be doubled; therefore CuA has approximately the same thiolate character as blue copper, but delocalized over the two thiolates (38% S 3p in blue 38 Copper, 46% S 3p in CuA). From Figure 1.3B, blue copper and CuA both have about the same Cu L-edge integrated intensity, therefore Cu d characters in their RAMO′s (41%

Cu d in blue copper, 44% Cu d in CuA), however again for CuA, this is delocalized over the two Cu centers.

The highly-covalent, delocalized ground state wavefunction of CuA makes major contributions to rapid ET.38 The high covalency of the sulfur bridges activates specific superexchange pathways for ET. The delocalized nature of CuA distributes the geometry change associated with redox over twice the number of bonds as in a localized site, but with half of the distortion in each bond. As the reorganization energy (λreorg) in Marcus

6 2 theory (Equation 1.1) goes as the distortion squared ( λreorg ≈ kdisn(Δr) ), this decreases

39 the reorganization energy by ½ and leads to an increase in kET.

Also from the pre-edge energies in Figure 1.3, the RAMO of CuA is 0.8 eV higher in energy than that of blue copper even though these have similar trigonal ligand fields.

This must then reflect a bonding interaction between the Cu′s at RCu-Cu = 2.4 Å. This has been probed directly by absorption (Abs) and resonance Raman (rR) spectroscopies.40, 41

Figure 1.8 shows the low temperature absorption spectrum of CuA. There are two regions: the bands at ~ 20 000 cm-1 are thiolate to Cu charge transfer (CT) transitions as indicated by Raman enhancement by these bands, which shows enhancement in the symmetric, in-phase breathing mode (ν4), and in the Cu-S and Cu-N bond distortions (ν2) -1 and (ν3). Alternatively, excitation into the Abs band at 13400 cm enhances only the breathing mode (ν4) and the symmetric out-of-phase accordion mode (ν1), which 40-42 indicates distortions only in ν4 and ν1, thus a change in the Cu-Cu distance. From the rR intensities, the exited state distortion associated with this electronic transition can be estimated as a 0.44 Å elongation in the Cu-Cu bond with no change in Cu-L bond lengths.43 This transition is assigned as the Cu-Cu ψ→ψ* transition (a transition between

11 the Cu-Cu bonding-to-antibonding molecular orbitals of the class III MV system) with distortions only along the symmetric vibrational modes of the Cu2S2 core, consistent with the completely delocalized nature of the ground state associated with a strong Cu-Cu interaction at 2.4 Å.

ν ν4 ν1 S ν S 3 2 S N Cu Cu Cu Cu Cu Cu N S

S S

Figure 1.8 Low temperature absorption and resonance Raman profiles of wild type CuA site from B. subtilis. Raman modes: ν1, “accordion” mode; ν2, mixed Cu-S/Cu-N stretching mode; ν3, out-of-phase “twisting” Cu-S stretching modes; ν4, symmetric breathing mode.

The bonding contributions to the Cu-Cu interaction in CuA were elucidated by comparison to a class III MV model complex (Figure 1.9A) reported by Tolman and coworkers.40, 44, 45 This complex has a Cu-Cu bond length of 2.9 Å, which eliminates the direct Cu-Cu bonding contribution to the interaction between the Cu′s leading to delocalization. By comparison of the low temperature Abs/MCD spectra in Figure 1.10 of the MV complex in red to the spectra of the homovalent analog (structure shown in Figure 1.9B) in blue, the band at 5600 cm-1 can be assigned as the ψ → ψ* transition of

12 the MV model. This reflects in the electronic coupling between the two Cu′s (2HAB) and derives from the superexchange type pathway associated with the bridging thiolates. From the schematic in Figure 1.11 top that is based on density functional theory (DFT) calculations, this produces an energy splitting of the dπ orbitals on the two Cu′s leading to a πu lowest unoccupied molecular orbital (LUMO) due to its antibonding interaction with the thiolate bridges. S K-edge data in Figure 1.12 bottom quantify the high sulfur covalency in this LUMO. Comparison of the MV model to CuA shows that the ψ → ψ* -1 transition in CuA shifted up in energy by 7800 cm , yet CuA has somewhat less S 3p character in its LUMO (Figure 1.12). This requires that there is an additional contribution to the electronic coupling between the Cu′s in CuA (2HAB) associated with a direct Cu-Cu bond. From the schematic based on DFT calculations in Figure 1.11 bottom, this involves a strong σ-type bonding/antibonding interaction between the d orbitals x 2 − y 2 on each Cu, leading to the σu* RAMO of CuA. This gives a net large 2HAB in CuA which is key to its delocalized electronic structure.

Figure 1.13 includes the effect of vibronic coupling in the Q- mode (i.e. equation 1.3) for CuA and the MV model. From Figure 1.13 right, the MV model is just at the -1 delocalized limit due to its 2HAB = 5600 cm . Alternatively for CuA, the large 2HAB =13400 cm-1 associated with the Cu-Cu bond at 2.4 Å gives a strongly stabilized, delocalized site, which is critical in keeping CuA delocalized even in its low symmetry protein environment.

A B

Figure 1.9 (A) Mixed-Valent (MV) Model; (B) Homovalent (II,II) Model.

13

5600cm-1 ψ→ψ*

Figure 1.10 Assignment of ψ–ψ* transition in the MV model complex. Comparison of low-temperature absorption (a) and MCD (b) spectra for the mixed valence model complex (red lines) and the homovalent model complex (blue lines) reveals the ψ–ψ* band at 5560 cm-1 that is present only in the MV complex.

Cu-Cu π b u Cu-Cu σu* β-LUMO β-LUMO

2H = 2HAB= 2|hCu-S| AB 2|hCu-S| -1 13400cm-1 5600cm

b Cu-Cu π b Cu-Cu σ g g

MV: Cu-Cu 2.9Å CuA: Cu-Cu 2.4Å

Figure 1.11. D2h-idealized molecular orbital splittings in MV model (left) and CuA (right), showing the separate contributions of Cu-S (hCu-S) and Cu-Cu (hCu-Cu) bonding interactions to 2HAB. σ and π labels refer to the symmetries of Cu-Cu interactions only.

14

Figure 1.12 The spectroscopic definition of the electronic structure of CuA. (A) Comparison of absorption spectra shows an increase in ψ–ψ* transition energy relative to the mixed valence model. (B) Sulfur K-edge XAS reveals similar bridging thiolate

covalency in CuA center and the MV model.

8000 Cu σ * 8000 A u MV π u ) ) -1 -1 4000 4000 Ψ* Ψ*

0 0 2H =13400cm-1 2H =5600cm-1 AB AB -4000 -4000 Relative Energy(cm Relative Energy(cm Ψ Ψ

-2-1012 -2 -1 0 1 2 x x - - Figure 1.13 Potential energy surfaces in the Q- mode for CuA and the MV model showing

the strong stabilization for valence delocalization in CuA due to metal-metal bonding.

15 1.4 Cyt c Cyts c fall into at least four general classes.9, 46 Mitochondrial cyt c (class I) is primarily known as an electron transfer (ET) heme protein. It transfers electrons to the membrane bound enzyme cytochrome c oxidase (CcO) as shown in Figure 1.6. The monoheme group in mitochondrial cyt c is covalently linked to the protein polymer through two thioether bonds associated with two Cys, and the heme attachment site (- Cys-Xxx-Yyy-Cys-His-) is towards the N- terminus. The heme iron has two axial ligands

(the fifth or proximal ligand is NHis, and the sixth or distal liagnd is SMet). Class II cyts c have monoheme covalently linked to the highly conserved –Cys-Xxx-Yyy-Cys-His- near their C-termini, divided into two subgroups IIa (known as cyt c’ with only one axial ligand NHis) and IIb with sixth SMet ligand. Triheme, tetraheme, octaheme, and 16-heme c from sulfate and sulfur-reducing bacterial are included in class III with bisHis ligation. Class IV cyst c are tetraheme photosynthetic reaction center with four heme in His/Met ligation or three heme in bisHis ligation and one heme in His/Met ligation. However, there are cyts c not homogenous to any of I-IV classes such as cyt c554 (tetraheme: three hemes are six coordinates with bisNHis axial ligands, and one heme is five coordinates with an axial NHis ligand) involved in biological nitrification pathway and cyt f

(monoheme with two axial liagnds: NHis and NTyr) from the cyt b6f complex of oxygenic photosynthesis. Considerable efforts have been made to investigate the structure and the electron transfer mechanism of mitochondrial cyt c by single crystal structure,47-49 NMR,50-53 Mössbauer, electron paramagnetic resonanace (EPR), electronic absorption, circlular dichroism (CD),54 magnetic CD55 and resonance Raman spectroscopies56-58. The typical absorption spectra of heme complexes are characterized by one intense Soret band (or B band) in the UV-Visible region and one relatively weak Q band in visible region. There are also weak features in the lower or higher energy regions. In the oxidized state, mitochondrial cyt c has a unique absorption band at 695 nm. Interestingly, the peak intensity of 695nm absorption band decreases at elevated temperature, which led to the proposal of a thermal dynamic equilibrium between a “hot” and a “cold” form of cyt c by Schejter and George.59 Further, the biphasic temperature dependence of 695nm peak was observed by Filosa and English and they proposed the existence of a third

16 alkaline isomer.60 Recently, Elliott and his coworkers61 detected a low reduction potential form of cyt c ascribed to the loss of the Met ligand. It is generally thought the Fe-Smet bond is stronger in the reduced rather than the oxidized state, which is based on the electrochemistry studies61, 62 and theoretical calculations63. While all these are related to the binding strength of the axial ligand SMet, the origin of the 695 nm band is in dispute and the Met-loss species are not clear as to whether the met is replaced by an alternative protein residue, water or no ligand. One new discovery concerning cyt c is its role in apoptosis, a controlled form of cell death used to kill cells in the process of development or in response to infection or DNA damage.64, 65 Cyt c is released by the mitochondria in response to pro-apoptotic stimuli. The release of small amounts of cyt c leads to an interaction with the IP3 receptor (IP3R) on the endoplasmic reticulum (ER), causing the ER calcium release. The overall increase in calcium triggers a massive release of cyt c, which then acts in the positive feedback loop to maintain ER calcium release through the IP3Rs. This release of cytochrome c in turn activates caspase 9, a cysteine protease. Caspase 9 can then go on to activate caspase 3 and caspase 7, which are responsible for destroying the cell from within. Cyt c is also suspected to be the functional complex in the so called Low-level laser therapy (LLLT). In LLLT, 670 nm laser light penetrates wounded and scarred tissue to stimulate cellular regeneration and tissue repair.66

(CcO Figure 1.14 Schematic of cytochrome c pathways.67

17 1.5 Overview of Thesis

This thesis presents studies on T1 Cu, CuA and cytochrome c electron transfer active sites, utilizing a wide-range of spectroscopies combined with density functional calculations to understand their electronic structures, the origin of their geometric structures, and how they contribute to function. Chapter 1 gives an overview of metal sites that are known to be involved in biological electron transfer, and focuses on the spectral features of T1 Cu, CuA sites and cytochrome c. Portions of this chapter are from a review article published in Chem. Soc. Rev. 2008, 37, 623-638. Chapter 2 investigates the nature of large temperature-dependent absorption spectral changes in Nitrite Reductases (NiRs). This leads to the discovery of a thermodynamic equilibrium between two species in NiR: a green form and a blue form. The green form is enthalpically favored with a short thioether Smet-Cu bond, and the blue form is entropically favored associated with the loss of the unconstrained thioether Smet-Cu bond. Extension to and the contrast of this result to the behavior of the blue copper site in Plastocyanin define the constraints imposed on the blue copper site by the protein. The concept of the protein constraint on the copper site is referred to as an entatic or rack induced state.68-70 Its biological importance will be discussed. These studies were perused in collaborating with Dr. Somdatta Ghosh and Dr. Abhishek Dey. The protein was provided by Prof. Charles P. Scholes and his student Yan Sun at SUNY, Albany. This work has appeared as a full paper in Proc. Natl. Acad. Sci. USA, 2009, 106, 4969- 4974. Chapter 3 extends the study of chapter 2 to a T1 Cu site of P. pantotrophus pseudoazurin. This T1 site shows an interesting absorption spectrum in which both the ~600 nm π (blue Cu) and ~450 nm σ (green Cu) CT bands are fairly intense at room temperature.71 In this study, the combination of spectroscopic methods were used to investigate whether the presence of these two bands in the absorption spectrum reflects a thermodynamic equilibrium between two different species as in NiR, or whether they reflect a single species as in plastocyanin thus also having protein constraints (but different from that in plastocyanin) on the active site. Based on NiR studies in chapter 2 and this extended investigation, we demonstrate that for T1 Cu sites and other

18 metalloproteins with weak axial ligation, the temperature dependence of their spectral features provides a direct probe as to whether the protein allows the ligand to dissociate from the metal. These studies were pursued in collaboration with several people. Protein samples were provided by Prof. Isabel Moura and Dr. Sofia R. Pauleta at Universidade Nova de Lisboa, Caparica, Portugal. The 95 GHz and 285 GHz EPR were conducted by Prof. Sun Un at Institut de Biologie et de Technologies de Saclay, CEA Saclay, Gif-sur- Yvette, France. Ryan G. Hadt provided help in data collection, data analysis and paper writing. This work has appeared as a full paper in J. Inorg. Biochem. 2009, 103, 1307- 1313.

Chapter 4 is a study of CuA sites to resolve a two-state issue. For the CuA site in the * protein, σu and πu are the ground and first energy excited states, respectively. EPR studies suggest that the first excited state energy is at 3,000-4,500 cm-1. However, from paramagnetic NMR studies, it has been observed that the first excited state is thermally- accessible and is at ~350 cm-1. This study addresses this apparent discrepancy and * evaluates the roles of the two electronic states, σ u and πu, in electron transfer (ET) of

CuA. Density functional theory calculations were done together with Dr. S. I. Gorelsky.

T. thermophilus CuA Protein samples were provided by Prof. James A. Fee and his student Ying Chen at The Scripps Research Institute, La Jolla, CA. This work has appeared as a communication in J. Am. Chem. Soc. 2006, 128, 16452-16453.

Chapter 5 is a detailed description of the perturbations on the CuA site (in an azurin 72 variant) upon changing pH. This CuA site exhibits a very interesting pH effect. CuA Azurin at pH = 7 displays a mixed valence EPR spectrum with a seven-line hyperfine pattern associated with the delocalization of the S=1/2 over two Cu centers (ICu=3/2). Upon lowering the pH to 4.0, the EPR hyperfine coupling changes to a four-line pattern, indicating apparent spin localization. Interestingly, the H120A mutant has very similar spectral features to the low-pH form, and the pH-dependent transition is eliminated. Multifrequency EPR data have indicated that the electron spin in the H120A mutant is localized based on a four-line Cu hyperfine pattern.73 However, ENDOR data indicate that the site is delocalized based on superhyperfine couplings of cysteine Cβ protons and hisidine nitrogens.73 In this study, we address this apparent discrepancy through a series of electronic spectroscopic studies of the high-pH and low-pH forms of CuA azurin and

19 its H120A mutant combined with DFT calculations. The possible contribution of the pH- dependent transition to proton coupled electron transfer (PCET) in CcO is also discussed. These studies were in collaboration with Prof. Yi Lu and his students Dewain Garner and

Hee Jung Hwang who provided CuA proteins and were involved in discussions. Dr. Ritimukta Sarangi collected and analyzed all the EXAFS data, working under the supervision of Prof. Keith O. Hodgson and Prof. Britt Hedman. The work presented in this chapter has appeared as a full paper in J. Am. Chem. Soc. 2008, 130, 5194-5205.

Chapter 6 presents initial studies on cyt c investigating the nature of its Fe-SMet bond and the role of methionine in tuning protein functions (ET and apoptosis). This is inspired by the studies of NiR in chapter 2. In blue copper sites, the Cu-Smet bond is stronger for Cu2+ than for Cu1+, while for cyt c it is generally thought to be stronger in the reduced rather than the oxidized state, which is based on the electrochemistry sudies,61, 62 ligand competition74, 75 and protein folding studies74, and a slightly shorter Fe-S bond in the reduced site from EXAFS.76 DFT calculations calibrated with experimental data provide insight into the nature of the Fe-SMet bond in cyt c. These studies were in collaboration with Ryan G. Hadt. Prof. Bozhen Chen was involved in DFT calculations. Dr. Ritimukta Sarangi generously provided horse heart cytchrome c protein. The Appendix focuses on water-soluble cobalt porphyrins in aqueous solution to study oxygen binding. The synthesis and complex characterizations were done by Dr. Yi-long Yan. My contributions were in EPR experiments including data collection and analysis, and scientific discussions.

In summary, this thesis presents a study of T1 Cu, CuA and cytochrome c using spectroscopic techniques and complementary density functional computations to probe geometric and electronic structures and their contribution to reactivity.

1.6 References 1. Holm, R. H.; Kennepohl, P.; Solomon, E. I., Structural And Functional Aspects of Metal Sites in Biology. Chem. Rev. 1996, 96, (Thematic "Bioinorganic Enzymology" Issue), 2239-2314. 2. Nocek, J. M.; Zhou, J. S.; De Forest, S.; Priyadarshy, S.; Beratan, D. N.; Onuchic, J. N.; Hoffman, B. M., Theory and Practice of Electron Transfer within Protein-

20 Protein Complexes: Application to the Multidomain Binding of Cytochrome c by Cytochrome c Peroxidase. Chem. Rev. 1996, 96, (7), 2459-2490. 3. Stephens, P. J.; Jollie, D. R.; Warshel, A., Protein Control of Redox Potentials of Iron-Sulfur Proteins. Chem. Rev. 1996, 96, (7), 2491-2513. 4. Solomon, E. I., Spectroscopic Methods in Bioinorganic Chemistry: Blue to Green to Red Copper Sites. Inorg. Chem. 2006, 45, (20), 8012-8025. 5. Solomon, E. I.; Xie, X.; Dey, A., Mixed Valent Sites in Biological Electron Transfer. Chem. Soc. Rev. 2008, 37, (4), 623-638. 6. Solomon, E. I.; Szilagyi, R. K.; DeBeerGeorge, S.; Basumallick, L., Electronic Structures of Metal Sites in Proteins and Models: Contributions to Function in Blue Copper Proteins. Chem. Rev. 2004, 104, (2), 419-458. 7. Tsukihara, T.; Aoyama, H.; Yamashita, E.; Tomizaki, T.; Yamaguchi, H.; Shinzawa-Itoh, K.; Nakashima, R.; Yaono, R.; Yoshikawa, S., Structures of metal sites of oxidized bovine heart cytochrome c oxidase at 2.8 Å. Science 1995, 269, (5227), 1069-1074. 8. Iwata, S.; Ostermeier, C.; Ludig, B.; Michel, H., Structure at 2.8 Å resolution of cytochrome c oxidase from Paracoccus denitrificans. Nature 1995, 376, (6542), 660-669. 9. Rodgers, K. R.; Lukat-Rodgers, G. S., Comprehensive Coordination Chemistry II. Elsevier Ltd: 2003; Vol. 8. 10. Marcus, R. A.; Sutin, N., Biochim. Biophys. Acta 1985, 811, 265-322. 11. Newton, M. D., Chem. Rev. 1991, 91, 767-792. 12. Solomon, E. I.; Penfield, K. W.; Gewirth, A. A.; Lowery, M. D.; Shadle, S. E.; Guckert, J. A.; LaCroix, L. B., Electronic structure of the oxidized and reduced blue copper sites: contributions to the electron transfer pathway, reduction potential, and geometry. Inorg. Chim. Acta 1996, 243, (1-2), 67-78. 13. Malmström, B. G.; Vänngård, T., J. Mol. Biol. 1960, 2, 118. 14. Penfield, K. W.; Gay, R. R.; Himmelwright, R. S.; Eickman, N. C.; Norris, V. A.; Freeman, H. C.; Solomon, E. I., Spectroscopic Studies on Plastocyanin Single Crystals: A Detailed Electronic Structure Determination of the Blue Copper Active Site. J. Am. Chem. Soc. 1981, 103, 4382.

21 15. Shadle, S. E.; Penner-Hahn, J. E.; Schugar, H. J.; Hedman, B.; Hodgson, K. O.; Solomon, E. I., X-ray absorption spectroscopic studies of the blue copper site: metal and ligand K-edge studies to probe the origin of the EPR hyperfine splitting in plastocyanin. J. Am. Chem. Soc. 1993, 115, (2), 767-776. 16. George, S. J.; Lowery, M. D.; Solomon, E. I.; Cramer, S. P., Copper L-Edge Spectral Studies: A Direct Experimental Probe of the Ground State Covalency in the Blue Copper Site in Plastocyanin. J. Am. Chem. Soc. 1993, 115, 2968. 17. Saraste, M. Q., Rev. Biophys. 1990, 23, 331-366. 18. Malmström, B. G., Chem. Rev. 1990, 90, 1247-1260. 19. Babcock, G. T.; Wikström, M., Nature 1992, 356, 301-309. 20. Musser, S. M.; Stowell, M. H. B.; Chan, S. I., Comparison of ubiquinol and cytochrome c terminal oxidases An alternative view. FEBS Lett. 1993, 327, 131- 136. 21. Garcίa-Horsman, J. A.; Barquera, B.; Rumbley, J.; Ma, J.; Gennis, R., B. J. Bacteriol. 1994, 176, 5587-5600. 22. Wikström, M., Cytochrome c oxidase: 25 years of the elusive proton pump. Biochim. Biophys. Acta-Bioenerg. 2004, 1655, 241. 23. Kroneck, P. M. H.; Riester, J.; Zumft, W. G.; Antholine, W. E., Biol. Met. 1990, 3, 103-109. 24. Zumft, W. G.; Kroneck, P. M. H.; Robert, K. P., Respiratory Transformation of Nitrous Oxide (N2O) to Dinitrogen by Bacteria and Archaea. Adv. Microb. Physiol. 2006, 52, 107. 25. Suharti; Strampraad, M. J. F.; Schroder, I.; de Vries, S., A Novel Copper A Containing Menaquinol NO Reductase from Bacillus azotoformans. Biochemistry 2001, 40, (8), 2632-2639. 26. Suharti; Heering, H. A.; deVries, S., NO Reductase from Bacillus azotoformans Is a Bifunctional Enzyme Accepting Electrons from Menaquinol and a Specific Endogenous Membrane-Bound Cytochrome c551. Biochemistry 2004, 43, (42), 13487-13495. 27. Blackburn, N. J.; Barr, M. E.; Woodruff, W. H.; van der Oost, J.; de Vries, S., Biochemistry 1994, 33, 10401.

22 28. Blackburn, N. J.; de Vries, S.; Barr, M. E.; Houser, R. P.; Tolman, W. B.; Sanders, D.; Fee, J. A., J. Am. Chem. Soc. 1997, 119, 6135. 29. Tsukihara, T.; Aoyama, H.; Yamashita, E.; Tomizaki, T.; Yamaguchi, H.; Shinazwaitoh, K.; Nakashima, R.; Yaono, R.; Yoshikawa, S., Science 1996, 272, 1136. 30. Williams, P. A.; Blackburn, N. J.; Sanders, D.; Bellamy, H.; Stura, E. A.; Fee, J. A.; McRee, D. E., Nat. Struct. Biol. 1999, 6, 509. 31. Ostermeier, C.; Harrenga, A.; Ermler, U.; Michel, H., Proc. Natl. Acad. Sci. USA 1997, 94, 10547. 32. Itawa, S.; Ostermeier, C.; Ludwig, B., Nature 1995, 376, 660. 33. Solomon, E.; Hanson, M., Bioinorganic Spectroscopy. Wiley: New York: 1999; Vol. II. 34. Robin, M. B.; Day, P., Adv. Inorg. Chem. Radiochem. 1967, 10, 247. 35. Kroneck, P.; Antholine, W.; Riester, J.; Zumft, W., The Cupric Site in Nitous- Oxide Reductase contains a mixed-valence [Cu(II),Cu(I)] Binuclear center - A Multifrequency Electron-Paramagnetic Resonance Investigation. FEBS Lett. 1988, 242, (1), 70-74. 36. Piepho, S. B.; Krausz, E. R.; Schatz, P. N., J. Am. Chem. Soc. 1978, 100, 2996- 3005. 37. Solomon, E. I.; Szilagyi, R. K.; George, S. D.; Basumallick, L., Electronic Structures of Metal Sites in Proteins and Models: Contributions to Function in Blue Copper Proteins. Chem. Rev. 2004, 104, 419. 38. George, S. D.; Metz, M.; Szilagyi, R. K.; Wang, H.; Cramer, S. P.; Lu, Y.; Tolman, W. B.; Hedman, B.; Hodgson, K. O.; Solomon, E. I., A Quantitative Description of the Ground State Wavefunction of CuA by X-ray Absorption Spectroscopy: Comparison to Plastocyanin and Relevance to Electron Transfer. J. Am. Chem. Soc. 2001, 123, 5757-5767. 39. Gorelsky, S. I.; Xie, X.; Chen, Y.; Fee, J. A.; Solomon, E. I., The Two-State Issue in the Mixed-Valence Binuclear CuA Center in Cytochrome c Oxidase and N2O Reductase. J. Am. Chem. Soc. 2006, 128, (51), 16452-16453.

23 40. Hay, M. T.; Ang, M. C.; Gamelin, D. R.; Solomon, E. I.; Antholine, W. E.; Ralle, M.; Blackburn, N. J.; Massey, P. D.; Wang, X.; Kwon, A. H.; Lu, Y.,

Spectroscopic Characterization of an Engineered Purple CuA Center in Azurin. Inorg. Chem. 1998, 37, 191-198. 41. Wallace-Williams, S. E.; James, C. A.; de Vries, S.; Saraste, M.; Lappalainen, P.; van der Oost, J.; Fabian, M.; Palmer, G.; Woodruff, W. H., Far-Red Resonance Raman Study of Copper A in Subunit II of Cytochrome c Oxidase. J. Am. Chem. Soc. 1996, 118, (16), 3986-3987. 42. Andrew, C. R.; Fraczkiewicz, R.; Czernuszewicz, R. S.; Lappalainen, P.; Saraste, M.; Sanders-Loehr, J., Identification and Description of Copper-Thiolate Vibrations in the Dinuclear CuA Site of Cytochrome c Oxidase. J. Am. Chem. Soc. 1996, 118, (43), 10436-10445. 43. Lee, S.-Y.; Heller, E. J., J. Chem. Phys. 1979, 71, 4777. 44. Houser, R. P.; Young, V. G., Jr.; Tolman, W. B., J. Am. Chem. Soc. 1996, 118, 2101-2102. 45. Williams, K.; Gamelin, D.; LaCroix, L. B.; Houser, R. P.; Tolman, W. B.; Mulder, T. C.; de Vries, S.; Hedman, B.; Hodgson, K. O.; Solomon, E. I., Influence of Copper-Sulfur Covalency and Copper-Copper Bonding on Valence Delocalization and Electron Transfer in the CuA Site of Cytochrome c Oxidase. J. Am. Chem. Soc. 1997, 119, (3), 613-614. 46. Bertini, I.; Cavallaro, G.; Rosato, A., Cytochrome c: Occurrence and Functions. Chem. Rev. 2006, 106, (1), 90-115. 47. Takano, T.; Dickerson, R. E., Conformational change of cytochrome c: I. Ferrocytochrome c structure refined at 1.5 Å resolution. J. Mol. Biol. 1981, 153, 79-94. 48. Takano, T.; Dickerson, R. E., Conformational change in cytochrome c: II. Ferricytochrome c refinement at 1.8 Å resolution and comparison with the ferrocytochrome structure. J. Mol. Biol. 1981, 153, 95-115. 49. Berghuis, A. M.; Brayer, G. D., Oxidation state-dependent conformational changes in cytochrome c. J. Mol. Biol. 1992, 223, 959-976.

24 50. Feng, Y.; Roder, H.; Englander, S. W., Redoxdependent structure change and hyperfine nuclear magnetic resonance shifts in cytochrome c. Biochemistry 1990, 29, 3494-3504. 51. Gochin, M.; Roder, H., Protein structure refinement based on paramagnetic NMR shifts: applications to wild-type and mutant forms of cytochrome c. Protein Sci. 1995, 4, 296-305. 52. Qi, P. X.; Beckman, R. A.; Wand, A. J., Solution structure of horse heart ferricytochrome c and detection of redox-related structural changes by highresolution 1H NMR. Biochemistry 1996, 35, 12275-12286. 53. Banci, L.; Bertini, I.; Huber, J. G.; Spyroulias, G. A.; Turano, P., Solution structure of reduced horse heart cytochrome c. J. Biol. Inorg. Chem. 1999, 4, 21- 31. 54. Smith, D. W.; Williams, R. J. P., The Spectra of Ferric Haems and Haemoproteins. Structure and Bonding 1969, 7, 1-45. 55. Cheesman, M. R.; Greenwood, C.; Thomson, A. J., Magnetic Circular Dlchrolsm of Hemoproteins. Advances In Inorganic Chemistry 1991, 36, 201-255. 56. Spiro, T. G.; Strekas, T. C., Resonance Raman spectra of heme proteins. J. Am. Chem. Soc 1974, 96, (2), 338-345. 57. Oellerich, S.; Wackerbarth, H.; Hildebrandt, P., Spectroscopic Characterization of Nonnative Conformational States of Cytochrome c. J. Phys. Chem. B 2002, 106, 6566-6580. 58. Spiro, T., Resonance Raman Spectroscopic Studies of Heme Proteins. Biochim. Biophys. Acta 1975, 416, 169-189. 59. Schejter, A.; George, P., The 695-mμ Band of Ferricytochrome c and Its Relationship to Protein Conformation. Biochemistry 1964, 3, (8), 1045-1049. 60. Filosa, A.; English, A. M., Probing local thermal stabilities of bovine, horse, and tuna ferricytochromes c at pH 7. J. Biol. Inorg. Chem. 2000, 5, (4), 448-454. 61. Ye, T.; Kaur, R.; Senguen, F. T.; Michel, L. V.; Bren, K. L.; Elliott, S. J., Methionine Ligand Lability of Type I Cytochromes c: Detection of Ligand Loss Using Protein Film Voltammetry. J. Am. Chem. Soc. 2008, 130, (21), 6682-6683.

25 62. Raphael, A. L.; Gray, H. B., Axial Ligand Replacement in Horse Heart Cytochrome c by Semisynthesis. Proteins: Structure, Function, and Genetics 1989, 6, (3), 338-340. 63. Rovira, C.; Carloni, P.; Parrinello, M., The Iron-Sulfur Bond in Cytochrome c. J. phys. Chem. B 1999, 103, 7031-7035. 64. Liu, X.; Kim, C. N.; Yang, J.; Jemmerson, R.; Wang, X., Induction of apoptotic program in cell-free extracts: requirement for dATP and cytochrome c. Cell 1996, 86, 147-157. 65. Ow, Y.-L. P.; Green, D. R.; Hao, Z.; Mak, T. W., Cytochrome c: functions beyond respiration. Nat. Rev. Mol. Cell Biol. 2008, 9, 532-542. 66. Karu, T. I.; Pyatibrat, L. V.; Afanasyeva, N. I., Cellular effects of low power laser therapy can be mediated by nitric oxide. Lasers Surg. Med. 2005, 36, (4), 307-314. 67. http://www.sigmaaldrich.com/life-science/metabolomics/enzyme- explorer/learning-center/cytochrome-c.html. 68. Malmstrom, B. G., Rack-induced bonding in blue-copper proteins. Eur. J. Biochem. 1994, 223, (3), 711-718. 69. Williams, R. J. P., Energised (entatic) states of groups and of secondary structures in proteins and metalloproteins. FEBS 1995, 234, (2), 363-381. 70. Gray, H. B.; Malmstrom, B. G.; Williams, R. J. P., Copper coordination in blue proteins. J. Biol. Inorg. Chem. 2000, 5, (5), 551-559. 71. Pauleta, S. R.; Guerlesquin, F.; Goodhew, C. F.; Devreese, B.; Beeumen, J. V.; Pereira, A. S.; Moura, I.; Pettigrew, G. W., Paracoccus pantotrophus Pseudoazurin Is an Electron Donor to Cytochrome c Peroxidase. Biochemistry 2004, 43, (35), 11214-11225. 72. Hwang, H. J.; Lu, L., pH-dependent transition between delocalized and trapped

valence states of a CuA center and its possible role in proton-coupled electron transfer. Proc. Natl. Acad. Sci. U.S.A. 2004, 101, 12842-12847. 73. Lukoyanov, D.; Berry, S. M.; Lu, Y.; Antholine, W. E.; Scholes, C. P., Role of Coordinating Histidine in Altering the Mixed Valency of CuA: An Electron Nuclear Double Resonance-Electron Paramagnetic Resonance Investigation. Biophys. J. 2002, 82, 2758-2766.

26 74. Hantgan, R. R.; Taniuchi, H., Conformational dynamics in cytochrome c. A fragment exchange study. J. Biol. Chem. 1978, 253, (15), 5373-5380. 75. Schejter, A.; Plotkin, B., The binding characteristics of the cytochrome c iron. Biochem. J. 1988, 255, (1), 353-356. 76. Cheng, M.-C.; Rich, A. M.; Armstrong, R. S.; Ellis, P. J.; Lay, P. A., Determination of Iron Ligand Bond Lengths in Ferric and Ferrous Horse Heart Cytochrome c Using Multiple-Scattering Analyses of XAFS Data. Inorg. Chem. 1999, 38, (25), 5703-5708.

27

Chapter 2

Thermodynamic Equilibrium between Blue and Green Copper Sites and the Role of the Protein in Controlling Function*

* Reproduced in part with permission from Proc. Natl. Acad. Sci., 2009, 106, 4969. Copyright © 2009 the National Academy of Science.

28 2.1 Introduction Blue copper (also called type 1, T1 copper)1 active sites are found in a variety of proteins including plastocyanins that undergo rapid electron transfer (ET).2-5 The first crystal structures were available for plastocyanin from poplar leaves showing an unusual active site geometry, which was a distorted tetrahedron with a short Cu-SCys bond at 2.1 6-8 2+ Å, a long Cu-SMet bond at 2.8-2.9 Å and two Cu-NHis bonds at ~ 2.0 Å. Normally Cu complexes are square planar due to the Jahn-Teller effect9 and it was thought that the protein constrained the copper site structure to be in between that of Cu+ (tetrahedral) and Cu2+ (square planar) to facilitate ET. This concept of the protein constraint on the copper site was referred to as an entatic or rack induced state.10-12 Associated with this unusual active site structure were novel spectral features relative to normal tetrahedral Cu2+ complexes.1, 13-15 In particular, the blue copper site exhibited an intense lower energy thiolate π to Cu charge transfer (CT) transition (and a weak higher energy σ CT transition) reflecting a π bonding interaction of the thiolate with the Cu2+ of the blue copper center (π ground state, Scheme 2.1A).16, 17 Normal tetragonal cupric complexes have σ ligand- metal bonds that result in an intense higher energy σ and weak lower energy π ligand to metal CT transition(σ ground state, Scheme 2.1B).5

2 2 Scheme 2.1 Schematic representation of Cu dx -y and SCys p-orbital ground states of A) blue copper (plastocyanin, π ground state), resulting in the high energy weak σ and low energy strong π CT transitions and B) normal and green copper (NiR, σ ground state), resulting in the high energy strong σ and low energy weak π CT transitions.

29 However the idea of protein constraint on the oxidized site was questioned both in calculations and spectroscopic studies. Total energy calculations were used to argue that the blue copper ligand set gives a structure similar to the protein site on geometry optimization.18 Photoemission spectroscopic studies on models 19 of the reduced site indicated that the geometric changes predicted based on the change in the electronic structure upon oxidation were equivalent to those observed experimentally in crystal + structure and EXAFS data. In fact it was suggested that the long Cu -SMet bond present in the reduced site was a constraint of the protein. The bond is ~ 2.9 Å in the crystal structure of the reduced site20 rather than the expected distance of ~ 2.3 Å.19, 21 The long thioether bond reduces its donor interaction with the copper, which is compensated by the

thiolate leading to the short Cu-SCys bond of 2.1 Å in the blue copper sites. This long thioether /short thiolate would eliminate all orbital degeneracy of the oxidized site thus, there would be no Jahn Teller distortion from the tetrahedral to a tetragonal structure. The above “coupled distortion” model provided an explanation for a series of blue copper related proteins (i.e. same Cys, Met, 2His ligand set) that changed from blue to green in color in going from plastocyanin to nitrite reductase (NiR).5, 22, 23 In the latter

protein the Cu-SMet bond has decreased to 2.45 Å, the Cu-SCys bond has elongated to 2.21 Å and the S-Cu-S plane has rotated relative to the N-Cu-N plane to a more tetragonal structure.24 Associated with this distortion the thiolate now σ-bonds to the Cu2+ and the CT spectrum exhibits a higher energy intense σ CT (reflecting a σ ground state) and a lower energy weaker π CT transition (Scheme 1B), similar to normal tetragonal Cu2+ complexes but inverted relative to the blue copper site in plastocyanin.22 In our past studies of the green copper site in NiR we observed a large temperature dependence of its absorption spectrum.22 In a study on a loop mutant of Amicyanin, a more limited temperature dependence was observed and assigned to the Boltzmann population of two electronic minima.25 In density functional theory (DFT) studies with a B3LYP hybrid functional, Ryde and coworkers predicated two minima, one similar to the blue copper site and a second similar to the green copper site at comparable energy.26 However this is dependent on the functional and amount of Hartree Fock mixing.5 Based on resonance Raman (rR) data on NiR, two species were suggested to be present,27 however, this was refuted by later rR data on the same enzyme.28

30 In this study we explore the temperature dependence of the green copper site in NiR using absorption and rR spectroscopies combined with DFT calculations. These studies show that in contrast to the earlier proposal on the amicyanin mutant, the temperature dependence in NiR reflects a thermodynamic equilibrium of two species driven by entropy. These studies demonstrate that the thioether-copper bond is unconstrained in NiR and allow us to experimentally determine the strength of this bond. Extension to and the contrast of these results to the behavior of the blue copper site in plastocyanin define the constraints imposed on the blue copper site by the protein and thus provide insight into the entatic/rack nature of the blue copper site and its contribution to ET function.

2.2 Experimental Methods 2.2.1 Materials All reagents were of the highest grade commercially available and used without further purification. Rhodobacter sphaeroides (Rs) NiR was isolated and purified (pH~7.2) as previously reported.29-31 Glassed samples for low temperature absorption

experiments were prepared by adding 50% (v/v) buffer/glycerol-(O-d)3 or buffer/ethylene

glycol . Addition of glycerol/ethylene glycol had no effect on the EPR spectrum. The approximate concentration of samples used for spectroscopy was 0.5mM in phosphate buffer, pH ~7.1. Temperature independent buffer32 was also used for spectroscopic studies and had identical results as phosphate buffer.

2.2.2 Spectroscopic Studies Low-temperature absorption spectroscopy (below 0 0C) was performed on a double beam spectrophotometer (Cary 500) using a liquid helium cryostat (Janis Research Super Vari-Temp). Higher temperature absorption spectra (above 0 0C) were collected using a UV-Vis diode array spectrophotometer (Agilent 8453). A buffer/ethylene glycol (40:60 mixture) solution was used to lower the freezing temperature in order to obtain low temperature absorption data of samples in the solution state. Raman spectra were obtained using a series of lines from Kr+ (Coherent 190CK) and Ar+ (Coherent Sabre 25/7) ion lasers with incident power ranging from 10 to 50 mW in an ~135° backscattering configuration. Scattered light was dispersed through a triple

31 monochromator (Spex 1877 CP, with 1200, 1800, and 2400 groove/mm gratings) and detected with a back-illuminated CCD camera (Princeton Instruments ST-135). A Dye laser (Rhodamine 6G, Coherent 599) was used for other spectral regions. Samples contained in NMR tubes were immersed in a liquid nitrogen finger dewar for 77 K measurements or kept at room temperature. Background spectra of charcoal in the same NMR tube were used for baseline subtraction. Solution samples (for resonance Raman data) in buffer/ethylene glycol were first cooled in a dry ice/methanol bath (-80 0C) and then frozen in liquid nitrogen.

2.2.3 Computational Details The oxidized and reduced T1 copper site was modeled by a 33-atom cluster in which dimethyl thioether, methyl thiolate and imidazole replaced the methionine, cysteine and histidine residues, respectively. An extended 48-atom model (one ethyl chain was added to each imidazole, and the methyl thiolate was replaced by an ethyl thiolate) was employed to impose the protein β-carbon constraints present in the X-ray structure of Plastocyanin or NiR (PDB ID: 1PLC7 and 1OE133, respectively). The optimized geometries with β-carbon constraints reproduced the crystallographic results. The axial thioether ligand was not constrained to model the green copper site and was constrained at specific distances for blue copper site geometry optimizations. Geometry optimizations on the 33-atom model reasonably reproduced the geometric features of the 48-atom model, and were used in frequency calculations. Protein solvation effects on the copper active site were estimated using a polarized continuum method (PCM)34 with ε = 4.0. DFT calculations (spin unrestricted for Cu2+) were performed using Gaussian 03.35 Geometries were optimized using a hybrid functional involving BP8636, 37 with 38% Hartree–Fock exchange, tight SCF convergence criteria (10-8 au) and a mixed triple- 38 zeta/double zeta (TZVP on Cu and S and 6-31G* on the other atoms) basis set. Wave function stability checks were performed to confirm that the calculations corresponded to the ground state. For the potential energy surfaces, single point calculations in the PCM were applied to the geometry optimized vacuum structures.

32 2.3 Results 2.3.1 Low Temperature Spectroscopy The low temperature absorption spectrum of NiR (Figure 2.1, green) is dominated by an intense band at ~21700 cm-1 (460 nm) from the T1 site that has been assigned as the 22 SCys Æ Cu σ charge transfer (CT) transition. A weaker SCys Æ Cu π CT transition is also observed at ~17500 cm-1 (570 nm). This is followed by lower energy ligand field -1 22 transitions. There is also a SMet Æ Cu CT transition at 25600 cm (390 nm).

)

-1 cm -1 M ( ε

Energy (cm-1) Figure 2.1 Absorption spectra of resting WT NiR at 7 K (green) room temperature (blue).

The rR spectrum of NiR excited at 458 nm (into the σ CT band) at 77 K is presented in Figure 2.2A (bold green). It has three intense peaks at 364, 376 and 403 cm-1 as reported previously.39 When NiR is excited at 568 nm (into the π CT transition) at 77 K (Figure 2.2B, bold green), the same vibrations (364, 376 and 403 cm-1) are enhanced but with a different intensity ratio. This shows that the weak band at 570 nm in the absorption spectrum at low temperature originates from the same species as the 460 nm band, indicating that there is a single species present at 77 K in NiR. This is the green copper center previously defined by spectroscopy22 and crystallography24 (reported at 5 K and 100 K, respectively) for this enzyme. Note that the different relative intensities of the vibrations obtained with 568 nm excitation compared to those obtained by excitation into the 458 nm band reflect differences in the π and σ CT excited state distortions.40 The vibrations of the green copper species that are resonance enhanced reflect kinematic mixing of the Cu-SCys vibration over several protein modes, with an effective frequency

33 41 (intensity weighted average at 568 nm excitation) of the Cu-SCys bond, i.e. <νCu-S> of ~388 cm-1. This is lower than the value for the blue copper sites in Plastocyanin in which -1 <νCu-S> = 403 cm , reflecting a weaker Cu-SCys bond for the green copper site. The

weaker Cu-SCys reflects the stronger donor interaction of the axial Met ligand with the

copper in the green site of NiR relative to Plastocyanin (Cu-SMet bond: 2.45 Å in green copper; 2.8 Å in plastocyanin).

A 364 403

y 376

Raman Intensit

-1 Raman Shift (cm )

B 408

y

420 403 375 376 Raman Intensit 364

Raman Shift (cm-1) Figure 2.2 Resonance Raman spectra of resting WT NiR excited at A) 458 nm and B) 568 nm. 77 K data are in bold lines and 25 0C in dotted lines. Inset: resonance Raman spectrum of the pure Blue component of NiR (blue spectrum, obtained by subtracting the spectrum of the green species at 77 K, 568 nm from the 25 0C, 568 nm spectrum and renormalization) and of the M182T variant of NiR (red spectrum).

34

Figure 2.3 Resonance Raman profiles overlaid with the absorption spectra. The 364 cm-1 peak associated with the green copper species, while the 420 cm-1 peak reflects the blue copper component of NiR.

2.3.2 Room Temperature Spectroscopy The absorption spectrum at room temperature (Figure 2.1, blue) is vastly different from the low temperature spectrum. There is a significant redistribution of intensity with the band at ~570 nm having gained intensity and red shifted to ~600 nm, while the absorption band at ~460 nm has lost ~50% of its intensity. As a large temperature dependence was observed in the absorption spectrum of NiR, rR data were also collected at 25 0C. The rR spectrum of NiR, excited at 458 nm at room temperature, has the same features as that obtained at 77 K, but with much lower intensity (Figure 2.4A, dotted green), consistent with the loss of absorption intensity of the 460 nm band. However when NiR is excited at 568 nm at 25 0C, the spectrum changes significantly relative to the low temperature rR spectrum (Figure 2.2B, blue). It now shows an intense peak at 408 cm-1 with a shoulder at 420 cm-1. There is also a broad weak feature at ~375 cm-1 which lacks the 364 cm-1 feature of the low temperature green copper site. The vibrational features enhanced by the 568 nm excitation at room temperature are different from those observed at low temperature, for green copper

35 species with the same excitation line and indicate that a second species, in addition to the green copper species, has grown in at elevated temperature. Subtraction of the low temperature green copper spectrum (scaled down by 50%, vide infra) from the room temperature data and renormalization yields the spectrum of the

pure high temperature component (Figure 2.2B, inset, blue) which has <νCu-S> of ~407 -1 -1 cm . This is significantly increased from the <νCu-S> ~388 cm for the green copper species, indicating that the species that grows in at elevated temperatures has a stronger

Cu-SCys bond. A mutant of NiR has been reported in which the axial Met ligand has been replaced by a threonine residue (M182T).29, 39 This mutation results in a blue T1 copper -1 site with <νCu-S> of ~408 cm . The rR spectrum of the high temperature species (Figure 2.2B, inset, bold blue) is essentially the same as that of the mutant. Note that the rR profile (Figure 2.3) of the 364 cm-1 mode (associated with only the green copper component) indicates that it is strongly enhanced by the 460 nm CT band and only weakly enhanced by the 600 nm CT band. This profiles well with the low temperature absorption spectrum of the green species, which has an intense σ CT band at 460 nm and a weak π CT band at 570 nm. It decreases in intensity with increasing temperature reflecting the decrease in the amount of green site present. The profile for the 420 cm-1 mode, characteristic of the high temperature species shows that it is strongly enhanced by the 600 nm π CT band and not by the 460 nm σ CT band. The very different Raman profile of the 420 cm-1 vibration indicates that it must originate from a second blue copper component of the T1 site in NiR. The high vibration frequency and the intense π CT transition for the room temperature species show that it is a blue copper site as in Plastocyanin that has a

relatively strong Cu-SCys bond reflecting a relatively weak Cu-S(Met) bond. The <νCu-S> of ~407 cm-1 for the blue species in NiR is higher than that of Plastocyanin, which has -1 axial Cu-SMet ~ 2.8 Å ( <νCu-S> of ~403 cm ) but lower than that of fungal laccases -1 5 5, 22, 23 (<νCu-S> of ~413 cm ) that has no axial ligand. From the coupled distortion model, there is a direct correlation between the observed <νCu-S> associated with the Cu-SCys bond and the Cu-SMet bond length. The longer Met has a decreased donor interaction with

the copper that is compensated by an increase in the Cu-SCys bond. The <νCu-S> of the

36 blue component of NiR indicates that its Cu-S(Met) bond length/strength is in-between that of Plastocyanin and fungal laccases. Thus the rR data clearly show the presence of two distinct chromophores (green and blue copper sites) at room temperature and only one chromophore (green copper) at low temperature.

2.3.3 Thermodynamics A series of solution absorption spectra were collected over a wide range of temperatures (215 K-323 K, Figure 2.4A, B). The data show that the intensity of the 460 nm band decreases while the intensity of the 600 nm band increases with increasing temperature. This could indicate a Boltzmann population of a second local minimum having a different geometry of the T1 site (as has been proposed in ref. 25) or the presence of two T1 sites in NiR in a thermodynamic equilibrium one site is enthalpically favored and the other is entropically favored. Using the absorption spectrum of a pure green site (absorption spectrum at 7 K in Figure 2.1 green) and the data for the M182T mutant of NiR as representative of a blue site (The M182T of NiR is a mutant with the axial methionine at the T1 Cu center is replaced by weaker ligand threonine),39 the relative amounts of the blue and the green species were obtained from the temperature dependent absorption spectra of NiR (Figure 2.4). Note that at 550C (328 K) there is 65% blue copper sites (which continues to increase with temperature) and 35% green sites demonstrating that there is a thermodynamic equilibrium and not a Boltzmann population of two levels, where for the latter a maximum of 50% population of the higher energy level is possible. The relative populations of the blue and the green copper sites have been used to obtain the

equilibrium constant, Keq, for the green to blue conversion at various temperatures

(Equation 2.1). A plot of the lnKeq vs 1/T (Figure 2.5, 215-328K) resulted in a ∆H ≈ 4.6 kcal/mol from the slope and ∆S ≈ 15 cal/molK (T∆S = 4.5 kcal/mol at 298 K) from the intercept, for the green to blue component conversion of NiR. Thus the green species is enthalpically favored while the blue species is entropically favored. As a consequence, below 215 K (i.e. T∆S = 3.2 kcal/mol), only the green species is present and contributes

to the absorption and resonance Raman spectra as Keq is dominated by the ΔH term. The fraction of the blue form increases with temperature and at room temperature (298 K) the

37 Keq between these two forms is close to unity, i.e. ∆G = 0. These observed ΔH and ΔS for the blue/green T1 copper site equilibrium are analyzed using DFT calculations in section 3.4.

Keq Green Blue Equation 2.1

A

) -1 cm

-1 (M ε

Energy (cm-1)

B

) -1 cm

-1 (M ε

Energy (cm-1) Figure 2.4 Temperature dependence of the absorption spectra of resting WT NiR A) in buffer with ethylene glycol (40:60) and B) in buffer solution.

38

Figure 2.5 Plot of lnKeq vs. 1/T, where Keq ~ 1 at 298 K.

2.4 Analysis 2.4.1 Oxidized Site The variable temperature spectroscopic data indicate that there are two forms of

oxidized T1 sites of NiR that are in a thermodynamic equilibrium. They differ in Cu-SCys bond strength based on the rR data. One has a dominant SCys → Cu σ CT and a relatively -1 weak Cu-SCys bond ( <νCu-S(Cys)> = ~388 cm ) reflecting a short Cu-SMet bond consistent with the structure of the green copper center reported in the low temperature crystal 42 structure. The Cu-SCys bond of the second species, which appears at higher temperature -1 and has a dominant SCys → Cu π CT transition, is stronger ( <νCu-S(Cys)> = ~ 413 cm ) than that of the green copper site. In fact it is even slightly stronger than that of the blue 43 copper site in plastocyanin (which has a Cu-SMet distance of 2.8 Å) , but weaker than that of the blue copper sites in Fet3p and fungal laccases that have no axial ligand.44-46

This indicates that the blue component in NiR has a very weak Cu-SMet bond. Note that the crystal structure of a NiR with atomic resolution (PDB:1OE147, blue absorption at room temperature, crystal structure at 100K) reveals that the thioether has two

39 conformations: one with a Cu-SMet distance of 2.45 Å (major) and a second with 4.26 Å (minor). The thermodynamic parameters obtained in the results section indicate that the green species of NiR is enthalpically favored while the blue species of NiR is entropically favored. DFT calculations are used here to evaluate the bond strength of this axial thioether ligand, and its contribution to the enthalpy and entropy of the oxidized T1 copper site. The geometry optimized structure with no constraints (Figure 2.6A green) reproduced the crystal structure (PDB ID: 1ZV2 42 at 100K, Figure 2.6A red) of the green copper site of Rs NiR (the deviations of the calculated bond distances and angles are within 0.04Å and 5°, repectively). It has a flattened tetrahedral (toward square planar) structure with a

Cu-SCys distance of 2.18 Å and a Cu-Sthioether distance of 2.41 Å (Figure 2.6A green).

Upon elimination of the thioether ligand, the reoptimized Cu-SCys bond length decreases by 0.05 Å. The copper is now in a trigonal ligand field, and has a structure in reasonable agreement (within 0.04Å and 3°) with the copper sites found in the fungal laccases and Fet3p. These two structures can be considered to be the limits of the coupled distortion

coordinate (i.e. unconstrained relatively strong Cu-SMet bond and no Cu-SMet bond, respectively).

The calculated total electronic energy for Cu-Sthioether bond dissociation for the green copper site is 6.8 kcal/mol. Upon loss of the thioether (Equation 2.2), the enthalpy change is 5.4 kcal/mol (Table 2.1, top row, in parenthesis where solvent effects were included) and the entropy gain is 39 cal/(mol⋅K) (TΔS = 11.7 kcal/mol at 298 K, Table 2.1 top row). The calculated enthalpy change is consistent with the experimental results (ΔHexp = 4.6 kcal/mol). However, the calculated entropy gain is larger than experimentally observed (TΔS = 4.5 kcal/mol at 298 K). This reflects the fact that in the NiR protein the thioether still remains near the active site as it is covalently linked into the polypeptide chain. This is consistent with the lower <νCu-S(thiolate)> for the blue component of NiR relative to the fungal laccases (no axial ligand) reflecting the presence of a very weak Cu-

SMet bond.

Keq [Cu-Met]Green [Cu]Blue + Met Equation 2.2

40 The solid line in Figure 2.7 shows the effect of elongation of the Cu-Sthioether bond from 2.4 Å of the green site to longer distances (i.e. constrained Cu-SMet with optimization of the rest of the structure). By 4 Å (modeling the blue component of NiR) the Met dissociation energy is < 2 kcal/mol. Geometry optimization of the structure with the Cu-Sthioether bond at 4.0 Å leads to a trigonal ligand field with a shorter copper thiolate bond (dCu-S(Cys) = 2.13 Å, Figure 2.6A blue) relative to the calculated unconstrained green

site (dCu-S(Cys) = 2.18 Å). In going to the calculated 4.0 Å Cu-SMet blue structure, the SCys-

Cu-SMet plane rotates further out of the NHis-Cu-NHis plane, which reflects a larger distortion toward a tetrahedral structure. This 4.0 Å blue species has essentially the same structure as the copper site without an axial ligand. Upon loss of the axial ligand from the 4 Å structure, the entropy change is 20 cal/(mol⋅K) (TΔS = 6.0 kcal/mol at 298 K, Table 2.1 bottom row, note that this calculation does not include solvation) which reflects the gain of rotational and translational freedom. This is about half the calculated entropy change of 39 cal/(mol⋅K) (TΔS = 11.7 kcal/mol at 298 K, Table 2.1 first row) for Met loss from the green species. In going from the green species to the 4.0 Å component, the calculated entropy change is 5.7 kcal/mol (i.e. the difference between the above two entropy terms). This is now consistent with the experimentally determined entropic contribution for the green to blue conversion in NiR (TΔSexp = 4.5 kcal/mol), which favors the blue species at elevated temperature. This entropy gain reflects an increase in

the number of low energy normal modes in the 4.0 Å blue species (dCu-S(thioether = 4.0 Å) relative to the green species (dCu-S(thioether = 2.4 Å). The population of the low energy modes entropically stabilizes the 4.0 Å blue species.

We now consider the blue copper site in plastocyanin with the Cu-SMet constrained at 2.8 Å. The optimized structure reproduces the classic blue copper site in Plastocyanin,

with the copper in a trigonal ligand field and a short copper thiolate bond (dCu-S(Cys) = 2.15 Å). The calculated enthalpy and entropy changes for the Met loss for the plastocyanin blue copper site are essentially the same as for the green species (Table 2.1, first and second row) while plastocyanin is enthalpically destabilized by ~ 1 kcal/mol. Importantly, if plastocyanin were unconstrained one would expect a similar behavior to that of NiR with loss of the axial ligand due to increased entropy at elevated temperature. However, experimentally, the oxidized blue copper sites in plastocyanins and related proteins show

41 no temperature dependent absorption behavior. This reflects constraints by the protein environment on the plastocyanin active site, referred to as the entatic/rack state (vide infra).

Scys

Scys A B

Smet

Smet

Nhis2

Nhis2 Nhis1 Nhis1

Figure 2.6 The geometry of A) oxidized and B) reduced type 1 Cu sites (only Cu and coordinated N, S atoms are shown for clarity, and the arrows indicate the major angle changes). Red – crystal structure of NiR at 100 K, short Cu-SMet bond length; Green –

fully geometry optimized structure, short Cu-SMet bond length; Blue – partial geometry

optimized structure with Cu-Sthioether distance constrained at 4 Å of the long bond length).

6

4

CuII CuI

2

kcal/mol ΔΕ

0 234567891011 Cu-S distance (Å)

Figure 2.7 Potential energy surface (electronic energies) as a function of Cu-Smet distance (solvation included using a PCM with ε = 4.0). Solid line-oxidized Cu site, dashed line- reduced Cu site.

42 Table 2.1 Calculated free energies for loss of an axial Met ligand (constrained at different

Cu-SMet distances) from an oxidized copper site at room temperature.

ΔE c ΔHd ΔS TΔS ΔG Oxidized Cu Site Models (kcal/mol) (kcal/mol) (cal/mol⋅K) (kcal/mol) (kcal/mol)

+ Cu(Im)2(SCH3)S(CH3)2] a 10.4 (6.8) 9.1 (5.4) 39 (42) 11.7 (12.5) -2.6 (7.1) dCu-S(thioether) = 2.4Å

+ [Cu(Im)2(SCH3)S(CH3)2] b 8.9 (5.2) 7.7 35 10.4 -2.7 dCu-S(thioether) = 2.8Å

+ [Cu(Im)2(SCH3)S(CH3)2] b 6.2 (1.9) 5.5 20 6.0 -0.5 dCu-S(thioether) = 4.0Å

a. The dCu-S(thioether) distance is obtained from full geometry optimization.

b. The dCu-S(thioether) distance is constrained as indicated. c. Total electronic energy change for loss of Met ligand. d. Also from c with zero-point and thermal corrections included

H = Eelectronic+EZPE+Evib+Erot+Etrans+RT, ZPE=zero point energy The results of PCM calculations are shown in the parentheses.

2.4.2 Reduced site Full optimization of the reduced model of the green copper site gave a tetrahedral

ligand field with a Cu-Sthioether distance of 2.5 Å (Figure 2.6B green), similar to the crystal structure of the reduced NiR at 100 K42 (Figure 2.6B red). Removal of the thioether ligand and reoptimization gives a geometry very similar to that of the reduced

T1 site in Fet3p, with the copper ion in the NHis-NHis-SCys plane. Elongation of the thioether bond results in a very small energy change (ΔE ≈ 1 kcal/mol, Figure 2.7 dashed line). This leads to a very small enthalpy change ((ΔH ≈ 1 kcal/mol) associated with the loss of an axial Met bond. However, the entropy change for this process is similar to that of the oxidized copper site (TΔS = 10.9 kcal/mol at 298K, Table 2.2, first row). Further,

in comparing the results for the fully optimized structure to those for the constrained dCu-

S(thioether) = 4.0 Å structure, the entropy change is ~ 4.5 kcal/mol at 298 K ((10.9 kcal/mol –(6.4 kcal/mol), Table 2.2, top and bottom rows, TΔS column, note these calculations do

43 not include solvation), indicating a similar entropy change in going from a green to a blue site in the reduced NiR as determined above for the oxidized site. Thus, for the reduced site the entropy term should be dominant and the Met ligand should not be bonded to the copper at room temperature. A low ΔH and large TΔS is also calculated for loss of a Met ligand for the reduced blue copper site in Plastocyanin (Table 2.2, second row, the values for the plastocyanin site (2.9 Å) to convert to a 4 Å Cu-S(Met) distance (as in NiR) are ΔH = 0.6 kcal/mol, TΔS = 3.7 kcal/mol). The fact that this does not occur based on the room temperature crystal structure20, 48, again reflects a protein constraint on the reduced copper site in plastocyanin.

Table 2.2 Calculated free energies for loss of an axial Met ligand (constrained at different

Cu-SMet distances) from a reduced copper site at room temperature.

Reduced Cu Site ΔEc ΔHd ΔS TΔS ΔG Models (kcal/mol) (kcal/mol) (cal/mol⋅K) (kcal/mol) (kcal/mol)

Cu(Im)2(SCH3)S(CH3)2] a 5.8 (0.5) 4.7 (0) 37 (38) 10.9 (11.3) -6.2 (11.9) dCu-S(thioether) = 2.5Å

[Cu(Im)2(SCH3)S(CH3)2] b 5.0 (0.4) 3.8 34 10.1 -6.3 dCu-S(thioether) = 2.9Å

[Cu(Im)2(SCH3)S(CH3)2] b 4.6 (0) 3.2 21 6.4 -3.2 dCu-S(thioether) = 4.0Å

a. The dCu-S(thioether) distance is obtained from full geometry optimization.

b. The dCu-S(thioether) distance is constrained as indicated. c. Total electronic energy change for loss of Met ligand. d. Also from c with zero-point and thermal corrections included

H = Eelectronic+EZPE+Evib+Erot+Etrans+RT, ZPE=zero point energy The results of PCM calculations are shown in the parentheses.

44 2.5 Discussion The oxidized green copper site of NiR at low temperature is tetragonally distorted with a strong predominantly σ bonding interaction of the thiolate ligand with the copper. Temperature dependent rR and absorption data indicate that the thioether ligand in NiR partially dissociates at higher temperatures resulting in a trigonal blue copper site, which has a strong π bonding interaction of the thiolate ligand with the copper. Experimentally the Cu2+-thioether bond strength in NiR has been determined here to be 4.6 kcal/mol, and at room temperature the gain in entropy (T∆S = 4.6 kcal/mol) on thioether dissociation approximately equals the enthalpy change due to loss of this ligand, which results in the formation of a ~1:1 mixture of green and blue copper species. This indicates that the T1 site in NiRs is unconstrained, allowing the association and dissociation of the Met ligand at low and high temperatures, respectively. The relatively unconstrained nature of the T1

site of NiR may reflect the long Cys-Xn-His loop, which would allow flexibility and limit intra-loop H-bonding. DFT calculations reproduce the experimentally observed ∆H and ∆S of this process. Further, in the reduced state, calculations indicate that the axial Met- Cu bond is very weak with a ∆H of ~1 kcal/mol for dissociation. However, the T∆S is comparable to that of the oxidized site (~4.6 kcal/mol observed experimentally, at room temperature) and hence is the dominant contributor to ∆G at room temperature for the reduced site. Thus, while the oxidized form exists as a 1:1 mixture of blue and green species, the reduced T1 site in NiR likely exists as one species with no thioether-Cu bonding interaction at room temperature. It is important to emphasize that our model for the large temperature dependence in NiR is conceptually different from that invoked previously to explain the more limited temperature dependence of absorption data in a loop mutation of amicyanin.25 It had been proposed that there is a low lying second electronic minimum of the T1 site that becomes Boltzmann populated with increasing temperature. However, this only allows for the maximum of a 1:1 population of the two species at high temperatures. Alternatively, our experimental results show that at elevated temperatures (550C, 328 K) the population of the blue site is 65±5%, indicating that there is in fact a thermodynamic equilibrium between the two species.

45 Classic blue copper sites like plastocyanin have elongated Met-Cu bonds at ~2.8 Å and do not show the large temperature dependent behavior observed for NiR. These results show that in contrast to the green/blue equilibrium of sites in NiR, the thioether is constrained in plastocyanin. Importantly, in plastocyanins the Met ligand stays bonded at room temperature in both oxidation states. Having a constrained thioether ligand has been referred to as an entatic11 or rack state10 for these classic blue copper sites. From the crystal structures7, 20 and spectroscopies19 the thioether is bonded to the copper at similar distances in both oxidation states. Thus the protein opposes the entropy effect of loss of a relatively weak Met ligand. From Figure 2.7 and Tables 1 and 2, the Met has a stronger bonding interaction with the oxidized than the reduced site. This can lead to a greater than 200 mV lowering of the reduction potential in these classic blue copper sites relative to sites without the axial methionine ligand. Indeed the fungal laccases and Fet3p have reduction potentials that range from 470-770 mV, higher than those of plastocyanin (370 mV) and azurin (310 mV). The green T1 copper sites in NiRs have a dominant σ ground state at low temperature and at elevated temperatures a blue T1 copper site is also present with a dominant π ground state (Scheme 1). This is in contrast with other “perturbed” T1 copper sites like cucumber basic blue (CBB) and pseudoazurin (PA) that have both σ and π character,23 but no temperature dependent behavior.40 This implies that there is a single species at all temperatures and the thioether remains bound to the copper in these active sites, though at a shorter distance relative to plastocyanin, which leads to the mixed π/σ character. Thus these sites are also constrained and represent different points along the coupled distortion coordinate. This can play a role in fine tuning the reduction potential of these

sites as these have different ΔEs of the Cu-SMet bond (Figure 2.7). In summary, the large temperature dependence observed in the absorption and rR spectra in green NiRs reflect an unconstrained T1 site where entropy favors the dissociation of the relatively weak axial Met ligand at higher temperature. The ∆H and T∆S of Met binding are comparable in the oxidized state (~4.5 kcal/mol), whereas, the ∆H in the reduced state is much lower. For the blue copper sites in the plastocyanins and azurins, the protein overcomes the entropy increase associated with ligand loss. This

46 stabilizes the oxidized more than the reduced site and can tune the reduction potential down by greater than 200 mV.

2.6 Acknowledgment

This research was supported by NSF Grant CHE 0446304 (E.I.S) and NIH EB00326929 (C.P.S). S. G. and X. X. are grateful for William S. Johnson fellowships.

2.7 References 1. Malkin, R.; Malmstrom, B. G., The State and Function of Copper in Biological Systems. Adv. Enzymol. Relat. Areas Mol. Biol. 1970, 33, 177-244. 2. Adman, E. T., Copper protein structures. Advances in protein chemistry 1991, 42, 145-197. 3. Gray, H. B., Long-range electron-transfer in blue copper proteins. Chemical Society Reviews 1986, 15, (1), 17-30. 4. Holm, R. H.; Kennepohl, P.; Solomon, E. I., Structural and functional aspects of metal sites in biology. Chem. Rev. (Washington, DC, United States) 1996, 96, 2239-2314. 5. Solomon, E. I.; Szilagyi, R. K.; George, S. D.; Basumallick, L., Electronic Structures of Metal Sites in Proteins and Models: Contributions to Function in Blue Copper Proteins. Chem. Rev. (Washington, DC, United States) 2004, 104, (2), 419-458. 6. Colman, P. M.; Freeman, H. C.; Guss, J. M.; Murata, M.; Norris, V. A.; Ramshaw, J. A. M.; Venkatappa, M. P., X-ray crystal structure analysis of plastocyanin at 2.7 Å resolution. Nature 1978, 272, (23 March), 319-324. 7. Guss, J. M.; Bartunik, H. D.; Freeman, H. C., Accuracy and precision in protein stucture analysis: restrained least-squares refinement of the structure of poplar plastocyanin at 1.33 ang. resolution. Acta Crystallographica, Section B: Structural Science 1992, B48, (6), 790-811. 8. Guss, J. M.; Freeman, H. C., Structure of oxidized poplar plastocyanin at 1.6 A resolution. Journal of molecular biology 1983, 169, (2), 521-563.

47 9. Jahn, H.; Teller, E., Stability of degenerate electronic states in polyatomic molecules. Physical Review A 1936, 49, 874. 10. Malmstrom, B. G., Rack-induced bonding in blue-copper proteins. European journal of biochemistry / FEBS 1994, 223, (3), 711-718. 11. Williams, R. J. P., Energised (entatic) states of groups and of secondary structures in proteins and metalloproteins. FEBS 1995, 234, (2), 363-381. 12. Gray, H. B.; Malmstrom, B. G.; Williams, R. J. P., Copper coordination in blue proteins. Journal of Biological Inorganic Chemistry 2000, 5, (5), 551-559. 13. Malmstrom, B. G.; Reinhammar, B.; Vanngard, T., The state of copper in stellacyanin and laccase from the lacquer tree Rhus vernicifera. Biochimica et biophysica acta 1970, 205, (1), 48-57. 14. Gray, H. B.; Solomon, E. I., Electronic structures of blue copper centers in proteins. Metal Ions in Biology. 1981, 3, (Copper Proteins), 1-39. 15. Solomon, E. I.; Lowery, M. D., Electronic structure contributions to function in bioinorganic chemistry. Science (Washington, DC, United States) 1993, 259, (5101), 1575-1581. 16. Gewirth, A. A.; Solomon, E. I., Electronic structure of plastocyanin: excited state spectral features. Journal of the American Chemical Society 1988, 110, (12), 3811-1938. 17. Solomon, E. I.; Penfield, K. W.; Gewirth, A. A.; Lowery, M. D.; Shadle, S. E.; Guckert, J. A.; LaCroix, L. B., Electronic structure of the oxidized and reduced blue copper sites: contributions to the electron transfer pathway, reduction potential, and geometry. Inorg. Chim. Acta 1996, 243, (1-2), 67-78. 18. Ryde, U.; Olsson, M. H. M.; Pierloot, K.; Roos, B. O., The cupric geometry of blue copper proteins is not strained. Journal of Molecular Biology 1996, 261, (4), 586-596. 19. Guckert, J. A.; Lowery, M. D.; Solomon, E. I., Electronic Structure of the Reduced Blue Copper Active Site: Contributions to Reduction Potentials and Geometry. Journal of the American Chemical Society 1995, 117, (10), 2817-2844.

48 20. Guss, J. M.; Harrowell, P. R.; Murata, M.; Norris, V. A.; Freeman, H. C., Crystal structure analyses of reduced (copper I) poplar plastocyanin at six pH values. Journal of Molecular Biology 1986, 192, (2), 361-387. 21. Solomon, E. I.; Randall, D. W.; Glaser, T., Electronic Structures of Active Sites in Electron Transfer Metalloproteins: Contributions to Reactivity. Coordination Chemistry Reviews 2000, 200, 595-632. 22. LaCroix, L. B.; Shadle, S. E.; Wang, Y.; Averill, B. A.; Hedman, B.; Hodgson, K. O.; Solomon, E. I., Electronic structure of the perturbed blue copper site in nitrite reductase: Spectroscopic properties, bonding, and implications for the entatic/rack state. Journal of the American Chemical Society 1996, 118, (33), 7755-7768. 23. LaCroix, L. B.; Randall, D. W.; Nersissian, A. M.; Hoitink, C. W. G.; Canters, G. W.; Valentine, J. S.; Solomon, E. I., Spectroscopic and Geometric Variations in Perturbed Blue Copper Centers: Electronic Structures of Stellacyanin and Cucumber Basic Protein. Journal of the American Chemical Society 1998, 120, (37), 9621-9631. 24. Adman, E. T.; Godden, J. W.; Turley, S., The structure of copper-nitrite reductase from Achromobacter cycloclastes at five pH values, with NO2- bound and with type II copper depleted. The Journal of biological chemistry 1995, 270, (46), 27458-17474. 25. Comba, P.; Mueller, V.; Remenyi, R., Interpretation of the temperature-dependent color of blue copper protein mutants. Journal of Inorganic Biochemistry 2004, 98, (5), 896-902. 26. Pierloot, K.; DeKerpel, J. O. A.; Ryde, U.; Olsson, M. H. M.; Roos, B. O., Relation between the Structure and Spectroscopic Properties of Blue Copper Proteins. Journal of the American Chemical Society 1998, 120, (50), 13156-13166. 27. Dooley, D. M.; Moog, R. S.; Liu, M. P.; Payne, W. J.; LeGall, J., Resonance Raman spectra of the copper-sulfur chromophores in Achromobacter cycloclastes nitrite reductase. Journal of Biological Chemistry 1988, 263, (29), 14625-14628. 28. Han, J.; Loehr, T. M.; Y. Lu, Y.; Valentine, J. S.; Averill, B. A.; Sanders-Loehr, J., Resonance Raman excitation profiles indicate multiple Cys Cu charge transfer

49 transitions in type 1 copper proteins. Journal of the American Chemical Society 1993, 115, (10), 4256-4263. 29. Olesen, K.; Veselov, A.; Zhao, Y.; Wang, Y.; Danner, B.; Scholes, C. P.; Shapleigh, J. P., Spectroscopic, Kinetic, and Electrochemical Characterization of Heterologously Expressed Wild-Type and Mutant Forms of Copper-Containing Nitrite Reductase from Rhodobacter sphaeroides 2.4.3. Biochemistry 1998, 37, (17), 6086-6094. 30. Veselov, A.; Olesen, K.; Sienkiewicz, A.; Shapleigh, J. P.; Scholes, C. P., Electronic Structural Information from Q-Band ENDOR on the Type 1 and Type 2 Copper Liganding Environment in Wild-Type and Mutant Forms of Copper- Containing Nitrite Reductase. Biochemistry 1998, 37, (17), 6095-6105. 31. Zhao, Y.; Lukoyanov, D. A.; Toropov, Y. V.; Wu, K.; Shapleigh, J. P.; Scholes, C. P., Catalytic function and local proton structure at the type 2 copper of nitrite reductase: the correlation of enzymatic pH dependence, conserved residues, and proton hyperfine structure. Biochemistry 2002, 41, (23), 7464-7474. 32. Sieracki, N. Y.; Hwang, H. J.; Lee, M. K.; Garner, D. K.; Lu, Y., A temperature independent pH (TIP) buffer for biomedical biophysical applications at low temperatures. Chemical Communications (Cambridge, United Kingdom) 2008, 7, 823-825. 33. Ellis, M. J.; Dodd, F. E.; Sawers, G.; Eady, R. R.; Hasnain, S. S., Atomic resolution structures of native copper nitrite reductase from Alcaligenes xylosoxidans and the active site mutant Asp92Glu. Journal of Molecular Biology 2003, 328, (2), 429-438. 34. Cossi, M.; Scalmani, G.; Rega, N.; Barone, V., New developments in the polarizable continuum model for quantum mechanical and classical calculations on molecules in solution. J. Chem. Phys. 2002, 117, 43. 35. Frisch, M. J. et. al. Gaussian 03, Revision C.02; Gaussian, Inc.: Wallingford, CT, 2004. 36. Becke, A. D., Density-functional exchange-energy approximation with correct asymptotic behavior. Phys. Rev. A 1988, 38, 3098.

50 37. Perdew, J. P., Density-functional approximation for the correlation energy of the inhomogeneous electron gas. Phys. Rev. B 1986, 33, 8822. 38. Szilagyi, R. K.; Metz, M.; Solomon, E. I., Spectroscopic Calibration of Modern Density Functional Methods Using [CuCl4]2-. Journal of Physical Chemistry A 2002, 106, (12), 2994-3007. 39. Basumallick, L.; Szilagyi, R. K.; Zhao, Y.; Shapleigh, J. P.; Scholes, C. P.; Solomon, E. I., Spectroscopic Studies of the Met182Thr Mutant of Nitrite Reductase: Role of the Axial Ligand in the Geometric and Electronic Structure of Blue and Green Copper Sites. Journal of the American Chemical Society 2003, 125, (48), 14784-14792. 40. Xie, X.; Hadt, R. G.; Pauleta, S. R.; González, P. J.; Un, S.; Moura, I.; Solomon, E. I., A Variable Temperature Spectroscopic Study on P. pantotrophus Pseudoazurin: Protein Constraints on the Blue Cu Site. J. Inorg. Biochem. 2009, 103, 1307-1313. 41. Blair, D. F.; Campbell, G. W.; Schoonover, J. R.; Chan, S. I.; Gray, H. B.; Malmstrom, B. G.; Pecht, I.; Swanson, B. I.; Woodruff, W. H.; Cho, W. K.; English, A. M.; Fry, H. A.; Lum, V.; Norton, K. A., Resonance Raman studies of blue copper proteins: effects of temperature and isotopic substitutions. Structural and thermodynamic implications. Journal of the American Chemical Society 1985, 107, (20), 5755-5766. 42. Jacobson, F.; Guo, H.; Olesen, K.; Okvist, M.; Neutze, R.; Sjolin, L., Structures of the oxidized and reduced forms of nitrite reductase from Rhodobacter sphaeroides 2.4.3 at high pH: changes in the interactions of the type 2 copper. Acta Crystallographica Section D 2005, 61, (9), 1190-1198. 43. Mitchell Guss, J.; Freeman, H. C., Structure of oxidized poplar plastocyanin at 1.6 Å resolution. Journal of Molecular Biology 1983, 169, (2), 521-563. 44. Ducros, V.; Brzozowski, A. M.; Wilson, K. S.; Brown, S. H.; Østergaard, P.; Schneider, P.; Yaver, D. S.; Pedersen, A. H.; Davies, G. J., Crystal structure of the type-2 Cu depleted laccase from Coprinus cinereus at 2.2 Å resolution. Nature Structural Biology 1998, 5, 310-316.

51 45. Solomon, E. I.; Sundaram, U. M.; Machonkin, T. E., Multicopper Oxidases and Oxygenases. Chem. Rev. 1996, 96, (7), 2563-2606. 46. Taylor, A. B.; Stoj, C. S.; Ziegler, L.; Kosman, D. J.; Hart, P. J., The copper-iron connection in biology: Structure of the metallo-oxidase Fet3p. Proc. Natl. Acad. Sci. USA 2005, 102, (43), 15459-15464. 47. Ellis, M. J.; Dodd, F. E.; Sawers, G.; Eady, R. R.; Hasnain, S. S., Atomic Resolution Structures of Native Copper Nitrite Reductase from Alcaligenes xylosoxidans and the Active Site Mutant Asp92Glu. Journal of Molecular Biology 2003, 328, (2), 429. 48. Inoue, T.; Gotowda, M.; Sugawara, H.; Kohzuma, T.; Yoshizaki, F.; Sugimura, Y.; Kai, Y., Structure Comparison between Oxidized and Reduced Plastocyanin from a Fern, Dryopteris crassirhizoma. Biochemistry 1999, 38, (42), 13853-13861.

52

Chapter 3

Variable Temperature Spectroscopic Study on Pseudoazurin: Effects of Protein Constraints on the Blue Cu Site*

* Reproduced in part with permission from J. Inorg. Biochem. 2009, 103.1307-1314. Copyright © 2009 Royal Society of Chemistry.

53 3.1 Introduction The blue or Type 1 (T1)1 copper proteins carry out rapid inter- and intra-molecular electron transfer (ET) in many biological systems.2-5 A classic and widely studied blue copper protein is plastocyanin in which the copper site has been thought to be constrained in a distorted tetrahedral coordination environment associated with four protein residue 6-9 derived ligands: 1SCys, 2NHis and 1SMet. The Cu-SCys bond is short at 2.07 Å, and the

Cu-SMet bond is long at 2.80 Å leading to an axially distorted tetrahedral structure with

the Cu close to the plane of the NHis1NHis2SCys. The idea of a protein constraint has been termed an entatic or rack state that has been thought to facilitate the ET function.10-12 The absorption spectrum of plastocyanin has an intense SCys π → Cu dx2-y2 charge transfer -1 -1 (CT) transition at ~600 nm (ε600 ~ 5000 M cm ) that is responsible for its blue color 13 and a weak higher energy SCys σ → Cu dx2-y2 CT transition. This CT intensity pattern reflects the π bonding nature of the thiolate to the Cu2+ in the ground state wavefunction. -4 -1 This site exhibits a very small hyperfine coupling A|| < 80 x 10 cm in the g|| region of its EPR spectrum, which is associated with the highly covalent bonding interaction 14, 15 between the Cu and SCys ligand. The covalency of the redox active molecular orbital (RAMO) has been quantified through Cu L- and S K-edge X-ray absorption spectroscopic (XAS) studies indicating that the RAMO has 42% Cu d and 38% S p 15-17 character. The high covalency activates the Cu-Scys bond for efficient ET. Recent studies on the T1 copper site in Rhodobacter sphaeroides nitrite reductase (NiR) have provided insight into the entatic/rack nature of the blue copper site.18 The low temperature crystal structure of NiR indicated that its T1 site with the same ligand set as plastocyanin has a tetragonally distorted structure with a shorter Cu-SMet bond at 2.48 Å 19 and a longer Cu-SCys bond at 2.21 Å. The low temperature absorption spectrum of NiR -1 shows a relatively weak SCys π → Cu dx2-y2 CT transition at ~570 nm (ε570 ~ 1600 M -1 -1 -1 cm ) and an intense SCys σ → Cu dx2-y2 CT transition (ε460 ~ 2700 M cm ) giving it a green color. This intense σ/weak π CT pattern reflects the dominant σ bonding nature of the ground state wavefunction for the tetragonally distorted green Cu site. However, the room temperature absorption spectrum of NiR shows increased intensity and a red-shift of the ~570 nm band, while the band at ~460 nm loses ~50% of its intensity. This temperature dependence has been determined to reflect a thermodynamic equilibrium

54 between a green Cu site and a blue Cu site, in which the Met ligand of the green site at low temperature is lost at higher temperature due to an increase in entropy. Thus, the

axial Met ligand in NiR is unconstrained by the protein in the sense that the Cu-Smet bond is weak and can be lost at elevated temperatures. In contrast, in plastocyanin, while the

Cu-Smet bond is still weak it remains bonded to the Cu at 2.8 Å over the same temperature range. In fact, in plastocyanin with a shorter Cys-Xn-His (n = 2) loop than that in NiR (n = 8, which allows flexibility and limits intraloop H-bonding), the protein imposes a constraint on the Met ligand keeping it bonded to the Cu at physiological temperature. This protein constraint or entatic/rack state stabilizes the oxidized state more than the reduced state and thus lowers the reduction potential.18 P. pantotrophus pseudoazurin has a T1 site that shows an interesting absorption spectrum in that both the ~600 nm and ~450 nm CT bands are fairly intense at room temperature.20 In this study, we use a combination of spectroscopic methods to investigate whether the presence of these two bands in the absorption spectrum reflects a thermodynamic equilibrium between two different species as in NiR, or whether they reflect a single species as in plastocyanin thus also having protein constraints on the T1 geometric structure. The latter is determined to be the case, but with the Met ligand at a shorter distance to the copper (from the crystal structure 21) leading to a ground state wavefunction with both π and σ character. This ground state is responsible for the absorption intensity in both CT transitions, reflecting a coupled distortion model in which

as the SMet to Cu bond gets shorter, the SCys to Cu bond gets longer and the site becomes more tetragonal, but not to the extent of the green Cu site. The coupled distortion in pseudoazurin leads to the mixed π/σ character in the ground state wavefunction and lowers the energy of the oxidized site en route to a reduction potential decrease compared to Plastocyanin.

3.2 Experimental Methods 3.2.1 Protein Purification P. pantotrophus pseudoazurin was heterologously expressed, isolated and purified as described previously.20 The buffer was exchanged to 10 mM Hepes, 10 mM NaCl, pH 7.5 and pseudoazurin was stored at -80 ºC. The concentration of the protein was determined

55 spectrophotometrically using the extinction coefficient at 590 nm, ε = 3000 M-1 cm-1 for fully oxidized pseudoazurin.20

3.2.2 Absorption and Magnetic Circular Dichroism Spectroscopy Low temperature absorption spectroscopy was performed using a Cary 500 double beam spectrophotometer modified with a Janis Research Super Vari-Temp liquid helium cryostat mounted in the optical path. 295 K absorption data were collected using a UV- Vis diode array spectrophotometer (Agilent 8453). Low temperature magnetic circular dichroism (MCD) experiments were conducted using two Jasco spectropolarimeters. Each is equipped with a modified sample compartment to accommodate focusing optics and an Oxford Instruments SM4000-7T superconducting magnet/cryostat. This arrangement allows data collection at temperatures from 1.6 to 290 K and fields up to 7 T. A Jasco J810 spectropolarimeter operating with an S-20 photomultiplier tube was used to access the visible and ultraviolet spectral region. A Jasco J200 spectropolarimeter operating with a cooled InSb detector was used for the near-infrared region. MCD samples were run in cells fitted with quartz disks and a 0.2 cm rubber gasket spacer. Simultaneous Gaussian fitting of the absorption, and MCD spectra was performed with the Peak-Fit program (Jandel).

3.2.3 Electron Paramagnetic Resonance X- and Q-band electron paramagnetic resonance (EPR) spectra were obtained using a Bruker EMX spectrometer. X-band Spectra were obtained at 77 K in a liquid nitrogen finger Dewar using an ER 041 XG microwave bridge, and an ER 4102ST cavity (Parameters for recording the X-band EPR: 9.39 GHz frequency and 20 G modulation amplitude). Q-band EPR spectra were obtained at 77 K using an ER 051 QR microwave bridge, an ER 5106QT resonator, and an Oxford continuous-flow CF935 cryostat (Parameters for recording the Q-band EPR: 33.83 GHz frequency, 20 G modulation amplitude). The protein concentrations for the EPR spectra were ~1 mM. EPR spectra were baseline-corrected and the X-band data were simulated using XSophe, g values were obtained from Q-band and high-frequency data. The high-frequency (95 GHz and 285 GHz) EPR spectrometers were described elsewhere.22 Samples of wild-type

56 pseudoazurin were 1.5 mM, in a buffer of 10 mM Hepes and 10 mM NaCl at pH 7.5. EPR spectra at 95 GHz and 285 GHz were recorded at 4.2 K.

3.2.4 Resonance Raman Resonance Raman (rR) spectra were obtained in a ~135 ° backscattering configuration with an incident power of 20 mW, using Rhodamine 6G and Stilbene dye lasers pumped by a Coherent Innova Sabre 25/7 Ar+ CW ion laser. Scattered light was dispersed through a triple monochromator (Spex 1877 CP, with 1200, 1800, and 2400 groove/mm gratings) and detected with a back-illuminated CCD camera (Princeton Instruments ST-135). Samples were contained in an NMR tube immersed in a liquid nitrogen finger dewar for low temperature measurements or in a spinning NMR tube for 295 K measurements. Background spectra were obtained using a buffer solution at either 77 K or 295 K for baseline subtraction in the same type of NMR tube. Frequencies are accurate to within 2 -1 cm . Raman energies were calibrated using Na2SO4 and citric acid.

The Cu-SCys bond strength was estimated from the effective frequency of the Cu-SCys vibration. This was obtained from the intensity-weighted average of the rR vibrations in -1 i 2 i i 330-500 cm region (<νCu-S> = ∑i [(I01 )(νi) ]/ ∑i(I01 )(νi), where I01 and νi are the intensity and frequency of a given peak i, respectively 23). Each peak was fit with a Gaussian equation to estimate the intensity and energy.

3.2.5 DFT Calculations DFT calculations were performed to understand the change in EPR spectra between plastocyanin and pseudoazurin. For the starting structure, a 54-atom model was used which included the α- and β-carbons of the protein residues fixed to match the T1 Cu site in plastocyanin (PDB ID: 1PLC) 8. Partial geometry optimization was performed, which reproduced the crystal structure (the deviations of the calculated bond distances and angles are within 0.04 Å and 5 °, repectively; note that the calculated angle between

NHis1CuNHis2 and SCysCuSMet is 81 ° (experiment: 82 °)). In order to impose the coupled distortion to model the Cu site in P. pantotrophus pseudoazurin, the angle between the

NHis1CuNHis2 and SCysCuSMet planes was distorted to 73 ° (this is the angle shown in the crystal structure PDB ID: 1ADW 21) and a new geometry optimization was performed to

57 obtain more accurate estimates of the metal-ligand bond distances (Figure 3.1). Spin- unrestricted DFT calculations were performed using Gaussian 03 24 with a hybrid functional B(38HF)LYP,25-27 tight SCF convergence criteria (10-8 au) and a mixed triple-

zeta/double zeta (TZVP on Cu and S, and 6-31G* on the other atoms) basis set. Wave function stability calculations were performed to confirm that the calculated wave functions corresponded to the ground state. EPR hyperfine coupling parameters were calculated on the above geometry optimized models using the program ORCA 28 with the hybrid functional B(38HF)LYP, the CoreProp basis set 29 for Cu, and the SVP basis set 30 for the other atoms. These are the TurboMole DZ bases developed by Ahlrichs and coworkers and obtained from the basis set library.31

α

β

S Cys SMet

NHis1 Cu

NHis2

Figure 3.1 Structures of T1 Cu sites (plastocyanin in black, pseudoazurin in grey)

3.3. Results and Analysis 3.3.1 Low Temperature Electronic Spectroscopy Figure 3.2 shows the 5 K electronic absorption and MCD spectra of P. pantotrophus pseudoazurin. These spectral features are perturbed relative to those of plastocyanin.13

58 We adopt the plastocyanin numbering scheme for spectral assignments.14 For pseudoazurin, there are two distinct regions in the low temperature absorption spectrum. From low to high energy, these are the ligand field (LF) transitions and CT transitions.

They can be identified by their |Co/Do| ratios (i.e. MCD to absorption intensity ratio, Table 3.1). These ratios were calculated using the equation 32

C0 kT Δε = ( ) max , D0 μ B B ε where T is the temperature, B is the external magnetic field strength, k is Boltzmann’s -1 -1 constant, μB is the Bohr magneton, ε is the absorption maximum in M cm , and Δε is -1 -1 the MCD intensity maximum also measured in M cm . k/μB can be taken as ~1.489 T K-1. For low symmetry sites, MCD intensity requires out of state spin orbit coupling (SOC). The SOC constant for Cu (~830 cm-1) is larger than those of the ligand atoms (N, ~70 cm-1 and S, 325 cm-1).33 Thus, the electronic transitions centered on Cu will show greater MCD intensity. Bands 5-8 have large |Co/Do|, thus they are assigned as LF transitions. The LF transitions of pseudoazurin show some differences when compared to plastocyanin. First, the ~5000 cm-1 band is blue-shifted ~100 cm-1 in pseudoazurin. This

is the Cu dz2 → dx2-y2 LF transition, and its increased energy indicates the axial LF is

stronger than that of plastocyanin. Also the increased absorption intensity of the Cu dxz+yz

→ dx2-y2 (band 6) and dxz-yz → dx2-y2 (band 7) transitions reflect increased CT character in

these LF transitions. The |Co/Do| ratios of bands 1 - 4 are smaller, thus, these bands are

assigned as CT transitions. Based on their |Co/Do| ratios and rR enhancements (vide infra),

we assign bands 4 and 3 as π SCys p → Cu dx2-y2 and σ SCys p → Cu dx2-y2 CT transitions, respectively. (Note that there is no clear evidence for the assignment of bands 1 and 2 in pseudoazurin; these were assigned as His and Met CT transitions in plastocyanin). The increased intensity in the 450 nm region (band 3) for pseudoazurin relative to plastocyanin can either indicate that pseudoazurin is a single T1 species with a mixed

π/σ bonding interaction between the Cu d and SCys p orbitals, or a frozen mixture of two species where one is a blue and the other is a green copper site.

59

Figure 3.2 Electronic absorption (5 K) and MCD (5 K, 7 T) spectra of P. pantotrophus pseudoazurin. Both spectra were taken on a glass sample of 1.5 mM pseudoazurin in a 10 mM phosphate buffer (pH 6.9)/glycerol (50:50 v/v).

60 Table 3.1 Low temperature (5 K) electronic spectroscopic parameters for pseudoazurin and plastocyanin 13, 34. ε Δε Band Assignment Wavenumber C /D (M-1cm-1) (M-1cm-1) o o Paz Pc Paz Pc Paz Pc Paz Pc

8 dz2 ~5100 ~5000 1.2 (+) (+) (+) 7 dxy 10400 10800 250 -5.3 -8.5 (-) -0.213 6 dxz+yz 12500 12800 1580 1425 11.4 20.9 0.054 0.092 5 dxz-yz 13500 13950 1970 500 -19.7 -41.4 -0.074 -0.518

4 SCys π 16850 16700 3360 5160 -13.6 -10.2 -0.030 -0.012 1 Smet 18700 500 1.6 0.023

3 SCys σ 22100 18700 2330 600 2.6 1.2 0.008 0.013

2 NHis π 24600 21390 1700 288 -2.1 -0.5 -0.009 -0.011 Paz and Pc denote pseudoazurin and plastocyanin, respectively.

3.3.2 Low Temperature Resonance Raman Spectroscopy Low temperature rR experiments were performed to determine whether the bands at ~600 and 450 nm arise from a single Cu site or a mixture of two species. Results of laser excitation at 593 and 450 nm at 77 K are shown in Figure 3.3. A rich vibrational spectrum is observed in the frequency region of ~320 – 500 cm-1 for both excitation -1 energies. A feature at 262 cm (not shown) has previously been assigned as the Cu-NHis stretching mode. In the rR spectrum obtained at 593 nm excitation, the large number of observed frequencies has been attributed to kinematic coupling of the Cu-SCys stretch 35 with the SCys-Cβ-Cα, Cβ-Cα-C(O), Cβ-Cα-C(N-H) bends. From the low temperature rR data, it is clear that the frequencies of the vibrational transitions observed at 593 nm have not changed upon excitation at 450 nm. There is, however, a redistribution in the observed rR intensity for the two CT regions. The facts that the frequencies of the observed rR enhanced normal modes are the same for both the 593 and 450 nm laser excitation and profiles of the enhanced vibrations follow both the σ and π SCys→ Cu CT absorption intensity, indicate that there is a single species present at low temperature. This is also supported by the multifrequency EPR results (vide infra). Thus, both the π and σ CT bands originate from a single species. The effective frequency of Cu-SCys vibration (intensity weighted average for the Raman spectrum obtained with 593 nm

61 -1 excitation), <νCu-Scys>, is 396 cm . Values for <νCu-Scys> have been previously determined to be 419, 403, 394 and 388 cm-1 for the type 1 Cu sites in fungal laccase, plastocyanin, cucumber basic protein (CBP) and NiR,5 respectively, indicating the strength of the Cu thiolate bond in pseudoazurin is similar to that of CBP, another perturbed blue copper protein with spectroscopic features similar to pseudoazurin (both π and σ CT transitions have significant intensity, and both proteins have a rhombic EPR signal, vide infra).

Figure 3.3 Resonance Raman spectra of P. pantotrophus pseudoazurin with laser excitation of 593 and 450 nm at 77 K. The sample consisted of ~1.5 mM pseudoazurin in 10 mM phosphate buffer (pH 6.9).

3.3.3 Electron Paramagnetic Resonance Experimental (solid line) and simulated (dotted line) X- and Q-band EPR data of pseudoazurin are shown in Figures 3.4A and B, and higher frequency (95 GHz and 285 GHz) spectra are shown in Figure 4C. The hyperfine couplings were best resolved at X- band frequency. By contrast, at 95 and 285 GHz the hyperfine features were lost, but the g-anisotropy was completely resolved. It was possible to "read-off" the g-values to good accuracy from the high-field spectra: gz = 2.200, gy = 2.065 and gx = 2.014 (±0.001). The Q-band spectra represented an intermediate case. The ensemble of the EPR data was indicative of a single species at low temperature.

62

Figure 3.4 X-Band (A) and Q-band (B) EPR data on pseudoazurin (solid line: experiment; dashed line: simulation; simulated results are listed in Table 2); high frequency EPR data 95 GHz and 285 GHz (C). Experimental condition: X-band data, 9.39 GHz, were taken at 77K; Q-band, 33.83 GHz, were taken at 50 K, EPR spectra at 95 GHz and 285 GHz were recorded at 4.2 K.

63 The X- and Q-band spectra were simulated using the same spin Hamiltonian parameters. These parameters are summarized in Table 3.2. A comparison of the g and A values to those of plastocyanin indicates that pseudoazurin has a larger rhombic splitting

(Δg⊥ = 0.051 in pseudoazurin and 0.017 in plastocyanin) and a larger perpendicular -4 -1 -4 -1 hyperfine coupling Ax (Ax=65x10 cm in pseudoazurin and 17x10 cm in plastocyanin). DFT calculations of EPR parameters reproduced the trends in going from plastocyanin to pseudoazurin (Table 3.3). The Löwdin population analyses indicate that the ground state wavefunctions of the Cu sites in plastocyanin and pseudoazurin are very similar.

However, there are 0.8% dz2 and 0.7% 4s characters mixed into the dx2-y2 ground state in pseudoazurin, in contrast to the essentially zero mixing in plastocyanin (Table 3.4). This 36 reflects the correlation developed by Gewirth et al that dz2 mixing is a major contribution to rhombic splitting. In going from plastocyanin to pseudoazurin, the Cu site symmetry changes from ~C3v toward ~C2v which allows dz2 mixing into the dx2-y2 ground state. Further, a small amount of Cu 4s character in the ground state wavefunction produces large positive Fermi contact contributions to the hyperfine coupling 37 which greatly reduces the negative Az and increases the small Ax for pseudoazurin relative to plastocyanin.

Table 3.2 EPR parameters of P. pantotrophus pseudoazurin and a comparison of those parameters to other blue copper sites. A A A Protein g g g Δg z y x Reference z y x ┴ (10-4 cm-1) (10-4 cm-1) (10-4 cm-1) plastocyanin 2.226 2.059 2.042 0.017 63 <17 <17 36 pseudoazurin 2.200 2.065 2.014 0.051 52 11 65 this work cucumber 2.207 2.07 2.02 0.05 55 10 60 38 basic protein nitrite 2.19 2.06 2.02 0.04 73 - - 34 reductase

Table 3.3 Calculated EPR parameters for Cu sites in plastocyanin and pseudoazurin.

Az Ay Ax |ΔA⊥| gz gy gx Δg⊥ -4 -1 -4 -1 -4 -1 -4 -1 (10 cm ) (10 cm ) (10 cm ) (10 cm ) plastocyanin 2.192 2.080 2.057 0.023 -92.3 15.6 16.4 0.8 pseudoazurin 2.190 2.088 2.044 0.044 -67.9 12.2 44.5 32.3

64 Table 3.4 Cu characters in β-LUMO by Löwdin reduced orbital population analysis (%).

s px py pz dz2 dxz dyz dx2-y2+dxy total

plastocyanin 0.2 0.5 0.2 0.0 0.1 0.1 0.1 47.6 48.8

pseudoazurin 0.7 0.1 0.7 0.7 0.8 0.1 0.1 45.1 48.3

Figure 3.5 Absorption spectra of pseudoazurin at 5 K (solid line) and 295 K (dotted line), Spectra were taken on a sample of 1.5 mM pseudoazurin in 10 mM phosphate buffer (pH 6.9)/glycerol (50:50 v/v).

3.3.4 Variable Temperature Absorption Spectra The absorption spectra of the Cu site in pseudoazurin at 5 and 295 K are shown in Figure 3.5. The spectra show a general broadening of peaks upon increasing the temperature from 5 to 295 K. The CT band maxima, however, do not change, and the ratio of total intensities does not change significantly ((εσ/επ)5K = 0.70 and (εσ/επ)295K =

0.62 where σ and π denote the SCys σ → Cu dx2-y2 and SCys π → Cu dx2-y2 CT transitions, respectively). Thus, pseudoazurin has a very different temperature dependent behavior

relative to NiR where the intensity ratio changes from εσ/επ = 1.7 at low temperature to 1.0 at room temperature.18 This indicates that the absorption bands at 450 and 600 nm for

65 pseudoazurin do not reflect a thermodynamic equilibrium of two species in which one

has a π and the other a σ ground state, but in fact reflect a single species with an Iσ/Iπ ratio in-between those of the blue copper site in plastocyanin (π ground state) and the green site in NiR (σ ground state).

Figure 3.6 Resonance Raman spectra of pseudoazurin with laser excitation at 607 nm at 77 and 295 K. Spectra were taken on a sample of 1.5 mM pseudoazurin in 10 mM phosphate buffer (pH 6.9).

3.3.5 Variable Temperature Resonance Raman Spectroscopy rR spectra obtained at 77 and 295 K upon excitation at 607 nm are presented in Figure 3.6. Upon warming the protein sample to 295 K, very little change is observed as only peak broadening occurs. The consistency of the frequencies and the intensity distribution pattern indicate that these rR data reflect a single blue copper species, but -1 -1 with a <νCu-Scys> of 396 cm , in between that of plastocyanin (<νCu-Scys> = 403 cm ) and -1 the green site ((<νCu-Scys> = 388 cm ). For NiR with excitation at 607 nm, the rR spectral

66 features at 364 and 420 cm-1 change upon going from liquid nitrogen temperature to room temperature reflecting the thermodynamic population of a second species with a different -1 Cu-SCys bond strength (<νCu-Scys> = 407 cm ). These temperature independent rR data for pseudoazurin are consistent with the absorption spectra, showing that only one species is present even at elevated temperature but with different spectral features from those of plastocyanin.

3.4. Discussion At low temperature, the Cu site in pseudoazurin exhibits essentially identical rR

vibrational frequencies when excited into the π SCys p → Cu dx2-y2 (band 4) and σ SCys p

→ Cu dx2-y2 (band 3) CT regions. This shows that bands 3 and 4 in the absorption spectrum at low temperature originate from the same species, i.e. a single species exists at low temperature, which is consistent with multifrequency EPR results. In going from low temperature to room temperature, no new features appear in the absorption or rR spectra indicating that only one species exists even at room temperature. This species has both π and σ CT intensity and its Cu-Smet bond strength is in-between that of the low temperature green site in NiR (dominant σ CT) and T1 center in plastocyanin (π CT). In the past, we defined the coupled distortion model shown in Figure 3.7 that describes geometric changes in blue copper proteins in going from plastocyanin to CBP to green NiR.38 Two extremes of the coupled distortion coordinate are represented by the blue copper site in fungal laccase that has no axial ligand, and the low temperature green -1 Cu component of NiR that has a strong axial Met ligand (<νCu-Scys> = 388 cm ) [18]. In the coupled distortion model, three dominant geometric changes occur in going from a 2+ blue to green site: 1) a decrease in the Cu -SMet bond length, which increases its donor 2+ 2+ interaction with the Cu , 2) this leads to an increase in the length of the Cu -SCys bond, and 3) a distortion from a more tetrahedral to a more tetragonal structure, which reflects a 2+ 2+ Jahn-Teller (eu) rotation of the SCys-Cu -SMet plane toward the NHis1-Cu -NHis2 plane. These geometric changes result in spectroscopic changes for the different proteins along the coupled distortion coordinate from blue to green. These spectroscopic changes are: 1) an increase in the ε450/ε600 ratio associated with an increase in the σ bonding interaction

67 between Cu d and SCys p orbitals in the ground state wavefunction; 2) a decreased

effective vibrational frequency, <νCu-Scys>; and 3) an increase in the LF transition energies. The spectral features of the Cu site in pseudoazurin indicate that it falls in the middle of the coupled distortion model between the blue Cu site in plastocyanin and the low temperature green site in NiR.

Figure 3.7 Continuum of coupled tetragonal distortions in a series of perturbed blue copper proteins (Pc: plastocyanin, CBP: cucumber basic protein, NiR: nitrite reductase; from reference 38)

It is important to emphasize that the classic blue copper site in plastocyanin does not show the thermodynamic equilibrium behavior observed in NiR.18 This has been

associated with a protein constraint such that the Cu-SMet bond is at 2.80 Å at low temperature and the Met ligand stays bonded to the Cu at room temperature. As presented in the introduction, protein constrained ligation has been referred to as an entatic or rack state for the classic blue copper sites.10-12 From reference 18, the protein constraint in plastocyanin overcomes the entropy increase that would be associated with loss of a weak axial Met ligand. The lack of a temperature dependence of the spectral features of pseudoazurin indicates that it is also a single species even at physiological temperature. Thus, as in plastocyanin, the met ligand in pseudoazurin is constrained by the protein. This is consistent with the fact that both plastocyanin and pseudoazurin have the same Cys-Xn-His (n = 2) loop. The spectral features of pseudoazurin are in-between those of

68 plastocyanin and NiR at low temperature. This indicates that the protein constrains the axial Met ligand at a shorter distance to the Cu than in plastocyanin but longer than in the green component of NiR. In summary, for blue copper and other metalloproteins that have weak axial ligation, the temperature dependence of their spectral features provides a direct probe as to whether the protein allows the ligand to dissociate from the metal. Controlling ligand loss can tune reduction potentials and regulate reactivity under physiological conditions.

3.5 Acknowledgments This research was supported by NSF Grant CHE 0446304 (E.I.S.). X. X. is grateful for William S. Johnson fellowship.

3. 6 References 1. Malkin, R.; Malmstrom, B. G., The State and Function of Copper in Biological Systems. Adv. Enzymol. Relat. Areas Mol. Biol. 1970, 33, 177-244. 2. Adman, E. T., Copper protein structures. Advances in protein chemistry 1991, 42, 145-197. 3. Gray, H. B., Long-range electron-transfer in blue copper proteins. Chemical Society Reviews 1986, 15, (1), 17-30. 4. Holm, R. H.; Kennepohl, P.; Solomon, E. I., Structural And Functional Aspects of Metal Sites in Biology. Chem. Rev. 1996, 96, (Thematic "Bioinorganic Enzymology" Issue), 2239-2314. 5. Solomon, E. I.; Szilagyi, R. K.; DeBeerGeorge, S.; Basumallick, L., Electronic Structures of Metal Sites in Proteins and Models: Contributions to Function in Blue Copper Proteins. Chem. Rev. 2004, 104, (2), 419-458. 6. Colman, P. M.; Freeman, H. C.; Guss, J. M.; Murata, M.; Norris, V. A.; Ramshaw, J. A. M.; Venkatappa, M. P., X-ray crystal structure analysis of plastocyanin at 2.7 Å resolution. Nature 1978, 272, (23 March), 319-324. 7. Guss, J. M.; Bartunik, H. D.; Freeman, H. C., Accuracy and precision in protein stucture analysis: restrained least-squares refinement of the structure of poplar plastocyanin at 1.33 ang. resolution. Acta Crystallogr., Sect B: Struct. Sci. 1992, B48, (6), 790-811.

69 8. Guss, J. M.; Freeman, H. C., Structure of Oxidised Poplar Plastocyanin at 1.6 Å Resolution. J. Mol. Biol. 1983, 169, 521-563. 9. Solomon, E. I., Spectroscopic Methods in Bioinorganic Chemistry: Blue to Green to Red Copper Sites. Inorg. Chem. 2006, 45, (20), 8012-8025. 10. Williams, R. J. P., Energised (entatic) states of groups and of secondary structures in proteins and metalloproteins. FEBS 1995, 234, (2), 363-381. 11. Malmstrom, B. G., Rack-induced bonding in blue-copper proteins. Eur. J. Biochem. 1994, 223, (3), 711-718. 12. Gray, H. B.; Malmstrom, B. G.; Williams, R. J. P., Copper coordination in blue proteins. J. Biol. Inorg. Chem. 2000, 5, (5), 551-559. 13. Gewirth, A. A.; Solomon, E. I., Electronic structure of plastocyanin: excited state spectral features. J. Am. Chem. Soc 1988, 110, (12), 3811-1938. 14. Penfield, K. W.; Gay, R. R.; Himmelwright, R. S.; Eickman, N. C.; Norris, V. A.; Freeman, H. C.; Solomon, E. I., Spectroscopic Studies on Plastocyanin Single Crystals: A Detailed Electronic Structure Determination of the Blue Copper Active Site. J. Am. Chem. Soc. 1981, 103, 4382. 15. Shadle, S. E.; Penner-Hahn, J. E.; Schugar, H. J.; Hedman, B.; Hodgson, K. O.; Solomon, E. I., X-ray absorption spectroscopic studies of the blue copper site: metal and ligand K-edge studies to probe the origin of the EPR hyperfine splitting in plastocyanin. J. Am. Chem. Soc. 1993, 115, (2), 767-776. 16. Scott, R. A.; Hahn, J. E.; Doniach, S.; Freeman, H. C.; Hodgson, K. O., Polarized x-ray absorption spectra of oriented plastocyanin single crystals. Investigation of methionine-copper coordination. J. Am. Chem. Soc 1982, 104, (20), 5364-5369. 17. Penfield, K. W.; Gewirth, A. A.; Solomon, E. I., Electronic Structure and Bonding of the Blue Copper Site in Plastocyanin. J. Am. Chem. Soc. 1985, 107, 4519. 18. Ghosh, S.; Xie, X.; Dey, A.; Sun, Y.; Scholes, C. P.; Solomon, E. I., Thermodynamic Equilibrium between Blue and Green Copper Sites and the Role of the Protein in Controlling Function. Proc. Natl. Acad. Sci. USA 2009, 106, 4969-4974. 19. Jacobson, F.; Guo, H.; Olesen, K.; Okvist, M.; Neutze, R.; Sjolin, L., Structures of the oxidized and reduced forms of nitrite reductase from Rhodobacter sphaeroides

70 2.4.3 at high pH: changes in the interactions of the type 2 copper. Acta Crystallographica Section D 2005, 61, (9), 1190-1198. 20. Pauleta, S. R.; Guerlesquin, F.; Goodhew, C. F.; Devreese, B.; Beeumen, J. V.; Pereira, A. S.; Moura, I.; Pettigrew, G. W., Paracoccus pantotrophus Pseudoazurin Is an Electron Donor to Cytochrome c Peroxidase. Biochemistry 2004, 43, (35), 11214-11225. 21. Williams, P.; Fülöp, V.; Leung, Y.; Chan, C.; Moir, J.; Howlett, G.; Ferguson, S.; Radford, S.; Hajdu, J., Pseudospecific docking surfaces on electron transfer proteins as illustrated by pseudoazurin, cytochrome c550 and cytochrome cd1 nitrite reductase. Nat. Struct. Biol. 1995, 2, (11), 975-982. 22. Un, S.; Dorlet, P.; Rutherford, A. W., Appl. Magn. Reson. 2001, 21, 341-361. 23. Blair, D. F.; Campbell, G. W.; Schoonover, J. R.; Chan, S. I.; Gray, H. B.; Malmstrom, B. G.; Pecht, I.; Swanson, B. I.; Woodruff, W. H.; Cho, W. K.; English, A. M.; Fry, H. A.; Lum, V.; Norton, K. A., Resonance Raman studies of blue copper proteins: effects of temperature and isotopic substitutions. Structural and thermodynamic implications. Journal of the American Chemical Society 1985, 107, (20), 5755-5766. 24. Frisch, M. J. T., G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.; Cheeseman, J. R.; Montgomery, J., J. A.; Vreven, T.; Kudin, K. N.; Burant, J. C.; Millam, J. M.; Lyengar, S. S.; Tomasi, J.; Barone, V.; Mennucci, B.; Cossi, M.; Scalmani, G.; Rega, N.; Petersson, G. A.; Nakatsuji, H.; Hada, M.; Ehara, M.; Toyota, K.; Fukuda, R.; Hasegawa, J.; Ishida, M.; Nakajima, T.; Honda, Y.; Kitao, O.; Nakai, H.; Klene, M.; Li, X.; Knox, J. E.; Hratchian, H. P.; Cross, J. B.; Adamo, C.; Jaramillo, J.; Gomperts, R.; Stratmann, R. E.; Yazyev, O.; Austin, A. J.; Cammi, R.; Pomelli, C.; Ochterski, J. W.; Ayala, P. Y.; Morokuma, K.; Voth, G. A.; Salvador, P.; Dannenberg, J. J.; Zakrzewski, V. G.; Dapprich, S.; Daniels, A. D.; Strain, M. C.; Farkas, O.; Malick, D. K.; Rabuck, A. D.; Raghavachari, K.; Foresman, J. B.; Ortiz, J. V.; Cui, Q.; Baboul, A. G.; Clifford, S.; Cioslowski, J.; Stefanov, B. B.; Liu, G.; Liashenko, A.; Piskorz, P.; Komaromi, I.; Martin, R. L.; Fox, D. J.; Keith, T.; Al-Laham, M. A.; Peng, C. Y.; Nanayakkara, A.; Challacombe, M.; Gill, P. M. W.; Johnson, B.; Chen, W.; Wong, M. W.;

71 Gonzalez, C.; Pople, J. A., Gaussian 03, Revision C.01., Gaussian 03, Revision C.02. In Gaussian, Inc.: Wallingford, CT, 2004. 25. Perdew, J. P., Density-functional approximation for the correlation energy of the inhomogeneous electron gas. Phys. Rev. B 1986, 33, 8822. 26. Becke, A. D., Density-functional exchange-energy approximation with correct asymptotic behavior. Phys. Rev. A 1988, 38, 3098. 27. Becke, A. D., Density-functional thermochemistry. III. The role of exact exchange. J. Chem. Phys. 1993, 98, 5648. 28. Neese, F. ORCA, version 2.6; Universität Bonn: Bonn, Germany. The program is available free of charge at http://www.thch.uni-bonn.de/tc/orca. In 2006. 29. Neese, F., Inorg. Chim. Acta 2002, 337, 181-192. 30. Schäfer, A.; Horn, H.; Ahlrichs, R., J. Chem. Phys. 1992, 97, 2571-2577. 31. ftp.chemie.uni-karlsruhe.de/pub/basen. 32. Piepho, S. B.; Schatz, P. N., Group Theory in Spectroscopy with Applications to Magnetic Circular Dichroism. John Wiley & Sons: New York, Chichester, Brisbane, Toronto, Singapore: 1983. 33. Solomon, E. I.; Lever, A. B. P., Inorganic Electronic Structure and Spectroscopy. John Wiley & Sons: Hoboken: New Jersey, 2006; Vol. I, p 110. 34. LaCroix, L. B.; Shadle, S. E.; Wang, Y.; Averill, B. A.; Hedman, B.; Hodgson, K. O.; Solomon, E. I., Electronic structure of the perturbed blue copper site in nitrite reductase: Spectroscopic properties, bonding, and implications for the entatic/rack state. Journal of the American Chemical Society 1996, 118, (33), 7755-7768. 35. Qiu, D.; Kilpatrick, L.; Kitajima, N.; Spiro, T. G., Modeling blue copper protein resonance Raman spectra with thiolate-CuII complexes of a sterically hindered tris(pyrazolyl)borate. J. Am. Chem. Soc. 1994, 116, (6), 2585–2590. 36. Gewirth, A. A.; Cohen, S. L.; Schugar, H. J.; Solomon, E. I., Spectroscopic and theoretical studies of the unusual EPR parameters of distorted tetrahedral cupric sites: correlations to x-ray spectral features of core levels. Inorg. Chem. 1987, 26, (7), 1133-1146. 37. Xie, X.; Gorelsky, S. I.; Sarangi, R.; Garner, D. K.; Hwang, H. J.; Hodgson, K. O.; Hedman, B.; Lu, Y.; Solomon, E. I., Perturbations to the Geometric and

72 Electronic Structure of the CuA Site: Factors that Influence Delocalization and Their Contributions to Electron Transfer. J. Am. Chem. Soc. 2008, 130, (15), 5194-5205. 38. LaCroix, L. B.; Randall, D. W.; Nersissian, A. M.; Hoitink, C. W. G.; Canters, G. W.; Valentine, J. S.; Solomon, E. I., Spectroscopic and Geometric Variations in Perturbed Blue Copper Centers: Electronic Structures of Stellacyanin and Cucumber Basic Protein. J. Am. Chem. Soc 1998, 120, (37), 9621-9631.

73

Chapter 4

The Two State Issue in the Mixed-Valence Binuclear CuA Center in

Cytochrome c Oxidase and N2O Reductase*

* Reproduced in part with permission from J. Am. Chem. Soc. 2006, 128, 16452-16453. Copyright © 2006 American Chemical Society.

74 4.1 Introduction 1 The binuclear CuA site found both in cytochrome c oxidase (CcO) and nitrous oxide 2 reductase (N2OR) functions in long range electron transfer (ET). Studies of CuA sites have been facilitated by the construction of inorganic models,3 and the perturbation of 4 5, 6 CuA sites by mutation and by inserting CuA into proteins containing cupredoxin folds.

The CuA site is defined as a Cu2(SCys)2 cluster (Figure 4.1A) with a short Cu-Cu distance 7, 8 (EXAFS dCu-Cu = 2.43-2.44 Å) . The Cu2S2 atoms are nearly-planar and each copper ion 1, 2, 9 is further coordinated equatorially by NHis and axially by either SMet or carbonyl O.

Figure 4.1 (Top) DFT-Optimized structures of CuA without the protein environment: (A) * σu state and (B) πu state (only Cu2S2 cluster is shown for simplicity, internuclear distances are given in Å, CuM denotes the Cu atom with the axial Met ligand); (bottom) + β-spin LUMO (contour value = 0.03) of the [Cu2(SCH3)2(imz)2] complex (Figure 4.S1a) * in the (A) σu and (B) πu states.

In its oxidized state, CuA is a delocalized (Class III) mixed-valence species Cu1.5+Cu1.5+.10-13 The ground state (GS) wave function, quantified by X-ray absorption spectroscopy (XAS), is highly covalent,13 with the spin density delocalized over the

Cu2S2 cluster. This greatly contributes to the redox properties by lowering the reorganization energy and providing superexchange hole coupling for long-range ET into 3 and out of the CuA center. From studies of a model complex (dCu-Cu = 2.92Å) relevant to

75 * the CuA protein site, there are two types of GS wave function, σ u and πu (Figure 4.1B) * which interchange depending on the Cu-Cu distance (σ u for CuA at a short dCu-Cu and πu 12 for the model complex at a long dCu-Cu). DFT calculations of the CuA site reveal that the GS potential energy surface of the mixed-valence, oxidized cluster is flat with two 14 * minima, which correspond to the σu (dCu-Cu = 2.49 Å, Figure 4.1A) and πu (dCu-Cu = 3.06 Å, Figure 4.1B) electronic states. 15-17 As first discussed by Neese and co-workers, πu is the lowest excited-state for the

CuA site in the protein but its energy has been controversial. EPR data on CuA proteins

show a low g|| value of 2.19 (Table 4.S1) which derives from spin-orbital coupling * between the σu GS and the πu excited-state, which requires an energy gap of 3,000-4,500 cm-1.12, 15 On the other hand, from paramagnetic NMR studies, it has been observed that the lowest-energy excited-state is thermally-accessible and the energy gap between the -1 18 GS and the thermally-accessible state is ~350 cm . This has been found in the CuA site from P. verstus and P. denitrificans, which have the same spectral features in UV-Vis absorption and equivalent EPR g-values (Table 4.S1). This study addresses this apparent * discrepancy and evaluates the possible role of the two electronic states (σ u and πu) in ET of CuA.

4.2 Experimental Details 4.2.1 Sample Preparation

The samples of CuA soluble domain from T. thermophilus (Tt) were prepared according to Slutter et al.19

4.2.2 Absorption and MCD Measurements Room temperature absorption data were collected on HP Agilent spectrometer. MCD measurements were performed on Jasco J730 (UV/vis, S1 and S20 PMT detection) and

J200 (NIR, liquid N2-cooled InSb detection) spectropolarimeters equipped with an Oxford Instruments SM4000-7 Tesla (T) superconducting magnet/cryostat. The MCD spectrum was corrected for the natural CD and Zero-field baseline effects caused by strain in the glasses by averaging the positive and negative filed data at 5K (i.e. [1T-(-

)1T]/2. Samples of ~ 0.5 mM Tt CuA-containing protein in trisHCl buffer (pH8) were

76 diluted with glycerol-d3 (~50% v:v) and injected into sample cells comprised of two quartz disks separated by a Viton O-ring spacer. The concentration of the protein is determined by EPR spin quantification.

4.2.3 Computational Details Density Functional Theory (DFT) calculations have been performed using the Gaussian 03 program.20 Optimized molecular geometries were calculated using the B3LYP 21 exchange-correlation functional.

Three CuA models for DFT calculations were used: 1) 30-atom model (no Cu axial

ligands, Figure 4.S1a), 2) 51-atom model (with CH3SCH3 and CH3CONHCH3 as axial ligands, Figure 4.S1b), 3) 217-atom model (Figure 4.S1c) based on the high-resolution structure of an engineered CuA azurin (PDB: 1CC3, water molecules were removed). 22 The mixed triple-zeta/double zeta (TZVP on Cu and S and 6-31G* on the other atoms) basis set and tight SCF convergence criteria were used for calculations. Spin- unrestricted DFT was employed to model the open-shell species. Wave function stability calculations were performed to confirm that the calculated wave functions corresponded to the ground state. Geometry optimizations using the 30- and 51-atom models were performed without any constrains. For the 217-atom model, the partial geometry optimization was performed. The positions of the protein backbone carbon, nitrogen and oxygen atoms (Fig. S2) were frozen to those derived from the 1.65 Å-resolution X-ray structure of an 23 engineered CuA azurin (PDB ID: 1CC3). Frequency calculations for the 30- and 51-atom models were performed to ensure that the stationary points were true energy minima. Time-dependent DFT (TD-DFT)24 was used to calculate the energies and intensities of the 40 lowest-energy electronic transitions. The calculated absorption energies and intensities were transformed with the SWizard program25 into simulated spectra, using Gaussian functions with half-widths of 2670 cm-1 based on the average experimental half-bandwidths in the spectrum of the Tt

CuA site. Atomic charges and spin densities were calculated using Mulliken26 and natural27 population analyses (MPA and NPA respectively) as implemented in Gaussian 03. The

77 compositions of molecular orbitals in terms of atomic contributions were derived using MPA as implemented in the AOMix program.28, 29 The Mayer bond orders (BOs)30 were obtained to analyze covalent contributions to chemical bonding using the AOMix-L program.31 The analysis of the MO compositions in terms of fragment MOs (FOs) was performed using AOMix-CDA.32 33, 34 Bond energy (ΔEtotal) decomposition analysis in terms of electrostatic interaction

(ΔVelstat), exchange repulsion (ΔEPauli) and orbital interactions (ΔEorbit) was performed using the ADF2006.01 package35 at the BP86/TZ2P level of theory. The two fragments II + I for the decomposition analysis were defined as (imz)Cu (SCH3) and (imz)Cu (SCH3)

from the 30-atom model of CuA. Inner-sphere reorganization energies for electron transfer were calculated using the equation 4.1, λ = E − E + E − E i ( g=ox g=red )reduced ( g=red g=ox )oxidized (4.1) in which E are the electronic energies, the g=ox and g=red subscripts refer to the +1.5 +1.5 +1 +1 geometry of the CuA site in the fully-relaxed Cu Cu and Cu Cu redox states, respectively, and the oxidized and reduced subscripts refer to the Cu+1.5Cu+1.5 and Cu+1Cu+1 redox states, respectively. The electron transfer (ET) rates were calculated using the Marcus equation:36

π ⎡ (ΔG o + λ) 2 ⎤ 2 (4.2) k ET = 2 ⋅ (H DA ) ⋅ exp⎢− ⎥ λk T 4λk T h B ⎣ B ⎦ o The driving forces (-ΔG ) for σu* CuA → heme a and heme c → σu* CuA ET in 37 38 cytochrome c oxidase are 0.031 eV and 0.030 eV , respectively. For the πu CuA →

heme a and heme c → πu CuA ET, the driving forces were calculated taking into account -1 that the πu GS of CuA is 350 cm higher in energy relative to the σu* GS of CuA. 13 The heme c → σu* CuA ET pathway involves the Cys200 ligand of CuA. The σu*

CuA → heme a ET can involve two competitive pathways: one through the His204 of 13, 39, 40 13, 41, 42 CuA and a second through the Cys200 . The greater Cu-SCys covalency

relative to the Cu-NHis covalency allows for enhancement of the ET rate for the Cys200 pathway and can make it competitive with the shorter His204 pathway. In the case of the

78 ET pathway that goes though the Cys200 ligand, the superexchange coupling is 2 42 proportional to the α S, (SCys character in the β-spin LUMO). Since for both σu* and πu

GSs, SCys characters are fairly similar (~40% for σu* GS and ~45% for πu GS, Table S4.2), the electronic coupling in the ET pathway is not affected significantly based on the

change in the Cu-Sthiolate covalency in going from the σu* GS to the πu GS. In the case of the ET pathway that goes though the His204 ligand, the superexchange coupling is 2 proportional to the α N, (NHis character in the β-spin LUMO). For the σu* GS, NHis character is approximately 2 times higher (Table 4.S2) relative to NHis character for the πu

GS. Thus, the (kET σu*) / (kET πu) ratios (Figure 4.S5) were calculated assuming that HDA

σu* = HDA πu for the Cys ET pathway and HDA σu* = 2 HDA πu for the His ET pathway.

4.3 Results and Analysis

UV-Vis absorption of CuA from Thermus thermophilus (Tt) is given in Figure 4.2A, which is similar to other CuA sites, and its EPR g-values are also equivalent to the other 15 CuA sites (Table S4.1). The g|| of 2.19 can be used through Equation 4.3 to estimate the * 2 energy (Δ) of the πu excited state relative to the σ u GS of CuA. α is the total Cu character * 13 in the σu GS, which is 44% from Cu L-edge XAS. The Cu character in the πu excited 2 * state, β , is 31% (Table 4.S2) from the DFT calculations below. These give a σu -πu -1 energy gap of ~5000 cm (Table 4.S1). Paramagnetic NMR behavior of Tt CuA is similar to the other CuA proteins (Table 4.S1) indicating a presence of the thermally-accessible -1 state within the energy range estimated for the other CuA proteins (~450 cm ).

8ζ Cuα 2 β 2 g = g + 3d (4.3) || e Δ

The MCD spectrum (Figure 4.2B) can be used to resolve individual electronic

transitions (Figure 4.2A, red lines) in the absorption spectrum of CuA. TD-DFT (Figure

4.2C) on CuA reproduces the experimental absorption spectrum well and gives the parity- -1 * forbidden, lowest-energy excited state πu at 3200 cm above the σu GS. This result is consistent with the EPR-derived energy gap. However, the thermally-accessible excited

state is not revealed in TD-DFT calculations in the CuA geometry.

79

Figure 4.2 (A) Room-temperature absorption, (B) low-temperature (5K) MCD spectra of Tt CuA; (C) TD-DFT calculated absorption spectrum of the CuA model (total absorption –black and individual components –red).

We can use TD-DFT calculations supported by the spectroscopic data on CuA to map the GS and excited state potential energy surfaces (PES). These calculations (Figure 4.3A) * reveal the electronic coupling between the σu and πu states, which is a function of the Cu-

Cu distance. The crossing point between the two states is at dCu-Cu = 2.66 Å and coincides with the inflection point of the Cu-Cu bond order30 curve (Figure 4.3B). These PES calculations show that both NMR and EPR results are consistent with the

electronic/geometric structure of CuA. The anti-Curie behavior observed in paramagnetic * NMR studies of CuA results from the thermal equilibrium between the σu and πu GSs. They are at very close energies in their respective equilibrium geometries (Figure 4.3A).

80 * Alternatively, the EPR g-value analysis involves the σu GS in the geometry with a short -1 dCu-Cu in which the πu is a Frank-Condon excited state calculated to be at 3200 cm . * Similar energies but different wave functions of the σu and πu GS of CuA suggest * different chemical bonding in the two states. Going from the σu GS with the short Cu-Cu

internuclear distance to πu, the Cu-Cu bond order decreases from 0.32 to 0.12 (Figure 4.3B), indicating a significant decrease of the direct Cu-Cu bonding interaction. However, the sum of the Cu-S and Cu-Cu bond orders (Figure 4.S3) has decreased only slightly in the two states, thus the loss of the Cu-Cu interaction is compensated by 1) the gain in

Cu-S interactions and 2) a reduction of the exchange repulsion in the πu GS relative to the * * σu GS (Table 4.S3). Thus the electronic energies of the σu GS and the πu GS remain very close.

Figure 4.3 (A) The ground state and the first excited state potential energy surfaces (black

lines refer to the CuA cluster in the vacuum and green – the cluster in the protein

environment) and (B) Mayer bond order between the two Cu atoms in the GS of CuA as a function of the Cu-Cu distance.

81 In the πu GS of CuA, the Cu-SCys covalency becomes localized resulting in two short

and two long Cu-SCys bonds (Figure 4.1B). The decreased Cu-Cu interaction (Figure 4.3B)

results in a spin distribution in the CuA cluster that is easier to localize under the low- symmetry ligand-field of metal axial ligand (Figure 4.S3) and electrostatic perturbations in the protein environment (Table 4.S2). Localization of the Cu-S covalency and valence- trapping lead to larger inner- and outer-sphere reorganization energies (λi and λo) of the * πu GS relative to the σu GS (Table 4.S4). The calculated λi of the πu GS is 1.6-times * higher than λi of the σu GS. * Since the σu and πu GSs of CuA have very similar energies, the ET driving force * changes by only ~43 mV in going from the σu GS to the πu GS. Thus, the factors that can

influence the ET rate between CuA and its redox partners are the donor-acceptor electronic coupling and the reorganization energies. Using Marcus theory it is possible to

evaluate relative efficiency, kσ/kπ, of the two states for ET. The calculated ET rate ratios

in the heme c→CuA and CuA→heme a pathways (Figure 4.4) are ~15 for both of these

processes, which suggests that the πu GS is much less efficient for long-range ET than the * σ u state.

20

15 π 10 / k σ k

5

CuA to heme a

heme c to CuA 0 -0.4 -0.2 0.0 0.2 0.4 ΔG (eV)

Figure 4.4 Calculated ET rate ratios kσ / kπ in the heme c→CuA and CuA→heme a pathways as a function of the ET driving force (ΔG).

82 DFT calculations of the CuA site without the protein environment (Figure 4.S1a,b) * result in the πu GS being a slightly lower-energy state relative to the σu GS (Figure 4.3A).

Expanding the QM calculations from the 51-atom CuA site model (Figure 4.S1b) to 217 * atoms (Figure 4.S1c) or all protein atoms (using the QM/MM method) results in the σu

GS of CuA as the lowest-energy state with dCu-Cu of 2.44Å (Figure 4.3A, green line) in 8 agreement with the EXAFS data (dCu-Cu of 2.43-2.44Å) . This relative stability derives from the fact that the non-covalent interactions, including H-bonds between the protein * backbone amides and the SCys atoms of the CuA site, stabilize the σu GS (which has less

spin density and more negative charge on the SCys atoms, Table 4.S2) relative to the πu * GS. Thus, the protein environment plays a role in maintaining CuA in the σu as a lowest- energy state with the lowest reorganization energy for efficient intra- and inter-molecular ET with a low driving force.

4.4 Acknowledgment

This research was supported by NSF Grant CHE 0446304 (E.I.S.) and NIH Grants DK31450 (E.I.S.) and GM35342 (J.A.F.). S.I.G. thanks NSERC (Ottawa) for a postdoctoral fellowship.

4.5 Supporting Information

Table 4.S1 The g-values and energy gap between the πu and σu* states from EPR, the

energy of the thermally accessible state probed by NMR for CuA proteins from various resources. X-ray EPR NMR Structure Resolution Δ i Δ f Δ Protein Source ref g ref ref ref (Å) || cm-1 cm-1 cm-1

CuA in 43 10 g this h 44 a P. stutzeri 1.6 2.18 4500 5100 585 N2OR study

CuA from P. e 1 16 this 18 b 2.8 2.19 4800 350 CcO denitrificans study T. CuA from ba3 45 46 this j 19 b thermophilu 1.6 2.19 4900 450 oxidase study s P. this Cu azurinc 1.65 23 2.17 47 5400 n/a - A aeruginosa study Cu from A P. versutus 18 CcO b,d n/a n/a - n/a 350

83 a) Native CuA protein. b) Soluble CuA domain. c) Engineered CuA protein. d) This protein has same spectral features in UV-Vis absorption and EPR to those of P. denitrificans. e) The crystal structure is on the whole protein of cytochrome c oxidase from P. denitrificans; other data in this row are of the soluble CuA domain. f) The experimental 2 2 2 α1 value is used for the Δ calculation (α1 =0.44 for CuA azurin, Cu L-edge), and β1 is adjusted to 0.31 based on the DFT calculation. The simple model for g-value analysis 17 neglects the influence of sulfur spin-orbit coupling which is ~-0.01 to g||. Therefore, Δ -1 -1 2 2 could be ~400 cm smaller. g) Δ is 3500-4500 cm , assuming β1 = α1 i.e. the Cu

contribution to the β-spin LUMO in the πu excited state is identical to that in the ground 2 15 state, α1 = 0.31 or 0.37 which is based on fitting EPR hyperfine-coupling constant A||. h) NMR data are on mutant N2OR which is defective in the biosynthesis of the catalytic

center, only one of the four Cys Hβ NMR signals are observed. CuA variant of amicyanin

and CuA center of B. subtilis has no hyperfine-shifted signals in the 100-400 ppm region;44 therefore not listed in this table. i) the original estimate from the EPR g-value. (j) 19 NMR data of Cys153 Hβ1(d) taken from Ref were fitted with the

2 −ΔE / kT 2 Aβi (b sin φ + c ) + e (b cos φ + c ) equation ( ) = 1 βi 1 2 βi 2 . The XAS-derived S h 1+ e−ΔE / kT

character (ρS =0.23) of CuA azurin and calculated S character (ρS =0.32) of the 217-atom

CuA model are used for σu* and πu respectively to give the energy difference of the two 2 states; b=BρS, B=100 MHz, c1=0, c2=-1/2b2φ cos2θ0 = 9.6 MHz, where φ=0.79 is the o effective amplitude of the tortional oscillation, θ0 =90 .

84 Table 4.S2 Bond distances, NPA-derived atomic charges and atomic spin densities, and the total Cu, Sthiolate and Nimidazole contributions to β-spin LUMO of the CuA models. Parameters a 30-atom model 51-atom 217-atom model model σu* GS d(Cu-Cu) c 2.43 2.49 2.44 c d(CuM-SI) 2.32 2.36 2.33 c d(CuM-SII) 2.32 2.35 2.33 c d(CuO-SI) 2.32 2.34 2.40 c d(CuO-SII) 2.32 2.33 2.35 q(CuM) 0.824 0.818 0.914 q(CuO) 0.824 0.933 0.938 q(SI) -0.416 -0.482 -0.515 q(SII) -0.419 -0.442 -0.476 SD(CuM) 0.237 0.201 0.266 SD(CuO) 0.237 0.281 0.245 SD(SI) 0.218 0.193 0.168 SD(SII) 0.218 0.251 0.247 Cu% β-LUMO 50.3 50.5 52.0 Sthiolate% β-LUMO 40.4 40.7 38.6 b b b Nimidazole% β-LUMO 5.8 (2.9 ) 2.8 (2.2 ) 4.5 (1.7 ) d,e Cu%, Sthiolate% β-HOMO 28.0, 65.6 30.8, 63.1 25.8, 52.1 πu GS d(Cu-Cu) 2.94 3.06 2.9 d(CuM-SI) 2.23 2.31 2.29 d(CuM-SII) 2.43 2.38 2.44 d(CuO-SI) 2.43 2.37 2.53 d(CuO-SII) 2.23 2.28 2.31 q(CuM) 0.799 0.797 0.919 q(CuO) 0.799 0.929 0.932 q(SI) -0.649 -0.432 -0.501 q(SII) -0.649 -0.418 -0.470 SD(CuM) 0.189 0.134 0.251 SD(CuO) 0.189 0.270 0.205 SD(SI) 0.284 0.282 0.200 SD(SII) 0.284 0.286 0.281 Cu% β-LUMO 43.6 44.9 48.5 Sthiolate% β-LUMO 49.6 49.5 42.6 b b b Nimidazole% β-LUMO 2.8 (1.4 ) 1.2 (1.0 ) 3.5 (1.2 ) a) Distances (d) are given in Angstroms; atomic charges (q) and spin densities (SD) are given in atomic units; the total Cu, Sthiolate and Nimidazole contributions to β-spin LUMO refer to orbital contributions from a pair of the corresponding atoms (Cu, S, N). b) Nimidazole contribution from the

N atom that corresponds to NHis that is involved in the CuA→heme a ET pathway through the

85 His204 ligand.13 c) These distances can be compared to the EXAFS data which show d(Cu-Cu) 8 and d(Cu-S) of 2.43-2.44Å and 2.29-2.30Å, respectively. d) Cu and Sthiolate contributions in the β- spin HOMO. e) In the 217-atom model, β-spin HOMO-4 corresponds to the β-HOMO in the 30- and 51-atom models.

-1 Table 4.S3 Bond energy decomposition analysis (kcal mol ) for the 30-atom CuA model.

electronic ground state ΔVelstat ΔEPauli ΔEorbit ΔEtotal

σu* (dCu-Cu=2.43 Å) -181.1 203.4 -116.0 -93.6

πu (dCu-Cu=2.94 Å) -121.3 151.3 -102.0 -72.0

Table 4.S4 Reorganization energies (λ) of the ET pathway with the CuA center in the

σu* and πu ground states.

CuA λtotal Heme a Heme c Cu heme c Inner Sphere Outer Sphere A →heme a →CuA 12 0.25 13 37 13, 48, 49 λ (eV) σ * a 0.15 0.4 1.2 0.4 0.8 u (0.40) 50 0.48 13, 51, 52 37 13, 48, 49 λ (eV) π a 0.32 0.4 1.2 0.6 1.0 u (0.66) a) The DFT-calculated values are shown in parenthesis.

Figure 4.S1 Structures of the CuA models for DFT calculations: (A) 30-atom model (no

Cu axial ligands), (B) 51-atom model (with CH3-S-CH3 and CH3-CO-NH-CH3 as axial ligands), (C) 217-atom model based on the high-resolution structure of an engineered CuA 23 azurin (PDB ID: 1CC3). The structures shown correspond to the geometries of the σu* ground state.

86

Figure 4.S2 Structure (top and side view) of the 217-atom CuA model. Gray-colored atoms indicate the frozen atoms in the geometry optimization; the coordinates of the atoms shown in blue were optimized.

Figure 4.S3 Mayer bond orders for the four Cu-S bonds (black lines and circles) of the

CuA cluster (51-atom model), the sum of the Cu-S bond orders (blue open circles) and the sum of the Cu-S and Cu-Cu bond orders (blue squares). CuM is the Cu atom of the Cu2S2 cluster with the thioether axial ligand.

87

0.30 A CuO

0.25 SII

SI 0.20

Atomic Spin Density 0.15

CuM

0.10 0.35

B SII 0.30

Cu 0.25 M

0.20

CuO o

Atomic Spin Density 0.15 SI

0.10 2.2 2.4 2.6 2.8 3.0 3.2 o Cu-Cu Distance (A) Figure 4.S4 NPA-derived atomic spin densities for the Cu and Sthiolate atoms of the CuA cluster (A: the 51-atom model , B: the 217-atom model ). CuM designate the Cu atom with the thioether axial ligand.

88 4.6 References 1. Iwata, S.; Ostermeier, C.; Ludig, B.; Michel, H., Structure at 2.8 Å resolution of cytochrome c oxidase from Paracoccus denitrificans. Nature 1995, 376, (6542), 660-669. 2. Brown, K.; Djinovic-Carugo, K.; Haltia, T.; Cabrito, I.; Saraste, M.; Moura, J. J. G.; Moura, I.; Tegoni, M.; Cambillau, C., Revisiting the catalytic CuZ cluster of

nitrous oxide (N2O) reductase. J. Biol. Chem. 2000, 275, 41133–41136. 3. Houser, R. P.; Young, V. G.; Tolman, W. B., A Thiolate-Bridged, Fully Delocalized Mixed-Valence Dicopper(I,II) Complex That Models the CuA Biological Electron-Transfer Site. J. Am. Chem. Soc 1996, 118, (8), 2101-2102. 4. Zumft, W. G.; Viebrock-Sambale, A.; Braun, C., Eur. J. Biochem. 1990, 192, 591-599. 5. Hay, M.; Richards, J.; Lu, Y., Construction and characterization of an azurin analog for the purple copper site in cytochrome c oxidase. Proc Natl. Acad. Sci. U.S.A. 1996, 93, (1), 461-464. 6. Dennison, C.; Vijgenboom, E.; de Vries, S.; van der Oost, J.; Canters, G.,

Introduction of a CuA site into the blue copper protein amicyanin from Thiobacillus versutus. FEBS Lett. 1995, 365, 92-94. 7. Blackburn, N. J.; Barr, M. E.; Woodruff, W. H.; van der Oost, J.; de Vries, S., Metal-Metal Bonding in Biology: EXAFS Evidence for a 2.5 Å Copper-Copper

Bond in the CuA Center of Cytochrome Oxidase. Biochemistry 1994, 33, 10401- 10407. 8. Blackburn, N. J.; de Vries, S.; Barr, M. E.; Houser, R. P.; Tolman, W. B.; Sanders, D.; Fee, J. A., X-ray Absorption Studies on the Mixed-Valence and Fully

Reduced Forms of the Soluble CuA Domains of Cytochrome c Oxidase. J. Am. Chem. Soc. 1997, 119, 6135-6143. 9. Haltia, T.; Brown, K.; Tegoni, M.; Cambillau, C.; Saraste, M.; Mattila, K.; Djinovic-Carugo, K., Crystal Structure of Nitrous Oxide Reductase from Paracoccus denitrificans at 1.6 Å Resolution. Biochem. J. 2003, 369, 77-88. 10. Kroneck, P.; Antholine, W.; Riester, J.; Zumft, W., The Cupric Site in Nitous- Oxide Reductase contains a mixed-valence [Cu(II),Cu(I)] Binuclear center - A

89 Multifrequency Electron-Paramagnetic Resonance Investigation. FEBS Lett. 1988, 242, (1), 70-74. 11. Kroneck, P. M. H.; William A. Antholine; Riester, J.; Zumft, W. G., The nature of

the cupric site in nitrous oxide reductase and of CuA in cytochrome c oxidase. FEBS Lett. 1989, 249, 212-213. 12. Gamelin, D. R.; Randall, D. W.; Hay, M. T.; Houser, R. P.; Mulder, T. C.; Canters, G. W.; de Vries, S.; Tolman, W. B.; Lu, Y.; Solomon, E. I., Spectroscopy

of Mixed-Valence CuA -Type Centers: Ligand-Field Control of Ground-State Properties Related to Electron Transfer. J. Am. Chem. Soc. 1998, 120, 5246-5263. 13. DeBeer George, S.; Metz, M.; Szilagyi, R. K.; Wang, H.; Cramer, S. P.; Lu, Y.; Tolman, W. B.; Hedman, B.; Hodgson, K. O.; Solomon, E. I., A Quantitative

Description of the Ground-State Wave Function of CuA by X-ray Absorption Spectroscopy:Comparison to Plastocyanin and Relevance to Electron Transfer. J. Am. Chem. Soc 2001, 123, (24), 5757-5767. 14. Olsson, M. H. M.; Ryde, U., Geometry, Reduction Potential, and Reorganization

Energy of teh Binuclear CuA Site, Studied by Density Functional Theory. J. Am. Chem. Soc. 2001, 123, 7866-7876. 15. Neese, F.; Zumft, W. G.; Antholine, W. E.; Kroneck, P. M. H., The Purple 63 65 Mixed-Valence CuA Center in Nitrous Oxide Reductase: EPR of the Cu , Cu , and Both Cu65 and [15N]Histidine-Enriched Enzyme and a Molecular Orbital Interpretation. J. Am. Chem. Soc. 1996, 118, 8692-8699. 16. Farrar, J. A.; Neese, F.; Lappalainen, P.; Kroneck, P. M. H.; Saraste, M.; Zumft,

W. G.; Thomson, A. J., The Electronic Structure of CuA: A Novel Mixed-Valence Dinuclear Copper Electron-Transfer Center. J. Am. Chem. Soc. 1996, 118, 11501- 11514. 17. Neese, F.; Kappl, R.; Hüttermann, J.; Zumft, W. G.; Kroneck, P. M. H., Probing the ground state of the purple mixed valence CuA center in nitrous oxide reductase: a CW ENDOR (X-band) study of the 65Cu, 15N-histidine labeled enzyme and interpretation of hyperfine couplings by molecular orbital calculations. J. Biol. Inorg. Chem. 1998, 3, 53-67.

90 18. Salgado, J.; Warmerdam, G.; Bubacco, L.; Canters, G. W., Understanding the

Electronic Properties of the CuA Site from the Soluble Domain of Cytochrome c Oxidase through Paramagnetic 1H NMR. Biochemistry 1998, 37, (20), 7378-7389. 19. Bertini, I.; Bren, K. L.; Clemente, A.; Fee, J. A.; Gray, H. B.; Luchinat, C.;

Malmstrom, B. G.; Richards, J. H.; Sanders, D.; Slutter, C. E., The CuA Center of

a Soluble Domain from Thermus Cytochrome ba3. An NMR Investigation of the paramagnetic Protein. J. Am. Chem. Soc 1996, 118, (46), 11658-11659. 20. Frisch, M. J. T., G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.; Cheeseman, J. R.; Montgomery, J., J. A.; Vreven, T.; Kudin, K. N.; Burant, J. C.; Millam, J. M.; Lyengar, S. S.; Tomasi, J.; Barone, V.; Mennucci, B.; Cossi, M.; Scalmani, G.; Rega, N.; Petersson, G. A.; Nakatsuji, H.; Hada, M.; Ehara, M.; Toyota, K.; Fukuda, R.; Hasegawa, J.; Ishida, M.; Nakajima, T.; Honda, Y.; Kitao, O.; Nakai, H.; Klene, M.; Li, X.; Knox, J. E.; Hratchian, H. P.; Cross, J. B.; Adamo, C.; Jaramillo, J.; Gomperts, R.; Stratmann, R. E.; Yazyev, O.; Austin, A. J.; Cammi, R.; Pomelli, C.; Ochterski, J. W.; Ayala, P. Y.; Morokuma, K.; Voth, G. A.; Salvador, P.; Dannenberg, J. J.; Zakrzewski, V. G.; Dapprich, S.; Daniels, A. D.; Strain, M. C.; Farkas, O.; Malick, D. K.; Rabuck, A. D.; Raghavachari, K.; Foresman, J. B.; Ortiz, J. V.; Cui, Q.; Baboul, A. G.; Clifford, S.; Cioslowski, J.; Stefanov, B. B.; Liu, G.; Liashenko, A.; Piskorz, P.; Komaromi, I.; Martin, R. L.; Fox, D. J.; Keith, T.; Al-Laham, M. A.; Peng, C. Y.; Nanayakkara, A.; Challacombe, M.; Gill, P. M. W.; Johnson, B.; Chen, W.; Wong, M. W.; Gonzalez, C.; Pople, J. A., Gaussian 03, Revision C.01., Gaussian 03, Revision C.02. In Gaussian, Inc.: Wallingford, CT, 2004. 21. Becke, A. D., Density-functional thermochemistry. III. The role of exact exchange. J. Chem. Phys. 1993, 98, 5648. 22. Schafer, A.; Huber, C.; Ahlrichs, R., Fully optimized contracted Gaussian basis sets of triple zeta valence quality for atoms Li to Kr. J. Chem. Phys. 1994, 100, 5289-5835. 23. Robinson, H.; Ang, M. C.; Gao, Y.-G.; Hay, M. T.; Lu, Y.; Wang, A. H.-J.,

Structural Basis of Electron Transfer Modulation in the Purple CuA Center. Biochemistry 1999, 38, (18), 5677-5683.

91 24. Stratmann, R. E.; Scuseria, G. E.; Frisch, M. J., An efficient implementation of time-dependent density-functional theory for the calculation of excitation energies of large molecules. J. Chem. Phys. 1998, 109, 8218-8224. 25. Gorelsky, S. I., SWizard, Version 4, http://www.sg-chem.net. In 2000. 26. Mulliken, R. S., Electronic Population Analysis on LCAOMO Molecular Wave Functions. I. J. Chem. Phys. 1955, 23, 1833. 27. Reed, A. E.; Curtiss, L. A.; Weinhold, F., Intermolecular interactions from a natural bond orbital, donor-acceptor viewpoint. Chem. Rev. 1988, 88, 899. 28. Gorelsky, S. I., AOMix: Program for Molecular Orbital Analysis, http://www.sg- chem.net. In. 29. Gorelsky, S. I.; Lever, A. B. P., J. Organomet. Chem. 2001, 635, 187. 30. Mayer, I., Chem. Phys. Lett. 1983, 97, 270. 31. Gorelsky, S. I.; Basumallick, L.; Vura-Weis, J.; Sarangi, R.; Hodgson, K. O.; Hedman, B.; Fujisawa, K.; Solomon, E. I., Spectroscopic and DFT Investigation of[M{HB(3,5-iPr2pz)3}(SC6F5)] (M = Mn, Fe, Co, Ni, Cu, and Zn) Model Complexes: Periodic Trends in Metal−Thiolate Bonding. Inorg. Chem. 2005, 44, (14), 4947-4960. 32. Gorelsky, S. I.; Ghosh, S.; Solomon, E. I., Mechanism of N2O reduction by the

μ4-S tetranuclear CuZ cluster of nitrous oxide reductase. J. Am. Chem. Soc. 2006, 128, (1), 278-290. 33. Bickelhaupt, F. M.; Baerends, E. J., Rev. Comput. Chem. 2000, 15, 1-86. 34. Ziegler, T.; Rauk, A., Calculation of Bonding Energies by Hartree-Fock Slater Method.1. Transition-State Method. Theoret. Chim. Acta 1977, 46, (1), 1-10. 35. ADF2006.01, SCM, Theoretical Chemistry, Vrije Universiteit, Amsterdam, The Netherlands, http://www.scm.com. In. 36. Marcus, R. A.; Sutin, N., Electron transfers in chemistry and biology. Biochim. Biophys. Acta 1985, 811, 265-322. 37. Farver, O.; Einarsdottir, O.; Pecht, I., Electron transfer rates and equilibrium within cytochrome c oxidase. Eur. J. Biochem. 2000, 267, 950-954.

92 38. Pan, L. P.; Hibdon, S.; Liu, R. Q.; Durham, B.; Millett, F., Intracomplex electron transfer between ruthenium-cytochrome c derivatives and cytochrome c oxidase. Biochemistry 1993, 32, 8492-8498. 39. Regan, J. J.; Ramirez, B. E.; Winkler, J. R.; Gray, H. B.; Malmstrom, B. G., Bioenerg. Biomembr. 1998, 30, (35). 40. Ramirez, B. E.; Malmstrom, B. G.; Winkler, J. R.; Gray, H. B., Proc. Natl. Acad. Sci. U.S.A. 1995, 92, 11949. 41. Medvedev, D. M.; Daizadeh, I.; Stuchebrukhov, A. A., J. Am. Chem. Soc. 2000, 122, 6571. 42. Williams, K.; Gamelin, D.; LaCroix, L. B.; Houser, R. P.; Tolman, W. B.; Mulder, T. C.; de Vries, S.; Hedman, B.; Hodgson, K. O.; Solomon, E. I., Influence of Copper-Sulfur Covalency and Copper-Copper Bonding on Valence Delocalization and Electron Transfer in the CuA Site of Cytochrome c Oxidase. J. Am. Chem. Soc. 1997, 119, (3), 613-614. 43. Haltia, T.; Brown, K.; Tegoni, M.; Cambillau, C.; Saraste, M.; Djinovic-Carugo,

K., The crystal stucture of nitrous oxide reductase (N2OR) from Paracoccus denitrificans at 1.6 Å resolution. J. Inorg. Biochem. 2001, 86, 246-246. 44. Holz, R. C.; Alvarez, M. L.; Zumft, W. G.; Dooley, D. M., 1H NMR Studies on the CuA Center of Nitrous Oxide Reductase from Pseudomonas stutzeri. Biochemistry 1999, 38, (34), 11164-11171. 45. Williams, P. A.; Blackburn, N. J.; Sanders, D.; Bellamy, H.; Stura, E. A.; Fee, J.

A.; McRee, D. E., The CuA domain of Thermus thermophilus ba3-type cytochrome c oxidase at 1.6 Å resolution. Nat. Struct. Biol. 1999, 6, 509-516. 46. Karpefors, M.; Slutter, C.; Fee, J.; Aasa, R.; Kallebring, B.; Larsson, S.; Vanngard,

T., Electron paramagnetic resonance studies of the soluble CuA protein from the

cytochrome ba3 of Thermus thermophilus. Biophys. J. 1996, 71, 2823-2829. 47. Hay, M. T.; Ang, M. C.; Gamelin, D. R.; Solomon, E. I.; Antholine, W. E.; Ralle, M.; Blackburn, N. J.; Massey, P. D.; Wang, X.; Kwon, A. H.; Lu, Y.,

Spectroscopic Characterization of an Engineered Purple CuA Center in Azurin. Inorg. Chem. 1998, 37, 191-198.

93 48. Wright, J. L.; Wang, K.; Geren, L.; Saunders, A. J.; Pielak, G. J.; Durham, B.; Millett, F., ACS AdV. Chem. Ser. 1998, 254, 99-110. 49. Wuttke, D. S.; Bjerrum, M. J.; Winkler, J. R.; Gray, H. B., Science 1992, 256, 1007-1009. 50. Fraga, E.; Webb, M. A.; Loppnow, G. R., J. Phys. Chem. 1996, 100, 3278-3287. 51. Di Bilio, A. J.; Hill, M. G.; Bonander, N.; Karlsson, B. G.; Villahermosa, R. M.; Malmstro¨m, B. G.; Winkler, J. R.; Gray, H. B. J., J. Am. Chem. Soc. 1997, 119, 9921-9922. 52. Skov, L. K.; Pascher, T.; Winkler, J. R.; Gray, H. B., J. Am. Chem. Soc. 1998, 120, 1102-1103.

94

Chapter 5

Perturbations to the Geometric and Electronic Structure of the CuA Site: Factors that Influence Delocalization and their Contributions to Electron Transfer *

* Reproduced in part with permission from J. Am. Chem. Soc. 2008, 128, 16452-16453. Copyright © 2008 American Chemical Society.

95 5.1 Introduction 1-6 The CuA site, found in cytochrome c oxidases (CcOs) , nitrous oxide reductases 7, 8 9, 10 (N2ORs), and a nitric oxide reductase (NOR), is responsible for rapid intra- and intermolecular electron transfer (ET) in biological systems. CcOs catalyze the terminal

step of enzymatic aerobic respiration by coupling the four electron reduction of O2 to

H2O to the generation of a proton electrochemical gradient that is the energetic driving

force for the synthesis of ATP. NOR reduces two NO to N2O + H2O and N2OR is the terminal denitrification enzyme reducing N2O to N2 + H2O. In all enzymes, the CuA site serves as an ET conduit between an electron source and the catalytic active site.

Previous studies of CuA sites in CcOs, and N2ORs have defined their geometric and electronic properties through a combination of X-ray crystallography, extended X-ray absorption fine structure (EXAFS), resonance Raman (rR), magnetic circular dichroism (MCD), nuclear magnetic resonance (NMR), electron paramagnetic resonance (EPR) and X-ray absorption spectroscopies (XAS).11-21 The expression and isolation of engineered 22-27 and mutant CuA sites , combined with the successful synthesis of CuA model 28 complexes have facilitated strongly the understanding of the electronic structure of CuA.

The binuclear CuA site contains two copper centers bridged by two thiolates, with each

copper further coordinated equatorially by one Nhis and axially by either a Smet or a

carbonyl OGlu. The Cu2S2 atoms form a diamond core, which is nearly planar, with a Cu- 13, 14, 19, 29 Cu distance of 2.40-2.44Å . In its oxidized state, CuA is a completely delocalized (class III30) mixed valence Cu1.5+-Cu1.5+ species.18, 19, 21, 31, 32 The ground state wave function quantified by XAS is highly covalent with 44% Cu character and 46% S character.33 This is very similar to the blue copper ground state wavefunction, however,

in CuA the covalency is distributed over the Cu2S2 dimer. ENDOR studies of CuA have 34 shown a total of ~4% spin density over two NHis ligands in CcO and ~4% NHis 35 character on at least one NHis in N2OR. The class III mixed valence ψ→ψ* transition of 16, 19 CuA was identified by rR. This transition is associated with the valence delocalization

of CuA providing a measure of the electronic coupling 2HAB between two Cu′s.

Correlation of the ψ→ψ* transition and Cu/S edge spectroscopies in CuA and a model have shown that both direct Cu-Cu and superexchange through the S bonds contribute to

HAB. Consistent with those studies, density functional theory (DFT) calculations of the

96 Cu1 Cu 2 Cu site reveal a completely delocalized σ * type ground state ( d 2 2 − d 2 2 ) (Chart A u x −y x −y 1A) that is the redox active molecular orbital (RAMO). Note that there is also a low-lying

Cu1 Cu 2 πu exited state ( d xy + d xy ) (Chart 5.1B), responsible for various spectroscopic features (vide infra).

Chart 5.1 A C

H120 CuO CuS B

36 Engineered CuA azurin exhibits a very interesting pH effect. It displays a mixed valence EPR spectrum with a seven-line hyperfine pattern associated with the delocalization of the S = 1/2 over two Cu centers (ICu = 3/2). Upon lowering the pH to 4.0, the EPR hyperfine coupling changed to a four-line pattern, indicating apparent spin localization. This pH dependent transition is reversible with a pKa value of ~5. It has been suggested that this change in EPR signal can be related to a perturbation of the H120 ligand, which has an imidazole nitrogen equatorially coordinated to the Cu with an axial carbonyl OGlu ligand (Chart 5.1C). The H120 ligand is located near the protein surface and could be protonated at low pH. Interestingly, the H120A mutant has very similar spectral features to the low-pH form, and the pH-dependent transition is eliminated. Multifrequency EPR data have indicated that the electron spin in the H120A mutant is localized based on a four-line Cu hyperfine pattern.37 However, ENDOR data indicate that the site is delocalized based on superhyperfine couplings of cysteine Cβ protons and hisidine nitrogens.37 In this study, we address this apparent discrepancy through a series of electronic spectroscopic studies of the high-pH and low-pH forms of CuA azurin and its H120A mutant combined with DFT calculations. Our studies define the geometric and electronic

97 structure of the low-pH form of CuA azurin and its H120A mutant, elucidate the extent of electron delocalization and the contribution of the low-lying πu excited state to the ground state spectral features, and provide further insight into the factors contributing to delocalization (and the associated facile ET of CuA) in the low symmetry protein environment. The possible contribution of the pH-dependent transition to proton coupled electron transfer (PCET) in CcO is discussed.

5.2 Experiments 5.2.1 Sample Preparation 26 The CuA construct engineered into Pseudomonas aeruginosa azurin and its mutant H120A38 were prepared as described previously.

5.2.2. Electron Paramagnetic Resonance EPR spectra were obtained using a Bruker EMX spectrometer, an ER 041 XG microwave bridge, and an ER 4102ST cavity. All X-band samples were run at 77 K in a liquid nitrogen finger Dewar. A Cu standard (1.0 mM CuSO4⋅5H2O with 1mM HCl and

2M Na2ClO4) was used for spin quantitation of the EPR spectra. All Q-band spectra were obtained at 77 K using an ER 051 QR microwave bridge, an ER 5106QT resonator, and an Oxford continuous-flow CF935 cryostat. EPR spectra were baseline-corrected and simulated using XSophe (Bruker). For each protein, X- and Q-band spectra were simultaneously fit in order to constrain the simulation parameters (g values were obtained from Q-band data and hyperfine coupling from X-band data).

5.2.3. Electronic Absorption and MCD Spectroscopies Low temperature absorption data were collected on a Cary-17 spectrometer using a Janis Research Super Vari-Temp cryogenic dewar mounted in the optical path. MCD measurements were performed on Jasco J810 (UV/vis, S1 and S20 PMT detection) and

J200 (NIR, liquid N2-cooled InSb detection) spectropolarimeters equipped with an Oxford Instruments SM4000-7 Tesla (T) superconducting magnet/cryostats. The MCD spectra were corrected for the natural CD and zero-field baseline effects caused by the strain in glass by averaging the positive and negative field data at 5K ([7T-(-)7T]/2).

98 Samples of ~1 mM CuA-containing protein in NH4OAc buffer (pH = 4 and pH = 7 for w.t. CuA azurin, pH = 5 for the H120A mutant) were diluted with glycerol-d3 (~50% v:v) and injected into sample cells comprised of two quartz disks separated by a Viton O-ring spacer. The concentrations of the protein were determined by EPR spin quantitation.

5.2.4. Resonance Raman Spectroscopy rR spectra were obtained using continuous wave (CW) excitation from Kr+ (Coherent I90K) and Ar+ (Coherent I18UV and CR18) ion laser sources. A Ti-sapphire (Coherent 890) laser was used for the NIR region. Incident power in the range of ~25 mW was used in a ~135° backscattering arrangement. Spectra were measured using a SPEX model 1877 CP triple monochromator with 1200, 1800, and 2400 grooves/mm holographic gratings, equipped with a Princeton Instruments back-illuminated CCD -1 detector. The resolution was 2 cm at all excitation energies. Solution samples were frozen in NMR tubes placed in a liquid nitrogen finger Dewar. Excitation Profile intensities of Raman scattering peaks were determined relative to the 230 cm-1 ice peak in protein samples.

5.2.5 XAS Data Acquisition The X-ray absorption spectra were measured at the Stanford Synchrotron Radiation Laboratory on the focused 16-pole 2.0 T wiggler beam line 9-3 under standard ring conditions of 3 GeV and 80-100 mA. A Si(220) double crystal monochromator was used for energy selection. A Rh-coated harmonic rejection mirror and a cylindrical Rh-coated bent focusing mirror were used on beam line 9-3 to reject components of higher harmonics. During data collection, the samples were maintained at a constant temperature of 10 K using an Oxford Instruments CF 1208 liquid helium cryostat. Fluorescence mode was used to measure data to k = 13.4 Å-1 employing a Canberra Ge 30-element solid array detector. Internal energy calibration was accomplished by simultaneous measurement of the absorption of a Cu-foil placed between two ionization chambers situated after the sample. The first inflection point of the foil spectrum was set at 8980.3 eV. Data presented here are thirteen-scan (w.t., pH6), eight-scan (WT, pH4) and nine- scan (H120A, pH6) averaged spectra, which were processed by fitting a second-order

99 polynomial to the pre-edge region and subtracting this from the entire spectrum as background. A three-region spline of orders 2, 3 and 3 was used to model the smoothly decaying post-edge region. The data were normalized by subtracting the cubic spline and assigning the edge jump to 1.0 at 9000 eV using the PySpline program.39 The crystal structure of CuA azurin (PDB code 1CC3) was used to calculate the theoretical EXAFS phase and amplitude parameters using FEFF (version 7.0)40, 41 and fit to the data using EXAFSPAK.42 The structural parameters allowed to vary during the fitting process were the bond length (R) and the bond variance σ2, which is related to the Debye-Waller factor resulting from thermal motion, and static disorder. The non-structural parameter E0 (the energy at which k =0) was also allowed to vary, but restricted to a common value for every component in a given fit. Coordination numbers were systematically varied in the course of the fit but were fixed within a given fit. For the Cu K-edge experiments, the protein concentrations (in Cu content) were ~2.5 mM for w.t. and ~5 mM for H120A mutant. The samples were prepared in NH4OAc buffer. 30% glycerol was added to the protein solution as a glassing agent. Samples were loaded into Lucite XAS cell holders and inserted into 2 mm Lucite XAS cells with 63.5

μm Mylar windows. All samples were rapidly frozen by immersion in liquid N2 and kept at or below this temperature.

5.2.6. DFT Calculations

A 96-atom model (47 heavy atoms) was used to match the high-pH CuA site. A partial geometry optimization was performed, where positions of the protein backbone carbon, nitrogen and oxygen atoms were frozen to those in the 1.65 Å-resolution X-ray structure of the engineered CuA azurin (PDB ID: 1CC3). In order to model the pH effect on the

CuA site, additional calculations were performed with ligand H120 protonated (97 atoms). DFT calculations were performed using the program Gaussian 03.43 Spin-unrestricted DFT was employed to model the open shell species. Optimized molecular geometries were calculated using the hybrid B3LYP exchange-correlation functional44-46 with tight -8 SCF convergence criteria (10 au) and the mixed triple-zeta/double zeta (TZVP on Cu and S and 6-31G* on the other atoms) basis sets. Wave function stability calculations were performed to confirm the calculated wave functions corresponded to the ground

100 state. Time-dependent DFT (TD-DFT) was used to calculate the energies and intensities of the 30 lowest-energy, spin-allowed electronic transitions. The calculated absorption energies and intensities were transformed with the SWizard program47 into simulated spectra, using Gaussian with half-widths of 2600 cm-1. EPR hyperfine coupling parameters were calculated on the above geometry optimized models using the program ORCA.48 This employed the hybrid functional B3LYP with the CoreProp basis set49 for Cu, and the SVP basis set50 for other atoms. These bases are the TurboMole DZ bases developed by Ahlrichs and coworkers and obtained from the basis set library.51

5.3 Results and Analysis 5.3.1 Spectroscopy 5.3.1.1 Electron Paramagnetic Resonance Figure 5.1 shows the X- and Q- band EPR spectra of the high pH (at pH = 7) and the low-pH (at pH = 4) forms of w.t. CuA-azurin protein, and the H120A mutant. Table 5.1 gives the g|| values and hyperfine coupling parameters obtained by XSophe simulation of the X-band data. The g values were extracted from the Q-band data and then correlated to X-band spectra.

At high pH, CuA azurin has a g|| value of 2.173 and a g⊥ of 2.022 (Figure 5.1 black,

Table 5.1). At low pH, CuA azurin has g|| = 2.222 and g⊥ = 2.010 (Figure 5.1 red, Table 5.1), which are similar to the g values for the H120A mutant (Figure 5.1 green, Table 5.1).

The gx-gy anisotropy was < 0.01 in all cases. The g|| > g⊥> 2.0023 pattern indicates that

all these forms of Cu have a d 2 2 ground state, where the electron can be delocalized A x − y over two coppers.

At high pH, CuA azurin displays a 7-line hyperfine coupling pattern in g|| region. The 7 hyperfine lines are associated with complete delocalization of the unpaired electron (S = 1/2) over two equivalent coppers, each with I=3/2. Consistent with this, XSophe simulations (Table 5.1) show that this high-pH form has equivalent hyperfine coupling of Cu1 Cu2 -4 -1 Cu1 Cu2 -4 -1 the two coppers: A|| = A|| =|53| x 10 cm and A⊥ = A⊥ =|26| x 10 cm .

101

A

2800 2900 3000 3100 3200 3300 3400 3500 3600 Field (Gauss)

B

2.28 2.24 2.20 2.16 2.12 2.08 2.04 2.00 1.96 g Figure 5.1 (A) X-band EPR spectra. Microwave frequency, 9.46 GHz; temperature 77K; (B) Q-band EPR spectra. Microwave frequency, 33.86GHz; temperature 77K; high pH form (⎯), low pH form (⎯), H120A mutant (⎯), and XSophe simulations (⎯).

102 Upon going to low pH, the EPR spectrum goes to a 4-line hyperfine coupling pattern in g|| region, consistent with a large inequivalence in the hyperfine coupling of each Cu nucleus to the S=1/2, and the unpaired electron appears to be localized. In Table 5.1, the Cu1 -4 -1 Cu2 -4 -1 Cu1 and Cu2 XSophe simulations give A|| =|43| x 10 cm , A|| =|12| x 10 cm , and A⊥ in -4 -4 -1 the range between |21|x10 and |26|x10 cm . The EPR spectra of the H120A mutant are equivalent to those of the low pH form (Figure 5.1 green, Table 5.1). The similarities in these spectra indicate that the ground state wavefunctions of the copper centers in the two proteins are very similar. The pH dependence is eliminated in the H120A mutant; thus for the low pH case, the His ligand (H120) is proposed to be protonated, and this changes the EPR signal such that the hyperfine coupling of the electron spin to one Cu is greatly decreased.

Table 5.1 EPR parameters of the high and low pH forms of CuA azurin, and its H120A mutant. Cu1 Cu2 Cu1 Cu2 Cu1 Cu2 A,x A x A y , A y A,z Az gx gy gz -4 -1 -4 -1 -4 -1 (10 cm ) (10 cm ) (10 cm ) High pH Form 2.022 2.022 2.173 25, 27 25, 26 53, 53 Low pH Form* 2.010 2.010 2.222 26, 21 21, 24 43, 12 H120A 2.010 2.010 2.225 24, 19 19, 24 46, 7

* Note that the low pH form still has ~10% of the high pH form as the pka is ~5. This is included in EPR spectral simulation.

5.3.1.2. Electronic Absorption and MCD Spectroscopy. A comparison of the low temperature Abs and MCD Spectra of the high-pH and low- pH forms, and the H120A mutant is shown in Figure 5.2. (In supporting information (Figure 5.S1), LT Abs and MCD peaks are fit simultaneously, and correlated to the peaks in the room temperature CD spectrum). Simulations of the Abs and MCD spectra identify a series of 7 electronic transitions in the range from 8000 to 22000cm-1. The high-pH form has two peaks (band 6 and 7) in

103 the Abs spectrum between 18000 and 22000 cm-1 and an associated intense derivative- shaped pseudo-A term in the low temperature MCD spectrum, which have been assigned 19 -1 as the Scys → Cu charge transfer (CT) transitions by rR. Band 2 at 13300 cm that is relatively intense in Abs, exhibits a negative C-term MCD signal and has been assigned by rR studies as the ψ→ψ* transition. This involves excitation of an electron from the

Cu A Cu B Cu A Cu B Cu-Cu bonding ( d 2 2 + d 2 2 ) orbital to the σ antibonding ( d 2 2 − d 2 2 ) σ g x −y x −y u x −y x −y orbital, and reflects a class III mixed valence system.19

7 6 6

5 A. LT Absorption

-1 4

cm -1 3 2 / mM ε

2

1

0 32000 28000 24000 20000 16000 12000 8000

Energy (cm-1)

1500 B. 5K, 7T MCD 1000

500 -1 cm

-1 0

/ M Δε -500

-1000

-1500 32000 28000 24000 20000 16000 12000 8000 Wavenumber Figure 5.2 Low-temperature absorption and MCD spectra of the high pH form (⎯), the low pH form (⎯) and the H120A mutant (⎯).

104 From Figure 5.2, the Abs and MCD spectra of the low-pH and mutant forms are equivalent. Both are dominated by three intense absorption features at ~19000 cm-1 (band 6), ~ 20400 cm-1 (band 7) and ~ 12200 cm-1 (band 2), with several additional weaker features also observed at lower and higher energies. These are also very similar to the Abs and MCD spectra of the high-pH form, but with an 1100 cm-1 shift of band 2 to lower energy both in Abs and MCD. The similarity among the spectra in Figure 5.2 confirms that the low-pH form and the H120A mutant are identical essentially, and can be treated as perturbations of the high-pH form.

A *

*

120 160 200 240 280 320 360 400 Raman Shift (cm-1)

* B *

120 160 200 240 280 320 360 400 -1 Raman Shift (cm ) Figure 5.3. Resonance Raman spectra of (A) S→Cu CT band exitation and (B) ψ→ψ* band excitation (High pH form ⎯, low pH form ⎯ and H120A mutant ⎯; solvent peaks are marked with stars).

105 5.3.1.3 Resonance Raman Spectroscopy Figure 5.3 shows the rR spectra obtained for the three proteins (rR excitation profiles are shown in Figure 5.S2 in supporting information). The rR spectra of the high-pH form are similar to those previously reported for the CuA site, but clarified by going to a high enough pH to eliminate low pH contributions. Excitation into the Scys → Cu CT transition at 477 nm produces strong resonance enhancement of the symmetric breathing -1 -1 mode at 337 cm (ν4), an out-of-phase “twisting” Cu-S stretching modes at 274 cm (ν3) -1 15 and a mixed Cu-S/Cu-N stretching mode at 256 cm (ν2), as described previously

(Chart 5.2). There is essentially no resonance enhancement of the Cu2S2 “accordion” mode (ν1, in Chart 5.2). However, excitation into absorption band 2 at 750 nm for the high pH form shows a dramatically different intensity distribution: the intensities of ν2 -1 52 and ν3 vibrations dramatically decrease, and that of the ν1 accordion mode at 158 cm -1 gains significant resonance enhancement relative to the 337 cm (ν4) breathing mode.

Thus excitation into band 2 activates the distortions of the Cu2S2 site along the ν1 and ν4 modes in the associated excited state, resulting in a large elongation of the Cu-Cu bond with little change in Cu-Scys and Cu-NHis bond lengths. Band 2 was assigned as the

ψ→ψ* transition since only totally symmetric vibrations of the Cu2S2 core are 16, 19 observed. The rR data are consistent with the valence delocalized ground state of CuA.

S S ν2 Chart 5.2 N Cu Cu Cu

S ν ν1 2

ν4 ν3 S Cu Cu Cu N S S

ν ν4 3

106 For the low-pH and H120A mutant forms, excitation into band 2 also gives a dramatic enhancement of the ν1 and ν4 modes with ν2 + ν3 greatly decreased; therefore, band 2 in the low-pH and H120A mutant forms can also be assigned as the ψ→ψ* transition, indicating an equivalent excited state distortion to that observed in the high-pH form.

Based on these excited state data, the low-pH and H120A mutant forms of CuA are still -1 delocalized. The ν1 goes up by 10 cm , indicating a stronger Cu-Cu bonding interaction.

Further, rR excitation into the Scys→Cu CT region in the low pH and mutant forms -1 shows that one of the two Cu-NHis vibrations contributing to ν2 at ~256 cm in the high pH form is eliminated (Note that there are intensity changes in the Cu-NHis vibrations at ~256 cm-1 in Figure 5.3A in going from the high pH form to the low pH form). This indicates that upon going to the low pH form, one histidine ligand is lost, as is also, of course, the case for the H120A mutant which exhibits the equivalent rR spectrum as the low pH form.

5.3.1.4 EXAFS Studies Figure 5.4A shows a comparison of the Cu K-edge EXAFS and the corresponding non- phase shift corrected Fourier transform data for the high pH and the low pH forms of CuA and the H120A mutant. Figure 5.S3-5 show the fits to these EXAFS data. EXAFS fit parameters are summarized in Table 5.2. The first shell of the high pH site was fit with one Cu-N contribution at 1.92 Å, two Cu-S contributions at 2.27 Å and one Cu-Cu contribution at 2.40 Å.53 The first shell of the low pH site was fit with a 0.5 Cu-N contribution at 1.95 Å, two Cu-S contributions at 2.26 Å and one Cu-Cu contribution at

2.36 Å. The first shell of H120A CuA was fit with a 0.5 Cu-N contribution per Cu at 1.93 Å, two Cu-S contributions at 2.26 Å and one Cu-Cu contribution at 2.36 Å. Second and third shells of the data for all three proteins were fit to the single and multiple scattering contributions from the histidine ring(s). The EXAFS fits indicate that the two Cu-S contributions (~2.27 Å) due to the two Cu-

SCys bonds remain very similar in the three sites. Interestingly, however, the EXAFS fits show that the low pH and H120A mutant data have only 0.5 Cu-N contribution to the first shell, indicating that one of the Cu-NHis ligands is lost at low pH and in the His to

Ala mutation. Note that DFT calculations (vide infra) indicate that the Cu-OGlu becomes

107 shorter for these two forms (~2.1 Å) and a weak Cu − O bond (2.26 Å) replaces the H2O

Cu-NHis bond (Table 5.3). Attempts to fit the EXAFS data with 0.5 Cu-O component at 2.1 Å and at 2.26 Å were unsuccessful indicating that, if present, the Cu-O bond does not significantly contribute to the EXAFS spectrum. This is consistent with the rR data, which indicates that one Cu-NHis ligand is lost at low pH or in the His to Ala mutation.

Although the first shell Cu-N component indicates the loss of one Cu-NHis ligand, a similar decrease in the second and third shell contribution is not observed (see Supporting

Information). This likely results from the decreased Cu-OGlu bond distance and associated increase in the multiple-scattering contribution from the carbon atoms of the glutamic acid ligand that can compensate for the loss of the Cu-NHis ligand.

Figure 5.4 (A) EXAFS data and (B) the corresponding non-phase shift corrected Fourier transforms of w.t. CuA azurin at high pH (⎯) and low pH (⎯) and H120A mutant (⎯).

108 The EXAFS data for the low pH sample and the H120A mutant overlap over the entire -1 k range. A significant difference in the range of k = 8-13 Å is observed between the EXAFS data for the high and low pH forms. The shift in the EXAFS beat-pattern to higher k in the low pH form and the H120A mutant is a clear indication of a change in the Cu-Cu distance, which has the maximum contribution to the EXAFS data in the high k region. Indeed, EXAFS fits show that the Cu-Cu component in the EXAFS data for the low pH sample and H120A mutant (2.36 Å) is shorter relative to that in the high pH form (2.40 Å) (Table 5.2). This is consistent with the rR data showing an increase in the frequency of ν1, which indicates an increase in the Cu-Cu bond strength.

Table 5.2 EXAFS Least Squares Fitting Results for WT CuA at high pH and low pH and for the H120A mutant.

a 2 2 b c Coord/Path R (Å) σ (Å ) E0 (eV) F

1 Cu-N/O 1.92 343 High pH 2 Cu-S 2.27 740 -14.7 0.10 form 1 Cu-Cu 2.40 635

1 Cu-N/O 1.95 343 Low pH 2 Cu-S 2.26 614 -12.6 0.05 form 1 Cu-Cu 2.36 668

1 Cu-N/O 1.92 389

H120A 2 Cu-S 2.26 648 -14.3 0.06

1 Cu-Cu 2.36 689

aThe estimated standard deviations for the distances are in the order of ± 0.02 Å (for the first shell). bThe σ2 values are multiplied by 105. c Error is given by 2 6 2 6 d 2 Σ[(χobsd–χcalcd) k ]/Σ[(χobsd) k ]. The σ factor of the multiple scattering path is linked to that of the corresponding single scattering path.

109 5.3.2 DFT Calculations 5.3.2.1 Optimized Structures. DFT calculations were used to evaluate the change in geometric and electronic structure of the CuA site associated with the protonation of one histidine ligand. In the geometry optimized 96-atom model of the high pH site (Figure 5.5A and Table

5.3), the Cu2S2 core is very symmetric with the Cu-Scys distances of 2.34 to 2.38 Å, and a

Cu-Cu distance of 2.50 Å. The equatorial Cu-Nhis distances are both 2.03 Å. The axial

Cu-Smet distance is calculated with 2.79 Å and the Cu-Oglu distance obtained is 2.34 Å.

Upon protonation of the His120 ligand on CuO (97-atom model, Figure 5.5B, the labels

CuS and CuO denote the copper with a weak axial methionine sulfur and carbonyl oxygen ligand, respectively), this His comes off and is replaced by a nearby H2O ligand with a Cu- O distance of 2.26 Å, and the Cu-Oglu distance is decreased to 2.08 Å. The O H 2O

Cu2S2 core is also perturbed with the Cu-Scys distances of 2.31 to 2.40 Å. The Cu-Cu distance is decreased to 2.46 Å and the remaining CuS-Nhis distance is not changed. The trends of bond length changes from 96-atom model to 97-atom model are consistent with the crystal structure data and EXAFS results of CuA site. From the crystal 54 structure of CuA azurin, two slightly different CuA sites, A and B in the asymmetric unit, were observed (Table 5.3). In both sites, the average Cu-S distances are similar (2.37 Å and 2.38 Å, respectively), and the CuS-Nhis distances are also very close (2.01 Å and 2.08 Å, respectively). However, for the B site relative to the A site, the Cu-Cu distance slightly decreased (2.42 Å to 2.35 Å), and the Cu-NHis distance at the CuO sites differs by 0.2 Å (2.06 Å and 2.26 Å). Since the protein was crystallized at pH = 5, near the pKa of

H120, it is possible that the two forms of CuA sites resolved in crystal structure are related to the high pH and low pH forms, although they can simply reflect independent refinements to disordered structures. The EXAFS data also reveal similar structural differences between the high pH and low pH forms (vide supra). In particular, the Cu-Cu distance of the high-pH form is 0.04 Å longer than that of the low-pH form, and only 0.5

Cu-NHis contribution with a distance of 1.92 Å was found in the low pH form. Thus, the 96-atom and 97-atom models reasonably reproduce the high pH and low pH forms, respectively, of CuA.

110

B A

CuS CuS

CuO

CuO

Glu H120 H120

Figure 5.5 DFT geometry optimized structures (A) 96-atom model of the high-pH form; (B) 97-atom model of the low-pH form/H120 mutant.

Table 5.3 The bond distances in the X-ray crystal structure of CuA azurin, DFT geometry optimized 96-atom and 97-atom models.

96-atom 97-atom CuA azurin crystal structure model model A B S1 – Cu 2.38 2.31 2.30 2.33 O S1 – CuS 2.34 2.41 2.29 2.44

S3 – CuS 2.36 2.40 2.42 2.30

S3 - Cu 2.34 2.33 2.46 3.43 O Cu - Cu 2.50 2.46 2.42 2.35

CuO - NHis/OH2O 2.02 2.26 2.06 2.26

CuS – NHis 2.04 2.03 2.01 2.08 SMet – CuS 2.79 2.64 2.98 3.16

CuO - Oglu 2.34 2.08 2.17 2.15

111 5.3.2.2 Ground State Electronic Structure. (i) Ground State Wavefunction

In the high-pH CuA model, the DFT calculations give a σu* ground state with a total of 48% electron spin (Mulliken population analysis) approximately delocalized over the two coppers, 44% electron spin over the two bridging Scys atoms, and a total of 7% electron spin over the two Nhis atoms (Figure 5.6A, Table 5.4). This spin density distribution is close to the experimental results (44% Cu character and 46% S character in the ground state wavefunction).33 For Cu atoms, the electron spin is mostly in 3d orbitals (Table 5.5). In the low pH model, the water ligand substitution of His120 produces a distorted ligand field on CuO (Figure 5.5B). However, the σu* ground state wavefunction of the low pH form still shows about the same electron spin delocalization over the two Cu atoms, two S atoms and the remaining N (Figure 5.6B, Table 5.4). Both the high pH and low pH forms are calculated to be valence delocalized which corroborates the correlation of the rR and Abs/MCD spectra. The electron spin on both centers is dominantly in

3d 2 2 orbitals consistent with the EPR g values (vide supra). However, the spin density x − y distribution of the low pH form has a limited but important difference relative to the high pH model. From Table 5.5, the distorted ligand field produces ~ 1% CuO 4s mixing into the ground state wavefunction of the binuclear site. As evaluated below, this ~ 1% 4s mixing dramatically changes the Fermi contact contribution to the hyperfine coupling for

CuO in the binuclear Cu center.

A B

Figure 5.6 β-LUMOs with the contour values of 0.03 a.u. (A) 96-atom model of the high pH form; (B) 97-atom model of the low pH form.

112 Table 5.4 Mulliken atomic spin densities of 96-atom and 97-atom models from Gaussian 03 calculations.

CuO CuS S1 S2 NO/O NS 96-atom model 26 22 19 25 4 3 97 atom model 24 25 21 23 1 4

Table 5.5 Löwdin population analyses of Cu Characters in β-LUMO of 96-atom and 97-atom models from Orca 2.5 calculations.

β-LUMO

s pz px py dz2 dxz dyz dx-y2 dxy dtotal*

96-atom CuO 0.1 0.2 0.2 0.0 0.2 1.0 0.0 28.6 0.1 29.9 model CuS 0.1 0.0 0.2 0.1 0.1 0.5 0.0 24.0 1.3 25.9

97-atom CuO 1.1 0.0 0.0 0.0 2.1 0.5 0.2 23.7 0.1 26.6 model CuS 0.0 0.1 0.2 0.0 0.8 2.5 0.0 25.7 0.3 29.3 * The Cu characters are slightly different from those in Table 4 because a different program, Orca, with a different population analysis (Löwdin) was employed (Hyperfine coupling calculations were also done in program Orca).

(ii) Cu Hyperfine Coupling These calculations have been performed with the program ORCA 2.5. Three terms contribute to the hyperfine coupling of the electron spin to the nuclear spin on the Cu centre: Fermi contact (FC), spin dipolar (SD) and orbital dipolar (OD).

From Table 5.6, the FC contributions to CuS in both high pH and low pH models are

CuS -4 -1 CuS -4 -1 very similar, Ahigh pH form (FC) = -53x10 cm , Alow pH form (FC)= -56x10 cm respectively.

-4 -1 However, this contribution to CuO has changed dramatically from -58x10 cm in the -4 -1 high-pH model to -20x10 cm in the low-pH model. The SD contributions to CuS for the two models are also very similar. In the low pH form, this term for CuO decreases A|| -4 -1 by 15x10 cm and results in a significantly rhombic A⊥. The OD terms for CuO and CuS both slightly increase in the A|| direction of the low pH form. The total calculated results for A|| reproduce the experimental results reasonably well, however, the calculated results for A⊥ are less satisfactory.

113 The EPR hyperfine coupling calculations are consistent with the experimental results for the high pH form, indicating that the two coppers equivalently contribute to the observed 7-line hyperfine coupling in the g|| range. Upon going to the low-pH form, the calculated total parallel hyperfine for CuS remains at the same, however, the calculated A|| is significantly decreased in CuO. Therefore, the hyperfine couplings for two Cu′s are no longer equivalent at low pH, and a 4-line pattern in the g|| region is expected and observed. As is clear from Table 5.6, this inequivalence dominantly reflects the difference in the FC term for CuO in going from high pH to low pH. This demonstrates that the 1% Cu 4s mixing into the ground state at low pH gives a large contribution to A|| as it produces direct positive spin density at the nucleus. In contrast to the high pH form, the A-tensor of the low pH structure is no longer colinear with the g tensor. The possible contribution of this effect on the EPR spectrum of the low pH form was evaluated through spectral simulations (Figure 5.S6 in supporting information). This A-tensor rotation reduces the hyperfine couplings of both Cu′s, but does not change the 4-line hyperfine coupling pattern. Thus, the colinear simulations of experimental spectra of the low pH form and the H120A mutant can reasonably reproduce the hyperfine pattern with only minor deviations in the intensity (Figure 5.1).

Table 5.6 Calculated Cu hyperfine Coupling Constants of the 96-atom and 97-atom models with Program ORCA 2.5 (10-4cm-1) A(FC) A(SD) A(OD) A(Tot)* -71 72 -52 A|| Cu -53 31 14 -8 96- S A 40 10 -3 ⊥ atom -78 86 -48 A model || Cu -58 31 18 -10 O A 47 12 1 ⊥

-76 91 -48 A|| Cu -56 29 21 -2 97- S A 48 12 4 ⊥ atom -63 91 17 A model || Cu -20 -1 27 -4 O A 64 4 49 ⊥ *Atotal is the eigenvalue of three contributions, which is collinear with g tensor in 96-atom model, but is noncollinear in 97-atom model

114

S

N Cu Cu N B3u y ↑ σ*u RAMO S x D2h

B2u

πu ↑↓ 2 2 x2-y2 x -y

xy xy

B1g ↑↓ π∗g

Ag

↑↓ σg z2 ↑↓ z2 ↑↓

↑↓ xz,yz ↑↓ xz,yz ↑↓ ↑↓

Figure 5.7 Simplified molecular orbital diagram of the in-plane Cu2S2 core in CuA site

demonstrates the coupling between two C NCuS monomers each with its 3d 2 2 orbital 2v 2 x − y

highest in energy, the 3d orbital second, followed by the3d 2 , 3d and 3d orbitals. xy z xz yz

5.3.2.3 Excited States

The bonding description developed for CuA is given in Figure 5.7. Each Cu has 5d orbitals, which split in D2h dimer symmetry into 10 levels due to the direct bonding interactions between Cu atoms and with the bridging S ligands. In addition, each thiolate has two valence 3p orbitals perpendicular to the S-C bond, which contribute to form molecular orbitals for electron delocalization in the Cu2S2 plane. With 19d electrons 1.5+ 1.5+ present in the Cu -Cu core, a half occupied ψ*(RAMO) is formed, which is the σu* (i.e. the Cu-Cu antibonding orbital) ground state. From Figure 5.7, a series of five parity-

115 allowed (g→u) metal d-orbital based electronic transitions are predicted for the mixed valence dimer, each involving the promotion of an electron from a doubly occupied orbital to the half occupied RAMO. In particular, the σg→σu* (ag→b3u) is the ψ→ψ* transition (band 2 in Figure 5.2). Alternatively, the four d-orbital based u→u transitions are parity forbidden. However, the energy of πu (b2u) has been estimated earlier from the 19, 31, 32 EPR g|| value of CuA and will be evaluated below from the MCD C-term associated with the ψ→ψ* transition. The sulfur based dimer orbitals give two parity 18, 19, 55 allowed transitions (Scys→Cu CT) involving the Sp orbitals in the xy plane. These correspond to bands 6 and 7 in Figure 5.2. The pH perturbations on these excited states

(in particular, the πu state), and their effects on the spectroscopy, are evaluated below. (i)TD-DFT

Calculations were applied to simulate the absorption spectrum of the high pH CuA and evaluate the effects of H120 protonation on this spectrum. In Figure 5.8, the calculated spectrum of the high-pH model reproduces well the experimental absorption spectrum. TD-DFT calculation of the low pH model shows no dramatic difference, with all peaks red shifted by ~1000 cm-1. The red shift in the low energy region of the low-pH form is consistent with experiment while the red shift in the CT region is not observed in experiment. The experimental (Figure 5.2) and computational (Figure 5.8) decrease of ~1000 cm-1 in the ψ→ψ* transition energy between the high and low pH forms reflects a ~10% decrease in 2HAB. Interestingly, the Cu-Cu bond distance decreases by 0.04 Å, the Raman -1 shift ν1 increases by 10 cm , and the spin density distribution over Cu2S2 exhibits only a minor change. Thus, a stronger direct Cu-Cu interaction is indicated, which is consistent with a calculated small increase in the Mayer bond order56 (from 0.37 to 0.39). However, the DFT optimized structure of the low pH form indicates a less symmetric Cu2S2 core with the Cu-S distance difference of up to 0.1 Å, a longer S-S distance (increased by 0.06

Å). This decreases the superexchange coupling contribution to 2HAB. Thus, the relatively limited experimental decrease in 2HAB reflects the net effects of opposite contributions of an increased direct Cu-Cu coupling and decreased superexchange contribution through the thiolate bridges.

116

10000 Sx,y→Cu CT

8000

6000

-1 cm

-1 /M

ε 4000 ψ→ψ*

2000

0 32000 28000 24000 20000 16000 12000 8000 -1 Energy (cm ) Figure 5.8 TD-DFT calculated absorption spectra of the high pH form model (solid line) and the low pH form model (dashed line).

-1 The TD-DFT calculations predict that the low-lying πu (b2u) state is 4000 cm above -1 the σu* (b3u) ground state in the high-pH model, and 2800 cm above the σu* ground state in the low-pH model. This πu state spin orbit (S.O.) couples into the σu* ground state and leads to the deviation of the EPR g|| from 2.0023. This is given by:

Cu 2 2 g|| ≈ ge + 8ζ 3d α β / Δ * , (5.1) σ u /π u

2 2 In which α is the total Cu character in the σu* ground state, β is the total Cu character in

Cu the πu excited-state, Δ * is the energy splitting between σu* and πu, andζ 3d is the Cu σ u /π u

-1 3d S.O. coupling constant. From the DFT calculations, Δ * deceases by 1200 cm in σ u /π u the low pH form while the Cu characters of the σu* and πu states are similar to those in the high pH form. Thus, the g|| value increases. This is consistent with the experiment: g||

= 2.173 for the high pH form, and g|| = 2.222 for the low pH form. Our previous study -1 gave an experimental estimate for Δ * of 5000 cm in the high pH form. From σ u /π u

117 -1 equation 1 and the experimental g|| value, a value of Δ * = 3900 cm is obtained for the σ u /π u low pH form, which is in qualitative agreement with the TD-DFT calculations.

(ii) The ψ → ψ* MCD C-term As given by simplified equation 5.2, MCD C-term intensity requires two perpendicular non-zero electronic dipole transition moments (Mi) that are also perpendicular to the Zeeman direction.

Δε ∝ g z M x M y + g x M y M z + g y M z M x (5.2)

In the < D2h symmetry of the CuA site, all states are nondegenerate, thus all the electronic transitions are unidirectional. Therefore, MCD intensity requires the S.O. coupling between states with perpendicular transition moments. From Figure 5.2, the LT MCD 57 spectra of CuA sites show two types of behavior: a derivative shaped, pseudo-A term in the Scys → Cu CT region (band 6 and 7), and a negative C-term feature in the ψ→ψ* transition region (band 2). The mechanism of the pseudo-A term in the CT region has been evaluated18, 19, 55. The non-zero C-term intensity can be obtained by S.O. coupling between two nearby excited states |J> and |K>, to which the orthogonal transitions are made from the ground state | A> (Chart 5.3 left). In CuA, the ground state is σu* and the two nearby excited states are correspond to the two thiolate ligand based CT states which are allowed and polarized in the Cu2S2 plane (i.e. the xy plane, x is along the Cu-Cu vector). The two excited states S.O. couple in the z direction. This produces the equal and opposite signed MCD features (the temperature dependent pseudo-A term) shown in Chart 5.3 left.

Chart 5.3

118 Alternatively, an isolated C-term feature would be obtained if there is a low-lying (non-thermally accessible) state |K′> with a transition to the same excited state |J> polarized perpendicular to the transition from the ground state |A>, that can S.O. couple in a third perpendicular direction (Figure 5.9 right). From the EPR and TD-DFT results of

CuA sites, the πu state or |K′> is energetically near the σu* ground state or |A> ( Δ * = π u /σ u

-1 5000 cm for the high pH form). This state can S.O. couple into the σu* ground state in the z direction. The ψ→ψ* transition (i.e. σg → σu*) is x polarized, and the σg → πu transition is y polarized as diagramed in Figure 5.9. The MCD signal for the ψ → ψ* transition is predicted to be negative. This is consistent with the negative C-term for the

ψ → ψ* transition of CuA.

σu* πu Acceptor MO Ground state MO Low lying state MO

b b σg σg Donor MO

Tranistion Density

mx my Tranistion Dipole Moment

Spin Orbit Rotation σu* πu

z y Lz

mx

x

my

RCP absorption negative MCD

b Figure 5.9 Graphic prediction of the C-term sign for the ψ→ψ* transition (i.e. σg →σu*).

119 In going from high pH to low pH, from both the TD-DFT calculations and EPR g|| -1 values, the Δ * energy is decreased by ~1000 cm . Thus, a more intense MCD signal π u /σ u of the ψ→ψ* transition is expected in the low pH form. Our experimental data show that the ratios of the MCD C-term intensity to the absorption intensity of band 2, the C0/D0 values, are 0.43 ± 0.06 in the high pH form and 0.55 ± 0.10 in the low pH form. The difference between these is within the experimental uncertainty, and the expected increase in the C0/D0 value can not be resolved unambiguously due to the overlap of band

2 and 3 in both the high and low pH forms. However, compared with the C0/D0 values of the d→d transitions of monomer Cu centers (i.e. a localized Cu site) that are ~0.1,58 the large C0/D0 value of band 2 in CuA (~0.5), strongly supports the valence delocalized descriptions for both the high and low pH forms. The large C0/D0 value reflects the presence of the low lying πu state that effectively S.O. couples into the ground state. In monomers and localized mixed valence systems, the d-d exited state energies are much higher, leading to less effective S.O. coupling into the ground state. This lowers the C0/D0 contribution from S.O. mixing into the ground state, thus, in localized systems the MCD C-term intensity mostly reflects S.O. coupling between ligand field excited states (i.e. pseudo-A terms).

5.4 Discussion 5.4.1 Spectral Probes of Delocalization

Apparent localization is observed from the EPR spectra of CuA at low-pH and in its H120A mutant. However, the rR and MCD C-term data provide clear evidence for valence delocalization. This is consistent with the Q-band ENDOR results on the H120A mutant in which the nitrogen ligand of the unperturbed histidine (His46) exhibits a very 37 similar hyperfine coupling to that of the delocalized CuA site. For the high-pH and the low-pH/H120A models, the DFT calculations reveal a delocalized σu* ground state with a similar spin density distribution over the Cu2S2NHis46 core. In the high pH form, each Cu has a relatively small negative contribution to the

hyperfine coupling due to the delocalization of the electron spin over d 2 2 orbitals on x − y the two Cu centers. In the low pH form, the ~1% 4s mixing due to the distorted ligand

120 field of CuA adds a direct Fermi contact contribution to the hyperfine coupling, which is large and positive. The net effect is to generate a smaller CuO hyperfine coupling to the electron spin, even though it has about as much total spin density as the non-perturbed

CuS center. Thus the EPR spectrum can be misleading with respect to delocalization, due to the potentially large effects of small contributions to the ground state wavefunction.

5.4.2 Factors Effecting Delocalization

The electron delocalization of the high pH form of CuA has been evaluated in terms of 19, 59, 60 the Q- mode in PKS model (equation 5.3).

2 1 ⎛ Λ2 ⎞ ⎡1 ⎛ Λ2 ΔE ⎞ ⎤ E ± = ⎜ ⎟x2 ± ⎢ ⎜ x + ⎟ + H 2 ⎥ (5.3) 2 ⎜ k ⎟ − 2 ⎜ k − 2 ⎟ AB ⎝ − ⎠ ⎣⎢ ⎝ − ⎠ ⎦⎥ The dimensionless coordinate x_ is along the Q_ mode (Q_ is the antisymmetric Λ2 combination of the breathing modes of the two Cu centers), is the vibronic trapping k− term associated with the change in the Cu-L bond lengths upon oxidation, ΔE is the potential difference between monomers, and HAB is the electronic coupling matrix element regulating electron delocalization between Cu atoms. In the high pH form of

CuA, the two coppers have approximately equivalent ligand field (ΔE ~ 0), and the large -1 2HAB (13300 cm ) overcomes the vibronic trapping and keeps the CuA center delocalized even in the low symmetry protein environment. Thus, the two Cu atoms of the high pH form have equivalent contributions to the ground state wavefuction, i.e. complete delocalization. In the low pH form, one histidine is protonated, and the two Cu atoms are no longer 2+ equivalent. From DFT calculations with either CuO or CuS replaced by Zn , we estimate an energy difference of ΔE ~ 0.120 eV favoring the CuO oxidation in the low pH geometry. The electronic coupling term 2HAB in the low pH form decreases by ~10%. This reflects the decrease in the superexchange contribution (vide supra). Using this experimental 2HAB and the calculated ΔE, regression analysis of equation 3 gives the relative contribution of CuO:CuS = 58:42 to the ground state wave function. The ground and excited state potential energy surfaces of the low pH form are plotted in Figure 5.10.

121 The two dashed lines indicate the potential energy surfaces of valence-trapped CuA. Only one minimum is observed in the ground state potential energy surface, indicating that the -1 low pH form is still a delocalized system and the 2HAB (~12000 cm ) is large enough to overcome vibronic trapping even with the inequivalent potentials of the two Cu centers.

10000

8000 Cu A 6000 )

-1 φ b 4000 Ψ* 2000 φ 0 a -1 -2000 12200cm -4000

Relative Energy(cm Ψ -6000

-2 -1 0 1 2 x -

Figure 5.10 Potential energy surfaces in Q- mode for the low-pH form of CuA site. The

-1 2 -1 specific parameters used are: 2HAB =12000 cm , Λ / k− = 2450 cm , ΔE = 0.120 eV.

5.4.3 Correlation of Electronic Structure to ET and its Regulation by [H+]

The valence delocalized electronic structure of CuA contributes to its efficiency in biological ET. According to semi-classical Marcus theory, the rate of long range ET 61-63 is dependent upon the following factors: λ, HDA, and ΔG°, where λ is the reorganization energy associated with the active site geometry change with redox, HDA is the through protein electronic coupling between the donor and acceptor, and ΔG° is the thermodynamic driving force. The rate of electron transfer is enhanced by maximizing

HDA while minimizing the sum of λ+ ΔG°. Here we consider how these factors change in going to the low pH form. (i) Reorganization Energy 64, 65 The high pH form of CuA has a reorganization energy of ~0.4 eV , which is ~½ that of the monomer (i.e. localized) blue copper site (~0.80 eV).58 This low

122 reorganization energy for CuA results from the delocalized nature of the ground state that distributes redox induced geometric changes over the 2Cu atoms.19, 66 In going to the low pH form, the reorganization energy increases by 0.18 eV, as determined by Pecht and coworkers.67 Consistent with the experimental results, the calculated inner sphere reorganization energies are 0.36 eV for the high pH model, and 0.50 eV for the low pH model. Pecht and coworkers attributed this increase to localization in the low pH form. However, from the above spectroscopic results, the low pH form is delocalized. In the low pH form, there is a H2O ligand on CuO, which will be lost on reduction. In the mononuclear (i.e. localized) red copper site, there is also a H2O ligand that is lost upon reduction. The reorganization energy of red copper is 1.2 eV.68 Thus, even in the low pH form of CuA, the delocalization over 2Cu atoms significantly decreases the effect on λ associated with the loss of a H2O ligand upon reduction. (ii) Superexchange Pathways

Two efficient intra-molecular ET pathways from CuA to heme a have been recognized33 (Figure 5.11). Pathway 1 (His204 to Heme a) has a calculated ET rate similar to the experimental rate (1x104 s-1). The calculated ET rate of Pathway 2 (Cys200 to Heme a) is two orders of magnitude less efficient than pathway 1 (8.0x101 s-1).

However, correcting this for the anisotropic covalency of the Cu-Scys bond of pathway 2 relative to the Cu-Nhis bond of pathway 1, pathway 2 becomes comparable to pathway 1.33

The H120 ligand in CuA Azurin corresponds to H204 in bovine heart CcO, which is along ET pathway 1 (Figure 5.11). There are three direct effects in the ET rate due to the

H204 protonation at low pH: the loss of the pathway 1 in CuA→ heme a ET, an increase 36 in λ, and an increase in the CuA reduction potential by ~70 mV. Considering the second

2 and third factors and assuming the same donor-acceptor coupling strength H DA in ET for both pH forms, the ET efficiency of the low pH form is predicated to be ~11 fold slower than the high pH form. Experimentally (in the related Rs. CcO and its H260N mutant69), this ratio is much larger (2000 fold in Table 5.7). This ~180 fold difference cannot result from the loss of one of two comparable pathways that can only give ~ 2 fold decrease. However, pathway 2 involves a 1.86 Å through space jump from an Ile to an Arg residue.

The Ile is next to the Glu that is axially coordinated to CuO. In the calculated geometric

123 structure of the low pH/His mutant model, the distance of CuO to Oglu ligand decreases by 0.26 Å with respect to the high pH model. This elongates the space between the Ile and Arg residues. In addition, based on the crystal structure (PDB:1V54), the imidazole ring of H204 in CcO is near the Ile residue, and its protonation can shift the Ile residue further away from the Arg residue. The calculated through space jump decays rapidly

2 −1.7(R−1.4) 2 61, 70 2 ( H DA ∝ (0.6e ) , where R is the through space distance); H DA can decrease by ~ 30 fold if the through space distance is increased by ~1.0 Å. Combined, these factors would contribute ~660 fold decrease in the ET rate in the low pH form.

Figure 5.11 Proposed ET pathways in bovine heart CcO based on Pathways analysis

(reference 34). The Cys200 and His204 CuA-to-heme a pathways are comparable in rate.

124

Table 5.7 CuA to Heme a ET rate comparison between the high and low pH forms of CuA.

CuA → heme a

Forms High pH Low pH λ (eV) 0.40 0.49

ΔG (eV)a -0.05 0.03

b highp H kET (Calc) c low pH 11 kET a -1 Experiment kET s 90,000 45

highp H kET (Exp) low pH 2000 kET a. Experimental results are from kinetic analyses of ET rate of the Rs. CcO and its H260N mutant. The CuA site in H260N mutant corresponds to the low pH form in this table. The reduction potential of CuA site in H260N mutant increases by ~90 mV (reference 68) (the reduction potential of the low pH form of CuA azurin increases by ~70 mV). o 2 π 2 ⎡ (ΔG + λ ) ⎤ k = ⋅ ( H ) ⋅ exp − b. ET 2 DA ⎢ ⎥ 4λk T h λk B T ⎣ B ⎦ 2 c. Assuming the same donor-acceptor coupling strength (HDA) for both pH forms.

(iii) Possible Contribution to Function While the identification of the exit pathway for pumped protons is still lacking, the direct role of CuA in proton pumping lost support with the discovery of the heme-Cu 71, 72 oxidase of E. Coli, cytochrome bo3 which pumps protons but does not contain CuA.

However, the reversible pH effect on CuA studied above could play a role in PCET through regulating ET upon local proton accumulation.

5.5 Summary This study provides a detailed understanding of the pH effect on the geometric and electronic structure of the CuA site. In particular, EXAFS studies show that the bond distances in Cu2S2 core of the high pH and low pH forms of CuA and its H120A mutant are very similar, and the rR and MCD spectra demonstrate electron delocalization. This is

125 due to the large 2HAB, and dominantly reflects a strong direct Cu-Cu interaction that keeps the CuA site delocalized in the low-symmetry protein environment, even with the protonation and loss of a His ligand. DFT calculations support this electron delocalization description and provide insight into the change in spectral features with pH. Importantly, the 4-line EPR hyperfine coupling pattern in the low pH form is not due to electron localization but rather reflects ~1% CuO 4s mixing into the ground state spin wavefunction. This reversible pH effect can play an important role in regulating ET under in vivo conditions.

5.6 Acknowledgement This research was supported by the NSF Grant CHE 0446304 (E.I.S), NIH RR-01209 (K.O.H.), NSF CHE98-76457 and NSF CHE-0552008 (Y.L). EXAFS experiments were performed at SSRL, which is funded by the DOE Office of Basic Energy Sciences. The SSRL Structural Molecular Biology Program is supported by the NIH National Center for Research Resources, Biomedical Technology Program, and by the DOE Office of Biological and Environmental Research. The project described was supported by Grant Number P41 RR-001209 from the National Center for Research Resources (NCRR), a component of the National Institutes of Health (NIH) and its contents are solely the responsibility of the authors and do not necessarily represent the official view of NCRR or NIH.

126

5.7 Supporting Information

A 5 5 20775 4 4

3 LT Abs 3 18734

2 2 13280 17320 1 22404 1 15549 14404 28835 23601 0 12023 0 7 6 5 4 3 2 1 20032000 24000 16000 8000200 18892 100 5K MCD 100 17325 28866 15490 0 23480 0 12000 22422 14404 13280 -100 -100

21223 -200 -200 32000 24000 16000 8000 200 200

100 RT CD 100

0 0

-100 -100

-200 -200 32000 24000 16000 8000

Figure 5.S1. (A) 5K 1T MCD, LT Abs and RT CD of the high pH form. Abs and MCD peaks are fitted simultaneously, and compared to RT CD peak fitting.

127

B 7 7 6 6 20407 5 19020 5 LT Abs 4 4

3 3

2 21687 2 12200 17136 1 15700 12971 1 28524 23492 0 9550 0 40032000 24000 7 6 516000 4 3 2 1 8000400 19020

200 5K MCD 200 17136 28522 15700 0 0 23410 12971 9550 21669 12200 -200 -200 20407

-400 -400 32000 24000 16000 8000 200 200

100 RT CD 100

0 0

-100 -100

-200 -200 32000 24000 16000 8000

Figure 5.S1 (B) 5K 1T MCD, LT Abs and RT CD of the low pH form. Abs and MCD peaks are fitted simultaneously, and compared to RT CD peak fitting.

128

5

ν 1 avg.(ν +ν3) 4 2 ν 4 pH7 LT Abs

-1 3 cm -1 mM 2

1

0 32000 30000 28000 26000 24000 22000 20000 18000 16000 14000 12000 10000 8000

Wavenumber(cm-1)

6

5 ν 1 avg.(ν +ν ) 2 3 ν 4 4 pH4 LT Abs

-1

cm -1 3

mM

2

1

0 32000 30000 28000 26000 24000 22000 20000 18000 16000 14000 12000 10000 8000

Wavenumber(cm-1)

Figure 5.S2 ~77K resonance Raman excitation profiles of CuA Azurin: (a) High-pH form; (b) Low-pH form.

129

Figure 5.S3 Fourier transforms (non phase shift corrected) and EXAFS data (Inset) of

WT P.a purple CuA construct at high pH. Data (―), Fit (―).

Table 5.S1 Complete EXAFS Least Squares Fitting Results for WT CuA at pH6. a 2 2 b c Coord/Path R (Å) σ (Å ) E0 (eV) F 1 Cu-N/O 1.92 343 2 Cu-S 2.27 740 1 Cu-Cu 2.40 635 2 Cu-C/N 2.93 606 -14.7 0.10 4 Cu-C/N-C/N 3.14 606d 2 Cu-C/N 4.20 142 8 Cu-C/N-C/N 4.35 142d aThe estimated standard deviations for the distances are in the order of ± 0.02 Å (for the first shell). bThe σ2 values are multiplied by 105. cError is given by 2 6 2 6 d 2 Σ[(χobsd–χcalcd) k ]/Σ[(χobsd) k ]. The σ factor of the multiple scattering path is linked to that of the corresponding single scattering path.

130

Figure 5.S4 Fourier transforms (non phase shift corrected) and EXAFS data (Inset) of

WT P.a purple CuA construct at low pH. Data (―), Fit (―).

Table 5.S2 Complete EXAFS Least Squares Fitting Results for WT CuA at pH4. a 2 2 b c Coord/Path R (Å) σ (Å ) E0 (eV) F 0.5 Cu-N/O 1.95 343 2 Cu-S 2.26 614 1 Cu-Cu 2.36 668 2 Cu-C/N 2.92 308 -12.6 0.05 4 Cu-C/N-C/N 3.08 308d 2 Cu-C/N 4.21 404 8 Cu-C/N-C/N 4.36 404d aThe estimated standard deviations for the distances are in the order of ± 0.02 Å (for the first shell). bThe σ2 values are multiplied by 105. cError is given by 2 6 2 6 d 2 Σ[(χobsd–χcalcd) k ]/Σ[(χobsd) k ]. The σ factor of the multiple scattering path is linked to that of the corresponding single scattering path.

131

Figure 5.S5 Fourier transforms (non phase shift corrected) and EXAFS data (Inset) of the

H120A mutant of P.a purple CuA. Data (―), Fit (―).

Table 5.S3 Complete EXAFS Least Squares Fitting Results for H120A mutant. a 2 2 b c Coord/Path R (Å) σ (Å ) E0 (eV) F 0.5 Cu-N/O 1.92 389 2 Cu-S 2.26 648 1 Cu-Cu 2.36 689 2 Cu-C/N 2.92 430 -14.3 0.06 4 Cu-C/N-C/N 3.06 430d 2 Cu-C/N 4.18 981 8 Cu-C/N-C/N 4.34 981d aThe estimated standard deviations for the distances are in the order of ± 0.02 Å (for the first shell). bThe σ2 values are multiplied by 105. cError is given by 2 6 2 6 d 2 Σ[(χobsd–χcalcd) k ]/Σ[(χobsd) k ]. The σ factor of the multiple scattering path is linked to that of the corresponding single scattering path.

132

A

B

C

D

E

2800 3000 3200 3400 3600 Gauss

Figure 5.S6 (A) X-band EPR spectrum of the high pH form of CuA; (B) XSophe simulations of A. (C) XSophe simulations of A with only the A-tensor rotation considered. (D) XSophe simulations of A with Fermi contact change only on CuO site. (E)

X-band EPR spectrum of the low pH form of CuA.

5.8 Reference 1. Saraste, M. Q., Rev. Biophys. 1990, 23, 331-366. 2. Malmström, B. G., Chem. Rev. 1990, 90, 1247-1260. 3. Babcock, G. T.; Wikström, M., Nature 1992, 356, 301-309. 4. Musser, S. M.; Stowell, M. H. B.; Chan, S. I., Comparison of ubiquinol and cytochrome c terminal oxidases An alternative view. FEBS Lett. 1993, 327, 131- 136. 5. Garcίa-Horsman, J. A.; Barquera, B.; Rumbley, J.; Ma, J.; Gennis, R., B. J. Bacteriol. 1994, 176, 5587-5600.

133 6. Wikström, M., Cytochrome c oxidase: 25 years of the elusive proton pump. Biochim. Biophys. Acta-Bioenerg. 2004, 1655, 241. 7. Kroneck, P. M. H.; Riester, J.; Zumft, W. G.; Antholine, W. E., Biol. Met. 1990, 3, 103-109. 8. Zumft, W. G.; Kroneck, P. M. H.; Robert, K. P., Respiratory Transformation of Nitrous Oxide (N2O) to Dinitrogen by Bacteria and Archaea. Adv. Microb. Physiol. 2006, 52, 107. 9. Suharti; Strampraad, M. J. F.; Schroder, I.; de Vries, S., A Novel Copper A Containing Menaquinol NO Reductase from Bacillus azotoformans. Biochemistry 2001, 40, (8), 2632-2639. 10. Suharti; Heering, H. A.; deVries, S., NO Reductase from Bacillus azotoformans Is a Bifunctional Enzyme Accepting Electrons from Menaquinol and a Specific Endogenous Membrane-Bound Cytochrome c551. Biochemistry 2004, 43, (42), 13487-13495. 11. Tsukihara, T.; Aoyama, H.; Yamashita, E.; Tomizaki, T.; Yamaguchi, H.; Shinzawa-Itoh, K.; Nakashima, R.; Yaono, R.; Yoshikawa, S., Structures of metal sites of oxidized bovine heart cytochrome c oxidase at 2.8 Å. Science 1995, 269, (5227), 1069-1074. 12. Iwata, S.; Ostermeier, C.; Ludig, B.; Michel, H., Structure at 2.8 Å resolution of cytochrome c oxidase from Paracoccus denitrificans. Nature 1995, 376, (6542), 660-669. 13. Blackburn, N. J.; Barr, M. E.; Woodruff, W. H.; van der Oost, J.; de Vries, S., Metal-Metal Bonding in Biology: EXAFS Evidence for a 2.5 Å Copper-Copper

Bond in the CuA Center of Cytochrome Oxidase. Biochemistry 1994, 33, 10401- 10407. 14. Blackburn, N. J.; de Vries, S.; Barr, M. E.; Houser, R. P.; Tolman, W. B.; Sanders, D.; Fee, J. A., X-ray Absorption Studies on the Mixed-Valence and Fully

Reduced Forms of the Soluble CuA Domains of Cytochrome c Oxidase. J. Am. Chem. Soc. 1997, 119, 6135-6143. 15. Andrew, C. R.; Fraczkiewicz, R.; Czernuszewicz, R. S.; Lappalainen, P.; Saraste, M.; Sanders-Loehr, J., Identification and Description of Copper-Thiolate

134 Vibrations in the Dinuclear CuA Site of Cytochrome c Oxidase. J. Am. Chem. Soc. 1996, 118, (43), 10436-10445. 16. Wallace-Williams, S. E.; James, C. A.; de Vries, S.; Saraste, M.; Lappalainen, P.; van der Oost, J.; Fabian, M.; Palmer, G.; Woodruff, W. H., Far-Red Resonance Raman Study of Copper A in Subunit II of Cytochrome c Oxidase. J. Am. Chem. Soc. 1996, 118, (16), 3986-3987. 17. Greenwood, C.; Hill, B. C.; Barber, D.; Eglinton, D. G.; Thomson, A. J., The

optical properties of CuA in bovine cytochrome c oxidase determined by low- temperature magnetic-circular-dichroism spectroscopy. Biochem. J. 1983, 215, (2), 303-316. 18. Farrar, J. A.; Neese, F.; Lappalainen, P.; Kroneck, P. M. H.; Saraste, M.; Zumft,

W. G.; Thomson, A. J., The Electronic Structure of CuA: A Novel Mixed-Valence Dinuclear Copper Electron-Transfer Center. J. Am. Chem. Soc. 1996, 118, 11501-11514. 19. Hay, M. T.; Ang, M. C.; Gamelin, D. R.; Solomon, E. I.; Antholine, W. E.; Ralle, M.; Blackburn, N. J.; Massey, P. D.; Wang, X.; Kwon, A. H.; Lu, Y.,

Spectroscopic Characterization of an Engineered Purple CuA Center in Azurin. Inorg. Chem. 1998, 37, 191-198. 20. Salgado, J.; Warmerdam, G.; Bubacco, L.; Canters, G. W., Understanding the Electronic Properties of the CuA Site from the Soluble Domain of Cytochrome c Oxidase through Paramagnetic 1H NMR. Biochemistry 1998, 37, 7378-7389. 21. Kroneck, P.; Antholine, W.; Riester, J.; Zumft, W., The Cupric Site in Nitous- Oxide Reductase contains a mixed-valence [Cu(II),Cu(I)] Binuclear center - A Multifrequency Electron-Paramagnetic Resonance Investigation. FEBS Lett. 1988, 242, (1), 70-74. 22. von Wachenfeldt, C.; de Vries, S.; van der Oost, J., FEBS Lett. 1994, 340, 109- 113. 23. Lappalainen, P.; Aasa, R.; Malmström, B. G.; Saraste, M., J. Biol. Chem. Rev. 1993, 268, 26416-26421. 24. Slutter, C. E.; Sanders, D.; Wittung, P.; Malmström, B. G.; Aasa, R.; Richards, J. H.; Gray, H. B.; Fee, J. A., Biochemistry 1996, 36, 3387-3395.

135 25. Zumft, W. G.; Viebrock-Sambale, A.; Braun, C., Eur. J. Biochem. 1990, 192, 591-599. 26. Hay, M.; Richards, J.; Lu, Y., Proc Natl. Acad. Sci. U.S.A. 1996, 93, 461-464. 27. Dennison, C.; Vijgenboom, E.; de Vries, S.; van der Oost, J.; Canters, G.,

Introduction of a CuA site into the blue copper protein amicyanin from Thiobacillus versutus. FEBS Lett. 1995, 365, 92-94. 28. Houser, R. P.; Young, V. G., Jr.; Tolman, W. B., J. Am. Chem. Soc. 1996, 118, 2101-2102. 29. Tsukihara, T.; Aoyama, H.; Yamashita, E.; Tomizaki, T.; Yamaguchi, H.; Shinzawa-Itoh, K.; Nakashima, R.; Yaono, R.; Yoshikawa, S., The whole structure of the 13-subunit oxidized cytochrome c oxidase at 2.8 Å. Science 1996, 272, (5265), 1136-1144. 30. Robin, M. B.; Day, P., Adv. Inorg. Chem. Radiochem. 1967, 10, 247. 31. Neese, F.; Zumft, W. G.; Antholine, W. E.; Kroneck, P. M. H., The Purple Mixed-Valence CuA Center in Nitrous Oxide Reductase: EPR of the Copper-63-, Copper-65-, and Both Copper-65- and [15N]Histidine-Enriched Enzyme and a Molecular Orbital Interpretation. J. Am. Chem. Soc. 1996, 118, 8692-8699. 32. Gorelsky, S. I.; Xie, X.; Chen, Y.; Fee, J. A.; Solomon, E. I., The Two-State Issue in the Mixed-Valence Binuclear CuA Center in Cytochrome c Oxidase and N2O Reductase. J. Am. Chem. Soc. 2006, 128, (51), 16452-16453. 33. George, S. D.; Metz, M.; Szilagyi, R. K.; Wang, H.; Cramer, S. P.; Lu, Y.; Tolman, W. B.; Hedman, B.; Hodgson, K. O.; Solomon, E. I., A Quantitative Description of the Ground State Wavefunction of CuA by X-ray Absorption Spectroscopy: Comparison to Plastocyanin and Relevance to Electron Transfer. J. Am. Chem. Soc. 2001, 123, 5757-5767. 34. Gurbiel, R. J.; Fann, Y. C.; Surerus, K. K.; Werst, M. M.; Musser, S. M.; Doan, P. E.; Chan, S. I.; Fee, J. A.; Hoffman, B. M., Detection of two histidyl ligands to CuA of cytochrome oxidase by 35-GHz ENDOR. 14,15N and 63,65Cu ENDOR studies of the CuA site in bovine heart cytochrome aa3 and cytochromes caa3 and ba3 from Thermus thermophilus. J. Am. Chem. Soc. 1993, 115, (23), 10888- 10894.

136 35. Neese, F.; Kappl, R.; Hüttermann, J.; Zumft, W. G.; Kroneck, P. M. H., Probing the ground state of the purple mixed valence CuA center in nitrous oxide reductase: a CW ENDOR (X-band) study of the 65Cu, 15N-histidine labeled enzyme and interpretation of hyperfine couplings by molecular orbital calculations. J. Biol. Inorg. Chem. 1998, 3, 53-67. 36. Hwang, H. J.; Lu, L., pH-dependent transition between delocalized and trapped

valence states of a CuA center and its possible role in proton-coupled electron transfer. Proc. Natl. Acad. Sci. U.S.A. 2004, 101, 12842-12847. 37. Lukoyanov, D.; Berry, S. M.; Lu, Y.; Antholine, W. E.; Scholes, C. P., Role of Coordinating Histidine in Altering the Mixed Valency of CuA: An Electron Nuclear Double Resonance-Electron Paramagnetic Resonance Investigation. Biophys. J. 2002, 82, 2758-2766. 38. Wang, X.; Berry, S. M.; Xia, Y.; Lu, Y., The Role of Histidine Ligands in the

Structure of Purple CuA Azurin. J. Am. Chem. Soc. 1999, 121, 7449-7450. 39. PySpline, Tenderholt, A. 2005, PySpline, Stanford University, Stanford, CA- 94305. In 2005. 40. Muestre de Leon, J.; Rehr, J. J.; Zabinsky, S. I.; Albers, R. C., Phys. Rev. B. 1991, 44, 4146-4156. 41. Rehr, J. J.; Muestre de Leon, J.; Zabinsky, S. I.; Albers, R. C., J. Am. Chem. Soc. 1991, 113, 5135-5140. 42. George, G. N. EXAFSPAK & EDG_FIT, EXAFSPAK & EDG_FIT, Stanford Synchrotron Radiation Laboratory, Stanford Linear Accelerator Center, Stanford University, Stanford, CA 94309, 2000. 43. 03, G., Frisch, M. J.; et al. Gaussian 03, Revision C.02; Gaussian, Inc.: Wallingford, CT, 2004. In. 44. Perdew, J. P., Phys. Rev. B. 1986, 33, 8822. 45. Becke, A. D., Phys. Rev. A 1988, 38, 3098. 46. Becke, A. D., J. Chem. Phys. 1993, 98, 5648. 47. SWizard, Gorelsky, S. I. SWizard; Department of Chemistry, York University: Toronto, ON, 1999 (http://www.sg-chem.net). In 1999.

137 48. Neese, Neese, F. ORCA, version 2.5; Universität Bonn: Bonn, Germany. The program is available free of charge at http://www.thch.uni-bonn.de/tc/orca. In 2007. 49. Neese, F., Inorg. Chim. Acta 2002, 337, 181-192. 50. Schäfer, A.; Horn, H.; Ahlrichs, R., J. Chem. Phys. 1992, 97, 2571-2577. 51. Basis.set.library-usedbyorca, ftp.chemie.uni-karlsruhe.de/pub/basen. -1 52. Accordion-Mode, This is increased from the 130 cm of past studies on B.s. CuA

due to a shorter Cu-Cu bond in CuA azurin. In 2007.

53. EXAFS, The Cu K-edge EXAFS data of w.t. CuA azurin at pH 5.1 has been

previously reported in reference 32. The pH titrations on w.t. CuA azurin show a pKa value of ~5 for the two different pH dependent electronic structures observed from UV-vis and EPR data. Thus, the reported EXAFS spectrum represents an approximately equal mixture of the two forms and hence quantitatively differs from the high pH spectrum presented here. In 2007. 54. Robinson, H.; Ang, M. C.; Gao, Y.-G.; Hay, M. T.; Lu, Y.; Wang, A. H.-J., Structural Basis of Eelctron Transfer Modulation in the Purple CuA Center. Biochemistry 1999, 38, (18), 5677-5683. 55. Neese, F.; Solomon, E. I., MCD C-Term Signs, Saturation Behavior, and Determination of Band Polarizations in Randomly Oriented Systems with Spin S >= 1/2. Applications to S = 1/2 and S = 5/2. Inorg. Chem. 1999, 38, (8), 1847- 1865. 56. Mayer, I., Chem. Phys. Lett. 1983, 97, 270. 57. pseudo-A, The derivative shape leads to the A-term description, but the temperature dependent of this signal shows that it is a combination of equal but opposite C-terms. In. 58. Solomon, E. I.; Szilagyi, R. K.; George, S. D.; Basumallick, L., Electronic Structures of Metal Sites in Proteins and Models: Contributions to Function in Blue Copper Proteins. Chem. Rev. 2004, 104, (2), 419-458. - 59. Q_-Mode, With the monomeric breathing modes QA and QB, the Q_ mode is 2 1/2 (QA-QB). The dimensionless coordinate x_ is along the Q_ mode, x_=Q_/(Λ/k_), where k_=4π2c2μ_ν2_ in which ν_ and μ_ are the frequency and modal mass of

138 the Q_ mode, respectively. The quantity Λ is the vibronic coupling parameter 2 1/2 2 where Λ /k_≈k_[n Δrredox] , in which n is the number of metal-ligand bond

length changes and Δrredox is the difference in metal-ligand bond lengths between the oxidized and reduced structures. In 2007. 60. Piepho, S. B.; Krausz, E. R.; Schatz, P. N., J. Am. Chem. Soc. 1978, 100, 2996- 3005. 61. Beratan, J. N.; Betts, J. N.; Onuchic, J. N., Science 1991, 252, 1285-1288. 62. Marcus, R. A.; Sutin, N., Biochim. Biophys. Acta 1985, 811, 265-322. 63. Newton, M. D., Chem. Rev. 1991, 91, 767-792. 64. Brezezinski, P., Biochemistry 1996, 35, 5611-5615. 65. Farver, O.; Lu, Y.; Ang, M. C.; Pecht, I., Proc. Natl. Acad. Sci. U.S.A. 1999, 96, 899-902.

66. Larrson, S.; Källebring, B.; Wittung, P.; Malmström, B. G., The CuA center of cytochrome-c oxidase: Electronic structure and spectra of models compared to the

properties of CuA domains. Proc. Natl. Acad. Sci. U.S.A. 1995, 92, 7167. 67. Farver, O.; Hwang, H. J.; Lu, Y.; Pecht, I., Reorganization Energy of the CuA Center in Purple Azurin: Impact of the Mixed Valence-to-Trapped Valence State Transition. J. Phys. Chem. B 2007, 111, (24), 6690-6694. 68. Basumallick, L.; Sarangi, R.; George, S. D.; Elmore, B.; Hooper, A. B.; Hedman, B.; Hodgson, K. O.; Solomon, E. I., Spectroscopic and Density Functional Studies of the Red Copper Site in Nitrosocyanin: Role of the Protein in Determining Active Site Geometric and Electronic Structure. J. Am. Chem. Soc. 2005, 127, (10), 3531-3544. 69. Wang, K.; Geren, L.; Zhen, Y.; Ma, L.; Ferguson-Miller, S.; Durham, B.; Millett, F., Mutants of the CuA Site in Cytochrome c Oxidase of Rhodobacter sphaeroides: II. Rapid Kinetic Analysis of Electron Transfer. Biochemistry 2002, 41, (7), 2298- 2304. 70. Beratan, D. N.; Onuchic, J. N.; Betts, J. N.; Bowler, B. E.; Gray, H. B., Electron tunneling pathways in ruthenated proteins. J. Am. Chem. Soc. 1990, 112, 7915- 7921.

139 71. Puustinen, A.; Finel, M.; Virkki, M.; Wikström, M., Cytochrome o (bo) is a proton pump in Paracoccus denitrificans and Escherichia coli. FEBS Lett. 1989, 249, (2), 163-167. 72. Hosler, J. P.; Ferguson-Miller, S.; Mills, D. A., Energy Transduction: Proton Transfer Through the Respiratory Complexes. Annu. Rev. Biochem. 2006, 75, 165-187.

140

Chapter 6

The Fe-Smet Bond in Ferricytochrome c

141 6.1 Introduction The mitochondrial cytochromes c (cyt c), are a unique class of heme containing proteins, participating in electron transfer and apoptosis in biological systems.1-3 The structures of mitochondrial cyt c from various sources have been determined by X-ray crystallography,4-6 NMR7-10 and EXAFS.6, 11 The heme group of cyt c is covalently linked to the protein polymer through thioether linkages via Cys14 and Cys17 (numberings of protein residues here and after are referred to horse heart cyt c 1HRC12). The heme and two axial ligands (the imidazole Nε atom of His18 and the S atom of the thioether chain of Met80) create a strong ligand field. The Fe center is at low spin in both oxidation states (Ferrocychrome c and ferricytochrome c). The absorption band at 695 nm in ferricychrome c has been experimentally correlated 3+ to the presence of a Fe -SMet bond. This transition has been assigned both as a porphyrin 2 13 2 14 to Fe dZ transition and a SMet to Fe dZ charge transfer (CT) . While the origin of this absorption is still in dispute, the diminishing peak intensity of the 695 nm absorption band at elevated temperature in ferricytochrome c led to the proposal of a thermodynamic equilibrium between a “hot” and a “cold” form by Schejter and George.15 Further studies indicate that the 695 nm band exhibits a biphasic temperature dependent behavior. This was analyzed by Filosa and English, who proposed that the first phase is an equilibrium between a native form and a form with a change in the Met chain orientation, and the second phase involves the actual Met substitution by a stronger alkaline ligand.16 Recently, Elliott and coworkers 17 detected a low reduction potential form of cyt c

ascribing it to the loss of the Met ligand. It is generally thought the Fe-SMet is stronger in the reduced rather than the oxidized state, which is based on protein folding energy,18 electrochemistry studies17, 19 and theoretical calculations20. All these are related to the temperature sensitive transition at 695 nm and the binding strength of the axial ligand

SMet, but it is not clear on the Met-loss species as to weather the Met is replaced by alternative protein residue, water or the axial position remains vacant. Here, we will address the origin of the 695 nm absorption band and nature of the Fe-

SMet bond in ferricychrome c using density function theory (DFT) calculations calibrated with experimental results.

142

A

Tyr67

B Met80

Cys17

Cys14 His18

Pro30

Figure 6.1 The structures of cytochrome c models: A) 55-atom model; B) 149 atom model (numberings of protein residues are referred to horse heart cyt c 1HRC).

6.2 Calculation Details

DFT calculations were performed to investigate the Fe-SMet bond and the electronic structure of the met-loss form in ferricytochrome c. Our 55-atom model, Fe-P-Im-DMS, is shown in Figure 6.1A (P = porphyrin, Im = Imidazole, DMS = dimethyl sulfide). The Met-loss species was modeled with removal of the axial thioether (DMS). The fully

143 geometry optimized structures were used to evaluate enthalpic and entropic contributions to a thermodynamic equilibrium. To consider protein matrix effects on the cyt c site, a 149-atom model was used, which included two thioether linkages to the heme group, a H- bond to the imidazole Nδ atom of His18, and a H-bond to the thioether S atom of Met80 as shown in Figure 6.1B. The 149-atom model was partially geometry optimized with the α- and β-carbons of the protein residues fixed. Spin-unrestricted DFT calculations were performed using Gaussian 03 with tight SCF convergence criteria (10-8 au) and a mixed

triple-zeta/double zeta basis set (6-311G* on Fe and S, and 6-31G* on the other atoms). Hybrid functional BP86 with 10% Hartree Fock mixing was used for DFT calculations, and the pure functional BP86 was used for TDDFT calculations. Wave function stability calculations were performed to confirm that the calculated wave functions corresponded to the ground state.

6.3 Results and Analysis DFT calculations are used to investigate the change in the geometric and electronic structure upon loss of the axial Met ligand in ferricyochrome c. The Met contributions to the enthalpy and entropy in a thermodynamic process will be discussed. TDDFT calculations are used to simulate and assign the 695 nm absorption band.

6.3.1 DFT Calculations 6.3.1.1 Geometric Structure The ground state of the 55-atom model was found to be at low spin (S=1/2), consistent with experimental results.21-23 The fully optimized geometry of the low spin states well

reproduced the experimental results as shown in Table 6.1 (Note that the Fe-SMet distance in the 55-atom model is 0.08 Å longer than in EXAFS results24). The geometry

optimizations show a longer Fe-SMet distance in the higher spin state (Fe-SMet: 2.41 Å (S=1/2), 2.81 Å (S=3/2), 2.75 Å (S=5/2), respectively). To consider protein matrix effects on the metal site, 149-atom models described in section 6.2.3 were applied. The 149- atom model was also found to be more stable at low spin state. As shown in Table 6.1, the bond distances in both models are very similar. The axial ligand distances are slightly longer in the 149-atom model.

144 Upon removal of the axial ligand (Met80) for both 55-atom and 149 models, geometry optimizations give similar bond distances as shown as 5C and summarized in Table 6.1. Both models indicate that the lowest electronic energy state is at an intermediate spin state (S=3/2). Two crystal structures of cyt c’ (1E83 without Met ligand (S=3/2),25 and 26 2CCY with a very weak Fe and SMet interaction at 3.7 Å (S=3/2) ) are listed in Table 6.1 for comparison.

Table 6.1 Experimental and calculated bond distances in heme sites with or without a Met ligand.

Crystal EXAFS Crystal Structure (5C) Calc. (6C) Calc. (5C) Calc. (6C) Calc. (5C) spin bond a b e f e f Structure Results No Metc With Metd 55-atom 55-atom 149-atom 149-atom

Fe-SMet 2.43 2.33(2) - - 2.41 2.45

½ Fe-Neq 2.00 1.98(2) - - 2.00 1.98 2..01 1.98

Fe-NHis 2.01 2.01(3) - - 1.96 1.89 1.94 1.90

Fe-SMet - - - - 2.81 2.85

3/2 Fe-Neq - - 2.03 2.01 2.00 1.97 2.01 1.99

Fe-NHis - - 2.01 2.00 2.18 2.13 2.10 2.12

Fe-SMet - - - - 2.75 2.77

5/2 Fe-Neq - - - - 2.05 2.05 2.07 2.06

Fe-NHis - - - - 2.14 2.06 2.09 2.07 a) Horse heart ferricytochrome c, PDB ID:1HRC, resolution 1.94Å at 288K;12 b) EXAFS for horse heart ferricytochrome c at 4K;24 c) ferricytochrome c’ without Met ligand, 25 PDB:1E83, resolution 2.05Å at 100K; d) ferricytochrome c’ with a Fe-SMet distance at 3.7 Å. PDB: 2CCY 1.67Å at 288 K;26 e) 6C denotes that heme Fe has two axial ligands

(SMet and NHis); f) 5C denotes that axial ligands SMet is removed from the 55- or 149-atom model.

6.3.1.2 Energies Table 6.2 and 6.3 list the calculated energy differences among different spin states for the Met-bound form and Met-loss form. Here we focus on the comparison between the lowest energy states for the Met-bound form (S=1/2) and that for the Met-loss form (S = 3/2).

145 The calculated electronic energy for Fe-SMet bond dissociation in ferricytochorme c is 6.9 kcal/mol in the 55-atom model in which solvent effects were included (Table 6.2 in bold face. Note that this energy in the 149-atom model is 7.6 kcal/mol, which is larger than that in the 55-atom model). Upon loss of the Met ligand in the 55-atom model, the enthalpy change (ΔH) is 5.3 kcal/mol (Table 6.3, in bold) and the entropy change is 14.4 kcal/mol at 298 K (Table 6.3 in bold italic)(Equation 6.1). The calculated enthalpy change is smaller than the experimental results (ΔHexp = 9.5 kcal/mol16) indicating that protein effects should be considered. However, the calculated entropy gain is larger than experimentally observed (TΔS = 8.5 kcal/mol at 298 K16). This entropy contribution upon the loss of the Met ligand is clearly overestimated because the Met residue is still covalently linked to protein matrix as we discussed on the T1 Cu site of nitrite reductase (NiR) in Chapter 2 (Entropy: 4.5 kcal/mol (experiment) vs 11.7 kcal/mol (calculated) in NiR). Note that the spin state change has only a small contribution to the calculated entropy change (Table 6.3 last row). With the ratio of experiment value/calculated value

in NiR, the calculated entropy change upon the Fe-SMet bond dissociation in ferricytochrome c is corrected to be 5.5 kcal/mol at 298 K, which is close to the experimental results (8.5 kcal/mol at 298 K).

Table 6.2 Calculated electronic energies for the loss of the axial Met ligand.

Energy Spin FeIII (kcal/mol) 1/2 3/2 5/2

Emet on (6C) 0.0 5.0 17.0

55-atom Emet off (5C+Met) 17.1 11.6 24.0

(ΔE6c vs 5c) pcm 10.9 6.9 19.2

Emet on (6C) 0.0 6.4 13.1 149-atom Emet off (5C+Met) 11.5 7.6 15.9

6C denotes that Fe has two axial ligands (SMet and NHis), and 5C denotes that heme Fe has

only one axial ligand NHis.

146 Table 6.3 Calculated free energies for the loss of the axial Met ligand (T = 298K)

Energy Spin State of 5C FeIII (kcal/mol) 1/2 3/2 5/2 ΔH 15.7 10.1 21.6 Vacuum TΔS 13.0 14.4 15.2 ΔG 2.7 -4.3 6.4 ΔH 9.5 5.3 16.7 Solvent TΔS 13.0 14.4 15.2 correction ΔG -3.5 -9.1 1.5

Spin S (cal/mol⋅K) 1.377 2.755 3.561 Contribution TΔS(6CS=1/2→ 5C) 0 0.41 0.65

6C denotes that Fe has two axial ligands (SMet and NHis), and 5C denotes that Fe has only

one axial ligand (NHis).

6.3.2 TD-DFT The 695 nm (14,400 cm-1) band in ferricytochrome c is z polarized.13 Ligand competition studies show that the 695 nm band is related to Met binding, which is also supported by the fact that no peak was found in this energy region in the case of cytochromes with two axial His ligands. TDDFT calculations on the MetHis model (Fe-

P-Im-DMS) and the 2His model (Fe-P-(Im)2) were performed to simulate the absorption spectra and define the nature of the 695 nm band.

The molecular orbital descriptions in the notation appropriate to D4h symmetry for Fe- P-Im-DMS and Fe-P-Im-DMS are given in Figure 6.2 and Figure 6.3 respectively. Based on TDDFT calculations, low energy electronic transitions are related to the five Fe d

orbitals which split in D4h symmetry into four levels, the nearby porphyrin orbitals (2b1u

> 4eg, unoccupied; 3a2u > 1a1u > 2b2u> 2a2u, occupied), His π1, His π2, and Met S b1. Note that other orbitals are at much deeper or higher energy and do not contribute to NIR- Visible spectra. The calculated energies of d-d transitions in Fe-P-Im-DMS and Fe-P-Im-DMS are listed in Table 6.4. These transitions are parity forbidden, therefore very weak in the absorption spectra. There are interesting features in the calculated region of 12,000 – 20,000 cm-1 as show

147 in Figures 6.2 for Fe-P-Im-DMS) and 6.3 for Fe-P-(Im)2), and Tables 6.5 for Fe-P-Im-

DMS) and 6.6 for Fe-P-(Im)2). In both Fe-P-Im-DMS and Fe-P-(Im)2, both the 2b2u →

dyz and 2a2u → dyz transitions have dominant y-polarization. However, the 2a2u → dyz transition in Fe-P-Im-DMS has a significant z-polarization component associated with the

S b1 (pz) mixing into the donor orbital. The intensity of the His π1, 2 → dyz transition in

Fe-P-(Im)2 is ~ twice of the Fe-P-Im-DMS model. 2 The 3a2u → dz transitions in both models are z-polarized. The energy of the intensity- weighted average of the α and β transitions is 2,600 cm-1 lower in Fe-P-Im-DMS -1 -1 (16200cm ) than that in Fe-P-(Im)2 (18800 cm ). The former is also more intense. Based on this energy shift and polarization, we are, at present, tempted to assign the -1 2 2 experimental 695 nm (14,400 cm ) band as the 3a2u → dz (porphyrin to dz ) and 2a2u → dyz (porphyrin to dyz) transitions both having z-polarization (Note that the models used here did not include the propionates and other charged group around the heme site in the protein. This will lead to some shifts in the calculated energy). The acceptor orbital in the former transition has significant S p (13%) along with dominant Fe dz2 (61%) while the donor orbital in the latter has significant S p (16%) character along with 2 dominant porphyrin character (80%). Note that the energy of S b1 (51% SMet) → dz transition (20,500 cm-1) is too high to be assigned as the 695 nm band in the absorption spectrum.

Table 6.4 Calculated d-d transitions for in Fe-P-Im-DMS and Fe-P-Im2

Fe-P-Im-DMS Fe-P-Im2

Energy/cm-1 No. Energy/cm-1 No. dxz → dyz 1,000 (β) 1 1,300 (β) 1 dxy → dyz 2,200 (β) 2 2,500 (β) 2 dxz → dz2 11,000(α+β) 5 12,800 (α+β) 5 dxy → dz2 12,000 (α+β) 6 14,000 (α+β) 7 dyz → dz2 19,500 (α) 31 22,000 (α) 37 17,600 (α+β) 25 17,200 (α+β) 22 dxz → dx2-y2 24,900 (α+β) 47 24,900 (α+β) 45 17,000 (α+β) 21 17,100 (α+β) 19 dxy → dx2-y2 24,400 (α+β) 45 23,300 (α+β) 41 dyz → dx2-y2 26,300 (α) 59 26,400 (α) 56

148

Table 6.5 Calculated dominant transitions in the region of 12,000 – 20, 000 cm-1 for Fe- P-Im-DMS f Energies polarization -1 No. cm X Y Z

2b2u → dyz (β) 13,200 0.0017 7 0.0106 0.1936 0.0663

2a2u → dyz (β) 14,400 0.0051 9 0.0587 0.3080 0.1237 His π →dyz (β) 15,900 0.0010 15 0.0121 0.0966 0.1022 2 3a2u → dz (α) 14,900 12 0.0172 0.0676 0.2132 2 0.0049 3a2u → dz (β) 17,400 24 0.0162 0.0253 0.2206

Table 6.6 Calculated dominant transitions in the region of 12,000 – 20, 000 cm-1 for Fe-

P-(Im)2

Energies polarization -1 f No cm X Y Z

2b2u → dyz (β) 13,200 0.0025 6 0.1095 0.2274 0.0082

2a2u → dyz (β) 15,300 0.0049 13 -0.0626 0.3162 0.0262 His π →dyz (β) 15,000 0.0026 11 0.0053 0.0363 -0.2451 His π → dyz (β) 15,600 0.0012 14 -0.0456 0.0836 0.1256

2 3a2u → dz (α) 17,100 19 -0.0049 0.0010 0.1421 0.0031 2 3a2u → dz (β) 19,700 30 -0.0012 0.0044 -0.1813

149

Figure 6.2 Schematic of molecular orbital diagram of Fe-P-Im-DMS (P = porphyrin, Im = Imidazole, DMS = dimethyl sulfide). The table-inset give the symmetry correlations between porphyrin and Fe d orbital.27

150

Figure 6.3 Schematic of molecular orbital diagram of Fe-P-(Im)2 (P = porphyrin, Im = Imidazole).

151 50000 1000 A 40000 800 600 30000 400 9

20000 200 24 12 7

0 20000 18000 16000 14000 12000 10000 10000 Energy/cm-1

0 30000 24000 18000 12000 6000 Energy/cm-1 30000 B 1000 25000

20000

13 15000 11 6 30 10000 19 14

0 22000 20000 18000 16000 14000 12000 10000 -1 5000 Energy/cm

0 30000 24000 18000 12000 6000 Energy/cm-1 C

Figure 6.4 TD-DFT calculated absorption spectrum of the Cyt c model of A) FeIII with His and Met axial ligand; B) FeIII with bisHis axial ligand; C) experimental absorption spectrum of ferricytochrome c.16

152 6.4 Discussion 3+ DFT calculations indicate that the Fe -SMet bond in the Fe-P-Im-DMS (6C) model is 3+ enthalpically favored. The loss of Fe -SMet bond is entropically favored resulting in a 5C species with an intermediate spin state (S = 3/2) (Equation 6.1). Keq1 [His-Fe-Met]6C [His-Fe]5C + Met Equation 6.1

The calculated enthalpy and entropy contributions are consistent with the experimental data for the first phase of the bi-phasic thermodynamic process observed in horse heart ferricytochrome c.16 However, Raman studies of porphyrin high frequency vibrations indicate that there are only low spin species present in the first phase thermodynamic process, which is not consistent with a 5C species. Alternatively, the Met can be replaced by a stronger ligand B (lysine or histidine) in the first phase. A possible model is that at elevated temperatures the ligand B would be deprotonated (Equation 6.2) and could then bind to Fe (Equation 6.3). If this is the case, three thermodynamic equilibria would be coupled and result in a 6C’ species with a low spin state. A systematic DFT evaluation of this possibility combined with obtaining additional Raman data is underway.

K + eq2 + B:H B: + H Equation 6.2 K eq3 [His-Fe]5C + B: [His-Fe-B]6C’ Equation 6.3

TDDFT calculations support the assignment of the 695 nm band as porphyrin to dz2 and porphyrin to dyz transitions. There is also significant SMet character involved in both transitions which could explain the weak intensity and z polarization of 695 nm band. The two-transition assignment is consistent with the observation of a derivative-shape peak in magnetic circular dichroism spectra in 695 nm region.14 We have been able to obtain resonance Raman features enhanced by the 695 nm band which are very different from those observed upon excitation into the B-band. Frequency calculations calibrated with nuclear resonance vibrational spectroscopy (NRVS) data will be used to assign the resonance Raman features and thus the 695 nm band.

153 In a series of cyts c, including horse heart cyt c, cyt c552 from Hydrogenobacter

thermophilus, cyt c551 form Pseudomina aeruginosa, different temperature-dependent decreases of their 695 nm absorption bands are observed16, 17, which likely reflect the protein’s role in tuning the Met binding. The dependence of the thermodynamic 3+ parameters of the Fe -Smet bond on the type of cyts c should, in parallel with our studies on green and blue Cu proteins, elucidate the role of the protein in stabilizing this bond and the nature of the activation of cyt c for enzymatic catalysis to initiate apoptosis.

6.5 References 1. Liu, X.; Kim, C. N.; Yang, J.; Jemmerson, R.; Wang, X., Induction of apoptotic program in cell-free extracts: requirement for dATP and cytochrome c. Cell 1996, 86, 147-157. 2. Ow, Y.-L. P.; Green, D. R.; Hao, Z.; Mak, T. W., Cytochrome c: functions beyond respiration. Nat. Rev. Mol. Cell Biol. 2008, 9, 532-542. 3. Bertini, I.; Cavallaro, G.; Rosato, A., Cytochrome c: Occurrence and Functions. Chem. Rev. 2006, 106, (1), 90-115. 4. Takano, T.; Dickerson, R. E., Conformational change of cytochrome c: I. Ferrocytochrome c structure refined at 1.5 Å resolution. J. Mol. Biol. 1981, 153, 79-94. 5. Takano, T.; Dickerson, R. E., Conformational change in cytochrome c: II. Ferricytochrome c refinement at 1.8 Å resolution and comparison with the ferrocytochrome structure. J. Mol. Biol. 1981, 153, 95-115. 6. Berghuis, A. M.; Brayer, G. D., Oxidation state-dependent conformational changes in cytochrome c. J. Mol. Biol. 1992, 223, 959-976. 7. Feng, Y.; Roder, H.; Englander, S. W., Redoxdependent structure change and hyperfine nuclear magnetic resonance shifts in cytochrome c. Biochemistry 1990, 29, 3494-3504. 8. Gochin, M.; Roder, H., Protein structure refinement based on paramagnetic NMR shifts: applications to wild-type and mutant forms of cytochrome c. Protein Sci. 1995, 4, 296-305.

154 9. Qi, P. X.; Beckman, R. A.; Wand, A. J., Solution structure of horse heart ferricytochrome c and detection of redox-related structural changes by highresolution 1H NMR. Biochemistry 1996, 35, 12275-12286. 10. Banci, L.; Bertini, I.; Huber, J. G.; Spyroulias, G. A.; Turano, P., Solution structure of reduced horse heart cytochrome c. J. Biol. Inorg. Chem. 1999, 4, 21- 31. 11. Louie, G. V.; Brayer, G. D., High-resolution refinement of yeast iso-1- cytochrome c and comparisons with other eukaryotic cytochromes c. J. Mol. Biol. 1990, 214, (2), 527. 12. Bushnell, G. W.; Louie, G. V.; Brayer, G. D., High-resolution three-dimensional structure of horse heart cytochrome c. J. Mol. Biol. 1990, 214, (2), 585. 13. Eaton, W. A.; Hochstrasser, R. M., Electronic Spectrum of Single Crystals of Ferricytochrome-c. J. Phys. Chem. 1967, 46, (7), 2533-2539. 14. Gadsby, P. M. A.; Thomson, A. J., Assignment of the axial ligands of ferric ion in low-spin hemoproteins by near-infrared magnetic circular dichroism and electron paramagnetic resonance spectroscopy. J. Am. Chem. Soc. 1990, 112, (13), 5003- 5011. 15. Schejter, A.; George, P., The 695-mμ Band of Ferricytochrome c and Its Relationship to Protein Conformation. Biochemistry 1964, 3, (8), 1045-1049. 16. Filosa, A.; English, A. M., Probing local thermal stabilities of bovine, horse, and tuna ferricytochromes c at pH 7. J. Biol. Inorg. Chem. 2000, 5, (4), 448-454. 17. Ye, T.; Kaur, R.; Senguen, F. T.; Michel, L. V.; Bren, K. L.; Elliott, S. J., Methionine Ligand Lability of Type I Cytochromes c: Detection of Ligand Loss Using Protein Film Voltammetry. J. Am. Chem. Soc. 2008, 130, (21), 6682-6683. 18. Pascher, T.; Chesick, J. P.; Winkler, J. R.; Gray, H. B., Protein folding triggered by electron transfer. Science 1996, 271, 1558-1560. 19. Raphael, A. L.; Gray, H. B., Axial Ligand Replacement in Horse Heart Cytochrome c by Semisynthesis. Proteins: Structure, Function, and Genetics 1989, 6, (3), 338-340. 20. Rovira, C.; Carloni, P.; Parrinello, M., The Iron-Sulfur Bond in Cytochrome c. J. phys. Chem. B 1999, 103, 7031-7035.

155 21. Cheesman, M. R.; Greenwood, C.; Thomson, A. J., Magnetic Circular Dlchrolsm of Hemoproteins. Advances In Inorganic Chemistry 1991, 36, 201-255. 22. Smith, D. W.; Williams, R. J. P., The Spectra of Ferric Haems and Haemoproteins. Structure and Bonding 1969, 7, 1-45. 23. Scheidt, W. R.; Reed, C. A., Spin-state/stereochemical relationships in iron porphyrins: implications for the hemoproteins. Chem. Rev. 1981, 81, (6), 543-555. 24. Cheng, M.-C.; Rich, A. M.; Armstrong, R. S.; Ellis, P. J.; Lay, P. A., Determination of Iron Ligand Bond Lengths in Ferric and Ferrous Horse Heart Cytochrome c Using Multiple-Scattering Analyses of XAFS Data. Inorg. Chem. 1999, 38, (25), 5703-5708. 25. Lawson, D. M.; Stevenson, C. E. M.; Andrew, C. R.; Eady, R. R., Unprecedented proximal binding of nitric oxide to heme: implications for guanylate cyclase. Embo J. 2000, 19, 5661-5671. 26. Finzel, B. C.; Weber, P. C.; Hardman, K. D.; Salemme, F. R., Structure of ferricytochrome c' from Rhodospirillum molischianum at 1.67 A resolution. J. Mol. Biol. 1985, 186, 627-643. 27. Cheng, R.-J.; Chen, P.-Y.; Lovell, T.; Liu, T.; Noodleman, L.; Case, D. A., Symmetry and Bonding in Metalloporphyrins. A Modern Implementation for the Bonding Analyses of Five- and Six-Coordinate High-Spin Iron(III)−Porphyrin Complexes through Density Functional Calculation and NMR Spectroscopy. J. Am. Chem. Soc 2003, 125, 6774-6783.

156

Appendix

Oxygen Binding of Water-Soluble Cobalt Porphyrins in Aqueous Solution*

* Reproduced with permission from Inorg. Chem. 2005, 44, 9628-9630. Copyright © 2005 American Chemical Society.

157 A.1 Introduction Hemoglobin (Hb) and myoglobin (Mb) are responsible for the storage and transport of molecular oxygen in biological systems. Using synthetic porphyrin models to mimic the 1-3 functions of hemoprotein has been extensively explored. Most reported synthetic O2 binding systems have been studied in aprotic organic solvents such as toluene, benzene, 1-3 + - and dichloromethane, because traces of water (H2O, H , OH ) can result in an irreversible decomposition of such dioxygen adducts in seconds (equation 1-3).3, 4 In

recent years, water-soluble O2 carriers have been sought because of proposed biomedical applications such as artificial red cells, photodynamic therapy (PDT) for tumor treatment, organ preservation, and DNA cleavage.5, 6 In contrast to the large literature describing dioxygen carrying systems in aprotic organic solvents,1-3 only a few examples of aqueous dioxygen-carrying systems have been reported.7-11 A major challenge in developing such aqueous systems is to build a relatively hydrophobic metalloporphyrin center within a water-soluble model system in order to prolong the lifetime of the oxygen adducts. Natural hemoglobin and myoglobin create such a structure by a protein matrix around the heme center. The globin moiety provides a relatively hydrophobic micro environment above each heme center thus inhibiting water-promoted autoxidation. The isolation of heme centers in protein pockets also prevents the formation of μ-oxo dimers, which result from another important autoxidation path of synthetic oxygen carriers (equation 4-5).3, 4

LFeII(O ) H O III O _ (eq. 1) O2 2 + 2 LFe (H2O) + 2

II II + III LFe (O2) Fe(II) LFe (O2) + H + H2O LFe (H2O) + O2H (eq. 2)

L II − III − O _ (eq. 3) LFe (O2) + OH LFe (OΗ ) + 2

II II III III IV (eq. 4) LFe (O2)LFe+ LFe -O2-Fe L 2LFe =O

IV II 2LFe =O +LFe LFeIII-O-FeIIIL (eq. 5)

Among the rare dioxygen-carrying models stable in aqueous media, Tsuchida et al. have used picket-fence porphyrins containing zwitter ionic phospholipids substituents.7, 8 Such models are not truly soluble in water, but form micelles dispersed in water. Other examples reported by Groves,9 Kano,10 and their co-workers employed a cyclodextrin- associated axial ligand to provide a hydrophobic environment encapsulating the iron(II) porphyrin center.

158 Cobalt porphyrins have also been investigated as dioxygen carriers.12-16 The cobalt analogs of hemoglobin and myoglobin are referred as coboglobins (CoHb, CoMb).17 II Their dioxygen binding complexes (LCo (O2)) are known to have greater kinetic stability toward decomposition compared with their heme counterparts. To the best of our knowledge, all oxygen binding cobalt porphyrins have been studied in organic solvents except that an aqueous oxygen-carrying cobalt porphyrin system was briefly mentioned in the literature.18 Thus we planned to utilize the attenuated reactivity of cobalt porphyrins to develop aqueous dioxygen-carrying systems. Herein we report the first water-soluble cobalt porphyrin that binds dioxygen in aqueous media.

A.2 Experiments Reagents and solvents were used as received unless otherwise indicated. Metallation

reactions of porphyrin compounds were performed under N2 atmosphere. All air sensitive samples for UV-Vis, EPR and oxygenation measurements were prepared in a Vacuum Atmospheres glovebox (VAC) under inert atmosphere. Solvents and buffer solution for air-sensitive experiments were deoxygenated before use.

Preparation, characterization of compound 1Co, 1H2, and ligand 2 are given in the supporting information section.

A.3 Results and Analysis The water soluble cobalt porphyrin 1Co was prepared from αααα-TAPP (Figure A.1) (see supporting information for details). The formation of a cage by the four carbamoylbenzyl quaternary amine groups is consistent with the symmetry manifest by

the NMR spectra of the demetalated porphyrin 1H2 and its precursor. Whereas the quaternary amine groups result in water solubility of 1Co, the four bulky carbamoylbenzyl groups over the porphyrin ring are expected to establish a relatively congested and hydrophobic micro environment over the metalloporphyrin center. This design is in accord with reports of native human hemoglobin mutants19: introduction of bulky or aromatic residues around heme centers markedly decreases the accessibility of this distal pocket to water, and should inhibit the autoxidation of the dioxygen complex. A water-soluble imidazole 2 was synthesized. Chelation of the imidazole 2 to 1Co should

159 mimic the proximal histidine residue in native hemoprotein.1-3 The triethylene glycol moiety in the ligand 2 not only results in high water solubility, but is also expected, because of the 2-methyl group, to provide steric hindrance preventing ligand 2 from fitting into the distal pocket of the porphyrin model 1Co, thus leaving the sixth coordination site vacant for dioxygen binding. It seems reasonable that the four positively charged carbamoylbenzyl quaternary amine groups at the porphyrin periphery should prevent the μ−oxo dimerization by electrostatic and steric repulsion (Figure A.2).

N N 4Cl

N N O O N NH HN O Me O N O O N 3 HN NNM NH O N 2

(αααα form) 1 (1H2: M=2H; 1Co: M=Co; 1Fe: M=Fe) Figure A1. Design of model compounds

N N 4Cl

N N O O NH HN O O2 1Co + 2 O N in water HN NNCo NH in water N

N Me O N O 3 2-1Co O

N N N 4Cl N 4Cl

N N N N O O O O HN NH HN NH O O O2 O O O N O N III HN NNCo NH HN NNCo NH N N

N N Me O N Me O N O 3 2-1Co(O2) O 3 O O Figure A2. Dioxygen binding of 1Co in the presence of ligand 2.

160

1.4 420 nm 1.2 (a) 434 nm (c) 1 422 nm (b) 0.8

A 0.6 0.4 0.2 0 -0.2320 370 420 470 520 570 620 Wavelength (nm)

Figure A3. UV-vis spectra of 1Co in pH 7.0 phosphate buffer solution (C = ~ 5×106 M):

(a) 1Co under N2; (b) 1Co + ligand 2 (~100 equivalents) under N2; (c) 1Co + ligand 2

(~100 equivalents) under O2.

Examination of dioxygen binding of 1Co was carried out in a pH 7.0 phosphate buffer solution at room temperature (22oC). The concentration of 1Co was adjusted to maintain a maximum absorption (A) at around 1.0 ~1.2 in the Soret band (Figure A.3). Under a nitrogen atmosphere, the 1Co solution has a Soret peak at 420 nm (Figure A.3, trace a). Upon addition of ligand 2 (~100 equivalents of 1Co), the Soret peak shifts slightly to 422 nm, and the absorption intensity decreases (Figure A.3, trace b). This indicates the formation of a five-coordinate adduct 2-1Co and is consistent with reports of five- coordinate cobalt porphyrins in organic solvents.14, 15 In the presence of dioxygen, the five-coordinate complex immediately reacts, leading to a 12 nm shift of the Soret peak from 422 nm to 434 nm (Figure 3, trace c), which suggests the formation of the six- 14, 15 coordinate O2 complex (2-1Co (O2)). A good isosbestic point (428 nm) was observed in the UV-vis absorption spectra in the dioxygen titration process (Figure A.4). dioxygen binding of the water soluble iron porphyrin 1Fe was also investigated in the presence of ligand 2 in an aqueous solution, but the oxygenated species was found to be too transient to analyze.7-11

161

1.4

1.2 434 nm 422 nm 1

0.8 oxygenation

A 0.6 0.4

0.2

0 320 370 420 470 520 -0.2 Wavelength (nm)

Figure A4. UV-vis absorption spectra of the oxygen titration course (changing from 2-

1Co to 2-1Co(O2)).

EPR spectroscopy is a definitive technique to examine oxygen binding of Co(II) porphyrins.6, 17, 19 The single electron in Co(II) (d7) is transferred to dioxygen forming •– Co(III)(O2 ) adduct upon oxygenation (Figure A.2 and 5). Introduction of an excess of the imidazole ligand 2 to an aqueous solution of 1Co (pH 7.0) led to the formation of the 2 five-coordinate adduct 2-1Co(II) (dz ground state; gll = 2.02, g┴ = 2.31). Due to the interaction between 59Co (I = 7/2) and 14N (I = 1), the expected octet hyperfine structure 59 with a triplet super-hyperfine coupling pattern was observed (Figure 5, trace a; All ( Co) 14 = ~86 G, All ( N) = ~16 G). This observation rules out coordination of a second base even when ligand 2 was present in large excess When the five-coordinate adduct 2-

1Co(II) is exposed to O2, a dramatic change in the EPR spectrum was observed (gll =

2.08, g┴ = 2.01; All = 17 G, A┴ = 11 G), indicating the formation of the six-coordinate •– complex 2-1Co(III) (O2 ) (Figure 5, trace b). The EPR signals of the oxygenated adduct lost intensity over time. After 1~2 hours, •– only a weak EPR signal of 2-1Co(III) (O2 ) was detected. This observation suggests •– decomposition of the oxygenation adduct 2-1Co(III) (O2 ) to an EPR inactive Co(III)-X   2 species (X denotes OH, H2PO4, or HPO4 depending on pH of buffered solution). The impact of pH on the kinetic stability of the oxygenation adduct was examined. A very •– weak EPR signal of 2-1Co(III) (O2 ) was detected at pH 5.0; in contrast, the oxygenated adduct was shown to be present in a pH 9.0 solution based on the EPR spectrum

162 (supporting information section A.5.8). This result implies that more acidic conditions + •– • 4 (H ) accelerate the decomposition of cobalt superoxide (O2 to HOO /H2O2).

a b

3000 3100 3200 3300

a

2600 2800 3000 3200 3400 3600 3800

Field (G) Figure A5. EPR Spectra of Complex 2-1Co under pH 7.0 buffer solution at 77K (1Co: C

> 1mM, ligand 2: 5~10 equivalents): (a) under N2; (b) after oxygenation

In summary, using cobalt porphyrin 1Co and imidazole ligand 2, we have developed the first cobalt porphyrin system that carries dioxygen in aqueous media. This metastable system demonstrates a new strategy for developing aqueous dioxygen carriers, and also provides an opportunity to probe the mechanisms related to the decomposition of hemoprotein dioxygen adducts in aqueous media, as most reported models for the study of structure-function correlations can only be studied in organic solvents.1-3

A.4 Acknowledgement

This material is based upon work supported by the NSF under Grant No. CHE-013206. Support form NIH Grant No. 017880 and No. DK31450 (E.I.S) also contributed to this work.

163 A.5 Supporting Information A.5.1 Preparation of compound 2

(i) (CH3O)2SO2, NaHCO3; (ii) (a) 2.0 N NaOH, (b) HCl; (iii) Ac2O, 95 °C; (iv) (a)

SOCl2, CH3CN, reflux, (b) triethylene glycol monomethyl ether, CH3CN, reflux.

1-methyl-5-imidazolecarboxylic acid was prepared following the literature.20 To a

solution of 1-methyl-5-imidazolecarboxylic acid (400 mg, 3.17 mmol) in CH3CN (10 mL) was added thionyl chloride (3.0 mL, 30 mmol). The mixture was refluxed for 3 h. Solvent and excess SOCl2 were evaporated under vacuum. The residue was redissolved in dry

CH3CN (10 mL). Triethylene glycol monomethyl ether (0.7 mL, 4.14 mmol) was added followed by triethylamine (1.0 mL, 7.17 mmol). The mixture was refluxed for 4 h. After concentration, the residue was purified by chromatography on silica gel with eluent 1 CH3OH/CH2Cl2 = 5/95 to give compound 2 (540 mg, 63%) as a light yellow oil. H

NMR (500 MHz, CDCl3): δ 7.70 (s, 1H), 7.59 (s, 1H), 4.36 (t, J = 5.0 Hz, 2 H), 3.86 (s, 3H), 3.74 (t, J = 4.5 Hz, 2H), 3.58 – 3.65 (m, 6 H), 3.47 (t, J = 4.5 Hz, 2H), 3.32 (s, 3H); 13 C NMR (CDCl3): 160.12, 142.37, 137.36, 123.04, 71.88, 70.65, 70.60, 70.56, 69.03, + 63.58, 59.03, 34.21; MS (ESI): m/e = 272.8 [M+H] for C12H21N2O5. HRMS (ESI) for + C12H21N2O5 [M+H] : calcd. 273.1450, found 273.1461.

164 A.5.2 Preparation of compound 1Co Porphyrin compound 1Co was prepared according following scheme.

A.5.3 Preparation of compound 3H2 αααα-TAPP was prepared following the literature.21, 22 αααα-TAPP (675 mg, 1.0 mmol)

was dissolved in dry CH2Cl2 (100 mL). To this solution was added 4-

Chloromethylbenzyl chloride (1.51 g, 8.0 mmol) followed by Et3N (1.12 mL, 8.0 mmol). The mixture was stirred at room temperature for 1 h. The reaction mixture was washed

with diluted NaHCO3 solution and dried over MgSO4. After concentration, the residue

was purified by chromatography on silica gel with eluent EtOAc/CH2Cl2=8/92 to give 1 compound 3H2 (810 mg, 63%) as a solid. H NMR (500 MHz, CDCl3): δ 8.96 (s, 8H), 8.80 (d, J = 8.0 Hz, 4H), 8.00 (s, 4H), 7.88 (t, J = 8.0 Hz, 4H), 7.80 (d, J = 7.5 Hz, 4H), 7.51 (t, J = 7.5 Hz, 4H), 6.86 (d, J = 8.0 Hz, 8H), 6.54 (d, J = 8Hz, 8H), 3.92 (s, 8H),

−2.63 (s, 2H); 13C NMR (CDCl3):165.41, 141.01, 138.12, 136.36, 134.37, 131.65, 130.39, 128.63, 127.18, 123.93, 122.15, 115.71, 45.05; MS (ESI): m/e = 1282.2 (M+) for

165 + C76H54Cl4N8O4. HRMS (ESI) for C76H55Cl4N8O4 [M+H] : calcd. 1283.3100, found

1283.3102; UV-Vis (CH2Cl2): 422, 514 nm.

A.5.4 Preparation of compound 3H2

Compound 3H2 (180 mg, 0.14 mmol) was stirred in a methylamine solution in ethyl

alcohol (33%, 15 mL) overnight at room temperature under N2 atmosphere. The mixture was concentrated under vacuum. The residue was recrystallized from CH3OH/Acetone 1 solvent to give compound 1H2 as a solid (173 mg, 83%). H NMR (500 MHz, DMSO-d6): δ 10.10 (s, 4H), 8.82 (s, 8H), 7.88 (d, J = 7.5 Hz, 4H), 7.84 (t, J = 7.5 Hz, 4H), 7.73 (d, J = 7.5 Hz, 4H), 7.60 (t, J = 7.5 Hz, 4H), 7.17 (d, J = 7.5 Hz, 8H), 7.10 (d, J = 7.5 Hz, 8H), 4.59 (s, 8H), 2.89 (s, 36H, NMe3), −2.99 (s, 2H); 13C NMR (CD3OD): 167.27, 138.08, 136.94, 136.50, 132.79, 130.85, 129.50, 127.76, 126.44, 125.43, 116.64, 67.85, 52.11; + MS (ESI): m/e = 1521.8 (M ) for C88H90Cl4N12O4; HRMS (ESI) for C88H90Cl3N12O4 + [M−Cl] : calcd. 1483.6274, found 1483.6263; UV-Vis (CH3OH): 422, 520 nm.

A.5.5 Preparation of compound 1Co

The reaction was carried out under a N2 atmosphere. To a solution of 1H2 (150 mg, 0.1

mmol) in dry CH2Cl2/CH3OH (v/v 5:1, 30 mL) was added CoCl2 (65 mg, 0.5 mmol). The

mixture was gently refluxed overnight under N2. Completion of the reaction was checked by UV-Vis spectra of an aliquot sample of the reaction mixture. After filtration, the filtrate was concentrated. The residue was dried under high vacuum to give product 1Co + as a solid in quantitative yield. MS (ESI): m/e = 1575 [M] for C88H88Cl4CoN12O4; + HRMS (ESI) for C88H88Cl3CoN12O4 [M−Cl] : calcd. 1540.5449, found 1540.5419; UV-

Vis (CH3OH): 420, 518 nm. The product mixture was contaminated with a little

unreacted CoCl2. As 1Co is very polar, the product could not be purified by chromatography on silica gel.

A.5.6 Preparation of compound 3Co (method B)

The reaction was carried out under N2. To a solution of 3H2 (150 mg, 0.12 mmol) in

dry THF (20 mL) was added CoCl2 (150 mg, 1.2 mmol) followed by 2,6-lutidine (350 μL, 2.6 mmol). The mixture was stirred at room temperature for 8 h. Completion of the

166 reaction was checked by UV-Vis spectra of an aliquot sample of the reaction mixture. After removal of solvent, the residue was dissolved in benzene (60 mL). The solution was washed with dilute dammonia solution, dried over MgSO4, and evaporated to dryness. The residue was purified by chromatography on silica gel with eluent

EtOAc/CH2Cl2=8/92 to give compound 3Co (125 mg, 82%) as a solid. MS (ESI): m/e = + + 1339.6 [M] for C76H52Cl4CoN8O4; HRMS for C76H52Cl4CoN8O4 [M] calcd. 1339.2198,

found 1339.2186; UV-Vis (CH2Cl2): 432, 548 nm.

A.5.7 Preparation of compound 1Co from 3Co (method B) Compound 3Co (140 mg, 0.10 mmol) was stirred in a methylamine solution in ethyl alcohol (33%, 15 mL) overnight at room temperature under N2. Excess methylamine and solvent were evaporated off under vacuum. The residue was dried under high vacuum to give 1Co in quantitative yield.

A.5.8 X-band (9.37GHz) EPR Spectra of 1Co Buffer Solutions (C > 1mM) Under Different Conditions (recorded at 4mW and 77K)

Figure A.S1 In the absence of ligand 2, at pH 7.0, under N2, g|| = 2.02, A|| = 110G, g⊥ is not resovled.

167

Figure A.S2 (a) In the presence of ligand 2, at pH 7.0, under N2, g|| = 2.02, A|| = 86G, AN

= 16G, g⊥ = 2.31; (b) In the presence of ligand 2, at pH 7.0, under O2, g|| = 2.08, A|| =

17G, g⊥ = 2.01, A⊥ = 11G. Spin quantitation shows a:b ~100:66 (intensity lost over time).

168

Figure A.S3 (a) In the presence of ligand 2, at pH 5.0, under N2, g|| = 2.02, A|| = 85G, AN

= 15G, g⊥ = 2.31; (b) In the presence of ligand 2, at pH 5.0, under O2 (oxygenation adduct decomposed). Spin quantitation shows a:b ~100:0.7.

169

Figure A.S4 (a) In the presence of ligand 2, at pH 9.0, under N2, g|| = 2.02, A|| = 83G, AN

= 15G, g⊥ = 2.31; (b) In the presence of ligand 2, at pH 9.0, under O2, g|| = 2.08, A|| =

17G, g⊥ = 2.01, A⊥ = 11G. Spin quantitation shows a:b ~100:4.

A.6 References 1. Collman, J. P.; Boulatov, R.; Sunderland, C. J.; Fu, L., Chem. Rev. 2004, 104, 61. 2. Collman, J. P.; Fu, L., Acc. Chem. Res. 1999, 32, 455.

170 3. Momenteau, M.; Reed, C. A., Chem. Rev. 1994, 94, 659. 4. Shikama, K., Chem. Rev. 1998, 98, 1357. 5. Kobayashi, K.; Tsuchida, E., Artificial Oxygen Carrier. Its Front Line. Springer- Verlag: Tokyo, 2005. 6. Kadish, K. M.; Smith, K. M.; Guilard, R., The Porphyrin Handbook. Academic Press: San Diego, 2000; Vol. 6&14. 7. Komatsu, T.; Moritake, M.; Nakagawa, A.; Tsuchida, E., Chem. Eur. J. 2002, 8, 5469. 8. Komatsu, T.; Hayakawa, S.; Yanagimoto, T.; Kobayakawa, M.; Nakagawa, A.; Tsuchida, E., Bull. Chem. Soc. Jpn. 2001, 74, 1703. 9. Zhou, H.; Groves, J. T., Biophysical Chem. 2003, 105, 639. 10. Kano, K.; Kitagishi, H.; Kodera, M.; Hirota, S., Angew. Chem. Int. Ed. 2005, 435. 11. Komatsu, T.; Hayakawa, S.; Tsuchida, E.; Hishide, H., Chem. Comm. 2003, 50. 12. Collman, J. P.; Zhang, X.; Wong, K.; Brauman, J. I., J. Am. Chem. Soc. 1994, 116, 6245. 13. Collman, J. P. B., J. I.; Doxsee, K. M.; Halbert, T. R.; Hayes, S. E.; Suslick, K. S., J. Am. Chem. Soc. 1978, 100, 2761. 14. Stynes, D. V. S., H. C.; Hames, B. R.; Ibers, J. A., J. Am. Chem. Soc. 1973, 95, 1796. 15. Stynes, D. V. S., H. C.; Ibers, J. A.; Hames, B. R., J. Am. Chem. Soc. 1973, 95, 1142. 16. Proniewicz, L. M. K., J. R., J. Am. Chem. Soc. 1990, 112, 675. 17. Jones, R. D. S., D. A.; Basolo, F., Chem. Rev. 1979, 79, 139. 18. Fiammengo, R. W., K.; Crego-Calama, M.; Timmerman, P.; Figoli, A.; Wessling, M.; Reinhoudt, D. N., Org. Lett. 2003, 5, 3367. 19. Mathews, A. J. R., R. J.; Olson, J. S.; Tame, J.; Renaud, J.-P.; Nagai, K., J. Biol. Chem. 1989, 264, 16573. 20. O’Connell, J. F. P., J.; Yelle, W. E.; Wang, W.; Rapoport, H., Synthesis 1988, 767. 21. Collman, J. P. G., R. R.; Reed, C.; Halbert, T. R.; Lang, G.; Robinson, W. T., J. Am. Chem. Soc. 1975, 97, 1427. 22. Lindsey, J., J. Org. Chem. 1980, 5215.

171