Document [original]

Vibrational spectroscopic studies on [NiFe] hydrogenases:

Insights into structure and function

vorgelegt von

M. Sc.

Chara Karafoulidi-Retsou

ORCID: 0000-0001-8572-408043

an der Fakultät II – Mathematik und Naturwissenschaften

der Technischen Universität Berlin

zur Erlangung des akademischen Grades

Doktorin der Naturwissenschaften

-Dr. rer. nat.-

genehmigte Dissertation

Promotionsausschuss:

Vorsitzender: Prof. Dr. Arne Thomas

Gutachter: Prof. Dr. Peter Hildebrandt

Gutachter: Prof. Dr. Holger Dobbek

Tag der wissenschaftlichen Aussprache: 21. März 2024

Berlin 2024

Abstract

Molecular hydrogen has signiﬁcant poten�al as an energy carrier in sustainable industrial produc�on

aiming to eliminate carbon dioxide emission. This potential, however, relies on the development of

electrolyzers for energy storage through water spli�ng and fuel cell technologies for conver�ng H2 and

O2 back into electricity, using cheap, abundant, and efficient catalysts. In nature, H2 cycling is orchestrated

by metalloenzymes known as hydrogenases, which operate with remarkable cataly�c rates using

exclusively nickel and/or iron metals. The discovery of hydrogenases has been a breakthrough towards

novel cataly�c strategies to replace noble metals like pla�num. Addi�onally, hydrogenases have also been

employed in H2-dependent biotechnological applica�ons such as NAD(P)H-cofactor regenera�on systems

and biosensors. Despite these achievements, these enzymes also display a few drawbacks limi�ng their

applica�on. First, they are not easy to produce as their matura�on depends on a complex biosynthe�c

machinery. Secondly, most hydrogenases are extremely sensi�ve to oxygen, which inhibits the cataly�c

sites. Thirdly, as hydrogenases are biomolecules with large molecular weights, the achievement of high

catalyst densi�es at electrode surfaces is o�en problema�c. To overcome these drawbacks, a profound

understanding of the cataly�c mechanism and the matura�on of these enzymes is required, thereby

deﬁning the objec�ve of this work which employs a combina�on of infrared (IR) spectroscopic techniques.

In the ﬁrst part, the cataly�c subunit HoxG of the membrane-bound [NiFe]-hydrogenase from Cupriavidus

necator, lacking the [Fe-S] cluster-containing small subunit HoxK, has been subjected to detailed

spectroscopic inves�ga�ons. This project revealed so far unknown matura�on intermediates of the

stepwise assembled [NiFe] cofactor. Among them, the metal-free (preHoxGΔFeNi) and Ni-depleted

(preHoxGΔNi) large subunit intermediates oﬀer novel possibili�es to introduce diﬀerent ac�ve site metals

aiming to develop in the future chemzymes with alterna�ve cataly�c func�ons. Subsequently, the HoxG

subunit containing a fully equipped NiFe(CN)2CO cofactor was analyzed in detail in solu�on and

subsequently also immobilized on surfaces using surface-enhanced infrared spectroscopy to understand

its supramolecular arrangement, stability and (redox) reac�vity. Experimental results were complemented

by theore�cal calcula�ons by the group of Prof. Mroginski to achieve a comprehensive insight regarding

orienta�on of the immobilized proteins and their distance from the electrode surface.

In the second part, the soluble NAD+-reducing [NiFe]-hydrogenase from Hydrogenophilus thermoluteolus

and the membrane-bound [NiFe]-hydrogenase from Cupriavidus necator were used as model enzymes for

IR spectroscopic inves�ga�ons targe�ng the elucida�on of certain cataly�c and (oxygen-) inhibited

intermediates. Both enzymes were inves�gated in a broad temperature range, focusing on the eﬀect of

temperature and light induced perturba�ons of the enrichment of certain redox species. One of the new

discoveries was a light triggered conversion of the fully reduced state of the ac�ve site, Nia-SR, to the one-

electron oxidized Nia-L. The light-driven Nia-SR → Nia-L reac�on represents a photochemical shortcut of

the cataly�c cycle and may be a milestone for the manipula�on of hydrogenases with light.

Finally, we addi�onally resolved the unexpected IR spectral contribu�ons of protonated cysteine residues

during the conversion of certain hydrogenase redox states. These ﬁndings, backed up by biochemical and

computa�onal data performed in collabora�on with other researchers highlight the importance of careful

interpreta�on of IR signals.

Zusammenfassung

Molekularer Wasserstoﬀ besitzt ein erhebliches Potenzial als Energieträger in einer nachhal�gen

industriellen Produk�on zu wirken, die darauf abzielt, Kohlenstoﬀdioxidemissionen zu vermeiden.

Voraussetzung dafür ist jedoch die Entwicklung von Elektrolyseuren zur Energiespeicherung durch

Spaltung von Wasser und Brennstoﬀzellentechnologien zur Rückverwandlung von H2 und O2 in Elektrizität

unter Verwendung preiswerter und eﬃzienter Katalysatoren. In der Natur wird der H2-Zyklus von

Metalloenzymen, den sogenannten Hydrogenasen, gesteuert, die mit bemerkenswerten kataly�schen

Geschwindigkeiten arbeiten und ausschließlich Nickel- und/oder Eisenmetalle verwenden. Die Entdeckung

der Hydrogenasen war ein Durchbruch auf dem Weg zu neuen kataly�schen Strategien, die darauf

abzielen, Edelmetalle wie Pla�n zu ersetzen. Darüber hinaus wurden Hydrogenasen auch in H2-abhängigen

biotechnologischen Anwendungen wie Systemen zur Regenera�on von NAD(P)H-Kofaktoren und als

Biosensoren eingesetzt. Trotz dieser Erfolge weisen diese Enzyme auch einige Nachteile auf, die ihre

Anwendung einschränken. Erstens sind sie nicht einfach herzustellen, da ihre Reifung von einer komplexen

Biosynthesemaschinerie abhängt. Zweitens reagieren die meisten Hydrogenasen extrem empﬁndlich auf

Sauerstoﬀ, der ihre kataly�schen ak�ven Zentren o� stark hemmt. Dritens sind Hydrogenasen große

Biomoleküle, so dass die Erzielung hoher Katalysatordichten an Elektrodenoberﬂächen o� problema�sch

ist.

Um diese Nachteile zu überwinden, bedarf es einer umfassenden Kenntnis der kataly�schen

Mechanismen der Enzyme und ihrer Assemblierung, wodurch die Ziele dieser spektroskopischen

Untersuchung deﬁniert sind. Im ersten Teil wurde die kataly�sche Untereinheit HoxG der

iii

membrangebundenen [NiFe]-Hydrogenase aus Cupriavidus necator, der die [Fe-S] Cluster enthaltende

kleine Untereinheit HoxK fehlt, einer detaillierten spektroskopischen Untersuchung unterzogent, wobei

bisher unbekannte Reifungsintermediate des schritweise aufgebauten [NiFe]-Cofaktors entdeckt wurden.

Unter ihnen bieten die metallfreien (preHoxGΔFeNi) und Ni-verarmten (preHoxGΔNi) Zwischenstufen der

großen Untereinheit neue Möglichkeiten, verschiedene Metalle in das ak�ve Zentrum einzuführen, um in

Zukun� sogenannte „Chemzyme“ mit alterna�ven kataly�schen Funk�onen zu entwickeln. Anschließend

wurde die HoxG-Untereinheit, die einen vollständig ausgestateten NiFe(CN)2CO-Cofaktor enthält, in

Lösung und anschließend auch immobilisiert auf Oberﬂächen mit der oberﬂächenverstärkten Infrarot-

Spektroskopie eingehend untersucht, um die supramolekulare Anordnung, Stabilität und (Redox-)

Reak�vität der Enzyme zu verstehen. Die experimentellen Ergebnisse wurden durch theore�sche

Rechnungen in der Gruppe von Prof. Mroginski ergänzt, so dass umfassende Einsichten über die

Orien�erung der immobilisierten Proteine und deren Abstand zur Elektrodenoberﬂäche gewonnen

werden konnten.

Im zweiten Teil wurden die lösliche NAD+-reduzierende [NiFe]-Hydrogenase aus Hydrogenophilus

thermoluteolus und die membrangebundene [NiFe]-Hydrogenase aus Cupriavidus necator als

Modellenzyme für IR-spektroskopische Untersuchungen verwendet, die auf die Au�lärung bes�mmter

kataly�scher und (Sauerstoﬀ-) inhibierter Zwischenprodukte abzielten. Beide Enzyme wurden in einem

weiten Temperaturbereich untersucht, wobei ein Schwerpunkt auf die Auswirkungen von Temperatur und

lich�nduzierten Störungen auf die Anreicherung bes�mmter Redoxspezies gelegt wurde. Bemerkenswert

ist, dass eine durch Licht ausgelöste Umwandlung des vollständig reduzierten Zustands des ak�ven

Zentrums, Nia-SR, in das mit einem Elektron oxidierte Nia-L nachwiesen werden konnte. Die

lichtgesteuerte Nia-SR → Nia-L-Reak�on stellt eine photochemische Abkürzung des kataly�schen Zyklus

dar und könnte ein Meilenstein für die Manipula�on von Hydrogenasen mit Licht sein.

Schließlich konnten auch die unerwarteten IR Signale von protonierten Cysteinresten während der

Umwandlung bes�mmter Hydrogenase-Redoxzustände aufgeklärt werden. Diese Ergebnisse, die durch

biochemische und rechnerische Daten anderer Arbeitsgruppen gestützt werden, unterstreichen die

Bedeutung einer sorgfäl�gen Interpreta�on der IR Spektren.

Publications

Related to this thesis:

Submitted

Chara Karafoulidi-Retsou; Christian Lorent; Sagie Katz; Yvonne Rippers; Hiroaki Matsuura; Yoshiki Higuchi;

Ingo Zebger*; Marius Horch*

Light-Induced Electron Transfer in a [NiFe] Hydrogenase Opens a Photochemical Shortcut for Catalytic

Dihydrogen Cleavage. (Submitted 01/2024)

Jovan Dragelj; Chara Karafoulidi-Retsou; Sagie Katz; Oliver Lenz; Ingo Zebger; Giorgio Caserta; Sophie

Sacquin-Mora; Maria Andrea Mroginski *

Conformational and mechanical stability of the isolated large subunit of membrane-bound [NiFe]-

hydrogenase from Cupriavidus necator Front.Microbiol.13:1073315

(2023). https://www.frontiersin.org/articles/10.3389/fmicb.2022.1073315/full

Giorgio Caserta; Sven Hartmann; Casey Van Stappen; Chara Karafoulidi-Retsou; Christian Lorent; Stefan

Yelin; Matthias Keck; Janna Schoknecht; Ilya Sergueev; Yoshitaka Yoda; Peter Hildebrandt; Christian

Limberg; Serena DeBeer; Ingo Zebger; Stefan Frielingsdorf *; Oliver Lenz *

Stepwise assembly of the active site of [NiFe]-hydrogenase. Nature Chemical Biology volume 19,

pages498–506 (2023). https://www.nature.com/articles/s41589-022-01226-w

Kulka-Peschke, C. J.; Schulz, A. C.; Lorent, C.; Rippers, Y.; Wahlefeld, S.; Preissler, J.; Schulz, C.;

Wiemann, C.; Bernitzky, C. C. M.; Karafoulidi-Retsou, C.; Wrathall, S. L. D.; Procacci, B.; Matsuura, H.;

Greetham, G. M.; Teutloff, C.; Lauterbach, L.; Higuchi, Y.; Ishii, M.; Hunt, N. T.; Lenz *, O.; Zebger *, I.;

Horch *, M.

Reversible Glutamate Coordination to High-Valent Nickel Protects the Active Site of a [NiFe]

Hydrogenase from Oxygen. J. Am. Chem. Soc. (2022), 144 (37), 17022–17032.

https://doi.org/10.1021/jacs.2c06400

To be Submitted

Chara Karafoulidi-Retsou; Jovan Dragelj; Sara Böning; Sophie Sacquin-Mora; Sagie Katz; Ingo Zebger;

Oliver Lenz; Giorgio Caserta; Maria Andrea Mroginski

Spectro-electrochemical control of the large subunit of a [NiFe] hydrogenase in solution and

immobilized. (In preparation)

Chara Karafoulidi-Retsou; Stefan Frielingsdorf; Yvonne Rippers; Marius Horch; Sagie Katz; Oliver Lenz;

Ingo Zebger; Giorgio Caserta

Spectroscopic studies of a membrane-bound [NiFe] hydrogenase under cryogenic temperatures:

Difference spectroscopy beyond the first coordination sphere. (In preparation)

Other publications in peer-reviewed journals:

Laun, K.; Duffus, B.R.; Kumar, H.; Oudsen, J.-P.; Karafoulidi-Retsou, C.; Tadjoung, A.F.; Hildebrandt, P.; Ly,

K.H.; Leimkühler, S.; Katz*, S.; Zebger*, I. A minimal light-driven system to study the enzymatic CO2

reduction of formate dehydrogenase. ChemCatChem 2022, 14, e202201067

first author, *corresponding author

Selected Talks and Posters

Immobilizing peptides and proteins: Interplay between theoretical and experimental approaches,

CECAM-FR-MOSER, Institut de Biologie Physico-Chimique, Sorbonne Université, Paris, France, October 6th,

2022. (Combined talk with Dr. Jovan Dragelj)

New insights on H2 catalysis and O2 tolerance of a soluble NAD+-reducing [NiFe] hydrogenase, Iron

Sulfur Proteins: Biogenesis, Regulation and Function meeting, Sainte-Maxime, France, September 26th-

30th, 2022.

Temperature-dependent and light-sensitive equilibria of catalytic intermediates in a thermostable

NAD+-reducing [NiFe] hydrogenase, 12th International Conference on Advanced Vibrational Spectroscopy,

Faculty of Chemistry Jagiellonian University, Krakow, Poland, August 27th – Sepember 1st, 2023.

vii

Acknowledgements

I would like to thank:

• Prof. Dr. Peter Hildebrandt and Dr. Ingo Zebger for the opportunity to conduct my PhD under their

supervision in this group

• Dr. Ingo Zebger in particular for the scientific exchange and the long meetings, important for the

evolution of my projects

• Dr. Giorgio Caserta, for his methodical scientific approach, the inspirational mentorship and

support throughout my doctoral studies

• Dr. Sagie Katz, for the inspirational long talks, boosting my creativity and all the contribution and

support to my research endeavors

• Dr. Enrico Forbrig for being the first to welcome me not only in the group but also in Berlin and

familiarizing me with the lab equipment

• Prof. Dr. Mroginski and her group, particularly Dr. Jovan Dragelj and Sarah Böning for the

collaboration and the scientific support

• Dr. Marius Horch for our collaboration, the scientific talks and all the knowledge that I gained

from him

• Dr. Stefan Frielingsdorf, Dr. Oliver Lenz and Dr. Andrea Schmidt for our collaboration

• Armel Waffo, Dr. Konstantin Laun, Dr. Christian Lorent, Lena Schäfer, Tamanna Ahammad, Victor

Nikolaus, Dr. Katharina Julia Kulka, Charlotte Wiemann and Dr. Anastasia Kraskov

• Prof. Dr. Inez Weidinger for recommending me to the group

• All the other colleagues from MVL, Marina Böttcher, Jürgen Krauss, Claudia Schulz and Dr. Uwe

Kuhlmann

• My family in Greece for all the support, my brother Christos, my mother Eugenia and my father

Argyris

• The biggest thanks goes to Dimitris, my devoted and supportive companion

viii

Abbrevia�ons

Arg

Arginine

ATP

Adenosine Triphosphate

ATR

Atenuated Total Reﬂec�on

COOH

8-Mercaptooctanoic acid thiol

8-Amino-1-octanthiol hydrochloride

1-Octanthiol

ca.

circa

Counter Electrode

Cupriavidus Necator

cryo

cryogenic

Cyclic Voltametry

Cys

Cysteine

Cyt

Cytochrome

DFT

Density Func�onal Theory

DvMF

Desulfovibrio vulgaris Miyazaki F

EDC

N-Ethyl-Nʹ-(3-dimethylaminopropyl)-carbodiimide –hydrochloride

EPR

Electron Paramagne�c Resonance

FMN

Flavin Mononucleo�de

Fourier Transform

FTIR

Fourier Transform Infrared

Glu

Glutamate

H/D

Hydrogen/Deuterium (exchange)

Het

Heterologous

His

His�dine

His-tag

hexahis�dine-tag (aﬃnity tag)

Hydrogenophilus Thermoluteolus

Infrared

IRE

Internal Reﬂec�on Element

LED

Light-Emi�ng Diode

MB/MeB

Methylene Blue

MBH

Membrane Bound Hydrogenase

Molecular Dynamics

NAD+

Nico�namide Adenine Dinucleo�de

NADPH

Nico�namide Adenine Dinucleo�de Phosphate

NaDT

, Sodium hydrosulﬁte

NHS

N-Hydroxysuccinimid

NMR

Nuclear Magne�c Resonance

OCP

Open Circuit Poten�al

PDB

Protein data base

QM/MM

Quantum Mechanics / Molecular Mechanics

Reference Electrode

Regulatory Hydrogenase

Resonance Raman

Room Temperature

ROS

Reac�ve Oxygen Species

SAM

Self-Assembled Monolayer

SEC

Size-Exclusion-Chromatography

SEIRA

Surface-Enhanced Infrared Absorp�on

Soluble Hydrogenase

Strep-tag

strep-tag II aﬃnity tag

Working Electrode

XAS

X-ray absorp�on Spectroscopy

Contents

Abstract ............................................................................................................................................................. i

Zusammenfassung ............................................................................................................................................ ii

Acknowledgements ......................................................................................................................................... vii

1. Introduction ................................................................................................................................................. 1

1.1 Hydrogenases ......................................................................................................................................... 2

1.2 Active site composition of [NiFe] – hydrogenases ................................................................................ 2

1.3 Redox structural Intermediates and their interconversion ................................................................... 4

1.4 Spectroscopic tools for the characterization of the redox structural states of the active site ............ 7

1.4.1 Large subunit HoxG-Maturation Process ..................................................................................... 11

1.5 Hydrogenophilus Thermoluteolous HtSH ............................................................................................. 13

1.6 References ........................................................................................................................................... 15

2. Methods, techniques and materials .......................................................................................................... 23

2.1 Molecular Vibrations ............................................................................................................................ 23

2.2 Normal modes ...................................................................................................................................... 25

2.3 Vibrational Spectroscopy ..................................................................................................................... 27

2.3.1 Infrared absorption ....................................................................................................................... 28

2.4 The Michelson Interferometer ............................................................................................................ 29

2.5 Attenuated Total Reflectance – Infrared Spectroscopy ...................................................................... 29

2.6 Surface-Enhanced Infrared Absorption Spectroscopy ........................................................................ 31

2.7 Electromagnetic (EM) Mechanism-plasmon resonance ..................................................................... 32

2.8 IR Spectroscopy on proteins ................................................................................................................ 34

2.8.1 Absorption spectroscopy .............................................................................................................. 34

2.8.2 Difference spectroscopy ............................................................................................................... 36

2.8.3 Functionalized gold surface-Self-Assembled Monolayer (SAM) .................................................. 38

2.8.4 SEIRA prism as the electrode - Electrochemistry (EC) .................................................................. 39

2.9 Electron Paramagnetic Resonance ...................................................................................................... 40

2.10 Experimental Procedures, Chemicals and Buffers ............................................................................ 45

2.11 References ......................................................................................................................................... 51

3. Spectroscopic studies of the large subunit HoxG of the membrane-bound hydrogenase from

Cupriavidus necator ....................................................................................................................................... 53

3.1 HoxG studies in bulk ............................................................................................................................ 53

3.1.1 Maturation Intermediates ............................................................................................................ 53

xii

3.1.2 Characterization of the fully-matured HoxG ................................................................................ 58

3.1.3 Monomer vs Homodimer of HoxG ............................................................................................... 66

3.2 HoxG immobilized on SAM functionalized electrodes-SEIRA studies ................................................. 72

3.2.1 Headgroup and protonation degree effect on the protein’s orientation .................................... 72

3.2.2 Comparison between experimental results and simulations ...................................................... 81

3.2.3 The influence of the ionic strength .............................................................................................. 83

3.2.4 Immobilizing HoxG via a covalent approach ................................................................................ 86

3.2.5 Immobilization of the CnMBH heterodimer ................................................................................. 93

3.3 Conclusions .......................................................................................................................................... 98

3.4 References ......................................................................................................................................... 101

4. Spectroscopic studies on temperature-dependent and light-sensitive equilibria of [NiFe] hydrogenases

...................................................................................................................................................................... 105

4.1 Temperature-dependent equilibria between redox-structural states of [NiFe] hydrogenases ....... 107

4.1.1 Hydrogenophilus Thermoluteolus Soluble Hydrogenase ........................................................... 107

4.1.2 Cupriavidus necator Membrane Bound Hydrogenase ............................................................... 114

4.2 Photoreactivity of Nia-SR species ...................................................................................................... 117

4.2.1. HtSH ........................................................................................................................................... 117

4.2.2. Charaterization of the various prosthetic groups in HtSH ........................................................ 127

4.3 Characterization of the Nia-L species in Hydrogenophilus Thermoluteolus Soluble Hydrogenase

(HtSH) ....................................................................................................................................................... 133

4.4 Characterization of the Nia-L species of Cupriavidus Necator Membrane Bound Hydrogenase

(CnMBH) ................................................................................................................................................... 137

4.4.1 C81S variant of CnMBH ............................................................................................................... 141

4.5 Conclusions ........................................................................................................................................ 145

4.6 References ......................................................................................................................................... 147

5. Outlook and final remarks ....................................................................................................................... 151

6. Appendix .................................................................................................................................................. 155

6.1 Appendix to Chapter 3 ....................................................................................................................... 155

6.2 Appendix to Chapter 4 ....................................................................................................................... 159

1. Introduction

In [NiFe] hydrogenases, the Fe metal of the catalytic center is coordinated by three unusual inorganic

ligands, namely one carbon monoxide molecule (CO) and two cyanides (CN–), which comprise also value

marker bands for infrared (IR) spectroscopy and are studied by different IR techniques. One advantage is

that their intramolecular stretching vibrations are located in a spectral region free of other IR signals.

Additionally, the σ-donor/π-acceptor properties of these ligands make them sensitive toward electronic

and structural changes of the active site.1 Therefore, IR spectroscopy is a powerful method to monitor

vibrational modes not only useful for the assignment of the different redox structural states of the active

site,1,2 but also for protonation events around the first coordination sphere, enlightening for example the

proton transfer pathway.3,4

Despite their obvious advantages, these enzymes are associated with several limitations hindering their

broad commercial utilization. The challenge lies in the maturation complexity and yet-to-be-fully-

understood biosynthetic machinery, making their production a non-trivial task.5,6 In case these enzymes

are immobilized, their large volume impedes the attainment of optimal catalyst density on electrode

surfaces. Additionally, after chemical or physical immobilization the enzymes undergo some unavoidable

structural changes. The final orientation determines the energy loss at the enzyme-electrode interface

during the electron transfer.

This thesis aims to address certain questions highlighted here and to overcome the challenges which were

previously documented. In Chapter 3 it is attempted to study and gain an insight on the complex

maturation process of the [NiFe] hydrogenases. A fully mature large subunit is characterized i.a. by IR

spectroscopy. As a next step in the direction of applicability, the protein is immobilized on functionalized

electrodes with the goal to follow its electrochemical control spectroscopically. This requires an

orientation-controlled adsorption to enable efficient electronic communication with the surface of the

electrode.

In Chapter 4 two hydrogenases are studied at cryogenic temperatures. The results show that under

illumination some of the catalytically active site states can be bypassed, due to a light-driven electron

transfer towards the enzyme’s iron-sulfur clusters. A shortcut of the catalytic cycle reveals new pathways

of accelerating or controlling [NiFe] hydrogenases by light. By delving into these uncharted territories, it

is attempted to uncover the potential applications that arise from this unique interplay, envisioning the

future utilization of hydrogenases for sustainable energy solutions.

1.1 Hydrogenases

Hydrogenases are ancient metalloenzymes responsible for the reversible cleavage of molecular hydrogen

into protons and electrons under ambient conditions.7–9 Comprehensive review articles on the importance

of H2 in the microbial world are available.9–13 For example, H2 production can serve as a mechanism for

the organism to get rid of excess electrons and protons. In other cases, hydrogen consumption is coupled

with other chemical transformations such as the reduction of carbon dioxide or the formation of a proton

gradient required for the production of ATP (Adenosine Triphosphate) .8,14,15 H2 cleavage takes place at a

specific active site based on transition metals nickel and iron or solely iron.

According to their metal-content hydrogenases are classified into 3 groups, i.e., [NiFe] hydrogenases,

[FeFe] hydrogenases, and [Fe] hydrogenases (Fig. 1.1). [NiFe] hydrogenases, which are the focus of this

study, can be further classified based on the subunits composition, cofactor content, cellular location, and

their physiological function.13

Figure 1.1: Structures of the catalytic center in [FeFe], [NiFe] and [Fe] hydrogenases. The substrate/inhibitor in the

presumed binding position is marked as X. The figure is reprinted from the open access journal, Molecules 2021, 26,

4852.16

All hydrogenase active sites house unusual and normally toxic carbon cyanide (CN–) and carbon monoxide

(CO) or solely CO ligands coordinated to the particular active sites. These ligands were first detected by

IR2,17–19 and this finding was quite surprising for the scientific community considering that those

compounds can be potentially lethal for living organisms.

1.2 Active site composition of [NiFe] – hydrogenases

The simplest [NiFe] hydrogenases is comprised by a heterodimeric functional unit, which consists of a

large (ca. 50-60 kDa) and a small (ca. 30-40 kDa) protein subunit.7 The large subunit is harboring the

NiFe(CN)2CO active site, whereas the small subunit contains at least one [FeS] cluster, which function(s)

as an electron relay. Additional subunits can also be present, together with various prosthetic groups (e.g.

FMN, NAD).7,13,14,20–22

Ni Fe

SCys

Figure 1.2: The catalytic site of a [NiFe] hydrogenase comprises a heterobimetallic NiFe(CN)2CO complex. Here the

nickel ions are bound to the protein scaffold of the large subunit via four cysteine residues, two of which are shared

with the Fe ion. The coordination sphere of the Fe site is completed by two strong-field CN– and one CO ligand. The

label X stands for a third bridging position between the two metals, which can be empty or occupied by e.g. OH– or

H– ligands.

The first X-ray structure of a [NiFe] hydrogenase was reported in 199523,24 , revealing that the large subunit

harbors a metal cofactor comprised of the Ni and Fe transition metals. This active site is bound to the

hydrogenase large subunit via four cysteine residues, two terminally bound to the Ni and two bridging the

Ni and Fe ions (Fig. 1.2). Significantly, the Fe ion retains a low spin (S=0) ferrous form in all catalytic and

inhibited states observed so far, while the Ni has been shown to change oxidation state (NiI, NiII and NiIII

or NiIV,25 which was recently proposed). The coordination sphere of the Fe is completed by two CN– and

one CO ligand. The vibrational modes of these ligands are sensitive toward various structural and

electronic aspects of the active site, and the CO stretching represents a convenient identifier for different

[NiFe] states.

As the active site is buried in the protein scaffold, gas and H+ channels connecting the surface to the active

site cleft have been recognized in hydrogenases. It is generally accepted that protons leave the [NiFe]

active site assisted by close-by residues. Among them, changes in the protonation state have been

detected for one of the Ni-bound cysteine residues and glutamate.4 Both residues are connected to a

network of amino acids and water molecules, leading to the protein surface.3,26 A few groups highlighted

also the importance of a close-by arginine that has been hypothesized to work together with the [NiFe]

site metal atoms, acting as a frustrated Lewis pair, that mediates the first step of H2 activation. 27

1.3 Redox structural Intermediates and their interconversion

Figure 1.3: Redox structural states and proposed interconversion for anaerobic [NiFe] hydrogenases.28 The Ni-A/B/C

refers to the first three paramagnetic redox states of the [NiFe], which were (EPR) spectroscopically identified. In

this thesis the focus lays on the active and ready states, the ‘’r’’ stands for the ready-to-activate states, typically

these states activate within 3 min. The index ‘’a’’ stands for the active states in H2 catalysis.

Each redox structural state of the active site is generally symbolized as Nix-Y, where the subscript “x” gives

a hint of whether the species is catalytically active, ready, or inactive. In detail, “a” stands for active, “r”

for ready, “u” for unready and “ia” for inactive species. Y is used to differentiate between the

paramagnetic redox states (Ni-A, Ni-B, Ni-C and Ni-L) from the diamagnetic ones, which are symbolized as

Ni-S, where S stands for EPR silent. Additionally, the paramagnetic hydrogenase states are usually named

following the order they were discovered, Ni-A, Ni-B, etc. with Ni-L being an exception as L stands for light.

This species is enriched by light exposure of the reduced Ni-C. A schematic depiction of the redox

structural states commonly observed in anaerobic [NiFe]-hydrogenases is shown in Fig. 1.3.

Niu-A and Nir-B are the most oxidized [NiFe] hydrogenase redox states.29–32 Both feature a bridging OH-

ligand between the two heterometals and a NiIII ion. Niu-A, however, is found (almost exclusively) in O2-

sensitive [NiFe]-hydrogenases that are strongly inhibited by O2. High-resolution X-ray structural data

revealed the presence of sulfoxygenated species in Niu-A involving one of the bridging cysteine

residues.33,34 Crystallographic data did not show sulfoxygenated species in Nir-B.35

One-electron reduction of Nir-B and the addition of one proton leads to removal of bridging OH– ligand as

H2O and formation of Nia-S, which is considered to be the first, hydrogen binding state of the catalytic

cycle.36-38 This species has been widely characterized using a plethora of techniques including IR,

resonance Raman, nuclear resonance vibrational spectroscopy (NRVS) and X-ray crystallography. Herein,

the active site exhibits a low spin Ni2+ ion in a distorted seesaw geometry. One electron reduction and the

addition of a H+ converts the diamagnetic Nia-S to the paramagnetic Nia-C,39 characterized by a NiIII-H-FeII

active site configuration. The presence of a bridging hydride has been ascertained by advanced isotope

specific EPR techniques using H/D (Hydrogen/Deuterium exchange) labelling of the active site.40-42 At this

stage the [NiFe] site can accept another electron and a proton, populating the “fully reduced” Nia-SR

species which is characterized by a NiII-H-FeII electronic configuration of the active site and a protonated

terminal cysteine.7,43-45 Nia-SR is diamagnetic and therefore EPR silent. Evidence of a bridging hydride has

been provided by an ultra-high resolution structure of the O2-sensitive [NiFe] hydrogenase from

Desulfovibrio vulgaris Miyazaki F (DvMF) and the detection of Fe-H wagging modes in NRVS experiments

on the same enzyme.45 The Nia-SR usually comprises at least three sub-forms, named Nia-SR, Nia-SR’, and

Nia-SR’’. All these species can be discriminated exclusively using IR spectroscopy, monitoring the CO

stretching vibration of the hydrogenase active site that is highly sensitive to perturbations such as

protonation changes at nearby residues and redox changes at the [NiFe] site and/or the proximal [FeS]

cluster. The structural differences of the various Nia-SR sub-forms as well as their role in catalysis are not

clear, yet,19,29,43,46 and a few groups have proposed that they might differ in the protonation state of nearby

amino acid residues.

The Nia-L intermediate was originally discovered upon light exposure of Nia-C under cryogenic conditions.

Nia-L represents a tautomer of the Nia-C, where the two electrons of the bridging hydride are stored

formally on the Ni ion, resulting in a formal NiI-FeII electronic configuration, and the residual proton is

located at one of the Ni-bound terminal cysteines. Evidence of cysteine protonation has been provided

by comprehensive IR investigations on the DvMF hydrogenase3 and recently also of the regulatory [NiFe]-

hydrogenase4 from Cupriavidus necator. There are at least three Nia-L sub-forms characterized by IR and

EPR spectroscopy.7,46-51 As for the Nia-SR sub-forms, the difference between the Nia-L states is not

completely enlightened and relevant studies are sparse.4,51,52 More about this intermediate will be

presented in Chapter 4.

Finally, the removal of one electron and one proton from Nia-L restores the Nia-S intermediate. There is

an ongoing debate on whether this step takes place via a proton-coupled electron transfer (PCET) or a

sequential electron-transfer (ET) - proton-transfer (PT) mechanism.52,53

1.4 Spectroscopic tools for the characterization of the redox structural states of the

active site

Infrared (IR) and Resonance Raman (RR):

IR and Raman spectroscopy offer valuable insights into various aspects of metalloenzymes. When a

metalloenzyme harbors IR or Raman active species in its catalytic center, such as the CO and CN– ligands

in hydrogenases, or when its substrate/inhibitor exhibits IR/Raman activity, the corresponding

spectroscopic method becomes a powerful tool for gaining information about the different redox

structural states of the active site.

Figure 1.4: Example of a baseline-corrected infrared spectrum of a reduced [NiFe] hydrogenase. The intramolecular

stretching vibrations of the diatomic CO and CN– (symmetric and asymmetric) ligands are marked in red. The

spectrum was measured by Armel Waffo Tadjoung.

IR spectroscopy is used for the characterization of the redox structural states of the catalytic cycle. In

[NiFe] hydrogenases the Fe(CO)(CN)2 moiety gives rise to one carbonyl stretching vibration v(CO) and the

symmetric and antisymmetric CN stretching modes vs(CN) and vas(CN), which are very sensitive to the

oxidation state, structure and geometry of the active site. The CO vibrations are detected at 1800-2020

cm–1 and the CN vibrations in the 2050-2100 cm–1 range.

Figure 1.5: RR spectra of [NiFe] active site. A: Low-frequency region upon excitation with 568 nm with (left) metal-S

modes and (right) Fe-CO/CN modes. B: [FeS] clusters signals observed upon 458 nm excitation.

RR spectroscopy focuses on the vibrational modes that involve metal ligand modes and can be used for a

more detailed characterization of the accessible redox state of the active site, also in respect to their

interaction with the other redox cofactors of the protein such as the [FeS] clusters. When several metal

sites are present in the protein the detection and correct assignment of the metal ligand vibrations

becomes a challenge. In [NiFe] hydrogenases the Fe-CO/CN coordinates give rise to normal modes in the

region of 450-700 cm–1 (Fig. 1.5A right) and metal-S vibrations below 400 cm–1 (Fig. 1.5A left). The active

site metal-S vibration signals are typically observed upon 568 nm excitation (or higher wavelength) and

those signals drop for a 458 nm excitation line. On the contrary, the [FeS] cluster signals are noted in the

region between 250-450 cm–1, when a 458 nm is used and the signals typically drop when a 568 nm

excitation line is used.

EPR

Redox states of the catalytic cycle with one or more unpaired electrons can be studied with electron

paramagnetic resonance (EPR). Among the information which can be obtained is the oxidation and spin

state of the metal. Additionally, the interaction between different metal centers in the protein can be

detected, such as between the Ni-Fe cofactor and the [FeS] clusters.7

XAS

X-ray absorbance spectroscopy (XAS) is a method that offers information about the oxidation state, ligand

and geometry of the metal center. An advantage of XAS compared to EPR is the ability to detect and study

EPR-silent states.54-60

Mössbauer

Mössbauer spectroscopy is used to study the nuclear structure with the absorption and emission of γ-

rays. The sample is exposed to a beam of γ radiation and the detector measures the intensity of the

transmitted beam. Solid or frozen samples, ensure that the nucleus stays intact upon gamma radiation

absorption/emission. In the [NiFe] hydrogenases, isotopic labelling of the Fe sites to 57Fe is required and

the method is utilized for the investigation of the electronic structure, spin state and coordination sphere

of the Fe ions. This method can be e.g. used for the study of the large subunit of the hydrogenases and

the maturation intermediates61,62 and redox structural states of the entire enzyme. In addition, it can also

be used for multi-cluster enzymes to characterize the electronic states and their coordination.19,63-65

NRVS

When a synchrotron source is used Nuclear Resonance Vibrational Spectroscopy (NRVS) measures the

vibrational properties of the Mössbauer-active nuclei, mostly Fe57. NRVS focuses on the vibrational

properties of nuclei whereas Mössbauer mainly focuses on electronic and magnetic properties of the

sample. In hydrogenases, similar to RR spectroscopy, one can observe the Fe-CO and Fe-CN vibrations.45,66

The advantage of this method over RR is that the NRVS spectra do not require coupling to electronic

transitions of the cofactor.67 Fe-CO and Fe-CN modes are strong in the region from 440 to 640 cm–1 and

the Fe-S cluster modes below 440 cm–1.45 However, to some extent Fe-CO vibrations are covered by the

Fe-S clusters’ vibrations, thus, the application of NRVS, for characterizing the isolated large subunits of

the hydrogenases, is valuable to disentangle the contributions of the different cofactors. In particular, if

they are in vitro reconstituted after selective Fe57 labelling with their small subunit. 66,68 Finally, they are

useful for the characterization of the catalytically active states, such as the Nia-SR by the observation of

the Fe-H stretching mode.45

1.4 Cupriavidus necator Membrane-Bound Hydrogenase

In this paragraph the structural and functional details of the O2-tolerant membrane-bound [NiFe]

hydrogenase from Cupriavidus necator are introduced, which is used as a model system for the

spectroscopic investigations of this dissertation. The host organism is a Gram-negative hydrogen-

metabolizing (Knallgas) chemolithotrophic bacterium. It hosts four different [NiFe] hydrogenases: a

regulatory hydrogenase (RH), a membrane-bound hydrogenase (MBH), which is described in more detail

here, a soluble NAD+ (nicotinamide adenine dinucleotide)-reducing hydrogenase (SH) and an

actinobacterial-like hydrogenase (AH).

Figure 1.6: The heterotrimeric form of the membrane-bound hydrogenase, composed of the large subunit HoxG

(blue), the small subunit HoxK (green), which hosts the [FeS] cluster and HoxZ (yellow), harboring the diheme

cytochrome (yellow), as natural electron acceptor, and anchors the protein to the membrane. This figure is adapted

by Chem. Rev. 2014, 114, 8, 4081–4148. 19

The MBH is located on the outer cytoplasmic membrane, recovering energy for the cell by channeling

electrons into the respiratory chain. Its native form is comprised of a trimeric complex, HoxGKZ, where

HoxG is the catalytic large subunit housing the [NiFe] active site, HoxK is the small electron-transferring

subunit using [FeS] clusters as electron relay and HoxZ is a membrane-integral diheme cytochrome b

harboring subunit that transfers the electrons from the hydrogenase module (HoxGK) to the respiratory

chain. In the cytoplasmic membrane, the MBH is found in a supramolecular arrangement, consisting of

three heterotrimers.69

The MBH is O2-tolerant, namely it retains the catalytic ability (for H2 oxidation) in the presence of O2.35,70

Several reports have related the O2-tolerance mainly to the presence of an unusual 6 Cys-ligated [4Fe–

3S] cluster in HoxK, able to shuttle two electrons for O2 reduction under physiologically relevant redox

potentials.35,70,71

1.4.1 Large subunit HoxG-Maturation Process

The maturation of [NiFe]-hydrogenases is a sophisticated process, the details of which are not yet fully

understood. This endeavor involves the gradual introduction of a bi-metal active site into the apo-protein,

a procedure facilitated by a complex machinery of at least six maturases, namely HypA-F.5,72

The most crucial step in [NiFe] hydrogenase biosynthesis is the assembly of the NiFe(CN)2CO core. The

CN– ligand is synthesized from carbamoyl phosphate and catalyzed by the HypE/HypF proteins,73 while

the CO ligand is formed from formyl-tetrahydrofolate.74,75 This process is assumed to follow a specific

mechanism due to the potential toxicity of these ligands for the cell. The details of the biosynthesis of the

CO ligand from formyl-tetrahydrofolate under anaerobic conditions remain elusive. Under aerobic

conditions an additional maturation protein namely HypX seems to be involved in the CO ligand

synthesis.74-76 The HypCD complex serves as a scaffold for the assembly of the Fe(CN)2(CO) moiety. The

cyanide group after being synthesized is transferred from HypE to the Fe, which is coordinated by the

HypCD complex. As depicted in Figure 1.7, the HypCD complex is then responsible for the insertion of the

Fe(CN)2CO core into the apo-protein. Subsequently, the synthesis and incorporation of the Ni ion are

carried out by the maturases HypA and HypB.19,77 Further the C-terminus extension, is cleaved by an

endopeptidase (HoxM),78,79 leading to the fully mature large subunit .

Figure 1.7: [NiFe]-hydrogenase large subunit assembly steps (1-4) and the different maturases involved. From top to

bottom, the first up to the last step of the maturation process are displayed. The fully mature large subunit (4)

accommodates the bimetallic [NiFe] site, with the Ni ion covalently bound to the protein scaffold via four cysteine

residues (Cys 1-4). Hereby, Cys2 and Cys4 serve as bridging cysteines between the Ni and the Fe in the fully mature

protein and a third bridging position may be occupied by a hydroxyl ligand, a bridging hydride, or it can remain

vacant, depending on the specific redox conditions. The figure is adapted from Nature Chemical Biology volume 19,

pages 498–506 (2023).62

The role of the C-terminal extension in the maturation process is not fully understood. Several [NiFe]-

hydrogenases carry a C-terminal extension, which is cleaved by a hydrogenase-specific endopeptidase

after the insertion of the nickel and the Fe(CN)2(CO) into the active-site.62,80,81 However, this was debated

when it was discovered that the large subunit of E.coli Hyd2 could form a complex with the small subunit,

even when it lacks the Ni and the Fe(CN)2(CO) complex.81 Already in 2020, Hartmann et.al.,82 followed

another approach to investigate the role of the C-terminal extension in the maturation process. The C-

terminal extension was removed by genetic engineering and resulted in fully-mature MBH but in low

concentrations in the membrane.82

1.5 Hydrogenophilus Thermoluteolous HtSH

The soluble cytosolic [NiFe] hydrogenase of Hydrogenophilus thermoluteolus (HtSH) consists of four

subunits, as shown in Fig. 1.8. This protein couples the H2 oxidation with the NAD+ reduction.20,83,84 HoxH

hosts the [NiFe] cofactor, HoxF the three [FeS] clusters (2 x [4Fe4S] clusters and one [2Fe2S]), HoxU one

FMN (Flavin Mononucleotide) and one [4Fe4S] cluster, HoxY one [4Fe4S] cluster.85-87

Hydrogenophilus thermoluteolus TH-1 is an aerobic, chemolithoautotrophic, H2-oxidizing β-

proteobacterium.88 HtSH is an oxygen-tolerant hydrogenase with a maximum catalytic activity at 80 0C.83

The above-mentioned properties make this hydrogenase a good candidate for biotechnological

applications. Not only because of its O2-tolerance but also its thermostability.83 Other soluble

hydrogenases such us the one of Cupriavidus necator, have been shown to be additionally involved in

NAD(P)H cofactor regeneration, similarly to CnSH.89

Figure 1.8: The soluble cytosolic [NiFe] hydrogenase of Hydrogenophilus thermoluteolus (HtSH). It couples NAD+

reduction with H

oxidation, taking place in the diaphorase and hydrogenase module, respectively. Each of the [FeS]

clusters functions as an electron transfer relay between the two active sites.

In the oxidized protein the C-terminus of HoxF ranges from the interface between HoxF and HoxY to the

NADH active site of HoxF. Upon H2 reduction, the C-terminus is located between HoxF and HoxY. Although

a crystallization artifact cannot be excluded, it is assumed that this happens to prevent the reduced FMN

from being involved in the formation of ROS species upon oxygen attack. It is questionable whether the

FMN dissociation takes place in the cell, when the H2 oxidation is coupled to the NAD+ reduction and the

clusters are not completely reduced.87,90

In 2017 the crystal structures of HtSH in its air-oxidized (as-isolated) form and after H2 reduction were

published. Herein, an unprecedented hexa-coordinated nickel ion was revealed, with one terminal

cysteine, three instead of two bridging cysteines to the iron ion and a carboxyl group of a glutamate,

serving as bidentate ligand to the nickel. Further insights were given later by Kulka et.al.25 One of the most

interesting findings was that this structure probably serves as a protection mechanism of this hydrogenase

from O2 attack. The authors observed that upon H2-reduction and subsequent fast O2 reoxidation, a

species with unusually high CO stretching vibration (1993 cm–1) was enriched. It was proposed, that Ni

exists in this species in +IV oxidation state, with three bridging cysteines and an E32 coordinating the Ni

in bidentate fashion, as suggested by the initial structural data. Herein, C462 which is terminally bound to

the Ni in the reduced structure, occupies the third bridging position between both active-site metals

completing so the sixth ligand coordination of the Ni. The Ni(IV)/Fe(II) closed cell structure was recently

debated and an alternative Ni(III)/Fe(III) structure was suggested.25,91 The formal oxidation state of the Fe

and Ni can presumably be resolved by 57Fe Mössbauer or/and EXAFS measurements, respectively, and

future studies may clarify this issue. In any case one has to consider that the representation of the

oxidation states by localized models might be oversimplified.91

In this study the focus lies mainly on the reduced protein which resides in a mixture of Nia-C and Nia-SR

species and on light and temperature triggered conversions between the species. The results show that

some species of the catalytic cycle are probably by-passed, opening new pathways in the potential

regulation of the [NiFe] active site with light.

1.6 References

1. Darensbourg, M. Y., Lyon, E. J. & Smee, J. J. The bio-organometallic chemistry of active site iron in

hydrogenases. Coord. Chem. Rev. 206–207, 533–561 (2000).

2. Volbeda, A. et al. Structure of the [NiFe] hydrogenase active site: Evidence for biologically

uncommon Fe ligands. J. Am. Chem. Soc. 118, 12989–12996 (1996).

3. Tai, H., Hirota, S. & Stripp, S. T. Proton Transfer Mechanisms in Bimetallic Hydrogenases. Acc.

Chem. Res. 54, 232–241 (2021).

4. Waffo, A. F. T. et al. Structural Determinants of the Catalytic Ni a ‑ L Intermediate of [ NiFe ] -

Hydrogenase. J. Am. Chem. Soc. 145, 13674–13685 (2023).

5. Lacasse, M. J. & Zamble, D. B. [NiFe]-Hydrogenase Maturation. Biochemistry 55, 1689–1701

(2016).

6. Casalot, L. & Rousset, M. Maturation of the [NiFe] hydrogenases. Trends Microbiol. 9, 228–237

(2001).

7. Lubitz, W., Ogata, H., Rüdiger, O. & Reijerse, E. Hydrogenases. Chem. Rev. 114, 4081–4148

(2014).

8. Vignais, P. M., Billoud, B. & Meyer, J. Classification and phylogeny of hydrogenases. FEMS

Microbiol. Rev. 25, 455–501 (2001).

9. Hydrogen as a Fuel: Learning from Nature. (Taylor & Francis, 2001).

10. Benoit, S. L., Maier, R. J., Sawers, R. G. & Greening, C. Molecular Hydrogen Metabolism: a

Widespread Trait of Pathogenic Bacteria and Protists. Microbiol. Mol. Biol. Rev. 84, 1–52 (2020).

11. Hernsdorf, A. W. et al. Potential for microbial H 2 and metal transformations associated with

novel bacteria and archaea in deep terrestrial subsurface sediments. ISME J. 11, 1915–1929

(2017).

12. Nandi, R. & Sengupta, S. Microbial Production of Hydrogen: An Overview. Crit. Rev. Microbiol. 24,

61–84 (1998).

13. Greening, C. et al. Genomic and metagenomic surveys of hydrogenase distribution indicate H 2 is

a widely utilised energy source for microbial growth and survival. ISME J. 10, 761–777 (2016).

14. Vignais, P. M. & Billoud, B. Occurrence, classification, and biological function of hydrogenases: An

overview. Chem. Rev. 107, 4206–4272 (2007).

15. Vincent, K. A., Parkin, A. & Armstrong, F. A. Investigating and Exploiting the Electrocatalytic

Properties of Hydrogenases. Chem.Rev. 107, 4366–4413 (2007).

16. Silveira, C. M. S. et al. Molecular Details on Multiple Cofactor Containing Redox Metalloproteins

Revealed by Infrared and Resonance Raman Spectroscopies. MDPI 26, (2021).

17. Happe, R. P., Roseboom, W., Plerik, A. J., Albracht, S. P. J. & Bagley, K. A. Biological activition of

hydrogen. Nature 385, 126 (1997).

18. Pierik, A. J., Schmelz, M., Lenz, O., Friedrich, B. & Albracht, S. P. J. Characterization of the active

site of a hydrogen sensor from Alcaligenes eutrophus. FEBS Lett. 438, 231–235 (1998).

19. Lubitz, W., Ogata, H., Ru, O. & Reijerse, E. Hydrogenases. Chem. Rev. 114, 4081–4148 (2014).

20. Lauterbach, L. et al. Nuclear resonance vibrational spectroscopy reveals the FeS cluster

composition and active site vibrational properties of an O2-tolerant NAD+-reducing [NiFe]

hydrogenase. Chem. Sci. 6, 1055–1060 (2015).

21. Horch, M., Lauterbach, L., Lenz, O., Hildebrandt, P. & Zebger, I. NAD(H)-coupled hydrogen cycling

- Structure-function relationships of bidirectional [NiFe] hydrogenases. FEBS Lett. 586, 545–556

(2012).

22. Fritsch, J., Lenz, O. & Friedrich, B. Structure, function and biosynthesis of O2-tolerant

hydrogenases. Nat. Rev. Microbiol. 11, 106–114 (2013).

23. Volbeda, A. et al. Crystal structure of the nickel-iron hydrogenase from Desulfovibrio gigas.

Nature 373, 580–587 (1995).

24. Volbeda, A. et al. Structure of the [NiFe] hydrogenase active site: Evidence for biologically

uncommon Fe ligands. J. Am. Chem. Soc. 118, 12989–12996 (1996).

25. Kulka-Peschke, C. J. et al. Reversible Glutamate Coordination to High-Valent Nickel Protects the

Active Site of a [NiFe] Hydrogenase from Oxygen. J. Am. Chem. Soc. 144, 17022–17032 (2022).

26. Tombolelli, D. & Mroginski, M. A. Proton Transfer Pathways between Active Sites and Proximal

Clusters in the Membrane-Bound [NiFe] Hydrogenase. J. Phys. Chem. B 123, 3409–3420 (2019).

27. Evans, R. M. et al. Mechanism of hydrogen activation by [NiFe] hydrogenases. Nat. Chem. Biol.

12, 46–50 (2016).

28. Kendall-Price, S. Mechanistic studies of H2 oxidation and evolution catalysed by NiFe

hydrogenase enzymes. PhD thesis, Fac. Phys. Sci. Univ. Oxford (2022).

29. Lubitz, W., Reijerse, E. & van Gastel, M. [NiFe] and [FeFe] hydrogenases studied by advanced

magnetic resonance techniques. Chem. Rev. 107, 4331–4365 (2007).

30. Fernandez, V. M., Hatchikian, E. C. & Cammack, R. Properties and reactivation of two different

deactivated forms of Desulfovibrio gigas. Biochim. Biophys. Acta 832, 69–79 (1985).

31. Albracht, S. P. J. Nickel hydrogenases: in search of the active site. Biochim. Biophys. Acta 1188,

167–204 (1994).

32. Cammack, R., Fernandez, V. M. & Scheider, K. Activation and active sites of nickel-containing

hydrogenases. Biochimie 68, 85–91 (1986).

33. Volbeda, A. et al. Crystallographic studies of [NiFe]-hydrogenase mutants: Towards consensus

structures for the elusive unready oxidized states. J. Biol. Inorg. Chem. 20, 11–22 (2015).

34. Barilone, J. L., Ogata, H., Lubitz, W. & Gastel, M. Van. Structural differences between the active

sites of the Ni-A and Ni-B states of the [ NiFe ] hydrogenase: an approach by quantum chemistry

and single crystal ENDOR spectroscopy. Phys. Chem. Chem. Phys. 17, 16204–16212 (2015).

35. Frielingsdorf, S. et al. Reversible [4Fe-3S] cluster morphing in an O 2 -tolerant [NiFe]

hydrogenase. Nat. Chem. Biol. 10, 378–385 (2014).

36. Horch, M. et al. Resonance Raman spectroscopy on [NiFe] hydrogenase provides structural

insights into catalytic intermediates and reactions. J. Am. Chem. Soc. 136, 9870–9873 (2014).

37. Bernhard, M. et al. The H2 sensor of Ralstonia eutropha: Biochemical characteristics,

spectroscopic properties, and its interaction with a histidine protein kinase. J. Biol. Chem. 276,

15592–15597 (2001).

38. Bruschi, M., Tiberti, M., Guerra, A. & De Gioia, L. Disclosure of key stereoelectronic factors for

efficient H2 binding and cleavage in the active site of [NiFe]-hydrogenases. J. Am. Chem. Soc. 136,

1803–1814 (2014).

39. Lacey, A. L. De, Ferna, M., Curie, M., Rousset, M. & Cammack, R. Activation and Inactivation of

Hydrogenase Function and the Catalytic Cycle : Spectroelectrochemical Studies. 4304–4330

(2007).

40. Brecht, M., Van Gastel, M., Buhrke, T., Friedrich, B. & Lubitz, W. Direct Detection of a Hydrogen

Ligand in the [NiFe] Center of the Regulatory H2-Sensing Hydrogenase from Ralstonia eutropha in

Its Reduced State by HYSCORE and ENDOR Spectroscopy. J. Am. Chem. Soc. 125, 13075–13083

(2003).

41. Foerster, S. et al. Single Crystal EPR Studies of the Reduced Active Site of [ NiFe ] Hydrogenase

from Desulfovibrio vulgaris Miyazaki F. J. Am. Chem. Soc. 9, 83–93 (2003).

42. Foerster, S., Gastel, M. Van & Brecht, M. An orientation-selected ENDOR and HYSCORE study of

the Ni-C active state of Desulfovibrio vulgaris Miyazaki F hydrogenase. J Biol Inorg Chem 51–62

(2005) doi:10.1007/s00775-004-0613-5.

43. De Lacey, A. L., Fernández, V. M., Rousset, M. & Cammack, R. Activation and inactivation of

hydrogenase function and the catalytic cycle: Spectroelectrochemical studies. Chem. Rev. 107,

4304–4330 (2007).

44. Ogata, H., Lubitz, W. & Higuchi, Y. [NiFe]hydrogenases: Structural and spectroscopic studies of

the reaction mechanism. Dalt. Trans. 9226, 7577–7587 (2009).

45. Ogata, H. et al. Hydride bridge in [NiFe]-hydrogenase observed by nuclear resonance vibrational

spectroscopy. Nat. Commun. 6, 1–8 (2015).

46. Ash, P. A., Hidalgo, R. & Vincent, K. A. Proton Transfer in the Catalytic Cycle of [NiFe]

Hydrogenases: Insight from Vibrational Spectroscopy. ACS Catal. 7, 2471–2485 (2017).

47. Brecht, M., Van Gastel, M., Buhrke, T., Friedrich, B. & Lubitz, W. Direct Detection of a Hydrogen

Ligand in the [NiFe] Center of the Regulatory H2-Sensing Hydrogenase from Ralstonia eutropha in

Its Reduced State by HYSCORE and ENDOR Spectroscopy. J. Am. Chem. Soc. 125, 13075–13083

(2003).

48. Pandelia, M. E. et al. Membrane-bound hydrogenase i from the hyperthermophilic bacterium

aquifex aeolicus: Enzyme activation, redox intermediates and oxygen tolerance. J. Am. Chem.

Soc. 132, 6991–7004 (2010).

49. Pierik, A. J., Roseboom, W., Happe, R. P., Bagley, K. A. & Albracht, S. P. J. Carbon monoxide and

cyanide as intrinsic ligands to iron in the active site of [NiFe]-hydrogenases. NiFe(CN)2CO,

biology’s way to activate H2. J. Biol. Chem. 274, 3331–3337 (1999).

50. Dole, F. et al. Spin-spin interactions between the Ni site and the [4Fe-4S] centers as a probe of

light-induced structural changes in active Desulfovibrio gigas hydrogenase. Biochemistry 35,

16399–16406 (1996).

51. Medina, M., Hatchikian, E. C. & Cammack, R. Studies of light-induced nickel EPR signals in

hydrogenase: comparison of enzymes with and without selenium. Biochimica et Biophysica Acta

vol. 1275 (1996).

52. Ash, P. A., Hidalgo, R. & Vincent, K. A. Proton Transfer in the Catalytic Cycle of [NiFe]

Hydrogenases: Insight from Vibrational Spectroscopy. ACS Catal. 7, 2471–2485 (2017).

53. Greene, B. L., Wu, C. H., McTernan, P. M., Adams, M. W. W. & Dyer, R. B. Proton-Coupled

Electron Transfer Dynamics in the Catalytic Mechanism of a [NiFe]-Hydrogenase. J. Am. Chem.

Soc. 137, 4558–4566 (2015).

54. Debeer, S. et al. In search of metal hydrides: an X-ray absorption and emission study of [NiFe]

hydrogenase model complexes. R. Soc. Chem. 18, 10688–10699 (2016).

55. Bagyinka, C., Whitehead, J. P. & Maroney, M. J. An X-ray Absorption Spectroscopic Study of

Nickel Redox Chemistry in Hydrogenase. J. Am. Chem. Soc. 115, 3576–3585 (1993).

56. Lambertz, C. et al. Electronic and molecular structures of the active-site H-cluster in [FeFe]-

hydrogenase determined by site-selective X-ray spectroscopy and quantum chemical

calculations. Chem. Sci. 5, 1187–1203 (2014).

57. Burgdorf, T. et al. Structural and Oxidation-State Changes at Its Nonstandard Ni - Fe Site during

Activation of the NAD-Reducing Hydrogenase from Ralstonia eutropha Detected by X-ray

Absorption, EPR, and FTIR Spectroscopy. J. Am. Chem. Soc. 576–592 (2005).

58. Fritsch, J. et al. [NiFe] and [FeS] Cofactors in the Membrane-Bound Hydrogenase of Ralstonia

eutropha Investigated by X-ray Absorption Spectroscopy: Insights into O 2 -Tolerant H 2

Cleavage. Biochemistry 50, 5858–5869 (2011).

59. Whitehead, J. P., Gurbiel, R. J., Bagyinka, C., Hoffman, B. M. & Maroney, M. J. The Hydrogen

Binding Site in Hydrogenase : 35-GHz ENDOR and XAS Studies of the Ni-C Active Form and the Ni-

L Photoproduct. J. Am. Chem. Soc. 115, 5629–5635 (1993).

60. Gu, Z. et al. Structure of the Ni Sites in Hydrogenases by X-ray Absorption Spectroscopy . Species

Variation and the Effects of Redox Poise. J. Am. Chem. Soc. 118, 11155–11165 (1996).

61. Caserta, G. et al. The large subunit of the regulatory [NiFe]-hydrogenase fromRalstonia eutropha-

a minimal hydrogenase? Chem. Sci. 11, 5453–5465 (2020).

62. Caserta, G. et al. Stepwise assembly of the active site of [NiFe]-hydrogenase. Nat. Chem. Biol. 19,

498–506 (2023).

63. Teixeira, M. et al. Redox Intermediates of Desulfovibrio gigas [NiFe] Hydrogenase Generated

Under Hydrogen. J. Biol. Chem. 264, 16435–16450 (1989).

64. Pereira, A. S., Tavares, P., Moura, I., Moura, J. G. & Quı, D. De. Mössbauer Characterization of the

Iron - Sulfur Clusters in Desulfovibrio Vulgaris Hydrogenase. J. Am. Chem. Soc. 123, 2771–2782

(2001).

65. Roncaroli, F. et al. Cofactor composition and function of a H2-sensing regulatory hydrogenase as

revealed by Mössbauer and EPR spectroscopy. Chem. Sci. 6, 4495–4507 (2015).

66. Caserta, G. et al. Hydroxy-bridged resting states of a [NiFe]-hydrogenase unraveled by cryogenic

vibrational spectroscopy and DFT computations. Chem. Sci. 12, 2189–2197 (2021).

67. Kamali, S. et al. Observation of the Fe-CN and Fe-CO Vibrations in the Active Site of [ NiFe ]

Hydrogenase by Nuclear Resonance Vibrational Spectroscopy. Angew. Commun. 52, 724–728

(2013).

68. Caserta, G. et al. In Vitro Assembly as a Tool to Investigate Catalytic Intermediates of [NiFe]-

Hydrogenase. ACS Catal. 10, 13890–13894 (2020).

69. Radu, V., Frielingsdorf, S., Evans, S. D., Lenz, O. & Jeuken, L. J. C. Enhanced oxygen-tolerance of

the full heterotrimeric membrane-bound [NiFe]-hydrogenase of Ralstonia eutropha. J. Am. Chem.

Soc. 136, 8512–8515 (2014).

70. Goris, T. et al. A unique iron-sulfur cluster is crucial for oxygen tolerance of a [NiFe]-hydrogenase.

Nat. Chem. Biol. 7, 310–318 (2011).

71. Schmidt, A. et al. Stepwise conversion of the Cys6[4Fe-S] to a Cys4[4Fe-4S] cluster and its impact

on the oxygen tolerance of [NiFe]-hydrogenase. Chem. Sci. 14, 11105–11120 (2023).

72. Böck, A. et. al. Maturation of Hydrogenases. Adv. Microb. Physiol. 51, 1–72 (2006).

73. Reissmann, S. et al. Taming of a Poison : Biosynthesis of the NiFe-Hydrogenase Cyanide Ligands.

Science (80-. ). 299, 1067–1070 (2003).

74. Schulz, A. et al. Formyltetrahydrofolate Decarbonylase Synthesizes the Active Site CO Ligand of O

2 ‑ Tolerant [ NiFe ] Hydrogenase. J. Am. Chem. Soc. 1457–1464 (2020)

doi:10.1021/jacs.9b11506.

75. Bürstel, I. et al. CO synthesized from the central one-carbon pool as source for the iron carbonyl

in O2-tolerant [NiFe]-hydrogenase. PNAS 113, 14722–14726 (2016).

76. Buhrke, T. & Friedrich, B. hoxX ( hypX ) is a functional member of the Alcaligenes eutrophus hyp

gene cluster. Arch. Microbiol. 170, 460–463 (1998).

77. Lacasse, M. J., Summers, K. L., Khorasani-motlagh, M., George, G. N. & Zamble, D. B. Bimodal

Nickel-Binding Site on Escherichia coli [NiFe]-Hydrogenase Metallochaperone HypA. Inorg. Chem.

58, 13604–13618 (2019).

78. Thiemermann, S. et al. Carboxyl-Terminal Processing of the Cytoplasmic NAD-Reducing

Hydrogenase of Alcaligenes eutrophus Requires the hoxW Gene Product. J. Bacteriol. 178, 2368–

2374 (1996).

79. Binder, U., Maier, T. & Böck, A. Nickel incorporation into hydrogenase 3 from Escherichia coli

requires the precursor form of the large subunit. 165, 69–70 (1996).

80. Senger, M., Stripp, S. T. & Soboh, B. Proteolytic cleavage orchestrates cofactor insertion and

protein assembly in [NiFe]-hydrogenase biosynthesis. J. Biol. Chem. 292, 11670–11681 (2017).

81. Pinske, C., Thomas, C., Nutschan, K. & Sawers, R. G. Delimiting the Function of the C-Terminal

Extension of the Escherichia coli [ NiFe ] -Hydrogenase 2 Large Subunit Precursor. 10, (2019).

82. Hartmann, S., Frielingsdorf, S., Caserta, G. & Lenz, O. A membrane-bound [ NiFe ] -hydrogenase

large subunit precursor whose C-terminal extension is not essential for cofactor incorporation

but guarantees optimal maturation. Microbiologyopen 9, 1–10 (2020).

83. Preissler, J. et al. Enzymatic and spectroscopic properties of a thermostable [NiFe]‑hydrogenase

performing H2-driven NAD+-reduction in the presence of O2. Biochim. Biophys. Acta - Bioenerg.

1859, 8–18 (2018).

84. Lauterbach, L., Lenz, O. & Vincent, K. A. H2-driven cofactor regeneration with NAD(P)+-reducing

hydrogenases. FEBS J. 280, 3058–3068 (2013).

85. Lauterbach, L., Idris, Z., Vincent, K. A. & Lenz, O. Catalytic properties of the isolated diaphorase

fragment of the NAD +-reducing [NiFe]-hydrogenase from Ralstonia eutropha. PLoS One 6,

(2011).

86. Horch, M., Lauterbach, L., Lenz, O., Hildebrandt, P. & Zebger, I. NAD(H)-coupled hydrogen cycling

- Structure-function relationships of bidirectional [NiFe] hydrogenases. FEBS Lett. 586, 545–556

(2012).

87. Shomura, Y. et al. Structural basis of the redox switches in the NAD+-reducing soluble [NiFe]-

hydrogenase. Science. 357, 928–932 (2017).

88. Hayashi, N. R., Ishida, T., Yokota, A., Kodama, T. & Igarashi, Y. Hydrogenophilus thermoluteolus

gen. nov., sp . nov ., a thermophilic , facultatively chemolithoautotrophic , hydrogen-oxidizing

bacterium. Int. J. Syst. Bacteriol. 49, 783–786 (1999).

89. Lauterbach, L. & Lenz, O. Catalytic production of hydrogen peroxide and water by oxygen-

tolerant [NiFe]-hydrogenase during H2 cycling in the presence of O2. J. Am. Chem. Soc. 135,

17897–17905 (2013).

90. Schulz, A. C. Biochemical and spectroscopic insights into peculiar active site structures and

biosynthesis of O2-tolerant [NiFe]-hydrogenases. (2021).

91. Kumar, R. & Stein, M. The Fully Oxidized State of the Glutamate Coordinated O2-Tolerant [NiFe]-

Hydrogenase Shows a Ni(III)/Fe(III) Open-Shell Singlet Ground State. J. Am. Chem. Soc. 145,

10954–10959 (2023).

2. Methods, techniques and materials

Molecular vibrations can be probed by IR absorption or Raman spectroscopy. Both techniques were

developed in the middle of the last century and are nowadays indispensable tools in analytical science. A

detailed discussion about the following methods can be found in 1,2 . The following chapters follow the

formalism of 1,3–7.

2.1 Molecular Vibrations

For a description of molecular vibrations, the atoms in a molecule are described as point masses,

connected by massless springs. The spring represents the interactions between the atoms. As an example,

a simplified system of a diatomic molecule A–B can be considered. In case the two atoms are forced to

move along the x-axis by 𝛥𝛥𝑥𝑥 from the equilibrium position, the displacement is related to the restoring

force 𝐹𝐹𝑥𝑥 which is given by Hooke’s law:

𝐹𝐹𝑥𝑥=−𝑘𝑘𝛥𝛥𝑥𝑥 2.1

where 𝑘𝑘 is the force constant, which is a measure of the rigidity of the spring or bond. The potential energy

𝑉𝑉 depends on the displacement from the equilibrium position:

𝑉𝑉=1

2𝑘𝑘𝛥𝛥𝑥𝑥2 2.2

The kinetic energy T of the oscillating motion is given by:

𝑇𝑇=1

2𝜇𝜇(𝛥𝛥𝑥𝑥󰇗)2

2.3

where 𝜇𝜇 is the reduced mass described by:

𝜇𝜇=𝑚𝑚𝐴𝐴𝑚𝑚𝐵𝐵

𝑚𝑚𝐴𝐴+𝑚𝑚𝐵𝐵 2.4

Due to the conservation of energy in a closed system without any external forces, the following relation

is given:

0 = 𝑑𝑑𝑇𝑇

𝑑𝑑𝑑𝑑+𝑑𝑑𝑉𝑉

𝑑𝑑𝑑𝑑 2.5

which can also be expressed as:

0 = 1

2𝑑𝑑(𝛥𝛥𝑥𝑥󰇗2)

𝑑𝑑𝑑𝑑 +1

2𝑘𝑘𝑑𝑑(𝛥𝛥𝑥𝑥2)

𝑑𝑑𝑑𝑑 2.6

This eventually leads to the Newton equation of motion:

𝑘𝑘𝜇𝜇𝛥𝛥𝑥𝑥+ 𝑑𝑑2𝛥𝛥𝑥𝑥

𝑑𝑑𝑑𝑑2= 0 2.7

The solution of the above equation has the form:

𝛥𝛥𝑥𝑥=𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴(𝜔𝜔𝑑𝑑+𝜑𝜑) 2.8

Here 𝐴𝐴, 𝜔𝜔, and 𝜑𝜑, are the amplitude, circular frequency and phase of the vibrational motion, respectively.

Introducing equation (2.8) into (2.7) leads to equation (2.9), which correlates the circular frequency with

the spring constant and the reduced mass.

𝜔𝜔=�𝑘𝑘𝜇𝜇 2.9

In order to transform the circular frequency into wavenumbers 𝑣𝑣� (cm–1), one obtains

𝑣𝑣�=1

2𝜋𝜋𝐴𝐴�𝑘𝑘𝜇𝜇 2.10

where c is the speed of light. From equation (2.10), it becomes clear that the frequency of a vibration

increases by increasing bond strength and decreasing reduced mass.

2.2 Normal modes

If we now consider a larger molecule, with N atoms, each atom can be displaced in all three x-, y- and z-

directions of the Cartesian coordinate system. Thus, the molecule has in total 3N degrees of freedom, but

not all of them are vibrational motions. A non-linear molecule has 3 translational and 3 rotational degrees

of freedom, whereas a linear molecule has 3 translational and 2 rotational degrees of freedom. Thus, the

remaining 3N-6 modes for a non-linear molecule and the respective 3N-5 for a linear molecule correspond

to vibrational degrees of freedom.

These vibrations are generally denoted as normal modes. To estimate the frequencies of these normal

modes, we express the kinetic and potential energy in terms of the displacement of each atom of the

molecule with regard to the directions of the Cartesian coordinate system. Then, the kinetic energy is

obtained by:

𝑇𝑇=1

2�𝑚𝑚𝑎𝑎��𝑑𝑑𝛥𝛥𝑥𝑥𝑎𝑎

𝑑𝑑𝑑𝑑 �2+�𝑑𝑑𝛥𝛥𝑦𝑦𝑎𝑎

𝑑𝑑𝑑𝑑 �2+�𝑑𝑑𝛥𝛥𝑧𝑧𝑎𝑎

𝑑𝑑𝑑𝑑 �2�

𝑁𝑁

𝑎𝑎=1 2.11

Using mass-weighted Cartesian displacement coordinates,

𝑞𝑞𝑖𝑖=�𝑚𝑚𝑖𝑖𝛥𝛥𝑥𝑥𝑖𝑖,𝑞𝑞𝑖𝑖+1=�𝑚𝑚𝑖𝑖𝛥𝛥𝑦𝑦𝑖𝑖,𝑞𝑞𝑖𝑖+2=�𝑚𝑚𝑖𝑖𝛥𝛥𝑧𝑧𝑖𝑖 ( 𝑓𝑓𝐴𝐴𝑓𝑓 𝑒𝑒𝑒𝑒𝐴𝐴ℎ 𝑒𝑒𝑑𝑑𝐴𝐴𝑚𝑚 𝑒𝑒=𝑖𝑖)

the following expression for the kinetic energy is obtained:

𝑇𝑇=1

2��𝑑𝑑𝑞𝑞𝑖𝑖

𝑑𝑑𝑑𝑑�

3𝑁𝑁

𝑖𝑖=1

2 2.12

Potential energy comprises all possible interactions of all individual atoms. This includes not only the

bonding interactions, but also non-bonding interactions such as electrostatic or van der Waals. The

potential energy is expressed in a Taylor series with respect to the coordinates 𝑞𝑞𝑖𝑖.

𝑉𝑉=𝑉𝑉0+��𝜕𝜕𝑉𝑉

𝜕𝜕𝑞𝑞𝑖𝑖�0

3𝑁𝑁

𝑖𝑖=1 𝑞𝑞𝑖𝑖+1

2�� 𝜕𝜕2𝑉𝑉

𝜕𝜕𝑞𝑞𝑖𝑖𝜕𝜕𝑞𝑞𝑗𝑗�0𝑞𝑞𝑖𝑖

3𝑁𝑁

𝑖𝑖,𝑗𝑗=1 𝑞𝑞𝑗𝑗+⋯ 2.13

We are interested in changes of the potential energy derived by displacements of the individual atoms

and thus the first term (potential energy at the equilibrium) and the second term (as infinitesimal changes

of the 𝑞𝑞𝑖𝑖 from the equilibrium position) can be omitted. Considering the harmonic approximation, higher

order terms can be neglected and therefore, the potential energy expression can be simplified to:

𝑉𝑉≈1

2�� 𝜕𝜕2𝑉𝑉

𝜕𝜕𝑞𝑞𝑖𝑖𝜕𝜕𝑞𝑞𝑗𝑗�0=1

2�𝑓𝑓𝑖𝑖𝑗𝑗𝑞𝑞𝑖𝑖𝑞𝑞𝑗𝑗

3𝑁𝑁

𝑖𝑖,𝑗𝑗=1 2.14

where 𝑓𝑓𝑖𝑖𝑗𝑗 are the force constants.

According to Newton’s equation of motions the total energy is given by (2.15) in analogy to equation (2.5):

𝑑𝑑

𝑑𝑑𝑑𝑑𝜕𝜕𝑇𝑇

𝜕𝜕𝑞𝑞𝚥𝚥+𝜕𝜕𝑉𝑉

𝜕𝜕𝑞𝑞𝑗𝑗= 0 2.15

Inserting (2.12) and (2.14) to (2.15), we obtain the following expression:

𝑞𝑞𝚥𝚥󰇘+�𝑓𝑓𝑖𝑖𝑗𝑗

3𝑁𝑁

𝑖𝑖=1 𝑞𝑞𝑖𝑖= 0 2.16

Equation (2.16) is equivalent to equation (2.17) for the diatomic harmonic oscillator, but it represents not

one, but instead 3N linear second order differential equations, with a general solution:

𝑞𝑞𝑖𝑖=𝐴𝐴𝑖𝑖cos�√𝜆𝜆𝑑𝑑+𝜑𝜑� 2.17

In the aforementioned equation 𝐴𝐴 and √𝜆𝜆 and 𝜑𝜑 are representing the amplitude, the frequency and the

phase, respectively.

This system of differential equations has 3N solutions for 𝜆𝜆, corresponding to 3N frequencies 𝜆𝜆1

2, among

them 6 (5) solutions are zero in the case of non-linear (linear) molecules, corresponding to the

translational and rotational degrees of freedom. The characteristic amplitudes 𝐴𝐴𝑖𝑖𝑖𝑖 are essential to

describe quantitatively the displacement of each atom with the respective normal mode 𝑘𝑘. In each normal

mode, all atoms vibrate in-phase with the same frequency, but different amplitudes.

For a mathematical description of the probability of vibrational transitions, a more simple and compact

presentation of the normal modes is desirable, rather than the Cartesian coordinates. Therefore, the

mass-weighted Cartesian coordinates 𝑞𝑞𝑖𝑖 are converted into normal coordinates 𝑄𝑄𝑖𝑖 via an orthogonal

transformation according to:

𝑄𝑄𝑖𝑖=�𝑙𝑙𝑖𝑖𝑖𝑖

3𝑁𝑁

𝑖𝑖=1 𝑞𝑞𝑖𝑖 2.18

The transformational coefficients 𝑙𝑙𝑖𝑖𝑖𝑖 are chosen, so that 𝑇𝑇 and 𝑉𝑉 can be expressed as in equations (2.12)

and (2.14). The potential energy does not depend on cross-products 𝑄𝑄𝑖𝑖∙𝑄𝑄𝑖𝑖′ (𝑤𝑤𝑖𝑖𝑑𝑑ℎ 𝑘𝑘≠𝑘𝑘′), resulting in

the following solution of the secular determinant:

𝑄𝑄𝑖𝑖=𝐾𝐾𝑖𝑖cos��𝜆𝜆𝑖𝑖𝑑𝑑+𝜑𝜑� 2.19

Finally, a third coordinate system can be employed which is based on intuitive internal coordinates such

as stretching, bending, out-of-plane deformation, and torsional coordinates.

2.3 Vibrational Spectroscopy

Light can be attenuated or transmitted by matter being absorbed as well as elastically or inelastically

scattered. In case that the incident photon matches the energy gap between the vibrational ground state

and the excited state, the photon may be absorbed and the molecule is promoted to a higher energy level,

named excited state. This change can be measured by absorption spectroscopy, which measures the loss

of energy. Another possibility is that the photon is scattered. The photon can be scattered elastically,

retaining its energy or wavelength (Rayleigh scattering) or inelastically, altering its wavelength (anti-

Stokes and Stokes Raman scattering). All the possible transitions are schematically represented by the

following energy diagram in Fig.2.1.

IR and Raman spectroscopy follow different selection rules, which determine different probabilities of the

active modes for each technique. The selection rules for each measurement differ and are determined i.a.

by symmetry and group theory. The two methods can be used complementarily for the characterization

of optically active modes of the molecules. Both methods are used in industry and research as analytical

tools providing structural information of the probed molecules.

Figure 2.1: Energy diagram, depicting different vibrational transitions involved, Rayleigh, Raman Stokes, anti-Stokes

and Resonance Raman scattering. The vibrational states of a molecule in the ground electronic state can be probed

either by directly measuring the absolute frequency, as in the case of IR absorption or the relative frequency named

Raman shift (Stokes and anti-Stokes) of the allowed transitions.8,9

2.3.1 Infrared absorption

Analogous to other optical methods, the extent of IR absorption is depicted as the absorbance (OD =

optical density) using the Lambert-Beer law:

𝐴𝐴=−lg �𝐼𝐼

𝐼𝐼0�=𝜀𝜀𝐴𝐴𝑙𝑙 2.20

Here, 𝐼𝐼 represents the amount of IR radiation, which is passing through an analyte solution. The reference

intensity is represented by 𝐼𝐼0. The obtained absorbance depends on the molar extinction coefficient 𝜀𝜀,

the concentration of the absorbing species 𝐴𝐴 and the optical path length 𝑙𝑙.

Nowadays, spectrometers work based on the Fourier Transform (FT) principle. The central unit of an (FT)IR

instrument is the Michelson interferometer (see Figure 2.2), that is based on the general set-up developed

by Michelson in 1891.10,11

2.4 The Michelson Interferometer

Figure 2.2: Scheme of a Michelson Interferometer. Reprinted from 3,5.

The interferometer consists of two plane mirrors, oriented perpendicular to each other and a beam

splitter, which divides the beam into two paths. These beams are reflected by the two mirrors and

subsequently recombined, leading to a constructive or destructive interference that depends on the

pathlength difference. The combined beam then passes through the sample and the modulation of the

interference due to the vibrational transitions is detected relative to a reference interferogram. To convert

the interferogram (intensity as a function of the pathlength difference) into a spectrum (intensity as a

function of the frequency) a Fourier transformation is required.

2.5 Attenuated Total Reflectance – Infrared Spectroscopy

Attenuated total reflection (ATR) IR spectroscopy is a special technique to probe vibrations of immobilized

molecules. The technique makes use of Snell’s law (Figure 2.3A), which describes the angle at which the

radiation is refracted when it passes from one transparent medium to the other. We assume that n1>n2

or in other words 1 is an optically dense medium and 2 an optically rare medium. When θ>θc, then the

beam reflects internally with an angle of reflection equal to the angle of incidence.

Figure 2.3: Illustrative depiction of the attenuated total reflection (ATR) principle and its mechanism of action

described by Snell’s law (A), showing the reflection on an interface between two materials with different refractive

indices n1 and n2. Inspired by 10. In (B) the application of Snell’s law in the case of an ATR prism is depicted. The IR

beam passes through the internal reflection element (IRE) and is totally reflected at the IRE-sample interface. An

evanescent wave, which decays with distance from the interface, penetrates the sample at each reflection. Inspired

by 12.

Any material that exhibits internal reflection is known as an internal reflection element (IRE), which is

commonly a prism or crystal. ATR IR spectroscopy is a specific method using an optical material such as a

prism with a high refractive index to create an evanescent wave when an IR beam is reflected on the

sample-prism interface (Figure 2.3B). The evanescent wave penetrates into the less dense sample and is

then absorbed or “attenuated” by the sample, if the angle of the incidence is smaller than the critical

angle. Materials with a high refractive index are silicon, diamond, zinc, selenite or germanium. In this

study, a silicon prism was used with a refractive index equal to 3.41 and a critical angle of 26°.

The strength of the oscillating electric field 𝐸𝐸R of the evanescent wave decays with the distance z (z-

direction), from the reflecting plane:

𝐸𝐸=𝐸𝐸0𝑒𝑒�− 𝑧𝑧

𝑑𝑑𝑝𝑝� 2.21

where 𝐸𝐸0 is the electric field strength of the evanescent wave at the reflecting plane. The penetration

depth 𝑑𝑑𝑝𝑝 , is the depth in which the wave penetrates the probed solution or material. The penetration

depth is a function of the incident radiation in a vacuum, λ, the ratio of the refractive indexes in between

the optically dense / rare medium and finally the angle of the incidence beam.

𝑑𝑑𝑝𝑝=𝜆𝜆

2𝜋𝜋𝜋𝜋1�𝐴𝐴𝑖𝑖𝜋𝜋2𝜃𝜃−�𝜋𝜋2

𝜋𝜋1�2 2.22

Typically, the penetration depth is found to be in the range of the wavelength of the incident radiation.

For an incident radiation angle of 600 and a spectral region of 1000 to 4000 cm–1 (10 – 2.5 μm), the

penetration depth of the evanescent wave is in between 2.6 and 0.7 μm. The parallel (p) and

perpendicular (s) component of the incident radiation cause a polarization of the evanescent wave in the

x- and y- directions.

Since the penetration depth of the evanescent wave is rather small, the strong absorption of media such

as water is less pronounced. Among others the advantage of ATR is also that the buffer in contact with

the target material can be exchanged and the effect of the buffer on the experiment can be studied (pH,

ionic strength). In parallel the sample can be dried above the ATR prism and the effect of the buffer

absorption can be eliminated. Additionally, on top of the ATR prism orientation-dependent experiments

can be performed and by addition of the substrate the kinetics of the reactions can be followed in-situ.

2.6 Surface-Enhanced Infrared Absorption Spectroscopy

A method that is often used when IR spectroscopy and electrochemistry are utilized at the same time is

the so-called surface-enhanced IR absorption spectroscopy (SEIRAS). In this context it is mostly employed

in the ATR mode as described above. The SEIRA effect refers to the enhancement of the IR signal from

molecules adsorbed on nanostructured Ag or Au films by a factor of 10 to 1000 depending on the metal

surface. Signals are selectively enhanced in the close vicinity up to approximately 8 nm from the metal

surface. The electromagnetic and the chemical mechanism (Figure 2.4) are mainly responsible for the

observed signal enhancement. The electromagnetic mechanism is based on plasmon resonance and the

perturbation of the optical properties, whereas the chemical mechanism can be attributed to donor-

acceptor interactions and charge-transfer effects. SEIRA-enhancement depends on the angle of the

incident light and the polarization of the IR radiation. Additionally, the Au film can also function as the

working electrode. Thus, electrochemical approaches can be applied. SEIRA experiments can be carried

out in static or time-resolved modes.

Figure 2.4: Representation of the electromagnetic (EM) and chemical mechanism of surface enhanced infrared

absorption (SEIRA). The metal particles are depicted as ellipses where a is the semimajor axis and b the semiminor

one. p is the dipole moment and δp the additionally induced dipole of the metal island, denoted to the molecular

vibration of the adsorbed molecules. Adapted from 3,5.

2.7 Electromagnetic (EM) Mechanism-plasmon resonance

The incident IR radiation polarizes the metal “islands” by excitation of collective electron resonances. The

term metal islands stands for metal atoms or nanoparticles that are typically supported on a substrate.

Metal islands can exhibit surface plasmon resonance, a phenomenon where the conduction electrons in

the metal oscillate collectively when exposed to electromagnetic radiation. The induced dipole p,

generates a local EM field, which is polarized perpendicularly to the surface at every point of the metal.

The local electromagnetic field consists of two components the incident electric field and the induced

electric field:

The enhanced electric field is polarized along the surface normal and its magnitude decays with the sixth

power of the distance from the metal surface:

𝐸𝐸

�



𝑙𝑙𝑙𝑙𝑙𝑙(𝑣𝑣0) = 𝐸𝐸

�



0(𝑣𝑣0) + 𝐸𝐸

�



𝑖𝑖𝑖𝑖𝑑𝑑(𝑣𝑣0) 2.23

|𝐸𝐸𝑙𝑙𝑙𝑙𝑙𝑙𝑎𝑎𝑙𝑙|2=4𝑝𝑝2

𝑑𝑑6

2.24

The equation 2.24 is the origin of the short-range enhancement. The enhancement factor 𝐹𝐹 at a distance

𝑑𝑑 from the surface of a nanoparticle with a radius 𝑒𝑒0 will be:

𝐹𝐹(𝑑𝑑)=𝐹𝐹(0) �𝑒𝑒0

𝑒𝑒0+𝑑𝑑�6 2.25

The electromagnetic field interacts with the adsorbed molecules causing vibrational state transitions. The

particle size determines the extent of the enhancement factor. By using the Bruggemann effective

medium theory, it is shown that the enhancement is particularly large for ellipsoidal particles 𝑎𝑎

𝑏𝑏 > 1. The

electromagnetic theory also accounts for the surface selection rules of SEIRA. Accordingly, the

enhancement is strongest for modes with a dipole moment perpendicular to the surface. The selection

rules can, therefore, be used to determine the orientation of a molecule or a part of a molecule with

respect to the surface.

Chemical Mechanism

The vibrations of chemisorbed molecules are better enhanced compared to those of the physisorbed

molecules. Chemical adsorption, also known as chemisorption, takes place when molecules are attached

to a solid surface by chemical bond. The interaction between the metal particles and the adsorbate

changes the vibrational polarizability of the molecules. A charge transfer exhibited by charge oscillations

between the metal nanoparticles and the adsorbate increases the absorption coefficient.

2.8 IR Spectroscopy on proteins

A typical IR spectrum of a protein contains information about its structure. It includes the modes of the

polypeptide backbone which provide insight into the secondary structure. Additionally, the IR spectra of

metalloenzymes containing CO, CN– ligands such as the [NiFe] hydrogenases, can also provide information

on the different redox states of the active site.

The folding of the primary structure results in the secondary structure such us a-helix, β-helix, β-sheet,

turns and random coils, stemming from the specific conformations of the peptide bonds. Hence,

information about the secondary structure can be gained by the analysis of the modes of the peptide

bonds, specifically of the amide I and amide II modes. The amide I is mainly due to the C=O stretching with

a minor contribution of the NH in-plane bending, whereas the amide II results from a combination of the

NH in-plane bending and the CN stretching. (Table 2.1).

Table 2.1: Wavenumber regions representing specific amide modes.

Wavenumber (cm–1) Group vibration

amide A

3310-3270

N-H stretching

amide B 3010-3030 N-H stretching with 1st amide II overtone

amide I

1700-1600

C=O stretching/minor N-H bending contribution

amide II

1580-1480

N-H bending/C-N stretching

amide III

1300-1230

N-H bending/C-N stretching

The most informative mode for the secondary structure of proteins are the amide I and amide II. Amide I

is very sensitive to the secondary structure. Amide II is also informative for the secondary structure but

the correlation between secondary structure and frequency is often less straightforward.

2.8.1 Absorption spectroscopy

The sample is placed in between two CaF2 windows with a refractive index equal to 1.4. One of the

windows contains a cavity ranging from 2- 6 µm. Between the CaF2 windows a spacer can be placed to

increase the optical path length. Without spacer, the effect of water absorbance is rather low and the

amide I band of a protein sample which is covered by the water bending absorption for pathlengths larger

than 10 μm, can be accessed.

The ratio between the initial light intensity 𝐼𝐼0 and 𝐼𝐼 (after passing through the sample) is expressed by

the transmittance (Eq. 2.26 and 2.27). The relationship between the absorbance, the transmittance and

the Lambert-Beer (Eq.2.20), are expressed by the following equations.

𝑇𝑇=𝐼𝐼

𝐼𝐼0 2.26

𝑇𝑇(%) = 100 𝐼𝐼

𝐼𝐼0 2.27

By Lambert-Beer’s law:

𝐴𝐴=log10𝐼𝐼0

𝐼𝐼=−log10𝑇𝑇=εcl 2.28

𝜀𝜀 = molar absorption coefficient 𝑀𝑀−1𝐴𝐴𝑚𝑚−1

𝐴𝐴= molar concentration M

𝑙𝑙=optical path length cm

2.8.2 Difference spectroscopy

As mentioned above the active site of hydrogenases consists of a bimetallic [NiFe] site. A four coordinated

Ni ion that is coordinated by four cysteine thiolates, two of which are bridging to the Fe ion that is ligated

by one CO and two CN–, forming a [Fe(CO)(CN)2] moiety. The stretching vibrations of the carbonyl and

cyanide ligands give rise to three distinct IR absorptions. The first one corresponds to the carbonyl

stretching vibration (v(CO)), whereas the second and third band are related to the symmetric and

antisymmetric stretching (vs(CN) and vas(CN)) modes from the cyanide ligands (Fig. 2.5). These vibrational

modes are very sensitive to the electronic structure of the [NiFe] active site. They typically appear

between 1800 – 2100 cm–1, a region of the IR spectrum which is free of spectral contributions from the

protein or water. Therefore, these peaks are suitable marker bands to study [NiFe] hydrogenases.

In the absorbance spectrum of a protein there are strongly overlapping regions containing a multitude of

modes which are derived from the different constituents or/and redox structural states of the active site.

Figure 2.5: The active site region of the IR spectrum. Reprinted from the open access journal Chem. Sci.,

2019,10, 8981

-8989.

Difference spectroscopy derived between two defined conditions is a common way to simplify a spectrum

and selectively highlight spectral changes induced by a trigger, such as light.

In Fig. 2.6 in spectrum (1) the CO/CN vibrations of the active site of a [NiFe] hydrogenase recorded in the

dark are shown. The enzyme resides in a mixture of active site states. Therefore, one can see multiple CO

and CN stretching modes in the spectral data.

In the CN stretching region several bands belonging to different active site states overlap with each other.

Spectrum (2) depicts the peaks upon illumination, in which some of the states did not photoconvert. One

can calculate the difference spectrum (2)-(1), which exclusively displays the bands that are related to light

driven changes, while all other bands cancel out.13–17

Figure 2.6: Difference transmission mode IR spectroscopy. An example of a difference spectrum derived from the

subtraction of spectrum 1 from spectrum 2, is shown as blue

trace,

in which the negative and positive peaks refer to

photoconverted species and photoproducts, respectively. Spectrum (1) is recorded in the dark, whereas spectrum

(2) is reco

rded under light exposure.

Difference spectroscopy is also used in this study to follow redox-induced changes in the sample at the

active site and to detect conformational changes, degradation, and reorientation or desorption when the

protein is immobilized on top of the surface (SEIRA).

2.8.3 Functionalized gold surface-Self-Assembled Monolayer (SAM)

To avoid denaturation, proteins are not immobilized directly on the Au surface, only after covering it with

a Self-Assembled Monolayer (SAM) of organic molecules. The charge of the headgroup can be positive,

negative, neutral or hydrophobic, resulting in a different orientation of the immobilized protein. SAMs are

created when organic molecules residing in a liquid or gas phase align themselves in an ordered manner

on a substrate after chemisorption. A suitable combination of substrate material and organic molecule is

essential for this process.18 SAM formation reduces the system's free energy, reaching a

thermodynamically favored state. They act as insulating films, with stable thiol-metal bonds, making them

resilient to various solvents and a wide potential and temperature range. The strength of the thiol-metal

bond is in the order of 100 kJ/mol, meaning that the functionalized surfaces are stable.18 Metal surfaces

like gold, silver, copper, aluminum oxide, silver oxide, or glass are commonly used substrates for SAMs.

Gold is widely used because closed-packed monolayers are formed on top of this metal, and it exhibits

substantial chemical stability. Amphiphilic alkanethiols, represented as R-(CH2)n-SH chains, form SAMs on

gold surfaces, stabilized by van der Waals forces.18 The choice of the functional group R depends on the

protein's properties.18

SAM thickness typically ranges from 1-2 nm, and a small amount of solution is sufficient to cover a large

area. Ethanol is commonly utilized because of its solvating capabilities, low cost, high purity, and low

toxicity. The chosen solvent also has an impact on the SAM formation kinetics. Non-polar solvents like

heptane or hexane can accelerate SAM formation, while longer hydrocarbons slow it down.18 The lower

the ethanol concentration, the longer is the required immersion time.18

The scientific community as well as the industry are interested in immobilizing proteins on surfaces,

especially for biotechnological applications like chromatography and biosensors.19 Immobilization can be

reversible (e.g., adsorption, disulfide bonds) or irreversible (e.g., covalent coupling, entrapment,

crosslinking). SAMs with different headgroups (positive, negative, hydrophobic) aim to immobilize

proteins with diverse orientations, and other factors like protonation degree, chain length, buffer

ionization strength and temperature are considered for electrochemical control of the protein.20–22 In

Figure 2.7 the immobilized large subunit HoxG, of CnMBH (Cupriavidus Necator Membrane Bound

Hydrogenase) is depicted as an example.

Figure 2.7: The large subunit (HoxG) of the CnMBH immobilized on top of carboxy-terminated-coated Au. MD

(Molecular Dynamics) simulated graph was kindly provided by Dr. Jovan Dragelj / Prof. Mroginski group.

For more stable, yet irreversible attachment, the formation of a covalent bond between the protein and

the SAM is sometimes required. In this study, we initially chose a negatively charged SAM which was then

esterified by the addition of N-(3-dimethylamino propyl)-N-ethyl carbodiimide hydrochloride (EDC) and

N-Hydroxysuccinimide (NHS). Herein, the SAM was first esterified and then the proteins were covalently

bound via their surface-exposed lysines.23–25

2.8.4 SEIRA prism as the electrode - Electrochemistry (EC)

Since SEIRA spectroscopy is based on the IR absorption, a prism is used as an ATR-element but it can be

concomitantly utilized by coating it chemically with nanostructured gold, as an electrode for the careful

control or measurement of the potential and electric current through the cell. It is typically used in a three-

electrode configuration. The redox reactions of interest are taking place at the gold (Au) working electrode

(WE). Generally, the counter electrode (CE) is an inert conductor like platinum (Pt), as used also in this

study. The potential is set against a third electrode, named the reference electrode (RE). The reference

electrode should have a constant potential since current flows only through the working and counter

electrode. When proteins are immobilized on the electrode surface, the goal is to achieve direct electron

transfer (DET) for adequate electrochemical control. If DET occurs, then the currents are proportional to

the catalytic turnover rate of the immobilized (bio)catalyst. In chronoamperometry the potential is set to

a constant value and the current is recorded as a function of time. In this case, the current is proportional

to the available amount of substrate. When cyclic voltammetry (CV) is used, the potential is swept with a

fixed scan rate back and forth, and the current response is recorded. Under non-turnover conditions, the

midpoint potential of the cofactors of the enzyme can probably be estimated, while under turnover

conditions the experimentalist can gain information related to the reversibility of the catalyzed reaction,

e.g. the onset potential, the potential of deactivation, and the catalytic turnover rate.

Figure 2.8: Schematic representation of the spectroelectrochemical SEIRA cell. The nanostructured Au film, which

serves as the IR signal amplifier, is also used as the working electrode. Usually, a Pt mesh is used as counter electrode

(CE) and a (3 M KCl) Ag/AgCl as reference electrode (RE). The SAM is formed on top of the Au coated Si surface. The

protein is immobilized on top of the functionalized electrode as shown above. Direct immobilization on top of the

gold could cause protein denaturation.

2.9 Electron Paramagnetic Resonance

Electron paramagnetic resonance (EPR) spectroscopy is based on a fundamental physical property of

molecules or in our case enzymes harboring several redox-active metal centers that have unpaired

electrons (paramagnetic). EPR spectroscopy examines the interaction of an external magnetic field with

the unpaired electrons of the molecule. A variety of information can be obtained such us the oxidation

state and spin state of a metal or the interactions between metal centers within the metalloenzyme.

Generally, isolated electrons are characterized by an intrinsic mechanical angular momentum called spin

𝑆𝑆. The angular momentum is a vector property, defined by the modulus and the direction in space. Thus,

the electron spin can be found in two states (𝑆𝑆= ± 1

2), with different orientations with respect to an

external magnetic field but the same magnitude of the angular momentum. The angular momenta of

quantum particles are in the order of ℎ2𝜋𝜋

⁄, where ℎ is the Planck constant. The magnitude of 𝑆𝑆 is equal

to:

Where S=1/2 is the electron quantum spin number and then |S|=�3 4

⁄. Sz can either be 1/2 or -1/2 and

these two spins have the same energy, if no external magnetic field is applied.

Zeeman Effect

Figure 2.9: Schematic illustration of the electron spin in a magnetic field.

A crucial point in the theory of EPR spectroscopy is 𝜇𝜇𝑒𝑒, the magnetic moment, which is associated with the

electron spin angular momentum, where 𝜇𝜇𝑒𝑒 is proportional to 𝑆𝑆 and 𝜇𝜇𝑒𝑒 and 𝑆𝑆 vectors are parallel to each

other, but with opposite directions.

|𝑆𝑆|=�𝑆𝑆(𝑆𝑆+ 1) 2.29

𝜇𝜇𝑒𝑒=𝑔𝑔𝜇𝜇𝛣𝛣𝑆𝑆 2.30

The g number is called Landé factor and for a free electron is equal to 𝑔𝑔= 2.002319, 𝜇𝜇𝛣𝛣=−|e|h

4πme=

9.27410−24JT−1, where 𝑚𝑚𝑒𝑒 is the electron mass, 𝑒𝑒 the electron charge, ℎ= 6.626𝑥𝑥10−34 𝐽𝐽𝐴𝐴 is the Planck

constant and 𝜇𝜇𝐵𝐵 the atomic unit of the magnetic moment, which is called the Bohr magneton.26

When a constant external magnetic field (𝐵𝐵) is applied, then the energy of the magnetic moment is given

by the following equation:

𝐸𝐸=𝑔𝑔|𝜇𝜇𝛣𝛣|𝐵𝐵𝑙𝑙𝑆𝑆𝑧𝑧 2.31

Where 𝐵𝐵0 is the magnetic field intensity. Considering that the electron spin can be in two states either

1/2 or -1/2 then:

The splitting of the electron spin energy level into two levels in the presence of a magnetic field is called

Zeeman effect.26

Electron Zeeman Interaction – g values of hydrogenases cofactors

The following term referred to as spin Hamiltonian was introduced to describe the interaction between

electrons and nuclear spins and is expressed as follows27,28.

𝐻𝐻0

�=𝛨𝛨𝛦𝛦𝛦𝛦

�+𝛨𝛨𝛮𝛮𝛦𝛦

�+𝛨𝛨𝐻𝐻𝐻𝐻

�+𝐻𝐻𝐸𝐸𝐸𝐸

�+𝐻𝐻𝑁𝑁𝑁𝑁

� 2.33

Where: 𝐻𝐻𝐸𝐸𝐸𝐸

�=𝐸𝐸𝑙𝑙𝑒𝑒𝐴𝐴𝑑𝑑𝑓𝑓𝐴𝐴𝜋𝜋 𝑍𝑍𝑒𝑒𝑒𝑒𝑚𝑚𝑒𝑒𝜋𝜋 𝑖𝑖𝜋𝜋𝑑𝑑𝑒𝑒𝑓𝑓𝑒𝑒𝐴𝐴𝑑𝑑𝑖𝑖𝐴𝐴𝜋𝜋

𝐻𝐻𝑁𝑁𝐸𝐸

�=𝑁𝑁𝑁𝑁𝐴𝐴𝑙𝑙𝑒𝑒𝑒𝑒𝑓𝑓 𝑍𝑍𝑒𝑒𝑒𝑒𝑚𝑚𝑒𝑒𝜋𝜋 𝑖𝑖𝜋𝜋𝑑𝑑𝑒𝑒𝑓𝑓𝑒𝑒𝐴𝐴𝑑𝑑𝑖𝑖𝐴𝐴𝜋𝜋

𝐻𝐻𝐻𝐻𝐻𝐻

�=𝐻𝐻𝑦𝑦𝑝𝑝𝑒𝑒𝑓𝑓𝑓𝑓𝑖𝑖𝜋𝜋𝑒𝑒 𝑖𝑖𝜋𝜋𝑑𝑑𝑒𝑒𝑓𝑓𝑒𝑒𝐴𝐴𝑑𝑑𝑖𝑖𝐴𝐴𝜋𝜋

𝐻𝐻𝐸𝐸𝐸𝐸

�=𝐸𝐸𝑙𝑙𝑒𝑒𝐴𝐴𝑑𝑑𝑓𝑓𝐴𝐴𝜋𝜋−𝑒𝑒𝑙𝑙𝑒𝑒𝐴𝐴𝑑𝑑𝑓𝑓𝐴𝐴𝜋𝜋 𝑖𝑖𝜋𝜋𝑑𝑑𝑒𝑒𝑓𝑓𝑒𝑒𝐴𝐴𝑑𝑑𝑖𝑖𝐴𝐴𝜋𝜋

𝐻𝐻𝑁𝑁𝑁𝑁

�=𝑁𝑁𝑁𝑁𝐴𝐴𝑙𝑙𝑒𝑒𝑒𝑒𝑓𝑓 𝑞𝑞𝑁𝑁𝑒𝑒𝑑𝑑𝑓𝑓𝐴𝐴𝑝𝑝𝐴𝐴𝑙𝑙 𝑖𝑖𝜋𝜋𝑑𝑑𝑒𝑒𝑓𝑓𝑒𝑒𝐴𝐴𝑑𝑑𝑖𝑖𝐴𝐴𝜋𝜋

𝐸𝐸±= ± �1

2�𝑔𝑔|𝜇𝜇𝛣𝛣|𝛣𝛣0 2.32

In our system the most relevant Hamiltonians are the 𝐻𝐻𝐸𝐸𝐸𝐸

�, 𝐻𝐻𝐻𝐻𝐻𝐻

�, 𝐻𝐻𝐸𝐸𝐸𝐸

�.

For the study of the hyperfine interaction, site selective isotopic labeling is needed, for example Ni61,

which is out of the focus of this study.

The 𝐻𝐻𝐸𝐸𝐸𝐸

� depends on different phenomena expressed by different Hamiltonians, which are the 𝐻𝐻

�𝐸𝐸𝐸𝐸,

expressing the exchange coupling, the 𝐻𝐻

�𝐷𝐷𝐷𝐷, expressing the dipolar coupling and the 𝐻𝐻𝐸𝐸𝐻𝐻𝑍𝑍

� ,expressing zero-

field splitting.

The exchange coupling expresses electron-electron interactions which are significant when the orbitals of

two spins are overlapping. This interaction is relevant for the iron-sulfur clusters of the [NiFe]

hydrogenases, but only for metal centers which are not separated by more than 15 Å, because the

exchange interaction is mainly isotropic and decays exponentially as a function of the distance.

The dipolar interaction on the other hand is highly anisotropic and less dependent on the distance.

Therefore, electron-electron interactions prevail for distances larger than 15 Å. These interactions are

relevant when interpreting spectra of different active sites.

The zero field splitting expresses the magnetic interaction between individual unpaired electrons in the

absence of external magnetic field for a spin system with 𝑆𝑆> 1 2

⁄.

For the characterization of the metal cofactors of a hydrogen-cycling hydrogenase studied by EPR at 80 K,

maybe the most significant component is the electron Zeeman interaction described by the following

equation, which is applied to isotropic systems:

𝐻𝐻𝐸𝐸𝐸𝐸

�=𝑔𝑔𝑒𝑒𝜇𝜇𝐵𝐵𝐵𝐵0𝑆𝑆󰆹 2.34

Additionally, the spin of electrons which are bound to an atom, experience additional internal fields and

interact with the orbital angular momentum 𝐿𝐿�. In this so-called spin coupling orbital, the g-value differs

to the 𝑔𝑔𝑒𝑒 value of the free electron. The spin orbital coupling significantly contributes to the electron

Zeeman interactions in the case of transition metals, such as Ni and Fe which constitute the active site or

the active site clusters. The following Hamiltonian describes these interactions:

𝐻𝐻𝐸𝐸𝐸𝐸

�+𝐻𝐻𝐿𝐿𝑍𝑍

�=𝜇𝜇𝛣𝛣𝛣𝛣0�𝐿𝐿�+𝑔𝑔𝑒𝑒𝑆𝑆󰆹�+𝜆𝜆𝐿𝐿�𝑆𝑆󰆹 2.35

where 𝜆𝜆 is the coupling constant and increases with growing atomic masses.

The anisotropic interaction between an electron spin and the external magnetic field is characterized by

the g-tensor and described by the following matrix.

𝑔𝑔=�𝑔𝑔𝑥𝑥𝑥𝑥 0 0

0𝑔𝑔𝑦𝑦𝑦𝑦 0

0 0 𝑔𝑔𝑧𝑧𝑧𝑧� 2.36

The obtained values give rise to isotropic, axial and rhombic signals that generate the typical EPR

signatures. The interaction of an unpaired electron with 𝑆𝑆= 1/2 and an external magnetic field results in

two split energy levels, where the momentum of the electron adapts two orientations, either parallel

(𝑚𝑚𝑠𝑠=−1/2) or antiparallel orientation (𝑚𝑚𝑠𝑠= +1/2). The energy difference (𝛥𝛥𝛥𝛥) is between the two

states is given by:

𝛥𝛥𝐸𝐸=ℎ𝑣𝑣=𝑔𝑔𝑒𝑒𝜇𝜇𝐵𝐵𝐵𝐵0 2.37

Summing up, the paramagnetic centers of metalloproteins can be detected by EPR spectroscopy such as

the NiI, NiIII oxidation state of the active site. The paramagnetic Niu-A, Nir-B, Nia-C, Nia-L have been

intensively characterized by EPR spectroscopy, while the diamagnetic states, Nir/a/u/ia-S and Nia-SR, are not

accessible by this technique.

The [Fe-S] clusters are also detectable by EPR spectroscopy. The most common [Fe-S] clusters in [NiFe]

hydrogenases are [2Fe-2S], [3Fe-4S] and [4Fe-4S] clusters. The [Fe-S] clusters of the EPR spectra are

observed at cryogenic temperatures as low as 4 K, which is the limit for a helium cryostat. The [4Fe-4S]

and the [2Fe-2S] cluster are EPR active when they are (by one electron) reduced and the [3Fe-4S], when

they are oxidized. The active [Fe-S] states are summarized in the table below.

Table 2.2: [Fe-S] clusters and their redox states.

Cluster

EPR silent

EPR active

[2Fe-2S]

+2 (oxidized)

+1 (reduced)

[3Fe-4S]

0 (reduced)

+1 (oxidized)

[4Fe-4S]

+2 (oxidized)

+1 (reduced)

2.10 Experimental Procedures, Chemicals and Buffers

Table 2.3: Used chemicals in this work

Chemical

Distributor

Ethanol

Merck

N-Ethyl-Nʹ-(3-dimethylaminopropyl)-carbodiimid –hydrochlorid (EDC)

Merck

N-Hydroxysuccinimid

Merck

Ag/AgCl reference electrode

World Precision Instruments, Inc.

HCl

Merck

Fluka

K2HPO4 Merck

KH2PO4 Merck

Na2S2O3 ∙5H2O Sigma Aldrich

Na2SO3 Merck

NaAuCl4∙2H2O Sigma Aldrich

NaCl

Merck

NH4Cl Sigma Aldrich

NH4F Sigma Aldrich

Tris

Merck

Glycerol, HOCH2CH(OH)CH2OH Merck

8-Mercaptooctanoic acid thiol

Merck

8-Amino-1-octanethiol hydrochloride (C8H19NS∙HCl) Merck

1-Octanthiol

Merck

Sodium dithionite (Na2O4S2) Merck

Potassium hexacyanoferrate (III)

Merck

IR and SEIRA experiments

• The HoxG sample was purified in 50 mM KiPO4 and 150 mM NaCl, pH 7, by Dr. Giorgio Caserta.29

For the SEIRA experiments the protein was diluted either to 50 mM KiPO4 and 150 mM NaCl, at

pH 7 or pH 5.5. Different concentrations of the monovalent and divalent ions were used, in which

the effect of the ionic strength on the immobilization was investigated.

• For the covalently attached HoxG, the carboxyl-terminated surface was exposed to a 0.1 M NHS,

0.2M EDC solution, 50 mM KiPO4 solution, at pH 7, unless otherwise stated.

• Homologous HtSH was stored in a 25 mM Tris HCl, 150 mM NaCl, pH 7.4. For the cryo-IR

experiments of the reduced sample, the protein concentration was increased to 1mM (by gently

heating up to 500C) and therefore the final salt concentration was increased to 75 mM Tris HCl,

450 mM NaCl, pH 7.4. Afterwards 50% glycerol was added.

• Heterologously expressed HtSH was stored in 50 mM KiPO4, 20% glycerol, 5 mM MgCl2, 0.5 mM

NiCl2, protease inhibitor, pH = 7.2.

• Homologously expressed HtSH was stored in 50mM Tris. For cryo-IR measurements, the

concentration was between 700 μM - 1 mM.

• Strep-tagged and His-tagged CnMBH samples were prepared by Dr. Stefan Frielingsdorf, Lenz

group.30,31 For SEIRA measurements the final concentration was equal to 4 μM, for cryo-IR the

concentration was between 500 μM-1 mM.

D2O exchange for the CnMBH, was carried out by Stefan Frielingsdorf. 32

Size Exclusion Chromatography for HoxG

Measurements were run on an ÄKTA pure 25 using a Superdex 200 Increase 10/300 GL (Cytiva) column

equilibrated with the purification buffer at 4 °C at a flow rate of 0.5 mL/min. A calibration curve was made

by measuring six protein standards with known molecular weights between 12 and 670 kDa: Thyroglobulin

(669 kDa, 9.34 mL), Apoferritin (443 kDa, 10.49 mL), β-Amylase (200 kDa, 11.8 mL), Bovine serum albumin

(66 kDa, 14.1 mL), Carbonic anhydrase (29 kDa, 16.8 mL) and cytochrome C (12.3 kDa, 18.24 mL).

Additionally, HoxC (the large subunit of the regulatory hydrogenase from C. necator) was included in the

calibration series as its oligomerization profile was recently elucidated.

Fourier Transform Infrared (FTIR)-Transmission studies

For infrared (IR) transmission studies either a Bruker Vertex 70v or Tensor 27 Fourier Transform IR

spectrometer was used. Both are equipped with a liquid N2 mercury cadmium telluride detector (MCT).

All spectroscopic data were obtained and analyzed using Bruker OPUS software.

For the measurements at ambient temperatures, the temperature was usually adjusted at 100 C. An

amount of sample, approximately equal to 10 μL, was inserted into the sandwich configuration consisting

of two CaF2 windows. A polytetrafluoroethylene (PTFE) spacer of 50 μm with a hole in the center, defines

the needed volume and contributed to the proper sealing of the cell.

For measurements at cryogenic temperatures, the setup was slightly different. 2-6 μL were pipetted

inside the cell. The exact volume of the sample needed depends on the thickness of the Teflon spacer

which varies can vary from 10-100 μm. In the cryogenic experiments presented below the thickness of the

spacer used was 20 μm. The cell was placed into a cryostat used for studies at cryogenic temperatures

and filled with helium (He) gas. For this measurement, a Bruker Tensor 27 FTIR spectrometer was used.

On either side of the sample, LED lamps were placed (λ = 460 nm), for illumination at cryogenic

temperatures. For one single spectrum, 200 scans were averaged with a spectral resolution of 2 cm–1. 10

single spectra were averaged and the absorbance was calculated according to the Lambert-Beer law.

Au electroless deposition for SEIRA studies

An Au film was prepared by electroless deposition. First, the prism was polished with alumina powder

(Microgrit WCA-9) and then rinsed with water. To remove the oxide layer, 700 μL of 400 g/L NH4F were

pipetted on top of the Si prism for 2 min. The prism was then rinsed, dried, and then immersed in a bath

to reach 65 0C. On top the plating solution was placed. The plating solution was composed of the reduction

solution comprised of 0.3 M Na2SO3, 0.1 M Na2S2O3∙5H2O, and 0.1 M NH4Cl, the 2 % (w/w) HF solution and

the 0.03 M NaAuCl4∙2H2O solution. The three solutions were mixed in 1:1:1 volume ratio. The electroless

deposition was based on the reduction of AuIII and the associated oxidation of the Si surface according to

the following mechanism:

𝑆𝑆𝑖𝑖0(𝐴𝐴)+ 6𝐹𝐹−(𝑒𝑒𝑞𝑞)→ 𝑆𝑆𝑖𝑖𝐹𝐹62−(𝑒𝑒𝑞𝑞)+ 4𝑒𝑒−

𝐴𝐴𝑁𝑁𝐴𝐴𝑙𝑙4–(𝑒𝑒𝑞𝑞)+ 3𝑒𝑒−(𝑒𝑒𝑞𝑞)→ 𝐴𝐴𝑁𝑁0(𝑒𝑒𝑞𝑞)+ 4𝐴𝐴𝑙𝑙−(𝑒𝑒𝑞𝑞)

Electrochemical cleaning of Au deposited prism

After the electroless deposition the Au film was electrochemically cleaned. Six oxidation/reduction cycles

were applied in the range of 0.1 to 1.4 V in a 0.1 M H2SO4 solution. Before (15 min) and during the cyclic

voltammetry the buffer was purged with Ar. In Figure 2.10 below a typical CV is depicted. A single

reduction peak is observed, which is centered at ca. 920 mV. In order to calculate the area, the Au-oxide

reduction charge density, was used. The charge needed to reduce the surface Au-oxide was determined

from the area of the reduction peak compared to the specific charge density of 400 μC cm–2.5

Figure 2.10: Cyclic Voltamogram depicting the electrochemical cleaning of the nanostructured Au electrode in 0.1 M

H2SO4. The area of the gold surface is obtained by the reductive peak of the Au-oxide. The position of the peak Ered

= 0.917 mV.

SAM formation

Au electrodes were immersed for 12-16 hrs in ethanolic ω-mercaptan solution to form self-assembled

monolayers (SAM) on the electrode.

Specifically, the following mercaptan solutions were used, diluted to ethanol in a final concentration equal

to 1 mM:

8-Amino-1-octane-thiol (HSCH2(CH2)6CH2NH2)

8-Mercaptooctanoic acid (C8H16O2S)

1-Octanthiol (CH3(CH2)6CH2SH)

HtSH crystalization

The protein was crystallized by Dr. Andrea Schmidt, directly for the as-iso crystals. Approximately one

month is needed for the crystals to reach their maximum size.33For the reduced crystals the sample was

first reduced for 1 hr at 50 °C and the preparation of the crystal set-up took place inside an anaerobic glove

box. For the incubation of the reduced crystals a custom-made, gas-tight set-up was used, made from

poly-(methylmethacrylate) (PMMA).

To obtain re-oxidized crystals, directly after reduction the crystals were exposed to oxygen and the

crystallization takes place in aerobic conditions. Every time crystals grow at slightly different parameters

in terms of PEG concentration and seeding solution-to-protein ratio. Crystals grow either at 10 or 20 °C.

Table 2.4: Crystallization conditions

0,1 M Tris pH 8,5

0,2 M MgCl2

100

PEG3350 (50 %)

100

110

120

130

140

150

hanging drops

2 2 2 2 2 2

(amount)

protein µL

buffer µL

seeding stock µL

0.4

Microscope – IR

The crystals were transferred for measurement on the top of an IR-transparent and cryo-stable plate

which consisted of two parts. A 1.5 mm thick magnesium fluoride window was attached to a custom-made

PMMA frame, which fits into a holder from Linkam Scientific Instruments. The setup was built by Dr.

Christian Lorent and it allows combined IR-RR measurements to the same crystal. For further information,

the reader is referred to 28,34.

Thin layer Cryo-IR

HtSH or CnMBH sample was reduced by exposure to humidified gas, 100% H2, for 1 hr, at 50 0C or 10 0C,

respectively, unless indicated otherwise. The final concentration of the sample was equal to 600-700 μM.

The buffer contained 50 mM Tris-HCl pH 8.5, 300 mM NaCl, 50 % glycerol for HtSH and 50 mM KiPO4, 150

mM NaCl, pH 5.5, 50 % glycerol. After reduction, the sample was pipetted into an air-tight sandwich cell

consisting of two CaF2 windows, separated by a 50 μm spacer. The cell was then transferred into a

homemade liquid nitrogen cryostat. The cell compartment was constantly purged with dry N2. The sample

was measured by Bruker Tensor 27 FTIR spectrometer with a mercury cadmium telluride (MCT) detector.

For the data analysis, Bruker OPUS software 7.8 was used. Spectra with a resolution of 2 cm–1 were

recorded by averaging 200 scans. For the calculation of the absorbance spectra, first the reference

spectrum was calculated as the average of single channel spectra. The sample’s single-channel spectra

were also averaged. Absorbance spectra were calculated using the averaged protein spectra versus the

averaged buffer spectra as reference using Lambert Beer’s Law (Eq.2.20). The difference spectra light

minus dark were calculated, using the averaged-dark single channel spectrum as reference. The sample

was illuminated with 460 nm LED light.

Spectra Calculation

To obtain an absolute spectrum, the spectrum of the buffer was subtracted from the spectrum of the

sample. First the transmission spectra of the buffer and the sample were measured and converted into

absorption spectrum according to the following formula.

𝐴𝐴𝑎𝑎𝑏𝑏𝑠𝑠=−𝑙𝑙𝐴𝐴𝑔𝑔𝑇𝑇𝑠𝑠𝑎𝑎𝑠𝑠𝑝𝑝𝑙𝑙𝑒𝑒

𝑇𝑇𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑒𝑒𝑏𝑏 2.38

EPR spectroscopic studies of reduced HtSH

100 μL of HtSH reduced either with H2 or with NaDT and H2 was filled in EPR tubes which were purged

before with H2 inside the anaerobic tent. X-Band cw EPR spectra were recorded at a microwave frequency

of 9.3 GHz on a Bruker EMXplus spectrometer equipped with a high-sensitivity cylindrical mode resonator.

Figure 2.11: CaF

windows (2) polytetrafluoroethylene (PTFE) spacer of 10-50 μm with an opening in the center,

defines the needed volume and contributes to the proper sealing of the cell. (3) pipetted sample, the volume of

which is between 5

-10 μL.

2.11 References

1. Siebert, F. & Hildebrandt, P. Vibrational Spectroscopy in Life Science. (WILEY-VCH Verlag GmbH &

Co. KGaA,Weinheim, 2008). doi:10.1007/s00396-007-1816-4.

2. Bright Wilson, E. J., Decius, J. C. & Cross, P. C. Molecular Vibrations. The Theory of Infrared and

Raman Vibrational Spectra. (Dover Publications, Inc., 1955).

3. Kozuch, J. Structure-Function Relationships of Membrane Proteins - Spectroelectrochemical

Investigation of Artificial Membranes. PhD thesis, Univ. der Tech. Univ. Berlin (2012).

4. Forbrig, E. Investigation of membrane-active peptides and proteins by vibrational spectroscopy.

PhD thesis, Univ. der Tech. Univ. Berlin (2018).

5. Heidary, N. IR spectro-electrochemistry of an adsorbed oxygen-tolerant [NiFe] hydrogenase on

conductive surfaces. PhD thesis, Univ. der Tech. Univ. Berlin (2017).

6. Katz, S. Spectroscopic studies on hydrogenases. PhD thesis, Univ. der Tech. Univ. Berlin (2019).

7. Staffa, J. K. Electric fields at biomimetic interfaces – Spectro-electrochemical studies on the

vibrational Stark effect of artificial membranes. PhD thesis, Univ. der Tech. Univ. Berlin (2019).

8. Laun, K. Vibrational spectroscopic study on the bis-MGD cofactor in DMSO reductase enzymes.

PhD thesis, Univ. der Tech. Univ. Berlin (2023).

9. Gkogkou, D. Anisotropic plasmonic nanoparticle arrays for surface-enhanced biosensors. PhD

thesis, Univ. der Tech. Univ. Berlin (2017).

10. Griffiths, R. P. & De Haseth, J. Fourier transform infrared spectrometry. vol. 10 (John Wiley &

Sons, Inc., Hoboken, New Jersey, 2007).

11. Michelson, A. A. XXXVIII. On the application of interference-methods to spectroscopic

measurements. J. Sci. 31, 338–346 (1891).

12. Kendall-Price, S. Mechanistic studies of H2 oxidation and evolution catalysed by NiFe

hydrogenase enzymes. PhD thesis, Fac. Phys. Sci. Univ. Oxford (2022).

13. Tai, H., Hirota, S. & Stripp, S. T. Proton Transfer Mechanisms in Bimetallic Hydrogenases. Acc.

Chem. Res. 54, 232–241 (2021).

14. Waffo, A. F. T. et al. Structural Determinants of the Catalytic Ni a ‑ L Intermediate of [ NiFe ] -

Hydrogenase. J. Am. Chem. Soc. 145, 13674–13685 (2023).

15. Lorent, C. et al. Shedding Light on Proton and Electron Dynamics in [FeFe] Hydrogenases. J. Am.

Chem. Soc. 142, 5493–5497 (2020).

16. Moldenhauer, M. et al. Parameterization of a single H-bond in Orange Carotenoid Protein by

atomic mutation reveals principles of evolutionary design of complex chemical photosystems.

Front. Mol. Biosci. 1–13 (2023) doi:10.3389/fmolb.2023.1072606.

17. Ataka, K., Stripp, S. T. & Heberle, J. Biochimica et Biophysica Acta Surface-enhanced infrared

absorption spectroscopy ( SEIRAS ) to probe monolayers of membrane proteins ☆. BBA -

Biomembr. 1828, 2283–2293 (2013).

18. Johannes, G. V., Robert, J. F. & Keyes, T. E. Interfacial Supramolecular Assemblies. (John Wiley &

Sons Ltd, 2003).

19. Kyrpel, T. et al. Hydrogenase-based electrode for hydrogen sensing in a fermentation bioreactor.

(2023) doi:10.1016/j.bios.2023.115106.

20. Love, J. C., Estroff, L. A., Kriebel, J. K., Nuzzo, R. G. & Whitesides, G. M. Self-assembled

monolayers of thiolates on metals as a form of nanotechnology. Chem. Rev. 105, 1103–1169

(2005).

21. Bhadra, P. & Siu, S. W. I. Effect of Concentration, Chain Length, Hydrophobicity, and an External

Electric Field on the Growth of Mixed Alkanethiol Self-Assembled Monolayers: A Molecular

Dynamics Study. Langmuir 37, 1913–1924 (2021).

22. Casalini, S., Bortolotti, C. A., Leonardi, F. & Biscarini, F. Self-assembled monolayers in organic

electronics. Chem. Soc. Rev. 46, 40–71 (2017).

23. Sam, S. et al. Semiquantitative study of the EDC/NHS activation of acid terminal groups at

modified porous silicon surfaces. Langmuir 26, 809–814 (2010).

24. Harris, T. G. A. A. et al. Electrografted Interfaces on Metal Oxide Electrodes for Enzyme

Immobilization and Bioelectrocatalysis. ChemElectroChem 8, 1329–1336 (2021).

25. Sam, S. et al. Covalent immobilization of amino acids on the porous silicon surface. Surf. Interface

Anal. 42, 515–518 (2010).

26. Brustolon, M. & Giamello, E. RESONANCE ELECTRON RESONANCE A Practitioner ’ s Toolkit. JOHN

WILEY & SONS, INC., PUBLICATION (2009).

27. Abragam, A. & Pryce, M. H. L. Theory of the nuclear hyperfine structure of paramagnetic

resonance spectra in crystals. Proc. R. Soc. London. Ser. A. Math. Phys. Sci. 205, 135–153 (1951).

28. Lorent, C. Combining Electron Paramagnetic Resonance and Vibrational Spectroscopy to Explore

Catalysis of Hydrogen-converting Enzymes. PhD thesis, Univ. der Tech. Univ. Berlin (2022).

29. Caserta, G. et al. Stepwise assembly of the active site of [NiFe]-hydrogenase. Nat. Chem. Biol. 19,

498–506 (2023).

30. Lenz, O., Lauterbach, L., Frielingsdorf, S. & Friedrich, B. Oxygen-tolerant hydrogenases and their

biotechnological potential. in Biohydrogen (ed. Rögner, M.) (DE GRUYTER, 2015).

doi:10.1515/9783110336733.

31. Lenz, O., Lauterbach, L. & Frielingsdorf, S. O2 -tolerant [ NiFe ] -hydrogenases of Ralstonia

eutropha H16 : Physiology , molecular biology , purification , and biochemical analysis. Methods

Enzymol. 1–35 (2018) doi:10.1016/bs.mie.2018.10.008.

32. Glasoe, K. P. & Long, F. A. Use of glass electrodes to measure acidities in deuterium oxide. J. Phys.

Chem. 64, 188–190 (1960).

33. Shomura, Y. et al. Structural basis of the redox switches in the NAD+-reducing soluble [NiFe]-

hydrogenase. Science. 357, 928–932 (2017).

34. Lorent, C. et al. Exploring Structure and Function of Redox Intermediates in [NiFe]-Hydrogenases

by an Advanced Experimental Approach for Solvated, Lyophilized and Crystallized

Metalloenzymes. Angew. Chemie - Int. Ed. 60, 15854–15862 (2021).

3. Spectroscopic studies of the large subunit HoxG of the membrane-

bound hydrogenase from Cupriavidus necator

A prerequisite for [NiFe] hydrogenase biosynthesis is the assembly of the NiFe(CN)2CO moiety. Some

details of the mechanism are still not fully unraveled. In this part of the thesis, various isolated maturation

intermediates of the HoxG protein, the large subunit of CnMBH (Cupriavidus necator Membrane-Bound

Hydrogenase), are characterized by means of IR spectroscopy. Here, the data derived by IR spectroscopic

techniques are shown, which probe the stretching modes of the CO and CN– ligands of the Fe in the active

site. These modes are sensitive marker bands that report changes in the charge distribution, electronic

configuration, and structure of the [NiFe] active site.

Besides an extensive examination of the HoxG subunit in solution, we also focused on the immobilization

of this protein on surfaces using mercaptan-functionalized Au-nanostructured electrodes forming self-

assembled monolayers (SAMs). SAMs of different headgroups are utilized, exhibiting diverse charge and

protonation degree. The rational beyond the choice of different SAMs is that the various headgroups

might induce different orientation of the protein according to cationic, anionic, polar and hydrophobic

interactions. Subsequently, we employ a combination of spectroscopic and computational methods to

ascertain the orientation of the protein when interfaced on surfaces with different charge distributions.

Furthermore, the electrochemical control of the HoxG protein was also attempted. For this purpose, the

protein was additionally covalently bound. Specifically, a COOH-terminated SAM was treated with a

mixture of 1-Ethyl-3-(3-dimethylaminopropyl) carbodiimide and N-hydroxysuccinimide (EDC/NHS),

forming an NHS-ester. This procedure allowed the formation of a covalent amide bond between the

functionalized surface and the protein. With this process mainly surface-exposed lysine residues were

targeted. The established protocol on the HoxG subunit was then extended to the whole MBH enzyme.

3.1 HoxG studies in bulk

3.1.1 Maturation Intermediates

The presented research was driven by the prospect of harnessing HoxG for reactivity modulation via active

site modifications, including the incorporation of different metals. By gaining insights into the in vivo

assembly process, we aimed to develop strategies for the artificial maturation of [NiFe] hydrogenases.

This entails the incremental insertion of a bi-metal active site into the apo-protein, a process facilitated in

vivo by a complicated machinery of maturases.

Figure 3.1: Proposed model of the sequential maturation process of the [NiFe] hydrogenase NiFe(CN)2CO site. The

large subunit of [NiFe] hydrogenases hosts the [NiFe] heterobimetallic active site. Adapted from Caserta et al. 1

As detailed in the introduction, the Ni ion of the hydrogenase active site is covalently bound to the protein

scaffold via four cysteine residues of the large subunit. A schematic representation of the hydrogenase

large subunit is shown in Fig. 3.1 and the four cysteine residues are labelled Cys1-4 (Figure 3.1). Cys2 and

Cys4 serve as bridging ligands for the two metals while Cys1 and Cys3 bind exclusively to the Ni ion.

Depending on the redox conditions, a third bridging position may be occupied by a hydroxy ligand or a

bridging hydride. The most complicated step in the maturation of [NiFe] hydrogenases is the assembly of

the NiFe(CN)2CO cofactor. This goal is served by the elaborate “Hyp” protein machinery that involves at

least six proteins: HypA-F.1–3

In short, it has been proposed that the insertion of the Fe(CN)2CO fragment into the apo-hydrogenase

large subunit (Fig. 3.1.1) is accomplished by the HypCD complex, forming a large subunit intermediate

devoid of Ni. This latter has been named preHoxGΔNi in case of the large subunit HoxG of MBH (Fig. 3.1.2).

The insertion of the Ni ion is accomplished by the maturases HypA and HypB, resulting in a new large

subunit intermediate, preHoxG, which should be equipped with a fully matured NiFe(CN)2CO site but still

in its unprocessed form due to the presence of a C-terminal tail (Fig. 3.1.3). This intermediate should then

undergo a proteolytic cleavage of its C-terminal extension by the HoxM endopeptidase, leading to the

fully matured large subunit intermediate mHoxG, prone to interact with the small subunit to form a fully

active [NiFe]-hydrogenase (Fig. 3.1.4). All these proposed intermediates have been recently isolated and

were characterized by IR spectroscopy in collaboration with the group of Oliver Lenz. The spectra were

normalized to the amide II intensity and the vCO/ vCN region of the spectrum is depicted in Figure 3.2.1

In apo-HoxG protein, named preHoxGΔFeNi, only minuscule (vibrational) band intensities of the diatomic

ligands were detected (Fig. 3.2). In the case of preHoxGΔNi three main peaks were identified at 1949, 1962

and 1997 cm–1 in the vCO and at 2072 and 2092 cm–1 in the vCN stretching region (Fig. 3.2).

Figure 3.2: Baseline corrected IR spectra of the various HoxG maturation intermediates. Intensities were normalized

to the amide II band. The spectra display the characteristic vCO and vCN stretching modes of the hydrogenase

cofactor. Adapted from Caserta et al. 1

The IR spectrum of preHoxG, which still contains the C-terminus, shows a main CO absorption at 1953

cm−1 and two main CNs at 2066 and 2076 cm–1, together with a minor signal at 2090 cm−1. The mHoxG

exhibits two predominant CO bands at 1929 and 1939 cm−1 and a weaker absorption at 1953 cm–1. The

latter is assigned to a remaining fraction of the unprocessed preHoxG intermediate. The CN modes in

mHoxG are detected at 2057 and 2070 cm–1 with a minor shoulder at 2090 cm–1. The latter is assigned to

preHoxG. The peaks observed for preHoxGΔNi are significantly broader compared to those observed for

preHoxG and mHoxG, which suggests a certain degree of structural flexibility in preHoxGΔNi possibly due

to the absence of the Ni ion (Fig. 3.2).

Previous studies showed that the C-terminal tail in MBH is not essential for the in vivo maturation. A fully

active enzyme could be isolated from strains with a deleted C-terminus, however, low protein yields were

reported.4,5

To investigate the impact of the C-terminal extension in HoxG two further intermediates were provided

by the Lenz group, the processed (proc) versions, procHoxGΔNi and procHoxG, in which the C-terminal

extension was removed by genetic engineering. The 2nd derivative of the CO / CN region of the IR spectra

is depicted in Figure 3.3. On the left side mHoxG is plotted in comparison to procHoxG, while on the right

side of the figure preHoxGΔNi and procHoxGΔNi are plotted.

Figure 3.3: A: 2nd derivative of the IR spectra of mHoxG protein and procHoxG, which is lacking the C-terminal

extension and B: 2nd derivative of preHoxGΔNi with procHoxGΔNi analogue which is genetically devoid of the C-terminal

extension.1

The spectra on the left, presenting the mHoxG and procHoxG, are almost identical with a slightly differing

ratio of the vCO absorptions at 1929 and 1939 cm–1. The 1953 cm–1 peak corresponds, as mentioned

above, to residual preHoxG species, while the peak at 1958 cm–1 has been attributed to trace amounts of

mHoxG with an oxidized cofactor. This assignment was corroborated by EPR measurements revealing sub-

stoichiometric NiIII species in mHoxG.1 ProcHoxG exhibits an additional CO band at 1950 cm–1 that is

probably hidden in the spectrum of the mHoxG sample, due to the broad absorption at 1953 cm−1 of the

unprocessed preHoxG. The broadness of this absorption might be attributed to some large subunit species

devoid of Ni.

The IR spectra shown in Fig. 3.3B, depicting preHoxGΔNi and the procHoxGΔNi are also very similar, with the

exception of CO absorption at 1962 cm–1 that is present in preHoxGΔNi and replaced by a peak at 1975 cm–

1 in procHoxGΔNi. These small changes might be related to subtle changes in the active site geometry

and/or coordination potentially induced by the cleavage of the C-terminal extension.

3.1.2 Characterization of the fully-matured HoxG

In this paragraph the focus lies on the spectroscopic characterization of the fully matured HoxG, which is

additionally compared to the large subunit of HoxC of the regulatory hydrogenase from C. necator (CnRH).

This latter has been extensively characterized in the past years.6,7 From now on the fully matured large

subunit mHoxG will be referred to simply as HoxG. It is known that in the as-isolated form HoxC resides in

two Nir-S states, namely Nir-SI and Nir-SII. By means of IR spectroscopy it was shown that those two species

interconvert to each other as a function of the pH and the temperature.6,7 Data for HoxC were provided

by Dr. Giorgio Caserta. HoxG also resides in diamagnetic ready states.1 In this study we will refer to them

as Nir-Sa and Nir-Sb rather than Nir-SI and Nir-SII, since it is not clear so far whether these states are identical

to the observed Nir-SI/II states in HoxC.

In the figure below the effect of the pH on the enrichment of the Nir-S states in both HoxG (Fig. 3.4 panels

A1/B1) and HoxC (Fig. 3.4 panels A2/B2) is shown.

Figure 3.4: A

: vCO and vCN region of the absorbance spectrum depicting the pH titration of HoxG at 10 °C in a 150

mM K

iPO4, 50 mM NaCl solution from pH 5.5 to pH 9.0. At pH 5.5 the predominant species is Nir-Sb with a v

absorption at 1930

cm−1. At pH 9.0 the predominant species is Nir-Sa with a vCO at 1940 cm−1. B1:

Spectra as in A1

examining the reversibility of the inter

-conversion between Nir-Sa and Nir-Sb. The sample is re-

buffered from pH 7.3

to pH 8.0, (blue trace) and measured in the IR regime. At pH 7.3 the sample resides in a 50:50 ratio between Ni

r-S

and N

ir-Sb. At pH 8.0 most of the Nir-Sb is converted into Nir-Sa

. The buffer of the same sample is then exchanged

from pH 8.0 to pH 6.0 (orange trace). At pH 6.0 most of Ni

r-Sa converts into Nir-Sb

. These data show the reversibility

of the inter

-conversion of both species. A2: vCO and v

CN region of the absorbance spectrum depicting the pH

titration of HoxC, 10

°C in 150 mM Tris, 50 mM NaCl. HoxC is stored at pH 8.0. B2: v

CO region of the HoxC active site.

When the buffer of the sample is lowered to 6.0 from pH

8.0 (orange trace), the two Nir-SI,II

species reside in equal

amounts. After returning the pH of the sample to 8.0, the original ratio of the two Ni

r-S species is restored.8

Figure 3.4 panel A1 shows that the two ready species of HoxG, represented by the vCO absorptions at

1930 (Nir-Sb) and 1940 cm–1 (Nir-Sa), reside in a pH-dependent equilibrium. Nir-Sb is enriched at low pHs,

accumulating as a pure state at pH 5.5. On the contrary Nir-Sa is dominant at pH 9.0. At pH 7.3 the two

active site states exist in a 50:50 ratio.

Similarly, the HoxC Nir-SII species (νCO at 1941 cm–1, Figure 3.4 panel A2) dominates at low pHs and Nir-SI

at high pH values (CO stretching at 1951 cm–1). However, Nir-SII could not be homogenously enriched to

more than 50 % due to poor protein stability at pHs lower than 5.0.

The conversion of Nir-SI to Nir-SII was shown to be reversible in HoxC, as exemplified in Figure 3.4 panel B2

where the original ratio of the two Nir-S species is restored upon re-buffering of the HoxC sample to pH

8.0 (blue trace). Similarly, we exchanged the protein buffer in HoxG from pH 7.4 to 8.0 (see Figure 3.4

panel B1). Afterwards, the sample was divided into two aliquots. One was measured directly in

transmission IR, whereas the second aliquot underwent a 2nd buffer exchange from pH 8.0 to pH 6.0. The

two Nir-Sa and Nir-Sb species interconvert reversibly.

Previous computational work on HoxC suggested that the structures depicted in Figure 3.4 panel B2 are

the most probable structures for the Nir-SI and Nir-SII states. Nir-SII is predicted to harbor a H2O molecule

bound terminally to the Ni center, while Nir-SI has been proposed to contain an OH– ligand in the bridging

position between the Ni and Fe and a protonated Ni-bound terminal cysteine. This led to the conclusion

that the two Nir-S species in HoxC are simply differing in the localization of a H+ bound either in the form

of a H2O (Nir-SII) or a protonated cysteine (Nir-SI). Surprisingly, the Nir-SI/II equilibrium was influenced by

the temperature as IR data collected at ca. 90 K revealed almost exclusive population of Nir-SI.8 Herein, it

is attempted to gain structural details on the Nir-Sa and Nir-Sb species. A recent spectroscopic

characterization of HoxG revealed that the Nir-Sa/b species bear oxygenous bridging ligands and feature a

hexacoordinated Fe ion.1 Notably and contrary to the observations on the HoxC subunit,7 the equilibrium

of Nir-Sa/b in HoxG is largely unaffected by temperature as the ratio of the two species at 283 K appears

almost fully preserved at 135 and 200 K (Fig. 3.5).1

Figure 3.5: IR spectra of the as-isolated HoxG recorded at different temperatures: 283 K (black trace), 200 K (grey

trace), and 135 K (red curve). Bands related to vCO and vCN stretching modes in the Nir-Sa and Nir-Sb states of the

active site are labeled in black and red, respectively. With the asterisk is marked a peak derived from unmatured

HoxG species (see Figure 3.2). The buffer is 50 mM KiPO4 buffer with 150 mM NaCl and 30 % glycerol, pH 7.4.

After the addition of glycerol a small shift and broadening of the Nir-S related peaks, was observed,

presumably due to changes in the H-bonding network so that they appear as a single band. Due to the

line sharpening at lower temperatures they can be distinguished at 135 K. It seems that Nir-Sa with a vCO

at 1940 cm–1, is more affected, compared to the Nir-Sb with a vCO at 1930 cm–1, suggesting a different local

environment (Check Fig. A.3.1 in the Appendix). Possible pH changes after the addition of glycerol can be

disregarded as reason for the observed band shift, as the pH of the sample is intentionally adjusted after

its admixture. In addition as shown in Fig. 3.4 the ratio of the Nir-Sa and Nir-Sb peaks is changing with pH

but their band positions remain fixed.

Therefore, one can assume that this shift is related to slight conformational changes of the protein due to

the addition of glycerol into the buffer, which is presumably related to the ability of glycerol to interact

with hydrophobic residues and ‘penetrate’ in areas of the matrix that water cannot.

To gain deeper information on the possible active site structures representing the Nir-Sa/b species,

computational work was carried out by the group of Prof. Maria-Andrea Mroginski (Dr. Jovan Dragelj and

Sarah Böning). Part of the spectroscopic and computational work has already been disseminated in a

recent publication.9 Various models of the active site in HoxG were built, which included amino acids in

the 1st and 2nd coordination sphere as shown in Fig. 3.6.10 After the model is built, it is utilized to predict

the IR features of the active site. The absolute position of the active CO/CN peaks is not considered

reliable, because it is limited to the 2nd coordination sphere, however the difference in wavenumbers

between the two peaks (ΔvCO) can be trusted.

Figure 3.6: Detailed view of the [NiFe]-active site and amino acid residues in the first and second coordination sphere

for the HoxG subunit of CnMBH (PDB entry 3RGW). For simplicity only the sulfur atoms of the four coordinating

cysteines are shown. Distances are marked with black dotted lines and given in Å. Reproduced with permission from

Springer Nature.11

In the following table the structures of the most prominent couples based on the vCO and vCN positions

and, particularly, the direction of the vCO shift (ΔvCO = 10 cm–1) are summarized. Based on the fact that

the two protein structures are in a pH-dependent equilibrium, one can assume that the one that

accumulates at low pH is more protonated than the one that accumulates at high pH. Below we suggest

a couple of models in which the vCO of the more protonated species is at 1943 cm–1, whereas the less

protonated appears at 1951 cm–1 (Table 3.1). For this couple there is a deviation of 10 cm–1 in absolute

numbers between the computed CO stretching mode and the experimental values. However, the shift in

wavenumbers of the Nir-Sb peak versus the Nir-Sa peak is 10 cm–1, both in the experimental and the

computed results. In Table 3.2, models where the states display the same amount of protons, similarly to

HoxC, are shown.

In general we assume that since the ratio of the Nir-S states is highly affected by the pH, a couple in which

the states differ by one proton is more plausible than a couple with the same amount of protons, where

a proton translocation from one position of the active site to another takes place. However, the latter

scenario cannot be completely excluded.

Table 3.1: Calculated frequencies for four different couples. The conjugate structures differ by one proton.

pH Dependent

vCO

(cm

–1

)

vCN

(cm

–1

)

vCN

(cm

–1

)

Couple I

a. Ni-H

O-Fe_Cys597H_H81H_E27H

1943

2075

2105

b. Ni-OH-Fe_Cys597H_H81H_E27H 1951 2067-8 2110

His82 (Nε) shares an H

with C600

Table 3.2: Calculated values for the models. All of the couples refer to proton translocation.

Proton Translocation

vCO

(cm

–1

)

vCN

(cm

–1

)

vCN

(cm

–1

)

Couple II a. Ni-OH-Fe_Cys597H_H81_E27H 1938 2044 2086

b. Ni-H2O-Fe_Cys597_H81H_E27 1947 2084 2107 H moves from His82 (Nε) tο C600

Couple III

a. Ni-H2OTFe_Cys597H_H81_E27 1939 2047 2074 H moves from

Cys597

E32

H2O between Ni and Fe

b. Ni-H2O-Fe_Cys597_H81H_E27 1947 2084 2107

H moves from His82 (Nε) tο C600

Ligand closer to Ni

Couple IV

a. Ni-OH-Fe_Cys597H_H81H_E27H

1951

2067-8

2110

H moves from His82 (Nε) tο C600

b. Ni-H2O-Fe_Cys597H_H81H_E27 1961 2085 2108 H moves from

Cys597

E32

H moves from His82 (Nε) tο C600

Despite some differences, all the modeled structures share structural similarities. In models where initially

a proton was bound to C597 but E27 was deprotonated, the proton was eventually transferred from C597

to E27 by the end of the simulation. C597 remained protonated only when E27 was protonated at the

same time.

His82 (Nε) was either forming a hydrogen bond with the bridging cysteine C600 or the proton was moving

to C600. In some models significant conformational changes of certain amino-acids of the outer 2nd

coordination sphere, such as the leucine L533, were observed. Notably, these conformational changes did

not result in any changes in the active site vibrational modes.

The IR spectrum of HoxG did not show any changes after incubation with H2, therefore, the protein is

considered catalytically inactive. However, we are interested in clarifying whether it is redox active.

Treatment with chemical oxidant, K3Fe(CN)6 and reductant, sodium dithionite (NaDT/Na2S2O4), resulted

in frequency upshift and downshift, respectively, revealing new active site species. The vCO and vCN

stretching regions of the spectrum are depicted in Figure 3.7.1

As shown in Figure 3.7 above, the spectrum after the addition of K3Fe(CN)6 revealed a CO stretching band

at 1958 cm–1 with the respective CN stretching modes located at 2037 and 2084 cm–1. Interestingly, it is

shown that both Nir-Sa and Nir-Sb can be oxidized and form the same Nir-B species. There are two additional

prominent bands, one at 2116 cm–1, which corresponds to the v(CN) stretching of Fe(CN)63– and one at

2036 cm–1, which corresponds to the v(CN) stretching of the one-electron reduced Fe(CN)64–.

Upon treatment with NaDT, the spectrum of the reduced sample did not change significantly compared

to the spectrum of the as-isolated sample. The 1933 cm–1 is the average of the two Nir-S species maybe

Figure 3.7: Baseline-corrected IR spectra of the different redox states of the mHoxG maturation intermediate,

normalized to the amide II intensity. The spectra show the CO / CN stretching region of the active site of as

isolated

(top), ferricyanide

-oxidized (middle) and NaDT-reduced (bottom), HoxG. Black / red trace, Nir-Sa and Nir-Sb

respectively, marked in blue are the bands of Ni

r-B. The yellow-marked band refers to Nia-L. The ferricyanide

–

treated mHoxG spectrum (middle) shows two additional bands at 2037 and 2115 cm

–1

, which correlate with the CN

stretching bands of Fe(CN)6

3–

and Fe(CN)6

4–

, respectively.

due to slight pH variations after the addition of NaDT. Only some traces of Nia-L were detected with vCO

at 1889 cm–1, whereas the vCN bands could not be resolved. The Nia-L species is characterized by a NiI-FeII

configuration of the active site, therefore it is paramagnetic, in contrast to Nir-S (NiI-FeII). EPR

measurements confirmed the presence of such a paramagnetic active site species, upon reduction with

NaDT.1 Notably, Nia-L was only detected after reduction with NaDT but not H2.

3.1.3 Monomer vs Homodimer of HoxG

After a comprehensive structural analysis of the Nir-S species in HoxG our attention has shifted towards

investigating the mechanical properties of the fully matured HoxG, aiming to discern the alterations

occurring on the protein, particularly in terms of rigidity, when it is isolated from the small subunit HoxB.

Additionally, we are exploring the propensity of the protein to form monomers, dimers, or oligomers in

relation to changes in concentration.

The oligomerization of HoxG was investigated utilizing size exclusion chromatography (SEC). More

specifically, the as-isolated protein was measured in various concentration ranges (0.5, 2.5, 12.5, and 60

mg/mL) in a 50 mM KiPO4 and 150 mM NaCl solution at pH 7.4 with SEC.

Figure 3.8: A: HoxG SEC measurements in different protein concentrations were performed in a 50 mM KiPO4 buffer

with 150 mM NaCl at pH 7.4 and 4 0C. B: Monomeric and dimeric forms were calculated based on reference proteins.

Figure adapted from the open-access journal, Frontiers in Microbiology. 9

All the protein solutions were run at an ÄKTA pure 25 system using a Superdex 200 Increase 10/300 GL

(Cytiva) column equilibrated with the purification buffer at 4 °C. A calibration curve was made by

measuring six protein standards with known molecular weights between 12 and 670 kDa: thyroglobulin

(669 kDa, 9.34 mL), apoferritin (443 kDa, 10.49 mL), β-amylase (200 kDa, 11.8 mL), bovine serum albumin

(66 kDa, 14.1 mL), carbonic anhydrase (29 kDa, 16.8 mL) and cytochrome C (12.3 kDa, 18.24 mL). HoxC,

the large subunit of the regulatory hydrogenase of C. necator, was also used as a calibrant. Its size is

appropriate and estimated to be close to HoxG. In this region there is a lack of other calibrants. Besides,

the oligomerization profile of HoxC is known.6,12

In concentrations below 12.5 mg/mL, HoxG mainly exists in the monomeric form (Rv 15.34 mL). A

significant amount of dimeric form (Rv 14.1 mL) was detected at higher concentrations. At 60 mg/mL, the

dimeric form is predominantly detected.

Figure 3.9: A: HoxG and pre-HoxG SEC measurements were performed in a 50 mM KiPO4 buffer, with 150 mM NaCl

at pH 7.4 and 4 0C, aiming to test the C-terminus extension effect on the creation of dimers. B: HoxG SEC

measurements aiming to examine the effect of the ionic strength on the dimerization. Measurements were

performed in 50 mM KiPO4 buffer with 150 mM and 20 mM NaCl at pH 7.4 and 4 °C.

To elucidate the influence of the C-terminus on dimerization, we subjected 60 mg/mL of HoxG (fully

matured without C-terminus) and preHoxG (fully matured with C-terminus) to SEC, as depicted in Figure

3.9 above. As illustrated in Figure 3.9A (blue curve), the presence of the C-terminus had a significant

inhibitory effect on dimer and oligomer formation. Given the proximity of the C-terminus to the active

site, these results provide a stronger indication that dimerization primarily occurs at the former HoxG-

HoxK interface.

Further, we investigated the impact of buffer concentration on dimerization, and the findings are

presented in Figure 3.9B. More specifically, the NaCl concentration was decreased from 150 to 20 mM,

while the KiPO4 concentration remained constant. The results reveal that the ion concentration had no

discernible effect on dimer formation.

Two HoxG samples were analyzed using time-dependent IR spectroscopy, with variations in their

concentrations, as shown in Figure 3.10. The first sample had a concentration of 34 mg/mL, while the

second sample had a concentration of 102 mg/mL. The results indicate, that in the first sample the active

site exists predominantly in its monomeric form. In contrast, in the more concentrated sample, the active

site is predominantly present in its dimeric form. The measurements were conducted continuously over

a time span of 7 hours.

Figure 3.10: Concentration dependent IR spectroscopic characterization of HoxG. A: Baseline-corrected IR spectra

for HoxG sample (34 mg/mL) at t0 and after 7 h. B: Baseline-corrected IR spectra for HoxG sample (102 mg/mL) at t0

and after 7 h. The spectral region in the top left and top right displays bands related to the CO stretching vibrations

A B

of the active site. The shown spectra have been normalized to the intensity of the amide II band (see figure insets).

C: Time evolution of the integral of the [NiFe] active site CO absorption bands for the HoxG sample at 30 (black

squares) and 100 mg/mL (blue squares), respectively. For an adequate comparison of the time evolution of the active

site absorptions in the top left and top right figures, we have chosen 120 min as the starting point (t0) for plotting

the integral intensity of the CO absorptions. Figure adapted from the open-access journal, Frontiers in Microbiology.9

The spectra measured after 2 hrs and 7 hrs are displayed above in the vCO region of the active site in

Figure 3.10. In Fig. 3.10A the 34 mg / mL sample is depicted, whereas in Fig. 3.10B the 102 mg / mL sample

is displayed. In the inset, the amide II region, is shown. The spectra of the lower and higher concentration

were first normalized to the amide II intensity of the most concentrated sample and the spectra measured

after 7 hours were multiplied by a factor of 3. The area below the CO stretching mode was integrated and

plotted as a function of time (Fig.3.10C).

As previously observed, the amide II intensity exhibited no significant changes throughout the

measurement period for both samples. However, the vCO intensity was reduced and this phenomenon

occurred at a higher rate in the case of the lower concentration sample (34 mg/mL) compared to the

higher concentration sample (102 mg/mL). For this reason it can be suggested that dimerization occurs

specifically at the HoxG/HoxK interface of the former. This, in turn, implies that the formation of a

homodimer within the [NiFe] active site's interface could provide a form of protection to the active site

interface by limiting solvent access and preventing the detachment of the [NiFe] cofactor from the protein

scaffold.

A further insight is given by Gaussian accelerated Molecular Dynamic simulations (GaMD). In our setup,

HoxG is equipped with a Strep-Tag II© at the N-terminus, a modification computed by Dr. Dragelj as

described in the referenced paper. In the figure the rigidity profiles of the large subunit HoxG of MBH

(HoxGMBH) are shown, HoxG without HoxK (HoxGC), HoxG monomer (HoxGm) and HoxG dimer (HoxGd). The

first two were extracted from the protein data base (pdb) file (3RGW) without further processing, whereas

HoxGm and HoxGd models were obtained using frames extracted after a 200 ns GaMD simulation. In Figure

3.11, the rigidity profiles of HoxGMBH and HoxGc, are shown as a mirror image of HoxGm and HoxGd. For

further information on the simulations refer to 9.

Figure 3.11: Rigidity profiles of HoxGm and HoxGd (upper purple and yellow line, respectively) and HoxGc, HoxGMBH

(PDB code: 3RGW) (black and red line, respectively). Force constants (in kcal mol–1Å–2) of HoxGm and HoxGd models

were obtained using frames extracted after 200 ns GaMD simulation. Figure reprinted from the open access journal,

Frontiers in Microbiology.

Notably, HoxGMBH exhibits force constants indicating high rigidity for residues coordinating the active site,

such as Cys75, Cys78, Cys597, and Cys600. The absence of HoxK has a limited effect on this trend, with

minor variations in the peak intensities of C597 and C600. In contrast, active site cysteines exhibit reduced

rigidity in HoxGm due to solvent exposure. Interestingly, the rigidity around the [NiFe] active site, lost in

HoxGm, is partially regained in the HoxGd model. These mechanical changes in the case of the HoxGm may

impact substrate channeling (H2) and active site binding efficiency, leading to cofactor leaching and may

partially explain why HoxGm is inactive.

Hence, we can conclude that the formation of homodimers in HoxG serve as a protective mechanism

against profound conformational changes that could potentially result in metal leaching. Additionally, our

spectroscopic and computational analyses of the large subunit have unveiled its capacity to form dimers

at higher sample concentrations. The formation of dimers enhances the long-term stability of the cofactor

as shown by the IR data. Additionally, based on SEC experiments on preHoxG, it was observed that without

cleavage of the C-terminus no dimers are formed. The C-terminus is very close to the active site. The fact

that dimerization enhances the long term stability of the protein and that the dimers do not form when

C-terminus is present (preHoxG), indicate that the HoxG-HoxK interface of MBH serves as the primary site

for dimerization, as shown in Figure 3.12, below.9

Figure 3.12: HoxG

structure after 150 ns of GaMD simulation. The protein backbone is displayed transparent,

while the interfaces of the two subunits are displayed as balls in magenta (HoxG I subunit) and orange (HoxG II

subunit) colors. Residues identified at the HoxGd dimer interface: A: HoxG

I subunit (magenta): Arg65, Leu233,

Gly225, Pro238, Gly247, Ile246, Gly245, Ala248, Ala506, Ser250, Thr505, Asn242, Cys239, Ala240, Arg257, Thr503,

Asn231, Cys597, Leu243, Val237, Lys226, Asn227, Phe70, Arg62, Val508, Ala69, Thr221, Lys60, Arg73, Lys214,

Arg53, Cys75, Glu271, Gly28, Glu27, Ile78, Gln213, His124, Val57, Phe173, Gln178, Arg169, Arg384, His29, Arg386,

Val217, Pro22, Asp21, Asn184, Ser176, Glu175, Gly177. B: HoxG

-I

I subunit (orange): Met183, Asn184, His116,

Asp165, Arg169, Phe173, Ser176, Gln178, Asp211, Val120, His124, Thr602, Gln213, Leu598, Ile26, Asp21, Val217,

Arg172, Glu27, Arg53, Val77, His29, Ala599, Gly76, Met51, Thr221, Arg73, Arg384, Glu175, Phe70, Lys226, Pro238,

Pro372, Thr385, Thr50, Val237, Ala249, Ile58, Ser250, Asn242, Arg62, Cys239, The505, Val508, Arg65, Asn227,

Trp511, Ala248, Ala506.

Figure is reprinted from the open access journal, Frontiers in Microbiology.9

3.2 HoxG immobilized on SAM functionalized electrodes-SEIRA studies

3.2.1 Headgroup and protonation degree effect on the protein’s orientation

Following the studies in solution, the fully matured large subunit HoxG of the CnMBH was immobilized

onto Self-Assembled Monolayer (SAM) coated electrodes, as an immobilization on the bare metal SEIRA

(Surface-Enhanced Infrared Absorption) electrode could result in unfolding and inactivation of proteins.

SAMs with different headgroups were studied to control and assess the orientation of the protein on the

surface.

The most important interactions between the protein and the SAM are electrostatic interactions which in

turn depend on the protonation state of both the protein and the SAM headgroup. Therefore, the first

step was to determine the pKa of the SAM.13,14

Three SAMs with different headgroups were selected. A carboxyl-terminated (S(CH2)7COOH), an amino-

terminated (S(CH2)8NH2) and a hydrophobic (S(CH2)7CH3) SAM. The first two were used at two different

pHs (pH 5.5 and 7.0), resulting in different protonation degrees. In contrast to cationic and anionic SAMs,

protein immobilization to methyl-terminated SAMs via hydrophobic interactions is largely independent of

the pH.

Various pKa values for different SAM's monolayers can be found in the literature depending on the method

used for pKa determination. For example, for an 11-mercaptoundecanoic acid monolayer, pKa values of

10.3 by double-layer-capacitance, 6.1 by quartz crystal microbalance and 5.7 by laser-induced

temperature jump technique were determined.15 For this reason, the decision was made to investigate

the pKas of the protonatable SAMs under similar experimental conditions as used in the SEIRA

experiments. Cyclic voltametry (CV) was utilized for the estimation of the pKa values which were compared

to the experimentally estimated values of already published work, in which the same protocols were

used.15

Figure 3.13: Cyclic voltammograms of 1.0 mM K3Fe(CN)6 on a carboxyl-terminated SAM-modified electrode at pH

3, pH 5, pH 6, pH 7 and pH 8 (from external to internal ) at 100 mV s–1.

A negatively and a positively charged SAM (SH-(CH2)7-COOH, 8-Mercaptooctanoic acid thiol and SH-(CH2)8-

NH2, 8-Amino-1-octanthiol hydrochloride respectively), herein referred as C7COOH and C8NH2, were

immobilized on a SEIRA electrode. The pKa values were determined following procedures reported by the

groups of Zhao and Degafa,16 whereby [Ru(NH3)6]3+/2+ or [Fe(CN)6]33–/4– were used as redox probes for

C7COOH and C8NH2, respectively. In the case that [Fe(CN)6]33–/4– is used as a redox probe for the estimation

of the C7COOH SAM, the electron transfer between the electrode and the electroactive probe is related

with the protonation state of the carboxylic acid group (Figure 3.13). At low pH, the terminal carboxylic

acid is fully protonated and the repulsion is negligible. When the pH is increased the terminal carboxylic

headgroup is getting more deprotonated and consequently, the repulsive forces between the terminal

carboxy-group and the Fe(CN)63– are increasing. The carboxylic acid headgroups can be expected to switch

from partially to fully dissociated when the repulsion between the SAM and the probe is maximum. The

dissociated terminal carboxylic acid is the limiting factor for the electron transfer. The current response

decreases with increasing pH and then reaches a minimum value. Titration curves were derived from peak

currents (ip) of cyclic voltammograms recorded at different pHs, as exemplarily shown in Figure 3.13. The

titration curves are shown in the Appendix Fig. A.3.2 and Fig. A.3.3. The experimentally calculated pKa

values were in good agreement with the pKa values described in literature.15,17 From the pKa values the

protonation degree of the different SAMs could be calculated, as shown in Table 3.3.

Table 3.5: pKa values and resulting SAM protonation on the working pH

pKa

Charged SAM

S(CH

)

COOH

6.7

5.5

6.7

71%

S(CH2)8NH2

5.5

24%

Regarding the HoxG surface, its binding domain to HoxK is hydrophilic, as shown in Figure 3.14 below, and

there is an accumulation of positively charged residues such as Lys, His+ and Arg. Due to this accumulation

of positive charges, one can expect that when immobilizing the protein on top of a negatively charged

SAM, a fraction of molecules will be oriented with the HoxG-HoxK interaction domain toward the

electrode.

Figure 3.14: HoxG subunit derived from the MBH crystal structure 3RGW. Electrostatic surface potential of the HoxG

structures, A: Interaction domain towards HoxK and B: solvent exposed side. The electrostatic potential surface is

displayed, where red and blue indicate the negatively and positively charged regions, respectively. Figure reprinted

from the open access journal, Frontiers in Microbiology.

Next, HoxG was immobilized onto a SEIRA electrode coated with differently charged SAMs. As an example,

the SEIRA spectrum of HoxG immobilized on a C7COOH SAM is presented in Figure 3.15, highlighting in

particular the absorption bands relevant for the analysis of protein orientation and the active site’s

oxidation state.

Figure 3.15: SEIRA spectrum of HoxG after subtraction of buffer contributions. Below 1700 cm–1, the vibrational

modes of the protein backbone are depicted as positive peaks, as well as the CO stretching and the symmetric and

asymmetric CN stretching modes in a region free of other signals (i.e. protein backbone or H2O absorptions) between

2115 and 1870 cm–1. The strong negative absorptions between 3050 and 3400 cm–1 are mainly related to negative

OH stretching vibrations. They are negative because during immobilization water molecules are replaced by protein

molecules. Additionally variations in this region are probably derived from ice on the detector window.

After immobilization of the protein, negative and positive peaks are detected. Two dominant absorption

bands are observed at 1659 (amide I) and 1550 cm–1 (amide II) that are related to protein backbone

vibrational modes. The amide I band mainly derives from the C=O stretching vibration of the peptide bond,

while the amide II band is associated with the respective N-H in-plane bending and C-N stretching

vibrations. Between 1870 and 2115 cm–1, positive bands assigned to the CO and CN stretching modes of

the active site ligands are detected. Accordingly, a single configuration of the active site exhibits in this

region one low frequency (vCO) and two high frequency (vCN) bands in the IR regime.

Additional positive bands are detected between approximately 2800-2970 cm–1, which originate from CH2

and CH3 stretching vibrations of amino acid residues. The negative peaks at 3400 cm–1 are derived from

vibrations of H2O molecules in proximity of the electrode surface, which are replaced during

immobilization by protein molecules. Negative peaks at 3246 cm–1 are presumably derived from variations

of ice on the detector window. The negative band at ca. 1640 cm–1 arises from the H-O-H bending modes

and partially overlaps with the amide I band which is sometimes challenging for the estimation of its

intensity.

The enhancement of vibrational modes of the immobilized protein is orientation-sensitive due to SEIRA

selection rules. Vibrational modes with a dipole moment change parallel to the surface normal are

strongly enhanced, whereas those with mainly a (respective) perpendicular to the surface normal

component gain a much weaker enhancement. As a result, some bands may appear with different

intensities in the SEIRA spectrum as compared to the isotropic IR measurements in solution. Exemplarily,

the active site regime of the spectrum and the amide I and amide II region are shown in Fig. 3.16.

Figure 3.16: SEIRA spectra of immobilized HoxG using differently charged SAMs analyzed at pH 5.5 and pH 7.0. A:

Spectral region of the CO and CN stretching modes of the active site. B: Amide I and amide II absorptions centred at

1659 and 1550 cm

–1, respectively. Their band intensity is use

d to monitor the portion of protein bound to different

SAMs. Experimental conditions:

50 mM phosphate buffer with 150 mM NaCl at pH 5.5 (red) or pH 7.0 (black). The

SEIRA investigations were carried out at 10

°C.

On panel A the active site region is depicted. The relative intensity of the CN stretching compared to the

CO mode is not the same for all the SAMs and also differs from their ratio in solution. Further, the ratio

of the Nir-S species differs in each SAM even at the same pH (see Figure 3.16A). As previously shown, the

as-isolated samples exist in two Nir-S species, whereby a band at 1930 cm–1 is characteristic for the CO

stretching of the Nir-Sb state and the band at 1940 cm–1 is associated with the Nir-Sa state. In solution, both

states exhibit a pH-dependent equilibrium. An additional band is observed at 1953 cm–1 and assigned to

a leftover amount of preHoxG, i.e, a fully matured HoxG version in which the C-terminal extension has not

been cleaved. The pH-dependent equilibrium of the Nir-Sa and Nir-Sb species in solution is conclusively

reproduced using the C8NH2 SAM (middle traces in Figure 3.16B). Opposite both ready silent states appear

equally enriched at both pH values using negatively charged SAMs, although the poor signal-to-noise ratio

at pH 7.0 prevents further data analysis. As described in the previous chapter, we should only have pure

Nir-Sb (1929 cm–1) species at pH 5.5. In the case of C8NH2 SAM at pH 5.5, almost pure Nir-Sb accumulates,

whereas in the case of the C7COOH SAM, a mixture of states is detected.

The intensity of the amide bands has been used as a first indicator of the quantity of the immobilization,

because the amide intensity is related to the coverage of the surface by protein-molecules. In the case of

the C7COOH SAM, at pH 5.5 the amide I and amide II intensity is 13.3 mOD and 12.23 mOD, whereas at

pH 7.0 the amide I and amide II intensity is equal to 4.6 mOD and 4.79 mOD, respectively. Remarkably, for

the SAM terminated with carboxylic acid, at pH 5.5, where the headgroup is expected to be mostly

protonated (i.e. neutral charge), immobilization was better than for all other SAMs. For C8NH2, the amide

I and amide II intensities are 7.17 mOD and 5.89 mOD at pH 5.5 and 5.82 mOD and 4.95 mOD at pH 7.0,

respectively. For C8NH2 SAM only a minimal intensity difference is observed at the two pHs. For the

hydrophobic SAM the amide I and amide II intensity is equal to 7.69 mOD and 7.99 mOD, respectively.

The intensity ratio of the CO and CN bands (ICO / ICN) can provide information regarding orientation of

HoxG on the SEIRA electrode, when compared to the isotropic case in solution and the results are

summarized in the table below.

Table 3.6: Ratio of the ratio of the integrated intensities of the CO and CN bands (ICO / ICN)

ICO / ICN

S(CH2)7COOH

pH 5.5

0.87

pH 7.0

2.5

S(CH2)8NH2

pH 5.5

1.5

pH 7.0

2.3

S(CH2)7CH3

pH 7.0

2.6

Solution

pH 7.0

2.3

The ICO / ICN value in solution is ca. 2.3, which relates to the intrinsic different absorption cross section of

the two modes. Notably, in the case of the C7COOH SAM, this ratio is 0.87 at pH 5.5, which significantly

differs to the value in solution. This finding implies that the transition from a polar but largely uncharged

carboxyl-terminated SAM to an anionic SAM results in a very different orientation of the protein which

corresponds to a net increase of the angle between the CO bond and the surface normal. Here, we can

refer to the angle between the surface normal and an active site bond as ‘theta’ (θ). The angle θ formed

between the CN vector and the surface normal is smaller compared to the angle θ formed between the

CO vector and the surface normal. The resulting enhancement of the vCN signal over the vCO can be

explained by the SEIRA selection rules. Accordingly, the enhancement increases when the dipole moment

of the corresponding mode is more parallel to the surface normal. Most likely, in this configuration the

HoxG binds to the SAM via its HoxK interaction domain in view of the relatively high concentration of

positively charged amino acid residues (vide supra).

In addition, the integrated band intensities of the amide I and amide II absorptions are plotted as a

function of time for all the SAMs, depicting the temporal evolution of their intensities for the first 78 min

of the protein’s immobilization (Figure 3.17).

Figure 3.17: Integrated peak intensities of the amide I and amide II bands plotted as a function of time. A mono-

exponential function could be fitted to the experimental data ( A = A0 + A1 exp –t/τ1) ). The experiments were carried

out at 10 °C in a phosphate buffer, 50 mM KiPO4 and 150 mM NaCl.

The curves shown in Figure 3.17(A - E) describe the evolution of the integrated peak intensity of all SAMs

with time. In each case, a mono-exponential function could be fitted to the data. Interestingly, the fastest

time constants are determined for immobilization on the C7COOH SAM at pH 7 despite the low coverage.

Whereas the previous measurements were carried out at open circuit potential (OCP), an oxidative

potential was applied to the potein versus reductive potential for all of the SAMs at 10 °C, but the attempt

to achieve an electrochemical redox control of the active site was not successful for any of the SAMs at

this temperature. It was then attempted to control the enzyme electrochemically at room temperature,

however, it was only successful in the case of the C7COOH SAM. Here, differences were detected between

the spectra recorded at 10 and 20 °C. Notably, the CO/CN ratio was 1.39 at 20 °C compared to 0.87 at 10

°C, but still considerably smaller than the in-solution value.

The SEIRA spectrum of the immobilized HoxG at room temperature is depicted in Figure 3.18 at OCP and

upon applied oxidative potential (350 mV vs a 3 M Ag/AgCl reference electrode). A portion of the protein

was successfully oxidized and reversibly reduced (Figure 3.18B and 18C). By comparing the intensities of

the CO and CN stretching modes (see Figure 3.18B) to the intensities of the negative-positive peaks as

shown in Figure 3.18C (difference spectrum) it is estimated that electrochemical control could be

established with ca. 25 % of the immobilized protein. This favorable immobilization of the protein on top

of the electrode enabled direct electron transfer by the application of potential without the use of

mediator, leading to the enrichment of the oxidized Nir-B species (NiIII-OH-FeII), with the characteristic CO

stretching frequency at 1958 cm–1 and the corresponding CN stretching at 2083 and 2074 cm–1.

Figure 3.18: A: SEIRA spectrum of the immobilized HoxG at room temperature on top of the C7COOH SAM, in KiPO4

buffer at pH 5.5. B: SEIRA spectrum of the active site at OCP and when an oxidative potential is applied. C: difference

spectra, the application of an oxidative versus a reductive potential (brown curves) displayed and the reversible

reaction, when the application of a reductive versus an oxidative potential is shown (black curves).

3.2.2 Comparison between experimental results and simulations

In order to extract more detailed information from the spectroscopic data on the immobilized HoxG

proteins, computational approaches using molecular dynamics (MD) simulations were performed by Dr.

Jovan Dragelj. In short, a gold electrode was modeled with different SAMs, whereby the protonation state

of the head groups was set according to values calculated from the experimental data. The structure of

the HoxG subunit was taken from the 3RGW pdb file and the affinity Strep-tag© was added to the

sequence. After structure optimization and pKa calculation, yielding adequately protonated amino acid

residues on the surface, the protein was placed near the electrode and rotated around the two axes

parallel to the electrode surface. Potential interactions with the SAM head groups were mapped and initial

orientations of the protein for the simulation were selected. MD simulation of 200 ns was followed to

generate possible final orientations of HoxG for each SAM and pH (5.5 and 7.0). Selected snapshots from

the MD simulations are shown in Figure 3.19.

Figure 3.19: Predicted structures of adsorbed HoxG on different SAMs and working pH simulated for 200 ns. As

starting point the energetic minimum was used.

At the end of the simulation time, after 200 ns, different orientations of HoxG were observed depending

on SAM composition and pH. For the C7COOH SAM at pH 5.5, the active site is located ca. 14 Å away from

the SAM surface, suggesting the possibility of direct electron transfer. For the C8NH2 SAM at pH 5.5, only

transient interactions between HoxG and the SAM surface were observed during MD simulations with a

minimum distance of 50 Å. For the C8NH2 SAM at pH 7.0, the minimum distance between the active site

and the SAM is 25 Å. Simulated immobilization of HoxG on the hydrophobic SAM indicated a minimum

[NiFe] to electrode distance of ca. 43 Å.

Summing up the predictions from the computational results, the minimum [NiFe] to active site distance

was observed for the C8COOH at pH 5.5. This is in line with the experimental results if we consider that

this is the only SAM on top of which we succeeded in electrochemically controlling the redox state of

HoxG. The protein was also immobilized on top of the C8COOH at pH 7.0, C7CH3 (1-Octanthiol) at pH 7.0

and C8NH2 at pH 7.0 but the [NiFe] to the electrode distance was larger than 25 Å, excluding the possibility

of direct electron transfer, which is also in line with the experimental data. A discrepancy between the

experimental and computational results is observed in the case of the C8NH2 SAM, at pH 5.5. According to

the computational results and in contrast to the experimental results the protein should not get

immobilized on top of the latter SAM.

3.2.3 The influence of the ionic strength

Motivated by the available literature18 we attempt to clarify if the detected discrepancies between the

experimental and the computational data are related to the influence of the buffer (ionic strength and ion

valency). Such an example is the immobilization attempts of cytochrome C (cytC). For immobilized cytC,

which carries a positive patch19,20 on the surface next to the redox active heme cofactor, direct electron

transfer was previously observed for both positively and negatively charged SAMs.18,21,22 One given

explanation was that a similar orientation was established due to the ion layer masking the repulsive

forces between the protein and the positive amino-headgroups mimicking the immobilization on top of a

negatively-charged SAM, when KiPO4 is used as buffer and NaCl as supporting electrolyte. This interaction

was proven to be ion-valency dependent and led to the elimination of the repulsive forces between the

positive surface and the positive cytC crevice. Overall, the study highlighted the importance of considering

anion type and concentration when investigating the adsorption of a protein.

Here, the influence of the salt is investigated for the most protonated, C8NH2-SAM at pH 5.5 and the most

deprotonated, C7COOH-SAM at pH 7.0. For the first one, a discrepancy between the experimental and

computational data was observed, whereas for the second SAM, computational data are not at the

moment available.

In Figure 3.20 the amide region of the absorbance spectrum is shown. This reflects the amount of protein

which is immobilized on top of the SAM. Concentration of NaCl, from 150 to 0 mM did not have any

influence on the amide band intensities. Moreover, KiPO4 concentration from 50 to 12 mM did not affect

the amount of the immobilized protein, since the amide I and amide II intensity did not significantly

change. This could be related to a considerable masking of the headgroups by the divalent phosphate

anions of the buffer (KiPO4), which are expected to form a more stable layer than the monovalent Cl–

anion, even at such low concentrations as 12 mM. Presumably, under these conditions, the protein

accumulates on top of the ion layer. Indeed, in experiments in water with minimal ion concentration

(adjusted to pH 5.5 with HCl), no immobilization was observed on C8NH2 SAM.

Figure 3.20: Effect of the buffer salt concentration and the ionic strength on the immobilization capability of the

most protonated (CH2)8NH2 SAM at pH 5.5, utilizing KiPO4 buffer. At this pH, the SAM was protonated by ca. 24%.

Based on the experiments conducted on the NH2-terminated SAM, two conclusions were drawn: Firstly,

the protein did not immobilize when pure H2O was used instead of a buffer, emphasizing the importance

of the phosphate anions. Secondly, the protein was successfully immobilized at 50 mM KiPO4, regardless

of the NaCl concentration within the range of 0 mM to 150 mM NaCl. This is consistent with a previous

study in which size and valence of the anion proved to be important.18 Specifically, the significance of

HPO42– was shown to be more important than the impact of H2PO4– and Cl–. Immobilization of the protein

in a buffer with 300 mM NaCl solution to 3 M NaCl solution, resulted to gradual precipitation of the protein

(check appendix Fig. A.3.4A). When the protein was measured in solution instead of its immobilized state

on the surface, such differences were not observed in the amide II region and the active site absorptions

were well resolved (check appendix Fig. A.3.4B).

Figure 3.21: Effect of the buffer salt concentration and the ionic strength of the KiPO4 buffer for the most

deprotonated SAM in this study, (CH2)7COOH, at pH 7. At this pH, the SAM was deprotonated by ca. 71%.

Then the influence of the buffer for the negatively charged SAM, was examined (Figure 3.21). In the case

of the C7COOH SAM there is a repulsion between the anions of the buffer and the negatively charged

surface. Therefore, the phosphate anions do not form a layer on top of the C7COOH SAM. For this purpose,

it was attempted to vary the NaCl concentration instead. Indeed, the reduction of the monovalent salt

concentration from 150 mM to 0 mM appears to have a significant effect on the amount of the

immobilized protein.

3.2.4 Immobilizing HoxG via a covalent approach

Electrostatic immobilization relies on the non-covalent interaction between charged surfaces and

proteins. However, there are some limitations. The electrostatic binding is relatively weak compared to

the covalent and the protein is prone to desorption. An electrostatic immobilization is reversible, meaning

that the protein has the ability to detach from the surface under certain conditions. Although this can be

advantageous for some applications it could be problematic if long-term stability of the immobilized

protein is required. Another way to achieve a preferential orientation is based on the covalent attachment

of the protein to the electrode. In that case, there is lower susceptibility to environmental factors, such

as changes in pH, ionic strength, or temperature, which could enhance protein desorption such as in the

case of electrostatic binding. All these factors can lead to limited reusability.

A standard procedure for a covalent immobilization of proteins involves the formation of a succinimidyl

ester (COO-Suc)-terminated surface layer and its reaction with primary amines, for example of lysine

residues, in order to form covalent peptide bonds (Fig. 3.22).23–25 In some studies, it has also been

reported that the imidazole ring of the histidine can react with the COO-Suc and the reaction may be

faster than with lysines, forming an N-acyl adduct. This product also hydrolyses very fast, however, it may

provide time for the nearby lysines to attach and may influence the orientation of His-tagged

(hexahistidine affinity tag) proteins.26

The C7COOH SAM was modified by adding EDC and NHS. NHS esters are formed when a carboxylate reacts

with NHS (N-hydroxysuccinimide) in the presence of a carbodiimide (EDC). First attempts to modify a

C7COOH SAM using the standard protocols from literature proved unsuccessful. Working with a SEIRA

electrode is beneficial because the whole immobilization process could be spectroscopically monitored

and the protocols can be improved. In such a way, the possible formation of side products could be

detected and eliminated. This is important as conditions for efficient esterification should be optimized

for each electrochemical set-up, presumably because electrode morphology or the buffer’s ions play a

role in the reaction progress, explaining why the initial attempts proved to be fruitless.

Figure 3.22: Scheme depicting the reaction for the covalent binding of the protein to the esterified SAM. Reprinted

with permission from Royal Society of Chemistry.27

EDC and NHS are mixed together in an aqueous solution (see Figure 3.22).27 The first step of the reaction

is the addition of the OH group of the carboxylic acid across one of the double bonds of the carbodiimide

reactant, forming an O-acylurea adduct (see Figure 3.22).27 However, the formation of the stable amide

bond is not the only pathway of the reaction. The alternative pathways are depicted in Figure 3.23 below.

Figure 3.23: The interaction between surface acid groups and the mixture of EDC and NHS results in the products

described above, encompassing succinimidyl ester, anhydride and urea. The kinetic competition between the various

reaction paths determines the final surface composition. Adapted from 23.

After the formation of O-acylurea, there are three possible pathways for the follow-up reaction. One

possibility is that NHS reacts with this intermediate, through a nucleophilic attack mechanism and it forms

the succinimidyl-ester, as shown in Fig. 3.23(3).23,24 A reaction with a neighboring carboxyl group can yield

an anhydride product, grafted at the surface after urea release, as shown in Fig. 3.23(1). However, this

anhydride product can still be activated if there are sufficient NHS amounts in the solution. Contrarily to

the anhydride, the N-acylurea product, as shown to Fig. 3.23(2), is believed to be non-reactive and

contributes to the IR absorption signals in the amide region.

The activation of the SAM and the stability of the NHS ester may depend on the conditions of the

experiment. The activation reaction with EDC and NHS is most efficient between pH 4.5 and pH 7.2. NHS

esters have a half-life of 4 - 5 hours at pH 7, 1 hour at pH 8 and only 10 minutes at pH 8.6. The reaction of

NHS-activated molecules with primary amines is most efficient at pH values between 7 and 8. At this

experiment, a buffer of pH 7 is used in order to assure NHS ester stability.28 Proteins may be attached via

more than one amine-containing residues.19 Hence, the formation of additional covalent bonds with the

protein molecules bound to the surface could also occur after buffer exchange, provided that the

COO-Suc product is not hydrolyzed. With SEIRA spectroscopy, the entire immobilization process can be

spectroscopically monitored and optimized depending on the particular conditions of the experiment.

In Figure 3.24, the time evolution of the esterification process of the SAM, as recorded with SEIRA, is

depicted. A spectrum of the plain phosphate buffer was used as a reference. tinitial indicates the first

spectrum (recorded within approx. 3 min) after the mixture of EDC and NHS solution is placed on top of

the C7COOH SAM. This spectrum represents a mixture of the educts and products of the reaction. The

final spectrum tfinal, obtained after approx. 21 minutes and before rinsing the surface, is dominated by the

products. A difference spectrum (tfinal – tinitial,) is calculated to show the reaction progress. Negative and

positive bands depict consumed and formed species, respectively. In the first spectrum a 1709 cm–1 band

is recorded, which diminishes with time, whereas all the other absorptions increase in intensity as a

function of time. The positive bands at 1814, 1782 and 1736 cm–1 are assigned together with the negative

contribution at ca. 1700 cm–1 to the activation of the carboxylates headgroups. The negative bands at

1753 and 1709 cm–1 belong to the vacid-CO stretches of transient anhydride species. These absorptions

almost disappear completely after the third spectrum when the esterification proceeds, indicating that

they do not represent a stable product. The positive bands at 1636 and 1572 cm–1 may originate from the

byproduct N-acyl urea.

Figure 3.24: SEIRA spectrum depicting the esterification treatment of a C7COOH SAM, with EDC and NHS in 50 mM

KiPO4 and 150 mM NaCl, at pH 7. The first spectrum is recorded after approximately 3 min (t=tinitial, dark red curve)

and the last spectrum after approximately 21 min (t=tfinal, yellow curve). Below is the difference spectrum (tfinal-tinitial).

Notably, the positions of the absorptions detected here do not fit precisely with the peak positions

reported in the literature.23,25 The three bands at 1814, 1782 and 1736 cm–1 may be considered as a

fingerprint for SAM esterification, but the peak positions are slightly different to those reported by Sam

et al. (1820, 1785, 1740 cm–1) or Ciue Wang et.al. (1815, 1780, 1740 cm–1). One possible reason for this

discrepancy may be the stronger interfacial electric field at the Au electrode which may result in an

appreciable frequency downshift (Stark effect).29,30

Figure 3.25: Last SEIRA spectrum from the esterification process of a C7COOH SAM, with EDC and NHS in 50 mM

KiPO4, 150 mM NaCl at pH 7 is depicted in orange. The blue trace relates to the spectrum recorded after removing

the EDC/NHS solution and rising with phosphate buffer. In the latter spectrum two bands are missing in the amide

region, namely those with at 1636 and 1572 cm–1, proving that they do not originate from N-acylurea.

The peaks at 1636 and 1572 cm–1 disappear after rinsing with buffer, as depicted in Figure 3.25. The

orange curve shows the last spectrum of esterification, whereas the blue spectrum was recorded after

rinsing. These two absorptions can either be related to urea released during the formation of the

succinimidyl ester or the N-acylurea. In contrast to N-acylurea, which is a stable byproduct and a dead end

for the reaction, O-acylurea can react with amino compounds, albeit very slowly.25 Given the fact that the

above peaks disappear upon rinsing, we can assign them to urea23, which remained on top of the surface.

Therefore, one can conclude that under the conditions of the experiment, the C7COOH SAM was

successfully activated by EDC and NHS, yielding a high amount of succ-ester compared to the byproducts.

Next, the activated SAMs were incubated with HoxG solution to form the final peptide bond with the

residues of the protein surface.

Figure 3.26: A: SEIRA difference spectrum of the amide region of the covalently bound HoxG. A spectrum recorded

after the formation of the NHS ester was used as a reference. B: SEIRA spectra taken in the region of the CO and CN

stretching modes. Trace (1) represents the spectrum measured at –350 mV, trace (2) the one at + 350 mV, and trace

(3) refers to the spectrum after setting the potential back to –350 mV. All potentials are recorded vs a Ag/AgCl

reference electrode. The grey curves depict the resulting difference spectra, highlighting the potential induced redox

changes and a certain loss of the cofactor for the back reaction.

In Figure 3.26A the negative NHS-peaks in combination with the positive amide I and amide II absorptions

reflect a successful covalent attachment. Both the amide region of the spectrum and the corresponding

active site region indicate that the protein remains intact after covalent attachment.

Figure 3.26B shows the active site region enlarged at different potentials. The grey curves (2) – (1) and (3)

– (2) depict the related difference spectra. Comparing the curves (1) and (3) and the resulting difference

spectra between the various applied potentials, indicates a significant loss of the active site, besides the

expected variation of the redox states upon multiple switching of the potential. Especially in the grey

difference spectrum (3) – (2), this is obvious, as the intensity of the negative peaks is bigger than the

intensity of the positive peaks. In the case of electrostatic binding, approx. 25% of the protein was

oxidized. In the case of covalent binding, the SAM 50% of the protein can be oxidized. Additionally the

SAM remains quite stable resulting in a high signal-to-noise ratio in the difference spectrum.

To confirm that the reason for the less intense active site signals in the (difference) spectrum is cofactor

loss, it is important to rule out any protein desorption. In Fig. 3.27, the respective stability of electrostatic

versus covalent binding for the adsorbed model protein is presented. Herein, the amide region of the

corresponding difference spectra (end versus start) of the various experiments, are plotted, in which the

protein is either electrostatically (black trace) or covalently (red trace) attached to the electrode surface.

In both experiments the steps after the protein’s immobilization were comparable. After protein

immobilization the surface was rinsed with buffer that was afterwards exchanged. Subsequently,

oxidative and reductive potentials were applied. In the case of electrostatic binding some protein was

desorbed upon potential switching, as can be inferred from the negative amide bands. In the case of

covalent binding, the SAM remained quite stable, resulting in a high signal-to-noise ratio of the difference

spectrum. Thus, the loss in the active site signal intensity, is attributed to a significant cofactor depletion.

Figure 3.27: The figure depicts the difference spectrum recorded at the end of the experiment for electrostatic

binding as well as covalent binding, upon switching the potential, displaying that in the latter case significant cofactor

loss occurs without protein’s desorption.

Figure 3.27 depicts the corresponding difference spectra, calculated form the (spectral) data taken at the

end of the experiments minus that recorded at the start, for HoxG attached covalently (red curve) or

electrostatically (black curve). In the case of the electrostatic binding, negative peaks were observed,

indicating protein desorption. Contrarily, in case of the covalent binding, protein desorption is profoundly

eliminated, as no negative band contributions are detectable in the amide I and amide II region, as

depicted in Figure 3.27. Thus, the fact that a lower amount of protein is reduced and oxidized in the

difference spectrum each time(see difference spectra at Figure 3.26), can be assigned rather to the loss

of the [NiFe] cofactor and not to protein desorption.

In summary, the covalent attachment of HoxG resulted in increased electrochemical control with respect

to an electrostatic immobilization, suggesting a preferred orientation for about 50 % of the redox active

protein with the active site close enough to the Au surface enabling a direct electron transfer. The

observed active site loss supposedly occurs because HoxG losses its rigidity when it lacks HoxK. In addition,

lysines that are located very close to the [NiFe] center could potentially lead to metal leaching, possibly

related to an induction of shearing forces, when a potential is applied. The conformational flexibility of

the protein, which could probably protect the active site or facilitate electron transfer for the fraction of

protein which is not optimally oriented is now hindered. As shown in previous studies the rate constant

of heterogeneous electron transfer (ET) expresses a convolution of orientation-dependent electron

tunneling probabilities, protein reorientation and protein dynamics.19,20 Protein molecules that were not

optimally oriented and exhibit low electron tunneling probabilities in the end electrochemically controlled

after rotational diffusion on the surface or in the intramolecular movements, upon applied potential. In

the next chapter, the effect of covalent attachment compared to the electrostatic binding of the

heterodimeric CnMBH, is examined.13

3.2.5 Immobilization of the CnMBH heterodimer

The new protocol for covalent binding was also applied to the entire heterodimeric CnMBH. For this

enzyme, activity can also be assessed electrochemically, which allows us to monitor possible protein

inactivation as a result of the conducted procedure. The results are then compared to electrostatic

immobilization.

Enzyme samples were immobilized onto SEIRA electrodes at pH 6. Here, either pure C7COOH SAM or an

esterified electrode were used. The gas atmosphere was changed to 100 % Ar to provide anoxic

conditions. Under argon, the CnMBH active site resides in the Nir-B resting state with characteristic CO

and CN stretching bands at 1948 cm–1 and 2081 /2098 cm–1, respectively.31–33 In SEIRA experiments, an

unusually high amount of the Niia-S species (CO stretching at 1929 cm–1)31–33 is often observed and its

(relative) amount tends to increase with time. After 20 min under Ar in the anaerobic buffer, the

atmosphere was changed to H2. To ensure anoxic conditions during reduction with H2, the Ar needle

remained in the headspace of the cell for further purging. In such a way, possible oxygen leakage was

mitigated as a layer of the heavy Ar was formed over the solution. A second H2 needle was immersed into

the solution allowing fast diffusion of the dissolved substrate gas. Upon reduction with H2 (Figure 3.28,

middle column), the protein was converted to the Nia-C (1957 cm–1) 31–33 state and the fully reduced

species, Nia-SR (1945 cm–1), Nia-SR’ (1925 cm–1) and Nia-SR’’ (1919 cm –1).31,32,34 In less than 1 min after the

gas exchange, the OCP decreased rapidly (from 30 ± 50 mV to –350 ± 50 mV). In the case of the covalently

bound His-tagged protein the OCP droped even 50 mV lower, down to –400 ± 50 mV.

Figure 3.28: A: SEIRA spectra of a His-tagged MBH immobilized covalently, and B: electrostatically. Left, the amide

and active site region of the spectrum is depicted. The middle panel displays the same spectra in an enlarged view

in the CO and CN stretching region. Right, difference spectra of the oxidized versus reduced protein and vice versa.

The first trace of each figure in the right column depicts the difference spectrum 150 minus –450 mV vs Ag/AgCl and

the second trace the related data when subsequently –450 mV minus 150 mV vs Ag/AgCl are applied. The intensity

of the peaks in the difference spectrum refers to the amount of protein which is in good contact with the electrode.

Comparing Fig. 3.28A to Fig. 3.28B, no significant differences in the coverage of the electrode between

the electrostatically and covalent bound His-tagged MBH can be assessed from the amide I and amide II

band intensities. This means that a similar packing density of the protein on the SAM is achieved. The

intensity ratio amide II /active site bands is similar for all cases. The ratio is 11.2 for the electrostatically

immobilized His-tagged CnMBH and 9.9 for the covalently immobilized His-tagged CnMBH.

The middle and right columns in Fig. 3.28 show the active site region of the SEIRA spectra and the resulting

difference spectra upon applied potentials, respectively. In general, most of the molecules are not

activated directly by hydrogen. When the distal cluster is facing towards the electrode, direct electron

transfer from the cluster to the electrode is possible and the rest of the molecules are activated by the

electrode. The electrode functions as a wire. Therefore, by comparing the amount of activated with non-

activated enzyme one can estimate the amount of protein which can be electrochemically controlled.

Similar information can be obtained if one compares the active site band intensities in the absorbance

spectra with the observed changes in the respective potential-dependent difference spectra. If one

compares the active site intensity, it is the same for both proteins (Fig. 3.28B). However, from the

difference spectra intensity, it seems that the intensity is slightly higher for the covalently (0.04 mOD)

over the electrostatically attached protein (0.03 mOD). In that respect, it seems that more molecules are

probably in direct electrode contact in the case of the covalently attached protein compared to the

electrostatically attached. Nevertheless, one has to consider that a reliable quantification is probably not

possible in Fig. 3.28B due to baseline induced errors and the relatively low signal to noise ratio. In addition,

the CO stretching of the Nir-B state is located closely in wavenumbers to the two electron reduced Nia-SR

species and is difficult to estimate how much residual Nir-B is present. The comparison is easier focusing

on the vCN stretching region at 2049, 2075 cm–1 for Nia-SR’ versus 2081, 2098 cm–1 for Nir-B. Furthermore,

the CO stretching of the two electron reduced Nia-SR’ (1927 cm–1)32 species, after H2 treatment is close to

that of the Niia-S (‘ia’, inactive) at 1930 cm–1, an inactive species, which often accumulates as a function of

time.

Cyclic voltammetry (CV) on thin protein films is a useful tool to estimate the amount of enzyme which is

in a suitable orientation and distance for direct electron transfer (DET) to [NiFe] or the distal [FeS] cluster.

Accordingly, we compare the catalytic current recorded under bare H2 (DET) with the one measured in

the presence of methylene blue (MB), corresponding to mediated electron transfer (MET).

The CV of hydrogenases displays three distinct regions (see Figure 3.29A). At lower potentials, typically

below the thermodynamic potential of H2/H+, the negative current region corresponds to H+ reduction.

The current curve intersects the zero-current line at the thermodynamic potential of the H2/H+ couple,

where the enzyme facilitates both H2 oxidation and H2 production. Above this point, two separate positive

current regions can be identified, referred to as the activation and deactivation regions of the

hydrogenase.

Figure 3.29: A: Cyclic voltammograms (CV) of the [NiFe] hydrogenase from Cupriavidus Necator covalently attached

to a C7COOH SAM after reacting with EDC/NHS, under Ar (black trace) and under H2 (red trace). The color coding

indicates hydrogen evolution reaction (HER) (orange area), H2 oxidation (green area) and anaerobic inactivation

(yellow area) of hydrogenase and its switch potential for inactivation (position of dark green vertical line at –165

mV. B: Eswitch is defined as the potential of maximum slope in the reductive activation direction, determined from

the 1st derivative plot.35–37

Hydrogenase deactivation is regulated through rigorous redox control via the electrode, leading to

anaerobic oxidation at the active site and the formation of the reversibly inactivated Nir − B state, which

results in a decline of the catalytic current. This state can also be induced by the addition of O2.38 During

the return scan, the electrocatalytic current recovers as the enzyme reactivates at lower potentials. In the

presence of hydrogen, the potential at which activity is recovered is highly sensitive to experimental

conditions, and is characterized by a specific parameter known as the "switch potential" (Eswitch). This value

is defined as the potential at which the reductive activation exhibits the maximum slope, as determined

from the 1st derivative (Fig.3.29B). Eswitch varies depending on pH and temperature, higher switch

potentials are expected to occur with decreasing pH and/or increasing temperature and its determination

makes sense only when full deactivation is observed (see Figure 3.29B). The corresponding CVs of the

electrostatically attached His-tagged CnMBH are shown in Figure 3.30.

Figure 3.30: Thin film protein voltammetry of A: His-tagged MBH, covalently attached, under argon (black trace),

under hydrogen (red trace) and under hydrogen and 15 μM MeB (green trace), B: His-tagged MBH, electrostatically

bound, under argon (black trace) and under hydrogen (red trace) and after addition of 5 mM MeB (blue trace) and

15 mM MeB (green trace) .

This protein is known to be strongly biased toward hydrogen oxidation over hydrogen production.

However under our experimental conditions negative currents are detected that correspond to H+

reduction in the case of the His-tagged CnMBH. The CV is Fig. 3.30 A crosses the zero-current axis close to

the equilibrium potential, which is approximately equal to –522 mV vs Ag/AgCl at 25 0C, pH 5.5, under

100% H2 provided that the inlet pressure is 1 atm. The value is calculated by the Nernst equation and may

slightly differ to the (experimental) value in our set-up.

The electrostatically attached His-tagged enzyme exhibits predominately currents indicative for H+

reduction rather than H2 oxidation (Fig. 3.30B). The main reason could be the orientation of the protein

which is electrostatically attached on top of the C7COOH SAM. The addition of MB enhanced the H2

oxidation. Therefore, the results seem to indicate that the change in the bias of the direction of H2

production over H2 oxidation depends on the difference of the relative orientation of the two samples on

top of the electrode. Possibly, the His-Tag contributes to the final enzyme orientation because such signals

were not observed in the Strep-tagged MBH (see Appendix Fig. A.3.6). As previously mentioned the

imidazole ring of the histidine reacts fast with the NHS-ester, but it is also hydrolyzed fast. Presumably the

His-Tag affects the final orientation of the protein if lysines in its proximity react with the NHS-ester,

forming amide bonds.

Notably this is not the first time that His-tagged MBH showed an unusually high proton reduction. In a

recently published paper,39 the protein was immobilized on top of tin-rich indium tin oxide (ITOTR) and the

same was observed only for the His-tagged and not the Strep-tagged protein. In our system H+ reduction

currents are also observed on some extent for the Strep-tagged protein, but less prominently compared

to His-tagged MBH (see Appendix Fig. A.3.6). Another common observation between these two studies

is that the unusually high proton reduction currents remain or are even enhanced, when the buffer is

saturated with H2, although this protein is known to be substrate inhibited. The explanation for this

observation remains elusive.

3.3 Conclusions

HoxG, in the as-isolated form, is found to be present in a couple of Nir-S species characterized by a NiII-X-

FeII active site structure. These two Nir-S species are in a pH-dependent equilibrium that remains largely

unaffected by changes in the temperature. It is postulated that they differ by the addition or removal of

a single proton. However, the possibility cannot be dismissed, that both states may possess an equal

proton count and the two conjugate species differ by the position of a proton within the active site. In any

case, computational analysis prominently underscores the significant role of His82, which is not present

in the large subunits of other hydrogenases, such as the CnRH of HoxC. This histidine can exist in a single

or double protonated state, occasionally forming a hydrogen bond with C600. In certain computational

models, the translocation of an H+ from His82 to C600 is observed. Another notable observation pertains

to the terminal glutamate E32, which consistently exhibits a propensity for protonation.

Despite being catalytically inactive, HoxG remains redox active. Treatment with K3Fe(CN)6 leads to the

accumulation of a Nir-B state, in which the active site adopts the NiIII-OH-FeII configuration. The protein’s

catalytic inactivity is likely attributable to conformational alterations when the larger HoxG subunit is

isolated without the presence of the smaller HoxK subunit. Computational modeling suggests that the

active site undergoes a loss of structural rigidity, a phenomenon that is partially reversed when one HoxG

molecule forms a dimeric complex with another HoxG molecule.

Experimental data reveal that HoxG homo-dimerization is dependent on the sample’s concentration. In

samples characterized by low concentration, in which the HoxG monomer predominates, a notable

decline in cofactor stability is observed as a function of time. This observation lends credence to the idea

that primary dimerization predominantly occurs at the former HoxG-HoxK interface. This proposition is

further substantiated by the fact that in the case of preHoxG (representing fully matured HoxG, with an

intact C-terminus that has not undergone cleavage), dimers fail to form due to the close proximity of the

C-terminus to the active site. Furthermore, computational analysis reveals that the structural rigidity in

the area around the [NiFe] cofactor is regained when homo-dimers (HoxGd) are formed, whereas rigidity

is lost in the HoxG monomer compared to the heterodimeric CnMBH (HoxG-HoxK),

Two different approaches were employed for the immobilization of HoxG i.e., the electrostatic and the

covalent. In the context of electrostatic immobilization, a particular surface, namely C7COOH SAM at pH

5.5, appears to induce a narrower range of molecular orientations and permits limited control over the

electrochemical behavior of a fraction of the protein. MD simulations revealed that for this SAM, the

distance between the [NiFe] center and the headgroup is minimized and equal to 14 Å, providing a

potential for direct electron transfer. With respect to the accumulation of positively charged amino acids

at the former HoxG/HoxK interface, it is anticipated that the protein immobilizes in a manner that exposes

this region to the C7COOH SAM-functionalized electrode.

Furthermore, the protein was covalently attached on top of the electrode. Through the applied protocol,

lysines are specifically targeted. Thus, one expects that a portion of the immobilized protein molecules

adopts a similar orientation as the electrostatically immobilized protein on top of the C7COOH surface. A

larger fraction of the protein is chemically oxidized and reduced, however, for this protein fraction a

gradual [NiFe] cofactor loss is observed besides a limited electrochemical control of the present redox

states. Notably, cofactor loss was even observed in solution for the monomeric HoxG, potentially related

to interfacial tensions. When considering the population of protein molecules, which is not subjected to

electrochemical control, it is important to note that electron transfer depends on the interplay of two

factors: protein dynamics and electron tunneling.19,20 For instance, when cytochrome C (cytC) molecule

was covalently atthached low-amplitude motions of the protein molecules result in low electron tunneling

probability.19 To enhance the electron tunneling probability, a rotational motion of the protein on top of

the SAM was required. However, intramolecular or macroscopic, motions are significant for electron

transfer which may be impeded in the covalent immobilization case if several covalent attachment sites

are employed.

Subsequently, the novel covalent attachment protocol was applied to the full-length membrane-bound

hydrogenase (MBH) to evaluate its catalytic activity subsequent to covalent binding. A His-tagged MBH

was successfully covalently immobilized. CV of the His-tagged protein revealed a discernible shift in the

catalytic bias in favor of hydrogen (H2) production. This shift was not that prominent in the case of the

Strep-tagged immobilized protein. Given that the sole distinction between the samples lies in the affinity

100

tag, it can be inferred that the latter exerts an influence on the final orientation of the protein when the

covalent attachment protocol is applied.

101

3.4 References

1. Caserta, G. et al. Stepwise assembly of the active site of [NiFe]-hydrogenase. Nat. Chem. Biol. 19,

498–506 (2023).

2. Casalot, L. & Rousset, M. Maturation of the [NiFe] hydrogenases. Trends Microbiol. 9, 228–237

(2001).

3. Lacasse, M. J. & Zamble, D. B. [NiFe]-Hydrogenase Maturation. Biochemistry 55, 1689–1701

(2016).

4. Hartmann, M. S. S. Lessons from active site biosynthesis of [NiFe]-hydrogenase-Towards the in

vitro reconstitution of the bimetallic catalytic center. (2021).

5. Hartmann, S., Frielingsdorf, S., Caserta, G. & Lenz, O. A membrane-bound [NiFe]-hydrogenase

large subunit precursor whose C-terminal extension is not essential for cofactor incorporation

but guarantees optimal maturation. Microbiologyopen 9, 1197–1206 (2020).

6. Caserta, G. et al. The large subunit of the regulatory [NiFe]-hydrogenase fromRalstonia eutropha-

a minimal hydrogenase? Chem. Sci. 11, 5453–5465 (2020).

7. Caserta, G. et al. Hydroxy-bridged resting states of a [NiFe]-hydrogenase unraveled by cryogenic

vibrational spectroscopy and DFT computations. Chem. Sci. 12, 2189–2197 (2021).

8. Caserta, G. et al. Supplementary Information Hydroxy-bridged Active Site States of [NiFe]-

Hydrogenase Unraveled by Cryogenic Vibrational Spectroscopy and DFT Computations. Chem.

Sci. 12, (2021).

9. Dragelj, J. et al. Conformational and mechanical stability of the isolated large subunit of

membrane-bound [NiFe]-hydrogenase from Cupriavidus necator. Front. Microbiol. 13, 1–15

(2023).

10. Fritsch, J. et al. The crystal structure of an oxygen-tolerant hydrogenase uncovers a novel iron-

sulphur centre. Nature 479, 249–253 (2011).

11. Frielingsdorf, S. et al. Reversible [4Fe-3S] cluster morphing in an O 2 -tolerant [NiFe]

hydrogenase. Nat. Chem. Biol. 10, 378–385 (2014).

12. Caserta, G. et al. Supplementary Information: The large subunit of the regulatory [NiFe]-

hydrogenase from. Chem. Sci. 11, (2020).

13. Heidary, N. et al. Orientation-controlled electrocatalytic efficiency of an adsorbed oxygen-

tolerant hydrogenase. PLoS One 10, (2015).

14. Utesch, T. et al. Effect of the protonation degree of a self-assembled monolayer on the

immobilization dynamics of a [NiFe] hydrogenase. Langmuir 29, 673–682 (2013).

15. Dai, Z. & Ju, H. Effect of chain length on the surface properties of ω-carboxy alkanethiol self-

assembled monolayers. Phys. Chem. Chem. Phys. 3, 3769–3773 (2001).

16. Degefa, T. H., Schön, P., Bongard, D. & Walder, L. Elucidation of the electron transfer mechanism

of marker ions at SAMs with charged head groups. J. Electroanal. Chem. 574, 49–62 (2004).

102

17. Marmisolle, W. A., Capdevila, D. A., Llave, E. De, Williams, F. J. & Murgida, D. H. Self-Assembled

Monolayers of NH 2 ‑ Terminated Thiolates: Order, pKa and Specific Adsorption. Langmuir 29,

5351–5359 (2013).

18. Peng, C., Liu, J., Xie, Y. & Zhou, J. Molecular Simulations of Cytochrome c Adsorption on Positively

Charged Surfaces: The Influence of Anion Type and Concentration. Phys. Chem. Chem. Phys. 18,

9979–9989 (2016).

19. Ly, H. K. et al. Thermal fluctuations determine the electron-transfer rates of cytochrome c in

electrostatic and covalent complexes. ChemPhysChem 11, 1225–1235 (2010).

20. Khoa Ly, H. et al. Electric-field effects on the interfacial electron transfer and protein dynamics of

cytochrome c. J. Electroanal. Chem. 660, 367–376 (2011).

21. Kranich, A. et al. Gated electron transfer of cytochrome c6 at biomimetic interfaces: a time-

resolved SERR study. Phys. Chem. Chem. Phys. 11, 7390–7397 (2009).

22. Zuo, P., Albrecht, T., Barker, P. D., Murgida, D. H. & Hildebrandt, P. Interfacial redox processes of

cytochrome b562. Phys. Chem. Chem. Phys. 11, 7430–7436 (2009).

23. Sam, S. et al. Semiquantitative study of the EDC/NHS activation of acid terminal groups at

modified porous silicon surfaces. Langmuir 26, 809–814 (2010).

24. Sam, S. et al. Covalent immobilization of amino acids on the porous silicon surface. Surf. Interface

Anal. 42, 515–518 (2010).

25. Wang, C., Yan, Q., Liu, H. B., Zhou, X. H. & Xiao, S. J. Different EDC/NHS activation mechanisms

between PAA and PMAA brushes and the following amidation reactions. Langmuir 27, 12058–

12068 (2011).

26. Cuatrecasas, P. & Parikh, I. Adsorbents for Affinity Chromatography. Use of N-

Hydroxysuccinimide Esters of Agarose. Biochemistry 11, 2291–2299 (1972).

27. Bart, J. et al. Room-temperature intermediate layer bonding for microfluidic devices. Lab Chip 9,

3481–3488 (2009).

28. Hermanson, G. T. The Reactions of Bioconjugation. in Bioconjugate Techniques 229–258 (2013).

29. Schkolnik, G. et al. Vibrational stark effect of the electric-field reporter 4-mercaptobenzonitrile as

a tool for investigating electrostatics at electrode/SAM/solution interfaces. Int. J. Mol. Sci. 13,

7466–7482 (2012).

30. Utesch, T. et al. Potential Distribution across Model Membranes. J. Phys. Chem. B 126, 7664–

7675 (2022).

31. Saggu, M. et al. Spectroscopic insights into the oxygen-tolerant membrane-associated [NiFe]

hydrogenase of Ralstonia eutropha H16. J. Biol. Chem. 284, 16264–16276 (2009).

32. Siebert, E. Vibrational Spectroscopy of the Active Site and Iron Sulfur Clusters of the Membrane

Bound Hydrogenase from Ralstonia Eutropha. (2015).

33. Katz, S. Spectroscopic studies on hydrogenases. PhD thesis, Univ. der Tech. Univ. Berlin (2019).

34. Waffo, A. F. T. et al. Structural Determinants of the Catalytic Ni a ‑ L Intermediate of [ NiFe ] -

Hydrogenase. J. Am. Chem. Soc. 145, 13674–13685 (2023).

103

35. Armstrong, F. A. & Hirst, J. Reversibility and efficiency in electrocatalytic energy conversion and

lessons from enzymes. 108, (2011).

36. Armstrong, F. A. et al. Reactions of complex metalloproteins studied by protein-film

voltammetry. 26, 169–179 (1997).

37. Vincent, K. A., Parkin, A. & Armstrong, F. A. Investigating and Exploiting the Electrocatalytic

Properties of Hydrogenases. Chem.Rev. 107, 4366–4413 (2007).

38. Vincent, K. A. et al. Electrocatalytic hydrogen oxidation by an enzyme at high carbon monoxide or

oxygen levels. Proc. Natl. Acad. Sci. U. S. A. 102, 16951–16954 (2005).

39. Davis, V. & Heidary, N. et. al. Immobilization of O2-tolerant [NiFe] hydrogenase from Cupriavidus

necator on Tin-rich Indium Oxide Alters the Catalytic Bias from H2 Oxidation to Proton Reduction.

ACS Catal. 13, 6312–6327 (2023).

104

105

4. Spectroscopic studies on temperature-dependent and light-sensitive

equilibria of [NiFe] hydrogenases

The following chapter focuses on mechanistic aspects within the catalytic cycle of [NiFe]-hydrogenases,

in which the influence of external factors such as temperature on the enrichment of certain intermediates

were studied. As described already in the introduction, the active site of a hydrogenase contains unusual

CO/CN- ligands, which reassemble valuable IR probes for monitoring redox, structural and protonation

changes at or in proximity of the catalytic site. Most of the IR spectroscopic investigations presented in

this chapter have been conducted at cryogenic temperatures, providing novel details on the electronic

and/or structural properties of the [NiFe]-hydrogenase active site in catalytic, or resting states. This

approach follows successfully recent achievements in [NiFe]/[FeFe]-hydrogenases, where cryogenic IR

deciphered new hydride-containing intermediates1 as well as the respective protonation state of some

residues of the proton transfer pathway.2,3 Besides temperature-induced perturbations, we also utilize

light as a trigger to induce either changes in the (redox) equilibrium of certain intermediates or the

enrichment of transient species.

As stated in section 1.3 the exact catalytic mechanism of [NiFe]-hydrogenases is still under debate, despite

decades of research. While there is consensus on the structures of Nia-C and Nia-S, those of Nia-SR and Nia-

L are still highly debated as both comprise various sub-forms, whose structural differences have not been

clearly addressed. Additionally, besides typical Nia-C and Nia-L interconversion, previous Resonance

Raman studies suggested the possible conversion of Nia-SR and Nia-L upon light exposure at cryogenic

conditions and high local power densities.4 The corresponding mechanism needs to be elucidated and has

driven some of the following research.

106

Figure 4.1: A: X-ray structure of the NAD+-reducing soluble [NiFe]-hydrogenase from Hydrogenophilus

thermoluteolus (HtSH) comprising four protein subunits HoxH, HoxU, HoxY and HoxF (PDB: 5XF9). The hydrogenase’s

active site is located in the large subunit HoxH. B: X-ray structure of the membrane-bound hydrogenase (MBH) from

Cupriavidus necator (Cn), comprising the large subunit HoxG and the [FeS]-containing small subunit HoxK (PDB:

3RGW).

Two oxygen tolerant [NiFe]-hydrogenases have been used as model enzymes in these investigations. The

first one is the thermophilic NAD+-reducing soluble [NiFe] hydrogenase from Hydrogenophilus

thermoluteolus (HtSH) TH-15-8 and the second is the membrane-bound hydrogenase (MBH) from

Cupriavidus necator, formerly Ralstonia eutropha (CnMBH), which have been introduced in sections 1.4

and 1.5.4,9-13 Both SH and MBH enzymes have been suggested as possible candidates in H2-driven

biotechnological applications (e.g., NAD(P)H regeneration14,15 and H2 oxidation16-18), as they preserve

catalytic activity in the presence of oxygen that is usually inhibiting and/or damaging irreversibly O2-

sensitive members of the hydrogenase family.

At the end of the chapter the reader can find comprehensive tables with the CO, CN stretching frequencies

of the observed redox structural states of the NiFe active site for the enzymes which are studied in the

chapter (HtSH, wild type CnMBH and its C81S variant).

107

4.1 Temperature-dependent equilibria between redox-structural states of [NiFe]

hydrogenases

4.1.1 Hydrogenophilus Thermoluteolus Soluble Hydrogenase

Figure 4.2: Redox-structural states of the [NiFe] active site in HtSH (1) comprising in the as-isolated form the Ni(IV)r-

Hex, Ni(III)r-Hex, and Ni(III)’r-Hex species. “r” is the abbreviation for ‘’ready-silent” resting states, which are not

catalytically active. “Hex” stands for ‘’Hexa-coordinated’’, referring to the coordination number of the Ni ion. ΔΤ

indicates a temperature increase in the direction of the arrow. Each state is characterized by the respective band

position of the vCO stretch in wavenumbers. (2) Redox state of the as-isolated CnMBH wildtype, which comprises

predominantly a Nir-B resting state, harboring a bridging hydroxo ligand between the Ni and Fe ions.

The HtSH enzyme in the as-isolated form resides in different redox species (Figure 4.2) that can be

visualized using IR spectroscopy monitoring the various absorptions of the CO stretching vibrations. In a

recent study we have spectroscopically elucidated the nature of these redox structural species and

identified one, presumed hexa-coordinated Ni(IV) and two Ni(III) species that are referred herein as

Ni(IV)r-Hex and Ni(III)r-Hex (Fig. 4.2, r stands for “ready species” as these states are not catalytic

intermediates which are usually symbolized by the subscript “a” for activated).19 The exact arrangement

of the cofactor in the two Ni(III)r states is still unknown. Crystallographic, spectroscopic and biochemical

108

data suggested that Ni(IV)r-Hex holds an unprecedented Ni ion coordinated by a bidentate glutamate

(E32), three bridging and one terminal cysteine (Fig. 4.2.1).19 The presence of a third bridging cysteine

(Cys462) is also quite unusual since homologous Cys residues in all structurally resolved [NiFe]-

hydrogenases are usually found as terminal ligands for the Ni ion. The spectroscopic features of Ni(IV)r-

Hex state were found to be enriched in both protein solutions19 and crystals (unpublished, Fig. 4.3) upon

fast re-oxidation by air of the H2-reduced sample. This structural arrangement of the active site has been

proposed to provide protection from reactive oxygen species (ROS), as no vacant coordination site is

available at both, the Fe and Ni ion, respectively.20 Upon slow re-oxidation of the previously H2-reduced

HtSH-sample during exposure to air in the thin layer cell (up to 5 days), the Ni(III)’r-Hex and Ni(III)r-Hex

are enriched. The resulting difference spectrum, in which the active-site and Glu region are displayed, is

shown in Fig. A.4.1, Appendix).

Figure 4.3: 2nd derivative of the IR absorbance spectrum of HtSH crystals recorded at 80 K in their as-isolated and re-

oxidized forms after exposure to air. In the as-isolated form HtSH resides in a mixture of states, namely in the two

Ni (III)r-Hex states and some traces of the 1995 cm–1 Ni(IV)r-Hex form. After H2 reduction and subsequent fast

exposure of the sample to air the Ni(IV)r-Hex species is enriched.

109

For the two Ni(III)r-Hex a rather similar active site structure has been hypothesized as the two species

exhibit almost identical positions for the CO stretching bands ( 1963 vs 1971 cm–1, Fig. 4.3).

The 2nd derivative of the low-temperature IR absorbance spectra, recorded at 85 K, of heterologously and

homologously produced HtSH protein solutions are shown in Fig. 4.4. The species represented by the CO

band at 1971 cm–1 appears less prominent in the heterologous form of HtSH isolated from Cupriavidus

necator, which additionally contains contributions from a Nir-S species highlighted by a νCO at 1936 cm–1.

This indicates certain differences in the active site redox composition for the homologous and

heterologous protein preparations.

Figure 4.4: 2nd derivative of the IR absorbance spectra of (a) heterologously and (b) homologously produced HtSH

proteins in their as-isolated forms at 85 K. In the heterologously-expressed sample two main species accumulate,

the Nir(III)-Hex at 1963 cm–1 and the Nir-S at 1936 cm–1, together with some amount of the Nir(III)-Hex’ at 1971 cm–

1. In the homologously expressed sample, on the other hand, the predominant redox structural species are the two

Nir(III)-Hex species and a small amount of the Nir(IV)-Hex species. The ratio between the different observed states

varies slightly between different protein preparations.

110

Herein, it is attempted to shed light on the nature of the two Ni(III)r-Hex species by investigating their

behavior in solution at different temperatures as certain hydrogenase redox-structural states have been

recently shown to be sensitive to temperature perturbations.21,22 Homologously produced HtSH in

solution (25 mM Tris-HCl pH 7.4, 150 mM NaCl) was investigated by IR spectroscopy at 293, 200 and 85 K

(Fig. 4.5). While IR spectra at 293 K (trace a) exhibit predominantly the 1963 cm–1 species, a band at 1971

cm–1 appears to be significantly populated at low temperatures (traces b and c), thereby suggesting a

thermal equilibrium between the two species. Upon raising the temperature of the solution back to 293

K (trace d), the two species returned to the original equilibrium (trace a) confirming full reversibility of the

process.23,24

Figure 4.5: 2nd derivative of the IR absorbance spectra of HtSH protein solutions in the as-isolated form at 293 (a),

200 (b), 85 K (c) and brought back to 293 K (d). These spectra highlight the different enrichment of 1971 cm–1 and

1963 cm–1 species at different temperatures. The thermal equilibrium appears to be fully reversible as the original

ratio of the two species could be restored by bringing back the sample’s temperature to 293 K (d). The spectrum in

c was scaled x 1/3 for a better illustration. The CO and CN absorptions related to the νCO and νCN stretching

vibrations of the diatomic ligands of the active site are labeled with their corresponding wavenumbers.

111

The IR spectrum of the reduced HtSH protein in solution displays various species (Fig. 4.5) that have been

assigned to Nia-C (vCO at 1973 cm–1) and certain Nia-SR sub-forms (Nia-SR and Nia-SR’’ with vCO at 1958

and 1935 cm–1, respectively). Both Nia-C and Nia-SR intermediates contain a bridging hydride, differing

mainly in the electronic configuration of the Ni ion (NiIII vs NiII). The various Nia-SR sub-forms are tentatively

assigned to active site states with identical electronic configurations, NiII-H-FeII, and possibly differ in the

protonation state of nearby residues.9,25-27 Among them, sub-forms containing a terminal protonated

cysteine thiolate have been identified. In this respect, convincing evidence is provided by an ultra-high-

resolution structure of the O2-sensitive DvMF (Desulfovibrio vulgaris Miyazaki F) [NiFe]-hydrogenase

enzyme (ca. 85 % Nia-SR) and a combined DFT/NRVS study of the same species from the

Lubitz/Cramer/Rauchfuss laboratories.26 The relative population of the Nia-C and Nia-SR species in HtSH

has also been shown to vary depending on the chosen reductive treatment type, as shown in Fig. 4.6 (e.g.,

pure H2 gas, H2-plus-NaDT).

Figure 4.6: 2nd derivative of IR absorbance spectra of reduced HtSH in the CO stretching region highlighting changes

in the redox equilibrium between Nia-C and Nia-SR species as a function of the reductive treatment (a: H2 reduced,

b: H2-plus-NaDT reduction) and temperature (indices “1”:298 K, “2”:200 K).

112

Herein, we observed that the relative population of Nia-C (vCO band at 1973 cm–1, vCN bands at 2080 and

2092 cm–1) and Nia-SR´´ (vCO band at 1936 cm–1, vCN bands at 2050 and 2067 cm–1) could be altered by

exposing the samples to different temperatures. Fig. 4.7 shows the IR spectrum of H2-reduced HtSH in a

wide temperature range from 298 to 222 K.

Figure 4.7: IR absorbance spectra of HtSH in a wide temperature range between 298 and 220K. The IR spectroscopic

data reveals a temperature-dependent equilibrium between the Nia-C (νCO stretching region at 1973 cm–1) and Nia-

SR´´ (νCO stretching region at 1936 cm–1) redox states. The ratio between the one-electron reduced, Nia-C and the

two-electron reduced Nia-SR´´ species changes, with the latter increasing at lower temperatures. CN bands marked

with an asterisk (*) refer to signals which were not observed previously for HtSH enzymes. At 222 K there is one

predominant CO stretching peak but four CN stretching peaks instead of two. Therefore, there are two or more

species under the predominant CO stretching vibration of Nia-SR’’. Similar observations were made for the as-

isolated states of the soluble hydrogenase from Cupriavidus Necator (CnSH).28,29

The two species appear to be in a thermal equilibrium shifted towards Nia-SR´´ at lower temperatures. A

third minor species is observed at higher temperatures and possibly represents the Nia-SR sub-form with

a vCO absorption at 1958 cm–1. The IR data recorded at different temperatures confirm that the Nia-C

113

species converts into Nia-SR’’. Notably, at 222 K the Nia-SR’’ state is predominantly populated with a vCO

band at 1936 cm-1. However, four CN bands are observed, suggesting the presence of other species,

presumably Nia-SR’’ sub-forms, with a very similar band position of the respective CO absorption.

Usually, the different redox structural active site states in hydrogenases differ in their CO stretching

frequency, as this mode is most sensitive to redox changes. However, there are also some cases described

in literature in which the CO absorption remained almost at the same position, while the CN stretching

frequencies were changing. An example is a study on the D. fructosivorans [NiFe] hydrogenase. The data

showed that the exchange of residues involved in H-bonding with the active site CN– ligands resulted in

larger shifts for the CN stretching vibrations, whereas the CO stretching mode remained almost

unaffected. Similarly, our IR data at 222 K suggest the enrichment of two Nia-SR’’ sub-forms possibly

differing in the H-bonding network of CN– ligands.30

Additionally, we should recall that Nia-C and Nia-SR’’ are not electronically equivalent (contrary to the well-

known tautomerism observed for Nia-C and Nia-L9) and their interconversion requires the

removal/addition of one electron at the Ni ion, presumably provided/received by the proximal [FeS]-

cluster that is located at close distance 13.5 Å (redox center to redox center distance between the

different cofactors) from the [NiFe] active site, as depicted in Fig. 4.8, below.31

Figure 4.8: Electron transfer pathway in HtSH. Given on top of the black connection lines are the distances between

the [NiFe] active site and the other redox cofactor. These numbers display the center to center distances between

the individual redox cofactors/centers. The numbers in parentheses indicate the edge-to-edge distance in Å. The

above figure is adapted from the crystal structure of NAD+-reducing [NiFe]-hydrogenase in the air-oxidized state

(PDB: 5XF9).31

114

4.1.2 Cupriavidus necator Membrane Bound Hydrogenase

The membrane-bound hydrogenase from C. necator is usually isolated predominantly in the Nir-B state

characterized by a CO stretching vibration at 1948 cm–1 and CN absorptions at 2081 and 2098 cm–1 , as

illustrated in Fig. 4.9 (trace a). After reduction with H2 gas for 1 hour at pH 5.5 and 283 K, the

corresponding room temperature (RT) IR spectra show typical signals of Nia-C (trace b), with a CO

stretching vibration at 1957 cm–1, while low temperature data reveal both Nia-C and Nia-SR’ species,

besides minor amounts of Nia-SR (νCO at 1945 cm–1)32. Upon slow anaerobic oxidation under He

atmosphere, we managed to enrich almost pure Nia/r-S species (Nia-S, harboring a NiII-FeII site, in which

the bridging position between the Ni and the Fe is vacant or Nir-S, harboring a NiII-H2O-FeII active site

configuration) with a CO stretching vibration at 1936 cm–1, identifying also the CN stretching modes at

2075 and 2093 cm–1 (trace c) while aerobic oxidation results in the accumulation of Nir-B (NiIII-OH-FeII) and

a Niia-S (NiII) species (trace d).

Although the vCO band for the Nia/r-S species was known from previous studies,32-34 the corresponding CN

absorptions (Fig. 4.9, trace c) could not unambiguously be resolved, because these species had not been

enriched as a pure state before. Additionally it is not clear so far if the state with CO/CN stretching

vibrations at 1936/2075, 2093 cm–1 is a Nia-S or Nir-S state. It is however important that in this study, the

enrichment of this state was successful. In the future, crystals can be formed and by the geometry of the

active site, which can be resolved by X-ray crystallography, we can have a very strong indication about the

active site state. These studies can be combined with RR spectroscopy. After aerobic oxidation under air,

the Nir-B (NiIII-OH-FeII, vCO at 1948 cm–1 and vCN at 2081 and 2098 cm–1) and the Niia-S (NiII, vCO at 1930

cm–1) states, were enriched.

115

Figure 4.9: 2nd derivative of the absorbance spectra of CnMBH in KiPO4 buffer at pH 5.5 recorded at 283 K, obtained

after different redox treatments. a: as- isolated; b: H2-reduced sample; c: after anaerobic oxidation under He; d:

after aerobic oxidation with air.

The ratio between Nia-C and Nia-SR´ sub-form in MBH has also been shown to vary depending on the

incubation with pure H2 or forming gas (H2:N2, 5:95 % v/v) and the pH of the buffer solution, both

influencing the reduction potential.32,33 After H2 treatment (H2, 100 % v/v), pH 5.5, the respective IR

spectra of the reduced MBH have been recorded in a wide range of temperatures from 298 to 210 K (Fig.

4.10), attempting to detect alterations from the established redox equilibrium at 298 K.

116

Figure 4.10: Absorbance IR spectra of the reduced CnMBH recorded between 210 and 298 K revealed the presence

of a temperature-dependent redox equilibrium between the Nia-C and Nia-SR´ redox states of the protein. Herein,

the ratio between the one-electron reduced, Nia-C and the two-electrons reduced Nia-SR´ species changes drastically

as well as the ratio of Nia-L sub-species along with the Nia/r-S species.

As in the case of the HtSH enzyme (Fig. 4.7, §4.1.1), the population of the Nia-SR´ sub-form in MBH appears to

increase at lower temperatures (vCO band at 1925 cm–1 and vCNs at 2049 and 2071 cm–1) while the amount of Nia-

C (vCO at 1957 cm–1 and vCNs at 2075 and 2097 cm–1) decreases. In addition, we observed the accumulation of some

Nia-L (no light exposure) and Nia/r-S species (vCO at 1936 cm–1). Nia-L species have been already shown to be

populated without the prerequisite of light exposure at cryogenic temperatures. 13,32,35–37

It should be emphasized that the detected changes in the Nia-C/Nia-SR ratios for both HtSH and CnMBH

have important implications for the understanding of the spectral data provided by other methods. For

example, if we would compare IR spectroscopic data of H2-reduced HtSH at 298 K, where the hydrogenase

is enriched in 70% Nia-C and 30% Nia-SR’’, with EPR (Electron Paramagnetic Resonance) measurements at

80 K (Fig. 4.11B) showing no detectable Nia-C signals, then one could assume that the paramagnetic Nia-C

in HtSH is not detectable by EPR. However, the low temperature IR data have now shown that below 200

117

K, Nia-C converts almost completely into the diamagnetic Nia-SR’’ species explaining the discrepancy

between room temperature IR data and low-temperature EPR analysis.

4.2 Photoreactivity of Nia-SR species

4.2.1. HtSH

While in §4.1 the effect of the temperature on the redox equilibrium between Nia-C/Nia-SR species on

HtSH and CnMBH enzymes was analyzed, the focus in this section is set on the photoinduced redox

structural changes in the reduced intermediates of both model enzymes at cryogenic temperatures. The

Nia-C → Nia-L photo-conversion is known and has been studied for almost 40 years. For instance, Nia-L

represents a tautomeric form of Nia-C in which electrons from the hydride are stored on the nickel yielding

a formal NiI species, while the resulting proton has been shown to bind to one of the terminal cysteines

in two Nia-L sub-forms. 2,3,24,38 A Nia-SR to Nia-L photo-transformation was proposed in a resonance Raman

spectroscopic study on the same MBH enzyme, where the authors suggested that the high photon-flux of

the laser during RR experiments on an H2-reduced preparation of MBH might have induced this light-

triggered reaction.4 The lack of mechanistic details of this latter reaction inspired some of the

investigations described in this paragraph.

For illumination experiments, an H2-reduced HtSH sample was irradiated with blue light from a LED source

(λmax 460 nm) at temperatures lower than that of the protein-glass transition (110 K). At temperatures

below the Tg, the glassy material is characterized by a high viscosity and low molecular mobility.39,40 When

the temperature is raised above Tg, the rubbery material exhibits increased molecular mobility, allowing

molecular rearrangements and flow. In a protein sample the glass transition point (Tg is ca. 190-200 K)

expresses macroscopically the increase of viscosity. Microscopically it reflects the intrinsic temperature

dependence of the molecular dynamics of the protein itself and its bound solvent molecules. Above this

temperature the dynamic behavior is dominated by large-scale motion of bonded and non-bonded groups

of atoms. At lower temperatures simple harmonic vibrations predominate.39,40

While hydrogenases are usually investigated by IR (absorption) spectroscopy with the aim to monitor

different redox states via the inherent stretching vibrations of the CO and CN- ligands of the active site,

we applied herein difference spectroscopy to investigate structural changes at the NiFe(CN)2CO cofactor

upon light exposure. IR difference spectra are calculated by subtracting the absorbance spectra recorded

in the dark from the data taken after and/or during light irradiation (light minus dark). The resulting

118

negative peaks represent the parent states, while the positive bands are related to the newly photo-

induced species.

Figure 4.11A shows the corresponding difference spectrum obtained from the spectral data of an HtSH

sample reduced with H2 after and before light exposure. Two negative CO stretching vibrations are

detected, representing the Nia-C (νCO at 1973 cm–1)8 and Nia-SR´´ (νCO at 1936 cm–1)8 species, which

appear to convert into a single (positive) CO band at 1922 cm–1 (corresponding νCN modes are located at

2047 cm–1 and 2062 cm–1) at 110 K. The resulting state is assigned to Nia-L, based on the CO and CN

stretching frequencies observed in other [NiFe]-hydrogenases and corroborated by EPR measurements

(see below)3,24,25,38,41.

Complementary EPR spectra at 80 K (in the dark) of HtSH reduced with H2 (Fig. 4.11B) exhibit characteristic

signals of the Nia-C intermediate (gx = 2.211, gy = 2.139, gz = 2.010). Additional signals of a flavine

mononucleotide radical (semiquinone, g = 2.003) and a reduced [2Fe2S] cluster (gx = 2.027, gy = 1.934) are

discernable in the EPR spectrum in line with previous studies.42–45 Nia-SR is a diamagnetic hydrogenase

intermediate and consequently, it is not observable using EPR.46 Upon illumination with a 460 nm (LED

light), the EPR features of Nia-C vanish in favor of characteristic signals of Nia-L intermediate (gx = 2.278,

gy = 2.110, gz = 2.046). During this process, the EPR signals (at 80 K) related to the [2Fe2S] and FMN (flavin

mononucleotide) remain unchanged. Therefore, we used the intensity of the [2Fe2S] cluster signals as an

internal standard for the quantification of the EPR signals of the Nia-C and Nia-L species. The amount of

Nia-L and Nia-C appears to be ca. 44 % and 26 % of the [2Fe2S] integrated area, suggesting that another

parent state is present (ca. 18 %, according to the subtracted integrated area of Nia-L minus Nia-C),

undetected by EPR that is turned into Nia-L (Fig. 4.11B).46 The related IR spectroscopic data identified this

species as Nia-SR´´ (νCO at 1936 cm–1).

119

120

Figure 4.11: Photoconversion of the reduced intermediates in the HtSH enzyme. The sample was first reduced under

100 % H2 atmosphere (1h, 50 0C) and subsequently the active site redox composition was probed by IR (Α) and EPR

(Β) spectroscopy. Α: Top traces: Absorbance spectra before (black trace) and after illumination (blue trace), bottom

traces: Difference spectrum (light minus dark). The negative peaks represent the CO and CN stretching vibrations of

the parent states, Nia-C and Nia-SR’’ and the positive peaks the light-induced species, namely several Nia-L species.

B: The respective EPR spectrum recorded before illumination (in the dark; black trace) and after illumination (under

light: blue trace) at 80 K. The [2Fe2S] cluster intensity remains unchanged before and after light exposure and can

thus be used as an internal standard for the quantification of Nia-C and Nia-L EPR signals.

Kinetic information on the observed Nia-C/Nia-SR´´ to Nia-L transformations were obtained by calculating

the integral of the CO bands related to the parent and the photo-produced states and subsequent

evaluation of the integrated absorption area as a function of time (Fig. 4.12). For the parent states, mono-

exponential decay curves can be fitted to the time evolution of the CO bands, while a bi-exponential curve

could better reproduce the experimental data for Nia-L formation. Notably, the two time constants

extrapolated from the bi-exponential fit of Nia-L induction appear rather similar to those estimated from

the mono-exponential Nia-C and Nia-SR’’ photo-conversion. This suggests that the Nia-L species is plausibly

formed from both parent states in two independent processes, envisaging a light sensitivity for the Na-

SR´´ intermediate. Furthermore, the analysis of the time constants reveals a much slower process for the

Nia-SR´´-to-Nia-L conversion, which -contrary to the well-studied Nia-C/Nia-L process- is not a simple

tautomerization but also requires the removal of one electron from the active site.

121

Figure 4.12: Kinetic profile of the Nia-C/Nia-SR´´-to-Nia-L photo-conversion at 110 K under 460 nm irradiation. The

integrated νCO area of the respective species in the difference spectra (Figure 4.11a) is plotted as a function of time.

A mono-exponential curve can be fitted to the νCO integral of the Nia-C and Nia-SR’’ species. For Nia-L formation a

bi-exponential curve fit has to be applied. The two time constants extrapolated from the bi-exponential fit of the

Nia-L data resemble those estimated for the Nia-C and Nia-SR´´ decays, supporting the existence of two parent states

converging into Nia-L upon light exposure. At 110 K, the back-reaction kinetics are negligible.

Additional experiments in a temperature range between 140 and 170 K (Fig. 4.13) confirmed the light

sensitivity of one of the Nia-SR´´ sub-forms of the HtSH enzyme at higher temperatures. On the top of each

panel in Fig. 4.13A the corresponding absorbance spectra recorded in the dark (black trace) and during

illumination (blue trace) are shown. On the bottom of each panel in Fig.4.13A the resulting difference

spectra light minus dark (dark blue trace) and after dark retrieval when the light is turned off minus light

(light blue trace) are depicted.

122

123

Figure 4.13: A: Nia-C and Nia-SR’’ to Nia-L transformation between 130 K and 170 K. On the top of each panel the

absorbance spectra recorded in the dark (black trace) and after illumination (blue trace) are depicted. On the bottom

of each panel the resulting difference spectra, light minus dark (dark blue trace) and the data taken after dark

retrieval minus light (light blue trace) are depicted. B: The related area of the CO absorption of each species (Nia-C,

Nia-SR’’ and Nia-L) is integrated as a function of time. For each of the parent states mono-exponential curves were

fitted and a bi-exponential growth curve to Nia-L, except for the data points recorded at 170 K. At 170 K mono-

exponential curves can be fitted to all of the points.

The vCO area of each species was integrated and plotted as a function of time (Fig. 4.13B). Similarly to

the data at 110 K, the decay of the parent states (Nia-C and Nia-SR’’) was best described by a mono-

exponential function compared to the photoinduction of Nia-L, for which a bi-exponential function was

required. In view of the lower photo-transformation of Nia-SR´´ into Nia-L at high temperatures, we

observed that at 160 and 170 K, Nia-L is almost exclusively formed by Nia-C and therefore in the latter case,

the Nia-L light-triggered accumulation can be better described by a mono-exponential curve.

At each temperature, we also studied the relaxation kinetics after switching off the light, monitoring the

time evolution of the integrated vCO band intensities (Fig. 4.14). At 110 K, Nia-L does not relax back to

the parent states. At temperatures above 110 K mono-exponential decay curves can be fitted both to the

parent states and the photoinduced species (Nia-L). Fitting a biexponential curve to the Nia-L relaxation at

130 and 140 K was attempted, but the fit did not converge at 130 K.

124

Figure 4.14: Time-dependence of the integrated vCO band intensities monitored in the dark after switching off

irradiation (see Fig. 4.13). The data display the evolution of the species Nia-C, Nia-SR’’ and Nia-L. In each case, a

monoexponential function was fitted to the data.

The corresponding time constants derived from the fits in Figures 4.13 and 4.14 are listed in Table 4.1.

Table 4.1: Time constants of the formation and decay of Nia-L from Nia-SR’’ and Nia-C, taken from the

monoexponential fits to the data in Figures 4.13 and 4.14.

Temperature Nia-SR’’ ⇄ Nia-L Nia-L

→

Nia-SR’’ Nia-C ⇄ Nia-L Nia-L

→

Nia-C

110 Κ 13.8 min * Nia-L does not relax 2.22 min * Nia-L does not relax

130 K

13.6 min

108.4 min

1.75 min

32.5 min

140 K

11.7 min

29.5 min

1.98 min

27.6 min

160 K

33.1 min

12.8 min

18.7 min

6.99 min

125

In order to rationalize these data, one may assume the simplest reaction mechanism that is based on two

independent photochemical reaction routes from Nia-C and Nia-SR´´ to Nia-L. This mechanism is expected

to hold for temperatures at which transitions between Nia-C and Nia-SR´´ can be neglected (T < 200 K; see

sect. 4.1.1.). Furthermore, the Nia-L species formed from Nia-C and Nia-SR´´ do not seem to interconvert

in the temperature range under consideration since, within the experimental error, the portion of

photoinduced Nia-L from Nia-C (Nia-SR´´) is the same as that recovered from Nia-C (Nia-SR´´) in the dark.

This suggests that in a first approximation we may analyze the transitions Nia-C-to-Nia-L and Nia-SR´´-to-

Nia-L separately in this temperature region.

Considering the data in Figures 4.13 and 4.14 we first note that the extent of Nia-L formation from Nia-

SR´´ relative to that from Nia-C strongly decreases with increasing temperature. Assuming that the

proportionality factor that relates the band intensities with the concentration is the same for all species,

this tendency is reflected by the ratio (Nia-SR´´→ Nia-L)/(Nia-C → Nia-L) that decreases from ca. 1 at 110 K

to ca. 0.38 at 140 K. At 160 K, the ratio is much lower since the photoreaction Nia-SR´´→ Nia-L is close to

the detection limit. Neglecting any reverse photoprocesses to the parent states and thermal formation

routes to Nia-L, the reciprocal time constant for Nia-L formation is given by the sum of the rate constants

for photochemical Nia-L formation l0 and dark reversion kr. The latter constant can be determined directly

from the dark reversion time constant such that also the photochemical rate constant is readily obtained.

Table 4.2: Rate constants for the formation and decay of Nia-L from Nia-SR’’ and Nia-C, assuming two independent

and non-coupled reaction pathways. Rate constants were calculated from the time constant in Tables 4.1 as

described in the text, taken from the monoexponential fits to the data in Figures 4.13 and 4.14.

Temperature l0(Nia-SR’’ ⇄ Nia-L)

x 10

−

3 s

−

kr(Nia-L → Nia-SR’’)

(Ni

-C

⇄

-L)

kr(Nia-L → Nia-C)

x 10

−

3 s

−

1 x 10

−

3 s

−

1 x 10

−

3 s

−

110 Κ

1.2

ca. 0

7.5

ca. 0

130 K

1.1

0.2

0.5

140 K

0.9

0.6

7.8

0.6

160 K

“−0.8”

1.3

“−1.5”

2.4

The data in Table 4.2 show that the assumed mechanism of two independent reaction pathways to and

from Nia-L provides consistent results only in the temperature range from 110 to 140 K, since at 160 K

physically meaningless negative values are obtained for l0. Between 110 and 140 K, the rate constants for

the photoinduced Nia-L formation seem to be largely temperature-independent. This finding suggests that

the underlying proton transfer is associated with a very low activation barrier; possibly the transfer occurs

126

via intramolecular proton tunneling. In contrast, dark reversion reveals an appreciable energy barrier,

particularly in the case of the Nia-L → Nia-SR’’ transition, which is reflected by the increases of the rate

constant with the temperature.

The present data also provide new insights on the catalytic mechanism of [NiFe]-hydrogenases. So far it

has been argued that sequential catalytic events comprise Nia-S ⇄ Nia-SR ⇄ Nia-C ⇄ Nia-L, returning to

Nia-S according to the scheme presented in the theoretical part. By making use of light under cryogenic

conditions, we observed that certain intermediates might be “partially” bypassed. Finally, the conversion

of Nia-SR’’ (NiII-H-Fe) to Nia-L (NiI-Fe) is not a simple tautomerization (as observed for Nia-C and Nia-L) but

requires the additional removal of one electron from the hydrogenase’s active site. Possible electron

acceptor candidates in HtSH are either the [FeS]-clusters or the FMN prosthetic groups.

127

4.2.2. Charaterization of the various prosthetic groups in HtSH

In chapter 4.2.1 we have elucidated a new photo-induced transformation for [NiFe]-hydrogenases, in

which a Nia-SR sub-form converts into Nia-L. As stated above this process also requires an electron removal

from the [NiFe] site, plausibly transferred to other (metal) cofactors in the enzyme. Herein, we

characterize these additional prosthetic groups using EPR spectroscopy, attempting to shed light on the

possible electron acceptor. Low temperatures are usually used for the EPR detection of [FeS]-clusters due

to their fast relaxation properties.9,47 An H2-reduced HtSH sample was investigated at 10 K. The EPR data

of HtSH in the dark (black curve) and after illumination (blue curve) are shown in Figure 4.15.

Figure 4.15: EPR data of the photoconversion of HtSH at 10 K. The black trace depicts the spectrum which is

measured in the dark while the trace in blue refers to the spectrum taken after illumination. Signals form the [NiFe]

active site, a [2Fe2S] cluster and traces of [4Fe4S] species have been detected. No detectable changes were noted

by comparing the spectra before and after illumination. The spectra were measured by Dr. Christian Lorent.

128

The EPR spectrum (black curve) taken in the dark reveals major contributions from a reduced [2Fe2S]

cluster, while the observed minor trace signals (g=1.86) could possibly be derived from reduced [4Fe4S]

clusters. In addition, this species does not change upon illumination (blue curve). In line with the EPR data

recorded at 80 K in the dark, Nia-C is also detected. Notably, illumination of the EPR sample results in the

formation of several Nia-L species (Fig. 4.15, blue trace). The appearance of various Nia-L species has also

been detected in the regulatory [NiFe]-hydrogenase from C. necator.2 According to available

crystallographic data, HtSH houses four [4Fe4S] clusters distributed within the hydrogenase small subunit

HoxY ([4Fe4S]Y1), the subunits HoxU ([4Fe4S]U2 and [4Fe4S]U3) and HoxF ([4Fe4S]F1) of the diaphorase

module of the enzyme (Fig. 4.15).31 In the as-isolated form of the enzyme the [4Fe4S] clusters reside in

the diamagnetic +2 state and are thus EPR silent. Upon reduction we would have expected to detect

signals from various [4Fe4S]+ species. The lack of EPR signatures of the [FeS]-clusters might be ascribed

predominantly to a very fast spin relaxation from these metal sites. Similarly, the [4Fe4S] cluster N5 in

Complex I from E. coli,48 which is homologous to cluster U3 of HtSH has been shown to be difficult to

detect and only measurements at 3 K revealed its presence.

To validate this assumption, we performed EPR measurements at 5 K. EPR data recorded at 5 K reveal

broadening and partial splitting of the Nia-C signal, indicative for a magnetic interaction between the

paramagnetic [NiFe] active site and the reduced proximal [4Fe-4S]+ cluster (Y1). This splitting is detected

exclusively at very low temperatures (T < 6 K), while higher temperatures (e.g., at 10 K) often result in a

fast relaxation and the spin-spin coupling is averaged out.44 Similar phenomena have been observed for

other [NiFe]-hydrogenases. In addition, the signals of [2Fe2S]+ cluster at 5 K are highly saturated and

distorted compared to data at 10 K. Illumination at 5 K led to the disappearance of Nia-C signals followed

by the appearance of various Nia-L species (as observed for data at 10 K). Upon illumination the EPR

spectrum exhibits a very broad feature at g=1.82 and a weaker signal at g=1.89 (F1). These values could

fit to the gz values of more than one [4Fe4S] cluster. Additional changes in intensity (upon illumination)

were observed for the [2Fe2S]+ species that could be attributed to changes in the power-dependent

saturation properties of the metal site. Likely, these changes originate from the coupling of the [2Fe2S]+

cluster (U1) to a nearby paramagnetic [4Fe4S]+ cluster. The best candidates, based on their distance from

U1, are either the [4Fe4S] cluster U2 or F1 (see Fig. 4.16).31

129

Figure 4.16: EPR data of the photoconversion in HtSH at 5 K. The black trace is the spectrum which is measured in

dark before illumination and the blue trace is the spectrum taken after illumination. At 5 K some signals from the

[4Fe4S] are additionally observed, besides those of the active site and the [2Fe2S]. The spectra were measured by

Dr. Christian Lorent.

As the EPR data of H2-reduced HtSH did not reveal significant contributions from [4Fe4S] centers, we

further assessed the protein sample using stronger reducing agents like NaDT. This procedure is supposed

to limit the shifts of electrons between cofactors, possibly affecting the photoconversion of Nia-SR’’ into

Nia-L (the latter transition requires an oxidized electron acceptor proximal to the [NiFe] cofactor).

Figure 4.17 shows the IR spectrum of the same HtSH protein batch recorded at 100 K either after reduction

with H2 (top) or an additional treatment with NaDT (H2-plus-NaDT, bottom trace). After a combined H2-

plus-NaDT exposure, a broad CO peak, possibly related to (various) Nia-SR species, appears at 1934 cm–1.

It can be deconvoluted in two absorptions at 1933.7 and 1936 cm–1, respectively. The predominant band

in the NaDT-treated HtSH (i.e., 1934 cm–1) appears slightly red-shifted compared to the major absorption

of the Nia-SR´´ at 1936 cm–1 of the H2-reduced species. Small shifts (1-3 cm–1) of the CO stretching vibration

130

for the same redox state have been often associated to either changes in the protonation state (acid-base

equilibria) in proximity of the [NiFe] site, i.e. not the first coordination shell, or redox changes of the

proximal [FeS] cluster. Considering that the only difference in the two reductive treatments is the

presence of NaDT, we ascribed the CO band at 1934 cm–1 to a Nia-SR´´ species in proximity of a reduced

proximal [FeS] cluster. The reduction of this [4Fe-4S] cluster, results in an increase of electron density at

the [NiFe] active site, an associated increase in the π* back-bonding in the Fe-CO bond and a red shift of

the vCO frequency.49 This may explain the observed spectral shift of the CO stretching vibration when

NaDT is present.

Figure 4.17: Absorbance spectra of the H2-reduced (top) and the H2-plus-NaDT-reduced sample (bottom), measured

at 95 K. The black trace represents the experimental spectra, whereas the grey and green traces are fitted band

shape functions. The green spectrum shows the Nia-SR’’ species that is known to photoconvert.

131

Figure 4.18, shows the 2nd derivative of the IR spectra presented in Figure 4.17. Traces 1 and 3 represent

the spectra in dark, while traces 2 and 4 the spectra of species trapped after light exposure. Only the Nia-

SR’’ sub-form with a CO band at 1936 cm–1 from the H2-reduced sample appears to photoconvert to Nia-

L. In contrast, the main Nia-SR species of the H2-plus-NaDT-treated sample proved to be not light-sensitive

corroborating the enrichment of a Nia-SR´´ species associated with a reduced proximal cluster that is

unable to accept electrons from the [NiFe] active site.

In Figure 4.18b the corresponding difference spectra “light minus dark” are shown for better visualization

of the spectral differences of the H2-reduced sample (black curve) and the H2-plus-NaDT reduced sample

(grey curve). In the first case, both CO absorptions of Nia-C (1971 cm–1) and the Nia-SR’’ sub-form (1936

cm–1) transform into Nia-L while the NaDT treatment impaired the latter transformation (no Nia-C can be

detected).

Additional EPR data (Fig. 4.18c) recorded at 10 K from the H2-plus-NaDT-reduced HtSH sample revealed

an increase of the intensities of the [2Fe2S] and [4Fe4S] cluster signals, suggesting a higher population of

reduced metal cofactors. When comparing the Nia-C fingerprint region of the EPR spectra of H2- and H2-

plus-NaDT-reduced HtSH samples, we noticed a significant decrease of the Nia-C state in the sample that

contains NaDT, in line with the related IR data (Fig 4.18a). That means that the active site predominantly

exists in an EPR-silent state, presumably the Nia-SR’’ species, containing a reduced proximal cluster. This

is further corroborated by an increase of the signal intensities in the [4Fe4S] region. In summary, these

data suggest that a combined HtSH treatment with H2 and NaDT hinders the photoconversion of Nia-SR’’

to Nia-L, by reducing the proximal [FeS] cluster.

132

133

Figure 4.18: Comparison of different reduction procedures (H2 vs. H2-plus-NaDT). a: 2nd derivative of the absorbance

spectra, measured at 110 K, top: H2-reduced sample recorded in the dark (black trace) and after illumination (orange

trace), bottom: H2-plus-NaDT-reduced sample recorded in the dark (grey trace) and after illumination (orange trace),

b: IR difference spectra derived from data taken at 110 K (light-minus-dark, CO stretching region), black trace: H2-

reduced only sample, Grey trace: H2-plus-NaDT-reduced sample, c: EPR spectra recorded at 10 K in the dark, black

trace: H2-reduced sample, grey trace: H2-plus-NaDT-reduced sample At 10 K some signals from the [4Fe4S] are also

observed, besides those deriving from the active site and the [2Fe2S]. The EPR spectra were measured by Dr.

Christian Lorent.

4.3 Characterization of the Nia-L species in Hydrogenophilus Thermoluteolus Soluble

Hydrogenase (HtSH)

In the previous chapter we revealed that a Nia-L species can be photo-induced upon illumination of

samples residing in the Nia-C and Nia-SR´´ redox states of HtSH. We recall that Nia-L represents a tautomeric

form of Nia-C in which electrons from the bridging hydride are stored on the nickel yielding a formal NiI

species, while the resulting proton may be transferred to different amino acid residues of the protein

backbone depending on the catalytic mechanism. Various studies on [NiFe]-hydrogenases reported the

existence of several Nia-L species dependent on pH, temperature, and light.25,50,51 Similar to the various

Nia-SR species, the differences between the different Nia-L species is not fully unraveled. In a recent report

on the RH (Regulatory Hydrogenase) enzyme from C. necator two Nia-L species, namely Nia-L1 and Nia-L2,

were stoichiometrically enriched. Both species were found to harbor a H+ at one of the Ni-bound terminal

cysteines. The detection of protonated cysteine thiolates in [NiFe]-hydrogenases proved to be challenging

and up to date only two studies on DvMF3,38 and CnRH2 enzymes resolved faint νS-H stretching modes.

After having focused on the Nia-C → Nia-L photoconversion prior to this chapter, the different Nia-L species

will be discussed in the following. Besides the Nia-L1 species (with a νCO at 1922 cm–1), a second Nia-L

species, was detected, herein named Nia-L2, which exhibits an SH stretching absorption peak in the IR

spectrum similar to that observed for the DvMF enzyme. The Nia-L2 is gradually enriched at temperatures

above 110 K.

Figure 4.19A shows the absorbance spectra measured in the dark at selected temperatures from 130 K to

170 K (black trace) and the respective spectral data taken after the sample was illuminated (grey curve),

whereby Figure 4.19B depicts the respective difference spectra. The black trace shows the difference

spectra “light minus dark”. The grey curve depicts the corresponding relaxation, “dark minus light’’, after

the light was turned off. These data show that the Nia-L1 to Nia-SR’’ relaxation is more complete at higher

134

in contrast to lower temperatures, when comparing the amount of Nia-L1 that has relaxed to Nia-SR’’ at

130 K and 140 K versus 110 K, where no relaxation is observed.

135

Figure 4.19: IR spectra recorded at selected temperatures, a. 130 K, b. 140 K, c. 160 K and d. 170 K. A: Shown

absorbance spectra were recorded before (black) and after illumination (grey). B: Depicted difference spectra ‘light-

minus-dark’ (black, photo conversion) and ‘dark-minus-light’ (grey, after thermal relaxation). The back-reactions,

namely Nia-L to Nia-C relaxation and Nia-L to Nia-SR’’, become faster at elevated temperatures. Above 160 K a 2nd Nia-

L can be unambiguously detected, denoted in the following as Nia-L2 with distinct vCO and vCN modes at 1924 and

2053, 2066 cm–1.

At temperatures lower than 140 K, the solely photo-produced Nia-L species exhibits a vCO mode at 1922

cm–1 and vCN modes at 2047 and 2062 cm–1, respectively. In contrast, above 140 K another species is

produced, herein denoted as Nia-L2, characterized by a CO stretching band at 1924 cm–1, and CN

absorptions at 2053 and 2066 cm–1, as shown in Fig. 4.19. At 160 K the Nia-L2 species accumulates. By

comparing the grey curve at 140 K (Figure 4.18 B, b), at which Nia-L1 is the predominant species, and the

grey curve at 160 K (Figure 4.18 B, c), where Nia-L2 prevails. Notably, the first species relaxes already at

lower temperatures than the latter one. In addition, this second Nia-L2 species exhibits a distinct EPR

signature with gx = 2.339, gy = 2.087, and gz = 2.050.52 (Fig. 4.20)

Figure 4.20: EPR spectrum of H2-reduced HtSH measured at 170 K after 60 min of illumination. Some signals from

the FMN cofactor are also observed, besides the dominant ones from the active site. The spectrum was measured

by Dr. Christian Lorent.

136

Further analysis of the SH stretching region reveals a characteristic peak at ca. 2500 cm–1 which is solely

correlated with the Nia-L2 enrichment and not to the Nia-L1 species. Figure 4.21, top curve, shows the

difference spectrum light minus dark (black trace) and the difference spectrum, after the light is turned

off versus dark, dark retrieved minus dark (red trace). It was shown before that the Nia-L1 relaxes faster

than Nia-L2. Therefore, the red curve in Fig. 4.21 depicts the difference spectrum after Nia-L1 already

relaxed. In panel 1, one can see that the SH stretching peak intensity is unaffected, therefore it is not

related to Nia-L1 and it is only related to Nia-L2. Further, the bottom black curve in Fig. 4.21 shows the

difference spectrum “light minus dark” at 110 K, in which the predominant species is Nia-L1. In this case

(Fig. 4.21, panel 2), no band is detectable in the SH stretching region.

Figure 4.21: (Top) “Light-minus-dark” difference spectrum recorded at 160 K (black) and “relaxation-minus-dark”

difference spectrum taken during thermal relaxation after illumination at the same temperature (red). In the dark

spectrum the Nia-L1 species accumulate only in traces (orange asterisk). The red trace shows the difference spectrum

after Nia-L1 has relaxed to the parent state. The fact that the SH stretching remains unchanged and is not observed

at 110 K means that it is related solely to the Nia-L2. (Bottom) “Light-minus-dark” difference spectrum recorded at

110 K (black). Insets: Expanded view on the SH stretching region.

137

These signals have been assigned to a protonation event of one of the terminal cysteines in the Ni

coordination sphere which was investigated in several studies such as the one by Tai et.al.38 and Waffo

Tadjoung et.al.2 This suggests that the hydride of the Nia-C state dissociates upon illumination and is

translocated from the bridging position between Ni and Fe to one of the terminal cysteines, presumably,

C80.

4.4 Characterization of the Nia-L species of Cupriavidus Necator Membrane Bound

Hydrogenase (CnMBH)

In paragraph 3.3 two Nia-L species could be revealed in HtSH enzyme. Herein, we focus on light-induced

species in our 2nd model enzyme, the MBH from C. necator. Upon reduction with 100% H2, CnMBH mainly

resides in the Nia-C and Nia-SR’ redox species (Fig. 4.22, trace A.1). At 110 K the main species is Nia-SR’

(Fig. 4.22, trace A.2). Upon illumination at 95 K, only Nia-C (νCO at 1961 cm–1) photo-converts into a Nia-L

species, characterized by a CO band at 1911 cm–1 (Fig. 4.22, trace A.3). In Fig. 4.22B, the corresponding

difference spectra “light minus dark” (3-2) are depicted.

An inspection of the IR spectral region of the hydrogenase CN– ligands (2150-2000 cm–1) shows that two

species are generated upon irradiation of Nia-C. The two species exhibit overlapping CO absorptions,

which results in a single band at 1911 cm–1, but distinct CN signals at 2040, 2045, and 2077 (broad) cm–1.

Such an effect is rarely observed in [NiFe]-hydrogenases,28 since the νCO band is usually more sensitive to

electron density changes at the Fe, being a stronger π-acceptor compared to the CN- ligands.53

Another interesting area is the SH stretching region in which an intense negative (2516 cm–1) and positive

band (2523 cm–1) are detected in close vicinity. In addition, a similar negative-positive peak is found in

the 1750 cm–1 region of the spectrum. However, the latter is hard to interpret. Peaks in this region may

be related to a protonation of the COOH chains of Glu or Asp residues. The position of the corresponding

C=O stretching modes may vary from ca. 1710 to 1750 cm–1 depending on the H-bonding interactions.54,55

138

139

Figure 4.22: A: Absorbance spectrum of CnMBH depicted 1. at RT, 2. at 95 K in the dark, and 3. after illumination. B:

IR difference spectrum showing the Nia-C-to-Nia-L photo-transformation in CnMBH at 110 K. The spectrum displays

the differences between the spectra before and after the light irradiation. Grey box: SH stretching region, Red box:

CN stretching region, Blue box: region of modes from Glu/Asp side chains.

As in the case of HtSH enzyme, we investigated this photo-induced transformation at different

temperatures from 95 K to 200 K. The difference spectra depicted in Fig.4.23 indicate the formation of

other light-induced species, here named Nia-L3, with a CO stretching mode at 1888 cm–1 and Nia-L3’

species with CO stretching vibration at 1881 cm–1. The Nia-L3 species seems to be enriched on cost of Nia-

L2 whereas the Nia-L3’ species occurs, when illuminating above 155 K. Additionally, at 200 K a new species

is formed with a CO position which potentially fits to Nia-S. This observation may indicate that Nia-L3 is in

equilibrium with Nia-S.

Figure 4.23: IR difference spectrum depicting the Nia-C-to-Nia-L transition of CnMBH in the temperature range from

155 K to 200 K. The data displays the difference between the spectra before and after the light irradiation and how

the ratio between the different Nia-L species changes as a function of the temperature.

140

So far, it is known that the Nia-L3 is enriched at higher temperatures. Nevertheless, some open questions

remain. Thus, it is not clear, whether or not, Nia-L3 is formed directly from Nia-C or from Nia-L2. In addition,

it is uncertain if this species is formed in the dark or upon irradiation. To answer this question the following

experiment was conducted. First, the sample was measured at 160 K in dark (1), then the sample was

measured at 95 K in dark (2). Further the sample was illuminated at 95 K (3), then the light was turned off

and the cell was heated up to 160 K (4), afterwards the sample was cooled down to 95 K again (5).

Figure 4.24: IR difference spectrum of the Nia-C-to-Nia-L transition in CnMBH at 95 K, before (black trace) and after

heating up the sample to 160 K. The shown data depict the difference between the spectra before and after sample

illumination.

Indeed it could be shown, that the Nia-L3 species can be enriched in the dark from Nia-L2 when the cell is

heated up even if the light is turned off after the formation of Nia-L2 at 110 K. The black curve depicts the

141

respective “light-minus-dark” difference spectrum at 95 K, corresponding to the difference between the

experiments (3)-(2). The red curve is the difference spectrum of the spectra recorded at the (5)-(3) steps

of the experiment.

4.4.1 C81S variant of CnMBH

Besides the CO/CN absorptions of the hydrogenase diatomic ligands, we observed in the IR difference

spectrum a negative absorption at 2516 cm–1 for the Nia-C species, which shifts to 2523 cm–1 in the photo-

induced Nia-L species. These signals appear in a spectral region usually dominated by S-H stretching

vibrations. The observation of two νSH absorptions in MBH is quite puzzling as protonated thiolate for the

Nia-C species was not expected, considering that all reported spectroscopic and structural data of [NiFe]-

hydrogenases favor a deprotonated Ni-bound cysteine for this intermediate. Secondly, both νSH bands

have a higher integrated absorption coefficient suggesting a stronger polarization of the S-H bond of the

involved cysteines. A sequence alignment of CnMBH with the RH from C. necator and the MBH of the

DvMF enzyme, reveals that CnMBH contains an additional cysteine residue in proximity to the

hydrogenase active site (Cys81). This residue is visualized in Fig. 4.25, in which a collection of x-ray

structural data reveals different conformations of protonated Cys81 in the corresponding reduced and

oxidized redox structural active states.

Figure 4.25: A: As-isolated CnMBH (PDB structure 3RGW)56 ; B: fully (H2) and partially reduced (ascorbate) CnMBH

(PDB structures 4IUC57 and 4IUD57). The structures depict the NiFe active site with the bridging cysteines C600 and

C78 as well as the H82 and C81 residues. As shown in the 3RGW structure, C81 (in the red-dashed oval) is very

flexible, adopting two distinct configurations. Depending on the configuration, C81 could form a hydrogen bond

either with H82 or C78. In the reduced and semi-reduced crystals two distinct orientations of C81 (in red-dashed

oval) were isolated.

142

Cys81 is located in hydrogen bond distance from Cys78 (2.9 Å), which bridges the Ni and Fe ions, and also

to the CO of the active site (3.3 Å). The proposed H-bonds might result in a strong polarization of the S-H

bond of Cys81 leading to higher integrated absorption coefficients for the two SH bands. To validate this

assumption, we exchanged Cys81 with the isosteric serine residue.

Figure 4.26: NiFe cofactor of the WT CnMBH, depicting the possible hydrogen bond distance between the bridging

C78 and C81 (2.9 Å). Structural representations are based on 4IUC.58

The biochemical data of the MBH variant have been previously described.59,60 In addition to some

spectroscopic studies.32 IR spectroscopic data at 283 K are shown in Fig. 4.27. In the as-isolated form, the

protein resides in two inactive redox states, Niu-S (presumably NiII-OOH-FeII) characterized by CO and CN

bands at 1943, 2081, and 2105 cm–1, and Nir-B (NiIII-OH-FeII) with CO and CN absorptions at 1955, 2081,

and 2098 cm–1, respectively. After ca. 1 h reduction with 100 % H2, minor Nia-C species are formed,

highlighted by the appearance of CO and CN modes at 1964 cm–1 and 2073 and 2101 cm–1, respectively.

Besides Nia-C, a new CO band is detected at 1929 cm–1. Considering that this peak disappears after 80 min

of reduction with H2, we attributed this band to remnants of a Nia/r-S population. After extensive reduction

with H2 and similar to the native MBH enzyme, the predominant species in the C81S protein variant is Nia-

SR’ with CO and CN stretching mode at 1932 cm–1, 2047 and 2076 cm–1. To confirm that these states are

healthy states of the active site we re-oxidized the protein. The difference spectra “re-oxidized versus

reduced” as well as the 2nd derivative of the absorbance spectra are depicted in the Appendix, Fig.A.4.2.

143

144

Figure 4.27: A: 2nd derivative of the absorbance IR spectra of the C81S variant of the CnMBH recorded at 10 0C

represent from the top to bottom the as-isolated (oxidized) sample and the corresponding reduced sample after

incubation with 100% H2 for 55 and 80 min. The as-isolated sample resides in a mixture of Nir-B and Niu-S species.

After 55 min of reduction, the Nia-C and Nia/r-S states are detected. After 80 min of reduction, the Nia-SR’ species is

enriched. B: IR difference spectrum of the WT CnMBH and the C81S variant both recorded at 95 K.

At cryogenic temperatures, Nia-SR’ is enriched on expense of Nia-C and other Nia-SR sub-forms, similar to

the observations in native MBH. After reduction of the C81S protein variant, the sample was cooled down

to 95 K and illuminated with 460 nm LEDs. Figure 4.27 B shows the IR difference spectra “light minus dark”

of the C81S protein variant (black curve) and WT CnMBH (red curve). The exchange of Cys81 by a serine

led to a splitting of the CO stretching band for both Nia-C and Nia-L species. In the difference spectrum

(Fig.4.27 B), two negative bands were observed for Nia-C with CO absorptions at 1954 cm–1 and 1967 cm–

1 which are light-converted into two different Nia-L species characterized by CO bands at 1903 and 1916

cm–1. The corresponding CN absorptions are listed in Table A.4.3, revealing that the amino acid exchange

has a rather drastic effect on the spectroscopic features of the CO ligand of the hydrogenase, while the

CN signals are only weakly perturbed. Possibly, the reduced O-H bond lengths compared to the S-H

analogues result in a reduced ability of serine to form H-bond with either Cys78 or the CO ligand (see

above Fig. 4.27), thus, inducing a splitting of the observed CO stretching bands for both Nia-C and Nia-L

species.

Taking into account an early computational work on the C81S MBH61 protein variant, we can attribute the

Nia-C and Nia-L species with CO bands at 1967 and 1916 cm–1 to an active site arrangement in which serine

has an identical conformation to that of Cys81. The substitution of Cys81 by serine induces also an upshift

of the CO stretching (Nia-C: νCO at 1967 vs 1960 cm–1; Nia-L: νCO at 1916 vs 1910 cm–1), presumably related

to the higher polarity of the serine side chain, which in turn reduces the back-bonding from the iron and

strengthens the CO bond compared to that of native MBH. In summary, these data suggest that Cys81 is

indeed localized at H-bond distance from the hydrogenase active site (Cys78 and/or CO ligand) and its

substitution with a serine has slightly altered the hydrogenase active site arrangement.61 Depending on

their orientation some serine molecules do not form hydrogen bonds with the bridging cysteine leading

to a downshift of the CO stretching compared to the wild type (Nia-C: νCO at 1954 vs 1960 cm–1; Nia-L: νCO

at 1903 vs 1910 cm–1), whereas other S81 molecules with similar orientation as the C81 in the wild type

protein, can presumably (even if serine is shorter) form a hydrogen bond with e.g. the bridging cysteine

leading to an upshift of the CO stretching compared to the values in the wild type protein (Nia-C: νCO at

145

1967 vs 1960 cm–1; Nia-L: νCO at 1916 vs 1910 cm–1). Interestingly illumination of the sample at 170 K

results in the accumulation of Nia-L3 species, similar to the wild type enzyme (Appendix Fig.A.4.3).

In the panel of Figure 4.27B the SH stretching region of the wild type MBH and the C81S variant is depicted

in the black and red curve, respectively. The inspection of the SH stretching region does not reveal any

peaks, neither negative nor positive, supporting the hypothesis that indeed the protonated thiolate of

Cys81 is probed in the IR difference spectra (light-minus-dark) of the wild type MBH. Further, it was

observed that the intensity of the SH stretching frequency in the WT CnMBH depends on the temperature,

as depicted in Appendix, Fig.A.4.4.

4.5 Conclusions

In this chapter, we investigated the pronounced temperature-dependent changes of the ratio between

two hydride-bound catalytic intermediates, namely Nia-C and Nia-SR, observed in the NAD+-reducing

[NiFe] hydrogenase from Hydrogenophilus thermoluteolus (HtSH) and the membrane bound hydrogenase

from Cupriavidus Necator (CnMBH), as model enzymes. In the case of HtSH, temperature-dependent

alterations are observed in the Nia-C-to-Nia-SR’’ ratio, whereas for CnMBH wild type temperature-

dependent variations of the ratio between Nia-C and Nia-SR’ are detectable.

At cryogenic temperatures the samples were illuminated with 460 nm LED light. For HtSH, experimental

data that are provided for the first time demonstrate a light-induced conversion of the fully reduced

species of the catalytic cycle of a [NiFe] hydrogenase. This process contains a light-triggered hydride

dissociation from the bridging position (NiII-H-FeII) of the Nia-SR, similar to the process occurring upon

illumination of the Nia-C species. However, there is a crucial difference. Although the Nia-C to Nia-L

represents a photo-tautomerization, because both species are reduced by one electron, the Nia-SR’’ (2 e–

reduced) to Nia-L (1 e– reduced) photo-conversion requires an additional electron transfer across the

borders of the active site.

Illumination of the H2-reduced sample at 5 K leads to an increase of the [4Fe4S] signal intensity compared

to the sample measured in the dark with EPR. This could imply that the [FeS] clusters function as electron

acceptors. The reduced [4Fe4S]+ clusters are paramagnetic. However, the Nia-L species which accumulate

at this temperature are different to those observed at 80 K and the results are not yet conclusive. IR

measurements carried out after reduction of the sample by NaDT and H2 show mainly an accumulation of

the 2-e– reduced Nia-SR’’ state, together with minimal contribution of other subspecies. Notably, none of

the Nia-SR’’ species photoconvert. EPR measurements reveal that reduction by H2 and NaDT resulted in

more reduced [FeS] clusters compared to the reduction only by H2. Thus, the data indicate that a stronger

146

[FeS] cluster reduction inhibits the photo-conversion, conceivably by impeding the capability of the

clusters to act as electron acceptors.

Illumination of HtSH above 140 K revealed peaks in the SH stretching region, providing structural details

on the Nia-L species. Herein, two Nia-L were accumulated, Nia-L1 below 95 K, and Nia-L2 above 140 K. For

Nia-L2, an absorption attributable to the SH stretching was discovered, which was not present in the Nia-

L1 species. Additionally, the two photo-induced species reveal similar CO stretch frequencies, indicating

that they feature the same protonation state. Therefore, it is proposed that the terminal cysteine (possibly

C80) is protonated in both cases, but in Nia-L2 it is hydrogen bonded to another amino acid, presumably

the Glu (E32), increasing the polarizability of the bond.

Distinct peaks in the SH stretching region are also observed in the corresponding light minus dark

difference spectrum when the CnMBH wild type sample is illuminated. The related negative/positive

signal pair in the difference spectrum as well as the intensity of the signals compared to the active site

signals, are not common in other hydrogenases. We assume that this is not related to the protonation of

the terminal cysteine (alone). It is rather more probable, considering the peak intensity, that more than

one cysteine, possibly connected via hydrogen bonds, are involved.

Interestingly, it is observed that when the vCO of the active site is blue-shifted the SH respective stretching

peak is red-shifted and vice versa, indicating that this intense peak belongs to a cysteine close to the active

site. An exchange of Cys81 with an isosteric serine residue resulted in the C81S variant and the

negative/positive signal pair was not detected anymore in the corresponding difference spectrum. Serine

is smaller than cysteine and possibly the formation of a hydrogen bond is largely impaired.

Therefore, the couple of peaks in the SH stretching region is related to the C81, which is unique in CnMBH

and located in the 2nd coordination sphere of the active site. These findings underpin the need for careful

interpretation of the peaks in this spectral region. The results also suggest that C81 may be involved in an

alternative H+ transfer pathway.

147

4.6 References

1. Lorent, C. et al. Shedding Light on Proton and Electron Dynamics in [FeFe] Hydrogenases. J. Am.

Chem. Soc. 142, 5493–5497 (2020).

2. Waffo, A. F. T. et al. Structural Determinants of the Catalytic Ni a ‑ L Intermediate of [ NiFe ] -

Hydrogenase. J. Am. Chem. Soc. 145, 13674–13685 (2023).

3. Tai, H., Hirota, S. & Stripp, S. T. Proton Transfer Mechanisms in Bimetallic Hydrogenases. Acc.

Chem. Res. 54, 232–241 (2021).

4. Siebert, E. et al. Resonance raman spectroscopy as a tool to monitor the active site of

hydrogenases. Angew. Chemie - Int. Ed. 52, 5162–5165 (2013).

5. Lauterbach, L. et al. Nuclear resonance vibrational spectroscopy reveals the FeS cluster

composition and active site vibrational properties of an O2-tolerant NAD+-reducing [NiFe]

hydrogenase. Chem. Sci. 6, 1055–1060 (2015).

6. Lauterbach, L., Idris, Z., Vincent, K. A. & Lenz, O. Catalytic properties of the isolated diaphorase

fragment of the NAD +-reducing [NiFe]-hydrogenase from Ralstonia eutropha. PLoS One 6,

(2011).

7. Happe, R. P. et al. Unusual FTIR and EPR properties of the H 2-activating site of the cytoplasmic

NAD-reducing hydrogenase from Ralstonia eutropha.

8. Preissler, J. et al. Enzymatic and spectroscopic properties of a thermostable [NiFe]‑hydrogenase

performing H2-driven NAD+-reduction in the presence of O2. Biochim. Biophys. Acta - Bioenerg.

1859, 8–18 (2018).

9. Lubitz, W., Ogata, H., Rüdiger, O. & Reijerse, E. Hydrogenases. Chem. Rev. 114, 4081–4148

(2014).

10. Goris, T. et al. A unique iron-sulfur cluster is crucial for oxygen tolerance of a [NiFe]-hydrogenase.

Nat. Chem. Biol. 7, 310–318 (2011).

11. Dragelj, J. et al. Conformational and mechanical stability of the isolated large subunit of

membrane-bound [NiFe]-hydrogenase from Cupriavidus necator. Front. Microbiol. 13, 1–15

(2023).

12. Hartmann, S., Frielingsdorf, S., Caserta, G. & Lenz, O. A membrane-bound [NiFe]-hydrogenase

large subunit precursor whose C-terminal extension is not essential for cofactor incorporation

but guarantees optimal maturation. Microbiologyopen 9, 1197–1206 (2020).

13. Caserta, G. et al. Stepwise assembly of the active site of [NiFe]-hydrogenase. Nat. Chem. Biol. 19,

498–506 (2023).

14. Lauterbach, L., Lenz, O. & Vincent, K. A. H2-driven cofactor regeneration with NAD(P)+-reducing

hydrogenases. FEBS J. 280, 3058–3068 (2013).

15. Mertens, R., Greiner, L., Ban, E. C. D. Van Den, Haaker, H. B. C. M. & Liese, A. Practical

applications of hydrogenase I from Pyrococcus furiosus for NADPH generation and regeneration.

25, 39–52 (2003).

148

16. Fritsch, J., Lenz, O. & Friedrich, B. Structure, function and biosynthesis of O2-tolerant

hydrogenases. Nat. Rev. Microbiol. 11, 106–114 (2013).

17. Fritsch, J., Lenz, O. & Friedrich, B. Oxygen-tolerant hydrogenases in hydrogen-based technologies.

Nat. Rev. Microbiol. 22, 358–364 (2011).

18. Young, J., Kim, H. & Cha, H. J. Recent progress in hydrogenase and its biotechnological application

for viable hydrogen technology. Korean J. Chem. Eng. 30, 1–10 (2013).

19. Kulka-Peschke, C. J. et al. Reversible Glutamate Coordination to High-Valent Nickel Protects the

Active Site of a [NiFe] Hydrogenase from Oxygen. J. Am. Chem. Soc. 144, 17022–17032 (2022).

20. Shomura, Y. Structural Insights into the Protective Mechanisms against O2 in the [NiFe]-

Hydrogenases. in Advances in Bioorganometallic Chemistry 365–377 (Elsevier, 2018).

21. Birrell, A. J. et al. Spectroscopic and Computational Evidence that [FeFe] Hydrogenases Operate

Exclusively with CO-Bridged Intermediates. J. Am. Chem. Soc. 142, 222–232 (2019).

22. Caserta, G. et al. Hydroxy-bridged resting states of a [NiFe]-hydrogenase unraveled by cryogenic

vibrational spectroscopy and DFT computations. Chem. Sci. 12, 2189–2197 (2021).

23. T. Stripp, S., Mebs, S. & Haumann, M. Temperature Dependence of Structural Dynamics at the

Catalytic Cofactor of [FeFe]-hydrogenase. Inorg. Chem. 59, 16474–16488 (2020).

24. Tai, H., Nishikawa, K., Inoue, S., Higuchi, Y. & Hirota, S. FT-IR Characterization of the Light-Induced

Ni-L2 and Ni-L3 States of [NiFe] Hydrogenase from Desulfovibrio vulgaris Miyazaki F. J. Phys.

Chem. B 119, 13668–13674 (2015).

25. Ash, P. A., Hidalgo, R. & Vincent, K. A. Proton Transfer in the Catalytic Cycle of [NiFe]

Hydrogenases: Insight from Vibrational Spectroscopy. ACS Catal. 7, 2471–2485 (2017).

26. Ogata, H. et al. Hydride bridge in [NiFe]-hydrogenase observed by nuclear resonance vibrational

spectroscopy. Nat. Commun. 6, 1–8 (2015).

27. Ogata, H., Nishikawa, K. & Lubitz, W. Hydrogens detected by subatomic resolution protein

crystallography in a [NiFe] hydrogenase. Nature 520, 571–574 (2015).

28. Horch, M., Lauterbach, L., Lenz, O., Hildebrandt, P. & Zebger, I. NAD(H)-coupled hydrogen cycling

- Structure-function relationships of bidirectional [NiFe] hydrogenases. FEBS Lett. 586, 545–556

(2012).

29. Horch, M., Hildebrandt, P. & Zebger, I. Concepts in bio-molecular spectroscopy: Vibrational case

studies on metalloenzymes. Phys. Chem. Chem. Phys. 17, 18222–18237 (2015).

30. Delacey, A. L. et al. Spectroscopic and kinetic characterization of active site mutants of

Desulfovibrio fructosovorans Ni-Fe hydrogenase. J Biol Inorg Chem 8, 129–134 (2003).

31. Shomura, Y. et al. Structural basis of the redox switches in the NAD+-reducing soluble [NiFe]-

hydrogenase. Science. 357, 928–932 (2017).

32. Saggu, M. et al. Spectroscopic insights into the oxygen-tolerant membrane-associated [NiFe]

hydrogenase of Ralstonia eutropha H16. J. Biol. Chem. 284, 16264–16276 (2009).

33. Siebert, E. Vibrational Spectroscopy of the Active Site and Iron Sulfur Clusters of the Membrane

Bound Hydrogenase from Ralstonia Eutropha. (2015).

149

34. Katz, S. Spectroscopic studies on hydrogenases. PhD thesis, Univ. der Tech. Univ. Berlin (2019).

35. Hidalgo, R., Ash, P. A., Healy, A. J. & Vincent, K. A. Infrared spectroscopy during electrocatalytic

turnover reveals the Ni-L active site state during H2 oxidation by a NiFe hydrogenase. Angew.

Chemie - Int. Ed. 54, 7110–7113 (2015).

36. Vansuch, G. E. et al. Metal-ligand cooperativity in the soluble hydrogenase-1 from: Pyrococcus

furiosus. Chem. Sci. 11, 8572–8581 (2020).

37. Caserta, G. et al. The large subunit of the regulatory [NiFe]-hydrogenase fromRalstonia eutropha-

a minimal hydrogenase? Chem. Sci. 11, 5453–5465 (2020).

38. Tai, H., Nishikawa, K., Higuchi, Y., Mao, Z. & Hirota, S. Cysteine SH and Glutamate COOH

Contributions to [NiFe] Hydrogenase Proton Transfer Revealed by Highly Sensitive FTIR

Spectroscopy. Angew. Chemie 131, 13419–13424 (2019).

39. Kim, C. U., Tate, M. W. & Gruner, S. M. Protein dynamical transition at 110 K. Proc. Natl. Acad.

Sci. U. S. A. 108, 20897–20901 (2011).

40. Waffo, A. F. T. et al. Supporting information Structural Determinants of the Catalytic Ni a-L

Intermediate of [NiFe]-Hydrogenase.

41. Pandelia, M. E. et al. Membrane-bound hydrogenase i from the hyperthermophilic bacterium

aquifex aeolicus: Enzyme activation, redox intermediates and oxygen tolerance. J. Am. Chem.

Soc. 132, 6991–7004 (2010).

42. Long, M. et al. Characterization of a HoxEFUYH type of [NiFe] hydrogenase from Allochromatium

vinosum and some EPR and IR properties of the hydrogenase module. J. Biol. Inorg. Chem. 12,

62–78 (2007).

43. Erkens, A. et al. The NAD-linked soluble hydrogenase from Alcaligenes eutrophus H16: detection

and characterization of EPR signals deriving from nickel and flavin. JBIC vol. 1 (1996).

44. Lubitz, W., Reijerse, E. & van Gastel, M. [NiFe] and [FeFe] hydrogenases studied by advanced

magnetic resonance techniques. Chem. Rev. 107, 4331–4365 (2007).

45. Horch, M. et al. Probing the active site of an O2-Tolerant NAD +-reducing [NiFe]-Hydrogenase

from Ralstonia eutropha H16 by in situ EPR and FTIR spectroscopy. Angew. Chemie - Int. Ed. 49,

8026–8029 (2010).

46. Lubitz, W., Ogata, H., Ru, O. & Reijerse, E. Hydrogenases. Chem. Rev. 114, 4081–4148 (2014).

47. Hagen, W. R. EPR spectroscopy of complex biological iron–sulfur systems. J. Biol. Inorg. Chem. 23,

623–634 (2018).

48. Bridges, H. R., Bill, E. & Hirst, J. Mössbauer Spectroscopy on Respiratory Complex I: The Iron −

Sulfur Cluster Ensemble in the NADH-Reduced Enzyme Is Partially Oxidized. Biochemistry 51,

149–158 (2012).

49. Greene, B. L., Vansuch, G. E., Wu, C. H., Adams, M. W. W. & Dyer, R. B. Glutamate Gated Proton-

Coupled Electron Transfer Activity of a [NiFe]-Hydrogenase. J. Am. Chem. Soc. 138, 13013–13021

(2016).

50. Greene, B. L., Wu, C. H., McTernan, P. M., Adams, M. W. W. & Dyer, R. B. Proton-Coupled

150

Electron Transfer Dynamics in the Catalytic Mechanism of a [NiFe]-Hydrogenase. J. Am. Chem.

Soc. 137, 4558–4566 (2015).

51. Greene, B. L., Wu, C. H., Vansuch, G. E., Adams, M. W. W. & Dyer, R. B. Proton Inventory and

Dynamics in the Nia-S to Nia-C Transition of a [NiFe] Hydrogenase. Biochemistry 55, 1813–1825

(2016).

52. Caserta, G. et al. In Vitro Assembly as a Tool to Investigate Catalytic Intermediates of [NiFe]-

Hydrogenase. ACS Catal. 10, 13890–13894 (2020).

53. Rippers, Y., Horch, M., Hildebrandt, P., Zebger, I. & Mroginski, M. A. Revealing the absolute

configuration of the CO and CN- ligands at the active site of a [NiFe] hydrogenase.

ChemPhysChem 13, 3852–3856 (2012).

54. Barth, A. The infrared absorption of amino acid side chains. Prog. Biophys. Mol. Biol. 74, 141–173

(2000).

55. Barth, A. Infrared spectroscopy of proteins. Biochim. Biophys. Acta - Bioenerg. 1767, 1073–1101

(2007).

56. Fritsch, J. et al. The crystal structure of an oxygen-tolerant hydrogenase uncovers a novel iron-

sulphur centre. Nature 479, 249–253 (2011).

57. Frielingsdorf, S. et al. Reversible [4Fe-3S] cluster morphing in an O 2 -tolerant [NiFe]

hydrogenase. Nat. Chem. Biol. 10, 378–385 (2014).

58. Siebert, E. et al. Resonance Raman spectroscopic analysis of the iron–sulfur cluster redox chain of

the Ralstonia eutropha membrane-bound [NiFe]-hydrogenase. J. Raman Spectrosc. 52, 2621–

2632 (2021).

59. Lenz, O. et al. H 2 Conversion in the Presence of O 2 as Performed by the Membrane-Bound [

NiFe ] -Hydrogenase of Ralstonia eutropha. ChemPhysChem 11, 1107–1119 (2010).

60. Ludwig, M. et al. Concerted action of two novel auxiliary proteins in assembly of the active site in

a membrane-bound [NiFe] hydrogenase. J. Biol. Chem. 284, 2159–2168 (2009).

61. Rippers, Y., Utesch, T., Hildebrandt, P. & Zebger, I. Insights into the structure of the active site of

the O 2 -tolerant membrane bound [ NiFe ] hydrogenase of R . eutropha H16 by molecular

modelling w. Phys. Chem. Chem. Phys. 12, 16146–16149 (2011).

151

5. Outlook and final remarks

Characteriza�on of the fully mature large subunit of a [NiFe] hydrogenase in solu�on

The fully mature large subunit HoxG of the membrane-bound hydrogenase from Cupriavidus necator exists

in a mixture of EPR silent Nir-S res�ng states. Although the sample is cataly�cally inac�ve, it appears redox

ac�ve. Upon chemical oxida�on with K3Fe(CN)6, the Nir-B species, with a NiII-OH-FeII, ac�ve site

conﬁgura�on is formed, while chemical reduc�on with dithionite induces some traces of Nia-L-like species

with a NiI-FeII site. These two Nir-S species are in pH-dependent equilibrium but their ra�o remains

unaﬀected by changes in temperature. The spectroscopic studies were accompanied by computa�onal

studies from the group of Professor Mroginski. Based on the concomitant computa�ons we suggest three

possible couples of Nir-S species conﬁgura�ons. The Nir-S states most presumably diﬀer by one proton.

However, the most important ﬁnding of this combined study was the role of H82, a his�dine residue which

is not present in other large subunits, like those of the Regulatory Hydrogenase from Cupriavidus necator

(CnRH), HoxC. H82 was always found single or double protonated and in some models it even forms a

hydrogen bond with C600. Another ﬁnding was that the E32, a glutamate residue, seems to form a

hydrogen bond with one of the terminal cysteines of the ac�ve site and always remains protonated.

Size exclusion chromatography data clearly show that the large subunit HoxG dimerizes at high sample

concentra�ons and IR data show that this contributes to the long-term [NiFe] cofactor stability. This is an

indica�on that the dimeriza�on takes place in the former HoxG-HoxK interface. This possibility is further

supported by the size exclusion chromatography on pre-HoxG, a HoxG molecule in which the C-terminus

is not yet cleaved and dimeriza�on does not occur. Computa�onal studies conﬁrmed that the dimeriza�on

between two HoxG molecules in the former large-small subunit interface restored the ac�ve site rigidity.

It is not completely clear why HoxG is cataly�cally inac�ve. It would be interes�ng though to atempt to

measure cataly�c ac�vity on the immobilized HoxG applying reduc�ve poten�al, where the electrode can

presumably replace the [FeS] clusters and pump the electrons to the ac�ve site.

First immobiliza�on and electrochemical control of an isolated large hydrogenase subunit

Herein, for the ﬁrst �me, HoxG was immobilized and its redox structural states were electrochemically

controlled. This was successfully accomplished on top of a slightly nega�vely charged C7COOH SAM at pH

5.5. Some of the immobilized protein molecules were successfully oxidized and reversibly reduced. Based

on the computa�onal results the [NiFe] cofactor to electrode distance for these molecules should be 14

Å. However, the electrosta�c immobiliza�on results in a distribu�on of molecular orienta�ons, poten�ally

152

resul�ng in a distribu�on of electron transfer rates. Especially in the case of the isolated hydrogenase

subunit HoxG, devoid of the corresponding small counterpart, a certain distribu�on of orienta�ons that

enlarges the distance of the [NiFe] site to electrode surface, combined with the insula�on proper�es of

the protein matrix may result in a hampered electron transfer from the electrode to the protein’s

cofactors. Further, this study highlights the importance of considering not only the protona�on degree of

the SAM but also the ionic strength of the buﬀer and the role of the monovalent and divalent ions, which

based on the data seem to be more complicated than expected.

With the covalent immobiliza�on approach HoxG a higher amount of immobilized protein was successfully

electrochemically controlled but on the other hand, this interac�on resulted in a gradual cofactor loss. As

shown from the studies in solu�on, a loss of the [NiFe] cofactor was preferen�ally observed in the case of

the monomeric HoxG. The reason could presumably be the solvent accessibility of the ‘open’ ac�ve site

pocket (devoid of the small subunit). This factor in combina�on with shearing forces, especially when a

poten�al is applied a�er covalent immobiliza�on, presumably enhanced the loss of the cofactor. Regarding

the molecules, which could not be controlled electrochemically, one has to consider that the ability of the

molecules to re-orient on the surface and their ﬂexibility (even at intramolecular level) is very important

for a successful electron tunneling, which is restricted in the case of the covalently immobilized protein.

As reported in the corresponding chapter, the C7COOH SAM was modiﬁed for the covalent immobiliza�on

with N-Ethyl-Nʹ-(3-dimethylaminopropyl)-carbodiimid–hydrochlorid (EDC) and N-Hydroxysuccinimid

(NHS). With this method, primary amines and mainly the lysines are targeted. Although plenty of lysine

molecules are found mainly at the former HoxGK interface, some are further distributed across the protein

surface. Therefore, it could be interes�ng to induce a poly-lysine tag and atempt to immobilize the protein

through it. In this way one can eliminate the distribu�on of diﬀerent orienta�ons and the protein will

retain some of its ﬂexibility allowing intramolecular movements, which are a prerequisite for the op�mal

electron tunneling. This approach could be used for the HoxG/HoxK recons�tu�on on the electrode.

Covalent surface atachment of a membrane bound [NiFe] hydrogenase – results in H2 produc�on

The aforemen�oned protocol of covalent protein immobiliza�on was also applied on the heterodimer His-

Tagged Cupriavidus necator Membrane Bound Hydrogenase (CnMBH). Notably, herein currents related to

H2 produc�on were observed. Such cataly�c currents are uncommon for CnMBH, which is normally biased

towards H2 oxida�on and H2 produc�on is substrate inhibited. This observa�on is atributed to the

orienta�on of the protein on top of the surface.

153

Temperature equilibrium between the reduced Nia-C and Nia-SR species

Both, for the HtSH and CnMBH, Nia-C was found to exist in a temperature dependent equilibrium with a

Nia-SR species, hereby the later becomes enriched at cryogenic temperatures. Apart from this similarity

we observe some diﬀerences. Firstly, in HtSH the Nia-C species is transformed to Nia-SR’’, whereas in

CnMBH it is in equilibrium with Nia-SR’. Secondly, in CnMBH the (rela�ve) ra�o between the species is not

changing below 236 K, while in HtSH such behavior is observed only below 200 K. In both cases, however,

a transloca�on of electrons, presumably from the proximal cluster to the ac�ve site during the Nia-C to

Nia-SR conversion seems to take place. In addi�on, these ﬁndings highlight the importance of inves�ga�ng

temperature dependent equilibria of redox structural states with IR spectroscopy. Otherwise it is not

possible to compare the IR data to those recorded by diﬀerent methods at cryogenic temperatures.

Deciphering the Enriched State of CnMBH upon anaerobic reoxida�on: Unveiling the Puzzle of Ac�vity

or Readiness

In this thesis we succeeded, for the ﬁrst �me, in enriching a redox structural state with vCO and vCN

stretching vibra�ons at 1936 cm–1 and 2075, 2093 cm–1. O�en in literature this state is denoted as Nia/r-S,

since it is not clear whether this EPR silent state appears to be ac�ve or ready (res�ng). The recent

enrichment, yielding an almost pure state of the ac�ve site, can poten�ally lead to the crystalliza�on of

this species and characteriza�on of the ac�ve site geometry with X-ray crystallography. Furthermore, the

lack of a hydride in the bridging posi�on between the Ni and the Fe could in future be conﬁrmed by NRVS

measurements, in combina�on with DFT calcula�ons.

Light-induced electron transfer opens a photochemical shortcut for cataly�c dihydrogen cleavage

For the ﬁrst �me, a fully reduced cataly�c intermediate of a [NiFe] hydrogenase has been converted

through light exposure. The fully reduced state with a hydride in the bridging posi�on, Nia-SR’’ converts to

a Nia-L state with a free coordina�on site between the two metals. This process, presupposes a net

electron transfer from the ac�ve site towards [FeS] clusters of the enzyme. Notably, this was not observed

for other Nia-SR substates of HtSH. We propose that this is due to a lack of charge compensa�on at the

[NiFe] site in this state since the proton derived from H2 cleavage is likely located in considerable distance

from the cofactor in the Nia-SR’’ form, which features the lowest CO stretching frequency than in all the

other Nia-SR sub-forms. The light-driven transforma�on Nia-SR’’ → Nia-L serves as a photochemical

shortcut in the cataly�c cycle bypassing the Nia-C intermediate. Combined with the light-driven electron

transfer described earlier, this ﬁnding induces new possibili�es for manipula�ng hydrogenases using light.

154

SH stretching region beyond the ﬁrst coordina�on sphere: Traps and Challenges

Closely adjacent nega�ve and posi�ve peaks emerge in the SH stretching region when the CnMBH sample

is exposed to light. This pair of signals in the diﬀerence spectrum, along with the peak's intensity, seems

to be an unusual feature, which has not yet been observed in other hydrogenases. This peculiarity cannot

be atributed solely to the protona�on of the terminal cysteine. Given its pronounced intensity, it is

conceivable that mul�ple cysteine residues which are poten�ally linked by hydrogen bonds, contribute to

these strong peaks. Notably, the following correla�on has been no�ced: when the ac�ve site's vCO

stretching vibra�on is blue-shi�ed, the SH stretching peak is red-shi�ed, and vice versa, sugges�ng that

this prominent peak is associated with a cysteine near the ac�ve site, which is involved in electronic

interac�on.

When the Cys81 was subs�tuted by an isosteric serine residue, yielding the C81S variant, the diﬀerence IR

spectrum (light minus dark) was devoid of the nega�ve/posi�ve signal pair in the SH stretching region.

Serine, which is shorter than cysteine, is likely unable to form hydrogen bonds, which may account for the

observed shi� of the main CO stretching peak in the Nia-C and Nia-L states of the C81S variant.

Consequently, the observed pair of peaks in the SH stretching region can be atributed to C81, which is a

unique feature of CnMBH, and is located in the second coordina�on sphere of the ac�ve site. These

ﬁndings underscore the necessity for a cau�ous interpreta�on of peaks in this region and suggest that C81

may be involved in an alterna�ve pathway for H+ transfer.

155

6. Appendix

6.1 Appendix to Chapter 3

Figure A.3.1: 2nd derivative of the absorbance spectra vCO/vCN active site region, A: Without glycerol and B: with

30 % glycerol, in KiPO4 buffer, pH 7 at 10 0C.

Figure A.3.2: pKa titration of a (CH2)8NH2-SAM coated onto Au electrodes. The surface pKa value was determined by

the peak current (ip) of a given redox couple, following the protocol described in Degefa et. al.16

156

Figure A.3.3: pKa titration of C7COOH-SAM coated on Au electrodes. The surface pKa value was determined by the

peak current (ip) of a given redox couple, following the protocol described in Degefa et. al.16

Figure A.3.4: Left, effect of the monovalent ion at the elevated concentration for the -NH

terminated indicating

denaturation of the protein which is not exclusively related to the salt concentration, but it is probably combined with

strong surface tension. On the right site and on the inlet of the figure the amide region is depicted and on the right side

strong

(OH) negative absorptions which are related to the removal of water molecules from the surface and their

replacement with protein molecules.

Alternatively the changes in the amide region

could also indicate reorientation of

the protein molecules.

Right, shows the 2nd

derivative of the absorbance spectrum recorded in the transmission mode. In

the figure, the

vCO/vCN stretching vibration region is depicted. Here, 300 μΜ HoxG was diluted with 50 mM KiPO4

, 2 M

NaCl, pH 5.5. A spectrum was measured after 15 and 180 min. With this experiment it is shown that the high salt

concentration has no effect on the active site o

f the protein when the latter is not immobilized.

157

Figure A.3.5: The figure depicts the 2nd derivative of immobilized proteins on top of a COOH:OH SAM, mixed in a

1:9 ratio, indicating that details about the secondary structure can be revealed by SEIRAS.

158

Figure Α.3.6: A: The amide and active site region of the spectrum is depicted. B: The panel displays the same spectra

in an enlarged view in the CO and CN stretching region. C: The difference spectra of the oxidized versus reduced

protein and vice versa. The first trace of each figure in the right column depicts the difference spectrum 150 minus

–450 mV vs Ag/AgCl and the second trace the related data when subsequently –450 mV minus 150 mV vs Ag/AgCl

are applied. The intensity of the peaks in the difference spectrum refers to the amount of protein which is in good

contact with the electrode. D: Strep-tagged MBH, covalently bound, under argon (black trace) and under hydrogen

(red trace).

159

6.2 Appendix to Chapter 4

Appendix to 4.1 a

HtSH

Figure A.4.1: a: Re-oxidation of the HtSH. difference spectrum, depicting the 2150-1000 cm–1 spectral range. In this

region the CO/CN stretching vibrations of the active site (1900-2150 cm–1), conformational changes (Amide II at

around 1550 cm–1 ) and (de)protonation events of single aminoacids in the 2nd coordination sphere, can be

concominantly detected. b: top, IR spectrum, vCO/vCN active site region, of the reduced (black trace) and the re-

oxidised enzyme (red trace), bottom (blue trace), the difference spectrum “oxidized minus reduced”. The spectrum

of the re-oxidized sample was recorded 5 days after the sample’s reduction.

160

4.5 C81S variant of CnMBH

Figure A.4.2: Difference spectrum, light minus dark of the wild type CnMBH, a: 90 K, b: after warming the cell up to

155 K, and c: after cooling down again to 90 K. The inset depicts the corresponding SH stretching region. At 155 K,

an enrichment of Nia-L3 species deriving from the Nia-L2 state. Subsequently, this redox state equilibrium is shock

frozen to 90 K. From this measurement the temperature-dependent band intensity of the band is revealed.

161

Figure A.4.3: IR data of the re-oxidized C81S variant. Left: difference spectrum relative to its reduced species, Right:

2nd derivative of the underlying absorbance spectra of the reduced (grey trace) and re-oxidized form (red trace).

In Fig. A.4.3, on the left side the difference spectra re-oxidized minus reduced are depicted. The IR spectra

on the right side depict the 2nd derivative of the H2-reduced (red trace), C81S sample and the reoxidized

C81S sample (grey trace).

Figure A.4.4: Illumination of the C81S variant at 90 K and 170 K, exhibiting the formation of the 1880 cm–1 band

assigned to a Nia-L3 in the corresponding light minus dark difference spectrum. It is not clear if the parent states are

exclusively one or both Nia-C species.

162

Band position of the observed redox structural state of the oxygen tolerant [NiFe]

hydrogenases studied in this thesis.

Table 4.A.1: HtSH

Redox

species

v(CO)

(cm

–1

)

v(CN)

(cm

–1

)

v(CN)

(cm

–1

)

Reference

Ni(IV)r–Hex

1993-6

2081-2

2090

Ni(III)r-Hex

1963-4

2087

2098

Ni(III)r-Hex’

1971-3

*2083

*2103

*This work

Nir-S

1936

2058

2071

Nia-S

1951

2076

2089

Nia-C

1971-3

*2080

*2092

*This work

Nia-SR

1958

2062

2076

Nia-SR’

1943

2048

2062

Nia-SR’’

1936

*2050

*2067

*This work/8

Nia-L1

1922

2047

2062

Nia-L2

1924

2053

2066

Niia-S ? 1959-60 This work+ Dr.candidate

Charlotte Wiemann

In this study the 1959-60 cm–1 band was significantly enriched during the slow re-oxidation of the protein

inside the thin layer cell. However, it was also reported (personal notes by Charlotte Wiemann) that these

species appear in all homologues samples after 4-10 hr. Therefore, we assume that it might be a Niia-S

species.

Table 4.A.2: CnMBH wildtype

Redox

species

v(CO)

(cm

–1

)

v(CN)

(cm

–1

)

v(CN)

(cm

–1

)

Reference

Nir-B

1948

2081

2098

Nia/r-S

1936

2075

2093

32/ This work

Nia-C

1957

2075

2097

Nia-SR

1945

2068

2087

Nia-SR’

1925

2049

2071

Nia-SR’’

1919

2046

2071

Niia-S

1930

2060

2076

Niia-S**

1930

2049

2066

Nia-L2

1909

2045

2079

This work

Nia-L2’

1910

2040

2077

This work

Nia-L3

1881

2028

2072

This work

Nia-L3’

1886

This work

163

Table 4.A.3: C81S variant of CnMBH

Redox

species

v(CO)

(cm–1)

v(CN)1

(cm–1)

v(CN)2

(cm–1) Reference

Niu/r-S

1943

2081

2105

Nir-B

1955-7

2080

2097

Nia/r-S

1929

2052

2066

This work

Nia-C

1967

2073

2103

This work

Nia-C’

1954

2103

This work

Nia-SR

1950

Nia-SR’

1930

2049

2072

Nia-SR’’

1924

2047

2070

Nia-L

1916

2040

2083

This work

Nia-L’

1903

2045

2077

This work

Nia-L3

1880

2026

2052

This work

The vCN peaks of Nia-C (a) and Nia-C’ (b) in Table 4.A.3 above, are overlapping. Therefore the precise

assignment is not possible. The Nia-C species of the C81S, MBH variant was first detected by Saggu et.al..

32 Here the CO/CN stretching vibration peaks slightly differ to those determined in literature (1967/2073,

2103 cm–1 versus 1962/2072, 2084 cm–1). Here the values are estimated by the difference spectra upon

illumination at cryogenic temperatures (below 110 K), when Nia-C is the main species photoconverting.

Opposite in Saggu et.al. the peaks are determined from the absorbance spectra when several peaks

coexist (mainly Nia-C and Nia-SR’’). In that work the Nia-C’ was not detected. 32