1 Dynamics and Mechanism of Short-Range Electron Transfer

Dynamics and Mechanism of Short-Range Electron Transfer Reactions in

Flavoproteins

Dissertation

Presented in Partial Fulfillment of the Requirements for the Degree Doctor of Philosophy

in the Graduate School of The Ohio State University

Mainak Kundu, M.Sc.

Graduate Program in Chemistry

The Ohio State University

2019

Dissertation Committee

Professor Dongping Zhong, Advisor

Professor Heather Allen

Professor Sherwin Singer

Copyrighted by

Mainak Kundu

2019

Abstract

Short-range electron transfer (ET) reactions are extensively used in light perception processes in flavoproteins. Such reactions an ultrafast and occur on the similar timescales as local protein-solvent fluctuations from femtoseconds to picoseconds and thus the two dynamics are expected to be coupled. Using the model protein of flavodoxin in its semiquinone state, we systematically characterized the photoinduced redox cycle of charge separation and charge recombination with mutations of different aromatic electron donors

(tryptophan and tyrosine) and local residues to change the redox properties. We observed the ET dynamics in a few picoseconds, strongly following a stretched behavior resulting from a coupling between local environment fluctuations and these ET processes. We further observed the hot vibrational-state formation through charge recombination and the subsequent cooling dynamics and both processes are also in a few picoseconds. Combined with our previous studies of femtosecond ET in oxidized flavodoxin, these results coherently reveal the evolution of the ET dynamics from single to stretched exponential behaviors and elucidate the coupling mechanism. The observed hot vibration-state formation is robust and general and should be considered in all photoinduced back ET processes in flavoproteins.

To analyze the role of tunneling distances in such reactions, we determined ET dynamics in a flavin–tryptophan pair at distances ranging from 3.2 to 18 Å, in 10

ii flavodoxin mutants with constant driving forces. We widely observed the stretched ET behavior with coupling from local solvation processes, and heterogenous ET dynamics from the large fluctuations (0.4 to 0.8 Å) in W donors located in flexible loop regions. We further observed the exponentially correlated distance–dependent ET severely impacted by local protein structures. At short distances, ET is highly favored by overlapping orbital interactions. At longer separation, tunneling pathways may uniquely differ within a protein and alter tunneling barriers to change the ET dynamics. Slow ET dynamics at long distances leads to energy loss via triplet dynamics, consistent with lower reaction yields.

These results reveal how donor–acceptor configurations control ET mechanisms and signal transduction in photoreceptors.

The accelerated growth in Oryza Sativa by the trimeric protein OsHAl3 is regulated by blue light through an unknown mechanism. We identified a photoreduction of the chromophore FMN from electron donors W79 and W82 on blue light illumination. At the active site, we obtained solvation dynamics in the ET-inert mutant in 1.9 ps (20%), 16 ps

(42%), and 480 ps (37%), with the large energy relaxation of 217 cm-1, thus revealing a water-exposed FMN environment. We identified ultrafast forward ET and back ET dynamics in 1.2 ps, β=0.92 and 0.6 ps, β=0.97 for the donor W79. For W82, we resolved a faster forward ET and a slower back ET reaction in 4.0 ps, β=0.96 and 0.41 ps, β=1.0, respectively. With faster formation and longer recombination dynamics, W82 is the main electron donor, and the charge-separated intermediate could be key to downstream signaling. The large β values indicate minor ET coupling with solvent motions due to the slower environmental relaxations. We also observed formation of vibrationally hot BET

iii product, consistent with the flavodoxin ET processes. These ET dynamics are essential for the blue light perception and is the primary event for signal transduction. For the special mutant C119S, we observed ET dynamics comparable to the wild-type, indicating absence of C119–C125 disulfide bond, in contrast to the light-inactive homolog AtHAL3. Together with the observation of C119-C125 disulfide bond in light structure of OsHAL3, the cysteines are possibly responsible for light induced structural changes.

Dedication

This document is dedicated to my friends and family.

Acknowledgments

I take this opportunity to express my gratitude to my advisor Dr. Dongping Zhong for his dependable support and patience throughout my graduate studies. I thank him for his leadership and all the discussions on planning experiments, data analysis and manuscript preparation at various stages of research. His consistent guidance and scientific insight have been instrumental in the successful completion of this dissertation.

I am grateful to Dr. Heather Allen and Dr. Sherwin Singer for their support and serving on my candidacy committee, annual progress reports, and the dissertation committee. I have also enjoyed learning the various physical chemistry courses they have taught.

I thank Ms. Lijuan Wang for her help with biochemistry experiments, particularly during OsHAL3 purification. I specially thank Dr. Ting-Fang He for her preceding contributions in studying the protein flavodoxin, that paved the way for my research in electron transfer. I thank Drs. Zheyun Liu, Yangzhong Qin, Meng Zhang and Xiankun Li for their help with experiments and scientific discussions. I appreciate discussions with Dr.

Xiaojing Yang from UIC, Chicago, and collaborations for the OsHAL3 project, and Dr.

Richard Swenson for the gift of Flavodoxin plasmid. I also thank National Science

Foundation for financial support and the Ohio Supercomputer Center for the computing facility.

I like to thank Drs. Pratik Sen, Tarasankar Pal, Anjali Pal, Samrajnee Dutta, and my family for their support and encouragement that inspired me to pursue graduate studies.

I am truly thankful to my friends, who have offered me help, advice and support on numerous occasions during my Ph.D. journey.

vii

Vita

2009 B.Sc. Chemistry. Presidency College, Calcutta University, India.

Awarded Certificate of Merit, for top-10 university rank.

2011 M.Sc. Chemistry. Indian Institute of Technology Kanpur, India.

Dissertation: Spectroscopic Investigation of Synergistic Chloroform-

Methanol Binary Mixture.

2012–present Graduate Teaching and Research Associate, Department of

Chemistry and Biochemistry, The Ohio State University.

Publications

4. Short-range electron transfer in reduced flavodoxin: Ultrafast nonequilibrium dynamics coupled

with protein fluctuations. M. Kundu, T-F. He, Y. Lu, L. Wang, and D. Zhong, J. Phys. Chem.

Lett. (2018).

3. Galvanic replacement of As(0) nanoparticles by Au(III) for nanogold fabrication and SERS

application. S. Saha, S. Maji, R. Sahoo, M. Kundu, A. Kundu, and A. Pal, New J. Chem. (2014).

2. Wet-chemical synthesis of spherical arsenic nanoparticles by a simple reduction method and its

characterization. S. Saha, S. Maji, M. Kundu, A. Kundu, and A. Pal, Adv. Mat. Lett. (2012).

viii

1. Origin of strong synergism in weakly perturbed binary solvent system: A case study of primary

alcohols and chlorinated methanes. S. Gupta, S. Rafiq, M. Kundu, and P. Sen, J. Phys. Chem.

B (2011).

Fields of Study

Major Field: Chemistry

Table of Contents

Abstract ...... ii Dedication ...... v Acknowledgments...... vi Vita ...... viii List of Tables ...... xii List of Figures ...... xiii Chapter 1. Introduction ...... 1 1.1 Photoreceptors and photoenzymes ...... 1 1.2 Flavoproteins...... 2 1.2.1 Flavodoxin and electron transfer processes ...... 2 1.2.2 The Hal3 gene and bioregulation ...... 4 1.4 Methodology: Femtosecond-resolved spectroscopy ...... 5 Chapter 2. Short-Range Electron Transfer in Reduced Flavodoxin: Ultrafast Nonequilibrium Dynamics Coupled with Protein Fluctuations...... 12 2.1. Introduction ...... 12 2.2. Materials and methods ...... 13 2.2.1. Protein purification ...... 13 2.2.2. Femtosecond–resolved spectroscopy ...... 13 2.3. Results and discussions ...... 14 2.3.1. Spectroscopic properties and probing strategies ...... 14 2.3.2. Forward electron transfer and stretched behavior ...... 16 2.3.3. Vibrationally-coupled charge recombination and ground-state cooling dynamics...... 17 2.3.4. Electron transfer rates and reorganization energies...... 21 Chapter 3. Distance–Dependent Electron Transfer in Oxidized Flavodoxin: Ultrafast Nonequilibrium Dynamics Coupled with Protein Fluctuations...... 39 x

3.1. Introduction ...... 39 3.2. Materials and methods ...... 40 3.2.1. Protein purification ...... 40 3.2.2. Femtosecond–resolved spectroscopy ...... 40 3.2.3 Molecular dynamics simulations...... 41 3.3. Results and discussions ...... 42 3.3.1. Spectroscopic properties and probing strategies ...... 42 3.3.2. Forward electron transfer and dynamic heterogeneity ...... 43 3.3.3. Back electron transfer and photoinduced redox cycle...... 46 3.3.4. Dynamic fluctuations and implications in electron transfer dynamics...... 48 3.4 Conclusion ...... 50 Chapter 4. Dynamics and Mechanism of Light–Induced Electron Transfer Reactions in OsHAL3...... 63 4.1. Introduction ...... 63 4.2. Materials and methods ...... 64 4.2.1. Protein purification ...... 64 4.2.2. Femtosecond–resolved spectroscopy ...... 65 4.3. Results and discussions ...... 65 4.3.1. OsHAL3: Spectroscopic properties and photoactivity ...... 65 4.3.2. Solvation dynamics and dynamic flavin environment ...... 68 4.3.3. Forward electron transfer and dynamic heterogeneity ...... 69 4.3.4. Vibrationally-coupled charge recombination and ground-state cooling dynamics...... 70 4.3.4. Summary ...... 73 Bibliography ...... 88 Appendix A. Additional information for characterization of ET dynamics in SQ flavodoxin ...... 96 Appendix B. Additional information for characterization of ET dynamics in OX flavodoxin ...... 107 Appendix C. Additional information for protein preparation and characterization of ET dynamics in OsHAL3 ...... 113

List of Tables

Table 2.1 Timescales of ET reactions and subsequent cooling dynamics in semiquinone flavodoxin...... 27 Table 2.2: Transient-absorption dynamics of W60F/Y98F in semiquinone flavodoxin. .28 Table 2.3 Energetics of cyclic electron-transfer reactions in semiquinone flavodoxin. ....29 Table 3.1: Timescales of distance–dependent ET reactions in flavodoxin obtained from the heterogeneity fitting...... 53 Table 3.2 Fitting parameters of distance–dependent ET reactions in flavodoxin obtained from the wobbling model...... 54 Table 4.1 Dynamics of photoinduced electron transfer cycle in OsHAL3...... 74 Table A.1 FET dynamics in semiquinone flavodoxin from the fluorescence upconversion detection...... 98

xii

List of Figures

Figure 1.1 Redox states of FMN and their absorption and emission spectra...... 6 Figure 1.2 Structure of water exposed flavodoxin active site...... 7 Figure 1.3 Photoinduced redox cycle of OX flavodoxin...... 8 Figure 1.4 Solvation dynamics of flavodoxin in three redox states...... 9 Figure 15 Phenotype and structural analysis of OsHAL3...... 10 Figure 1.6 Experimental design of ultrafast pump-probe spectroscope...... 11 Figure 2.1 Structural and Electron Transfer properties of flavodoxin...... 30 Figure 2.2 Steady-states spectra of semiquinone flavodoxin...... 31 Figure 2.3 Forward ET dynamics of WT flavodoxin mutants...... 32 Figure 2.4 Normalized fs-resolved absorption transients of the mutant Y98R...... 33 Figure 2.5 Normalized fs-resolved absorption transients of the mutant W60F...... 34 Figure 2.6 Normalized fs-resolved absorption transients with dual donors...... 35 Figure 2.7 Normalized fs-resolved absorption transients of flavodoxin mutants...... 36 Figure 2.8 Photoinduced redox cycle coupled with environmental relaxations...... 38 Figure 3.1 Structural and spectral properties of oxidized flavodoxin...... 55 Figure 3.2 Forward electron transfer reaction dynamics and FMN–W configurations. ....56 Figure 3.3 Normalized fs-resolved absorption transients of the mutant D95W...... 57 Figure 3.4 Normalized fs-resolved absorption transients of the mutant E66W...... 58 Figure 3.5 Normalized fs-resolved absorption transients of the mutant D62W...... 59 Figure 3.6 Normalized fs-resolved transients of various flavodoxin mutants...... 60 Figure 3.7: Nonequilibrium ET dynamics modelled with a wobbling W donor...... 61 Figure 3.8: MD snapshot displaying the dynamic processes in intraprotein ET...... 62 Figure 4.1 Structure of OsHAL3 obtained from 3 ns MD simulation...... 75 Figure 4.2 Spectroscopic characterization of OsHAL3...... 76

xiii

Figure 4.3: Solvation dynamics of the ET inert mutant W79F/W82F...... 77 Figure 4.4: Normalized fs-resolved fluorescence transients of OsHAL3 mutants...... 79 Figure 4.5: Normalized fs-resolved absorption transients of the mutant W82F...... 80 Figure 4.6 Normalized femtosecond-resolved absorption transients of mutant W79F. ...82 Figure 4.7 Normalized fs-resolved absorption transients of WT OsHAL3...... 84 Figure 4.8: Normalized fs-resolved absorption transients of mutants C119S and H91F. .86 Figure 4.9 Summary of photoinduced electron transfer cycle in OsHAl3...... 87 Figure A.1 Steady-state absorption spectra of flavodoxin mutants...... 99 Figure A.2 Normalized fs-resolved fluorescence transients of flavodoxin mutants...... 100 Figure A.3 Normalized fs-resolved absorption transients of the mutant W60F/Y98F. ...102 Figure A.4 Deconvolution of fs-resolved absorption transients of mutant Y98F...... 103 Figure A.5 Deconvolution of fs-resolved absorption transients of mutant W60F...... 105 Figure B.1 Steady-state absorption spectra of OX flavodoxin mutants...... 109 Figure B.2 Gaussian fitting and histogram analysis of edge–to–edge FMN-W distances...... 110 Figure B.3 Normalized fs-resolved absorption transients of the mutant Y98F/W60F in different timescales...... 111 Figure B.4 Normalized fs-resolved absorption transients of the mutant L67W...... 112 Figure C.1 Fs-resolved fluorescence and absorption transients of the mutant W79F/W82F...... 117 Figure C.2 Fs-resolved transients of the mutant W79F/W82F/H91F gated at 495, 500 and 540 nm...... 118 Figure C.3 SQ ET dynamics of the mutant W79F using 580 nm excitation...... 119 Figure C.4 Comparison of fs-resolved absorption transients between the mutants H59/W82F and W82F...... 120

xiv

Chapter 1. Introduction

1.1 Photoreceptors and photoenzymes

Photoreceptor and photoenzymes are specialized proteins that carry out light induced biological functions in plants and animals.1-2 With the naturally occurring amino acids absorbing in the UV region, proteins absorb in the visible range using special chromophore molecules, as observed in xanthopsins,3 phytochromes,4-5 sensors of blue- light utilizing FAD (BLUF),6-7 cryptochromes,8-9 photolyases10-12 and rhodopsins.13 Each cofactor is a specially designed chromophore and is optimized for chemical processes such as electron transfer, formation/rupture of a chemical bond, bond-isomerization or conformational changes. The highly specific photochemistry and photophysics have evolved to maximize the reaction yield for signal transduction and compete with rapid energy dissipation processes such as vibrational/thermal relaxations, fluorescence and triplet formation.1,14-16 A major challenge in photochemistry research is to understand understand how photoreceptors achieve signaling at the molecular level through structural studies,17 phenotype analysis,18 and resolving the mechanism of and dynamics of photoinitiated reactions.14-16 Here we report the mechanism and dynamics of ultrafast electron transfer (ET) reactions as the primary photochemical events in flavoproteins.

1.2 Flavoproteins

Flavins are a family of isoalloxazine compounds that can readily switch between three oxidation states, namely oxidized quinine (FMN), one-electron reduced neutral semiquinone (FMNH˙), and completely reduced anionic hydroquinone (FMNH-), thereby allowing it to behave proficiently in several ET reactions (Figure 1.1A).19-20 This reaction mechanism is observed in damaged-DNA repair,10-12 magnetoreception,8-9 and fatty-acid decarboxylation21-22 processes in biological systems, where their flavin cofactors function as a redox center or a light sensor (Figure 1.2 B) to perform subsequent reactions and/or downstream signaling.

Fundamental to the mechanism of cyclic electron transfer (ET) in biological photo– machines lies the fact that the energy of absorbed photon is channeled to a highly specific reaction pathway via a charge-separated intermediate to guarantee favorable efficiency over competing futile processes.10-15 The ET dynamics are key to understanding the functions of these proteins as they control crucial steps such as chemical selectivity, time scales and reaction yields, and thus directly govern molecular mechanisms. In our efforts to understand the mechanism of such ET in proteins, we characterized the mechanism of ultrafast ET reactions in the protein flavodoxin,22-24 and the recently discovered photoreceptor OsHAL3.25-27

1.2.1 Flavodoxin and electron transfer processes

Flavodoxin is a redox-active protein known for its role as an ET carrier.28-29 It has a noncovalently bound flavin molecule (FMN) that is partially exposed to surface water molecules, making it an excellent system to study solvation and short-range ET processes 2

(Figures 1.2). At the active site, two aromatic residues (W60 and Y98) are active electron donors.24 Upon blue-light excitation of the oxidized state, an electron readily jumps from the W60/Y98F to the excited oxidized flavin. These ET rates were observed to be ultrafast owing to the stacked nature of flavin with both W60 and Y98 within van der Waals contacts. Ultrafast dynamics of forward ET (FET) and backward ET (BET) was observed to occur on the timescales of 135–340 fs and 0.95–1.56 ps, respectively, with a subsequent cooling from vibrationally hot BET products in 2.5–4.3 ps (Figure 1.3).30 This vibrational coupling in the Marcus inverted region lowers the activation energy and accelerates the charge recombination process and should be present in all flavoproteins with comparable

ET configurations. These ultrafast ET dynamics are nearly independent of environment relaxations and follow single-exponential behaviors. The environment response at oxidized state using an ET-inert mutant W60F/Y968F and observed the active-site solvation dynamics in 1.0 ps (53%), 25 ps (26%) and 670 ps (21%), corresponding to local water- network and protein-sidechains relaxations (Figure 1.4).31 Thus, the environment response is slower than the charge separation and FET essentially occurs in a frozen protein environment, following single-exponential dynamics. In Chapter 2, we use the flavodoxin system in the SQ state, where the slower ET dynamics from less favorable driving forces may overlap with environment relaxations and solvation coupled ET reactions. In Chapter

3, we characterized distance-dependent ET reactions from various FMN-W donors with different configurations using extensive mutations.

1.2.2 The Hal3 gene and bioregulation

The HAL3 gene has been discovered to positively regulate osmotic tolerance and phosphopantothenoylcysteine (PPC)-decarboxylase activity in plants.32-35

Specifically in Arabidopsis thaliana, the protein AtHAL3 regulates intracellular levels of Na+ and K+ under toxic salt concentrations in soil.32 Crystal structures also show the

PPC-AtHAL3 complex, where AtHAL3 exists as trimer and reduces PPC by a flavin mediated ET pathway, necessary for coenzyme-A biosynthesis.34 Recently, a new pathway has been discovered in the plant Oryza Sativa (common rice), where OsHAL3 promotes growth in rice seedlings by accelerating cell division, along with PPC- decarboxylase and salinity tolerance functions.26 Phenotype studies show overexpressing OsHAL3 increases height of plant seedlings by a interaction with

OsHAL3-interacting protein 1 (OsHIP1), that inactivates ubiquitination pathways(Figure 1.5). However, the accelerated growth is only observed under white- light/dark cycles, and absent under blue-light/dark cycles. Cross-linking experiments and homology in HAL3 proteins shows OsHAL3 exists as a trimer in dark or normal environments, and dissociates to monomers with blue light exposure. The monomer is unable to bind with OsHIP1, correlated with the observed light dependent phenotype. A similar phenotype of early flowering was discovered by interaction of OsHAL3 with

Heading date protein 1 (Hd1), and negative regulation by blue light exposure.27

OsHAL3 thus functions as a blue light receptor that regulates multiple phenotype through structural recognition, however the mechanism of light perception is unknown.

Interestingly, the homolog AtHAL3 do not display such light dependent biological

4 functions. Using mutagenesis and ultrafast spectroscopy, we resolved the mechanism of light induced ET reactions in OsHAL3 in Chapter 4.

1.4 Methodology: Femtosecond-resolved spectroscopy

To characterize the dynamics of ultrafast ET reactions we used fs-resolved pump- probe spectroscopy.36 The reaction is initiated by a pump pulse which generates an optical signal that depends upon the reaction dynamics. The fluorescence intensity of the excited chromophore at any time-delay can by determined by focusing the emission on a nonlinear

BBO crystal, and upconverting by a 800 beam light, called the probe pulse (Figure 1.6A).

The time delay between the two pulses can be controlled from a few femtoseconds to nanoseconds using a high-precision optical delay stage, and by repeating the experiment with a series of time-delays, the original fs-resolved fluorescence signal can be reconstructed.31 The emission can further be scanned at a series of wavelengths by altering the BBO angle using the phase-matching conditions. By using laser pulses with a temporal width of only 35 fs, we can achieve high time resolution in our experiments.

The dynamics of various intermediates and ET products can be characterized using the time resolved absorption changes. We use a similar setup with pump and probe beams.

Here, the pump beam similarly excites the sample at time zero, and changes in the intensity of probe beam is measured at desired wavelengths (Figure 1.6B).30 By knowing the absorption ranges of various reactants, intermediates and products involved, we probe the absorption dynamics systematically at multiple wavelengths, and resolve the dynamics of all chemical species involved using modelling and global fitting.

Figure 1.1 Redox states of FMN and their absorption and emission spectra.

(A) Existence of three redox states makes FMN an versatile electron acceptor and donor in ET reactions. (B) Different absorption and emission properties of the redox states enables characterization of ET dynamics through optical spectroscopy possible.

Figure 1.2 Structure of water exposed flavodoxin active site.

The partially exposed active site of flavin with two ET donors of W60 and Y98 makes flavodoxin an excellent system to study solvation coupled ET processes.

Figure 1.3 Photoinduced redox cycle of OX flavodoxin

Photoinduced ET reactions in OX flavodoxin is ultrafast and single-exponential in nature without coupling from environmental relaxations due to difference in timescales.

Figure 1.4 Solvation dynamics of flavodoxin in three redox states.

Femtosecond-resolved emission spectra of the ET inert mutant W60F/Y98F in three states with wavelength dependent energy relaxations in 1-2.6 ps, 20-40 ps and 200-670 ps timescales corresponding to local water-network relaxation, coupled local water−protein fluctuation, and flexible loop motions.

Figure 1.5 Phenotype and structural analysis of OsHAL3.

(a) Enhanced growth of rice seedlings with overexpression of OsHAL3. (b) Dissociation of OsHAL3 to monomers with blue-light exposure, and (c) regulation of ubiquitination pathways with HIP1 binding.

Figure 1.6 Experimental design of ultrafast pump-probe spectroscope.

(A) Upconversion detection of fluorescence using nonlinear mixing of emission from excited sample with BBO crystal to resolve forward ET rates and solvation dynamics (B) Transient absorption detection of changes in intensity in probe beam to comprehensively resolve dynamics of the complete ET cycle.

Chapter 2. Short-Range Electron Transfer in Reduced Flavodoxin: Ultrafast Nonequilibrium Dynamics Coupled with Protein Fluctuations.

2.1. Introduction

ET reactions have been theoretically described by the Marcus theory over past several decades37-38 and this equilibrium theory has undergone a few modifications to include effects such as local solvation or vibrational coupling39-43 as observed in recent photoinduced short-range ET reactions.16,30,44,45 Due to the ultrafast nature of these ET processes, the ET reactions can occur in a vibrationally unrelaxed excited state or couple with environment relaxation.45 Thus, more studies are needed to further elucidate the nonequilibrium short-range ET dynamics in proteins.

In our present work, we reduce the cofactor flavin to a semiquinone state (Figure

2.1A-B). Due to its smaller reduction potential,46 we expect the ET reactions to slow down, entering the time region of the local environment relaxations (Figure 2.1C). Thus, the ET dynamics may directly couple with local protein-water motions. We will systematically characterize the ET rates of the complete ET cycle and its dynamical relation with environment relaxations with a series of mutations using femtosecond-resolved fluorescence and absorption spectroscopy. Specifically, the designed various mutants are classified into four categories: (I) W60 as the sole electron donor by a single mutation of

Y98F, Y98A, Y98H and Y98R; (II) Y98 as the sole electron donor with a single mutation 12 of W60F and W60A; (III) both W60 and Y98 as electron donors of WT and single mutation of D95N, G61A and G61V; and (IV) double donors of W60/W98 and Y60/Y98 by mutation of Y98W and W60Y, respectively. We dissected FET, BET and subsequent vibrational cooling dynamics of each mutant protein and measured all the reaction timescales. These results will provide deep insights into the molecular mechanism of ultrafast short-range ET in proteins.

2.2. Materials and methods

2.2.1. Protein purification

The procedures of the expression and purification for Desulfovibrio vulgaris flavodoxin wild type and mutants have been well established.47,48 The 12 mutants (Y98F,

Y98A, Y98H, Y98R, Y98W, W60F, W60A, W60Y, D95N, G61A, G61V, W60F/Y98F) were designed and prepared by site-directed mutagenesis following the previous procedures. Upon purification, the samples were in oxidized form (FMN) and converted to the neutral radical semiquinone form (FMNH˙) by UV irradiation in anaerobic conditions.25 Due to the change of redox potentials by mutations, the duration of UV irradiation differs for different mutants. For fs-resolved experiments, the protein concentration of ~150 μM was used.

2.2.2. Femtosecond–resolved spectroscopy

All fs–resolved measurements were carried out using fluorescence upconversion and transient absorption methods. The experimental layout has been described elsewhere.36

Briefly, the pump wavelength was set at 580 nm to avoid excitation of any contamination

13 of oxidized states, if any. The pulse energy was attenuated to 100–150 nJ before being focused onto the sample cell. All samples were maintained under anaerobic conditions to prevent oxidation to quinone forms. For the fluorescence upconversion experiments, the emission was gated by another 800 nm laser beam in a 0.5 mm thick β-barium borate crystal

(BBO, type I). The emission was probed at different wavelengths by tuning the mixing crystal angle. For transient-absorption experiments, the probe pulses at the desired wavelengths between 525 and 800 nm were obtained via an optical parametric amplifier

(TOPAS, Spectra-Physics). The instrument responses are 300–350 fs and 150–200 fs for the fluorescence and transient absorption detection, respectively. All experiments were conducted at the magic angle (54.7°). To prevent heating and photobleaching, the sample was kept stirring in quartz cells with 1 or 5-mm thickness during laser irradiation.

Nonlinear signals at time zero in transient-absorption experiments were subtracted throughout data analyses.

2.3. Results and discussions

2.3.1. Spectroscopic properties and probing strategies

Figure 2 shows the absorption and fluorescence spectra of wild-type semiquinone flavodoxin as well as the absorption spectrum of the mutant Y98W. The WT absorption

49 spectrum shows its characteristic D1 D0 transition peaking at 580 nm with a shoulder near 620 nm. Except for Y98W (W60/W98), all mutants follow the WT absorption spectrum with Y mutants shifting to the red side by ~10 nm, W mutants shifting to the blue side by ~10 nm, and D95N and G61A mutants resulting in a blue shift, indicating different

14 electronic stabilization by these mutations (Figure A1). The G61V mutant is very difficult to reduce to the semiquinone state probably due to the bulky hydrophobic residue V to exclude nearby water molecules. The absorption of G61V is usually a mixing of oxidized and semiquinone states (Figure S1). With closely spaced two Trp residues in a sandwich configuration, the Y98W mutant experiences electronic delocalization owing to its charge- transfer nature in the molecule with an absorption tail extending to the red side (Figure

2.2), as also observed in its oxidized state.30 The emission spectrum of semiquinone state, peaking around 700 nm, was first reported in our early studies20 and obtained from the redox-inert W60F/Y98F mutant in the absence of ET quenching. It shows a typical non- lognormal shape like other flavoproteins50,51 possessing a heterogenous electrostatic protein environment.

Owing to the dynamic solvation behavior as we reported before,31,50 the excited semiquinone dynamics was probed at multiple wavelengths (652, 670 and 725 nm) from the blue to red side of the emission spectrum to detect solvent-interlaced ET behavior. For transient-absorption measurements, the probe is systematically tuned in the visible region over a wide range of wavelengths to capture the entire dynamics of all intermediates and products involved (Figure 2.2, upper panel). At the red side (800–740 nm), we exclusively detect the excited-state (FMNH˙*) decay, representing direct ET rates. The intermediate

W+ is detected at wavelengths mainly shorter than 700 nm.52 Possible hot ground states of

FMNH˙† is detected at as longer as 670 nm and portrayed a red shifted FMNH˙ absorption spectrum. The ground-state FMNH˙ is dominant at the wavelengths less than 640 nm.

2.3.2. Forward electron transfer and stretched behavior

To determine intraprotein electron transfer rates, we directly measured the fluorescence quenching dynamics of excited semiquinone flavin. We took the fluorescence transients of WT and 11 mutants and they all show ultrafast decays on the picosecond timescale (Figure 2.3 and Figure A2). Figure 2.3A shows several typical transients of the

ET quenching dynamics gated at 725 nm for WT and 5 mutants and Figure 3B shows the transients of Y98 (W60F) and W60 (Y98F) gated at three different wavelengths of 652,

670 and 725 nm. As reported before,20 the effective lifetime of FMNH˙* is ~230 ps using the redox-inert mutant W60F/Y98F, indicating the observed ultrafast decays in Figure 3 being the ET dynamics with W60 and/or Y98. The fluorescence transients at 725 nm of the red-side emission best represent the ET dynamics with minor solvation contributions, consistent with the transient-absorption measurements (see below). All these transients at

훽 725 nm can be best fit by 퐴푒−(푡/휏) (A, an amplitude; τ, a time constant; and β, a stretched parameter) (Table 2.1), although they can also be represented by a double-exponential decay (Table A1).

Specifically, for WT with the two electron donors, the ET dynamics occurs in 1.5 ps with β=0.78. For D95N, the ET dynamics with the same donors becomes faster in 1.4 ps with β=0.81, indicating a favorable driving force due to the mutation. For Y98F with only one donor of W60, the dynamics becomes slower in 2.9 ps with β=0.83. For the mutants of W60A and W60F with the same donor of Y98, the ET dynamics are in 4.2 ps

(β=0.83) for the former and 7.0 ps (β=0.83) for the latter, reflecting the different electrostatic environment. The addition of rates of W60A and Y98F is equal to that of WT.

For W60Y with two Y donors, surprisingly, the dynamics is very similar to W60F and in

7.0 ps (β=0.80), indicating that the Y60 donor may make a minor contribution. All resulting average times and β values are listed in Table 2.1. As also shown in Figure 3B, the transients gated at bluer wavelengths become gradually faster, a signature of ultrafast solvation dynamics mixing with the ET process. As reported before,31 the solvation dynamics at the binding site using W60F/Y98F mutant occur in 2.6 ps (75%) and 40 ps

(25%). The even longer solvation process in hundreds of picoseconds was not resolved due to the short lifetime of ~230 ps. Clearly, the ET dynamics in a few picoseconds observed here does overlap with the local solvation process, leading to the observation of a stretched behavior.

2.3.3. Vibrationally-coupled charge recombination and ground-state cooling dynamics.

To switch to transient-absorption detection of the ET dynamics, we first characterized the W60F/Y98F mutant as a control and the transients probed from 800 to

525 nm are shown in Figure 4. All transients show a long lifetime component of ~300–400 ps except that at 630-690 nm the transients also exhibit two initial rise components in 1.8–

2.6 ps and 8.5–26.6 ps (Table 2.2). These two components are the reflection of solvation processes as observed in the fluorescence measurements. These initial dynamics are incorporated in our transient-absorption ET data analyses whenever applicable. In Figures

4.5-7 (and Figures A3-A5), the transients probed at 800 nm directly reflect the forward ET dynamics and those probed at shorter wavelengths are mixture of the initial excited state, reaction intermediates and final products.

The resulting FET reaction times for WT and mutants probed at 800 nm are similar to those obtained from the fluorescence dynamics at 725 nm; these values are also listed in

Table 1. Clearly, the FET dynamics with one W donor are faster than that with one Y due to favorable free energy for the former by ~0.3 eV.53 For WT (Figure 2.7A), the ET rate is faster than that of W60 (Figure 2.5) or Y98 (Figure 2.6) and is closely the sum of both channels of W60 and Y98. Within class I with one W donor, the ET rates increase from the Y mutant of F to A, H and R reflecting stabilization of charge-separated states.

Similarly, for class II, the FET rates increase from the W mutant of F to A. Within class

III, the D95 mutation places the positive asparagine residue in the vicinity of flavin and stabilizes FMNH– intermediate, thereby accelerating ET (Figure 2.3 and Figure 2.7B). On mutating G61 to a nonpolar residue G61A, the H-bond to N(5) flavin is eliminated and the local structure could become loose.54 The FET process takes place over a longer distance with a slower rate. For G61V, the ET rate is fastest, reflecting a possible different donor- acceptor configuration around the active site due to a bulky valine residue. For class IV, the effect of Y60 as an additional electron donor in the mutant W60Y (Figure 2.7C) with two Y donors is not noticed, probably due to the flexible gate position of the binding pocket to the surface.55 For Y98W with two W donors (Figure 2.7D), surprisingly we can detect the ultrafast ET dynamics in 360 fs with β=1.0, the fastest ET rate from the two efficient

W donors being stacked in a sandwich configuration with the isoalloxazine ring. The ET dynamics is much faster than the initial solvation relaxation (2.6 ps), the environment is relatively frozen, and thus the FET dynamics becomes a single-exponential decay.

Interestingly, we could not detect such ultrafast ET dynamics from the fluorescence

18 measurement, indicating a delocalized electronic state as shown in Figure 2.2 with a long tail absorption. But, we still can detect the complete charge-separation dynamics by the transient-absorption method. For the oxidized state of the same mutant, due to the significant delocalization, we did not detect the FET dynamics by both fluorescent and transient-absorption methods.30

To capture the ET intermediates such as W+, we probed the dynamics from 800 to

525 nm. As shown in Figures 5-7, the transients systematically show the charge-separation dynamics, intermediate evolution and product formation probed from the red to blue side.

Mostly, the transients probed at 800 and 740 nm show the forward ET dynamics. From 690 to 650 nm, we observed that the transients progressively get slower. From 690 nm, we detected a rise-decay signal from W+ intermediate. Similar to the observation for oxidized flavodoxin,30 the charge recombination leads to the formation of a hot ground-state product

FMNH˙†. This hot state has a red-shifted absorption of the initial ground state (Figure 2.2), resulting in our observation of FMNH˙† in the 650–690 nm region (Figure 2.5-7).

Specifically, in Figure 2.4, we observed the charge recombination for W60 in 0.87 ps with =0.93. This BET is faster than the forward ET, similar to the semiquinone dynamics observed in photolyase,16 resulting in an apparent reverse kinetics with the rise representing the charge recombination and the delay being the charge separation. The  values is larger than the corresponding FET stretched one (=0.83), reflecting less extent of solvent coupling due to the faster BET dynamics. Besides the FET and BET signals, we detected distinct cooling dynamics involving vibrationally-excited BET product FMNH˙†

(B, C panels in Figure 2.4). The cooling relaxation takes about 3.2–5.0 ps. The more 19 contributions of the cooling process toward 650 nm gives rise to the observed slower pattern from 690 to 650 nm. From 640 to 525 nm, we detected FMNH˙ recovery from the hot FMNH˙† (also in Figure A.3 for deconvolution). Interestingly, the overall signal becomes faster due to partial signal cancellation of hot-state cooling with ground-state recovery. Similar patterns were observed in the ET cycle of all class I mutants, with faster recombination ranging in 0.80–0.92 ps, β=0.90–0.93 and the cooling dynamics in 3.0–5.0 ps.

For class II with sole Y donor, Figure 2.5 shows all the transients for Y98 (W60F), characteristic to the ET behavior involving only one Y donor. Similarly, the transients probed at 800 and 740 nm are exactly the same, representing pure FET dynamics in 6.3 ps and significantly slower than those in class I with one W donor. From 690–640 nm, we clearly observed an initial rise component in 1.1–2.1 ps, similar to what we observed in

Figure 2.4 for W60F/Y98F mutant (also Table 2.2) and resulting from the detection of initial solvation process. Because the Y+ intermediate only absorbs at wavelengths less than

510 nm,56 we only detected the species of FMNH˙*, FMNH˙† and FMNH˙. By systematic fitting (also in Figure A.4) and assuming the similar cooling dynamics in 3–5 ps, we observed the BET in 8.0 ps with =0.81. Similarly, for W60A, we observed BET in 6.0 ps

(=0.79). These BET dynamics are slightly slower than their FET dynamics. The cooling dynamics were found to be in 3.5–5.0 ps, similar to the observation for the class I mutants and also for the oxidized state. Note that for such faster cooling with slow charge recombination, the resulting dynamics appears a reverse kinetics with less accumulation of the hot FMNH˙†. Also, the gradually slowing pattern results from the more contributions 20 of the hot ground state but the decay here, different from the class I, is the charge recombination dynamics which are the same for all the probed wavelengths.

Figure 2.6 shows transient-absorption signals for WT, D95N, W60Y and Y98W, each having two electron donors present. For the former two (Figures 2.6A and 2.6B), we detected the FET dynamics in 1.8 ps (=0.83) and 1.4 ps (=0.81), respectively, by probing

800 and 740 nm. According to the two-channels model, we obtained the BET dynamics of charge recombination of W60 for WT and D95N in 1.1 ps with =0.89 and 0.91, respectively. We could not resolve the BET dynamics for the Y98 donor in these two-donor systems due to minor contributions of Y+ signals at the probed wavelengths (690–525 nm).

The cooling dynamics for the two proteins are similar in 3–5 ps. For W60Y in Figure 2.6C, the transient pattern is very similar to W60F, leading to a similar FET in 6.1 ps (β=0.80),

BET in 8.8 ps (β=0.81) and the cooling dynamics in 3–5 ps. As discussed earlier, the mutation of W60Y may make a minor contribution to the ET dynamics due to possible structural changes. For Y98W in Figure 2.6D, as mentioned early, we detected the FET dynamics at 800 nm. Furthermore, we observed a W+ intermediate signal at 740 nm, the only one among WT and 11 mutants, indicating a red shift of W98+ absorption due to the unique sandwich structure. We observed the BET dynamics in 1.0 ps with similar cooling dynamics in 3–4 ps. We similarly resolved ET dynamics of all other mutants in Figure 2.7.

2.3.4. Electron transfer rates and reorganization energies.

Using ultrafast spectroscopy, we measured the timescales of charge separation, charge recombination and cooling dynamics. With the information, we apply the Marcus theory to further analyze the ET mechanism. The ET rate can be expressed: 21

3 표 2 2 4 휋 (훥퐺 +휆) 푘 = 퐻퐴퐵 √ 2 푒푥푝 [− ] [1] ℎ 퐾퐵푇휆 4휆퐾퐵푇

-1 where k is ET rate in seconds , HAB is coupling constant in eV, λ is the reorganization energy in eV and ΔGo is the total free energy in eV. Although the ET dynamics are in nonequilibrium, the estimated ET parameters using the Marcus theory provide valuable insights into the ET mechanism.

57 Using the average forward and backward ET rates, the energy of D1 D0 transition at 620 nm as 2.0 eV, the reported reduction potentials for the mutants,47,54,58 and the constraints of a larger reorganization energy for backward ET than that for forward ET, we determined the λFET, λBET and HAB for all donor-acceptor pairs (Table

3). For ET in class I mutants, λFET is from 0.91 to 1.01 eV with a larger λBET ranging from 1.07 to 1.22 eV and HAB = 13.6 meV. As a result, the forward ET occurs in the

-1 normal region while backward ET in the inverted region. The large HAB (109 cm ) value signifies a significant coupling between the flavin and W60. Among these mutants, the flavin potential is significantly lower in Y98F (Table 3), possibly due to unfavorable interactions between anion FMNH- and the stacked F residue. This accounts for the observed slowest ET in 3.1 ps among class I mutants. For class II mutants, we obtained smaller reorganization energies for both forward and backward ET processes. Being in the hydrophobic pocket, Y98 experiences less exposure than W60, which lies on the flexible outer loop.55 Also the coupling between flavin and Y98 is greater with 14.8 meV (120 cm-1) resulting from a shorter separation between the two. The reduction potential of Y98/Y98+ in the hydrophobic environment was observed to be 1.25 eV, 0.22 eV smaller than the reported tyrosine potentials at 1.47 eV.53 22

For class III mutants with both W60 and Y98 as electron donors, we analyzed the data by assuming that the addition of individual rate contributions from the W60 and Y98 donors using the Marcus theory is equal to the experimentally observed rate (kET(expt) = kW60

+ kY98). We used the derived coupling values from the single-donor mutants as the coupling constants for the two donors. In WT, the λFET values were 0.85 and 0.82 eV for W60 and

Y98, respectively, smaller than those in single mutants, arising from the fast ET timescale with less solvation coupling. For D95N mutant, the mutation changes the redox environment around Y98 and the flavin, thus resulting in changes of λFET, 0.89 and 0.87 eV for W60 and Y98, respectively. For G61 mutants, the mutation is on the W60 side and thus λFET remains constant for Y98 donor at 0.90 eV, while λFET shifts to 1.05 and 1.08 eV for the W60 donor in G61A and G61V, respectively.

We could determine λBET for the donor W60, but not for the donor Y98 in this class III. λBET in WT is 1.20 eV, closer to that of Y98F. For D95N, λBET shifts to

1.16 eV due to the change of the driving force. G61A and G61V mutants experience a larger change in λBET of 1.2 and 1.1 eV, respectively, resulting from the different protein environment associated with the mutation as discussed earlier. For the λBET values of the

Y98 donor, because we could not resolve the weak Y98+ signals in our probing range, we cannot deduce the BET rates of the Y98 donor and thus we did not extract the λBET values for Y98. All other values are reported in Table 2.3.

The final photoinduced redox cycle is summarized in Figure 2.8. Although flavodoxin is not a protein with photoinduced function, it is an excellent system to investigate short-range ET in proteins. Together with our earlier studies of oxidized

23 flavin,30 we clearly observed how the protein fluctuations modulate the ET process, a coupling process of the ET configurations with local protein-solvent coordinates. By our measurements of protein relaxations on the picosecond time scale, the ultrafast ET in a few hundreds of femtoseconds in the oxidized state sees a “frozen” protein environment with less extent of coupling, leading to our observation of a single-exponential decay of

ET dynamics. But for semiquinone flavin, the ET dynamics in a few picoseconds enters the protein fluctuation region, as illustrated in Figure 1C, a strong coupling is expected and clearly, we observed a stretched exponential or multiple-exponential decay. Such coupling effects should be common in many ET dynamics occurring on the picosecond timescales in complex systems. Clearly, these ET dynamics are nonequilibrium processes and the Marcus theory should be cautiously used as discussed above. As a better approximation, we may apply the Sumi-Marcus model40 to treat such coupling processes to get deeper insights and the work is in progress.

The observation of vibrationally-excited ground-state flavin after charge recombination is significant. From both oxidized and semiquinone states, we observed the same pattern of vibrational cooling in ~3–5 ps at slightly long wavelengths of the ground- state absorption. Clearly, the ultrafast forward and backward ET dynamics provide an opportunity to observe the slower cooling dynamics. Such vibrational excitation in back

ET is general to all flavin-involved BET processes and in many cases, the slow BET masks the fast cooling dynamics. The recent report of the ET dynamics in the TrmFO protein unfortunately missed to analyze such cooling dynamics around 500–550 nm but attributed the longer dynamics to the absorption of the tyrosyl cation.59 It needs more caution in

24 analyses of these ultrafast ET dynamics, for example, the charge-separated anion flavin

(FMNH–) intermediate could also be formed in vibrationally hot states and our careful simulations including a similar cooling process and BET dynamics for hot FMNH– give no difference of our current results, i.e., the intermediate anion flavin FMNH– could also be vibrationally hot and we could not presently resolve it. More theoretical work is needed for further understanding of such ultrafast ET processes.

2.4. Conclusion

We investigated the short-range ET dynamics in semiquinone flavodoxin to further examine the coupling of the local protein-solvent fluctuations with ET processes using femtosecond spectroscopy and extensive mutations. Compared with our previous studies of oxidized flavodoxin, the smaller driving force in semiquinone flavodoxin makes ET slow down, entering the time region of local protein fluctuations. We thus widely observed a stretched dynamic behavior. For the four different groups of mutants with the donor(s) of W, Y, W/Y and Y/Y, we observed the cyclic ET dynamics in a few picoseconds with a stretched parameter of β ranging from 0.76 to 0.93. Specifically, we observed the forward

ET dynamics for the W donor in 1.4–3.4 ps with β=0.76–0.86 and the subsequent back ET in 0.84–0.96 ps with β=0.89–0.90. The BET processes were observed faster than their FET dynamics. For the Y donor, we observed the forward ET dynamics in 4.2–7.3 ps with

β=0.78–0.80 and subsequent back ET in 6.8–9.9 ps with β=0.79–0.81. The BET dynamics here are slower than their FET process. For the special mutant (Y98W) with W60/W98 donors, we observed the fastest ET in 0.36 ps and the back ET in 1.0 ps without a stretched behavior, due to the stacked nature of two W residues sandwiching the flavin acceptor and 25 also the relatively slow solvation dynamics. All the BET processes for WT and 11 mutants involve the hot vibrational-state formation and we also observed the subsequent cooling dynamics in 3–5 ps for all the proteins.

The observed stretched ET behaviors and hot vibrational-state formation during charge recombination are significant. As we sketched in Figure 2.1C, depending on the local protein-water fluctuations, the ET dynamics could be faster than the local solvation, as we observed in oxidized flavodoxin, leading to a single-exponential decay behavior with a "frozen" environment and thus a nearly zero of solvent reorganization energy (0). When the ET slows down, entering the time region of the local fluctuations, the ET dynamics gradually couple with the protein-water motions, resulting in a non-single-exponential dynamic behavior. The solvent reorganization energy (0) seems gradually increased while the ET slows down. When the ET is much slower than the local environment fluctuations, the ET process follows the traditional Marcus theory and the ET dynamics is always in equilibrium with the local protein-solvent configurations. Here, the short-range protein ET is always ultrafast in the crossing time region of the local environment fluctuations and in nonequilibrium. The vibrational ground-state formation during the BET is robust and should be general to any flavin-involved back ET process. More theoretical work is needed for the complete understanding of such ultrafast nonequilibrium dynamics in complex proteins.

Table 2.1 Timescales of ET reactions and subsequent cooling dynamics in

semiquinone flavodoxin.a

FET BET c Mutants Donor(s) b <τ> c c c τcooling <τ> β <τ> β (<τW>,<τY>) I Y98F W60 3.2 3.40 0.83 0.84 0.90 5 Y98A W60 2.20 0.79 0.93 0.93 3-5 Y98H W60 1.80 0.76 0.96 0.92 3-5 Y98R W60 1.40 0.83 0.90 0.93 3-5 III W60F Y98 7.7 7.30 0.78 9.00 0.81 5 W60A Y98 4.6 4.20 0.78 6.80 0 0.79 5 IIId WT W60, Y98 1.7 1.90 0.86 1.2, NR 0.89 4 (2.8, 5.6)0 D95N W60, Y98 1.6 1.60 0.81 1.2, NR 0.91 3-5 (2.4, 4.9)0 G61A W60, Y98 2.50 0.81 1.2, NR 0.90 3-5 (6.0, 4.2)0 G61V W60, Y98 1.50 0.80 0.92, NR 0.91 3-5 (4.2, 2.3)0 IV Y98W W60, W98 0.36 1.0 1.0 0 1.0 3-4 W60Y Y60, Y98 8.2 6.90 0.80 9.90 0.81 5   a  1 b Time constants are in units of ps. All average times are calculated using <τ> =   Obtained from the fluorescence     dynamics gated at 725 nm. cObtained from the transient absorption detection. dFor two donors, each individual ET timescale can be resolved from the total ET time by knowing the driving forces, coupling constants for the W and Y donors and the ranges of two reorganization energies. NR: not resolved for the Y donor.

Table 2.2: Transient-absorption dynamics in W60F/Y98F in semiquinone flavodoxin.a

Probe wavelengths 525 630 640 650 655 660 670 690 740 800 τ1 2.1 2.1 1.86 1.7 1.1 1.8 τ2 22.7 22.7 19.8 15.7 26.6 8.5 14.5 τ3 302 447 373 359 354 343 348 302 322 316

A1 -2 -53 -25 -21 -20 -12 A2 -46 -17 -12 -7 -9 -16 -13 A3 -98 30 100 100 100 100 100 100 100 100 Amp3aTime constants are in units of ps. Wavelengths are in units of nm, amplitudes are calculated relative to each other and negative values indicate rise components. -

Table 2.3 Energetics of cyclic electron-transfer reactions in semiquinone flavodoxin.a

b 표 표 Mutants Donor(s) E(SQ/HQ) HAB 훥퐺퐹퐸푇 휆퐹퐸푇 훥퐺퐵퐸푇 휆퐵퐸푇 I Y98F W60 -0.414 13.6 -0.436 0.91 -1.564 1.24 Y98A W60 -0.304 13.6 -0.546 0.99 -1.454 1.11 Y98H W60 -0.262 13.6 -0.588 1.01 -1.412 1.07 Y98R W60 -0.265 13.6 -0.585 0.97 -1.415 1.08 II W60Fc Y98 14.8 -0.340 0.90 -1.660 1.05 W60A Y98 -0.357 14.8 -0.393 0.90 -1.827 1.02 IIId WT W60, Y98 -0.443 13.6, 14.8 -0.407, - 0.85, 0.82 -1.593,- 1.20, NR D95N W60, Y98 -0.395 13.6, 14.8 -0.455, - 0.89, 0.87 -1.545,- 1.16, NR G61A W60, Y98 -0.359 13.6, 14.8 -0.491,0.307 - 1.05, 0.90 -1.509,1.693 - 1.12, NR G61V W60, Y98 -0.299 13.6, 14.8 -0.551,0.355 - 1.08, 0.90 -1.449,1.645 - 1.11, NR a 표 b 47,54,58 c 표 E(SQ/HQ), λ and Δ퐺 are in units of eV; HAB is in meV. Calculated0.391 from reported redox potentials.1.609 Δ퐺 values for the mutant W60F were estimated from the Marcus theory. dFor BET 0.451involving dual donors, we obtained1.549 λ values only for the W60 channel. NR: not resolved for the Y donor.

Figure 2.1 Structural and Electron Transfer properties of flavodoxin.

(A) X-ray crystal structure of semiquinone D. vulgaris flavodoxin (PDB code 4FX2) with the active site highlighted. (B) Close-up view of the active site with hydration water molecules obtained from a 500-ps MD simulation snapshot. FMNH˙ (purple) is sandwiched between W60 (green) and Y98 (yellow) at 3.5 Å and 3.3 Å, respectively. Neighboring residues of D95 (teal) and G61 (orange) are also shown. (C) Generalized timelines of photoinduced ET dynamics and active-site solvation relaxations in flavodoxin. The active site relaxations start from ~1 ps. For oxidized flavodoxin, the photoinduced cyclic ET is ultrafast within ~1 ps and senses a frozen environment. For semiquinone flavodoxin, the cyclic ET is in a few picoseconds and couples with active-site motions.

Figure 2.2 Steady-states spectra of semiquinone flavodoxin. The absorption spectra of WT (blue) and Y98W (red) and the emission spectrum of the ET-inert mutant W60F/Y98F at longer wavelengths. Note the long-tail absorption of Y98W due to the electron delocalization with a charge-transfer character. The pump wavelength and the gated fluorescence wavelengths are marked in arrows. The top bars represent the absorption ranges of the excited state, various intermediates and final product. The transient-absorption probe wavelengths are also marked by top arrows.

Figure 2.3 Forward ET dynamics of WT flavodoxin mutants. (A) Normalized femtosecond-resolved fluorescence transients of WT flavodoxin and mutants gated at 725 nm. (B) Normalized femtosecond-resolved fluorescence transients of Y98F and W60F gated at 652, 670 and 725 nm. Note the faster dynamics at the blue side resulting from the mixing of solvation and ET reaction.

Figure 2.4 Normalized fs-resolved absorption transients of the mutant Y98R. Inset A shows the gradual changes with the different probe wavelengths. Note the longer dynamics probed at 690–650 nm. Insets B and C show the deconvolution of the transients into constitutive species probed at 655 and 630 nm, respectively. The excited-state signal of FMNH˙* in inset C is relatively small and not shown.

Figure 2.5 Normalized fs-resolved absorption transients of the mutant W60F.

Inset A shows the gradual changes of the dynamics with the different probe wavelengths. Note the longer dynamics probed at 690–640 nm. Insets B and C show the deconvolution of the transients into constitutive species probed at 650 and 630 nm, respectively.

Figure 2.6 Normalized fs-resolved absorption transients with dual donors.

(A) W60/Y98 (WT), (B) W60/Y98 (D95N), (C) Y60/Y98 (W60Y) and (D) W60/W98 (Y98W). Note similar patterns with longer dynamics in the probe region of 690–640 nm and the difference in timescales for these mutants.

Figure 2.7 Normalized fs-resolved absorption transients of flavodoxin mutants.

(A-C) Transient-absorption signals of class I mutants with W60 as electron donor. (D) Transient-absorption signals of class II mutants with Y98 as the electron donor. (E and F) Transient-absorption signals of class III mutants with W60 and Y98 as electron donors. Note that G61V transient probed at 650 nm has been removed due to low signal-to-noise ratio.

Figure 2.7 continued

Figure 2.8 Photoinduced redox cycle coupled with environmental relaxations.

ET dynamics overlapping with active-site relaxations including hydration water molecules. Note the formation of vibrationally excited products FMNH˙† after charge recombination that subsequently decay in 3–5 ps to complete the ET cycle.

Chapter 3. Distance–Dependent Electron Transfer in Oxidized Flavodoxin: Ultrafast Nonequilibrium Dynamics Coupled with Protein Fluctuations.

3.1. Introduction

The mechanisms of these ET processes depend on factors such as the driving force

o (ΔG ), the electronic coupling between donor and acceptor (HAB) and the solvent reorganization energy (λ) as described by the equilibrium Marcus theory.37,38 To understand the nature of photoinduced ET reactions and their implications in biology, it is

16,60,61 critical to understand how these parameters regulate ET behavior in proteins. HAB is exponentially related to ET tunneling distances (R) as, 퐻퐴퐵(푅) =

1 퐻 (푅 ). exp [− 훽 (푅 − 푅 )], where 훽 is the decay constant in Å-1, making donor– 퐴퐵 0 2 푇 0 푇 acceptor distances vital in determining ET rates and photoreceptor design.39,62-64 Contrary to the classical Marcus theory, recent studies in short–range intraprotein ET reveal nonequilibrium dynamics,15,16,30,65 and further studies are required to analyze the distance– dependent ET mechanism in such short–range systems.

Here, we use an FMN–W redox pair in OX flavodoxin to investigate distance– dependent ET behavior with identical driving forces. Specifically, we mutated the nearby electron donors (W60, Y98, and Y100) to the redox inert residue F and used the mutant

W60F/Y98F/Y100F as a template to introduce an additional W donor at sites D95, F101,

D62, I65, E66, G103, A104, L67, and R131 to investigate the effect of local protein configurations in distance–dependent ET reactions (Figure 3.1A). The resulting mutants namely W60F/Y98F/Y100F/ D95W etc. are represented as D95W hereafter, for simplicity.

We systematically determined dynamics of charge separation, charge recombination, triplet formation and cooling processes, and analyzed fluctuations in FMN–W distances in the mutants. Our results present a detailed picture of nonequilibrium heterogenous ET dynamics, and the significance of local protein environment in the ET process and provide deep insight into the ET mechanism in photoreceptors.

3.2. Materials and methods

3.2.1. Protein purification

Wild-type flavodoxin was expressed and purified according to previously established protocols.47-48 We designed the triple mutant W60F/Y98F/Y101F as the ET–inert template to introduce an additional tryptophan as the flavin redox partner by performing the single mutations: D95W, I65W, F101W, D62W, E66W, G103W, A104W, L67W, R131W on the template. All mutants were prepared by sequential site-directed mutagenesis and purified following the same procedure. The protein concentration was maintained at ~150 M for time-resolved experiments.

3.2.2. Femtosecond–resolved spectroscopy

Femtosecond-resolved upconversion and transient absorption spectroscopy were carried out by pump-probe technique as described in our previous work.36 Briefly, the protein samples were pumped at 400 nm by using the frequency–doubled 800 nm beam. 40

The pump energy was attenuated to ~100 nJ. For the upconversion experiments, the FMN* emission at 538 nm was gated by another 800 nm laser beam in a 0.5 mm thick β-barium borate crystal (BBO, type I). For time-resolved absorption experiments, the probe wavelengths were tuned at 800, 740, 630, 515 and 480 nm via an optical parametric amplifier (TOPAS, Spectra-Physics). The instrument responses were 300–350 fs and 150–

200 fs for the fluorescence and transient absorption detection, respectively. All experiments were conducted at the magic angle (54.7°). All nonlinear signals at time zero in TA experiments, and dynamics arising from free FMN* in solution, if present were systematically subtracted throughout data analyses.

3.2.3 Molecular dynamics simulations.

Molecular Dynamics Simulations were performed using AMBER16 package with ff14SB and reported flavin forcefields.65 The initial starting geometry was taken from the reported flavodoxin structure (PDB code 2FX2).24 Mutant structures were created by replacing W60, Y98, and Y100 with Phe and then incorporating Trp at desired sites via

Pymol. For all mutations, the most probable conformations were chosen for the new W.

The structures were then optimized by tleap and solvated by using the TIP3P water model.

We performed standard energy minimization, heating, and equilibration procedures followed by the final run for a total of 12500 frames with 2 fs intervals to capture 25 ns dynamics. The resulting trajectories were analyzed using CPPTRAJ to obtain the shortest

FMN-W distances between the ring atoms. Representative distances in 5 ns timescales are plotted in Figure 3.2C, and analyzed in Figure B2.

3.3. Results and discussions

3.3.1. Spectroscopic properties and probing strategies

Figure 3.1B shows the steady state spectra of flavin in the active site. The absorption spectrum (red) of wild–type (WT) flavodoxin shows the S1 S0 peak at 450 nm

20,31 with a shoulder near 480 nm, and an additional S2 S0 peak at 380 nm. With mutations of W60 and Y98 aromatic residues, the mutants have blue–shifted absorption of 5–10 nm,

(Figure B1), indicating minor differences in their FMN environment. The steady–state emission spectrum was obtained from the ET–inert mutant W60F/Y98F and represents a non–lognormal profile arising from the heterogeneous electrostatic protein environment, commonly observed in other flavoproteins.50,51 We gated fluorescence dynamics of FMN* at emission maximum of 538 nm to characterize FET dynamics.

Using transient–absorption detection, we systematically probed the visible region to resolve the dynamics of all reactants, intermediates and products involved. Briefly, in the 800–740 nm region we identified the FET rates from FMN* absorption. At 630 nm, we resolved the W+ signal to identify BET dynamics. We probed 515 nm, to detect cooling dynamics of FMN†, in the red-shifted region of the FMN absorption. In the blue side, we monitored the 480 nm signal to observe FMN recovery dynamics as the final ET product.

We additionally probed the mutant W60F/Y98F, in the 800–480 nm region at 10 wavelengths to distinguish nature of FMN*, 3FMN, solvation dynamics (Figure B3). We observed solvation processes in 0.9–1.8 ps, 9–30 ps, and 600–700 ps, and identified 3FMN formation in 6.7 ps, along with a lifetime decay of FMN* in 6.0 ns, consistent with reported 42 results.20,31 In the 800–740 nm region, we detected minor 3FMN signals, which increased in the 700–630 nm as observed from overall slower transients. In 610–530 nm, due to cancelation of 3FMN rise dynamics with the stimulated emission signals of FMN*, the transients become faster. At the extreme blue side of 480 nm, the transient is negative due to dominant signal of FMN recovery. All these dynamics are incorporated in our analyses of ET processes as applicable. Especially, the triplet dynamics were subtracted to

determine charge separation rates (푘퐹퐸푇 = 푘퐹푀푁∗– 푘 3퐹푀푁).

3.3.2. Forward electron transfer and dynamic heterogeneity

To characterize the charge separation rates, we directly measured the fluorescence dynamics of FMN*. Figure 3.2(A–C) show the fluorescence transients obtained from 10 mutants, each with a different tryptophan donor, and the ET–inert mutant

W60F/Y98F/Y100F. The mutants display a wide range of FET dynamics, from hundreds of femtoseconds to a few nanoseconds, generally slowing down with an increasing FMN–

W distance. The Y98F mutant undergoes the fastest charge separation (Figure 3.2A) with a single exponential decay in 260 fs, as was determined from our previous studies.30 The ultrafast rate arises from the short tunneling distance of 3.4 ±0.2 Å between the electron donor W60, and FMN (Figure 3.2D). The single–exponential nature is significant, indicating the ET process is independent of local protein–water relaxation dynamics which are slower (> 1 ps) and charge separation occurs in a static environment.

In the D95W mutant, with an increased FMN–W separation at 3.8 ±0.5 Å, we observed charge separation dynamics in a few picoseconds (Figure 3.2A). With a slower

43 reaction timescale, the FET reaction overlaps with local protein-water motions, resulting

푡 훽 −( ) 휏 in a stretched-exponential behavior. The FMN* dynamics is fitted as 퐴1. 푒 1 +

푡 훽 −( ) 휏 퐴2. 푒 2 , and the resulting average FET timescales are in 2.2 ps (46%) and 10 ps (54%),

휏1 1 and β=0.84 (⟨휏1⟩ = ⁄훽 . Γ ( ⁄훽) etc.). The donor, W95 being present in the flexible loop undergoes large fluctuations (Figure 3.2D). With a unique ET rate present in each of the possible FMN–W configurations, a heterogeneity is introduced in the ensemble–averaged

ET dynamics, and the two stretched exponentials statistically describe such heterogenous behavior. The identical stretch for the two ET timescales indicates comparable coupling with the environment. Such heterogeneity was not observed in our previous studies with

W60 as the donor, because at close van der Waals distance, W60 has stacking interactions with the flavin and forms a stable pair (Figure 3.2D–E).24 However, in our current report owing to larger distance fluctuations, heterogeneity was commonly observed in various mutants.

For F101W, with the donor W101, we detected a dominant ET in 60 ps (95%), and a minor component in 7.2 ps (5%), each with β =0.94. Although W101 has a shorter edge- to-edge distance with flavin than W60 (Figure 3.2D), the FET rate is significantly slow here. Interestingly, W60 and W101 have different interactions with the FMN due to their locations. W60 is stacked with the FMN and is close to the o-xylene ring and the central pyrazine ring, while W101 is located towards the uracil group and is orthogonal to the

FMN (Figure 3.2E). Since the FMN* accepts an electron to the singly occupied HOMO, which is highly localized on o-xylene and pyrazine rings,67-69 ET is highly favorable with

W60, and hindered for W101, and thus the edge-to-edge distance is misleading in understanding ET tunneling in short range.

For I65W, we observed similar heterogeneity dynamics in 30 ps (65%), 60 ps (35

%) with β =0.79. For D62W, we observed a fast ET in 8.1 ps, β =0.70 (57%), and another significantly slow dynamic in 282 ps, β =0.57 (43%). The drastic differences in ET timescale and stretch indicate a high degree of heterogeneity, implicating strong fluctuations in the FMN–W62 pair. The MD data not only shows high oscillations in the distance but also conformational changes in the W62 (Figure 3.2D–E), leading to the complex ET behavior. 3FMN formation rates were subtracted from FMN* dynamics to calculate the ET rates in D62W and in mutants with slower charge separation. For, E66W charge separation occurs in and 40 ps (14%) and 320 ps (86%), β =0.72 with an FMN–W distance at 8.1 ±0.6 Å.

For G103W and A104W, we obtained single stretched ET decay in 780 ps, β=0.84 and 770 ps, β=0.78, respectively. The donors W103 and W104 are located in the α-helix

(Fig. 1A) and are bound to the rigid protein structure, resembling W60 in minor fluctuations (Figure 3.2D) and absence of distance–heterogeneity. However, G103W and

A104W have similar ET rates from different FMN–W distances in 7.0 and 9.6 Å, respectively. In A104W, the presence of residues 58–60 in between W104 and FMN

(Figure 3.2E) lowers the tunneling barrier (βT) and provides a facile route for the ET

70 reaction. For W103 and FMN, in the absence of such protein media HAB is higher, resulting in the relatively slow ET.

Slow FMN* dynamics were observed for L67W, R131W, and the ET inert mutant

W98F/Y60F/Y100F. For W60F/Y98F/Y100F, in absence of electron donors we observed an FMN* lifetime decay in 3.0 ns. Surprisingly, the dynamics is faster than the 6.0 ns decay in W60F/Y98F,20 possibly arising from changes in the local protein-water structure, with the Y100F mutation in the solvent–exposed flavin binding site. For L67W, we observed

ET dynamics in 106 ps, (β= 0.89) and a long FMN* decay in 3.0 ns from the two FMN–

W distances (Figure 3.2D). We observed identical decays for R131W and

W60F/Y98F/Y100F, indicating the FET is extremely slow or nonexistent at 18 Å.

3.3.3. Back electron transfer and photoinduced redox cycle.

Using time-resolved absorption spectroscopy, we systematically probed the ET reaction from 800–480 nm to characterize the complete ET reaction cycle (Figure 3.3–6,

B4). At the red wavelengths of 800 and 740 nm, we detected FMN* absorption to identify

FET rates. We obtained identical heterogeneities and similar FET rates compared to the upconversion results (Table 3.1). Minor differences in the timescales arise as the TA experiments detect more donor-acceptor configurations than the upconversion technique, also noted in other systems.30,71 For Y98F, we observed a single exponential FET in 0.4 ps

(= 1.0).30 Uniquely for W60, we detected red-shifted W+ absorption signals at 800–740 nm, possibly arising from the stacked nature of W60, also liable for the ultrafast FET rate.

For the mutants D95W, F101W, and I65W, we detected 2 stretched exponentials, confirming the solvation–coupled ET behavior and the fluctuating nature of the W donors in the loop region. For D62W, consistent with larger fluctuations, we observed a fast ET in 5.8 ps, =0.70 (56.5%) along with a slow decay in 294 ps, =0.57 (43.5%). 46

For E66W, in Figure 3.4, we observed FET dynamics in 32 ps (14%) and 375 ps

(86%), with =0.72. Due to the slower charge separation, FMN* significantly branches towards 3FMN formation and is also detected as a rise signal, leading to overall longer transients. In the absence of 3FMN decay dynamics in our experimental time–window, no

3FMN–ET is observed, and reportedly occur in sub–nanosecond timescales.72,73 Similarly, for G103W and A104W, we observed a single–stretched ET decay from the tightly held W donors in 740 ps, =0.89 and 845 ps, =0.89, respectively, For L67W, we detected a fast– stretched decay in 120 ps, =0.89 and the longer ET was not resolved.

To characterize BET dynamics, we tuned the TA probe to the 640–480 nm region and detected the charge transfer intermediate (W+), and ET products (FMN† and FMN). In our previous studies,30 we observed charge recombination in 1.6 ps, β =1.0 in Y98F. No stretched behavior was observed for ET reactions faster than solvation processes. In the

540–500 nm region, we detected the vibrationally excited BET product, FMN† with subsequent cooling dynamics in 3–4 ps.

Figure 3.3 shows the ultrafast transients for the mutant D95W. By knowing the heterogeneity ratio and the FET dynamics from 800 and 740 nm, we obtained BET dynamics in (46%) 1.7 ps and (54%) 2.8 ps, β =0.93. The smaller stretch than FET, results from reduced coupling with the environment due to faster BET rate, as was also observed in SQ flavodoxin system. Similarly, in Figure 3.6, we resolved charge recombination dynamics for F101W in (46%) 1.7 ps and (54%) 2.8 ps, β =0.93, and for I65W in (46%)

1.7 ps and (54%) 2.8 ps, β =0.93. For F101W with slower ET rates and similar cooling dynamics in ~4 ps, FMN† accumulation is lower, leading to the minor cooling signal.

Similar cooling dynamics should also be present in mutants with slower ET dynamics, but is not detected due to limited population of FMN†.

For D62W, in Figure 3.5 all transients show a fast decay in a few picoseconds (Inset

A) and a slower component in hundreds of picoseconds from different FMN-W distances.

At 630 nm, we obtained BET dynamics in (57%) 2.1 ps, β =0.93 and (43%) 8.3 ps, β =0.66, and observed longer 3FMN dynamics, due to slow charge separation dynamics in 294 ps.

No 3FMN was observed at 800 or 480 nm due to relatively lower 3FMN absorption (Fig.

B3). The faster BET processes indicate low stability in the charge–separated state and ET is not favored. For E66W, we detected BET in 3.8 ps (14%) and 8.0 ps (86%). With relatively faster BET rates, and smaller W+ signal due to 3FMN detection (Figure 3.4), we could not resolve stretch values for charge recombination, and used β =1.0. A similar pattern in transients was observed For G103W and A104W in Figure 3.6, with a single

BET dynamic in 48 ps (β =1.0) and 130 ps (β =1.0), respectively from mutations in the α– helix. For L67W (Figure B4), charge recombination dynamics are not resolved.

3.3.4. Dynamic fluctuations and implications in electron transfer dynamics.

We further examined the relation between heterogenous ET behavior and the dynamic nature of the wobbling W donors.74,75 Assuming an W donor with oscillating distances, 푟(푡) = 푟푒푞– 훥푟. sin 휔푡 about an equilibrium distance 푟푒푞, with the angular wobbling frequency 휔푤표푏 and amplitude 훥푟, we obtained the distance-dependent ET rate

훥푟.훽푇.sin(휔푡) 75 as 푘퐸푇(푟,푡) = 푘푒푞. 푒 , where 푘푒푞 is the ET rate at the equilibrium distance. We also assume distribution in initial position and determined the ensemble averaged ET

48 dynamics by adding dynamics from 9 starting positions between – 훥푟 and 훥푟, each weighted by their Boltzmann probability (See Appendix B). By fitting the transients, we determined τET, β, τwob, and βT for the mutants.

Specifically for F101 in Fig. 3.7, the small fluctuations (r = 0.40 Å), leads to the minor heterogeneity. At short distance of 3.2 Å, minor fluctuations in W101 may lead to orbital interactions leading to the minor heterogeneity dynamics in 6.4 ps. Also, for F101W and E65W the β values are similar to the heterogeneity model. For D95W, and E66W, in

Figure 3.7, we obtained fluctuations at 0.69–0.81 Å, with charge separation and recombination dynamics in 5.0 ps, β=0.90, and 1.7 ps, β=0.98, respectively. For D62W, we could not fit the dynamics with the wobbling model, possibly due two local conformations of W62, rather than a oscillating behavior (Figure B2).

Figure 3.8 shows the dynamic processes involved in photoinduced ET processes in flavodoxin. The excitation energy triggers ET processes along with local water– protein relaxations. With overlapping ET and solvation timescales, the excited state

FMN* is in nonequilibrium leading to a solvation–heterogeneity in the ET rates and the stretched behavior. The further observed heterogeneity in the ET reactions arising from residue fluctuations in hundreds of picoseconds. Here, the changes the distances with the electronic coupling, produces a further heterogeneity. Typically, fluctuations smaller than 0.4 Å, the distance heterogeneity is minor, and the ET dynamics can be expressed as a single ET process as was observed in mutants Y98F, G103W, and A104W where the donors were stable with stacking interactions or were part of helical structure. For mutants D95W, F101W, I65W and E66W, we observed Δr at 0.40–0.80 Å and alters ET 49 behavior and should be considered in all ET behavior in proteins. Empirically, ET is expressed as, and at longer distances, a change in (βT) can significantly affect ET dynamics. In proteins, the electronic coupling is between FMN and W is not only determined by the distances but also the local protein structure and can vary significantly within a protein.

3.4 Conclusion

We investigated distance–dependent ET behavior by resolving charge separation and recombination dynamics in 10 flavodoxin mutants using fs–resolved spectroscopy and

MD simulation. By varying the FMN–W distances from 3.2 to 18 Å, we observed wide range of FET dynamics in 0.3–845 ps, β=1.0–0.72, generally slowing down with increasing distances. For the intrinsic donor W60, we observed single–exponential FET dynamics in

0.4 ps without solvation coupling from a stacked FMN–W60 pair at short distance of 3.4

Å. For W donors located in flexible loops, we observed heterogeneous solvation–coupled

FET as a sum of two stretched exponentials from large fluctuations in the FMN–W distances, also obtained in our MD data. By modelling ET with wobbling W donors, we obtained unified FET dynamics in 4.7–330 ps, β=0.72–0.94 and estimated fluctuations in

0.40–0.81 Å. For the special mutant D62W, we observed extremely different FET components in 5.8 ps (β=0.70), and 294 ps (β=0.57), possibly from two locally stabilized conformations. For W donors in α-helix, with smaller fluctuations in 0.33–0.39 Å, we observed FET in 746–840 ps, β=0.80–0.89 without heterogeneity. For FMN–W distances longer than 10 Å, charge separation was slower than 3.0 ns, and were not resolved. For

FET slower 200 ps, we detected significant 3FMN formation, from longer dynamics in transient absorption signals. We similarly resolved BET dynamics in 0.9–130 ps, β=1.0–

0.66 with identical heterogeneities and distance–fluctuations. The faster BET dynamics and higher stretch values arises from insignificant stabilization of charge–transferred state and smaller mixing with solvation processes. Significantly, we dissected cooling dynamics in ~4 ps for faster ET reactions, and although not measurable should be present for all slow

BET processes.

Our extensive ET analyses show the role of electronic coupling, solvation timescales, orbital orientations, local protein fluctuations, and tunneling pathways in ET mechanism. For ET reactions faster than environmental relaxations, ET follows single– exponential dynamics, but with overlapping timescales, the processes couple leading to the widely observed stretched behavior. The observed heterogeneity in ET processes and fluctuating edge–to–edge distances clearly demonstrate the flexible nature of donor– acceptor configurations. With larger fluctuations, especially above 0.4 Å in the loop region, the varying distances significantly change the electronic coupling, leading to distribution of ET rates and the heterogeneity. This heterogeneity is often misinterpreted in protein dynamics as a static phenomenon and should be carefully analyzed. With localized orbitals in FMN, ET at shorter distances is dictated by donor–acceptor orientations and “edge–to– edge separation” may not represent accurate tunneling distance. Over longer distances, ET with slower timescales is inefficient, and dissipates excitation energy by futile pathways such as 3FMN formation, and is thus widely replaced by sequential or hopping mechanisms. By connecting dynamic donor–acceptor properties and ET behavior, these

51 findings a provide deeper understanding of the molecular basis of the photo-induced electron transfer mechanism in proteins.

Table 3.1: Timescales of distance–dependent ET reactions in flavodoxin obtained from the heterogeneity fittinga.

b c c Heterogeneity FET (TF ) FET (TA ) BET (TA ) Mutant (%) <τ> β <τ> β <τ> β Y98Fd 100 0.3 1.00 0.4 1.0 1.6 1.0

45.7 2.2 2.0 1.7 D95W 0.84 0.84 0.93 54.3 10 10 2.8

4.8 7.2 6.4 0.9 F101W 0.94 0.94 0.98 95.2 60 60 8.0

64.9 30 38 3.0 I65W 0.79 0.79 0.96 35.1 60 73 3.3

56.5 8.1 0.70 5.8 0.70 2.1 0.81 D62W 43.5 282 0.57 294 0.57 8.3 0.66

14.1 40 32 3.8 E66Wf 0.72 0.72 ~1 85.9 321 375 8.0 G103W f 100 780 0.87 740 0.89 48 ~1 A104W f 100 848 0.78 845 0.78 130 ~1 12.2 106 0.89 120 0.89 ND - L67We 87.8 ND - ND - ND - aTime constants are in units of ps. Triplet formation dynamics were subtracted to calculate ET rates. bObtained from transient fluorescence dynamics gated at 538 nm. cObtained from transient absorption data. dET dynamics in W60 were observed to be single exponential (23). eFMN* dynamics were identical to the ET–inert mutant W60F/Y98F/Y100F and ET rates were not determined (ND). fBET stretch could not be resolved and dynamics were assumed to be single–exponential.

Table 3.2 Fitting parameters of distance–dependent ET reactions in flavodoxin obtained from the wobbling model.a

FET BET Mutant Δr τwob βT <τ> β <τ> β D95W 0.69 60 1.21 4.7 0.89 1.7 0.95 F101W 0.40 80 1.24 58 0.94 8.4 0.98 I65W 0.50 200 1.30 63 0.80 2.3 0.96 E66WW101b 0.81 200 1.14 333 0.70 6.2 ~1.0 a -1 b Time constants are in units of ps, Δr is in Å, βT is in Å . Triplet formation dynamics were subtracted to calculate ET rates. BET stretch could not be resolved and dynamics were assumed to be single–exponential.

Figure 3.1 Structural and spectral properties of oxidized flavodoxin. (A) X-ray crystal structure of oxidized flavodoxin (PDB: 2FX2) showing the flavin cofactor (purple) along with W60, and various mutation sites (yellow) designed to vary ET tunneling. (B) The absorption spectrum (red) of WT flavodoxin, and emission spectrum (blue) of the ET inert mutant W60F/Y98F. The top panel displays the absorption ranges of the different species involved in ET reactions. Arrows indicate the pump and probe wavelengths used in time-resolved experiments.

Figure 3.2 Forward ET reaction dynamics and FMN–W configurations.

(A-C) Normalized femtosecond-resolved fluorescence transients of flavodoxin mutants probed at 538 nm in various timescales. (D) Fluctuations in shortest edge-to-edge distance between FMN and tryptophan ring atoms from MD simulations. Note correlation of distance fluctuations and heterogeneity ET parameters (τ1, τ2, β1, β2). (E) Unique FMN–W configurations and tunneling pathways that produce anomalous ET behavior.

Figure 3.3 Normalized fs-resolved absorption transients of the mutant D95W. Comparison of signals in Inset A shows longer transients at 630 and 515 nm from additional W+ and/or FMN† absorption. The 480 nm transient is negative due to the dominant FMN recovery signal. Insets B and C show the deconvolution of the transients into constitutive species at 630 and 515 nm, respectively.

Figure 3.4 Normalized fs-resolved absorption transients of the mutant E66W.

Note longer signals at all wavelengths due to 3FMN detection from slow FET dynamics in Inset A. Insets B and C show the deconvolution of the transients into constitutive species at 630 and 515 nm, respectively.

Figure 3.5 Normalized fs-resolved absorption transients of the mutant D62W.

Inset A shows faster heterogenous ET dynamics in short timescale. The 630 nm transient is longer due to 3FMN detection. Insets B and C show the deconvolution of the transients into constitutive species at 630 and 515 nm, respectively.

Figure 3.6 Normalized fs-resolved transients of various flavodoxin mutants.

(A) F101W, (B) I65W, (C) G103W, and (D) A104W with different FMN–W distances. (A–B) Note similar pattern in transients with ET in comparable timescales. (C–D) Differences in 630 nm signal arise from different BET rates. Also, longer transients are obtained at all wavelengths from 3FMN detection.

Figure 3.7: Nonequilibrium ET dynamics modelled with a wobbling W donor.

(A–B, D–E) Fitting comparison of heterogeneity vs wobbling model of femtosecond- resolved absorption transients in mutants D95W and I65W. (C, F) Deconvolution of the transients into FMN* and W+ signals to identify ET dynamics, wobbling frequency, and tunneling barrier.

Figure 3.8: MD snapshot displaying the dynamic processes in intraprotein ET.

Shown are FET reactions (τFET = 0.3–845 ps, β=1.0–0.57), subsequent BET process (τBET=

0.9–130 ps, β=1.0–0.66), local water/protein relaxations (τSOL =1 ps (53%) 25 ps (26 %),

670 ps (21 %)), and wobbling W donors (τwob= 60–200 ps) that lead to heterogeneous stretched ET dynamics.

Chapter 4. Dynamics and Mechanism of Light–Induced Electron Transfer Reactions in OsHAL3.

4.1. Introduction

OsHAL3 functions as a blue light receptor and regulates plant growth and early flowering in rice by forming a complex with the proteins HIP1 and HD1, respectively.26,27

Fig 4.1A shows the crystal structure of dark–adapted OsHAL3 as a trimer with a partially solvent–exposed flavin molecule (FMN) at the interface. The FMN is noncovalently bonded to the protein via a series of H-bonds with residues G29, V31, T54, S107, A108,

T110 and A141. The H-bonding is critical to the stability of the trimer as the mutant

S107P/A108P/N109Y/T110P is a monomer and only present in the apo form.26 At the active site, electron donors, W79 and W82 are present at 3.0 Å and 5.2 Å, respectively, indicating possible photoinduced ET reactions (Figure 4.1B). The neighboring charged residues, H59 and H91 tune the redox environment and may regulate ET dynamics.

Interestingly, C119 and C125 are present in the active site, without the disulfide bond, contrary to the homologous cysteines in AtHAL3.33 However, a C119–C125 disulfide bond is detected in light-exposed crystal structure of OsHAL3, together with a distorted trimer architecture. Blue light excitation of OsHAL3 possibly initiates ET reactions with nearby

W donors that leads to the disulfide bond formation and destabilizes the trimer. However, the mechanism of ET processes is unknown. 63

In this chapter, we analyzed the role of key residues in the active site and resolved the mechanism of light induced ET reactions in OsHAL3. Through spectroscopic analysis, we observed blue–light induced photoreduction of flavin in the presence of W79 and W82.

Using ultrafast spectroscopy, we characterized the complete ET cycle involving elementary steps of ultrafast FET, BET and cooling dynamics with various mutations. We analyzed

ET rates of Wild-type OsHAL3, and that of individual W79 and W82 ET pathways from

W82F and W79F mutants respectively. We observed a charge-transferred state as a potential precursor to downstream signaling. This study comprehensively determines the mechanism of ET reactions, as the primary photochemistry event that initiates signal transduction.

4.2. Materials and methods

4.2.1. Protein purification

The OsHAL3 plasmid was transformed in E-coli (BL21-DE3) and grown in Luria

Broth at 37 °C to OD 0.6 and induced overnight. The protein was extracted by lysing the cell and purifying the crude extract by TALON–affinity and DEAE sepharose columns.

The detailed purification steps are mentioned in Appendix C. The proteins were dialyzed in presence of 20 mM βME to prevent aggregation from the surface-exposed C176S.

Various mutants were prepared using site-directed mutagenesis with designed primers and further purified using identical protocols. All protein samples were further characterized by size-exclusion chromatography to determine purity and trimer formation. For ultrafast experiments, samples were concentrated to 175 µM, and dithiothreitol (dTT) was added to a final concentration of 20 mM. 64

4.2.2. Femtosecond–resolved spectroscopy

All fs–resolved measurements were carried out using fluorescence upconversion and transient absorption methods.36 Briefly, the pump wavelength was set at 480 nm and the pulse energy was attenuated to 70 nJ before being focused onto the sample. For the fluorescence upconversion experiments, the emission at 540 nm was gated by a 800 nm laser beam in a 0.5 mm thick β-barium borate crystal (BBO, type I). For TA experiments, the probe pulses were tuned between 500 and 800 nm via an optical parametric amplifier

(TOPAS, Spectra-Physics). The instrument responses are 360 fs and 150 fs for fluorescence and transient absorption detection, respectively. All experiments were conducted at the magic angle (54.7°) configuration. To prevent heating and photobleaching, the sample was kept stirring in 5 mm quartz cells thickness during laser irradiation. Non-linear artefacts at time zero, wherever present were subtracted throughout data analysis.

4.3. Results and discussions

4.3.1. OsHAL3: Spectroscopic properties and photoactivity

Figure 4.2A shows the steady-state spectra of WT OsHAL3 and mutant proteins.

The absorption curves (left) resemble the characteristic flavin spectrum with the S1S0 transition peaked at 455 nm together with a shoulder band at 483 nm, and the S2S0 transition at 340 nm.20,31 The pump wavelength was set at 480 nm, near the red edge of the absorption band to minimize vibrational relaxations in the excited state. With the exception of the mutant H91F, which is red–shifted by 4 nm with the removal of histidine at 3 Å 65 away, other mutants closely follow the S0S1 spectral shape implying negligible electrostatic changes in the flavin environment through mutations. We observed a long tail in the 520–700 nm for the ET proteins, due to possible charge character state involving the tryptophan donors at van der Waals distances.30 Variations observed below 400 nm are due to minor HQ formation in the samples during protein purification.20 For W79F we observed

~12% SQ content from the absorption in the 500-700 nm region. The emission spectrum

(right) was obtained from the ET inert mutant W79F/W82F in the absence of electron donors. The emission is peaked at 535 nm, and deviates from a non-lognormal emission profile with a flattened top, representing a heterogeneous electrostatic environment within the binding site, also observed in other flavoproteins.50,51 No fluorescence spectra were detected for in the presence of either W donor, indicating efficient quenching of the excited flavin.

To probe the nature of photoactivity in OsHAL3, we irradiated protein samples with blue light in the presence of 20 mM dTT. In presence of oxygen, no change in the

FMN absorption was observed. However under anaerobic conditions, we detected gradual photoreduction with blue light exposure (Figure 4.2 B-D).18 Figure 4.2B shows the progressive decrease of FMN spectra in WT. With favoring reduction potentials, the excited flavin (FMN*) accepts an electron from redox active W79 and W82, as indicated previously by the fluorescence quenching, and generates FMN˙- and W+ radicals. The charge separated pair can form FMN and W by a back electron transfer. Alternatively, dTT can react with the surface–exposed W+ and produce W and FMN˙-, responsible for the steady decrease in FMN absorption. FMN˙- can then form FMNH˙ with a proton transfer

66 reaction, or under continuous light exposure, produce FMNH– via repeated electron transfers.

In the WT photoreduction, no FMNH˙ absorption was detected in the 550-700 nm range, and forms HQ directly. Thus in WT, a dynamic equilibrium exists between FMN and FMNH-, and the SQ states are unfavored in OsHAL3. Uniquely for C119S, we observed a well-defined formation of SQ (Figure 4.2C), with a slow photoreduction rate, and the flavin equilibrium is shifted towards the OX and SQ states through changes in the flavin environment. Figure 4.2D shows the time rate of photoreduction. WT, W79F and

W82F have similar rates photoreduction rates indicating similar electron donating nature for both donors. The faster rates in H91F arise from changes in redox environment. H91F mutation enhances the photoreduction rate by favoring FMNH– formation. For

W79F/W82F, no FMN-W ET is produced, but the smaller F79/F82 residues shield FMN from the solvent poorly, and flavin is directly reduced by dTT. Without the possibility of

BET reaction with W donors, the rate observed here is faster.

Figure 4.2E shows the exclusion chromatography profile obtained for WT OsHAL3 and all mutants, and agrees well with the theoretical elution volume for the trimer. The trimer dissociation would terminate the FMN and W79/W82 interaction, from the neighboring monomer, and increase the fluorescence quantum yield. However, samples after photoreduction did not show any emission properties, and thus did not fully dissociate during blue light illumination. It is possible that OsHAL3 may only undergo minor conformational changes for biological signaling and dissociation to monomers is not required.

4.3.2. Solvation dynamics and dynamic flavin environment

To characterize the nature of environmental relaxations at the partially-exposed active site, we characterized the solvation processes in the ET-inert mutant W79F/W82F using intrinsic FMN, with a long lifetime dynamic in 180 ps (45%) and 560 ps (55%), as the probe (Figure C.1). Due to the non-lognormal fluorescence profile of the flavin (Figure

4.2A), we directly measured the time resolved emission spectra at various wavelengths to identify the solvation timescales. Figure 4.3A-B shows the reconstructed emission spectra, and representative snapshots at various time delays. We observed a gradual change in the shape of emission curve, with shrinking in the blue side and broadening at the red side, commonly observed in flavoproteins. We calculated the averaged frequencies and determined the energy relaxation (∆퐸(푡)) (Figure 4.3C). We observed three distinct relaxation timescales in 1.9 ps (20%), 16 ps (42%), and 480 ps (37%) corresponding to water motions, protein coupled water relaxations and collective protein motions in the flexible loops, respectively, as reported in flavodoxin. The dynamic energy relaxation can

푡 푡 푡 − − −    be obtained as ∆퐸(푡) = ∆퐸1 푒 1 + ∆퐸2 푒 2 + ∆퐸3 푒 3 + 퐸∞ ; ∆퐸1 + ∆퐸2 + ∆퐸3 +

-1 퐸∞ = 0, and obtained E1, E2 and E3 in 44, 92, 81 cm , respectively. The total energy relaxation in 217 cm-1 agrees well with OX flavodoxin and OX photolyase and describes a largely solvated protein environment. Thus, ET reactions in OsHAL3 may potentially couple with local relaxations if the timescales overlap. Similar solvation components were also observed in the mutant W79F/W82F/H91F mutant, with a longer FMN* lifetime in

1.2 ns (Figure C.2).

4.3.3. Forward electron transfer and dynamic heterogeneity

To identify FET rates, we directly measured the FMN* emission dynamics by the upconversion method. Figure 4.4 shows the fluorescence transients gated at 540 nm from single donors W79, W82 and double donors W79/W82 in WT and various mutants. For

W82F, we observed stretched ET dynamics in 1.1 ps, β=0.92 (87%) and 4.0 ps, β=~1

(12%). The minor heterogeneity dynamics arises from sample impurity during purification and is present in all ET dynamics, and the dominant component is considered as the principal ET pathway in OsHAL3. The stretch is observed due to solvation-coupled ET behavior arising from comparable timescales. For the heterogeneity component, the stretch parameter could not be resolved due to smaller amplitude and β is assumed as 1. Similarly, in W79F we observed FET dynamics in 0.44 ps, β=1.0 (89%) and 2.0 ps, β=~1 (11%).

For WT, we observed faster charge separation dynamics in 0.38 ps, β=1.0 (90%) and 1.3 ps, β=~1 (10%) with the addition of ET rates from both W79 and W82 donors.

Also, with the faster charge separation rate, W82 is the major electron donor with the higher branching ratio. No stretch is observed for dual donors of W79/W82 as FET reaction is faster than the solvation processes and occur in a static environment. For C119S, we detected FET in 0.42 ps, β=1.0 (97%) and 3.5 ps, β=~1 (3%). Nearly identical ET rates with WT reveals similar donor-acceptor configurations and that mutation of a single cysteine does not affect trimer stability, consistent with the absence of C119-C125 disulfide bond. For AtHAL3, mutating the homolog C118, may affect the protein folding and hence alter ET rates with W donors from the neighboring monomer. In H91F, we observed the fastest FET in 0.32 ps, β=1.0 (94 %) and 1.4 ps, β=~1 (6%). At close proximity, H91 69 hinders ET with electronic repulsion between the aromatic FMN. In summary, FET dynamics in OsHAL3 is ultrafast in 0.38–5.2 ps, β=0.92-1.0, necessary for efficient conversion of solar energy to chemical energy as a photoreceptor.

4.3.4. Vibrationally-coupled charge recombination and ground-state cooling dynamics.

We switched to transient absorption technique to characterize the dynamics of the photoinduced ET cycle (Figures 4.5-8). By systematically tuning the probe wavelengths from 800 nm to 500 nm, we characterized dynamics of all reactants, charged-intermediates and ET products. Specifically, at 800, 580 and 540, nm we detected FMN* and W+. By knowing the FET dynamics from fluorescence studies, we resolved BET rates from W+ signals. At slightly longer wavelengths of FMN absorption spectrum, in 510 and 500 nm, we detected FMN† dynamics formed from vibrationally excited BET. At 500 nm, we additionally detected FMN recovery signals and characterized the complete redox cycle.

We first characterized FMN* dynamics in the ET-inert mutant W79F/W82F, and observed a lifetime in 180 ps (45%) and 560 ps (55%) at all wavelengths (Figure C.1). Additionally, at 500-580 nm, we detected minor solvation components, and 3FMN absorption from longer absorption transients.

Figure 4.5 shows the time-resolved absorption transients for W82F with the donor

W79. At 800 nm (Inset A), we detected a stretched FMN* decay in 1.2 ps, β=0.92 (87%) and 4 ps, β=~1 (12.7%), and a minor W79+ rise-decay signal with BET dynamics in 0.6 ps,

β=0.97 (87%) and 1.4 ps, β=~1 (13%). For W79+ dynamics, we observed identical heterogeneities, and an apparent reverse kinetics, with BET being faster than the charge 70 separation process. At 580 and 540 nm, we detected an early negative signal from the stimulated FMN* emission, and an overall positive signal at later times from strong absorption of W79+ (Inset C). At 510 and 500 nm, we observed significantly longer dynamics from FMN† absorption with cooling dynamics in 3.5-4.0 ps, similar to the flavodoxin ET cycle.

For W79F, in Figure 4.6, we similarly resolved FET in 0.41 ps, β=1.0 (89%) and

2.0 ps, β=~1 (11%), and BET in 4.0 ps, β=0.96 (89%) and 19 ps, β=~1 (11%) We additionally detected a longer ET signal arising from the 12% SQ content in the sample

(Figure C4). With the faster FET rate, and longer BET timescale, W82+ forms a longer- lived intermediate, forms the major ET pathway and putatively causes the downstream signaling. With the observed BET timescale being similar to the cooling dynamics in 4-5 ps, FMN˙– may have additional vibrational relaxation dynamics, but were not detected in our transient–absorption experiments in the visible range.76 The vibrational relaxation can also be delocalized in the charge-transferred state with similar relaxations in W82+,77-79 but such dynamics could not be resolved in our data analyses.

Surprisingly with the shorter FMN-W distance, we observed a slower FET rate for

W79, and an unusually faster BET rate. To determine if this unique ET behavior of is due to H59 at a close distance of 3.0 Å from W79, we measured ET dynamics of the mutant

H59F/W82F (Figure C.4). No difference in ET dynamics were observed, and H59 is not positively charged at the buffer pH of 8.0. The W79 ET behavior may be due to its orientation with the FMN group, or via a tunneling mechanism through methyl group in the isoalloxazine ring.

For WT, in Figure 4.7, we observed faster FET dynamics in in 0.38 ps, β=1.0 (90%) and 1.3 ps, β=~1 (10%) from dual donors W79/W82. We deconvoluted the W79+ and W82+ signals (Inset A-D) and observed BET in 0.60 ps, β=1.0 (90%) and 1.4 ps, β=~1 (10%), and 4.3 ps, β=0.95 (90%) and 15 ps, β=~1, respectively for the donors. We did not detect an electron exchange reaction between the donors. With a longer distance of 5.0 Å between

W79 and W82, exchange reactions are typically slower without favorable driving forces, and cannot compete with the faster BET reactions. We further detected cooling dynamics in 5.0-5.5 ps as the final step in the ET process (Inset C)

Fig 4.8 shows the absorption transients from dual donors W79/W82 in mutants

C119S and H91F. For C119S, we detected FET in 0.42 ps, β=1.0 (97%) and 3.6 ps, β=~1

(3%), similar to the WT dynamics. We further resolved W79+ BET dynamics in 0.50 ps,

β=1.0 (97%) and 1.4 ps, β=~1 (3%), and W82+ BET in 5.3 ps, β=~0.95 (97%) and 19 ps,

β=~1 (3%), The comparable ET timescales, denote a similar protein environment, and

C119 is not critical for trimer stability without the disulfide bond formation. For H91F, we observed faster ET dynamics in 0.37 ps, β=1.0 (94%) and 1.4 ps, β=~1 (6%), from redox tuning in the flavin environment. We observed W79+ BET in 0.50 ps β=1.0 (94%) and 1.4 ps, β=~1 (6%), while BET with W82 occurred in 5.3 ps, β=0.93 (94%) and 18 ps, β=~1

(6%), and subsequent cooling dynamics in 3.5-4.0 ps. Due to the red-shifted flavin absorption in H91F, we detected FMN recovery dynamics in both 500 and 510 nm transients. The overall ET dynamics in H91F and C119S are similar to WT ET behavior and no key role in signal transduction is discovered for these residues.

4.3.4. Summary

We report here the systematic characterization of the of blue light induced photoactivity of

OsHAL3 with femtosecond-resolved spectroscopy and mutagenesis. We observed photoreduction of FMN in presence if electron donors W79 and W82 by ET reactions in the trimer with blue light exposure. In the active site, we observed a water exposed flavin environment with solvation dynamics in 1.9 ps (20%), 16 ps (42%), and 480 ps (37%), and total energy relaxation in 217 cm-1. We identified ultrafast ET in 1.2 ps, β=0.92 and 0.44 ps, β=1.0 from the donors W79 and W82, respectively. For WT and mutants with dual donors of W79/W82, we observed FET in 0.37-0.42 ps, β=1.0, that efficiently generates charge-transferred intermediates for signal transduction. For all mutants, W79+ BET occurs in 0.50–0.60 ps, β=1.0-0.97 while W82+ charge recombination dynamics is in 4.0–5.0 ps,

β=0.93–0.96. The distinct differences in ET timescales of the donors possibly arise through unique tunneling mechanisms or orbital interactions at short distances. Due to ultrafast nature of the ET reactions, the observed stretch in ET reactions is minor from reduced coupling with the environment. The comparable ET dynamics in WT and C119S, consistent with the absence of C119-C125 disulfide bond is significant, as the disulfide bond formation could be a necessary step for photoreceptor function, and explains lack of blue light activity in AtHAL3. With the higher branching ratio from faster charge separation and long-lived intermediate from slower charge recombination, we identify

W82 as the critical residue required for the light harnessing mechanism of OsHAL3, and further phenotype and structural studies are required to understand the connection of ET dynamics and downstream signaling events.

Table 4.1 Dynamics of photoinduced electron transfer cycle in OsHAL3a

b b c c Mutant Donor(s) Heterogeneity (%) <τFET> βFET <τET> βFET <τBET> βBET τcooling s 87.3 1.1 0.92 1.2 0.92 0.60 0.97 W82F W79 3.5-4.0 12.7 4.0 ~1 4.0 ~1 1.4 ~1 89.3 0.44 1.0 0.41 1.0 4.0 0.96 W79F W82 4.0-5.0 10.7 2.0 ~1 2.0 ~1 19 ~1 89.9 0.38 1.0 0.38 1.0 0.60, 4.3 1.0,0.95 WT W79, W82 5.0-5.5 10.1 1.3 ~1 1.3 ~1 1.4, 15 ~1, ~1 97.2 0.42 1.0 0.42 1.0 0.50, 5.3 1.0,0.95 C119S W79, W82 5.0-5.5 2.8 3.5 ~1 3.6 ~1 1.4, 19 ~1, ~1

94.1 0.32 1.0 0.37 1.0 0.50, 5.3 1.0,0.93 H91F W79, W82 3.5-4.0 5.9 1.4 ~1 1.4 ~1 1.2, 18 ~1, ~1   a  1 Time constants are in units of ps. All average ET times are calculated using <τ> =   . Stretch values for     minor heterogeneity dynamics could not be resolved due to smaller amplitude. cCalculated from upconversion dynamics gated at 540 nm. cObtained from transient absorption detection.

Figure 4.1 Structure of OsHAL3 obtained from 3 ns MD simulation.

(A) Trimeric structure, showing non-covalently bonded FMN at the interfaces, and water molecules within 5 Å of the protein. (B) Close up view of active site showing key residues with electron donors of W79 and W82, and cysteines C119 and C125 without disulfide bond formation. Note presence of water molecules in the flavin environment.

Figure 4.2 Spectroscopic characterization of OsHAL3

(A) (left) Absorption of WT OsHAL3 and mutants containing Flavin in oxidized form. (right) Emission spectra of (W79F/W82F). Note presence of SQ in W79F. (B) Photoreduction dynamics of WT OsHAL3 with blue light, in anaerobic conditions, and in the presence of 20 mM dTT. Inset shows changes in spectrum compared to dark state. Note zero changes in 520-720 nm from absence of FMNH˙ intermediates. (C) Photoreduction profile in C119S mutant with unique formation of FMNH˙. (D) Changes in absorption at 480 nm with time for various mutants. (E) Size exclusion chromatography of WT OsHAL3 trimers.

Figure 4.3: Solvation dynamics of the ET inert mutant W79F/W82F

(A) Three-dimensional reconstruction of fs-resolved emission spectra of FMN*. Spectra beyond 500–600 nm range were obtained from lognormal fitting. (B) Snapshots of fs- resolved spectra at various time delays along with steady-state emission with changes in the spectral shape. (C) Solvation responses at the active site obtained from averaged- frequency changes with time. Inset shows environmental relaxations in early timescales.

Figure 4.3 continued

Figure 4.4: Normalized fs-resolved fluorescence transients of OsHAL3 mutants.

Note minor component at longer timescales from sample heterogeneity. Faster dynamics of W79 indicate better electron donor properties in comparison to W82. Shorter dynamics observed in dual donors of W79/W82 with addition of FET rates from both donors.

Figure 4.5: Normalized fs-resolved absorption transients of the mutant W82F.

Inset A shows all transients jointly for quick comparison. Insets B, C and D show the deconvolution of the 800, 580 and 510 nm transients into constitutive species. Note 540 and 580 nm transients are predominantly negative owing to fast charge recombination charge separated state. Longer transients at blue sides are obtained from detection of long time FMN† dynamics.

Figure 4.5 continued

Figure 4.6 Normalized femtosecond-resolved absorption transients of mutant W79F.

Inset A displays gradual slower dynamics in smaller wavelength probes as a result of detecting. Note small signal from SQ ET dynamics resulting in 800 nm dynamics. Insets B, C and D show the deconvolution of the transients into constitutive species. All dashed lines represent 10.7% heterogeneity. Note distinct change in 540 and 580 nm transients compared to W82F. The signal recovers from the negative stimulated emission owing to long lived charge separated state. Similar slow behavior at blue sides is observed recurring feature from hot BET product formation.

Figure 4.6 continued

Figure 4.7 Normalized fs-resolved absorption transients of WT OsHAL3.

The comprehensive WT dynamics with a similar pattern of W79F. Note small all ET dynamics resulting from dual Trp donors. Insets A, B and C show the deconvolution of the 800, 580 and 510 nm transients into constitutive species. Note the presence of dual ET pathways with majority by W82 owing to faster FET and slower BET timescales. The signal recovers from the negative stimulated emission owing to long lived charge separated state. Similar slow behavior at blue sides is observed recurring feature from hot BET product formation

Figure 4.7 continued

Figure 4.8 Normalized fs-resolved absorption transients of mutants C119S and H91F. Note linear time axis before 2 ps and log scale thereafter. The C119S dynamics with a similar pattern of WT. Note H91F having early negative signal at 510 nm as we detect FMNH recovery.

Figure 4.9 Summary of photoinduced electron transfer cycle in OsHAl3

Ultrafast ET pathways of donors W79 and W82 for blue-light perception with forward ET, back ET and cooling processes. Note parallel and perpendicular orientations with flavin for W79 and W82 respectively. The charge recombination dynamics for FMN– ...W82+ overlaps with FMN cooling dynamics and vibrational relaxations may occur in the charge separated state.

Bibliography

(1) Möglich, A.; Yang, X.; Ayers, R. A.; Moffat, K. Annual Review of Plant Biology

2010, 61 (1), 21.

(2) Van Der Horst, M. A.; Hellingwerf, K. J. Accounts of Chemical Research 2004, 37

(1), 13.

(3) Perman, B.; Šrajer, V.; Ren, Z.; Teng, T. Y.; Pradervand, C.; Ursby, T.; Bourgeois,

D.; Schotte, F.; Wulff, M.; Kort, R.; Hellingwerf, K.; Moffat, K. Science 1998, 279

(5358), 1946.

(4) Rockwell, N. C.; Su, Y.-S.; Lagarias, J. C. Annual Review of Plant Biology 2006,

57 (1), 837.

(5) Quail, P. H.; Boylan, M. T.; Parks, B. M.; Short, T. W.; Xu, Y.; Wagner, D. Science

(New York, N.Y.) 1995, 268 (5211), 675.

(6) Gauden, M.; van Stokkum, I. H.; Key, J. M.; Luhrs, Dc.; van Grondelle, R.;

Hegemann, P.; Kennis, J. T. Proceedings of the National Academy of Sciences 2006,

103 (29), 10895.

(7) Gauden, M.; van Stokkum, I. H. M.; Key, J. M.; Lührs, D. C.; van Grondelle, R.;

Hegemann, P.; Kennis, J. T. M. Proceedings of the National Academy of Sciences

of the United States of America 2006, 103 (29), 10895.

(8) Ozturk, N.; Selby, C. P.; Annayev, Y.; Zhong, D.; Sancar, A. Proceedings of the

National Academy of Sciences 2011, 108 (2), 516.

(9) Liu, H.; Liu, B.; Zhao, C.; Pepper, M.; Lin, C. Trends in plant science 2011, 16 (12),

684.

(10) Zhang, M.; Wang, L.; Shu, S.; Sancar, A.; Zhong, D. Science 2016, 354 (6309), 209.

(11) Zhong, D. Annu. Rev. Phys. Chem 2015, 66, 691.

(12) Li, J.; Liu, Z.; Tan, C.; Guo, X.; Wang, L.; Sancar, A.; Zhong, D. Nature 2010, 466

(7308), 887.

(13) Nagel, G.; Ollig, D.; Fuhrmann, M.; Kateriya, S.; Musti, A. M.; Bamberg, E.;

Hegemann, P. Science 2002, 296 (5577), 2395.

(14) Zhang, M.; Wang, L.; Zhong, D. Archives of Biochemistry and Biophysics 2017,

632, 1.

(15) Tan, C.; Liu, Z.; Li, J.; Guo, X.; Wang, L.; Sancar, A.; Zhong, D. Nature

Communications 2015, 6, 7302.

(16) Liu, Z.; Tan, C.; Guo, X.; Li, J.; Wang, L.; Zhong, D. Journal of Physical Chemistry

Letters 2014, 5 (5), 820.

(17) Tenboer, J.; Basu, S.; Zatsepin, N.; Pande, K.; Milathianaki, D.; Frank, M.; Hunter,

M.; Boutet, S.; Williams, G. J.; Koglin, J. E.; Oberthuer, D.; Heymann, M.; Kupitz,

C.; Conrad, C.; Coe, J.; Roy-Chowdhury, S.; Weierstall, U.; James, D.; Wang, D.;

Grant, T.; Barty, A.; Yefanov, O.; Scales, J.; Gati, C.; Seuring, C.; Srajer, V.;

Henning, R.; Schwander, P.; Fromme, R.; Ourmazd, A.; Moffat, K.; Van Thor, J. J.;

Spence, J. C. H.; Fromme, P.; Chapman, H. N.; Schmidt, M. Science (New York,

N.Y.) 2014, 346 (6214), 1242.

(18) Gao, J.; Bian, M.; Wang, X.; Deng, W.; Yang, Z.; Zhong, D.; Zuo, Z.; Zhang, M.;

Lin, C. Proceedings of the National Academy of Sciences 2015, 112 (29), 9135.

(19) Domratcheva, T.; Udvarhelyi, A.; Rehaman, A.; Shahi, M. Methods in Molecular

Biology 2014, 1146, 191.

(20) Kao, Y. T.; Saxena, C.; He, T. F.; Guo, L.; Wang, L.; Sancar, A.; Zhong, D. Journal

of the American Chemical Society 2008, 130 (39), 13132.

(21) Sorigué, D.; Légeret, B.; Cuiné, S.; Blangy, S.; Moulin, S.; Billon, E.; Richaud, P.;

Brugière, S.; Couté, Y.; Nurizzo, D.; Müller, P.; Brettel, K.; Pignol, D.; Arnoux, P.;

Li-Beisson, Y.; Peltier, G.; Beisson, F. Science 2017, 357 (6354), 903.

(22) Huijbers, M. M. E.; Zhang, W.; Tonin, F.; Hollmann, F. Angewandte Chemie -

International Edition 2018, 57 (41), 13648.

(23) Zhou, Z.; Swenson, R. P. Biochemistry 1995, 34 (10), 3183.

(24) Watt, W.; Tulinsky, A.; Swenson, R. P.; Watenpaugh, K. D. Journal of Molecular

Biology 1991, 218 (1), 195.

(25) Swenson, R. P.; Krey, G. D. Biochemistry 1994, 33 (28), 8505.

(26) Sun, S.-Y.; Chao, D.-Y.; Li, X.-M.; Shi, M.; Gao, J.-P.; Zhu, M.-Z.; Yang, H.-Q.;

Luan, S.; Lin, H.-X. Nature cell biology 2009, 11 (7), 845.

(27) Su, L.; Shan, J. X.; Gao, J. P.; Lin, H. X. Molecular Plant 2016, 9 (2), 233.

(28) Sétif, P. Biochimica et Biophysica Acta - Bioenergetics 2001, 1507 (1–3), 161.

(29) Goñi, G.; Herguedas, B.; Hervás, M.; Peregrina, J. R.; De la Rosa, M. A.; Gómez-

Moreno, C.; Navarro, J. A.; Hermoso, J. A.; Martínez-Júlvez, M.; Medina, M.

Biochimica et Biophysica Acta - Bioenergetics 2009, 1787 (3), 144.

(30) He, T. F.; Guo, L.; Guo, X.; Chang, C. W.; Wang, L.; Zhong, D. Biochemistry 2013,

52 (51), 9120.

(31) Chang, C. W.; He, T. F.; Guo, L.; Stevens, J. A.; Li, T.; Wang, L.; Zhong, D. Journal

of the American Chemical Society 2010, 132 (36), 12741.

(32) Espinosa-Ruiz, A.; Bellés, J. M.; Serrano, R.; Culiáñez-Macià, F. A. Plant Journal

1999, 20 (5), 529.

(33) Kupke, T.; Hernández-Acosta, P.; Steinbacher, S.; Culiáñez-Macià, F. A. Journal of

Biological Chemistry 2001, 276 (22), 19190.

(34) Steinbacher, S.; Hernández-Acosta, P.; Bieseler, B.; Blaesse, M.; Huber, R.;

Culiáñez-Macià, F. A.; Kupke, T. Journal of Molecular Biology 2003, 327 (1), 193.

(35) Hernández-Acosta, P.; Schmid, D. G.; Jung, G.; Culiáñez-Macià, F. A.; Kupke, T.

Journal of Biological Chemistry 2002, 277 (23), 20490.

(36) Saxena, C.; Sancar, A.; Zhong, D. Journal of Physical Chemistry B 2004, 108 (46),

18026.

(37) Marcus, R. A.; Sutin, N. Biochimica et Biophysica Acta (BBA) - Reviews on

Bioenergetics 1985, 811 (3), 265.

(38) Marcus, R. A. Reviews of Modern Physics 1993, 65 (3), 599.

(39) Page, C. C.; Moser, C. C.; Chen, X.; Dutton, P. L. Nature 1999, 402 (6757), 47.

(40) Sumi, H.; Marcus, R. A. The Journal of Chemical Physics 1986, 84 (9), 4894.

(41) Yartsev, A.; Nagasawa, Y.; Douhal, A.; Yoshihara, K. Chemical Physics Letters

1993, 207 (4–6), 546.

(42) Weaver, M. J. Chem. Rev. 1992, 92 (3), 463.

(43) Barbara, P. F.; Meyer, T. J.; Ratner, M. A. The Journal of Physical Chemistry 1996,

100 (31), 13148.

(44) Wang, H.; Lin, S.; Allen, J. P. J.; Williams, J. J. C.; Blankert, S.; Laser, C.;

Woodbury, N. W. Science (New York, N.Y.) 2007, 316 (5825), 747.

(45) Liu, Z.; Tan, C.; Guo, X.; Li, J.; Wang, L.; Sancar, A.; Zhong, D. Proceedings of

the National Academy of Sciences of the United States of America 2013, 110 (32),

12966.

(46) McCarthy, A. A.; Mayhew, S. G.; Walsh, M. A.; Verma, C. S.; O’Connell, D. P.;

Reinhold, M.; Yalloway, G. N.; D’Arcy, D.; Higgins, T. M.; Voordouw, G.

Biochemistry 2002, 41 (36), 10950.

(47) Mayhew, S. G.; Massey, V. The Journal of biological chemistry 1969, 244 (3), 794.

(48) Massey, V.; Stankovich, M.; Hemmerich, P. Biochemistry 1978, 17 (1), 1.

(49) Kodali, G.; Siddiqui, S. U.; Stanley, R. J. Journal of the American Chemical Society

2009, 131 (13), 4795.

(50) Chang, C.-W.; Guo, L.; Kao, Y.-T.; Li, J.; Tan, C.; Li, T.; Saxena, C.; Liu, Z.; Wang,

L.; Sancar, A.; Zhong, D. Proceedings of the National Academy of Sciences of the

United States of America 2010, 107, 2914.

(51) Kao, Y.-T.; Tan, C.; Song, S.-H.; Öztürk, N.; Li, J.; Wang, L.; Sancar, A.; Zhong,

D. Journal of the American Chemical Society 2008, 130 (24), 7695.

(52) Brettel, K.; Aubert, C.; Vos, M. H.; Mathis, P.; Eker, A. P. M. Nature 2000, 405

(6786), 586.

(53) DeFelippis, M. R.; Murthy, C. P.; Broitman, F.; Weinraub, D.; Faraggi, M.; Klapper,

M. H. The Journal of Physical Chemistry 1991, 95 (8), 3416.

(54) O’Farrell, P. A.; Walsh, M. A.; McCarthy, A. A.; Higgins, T. M.; Voordouw, G.;

Mayhew, S. G. Biochemistry 1998, 37 (23), 8405.

(55) Genzor, C. G.; Perales-Alcón, A.; Sancho, J.; Romero, A. Nature Structural &

Molecular Biology 1996, 3 (4), 329.

(56) Karen, A.; Sawada, M. T.; Tanaka, F.; Mataga, N. Photochemistry and Photobiology

1987, 45 (1), 49.

(57) Kao, Y. T.; Saxena, C.; Wang, L.; Sancar, A.; Zhong, D. Cell Biochemistry and

Biophysics 2007, 48 (1), 32.

(58) Mayhew, S. G.; O’Connell, D. P.; O’Farrell, P. A.; Yalloway, G. N.; Geoghegan, S.

M. Biochemical Society transactions 1996, 24 (1), 122.

(59) Nag, L.; Sournia, P.; Myllykallio, H.; Liebl, U.; Vos, M. H. Journal of the American

Chemical Society 2017, 139 (33), 11500.

(60) Migliore, A.; Polizzi, N. F.; Therien, M. J.; Beratan, D. N. Chemical Reviews 2014,

114 (7), 3381.

(61) Liu, Z.; Tan, C.; Guo, X.; Kao, Y.-T.; Li, J.; Wang, L.; Sancar, A.; Zhong, D.

Proceedings of the National Academy of Sciences of the United States of America

2011, 108 (36), 14831.

(62) Winkler, J. R. J. R. Current Opinion in Chemical Biology. 2000, pp 192–198.

(63) Moser, C. C.; Keske, J. M.; Warncke, K.; Farid, R. S.; Dutton, P. L. Nature 1992,

355 (6363), 796.

(64) Gray, H. B.; Winkler, J. R. Quarterly Reviews of Biophysics. Cambridge University

Press August 2003, pp 341–372.

(65) Kundu, M.; He, T.-F.; Lu, Y.; Wang, L.; Zhong, D. The Journal of Physical

Chemistry Letters 2018, 9 (11), 2782.

(66) Schneider, C.; Sühnel, J. Biopolymers 1999, 50 (3), 287.

(67) Kammler, L.; Van Gastel, M. Journal of Physical Chemistry A 2012, 116 (41),

10090.

(68) Neiss, C.; Saalfrank, P. Photochemistry and photobiology 2003, 77 (1), 101.

(69) Fedorov, R.; Schlichting, I.; Hartmann, E.; Domratcheva, T.; Fuhrmann, M.;

Hegemann, P. Biophysical Journal 2003, 84 (4), 2474.

(70) Gray, H. B.; Winkler, J. R. Proceedings of the National Academy of Sciences 2005,

102 (10), 3534.

(71) Wan, C.; Xia, T.; Becker, H. C.; Zewail, A. H. Chemical Physics Letters 2005, 412

(1–3), 158.

(72) Huvaere, K.; Skibsted, L. H. Journal of the American Chemical Society 2009, 131

(23), 8049.

(73) Kopka, B.; Magerl, K.; Savitsky, A.; Davari, M. D.; Röllen, K.; Bocola, M.; Dick,

B.; Schwaneberg, U.; Jaeger, K. E.; Krauss, U. Scientific Reports 2017, 7 (1), 13346.

(74) Zhang, L.; Yang, Y.; Kao, Y. T.; Wang, L.; Zhong, D. Journal of the American

Chemical Society 2009, 131 (30), 10677.

(75) Kao, Y. T.; Guo, X.; Yang, Y.; Liu, Z.; Hassanali, A.; Song, Q. H.; Wang, L.; Zhong,

D. Journal of Physical Chemistry B 2012, 116 (30), 9130.

(76) Wolf, M. M. N.; Schumann, C.; Gross, R.; Domratcheva, T.; Diller, R. The Journal

of Physical Chemistry B 2008, 112 (42), 13424.

(77) Petersson, J.; Eklund, M.; Davidsson, J.; Hammarström, L. Journal of Physical

Chemistry B 2010, 114 (45), 14329.

(78) Gladkikh, V.; Burshtein, A. I.; Feskov, S. V.; Ivanov, A. I.; Vauthey, E. The Journal

of Chemical Physics 2005, 123 (24), 244510.

(79) Kumpulainen, T.; Lang, B.; Rosspeintner, A.; Vauthey, E. Chemical Reviews. 2017,

pp 10826–10939.

Appendix A. Additional information for characterization of ET dynamics in SQ flavodoxin

Kinetic Data Analyses for ET Cycle ET reactions in SQ flavodoxin involve charge separation, charge recombination and hot- state cooling that occur in a sequential order (see the complete scheme in Figure 8). * Assuming the total concentration of FMNH˙ is n0 upon excitation at time zero, the electron-transfer kinetics using the W donor as an example can be analyzed by fitting the equations below:

* [FMNH˙ ]t=0 = n0 [A1] d * * [FMNH˙ (t)] = –kFET [FMNH˙ (t)] [A2] dt d - + * + [FMNH (t)] = [W (t)] = kFET [FMNH˙ (t)] – kBET [W (t)] [A3] dt

† + † [FMNH˙ (t)] = kBET [W (t)] – kc [FMNH˙ (t)] [A4]

* + † [FMNH˙(t)] = n0 – [FMNH˙ (t)] – [W (t)] – [FMNH˙ (t)] [A5]

For ET reactions coupled with environment relaxations, we observed the stretched

훽 decays, 퐴푒−(푡/휏) . For an excited fluorophore (F*) following the stretched dynamics, the modified kinetics is:

* β * β-1 * [F (t)] = – z k [F (t)], z = β t , [F ]t=0 = n0

Therefore, in our data analyses for ET reactions coupled with environment relaxations, the following equations are used: d * β1 * [FMNH˙ (t)] = – z1 kFET [FMNH˙ (t)] [A6] dt

- + β1 * β2 + [FMNH (t)] = [W (t)] = z1 kFET [FMNH˙ (t)] – z2 kBET [W (t)] [A7]

† β2 + † [FMNH˙ (t)] = z2 kBET [W (t)] – kc [FMNH˙ (t)] [A8]

Table A.1 FET dynamics in semiquinone flavodoxin from the fluorescence

upconversion detection.a

Mutants Donor(s) Probe wavelengths 652 nm 670 nm 725 nm I Y98F W60 0.98(85%) 3.07(15%) 0.83(60%) 4.59(40%) 1.37(50%) 5.25(50%) Y98A W60 0.61(67%) 2.60(33%) Y98H W60 0.71(64%) 2.92(36%) Y98R W60 0.47(66%) 1.63(34%) II W60F Y98 1.46(47%) 5.30(53%) 1.93(44%) 8.52(56%) 2.95(33%) 9.73(67%) W60A Y98 1.10(77%) 5.75(22%) 1.28(50%) 5.50(50%) 1.76(30%) 5.67(70%) III WT W60,Y98 0.73(100%) 0.65(66%) 2.99(34%) 0.55(41%) 2.65(59%) D95N W60,Y98 0.63(95%) 6.00(5.%) 0.62(65%) 2.32(35%) 1.03(80%) 3.93(20%) G61A W60,Y98 0.78(63%) 3.37(37%) G61V W60,Y98 0.48(60%) 3.64(40%) IV Y98Wb W60,W98 N/A N/A N/A W60Y Y60,Y98 0.75(51%) 4.49(49%) 1.76(44%) 7.69(56%) 4.84(70%) 17.51(30%) aTime constants are in units of ps. bNo fluorescence upconversion signal was observed for Y98W.

Figure A.1 Steady-state absorption spectra of flavodoxin mutants.

(A) Mutants with a blue shift with respect to WT are represented by dashed lines. Solid lines represent mutants with a red shift. Note that G61V absorption spectrum has significant absorption from oxidized state with a weak semiquinone component. (B) Close- up view of the absorption spectra showing shifts relative to WT spectrum. Clearly, mutation of Y98 shifts to the right while mutating W60 results in a left shift.

Figure A.2 Normalized fs-resolved fluorescence transients of flavodoxin mutants.

(A) Up-conversion transients gated at 670 nm depicting ET behaviors of W60/Y98 and W60 donors obtained through various mutants. (B-C) Up-conversion transients probed at 725, 670 and 652 nm showing wavelength-dependent dynamics of solvation and ET reaction with W and Y donors in mutants of W60A, D95N, W60Y and WT.

100

Figure A.2 continued

101

Figure A.3 Normalized fs-resolved absorption transients of the mutant W60F/Y98F.

Initial energy dissipation measured at 1.8-2.1 ps and 8.5-26.6 ps mixes with lifetime decay of FMNH˙ (~300 ps) and generates the rise decay pattern in intermediate wavelengths. These dynamics are used as reference for analyzing ET data.

102

Figure A.4 Deconvolution of fs-resolved absorption transients of mutant Y98F.

(A) 800 and 740 nm wavelength probes FMNH˙* exclusive detection by 800 and 740 nm and kFET and β1 are obtained. (B) With known FET dynamics, the 690-nm transient is fitted + + with an additional W signal. The kBET and β2 were obtained from the deconvoluted W signal. (C-F) The transients are fitted with varying amplitudes of FMNH˙*, W+ and an additional FMNH˙† component. (G) A weak signal is detected at 630 nm with no contributions from FMNH˙*. (H) The 525-nm transient is probed within the FMNH˙ absorption range and is dominated by a negative FMNH˙ recovery signal. Note partial cancellation of cooling and recovery signals resulting in an overall faster signal

103

Figure A.4 continued

104

Figure A.5 Deconvolution of fs-resolved absorption transients of mutant W60F.

* (A-B) FMNH˙ exclusive detection by 800, 740 and 690 nm and kFET and β1 values are obtained. (C-F) With known FET dynamics and the similar cooling dynamics, the transients were fitted with an additional cooling signal and kBET and β2 were (G) The 630- nm transient was fitted with a negative FMNH˙* signal, similar to the early dynamics of W60F/Y98F mutant at the same wavelength. (H) The signal is probed within the FMNH˙ absorption range and is dominated by a negative FMNH˙ recovery signal. Note partial cancellation of cooling and recovery signals resulting in an overall faster signal

105

Figure A.5 continued

106

Appendix B. Additional information for characterization of ET dynamics in OX flavodoxin

Kinetic Data Analyses for ET Cycle: Heterogeneity Model ET reactions in FMN–W pair involve charge separation, triplet formation, charge * recombination and hot-state cooling. Assuming the total population of FMN is n0 upon excitation at time zero, various reaction kinetics in the ET cycle, can be analyzed by fitting the equations below:

* [FMN ]t=0 = n0 [B1] d * * * [FMN (t)] = –kFET [FMN (t)] – kTR [FMN (t)] [B2] dt d - + * + [FMN (t)] = [W (t)] = kFET [FMN (t)] – kBET [W (t)] [B3] dt

† + † [FMN (t)] = kBET [W (t)] – kc [FMN (t)] [B4]

* + † [FMN(t)] = n0 – [FMN (t)] – [W (t)] – [FMN (t)] [B5]

Where, kFET, kTR, kBET, and kc are rate constants of FET, triplet formation, BET and cooling dynamics, respectively.

For ET reactions coupled with environment relaxations, we observed the stretched decays,

훽 퐴푒−(푡/휏) . For an excited fluorophore (F*) following stretched dynamics, the modified kinetics is:

* β * β-1 * [F (t)] = – z k [F (t)], z = β t , [F ]t=0 = n0

Therefore, in our data analyses for ET reactions coupled with environment relaxations, the following equations are used: 107 d * β1 * * [FMN (t)] = – z1 kFET [FMN (t)]– kTR [FMN (t)] [B6] dt

+ β1 * β2 + [FMN (t)] = [W (t)] = z1 kFET [FMN (t)] – z2 kBET [W (t)] [B7]

† β2 + † [FMN (t)] = z2 kBET [W (t)] – kc [FMN (t)] [B8]

By fitting the fs-resolved transients, we determine [FMN*(t)], [W+(t)], [FMN†(t)], and [FMN(t)] kinetics.

For heterogeneous ET, we assume dynamics of each species as a sum of two such ET cycles at a fixed ratio. Thus,

* * * [FMN (t)] ∝ A1[FMN1 (t)] + A2[FMN2 (t)],

+ + + [W (t)] ∝ A1[W1 (t)] + A2 [W2 (t)], and

† † † [FMN (t)] ∝ A1[FMN1 (t)] + A2[FMN2 (t)]

108

Figure B.1 Steady-state absorption spectra of OX flavodoxin mutants.

(A) Flavin spectrum of the mutants with the S1S0 and S2S0 transitions peaked at 455 and 380 nm, respectively. (B) Close-up view of the absorption spectra showing shifts relative to WT spectrum. Mutation of aromatic Y98 residue blue–shifts the Y98F absorption by 5nm. Further mutation of both Y98F and W60 blue shifts the absorption in 5–10 nm indicating various electronic stabilization of FMN.

109

Figure B.2 Gaussian fitting and histogram analysis of edge–to–edge FMN-W distances. Large fluctuations in D95W, I65W, E66W from flexible W donors are strongly correlated with the observed heterogeneity in ET dynamics. Specially for D62W, we observed local configurations responsible for the unique τ1, τ2, β1, β2 ET dynamics.

110

Figure B.3 Normalized fs-resolved absorption transients of the mutant W60F/Y98F

in different timescales.

Note wavelength–dependent solvation behavior in earlier timescales (left) and longer rise dynamics (right) with detection of 3FMN signals.

111

Figure B.4 Normalized fs-resolved absorption transients of the mutant L67W.

For the donor W67, we detected FMN* decay in 106 ps, (β= 0.89) and a long component in 3.0 ns together with 3FMN absorption and solvation dynamics. No BET dynamics were resolved due to overall slower ET rates. Note similarity with signals of the ET–inert mutant W60F/Y98F.

112

Appendix C. Additional information for protein preparation and characterization of ET dynamics in OsHAL3

Protocol for OsHAL3 purification: The following is the scheme for expression/purification 6 L of cell culture.

1. Transform the OsHAL3 plasmid DNA (0.5ul) to 40ul E.coli BL21(DE3)

competent cell. Plate it on LB+Kan agar plates. 37C O/N.

2. Pick single colonies and inoculate 50mL LB+Kan+ O/N.

3. Transfer 10mL o/n culture to a 6X 1L culture with LB+Kan+ . Grow at 37C

(250rpm).

4. Induce the culture with 0.6 mM IPTG at the cell density reaches OD600 ~0.4-0.6.

Grow overnight at 18C (200rpm).

5. Spin down cells 5K for 10min and re-suspend the cell pellet in autoclaved cold

0 ddH2O and spin at 6K-5’ put cell pellet in -80 C, about 6-7g/Lcell

6. Lyse the cell in 300mM NaCl and 50mM Tris-HCl (pH8.0) 20mM βMEP

7. sonicate cell on ice at 60% duty cycle, output 5 about 45”X6-8

8. Centrifuge the cell lysate at 20 K for 45min.

9. Collect supernatant and loading onto talon column (3L cell about 4ml resin)

10. Wash the column with lysis buffer

11. Elute with elution buffer 300mM NaCl and 50 mM Tris.HCl, pH 8.0 20mM

βMEP containing 250mM imizadole.

113

12. Collect all yellow fractions with 1mL tubes (about 7ml).

13. Run SDS-PAGE on all fractions to check the purit, if protein is not pure

14. Combine all fractions and dilute at least 6 folds with no salt buffer 50mM Tris

pH8.0 20mM βMEP

15. Load onto DEAE sepharose column (20ml) wash with 50mM Tris pH8.0 50mM

NaCl 20mM βMEP

16. Step elution with 50mM Tris pH8.0 300mM NaCl 20mM βMEP or gradient elute

with 0-1M NaCl

17. Run SDS-PAGE to check purity and combine all pure fractions and concentrate to

small volume

18. Dialysis against 50mM Tris pH8.0 50 mM NaCl 20mM βMEPo/n

114

Kinetic data analyses for OsHAL3 ET cycle:

ET reactions in flavodoxin involve charge separation, charge recombination and hot-state

* cooling leading to FMN formation. Assuming the total concentration of FMN is n0 upon excitation at time zero, the electron-transfer dynamics with W79 and W82 donors can be described by:

* [FMN ]t=0 = n0 [C1] d * * [FMN (t)] = – kFLR [FMN (t)]; kFLR = kFETW79 + kFETW82 [C2] dt d + * + [W79 (t)] = kFETW79 [FMN (t)] – kBETW79 [W79 (t)] [C3] dt

+ * + [W82 (t)] = kFETW82 [FMN (t)] – kBETW82 [W82 (t)] [C4]

† + † [FMNW79 (t)] = kBETW79 [W79 (t)]– kc [FMNW79 (t)] [C5]

† + † [FMNW82 (t)] = kBETW82 [W79 (t)]– kc [FMNW82 (t)] [C6]

† † † [FMN (t)] = [FMNW79 (t)] + [FMNW82 (t)] [C7]

* + † [FMNH˙(t)] = n0 – [FMNH˙ (t)] – [W (t)] – [FMNH˙ (t)] [C8]

With possible overlap between ET and solvation timescales, the modified stretched-ET kinetics were defined as:

115 d * β1 * [FMN (t)] = – z1 kFLR [FMN (t)]; [C9] dt

d + β1 * β2 + [W79 (t)] = kFETW79 [FMN (t)] – kBETW79 [W79 (t)] [C10] dt

+ β1 * β3 + [W82 (t)] = kFETW82 [FMN (t)] – kBETW82 [W82 (t)] [C11]

† β2 + † [FMNW79 (t)] = kBETW79 [W79 (t)]– kc [FMNW79 (t)] [C12]

† β3 + † [FMNW82 (t)] = kBETW82 [W79 (t)]– kc [FMNW82 (t)] [C13]

† † † [FMN (t)] = [FMNW79 (t)] + [FMNW82 (t)] [C14]

For the heterogeneity ET dynamics were similarly modelled, and the heterogeneity concentrations were added at a constant ratio.

116

Figure C.1 Fs-resolved fluorescence and absorption transients of the mutant

W79F/W82F.

We identified the FMN* dynamics in 180 ps (45%) and 560 ps (55%) in the upconversion signal and absorption transient at 800 nm. Longer signals in the absorption signals in 500– 580 nm are obtained from 3FMN dynamics with identical dynamics. We additionally observed minor observed solvation dynamics in 1.8 and 18 ps in the absorption transients, but were not present in ET mutants due to ultrafast charge separation rates.

117

Figure C.2 Fs-resolved transients of the mutant W79F/W82F/H91F gated at 495,

500 and 540 nm.

FMN* dynamics with the longer lifetime of 1.2 ns showing fast solvation components in 1.8 ps, 16 ps (left), along with a longer component in 550 ps (right).

118

Figure C.3 SQ ET dynamics of the mutant W79F using 580 nm excitation.

SQ dynamics of FET and BET reactions in 3 ps and 72 ps using selective excitation of FMNH˙ with 580 nm light.

119

Figure C.4 Comparison of fs-resolved absorption transients between the mutants

H59/W82F and W82F.

Transient absorption signals of W82F and H59F/W82F at 800 and 500 nm show identical ET and BET dynamics. H59 is thus not protonated at the buffer pH of 8.0 and no electrostatic effect of was observed on the ET dynamics.

120