rights / license: research collection in copyright - non ...6/eth-6-02.pdf · strutturali di un...

Research Collection

Doctoral Thesis

EPR studies of copper coordination to DNA and of supercooledwater dynamics

Author(s): Santangelo, Maria Grazia

Publication Date: 2009

Permanent Link: https://doi.org/10.3929/ethz-a-005843060

Rights / License: In Copyright - Non-Commercial Use Permitted

This page was generated automatically upon download from the ETH Zurich Research Collection. For moreinformation please consult the Terms of use.

ETH Library

https://doi.org/10.3929/ethz-a-005843060

http://rightsstatements.org/page/InC-NC/1.0/

https://www.research-collection.ethz.ch

https://www.research-collection.ethz.ch/terms-of-use

Diss. ETH No. 18278

EPR studies of coppercoordination to DNA and ofsupercooled water dynamics

A dissertation submitted to the

SWISS FEDERAL INSTITUTE OF TECHNOLOGYZURICH

for the degree ofDoctor of Sciences

presented by

MARIA GRAZIA SANTANGELO

Laurea in Fisica, University of Palermo (Italy)born July 5, 1976citizen of Italy

accepted on the recommendation of

Prof. Dr. G. Jeschke, examinerProf. Dr. R. Riek, co-examinerDr. B. Spingler, co-examiner

2009

to Matteo

Contents

Abstract v

Sommario vii

1 General introduction 11.1 Electron paramagnetic resonance . . . . . . . . . . . . . . . . . . 11.2 Spin Hamiltonian . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

1.2.1 Electron Zeeman interaction . . . . . . . . . . . . . . . . . 31.2.2 Crystal-field splitting and spin-orbit coupling . . . . . . . 41.2.3 g anisotropy . . . . . . . . . . . . . . . . . . . . . . . . . . 61.2.4 Nuclear Zeeman interaction . . . . . . . . . . . . . . . . . 71.2.5 Hyperfine interaction . . . . . . . . . . . . . . . . . . . . . 81.2.6 Nuclear quadrupole interaction . . . . . . . . . . . . . . . 10

1.3 Continuous wave EPR . . . . . . . . . . . . . . . . . . . . . . . . 111.3.1 Molecular and chemical dynamics . . . . . . . . . . . . . . 111.3.2 Geometric and electronic structure . . . . . . . . . . . . . 15

1.4 Pulse EPR . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 161.4.1 MW pulse and spin dynamics . . . . . . . . . . . . . . . . 171.4.2 Relaxation time and linewidth . . . . . . . . . . . . . . . . 181.4.3 Electron spin echo . . . . . . . . . . . . . . . . . . . . . . 191.4.4 Origin of the nuclear modulation effect . . . . . . . . . . . 201.4.5 Orientation selection in pulse EPR . . . . . . . . . . . . . 241.4.6 Two-pulse ESEEM . . . . . . . . . . . . . . . . . . . . . . 251.4.7 Three-pulse ESEEM . . . . . . . . . . . . . . . . . . . . . 261.4.8 Hyperfine sublevel correlation experiment . . . . . . . . . . 271.4.9 Electron nuclear double resonance . . . . . . . . . . . . . . 30

2 EPR investigation of a DNA model system 332.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 332.2 Materials and methods . . . . . . . . . . . . . . . . . . . . . . . . 33

2.2.1 Sample preparation . . . . . . . . . . . . . . . . . . . . . . 332.2.2 Spectroscopy . . . . . . . . . . . . . . . . . . . . . . . . . 342.2.3 Data manipulation . . . . . . . . . . . . . . . . . . . . . . 35

2.3 Results and discussion . . . . . . . . . . . . . . . . . . . . . . . . 352.3.1 CW EPR spectra . . . . . . . . . . . . . . . . . . . . . . . 35

iii

2.3.2 14N Davies ENDOR . . . . . . . . . . . . . . . . . . . . . 362.3.3 1H HYSCORE . . . . . . . . . . . . . . . . . . . . . . . . 382.3.4 14N, 2H HYSCORE . . . . . . . . . . . . . . . . . . . . . . 402.3.5 13C HYSCORE . . . . . . . . . . . . . . . . . . . . . . . . 422.3.6 31P Mims ENDOR . . . . . . . . . . . . . . . . . . . . . . 432.3.7 Structural considerations . . . . . . . . . . . . . . . . . . . 44

2.4 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46

3 EPR investigation of copper-poly(dG-dC) interaction 473.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 473.2 DNA structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . 473.3 Materials and methods . . . . . . . . . . . . . . . . . . . . . . . . 49

3.3.1 Sample preparation . . . . . . . . . . . . . . . . . . . . . . 493.3.2 Spectroscopy and data manipulation . . . . . . . . . . . . 49

3.4 Results and discussion . . . . . . . . . . . . . . . . . . . . . . . . 503.4.1 Evidence for two copper species . . . . . . . . . . . . . . . 503.4.2 Properties in common to both copper species . . . . . . . 513.4.3 Assignment of the copper species . . . . . . . . . . . . . . 563.4.4 Cu(II)-(1-methyl-cytosine)4 model system . . . . . . . . . 583.4.5 Properties of the copper species I . . . . . . . . . . . . . . 593.4.6 Properties of the copper species II . . . . . . . . . . . . . . 603.4.7 Structural considerations and conclusions . . . . . . . . . . 633.4.8 Outlook . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64

4 Supercooled water in a silica hydrogel 654.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 654.2 Sol-Gel encapsulation . . . . . . . . . . . . . . . . . . . . . . . . . 664.3 Materials and methods . . . . . . . . . . . . . . . . . . . . . . . . 68

4.3.1 Silica hydrogel for EPR measurements . . . . . . . . . . . 684.3.2 Silica hydrogel for DSC measurements . . . . . . . . . . . 694.3.3 EPR spectroscopy . . . . . . . . . . . . . . . . . . . . . . . 694.3.4 DSC spectroscopy . . . . . . . . . . . . . . . . . . . . . . . 694.3.5 Theory and analysis of EPR data . . . . . . . . . . . . . . 704.3.6 Theory and DSC data analysis . . . . . . . . . . . . . . . . 73

4.4 Results and discussion . . . . . . . . . . . . . . . . . . . . . . . . 744.5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82

5 Summary and Outlook 85

Bibliography 87

Acknowledgements 95

Publications 97

Curriculum Vitae 99

iv

Abstract

In the first part of this thesis pulse EPR spectroscopy has been used to de-termine electronic and geometric structure of a mononucleotide model systemCu(II)-guanosine 5′-monophosphate, and to develop an approach for more de-tailed structural characterization of specifically bound metal ions in a variety ofnucleic acids of biological interest. In the second part this approach is extendedto a study of the copper-polynucleotide interaction. Results of a pulse EPRstudy on Cu(II)-poly(dG-dC)·poly(dG-dC) complex are presented. In the lastpart, continuous wave EPR spectroscopy together with the spin probe techniquehas been used to characterize the dynamics of supercooled water encapsulated insilica hydrogel. Such an aqueous matrix in a supercooled or glassy state allowsfor low-temperature studies without the use of cryoprotectants.

Simple copper salts are known to denature poly(dG-dC), whereas copper inthe form of an azamacrocyclic complex is able to convert the right-handed B formof the same DNA sequence to the corresponding left-handed Z form. Pulse EPRmethods provide unique information about the electronic and geometric structureof the Cu(II)-5′-GMP model system through a detailed mapping of the hyperfineinteractions between the unpaired electron of the Cu(II) ion and the magneticnuclei of the nucleotide ligand, and nuclear quadrupole interactions. It was foundthat the Cu(II) ion is directly bound to N7 of 5′-guanosine monophosphate andindirectly bound via a water of hydration to a phosphate group.

Pulse EPR spectroscopy has been used to investigate the structural features ofa copper ion coordinated to poly(dG-dC)·poly(dG-dC) and to predict the struc-tural changes of the polynucleotide induced by the presence of the copper ion. Itwas found that the Cu(II)-poly(dG-dC)·poly(dG-dC) complex presents a Z-formand is characterized by two different copper species. A molecular model has beenproposed to assign them. In this model one copper species is exclusively coordi-nated to the N7 of the guanine. In the other species, both guanine and cytosineare involved in the formation of the complex Cu(II)-poly(dG-dC)·poly(dG-dC).A copper crosslink between the N7 of guanine and the N3 of cytosine as the mostprobable coordination site is proposed. Moreover, no evidence was found for in-teraction of the copper species with the phosphate group and with equatorialwater.

v

EPR spectroscopy on a nitroxide probe, in combination with differential scan-ning calorimetry (DSC), was used to characterize the structural and dynamicproperties of water confined in silica hydrogels. In these silica hydrogels the ma-jor fraction of water remains in a supercooled state down to temperatures of atleast 198 K, whereas the minor fraction crystallizes at about 236 K during thecooling and melts at 246 K during heating. The liquid domains are of sufficientsize to solvate the nearly spherical paramagnetic probe molecule TEMPO witha diameter of about 6 A. Analysis of EPR spectra provides the rotational cor-relation time of the probe that is further used to compare the viscosity of thesupercooled water with the one of bulk water. In the temperature interval inves-tigated the supercooled water behaves as a fragile liquid and eventually solidifiesat 120 K to a glass that incorporates the probe molecules.

vi

Sommario

Nella prima parte di questa tesi la spettroscopia di EPR pulsato e stata usata perdeterminare la struttura elettronica e geometrica di un sistema modello mononu-cleotide Cu(II)-guanosina 5′-monofosfato e per sviluppare un approccio per unapiu dettagliata caratterizzazione strutturale di ioni metallici in una varieta diacidi nucleici di interesse biologico. Nella seconda parte questo approccio eesteso allo studio dell’interazione rame-polinucleotide. I risultati dello studiotramite EPR pulsato del sistema Cu(II)-poly(dG-dC)·poly(dG-dC) sono presen-tati. Nell’ultima parte, la spettroscopia EPR in onda continua insieme con latecnica della sonda di spin e stata usata per caratterizzare la dinamica dell’acquasottoraffreddata incapsulata in gel di silice. Questa matrice acquosa, che si trovain uno stato di sottoraffreddamento o in uno stato vetroso, permette studi a bassetemperature senza l’uso di crioprotettori.

E’ noto che semplici sali di rame sono capaci di denaturare polimeri di deos-siguanosina e deossicitidina, mentre complessi di rame sono in grado di convertirequesti stessi polimeri dallla forma destrorsa del B-DNA alla corrispondente formasinistrorsa del Z-DNA. Metodi di EPR pulsato forniscono informazioni sulla strut-tura elettronica e geometrica del sistema modello Cu(II)-5′-GMP attraverso unaindagine dettagliata delle interazioni iperfine tra l’elettrone spaiato del Cu(II) e inuclei magnetici del nucleotide, e dell’interazione quadrupolare nucleare. E’ statoscoperto che lo ione Cu(II) e direttamente legato all’azoto N7 della guanosina 5′-monofosfato e indirettamente legato, attraverso una molecola d’acqua, al gruppofosfatico.

La spettroscopia di EPR pulsato e stata usata per studiare le caratteristichestrutturali di un ione di rame legato al polinucleotide poly(dG-dC)·poly(dG-dC)e determinare i cambiamenti strutturali del polinucleotide indotti dalla presenzadello ione di rame. E’ stato scoperto che il sistema Cu(II)-poly(dG-dC)·poly(dG-dC) ha una forma sinistrorsa come lo Z-DNA ed e caratterizzato a due differentispecie di rame. Noi abbiamo proposto un modello molecolare per distinguerequeste due specie. In questo modello, una delle due specie e legata esclusiva-mente all’azoto N7 della guanina. Nell’altra specie, entrambe la guanina e lacitosina sono coinvolte nella formazione del sistema Cu(II)-poly(dG-dC)·poly(dG-dC). Noi proponiamo un collegamento tra l’azoto N7 della guanina e l’azoto N3della citosina tramite lo ione di rame. Inoltre non e stata ottenuta nessuna evi-

vii

denza di coordinazione tra il rame e il gruppo fosfatico e tra il rame e l’acqua inposizione equatoriale.

La spettroscopia EPR applicata ad una sonda nitrossile, insieme con la calori-metria a scansione differenziale, e stata usata per caratterizzare le proprieta strut-turali e dinamiche dell’acqua confinata in un gel di silice. In questi gel di silicela maggior parte dell’acqua viene sottoraffreddata fino ad una temperatura di198 K, mentre la minor parte cristallizza a 236 K durante il raffreddamento einvece fonde a 246 K durante il riscaldamento. La larghezza dei pori e sufficienteda permettere lo scioglimento in acqua della TEMPO, una sonda paramagnet-ica quasi sferica, che presenta un diametro di circa 6 A. L’analisi degli spettriEPR ci forniscono il tempo di rotazione della sonda, il quale e a sua volta us-ato per confrontare la viscosita dell’acqua sottoraffreddata con quella dell’acquapura. Nell’intervallo di temperature analizzato, l’acqua sottoraffreddata si com-porta come un liquido fragile e infine a 120 K diventa un vetro con incorporatele molecole della sonda.

viii

Chapter 1

General introduction

1.1 Electron paramagnetic resonance

Electron paramagnetic resonance (EPR) is a spectroscopic method for determin-ing the structure and dynamics and the spatial distribution of species with atleast one unpaired electron (paramagnetic species). The unpaired electrons leadto a non-vanishing spin of a particle that can be used as a spectroscopic probe.It turns out that transitions between electron spin states can be induced byon-resonant electromagnetic radiation which is chemically non-destructive. Themeasurement of electron spin transition frequencies and the evaluation of struc-tural information from them is the main purpose of EPR [Schweiger and Jeschke,2001].

Continuous wave (cw) and pulse EPR spectroscopy can provide unique in-formation on the electronic structure since the magnetic parameters are relatedto the electronic wavefunction and the configuration of the surrounding nucleiwith non zero-spin. The g values and, for species with several unpaired elec-trons, the zero-field splitting often provide fingerprint information on the typeof paramagnetic species. Hyperfine couplings characterize the bonding situationin more detail, and the exchange coupling between electron spins is related tothe delocalization and overlap of the singly occupied molecular orbitals of a pairof electron spins. In many cases, electronic structure can also be interpreted interms of geometric structure. Hyperfine couplings give access to distances be-tween the nuclei and the unpaired electron up to about 0.5 nm, while couplingsbetween electron spins provide geometric information up to distances of at least 5nm [Eaton et al., 2000, Jeschke, 2002, Jeschke and Polyhach, 2007]. The nuclearquadrupole interactions provide information on the bonding of nuclei. CW andpulse EPR spectroscopy are thus able to characterize the structure of systemslacking long-range order on length scales which are not easily accessible by othertechniques. Moreover, with a frequency range up to several hundred gigahertz(S-band: 3 GHz; X-band: 9.75 GHz; Q-band: ∼35 GHz; W-band: ∼95 GHz),EPR spectroscopy can also access molecular and chemical dynamics down to thenanoscale timescale.

1

2 Chap. 1 - General introduction

A rigorous theory of EPR must be based on quantum mechanics, becausequantum objects are involved on a microscopic level. However, experiments areusually done on ensembles of electron spins1. The macroscopic magnetizationof the sample due to these spins is created and manipulated by applying staticand time dependent magnetic fields. Furthermore, the behaviour of quantumobjects is governed by a Hamilton operator (Hamiltonian). At this point it isimportant to stress that the electron spin is almost always coupled to some extentto the orbital momentum of the electron. Nevertheless, the coupling of orbitalmomenta always results in another angular momentum, and thus does not leadto qualitative changes. It is then always possible to introduce an effective spin Swithout any changes in the theory (more details will be given later). The staticHamiltonian that describes the energies of states within the ground state of aparamagnetic species with an effective electron spin S and m nuclei with spins Iis discussed in the following paragraphs.

1.2 Spin Hamiltonian

Abragam and Pryce [Abragam and Pryce, 1951] developed the concept of the spinHamiltonian. Using a perturbation approach, it can be shown that the energies ofstates within the ground state of a paramagnetic species with an effective electronspin S and m nuclei with spin I are described by

H = HEZ +HZFS +HHF +HNZ +HNQ +HNN (1.1)

The terms in Eq. 1.1, refer to the following interactions: Electron Zeemaninteraction (HEZ), Zero-Field Splitting (HZFS), Hyperfine couplings between theelectron spin and the m nuclear spins (HHF ), Nuclear Zeeman interaction (HNZ),Nuclear Quadrupole interactions for spins with nuclear spin quantum numbers I>1/2 (HNQ), and spin-spin interactions between pairs of nuclear spins (HNN).The spin Hamiltonian contains only spin coordinates described by the electronspin vector operator S and the nuclear spin operator I. The effective spin, S, isdefined by the number (2S+1) of the low-lying energy levels which are responsiblefor the EPR spectrum and which are split in a magnetic field. The electronZeeman and nuclear Zeeman interactions are field dependent, hence the possibilityto perform EPR at different magnetic fields will be a powerful tool to separatedifferent interactions from each other. For the calculation of the eigenvalues andeigenstates of a Hamiltonian it is important to know the magnitude of the termsin Eq. 1.1. This allows one to decide whether the relevant information aboutthe system can be obtained by cw or pulse EPR. In the following we will limitour discussion to those terms of the Hamiltonian that are relevant in this thesis,namely HEZ , HHF , HNZ , and HNQ.

1An ensemble of spin systems, where the electron and nuclear spins in each single systemexperience the same time-averaged local fields, is called spin packet.

1.2 - Spin Hamiltonian 3

1.2.1 Electron Zeeman interaction

In the absence of any magnetic field the magnetic moment associated with theelectron spin S is randomly oriented and, for S = 1/2, the two energy levels aredegenerate. The application of an external magnetic field B0 results in a splittingof the two energy levels as the electron spin can only be oriented parallel or anti-parallel to the magnetic field vector. The quantization of the energy levels is dueto the quantum-mechanical nature of the electron spin. The splitting betweenthe two energy states is called the electron Zeeman interaction (EZI) and isproportional to the magnitude of B0, as shown in Figure 1.1

Figure 1.1: Illustration of the electron Zeeman splitting for an S = 1/2 system inan external magnetic field. Transitions between energy levels by microwave irradiationare shown.

The potential energy of this system is derived from the classical expressionfor the energy of a magnetic dipole in a magnetic field and is described by theelectron Zeeman term

HEZ =βeB0gS

~(1.2)

where βe denotes the Bohr magneton, ~/2π is the Planck constant, and B0 is thetranspose of the magnetic field vector B0. The g tensor2 contains the orientationdependence of the electron Zeeman splitting. The energies are then given by theeigenvalues of HEZ in Eq. 1.2

E± = ±1

2gβeB0 (1.3)

with the effective g value.

2Strictly speaking, g does not have the transformation properties of a tensor, as it connectstwo indipendent spaces (spin space and laboratory space). Therefore it is sometimes called aninteraction matrix.


In cw EPR, a transition is usually measured at fixed frequency and the mag-netic field is varied until resonance is reached; therefore

(E+ − E−) = hν = gβeB0 (1.4)

The spin can then flip from one orientation to the other (see Figure 1.1). Atthermal equilibrium, an excess of population exists in the lower energy state,the EPR signal strength depends on this Boltzmann population difference. FromEq. 1.4 by using values of βe and ~ in frequency unit, the resonance conditionbecomes

B0(mT ) =71.448× ν(GHz)

g(1.5)

For a free electron (ge = 2.0023) in a magnetic field of 330 mT, the resonantfrequency is 9.248 GHz.

1.2.2 Crystal-field splitting and spin-orbit coupling

Although there are a number of paramagnetic centers that can be describedas two-level systems with resonances around ge = 2.0023 (e.g. free radicals insolution), more complicated spectra are often observed as additional magnetic andelectric fields are exerted by the environment of the unpaired electron. In manycases further complications arise from the presence of more than one electron,e.g. for transition metal ions with several unpaired d -electrons (up to five forhigh-spin Mn2+ or Fe3+) or organic molecules in triplet states.

For transition metal ions like Fe3+ or Ni2+ the other interactions can be muchlarger than the electron Zeeman term, depending on the nature and the symmetryof the chemical environment of the ion. In this case the energy of the microwavefrequency may be too small to excite all the transitions. Furthermore only thelowest energy levels are populated if the energy splitting is much larger than kBT ,where kB is the Boltzmann constant and T the temperature. In this case EPRspectra can only be observed in the ground state manifold of the paramagneticsystem. For transition metal ions in the condensed phase the interactions respon-sible for the large splittings are the crystal-field (CF) splitting and the spin-orbitcoupling (SOC). The combined effect of these two interactions removes the or-bital degeneracy of the energy levels for most transition metal ions completely,leaving a non-degenerate ground state, often with zero orbital angular momen-tum (quenching of the orbital angular momentum). One example of the effect ofan octahedral CF on a transition metal ion is illustrated in Figure 1.2 with thearrangement of the ligands drawn schematically in the inset. In the special caseof a 3d9 ion in Oh symmetry such as for instance Cu2+, the degeneracy of the Eg

term is not affected by the SOC and the Jahn-Teller theorem applies. The orbitaldegeneracy is lifted and the energy of the system lowered by a displacement ofthe ligands on the z-axis (static Jahn-Teller effect). An elongation of the coordi-nation octahedron leads to the energy level scheme shown in Figure 1.2 with theunpaired electron in the dx2−y2 orbital.


Figure 1.2: Illustration of the effect of an octahedral CF on the energy levels of a3d9 transition metal ion. The resulting energy splitting is ∆. The degeneracy of theorbitals is further lifted by a Jahn-Teller effect through an elongation of the axial bonds.The geometrical arrangement of the ligands is schematically shown in the inset. TheZeeman splitting of the ground level (B1g) is indicated in the small shaded box.

The measurement of EPR spectra is limited to the Zeeman splitting imposedby an external field on the unpaired electron in the non-degenerate dx2−y2 orbital(small shaded area in Figure 1.2) and the spin system can now be described by thespin Hamiltonian in Eq. 1.2 with S = 1/2 since the ground state is non-degenerateand has only associated spin angular momentum. The ligand field does, however,act on the electron spin through the second order effect of the SOC. Since theground state of most molecules has zero orbital angular momentum due to thequenching by chemical bonding (e.g. radicals) or large CF’s (e.g. transition metalions) the SOC is zero to first order. The g-factor should thus have precisely thefree electron value. Interaction between the ground state and excited states,however, admixes small amounts of the orbital angular momentum to the groundstate which couple to the electron spin and make the electron spin sensitive toits crystalline environment. As a result the g-factor is no longer isotropic - thesplitting of the Zeeman levels depends on the symmetry of the ligand field andthe orientation of the system in the external magnetic field. This effect is takeninto account in Eq. 1.2 by writting g as a tensor rather than using the scalarvalue ge.


For the description of the EPR spectra in the ground state of a paramagneticion, an effective spin Hamilton operator is introduced which strongly depends onthe involved energy terms and thus on the individual ion and its environment. Thespin operator S used in the spin Hamiltonian is an effective spin, reflecting thenumber of energy levels involved in the EPR spectrum and does not necessarilyrefer to the true electron spin quantum number. In the case of the Cu(II) ion andof organic radicals, electron spin and effective spin are the same. Even if they arenot, this has no influence on the theory and no difference will be made betweeneffective spin and true electron spin in the following.

1.2.3 g anisotropy

When dealing with magnetic moments in a crystalline or molecular environmentone has to be aware that the cw and pulse EPR spectra may either depend onthe orientation of the sample in the magnetic field B0 (e.g. single crystals) or aresuperpositions of many different single crystal spectra with random orientation(e.g. powder samples). Intermediate situations occur in liquid crystals or partiallyordered systems.

In the easier case the only magnetic field experienced by the electron spin isthe external magnetic field B0 and the spin vector S is oriented either parallelor antiparallel to B0. As shown above additional inner fields, transmitted to theelectron spin by the SOC, are present in molecular systems which may be of thesame order of magnitude or even larger than the external field. The effective fieldBeff , experienced by the electron spin in a molecular or crystalline environmentis a superposition of the external and internal fields. The orientation dependentvariation of the Zeeman splitting is expressed by a g tensor.

Second-order pertubation theory yields for the g tensor [Wiel et al., 1994]

g = ge1l + 2λΛ (1.6)

where 1l is the 3× 3 unit matrix, λ is the spin-orbit coupling constant and Λ is asymmetric tensor that describes SOC in excited states and can be computed byperturbation theory. Any anisotropy or deviation from ge results from the tensorΛ, and involves only contibutions of the orbital angular momentum from excitedstates. The g tensor in Eq. 1.6, which contains information about the symmetryof the inner fields, can be obtained experimentally. In a compound with a para-magnetic metal ion the g tensor is essentially determined by this metal ion andthe directly coordinated ligand atoms.

In the principal axes frame the g tensor is expressed by

g =

gx

gy

gz

(1.7)


where gx, gy and gz are the principal values of g. For orthorhombic symmetry,the orientational dependence of the g value is described by

g(θ, φ) = (g2x sin2 θ cos2 φ + g2

y sin2 θ sin2 φ + g2z cos2 θ)

12 (1.8)

where (θ, φ) are the polar angles describing the orientation between B0 and theprincipal axes of the g tensor. For axial symmetry, where the z axis is the ro-tational symmetry axis, the principal values are reduced to g⊥ = gx = gy andg‖ = gz with

g(θ) = (g2⊥ sin2 θ + g2

‖ cos2 θ)12 (1.9)

An axial g tensor with g‖ > g⊥, represented by a rotational ellipsoid, and theline shape of the EPR spectrum obtained for a large number of paramagneticsystems with random orientation of their g ellipsoids with respect to the staticmagnetic field B0 are shown in Figure 1.3.

Figure 1.3: Schematic drawings of an axial g ellipsoid and the corresponding EPRspectrum. The powder line shape results from the contributions of a large number ofindividual spin packets with different resonance positions. The arrows correspond toprincipal g values.

1.2.4 Nuclear Zeeman interaction

As the electron spin S, also the nuclear spin I is quantized in a magnetic field,resulting in the Zeeman splitting of the nuclear spin states. This nuclear Zeemaninteraction (NZI) is described by


HNZ = −gnβn

~B0I (1.10)

In principle, there exists a small anisotropic deviation of the nuclear g valuefrom gn (chemical shift). However, this chemical shift is much smaller thanlinewidths in spectra of paramagnetic species and can thus be neglected in EPRspectroscopy. For stable isotopes, I covers the range from 1/2 (e.g. 1H, 13C ) to 6(50V). The dimensionless gn factor, as well as the nuclear spin quantum numberI, is an inherent property of the nucleus; e.g., gn(1H) = 5.58; gn(2H) = 0.857;gn(14N) = 0.40. The NZI usually has no influence on cw EPR spectra since thisinteraction is much smaller than the EZI (e.g. for protons the NZI is 1/658 of theEZI). For a given nuclear spin quantum number I a splitting into (2I+1) energylevels is observed, each level being characterized by the nuclear magnetic spinquantum number mI=I,I-1, ...,-I.

1.2.5 Hyperfine interaction

The interaction between the electron spin and the nuclear spin is denoted ashyperfine interaction (HFI) and is described by the Hamiltonian

HHF = SAI (1.11)

Eq. 1.11 with the hyperfine tensor A can be written as the sum of the isotropicor Fermi contact interaction HF , and the electron-nuclear dipole-dipole couplingHDD.

The Fermi contact interaction plays a role if there is finite (unpaired) electronspin density at the nucleus. It depends on the symmetry of any spin polarizedorbitals. There will be no isotropic interaction with a nucleus if the electronresides in an orbital with a node at the site of the nucleus (p, d, f , . . . orbitals).The Fermi contact interaction is given by

HF = aisoSI (1.12)

with

aiso =2

3

µ0

~geβegnβn | Ψ0(0) |2 (1.13)

where |Ψ0(0)|2 is the electron spin density at the nucleus and aiso is the isotropichyperfine coupling constant. The latter (assuming 100% spin density at thenucleus) can be obtained from standard tables (e.g. aiso(

1H) = 50.68 mT, aiso(19F)

= 1886.53 mT). The isotropic interaction aiso can be also significant when theunpaired electron mainly resides in a p or a d orbital; in such cases the spin densityat the nucleus is induced by configuration interaction or spin polarization, whichtransfers spin density from the singly occupied molecular orbital (SOMO) to thenucleus via s-orbital.


Interaction between an electron and a nuclear dipole is the source for theanisotropic component of the hyperfine coupling. The orientation dependentdipole-dipole interaction between the magnetic moments of the electron and thenucleus is described by electron-nuclear dipole-dipole interaction

HDD =µ0

4π~geβegnβn

[(3Sr)(rI)

r5− SI

r3

]= STI (1.14)

where r is the vector connecting the electron and the nuclear spin and T isthe symmetric and traceless dipolar coupling matrix. The dipolar interaction(described by T) acts through space between the nucleus and the electron. As theunpaired electron is delocalized in its orbital, the dipolar interaction depends onthe integral over the SOMO. For orbitals with inversion symmetry, this interactionvanishes. However, different interactions of the unpaired electron with electrons ofdifferent spins can spin-polarize other doubly occupied orbitals. These polarizedorbitals will also interact with the nucleus. For 3d complexes, there is oftensignificant spin density in the 2p and 3p orbitals. One of the useful parameterthat can be derived from the matrix T is the so-called uniaxiality parameter

T0 =2

5

µ0

4π~gnβn〈r−3〉 (1.15)

where the angular brackets imply integration over a p or d orbital. The anisotropichyperfine couplings for one unpaired electron in a p orbital centered on an atomcan be found in tables (e.g. 14N:1.981 mT). When the ligand nuclei are at a suf-ficient distance (r > 0.25 nm) from the electron, the hyperfine coupling betweenthe electron and the nuclear spin is often treated in the point-dipole approxima-tion. Here, the electron spin density is considered to be located at a single pointin the space (for example on the transition metal ion)

Tij = ρM µ0

4π

gnβeβn

r3~gi(3rirj − δij) [i, j = x, y, z] (1.16)

Eq. 1.16 can the be used to determine the positions and the distances of theligand nuclei from the metal center by their orientationally dependent hyperfineinteraction. This derivation is predominantly used for protons since for them thedipolar interaction is the only source of anisotropy.

The total hyperfine tensor A can be written as the sum of the isotropic hy-perfine coupling, aiso, and the dipolar part

A = 1laiso + R(α, β, γ)

Tx

Ty

Tz

R(α, β, γ) (1.17)

where the second term represents the dipolar coupling with the matrix R whichtransforms the matrix T from the laboratory frame to the molecular frame withEuler angels α, β, γ. Since usually the g tensor defines the molecular reference


frame and is not coaxial with the A tensor, a rotation of the A tensor relativeto the g frame is needed. The dipolar part T is composed from contributionsof different molecular orbitals, e.g., for 14N the dipolar part can be split intoorthorhombic matrices due to the contribution of different 3p (px, py, pz) orbitals.For an axial system Eq. 1.17 becomes

A = 1laiso + R(α, β, γ)

−T−T

2T

R(α, β, γ) (1.18)

The largest dipolar coupling (2T ) is found along the vector r connecting theelectron and nuclear spin.

1.2.6 Nuclear quadrupole interaction

Nuclei with nuclear spin quantum number I>1/2 are distinguished by a nonspherical charge distribution resulting in a nuclear electric quadrupole momentQ. The interaction of this charge distribution with the electric field gradient,caused by the electrons and nuclei in its close vicinity, is described by

HNQ = IPI (1.19)

where P is the traceless nuclear quadrupole tensor. In its principal axes systemEq. 1.19 can be written as

HNQ = PxI2x + PyI

2y + PzI

2z (1.20)

=e2qQ

4I(2I − 1)~[3I2

z − I(I + 1) + η(I2x − I2

y )]

where the nuclear quadrupole tensor is

P =e2qQ

4I(2I − 1)~

−(1− η)−(1 + η)

2

(1.21)

where e is the elementary charge, eq is the magnitude of the Electric Field Gradi-ent (EFG) seen by the nucleus, and Q is the quadrupole moment (the electricalshape of the nucleus is a fixed parameter for each isotopes species). The asymme-try parameter η = (Px−Py)/Pz, with |Pz|>|Py|>|Px| and 0 < η < 1, describes thedeviation of the field gradient from axial symmetry. The largest principal valueof the quadrupole tensor is given by Pz = e2qQ/2I(2I − 1)~. In the literaturethe two quantities e2qQ/~ and η are usually given. In an EPR spectrum, the nu-clear quadrupole interaction manifests itself as second order shifts of the allowedresonance lines and in the appearance of forbidden transitions, both of these ef-fects are not easy to observe in disordered systems. In nuclear frequency spectrameasured by ESEEM and ENDOR methods, the nuclear quadrupole interactionmanifests itself as a first order line splitting.

1.3 - Continuous wave EPR 11

1.3 Continuous wave EPR

Despite the development of sophisticated pulse EPR methods, studies of parama-gnetic systems usually still begin with analysis of the cw EPR spectrum. Thisis because for most species, cw EPR spectroscopy is somewhat more sensitive,and because the dominanting interactions are revealed in cw EPR spectra. CWEPR spectra are recorded by putting a sample into a microwave (mw) irradia-tion field of constant frequency ν and sweeping the external magnetic field B0

through resonance. In the experimental set up the mw field is built up in a res-onator (typically a rectangular cavity), into which the sample tube is introduced.The resonator is critically coupled which means that, off resonance from the spinsystem, the incident power is completely absorbed by the resonator. Additionalabsorption by the sample (spin system) on resonance leads to a detuning of theresonator and the reflection of mw power. The recording of this reflected mwpower as a function of the magnetic field yields the cw EPR spectrum. Am-plitude modulation of the magnetic field with a frequency of typically 100 kHzand lock-in detection increases the signal-to-noise (S/N) ratio considerably andis responsible for the derivative shape of the spectra. In the following we discusshow information on dynamics and on structural details can be extracted fromlineshape of cw EPR spectra.

1.3.1 Molecular and chemical dynamics

Spectroscopic experiments are influenced by dynamics on different time scales.According to the uncertainty relation ∆E∆t ≥ h/(4π), a transition cannot havea well defined energy if the lifetime of the excited state is too short. This phe-nomenon is called lifetime broadening. In magnetic resonance lifetime broadeningmanifests itself as line broadening due to a short transverse relaxation time. Typi-cal lifetimes of excited spin states in EPR spectroscopy are in the µs to ms range.Furthermore, lines may be broadened if the transition frequency is changed bya dynamics process such as a chemical reaction or reorientation of the molecule.This happens when the product of the characteristic time τc of the dynamic pro-cess with the frequency change ∆ν during the process is close to unity. If theprocess is fast compared to the frequency change (τc∆ν 1), a narrow line at theaverage frequency is observed. On the other hand, if the process is much slower(τc∆ν 1), the spectrum is a superposition of the spectra of all orientations.EPR spectra are sensitive to molecular reorientation on the ps to µs time scale,which is typical for slow tumbling of molecules in viscous solutions and in poly-mer matrices close to the glass transition temperature. The sensitivity of EPRline shapes to such motion is often used in a combination with spin probe or spinlabel techniques to obtain information on dynamic processes. Stable free radicals,usually nitroxide radicals, are mixed into the material of interest (spin probes) orare chemically attached to a macromolecule (spin labels). For nitroxides in dilutesolution, many of the terms in the Hamiltonian of Eq. 1.1 are irrelevant and onlythe EZI and the hyperfine coupling to the 14N (I = 1) remain. Due to the fast


motion of the spin probes the g and the hyperfine values are averaged and theHamiltonian describing a nitroxide becomes

Hnitroxide =βeB0gisoS

~+ aisoSI (1.22)

where giso = 13(gx+gy+gz). Solving this Hamiltonian yields the energy eigenvalues

and assuming that the constant magnetic field B0 is applied in the z-direction,the spin vectors S and I are replaced by the magnetic quantum number mS

(mS = ±1/2) and mI (mI = 0,±1)

Enitroxide = βeB0gisomS + aisomSmI (1.23)

The selection rules for EPR on this system are ∆mS = ±1 and ∆mI = 0, sothat the resonance condition becomes

(E+ − E−) = hν = βeB0giso + aisomI (1.24)

which explains the three-line pattern of the nitroxide cw EPR spectra. Figure 1.4depicts the energy states of Eq. 1.23 as a function of the magnetic field (e.g.during a field sweep) and also shows the three allowed transitions that lead tothe three-line nitroxide spectrum.

Figure 1.4: B0-field dependence of the energy states of a nitroxide; also shown arethe allowed EPR transition fulfilling the selection rules ∆mS = ±1 and ∆mI = 0.

Eq. 1.24 also shows that the major advantage of performing high-field/high-frequency cw EPR, e.g., going to W-band frequencies, is the improved g-resolution,while the hyperfine resolution remains unaltered. It thus becomes possible to sep-arate contributions to the spectra from electron Zeeman and hyperfine anisotropies.

Figure 1.5 presents the molecular coordinate system that is usually chosen forthe description of the interactions of the nitroxide molecules. The 2pz orbital of


the nitrogen atom defines the z-axis, while the x-axis is along the N-O bond andthe y-axis perpendicular to both. In most nitroxides, the coordinate frames (i.e.,the directions of the main tensor elements) of the electron Zeeman and hyperfineinteraction are almost collinear.

Figure 1.5: Chemical structure of nitroxide molecules and principal axis system ofelectron Zeeman and hyperfine tensor. The cw EPR spectra due to nitroxide moleculesoriented along the principal axes are also shown.

The nitroxide EPR spectra are characterized by an anisotropic g tensor andan anisotropic hyperfine tensor. The hyperfine coupling is maximum along thedirection of the p orbital on the nitrogen (Azz) and minimum in the plane per-pendicular to that direction (Axx ≈ Ayy). On the other hand, the g value ismaximum along the N-O bond (gxx), minimum along the direction of the p or-bital (gzz), and intermediate along the molecular y-axis (gyy). If the nitroxiderotates with a rate that exceeds the magnitude of the anisotropy of the hyperfineinteraction (fast motion regime), the resulting spectrum shows a simple tripletpattern characterized either by uniform (average or isotropic spectra, τc ≤ 10 ps)or nonuniform linewidths (10 ps ≤ τc ≤ 4 ns). For correlation times τc ≥ 4 ns,dynamics is slow on the EPR timescale (slow motion regime), and the lineshapebecomes more complicated. Finally, if the nitroxide rotates with a rate that issmaller than the intrinsic linewidth (τc ≥ 1 µs), a powder pattern results thatis almost insensitive to changes in τc (rigid limit spectra). In Figure 1.6 nitrox-ide EPR spectra corresponding to different rotational rates are shown. In thefast limit we observe an isotropic spectrum and the splitting of outer peaks 2A

′zz

corresponds to twice the isotropic hyperfine coupling (aiso). In the fast motionregime, the dependence of the peak-to-peak width of an individual hyperfine lineon the nuclear spin state (mI) can be expressed as

∆B(mI) = A + BmI + Cm2I (1.25)

where A, B, and C are only weakly mI-dependent quantities that are calculatedfrom g and A values and the rotational correlation time. In the fast motion


Figure 1.6: Dependence of the nitroxide cw EPR spectrum on the rotational correla-tion time τc. Simulated spectra with different rotational correlation times.

regime the rotational correlation times are gained from linewidth analysis. In theslow motion regime the lines are stringly and asymmetrically broadened and therotational correlation times can be determined by simulating the lineshape of theEPR spectra. In the rigid limit case, the spectrum corresponds to a superpositionof the spectra of all possible orientations of the nitroxide molecule with respect tothe constant magnetic field B0 and the splitting of outer peaks 2A

′zz corresponds

to the biggest hyperfine component (2Azz).In addition to spin probe dynamics, cw EPR spectra also contain information

on the chemical environment of the spin probe. This information is present inthe spectra mainly in the 14N hyperfine splitting: all interactions of the nitroxidemolecule with its surroundings that lead to a change in the electronic structureof the spin probe also change the hyperfine interaction. Therefore, the same kindof spin probe in solvents of different polarity, pH or hydrophobicity will showdeviations in the hyperfine splitting aiso in liquid state cw EPR spectra, anddifferent values for Azz in powder-like cw EPR spectra. The reason for this canbe understood when inspecting the (coincident) molecular g- and A-coordinateframe shown in Figure 1.5. It becomes clear that hydrogen bonding of the NO-


oxygen atom to, e.g., solvent protons affects the Axx value and, to a much strongerdegree, also the gxx value, while due to molecular geometry the changes in polaritymainly influence the Azz value (and to a lesser degree also gxx).

1.3.2 Geometric and electronic structure

We already said that in a compound with a paramagnetic metal ion the g matrixis essentially determined by this metal ion and the directly coordinated ligandatoms. The determination of the g values by simulating the cw EPR spectra,allows one to get more insights about the paramagnetic ion. For example, thecw EPR spectra of 3d9 ions, with ground state electronic configuration dx2−y2

are characterized by an axial symmetry g-tensor with g‖ > g⊥. On the otherhand, when the ground state electronic configuration is dz2 , the cw EPR spectraof these ions are characterized by an axial symmetry g-tensor with g‖ < g⊥. Thedirect coordination of the metal ion to ligand atoms also influences the g values.This can be a useful tool to detect the formation of metal ion complexes.

A typical X-band cw EPR spectrum for an axially elongated copper complexis shown in Figure 1.7.a. The spin Hamiltonian for a spin system consisting of aCu(II) ion (S = 1/2, I = 3/2) with weakly coupled nuclei from ligands in the firstand second coordination spheres (e.g. 13C nuclei and protons) can be written asfollows

H =βeB0gS

~+ SACuI +HN (1.26)

The electron Zeeman interaction is much larger than the hyperfine couplingto the copper nucleus, expressed by ACu. The copper quadrupole interaction isneglected. The g and the ACu tensors are often assumed to be coaxial and axiallysymmetric. The third term HN describes the 13C and proton interactions whichare small compared to the first and second term. The cw EPR spectrum is thenobtained by neglecting the last term, and is described by the principal values g‖,g⊥, A‖, A⊥ of the g and ACu tensors. The parameters chosen for the simulationof the spectrum are those for a copper-hexaquo complex with g‖ > g⊥ and A‖ >A⊥. The spectrum is a superposition of four axial powder line shapes which canbe assigned to the different mI states. The 2D plot (Figure 1.7.b.) shows thecontributions of the different mI states as a function of the angle θ (defined inFigure 1.3) as the angle between the z-axis and the external magnetic field B0.In experimental spectra the splitting in the g‖ region is often not resolved due tolarge linewidths whereas the g⊥ and A⊥ values can usually be determined withoutdifficulties. The interaction of the electron spin with the weakly coupled nuclei ofthe ligands described by HN is usually not resolved in the cw EPR spectra andcan only be investigated by applying pulse EPR. Due to the limited bandwidthof the mw pulses only a small part of the EPR spectrum calculated from theHamiltonian in Eq. 1.26 is excited. The dashed lines in Figure 1.7.b indicate theexcitation bandwidth (∼50 MHz) for a pulse EPR experiment performed at the


Figure 1.7: (a) CW EPR spectrum of an axially elongated copper complex (firstderivative); (b) same spectrum drawn as absorption spectrum with a 2D plot showingthe different mI transitions as a function of the angle θ. The g and the A tensor uniqueaxes deviate slightly from each other.

B0 position indicated by the arrow. At a given B0 position spin packets fromdifferent mI states with different θ values can thus contribute to the spectrum.

1.4 Pulse EPR

In pulse EPR experiments the spectrum is recorded by exciting a large frequencyrange simultaneously with a single high-power mw pulse of given frequency νmw

at a constant magnetic field B0. The length of the mw pulse determines theexcitation bandwidth of the pulse. The excitation profile of the pulse is obtainedin the limits of the linear response theory by Fourier Transformation of the pulseshape. If tp is the length of the pulse, a homogeneous excitation is expected in anarea of width 1/tp. An excitation bandwidth of 100 MHz is thus obtained for a10 ns pulse, which corresponds to 3 mT at g = 2. Reduction of the pulse length

1.4 - Pulse EPR 17

leads to broader excitation range. Due to technical limitations (i.e., deadtime ofthe spectrometer and limited mw power) it is, however, in most cases impossibleto excite and observe the whole spectrum by a single EPR pulse, which is a severelimitation as compared to pulse NMR. The length of the pulse thus determinesthe selectivity of the excitation.

Many EPR applications still make use of cw methods as the recording andinterpretation of the pulse EPR spectra requires more sophisticated technicalequipment and a more advanced theoretical background. A significant advan-tage of cw EPR with respect to the pulse methods is the higher sensitivity. Afurther limitation of pulse EPR is the low measuring temperatures imposed bythe fast relaxation of the transverse magnetization involved in the pulse experi-ments. However, additional information about weakly coupled nuclei and relax-ation properties of the spin system can be obtained by manipulating the spinswith sequences of mw pulses. In the following we will proceed in this way: firstwe describe the effects of a mw pulse on a spin system and we discuss the conceptof relaxation. Then the electron spin echo phenomenon and the model system forpulse EPR experiments are introduced. Finally, the pulse EPR techniques usedin this thesis are described.

1.4.1 MW pulse and spin dynamics

In this section we introduce the concepts of macroscopic magnetization and ro-tating frame to describe the effects of mw pulses on a given spin system.

The macroscopic magnetization M of a sample is obtained by adding up thecontribution of all microscopic magnetic moments in a given sample volume.In the presence of an external magnetic field B0 (taken along the z-axis) andafter the magnetization vector is tilted with respect to the z-axis by mw field,the macroscopic magnetization M starts to precess around B0 with a Larmorfrequency ν = |γ|B0 with |γ| the gyromagnetic ratio. At thermal equilibriumthe precession of the magnetic moments about the external magnetic field B0 isout of phase, the magnetic xy components are randomly distributed and a netmagnetization Mz is observed only along B0 in the Boltzmann equilibrium. If wechoose a rotating frame where the axes system rotates with the frequency νmw ofthe mw excitation about the z-axis, the excitation becomes time independent andthe precession frequency for the electron spins is given by the offset Ω = ν − νmw

of the Larmor frequency ν, obtained from the resonance condition in Eq. 1.4, andthe excitation frequency νmw.

In the most simple pulse experiment mw radiation is applied with frequencyνmw and polarized in the xy plane. The oscillating B1 field of this radiation canbe decomposed into two counter rotating components one of which is rotating inthe same direction as the Larmor precession of the spins. Hence, in the rotatingframe B1 is a stationary field, defined along x. A π/2 pulse along x thus transfersthe longitudinal magnetization Mz into transverse y magnetization. After themw pulse is switched off the magnetization vector moves back to its equilibrium


position along z, thereby precessing with the offset Ω (see Figure 1.8). Thisbehavior is called relaxation. The precession leads to oscillating component Mx

and My of the magnetization vector in the xy-plane which can be recorded asfree induction decay (FID) after the mw pulse.

Figure 1.8: Illustration of the simplest pulse experiment with a single mw π/2 pulse(top) and the behavior of the magnetization vector M in the rotating frame (bottom).(a) Boltzmann equilibrium, the magnetization vector M is aligned along the static fieldB0; (b) M is tilted by 90 in the xz plane under the pertubation B1; (c) after the pulsethe magnetization precesses with Ω back to its equilibrium position.

1.4.2 Relaxation time and linewidth

Now we discuss the effects which bring the magnetization vector back to itsequilibrium position. In an S = 1/2 system the spins can be oriented paral-lel or anti-parallel to the external magnetic field and a pertubation (e.g. pulse)induces transitions between the two energy levels. The Mz magnetization atequilibrium is the result of a small surplus of spins parallel to their quantizationaxis. Changing the Mz magnetization involves reorientation of the microscopicmagnetic moments. Transversal magnetization Mx and My on the other hand isgiven by an in-phase precession of the magnetic moments about the external field,induced e.g. by a mw pulse. Upon relaxation the Mz magnetization is restoredand the transversal magnetic components Mx and My vanish. These changes inmagnetization are associated with the spin-lattice or longitudinal relaxation timeT1 and the spin-spin relaxation time T2, respectively. The spin-lattice relaxationrelates to the characteristic lifetime of the spin state and is determined by thedissipation of energy via the thermal vibration of the lattice. T1 is related throughthe uncertainty relation ∆E∆t ≥ h/(4π) to the linewidth of an individual spin

1.4 - Pulse EPR 19

packet. A small T1, taken as a measure for ∆t, leads to a smearing out of theenergy levels (large ∆E) and thus a broad resonance line. Large T1 values areusually found for systems with isolated electronic ground states, well separatedfrom the lowest excited states. The spin-spin relaxation is concerned with themutual spin flips caused by dipolar and exchange interactions between the as-sembly of spin in the sample. T2 is usually much shorter than T1 and is thusthe dominant contribution to the linewidth. By adding the two contributions theresultant linewidth of a single spin packet is

1

T′2

=1

T2

+1

2T1

(1.27)

T′2 is the relaxation time which is of importance in connection with the transverse

magnetization generated and observed in pulse EPR experiments. T′2 is tempera-

ture dependent and determines the homogeneous linewidth of a single spin packet.For transition metal ion complexes, T

′2 values in the range of several µs are com-

mon at liquid helium temperatures. Among the various effects influencing T′2 we

only mention high spin concentrations or clustering of paramagnetic moleculeswhich may lead to a considerable shortening of the relaxation time.

1.4.3 Electron spin echo

The spin echo phenomenon was first observed by Erwin Hahn in 1950 [Hahn,1950]. It is based on the non-linear behaviour of an ensemble of spins withdifferent Larmor frequencies. In Figure 1.9 the pulse sequence and the diagramdescribing the motion of the magnetization vectors in a primary echo experimentis shown.

Figure 1.9: Pulse sequence (a) and the diagram (b) describing the motion of themagnetization vectors in a primary echo experiment. The numbers label the points intime shown in the diagram; resonant excitation with mw pulses along the x-axis.

At thermal equilibrium, the magnetization vector is oriented along the z-axis.The π/2 pulse along the x-axis rotates the magnetization to the −y-axis (timepoint 1). After the pulse, the different spin packets begin to precess with their


individual Larmor frequencies Ω around the z-axis, resulting in a defocusing of thetransverse magnetization. After time τ (time point 2), a π pulse again along thex-axis turns all the magnetization vectors through 180 about this axis (time point3). Since the directions of rotation of the individual spin packets are not changedby the refocusing pulse, after another time τ , all the vectors are aligned alongthe y-axis, giving rise a net y-magnetization called electron spin echo (time point4). The importance of spin echo experiments in pulse EPR is explained by thebroad lines encountered in solid-state EPR which make the recording of FID oftenimpossible. Anisotropic interaction like the g tensor or the B0 inhomogeneitieslead to an inhomogeneous broadening of the EPR line. Unlike the homogeneouslinewidth, the inhomogeneous linewidth is not related to any relaxation time.The dephasing of the magnetic moments under the action of the magnetic fielddistribution leads to a rapid vanishing of the transversal magnetization, whichoften occurs already during the spectrometer deadtime.

1.4.4 Origin of the nuclear modulation effect

In 1965, Rowan, Hahn, and Mims [Rowan et al., 1965] observed that for cer-tain samples the decays of primary electron spin echoes were modulated withfrequencies that correspond to nuclear frequencies and their combinations (differ-ences and sums). This electron spin echo envelope modulation effect (ESEEM)is caused by the hyperfine and nuclear quadrupole interaction, and it allows fora detection of nuclear spectra without directly exciting the nuclear spins. Thebasic principles of many pulse EPR experiments can be explained by consideringa spin system consisting either of one electron spin S = 1/2 and one nuclearspin I = 1/2 or of one electron spin S = 1/2 and one nuclear spin I = 1. Thetreatment here is restricted to these two cases.

Spin system S = 1/2, I = 1/2

The nuclear spin Hamiltonian for an S = 1/2, I = 1/2 system with an isotropic gtensor and an anisotropic hyperfine interaction, can be expressed in the laboratoryframe, where the static magnetic field vector B0 is taken along the laboratoryz-axis, as

H0 = ωSSz + ωIIz + SAI (1.28)

where ωS = geβeB0/~ and ωI = −gnβnB0/~ are the electron and the nuclear Zee-man frequency, respectively, expressed in angular frequency units. The hyperfinetensor contains secular (SzIz), pseudo-secular (SzIx, SzIy) and non-secular (SxIx,SxIy, SyIx, SyIy) terms. In the high-field approximation, the non-secular termscan be neglected, since these off-diagonal elements are small compared to thefrequency difference of the diagonal elements. If we switch to a frame where thenucleus lies in the xy plane of the laboratory frame, Eq. 1.28 becomes

1.4 - Pulse EPR 21

H0 = ωSSz + ωIIz + ASzIz + BSzIx (1.29)

with B = (B2x + B2

y)1/2. In the rotating frame the Hamoltonian is given by

H0 = ΩSSz + ωIIz + ASzIz + BSzIx (1.30)

For an axial hyperfine interaction the coefficients A and B are related to theprincipal values A⊥ and A‖ of the hyperfine matrix and to the dipolar couplingconstant T by

A = A‖ cos2 θ + A⊥ sin2 θ = aiso + T (3 cos2 θ − 1) (1.31)

B = (A‖ − A⊥) sin θ cos θ = 3T sin θ cos θ (1.32)

Diagonalization of the Hamiltonian in Eq. 1.30 yields the following expressionsfor the two nuclear frequencies

ωα =| ω12 |=[(ωI +

A

2)2 +

B2

4

] 12

(1.33)

and

ωβ =| ω34 |=[(ωI −

A

2)2 +

B2

4

] 12

(1.34)

and the allowed (∆mS = ±1, ∆mI = ±0) and forbidden (∆mS = ±1, ∆mI =±1) electron transition frequencies ω13 = ωS + ω−/2, ω24 = ωS − ω−/2 andω14 = ωS + ω+/2, ω23 = ωS − ω+/2, respectively, with ω+ = ω12 + ω34 and ω−= |ω12 − ω34| (for the definition of ωα and ωβ see Figure 1.10). The transitionprobabilities of the allowed (Ia) and forbidden (If ) electron transition are givenby

Ia = cos2 η =| ω2

I − 14ω2− |

ωαωβ

(1.35)

and

If = sin2 η =| ω2

I − 14ω2

+ |ωαωβ

(1.36)

where 2η is the angle between the nuclear quantization axes in the two mS man-ifolds, and ωI is the Larmor frequency.The energy level scheme for an S = 1/2, I = 1/2 system, together with theEPR and nuclear frequency spectra, is shown in Figure 1.10. Depending on thedominating interaction one has to consider the weak coupling |A|<2|ωI | or strongcoupling |A|>2|ωI | case. The effective quantization vector, experienced by the nu-cleus is obtained by addition of the HF and NZ interactions. For an isotropic case


(B = 0), the NZ interaction and the HF field act in the same direction z for bothweak and strong coupling in the mS = −1/2 manifold. The splitting between thenuclear Zeeman states is thus increased by both interactions. In the mS = 1/2manifold the NZ interaction and the HF field act in the opposite direction. Atexact cancellation (|A|=2|ωI |) these two terms cancel and the nuclear frequencyis zero in this manifold. In the strong coupling case, since HF>NZ, the energyof the |αβ> and |αα> states are interchanged in their order, as compared to theweak coupling case.

Figure 1.10: Energy level scheme for an S = 1/2, I = 1/2 system in the weak andstrong coupling case together with electron (full arrows: allowed transition; dashed ar-rows: forbidden transitions) and nuclear transitions (dash-dotted arrows) for the strongcoupling case. For an isotropic hyperfine interaction the values of nuclear frequenciesare shown.

ESEEM techniques rely on the excitation of both allowed and forbiden EPRtransitions using mw radiation. There is no ESEEM effect if If = 0. Thisexcludes the measurement of isotropic hyperfine couplings (T = 0 → B = 0)and the principal values of the hyperfine couplings (θ = 0, π/2 → B = 0). Theformer precludes the measurement of hyperfine couplings of paramagnetic centersin solution if the observed couplings are averaged to aiso by rapid moleculartumbling.

Spin system S = 1/2, I = 1

In the section above, the influence of the magnitude of the hyperfine term rela-tive to the nuclear Zeeman term on the energy levels for a nucleus with I = 1/2was shown. For nuclei with I = 1, with the hyperfine interaction approximatelytwice the NZI, the level splitting of one mS manifold is primarily determinedby the NQI. In this situation of exact cancellation, when the hyperfine and thenuclear Zeeman terms match, the effective field experienced by an I = 1 nucleusvanishes in the mS = 1/2 manifold. The nuclear frequencies within this mani-fold therefore correspond to the nuclear quadrupole resonance (NQR) frequencies

1.4 - Pulse EPR 23

ω0 = 2Kη, ω− = K(3 − η), and ω+ = K(3 + η) with K = e2qQ/4~. This con-dition leads to a line narrowing and consequently the intensity increases. Themanifold mS = −1/2 where the NZ and HF interactions are additive gives rise tomuch broader resonances and often the only distinguishable feature is the doublequantum (dq) transition, ∆mI = 2. This is because to first order the dq frequen-cies are free from quadrupole broadenings. The energy level diagram illustratingthis situation is shown in Figure 1.11. If the anisotropic hyperfine interactionis small compared to the isotropic and the nuclear quadrupole interaction thedouble quantum frequency is given by

ωdq = 2[(mSaiso + ωI)

2 + K2(3 + η)2] 1

2 (1.37)

Figure 1.11: Energy level diagram for a S = 1/2, I = 1 spin system under thecondition of exact cancellation (|ωI | = |A/2|).

If the hyperfine energy with coupling constant A exceeds both the NZI andthe NQI for an axially symmetric system with I = 1 the energy level diagramshown in Figure 1.12 is obtained.

In ESEEM experiments, usually both single quantum (sq) and double quan-tum (dq) nuclear transitions can be observed, while in electron nuclear doubleresonance (ENDOR) spectroscopy only the sq nuclear transitions are observed,and the corresponding frequencies are given in general by the first order equations

ωsq =A

2± ωI ± 3

Pz

2(1.38)

The four ENDOR lines predicted by Eq. 1.38 are often not resolved. When P issmaller than the ENDOR linewidth, 0.1-0.5 MHz for frozen solution of transitionmetal ions, only two lines centered at A/2 and separated by 2ωI are observed.


Figure 1.12: Energy level diagram for a S = 1/2, I = 1 spin system in the strongcoupling case, |A|>2|ωI | and a positive nuclear quadrupole interaction along this inter-action.

1.4.5 Orientation selection in pulse EPR

The absorption spectrum of a paramagnetic center in a frozen solution or in apowder is anisotropic. Usually, the electron Zeeman interaction gives the largestcontribution. For organic radicals the anisotropy is small, and it is possible toexcite the entire spectrum with a short intense mw pulse. In transition metalcomplexes the situation is typically very different; the width of the absorptionspectrum (≈1 GHz) is much larger than the effective excitation bandwidth of themw pulse (≈50 MHz). This orientationally selective excitation allows the mag-netic interactions with respect to the g tensor coordinate system to be estimated.By setting the observed field B0 at the extreme edges of the spectrum only a verylimited range of orientations contributes to the ESEEM or ENDOR spectrum;this is commonly called a ”single crystal like” position (position 1 in Figure 1.13.a,position 2 in Figure 1.13.b, positions 1 and 3 in Figure 1.13.c). At these observerpositions the nuclear frequency spectra exhibit sharp lines. Nuclear frequencyspectra obtained with an arbitraty B0 orientation show a much less pronouncedstructure and are more difficult to interpret (position 2 in Figure 1.13.c) sincethe spectrum consist of many orientations. For systems with an axial g tensor,an observer position exists in the EPR spectrum which corresponds to all the B0

orientations in the plane defined by g⊥ (position 2 in Figure 1.13.a). Thus, set-ting the magnetic field at such a position results in a nuclear frequency spectrumwhich is a superposition of the spectra arising from all selected B0 orientations.Depending on the relative orientation of the g and hyperfine tensors the low-and high-field end of a spectrum may be dominated by orientations far awayfrom canonical orientations (position 2 in Figure 1.13.a: extra absorption peakcorresponding to a non-canonical orientation). The orientation selection can beimproved by performing the experiment at higher mw frequencies (increase of theEZI with costant HFI).

1.4 - Pulse EPR 25

Figure 1.13: Typical field-swept powder EPR spectra (in absorption) and orienta-tion selections on the unit sphere. Full arrows: observed positions for recording singlecrystal-like nuclear frequency spectra. Double-line arrow: observer position for record-ing all orientations in the g⊥ plane. Dashed arrow: extra absorption peak. Dottedarrow: observer position ath the gy feature. (a) Axially symmetric spectrum withg‖>g⊥ and I = 3/2 (e.g. Cu(II) complex); (b) axially symmetric spectrum with g‖<g⊥ and I = 7/2 with A‖A⊥ (e.g. Co(II) low-spin complex); (c) orthorhombic spec-trum (e.g Ni(I) complex or Fe(III) low-spin complex) (Modified from [Schweiger andJeschke, 2001]).

1.4.6 Two-pulse ESEEM

In the two-pulse ESEEM experiment (see Figure 1.9) the intensity of the primaryecho is recorded as a function of the time interval τ between the π/2 and π pulses.The echo amplitude modulation formula for an S = 1/2, I = 1/2 system is givenby [Schweiger and Jeschke, 2001]

V2p = 1− k

4[2− 2 cos(ωατ)− 2 cos(ωβτ) + cos(ω−τ) + cos(ω+τ)] (1.39)

where k is the orientation dependent modulation depth parameter given by


k(θ) = sin2 2η =

(BωI

ωαωβ

)2

(1.40)

For the case of an isotropic hyperfine interaction A = aiso1l, or B0 orientedalong one of the principal axis of the hyperfine tensor (θ = 0 or θ = π/2), theecho modulation disappears since B in Eq. 1.40 is zero. The modulation depthis maximum when |A|≈2|ωI |. Here, A is the hyperfine coupling at an arbitraryorientation.

The main shortcoming of the two-pulse experiment is that the primary echodecays with the phase memory time, TM ,3 which is often very short. This disad-vantage can be avoided with the three-pulse ESEEM experiment.

1.4.7 Three-pulse ESEEM

In the three-pulse ESEEM experiment (see Figure 1.14) the first π/2 pulse createsallowed and forbidden electron coherence evolving during τ . The second π/2pulse partially transfers this electron coherence to nuclear coherence which evolvesduring the evolution time T and decays with the transverse nuclear relaxationtime T2n, which is usually much longer than the corresponding relaxation timeTM of the electrons. The third π/2 pulse transfers the nuclear coherence back toobservable electron coherence. The modulation of the stimulated echo is givenby

V3p(τ, T ) =1

2[V α(τ, T ) + V β(τ, T )] (1.41)

with the contribution from the α = 1/2 (β = -1/2) electron spin manifold

V α(β)(τ, T ) = 1− k

2[1− cos(ωβ(α)τ)][1− cos(ωα(β)(τ + T ))] (1.42)

Figure 1.14: Pulse sequence for three-pulse ESEEM experiment.

When T is varied the echo envelope is modulated ony by the two basic fre-quencies ωα and ωβ. In contrast to the two-pulse ESEEM experiment, the sumω+ and the difference ω− frequencies do not appear. The three-pulse ESEEMamplitudes depend on τ which results in blind spots at ωα(β) when τ = 2πn/ωβ(α)

(n = 1, 2, . . .). As a consequence, the experiment often has to be performed fordifferent τ values and spectra have to be added to avoid blind spots artefacts.

3The phase memory time TM is an empirical parameter usually associated with the decayof the primary echo. TM corresponds to the inverse homogeneous linewidth and is sometimessimply called T2 in literature.

1.4 - Pulse EPR 27

1.4.8 Hyperfine sublevel correlation experiment

The four-pulse HYSCORE experiment (see Figure 1.15) [Hoefer et al., 1986] isa 2D technique where an additional π pulse is introduced between the secondand third π/2 pulse of the three-pulse ESEEM. During the first evolution periodt1, the nuclear coherence created by the π/2 − τ − π/2 sub-sequence evolvesin the α(β) electron spin manifold. The non-selective π pulse interchanges thenuclear coherences between the electron spin α and β manifolds. During thesecond evolution period t2, the transferred nuclear coherence evolves in the β(α)electron spin manifold. Finally, the nuclear coherence is transferred to electroncoherence by the last π/2 pulse and is detected as an electron spin echo, which ismodulated with the nuclear frequencies. For the two spin system cross-peaks canbe observed at (±ωα, ±ωβ) and (±ωβ, ±ωα). If the modulation depth is small(i.e. 2|ωI ||A| or |A|2|ωI |), non-ideal, so called matched, mw pulses with anoptimized strength and length can be used [Jeschke et al., 1998].

Figure 1.15: Pulse sequence and experimental time traces of a typical four-pulseHYSCORE experiment with different mw channel phases recorded in the transientmode of the spectrometer. The position of the events (pulses, wanted echo) are indi-cated by the dashed lines. The huge amount of unwanted echoes is visible in the upper8 traces. The lowest trace is the result of the summation with the phase +-+- -+-+.


A drawback of the HYSCORE sequence is the occurence of unwanted echoesand thus echo crossings caused by the number of pulses. However, this problemcan be circumvented by applying phase cycling during the experiment. The resultof such an 8-step phase cycle is shown in Figure 1.15. The residual unwantedechoes due to anti-echo pathways can be neglected in practice.

Figure 1.16 shows typical HYSCORE powder patterns from an S = 1/2,I = 1/2 spin system. In the strong coupling case, |A|>2|ωI |, the correlationridges are oriented parallel to the diagonal and are separated by approximately2|ωI |. In the weak coupling case, |A|<2|ωI |, the two arcs are displaced fromthe anti-diagonal at |ωI |, with a maximum frequency shift given by [Poeppl andKevan, 1996]

∆ωmax =9

32

T 2

| ωI |(1.43)

The advantage of ∆ωmax for inferring the anisotropic part of the hyperfinecoupling arises from the fact that the intensities of the end points of the arcsvanish since they correspond to a depth parameter k = 0.

Figure 1.16: Theoretical HYSCORE powder patterns for an S = 1/2, I = 1/2 spinsystem with an axial hyperfine tensor. (a) Strong coupling case with νI = 3.5 MHz,aiso = 18 MHz and T = 6 MHz (black curves), aiso = 2 MHz and T = 14 MHz (graycurves). The additional patterns appearing in the (+, +) quadrant have been omittedfor clarity reasons. (b) Weak coupling case with νI = 14 MHz, aiso = 2.5 MHz and T= 6 MHz. Units in MHz. (Modified from [Schweiger and Jeschke, 2001])

The maximum frequency shift ∆ωmax from the anti-diagonal at |ωI | corre-sponds to an orientation of θ = 45 between the B0 field and the z-component ofthe hyperfine tensor. According to Eq. 1.43, the anisotropic hyperfine couplingcan be calculated from the maximum shift of the cross peaks with respect tothe (ωI , ωI) peak. An additional advantage of this method is that in case of an

1.4 - Pulse EPR 29

anisotropic g tensor, the relative orientation of the hyperfine tensor versus theg tensor can be deduced directly from the spectrum. In many cases, however, acommon problem is that the sensitivity at the edges is poor, since the modulationintensity is zero at canonical orientations of the hyperfine tensor.

For nuclei with I>1/2, the frequencies of the ∆mI = 2 transitions to first orderdo not depend on the nuclear quadrupole coupling. The correlation patterns ofall the other transitions depend on the nuclear quadrupole coupling to first order,which causes an orientation dependent shift of the correlation features along thediagonal. The sign of the shift is opposite for the transitions ∆mS = 0, 1 and∆mI = 0, 1, hence the first order patterns are symmetric with respect to theanti-diagonal at ωI . The pattern depends also on the relative orientations ofhyperfine and nuclear quadrupole interactions. It can be seen from Eq. 1.38 thatthe double quantum transitions (mI , mI+2) = (-1, 1) do not depend to firstorder on the nuclear quadrupole coupling. Thus, correlation patterns similar tothose found for nuclei with I = 1/2 are expected. However, the single quantumtransitions (mI , mI+1) = (0, 1) and (-1, 0) do depend to first order on the nuclearquadrupole coupling and, therefore, are usually broad for disordered systems.This is the reason that usually the double quantum cross peaks are the mostprominent features of the 14N HYSCORE spectra. When the nuclear quadrupoleinteraction is much weaker than the nuclear Zeeman and hyperfine interactions,the transition probabilities of the double quantum transitions are very small andthe single quantum correlation peaks dominate the spectrum. This is the casefor deuterium nuclei. The 14N HYSCORE spectra of the Cu(II)NCTPP complexis shown in Figure 1.17.

Figure 1.17: HYSCORE spectra of the outer 14N of Cu(II)NCTPP magneticallydiluted in ZnTPP powder, measured at the outer edge of the low-field mI transition ofg‖. (a) X-band spectrum; ν14N = 0.9 MHz; τ = 100 ns. (b) Q-band spectrum; ν14N =3.6 MHz; τ = 100 ns. (Modified from [Mitrikas et al., 2005]).


1.4.9 Electron nuclear double resonance

ENDOR is a well-established magnetic resonance technique for obtaining detailedmolecular and electronic structure information about paramagnetic species. Theshapes of ESEEM and ENDOR spectra are generally different, and the methodsgive complementary information. In ESEEM experiments only mw radiation isused and the indirect excitation of the nuclear spins arises from the excitationof EPR forbidden transitions. ESEEM is very well suited for measuring weakhyperfine couplings of the order of the nuclear Zeeman frequencies, while ENDORgenerally works better for strong hyperfine couplings. In ENDOR experimentsboth mw and radio frequency (rf) radiations are used. The ENDOR effect isbased on the transfer of polarization between electron and nuclear transitions.The nuclear polarization, created by a single mw pulse is changed by a selectiverf pulse. During the detection period, the nuclear polarization is transferred toelectron coherence and indirectly observed via an FID or an echo of the electronspins.

Davies ENDOR

The pulse ENDOR scheme (Figure 1.18) introduced by Davies [Davies, 1974] isbased on selective mw pulses. A first selective mw π pulse inverts the polarizationof a particular EPR transition (see Figure 1.19 for a S = 1/2, I = 1/2 modelsystem). During the mixing period, a selective rf pulse with flip angle β is appliedon-resonance with one of the nuclear frequencies, the polarization of this transi-tion is chaged accordingly. This change which also alters the polarization of theEPR observed transition, is then measured via a primary electron spin echo. TheENDOR spectrum is thus recorded by monitoring the primary echo intensity asthe rf frequency is incremented stepwise over the desired frequency range.

Figure 1.18: Pulse sequence for the Davies ENDOR experiment. All interpulse delaysare fixed and the radio frequency is varied.

The echo can disappear completely if the rf matches exactly one of the twoallowed nuclear transition frequencies and if the rf pulse flip angle is π. Thiscorresponds to an ENDOR efficiency of 50%.

The intensities of the NMR lines in the Davies ENDOR spectrum depend onthe length tmw (selectivity) of the first mw pulse. This effect is described by theselectivity parameter ηS = aisotmw/2π [DeRose and Hoffman, 1995]. For ηS1,

1.4 - Pulse EPR 31

Figure 1.19: Effect of an rf pulse resonant with the nuclear transition with frequencyωα on the population of the energy levels of the EPR observer transition (highlight) ina Davies ENDOR experiment.

the mw pulse excites one EPR transition of each spin packet, whereas in the caseof ηS<1, both transitions are affected. This is called self-ELDOR effect, since twoelectron transitions with different frequencies are excited. The absolute ENDORintensity is described as a function of the selectivity parameter ηS by

V (ηS) = Vmax

√2ηs

η2S + 1

2

(1.44)

where Vmax is the maximum ENDOR intensity obtained with

ηmaxS =

√2

2(1.45)

Mims ENDOR

The pulse ENDOR scheme (see Figure 1.20) introduced by Mims [Mims, 1965] isbased on the stimulated echo sequence with three non-selective mw π/2 pulses.

Figure 1.20: Pulse sequence for the Mims ENDOR experiment. All interpulse delaysare fixed and the radio frequency is varied.


The preparation sequence, π/2−τ−π/2, creates a grated polarization patternwhich is changed by the selective rf pulse and is recorded as a function of rffrequency via the stimulated electron spin echo. The ENDOR efficiency is givenby [Schweiger and Jeschke, 2001]

FENDOR =1

4(1− cos(aisoτ)) (1.46)

and depends upon the hyperfine coupling constant aiso and the time τ . It ismaximum for τ = (2n + 1)π/aiso, and zero for τ = 2nπ/aiso, with n = (0, 1,2, . . .). Mims ENDOR thus exhibits a blind spot behavior similar to the three-pulse ESEEM and HYSCORE experiments, but which now depends upon aiso

(in three-pulse ESEEM the blind spots depend upon ωα and ωβ). Note that thedeadtime of the spectrometer prevents very small τ values from being used, sothat at X-band frequencies typically τ > 100 ns are used. For τ = 100 ns blindspots occur when aiso = (0, 10, 20, . . .) MHz. For large hyperfine couplings itis thus usually preferable to employ the Davies ENDOR sequence with a wellchosen length for the inversion π mw pulse. On the other hand, Mims ENDORcan be particularly sensitive for measuring small hyperfine couplings if the phasememory time TM of the sample is sufficiently long to allow an optimal τ value tobe used [Zanker et al., 2005].

Chapter 2

EPR investigation of a DNAmodel system

2.1 Introduction

The interaction of metal complexes with nucleic acids is one of the central topicsin bioinorganic chemistry. Copper in the form of an azamacrocyclic complex isable to convert the right-handed B form of poly(dG-dC) to the correspondingleft-handed Z form [Spingler and Pieve, 2005]. Preformed Z-DNA crystals thatwere soaked with copper salts bound the metal ions exclusively via N7 of guanine[Geierstanger et al., 1991]. On the other hand, simple copper salts are known todenature the same DNA sequence [Hiai et al., 1965]. We are interested to under-stand why Cu(II) as an aquated ion denatures DNA on the one hand and inducesthe conformational change to Z-DNA in the form of an azamacrocyclic complexon the other hand. Copper is an ideal metal for electron paramagnetic resonancespectroscopy [Andersson et al., 2003, Calle et al., 2006, Farrar et al., 1996]. Inorder to understand the interaction between copper and oligonucleotides, one hasfirst to study the copper-mononucleotide system. Pulse EPR techniques allow,together with the use of isotope-labeled nucleotides, to determine the distancebetween the paramagnetic copper center and selected atoms around within a ra-dius of not more than 6 A. In particular, we are able to describe the geometryof copper(II) bound to 5′-guanosine-monophosphate (5′-GMP) in solution. Thisstudy is the entry point to further characterize copper-oligonucleotide systems.

2.2 Materials and methods

2.2.1 Sample preparation

To charactarize the copper-mononucleotide interaction, the following sampleshave been prepared :

– sample 1, Cu(II) trifluoromethanesulfonate (triflate; Cu(CF3SO3)2) in 50%dimethyl sulfoxide (DMSO) and 50% H2O;

33

34 Chap. 2 - EPR investigation of a DNA model system

– sample 2, Cu(II) triflate and 5′-GMP (Figure 2.1.a) in 50% DMSO and 50%H2O

– sample 3, Cu(II) triflate and 5′-GMP selectively deuterated at position 8(D8-5′-GMP, Figure 2.1.b) in 50% DMSO and 50% H2O

– sample 4, Cu(II) triflate and 5′-GMP deuterated at all the exchangeableprotons (H8-d5-5

′-GMP, Figure 2.1.c) in 50% DMSO-d6 and 50% D2O

All reagents and nucleotides were purchased from Fluka at the highest purityavailable. In all samples the metal concentration was 2 mM. For the 5′-GMP-containing samples (2, 3, 4) the metal to ligand ratio was 1:1. 5′-GMP deuteratedat the 8 position (D8-5′-GMP) was prepared by overnight treatment of 5′-GMPat 60 C with 4 equiv. of triethylamine in D2O. Excess reagent was removed andexchangeable hydrogens back exchanged to protons by repeated lyophilization todryness from H2O [Hoogstraten et al., 2002]. 1H NMR showed the absence of anytriethylamine. The H8-d5-5

′-GMP ligand was prepared by overnight treatment inD2O. Excess reagent was removed by lyophilization. The pH of all samples wasdeterminated to be 6.4.

Figure 2.1: Chemical structures of guanosine 5′-monophosphate (5′-GMP) and selec-tively deuterated 5′-GMP: (a) 5′-GMP; (b) D8-5′-GMP; (c) H8-d5-5′-GMP.

2.2.2 Spectroscopy

CW EPR measurements at X-band were carried out on a Bruker ELEXSYS E500spectrometer equipped with a Bruker super-high-Q cavity. Experimental condi-tions were: mw frequency ≈9.5 GHz; mw power incident to the cavity = 20mW;modulation frequency = 100kHz; modulation amplitude = 0.5 mT. Cooling ofthe sample was performed with a liquid nitrogen dewar vessel to 130 K.

The X-band pulse EPR experiments were performed with a Bruker E580 spec-trometer (mw frequency ≈9.7 GHz) equipped with a liquid helium cryostat fromOxford Instruments. All pulse experiments were done at 20 K and at a repetitionrate of 1 kHz. The magnetic field was measured with a Bruker ER083 CS Gaussmeter.

The HYSCORE spectra were recorded with the following instrumental pa-rameters: tπ/2 = tπ = 16 ns; starting values of the two variable times t1 and t2,

2.3 - Results and discussion 35

56 ns; time increment, ∆t = 16 ns (data matrix 350×350). To avoid blind spots,spectra with different τ values were recorded and added. An eight-step phasecycle was used to remove unwanted echoes.

The Davies ENDOR experiments were carried out with the pulse sequenceπ-T -π/2-τ -π-τ -echo, with mw pulse lengths of 32, 16, and 32 ns, respectively,and an interpulse time τ of 400 ns. An rf π pulse of variable frequency and alength of 9 µs was applied during time T (10 µs).

The Mims ENDOR experiments were performed using the pulse sequence π/2-τ -π/2-T -π/2-τ -echo, with mw pulse lengths of 16 ns and an interpulse time of τ= 500 ns. During time T (10 µs) a rf pulse with variable frequency and a lengthof 9 µs was applied. Both HYSCORE and ENDOR experiments were carriedout at different observer positions that correspond to different orientations of themolecules with respect to B0 (orientation selectivity).

2.2.3 Data manipulation

The data were processed with the program MATLAB 7.2 (MathWorks, Natick,MA, USA). The time traces of the HYSCORE spectra were baseline correctedwith an exponential function, apodized with a Gaussian window and zero filled.After a two-dimensional Fourier transformation, the absolute value spectra werecalculated. The HYSCORE spectra recorded with different τ values were addedto eliminate τ -dependent blind spots. The cw EPR and ENDOR spectra weresimulated with the EasySpin package [Stoll and Schweiger, 2006]. For the cw EPRsimulations the hyperfine coupling shift due to the two copper isotopes 63Cu and65Cu was taken into account. The HYSCORE spectra were simulated with aprogram written in-house [Madi et al., 2002].

2.3 Results and discussion

The pKa value of the second deprotonation step of the phosphate group in 5′-GMP was determined to be 6.25 ± 0.02 [Sigel et al., 1994]. Addition of one equiv.of copper lowers this pKa value down to 4.9 ± 0.3 [Sigel and Song, 1996]. The pHof all measured samples was 6.4 ± 0.1, therefore all the formed Cu(II)-5′-GMPcomplexes had a neutral charge.

2.3.1 CW EPR spectra

The frozen solution X-band EPR spectra of the Cu(II) triflate sample (sample1) and the Cu(II)-5′-GMP samples (samples 2, 3, and 4) together with theirsimulations are shown in Figure 2.2. All spectra revealed features that are typicalfor Cu(II) complexes with a (dx2−y2)1 ground state (g‖>g⊥) [McGarvey, 1966] (seeTable 2.1).

The comparison of the g values and the Cu(II) hyperfine parameters betweenCu(II) triflate (sample 1) and Cu(II)-5′-GMP (samples 2, 3, and 4) indicates a


metal-guanosine interaction. Indeed the EPR parameters of Cu(II) triflate are

Table 2.1: g values and 63Cu hyperfine parameters.

sample g⊥(±0.005) g‖(±0.005) |A⊥|(±10) |A‖|(±10)(MHz) (MHz)

1 2.079 2.406 25 3822 2.072 2.358 15 4553 2.067 2.340 15 4604 2.069 2.353 15 455

typical for Cu(II) complexes with four oxygen atoms as equatorial ligands [Peisachand Blumberg, 1974]. Instead the g values and Cu(II) hyperfine parametersof Cu(II)-5′-GMP complexes are typical for Cu(II) complexes with either threeoxygen atoms and one nitrogen atom as equatorial ligands or four oxygen asequatorial ligands. We cannot distinguish between the two possibilities by theinspection of cw EPR spectra. However application of pulse EPR spectroscopy(see 14N Davies ENDOR section) ruled out the second possibility, confirming thatthe Cu(II) ion is coordinated to three oxygen atoms and one nitrogen atom. Theclear changes in g‖ and A‖ prove that the surrounding environment of the Cu(II)ion changed when the 5′-GMP ligand was added. Moreover the g values and thehyperfine parameters of the samples 2, 3, and 4 very similar to each other. Thisis an indication that the interaction between the Cu(II) ion and the guanosinenucleotide is the same for all three slightly different kinds of 5′-GMP ligandsthat we used. Although the cw EPR data give evidence of a coordination ofthe copper metal by the GMP nucleotide, detailed information about the metalion site coordination cannot be obtained. Further insight into the metal ion sitecoordination is provided by pulse EPR techniques.

2.3.2 14N Davies ENDOR

The Davies ENDOR spectra of Cu(II)-H8-d5-5′-GMP (sample 4) at different mag-

netic field positions, together with their simulations are shown in Figure 2.3.Davies ENDOR experiments are very well suited for measuring strong hyperfinecouplings. The spectra are characterized by two broad peaks centered at ap-proximately |A/2| = 16 MHz and split by twice the nitrogen nuclear Zeemanfrequency, 2νN . This is typical for a strongly coupled nitrogen nucleus with|A/2|>|νN |. This result combined with the cw EPR parameters, validates thehypothesis that we have a copper complex with three oxygen atoms and one ni-trogen atom as equatorial ligands. The additional splitting observed in all spectrais assigned to the nuclear quadrupole interaction. Moreover, signals from weaklycoupled protons, which overlap with signals from the strongly coupled nitrogen,


Figure 2.2: X-band cw EPR spectra taken at 130 K: (a) Cu(II) triflate in DMSO/H2Osolution; (b) Cu(II)-5′-GMP complex in DMSO/H2O solution; (c) Cu(II)-D8-5′-GMPcomplex in DMSO/H2O solution; (d) Cu(II)-H8-d5-5′-GMP complex in DMSO-d6/D2Osolution. Thin traces: experiments; thick traces: simulations. A-F: observer positionsused for the Davies and Mims ENDOR measurements.


have been suppressed by using short mw pulses. This effect is known as hyper-fine contrast selective ENDOR [Schweiger and Jeschke, 2001]. Davies ENDORmeasurements at different observer positions (thin traces) and their simulations(thick traces) allow for the estimation of the principal values of the hyperfine, A,and nuclear quadrupole, Q, coupling tensors: (Ax, Ay, Az) = (32.0, 38.0, 31.0)± 0.2 MHz, and (Qx, Qy, Qz) = (2.1, -1.5, -0.6) ± 0.5 MHz. Since the larger hy-perfine coupling occurs at the observer position corresponding to g⊥, we concludethat this strongly coupled nitrogen occupies an equatorial position. Althoughthe atom N7 is the most plausible binding site of the ligand, from these data weare not able to identify, in a definitive way, the identity of this strongly couplednitrogen. However, with the knowledge of the hyperfine interactions between theunpaired electron of the Cu(II) ion and the other magnetic nuclei, and of thenuclear quadrupole interaction, it is possible to obtain detailed insight into thelocal environment of the paramagnetic center, and validate the hypothesis thatthe Cu(II) ion is directly coordinated to atom N7.

Figure 2.3: X-band Davies ENDOR spectra of Cu(II)-H8-d5-5′-GMP in DMSO-d6/D2O (sample 4) taken at observer position A-C (see Figure 2.2, spectra d). Thintraces: experiments, thick traces: simulations.

2.3.3 1H HYSCORE

To get information about the weakly coupled nuclei, the HYSCORE techniquewas utilized. The X-band proton spectra of samples 2, 3, and 4 are shown in


Figure 2.4 together with the corresponding simulations. The HYSCORE spec-trum of Cu(II)-5′-GMP complex (sample 2, Figure 2.4.a) shows at least two typesof weakly coupled protons. One type is characterized by an intense ridge closeto the anti-diagonal at the proton Larmor frequency (∼14 MHz) and can eitherbe assigned to protons of water molecules coordinated in axial positions or/andweakly coupled protons of solvent molecules [Schosseler et al., 1997]. The othertype of proton HYSCORE signal is distinguished by a broad ridge shifted awayfrom the anti-diagonal. This last proton signal can be simulated with the hyper-fine parameters (Ax, Ay, Az) = (-6.3, -6.3, 9.3) ± 0.2 MHz and Euler angles [α,β, γ] = [0, 90, 0], which correspond to an isotropic hyperfine coupling constantof aiso = -1.1 ± 0.2 MHz and a dipolar coupling constant of T = 5.2 ± 0.2 MHz.In the point-dipole approximation, the latter value corresponds to a point-dipoledistance of 2.5 ± 0.1 A. These weakly coupled protons can either be assigned toligand protons of the nucleotide or to water molecules directly coordinated to thecopper. In order to be able to distinguish between these options we repeated thesame experiment for the Cu(II)-D8-5′-GMP (sample 3) with a deuterium nucleusin the 8 position (see Figure 2.1.b).

Figure 2.4: X-band proton HYSCORE spectra measured at a magnetic field posi-tion close to g⊥. (a) Cu(II)-5′-GMP (sample 2), τ = 100, 120, 140, and 160 ns; (b)Cu(II)-D8-5′-GMP (sample 3), τ = 100, 120, 140, and 160 ns; (c) Cu(II)-H8-d5-5′-GMP (sample 4), τ = 100, 120, 140, 160, 400, 450, and 500 ns; (d)-(f) correspondingsimulations. The antidiagonal lines are given for ν1H .

New peaks from a deuterium nucleus appeared at low frequency (for details seethe 14N, 2H HYSCORE section, Figure 2.5.b). Moreover, the proton HYSCOREspectrum of Figure 2.4.b can be simulated with the same hyperfine parameters


we found for sample 2 (see Table 2.2). This finding indicates that in both com-plexes we can detect the same type of weakly coupled protons. This proton peakcertainly can not be assigned to H8 because in sample 3 there is a deuteriumnucleus at this position. Moreover, the hyperfine couplings and the distance ob-tained from our analysis are compatible with the hyperfine couplings and the dis-tance of water protons in Cu[(H2O)6]

2+ complexes determined in a HYSCORE[Schosseler et al., 1997] and a single crystal cw ENDOR study [Atherton andHorsewill, 1979]. These considerations confirm the hypothesis that the aforemen-tioned HYSCORE signal can be assigned to protons of water molecules directlycoordinated to the copper center. To study the hyperfine interaction between theunpaired electron of the copper ion and the H8 proton of the nucleotide we havemeasured the HYSCORE spectrum of Cu(II)-H8-d5-5

′-GMP (sample 4 in 50%DMSO-d6 and 50% D2O). In this sample, all the exchangeable protons have beenreplaced with deuterium nuclei (see Figure 2.1.c). The experimental spectrumof Figure 2.4.c consists of proton peaks close to the anti-diagonal at the protonLarmor frequency and combination peaks between the proton and the deuteriumfrequency. The proton signal can be simulated with the hyperfine parameters(Ax,Ay, Az) = (-0.3, -0.3, 7.5) ± 0.2 MHz and Euler angles [α, β, γ] = [0, 30, 0],which correspond to an isotropic hyperfine coupling constant of aiso = 2.3 ± 0.2MHz and a dipolar coupling constant of T = 2.6 ± 0.2 MHz. In the point-dipoleapproximation, the latter value corresponds to a point-dipole distance of 3.1 ±0.1 A. This weakly coupled proton can be exclusively assigned to H8 since in thissample there are no other protons in such a close proximity of the copper center.Moreover, our data are in agreement with the imidazole proton measured in acopper-histidine complex by W-band Davies ENDOR [Manikandan et al., 2001].The distance between the H8 proton and the paramagnetic center is compatiblewith the hypothesis that the Cu(II) ion is directly coordinated to nitrogen N7.

Table 2.2: 1H hyperfine parameters, Euler angles, and derived distances.

sample aiso(±0.2)(MHz) T (±0.2)(MHz) [α, β, γ]() r(±0.1)( A)

2 -1.1 5.2 [0, 90, 0] 2.53 -1.1 5.2 [0, 90, 0] 2.54 2.3 2.6 [0, 30, 0] 3.1

2.3.4 14N, 2H HYSCORE

The low frequency region of the HYSCORE spectra of samples 2, 3, and 4 areshown in Figure 2.5 together with the corresponding simulations.

The HYSCORE spectrum of Cu(II)-5′-GMP (sample 2, Figure 2.5.a) is domi-nated by cross-peaks that are assigned to dq correlation peaks from 14N and theirstrong intensity is typical for disordered S = 1/2, I = 1 spin systems with negligi-


Figure 2.5: X-band nitrogen and deuterium HYSCORE spectra measured at a mag-netic field position close to g⊥. (a) Cu(II)-5′-GMP (sample 2), τ = 100, 120, 140,and 160 ns; (b) Cu(II)-D8-5′-GMP (sample 3), τ = 100, 120, 140, and 160 ns; (c)Cu(II)-H8-d5-5′-GMP (sample 4), τ = 100, 120, 140, 160, 400, 450, and 500 ns; (d)-(f)corresponding simulations. The antidiagonal lines are given for ν2H and 2ν2H .

ble anisotropic hyperfine coupling [Dikanov et al., 1996]. Numerical simulationsof the set of spectra recorded at different observer positions (not shown here)yield the following hyperfine and nuclear quadrupole parameters: (Ax, Ay, Az) =(1.51, 1.51, 1.63) ± 0.05 MHz, and Euler angles [α, β, γ] = [0, 10, 0], |e2qQ/h|= 2.44 ± 0.05 MHz with Euler angles [α, β, γ] = [90, 90, 0] and η = 0.56 ±0.05. This nitrogen is characterized by a fairly isotropic hyperfine interaction anda significant quadrupole interaction. The latter can be used as a probe in orderto identify this nitrogen. Its comparison with nuclear quadrupole parameters ofguanine nitrogens obtained by nuclear quadrupole resonance studies [Garcia andSmith, 1983] indicates that N1 (e2qQ/h = 2.63 MHz, η = 0.60) is a more likelycandidate than N7 (e2qQ/h = 3.27 MHz, η = 0.16). Although other GMP nitro-gens like N9 (e2qQ/h = 1.91 MHz, η = 0.75) or N3 (data not available) cannot beexcluded, our nuclear quadrupole parameters together with the weak hyperfinecoupling strongly suggest that the signal can be assigned to a remote nitrogen(other than N7) of the GMP ligand. This finding is in line with the hypothesisthat the Cu(II) ion is directly coordinated to nitrogen N7, as was suggested bythe previous analysis of the proton HYSCORE spectra.

The HYSCORE spectrum of the Cu(II)-D8-5′-GMP sample (sample 3) (Fig-ure 2.5.b) also shows a similar remote nitrogen pattern. This weakly coupled


nitrogen is characterized by the same hyperfine and nuclear quadrupole parame-ters of the nitrogen in sample 2. Moreover in sample 3, a new cross peak appearsclose to the 2H Larmor frequency at ∼ (2.3, 2.3) MHz, which can be assigned tothe deuterium nucleus D8. Since the gyromagnetic ratio for deuterium is approx-imately 6.5 times smaller than for proton, the hyperfine interaction is expectedto scale as A(2H) = A(1H)/6.5. By scaling the hyperfine parameters of the H8proton observed in sample 4, (Ax, Ay, Az) = (-0.05, -0.05, 1.15) MHz, and Eulerangles [α, β, γ] = [0, 30, 0], and assuming a small quadrupole interaction with|e2qQ/h| = 0.4 MHz and η = 0.03 (which is typical for deuterium nuclei bonded tocarbon), we could accurately simulate the peak assigned to the deuterium nucleus(see Figures 2.5.b, e). This cross-check further supports our previous assignmentof the proton H8.In Figure 2.5.c, the HYSCORE spectrum of the Cu(II)-H8-d5-5

′-GMP sample(sample 4) is shown. In this sample we have abundant deuterium nuclei fromthe ligand and the solvent. This together with suppression effects often encoun-tered in ESEEM spectroscopy [Stoll et al., 2005] could account for the absenceof the remote nitrogen peaks observed in the other two samples. The spectrumis characterized by single quantum, double quantum and combination peaks ofdeuterium nuclei. It can be simulated by scaling the hyperfine parameters of theprotons detected in samples 2 and 3 (ascribed to water molecules directly coor-dinated to copper), (Ax, Ay, Az) = (-0.97, -0.97, 1.43) MHz, and Euler angles [α,β, γ] = [0, 90, 0], assuming a negligible quadrupole interaction. This result isin agreement with the assignment of the proton HYSCORE spectra of samples 2and 3 discussed in the previous section.

2.3.5 13C HYSCORE

To detect any coordination between the copper and the solvent we prepared thesample 2 in DMSO labeled with 13C. The HYSCORE spectrum of Cu(II)-5′-GMP in 50% 13C-DMSO and 50% H2O is shown in Figure 2.6 together with thecorresponding simulation. The spectrum shows a similar remote nitrogen patternto the one recorded for the sample 2 in DMSO/H2O (see Figure 2.5.a). A newmatrix peak appears at the 13C Larmor frequency at ∼ (3.6, 3.6) MHz, which canbe assigned to weakly coupled 13C nuclei (this matrix peak is not included in thesimulation). Moreover, the spectrum is characterized by two arcs displaced fromthe antidiagonal at the 13C Larmor frequency which can be assigned to directlycoordinated DMSO molecules. The hyperfine values and the Euler angles usedto simular this last pattern are (Ax, Ay, Az) = (2.4, 2.4, 5.1) ± 0.2 MHz and [α,β, γ] = [0, 70, 0]. We can conclude that some complexes are also coordinatedwith at least one DMSO molecule.


Figure 2.6: X-band carbon HYSCORE spectrum of Cu(II)-5′-GMP in 50% 13C-DMSO and 50% H2O measured at a magnetic field position close to g⊥. (a) Experi-mental spectrum, τ = 100, 120, 140, and 160 ns; (b) simulation. The antidiagonal linesis given for ν13C .

2.3.6 31P Mims ENDOR

To detect the interaction between the copper center and the 31P nucleus of thephosphate group, the Mims ENDOR technique has been used. This method isvery useful to detect weak hyperfine couplings from nuclei that are at relativelylong distances from the paramagnetic center (typically r≥5 A) [Zanker et al.,2005]. A weak 31P coupling should produce a pair of transitions positioned sym-metrically about the Zeeman frequency for that nucleus. Mims ENDOR spectraof the Cu(II)-5′-GMP sample (sample 2) recorded at different magnetic field po-sitions, and their simulations are shown in Figure 2.7.

The hyperfine values and Euler angles used for the simulation are: (Ax, Ay,Az) = (-0.20, -0.20, 0.45) ± 0.05 MHz and [α, β, γ] = [0, 40, 0]. For this interac-tion, essentially no electron delocalization into the 31P nucleus is seen (aiso = 0.02± 0.02 MHz) and a dipolar coupling of T = 0.22 ± 0.02 MHz is obtained. Thisdipolar coupling corresponds to a point-dipole distance r = 5.3 ± 0.2 A. The X-ray structure analysis of Cu(II) with 5′-GMP not only contains three Cu-5′-GMPmolecules in the asymmetric unit but also three different types of interactionsbetween the copper ion and the phosphate group [Sletten and Lie, 1976]. Thecopper ion either binds by direct coordination to the phosphate, or via two hy-drogen bridges mediated via two cis coordinated water molecules or finally via asingle hydrogen bridge to the phosphate mediated by one water molecule that iscopper coordinated. These three different interactions have the following increas-ing copper to phosphate distance ranges: 3.14-3.27, 4.63 and 5.10-5.52 A. Thedistance of 5.3 ± 0.2 A that we have found by the Mims ENDOR method is clearlycompatible with the latter type of a metal-phosphorus interaction distances ob-


Figure 2.7: X-band Mims ENDOR spectra of Cu(II)-5′-GMP (sample 2) taken atobserver positions D-F (see Figure 2.2, spectra b). Thin traces: experiments, thicktraces: simulations.

served for water–mediated phosphate ligation. Therefore we can conclude thatthe Cu(II) ion is indirectly bound via a single water molecule to a phosphategroup.

2.3.7 Structural considerations

The g values obtained for all the samples indicate that the copper ion in all thesamples studied in this work has a (dx2−y2)1 ground state with a square planarsymmetry. From Davies ENDOR spectra is not possible to exactly determinethe number of nitrogen directly coordinated to the copper center. However, fromthe cw spectrum (Figure 2.2, spectra b), we can exclude that more than onenitrogen is directly coordinated to the copper because, since in our samples themetal to ligand ratio was 1:1, we do not have evidence of the presence of twospecies that instead would have formed if more nitrogen atoms were directlycoordinated to the copper center. Combining the results of the Davies ENDORand cw analysis we can conclude that the copper ion is directly coordinated toonly one nitrogen of the nucleotide. Since this latter nitrogen is characterizedby a very large isotropic and a very small anisotropic hyperfine constant, we cansay that the nitrogen is equatorially coordinated to the copper ion. Moreover we


can exclude the presence of a dimer because the cw spectra of all the samplesdid not show any signals at half the magnetic field (data not shown). A searchin the Cambridge Structure Database [Allen and Kennard, 1993] was undertakenin order to determine to which nitrogen atom(s) of any 9-substituted guaninecopper was found to bind. Actually all published X-ray structures of copperbinding to a nitrogen of 9-substituted guanine show that the copper is exclusivelybinding to N7 (see for example the following ref. [Sheldrick, 1981, Sletten andLie, 1976]). From the proton HYSCORE spectra we have evidence of a directcoordination of the copper ion by water molecules and in addition by N7 (becauseof the H8 signal). Moreover, from the HYSCORE spectrum of Cu(II)-5′-GMPin 50% 13C-DMSO and 50% H2O we cannot exclude that some complexes couldbe coordinated also with some DMSO molecules. The molecular model of theCu(H2O)5-5

′-GMP complex consistent with all our data is shown in Figure 2.8

Figure 2.8: Molecular model of the Cu(H2O)5-5′-GMP complex consistent with EPR,ENDOR and HYSCORE data. The g-frame is also shown and the main nuclei arelabeled. The figure was created with the help of the programm Hyperchem 7.5.1 (Hy-percube, Gainesville, FL, USA).


2.4 Conclusions

Metal ion coordination to nucleic acids is essential for the biological function ofnucleic acids. In this first part of the thesis we present a cw and pulse EPRstudy of the mononucleotide model system Cu(II)-5′-GMP. We obtained a com-plete characterization of the structural features of the metal ion bound to thenucleotide. The copper is directly coordinated to N7 of the guanine and to fivesolvent molecules, one of them forms a hydrogen bridge to the phosphate group.The complete characterization of the model system Cu(II)-5′-GMP provides abasis for the characterization of copper-oligonuleotides complexes, described inthe following Chapter 3 of this thesis.

Chapter 3

EPR investigation ofcopper-poly(dG-dC) interaction

3.1 Introduction

Metal ions are present in practically all biological materials: they are known tobind tightly to membranes [Sasaki, 2003] and they participate in the stabilizationof numerous biological structures. Various protein-metal ions complexes havebeen described [Gray, 2003] and several enzymatic reactions depend on the pres-ence of metal ions [Bertini et al., 1994, Lippard and Berg, 1994]. Many metalions are found in intimate association with nucleic acids in their natural envi-ronment and these ions may stabilize the structure of nucleic acids. Studies ofthe interactions of metal ions with DNA have been made at two levels: addi-tion of metal ions to natural DNA preparations and studies of the metal-DNAcomplexes obtained, and search for metal ions already present in DNA moleculesextracted from various biological materials. In this thesis we study the structuralproperties of a copper-DNA complex. It is known that the simple copper saltsare able to denaturate the DNA [Hiai et al., 1965] and copper in the form of anazamacrocyclic complex is able to induce the conformational change from B formof poly(dG-dC) to the corresponding Z form [Spingler and Pieve, 2005]. In thispart of the thesis we used cw and pulse EPR spectroscopy to study the structuralchanges of polynucleotides induced by the presence of a copper ion.

3.2 DNA structure

DNA is a flexible and dynamic molecule that can assume a variety of intercon-verting forms. Each form has its own biological role in the regulation of the lifeof the cell. Genetic information is provided by the DNA in at least two differentways. First, the sequence of the nucleotides determines the primary structureof the proteins. Second, DNA can regulate gene expression through its shape,which may in turn depend on adaption of various conformations. Selective stabi-lization of one of these interconverting forms can help to better understand their

47

48 Chap. 3 - EPR investigation of copper-poly(dG-dC) interaction

biological roles.

DNA consists of two long polymers of simple units called nucleotides, withbackbones made of sugars and phosphate groups joined by ester bonds. These twostrands run in opposite directions to each other and are therefore anti-parallel.They entwine like vines, in the shape of a double helix. The DNA double helixis stabilized by hydrogen bonds between the bases attached to the two strands.The adenine forms a base pair with thymine, as does guanine with cytosine. Inthe canonical Watson-Crick base pairing there are three hydrogen bonds betweenthe guanine and the cytosine bases. Instead, in the less common Hoogsteen basepairing these two bases are hold together by only two hydrogen bonds and theguanine is rotated by 180 with respect to the cytosine. B-DNA is the mostcommon form of DNA under the conditions found in cells and it has a right-handed double helical structure. Another possible double helical structure ofDNA is the Z-DNA. It is a left-handed double helical structure in which thedouble helix winds to the left (instead to the right) in a zig-zag pattern. InFigure 3.1 the different structures of B-DNA and Z-DNA are shown.

Figure 3.1: Structures of Z-DNA and B-DNA. (Taken fromwww.cmgm.stanford.edu/biochem/biochem2001 website).

Formation of Z-DNA necessitates a polymer of mainly alternating purine-pyrimidine nucleobase sequences. In addition, Z-DNA requires as an environmenteither high salt concentrations or the presence of multiple-charged cations [Fuertes

3.3 - Materials and methods 49

et al., 2006].We are interested to characterize the interaction between the copper ion and

the poly(dG-dC)·poly(dG-dC) that presents a Z form. In particular, using cwand pulse EPR techniques, we are able to determine the structural features ofthe metal ion bound to the polynucleotide and predict the structural changes ofthe polynucleotide induced by the presence of the metal ion.


3.3.1 Sample preparation

To charactarize the copper-polynucleotide interaction, we prepared the followingsamples:

– sample 5, Cu(II) triflate and poly(dG-dC)·poly(dG-dC) in 50% DMSO and50% H2O;

– sample 6, Cu(II) triflate and poly(dG-dC)·poly(dG-dC) in 50% 13C-DMSOand 50% H2O

– sample 7, Cu(II) triflate and poly(dG-dC)·poly(dG-dC) deuterated at allthe exchangeable protons in 50% DMSO-d6 and 50% D2O.

All reagents were purchased from Fluka at the highest purity available; thepoly(dG-dC)·poly(dG-dC) was purchased from GE Healthcare and used withoutpurification. In all samples, the concentration of poly(dG-dC)·poly(dG-dC) was10 mM (in bases) and the metal concentration was 0.5 mM. To check a possibledependence of copper coordination on the copper-to-nucleotide ratio, sampleswith metal concentrations of 1 mM and 2 mM have also been prepared andthe cw EPR spectra measured. The poly(dG-dC)·poly(dG-dC) deuterated at allthe exchangeable protons was prepared by overnight treatment in D2O. Excessreagent was removed by lyophilization. The pH of all samples was determinatedto be 6.4. For comparison, the cw EPR spectrum of Cu(II) triflate and poly(dG-dC)·poly(dG-dC) in 100% H2O was also measured.

3.3.2 Spectroscopy and data manipulation

In this part of the thesis we used the same cw and pulse EPR techniques thathave been used to characterize the Cu(II)-5′-GMP complex (see Spectroscopy andData Manipulation sections in Chapter 2 for details). However, the HYSCOREexperiments have been performed in two different ways: the standard HYSCOREwith the sequence π/2-τ -π/2-t1-π-t2-π/2 and tπ/2 = tπ = 16 ns; the proton-matched HYSCORE with the same pulse sequence like the standard HYSCOREand matched pulses (the second and fourth one) for the weak coupling case (pro-ton enhancement). The amplitude of the mw field for the matched pulses was


adjusted by maximizing the two pulse echo with 16- and 32-pulses. This condi-tion corresponds to ω/2π = 15.6 MHz, a value that was found to be sufficientlyclose to the matching field ω/2π for the proton Zeeman frequency at X-band.The length pulse sequence was: 16-τ -72-t1-16-t2-72-τ (pulse lenghts given in ns).

Pulse measurements at Q-band were performed on a home-built spectrometer[Gromov et al., 2001] at 20 K. The Davies ENDOR experiments were carried outwith the pulse sequence π-T -π/2-τ -π-τ -echo, with mw pulse lengths of 400, 200,and 400 ns, respectively, and an interpulse time τ of 900 ns. An rf π pulse ofvariable frequency and a length of 9.4 µs was applied during time T (10 µs).

Circular dichroism (CD) measurements were performed at the University ofZurich on a Jasco J-810 spectropolarimeter equipped with a Jasco PFD-4255Peltier temperature controller running at 25 C. The poly(dG-dC)·poly(dG-dC)and the metal concentration was about 0.14 mM, corresponding to an UV ab-sorption at 260 nm of about 1, and about 285 µM, respectively. The solution was50% DMSO and 50% H2O. Background correction was done by subtracting thespectra of the corresponding solutions without DNA and metal ion.


3.4.1 Evidence for two copper species

The frozen solution X-band EPR spectrum of the Cu(II)-poly(dG-dC)·poly(dG-dC) complex in DMSO/H2O solution (sample 5) together with its simulation areshown in Figure 3.2.a.

It is immediately evident that the spectrum contains two copper species (thesame cw EPR spectrum was obtained for the samples 6 and 7). To test whetherthe two species are induced by the presence of DMSO as a cryoprotectant, wetried to measure the cw EPR spectrum of the Cu(II)-poly(dG-dC)·poly(dG-dC)complex in pure water (Figure 3.2.d). We can see that the spectrum is charac-terized by similar copper species, even if the sensitivity is much lower than inDMSO/H2O solution since probably most of the copper-coordinated DNA pre-cipitates during shock-freezing. This last effect is related to crystallization ofwater. This assumption is supported by the fact that no resolved spectrum isobserved when the slower freezing technique of immersion of the sample tube inliquid nitrogen is used. Even with fast freeze-quenching in dry ice/ethanol theresolution was too poor, and it was not possible to run pulse experiments. Fur-thermore, to demostrate the reproducibility of the data, samples with differentmetal concentration have been measured (Figure 3.2.b,c). The spectra show thesame copper species and the same ratio of the two species.

We can conclude that the copper ion is bound to the polynucleotide in twodifferent ways. From the simulation it was possible to determine the g valuesand the Cu(II) hyperfine parameters of both species (see Table 3.1), and the per-centage of the two populations. These parameters indicate that the two speciesare slightly different, and there is no evidence of free copper in solution (see Ta-


Figure 3.2: X-band cw EPR spectrum of Cu(II)-poly(dG-dC)·poly(dG-dC) complexin DMSO/H2O solution taken at 130 K: (a) metal concentration 0.5 mM; thick trace:experiment, thin trace: simulation; (b) metal concentration 1 mM; (c) metal concentra-tion 2 mM. For comparison the cw EPR spectrum of Cu(II)-poly(dG-dC)·poly(dG-dC)complex in pure water, metal concentration 0.5 mM, is shown (d). The arrows indicatethe observer positions used for the pulse measurements.

ble 2.1). From now on, we shall call ”species I” the one that corresponds tolower population (45%) and the higher field g‖ features, and ”species II” the onecharacterized by a higher population (55%).

3.4.2 Properties in common to both copper species

Combining the results of the ENDOR, HYSCORE and cw analysis we couldobtain a characterization of the properties that can be assigned to both species.Both species revealed features that are typical for Cu(II) complexes with a (dx2−y2)1

ground state (g‖>g⊥). The g values and the Cu(II) hyperfine parameters are typ-ical for Cu(II) complexes with either three oxygen atoms and one nitrogen atom


Table 3.1: g values, 63Cu hyperfine parameters of Cu(II)-poly(dG-dC)·poly(dG-dC)complex in DMSO/H2O, and percentage of population.

species g⊥(±0.005) g‖(±0.005) |A⊥|(±10) |A‖|(±10) population(MHz) (MHz) (%)

I 2.062 2.326 25 442 45II 2.062 2.290 28 465 55

as equatorial ligands or two oxygen atoms and two nitrogen atoms as equatorialligands [Peisach and Blumberg, 1974]. In Figure 3.3 the relationship of g‖ andA‖ for both Cu(II)-poly(dG-dC)·poly(dG-dC) complex species is shown.

Figure 3.3: Relationship of g‖ and A‖ for Cu(II) triflate in DMSO/H2O (circle),Cu(II)-5′-GMP complex in DMSO/H2O (asterisk), Cu(II)-poly(dG-dC)·poly(dG-dC)complex in DMSO/H2O species I (square), Cu(II)-poly(dG-dC)·poly(dG-dC) complexin DMSO/H2O species II (diamond). The regions set off by dashed lines indicatethe area of g‖ and A‖ corresponding to copper complexes with four oxygen atoms asequatorial ligands, to copper complexes with three oxygen atoms and one nitrogen atomas equatorial ligands, and copper complexes with two oxygen atoms and two nitrogenatoms as equatorial ligands.

It is evident that the g values and the Cu(II) hyperfine parameters are inthe regions corresponding to copper complexes with three oxygen atoms and onenitrogen atom as equatorial ligands or two oxygen atoms and two nitrogen atomsas equatorial ligands. Moreover, we can see that g‖ and A‖ values of both speciesof the Cu(II)-poly(dG-dC)·poly(dG-dC) complex are very close to g‖ and A‖values of the Cu(II)-5′-GMP complex (shown as asterisk), even if they do not


match perfectly. This is an indication that both species could be, in principle,similar to the Cu(II)-5′-GMP model system.

To further prove the direct coordination of the copper ion to a nitrogen atomwe measured the Davies ENDOR spectrum of Cu(II)-poly(dG-dC)·poly(dG-dC)complex deuterated at all the exchangeable protons (sample 7) in DMSO-d6/D2O.The Davies ENDOR spectrum shows a peak typical of strongly coupled nitrogen(see Figure 3.6). This is an indication that at least one species is coordinatedwith at least one nitrogen atom in agreement with the g values and the Cu(II)hyperfine parameters obtained from the simulation.

To detect any interaction between the copper center and the 31P nuclei of thephosphate groups, the Mims ENDOR technique has been used. In contrast tothe result of similar experiments on Cu(II)-5′-GMP, no evidence for 31P in closeproximity of the Cu(II) ion has been obtained. We may not completely excludethat 31P is only slightly further away from the Cu(II) ion than in Cu(II)-5′-GMP,but it is most probable that for sterical and electrostatical reasons, in the Cu(II)-poly(dG-dC)·poly(dG-dC) complex the phosphate group is not hydrogen-bondedto a water molecule that coordinates the Cu(II) ion. Indeed, in our complexthe 31P atoms are in a relatively fixed position in the backbone of the nucleotidepolymer, whereas the phosphate group in Cu(II)-5′-GMP complex has larger con-formational freedom.

To get information about equatorial water coordination the Davies ENDORtechnique at Q-band was utilized. The Q-band Davies ENDOR spectra of samples5 and 7 are shown in Figure 3.4.

Figure 3.4: Q-band Davies ENDOR spectra of Cu(II)-poly(dG-dC)·poly(dG-dC) com-plex in DMSO/H2O, sample 5 (thin trace) and of Cu(II)-poly(dG-dC)·poly(dG-dC)complex deuterated at all the exchangeable protons in DMSO-d6/D2O, sample 7 (thicktrace), taken at observer position B (see Figure 3.2.a).


Both spectra are characterized by two peaks centered at the proton Larmorfrequency and that can be assigned to weakly coupled protons. We refrainedfrom simulations of these spectra, since due to the presence of two species, is notpossible to find a unique hyperfine tensor that fits our data. However, we canconclude that the sizeable proton couplings measured for sample 5 do not changesignificantly when the solvent protons and the exchangeable protons of the DNAare completely exchanged by deuterium. This excludes equatorial water coordi-nation. This last result is further supported by comparing the X-band protonHYSCORE spectra of the samples 5 and 7 (see Figure 3.5). Even if the pro-ton signals are slightly different, we do not observe any suppression of the moststrongly coupled protons in the deuterated sample. In contrast, such suppressionhas been observed for the Cu(II)-5′-GMP complex (see Figure 2.4.a, c) where wehave evidence of equatorial water coordination.

Figure 3.5: X-band proton HYSCORE spectra taken at observer position B (seeFigure 3.2.a). (a) Cu(II)-poly(dG-dC)·poly(dG-dC) complex in DMSO/H2O, sample5, τ = 100, 120, 140, 160 ns; (b) Cu(II)-poly(dG-dC)·poly(dG-dC) complex deuteratedat all the exchangeable protons in DMSO-d6/D2O, sample 7, τ = 100, 120, 140, 160,400 ns. The antidiagonal lines are given for ν1H .

In Figure 3.6 the X-band Davies ENDOR spectra of Cu(II)-H8-d5-5′-GMP

complex (4) and of Cu(II)-poly(dG-dC)·poly(dG-dC) complex (sample 7) areshown. Comparing the two spectra we can see that the copper in both complexesis directly coordinated by at least one nitrogen atom, but the hyperfine couplingbetween the copper ion and the nitrogen atom is stronger in the model systemCu(II)-H8-d5-5

′-GMP than in the Cu(II)-poly(dG-dC)·poly(dG-dC) complex. In-deed, the spectrum of Cu(II)-H8-d5-5

′-GMP complex is centered at approximately16 MHz corresponding to a hyperfine coupling of about 32 MHz. Instead, thespectrum of the Cu(II)-poly(dG-dC)·poly(dG-dC) complex is centered at approx-imately 12 MHz corresponding to a hyperfine coupling of about 24 MHz. This isan indication that the overlapping between the copper orbital and the nitrogenorbital is not optimized in the Cu(II)-poly(dG-dC)·poly(dG-dC) complex due to


restricted ligand flexibility of the nucleobases in the context of a double strandedDNA.

Figure 3.6: X-band Davies ENDOR spectra of Cu(II)-poly(dG-dC)·poly(dG-dC) com-plex deuterated at all the exchangeable protons in DMSO-d6/D2O, sample 7, (thintrace) and of Cu(II)-H8-d5-5′-GMP complex, sample 4, (thick trace), taken at observerposition B (see Figure 3.2.a).

In Figure 3.7 the X-band HYSCORE spectrum of the Cu(II)-poly(dG-dC)·poly-(dG-dC) complex (sample 6) in 50% 13C-DMSO and 50% H2O is shown. Thespectrum reveals the same features obtained for the model system Cu(II)-5′-GMP(see Figure 2.6.a). Moreover, the carbon signal can be simulated (see Figure 2.6.b)with the same hyperfine parameters. We can conclude that in the Cu(II)-poly(dG-

Figure 3.7: X-band carbon HYSCORE spectrum of Cu(II)-poly(dG-dC)·poly(dG-dC)complex in 50% 13C-DMSO and 50% H2O, sample 6, taken at observer position B (seeFigure 3.2.a), τ = 100, 120, 140, 160 ns; The antidiagonal line is given for ν13C .

dC)·poly(dG-dC) complex at least one species is coordinated with the DMSOsolvent and the interaction between the copper ion and the DMSO is the samelike in the model system Cu(II)-5′-GMP.


3.4.3 Assignment of the copper species

Up to now we have tried to characterize the properties that can be assignedto both species of the Cu(II)-poly(dG-dC)·poly(dG-dC) complex. In order todistinguish the properties of each single species, we tried to find a molecularmodel that fits with our experimental data and that can help us to get furtherinsight about the structure of Cu(II)-poly(dG-dC)·poly(dG-dC) complex. Thedetermination of the binding model of Cu(II) in DNA has been the subject ofmany studies. Eichhorn and Clark [Eichhorn and Clark, 1965] have postulated onthe basis of their results a model for the copper-B-DNA complex in which the twostrands are held in alignment by formation of a Cu(II) complex between guanineand cytosine residues. Also Zimmer et al. [Zimmer et al., 1971] proposed a similarmodel for the copper-B-DNA complex with the Cu(II) crosslinks involving N7 ofguanine and N3 of cytosine as the most probable coordination sites. Moreover,they proposed the following scheme for Cu(II) base interaction and crosslinkformation at GC base-pair sites in B-DNA. In a first step, the Cu(II) ion bindsexclusively to a guanine residue. Breaking of hydrogen bonds, loosening of base-base interactions and tilt of the bases would then alter the structure for the Cu(II)crosslink between the guanine and the cytosine. The usual anti conformation ofthe deoxyguanosine residue is altered to the syn conformation.

In order to assign the two species of our complex, we propose an analogousmodel for the copper-Z-DNA complex. Indeed, CD measurements have been doneto test whether the DNA is present in the B- or Z-form (see Figure 3.8). The

Figure 3.8: Circular dichroism response versus wavelength for Cu(II)-poly(dG-dC)·poly(dG-dC) complex in water solution (solid line), for poly(dG-dC)·poly(dG-dC)in 50% DMSO and 50% H2O (dashed line), and for Cu(II)-poly(dG-dC)·poly(dG-dC)complex in 50% DMSO and 50% H2O (dotted line). The metal concentration was 20%of DNA bases concentration.

Cu(II)-poly(dG-dC)·poly(dG-dC) complex in water solution presents a B-form


(solid line). The poly(dG-dC)·poly(dG-dC) in 50% aqueous DMSO solution,that contains no copper, can be described in a state in between B- and Z-DNA(dashed line). Addition of 0.2 equiv. of Cu(II) versus DNA bases moves thisequilibrium more towards the Z-form (dotted line).

In our model we propose that one copper species is coordinated exclusively toN7 of guanine (Figure 3.9.a). In Z-DNA the guanine residue in such a complexis in syn conformation. In the other copper species, a copper crosslink betweenthe guanine and the cytosine is proposed, with a corresponding change of thelocal conformation of the guanine residue. The guanine moves from syn to anticonformation (Figure 3.9.b). To test this proposed molecular model, we triedto verify that our experimental results fit with this model. In particular, it wasnecessary to distinguish the properties of each single species, and then verify thatthey are consistent with our model.

Figure 3.9: Schematic representation of a complex binding model of Cu(II) in DNAat a GC base pair. (a) Cu(II) coordinated exclusively to guanine; (b) complexing ofCu(II) between guanine and cytosine moieties. The figures were created with the helpof the programs: B (formerly Biomer, Neill White, 1999), Chem 3D (CambridgeSoft),and DS Visualizer (Accelrys).


3.4.4 Cu(II)-(1-methyl-cytosine)4 model system

Before describing the properties of each copper species, it was necessary to char-acterize the interaction of the copper ion with cytosine. We were not interestedto reach a complete characterization of this model system, but information aboutthe hyperfine couplings between the copper ion and the non-exchangeable pro-tons of the cytosine were required to prove the validity of our proposed molecularmodel.

A sample of Cu(II) triflate and 1-methyl-cytosine deuterated at all the ex-changeable protons in 50% DMSO-d6 and 50% D2O was prepared and the cwEPR and the proton matched HYSCORE spectra were measured. The 1-methyl-cytosine was synthesized at the Institute of Inorganic Chemistry of the Universityof Zurich by Marcel Ziegert using the procedure described by Kistenmacher etal. [Kistenmacher et al., 1979]. The metal concentration was 2 mM and the con-centration of 1-methyl-cytosine was 8 mM. In Figure 3.10.a the frozen solutionX-band EPR spectrum is shown.

Figure 3.10: X-band spectra of Cu(II)-(1-methyl-cytosine)4 system deuterated at allthe exchangeable protons in 50% DMSO-d6 and 50% D2O; (a) cw EPR spectrum takenat 130 K, the arrow indicates the observer positions used for the pulse measurements;(b) proton matched HYSCORE spectrum taken at observer position C, τ = 100, 120,140, 160, 400 ns; the antidiagonal is given for ν1H .

As with Cu(II)-5′-GMP, the cw EPR spectrum of Cu(II)-(1-methyl-cytosine)4

model system reveals features that are typical for Cu(II) complexes with a (dx2−y2)1

ground state. In Figure 3.10.b the X-band proton matched HYSCORE spectrumis shown. The spectrum consists of proton peaks close to the antidiagonal atthe proton Larmor frequency and combination peaks between the proton and thedeuterium frequency. These weakly coupled protons can be exclusively assignedto H5 and H6 (see Figure 3.9) since in our sample there are no other protonsin such close proximity to the copper center. Moreover, comparing the protonsignal of the Cu(II)-(1-methyl-cytosine)4 model complex with the proton signal of


Cu(II)-H8-d5-5′-GMP model complex (Figure 2.4.c), it is evident that the width

of the ridge is smaller in the Cu(II)-(1-methyl-cytosine)4 model complex indicat-ing smaller hyperfine couplings between the copper ion and the H5 and H6 ofcytosine than between the copper ion and H8 of guanine.

3.4.5 Properties of the copper species I

The ”species I” is the one that is characterized by the 63Cu hyperfine tensor (Ax,Ay, Az) = (25, 25, 442) ± 10 MHz and 45% of population. To separate the signalcontribution of species I from the signal contribution of species II we performedthe HYSCORE experiments at a very low magnetic field position (observer po-sition A Figure 3.2.a). It is reasonable to assume that in this magnetic fieldposition the main contribution comes from species I, even if we may have also asmall contribution due to species II. In Figure 3.11.a the low-frequency region ofthe HYSCORE spectrum of Cu(II)-poly(dG-dC)·poly(dG-dC) complex (sample5) is shown. The spectrum is dominated by cross-peaks similar to those ob-served for the Cu(II)-5′-GMP complex (Figure 3.11.b), which have been assignedto a weakly coupled remote nitrogen. This nitrogen signal is not detected in theHYSCORE spectrum of the Cu(II)-poly(dG-dC)·poly(dG-dC) complex taken atobserver position B where it is reasonable to assume that the main signal contri-bution comes from species II.

Figure 3.11: X-band nitrogen HYSCORE spectra (a) of Cu(II)-poly(dG-dC)·poly(dG-dC) complex in DMSO/H2O, sample 5, taken at observer position A(see Figure 3.2.a), τ = 100, 120, 140, 160 ns; (b) of Cu(II)-5′-GMP complex, sample2, measured at a magnetic field position close to g‖, τ = 100, 120, 140, 160, ns. Theantidiagonal lines are given for ν2H and 2ν2H .

In Figure 3.12.a the proton HYSCORE spectrum of Cu(II)-poly(dG-dC)·poly-(dG-dC) complex deuterated at all the exchangeable protons (sample 7) is shown.The spectrum is characterized by a broad ridge close to the antidiagonal at the


proton Larmor frequency. This signal is very similar to the proton signal detectedfor the Cu(II)-H8-d5-5

′-GMP complex measured at a magnetic field position closeto g‖ (Figure 3.12.b), which has been assigned to proton H8 of the guanine.

Figure 3.12: X-band proton HYSCORE spectra (a) of Cu(II)-poly(dG-dC)·poly(dG-dC) complex deuterated at all the exchangeable protons in DMSO-d6/D2O, sample 7,taken at observer position A (see Figure 3.2.a), τ = 100, 120, 140, 160, 400 ns; (b) ofCu(II)-H8-d5-5′-GMP complex, sample 4, measured at a magnetic field position closeto g‖, τ = 100, 120, 140, 160, 400 ns. The antidiagonal line is given for ν1H .

Using these last results we could conclude that the species I, even if slightlydifferent, is rather similar to the model system Cu(II)-5′-GMP where the copperion is directly coordinated to the nitrogen N7 of guanine and where the proton H8is weakly coupled to the copper center. This is in agreement with our proposedmodel where one copper species is exclusively coordinated to the guanine of thedouble-stranded Z-DNA (Figure 3.9.a). The slight difference between the twosystems may be due to lack of interaction between the copper ion and the phos-phate group in the Cu(II)-poly(dG-dC)·poly(dG-dC) complex, and differences inwater coordination.

3.4.6 Properties of the copper species II

We called ”species II” the one that is characterized by the 63Cu hyperfine tensor(Ax, Ay, Az) = (28, 28, 465) ± 10 MHz and 55% of population. To detect theproperties of this species we measured the matched HYSCORE spectrum at ob-server position B (see Figure 3.2.a), since in this position it is reasonable to assumethat the main signal contribution comes from the species II. In Figure 3.13.a-bthe proton matched HYSCORE spectrum of Cu(II)-poly(dG-dC)·poly(dG-dC)complex deuterated at all the exchangeable protons (sample 7) is shown. Thespectrum is characterized by a broad ridge close to the antidiagonal at the protonLarmor frequency (Figure 3.13.a) and by a broad ridge close to the antidiagonal


at two times the proton Larmor frequency (Figure 3.13.b). The presence of a sig-nal near 2ν1H , stemming from double-quantum nuclear coherence, indicates thatthere are at least two non-exchangeable protons weakly coupled to the coppercenter. This is an indication that in species II the copper ion is coordinated withtwo nucleotide bases. By inspection of the experimental HYSCORE spectrum,however, it is not possible to determine if the copper ion is coordinated eitherwith two equal nucleotide bases (two guanine or two cytosine) or with one gua-nine and one cytosine, as proposed in our model. Since our sample is deuteratedat all the exchangeable protons, the only protons close to the copper center thatcan give a contribution at the proton signal of the matched HYSCORE spec-trum are H8 of the guanine and H6 and H5 of the cytosine (see Figure 3.9). Weused the molecular model described before to simulate our experimental matchedHYSCORE spectrum. In our model, for the copper species coordinated with oneguanine and one cytosine some constraints have been imposed. In particular, weimposed a distance between the copper ion and the nitrogen atoms (N7 and N3)of about 2 A and a hydrogen bond between the guanine and the cytosine with adistance of 3 A, as shown in Figure 3.9.b. Using this molecular model it was pos-sible to predict the distance between the copper ion and the protons H5 and H6of the cytosine. We obtained the following distances: for Cu(II)-H5 r = 4.57 A,for Cu(II)-H6 r = 5.37 A. Using the point-dipole approximation we obtained adipolar coupling constant of T = 0.83 MHz, 0.51 MHz, respectively. To evaluatethe isotropic hyperfine coupling we made the following assumptions: since theprotons H5 and H6 are very far away from the copper center, and therefore thereis no significant electron delocalization into these proton nuclei, it is reasonableto assume that the isotropic hyperfine couplings are zero. Indeed, the hyperfinecouplings in Cu(II)-(1-methyl-cytosine)4 (Figure 3.10) are as small as predictedby these considerations. Concerning the hyperfine coupling of the proton H8, weused the parameters aiso = 6.8 MHz and T = 1.5 MHz that have been obtainedfitting the width of the ridge at ν1H of the experimental spectrum. After we havepredicted the hyperfine tensors of all non-exchangeable protons, we used thesevalues to simulate the experimental matched HYSCORE spectrum. The resultsof the simulation are shown in Figure 3.13.c-f. Two different kinds of simulationshave been done. For the hypothesis of copper coordinated with two guanine baseswe simulated the matched HYSCORE spectrum considering two equivalent H8of guanine (Figure 3.13.c-d). Instead, for the hypothesis of copper coordinatedwith one guanine and one cytosine base, we simulated the matched HYSCOREspectrum considering the H8 of guanine and the H5 of cytosine (Figure 3.13.e-f). The same results have been obtained for the simulated spectrum consideringthe H8 of guanine and the H6 of cytosine. Even if we assumed that the mainexperimental signal contribution in a spectrum measured at observer position Bcomes from species II, there is also a small signal contribution due to species I.Moreover, in the simulation we considered only the contributions of the protonsH8, H5, and H6 even if there are other non-exchangeable but very weakly coupledprotons in our sample.


Figure 3.13: X-band proton matched HYSCORE spectra of Cu(II)-poly(dG-dC)·poly(dG-dC) complex deuterated at all the exchangeable protons (sample 7) takenat observer position B (see Figure 3.2.a), τ = 100, 120, 140, 160, 400 ns; (a)-(b) low-and high-frequency region; (c)-(d): low- and high-frequency region of simulated spec-tum considering two equivalent H8 of guanine; (e)-(f): low- and high-frequency regionof simulated spectum considering H8 of guanine and H5 of cytosine. See the text formore details about the simulation. The antidiagonal lines are given for ν1H and 2ν1H .


For that reason we were not able to simulate exactly the width of the ridgesat ν1H and 2ν1H . However, it was possible to come to the following conclusions.In the experimental spectrum the ridge at 2ν1H is spread over 9.3 MHz againstthe 8.30 MHz of the ridge at ν1H . Therefore, the width of the ridge at 2ν1H

exceeds the length of the one at ν1H by about 1 MHz. In the simulated spectrumconsidering the H8 protons of two guanine bases, the width of the ridge at 2ν1H

exceeds the one at ν1H by about 8.3 MHz. Instead, in the simulated spectrumconsidering the H8 of guanine and the H5 of cytosine the width of the ridge at2ν1H exceeds the one at ν1H by about 0.6 MHz, which is much more constistentwith the experimental result. We can conclude that in species II the copper ion iscoordinated with two chemically not equivalent protons with very different hyper-fine couplings which is probably inconsistent with coordination by two guaninebases, certainly inconsistent with coordination by two cytosine bases, but con-sistent with coordination by one guanine and one cytosine base. Moreover, it isunlikely that the signal at twice the proton Larmor frequency can be explained byone or two cytosine bases without coordination to guanine because, first, there isno evidence of a proton signal at twice the proton Larmor frequency in a Cu(II)-(cytosine)4 complex; second, the small hyperfine coupling between the copper ionand the H5 and H6 of cytosine (see Figure 3.10.b) cannot explain the broad ridgesat ν1H and 2ν1H that charaterized the Cu(II)-poly(dG-dC)·poly(dG-dC) complex.These last results are all in agreement with our proposed model where one copperspecies is coordinated with one guanine and one cytosine (Figure 3.9.b).

3.4.7 Structural considerations and conclusions

Even if with the cw and pulse EPR techniques we were not able to obtain acomplete characterization of the structural features of the metal ion bound topoly(dG-dC)·poly(dG-dC), with our study new insights have been obtained. Thecw spectrum of Cu(II)-poly(dG-dC)·poly(dG-dC) complex shows the presence oftwo copper species. Their EPR parameters and relative populations do barelyvary in the range of 0.05-0.2 equiv. of Cu(II) versus DNA bases. Both species arecharacterized by a (dx2−y2)1 ground state with a square-planar symmetry. FromDavies ENDOR, HYSCORE and cw analysis it was possible to assign the twodifferent species. Our experimental data fit very well with a molecular modelanologous to the one proposed by Zimmer et al. In our model, one copper speciesmost probably is coordinated to the N7 of the guanine. This is consistent with ourdata that indicate that the species I is very similar to the Cu(II)-5′-GMP com-plex. Additionally, in our model a copper crosslink between a guanine and a cyto-sine residues is proposed. From the experimental and simulated proton matchedHYSCORE spectra we have evidence of coordination of the copper species II totwo chemically non equivalent non-exchangeable protons that is most consistentwith coordination by one guanine and one cytosine. Furthermore, we do not haveevidence of interaction of both copper species with the phosphorous nucleus ofthe phosphate group, and of equatorial water coordination.


3.4.8 Outlook

In principle, we can not exclude that the presence of the two species could besomehow influenced by the solvent used for cryoprotection, namely DMSO. In thelast part of the thesis we studied the properties of water encapsulated in silica hy-drogels. In these silica hydrogels the major fraction of water is supercooled downto temperatures of at least 198 K, given the opportunity to study the propertiesof biological molecules, like proteins or polynucleotides, below the crystallizationtemperature of water without using cryoprotectants. It would be very interestingto try to encapsulate the Cu(II)-poly(dG-dC)·poly(dG-dC) complexes in such asilica hydrogel and to study their properties in a purely aqueous environment.

Chapter 4

Supercooled water in a silicahydrogel

4.1 Introduction

Water plays a fundamental and ubiquitous role in all aspects of life phenomena.Although water has been the topic of considerable research, its physical proper-ties are not fully understood. Below the melting point at 273 K it is possible tosupercool bulk water down to ∼235 K where it crystallizes [Mishima and Stanley,1998]. It is also possible to rapidly quench small quantities of water to temper-atures below 100 K to form amorphous solid water, i.e. glassy water. Whenheating glassy water, it exhibits a glass-liquid transition at a temperature, Tg, inthe 124-136 K temperature interval [Handa and Klug, 1988, Johari et al., 1987]and then, because of the increased mobility of the molecules, it crystallizes at150 K [Mishima and Stanley, 1998]. There is thus an experimentally inaccessibletemperature region (so-called no man’s land) for bulk supercooled water betweenroughly 150 and 235 K. See Figure 4.1 for an overview of the water behavior atdifferent temperature. The temperature dependence of the viscosity and relax-ation time above the no man’s land region follows a power law that diverges ata critical temperature Ts ≈ 228 K [Speedy and Angell, 1976] in the inaccessibletemperature region.

Confined water is of importance because it provides a means of entering intothe inaccessible temperature region for bulk supercooled water [Bergman andSwenson, 2000, Faraone et al., 2004]. In given geometrical confinements wa-ter molecules are unable to form a crystalline structure. Thus, water remainsfluid and, like any other supercooled liquid, becomes increasingly viscous as thetemperature approaches Tg. Geometric confinement is expected to induce mod-ifications on both the structural and dynamic properties of bulk water. Studieson water in confined geometries are of relevance also for biological implications,since water confined at protein surfaces has been suggested to be important toprotein stability and function [Rupley and Careri, 1991, Santangelo et al., 2003].

Sol-gel techniques for the preparation of silica hydrogels offer a new way of

65

66 Chap. 4 - Supercooled water in a silica hydrogel

Figure 4.1: Overview of the water behavior at different temperatures.

obtaining water confined in a three-dimensional disordered and hydrophilic ma-trix that provides an environment for water similar to the one near protein sur-faces. Studies on water confined in silica hydrogels, using optical absorptionspectroscopy in the near infrared region (NIR), have provided information on thethermodynamic equilibrium between different classes of water molecules [Cupaneet al., 2002]. In particular, in a three-months aged ”dry” hydrogel (Figure 4.2.a),a fraction (∼5%) of weakly bonded water molecules is always present, even at5 K. In contrast, in the case of a one-day aged ”wet” hydrogel (Figure 4.2.b),crystallization of water occurs between 265 and 250 K thus preventing investi-gation with optical spectroscopy at lower temperatures; moreover, saturation ofthe signal does not allow evaluating the fraction of water that crystallizes in wethydrogel.

To characterize the structural and dynamic properties of water confined insilica hydrogels, we have encapsulated nitroxide TEMPO molecules in these sil-ica hydrogels and we have used EPR and differential scanning calorimetry (DSC)techniques. Indeed, the cw EPR spectrum of nitroxide molecules is highly sen-sitive to the mobility of the spin probes, thus information about the local envi-ronment of the probes can be obtained. On the other hand, DSC measurementsprovide quantitative and qualitative information about physical and chemicalchanges that involve endothermic or exothermic processes like glass transition,crystallization and melting transitions.

4.2 Sol-Gel encapsulation

At the beginning of 1990’s, Zink and coworkers [Dave et al., 1994, Ellerby et al.,1992] have developed a technique that enables one to encapsulate water moleculesin optically transparent, porous silica glass matrices. This sol-gel process is es-sentially composed by two steps that make use of alkoxide precursors such as the

4.2 - Sol-Gel encapsulation 67

Figure 4.2: (a) Absorption spectra (∼1.45 µm water overtone band) of the 3 monthsaged dry hydrogel, at selected temperatures, in the range 325-5 K. The spectrum of iceIh is reported for comparison. (b) Absorption spectra (∼1.45 µm water overtone band)of one day aged wet hydrogel at 295 K, of liquid water at 295 K, of wet hydrogel at160 K, and of ice Ih at 233 K. The arrows indicate the wavelength of 1.41 µm, where”weakly bonded” water molecules absorb. (Modified from [Cupane et al., 2002]).

tetramethyl orthosilicate (TMOS):

A first reaction — the hydrolysis — leads, in the presence of water, to the for-mation of an Si-OH group and a methanol molecule (CH3OH):

A second reaction — the polycondensation — involves Si-OH groups of two dis-tinct molecules and leads to the formation of an Si-O-Si bond between them, pluswater:

The Si atom in a TMOS molecule is bound to four -OCH3 groups, each of themcan give rise to a hydrolysis reaction. It is clear that the polycondensation reac-tion can go on and lead to a Si-O-Si network.


The resulting material is optically transparent and has a structure similarto amorphous silica, though it is porous and less resistant than silica at hightemperatures. It consists of an amorphous phase (the Si-O-Si bond network) anda liquid phase (essentially water) that remains encapsulated in the interior of thepores.

In order to encapsulate nitroxide molecules in this glassy matrix, one has tomix the TMOS solution with the nitroxide solution. Nitroxide molecules (togetherwith some solvent) remain encapsulated inside the pores of the matrix during theformation of gel. It has been shown that pore dimensions are of the order ofseveral tens of angstrom [Dave et al., 1994].


In this project we measured the EPR spectra and DSC scans of ”wet” silicahydrogel having an hydration level, h, equal to 4.5 (h is defined as the massof water present in the gel divided by the mass of nonaqueous material) andcontaining the nitroxide probe TEMPO. Solutions of TEMPO in pure water andin a glycerol/water mixture (80% glycerol/20% water, vol/vol) have also beenprepared and investigated with EPR as reference samples.

4.3.1 Silica hydrogel for EPR measurements

The nitroxide radical 2,2,6,6-tetramethylpiperidine-1-oxyl (TEMPO) and the alko-xide tetramethyl orthosilicate (TMOS) were purchased from Aldrich and Merck,respectively, and used without further purification. A solution containing 75%vol/vol TMOS, 21% H2O (Millipore purified), and 4% 0.05 mM HCl was sonicatedfor ∼20 min and diluted with an equal volume of a 400 µM solution of TEMPO inwater. Approximately 15 mg of the resulting mixture were poured into an EPRcapillary tube (I.D. ∼700 µm) and left to age for about one day until the samplehad solidified macroscopically. The gel was then purged for several minutes withdry nitrogen until its hydration level had dropped to h = 4.5. The final mass ofwater in this sample was about 10 mg. Measurements on samples with differenthydration levels between h = 4 and h = 5, and the same aging time (∼1 day),have also been performed and yielded essentially the same results. Thus, in thefollowing, we will call ”wet” all hydrogels having a hydration level h = 4.5 ± 0.5and an aging time of ∼1 day.

A water sample and a glycerol/water mixture (80% glycerol/20% water, vol/vol)both containing 200 µM TEMPO were also prepared starting from a 2 mM stocksolution of TEMPO in water and were poured into a quartz EPR tube havingI.D. of 500 µm and 2 mm, respectively. All aqueous solutions were prepared withMillipore ultrapure water. Reagent grade glycerol was purchased from ABCRGmbH & Co. KG.


4.3.2 Silica hydrogel for DSC measurements

Wet hydrogels used for DSC measurements have been prepared following the sameprocedure used for EPR experiments but with two different sample geometries.In one case, after mixing the reagents as described previously, the resulting liquidwas poured into a standard stainless steel pan for DSC measurements. Becauseof the large surface-to-volume ratio, the water evaporation rate is much higherthan in the case of the sample prepared in a capillary tube. After a few minutesfrom preparation, a hydration level h = 4.4 was reached; the DSC pan was thenimmediately sealed and left to age for about one day before performing the DSCmeasurements. This sample contained about 34 mg of water. In order to be ableto compare more directly the EPR and DSC experiments, DSC scans were alsoperformed on a wet hydrogel prepared into a capillary tube identical to that usedfor EPR measurements. In particular, after pouring the sample into the capillarytube, it was left to age for about one day until it had solidified macroscopically.The sample was then purged for several minutes with dry nitrogen until its hy-dration level had dropped to h = 4.1. At this point, the bottom part (about5 mm) of the capillary tube containing the silica hydrogel was cut, put insidea DSC stainless steel pan, and sealed. The DSC measurements were performedright after sealing the pan. The final mass of water in this sample was about 1.6mg.

4.3.3 EPR spectroscopy

CW EPR measurements at X-band were carried out using a Bruker ELEXSYSE500 spectrometer equipped with a Bruker super-high-Q ER-4122SHQ cavity.Cooling of the sample was performed with a liquid nitrogen dewar vessel and aBruker ER-4111VT temperature control unit. Measurement parameters were asfollows: microwave frequency ≈ 9.5 GHz and modulation frequency = 100 kHz.To avoid signal distortion due to excessive magnetic field modulation, at eachtemperature, the modulation amplitude was chosen to be about one third of theEPR linewidth. A saturation study on the samples allowed us to ensure that, ateach temperature, the microwave power incident on the sample did not inducesaturation.

4.3.4 DSC spectroscopy

DSC measurements were performed at the Department of Physics of University ofPalermo with a PerkinElmer PYRIS Diamond differential scanning calorimeterequipped with a liquid nitrogen cooling system that allows operation down toabout 95 K. Both the temperature and heat-flow scale of the DSC instrumentwere calibrated using a high purity indium reference sample. Measurements havebeen performed on samples sealed inside standard DSC stainless steel pans againstan empty pan placed in the reference vessel of the instrument. The temperaturewas lowered from ambient to 95 K at 10 K/min, held at 95 K for 5 minutes,


and then raised at 10 K/min toward room temperature while monitoring for anyexothermic or endothermic process. Scans at 3 K/min were also performed. Thesmall differences in transition temperatures between the faster and the slowerscans do not affect interpretation of the data.

4.3.5 Theory and analysis of EPR data

It was already mentioned in Chapter 1 that the lineshapes of the cw EPR spectraof nitroxide probes at X-band frequencies are dominated by the hyperfine cou-pling of the unpaired electron of the N-O group to the 14N nucleus with nuclearspin I = 1. For a molecule at a given orientation this hyperfine coupling causesa splitting of the EPR line into three lines corresponding to magnetic quantumnumbers mI = -1, 0, and 1 of the 14N nucleus. The temperature dependence ofEPR spectra of nitroxide probes is due to changes of the probe’s motion thatcan be characterized by its rotational correlation time, τc. The slow down ofthe rotational correlation time τc as the temperature is lowered (or the viscosityincreased) can be tracked by monitoring the separation, 2A

′zz, between the outer

lines of the nitroxide EPR spectrum (see Figure 1.6). Apart from the sensitivityof the EPR technique to the spin probe motion, the different EPR parametersassociated with a spin probe show a solvent dependence. In particular, the hy-perfine coupling component perpendicular to the plane of the ring structure ofTEMPO, Azz, the isotropic hyperfine coupling, aiso [Griffith et al., 1974] and thegx element of the g tensor [Kawamura et al., 1967], are sensitive to the polarityof the local environment [Kurad et al., 2003, Owenius et al., 2001]. Thus, bycomparing the EPR parameters obtained for TEMPO in wet hydrogel and inbulk water, it is possible to get information about the relative polarity of oursamples. Since with the X-band spectra we could not have accurate estimates ofthe gx and Azz parameters, we restrict ourselves to comparing aiso parameters toobtain information about the polarity of our samples.

In the case of spherical molecules of radius r dissolved in an isotropic solventat temperature T , the rotational correlation time is related to the solvent viscosityη through the Debye-Stokes-Einstein equation

τc =4πηr3

3kBT(4.1)

where kB is the Boltzmann constant. Since the specific interactions of the probemolecule with the solvent influence the dynamics of the probe [Zager and Freed,1982], Eq. 4.1 cannot be used to obtain accurate values of the macroscopic solventviscosity. However, Eq. 4.1 can still be used to evaluate the relative viscosity ofwater confined in wet hydrogels with respect to that of bulk water or an 80%glycerol/H2O mixture by comparing the temperature dependence of the rotationalcorrelation times.


Fast motion regime

A general theory that relates τc to the spectral width of the EPR lines has beenderived in the past by Kivelson [Kivelson, 1960]. In the case of nitroxide probesin fast motion, the equations derived by Kivelson can be combined to obtain thefollowing expression

τc =π√

3

b

[b

8− 4∆γB0

15

]−1

∆ν0

[∆ν−1

∆ν0

− 1

](4.2)

where b is a constant that depends on the anisotropic hyperfine coupling, ∆γis a constant that depends on g anisotropy, B0 is the laboratory magnetic field,and ∆ν0 and ∆ν−1 are the peak-to-peak widths (in Hz) corresponding to themid- and high-field lines, respectively. For convenience of measurement the ratio∆ν−1/∆ν0 can be replaced by (h0/h−1)

1/2, where h0 and h−1 are the heights ofthe mid- and high-field lines of the first-derivative absorption spectrum

τc =π√

3

b

[b

8− 4∆γB0

15

]−1g0µB

hW0

[√h0

h−1

− 1

](4.3)

= kW0

[√h0

h−1

− 1

]where the mid-field peak-to-peak width, ∆ν0, has been converted to peak-to-peakwidth in magnetic field units, W0, by the factor g0µBh−1, where µB=βe is theBohr magneton, h is the Planck constant, and g0 is the isotropic g-factor

g0 =1

3(gx + gy + gz) (4.4)

The parameters b and ∆γ are defined by the following equations

b =4π

3

[Azz −

Axx + Ayy

2

](4.5)

∆γ =2πµB

h

[gz −

gx + gy

2

](4.6)

In order to estimate τc in the case of TEMPO in wet silica hydrogel we haveused the following values for the parameters appearing in the above equations:(gx, gy, gz) = (2.0090, 2.0061, 2.0021), (Axx, Ayy, Azz) = (22.2, 22.2, 106.5) MHzand B0 = 3399 Gauss; g values are those found by others for TEMPO in a glassyaqueous solvent [Hu et al., 2008], Azz has been estimated from the splitting of thespectrum at the lowest temperature investigated, while, for Axx and Ayy it hasbeen assumed thus Axx = Ayy = 0.5(3aiso-Azz), and aiso has been estimated from


the splitting of the spectrum of TEMPO in water at room temperature. Usingthese values one obtains

k =π√

3

b

[b

8− 4∆γB0

15

]−1g0µB

h= 4.9× 10−10 s

G(4.7)

Analogous calculations have been done also for TEMPO in water and in 80%glycerol/H2O and very similar values for the constant k were obtained.

Eq. 4.3 assumes an isotropic rotational diffusion tensor, characterized by thecorrelation time τc. This is appropriate for the almost spherical probe TEMPO,as long as specific interactions, such as hydrogen bonding to the N-O group, arenegligible. However, starting from the theory developed by Kivelson it is possibleto derive two alternative equations for τc, which depend also on the intensity,h+1, of the low-field line. In particular

τc =π√

3

b

[4∆γB0

15

]−1g0µB

hW0

1

2

[√h0

h+1

−

√h0

h−1

](4.8)

τc =π√

3

b

[b

8

]−1g0µB

hW0

1

2

[√h0

h+1

+

√h0

h−1

− 2

](4.9)

If the hypothesis of isotropic rotation is correct, Eqs. 4.3, 4.8, and 4.9 shouldall give the same rotational correlation time (within experimental errors). Inthe case of TEMPO in wet silica hydrogel at 293 K, we obtain τc = 6.9×10−11

s using Eq. 4.3, and 5.2×10−11 s and 8.6×10−11 s using Eq. 4.8 and 4.9, re-spectively. The deviations between the three values are small on the logarithmicscale used for discussing temperature dependence. Moreover, good agreementhas been obtained for all the temperatures in the fast motion regime, and for allthe samples investigated. This indicates that the rotational motion of the probemolecules is similar in bulk and in confined water and can be characterized to agood approximation by an isotropic rotational diffusion tensor.

A further assumption of Eq. 4.3 is that the intrinsic lineshapes of the spinprobe EPR spectrum are Lorentzian. In fact, due to the presence of unresolvedproton hyperfine structure, an inhomogeneous (typically Gaussian) broadening isalso present [Marsh, 1989]. In order to further test the validity of the isotropichypothesis and investigate the effect of a Gaussian broadening on the correlationtime calculations, we have simulated the experimental spectra using the EasySpin[Stoll and Schweiger, 2006] function ”chili”, which is based on the stochastic Liou-ville equation (SLE). The simulations assumed an isotropic rotational diffusiontensor. Moreover, in order to reproduce the experimental line shapes, both aGaussian and a Lorentzian contribution to the total linewidth had to be assumed.The agreement between the simulations and the experimental data was good forall three different samples investigated (TEMPO in wet hydrogel, in water, andin 80% glycerol/H2O mixture), and the τc values obtained from the simulations


were very similar to the τc values obtained using Eq. 4.3. The similarity of the τc

values between SLE fits and the Kivelson formula prove that, despite the inhomo-geneous Gaussian broadening in our experimental spectra, the Kivelson formulais adequate for calculating rotational correlation times.

Slow motion regime

In the slow motion regime, we have estimated τc by using the correlation betweenA

′zz and τc. Assuming isotropic Brownian diffusion motion, spectra for 121 dif-

ferent values of τc between 10−11 and 10−5 s were simulated using software fromthe Freed group [Schneider and Freed, 1989] and a spline interpolation of the A

′zz

values was performed (Figure 4.3). Experimental A′zz values were converted to

correlation times τc by inverting the functional dependence A′zz(τc) defined by the

interpolating splines.

Figure 4.3: Dependence of the outer extrema splitting 2A′zz on the rotational corre-

lation time (semi logarithmic plot).

4.3.6 Theory and DSC data analysis

Differential scanning calorimetry measures the temperatures and the heat flowsassociated with transitions in materials. In particular, a differential scanningcalorimeter measures the heat flow into or out of a sample relative to a refer-ence with a linear temperature ramp. These measurements provide quantitativeand qualitative information about physical and chemical changes that involveendothermic or exothermic process, or changes in heat capacity. For example,when we heat a material and it has just gone through the glass transition, thematerial has a higher heat capacity. With DSC we can measure this change ofheat capacity which is apparent as a step in the plot and determine the glass tran-sition temperature. In a crystallization transition, when the material becomes acrystal, it gives off heat, with a consequent drop of heat flow. This drop is seen


as a pronounced dip in the plot of heat flow versus temperature. Instead, in amelting transition, the material absorbs heat in order to melt the crystal. Thisextra heat flow during melting shows up as a pronounced peak in the DSC plot.The crystallization and melting temperatures of water encapsulated in our wethydrogels were directly determined from the DSC scans. The fraction of watermolecules that crystallizes/melts in wet hydrogels was obtained by dividing thearea of the crystallization/melting peaks by the water latent heat of melting,assuming that water encapsulated in our silica hydrogels is characterized by alatent heat for melting very close to that of bulk water.


The cw EPR spectra at different temperatures of TEMPO in wet hydrogel, water,and 80% glycerol/H2O are shown in Figure 4.4, panels a, b, and c respectively.The spectra of TEMPO in wet hydrogel (Figure 4.4.a) between room temperatureand about 198 K are typical of nitroxides in the fast motion regime. This provesthat the macroscopically solid sample contains water that can be supercooledat least down to 198 K (-75 C). Decreasing the temperature, the dynamics ofTEMPO molecules becomes slower and the spectra between 195 K and 101 Kare typical of probe molecules in a rigid environment. When TEMPO is dis-solved in pure water (Figure 4.4.b) fast motion regime spectra are observed downto 255 K; at lower temperatures (between 255 and 240 K) water freezes and theTEMPO molecules tend to precipitate out of the crystalline lattice formed by wa-ter molecules. The cw EPR spectrum is now characterized by one broad peak dueto exchange and dipolar interactions between the aggregated TEMPO molecules.Such behavior (well known for solutions of TEMPO in bulk water) is not ob-served for TEMPO encapsulated in our wet hydrogel: the TEMPO molecules donot precipitate as they do in pure water. Nevertheless, from just the EPR data,we cannot exclude the formation of ice in wet hydrogels. We could either have amixture of ice and liquid water, with the spin probes in the liquid fraction, or wemight have a measurable fraction of the spin probes as impurities (defects) in theice crystal, i.e. a deformed ice lattice around the probe. The second hypothesisis less probable because, if an ice crystal forms inside a pore, it is reasonable tosuppose that TEMPO molecules would tend to precipitate out of the crystallinephase as is observed in the case of bulk water. Moreover, the shape of the cwEPR spectra above 200 K shows no evidence of TEMPO molecules in a rigid(ice-like) environment. Hence, we can conclude that TEMPO molecules are sol-vated in liquid water. The results further indicate that water molecules aroundTEMPO form a three-dimensionally continuous liquid phase. Indeed, in the caseof the formation of a two-dimensional film around TEMPO we would expect spe-cific directional interactions of the spin probes with the silica matrix resulting inan increase of the rotational diffusion anisotropy and a marked slowdown of therotational diffusion time, which we do not observe.

Another feature of the wet silica hydrogel is revealed by comparing its cw


Figure 4.4: X-band band EPR spectra of TEMPO in (a) a wet hydrogel (having a hy-dration level h = 4.5), (b) bulk water, (c) a glycerol/water mixture (80% glycerol/20%water, vol/vol) at different temperatures. The extreme outer peak separation, 2A

′zz, is

indicated in panel c in the case of TEMPO in 80% glycerol/H2O at 220 K.

EPR spectrum at a given temperature (Figure 4.4.a) with the spectra, at thesame temperature, of TEMPO in 80% glycerol/H2O (Figure 4.4.c) and in watersolution (Figure 4.4.b). At 220 K, while TEMPO molecules in the hydrogel arestill in the fast motion regime, the TEMPO molecules in 80% glycerol/H2O arealready close to the rigid limit; this means that the viscosity of water within wetsilica hydrogels is lower than that of an 80% glycerol/H2O mixture at the sametemperature. Conversely, at 293 K, the spectrum of TEMPO molecules in watersolution corresponds to motion faster than in the hydrogel; this means that theviscosity of water within wet silica hydrogels is higher than that of a water solu-tion at the same temperature.

A simple method for characterizing the nitroxide probe dynamics is to plotthe extreme separation of the EPR outer lines (2A

′zz) as a function of temperature

(Figure 4.5). In our wet hydrogel, the behavior of 2A′zz is similar to that observed

in a large number of synthetic polymers and low-molecular weight glass-formers[Cameron, 1989, Svajdlenkova et al., 2008, Veksli et al., 2000]. Usually a charac-teristic temperature T50G is identified, which corresponds to the temperature at


which 2A′zz = 50 Gauss. If the probe size matches a characteristic length of the

glass-forming matrix, the temperature T50G corresponds to a ”high frequency”glass transition temperature since it corresponds to the temperature at whichthe relaxation time crosses the window of ∼10−8 s; such a temperature is gener-ally expected to be higher than, but correlated with, Tg (the calorimetric glasstransition temperature). For probes that are much smaller than the molecules ofthe glass-forming matrix, T50G may be lower than Tg. In the case at hand, theprobe has a diameter of ∼6 A larger than that of a water molecule, so that thefirst case applies. From Figure 4.5, it is observed that 2A

′zz undergoes a sharp

transition between 200 and 195 K. We find T50G = 197 K, indicating that waterencapsulated in silica hydrogels can be supercooled at least down to about 198K.

Figure 4.5: Temperature dependence of the outer peak separation (2A′zz) character-

izing the EPR spectra of TEMPO in a wet hydrogel (having a hydration level h = 4.5);data acquired on both lowering the temperature (circles) and raising the temperature(asterisks) are reported. The line is a reference at 197 K.

Although the EPR experiments have been carried out cooling the sample fromroom temperature to 101 K, spectra at several selected temperatures have beenrecorded during a heating cycle after the sample had been brought down to 101 K.These data points (shown as asterisks in Figure 4.5) do not deviate significantlyfrom the dependence observed during cooling, which suggests that no hysteresisis present. Further experiments showed that T50G is the same during the heatingand the cooling cycle (data not shown) thus excluding the presence of hysteresis.In contrast, for TEMPO in bulk water, while we were able to supercool the samplebelow its freezing point (down to 255 K), evidence of a liquid phase on heatingwas observed only when the sample was brought back above the water meltingpoint (∼273 K).


In the fast motion regime, 2A′zz approaches 2aiso. In the case of TEMPO in

silica hydrogel we obtained aiso = 49.9 MHz, while in the case of TEMPO inwater aiso = 50.3 MHz. The difference of the two values does not exceed ex-perimental error. This is an indication that the polarity of the local TEMPOenvironment is similar in bulk water and in wet hydrogel, and it thus confirmsthat the probe in wet hydrogel, at least at room temperature, is solvated by water.

Figure 4.6 shows plots of the correlation time τc for TEMPO molecules in wethydrogel and in 80% glycerol/H2O as a function of the inverse temperature; thebehavior of TEMPO in water solution is also reported for comparison. For allsamples, τc values increase by decreasing the temperature; a saturating behavior isobserved at temperatures well below T50G where free rotational motion no longeroccurs on the EPR timescale and the spectral lineshapes may be dominated bylibrational motion [Dzuba, 2000] or reorganization of the solvent cage [Kirilinaet al., 2004]. The values of τc obtained by fitting lineshapes with a model of

Figure 4.6: Temperature dependence of the rotational correlation time, τc, of TEMPOin wet hydrogel (circles), in a glycerol/water mixture (80% glycerol/20% water, vol/vol)(squares) and in bulk water (triangles). The filled symbols correspond to the fast motionregime, the open symbols correspond to the slow motion regime.

isotropic rotational diffusion thus do not have a clear physical interpretation attemperatures well below T50G.

Although in the whole temperature range the correlation times of both sam-ples have a similar behavior, we could observe that the dynamics of TEMPOmolecules in silica hydrogels is always faster than that of TEMPO in a 80%


glycerol/H2O mixture; moreover, in the temperature interval of 290-255 K, thedynamics of TEMPO molecules in silica hydrogels is always slower than that ofTEMPO in water solution. We may thus conclude that water confined in wethydrogels is less viscous than an 80% glycerol/H2O mixture but more viscousthan bulk water. The correlation times and viscosity of water confined in our wetsilica hydrogels (having a hydration level h = 4.5) are larger than those for purewater (about a factor of 10 at 290 K), in agreement with results from dielectricrelaxation spectroscopy and quasi-elastic neutron scattering on analogous silicahydrogels having a lower hydration level [Cammarata et al., 2003, Schiro et al.,2008].

The glass transition regime of supercooled fluids is characterized by a numberof typical kinetic phenomena. On cooling, the correlation time can increase ina non-Arrhenius way by many orders of magnitude in a narrow temperaturerange. The deviation from thermally activated dynamics is called fragility andtypically the temperature dependence of the correlation time can be describedusing a Vogel-Fulcher-Tamman (VFT) law. In order to evaluate the fragilityof water confined in wet silica hydrogels, we have tried to fit the data relativeto the fast motion regime in terms of either an Arrhenius law or a cooperativeVFT law. Values of the parameters obtained by fitting in terms of the Arrheniusexpression τc = τ0·exp(∆H/RT) are shown in Table 4.1. A pre-exponential factorof 10−19 s, obtained for TEMPO in 80% glycerol/H2O, appears quite unphysical;moreover, although a linear fit agrees reasonably well with both the experimentaldata sets in the rather limited temperature interval available, the extrapolationsto low temperatures give values of T100s (i.e. the temperature at which thecorrelation time reaches the value of 100 s, corresponding to the calorimetricglass transition) of 65 K and 125 K for TEMPO in wet hydrogel and in 80%glycerol/H2O, respectively, much lower than the calorimetrically measured valuesof 120 K (see below) and 180 K [Chen et al., 2006a].

Table 4.1: Values of the Arrhenius and VFT parameters obtained by fitting the datain the fast motion regime.

Arrhenius VFTlog(τ0/s) ∆H log(τ0/s) D T0

sample (kJ/mol) (K)wet silica hydrogel -14 20 -12.14 10.13 91.580% glycerol/H2O -19 52 -13.85 9.21 143.7

For the above reasons we performed fits in terms of the VFT expression τc =τ0·exp[DT0/(T-T0)] imposing a τc of 100 s at the glass transition temperature;values of the VFT parameters are also shown in Table 4.1. Figure 4.7 reportsan ”Angell plot” [Bohmer et al., 1993] of our experimental data; the resultingVFT fits are reported as continuous lines and give excellent agreement with the


experimental data indicating that both water confined in wet silica hydrogel andan 80% glycerol/H2O mixture behave as fragile glass-formers [Cook et al., 1994].

Figure 4.7: Angell plots for TEMPO in wet hydrogel (circles) and in a glycerol/watermixture (80% glycerol/20% water, vol/vol) (squares). The solid lines are best fits to aVFT law.

The fragility parameter [Bohmer et al., 1993] m defined as

m =dlog(τc)

d(Tg/T )|T=Tg (4.10)

can give a measure of the fragility and how large is the deviation from the Ar-rhenius law. From the extrapolation of the VFT behavior down to the glasstransition temperature region, we have obtained a fragility parameter m = 60for water in silica hydrogel and m = 79 for 80% glycerol/H2O respectively. Wenote, however, that for water confined in wet silica hydrogel, we cannot excludethe onset of a fragile-to-strong transition [Ito et al., 1999, Mallamace et al., 2006]at temperatures lower than 200 K. Further experiments with techniques ableto probe the dynamics of water in the time window from 100 ns to 100 s, suchas dielectric spectroscopy [Capaccioli et al., 2007], are needed to clarify this point.

In Figure 4.8.a we show the DSC scans of TEMPO in a wet hydrogel prepareddirectly into a steel pan. On cooling the sample, a crystallization peak is seen atabout 240 K. This peak corresponds to the freezing of about 25% of the samplewater content; the remaining water does not crystallize at least down to 95 K. Onheating the sample, a glass transition is observed at about 120 K (see left inset).A similar result has been recently obtained by the Oguni group using adiabatic


calorimetry [Oguni et al., 2007]. At higher temperatures, two melting peaks areseen: a small one at about 175 K and a large and broad one at about 255 K. Thefirst melting peak (see right inset) is preceded by an exothermic peak at about168 K likely due to crystallization of a small fraction of supercooled water. Themelting peak at 255 K is spread over a relatively large temperature interval. Weinterpret this spread as a result of the pore size distribution of the silica matrix;indeed, due to surface effects, the melting point depends on pore size [Alcoutlabiand McKenna, 2005]. Consistently, the amount of water that, on re-heating, meltsat about 255 K, corresponds to the amount that had crystallized, on cooling, atabout 240 K.

In Figure 4.8.b we show the DSC scans of TEMPO in a wet hydrogel preparedinto a capillary tube and then sealed inside a steel pan. The results are closelysimilar to the previous ones. On cooling the sample, a crystallization peak isseen at about 236 K and it is due to the freezing of a fraction of water (about9%). We note that the capillary tube facilitates supercooling of water inside oursamples; indeed, it is possible to supercool a larger fraction of water (91% ver-sus 75%) in the capillary tube sample with respect to that prepared in the DSCpan. The remaining water does not crystallize at least down to 95 K. On heatingthe sample, no glass transition is clearly detected. We attribute this failure toinsufficient sensitivity for such a low amount of water; indeed, the capillary tubesample contains only 1.6 mg of water, as compared to 34 mg of the steel pansample. A melting peak is observed at about 175 K, again preceded by a smallcrystallization peak at about 165 K (see right inset). A second much broadermelting peak is observed at about 246 K. As in the steel pan sample, the amountof water that melts (at about 246 K) on re-heating corresponds to the amountthat had crystallized (at about 236 K) on cooling.

Combining the EPR results (Figure 4.5) and the DSC results (Figure 4.8)we could reach a characterization of the structural and dynamic properties ofwater confined in wet silica hydrogels. From the DSC scans we have seen that allwater encapsulated in our silica hydrogels remains liquid from room temperaturedown to 236 K. This is seen also from the cw EPR spectra (Figure 4.5) showingthat, between room temperature and 236 K, TEMPO molecules are in the fastmotion regime: water molecules around the probe are in liquid state. At about236 K, only a minor fraction of the total mass of water contained in our samplefreezes while the large majority is still in a supercooled state at least down to95 K. From the EPR data we see that TEMPO molecules below 236 K are stillin the fast motion regime down to 198 K. This means that TEMPO moleculesare solvated by the fraction of water that does not crystallize, confirming ourprevious hypothesis that the hydrogel contains both ice and supercooled water.Not surprisingly, the TEMPO molecules are excluded from the crystalline phaseand are dissolved only in the liquid phase. Although from the DSC data wehave evidence that the majority of water in our hydrogels does not crystallize atleast down to 95 K, EPR data reveal that a sharp transition from the fast to the


Figure 4.8: DSC scans of (a) a wet hydrogel (h = 4.4) containing TEMPO andprepared directly in a DSC steel pan, this sample contains 34 mg of water; (b) a wethydrogel (h = 4.1) containing TEMPO and prepared inside a capillary tube that hasbeen sealed inside a steel pan (see text for details), this sample contains 1.6 mg of water.Insets are magnified plots of regions in the heating cycle where glass transitions (leftinsets) or weak crystallization/melting peaks (right insets) are observed or expected.


slow motion regime occurs at about 197 K (Figure 4.5). At temperatures slightlylower than 197 K, TEMPO molecules are immobilized with respect to the EPRtimescale. Apparently, the rotational correlation time of TEMPO molecules witha diameter of 6 A drops below 10−8 s at a temperature that is about 77 K higherthan the DSC glass transition temperature of the solvent (120 K). The cw EPRspectra below about 190 K are typical of probe molecules in a rigid environment(Figure 4.4.a) as evidenced by the sharp transition of the 2A

′zz parameter at

approximately 197 K (Figure 4.5). The glass transition is hardly seen in the DSCscans of wet hydrogels prepared in capillary tubes (Figure 4.8.b) whereas it isclearly observed in the DSC scans of the sample prepared directly into the steelpan (Figure 4.8.a). Since the samples are practically the same (the only differenceis the presence of the capillary tube), we assume that the glass transition takesplace at the same temperature and is missed in the DSC scans of the smallersample for lack of sensitivity. The solvated TEMPO molecules become mobilewith respect to the EPR timescale at temperatures higher than about 200 Kwithout any significant hysteresis between cooling and heating; on re-heating atabout 246 K all the water encapsulated in wet silica hydrogels is again in theliquid state.

4.5 Conclusions

Two different techniques, EPR and DSC, have been used to characterize thestructural and dynamic properties of water confined in wet silica hydrogels. Thesemacroscopically solid samples prepared through a sol-gel protocol have a hydra-tion level much higher than most samples used for studies on the dynamics ofwater in confinement [Bruni et al., 1998, Cammarata et al., 2003, Chen et al.,2006b]. In our samples, TEMPO spin probes and water molecules are confined in-side silica matrix pores. In previous work, using optical absorption spectroscopyin the near infrared region, we found water crystallization in wet hydrogels below265 K. The new results reveal that, in fact, only a small fraction of water confinedin our matrix freezes, and the rest is supercooled at least down to 95 K; at thistemperature (lower than the glass transition temperature) the sample containsan heterogeneous mixture of ice and amorphous solid water. Probably, as thetemperature is decreased below 236 K, ice nanocrystals form inside the pores buta substantial fraction (between 75% and 90%) of water remains liquid. Therefore,it is possible with such silica hydrogels to access the ”no man’s land” for bulkwater, i.e. the temperature region between roughly 150 and 235 K. It should thusbe possible, using different spectroscopic techniques like Broad-Band Dielectricspectroscopy, NMR, FTIR, and Quasi-Elastic Neutron scattering, to study thestructural and dynamic properties of water in this region. The samples exhibit aglass transition at approximately 120 K. Furthermore, we observed that TEMPOmolecules encapsulated in our silica hydrogels are always dissolved in the waterliquid phase and are in the fast motion regime with respect to EPR timescalesdown to 198 K. TEMPO molecules do not phase separate, upon immobilization,

4.5 - Conclusions 83

from water in wet hydrogels, they rather remain solvated. This is in contrast tothe behavior observed in pure water, but characteristic for a glass-former. Thesize of amorphous or liquid domains exceeds the probe size of 6 A. Larger EPRprobes could be used to obtain an upper limit for the characteristic domain size. Apossible application of the above results is the encapsulation of metallo-protein orDNA water solutions in silica hydrogels; without using cryoprotectants (like glyc-erol or DMSO), one could thus study the dynamics of metallo-proteins by EPRspectroscopy below the crystallization temperature of water or the coordinationbehaviour of DNA without influencing DNA conformation by a cryoprotectant.

Chapter 5

Summary and Outlook

In this thesis a diverse arsenal of electron paramagnetic resonance spectroscopytechniques for the investigation of structure and dynamics was utilized. On theone hand, ENDOR and ESEEM techniques together with cw EPR experimentsto obtain a detailed description of the structural features of a copper-DNA com-plex. On the other hand, with a cw EPR technique and lineshape analysis it waspossible to access dynamic properties of confined supercooled water.

In Chapter 2 a study of the interaction of copper ion with the DNA model sys-tem guanosine-monophosphate was presented. The application of ENDOR andHYSCORE spectroscopies allowed for an elaborate mapping of the interactionsbetween the copper ion and the magnetic nuclei of the nucleotide ligand, and forthe determination of the metal ion site coordination. The phosphate group of thenucleotide could be localized by direct ENDOR observation, although it belongsto the second coordination sphere and interacts with directly coordinated wateronly via a hydrogen bond. This study could be considered a start point to furtherstudy metal ion binding to mononucleotides or oligonucleotides in solution.

In Chapter 3 EPR spectroscopy was further used to study the interaction ofcopper with a polymer of alternating deoxyguanosine and deoxycytosine. CWEPR data show that two copper species are present in this complex. Pulse EPRdata enabled us to formulate a molecular model to assign them. According to thismodel, one species is exclusively coordinated to the guanine. Additionally, for theother species a copper crosslink between a guanine and a cytosine is proposed.This study allowed a better understanding of how the presence of copper ions af-fects the DNA conformation. The main relevant aspect of this work is that metalion interaction with oligonucleotides, like DNA or RNA, can be studied in frozensolution rather than in crystals. In this context, EPR spectroscopy can give in-sight which is not easily accessibly by other techniques such as soaking of DNAcrystals, where conformational changes are blocked by crystal packing. It couldbe interesting to further study the role of more complicated copper complexes, likeazamacrocyclic complexes, in the stabilization of the DNA conformation. With

85

86 Chap. 5 - Summary and Outlook

respect to the occurence of two copper binding sites it would be interesting to trywhether one of them can be saturated by a diamagnetic substitute for copper. Ifit is possible, the methodology developed for guanosine-monophosphate could befully applied to the DNA, without the complication of signal overlap.

Finally, EPR spectroscopy has been used to study the dynamics of confinedsupercooled water around the spin probe TEMPO. We obtained evidence thatwater encapsulated in our wet silica hydrogels could be supercooled down to atleast 198 K, and that the major fraction of water does not crystallize down to95 K. The preparation of these silica hydrogels gives us the possibility to studymetal ion complexes at very low temperature without the use of cryprotectants.In this context, hemoproteins or metal-DNA complexes, could be studied byEPR spectroscopy in water solution, an environment much more similar to theone found in cells. Moreover, the possibility to regulate the pH in the silicahydrogels, for example using water/buffer solutions, allows for encapsulation ofbiomacromolecules without denaturation.

Bibliography

A. Abragam and M. H. L. Pryce. Theory of the nuclear hyperfine structure of theparamagnetic resonance spectra in crystals. Proc. Roy. Soc. (London), A205:135–153, 1951.

M. Alcoutlabi and G. B. McKenna. Effects of confinement on material behaviourat the nanometre size scale. J. Phys. Condens. Matter, 17:R461–R524, 2005.

F. H. Allen and O. Kennard. Chem. Des. Automat. News, 8:31–37, 1993.

K. K. Andersson, P. P. Schmidt, B. Katterle, K. R. Strand, A. E. Palmer, S. K.Lee, E. I. Solomon, A. Graslund, and A. L. Barra. Examples of high-frequencyEPR studies in bioinorganic chemistry. J. Biol. Inorg. Chem., 8:235–247, 2003.

N. M. Atherton and A. J. Horsewill. Proton ENDOR of Cu(H2O)62+ in

Mg(NH4)2(SO2)4·6H2O. Mol. Phys., 37:1349–1361, 1979.

R. Bergman and J. Swenson. Dynamics of supercooled water in confined geometry.Nature, 403:283–286, 2000.

I. Bertini, H. B. Gary, S. J. Lippard, and J. S. Valentine. Bioinorganic Chemistry.University Science Books, Mill Valley, CA, 1994.

R. Bohmer, K. L. Ngai, C. A. Angell, and D. J. Plazek. Nonexponential re-laxations in strong and fragile glass formers. J. Chem. Phys., 99:4201–4209,1993.

F. Bruni, M. A. Ricci, and A. K. Soper. Water confined in vycor glass. I. Aneutron diffraction study. J. Chem. Phys., 109:1478–1485, 1998.

C. Calle, A. Sreekanth, M. V. Fedin, J. Forrer, I. Garcia-Rubio, I. A. Gromov,D. Hinderberger, B. Kasumaj, P. Leger, B. Mancosu, G. Mitrikas M. G. San-tangelo, S. Stoll, A. Schweiger, R. Tschaggelar, and J. Harmer. Pulse EPRmethods for studying chemical and biological samples containing transitionmetals. Helv. Chim. Acta, 89:2495–2521, 2006.

G. G. Cameron. Comprehensive Polymer Science: The Synthesis, Characteri-zation, Reactions and Applications of Polymers. Vol. 1, Pergamon, Oxford,1989.

87

88 BIBLIOGRAPHY

M. Cammarata, M. Levantino, A. Cupane, A. Longo, A. Martorana, and F. Bruni.Structure and dynamics of water confined in silica hydrogels: X-ray scatteringand dielectric spectroscopy studies. Eur. Phys. J., 12:S63–S66, 2003.

S. Capaccioli, K. L. Ngai, and N. Shinyashiki. The Johari-Goldstein beta-relaxation of water. J. Phys. Chem. B, 111:8197–8209, 2007.

B. Chen, E. E. Sigmund, and W. P. Halperin. Stokes-Einstein relation in super-cooled aqueous solutions of glycerol. Phys. Rev. Lett., 96:145502, 2006a.

S. H. Chen, F. Mallamace, C. Y. Mou, M. Broccio, C. Corsaro, A. Faraone, andL. Liu. The violation of the Stokes-Einstein relation in supercooled water. Proc.Natl. Acad. Sci. U.S.A., 103:12974–12978, 2006b.

R. L. Cook, H. E. King, C. A. Herbst, and D. R. Herschbach. Pressure andtemperature dependent viscosity of two glass forming liquids: Glycerol anddibutyl phthalate. J. Chem. Phys., 100:5178–5189, 1994.

A. Cupane, M. Levantino, and M. G. Santangelo. Near infrared spectra of waterconfined in silica hydrogels in the temperature interval 365-5 K. J. Phys. Chem.B, 106:11323–11328, 2002.

B. C. Dave, B. Dunn, J. S. Valentine, and J. I. Zink. Sol-gel encapsulationmethods for biosensors. Anal. Chem., 66:1120A–1127A, 1994.

E. R. Davies. New pulse ENDOR technique. Phys. Lett. A, 47:1–2, 1974.

V. J. DeRose and B. M. Hoffman. Methods in Enzymology. Vol. 246, AcademicPress, San Diego, 1995.

S. A. Dikanov, L. Y. Xun, A. B. Karpiel, A. M. Tyryshkin, and M. K. Bowman.Orientationally selected two-dimensional ESEEM spectroscopy of the Rieske-type iron-sulfur cluster in 2,4,5-trichlorophenoxyacetate monooxygenase fromBurkholderia cepacia AC1100. J. Am. Chem. Soc., 118:8408–8416, 1996.

S. A. Dzuba. Libration motion of guest spin probe molecules in organic glasses:CW EPR and electron spin echo study. Spectrochim. Acta, 56A:227–234, 2000.

G. R. Eaton, S. S. Eaton, and L.J. Berliner (Eds). Biological magnetic resonance.Vol. 19, Kluwer Publishing, Amsterdam, 2000.

G. L. Eichhorn and P. Clark. Interaction of metal ions with polynucleotides andrelated compounds. V. The unwindind and rewindind of DNA strands underthe influence of copper(II) ions. Proc. Nat. Acad. Sci. U.S.A, 47:586–592, 1965.

L. M. Ellerby, C. R. Nishida, F. Nishida, S. A. Yamanaka, B. Dunn, J. S. Valen-tine, and J. I. Zink. Encapsulation of proteins in transparent porous silicateglasses prepared by the sol-gel method. Science, 255:1113–1115, 1992.

BIBLIOGRAPHY 89

A. Faraone, L. Liu, Mou C. Y, C. W. Yen, and S. H. Chen. Fragile to strongliquid transition in deeply supercooled confined water. J. Chem. Phys., 121:10843–10846, 2004.

J. A. Farrar, F. Neese, P. Lappalainen, P. M. H. Kroneck, M. Saraste, W. G.Zumft, and A. J. Thomson. The electronic structure of CuA: A novel mixed-valence dinuclear copper electron-transfer center. J. Am. Chem. Soc., 118:11501–11514, 1996.

M. A. Fuertes, V. Cepeda, C. Alonso, and J. M. Perez. Molecular mechanisms forthe B-Z transition in the example of poly[d(G-C)·d(G-C)] polymers. A criticalreview. Chem. Rev., 106:2045–2064, 2006.

M. L. S. Garcia and J. A. S. Smith. N-14 and H-2 quadrupole double-resonancein salts of Cytosine and Adenine. J. Chem. Soc. Perkin Trans. II, pages 1401–1408, 1983.

B. H. Geierstanger, T. F. Kagawa, S. L. Chen, G. J. Quigley, and P. S. Ho. Base-specific binding of copper(II) to Z-DNA. The 1.3 A single crystal structure ofd(m5CGUAm5CG) in the presence of CuCl2. J. Biol. Chem., 266:20185–20191,1991.

H. B. Gray. Biological inorganic chemistry at the beginning of the 21st century.Proc. Natl. Acad. Sci. U.S.A., 100:3563–3568, 2003.

O. H. Griffith, P. J. Dehlinger, and S. P. Van. Shape of hydrophobic barrier ofphospholipid bilayers (evidence for water penetration in biological membranes).J. Membr. Biol., 15:159–192, 1974.

I. Gromov, J. Shane, J. Forrer, R. Rakhmatoullin, Y. Rozentzwaig, andA. Schweiger. A Q-band pulse EPR/ENDOR spectrometer and the implemen-tation of advanced 1- and 2-dimensional pulse EPR methodology. J. Magn.Res., 149:196–203, 2001.

E. L. Hahn. Spin echoes. Phys. Rev., 80:580–594, 1950.

Y. P. Handa and D. D. Klug. Heat capacity and glass transition behavior ofamorphous ice. J. Phys. Chem., 92:3323–3325, 1988.

S. Hiai, K. Suzuki, and E. Moriguchi. Effects of Cupric ions on thermal denatu-ration of nucleic acid - transforming activity and DNA denatured with Curpicions. J. Mol. Biol., 11:672–691, 1965.

P. Hoefer, A. Grupp, H. Nebenfuhr, and M. Mehring. Hyperfine sublevel corre-lation (HYSCORE) spectroscopy - a 2D ESR investigation of the squaric acidradical. Chem. Phys. Lett., 132:279–282, 1986.

90 BIBLIOGRAPHY

C. G. Hoogstraten, C. V. Grant, T. E. Horton, V. J. DeRose, and R. D. Britt.Structural analysis of metal ion ligation to nucleotides and nucleic acids usingpulsed EPR spectroscopy. J. Am. Chem. Soc., 124:834–842, 2002.

K. N. Hu, C. Song, H. H. Yu, T. M. Swager, and R. G. Griffin. High frequencydynamic nuclear polarization using biradicals: A multifrequency EPR lineshapeanalysis. J. Chem. Phys., 128:052302, 2008.

K. Ito, C. T. Moynihan, and C. A. Angell. Thermodynamic determination offragility in liquids and a fragile-to-strong liquid transition in water. Nature,398:492–495, 1999.

G. Jeschke. Distance measurements in the nanometer range by pulse EPR.ChemPhysChem, 3:927–932, 2002.

G. Jeschke and Ye. Polyhach. Distance measurements on spin-labelled biomacro-molecules by pulsed electron paramagnetic resonance. Phys. Chem. Chem.Phys., 9:1895–1910, 2007.

G. Jeschke, R. Rakhmatullin, and A. Schweiger. Sensitivity enhancement bymatched microwave pulses in one- and two-dimensional electron spin echo en-velope modulation spectroscopy. J. Magn. Reson., 131:261–271, 1998.

G. P Johari, A. Hallbrucker, and E. Mayer. The glass-liquid transition of hyper-quenched water. Nature, 330:552–553, 1987.

T. Kawamura, S. Matsunam, and T. Yonezawa. Solvent effects on g-value ofdi-t-butyl nitric oxide. Bull. Chem. Soc. Jap., 40:1111–&, 1967.

E. P. Kirilina, I. A. Grigoriev, and S. A. Dzuba. Orientational motion of nitroxidesin molecular glasses: Dependence on the chemical structure, on the molecularsize of the probe, and on the type of the matrix. J. Chem. Phys., 121:12465–12471, 2004.

T. J. Kistenmacher, M. Rossi, J. P. Caradonna, and L. G. Marzilli. Solutionand solid state studies on the interactions of protonated cytosine salts. III.Interpyrimidine base stacking and asymmetric interbase hydrogen bonding inthe structure of 1-methylcytosine hemihydroiodide hemihydrate. Adv. Mol.Relax. Interact. Proc, 15:119–133, 1979.

D. Kivelson. Theory of ESR linewidths of free radicals. J. Chem. Phys., 33:1094–1106, 1960.

D. Kurad, G. Jeschke, and D. Marsh. Lipid membrane polarity profiles by high-field EPR. Biophys. J., 85:1025–1033, 2003.

S. J. Lippard and J. M. Berg. Principles of Bioinorganic Chemistry. UniversityScience Books, Mill Valley, CA, 1994.

BIBLIOGRAPHY 91

Z. L. Madi, S. Van Doorslaer, and A. Schweiger. Numerical simulation of one- andtwo-dimensional ESEEM experiments. J. Magn. Reson., 154:181–191, 2002.

F. Mallamace, M. Broccio, C. Corsaro, A. Faraone, U. Wanderlingh, L. Liu,C. Y. Mou, and S. H. Chen. The fragile-to-strong dynamic crossover transitionin confined water: Nuclear magnetic resonance results. J. Chem. Phys., 124:161102, 2006.

P. Manikandan, B. Epel, and D. Goldfarb. Structure of copper(II)-histidine basedcomplexes in frozen aqueous solutions as determined from high-field pulsedelectron nuclear double resonance. Inorg. Chem., 40:781–787, 2001.

D. Marsh. Biological Magnetic Resonance Spin Labeling: Theory and Applica-tions. Vol. 8, Academic Press, New York and London, 1989.

B. R. McGarvey. In: Carlin, R.L. (ed) Transition Metal Chemistry. Dekker, NewYork, pp 89-201, 1966.

W. B. Mims. Pulsed ENDOR experiment. Proc. R. Soc. Lond, 283:452–457,1965.

O. Mishima and H. E. Stanley. The relationship between liquid, supercooled andglassy water. Nature, 396:329–335, 1998.

G. Mitrikas, C. Calle, and A. Schweiger. Asymmetric spin density distribution inthe copper(II) complex of N-confused tetraphenylporphyrin: A multifrequencycontinuous-wave and pulse EPR study. Angew. Chem. Int. Ed., 44:3301–3303,2005.

M. Oguni, S. Maruyama, K. Wakabayashi, and A. Nagoe. Glass transitions ofordinary and heavy water within silica-gel nanopores. Chem. Asian J., 2:514–520, 2007.

R. Owenius, M. Engstrom, M. Lindgren, and M. Huber. Influence of solventpolarity and hydrogen bonding on the EPR parameters of a nitroxide spinlabel studied by 9 GHz and 95 GHz EPR spectroscopy and DFT calculations.J. Phys. Chem. A, 105:10967–10977, 2001.

J. Peisach and W. E. Blumberg. Structural implications derived from the analysisof electron paramagnetic resonance spectra of natural and artificial copperproteins. Arch. Biochem. Biophys., 165:691–708, 1974.

A. Poeppl and L. Kevan. A pratical strategy for determination of proton hyper-fine interaction parameters in paramagnetic transition metal ion complexes bytwo-dimensional HYSCORE electron spin resonance spectroscopy in disorderedsystems. J. Phys. Chem., 100:3387–3394, 1996.

L. G. Rowan, E. L. Hahn, and W. B. Mims. Electron-spin-echo envelope modu-lation. Phys. Rev., 137:A61–A71, 1965.

92 BIBLIOGRAPHY

G. A. Rupley and G. Careri. Protein hydration and function. Adv. Protein Chem.,41:37–172, 1991.

M. G. Santangelo, M. Levantino, E. Vitrano, and A. Cupane. Ferrycytochrome cencapsulated in silica hydrogels: correlation between active site dynamics andsolvent structure. Biophys. Chem., 103:67–75, 2003.

D. Y. Sasaki. Control of membrane structure and organization through chemicalrecognition. Cell Biochem. Biophys., 39:145–161, 2003.

G. Schiro, M. Sclafani, C. Caronna, F. Natali, M. Plazanet, and A. Cupane.Dynamics of myoglobin in confinement: an elastic and quasi-elastic neutronscattering study. Chem. Phys., 345:259–266, 2008.

D. J. Schneider and J. H. Freed. Calculating slow motional magnetic resonancespectra. A user’s guide. Biol. Magn. Reson., 8:1–76, 1989.

P. M. Schosseler, B. Wehrli, and A. Schweiger. Complexation of copper(II) withcarbonate ligands in aqueous solution: a cw and pulse EPR study. Inorg.Chem., 36:4490–4499, 1997.

A. Schweiger and G. Jeschke. Principles of pulse electron paramagnetic resonance.Oxford University Press, New York, 2001.

W. S. Sheldrick. Metal-ribose binding in a transition metal-nucleotide complex- preparation and structure of a polymeric copper(II) complex of Guanosine2’-monophosphate. Acta Crystallogr. Sect. B, 37:1820–1824, 1981.

H. Sigel and B. Song. Solution structure of nucleotide-metal ion complexes. Isomerequilibria. Met. Ions Biol. Syst., 32:135–205, 1996.

H. Sigel, S. S. Massoud, and N. A. Corfu. Comparison of the extent ofmacrochelate formation in complexes of divalent metal ions with Guanosine(GMP2-), Inosine (IMP2-), and Adenosine 5’-Monophosphate (AMP2-). Thecrucial role of N-7 basicity in metal ion-nucleic base recognition. J. Am. Chem.Soc., 116:2958–2971, 1994.

E. Sletten and B. Lie. Copper complex of Guanosine-5’-monophosphate. ActaCrystallogr. Sect. B, 32:3301–3304, 1976.

R. J. Speedy and C. A. Angell. Isothermal compressibility of supercooled waterand evidence for a thermodynamic singularity at -45 C. J. Chem. Phys., 65:851–858, 1976.

B. Spingler and C. Da Pieve. Induction of B- to Z-DNA transition by copper andzinc complexes with C(15) substituted macrocyclic pentaaza ligands. DaltonTrans., pages 1637–1643, 2005.

S. Stoll and A. Schweiger. Easyspin, a comprehensive software package for spectralsimulation and analysis in EPR. J. Magn. Reson., 178:42–55, 2006.

S. Stoll, C. Calle, G. Mitrikas, and A. Schweiger. Peak suppression in ESEEMspectra of multinuclear spin system. J. Magn. Reson., 177:93–101, 2005.

H. Svajdlenkova, D. Racko, and J. Bartos. Spin probe reorientation and itsconnections with free volume and relaxion dynamics: Diglycidyl-ether of bis-phenol A. J. Non-Cryst. Solids, 354:1855–1861, 2008.

Z. Veksli, M. Andreis, and B. Rakvin. EPR spectroscopy for the study of polymerheterogeneity. Prog. Polym. Sci., 25:949–986, 2000.

J. A. Wiel, J. R. Bolton, and J.E. Wertz. Electron paramagnetic resonance -elementary theory and pratical applications. Wiley, New York, 1994.

S. A. Zager and J. H. Freed. Electron-spin relaxation and molecular dynamics inliquids. I. Solvent dependence. J. Chem. Phys., 77:3344–3359, 1982.

P. P. Zanker, G. Jeschke, and D. Goldfarb. Distance measurements betweenparamagnetic centers and a planar object by matrix Mims electron nucleardouble resonance. J. Chem. Phys., 122:24515, 2005.

C. Zimmer, G. Luck, H. Fritzsche, and H. Triebel. DNA-copper(II) complex andthe DNA conformation. Biopolymers, 10:441–463, 1971.

Acknowledgements

Many people have supported, stimulated and motivated me during my PhD the-sis. I would therefore like to thank them:

Prof. Gunnar Jeschke who agreed to take over the supervision of my thesis afterthe loss of Prof. Arthur Schweiger. His advice and encouragement were invalu-able. It has been a wonderful experience working with him.

Prof. Arthur Schweiger who first gave me the opportunity to discover the won-derful world of EPR.

Dr. Bernhard Spingler for the trust he has posed in me and for having alwaysencouraged me with enthusiasm.

Prof. Beat H. Meier who generously took care of the EPR group during thetransition phase.

Prof. Roland Riek for agreeing to be co-examiner of this thesis and for providingvaluable input.

Prof. Ernesto Di Iorio who first suggested me to study EPR and introduced meto Arthur.

Prof. Antonio Cupane, supervisor of my undegraduate thesis in Palermo, whohas always encouraged me and supported my career choices. I have also enjoyedcollaborating with him on one of the projects presented in this thesis.

Dr. Ines Garcia-Rubio for all her unconditional help, for her friendship and forall the good times we spent together. I feel lucky to have met her.

Dr. Cinzia Finazzo for her help and useful advices on all the aspects, both sci-entific and practical, of my stay in Switzerland.

Dr. Enrica Bordignon for her patience and good humor in the face of my igno-rance of some aspects of chemistry.

95

Dr. Fabio La Mattina who invited me to visit Zurich and showed me around thecity. I have immediately realized that Zurich was a wondeful city where to spendsome years of my life.

Dr. George Mitrikas for his support, for training me in the use of the spectrom-eters and simulating programs during the realization of the first part of this work.

Jorg Forrer for explaining me everything about the EPR instrumentation in Ital-ian and for motivating the whole group in the tough moments.

All the present and former members of the EPR group at ETH for the niceconversations and the help they gave to me: Besnik Kasumaj, Dr. Maxim Yu-likov, Dr. Yevhen Polyhach, Petra Luders, Thomas Kohn, Udo Kielmann, ReneTschaggelar, Dr. Igor Gromov, Dr. Jeffrey Harmer, Dr. Anandaram Sreekanth,Dr. Carlos Calle, Dr. Stefan Stoll, Dr. Dariush Hinderberger, Bruno Mancosu.

Alfredo and Philipp for the preparation of the nucleotide samples.

Dr. Erich Meister for ensuring the outstanding functioning of praktikum labs.

Veronika Sieger for answering all the questions about administrative proceduresI had.

Irene Muller who was always there when I needed help, either for burocratic orprivate matters.

my friend Francesco for helping me out of Latex traps.

my friends Marco, Chiara, Valeria, Rino, Leonardo, Alice, Antonella for all thesupport that I always get from them.

My family – dad, mom, Rosa, Enzo, Filippo, Pamela – for their love and under-standing.

I would like to mentions my niece Sara and my nephews Damiano and Federicowho have made my life happier.

Finally, I would like to thank my love Matteo. He has been always present and Icould rely on him all the time. He has always believed in me.

Publications

Papers

M.G. Santangelo, M. Levantino, A. Cupane, and G. Jeschke. Solvation of a probemolecule by fluid supercooled water in a hydrogel at 200 K. J. Phys. Chem. B,112:15546-15553, 2008.

M.G. Santangelo, A. Medina-Molner, A. Schweiger, G. Mitrikas, and B. Spingler.Structural analysis of Cu(II) ligation to the 5′-GMP nucleotide by pulse EPRspectroscopy. J. Biol. Inorg. Chem., 12:767-775, 2007.

M.G. Santangelo, M. Levantino, E. Vitrano, and A. Cupane. Ferricytochromec encapsulated in silica hydrogels: correlation between active site dynamics andsolvent structure. Biophys. Chem., 103:67-75, 2003.

A. Cupane, M. Levantino, and M.G. Santangelo. Near-infrared spectra of waterconfined in silica hydrogels in the temperature interval 365-5 K. J. Phys. Chem.B, 106:11323-11328, 2002.

Reviews

B. Spingler, C. Da Pieve, A. Medina-Molner, P.M. Antoni, and M.G. Santangelo.Interaction of novel metal complexes with DNA: synthetic and structural aspects.Chimia, 63:153-156, 2009.

C. Calle, A. Sreekanth, M.V. Fedin, J. Forrer, I. Garcia-Rubio, I.A. Gromov, D.Hinderberger, B. Kasumaj, P. Leger, B. Mancosu, G. Mitrikas, M.G. Santan-gelo, S. Stoll, A. Schweiger, R. Tschaggelar, and J. Harmer. Pulse EPR methodsfor studying chemical and biological samples containing transition metals. Helv.Chim. Acta, 89:2495-2521, 2006.

97

Curriculum Vitae

Personal data

Name : Maria Grazia Santangelo

Birthdate : 5th July 1976

Birthplace : Castelvetrano (TP), Italy

Citizenship : Italian

Education

1995-2002 Degree in Physics, University of Palermo, Palermo, Italy.

2005-2009 Ph.D student in the Laboratory of Physical Chemistry ofETH Zurich (Switzerland) under the supervision of Prof. A.Schweiger and Prof. G. Jeschke.

Professional Experience

2003-2004 IT Consultant, Accenture Technology Solutions, Milan(Italy).

2005-2009 Teaching assistant (class lectures and laboratory experiments)in physical chemistry courses, ETH Zurich (Switzerland).

99

rights / license: research collection in copyright - non ...6/eth-6-02.pdf · strutturali di un...

Documents